Split (1)

train · ~9.11M rows (showing the first 93.5k)

identifier stringlengths 11 32	pdf_url stringlengths 17 4.62k ⌀	lang stringclasses 120 values	error stringclasses 1 value	title stringlengths 2 500 ⌀	source_name stringlengths 1 435 ⌀	publication_year float64 1.9k 2.02k	license stringclasses 3 values	word_count int64 0 1.64M	text stringlengths 1 9.75M
https://openalex.org/W3098387096	https://ogsh.abvpress.ru/jour/article/download/561/438	Russian	null	Does morphology have real impact on local and distant recurrences in head and neck cutaneous melanoma?	Opuholi golovy i šei	2,020	cc-by	9,551	Опухоли ГОЛОВЫ и ШЕИ HEAD and NECK tumors 3’2020 Оригинальное исследование Опухоли ГОЛОВЫ и ШЕИ HEAD and NECK tumors 3’2020 Оригинальное исследование Том 10 Vol. 10 Does morphology have real impact on local and distant recurrences in head and neck cutaneous melanoma? We consider that clear surgical margins for any thickness of cutaneous melanoma of head and neck should be as follows: <2 mm – 0.46 mm (p = 0.13), 2.01–4.00 mm – 0.58 mm (p = 0.002), >4 mm – 0.72 mm (p = 0.016). In our work, the influence of the main prog- nostic factors, such as tumor thickness according to Breslow, level of invasion according to Clark and ulceration on the frequency of head and neck cutaneous melanoma local recurrences had no impact. Key words: cutaneous melanoma, head and neck, local recurrence, surgical resection margin, progression-free survival For citation: Mudunov A. M., Pak M. B., Demidov L. V., Baryshnikov K. A. Does morphology have real impact on local and distant recurrences in head and neck cutaneous melanoma? Opukholi golovy i shei = Head and Neck Tumors 2020;10(3):55–64. DOI: 10.17650/2222-1468-2020-10-3-55-64 DOI: 10.17650/2222-1468-2020-10-3-55-64 Does morphology have real impact on local and distant recurrences in head and neck cutaneous melanoma? Progression-free survival didn’t correlate with the size of surgical resection margins. The survival rates were the best with the lowest resection margin under 0.5 cm (77.3 %) and the worst with the highest resection margin more then 1.0 cm (38.7 %). That means that the treat ment results don’t depend on the width increase of tumor resection margin. f p y , y , g f f g The aim of the study is to define influence of the main prognostic factors such as tumor thickness according to Breslow, the level of invasion according to Clark and the presence of ulceration on the frequency of local recurrence with cutaneous melanoma of head and neck. The aim of the study is to define influence of the main prognostic factors such as tumor thickness according to Breslow, the level of invasion according to Clark and the presence of ulceration on the frequency of local recurrence with cutaneous melanoma of head and neck. Materials and methods. The research involved 174 patients with cutaneous melanoma of head and neck (1995–2014). According to our index of contraction of a skin flap (median 30 %) the true borders of resection were clearly defined within all the patients. Thereby, 3 groups were identified with the following resection margin: <0.5; 0.6–1.0 and >1.0 cm, where followed-up treatment results were analyzed. Results. Progression-free survival didn’t correlate with the size of surgical resection margins. The survival rates were the best with the lowest resection margin under 0.5 cm (77.3 %) and the worst with the highest resection margin more then 1.0 cm (38.7 %). That means that the treat ment results don’t depend on the width increase of tumor resection margin. were identified with the following resection margin: <0.5; 0.6–1.0 and >1.0 cm, where followed-up treatment results were analyzed. Results. Progression-free survival didn’t correlate with the size of surgical resection margins. The survival rates were the best with the lowest resection margin under 0.5 cm (77.3 %) and the worst with the highest resection margin more then 1.0 cm (38.7 %). That means that the treat ment results don’t depend on the width increase of tumor resection margin. p f g Conclusions. Does morphology have real impact on local and distant recurrences in head and neck cutaneous melanoma? A. M. Mudunov, M. B. Pak, L. V. Demidov, K. A. Baryshnikov N. N. Blokhin National Medical Research Center of Oncology, Ministry of Health of Russia; Russia, 115478 Moscow, Kashirskoye Hwy, 24 Contact: Maxim Pak mbpak@yandex.ru Introduction. The term “local recurrence” is usually understood as regrowth of a tumor after surgical treatment. The regrowth appears within 3–5 cm from postoperative cicatrix. The causes for such prolonged tumor growth or recurrence of patients with cutaneous melanoma are nonradical surgical treatment as well as satellite or transit metastases that were not removed in-block with primary tumor. A great num- ber of clinical researches, aimed at examination of melanoma, its patterns, anatomical criteria and features of clinical course, gave an op- portunity to separate satellite or transit metastases into an independent group. Such metastases are realized inside or subdermally, up to 2 cm or more than 2 cm from the primary tumor, yet, not reaching the location of the first regional barrier. Introduction. The term “local recurrence” is usually understood as regrowth of a tumor after surgical treatment. The regrowth appears within 3–5 cm from postoperative cicatrix. The causes for such prolonged tumor growth or recurrence of patients with cutaneous melanoma are nonradical surgical treatment as well as satellite or transit metastases that were not removed in-block with primary tumor. A great num- ber of clinical researches, aimed at examination of melanoma, its patterns, anatomical criteria and features of clinical course, gave an op- portunity to separate satellite or transit metastases into an independent group. Such metastases are realized inside or subdermally, up to 2 cm or more than 2 cm from the primary tumor, yet, not reaching the location of the first regional barrier. The aim of the study is to define influence of the main prognostic factors such as tumor thickness according to Breslow, the level of invasion according to Clark and the presence of ulceration on the frequency of local recurrence with cutaneous melanoma of head and neck. Materials and methods. The research involved 174 patients with cutaneous melanoma of head and neck (1995–2014). According to our index of contraction of a skin flap (median 30 %) the true borders of resection were clearly defined within all the patients. Thereby, 3 groups were identified with the following resection margin: <0.5; 0.6–1.0 and >1.0 cm, where followed-up treatment results were analyzed. Results. Introduction Thus, over the past 30 years, the main problem of the surgical treatment of primary cutaneous melanoma all over the world was the question about the size of resection mar- gins [10, 15–21]. Cutaneous melanoma refers to a range of tumors with unexpected clinical outcome. On the one hand, it is connec ted with the aggressive character of the tumor and severity of prognosis, on the other hand with the possibility of curing of the disease on the stage of local process [1, 2]. Despite the fact that this pathology constitutes not more than 12 % among all malignant skin neoplasms, death rate due to this pathology constitutes more than 75 % among all death oc- currences induced by malignant dermal diseases. Accord- ing to the data from various authors, cutaneous melanoma of head and neck occurs in approximately 20–30 % of cases [3]. Over the past 3 decades there has been a sharp increase in the incidence of cutaneous melanoma disease [4–6]. There were 10 454 new cases of cutaneous melanoma pa- tients in the year of 2016 in Russia. From 2006 to 2016 average annual increase of diseased constituted 3.07 % [7]. The peak incidence of diseases is observed in the age group of 60–64. Since 1985 6 prospective clinical researches studying the optimum resection margin borders of primary cutaneous melanoma resection margin of the skin with different tumor thickness have been conducted. It should be pointed out that patients with head and neck cutaneous melanoma were included in only one of such large studies (European Trial French, 1993 (Khayat D. et al.)). According to table 1 it is clear that all researches were conducted more than 10–20 years ago and it was possible to study the question of safe resection margin at any stage of the neoplastic process throughout this time, except for the stage IIC (cutaneous melanoma >4 mm thick with ulceration “T4b”), which, according to our data, is the most common in the Russian population due to late diagnosis, and the least studied in prospective randomized studies. In Russia, according to the data for the year of 2018, around one third of patients were diagnosed with I–II stag- es of the disease – 79.2 %, and III and IV stages of the disease – 10.5 and 8.6 %, respectively. Introduction However, if one separates I and II stages, the incidence of this disease will not seem so optimistic. Thus patients with “early” cutaneous melanoma (I stage) are detected in approximately 10–12 % of cases, but in a more significant amount of cases – 60 % of patients are exposed with “interim” neoplastic process (II stage), a long-term prognosis of life that cannot be con- sidered satisfactory. Statistical analysis of the conducted research showed that reducing the boundaries of the resection margin does not lead to increase of the number of recurrences and overall survival rate remains the same in both groups. Thus, the result has always been in favor of an economical resection of primary cutaneous melanoma. Nevertheless, in 2015 Keith Wheatley and his co-authors carried out a meta-analysis of 6 prospective researches described above. According to the authors, economical resection is not inferior to a wider one; however, “a slight difference does not mean that it does not exist”. There were 4233 patients with cutaneous melanoma included in the research, where a wide resection (3, 4 and 5 cm) was compared to a more narrow one (1 or 2 cm). In the result of this systematic overview and meta-analysis a slight worsening of followed-up treatment results was detected [33]. An increase in the incidence of cutaneous melanoma indicates that diagnostic methods for this pathology have improved in recent years, at the same time high mortality from disseminated melanoma is recorded. The main treatment for primary cutaneous melanoma remains surgical removal. Up till the 50–60s of the XX cen- tury, most oncologists performed very extensive, ultraradical operations in order to improve treatment results [8]. Most surgeons had no doubt that a narrow melanoma resection would lead to an inevitable recurrence of the disease. In the literature, one could find recommendations for the surgical treatment of primary cutaneous melanoma with resection margin borders from 2 cm [9] to 15 cm [10–14]. At the same time, more and more works on clinical trials of a more eco- nomical resection began to appear. Speaking of followed-up treatment results of cutaneous melanoma, one should notice that life prognosis isn’t influenced by safe resection margin. The main morphological factors of primary tumor such as tumor thickness according to Breslow and the presence of invasion [34]. Опухоли ГОЛОВЫ и ШЕИ HEAD and NECK tumors 3’2020 Оригинальное исследование Том 10 Vol. 10 Заключение. С использованием ROC-кривых определены безопасные границы хирургического отступа для меланомы кожи головы и шеи: при толщине меланомы кожи головы и шеи <2 мм отступ 0,46 мм (p = 0,13), при толщине меланомы 0,58 мм – 2,01–4,00 мм (р = 0,002), при толщине меланомы >4 мм – 0,72 мм (р = 0,016). В нашей работе такие факторы, как толщина опухоли по Breslow, уровень инвазии по Clark и наличие изъязвлений, не влияли на частоту местных рецидивов меланомы кожи головы и шеи, что дает основание утверждать, что уменьшение хирургического отступа действительно не приводит к ухудшению отдаленных результатов лечения. Ключевые слова: меланома кожи, голова и шея, местный рецидив, край резекции, выживаемость без прогрессирования Для цитирования: Мудунов A. M., Пак M. Б., Демидов Л. В., Барышников К. A. Влияют ли морфологические признаки на часто- ту местных рецидивов и отдаленных метастазов при меланоме кожи головы и шеи? Опухоли головы и шеи 2020;10(3):55–64. (На англ.). A. M. Мудунов, M. Б. Пак, Л. В. Демидов, К. A. Барышников ФГБУ «Национальный медицинский исследовательский центр онкологии им. Н. Н. Блохина» Минздрава России; Россия, 115478 Москва, Каширское шоссе, 24 A. M. Мудунов, M. Б. Пак, Л. В. Демидов, К. A. Барышников ФГБУ «Национальный медицинский исследовательский центр онкологии им. Н. Н. Блохина» Минздрава России; Россия, 115478 Москва, Каширское шоссе, 24 A. M. Мудунов, M. Б. Пак, Л. В. Демидов, К. A. Барышников ФГБУ «Национальный медицинский исследовательский центр онкологии им. Н. Н. Блохина» Минздрава России; Россия, 115478 Москва, Каширское шоссе, 24 Введение. Под термином «местный рецидив» обычно понимают повторный рост опухоли после хирургического вмешательства, который возникает в пределах 3–5 см от послеоперационного рубца. Причины продолженного роста опухоли или рецидива у па- циентов с меланомой кожи – нерадикальное хирургическое лечение, а также сателлитные или транзитные метастазы, которые не были удалены единым блоком вместе с первичной опухолью. Многочисленные исследования особенностей биологии меланомы, морфологических критериев ее диагностики и особенностей ее клинического течения позволили выделить в отдельную группу сателлитные и транзитные метастазы, которые реализуются внутри- или подкожно на расстоянии <2 см или >2 см от пер- вичной опухоли, но не достигая места расположения первого регионарного барьера. Цель исследования – определить влияние основных прогностических факторов, таких как толщина опухоли по Breslow, уровень инвазии по Clark и наличие изъязвлений, на частоту местных рецидивов при меланоме кожи головы и шеи. Цель исследования – определить влияние основных прогностических факторов, таких как толщина опухоли по Breslow, уровень инвазии по Clark и наличие изъязвлений, на частоту местных рецидивов при меланоме кожи головы и шеи. Материалы и методы. В исследовании приняли участие 174 пациента с меланомой кожи головы и шеи, проходившие лечение в 1995–2014 гг. С помощью полученного нами индекса сокращения кожного лоскута (медиана 30 %) истинные границы резекции Материалы и методы. В исследовании приняли участие 174 пациента с меланомой кожи головы и шеи, проходившие лечение в 1995–2014 гг. С помощью полученного нами индекса сокращения кожного лоскута (медиана 30 %) истинные границы резекции были точно определены у всех пациентов. В зависимости от этого были сформированы 3 группы с отступом <0,5; 0,6–1,0 и >1,0 см, в которых проанализированы отдаленные результаты лечения. Результаты. Безрецидивная выживаемость не коррелировала с величиной хирургического отступа. Она была лучшей при мини- мальном хирургическом отступе (77,3 %) и худшей при максимальном отступе (38,7 %), т. е. отдаленные результаты лечения не зависят от ширины хирургического отступа. 55 55 Introduction At the same time, the size of the surgical resection margin affects the frequency of local recurrences, reflecting the radical nature of surgical intervention in the area of the primary site [35]. The term “local recurrence” is usually understood as regrowth of a tumor after surgical treatment. The regrowth 56 56 Опухоли ГОЛОВЫ и ШЕИ HEAD and NECK tumors 3’2020 Оригинальное исследование Том 10 Vol. 10 Опухоли ГОЛОВЫ и ШЕИ Table 1. Clinical research evaluating the boundaries of resection margin borders for radical removal of primary cutaneous melanoma Research Amount of patients Tumor thickness Resection margin borders TNM / AJCC 2009 Overall survival Periodicity of local recurrences European Trial French, 1993 (Khayat D. et al.) [22, 23] 336 ≤2 mm 2 vs 5 сm I–IIA 87 and 86 % (10‑year) 13.6 and 20 % Swedish I MSG Trial Swedish, 2000 (Cohn- Cedermark G. et al., Ringborg U. et al.) [24, 25] 989 0.8–2.0 mm 2 vs 5 сm I–IIA 79 and 76 % (5‑year) 0.6 and 2.4 % WHO Melanoma Trial, 1991 (Cascinelli N. et al., Veronesi U. et al.) [26, 27] 612 ≤2 mm (0.99 mm in group 1 cm, 1.02 mm in group 3 cm) 1 vs 3 сm I–IIA 96.4 and 96.4 % (10‑year) 0.98 and 0.97 % Intergroup Melanoma Trial Intergroup, 1996 (Balch C. et al., Karakousis C. P. et al.) [28–30] 740 1–4 mm (average thickness 1.8 mm) 2 vs 4 сm I–IIB 70 and 77 % (6‑year) 2.1 and 2.6 % UK Trial BAPS / MSG, 2004 (Thomas J. M. et al.) [31] 900 ≥2 mm (average thickness in group 1 cm – 3 mm, in group 3 cm – 3.1 mm) 1 vs 3 сm I–IIC 32.2 and 30.9 % (5‑year) 8.27 and 5.64 % Swedish II, 2005 (Gillgren P. et al.) [32] 1000 ≥2 mm (pT3, pT4) 2 vs 4 сm I–IIC Preliminary findings: no differences with 5‑year survival detected – cal research evaluating the boundaries of resection margin borders for radical removal of primary cutaneous melanoma present as a local process as well as other manifestations of the disease may not occur, while in the other case reali- zation of metastasis to regional lymph node of neck takes place and the situation meets the “N2c” criterion, 5‑year survival of which does not exceed 40–45 %. Introduction According to many authors, satellitosis happens due to the phenomenon of “sticking” of melanocytes in lymphatic fissure. The issue of likelihood of the development of satellites on this or that particular area of skin with the realization of metastasis to regional lymph node of neck with various thicknesses of primary melanoma remains unsolved. This would allow a more selective approach to the concept of “wide resection of head and neck cutaneous melanoma”, where the issue of economical removal is most relevant. appears within 3–5 cm from postoperative cicatrix. The causes for such prolonged tumor growth or recurrence of patients with cutaneous melanoma are nonradical surgical treatment as well as satellite or transit metastases that were not removed in-block with primary tumor [1, 36, 37]. A great number of clinical researches, aimed at examination of melanoma, its patterns, anatomical criteria and features of clinical course, gave an opportunity to separate satellite or transit metastases into an independent group. Such metastases are realized inside or subdermally, up to 2 cm or more than 2 cm from the primary tumor, yet, not reaching the location of the first regional barrier. We consider it to be necessary to differentiate two fol- lowing notions – recurrence, which appeared straight in cic- atrix, and the realization of satellite or transit metastasis, which is interpreted by many authors as “local recurrence”. We believe that it is wrong to consider their presence as a “lo cal recurrence”, since they represent a sign of the systemic character of the disease and are classified according to the TNM AJCC 8th edition (2017) in the “N” category [38]. It should be noted that the life prognosis of the patients from both groups is different. In the first case a repeated event is However, there is another point of view on the nature of relapsing lesions with cutaneous melanoma. This theory is based on paracrine stimulation of the primary tumor of normal melanocytes with their subsequent transformation into tumorous ones (Prof. L.V. Demidov). Such mechanism arises due to the humoral effect of growth factors of the pri- mary tumor, which can circulate throughout the whole body even after the removal of the primary tumor. These factors 57 57 Опухоли ГОЛОВЫ и ШЕИ HEAD and NECK tumors 3’2020 Оригинальное исследование Том 10 Vol. 10 of the research was to evaluate the skin flap reduction index. Introduction The study included 21 patients with a local form of head and neck cutaneous melanoma. 25 skin flaps were examined. The most frequent anatomical localization of the primary lesion was the facial area – 12 (48 %) cases. In 8 (32 %) cases, the head and in 5 (20 %) cases, the neck. can stimulate the production of the same factors in normal melanocytes, involving an increasing number of normal cells in the process of tumor transformation. Taking this point of view into account, it becomes absolutely clear that such mechanism of a recurrence does not depend on the volume of a removed tissue. In support of this theory, there are a num- ber of individual clinical observations in which a local re- currence occurs simultaneously with remote metastasis. This fact confirms the consistency of existence of paracrine mech- anism. However, further study of the pathogenesis of relaps- ing processes is necessary to confirm this theory. When a skin flap was formed after resection, the distance from the edge of resection margin to the edge of the tumor reduced from 14.1 to 53.3 % and averaged 31.2 ± 10.7 %. Thus, the median contraction of the skin flap was 30 % (fig. 1). It appeared that the skin flap reduction index didn’t signi ficantly differ in relation to the location of the tumor and the age of the patients. Thus, it seems interesting to analyze the influence of dif- ferent factors on the frequency of local recurrences after radical removal of cutaneous melanoma of head and neck. The conducted analysis allowed us to calculate the true value of resection margin of all patients in the retrospective group (n = 174). We stratificated the main group of patients into 3 subgroups depending on the adjusted value of the re- section margin: <0.5 cm, 0.6–1.0 cm and >1.0 cm. * Significant difference compared to the position of the tumor in the facial area. † Significant difference compared to the position of the tumor in the hairy part of the head. Materials and methods This study is based on retrospective data from 174 pa- tients with head and neck cutaneous melanoma who received treatment at the N. N. Blokhin National Medical Research Center of Oncology in 1995–2014. The age of patients ranged from 18 to 92 years (mean age 54.4 ± 16.3 years, median 54 years). When evaluating the results of treatment, depending on the resection margin, it should be also taken into consideration that there is such an issue as a reduction of the skin flap from the moment of resection till the de- scription of it by a pathomorphologist, and there are differ- ences between surgical and pathological resection margins. Due to the fact that the main source of information was the data of histological findings, where the pathomorpho- logical margin was indicated, this circumstance had to be taken into account in order to determine the true boundaries of the resection in the future. Therefore, there was a need to find a method that would objectively restore the original size of the skin flap. This study is based on retrospective data from 174 pa- tients with head and neck cutaneous melanoma who received treatment at the N. N. Blokhin National Medical Research Center of Oncology in 1995–2014. The age of patients ranged from 18 to 92 years (mean age 54.4 ± 16.3 years, median 54 years). When evaluating the results of treatment, depending on the resection margin, it should be also taken into consideration that there is such an issue as a reduction of the skin flap from the moment of resection till the de- scription of it by a pathomorphologist, and there are differ- ences between surgical and pathological resection margins. The minimal resection margin (<0.5 cm) was most often used in the group of patients with primary tumor localization in the facial area – 48.9 % (n = 44) (compared to hairy part of the head – 20.8 % (n = 11) and neck area – 6.4 % (n = 2), p <0.005) (table 2). Surgical margin of 0.6–1.0 cm was most Fig. 1. Materials and methods often recorded in the neck area – 58.1 % (n = 18) compared to hairy part of the head and facial area: 30.2 and 33.3 %, respectively (p <0.005). Surgical resection margin of more than 1.0 cm was most often found in groups of patients with localization of the tumor on the hairy part of the head 49 % (n = 26) and 35.5 % of the neck (n = 11) (compared with the facial area 17.8 %, p <0.005). It is common knowledge, that the presence of metasta- sis in regional lymphatic nodes (pN+) is one of the main fac- tors which significantly worsens the followed-up treatment results of patients with cutaneous melanoma. For that mat- ter, for obtaining an objective assessment of the impact of such a factor as the size of the surgical resection margin, we decided to exclude patients with regional metastasis from the study groups for further analysis. Thus, the group without regional metastasis (pN0) constituted 131 patients. When localizing the tumour in the facial area, the ave rage resection margin value constituted 0.61 ± 0.41 and was significantly smaller (compared to the hairy part of the head 1.04 ± 0.63 (p = 0.000002) and the neck area 1.02 ± 0.50 (p = 0.000017). Since the thickness of the tumor according to Breslow is the most important prognostic factor for the local cuta- neous melanoma, we analyzed the size of the surgical resec- tion margin depending on the thickness of the primary me lanoma, and the following tendency was detected. In the group of patients with “thin” melanoma (≤1 mm), in most cases the minimum (<0.5 cm) surgical resection margin was performed – 16 (55 %), a bit less often 0.6–1.0 cm – 11 (38 %), and margin >1.0 cm in only 2 (7 %) patients (table 3). A detailed analysis of patients with a surgical margin tumor thickness (1.01–2.00 and 2.01–4.00 mm) showed almost the same distribution of patients in groups depending on the size of the surgical resection margin. At the same time, surgical resection margin of up to 1.0 cm was performed in these groups, in most cases (79 and 76.9 %). In the group of patients with “thick” cutaneous melanoma (>4 mm), the maximum surgical margin (>1.0 cm) was performed with 41 (41.5 %) patients, and twice less often – 22 times (22.2 %) – the minimum one was (<0.5 cm). Materials and methods Skin flap reduction index 25 0 24 23 22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Skin flap reduction index Median Due to the fact that the main source of information was the data of histological findings, where the pathomorpho- logical margin was indicated, this circumstance had to be taken into account in order to determine the true boundaries of the resection in the future. Therefore, there was a need to find a method that would objectively restore the original size of the skin flap. For this reason, we introduced a skin flap reduction index – the amount of tissue volume loss, which is estimat- ed as the reduction in the distance from the edge of the re- section to the edge of the tumor before and after surgery and which is expressed as a percentage. A prospective part of the research was performed in the Department of Head and Neck Tumors of the N. N. Blokhin National Medical Research Center of Oncology in 2013–2015. The purpose Fig. 1. Skin flap reduction index Table 2. The size of resection margin depending on localization of position of the primary tumor (n = 174) Resection margin Hairy part of the head (n = 53) Facial area (n = 90) Regions of neck (n = 31) Overall (n = 174) <0.5 cm 11 (20.8 %)* 44 (48.9 %) 2 (6.4 %)* 57 (32.8 %) 0.6–1.0 cm 16 (30.2 %) 30 (33.3 %) 18 (58.1 %)† 64 (36.8 %) >1.0 cm 26 (49.0 %) 16 (17.8 %) 11 (35.5 %)* 53 (30.4 %) Average, cm 1.04 ± 0.63* 0.61 ± 0.41 1.02 ± 0.50* 0.82 ± 0.54 * Significant difference compared to the position of the tumor in the facial area. Table 2. The size of resection margin depending on localization of position of the primary tumor (n = 174) * Significant difference compared to the position of the tumor in the facial area. † Significant difference compared to the position of the tumor in the hairy part of the head. 58 58 Опухоли ГОЛОВЫ и ШЕИ HEAD and NECK tumors 3’2020 Оригинальное исследование Том 10 Vol. 10 Also, all the patients were divided into two groups according to the presence or absence of ulceration of pri- mary skin melanoma: 105 (60.3 %) against 68 (39.1 %) (table 4). Materials and methods A detailed analysis of the surgical resection margin de- pending on the stage of TNM / AJCC (the distribution was performed in the group without regional metastasis “pN0”) generally showed a similar tendency, as when analyzing the general group of patients with surgical resection margin depending on the thickness of the tumor according to Bre slow (n = 174) (table 3). Surgical resection margin ≤0.5 cm was most often performed at the stage of melanoma 0 – IA / B – 41.6 % (table 5). In stages IIA and IIB, all types of surgical resection margin were used approximately similarly often. The maximum surgical resection margin (>1.0 cm) was most often (3 times) used in the IIC stage of the tumor process. ( ) Thus, a greater surgical resection margin was used with a greater local prevalence of the tumor process. In order to determine the reference range of the neces- sary resection margin that affects the prognosis of the disease course, we analyzed the ROC curves (Receiver Operating Characteristic) at different tumor thicknesses in relation to the frequency of progression (fig. 2). Table 3. The size of the surgical resection margin depending on the thickness of the tumor according to Breslow (n = 174) Resection margin Thickness of the tumor according to Breslow ≤1 mm (n = 29) 1.01–2.00 mm (n = 19) 2.01–4.00 mm (n = 26) >4 mm (n = 99) in situ (n = 1) <0.5 cm 16 (55.0 %)* 9 (47.5 %)* 9 (34.6 %) 22 (22.2 %) 1 (100.0 %) 0.6–1.0 cm 11 (38.0 %) 6 (31.5 %) 11 (42.3 %) 36 (36.3 %) 0 % >1.0 cm 2 (7.0 %)* 4 (21.0 %) 6 (23.1 %) 41 (41.5 %) 0 % * Significant differences compared to a tumor with thickness more than 4 mm. Thus, with an increase in the thickness of the tumor, the size of the surgical resection margin gradually increased. Table 4. Patient distribution according to ulceration Ulceration Hairy part of the head (n = 53) Facial area (n = 90) Neck area (n = 31) Total (n = 174) N 17 (38 1 %) 39 (43 3 %) 12 (38 7 %) 68 (39 1 %) Table 3. Materials and methods The size of the surgical resection margin depending on the thickness of the tumor according to Breslow (n = 174) Resection margin Thickness of the tumor according to Breslow ≤1 mm (n = 29) 1.01–2.00 mm (n = 19) 2.01–4.00 mm (n = 26) >4 mm (n = 99) in situ (n = 1) <0.5 cm 16 (55.0 %)* 9 (47.5 %)* 9 (34.6 %) 22 (22.2 %) 1 (100.0 %) 0.6–1.0 cm 11 (38.0 %) 6 (31.5 %) 11 (42.3 %) 36 (36.3 %) 0 % >1.0 cm 2 (7.0 %)* 4 (21.0 %) 6 (23.1 %) 41 (41.5 %) 0 % Table 3. The size of the surgical resection margin depending on the thickness of the tumor according to Breslow (n = 17 Table 3. The size of the surgical resection margin depending on the thickness of the tumor according to Breslow (n = 174) * Significant differences compared to a tumor with thickness more than 4 mm. Th ith i i th thi k f th t th i f th i l ti i d ll * Significant differences compared to a tumor with thickness more than 4 mm. Thus, with an increase in the thickness of the tumor, the size of the surgical resection margin gradually increased. Table 4. Patient distribution according to ulceration Ulceration Hairy part of the head (n = 53) Facial area (n = 90) Neck area (n = 31) Total (n = 174) No 17 (38.1 %) 39 (43.3 %) 12 (38.7 %) 68 (39.1 %) Yes 36 (67.9 %) 50 (55.6 %) 19 (61.3 %) 105 (60.3 %) In situ 0 % 1 (1.1 %) 0 % 1 (0.6 %) * Significant differences compared to a tumor with thickness more than 4 mm. Thus, with an increase in the thickness of the tumor, the size of the surgical resection margin gradually increa 59 59 Опухоли ГОЛОВЫ и ШЕИ HEAD and NECK tumors 3’2020 Оригинальное исследование Том 10 Vol. 10 Опухоли ГОЛОВЫ и ШЕИ Table 5. Materials and methods The amount of surgical resection margin depending on the stage of cutaneous melanoma in the pN0 group (n = 131) Resection margin Amount of patients Stage of TNM / AJCC 0 – IA / B IIA IIB IIC ≤0.5 cm 48 20 (41.6 %) 7 (14.6 %) 13 (27.1 %) 8 (16.7 %) 0.5–1.0 cm 51 11 (21.6 %)* 7 (13.7 %) 13 (25.5 %) 20 (39.2 %)* >1.0 cm 32 3 (9.4 %)* 3 (9.4 %) 9 (28.1 %) 17 (53.1 %)* Total 131 34 (25.9 %) 17 (13.0 %) 35 (26.7 %) 45 (34.4 %) Credible differences compared to resection margin of ≤0.5 cm, р <0.05. Fig. 2. ROC-curve for the thickness of tumor: a – less 2 mm; b – 2.01–4.0 mm; c – more than 4 mm Sensitivity Sensitivity Sensitivity 0 0.2 0.4 0.6 0.8 1.0 Specyfity 0 0.2 0.4 0.6 0.8 1.0 Specyfity 0 0.2 0.4 0.6 0.8 1.0 Specyfity 1.0 0.8 0.6 0.4 0.2 0 1.0 0.8 0.6 0.4 0.2 0 1.0 0.8 0.6 0.4 0.2 0 а b c Table 5. The amount of surgical resection margin depending on the stage of cutaneous melanoma in the pN0 group (n = 131) Resection margin Amount of patients Stage of TNM / AJCC 0 – IA / B IIA IIB IIC ≤0.5 cm 48 20 (41.6 %) 7 (14.6 %) 13 (27.1 %) 8 (16.7 %) 0.5–1.0 cm 51 11 (21.6 %) 7 (13.7 %) 13 (25.5 %) 20 (39.2 %)* >1.0 cm 32 3 (9.4 %)* 3 (9.4 %) 9 (28.1 %) 17 (53.1 %)* Total 131 34 (25.9 %) 17 (13.0 %) 35 (26.7 %) 45 (34.4 %) Credible differences compared to resection margin of ≤0.5 cm, р <0.05. Fig. 2. ROC-curve for the thickness of tumor: a – less 2 mm; b – 2.01–4.0 mm; c – more than 4 mm Sensitivity Sensitivity Sensitivity 0 0.2 0.4 0.6 0.8 1.0 Specyfity 0 0.2 0.4 0.6 0.8 1.0 Specyfity 0 0.2 0.4 0.6 0.8 1.0 Specyfity 1.0 0.8 0.6 0.4 0.2 0 1.0 0.8 0.6 0.4 0.2 0 1.0 0.8 0.6 0.4 0.2 0 а b c able 5. The amount of surgical resection margin depending on the stage of cutaneous melanoma in the pN0 group (n = 131) Credible differences compared to resection margin of ≤0.5 cm, р <0.05. Materials and methods Sensitivity 0 0.2 0.4 0.6 0.8 1.0 Specyfity 1.0 0.8 0.6 0.4 0.2 0 b Sensitivity 0 0.2 0.4 0.6 0.8 1.0 Specyfity 1.0 0.8 0.6 0.4 0.2 0 c Sensitivity 0 0.2 0.4 0.6 0.8 1.0 Specyfity 1.0 0.8 0.6 0.4 0.2 0 а Fig. 2. ROC-curve for the thickness of tumor: a – less 2 mm; b – 2.01–4.0 mm; c – more than 4 mm The information from the ROC-curves was as follows. For thickness less than 2 mm, the critical point was the re- section margin value of 0.46 cm, which turned out to be statistically insignificant, the sensitivity of this one was 67 % with a specificity of 61 % (fig. 2a). For the thickness of the primary tumor 2.01–4.00 mm (fig. 2b) and more than 4 mm (fig. 2c), the critical points became 0.58 cm (sensitivity 93 % with a specificity of 58 %) and 0.72 cm (sensitivity 68 % at specificity of 60 %), these indicators were significant. Results Results While analyzing the followed-up treatment results in the general group of patients (n = 174), progression at different timescales after the end of treatment occurred in 51 % of the cases (fig. 3). And this progression is most often Results While analyzing the followed-up treatment results in the general group of patients (n = 174), progression at different timescales after the end of treatment occurred in 51 % of the cases (fig. 3). And this progression is most often While analyzing the followed-up treatment results in the general group of patients (n = 174), progression at different timescales after the end of treatment occurred in 51 % of the cases (fig. 3). And this progression is most often Fig. 3. Followed-up treatment results in the general group of patients (n = 174) n = 88 n = 3 n = 37 n = 53 50.6 21.3 30.5 1.7 60 – 50 – 40 – 30 – 20 – 10 – 0 The progression rate with a tumor thickness less than 2 mm with a surgical resection margin of up to 0.46 cm was 16.0 % (n = 4), and with more than 0.46 cm it was 34.8 % (n = 8) (p = 0.13). With a melanoma thickness of 2.01– 4.0 mm with a surgical resection margin of up to 0.58 cm, progression occurred in 6.7 % (n = 1) with more than 0.58 cm – 56.5 % (n = 13) (p = 0.002). And finally, with a tumor thickness of more than 4 mm with a surgical margin of up to 0.72 cm, progression was noted in 57.1 % (n = 20), and with an surgical margin of more than 0.72 cm – 80.8 % (n = 42), p = 0.016. Thus, it is seen that the increase of a surgical resection margin does not improve followed-up treatment results. An increase in significance was noted with an increase in the thickness of primary melanoma. This suggests that the in- fluence of surgical resection margin is most important for cutaneous melanoma of >2 mm. Fig. 3. Followed-up treatment results in the general group of patients (n = 174) 60 60 Опухоли ГОЛОВЫ и ШЕИ HEAD and NECK tumors 3’2020 Оригинальное исследование Том 10 Vol. 10 period from 7.3 to 190.4 months (average period of moni- toring constituted 65.1 ± 45.2 months, median 59.3 months). Results manifested in the form of implementation of distant (30.5 %) and regional (21 %) metastasis. And only in 2 % (n = 3) of cases the occurrence of local recurrence was re- corded. Recurrence-free survival did not correlate with the size of the surgical resection margin. It was the best with a minimal resection margin of 77.3 ± 6.4 % (median not achieved) and the worst with a maximum surgical margin of 38.7 ± 8.8 % (median of 20 months) (fig. 4), that means, that followed-up treatment results did not depend on the increase in the width of the surgical resection margin. For further understanding of the influence of surgical resection margin on followed-up treatment results and for obtaining more objective data from the study group, patients with regional pN+ metastases were excluded; therefore, the study was performed in the group without regional pN0 metastases (n = 131). During the detailed analysis of the “N0” group, depend- ing on the size of the surgical resection margin, it turned out that with an increase of surgical margin, the frequency of progression of patients rapidly grew (fig. 5). So, with surgical margin of ≤0.5 cm, progression occurred in 22.9 %, 0.6–1.0 cm – 47 % (p <0.05) and with surgical margin more than >1 cm – 62.5 % (p <0.05). This happened mainly due to an increase in the frequency of regional and distant me- tastasis process. Regional metastases with a minimum in- dentation (≤0.5 cm) were realized in 10.4 %, while with a maximum surgical resection margin (>1 cm) this indicator was 37.5 %. A similar tendency is observed when assess- ing the frequency of distant metastasis process: ≤0.5 cm – 12.5 %, and at >1 cm – 25 %, respectively. It is important to note that local recurrence of cutaneous melanoma over the entire observation period developed in 2 (3.9 %) patients with a surgical indentation of 0.6–1.0 cm. Thus, it can be seen that the area of the surgical resection margin does not affect the frequency of local recurrence. Among the 131 of analyzed patients, progression occurred to 55 (42.0 %) patients in the period from 1 to 121 months (the average time of progression constituted 18.7 ± 23.6 months). The age of patients ranged from 20 to 89 years (mean age 53.3 ± 15.9 years, median 54 years). 47 (35.8 %) patients died from the underlying disease. Results Followed-up treatment results depending on the thickness of melanoma according to Breslow in the group of patients without regional metastases (pN0, n = 131) 35 30 25 20 15 10 5 0 % Local relapses Regional metastases Distant metastases 29.7 32.8 11.5 3.8 3.8 1.6 ≤1 mm (n = 26) >4 mm (n = 64) Distant metastases Fig. 6. Followed-up treatment results depending on the thickness of melanoma according to Breslow in the group of patients without regional metastases (pN0, n = 131) Fig. 7. Followed-up treatment results depending on the level of invasion according to Clark in the group of patients without regional metastases (pN0, n = 131) 40 30 20 10 0 % Local relapses Regional metastases Distant metastases 33.4 37.5 4.2 0 8.4 4.2 II level (n = 24) V level (n = 24) II level (n = 24) V level (n = 24) Distant metastases Regional metastases Fig. 7. Followed-up treatment results depending on the level of invasion according to Clark in the group of patients without regional metastases (pN0, n = 131) Fig. 7. Followed-up treatment results depending on the level of invasion according to Clark in the group of patients without r Fig. 8. Followed-up treatment results depending on the presence of ulceration in the group of patients without regional metastases (pN0, n = 131) 30 25 20 15 10 5 0 % Local relapses Regional metastases Distant metastases 12.3 28.4 19.3 23 1.8 1.4 Without ulceration (n = 57) With ulceration (n = 74) eatment results depending on the presence of ulceration in the group of patients without regional metastases (pN0, n = 131) Fig. 8. Followed-up treatment results depending on the presence of ulceration in the group of patients without regional meta There was still no effect of the invasion level according to Clark on the local recurrence rate. carcinelcosis of the primary tumor, it can be seen that wors- ening of overall survival occurred also due to an increase in the frequency of regional and distant metastasis (fig. 8). It should be noted that the frequency of regional metastases carcinelcosis of the primary tumor, it can be seen that wors- ening of overall survival occurred also due to an increase in the frequency of regional and distant metastasis (fig. 8). Results They were tracked in the Fig. 4. Recurrence-free survival curve depending on the size of the surgical resection margin in the group of patients without regional metastases (pN0, n = 131, р = 0.0022) 0 20 40 60 80 100 120 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 up to 0.5 cm 0.6–1 cm >1 Cumulative Proportion Surviving Time Cumulative Proportion Surviving (Kaplan–Meier) Progression No progression Cumulative Proportion Surviving (Kaplan–Meier) Cumulative Proportion Surviving (Kaplan–Meier) Progression No progression The frequency of regional and distant metastases in the group of patients with a tumor thickness of more than 4 mm, increased by more than 3 times compared to the group of pa- tients with up to 1 mm one (fig. 6). Moreover, the frequen- cy of local recurrence did not depend on the thickness of the primary tumor according to Breslow. In the group of patients with V-level of invasion, the fre- quency of regional and distant metastasis also sharply in- creased compared to the group of patients with II level (fig. 7). Fig. 4. Recurrence-free survival curve depending on the size of the surgical resection margin in the group of patients without regional metastases (pN0, n = 131, р = 0.0022) Fig. 5. Followed-up treatment results depending on the size of the surgical resection margin in the group of patients without regional metastases (pN0, n = 131) 40 30 20 10 0 % Local relapses Regional metastases Distant metastases 0 3.9 21.6 10.4 37.5 12.5 27.5 25 ≤0.5 cm 0.6–1 cm >1 cm 0 37.5 Fig. 5. Followed-up treatment results depending on the size of the surgical resection margin in the group of patients without regional metastases (pN0, n = 131) Fig. 5. Followed-up treatment results depending on the size of the surgical resection margin in the group of patients without r 61 61 Опухоли ГОЛОВЫ и ШЕИ HEAD and NECK tumors 3’2020 Оригинальное исследование Том 10 Vol. 10 Fig. 6. Followed-up treatment results depending on the thickness of melanoma according to Breslow in the group of patients without regional metastases (pN0, n = 131) 35 30 25 20 15 10 5 0 % Local relapses Regional metastases Distant metastases 29.7 32.8 11.5 3.8 3.8 1.6 ≤1 mm (n = 26) >4 mm (n = 64) Fig. 6. R E F E R E N C E S / Л И Т Е Р А Т У Р А 1. Аллахвердян Г.С., Чекалова М.А., Ко- косадзе Н.В. Дооперационная оценка местного распространения первичной меланомы кожи ультразвуковым мето- дом. Ультразвуковая и функциональная диагностика 2007;(4):238–9. [Allakhverdyan G.S., Chekalova M.A., Kokosadze N.V. Preoperative assessment of the local spread of primary cutaneous melanoma using ultrasound. Ultra zvukovaya i funktsionalnaya diagnostika = Ultrasound and functional diagnostics 2007;(4):238–9. (In Russ.)]. J Am Acad Dermatol 1987;17(6):1050–3. DOI: 10.1016/s0190-9622(87)80487-5. J Am Acad Dermatol 1987;17(6):1050–3. DOI: 10.1016/s0190-9622(87)80487-5. analysis and results of surgical treatment. Ann Surg 1984;200(6):769–75. DOI: 10.1097/00000658-198412000-00017 7. Малишевская Н.П., Соколова А.В., Демидов Л.В. Современное состояние заболеваемости меланомой кожи. Медицинский совет 2018;(10):161–5. [Malishevskaya N.P., Sokolova A.B., Demidov L.V. The incidence of skin melanoma in the russian federation and federal districts. Meditsinsky sovet = Medi cal Сouncil 2018;(10):161–5. (In Russ.)]. 14. Urist M.M., Maddox W.A., Kennedy J., Balch C.M. Patient risk factors and surgical morbidity after regional lymphadenectomy in 204 melanoma patients. Cancer 1983;51(11):2l52–6. 15. Анисимов В.В. Содержание понятия «местный рецидив» после хирургиче- ского лечения злокачественной мела- номы кожи. Вопросы онкологии 1985;31(1):32–7. [Anisimov V.V. Content of the concept of “local recurrence” after surgical treatment of cutaneous malignant melanoma. Voprosy onkologii = Problems in Oncology 1985;31(1):32–7. (In Russ.)]. cal Сouncil 2018;(10):161–5. (In Russ.)]. 8. Balch C.M., Hjugton A.N., Peters L.I. Cutaneous melanoma. In: Principles and practice of oncology. Ed. by V.I. De Vita, S. Hellman, S.A. Rosenberg. 4th edn. Philadelphia: Lippencott Co., 1993. P. 1612–1661. 2. Balch C.M., Soong S.J., Murad T.M. et al. A multi factorial analysis of melanoma. II. Prognostic factors in patients with stage I (localized) melanoma. Surgery 1979;86(2):343–51. 9. Elder D.E., Guerry D.V., Heiberger R.M. et al. Optimal resection margin for cu taneous malignant melanoma. Plast Reconstr Surg 1983;71(1):66–72. DOI: 10.1097/00006534-198301000-00015. 16. Трапезников Н.Н., Рабен А.С., Яворский В.В., Титинер Г.Б. Пигмент- ные невусы и новообразования кожи. М.: Медицина, 1976. С. 177. [Trapeznikov N.N., Raben A.S., Yavorski V.V., Titiner G.B. Pigmented nevi and skin neoplasms. Moscow: Meditsina, 1976. P. 177. (In Russ.)]. 3. Пачес А.И. Опухоли головы и шеи: Клиническое руководство. М.: Практи- ческая медицина, 2013. С. 60–69. [Paches A.I. Head and neck tumors: Clinical guideline. Moscow: Praktiches kaya meditsina, 2013. P. 60–69. (In Russ.)]. 10. Bodurtha A.J. Spontaneous regression of malignant melanoma. In: Human malignant melanoma. Ed. by W.H. Clark, L.J. Goldmarm, M.J. Mastrangelo. New York; San Francisco; London: Game & Stratton, 1979. P. 227–241. 4. Friedman R.J., Rigel S., Kopf A.W. Results It should be noted that the frequency of regional metastases During the analysis of followed-up treatment results, depending on such an important prognostic factor as 62 62 Опухоли ГОЛОВЫ и ШЕИ HEAD and NECK tumors 3’2020 Оригинальное исследование Том 10 Vol. 10 in the group with carcinelcosis of the primary tumor in- creased by more than 50 %. At the same time, no influence of this factor on the frequency of local recurrence was noted. of regional and distant metastases, which fits into specific- ity of tumor biology. The local recurrence can be affected not only by surgical resection margin and pathomorpholog- ical prognostic factors such as tumor thickness according to Breslow, level of invasion according to Clark and the pres- ence of ulceration, as well as the biology of the tumor itself, which makes the question of further study, within the frame- work of a prospective multicenter study, incredibly relevant. Discussion The current study shows that cutaneous melanoma of head and neck is not characterized as a disease of local recurrences but manifests itself in the form of high frequen- cy regional and distant metastases, which fits into the nature of tumor biology. Local recurrences can be affected not only by surgical resection margin and pathomorphological prognostic factors such as tumor thickness according to Breslow, level of invasion according to Clark, ulceration, as well as the tumor biology, which makes the further studies within the prospective multicenter research incredibly rele vant. Conclusions We consider that clear surgical margins for any thickness of cutaneous melanoma of head and neck should be as fol- lows: <2 mm – 0.46 mm (p = 0.13), 2.01–4.00 mm – 0.58 mm (p = 0.002), >4 mm – 0.72 mm (p = 0.016). In our work, the influence of the main prognostic fac- tors, such as tumor thickness according to Breslow, level of invasion according to Clark and ulceration on the fre- quency of head and neck cutaneous melanoma local recur- rences had no impact. The present study demonstrates that cutaneous mela- noma of head and neck is not a disease of local recurrences, but it manifests itself in the form of a high frequency Опухоли ГОЛОВЫ и ШЕИ HEAD and NECK tumors 3’2020 Оригинальное исследование Том 10 Vol. 10 melanoma: the WHO Melanoma Program experience. Ann Surg Oncol 2000;7(6):469–74. DOI: 10.1007/s10434-000-0469-z. margin of 2 cm versus 4 cm for cutaneous malignant melanoma with a tumor thickness of more than 2 mm. In: World Congress on Melanoma. Vancouver, 2005. margin of 2 cm versus 4 cm for cutaneous malignant melanoma with a tumor thickness of more than 2 mm. In: World Congress on Melanoma. Vancouver, 2005. satellites” in the reticular dermis and subcutaneous fat. Ann Surg 1981;194(1):108–12. DOI: 10.1097/ 00000658-198107000-00019. ; ( ) DOI: 10.1007/s10434-000-0469-z. 34. Wheatley K., Wilson J.S., Gaunt P., Marsden J.R. Surgical excision margins in primary cutaneous melanoma: a meta- analysis and Bayesian probability eva luation. Cancer Treat Rev 2016;42:73–81. DOI: 10.1016/j.ctrv.2015.10.013. 21. Levine A. On the histological diagnosis and prognosis of malignant melanoma. J Clin Pathol 1980;33(2):101–24. DOI: 10.1136/jcp.33.2.101. 27. Veronesi U., Cascinelli N. Narrow excision (1-cm margin). A safe procedure for thin cutaneous melanoma. Arch Surg 1991;126(4):438–41. DOI: 10.1001/ archsurg.1991.01410280036004. 22. Banzet P., Thomas A., Vuillemin E. Wide versus narrow surgical excision in thin (<2 mm) stage I primary cutaneous melanoma: long term results of a French multicentric prospective randomized trial on 319 patients. Proc Am Assoc Clin Oncol 1993;12:387. 28. Balch C.M., Urist M.M., Karakousis C.P. et al. Efficacy of 2-cm surgical margins for intermediate-thickness melanomas (1 to 4 mm). Results of a multi- institutional randomized surgical trial. 35. Breslow A. Thickness, cross-sectional areas and depth of invasion in prognosis of cutaneous melanoma. Ann Surg 1970;172:902–8. 29. Ann Surg 1993;218(3):262–7. DOI: 10.1097/00000658-199309000-00005. 36. Balch C.M., Houghton A.N., Sober A.J. et al. Thompson complex closures of melanoma excisions. In: Cutaneous melanoma. 5th edn. St. Louis: Quality Medical Publishing, 2009. P. 275–319. 23. Khayat D., Rixe O., Martin G. et al. Surgical margins in cutaneous melanoma (2 cm versus 5 cm for lesions measuring less than 2.1-mm thick). Cancer 2003; 97(8):1941–6. DOI: 10.1002/cncr.11272. 30. Balch C.M., Soong S.J., Smith T. et al. Long-term results of a prospective surgical trial comparing 2 cm vs. 4 cm excision margins for 740 patients with 1–4 mm me lanomas. Ann Surg Oncol 2001;8(2):101–8. DOI: 10.1007/s10434-001-0101-x. 37. Барчук Л.С. Хирургическое лечение меланом. Практическая онкология 2001;(34):30–6. [Barchuk L.S. Surgical treatment of melanoma. Prakticheskaya onkologiya = Practical Oncology 2001;(34):30–6. (In Russ.)]. 24. Cohn-Cedermark G., Rutqvist L.E., Ander- sson R. et al. Опухоли ГОЛОВЫ и ШЕИ HEAD and NECK tumors 3’2020 Оригинальное исследование Long term results of a ran domized study by the Swedish Melanoma Study Group on 2-cm versus 5-cm resec tion margins for patients with cutaneous melanoma with a tumor thickness of 0.8– 2.0 mm. Cancer 2000;89(7):1495–501. 31. Karakousis C.P., Balch C.M., Urist M.M. et al. Local recurrence in malignant melanoma: long-term results of the multiinstitutional randomized surgical trial. Ann Surg Oncol 1996;3(5):446–52. DOI: 10.1007/BF02305762. 38. Bagley F.H., Cady B., Lee A. et al. Changes in clinical preservation and management of malignant melanoma. Cancer 1981;47:2126–34. 25. Ringborg U., Andersson R., Eldh J. et al. Resection margins of 2 versus 5 cm for cu taneous malignant melanoma with a tumor thickness of 0.8 to 2.0 mm: randomized study by the Swedish Melanoma Study Group. Cancer 1996;77(9):1809–14. 32. Thomas J.M., Newton-Bishop J., A’Hern R. et al. Excision margins in high- risk malignant melanoma. N Engl J Med 2004;350(8):757–66. DOI: 10.1056/NEJMoa030681. 39. Gershenwald J.E., Scolyer R.A., Hess K.R. et al. Melanoma staging: Evidence-based changes in the American Joint Committee on Cancer eighth edition cancer staging manual. CA Cancer J Clin 2017;67(6):472–92. DOI: 10.3322/caac.21409. 26. Cascinelli N., Belli F., Santinami M. et al. Sentinel lymph node biopsy in cutaneous 33. Ringborg U., Brahme E.M., Drewiecki K. Randomized trial of a resection 33. Ringborg U., Brahme E.M., Drewiecki K. Randomized trial of a resection Authors’ contributions A.M. Mudunov: developing the research design, surgical treatment, scientific editing, scientific advice, article writing; M B P k ti l iti i i f bli ti th ti l ' th l i f th bt i d d t i l t t t i t A.M. Mudunov: developing the research design, surgical treatment, scientific editing, scientific advice, article writing M.B. Pak: article writing, reviewing of publications on the article's theme, analysis of the obtained data, surgical treat A.M. Mudunov: developing the research design, surgical treatment, scientific editing, scientific advice, article writing; g, g p , y L.V. Demidov: developing the research design, scientific editing, scientific advice; L.V. Demidov: developing the research design, scientific editing, scientific advice; K.A. Baryshnikov: surgical treatment, assistance, patient supervision. Вклад авторов Вклад авторов A.M. Мудунов: разработка дизайна исследования, проведение операций, научное редактирование, научное консультирование, написание текста статьи; ов: разработка дизайна исследования, проведение операций, научное редактирование, научное консультирова д р A.M. Мудунов: разработка дизайна исследования, проведение операций, научное редактирование, научное консультирование, написание текста статьи; р A.M. Мудунов: разработка дизайна исследования, проведение операций, научное редактирование, научное текста статьи; M.Б. Пак: написание текста статьи, обзор публикаций по теме статьи, анализ полученных данных, проведение операций, ассистирование; Л.В. Демидов: разработка дизайна исследования, научное редактирование, научное консультирование; M.Б. Пак: написание текста статьи, обзор публикаций по теме статьи, анализ полученных данных, проведение операций, ассистирование; Л.В. Демидов: разработка дизайна исследования, научное редактирование, научное консультирование; M.Б. Пак: написание текста статьи, обзор публикаций по теме статьи, анализ полученных данных, проведение операций, ассистирование; Л.В. Демидов: разработка дизайна исследования, научное редактирование, научное консультирование; Л.В. Демидов: разработка дизайна исследования, научное редактирова К.A. Барышников: проведение операций, ассистирование, курирование пациентов. ORCID of authors / ORCID авторов A.M. Mudunov / A.M. Мудунов: https://orcid.org/0000-0002-0918-3857 A.M. Mudunov / A.M. Мудунов: https://orcid.org/0000-0002-0918-3857 M.B. Pak / М.Б. Пак: https://orcid.org/0000-0003-4546-0011 M.B. Pak / М.Б. Пак: https://orcid.org/0000-0003-4546-0011 L.V. Demidov / Л.В. Демидов: https://orcid.org/0000-0002-8562-6082 K.A. Baryshnikov / К.А. Барышников: https://orcid.org/0000-0002-9532-4264 K.A. Baryshnikov / К.А. Барышников: https://orcid.org/0000-0002-9532-4264 Conflict of interest. The authors declare no conflict of interest. Конфликт интересов. Авторы заявляют об отсутствии конфликта интересов. Conflict of interest. The authors declare no conflict of interest. Конфликт интересов. Авторы заявляют об отсутствии конфликта интересов. Конфликт интересов. Авторы заявляют об отсутствии конфликта интересов. Financing. The work was performed without external funding. Financing. The work was performed without external funding. Финансирование. Исследование выполнено без спонсорской поддержки. Финансирование. Исследование выполнено без спонсорской поддержки. R E F E R E N C E S / Л И Т Е Р А Т У Р А Early detection of malignant melanoma: the role of physician examination and self-exa mination of the skin. CA Cancer J Clin 1985;35(3):130–51. DOI: 10.3322/canjclin.35.3.130. 17. Beardmore G.L., Davis N.C. Multiple primary cutaneous melanomas. Arch Dermatol 1975;11:603–9. 11. Spellman J.E. Jr, Driscoll D., Velez A., Karakousis C. Thick cutaneuos melanoma of the trunk and extremities: an institu tional review. Surg Oncol 1994;3(6):335–43. DOI: 10.1016/0960-7404(94)90072-8. 18. Cascinelli N., Bajetta E., Vaglini M. et al. Present status and future perspectives of adjuvant treatment of cutaneous malignant melanoma. Pigment Cell Res 1983;6:187–98. 5. Friedman R.J., Rigel D.S., Silver- man M.K. et. al. Malignant melanoma in the 1990’s: The continued importance of early detection and the role of physician examination and self-examination of the skin. CA Cancer J Clin 1991;41(4):201–26. DOI: 10.3322/canjclin.41.4.201. 19. Stage I melanoma of the skin: the problem of resection margins. W.H.O. Collaborating Centres for Evaluation of Methods of Diagnosis and Treatment of Melanoma. Eur J Cancer 1980;16(8):1079–85. 12. Taylor B.A., Hughes L.E. A policy of selective excision for primary cutaneous malignant melanoma. Eur J Surg Oncol 1985;11(1):7–13. 6. Rigel D.S., Kopf A.W., Friedman R.J. The rate of malignant melanoma in the United States: are we making an impact? 13. Urist M.M., Balch C.M., Soong S.J. et al. Head and neck melanoma in 534 clinical stage I patients. A prognostic factors 20. Day C.L., Harrist T.J., Gorstein F. et al. Prognostic significance of “microscopic 63 63 Опухоли ГОЛОВЫ и ШЕИ HEAD and NECK tumors 3’2020 Оригинальное исследование Compliance with patient rights and principles of bioethics tients gave written informed consent to participate in the stu облюдение прав пациентов и правил биоэтики Соблюдение прав пациентов и правил биоэтики Протокол исследования одобрен комитетом по биомедицинской этике Первого Московского государственного медицинского универси- тета им. И.М. Сеченова. Все пациенты подписали информированное согласие на участие в исследовании. Все пациенты подписали информированное согласие на участие в исследовании. Статья поступила: 22.07.2020. Принята к публикации: 15.09.2020. Article submitted: 22.07.2020. Accepted for publication: 15.09.2020. Статья поступила: 22.07.2020. Принята к публикации: 15.09.2020. Article submitted: 22.07.2020. Accepted for publication: 15.09.2020. Статья поступила: 22.07.2020. Принята к публикации: 15.09.2020. Article submitted: 22.07.2020. Accepted for publication: 15.09.2020. Статья поступила: 22.07.2020. Принята к публикации: 15.09.2020. Article submitted: 22.07.2020. Accepted for publication: 15.09.2020. 64 64
https://openalex.org/W2800490336	https://dash.harvard.edu/bitstream/1/37160128/1/5923014.pdf	English	null	Migalastat improves diarrhea in patients with Fabry disease: clinical-biomarker correlations from the phase 3 FACETS trial	Orphanet journal of rare diseases	2,018	cc-by	5,866	Permanent link http://nrs.harvard.edu/urn-3:HUL.InstRepos:37160128 Terms of Use This article was downloaded from Harvard University’s DASH repository, and is made available under the terms and conditions applicable to Other Posted Material, as set forth at http:// nrs.harvard.edu/urn-3:HUL.InstRepos:dash.current.terms-of-use#LAA Published Version doi:10.1186/s13023-018-0813-7 Published Version doi:10.1186/s13023-018-0813-7 Citation Schiffmann, R., D. G. Bichet, A. Jovanovic, D. A. Hughes, R. Giugliani, U. Feldt-Rasmussen, S. P. Shankar, et al. 2018. “Migalastat improves diarrhea in patients with Fabry disease: clinical- biomarker correlations from the phase 3 FACETS trial.” Orphanet Journal of Rare Diseases 13 (1): 68. doi:10.1186/s13023-018-0813-7. http://dx.doi.org/10.1186/s13023-018-0813-7. © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Open Access RESEARCH Open Access Migalastat improves diarrhea in patients with Fabry disease: clinical-biomarker correlations from the phase 3 FACETS trial Raphael Schiffmann1,14, Daniel G. Bichet2, Ana Jovanovic3, Derralynn A. Hughes4, Roberto Giugliani5, Ulla Feldt-Rasmussen6, Suma P. Shankar7,13, Laura Barisoni8, Robert B. Colvin9, J. Charles Jennette10, Fred Holdbrook11, Andrew Mulberg11, Jeffrey P. Castelli11, Nina Skuban11, Jay A. Barth11 and Kathleen Nicholls12 Share Your Story The Harvard community has made this article openly available. Please share how this access benefits you. Submit a story . Accessibility Schiffmann et al. Orphanet Journal of Rare Diseases (2018) 13:68 https://doi.org/10.1186/s13023-018-0813-7 Abstract Background: Fabry disease is frequently characterized by gastrointestinal symptoms, including diarrhea. Migalastat is an orally-administered small molecule approved to treat the symptoms of Fabry disease in patients with amenable mutations. Methods: We evaluated minimal clinically important differences (MCID) in diarrhea based on the corresponding domain of the patient-reported Gastrointestinal Symptom Rating Scale (GSRS) in patients with Fabry disease and amenable mutations (N = 50) treated with migalastat 150 mg every other day or placebo during the phase 3 FACETS trial (NCT00925301). Results: After 6 months, significantly more patients receiving migalastat versus placebo experienced improvement in diarrhea based on a MCID of 0.33 (43% vs 11%; p = .02), including the subset with baseline diarrhea (71% vs 20%; p = .02). A decline in kidney peritubular capillary globotriaosylceramide inclusions correlated with diarrhea improvement; patients with a reduction > 0.1 were 5.6 times more likely to have an improvement in diarrhea than those without (p = .031). Conclusions: Migalastat was associated with a clinically meaningful improvement in diarrhea in patients with Fabry disease and amenable mutations. Reductions in kidney globotriaosylceramide may be a useful surrogate endpoint to predict clinical benefit with migalastat in patients with Fabry disease. Trial registration: NCT00925301; June 19, 2009. Trial registration: NCT00925301; June 19, 2009. Trial registration: NCT00925301; June 19, 2009. Keywords: Amenable mutation, Diarrhea, Fabry disease, Gastrointestinal, Globotriaosylceramide, GSRS, Lyso-Gb3, Migalastat, Pharmacological chaperone Migalastat improves diarrhea in patients with Fabry disease: clinical-biomarker correlations from the phase 3 FACETS trial Raphael Schiffmann1,14, Daniel G. Bichet2, Ana Jovanovic3, Derralynn A. Hughes4, Roberto Giugliani5, Ulla Feldt-Rasmussen6, Suma P. Shankar7,13, Laura Barisoni8, Robert B. Colvin9, J. Charles Jennette10, Fred Holdbrook11, Andrew Mulberg11, Jeffrey P. Castelli11, Nina Skuban11, Jay A. Barth11 and Kathleen Nicholls12 RESEARCH Open Access Background debilitating signs, symptoms, and life-limiting sequelae of Fabry disease [3]. Intrafamilial phenotypic variability is common in Fabry disease [4] and other genetic disorders, such as muscular dystrophy [5], making it difficult to pro- vide an accurate prognosis to patients based only on family history. Levels of disease substrate have been used as biomarkers in various clinical studies in Fabry disease [6, 7]; however, the correlation of changes in these bio- markers with clinical variables remains limited. Fabry disease is a rare, progressive, life-threatening X- linked lysosomal storage disorder, affecting males and fe- males, with an estimated prevalence of 1:117,000 to 1: 40,000 [1, 2]. Mutations in the GLA gene can lead to a deficiency of the lysosomal enzyme α-galactosidase A, which in turn results in an accumulation of glycosphingo- lipids, including globotriaosylceramide (GL-3) and plasma globotriaosylsphingosine (lyso-Gb3), and subsequently the Gastrointestinal signs and symptoms are a prominent and clinically important manifestation of Fabry disease and are reported by at least half of patients [8, 9]. Com- mon gastrointestinal signs and symptoms associated * Correspondence: raphael.schiffmann@BSWhealth.org 1Baylor Scott & White Research Institute, Dallas, TX, USA 14Institute of Metabolic Disease, 3812 Elm Street, Dallas, TX 75226, USA Full list of author information is available at the end of the article * Correspondence: raphael.schiffmann@BSWhealth.org 1Baylor Scott & White Research Institute, Dallas, TX, USA 14Institute of Metabolic Disease, 3812 Elm Street, Dallas, TX 75226, USA Full list of author information is available at the end of the article Schiffmann et al. Orphanet Journal of Rare Diseases (2018) 13:68 Page 2 of 7 in patients with GLA mutations amenable to migalastat (N = 50) [13]. with Fabry disease include diarrhea, nausea, vomiting, abdominal pain, and constipation [10, 11]. Gastrointes- tinal manifestations of Fabry disease are reported from an early age, and often have profound negative effects on social and economic functioning and quality of life in patients [11, 12]. Statistical analyses Statistical analyses The mean change in GSRS scores from baseline to month 6 was a pre-specified endpoint in the FACETS study. Change from baseline was presented descriptively; statistical tests of significance were performed using an ANCOVA model that included treatment, baseline, and treatment-by-baseline interaction. The p-value was calcu- lated based on the comparison of the least squares means. A response in the GSRS-D was defined as a reduction of 0.33 from baseline (i.e., MCID). The MCID was based on estimates in the literature for several non-Fabry gastrointestinal disorders in which diarrhea is a promin- ent symptom, and is consistent with an estimate of MCID based on data in Fabry patients from the FACETS study. Specifically, the MCID of 0.33 was derived from The mean change in GSRS scores from baseline to month 6 was a pre-specified endpoint in the FACETS study. Change from baseline was presented descriptively; statistical tests of significance were performed using an ANCOVA model that included treatment, baseline, and treatment-by-baseline interaction. The p-value was calcu- lated based on the comparison of the least squares means. GL-3 levels in kidney peritubular capillaries y p p Detailed methodology and results of the qualitative assess- ments of kidney biopsies have been reported [15, 19]. Briefly, kidney biopsies were performed at baseline and at months 6 and 12; these were assessed by 3 independent pathologists using whole slide images at 100× magnifica- tion in at least 300 peritubular capillaries in each biopsy to quantify the average number of GL-3 inclusions per PTC. Response to treatment was defined as a reduction of > 0.1 inclusions per capillary (which is above the level of background staining). Plasma lyso-Gb3 Plasma lyso-Gb3 levels were assessed at baseline and at months 6 and 12, and analyzed by means of liquid chromatography-mass spectroscopy [15]. The liquid chromatography-mass spectroscopy plasma lyso-Gb3 method used a novel stable isotope-labeled internal standard, 13C6-lyso-Gb3 (lower-limit-of-quantification: 0. 200 ng/mL, 0.254 nmol/L) [20, 21]. Response to treatment was defined as any reduction from baseline. We report here the results of further analyses using minimal clinically important difference (MCID) to evalu- ate improvements in diarrhea using the patient-reported Gastrointestinal Symptom Rating Scale (GSRS), in patients with Fabry disease treated with migalastat in the FACETS study. We also examine whether reductions in kidney peritubular capillary (PTC) GL-3 or lyso-Gb3 can be used as a surrogate endpoint to predict clinical bene- fit with migalastat. Gastrointestinal symptoms rating scale The GSRS comprises 15 questions that assess the sever- ity of 5 domains: diarrhea (“GSRS-D”), abdominal pain, constipation, indigestion, and reflux. Each domain consists of 2–4 questions, each rated on a 7-point Likert scale (from 1—absence of burden to 7—very severe dis- comfort) [18]. The GSRS-D has 3 questions to assess diarrhea frequency, consistency, and urgency; scores were determined by calculating the mean of the items within this domain. Results were collected at baseline and months 6, 12, 18, and 24 for all patients with amen- able mutations, and for the subset of patients presenting with gastrointestinal signs and symptoms at baseline. Migalastat is a pharmacological chaperone designed to bind selectively and reversibly with high affinity to the active sites of certain mutant forms of α-galactosidase (amenable GLA mutations) [13, 14]. Chaperoning mutated α-galactosidase A to lysosomes may mimic natural en- zyme trafficking, which has been suggested to result in more consistent α-galactosidase A activity than current standard of care enzyme replacement therapy (ERT) [15]. y p py ( ) [ ] In the phase 3 FACETS trial, which included a 6- month placebo-controlled stage, treatment with migala- stat maintained stable renal function, reduced cardiac mass, and reduced the severity of gastrointestinal signs and symptoms (diarrhea, reflux, and indigestion do- mains) in patients with Fabry disease and amenable mutations [15]. In the phase 3, active-controlled AT- TRACT study, migalastat and ERT had similar effects on renal function in patients with Fabry disease and amen- able mutations, and cardiac mass decreased significantly with migalastat treatment (compared with no change with ERT); furthermore, migalastat was generally safe and well-tolerated [16]. These results led to the approval of migalastat in the European Union, Switzerland, Canada, Australia, Republic of Korea, Japan, and Israel for the treatment of Fabry disease in patients aged 16 years and older with amenable mutations [14, 17]. Study design and patients The FACETS trial (AT1001–011, NCT00925301) has been described previously [15]. In brief, the study consisted of a 6-month randomized, double-blind, placebo-controlled phase, followed by a 6-month open-label phase with cross-over of placebo-treated patients to migalastat, and a 12-month extension phase. Male and female patients aged 16 to 74 years with Fabry disease, who were naive to ERT or had not received ERT for at least 6 months before screening, were eligible for randomization [15]. The effect of migalastat on gastrointestinal symptoms was evaluated A response in the GSRS-D was defined as a reduction of 0.33 from baseline (i.e., MCID). The MCID was based on estimates in the literature for several non-Fabry gastrointestinal disorders in which diarrhea is a promin- ent symptom, and is consistent with an estimate of MCID based on data in Fabry patients from the FACETS study. Specifically, the MCID of 0.33 was derived from Schiffmann et al. Orphanet Journal of Rare Diseases (2018) 13:68 Page 3 of 7 improvement of 0.33 compared with 2/10 (20%) of placebo-treated patients (p = .02) (Fig. 1a, b). improvement of 0.33 compared with 2/10 (20%) of placebo-treated patients (p = .02) (Fig. 1a, b). anchor-based methodologies from liver transplant pa- tients with gastrointestinal symptoms (MCID = 0.33) [22], patients with autoimmune disease with and without gastrointestinal symptoms (MCID = 0.33) [23], and renal transplant patients with and without gastrointestinal symptoms (MCID = 0.40) [24]. A distribution-based esti- mate of MCID in Fabry disease was derived from the change from baseline data in the placebo arm of the FACETS study. Using this approach, an MCID of 0.35 was generated, based on half the standard deviation [22, 23], supporting an MCID for GSRS-D of 0.33 in Fabry pa- tients. A sensitivity analysis using a higher threshold of 0. 66 was also performed to confirm the results. GSRS-D and kidney GL-3 inclusions As previously reported, 6 months of migalastat treatment was associated with a significantly greater reduction in the mean number of GL-3 inclusions per PTC compared with placebo (−0.25 vs + 0.07; p = .008) [15]. An analysis conducted on a combined endpoint of mean change from baseline in PTC GL-3 inclusions and GSRS-D using Xu’s statistic demon- strated a significant treatment effect of migalastat ver- sus placebo (1-sided; p = .009). Pre-specified analyses were conducted for all patients with amenable mutations and, post hoc, for the subset of patients with amenable mutations who reported diar- rhea symptoms at baseline. Assessment of patient-level responses demonstrated a consistent beneficial effect of migalastat on PTC GL-3 inclusions and GSRS-D. A majority of patients with amenable mutations (15/18; 83%) treated with migalastat demonstrated a response in PTC GL-3 and/or GSRS- D when either or both of these endpoints were ele- vated at baseline, compared with 5/15 (33%) patients treated with placebo. Sensitivity analyses The results of the MCID analysis for observed im- provement in diarrhea were confirmed in a sensitivity analysis. Using an improvement threshold of 0.66, 9/ 28 (32%) migalastat-treated patients experienced a clinically relevant change compared with 1/19 (5%) placebo-treated patients (p = .03). In patients with diarrhea symptoms at baseline (baseline GSRS-D score of ≥1), 9/13 (69%) migalastat-treated patients experienced a clinically relevant change compared with 1/9 (11%) placebo-treated patients (p = .01) (Fig. 1c). The number of patients demonstrating a response in GSRS-D and/or PTC GL-3 from baseline to month 6 was compared between treatment groups. A retrospect- ive analysis using Xu’s statistic, a multivariate test used to assess if treatment has a beneficial effect on multiple outcomes simultaneously [25], evaluated whether treat- ment impacted both parameters as a combined endpoint. Logistic regression was used to assess the correlation be- tween changes in GSRS-D and PTC GL-3. A similar re- gression analysis was performed to assess the correlation between changes in GSRS-D and plasma lyso-Gb3. Summary of GSRS findings Of the 50 patients with Fabry disease and amenable mu- tations who were enrolled in the FACETS study, 28 (56%) reported diarrhea symptoms at baseline. A logistic regression modeling the improvement in GSRS-D (ie, change from baseline to Month 6 < −0. 33) as the dependent variable, and reduction in PTC GL-3 inclusions (ie, change from baseline <−0.1, ≥−0.1) and treatment group as independent variables, indicated that reductions in PTC GL-3 inclusions were strongly associated with improvement in diarrhea (Table 1). Patients who had a reduction of > 0.1 in PTC GL-3 inclusions (ie, change from baseline <−0.1) were 5.6 times more likely to have an improvement in diarrhea symptoms than patients who did not have a reduction (p = .031). As previously reported [15], in patients randomly assigned to migalastat (n = 28), symptoms of diarrhea, based on the GSRS-D, improved within the first 6 months of treatment (change from baseline, −0.3), whereas diar- rhea in the placebo-treated group (n = 22) worsened (change from baseline, + 0.2; p = .03). A numerically larger reduction in GSRS-D scores was also observed in the sub- group of patients who reported gastrointestinal symptoms at baseline (migalastat change from baseline, −0.6; pla- cebo change from baseline, + 0.2). These improvements continued through 24 months of treatment [15]. GSRS-D and plasma lyso-Gb3 missing data, the odds ratio was similar for PTC GL- 3 (odds ratio = 6.6; 95% confidence interval, 1.3–33.0; p = .02) and plasma lyso-Gb3 (odds ratio = 6.2; 95% confidence interval, 0.6–64.3; p = .12); however, statis- tical significance was achieved only for PTC GL-3, possibly due to the smaller sample size for plasma lyso-Gb3 (PTC GL-3, n = 44; plasma lyso-Gb3, n = 31). missing data, the odds ratio was similar for PTC GL- 3 (odds ratio = 6.6; 95% confidence interval, 1.3–33.0; p = .02) and plasma lyso-Gb3 (odds ratio = 6.2; 95% confidence interval, 0.6–64.3; p = .12); however, statis- tical significance was achieved only for PTC GL-3, possibly due to the smaller sample size for plasma lyso-Gb3 (PTC GL-3, n = 44; plasma lyso-Gb3, n = 31). GSRS-D and plasma lyso-Gb3 y p y After 6 months of treatment, 12/28 (43%) migalastat- treated patients experienced a GSRS-D score improve- ment of 0.33 (i.e., MCID) compared with 2/19 (11%) pa- tients receiving placebo (p = .02) (Fig. 1a, b). In the subset of patients with diarrhea symptoms at baseline (baseline GSRS-D scores ≥1), 12/17 (71%) of the migalastat-treated patients experienced a clinically relevant When the same correlation analysis was performed for changes in plasma lyso-Gb3, defined as any reduc- tion from baseline, and a response in GSRS-D, de- fined as any reduction ≥0.33, the linear correlation was not statistically significant (odds ratio = 2.5; 95% confidence interval 0.63–9.6; p = .2). When the logistic regression was conducted excluding patients with Schiffmann et al. Orphanet Journal of Rare Diseases (2018) 13:68 Page 4 of 7 missing data, the odds ratio was similar for PTC GL- 3 (odds ratio = 6.6; 95% confidence interval, 1.3–33.0; p = .02) and plasma lyso-Gb3 (odds ratio = 6.2; 95% fid i l 0 6 64 3 12) h i Discussion In this investigation, migalastat was associated with a clinically relevant improvement in diarrhea symptoms b d h GSRS D i i i h F b di a b c Fig. 1 Patients experiencing a minimal clinically important difference in GSRS-D scores after 6 months of treatment. a Improvement of 0.33. b Forest plot showing the effect size of migalastat treatment vs placebo. c Sensitivity analysis: improvement of 0.66 a b a b c Fig. 1 Patients experiencing a minimal clinically important difference in GSRS-D scores after 6 months of treatment. a Improvement of 0.33. b Forest l h i h ff i f i l l b S i i i l i i f 066 c Fig. 1 Patients experiencing a minimal clinically important difference in GSRS-D scores after 6 months of treatment. a Improvement of 0.33. b Forest plot showing the effect size of migalastat treatment vs placebo. c Sensitivity analysis: improvement of 0.66 missing data, the odds ratio was similar for PTC GL- 3 (odds ratio = 6.6; 95% confidence interval, 1.3–33.0; p = .02) and plasma lyso-Gb3 (odds ratio = 6.2; 95% confidence interval, 0.6–64.3; p = .12); however, statis- tical significance was achieved only for PTC GL-3, possibly due to the smaller sample size for plasma lyso-Gb3 (PTC GL-3, n = 44; plasma lyso-Gb3, n = 31). Discussion In this investigation, migalastat was associated with a clinically relevant improvement in diarrhea symptoms based on the GSRS-D in patients with Fabry disease treated during the phase 3 FACETS trial. Based on the estimated MCID of 0.33, statistically significantly more patients treated with migalastat achieved a clinically Page 5 of 7 Schiffmann et al. Orphanet Journal of Rare Diseases (2018) 13:68 Schiffmann et al. Orphanet Journal of Rare Diseases (2018) 13:68 Schiffmann et al. Orphanet Journal of Rare Diseases (2018) 13:68 Table 1 Logistic regression assessing correlation between PTC GL-3 reductions and GSRS-D improvement (patients with amenable mutations) Parameter and Criteria Odds Ratio 95% CI of Odds Ratio GSRS-D Reduction From Baseline of 0.33 (n = 50) 5.55 (1.17–26.26) p = .031 Kidney Peritubular Capillary GL-3 Reduction From Baseline > 0.1 CI confidence interval, GL-3 globotriaosylceramide, GSRS-D Gastrointestinal Symptoms Rating Scale—diarrhea symptoms were common, with diarrhea occurring in 56% of patients with amenable mutations at baseline based on the diarrhea domain of the patient-reported GSRS [15]. As previously reported, 6 months of treatment with migalastat resulted in a significant improvement in diarrhea and reflux compared with placebo-treated pa- tients; for diarrhea, this improvement was sustained over 24 months [15]. Over this time, there was also a signifi- cant improvement in indigestion and a trend towards improvement in constipation [15]. CI confidence interval, GL-3 globotriaosylceramide, GSRS-D Gastrointestinal Symptoms Rating Scale—diarrhea p p [ ] In the FACETS trial, migalastat reduced substrate levels of GL-3 in patients with Fabry disease [15]. It has been postulated in the literature that GL-3 deposition in endothelial intestinal vasculature and enteric ganglia may contribute to the gastrointestinal manifestations of Fabry disease, with both cell types shown to accumulate GL-3 in Fabry patients with gastrointestinal symptoms [10, 29, 30]. Abnormal function of the enteric plexi is recognized as a potential mechanism causing irritable bowel syndrome, for which patients report similar gastrointestinal symptoms to Fabry disease [28]. Al- though GL-3 levels in the gastrointestinal tract were not assessed in this study, we hypothesized that other mea- sures of disease substrate (i.e., PTC GL-3 and plasma lyso-Gb3) would likely reflect GL-3 changes in the gastrointestinal system, and may be correlated with im- provements in gastrointestinal symptoms. Thus, we ex- plored the correlation between reduction in kidney GL-3 inclusions and improvement in GSRS-D scores. Discussion The re- sults indicate that reductions in PTC GL-3 inclusions were significantly associated with improvements in diar- rhea (GSRS-D scores). Possibly due to the smaller data set available for lyso-Gb3, no statistically significant asso- ciation was observed between plasma lyso-Gb3 and GSRS-D scores. Additional studies exploring the poten- tial for plasma lyso-Gb3 to be used as a surrogate end- point are warranted. Based on these results, reductions in GL-3 could be useful as a surrogate endpoint for predicting clinical benefit (i.e., improvement in diarrhea) with migalastat in patients with Fabry disease. meaningful improvement in diarrhea symptoms versus those receiving placebo (43% vs 11%; p = .02) after 6 months of treatment. This statistical significance was maintained when the analysis was restricted to only pa- tients with diarrhea symptoms at baseline (71% vs 20%; p = .02). A sensitivity analysis using a higher MCID threshold of 0.66 supports these findings. Additionally, a correlation between reduction of PTC GL-3 inclusions and diarrhea improvement was observed; patients with a PTC GL-3 inclusion reduction of > 0.1 were 5.6 times more likely to also have improvement in diarrhea symp- toms (p = .031). The correlation between reduction in the other disease substrate, plasma lyso-Gb3, and im- provement in GSRS-D was nonsignificant; a small sam- ple size may have confounded this result. Diarrhea is among the most common and most troublesome of the gastrointestinal symptoms experi- enced by Fabry patients. The abnormal accumulation of GL-3 in neurons of the peripheral nervous system, resulting in altered autonomic function and gastrointes- tinal disturbances, may contribute to the high prevalence of gastrointestinal symptoms in patients with Fabry dis- ease [26, 27]. Since up to 67% of patients report that they experience gastrointestinal symptoms, including diarrhea, some or all of the time [12], and patients re- port as many as 12 bouts of diarrhea per day [28], symp- toms can result in significantly reduced quality of life [11]. Thus, reduction in the severity and frequency of diarrhea can be particularly important for patients. Data from open-label studies with ERT have previously re- ported improvements in gastrointestinal symptoms fol- lowing treatment with agalsidase alfa [11] or agalsidase beta [10]. Discussion After 12 months of ERT, agalsidase alfa re- duced the prevalence of diarrhea by 8% [11]; similarly, following 6–7 months of therapy with agalsidase beta, episodes of diarrhea were reduced, and remained rare or occasional while therapy was maintained (≥3 years) [10]. Our results suggest a similar, clinically relevant improve- ment in diarrhea symptoms with migalastat treatment. One limitation of these analyses is that the GSRS has not been validated specifically in Fabry disease. Nonethe- less, acceptable psychometric properties of the GSRS, including reliability, stability, and construct validity, have been established in patients with irritable bowel syn- drome, gastroesophageal reflux disorder, and dyspepsia [18, 31–33]. Across these studies, the GSRS-D has demonstrated consistently strong psychometric proper- ties with reliability (Cronbach’s α) between 0.72–0.84, test-retest stability (intra-class correlation coefficient) between 0.38–0.70, and construct validity, as evidenced by correlations with various health-related quality of life instruments including SF-36, Quality of Life in Reflux and Dyspepsia (QOLRAD), and Psychological General The analyses presented add to previously published data demonstrating that migalastat improves gastrointes- tinal signs and symptoms in patients with Fabry disease. In the FACETS study, gastrointestinal signs and Page 6 of 7 Page 6 of 7 Schiffmann et al. Orphanet Journal of Rare Diseases (2018) 13:68 Page 6 of 7 Well-being (PGWB) [32, 33]. These psychometric prop- erties make a case for further use, examination, and per- haps, validation, of the GSRS in the Fabry patient population. An additional limitation is that the reported p-values are nominal p-values that have not been adjusted for multiplicity. As such, these should be inter- preted with caution. Conclusions To our knowledge, the FACETS study is the only double- blind, placebo-controlled study to evaluate gastrointestinal signs and symptoms in patients with Fabry disease. Re- sponder analyses demonstrate that migalastat provided a clinically meaningful reduction in diarrhea in patients with Fabry disease and amenable mutations. These data add to the evidence that migalastat improves gastrointestinal signs and symptoms, including diarrhea, reflux, and indi- gestion, in patients with Fabry disease [15]. Correlations between GSRS-D scores and PTC GL-3 inclusion reduc- tions suggest that PTC GL-3 inclusions are a potential surrogate endpoint that may predict clinical outcomes with migalastat treatment in Fabry disease. Given that diarrhea occurs frequently and is highly troublesome in Fabry disease, patients with an amenable GLA mutation may derive meaningful symptom relief from treatment with migalastat. Acknowledgements The authors acknowledge the scientific writing services of Virginia Owen, PhD. Additional writing and editorial assistance was provided by Sally Mitchell, PhD, and Dana Francis, PhD, from ApotheCom (Yardley, PA); this assistance was funded by Amicus. Competing interests RS has served as a consultant for and received research funding from Protalix Biotherapeutics and Amicus. DGB has received research funding, serves as a consultant, and is on the speaker’s bureau for Amicus and Genzyme, and has received research funding from Shire. AJ has received advisory honoraria and speaker’s fees from Shire, Amicus, Biomarin, and Genzyme. DAH has served as a consultant for and received research funding and honoraria from Amicus, Shire, Genzyme, Protalix, and Actelion. RG has received honoraria from Amicus, Biomarin, Genzyme, and Shire. UFR reports other support from Amicus during the conduct of the study, grant support and speaker’s honoraria from Amicus, Genzyme, and Shire HGT outside the submitted work, and research funding from Novo Nordisk Research Foundation. SPS reports grants and non-financial support from Amicus dur- ing the conduct of the study. LB has served as a consultant for Protalix. RBC has served as a consultant for Amicus and has received grants from the Na- tional Institutes of Health. JCJ does not have anything to disclose. KN has served as an advisor for Amicus, Shire HGT, and Genzyme, has received re- search support from Amicus and Shire HGT, and has received travel support from Genzyme. FH, AM, JPC, NS, and JAB are employees of and hold stock in Amicus. RS has served as a consultant for and received research funding from Protalix Biotherapeutics and Amicus. DGB has received research funding, serves as a consultant, and is on the speaker’s bureau for Amicus and Genzyme, and has received research funding from Shire. AJ has received advisory honoraria and speaker’s fees from Shire, Amicus, Biomarin, and Genzyme. DAH has served as a consultant for and received research funding and honoraria from Amicus, Shire, Genzyme, Protalix, and Actelion. RG has received honoraria from Amicus, Biomarin, Genzyme, and Shire. UFR reports other support from Amicus during the conduct of the study, grant support and speaker’s honoraria from Amicus, Genzyme, and Shire HGT outside the submitted work, and research funding from Novo Nordisk Research Availability of data and materials Not applicable; data are presented in the main paper. 4. Rigoldi M, Concolino D, Morrone A, et al. Intrafamilial phenotypic variability in four families with Anderson-Fabry disease. Clin Genet. 2014;86(3):258–63. 5. Messina S, Tortorella G, Concolino D, et al. Congenital muscular dystrophy with defective alpha-dystroglycan, cerebellar hypoplasia, and epilepsy. Neurology. 2009;73(19):1599–601. Publisher’s Note S i N i Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Abbreviations ERT: Enzyme replacement therapy; GL-3: Globotriaosylceramide; GSRS: Gastrointestinal Symptom Rating Scale; GSRS-D: Gastrointestinal Symptom Rating Scale—diarrhea; lyso-Gb3: Globotriaosylsphingosine; MCID: Minimal clinically important difference; PGWB: Psychological General Well-being; PTC: Kidney peritubular capillary; QOLRAD: Quality of Life in Reflux and Dyspepsia Received: 29 November 2017 Accepted: 18 April 2018 References kl 1. Meikle PJ, Hopwood JJ, Clague AE, Carey WF. Prevalence of lysosomal storage disorders. JAMA. 1999;281(3):249–54. 1. Meikle PJ, Hopwood JJ, Clague AE, Carey WF. Prevalence of lysosomal storage disorders. JAMA. 1999;281(3):249–54. 2. Desnick RJ, Ioannou Y, Eng CM. α-Galactosidase A deficiency: Fabry disease. The online metabolic and molecular bases of inherited disease. 2016; https://doi.org/10.1036/ommbid.181. 2. Desnick RJ, Ioannou Y, Eng CM. α-Galactosidase A deficiency: Fabry disease. The online metabolic and molecular bases of inherited disease. 2016; https://doi.org/10.1036/ommbid.181. Authors’ contributions The study (AT1001–011) was designed by the authors and the sponsor (Amicus). Data collection and analyses were undertaken by the sponsor (Amicus) in collaboration with a core group of investigators. The first draft of the manuscript was written by RS and reviewed by all the authors. All authors vouch for the completeness and accuracy of the data and analyses and for the fidelity of the study to the protocol. All authors made the decision to submit the manuscript for publication. All authors read and approved the final manuscript. 6. Eng CM, Banikazemi M, Gordon RE, et al. A phase 1/2 clinical trial of enzyme replacement in fabry disease: pharmacokinetic, substrate clearance, and safety studies. Am J Hum Genet. 2001;68(3):711–22. 7. Schiffmann R, Murray GJ, Treco D, et al. Infusion of alpha-galactosidase a reduces tissue globotriaosylceramide storage in patients with Fabry disease. Proc Natl Acad Sci U S A. 2000;97(1):365–70. 8. Buda P, Ksiazyk J, Tylki-Szymanska A. Gastroenterological complications of Anderson-Fabry disease. Curr Pharm Des. 2013;19(33):6009–13. Author details 1 1Baylor Scott & White Research Institute, Dallas, TX, USA. 2Hôpital du Sacré-Coeur, University of Montreal, Montreal, Quebec, Canada. 3Salford Royal Foundation Trust, Manchester, Greater Manchester, UK. 4NHS Foundation Trust, Royal Free Hospital, London, UK. 5Medical Genetics Service, HCPA/UFRGS, Porto Alegre, Brazil. 6Rigshospitalet, University of Copenhagen, Copenhagen, Denmark. 7Emory University School of Medicine, Atlanta, GA, USA. 8Miller School of Medicine, University of Miami, Miami, FL, USA. 9Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA. 10School of Medicine, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA. 11Amicus Therapeutics, Inc., Cranbury, NJ, USA. 12Department of Nephrology, Royal Melbourne Hospital, Parkville, VIC, Australia. 13Present Address: UC Davis MIND Institute, Sacramento, CA, USA. 14Institute of Metabolic Disease, 3812 Elm Street, Dallas, TX 75226, USA. Funding Funding This study was funded by Amicus Therapeutics, Inc., Cranbury, NJ. Funding This study was funded by Amicus Therapeutics, Inc., Cranbury, NJ. This study was funded by Amicus Therapeutics, Inc., Cranbury, NJ. This study was funded by Amicus Therapeutics, Inc., Cranbury, NJ. 3. Germain DP. Fabry disease. Orphanet J Rare Dis. 2010;5:30. Availability of data and materials Not applicable; data are presented in the main paper. Ethics approval and consent to participate 9. Mehta A, Ricci R, Widmer U, et al. Fabry disease defined: baseline clinical manifestations of 366 patients in the Fabry outcome survey. Eur J Clin Investig. 2004;34(3):236–42. Written consent to participate in the study was obtained from all patients. The study was approved by the institutional review board or ethics committee at each participating center and was conducted in accordance with the International Conference on Harmonisation Good Clinical Practice guidelines and the principles of the Declaration of Helsinki. Written consent to participate in the study was obtained from all patients. The study was approved by the institutional review board or ethics committee at each participating center and was conducted in accordance with the International Conference on Harmonisation Good Clinical Practice guidelines and the principles of the Declaration of Helsinki. 10. Banikazemi M, Ullman T, Desnick RJ. Gastrointestinal manifestations of Fabry disease: clinical response to enzyme replacement therapy. Mol Genet Metab. 2005;85(4):255–9. Page 7 of 7 Schiffmann et al. Orphanet Journal of Rare Diseases (2018) 13:68 11. Hoffmann B, Schwarz M, Mehta A, Keshav S. Gastrointestinal symptoms in 342 patients with Fabry disease: prevalence and response to enzyme replacement therapy. Clin Gastroenterol Hepatol. 2007;5(12):1447–53. y 12. Gold KF, Pastores GM, Botteman MF, et al. Quality of life of patients with Fabry disease. Qual Life Res. 2002;11(4):317–27. 13. Benjamin ER, Della Valle MC, Wu X, et al. The validation of 13. Benjamin ER, Della Valle MC, Wu X, et al. The validation of pharmacogenetics for the identification of Fabry patients to be migalastat. Genet Med. 2017;19(4):430–8. 14. Galafold [summary of product characteristics]. Buckinghamshire, UK; Amicus Therapeutics UK Ltd: 2017. http://www.ema.europa.eu/docs/en_GB/document_ library/EPAR_-_Product_Information/human/004059/WC500208434.pdf. 15. Germain DP, Hughes DA, Nicholls K, et al. Treatment of Fabry's disease with the pharmacologic chaperone migalastat. N Engl J Med. 2016;375(6):545–55. 16. Hughes DA, Nicholls K, Shankar SP, et al. Oral pharmacological chaperone migalastat compared with enzyme replacement therapy in Fabry disease: 18-month results from the randomised phase III ATTRACT study. J Med Genet. 2017;54(4):288–96. 17. Markham A. Migalastat: first global approval. Drugs. 2016;76(11):1147–52. 18. Svedlund J, Sjodin I, Dotevall G. GSRS—a clinical rating scale for gastrointestinal symptoms in patients with irritable bowel syndrome and peptic ulcer disease. Dig Dis Sci. 1988;33(2):129–34. 19. Barisoni L, Jennette JC, Colvin R, et al. Novel quantitative method to evaluate globotriaosylceramide inclusions in renal peritubular capillaries by virtual microscopy in patients with Fabry disease. Arch Pathol Lab Med. 2012;136(7):816–24. 20. Schiffmann et al. Orphanet Journal of Rare Diseases (2018) 13:68 Schiffmann et al. Orphanet Journal of Rare Diseases (2018) 13:68 Ethics approval and consent to participate Boutin M, Auray-Blais C. Multiplex tandem mass spectrometry analysis of novel plasma lyso-Gb(3)-related analogues in Fabry disease. Anal Chem. 2014;86(7):3476–83. 21. Benjamin ER, Hamler R, Brignol N, Boyd R, Yu J, Bragat A, et al. Migalastat reduces plasma globotriaosylsphingosine (lyso-Gb3) in Fabry patients: results from the FACETS phase 3 study. J Inherit Metab Dis. 2014;37(Suppl 1):S161. 22. Sterneck M, Settmacher U, Ganten T, et al. Improvement in gastrointestinal and health-related quality of life outcomes after conversion from mycophenolate mofetil to enteric-coated mycophenolate sodium in liver transplant recipients. Transplant Proc. 2014;46(1):234–40. 23. Manger B, Hiepe F, Schneider M, et al. Impact of switching from mycophenolate mofetil to enteric-coated mycophenolate sodium on gastrointestinal side effects in patients with autoimmune disease: a phase III, open-label, single-arm, multicenter study. Clin Exp Gastroenterol. 2015;8:205–13. 24. Chan L, Mulgaonkar S, Walker R, Arns W, Ambuhl P, Schiavelli R. Patient- reported gastrointestinal symptom burden and health-related quality of life following conversion from mycophenolate mofetil to enteric-coated mycophenolate sodium. Transplantation. 2006;81(9):1290–7. 25. Xu X, Tian L, Wei LJ. Combining dependent tests for linkage or association across multiple phenotypic traits. Biostatistics. 2003;4(2):223–9. 26. Tuttolomondo A, Pecoraro R, Simonetta I, et al. Neurological complications of Anderson-Fabry disease. Curr Pharmacy Des. 2013;19(33):6014–30. 27. Tuttolomondo A, Pecoraro R, Simonetta I, et al. Anderson-Fabry disease: a multiorgan disease. Curr Pharmacy Des. 2013;19(33):5974–96. 28. Keshav S. Gastrointestinal manifestations of Fabry disease. In: Mehta A, Beck M, Sunder-Plassmann G, editors. Fabry disease: perspectives from 5 years of FOS. Oxford: Oxford PharmaGenesis; 2006. 29. O'Brien BD, Shnitka TK, McDougall R, et al. Pathophysiologic and ultrastructural basis for intestinal symptoms in Fabry’s disease. Gastroenterology. 1982;82(5 Pt 1):957–62. 30. Politei J, Thurberg BL, Wallace E, et al. Gastrointestinal involvement in Fabry disease. So important, yet often neglected. Clin Genet. 2016;89(1):5–9. 31. Wiklund IK, Fullerton S, Hawkey CJ, et al. An irritable bowel syndrome- specific symptom questionnaire: development and validation. Scand J Gastroenterol. 2003;38(9):947–54. 32. Kulich KR, Madisch A, Pacini F, et al. Reliability and validity of the gastrointestinal symptom rating scale (GSRS) and quality of life in reflux and dyspepsia (QOLRAD) questionnaire in dyspepsia: a six-country study. Health Qual Life Outcomes. 2008;6:12. 33. Revicki DA, Wood M, Wiklund I, Crawley J. Reliability and validity of the gastrointestinal symptom rating scale in patients with gastroesophageal reflux disease. Qual Life Res. 1998;7(1):75–83.
https://openalex.org/W4384115648	https://zenodo.org/records/8087806/files/REPEAT_Climate_Progress_and_the_117th_Congress.pdf	English	null	Climate Progress and the 117th Congress: The Impacts of the Inflation Reduction Act and Infrastructure Investment and Jobs Act	Zenodo (CERN European Organization for Nuclear Research)	2,023	cc-by	49,477	Climate Progress and the 117th Congress: The Impacts of the Inflation Reduction Act and Infrastructure Investment and Jobs Act July 2023 repeatproject.org Climate Progress and the 117th Congress: The Impacts of the Inflation Reduction Act and Infrastructure Investment and Jobs Act July 2023 repeatproject.org repeatproject.org Climate Progress and the 117th Congress: The Impacts of the Inflation Reduction Act and Infrastructure Investment and Jobs Act Jesse D. Jenkins1, Erin N. Mayfield2, Jamil Farbes3, Greg Schivley1, Neha Patankar4, and Ryan Jones3 nceton University, Zero-carbon Energy Systems Research and Optimization Laboratory (ZERO Lab) rtmouth College, Sustainable Transition Lab olved Energy Research ghamton University ed citation: Jenkins, J.D., Mayfield, E.N., Farbes, J., Schivley, G., Patankar, N., and Jones, R., “Climate Progress and the 117th Congress: acts of the Inflation Reduction Act and the Infrastructure Investment and Jobs Act ,” REPEAT Project, Princeton, NJ, July 2023. 5281/zenodo.8087806 ion: July 12, 2023 2 Climate Progress and the 117th Congress: The Impacts of the Inflation Reduction Act and Infrastructure Investment and Jobs Act Jesse D. Jenkins1, Erin N. Mayfield2, Jamil Farbes3, Greg Schivley1, Neha Patankar4, and Ryan Jones3 In this report, you will find results for greenhouse gas emissions, clean energy and electri more, along with estimated impacts on energy expenditures, capital investment, energy s Given the significant uncertainty about future outcomes, all results in this report should analysis regularly as new data and inputs become available and new policies are propose Note that this work has not been subject to formal peer review. 1 – This includes an increase in near-term fossil fuel prices due to Russia’s inv 2 – Note that this report does not include the impact of light duty vehicle tailp However, modeled results are very close to compliant with these rules in all c 3 – See http://bit.ly/REPEAT-Policies for a complete list of policies in IRA and Climate Progress and the 117th Congress: The Impacts of the Inflation Reduction Act and Infrastructure Investment and Jobs Act Jesse D. Jenkins1, Erin N. Mayfield2, Jamil Farbes3, Greg Schivley1, Neha Patankar4, and Ryan Jones3 Suggested citation: Jenkins, J.D., Mayfield, E.N., Farbes, J., Schivley, G., Patankar, N., and Jones, R., “Climate Progress and the 117th Congress: The Impacts of the Inflation Reduction Act and the Infrastructure Investment and Jobs Act ,” REPEAT Project, Princeton, NJ, July 2023. DOI: 10.5281/zenodo.8087806 This version: July 12, 2023 for the REPEAT Project was provided by a grant from the Hewlett Foundation 2 Forward With the close of the 117th Congress in Jan legislation passed during this landmark ses of 2022 (IRA) and H.R. 3684, the Infrastruc impact of these laws on the energy system In this revised analysis, we have updated a data and analysis on oil and gas sector met Report on the Inflation Reduction Act relea (‘Conservative’, ‘Mid-range’, and ‘Optimisti of constraints on supply chains and other r scenario which only reflects policies enacte cost-effective pathway to reduce U.S. green United States’ mid- and long-term climate In this report, you will find results for green more, along with estimated impacts on ene Given the significant uncertainty about fut analysis regularly as new data and inputs b Note that this work has not been subject to Forward With the close of the 117th Congress in January 2023, REPEAT Project has completed a rev legislation passed during this landmark session. This includes detailed analysis of the com of 2022 (IRA) and H.R. 3684, the Infrastructure Investment and Jobs Act of 2021 (IIJA). T impact of these laws on the energy system and greenhouse gas emissions trajectory of th In this revised analysis, we have updated all assumptions to reflect the latest data availab data and analysis on oil and gas sector methane emissions and abatement opportunities Report on the Inflation Reduction Act released in August, 2022. This revised analysis now (‘Conservative’, ‘Mid-range’, and ‘Optimistic’) to better reflect uncertainty about the effec of constraints on supply chains and other rate-limiting factors.3 This report also presents scenario which only reflects policies enacted as of the start of the 117th Congress in Janua cost-effective pathway to reduce U.S. greenhouse gas emissions to 50-52% below 2005 le United States’ mid- and long-term climate mitigation goals. 3 1 – This includes an increase in near-term fossil fuel prices due to Russia’s invasion of Ukraine and revised assumptions on electric vehicle uptake reflecting current market trends. 2 – Note that this report does not include the impact of light duty vehicle tailpipe emissions standards through MY2026 or heavy-duty vehicle soot rule for MY2027+ finalized in late 2022. However, modeled results are very close to compliant with these rules in all cases. Both will be explicitly treated in subsequent REPEAT Project analysis later this year. 3 – See http://bit.ly/REPEAT-Policies for a complete list of policies in IRA and IIJA and assumptions and treatment of each policy under the three Current Policies scenarios. 3 1 – This includes an increase in near-term fossil fuel prices due to Russia’s invasion of Ukraine and revised assumptions on electric vehicle uptake reflecting current market trends. 2 – Note that this report does not include the impact of light duty vehicle tailpipe emissions standards through MY2026 or heavy-duty vehicle soot rule for MY2027+ finalized in late 2022. However, modeled results are very close to compliant with these rules in all cases. Both will be explicitly treated in subsequent REPEAT Project analysis later this year. 3 – See http://bit.ly/REPEAT-Policies for a complete list of policies in IRA and IIJA and assumptions and treatment of each policy under the three Current Policies scenarios. Forward close of the 117th Congress in January 2023, REPEAT Project has completed a revised analysis of the climate and energy system impacts of n passed during this landmark session. This includes detailed analysis of the combined impacts of H.R. 5376, the Inflation Reduction Act RA) and H.R. 3684, the Infrastructure Investment and Jobs Act of 2021 (IIJA). This report presents REPEAT Project’s revised findings on the these laws on the energy system and greenhouse gas emissions trajectory of the United States. his revised analysis, we have updated all assumptions to reflect the latest data available at year-end 20221 and improved the quality of source a and analysis on oil and gas sector methane emissions and abatement opportunities in agriculture and forestry sectors relative to our Preliminary ort on the Inflation Reduction Act released in August, 2022. This revised analysis now includes a range of three Current Policies2 scenarios nservative’, ‘Mid-range’, and ‘Optimistic’) to better reflect uncertainty about the effectiveness of IRA and IIJA provisions and the potential impacts onstraints on supply chains and other rate-limiting factors.3 This report also presents two benchmark scenarios: a Frozen Policies (Jan. ‘21) nario which only reflects policies enacted as of the start of the 117th Congress in January 2021; and a Net-Zero Pathway scenario, which reflects a t-effective pathway to reduce U.S. greenhouse gas emissions to 50-52% below 2005 levels by 2030 and net-zero by 2050, consistent with the ted States’ mid- and long-term climate mitigation goals. will find results for greenhouse gas emissions, clean energy and electric vehicle deployment, fossil energy production and use, and timated impacts on energy expenditures, capital investment, energy supply-related employment, air pollution, and public health. uture outcomes, all results in this report should be considered approximate. REPEAT Project updates our become available and new policies are proposed and enacted. Given the significant uncertainty about future outcomes, all results in this report should be considered approximate. REPEAT Project updates our analysis regularly as new data and inputs become available and new policies are proposed and enacted. en the significant uncertainty about future outcomes, all results in this report should be considered approximate. REPEAT Project updates our ysis regularly as new data and inputs become available and new policies are proposed and enacted. Note that this work has not been subject to formal peer review. Table of Contents Section Page Executive Summary 5 About REPEAT Project 20 Summary of Methods 23 Results 37 Greenhouse Gas Emissions 38 Energy Demand 47 Energy Supply 63 Electricity 68 Hydrogen 86 Carbon Capture, Use and Storage 90 Annual Energy Expenditures 93 Capital Investment in Energy Supply Infrastructure 95 Employment Impacts 98 Air Pollution and Public Health Impacts 103 Additional Impacts 107 Section Greenhouse Gas Emissions (Including Land Carbon Sinks) billion metric tons CO2-equivalent (Gt CO2-e)1 Infrastructure Law (IIJA) only: ~4.8 billion tons in 2030 (~28% below 2005)3 Net-Zero Pathway ~3.3 billion tons in 2030 (51% below 2005) Current Policies, including the Inflation Reduction Act (IRA): ~4.0-4.2 billion tons in 2030 (~37%-41% below 2005)3 2030 target: ≥50% below 2005 2021 emissions2: ~5.6 billion tons 2005 emissions2: ~6.7 billion tons Frozen Policies (Jan. ‘21) ~4.8 billion tons in 2030 (~28% below 2005) Legislation enacted by the 117th Congress could: • roughly double the pace of annual U.S. decarbonization to ~4%/year. • cut annual emissions in 2030 by ~0.5-0.8 billion metric tons relative to the Frozen Policies scenario. • get the U.S. to ~37-41% below 2005 historical GHG emissions (vs national target of 50-52%) • reduce cumulative GHG emissions by about 3.4-5.6 billion tons over the next decade (2023-2032). 3 4 5 6 Historical emissions2 Modeled emissions 2030 target: ≥50% below 2005 2021 emissions2: ~5.6 billion tons 2005 emissions2: ~6.7 billion tons Frozen Policies (Jan. ‘21) ~4.8 billion tons in 2030 (~28% below 2005) Section 4 Executive Summary 5 10 2015 2020 2025 2030 2035 Historical emissions2 Modeled emissions odeled Net U.S. Greenhouse Gas Emissions (Including Land Carbon Sinks) -equivalent (Gt CO2-e)1 Infrastructure Law (IIJA) only: ~4.8 billion tons in 2030 (~28% below 2005)3 Net-Zero Pathway ~3.3 billion tons in 2030 (51% below 2005) Current Policies, including the Inflation Reduction Act (IRA): ~4.0-4.2 billion tons in 2030 (~37%-41% below 2005)3 lculations use IPCC AR4 100 year global warming potential as per EPA Inventory of Greenhouse Gas Emissions and Sinks. All values should be regarded as approximate given uncertainty in future outcomes. A Inventory for 2005-2020; 2021 estimate from February 2023 draft EPA Inventory. any changes in passenger and freight miles traveled due to surface transportation, rail, and transit investments in IIJA. According to the Georgetown Climate Center, emissions impact of these changes depend heavily on state implementation of funding anywhere from -14 Mt/yr to +25 Mt/yr change in CO2 emissions from transportation in 2030. 2030 target: ≥50% below 2005 Legislation enacted by the 117th Congress could: • roughly double the pace of annual U.S. decarbonization to ~4%/year. • cut annual emissions in 2030 by ~0.5-0.8 billion metric tons relative to the Frozen Policies scenario. • get the U.S. to ~37-41% below 2005 historical GHG emissions (vs national target of 50-52%) • reduce cumulative GHG emissions by about 3.4-5.6 billion tons over the next decade (2023-2032). 2021 emissions2: ~5.6 billion tons Frozen Policies (Jan. ‘21) ~4.8 billion tons in 2030 (~28% below 2005) 7 Historical and Modeled Net U.S. Greenhouse Gas Emissions (Including Land Carbon Sinks) billion metric tons CO2-equivalent (Gt CO2-e)1 2 3 4 5 6 7 Historical emissions2 Modeled emissions Historical and Modeled Net U.S. Greenhouse Gas Emissions (Including Land Carbon Sinks) billion metric tons CO2-equivalent (Gt CO2-e)1 Infrastructure Law (IIJA) only: ~4.8 billion tons in 2030 (~28% below 2005)3 Net-Zero Pathway ~3.3 billion tons in 2030 (51% below 2005) Current Policies, including the Inflation Reduction Act (IRA): ~4.0-4.2 billion tons in 2030 (~37%-41% below 2005)3 2030 target: ≥50% below 2005 2021 emissions2: ~5.6 billion tons 2005 emissions2: ~6.7 billion tons Frozen Policies (Jan. ‘21) ~4.8 billion tons in 2030 (~28% below 2005) 2 3 4 5 6 7 Historical emissions2 Modeled emissions Historical and Modeled Net U.S. Legislation enacted by the 117th 2021 emissions2: ~5.6 billion tons 6 • roughly double the pace of annual U.S. decarbonization to ~4%/year. Infrastructure Law (IIJA) only: ~4.8 billion tons in 2030 (~28% below 2005)3 • cut annual emissions in 2030 by ~0.5-0.8 billion metric tons relative to the Frozen Policies scenario. • get the U.S. to ~37-41% below 2005 historical GHG emissions (vs national target of 50-52%) • get the U.S. to ~37-41% below 2005 historical GHG emissions (vs national target of 50-52%) Current Policies, including the Inflation Reduction Act (IRA): ~4.0-4.2 billion tons in 2030 (~37%-41% below 2005)3 2030 target: ≥50% below 2005 Net-Zero Pathway ~3.3 billion tons in 2030 (51% below 2005) 0 021 estimate from February 2023 draft EPA Inventory. nd freight miles traveled due to surface transportation, rail, and transit investments in IIJA. According to the Georgetown Climate Center, emissions impact of these changes depend heavily on state implementation of funding o +25 Mt/yr change in CO2 emissions from transportation in 2030. nts in IIJA. According to the Georgetown Climate Center, emissions impact of these changes depend heavily on state implementation of funding r and freight miles traveled due to surface transportation, rail, and transit investments in IIJA. According to the Georgetown Climate Center, emissio r to +25 Mt/yr change in CO2 emissions from transportation in 2030. 6 Modeled Net U.S. Greenhouse Gas Emissions (Including Land Carbon Sinks) percent below 2005 historical emissions1 Modeled Net U.S. Greenhouse Gas Emissions (Including Land Carbon Sinks) ercent below 2005 historical emissions1 022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 eled Net U.S. Greenhouse Gas Emissions (Including Land Carbon Sinks) below 2005 historical emissions1 Net-Zero Pathway Current Policies, incl. IRA historical net U.S. greenhouse gas emissions were 6,686 million metric tons of CO2-equivalent (EPA Inventory of Greenhouse Gas Emissions and Sinks). CO2-equivalent emissions calculations use IPCC AR4 100 year global warming as per EPA Inventory of Greenhouse Gas Emissions and Sinks. 2030 target: ≥50% below 2005 Frozen Policies (Jan. ‘21) Under Current Policies scenarios, U.S. emissions in 2030 fall to ~37-41% below 2005 emissions levels, falling short of the 50-52% goal set by President Biden. 37-41% below 2005 46-53% below 2005 29% below 2005 33% below 2005 51% below 2005 63% below 2005 Under Current Policies scenarios, U.S. emissions in 2030 fall to ~37-41% below 2005 emissions levels, falling short of the 50-52% goal set by President Biden. Legislation enacted by the 117th 2-equivalent (EPA Inventory of Greenhouse Gas Emissions and Sinks). CO2-equivalent emissions calculations use IPCC AR4 100 year global warming 3% 1% -1% -5% -9% 4% 2% -1% -5% -9% -15% 4% 1% -3% -8% -14% -22% 4% 1% -4% -10% -17% -26% 2024 2026 2028 2030 2032 2035 Frozen Policy (Jan '21) Conservative Mid-range Optimsitic Policies enacted by the 117th Congress will spur the first sustained decline in petroleum product consumption in U.S. history, with consumption of gasoline, diesel, and other petroleum-based fuels declining 5-10% from 2021 levels by 2030 and 15-26% by 2035. Current Policies Scenarios ual U.S. Petroleum Product Consumption vs 2021 (mmbbl/y) Changes in Annual U.S. Petroleum Product Consumption vs 2021 million barrels per year (mmbbl/y) Changes in Annual U.S. Petroleum Product Consumption vs 2021 million barrels per year (mmbbl/y) 2035 -26% 8 Changes in Annual U.S. Natural Gas Consumption vs 2021 trillion cubic feet per year (Tcf/year) -10% -10% -13% -14% -14% -13% -15% -17% -21% -24% -30% -10% -9% -13% -18% -21% -29% -13% -14% -16% -20% -24% -33% 2024 2026 2028 2030 2032 2035 Frozen Policy (Jan '21) Conservative Mid-range Optimsitic 9 Current Policies Scenarios Policies enacted by the 117th Congress reduce U.S. consumption of natural gas, as renewable energy displaces natural gas for power generation and efficiency and electrification lower gas use in buildings and industry. Under Current Policies scenarios, U.S. natural gas consumption declines by about one-fifth from 2021 levels by 2030 and by about one-third by 2035. Note – The magnitude of reductions in natural gas demand is affected by the balance of wind and solar power deployment rates (which offset gas consumption for power generation, all else equal) and the pace of electrification (which increases electricity demand and thus natural gas consumption for power generation, all else equal). As a result, the Conservative scenario with slower electrification results in lower gas demand in many years than the Middle or Optimistic Current Policies scenarios, as these scenarios have both faster wind and solar deployment and higher electricity demand from more rapid EV adoption and other electrification. Legislation enacted by the 117th 9 - 1,000 2,000 3,000 4,000 5,000 6,000 2021 2022 2023 2024 2025 2026 2027 2028 3,190 3,146 863 934 388 889 23 189 268 434 - 1,000 2,000 3,000 4,000 5,000 6,000 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 3,193 3,189 845 890 287 609 25 160 270 446 - 1,000 2,000 3,000 4,000 5,000 6,000 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 3,026 3,215 3,324 786 790 799 29 203 476 - 1,000 2,000 3,000 4,000 5,000 6,000 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 hydrogen electrolysis industrial steam transportation heating other Electricity Demand by Sector terawatt-hours per year (TWh/year) Current Policies Scenarios Frozen Policies (Jan ’21) Mid-range Optim Conservative 4,218 4,635 4,619 5,294 4,732 5,592 3,841 3,097 3,120 861 1,049 203 678 175 393 40 142 - 1,000 2,000 3,000 4,000 5,000 6,000 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 3,172 3,049 871 957 449 1,023 35 273 269 436 - 1,000 2,000 3,000 4,000 5,000 6,000 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 3,190 3,146 863 934 388 889 23 189 268 434 - 1,000 2,000 3,000 4,000 5,000 6,000 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 3,193 3,189 845 890 287 609 25 160 270 446 - 1,000 2,000 3,000 4,000 5,000 6,000 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 15 3,324 0 799 3 476 2030 2031 2032 2033 2034 2035 olysis by Sector Wh/year) Current Policies Scenarios Jan ’21) Mid-range Optimistic Net-Zero Pathway Conservative Growing adoption of electric vehicles, heat pumps, industrial electric boilers, and hydrogen electrolysis will drive a sustained increase in U.S. electricity consumption for the first time since the mid-2000s. Under Current Policies scenarios, U.S. electricity demand grows 20-25% from 2022 to 2030 and reaches 38-49% higher than 2022 demand by 2035. Legislation enacted by the 117th p et per year (Tcf/year) -8% -10% -10% -13% -14% -14% -13% -15% -17% -10% -9% -13% -18% -13% -14% -16% 20% 2024 2026 2028 2030 2032 2035 Frozen Policy (Jan '21) Conservative Mid-range Optimsitic Current Policies Scenarios ubic feet per year (Tcf/year) 2030 ervative Mid-range Optimsitic Current Policies Scenarios Note – The magnitude of reductions in natural gas demand is affected by the balance of wind and solar power deployment rates (which offset gas consumption for power generation, all else equal) and the pace of electrification (which increases electricity demand and thus natural gas consumption for power generation, all else equal). As a result, the Conservative scenario with slower electrification results in lower gas demand in many years than the Middle or Optimistic Current Policies scenarios, as these scenarios have both faster wind and solar deployment and higher electricity demand from more rapid EV adoption and other electrification. Note – The magnitude of reductions in natural gas demand is affected by the balance of wind and solar power deployment rates (which offset gas consumption for power generation, all else equal) and the pace of electrification (which increases electricity demand and thus natural gas consumption for power generation, all else equal). As a result, the Conservative scenario with slower electrification results in lower gas demand in many years than the Middle or Optimistic Current Policies scenarios, as these scenarios have both faster wind and solar deployment and higher electricity demand from more rapid EV adoption and other electrification. Legislation enacted by the 117th 18 4,635 4,619 5,294 4,732 5,592 4,797 5,739 4,376 5,382 3,097 3,120 861 1,049 203 678 175 393 40 142 - 1,000 2,000 3,000 4,000 5,000 6,000 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 3,172 3,049 871 957 449 1,023 35 273 269 436 - 1,000 2,000 3,000 4,000 5,000 6,000 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 3,190 3,146 863 934 388 889 23 189 268 434 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Current Policies Scenarios Mid-range Optimistic Net-Zero Pathway Growing a vehicle industrial e hydroge drive a sus U.S. elect for the fi m Under C scenario demand g 2022 to 2 38-49% dem 4,732 5,592 4,797 5,739 4,376 5,382 Current Policies Scenarios Mid-range 3,097 3,120 861 1,049 203 678 175 393 40 142 - 1,000 2,000 3,000 4,000 5,000 6,000 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 3,172 3,049 871 957 449 1,023 35 273 269 436 - 1,000 2,000 3,000 4,000 5,000 6,000 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 3,190 3,146 863 934 388 889 23 189 268 434 - 1,000 2,000 3,000 4,000 5,000 6,000 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 3,189 890 609 160 446 2032 2033 2034 2035 Current Policies Scenarios Mid-range Optimistic Net-Zero Pathway Growing adoption of electric vehicles, heat pumps, industrial electric boilers, and hydrogen electrolysis will drive a sustained increase in U.S. electricity consumption for the first time since the mid-2000s. Under Current Policies scenarios, U.S. electricity demand grows 20-25% from 2022 to 2030 and reaches 38-49% higher than 2022 demand by 2035. 5,294 4,732 5,592 4,797 5,739 4,376 5,382 Net-Zero Pathway Net-Zero Pathway Optimistic Growing adoption of electric vehicles, heat pumps, industrial electric boilers, and hydrogen electrolysis will drive a sustained increase in U.S. electricity consumption for the first time since the mid-2000s. Under Current Policies scenarios, U.S. electricity demand grows 20-25% from 2022 to 2030 and reaches 38-49% higher than 2022 demand by 2035. Legislation enacted by the 117th Modeled Annual Average Capacity Additions Un average gigawatts/year (GW/year) Under Current Policies scenarios, wind and solar PV capacity additions set new records over the coming decade. The average annual rate of solar PV additions more than doubles from a 19 GW peak in 2021 to an average rate of 44-51 GW/year from 2023-2030. Onshore wind capacity additions reach 39-43 GW/year from 2023-2030, nearly triple the peak of 15 GW in 2020. Solar PV additions increase further to 123-167 GW/year on average from 2031-2035, while wind additions remain roughly steady at 26-41 GW/year during that period. According to the U.S. Energy Information Administration’s Short-term Energy Outlook, the solar industry is currently on track and expected to average 41 GW/year of new capacity in 2023 and 2024, while the onshore wind industry lags behind our modeled pace, with an expected rate of just 7 GW/year over 2023 and 2024.3 Variation in modeled capacity across Current Policies scenarios from 2023-2030 primarily reflects exogenous constraints on annual capacity additions designed to approximate several challenges that are difficult to model explicitly but may limit renewable energy deployment, including the ability to site and permit projects at requisite pace and scale, expand supply chains, interconnect generating capacity, and hire and train the expanded energy workforce to build these projects. See p. 32 for details on these constraints. Capacity additions in the 2031-2035 period are generally not limited by exogenous constraints and reflect economic equilibrium outcomes. 15 4 59 10 6 15 19 4 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 storage solar PV (all) onshore wind offshore wind hydro nuclear natural gas w/CCS natural gas other fossil coal 12 1 - Historical data labels represent peak historical annual additions (source, 1950-2020: EIA Form 860; 2021-2022: EIA Short-term Energy Outlook, June 6, 2023) 2 - Modeled data labels reflect range across Current Policies scenarios (Conservative, Mid-range and Optimistic). Depicted bars represent results for the Mid-range scenario. 3 - EIA Short-term Energy Outlook, June 6, 2023 edition cal Annual Electricity Capacity Additions vs. Modeled Annual Average Capacity Additions Under Current Policies Scenarios gawatts/year (GW/year) Historical (EIA)1 Modeled (REPEAT Project)2 2023-2030 2031-2035 Under Current Policies scenarios, wind and solar PV capacity additions set new records over the coming decade. Legislation enacted by the 117th 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 10 Electricity Generation by Resource thousand terawatt-hours per year (TWh/year) Frozen Policies (Jan ’21) Electricity Generation by Resource thousand terawatt-hours per year (TWh/year) Frozen Policies (Jan ’21) 53% 69% 81% 91% 93% 95% 0 % 1 0% 2 0% 3 0% 4 0% 5 0% 6 0% 7 0% 8 0% 9 0% 1 0 % 0 1 2 3 4 5 6 7 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 50% 62% 67% 77% 83% 91% 0 % 1 0% 2 0% 3 0% 4 0% 5 0% 6 0% 7 0% 8 0% 9 0% 1 0 % 0 1 2 3 4 5 6 7 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 47% 55% 62% 75% 80% 89% 0 % 1 0% 2 0% 3 0% 4 0% 5 0% 6 0% 7 0% 8 0% 9 0% 1 0 % 0 1 2 3 4 5 6 7 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 50% 60% 66% 77% 81% 90% 0 % 1 0% 2 0% 3 0% 4 0% 5 0% 6 0% 7 0% 8 0% 9 0% 1 0 % 0 1 2 3 4 5 6 7 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 5% 59% 67% 71% 75% 0 % 1 0% 2 0% 3 0% 4 0% 5 0% 6 0% 7 0% 8 0% 9 0% 1 0 % 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Generation by Resource watt-hours per year (TWh/year) Current Policies Scenarios en Policies (Jan ’21) Mid-range Optimistic Net-Zero Pathway Conservative Current Policies Scenarios Mid-range Current Policies Scenarios Mid-range 7 cenarios Conservative Optimistic Net-Zero Pathway Conservative Optimistic 81% 0 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 0 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 0 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 11 15 4 59 10 6 0 50 100 150 200 250 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 1 - Historical data labels represent peak historical annual additions (source, 1950-2020: EIA Form 860; 20 2 - Modeled data labels reflect range across Current Policies scenarios (Conservative, Mid-range and Opt 3 - EIA Short-term Energy Outlook, June 6, 2023 edition Historical Annual Electricity Capacity Additions vs. Legislation enacted by the 117th The average annual rate of solar PV additions more than doubles from a 19 GW peak in 2021 to an average rate of 44-51 GW/year from 2023-2030. Onshore wind capacity additions reach 39-43 GW/year from 2023-2030, nearly triple the peak of 15 GW in 2020. Solar PV additions increase further to 123-167 GW/year on average from 2031-2035, while wind additions remain roughly steady at 26-41 GW/year during that period. According to the U.S. Energy Information Administration’s Short-term Energy Outlook, the solar industry is currently on track and expected to average 41 GW/year of new capacity in 2023 and 2024, while the onshore wind industry lags behind our modeled pace, with an expected rate of just 7 GW/year over 2023 and 2024.3 Variation in modeled capacity across Current Policies scenarios from 2023-2030 primarily reflects exogenous constraints on annual capacity additions designed to approximate several challenges that are difficult to model explicitly but may limit renewable energy deployment, including the ability to site and permit projects at requisite pace and scale, expand supply chains, interconnect generating capacity, and hire and train the expanded energy workforce to build these projects. See p. 32 for details on these constraints. Capacity additions in the 2031-2035 period are generally not limited by exogenous constraints and reflect economic equilibrium outcomes. 8-12 42-62 44-51 123-167 39-43 26-41 3 6 9-13 2-3 Historical Annual Electricity Capacity Additions vs. Modeled Annual Average Capacity Additions Under Current Policies Scenarios average gigawatts/year (GW/year) 250 Under Current Policies scenarios, wind and solar PV capacity additions set new records over the coming decade. The average annual rate of solar PV additions more than doubles from a 19 GW peak in 2021 to an average rate of 44-51 GW/year from 2023-2030. Onshore wind capacity additions reach 39-43 GW/year from 2023-2030, nearly triple the peak of 15 GW in 2020. Solar PV additions increase further to 123-167 GW/year on average from 2031-2035, while wind additions remain roughly steady at 26-41 GW/year during that period. According to the U.S. Energy Information Administration’s Short-term Energy Outlook, the solar industry is currently on track and expected to average 41 GW/year of new capacity in 2023 and 2024, while the onshore wind industry lags behind our modeled pace, with an expected rate of just 7 GW/year over 2023 and 2024.3 According to the U.S. Legislation enacted by the 117th transmission planning, siting, permitting and cost allocation practices can all potentially impede the real-world pace of transmission expansion. We explore the impacts of more constrained transmission expansion on the following page. Note that U.S. electricity demand has been roughly flat since the mid-2000s, and modeled transmission expansion rates under Current Policies are roughly equal to the historical pace achieved from the 1970s to the 1990s1, the last period during which U.S. electricity demand steadily increased. The pace of transmission expansion under the Net-Zero Pathway exceeds the historical 1978-1999 rate and is twice as fast as the more recent 2004-2016 period. 1 1.9% 1.2% 2.4% 1978-1999 2004-2016 Current Policies Net-Zero Pathway 1.5-1.8% 1 2 1 - Reported by U.S. Department of Energy and cited in Cembalest (2022), “Eye on the Market 12th Edition: 2022 Annual Energy Paper” p.12. 2 - Reported by UT Austin and cited in Cembalest (2022), p.12. To achieve the maximum emissions reduction under Current Policies, U.S. transmission capacity must expand roughly 50% faster through 2035 than the recent historical rate.2 While our modeling finds this outcome makes economic sense given incentives under IRA, current U.S. transmission planning, siting, permitting and cost allocation practices can all potentially impede the real-world pace of transmission expansion. We explore the impacts of more constrained transmission expansion on the following page. Note that U.S. electricity demand has been roughly flat since the mid-2000s, and modeled transmission expansion rates under Current Policies are roughly equal to the historical pace achieved from the 1970s to the 1990s1, the last period during which U.S. electricity demand steadily increased. The pace of transmission expansion under the Net-Zero Pathway exceeds the historical 1978-1999 rate and is twice as fast as the more recent 2004-2016 period. To achieve the maximum emissions reduction under Current Policies, U.S. transmission capacity must expand roughly 50% faster through 2035 than the recent historical rate.2 While our modeling finds this outcome makes economic sense given incentives under IRA, current U.S. transmission planning, siting, permitting and cost allocation practices can all potentially impede the real-world pace of transmission expansion. We explore the impacts of more constrained transmission expansion on the following page. Note that U.S. electricity demand has been roughly flat since the mid-2000s, and modeled transmission expansion rates under Current Policies are roughly equal to the historical pace achieved from the 1970s to the 1990s1, the last period during which U.S. Legislation enacted by the 117th Energy Information Administration’s Short-term Energy Outlook, the solar industry is c on track and expected to average 41 GW/year of new capacity in 2023 and 2024, while the onshore wind in lags behind our modeled pace, with an expected rate of just 7 GW/year over 2023 and 2024.3 storage solar PV (all) onshore wind offshore wind hydro nuclear natural gas w/CCS natural gas other fossil coal Variation in modeled capacity across Current Policies scenarios from 2023-2030 primarily reflects exogenous constraints on annual capacity additions designed to approximate several challenges that are difficult to model explicitly but may limit renewable energy deployment, including the ability to site and permit projects at requisite pace and scale, expand supply chains, interconnect generating capacity, and hire and train the expanded energy workforce to build these projects. See p. 32 for details on these constraints. Capacity additions in the 2031-2035 period are generally not limited by exogenous constraints and reflect economic equilibrium outcomes. 50 1 - Historical data labels represent peak historical annual additions (source, 1950-2020: EIA Form 860; 2021-2022: EIA Short-term Energy Outlook, June 6, 2023) 2 - Modeled data labels reflect range across Current Policies scenarios (Conservative, Mid-range and Optimistic). Depicted bars represent results for the Mid-range scenario. 3 - EIA Short-term Energy Outlook, June 6, 2023 edition 1 - Historical data labels represent peak historical annual additions (source, 1950-2020: EIA Form 860; 2021-2022: EIA Short-term Energy Outlook, June 6, 2023) 2 - Modeled data labels reflect range across Current Policies scenarios (Conservative, Mid-range and Optimistic). Depicted bars represent results for the Mid-range scenario. 3 - EIA Short-term Energy Outlook, June 6, 2023 edition 12 ross Current Policies scenarios (Conservative, Mid-range and Optimistic). Depicted bars represent results for the Mid-range scenario. 6 2023 edition Compound Annual Growth in Electricity Transmission Capacity, 2020-2035 vs. Historical Periods percent annual growth in gigawatt-miles 1.2% 2.4% 9 2004-2016 Current Policies Net-Zero Pathway 1.5-1.8% 1 2 1 - Reported by U.S. Department of Energy and cited in Cembalest (2022), “Eye on the Market 12th Edition: 2022 Annual Energy Paper” p.12. 2 - Reported by UT Austin and cited in Cembalest (2022), p.12. To achieve the maximum emissions reduction under Current Policies, U.S. transmission capacity must expand roughly 50% faster through 2035 than the recent historical rate.2 While our modeling finds this outcome makes economic sense given incentives under IRA, current U.S. Legislation enacted by the 117th electricity demand steadily increased. The pace of transmission expansion under the Net-Zero Pathway exceeds the historical 1978-1999 rate and is twice as fast as the more recent 2004-2016 period. Net-Zero Pathway 1 - Reported by U.S. Department of Energy and cited in Cembalest (2022), “Eye on the Market 12th Edition: 2022 Annual Energy Paper” p.12 2 - Reported by UT Austin and cited in Cembalest (2022), p.12. 1 - Reported by U.S. Department of Energy and cited in Cembalest (2022), “Eye on the Market 12th Edition: 2022 Annual Energy Paper” p.12 2 - Reported by UT Austin and cited in Cembalest (2022), p.12. 1 - Reported by U.S. Department of Energy and cited in Cembalest (2022), “Eye on the Market 12th Edition: 2022 Annual Energy Paper” p.12. 2 - Reported by UT Austin and cited in Cembalest (2022), p.12. ” p.12. 13 -50% -41% 1% -53% -34% -43% -51% -63% Mid-range Mid-Range, Transmission Constrained Optimistic Optimistic, Transmission Constrained Net-Zero Pathway 1 Expansion Constraints on Modeled Net U.S. Greenhouse Gas Emissions (Including Land Carbon Sinks) s1 1 - CO2-equivalent emissions calculations use IPCC AR4 100 year global warming potential as per EPA Inventory of Greenhouse Gas Emissions and Sinks. 2 - Transmission constrained cases limit total transmission capacity expansion to a compound annual growth rate of 1%/year, roughly equivalent to the 2004-2016 average historical pace. The maximum total increase in GW-miles for each model year is allocated as constraints on expansion of inter-regional transmission, interconnection lines for wind and interconnection lines for solar PV respectively in proportion to the total expansion for each category of lines under unconstrained Current Policies cases. If the United States cannot build new transmission capacity faster than the recent historical pace (~1%/year)2, the ‘emissions gap’ between Current Policies scenarios and U.S. climate goals widens to over 1 billion metric tons in 2030. 30 2035 2 2 1,070 MMT gap target Modeled Net U.S. Greenhouse Gas Emissions (Including Land Carbon Sinks t of Transmission Expansion Constraints on Modeled Net U.S. Greenhouse hange vs. 2005 emissions1 Impact of Transmission Expansion Constraints on Modeled Net U.S. Greenhouse Gas Emissions (Including Land C percent change vs. 2005 emissions1 Impact of Transmission Expansion Constraints on Modeled Net U.S. Greenhouse Gas Emissions (Including Land Carbon Sinks) percent change vs. Legislation enacted by the 117th 2005 emissions1 -38% -50% -33% -41% -41% -53% -34% -43% -51% -63% -70% -60% -50% -40% -30% -20% -10% 0% Mid-range Mid-Range, Transmission Constrained Optimistic Optimistic, Transmission Constrained Net-Zero Pathway Impact of Transmission Expansion Constraints on Modeled Net U.S. Greenhouse Gas Emissions (Including Land Carbo percent change vs. 2005 emissions1 1 - CO2-equivalent emissions calculations use IPCC AR4 100 year global warming potential as per EPA Inventory of Greenhouse Gas Emissions and 2 - Transmission constrained cases limit total transmission capacity expansion to a compound annual growth rate of 1%/year, roughly equivalent historical pace. The maximum total increase in GW-miles for each model year is allocated as constraints on expansion of inter-regional transmiss for wind and interconnection lines for solar PV respectively in proportion to the total expansion for each category of lines under unconstrained C If the United States cannot build capacity faster than the recen (~1%/year)2, the ‘emissions gap Policies scenarios and U.S. clima over 1 billion metric ton 2030 2035 2 2 1,070 MMT gap 1,140 MMT gap 2030 target -38% -50% -33% -41% -41% -53% -34% -43% -51% -63% Mid-range Mid-Range, Transmission Constrained Optimistic Optimistic, Transmission Constrained Net-Zero Pathway t change vs. 2005 emissions1 1 - CO2-equivalent emissions calculations use IPCC AR4 100 year global warming potential as per EPA Inventory of Greenhouse Gas Emissions and Sinks. 2 Transmission constrained cases limit total transmission capacity expansion to a compound annual growth rate of 1%/year roughly equivalent to the 2004 2016 average If the United States cannot build new transmissio capacity faster than the recent historical pace (~1%/year)2, the ‘emissions gap’ between Curre Policies scenarios and U.S. climate goals widens over 1 billion metric tons in 2030. 2030 2035 2 2 1,070 MMT gap 1,140 MMT gap 2030 target -38% -33% -41% -34% -51% -70% -60% -50% -40% -30% -20% -10% 0% 2030 1,070 MMT gap 1,140 MMT gap 2030 target 0% If the United States cannot build new transmission capacity faster than the recent historical pace (~1%/year)2, the ‘emissions gap’ between Current Policies scenarios and U.S. climate goals widens to over 1 billion metric tons in 2030. 1 - CO2-equivalent emissions calculations use IPCC AR4 100 year global warming potential as per EPA Inventory of Greenhouse Gas Emissions and Sinks. 2 - Transmission constrained cases limit total transmission capacity expansion to a compound annual growth rate of 1%/year, roughly equivalent to the 2004-2016 average historical pace. Legislation enacted by the 117th The maximum total increase in GW-miles for each model year is allocated as constraints on expansion of inter-regional transmission, interconnection lines for wind and interconnection lines for solar PV respectively in proportion to the total expansion for each category of lines under unconstrained Current Policies cases. -13 -30 -46 -67 -88 -100 -113 -127 -140 -60 -65 -71 -325 -275 -225 -175 -125 -75 -25 25 75 125 175 225 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 -6 -17 -28 -45 -62 -73 -85 -96 -107 -45 -49 -54 -325 -275 -225 -175 -125 -75 -25 25 75 125 175 225 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 -8-16-25-35-45-52-59-58-56-24-25-26 fixed o&m capital - supply side electricity transmission electricity distribution capital - demand side variable o&m fuels delivery fuel and commodities supply-side incentives demand-side incentives net difference Mid-range Change in Annual Energy Expenditures vs Frozen Policies as of January 2021 billions of 2023 US dollars1 Optimistic Current Policies Scenarios Conservative T I U A e 1 1 - Dollar values modeled internally as 2018 USD, and converted for display to 2023 USD using Bureau of Labor Statistics CPI Inflation Calc -8 -16 -25 -35 -45 -52 -59 -58 -56 -24 -25 -26 -325 -275 -225 -175 -125 -75 -25 25 75 125 175 225 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 0 -13 -30 -46 -67 -88 -100 -113 -127 -140 -60 -65 -71 -325 -275 -225 -175 -125 -75 -25 25 75 125 175 225 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 -6 -17 -28 -45 -62 -73 -85 -96 -107 -45 -49 -54 -325 -275 -225 -175 -125 -75 -25 25 75 125 175 225 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 25-35-45-52-59-58-56-24-25-26 capital - supply side electricity transmission electricity distribution mand side variable o&m fuels delivery fuel and commodities incentives demand-side incentives net difference Mid-range 15 ergy Expenditures vs Frozen Policies as of January 2021 Optimistic Current Policies Scenarios onservative The Inflation Reduction Act and Infrastructure Law lower annual U.S. energy expenditures ~3-7% in 2030, a savings of $59-$113 billion for U.S. households, businesses, and industry. Legislation enacted by the 117th Annual energy cost savings peak in 2032 prior to the scheduled expiration of several IRA incentives (i.e., tax credits for EV adoption, efficiency etc.) but remain about 1-3% lower than the Frozen Policies scenario through 2035. 1 - Dollar values modeled internally as 2018 USD, and converted for display to 2023 USD using Bureau of Labor Statistics CPI Inflation Calculator: 1 2018 UDS = 1.207 2023 USD. -59 -58 -56 -24 -25 -26 2030 2031 2032 2033 2034 2035 Mid-range Current Policies Scenarios Mid-range Current Policies Scenarios 175 225 electricity transmission fuels delivery net difference 5 5 capital - supply side variable o&m demand-side incentives electricity distribution fuel and commodities The Inflation Reduction Act and Infrastructure Law lower annual U.S. energy expenditures ~3-7% in 2030, a savings of $59-$113 billion for U.S. households, businesses, and industry. Annual energy cost savings peak in 2032 prior to the scheduled expiration of several IRA incentives (i.e., tax credits for EV adoption, efficiency etc.) but remain about 1-3% lower than the Frozen Policies scenario through 2035. The Inflation Reduction Act and Infrastructure Law lower annual U.S. energy expenditures ~3-7% in 2030, a savings of $59-$113 billion for U.S. households, businesses, and industry. Annual energy cost savings peak in 2032 prior to the scheduled expiration of several IRA incentives (i.e., tax credits for EV adoption, efficiency etc.) but remain about 1-3% lower than the Frozen Policies scenario through 2035. Annual energy cost savings peak in 2032 prior to the scheduled expiration of several IRA incentives (i.e., tax credits for EV adoption, efficiency etc.) but remain about 1-3% lower than the Frozen Policies scenario through 2035. Annual energy cost savings peak in 2032 prior to the scheduled expiration of several IRA incentives (i.e., tax credits for EV adoption, efficiency etc.) but remain about 1-3% lower than the Frozen Policies scenario through 2035. -30 -46 -67 -88 1 - Dollar values modeled internally as 2018 USD, and converted for display to 2023 USD using Bureau of Labor Statistics CPI Inflation Calculator: 1 2018 UDS = 1.207 2023 USD. 1 - Dollar values modeled internally as 2018 USD, and converted for display to 2023 USD using Bureau of Labor Statistics CPI Inflation Calculator: 1 2018 UDS = 1.207 2023 USD. Legislation enacted by the 117th 16 Mid-range Change in Energy Supply Related Employment by Resource vs Frozen Policies as of January 2021 thousands of jobs1 Optimistic Current Policies Scenarios Conservative Optimistic The Inflation Reduction Act and Infrastructure Law could increase energy supply related employment1 by about 1.4-1.7 million additional jobs in 2030 and 2.2-2.9 million by 2035. Solar, wind, and grid related jobs expand rapidly under Current Policies scenarios accounting for the vast majority of additional employment. Oil, natural gas, and coal related employment declines by ~50,000-70,000 jobs in 2030, roughly equal to the additional jobs created in CO2 transport & storage. The Inflation Reduction Act and Infrastructure Law could increase energy supply related employment1 by about 1.4-1.7 million additional jobs in 2030 and 2.2-2.9 million by 2035. Solar, wind, and grid related jobs expand rapidly under Current Policies scenarios accounting for the vast majority of additional employment. Oil, natural gas, and coal related employment declines by ~50,000-70,000 jobs in 2030, roughly equal to the additional jobs created in CO2 transport & storage. The Inflation Reduction Act and Infrastructure Law could increase energy supply related employment1 by about 1.4-1.7 million additional jobs in 2030 and 2.2-2.9 million by 2035. Legislation enacted by the 117th 15 96 332 357 415 476 950 938 1,131 842 358 640 707 712 1,011 36 120 123 176 116 137 90 126 113 130 86 91 25 33 38 24 200 36 39 45 62 63 62 61 47 113 116 117 117 117 500 1,000 1,500 2,000 2,500 3,000 3,500 electricity transmission electricity generation - solar electricity generation - wind electricity generation - storage electricity generation - natural gas electricity generation - natural gas w/ cc electricity generation - nuclear CO2 transport & storage biofuels industrial heat electrolysis (hydrogen) methane reforming (hydrogen) 'e-fuels' (fischer-tropsch liquids) liquefied natural gas export 16 e Capital Investment in Energy Supply Related Infrastructure, 2023-2035 US dollars1 1 - Dollar values modeled internally as 2018 USD, and converted for display to 2023 USD using Bureau of Labor Statistics CPI Inflation Calculator: 1 2018 UDS = 1.207 2023 USD. Values less than $20 billion not displayed in data labels. Capital investment in fuel and electricity distribution are not included. 2 - Includes associated transmission investments. 3,157 2,798 2,521 2,373 1,327 The Inflation Reduction Act and Infrastructure Law spur over a trillion dollars of additional capital investment in energy supply related infrastructure through 2035. Over four-fifths of this ~$2.4-2.8 trillion in cumulative investment is directed to wind and solar power and battery energy storage and related transmission grid investments. The laws drive additional investment in clean energy manufacturing and supply chains, distribution networks and demand side equipment that are not depicted here. Cumulative Capital Investment in Energy Supply Related Infrastructure, 2023-2035 billions of 2023 US dollars1 Capital Investment in Energy Supply Related Infrastructure, 2023-2035 US dollars1 Cumulative Capital Investment in Energy Supply Related Infrastructure, 2023-2035 billions of 2023 US dollars1 The Inflation Reduction Act and Infrastructure Law spur over a trillion dollars of additional capital investment in energy supply related infrastructure through 2035. 0 3,500 1 - Dollar values modeled internally as 2018 USD, and converted for display to 2023 USD using Bureau of Labor Statistics CPI Inflation Calculator: 1 2018 UDS = 1.207 2023 USD. Values less than $20 billion not displayed in data labels. Capital investment in fuel and electricity distribution are not included. 2 - Includes associated transmission investments. Legislation enacted by the 117th 137 249 343 679 1,057 1,178 203 260 300 423 553 729 60 118 188 242 266 390 -8 -11 -17 -22 -28 66 98 119 176 242 313 41 55 65 98 141 186 35 42 46 71 107 -500 0 500 1,000 1,500 2,000 2,500 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 73 165 357 606 793 953 163 203 260 355 420 592 39 75 153 209 228 335 -6 -12 -16 -22 47 62 96 148 193 261 31 39 55 83 109 153 29 36 42 58 76 112 -500 0 500 1,000 1,500 2,000 2,500 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 259 321 349 310 334 324 285 415 527 590 613 705 741 701 432 510 566 513 550 575 602 51 51 51 51 50 49 49 415 393 358 325 294 267 231 255 299 334 333 368 383 370 249 268 280 276 292 296 278 33 32 36 36 36 35 34 24 40 49 58 78 88 78 126 222 354 575 733 899 190 238 288 359 394 544 54 96 166 204 201 288 -5 -9 -10 -13 62 85 113 150 179 235 38 51 65 86 105 144 33 40 47 60 73 106 -500 0 500 1,000 1,500 2,000 2,500 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 The Inflation Re and Infrastructur increase ener related employ about 1.4-1.7 additional jobs 2.2-2.9 million That includes ab additional man jobs in 2030 an one million manufacturin 2035, primarily and wind turbine manufactu 1 - Employment in petroleum fuel refining, distribution, and retailing; hydrogen production, distribution and retailing; energy storage manufacturing, installation and operations; automotive supply chains and assembly; and energy efficiency are excluded from this analysis. Legislation enacted by the 117th oil & gas) professional trade agriculture other Mid-range Energy Supply Related Employment by Sector vs Frozen Policies as of January 2021 obs1 Optimistic Current Policies Scenarios Conservative 137 249 343 679 1,057 1,178 203 260 300 423 553 729 60 118 188 242 266 390 -8 -11 -17 -22 -28 66 98 119 176 242 313 41 55 65 98 141 186 35 42 46 71 107 144 -500 0 500 1,000 1,500 2,000 2,500 3,000 3,500 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 73 165 357 606 793 953 163 203 260 355 420 592 39 75 153 209 228 335 -6 -12 -16 -22 47 62 96 148 193 261 31 39 55 83 109 153 29 36 42 58 76 112 -500 0 500 1,000 1,500 2,000 2,500 3,000 3,500 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 9 310 334 324 285 0 613 705 741 701 6 513 550 575 602 51 50 49 49 8 325 294 267 231 4 333 368 383 370 0 276 292 296 278 36 36 35 34 58 78 88 78 onstruction utilities pipeline mining (incl. oil & gas) professional trade agriculture other 575 733 899 359 394 544 204 201 288 -9 -10 -13 150 179 235 86 105 144 60 73 106 2029 2030 2031 2032 2033 2034 2035 Mid-range 18 Related Employment by Sector vs Frozen Policies as of January 2021 Optimistic Current Policies Scenarios ative The Inflation Reduction Act and Infrastructure Law could increase energy supply related employment1 by about 1.4-1.7 million additional jobs in 2030 and 2.2-2.9 million by 2035. That includes about 600,000 additional manufacturing jobs in 2030 and roughly one million more manufacturing jobs in 2035, primarily in solar PV and wind turbine component manufacturing. 1 - Employment in petroleum fuel refining, distribution, and retailing; hydrogen production, distribution and retailing; energy storage manufacturing, installation and operations; automotive supply chains and assembly; and energy efficiency are excluded from this analysis. Values less than 5 thousand jobs not displayed in labels 1,428 1,674 2,207 387 585 963 1,451 1,800 2,388 543 820 1,056 1,676 2,007 2,911 The Inflation Reduction Act and Infrastructure Law could increase energy supply related employment1 by about 1.4-1.7 million additional jobs in 2030 and 2.2-2.9 million by 2035. Legislation enacted by the 117th Values less than 5 thousand jobs not displayed in labels Net: 505 733 1,033 1,428 1,674 2,207 387 585 963 1,451 1,800 2,388 543 820 1,056 1,676 2,007 137 249 343 679 1,057 1,178 203 260 300 423 553 729 60 118 188 242 266 390 -8 -11 -17 -22 -28 66 98 119 176 242 313 41 55 65 98 141 186 35 42 46 71 107 0 500 1,000 1,500 2,000 2,500 73 165 357 606 793 953 163 203 260 355 420 592 39 75 153 209 228 335 -6 -12 -16 -22 47 62 96 148 193 261 31 39 55 83 109 153 29 36 42 58 76 112 0 500 1,000 1,500 2,000 2,500 259 321 349 310 334 324 285 415 527 590 613 705 741 701 432 510 566 513 550 575 602 51 51 51 51 50 49 49 415 393 358 325 294 267 231 255 299 334 333 368 383 370 249 268 280 276 292 296 278 33 32 36 36 36 35 34 24 40 49 58 78 88 78 126 222 354 575 733 899 190 238 288 359 394 544 54 96 166 204 201 288 -5 -9 -10 -13 62 85 113 150 179 235 38 51 65 86 105 144 33 40 47 60 73 106 0 500 1,000 1,500 2,000 2,500 and Infras increas related about additiona 2.2-2.9 That inclu addition jobs in 2 one manuf 2035, pr and wind ma Net: 505 733 1,033 1,428 1,674 2,207 387 585 963 1,451 1,800 2,388 543 820 1,056 1,676 2,007 That includes about 600,000 additional manufacturing jobs in 2030 and roughly one million more manufacturing jobs in 2035, primarily in solar PV and wind turbine component manufacturing. That includes about 600,000 additional manufacturing jobs in 2030 and roughly one million more manufacturing jobs in 2035, primarily in solar PV and wind turbine component manufacturing. 500 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 500 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 500 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 1 - Employment in petroleum fuel refining, distribution, and retailing; hydrogen production, distribution and retailing; energy storage manufacturing, installation and operations; automotive supply chains and assembly; and energy efficiency are excluded from this analysis. Legislation enacted by the 117th 57 60 60 61 92 -26 -39 -55 -75 -23 -21 -58 64 192 329 444 539 736 374 453 577 1,013 1,660 1,986 63 154 134 230 221 246 -500 0 500 1,000 1,500 2,000 2,500 3,000 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 56 59 59 59 90 -30 -41 -57 -42 21 99 250 376 464 611 312 430 551 819 1,132 1,544 124 250 267 250 -500 0 500 1,000 1,500 2,000 2,500 3,000 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 6 8 8 8 9 10 810 791 757 719 682 643 586 134 133 116 96 77 68 59 398 400 463 385 381 379 377 97 92 101 99 97 96 90 437 583 621 601 681 739 789 49 49 49 47 49 51 54 68 188 239 325 467 523 390 138 200 259 235 263 257 272 52 53 54 54 82 -22 -30 -22 -50 58 163 289 380 415 532 355 460 595 843 1,110 1,490 55 94 121 201 195 181 -500 0 500 1,000 1,500 2,000 2,500 3,000 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Net: 505 733 1,033 1,428 1,674 2,207 387 585 963 1,451 1,800 2,388 543 820 1,056 1,676 2,007 1 - Employment in petroleum fuel refining, distribution, and retailing; hydrogen production, distribution and retailing; energy storage manufacturin operations; automotive supply chains and assembly; and energy efficiency are excluded from this analysis. Values less than 20 thousand jobs not di Solar, wind, and grid related jobs expand rapidly under Current Policies scenarios accounting for the vast majority of additional employment. 1 - Employment in petroleum fuel refining, distribution, and retailing; hydrogen production, distribution and retailing; energy storage manufacturing, installation and operations; automotive supply chains and assembly; and energy efficiency are excluded from this analysis. Values less than 20 thousand jobs not displayed in labels 1 - Employment in petroleum fuel refining, distribution, and retailing; hydrogen production, distribution and retailing; energy storage manufacturing, installation and operations; automotive supply chains and assembly; and energy efficiency are excluded from this analysis. Values less than 20 thousand jobs not displayed in labels 17 17 3,500 3,500 manufacturing construction utilities pipeline mining (incl. Legislation enacted by the 117th Values less than 5 thousand jobs not displayed in labels 1 - Employment in petroleum fuel refining, distribution, and retailing; hydrogen production, distribution and retailing; energy storage manufacturing, installation and operations; automotive supply chains and assembly; and energy efficiency are excluded from this analysis. Values less than 5 thousand jobs not displayed in labels 1 - Employment in petroleum fuel refining, distribution, and retailing; hydrogen production, distribution and retailing; energy storage manufacturing, installation and operations; automotive supply chains and assembly; and energy efficiency are excluded from this analysis. Values less than 5 thousand jobs not displayed in labels 1 - Employment in petroleum fuel refining, distribution, and retailing; hydrogen production, distribution and retailing; energy storage manufac operations; automotive supply chains and assembly; and energy efficiency are excluded from this analysis. Values less than 5 thousand jobs no 18 19 e Avoided Premature Deaths From Exposure to Fine Particulate Matter From Energy Activities vs Frozen Policies, 2023-2035 17.6 28.1 36.8 Reductions in fine particulate pollution spurred by the Inflation Reduction Act and Infrastructure Law could avoid roughly 17,000-37,000 premature deaths from 2023-2035, saving ~$150-325 billion in economic damages from avoided mortalities alone. lative Avoided Premature Deaths From Exposure to Fine Particulate Matter From Energy Activities vs Frozen Policies, 2023-2035 ds 17.6 28.1 36.8 Reductions in fine particulate pollution spurred by the Inflation Reduction Act and Infrastructure Law could avoid roughly 17,000-37,000 premature deaths from 2023-2035, saving ~$150-325 billion in economic damages from avoided mortalities alone. Cumulative Avoided Premature Deaths From Exposure to Fine Particulate Matter From Energy Activities vs Frozen Policies, 2023-2035 housands Current Policies Scenarios 28.1 36.8 Mid-range Optimistic Reductions in fine particulate pollution spurred by the Inflation Reduction Act an Infrastructure Law could avoid roughly 17,000-37,000 premature deaths from 2023-2035, saving ~$150-325 billion in economic damages from avoided mortalities alone. Legislation enacted by the 117th Cumulative Avoided Premature Deaths From Exposure to Fine Particulate Matter From Energy A thousands Current Policies Scenarios 17.6 28.1 Conservative Mid-range Optimistic ne Particulate Matter From Energy Activities vs Frozen Policies, 2023-203 Cumulative Avoided Premature Deaths From Exposure to Fine Particulate thousands Cumulative Avoided Premature Deaths From Exposure to Fine Particulate Matter From Energy Activities thousands Premature Deaths From Exposure to Fine Particulate Matter From Energy Activities vs Frozen Policies, 2023-2035 Current Policies Scenarios Reductions in fine particulate pollution spurred by the Inflation Reduction Act and Infrastructure Law could avoid roughly 17,000-37,000 premature deaths from 2023-2035, saving ~$150-325 billion in economic damages from avoided mortalities alone. 19 About REPEAT Project About REPEAT Project 20 nd independent environmental and economic evaluation of federal energy and climate , offering a detailed look at the United States’ evolving energy and climate policy on the path to net-zero greenhouse gas emissions. g and analysis tools coupled with detailed macro-energy system optimization models to atory proposals at politically-relevant spatial resolutions (e.g., state, county, and finer ods used in the Princeton Net-Zero America study. redible information and analysis for broad educational purposes, including as a resource on-makers, and the media. was provided by a grant from the Hewlett Foundation. REPEAT Project played a central role in informing debate, media coverage, and public d and enacted legislation. The project continues to provide regular analysis of pending and progress towards climate goals, and other analysis at repeatproject.org REPEAT Project provides regular, timely and independent environmental and economic evaluation of federal energy and climate policies as they’re proposed and enacted, offering a detailed look at the United States’ evolving energy and climate policy environment and the country’s progress on the path to net-zero greenhouse gas emissions. Approach: we employ geospatial planning and analysis tools coupled with detailed macro-energy system optimization models to rapidly evaluate federal policy and regulatory proposals at politically-relevant spatial resolutions (e.g., state, county, and finer resolutions). This is a refinement of methods used in the Princeton Net-Zero America study. Goal: provide independent, timely, and credible information and analysis for broad educational purposes, including as a resource available publicly for stakeholders, decision-makers, and the media. available publicly for stakeholders, decision-makers, and the media. Funding: funding for the REPEAT Project was provided by a grant from the Hewlett Foundation pact: throughout the 117th Congress, REPEAT Project played a central role in informing debate, media coverage, and public erstanding of the impacts of proposed and enacted legislation. The project continues to provide regular analysis of pending and lized federal regulations, updates on progress towards climate goals, and other analysis at repeatproject.org 21 The REPEAT Project Team Princeton ZERO Lab: Prof. Jesse D. Jenkins (PI), Dr. G Dartmouth College: Prof. Erin Mayfield (co-PI); Binghamton University: Prof. Neha Patankar; Evolved Energy Research: Ryan Jones, Jamil Farbes; Former contributors: Princeton: Dr. Qingyu Xu; Annie Jacobson, Claire Wayner, Aneesha Manocha, Riti Bhandakar, Cady Feng; Montara Mountain Energy: Emily Leslie, Dr. Andrew Pascale. Website development by Hyperobjekt. For more, see repeatproject.org/about t of interests: Jesse D. Jenkins is part owner of DeSolve, LLC, which provides techno-economic analysis and decision support for clean energy technology nd investors. A list of clients can be found at https://www.linkedin.com/in/jessedjenkins. He serves on the advisory boards of Eavor Technologies Inc., a p geothermal technology company, and Rondo Energy, a provider of high-temperature thermal energy storage and industrial decarbonization solutions, s an equity interest in both companies. He also provides policy advisory services to Clean Air Task Force, a non-profit environmental advocacy group, and a technical advisor to MUUS Climate Partners and Energy Impact Partners, both investors in early-stage climate technology companies. 22 Summary of Methods 23 3. Geospatially-resolved downscaling & mapping 3. Geospatially-resolved downscaling & mapping Analysis Framework 3. Geospatially-resolved downscaling & mapping 4. Impact modeling (employment & air pollution) 2. Macro-energy systems modeling EnergyPATHWAYS scenario tool + RIO optimization tool Note that with this revised analysis, we have updated the RIO U.S. model to 27 zones for finer spatial resolution and accuracy 1. Geospatially-resolved inputs 3. Geospatially-re downscaling & m 4. Impact modeli (employment & a 2. Macro-energy systems modeling EnergyPATHWAYS scenario tool + RIO optimization tool Note that with this revised analysis, we have updated the RIO U.S. model to 27 zones for finer spatial resolution and accuracy Analysis Framework 2. Macro-energy systems modeling EnergyPATHWAYS scenario tool + RIO optimization tool Note that with this revised analysis, we have updated the RIO U.S. model to 27 zones for finer spatial resolution and accura 4. Impact modeling (employment & air pollution) Note that with this revised analysis, we have updated the RIO U.S. model to 27 zones for finer spatial resolution and accuracy 24 Scenarios and policies modeled in this repo Frozen Policies (Jan. ‘21) Benchmark – no new policies or regulations after Jan Net-Zero Pathway Benchmark – cost-effective pathway to reduce GHG emissio by 2030 and net-zero by 2050, consistent with President Biden’s climate mitiga Current Policies – three scenarios (‘Conservative’, ‘Mid-range’, and ‘Optimistic Policies, including the Inflation Reduction Act and Infrastructure Investment an The three scenarios are intended to reflect uncertainty about the effectivenes impacts of constraints on supply chains and other rate-limiting factors. Note th macro-economic uncertainties such as the pace of economic growth, variation factors that may impact outcomes. Infrastructure Law Only – reflects the impact of the Infrastructure Investment Inflation Reduction Act policies from the Current Policies scenario above to iso See https://bit.ly/REPEAT-Policies for detailed section-by-section descriptio in each legislation and explanation of treatment in RE 1 – Note that this report does not include the impact of light duty vehicle tailpipe emi late 2022. However, modeled results are very close to compliant with these rules in a d sink assumptions drive modeled CO2 limits Net-Zero Pathway: Non-CO2 and land sink assumptions drive modeled CO2 limits Mt CO2-equivalent1 Year2 CO2 Non-CO2 Total Land sink Total GHGs 1990 5,121 1,424 -939 5,606 2005 6,132 1,418 -854 6,696 2010 5,680 1,435 -807 6,307 2015 5,376 1,441 -752 6,065 2019 5,262 1,420 -768 5,914 2022 4,980 1,350 -777 5,553 2024 4,529 1,257 -795 4,991 2026 3,862 1,187 -812 4,237 2028 3,444 1,119 -830 3,733 2030 3,026 1,071 -848 3,249 2032 2,713 1,072 -851 2,933 2035 2,239 1,067 -855 2,451 2040 1,446 842 -863 1,425 2050 -43 802 -877 -118 1 - CO2-equivalent emiss • The Net-Zero Pathway achieves an overall GHG emissions trajectory that reaches 51% below 2005 emissions levels by 2030 and net-zero anthropogenic GHG emissions by 2050, consistent with the United States’ mid- and long-term climate mitigation goals. • The Net-Zero Pathway achieves an overall GHG emissions trajectory that reaches 51% below 2005 emissions levels by 2030 and net-zero anthropogenic GHG emissions by 2050, consistent with the United States’ mid- and long-term climate mitigation goals. • A limit on total CO2 emissions from energy and industrial activities is explicitly modeled as a constraint in RIO. The emissions limit is based on the total GHG limit less exogenous assumptions about non-CO2 GHG mitigation opportunities and improvements in land carbon sinks described below: • A limit on total CO2 emissions from energy and industrial activities is explicitly modeled as a constraint in RIO. The emissions limit is based on the total GHG limit less exogenous assumptions about non-CO2 GHG mitigation opportunities and improvements in land carbon sinks described below: • The non-CO2 trajectory is based on EPA’s 2019 non-CO2 GHG Marginal Abatement Cost Curve and assumes all mitigation opportunities available at <$100/t CO2-e plus the impacts of the American Innovation and Manufacturing Act of 2020, which requires phase down of HFCs and will result in an additional ~120 MMt CO2e reduction in 2030 and 158 MMt by 2050 as per EPA analysis. Scenarios and policies modeled in this report 21) Benchmark – no new policies or regulations after January 2021 Net-Zero Pathway Benchmark – cost-effective pathway to reduce GHG emissions 50-52% below 2005 levels by 2030 and net-zero by 2050, consistent with President Biden’s climate mitigation goals. Current Policies – three scenarios (‘Conservative’, ‘Mid-range’, and ‘Optimistic’) reflecting the potential impact of Current Policies, including the Inflation Reduction Act and Infrastructure Investment and Jobs Act enacted by the 117th Congress.1 The three scenarios are intended to reflect uncertainty about the effectiveness of IRA and IIJA provisions and the potential impacts of constraints on supply chains and other rate-limiting factors. Note that these scenarios do not reflect broader macro-economic uncertainties such as the pace of economic growth, variation in global commodity prices, and other factors that may impact outcomes. astructure Law Only – reflects the impact of the Infrastructure Investment and Jobs Act alone. This scenario removes all ation Reduction Act policies from the Current Policies scenario above to isolate the impact of IIJA. the impact of the Infrastructure Investment and Jobs Act alone. This scenario removes all m the Current Policies scenario above to isolate the impact of IIJA. tps://bit.ly/REPEAT-Policies for detailed section-by-section descriptions of climate and clean energy related policies in each legislation and explanation of treatment in REPEAT Project modeling. 1 – Note that this report does not include the impact of light duty vehicle tailpipe emissions standards through MY2026 or heavy-duty vehicle soot rule for MY2027+ finalized in ate 2022. However, modeled results are very close to compliant with these rules in all cases. Both will be explicitly treated in subsequent REPEAT Project analysis later this year. 25 1 - CO2-equivalent emissions calculations use IPCC AR4 100 year global warming potential as per EPA Inventory of Greenhouse Gas Emissions and Sinks. 2 - Historical data for 1990-2019 from U.S. EPA Inventory. y: Non-CO2 and land sink assumptions drive modeled CO2 limits thway: Non-CO2 and land sink assumptions drive modeled CO2 limits • The land carbon sink trajectory is based on the Net-Zero America study and increase from ~770 MMt in 2019 to ~850 MMt in 2030 and ~880 MMt in 2050. • The land carbon sink trajectory is based on the Net-Zero America study and increase from ~770 MMt in 2019 to ~850 MMt in 2030 and ~880 MMt in 2050. • Methane emissions associated with changes in oil, natural gas, and coal production are also estimated. As these emissions are calculated after solving RIO under the previously determined CO2 emissions constraint, total net emissions in 2050 are slightly negative (-118 MMt). 1 - CO2-equivalent emissions calculations use IPCC AR4 100 year global warming potential as per EPA Inventory of Greenhouse Gas Emissions and Sinks. 2 - Historical data for 1990-2019 from U.S. EPA Inventory. 26 Methane abatement 1 - We assume covered entities capture all abatement <$50/tCO2e (methane fee is $60/tCO2e at AR4 GWP100) beginning with the EPA 2019 MAC curve, which is then scaled down to reflect the ratio of CH4 emissions reported by GHG Reporting Program (GHGRP) covered entities subject to the IRA methane fee to total EPA inventory for CH4 from Oil & Gas Systems (39%). We then further scaled down estimated abatement opportunities to reflect the ratio between modeled GHGI Baseline and EPA MAC baseline (which is larger and reflects an older edition of the GHGI report). 1 - We assume covered entities capture all abatement <$50/tCO2e (methane fee is $60/tCO2e at AR4 GWP100) beginning with the EPA 2019 MAC curve, which is then scaled down to reflect the ratio of CH4 emissions reported by GHG Reporting Program (GHGRP) covered entities subject to the IRA methane fee to total EPA inventory for CH4 from Oil & Gas Systems (39%). We then further scaled down estimated abatement opportunities to reflect the ratio between modeled GHGI Baseline and EPA MAC baseline (which is larger and reflects an older edition of the GHGI report). Methane abatement Methane emissions under each scenario reflect the following analysis and assumptions: Methane emissions under each scenario reflect the following analysis an Methane emissions under each scenario reflect the following analysis and assumptions: • All scenarios: all scenarios include estimated mitigation from regulations enacted prior to January 2021 as well as the impact of changes in modeled oil, natural gas and coal production volumes. • All scenarios: all scenarios include estimated mitigation from regulations enacted prior to January 2021 in modeled oil, natural gas and coal production volumes. • Infrastructure Law Only: additionally includes an estimated 0.25 MMt CO2-e per year in methane reduct 2022-2030) from plugging abandoned oil and gas wells as funded by Section 40601 of the law (“Orphane and restoration”), based on the RFF report, "Plugging Abandoned Wells: Effects of the Draft Energy Infra y includes an estimated 0.25 MMt CO2-e per year in methane reductions (increasing linearly from ed oil and gas wells as funded by Section 40601 of the law (“Orphaned well site plugging, remediation, report, "Plugging Abandoned Wells: Effects of the Draft Energy Infrastructure Act" (RFF, July 2021). • Infrastructure Law Only: additionally includes an estimated 0.25 MMt CO2-e per year in methane reductions (increasing linearly from 2022-2030) from plugging abandoned oil and gas wells as funded by Section 40601 of the law (“Orphaned well site plugging, remediation and restoration”), based on the RFF report, "Plugging Abandoned Wells: Effects of the Draft Energy Infrastructure Act" (RFF, July 2021). • Current Policies scenarios: additionally include the impacts of the Inflation Reduction Act’s Methane Emissions Reduction Program (Section 60113, aka the “methane fee”), based on oil & gas methane mitigation modeling performed for REPEAT Project by Clean Air Task Force and considering estimated marginal abatement opportunities for covered entities1 available at or below the methane fee level. For documentation of CATF modeling approach, see "Reducing Methane from Oil and Gas: A Path to a 65% Reduction in Sector Emissions" (CATF, December 2020). Note that this analysis accounts for changes in modeled oil and natural gas production volumes across Conservative, Mid-range, and Optimistic scenarios, which impacts available abatement opportunities relative to the baseline. • Net-Zero Pathway: assumes all mitigation opportunities available at <$100/t CO2-e. • Net-Zero Pathway: assumes all mitigation opportunities available at <$100/t CO2-e. 27 Agricultural non-CO2 abatement and land carbon sinks uctions in agriculture related non-CO2 greenhouse gases (CH4 and N2O) and improvements in land carbon sinks are based on marginal tement cost estimates from a meta-study conducted by ICF for the Environmental Defense Fund report "Ambitious Climate Mitigation hways for U.S. Agriculture and Forestry: Vision for 2030” (EDF and ICF, September 2022). Reductions under each scenario reflect the owing analysis and assumptions: • Frozen Policies (Jan. ’21) and Infrastructure Law Only: no additional mitigation relative to baseline from “Gl Gas Emission Projections & Mitigation Potential: 2015-2050” (EPA 2019). • Frozen Policies (Jan. ’21) and Infrastructure Law Only: no additional mitigation relative to baseline from “Global Non-CO2 Greenhouse Gas Emission Projections & Mitigation Potential: 2015-2050” (EPA 2019). • Current Policies: assume additional reductions from the following Inflation Reduction Act programs: • Current Policies: assume additional reductions from the following Inflation Reduction Act programs: cies: assume additional reductions from the following Inflation Reduction A e additional reductions from the following Inflation Reduction Act programs ‐ Section 21001 - Additional Agricultural Conservation Investments or all scenarios, we assume 10% of program budgets for overhead and the remaining 90% allocated across relevant categories of stimated marginal abatement costs from the EDF/ICF (2022) study. Across the Conservative / Mid-range / Optimistic cases, we assume 0% / 20% / 25% of the total technical abatement potential at each price point is available/responsive to incentives, creating variation in he effectiveness of the programs. • Net-Zero Pathway: assumes all mitigation opportunities in the EDF/ICF (2022) study available at <$100/t CO2-e. ro Pathway: assumes all mitigation opportunities in the EDF/ICF (2022) stud • Net-Zero Pathway: assumes all mitigation opportunities in the EDF/ICF ( 28 Treatment of policy incentives for demand-side adoption (other than vehicles) de choices other than vehicles (e.g. heating, cooling, building efficiency), sales traject an vehicles (e.g. heating, cooling, building efficiency), sales trajectories are scenario b For demand-side choices other than vehicles (e.g. heating, cooling, building efficiency For demand-side choices other than vehicles (e.g. heating, cooling, building efficiency), sales trajectories are scenario based in all cases. • Frozen Policies (Jan ‘21): demand-side technology adoption is based on the EIA’s A • Frozen Policies (Jan ‘21): demand-side technology adoption is based on the EIA’s Annual Energy Outlook 2022. Agricultural non-CO2 abatement and land carbon sinks ‐ After funds are exhausted, sales shares stay fixed at the highest level achieved at that time. Agricultural non-CO2 abatement and land carbon sinks • Net-Zero Pathway: assumes an S-curve change in sales patterns that see lower carbon alternatives saturate markets in 2035, comprehensive and sustained effort to achieve economy-wide net-zero GHG at low cost (similar to scenarios from Evolved En Annual Decarbonization Perspectives 2022). • Net-Zero Pathway: assumes an S-curve change in sales patterns that see lower carbon alternatives saturate markets in 2035, consistent with a comprehensive and sustained effort to achieve economy-wide net-zero GHG at low cost (similar to scenarios from Evolved Energy Research’s Annual Decarbonization Perspectives 2022). • Infrastructure Law Only and Current Policies: we allocate funds from demand-side incentives to specific end-use subsectors and assume these fun our way up the S-curve” of adoption for each technology relative to the Net-Zero Pathway as follows: • Infrastructure Law Only and Current Policies: we allocate funds from demand-side incentives to specific end-use subsectors and assume these funds “buy our way up the S-curve” of adoption for each technology relative to the Net-Zero Pathway as follows: ‐ We calculate incremental up-front (capital) cost of all demand-side subsectors in the Net-Zero Pathway vs Frozen Policies (i.e., commercial ventilation, residential building shells, residential heating, etc.). cremental up-front (capital) cost of all demand-side subsectors in the Net-Zero Pathway vs Frozen Policies (Jan. ‘21) scenario al ventilation, residential building shells, residential heating, etc.). ‐ We total all available budgets for incentives for each demand-side subsector (with some judgement applied as to allocation of budgets that apply to multiple sub-sectors; see Policy Worksheet for details). ‐ We reduce effective budgets for all demand-side measures (incl. vehicles) by varying amounts depending on scenario to reflect administration, implementation costs, programmatic inefficiencies, and funding for inframarginal purchases that would have occurred otherwise (a simplifying assumption reflecting that programs are not perfectly efficient in allocating available funds). The program effectiveness ‘haircut’ for each scenario follows: Current Policies (Conservative): 40%; Current Policies (Mid-range) and Infrastructure Law Only: 20%; Current Policies (Optimistic): 10%. ‐ Then we follow the sales curve for the Net-Zero Pathway scenario, using the policy funds to cover any incremental costs of the subsector in the Net-Zero Pathway relative to the Frozen Policies (Jan. ‘21) scenario until the available funds are exhausted. or the Net-Zero Pathway scenario, using the policy funds to cover any incremental costs of the subsector in the Frozen Policies (Jan. ‘21) scenario until the available funds are exhausted. 29 Light duty vehicle transition assumptions all scenarios, we assume 2022-2024 sales shares follow the forecast from the BloombergNEF “Electric Vehicle Outlook 2022” (BNEF EVO22), published prior A, assuming that EV supply chains constrain demand during this period and sales volumes are thus unaffected by vehicle purchase incentives. This esponds to the following EV sales shares1: 4.7% in 2021; 6% in 2022; and 8% in 2023; 11% in 2024. For all scenarios, we assume 2022-2024 sales shares follow the forecast from the BloombergNEF “Electric Vehicle Outlook 2022” (BNEF EV to IRA, assuming that EV supply chains constrain demand during this period and sales volumes are thus unaffected by vehicle purchase inc corresponds to the following EV sales shares1: 4.7% in 2021; 6% in 2022; and 8% in 2023; 11% in 2024. After 2024, sales trajectories vary across scenarios as follows: n the U.S. lag behind BNEF EVO22 by two years, assuming that IIJA’s impact accelerates changes in sales shares. This ZEV sales shares: 14% in 2026; 20% in 2028; 31% in 2030; 42% in 2032; 49% in 2035. • Frozen Policies: sales shares in the U.S. lag behind BNEF EVO22 by two years, assuming that IIJA’s impact accelerates changes in sales shares. This corresponds to the following ZEV sales shares: 14% in 2026; 20% in 2028; 31% in 2030; 42% in 2032; 49% in 2035. rozen Policies: sales shares in the U.S. lag behind BNEF EVO22 by two years, assuming that IIJA’s impact accelerates changes in sales shares. This orresponds to the following ZEV sales shares: 14% in 2026; 20% in 2028; 31% in 2030; 42% in 2032; 49% in 2035. nfrastructure Law Only, and Current Policies (Conservative): U.S. sales shares follow shares projected in BNEF EVO22. That is, these scenarios assume no ncremental increase in sales from IIJA or IRA incentives. This corresponds to the following EV sales shares: 20% in 2026; 31% in 2028; 42% in 2030; 49% in 2032; 59% in 2035. Note additional 2% of light vehicle sales are hydrogen FCVs in 2030 and 4% in 2035. • Infrastructure Law Only, and Current Policies (Conservative): U.S. sales sha incremental increase in sales from IIJA or IRA incentives. This corresponds t 2032; 59% in 2035. 2 - 18 states + DC follow CARB low-emissions tailpipe standards now: CA, NY, MA, VT, ME, PA, CT, RI, WA, OR, NJ, MD, DE, CO, MN, NV, VA, NM plu 1 - Overall EV sales shares between light autos and light trucks reflect EIA Annual Outlook 2022 sales projections between the two categories, skew two to one 30 1 - Overall EV sales shares between light autos and light trucks reflect EIA Annual Outlook 2022 sales projections between the two categories, skewing toward light trucks at roughly two-to-one. Light duty vehicle transition assumptions Note additional 2% of light vehicle sales are hydrogen FC Current Policies (Mid-range): California and 17 states + DC that historically follow the California Zero Emissions Vehicle (ZEV) program2 are assumed to meet California’s Advanced Clean Cars 2 standard, which requires zero emissions vehicle sales to reach 100% by 2035, while all other states are assumed to follow he Conservative scenario sales shares above. We linearly interpolate the sales shares from 2024-2035, resulting in the following national EV sales shares: 22% in 2026; 36% in 2028; 52% in 2030; 66% in 2032; and 82% in 2035. Note an additional 2% of light sales are hydrogen FCVs in 2030 and 4% in 2035. t Policies (Optimistic): sales shares across the entire country are assumed to meet California’s Advanced Clean Cars 2 standard, which requires zero ons vehicle sales to reach 100% by 2035, resulting in the following national EV sales shares: 27% in 2026; 46% in 2028; 65% in 2030; 81% in 2032; and 2035. Note an additional 2% of light sales are hydrogen FCVs in 2030 and 5% in 2035. For purposes of calculating the cost of adoption and consumer energy expenditures, we phase in the value of the light duty vehicle tax credits over the 2025-2032 period, assuming the automotive sector progressively reorients supply chains to meet the battery and materials sourcing requirements of the IRA EV personal income tax credit. We assume 70% of vehicle sales receive either the personal or business tax credit (including leases and fleet purchases) and assume the following average credit values: $5,000 in 2024-2026 period; $6,000 in 2027-2028 period; and $7,500 in 2029-2032 period. doption and consumer energy expenditures, we phase in the value of the light duty vehicle tax credits over the otive sector progressively reorients supply chains to meet the battery and materials sourcing requirements of the IRA EV 70% of vehicle sales receive either the personal or business tax credit (including leases and fleet purchases) and assume 000 in 2024-2026 period; $6,000 in 2027-2028 period; and $7,500 in 2029-2032 period. 30 Medium and heavy duty vehicle transition assumptions For all scenarios, we assume 2022-2024 sales shares follow the forecast from the BloombergNEF “Electric Vehicle Outlook 2022” (BNEF EVO22), published prior to IRA, assuming that supply chains constrain demand during this period and sales volumes are thus unaffected by vehicle purchase incentives. This corresponds to the following ZEV sales shares: 0.2% in 2021; 0.4% in 2022; 0.9% in 2023; 1.8% in 2024. After 2024, sales trajectories vary across scenarios as follows: After 2024, sales trajectories vary across scenarios as follows: • Frozen Policies: sales shares in the U.S. lag behind BNEF EVO22 by two years, assuming that IIJA’s impact accelerates change corresponds to the following ZEV sales shares: 1.8% in 2026; 4.9% in 2028; 9.3% in 2030; 16% in 2032; 25% in 2035. • Infrastructure Law Only, and Current Policies (Conservative): U.S. sales shares follow shares projected in BNEF EVO22 increase in sales from IRA incentives. This corresponds to the following ZEV sales shares: 4.9% in 2026; 9.3% in 2028; 1 2035. • Current Policies (Mid-range): IRA incentives drive rapid growth in this scenario across all states, driving sales share increases for zero-e HDV on a trajectory roughly inline with California’s Advanced Clean Trucks standard. Once incentives roll off, sales shares decline slight with 17 states + DC accelerate again by 2036 on a pace to reach 100% ZEV in 2045. All other states follow the same increases in sales s the Conservative scenario sales shares above. US-wide ZEV sales shares are: 16% in 2026; 38% in 2028; 58% in 2030; 67% in 2032; and ncentives drive rapid growth in this scenario across all states, driving sales share increases for zero-emission MDV and with California’s Advanced Clean Trucks standard. Once incentives roll off, sales shares decline slightly, and CA along n by 2036 on a pace to reach 100% ZEV in 2045. All other states follow the same increases in sales share by year as ares above. US-wide ZEV sales shares are: 16% in 2026; 38% in 2028; 58% in 2030; 67% in 2032; and 53% in 2035. • Current Policies (Mid-range): IRA incentives drive rapid growth in this scenario across all states, driving sales share increases for zero-emission MDV and HDV on a trajectory roughly inline with California’s Advanced Clean Trucks standard. Medium and heavy duty vehicle transition assumptions Based on assumed vehicle cost inputs across MDV and HDV fleets, this results in average incentives of: $35,000 in 2024-2026 period; $33,000 in 2027-2028 period; and $30,500 in 2029-2032 period 31 3 0 200 400 600 800 1000 1200 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 Maximum annual onshore wind capacity additions gigawatts/year (GW/year) Conservative Mid-range Optimistic 0 200 400 600 800 1000 1200 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 Maximum annual utility-scale solar PV capacity additions gigawatts/year (GW/year) Conservative Mid-range Optimistic ted historical maximum additions for 2023 as reported by DOE “Wind Technologies Market Report, rgy Industrial Association “Solar Market Insights Report, Q3 2022 edition.” 0 200 400 600 800 1000 1200 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 Maximum annual onshore wind capacity additions gigawatts/year (GW/year) Conservative Mid-range Optimistic 0 200 400 600 800 1000 1200 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 Maximum annual onshore wind capacity additions gigawatts/year (GW/year) Conservative Mid-range Optimistic Onshore wind and solar annual build constraints: For all Current Policies scenarios, we assume onshore wind and utility-scale solar additions are fixed at 11 GW of wind and 6 GW of solar in 2022 and 17 GW of wind and 16 GW of solar in 2023.1 All other scenarios have the same 2022 annual build and no minimum annual build in 2023. Beyond 2023, annual build constraints increase at different rates by scenario. These constraints are intended to approximate several challenges that are difficult to model explicitly but may limit renewable energy deployment, including the ability to site and permit projects at requisite pace and scale, expand supply chains, interconnect generating capacity, and hire and train the expanded energy workforce to build these projects. 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 • Current Policies (Conservative): maximum annual additions for both onshore wind and utility-scale solar grow at a compound annual growth rate (CAGR) of 17% from 2023-2032. From 2033-2035 this drops to 10%, and for 2036 and after is limited to 5% as incentives sunset and the markets for these technologies reach maturity at large scale. 0 200 400 600 800 1000 1200 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 Maximum annual utility-scale solar PV capacity additions gigawatts/year (GW/year) Conservative Mid-range Optimistic 0 200 400 600 800 1000 1200 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 Maximum annual utility-scale solar PV capacity additions gigawatts/year (GW/year) Conservative Mid-range Optimistic • Frozen Policies, Infrastructure Law Only, and Current Policies (Mid-range): maximum annual additions for both renewable technologies grow at a CAGR of 20% from 2023-2032. The same assumptions as the Conservative scenario are applied after 2032 (2033-2035 at 10% CAGR, after 2035 at 5%). • Frozen Policies, Infrastructure Law Only, and Current Policies (Mid-range): maximum annual additions for both renewable technologies grow at a CAGR of 20% from 2023-2032. The same assumptions as the Conservative scenario are applied after 2032 (2033-2035 at 10% CAGR, after 2035 at 5%). Medium and heavy duty vehicle transition assumptions Once incentives roll off, sales shares decline slightly, and CA along with 17 states + DC accelerate again by 2036 on a pace to reach 100% ZEV in 2045. All other states follow the same increases in sales share by year as the Conservative scenario sales shares above. US-wide ZEV sales shares are: 16% in 2026; 38% in 2028; 58% in 2030; 67% in 2032; and 53% in 2035. • Current Policies (Optimistic): sales shares across the entire country are assumed to roughly follow the California’s Advanced Clean Trucks standard. Afte 2032 all states follow the same trajectory as states on the CA ZEV trajectory in the Mid-range scenario above achieving 100% ZEV by 2045. Through 2035 sales shares are identical to the Mid-range scenario, they begin to diverge in subsequent years. • Current Policies (Optimistic): sales shares across the entire country are assumed to roughly follow the California’s Advanced Clean Trucks standard. Afte 2032 all states follow the same trajectory as states on the CA ZEV trajectory in the Mid-range scenario above achieving 100% ZEV by 2045. Through 2035 sales shares are identical to the Mid-range scenario, they begin to diverge in subsequent years. Note: For purposes of calculating the cost of adoption, we assume commercial vehicles capture the lesser of $40,000 or 30% of the capital cost of the ZEV from 2023-2032, assuming the incentives will be a critical factor in vehicle purchases, and vehicle prices will decline over time. Onshore wind and solar annual build constraints: • Current Policies (Optimistic): starting in 2022, the injection constraint rises from 0 to 75 MMt/yr in 2024, 300 MMt/yr in 2030, 600 MMt/yr in 2035, 1,500 MMt/yr in 2040 and 4,000 MMt/yr in 2050. All scenarios share the same assumption on maximum annual deployment of electrolyzer capacity, starting at 1.3 MMt /yr of hydrogen production capacity in 2023, and growing at 20% compound annual growth in subsequent years. 0 25 50 75 100 125 150 175 200 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 Maximum annual electrolyzer capacity additions million metric tons of H2 output per year (MMt/year) 0 500 1000 1500 2000 2500 3000 3500 4000 4500 2022 2024 2030 2035 2040 2050 Maximum annual CO2 storage injection rate million metric tons per year (MMt/year) Conservative Mid-range Optimistic Maximum annual CO2 storage injection rate million metric tons per year (MMt/year) Constraints on technology growth rates 0 500 1000 1500 2000 2500 3000 3500 4000 4500 2022 2024 2030 2035 2040 2050 g j million metric tons per year (MMt/year) Conservative Mid-range Optimistic Onshore wind and solar annual build constraints: • Current Policies (Optimistic): from 2023-2032 maximum annual additions for onshore wind and utility-scale solar grow at a CAGR of 30%. The same post-2032 assumptions as the Conservative scenario apply. 0 200 2022 2023 2024 2025 2026 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 1 - Maximum wind capacity additions reflect projected additions for 2022 and reported historical maximum additions for 2023 as reported by DOE “Wind Technologies Market Report, 2022 Edition.” Maximum solar capacity additions in 2022 and 2023 reflect Solar Energy Industrial Association “Solar Market Insights Report, Q3 2022 edition.” reflect projected additions for 2022 and reported historical maximum additions for 2023 as reported by DOE “Wind Technologies Market Report, ity additions in 2022 and 2023 reflect Solar Energy Industrial Association “Solar Market Insights Report, Q3 2022 edition.” 32 Constraints on technology growth rates Ramp-up of CO2 injection and Electrolysis constraints: Each scenario incorporates different assumptions about the maximum annual growth rate in infrastructure to enable CO2 injection into geological formations. Development of network infrastructure for CO2 transport as well as CO2 storage basins and injections wells are likely to be binding constraints on the growth of carbon capture and storage, at least in initial years. These constraints are intended to proxy for variations in the potential scale-up of this enabling infrastructure for CO2 capture, transport, and storage and vary by scenario as follows: • Current Policies (Conservative): the injection constraint is 5 MMt/yr in 2024, rising to 100 MMt/yr in 2030, 180 MMt/yr in 2035, 500 MMt/yr in 2040 and 2,000 MMt/yr in 2050. Much of this constraint is taken up by the exogenous assumptions about industrial carbon capture such as gas processing, cement, and iron and steel.1 • Frozen Policies, Infrastructure Law Only, and Current Policies (Mid-range): the injection constraint is 50 MMt/yr in 2024, 200 MMt/yr in 2030, 400 MMt/yr in 2035, 1,000 MMt/yr in 2040 and 4,000 MMt/yr in 2050. Ramp-up of CO2 injection and Electrolysis constraints: Each scenario incorporates different assumptions about the maximum annual growth rate in infrastructure to enable CO2 injection into geological formations. Development of network infrastructure for CO2 transport as well as CO2 storage basins and injections wells are likely to be binding constraints on the growth of carbon capture and storage, at least in initial years. These constraints are intended to proxy for variations in the potential scale-up of this enabling infrastructure for CO2 capture, transport, and storage and vary by scenario as follows: • Current Policies (Conservative): the injection constraint is 5 MMt/yr in 2024, rising to 100 MMt/yr in 2030, 180 MMt/yr in 2035, 500 MMt/yr in 2040 and 2,000 MMt/yr in 2050. Much of this constraint is taken up by the exogenous assumptions about industrial carbon capture such as gas processing, cement, and iron and steel.1 0 25 50 75 100 125 150 175 200 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 Maximum annual electrolyzer capacity additions million metric tons of H2 output per year (MMt/year) Maximum annual electrolyzer capacity additions million metric tons of H2 output per year (MMt/year) • Frozen Policies, Infrastructure Law Only, and Current Policies (Mid-range): the injection constraint is 50 MMt/yr in 2024, 200 MMt/yr in 2030, 400 MMt/yr in 2035, 1,000 MMt/yr in 2040 and 4,000 MMt/yr in 2050. • Frozen Policies, Infrastructure Law Only, and Current Policies (Mid-range): the injection constraint is 50 MMt/yr in 2024, 200 MMt/yr in 2030, 400 MMt/yr in 2035, 1,000 MMt/yr in 2040 and 4,000 MMt/yr in 2050. • Current Policies (Optimistic): starting in 2022, the injection constraint rises from 0 to 75 MMt/yr in 2024, 300 MMt/yr in 2030, 600 MMt/yr in 2035, 1,500 MMt/yr in 2040 and 4,000 MMt/yr in 2050. All scenarios share the same assumption on maximum annual deployment of electrolyzer capacity, starting at 1.3 MMt /yr of hydrogen production capacity in 2023, and growing at 20% compound annual growth in subsequent years. All scenarios share the same assumption on maximum annual deployment of electrolyzer capacity, starting at 1.3 MMt /yr of hydrogen production capacity in 2023, and growing at 20% compound annual growth in subsequent years. Ramp-up of CO2 injection and Electrolysis constraints: 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 1 - The CO2 injection constraint only turns out to be binding in the Conservative scenario (through 2035). 1 - The CO2 injection constraint only turns out to be binding in the Conservative scenario (through 2035). 33 Treatment of tax credit ‘bonuses’ and ‘adders’ 34 Treatment of the Section 45X Advanced Manufa The Section 45X Advanced Manufacturing Production Tax Credit provides qualif tax credits for the manufacture of: solar PV modules and components; inverter and service vessels; battery cells, packs/modules, and electroactive materials; a tax credits have the potential to substantially reduce the cost of domestic manu the cost of competing imports.1 In this modeling, we assume the 45X tax credits help increase domestic conten deployment rates. However, we assume no passthrough of the 45X tax credit t assumption that demand for domestic content will exceed supply through 2035 competitively against available imports without affect marginal project prices. 1 - See Min et al. (2023), “Effects of Renewable Energy Provisions o Sustainable Transition Lab and Princeton ZERO Lab, June 2023, availab We assume the 45X manufacturing tax credit and the domestic content ‘adder’ available to clean electricity projects will drive strong demand for U.S. manufactured solar PV and wind turbine components that exceeds available supply, inducing investment to steadily increase U.S. manufacturing capacity. For the purposes of modeled employment in energy supply related sectors, we assume gradual increases in domestic content shares for solar and wind projects as depicted in the figure at right.1 ent of the Section 45X Advanced Manufactur Section 45X Advanced Manufacturing Production Tax Credit provides qualifying manufacturers located in the United States with income credits for the manufacture of: solar PV modules and components; inverters; wind turbines and components; offshore wind foundations service vessels; battery cells, packs/modules, and electroactive materials; and the production and processing of critical minerals. These credits have the potential to substantially reduce the cost of domestic manufacturing and enable U.S. manufacturers to price at or below f i i 1 35 is modeling, we assume the 45X tax credits help increase domestic content shares and help expand supply chains to meet modeled oyment rates. However, we assume no passthrough of the 45X tax credit to modeled project prices, which is consistent with the mption that demand for domestic content will exceed supply through 2035 and thus domestic manufacturers will price final products petitively against available imports without affect marginal project prices. 1 - See Min et al. (2023), “Effects of Renewable Energy Provisions of the Inflation Reduction Act on Technology Costs, Materials Demand, and Labor,” Dartmouth College Sustainable Transition Lab and Princeton ZERO Lab, June 2023, available at https://zenodo.org/record/8027939 Domestic Content Shares for Wind & Solar PV1 percent assume the 45X manufacturing tax credit and the domestic content er’ available to clean electricity projects will drive strong demand for manufactured solar PV and wind turbine components that exceeds able supply, inducing investment to steadily increase U.S. manufacturing acity. For the purposes of modeled employment in energy supply ted sectors, we assume gradual increases in domestic content shares olar and wind projects as depicted in the figure at right.1 In this modeling, we assume the 45X tax credits help increase domestic content shares and help expand supply chains to meet modeled deployment rates. However, we assume no passthrough of the 45X tax credit to modeled project prices, which is consistent with the assumption that demand for domestic content will exceed supply through 2035 and thus domestic manufacturers will price final products competitively against available imports without affect marginal project prices. Treatment of tax credit ‘bonuses’ and ‘adders’ eral tax credits in the Inflation Reduction Act, notably incentives for clean electricity, energy storage, carbon capture, storage & use, and ting nuclear power plants in Title I, Subtitle D, require qualified projects to meet certain prevailing wage and minimum apprenticeship uirements to claim the full value of the tax credit, referred to in the law as the ‘bonus’ value (see bit.ly/REPEAT-Policies for details). Failure meet these requirements reduces the credit value to a ‘base’ value worth 20% of the ‘bonus’ value. Throughout our modeling, we assume ligible projects meet the requirements for the ‘bonus’ credit value, as projects that fail to do so will be significantly disadvantaged nomically relative to projects that do meet these requirements and are thus unlikely to be competitive. uction and investment tax credits available to new clean electricity generation and energy storage projects also offer ‘adders’ that the value of the production tax credit by 10% and the investment tax credit by 10 percentage points if (a) projects meet certain content requirements and/or (b) projects are built in certain ‘energy communities’ defined in the law (both adders can be claimed if meets both requirements). In this analysis, neither the domestic content adder or energy communities adder are considered to the marginal projects represented in our modeling, with the exception of new nuclear power plants. We assume demand for content will exceed supply through 2035 and thus projects receiving the domestic content adder are likely to be inframarginal and fect modeled equilibrium outcomes. For nuclear power plants, we assume the most competitive projects would be built at d coal power plant sites and thus receive the bonus 10% energy communities adder (for a 40% ITC value), although limited nuclear s built in practice in any Current Policies cases. For other technologies, the implicit assumption is that marginal projects in each ne are not located in an energy community, so the energy community bonus credit is disregarded. This is a notably conservative on given the breadth of energy communities as defined by the law. In forthcoming revisions of our work, we will incorporate recent on energy communities into our geospatial data sets and adjust our clean electricity resource supply (cost) curves accordingly. ent of the Section 45X Advanced Manufactur Domestic Content Shares for Wind & Solar PV1 ume the 45X tax credits help increase domestic content shares and help expand supply chains to meet modeled ever, we assume no passthrough of the 45X tax credit to modeled project prices, which is consistent with the d for domestic content will exceed supply through 2035 and thus domestic manufacturers will price final products il bl i i h ff i l j i Domestic Content Shares for Wind & Solar PV1 percent Domestic Content Shares for Wind & Solar PV1 percent We assume the 45X manufacturing tax credit and the domestic content ‘adder’ available to clean electricity projects will drive strong demand for U.S. manufactured solar PV and wind turbine components that exceeds available supply, inducing investment to steadily increase U.S. manufacturing capacity. For the purposes of modeled employment in energy supply related sectors, we assume gradual increases in domestic content shares for solar and wind projects as depicted in the figure at right.1 1 - See Min et al. (2023), “Effects of Renewable Energy Provisions of the Inflation Reduction Act on Technology Costs, Materials Demand, and Labor,” Dartmouth College Sustainable Transition Lab and Princeton ZERO Lab, June 2023, available at https://zenodo.org/record/8027939 1 - See Min et al. (2023), “Effects of Renewable Energy Provisions of the Inflation Reduction Act on Technology Costs, Materials Demand, and Labor,” Dartmouth College Sustainable Transition Lab and Princeton ZERO Lab, June 2023, available at https://zenodo.org/record/8027939 35 1 – These assumptions include variation in the portion of program budgets or tax credit incentives effectively incentivizing deployment or adoption vs administrative overhead, transaction costs, or funding for inframarginal purchases. See other methods slides and section-by-section explanation of uncertainty treatment for each policy at https://bit.ly/REPEAT-Policies A note on interpretation of modeled results Optimization modeling used in this work assumes rational economic behavior from all actors. The modeling also has limited ‘frictions’ on deployment of infrastructure (e.g., power generation or transmission capacity), scale-up of industry supply chains (e.g., wind and solar), or consumer adoption of alternative products (e.g., EVs, heat pumps). Modeling results should thus be interpreted primarily as indications of the relative alignment of economic incentives as a result of policy changes. In other words, these results indicate what decisions make good economic sense for consumers and businesses to make. This is likely a necessary condition, but whether or not actors make such decisions in the real world depends on many factors we are unable to explicitly model. Real world outcomes will also contend with various non cost related challenges that may slow the pace of change relative to modeled resul changes. In other words, these results indicate what decisions make good economic sense for consumers and businesses to make. This is likely a necessary condition, but whether or not actors make such decisions in the real world depends on many factors we are unable to explicitly model. world outcomes will also contend with various non-cost related challenges that may slow the pace of change relative to modeled results, uding the ability to site and permit projects at requisite pace and scale, expand electricity transmission and CO2 transport and storage to ommodate new generating capacity, and hire and train the expanded energy workforce to build these projects. These non-cost related challenges are roughly approximated in our modeling via exogenous assumptions constraining the scale-up of various technologies and modulating the assumed effectiveness of various policy programs, which vary across Conservative, Mid-range and Optimistic variants of the Current Policies scenarios.1 n-cost related challenges are roughly approximated in our modeling via exogenous assumptions constraining the scale-up of echnologies and modulating the assumed effectiveness of various policy programs, which vary across Conservative, Mid-range mistic variants of the Current Policies scenarios.1 ally, modeled outcomes reflect a least-cost optimization. There are likely many alternative outcomes with similar costs within a few of these outcomes that may offer advantages in terms of other important outcomes related to the distribution of costs and benefits d with energy systems. Various stakeholders may prefer one or more of these alternative portfolios to the outcomes presented here. hould interpret modeled results accordingly. 1 – These assumptions include variation in the portion of program budgets or tax credit incentives effectively incentivizing deployment or adoption vs administrative overhead, transaction costs, or funding for inframarginal purchases. See other methods slides and section-by-section explanation of uncertainty treatment for each policy at https://bit.ly/REPEAT-Policies A note on interpretation of modeled results 1 – These assumptions include variation in the portion of program budgets or tax credit incentives effectively incentivizing deployment or adoption vs administrative overhead, transaction costs, or funding for inframarginal purchases. See other methods slides and section-by-section explanation of uncertainty treatment for each policy at https://bit.ly/REPEAT-Policies 36 Results 37 37 Greenhouse Gas Emissions Greenhouse Gas Emissions 38 7 Historical and Modeled Net U.S. Greenhouse Gas Emissions (Including Land Carbon Sinks) billion metric tons CO2-equivalent (Gt CO2-e)1 2 3 4 5 6 7 Historical emissions2 Modeled emissions Historical and Modeled Net U.S. Greenhouse Gas Emissions (Including Land Carbon Sinks) billion metric tons CO2-equivalent (Gt CO2-e)1 Infrastructure Law (IIJA) only: ~4.8 billion tons in 2030 (~28% below 2005)3 Net-Zero Pathway ~3.3 billion tons in 2030 (51% below 2005) Current Policies, including the Inflation Reduction Act (IRA): ~4.0-4.2 billion tons in 2030 (~37%-41% below 2005)3 2030 target: ≥50% below 2005 2021 emissions2: ~5.6 billion tons 2005 emissions2: ~6.7 billion tons Frozen Policies (Jan. ‘21) ~4.8 billion tons in 2030 (~28% below 2005) 2 3 4 5 6 7 Historical emissions2 Modeled emissions Historical and Modeled Net U.S. Greenhouse Gas Emissions (Including Land Carbon Sinks) billion metric tons CO2-equivalent (Gt CO2-e)1 Infrastructure Law (IIJA) only: ~4.8 billion tons in 2030 (~28% below 2005)3 Net-Zero Pathway ~3.3 billion tons in 2030 (51% below 2005) Current Policies, including the Inflation Reduction Act (IRA): ~4.0-4.2 billion tons in 2030 (~37%-41% below 2005)3 2030 target: ≥50% below 2005 2021 emissions2: ~5.6 billion tons 2005 emissions2: ~6.7 billion tons Frozen Policies (Jan. ‘21) ~4.8 billion tons in 2030 (~28% below 2005) Legislation enacted by the 117th Congress could: • roughly double the pace of annual U.S. decarbonization to ~4%/year. • cut annual emissions in 2030 by ~0.5-0.8 billion metric tons relative to the Frozen Policies scenario. • get the U.S. to ~37-41% below 2005 historical GHG emissions (vs national target of 50-52%) • reduce cumulative GHG emissions by about 3.4-5.6 billion tons over the next decade (2023-2032). 3 4 5 6 Historical emissions2 Modeled emissions 2030 target: ≥50% below 2005 2021 emissions2: ~5.6 billion tons 2005 emissions2: ~6.7 billion tons Frozen Policies (Jan. ‘21) ~4.8 billion tons in 2030 (~28% below 2005) Legislation enacted by the 117th 2021 emissions2: ~5.6 billion tons 6 • roughly double the pace of annual U.S. decarbonization to ~4%/year. Infrastructure Law (IIJA) only: ~4.8 billion tons in 2030 (~28% below 2005)3 • cut annual emissions in 2030 by ~0.5-0.8 billion metric tons relative to the Frozen Policies scenario. • get the U.S. to ~37-41% below 2005 historical GHG emissions (vs national target of 50-52%) • get the U.S. to ~37-41% below 2005 historical GHG emissions (vs national target of 50-52%) Current Policies, including the Inflation Reduction Act (IRA): ~4.0-4.2 billion tons in 2030 (~37%-41% below 2005)3 2030 target: ≥50% below 2005 Net-Zero Pathway ~3.3 billion tons in 2030 (51% below 2005) 0 021 estimate from February 2023 draft EPA Inventory. nd freight miles traveled due to surface transportation, rail, and transit investments in IIJA. According to the Georgetown Climate Center, emissions impact of these changes depend heavily on state implementation of funding o +25 Mt/yr change in CO2 emissions from transportation in 2030. nts in IIJA. According to the Georgetown Climate Center, emissions impact of these changes depend heavily on state implementation of funding r and freight miles traveled due to surface transportation, rail, and transit investments in IIJA. According to the Georgetown Climate Center, emissio r to +25 Mt/yr change in CO2 emissions from transportation in 2030. 39 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 eled Net U.S. Greenhouse Gas Emissions (Including Land Carbon Sinks) metric tons CO2-equivalent (Gt CO2-e)1 Net-Zero Pathway Current Policies, incl. IRA2 quivalent emissions calculations use IPCC AR4 100 year global warming potential as per EPA Inventory of Greenhouse Gas Emissions and Sinks. All values should be regarded as approximate given uncertainty in future outcomes. led emissions exclude any changes in passenger and freight miles traveled due to surface transportation, rail, and transit investments in IIJA. According to the Georgetown Climate Center, emissions impact of these changes depend heavily on state implementation of funding which could result in anywhere from -14 Mt/yr to +25 Mt/yr change in CO2 emissions from transportation in 2030. Frozen Policies (Jan. ‘21) Despite historic progress, U.S. emissions under Current Policies scenarios still fall about 0.6-0.9 GtCO2-e per year short of the country’s 2030 climate goals. 4.0-4.2 Gt CO2-e 4.8 Gt CO2-e 3.3 Gt CO2-e 3.1-3.6 GtCO2-e 4.5 Gt CO2-e 2.5 Gt CO2-e 2030 target: ≥50% below 2005 6 Modeled Net U.S. Legislation enacted by the 117th Greenhouse Gas Emissions (Including Land Carbon Sinks) billion metric tons CO2-equivalent (Gt CO2-e)1 6 Modeled Net U.S. Greenhouse Gas Emissions (Including Land Carbon Sinks) billion metric tons CO2-equivalent (Gt CO2-e)1 6 Despite historic progress, U.S. emissions under Current Policies scenarios still fall about 0.6-0.9 GtCO2-e per year short of the country’s 2030 climate goals. 5 2030 target: ≥50% below 2005 2030 target: ≥50% below 2005 ssions and Sinks. All values should be regarded as approximate given uncertainty in future outcomes. g potential as per EPA Inventory of Greenhouse Gas Emissions and Sinks. All values should be regarded as approximate given uncertainty in future outcomes. veled due to surface transportation, rail, and transit investments in IIJA. According to the Georgetown Climate Center, emissions impact of these changes depend heavily on state implementation of funding in CO2 emissions from transportation in 2030. emissions calculations use IPCC AR4 100 year global warming potential as per EPA Inventory of Greenhouse Gas Emissions and Sinks. All values should be regarded as approximate given uncertainty in future outcomes. ons exclude any changes in passenger and freight miles traveled due to surface transportation, rail, and transit investments in IIJA. According to the Georgetown Climate Center, emissions impact of these changes depend heavily on state implementation of funding uld result in anywhere from -14 Mt/yr to +25 Mt/yr change in CO2 emissions from transportation in 2030. l values should be regarded as approximate given uncertainty in future outcomes. ording to the Georgetown Climate Center, emissions impact of these changes depend heavily on state implementation of funding g pp g y nts in IIJA. According to the Georgetown Climate Center, emissions impact of these changes depend heavily on state implementation of funding 40 Modeled Net U.S. Greenhouse Gas Emissions (Including Land Carbon Sinks) percent below 2005 historical emissions1 Modeled Net U.S. Greenhouse Gas Emissions (Including Land Carbon Sinks) ercent below 2005 historical emissions1 022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 eled Net U.S. Greenhouse Gas Emissions (Including Land Carbon Sinks) below 2005 historical emissions1 Net-Zero Pathway Current Policies, incl. IRA historical net U.S. greenhouse gas emissions were 6,686 million metric tons of CO2-equivalent (EPA Inventory of Greenhouse Gas Emissions and Sinks). CO2-equivalent emissions calculations use IPCC AR4 100 year global warming as per EPA Inventory of Greenhouse Gas Emissions and Sinks. 2030 target: ≥50% below 2005 Frozen Policies (Jan. Legislation enacted by the 117th ‘21) Under Current Policies scenarios, U.S. emissions in 2030 fall to ~37-41% below 2005 emissions levels, falling short of the 50-52% goal set by President Biden. 37-41% below 2005 46-53% below 2005 29% below 2005 33% below 2005 51% below 2005 63% below 2005 Under Current Policies scenarios, U.S. emissions in 2030 fall to ~37-41% below 2005 emissions levels, falling short of the 50-52% goal set by President Biden. 2-equivalent (EPA Inventory of Greenhouse Gas Emissions and Sinks). CO2-equivalent emissions calculations use IPCC AR4 100 year global warming 41 Difference in Sectoral Emissions Under Current Policies vs Frozen Policies as of January 2021 million metric tons CO2-equivalent (Mt CO2-e)1,2 -870 -1140 -520 -600 -770 -1330 Conservative Mid-range Optimistic Difference in Sectoral Emissions Under Current Policies vs Frozen Policies as of January 2021 million metric tons CO2-equivalent (Mt CO2-e)1,2 • Changes in the transportation sector (e.g. electrification) and power sector (e.g. renewable energy deployment) are responsible for roughly half of all emissions reductions across all Current Policies scenarios. • Yet IRA delivers emissions reductions across all major emitting sectors of the economy including industry, buildings, and agricultural and forestry lands. • Changes in the transportation sector (e.g. electrification) and power sector (e.g. renewable energy deployment) are responsible for roughly half of all emissions reductions across all Current Policies scenarios. • Changes in the transportation sector (e.g. electrification) and power sector (e.g. renewable energy deployment) are responsible for roughly half of all emissions reductions across all Current Policies scenarios. • Yet IRA delivers emissions reductions across all major emitting sectors of the economy including industry, buildings, and agricultural and forestry lands. 2 - Modeled emissions exclude any changes in passenger and freight miles traveled due to surface transportation, rail, and transit investments in IIJA. According to the Georgetown Climate Center, emissions impact of these changes depend heavily on state implementation of funding from IIJA, which could result in anywhere from -14 Mt/yr to +25 Mt/yr change in CO2 emissions from transportation in 2030. 2 - Modeled emissions exclude any changes in passenger and freight miles traveled due to surface transportation, rail, and transit investments in IIJA. According to the Georgetown Climate Center, emissions impact of these changes depend heavily on state implementation of funding from IIJA, which could result in anywhere from -14 Mt/yr to +25 Mt/yr change in CO2 emissions from transportation in 2030. Legislation enacted by the 117th 42 Difference in Sectoral Emissions Under Current Policies vs Net-Zero Pathway million metric tons CO2-equivalent (Mt CO2-e)1,2 +1000 +1180 +900 +910 +750 +710 The Net-Zero Pathway reaches 51% below 2005 emissions in 2030 or nearly 0.1 GtCO2-e greater reduction than required to reach 50% below 2005 levels, leaving a ~0.6-0.9 GtCO2-e/y gap between Current Policies trajectory and U.S. 2030 climate goals. Conservative Mid-range Optimistic Difference in Sectoral Emissions Under Current Policies vs Net-Zero Pathway million metric tons CO2-equivalent (Mt CO2-e)1,2 Optimistic Notes: See additional Notes 1-2 on prior page 3 - While transportation electrification under IRA scenarios largely aligns with the Net-Zero Pathway, accelerated improvements in internal combustion vehicle fuel efficiency could further reduce transport sector emissions by ~10 Mt/yr in 2030 and ~10-30 Mt/yr in 2035. 4 – In Current Policies scenarios, solar capacity additions are already constrained through 2035 and wind capacity additions are close to constraints. Additional reductions in coal-fired generation would therefore likely be compensated primarily by increases in gas-fired generation. Displacing ~400-450 TWh of remaining coal generation in 2030 with gas CCGTs would result in a net reduction of ~0.2 GtCO2-e/year. Deeper reductions could potentially be achieved via deployment of carbon capture in the power sector, which is minimal in these revised REPEAT Project results. Further reductions are needed to close the gap with the Net-Zero Pathway and reach 2030 targets. Major opportunities3 include: • ~0.2 Gt in the power sector by accelerating coal plant retirements.4 • ~0.1 Gt via improved industrial process efficiency. • ~0.2-0.3 Gt via additional abatement of non-CO2 GHGs and improved agricultural and forestry practices. Further reductions are needed to close the gap with the Net-Zero Pathway and reach 2030 targets. Major opportunities3 include: • ~0.2 Gt in the power sector by accelerating coal plant retirements.4 • ~0.1 Gt via improved industrial process efficiency. • ~0.2-0.3 Gt via additional abatement of non-CO2 GHGs and improved agricultural and forestry practices. Further reductions are needed to close the gap with the Net-Zero Pathway and reach 2030 targets. +1000 +1180 +900 +910 +750 +710 below 2005 levels, leaving a ~0.6-0.9 GtCO2-e/y gap between Current Policies trajectory and U.S. 2030 climate goals. • ~0.2 Gt in the power sector by accelerating coal plant retirements.4 • ~0.1 Gt via improved industrial process efficiency. See additional Notes 1-2 on prior page 3 - While transportation electrification under IRA scenarios largely aligns with the Net-Zero Pathway, accelerated improvements in internal combustion vehicle fuel efficiency could further reduce transport sector emissions by ~10 Mt/yr in 2030 and ~10-30 Mt/yr in 2035. 4 – In Current Policies scenarios, solar capacity additions are already constrained through 2035 and wind capacity additions are close to constraints. Additional reductions in coal-fired generation would therefore likely be compensated primarily by increases in gas-fired generation. Displacing ~400-450 TWh of remaining coal generation in 2030 with gas CCGTs would result in a net reduction of ~0.2 GtCO2-e/year. Deeper reductions could potentially be achieved via deployment of carbon capture in the power sector, which is minimal in these revised REPEAT Project results. 43 4990 4230 3730 3250 2930 2430 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 5500 5090 4770 4250 4050 3630 2022 2024 2026 2028 2030 2032 2034 O2 - Coal and Coke CO2 - Natural Gas CO2 - Petroleum Products CO2 - Process Emissions CO2 - Geologic Storage O2 - Products and Bunkering CH4 - Petroleum & Natural Gas Other Non-CO2 GHGs Land Carbon Sinks Net 5090 4750 4670 4490 2028 2029 2030 2031 2032 2033 2034 2035 44 5500 5090 4770 4250 4050 3630 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 1 - CO2-equivalent emissions calculations use IPCC AR4 100 year global warming potential as per EPA Inventory of Greenhouse Gas Emissions and Sinks. All values rounded to nearest 10 M/yr. All values should be regarded as approximate given uncertainty in future outcomes. 2 - Modeled emissions exclude any changes in passenger and freight miles traveled due to surface transportation, rail, and transit investments in IIJA. See additional Notes 1-2 on prior page ouse Gas Sources and Sinks 2-equivalent (Mt CO2-e)1,2 Current Policies Scenarios olicies (Jan ’21) Conservative Mid-range Optimistic Net-Zero Pathway 5450 4860 4540 3980 3700 3150 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 5550 5090 4730 4160 3870 3340 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Current Policies Scenarios Mid-range Current Policies Scenarios Mid-range CO2 - Natural Gas CH4 - Petroleum & Natural Gas Conservative CO2 - Process Emissions Land Carbon Sinks Optimistic CO2 - Process Emissions Land Carbon Sinks Optimistic CO2 - Petroleum Products Other Non-CO2 GHGs 0 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 44 1 - CO2-equivalent emissions calculations use IPCC AR4 100 year global warming potential as per EPA Inventory of Greenhouse Gas Emissions and Sinks. All values rounded to nearest 10 M/yr. All values should be regarded as approximate given uncertainty in future outcomes. 44 1 - CO2-equivalent emissions calculations use IPCC AR4 100 year global warming potential as per EPA Inventory of Greenhouse Gas Emissions and Sinks. All values rounded to nearest 10 M/yr. All values should be regarded as approximate given uncertainty in future outcomes. 1 - CO2-equivalent emissions calculations use IPCC AR4 100 year global warming potential as per EPA Inventory of Greenhouse Gas Emissions and Sin nearest 10 M/yr. All values should be regarded as approximate given uncertainty in future outcomes. 44 0 1 2 3 4 5 6 7 2005 2010 2015 2020 2025 2030 2035 Historical emissions2 Modeled emissions Revised and Preliminary REPEAT Analysis of the Inflation Reduction Act billion metric tons CO2-equivalent (Gt CO2-e) net U.S. greenhouse gas emissions (including land carbon sinks)1 1 - CO2-equivalent emissions calculations use IPCC AR4 100 year global warming potential as per EPA Inventory of Greenhouse Gas Emissions and Sinks. All values should be regarded as approximate given uncertainty in future out 2 - Historical data from U.S. EPA Inventory for 2005-2030; 2021 estimate from February 2023 draft EPA Inventory. See additional Notes 1-2 on prior page 3 - Modeled emissions exclude any changes in passenger and freight miles traveled due to surface transportation, rail, and transit investments in IIJA. According to the Georgetown Climate Center, emissions impact of these changes from IIJA, which could result in anywhere from -14 Mt to +25 Mt change in CO2 emissions from transportation in 2030. 2030 target: 50% below 2005 Reflecting c other rate-li estimates a systems tra Policies (inc results repo • This revis range of t (‘Conserv ‘Optimisti about the • ‘Prelimina more rap and more deployme • The ‘Optim & ‘Prelimi • See p. 46 of Revise 2021 emissions2: ~5.6 billion tons 2005 emissions2: ~6.6 billion tons Preliminary REPEAT IRA Analysis: ~3.8 billion tons in 2030 (~43% below 2005) Revised Current Policies Scenarios (incl. IRA): ~4.0-4.2 billion tons in 2030 (~37%-41% below 2005) 4 – Jenkins et al. (20 Impacts of the Infla REPEAT Project,, up https://doi.org/10.5 Revised and Preliminary REPEAT Analysis of the Inflation Reduction Act billion metric tons CO2-equivalent (Gt CO2-e) net U.S. greenhouse gas emissions (including land carbon sinks)1 Revised and Preliminary REPEAT Analysis of the Inflation Reduction Act billion metric tons CO2-equivalent (Gt CO2-e) net U.S. greenhouse gas emissions (including land carbon 2015 2020 2025 2030 2035 Historical emissions2 Modeled emissions EPEAT Analysis of the Inflation Reduction Act CO2-e) net U.S. greenhouse gas emissions (including land carbon sinks)1 AR4 100 year global warming potential as per EPA Inventory of Greenhouse Gas Emissions and Sinks. All values should be regarded as approximate given uncertainty in future outcomes. -2030; 2021 estimate from February 2023 draft EPA Inventory. enger and freight miles traveled due to surface transportation, rail, and transit investments in IIJA. According to the Georgetown Climate Center, emissions impact of these changes depend heavily on state implementation of funding Mt to +25 Mt change in CO2 emissions from transportation in 2030. get: 50% below 2005 Reflecting constraints on supply chains and other rate-limiting factors, this analysis estimates a slower start to the energy systems transformation under Current Policies (including IRA) than preliminary results reported in August, 20224: • This revised analysis now includes a range of three Current Policies scenarios (‘Conservative’, ‘Mid-range’, and ‘Optimistic’), better reflecting uncertainty about the impacts of IRA. • ‘Preliminary’ results generally estimated more rapid increases in EV sales share and more rapid solar PV and wind deployment rates than revised results. See additional Notes 1-2 on prior page • The ‘Optimistic’ Current Policies scenario & ‘Preliminary’ scenario converge by 2032. • See p. 46 for sector-by-sector comparison of Revised & Preliminary results 2021 emissions2: ~5.6 billion tons Preliminary REPEAT IRA Analysis: ~3.8 billion tons in 2030 (~43% below 2005) Revised Current Policies Scenarios (incl. IRA): ~4.0-4.2 billion tons in 2030 (~37%-41% below 2005) 4 – Jenkins et al. (2022), “Preliminary Report: The Climate and Energy Impacts of the Inflation Reduction Act of 2022, REPEAT Project,, updated September 21, 2022, https://doi.org/10.5281/zenodo.7106218 uncertainty in future outcomes. Reflecting constraints on supply chains and other rate-limiting factors, this analysis estimates a slower start to the energy systems transformation under Current Policies (including IRA) than preliminary results reported in August, 20224: • This revised analysis now includes a range of three Current Policies scenarios (‘Conservative’, ‘Mid-range’, and ‘Optimistic’), better reflecting uncertainty about the impacts of IRA. • ‘Preliminary’ results generally estimated more rapid increases in EV sales share and more rapid solar PV and wind deployment rates than revised results. • The ‘Optimistic’ Current Policies scenario & ‘Preliminary’ scenario converge by 2032. • See p. 46 for sector-by-sector comparison of Revised & Preliminary results 4 – Jenkins et al. (2022), “Preliminary Report: The Climate and Energy Impacts of the Inflation Reduction Act of 2022, REPEAT Project,, updated September 21, 2022, https://doi.org/10.5281/zenodo.7106218 Reflecting constraints on supply chains and other rate-limiting factors, this analysis estimates a slower start to the energy systems transformation under Current Policies (including IRA) than preliminary results reported in August, 20224: 2021 emissions2: ~5.6 billion tons 6 • This revised analysis now includes a range of three Current Policies scenarios (‘Conservative’, ‘Mid-range’, and ‘Optimistic’), better reflecting uncertainty about the impacts of IRA. • This revised analysis now includes a range of three Current Policies scenarios (‘Conservative’, ‘Mid-range’, and ‘Optimistic’), better reflecting uncertainty about the impacts of IRA. Revised Current Policies Scenarios (incl. IRA): ~4.0-4.2 billion tons in 2030 (~37%-41% below 2005) 2030 target: 50% below 2005 2030 target: 50% below 2005 • ‘Preliminary’ results generally estimated more rapid increases in EV sales share and more rapid solar PV and wind deployment rates than revised results. Preliminary REPEAT IRA Analysis: ~3.8 billion tons in 2030 (~43% below 2005) 4 – Jenkins et al. (2022), “Preliminary Report: The Climate and Energy Impacts of the Inflation Reduction Act of 2022, REPEAT Project,, updated September 21, 2022, https://doi.org/10.5281/zenodo.7106218 0 of Greenhouse Gas Emissions and Sinks. See additional Notes 1-2 on prior page All values should be regarded as approximate given uncertainty in future outcomes. y 021 estimate from February 2023 draft EPA Inventory. nd freight miles traveled due to surface transportation, rail, and transit investments in IIJA. According to the Georgetown Climate Center, emissions impact of these changes depend heavily on state implementation of funding 25 Mt change in CO2 emissions from transportation in 2030. r and freight miles traveled due to surface transportation, rail, and transit investments in IIJA. According to the Georgetown Climate Center, emissio o +25 Mt change in CO2 emissions from transportation in 2030. 45 Differences Between Revised Current Policies Scenarios and Preliminary REPEAT Analysis of the Inflation Reduction million metric tons CO2-equivalent (Mt CO2-e)1,2,3 +420 +320 +160 +470 +200 -10 Conservative Mid-range Optimistic Scenarios and Preliminary REPEAT Analysis of the Inflation Reduction Ac Differences Between Revised Current Policies Scenarios and Preliminary REPEAT Analysis of the Inflation Reduction Act million metric tons CO2-equivalent (Mt CO2-e)1,2,3 Notes: 1 - CO2-equivalent AR4 100 year glob Inventory of Gree All values rounded <10 Mt omitted fr regarded as appro outcomes. 2 - Modeled emiss passenger and fre transportation, ra According to the G emissions impact state implementat could result in any change in CO2 em 3 – Differences vs (2022), “Prelimina Impacts of the Infl REPEAT Project,, u https://doi.org/10 +420 +320 +160 +470 +200 -10 Conservative Mid-range Optimistic Differences Between Revised Current Policies Scenarios and Preliminary REPEAT Analysis of the Inflation R million metric tons CO2-equivalent (Mt CO2-e)1,2,3 Differences Between Revised Current Policies Scenarios and Preliminary REPEAT Analysis of the Inflation Reduction Act million metric tons CO2-equivalent (Mt CO2-e)1,2,3 Conservative Mid-range Optimistic een Revised Current Policies Scenarios and Preliminary REPEAT Analysis of the Inflation Reduction Act quivalent (Mt CO2-e)1,2,3 Notes: 1 - CO2-equivalent emissions calculations use IPCC AR4 100 year global warming potential as per EPA Inventory of Greenhouse Gas Emissions and Sinks. All values rounded to nearest 10 Mt and all values <10 Mt omitted from labels. All values should be regarded as approximate given uncertainty in future outcomes. 2 - Modeled emissions exclude any changes in passenger and freight miles traveled due to surface transportation, rail, and transit investments in IIJA. See additional Notes 1-2 on prior page According to the Georgetown Climate Center, emissions impact of these changes depend heavily on state implementation of funding from IIJA, which could result in anywhere from -14 Mt/yr to +25 Mt/yr change in CO2 emissions from transportation in 2030. 3 – Differences vs values reported in Jenkins et al. (2022), “Preliminary Report: The Climate and Energy Impacts of the Inflation Reduction Act of 2022, REPEAT Project,, updated September 21, 2022, https://doi.org/10.5281/zenodo.7106218 +420 +320 +160 +470 +200 -10 Conservative Mid-range Optimistic Optimistic Notes: 46 47 Final Energy Demand by Energy Carrier/Fuel trillion Btu per year 2,769 1,553 14,796 744 4,893 3,459 13,198 7,338 10,418 2,764 0 10,000 20,000 30,000 40,000 50,000 60,000 70,000 80,000 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2,897 1,612 15,448 754 4,996 3,731 11,897 7,047 10,887 2,811 0 10,000 20,000 30,000 40,000 50,000 60,000 70,000 80,000 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2,902 1,622 14,842 754 5,023 3,731 12,804 7,530 11,210 2,811 0 10,000 20,000 30,000 40,000 50,000 60,000 70,000 80,000 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 mand by Energy Carrier/Fuel Current Policies Scenarios Policies (Jan ’21) Mid-range Optimistic Net-Zero Pathway 2,822 1,628 14,717 754 5,057 3,731 12,805 7,612 11,747 2,811 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Conservative 2,820 1,611 14,374 755 5,019 3,731 13,379 7,759 11,974 2,812 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Total: 65,232 64,264 63,721 63,239 62,980 Net-Zero Pathway Net-Zero Pathway Optimistic 80,000 0 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 0 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 0 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2029 2030 2031 2032 2033 2034 2035 48 ergy Demand by Energy Carrier/Fuel vs Frozen Policies as of January 2021 -319 -737 -1,342 -2,258 -3,547 -6,091 -10,000 -8,000 -6,000 -4,000 -2,000 0 2,000 4,000 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 -135 -419 -917 -1,517 -2,023 -2,884 -10,000 -8,000 -6,000 -4,000 -2,000 0 2,000 4,000 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 -135 -419 -917 -1,517 -2,023 -2,884 -10,000 -8,000 -6,000 -4,000 -2,000 0 2,000 4,000 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Final Energy Demand by Energy Carrier/Fuel vs Frozen Policies as of January 2021 year Current Policies Scenarios Mid-range Optimistic Net-Zero Pathway Conservative -354 -666 -974 -1,178 -1,332 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 -319 -737 -1,342 -2,258 -3,547 6000 -4,000 -2,000 0 2,000 4,000 -135 -419 -917 -1,517 -2,023 -2,884 6000 -4,000 -2,000 0 2,000 4,000 -135 -419 -917 -1,517 -2,023 -2,884 -6000 -4,000 -2,000 0 2,000 4,000 gy y gy y ear Current Policies Scenarios Mid-range Optimistic Net-Zero Pathway Conservative -354 -666 -974 -1,178 -1,332 Net-Zero Pathway -1,517 -1,517 -2,023 -4,000 -2,884 -6,091 49 Energy Demand by Sector Btu per year Current Policies Scenarios Frozen Policies (Jan ’21) Mid-range Optimistic Net-Zero Pathway Conservative Frozen Policy (Jan' 2021) IRA, Conservative IRA, Midrange IRA, Optimistic Net-Zero Pathway 0 2025 2030 2035 2020 2025 2030 2035 2020 2025 2030 2035 2020 2025 2030 2035 2020 2025 2030 2035 transportation residential commercial industry transportation residential commercial industry transportation residential commercial industry transportation residential commercial industry transportation residential commercial industry nergy by Sector Electrification and end-use efficiency lower final energy consumption in Current Policies and Net-Zero Pathway scenarios, even as total demand for energy services increases. Electric vehicles and heat pumps are much more efficient at converting final energy carriers to energy services (vehicle miles traveled, square footage conditioned) than internal combustion engines and fossil boilers. Current Policies Scenarios Mid-range IRA, Midrange Current Policies Scenarios Mid-range IRA, Midrange Conservative RA, Conservativ Net-Zero Pathway Net-Zero Pathway Optimistic A, Optimisti Electrification and end-use efficiency lower final energy consumption in Current Policies and Net-Zero Pathway scenarios, even as total demand for energy services increases. Electric vehicles and heat pumps are much more efficient at converting final energy carriers to energy services (vehicle miles traveled, square footage conditioned) than internal combustion engines and fossil boilers. Electrification and end-use efficiency lower final energy consumption in Current Policies and Net-Zero Pathway scenarios, even as total demand for energy services increases. Electric vehicles and heat pumps are much more efficient at converting final energy carriers to energy services (vehicle miles traveled, square footage conditioned) than internal combustion engines and fossil boilers. 2035 2020 2035 50 3% 1% -1% -5% -9% 4% 2% -1% -5% -9% -15% 4% 1% -3% -8% -14% -22% 4% 1% -4% -10% -17% -26% 2024 2026 2028 2030 2032 2035 Frozen Policy (Jan '21) Conservative Mid-range Optimsitic Current Policies Scenarios ual U.S. Petroleum Product Consumption vs 2021 (mmbbl/y) Policies enacted by the 117th Congress will spur the first sustained decline in petroleum product consumption in U.S. history, with consumption of gasoline, diesel, and other petroleum-based fuels declining 5-10% from 2021 levels by 2030 and 15-26% by 2035. Changes in Annual U.S. Petroleum Product Consumption vs 2021 million barrels per year (mmbbl/y) Changes in Annual U.S. Petroleum Product Consumption vs 2021 million barrels per year (mmbbl/y) 2035 -26% 51 Changes in Annual U.S. Natural Gas Consumption vs 2021 trillion cubic feet per year (Tcf/year) -10% -10% -13% -14% -14% -13% -15% -17% -21% -24% -30% -10% -9% -13% -18% -21% -29% -13% -14% -16% -20% -24% -33% 2024 2026 2028 2030 2032 2035 Frozen Policy (Jan '21) Conservative Mid-range Optimsitic 52 Current Policies Scenarios Policies enacted by the 117th Congress reduce U.S. consumption of natural gas, as renewable energy displaces natural gas for power generation and efficiency and electrification lower gas use in buildings and industry. Under Current Policies scenarios, U.S. natural gas consumption declines by about one-fifth from 2021 levels by 2030 and by about one-third by 2035. Note – The magnitude of reductions in natural gas demand is affected by the balance of wind and solar power deployment rates (which offset gas consumption for power generation, all else equal) and the pace of electrification (which increases electricity demand and thus natural gas consumption for power generation, all else equal). As a result, the Conservative scenario with slower electrification results in lower gas demand in many years than the Middle or Optimistic Current Policies scenarios, as these scenarios have both faster wind and solar deployment and higher electricity demand from more rapid EV adoption and other electrification. p et per year (Tcf/year) -8% -10% -10% -13% -14% -14% -13% -15% -17% -10% -9% -13% -18% -13% -14% -16% 20% 2024 2026 2028 2030 2032 2035 Frozen Policy (Jan '21) Conservative Mid-range Optimsitic Current Policies Scenarios ubic feet per year (Tcf/year) 2030 ervative Mid-range Optimsitic Current Policies Scenarios Note – The magnitude of reductions in natural gas demand is affected by the balance of wind and solar power deployment rates (which offset gas consumption for power generation, all else equal) and the pace of electrification (which increases electricity demand and thus natural gas consumption for power generation, all else equal). As a result, the Conservative scenario with slower electrification results in lower gas demand in many years than the Middle or Optimistic Current Policies scenarios, as these scenarios have both faster wind and solar deployment and higher electricity demand from more rapid EV adoption and other electrification. Note – The magnitude of reductions in natural gas demand is affected by the balance of wind and solar power deployment rates (which offset gas consumption for power generation, all else equal) and the pace of electrification (which increases electricity demand and thus natural gas consumption for power generation, all else equal). As a result, the Conservative scenario with slower electrification results in lower gas demand in many years than the Middle or Optimistic Current Policies scenarios, as these scenarios have both faster wind and solar deployment and higher electricity demand from more rapid EV adoption and other electrification. 52 - 1,000 2,000 3,000 4,000 5,000 6,000 2021 2022 2023 2024 2025 2026 2027 2028 3,190 3,146 863 934 388 889 23 189 268 434 - 1,000 2,000 3,000 4,000 5,000 6,000 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 3,193 3,189 845 890 287 609 25 160 270 446 - 1,000 2,000 3,000 4,000 5,000 6,000 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 3,026 3,215 3,324 786 790 799 29 203 476 - 1,000 2,000 3,000 4,000 5,000 6,000 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 hydrogen electrolysis industrial steam transportation heating other Electricity Demand by Sector terawatt-hours per year (TWh/year) Current Policies Scenarios Frozen Policies (Jan ’21) Mid-range Optim Conservative 4,218 4,635 4,619 5,294 4,732 5,592 3,841 3,097 3,120 861 1,049 203 678 175 393 40 142 - 1,000 2,000 3,000 4,000 5,000 6,000 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 3,172 3,049 871 957 449 1,023 35 273 269 436 - 1,000 2,000 3,000 4,000 5,000 6,000 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 3,190 3,146 863 934 388 889 23 189 268 434 - 1,000 2,000 3,000 4,000 5,000 6,000 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 3,193 3,189 845 890 287 609 25 160 270 446 - 1,000 2,000 3,000 4,000 5,000 6,000 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 15 3,324 0 799 3 476 2030 2031 2032 2033 2034 2035 olysis by Sector Wh/year) Current Policies Scenarios Jan ’21) Mid-range Optimistic Net-Zero Pathway Conservative Growing adoption of electric vehicles, heat pumps, industrial electric boilers, and hydrogen electrolysis will drive a sustained increase in U.S. electricity consumption for the first time since the mid-2000s. Under Current Policies scenarios, U.S. electricity demand grows 20-25% from 2022 to 2030 and reaches 38-49% higher than 2022 demand by 2035. 18 4,635 4,619 5,294 4,732 5,592 4,797 5,739 4,376 5,382 3,097 3,120 861 1,049 203 678 175 393 40 142 - 1,000 2,000 3,000 4,000 5,000 6,000 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 3,172 3,049 871 957 449 1,023 35 273 269 436 - 1,000 2,000 3,000 4,000 5,000 6,000 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 3,190 3,146 863 934 388 889 23 189 268 434 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Current Policies Scenarios Mid-range Optimistic Net-Zero Pathway Growing a vehicl industrial hydroge drive a su U.S. elect for the f m Under scenario demand g 2022 to 38-49% dem 4,732 5,592 4,797 5,739 4,376 5,382 Current Policies Scenarios Mid-range 3,097 3,120 861 1,049 203 678 175 393 40 142 - 1,000 2,000 3,000 4,000 5,000 6,000 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 3,172 3,049 871 957 449 1,023 35 273 269 436 - 1,000 2,000 3,000 4,000 5,000 6,000 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 3,190 3,146 863 934 388 889 23 189 268 434 - 1,000 2,000 3,000 4,000 5,000 6,000 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 3,189 890 609 160 446 2032 2033 2034 2035 Current Policies Scenarios Mid-range Optimistic Net-Zero Pathway Growing adoption of electric vehicles, heat pumps, industrial electric boilers, and hydrogen electrolysis will drive a sustained increase in U.S. electricity consumption for the first time since the mid-2000s. Under Current Policies scenarios, U.S. electricity demand grows 20-25% from 2022 to 2030 and reaches 38-49% higher than 2022 demand by 2035. 5,294 4,732 5,592 4,797 5,739 4,376 5,382 Net-Zero Pathway Net-Zero Pathway Optimistic Growing adoption of electric vehicles, heat pumps, industrial electric boilers, and hydrogen electrolysis will drive a sustained increase in U.S. electricity consumption for the first time since the mid-2000s. Under Current Policies scenarios, U.S. electricity demand grows 20-25% from 2022 to 2030 and reaches 38-49% higher than 2022 demand by 2035. 54 42% 49% 59% 52% 66% 82% 65% 81% 94% 50% 69% 86% 40% 39% 60% 67% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Frozen Policies/Current Policies (Conservative) Current Policies (Mid-range) Curre Net-Zero Pathway EPA Current Policies Scenario EPA P 13% 17% 24% 44% 48% 42% 44% 48% 48% 11% 23% 49% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 2022 2023 2024 2025 Light vehicles Medium vehicles Electric Vehicle Sales Shares percent of new vehicle sales per year1 2 1 - See p. 30-31 for details on REPEAT Project treatment of electric and zero emissions vehicle adoption. 2 - EPA scenarios provided for comparison purposes from “Multi-Pollutant Emissions Standards for Model Years 2027 and Lat Draft Regulatory Impact Analysis” U.S. EPA (April 2023), at p. 13-35 & 13-37. “Proposed Regulations” scenario includes model multi-pollution emissions standard, which is not included in REPEAT Project Current Policies scenarios. 42% 49% 59% 52% 66% 82% 65% 81% 94% 50% 69% 86% 40% 39% 60% 67% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Frozen Policies/Current Policies (Conservative) Net-Zero Pathway 1 Light vehicles percent of new vehicle sales per year1 55 42% 49% 59% 52% 66% 82% 65% 81% 94% 50% 69% 86% 40% 39% 60% 67% 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Frozen Policies/Current Policies (Conservative) Current Policies (Mid-range) Current Policies (Optimistic) Net-Zero Pathway EPA Current Policies Scenario EPA Proposed Regulations Scenario 13% 17% 24% 44% 48% 42% 44% 48% 48% 11% 23% 49% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 11% 30% 40% 56% 51% 40% 56% 53% 10% 44% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Light vehicles Medium vehicles Heavy vehicles vehicle sales per year1 2 2 1 - See p. 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 53 Transportation Sector Changes Transportation Sector Changes Transportation Sector Changes Frozen Policies Vehicle stock vehicles Final energy use by carrier/fuel trillion Btu per year Conservative Mid-range Optimistic Net-Zero Pathway Current Policies Scenarios Frozen Policies Conservative Mid-range Optimistic Net-Zero Pathway Current Policies Scenarios Penetration and im under Current Polic Light duty vehicle s 42-65% by 2030, 59 Light duty vehicle f 13-19% by 2030, 29 36-51 million EVs o 2030. Medium/heavy truc ~12-42% by 2030, 2 Medium/heavy truc ~3-10% by 2030, 9-2 0.5-1.6 million EV m trucks on U.S. roads g Final energy use by carrier/fuel trillion Btu per year Conservative Mid-range Optimistic Net-Zero Pathway Current Policies Scenarios Frozen Policies Conservative Mid-range Optimistic Net-Zero Pathway Current Policies Scenarios Penetration and impact of EVs under Current Policies scenarios Light duty vehicle sales share: 42-65% by 2030, 59-94% by 2035. Light duty vehicle fleet share: 13-19% by 2030, 29-44% by 2035. 36-51 million EVs on U.S. roads in 2030. Medium/heavy truck sales share: ~12-42% by 2030, 27-51% by 2035. Medium/heavy truck fleet share: ~3-10% by 2030, 9-24% by 2035. 0.5-1.6 million EV medium/heavy trucks on U.S. roads in 2030. Final energy use by carrier/fuel illi Final energy use by carrier/fuel trillion Btu per year Frozen Policies Conservative M Current Po Final energy use by carrier/fu trillion Btu per year Frozen Policies Conservative Curren ve Mid-range Current Policies Scenarios Net-Zero Pathway Optimistic s Frozen Policies Net-Zero Pathway Optimistic Conservative Medium/heavy truck sales share: ~12-42% by 2030, 27-51% by 2035. Medium/heavy truck fleet share: ~3-10% by 2030, 9-24% by 2035. 0.5-1.6 million EV medium/heavy trucks on U.S. roads in 2030. 30-31 for details on REPEAT Project treatment of electric and zero emissions vehicle adoption. 2 - EPA scenarios provided for comparison purposes from “Multi-Pollutant Emissions Standards for Model Years 2027 and Later Light-Duty and Medium-Duty Vehicles: Draft Regulatory Impact Analysis” U.S. EPA (April 2023), at p. 13-35 & 13-37. “Proposed Regulations” scenario includes modeled compliance with proposed MY2027+ multi-pollution emissions standard, which is not included in REPEAT Project Current Policies scenarios. 42% 49% 59% 52% 66% 82% 65% 81% 94% 50% 69% 86% 40% 39% 60% 67% 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Frozen Policies/Current Policies (Conservative) Current Policies (Mid-range) Net-Zero Pathway EPA Current Policies Scenario 13% 17% 24% 44% 48% 42% 44% 48% 48% 11% 23% 49% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 1 2 3 4 5 6 7 8 9 10 Light vehicles Medium vehicles 2 1 - See p. 30-31 for details on REPEAT Project treatment of electric and zero emissions vehicle ado 2 - EPA scenarios provided for comparison purposes from “Multi-Pollutant Emissions Standards fo Draft Regulatory Impact Analysis” U.S. EPA (April 2023), at p. 13-35 & 13-37. “Proposed Regulatio multi-pollution emissions standard, which is not included in REPEAT Project Current Policies scen 11% 30% 40% 56% 51% 40% 56% 53% 10% 44% 0% 10% 20% 30% 40% 50% 60% 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 p j p 2 - EPA scenarios provided for comparison purposes from “Multi-Pollutant Emissions Standards for Model Years 2027 and Later Light-Duty and Medium-Duty Vehicles: Draft Regulatory Impact Analysis” U.S. EPA (April 2023), at p. 13-35 & 13-37. “Proposed Regulations” scenario includes modeled compliance with proposed MY2027+ multi-pollution emissions standard, which is not included in REPEAT Project Current Policies scenarios. 55 Frozen Policy (Jan' 2021) IRA, Conservative IRA, Midrange IRA, Optimistic Net-Zero Pathway 2020 2025 2030 2035 2020 2025 2030 2035 2020 2025 2030 2035 2020 2025 2030 2035 2020 2025 2030 2035 0K 2K 4K 6K 8K 10K 12K 14K 16K 18K 20K 22K 24K 26K jet fuel diesel fuel gasoline fuel jet fuel diesel fuel gasoline fuel jet fuel diesel fuel gasoline fuel jet fuel diesel fuel gasoline fuel jet fuel diesel fuel gasoline fuel Transportation Final Energy Demand by Energy Carrier trillion Btu ammonia hydrogen electricity lpg fuel and feeds pipeline gas jet fuel diesel fuel gasoline fuel other petroleum Transportation Final Energy Demand by Energy Carrier/Fuel trillion Btu per year Current Policies Scenarios Frozen Policies (Jan ’21) Mid-range Optimistic Net-Zero Pathway Conservative Vehicle elec energy dem because efficient t Total final falls to 12-16 Current Polic and fre Demand for rapidly, fa Transportation Final Energy Demand by Energy Carrier Transportation Final Energy Demand by Energy Carrier/Fuel trillion Btu per year rozen Policy (Jan' 2021) IRA, Conservative IRA, Midrange IRA, Optimistic Net-Zero Pathway 20 2025 2030 2035 2020 2025 2030 2035 2020 2025 2030 2035 2020 2025 2030 2035 2020 2025 2030 2035 jet fuel diesel fuel gasoline fuel jet fuel diesel fuel gasoline fuel jet fuel diesel fuel gasoline fuel jet fuel diesel fuel gasoline fuel jet fuel diesel fuel gasoline fuel portation Final Energy Demand by Energy Carrier u ammonia hydrogen electricity lpg fuel and feedstocks pipeline gas jet fuel diesel fuel gasoline fuel other petroleum tu per year Current Policies Scenarios rozen Policies (Jan ’21) Mid-range Optimistic Net-Zero Pathway Conservative Vehicle electrification significantly reduces final energy demand in the transportation sector, because electric vehicles are much more efficient than internal combustion engines. Total final energy demand for transportation falls to 12-16% below 2022 levels in 2030 under Current Policies scenarios, even as total vehicle and freight miles traveled increases. Demand for gasoline and diesel fuel drops more rapidly, falling 17-24% from 2022 to 2030. olicy (Jan' 2021) IRA, Conservative IRA, Midrange IRA, Optimistic Net-Zero Pathway 5 2030 2035 2020 2025 2030 2035 2020 2025 2030 2035 2020 2025 2030 2035 2020 2025 2030 2035 fuel soline fuel jet fuel diesel fuel gasoline fuel jet fuel diesel fuel gasoline fuel jet fuel diesel fuel gasoline fuel jet fuel diesel fuel gasoline fuel ion Final Energy Demand by Energy Carrier ammonia hydrogen electricity lpg fuel and feedstocks pipeline gas jet fuel diesel fuel gasoline fuel other petroleum ear Current Policies Scenarios olicies (Jan ’21) Mid-range Optimistic Net-Zero Pathway Conservative Vehicle electrification significantly reduces final energy demand in the transportation sector, because electric vehicles are much more efficient than internal combustion engines. Total final energy demand for transportation falls to 12-16% below 2022 levels in 2030 under Current Policies scenarios, even as total vehicle and freight miles traveled increases. Demand for gasoline and diesel fuel drops more rapidly, falling 17-24% from 2022 to 2030. IRA, Midrange Current Policies Scenarios Mid-range IRA, Midrange Current Policies Scenarios Mid-range ammonia hydrogen electricity lpg fuel and feedstocks pipeline gas jet fuel diesel fuel gasoline fuel other petroleum Vehicle electrification significantly reduces final energy demand in the transportation sector, because electric vehicles are much more efficient than internal combustion engines. Total final energy demand for transportation falls to 12-16% below 2022 levels in 2030 under Current Policies scenarios, even as total vehicle and freight miles traveled increases. Demand for gasoline and diesel fuel drops more rapidly, falling 17-24% from 2022 to 2030. 2035 56 Residential Building Sector Changes Residential Building Sector Changes Frozen Policies Device stock evices Final energy use by carrier/fuel trillion Btu per year Conservative Mid-range Optimistic Net-Zero Pathway Current Policies Scenarios Frozen Policies Conservative Mid-range Optimistic Net-Zero Pathway Current Policies Scenarios Penetration residential h Current Poli Space heati 34-59% by 2 ~15% in Fro Space heati 19-24% by 2 25-31 millio homes in 20 Water heatin 12-21% by 2 ~0% in Froz Water heatin 8-12% by 20 11-16 millio heaters in U Final energy use by carrier/fuel trillion Btu per year Frozen Policies Conservative M Current Po Final energy use by carrier/fuel trillion Btu per year Frozen Policies Conservative M Current P Final energy use by carrier/fuel t illi Bt Conservative Mid-range Optimistic Net-Ze Pathw Current Policies Scenarios Net-Zero Pathway Net-Zero Pathway Space heating sales share: 34-59% by 2030 and after (up from ~15% in Frozen Policies). 57 Commercial Building Sector Changes mercial Building Sector Changes Frozen Policies e stock u of heating capacity Final energy use by carrier/fuel trillion Btu per year Conservative Mid-range Optimistic Net-Zero Pathway Current Policies Scenarios Frozen Policies Conservative Mid-range Optimistic Net-Zero Pathway Current Policies Scenarios Penetration and impact of commercial heat pumps under Current Policies scenarios Space heating sales share: 43-70% by 2030 and after (up from 7% in Frozen Policies). Space heating stock: 20-23% by 2030, 30-42% by 2035. Water heating sales share: ~2% and unaffected by policy scenarios (vs. 24% in 2030 in Net-Zero Pathway). Water heating stock: ~0.2% (vs 6% in 2030 in Net-Zero Pathway) lding Sector Changes ity Final energy use by carrier/fuel trillion Btu per year Conservative Mid-range Optimistic Net-Zero Pathway Current Policies Scenarios Frozen Policies Conservative Mid-range Optimistic Net-Zero Pathway Current Policies Scenarios Penetration and impact of commercial heat pumps under Current Policies scenarios Space heating sales share: 43-70% by 2030 and after (up from 7% in Frozen Policies). Space heating stock: 20-23% by 2030, 30-42% by 2035. Water heating sales share: ~2% and unaffected by policy scenarios (vs. 24% in 2030 in Net-Zero Pathway). Water heating stock: ~0.2% (vs 6% in 2030 in Net-Zero Pathway) Final energy use by carrier/fuel t illi Bt Net-Zero Pathway Space heating sales share: 43-70% by 2030 and after (up from 7% in Frozen Policies). 58 Final Energy Demand in Buildings trillion Btu Building Final Energy Demand by Energy Carrier/Fuel trillion Btu per year Frozen Policy (Jan' 2021) IRA, Conservative IRA, Midrange IRA, Optimistic Net-Zero Pathway 2020 2025 2030 2035 2020 2025 2030 2035 2020 2025 2030 2035 2020 2025 2030 2035 2020 2025 2030 2035 0K 2K 4K 6K 8K 10K 12K 14K 16K 18K 20K electricity pipeline gas electricity pipeline gas electricity pipeline gas electricity pipeline gas electricity pipeline gas Final Energy Demand in Buildings trillion Btu Building Final Energy Demand by Energy Carrier/Fuel trillion Btu per year Current Policies Scenarios Frozen Policies (Jan ’21) Mid-range Optimistic Net-Zero Pathway Conservative IRA, Conservative IRA, Midrange IRA, Optimistic Net-Zero Pathway 2020 2025 2030 2035 2020 2025 2030 2035 2020 2025 2030 2035 2020 2025 2030 2035 electricity pipeline gas electricity pipeline gas electricity pipeline gas electricity pipeline gas biomass & waste electricity steam lpg fuel and feedstocks pipeline gas diesel fuel other petroleum Current Policies Scenarios Mid-range Optimistic Net-Zero Pathway Conservative As in transportation, building electrification reduces total final energy demand, thanks to the improved efficiency of electric heat pumps relative to fossil boilers and furnaces and electric resistance heating. Total final energy demand in buildings falls by 1-3% from 2022 to 2030 under Current Policies scenarios, despite growth in population and building square footage. Demand for natural gas in buildings contracts by 10-13% from 2022 to 2030. IRA, Midrange Current Policies Scenarios Mid-range IRA, Midrange Current Policies Scenarios Mid-range Net-Zero Pathway Net-Zero Pathway y ( 2021) Frozen Policies (Jan ’21) RA, Optimist Optimistic biomass & waste electricity steam lpg fuel and feedstocks pipeline gas diesel fuel other petroleum biomass & waste electricity steam lpg fuel and feedstocks pipeline gas diesel fuel other petroleum As in transportation, building electrification reduces total final energy demand, thanks to the improved efficiency of electric heat pumps relative to fossil boilers and furnaces and electric resistance heating. Total final energy demand in buildings falls by 1-3% from 2022 to 2030 under Current Policies scenarios, despite growth in population and building square footage. Demand for natural gas in buildings contracts by 10-13% from 2022 to 2030. As in transportation, building electrification reduces total final energy demand, thanks to the improved efficiency of electric heat pumps relative to fossil boilers and furnaces and electric resistance heating. Total final energy demand in buildings falls by 1-3% from 2022 to 2030 under Current Policies scenarios, despite growth in population and building square footage. 2035 59 Industrial Final Energy Demand by Energy Carrier/Fuel trillion Btu per year Industry Final Energy Demand by Energy Carrier trillion Btu Energy Demand by Energy Carrier Industrial Final Energy Demand by Energy Carrier/Fuel trillion Btu per year Frozen Policy (Jan' 2021) IRA, Conservative IRA, Midrange IRA, Optimistic Net-Zero Pathway 2020 2025 2030 2035 2020 2025 2030 2035 2020 2025 2030 2035 2020 2025 2030 2035 2020 2025 2030 2035 0K 2K 4K 6K 8K 10K 12K 14K 16K 18K 20K 22K Industry Final Energy Demand by Energy Carrier trillion Btu steam biomass & waste hydrogen electricity lpg fuel and feedstocks pipeline gas diesel fuel gasoline fuel coal & coking coal other petroleum Policy (Jan' 2021) IRA, Conservative IRA, Midrange IRA, Optimistic Net-Zero Pathway 2030 2035 2020 2025 2030 2035 2020 2025 2030 2035 2020 2025 2030 2035 2020 2025 2030 2035 al Energy Demand by Energy Carrier steam biomass & waste hydrogen electricity lpg fuel and feedstocks pipeline gas diesel fuel gasoline fuel coal & coking coal other petroleum Current Policies Scenarios Frozen Policies (Jan ’21) Mid-range Optimistic Net-Zero Pathway Conservative A Net-Zero progress in (efficienc Cur This reflects incentivizing efficiency in steam biomass & waste hydrogen electricity lpg fuel and feedstocks pipeline gas diesel fuel gasoline fuel coal & coking coal other petroleum steam biomass & waste hydrogen electricity lpg fuel and feedstocks pipeline gas diesel fuel gasoline fuel coal & coking coal other petroleum A Net-Zero Pathway would see more rapid progress in industrial energy productivity (efficiency) gains than observed under Current Policies scenarios. This reflects a lack of comprehensive policy incentivizing industrial decarbonization and efficiency in the Inflation Reduction Act and Infrastructure Law. ear Policy (Jan' 021) IRA, Conservative IRA, Midrange IRA, Optimistic Net-Zero Pathway 2030 2035 2020 2025 2030 2035 2020 2025 2030 2035 2020 2025 2030 2035 2020 2025 2030 2035 steam biomass & waste hydrogen electricity lpg fuel and feedstocks pipeline gas diesel fuel gasoline fuel coal & coking coal other petroleum IRA, Conservative IRA, Midrange IRA, Optimistic Net-Zero Pathway steam biomass & waste hydrogen electricity lpg fuel and feedstocks pipeline gas diesel fuel gasoline fuel coal & coking coal other petroleum Current Policies Scenarios icies (Jan ’21) Mid-range Optimistic Net-Zero Pathway Conservative A Net-Zero Pathway would see more rapid progress in industrial energy productivity (efficiency) gains than observed under Current Policies scenarios. This reflects a lack of comprehensive policy incentivizing industrial decarbonization and efficiency in the Inflation Reduction Act and Infrastructure Law. IRA, Midrange IR idrange IRA, Op Current Policies Scenarios Mid-range IRA, Midrange I idrange IRA, Op Current Policies Scenarios Mid-range RA, Optimisti imistic Optimistic Frozen Policy (Jan 2021) 2020 2025 2030 2035 2 0K 2K 4K 6K 8K 10K 12K 14K 16K 18K 20K 22K y ( 021) IRA, Conse Frozen Policies (Jan ’21) A Net-Zero Pathway would see more rapid progress in industrial energy productivity (efficiency) gains than observed under Current Policies scenarios. This reflects a lack of comprehensive policy incentivizing industrial decarbonization and efficiency in the Inflation Reduction Act and Infrastructure Law. A Net-Zero Pathway would see more rapid progress in industrial energy productivity (efficiency) gains than observed under Current Policies scenarios. This reflects a lack of comprehensive policy incentivizing industrial decarbonization and efficiency in the Inflation Reduction Act and Infrastructure Law. 2030 2035 25 2030 60 Industrial Steam Production Capacity Stock gigawatts thermal (GW-th) 3 4 30 36 29 55 0 20 40 60 80 100 120 140 160 180 200 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 5 9 12 57 0 20 40 60 80 100 120 140 160 180 200 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 6 10 13 44 0 20 40 60 80 100 120 140 160 180 200 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 3 7 10 39 0 20 40 60 80 100 120 140 160 180 200 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 t a Stea oduct o Capac ty Stoc ts thermal (GW-th) Current Policies Scenarios Frozen Policies (Jan ’21) Mid-range Optimistic Net-Zero Pathway Conservative 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Under Current Policies, lower-cost renewable electricity and tax credits for hydrogen production spur increased adoption of electric resistance heating with thermal storage and hydrogen boilers for industrial process heat and steam requirements Current Policies Scenarios Mid-range Current Policies Scenarios Mid-range Net-Zero Pathway 0 61 175 393 0 200 400 600 800 1,000 1,200 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 35 273 0 200 400 600 800 1,000 1,200 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 23 189 0 200 400 600 800 1,000 1,200 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 25 160 0 200 400 600 800 1,000 1,200 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Steam Production thermal (TWh-th) Current Policies Scenarios zen Policies (Jan ’21) Mid-range Optimistic Net-Zero Pathway Conservative ctric resistance heating coupled with thermal age helps decarbonize industrial process heat Current Policies, taking advantage of low-cost renewable electricity incentivized by IRA. Current Policies Scenarios Mid-range Current Policies Scenarios Mid-range 1,200 Optimistic Net-Zero Pathway 800 0 0 62 63 zen Policy (Jan' 2021) IRA, Conservative IRA, Midrange IRA, Optimistic Net-Zero Pathway 5 2030 2035 Year 2025 2030 2035 Year 2025 2030 2035 Year 2025 2030 2035 Year 2025 2030 2035 Year nergy solar onshore wind offshore wind biomass geothermal uranium hydro oil natural gas fossil solids 64 ergy Supply per year (Quads/year) Current Policies Scenarios en Policies (Jan ’21) Mid-range Optimistic Net-Zero Pathway Conservative IRA, Midrange IRA, Optimistic Net-Zero Pathway solar onshore wind offshore wind biomass geothermal uranium hydro oil natural gas fossil solids Primary energy use falls in Current Policies and Net-Zero Pathway cases as renewable energy (and electrification) displaces fossil combustion and energy efficiency improves in buildings and industry. 2035 % reduction from 2022: -3% -4% -6% -16% coal and coke Primary Energy quads Primary Energy Supply quadrillion Btu per year (Quads/year) zen Policy (Jan' 2021) IRA, Conservative IRA, Midrange IRA, Optimistic Net-Zero Pathway 25 2030 2035 Year 2025 2030 2035 Year 2025 2030 2035 Year 2025 2030 2035 Year 2025 2030 2035 Year nergy solar onshore wind offshore wind biomass geothermal uranium hydro oil natural gas fossil solids per year (Quads/year) Current Policies Scenarios en Policies (Jan ’21) Mid-range Optimistic Net-Zero Pathway Conservative IRA, Midrange IRA, Optimistic Net-Zero Pathway solar onshore wind offshore wind biomass geothermal uranium hydro oil natural gas fossil solids Primary energy use falls in Current Policies and Net-Zero Pathway cases as renewable energy (and electrification) displaces fossil combustion and energy efficiency improves in buildings and industry. 2035 % reduction from 2022: -3% -4% -6% -16% coal and coke Frozen Policy (Jan' 2021) IRA, Conservative IRA, Midrange IRA, Optimistic Net-Zero Pathway 2025 2030 2035 Year 2025 2030 2035 Year 2025 2030 2035 Year 2025 2030 2035 Year 2025 2030 2035 Year ary Energy solar onshore wind offshore wind biomass geothermal uranium hydro oil natural gas fossil solids on Btu per year (Quads/year) Current Policies Scenarios Frozen Policies (Jan ’21) Mid-range Optimistic Net-Zero Pathway Conservative tive IRA, Midrange IRA, Optimistic Net-Zero Pathway solar onshore wind offshore wind biomass geothermal uranium hydro oil natural gas fossil solids Primary energy use falls in Current Policies and Net-Zero Pathway cases as renewable energy (and electrification) displaces fossil combustion and energy efficiency improves in buildings and industry. 2035 % reduction from 2022: -3% -4% -6% -16% coal and coke IRA, Midrange Current Policies Scenarios Mid-range IRA, Optim IRA, Midrange Current Policies Scenarios Mid-range IRA, Optim Frozen Policy (Jan' 2021) Frozen Policies (Jan ’21) tive IRA, Optimistic Optimistic IRA, Conservative Conservative Midrange 2035 % reduction from 2022: -3% Primary energy use falls in Current Policies and Net-Zero Pathway cases as renewable energy (and electrification) displaces fossil combustion and energy efficiency improves in buildings and industry. Primary energy use falls in Current Policies and Net-Zero Pathway cases as renewable energy (and electrification) displaces fossil combustion and energy efficiency improves in buildings and industry. 64 Change in Primary Energy Supply vs Frozen Policies as of January 2021 quadrillion Btu per year (Quads/year) Current Policies Scenarios Mid-range Conservative IRA, Conservative IRA, Midrange IRA, Op IRA, Conservative IRA, Midrange Primary Energy Relative to Frozen Policy quads Change in Primary Energy Supply vs Frozen Policies as of January 2021 quadrillion Btu per year (Quads/year) Current Policies Scenarios Mid O ti i ti C ti IRA Conservative IRA Midrange IRA Optimistic Primary Energy Relative to Frozen Policy quads Change in Primary Energy Supply vs Frozen Policies as of January 2021 uadrillion Btu per year (Quads/year) Current Policies Scenarios Mid-range Optimistic Net-Zero Pathway Conservative RA, Conservative IRA, Midrange IRA, Optimistic Net-Zero Pathway solar onshore wind offshore wind biomass geothermal uranium hydro oil natural gas fossil solids IRA, Conservative IRA, Midrange IRA, Optimistic Net-Zero Pathway 2022 2024 2026 2028 2030 2032 2035 2022 2024 2026 2028 2030 2032 2035 2022 2024 2026 2028 2030 2032 2035 2022 2024 2026 2028 2030 2032 2035 25 20 5 0 -5 0 5 Primary Energy Relative to Frozen Policy uads solar onshore wind offshore wind biomass geothermal uranium hydro oil natural gas fossil solids coal and coke -2.0 (-2%) -2.0 (-2%) -2.9 (-3%) -10.5 (-9%) Net change vs Frozen Policies: -3.9 (-4%) -5.0 (-5%) -7.0 (-7%) -17.7 (-16%) Net-Zero Pathway t-Zero Pathway Net-Zero Pathway 65 20 22 24 26 28 30 32 34 36 38 40 2022 2024 2026 2028 2030 2032 2034 20 120 220 320 420 520 620 720 2022 2024 2026 2028 2030 2032 2034 Frozen Policy (Ja Current Policies ( Current Policies ( Current Policies ( Net-Zero Pathwa Modeled U.S. Dry Natural Gas Production trillion cubic feet per year (Tcf/year)1 Modeled U.S. Coal Production million short tons per year (million t/year)1 1 – Modeled U.S. '21) Current Policies (Conservative) Current Policies (Mid-range) Current Policies (Optimistic) Net-Zero Pathway Modeled U.S. Crude Oil and Natural Gas Liquids Production billion barrels of oil equivalent per year (billion boe/year)1 d U.S. fossil fuel production volumes assume export volumes remain constant at levels from the EIA Annual Energy Outlook 2022 (AEO2022) Reference scenario and in domestic fuel consumption reduce both domestic production and imports in proportion to the historical share of domestic and imported supply in U.S. If d ti ti i b AEO2022 l l ll i t l d d i d t b ti fi d b d ti d ti 5 0 5.5 6.0 6.5 7.0 Frozen Policy (Jan. '21) Current Policies (Conservative) Current Policies (Mid-range) Current Policies (Optimistic) Net-Zero Pathway Modeled U.S. Crude Oil and Natural Gas Liquids Production billion barrels of oil equivalent per year (billion boe/year)1 Modeled U.S. Crude Oil and Natural Gas Liquids Production billion barrels of oil equivalent per year (billion boe/year)1 67 4 2026 2028 2030 2032 2034 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 2022 2024 2026 2028 2030 2032 2034 Frozen Policy (Jan. '21) Current Policies (Conservative) Current Policies (Mid-range) Current Policies (Optimistic) Net-Zero Pathway S. Crude Oil Production r year (billion bbl/year)1 Modeled U.S. Crude Oil and Natural Gas Liquids Production billion barrels of oil equivalent per year (billion boe/year)1 1 – Modeled U.S. fossil fuel production volumes assume export volumes remain constant at levels from the EIA Annual Energy Outlook 2022 (AEO2022) Reference scenario and reductions in domestic fuel consumption reduce both domestic production and imports in proportion to the historical share of domestic and imported supply in U.S. consumption. If domestic consumption increases above AEO2022 levels, all incremental demand is assumed to be satisfied by domestic production. 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 2022 2024 2026 2028 2030 2032 2034 Modeled U.S. Crude Oil Production billion barrels per year (billion bbl/year)1 1 – Modeled reductions i consumptio 5 0 5.5 6.0 6.5 7.0 Frozen Policy (Jan. '21) Current Policies (Conservative) Current Policies (Mid-range) Current Policies (Optimistic) Net-Zero Pathway Modeled U.S. Crude Oil and Natural Gas Liquids Production billion barrels of oil equivalent per year (billion boe/year)1 Modeled U.S. Crude Oil Production billion barrels per year (billion bbl/year)1 67 1 – Modeled U.S. fossil fuel production volumes assume export volumes remain constant at levels from the EIA Annual Energy Outlook 2022 (AEO2022) Refere reductions in domestic fuel consumption reduce both domestic production and imports in proportion to the historical share of domestic and imported supply consumption. If domestic consumption increases above AEO2022 levels, all incremental demand is assumed to be satisfied by domestic production. Modeled U.S. Coal Production million short tons per year (million t/year)1 Modeled U.S. Dry Natural Gas Production trillion cubic feet per year (Tcf/year)1 20 22 24 26 28 30 32 34 36 38 40 2022 2024 2026 2028 2030 2032 2034 Modeled U.S. Dry Natural Gas Production trillion cubic feet per year (Tcf/year)1 1 – Modeled reductions i consumptio 20 120 220 320 420 520 620 720 2022 2024 2026 2028 2030 2032 2034 Frozen Policy (Jan. '21) Current Policies (Conservative) Current Policies (Mid-range) Current Policies (Optimistic) Net-Zero Pathway Modeled U.S. Coal Production million short tons per year (million t/year)1 66 2028 2030 2032 2034 20 120 220 320 420 520 620 720 2022 2024 2026 2028 2030 2032 2034 Frozen Policy (Jan. '21) Current Policies (Conservative) Current Policies (Mid-range) Current Policies (Optimistic) Net-Zero Pathway ural Gas Production /year)1 Modeled U.S. Coal Production million short tons per year (million t/year)1 1 – Modeled U.S. fossil fuel production volumes assume export volumes remain constant at levels from the EIA Annual Energy Outlook 2022 (AEO2022) Reference scenario and reductions in domestic fuel consumption reduce both domestic production and imports in proportion to the historical share of domestic and imported supply in U.S. consumption. If domestic consumption increases above AEO2022 levels, all incremental demand is assumed to be satisfied by domestic production. Frozen Policy (Jan. '21) Current Policies (Conservative) Current Policies (Mid-range) Current Policies (Optimistic) Net-Zero Pathway 66 1 – Modeled U.S. fossil fuel production volumes assume export volumes remain constant at levels from the EIA Annual Energy Outlook 2022 (AEO2022) Reference scenario and reductions in domestic fuel consumption reduce both domestic production and imports in proportion to the historical share of domestic and imported supply in U.S. consumption. If domestic consumption increases above AEO2022 levels, all incremental demand is assumed to be satisfied by domestic production. 66 6 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 2022 2024 2026 2028 2030 2032 2034 Frozen Policy (Jan. fossil fuel production volumes assume export volumes remain constant at levels from the EIA Annual Energy Outlook 2022 (AEO2022) Reference scenario and reductions in domestic fuel consumption reduce both domestic production and imports in proportion to the historical share of domestic and imported supply in U.S. consumption. If domestic consumption increases above AEO2022 levels, all incremental demand is assumed to be satisfied by domestic production. 67 68 0 200 400 600 800 1000 1200 1400 1600 1800 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 C CO2 Emissions from Electricity Generation million metric tons per year (MMt/year) % below 2022: 38% 51% 44% 54% 53-57% 77-79% 88% 92% af fr Ele a Pa M m 75% below 2022 CO2 Emissions from Electricity Generation million metric tons per year (MMt/year) 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Net-Zero Pathway Current Policies, incl. IRA Frozen Policies (Jan. ‘21) Infrastructure Law Only y ons per year (MMt/year) % below 2022: 38% 51% 44% 54% 53-57% 77-79% 88% 92% Under IRA, clean electricity production and investment tax credits begin phasing out only after electricity sector CO2 emissions decline 75% from 2022 levels. This threshold is reached under Current Policies scenarios around 2035. Electricity sector emissions under Current Policies are considerably higher than under the Net-Zero Pathway (~500-550 MMt/y in 2030 and ~200-240 MMt/y in 2035). Coal power plants retire much more rapidly under a cost-optimized trajectory to reach U.S. climate goals. 75% below 2022 Under IRA, clean electricity production and investment tax credits begin phasing out only after electricity sector CO2 emissions decline 75% from 2022 levels. This threshold is reached under Current Policies scenarios around 2035. Electricity sector emissions under Current Policies are considerably higher than under the Net-Zero Pathway (~500-550 MMt/y in 2030 and ~200-240 MMt/y in 2035). Coal power plants retire much more rapidly under a cost-optimized trajectory to reach U.S. climate goals. Electricity sector emissions under Current Policies are considerably higher than under the Net-Zero Pathway (~500-550 MMt/y in 2030 and ~200-240 MMt/y in 2035). Coal power plants retire much more rapidly under a cost-optimized trajectory to reach U.S. climate goals. 69 CO2 Emissions from Electricity Generation by Source/Sink million metric tons per year (MMt/year) 25 11 238 151 -89 -47 174 114 -100 100 300 500 700 900 1,100 1,300 1,500 1,700 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 392 184 284 139 675 323 -100 100 300 500 700 900 1,100 1,300 1,500 1,700 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 434 209 307 160 740 369 -100 100 300 500 700 900 1,100 1,300 1,500 1,700 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 441 212 254 124 693 336 -100 100 300 500 700 900 1,100 1,300 1,500 1,700 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 614 462 367 313 980 774 -100 100 300 500 700 900 1,100 1,300 1,500 1,700 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 missions from Electricity Generation by Source/Sink metric tons per year (MMt/year) 024 551 441 64 331 285 588 882 726 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Frozen Policies (Jan ’21) Net-Zero Pathway Current Policies Scenarios Mid-range Optimistic Conservative Infrastructure Law Only 1 700 Current Policies Scenarios Mid-range Net-Zero Pathway 1, Optimistic -100 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 -100 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 -100 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 -100 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 -100 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 100 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 70 Electricity Generation by Resource thousand terawatt-hours per year (TWh/year) Frozen Policies (Jan ’21) Electricity Generation by Resource thousand terawatt-hours per year (TWh/year) Frozen Policies (Jan ’21) 53% 69% 81% 91% 93% 95% 0 % 1 0% 2 0% 3 0% 4 0% 5 0% 6 0% 7 0% 8 0% 9 0% 1 0 % 0 1 2 3 4 5 6 7 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 50% 62% 67% 77% 83% 91% 0 % 1 0% 2 0% 3 0% 4 0% 5 0% 6 0% 7 0% 8 0% 9 0% 1 0 % 0 1 2 3 4 5 6 7 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 47% 55% 62% 75% 80% 89% 0 % 1 0% 2 0% 3 0% 4 0% 5 0% 6 0% 7 0% 8 0% 9 0% 1 0 % 0 1 2 3 4 5 6 7 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 50% 60% 66% 77% 81% 90% 0 % 1 0% 2 0% 3 0% 4 0% 5 0% 6 0% 7 0% 8 0% 9 0% 1 0 % 0 1 2 3 4 5 6 7 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 5% 59% 67% 71% 75% 0 % 1 0% 2 0% 3 0% 4 0% 5 0% 6 0% 7 0% 8 0% 9 0% 1 0 % 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Generation by Resource watt-hours per year (TWh/year) Current Policies Scenarios en Policies (Jan ’21) Mid-range Optimistic Net-Zero Pathway Conservative Current Policies Scenarios Mid-range Current Policies Scenarios Mid-range 7 cenarios Conservative Optimistic Net-Zero Pathway Conservative Optimistic 81% 0 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 0 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 0 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 71 188 40 21 10 6 5 489 494 481 463 443 433 102 105 78 78 98 103 181 235 313 426 545 583 98 125 101 151 224 329 480 790 28 35 51 119 167 283 0 500 1,000 1,500 2,000 2,500 3,000 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 188 145 119 102 90 73 496 468 469 464 440 439 95 85 78 78 18 41 208 317 371 455 515 586 124 158 116 170 249 381 604 1,016 29 43 73 119 190 369 0 500 1,000 1,500 2,000 2,500 3,000 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 190 154 126 108 96 77 496 479 480 477 464 453 95 94 78 78 18 41 181 235 313 426 517 590 127 161 101 151 224 329 480 807 28 36 58 91 141 280 0 500 1,000 1,500 2,000 2,500 3,000 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 190 159 131 112 102 79 493 452 449 446 426 413 95 92 78 78 18 41 204 287 346 425 483 528 125 158 113 171 249 357 503 816 28 39 62 98 141 267 0 500 1,000 1,500 2,000 2,500 3,000 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 72 city by Resource Current Policies Scenarios icies (Jan ’21) Mid-range Optimistic Net-Zero Pathway Conservative 1 - Plot excludes small quantities of capacity of other renewables (biomass, geothermal, landfill gas, municipal waste and solar thermal), other fossil (petroleum, fossil waste) and fuel cells. Energy Information Administration’s Short-term Energy Outlook, the solar industry is currently on track and expected to average 41 GW/year of new capacity in 2023 and 2024, while the onshore wind industry lags behind our modeled pace, with an expected rate of just 7 GW/year over 2023 and 2024.3 Variation in modeled capacity across Current Policies scenarios from 2023-2030 primarily reflects exogenous constraints on annual capacity additions designed to approximate several challenges that are difficult to model explicitly but may limit renewable energy deployment, including the ability to site and permit projects at requisite pace and scale, expand supply chains, interconnect generating capacity, and hire and train the expanded energy workforce to build these projects. See p. 32 for details on these constraints. Capacity additions in the 2031-2035 period are generally not limited by exogenous constraints and reflect economic equilibrium outcomes. 15 4 59 10 6 15 19 4 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 storage solar PV (all) onshore wind offshore wind hydro nuclear natural gas w/CCS natural gas other fossil coal 73 1 - Historical data labels represent peak historical annual additions (source, 1950-2020: EIA Form 860; 2021-2022: EIA Short-term Energy Outlook, June 6, 2023) 2 - Modeled data labels reflect range across Current Policies scenarios (Conservative, Mid-range and Optimistic). Depicted bars represent results for the Mid-range scenario. 3 - EIA Short-term Energy Outlook, June 6, 2023 edition cal Annual Electricity Capacity Additions vs. Modeled Annual Average Capacity Additions Under Current Policies Scenarios gawatts/year (GW/year) Historical (EIA)1 Modeled (REPEAT Project)2 2023-2030 2031-2035 Under Current Policies scenarios, wind and solar PV capacity additions set new records over the coming decade. The average annual rate of solar PV additions more than doubles from a 19 GW peak in 2021 to an average rate of 44-51 GW/year from 2023-2030. Onshore wind capacity additions reach 39-43 GW/year from 2023-2030, nearly triple the peak of 15 GW in 2020. Solar PV additions increase further to 123-167 GW/year on average from 2031-2035, while wind additions remain roughly steady at 26-41 GW/year during that period. According to the U.S. Energy Information Administration’s Short-term Energy Outlook, the solar industry is currently on track and expected to average 41 GW/year of new capacity in 2023 and 2024, while the onshore wind industry lags behind our modeled pace, with an expected rate of just 7 GW/year over 2023 and 2024.3 Variation in modeled capacity across Current Policies scenarios from 2023-2030 primarily reflects exogenous constraints on annual capacity additions designed to approximate several challenges that are difficult to model explicitly but may limit renewable energy deployment, including the ability to site and permit projects at requisite pace and scale, expand supply chains, interconnect generating capacity, and hire and train the expanded energy workforce to build these projects. See p. 32 for details on these constraints. Capacity additions in the 2031-2035 period are generally not limited by exogenous constraints and reflect economic equilibrium outcomes. 8-12 42-62 44-51 123-167 39-43 26-41 3 6 9-13 2-3 Historical Annual Electricity Capacity Additions vs. Modeled Annual Average Capacity Additions Under Current Policies Scenarios average gigawatts/year (GW/year) 250 Under Current Policies scenarios, wind and solar PV capacity additions set new records over the coming decade. The average annual rate of solar PV additions more than doubles from a 19 GW peak in 2021 to an average rate of 44-51 GW/year from 2023-2030. Onshore wind capacity additions reach 39-43 GW/year from 2023-2030, nearly triple the peak of 15 GW in 2020. Solar PV additions increase further to 123-167 GW/year on average from 2031-2035, while wind additions remain roughly steady at 26-41 GW/year during that period. According to the U.S. Energy Information Administration’s Short-term Energy Outlook, the solar industry is currently on track and expected to average 41 GW/year of new capacity in 2023 and 2024, while the onshore wind industry lags behind our modeled pace, with an expected rate of just 7 GW/year over 2023 and 2024.3 According to the U.S. 134 118 113 100 480 486 486 483 95 94 78 78 18 41 255 285 304 325 32 32 214 310 417 508 34 47 70 132 2028 2029 2030 2031 2032 2033 2034 2035 Current Policies Scenarios Mid-range Net-Zero Pathway Conservative Optimistic 3,000 1 - Plot excludes small quantities of capacity of other renewables (biomass, geothermal, landfill gas, municipal waste and solar thermal), other fossil (petroleum, fossil waste) and fuel cells. 1 - Plot excludes small quantities of capacity of other renewables (biomass, geothermal, landfill gas, municipal waste and solar thermal), other fossil (petroleum, fossil waste) and fuel cells. 1 - Plot excludes small quantities of capacity of other renewables (biomass, geothermal, landfill gas, municipal waste and solar thermal), other fossil (petroleum, fossil waste) and fuel cells. 72 15 4 59 10 6 0 50 100 150 200 250 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 1 - Historical data labels represent peak historical annual additions (source, 1950-2020: EIA Form 860; 20 2 - Modeled data labels reflect range across Current Policies scenarios (Conservative, Mid-range and Opt 3 - EIA Short-term Energy Outlook, June 6, 2023 edition Historical Annual Electricity Capacity Additions vs. Modeled Annual Average Capacity Additions Un average gigawatts/year (GW/year) Under Current Policies scenarios, wind and solar PV capacity additions set new records over the coming decade. The average annual rate of solar PV additions more than doubles from a 19 GW peak in 2021 to an average rate of 44-51 GW/year from 2023-2030. Onshore wind capacity additions reach 39-43 GW/year from 2023-2030, nearly triple the peak of 15 GW in 2020. Solar PV additions increase further to 123-167 GW/year on average from 2031-2035, while wind additions remain roughly steady at 26-41 GW/year during that period. According to the U.S. '21) Conservative Mid-range Optimistic Net-Zero Pathway Inter-regional transmission Interconnection lines - offshore wind Interconnection lines - onshore wind Interconnection lines - solar 2030 2035 35,000 35,300 41,500 72,700 23,600 56,200 63,300 71,600 100,100 To keep up with growing electricity demand and connect to the best wind and solar resources across the country, the pace of electricity transmission capacity more than doubles under Current Policies scenarios relative to the Frozen Policies scenario (and increases more than four-fold under the Net- Zero Pathway). High voltage transmission capacity expands roughly 15-18% from 2020 to 2030 under Current Policies and about 24-31% by 2035.1 Transmission capacity expands 32% by 2030 and 43% by 2035 under the Net-Zero Pathway. The majority of new transmission capacity is driven by the need to interconnect wind farms. Wind power quality varies much more across sites than it does for solar PV, and populated areas tend to be farther from the best wind resources. That means solar PV projects have greater siting flexibility and generally require less long-distance transmission per gigawatt. Current Policies Scenarios Current Policies Scenarios 1 - Estimated total U.S. transmission capacity stood at roughly 232,000 GW-miles in 2020. This is based on ~200,000 GW-miles circa 2008 as reported by Homeland Infrastructure Foundation Layer Database (2008) and cited in NREL “Renewable Electricity Futures Study” (2012) p. 26 and assuming 1.2% annual average growth in transmission capacity, consistent with the historical average from 2004-2016 reported by UT Austin and cited in Cembalest (2022), “Eye on the Market 12th Edition: 2022 Annual Energy Paper” p.12. Interconnection lines - offshore wind Interconnection lines - solar Interconnection lines - offshore wind Interconnection lines - solar 120,000 2030 To keep up with growing electricity demand and connect to the best wind and solar resources across the country, the pace of electricity transmission capacity more than doubles under Current Policies scenarios relative to the Frozen Policies scenario (and increases more than four-fold under the Net- Zero Pathway). 1,200 9,800 10,300 13,400 36,600 1,900 11,100 12,600 16,100 41,700 1,600 1,600 1,600 1,600 12,600 4,600 4,600 4,600 4,600 13,400 10,100 21,200 21,400 23,800 21,500 12,800 30,200 35,700 35,300 35,800 2,000 2,500 2,100 2,700 2,100 4,300 10,400 10,400 15,600 9,200 - 20,000 40,000 60,000 80,000 100,000 Frozen Policies (Jan. '21) Conservative Mid-range Optimistic Net-Zero Pathway Frozen Policies (Jan. Energy Information Administration’s Short-term Energy Outlook, the solar industry is c on track and expected to average 41 GW/year of new capacity in 2023 and 2024, while the onshore wind in lags behind our modeled pace, with an expected rate of just 7 GW/year over 2023 and 2024.3 storage solar PV (all) onshore wind offshore wind hydro nuclear natural gas w/CCS natural gas other fossil coal Variation in modeled capacity across Current Policies scenarios from 2023-2030 primarily reflects exogenous constraints on annual capacity additions designed to approximate several challenges that are difficult to model explicitly but may limit renewable energy deployment, including the ability to site and permit projects at requisite pace and scale, expand supply chains, interconnect generating capacity, and hire and train the expanded energy workforce to build these projects. See p. 32 for details on these constraints. Capacity additions in the 2031-2035 period are generally not limited by exogenous constraints and reflect economic equilibrium outcomes. 50 1 - Historical data labels represent peak historical annual additions (source, 1950-2020: EIA Form 860; 2021-2022: EIA Short-term Energy Outlook, June 6, 2023) 2 - Modeled data labels reflect range across Current Policies scenarios (Conservative, Mid-range and Optimistic). Depicted bars represent results for the Mid-range scenario. 3 - EIA Short-term Energy Outlook, June 6, 2023 edition 1 - Historical data labels represent peak historical annual additions (source, 1950-2020: EIA Form 860; 2021-2022: EIA Short-term Energy Outlook, June 6, 2023) 2 - Modeled data labels reflect range across Current Policies scenarios (Conservative, Mid-range and Optimistic). Depicted bars represent results for the Mid-range scenario. 3 - EIA Short-term Energy Outlook, June 6, 2023 edition 73 ross Current Policies scenarios (Conservative, Mid-range and Optimistic). Depicted bars represent results for the Mid-range scenario. 6 2023 edition 74 ) 9,800 10,300 13,400 36,600 1,900 11,100 12,600 16,100 41,700 1,600 1,600 1,600 12,600 4,600 4,600 4,600 4,600 13,400 21,200 21,400 23,800 21,500 12,800 30,200 35,700 35,300 35,800 2,500 2,100 2,700 2,100 4,300 10,400 10,400 15,600 9,200 Conservative Mid-range Optimistic Net-Zero Pathway Frozen Policies (Jan. '21) Conservative Mid-range Optimistic Net-Zero Pathway 14,900 35,000 35,300 41,500 72,700 23,600 56,200 63,300 71,600 100,100 the country, the capacity more th scenarios relativ (and increases m High voltage trans 15-18% from 20 and about 24-31% expands 32% by N The majority of n by the need to int quality varies mu for solar PV, and from the best win projects have gre require less long-d Current Policies Scenarios Current Policies Scenarios 1 - Estimated total U.S. transmission capacity stood at roughly 232,000 GW-miles in 2020. This is based on ~200,000 GW-miles circa 2008 as re 80,000 High voltage transmission capacity expands roughly 15-18% from 2020 to 2030 under Current Policies and about 24-31% by 2035.1 Transmission capacity expands 32% by 2030 and 43% by 2035 under the Net-Zero Pathway. High voltage transmission capacity expands roughly 15-18% from 2020 to 2030 under Current Policies and about 24-31% by 2035.1 Transmission capacity expands 32% by 2030 and 43% by 2035 under the Net-Zero Pathway. 60,000 The majority of new transmission capacity is driven by the need to interconnect wind farms. Wind power quality varies much more across sites than it does for solar PV, and populated areas tend to be farther from the best wind resources. That means solar PV projects have greater siting flexibility and generally require less long-distance transmission per gigawatt. The majority of new transmission capacity is driven by the need to interconnect wind farms. Wind power quality varies much more across sites than it does for solar PV, and populated areas tend to be farther from the best wind resources. That means solar PV projects have greater siting flexibility and generally require less long-distance transmission per gigawatt. The majority of new transmission capacity is driven by the need to interconnect wind farms. Wind power quality varies much more across sites than it does for solar PV, and populated areas tend to be farther from the best wind resources. That means solar PV projects have greater siting flexibility and generally require less long-distance transmission per gigawatt. 1 - Estimated total U.S. transmission capacity stood at roughly 232,000 GW-miles in 2020. This is based on ~200,000 GW-miles circa 2008 as reported by Homeland Infrastructure Foundation Layer Database (2008) and cited in NREL “Renewable Electricity Futures Study” (2012) p. 26 and assuming 1.2% annual average growth in transmission capacity, consistent with the historical average from 2004-2016 reported by UT Austin and cited in Cembalest (2022), “Eye on the Market 12th Edition: 2022 Annual Energy Paper” p.12. 1 - Estimated total U.S. transmission capacity stood at roughly 232,000 GW-miles in 2020. This is based on ~200,000 GW-miles circa 2008 as repor Foundation Layer Database (2008) and cited in NREL “Renewable Electricity Futures Study” (2012) p. 26 and assuming 1.2% annual average growt consistent with the historical average from 2004-2016 reported by UT Austin and cited in Cembalest (2022), “Eye on the Market 12th Edition: 202 ‘Downscaled’ Wind, Solar & Transmission Expansion Through 2030 Frozen Policies (Jan. ‘21) Mapped wind and solar deployment locations reflect geo-spatially resolved land availability screening methods and least-cost siting algorithm. This is one of many possible patterns of wind and solar deployment consistent with modeled macro-energy systems outcomes and is presented for illustrative purposes only. ‘Downscaled’ Wind, Solar & Transmission Expansion Through 2030 Frozen Policies (Jan. ‘21) 75 ‘Downscaled’ Wind, Solar & Transmission Expansion Through 2030 Current Policies (Mid-range) Mapped wind and solar deployment locations reflect geo-spatially resolved land availability screening methods and least-cost siting algorithm. This is one of many possible patterns of wind and solar deployment consistent with modeled macro-energy systems outcomes and is presented for illustrative purposes only. ‘Downscaled’ Wind, Solar & Transmission Expansion Through 2030 Current Policies (Mid-range) locations vailability iting algorithm. ns of wind and solar d macro-energy systems rative purposes only. ‘Downscaled’ Wind, Solar & Transmission Expansion Through 2030 Current Policies (Mid-range) ‘Downscaled’ Wind, Solar & Transmission Expansion Through 2030 Current Policies (Mid-range) 76 ‘Downscaled’ Wind, Solar & Transmission Expansion Through 2030 Net-Zero Pathway Mapped wind and solar deployment locations reflect geo-spatially resolved land availability screening methods and least-cost siting algorithm. This is one of many possible patterns of wind and solar deployment consistent with modeled macro-energy systems outcomes and is presented for illustrative purposes only. ‘Downscaled’ Wind, Solar & Transmission Expansion Through 2030 Net-Zero Pathway locations vailability ting algorithm. s of wind and solar d macro-energy systems rative purposes only. ‘Downscaled’ Wind, Solar & Transmission Expansion Through 2030 Net-Zero Pathway 77 77 ‘Downscaled’ Wind, Solar & Transmission Expansion Through 2035 Frozen Policies (Jan. ‘21) Mapped wind and solar deployment locations reflect geo-spatially resolved land availability screening methods and least-cost siting algorithm. This is one of many possible patterns of wind and solar deployment consistent with modeled macro-energy systems outcomes and is presented for illustrative purposes only. ‘Downscaled’ Wind, Solar & Transmission Expansion Through 2035 Frozen Policies (Jan. ‘21) locations vailability ting algorithm. ns of wind and solar d macro-energy systems rative purposes only. ‘Downscaled’ Wind, Solar & Transmission Expansion Through 2035 Frozen Policies (Jan. ‘21) 78 ‘Downscaled’ Wind, Solar & Transmission Expansion Through 2035 Current Policies (Mid-range) Mapped wind and solar deployment locations reflect geo-spatially resolved land availability screening methods and least-cost siting algorithm. This is one of many possible patterns of wind and solar deployment consistent with modeled macro-energy systems outcomes and is presented for illustrative purposes only. ‘Downscaled’ Wind, Solar & Transmission Expansion Through 2035 Current Policies (Mid-range) ‘Downscaled’ Wind, Solar & Transmission Expansion Through 2035 Current Policies (Mid-range) 79 ‘Downscaled’ Wind, Solar & Transmission Expansion Through 2035 Net-Zero Pathway Mapped wind and solar deployment locations reflect geo-spatially resolved land availability screening methods and least-cost siting algorithm. This is one of many possible patterns of wind and solar deployment consistent with modeled macro-energy systems outcomes and is presented for illustrative purposes only. ‘Downscaled’ Wind, Solar & Transmission Expansion Through 2035 Net-Zero Pathway locations vailability ting algorithm. s of wind and solar d macro-energy systems rative purposes only. ‘Downscaled’ Wind, Solar & Transmission Expansion Through 2035 Net-Zero Pathway 80 Compound Annual Growth in Electricity Transmission Capacity, 2020-2035 vs. Historical Periods percent annual growth in gigawatt-miles 1.2% 2.4% 9 2004-2016 Current Policies Net-Zero Pathway 1.5-1.8% 1 2 1 - Reported by U.S. Department of Energy and cited in Cembalest (2022), “Eye on the Market 12th Edition: 2022 Annual Energy Paper” p.12. 2 - Reported by UT Austin and cited in Cembalest (2022), p.12. To achieve the maximum emissions reduction under Current Policies, U.S. transmission capacity must expand roughly 50% faster through 2035 than the recent historical rate.2 While our modeling finds this outcome makes economic sense given incentives under IRA, current U.S. transmission planning, siting, permitting and cost allocation practices can all potentially impede the real-world pace of transmission expansion. We explore the impacts of more constrained transmission expansion on the following page. Note that U.S. electricity demand has been roughly flat since the mid-2000s, and modeled transmission expansion rates under Current Policies are roughly equal to the historical pace achieved from the 1970s to the 1990s1, the last period during which U.S. electricity demand steadily increased. The pace of transmission expansion under the Net-Zero Pathway exceeds the historical 1978-1999 rate and is twice as fast as the more recent 2004-2016 period. 8 1.9% 1.2% 2.4% 1978-1999 2004-2016 Current Policies Net-Zero Pathway 1.5-1.8% 1 2 1 - Reported by U.S. Department of Energy and cited in Cembalest (2022), “Eye on the Market 12th Edition: 2022 Annual Energy Paper” p.12. 2 - Reported by UT Austin and cited in Cembalest (2022), p.12. To achieve the maximum emissions reduction under Current Policies, U.S. transmission capacity must expand roughly 50% faster through 2035 than the recent historical rate.2 While our modeling finds this outcome makes economic sense given incentives under IRA, current U.S. transmission planning, siting, permitting and cost allocation practices can all potentially impede the real-world pace of transmission expansion. We explore the impacts of more constrained transmission expansion on the following page. Note that U.S. electricity demand has been roughly flat since the mid-2000s, and modeled transmission expansion rates under Current Policies are roughly equal to the historical pace achieved from the 1970s to the 1990s1, the last period during which U.S. electricity demand steadily increased. The pace of transmission expansion under the Net-Zero Pathway exceeds the historical 1978-1999 rate and is twice as fast as the more recent 2004-2016 period. To achieve the maximum emissions reduction under Current Policies, U.S. transmission capacity must expand roughly 50% faster through 2035 than the recent historical rate.2 While our modeling finds this outcome makes economic sense given incentives under IRA, current U.S. transmission planning, siting, permitting and cost allocation practices can all potentially impede the real-world pace of transmission expansion. We explore the impacts of more constrained transmission expansion on the following page. Note that U.S. electricity demand has been roughly flat since the mid-2000s, and modeled transmission expansion rates under Current Policies are roughly equal to the historical pace achieved from the 1970s to the 1990s1, the last period during which U.S. electricity demand steadily increased. The pace of transmission expansion under the Net-Zero Pathway exceeds the historical 1978-1999 rate and is twice as fast as the more recent 2004-2016 period. Net-Zero Pathway 1 - Reported by U.S. Department of Energy and cited in Cembalest (2022), “Eye on the Market 12th Edition: 2022 Annual Energy Paper” p.12 2 - Reported by UT Austin and cited in Cembalest (2022), p.12. 1 - Reported by U.S. Department of Energy and cited in Cembalest (2022), “Eye on the Market 12th Edition: 2022 Annual Energy Paper” p.12 2 - Reported by UT Austin and cited in Cembalest (2022), p.12. 1 - Reported by U.S. Department of Energy and cited in Cembalest (2022), “Eye on the Market 12th Edition: 2022 Annual Energy Paper” p.12. 2 - Reported by UT Austin and cited in Cembalest (2022), p.12. ” p.12. 81 odeled Net U.S. Greenhouse Gas Emissions (Including Land Carbon Sinks) Impact of Transmission Expansion Constraints on Modeled Net U.S. Greenhouse Gas Emissions (Including Land Carbon Sinks) billion metric tons CO2-equivalent (Gt CO2-e)1 Impact of Transmission Expansion Constraints on Modeled Net U.S. Greenhouse Gas Emissions (Includ billion metric tons CO2-equivalent (Gt CO2-e)1 nts on Modeled Net U.S. Greenhouse Gas Emissions (Including Land Carbon Sinks) 4.5 3.4 3.9 2.5 2032 2033 2034 2035 4.8 4.5 4.0 3.1 4.4 3.8 5.6 3.2 2.5 2 3 4 5 6 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Current Policies (Optimistic) Net-Zero Pathway Current Policies, incl. IRA Frozen Policies (Jan. ‘21) Current Policies, Transmission Constrained2 Failing to accelerate transmission expansion beyond the recent historical pace (~1%/year)2 could forfeits about half of the emissions reductions that might otherwise be achieved under Current Policies. 82 p ( g ) nt (Gt CO2-e)1 4.8 4.5 4.2 3.4 4.5 3.9 3.2 2.5 2027 2028 2029 2030 2031 2032 2033 2034 2035 4.8 4.5 4.0 3.1 4.4 3.8 5.6 3.2 2.5 2 3 4 5 6 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 1 - CO2-equivalent emissions calculations use IPCC AR4 100 year global warming potential as per EPA Inventory of Greenhouse Gas Emissions and Sinks. 2 - Transmission constrained cases limit total transmission capacity expansion to a compound annual growth rate of 1%/year, roughly equivalent to the 2004-2016 average historical pace. The maximum total increase in GW-miles for each model year is allocated as constraints on expansion of inter-regional transmission, interconnection lines for wind and interconnection lines for solar PV respectively in proportion to the total expansion for each category of lines under unconstrained Current Policies cases. rrent Policies (Mid-range) Current Policies (Optimistic) Net-Zero Pathway Current Policies, incl. IRA Frozen Policies (Jan. ‘21) Current Policies, Transmission Constrained2 Failing to accelerate transmission expansion beyond the recent historical pace (~1%/year)2 could forfeits about half of the emissions reductions that might otherwise be achieved under Current Policies. Current Policies (Optimistic) 4.8 4.5 4.2 3.4 4.5 3.9 5.6 3.2 2.5 2 3 4 5 6 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2 3 4 5 6 1 - CO2-equivalent emissions calc 2 - Transmission constrained cas historical pace. The maximum to Current Policies (Mid-range) Current Policies (Mid-range) Failing to accelerate transmission expansion beyond the recent historical pace (~1%/year)2 could forfeits about half of the emissions reductions that might otherwise be achieved under Current Policies. 6 5 5 4 3.1 3 2 2035 1 - CO2-equivalent emissions calculations use IPCC AR4 100 year global warming potential as per EPA Inventory of Greenhouse Gas Emissions an 2 - Transmission constrained cases limit total transmission capacity expansion to a compound annual growth rate of 1%/year, roughly equivalent to the 2004-2016 average historical pace. The maximum total increase in GW-miles for each model year is allocated as constraints on expansion of inter-regional transmission, interconnection lines for wind and interconnection lines for solar PV respectively in proportion to the total expansion for each category of lines under unconstrained Current Policies cases. 2 - Transmission constrained cases limit total transmission capacity expansion to a compound annual growth rate of 1%/year, roughly equivale historical pace. The maximum total increase in GW-miles for each model year is allocated as constraints on expansion of inter-regional transmi for wind and interconnection lines for solar PV respectively in proportion to the total expansion for each category of lines under unconstrained -50% -41% 1% -53% -34% -43% -51% -63% Mid-range Mid-Range, Transmission Constrained Optimistic Optimistic, Transmission Constrained Net-Zero Pathway 8 Expansion Constraints on Modeled Net U.S. Greenhouse Gas Emissions (Including Land Carbon Sinks) ns1 1 - CO2-equivalent emissions calculations use IPCC AR4 100 year global warming potential as per EPA Inventory of Greenhouse Gas Emissions and Sinks. 2 - Transmission constrained cases limit total transmission capacity expansion to a compound annual growth rate of 1%/year, roughly equivalent to the 2004-2016 average historical pace. The maximum total increase in GW-miles for each model year is allocated as constraints on expansion of inter-regional transmission, interconnection lines for wind and interconnection lines for solar PV respectively in proportion to the total expansion for each category of lines under unconstrained Current Policies cases. If the United States cannot build new transmission capacity faster than the recent historical pace (~1%/year)2, the ‘emissions gap’ between Current Policies scenarios and U.S. climate goals widens to over 1 billion metric tons in 2030. 030 2035 2 2 1,070 MMT gap target Modeled Net U.S. Greenhouse Gas Emissions (Including Land Carbon Sinks t of Transmission Expansion Constraints on Modeled Net U.S. Greenhouse hange vs. 2005 emissions1 Impact of Transmission Expansion Constraints on Modeled Net U.S. Greenhouse Gas Emissions (Including Land C percent change vs. 2005 emissions1 Impact of Transmission Expansion Constraints on Modeled Net U.S. Greenhouse Gas Emissions (Including Land Carbon Sinks) percent change vs. 2005 emissions1 -38% -50% -33% -41% -41% -53% -34% -43% -51% -63% -70% -60% -50% -40% -30% -20% -10% 0% Mid-range Mid-Range, Transmission Constrained Optimistic Optimistic, Transmission Constrained Net-Zero Pathway Impact of Transmission Expansion Constraints on Modeled Net U.S. Greenhouse Gas Emissions (Including Land Carbo percent change vs. 2005 emissions1 1 - CO2-equivalent emissions calculations use IPCC AR4 100 year global warming potential as per EPA Inventory of Greenhouse Gas Emissions and 2 - Transmission constrained cases limit total transmission capacity expansion to a compound annual growth rate of 1%/year, roughly equivalent historical pace. The maximum total increase in GW-miles for each model year is allocated as constraints on expansion of inter-regional transmiss for wind and interconnection lines for solar PV respectively in proportion to the total expansion for each category of lines under unconstrained C If the United States cannot build capacity faster than the recen (~1%/year)2, the ‘emissions gap Policies scenarios and U.S. clima over 1 billion metric ton 2030 2035 2 2 1,070 MMT gap 1,140 MMT gap 2030 target -38% -50% -33% -41% -41% -53% -34% -43% -51% -63% Mid-range Mid-Range, Transmission Constrained Optimistic Optimistic, Transmission Constrained Net-Zero Pathway t change vs. 2005 emissions1 1 - CO2-equivalent emissions calculations use IPCC AR4 100 year global warming potential as per EPA Inventory of Greenhouse Gas Emissions and Sinks. 2 Transmission constrained cases limit total transmission capacity expansion to a compound annual growth rate of 1%/year roughly equivalent to the 2004 2016 average If the United States cannot build new transmissi capacity faster than the recent historical pace (~1%/year)2, the ‘emissions gap’ between Curre Policies scenarios and U.S. climate goals widens over 1 billion metric tons in 2030. 2030 2035 2 2 1,070 MMT gap 1,140 MMT gap 2030 target -38% -33% -41% -34% -51% -70% -60% -50% -40% -30% -20% -10% 0% 2030 1,070 MMT gap 1,140 MMT gap 2030 target 0% If the United States cannot build new transmission capacity faster than the recent historical pace (~1%/year)2, the ‘emissions gap’ between Current Policies scenarios and U.S. climate goals widens to over 1 billion metric tons in 2030. 1 - CO2-equivalent emissions calculations use IPCC AR4 100 year global warming potential as per EPA Inventory of Greenhouse Gas Emissions and Sinks. 2 - Transmission constrained cases limit total transmission capacity expansion to a compound annual growth rate of 1%/year, roughly equivalent to the 2004-2016 average historical pace. The maximum total increase in GW-miles for each model year is allocated as constraints on expansion of inter-regional transmission, interconnection lines for wind and interconnection lines for solar PV respectively in proportion to the total expansion for each category of lines under unconstrained Current Policies cases. 83 Impact of Transmission Expansion Constraints on Electricity Capacity by Resource gigawatts (GW)1 Current Policies (Mid-range) Current Policies (Optimistic) Impact of Transmission Expansion Constraints on Electricity Capacity by Resource gigawatts (GW)1 2 0.5 0.7 7 13 4 33 68 87 122 128 0.5 2 13 -39 -105 -112 -142 -164 -189 -26 -41 -88 -169 -306 -523 -4 -30 -52 -79 -143 -1,000 -800 -600 -400 -200 0 200 2024 2026 2028 2030 2032 2035 84 Transmission Expansion Constraints on Electricity Capacity by Resource )1 1 - Transmission constrained cases limit total transmission capacity expansion to a compound annual growth rate of 1%/year, roughly equivalent to the 2004-2016 average historical pace. The maximum total increase in GW-miles for each model year is allocated as constraints on expansion of inter-regional transmission and interconnection lines for onshore wind, offshore wind, solar PV respectively in proportion to the total expansion for each category of lines under unconstrained Current Policies cases. Current Policies (Mid-range) Current Policies (Optimistic) 3 2 3 9 16 15 44 57 80 90 -29 -58 -114 -166 -193 -33 -66 -114 -182 -329 -18 -30 -42 -83 4 2026 2028 2030 2032 2035 To achieve the full emissions reduction potential of Current Policies, new clean electricity must be rapidly added to simultaneously meet growing demand from electric vehicles, heat pumps, and other electrification and reduce fossil fuel use in the power sector. Constraining transmission growth1 severely limits the expansion of wind and solar power (and indirectly reduces the economic deployment of energy storage). As a result, the lifespan of dozens of coal-fired power plants are extended and hundreds of new natural gas power plants are constructed to meet growing demand in transmission constrained scenarios. Current Policies (Mid-range) 200 To achieve the full emissions reduction potential of Current Policies, new clean electricity must be rapidly added to simultaneously meet growing demand from electric vehicles, heat pumps, and other electrification and reduce fossil fuel use in the power sector. Constraining transmission growth1 severely limits the expansion of wind and solar power (and indirectly reduces the economic deployment of energy storage). As a result, the lifespan of dozens of coal-fired power plants are extended and hundreds of new natural gas power plants are constructed to meet growing demand in transmission constrained scenarios. To achieve the full emissions reduction potential of Current Policies, new clean electricity must be rapidly added to simultaneously meet growing demand from electric vehicles, heat pumps, and other electrification and reduce fossil fuel use in the power sector. Constraining transmission growth1 severely limits the expansion of wind and solar power (and indirectly reduces the economic deployment of energy storage). As a result, the lifespan of dozens of coal-fired power plants are extended and hundreds of new natural gas power plants are constructed to meet growing demand in transmission constrained scenarios. 1 - Transmission constrained cases limit total transmission capacity expansion to a compound annual growth rate of 1%/year, roughly equivalent to the 2004-2016 average Impact of Transmission Expansion Constraints on Electricity Generation by Resource terawatt-hours per year (TWh/year)1 Current Policies (Mid-range) Current Policies (Optimistic) Impact of Transmission Expansion Constraints on Electricity Generation by Resource terawatt-hours per year (TWh/year)1 Current Policies (Mid-range) Current Policies (Optimistic) Current Policies (Mid-range) 1,500 37 114 28 190 278 282 162 337 552 630 800 881 13 108 -147 -380 -400 -525 -584 -618 -56 -90 -206 -384 -685 -1,111 -2,000 -1,500 -1,000 -500 0 500 1,000 1,500 2024 2026 2028 2030 2032 2035 66 4 169 260 281 53 121 357 465 630 711 -46 -114 -216 -428 -621 -638 -22 -74 -154 -256 -396 -674 2024 2026 2028 2030 2032 2035 85 If new transmission capacity cannot be added at a faster pace1, growth of wind and solar power will be substantially constrained. The United States would thus be more reliant on coal and natural gas power plants to meet growing demand from electric vehicles and other electrification. While ~75-77% of generation is supplied by low-carbon sources in 2030 and 89-91% in 2035 under Current Policies scenarios, this share falls to 61-62% in 2030 and 71-72%, in 2035, if transmission expansion is constrained. 1 - Transmission constrained cases limit total transmission capacity expansion to a compound annual growth rate of 1%/year, roughly equivalent to the 2004-2016 average historical pace. The maximum total increase in GW-miles for each model year is allocated as constraints on expansion of inter-regional transmission and interconnection lines for onshore wind, offshore wind, solar PV respectively in proportion to the total expansion for each category of lines under unconstrained Current Policies cases. 37 114 28 190 278 282 162 337 552 630 800 881 13 108 -147 -380 -400 -525 -584 -618 -56 -90 -206 -384 -685 -1,111 -2,000 -1,500 -1,000 -500 0 500 1,000 2024 2026 2028 2030 2032 2035 66 4 169 260 281 121 357 465 630 711 114 -216 -428 -621 -638 -74 -154 -256 -396 -674 026 2028 2030 2032 2035 85 If new transmission capacity cannot be added at a faster pace1, growth of wind and solar power will be substantially constrained. To achieve the full emissions reduction potential of Current Policies, new clean electricity must be rapidly added to simultaneously meet growing demand from electric vehicles, heat pumps, and other electrification and reduce fossil fuel use in the power sector. Constraining transmission growth1 severely limits the expansion of wind and solar power (and indirectly reduces the economic deployment of energy storage). As a result, the lifespan of dozens of coal-fired power plants are extended and hundreds of new natural gas power plants are constructed to meet growing demand in transmission constrained scenarios. The United States would thus be more reliant on coal and natural gas power plants to meet growing demand from electric vehicles and other electrification. While ~75-77% of generation is supplied by low-carbon sources in 2030 and 89-91% in 2035 under Current Policies scenarios, this share falls to 61-62% in 2030 and 71-72%, in 2035, if transmission expansion is constrained. 1 - Transmission constrained cases limit total transmission capacity expansion to a compound annual growth rate of 1%/year, roughly equivalent to the 2004-2016 average historical pace. The maximum total increase in GW-miles for each model year is allocated as constraints on expansion of inter-regional transmission and interconnection lines for onshore wind, offshore wind, solar PV respectively in proportion to the total expansion for each category of lines under unconstrained Current Policies cases. 282 881 108 -618 -1,111 If new transmission capacity cannot be added at a faster pace1, growth of wind and solar power will be substantially constrained. The United States would thus be more reliant on coal and natural gas power plants to meet growing demand from electric vehicles and other electrification. While ~75-77% of generation is supplied by low-carbon sources in 2030 and 89-91% in 2035 under Current Policies scenarios, this share falls to 61-62% in 2030 and 71-72%, in 2035, if transmission expansion is constrained. 37 114 28 190 278 282 162 337 552 630 800 881 13 108 -147 -380 -400 -525 -584 -618 -56 -90 -206 -384 -685 -1,111 -1,000 -500 0 500 1,000 66 4 169 260 281 53 121 357 465 630 711 -46 -114 -216 -428 -621 -638 -22 -74 -154 -256 -396 -674 -1,000 -500 0 500 1,000 If new added at sol constrai be mor power pla electric While ~7 low-carb 2035 un share fal 203 If new transmission capacity cannot be added at a faster pace1, growth of wind and solar power will be substantially constrained. The United States would thus be more reliant on coal and natural gas power plants to meet growing demand from electric vehicles and other electrification. If new transmission capacity cannot be added at a faster pace1, growth of wind and solar power will be substantially constrained. The United States would thus be more reliant on coal and natural gas power plants to meet growing demand from electric vehicles and other electrification. To achieve the full emissions reduction potential of Current Policies, new clean electricity must be rapidly added to simultaneously meet growing demand from electric vehicles, heat pumps, and other electrification and reduce fossil fuel use in the power sector. Constraining transmission growth1 severely limits the expansion of wind and solar power (and indirectly reduces the economic deployment of energy storage). As a result, the lifespan of dozens of coal-fired power plants are extended and hundreds of new natural gas power plants are constructed to meet growing demand in transmission constrained scenarios. 1 - Transmission constrained cases limit total transmission capacity expansion to a compound annual growth rate of 1%/year, roughly equivalent to the 2004-2016 average historical pace. The maximum total increase in GW-miles for each model year is allocated as constraints on expansion of inter-regional transmission and interconnection lines for onshore wind, offshore wind, solar PV respectively in proportion to the total expansion for each category of lines under unconstrained Current Policies cases. 1 - Transmission constrained cases limit total transmission capacity expansion to a compound annual growth rate of 1%/year, roughly equivalent to the 2004-2016 average historical pace. The maximum total increase in GW-miles for each model year is allocated as constraints on expansion of inter-regional transmission and interconnection lines for onshore wind, offshore wind, solar PV respectively in proportion to the total expansion for each category of lines under unconstrained Current Policies cases. 84 84 37 114 28 190 278 282 162 337 552 630 800 881 13 108 -147 -380 -400 -525 -584 -618 -56 -90 -206 -384 -685 -1,111 -2,000 -1,500 -1,000 -500 0 500 1,000 1,500 2024 2026 2028 2030 2032 2035 66 4 169 260 281 53 121 357 465 630 711 -46 -114 -216 -428 -621 -638 -22 -74 -154 -256 -396 -674 2024 2026 2028 2030 2032 2035 ct of Transmission Expansion Constraints on Electricity Generation by Resource t-hours per year (TWh/year)1 Current Policies (Mid-range) Current Policies (Optimistic) If new transmission capacity cannot be added at a faster pace1, growth of wind and solar power will be substantially constrained. The United States would thus be more reliant on coal and natural gas power plants to meet growing demand from electric vehicles and other electrification. While ~75-77% of generation is supplied by low-carbon sources in 2030 and 89-91% in 2035 under Current Policies scenarios, this share falls to 61-62% in 2030 and 71-72%, in 2035, if transmission expansion is constrained. To achieve the full emissions reduction potential of Current Policies, new clean electricity must be rapidly added to simultaneously meet growing demand from electric vehicles, heat pumps, and other electrification and reduce fossil fuel use in the power sector. Constraining transmission growth1 severely limits the expansion of wind and solar power (and indirectly reduces the economic deployment of energy storage). As a result, the lifespan of dozens of coal-fired power plants are extended and hundreds of new natural gas power plants are constructed to meet growing demand in transmission constrained scenarios. 1,000 500 While ~75-77% of generation is supplied by low-carbon sources in 2030 and 89-91% in 2035 under Current Policies scenarios, this share falls to 61-62% in 2030 and 71-72%, in 2035, if transmission expansion is constrained. 1 - Transmission constrained cases limit total transmission capacity expansion to a compound annual growth rate of 1%/year, roughly equivalent to the 2004-2016 average historical pace. The maximum total increase in GW-miles for each model year is allocated as constraints on expansion of inter-regional transmission and interconnection lines for onshore wind, offshore wind, solar PV respectively in proportion to the total expansion for each category of lines under unconstrained Current Policies cases. 85 Hydrogen Production Capacity 8.7 0.8 5.4 6.9 4.4 9.9 9.9 0.4 2.2 6.6 3.9 0 5 10 15 20 25 30 35 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 7.3 0.6 0.7 6.0 13.7 14.2 0.6 1.4 1.4 2.0 11.3 18.4 0 5 10 15 20 25 30 35 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Optimistic 7.4 0.9 1.0 5.9 12.6 13.6 0.6 1.3 1.3 2.0 11.3 18.3 0 5 10 15 20 25 30 35 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 10.8 4.8 5.0 2.3 9.0 9.6 0.6 1.2 1.2 2.0 11.3 18.7 0 5 10 15 20 25 30 35 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 87 ons of hydrogen per year (MMt H2/year) Current Policies Scenarios zen Policies (Jan ’21) Mid-range Net-Zero Pathway Conservative 13.6 13.8 0.3 0.9 0.5 0.5 0.3 0.9 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 gasification w/cc sis mal methane reforming w/cc ethane reforming w/cc ethane reforming 14.7 16.1 15.7 26.4 34.5 16.0 26.1 34.2 16.0 27.0 34.7 14.3 17.8 27.6 1 1 - Note that in Current Policies scenarios in this analysis, we model hydrogen electrolysis facilities to be eligible for the 45V clean hydrogen production tax credit if they use only new, carbon-free electricity from within the same model region on an annual matching basis. This is less stringent than the hourly matching requirements likely required to ensure indirect CO2 emissions from hydrogen production do not exceed statutory requirements under IRA. See Ricks, Xu & Jenkins (2022), “Minimizing emissions from grid-based hydrogen production in the United States,” Environmental Research Letters and “45V Tax Credit: Three-Pillars Impact Analysis,” (Evolved Energy Research, June 23, 2023) for more. 86 Hydrogen Production Capacity Current Policies Scenarios Mid-range Current Policies Scenarios Mid-range 0 5 10 15 20 25 30 35 7.4 0.9 1.0 5.9 12.6 13.6 0.6 1.3 1.3 2.0 11.3 18.3 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Current Policies Scenarios Mid-range 16.0 26.1 34.2 0 5 10 15 20 25 30 35 10.8 4.8 5.0 2.3 9.0 9.6 0.6 1.2 1.2 2.0 11.3 18.7 0 5 10 15 20 25 30 35 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Frozen Policies (Jan ’21) Conservative 13.7 13.7 13.6 13.8 0.3 0.9 0.5 0.5 0.3 0.9 0 5 10 15 20 25 30 35 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 biomass gasification w/cc electrolysis autothermal methane reforming w/cc steam methane reforming w/cc steam methane reforming 13.9 13.7 14.7 16.1 15.7 26.4 34.5 1 1 - Note that in Current new, carbon-free electr indirect CO2 emissions f hydrogen production in Net-Zero Pathway 8.7 0.8 5.4 6.9 4.4 9.9 9.9 0.4 2.2 6.6 3.9 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 14.3 17.8 27.6 25 20 0 5 10 15 7.3 0.6 0.7 6.0 13.7 14.2 0.6 1.4 1.4 2.0 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 16.0 1 - Note that in Current Policies scenarios in this analysis, we model hydrogen electrolysis facilities to be eligible for the 45V clean hydrogen production tax credit if they use only new, carbon-free electricity from within the same model region on an annual matching basis. This is less stringent than the hourly matching requirements likely required to ensure indirect CO2 emissions from hydrogen production do not exceed statutory requirements under IRA. See Ricks, Xu & Jenkins (2022), “Minimizing emissions from grid-based hydrogen production in the United States,” Environmental Research Letters and “45V Tax Credit: Three-Pillars Impact Analysis,” (Evolved Energy Research, June 23, 2023) for more. Hydrogen Production Capacity Tax credits for clean hydrogen spur both carbon capture retrofits at existing methane reforming facilities and robust growth of electrolysis Hydrogen produced via biomass gasification with carbon capture is a key part of the Net-Zero Pathway, but not economic under Current Policies Current Policies Scenarios Mid-range 0 2 4 6 8 10 12 14 16 18 20 4.9 0.6 0.7 4.1 8.7 9.2 0.4 0.9 0.9 0.8 4.7 7.6 0 2 4 6 8 10 12 14 16 18 20 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 7.2 2.0 3.5 1.6 6.1 4.0 0.4 0.8 0.4 0.8 4.7 7.8 0 2 4 6 8 10 12 14 16 18 20 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 9.5 0.6 0.3 2032 2033 2034 2035 cc forming w/cc g w/cc g 10.8 10.0 13.6 15.8 10.2 14.9 18.4 Current Policies Scenarios n ’21) Mid-range Conservative Tax credits for clean hydrogen spur both carbon capture retrofits at existing methane reforming facilities and robust growth of electrolysis 0 2 4 6 8 10 12 14 16 18 20 4.9 0.6 0.7 4.1 8.7 9.2 0.4 0.9 0.9 0.8 4.7 7.6 0 2 4 6 8 0 2 4 6 8 0 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 8 10.2 14.9 18.4 Current Policies Scenarios Mid-range rogen spur both carbon ing methane reforming rowth of electrolysis 5.9 0.5 3.7 4.4 3.0 6.8 6.8 0.7 2.5 2.7 0 2 4 6 8 10 12 14 16 18 20 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 4.8 0.4 0.5 4.1 9.4 9.6 0.4 0.9 1.0 0.8 4.7 7.6 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 10.2 15.5 18.7 9.5 11.4 16.6 Optimistic Net-Zero Pathway Hydrogen produced via biomass gasification with carbon capture is a key part of the Net-Zero Pathway, but not economic under Current Policies Optimistic Optimistic Net-Zero Pathway 16 18 20 20 biomass gasification w/cc electrolysis autothermal methane reforming w/cc steam methane reforming w/cc steam methane reforming 1 18 20 Tax credits for clean hydrogen spur both carbon capture retrofits at existing methane reforming facilities and robust growth of electrolysis 15 13.6 10 5 0 1 - Note that in Current Policies scenarios in this analysis, we model hydrogen electrolysis facilities to be eligible for the 45V clean hydrogen production tax credit if they use only new, carbon-free electricity from within the same model region on an annual matching basis. Hydrogen Production Capacity 5.9 0.5 3.7 4.4 3.0 6.8 6.8 0.7 2.5 2.7 0 2 4 6 8 10 12 14 16 18 20 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 4.8 0.4 0.5 4.1 9.4 9.6 0.4 0.9 1.0 0.8 4.7 7.6 0 2 4 6 8 10 12 14 16 18 20 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 4.9 0.6 0.7 4.1 8.7 9.2 0.4 0.9 0.9 0.8 4.7 7.6 0 2 4 6 8 10 12 14 16 18 20 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 7.2 2.0 3.5 1.6 6.1 4.0 0.4 0.8 0.4 0.8 4.7 7.8 0 2 4 6 8 10 12 14 16 18 20 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 9.5 0.6 0.3 2032 2033 2034 2035 cc forming w/cc ng w/cc ng 88 n per year (MMt H2/year) 10.8 10.0 13.6 15.8 10.2 14.9 18.4 10.2 15.5 18.7 9.5 11.4 16.6 Optimistic Current Policies Scenarios n ’21) Mid-range Net-Zero Pathway Conservative 1 - Note that in Current Policies scenarios in this analysis, we model hydrogen electrolysis facilities to be eligible for the 45V clean hydrogen production tax credit if they use only new, carbon-free electricity from within the same model region on an annual matching basis. This is less stringent than the hourly matching requirements likely required to ensure indirect CO2 emissions from hydrogen production do not exceed statutory requirements under IRA. See Ricks, Xu & Jenkins (2022), “Minimizing emissions from grid-based hydrogen production in the United States,” Environmental Research Letters and “45V Tax Credit: Three-Pillars Impact Analysis,” (Evolved Energy Research, June 23, 2023) for more. Hydrogen Production Capacity 0 5 10 15 20 2022 2.5 2.3 2.0 7.2 8.6 9.8 0.3 1.1 2.0 0.2 0.9 2.6 3.6 0 5 10 15 20 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2.5 2.3 1.9 7.2 8.6 10.0 0.3 1.1 2.0 0.7 0.7 2.8 4.1 0 5 10 15 20 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2.5 2.4 2.1 7.0 7.7 8.6 0.3 1.1 2.0 0.3 2.2 2.9 0 5 10 15 20 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2.4 2.5 2.5 2.3 6.8 7.0 7.5 8.3 0 5 10 15 20 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 e-fuels (fischer-tropsch liquids) hydrogen boiler electricity generation ammonia fuel (haber-bosch) bulk chemicals (incl. fertilizer) refining ( ) Optimistic Current Policies Scenarios Frozen Policies (Jan ’21) Mid-range Conservative 9.5 9.2 10.0 10.8 10.0 13.6 15.8 10.2 14.9 18.4 10.2 15.5 18.7 Low-carbon hydrogen produced under Current Policies meets the bulk of traditional hydrogen demands in refining & bulk chemicals (e.g. fertilizer) and new uses in industrial heat and low-carbon fuels. Current Policies Scenarios Mid-range Current Policies Scenarios Mid-range 0 5 10 15 20 2.5 2.3 1.9 7.2 8.6 10.0 0.3 1.1 2.0 0.7 0.7 2.8 4.1 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Current Policies Scenarios Mid-range 10.2 14.9 18.4 d under Current tional hydrogen cals (e.g. fertilizer) d low-carbon fuels. Net-Zero Pathway Optimistic 2.5 2.3 1.9 7.0 7.8 12.3 0.6 2.0 0.7 0.3 0 5 10 15 20 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2.5 2.3 2.0 7.2 8.6 9.8 0.3 1.1 2.0 0.2 0.9 2.6 3.6 0 5 10 15 20 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2.5 2.3 1.9 7.2 8.6 10.0 0.3 1.1 2.0 0.7 0.7 2.8 4.1 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Optimistic Mid range Net Zero Pathway 10.2 14.9 18.4 10.2 15.5 18.7 9.5 11.4 16.6 Current drogen fertilizer) rbon fuels. Hydrogen Production Capacity This is less stringent than the hourly matching requirements likely required to ensure indirect CO2 emissions from hydrogen production do not exceed statutory requirements under IRA. See Ricks, Xu & Jenkins (2022), “Minimizing emissions from grid-based hydrogen production in the United States,” Environmental Research Letters and “45V Tax Credit: Three-Pillars Impact Analysis,” (Evolved Energy Research, June 23, 2023) for more. 88 2.5 2.3 1.9 7.0 7.8 12.3 0.6 2.0 0.7 0.3 0 5 10 15 20 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2.5 2.3 2.0 7.2 8.6 9.8 0.3 1.1 2.0 0.2 0.9 2.6 3.6 0 5 10 15 20 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2.5 2.3 1.9 7.2 8.6 10.0 0.3 1.1 2.0 0.7 0.7 2.8 4.1 0 5 10 15 20 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2.5 2.4 2.1 7.0 7.7 8.6 0.3 1.1 2.0 0.3 2.2 2.9 0 5 10 15 20 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2.5 2.3 7.5 8.3 2029 2030 2031 2032 2033 2034 2035 tropsch liquids) ration haber-bosch) (incl. fertilizer) ydrogen per year (MMt H2/year) Optimistic Current Policies Scenarios cies (Jan ’21) Mid-range Net-Zero Pathway Conservative 10.0 10.8 10.0 13.6 15.8 10.2 14.9 18.4 10.2 15.5 18.7 9.5 11.4 16.6 Low-carbon hydrogen produced under Current Policies meets the bulk of traditional hydrogen demands in refining & bulk chemicals (e.g. fertilizer) and new uses in industrial heat and low-carbon fuels. Hydrogen Production Capacity 0 5 10 15 20 2.5 2.3 2.0 7.2 8.6 9.8 0.3 1.1 2.0 0.2 0.9 2.6 3.6 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 10.2 15.5 18.7 2.5 2.3 7.0 7.8 0.6 0.7 0 5 10 15 20 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2.5 2.3 2.0 7.2 8.6 9.8 0.3 1.1 2.0 0.2 0.9 2.6 3.6 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 10.2 15.5 18.7 9.5 11.4 20 Low-carbon hydrogen produced under Current Policies meets the bulk of traditional hydrogen demands in refining & bulk chemicals (e.g. fertilizer) and new uses in industrial heat and low-carbon fuels. 2.5 2.3 1.9 7.0 7.8 12.3 0.6 2.0 0.7 0.3 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 9.5 11.4 16.6 0 5 10 15 2.5 2.4 2.1 7.0 7.7 8.6 0.3 1.1 2.0 0.3 2.2 2.9 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 10.0 13.6 15.8 0 5 10 2.4 2.5 2.5 2.3 6.8 7.0 7.5 8.3 0 5 10 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 9.5 9.2 10.0 10.8 0 0 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 0 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 0 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 89 90 0 50 100 150 200 250 300 350 400 2024 2025 2026 14.4 14.6 54.6 11.5 52.5 92.7 94.4 11.5 14.1 15.3 26.7 0 50 100 150 200 250 300 350 400 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 14.4 14.6 54.6 11.5 52.5 87.0 90.4 11.0 15.2 14.9 21.3 0 50 100 150 200 250 300 350 400 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 14.4 14.6 54.6 11.5 52.5 68.5 49.8 10.3 12.6 12.5 7.4 0 50 100 150 200 250 300 350 400 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 7.4 8.8 11.0 0 50 100 150 200 250 300 350 400 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 direct air capture natural gas power plants ethanol biomass pyrolysis biomass gasification methane reforming cement iron/steel refineries natural gas processing Carbon Dioxide Capture by Source million metric tons of CO2 per year (MMt CO2/year)1 Optimistic Current Policies Scenarios Frozen Policies (Jan ’21) Mid-range Conservative 1 1 1 1 1 - Excludes currently operating CCS projects with up to 18.6 MMT/y of capacity that exclusively supply CO2 for enhanced oil recovery (see https Capture volumes in the cement, iron/steel, refineries and natural gas processing sectors are exogenous inputs based on Larson et al. “A Turning Assessing the Climate and Clean Energy Provisions in the Inflation Reduction Act ,” (Rhodium Group: August 12, 2022) in Current Policies scenari Perspective“ (Evolved Energy Research, August 19, 2022) for Net-Zero Pathway. All other capture volumes are endogenously optimized in RIO m 1.5 13.0 18.5 108 203 129 261 136 269 The enhanced value of the 45Q tax credit for carbon dioxide storage and use is likely to make carbon capture economic across a range of sectors and industries. 24.3 78.1 83.1 51.8 65.9 65.9 9.1 14.6 86.5 49.2 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 241 289 24.3 78.1 83.1 51.8 65.9 65.9 9.1 14.6 86.5 49.2 0 50 100 150 200 250 300 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 14.4 14.6 54.6 11.5 52.5 92.7 94.4 11.5 14.1 15.3 26.7 0 50 100 150 200 250 300 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 14.4 14.6 54.6 11.5 52.5 87.0 90.4 11.0 15.2 14.9 21.3 0 50 100 150 200 250 300 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 14.4 14.6 54.6 11.5 52.5 68.5 49.8 10.3 12.6 12.5 7.4 0 50 100 150 200 250 300 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 7.4 8.8 11.0 2027 2028 2029 2030 2031 2032 2033 2034 2035 ne reforming t eel ies l gas processing 1 1 1 1 1 - Excludes currently operating CCS projects with up to 18.6 MMT/y of capacity that exclusively supply CO2 for enhanced oil recovery (see https://www.catf.us/ccsmapus/). Capture volumes in the cement, iron/steel, refineries and natural gas processing sectors are exogenous inputs based on Larson et al. “A Turning Point for US Climate Progress: Assessing the Climate and Clean Energy Provisions in the Inflation Reduction Act ,” (Rhodium Group: August 12, 2022) in Current Policies scenarios and on “Annual Decarbonizaton Perspective“ (Evolved Energy Research, August 19, 2022) for Net-Zero Pathway. All other capture volumes are endogenously optimized in RIO model. 24.3 78.1 83.1 51.8 65.9 65.9 9.1 14.6 86.5 49.2 0 50 100 150 200 250 300 350 400 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 14.4 14.6 54.6 11.5 52.5 92.7 94.4 11.5 14.1 15.3 26.7 0 50 100 150 200 250 300 350 400 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 14.4 14.6 54.6 11.5 52.5 87.0 90.4 11.0 15.2 14.9 21.3 0 50 100 150 200 250 300 350 400 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 91 14.4 14.6 54.6 11.5 52.5 68.5 49.8 10.3 12.6 12.5 7.4 0 50 100 150 200 250 300 350 400 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 7.4 11.0 2033 2034 2035 s ear (MMt CO2/year)1 Optimistic Current Policies Scenarios ’21) Mid-range Net-Zero Pathway Conservative 1 - Excludes currently operating CCS projects with up to 18.6 MMT/y of capacity that exclusively supply CO2 for enhanced oil recovery (see https://www.catf.us/ccsmapus/). Capture volumes in the cement, iron/steel, refineries and natural gas processing sectors are exogenous inputs based on Larson et al. “A Turning Point for US Climate Progress: Assessing the Climate and Clean Energy Provisions in the Inflation Reduction Act ,” (Rhodium Group: August 12, 2022) in Current Policies scenarios and on “Annual Decarbonizaton Perspective“ (Evolved Energy Research, August 19, 2022) for Net-Zero Pathway. All other capture volumes are endogenously optimized in RIO model. 18.5 108 203 129 261 136 269 241 289 The enhanced value of the 45Q tax credit for carbon dioxide storage and use is likely to make carbon capture economic across a range of sectors and industries. 350 400 350 400 350 400 350 400 ants n r year (MMt CO2/year)1 Optimistic Current Policies Scenarios an ’21) Mid-range Net-Zero Pathway Conservative The enhanced value of the 45Q tax credit for carbon dioxide storage and use is likely to make carbon capture economic across a range of sectors and industries. Current Policies Scenarios Mid-range Net-Zero Pathway Conservative Conservative Optimistic Optimistic 350 400 350 400 The enhanced value of the 45Q tax credit for carbon dioxide storage and use is likely to make carbon capture economic across a range of sectors and industries. 13.0 18.5 108 203 129 261 136 269 241 289 24.3 78.1 83.1 51.8 65.9 65.9 9.1 14.6 86.5 49.2 0 50 100 150 200 250 300 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 14.4 14.6 54.6 11.5 52.5 92.7 94.4 11.5 14.1 15.3 26.7 0 50 100 150 200 250 300 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 14.4 14.6 54.6 11.5 52.5 87.0 90.4 11.0 15.2 14.9 21.3 0 50 100 150 200 250 300 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 9 14.4 14.6 54.6 11.5 52.5 68.5 49.8 10.3 12.6 12.5 7.4 0 50 100 150 200 250 300 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 7.4 11.0 2031 2032 2033 2034 2035 ng essing1 1 - Excludes currently operating CCS projects with up to 18.6 MMT/y of capacity that exclusively supply CO2 for enhanced oil recovery (see https://www.catf.us/ccsmapus/). Capture volumes in the cement, iron/steel, refineries and natural gas processing sectors are exogenous inputs based on Larson et al. “A Turning Point for US Climate Progress: Assessing the Climate and Clean Energy Provisions in the Inflation Reduction Act ,” (Rhodium Group: August 12, 2022) in Current Policies scenarios and on “Annual Decarbonizaton 0 18.5 108 203 129 261 136 269 241 289 0 50 100 150 200 14.4 14.6 54.6 11.5 52.5 68.5 49.8 10.3 12.6 12.5 7.4 0 50 100 150 200 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 7.4 8.8 11.0 0 50 100 150 200 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 natural gas processing1 1.5 13.0 18.5 108 203 50 1 - Excludes currently operating CCS projects with up to 18.6 MMT/y of capacity that exclusively supply CO2 for enhanced oil recovery (see https://www.catf.us/ccsmapus/). Capture volumes in the cement, iron/steel, refineries and natural gas processing sectors are exogenous inputs based on Larson et al. “A Turning Point for US Climate Progress: Assessing the Climate and Clean Energy Provisions in the Inflation Reduction Act ,” (Rhodium Group: August 12, 2022) in Current Policies scenarios and on “Annual Decarbonizaton Perspective“ (Evolved Energy Research, August 19, 2022) for Net-Zero Pathway. All other capture volumes are endogenously optimized in RIO model. 91 Carbon Dioxide Use and Sequestration million metric tons of CO2 per year (MMt CO2/year) 9 se and Sequestration O2 per year (MMt CO2/year) Optimistic Current Policies Scenarios es (Jan ’21) Mid-range Net-Zero Pathway Conservative1 240.6 288.5 0 50 100 150 200 250 300 350 400 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 45.2 117.8 242.3 17.9 26.5 0 50 100 150 200 250 300 350 400 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 112.2 237.1 17.1 23.4 0 50 100 150 200 250 300 350 400 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 21.4 93.5 184.2 13.9 18.8 0 50 100 150 200 250 300 350 400 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 3.0 18.5 2030 2031 2032 2033 2034 2035 sch liquids) 3.0 18.5 108 203 129 261 136 269 241 289 The bulk of captured CO2 is injected for geologic storage2 although a small but growing share of CO2 from biogenic sources (e.g. ethanol) or direct air capture is used to produce synthetic liquid hydrocarbons (aka ‘e-fuels’) via combination with hydrogen in the Fischer-Tropsch process. 2 1 - Note that geologic storage in the Conservative scenarios is subject to a binding exogenous constraint on the increase in annual CO2 injection through 2035. See p. 33. 2 - We model geologic storage as dedicated CO2 sequestration (earning $85/t CO2 45Q credit value). In practice, some projects may make use of CO2 in oil fields for enhanced oil recovery (EOR), claiming the lower $60/t CO2 credit value for use of captured CO2. However, as amended by IRA, 45Q now provides greater economic incentive for dedicated storage relative to EOR and only 1 out of 171 projects announced since January 2021 has stated it plans to inject CO2 for EOR (see https://www.catf.us/ccsmapus/). Current Policies Scenarios Mid-range Net-Zero Pathway 350 400 350 400 The bulk of captured CO2 is injected for geologic storage2 although a small but growing share of CO2 from biogenic sources (e.g. ethanol) or direct air capture is used to produce synthetic liquid hydrocarbons (aka ‘e-fuels’) via combination with hydrogen in the Fischer-Tropsch process. 350 240.6 288.5 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 241 289 240.6 288.5 0 50 100 150 200 250 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 45.2 117.8 242.3 17.9 26.5 0 50 100 150 200 250 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 112.2 237.1 17.1 23.4 0 50 100 150 200 250 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 21.4 93.5 184.2 13.9 18.8 0 50 100 150 200 250 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 18.5 2034 2035 18.5 108 203 129 261 136 269 241 289 1 N t th t l i t i th C ti i i bj t t bi di t i t th i i l CO i j ti th h 2035 S 33 50 1 - Note that geologic storage in the Conservative scenarios is subject to a binding exogenous constraint on the increase in annual CO2 injection through 2035. See p. 33. 2 - We model geologic storage as dedicated CO2 sequestration (earning $85/t CO2 45Q credit value). In practice, some projects may make use of CO2 in oil fields for enhanced oil recovery (EOR), claiming the lower $60/t CO2 credit value for use of captured CO2. However, as amended by IRA, 45Q now provides greater economic incentive for dedicated storage relative to EOR and only 1 out of 171 projects announced since January 2021 has stated it plans to inject CO2 for EOR (see https://www.catf.us/ccsmapus/). 1 - Note that geologic storage in the Conservative scenarios is subject to a binding exogenous constraint on the increase in annual CO2 injection through 2035. See p. 33. 2 - We model geologic storage as dedicated CO2 sequestration (earning $85/t CO2 45Q credit value). In practice, some projects may make use of CO2 in oil fields for enhanced oil recovery (EOR), claiming the lower $60/t CO2 credit value for use of captured CO2. However, as amended by IRA, 45Q now provides greater economic incentive for dedicated storage relative to EOR and only 1 out of 171 projects announced since January 2021 has stated it plans to inject CO2 for EOR (see https://www.catf.us/ccsmapus/). 93 -13 -30 -46 -67 -88 -100 -113 -127 -140 -60 -65 -71 -325 -275 -225 -175 -125 -75 -25 25 75 125 175 225 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 -6 -17 -28 -45 -62 -73 -85 -96 -107 -45 -49 -54 -325 -275 -225 -175 -125 -75 -25 25 75 125 175 225 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 -8-16-25-35-45-52-59-58-56-24-25-26 fixed o&m capital - supply side electricity transmission electricity distribution capital - demand side variable o&m fuels delivery fuel and commodities supply-side incentives demand-side incentives net difference Mid-range Change in Annual Energy Expenditures vs Frozen Policies as of January 2021 billions of 2023 US dollars1 Optimistic Current Policies Scenarios Conservative T I U A e 1 1 - Dollar values modeled internally as 2018 USD, and converted for display to 2023 USD using Bureau of Labor Statistics CPI Inflation Calc -8 -16 -25 -35 -45 -52 -59 -58 -56 -24 -25 -26 -325 -275 -225 -175 -125 -75 -25 25 75 125 175 225 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 0 -13 -30 -46 -67 -88 -100 -113 -127 -140 -60 -65 -71 -325 -275 -225 -175 -125 -75 -25 25 75 125 175 225 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 -6 -17 -28 -45 -62 -73 -85 -96 -107 -45 -49 -54 -325 -275 -225 -175 -125 -75 -25 25 75 125 175 225 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 25-35-45-52-59-58-56-24-25-26 capital - supply side electricity transmission electricity distribution mand side variable o&m fuels delivery fuel and commodities incentives demand-side incentives net difference Mid-range 94 ergy Expenditures vs Frozen Policies as of January 2021 Optimistic Current Policies Scenarios onservative The Inflation Reduction Act and Infrastructure Law lower annual U.S. energy expenditures ~3-7% in 2030, a savings of $59-$113 billion for U.S. households, businesses, and industry. Annual energy cost savings peak in 2032 prior to the scheduled expiration of several IRA incentives (i.e., tax credits for EV adoption, efficiency etc.) but remain about 1-4% lower than the Frozen Policies scenario through 2035. 1 - Dollar values modeled internally as 2018 USD, and converted for display to 2023 USD using Bureau of Labor Statistics CPI Inflation Calculator: 1 2018 UDS = 1.207 2023 USD. nvestment in Energy Supply Infrastructure Capital Investment in Energy Supply Infrastructure -59 -58 -56 -24 -25 -26 2030 2031 2032 2033 2034 2035 Mid-range Current Policies Scenarios Mid-range Current Policies Scenarios 175 225 electricity transmission fuels delivery net difference 5 5 capital - supply side variable o&m demand-side incentives electricity distribution fuel and commodities The Inflation Reduction Act and Infrastructure Law lower annual U.S. energy expenditures ~3-7% in 2030, a savings of $59-$113 billion for U.S. households, businesses, and industry. Annual energy cost savings peak in 2032 prior to the scheduled expiration of several IRA incentives (i.e., tax credits for EV adoption, efficiency etc.) but remain about 1-4% lower than the Frozen Policies scenario through 2035. The Inflation Reduction Act and Infrastructure Law lower annual U.S. energy expenditures ~3-7% in 2030, a savings of $59-$113 billion for U.S. households, businesses, and industry. Annual energy cost savings peak in 2032 prior to the scheduled expiration of several IRA incentives (i.e., tax credits for EV adoption, efficiency etc.) but remain about 1-4% lower than the Frozen Policies scenario through 2035. Annual energy cost savings peak in 2032 prior to the scheduled expiration of several IRA incentives (i.e., tax credits for EV adoption, efficiency etc.) but remain about 1-4% lower than the Frozen Policies scenario through 2035. Annual energy cost savings peak in 2032 prior to the scheduled expiration of several IRA incentives (i.e., tax credits for EV adoption, efficiency etc.) but remain about 1-4% lower than the Frozen Policies scenario through 2035. -30 -46 -67 -88 1 - Dollar values modeled internally as 2018 USD, and converted for display to 2023 USD using Bureau of Labor Statistics CPI Inflation Calculator: 1 2018 UDS = 1.207 2023 USD. 1 - Dollar values modeled internally as 2018 USD, and converted for display to 2023 USD using Bureau of Labor Statistics CPI Inflation Calculator: 1 2018 UDS = 1.207 2023 USD. 94 Annual Capital Investment in Energy Supply Related Infrastructure billions of 2023 US dollars1 35 31 26 37 76 104 114 75 25 14 24 26 16 27 40 16 13 8 11 9 0 50 100 150 200 250 300 350 022 023 024 025 026 027 028 029 030 031 032 033 034 035 21 35 43 56 93 137 69 78 55 8 42 23 5 8 7 10 9 022 023 024 025 026 027 028 029 030 031 032 033 034 035 eration - wind electricity generation - storage eration - nuclear CO2 transport & storage ydrogen) methane reforming (hydrogen) Optimistic Net-Zero Pathway 181 243 304 243 294 218 35 31 26 37 76 104 114 75 25 14 24 26 16 27 40 16 13 8 11 9 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 ricity generation - storage transport & storage ane reforming (hydrogen) Net-Zero Pathway 243 294 218 21 56 69 23 0 50 100 150 200 250 13 37 36 51 78 111 37 97 56 5 32 27 6 6 7 10 9 0 50 100 150 200 250 18 30 31 56 79 106 57 75 44 6 30 18 55 7 10 9 0 50 100 150 200 250 13 15 14 26 51 29 36 26 34 8 27 10 9 13 15 14 26 51 29 36 26 34 8 27 10 9 104 110 101 160 215 235 139 244 261 181 0 50 100 150 200 250 18 30 31 56 79 106 57 75 44 6 30 18 55 7 10 9 0 50 100 150 200 250 13 15 14 26 51 29 36 26 34 8 27 10 9 13 15 14 26 51 29 36 26 34 8 27 10 9 0 50 100 150 200 250 73 104 110 101 160 215 235 0 50 100 150 13 15 14 26 51 29 36 26 34 8 27 10 9 13 15 14 26 51 29 36 26 34 8 27 10 9 0 50 100 150 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 73 104 110 101 0 0 2022 2023 2024 2025 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2027 2028 2029 2030 2031 2032 2033 2034 2035 1 - Dollar values modeled internally as 2018 USD, and converted for display to 2023 USD using Bureau of Labor Statistics CPI Inflation Calculator: 1 2018 UDS = 1.207 2023 USD. 95 96 332 357 415 476 950 938 1,131 842 358 640 707 712 1,011 36 120 123 176 116 137 90 126 113 130 86 91 25 33 38 24 200 36 39 45 62 63 62 61 47 113 116 117 117 117 500 1,000 1,500 2,000 2,500 3,000 3,500 electricity transmission electricity generation - solar electricity generation - wind electricity generation - storage electricity generation - natural gas electricity generation - natural gas w/ cc electricity generation - nuclear CO2 transport & storage biofuels industrial heat electrolysis (hydrogen) methane reforming (hydrogen) 'e-fuels' (fischer-tropsch liquids) liquefied natural gas export 96 e Capital Investment in Energy Supply Related Infrastructure, 2023-2035 US dollars1 1 - Dollar values modeled internally as 2018 USD, and converted for display to 2023 USD using Bureau of Labor Statistics CPI Inflation Calculator: 1 2018 UDS = 1.207 2023 USD. Values less than $20 billion not displayed in data labels. Capital investment in fuel and electricity distribution are not included. 2 - Includes associated transmission investments. 3,157 2,798 2,521 2,373 1,327 The Inflation Reduction Act and Infrastructure Law spur over a trillion dollars of additional capital investment in energy supply related infrastructure through 2035. Over four-fifths of this ~$2.4-2.8 trillion in cumulative investment is directed to wind and solar power and battery energy storage and related transmission grid investments. The laws drive additional investment in clean energy manufacturing and supply chains, distribution networks and demand side equipment that are not depicted here. Cumulative Capital Investment in Energy Supply Related Infrastructure, 2023-2035 billions of 2023 US dollars1 Capital Investment in Energy Supply Related Infrastructure, 2023-2035 US dollars1 Cumulative Capital Investment in Energy Supply Related Infrastructure, 2023-2035 billions of 2023 US dollars1 The Inflation Reduction Act and Infrastructure Law spur over a trillion dollars of additional capital investment in energy supply related infrastructure through 2035. 0 3,500 1 - Dollar values modeled internally as 2018 USD, and converted for display to 2023 USD using Bureau of Labor Statistics CPI Inflation Calculator: 1 2018 UDS = 1.207 2023 USD. Values less than $20 billion not displayed in data labels. Capital investment in fuel and electricity distribution are not included. 2 - Includes associated transmission investments. 96 Annual Capital Investment in Energy Supply Related Infrastructure billions of 2023 US dollars1 Annual Capital Investment in Energy Supply Related Infrastructure billions of 2023 US dollars1 35 31 26 37 76 104 114 75 25 14 24 26 16 27 40 16 13 8 11 9 0 50 100 150 200 250 300 350 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 21 35 43 56 93 137 69 78 55 8 42 23 5 8 7 10 9 0 50 100 150 200 250 300 350 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 13 37 36 51 78 111 37 97 56 5 32 27 6 6 7 10 9 0 50 100 150 200 250 300 350 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 18 30 31 56 79 106 57 75 44 6 30 18 55 7 10 9 0 50 100 150 200 250 300 350 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 13 15 14 26 51 29 36 26 34 8 27 10 9 electricity transmission electricity generation - solar electricity generation - wind electricity generation - storage electricity generation - natural gas electricity generation - natural gas w/ cc electricity generation - nuclear CO2 transport & storage biofuels industrial heat electrolysis (hydrogen) methane reforming (hydrogen) e-fuels' (fischer-tropsch liquids) liquefied natural gas export 15 14 6 51 29 6 26 34 8 7 10 9 0 5 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 97 US dollars1 Optimistic Current Policies Scenarios en Policies (Jan ’21) Mid-range Net-Zero Pathway Conservative 4 110 101 160 215 235 139 244 261 181 243 304 243 294 218 1 - Dollar values modeled internally as 2018 USD, and converted for display to 2023 USD using Bureau of Labor Statistics CPI Inflation Calculator: 1 2018 UDS = 1.207 2023 USD. Values less than $5 billion not displayed in data labels. Capital investment in fuel and electricity distribution are not included. Annual Capital Investment in Energy Supply Related Infrastructure billions of 2023 US dollars1 Values less than $5 billion not displayed in data labels. Capital investment in fuel and electricity distribution are not included. 97 98 Mid-range Change in Energy Supply Related Employment by Resource vs Frozen Policies as of January 2021 thousands of jobs1 Optimistic Current Policies Scenarios Conservative Optimistic The Inflation Reduction Act and Infrastructure Law could increase energy supply related employment1 by about 1.4-1.7 million additional jobs in 2030 and 2.2-2.9 million by 2035. Solar, wind, and grid related jobs expand rapidly under Current Policies scenarios accounting for the vast majority of additional employment. Oil, natural gas, and coal related employment declines by ~50,000-70,000 jobs in 2030, roughly equal to the additional jobs created in CO2 transport & storage. The Inflation Reduction Act and Infrastructure Law could increase energy supply related employment1 by about 1.4-1.7 million additional jobs in 2030 and 2.2-2.9 million by 2035. Solar, wind, and grid related jobs expand rapidly under Current Policies scenarios accounting for the vast majority of additional employment. Oil, natural gas, and coal related employment declines by ~50,000-70,000 jobs in 2030, roughly equal to the additional jobs created in CO2 transport & storage. The Inflation Reduction Act and Infrastructure Law could increase energy supply related employment1 by about 1.4-1.7 million additional jobs in 2030 and 2.2-2.9 million by 2035. Solar, wind, and grid related jobs expand rapidly under Current Policies scenarios accounting for the vast majority of additional employment. Oil, natural gas, and coal related employment declines by ~50,000-70,000 jobs in 2030, roughly equal to the additional jobs created in CO2 transport & storage. The Inflation Reduction Act and Infrastructure Law could increase energy supply related employment1 by about 1.4-1.7 million additional jobs in 2030 and 2.2-2.9 million by 2035. Mid-range Current Policies Scenarios Optimistic 6,000 6,000 natural gas biomass & biofuels electricity transmission & distribution nuclear power s 6,000 000 electricity transmission & distribution nuclear power 63 68 69 69 8 102 810 789 745 693 644 588 511 134 132 110 87 67 56 45 398 387 440 372 360 364 319 97 93 102 100 98 96 85 733 941 1,137 1,354 1,578 1,762 1,923 49 49 49 50 47 47 58 68 562 692 902 1,480 2,183 2,376 138 263 414 369 493 478 518 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 ower solar PV wind power 63 68 69 69 8 102 810 789 745 693 644 588 511 134 132 110 87 67 56 45 398 387 440 372 360 364 319 97 93 102 100 98 96 85 733 941 1,137 1,354 1,578 1,762 1,923 49 49 49 50 47 47 58 68 562 692 902 1,480 2,183 2,376 138 263 414 369 493 478 518 0 1,000 2,000 3,000 4,000 5,000 6,000 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 62 67 67 67 8 100 810 789 748 699 653 601 530 134 134 115 93 70 60 47 398 396 459 382 365 365 335 97 93 102 100 98 96 91 733 898 1,043 1,275 1,511 1,687 1,798 49 49 49 50 47 47 58 68 500 669 876 1,286 1,655 1,934 138 203 270 359 512 524 522 0 1,000 2,000 3,000 4,000 5,000 6,000 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 57 61 62 63 7 92 810 790 752 708 667 624 565 134 133 114 94 72 62 49 398 387 432 367 351 357 327 97 93 102 100 98 96 91 733 935 1,107 1,314 1,514 1,638 1,719 49 49 49 50 47 47 57 68 542 699 920 1,311 1,633 1,880 138 255 353 356 464 452 453 0 1,000 2,000 3,000 4,000 5,000 6,000 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 6 8 8 8 9 10 810 791 757 719 682 643 586 134 133 116 96 77 68 59 398 400 463 385 381 379 377 97 92 101 99 97 96 90 437 583 621 601 681 739 789 49 49 49 47 49 51 54 68 188 239 325 467 523 390 138 200 259 235 263 257 272 CO2 transport & storage oil coal natural gas biomass & biofuels electricity transmission & distribution nuclear power solar PV wind power 100 Under Current Policies, over 4.2 million Americans will be employed in energy supply related sectors1 by 2030 (up from ~2.4 million today), rising to 5.2-5.9 million by 2035. Mid-range Current Policies Scenarios Employment grows rapidly in solar PV, wind power, and electricity grid related jobs. 1 - Employment in petroleum fuel refining, distribution, and retailing; hydrogen production, distribution and retailing; energy storage manufacturing, installation and operations; automotive supply chains and assembly; and energy efficiency are excluded from this analysis. Values less than 5 thousand jobs not displayed in labels 6 8 8 8 3 10 810 791 757 719 682 643 586 134 133 116 96 77 68 59 398 400 463 385 381 379 377 97 92 101 99 97 96 90 733 877 945 1,025 1,134 1,223 1,187 49 49 49 47 49 51 54 68 188 239 325 467 523 390 138 200 259 235 263 257 272 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 63 68 69 810 789 745 693 134 132 110 87 398 387 440 372 97 93 102 100 733 941 1,137 1,354 49 49 49 50 68 562 692 902 138 263 414 369 0 1,000 2,000 3,000 4,000 5,000 2022 2023 2024 2025 2026 2027 2028 62 67 67 67 8 100 810 789 748 699 653 601 530 134 134 115 93 70 60 47 398 396 459 382 365 365 335 97 93 102 100 98 96 91 733 898 1,043 1,275 1,511 1,687 1,798 49 49 49 50 47 47 58 68 500 669 876 1,286 1,655 1,934 138 203 270 359 512 524 522 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 0 1,000 2,000 3,000 4,000 5,000 62 67 67 67 8 100 810 789 748 699 653 601 530 134 134 115 93 70 60 47 398 396 459 382 365 365 335 97 93 102 100 98 96 91 733 898 1,043 1,275 1,511 1,687 1,798 49 49 49 50 47 47 58 68 500 669 876 1,286 1,655 1,934 138 203 270 359 512 524 522 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 1 - Employment in petroleum fuel refining, distribution, and retailing; hydrogen production, distribution and retailing; energy storage manufacturing, installation and operations; automotive supply chains and assembly; and energy efficiency are excluded from this analysis. 57 60 60 61 92 -26 -39 -55 -75 -23 -21 -58 64 192 329 444 539 736 374 453 577 1,013 1,660 1,986 63 154 134 230 221 246 -500 0 500 1,000 1,500 2,000 2,500 3,000 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 56 59 59 59 90 -30 -41 -57 -42 21 99 250 376 464 611 312 430 551 819 1,132 1,544 124 250 267 250 -500 0 500 1,000 1,500 2,000 2,500 3,000 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 6 8 8 8 9 10 810 791 757 719 682 643 586 134 133 116 96 77 68 59 398 400 463 385 381 379 377 97 92 101 99 97 96 90 437 583 621 601 681 739 789 49 49 49 47 49 51 54 68 188 239 325 467 523 390 138 200 259 235 263 257 272 52 53 54 54 82 -22 -30 -22 -50 58 163 289 380 415 532 355 460 595 843 1,110 1,490 55 94 121 201 195 181 -500 0 500 1,000 1,500 2,000 2,500 3,000 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Net: 505 733 1,033 1,428 1,674 2,207 387 585 963 1,451 1,800 2,388 543 820 1,056 1,676 2,007 1 - Employment in petroleum fuel refining, distribution, and retailing; hydrogen production, distribution and retailing; energy storage manufacturin operations; automotive supply chains and assembly; and energy efficiency are excluded from this analysis. Values less than 20 thousand jobs not di Solar, wind, and grid related jobs expand rapidly under Current Policies scenarios accounting for the vast majority of additional employment. 1 - Employment in petroleum fuel refining, distribution, and retailing; hydrogen production, distribution and retailing; energy storage manufacturing, installation and operations; automotive supply chains and assembly; and energy efficiency are excluded from this analysis. Values less than 20 thousand jobs not displayed in labels 1 - Employment in petroleum fuel refining, distribution, and retailing; hydrogen production, distribution and retailing; energy storage manufacturing, installation and operations; automotive supply chains and assembly; and energy efficiency are excluded from this analysis. Values less than 20 thousand jobs not displayed in labels 99 Total Energy Supply Related Employment by Resource thousands of jobs1 Mid-range Current Policies Scenarios Values less than 5 thousand jobs not displayed in labels 1 - Employment in petroleum fuel refining, distribution, and retailing; hydrogen production, distribution and retailing; energy storage manufacturing, installation and operations; automotive supply chains and assembly; and energy efficiency are excluded from this analysis. Values less than 5 thousand jobs not displayed in labels 100 3,500 3,500 manufacturing construction utilities pipeline mining (incl. oil & gas) professional trade agriculture other Mid-range Energy Supply Related Employment by Sector vs Frozen Policies as of January 2021 obs1 Optimistic Current Policies Scenarios Conservative 137 249 343 679 1,057 1,178 203 260 300 423 553 729 60 118 188 242 266 390 -8 -11 -17 -22 -28 66 98 119 176 242 313 41 55 65 98 141 186 35 42 46 71 107 144 -500 0 500 1,000 1,500 2,000 2,500 3,000 3,500 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 73 165 357 606 793 953 163 203 260 355 420 592 39 75 153 209 228 335 -6 -12 -16 -22 47 62 96 148 193 261 31 39 55 83 109 153 29 36 42 58 76 112 -500 0 500 1,000 1,500 2,000 2,500 3,000 3,500 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 9 310 334 324 285 0 613 705 741 701 6 513 550 575 602 51 50 49 49 8 325 294 267 231 4 333 368 383 370 0 276 292 296 278 36 36 35 34 58 78 88 78 onstruction utilities pipeline mining (incl. oil & gas) professional trade agriculture other 575 733 899 359 394 544 204 201 288 -9 -10 -13 150 179 235 86 105 144 60 73 106 2029 2030 2031 2032 2033 2034 2035 Mid-range 101 Related Employment by Sector vs Frozen Policies as of January 2021 Optimistic Current Policies Scenarios ative The Inflation Reduction Act and Infrastructure Law could increase energy supply related employment1 by about 1.4-1.7 million additional jobs in 2030 and 2.2-2.9 million by 2035. That includes about 600,000 additional manufacturing jobs in 2030 and roughly one million more manufacturing jobs in 2035, primarily in solar PV and wind turbine component manufacturing. Mid-range Current Policies Scenarios 1 - Employment in petroleum fuel refining, distribution, and retailing; hydrogen production, distribution and retailing; energy storage manufacturing, installation and operations; automotive supply chains and assembly; and energy efficiency are excluded from this analysis. Values less than 5 thousand jobs not displayed in labels 1,428 1,674 2,207 387 585 963 1,451 1,800 2,388 543 820 1,056 1,676 2,007 2,911 The Inflation Reduction Act and Infrastructure Law could increase energy supply related employment1 by about 1.4-1.7 million additional jobs in 2030 and 2.2-2.9 million by 2035. 137 249 343 679 1,057 1,178 203 260 300 423 553 729 60 118 188 242 266 390 -8 -11 -17 -22 -28 66 98 119 176 242 313 41 55 65 98 141 186 35 42 46 71 107 -500 0 500 1,000 1,500 2,000 2,500 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 73 165 357 606 793 953 163 203 260 355 420 592 39 75 153 209 228 335 -6 -12 -16 -22 47 62 96 148 193 261 31 39 55 83 109 153 29 36 42 58 76 112 -500 0 500 1,000 1,500 2,000 2,500 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 259 321 349 310 334 324 285 415 527 590 613 705 741 701 432 510 566 513 550 575 602 51 51 51 51 50 49 49 415 393 358 325 294 267 231 255 299 334 333 368 383 370 249 268 280 276 292 296 278 33 32 36 36 36 35 34 24 40 49 58 78 88 78 126 222 354 575 733 899 190 238 288 359 394 544 54 96 166 204 201 288 -5 -9 -10 -13 62 85 113 150 179 235 38 51 65 86 105 144 33 40 47 60 73 106 -500 0 500 1,000 1,500 2,000 2,500 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 The Inflation Re and Infrastructur increase ener related employ about 1.4-1.7 additional jobs 2.2-2.9 million That includes ab additional man jobs in 2030 an one million manufacturin 2035, primarily and wind turbine manufactu 1 - Employment in petroleum fuel refining, distribution, and retailing; hydrogen production, distribution and retailing; energy storage manufacturing, installation and operations; automotive supply chains and assembly; and energy efficiency are excluded from this analysis. Mid-range Current Policies Scenarios Values less than 5 thousand jobs not displayed in labels Net: 505 733 1,033 1,428 1,674 2,207 387 585 963 1,451 1,800 2,388 543 820 1,056 1,676 2,007 137 249 343 679 1,057 1,178 203 260 300 423 553 729 60 118 188 242 266 390 -8 -11 -17 -22 -28 66 98 119 176 242 313 41 55 65 98 141 186 35 42 46 71 107 0 500 1,000 1,500 2,000 2,500 73 165 357 606 793 953 163 203 260 355 420 592 39 75 153 209 228 335 -6 -12 -16 -22 47 62 96 148 193 261 31 39 55 83 109 153 29 36 42 58 76 112 0 500 1,000 1,500 2,000 2,500 259 321 349 310 334 324 285 415 527 590 613 705 741 701 432 510 566 513 550 575 602 51 51 51 51 50 49 49 415 393 358 325 294 267 231 255 299 334 333 368 383 370 249 268 280 276 292 296 278 33 32 36 36 36 35 34 24 40 49 58 78 88 78 126 222 354 575 733 899 190 238 288 359 394 544 54 96 166 204 201 288 -5 -9 -10 -13 62 85 113 150 179 235 38 51 65 86 105 144 33 40 47 60 73 106 0 500 1,000 1,500 2,000 2,500 and Infras increas related about additiona 2.2-2.9 That inclu addition jobs in 2 one manuf 2035, pr and wind ma Net: 505 733 1,033 1,428 1,674 2,207 387 585 963 1,451 1,800 2,388 543 820 1,056 1,676 2,007 That includes about 600,000 additional manufacturing jobs in 2030 and roughly one million more manufacturing jobs in 2035, primarily in solar PV and wind turbine component manufacturing. That includes about 600,000 additional manufacturing jobs in 2030 and roughly one million more manufacturing jobs in 2035, primarily in solar PV and wind turbine component manufacturing. 500 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 500 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 500 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 1 - Employment in petroleum fuel refining, distribution, and retailing; hydrogen production, distribution and retailing; energy storage manufacturing, installation and operations; automotive supply chains and assembly; and energy efficiency are excluded from this analysis. Mid-range Current Policies Scenarios Values less than 5 thousand jobs not displayed in labels 1 - Employment in petroleum fuel refining, distribution, and retailing; hydrogen production, distribution and retailing; energy storage manufacturing, installation and operations; automotive supply chains and assembly; and energy efficiency are excluded from this analysis. Values less than 5 thousand jobs not displayed in labels 1 - Employment in petroleum fuel refining, distribution, and retailing; hydrogen production, distribution and retailing; energy storage manufacturing, installation and operations; automotive supply chains and assembly; and energy efficiency are excluded from this analysis. Values less than 5 thousand jobs not displayed in labels 1 - Employment in petroleum fuel refining, distribution, and retailing; hydrogen production, distribution and retailing; energy storage manufac operations; automotive supply chains and assembly; and energy efficiency are excluded from this analysis. Mid-range Current Policies Scenarios Values less than 5 thousand jobs no Total Energy Supply Related Employment by Sector thousands of jobs1 Mid-range Current Policies Scenarios 280 479 621 683 1,044 1,415 1,490 494 808 937 1,028 1,251 1,427 1,541 572 709 837 902 1,006 1,070 1,180 51 57 56 55 54 45 52 415 389 350 314 277 245 202 289 399 469 500 595 678 726 265 325 354 365 416 464 486 33 32 36 36 35 35 31 29 80 97 113 157 203 229 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 ulture other Optimistic Optimistic 280 479 621 683 1,044 1,415 1,490 494 808 937 1,028 1,251 1,427 1,541 572 709 837 902 1,006 1,070 1,180 51 57 56 55 54 45 52 415 389 350 314 277 245 202 289 399 469 500 595 678 726 265 325 354 365 416 464 486 33 32 36 36 35 35 31 29 80 97 113 157 203 229 0 1,000 2,000 3,000 4,000 5,000 6,000 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 280 415 537 697 972 1,151 1,266 494 768 880 988 1,183 1,294 1,403 572 689 795 866 973 1,032 1,125 51 57 58 56 55 46 53 415 392 356 319 282 251 209 289 380 433 477 566 629 674 265 315 338 355 400 433 453 33 32 36 36 36 35 34 29 74 91 108 144 173 197 0 1,000 2,000 3,000 4,000 5,000 6,000 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 280 468 594 694 940 1,091 1,212 494 795 915 1,016 1,186 1,268 1,356 572 703 815 880 968 1,004 1,078 51 56 56 55 54 46 54 415 390 354 320 285 257 218 289 395 456 494 568 615 648 265 323 350 365 404 429 444 33 32 36 36 36 35 34 29 78 95 113 146 170 190 0 1,000 2,000 3,000 4,000 5,000 6,000 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 259 321 349 310 334 324 285 415 527 590 613 705 741 701 432 510 566 513 550 575 602 51 51 51 51 50 49 49 415 393 358 325 294 267 231 255 299 334 333 368 383 370 249 268 280 276 292 296 278 33 32 36 36 36 35 34 24 40 49 58 78 88 78 manufacturing construction utilities pipeline mining (incl. Mid-range Current Policies Scenarios oil & gas) professional trade agriculture other Mid range 102 Optimistic Frozen Policies (Jan 21) Conservative 342 372 340 365 358 313 605 677 728 828 874 811 649 719 713 764 804 790 51 51 51 50 49 49 393 358 325 294 267 231 333 371 381 418 436 413 285 298 300 318 324 300 32 36 36 36 35 34 45 55 66 86 97 85 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Under Current Policies, over 4.2 million mericans will be employed in energy supply ated sectors1 by 2030 (up from ~2.4 million today), rising to 5.2-5.9 million by 2035. t includes roughly a million manufacturing positions in 2030 and ~1.2-1.5 million manufacturing jobs in 2035. 1 - Employment in petroleum fuel refining, distribution, and retailing; hydrogen production, distribution and retailing; energy storage manufacturing, installation and operations; automotive supply chains and assembly; and energy efficiency are excluded from this analysis. Values less than 5 thousand jobs not displayed in labels 6,000 6,000 utilities pipeline mining (incl. oil & gas) professional trade agriculture other 1 - Employment in petroleum fuel refining, distribution, and retailing; hydrogen production, distribution and retailing; energy storage manufacturing, installation and operations; automotive supply chains and assembly; and energy efficiency are excluded from this analysis. Values less than 5 thousand jobs not displayed in labels 1 - Employment in petroleum fuel refining, distribution, and retailing; hydrogen production, distribution and retailing; energy storage manufacturing, installation and operations; automotive supply chains and assembly; and energy efficiency are excluded from this analysis. Values less than 5 thousand jobs not displayed in labels 102 Air Pollution and Public Health Impacts 103 104 e Avoided Premature Deaths From Exposure to Fine Particulate Matter From Energy Activities vs Frozen Policies, 2023-2035 17.6 28.1 36.8 Reductions in fine particulate pollution spurred by the Inflation Reduction Act and Infrastructure Law could avoid roughly 17,000-37,000 premature deaths from 2023-2035, saving ~$150-325 billion in economic damages from avoided mortalities alone. lative Avoided Premature Deaths From Exposure to Fine Particulate Matter From Energy Activities vs Frozen Policies, 2023-2035 ds 17.6 28.1 36.8 Reductions in fine particulate pollution spurred by the Inflation Reduction Act and Infrastructure Law could avoid roughly 17,000-37,000 premature deaths from 2023-2035, saving ~$150-325 billion in economic damages from avoided mortalities alone. Cumulative Avoided Premature Deaths From Exposure to Fine Particulate Matter From Energy Activities vs Frozen Policies, 2023-2035 housands Current Policies Scenarios 28.1 36.8 Mid-range Optimistic Reductions in fine particulate pollution spurred by the Inflation Reduction Act an Infrastructure Law could avoid roughly 17,000-37,000 premature deaths from 2023-2035, saving ~$150-325 billion in economic damages from avoided mortalities alone. Cumulative Avoided Premature Deaths From Exposure to Fine Particulate Matter From Energy A thousands Current Policies Scenarios 17.6 28.1 Conservative Mid-range Optimistic ne Particulate Matter From Energy Activities vs Frozen Policies, 2023-203 Cumulative Avoided Premature Deaths From Exposure to Fine Particulate thousands Cumulative Avoided Premature Deaths From Exposure to Fine Particulate Matter From Energy Activities thousands Premature Deaths From Exposure to Fine Particulate Matter From Energy Activities vs Frozen Policies, 2023-2035 Current Policies Scenarios Reductions in fine particulate pollution spurred by the Inflation Reduction Act and Infrastructure Law could avoid roughly 17,000-37,000 premature deaths from 2023-2035, saving ~$150-325 billion in economic damages from avoided mortalities alone. 104 nnual Avoided Premature Deaths From Exposure to Fine Particulate Matter Annual Avoided Premature Deaths From Exposure to Fine Particulate Matter From Energy Activities vs Frozen Policies as of January 2021 thousands housands -0.5 -0.6 -0.7 -1.6 -2.6 -4.1 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 -0.2 -0.5 -1.3 -2.8 -4.6 -7.0 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 -0.7 -1.5 -1.9 -3.3 -5.4 -8.2 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 electric generation - coal electric generation - gas commercial & institutional - coal commercial & institutional - gas commercial & institutional - oil commercial & institutional - other residential - gas residential - oil residential - other coal mining oil and gas production light duty autos light duty trucks medium duty trucks heavy duty trucks buses Optimistic Current Policies Scenarios Mid-range Conservative By 2030, the Inflation Reduction Act and Infrastructure Law could save about 1,600-3,300 American lives annually from avoided exposure to fine particulate pollution from energy activities, rising to about 4,100-8,200 lives per year in 2035. Current Policies Scenarios Mid-range Current Policies Scenarios Mid-range Optimistic Conservative 105 33 29 27 24 23 19 0 5 10 15 20 25 30 35 40 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 electric generation - coal electric generation - gas commercial & institutional - coal commercial & institutional - gas commercial & institutional - oil commercial & institutional - other residential - gas residential - oil residential - other coal mining oil and gas production light duty autos light duty trucks medium duty trucks heavy duty trucks buses 33 30 27 25 24 21 0 5 10 15 20 25 30 35 40 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 33 29 28 26 26 24 0 5 10 15 20 25 30 35 40 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 e Deaths From Exposure to Fine Particulate Matter From Energy Activities Optimistic Current Policies Scenarios Mid-range Conservative es (Jan. ‘21) 9 27 29 28 2029 2030 2031 2032 2033 2034 2035 Relative to 2022 levels, total annual deaths from fine particulate pollution decline 23-28% by 2030 and 30-42% by 2035 under Current Policies. Annual Premature Deaths From Exposure to Fine Particulate Matter From Energy Activities thousands Current Policies Scenarios Mid-range Conservative Frozen Policies (Jan. ‘21) 33 29 27 24 23 19 0 5 10 15 20 25 30 35 40 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 e e c c c c r r r c o l l m h b 33 30 27 25 24 21 0 5 10 15 20 25 30 35 40 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 33 29 28 26 26 24 0 5 10 15 20 25 30 35 40 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Annual Premature Deaths From Exposure to Fine Particulate Matter From Energy Activities thousands Optimistic Current Policies Scenarios Mid-range Conservative Frozen Policies (Jan. ‘21) 34 33 30 29 27 29 28 0 5 10 15 20 25 30 35 40 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2 Current Policies Scenarios Mid-range Optimistic Optimistic Conservative 40 Relative to 2022 levels, total annual deaths from fine particulate pollution decline 23-28% by 2030 and 30-42% by 2035 under Current Policies. 106 107 Additional impacts of Inflation Reduction Act policies Beyond the direct emissions reduction impacts of the policies modeled in this report, the Inflation Reduction Act contains important policy measures and programs that will spur innovation and maturation of nascent advanced energy industries, build U.S. clean energy manufacturing and supply chains, improve public health and environmental justice, and drive investment and economic opportunities in communities across the United States. IRA builds on the demonstration and hubs funding in the Infrastructure Law by providing early market deployment opportunities over the next decade that will drive innovation and maturation of important nascent clean technologies that need to be ready for wide-scale deployment in the 2030s and 2040s, including clean hydrogen, carbon capture, zero-carbon liquid fuels, direct air capture, advanced nuclear and geothermal energy, and more. These technologies all have access to robust deployment subsidies (many for the first time)1 that are likely to have a similar catalytic impact as the tax credits that cultivated wind and solar energy industries and drove costs down by ~90% for solar and ~70% for wind from 2010 to 2021. lso contains robust support for the development of American manufacturing of solar, wind, battery and electric vehicle components and mbly as well as critical minerals processing. The bill ties bonus tax incentives for clean electricity and credits for consumer clean vehicles purchases omestic content sourcing standards, providing strong demand for U.S. materials and manufacturing.2 It also provides $2 billion in grants and $40 n in loans to retool American auto manufacturing to produce clean vehicles and $37 billion in new tax credits to spur investment in America’s city to produce and assemble wind and solar PV components, batteries and clean vehicles, and process critical minerals.3 An additional $0.5 billion is appropriated for the President to use the Defense Production Act to build American supply chains for heat pump and battery manufacturing, critical rals, and other strategic priorities.4 Those policies are important to expand supply chains and enable rapid scale-up of these technologies, and they also create hundreds of thousands of manufacturing jobs across the country, giving countless communities a direct, tangible, near-term stake in the n energy transition. IRA also contains robust support for the development of American manufacturing of solar, wind, battery and electric vehicle components and assembly as well as critical minerals processing. Additional impacts of Inflation Reduction Act poli ckage of environmental justice provisions in IRA provide at least $60 billion to reduce harmful pollution in environmentally overburdened munities, ensure more equitable access to renewable energy and energy efficiency and building electrification opportunities, and improve public th and climate resiliency. 109 cut pollution in low-income communities and areas burdened by the worst air pollution in the country. This munity-led environmental and climate justice projects and more than $4 billion in funds to reduce air pollution vehicles like garbage trucks and city buses with zero-emissions vehicles, and improve interior air quality in aw funds ‘fenceline’ air pollution monitoring to empower EPA and local air quality agencies to track and reduce ommunities, and it appropriates needed funding for the White House to map and identify environmental lution.2 IRA reinstates (and adjusts for inflation) the ‘polluter pays’ Superfund Tax to cover the cost of taminated industrial sites, and it invests $1 billion to improve energy and water efficiency, indoor air quality, and over $3 billion to improve neighborhood walkability, safety, and affordable transportation access.3 llars to expand equitable access to clean and efficient technologies. The $27 billion Greenhouse Gas his funding to deploy clean energy and pollution-reducing technologies in low-income and disadvantaged to provide financial assistance for clean energy projects benefiting disadvantaged communities.4 Hundreds of hority will help Tribal and Native Hawaiian communities improve climate resilience, access clean electricity, programs also provide $8.8 billion to ensure access to energy efficiency and building electrification funds for ck tax liability to take advantage of other tax credits.6 (Sec. 60201), $3 billion for Grants to Reduce Air Pollution at Ports (Sec. 60102), $1 billion for Clean Heavy-Duty Vehicles (Sec. 60101), $60 million for Diesel Emissions Reductions (Sec. 60104), and 5 million Environmental and Climate Data Collection program for the White House Council on Environmental Quality (Sec. 60401). ncy or Water Efficiency or Climate Resilience of Affordable Housing (Sec. 30002), and Neighborhood Access and Equity Grant Program (Sec. 60501). lion out of $27 billion in total funding to low-income and disadvantaged communities. limate Resilience and Native Hawaiian Climate Resilience (Sec. 80001 and 80002), $150m Tribal Electrification Program (Sec, 80003) and $13m Emergency Drought Relief for Tribes (Sec. 80004). s Program (Sec. 50121) and $4.5 billion High-Efficiency Electric Home Rebate Program (Sec. 50122). A variety of programs will direct funding to cut pollution in low-income communities and areas burdened by the worst air pollution in the country. Additional impacts of Inflation Reduction Act policies The bill ties bonus tax incentives for clean electricity and credits for consumer clean vehicles purchases to domestic content sourcing standards, providing strong demand for U.S. materials and manufacturing.2 It also provides $2 billion in grants and $40 billion in loans to retool American auto manufacturing to produce clean vehicles and $37 billion in new tax credits to spur investment in America’s capacity to produce and assemble wind and solar PV components, batteries and clean vehicles, and process critical minerals.3 An additional $0.5 billion is also appropriated for the President to use the Defense Production Act to build American supply chains for heat pump and battery manufacturing, critical minerals, and other strategic priorities.4 Those policies are important to expand supply chains and enable rapid scale-up of these technologies, and they will also create hundreds of thousands of manufacturing jobs across the country, giving countless communities a direct, tangible, near-term stake in the clean energy transition. 1 – These include the clean hydrogen PTC (Sec. 13204), 45q tax credit for CCS (Section 13104), and new technology-neutral production and investment tax credits for all carbon-free electricity generation (Sec. 13701 and 13702) and a clean fuel production tax credit (Sec. 13704). The bill also provides $40 billion in expanded loan authority for the DOE Loan Programs Office (LPO) to support investment in nascent clean energy sectors. 2 – A bonus 10% increase in the value of the production tax credit (Sec. 13101 and 13701) and 10 percentage point increase in the investment tax credit (Sec. 13102 and 13702) are available for clean electricity projects that meet domestic content requirements for materials and manufactured components. The consumer clean vehicles tax credit (Section 13401) is also tied to increasing requirements for sourcing of batteries and critical minerals from North America or our trade partners. 3 – See the Domestic Manufacturing Conversion Grants (Sec. 50143), Advanced Vehicle Technology Manufacturing loan program at DOE (Sec. 50142), and the 48C Advanced Energy Project Credit (Sec. 13501) and Advanced Manufacturing Production Credit (Sec. 13502). 4 – Enhanced Use of Defense Production Act of 1950 (Sec. 30001). p ( ) g q g p cturing Conversion Grants (Sec. 50143), Advanced Vehicle Technology Manufacturing loan program at DOE (Sec. 50142), and the 48C Advanced Energy Project Credit (Sec. 13501) and Advanced Manufacturing Production Credit (Sec. 13502) d f ( ) 1 – See the $3 billion Environmental and Climate Justice Block Grants program (Sec. 60201), $3 billion for Grants to Reduce Air Pollution at Ports (Sec. 60102), $1 billion for Clean Heavy-Duty Vehicles (Sec. 60101), $60 million for Diesel Emissio $50 million for Funding to Address Air Pollution at Schools (Sec. 60106). 2 – See the $281 million Funding to Address Air Pollution (Sec. 60105) and $32.5 million Environmental and Climate Data Collection program for the White House Council on Environmental Quality (Sec. 60401). 3 – See the Reinstatement of Superfund (Sec. 13601), Improving Energy Efficiency or Water Efficiency or Climate Resilience of Affordable Housing (Sec. 30002), and Neighborhood Access and Equity Grant Program (Sec. 60501). 4 – The Greenhouse Gas Reduction Fund (Sec. 60103) dedicates at least $15 billion out of $27 billion in total funding to low-income and disadvantaged communities. 5 – See Tribal Energy Loan Guarantee Program (Sec. 50145), $260m for Tribal Climate Resilience and Native Hawaiian Climate Resilience (Sec. 80001 and 80002), $150m Tribal Electrification Program (Sec, 80003) and $13m Emergency Drought 6 – See the $4.3 billion Home Energy Performance-Based Whole-House Rebates Program (Sec. 50121) and $4.5 billion High-Efficiency Electric Home Rebate Program (Sec. 50122). 1 Sec. 50145), $260m for Tribal Climate Resilience and Native Hawaiian Climate Resilience (Sec. 80001 and 80002), $150m Tribal Electrification Program (Sec, 80003) and $13m Emergency Drought Relief for Tribes (Sec. 80004). ce-Based Whole-House Rebates Program (Sec. 50121) and $4.5 billion High-Efficiency Electric Home Rebate Program (Sec. 50122). Additional impacts of Inflation Reduction Act poli Bonus tax credits are available for investments in clean electricity generation sited in traditional ‘energy communities’ across d as areas with significant historical employment in energy resource, extraction, processing or transportation or where coal plants or mines ecent decades, providing a strong financial incentive to retool and repower existing power plant sites and reinvest in energy producing hese tax incentives complement $9.7 billion in financial assistance for rural electric cooperatives to install zero-emissions generation, carbon upgrades and $5 billion in appropriations to create a new energy community reinvestment financing program that will support up to $250 arantees to retool, repower, repurpose, or replace aging energy infrastructure and install carbon capture and other low emissions retrofits neration and fuels production and refining facilities.2 The Act also sets aside $4 billion in tax incentives that will spur at least $13 billion in nufacturing investments in these energy communities.3 of most of these provisions are beyond the scope of this project to model, but they will nevertheless deliver real, salient benefits for diverse ross the country. Additional impacts of Inflation Reduction Act The IRA provides grants, loans, and tax incentives that will drive hundreds o between now and 2030. Bonus tax credits are available for investments in cle America, defined as areas with significant historical employment in energy re have closed in recent decades, providing a strong financial incentive to retoo communities.1 These tax incentives complement $9.7 billion in financial assis capture, or grid upgrades and $5 billion in appropriations to create a new ene billion in loan guarantees to retool, repower, repurpose, or replace aging ene at electricity generation and fuels production and refining facilities.2 The Act clean energy manufacturing investments in these energy communities.3 The full impacts of most of these provisions are beyond the scope of this proj communities across the country. Additional impacts of Inflation Reduction Act poli This includes $3 billion for block grants for community-led environmental and climate justice projects and more than $4 billion in funds to reduce air pollution at America’s ports, replace dirty heavy duty vehicles like garbage trucks and city buses with zero-emissions vehicles, and improve interior air quality in schools in low-income communities.1 The law funds ‘fenceline’ air pollution monitoring to empower EPA and local air quality agencies to track and reduc pollution burdens on the most vulnerable communities, and it appropriates needed funding for the White House to map and identify environmental justice communities on the frontlines of pollution.2 IRA reinstates (and adjusts for inflation) the ‘polluter pays’ Superfund Tax to cover the cost of remediating the worst environmentally contaminated industrial sites, and it invests $1 billion to improve energy and water efficiency, indoor air quality, and climate resiliency of affordable housing and over $3 billion to improve neighborhood walkability, safety, and affordable transportation access.3 The IRA also dedicates tens of billions of dollars to expand equitable access to clean and efficient technologies. The $27 billion Greenhouse Gas Reduction Fund devotes more than half of this funding to deploy clean energy and pollution-reducing technologies in low-income and disadvantaged communities and to establish ‘green banks’ to provide financial assistance for clean energy projects benefiting disadvantaged communities.4 Hundreds of millions in grants and $20 billion in loan authority will help Tribal and Native Hawaiian communities improve climate resilience, access clean electricity, and electrify buildings.5 Finally, two rebate programs also provide $8.8 billion to ensure access to energy efficiency and building electrification funds for low- and middle-income households that lack tax liability to take advantage of other tax credits.6 109 1 5 – See Tribal Energy Loan Guarantee Program (Sec. 50145), $260m for Tribal Climate Resilience and Native Hawaiian Climate Resilience (Sec. 80001 and 80002), $150m Tribal Electrification Program (Sec, 80003) and $13m Emergency Drought Relief for Tribes (Sec. 80004). 6 – See the $4.3 billion Home Energy Performance-Based Whole-House Rebates Program (Sec. 50121) and $4.5 billion High-Efficiency Electric Home Rebate Program (Sec. 50122). l impacts of Inflation Reduction Act policies (continued) s grants, loans, and tax incentives that will drive hundreds of billions of dollars in cumulative investment in American energy communities nd 2030. Additional impacts of Inflation Reduction Act policies (continued) IRA provides grants, loans, and tax incentives that will drive hundreds of billions of dollars in cumulative investment in American energy communities ween now and 2030. Bonus tax credits are available for investments in clean electricity generation sited in traditional ‘energy communities’ across rica, defined as areas with significant historical employment in energy resource, extraction, processing or transportation or where coal plants or mines closed in recent decades, providing a strong financial incentive to retool and repower existing power plant sites and reinvest in energy producing munities.1 These tax incentives complement $9.7 billion in financial assistance for rural electric cooperatives to install zero-emissions generation, carbon ure, or grid upgrades and $5 billion in appropriations to create a new energy community reinvestment financing program that will support up to $250 on in loan guarantees to retool, repower, repurpose, or replace aging energy infrastructure and install carbon capture and other low emissions retrofits ectricity generation and fuels production and refining facilities.2 The Act also sets aside $4 billion in tax incentives that will spur at least $13 billion in n energy manufacturing investments in these energy communities.3 IRA provides grants, loans, and tax incentives that will drive hundreds of billions of dollars in cumulative investment in American energy communities ween now and 2030. Additional impacts of Inflation Reduction Act policies (continued) Bonus tax credits are available for investments in clean electricity generation sited in traditional ‘energy communities’ across rica, defined as areas with significant historical employment in energy resource, extraction, processing or transportation or where coal plants or mines closed in recent decades, providing a strong financial incentive to retool and repower existing power plant sites and reinvest in energy producing munities.1 These tax incentives complement $9.7 billion in financial assistance for rural electric cooperatives to install zero-emissions generation, carbon ure, or grid upgrades and $5 billion in appropriations to create a new energy community reinvestment financing program that will support up to $250 on in loan guarantees to retool, repower, repurpose, or replace aging energy infrastructure and install carbon capture and other low emissions retrofits ectricity generation and fuels production and refining facilities.2 The Act also sets aside $4 billion in tax incentives that will spur at least $13 billion in n energy manufacturing investments in these energy communities.3 s are beyond the scope of this project to model, but they will nevertheless deliver real, salient benefits for diverse The full impacts of most of these provisions are beyond the scope of this project to model, but they will nevertheless deliver real communities across the country. full impacts of most of these provisions are beyond the scope of this project to model, but they will nevertheless deliver real, salient benefits for diverse munities across the country 1 – A bonus 10% increase in the value of the production tax credit (Section 13101 and 13701) and 10 percentage point increase in the investment tax credit (Section 13102 and 13702) are available for clean electricity projects installed in 2 – See the USDA Assistance for Rural Electric Cooperatives program (Sec. 22004) and new DOE Section 1703 Energy Infrastructure Reinvestment Financing program (Sec. 50144). 3 – The 48C Advanced Energy Project Credit (Sec. 13501) specifically sets aside $4 billion (out of $10 billion in total incentives) for energy communities; the 30% investment tax credit will spur at least $13.3 billion in investment. 110 repeatproject.org
https://openalex.org/W4361921540	https://aacr.figshare.com/articles/journal_contribution/SFigure_5_from_Macropinocytosis_of_Bevacizumab_by_Glioblastoma_Cells_in_the_Perivascular_Niche_Affects_their_Survival/22462761/1/files/39914064.pdf	English	null	SFigure 5 from Macropinocytosis of Bevacizumab by Glioblastoma Cells in the Perivascular Niche Affects their Survival	null	2,023	cc-by	340	Suppl Figure 5 Suppl Figure 5 uppl Figure 5 1 1 SFigure 5. Co-localization of a fraction of the bevacizumab-VEGF complex with Rab4 and a fraction with LAMP1 in CD133+ GBM tumor cells. CD133+ GBM cells (08-387) were plated for 18 h as in Figure 3. Bevacizumab (250 µg/ml) was incubated with biotinylated-rec-human-VEGF (100 ng/ml) for 1 h, and the mixture added to the cells for 5 min, the cells washed and fixed or the media replaced and the cells washed and fixed at 3 h. A&B, The cells were reacted with Alexa-488-anti-human IgG and anti-Rab4 or anti-LAMP1 antibody, and Alexa-594-conjugated secondary antibody, as well as Alexa- 647-Streptavidin, followed by DAPI nuclear stain and confocal microscopy. Arrows denote bevacizumab (green), Rab4 or LAMP1 (red) and VEGF (magenta), as well as their co-localization (A&B). Triple labeling is indicated by the arrows in both A & B. C&D, The percent bevacizumab-VEGF complex co-localized with Rab4 (5 min) (C) or LAMP1 (3 h) (D) is plotted as the mean+SEM based on the Mander’s coefficient. Scale bars denote 5-µm. SFigure 5. Co-localization of a fraction of the bevacizumab-VEGF complex with Rab4 and a fraction with LAMP1 in CD133+ GBM tumor cells. CD133+ GBM cells (08-387) were plated for 18 h as in Figure 3. Bevacizumab (250 µg/ml) was incubated with biotinylated-rec-human-VEGF (100 ng/ml) for 1 h, and the mixture added to the cells for 5 min, the cells washed and fixed or the media replaced and the cells washed and fixed at 3 h. A&B, The cells were reacted with Alexa-488-anti-human IgG and anti-Rab4 or anti-LAMP1 antibody, and Alexa-594-conjugated secondary antibody, as well as Alexa- 647-Streptavidin, followed by DAPI nuclear stain and confocal microscopy. Arrows denote bevacizumab (green), Rab4 or LAMP1 (red) and VEGF (magenta), as well as their co-localization (A&B). Triple labeling is indicated by the arrows in both A & B. C&D, The percent bevacizumab-VEGF complex co-localized with Rab4 (5 min) (C) or LAMP1 (3 h) (D) is plotted as the mean+SEM based on the Mander’s coefficient. Scale bars denote 5-µm. 2
https://openalex.org/W4237866448	https://gi.copernicus.org/articles/9/499/2020/gi-9-499-2020.pdf	English	null	Radiation tolerance of the PNI RM3100 magnetometer for a Europa lander mission	null	2,020	cc-by	6,025	Leonardo H. Regoli1,2, Mark B. Moldwin1, Connor Raines1, Tom A. Nordheim3, Cameron A. Miller4, Martin Carts5, d S A P i4 Leonardo H. Regoli1,2, Mark B. Moldwin1, Connor Raines1, Tom A. Nordheim3, Cameron A. Miller4, Martin Carts5, and Sara A. Pozzi4 1Climate and Space Sciences and Engineering, University of Michigan, Ann Arbor, Michigan, USA 1Climate and Space Sciences and Engineering, University of Michigan, Ann Arbor, Michigan, USA 2The Johns Hopkins University Applied Physics Laboratory, Laurel, Maryland, USA 3 1Climate and Space Sciences and Engineering, University of Michigan, Ann Arbor, Michigan, USA 2The Johns Hopkins University Applied Physics Laboratory, Laurel, Maryland, USA 3Jet Propulsion Laboratory, California Institute of Technology, Pasadena, California, USA 4Department of Nuclear Engineering and Radiological Sciences, University of Michigan, Ann Arbor, Michigan, USA 5Civil Servant, Radiation Effects and Analysis Group, NASA Goddard Space Flight Center, Greenbelt, Maryland, USA 1Climate and Space Sciences and Engineering, University of Michigan, Ann Arbor, Michig 2The Johns Hopkins University Applied Physics Laboratory, Laurel, Maryland, USA e Jo s op s U ve s ty pp ed ys cs abo ato y, au e , a y a d, US 3Jet Propulsion Laboratory, California Institute of Technology, Pasadena, California, USA p y, gy, , , 4Department of Nuclear Engineering and Radiological Sciences, University of Michigan, Ann Arbor, Michigan, USA 5Civil Servant, Radiation Effects and Analysis Group, NASA Goddard Space Flight Center, Greenbelt, Maryland, USA Correspondence: Leonardo H. Regoli (leonardo.regoli@jhuapl.edu) Received: 28 April 2020 – Discussion started: 25 May 2020 Revised: 16 October 2020 – Accepted: 12 November 2020 – Published: 23 December 2020 Received: 28 April 2020 – Discussion started: 25 May 2020 Revised: 16 October 2020 – Accepted: 12 November 2020 – Published: 23 December 2020 Abstract. The results of two radiation test campaigns on a low-cost commercial off-the-shelf magnetometer are pre- sented. The test setup and the total ionization dose (TID) levels studied were designed to meet the requirements of a mission to land on Europa. Based on the Europa Lander Sci- ence Deﬁnition Team report, instruments inside an aluminum vault at the surface of Europa would need to withstand TID of up to 300 krad(SI). In order to evaluate the performance of the PNI RM3100 magnetometer, nine separate sensors were irradiated at two different facilities during two separate cam- paigns and under different conﬁgurations, including passive and active tests. Of the nine sensors, seven survived the TID of 300 krad(SI) while the other two sensors started present- ing failures after reaching 150 krad(SI). Leonardo H. Regoli1,2, Mark B. Moldwin1, Connor Raines1, Tom A. Nordheim3, Cameron A. Miller4, Martin Carts5, d S A P i4 Post-irradiation tests showed that eight of the nine sensors continued to work with- out appreciable degradation after stopping exposure, while one sensor stopped working altogether. (Prinzie et al., 2018). Single-event upsets (SEUs) refer to nondestructive events that can alter the logic state of circuits, introducing errors into measurements (Hands et al., 2018). With time, radiation exposure will degrade the semiconduc- tors by accumulating positively charged holes at the silicon interface, leading to long-term effects that can cause com- plete malfunction if the total amount of exposure exceeds a certain threshold. This type of damage is measured in terms of the total ionization dose (TID). In this paper, we focus on the TID effects on a commercial off-the-shelf magnetometer that makes use of a measurement principle known as magneto-induction. The importance of TID lies in the fact that its effects will determine the time the sensor will survive under any given environment. The sensor provides magnetic ﬁeld measurements in the three dimen- sions; however, the entire analysis presented in this paper is related to the magnitude of the ﬁeld in order to account for the three axes when evaluating the performance. 1 Introduction Radiation exposure of instruments in space can vary signif- icantly depending on where the instruments are expected to operate. The low-Earth-orbit (LEO) environment has been well characterized in order to account for effects of radiation on humans. In general, the main sources of charged-particle radiation are galactic cosmic rays (GCRs), solar energetic particles (SEPs), and trapped particles in the inner radiation When designing electronics, one of the aspects to take into account is the susceptibility of semiconductors to radiation effects. While this is a minor concern for everyday consumer electronics, it becomes an important aspect in the design of instrumentation for space applications. There are different ways in which radiation can affect elec- tronics, and there are different ways to quantify these effects L. H. Regoli et al.: Radiation tolerance PNI RM3100 These conditions would result in a TID of approximately 10 krad(SI) per year. They found different failure doses, with the MSP430 failing at 240 krad(SI). The rest of the components irradiated failed at signiﬁcantly lower doses, on the order of a few tens of krad(SI). The signals to be detected at the surface of Europa are a combination of electromagnetic waves from the surrounding environment (e.g., ion cyclotron waves arising from the mass loading of the co-rotating plasma) and induction signals gen- erated by the presence of the subsurface ocean and, possi- bly, water ﬂows near the landing site. The induction signal arising from the change in magnetic ﬁeld experienced by Eu- ropa along its orbit is quite strong (estimated to be about 12– 20 nT as reported in Khurana et al., 2009), well within the measurement capabilities of the PNI RM3100 (Regoli et al., 2018). The induction signals from water ﬂows are expected to be much smaller, but no published estimates are available (to the best of our knowledge), and assessing the suitability of the PNI RM3100 for those measurements is beyond the scope of this paper. Even with the presence of the radiation belts, the space environment in the vicinity of the Earth is relatively benign when compared with other environments in the solar system. The most extreme radiation environment in the solar system corresponds to the radiation belts of Jupiter, with high ﬂuxes of energetic ions up to 100 MeV and electrons up to 700 keV, as measured by Galileo (Cooper et al., 2001; Paranicas et al., 2001). Europa, one of the four Galilean moons and one of the most promising extraterrestrial objects in the solar system for the detection of life, is located inside Jupiter’s radiation belt, and thus any mission designed to land on the moon’s surface will have to be able to withstand these extreme conditions. For the Europa Lander mission concept currently under development by NASA, instruments would be placed inside a radiation vault similar to that used on the Juno spacecraft (Europa Lander Science Deﬁnition Team, 2016). This vault would reduce the radiation dose to roughly 150 krad(SI) over the 20 d surface mission. Implementing a radiation design factor of 2, this means that any instrument placed inside the radiation vault will have to survive and function up to at least 300 krad(SI) (Europa Lander Science Deﬁnition Team, 2016). L. H. Regoli et al.: Radiation tolerance PNI RM3100 500 L. H. Regoli et al.: Radiation tolerance PNI RM3100 Figure 1. Basic circuit showing the working principle of the magneto-inductive technology. The Schmitt trigger ﬂips the direc- tion of the current that causes the sensor to charge and discharge in both directions (from Leuzinger and Taylor, 2010). belt, mainly consisting of electrons and protons with energies ranging from 100 keV to several hundreds of mega-electron volts (Badhwar, 1997; Kovtyukh, 2018). The local Earth environment has also been studied from the point of view of the effect of radiation on electron- ics, mostly on CMOS devices. Sajid et al. (2018) analyzed the effects of TID on 65 and 130 nm N-type metal–oxide– semiconductor (NMOS) technology by simulating the space environment of a LEO satellite for a total mission duration of 3 years. They found that for both types of circuits, the leakage drain-source current for zero gate voltage increases signiﬁcantly with TID, with the effect being stronger for the 65 nm device. At the same time, the saturation current (when the gate voltage increases) remained constant, regardless of the TID level. This leakage current occurs as a consequence of an increased potential accumulated in the device, which subsequently affects the voltage threshold of the gate. Figure 1. Basic circuit showing the working principle of the magneto-inductive technology. The Schmitt trigger ﬂips the direc- tion of the current that causes the sensor to charge and discharge in both directions (from Leuzinger and Taylor, 2010). radiation environment that is particularly important for the electronics of the instrument, placing the sensing coils in- side the vault could lead to the detection of magnetic noise coming from other instruments and spacecraft subsystems. To remediate this, the sensing coils can be separated from the electronics and placed outside the vault. While the radia- tion tolerance of the PNI RM3100 coils has not been studied, coils are generally signiﬁcantly less susceptible to radiation than electronic components. Netzer et al. (2014) evaluated a series of components suit- able for the development of CubeSat missions, including an MSP430 microcontroller, the BeagleBone Black com- puter development platform, and a series of junction ﬁeld effect transistor (JFET) and complementary metal–oxide– semiconductor (CMOS) operational ampliﬁers under condi- tions similar to those encountered by missions in LEO at inclinations between 45 and 90◦and with 100 mm of alu- minum shielding. L. H. Regoli et al.: Radiation tolerance PNI RM3100 The tests presented in this work were designed to reach and surpass that value, with the aim of providing an initial evaluation of the survivability of the PNI RM3100 magnetometer. While the vault provides shielding against the Published by Copernicus Publications on behalf of the European Geosciences Union. Published by Copernicus Publications on behalf of the European Geosciences Union. Geosci. Instrum. Method. Data Syst., 9, 499–507, 2020 3 Magneto-inductive technology The RM3100 magnetometer works with a completely digital technology called the magneto-inductive principle. In order to detect the ambient magnetic ﬁeld, the magneto-inductive measurement principle makes use of an oscillator circuit con- sisting of an inductor and resistor (LR) circuit and a Schmitt trigger to ﬂip the direction of the current (Fig. 1). By ﬂipping the direction of the current, the circuit causes the inductor to go through consecutive charge and discharge cycles in both directions. The ﬁeld sensed by the coil is a combination of the ambient ﬁeld and the self-induced ﬁeld due to the current ﬂow. The selection of the current dictates Geosci. Instrum. Method. Data Syst., 9, 499–507, 2020 https://doi.org/10.5194/gi-9-499-2020 L. H. Regoli et al.: Radiation tolerance PNI RM3100 501 Figure 2. Permeability curve of the core material of the sensors, showing the region of operation and difference in charge and discharge times in the absence (left) and presence (right) of an external magnetic ﬁeld (from Leuzinger and Taylor, 2010). Figure 2. Permeability curve of the core material of the sensors, showing the region of operation and difference in charge and discharge times in the absence (left) and presence (right) of an external magnetic ﬁeld (from Leuzinger and Taylor, 2010). Figure 2. Permeability curve of the core material of the sensors, showing the region of operation and diff times in the absence (left) and presence (right) of an external magnetic ﬁeld (from Leuzinger and Taylor, 201 Figure 3. Zero-Gauss chamber at the Climate and Space Research Building. tal, with no power-consuming and radiation-sensitive com- ponents such as an analogue-to-digital converter (ADC) or ampliﬁers. This helps the sensor provide magnetic ﬁeld mea- surements with very low power consumption and good toler- ance to radiation. This paper presents the results obtained for TID on nine different RM3100 sensors during two different test sessions completed between April and July 2019. The original aim of the tests was to study the survivability of the sensor to a TID of 300 krad(SI), in preparation for a potential landed mission to the Jovian moon Europa. Geosci. Instrum. Method. Data Syst., 9, 499–507, 2020 4 Testing facilities Figure 3. Zero-Gauss chamber at the Climate and Space Research Building. The tests presented here were performed in four different stages. The ﬁrst and fourth stages involved the pre- and post- exposure characterization of the sensors being used in the study. For both stages, a zero-Gauss chamber that is located at the Climate and Space Research Building, at the Univer- sity of Michigan was used (Fig. 3). Inside the zero-Gauss chamber, periodic and small variations in the ﬁeld that might take place during the measurements are reduced to a value that is below the resolution of the instrument, as reported in Regoli et al. (2018). the region in the permeability curve where the sensor works (Fig. 2). In the absence of an external magnetic ﬁeld, the charge and discharge times in both directions are the same. If an external ﬁeld is present, the disturbance represented by it will cause the sensor to work in a different region of the permeability curve, causing a change in the charge and discharge times in each direction. Due to the particular shape of the permeabil- ity curve, the magnitude of this change is different for each direction and, by measuring the time difference between both cycles, the external ﬁeld can be determined. The second stage involved the exposure to an X-ray beam of three of the sensors and it was carried out at the De- partment of Nuclear Engineering and Radiological Sciences at the University of Michigan. The radiation source used was a commercially available Varian M9 linear acceler- ator (Fig. 4). The output of the linear accelerator is an One of the advantages of the magneto-inductive technol- ogy is that the measurement principle is completely digi- Geosci. Instrum. Method. Data Syst., 9, 499–507, 2020 https://doi.org/10.5194/gi-9-499-2020 L. H. Regoli et al.: Radiation tolerance PNI RM3100 Table 1. Noise ﬂoor (nT / √ Hz@1Hz). Sensor Pre-irradiation Post-irradiation 1 2.7202 2.8616 2 2.1505 2.7336 3 2.3379 2.0636 4 4.7632 2.9460 5 1.9776 − 6 2.8023 2.6169 7 2.2259 2.4825 8 2.5885 1.9655 9 2.4596 3.1086 L. H. Regoli et al.: Radiation tolerance PNI RM3100 502 Figure 4. Detail of the output end of the Varian M9 linear accelera- tor at the Department of Nuclear Engineering and Radiological Sci- ences at the University of Michigan. 4 Testing facilities The exit of the beam is located at the right-hand side of the ﬁgure, and the magnetometer being ir- radiated is located behind the wall on the right. Table 1. Noise ﬂoor (nT / √ Hz@1Hz). Sensor Pre-irradiation Post-irradiation teract with mass. As photon energy decreases from 1 MeV, the likelihood of interaction with matter increases, and this makes lower-energy photons more damaging than the de- sired, higher-energy, photons. Attenuation of the substantial lower-energy portion of the spectrum is the aim of the ﬁlter box. Figure 4. Detail of the output end of the Varian M9 linear accelera- tor at the Department of Nuclear Engineering and Radiological Sci- ences at the University of Michigan. The exit of the beam is located at the right-hand side of the ﬁgure, and the magnetometer being ir- radiated is located behind the wall on the right. Figure 5. Magnetometers and dosimeters ready to be tested at the GSFC Radiation Effects Facility. 5 Tests and results This section summarizes the test setup and the results ob- tained for each individual magnetometer. To make the infor- mation easier to follow, a subsection for each of the test cam- paigns is provided, with a detailed description of each partic- ular test and for any special behavior of the sensors when ap- plicable. In addition, each of the nine magnetometers is num- bered according to the order in which they were exposed, and this numeration is maintained consistently throughout the pa- per. Two different types of tests were performed in terms of whether the sensor was capturing data during exposure or not. The tests when the sensors were capturing data are re- ferred to as active, while the other ones, when the sensors were unplugged from their power supply, are referred to as passive. For all the active tests, the magnetometer worked at a data rate of 40 Hz. Figure 5. Magnetometers and dosimeters ready to be tested at the GSFC Radiation Effects Facility. 9 MeV bremsstrahlung (X-ray) beam with nominal energy of 9 MeV. The maximum dose rate of the Varian M9 is of 3 krad(SI)min−1 at 1 m, but all the tests performed were run at 20 % of that capacity, 600 rad(SI) min−1 at 1 m. The sen- sors were placed at 25 cm from the source, and since the dose rate decreases according to a 1/r2 factor, the radiation rate at the sensors during the tests was of 9.6 krad(SI)min−1. Geosci. Instrum. Method. Data Syst., 9, 499–507, 2020 5.1 Test campaign 1: University of Michigan Using the Varian M9 linear accelerator described in the pre- vious section, the ﬁrst sensor (sensor 1) was exposed while operating (active test) from 0 to 300 krad(SI) in steps of 50 krad(SI) and the output brieﬂy analyzed after each set of exposures. During the test, the magnetometer survived with- out any apparent degradation of the signal until 300 krad(SI), so two extra steps of 100 krad(SI) each were used to bring the TID to 500 krad(SI). The third stage involved the exposure to a gamma ray (X- rays originating within the nucleus) ﬁeld of six of the sen- sors and it was carried out at the NASA Goddard Space Flight Center Radiation Effects Facility high-dose-rate total- ionizing dose irradiator (Fig. 5). The sensors were situated within a spectrum-modifying PbAl ﬁlter box, which prefer- entially reduces lower-energy photons. A standard assump- tion is that TID radiation is composed entirely of 1 MeV photons, but in reality that pure original radioactive decay spectrum is smeared downwards in energy as the photons in- Figure 6 shows the magnitude of the magnetic ﬁeld mea- sured by sensor 1 in nanoteslas during each of the exposure steps, all in a single plot. Each exposure is plotted with a dif- ferent color. The ﬁrst six exposures were 50 krad(SI) steps Geosci. Instrum. Method. Data Syst., 9, 499–507, 2020 https://doi.org/10.5194/gi-9-499-2020 L. H. Regoli et al.: Radiation tolerance PNI RM3100 L. H. Regoli et al.: Radiation tolerance PNI RM3100 503 Figure 6. Magnetic ﬁeld magnitude measured during the radiation tests at the Department of Nuclear Engineering and Radiological Sciences at the University of Michigan. Figure 6. Magnetic ﬁeld magnitude measured during the radiation tests at the Department of Nuclear Engineering and Radiological Sciences at the University of Michigan. Figure 6. Magnetic ﬁeld magnitude measured during the radiation tests at the Department of Nuclear Engineering and Radiological Sciences at the University of Michigan. Figure 7. Magnitude of the magnetic ﬁeld as measured by sensors 4 (a) and 5 (b) during the test campaign at the GSFC. The data shown between vertical lines correspond to 2 min intervals collected between exposures. The vertical lines mark the TID that the sensors went through before taking the measurements. Figure 7. Magnitude of the magnetic ﬁeld as measured by sensors 4 (a) and 5 (b) during the test campaign at the GSFC. The data shown between vertical lines correspond to 2 min intervals collected between exposures. The vertical lines mark the TID that the sensors went through before taking the measurements. with a ﬁnal dose of 300 krad(SI). The last two steps were of 100 krad(SI) each for a total ﬁnal dose of 500 krad(SI). The plot only shows results up to 450 krad(SI) since the sensor started failing at around that level. tion of the test that becomes particularly pronounced after the TID reaches approximately 225 krad(SI) and continues right until failure of the sensor at above 450 krad(SI). Addition- ally, for each exposure step, a small change in the ﬁeld being measured and an increase in the noise is present. The source of this variability is the accelerator itself. At the beginning Two variabilities of interest are visible in the plot. The ﬁrst one is a steady increase in the total ﬁeld throughout the dura- Geosci. Instrum. Method. Data Syst., 9, 499–507, 2020 Geosci. Instrum. Method. Data Syst., 9, 499–507, 2020 https://doi.org/10.5194/gi-9-499-2020 https://doi.org/10.5194/gi-9-499-2020 504 L. H. Regoli et al.: Radiation tolerance PNI RM3100 Figure 8. Magnitude of the magnetic ﬁeld as measured by sensors 8 (orange) and 9 (black) during the test campaign at the GSFC. Sensor 8 started failing after a TID of 150 krad(SI). L. H. Regoli et al.: Radiation tolerance PNI RM3100 L. H. Regoli et al.: Radiation tolerance PNI RM3100 504 Figure 8. Magnitude of the magnetic ﬁeld as measured by sensors 8 (orange) and 9 (black) during the test campaign at the GSFC. Sensor 8 started failing after a TID of 150 krad(SI). Figure 9. Pre- and post-exposure (left and right panels, respectively) measurements of the magnetic ﬁeld inside the zero-Gauss chamber at the University of Michigan for the three sensors used during the University of Michigan test campaign. Figure 9. Pre- and post-exposure (left and right panels, respectively) measurements of the magnetic ﬁeld inside the zero-Gauss chamber at the University of Michigan for the three sensors used during the University of Michigan test campaign. Figure 9. Pre- and post-exposure (left and right panels, respectively) measurements of the magnetic ﬁeld the University of Michigan for the three sensors used during the University of Michigan test campaign. and they were tested afterwards inside the zero-Gauss cham- ber for functioning, as well as performance degradation. The results for this post-exposure tests are presented at the end of this section. and end of each step, data were collected during a short pe- riod of a few seconds without exposure, while the accelerator was switched on and off. The two other magnetometers exposed to the beam from the Varian M9 linear accelerator (sensors 2 and 3) were ex- posed while disconnected (passive test) in a single exposure from 0 to 300 krad(SI). With a rate of 9.6 krad(SI)min−1, the total exposure lasted just over 31 min. Once the exposure was ﬁnished, the sensors were removed from the testing facility 5.2 Test campaign 2: Goddard Space Flight Center A total of six individual sensors were irradiated using the TID irradiator located at the Goddard Space Flight Center Geosci. Instrum. Method. Data Syst., 9, 499–507, 2020 https://doi.org/10.5194/gi-9-499-2020 L. H. Regoli et al.: Radiation tolerance PNI RM3100 L. H. Regoli et al.: Radiation tolerance PNI RM3100 505 Figure 10. Pre- and post-exposure (left and right panels, respectively) measurements of the magnetic ﬁeld inside the zero-Gauss chamber at the University of Michigan for the six sensors used during the GSFC test campaign. Figure 10. Pre- and post-exposure (left and right panels, respectively) measurements of the magnetic ﬁeld inside the zero-Gauss chamber at the University of Michigan for the six sensors used during the GSFC test campaign. (GSFC). The ﬁrst two sensors (sensors 4 and 5) were tested passively but this time in steps of 25 krad(SI), collecting data after each exposure to test their survivability and per- formance degradation. grated circuit (ASIC) that produce random upset events. In- terestingly, even with some failure rates, the sensor survived up to the 300 krad(SI) exposure level and continued deliver- ing relatively steady data. Figure 7 shows the results of these measurements for sensors 4 (top panel) and 5 (bottom panel). Each panel shows a series of 2 min measurements taken right after each 25 krad(SI) exposure step (marked by the vertical line at the right of each measurement). Since the lab environment where the measurements were taken was not controlled in terms of magnetic noise, the absolute values for each magnetometer and each step are irrelevant. However, what can be seen in both plots is the failure rate (seen in the plots as extremely high or low values of the measured ﬁeld) of the sensors with increasing TID. In order to study the effect of dose rate, sensors 6 and 7 were tested passively. First, a single exposure between 0 and 250 krad(SI) at a low rate of 0.26 krad(SI) h−1 was completed. After this, a second exposure between 250 and 300 krad(SI) at a higher rate of 3.17 krad(SI) h−1 was com- pleted. In this case, both sensors survived the total exposure and none of them showed signs of failure or signal degrada- tion in the form of an appreciable change in the noise levels, measured as the standard deviation of the detected signal. Geosci. Instrum. Method. Data Syst., 9, 499–507, 2020 https://doi.org/10.5194/gi-9-499-2020 5.3 Post-exposure evaluation In order to evaluate the performance degradation of the sen- sors with time, all nine sensors exposed during the two cam- paigns reported in this paper were tested in the lab before and after the campaigns. Both sets of measurements were taken for each sensor for a total of 10 min. The measurements were performed inside the zero-Gauss chamber at the Uni- versity of Michigan. The pre-exposure measurements were taken within 2 d of each exposure campaign, while the post- exposure measurements were taken approximately a month after the campaigns. For a Europa lander mission, based on the Europa Sci- ence Deﬁnition Team (SDT) report (Europa Lander Science Deﬁnition Team, 2016, Fig. 6.5), the dose rate expected inside a 7.62 mm aluminum shielding is of approximately 7.7 krad(SI)d−1 or about 320 rad(SI) h−1. The sensors were exposed at a different range of dose rates going from 260 rad(SI) h−1 to over 3 krad(SI) h−1 at the GSFC facility, and up to 9.6 krad(SI)min−1 at the University of Michigan facil- ity, thus mimicking and exceeding the conditions expected during a real mission, not only in terms of TID but also of dose rate. p g Figures 9 and 10 show the magnitude of the magnetic ﬁeld measured during the pre- (left panels) and post-exposure (right panels) tests for each of the nine magnetometers stud- ied. The title of each plot shows the standard deviation of the measured signal, which is a measure of the instrument noise. The ﬁrst thing to notice is that eight of the nine magne- tometers continued to work without failure after the exposure was stopped, with only sensor 5 (data shown in bottom panel of Fig. 7) not showing any data (complete failure). The other sensor that showed some level of failure during the tests (sen- sor 8) went back to normal functioning. These two results mean that there are at least two different failure modes in the sensor. While the failure present during testing for sensor 8 might have been related to charge buildup (something also indicated by the fact that the sensor was able to run, chang- ing the cycle count even while being irradiated), the failure in sensor 5 must have caused a permanent damage in the ASIC, possibly a charge-buildup-induced short circuit. 5.3 Post-exposure evaluation All the tests performed during the reported campaigns were with the whole magnetometer, including not only the sensing coils but also the electronics. In reality, the coils can be separated from the electronics and, since the part that is really susceptible to radiation damage are the electronics, a small enclosure with extra shielding could easily be accom- modated inside the planned vault in order to further shield the instrument. This is possible due to the very low mass (less than 3 g) and volume (about 6.4516 cm2) of the mag- netometer. This would allow the sensing coils to be far from the magnetically noisy environment that is expected to be present inside the vault due to the presence of other instru- ments and spacecraft subsystems. It is also interesting to look at the difference in standard de- viation between the measurements taken prior to and after the exposure campaigns. No appreciable difference is present for any of the working magnetometers, meaning that no degra- dation in the performance occurred. The same behavior can be observed with the noise ﬂoors of the sensors before and after irradiation, given as the value of the power spectrum density at 1 Hz. These are presented in Table 1. Increasing the shielding by 2.54 mm would bring the TID over the 20 d of the mission down to approximately 100 krad(SI) (200 krad(SI) using the design factor of 2), and by adding extra 2.54 mm the total TID would be approxi- mately 75 krad(SI) (150 krad(SI) with the design factor of 2). Given that the sensor that failed at the lowest TID level (sen- sor 5) started failing at 150 krad(SI), this means that an extra shielding of 5 mm for the electronics would guarantee 100 % survivability without any failure for all the sensors used dur- ing the test campaigns. L. H. Regoli et al.: Radiation tolerance PNI RM3100 506 due to the buildup of charges inside the semiconductors aris- ing from the radiation exposure. Although no ground-truth values of the external ﬁeld were available during the tests, the fact that a step is visible in the readings of sensor 9 be- tween the moment when the test was brieﬂy stopped and re- initiated seems to indicate that this slope is in fact produced by charge buildup. after changing the sampling frequency. This gives a success rate, just in terms of functionality (without considering per- formance), of 78 % to 89 %. There was no apparent correla- tion between failure rates and the nature of the irradiation in terms of active vs. passive irradiation or dose rate. p After the end of all the tests, the irradiated sensors were stored and later prepared for a post-irradiation test, similar to that performed at the beginning, during the characterization period. For this stage, the nine sensors were tested inside the zero-Gauss chamber for a period of 5 min. Eight of the nine sensors provided data and for those that continued working no degradation in performance was detected. This means that at least one magnetometer that presented errors during irra- diation (sensor 8) completely recovered, while sensor 5 did not. While the tests presented in this paper do not allow us to speciﬁcally identify which part of the sensor failed, given that all the logic is carried out inside an application-speciﬁc integrated circuit (ASIC), it is expected that this is the com- ponent that is being affected during the exposure. 5.2 Test campaign 2: Goddard Space Flight Center The last two sensors (sensors 8 and 9) were actively tested while being irradiated in a single step from 0 to 300 krad(SI). The results for both sensors are shown in Fig. 8. Shortly after reaching 150 krad(SI), sensor 8 (orange curve) started to fail, while sensor 9 continued working continuously until reach- ing 300 krad(SI). At around 180 krad(SI) some discontinu- ities are visible in the data returned by sensor 9. At this point, the test was temporarily suspended to check on the function- ing of sensor 8. So the visible spikes are noisy data from the times when the test was stopped and re-started. While sensor 4 shows no signs of failure whatsoever throughout the duration of the test, sensor 5 started failing right after reaching the 50 krad(SI) level. The rate of failure (measured as the ratio of invalid measurements to valid mea- surements) changes with time, although not monotonically as it would be expected, with two particularly faulty sets of measurements after reaching 150 krad(SI) and after reach- ing 200 krad(SI) with failure rates of 38 % and 54 %, respec- tively. However, even after these periods with high failure rate, after another exposure bringing the TID to 225 krad(SI), the failure rate went down to zero. This erratic behavior is probably due to damages in the application-speciﬁc inte- There is a clear change in slope that occurs at around 120 krad(SI) where both sensors start to show a steady in- crease in the measured magnitude of the ﬁeld. This steady increase can be due to a real increase in the external ﬁeld or https://doi.org/10.5194/gi-9-499-2020 https://doi.org/10.5194/gi-9-499-2020 Geosci. Instrum. Method. Data Syst., 9, 499–507, 2020 References Badhwar, G. D.: The Radiation Environment in Low-Earth Or- bit, Radiat. Res., 148, S3–S10, https://doi.org/10.2307/3579710, 1997. Cooper, J. F., Johnson, R. E., Mauk, B. H., Garrett, H. B., and Gehrels, N.: Energetic Ion and Electron Irradia- tion of the Icy Galilean Satellites, Icarus, 149, 133–159, https://doi.org/10.1006/icar.2000.6498, 2001. Author contributions. LHR took part in the design of the tests, par- ticipated in the irradiation campaigns, ran all the post-irradiation tests, and prepared the manuscript. MBM took part in the design of the tests. CR took part in the design of the tests, ran all the pre- irradiation tests, and participated in the irradiation campaigns. TAN contributed with his expertise in the radiation environment in the vicinity of Europa and provided calculations to estimate expected TID during a potential mission. CMA and SAP managed the linear accelerator at the University of Michigan. Martin Carts managed the irradiator at the Goddard Space Flight Center facility. All co-authors participated in the revision of the manuscript. Europa Lander Science Deﬁnition Team: Europa Lander Study, Tech. rep., NASA, Pasadena, CA, 2016. Hands, A. D. P., Ryden, K. A., Meredith, N. P., Glauert, S. A., and Horne, R. B.: Radiation Effects on Satellites During Ex- treme Space Weather Events, Space Weather, 16, 1216–1226, https://doi.org/10.1029/2018SW001913, 2018. Khurana, K. K., Kivelson, M. G., Hand, K. P., and Russell, C. T.: Electromagnetic Induction from Europa’s Ocean and Deep Inte- rior, in: Europa, University of Arizona Press, Tucson, Arizona, USA, 2009. Kovtyukh, A. S.: Ion Composition of the Earth’s Radiation Belts in the Range from 100 keV to 100 MeV/nucleon: Fifty Years of Research, Space Sci. Rev., 214, 124, https://doi.org/10.1007/s11214-018-0560-z, 2018. Competing interests. The authors declare that they have no conﬂict of interest. Leuzinger, A. and Taylor, A.: Magneto-Inductive Technology Overview, Tech. rep., PNI Sensor Corporation, Santa Rosa, Cal- ifornia, USA, 2010. Acknowledgements. Tom A. Nordheim acknowledges support from the Jet Propulsion Laboratory, California Institute of Technology, under a contract with NASA. Netzer, R., Avery, K., Kemp, W., Vera, A., Zufelt, B., and Alexan- der, D.: Total Ionizing Dose Effects on Commercial Electronics for Cube Sats in Low Earth Orbits, in: 2014 IEEE Radiation Ef- fects Data Workshop (REDW), Boston, Massachusetts, USA, 1– 7, https://doi.org/10.1109/REDW.2014.7004607, 2014. Financial support. This work was supported by a NASA Instru- ment Concepts for Europa Exploration (80NSSC19K0608) and a NASA Heliophysics Technology and Instrument Development for Science (80NSSC18K1240). L. H. Regoli et al.: Radiation tolerance PNI RM3100 Data availability. All the data obtained during the different tests presented in this paper can be accessed through the University of Michigan’s Deep Blue repository system under the DOI number https://doi.org/10.7302/hs4j-7064 (Regoli, 2020). References 6 Conclusions A total of nine individual RM3100 magnetometers were ir- radiated at different dose rates, all of them up to a TID of at least 300 krad(SI). Of the nine sensors tested, two showed some type of failure during the irradiation proce- dure, although one of them was able to continue working https://doi.org/10.5194/gi-9-499-2020 Geosci. Instrum. Method. Data Syst., 9, 499–507, 2020 507 Geosci. Instrum. Method. Data Syst., 9, 499–507, 2020 References This work was also supported by the Consortium for Veriﬁcation Technology under the Depart- ment of Energy National Nuclear Security Administration (DE- NA0002534) and the US Department of Homeland Security, Coun- tering Weapons of Mass Destruction Ofﬁce, Academic Research Initiative (2016-DN-077-ARI106). Paranicas, C., Carlson, R. W., and Johnson, R. E.: Electron bombardment of Europa, Geophys. Res. Lett., 28, 673–676, https://doi.org/10.1029/2000GL012320, 2001. Prinzie, J., Steyaert, M., and Leroux, P.: Radiation Effects in CMOS Technology, pp. 1–20, Springer International Publishing, Cham, https://doi.org/10.1007/978-3-319-78616-2_1, 2018. Regoli, L. H.: Sensor data for “Radiation tolerance of the PNI RM3100 magnetometer for a Europa lander mission”, https://doi.org/10.7302/hs4j-7064, last access: 22 December 2020. Review statement. This paper was edited by Olivier Witasse and reviewed by two anonymous referees. Regoli, L. H., Moldwin, M. B., Pellioni, M., Bronner, B., Hite, K., Sheinker, A., and Ponder, B. M.: Investigation of a low-cost magneto-inductive magnetometer for space science applications, Geosci. Instrum. Meth., 7, 129–142, https://doi.org/10.5194/gi- 7-129-2018, 2018. Sajid, M., Chechenin, N., Torres, F. S., Hanif, M. N., Gulzari, U. A., Arslan, S., and Khan, E. U.: Analysis of To- tal Ionizing Dose effects for highly scaled CMOS devices in Low Earth Orbit, Nucl. Instrum. Meth. B, 428, 30–37, https://doi.org/10.1016/j.nimb.2018.05.014, 2018. Geosci. Instrum. Method. Data Syst., 9, 499–507, 2020 https://doi.org/10.5194/gi-9-499-2020 https://doi.org/10.5194/gi-9-499-2020
W2005536579.txt	https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0087647&type=printable	en		Fkh1 and Fkh2 Bind Multiple Chromosomal Elements in the S. cerevisiae Genome with Distinct Specificities and Cell Cycle Dynamics	PloS one	2,014	cc-by	9,670	Fkh1 and Fkh2 Bind Multiple Chromosomal Elements in the S. cerevisiae Genome with Distinct Specificities and Cell Cycle Dynamics A. Zachary Ostrow1., Tittu Nellimoottil1., Simon R. V. Knott1¤, Catherine A. Fox2, Simon Tavaré1, Oscar M. Aparicio1* 1 Molecular and Computational Biology Program, University of Southern California, Los Angeles, California, United States of America, 2 Department of Biomolecular Chemistry, University of Wisconsin Medical School, Madison, Wisconsin, United States of America Abstract Forkhead box (FOX) transcription factors regulate a wide variety of cellular functions in higher eukaryotes, including cell cycle control and developmental regulation. In Saccharomyces cerevisiae, Forkhead proteins Fkh1 and Fkh2 perform analogous functions, regulating genes involved in cell cycle control, while also regulating mating-type silencing and switching involved in gamete development. Recently, we revealed a novel role for Fkh1 and Fkh2 in the regulation of replication origin initiation timing, which, like donor preference in mating-type switching, appears to involve long-range chromosomal interactions, suggesting roles for Fkh1 and Fkh2 in chromatin architecture and organization. To elucidate how Fkh1 and Fkh2 regulate their target DNA elements and potentially regulate the spatial organization of the genome, we undertook a genome-wide analysis of Fkh1 and Fkh2 chromatin binding by ChIP-chip using tiling DNA microarrays. Our results confirm and extend previous findings showing that Fkh1 and Fkh2 control the expression of cell cycle-regulated genes. In addition, the data reveal hundreds of novel loci that bind Fkh1 only and exhibit a distinct chromatin structure from loci that bind both Fkh1 and Fkh2. The findings also show that Fkh1 plays the predominant role in the regulation of a subset of replication origins that initiate replication early, and that Fkh1/2 binding to these loci is cell cycle-regulated. Finally, we demonstrate that Fkh1 and Fkh2 bind proximally to a variety of genetic elements, including centromeres and Pol IIItranscribed snoRNAs and tRNAs, greatly expanding their potential repertoire of functional targets, consistent with their recently suggested role in mediating the spatial organization of the genome. Citation: Ostrow AZ, Nellimoottil T, Knott SRV, Fox CA, Tavaré S, et al. (2014) Fkh1 and Fkh2 Bind Multiple Chromosomal Elements in the S. cerevisiae Genome with Distinct Specificities and Cell Cycle Dynamics. PLoS ONE 9(2): e87647. doi:10.1371/journal.pone.0087647 Editor: Yanchang Wang, Florida State University, United States of America Received October 15, 2013; Accepted December 26, 2013; Published February 4, 2014 Copyright: ß 2014 Ostrow et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: This study was supported by National Institutes of Health (NIH) grants 5R01-GM065494 and 3R01-GM065494-S1 (to O.M.A.), P50-HG002790 (to T.N., S.R.V.K., and S.T.), and R01-GM56890 (to C.A.F.). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. * E-mail: oaparici@usc.edu ¤ Current address: Cold Spring Harbor Laboratory, Cold Spring Harbor, New York, United States of America . These authors contributed equally to this work. repression through mechanisms involving recruitment of coactivators or co-repressors, including chromatin modifiers [1]. In Saccharomyces cerevisiae, four proteins contain a Fkh-DBD, including Fkh1, Fkh2, Hcm1, and Fhl1 (reviewed in [2]). Fhl1 has diverged substantially and binds unrelated DNA sequence(s). Hcm1 regulates the expression of a set of genes expressed during S-phase, including Fkh1 and Fkh2 [3]. Fkh1 and Fkh2 share the greatest sequence similarity with each other and recognize similar DNA sequences, which are largely distinct from those recognized by Hcm1 [3–13]. Fkh1 and Fkh2 also share a ForkHeadAssociated (FHA) domain, a phosphothreonine-binding motif, while Fkh2 contains an additional C-terminal domain [6,12]. Fkh1 and Fkh2 regulate a set of ,33 genes, referred to as the CLB2cluster, which are expressed during late S/G2-phase to regulate subsequent mitotic events [11]. Combined deletion of FKH1 and FKH2 severely diminishes expression of CLB2-cluster genes and induces pseudohyphal growth, normally a starvation response, whereas deletion of either Introduction Forkhead Box (Fox) transcription factors comprise a large and diversified family of DNA binding proteins that function in a wide range of processes from yeast to humans, including cell cycle control, development, stress response, and apoptosis (reviewed in [1]). Common to these proteins is the Forkhead DNA-Binding Domain (Fkh-DBD) that binds DNA as a monomer through a conserved helix-turn-helix motif variant, known as a winged-helix. The Fkh-DBD typically recognizes a conserved core DNA sequence (RYMAAYA) with flanking nucleotides providing additional DNA sequence specificity for different Fkh-DBDs. In animals, Fox proteins have been characterized as pioneer transcription factors for their intrinsic ability to bind with sequence specificity to DNA within a compacted, nucleosomal context, and to remodel chromatin for transactivator accessibility and gene activation. Additionally, Fox proteins act in gene activation and PLOS ONE \| www.plosone.org 1 February 2014 \| Volume 9 \| Issue 2 \| e87647 Genome-Wide Binding of Forkhead Proteins gene alone has less severe effects on CLB2-cluster expression and does not cause pseudohyphal growth [6,8,11,14,15]. Thus, Fkh1 and Fkh2 can partially complement loss of each other’s function. However, the phenotypes of the single deletions are different on CLB2-cluster gene expression, with FKH1 deletion being defective in transcriptional repression during G1-phase and FKH2 deletion being defective in timing and peak transcriptional activation levels during late-S/G2 [6,8,11,14]. Both proteins are thought to participate in CLB2-cluster gene repression, however, Fkh2, but not Fkh1, exhibits cooperative DNA-binding interaction with transcription factor Mcm1 that is key to transcriptional activation [5,16]. In addition, Clb5-Cdk1-mediated phosphorylation of the unique C-terminus of Fkh2 promotes interaction with transcription factor Ndd1 that is reinforced by binding of Clb2-Cdk1phosphorylated Ndd1 with the FHA domain of Fkh2, culminating in transcriptional activation [17–19]. NDD1 is essential for CLB2cluster gene activation and its deletion is lethal; however, this lethality is suppressed by deletion of FKH2, but not FKH1, consistent with the idea that Ndd1 overcomes repression by Fkh2 [15]. The interactions of Fkh2 with Mcm1 and Ndd1 have led to a greater focus in previous studies on Fkh2 rather than on Fkh1, and hence, how Fkh1 normally participates in CLB2-cluster regulation is less clear. In contrast, FKH1 has been uniquely implicated in regulation of mating-type switching (reviewed in [20]). Mating-type switching in budding yeast involves repair of a dsDNA break targeted to the MAT locus, resulting in a gene conversion event at MAT. The break is repaired by homologous recombination using one of two homologous donor mating-type alleles (a or a) on either distal arm of the chromosome. Mata cells preferentially (,95%) use HMLa versus HMRa as the donor locus, resulting in a mating-type switch. This preference acts through a Recombination Enhancer (RE) element near HMLa that binds Fkh1. Deletion of the RE or FKH1, but not FKH2, eliminates donor preference, and tethering of the Fkh1-FHA domain in place of the RE is sufficient to restore donor preference [21–23]. Thus, Fkh1 regulates the physical interaction between chromosomally distal DNA sequences. More recently, FKH1 and FKH2 were reported to regulate replication origin timing through a mechanism also involving longrange chromosomal interactions resulting in clustering of earlyfiring origins [24]. Combined deletion of FKH1 and FKH2 alters the replication timing of most of the earliest- and latest-firing replication origins throughout the genome. Early origins that are delayed in fkh1D fkh2D cells (referred to as Fkh-activated origins) are locally enriched for Fkh1 and/or Fkh2 (Fkh1/2) consensus binding sequences, and deletion of these consensus binding sequences near an early origin deregulates its timing. Deletion of FKH1 alone has a more modest effect, with ,50 replication origins (early and late) detectably altered, while deletion of FKH2 alone has no effect. Thus, FKH1 appears to play the primary role in regulating replication origin timing while FKH2 can partially substitute for FKH1 in this function. The basis for this difference remains to be elucidated. Previous studies of Fkh1 and Fkh2 chromatin binding using chromatin immunoprecipitation analyzed by DNA microarray (ChIP-chip) combined with analysis of consensus sequence conservation revealed a few hundred genomic binding loci for each protein [4,7,13]. However, these datasets did not report binding of Fkh1 or Fkh2 at many Fkh-activated origins, despite the recently reported enrichment of consensus binding sequences near these origins, suggesting that the existing data are incomplete. Indeed, the previous ChIP-chip study used early microarray technology with coverage of intergenic regions only, in most cases by a single cDNA probe per intergenic region. In addition, the PLOS ONE \| www.plosone.org previous study analyzed unsynchronized cell populations, which might miss cell cycle-regulated binding. We wished to generate more comprehensive and higher-resolution binding data for Fkh1 and Fkh2, and examine cell cycle regulation. Given the improvement in microarray platforms, instruments and reagents available for ChIP-chip studies, we undertook a new analysis of Fkh1 and Fkh2 binding. Our results indicate highly abundant binding of Fkh1 and Fkh2 throughout the genome with many shared and unique binding loci. Nucleosomal architecture differs at loci unique to Fkh1 versus loci that also bind Fkh2. We also observe cell cycle regulation of binding in the proximity of specific elements such as replication origins, and observe robust association with a variety of other genetic elements not previously reported, including RNA Pol III-transcribed genes. These findings provide an expanded map of Fkh1 and Fkh2 chromatin binding, provide novel insight into origin regulation, and suggest novel roles for Fkh1 and Fkh2 in genome regulation. Results An Expanded Map of Fkh1 and Fkh2 Binding to the S. cerevisiae Genome To assess the genome-wide distribution of Fkh1 and Fkh2, we performed ChIP-chip using several immunologic approaches. First, we used a polyclonal antibody that immunoprecipitates Fkh1 and Fkh2 (herein referred to as ‘‘anti-Fkh1/2 poly’’) and carried out experiments in wild type (WT) and fkh1D fkh2D (control) strains. To validate and supplement these results, we also performed the analysis with anti-MYC monoclonal antibody in WT strains expressing C-terminally epitope-tagged Fkh1 (Fkh1Myc9), Fkh2 (Fkh2-Myc13), and an untagged (control) strain. Experiments were performed in triplicate and analyzed with tiling microarrays covering unique sequences of the S. cerevisiae genome (one ,60 bp oligonucleotide probe every ,80 bp of unique sequence). Data from individual replicates were analyzed to identify significantly enriched regions (p#0.05) having a minimum length of 500 bp (see Methods). Segments of these enriched regions that overlapped by at least 500 bp in at least two replicates were deemed ‘‘bound loci’’, while any such regions overlapping substantially ($ 50% of length) with regions deemed bound in the control strains (fkh1D fkh2D for anti-Fkh1/2 poly and untagged for anti-Myc) were excluded from the set. Plots of the data across chromosome VI show the average from the three replicates of each experiment with bound loci colored (Fig. 1; plots of all chromosomes are presented in Fig. S1). Analysis with anti-Fkh1/2 poly identified 1503 Fkh1 and/or Fkh2 (Fkh1/2)-bound loci that were not detected in the control fkh1D fkh2D cells (Table S1). To investigate the dependence of these bound loci on Fkh1 and Fkh2, we performed ChIP-chip on fkh1D and fkh2D strains with anti-Fkh1/2 poly (Fig. 1, Table S1). We analyzed the resulting binding maps to identify overlapping regions (see Methods), which are indicated in the corresponding intersection of the Venn diagram (Fig. 2A). Focusing on the intersection of the WT with the fkh1D and fkh2D sets, 702 bound loci in WT and fkh2D cells were not bound in fkh1D cells, defining these as Fkh1-dependent loci and suggesting these loci specifically bind Fkh1 (Fig. 2A). 63 sites bound in WT and fkh1D cells were not bound in fkh2D cells, defining these as Fkh2-dependent loci and suggesting that these sites specifically bind Fkh2. The remaining 605 loci are defined as Fkh1/2-dependent loci, suggesting that these sites can bind both Fkh1 and Fkh2, either simultaneously or in the absence of the other. Analysis with anti-Myc identified 1013 Fkh1-Myc- and 700 Fkh2-Myc-bound loci, which were not detected in the untagged 2 February 2014 \| Volume 9 \| Issue 2 \| e87647 Genome-Wide Binding of Forkhead Proteins Figure 1. Genome-wide analysis of Fkh1 and Fkh2 chromatin binding. Plots show averaged ChIP-chip signal (M) from three experimental replicates along chromosome VI, with enriched regions plotted in purple. The antibody and strain genotype used for each experiment are indicated to the left of each panel; the corresponding strains from top to bottom are: CVy43, ZOy1, CVy138, CVy139, ZOy3, ZOy4, and CVy43. Triangles on the bottom panel indicate the position of determined binding sites as described in the text, color-coded by classification. doi:10.1371/journal.pone.0087647.g001 which were detected to bind Fkh1 and Fkh2 (see Methods, Table S2). To examine these sets of Fkh1 and Fkh2 binding loci further, we searched for Fkh1 and Fkh2 consensus binding sequences within the called regions. Using previously reported position-weight matrices of Fkh1 and Fkh2 consensus sequences [9], we determined coordinates for Fkh1 and Fkh2 consensus sequences in the yeast genome (Table S3). The Fkh1 and Fkh2 consensus sequences are very similar to each other, so we searched for the presence of either one, within each set of bound loci. 72%, 45%, and 81% of the Fkh1-only, Fkh2-only, and Fkh1and2 bound loci, respectively, contained at least one Fkh1/2 consensus sequence match (Fig. 2E). strain (Table S1). These sets showed substantial overlap with the Fkh1/2-poly set, with 81% of the Fkh1-Myc and 70% of the Fkh2Myc bound loci intersecting with the Fkh1/2 poly set, while the union of Fkh1-Myc and Fkh2-Myc sets intersected with 61% of the larger Fkh1/2 poly set (Fig. 2B). The Fkh1-Myc and Fkh2-Myc sets also showed substantial overlap with each other, with 452 loci exhibiting binding to both proteins. An additional 530 loci bound Fkh1-Myc specifically, and 221 loci bound Fkh2-Myc specifically. To test these inferred specificities, we examined Fkh1-Myc and Fkh2-Myc binding at Fkh1- and Fkh2-dependent loci determined in the experiments with anti-Fkh1/2 poly. Fkh1-dependent loci showed greater overlap with Fkh1-Myc (58%) than Fkh2-Myc (19%) loci (Fig. 2C), whereas a more balanced proportion of all Fkh1/2 poly loci overlapped with Fkh1-Myc (55%) and Fkh2-Myc (34%) loci (Fig. 2B), consistent with specific or preferential binding of Fkh1 to the set of Fkh1-dependent loci. In contrast, the comparatively small number of Fkh2-dependent loci showed similar overlap with Fkh2-Myc (22%) and Fkh1-Myc (21%) loci (Fig. 2D). Overall, the multiple approaches, use of controls, and good overlap between datasets suggests we have generated robust Fkh1 and Fkh2 binding data. We consolidated the data into three, non-overlapping sets for further analysis, yielding: 828 Fkh1-only loci, which were only detected to bind Fkh1, 285 Fkh2-only loci, which were only detected to bind Fkh2, and 541 Fkh1and2 loci, PLOS ONE \| www.plosone.org Fkh1 and Fkh2 are Associated with Distinct Chromatin Architectures Fkh1 and Fkh2 have been implicated in the regulation of chromatin structure through the recruitment of chromatin modifiers and remodelers [25–27], so we examined the chromatin structure associated with Fkh1 and Fkh2 binding. To achieve basepair resolution necessary to compare binding with nucleosome positioning, we examined Fkh1- and Fkh2-bound loci containing a single Fkh1/2 consensus sequence(s) and aligned these sequences with a published map of nucleosome positions [28]. We plotted the nucleosome density in a 2 kb region surrounding each consensus 3 February 2014 \| Volume 9 \| Issue 2 \| e87647 Genome-Wide Binding of Forkhead Proteins associated with Fkh1-only loci and 21% associated with Fkh1and2 loci. Only 11% of Fkh1-only and no Fkh2-only loci are associated with Fkh-repressed origins, however, 20% of Fkh-repressed origins are associated with Fkh1and2 binding loci. These results are consistent with our previous demonstration that Fkh1/2 consensus binding sequences are enriched near Fkh-activated origins and required for their regulation, whereas Fkh1/2 consensus sequences are depleted near Fkh-repressed origins [24]. However, these results also suggest that Fkh1/2 binding is not sufficient to establish Fkh-activation or that Fkh-unregulated origins are associated with factors that oppose Fkh-origin regulation (see Discussion). The results further suggest that Fkh-repression of origins may in some cases derive from direct binding by Fkh1/2. The predominance of Fkh1 over Fkh2 binding near origins was consistent with our previous finding that fkh1D cells deregulate origin timing whereas fkh2D cells do not (see Introduction). However, our previous study also showed that fkh1D fkh2D cells deregulate many additional origins than fkh1D cells, suggesting a primary role for Fkh1 in origin timing regulation and a secondary role for Fkh2 [24]. Given our previous findings that both Fkh1 and Fkh2 consensus binding sequences are enriched near Fkhactivated origins, the preference for Fkh1 binding indicates the existence of additional determinants of Fkh1 versus Fkh2 binding specificity. Possible candidates for determining Fkh1 versus Fkh2 binding specificity are Mcm1 and Ndd1. In vitro, Mcm1 binds cooperatively with Fkh2, but not Fkh1, to DNA sequences containing closely juxtaposed Fkh1/2 and Mcm1 consensus binding sequences [5,16]. In vivo, Fkh2 recruits Ndd1 to CLB2cluster gene promoters through interactions involving the unique C-terminus of Fkh2 [18,19]. To examine the relationship of Mcm1 and Ndd1 with Fkh1 and Fkh2 binding, we plotted Fkh1-only, Fkh2-only, and Fkh1and2 binding loci for 10 kb regions centered on 79 Mcm1 and 315 Ndd1 binding sites, which were previously reported to bind the respective protein in ChIP experiments and contain a recognizable consensus sequence for the respective protein [4,7]. The heat maps show strong enrichment of Fkh1and2-bound loci proximal to Mcm1 binding sites, with 41% of Mcm1 binding sites overlapping with a Fkh1and2 locus. A few Fkh1-only and almost no Fkh2-only loci were associated with Mcm1 binding sites (Fig. 5B). Ndd1 exhibited a similar pattern of association, with 52% of Ndd1 binding sites proximal to Fkh1and2 loci, 13% of Fkh1-only loci and almost no Fkh2-only loci are proximal to Ndd1 binding (Fig. 5B). Because Fkh-activated replication origins are associated predominantly with Fkh1-only binding loci, this result implies that neither Mcm1 nor Ndd1 associates with most Fkh-activated origins. We tested this directly by searching for Mcm1 and Ndd1 binding sites proximal to replication origins, and for comparison, to CLB2-cluster genes. The results show no instances of Mcm1 binding sites within 500 bp of any of the replication origin classes, whereas 19% of CLB2-cluster genes are within 500 bp of an Mcm1 binding site (Fig. 5C). Like Mcm1, Ndd1 binding sites are also enriched at CLB2-cluster genes, with 22% of CLB2-cluster genes proximal to an Ndd1 site. In contrast to Mcm1, however, Ndd1 binding sites are associated with 10% of Fkh-unregulated origins, representing significant enrichment with this origin class, and with 3% and 4% of Fkh-activated and Fkh-repressed origins, respectively (Fig. 5C). These results suggest that recruitment of Ndd1 to replication origins might counteract Fkh1/2-regulation of origin function (see Discussion). sequence bound by Fkh1-only, Fkh2-only, and Fkh1and2 loci, as separate sets (Fig. 3A). The data show differences in the nucleosome densities associated with these bound loci, with Fkh1-only loci localizing to narrower nucleosome-depleted regions than Fkh2-only and Fkh1and2 loci. We consolidated the data into an average nucleosome density profile for each set and plotted the profiles together for comparison (Fig. 3B). Estimation of the size of the nucleosome-depleted regions indicates a length of ,400 bp at Fkh1and2 loci versus ,275 bp at Fkh1-only loci, a difference of approximately one nucleosome. Fkh1 and Fkh2 Binding at Regulated Genes Next, we examined the Fkh1 and Fkh2 binding data at genes previously reported to be under Fkh1/2 regulation. We generated heat maps of Fkh1-only, Fkh2-only, and Fkh1and2 binding frequency for 10 kb regions centered and oriented on the start codons of 32 CLB2-cluster genes and, for comparison, two additional groups of co-regulated genes: 13 ‘‘CLN2-cluster’’ genes expressed in late G1-phase and 18 ‘‘SIC1-cluster’’ genes expressed in late M-early G1-phase (Fig. 4A) [29]. The heat maps show enrichment of Fkh1 and Fkh2 over the promoter regions of CLB2cluster genes, with 38% of these regions binding both proteins, an additional 21% binding only Fkh2, and an additional 8% binding only Fkh1. In comparison, Fkh1 and Fkh2 were not enriched over the promoters of the CLN2-cluster genes, as expected. Interestingly, some enrichment of Fkh1 and Fkh2 was apparent over SIC1cluster genes, which is consistent with Fkh1 and Fkh2 acting as anti-activators of a subset of SIC1-cluster genes resulting in their activation by Ace2 but not by Swi5 [27]. To examine this more closely, we divided the SIC1 gene cluster into subsets activated by transcription factor Ace2 only, Swi5 only, or either factor, and generated heat maps of Fkh1 and Fkh2 binding frequencies (Fig. 4B). The results show occupancy of Fkh1 and Fkh2 at 38% of Ace2-only genes, but little to no occupancy at other SIC1-cluster genes, confirming that Fkh1 and Fkh2 specifically bind Ace2-only genes [27]. These findings demonstrate that our data recapitulate known features of Fkh1 and Fkh2 binding. Fkh1 and Fkh2 Binding at Replication Origins Fkh1/2 were recently identified as regulators of the initiation timing of replication origins throughout the budding yeast genome [24,30]. In fkh1D fkh2D cells, the initiation of many early origins is delayed, and these origins are locally enriched for Fkh1/2 consensus binding sequences. For a few tested origins, Fkh1/2 binding sequences in cis were shown to be essential for regulation of the proximal origin. However, previous ChIP-chip analysis did not report Fkh1/2 binding at many Fkh-regulated origins [4,7,13], suggesting that Fkh1/2 might act over longer distances to regulate some origins. To examine the Fkh1- and Fkh2-bound loci we have identified in relation to replication origins, we divided origins (termed ARS in yeast) into three groups defined by their change in origin activity in fkh1D fkh2D cells in our previous study: Fkhactivated origins, which showed reduced early firing, Fkhrepressed origins, which showed increased early firing, and Fkhunregulated origins, which showed no significant change in early firing [24]. For each set of origins, we generated heat maps representing the frequency of Fkh1-only, Fkh2-only and Fkh1and2 bound loci for a 10 kb region centered and oriented on the ARS Consensus Sequence (ACS), which is the essential origin-defining sequence (Fig. 5A). Fkh-activated origins are enriched for proximal Fkh1 binding, with 42% of these origins associated with Fkh1-only loci and an additional 27% associated with Fkh1and2 loci, while only 2% are associated with Fkh2-only loci. Fkh-unregulated origins are also enriched for Fkh1, with 31% of these origins PLOS ONE \| www.plosone.org 4 February 2014 \| Volume 9 \| Issue 2 \| e87647 Genome-Wide Binding of Forkhead Proteins Figure 2. Correlation of Fkh1 and Fkh2 binding sites identified in different experiments. A–D) Venn diagrams show overlap between binding regions identified and/or categorized in different experiments. The area of the circle representing each group and the degree of intersection PLOS ONE \| www.plosone.org 5 February 2014 \| Volume 9 \| Issue 2 \| e87647 Genome-Wide Binding of Forkhead Proteins between groups are proportional to the number of binding loci in each group and degree of intersection, respectively. Discrepancies in number of total binding loci corresponding to datasets between the different Venn diagrams result from the method for calculating intersection between the sets (see Methods S1). E) Pie charts show the percentage of binding loci in each group for which the indicated number of matches to Fkh1 and/or Fkh2 consensus binding site(s) were identified. Because the values were rounded to the nearest whole number, the sum of percentages in two of the pie charts exceeds 100%. doi:10.1371/journal.pone.0087647.g002 binding loci near different sets of genomic elements (as defined in Saccharomyces Genome Database) (Fig. 6). Fkh1 and Fkh2 showed remarkable occupancy near several of these elements, with occupancy rates comparable to those at CLB2-cluster genes and Other Genetic Elements Associated with Fkh1 and Fkh2 Binding To determine whether Fkh1 and Fkh2 bind and potentially regulate other genomic elements, we plotted Fkh1 and Fkh2 Figure 3. Distinct nucleosome positioning at Fkh1-only loci versus loci that bind Fkh2. A) The heat maps show density of MNaseprotected sequences (Eaton et al 2010) for 2 kb regions centered on Fkh1/2 consensus sequences within enriched regions that have only a single Fkh1/2 consensus sequence. B) Averaged signal intensities from (A) are plotted. Arrows indicate the positions used to estimate length of nucleosome-depleted regions reported in the Results. doi:10.1371/journal.pone.0087647.g003 PLOS ONE \| www.plosone.org 6 February 2014 \| Volume 9 \| Issue 2 \| e87647 Genome-Wide Binding of Forkhead Proteins Figure 4. Fkh1 and Fkh2 binding with target genes. Heat maps show 10 kb regions of summed binding data for the indicated types of binding loci (Fkh1-only, Fkh2-only, Fkh1and2) surrounding the groups of features indicated above the heat map. The color represents the frequency of enriched binding sequences called for each group of features, amongst total number of features (n) included in each group. A) Fkh1 and Fkh2 enrichment frequencies surrounding CLB2-, CLN2-, and SIC1-cluster genes are plotted as separate groups, with the respective ORFs aligned by their start codons at coordinate 0, with transcription toward positive coordinates to the right. B) Fkh1 and Fkh2 enrichment frequencies surrounding Ace2only-regulated genes, Ace2- or Swi5-regulated genes, and Swi5-only-regulated genes are plotted with the ORFs aligned and oriented as in (A). doi:10.1371/journal.pone.0087647.g004 ‘‘High-S’’, shows higher binding in early S-phase and lower binding in G2/M and late G1. The High-G1 cluster shows higher binding in late G1 and lower binding in G2/M and early S. The High-G2/M cluster shows higher binding in G2/M and lower binding in late G1 and early S, while the Low G1 cluster shows lower binding in late G1 and higher binding in early S and G2/M. To ascertain whether these cell cycle binding patterns are associated with specific functional classes of Fkh1/2 binding loci such as those associated with CLB2-cluster genes or replication origins, we determined the binding patterns of Fkh1/2 binding loci within 500 bp of specific classes of genomic features analyzed above (Fig. 7B). This analysis indicates that Fkh1/2 binding loci proximal to distinct genomic elements exhibit significantly distinct cell cycle patterns of Fkh1/2 binding (see Methods). For example, the High G1 binding pattern, which is the least frequent overall when all binding loci are considered, is the most frequent pattern associated with ARS and X elements, and is also significantly enriched at LTRs, ncRNAs, retrotransposons, tRNAs, and telomeres. The High G1 pattern is also depleted at snoRNAs. The Low G1 pattern, which is infrequent in the overall distribution, is significantly enriched at Introns, 59 UTR Introns, snoRNAs, and tRNAs; this pattern is also depleted at ARSs, X elements and telomeres. The High G2/M pattern is modestly enriched at Introns, LTRs, and tRNAs, and is most notably depleted near ARSs. The High-S pattern, which is most frequent overall, is correspondingly depleted at most of the aforementioned elements that are enriched for another pattern. However, the High-S pattern is not depleted at binding loci proximal to ORFs, telomeres, centromeres, and 59 UTR Introns. Fkh-activated origins. As a group, ORFs show minor enrichment of Fkh1 or Fkh2 relative to flanking sequences. ARSs, telomeres, and subtelomeric X and Y elements, are associated predominantly with Fkh1-only, with 15–20% of these elements proximal to a Fkh1-only locus. In contrast, centromeres, 59 UTR introns, snoRNAs, and tRNAs are more frequently associated with Fkh1and2 binding loci, which are proximal to 40–60% of these elements; these elements show more modest levels of enrichment for Fkh2-only loci (see Table S4 for list of genes with Fkh1/2 enrichment upstream). Fkh1and2 binding loci are also proximal to 20–30% of introns, ncRNAs, retrotransposons, and dispersed long terminal repeats (LTRs). Interestingly, ncRNAs were associated with Fkh1-only binding loci at a similar frequency as with Fkh1and2 loci. These findings suggest that Fkh1 and Fkh2 have unrecognized roles in the regulation of Pol III-transcribed genes, intron processing, and centromere function. Cell Cycle Dynamics of Fkh1 and Fkh2 Binding To gain further insight into the mechanisms that Fkh1/2 use to regulate genes and origins in the cell cycle context, we performed ChIP-chip of Fkh1 and Fkh2 with anti-Fkh1/2 poly in cells synchronized in G2/M with nocodazole, in late G1 with a-factor, and in early S with hydroxyurea. Data from these experiments corresponding to the Fkh1/2 binding loci identified above were subjected to k-means clustering analysis according to the binding patterns of individual loci across the three cell cycle stages (Fig. 7A, Table S2, see Methods). This analysis revealed four distinct clusters that can account for most of the data, with each cluster representing a distinct binding pattern across the cell cycle (Fig. 7A). The largest cluster of ,865 binding loci, named PLOS ONE \| www.plosone.org 7 February 2014 \| Volume 9 \| Issue 2 \| e87647 Genome-Wide Binding of Forkhead Proteins Figure 5. Fkh1 and Fkh2 binding with replication origins. (A–B) Heat maps show 10 kb regions of summed binding data for the indicated types of binding loci (e.g.: Fkh1-only, Fkh2-only, Fkh1and2) surrounding the groups of features indicated above the heat map. The color represents the frequency of enriched binding sequences called for each group of features, amongst total number of features (n) included in each group. A) Fkh1 and Fkh2 enrichment frequencies surrounding Fkh-activated, Fkh-unregulated, and Fkh-repressed origins are plotted with each group aligned and oriented at coordinate 0 by each origin’s ARS consensus sequence (ACS). B) Fkh1 and Fkh2 enrichment frequencies are plotted around Mcm1 and Ndd1 binding sites, which are aligned and oriented by Mcm1 and Ndd1 consensus sequences, respectively. C) The graph shows the percentage of each element class having an Mcm1 or Ndd1 binding site within 500 bp. Asterisks indicate values significantly greater than expected on a random basis at p,0.01 (see Methods S1). doi:10.1371/journal.pone.0087647.g005 lower. With the exception of the very low binding at telomeres in G2/M, binding levels show the greatest differences amongst elements in late G1. To examine Fkh1/2 binding at specific loci, particularly Fkhactivated origins, we plotted the cell cycle ChIP data for a 100 kb region of chromosome III (Fig. 8, see Fig. S2 for plots of all chromosomes). This region includes early-efficient origins ARS305 and ARS306, the silent mating-type locus HML, the Recombination Enhancer (RE) for mating-type donor preference, and BUD3, a Fkh1/2-regulated CLB2-cluster gene, all of which are associated with Fkh1/2 binding. A previous study reported binding of Fkh1 To scrutinize the binding dynamics more directly at these genomic elements, we plotted Fkh1/2 binding profiles at loci specifically proximal to each set of elements (Fig. 7C). The plots show distinct binding patterns associated with different element types. For example, ARSs and telomeres show lower signals in G2/M and sharply higher signals in late G1 and in early S. In contrast, centromeres and 59 UTR Introns showed intermediate signals in G2/M decreasing in early G1 followed by strikingly higher signals in early S. The remaining elements also generally showed higher signals in early S compared with G2/M and early G1, however, the overall degree of fluctuation was somewhat PLOS ONE \| www.plosone.org 8 February 2014 \| Volume 9 \| Issue 2 \| e87647 Genome-Wide Binding of Forkhead Proteins Figure 6. Analysis of Fkh1 and Fkh2 binding proximal to various genetic elements. Fkh1 and Fkh2 enrichment frequencies surrounding different classes of genetic elements are oriented and aligned at coordinate 0 according to the first base position of each element. The maximum frequency reached within 100 bp (500 bp for Y9) of coordinate 0 is indicated above each heat map. The asterisk indicates significant enrichment (p, 0.001) near coordinate 0 (see Methods). doi:10.1371/journal.pone.0087647.g006 notion that Fkh1/2 function through distinct mechanisms to regulate distinct classes of genetic elements. and Fkh2 to CLB2-cluster target genes in late G1- and G2/Msynchronized cells, suggesting that Fkh1/2 bind constitutively to CLB2-cluster target genes [15]. In agreement with these previous reports, Fkh1/2 binding was strongly enriched at BUD3 at all cell cycle times tested. In contrast, previous analysis of Fkh1 binding at the RE showed binding in G2/M but not in late G1 [31]. However, our data show binding of Fkh1/2 at all three cell cycle times, though we note a decreased signal in late G1. At HML-I, Fkh1/2 binding was detected at all cell cycle times, though the signal was decreased in G2/M. Unlike Fkh1/2 binding at all of these loci, however, Fkh1/2 binding at Fkh-activated origins ARS305 and ARS306 showed strong enrichment in G1-phase, but little or no enrichment in S- or G2/M-phases. These findings reveal a new dimension of Fkh1/2 regulation and support the PLOS ONE \| www.plosone.org Discussion An Expanded Map of Fkh1 and Fkh2 Binding to the S. cerevisiae Genome The recent discovery that Fkh1 and Fkh2 regulate replication initiation timing [24,30], along with exciting new mechanistic insight into how Fkh1 regulates donor preference in mating-type switching [21], in addition to their well-established roles as transcription factors, have stoked new interest into these versatile regulators of the genome. A primary goal of this study was to gain a greater understanding of the relationship between Fkh1/2 9 February 2014 \| Volume 9 \| Issue 2 \| e87647 Genome-Wide Binding of Forkhead Proteins Figure 7. Cell cycle analysis of Fkh1 and Fkh2 binding. A) The k-means cluster-gram on the left shows the average signal intensity of individual binding loci across the three experiments, divided into four groups, which was found to account well for the data. The graphs to the right of each cluster show the averaged signal of all sites in the cluster. B) For each class of genetic element indicated, the number of proximal (+/2 500 bp) Fkh1 and Fkh2 binding loci in each of the four clusters in (A) was counted to determine the distribution of these binding sites amongst the four clusters. The colors in the graph correspond to the colors of the four clusters in (A). C) The average signals of Fkh1 and Fkh2 binding loci proximal (+/2500 bp) to the indicated genetic element class was determined and plotted for the three cell cycle points. doi:10.1371/journal.pone.0087647.g007 identified here. To provide further confidence for our sets of identified binding loci, we searched for matches to Fkh1/2 consensus binding sequences. We found that a large majority of Fkh1-only and Fkh1and2 loci contained at least one consensus match within the enriched region, however, only slightly fewer than half of the Fkh2-only loci contained a match. We chose not to use the presence of a consensus sequence as a filter to reduce the number of called loci to avoid imposing this possible bias, as it remains possible that close matches to the consensus sequence were missed, or that Fkh1/2 binds some sequences independently of a consensus sequence. A related possibility is that binding loci lacking a consensus sequence represent sites of ‘‘indirect’’ binding (as coined by Bulyk and colleagues in [32]) where Fkh1/2 do not binding and regulation of replication origins. Elucidating a more complete and dynamic map of Fkh1 and Fkh2 binding loci throughout the genome enabled robust, genome-scale analyses of these binding loci in relation to replication origins, as well as other functional genetic elements. We identified hundreds of novel binding loci for both proteins, including shared and specific loci. Analyses of these data showed binding to known binding loci and targets of regulation such as CLB2-cluster genes, serving to validate these results. These new genomic maps of Fkh1 and Fkh2 binding also provide a valuable resource for future genome-wide and locus-specific studies. Our analysis of Fkh1/2 binding throughout the genome paints a somewhat different picture than previous studies [4,13], with several-fold more binding loci, especially loci binding only Fkh1, PLOS ONE \| www.plosone.org 10 February 2014 \| Volume 9 \| Issue 2 \| e87647 Genome-Wide Binding of Forkhead Proteins Figure 8. G1-specific binding of Fkh1/2 at Fkh-activated origins. Plots show averaged ChIP-chip signal from three experimental replicates along a 100 kb region of the left arm of chromosome III, with enriched regions plotted in purple. The cell cycle arrest for each experiment is indicated to the left of each panel. Boxed loci are discussed in the Results. doi:10.1371/journal.pone.0087647.g008 is also bound by Fkh2, almost none binds only Fkh2. These findings are consistent with the differential effects on individual origin function when either FKH1 or FKH2 is deleted [24]. These results also reinforce previous findings that Fkh1/2 act directly in cis to regulate origin function [24,30]. Nevertheless, we did not detect Fkh1/2 binding near one-third of Fkh-activated origins, leaving open the possibility that the regulation of some origins occurs over a longer distance or indirectly. We also detected Fkh1, and to a lesser degree Fkh1 and Fkh2, binding at a fraction of origins in the Fkh-unregulated group. Some of these may represent bona-fide Fkh-activated origins within this set that did not reach the significance threshold to be classified as Fkh-activated in the previous study. However, another possibility is that additional chromatin regulators binding in the vicinity of these origins oppose Fkh1/2 function, resulting in their Fkh-unregulated phenotype. Indeed, the presence of Ndd1 binding sites near Fkh-unregulated origins may explain why some of these origins are Fkh-unregulated despite many of these origins being bound by Fkh1 and Fkh2. The cell cycle-regulated association of Fkh1/2 with replication origins reported here is an important advance toward a complete understanding of the mechanism of Fkh1/2-regulation of origin timing. Previous studies have indicated that the establishment of the replication-timing program occurs in the M to early G1 period [34,35]. More recent studies indicate that the selective recruitment of replication initiation factors to licensed origins during G1-phase determines early origin firing, and Fkh1/2 are required for this recruitment (reviewed in [36]). This strongly suggests that the G1phase recruitment of Fkh1/2 is essential for initiation factor recruitment and is linked to the origin licensing cycle. This might bind DNA directly but bind chromatin through interaction with other DNA-binding proteins. The much larger number of Fkh1-only versus Fkh2-only loci suggests that Fkh2 binding is more specific or otherwise restricted. This might be explained by additional specificity provided by its interacting partners Mcm1 and/or Ndd1. Hence, it is surprising that Mcm1 and Ndd1 binding sites are located proximal to Fkh1and2 loci but not to Fkh2-only loci. This finding suggests that a different factor is responsible for the exclusive binding of Fkh2 at Fkh2-only loci. Whereas the more extensive nucleosome-depleted regions associated with Fkh1and2 binding loci may be related to Mcm1 and/or Ndd1 binding, this does not explain the similar nucleosomal structure observed at Fkh2-only loci, which are not associated with Mcm1 or Ndd1. Instead, the narrower nucleosome-depleted regions associated with Fkh1-only loci and the larger numbers of loci that bind Fkh1 (i.e., Fkh1-only and Fkh1and2 loci) suggest that Fkh1 is better able to access potential binding sequences in chromatin than Fkh2. A related possibility is that greater abundance of Fkh1 (1720 molecules/cell) versus Fkh2 (656 molecules/cell) results in a more restricted set of binding loci for Fkh2 [33]. Alternatively, Fkh1 and Fkh2 binding may regulate the remodeling of chromatin in distinct ways resulting in the observed differences. This is currently under investigation. Fkh1 and Fkh2 Binding at Replication Origins In contrast to the high occupancy of both Fkh1 and Fkh2 at CLB2-cluster genes, Fkh-activated replication origins are most frequently bound by Fkh1 only, and whereas a minority of origins PLOS ONE \| www.plosone.org 11 February 2014 \| Volume 9 \| Issue 2 \| e87647 Genome-Wide Binding of Forkhead Proteins ZOy4 were constructed by deletion of BAR1 in strains Z1448 and Z1370, respectively, using BamHI-BglII-digested plasmid pDbar1::URA3 with lithium acetate transformation [41]. Cells were grown at 23uC for all experiments and synchronized in late G1, early S, and G2/M by incubation for 3 h in 7.5 nM a-factor (Sigma, T6901), 200 mM hydroxyurea (Sigma, H8627), or 10 mg/mL nocodazole (Sigma, M1404), respectively. ChIP-chip experiments were performed as described previously [42], with the following modifications and reagents: chromatin was sheared to an average size of 300 bp using a Covaris S2 instrument; immunoprecipitations were performed with 9E10 (Covance, MMS150) at 1:100 followed by pull-down with Protein G Dynabeads (Invitrogen, 10004D), or with anti-Fkh1/2 polyclonal antibody [43], which was pre-crosslinked to protein A-Sepharose 4B beads (Invitrogen, 10– 1041), at 1:40 (packed bed volume). Up to 10 ng immunoprecipitated (IP) and total DNA samples were subjected to whole genome amplification (Sigma, WGA2), followed by primer extension labeling with Cy5 and Cy3 end-labeled random nonamers, as described previously [42]. Cy5-labeled IP and Cy3-labeled total DNA samples were combined and hybridized to custom oligonucleotide tiling microarrays (Roche-Nimblegen, 124 k HX12) that tile one ,60 bp oligonucleotide probe per ,80 bp of unique genomic sequence; the Maui hybridization system and reagents (Roche) were used according to the manufacturer’s instructions, and image capture was performed using an Axon 4100A scanner. involve interactions with protein(s) that license origins in early G1phase such as Mini-Chromosome Maintenance proteins, and/or might involve regulation by CDK or DDK activities. Experiments are in progress to determine the mechanism of cell cycle-regulated binding of Fkh1/2 to replication origins. Novel Genetic Elements Associated with Fkh1 and Fkh2 Binding A novel finding of this study is the association of Fkh1/2 with a large number of functional genetic elements, including centromeres, telomeres, transposable elements, introns and RNA Pol IIItranscribed genes, suggesting a possible role for Fkh1/2 in regulating the function of these elements. Enrichment of Fkh1 upstream of tRNA genes has been previously reported [37]. The high Fkh1/2 occupancy at tRNAs and snoRNAs, which are transcribed by RNA Pol III is particularly intriguing given the known role of Fkh1/2 as a regulator of some Pol II-transcribed genes. Furthermore, Fkh1/2 are thought to regulate origin timing and mating-type donor preference by mediating long-range intraand/or inter-chromosomal interactions (reviewed in [20,36]), while highly expressed tRNAs aggregate into clusters surrounding the nucleolus (reviewed in [38]). It will be interesting to determine whether Fkh1/2 regulate tRNA clustering or expression. Similarly, the association of Fkh1/2 with transposable elements, centromeres and telomeres, all suggest a function in chromosomal organization. The Fkh1/2 association with one or more of these element classes may reflect co-localization of two or more element classes where a single class is the functional target of Fkh1/2. A possible case is the enrichment of Fkh1 with telomeres and subtelomeric X and Y9 elements, which are associated with a high density of ARS elements [39,40]. Thus, the binding of Fkh1 near subtelomeric origins likely explains their observed proximity to subtelomeric elements and telomeres. Although telomeres and subtelomeres are late-replicating, many of these regions become even later replicating in fkh1D fkh2D cells, consistent with Fkh1/2 regulating subtelomeric origins [24]. tRNAs and retrotransposons also colocalize with replication origins more frequently than expected at random [40]; however, this relationship probably does not explain the Fkh1/2 association with these elements because tRNAs and retrotransposons are primarily associated with Fkh1and2 binding loci whereas origins are primarily associated with Fkh1-only loci. Nevertheless, yeast transposable elements frequently co-localize with tRNAs and Pol III-transcribed genes so the association seen with these various elements may result from this co-localization. Given the much higher occupancy of Fkh1/2 at tRNAs and snoRNAs and the larger number of these elements compared with retrotransposons, we think it is more likely that the association with retrotransposons reflects functional Fkh1/2 binding near tRNAs and snoRNAs, rather than the converse. Whereas further studies will be required to elucidate fully the role(s) of Fkh1 and Fkh2 at these various elements, these remarkably robust associations strongly suggest that Fkh1 and Fkh2 have more global functions than previously appreciated. It remains to be seen whether the association of Fkh1 and Fkh2 with a broad array of genetic elements can be explained by a common mechanism involving higher-order chromatin organization. Microarray Data Analysis and Peak Calling We used RINGO package (http://www.biomedcentral.com/ 1471-2105/8/221) in BIOCONDUCTOR suite to perform the microarray normalization. The ChIP peaks were calculated with a distCutOff value of 5000. The upperBoundNull method with a p-value of 0.05 was used to calculate the threshold for calculating the enriched regions. M is the log2 ratio of bound to total signal. From each microarray experiment, we obtained a set of enriched regions defined by chromosome number, start, stop, maxLevel, and score of each peak. For experimental triplicates, all nucleotides were examined to identify those enriched in at least two of the replicates. Nucleotides pertaining to contiguous stretches of enriched nucleotides $500 bp were identified. Finally, these enriched regions were eliminated if 50% or more of their nucleotides overlaps with enriched nucleotides in the control datasets. The remaining enriched regions are deemed ‘‘bound’’. Analysis of Intersection between Datasets Bound regions from different datasets that overlap by $100 bp were deemed to intersect and were enumerated within the intersecting region of the Venn diagrams. Details on set functions and construction of the Venn diagrams are described in Methods S1. Calling Fkh1-only, Fkh2-only, and Fkh1and2 Sets Fkh1-only loci were defined as the union of Fkh1-dependent and Fkh1-Myc loci followed by subtraction of Fkh2-Myc loci. Fkh2-only loci were defined as the union of Fkh2-dependent and Fkh2-Myc loci followed by subtraction of Fkh1-Myc loci. Fkh1and2 loci were defined as all loci with subtraction of Fkh1only and Fkh2-only loci. For union of sets, all nucleotides in the sets being combined were included in the union. For subtraction of a set B from a set A, enriched regions in set A were entirely eliminated from set A if they overlapped by $100 bp with enriched region(s) from set B. For smaller overlaps, only the overlapping nucleotides were eliminated from set A. The remaining enriched sequences of set A comprise the subtracted set. Methods Yeast Strains and Methods All strains (see Table S5) are congenic with the W303 background, including FKH1 and FKH2 MYC-tagged strains, Z1448 and Z1370 respectively, from the Young lab [4]. ZOy3 and PLOS ONE \| www.plosone.org 12 February 2014 \| Volume 9 \| Issue 2 \| e87647 Genome-Wide Binding of Forkhead Proteins Analysis of Fkh1/2 Enrichment at Genetic Elements Supporting Information Heat maps of Fkh1/2 binding proximal to features of interest were constructed from two-dimensional binary matrices. Each row of the matrix represents nucleotides on either side of one instance of the chromosomal element of interest; there are as many rows as there are instances of the element class under analysis. The central column (plotted as coordinate 0) represents a central reference nucleotide for each instance of that chromosomal element, and on either side are the surrounding nucleotides, with one nucleotide per column. A matrix value of 1 indicates that the nucleotide position was called as enriched in the ChIP analysis, whereas a value of 0 indicates that the nucleotide was not enriched. The average value for each column was plotted as the binding frequency. Values given in the text and figure are the maximum binding frequency within 100 bp (500 bp for Y9) of coordinate 0. Coordinates for all genetic elements were acquired from SGD, with the exception that coordinates for ACSs in Figure 5A were taken from [28]. To test the significance of enrichment of Fkh1/2 binding in the vicinity of genetic elements, we performed simulations to model the null distribution and then tested whether the actual distribution was significantly higher than the null distribution. This method was not applicable to X, Y9, telomeres, or retrotransposons because of the lack of unique sequences downstream of these elements. Details of the simulation and statistical tests are described in Methods S1. Figure S1 Genome-wide analysis of Fkh1 and Fkh2 chromatin binding. Plots show averaged ChIP-chip signal from three experimental replicates along each chromosome, with enriched regions plotted in purple. The antibody and strain genotype used for each experiment are indicated to the left of each panel. The corresponding strains from top to bottom are: CVy43, ZOy1, CVy138, CVy139, ZOy3, ZOy4, and CVy43. Triangles on the bottom panel indicate the position of determined binding sites as described in the text, color-coded by classification. (RAR) Figure S2 Cell cycle binding of Fkh1/2 genome-wide. Plots show averaged ChIP-chip signal from three experimental replicates along each chromosome, with enriched regions plotted in purple. The cell cycle arrest for each experiment is indicated to the left of each panel. (RAR) Table S1 Enriched regions for each experiment performed in triplicate. Each row gives genomic coordinates of enriched regions from data combined from ChIP-chip experiments performed in triplicate. Strain and antibody used are indicated in the key. (RAR) Table S2 Genomic coordinates of Fkh1 and Fkh2 binding sites organized by class. Enriched regions indicated for Fkh1-only, Fkh2-only, and Fkh1and2. (CSV) Cell Cycle Analysis of Binding Each enriched region identified by RINGO is associated with a total score, which is a measure of enrichment across the entire region. We normalized the total score to a score per nucleotide by dividing the total score by the length of the enriched region. Next, we calculated a union set of all the enriched regions across the three cell cycle experiments (G2/M; late G1; early S), which included all nucleotides within enriched regions in any of the sets. The score associated with each enriched region in the union set was calculated as the total of the per nucleotide score of each nucleotide that belongs to that enriched region. Hence we ended up with three tracks of enriched regions with the same chromosomal coordinates, but different total scores. These three sets of total scores were subjected to k-means clustering with k = 4, and distance measure being Pearson’s correlation coefficient. Fkh1/2 binding loci that were not enriched in any of the three cell cycle experiments were excluded from this analysis. We also assigned subsets of these union sets to genetic elements from SGD annotation file, if the enriched region overlapped with or was ,100 bp from a boundary of the feature. Then we determined the class to which each feature-associated enriched region belonged and constructed the stacked bar graphs of their distribution for each genetic element. A chi-squared test was applied to the corresponding ratios of each set of Fkh1/2 binding loci associated with a particular genetic element to test whether it was significantly different from the null distribution after Bonferroni correction. The null distribution was chosen as the membership ratios of all Fkh1/2 binding loci in the four cell cycle clusters. The distributions at all individual classes of genetic elements were found to be significantly different from the null. Table S3 Genomic coordinates for Fkh1 and Fkh2 consensus sites. Each row gives coordinates of a single Fkh1 or Fkh2 consensus site as indicated. (CSV) Table S4 Genes with upstream Fkh1/2 enrichment. Genes are listed for which the upstream region overlaps with a Fkh1 or Fkh2 enriched region. 500 bp regions upstream of transcription start sites for ORFs and snoRNA and tRNA genes acquired from SGD were analyzed for overlap with Fkh1 or Fkh2 enriched regions. (XLSX) Table S5 Strain information. Name, genotype and source of each strain used in this study. (XLSX) Methods S1 Additional details of methods are given along with schematics of methods used to define intersections, unions, and subtractions, as well as methods and formulas used to calculate Venn diagrams. (DOC) Acknowledgments We thank Michelle Arbeitman for critical reading of the manuscript and Rick Young for strains. Author Contributions Conceived and designed the experiments: OA CAF TN AZO. Performed the experiments: AZO. Analyzed the data: OA SRVK TN AZO ST. Contributed reagents/materials/analysis tools: CAF. Wrote the paper: OA CAF AZO TN. Data Accession Microarray pair files are available at GEO, accession number: GSE42567. PLOS ONE \| www.plosone.org 13 February 2014 \| Volume 9 \| Issue 2 \| e87647 Genome-Wide Binding of Forkhead Proteins References 21. Li J, Coic E, Lee K, Lee CS, Kim JA, et al. (2012) Regulation of budding yeast mating-type switching donor preference by the FHA domain of Fkh1. PLoS Genet 8: e1002630. 22. Sun K, Coic E, Zhou Z, Durrens P, Haber JE (2002) Saccharomyces forkhead protein Fkh1 regulates donor preference during mating-type switching through the recombination enhancer. Genes Dev 16: 2085–2096. 23. Wu X, Haber JE (1996) A 700 bp cis-acting region controls mating-type dependent recombination along the entire left arm of yeast chromosome III. Cell 87: 277–285. 24. Knott SR, Peace JM, Ostrow AZ, Gan Y, Rex AE, et al. (2012) Forkhead transcription factors establish origin timing and long-range clustering in S. cerevisiae. Cell 148: 99–111. 25. Sherriff JA, Kent NA, Mellor J (2007) The Isw2 chromatin-remodeling ATPase cooperates with the Fkh2 transcription factor to repress transcription of the Btype cyclin gene CLB2. Mol Cell Biol 27: 2848–2860. 26. Veis J, Klug H, Koranda M, Ammerer G (2007) Activation of the G2/Mspecific gene CLB2 requires multiple cell cycle signals. Mol Cell Biol 27: 8364– 8373. 27. Voth WP, Yu Y, Takahata S, Kretschmann KL, Lieb JD, et al. (2007) Forkhead proteins control the outcome of transcription factor binding by antiactivation. EMBO J 26: 4324–4334. 28. Eaton ML, Galani K, Kang S, Bell SP, MacAlpine DM (2010) Conserved nucleosome positioning defines replication origins. Genes Dev 24: 748–753. 29. Spellman PT, Sherlock G, Zhang MQ, Iyer VR, Anders K, et al. (1998) Comprehensive identification of cell cycle-regulated genes of the yeast Saccharomyces cerevisiae by microarray hybridization. Mol Biol Cell 9: 3273– 3297. 30. Looke M, Kristjuhan K, Varv S, Kristjuhan A (2012) Chromatin-dependent and -independent regulation of DNA replication origin activation in budding yeast. EMBO Rep 14: 191–198. 31. Coic E, Richard GF, Haber JE (2006) Saccharomyces cerevisiae donor preference during mating-type switching is dependent on chromosome architecture and organization. Genetics 173: 1197–1206. 32. Gordan R, Hartemink AJ, Bulyk ML (2009) Distinguishing direct versus indirect transcription factor-DNA interactions. Genome Res 19: 2090–2100. 33. Ghaemmaghami S, Huh WK, Bower K, Howson RW, Belle A, et al. (2003) Global analysis of protein expression in yeast. Nature 425: 737–741. 34. Dimitrova DS, Prokhorova TA, Blow JJ, Todorov IT, Gilbert DM (2002) Mammalian nuclei become licensed for DNA replication during late telophase. J Cell Sci 115: 51–59. 35. Raghuraman MK, Brewer BJ, Fangman WL (1997) Cell cycle-dependent establishment of a late replication program. Science 276: 806–809. 36. Aparicio OM (2013) Location, location, location: it’s all in the timing for replication origins. Genes Dev 27: 117–128. 37. Venters BJ, Wachi S, Mavrich TN, Andersen BE, Jena P, et al. (2011) A comprehensive genomic binding map of gene and chromatin regulatory proteins in Saccharomyces. Mol Cell 41: 480–492. 38. Hopper AK, Pai DA, Engelke DR (2010) Cellular dynamics of tRNAs and their genes. FEBS Lett 584: 310–317. 39. Chan CS, Tye BK (1983) Organization of DNA sequences and replication origins at yeast telomeres. Cell 33: 563–573. 40. Wyrick JJ, Aparicio JG, Chen T, Barnett JD, Jennings EG, et al. (2001) Genome-wide distribution of ORC and MCM proteins in S. cerevisiae: highresolution mapping of replication origins. Science 294: 2357–2360. 41. Gietz RD, Schiestl RH (2007) High-efficiency yeast transformation using the LiAc/SS carrier DNA/PEG method. Nat Protoc 2: 31–34. 42. Viggiani CJ, Aparicio JG, Aparicio OM (2009) ChIP-chip to analyze the binding of replication proteins to chromatin using oligonucleotide DNA microarrays. Methods Mol Biol 521: 255–278. 43. Casey L, Patterson EE, Muller U, Fox CA (2008) Conversion of a replication origin to a silencer through a pathway shared by a Forkhead transcription factor and an S phase cyclin. Mol Biol Cell 19: 608–622. 1. Lalmansingh AS, Karmakar S, Jin Y, Nagaich AK (2012) Multiple modes of chromatin remodeling by Forkhead box proteins. Biochim Biophys Acta 1819: 707–715. 2. Murakami H, Aiba H, Nakanishi M, Murakami-Tonami Y (2010) Regulation of yeast forkhead transcription factors and FoxM1 by cyclin-dependent and pololike kinases. Cell Cycle 9: 3233–3242. 3. Pramila T, Wu W, Miles S, Noble WS, Breeden LL (2006) The Forkhead transcription factor Hcm1 regulates chromosome segregation genes and fills the S-phase gap in the transcriptional circuitry of the cell cycle. Genes Dev 20: 2266–2278. 4. Harbison CT, Gordon DB, Lee TI, Rinaldi NJ, Macisaac KD, et al. (2004) Transcriptional regulatory code of a eukaryotic genome. Nature 431: 99–104. 5. Hollenhorst PC, Pietz G, Fox CA (2001) Mechanisms controlling differential promoter-occupancy by the yeast forkhead proteins Fkh1p and Fkh2p: implications for regulating the cell cycle and differentiation. Genes Dev 15: 2445–2456. 6. Kumar R, Reynolds DM, Shevchenko A, Goldstone SD, Dalton S (2000) Forkhead transcription factors, Fkh1p and Fkh2p, collaborate with Mcm1p to control transcription required for M-phase. Curr Biol 10: 896–906. 7. MacIsaac KD, Wang T, Gordon DB, Gifford DK, Stormo GD, et al. (2006) An improved map of conserved regulatory sites for Saccharomyces cerevisiae. BMC Bioinformatics 7: 113. 8. Pic A, Lim FL, Ross SJ, Veal EA, Johnson AL, et al. (2000) The forkhead protein Fkh2 is a component of the yeast cell cycle transcription factor SFF. EMBO J 19: 3750–3761. 9. Zhu C, Byers KJ, McCord RP, Shi Z, Berger MF, et al. (2009) High-resolution DNA-binding specificity analysis of yeast transcription factors. Genome Res 19: 556–566. 10. Zhu G, Davis TN (1998) The fork head transcription factor Hcm1p participates in the regulation of SPC110, which encodes the calmodulin-binding protein in the yeast spindle pole body. Biochim Biophys Acta 1448: 236–244. 11. Zhu G, Spellman PT, Volpe T, Brown PO, Botstein D, et al. (2000) Two yeast forkhead genes regulate the cell cycle and pseudohyphal growth. Nature 406: 90–94. 12. Kaufmann E, Knochel W (1996) Five years on the wings of fork head. Mech Dev 57: 3–20. 13. Simon I, Barnett J, Hannett N, Harbison CT, Rinaldi NJ, et al. (2001) Serial regulation of transcriptional regulators in the yeast cell cycle. Cell 106: 697–708. 14. Hollenhorst PC, Bose ME, Mielke MR, Muller U, Fox CA (2000) Forkhead genes in transcriptional silencing, cell morphology and the cell cycle. Overlapping and distinct functions for FKH1 and FKH2 in Saccharomyces cerevisiae. Genetics 154: 1533–1548. 15. Koranda M, Schleiffer A, Endler L, Ammerer G (2000) Forkhead-like transcription factors recruit Ndd1 to the chromatin of G2/M-specific promoters. Nature 406: 94–98. 16. Boros J, Lim FL, Darieva Z, Pic-Taylor A, Harman R, et al. (2003) Molecular determinants of the cell-cycle regulated Mcm1p-Fkh2p transcription factor complex. Nucleic Acids Res 31: 2279–2288. 17. Darieva Z, Pic-Taylor A, Boros J, Spanos A, Geymonat M, et al. (2003) Cell cycle-regulated transcription through the FHA domain of Fkh2p and the coactivator Ndd1p. Curr Biol 13: 1740–1745. 18. Pic-Taylor A, Darieva Z, Morgan BA, Sharrocks AD (2004) Regulation of cell cycle-specific gene expression through cyclin-dependent kinase-mediated phosphorylation of the forkhead transcription factor Fkh2p. Mol Cell Biol 24: 10036–10046. 19. Reynolds D, Shi BJ, McLean C, Katsis F, Kemp B, et al. (2003) Recruitment of Thr 319-phosphorylated Ndd1p to the FHA domain of Fkh2p requires Clb kinase activity: a mechanism for CLB cluster gene activation. Genes Dev 17: 1789–1802. 20. Haber JE (2012) Mating-type genes and MAT switching in Saccharomyces cerevisiae. Genetics 191: 33–64. PLOS ONE \| www.plosone.org 14 February 2014 \| Volume 9 \| Issue 2 \| e87647
https://openalex.org/W2037801853	https://ccsenet.org/journal/index.php/ies/article/download/30783/18122	English	null	Business English as a Lingua Franca	International education studies	2,013	cc-by	5,170	International Education Studies; Vol. 6, No. 10; 2013 ISSN 1913-9020 E-ISSN 1913-9039 Published by Canadian Center of Science and Education International Education Studies; Vol. 6, No. 10; 2013 ISSN 1913-9020 E-ISSN 1913-9039 Published by Canadian Center of Science and Education International Education Studies; Vol. 6, No. 10; 2013 ISSN 1913-9020 E-ISSN 1913-9039 Abstract This paper examines BELF (Business English as a Lingua Franca) teaching and researching in China. A literature review is conducted using the China National Knowledge Infrastructure (CNKI) database. This survey includes a cursory literature search on BELF and a thorough literature study of 12 Chinese major academic journals. From the data collected in this study several findings emerge. First, BELF teaching and research has been on the rise since the founding of the People’s Republic of China; and it appears that there will continue to be a growth in BELF literature and research. Secondly, BELF literature during this period has covered a wide range of areas, such as teaching methodology, mode and learning strategies, syllabus design, course assessment and reform, translation studies, systemic functional based studies, discourse studies, as well as other areas. Thirdly, some drawbacks are also exposed. For example, the paucity of empirical studies indicates a strong need to improve the quantity of BELF research literature. Keywords: Business English (BE), Lingua Franca (LF), BELF teaching and research Keywords: Business English (BE), Lingua Franca (LF), BELF teaching and research Business English as a Lingua Franca (BELF) Yan Wu1 1 College of Foreign Languages, Hebei United University, Tangshan, Hebei, China Correspondence: Yan Wu, College of Foreign Languages, Hebei United University, No. 46 West Xinhua Road, Tangshan, Hebei, China. Tel: 86-139-3155-3695. E-mail: helenawuyan@126.com Yan Wu1 1 College of Foreign Languages, Hebei United University, Tangshan, Hebei, China Correspondence: Yan Wu, College of Foreign Languages, Hebei United University, No. 46 West Xinhua Road, Tangshan, Hebei, China. Tel: 86-139-3155-3695. E-mail: helenawuyan@126.com Received: August 7, 2013 Accepted: September 22, 2013 Online Published: Septemb doi:10.5539/ies.v6n10p130 URL: http://dx.doi.org/10.5539/ies.v6n10p130 1. The Definition of BELF BELF (Business English as a Lingua Franca or English Lingua Franca for Business Purposes) is employed as a substitute for BE (Business English) because of the wider application of Business English in international business communications among nonnative speakers. In addition to it, the focus of teaching and researching in BELF is certainly wider than that in BE. BELF covers both the teaching and researching of Business English with the purposes of facilitating both business communications between speakers of other languages and NS (native English speakers), and between NNS (nonnative speakers). Although the core definition of BELF has much in common with that of BE, the former differs from the latter in some aspects. There is a brief introduction to the definition of BE by domestic and overseas scholars. Mark Ellis and Christine Johnson in their book “Teaching Business English” (1994, pp. 7-13) have defined the characteristics of BE from five aspects: Mark Ellis and Christine Johnson in their book “Teaching Business English” (1994, pp. 7-13) have defined the characteristics of BE from five aspects: 1) “Much of the language needed by business people (apart from social language) will be transactional.” ocial contacts are often highly ritualized.” 2) “Social contacts are often highly ritualized.” 3) Clear information should be conveyed within a short time. 3) Clear information should be conveyed within a short time. 4) The language used in business “will be neither as rich in vocabulary and expression nor as culture-bound, as that used by native speakers, but will be based on a core of the most useful and basic structures and vocabulary.” 5) Business English courses differ greatly in some aspects like needs analysis, assessment of level, syllabus, course objectives, etc. Jiang and Guo (1997, pp. 43-50) have explored BE in comparison with General English (GE). The major differences between them lie in the goal of teaching and materials in use. Lin (2004) defines international business English from four aspects: Lin (2004) defines international business English from four aspects: a) From the perspective of ESP, it can be defined as an attitude or viewpoint in dealing with language in international business communications, rather than an innovative or creative work. 130 Vol. 6, No. 10; 2013 International Education Studies www.ccsenet.org/ies b) It is an international language, which draws distinction in accordance with native English or Standard English. 1. The Definition of BELF c) As a major well-developed branch of ESP, its teaching falls into EGBP (English for General Business Purposes) and ESBP (English for Specific Business Purposes). d) It should take learners’ needs into account and show respect for individual differences. As is shown above, there is not a universally acknowledged definition of BE. The multi-disciplinary characteristics of BE may contribute to the diversity of its definitions. While there is some overlap with BE, BELF covers a wide variety of English usage in many industries, such as finance, economics, business law, foreign trade, tourism, advertising, etc. Taking this into account, BELF falls into two categories, namely, ELFGBP (English Lingua Franca for General Business Purposes) and ELFSBP (English Lingua Franca for Specific Business Purposes). The former is defined in accordance with the core features of Business English. While the latter is definedin accordance with the diversities in various levels of language as well as the specific context English language in use. 2. BELF Teaching and Research in China -A Historial Review Business English (BE) has been taught in China ever since the founding of the People’s Republic of China. The development of Business English can be observed in teaching goals, courses offered, and materials used. 1n the 1950s, Business English teaching was aimed at producing translators and interpreters in the field of foreign trade. Only a few colleges offered Business English Translation programs. In the 1980s, it was aimed at teaching students both the English language and a way of conducting business. Content-based courses were offered in English, such as marketing, management, and international co-operation. It was believed that by providing students with an English language environment, they could learn both language and business knowledge naturally. During the 1990s, Business English underwent a considerable development. And business English teaching was considered a way to cultivate multi-talents. This transformation led to colleges and universities at different levels to start offering a variety of business English courses. Since the beginning of 21st century, although Business English remains a way of cultivating multi-talents, the literature research (see Table 1) in this field has shown that many colleges, universities, scholars and teachers have taken initiative to explore BE in terms of teaching methods and modes, teaching materials, etc., and they have endeavored to relate theories to practice which is totally different from intuition-oriented (See Table 3 & 4). 3.1 Research Methods Following is a literature research on journal papers already published in China. All the literature is from Chinese major academic journals in the field of foreign languages, including 12 journals: Foreign Language Teaching and Research, Foreign Language World, Foreign Language and Literature, Foreign Languages Research, Foreign Language Learning Theory and Practice, Modern Foreign Languages, Foreign Languages and Their Teaching, Foreign Language Education, Journal of PLA University of Foreign Languages, Foreign Language Research, Shandong Foreign Language Teaching Journal, Journal of Foreign Language. All the data are provided by CNKI (China National Knowledge Infrastructure) . 3.2 Data Analysis The possible reasons m Why is there a sharp increase in the amount of literature during the last decade? The possible reasons may be: 1) After implementing the China Open Door Policies in 1978, especially after China’s successfully entry into the World Trade Organization (WTO), there’s been an increasing demand for proficient English language users with business expertise. 2) In the 1980s, a lot of theories were introduced to China, such as ESP and Systemic Functional Linguistics. These have profound influences on BELF researching and teaching methods. 3) During the 1990s, the Ministry of Education recognized that Business English teaching was a method of cultivating multi-talented individuals. Since this point, colleges and universities at various levels started offering Business English courses. 3) During the 1990s, the Ministry of Education recognized that Business English teaching was a method of cultivating multi-talented individuals. Since this point, colleges and universities at various levels started offering Business English courses. 4) The globalization trend has also contributed to the rapid growth in BELF literature. In order to develop a thorough understanding of BELF teaching and researching in China, a literature search was conducted of 12 major academic journals in the field of foreign languages. These journals were selected because they are highly-regarded in China. In addition, the papers included in these journals are professional and of higher quality. In order to develop a thorough understanding of BELF teaching and researching in China, a literature search was conducted of 12 major academic journals in the field of foreign languages. These journals were selected because they are highly-regarded in China. In addition, the papers included in these journals are professional and of higher quality. The results show that there are 87 papers related to BELF teaching and research published since 1981. The distribution of which is shown in Table 2. Only eleven journals are included in Table 2 because no literature in BELF was found in the Foreign Language Teaching and Research journal. 132 Table 2. 3.2 Data Analysis By searching the CNKI (China National Knowledge Infrastructure) database for key words: Business English, or its Chinese correspondence, 商务英语，商贸英语，经贸英语，外贸英语， the data collected shows that during the period of Jan. 1st, 1949 to Jan. 1st, 2012, we can see that there are 8672 articles in this field. Only one of the articles is from before Jan. 1st, 1978 and 35 are during the period of Jan. 2nd, 1978 to Jan. 1st, 1990. However, the amount of articles rose from 623 (between Jan. 2nd, 1990 and Jan. 1st, 2000) to 8013 (between Jan. 2nd, 2000 and Jan. 1st, 2012). Table 1 shows this sharp increase in the amount of BELF articles during the last decade. Table 1. Brief Overview on BELF Teaching and Researching in China since the Founding of the PRC Period of time Before Jan. 1st, 1978 Jan. 2nd，1978 – Jan. 1st, 1990 Jan. 2nd， 1990 --- Jan. 1st, 2000 Jan. 2nd，2000 – Jan. 1st，2012 Total Search results 1 35 623 8013 8672 e 1. Brief Overview on BELF Teaching and Researching in China since the Founding of the PRC 131 International Education Studies Vol. 6, No. 10; 2013 www.ccsenet.org/ies Though BELF teaching started shortly after the founding of the PRC, Table 1 illustrates a couple of facts. As is known, BE research dates back to the 1980s overseas. Therefore, BELF research had developed slowly for 10 years in the PRC. Secondly, since 1990s, BELF teaching and researching has been developing from intuition-oriented towards rational or theory-guided practice. Scholars started thinking and conducting research on BELF in a more scientific way. Thirdly, BELF literature has increased since Jan. 1st, 1990, and the last decade has witnessed a considerable increase. Though BELF teaching started shortly after the founding of the PRC, Table 1 illustrates a couple of facts. As is known, BE research dates back to the 1980s overseas. Therefore, BELF research had developed slowly for 10 years in the PRC. Secondly, since 1990s, BELF teaching and researching has been developing from intuition-oriented towards rational or theory-guided practice. Scholars started thinking and conducting research on BELF in a more scientific way. Thirdly, BELF literature has increased since Jan. 1st, 1990, and the last decade has witnessed a considerable increase. Why is there a sharp increase in the amount of literature during the last decade? 3.2 Data Analysis However, just a small number of journal articles in this field were published each year. Since 2006 there has been a steady increase in the number of articles published in this field. In 2011, it reached to a peak when 11 published papers appeared in these prestigious journals. Further, Table 2 shows that at least 6 papers were published each year from 2006 to 2011. Based on the data collected, it can be concluded that the quantity of journal papers in BELF have been improved over the past several decades. Nevertheless, among these 87 papers identified, only 4 papers were empirical studies, and accounted for 4.60% (see Table 3). This fact reflects a limitation in BELF studies. From Table 2, it can be clearly seen that the literature on BELF has been on rise ever since 1981. However, just a small number of journal articles in this field were published each year. Since 2006 there has been a steady increase in the number of articles published in this field. In 2011, it reached to a peak when 11 published papers appeared in these prestigious journals. Further, Table 2 shows that at least 6 papers were published each year from 2006 to 2011. Based on the data collected, it can be concluded that the quantity of journal papers in BELF have been improved over the past several decades. Nevertheless, among these 87 papers identified, only 4 papers were empirical studies, and accounted for 4.60% (see Table 3). This fact reflects a limitation in BELF studies. Table 3. The Proportion of Empirical and Non-empirical Studies Research methods Empirical studies Non-empirical studies Total 4 83 Percentage (%) 4.60 95.40 Table 4. The Distribution of the 87 Papers by Research Area Research areas Total Percentage (%) Teaching methodology, mode and learning strategies 25 28.74 Syllabus design ,course assessment and reform 8 9.20 Testing 3 3.45 Translation studies 17 19.54 Teacher studies 2 2.30 Teaching material studies 2 2.30 Genre analysis 7 8.05 Pragmatics study 3 3.45 Systemic functional based 8 9.20 Needs analysis 2 2.30 Discourse studies 13 14.94 Other 6 6.90 Table 3. The Proportion of Empirical and Non-empirical Studies Table 4. The Distribution of the 87 Papers by Research Area Discourse studies The research areas of the 87 papers can be grouped into 12 categories (see Table 4). Some papers focused on two or more research areas are included in others. 3.2 Data Analysis Distribution of Academic Journals and BELF Articles (1981—2012)Journals Year 1 2 3 4 5 6 7 8 9 10 11 Total Percentage (%) 1981 1 1 2 2.30 1993 1 1 2 2.30 1994 1 1 1.15 1995 1 1 1 3 3.45 1996 1 1 1.15 1997 1 1 1 1 1 5 5.75 1998 2 1 3 3.45 1999 2 2 4 4.60 2000 1 1 2 2.30 2001 1 1 2 2.30 2002 2 2 2.30 2003 1 1 2 2.30 2004 1 1 3 5 5.75 2005 1 2 2 1 6 6.90 2006 2 1 2 2 1 8 9.20 2007 1 1 2 2 6 6.90 2008 2 2 2 6 6.90 2009 2 1 1 1 1 6 6.90 2010 2 1 4 1 1 9 10.35 132 Table 2. Distribution of Academic Journals and BELF Articles (1981—2012)Journals Year 1 2 3 4 5 6 7 8 9 10 11 Total Percentage (%) 1981 1 1 2 2.30 1993 1 1 2 2.30 1994 1 1 1.15 1995 1 1 1 3 3.45 1996 1 1 1.15 1997 1 1 1 1 1 5 5.75 1998 2 1 3 3.45 1999 2 2 4 4.60 2000 1 1 2 2.30 2001 1 1 2 2.30 2002 2 2 2.30 2003 1 1 2 2.30 2004 1 1 3 5 5.75 2005 1 2 2 1 6 6.90 2006 2 1 2 2 1 8 9.20 2007 1 1 2 2 6 6.90 2008 2 2 2 6 6.90 2009 2 1 1 1 1 6 6.90 2010 2 1 4 1 1 9 10.35 132 International Education Studies Vol. 6, No. 10; 2013 www.ccsenet.org/ies 2011 2 3 1 1 3 1 11 12.64 2012 1 1 1.15 Total 6 8 1 7 3 9 16 1 5 18 13 87 100% Notes: 1. Foreign Language World; 2. Foreign Language and Literature; 3. Foreign Languages Research; 4. Foreign Language Learning Theory and Practice; 5. Modern Foreign Languages; 6. Foreign Language Research; 7. Shandong Foreign Language Teaching Journal; 8. Journal of Foreign Language; 9. Journal of PLA University of Foreign Languages; 10. Foreign Language Education; 11. Foreign Languages and Their Teaching From Table 2, it can be clearly seen that the literature on BELF has been on rise ever since 1981. 3.2 Data Analysis From the data displayed in Table 4, several conclusions are apparent. First, BELF literature has focused on various BELF content areas. Second, literature on teaching methodology and mode, learning strategies has made up the largest proportion 28.74% of work with translation studies 19.54%, the second largest. Taken together this data indicate a close relationship between BELF teaching and research. Translation studies on BELF have been given a lot of attention in English as a foreign language (EFL) context. Third, the fact that literature on syllabus design, course assessment and reform, genre analysis, and systemic functional based studies has accounted for relatively higher proportions of articles, 9.20%, 8.05% and 9.20% respectively, implies that BELF course design has attracted much more attention. This finding also indicates that the time for relating theory to teaching practice has become shorter. 133 Vol. 6, No. 10; 2013 International Education Studies www.ccsenet.org/ies Literature in areas such as teacher studies, teaching material studies and testing has made up a small proportion of articles. Therefore, more research work is urgently needed in these fields. 4. Conclusion To sum up, BELF teaching and research in China has had a rapid and steady growth. BELF teaching and research is deeply rooted in the growth of the domestic economy and influenced by the process of globalization. Consequently, each achievement in BELF teaching and research will have a positive impact on economic development and global communications. There is no other subject that can surpass the great effect BELF has produced on international communications. This research is focused on journal papers published in 12 major academic journals in the field of foreign languages in China. The papers collected may to some extent indicate the trend of the development of BELF teaching and researching and provide readers with reliable references. However, more research is needed to ensure a promising future for BELF teaching and research. References Bai, J. Y. (2011). Research into the Construction of Teaching Contents and Curriculum Consistent with the Industry Standard. Shangdong Foreign Language Teaching Journal, 3, 5-9. Bao, W., & Ji, S. J. (2007). A Survey of Competence System for Business English Talents in Higher Vocational Colleges. Shangdong Foreign Language Teaching Journal, 2, 110-114. Bao, W. (2011). On Teacher Professional Development for Business English Discipline. Foreign Language Research, 2, 122-123. Cai, L. (2006). Oral business English teaching under the guidance of constructivism. Foreign Language World, 2, 32-36. Cai, Y. (2001). An Effective Way of Cultivating Multi-talents- a review over business English courses. Foreign Languages and Their Teaching, 4, 36-38. Chen, D. C. (2003). Genre Analysis of Business English Letters and its Writing Strategies. Foreign Language Teaching, 2, 52-56. Chen, D. C. (2007). Ritualized Language in English-Chinese Business Letters and its Translation Principles. Foreign Language Teaching Abroad, 4, 43-47. Chen, D. C. (2007). Ritualized language in English-Chinese Foreign Trade Letters and its Translation Principles. Foreign Language Research, 4, 43-47. Chen, J. P., & Nie, L. L. (2009). On Disciplinary Framework of Business English: Based on Current Researches. Foreign Language Education, 5, 73-76. Chen, P. (1999). Innovations in the Foreign Trade English Writing Instruction: Genre Analysis and its Applications. Foreign Language Education, 3, 28-30. Chen, X. J., Wu, J., & Tang, X. Q. (2010). A Report on the First Forum of Languages and International Business Across Disciplines. Modern Foreign Languages, 2, 70. Chen, X. L. (2010). Theoretical Issues in Business English Tests. Foreign Language Research, 6, 93-9 Chen, X. L., Xu, S. H., & Wang, L. L. (2011). The Construction Model of Business Language-Double Dimension of Ad-Discourse Construction. Foreign Language Education, 3, 23-27. Chen, Y. (1998). Communicative Approaches in Business English Teaching. Foreign Languages and Their Teaching, 8, 41-42. Chen, Z. C. (2002). Inconsistent or conflicting English translations of identical orsynonymic Chinese vocabular items:A prominent defect of New Age C-E Dictionary (2000) exposed in 101 specific instances. Journal of Sichuan International Studies University, 3, 130-134. Chen, Z. M. (2001). Dui Wai Jing Ji Mao Yi Da Xue Xiao Zhi/Annuals of the University of International Business and Economics. Beijing: Press of University of International Business and Economics. Chen, Z. S. (2002). A Review on the New Age Chinese-English Dictionary. Journal of PLA University of Foreign Languages, 4, 80-83. Dou, D. Y., & Cao, Y. (2006). A Multi-level Perspective in Curricula and Teaching of Business English. References Foreign Language Education, 6, 53-55. 134 Vol. 6, No. 10; 2013 International Education Studies www.ccsenet.org/ies Dun, G. G. (2002). Characteristics and Translation of Words in English for Business and Economics. Shangdong Foreign Language Teaching Journal, 3, 37-40. Fu, M. R. (1999). The Modern Business Styles of the Structure and Language of Business English Practical Writing-LCCI Remarks on Business English Practical Writing Code. Foreign Language Education, 4, 45-47. Fu, M. R. (2000). Business English Writing Standard and Business English Letter Writing Skills. Foreign Language Teaching Abroad, 1, 33-37. Gan, C. Y. (2006). Post-Modernist Cultural Context and International Business Negotiation. Journal of Sichuan International Studies University, 5, 73-76. Gan, C. Y. (2007). The Cultural Logic and Thinking Tracks of Post-Modernist International Business Negotiation. Journal of Sichuan International Studies University, 4, 107-110. Gu, Z. Z. (2006). Applying Task-based Approach to Business English Teaching. Shangdong Foreign Language Teaching Journal, 3, 79-82. Guan, C. L. (2005). A Discussion of A needs Analysis Necessary for an Effective Business-oriented English. Foreign Languages and Their Teaching, 3, 40-43. Hou, Y. F. (1993). Exemplifications of Specialist English in Letter of Credit. Modern Foreign Language, 2, 50-56. Hu, G. S. (1990). The Basic Linguistic Characteristics and Techniques in Foreign Set-Tech Trade Negotiations. Modern Foreign Languages, 4, 4-11. Hu, G. S. (1993). Euphemisms in International Communications. Foreign Language Education, 1, 25-32. Huang, Y. (1997). The Characteristics and Requirements of the Teaching of International Trade major. Journal of PLA University of Foreign Languages, 5, 76-77. Inman, M., & Fu, H. C. (1981). Foreign Languages and America’s International Companies. Foreign Language Teaching Abroad, 2, 16-24. Jiang, Y. J., & Guo J. C. (1997). Business English and its researching and teaching. Foreign Language Education, 1, 43-50. Kong, Q. Y. (1997). Practical Professional English-A Course-book Fostering Practical Abilities. Foreign Language World, 4, 28-30. Li, G., & Li, B. (2011). On the Translation of Business Contract from the Perspective of Systemic Functional Linguistics. Foreign Language and Literature, 1, 112-115. Li, H. Y. (2004). Focus on Lexical Chunk Productive Practice in Vocabulary Drilling-A Discussion on Vocabulary Exercise Design of the Course-book “English for Economic and Trade Purposes: an Integrated Course-book”. Foreign Language Teaching Abroad, 1, 59-64. 007). Genre Analysis of Business English Emails. Foreign Languages and Their Teaching, 7, 24-28. Li, M. Q. (2009). A Research of Business English Translation Based on the Principle of “Innovation”. Foreign Language Research, 1, 124-126. Li, M. Q. (2009). References A Research of Business English Translation Based on the Principle of "Innovation". Journal of Sichuan International Studies University, S1, 121-124. Li, R. F. (2004). The Application of Genre-based Approach in Business English Teaching Practice. Journal of Xi’an International Studies University, 1, 68-70. Li, T. C. (1998). A Discussion on Vocabulary Features of Business English Writing. Foreign Language Education, 2, 53-56. Li, T. Z. (2006). Advantages of lexical chunks in English writing for foreign trade and their productivity training methods. Foreign Language World, 1, 36-41. Li, X. P. (2011). Thoughts on the Construction of Communicative Situation in the Teaching of L2 Writing Based on the Practice of Business English Writing. Foreign Language and Literature, 3, 137-140, 148. Li, X. S. (1997). A Study on the Strategies and Skills of Business Negotiation Languages. Shangdong Foreign Language Teaching Journal, 1, 24-28. 135 International Education Studies Vol. 6, No. 10; 2013 www.ccsenet.org/ies Li, X. S. (1997). Genre Analysis and Business Discourses. Journal of Foreign Language, 2, 26-29. Li, Y. X. (2008). How to Translate English Company Names. Shangdong Foreign Language Teaching Journal, 3, 109-112. Liao, G. R. (2007). Expressive Conciseness on the Basis of Frames and Contexts-A case study on foreign trade English. Foreign Languages and Their Teaching, 11, 12-15. Lin, T. H. (2004). Three Reflections on the Teaching of International Business English. International Business Studies, 1, 57-61. Liu, F. G. (1995). Translation in China’s Economy and Trade. Foreign Languages and Their Teaching, 3, 44-47. Liu, F. G. (1999). Studies on Business Chinese-English Translation-Based on language features of Business English. Journal of PLA University of Foreign Languages, 2, 76-79. Liu, Q. J., & Li, J. (2006). Research on the Curriculum and Mode of Business English Teaching. Shangdong Foreign Language Teaching Journal, 6, 77-79 . Liu, Q. J. (2002). Innovation Teaching and Ability Cultivation in Business English. Shangdong Foreign Language Teaching Journal, 2, 46-48. Liu, Q. J. (2008). The Practice and Research of Interpretation Course in Business English Teaching. Shangdong Foreign Language Teaching Journal, 3, 78-80. Liu, Q. J. (2010). Analysis on the Characteristics of Business English Vocabulary and Comments on Business English Translation. Foreign Language Teaching Journal, 3, 89-93. Liu, S. L. (2005). Trends in the Teaching of International Business Negotiation English. Foreign Language Education, 5, 32-36. Mark Ellis, & Christine Johnson. (1994). Teaching Business English. New York: Oxford University Press. Mo, Z. S., & Sun, W. J. (2010). References A Survey on the Current Situation of Business English Writing Course books. Foreign Language Education, 5, 81-85. Mo, Z. S. (2008). Research on business English education based on economics of language. Foreign Language World, 2, 67-74. Pan, D. T. (2010). On the Translation of Business Text from the Perspective of Skopos Theory. Foreign Language Research, 5, 120-122. Peng, A. H. (2009). Exploring the Practical Principles of Translating: Business Publicity Texts of Traditional Chinese Medicine from an Intercultural Perspective. Journal of Sichuan International Studies University, S2, 107-109. Qin, L. L. (2004). Studies on the Translation of Business English-From a multi-perspective. Foreign Languages Research, 4, 62-65. Qiu, T. H. (2000). Positive Pragmatic Strategy Used in International Business and Trade Negotiations. Foreign Languages and Their Teaching, 4, 41-44. Ruan, J. Z. (2005). On the Course Design of Business English Writing and Teaching Model. Foreign Language Teaching Journal, 1, 57-59. Ruan, J. Z. (2005). On the course objectives of college Business English and its teaching principles. Foreign Language World, 3, 28-33. Rui, Y. P. (2011). Subjective Function in the Discourse Construction of Business English. Foreign Language Research, 5, 136-138. Sang, S. M. (1996). Language Features of Foreign Trade English Contract and its Translation. Foreign Languages and Their Teaching, 6, 14-17. Shi, C. R., & Bai, Y. (2012). A Review of the Research on Business English Translation in the Past Decade. Journal of PLA University of Foreign Languages, 1, 82-87. Sun, X. C. (1993). The Meaning of Compound Words Composed of here(there) + prep. In Business English. Shangdong Foreign Language Teaching Journal, 4, 39-40. Tan, H. J. (1999). Business English Teaching and Functional Approaches. Journal of PLA University of Foreign Languages, S1, 62-64. 136 Vol. 6, No. 10; 2013 International Education Studies www.ccsenet.org/ies Tang, J. (1998). Research on Typical Polite Business English Language Mechanisms. Foreign Language Education, 4, 18-21. Wang, F. (2009). Instructional Design and Evaluation of Business English(BE) Based on Networked Connectivism. Foreign Languages and Their Teaching, 11, 38-41. Wang, H. L., & Guo, J. R. (2006). Genre Analysis and Promotional Texts in Business Settings. Foreign Language Education, 4, 36-41. Wang, H. L., & Guo, J. R. (2005). Genre Analysis of English Print Advertisements. Foreign Language Education, 1, 29-32. Wang, L. (2004). The Register and Language Features of Foreign Trade Negotiation English. Foreign Languages and Their Teaching, 4, 13-15. Wang, L. F., & Jiang, J. L. (2011). References English Test for International Business of China: Its Design, Reliability & Validity Analysis. Foreign Languages and Their Teaching, 6, 38-43. Wang, L. N., & Duan, J. (2010). The Case Study Method in Business English Teaching. Foreign Language and Literature, 4, 100-104. Wang, Z. Y. (1995). Problems with Foreign Trade English Teaching. Foreign Language Research, 2, , B. L. (1994). British Cambridge Business English Test. Foreign Language Teaching Abroad, 1, 8, 3 Wen, J. (1995). In Terms of English for Economic and Trade Purposes Teaching: The Necessity of Composing a dictionary of English for Economic and Trade Purposes Teaching. Foreign Language Teaching Abroad, 1, 14-16. Wen, X. M. (2010). An Empirical Study on Vocabulary Learning Strategies among Business English Majors. Foreign Language and Literature, 5, 138-140, 149. Wu, J. X. (1990). Fuzzy Language in Foreign Trade Negotiation. Modern Foreign Languages, 4, 17-2 Xiao, H. (2008). On the Reform of the English Curriculum for Non-English Major Postgraduates in Financial Institute: a Case Study. Foreign Languages and Their Teaching, 8, 37-39. Xiao, Q. (2008). The Appraisal and Its Functions in Business English Conversation. Foreign Languages and Their Teaching, 6, 15-17. Xiao, Y. P. (2011). A Pragmatic Analysis of the Requesting Strategies in International Business Emails. Foreign Language Education, 6, 35-39. Xu, J. (2004). Genre Analysis on Business English Letters. Journal of Xi’an International Studies University, 1, 26-29. Xu, J. (2010). Business English Studies: A Functional Linguistics Perspective. Foreign Language Research, 5, 56-60. Xu, J. (2011). Translation Studies on Business English from the Perspective of Evaluation Theory. Journal of PLA University of Foreign Languages, 6, 90-93, 111. Yan, M. (2010). The Theoretical Construction of Discourse Community. Foreign Language Research, 6, 89-92. Yu, G. (2007). Social Reflection by the Evolution of Business English Words and Expressions. Shangdong Foreign Language Teaching Journal, 2, 69-73. Zeng, L. S. (2011). A Contextual Parameters-based Study of the Cognitive Mechanism for Interpreting Conceptual Meaning Transmutation —Theories and Methods in Teaching Translation of English News about Current International. Foreign Language Education, 6, 8-12. Zhao, S. R. (2009). Discussion on Application of Situational Language Teaching in Business English from the Perspective of Need Analysis. Foreign Language Teaching Journal, 2, 68-71. Zhou, M. Z. (2006). The Contextual Factors and Meaning-Determination in the Business Text Translation. Foreign Languages and Their Teaching, 11, 59-61. Zhou, R. Q. (2001). A Pragmatic Analysis of the Apologizing Strategies in Business Settings. Foreign Language Education, 4, 92-95. Zhou, W. References X. (2011). Genre Analysis of Business English Based on Discourse Community. Foreign Language Research, 6, 48-50. 137 International Education Studies Vol. 6, No. 10; 2013 www.ccsenet.org/ies Zhu, Y, Li, & J. Y. (2004). A Contrastive Study on the Chinese and English Speech Act of Refusal in Shopping Activities. Foreign Language Education, 1, 19-22. Zhu, Y, Li, & J. Y. (2004). A Contrastive Study on the Chinese and English Speech Act of Refusal in Shopping Activities. Foreign Language Education, 1, 19-22. Copyrights Copy g s Copyright for this article is retained by the author(s), with first publication rights granted to the journa py g y ( ) p g g j This is an open-access article distributed under the terms and conditions of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/3.0/). This is an open-access article distributed under the terms and conditions of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/3.0/). 138
https://openalex.org/W2802918931	https://orca.cardiff.ac.uk/id/eprint/133374/1/c8cc01661e.pdf	English	null	Long-wavelength TCF-based fluorescence probes for the detection and intracellular imaging of biological thiols	Chemical communications	2,018	cc-by	3,918	Open Access Article. Published on 17 April 2018. Downloaded on 7/13/2020 11:23:05 AM. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. Cite this: Chem. Commun., 2018, 54, 4786 Cite this: Chem. Commun., 2018, 54, 4786 Adam C. Sedgwick, a Jordan E. Gardiner, a Gyoungmi Kim,b Maksims Yevglevskis, c Matthew D. Lloyd, c A. Toby A. Jenkins, a Steven D. Bull, a Juyoung Yoon b and Tony D. James a Received 28th February 2018, Accepted 16th April 2018 Received 28th February 2018, Accepted 16th April 2018 DOI: 10.1039/c8cc01661e With our research, we are interested in the development of reaction based fluorescent probes for the detection of biologically relevant species to be used as powerful tools for the understanding of diseases.9–13 Currently, a number of fluorescent probes exist for the detection of biological thiols.14–20 However, long excitation/ emission wavelength fluorescent probes are highly desirable as they allow deeper tissue penetration, minimal background auto- fluorescence from proteins and photodamage to the biological samples. Therefore, in this work we looked to develop TCF-based systems for the long wavelength detection of GSH.11 Two ‘turn on’ TCF-based fluorescence probes were developed for the detection of biological thiols (TCF-GSH and TCFCl-GSH). TCF-GSH was shown to have a high sensitivity towards glutathione (GSH) with a 0.28 lM limit of detection. Unfortunately, at higher GSH concentrations the fluorescence intensity of TCF-GSH decreased and toxicity was observed for TCF-GSH in live cells. However, TCFCl-GSH was shown to be able to detect GSH at biologically relevant concentrations with a 0.45 lM limit of detection. No toxicity was found for TCFCl-GSH and a clear ‘turn on’ with good photostability was observed for the exogenous addition of GSH, Cys and HCys. Furthermore, TCFCl-GSH was used to evaluate the eﬀects of drug treatment on the levels of GSH in live cells. TCF-based fluorophores have an internal charge transfer (ICT) donor–p–acceptor (D–p–A) structure with long emission wavelengths (see ESI† – Scheme S1). As a result, TCF fluoro- phores have been used in many applications such as non-linear optic chromophores and fluorescent probes.21–25 Hilderbrand et al. previously developed a ‘turn on’ sulfonamide based TCF fluorescent probe for the detection of biological thiols.26 How- ever, a PEG unit was required to provide aqueous solubility and cell permeability. The probe was successfully shown to detect biological thiols in 3T3 cells. We believed the synthesis of the analogous sulfonate ester would overcome the need for a PEG unit and provide a much simpler synthesis. a Department of Chemistry, University of Bath, Bath, BA2 7AY, UK. E-mail: t.d.james@bath.ac.uk, s.d.bull@bath.ac.uk b Department of Chemistry and Nano Science, Ewha Womans University, Seoul 120-750, Korea. E-mail: jyoon@ewha.ac.kr c Drug & Target Development, Department of Pharmacy & Pharmacology, University of Bath, Claverton Down, Bath, BA2 7AY, UK † Electronic supplementary information (ESI) available: All data supporting this study. See DOI: 10.1039/c8cc01661e ChemComm This journal is ©The Royal Society of Chemistry 2018 Open Access Article. Published on 17 April 2018. Downloaded on 7/13/2020 11:23:05 AM. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. Fig. 1 TCF-based fluorescent probes for the detection of biological thiols (TCF-GSH and TCFCl-GSH). Remarkably, TCF-GSH was very sensitive towards GSH producing a full ‘turn on’ fluorescence response in the presence of 25 mM GSH. Unfortunately, at concentrations 450 mM the fluorescence intensity of TCF-GSH began to drop dramatically. This is due to attack of the TCF fluorophores by nucleophiles (Fig. 2)28 (see ESI† – Fig. S5–S11). Fig. 3 Fluorescence spectra of TCFCl-GSH (5 mM) with addition of GSH (0–750 mM) and 15 min wait between additions in PBS buffer solution, 20% v/v DMSO, pH 8.00 at 25 1C. lex = 560 15 nm. response, TCFCl-GSH also required pH 8.0 buﬀer solution (20% v/v DMSO). However, TCFCl-GSH was shown to be less sensi- tive towards the biological thiols and no decrease in fluorescence intensity was observed at higher concentrations (Fig. 3). We then evaluated the selectivity of TCF-GSH towards other biologically relevant thiols and amino acids (see ESI† – Fig. S12 and S13). As predicted, TCF-GSH reacted with the other sulphydryl (R-SH) compounds, Cys and HCys with Cys producing the largest fluorescent response. However, the overall concentrations of both Cys and HCys are low in comparison to GSH in cells.29,30 TCF-GSH demonstrated an excellent selectivity for GSH against other amino acids. This excellent selectivity permitted the evaluation of TCF- GSH for the detection of exogenous and endogenously generated thiols in live cells. Sadly, despite TCF-GSH being sensitive towards GSH, we only observed a clear ‘off–on’ response for the exogenous addition of Cys in HeLa cells. Furthermore, TCF- GSH was shown to have toxicity in cell viability experiments (see ESI† – Fig. S16 and S17). TCF-GSH is therefore unsuitable for cell imaging experiments for the detection of biothiols. Interestingly, cellular imaging experiments did not require any additional addi- tives to compensate for the 20% v/v DMSO required in the in vitro experiments. We then evaluated the selectivity of TCFCl-GSH towards other biologically relevant thiols and amino acids (see ESI† – Fig. S14 and S15). As for TCF-GSH, TCFCl-GSH reacted with the R-SH containing amino acids Cys and HCys. While excellent selectivity for GSH was observed against other amino acids. This permitted the evaluation of TCFCl-GSH for the detection of exogenous and endogenous thiols in live cells. As shown in Fig. 4 Fluorescence imaging in live cells. Open Access Article. Published on 17 April 2018. Downloaded on 7/13/2020 11:23:05 AM. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. The TCF fluoro- phore unit was synthesised as previously reported using the reaction of 3-hydroxy-3-methyl-2-butanone, malonitrile and NaOEt in EtOH. With the TCF unit in hand, the (D–p–A) systems TCF-OH and TCFCl-OH were isolated in high yield using microwave reaction conditions.27 The TCF phenols were then reacted with 2,4-dinitrobenzenesulfonylchloride to afford the desired fluorescent probes TCF-GSH and TCFCl-GSH in satisfactory yields (55% and 64%) (Fig. 1). Glutathione (GSH), cysteine (Cys) and homocysteine (HCys) play a vital role in maintaining the biological redox homeostasis.1,2 GSH is a natural tripeptide (g-L-glutamyl-L-cysteinyl-glycine), which exists in the thiol reduced form (GSH) and disulphide- oxidised (GSSG) form.2 GSH is the predominant form, which exists in millimolar concentrations in most cells where it func- tions as an antioxidant.3 Elevated levels of GSH are common in the presence of oxidative stress and the susceptibility of a cell towards reactive oxygen or nitrogen species (ROS/RNS) largely depends on the concentration of intracellular GSH.4–7 Therefore, the change in the level of GSH concentration has been associated with a number of diseases such as AIDS, liver damage, cancer and neurodegenerative disease (Alzheimer’s disease).6,7 Interestingly, it was reported that at early stages of cell proliferation (S, G2 and M phases), GSH was found to localise at the nucleus. This was believed to prevent apoptosis and provide a reduced environment for transcription factors to bind to DNA.8 On the addition of GSH, both probes TCF-GSH and TCFCl- GSH change colour from yellow to purple (see ESI† – Fig. S1 and S2). We evaluated the fluorescence behaviour of TCF-GSH, in pH 8.0 buffer solution (20% v/v DMSO) (see ESI†–Fig. S3 and S4). Interestingly, 20% v/v DMSO was required for the reaction between the probe and the chosen biological thiol to take place. We then evaluated TCF-GSH for the detection of GSH, given that it is the most predominant biological thiol in cells. 4786 \| Chem. Commun., 2018, 54, 4786--4789 This journal is ©The Royal Society of Chemistry 2018 View Article Online Fig. 3 Fluorescence spectra of TCFCl-GSH (5 mM) with addition of GSH (0–750 mM) and 15 min wait between additions in PBS buffer solution, 20% v/v DMSO, pH 8.00 at 25 1C. lex = 560 15 nm. ChemComm Communication Communication Fig. 1 TCF-based fluorescent probes for the detection of biological thiols (TCF-GSH and TCFCl-GSH). Communication Fig. 1 TCF-based fluorescent probes for the detection of biological thiols (TCF-GSH and TCFCl-GSH). Communication This journal is ©The Royal Society of Chemistry 2018 Open Access Article. Published on 17 April 2018. Downloaded on 7/13/2020 11:23:05 AM. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. HeLa cells were preincubated with 0.2 mM NMM for 20 min and washed with Dulbecco’s phosphate- buﬀered saline (DPBS) and incubated with 200 mM cysteine, homocysteine and GSH-MEE for 20 min. After washing with DPBS, cells were stained with 20 mM TCFCl-GSH for 20 min and fluorescence images acquired by confocal microscopy. (a) Only TCFCl-GSH, (b) NMM + TCFCl-GSH, (c) NMM + cysteine + TCFCl-GSH, (d) NMM + homocysteine + TCFCl- GSH and (e) NMM + GSH-MEE + TCFCl-GSH. Top: Fluorescence image (ex. 559 nm/em. 575–675 nm), bottom: merged image with DIC. Scale bar: 20 mm. Quantitative data of fluorescence intensity was calculated by FV10-ASW 4.0 software and measured per one cell. Results are expressed as mean standard deviation of three independent experiments. We therefore turned our attention towards the fluorescence properties of TCFCl-GSH. In order to produce a fluorescence Fig. 2 Fluorescence spectra of TCF-GSH (5 mM) with addition of GSH (0–500 mM) and 15 min wait between additions in PBS buffer solution, 20% v/v DMSO, pH 8.00 at 25 1C. lex = 560 15 nm. Orange dashed lines indicate fluorescence decrease at high GSH concentrations. Fig. 4 Fluorescence imaging in live cells. HeLa cells were preincubated with 0.2 mM NMM for 20 min and washed with Dulbecco’s phosphate- buﬀered saline (DPBS) and incubated with 200 mM cysteine, homocysteine and GSH-MEE for 20 min. After washing with DPBS, cells were stained with 20 mM TCFCl-GSH for 20 min and fluorescence images acquired by confocal microscopy. (a) Only TCFCl-GSH, (b) NMM + TCFCl-GSH, (c) NMM + cysteine + TCFCl-GSH, (d) NMM + homocysteine + TCFCl- GSH and (e) NMM + GSH-MEE + TCFCl-GSH. Top: Fluorescence image (ex. 559 nm/em. 575–675 nm), bottom: merged image with DIC. Scale bar: 20 mm. Quantitative data of fluorescence intensity was calculated by FV10-ASW 4.0 software and measured per one cell. Results are expressed as mean standard deviation of three independent experiments. Fig. 4 Fluorescence imaging in live cells. HeLa cells were preincubated with 0.2 mM NMM for 20 min and washed with Dulbecco’s phosphate- buﬀered saline (DPBS) and incubated with 200 mM cysteine, homocysteine and GSH-MEE for 20 min. After washing with DPBS, cells were stained with 20 mM TCFCl-GSH for 20 min and fluorescence images acquired by confocal microscopy. Open Access Article. Published on 17 April 2018. Downloaded on 7/13/2020 11:23:05 AM. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. (a) Only TCFCl-GSH, (b) NMM + TCFCl-GSH, (c) NMM + cysteine + TCFCl-GSH, (d) NMM + homocysteine + TCFCl- GSH and (e) NMM + GSH-MEE + TCFCl-GSH. Top: Fluorescence image (ex. 559 nm/em. 575–675 nm), bottom: merged image with DIC. Scale bar: 20 mm. Quantitative data of fluorescence intensity was calculated by FV10-ASW 4.0 software and measured per one cell. Results are expressed as mean standard deviation of three independent experiments. Fig. 2 Fluorescence spectra of TCF-GSH (5 mM) with addition of GSH (0–500 mM) and 15 min wait between additions in PBS buffer solution, 20% v/v DMSO, pH 8.00 at 25 1C. lex = 560 15 nm. Orange dashed lines indicate fluorescence decrease at high GSH concentrations. This journal is ©The Royal Society of Chemistry 2018 Chem. Commun., 2018, 54, 4786--4789 \| 4787 View Article Online ChemComm Communication Communication Fig. 4 (see ESI† – Fig. S18 and S19 for TCF-GSH), TCFCl-GSH displayed an already strong fluorescence response in live cells Fig. 4(a). This observation was due to the presence of endogenous thiols reacting with TCFCl-GSH. However, pre-treatment of HeLa cells with the thiol reactive N-methylmaleimide (NMM) led to the reduction of endogenous thiols and therefore low fluorescence intensity was observed when TCFCl-GSH was added Fig. 4(b). The addition of 200 mM of an exogenous thiol (Cys, HCys, or GSH-Methyl ester) led to a clear change in fluorescence intensity demonstrating the ability of TCFCl-GSH to detect thiols in cells. (see ESI† – Fig. S20 and S21 for TCF-GSH), the addition of H2O2 (500 mM) or Cisplatin (200 mM) resulted in the depletion of the endogenous thiols and consequently reduced fluorescence was observed when TCFCl-GSH was added. Subsequently, the addition of the GSH producing drug N-acetylcysteine31 recovered the GSH levels resulting in a large increase in fluorescence intensity. In summary, two ‘turn on’ TCF-based fluorescent probes have been developed for the detection of biological thiols (TCF-GSH and TCFCl-GSH). TCF-GSH was shown to have a high sensitivity towards glutathione (GSH). Unfortunately, at higher GSH concentrations the fluorescence intensity of TCF-GSH decreased and toxicity was observed in live cells making it unsuitable for cellular imaging. However, TCFCl-GSH was shown to be able to detect GSH at biological relevant concentrations. Also, no toxicity was observed for TCFCl-GSH and a clear ‘turn on’ response was observed upon the exogenous addition of GSH, Cys and HCys. Conflicts of interest Fig. 6 Intracellular fluorescence change caused by drug treatment. (B) HeLa cells were incubated with 200 mM cisplatin with or without 2 mM NAC for 6 h and stained with 20 mM TCFCl-GSH for 20 min. Quantitative data of fluorescence intensity was calculated by FV10-ASW 4.0 software and measured per one cell. Results are expressed as mean standard deviation of three independent experiments. Ex. 559 nm/em. 575–675 nm. Scale bar: 20 mm. No conflicts of interest. Open Access Article. Published on 17 April 2018. Downloaded on 7/13/2020 11:23:05 AM. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. (A) HeLa cells were incubated with 500 mM H2O2 with or without 2 mM NAC for 6 h and stained with 20 mM TCFCl-GSH for 20 min. Quantitative data of fluorescence intensity was calculated by FV10-ASW 4.0 software and measured per one cell. Results are expressed as mean standard deviation of three independent experiments. Ex. 559 nm/em. 575–675 nm. Scale bar: 20 mm. This article is licensed under a Creative Commons Attrib We would like to thank the EPSRC, the University of Bath and Prostate Cancer UK (PG14-009) for funding. ACS and JEG thank the EPSRC for studentships. TDJ wishes to thank the Royal Society for a Wolfson Research Merit Award. NMR characterisation facilities were provided through the Chemical Characterisation and Analysis Facility (CCAF) at the University of Bath (www.bath.ac.uk/ccaf). The EPSRC UK National Mass Spectrometry Facility at Swansea University is thanked for analyses. JY thanks the support from the National Research Foundation of Korea (NRF), which was funded by the Korea government (MSIP) (No. 2012R1A3A2048814). ACS, MY, MDL and TDJ are members of the Cancer Research@Bath (CR@B) network. Fig. 5 Intracellular fluorescence change caused by drug treatment. (A) HeLa cells were incubated with 500 mM H2O2 with or without 2 mM NAC for 6 h and stained with 20 mM TCFCl-GSH for 20 min. Quantitative data of fluorescence intensity was calculated by FV10-ASW 4.0 software and measured per one cell. Results are expressed as mean standard deviation of three independent experiments. Ex. 559 nm/em. 575–675 nm. Scale bar: 20 mm. Fig. 5 Intracellular fluorescence change caused by drug treatment. (A) HeLa cells were incubated with 500 mM H2O2 with or without 2 mM NAC for 6 h and stained with 20 mM TCFCl-GSH for 20 min. Quantitative data of fluorescence intensity was calculated by FV10-ASW 4.0 software and measured per one cell. Results are expressed as mean standard deviation of three independent experiments. Ex. 559 nm/em. 575–675 nm. Scale bar: 20 mm. Open Access Article. Published on 17 April 2018. Downloaded on 7/13/2020 11:23:05 AM. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. Furthermore, TCFCl-GSH was able to evaluate the eﬀects of drug treatment and the addition of ROS (H2O2) on live cells, both of which resulted in a depletion of cellular GSH levels and a reduced fluorescence intensity. Subsequent, addition of NAC increased the GSH levels and enhanced the observed fluorescence intensity. m g ( y , y , GSH-Methyl ester) led to a clear change in fluorescence intensity demonstrating the ability of TCFCl-GSH to detect thiols in cells. We then evaluated the ability of TCFCl-GSH to detect changes in the concentration levels of endogenous thiols through the addition of drugs and reactive oxygen species (ROS) such as H2O2. It is well known that GSH protects against drug induced toxicity and acts as a ROS scavenger. Therefore in Fig. 5 and 6 Fig. 5 Intracellular fluorescence change caused by drug treatment. (A) HeLa cells were incubated with 500 mM H2O2 with or without 2 mM NAC for 6 h and stained with 20 mM TCFCl-GSH for 20 min. Quantitative data of fluorescence intensity was calculated by FV10-ASW 4.0 software and measured per one cell. Results are expressed as mean standard deviation of three independent experiments. Ex. 559 nm/em. 575–675 nm. Scale bar: 20 mm. Fig. 6 Intracellular fluorescence change caused by drug treatment. (B) HeLa cells were incubated with 200 mM cisplatin with or without 2 mM NAC for 6 h and stained with 20 mM TCFCl-GSH for 20 min. Quantitative data of fluorescence intensity was calculated by FV10-ASW 4.0 software and measured per one cell. Results are expressed as mean standard deviation of three independent experiments. Ex. 559 nm/em. 575–675 nm. Scale bar: 20 mm. Open Access Article. Published on 17 April 2018. Downloaded on 7/13/2020 11:23:05 AM. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. Open Access Article. Published on 17 April 2018. Downloaded on 7/13/2020 11:23:05 AM. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. We then evaluated the ability of TCFCl-GSH to detect changes in the concentration levels of endogenous thiols through the addition of drugs and reactive oxygen species (ROS) such as H2O2. It is well known that GSH protects against drug induced toxicity and acts as a ROS scavenger. Therefore in Fig. 5 and 6 Fig. 5 Intracellular fluorescence change caused by drug treatment. Notes and references Commun., 2018, 54, 4786--4789 This journal is ©The Royal Society of Chemistry 2018 View Article Online Communication ChemComm 21 Y. J. Wang, Y. Shi, Z. Y. Wang, Z. F. Zhu, X. Y. Zhao, H. Nie, J. Qian, A. J. Qin, J. Z. Sun and B. Z. Tang, Chem. – Eur. J., 2016, 22, 9784–9791. 11 A. C. Sedgwick, H. H. Han, J. E. Gardiner, S. D. Bull, X. P. He and T D James Chem Commun 2017 53 12822–12825 11 A. C. Sedgwick, H. H. Han, J. E. Gardiner, S. D. Bull, X. P. He and T. D. James, Chem. Commun., 2017, 53, 12822–12825. D. James, Chem. Commun., 2017, 53, 12822–12825. 12 A. C. Sedgwick, H.-H. Han, J. E. Gardiner, S. D. Bull, X.-P. He and T. D. James, Chem. Sci., 2018, 9, 3672–3676. 22 W. Shu, L. G. Yan, Z. K. Wang, J. Liu, S. Zhang, C. Y. Liu and B. C. Zhu, Sens. Actuators, B, 2015, 221, 1130–1136. T. D. James, Chem. Sci., 2018, 9, 3672–3676 13 A. C. Sedgwick, A. Hayden, B. Hill, S. D. Bull, R. B. P. Elmes and T. D. James, Front. Chem. Sci. Eng., 2018, DOI: 10.1007/s11705-017-1697-0. 23 Y. H. Yang, J. L. Liu, H. Y. Xiao, Z. Zhen and S. H. Bo, Dyes Pigm., 2017, 139, 239–246. 14 D. Wu, A. C. Sedgwick, T. Gunnlaugsson, E. U. Akkaya, J. Yoon and T. D. James, Chem. Soc. Rev., 2017, 46, 7105–7123. 24 T. Yu, G. X. Yin, P. Yin, Y. Zeng, H. T. Li, Y. Y. Zhang and S. Z. Yao, RSC Adv., 2017, 7, 24822–24827. 15 S. Q. Wang, S. L. Shen, Y. R. Zhang, X. Dai and B. X. Zhao, Chin. J. Org. Chem., 2014, 34, 1717–1729. 25 B. C. Zhu, H. Kan, J. K. Liu, H. G. Liu, Q. Wei and B. Du, Biosens. Bioelectron., 2014, 52, 298–303. 16 L. L. Yin, Z. Z. Chen, L. L. Tong, K. H. Xu and B. Tang, Chin. J. Anal. Chem., 2009, 37, 1073–1081. 26 J. Bouﬀard, Y. Kim, T. M. Swager, R. Weissleder and S. A. Hilderbrand, Org. Lett., 2008, 10, 37–40. 17 Y. M. Yang, Q. Zhao, W. Feng and F. Y. Li, Chem. Rev., 2013, 113, 192–270. 27 M. Ipuy, C. Billon, G. Micouin, J. Samarut, C. Andraud and Y. Bretonniere, Org. Biomol. Chem., 2014, 12, 3641–3648. 18 Y. Qi, Y. Huang, B. W. Li, F. Zeng and S. Z. This journal is ©The Royal Society of Chemistry 2018 Notes and references 1 L. B. Poole, Free Radical Biol. Med., 2015, 80, 148–157. 1 L. B. Poole, Free Radical Biol. Med., 2015, 80, 148–157. 2 G. Y. Wu, Y. Z. Fang, S. Yang, J. R. Lupton and N. D. Turner, J. Nutr., 2004, 134, 489–492. 2 G. Y. Wu, Y. Z. Fang, S. Yang, J. R. Lupton and N. D. Turner, J. Nutr., 2004, 134, 489–492. 3 S. C. Lu, Mol. Aspects Med., 2009, 30, 42–59. 4 J. S. Armstrong, K. K. Steinauer, B. Hornung, J. M. Irish, P. Lecane, G. W. Birrell, D. M. Peehl and S. J. Knox, Cell Death Diﬀer., 2002, 9, 252–263. 5 M. Valko, D. Leibfritz, J. Moncol, M. T. D. Cronin, M. Mazur and J. Telser, Int. J. Biochem. Cell Biol., 2007, 39, 44–84. 6 C. B. Pocernich and D. A. Butterfield, Biochim. Biophys. Acta, 2012, 1822, 625–630. 7 N. Ballatori, S. M. Krance, S. Notenboom, S. J. Shi, K. Tieu and C. L. Hammond, Biol. Chem., 2009, 390, 191–214. Fig. 6 Intracellular fluorescence change caused by drug treatment. (B) HeLa cells were incubated with 200 mM cisplatin with or without 2 mM NAC for 6 h and stained with 20 mM TCFCl-GSH for 20 min. Quantitative data of fluorescence intensity was calculated by FV10-ASW 4.0 software and measured per one cell. Results are expressed as mean standard deviation of three independent experiments. Ex. 559 nm/em. 575–675 nm. Scale bar: 20 mm. Fig. 6 Intracellular fluorescence change caused by drug treatment. (B) HeLa cells were incubated with 200 mM cisplatin with or without 2 mM NAC for 6 h and stained with 20 mM TCFCl-GSH for 20 min. Quantitative data of fluorescence intensity was calculated by FV10-ASW 4.0 software and measured per one cell. Results are expressed as mean standard deviation of three independent experiments. Ex. 559 nm/em. 575–675 nm. Scale bar: 20 mm. 8 J. Markovic, C. Borras, A. Ortega, J. Sastre, J. Vina and F. V. Pallardo, J. Biol. Chem., 2007, 282, 20416–20424. 9 A. C. Sedgwick, R. S. L. Chapman, J. E. Gardiner, L. R. Peacock, G. Kim, J. Yoon, S. D. Bull and T. D. James, Chem. Commun., 2017, 53, 10441–10443. 10 A. C. Sedgwick, X. L. Sun, G. Kim, J. Yoon, S. D. Bull and T. D. James, Chem. Commun., 2016, 52, 12350–12352. 4788 \| Chem. Chem. Commun., 2018, 54, 4786--4789 \| 4789 Notes and references Wu, Anal. Chem., 2018, 90, 1014–1020. 28 W. L. Wu, Z. Y. Wang, X. Dai, J. Y. Miao and B. X. Zhao, Sci. Rep., 2016, 6, 25315. 19 J. Li, Y. Kwon, K. S. Chung, C. S. Lim, D. Lee, Y. K. Yue, J. Yoon, G. Kim, S. J. Nam, Y. W. Chung, H. M. Kim, C. X. Yin and J. H. Ryu, Theranostics, 2018, 8, 1411–1420. 29 R. Banerjee, J. Biol. Chem., 2012, 287, 4397–4402. 20 L. Chen, J.-S. Park, D. Wu, C.-H. Kim and J. Yoon, Sens. Actuators, B, 2018, 262, 306–312. 30 P. Ganguly and S. F. Alam, J. Nutr., 2015, 14, 6. 31 S. Y. Sun, Cancer Biol. Ther., 2010, 9, 109–110.
https://openalex.org/W2221498111	https://www.banglajol.info/index.php/BJP/article/download/24238/17492	English	null	Antitubercular activity of the pigment from forest soil <i>Streptomyces</i> sp SFA5	Bangladesh journal of pharmacology	2,015	cc-by	2,288	Abstract Extracellular pigment from the forest soil Streptomyces sp SFA5 was produced by submerged fermentation using yeast extract malt extract broth. Crude pigment from the medium was extracted using ethyl acetate. Antitubercular activity of the pigment was tested against M. tuberculosis H37Rv by microplate alamar blue assay and luciferase reporter phage assay. The pigment was also tested for inhibitory activity against M. tuberculosis lysine aminotransferase by colorimetric method. In both microplate alamar blue and luciferase reporter phage assay, the crude pigment showed activity against M. tuberculosis H37Rv at 125 and 250 µg/mL concentration, respectively. The M. tuberculosis lysine aminotransferase was inhibited at the IC50 value of 4.5 µg/ mL concentration. sources have yielded an impressive number of com- pounds over the past 50 years (Busti et al., 2006). Among the microbial sources, Actinobacteria are the well-explored source in terms of secondary metabolites especially antibiotics. The diversity of Actinobacteria in various natural including extreme and man-made environments is well documented. They are the most economically valuable prokaryotes which are well- known to produce chemically diverse metabolites with a wide range of biological activities (Berdy, 2015; Velho- Pereira and Kamat, 2013). From January 1990 to December 2012, in total 949 anti-mycobacterium natural products were reported in the literature (Dashti et al., 2014). Recently the rate of discovering new compounds from terrestrial actinobacteria has decreased but the rate of re -isolation of known actinomycetes and antibiotics is on the increase. Antitubercular activity of the pigment from forest soil Streptomyces sp SFA5 Radhakrishnan Manikkam1, Sekar Ponnuswamy2, Jerrine Joseph1 and Vanaja Kumar1 Radhakrishnan Manikkam1, Sekar Ponnuswamy2, Jerrine Joseph1 and Vanaja 1Centre for Drug Discovery and Development, Sathyabama University, Chennai 600 119, India; 2Department of Chemical Engineering, SSN College of Engineering, Chennai 603 110, India. Article Info Received: 17 July 2015 Accepted: 28 August 2015 Available Online: 25 December 2015 DOI: 10.3329/bjp.v11i1.24238 Cite this article: Radhakrishnan M, Shekar P, Jerrine J, Vanaja K. Antitubercular activity of the pigment from forest soil Strepto- myces sp SFA5. Bangladesh J Pharmacol. 2016; 11: 138-40. Bangladesh J Pharmacol 2016; 11: 138-140 Bangladesh J Pharmacol 2016; 11: 138-140 A Journal of the Bangladesh Pharmacological Society (BDPS) Bangladesh J Pharmacol 2016; 11: 138-140 Journal homepage: www.banglajol.info Abstracted/indexed in Academic Search Complete, Asia Journals Online, Bangladesh Journals Online, Biological Abstracts, BIOSIS Previews, CAB Abstracts, Current Abstracts, Directory of Open Access Journals, EMBASE/Excerpta Medica, Google Scholar, HINARI (WHO), International Pharmaceutical Abstracts, Open J-gate, Science Citation Index Expanded, SCOPUS and Social Sciences Citation Index; ISSN: 1991-0088 This work is licensed under a Creative Commons Attribution 4.0 International License. You are free to copy, distribute and per- form the work. You must attribute the work in the manner specified by the author or licensor Bangladesh Journal of Pharmacology Research Article Antitubercular activity of the pig- Antitubercular activity of the pig- Antitubercular activity of the pig- ment from forest soil ment from forest soil ment from forest soil Streptomyces Streptomyces Streptomyces sp SFA5 sp SFA5 sp SFA5 BJP This article was downloaded by you on: Jun 26, 2017 Bangladesh J Pharmacol 2016; 11: 138-140 Introduction The incidence of emerging infectious diseases in humans has increased within the recent past or threatens to increase in the near future. Mycobacterium tuberculosis still remains a deadly pathogen two decades after the announcement of tuberculosis as a global health emergency by the World Health Organization. In 2013, an estimated 9.0 million people developed TB and 1.5 million died from the disease, 360,000 of whom were HIV-positive (WHO report, 2014). In last few years new drug combinations have shown promising potential to significantly shorten tuberculosis treatment times. However, there are very few new chemical entities being developed to treat this global threat (Riccardi and Pasca, 2014). Hence, there is a dire need for new chemical entities to develop as potential antitubercular molecules. Hence researchers are now searching rare ecosystems, instead of normal terrestrial sources, such as marine, mountain and forest ecosystems for Actinobacteria for potentially new biosynthetic diversity (Lam, 2006; Berdy, 2012). With this view, the present study was Microorganisms of both terrestrial and marine origins have proven to be excellent sources of novel natural products. Programs aimed at the discovery of antibio- tics and other bioactive metabolites from microbial Bangladesh J Pharmacol 2016; 11: 138-140 139 added with 50 L of crude extract in order to get the final concentration of 250 g/mL and 500 µg/mL. One hundred microliter of M. tuberculosis cell suspension was added to all the vials. DMSO (1%) was also included in the assay as a solvent control. All the vials were incubated at 37°C for 72 hours. After incubation, 50 L of high titre phage phAE129 and 40 L of 0.1M CaCl2 solutions were added to the test and control vials. All the vials were incubated at 37°C for 4 hours. After incubation, 100 L from each vial was transferred to luminometer cuvette. About 100 L of D-luciferin was added and relative light unit (RLU) was measured in luminometer. RLU reduction by 50% or more when compared to control was considered as having antitubercular activity. attempted to study the antitubercular activity of forest soil Streptomyces sp SFA5. Results Streptomyces sp SFA5 showed good growth with soluble yellow pigment production on YEME broth. In microplate alamar blue assay, the growth of M. tuberculosis H37Rv was inhibited at 125 µg/mL concentration. In luciferase reporter phage assay, the crude pigment showed 60 and 74% reduction in RLU at 250 and 500 µg/mL concentration, respectively. Results of both the method showed that the yellow pigment produced by Streptomyces sp SFA5 was active against M. tuberculosis H37Rv. Standard drugs isoniazid and rifampicin showed more than 95% reduction in RLU in luciferase reporter phage assay. In enzyme inhibition assay, the MTB-LAT enzyme was inhibited by 4.98 µg/ mL concentration of crude pigment. Lysine aminotransferase enzyme inhibition assay MTB lysine aminotransferase activity was determined by the detection of piperidine 6-carboxylate. The crude pigment was tested at <1 to 100 µg/mL concentration. Briefly, 15 mM enzyme solution was added to 1.0 mL 200 mM phosphate buffer pH 7.2 containing 1 mM L- lysine–HCl, 1 mM a-ketoglutarate and 15 mM PLP. The mixture was incubated at 310 K for 1 hour. The reaction was terminated by the addition of 500 mL 10% trichloroacetic acid in ethanol. Piperidine 6-carboxylate was detected by measuring the color intensity of its adduct with o-aminobenzaldehyde spectroscopically at 465 nm (Samala et al., 2014). Description of Streptomyces sp SFA5 Streptomyces sp SFA5 was isolated from the soil sample collected from Sabarimalai forest (Western Ghats), Kerala, India using starch casein agar medium. The soluble yellow pigment produced by the strain SFA5 on yeast extract malt extract agar showed good activity against bacterial pathogens. The viability of strain SFA5 was maintained on ISP2 agar slants as well as in 30% glycerol broth. %RLU reduction = Control RLU − Test RLU / Control RLU × 100 %RLU reduction = Control RLU − Test RLU / Control RLU × 100 Microplate alamar blue assay Briefly, the inoculum was prepared from fresh LJ medium resuspended in 7H9 medium, adjusted to a McFarland tube number 1, and diluted 1:20; 100 L was used as inoculum. Each drug stock solution was thawed and diluted in 7H9-S at 4-fold the final highest concentration tested. Serial 2-fold dilutions of crude pigment were prepared 6.125 to 500 µg/mL concentra- tion directly in a sterile 96-well microtiter plate using 100 L 7H9-S. A growth control containing no antibiotic and a sterile control was also prepared on each plate. Sterile water was added to all perimeter wells to avoid evaporation during the incubation. The plate was covered, sealed in plastic bags and incubated at 37°C in normal atmosphere. After 7 days of incubation, 30 L of alamar blue solution was added to each well, and the plate was re-incubated overnight. A change in color from blue (oxidized state) to pink (reduced) indicated the growth of bacteria, and the MIC was defined as the lowest concentration of drug that prevented this change in color (Samala et al., 2014). Production and extraction of pigment Spores of Streptomyces sp SFA5 was transferred into 50 mL of YEME broth and incubated in a rotary shaker at 28°C for 48 hours. About 10% of inoculum was transferred into each 100 mL of YEME broth and incubated at 28°C for 120 hours. After incubation, the cell-free supernatant was prepared by centrifugation at 5,000 rpm for 30 min. The pigment present in the cell free supernatant was extracted by liquid-liquid extraction using ethyl acetate at 1:1 ratio. Then the ethyl acetate portion was concentrated and dried using rotary evaporator (Radhakrishnan et al., 2007). References Ashforth EJ, Fu C, Liu X. Bioprospecting for antituberculosis leads from microbial metabolites. Nat Prod Rep. 2010; 27: 1709–19. Berdy J. Thoughts and facts about antibiotics: Where we are now and where we are heading? J Antibiot. 2012; 65: 385–95. The inability to convert target inhibition into growth inhibition and eventually to bacterial cell death, which bedevils the target-driven approach, is circumvented by identifying compounds with potent antitubercular activity by whole-cell screening, which is clearly a feasible starting point (Manjunatha and Smith, 2015). In the present study, crude pigment exhibited good antitubercular activity in both rapid and whole-cell screening based methods such as microplate alamar blue assay and luciferase reporter phage assay. Previously there are many authors reported the anti- tubercular activity of natural products by microplate alamar blue assay. However, the luciferase reporter phage assay was rapid which produced results in 3 days when compared to microplate alamar blue assay which took 7-10 days to produce results. Berdy J. Bioactive microbial metabolites. J Antibiot. 2005; 58: 1– 26 Busti E, Monciardini P, Cavaletti L, Bamonte R, Lazzarini A, Sosio M, Donadio S. Antibiotic-producing ability by repre- sentatives of a newly discovered lineage of actinomycetes. Microbiology 2006; 152: 675–83. Dashti Y, Grkovic T, Quinn RJ. Predicting natural product value, an exploration of anti-TB drug space. Nat Prod Rep. 2014; 31: 990-98. Lam KS. Discovery of novel metabolites from marine actino- mycetes. Curr Opin Microbiol. 2006; 9: 245–51. Manjunatha UH, Smith PW. Perspectives: Challenges and opportunities in TB drug discovery from phenotypic screening. Bioorg Med Chem. 2015; 23: 5087-97. Nguta JM, Appiah-Opong R, Nyakro AK, Yeboah-Manu D, Addo PGA. Current perspectives in drug discovery against tuberculosis from natural products. Int J Mycobacteriol. 2015; 4: 165-83. Lysine aminotransferase in M. tuberculosis is an important enzyme for its long-term persistence. It is implicated in mycobacterial stress response and is up- regulated approximately 40-fold in nutrient-starved models designed to mimic the persistent/latent state of tuberculosis. It catalyzes an overall reaction involving the transfer of the epsilon-amino group of L-lysine to a- ketoglutarate to yield L-glutamate and alpha-amino- adipate-D-semialdehyde and then to piperidine 6- carboxylate. In this study, lysine aminotransferase inhibition activity of crude pigment was determined by the detection of reaction product piperidine 6- carboxylate by measuring the color intensity of its adduct with o-aminobenzaldehyde (Samala et al., 2014). References The low IC50 value exhibited by the crude pigment indicated that this pigment will be a promising candidate to fight against latent tubercle bacilli also. Radhakrishnan M, Balagurunathan R, Selvakumar N, Mukesh Doble, Vanaja Kumar. Bioprospecting of marine derived actinomycetes with special reference to antimycobacterial activity. Indian J Geo-Marine Sci. 2011; 40: 407-10. Radhakrishnan M, Balaji S, Balagurunathan R. Thermotolerant actinomycetes from Himalayan Mountain: Antagonistic potential, characterization and identification of selected strains. Malaysian Appl Biol. 2007; 36: 59-65. Radhakrishnan M, Gopikrishnan V, Balaji S, Balagurunathan R, Vanaja K.. Bioprospecting of actinomycetes from certain less explored ecosystems active against Mycobacterium tuber- culosis and other non-mycobacterial pathogens. Int Scholarly Res Notices. 2014; 2014. Riccardi G, Pasca MR. Trends in discovery of new drugs for tuberculosis therapy. J Antibiot. 2014; 67: 655–59. Crude natural product extracts are complex mixtures of perhaps hundreds of different compounds working together in synergy when the extract is administered as a whole. Discovery of natural product hits and their progression towards development includes extraction of the crude extract from the source, concentration, fractionation and purification to yield a single bioactive compound (Nguta et al., 2015). Samala G, Kakan SS, Nallangi R, Brindha Devi P, Sridevi JP, Saxena S, Yogeeswari P, Sriram D. Investigating structure– activity relationship and mechanism of action of anti- tubercular1-(4-chlorophenyl)-4-(4-hydroxy-3-methoxy-5- nitrobenzylidene) pyrazolidine-3,5- dione [CD59]. Int J Mycobacteriol. 2014; 3: 117–26. Velho-Pereira S, Kamat NM. Actinobacteriological research in India. Indian J Exp Biol. 2013; 51: 573-96. World Health Organization. WHO Global tuberculosis report. 2014. Discussion Antitubercular activity of pigment was also studied against standard laboratory strain M. tuberculosis H37Rv by adopting luciferase reporter phage assay (Radhakrishnan et al., 2014). About 350 L of G7H9 broth supplemented with 10% albumin dextrose complex and 0.5% glycerol was taken in cryovials and The majority of natural product collections usually start as crude extracts of fresh or dried material processed by different methods using various chemical solvents. In the present study, antitubercular activity exhibited by the crude pigment extracted from the cell-free Bangladesh J Pharmacol 2016; 11: 138-140 140 supernatant indicated its extracellular nature. Most of the actinobacterial compounds are extracellular in nature (Radhakrishnan et al., 2011). Further, action- bacterial pigments are reported to exhibit wide range of biological activities such as antibiotic, anti-cancer, antifungal etc. Acknowledgement Authors thank the management of Sathyabama Univer- sity for the research facilities provided. Authors also thank Dr. D. Sriram, BITS-Pilani, Hyderabad campus for his kind help in antitubercular activity screening. Author Info Radhakrishnan Manikkam (Principal contact) e-mail: mrkactinos@yahoo.com Author Info Radhakrishnan Manikkam (Principal contact) e-mail: mrkactinos@yahoo.com Your feedback about this paper Your feedback about this paper Your feedback about this paper 1. Number of times you have read this paper 2. Quality of paper 3. Your comments 0 Click 1. Number of times you have read this paper 3. Your comments
https://openalex.org/W2586772191	https://boris.unibe.ch/111767/1/Mhimbira%20PLoSNeglTropDis%202017.pdf	English	null	Prevalence and clinical relevance of helminth co-infections among tuberculosis patients in urban Tanzania	PLoS neglected tropical diseases	2,017	cc-by	11,888	RESEARCH ARTICLE Editor: Michael H. Hsieh, George Washington University, UNITED STATES Editor: Michael H. Hsieh, George Washington University, UNITED STATES Editor: Michael H. Hsieh, George Washington University, UNITED STATES Received: September 12, 2016 Accepted: January 20, 2017 Published: February 8, 2017 Copyright: © 2017 Mhimbira et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Received: September 12, 2016 Accepted: January 20, 2017 Published: February 8, 2017 Methodology Copyright: © 2017 Mhimbira et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Between November 2013 and October 2015, we enrolled adult (18 years) sputum smear- positive TB patients and household contact controls without TB during an ongoing TB cohort study in Dar es Salaam, Tanzania. We used Baermann, FLOTAC, Kato-Katz, point-of-care circulating cathodic antigen, and urine filtration to diagnose helminth infections. Multivariable logistic regression models with and without random effects for households were used to assess for associations between helminth infection and TB. Data Availability Statement: Due to ethical restrictions imposed by the Ifakara Health Institute Institutional Review Board and National Health Research Ethics sub-Committee of the National Institute for Medical Research of Tanzania on protecting patient confidentiality, data cannot be made publicly available. Some of identified reasons include many unique identifiers such as TB district number which is the unique number assigned to TB patients who are diagnosed and treated at the health facility where also this study is being conducted. In Abstract Citation: Mhimbira F, Hella J, Said K, Kamwela L, Sasamalo M, Maroa T, et al. (2017) Prevalence and clinical relevance of helminth co-infections among tuberculosis patients in urban Tanzania. PLoS Negl Trop Dis 11(2): e0005342. doi:10.1371/journal. pntd.0005342 Prevalence and clinical relevance of helminth co-infections among tuberculosis patients in urban Tanzania Francis Mhimbira1,2,3, Jerry Hella1,2,3, Khadija Said1,2,3, Lujeko Kamwela1, Mohamed Sasamalo1,2,3, Thomas Maroa1, Magreth Chiryamkubi4, Grace Mhalu1,2,3, Christian Schindler3,5, Klaus Reither1,2,3, Stefanie Knopp3,5,6, Ju¨rg Utzinger3,5, Se´bastien Gagneux2,3, Lukas Fenner1,2,3,7 1 Department of Intervention and Clinical Trials, Ifakara Health Institute, Dar es Salaam, Tanzania, 2 Department of Medical Parasitology and Infection Biology, Swiss Tropical and Public Health Institute, Basel, Switzerland, 3 University of Basel, Basel, Switzerland, 4 Department of Curative Services, Ministry of Health, Community Development, Gender, Elderly and Children, Dar es Salaam, Tanzania, 5 Department of Epidemiology and Public Health, Swiss Tropical and Public Health Institute, Basel, Switzerland, 6 Wolfson Wellcome Biomedical Laboratories, Department of Life Sciences, Natural History Museum, London, United Kingdom, 7 Institute of Social and Preventive Medicine, University of Bern, Bern, Switzerland * fmhimbira@ihi.or.tz (FM); lukas.fenner@unibas.ch (LF) Background Helminth infections can negatively affect the immunologic host control, which may increase the risk of progression from latent Mycobacterium tuberculosis infection to tuberculosis (TB) disease and alter the clinical presentation of TB. We assessed the prevalence and deter- mined the clinical relevance of helminth co-infection among TB patients and household con- tact controls in urban Tanzania. * fmhimbira@ihi.or.tz (FM); lukas.fenner@unibas.ch (LF) Epidemiology of TB and helminth co-infection interval (CI): 0.88–1.80, p = 0.22), but S. mansoni infection was (aOR 2.15, 95% CI: 1.03– 4.45, p = 0.040). Moreover, S. mansoni infection was associated with lower sputum bacte- rial load (aOR 2.63, 95% CI: 1.38–5.26, p = 0.004) and tended to have fewer lung cavita- tions (aOR 0.41, 95% CI: 0.12–1.16, p = 0.088). interval (CI): 0.88–1.80, p = 0.22), but S. mansoni infection was (aOR 2.15, 95% CI: 1.03– 4.45, p = 0.040). Moreover, S. mansoni infection was associated with lower sputum bacte- rial load (aOR 2.63, 95% CI: 1.38–5.26, p = 0.004) and tended to have fewer lung cavita- tions (aOR 0.41, 95% CI: 0.12–1.16, p = 0.088). addition, the dataset has the following variables that may breach patient privacy such as; place of treatment, gender, geographical position system (GPS) of residence of TB patients enrolled into this study, household size, and year of birth. Interested researchers should contact Dr. Frederick Haraka (fharaka@ihi.or.tz) for further information related to data access. Author summary Tuberculosis (TB), caused by the bacterium Mycobacterium tuberculosis, and parasitic worm infections are typical diseases of poverty. They often overlap geographically, and can occur in the same individual. Parasitic worm infections contribute to the down-regu- lation of the essential immune response against TB, and therefore can increase progres- sion from latent M. tuberculosis infection to active TB. We conducted a case-control study in Dar es Salaam, the economic capital of Tanzania, where TB and helminths constitute a considerable burden. We found that infection with the blood fluke Schistosoma mansoni was associated with active TB, while none of the other parasitic worms showed such an association. Interestingly, TB patients infected with S. mansoni had significantly lower sputum bacterial load at diagnosis and tended to have fewer lung cavitations compared with TB patients without any parasitic worm infection. Diagnosis and treatment of para- sitic worm infections, particularly schistosomiasis, should be considered during the man- agement of TB patients and in the context of TB control programs. This could help to reduce the TB burden in settings where TB and parasitic worms co-exist. Competing interests: The authors have declared that no competing interests exist. Conclusions/Significance S. mansoni infection was an independent risk factor for active TB and altered the clinical pre- sentation in TB patients. These findings suggest a role for schistosomiasis in modulating the pathogenesis of human TB. Treatment of helminths should be considered in clinical man- agement of TB and TB control programs. Funding: LF received funding from Rugolf Geigy Foundation (http://www.geigystiftung.ch/en/). FM received funding for his PhD studies from Ifakara Health Institute (www.ihi.or.tz) and Amt fu¨r Ausbildungsbeitra¨ge (http://www.hochschulen.bs. ch/ueber-uns/organisation/amt- ausbildungsbeitraege.html). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. ausbildungsbeitraege.html). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. PLOS Neglected Tropical Diseases \| DOI:10.1371/journal.pntd.0005342 February 8, 2017 Principal findings A total of 597 TB patients and 375 household contact controls were included. The median age was 33 years and 60.2% (585/972) were men. The prevalence of any helminth infec- tion among TB patients was 31.8% (190/597) and 25.9% (97/375) among controls. Stron- gyloides stercoralis was the predominant helminth species (16.6%, 161), followed by hookworm (9.0%, 87) and Schistosoma mansoni (5.7%, 55). An infection with any hel- minth was not associated with TB (adjusted odds ratio (aOR) 1.26, 95% confidence 1 / 19 PLOS Neglected Tropical Diseases \| DOI:10.1371/journal.pntd.0005342 February 8, 2017 Study design The study was conducted within the frame of an ongoing prospective cohort study of TB patients and household contact controls in Dar es Salaam (TB-DAR). We assessed the associa- tion of TB and helminth infection in a case-control study design of TB patients (sputum smear-positives for acid-fast bacilli [AFB]) and household contact controls (Xpert MTB/RIF negative), who were matched by age (±5 years) and whenever possible by sex. We prospectively followed-up TB patients and assessed the clinical outcomes comparing TB patients with and without helminth infection at 6 and 12 months after recruitment. Epidemiology of TB and helminth co-infection helminth infections is characterized by the induction of CD4+ T-helper 2 (Th2) and down-reg- ulation of CD4+ T-helper 1 (Th1) cells [12–15]. This immunological imbalance has been sug- gested to increase the risk of progression from latent M. tuberculosis infection to active TB and to worsen the clinical outcomes. We aimed to study the interaction between TB and helminth co-infections by comparing the prevalence of helminth infections, using a suite of diagnostic techniques, between TB patients and household contact controls without TB in an ongoing cohort study in Dar es Salaam, Tanzania, and to assess the effects of helminth infection on the clinical presentation and outcomes of TB disease. Ethics statement The study protocol was approved by the institutional review board of the Ifakara Health Insti- tute (IHI; reference no. IHI/IRB/No 04–2015) and the Medical Research Coordinating Com- mittee of the National Institute of Medical Research (NIMR; reference no. NIMR/HQ/R.8c/ Vol.I/357) in Tanzania, and the ethics committee of north-west and central Switzerland (EKNZ; reference no.: UBE-15/42). Written informed consent was obtained from all study participants. TB patients were treated according to the National TB and Leprosy Programme (NTLP) treatment guideline [8]. Individuals with a Schistosoma spp. infection were treated with praziquantel (40 mg/kg). Other helminth infections were treated with albendazole (400 mg) immediately after diagnosis, as recommended by the national treatment guidelines [16]. HIV-positive patients were clinically managed according to the Tanzania National HIV and acquired immune deficiency syndrome (AIDS) treatment guideline [17]. Study setting The study was conducted in the densely populated urban setting of Temeke district in Dar es Salaam, which is the economic capital of Tanzania. The population of Temeke is estimated at 1.4 million. In 2014, about one third of all TB patients from Dar es Salaam were notified in Temeke district (4,373; 32%) [18]. The overall HIV prevalence in the general adult population in Dar es Salaam is 5.2% [19]. The study area includes two TB sub-districts, Wailes I and Wailes II, whose patients are clinically managed at the Temeke district hospital and the two associated TB diagnostic and treatment centers of Tambukareli and Pasada [20]. Introduction Tuberculosis (TB), caused by Mycobacterium tuberculosis remains a challenging disease to con- trol. Indeed, over two billion people are estimated to be infected with M. tuberculosis world- wide [1]. Moreover one billion people are infected with soil-transmitted helminths, Tuberculosis (TB), caused by Mycobacterium tuberculosis remains a challenging disease to con- trol. Indeed, over two billion people are estimated to be infected with M. tuberculosis world- wide [1]. Moreover one billion people are infected with soil-transmitted helminths, schistosomes, filarial worms, and food-borne trematodes [2–4]. In 2014, an estimated 9.6 mil- lion new TB patients were notified and 1.5 million TB patients died from the disease [1]. TB is a leading cause of deaths from an infectious disease [5]. p p wide [1]. Moreover one billion people are infected with soil-transmitted helminths, schistosomes, filarial worms, and food-borne trematodes [2–4]. In 2014, an estimated 9.6 mil- lion new TB patients were notified and 1.5 million TB patients died from the disease [1]. TB is a leading cause of deaths from an infectious disease [5]. TB and helminthiases overlap geographically, particularly in areas where poverty persists, for example in countries of sub-Saharan Africa [1,6]. Where TB and helminth infections co- occur, they can affect the same individual and thus exacerbate the course of disease [6]. Several conditions such as diabetes mellitus, malnutrition, and malignancies are known to increase the risk of progressing from latent M. tuberculosis infection to active TB [7]. Human immuno- deficiency Virus (HIV)-induced immunodeficiency is by far the most important risk factor for developing TB [1,8], but parasitic co-infections such as with helminths can also contribute to the development of TB [9–11]. Immune dysregulations caused by helminth infections are known to negatively affect the prognosis of HIV and malaria [6,12]. The immune response to 2 / 19 PLOS Neglected Tropical Diseases \| DOI:10.1371/journal.pntd.0005342 February 8, 2017 Study procedures TB patients and household contact controls were interviewed and underwent physical exami- nation during recruitment at the study site (see under “Data Collection and Definitions”). We collected skinfold measurements from four body sites (biceps, triceps, subscapular, and suprai- liac) using the Harpenden skinfold caliper [22]. The percentage body fat was calculated as pre- viously described [23]. Household contacts with no symptoms or signs of TB submitted a sputum sample for Gene Xpert MTB/RIF to rule-out TB. We collected blood, stool, and urine samples from TB patients and controls for subsequent laboratory investigations. Chest X-rays for TB patients were done at the Temeke district hospital and were interpreted by an experi- enced board certified radiologist who was blinded to patients’ clinical data. Trained field work- ers collected geographic coordinates (global positioning system [GPS]) from the patients’ homes using Samsung Tab 4 android tablets (Samsung; Suwon, South Korea). Epidemiology of TB and helminth co-infection contact control. Controls at recruitment were free of symptoms and signs suggestive of TB, healthy on physical examination, and had a negative Xpert MTB/RIF result (Cepheid; Califor- nia, United States of America). Assuming a helminth prevalence of 45% in TB patients and 26% in controls based on results from previous publications [21] and a power of 80%, the target sample size was 109 study par- ticipants (for each group) to detect a prevalence difference of 19% between the two groups with a significance level of test 0.05, two-tailed and calculated with Stata version 14.0 (Stata Corp; Texas, United States of America). Study population and sample size We consecutively enrolled study participants starting in November 2013 until October 2015 to reach the required sample size. Over this period, we included adult TB patients (18 years of age and sputum-smear positive) and household contact controls. Any individual living in the same household as the index TB patients enrolled in the study is referred to as a household 3 / 19 PLOS Neglected Tropical Diseases \| DOI:10.1371/journal.pntd.0005342 February 8, 2017 PLOS Neglected Tropical Diseases \| DOI:10.1371/journal.pntd.0005342 February 8, 2017 Data collection and definitions We collected socio-demographic indicators including age, sex, ethnicity, education, and household income. Anthropometric data included weight, height, and skinfold measurements. Clinical data collected pertained to presenting symptoms of TB patients, TB treatment cate- gory, and treatment outcomes. Laboratory data included ZN sputum smear results and Gene Xpert MTB/RIF results, helminth species infections, HIV status, full blood cell count, and CD4+ cell count. All study participants were asked about their use of anthelmintic treatment in the last 12 months prior to the enrollment into the study. Study data were captured by elec- tronic case report forms using the open-source data collection software ODK on Android PC tablets [30]. Data management was done using the eManagement tool “odk_planner”, as previ- ously described [30]. Data were uploaded to a password protected secure server with regular back-ups. In order to grade the clinical severity of TB, we adopted a previously published clinical TB score [31], with the following modification: 12 points TB score parameters instead of 13 points as tachycardia was not systematically measured. The following TB score parameters were used: (i) coughing; (ii) hemoptysis; (iii) chest pain; (iv) dyspnea; (v) night sweating; (vi) anemic con- junctivae; (vii) positive finding at auscultation; (viii) axillary temperature >37.0˚C; (ix) mid upper arm circumference (MUAC) <220 mm; (x) MUAC <200 mm; (xi) body mass index (BMI) <18 kg/m2; and (xii) BMI <16 kg/m2. TB score was then categorized into mild (score of 1–5) and severe (score of 6). Low BMI was defined as BMI <18 kg/m2; high sputum bacte- rial load as AFB sputum smear result 2+ (quantitative scoring), which correlates with Gen- eXpert Ct values [32]. To assess the clinical outcomes among TB patients, we defined poor gain as a change in absolute body weight (<7 and 7 kg), BMI (<2.6 and 2.6 kg/m2) and body fat (<0 and 0%) from recruitment to month 6 of follow-up. “Any helminth infection” was defined as infection with any of the following helminth spe- cies: A. lumbricoides, E. vermicularis, hookworm, Hymenolepis diminuta, S. haematobium, S. mansoni, S. stercoralis and T. trichiura. High occupational risk for schistosomiasis was defined as working in rice fields, sand harvesting, washing cars, and fishing in freshwater. The intensity of helminth infection was defined according to WHO classification [33]. Epidemiology of TB and helminth co-infection Blood testing. In line with national HIV testing algorithms, screening was done using the Alere Determine HIV rapid test (Alere, USA). The Uni-gold HIV (Trinity Biotech; Wicklow, Ireland) rapid test served as a confirmatory test in case of a positive screening test. The CD4+ T-cells counts were determined using a FACSCount machine (Becton Dickinson Biosciences; California, United States of America). A full blood cell count was done with a MS4 Vet hema- tology analyzer (Diamond Diagnostics; Massachusetts, United States of America). All blood tests were performed at the Temeke district hospital laboratory, which is under supervision and quality control by the regional laboratory technician. Data collection and definitions The average egg counts from the triplicate Kato-Katz thick smears per stool sample and per individual were multiplied by a factor of 24 to obtain eggs per gram (EPG) of stool [25]. Laboratory procedures Microbiological investigations. A patient was considered as having TB when any of the two submitted sputum samples were positive for AFB by staining sputum smears using the Ziehl-Nielsen (ZN) method, and a positive mycobacterial culture. Sputum smear microscopy was done at the Temeke district hospital under continuous quality control by the central tuber- culosis reference laboratory (Dar es Salaam, Tanzania). AFB smear-positive results were graded according to World Health Organization/International Union Against Tuberculosis and Lung Disease (WHO/IUATLD) guidelines: “scanty” with 1–9 AFB per 100 oil immersion fields; “1+” with 10–99 AFB per 100 immersion fields; “2+” with 1–10 AFB per 1 immersion field, and “3+” with >10 AFB per immersion field [8,24]. To rule out TB among household controls, an additional sputum sample from TB patients and controls was sent to the TB labo- ratory at the Bagamoyo Research and Training Center (BRTC), IHI, for GeneXpert MTB/RIF (controls) and for culture on Lo¨wenstein-Jensen media (TB patients and controls). Helminthological investigations. For the diagnosis of helminth infections, single stool and urine samples were collected from each participant before the start of TB treatment (TB patients) and at the time of enrolment (controls). All stool and urine samples were transferred to the Helminth Unit at BRTC and examined for helminth infections using standardized, qual- ity-controlled procedures as described elsewhere [25–27]. The Kato-Katz (triplicate thick smears per stool sample) and the FLOTAC methods were used to diagnose Ascaris lumbricoides, hookworm, S. mansoni, and Trichuris trichiura infections. The Baermann method was used to identify Strongyloides stercoralis infections [28]. The adhesive tape test was used to diagnose Enterobius vermicularis infections [26]. In addition, a rapid point-of-care circulating cathodic antigen (POC-CCA) urine cassette test was employed for the diagnosis of S. mansoni [29]. The urine filtration method was applied to detect S. haematobium infections [26]. For quality con- trol, 10% of Kato-Katz slides were randomly selected and re-examined by a second reader. PLOS Neglected Tropical Diseases \| DOI:10.1371/journal.pntd.0005342 February 8, 2017 4 / 19 Characteristics of study participants A total of 597 TB patients and 375 household contact controls were included. Table 1 summa- rizes the socio-demographic and clinical characteristics of TB patients and controls. The study participants’ flow diagram is shown in Fig 1. Among all study participants, the median age was 33 years (interquartile range [IQR]: 26–41 years) and 60.2% (585/972) were men. HIV preva- lence was 20.4% (95% confidence interval (CI): 17.9–23.0%). TB patients were more frequently male compared with controls (68.8% [411/597] vs. 46.4% [174/375]), HIV-positive (27.3% [163] vs. 9.3% [35]), and smokers (18.1% [108] vs. 8.8% [33]). TB patients also had a lower median BMI (18.3 kg/m2, IQR: 16.5–20.4 kg/m2 vs. 23.9 kg/m2, IQR: 21.6–28.1 kg/m2) and a lower median hemoglobin level (11.3 g/dl, IQR: 9.9–12.7 g/dl vs. 12.8 g/dl, IQR: 11.5–14.1 g/dl). The patient characteristics, stratified by HIV status, are shown in S1 Table. Epidemiology of TB and helminth co-infection control. In addition, we also performed conditional logistic regression among matched pairs to confirm the results. Additional analyses assessed the association of TB and with specific hel- minth species separately. We also examined whether the association between the presence of a helminth infection and a recent history of deworming drugs depended on HIV infection status by including an interaction term in the logistic regression model. Among TB patients, logistic regression models were used to study associations between helminth infection and clinical pre- sentation at the time of TB diagnosis (such as TB score, high sputum bacterial load, lung infil- tration, and cavitation), and to study the association between helminth infection and clinical outcomes after 6 months of TB treatment (change in absolute weight, BMI, and percentage body fat). Associations were expressed as crude odds ratios (ORs) and adjusted ORs (aORs). All analyses were performed in Stata version 14.0 (Stata Corp; Texas, United States of America). We used the geographic coordinates of the TB patients’ homes to analyze the spatial distri- bution of TB and helminth co-infections. The prevalence of helminths and helminth species was analyzed at the ward level for optimal readability. The average area per ward in the Dar es Salaam region is 15.5 km2 [19]. The maps were produced using the software package ArcGIS Desktop version 10.2 (ESRI; California, United States of America) and the shape files from the National Bureau of Statistics of Tanzania [34]. Statistical analysis We compared the characteristics of TB patients and household contact controls at the time of TB diagnosis or enrolment. The prevalence of helminth infection was calculated from the gen- eralized estimations equation adjusting for clustering at the household level. We used multi- level mixed-effects logistic regression with random intercepts at the level of households to assess risk factors for helminth infection. To assess risk factors for TB, we compared cases and controls using unconditional logistic regression because not all TB cases could be assigned a 5 / 19 PLOS Neglected Tropical Diseases \| DOI:10.1371/journal.pntd.0005342 February 8, 2017 BMI, body mass index; HIV, human immunodeﬁency virus; Hb, hemoglobin level; IQR, inter-quartile range; MUAC, mid-upper arm circumference; US$, United States dollars (1 US$ = 2,190 Tanzanian Shillings in March 2016) Prevalence and risk factors for helminth infection Among all participants, the prevalence of any helminth infection was 29.5% (95% CI: 26.7– 32.6%). S. stercoralis (16.5%, 161) was the predominant helminth species, followed by hook- worm (9.0%, 87), S. mansoni (5.7%, 55) and S. haematobium (2.0%, 19). Overall, TB patients were more frequently co-infected with any helminth species compared with controls (OR 1.34, 95% CI: 1.00–1.78, p = 0.048; Table 2). The prevalence of helminth infection was lower in HIV-positive (22.7%, 45) compared with HIV-negative study participants (31.3%, 242; S1 Table). Similarly, helminth infection was lower among TB patients co-infected with HIV (22.7%, 37) compared with HIV-negative TB patients (35.3%, 153; S2 Table). We found that most study participants had light-intensity helminth infection. For example, 96.4% (54) of study participants had light-intensity hookworm infection as determined by the Kato-Katz method (S3 Table). The prevalence and geographic distribution of species-specific helminth infections in the study area is shown in S1 Fig. Study participants with occupational risk for acquiring schistosomiasis, such as working in rice fields, sand harvesting, washing cars, and fishing had higher odds of being infected with any helminth species (aOR 1.42, 95% CI: 1.04–1.95, p = 0.029). HIV-positive patients were less 6 / 19 PLOS Neglected Tropical Diseases \| DOI:10.1371/journal.pntd.0005342 February 8, 2017 a Occupational risk for acquiring schistosomiasis (working in rice ﬁelds, sand harvesting, washing cars, and ﬁshing) pational risk for acquiring schistosomiasis (working in rice ﬁelds, sand harvesting, washing cars, and ﬁshing) BMI, body mass index; HIV, human immunodeﬁency virus; Hb, hemoglobin level; IQR, inter-quartile range; MUAC United States dollars (1 US$ = 2,190 Tanzanian Shillings in March 2016) Frequency distribution of helminth infections, stratified by TB patients and household contact controls. Helminth infection All TB patients Controls Comparing TB patients and controlsa (n = 972) (n = 597) (n = 375) n (%) n (%) n (%) OR (95% CI) p-value Any helminth 287 (29.5) 190 (31.8) 97 (25.9) 1.34 (1.00–1.78) 0.048 Helminth species Strongyloides stercoralis 161 (16.6) 111 (18.6) 50 (13.3) 1.48 (1.03–2.13) 0.032 Hookworm 87 (9.0) 55 (9.2) 32 (8.5) 1.09 (0.69–1.72) 0.72 Ascaris lumbricoides 6 (0.6) 3 (0.5) 3 (0.8) 0.63 (0.13–3.12) 0.57 Enterobius vermicularis 5 (0.5) 1 (0.2) 4 (1.1) NA NA Trichuris trichiura 9 (0.9) 6 (1.0) 3 (0.8) 1.25 (0.31–5.06) 0.75 Hymenolepis diminuta 2 (0.2) 1 (0.2) 1 (0.3) NA NA Schistosoma spp. 70 (7.2) 49 (8.2) 21 (5.6) 1.51 (0.89–2.56) 0.13 Schistosoma mansoni 55 (5.7) 40 (6.7) 15 (4.0) 1.72 (0.94–3.17) 0.079 Schistosoma haematobium 19 (2.0) 11 (1.8) 8 (2.1) 0.86 (0.34–2.16) 0.75 Helminth infection 0.13 None 685 (70.5) 407 (68.2) 278 (74.1) 1 Mono-infection 237 (24.4) 158 (26.5) 79 (21.1) 1.37 (1.00–1.86) Infection with 2 species 50 (5.1) 32 (5.3) 18 (4.8) 1.21 (0.67–2.21) a Estimates from an unadjusted mixed-effect models with household as a random intercept NA, not applicable; OR, odds ratio doi:10.1371/journal.pntd.0005342.t002 Epidemiology of TB and helminth co-infection Epidemiology of TB and helminth co-infection Fig 1. Study participants’ flow diagram. doi:10.1371/journal.pntd.0005342.g001 Fig 1. Study participants’ flow diagram. doi:10.1371/journal.pntd.0005342.g001 Table 2. Frequency distribution of helminth infections, stratified by TB patients and household contact controls. Helminth infection All TB patients Controls Comparing TB patients and controlsa (n = 972) (n = 597) (n = 375) n (%) n (%) n (%) OR (95% CI) p-value Any helminth 287 (29.5) 190 (31.8) 97 (25.9) 1.34 (1.00–1.78) 0.048 Helminth species Strongyloides stercoralis 161 (16.6) 111 (18.6) 50 (13.3) 1.48 (1.03–2.13) 0.032 Hookworm 87 (9.0) 55 (9.2) 32 (8.5) 1.09 (0.69–1.72) 0.72 Ascaris lumbricoides 6 (0.6) 3 (0.5) 3 (0.8) 0.63 (0.13–3.12) 0.57 Enterobius vermicularis 5 (0.5) 1 (0.2) 4 (1.1) NA NA Trichuris trichiura 9 (0.9) 6 (1.0) 3 (0.8) 1.25 (0.31–5.06) 0.75 Hymenolepis diminuta 2 (0.2) 1 (0.2) 1 (0.3) NA NA Schistosoma spp. Epidemiology of TB and helminth co-infection Table 1. Socio-demographic and clinical characteristics of tuberculosis (TB) patients and household contact controls without TB. Characteristics Total (n = 972) TB patient (n = 597) Controls (n = 375) Age in years, median (IQR) 33 (26–41) 33 (26–40) 33 (26–42) Age groups (years) 18–24 194 (20.0) 107 (17.9) 87 (23.2) 25–34 347 (35.7) 226 (37.9) 121 (32.3) 35–44 266 (27.4) 169 (28.3) 97 (25.9) 45 165 (17.0) 95 (15.9) 70 (18.7) Sex Female 387 (39.8) 186 (31.2) 201 (53.6) Male 585 (60.2) 411 (68.8) 174 (46.4) HIV status Negative 774 (79.6) 434 (72.7) 340 (90.7) Positive 198 (20.4) 163 (27.3) 35 (9.3) Education level No/primary 806 (82.9) 500 (83.8) 306 (81.6) Secondary/University 166 (17.1) 97 (16.2) 69 (18.4) Occupation Unemployed 349 (35.9) 204 (34.2) 145 (38.7) Employed 623 (64.1) 393 (65.8) 230 (61.3) Smoking status No 831 (85.5) 489 (81.9) 342 (91.2) Yes 141 (14.5) 108 (18.1) 33 (8.8) People in the household 3 731 (75.2) 442 (74.0) 289 (77.1) >3 241 (24.8) 155 (26.0) 86 (22.9) Household income per month (US$) 100 763 (78.5) 473 (79.2) 290 (77.3) >100 209 (21.5) 124 (20.8) 85 (22.7) Body weight at diagnosis, [in kg], (IQR) 54 (48–61) 51 (46–57) 59 (53–67) BMI (kg/m2), median (IQR) 20.0 (17.6–23.4) 18.3 (16.6–20.4) 23.9 (21.6–28.1) BMI categories (kg/m2) Underweight <18.5 337 (34.7) 318 (53.3) 19 (5.1) Normal, 18.5–24.9 454 (46.7) 256 (42.9) 198 (52.8) Overweight 25.0–29.9 119 (12.2) 21 (3.5) 98 (26.1) Obese 30 62 (6.4) 2 (0.3) 60 (16.0) Body fat (%) 10.1 (7.7–14.7) 9.5 (6.8–13.7) 11.5 (8.5–17.0) MUAC (cm), median (IQR) 24.3 (22.7–26.2) 23.3 (22.0–25.3) 25.3 (23.7–28.0) Waist hip ratio, median (IQR) 0.89 (0.86–0.94) 0.89 (0.86–0.94) 0.89 (0.86–0.94) Occupational riska No 521 (54.2) 322 (54.2) 199 (54.1) Yes 441 (45.8) 272 (45.8) 169 (45.9) Individual deworming (past 12 months) Yes 797 (82.0) 484 (81.1) 313 (83.5) No 175 (18.0) 113 (18.9) 62 (16.5) Hb level (g/dl), median (IQR) 12 (10.4–13.3) 11.3 (9.9–12.7) 12.8 (11.5–14.1) a Occupational risk for acquiring schistosomiasis (working in rice ﬁelds, sand harvesting, washing cars, and ﬁshing) BMI, body mass index; HIV, human immunodeﬁency virus; Hb, hemoglobin level; IQR, inter-quartile range; MUAC, mid-upper arm circumference; US$, nical characteristics of tuberculosis (TB) patients and household contact controls without TB. PLOS Neglected Tropical Diseases \| DOI:10.1371/journal.pntd.0005342 February 8, 2017 7 / 19 Fig 1. Study participants’ flow diagram. doi:10.1371/journal.pntd.0005342.g001 Table 2. a Estimates from an unadjusted mixed-effect models with household as a random intercept a Estimates from an unadjusted mixed-effect models with household as a random intercept a Estimates from an unadjusted mixed-effect models with household as a random intercept NA, not applicable; OR, odds ratio 70 (7.2) 49 (8.2) 21 (5.6) 1.51 (0.89–2.56) 0.13 Schistosoma mansoni 55 (5.7) 40 (6.7) 15 (4.0) 1.72 (0.94–3.17) 0.079 Schistosoma haematobium 19 (2.0) 11 (1.8) 8 (2.1) 0.86 (0.34–2.16) 0.75 Helminth infection 0.13 None 685 (70.5) 407 (68.2) 278 (74.1) 1 Mono-infection 237 (24.4) 158 (26.5) 79 (21.1) 1.37 (1.00–1.86) Infection with 2 species 50 (5.1) 32 (5.3) 18 (4.8) 1.21 (0.67–2.21) a Estimates from an unadjusted mixed-effect models with household as a random intercept equency distribution of helminth infections, stratified by TB patients and household contact controls. Table 2. Frequency distribution of helminth infections, stratified by TB patient PLOS Neglected Tropical Diseases \| DOI:10.1371/journal.pntd.0005342 February 8, 2017 8 / 19 likely to be infected with any helminth species (aOR 0.57, 95% CI: 0.37–0.87, p = 0.010; Table 3). Study participants who did not take anthelmintic treatment in the past 12 months did not have significant higher odds of being co-infected with any helminth species (aOR 1.35, 95% CI: 0.92–1.99, p = 0.12). There was no statistically significant interaction between the Epidemiology of TB and helminth co-infection Table 3. Risk factors for any helminth infection among TB patients and household controls without TB. Characteristic Helminth infection, n (%) Unadjusted Adjusted Yes No OR (95% CI) p-value aOR (95% CI) p-value Participant 0.054 0.18 Controls 97 (33.8) 278 (40.6) 1.00 1.00 TB patients 190 (66.2) 407 (59.4) 1.35 (1.00–1.82) 1.29 (0.88–1.87) Age group (years) 0.30 0.46 18–24 50 (17.4) 144 (21.0) 1.00 1.00 25–34 115 (40.1) 232 (33.9) 1.46 (0.96–2.23) 1.38 (0.89–2.17) 35–44 75 (26.1) 191 (27.9) 1.13 (0.72–1.78) 1.11 (0.68–1.82) 45 47 (16.4) 118 (17.2) 1.16 (0.70–1.92) 1.18 (0.69–2.03) Sex 0.003 0.24 Female 93 (32.4) 294 (42.9) 1.00 1.00 Male 194 (67.6) 391 (57.1) 1.60 (1.17–2.18) 1.23 (0.87–1.75) HIV status 0.022 0.010 Negative 242 (84.3) 532 (77.7) 1.00 1.00 Positive 45 (15.7) 153 (22.3) 0.63 (0.43–0.94) 0.57 (0.37–0.87) BMI category (kg/m2) 0.077 0.47 BMI 18 175 (61.0) 460 (67.2) 1.00 1.00 BMI <18 112 (39.0) 225 (32.8) 1.32 (0.97–1.79) 1.14 (0.79–1.64) Education level 0.28 0.50 No/primary 243 (84.7) 563 (82.2) 1.00 1.00 Secondary/University 44 (15.3) 122 (17.8) 0.80 (0.53–1.20) 0.86 (0.55–1.34) Employment status 0.13 0.42 Unemployed 93 (32.4) 256 (37.4) 1.00 1.00 Employed 194 (67.6) 429 (62.6) 1.28 (0.93–1.76) 1.16 (0.81–1.65) Number of people in the household 0.66 0.97 3 218 (76.0) 69 (24.0) 1.00 1.00 >3 513 (74.9) 172 (25.1) 0.93 (0.65–1.32) 0.99 (0.69–1.42) Household income per month (US$) 0.47 0.75 100 229 (79.8) 534 (78.0) 1.00 1.00 >100 58 (20.2) 151 (22.0) 0.87 (0.60–1.26) 0.94 (0.63–1.40) Individual deworming (past 12 months) 0.043 0.12 Yes 224 (78.0) 573 (83.6) 1.00 1.00 No 63 (22.0) 112 (16.4) 1.48 (1.01–2.15) 1.35 (0.92–1.99) Occupational riska 0.009 0.029 No 136 (47.7) 385 (56.9) 1.00 1.00 Yes 149 (52.3) 292 (43.1) 1.50 (1.11–2.03) 1.42 (1.04–1.95) a Occupational risk for acquiring schistosomiasis (working in rice ﬁelds, sand harvesting, washing cars, and ﬁshing) BMI, body mass index; HIV, human immunodeﬁcieny virus; US$, United States dollars (1 US$ = 2,190 Tanzanian Shillings in March 2016) Multilevel mixed-effects logistic regression model with household as a random intercept, adjusted for TB status, age-groups, sex, HIV status, BMI, a Occupational risk for acquiring schistosomiasis (working in rice ﬁelds, sand harvesting, washing cars, and ﬁshing) BMI, body mass index; HIV, human immunodeﬁcieny virus; US$, United States dollars (1 US$ = 2,190 Tanzanian Shillings in March 2016) Multilevel mixed-effects logistic regression model with household as a random intercept, adjusted for TB status, age-groups, sex, HIV status, BMI, education level, employment status, number of people living in the same household, individual deworming status, occupational risk, and income level. Helminth infection as a risk factor for TB Multiple logistic regression models adjusted for patient characteristics and known risk factors for TB showed that any helminth infection was not statistically significantly associated with TB (aOR 1.26, 95% CI: 0.88–1.80, p = 0.22, Table 4 and S7 Table). However, when analyzing each helminth species separately, we found that S. mansoni infection was significantly associated with TB (aOR 2.15, 95% CI: 1.03–4.45, p = 0.040), but there was no significant association between TB and S. stercoralis or hookworm infection (S8 Table). Other co-factors that were significantly associated with TB included: male sex, HIV co-infection, smoking, living in a household with 3 people, and a low BMI (Table 4). The unadjusted and adjusted ORs for any helminth infection and S. mansoni are shown in S7 Table. Results were more pronounced when using a conditional logistic regression model (S9 Table). Epidemiology of TB and helminth co-infection effects of HIV infection and deworming status on TB incidence (P-value from test for interac- tion: 0.5). When analyzing the risk factors for helminth infection separately for TB patients and household controls without TB, we found similar results (see S5 and S6 Tables). Effect of helminth infection on clinical presentation and disease severity in TB patients TB patients co-infected with any helminth infection were more likely than helminth un- infected TB patients to present with hemoptysis (74 [38.9%] vs. 123 [30.2%]), had higher median hemoglobin levels (11.7 g/dl, IQR: 10.1–13.0 g/dl vs. 11.3 g/dl, IQR: 9.8–12.5 g/dl) and higher median eosinophil counts (0.2, IQR: 0.1–0.4 cells/μl vs. 0.1, IQR: 0.05–0.2 cells/μl; Table 5). TB patients co-infected with S. mansoni were more likely to have lower sputum bacte- rial load than helminth-uninfected TB patients (aOR 2.63; 95% CI: 1.38–5.26, p = 0.004). Fur- thermore, we found that TB patients co-infected with S. mansoni tended to have fewer lung cavities, although this association lacked statistical significance (aOR 0.41, 95% CI: 0.12–1.16, p = 0.088; Table 6). There were no statistically significant differences in radiological features between TB patients with and without any helminth infection as shown in S10 Table. Effect of helminth infection on clinical outcomes in TB patients Overall, 81.7% (273 TB patients) were cured at the end of TB treatment (at 6 months), 17.4% (58) completed treatment (AFB smear results not available at 6 months, but documented com- pletion of treatment), and 0.9% (3) were treatment failures (positive AFB smear result at 6 months). We found no significant associations between helminth infection (at time of recruit- ment) and poor gain in absolute weight (aOR 0.89, 95% CI: 0.55–1.45, p = 0.63), BMI (aOR 0.74, 95% CI: 0.46–1.21, p = 0.23), and body fat percentage (aOR 0.92, 95% CI: 0.55–1.56, p = 0.78) after 6 months on TB treatment, as shown in S11 Table. Note: interaction between the effect of HIV and deworming status on the risk for any helminth infection: p = 0.50 d i 10 1371/j l td 0005342 t003 likely to be infected with any helminth species (aOR 0.57, 95% CI: 0.37–0.87, p = 0.010; Table 3). Study participants who did not take anthelmintic treatment in the past 12 months did not have significant higher odds of being co-infected with any helminth species (aOR 1.35, 95% CI: 0.92–1.99, p = 0.12). There was no statistically significant interaction between the PLOS Neglected Tropical Diseases \| DOI:10.1371/journal.pntd.0005342 February 8, 2017 9 / 19 Discussion We present findings on the prevalence and association of TB and helminth co-infection among adult TB patients and household contact controls in a highly-urbanized setting of Dar es Salaam, Tanzania. We found that S. mansoni infection was a risk factor for TB disease. This association remained significant after adjustment for other known risk factors for TB, such as HIV infection, smoking, and underweight [35]. None of the other investigated helminth spe- cies or the surrogate measure of “any helminth infection” were significantly associated with TB. Importantly, associations between any helminth co-infection and TB were reported in pre- vious epidemiologic studies [21,36,37], as well as in experimental work using animal or 10 / 19 PLOS Neglected Tropical Diseases \| DOI:10.1371/journal.pntd.0005342 February 8, 2017 Logistic regression model for TB disease status as the outcome. Model adjusted for any helminth infection/S. mansoni, age, sex, HIV status, BMI, education level, employment status, smoking status, number of people living in the same household, individual deworming status, helminth risk occupation and income level. Epidemiology of TB and helminth co-infection Table 4. Associations of TB disease with helminth infection and other patient characteristics. Characteristics Any helminth infection (n = 972) S. mansoni infection (n = 972) TB patients Controls Adjusted Adjusted n (%) n (%) aOR (95% CI) p-value aOR (95% CI) p-value Helminth infection 0.22 0.040 No 407 (68.2) 278 (74.1) 1.00 1.00 Yes 190 (31.8) 97 (25.9) 1.26 (0.88–1.80) 2.15 (1.03–4.45) Age group (years) 0.49 0.25 18–24 107 (17.9) 87 (23.2) 1.00 1.00 25–34 226 (37.9) 121 (32.3) 1.22 (0.77–1.94) 1.24 (0.78–1.97) 35–44 169 (28.3) 97 (25.9) 1.00 (0.60–1.67) 1.02 (0.61–1.7) 45 95 (15.9) 70 (18.7) 0.85 (0.48–1.48) 0.88 (0.51–1.54) Sex <0.001 <0.001 Female 186 (31.2) 201 (53.6) 1.00 1.00 Male 411 (68.8) 174 (46.4) 3.12 (2.13–4.56) 3.16 (2.16–4.63) HIV status <0.001 <0.001 Negative 434 (72.7) 340 (90.7) 1.00 1.00 Positive 163 (27.3) 35 (9.3) 6.18 (3.83–9.95) 6.23 (3.86–10.05) Education level 0.55 0.57 No/primary 500 (83.8) 306 (81.6) 1.00 1.00 Secondary/University 97 (16.2) 69 (18.4) 1.15 (0.73–1.80) 1.14 (0.72–1.79) Employment status 0.63 0.66 Unemployed 204 (34.2) 145 (38.7) 1.00 1.00 Employed 393 (65.8) 230 (61.3) 0.91 (0.62–1.33) 0.92 (0.63–1.34) Smoking status 0.012 0.011 No 489 (81.9) 342 (91.2) 1.00 1.00 Yes 108 (18.1) 33 (8.8) 1.92 (1.15–3.21) 1.95 (1.16–3.25) Number of people in the household 0.018 0.015 3 people 442 (74.0) 289 (77.1) 1.00 1.00 >3 people 155 (26.0) 86 (22.9) 1.58 (1.08–2.30) 1.60 (1.09–2.34) Household income per month (US$) 0.85 0.95 100 473 (79.2) 290 (77.3) 1.00 1.00 >100 124 (20.8) 85 (22.7) 1.04 (0.69–1.56) 1.01 (0.68–1.52) BMI category (kg/m2) <0.001 <0.001 BMI 18 279 (46.7) 318 (53.3) 1.00 1.00 BMI <18 356 (94.9) 19 (5.1) 23.20 (13.91–38.69) 23.52 (14.1–39.24) Occupational riska 0.24 0.26 No 322 (54.2) 199 (54.1) 1.00 1.00 Yes 272 (45.8) 169 (45.9) 0.82 (0.59–1.15) 0.83 (0.59–1.15) Individual deworming (past 12 months) 0.20 0.21 Yes 484 (81.1) 313 (83.5) 1.00 1.00 No 113 (18.9) 62 (16.5) 0.75 (0.48–1.16) 0.76 (0.49–1.17) BMI, body mass index; CI, conﬁdence interval; HIV, human immunodeﬁcieny virus; OR, odds ratio; US$, United States dollars (1 US$ = 2,190 Tanzanian Shillings in March 2016) sociations of TB disease with helminth infection and other patient characteristics. PLOS Neglected Tropical Diseases \| DOI:10.1371/journal.pntd.0005342 February 8, 2017 11 / 19 Epidemiology of TB and helminth co-infection Table 5. Patient characteristics of TB patients infected and not infected with helminths at the time of TB diagnosis. Characteristics Total TB and helminth TB only p-value (n = 597) (n = 190) (n = 407) Age, median (IQR) (years) 33 (26–40) 31 (26–39) 34 (27–40) 0.22 Age groups (years) 0.13 18–24 107 (17.9) 35 (18.4) 72 (17.7) 25–34 226 (37.9) 81 (42.6) 145 (35.6) 35–44 169 (28.3) 42 (22.1) 127 (31.2) 45 95 (15.9) 32 (16.8) 63 (15.5) Sex 0.007 Female 186 (31.2) 45 (23.7) 141 (34.6) Male 411 (68.8) 145 (76.3) 266 (65.4) HIV status 0.003 Negative 434 (72.7) 153 (80.5) 281 (69.0) Positive 163 (27.3) 37 (19.5) 126 (31.0) CD4+ count, cells/mla 202 (94–273) 185 (90–259) 203 (100–273) 0.74 Education level 0.49 No/primary 500 (83.8) 162 (85.3) 338 (83.0) Secondary/University 97 (16.2) 28 (14.7) 69 (17.0) Occupation 0.99 Unemployed 204 (34.2) 65 (34.2) 139 (34.2) Employed 393 (65.8) 125 (65.8) 268 (65.8) Number of people in the household 0.89 3 people 442 (74.0) 140 (73.7) 302 (74.2) > 3 people 155 (26.0) 50 (26.3) 105 (25.8) Smoking status <0.004 No 489 (81.9) 143 (75.3) 346 (85.0) Yes 108 (18.1) 47 (24.7) 61 (15.0) Household income per month (US$) 0.45 100 473 (79.2) 154 (81.1) 319 (78.4) >100 124 (20.8) 36 (18.9) 88 (21.6) Body weight (kg), median (IQR) 51 (46–57) 50.9 (46–56) 51.7 (46–57.5) 0.40 BMI (kg/m2), median(IQR) 18.3 (16.6–20.4) 18.2 (16.5–20.2) 18.5 (16.6–20.4) 0.22 BMI (kg/m2) groups, n (%) 0.40b Underweight <18.5 318 (53.3) 108 (56.8) 210 (51.6) Normal, 18.5–24.9 256 (42.9) 78 (41.1) 178 (43.7) Overweight 25.0–29.9 21 (3.5) 4 (2.1) 17 (4.2) Obese 30 2 (0.3) 0 2 (0.5) Body fat (%) 9.5 (6.8–13.7) 9.1 (6.0–12.7) 9.8 (7.4–14.0) 0.008 MUAC (cm), median (IQR) 23.3 (22.0–25.3) 23.7 (22.0–25.0) 23.3 (22.0–25.7) 0.99 Waist hip ratio, median (IQR) 0.89 (0.85–0.94) 0.89 (0.85–0.94) 0.89 (0.86–0.94) 0.75 Occupational risk 0.095 No 322 (54.2) 93 (49.2) 229 (56.5) Yes 272 (45.8) 96 (50.8) 176 (43.5) Individual deworming (past 12 months) 0.013 Yes 484 (81.1) 143 (75.3) 341 (83.8) No 113 (18.9) 47 (24.7) 66 (16.2) Symptomsc Cough 594 (99.5) 189 (99.5) 405 (99.5) 0.96 Fever 551 (92.3) 174 (91.6) 377 (92.6) 0.65 (Continued) Table 5. (Continued) Characteristics Total TB and helminth TB only p-value (n = 597) (n = 190) (n = 407) Weight loss 573 (96.0) 181 (95.3) 392 (96.3) 0.54 Night sweats 566 (94.8) 184 (96.8) 382 (93.9) 0.13 Hemoptysis 197 (33.0) 74 (38.9) 123 (30.2) 0.035 TB score, median (IQR) 5 (4–6) 5 (4–6) 5 (4–6) 0.22 TB score (0–5) 372 (62.3) 115 (60.5) 257 (63.1) TB score (6–12) 225 (37.7) 75 (39.5) 150 (36.9) TB treatment categories 0.40 Retreatment 14 (2.3) 3 (1.6) 11 (2.7) New patients 583 (97.7) 187 (98.4) 396 (97.3) Blood parameters c Hemoglobin level 11.3 (9.9–12.7) 11.7 (10.1–13) 11.3 (9.8–12.5) 0.044 Eosinophil, cells per μld 0.15 (0.06–0.32) 0.2 (0.1–0.4) 0.1 (0.05–0.2) 0.003 AFB id f t b illi BMI b d i d HIV h i d ﬁ i IQR i t til MUAC id i f US$ AFB, acid-fast bacilli; BMI, body mass index; HIV, human immunodeﬁency virus; IQR, interquartile range; MUAC, mid-upper arm circumference; US$, United States dollars (1 US$ = 2,190 Tanzanian Shillings in March 2016) Helminth infection occupation risk (working in rice ﬁelds, sand harvesting, washing cars, and ﬁshing) a TB patient co-infected with HIV and have CD4+ count values (n = 80) b Fisher’s exact test c“Symptoms”, and “blood parameters”: categories not mutually exclusive d TB patients with an available full blood count result (n = 322) d i 10 1371/j l d 0005342 005 macrophage infection models [9,13,15]. In line with our findings, a recent study with human peripheral mononuclear cells exposed to M. tuberculosis and S. mansoni antigens showed that S. mansoni-induced CD4+ T cells disrupt the control of M. tuberculosis in infected macro- phages [9]. Several studies in humans suggested that helminth infections may increase the risk for pro- gression of latent M. tuberculosis infection to active TB [15,21,37] as well as for exacerbating the disease [15]. However, the results of these studies are conflicting, and no differentiation at the helminth species level was made in these analyses. Indeed, the hypothesis of a helminth species-specific impact on the host response is supported by a recent systematic review, which revealed a trend toward an association between a decrease in HIV viral loads and treatment for S. mansoni, but not for other helminth species [38]. Patient characteristics of TB patients infected and not infected with helminths at the time of TB dia PLOS Neglected Tropical Diseases \| DOI:10.1371/journal.pntd.0005342 February 8, 2017 PLOS Neglected Tropical Diseases \| DOI:10.1371/journal.pntd.0005342 February 8, 2017 12 / 19 Epidemiology of TB and helminth co-infection Table 5. (Continued) Characteristics Total TB and helminth TB only p-value (n = 597) (n = 190) (n = 407) Weight loss 573 (96.0) 181 (95.3) 392 (96.3) 0.54 Night sweats 566 (94.8) 184 (96.8) 382 (93.9) 0.13 Hemoptysis 197 (33.0) 74 (38.9) 123 (30.2) 0.035 TB score, median (IQR) 5 (4–6) 5 (4–6) 5 (4–6) 0.22 TB score (0–5) 372 (62.3) 115 (60.5) 257 (63.1) TB score (6–12) 225 (37.7) 75 (39.5) 150 (36.9) TB treatment categories 0.40 Retreatment 14 (2.3) 3 (1.6) 11 (2.7) New patients 583 (97.7) 187 (98.4) 396 (97.3) Blood parameters c Hemoglobin level 11.3 (9.9–12.7) 11.7 (10.1–13) 11.3 (9.8–12.5) 0.044 Eosinophil, cells per μld 0.15 (0.06–0.32) 0.2 (0.1–0.4) 0.1 (0.05–0.2) 0.003 AFB, acid-fast bacilli; BMI, body mass index; HIV, human immunodeﬁency virus; IQR, interquartile range; MUAC, mid-upper arm circumference; US$, United States dollars (1 US$ = 2,190 Tanzanian Shillings in March 2016) Helminth infection occupation risk (working in rice ﬁelds, sand harvesting, washing cars, and ﬁshing) a TB patient co-infected with HIV and have CD4+ count values (n = 80) b Fisher’s exact test c“Symptoms”, and “blood parameters”: categories not mutually exclusive d TB patients with an available full blood count result (n = 322) doi:10 1371/journal pntd 0005342 t005 Table 5. PLOS Neglected Tropical Diseases \| DOI:10.1371/journal.pntd.0005342 February 8, 2017 This is in line with our findings that TB patients co-infected with S. mansoni tended to present less frequently with lung cavitations compared with S. mansoni-negative TB patients. Any hel- minth co-infection did not appear to have an effect on clinical outcomes during follow-up. We found no evidence for an effect of helminth co-infection on the gain in the percentage of body fat and BMI after 6 months (e.g., at the time of completed TB treatment). This might be explained by the fact that the administration of anthelmintic treatment offered to the study participants after diagnosis might have reversed the Th1 immune response [15], and thus attenuated the effect of helminth infections on clinical outcomes. However, the effect of a reversal of the Th1 immune response could be minimal as the anthelmintic drugs target the worms [42], which are less immunogenic compared with deposited S. mansoni eggs [9]. loads in the sputum compared with HIV-negative patients [40,41]. Hence, the helminth- induced Th1 immunological impairment might have an effect on the sputum bacterial load. Moreover, TB patients with an impaired host immune system rarely present with lung cavita- tion resulting in fewer M. tuberculosis bacilli being expectorated in the sputum [40,41]. This is in line with our findings that TB patients co-infected with S. mansoni tended to present less frequently with lung cavitations compared with S. mansoni-negative TB patients. Any hel- minth co-infection did not appear to have an effect on clinical outcomes during follow-up. We found no evidence for an effect of helminth co-infection on the gain in the percentage of body fat and BMI after 6 months (e.g., at the time of completed TB treatment). This might be explained by the fact that the administration of anthelmintic treatment offered to the study participants after diagnosis might have reversed the Th1 immune response [15], and thus attenuated the effect of helminth infections on clinical outcomes. However, the effect of a reversal of the Th1 immune response could be minimal as the anthelmintic drugs target the worms [42], which are less immunogenic compared with deposited S. mansoni eggs [9]. We found that TB patients had a higher crude prevalence of helminth infections, as com- pared with household contact controls. The higher prevalence of helminth infections among TB patients could be the result of the pathogenic role of helminth infection in the progression from M. tuberculosis infection to active TB. Epidemiology of TB and helminth co-infection Table 6. Effect of helminth infection on the clinical severity and clinical presentation in TB patients at the time of TB diagnosis. Helminth infection Severe TB scorea High sputum bacterial loadb Lung inﬁltration Lung cavitation aOR (95% CI) p-value aOR (95% CI) p-value aOR (95% CI) p-value aOR (95% CI) p-value Any helminth infection 0.55 0.12 0.42 0.82 No 1.00 1.00 1.00 1.00 Yes 1.12 (0.78–1.61) 0.75 (0.51–1.08) 0.82 (0.50–1.33) 0.95 (0.60–1.50) Strongyloides stercoralisc 0.44 0.39 0.17 0.76 No 1.00 1.00 1.00 1.00 Yes 1.19 (0.76–1.86) 0.82 (0.52–1.29) 1.56 (0.83–2.92) 1.09 (0.62–1.91) Schistosoma mansoni d 0.75 0.004 0.15 0.088 No 1.00 1.00 1.00 1.00 Yes 0.89 (0.45–1.78) 0.37 (0.19–0.72) 0.51 (0.21–1.27) 0.41 (0.12–1.16) Hookworme 0.55 0.40 0.086 0.54 No 1.00 1.00 1.00 1.00 Yes 1.20 (0.67–2.15) 0.77 (0.42–1.42) 0.51 (0.23–1.10) 0.79 (0.37–1.69) Multiple infections 0.82 0.020 0.19 0.40 None 1.00 1.00 1.00 1.00 Mono 1.13 (0.77–1.67) 0.88 (0.59–1.31) 0.85 (0.51–1.43) 1.06 (0.66–1.72) Double or more 1.04 (0.48–2.22) 0.34 (0.16–0.73) 0.67 (0.24–1.84) 0.50 (0.17–1.44) Logistic regression model adjusted for age, sex, HIV infection, and smoking status. a TB score (mild [score of 1–5] and severe [score of 6–12]) b Sputum bacterial load (according to qualitative AFB smear microscopy grading): mild (scanty and 1+) and severe (+2) c 79 TB i i h h l i h i f i h h S li l d d on the clinical severity and clinical presentation in TB patients at the time of TB diagnosis. Logistic regression model adjusted for age, sex, HIV infection, and smoking status. Logistic regression model adjusted for age, sex, HIV infection, and smoking status. a TB score (mild [score of 1–5] and severe [score of 6–12]) b Sputum bacterial load (according to qualitative AFB smear microscopy grading): mild (scanty and 1+) and severe (+2) c 79 TB patients with any helminth infection other than S. stercoralis were excluded d 150 TB patients with helminth co-infection other than S. mansoni were excluded e 72 TB patients with helminth co-infection other than hookworm were excluded doi:10.1371/journal.pntd.0005342.t006 loads in the sputum compared with HIV-negative patients [40,41]. Hence, the helminth- induced Th1 immunological impairment might have an effect on the sputum bacterial load. Moreover, TB patients with an impaired host immune system rarely present with lung cavita- tion resulting in fewer M. tuberculosis bacilli being expectorated in the sputum [40,41]. A case-control study from Ethiopia also found an association between TB and helminth infections, and the association was stron- ger in patients that were infected with multiple helminth species [21]. The small number of study participants with S. mansoni infection (31 among TB cases, nine among controls) may have masked an association between TB and schistosomiasis in that study [21]. In contrast, a cohort study from India showed no difference in TB incidence rates in helminth-infected and helminth-free individuals after 2.5 years of follow-up [39]. y p We also found that S. mansoni, but not other helminth species, was associated with the clin- ical presentation among TB patients. Patients co-infected with S. mansoni had lower sputum bacterial loads at the time of TB diagnosis than S. mansoni-negative TB patients. Similarly, a study in Ethiopia observed lower sputum bacterial loads at TB diagnosis in TB patients co- infected with any helminth species [40]. Interestingly, our observation in TB patients co- infected with S. mansoni resembles the paucibacillary disease in HIV-positive individuals with severe immunosuppression, who frequently have negative or low bacterial M. tuberculosis 13 / 19 PLOS Neglected Tropical Diseases \| DOI:10.1371/journal.pntd.0005342 February 8, 2017 PLOS Neglected Tropical Diseases \| DOI:10.1371/journal.pntd.0005342 February 8, 2017 Epidemiology of TB and helminth co-infection helminth infection in our study was 32% and lower compared with the 71% observed in the latter study [21]. It is conceivable that the high proportion of self-reported previous use of anthelmintic drugs in our study (approximately 80%) could have reduced the overall preva- lence of helminth infection. Hence, we may have underestimated the effects of helminth infec- tion seen in our study. We also found that occupation exposing people to regular water contacts (for instance rice field workers, sand harvesters, car washers, and fishermen) were associated with helminth infections. Being exposed to freshwater bodies and being involved in water-related activities have previously been reported to increase the risk of helminth infections [44]. In the current study, HIV-positive individuals were less likely to be co-infected with helminths. A lower prev- alence of helminth infections in HIV-positive patients has also been reported in a study con- ducted in Mwanza in northern Tanzania, which is a highly endemic area for helminthiases [45]. Of note, current clinical practice in Tanzania is to treat any helminth infection in HIV- positive patients at enrolment into HIV care and in case of clinical suspicion of helminth infec- tion during follow-up, as specified in the HIV/AIDS management guideline [17]. The use of anthelmintic drugs is safe and might be beneficial in HIV-positive patients by possibly reduc- ing the HIV-RNA viral load and subsequently improving clinical outcomes [46]. Furthermore, cotrimoxazole preventive therapy (CPT), which is recommended for HIV-positive patients, has also been reported to have limited anthelmintic properties [43,47]. This might explain the lower prevalence of helminth infection among HIV-positive individuals in our study [17]. Our research has several strengths and limitations that warrant consideration. An impor- tant strength of our study is the large sample size and the recruitment of both TB patients and household contact controls with similar socioeconomic profiles and exposure patterns to both TB and helminth infection. Our findings may well apply to other settings with a similar preva- lence of TB, HIV, and helminth infections in sub-Saharan Africa. Furthermore, we used rec- ommended TB diagnostics and a suite of standardized, quality-controlled helminth diagnostics, which have comparable diagnostic performance to resource-intensive molecular test assays [25]. Study limitations include the following. First, this is an observational study which cannot establish a causal relationship between helminth infections and TB disease. PLOS Neglected Tropical Diseases \| DOI:10.1371/journal.pntd.0005342 February 8, 2017 The higher prevalence of helminth co-infection in TB patients has also been noted in other studies from different settings [9,43]. For example, a study conducted in Ethiopia reported a higher prevalence of helminth infection among TB patients as compared with household contact controls [21]. Overall, the prevalence of 14 / 19 PLOS Neglected Tropical Diseases \| DOI:10.1371/journal.pntd.0005342 February 8, 2017 Epidemiology of TB and helminth co-infection compare helminth-induced immune regulation by different helminth species. Prospective cohort studies that evaluate the effect of preventive anthelmintic chemotherapy on the inci- dence of M. tuberculosis infection and active TB could further help to understand the interac- tion between these diseases at the population level. Helminthiasis control measures, in combination with traditional TB control strategies, could potentially contribute to the global efforts to reduce TB incidence by 80% until 2030, as stipulated in WHO’s ambitious End TB Strategy [5]. Supporting information S1 Table. Socio-demographic and clinical characteristics of TB patients and household contact controls, stratified by HIV infection status. (DOCX) S2 Table. Frequency distribution of helminth infections among TB patients and household controls without TB, stratified by HIV status. (DOCX) S3 Table. Frequency distribution and intensity of helminth infection in TB patients and household contact controls, as determined by the Kato-Katz method (triplicate slides). (DOCX) S4 Table. Full blood count and hematological parameters in TB patients, stratified by hel- minth infection status. (DOCX) S5 Table. Additional analysis: Risk factors for any helminth infection among TB patients only. (DOCX) S6 Table. Additional analysis: Risk factors for any helminth infection among household controls without TB only. (DOCX) S7 Table. Full table with unadjusted and adjusted odds ratios. Associations of TB disease with helminth infection and other patient characteristics comparing TB patients and house- hold contact controls without TB. (DOCX) S8 Table. Associations of TB disease with Strongyloides stercoralis and hookworm infection (unadjusted and adjusted odds ratios). (DOCX) S9 Table. Additional analysis: Helminth infection and patient characteristics associated with TB among TB patients and household contact controls, using conditional logistic regression. (DOCX) S10 Table. Radiological findings of chest X-rays in TB patients at the time of TB diagnosis, stratified by helminth infection status. (DOCX) S11 Table. Association of helminth infection with poor recovery of BMI, poor gain of abso- lute weight, and percentage body fat in TB patients, between recruitment and after 6 Second, we could not fully verify whether or not the household contact controls were latently infected with M. tuberculosis, which is a prerequisite to develop TB. However, because Dar es Salaam is a high-burden setting for TB with considerable risk of transmission, and because living with a TB patient is a strong risk factor for TB [35], it is reasonable to assume that the controls have previously been exposed and infected with M. tuberculosis. Third, we did not check the hel- minth infection status for TB patients during and after completion of TB treatment, which could influence the clinical outcomes. However, we do not expect a high helminth re-infection rate after 6 months in our study area [48]. Fourth, we did not use molecular diagnostics such as polymerase chain reaction (PCR) which might have identified some more cases, but one of our previous studies revealed that also PCR approaches miss in particular very light intensity infections. Moreover, its performance and sensitivity vary with the helminth species under examination [25]. Hence, also a PCR cannot be considered as the diagnostic gold-standard. In conclusion, co-infection with S. mansoni, but not other helminth species, was found to be an independent risk factor for active TB in our study and was associated with the clinical presentation in TB patients. These findings suggest a role for S. mansoni, or helminth infection in general, in immunomodulation of human TB. Treatment of helminth infections should be considered in the clinical management of TB patients, and helminthiasis control/elimination through preventive chemotherapy might prove to be useful as an additional component of TB control programs. Further research is needed to establish the underlying mechanisms, and PLOS Neglected Tropical Diseases \| DOI:10.1371/journal.pntd.0005342 February 8, 2017 15 / 19 S4 Table. Full blood count and hematological parameters in TB patients, stratified by hel- minth infection status. (DOCX) S4 Table. Full blood count and hematological parameters in TB patients, stratified by hel- minth infection status. (DOCX) S5 Table. Additional analysis: Risk factors for any helminth infection among TB patients only. (DOCX) S5 Table. Additional analysis: Risk factors for any helminth infection among TB patients only. (DOCX) Supporting information S1 Table. Socio-demographic and clinical characteristics of TB patients and household contact controls, stratified by HIV infection status. (DOCX) S2 Table. Frequency distribution of helminth infections among TB patients and household controls without TB, stratified by HIV status. (DOCX) S3 Table. Frequency distribution and intensity of helminth infection in TB patients and household contact controls, as determined by the Kato-Katz method (triplicate slides). (DOCX) Epidemiology of TB and helminth co-infection months of completed TB treatment. (DOCX) S1 Fig. Geographic distribution of helminth infection in the study area. (A) The prevalence of helminth infection summarized at the ward level. (B) The helminth species distribution at the study area. Other helminth infections include: Ascaris lumbricoides, Enterobius vermicu- laris, Trichuris trichiura, and Hymenolepis diminuta. (DOCX) S1 Fig. Geographic distribution of helminth infection in the study area. (A) The prevalence of helminth infection summarized at the ward level. (B) The helminth species distribution at the study area. Other helminth infections include: Ascaris lumbricoides, Enterobius vermicu- laris, Trichuris trichiura, and Hymenolepis diminuta. (DOCX) Acknowledgments We thank all the study participants whose data were used in this study. We are grateful for the assistance and support from the office of the Temeke District Medical Officer, the Temeke dis- trict hospital staff, and the National TB Program and the District TB coordinators of the Temeke district, Dar es Salaam, Tanzania. We also thank the study team at the Temeke district hospital for recruiting patients. PLOS Neglected Tropical Diseases \| DOI:10.1371/journal.pntd.0005342 February 8, 2017 S6 Table. Additional analysis: Risk factors for any helminth infection among household controls without TB only. (DOCX) S6 Table. Additional analysis: Risk factors for any helminth infection among household controls without TB only. (DOCX) S7 Table. Full table with unadjusted and adjusted odds ratios. Associations of TB disease with helminth infection and other patient characteristics comparing TB patients and house- hold contact controls without TB. (DOCX) S8 Table. Associations of TB disease with Strongyloides stercoralis and hookworm infection (unadjusted and adjusted odds ratios). (DOCX) S9 Table. Additional analysis: Helminth infection and patient characteristics associated with TB among TB patients and household contact controls, using conditional logistic regression. (DOCX) S10 Table. Radiological findings of chest X-rays in TB patients at the time of TB diagnosis, stratified by helminth infection status. (DOCX) S11 Table. Association of helminth infection with poor recovery of BMI, poor gain of abso- lute weight, and percentage body fat in TB patients, between recruitment and after 6 S11 Table. Association of helminth infection with poor recovery of BMI, poor gain of abso- lute weight, and percentage body fat in TB patients, between recruitment and after 6 16 / 19 PLOS Neglected Tropical Diseases \| DOI:10.1371/journal.pntd.0005342 February 8, 2017 Conceptualization: FM JH KS SK JU SG LF. Conceptualization: FM JH KS SK JU SG LF. Formal analysis: FM JH KS GM CS LF. Funding acquisition: FM SG LF. Investigation: FM JH KS LK MS TM MC GM. Methodology: FM JH KS SK GM LF. Project administration: FM JH KS SK GM LF. Supervision: FM SG LF. Formal analysis: FM JH KS GM CS LF. Funding acquisition: FM SG LF. Investigation: FM JH KS LK MS TM MC GM. Methodology: FM JH KS SK GM LF. Project administration: FM JH KS SK GM LF. Supervision: FM SG LF. Formal analysis: FM JH KS GM CS LF. Supervision: FM SG LF. Visualization: FM JH SK LF. Writing – original draft: FM. Writing – review & editing: FM JH KS LK MS TM MC GM CS KR SK JU SG LF. Epidemiology of TB and helminth co-infection 7. Lo¨nnroth K, Jaramillo E, Williams BG, Dye C, Raviglione M. Drivers of tuberculosis epidemics: the role of risk factors and social determinants. Soc Sci Med. 2009; 68: 2240–2466. doi: 10.1016/j.socscimed. 2009.03.041 PMID: 19394122 8. NTLP, MoHSW. Manual for the management of tuberculosis and leprosy. 6th ed. Dar es Salaam: Min- istry of Health and Social Welfare, Dar es Salaam; 2013. 9. DiNardo AR, Mace EM, Lesteberg K, Cirillo JD, Mandalakas AM, Graviss EA, et al. Schistosome solu- ble egg antigen decreases Mycobacterium tuberculosis-specific CD4+ T-cell effector function with con- comitant arrest of macrophage phago-lysosome maturation. J Infect Dis. 2016; 214: 479–488. doi: 10. 1093/infdis/jiw156 PMID: 27389351 10. Potian JA, Rafi W, Bhatt K, McBride A, Gause WC, Salgame P. Preexisting helminth infection induces inhibition of innate pulmonary anti-tuberculosis defense by engaging the IL-4 receptor pathway. J Exp Med. 2011; 208: 1863–1874. doi: 10.1084/jem.20091473 PMID: 21825018 11. Rafi W, Ribeiro-Rodrigues R, Ellner JJ, Salgame P. Coinfection-helminthes and tuberculosis. Curr Opin HIV AIDS. 2012; 7: 239–244. doi: 10.1097/COH.0b013e3283524dc5 PMID: 22411453 12. Salgame P, Yap GS, Gause WC. Effect of helminth-induced immunity on infections with microbial path- ogens. Nat Immunol. 2013; 14: 1118–1126. doi: 10.1038/ni.2736 PMID: 24145791 13. Babu S, Nutman TB. Helminth-tuberculosis co-infection: an immunologic perspective. Trends Immunol. 2016; 37: 597–607. doi: 10.1016/j.it.2016.07.005 PMID: 27501916 14. Mishra PK, Palma M, Bleich D, Loke P, Gause WC. Systemic impact of intestinal helminth infections. Mucosal Immunol. 2014; 7: 753–762. doi: 10.1038/mi.2014.23 PMID: 24736234 15. Monin L, Griffiths KL, Lam WY, Gopal R, Kang DD, Ahmed M, et al. Helminth-induced arginase-1 exac- erbates lung inflammation and disease severity in tuberculosis. J Clin Invest. 2015; 125: 4699–4713. doi: 10.1172/JCI77378 PMID: 26571397 16. MoHSW. Standard treatment guidelines and essential medicines list. 4th ed. Dar es Salaam: Ministry of Health and Social Welfare; 2013. 17. NACP. National guidelines for the management of HIV and AIDS. 5th ed. Dar es Salaam: Ministry of Health and Social Welfare; 2015. 18. NTLP, MoHSW. National Tuberculosis and Leprosy 2014 Annual Report. Dar es Salaam: Ministry of Health and Social Welfare; 2015. 19. PMORALG. Dar es Salaam region socio-economic profile. Dar es Salaam: Prime Minister’s Office, Regional Administration and Local Government; 2014. 20. Mhimbira F, Hella J, Maroa T, Kisandu S, Chiryamkubi M, Said K, et al. Home-based and facility-based directly observed therapy of tuberculosis treatment under programmatic conditions in urban Tanzania. PLoS One. References References 1. WHO. Global tuberculosis report 2015. Geneva: World Health Organization; 2015. 2. Knopp S, Steinmann P, Keiser J, Utzinger J. Nematode infections: soil-transmitted helminths and Trichi- nella. Infect Dis Clin North Am. 2012; 26: 341–358. doi: 10.1016/j.idc.2012.02.006 PMID: 22632643 3. Pullan RL, Smith JL, Jasrasaria R, Brooker SJ. Global numbers of infection and disease burden of soil transmitted helminth infections in 2010. Parasit Vectors. 2014; 7:37. doi: 10.1186/1756-3305-7-37 PMID: 24447578 4. Utzinger J, Becker S, Knopp S, Blum J, Neumayr A, Keiser J, et al. Neglected tropical diseases: diagno- sis, clinical management, treatment and control. Swiss Med Wkly. 2012; 142:w13727. doi: 10.4414/ smw.2012.13727 PMID: 23180107 5. WHO. The end TB strategy. Geneva: World Health Organization; 2015. 6. Simon GG. Impacts of neglected tropical disease on incidence and progression of HIV/AIDS, tuberculo- sis, and malaria: scientific links. Int J Infect Dis. 2016; 42: 54–57. doi: 10.1016/j.ijid.2015.11.006 PMID: 26594012 1. WHO. Global tuberculosis report 2015. Geneva: World Health Organization; 2015. 1. WHO. Global tuberculosis report 2015. Geneva: World Health Organization; 2015. 2. Knopp S, Steinmann P, Keiser J, Utzinger J. Nematode infections: soil-transmitted helminths and Trichi- nella. Infect Dis Clin North Am. 2012; 26: 341–358. doi: 10.1016/j.idc.2012.02.006 PMID: 22632643 3. Pullan RL, Smith JL, Jasrasaria R, Brooker SJ. Global numbers of infection and disease burden of soil transmitted helminth infections in 2010. Parasit Vectors. 2014; 7:37. doi: 10.1186/1756-3305-7-37 PMID: 24447578 4. Utzinger J, Becker S, Knopp S, Blum J, Neumayr A, Keiser J, et al. Neglected tropical diseases: diagno- sis, clinical management, treatment and control. Swiss Med Wkly. 2012; 142:w13727. doi: 10.4414/ smw.2012.13727 PMID: 23180107 4. Utzinger J, Becker S, Knopp S, Blum J, Neumayr A, Keiser J, et al. Neglected tropical diseases: diagno- sis, clinical management, treatment and control. Swiss Med Wkly. 2012; 142:w13727. doi: 10.4414/ smw.2012.13727 PMID: 23180107 5. WHO. The end TB strategy. Geneva: World Health Organization; 2015. 5. WHO. The end TB strategy. Geneva: World Health Organization; 2015. 5. WHO. The end TB strategy. Geneva: World Health Organization; 2015. 6. Simon GG. Impacts of neglected tropical disease on incidence and progression of HIV/AIDS, tuberculo- sis, and malaria: scientific links. Int J Infect Dis. 2016; 42: 54–57. doi: 10.1016/j.ijid.2015.11.006 PMID: 26594012 PLOS Neglected Tropical Diseases \| DOI:10.1371/journal.pntd.0005342 February 8, 2017 17 / 19 PLOS Neglected Tropical Diseases \| DOI:10.1371/journal.pntd.0005342 February 8, 2017 Epidemiology of TB and helminth co-infection 29. Colley DG, Binder S, Campbell C, King CH, Tchuem Tchuente´ L-A, N’Goran EK, et al. A five-country evaluation of a point-of-care circulating cathodic antigen urine assay for the prevalence of Schistosoma mansoni. Am J Trop Med Hyg. 2013; 88: 426–432. doi: 10.4269/ajtmh.12-0639 PMID: 23339198 30. Steiner A, Hella J, Gru¨ninger S, Mhalu G, Mhimbira F, Cercamondi CI, et al. Managing research and surveillance projects in real-time with a novel open-source eManagement tool designed for under- resourced countries. J Am Med Inform Assoc. 2016; ocv185. 31. Wejse C, Gustafson P, Nielsen J, Gomes VF, Aaby P, Andersen PL, et al. TB score: signs and symp- toms from tuberculosis patients in a low-resource setting have predictive value and may be used to assess clinical course. Scand J Infect Dis. 2008; 40: 111–120. doi: 10.1080/00365540701558698 PMID: 17852907 32. Blakemore R, Nabeta P, Davidow AL, Vadwai V, Tahirli R, Munsamy V, et al. A multisite assessment of the quantitative capabilities of the Xpert MTB/RIF assay. Am J Respir Crit Care Med. 2011; 184: 1076– 1084. doi: 10.1164/rccm.201103-0536OC PMID: 21836139 33. WHO. Helminth control in school-age children: a guide for managers of control programmes. Geneva: World Health Organization; 2011. 34. Tanzania: shapefiles for EAs, villages, districts and regions. In: openmicrodata.org [Internet]. [cited 3 Apr 2016]. https://openmicrodata.wordpress.com/2010/12/16/tanzania-shapefiles-for-eas-villages- districts-and-regions/ 35. Rieder HL. Epidemiologic basis of tuberculosis control. 1st ed. Paris: International Union Against Tuberculosis and Lung Disease; 1999. 36. Resende Co T, Hirsch CS, Toossi Z, Dietze R, Ribeiro-Rodrigues R. Intestinal helminth co-infection has a negative impact on both anti-Mycobacterium tuberculosis immunity and clinical response to tuberculo- sis therapy. Clin Exp Immunol. 2006; 147: 45–52. 37. Tristão-Sa´ R, Ribeiro-Rodrigues R, Johnson LT, Pereira FEL, Dietze R. Intestinal nematodes and pul- monary tuberculosis. Rev Soc Bras Med Trop. 2002; 35: 533–535. PMID: 12621678 38. Sangare´ LR, Herrin BR, John-Stewart G, Walson JL. Species-specific treatment effects of helminth/ HIV-1 co-infection: a systematic review and meta-analysis. Parasitology. 2011; 138: 1546–1558. doi: 10.1017/S0031182011000357 PMID: 21729353 39. Chatterjee S, Kolappan C, Subramani R, Gopi PG, Chandrasekaran V, Fay MP, et al. Incidence of active pulmonary tuberculosis in patients with coincident filarial and/or intestinal helminth infections fol- lowed longitudinally in South India. PLoS One. 2014; 9: e94603. doi: 10.1371/journal.pone.0094603 PMID: 24728010 40. Abate E, Belayneh M, Idh J, Diro E, Elias D, Britton S, et al. 2016; 11: e0161171. doi: 10.1371/journal.pone.0161171 PMID: 27513331 21. Elias D, Mengistu G, Akuffo H, Britton S. Are intestinal helminths risk factors for developing active tuber- culosis? Trop Med Int Health. 2006; 11: 551–558. doi: 10.1111/j.1365-3156.2006.01578.x PMID: 16553939 22. Baty International (BI). The Harpenden skinfold caliper HSB-BI. Baty International, West Sussex; 2007. 23. Durnin JVGA, Womersley J. Body fat assessed from total body density and its estimation from skinfold thickness: measurements on 481 men and women aged from 16 to 72 years. Br J Nutr. 1974; 32: 77– 97. PMID: 4843734 24. WHO. The handbook: laboratory diagnosis of tuberculosis by sputum microscopy. Global ed. Adelaide: Stop TB Partnership-Global Laboratory Initiative Working Group; 2013. 25. Knopp S, Salim N, Schindler T, Karagiannis Voules DA, Rothen J, Lweno O, et al. Diagnostic accuracy of Kato–Katz, FLOTAC, Baermann, and PCR methods for the detection of light-intensity hookworm and Strongyloides stercoralis infections in Tanzania. Am J Trop Med Hyg. 2014; 90: 535–545. doi: 10.4269/ ajtmh.13-0268 PMID: 24445211 26. Salim N, Knopp S, Lweno O, Abdul U, Mohamed A, Schindler T, et al. Distribution and risk factors for Plasmodium and helminth co-infections: a cross-sectional survey among children in Bagamoyo district, Coastal Region of Tanzania. PLoS Negl Trop Dis. 2015; 9: e0003660. doi: 10.1371/journal.pntd. 0003660 PMID: 25837022 27. Salim N, Schindler T, Abdul U, Rothen J, Genton B, Lweno O, et al. Enterobiasis and strongyloidiasis and associated co-infections and morbidity markers in infants, preschool- and school-aged children from rural coastal Tanzania: a cross-sectional study. BMC Infect Dis. 2014; 14: 644. doi: 10.1186/ s12879-014-0644-7 PMID: 25486986 28. Garcia LS, Bruckner DA (2001) Diagnostic medical Parasitology; Garcia LS, Bruckner DA, eds. Wash- ington D.C.: American Society for Microbiology. pp 1–791. 18 / 19 PLOS Neglected Tropical Diseases \| DOI:10.1371/journal.pntd.0005342 February 8, 2017 PLOS Neglected Tropical Diseases \| DOI:10.1371/journal.pntd.0005342 February 8, 2017 Asymptomatic helminth infection in active tuberculosis is associated with increased regulatory and Th-2 responses and a lower sputum smear positivity. PLoS Negl Trop Dis. 2015; 9: e0003994. doi: 10.1371/journal.pntd.0003994 PMID: 26248316 41. Sharma SK, Mohan A, Kadhiravan T. HIV-TB co-infection: epidemiology, diagnosis and management. Indian J Med Res. 2005; 121: 550–567. PMID: 15817963 42. Martin RJ, Puttachary S, Buxton SK, Verma S, Robertson AP. The Conqueror Worm: recent advances with cholinergic anthelmintics and techniques excite research for better therapeutic drugs. J Helminthol. 2015; 89: 387–397. doi: 10.1017/S0022149X1400039X PMID: 24871674 43. Abate E, Belayneh M, Gelaw A, Idh J, Getachew A, Alemu S, et al. The impact of asymptomatic hel- minth co-infection in patients with newly diagnosed tuberculosis in North-West Ethiopia. PLoS One. 2012; 7: e42901. doi: 10.1371/journal.pone.0042901 PMID: 22952620 44. Woodburn PW, Muhangi L, Hillier S, Ndibazza J, Namujju PB, Kizza M, et al. Risk factors for helminth, malaria, and HIV infection in pregnancy in Entebbe, Uganda. PLoS Negl Trop Dis. 2009; 3: e473. doi: 10.1371/journal.pntd.0000473 PMID: 19564904 45. Range N, Magnussen P, Mugomela A, Malenganisho W, Changalucha J, Temu MM, et al. HIV and par- asitic co-infections in tuberculosis patients: a cross-sectional study in Mwanza, Tanzania. Ann Trop Med Parasitol. 2007; 101: 343–351. doi: 10.1179/136485907X176373 PMID: 17524249 46. Modjarrad K, Vermund SH. Effect of treating co-infections on HIV-1 viral load: a systematic review. Lan- cet Infect Dis. 2010; 10: 455–463. doi: 10.1016/S1473-3099(10)70093-1 PMID: 20610327 47. Janssen S, Hermans S, Knap M, Moekotte A, Rossatanga EG, Adegnika AA, et al. Impact of anti-retro- viral treatment and cotrimoxazole prophylaxis on helminth infections in HIV-Infected patients in Lamba- re´ne´, Gabon. PLoS Negl Trop Dis. 2015; 9: e0003769. doi: 10.1371/journal.pntd.0003769 PMID: 25993501 48. Tukahebwa EM, Vennervald BJ, Nuwaha F, Kabatereine NB, Magnussen P. Comparative efficacy of one versus two doses of praziquantel on cure rate of Schistosoma mansoni infection and re-infection in Mayuge District, Uganda. Trans R Soc Trop Med Hyg. 2013; 107: 397–404. doi: 10.1093/trstmh/trt024 PMID: 23596262 PLOS Neglected Tropical Diseases \| DOI:10.1371/journal.pntd.0005342 February 8, 2017 19 / 19
https://openalex.org/W3176256411	https://www.scielo.br/j/qn/a/VJ58TBjHVqDZsvWLckcFbTQ/?lang=pt&format=pdf	Portuguese	null	MICROPLÁSTICOS: OCORRÊNCIA AMBIENTAL E DESAFIOS ANALÍTICOS	Química Nova	2,021	cc-by	27,299	INTRODUÇÃO Os MP são onipresentes no ambiente e são classificados atualmente como contaminantes emergentes.10,11 Seus potenciais riscos aos seres vivos, bem como os níveis da contaminação em diferentes compartimentos ambientais, precisam ser mais bem elucidados. Por isso, essa nova classe tem sido foco de pesquisas no mundo todo. A poluição plástica é atualmente um problema de grande relevância sob o ponto de vista ambiental e socioeconômico, consequência principalmente da má gestão dos resíduos sólidos, dentre outros fatores. Uma ramificação da questão é dada pela poluição causada por plásticos cujos tamanhos estejam nas escalas micrométrica e milimétrica. O primeiro relato envolvendo a presença dessas partículas plásticas no ambiente foi publicado em 1972, quando esse material foi encontrado retido em redes de plâncton, em águas costeiras na região da Nova Inglaterra, nos Estados Unidos.1 No mesmo ano, partículas de polietileno e poliestireno também foram identificadas no litoral do Rio Grande do Sul, no Brasil.2 No entanto, a comunidade científica ainda não explorava a problemática e poucos trabalhos foram publicados ao longo da década de 70. Com o aumento da preocupação sobre as consequências ecológicas da presença de tais partículas em diferentes ecossistemas, os estudos envolvendo os resíduos plásticos foram incorporando novos conceitos e o termo microplástico (MP) foi introduzido em 2004.3-5 O objetivo deste trabalho de revisão é destacar aspectos importantes da temática da poluição por MP, do ponto de vista da química ambiental, abordando sobre a presença deles no ambiente e discutindo sobre os métodos e os desafios analíticos para a determinação de MP em matrizes ambientais. Além disso, o trabalho também tem por meta apresentar o cenário das pesquisas no Brasil. A primeira parte desta revisão contextualiza o problema em nível mundial e apresenta uma discussão sobre as fontes e algumas consequências dos MP no ambiente. A segunda parte discute as etapas de controle de qualidade, amostragem, preparo de amostras e caracterização, que têm sido empregadas mundialmente para a determinação de MP nos compartimentos ambientais solo/ sedimento, água e ar. Finalmente, a terceira parte apresenta, com base nos trabalhos publicados em literatura indexada, o cenário de ocorrência de MP no Brasil, destacando os métodos analíticos que foram empregados nas pesquisas. Várias definições são encontradas para o termo MP, de acordo com a faixa de tamanho das partículas,6,7 sendo a mais utilizada a que se refere a esses materiais como partículas de polímeros orgânicos sintéticos com tamanho inferior a 5 mm. MICROPLÁSTICOS: OCORRÊNCIA AMBIENTAL E DESAFIOS ANALÍTICOS MICROPLÁSTICOS: OCORRÊNCIA AMBIENTAL E DESAFIOS ANALÍTICOS Revisão Revisão Revisão Cassiana C. Montagnera,, , Mariana Amaral Diasa, Eduardo Maia Paivaa e Cristiane Vidala aDepartamento de Química Analítica, Instituto de Química, Universidade Estadual de Campinas, 13083-97 Cassiana C. Montagnera,, , Mariana Amaral Diasa, Eduardo Maia Paivaa e Cristiane Vidala aDepartamento de Química Analítica, Instituto de Química, Universidade Estadual de Campinas, 13083-970 Campinas – SP, Brasi Recebido em 27/02/2021; aceito em 08/06/2021; publicado na web em 29/06/2021 Cassiana C. Montagnera,, , Mariana Amaral Diasa, Eduardo Maia Paivaa e Cristiane Vidala aDepartamento de QuímicaAnalítica, Instituto de Química, Universidade Estadual de Campinas, 13083 Cassiana C. Montagnera,, , Mariana Amaral Diasa, Eduardo Maia Paivaa e Cristiane Vidala Recebido em 27/02/2021; aceito em 08/06/2021; publicado na web em 29/06/2021 MICROPLASTICS: ENVIRONMENTAL OCCURRENCE AND ANALYTICAL CHALLENGES. Plastic pollution is a major societal, economic, and environmental issue. Upon release into the environment, plastics are altered by biological and physicochemical processes that influence their fate and transport within ecosystems. Large plastic debris can fragment into smaller pieces and are called microplastics (MPs) when occurring at sizes between 1 µm and 5 mm, fragmented or produced in this range. These small pieces of plastics are ubiquitous anthropogenic contaminants found throughout the world; however, the environmental consequences and effects on biota are not clearly understood. Scientific literature on the environmental implications of MP pollution is expanding rapidly and critical review of this literature is necessary to identify areas of evolving consensus and remaining gaps in knowledge. Herein a comprehensive literature review was performed to assess (i) the sources and distribution of MPs in different environmental compartments, (ii) the analytical methods that have been applied worldwide, and the key analytical challenges that remain in assessing MPs in the environment, and, finally, (iii) to integrate the findings of Brazilian MP research, which showed that since 2004 (80 papers up to 2020) has documented the presence of MPs in aquatic matrices and sediments with focus on marine ecosystems, mainly identified by visual inspection. Keywords: plastic pollution; environmental contaminants; microplastic sampling; microplastic characterization; polymer identification. *e-mail: ccmonta@unicamp.br INTRODUÇÃO Essa definição foi proposta em 2009 pela National Oceanic and Atmospheric Administration (NOAA) e, desde então, a maioria das publicações tem a adotado como referência, inclusive este trabalho.3,8 Após um pouco mais de uma década, em 2020, foi publicada a norma intitulada “Plastics – Environmental Aspects – State of Knowledge and Methodologies” (ISO/TR 21960:2020), em que o termo MP é definido como qualquer partícula plástica sólida insolúvel em água com dimensões entre 1 µm e 1000 µm.9 A norma também define o termo “large microplastic” (microplástico grande, em tradução livre), para a faixa de tamanho de 1 a 5 mm. Deste modo, esta revisão é apresentada de forma a levar aos químicos ambientais e a outros pesquisadores das demais áreas de conhecimento deste tema transdisciplinar, as ferramentas que darão subsídios para políticas públicas e elaboração de projetos nessa área, que ainda é carente de pesquisa no Brasil. Quim. Nova, Vol. 44, No. 10, 1328-1352, 2021 Quim. Nova, Vol. 44, No. 10, 1328-1352, 2021 http://dx.doi.org/10.21577/0100-4042.20170791 A PRODUÇÃO E O DESTINO DOS PLÁSTICOS A energia da ligação química C-H é menor em átomos de carbono terciário do que nos secundários, de forma que o processo de degradação pode ser iniciado mais facilmente em polímeros de cadeias ramificadas quando comparados aos polímeros lineares, os quais possuem majoritariamente átomos de carbono secundário.32 Os processos de degradação resultam em perda da massa molar do polímero, aumento da molhabilidade e da cristalização, bem como a formação de fissuras, gerando os fragmentos de MP.33 A degradação física das partículas leva à geração de diferentes formas de MP, por exemplo, fibras, fragmentos e filmes. Os pellets, por sua vez, permanecem em sua forma física original após a degradação, mas podem ter algumas de suas propriedades alteradas.34,35 No Brasil, os dados da Associação Brasileira da Indústria do Plástico (Abiplast), que monitora os números desse setor, apontam crescimentos anuais na indústria de transformados plásticos.17 Ao mesmo tempo, o país ocupa a quarta posição em geração de resíduo plástico (posicionado após os Estados Unidos, a China e a Índia), com aproximadamente 11 milhões de toneladas desses resíduos sendo gerados por ano, o que equivale a cerca de 11% do total mundial. Além disso, apesar de haver programas de incentivo ao processo de reciclagem, o qual pode afetar positivamente a cadeia socioeconômica do país, somente 1,28% tem essa destinação no Brasil.18 Um aspecto especialmente importante que pode ser atribuído ao baixo índice de reciclagem é o fato de que muitos desses materiais têm sido usados na produção de plásticos chamados de “uso único”, por exemplo os descartáveis, os quais, em sua maioria, são de difícil reciclagem e representam uma parcela significativa da poluição plástica global. MICROPLÁSTICOS NO AMBIENTE Contaminantes de origem antrópica, como bifenilas policloradas (PCB, do inglês Polychlorinated Biphenyls), pesticidas, hidrocarbo- netos policíclicos aromáticos e dioxinas foram identificados sorvi- dos em MP ao redor do mundo.37 O projeto chamado International Pellet Watch, criado em 2006 é uma iniciativa de monitorar, em nível mundial, a quantidade de contaminantes orgânicos associados aos pellets.38 Do Brasil, 21 amostras foram analisadas e concentrações de PCB foram detectadas entre 43 ng g-1 e 3892 ng g-1 de pellet.38 A PRODUÇÃO E O DESTINO DOS PLÁSTICOS Os plásticos são materiais poliméricos sintéticos leves, impermeáveis e duráveis, que podem ser formulados para serem rígidos ou flexíveis, transparentes ou coloridos, e apresentam baixo custo. Por tantas vantagens, são amplamente empregados em todos os setores da sociedade moderna. No entanto, problemas relacionados Microplásticos: ocorrência ambiental e desafios analíticos Vol. 44, No. 10 1329 ao gerenciamento inadequado dos resíduos sólidos fazem com que estes materiais sejam considerados contaminantes onipresentes no ambiente.12,13 sabonetes e cremes dentais. Além disso, são, também, utilizados na indústria farmacêutica, na indústria de plásticos, como matéria prima (pellets), e na indústria química em geral, como abrasivos industriais.25-27 A produção mundial de plásticos, que se iniciou em 1950, cresceu consideravelmente nos últimos 60 anos, com estimativas de que 8,3 bilhões de toneladas de plásticos virgens tenham sido produzidos para as mais diferentes aplicações. Os números atuais mostram que 6,3 bilhões de toneladas de resíduos plásticos foram gerados entre o início da produção, na década de 50, e 2015. Dessa quantidade, 9% foram reciclados, 12% incinerados e 79% foram dispostos em aterros ou no ambiente, demonstrando as deficiências no saneamento e no controle dos resíduos sólidos. Até 2050, estão previstos que cerca de 12 bilhões de toneladas de resíduos plásticos sejam lançados no ambiente, caso a produção atual de plásticos permaneça nesse ritmo acelerado e sem melhoria da gestão de resíduos.14,15 Os MP secundários, por sua vez, são aqueles que resultam da fragmentação dos resíduos plásticos maiores (ex: embalagens) expostos às intempéries no ambiente e outros agentes estressores, degradando-se em fragmentos cada vez menores, até atingir o tamanho dos MP.28,29 A degradação de um polímero está relacionada ao rompimento de ligações químicas covalentes, seja na cadeia principal ou em cadeia lateral. Esse rompimento é consequente da geração de espécies reativas (na maioria dos casos, radicais livres) que são também os responsáveis pela propagação do processo de degradação.30 A iniciação do processo pode ser causada por fotodegradação e/ou degradação física, química e biológica.31 Todas essas formas de iniciação implicam no fornecimento de energia para o rompimento das ligações químicas. Dessa forma, é possível entender porque alguns polímeros se degradam mais facilmente do que outros. Microplásticos como vetores de transporte de outros contaminantes Além da fragmentação, a degradação dos polímeros também favorece a liberação dos aditivos químicos, tais como, estabilizantes, corantes, plastificantes, retardantes de chama, entre outros, os quais são empregados na formulação do polímero para atingir as características necessárias ao uso pretendido no produto final. Durante a degradação, esses compostos podem ser lixiviados para o ambiente através da difusão até a superfície do MP.4,36 Além disso, principalmente em ambientes aquáticos, a via contrária, ou seja, a sorção de outros contaminantes no MP, tem sido objeto de estudos que visam elucidar a ação dos MP como vetores de transporte de outros contaminantes químicos orgânicos utilizados ou não no processamento do polímero. Tais contaminantes, como ftalatos, bisfenol A e éteres difenílicos polibromados, por exemplo, são conhecidos por interferirem no sistema endócrino. Além desses, compostos inorgânicos, como metais também podem se associar aos MP.37 Os impactos do ponto de vista ecológico e ambiental, aliados aos prejuízos estéticos e econômicos causados pela disposição inadequada desses resíduos, tornaram a poluição plástica alvo de ações e pesquisas nas mais diferentes áreas em âmbito mundial. Se o plástico é o material mais utilizado, é esperado que ele represente uma fração considerável dentre os resíduos sólidos. Além disso, é sabido que o destino final majoritário dos plásticos é o oceano e que cerca de 80% desses resíduos provém dos sistemas terrestres.19,20 Como consequência, há, atualmente, pelo menos cinco grandes ilhas de plásticos nos oceanos. Carregadas pelas correntes marítimas, toneladas de resíduos plásticos flutuantes se acumulam nos vórtices oceânicos, causando prejuízos inestimáveis ao ecossistema marinho, incluindo a morte de milhares de animais de acordo com o Panorama da Biodiversidade Global da Organização das Nações Unidas.21 A PRODUÇÃO E O DESTINO DOS PLÁSTICOS Os MP secundários, por sua vez, são aqueles que resultam da fragmentação dos resíduos plásticos maiores (ex: embalagens) expostos às intempéries no ambiente e outros agentes estressores, degradando-se em fragmentos cada vez menores, até atingir o tamanho dos MP.28,29 A degradação de um polímero está relacionada ao rompimento de ligações químicas covalentes, seja na cadeia principal ou em cadeia lateral. Esse rompimento é consequente da geração de espécies reativas (na maioria dos casos, radicais livres) que são também os responsáveis pela propagação do processo de degradação.30 A iniciação do processo pode ser causada por fotodegradação e/ou degradação física, química e biológica.31 Todas essas formas de iniciação implicam no fornecimento de energia para o rompimento das ligações químicas. Dessa forma, é possível entender porque alguns polímeros se degradam mais facilmente do que outros. Nos polímeros ramificados, por exemplo, a presença da ramificação implica na existência de átomos de carbono terciário na cadeia polimérica principal. A energia da ligação química C-H é menor em átomos de carbono terciário do que nos secundários, de forma que o processo de degradação pode ser iniciado mais facilmente em polímeros de cadeias ramificadas quando comparados aos polímeros lineares, os quais possuem majoritariamente átomos de carbono secundário.32 Os processos de degradação resultam em perda da massa molar do polímero, aumento da molhabilidade e da cristalização, bem como a formação de fissuras, gerando os fragmentos de MP.33 A degradação física das partículas leva à geração de diferentes formas de MP, por exemplo, fibras, fragmentos e filmes. Os pellets, por sua vez, permanecem em sua forma física original após a degradação, mas podem ter algumas de suas propriedades alteradas.34,35 Estudos apontam que, dentre os plásticos mais encontrados no ambiente, estão os polímeros termoplásticos polipropileno (PP), polietileno (PE) (podendo ser PEBD - polietileno de baixa densidade ou PEAD - polietileno de alta densidade), poliestireno (PS), policloreto de vinila (PVC), politereftalato de etileno (PET), poliamida (PA) e o polímero termorrígido poliuretano (PU).16 Nos polímeros ramificados, por exemplo, a presença da ramificação implica na existência de átomos de carbono terciário na cadeia polimérica principal. Distribuição e fontes dos microplásticos no ambiente Uma expedição ao local mais profundo dos oceanos, a Fossa das Marianas, revelou a presença de lixo plástico na coluna d’água. Foram encontrados MP em organismos dessa e de outras cinco fossas profundas do oceano.52 Outros estudos recentes relevaram a presença de MP na neve da montanha de maior altitude do planeta, o Monte Everest,53 bem como no ar de cidades populosas na China.54 Esses exemplos ilustram que os MP são contaminantes que podem ser encontrados em todas as matrizes ambientais e nos seres vivos, tanto em grandes centros urbanos, quanto em regiões remotas no nosso planeta, dada a dinâmica com que são lançados e transportados no ambiente.55-58 Algumas atividades cotidianas, como o uso de cosméticos ou lavagem de roupa, também contribuem para o lançamento de MP em ambientes aquáticos. As microesferas contidas em cosméticos e PHP (Figura 1a, exemplo 4) podem atingir concentrações de mais de 50 mil partículas por grama de produto.60 Em um estudo realizado na Eslovênia, foi estimada uma média diária de lançamento de microesferas no sistema de esgoto de 15 mg por pessoa. Utilizando um processo de tratamento de esgoto com sistema biológico, em escala de laboratório, foi concluído que cerca de 52% das microesferas são retidas no lodo ativado. Assim, para a região avaliada, é previsto um aporte diário de mais de 1 bilhão de microesferas no rio receptor do esgoto tratado.61i Diversas práticas e atividades usam e lançam resíduos plásticos no ambiente, de maneira proposital ou não. Os resíduos plásticos relacionados a essas atividades são as principais fontes de MP para o ambiente. A Figura 1a apresenta exemplos das principais atividades com potencial de emissão de plásticos (que serão geradores de MP) e MP (primários e/ou secundários) ao ambiente, bem como a distribuição deles nos diferentes compartimentos ambientais. A Figura 1b ilustra a distribuição dos MP ao longo da coluna d’água. Inúmeras fibras de tecidos naturais ou sintéticos são liberadas no efluente do processo de lavagem de roupas (Figura 1a, exemplo 5). As fibras sintéticas são constituídas em sua maioria de poliéster, acrílico e/ou PA. Uma única peça de roupa pode produzir mais de 1900 fibras por lavagem e suas concentrações podem superar 300 mg kg-1 de tecido lavado, porém, esses valores variam de acordo com as características de lavagem.62-64 Essas fibras são despejadas Figura 1. Distribuição e fontes dos microplásticos no ambiente (a) Principais exemplos de práticas e atividades que são fontes de emissão de microplásticos (MP) primários e/ou secundários e sua distribuição em diferentes compartimentos ambientais, como: (1) descarte de resíduos plásticos em lixões e aterros; (2) descarte inadequado de resíduos sólidos no ambiente; (3) desgaste de pneus de borracha; (4) utilização de produtos de higiene pessoal e cosméticos compostos por microesferas poliméricas; (5) lavagem de roupas sintéticas com desprendimento de fibras poliméricas; (6) aporte de microfibras e microesferas no esgoto e entrada na Estação de Tratamento de Esgoto (ETE); (7) lançamento do esgoto diretamente no ambiente, sem nenhum tratamento; (8) transporte de MP pelo ar na poeira de ambientes internos; (9) transporte de MP pelo ar em ambientes externos via deposição seca e (10) úmida; (11) uso de filmes plásticos na agricultura; (12) perda logística nas atividades industriais com a utilização de pellets; (13) uso de materiais plásticos em atividades pesqueiras e (14) desgaste de pinturas de embarcações. (b) Principais exemplos de fenômenos que ocorrem em ambientes aquáticos e afetam a distribuição de MP na coluna d’água, como: (15) formação de um biofilme na superfície das partículas de MP - bioincrustação; (15) desprendimento da camada de biofilme no MP – desincrustação e (16) desprendimento de MP aderidos ao sedimento - bioturvação Figura 1. (a) Principais exemplos de práticas e atividades que são fontes de emissão de microplásticos (MP) primários e/ou secundários e sua distribuição em diferentes compartimentos ambientais, como: (1) descarte de resíduos plásticos em lixões e aterros; (2) descarte inadequado de resíduos sólidos no ambiente; (3) desgaste de pneus de borracha; (4) utilização de produtos de higiene pessoal e cosméticos compostos por microesferas poliméricas; (5) lavagem de roupas sintéticas com desprendimento de fibras poliméricas; (6) aporte de microfibras e microesferas no esgoto e entrada na Estação de Tratamento de Esgoto (ETE); (7) lançamento do esgoto diretamente no ambiente, sem nenhum tratamento; (8) transporte de MP pelo ar na poeira de ambientes internos; (9) transporte de MP pelo ar em ambientes externos via deposição seca e (10) úmida; (11) uso de filmes plásticos na agricultura; (12) perda logística nas atividades industriais com a utilização de pellets; (13) uso de materiais plásticos em atividades pesqueiras e (14) desgaste de pinturas de embarcações. Microplásticos primários e secundários Os MP podem ser classificados como primários ou secundários de acordo com a sua fonte, ou seja, a origem do MP.4,22-24 Os MP de fonte primária são aqueles produzidos propositalmente para serem usados na escala de até 5 mm e, consequentemente, são lançados no ambiente nesse tamanho. Esses podem se apresentar em escala microscópica (microesferas), como os utilizados na formulação de cosméticos (dentre eles, os que contém glíter tem recebido especial destaque) e produtos de higiene pessoal (PHP) como esfoliantes, Uma vez no ambiente, a sorção de contaminantes pode potencializar os riscos relacionados à ingestão acidental de MP pela biota, deixando de ser apenas um efeito físico (obstrução do trato 1330 Montagner et al. Quim. Nova Quim. Nova O descarte inadequado de resíduos sólidos é uma prática que culmina para a poluição plástica em ambientes terrestres e aquáticos, com a possível formação de MP nesses compartimentos. Em ambiente terrestre, os resíduos plásticos presentes em lixões e aterros (Figura 1a, exemplo 1) são responsáveis pela formação de MP secundários, os quais são determinados em amostras de lixiviado de aterros sanitários urbanos.59 Além disso, os resíduos plásticos descartados inadequadamente no ambiente terrestre (Figura 1a, exemplo 2) podem atingir os corpos de água e também contribuem para a formação de MP em ambientes aquáticos. Outras fontes de emissão de MP em ambientes terrestres são o processo de desgaste de pneus (Figura 1a, exemplo 3), os quais são constituídos de borracha de estireno butadieno e inúmeros aditivos e o desgaste de pinturas em rodovias, através da abrasão dos veículos e ação do intemperismo.24 digestivo, sufocamento e estresse), mas também provocando outros efeitos fisiológicos (alterações hormonais, distúrbios na produção de enzimas, reprodução e crescimento) que estariam correlacionados aos elevados níveis de degradação dos corpos de água aos quais os organismos estão expostos.39-48 Os MP se comportam ainda como potenciais vetores de transporte de microrganismos, incluindo patógenos, através da formação de um biofilme na superfície do MP.49-51 Espécies invasoras também são transportadas por MP e seus efeitos à biodiversidade do ecossistema ainda são desconhecidos, bem como os prejuízos relacionados à migração de espécies exóticas para outros habitats.37,42 Distribuição e fontes dos microplásticos no ambiente Entretanto, as fibras são produzidas para serem longas e o rompimento físico delas é considerado por outros autores como um mecanismo de degradação, sendo nesse caso classificadas como MP secundários.62 O Guia do GESAMP (do inglês Joint Group of Experts on the Scientific Aspects of Marine Environmental Protection) classifica as fibras têxteis como MP secundários.66 Atividades pesqueiras também geram resíduos plásticos, como as redes e linhas de nylon (PA) e PE, que podem eliminar fragmentos plásticos no ambiente (Figura 1a, exemplo 13). As embarcações podem também contaminar o ambiente aquático com MP através do desgaste de suas pinturas (Figura 1a, exemplo 14), as quais são constituídas de tintas poliméricas como PU e epóxido para evitar a corrosão e a incrustação.24 Estudos relataram quantidades significativas de fibras na poeira de ambientes internos (Figura 1a, exemplo 8).54,67,68 As concentrações de fibras coletadas em amostras de poeira dentro de apartamentos foram cerca de 40 vezes maiores do que em ambientes externos. Devido à dificuldade em caracterizar quimicamente essas fibras, os autores concluíram que somente 33% das amostras correspondiam a polímeros sintéticos (PP e PA, em sua maioria). O restante das fibras encontradas na poeira foi classificado como de origem natural, como algodão, acetato de celulose e lã.69 Em ambientes aquáticos, a densidade e a forma dos MP são, via de regra, os fatores determinantes quanto à sua distribuição ao longo da coluna d’água (Figura 1b). Tabelas com os valores de densidade dos principais polímeros são facilmente encontradas na literatura, porém, essas informações referem-se aos materiais puros.23,81-83 A incorporação de ar e aditivos promove um aumento ou diminuição na densidade do material polimérico processado, tornando-o mais susceptível ou não à decantação. Algumas exceções à regra são encontradas, visto que na prática nem sempre a densidade de um polímero justifica sua presença em determinada profundidade. Isso ocorre porque os polímeros são materiais altamente persistentes e os processos de degradação a que estão expostos no ambiente alteram as suas propriedades físico-químicas se comparados ao polímero virgem.31 Em ambientes externos, os MP no ar podem ser transportados a outros compartimentos ambientais via deposição seca (Figura 1a, exemplo 9) e úmida (Figura 1a, exemplo 10). Em 2015 foi publicado o primeiro estudo sobre a possível presença de MP na atmosfera, que reportou uma média de 118 partículas m-2 d-1 na cidade de Paris. Distribuição e fontes dos microplásticos no ambiente (b) Principais exemplos de fenômenos que ocorrem em ambientes aquáticos e afetam a distribuição de MP na coluna d’água, como: (15) formação de um biofilme na superfície das partículas de MP - bioincrustação; (15) desprendimento da camada de biofilme no MP – desincrustação e (16) desprendimento de MP aderidos ao sedimento - bioturvação Microplásticos: ocorrência ambiental e desafios analíticos 1331 Vol. 44, No. 10 de esgoto.77,78 Dependendo do sistema de tratamento utilizado na ETE, a remoção de MP no afluente pode chegar até mais que 99%.22 Estudos realizados em ETE na Dinamarca estimam concentrações superiores a 18 mil partículas de MP L-1 d-1 no afluente de uma estação com tratamento secundário.79 No entanto, no Brasil, a aplicação do lodo na agricultura está protegida por legislações rigorosas como a Resolução CONAMA N°375/2006, o que torna uma prática pouco comum no país.80 no sistema de esgoto doméstico e encaminhadas para as Estações de Tratamento de Esgoto (ETE) (Figura 1a, exemplo 6). No entanto, em países em que grande parte da população não possui rede coletora e/ou tratamento de esgoto, o aporte de MP é agravado no ambiente, pois o esgoto é lançado diretamente nos corpos aquáticos (Figura 1a, exemplo 7). Por exemplo, no Brasil, a taxa de coleta e afastamento de esgoto é cerca de 60% e aproximadamente 50% de todo o esgoto bruto gerado no país é lançado diretamente nos rios, sem nenhum tratamento.65 Atividades industriais e logísticas relacionadas à produção de plásticos também são fontes de MP primários. Os pellets são perdidos durante seu transporte até seu uso como matéria prima na manufatura de produtos diversos (Figura 1a, exemplo 12), entrando acidentalmente no ambiente. Zonas portuárias são conhecidas por apresentarem altas concentrações de pellets e outros MP, pois estão expostas a grandes quantidades de descarte de resíduos provenientes de diversas fontes, por exemplo embarcações comerciais e turísticas. A comunidade científica ainda discute a classificação das fibras têxteis, consideradas uma das principais fontes de contaminação de MP no ambiente. Alguns autores classificam as fibras como sendo de origem primária, uma vez que são lançadas no ambiente já nesse tamanho pelo descarte da lavagem. Distribuição e fontes dos microplásticos no ambiente Dentre as partículas observadas, mais de 90% foram classificadas como fibras.70 A necessidade de confirmação da composição das partículas e a correlação com as características locais e dados atmosféricos como condições do vento, impulsionou outros trabalhos nessa matriz ambiental em áreas urbanas e rurais. Recentemente, pesquisadores encontraram MP em água de chuva coletada em áreas de proteção ambiental em concentrações médias de 132 partículas m-2 d-1, e estimaram que mais de mil toneladas de partículas plásticas são depositadas pela atmosfera anualmente no oeste dos Estados Unidos.71 A bioincrustação (Figura 1b, exemplo 15), causada pela formação de biofilme e colonização de microrganismos, também é capaz de promover um aumento na densidade da partícula, tornando-a facilmente encontrada em camadas mais profundas da coluna d’água e até no sedimento. O fenômeno inverso, chamado de desincrustação (Figura 1b, exemplo 15), ocorre através do desprendimento do biofilme pela ação de microrganismos e possibilita que os MP retornem à interface água-ar.31 Outros pesquisadores encontraram ainda partículas de MP em regiões montanhosas localizadas entre a Espanha e a França. Essa região é considerada intocada, devido ao difícil acesso e à distância de grandes cidades e centros industriais. A predominância da composição das partículas encontradas foi de PS e PE, polímeros muito utilizados em materiais de uso único e embalagens.72 A presença de MP em áreas remotas retrata, portanto, a facilidade com que essas partículas podem ser transportadas por longas distâncias pela ação do vento.72-74 O aparecimento de ranhuras na superfície, alterações na cristalinidade causadas pela degradação e a presença de bolsas de ar no interior do polímero são capazes de levar a uma diminuição em sua densidade, tornando-os mais propensos à flutuação. Além disso, o movimento da água oriundo das correntes promove a agitação dos MP, o que ocasiona sua presença em toda a coluna d’água, desde a superfície até o sedimento.23,31 Nos oceanos, há ainda um processo de sedimentação de partículas com materiais provenientes da decomposição de células, chamado de neve marinha. Além de prover alimento para o fundo do mar, a neve marinha é também uma via de transporte de MP. Controle de qualidade na determinação de microplásticos em amostras ambientais Recentemente, pesquisadores também encontraram MP de diferentes composições em fezes de pinguins na Antártica, evidenciando a ingestão desses materiais inclusive em áreas pouco habitadas do nosso planeta.114 Algumas consequências da ingestão de MP pelos organismos são a perda da habilidade de capturar e digerir alimentos, perda de apetite devido ao bloqueio do canal alimentar e dificuldade de locomoção.37 Além da ingestão, em alguns organismos, como caranguejos, os MP são encontrados nas brânquias devido aos mecanismos de respiração.86 Após ingeridos, os MP também podem entrar no sistema circulatório causando danos em tecidos e células.87 Estudos in vitro mostraram que o contato de partículas de PP com células mononucleares do sangue periférico pode causar problemas à saúde, induzindo a produção de citotoxinas de células imunes.115 Independentemente das muitas possibilidades do método analítico, uma etapa deve estar presente em todo o processo: evitar a contaminação cruzada. Com o passar dos anos, grandes avanços analíticos têm sido observados nos trabalhos envolvendo a análise de MP. Um deles está relacionado ao controle cada vez mais rigoroso da qualidade dos resultados. Sabe-se que partículas de MP provenientes de outras fontes são responsáveis pela contaminação durante cada etapa da análise de MP. Portanto, o cuidado para evitar a contaminação das amostras inicia-se logo na etapa de amostragem. As duas principais fontes de contaminação de MP são: o local onde as amostras são manuseadas e o analista. Alguns materiais plásticos utilizados na coleta, como as redes e os copos de amostragem, podem contribuir para a contaminação das amostras. A caracterização química desses materiais é uma opção para realizar a identificação do polímero e possivelmente descartá-lo das amostras. Além disso, evita-se, sempre que possível, o uso de materiais plásticos durante a análise de MP, recomendando-se fortemente o uso de utensílios de vidro e de metal. No entanto, o cuidado com a limpeza desses utensílios também é fundamental para garantir a qualidade dos resultados.125,126 Alguns procedimentos de limpeza podem ser encontrados no trabalho de Dehaut e colaboradores.125 Na maioria dos casos, os MP são encontrados no trato gastrointestinal de peixes, portanto, raramente entram na dieta humana, pois são frações não comumente consumidas.116 Porém, em peixes processados, como as sardinhas, essas partículas podem ser ingeridas pelos seres humanos com maior facilidade. Exposição da biota e dos seres humanos aos microplásticos De maneira geral, os MP são facilmente ingeridos quando presentes no ambiente, diretamente por organismos ou indiretamente através do consumo de espécies de nível trófico inferior. A ingestão de MP é observada em organismos desde o primeiro nível trófico, como fitoplâncton e zooplâncton, até espécies maiores, como tartarugas e pássaros.86 Alguns fatores influenciam a biodisponibilidade de MP no ambiente, por exemplo, tamanho, densidade e coloração. Os MP menos densos são ingeridos pelas espécies que habitam superfícies dos corpos de água e os mais densos afetam as espécies presentes na coluna d’água e sedimento. Por serem resistentes aos processos de metabolização, os MP podem se bioacumular em diferentes organismos e, de acordo com a distribuição ao longo dos níveis tróficos, também podem ser biomagnificados.87 Dentre os desafios na elucidação da distribuição dos MP nos diferentes compartimentos ambientais está a harmonização de métodos analíticos, por conta da complexidade de cada compartimento e do tamanho alvo do MP. Apesar de existir alguns direcionamentos para determinadas matrizes, como o guia do GESAMP66 e da NOAA,124 ainda não existem métodos oficiais para a determinação de MP em cada compartimento ambiental. Além disso, são encontradas na literatura diversas unidades de concentração para expressar a quantidade de MP em determinado ambiente, o que pode causar confusão na tentativa de comparar resultados.23 Um panorama das várias possibilidades de um método analítico para a investigação de MP no ambiente que tem sido estudada internacionalmente é resumido na Figura 2. Esse panorama será discutido e detalhado a seguir. Partículas de lixo marinho foram encontradas no intestino de tartarugas no litoral do Rio Grande do Sul, dentre as quais mais de 70% eram MP. Resultados semelhantes foram observados em pássaros marinhos na mesma região, onde mais de 95% das partículas encontradas foram identificadas como MP.88 A presença de MP e resíduos plásticos também foi reportada em diferentes organismos, como plâncton,89 peixes,90-101 tartarugas,88,102 aves marinhas,88,103-105 mexilhões,106 anêmonas,107 peixe-boi,108 pinguins,109 tubarão baleia,110 cachalote,111 caranguejo,112 além de outros organismos que habitam a areia das praias.113 Distribuição e fontes dos microplásticos no ambiente Partículas com tamanho ≥ 200 µm como argila e detritos orgânicos, bem como fitoplâncton presentes na coluna d’água, aderem-se aos MP e podem, assim, deslocá-los a diferentes profundidades até atingir o sedimento.84,85 Uma vez no sedimento, alguns organismos são capazes de promover a bioturvação (Figura 1b, exemplo 16), que possibilita o desprendimento de MP através da abertura de caminhos nas camadas superficiais, sendo assim carregados pela corrente marítima. Essa Outros exemplos de fontes terrestres de emissão de MP estão relacionados à atividade agrícola. Os MP também são provenientes da degradação de filmes plásticos (Figura 1a, exemplo 11), normalmente de PE, utilizados na agricultura para aplicação da técnica mulching, a qual é destinada à proteção da plantação.75 Concentrações de MP entre 80,3 a 1075,6 partículas kg-1 de solo foram identificadas em áreas agrícolas, evidenciando que os filmes plásticos podem contribuir significativamente na contaminação do solo.76 Na agricultura, a aplicação do lodo de esgoto como fertilizante orgânico também pode ser uma via de contaminação do solo por MP, pois o lodo é a fração onde os MP ficam majoritariamente retidos após o tratamento 1332 Montagner et al. Quim. Nova MICROPLÁSTICOS E A QUÍMICA AMBIENTAL: DESAFIOS ANALÍTICOS dinâmica de distribuição permite que os MP ocupem diferentes profundidades em momentos distintos ao longo do seu tempo de vida no ambiente.34,47 Os estudos envolvendo a elucidação dos níveis de contaminação e, consequentemente, o impacto causado pela presença dos MP no ambiente devem considerar inúmeros fatores, dentre eles, alguns estão intrinsicamente ligados à química ambiental, como: i) a extensão da contaminação e a variação sazonal das concentrações; ii) a identificação e a caracterização dos MP incluindo os estágios de degradação dos mesmos; iii) a capacidade de sorção de outros contaminantes incluindo compostos orgânicos, inorgânicos e patógenos e iv) a capacidade de transporte deste material, bem como a identificação de fontes pontuais e difusas de contaminação. Controle de qualidade na determinação de microplásticos em amostras ambientais Um estudo relatou a presença de partículas de MP em sardinhas enlatadas provenientes de diferentes regiões do mundo, como Canadá, Alemanha, Irã, Japão, Letônia, Malásia, Marrocos, Polônia, Portugal, Rússia, Escócia, Tailândia e Vietnã.117 Outros estudos detectaram a presença de MP em água envasada comercializada na Itália e em vários países,118 inclusive no Brasil.119 Os MP também têm sido encontrados em diversos alimentos presentes na dieta humana, como mel, sal, cerveja, vinho, refrigerante, leite, entre outros.120 Com isso, os seres humanos estão constantemente expostos aos MP pela dieta alimentar, sendo estimada uma média de ingestão semanal de 0,1 a 5 g de MP por pessoa.121 E, finalmente, outra via de exposição humana aos MP se dá através da inalação de partículas no ar durante a respiração. Estima-se que um homem adulto inale 170 partículas diariamente,122 no entanto, os efeitos na saúde do sistema respiratório relacionados a essa exposição ainda são desconhecidos.123 Fibras poliméricas sintéticas são as principais responsáveis pela contaminação das amostras e, consequentemente, pela obtenção de resultados superestimados. Essas fibras podem se desprender das roupas do analista e também se aderirem na superfície das luvas. Para amenizar esse problema, a utilização de tecidos de algodão (material polimérico natural) é uma alternativa simples, que pode ser facilmente adotada na rotina do laboratório. Além disso, o cuidado com as luvas também merece atenção. Indica-se que tais luvas sejam guardadas em recipientes fechados e lavadas com água e/ou etanol previamente filtrados.125-127 1333 Microplásticos: ocorrência ambiental e desafios analíticos Microplásticos: ocorrência ambiental e desafios analíticos Vol. 44, No. 10 Figura 2. Panorama das possibilidades de um método analítico para a determinação de microplásticos no ambiente, considerando amostragem do comparti- mento ambiental de interesse, preparo de amostras e caracterização Figura 2. Panorama das possibilidades de um método analítico para a determinação de microplásticos no ambiente, considerando amostragem do comparti- mento ambiental de interesse, preparo de amostras e caracterização Figura 2. Panorama das possibilidades de um método analítico para a determinação de microplásticos no ambien mento ambiental de interesse, preparo de amostras e caracterização ades de um método analítico para a determinação de microplásticos no ambiente, considerando amostragem do comparti- ro de amostras e caracterização Em 2012 foi criada uma classificação para a amostragem de MP em ambientes marinhos, baseada em três abordagens diferentes, e essa classificação tem sido utilizada para as demais matrizes ambientais. Controle de qualidade na determinação de microplásticos em amostras ambientais As amostras de MP que são coletadas por extração direta no ambiente e reconhecidas a olho nu (partículas com diâmetro entre 1 a 5 mm) fazem parte da abordagem de amostragem seletiva, como a amostragem de pellets diretamente na areia em áreas litorâneas. A amostragem no bulk é aquela onde todo o volume de amostra é utilizado, sem a necessidade de redução durante o processo de coleta, como a amostragem de sedimentos marinhos. A amostragem realizada a partir da redução da amostra global, preservando somente a porção de interesse, é chamada de amostragem de volume reduzido, como a coleta de MP em águas superficiais através de redes de plâncton.128 Tendo em vista essas diversas fontes de MP, a adoção de medidas que possam reduzir a contaminação das amostras é uma prática bastante comum e cada vez mais necessária. Para isso, no laboratório, tem-se recomendado a utilização de ambientes limpos e com baixa circulação de pessoas durante a manipulação das amostras de MP. Algumas opções, como a utilização de capelas de fluxo laminar ou capelas com exaustão, têm se mostrado bastante eficientes para reduzir a contaminação. Além disso, a inserção de brancos/controles no estudo de quantificação de MP permite identificar possíveis fontes de contaminação cruzada. A prática mais comum é a inclusão de brancos de procedimento, os quais seguem o mesmo procedimento das amostras. A frequência em que tais brancos são preparados fica a critério do analista, visto que não existe procedimento padrão para a análise de MP em matrizes ambientais. Além disso, controles atmosféricos também podem ser inseridos durante a análise, a fim de quantificar a deposição de microfibras e podem tramitar em paralelo aos brancos de procedimento.125-127 Uma vez que cada matriz ambiental tem suas particularidades quanto à amostragem e o preparo de amostras, a discussão desse tópico será dividida entre as três principais matrizes estudadas na química ambiental: água, solo/sedimento e ar. Nesta revisão não serão abordadas as técnicas de amostragem e preparo de amostras biológicas, para tal, a leitura de outros trabalhos mais específicos é recomendada.34,81,129 A eliminação das fontes de contaminação cruzada de MP durante a análise de amostras ambientais ainda é considerada um grande desafio para os analistas, pois a identificação dessas fontes é uma tarefa laboriosa. Assim, a quantificação da contaminação cruzada através do uso de brancos/controles é uma alternativa que pode ser adotada. Controle de qualidade na determinação de microplásticos em amostras ambientais No entanto, alterações nos brancos/controles podem ocorrer, por exemplo, devido à variação da concentração de fibras no ar dentro do laboratório durante todo o procedimento, e também em cada dia de análise. Portanto, a adoção de protocolos padronizados no laboratório permite que os resultados obtidos sejam comparáveis, reprodutíveis e confiáveis. Matriz: água A escolha da técnica de amostragem de MP em matrizes aquáticas é dependente da profundidade de estudo e da logística de acesso ao ponto amostral, pois existe uma ampla variedade de equipamentos destinados à coleta nesse compartimento ambiental. As redes de coleta são, sem dúvida, a técnica mais comum para a obtenção de amostras de MP nessa matriz. São utilizadas para a amostragem em superfície e em profundidades rasas de até 1 m, tanto em ambiente marinho,56,57,130 quanto em água doce70,131,132 e salobra, como os estuários.130,133,134 Porém, há a necessidade do uso de embarcações ou pontes para ter acesso ao local de amostragem, além de profundidade e vazão mínima para que a rede possa ser armada. Montagner et al. Montagner et al. Quim. Nova Quim. Nova Amostragem discreta utilizando baldes de metal146 e de plástico,147 frascos de vidro131,148,149 e de aço inoxidável150 são opções de baixo custo para a coleta de MP em ambientes aquáticos e bastante atrativos quando o acesso ao local de amostragem é restrito, a vazão ou a profundidade do corpo hídrico são baixas. Os coletores são inseridos manualmente em águas superficiais, o que torna a coleta bastante laboriosa ao considerar que, na maioria das vezes, a introdução do coletor no corpo hídrico deve ser realizada várias vezes a fim de obter um volume de amostra representativo. O volume amostrado pode ser medido com precisão empregando balanças portáteis. O uso de copos de amostragem ou peneiras com diferentes granulometrias facilita o tratamento da amostra, que é posteriormente manuseada no laboratório.148-150 Em comparação com as redes nêuston148 e com as redes de plâncton, bongô e manta,149 a amostragem discreta obteve concentrações de MP de 3 a 4 vezes maiores. Essas altas concentrações podem ser explicadas devido à capacidade de obtenção de MP menores (entre 100 µm e 1,5 mm). Com isso, essa técnica de amostragem, embora laboriosa, pode ser considerada uma opção eficiente para a captura de partículas que passariam através da malha das redes.148 Outros modelos de frascos coletores, porém menos utilizados, também são encontrados para a amostragem discreta, por exemplo garrafas de Van Dorn151 e frascos Niskin.152,153 Para a quantificação de MP é preciso conhecer o volume exato de água amostrado, o qual é adquirido por meio de informações geradas por medidores de vazão acoplados nas redes e de sua área de abertura.135,136 Porém, a imprecisão de volume é algo recorrente, podendo ser gerada pela ação do movimento da água, por exemplo, ondas e correntezas, em que a submersão da rede pode não ser constante durante toda a amostragem, impedindo a entrada de água na rede.135 A abertura das redes pode ser circular ou retangular, com dimensões variáveis, as quais são essenciais para o posterior cálculo da área de abertura. Diversos modelos de rede são encontrados na literatura (redes de plâncton, nêuston e bongô) e, em geral, são fabricadas em nylon (PA) e acopladas a um recipiente para armazenamento da amostra (copo de amostragem) produzido em PVC, com uma malha de mesmo tamanho da rede. Matriz: solo/sedimento A ã A preocupação com a presença de MP em ambientes terrestres vem ganhando espaço na pesquisa. No entanto, assim como em ambientes aquáticos, até o momento, a amostragem de MP em solo e sedimento não possui protocolo padrão e a falta de consenso sobre as técnicas de amostragem geram uma variedade de unidades de concentração nessa matriz ambiental. Em geral, a quantificação de MP é expressa em itens por unidade de massa seca ou úmida (g, kg), de volume (mL, L, m3) ou de área (m2).23,46,154 As amostras das camadas superficiais podem ser retiradas manualmente com a utilização de utensílios de metal como espátulas, pás, colheres e pinças. Para a amostragem de MP em profundidade, normalmente são utilizados equipamentos específicos destinados a esse tipo de matriz ambiental como trados, dragas e corer (gravity corer, box corer).154,155 Em um estudo comparativo de amostragem em sedimentos marinhos na profundidade de 42 m, não foi observada diferença significativa na abundância de MP entre as coletas com draga Van Veen, box corer e gravity corer.155 Um histórico de sedimentação pode ser adquirido através de testemunhos, em que um tubo de PVC é utilizado para a obtenção de várias camadas de amostra. Com isso, é possível estimar a taxa de deposição de MP em sedimentos, por exemplo. Na areia de praia, a coleta de MP ou resíduos plásticos de tamanho superior ao de MP geralmente é realizada por meio de transectos com larguras e distâncias previamente definidas.156 Essa e outras abordagens também podem ser utilizadas para amostragem em solo.154 Embora não exista um padrão de profundidade, bem como de massa de amostra, geralmente coletam-se massas entre 500 g a 10 kg de sedimento e entre 50 g a 4 kg de solo.34,81,154 A coleta de MP em ambiente aquático também pode ser realizada com a utilização de bombas de amostragem, as quais possibilitam a separação in situ das partículas através do uso de peneiras com diferentes aberturas ou de copos de amostragem. Matriz: solo/sedimento A ã A coleta com bombas de amostragem torna-se viável devido à capacidade de separação de partículas com diferentes tamanhos e à operação em diferentes profundidades, ilustrando, assim, que essas técnicas podem ser consideradas complementares e a escolha deve priorizar as logísticas da amostragem como acesso ao local de coleta, vazão, profundidade, condições climáticas, entre outras.140,141 Além disso, essa coleta pode ser aplicada tanto em águas superficiais quanto ao longo da coluna d’água, em ambientes marinhos135,138,141,142 e de água doce,143 incluindo lagos140 e ETE.144 As bombas permitem maior precisão de volume quando comparadas às redes, porém a quantidade de água amostrada geralmente é menor, considerando uma coleta com o mesmo tempo de duração.135,141 Isso porque a rede abrange uma área de amostragem maior devido ao seu tamanho e, ao mesmo tempo, garante a amostragem na superfície, o que é um grande desafio para a amostragem realizada com bombas que necessitam estar minimamente mergulhadas nos corpos aquáticos para evitar danos ao sistema. Além disso, as bombas necessitam de uma fonte de energia próxima ao local para mantê-las em funcionamento durante toda a coleta e esse recurso nem sempre é viável. Dos pontos já discutidos, outro fator que deve ser levado em consideração durante a amostragem de MP, seja ela realizada com redes ou com bombas, é o volume de amostra, o qual também está ligado ao tempo de coleta. Para amostragens mais representativas em águas superficiais, recomenda-se a coleta de um volume de água maior que 500 L, no entanto, esse volume é dependente da vazão do corpo hídrico e do material utilizado na coleta.145 Montagner et al. Além disso, as redes podem conter um sistema de estabilização em águas turbulentas, como ocorre nas redes manta e catamarã.137 O trabalho de Silva e colaboradores descreve com mais detalhes os diferentes modelos de rede.137 Redes com malhas na faixa de 300 a 350 µm são as mais utilizadas para a amostragem de MP.81 Com isso, é esperado que partículas com tamanhos inferiores ao da malha não sejam encontradas, pois são perdidas ao passarem em seu interior, levando à subestimação da concentração de MP.138 Assim, a utilização de redes com malhas de tamanhos inferiores, possibilita uma amostragem mais representativa, obtendo partículas de MP em uma faixa de tamanho mais ampla. Em um estudo de comparação, onde foram utilizadas redes com três tamanhos de malha (100, 300 e 500 µm), foi observado que na rede de 100 µm a quantidade de MP foi 10 vezes maior do que na rede de 500 µm e 2,5 vezes maior do que na rede de 300 µm.139 No entanto, com a diminuição do tamanho da malha, as redes tornam-se mais propensas a sofrer entupimento com material orgânico, podendo resultar em perda da sua eficiência e em diminuição do volume de água coletado.82,139 Amostragem O primeiro desafio envolvendo a amostragem é a estratégia utilizada, pois MP são contaminantes insolúveis, de diferentes composições e propriedades físico-químicas, estando dispersos de forma não homogênea na amostra. Porém, há a necessidade do uso de embarcações ou pontes para ter acesso ao local de amostragem, além de profundidade e vazão mínima para que a rede possa ser armada. 1334 Preparo de amostras Em geral, o preparo de amostras de MP coletados em ambientes aquáticos é composto por duas etapas: (i) separação das partículas e (ii) remoção da matéria orgânica. A separação das partículas coletadas é um processo fundamental, pois detritos orgânicos e inorgânicos (por exemplo, minerais e impurezas não dissolvidas) são facilmente confundidos com MP, quando apenas uma inspeção visual é realizada. A separação pode ser realizada por filtração, peneiração ou por diferença de densidade. A filtração ou peneiração da água coletada pode ser realizada in situ ou ex situ. Para isso, comumente são utilizadas peneiras granulométricas ou membranas de filtração. A abertura ou porosidade desses materiais varia de acordo com o tamanho das partículas a serem estudadas. p Para isolar os MP da matriz de estudo, soluções salinas com altas densidades, como NaCl (d = 1,2 g cm-3), NaI (d = 1,6‑1,8 g cm‑3), ZnCl2 (d = 1,5-1,8 g cm-3) ou ZnBr2 (d = 1,7 g cm-3) podem ser adicionadas à amostra. Assim, de acordo com a diferença de densidade das partículas, duas frações são obtidas no sistema, uma com baixa e outra com alta densidade. A separação pode ser realizada em funis de separação com ação da gravidade ou em sistema de centrifugação. Trata-se de uma técnica simples na qual é possível separar os MP de outros componentes da matriz e, também, separar os MP com diferentes composições poliméricas, facilitando a primeira triagem da amostra. No entanto, alguns erros podem ser cometidos devido à degradação do polímero e incrustação de matéria orgânica.34,81,82 Além das soluções citadas anteriormente, para amostragens em ambiente marinho, a própria água do mar (d = 1,03 g cm-3) pode ser utilizada nessa etapa.82 Cada solução apresenta uma particularidade e a escolha do reagente deve se basear em seu custo-benefício e facilidade de obtenção e manuseio. Com a utilização da solução de NaCl é possível separar os polímeros com densidade até 1,2 g cm-3 (PP, PU, PE, PA e PS), enquanto os polímeros com densidade superior, como PET e PVC, permanecerão no precipitado. Por outro lado, com a solução de NaI ou ZnBr2, uma gama maior de polímeros pode ser separada, conforme observado na Figura 3. Utilizando a água do mar para separação, somente os polímeros PP e PE ficarão na superfície devido às suas densidades serem inferiores a 1,0 g cm-3. Preparo de amostras A falta de harmonização de procedimentos de coleta e preparo de amostras de MP em ambientes aquáticos possibilita a enorme variedade de métodos encontrados na literatura, ficando a critério do pesquisador a escolha do qual será aplicado, em função de seu objetivo. O protocolo apresentado pela NOAA é um dos mais completos para ser adotado quando se deseja estudar MP nesses ambientes, pois conta com etapas de separação de partículas e remoção de matéria orgânica. No entanto, a maior parte dos estudos envolvendo essa temática não apresenta nenhuma etapa de preparo de amostra.82 Isso também está relacionado à falta de caracterização química das partículas, em que somente a identificação visual (a olho nu ou por microscopia) é realizada. Portanto, a escolha do método de preparo é também dependente da técnica utilizada para caracterizar as partículas, visto que possíveis interferências podem ocorrer prejudicando a análise. O preparo de amostras em matrizes ambientais sólidas (sedimento e solo) em geral, pode seguir as mesmas etapas das amostras aquosas, como citado anteriormente.154,168,169 A NOAA estabeleceu um protocolo que utiliza uma solução de metatungstato de lítio (d = 1,6 g cm-3) na separação por densidade para MP coletados na areia e no sedimento marinho. Além disso, para a digestão da matéria orgânica em ambas as amostras, o uso do reagente de Fenton é recomendado.124 No entanto, as mesmas soluções salinas citadas no preparo de amostras aquosas para a separação por densidade, também podem ser utilizadas para amostras coletadas em ambientes terrestres.170-174 Figura 3. Separação por densidade contendo microplásticos de diferentes composições poliméricas utilizando as principais soluções salinas saturadas (NaCl, NaI, ZnBr2) e água do mar Quando as amostras de areia e sedimento são submetidas somente à inspeção visual, a remoção da matéria orgânica não é uma etapa de preparo muito utilizada. No entanto, com a utilização de técnicas para a caracterização química das partículas, esse processo é fundamental, pois amostras de solo e sedimento são consideradas complexas e heterogêneas. Para tal, diferentes estudos descrevem a utilização de agentes oxidantes, digestão ácida, alcalina e enzimática para a remoção da matéria orgânica.154,168,169,175 Finalmente, alguns procedimentos de preparo de amostra de MP coletados no ar atmosférico e na poeira estão descritos na literatura. Matriz: ar Uma das formas de coletar amostras de MP na atmosfera é por amostragem passiva. Nesse método, as amostras são depositadas naturalmente e podem ser obtidas por via seca ou via úmida (através da água da chuva). Alguns trabalhos que utilizam esse método são encontrados na literatura e a concentração de MP pode ser expressa em partículas m-2 d-1.70,72,157-161 Outra forma de obter amostras de MP da atmosfera é com a utilização de amostradores ativos. Nesse caso, o ar é forçado a passar por um suporte sólido, como filtros, utilizando bombas de amostragem com vazões conhecidas possibilitando que a concentração de MP seja expressa em partículas por unidade de Microplásticos: ocorrência ambiental e desafios analíticos Vol. 44, No. 10 1335 volume.69,162-165 Filtros de diferentes composições e porosidades são utilizados nessa amostragem e sua escolha é baseada na técnica de caracterização química que será utilizada.166 Devido à presença de matéria orgânica na superfície do MP, possíveis interferências na etapa de caracterização química podem ser observadas. Portanto, é necessária a utilização de métodos para eliminação da matéria orgânica que não danifiquem as propriedades químicas e estruturais dos polímeros. Os métodos mais comuns encontrados na literatura são aqueles baseados em digestão com soluções ácidas ou alcalinas, digestão enzimática e digestão branda utilizando oxidantes como H2O2 ou NaClO.82 A digestão ácida e alcalina pode promover degradação e descoloração de alguns polímeros, prejudicando sua caracterização química.81,82 A oxidação com H2O2 é o método mais encontrado na literatura, pois promove a digestão da matéria orgânica de uma forma mais branda. No entanto, pode afetar a integridade de alguns polímeros e, o tempo de exposição dos MP ao H2O2 é um fator que deve ser controlado.145,167 A NOAA estabeleceu um protocolo para tratamento de amostras de água e sedimento marinho, que consiste na utilização do reagente de Fenton para oxidação da matéria orgânica e posterior separação por densidade com solução de NaCl.124 volume.69,162-165 Filtros de diferentes composições e porosidades são utilizados nessa amostragem e sua escolha é baseada na técnica de caracterização química que será utilizada.166 Espectroscopia no infravermelho Essa tem sido a técnica analítica instrumental mais utilizada para confirmação da identidade de MP. Por meio da interação do polímero com a radiação eletromagnética na região do IR (comprimentos de onda entre 780 nm e 50 µm), obtém-se o espectro de absorção do polímero através de medidas de transmitância ou reflectância, que é característico para cada tipo de material. Assim, a confirmação da identidade do polímero é fundamentada pela atribuição das bandas. Para uma visão geral sobre a caracterização morfológica por inspeção visual e outras categorizações físicas, que não são o foco desta revisão, sugere-se a leitura de Lusher e colaboradores,176 bem como do Guia proposto pelo GESAMP.66 Nesta revisão, serão abordadas as principais técnicas para caracterização química da identidade do polímero, sendo que outras informações como tamanho, massa e forma também podem ser levantadas nessas análises em função da instrumentação utilizada. O tamanho mínimo de MP que a espectroscopia no IR pode acessar será dependente da instrumentação utilizada, podendo variar de dezenas a centenas de micrometros. Assim, reforça-se que o tamanho do MP em estudo é um fator chave para selecionar instrumentos adequados dentro da mesma técnica, impactando diretamente no custo de aquisição e manutenção das pesquisas. Embora as técnicas de caracterização química de materiais plásticos sejam bastante consolidadas, a temática de ocorrência de MP no ambiente trouxe o desafio de caracterizar polímeros pós- formulados com aditivos e processados, em tamanho reduzido, resultante de materiais envelhecidos por ação ambiental e dispersos em uma matriz complexa de onde serão extraídos. Assim, o MP encontrado no ambiente pode ter sofrido os mais diversos efeitos físico-químicos e microbiológicos, trazendo complexidade à interpretação dos resultados das análises, mesmo para técnicas já amplamente empregadas na caracterização de materiais. q ç ç p q Para exemplificar as diferentes aplicações e instrumentações, considera-se a técnica de espectroscopia no IR médio (MIR, do inglês mid-infrared, de 400 a 4000 cm-1). Por exemplo, pellets ou MP na forma de fragmentos medindo 3 x 3 x 1 mm são facilmente isolados da matriz ambiental e podem ser identificados por meio da análise no instrumento espectrômetro MIR com transformada de Fourier (FTIR, do inglês Fourier Transform Infrared) convencional, usando uma unidade ou acessório para aquisição de dados por reflectância total atenuada (espectroscopia ATR-FTIR, do inglês Attenuated Total Reflection–Fourier Transform Infrared). Nela, cada partícula de MP é prensada individualmente sobre um cristal (ZnSe, ZnS, diamante ou Si) para análise. Espectroscopia no infravermelho Como exemplo, MP maiores que 630 μm coletados de sedimento de rio foram analisados por essa instrumentação,177 assim como MP coletados do trato intestinal de baleias.178 Observa-se que nesse caso a instrumentação não tem um microscópio acoplado. Assim, a técnica é de fácil exequibilidade para partículas maiores que 500 µm, sendo limitada pelo tamanho. A limitação se dá tanto pelo manuseio, quanto pela qualidade do espectro. Sem um tamanho ou formato adequado para total contato da partícula com o cristal, a intensidade das bandas pode ser afetada. Isso ocorre especialmente naquelas de comprimentos de onda mais curtos na região de 2800 a 4000 cm-1, devido à menor penetração da radiação na amostra e, portanto, interferindo em bandas resultantes dos modos vibracionais de C-H, O-H e N-H, importantes para a identificação de polímeros.23 p p g ç Para a caracterização química de MP já foram reportados o uso do teste clássico de ponto de fusão (hot-stage melting ou “hot-needle”),176 das técnicas instrumentais como espectroscopia vibracional (infravermelho – IR, do inglês infrared, e Raman), técnicas termoanalíticas (pirólise – py, análises térmicas como termogravimetria ou análise termogravimétrica – TGA) hifenadas à cromatografia gasosa e/ou espectrometria de massas (pirólise acoplada à cromatografia gasosa – GC do inglês gas chromatography e espectrometria de massas – MS do inglês mass spectrometry, py- GC-MS). Essas técnicas instrumentais são as mais frequentemente encontradas na literatura e serão detalhadas nas próximas seções. No entanto, outras técnicas como cromatografia de exclusão de tamanho – SEC (do inglês Size Exclusion Chromatography); microscopias (microscopia eletrônica de varredura – SEM, do inglês Scanning Electron Microscopy, acoplada à espectroscopia de raios X por dispersão em energia, do inglês Energy-Dispersive X-Ray Spectroscopy - EDS, EDX, EDXS) também já foram reportadas na análise de MP.7 Assim, um ATR-FTIR convencional deixa de ser adequado para a investigação de MP menores, por exemplo, quando o objetivo da análise é investigar uma fibra de 50 µm de diâmetro depositada em um substrato após filtração de água de rio ou amostragem atmosférica. Para analisar o MP nesse tamanho, sobre a matriz, é necessário o uso do espectrômetro MIR acoplado a um microscópio (em geral referido como µ-FTIR), adquirindo espectros no modo de transmitância, reflectância ou ATR (µ-ATR-FTIR). Preparo de amostras No entanto, muitos estudos utilizam somente uma simples filtração das amostras que se encontram em meio líquido, para a retenção das partículas em um suporte sólido (filtros de fibra de vidro, quartzo, celulose, PTFE, entre outros) e a amostra é submetida diretamente para inspeção visual e/ou caracterização química.70,157,158,162,164 A utilização de técnicas de separação por diferença de densidade também pode ser utilizada, bem como de oxidação de matéria orgânica.54,67,69,159 No entanto, assim como para amostras coletadas em ambiente aquático e terrestre, a necessidade dessas etapas de preparo Figura 3. Separação por densidade contendo microplásticos de diferentes composições poliméricas utilizando as principais soluções salinas saturadas (NaCl, NaI, ZnBr2) e água do mar 1336 Montagner et al. Quim. Nova Quim. Nova deve ser avaliada considerando, principalmente, a técnica que será utilizada para a caracterização química. são, a priori, capazes de fornecer essa informação. Entretanto, as técnicas termoanalíticas são destrutivas e as espectroscópicas, em sua maioria, não. Caracterização química e quantificação de microplásticos Além disso, uma mesma técnica analítica pode ter diferentes arranjos instrumentais que permitirão ou não acessar a informação desejada. Logo, a consideração de outros fatores (tamanho alvo do MP investigado e matriz em estudo) é muito importante para guiar a escolha do instrumento e, consequentemente, o preparo de amostras que antecede a análise instrumental. Os MP são caracterizados majoritariamente por tamanho, forma (fibras, fragmentos, filmes, esferas ou pellets), cor, tipo (composição polimérica) e origem (primário ou secundário), tanto por técnicas que averiguam se o material é um plástico, quanto pelas que identificam o polímero majoritário. A inspeção visual com ou sem microscópio óptico foi e ainda é muito usada em triagem de suspeitos MP, podendo ser uma identificação subjetiva se não houver uma confirmação química com outra técnica instrumental. A vantagem é seu baixo custo e acessibilidade, inclusive em campo, com a ciência de que uma maior taxa de falsos positivos e negativos podem ocorrer sem a etapa de confirmação. Por outro lado, são coletadas centenas ou milhares de partículas nas amostragens, tornando o tempo de análise um fator importante para a decisão de realizar a confirmação química de todas as partículas. A seguir serão discutidas as técnicas mais utilizadas, bem como as estratégias para a identificação e/ou caracterização e os avanços analíticos utilizados para análise de MP, em amostras padrão (como prova de conceito) e em amostras ambientais. Espectroscopia no infravermelho Lindeque e colaboradores utilizaram as duas instrumentações, sendo um ATR‑FTIR convencional A escolha da técnica de caracterização para a análise ambiental deverá ser fundamentada em função da pergunta que se deseja responder no estudo, buscando aquela capaz de fornecer a informação relevante para a resposta, após considerado o seu contexto. Assim, uma técnica pode ser suficiente ou pode ser necessário o uso de técnicas complementares. Por exemplo, caso o objetivo seja a identificação do polímero majoritário do MP, as técnicas termoanalíticas acopladas (como py-GC-MS) e as técnicas espectroscópicas vibracionais Microplásticos: ocorrência ambiental e desafios analíticos Vol. 44, No. 10 1337 para partículas maiores que 100 μm e um μ-FTIR para as partículas menores.139 Utilizando µ-FTIR, há trabalhos sobre a ocorrência de MP no trato digestivo de peixes coletados no estuário do Rio Amazonas,94 MP coletados no mar em regiões remotas da Islândia,149 MP coletados em amostras de deposição atmosférica,69,161 entre outros.150 presença de água, portanto, é preciso controlar a umidade da amostra antes da medida, caso contrário uma larga banda do estiramento OH em 3300 cm-1 pode sobrepor outras bandas, dificultando a análise. Outro aspecto importante a ser considerado na escolha da técnica é o grau de envelhecimento dos polímeros. Fragmentos degradados podem se quebrar com facilidade durante uma análise por ATR-FTIR ou µ-ATR-FTIR, bem como podem apresentar biofilme ou matéria orgânica sorvida em sua superfície e aderir ao cristal, trazendo problemas à medida. Por isso, é comum realizar a digestão da amostra antes da análise por FTIR, para diminuir interferências. Por outro lado, FTIR é uma ferramenta consolidada quando o objetivo da análise é estudar a degradação dos polímeros, por exemplo, por meio dos índices de carbonila e hidroxila, grupos formados na cadeia polimérica de poliolefinas devido às oxidações que ocorrem no processo de degradação fotoquímica do plástico.33 Quando a quantidade de MP coletados é muito grande durante a amostragem ambiental, é comum uma subamostragem das partículas para reportar a caracterização de somente uma fração (ou um subset). Espectroscopia no infravermelho Embora seja uma prática que deveria ser evitada para redução de viés, fatores como tempo de processamento, custo analítico, operacional ou indisponibilidade de equipamentos tornam muitas vezes inviável a caracterização da composição de toda a amostragem.176 Em especial, no caso de MP pré-concentrados ou depositados sobre filtros (o substrato mais utilizado para análises microscópicas por μ‑FTIR), recomenda-se que pelo menos 25% do filtro seja analisado.145 O analista pode procurar possíveis MP sobre o filtro realizando uma inspeção visual não padronizada pelo microscópio para posterior aquisição do espectro do suspeito MP, mas é um trabalho muito laborioso e pouco preciso. Nesse caso, para buscar partículas em uma maior extensão do filtro de forma padronizada sem prévia inspeção visual, diminuindo a subjetividade, pode-se utilizar as técnicas de mapeamento ou imageamento hiperespectral (HSI, do inglês hyperspectral imaging).179 Nesse caso, a aquisição de espectros segue uma ordenação espacial por meio de varredura pontual (aquisição de um espectro em um determinado ponto da amostra, seguindo sucessivamente a ordenação espacial estabelecida, usando um detector monodimensional), ou pela aquisição simultânea dos espectros em vários pontos da amostra por meio de um detector bidimensional, em arranjo no plano focal, FPA (do inglês focal plane array). A análise de uma mesma área do filtro utilizando detectores bidimensionais tipo FPA é mais rápida que a aquisição pelos detectores monodimensionais.180 O processamento de dados fornece uma imagem em cores falsas em que é possível localizar os MP, obter tamanho, forma e a identidade do polímero. Os trabalhos abordados até o momento nessa seção utilizaram IR médio (MIR, de 400 cm-1 a 4000 cm-1 ou 2500 nm a 25000 nm). Outra região espectral do IR começou a ser explorada mais recentemente,190 em comprimentos de onda de 780-2500 nm: o IR próximo (NIR, do inglês near-infrared).191,192 Uma das modalidades exploradas foi pela utilização de instrumentos portáteis, com potencial de análises em campo.193,194 Outra modalidade foi a espectroscopia NIR combinada com imageamento em linha (line-scan).195 Diferente do modo de escaneamento pontual (point-scan), comum em microscópios, a instrumentação em linha permite análises mais rápidas por adquirir dados de várias amostras simultaneamente e escanear uma área amostral da ordem de centímetros em menos de um minuto,196 a depender do instrumento. Essa técnica é denominada HSI-NIR (do inglês near-infrared hyperspectral imaging), em região espectral também reportada como SWIR (do inglês shortwave infrared). Espectroscopia no infravermelho A maior vantagem da técnica é a sua rapidez de aquisição dos espectros, entretanto, demanda um tratamento de dados mais complexo. Devido ao perfil espectral com bandas mais largas, efeitos de espalhamento da luz na aquisição por reflectância e uma quantidade expressiva de dados a serem tratados (uma imagem pode conter centenas de milhares de espectros), ela é uma técnica que deve ser combinada com quimiometria e tratamento de imagem. A técnica já foi reportada para a identificação de MP coletados em água, solo e em peixes, por meio de análise exploratória e modelos supervisionados de classificação, detectando MP na faixa de tamanho de 200 μm a 5 mm nestes trabalhos.152,190,196-198 Estudos recentes utilizando imageamento µ-FPA-FTIR versaram sobre a ocorrência de MP em neve dos alpes ao ártico,181 MP no oceano Atlântico,182 em estudo de bacias de retenção de águas pluviais como fonte terrestre de MP a ambientes aquáticos183 e em estudo de remoção de MP em ETE.184 O tipo de substrato requerido para apresentar a amostra ao μ-FTIR depende do modo de aquisição. No modo reflectância usando um μ-ATR-FTIR, em geral usam-se membranas de filtração ou superfícies metálicas, como Au e Ag.161 Entretanto, o substrato deve ser transparente à faixa espectral MIR no caso de medidas por transmitância, sendo utilizado cristal de KBr ou CaF2, membrana de óxido de alumínio ou de silício,185-187 sendo que as janelas espectrais acessadas variam de acordo com a membrana.23 Ao se utilizar μ‑FTIR no modo transmitância, é possível detectar MP de até 20 μm, sendo um tamanho próximo do limite de difração no IR (10 µm em 1000 cm‑1). Porém, pode haver saturação do sinal quando empregada medida μ-FTIR por transmissão.188 Já o µ-ATR-FTIR apresenta a limitação das partículas eventualmente se aderirem ao cristal, o que limitaria o menor tamanho a centenas de micrometros.185 Uma melhoria recente para a aquisição de dados por imageamento μ‑FPA-FTIR foi o uso de uma janela de BaF2, transparente ao IR, sobre a amostra depositada na membrana de óxido de alumínio. O objetivo do uso da janela foi para nivelar a superfície da amostra e apresentar os MP (inclusive fibras) no plano focal do microscópio, sendo que a falta de nivelamento era considerada uma dificuldade para ajuste de foco na medida, principalmente de fibras. Como resultado, a quantidade de MP detectados com a amostra coberta foi maior do que das mesmas amostras analisadas sem a janela de BaF2.189 Raman A resolução espacial foi de 1 µm pixel-1 e a faixa espectral foi de 0 a 4000 cm-1. Cinco tipos de MP (PS, PET, PE, PVC e PP) foram monitorados nas amostras com e sem areia, sendo essa última também identificada pelo espectro Raman. Utilizando um tempo de integração de sinal Raman de 0,5 s, as imagens com 88 µm2 foram adquiridas em 64,5 min. Os autores concluíram que a técnica distinguiu os diferentes MP em uma mesma amostra, de forma a obter sinais separados de cada partícula, mas isso requereu um cuidadoso preparo da amostra a fim de obter uma superfície plana, em virtude da alta resolução espacial que o instrumento apresenta.225 A ingestão, digestão e aderência de partículas de PS, entre 1,4 a 30 µm de diâmetro, foi identificada em zooplânctons. Essa identificação foi realizada por meio de imagens químicas CARS sintonizando o laser sonda em bandas específicas do polímero (2845 cm-1 do estiramento C-H e 3050 cm-1 do estiramento C-H de aromáticos).215 Um método para análise de MP baseado na microscopia SRS foi descrito na literatura, em que foram adquiridas imagens de uma amostra teste preparada em laboratório, de uma amostra de esmalte de unha que continha PET e de uma amostra de sedimento do estuário do rio Reno. Os plásticos foram investigados, sintonizando o laser sonda em bandas específicas de 5 polímeros (PA, PET, PS, PP e PE), encontrados frequentemente nas amostras ambientais de MP. Foi utilizada uma membrana de óxido de alumínio devido ao baixo sinal de fundo da mesma e à sua superfície plana, uma vez que a resolução espacial da técnica é alta. As imagens foram adquiridas em uma área de 1 cm2 do filtro e duraram cerca de 5 h, com espaçamento de 3 µm entre os pixels e tempo de 100 µs pixel-1. Ressaltou-se que a análise dessa mesma área da membrana, com o mesmo tamanho de passo, mas utilizando o Raman convencional que requer em média um tempo de integração de 1 s pixel-1, demoraria 116 dias para aquisição de dados.226 Vale a pena destacar que o volume de bytes de dados é drasticamente reduzido utilizando as técnicas CARS e SRS, pois ao invés de adquirir milhares de comprimentos de onda por pixel, somente um ou algumas unidades por pixel são monitorados por imagem. Tais técnicas possuem características muito atrativas, mas dispõem de instrumentação bastante complexa, requerendo analistas especializados. Raman Por outro lado, as HSI carregam consigo uma riqueza de informações físico-químicas que permitem responder sobre a composição química de cada partícula da amostra com alto contraste molecular dada pela espectroscopia, sobre o tamanho e a morfologia das partículas e, por fim, a contagem delas. O primeiro trabalho a utilizar a espectroscopia Raman com a finalidade de analisar MP na temática de análise ambiental foi publicado em 2012.222 No entanto, esse estudo empregou amostras simuladas em laboratório utilizando sedimento e plásticos comerciais que não sofreram ação ambiental. O trabalho descreve uma prova de conceito de um método de separação, por meio do desenvolvimento de um aparato que favorecesse a recuperação de partículas de unidades de micrometros, utilizando ZnCl2 como solução para separação por densidade. As partículas foram classificadas como “large microplastic particles” (L-MPP; entre 5 e 1 mm) e “small microplastic particles” (SMPP; < 1 mm). Após o processo de separação, as L-MPP foram coletadas manualmente de uma membrana de filtração de quartzo e apresentadas ao microscópio para a aquisição espectral, enquanto as S-MPP foram analisadas diretamente na membrana. A caracterização química dos MP foi realizada utilizando espectroscopia Raman com laser em 633 nm. A inspeção visual e aquisição espectral das partículas S-MPP puderam ser feitas simultaneamente, pelo fato do espectrômetro Raman estar acoplado a um microscópio óptico confocal. Os espectros obtidos, na faixa de 50-4000 cm-1, foram comparados por meio da sobreposição dos dados de espectros obtidos de materiais de referência contidos em bases de dados comerciais, variando parâmetros instrumentais para otimizar a aquisição de dados. Foi possível identificar partículas de PE e PS de 5 µm.222 A base desse método ainda é bastante utilizada hoje em dia, a qual consiste na separação/filtração da amostra, seleção das partículas seguindo um critério visual (utilizando ou não microscopia óptica), aquisição dos espectros das partículas selecionadas e, por fim, confirmação da composição do MP comparando com o espectro de um padrão. id d i ífi l l bl d é d A HSI utilizando a espectroscopia Raman foi aplicada na análise de MP em amostras de areia de praia. Imagens com áreas de 88 × 88 µm e 55 × 55 µm foram adquiridas em amostras peneiradas com tamanho de poro de 125 µm. As amostras foram lavadas com etanol e água ultrapura antes das medidas. Raman A espectroscopia Raman já foi reportada para comprovar a ocorrência de MP em ambientes aquáticos (oceanos, mares, rios, lagos, estuários e seus respectivos sedimentos),199-214 terrestres,72 atmosféricos159,166 e em organismos.86,97,215-220 Complementar às espectroscopias NIR e MIR, a espectroscopia Raman se destaca pela sua capacidade de alcançar resolução espacial sub-micrométrica, permitindo preencher uma lacuna deixada pela espectroscopia MIR, que é a análise de MP menores do que 20 µm.221 A espectroscopia Raman é uma técnica essencialmente microscópica que apresenta limite de difração muito menor que aquele alcançado pela μ-FTIR (de até 266 nm utilizando laser 532 nm, por exemplo). Não apresenta influência significativa do sinal da água e possui bandas estreitas que facilitam a caracterização química. Porém, apresenta alguns inconvenientes como: i) um sinal de fundo devido à fluorescência e ao aquecimento da amostra, provocado pelo laser, que pode degradar amostras sensíveis alterando seu estado físico-químico; ii) necessita de maior tempo de integração do sinal espectral devido ao fraco sinal Raman e, por isso, maior tempo de análise e iii) possui mais variáveis instrumentais (comprimento do laser, lente objetiva, ajuste do foco do laser na amostra, grade de difração, entre outros) que dificultam a padronização Importante observar que MIR é uma técnica muito sensível à 1338 Montagner et al. Quim. Nova Quim. Nova do método, mas também, abrange as possibilidades de análise. do método, mas também, abrange as possibilidades de análise. conhecidos dessa técnica vibracional. Além disso, técnicas como aquisição de imagens hiperespectrais (HSI) por Raman, Espalhamento Coerente de Raman Anti-Stokes (CARS, do inglês Coherent Anti- Stokes Raman Scattering) e Espalhamento Raman Estimulado (SRS, do inglês Stimulated Raman Scattering) também têm sido exploradas para análise de MP. Considerando os aspectos citados acima, a resolução espacial é a principal característica que trouxe a espectroscopia Raman para análise de MP em amostras ambientais e impulsionada pela demanda por análises de MP menores que 20 μm. Nesse tamanho, eles são mais facilmente absorvidos pelos organismos e possuem propriedades físico-químicas diferentes dos MP maiores. HSI Raman, em que na maior parte dos casos utiliza-se a técnica point-scan, tem um custo de tempo de aquisição muito elevado, podendo facilmente passar de dias, se for utilizada uma alta resolução espacial (1 µm, por exemplo) para medir uma área com centenas de micrômetros quadrados. Outro ponto importante é o elevado volume de dados adquiridos que chega facilmente a centenas de gigabytes por imagem. Raman A comunidade científica levantou alguns problemas desse método de análise utilizando a inspeção visual como base de seleção de partículas apresentadas à espectroscopia Raman para análise de MP < 20 μm. O principal é o viés entre partículas elencadas como suspeitas e os MP realmente presentes. A taxa de sucesso na identificação de partículas de MP é baixa para aquelas menores que 100 μm e com cores e formas similares aos detritos orgânicos e minerais da matriz, confundindo o analista, enquanto que as maiores taxas de sucesso foram obtidas para partículas maiores que 100 μm, com cores mais vivas (azul por exemplo) e com forma de fibra.223 O grau de degradação, presença de aditivos e variabilidades de composição influenciam na análise dos MP por espectroscopia Raman. As fragmentações das cadeias poliméricas causam alterações nas intensidades das bandas no espectro. A oxidação, a sorção de compostos e os aditivos podem acrescentar bandas ao espectro ou mascarar as bandas dos MP. Os biofilmes provocam fluorescência que pode impedir a caracterização do MP, sendo esse último um dos principais inconvenientes da análise de MP em amostras ambientais utilizando Raman. Nesse sentido, alguns autores propuseram métodos de digestão branda da matéria orgânica nas amostras de MP. A utilização de solução de KOH como agente de digestão faz parte de um desses métodos e com a utilização da espectroscopia Raman como método de confirmação, vantagens e desvantagens desse preparo de amostra foram observados na análise dos MP degradados.224 Espectroscopia fototérmica no infravermelho Espectroscopia fototérmica no infravermelho As inovações publicadas na literatura que descrevem métodos de análise de MP em amostras ambientais utilizando a espectroscopia Raman, em sua maior parte se resumem a novos preparos de amostra a fim de melhorar a resposta espectral ou contornar os problemas já A espectroscopia fototérmica excitada por radiação IR (MIP, do inglês mid-IR Photothermal) é uma técnica bastante recente.227 Essa técnica tem várias implicações como a não influência de sinal de Microplásticos: ocorrência ambiental e desafios analíticos Vol. 44, No. 10 1339 referência e o MP formulado e degradado do mesmo polímero pode ser prejudicada. fundo e alta velocidade de aquisição, pois é limitada somente pelo tempo de relaxação térmica. Além disso, possui alta sensibilidade assim como no IR e permite aquisição de espectros IR por meio de medidas por reflectância. Uma das vantagens da técnica é a resolução espacial similar aos espectrômetros Raman, devido ao uso do mesmo tipo de laser no visível, ou seja, permite a aquisição de espectros no IR com resolução sub-micrométrica.198 Uma biblioteca livre contendo mais de 250 espectros Raman de plásticos comerciais e de amostras de MP coletadas no ambiente foi publicada como parte de um esforço de ampliar a acessibilidade da espectroscopia Raman neste contexto. Os materiais obtidos apresentaram variadas cores e morfologias. Algumas variáveis instrumentais foram acrescentadas como fontes de variabilidade espectral como o tipo do laser, filtro de densidade, grade de difração, tamanho da fenda, tempo de aquisição, número de acumulações, entre outras, a fim de enriquecer o banco de dados.233 Bancos de dados espectrais Raman e FTIR de MP coletados no ambiente também foram adquiridos e nesse caso o principal objetivo foi o estudo das modificações físico-químicas causadas na superfície dessas partículas e, por isso, imagens obtidas por SEM também foram utilizadas na análise.234 Embora seja uma técnica bastante recente, um equipamento comercial que emprega essa tecnologia já está disponível. O instrumento permite medir o espectro IR pelo efeito fototérmico e o espalhamento Raman simultaneamente.228 Na Figura 4 é mostrada a análise de MP obtido de ambiente marinho por meio de imagem hiperespectral, onde foram identificadas partículas de PS e polimetilmetacrilato (PMMA) dispersas em sal sobre uma lâmina de CaF2. A imagem química de dimensões 200 x 150 µm levou 20 min para ser adquirida monitorando somente dois comprimentos de onda, sintonizados no máximo de bandas do PMMA e do PS, 1730 e 1492 cm-1, respectivamente. Espectroscopia fototérmica no infravermelho Em destaque na imagem óptica pode-se observar a identificação de uma partícula de PS com aproximadamente 900 nm de diâmetro. A espectroscopia MIP foi descrita recentemente em um estudo de revisão dentre a perspectiva global sobre MP como uma técnica capaz de analisar partículas sub-micrométricas.229 Outra tendência para análises espectroscópicas mais rápidas e confiáveis é a identificação do polímero empregando algoritmos para processamento dos espectros e para reconhecimento de padrões espectrais, desenvolvido por meio de métodos quimiométricos (um tipo de aprendizado de máquinas, do inglês machine learning, voltado ao tratamento de dados químicos), os quais são embutidos no software dos equipamentos ou desenvolvidos in-house.235-243 Apesar da análise de MP utilizando técnicas quimiométricas e de processamento de HSI ser bastante informativa, a complexidade e o custo não a tornam tão popular quanto a inspeção visual e espectroscopia convencional, que são mais simples e de menor custo para serem implementadas às análises de rotina. No entanto, a capacitação de especialistas que tenham domínio sobre a técnica de aquisição e de tratamento dos dados multivariados, tem se mostrado uma demanda crescente importante para a área de química ambiental na temática de MP. Confirmação da identidade dos polímeros e bibliotecas espectroscópicas Confirmação da identidade dos polímeros e bibliotecas espectroscópicas Como pode ser notado, a identificação do MP de origem ambiental é um desafio de alta complexidade. Um dos fatores é que as confirmações baseadas em bibliotecas espectrais de polímeros puros ignoram as alterações dos MP que sofreram ação ambiental podem apresentar. Outra dificuldade está em produzir bibliotecas que englobem a enorme variabilidade química que MP podem apresentar, especialmente quando expostos às intempéries.230 Técnicas termoanalíticas A d i é Uma das bibliotecas de espectros mais utilizadas para caracterização dos materiais plásticos é chamada KnowItAllTM database, a qual contém dados espectrais de diversas técnicas incluindo IR e Raman.166,223,231 Nessa base de dados espectrais são encontrados espectros de praticamente todos os polímeros que compõem os MP. O software calcula um índice de similaridade espectral (chamado de HQI, do inglês Hit Quality Index) e ranqueia os compostos mais prováveis para o material desconhecido. As considerações para o cálculo do HQI são demonstradas na norma ASTM E2310-04.232 No entanto, os dados dessa biblioteca são de materiais de referência, sendo que a construção de base de dados com polímeros degradados é uma demanda emergente para a análise de MP, visto que a similaridade de espectros entre um polímero de A decomposição térmica controlada de um material, sob atmosfera inerte, é denominada pirólise (py). Ao serem aquecidos, os MP passam por transformações físico-químicas, degradam-se e perdem massa nesse processo. A caracterização de MP por técnicas térmicas é fundamentada nesses fenômenos e a principal informação que elas fornecem é a identidade do polímero por meio da análise de seus produtos de degradação, podendo inclusive ser feita a detecção de aditivos ou contaminantes sorvidos nos MP. Os fragmentos de maior abundância são geralmente monômeros, dímeros ou trímeros formados pela decomposição da cadeia polimérica do MP.244 A priori, as técnicas por degradação térmica não fornecem a informação de tamanho do MP, entretanto, ele pode ser conhecido se houver um pré-tratamento de amostra que separe as frações antes da análise. Figura 4. Imagem óptica, imagem química infravermelho (IR) e espectros das partículas indicadas pelas setas, adquiridas utilizando espectroscopia fototérmica no IR. Os dados são uma cortesia do fabricante do instrumento mIRage da empresa Photothermal Spectroscopy Corp Figura 4. Imagem óptica, imagem química infravermelho (IR) e espectros das partículas indicadas pelas setas, adquiridas utilizando espectroscopia fototérmica no IR. Os dados são uma cortesia do fabricante do instrumento mIRage da empresa Photothermal Spectroscopy Corp 1340 Montagner et al. Quim. Nova Quim. Nova A técnica de py-GC-MS consiste em um pirolisador acoplado a um GC‑MS ou diretamente no espectrômetro de massas (py-MS), em que o MP é inserido no pirolisador. Estratégias multitécnicas A comparabilidade do desempenho das mesmas técnicas entre diferentes trabalhos fica limitada pela falta de padronização dos métodos, sendo que a comunidade científica discute sobre a harmonização de procedimentos. O princípio de cada técnica leva a resultados quantitativos expressos em unidades diferentes, baseados em massa ou baseados em número de partícula, o que compromete a comparação de resultados de ocorrência entre estudos, ou mesmo de remoção.184 Por exemplo, as técnicas termoanalíticas acopladas à cromatografia tem resultados expressos em concentração, μg L‑1 para matrizes líquidas ou mg kg-1 para matrizes sólidas e fornecem informação de teores totais, independentemente do número e distribuição de tamanho de partículas inseridos no instrumento. As técnicas espectroscópicas reportam contagens, ou seja, número de MP extraídos de um determinado volume ou massa de amostra, sem a informação da massa total de MP durante a análise. Outra consideração são as particularidades de cada MP, por exemplo, um MP muito degradado poderia quebrar durante uma análise por ATR-FTIR, sugerindo que fosse analisado por Raman. Porém, se o mesmo MP fosse muito pigmentado ou escuro, teria interferência de fluorescência ou não teria sinal no Raman, sugerindo que fosse analisado por FTIR. Outra maneira de explorar os produtos de degradação de MP para identificá-los é pela utilização de análises térmicas. Tanto TGA quanto calorimetria exploratória diferencial (DSC, do inglês differential scanning calorimetry) fornecem informação das relações entre uma propriedade da amostra e temperatura, ao submetê-la a rampas de aquecimentos e/ou resfriamentos controlados em determinada atmosfera. Essas análises térmicas não são em geral utilizadas isoladamente no contexto de análise de MP, dado que os MP têm histórias térmicas únicas pela exposição ambiental, o que influencia suas propriedades e a curva obtida,254 além dos MP secundários terem sido processados com formulações e condições diversas. Assim, o mais comum é a utilização de TGA combinada ou acoplada a outras técnicas, obtendo as informações de variação de massa em função de temperatura por TGA e a caracterização do gás liberado durante a pirólise por MS ou GC-MS,248 sendo que os gases liberados podem ou não ser retidos em uma fase sólida para posterior dessorção térmica em uma unidade de dessorção térmica (TDU, do inglês thermal desorption unit ou TDS, thermal desorption system) e análise por MS ou GC-MS. Estratégias multitécnicas Essa configuração combinando TGA, sorção dos voláteis na fase sólida e posterior dessorção térmica é encontrada na literatura como TED-GC-MS, sendo TED do inglês thermal extraction desorption.255,256 Essa combinação permitiu analisar uma quantidade maior de amostras, sendo uma vantagem de TED-GC- MS sobre py-GC-MS. Por conta de temperaturas diferentes, limpeza da amostra e etapa de extração/dessorção após a pirólise, os perfis de decomposição obtidos por TED-GC-MS podem ser ligeiramente diferentes que os obtidos por py-GC-MS.244 Assim, não há técnicas melhores ou piores, mas aquelas mais adequadas em função do objetivo do estudo, dado que tamanhos, particularidades e matrizes podem favorecer uma ou outra técnica de confirmação química. Em função das potencialidades e limitações de cada uma, existe uma tendência para se utilizar uma estratégia multitécnica na caracterização química de MP, explorando que as técnicas são, na verdade, complementares. Uma contribuição nesse sentido foi dada por Elert e colaboradores, que padronizaram uma amostra para ser analisada qualitativamente por quatro técnicas, com o objetivo de verificar se detectavam o MP ao final. Partículas de até 125 μm de PE, PP, PS e PET foram intencionalmente adicionadas em um solo a uma concentração de 1% m/m de cada polímero, totalizando 4% m/m no solo e, em seguida, a amostra sintética foi analisada por espectroscopia vibracional combinada com imagem (µ-FPA-FTIR em modo de transmissão e Raman), TED-GC-MS e SEC.187 Os polímeros não continham aditivos, o que a priori excluiu interferências causadas por eles, dado que, por exemplo, a fluorescência de um pigmento poderia desfavorecer o Raman, enquanto que a presença de outros aditivos poderia ser um fator desfavorável para TED-GC-MS ou µ-FPA-FTIR. Observa-se que apesar da amostra ter sido padronizada em função do tipo, concentração e tamanho de MP, cada técnica demandou uma quantidade diferente de amostra, consequentemente, na quantidade de MP introduzida ao instrumento. Como resultado, exceto por SEC, as técnicas detectaram os 4 polímeros, entretanto, diferentes aspectos foram adicionalmente discutidos. As técnicas baseadas em massa, TED-GC-MS e SEC, utilizaram quantidades maiores de solo, sendo menos vulneráveis à falta de homogeneidade; ressalta-se que elas possibilitam calibração para posterior quantificação, mas não fornecem diretamente informação de tamanho e número de partículas. Técnicas termoanalíticas A d i é O uso de atmosfera inerte é importante para garantir a reprodutibilidade da pirólise, que também é influenciada pelo tipo de pirolisador utilizado e a matéria orgânica residual no MP, podendo-se utilizar um padrão interno no processo de pirólise.245,246 Quando a introdução da amostra ao pirolisador antes da análise cromatográfica é feita com o MP isolado, o tamanho do MP a ser analisado é limitado pela habilidade do analista no manuseio da amostra e é da ordem de 100 μm.247,248 Entretanto, avanços recentes no desenvolvimento de métodos por py-GC-MS reportam a análise do MP diretamente na matriz, o que pode possibilitar o estudo de MP menores.246 Solos fortificados com PE, PP e PS na concentração de 50 a 250 μg g-1 foram solubilizados com 1,2,4-triclorobenzeno (TCB) e alíquotas com até 4 g de solo foram analisadas por py-GC-MS, apresentando recuperações de 70-128% para o nível de 250 μg g-1 para solos com até 2,5% de carbono orgânico.249 Em outro trabalho, a análise de MP menores que 100 μm em água foi feita pela filtração da água e deposição dos MP em uma membrana de PTFE, posteriormente cortada a laser e então submetida à py-GC-MS, utilizando MS de alta resolução para maior detectabilidade dos produtos de degradação. A membrana de PTFE foi escolhida pela alta resistência térmica do material e uma prova de conceito feita em 200 mL de água do rio Tawe (Reino Unido), sugerindo a ocorrência de PS em tamanho da ordem submicrométrica nesse rio.250 A combinação de extração com líquido pressurizado (PLE, do inglês pressurized liquid extraction) e py-GC-MS atingiu limites de detecção dos MP mais comuns (PE, PET, PS, PP) na faixa de 7 a 30 μg g-1 para matrizes complexas como solo, sedimento e lodo de esgoto.251,252 Por fim, outro avanço foi a avaliação de instrumento py-MS portátil como potencial ferramenta para análise de MP em campo.253 sido menos utilizadas que as espectroscópicas na análise de MP,82 elas têm ganhado importância e são consideradas promissoras para a análise de nanoplásticos. Duas revisões recentes sobre o uso de técnicas térmicas apresentam uma tendência de sua aplicação para a matriz sedimento, bem como discutem as particularidades das técnicas em função de fundamentos, preparo de amostra, interferência da matriz, representatividade da subamostragem, limites de detecção, quantificação e validação na análise de MP.244,246 Estratégias multitécnicas Por outro lado, as técnicas espectroscópicas combinadas com imagem analisaram uma quantidade menor de amostra, o que Como características, tanto py-GC-MS quanto TED-GC-MS são técnicas destrutivas e fornecem resultados em base de concentração para a quantificação de MP, permitindo análise simultânea de diferentes polímeros, podendo atingir níveis de detecção e quantificação na faixa de nanogramas. Embora até o momento as técnicas térmicas tenham Microplásticos: ocorrência ambiental e desafios analíticos 1341 Vol. 44, No. 10 digitais264,271,272 das HSI que foram discutidas na seção de técnicas espectroscópicas. Ao final do processamento, ambas fornecem informações sobre forma e tamanho de partículas, entretanto, apenas HSI fornecem identidade química do polímero (PE, PP, entre outros), dado que a imagem é obtida em função do espectro IR ou Raman de cada pixel. poderia comprometer um protocolo de quantificação pela baixa amostragem. No entanto, forneceram diretamente a identidade, número e tamanho de partículas. Assim, Elert e colaboradores sugerem que o mais importante na escolha da técnica é decidir qual a questão que se quer responder da mesma: em casos que tamanho de partícula e grau de degradação do MP importam, as espectroscópicas seriam mais indicadas, enquanto TED-GC-MS e SEC responderiam mais rapidamente com quais tipos de MP o solo estaria contaminado, sem fornecer outras informações de cada partícula individualmente. Além das técnicas já discutidas nesse trabalho, já foi reportado na literatura o uso de TGA-FTIR-GC-MS, sendo que nesse caso a análise de IR é feita no gás evoluído da TGA,273 técnicas de ionização e dessorção acopladas à MS de alta resolução para a caracterização química.274 Assim, estabelecido o objetivo da análise, árvores de decisão baseadas na pergunta que se deseja responder no estudo podem ser úteis para guiar a escolha da técnica de análise, como a proposta por Zarfl275 e Lusher e colaboradores.176ii A técnica SEM-EDS fornece detalhes da morfologia na superfície da partícula e da composição elementar (de carbono e outros elementos), mas sem identificar de fato o polímero. Para a finalidade de identificar, é necessário associar um preparo de amostras que vise separar as partículas (por densidade, por exemplo) e considera-se que as partículas são MP em função do isolamento e da remoção de matéria orgânica durante o preparo de amostras. O CENÁRIO NO BRASIL Para finalizar esse trabalho de revisão, uma pesquisa na literatura sobre a ocorrência de MP em matrizes ambientais brasileiras foi realizada a fim de trazer o cenário de contaminação reportado até o momento sobre a presença de MP e os métodos analíticos que vêm sendo empregados (Tabela 1). A revisão foi baseada em artigos indexados nas bases de dados Scielo e Web of Science, publicados entre 2004 (primeiro trabalho nestas matrizes no Brasil) e dezembro de 2020, utilizando uma busca avançada combinando os termos “microplastic/microplástico”, “marine litter/lixo marinho”, “plastic debris/detritos de plástico”, “plastic pellets/pellets plásticos”, “plastic pollution/poluição plástica”, “marine debris/detritos marinhos”, “Brazil/Brasil”, selecionando aqueles com foco em ocorrência. O procedimento de corar os MP, a fim de reconhecê-los mais facilmente através da fluorescência, chamou a atenção de diversos grupos de pesquisa, que logo desenvolveram e otimizaram as condições de coloração em amostras ambientais.262-264 Diversos trabalhos passaram a adotar o corante em seus métodos de análise de MP, a fim de melhorar a taxa de sucesso na identificação dessas partículas utilizando a inspeção visual.262,265-270 Ao todo, foram considerados 80 trabalhos, dos quais 61 referem- se aos estudos na matriz sedimento (incluindo areia de praia) e 19 na matriz água. Nenhum trabalho sobre MP na atmosfera e no solo foi encontrado no Brasil, considerando o método de busca de publicações empregado neste processo de revisão. Dos 19 trabalhos sobre a matriz água, 7 referem-se à água salobra, 9 à água salgada e 3 à água doce, conforme mostra a Figura 5a. Os ecossistemas marinhos foram de forma geral, os mais estudados quando comparados aos sistemas de água doce. Embora esses últimos sejam considerados como as maiores fontes de aporte dos resíduos plásticos para os oceanos, apenas recentemente vêm recebendo mais atenção da comunidade científica. Do total de trabalhos selecionados nesta revisão, apenas 8% avaliaram a presença de MP em ecossistemas (água/sedimento) de águas doces, enquanto 92% foram realizados nos oceanos e regiões costeiras (incluindo os sistemas de água salobra). Considerando a abrangência dos estudos ao longo de todo o território brasileiro, 44% das amostragens foram realizadas na região sudeste, seguido pela região nordeste (35%), sul (17%) e norte (4%), como pode ser visualizado na Figura 5b. Em artigo recente, Prata e colaboradores264 concluíram que dentre uma série de corantes testados, o Nile Red foi o mais adequado para o desenvolvimento do método de baixo custo proposto para contagem de partículas de MP. Estratégias multitécnicas A posterior análise por SEM visaria uma caracterização morfológica de um MP já identificado e caracterizado por uma técnica espectroscópica, fornecendo detalhes da superfície degradada em resoluções na ordem nanométrica, ou a presença de outros elementos por SEM-EDS. Em um trabalho de avaliação da exposição humana a MP, reportou-se o emprego de SEM-EDS como método de identificação e quantificação dos MP,118 entretanto, foi refutado sob o argumento de que a utilização do teor de carbono não seria suficiente para a confirmação, além de outras falhas analíticas.257,258 Por fim, a caracterização química e/ou quantificação de MP é a última etapa do método analítico, que já propagou os erros devido à amostragem, subamostragem e preparo de amostra, e isso deve ser considerado. A qualidade analítica do método como um todo foi avaliada por Koelmans e colaboradores, baseada na publicações envolvendo a análise de MP em água doce e água potável.145 Foram avaliados os parâmetros como tipo e volume de amostragem, tamanho de partícula alvo, armazenamento e processamento de amostras, local apropriado (ex: sala limpa) para evitar contaminação durante o processamento da amostra, controles negativos como amostra branco, controles positivos e ensaios de recuperação com amostras enriquecidas com quantidades ou concentrações conhecidas de MP e, por fim, o método de caracterização química do tipo do polímero, demonstrando os desafios no contexto global da análise de MP.145 Outra estratégia bastante utilizada na análise de MP é o uso do corante vermelho do Nilo (do inglês Nile Red), um composto fluorescente que auxilia a inspeção visual. Esse corante é originalmente utilizado para corar o tecido lipídico de células, pois é um composto bastante hidrofóbico. Isso faz com que ele core facilmente vários tipos de materiais plásticos, assim, sua visualização na microscopia de fluorescência torna-se favorecida.259,260 Por exemplo, foi desenvolvido um método de fabricação de microfibras de materiais plásticos (PA, PET e PP com 10-28 µm de diâmetro e 40-100 µm de comprimento), para uso em estudos sobre a ingestão de MP por organismos. A motivação foi a dificuldade de encontrar microfibras comerciais para tais estudos. Nesse trabalho, o corante foi usado para imagear as microfibras dentro do trato intestinal de artêmias.261 i O CENÁRIO NO BRASIL No entanto, como as partículas podem se aderir ao restante do material retido nas redes durante a coleta, foram encontrados, por exemplo, MP com tamanho entre 5 e 100 µm em uma abundância de cerca de 20% no lago Guaíba no Rio Grande do Sul.286 Além disso, a maioria dos trabalhos realizados na matriz água Figura 5. Do total de 80 trabalhos publicados em revistas indexadas até 2020 sobre microplásticos e resíduos plásticos nas matrizes sedimento e água no Brasil temos, (a) porcentagens de estudos por matrizes ambientais e (b) porcentagens de estudos realizados nas diferentes regiões do Brasil A Tabela 1 descreve os trabalhos publicados com foco em ocorrência de MP e de resíduos plásticos no Brasil em matrizes ambientais (água salobra, água salgada, água doce e sedimento) e ilustra a falta de harmonização de métodos discutida anteriormente. Tabela 1. O CENÁRIO NO BRASIL O método consiste na análise de imagem digital, da fluorescência do corante Nile Red, depois de corar as partículas de MP da amostra. As imagens foram adquiridas utilizando uma câmera fotográfica convencional. Vários comprimentos de onda de excitação da fluorescência foram testados bem como vários tipos de MP virgens e degradados. A validação da eficiência do método proposto foi realizada com base em espectros no IR, a qual obteve um percentual de recuperação entre 89% e 111%. O método proposto tem forte apelo de aplicação devido seu baixo custo e devido ao ganho de velocidade e automação na contagem das partículas que eliminam a subjetividade da inspeção visual. Por outro lado, o método não responde sobre a composição química das partículas de MP e sobre a cor original, uma vez que a amostra necessita passar por processo de digestão. Nesse caso, o método poderia ser empregado como uma análise complementar. Importante diferenciar as técnicas de imagens 1342 Montagner et al. Quim. Nova Quim. Nova Figura 5. Do total de 80 trabalhos publicados em revistas indexadas até 2020 sobre microplásticos e resíduos plásticos nas matrizes sedimento e água no Brasil temos, (a) porcentagens de estudos por matrizes ambientais e (b) porcentagens de estudos realizados nas diferentes regiões do Brasil A areia de praia foi a matriz mais estudada em território brasileiro e isso pode ser facilmente justificado devido à facilidade de amostragem nesse compartimento, em que a coleta é feita manualmente, sem o uso de equipamentos específicos. Os pellets são um bom exemplo de materiais que podem ser coletados a olho nu, com o auxílio de pinças. Na região litorânea do Brasil, a presença de pellets tem sido evidenciada tanto em estudos que visam quantificar contaminantes associados311,312,315,320,321,334 ou somente para identificá- los em meio ao resíduo marinho.89,174,276,299 Em relação à amostragem na matriz água (salobra, salgada e doce), observou-se uma ampla abrangência de faixas de tamanho de MP devido à utilização de redes com diferentes tamanhos de malha, pois esses variaram de 60 µm a 42 mm. Com isso, seria esperado que MP com tamanho inferior a 60 µm não fossem encontrados nesse compartimento ambiental. O CENÁRIO NO BRASIL Ocorrência de microplásticos (MP) e resíduos plásticos em matrizes ambientais de água e sedimento no Brasil Local Amostragem e preparo de amostra Identificação Caracterização química Quantificação Referência Água salobra Estuário do rio Goiana/PE Rede de plâncton de 300 µm; secagem a 60 °C Microscopia óptica - 14.724 MP (26,04 MP 100 m-3) 133 Estuário do rio Goiana/PE Rede de plâncton de 300 µm; secagem a 60 °C Microscopia óptica - 26,1 MP 100 m-3 276 Estuário do rio Goiana/PE Rede de 1000 µm Microscopia óptica - 1662 MP (3,4 MP 100 m-3) e 2710 plásticos (1,4 itens 100 m-3) 134 Estuário do rio Goiana/PE Rede de plâncton de 300 µm Microscopia óptica - 26,06 MP 100 m-3 277 Estuário do rio Paraíba/PB Rede de 5 mm; secagem a 60 °C Inspeção visual - 88,37% dos itens coletados eram plásticos 278 Estuário de Paranaguá/PR Rede de arraste de 26 e 42 mm Inspeção visual - 291 itens (92,4% plásticos) 279 Estuário Santos-São Vicente/SP Coleta manual de resíduos flutuantes Inspeção visual - 2339 itens (89,64% plásticos) 280 Água salgada Arquipélagos São Pedro e São Paulo/PE Rede de plâncton de 300 µm Microscopia óptica - 99 MP e 22 itens plásticos (47,2% das amostras contaminadas com plásticos) 57 Arquipélagos São Pedro e São Paulo/PE Rede de plâncton de 300 µm Microscopia óptica 71 itens (1 item 100 m-3) 89 Ilhas de Fernando de Noronha/PE, Abrolhos/BA e Trindade/ES Rede manta de 300 µm; filtração e secagem Microscopia óptica - 243 partículas plásticas (1,52 partículas por rede) 56 Litoral do Atlântico entre Maranhão e Ceará Redes de plâncton de 120 e 300 µm Microscopia óptica - 0,14 MP m-3 (rede de 120 µm) e 0,02 MP m-3 (rede de 300 µm) 281 Enseada de Jurujuba/RJ Rede de plâncton 150 µm; filtração em membrana de acetato de celulose e peneiração em 7 malhas Microscopia óptica ATR-FTIR (partículas > 1 mm) 16,4 MP m-3 130 Baía de Guanabara/RJ Rede de plâncton 300 µm; peneiração em 2 malhas, remoção de matéria orgânica com H2O2, filtração em 300 µm Microscopia óptica ATR-FTIR (partículas > 1 mm) 1,4-21,3 MP m-3 282 Ocorrência de microplásticos (MP) e resíduos plásticos em matrizes ambientais de água e sedimento no Brasil Microplásticos: ocorrência ambiental e desafios analíticos 1343 Vol. 44, No. 10 Microplásticos: ocorrência ambiental e desafios analíticos 1343 Vol. 44, No. O CENÁRIO NO BRASIL 10 Local Amostragem e preparo de amostra Identificação Caracterização química Quantificação Referência Baía de Guanabara/RJ Água superficial em frascos e plásticos flutuantes com rede de pesca Inspeção visual ATR-FTIR 14 itens caracterizados 283 Baía de Guanabara/RJ Água superficial com rede de plâncton 200 e 64 µm e nêuston 64 µm Microscopia óptica µ-FTIR 0,6 a 11 MP m-3 284 Niterói/RJ Água superficial com rede de plâncton 120 µm; separação com solução de NaCl, filtração em 0,45 µm, secagem a temperatura ambiente Microscopia óptica ATR-FTIR 1319 itens (7,62 itens m-3) 285 Água doce Lago Guaíba/RS Água superficial com rede de plâncton 60 µm; peneiração em 63 µm, remoção de matéria orgânica com H2O2/Fe(II)/ H2SO4, filtração em membrana de acetato de celulose 0,45 µm, separação com NaI, filtração em membrana de fibra de vidro 0,45 µm Microscopia óptica µ-FTIR e Raman 9519 MP (11,9-61,2 itens m-3) 286 Rio Sinos/RS Água superficial e água tratada com frascos de vidro; remoção de matéria orgânica com KOH, filtração em membrana de fibra de vidro 7 µm, separação com NaCl, filtração em membrana de fibra de vidro 7 µm, adição de solução de vermelho do Nilo, secagem a 40 °C Microscopia de fluorescência - 330,2 partículas L-1 (água superficial) e 105,8 partículas L-1 (água tratada) 287 Lagoa Acaraí/SC Água superficial com rede de plâncton 300 µm; filtração em 500 µm, secagem a 70 °C Microscopia óptica - 1,4-3,4 itens 100 m-3 288 Sedimento Rios do Amazonas Sedimento; draga Van Veen, secagem a 50 °C, peneiração em 63 µm, remoção de matéria orgânica com H2O2, separação com ZnCl2, filtração em membrana de celulose 18 µm Microscopia óptica - 417-8178 MP kg-1 (faixa de 0,063-5 mm) e 0-5725 MP kg-1 (faixa de 0,063-1 mm) 289 Ilha Cotijuba/PA Areia de praia; diferentes profundidades; peneiração em 300 µm; separação com solução de NaCl; filtração em membrana de celulose 2 µm, secagem a 35 °C Microscopia óptica - 3040-20.300 partículas m-3 290 Praia da Corvina/PA Areia de praia; peneiração em 250-500 µm e 500-5000 µm Microscopia óptica - 5819 partículas (492,5 partículas m-3) 291 Lençóis Maranhenses/MA Areia de praia Inspeção visual ATR-FTIR 88 partículas 292 Rio Potengi/RN Sedimento Inspeção visual - ~400 plásticos no período chuvoso e ~900 plásticos no período seco 293 Praia Mansa de Fortaleza/CE Areia de praia Inspeção visual - 7510 itens (0,21-1,15 itens m-2) 294 Estuário do rio Goiana/PE Sedimento; corer cilíndrico, peneiração em 1 mm Inspeção visual - 6,36–15,89 itens m-2 295 Estuário do rio Goiana/PE Areia de estuário Inspeção visual - 6944 itens (> 95% plásticos) 296 P i d B Vi /PE A i d i I ã i l 32.045 itens 297 Tabela 1. O CENÁRIO NO BRASIL Ocorrência de microplásticos (MP) e resíduos plásticos em matrizes ambientais de água e sedimento no Brasil (cont.) 1. Ocorrência de microplásticos (MP) e resíduos plásticos em matrizes ambientais de água e sedimento no Brasil (cont.) 1344 Montagner et al. Montagner et al. Quim. O CENÁRIO NO BRASIL Nova Local Amostragem e preparo de amostra Identificação Caracterização química Quantificação Referência Praia de Boa Viagem/PE Areia de praia; peneiração em 1 mm; separação com solução de NaCl; filtração em membrana de celulose 2 µm; secagem a 40 °C Microscopia óptica - 838 itens 298 Praia de Boa Viagem/PE Areia de praia; secagem a 100 °C e peneiração em 2 malhas Microscopia óptica - 90 pellets (0,01 pellet cm-2) 2661 fragmentos plásticos (2-231 fragmentos cm-2) 299 Litoral de Pernambuco Areia de praia Inspeção visual - 12.815 itens (57,3% plásticos) 300 Praias de Fernando de Noronha/PE Areia de praia; peneiração em 1 mm, separação com solução de NaCl, filtração em membrana de fibra de vidro 1 µm, secagem a 40 °C Microscopia óptica - 504 partículas (11,1-455,6 partículas m-2) 301 Arquipélago de Fernando de Noronha/PE Areia de praia; secagem a 100 °C e peneiração em 1 mm Inspeção visual - 207 itens (93,5% plásticos) 55 Costa dos Coqueiros/BA Areia de praia Inspeção visual - 1974 itens (69,8% plásticos) 302 Praia da Costa do Dendê/BA Areia de praia Inspeção visual - 7189 itens (76% plásticos) 303 Litoral da Bahia Areia de praia Inspeção visual - 6751 itens (52-94% plásticos) 304 Praias de Salvador/BA Areia de praia Inspeção visual - 7858 itens (87,1% plásticos) 305 Litoral sul da Bahia Areia de praia Inspeção visual - 1430 pellets 306 Praias de Salvador/BA Areia de praia; separação com água do mar Inspeção visual - 1967 pellets 307 Praias da Ilha de Itaparica/BA Areia de praia Inspeção visual - 2220 itens na estação (49% plásticos) e 508 itens na baixa estação (59% plásticos) 308 Litoral de Sergipe Areia de praia Inspeção visual - 1484 itens (~ 85% plásticos) 309 Reservatório de água de chuva/SP Sedimento; separação com solução de ZnCl2; filtração em membrana de nitro celulose e secagem a temperatura ambiente Microscopia óptica ATR-FTIR e Raman 57.542 MP kg-1 310 Praias de Santos/SP Areia de praia; pinça para coleta de pellets Inspeção visual - Concentração de orgânicos sorvidos em pellets 311 Praias de Santos/SP Areia de praia; pinça para coleta de pellets, separação com solução de H2O/etanol Inspeção visual - Concentração de orgânicos sorvidos em pellets 312 Praias de Santos/SP Areia de praia; profundidade de até 1 m com trado, separação com água do mar e peneiração em 1 mm Inspeção visual - Concentrações de pellets em diferentes profundidades 171 Praias de Santos/SP Areia de praia Inspeção visual - Até 260-300 pellets m-2 313 Praias de Santos/SP Areia de praia Inspeção visual - 456 itens (84% plásticos) 314 Praias de Santos/SP Areia de praia; profundidade de até 25 cm com corer cilíndrico, separação com água do mar e peneiração em 1 mm Inspeção visual - 13.138 pellets 174 Ponta da Praia em Santos/SP Areia de praia; profundidade de até 1 m com pá de metal, peneiração em malha fina Inspeção visual - Concentração de orgânicos sorvidos em pellets 315 Tabela 1. O CENÁRIO NO BRASIL Ocorrência de microplásticos (MP) e resíduos plásticos em matrizes ambientais de água e sedimento no Brasil (cont.) cos (MP) e resíduos plásticos em matrizes ambientais de água e sedimento no Brasil (cont.) Microplásticos: ocorrência ambiental e desafios analíticos 1345 Vol. 44, No. 10 Vol. 44, No. O CENÁRIO NO BRASIL 10 Local Amostragem e preparo de amostra Identificação Caracterização química Quantificação Referência Estuário de São Vicente/SP Areia de praia Inspeção visual - 2129 itens (37,28% plásticos) 316 Praias de Santos/SP Areia de praia; peneiração em 0,1 mm Inspeção visual - Até 2,5 pellets m-1 317 Litoral de São Paulo Areia de praia; profundidade de até 1 m com trado, peneiração em 1 mm Inspeção visual Raman 18% PP, 78,2% PE e 3,8% mistura de PP e PE 318 Litoral de São Paulo Areia de praia; profundidade de até 1 m com trado, peneiração em 1 mm Inspeção visual - Média máxima de cerca de 140 pellets m-2 319 Litoral de São Paulo Areia de praia; pinça para coleta de pellets Inspeção visual - Concentração de orgânicos sorvidos em pellets 320 Litoral de São Paulo Areia de praia Inspeção visual - Concentração de metais sorvidos em pellets 321 Estuário de Santos e São Vicente/SP Sedimento; draga Van Veen, secagem a 50 °C, peneiração em 4 malhas Microscopia óptica py-GC-MS 1000-30.000 partículas g-1 322 Praias de Niterói/RJ Areia de praia Inspeção visual - 27.372 itens (52,7% plásticos) 323 Praias de Niterói/RJ Areia de praia Inspeção visual - 2789 itens (71% plásticos) 324 Litoral do Rio de Janeiro Areia de praia Inspeção visual - 391 itens (97% plásticos) 325 Praia de Camboinhas/RJ Areia de praia Inspeção visual - 41,5-58,34% plásticos 326 Baía de Guanabara/RJ Areia de praia; profundidade de 5 cm, separação com solução de NaCl, filtração em membrana de papel e secagem a 60 °C Microscopia óptica - 8766 partículas de MP 170 Baía de Guanabara/RJ Sedimento; draga Van Veen; profundidade de 5 cm; secagem a 60 °C; separação com solução de NaCl, peneiração em 15 µm Microscopia óptica ATR-FTIR 160-1000 itens kg-1 ou 4367-25.794 itens m-2 327 Baía de Guanabara/RJ Sedimento; draga Van Veen; profundidade de 20 a 55 m, separação com solução de NaCl, filtração em membrana de nitro celulose e secagem a 21 °C Microscopia óptica - 1 MP por 10 g de sedimento 173 Praia Grande e Praia Grussaí/RJ Areia de praia Inspeção visual - 20.040 itens (> 80% plásticos) 328 Praias de Arraial do Cabo/RJ Areia de praia Inspeção visual - 1080 itens no inverno (61,4% plásticos) e 1533 itens no verão (54,5% plásticos) 329 Niterói/RJ Sedimento e areia de praia; draga Van Veen, profundidade de 2 a 8 m; secagem a 60 °C; peneiração em 3 malhas; separação com solução de NaCl; filtração em membrana de nitrocelulose e secagem a 60 °C Microscopia óptica ATR-FTIR 563 itens no sedimento (20,74 itens kg-1) e 6912 itens na areia de praia (166,50 itens kg-1) 285 Praias de Armação dos Búzios/RJ Areia de praia Inspeção visual - 15.832 itens (64% plásticos) 330 Ilha da Trindade/ES Areia de praia Inspeção visual ATR-FTIR 1057 itens (77% plásticos) 58 Baía de Vitória/ES Sedimento; draga Van Veen; separação com solução de NaCl, filtração em membrana de nitrocelulose e secagem a temperatura ambiente Microscopia óptica e SEM - 0 a 38 partículas por amostra 172 Tabela 1. O CENÁRIO NO BRASIL Ocorrência de microplásticos (MP) e resíduos plásticos em matrizes ambientais de água e sedimento no Brasil (cont.) 1. Ocorrência de microplásticos (MP) e resíduos plásticos em matrizes ambientais de água e sedimento no Brasil (cont.) 1346 Montagner et al. Quim. Nova Tabela 1. Ocorrência de microplásticos (MP) e resíduos plásticos em matrizes ambientais de água e sedimento no Brasil (cont.) Local Amostragem e preparo de amostra Identificação Caracterização química Quantificação Referência Praias de Vitória, Fundão e Regên- cia/ES Areia de praia Inspeção visual - 4752 itens (73,2-85,9% plásticos) 331 Praia de Pontal do Sul/PR Areia de praia; peneiração em 2 malhas Inspeção visual - Máximo de cerca de 100 pellets 332 Estuário de Paranaguá/PR Areia de praia Inspeção visual - 924 itens (81,3% plásticos) 333 Litoral do Paraná Areia de praia; pinça para coleta de pellets Inspeção visual - Concentração de orgânicos sorvidos em pellets 334 Litoral do Paraná Areia de praia Inspeção visual - 12.048 itens (74,8% plásticos) 335 Litoral de Santa Catarina Areia de praia Inspeção visual - 6953 itens (69% plásticos) 336 Ilha de Florianópolis/SC Areia de praia Inspeção visual - 4291 itens (97% plásticos) 337 Praias de Florianópolis/SC Areia de praia Inspeção visual - 10.226 itens (~90 itens 100 m-2) 338 Praias de Laguna/SC Areia de praia Inspeção visual - 2142 itens (17 plásticos) 339 Praia do Cassino/RS Areia de praia Inspeção visual - 0,6-6,6 itens m-1 d-1 340 Praia do Cassino/RS Areia de praia Inspeção visual - 0,3-60,7 itens m-1 (56,1% plásticos) 341 (-): Caracterização química não realizada; ATR: reflectância total atenuada; FTIR: espectroscopia no infravermelho com transformada de Fourier; SEM: microscopia eletrônica de varredura; py-GC-MS: pirolisador acoplado a um cromatógrafo a gás e espectrômetro de massas; PP: polipropileno; PE: polietileno. a 1. Ocorrência de microplásticos (MP) e resíduos plásticos em matrizes ambientais de água e sedimento no Brasil (cont.) ATR: reflectância total atenuada; FTIR: espectroscopia no infravermelho com transformada de Fourier; SEM: C-MS: pirolisador acoplado a um cromatógrafo a gás e espectrômetro de massas; PP: polipropileno; PE: polietileno. ATR: reflectância total atenuada; FTIR: espectroscopia no infravermelho com transformada de Fourier; SEM: C-MS: pirolisador acoplado a um cromatógrafo a gás e espectrômetro de massas; PP: polipropileno; PE: polietileno. (n = 12) e na matriz sedimento (n = 37) não utilizou nenhuma etapa de preparo da amostra. O CENÁRIO NO BRASIL Nos estudos que realizaram a separação por densidade para isolar as partículas foram utilizadas soluções salinas distintas (NaCl, NaI, ZnCl2 e água do mar), permitindo que MP de diferentes composições poliméricas tenham sido encontrados. e as propriedades físico-químicas do polímero, que influenciam diretamente sua distribuição. Apesar dos avanços e das diversas possibilidades já descritas na literatura para os métodos analíticos – panorama que foi explorado neste trabalho – ainda não há métodos oficiais. A caracterização e/ou quantificação de MP pode requerer a determinação do seu tamanho, a composição química do polímero, a morfologia, a origem do MP (primária ou secundária) e o grau de degradação deles no ambiente, uma vez que essas informações ajudam a explicar a toxicidade, os mecanismos de transporte, a capacidade de lixiviação e de sorção dos MP e, consequentemente, o nível de contaminação a que os seres vivos estão expostos. As quantidades de MP reportadas no Brasil foram bastante variadas entre as diferentes regiões do país, tanto em valores absolutos quanto pelas unidades expressas, dificultando uma comparação de cenário, conforme discutido anteriormente neste trabalho. Além disso, 48 trabalhos usaram a inspeção visual e 19 trabalhos utilizaram a microscopia como método de identificação (atribuição se a partícula era ou não MP), sem posterior confirmação por meio de caracterização química usando técnicas analíticas. Somente 4 trabalhos utilizaram alguma técnica instrumental para a caracterização química das partículas após realizar a inspeção visual, sendo que desses, 3 utilizaram FTIR e 1 utilizou Raman. Ao todo, 9 trabalhos utilizaram a microscopia seguida de uma técnica analítica para caracterização química dos MP, sendo que desses, 6 utilizaram FTIR, 1 utilizou py-GC-MS e 2 utilizaram duas técnicas (FTIR e Raman). Até 2018, somente dois trabalhos utilizaram técnicas para caracterizar quimicamente os MP e de 2018 a 2020 foi observado um crescimento na quantidade de estudos que realizaram essa caracterização (n = 7), indicando uma tendência crescente em confirmar a composição das partículas coletadas no ambiente. No entanto, os dados sobre a identidade dos polímeros mais encontrados no Brasil ainda são escassos. Os químicos ambientais têm um compromisso importante nessa área, pois os protocolos analíticos estão em crescente desenvolvimento e, em sua maioria, estão baseados na aplicação de técnicas comumente empregadas em análise de materiais, mas agora adaptadas ao contexto ambiental. O CENÁRIO NO BRASIL De maneira geral, técnicas complementares são necessárias para responder a uma ou mais perguntas da área, tornando a escolha dos métodos analíticos estritamente relacionada com a hipótese da pesquisa, a logística de amostragem e a complexidade das amostras coletadas. A inspeção visual tem sido a maneira mais usual de se caracterizar os MP em amostras ambientais, no entanto, há uma tendência e demanda importante pela caracterização química, o que tem impulsionado o desenvolvimento de diferentes métodos analíticos nessa área. Os maiores avanços relacionados à caracterização são consequentes da necessidade de diminuir cada vez mais a subjetividade da análise, possibilitados pelos avanços na instrumentação analítica. Assim, a caracterização vem se tornando mais precisa em evitar falsos positivos, bem como a produtividade aumentada pela possibilidade de análises de MP cada vez menores e com técnicas cada vez mais rápidas e modernas, como o µ-FPA-FTIR e TED-GC-MS. A automatização pode ser vista como uma tendência a ser explorada. CONSIDERAÇÕES FINAIS E CONCLUSÃO Os MP são considerados contaminantes onipresentes no ambiente, tendo sido detectados mundialmente em todos os compartimentos ambientais. São oriundos de diversas fontes, sejam terrestres ou aquáticas. A elucidação da dimensão e dos impactos relacionados à poluição por MP torna essa uma pesquisa multidisciplinar e bastante complexa do ponto de vista ambiental. A amostragem e o preparo de amostras sempre foram etapas desafiadoras nos métodos analíticos e não são diferentes para a análise dos MP, dada a complexidade das matrizes ambientais, o tamanho das partículas, a identidade Como perspectivas para estabelecer diretrizes e as bases para futuro monitoramento, as discussões estratégicas abordam a decisão do compartimento ambiental, localização e frequência de amostragem; a harmonização dos métodos (protocolos de amostragem, preparo de amostras, parâmetros como a faixa de tamanho, critérios e unidades de Microplásticos: ocorrência ambiental e desafios analíticos Vol. 44, No. 10 1347 quantificação) para compilação e comparação de dados, respeitando o controle analítico e de qualidade, em especial com relação a evitar contaminações cruzadas e confirmação da identidade dos polímeros, e incluindo testes de proficiência. 20. Jambeck, J. R.; Geyer, R.; Wilcox, C.; Siegler, T. R.; Perryman, M.; Andrady, A.; Narayan, R.; Law, K. L.; Science 2015, 347, 768. 21. WHO Global Biodiversity Outlook 5; 2020. 22. Picó, Y.; Barceló, D.; ACS Omega 2019, 4, 6709 Finalmente, ainda há um campo bastante vasto de pesquisas sobre MP para serem realizadas no Brasil, desde o monitoramento deles nos diferentes compartimentos ambientais até a avaliação do comportamento, remoção pelos sistemas convencionais de tratamentos de águas e efluentes e os riscos ou danos ambientais às espécies nativas expostas a esse tipo de contaminante. A revisão da literatura mostrou 80 trabalhos publicados, os quais exploraram apenas matrizes aquáticas e sedimento. Este artigo pode ser usado como um importante ponto de partida nesse sentido, inclusive de forma a subsidiar o trabalho dos químicos analíticos em uma linha de pesquisa tão interdisciplinar na qual as inovações estão constantemente sendo publicadas na literatura. 23. Crawford, C. B.; Quinn, B.; Microplastic Pollutants; Elsevier: Amsterdam, 2017. 24. Boucher, J.; Friot, D.; Primary microplastics in the oceans: A global evaluation of sources; IUCN International Union for Conservation of Nature: Gland, Switzerland, 2017. 25. Fendall, L. S.; Sewell, M. A.; Mar. Pollut. Bull. 2009, 58, 1225. 26. Napper, I. E.; Bakir, A.; Rowland, S. J.; Thompson, R. C.; Mar. Pollut. Bull. 2015, 99, 178. 27. Browne, M. REFERÊNCIAS 36. Rabello, M.; De Paoli, M. A.; Aditivação de Termoplásticos, Editora Artliber: São Paulo, 2013. 1. Carpenter, E. J.; Anderson, S. J.; Harvey, G. R.; Miklas, H. P.; Peck, B. B.; Science 1972, 178, 749. 37. Avio, C. G.; Gorbi, S.; Regoli, F.; Mar. Environ. Res. 2017, 128, 2. 38. Takada, H.; Int. Pellet Watch 2006, 52, 1547. 2. Gomes, C. M. B.; Lançamento de partículas de polietileno à costa do Rio Grande do Sul, Separatas da Revista Veritas, 70/73, pp. 174-206. 39. Endo, S.; Takizawa, R.; Okuda, K.; Takada, H.; Chiba, K.; Kanehiro, H.; Ogi, H.; Yamashita, R.; Date, T.; Mar. Pollut. Bull. 2005, 50, 1103. 3. Andrady, A. L.; Mar. Pollut. Bull. 2011, 62, 1596. 40. Rios, L. M.; Moore, C.; Jones, P. R.; Mar. Pollut. Bull. 2007, 54, 1230. 40. Rios, L. M.; Moore, C.; Jones, P. R.; Mar. Pollut. Bull. 200 4. Cole, M.; Lindeque, P.; Halsband, C.; Galloway, T. S.; Mar. Pollut. Bull. 2011, 62, 2588. 41. Duis, K.; Coors, A.; Environ. Sci. Eur. 2016, 28, 2. 5. Thompson, R. C.; Science 2004, 304, 838. 42. Derraik, J. G. B.; Mar. Pollut. Bull. 2002, 44, 842. 43. Gregory, M. R.; Philos. Trans. R. Soc., B 2009, 364, 2013. 6. Olivatto, G. P.; Carreira, R.; Tornisielo, V. L.; Montagner, C. C.; Rev. Virtual Quim. 2018, 10, 1968. 44. Long, M.; Moriceau, B.; Gallinari, M.; Lambert, C.; Huvet, A.; Raffray, J.; Soudant, P.; Mar. Chem. 2015, 175, 39. 7. Hanvey, J. S.; Lewis, P. J.; Lavers, J. L.; Crosbie, N. D.; Pozo, K.; Clarke, B. O.; Anal. Methods 2017, 9, 1369. J.; Soudant, P.; Mar. Chem. 2015, 175, 39. 45. Moore, C. J.; Environ. Res. 2008, 108, 131. 46. Van Cauwenberghe, L.; Devriese, L.; Galgani, F.; Robbens, J.; Janssen, C. R.; Mar. Environ. Res. 2015, 111, 5. 8. Arthur, C.; Baker, J. E.; Bamford, H. A. In Proceedings of the International Research Workshop on the Occurrence, Effects, and Fate of Microplastic Marine Debris; NOAA Marine Debris Program, 2009; p. 530. C. R.; Mar. Environ. Res. 2015, 111, 5. 47. Wright, S. L.; Thompson, R. C.; Galloway, T. S.; Environ. Pollut. 2013, 178, 483. 9. ISO Plastics - Environmental aspects - State of knowledge and methodologies. ISO/TR 21960:2020. 48. Wright, S. L.; Kelly, F. J.; Environ. Sci. Technol. 2017, 51, 6634. 49. Naik, R. K.; Naik, M. M.; D’Costa, P. M.; Shaikh, F.; Mar. Pollut. Bull. AGRADECIMENTOS 30. Ceccarini, A.; Corti, A.; Erba, F.; Modugno, F.; La Nasa, J.; Bianchi, S.; Castelvetro, V.; Environ. Sci. Technol. 2018, 52, 5634. Os autores agradecem à Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP Proc. 2018/21733-0), ao Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq Proc. 428826/2018-0) e ao Instituto Nacional de Ciências e Tecnologias Analíticas Avançadas-INCTAA (FAPESP Proc. 2014/50951-4 e CNPq Proc. 465768/2014-8) pelo auxílio financeiro concedido aos projetos. CNPq Proc. 311422/2020-9 pela bolsa de produtividade em pesquisa de CCM. O presente trabalho foi realizado com apoio da Coordenação de Aperfeiçoamento de Pessoal de Nível Superior – Brasil (CAPES) – Código de Financiamento 001 31. Wagner, M.; Lambert, S. In The Handbook of Environmental Chemistry; Barceló, D.; Kostianoy, A. G., ed.; Springer International Publishing: New York, 2018; Vol. 58. 32. De Paoli, M. A.; Degradação e Estabilização de Polímeros, 2a ed., Chemkeys: Campinas, 2008. 33. Waldman, W. R.; Rillig, M. C.; Environ. Sci. Technol. 2020, 54, 7751. 34. Mai, L.; Bao, L.-J.; Shi, L.; Wong, C. S.; Zeng, E. Y.; Environ. Sci. Pollut. Res. 2018, 25, 11319. 35. Wu, P.; Huang, J.; Zheng, Y.; Yang, Y.; Zhang, Y.; He, F.; Chen, H.; Quan, G.; Yan, J.; Li, T.; Gao, B.; Ecotoxicol. Environ. Saf. 2019, 184, 109612. CONSIDERAÇÕES FINAIS E CONCLUSÃO A.; Crump, P.; Niven, S. J.; Teuten, E.; Tonkin, A.; Galloway, T.; Thompson, R.; Environ. Sci. Technol. 2011, 45, 9175. 28. Barnes, D. K. A.; Galgani, F.; Thompson, R. C.; Barlaz, M.; Philos. Trans. R. Soc., B 2009, 364, 1985. 29. Rocha-Santos, T.; Duarte, A. C.; TrAC – Trends Anal. Chem. 2015, 65, 47. REFERÊNCIAS GESAMP Guidelines for the Monitoring and Assessment of Plastic Litter in the Ocean; 2019. 98. Silva, J. D. B.; Barletta, M.; Lima, A. R. A.; Ferreira, G. V. B.; Environ. Pollut. 2018, 242, 1010. 67. Dehghani, S.; Moore, F.; Akhbarizadeh, R.; Environ. Sci. Pollut. Res. 2017, 24, 20360. 99. Lima, A. R. A.; Silva, M. D.; Possatto, F. E.; Ferreira, G. V. B.; Krelling, A. P. In The Handbook of Environmental Chemistry, 2020; pp. 1-34. 68. Abbasi, S.; Keshavarzi, B.; Moore, F.; Turner, A.; Kelly, F. J.; Dominguez, A. O.; Jaafarzadeh, N.; Environ. Pollut. 2019, 244, 153. 100. Silva-Cavalcanti, J. S.; Silva, J. D. B.; de França, E. J.; de Araújo, M. C. B.; Gusmão, F.; Environ. Pollut. 2017, 221, 218. B.; Gusmão, F.; Environ. Pollut. 2017, 221, 218. 69. Dris, R.; Gasperi, J.; Mirande, C.; Mandin, C.; Guerrouache, M.; Langlois, V.; Tassin, B.; Environ. Pollut. 2017, 221, 453. 101. Urbanski, B. Q.; Denadai, A. C.; Azevedo-Santos, V. M.; Nogueira, M. G.; Biota Neotrop. 2020, 20, e20201005. 70. Dris, R.; Gasperi, J.; Rocher, V.; Saad, M.; Renault, N.; Tassin, B.; Environ. Chem. 2015, 12, 592. 102. Rizzi, M.; Rodrigues, F. L.; Medeiros, L.; Ortega, I.; Rodrigues, L.; 102. Rizzi, M.; Rodrigues, F. L.; Medeiros, L.; Ortega, I.; Rodrigues, L.; Monteiro, D. S.; Kessler, F.; Proietti, M. C.; Mar. Pollut. Bull. 2019, 140, 536. 71. Brahney, J.; Hallerud, M.; Heim, E.; Hahnenberger, M.; Sukumaran, S.; Science 2020, 368, 1257. 103. Petry, M. V.; Benemann, V. R. F.; Mar. Pollut. Bull. 2017, 117, 131. 72. Allen, S.; Allen, D.; Phoenix, V. R.; Le Roux, G.; Durántez Jiménez, P.; Simonneau, A.; Binet, S.; Galop, D.; Nat. Geosci. 2019, 12, 339. monneau, A.; Binet, S.; Galop, D.; Nat. Geosci. 2019, 12, 339 104. Rossi, L. C.; Scherer, A. L.; Petry, M. V.; Mar. Pollut. Bull. 2019, 138, 235. 73. Ambrosini, R.; Azzoni, R. S.; Pittino, F.; Diolaiuti, G.; Franzetti, A.; Parolini, M.; Environ. Pollut. 2019, 253, 297. 105. Brentano, R.; de Brum, A. C.; Montone, R. C.; Petry, M. V.; Mar. Pollut. Bull. 2020, 151, 110815. 74. Zhang, Y.; Gao, T.; Kang, S.; Sillanpää, M.; Environ. Pollut. 2019, 254, 1. 106. Santana, M. F. M.; Ascer, L. G.; Custódio, M. R.; Moreira, F. T.; Turra, A.; Mar. Pollut. Bull. 2016, 106, 183. 75. Steinmetz, Z.; Wollmann, C.; Schaefer, M.; Buchmann, C.; David, J.; Tröger, J.; Muñoz, K.; Frör, O.; Schaumann, G. E.; Sci. REFERÊNCIAS N.; Amorim, A. L. A.; Patrício, J.; Palma, A. R. T.; Mar. Pollut. Bull. 2017, 117, 448. 91. Vendel, A. L.; Bessa, F.; Alves, V. E. N.; Amorim, A. L. A.; Patrício, J.; Palma, A. R. T.; Mar. Pollut. Bull. 2017, 117, 448. 92. Ferreira, G. V. B.; Barletta, M.; Lima, A. R. A.; Dantas, D. V.; Justino, A. K. S.; Costa, M. F.; ICES J. Mar. Sci. 2016, 73, 2695. 92. Ferreira, G. V. B.; Barletta, M.; Lima, A. R. A.; Dantas, D. V.; Justino, 58. Andrades, R.; Santos, R. G.; Joyeux, J.; Chelazzi, D.; Cincinelli, A.; Giarrizzo, T.; Mar. Pollut. Bull. 2018, 137, 180. A. K. S.; Costa, M. F.; ICES J. Mar. Sci. 2016, 73, 2695. 59. He, P.; Chen, L.; Shao, L.; Zhang, H.; Lü, F.; Water Res. 2019, 159, 38. 93. Ferreira, G. V. B.; Barletta, M.; Lima, A. R. A.; Morley, S. A.; Justino, A. K. S.; Costa, M. F.; Environ. Pollut. 2018, 236, 706. 93. Ferreira, G. V. B.; Barletta, M.; Lima, A. R. A.; Morley, S. A.; Justino, A. K. S.; Costa, M. F.; Environ. Pollut. 2018, 236, 706. 60. UNEP Plastic in Cosmetics - Are we polluting the environment through our personal care?, 2015. 94. Pegado, T. de S. e S.; Schmid, K.; Winemiller, K. O.; C 94. Pegado, T. de S. e S.; Schmid, K.; Winemiller, K. O.; Chelazzi, D.; Cincinelli, A.; Dei, L.; Giarrizzo, T.; Mar. Pollut. Bull. 2018, 133, 814. 61. Kalčíková, G.; Alič, B.; Skalar, T.; Bundschuh, M.; Gotvajn, A. Ž.; Chemosphere 2017, 188, 25. Cincinelli, A.; Dei, L.; Giarrizzo, T.; Mar. Pollut. Bull. 2018, 133, 814. 95. Andrade, M. C.; Winemiller, K. O.; Barbosa, P. S.; Fortunati, A.; Chelazzi, D.; Cincinelli, A.; Giarrizzo, T.; Environ. Pollut. 2019, 244, 766. 62. Salvador Cesa, F.; Turra, A.; Baruque-Ramos, J.; Sci. Total Environ. 2017, 598, 1116. 63. Napper, I. E.; Thompson, R. C.; Mar. Pollut. Bull. 2016, 112, 39. 96. Ferreira, G. V. B.; Barletta, M.; Lima, A. R. A.; Morley, S. A.; Costa, M. F.; Sci. Rep. 2019, 9, 13514. 64. De Falco, F.; Di Pace, E.; Cocca, M.; Avella, M.; Sci. Rep. 2019, 9, 1. 97. Dantas, N. C. F. M.; Duarte, O. S.; Ferreira, W. C.; Ayala, A. P.; Rezende, C. F.; Feitosa, C. V.; Mar. Pollut. Bull. 2020, 153, 110959. 65. IBGE Pesquisa Nacional de Saneamento Básico 2020. 66. REFERÊNCIAS 2019, 149, 110525. 10. Richardson, S. D.; Kimura, S. Y.; Anal. Chem. 2016, 88, 546. 50. Rummel, C. D.; Jahnke, A.; Gorokhova, E.; Kühnel, D.; Schmitt-Jansen, M.; Environ. Sci. Technol. Lett. 2017, 4, 258. 11. Richardson, S. D.; Kimura, S. Y.; Anal. Chem. 2020, 92, 473. 12. Spinacé, M. A. da S.; De Paoli, M. A.; Quim. Nova 2005, 28, 65. 51. Kirstein, I. V.; Kirmizi, S.; Wichels, A.; Garin-Fernandez, A.; Erler, R.; d d 2016 51. Kirstein, I. V.; Kirmizi, S.; Wichels, A.; Garin-Fernandez, A.; Erler, R.; Löder, M.; Gerdts, G.; Mar. Environ. Res. 2016, 120, 1. 13. do Sul, J. A. I.; Costa, M. F.; Environ. Pollut. 2014, 185, 352. Löder, M.; Gerdts, G.; Mar. Environ. Res. 2016, 120, 1. 14. Tourinho, P. S.; Kočí, V.; Loureiro, S.; van Gestel, C. A. M.; Environ. Pollut. 2019, 252, 1246. 52. Jamieson, A. J.; Brooks, L. S. R.; Reid, W. D. K.; Piertney, S. B.; Narayanaswamy, B. E.; Linley, T. D.; R. Soc. Open Sci. 2019, 6, 180667. 15. Geyer, R.; Jambeck, J. R.; Law, K. L.; Sci. Adv. 2017, 3, e1700782. 16. Akdogan, Z.; Guven, B.; Environ. Pollut. 2019, 254, 113011. 53. Napper, I. E.; Davies, B. F. R.; Clifford, H.; Elvin, S.; Koldewey, H. J.; Mayewski, P. A.; Miner, K. R.; Potocki, M.; Elmore, A. C.; Gajurel, A. P.; Thompson, R. C.; One Earth 2020, 3, 621. 17. ABIPLAST Perfil 2018; 2018. 18. WWF Brasil é o 4° país do mundo que mais gera lixo plástico, WWF, 2019. 19. Kataoka, T.; Nihei, Y.; Kudou, K.; Hinata, H.; Environ. Pollut. 2019, 244, 958. 54. Liu, C.; Li, J.; Zhang, Y.; Wang, L.; Deng, J.; Gao, Y.; Yu, L.; Zhang, J.; Sun, H.; Environ. Int. 2019, 128, 116. 1348 Montagner et al. Quim. Nova 55. do Sul, J. A. I.; Spengler, Â.; Costa, M. F.; Mar. Pollut. Bull. 2009, 58, 1236. 89. do Sul, J. A. I.; Costa, M. F.; Barletta, M.; Cysneiros, F. J. A.; Mar. Pollut. Bull. 2013, 75, 305. 56. do Sul, J. A. I.; Costa, M. F.; Fillmann, G.; Water, Air, Soil Pollut. 2014, 225, 2004. 90. Possatto, F. E.; Barletta, M.; Costa, M. F.; do Sul, J. A. I.; Dantas, D. V.; Mar. Pollut. Bull. 2011, 62, 1098. 57. Lima, A. R. A.; Barletta, M.; Costa, M. F.; Front. Environ. Sci. 2016, 4, 56. 91. Vendel, A. L.; Bessa, F.; Alves, V. E. REFERÊNCIAS Total Environ. 2016, 550, 690. 107. Morais, L. M. S.; Sarti, F.; Chelazzi, D.; Cincinelli, A.; Giarrizzo Martinelli Filho, J. E.; Environ. Pollut. 2020, 265, 114817. Martinelli Filho, J. E.; Environ. Pollut. 2020, 265, 114817. 76. Huang, Y.; Liu, Q.; Jia, W.; Yan, C.; Wang, J.; Environ. Pollut. 2020, 260, 114096. 108. Attademo, F. L. N.; Balensiefer, D. C.; Freire, A. C. da B.; de Sousa, G. P.; da Cunha, F. A. G. C.; Luna, F. de O.; Mar. Pollut. Bull. 2015, 101, 284. 77. Corradini, F.; Meza, P.; Eguiluz, R.; Casado, F.; Huerta-Lwanga, E.; Geissen, V.; Sci. Total Environ. 2019, 671, 411. 109. Di Beneditto, A. P. M.; Siciliano, S.; Mar. Pollut. Bull. 2017, 125, 330. 78. Westphalen, H.; Abdelrasoul, A. In Water Challenges of an Urbanizing World; InTech, 2018; p. 83. 110. Sampaio, C. L. S.; Leite, L.; Reis-Filho, J. A.; Loiola, M.; Miranda, R. J.; de Anchieta, C. C.; Nunes, J.; Macena, B. C. L.; Environ. Biol. Fishes 2018, 101, 1285. 79. Simon, M.; van Alst, N.; Vollertsen, J.; Water Res. 2018, 142, 1. 80. CONAMA Resolução CONAMA N° 375 de 29 de agosto de 2006 - Ministério do Meio Ambiente 2006. 111. Brentano, R.; Petry, M. V.; Mar. Pollut. Bull. 2020, 150, 110595. 81. Stock, F.; Kochleus, C.; Bänsch-Baltruschat, B.; Brennholt, N.; Reifferscheid, G.; TrAC – Trends Anal. Chem. 2019, 113, 84. 112. Costa, L. L.; Arueira, V. F.; da Costa, M. F.; Di Beneditto, A. P. M.; Zalmon, I. R.; Mar. Pollut. Bull. 2019, 145, 5. Reifferscheid, G.; TrAC – Trends Anal. Chem. 2019, 113, 84 82. Prata, J. C.; da Costa, J. P.; Duarte, A. C.; Rocha-Santos, T.; TrAC – Trends Anal. Chem. 2019, 110, 150. 113. Gusmão, F.; Domenico, M. Di; Amaral, A. C. Z.; Martínez, A.; Gonzalez, B. C.; Worsaae, K.; do Sul, J. A. I.; Lana, P. C.; Environ. Pollut. 2016, 216, 584. 83. Van Melkebeke, M.; Janssen, C.; De Meester, S.; Environ. Sci. Technol. 2020, 54, 8668. 114. Bessa, F.; Ratcliffe, N.; Otero, V.; Sobral, P.; Marques, J. C.; Waluda, C. M ; Trathan P N ; Xavier J C ; Sci Rep 2019 9 14191 114. Bessa, F.; Ratcliffe, N.; Otero, V.; Sobral, P.; Marques, J. C.; Waluda, C. M.; Trathan, P. N.; Xavier, J. C.; Sci. Rep. 2019, 9, 14191. 114. Bessa, F.; Ratcliffe, N.; Otero, V.; Sobral, P.; Marques, J. C.; M.; Trathan, P. N.; Xavier, J. REFERÊNCIAS Fisner, M.; Majer, A. P.; Balthazar-Silva, D.; Gorman, D.; Turra, A.; Environ. Sci. Pollut. Res. 2017, 24, 13732. 139. Lindeque, P. K.; Cole, M.; Coppock, R. L.; Lewis, C. N.; Miller, Z.; Watts, A. J. R. R.; Wilson-McNeal, A.; Wright, S. L.; Tamara, S.; Miller, R. Z.; Watts, A. J. R. R.; Wilson-McNeal, A.; Wright, S. L.; Galloway, T. S.; Environ. Pollut. 2020, 265,114721. 172. Baptista Neto, J. A.; Gaylarde, C.; Beech, I.; Bastos, A. C.; Quaresma, V. S.; de Carvalho, D. G.; Ocean & Coastal Management 2019, 169, 247. T. S.; Environ. Pollut. 2020, 265,114721. 140. Tamminga, M.; Stoewer, S.-C.; Fischer, E. K.; Environ. Pollut. 2019, 254, 112970. 173. Baptista Neto, J. A.; de Carvalho, D. G.; Medeiros, K.; Drabinski, T. L.; de Melo, G. V.; Silva, R. C. O.; Silva, D. C. P.; Batista, L. S.; Dias, G. T. M.; da Fonseca, E. M.; dos Santos Filho, J. R.; Mar. Pollut. Bull. 2019, 149, 110558. 141. Karlsson, T. M.; Kärrman, A.; Rotander, A.; Hassellöv, M.; Environ. Sci. Pollut. Res. 2020, 27, 5559. 174. Izar, G. M.; Morais, L. G.; Pereira, C. D. S.; Cesar, A.; Abessa, D. M S.; Christofoletti, R. A.; Reg. Stud. Mar. Sci. 2019, 29, 100705. 142. Ng, K. L.; Obbard, J. P.; Mar. Pollut. Bull. 2006, 52, 761. 143. Zhang, L.; Liu, J.; Xie, Y.; Zhong, S.; Yang, B.; Lu, D.; Zhong, Q.; Sci. Total Environ. 2020, 708, 135176. 175. Dioses-Salinas, D. C.; Pizarro-Ortega, C. I.; De-la-Torre, G. E.; Sci. Total Environ. 2020, 730, 139164. 144. Bannick, C. G.; Szewzyk, R.; Ricking, M.; Schniegler, S.; Obermaier, 144. Bannick, C. G.; Szewzyk, R.; Ricking, M.; Schniegler, S.; Obermaier, N.; Barthel, A. K.; Altmann, K.; Eisentraut, P.; Braun, U.; Water Res. 2019, 149, 650. 176. Lusher, A. L.; Bråte, I. L. N.; Munno, K.; Hurley, R. R.; Welden, N. A.; Appl. Spectrosc. 2020, 74, 1139. 145. Koelmans, A. A.; Mohamed Nor, N. H.; Hermsen, E.; Kooi, M.; Mintenig, S. M.; De France, J.; Water Res. 2019, 155, 410. 177. Klein, S.; Worch, E.; Knepper, T. P.; Environ. Sci. Technol. 2015, 49, 6070. 146. Ryan, P. G.; Suaria, G.; Perold, V.; Pierucci, A.; Bornman, T. G.; Aliani, S.; Environ. Pollut. 2020, 258, 113413. 178. Jung, M. R.; Horgen, F. D.; Orski, S. V.; Rodriguez C., V.; Beers, K. L.; Balazs, G. H.; Jones, T. T.; Work, T. M.; Brignac, K. C.; Royer, S.-J.; Hyrenbach, K. REFERÊNCIAS Cox, K. D.; Covernton, G. A.; Davies, H. L.; Dower, J. F.; Juanes, F.; Dudas, S. E.; Environ. Sci. Technol. 2019, 53, 7068. 151. McEachern, K.; Alegria, H.; Kalagher, A. L.; Hansen, C.; Morrison, S.; Hastings, D.; Mar. Pollut. Bull. 2019, 148, 97. 123. Amato-Lourenço, L. F.; Galvão, L. S.; de Weger, L. A.; Hiemstra, P. S.; Vij M G M d T S i T l E i 2020 749 141676 123. Amato-Lourenço, L. F.; Galvão, L. S.; de Weger, L. A.; Hiemstra, P. S.; Vijver, M. G.; Mauad, T.; Sci. Total Environ. 2020, 749, 141676. 152. Schmidt, L. K.; Bochow, M.; Imhof, H. K.; Oswald, S. E.; Environ. Pollut. 2018, 239, 579. 123. Amato Lourenço, L. F.; Galvão, L. S.; de Weger, L. A.; Hiemstra, P. S.; Vijver, M. G.; Mauad, T.; Sci. Total Environ. 2020, 749, 141676. Vijver, M. G.; Mauad, T.; Sci. Total Environ. 2020, 749, 141676 153. Courtene-Jones, W.; Quinn, B.; Gary, S. F.; Mogg, A. O. M.; Narayanaswamy, B. E.; Environ. Pollut. 2017, 231, 271. 124. Masura, J.; Baker, J. E.; Foster, G. D.; Arthur, C.; Herring, C.; Laboratory methods for the analysis of microplastics in the marine environment: recommendations for quantifying synthetic particles in waters and sediments. NOAA Mar. Debris Progr. 2015. 154. Möller, J. N.; Löder, M. G. J.; Laforsch, C.; Environ. Sci. Technol. 2020, 54, 2078. 155. Pagter, E.; Frias, J.; Nash, R.; Mar. Pollut. Bull. 2018, 135, 932. 125. Dehaut, A.; Hermabessiere, L.; Duflos, G.; TrAC – Trends Anal. Chem. 2019, 116, 346. 156. de Araújo, M. C. B.; Santos, P. J. P.; Costa, M. F.; Mar. Pollut. Bull. 2006, 52, 957. 126. Prata, J. C.; Reis, V.; da Costa, J. P.; Mouneyrac, C.; Duarte, A. C.; Rocha-Santos, T.; J. Hazard. Mater. 2021, 403, 123660. 157. Dris, R.; Gasperi, J.; Saad, M.; Mirande, C.; Tassin, B.; Mar. Pollut. Bull. 2016, 104, 290. 127. Brander, S. M.; Renick, V. C.; Foley, M. M.; Steele, C.; Woo, M.; Lusher, A.; Carr, S.; Helm, P.; Box, C.; Cherniak, S.; Andrews, R. C.; Rochman, C. M.; Appl. Spectrosc. 2020, 74, 1099. 158. Cai, L.; Wang, J.; Peng, J.; Tan, Z.; Zhan, Z.; Tan, X.; Chen, Q.; Environ. Sci. Pollut. Res. 2017, 24, 24928. 159. Klein, M.; Fischer, E. K.; Sci. Total Environ. 2019, 685, 96 128. Hidalgo-Ruz, V.; Gutow, L.; Thompson, R. C.; Thiel, M.; Environ. Sci. Technol. 2012, 46, 3060. 160. REFERÊNCIAS Stanton, T.; Johnson, M.; Nathanail, P.; MacNaughtan, W.; Gomes, R. L.; Sci. Total Environ. 2019, 666, 377. 129. Bergmann, M.; Gutow, L.; Klages, M. In Marine Anthropogenic Litter; Bergmann, M.; Gutow, L.; Klages, M., eds.; Springer International Publishing, 2015. 161. Wright, S. L.; Ulke, J.; Font, A.; Chan, K. L. A.; Kelly, F. J.; Environ. Int. 2020, 136, 105411. 130. Castro, R. O.; Silva, M. L.; Marques, M. R. C.; de Araújo, F. V.; Mar. Pollut. Bull. 2016, 110, 555. 162. Liu, K.; Wu, T.; Wang, X.; Song, Z.; Zong, C.; Wei, N.; Li, D.; Environ. Sci. Technol. 2019, 53, 10612. 131. Kapp, K. J.; Yeatman, E.; Environ. Pollut. 2018, 241, 1082. 163. Liu, K.; Wang, X.; Fang, T.; Xu, P.; Zhu, L.; Li, D.; Sci. Total Environ. 2019, 675, 462. 132. Bruge, A.; Dhamelincourt, M.; Lanceleur, L.; Monperrus, M.; Gasperi, J.; Tassin, B.; Sci. Total Environ. 2020, 718, 137319. 164. Liu, K.; Wang, X.; Wei, N.; Song, Z.; Li, D.; Environ. Int. 2019, 132, 105127. 133. Lima, A. R. A.; Costa, M. F.; Barletta, M.; Environ. Res. 2014, 132, 146. 165. Li, Y.; Shao, L.; Wang, W.; Zhang, M.; Feng, X.; Li, W.; Zhang, D.; Sci. Total Environ. 2020, 705, 135967. 134. Lima, A. R. A.; Barletta, M.; Costa, M. F.; Ramos, J. A. A.; Dantas, D. V.; Melo, P. A. M. C.; Justino, A. K. S.; Ferreira, G. V. B.; J. Fish Biol. 2016, 89, 619. 166. Wright, S. L.; Levermore, J. M.; Kelly, F. J.; Environ. Sci. Technol. 2019, 53, 8947. 135. Schönlau, C.; Karlsson, T. M.; Rotander, A.; Nilsson, H.; Engwall, M.; van Bavel, B.; Kärrman, A.; Mar. Pollut. Bull. 2020, 153, 111019. 167. Hurley, R. R.; Lusher, A. L.; Olsen, M.; Nizzetto, L.; Environ. Sci. Technol. 2018, 52, 7409. 136. Baini, M.; Fossi, M. C.; Galli, M.; Caliani, I.; Campani, T.; Finoia, M. G.; Panti, C.; Mar. Pollut. Bull. 2018, 133, 543. 168. Bläsing, M.; Amelung, W.; Sci. Total Environ. 2018, 612, 422. 137. Silva, A. B.; Bastos, A. S.; Justino, C. I. L.; da Costa, J. P.; Duarte, A. C.; Rocha-Santos, T. A. P.; Anal. Chim. Acta 2018, 1017, 1. 169. Li, J.; Song, Y.; Cai, Y.; Environ. Pollut. 2020, 257, 113570 170. Carvalho, D. G.; Baptista Neto, J. A.; Ocean & Coastal Management 2016, 128, 10. 138. Setälä, O.; Magnusson, K.; Lehtiniemi, M.; Norén, F.; Mar. Pollut. Bull. 2016, 110, 177. 171. REFERÊNCIAS C.; Sci. Rep. 2019, 9, 14191. 84. Van Cauwenberghe, L.; Vanreusel, A.; Mees, J.; Janssen, C. R.; Environ. Pollut. 2013, 182, 495. M.; Trathan, P. N.; Xavier, J. C.; Sci. Rep. 2019, 9, 14191. 115. Hwang, J.; Choi, D.; Han, S.; Choi, J.; Hong, J.; Sci. Total Environ. 2019, 684, 657. 85. Zhao, S.; Danley, M.; Ward, J. E.; Li, D.; Mincer, T. J.; Anal. Methods 2017, 9, 1470. 116. Rist, S.; Almroth, B. C.; Hartmann, N. B.; Karlsson, T. M.; Sci. Total Environ. 2018, 626, 720. 86. Watts, A. J. R.; Lewis, C.; Goodhead, R. M.; Beckett, S. J.; Moger, J.; Tyler, C. R.; Galloway, T. S.; Environ. Sci. Technol. 2014, 48, 8823. 117. Karami, A.; Golieskardi, A.; Choo, C. K.; Larat, V.; Karbalaei, S.; S l ti i B S i T t l E i 2018 612 1380 117. Karami, A.; Golieskardi, A.; Choo, C. K.; Larat, V.; Karbalaei, S.; Salamatinia, B.; Sci. Total Environ. 2018, 612, 1380. 117. Karami, A.; Golieskardi, A.; Choo, C. K.; Larat, V Salamatinia, B.; Sci. Total Environ. 2018, 612, 1380. Salamatinia, B.; Sci. Total Environ. 2018, 612, 1380. 87. Barboza, L. G. A.; Vethaak, A. D.; Lavorante, B. R. B. O.; Lundebye, A.-K.; Guilhermino, L.; Mar. Pollut. Bull. 2018, 133, 336. 118. Zuccarello, P.; Ferrante, M.; Cristaldi, A.; Copat, C.; Grasso, A.; 118. Zuccarello, P.; Ferrante, M.; Cristaldi, A.; Copat, C.; Grasso, A.; Sangregorio, D.; Fiore, M.; Conti, G. O.; Water Res. 2019, 157, 365. 118. Zuccarello, P.; Ferrante, M.; Cristaldi, A.; Copat, C.; Grasso, A.; Sangregorio, D.; Fiore, M.; Conti, G. O.; Water Res. 2019, 157, 365. 88. Tourinho, P. S.; do Sul, J. A. I.; Fillmann, G.; Mar. Pollut. Bull. 2010, 60, 396. ngregorio, D.; Fiore, M.; Conti, G. O.; Water Res. 2019, 157, 3 119. Mason, S. A.; Welch, V. G.; Neratko, J.; Front. Chem. 2018, 6, 407. Microplásticos: ocorrência ambiental e desafios analíticos Vol. 44, No. 10 1349 149. Green, D. S.; Kregting, L.; Boots, B.; Blockley, D. J.; Brickle, P.; da Costa, M.; Crowley, Q.; Mar. Pollut. Bull. 2018, 137, 695. 120. Shruti, V. C.; Pérez-Guevara, F.; Elizalde-Martínez, I.; Kutralam- Muniasamy, G.; Environ. Pollut. 2021, 268, 115811. 121. Senathirajah, K.; Attwood, S.; Bhagwat, G.; Carbery, M.; Wilson, S.; Palanisami, T.; J. Hazard. Mater. 2021, 404, 124004. 150. Barrows, A. P. W.; Christiansen, K. S.; Bode, E. T.; Hoellein, T. J.; Water Res. 2018, 147, 382. 122. REFERÊNCIAS D.; Jensen, B. A.; Lynch, J. M.; Mar. Pollut. Bull. 2018, 127, 704. 147. Song, Y. K.; Hong, S. H.; Jang, M.; Kang, J.-H.; Kwon, O. Y.; Han, G. 147. Song, Y. K.; Hong, S. H.; Jang, M.; Kang, J.-H.; Kwon, O. Y.; Han, G. M.; Shim, W. J.; Environ. Sci. Technol. 2014, 48, 9014. M.; Shim, W. J.; Environ. Sci. Technol. 2014, 48, 9014. 179. Amigo, J. M.; Babamoradi, H.; Elcoroaristizabal, S.; Anal. Chim. Acta 2015, 896, 34. 148. Barrows, A. P. W.; Neumann, C. A.; Berger, M. L.; Shaw, S. D.; Anal. Methods 2017, 9, 1446. 1350 Montagner et al. Quim. Nova 180. Harrison, J. P.; Ojeda, J. J.; Romero-González, M. E.; Sci. Total Environ. 2012, 416, 455. 217. Horton, A. A.; Jürgens, M. D.; Lahive, E.; van Bodegom, P. M.; Vijver, M. G.; Environ. Pollut. 2018, 236, 188. 181. Bergmann, M.; Mützel, S.; Primpke, S.; Tekman, M. B.; Trachsel, J.; Gerdts, G.; Sci. Adv. 2019, 5, 1. 218. Karlsson, T. M.; Vethaak, A. D.; Almroth, B. C.; Ariese, F.; van Velzen, M.; Hassellöv, M.; Leslie, H. A.; Mar. Pollut. Bull. 2017, 122, 403. 182. Pabortsava, K.; Lampitt, R. S.; Nat. Commun. 2020, 11, 4073. 219. Wagner, J.; Wang, Z. M.; Ghosal, S.; Rochman, C.; Gassel, M.; Wall, S.; Anal. Methods 2017, 9, 1479. 183. Liu, F.; Olesen, K. B.; Borregaard, A. R.; Vollertsen, J.; Sci. Total Environ. 2019, 671, 992. 184. Liu, F.; Nord, N. B.; Bester, K.; Vollertsen, J.; Water 2020, 12, 1085. 220. Watts, A. J. R.; Urbina, M. A.; Goodhead, R.; Moger, J.; Lewis, C.; Galloway, T. S.; Environ. Sci. Technol. 2016, 50, 5364. 185. Löder, M. G. J.; Kuczera, M.; Mintenig, S.; Lorenz, C.; Gerdts, G.; Environ. Chem. 2015, 12, 563. 221. Araujo, C. F.; Nolasco, M. M.; Ribeiro, A. M. P.; Ribeiro-Claro, P. J. A.; Water Res. 2018, 142, 426. 186. Käppler, A.; Windrich, F.; Löder, M. G. J.; Malanin, M.; Fischer, D.; Labrenz, M.; Eichhorn, K. J.; Voit, B.; Anal. Bioanal. Chem. 2015, 407, 6791. 186. Käppler, A.; Windrich, F.; Löder, M. G. J.; Malanin, M.; Fischer, D.; 222. Imhof, H. K.; Schmid, J.; Niessner, R.; Ivleva, N. P.; Laforsch, C.; Limnol. Oceanogr. Methods 2012, 10, 524. Labrenz, M.; Eichhorn, K. J.; Voit, B.; Anal. Bioanal. Chem. 2015, 407, 6791. 187. Elert, A. M.; Becker, R.; Duemichen, E.; Eisentraut, P.; Falkenhagen, J.; Sturm, H.; Braun, U.; Environ. Pollut. 2017, 231, 1256. REFERÊNCIAS KnowItAll U - Spectra Database. Wiley Sci. Solut. 2020. 199. Ballent, A.; Corcoran, P. L.; Madden, O.; Helm, P. A.; Longstaffe, F. J.; Mar. Pollut. Bull. 2016, 110, 383. 232. ASTM E2310-04 Standard Guide for Use of Spectral 232. ASTM E2310-04 Standard Guide for Use of Spectral Searching by Curve Matching Algorithms with Data Recorded Using Mid-infrared Spectroscopy. ASTM 2004, i, 1-9. 200. Absher, T. M.; Ferreira, S. L.; Kern, Y.; Ferreira, A. L.; Christo, S. W.; Ando, R. A.; Environ. Sci. Pollut. Res. 2019, 26, 292. 200. Absher, T. M.; Ferreira, S. L.; Kern, Y.; Ferreira, A. L.; 233. Munno, K.; De Frond, H.; O’donnell, B.; Rochman, C. M.; Anal. Chem. 2020, 92, 2443. Ando, R. A.; Environ. Sci. Pollut. Res. 2019, 26, 292. 201. Di, M.; Wang, J.; Sci. Total Environ. 2018, 616-617, 1620. 234. Dong, M.; Zhang, Q.; Xing, X.; Chen, W.; She, Z.; Luo, Z.; Sci. Total Environ. 2020, 739, 139990. 202. Enders, K.; Lenz, R.; Stedmon, C. A.; Nielsen, T. G.; Mar. Pollut. Bull. 2015, 100, 70. 235. Shan, J.; Zhao, J.; Liu, L.; Zhang, Y.; Wang, X.; Wu, F.; Environ. Pollut. 2018, 238, 121. 203. Frère, L.; Paul-Pont, I.; Rinnert, E.; Petton, S.; Jaffré, J.; Bihannic, I.; Soudant, P.; Lambert, C.; Huvet, A.; Environ. Pollut. 2017, 225, 211. 204. Horton, A. A.; Svendsen, C.; Williams, R. J.; Spurgeon, D. J.; Lahive, E.; Mar. Pollut. Bull. 2017, 114, 218. 236. Wander, L.; Vianello, A.; Vollertsen, J.; Westad, F.; Braun, U.; Paul, A.; Anal. Methods 2020, 12, 781. 4. Horton, A. A.; Svendsen, C.; Williams, R. J.; Spurgeon, D. E.; Mar. Pollut. Bull. 2017, 114, 218. 237. Da Silva, V. H.; Murphy, F.; Amigo, J. M.; Stedmon, C.; Strand, J.; Anal. Chem. 2020, 92, 13724. 205. Imhof, H. K.; Wiesheu, A. C.; Anger, P. M.; Niessner, R.; Ivleva, N. P.; Laforsch, C.; Sci. Total Environ. 2018, 613-614, 530. 238. Anger, P. M.; Prechtl, L.; Elsner, M.; Niessner, R.; Ivleva, N. P.; Anal. Methods 2019, 11, 3483. 206. Kazour, M.; Jemaa, S.; Issa, C.; Khalaf, G.; Amara, R.; Sci. Total Environ. 2019, 696, 133933. 239. Ellis, J. D.; Iqbal, R.; Yoshimatsu, K.; Anal. Chim. Acta 2020, 1103, 49. 207. Kniggendorf, A. K.; Wetzel, C.; Roth, B.; Sensors 2019, 19, 12. 240. Frère, L.; Paul-Pont, I.; Moreau, J.; Soudant, P.; Lambert, C.; Huvet, A.; i ll ll 2016 113 461 208. REFERÊNCIAS 223. Lenz, R.; Enders, K.; Stedmon, C. A.; MacKenzie, D. M. A.; Nielsen, T. G.; Mar. Pollut. Bull. 2015, 100, 82. Sturm, H.; Braun, U.; Environ. Pollut. 2017, 231, 1256. 224. Thiele, C. J.; Hudson, M. D.; Russell, A. E.; Mar. Pollut. Bull. 2019, 142, 384. 188. Käppler, A.; Fischer, D.; Oberbeckmann, S.; Schernewski, G.; Labrenz, M.; Eichhorn, K. J.; Voit, B.; Anal. Bioanal. Chem. 2016, 408, 8377. 189. Primpke, S.; Dias, P. A.; Gerdts, G.; Anal. Methods 2019, 11, 2138. 225. Sobhani, Z.; Al Amin, M.; Naidu, R.; Megharaj, M.; Fang, C.; Anal. Chim. Acta 2019, 1077, 191. 190. Karlsson, T. M.; Grahn, H.; Van Bavel, B.; Geladi, P.; J. Near Infrared Spectrosc. 2016, 24, 141. 226. Zada, L.; Leslie, H. A.; Vethaak, A. D.; Tinnevelt, G. H.; Jansen, J. J.; de Boer J F ; Ariese F ; J Raman Spectrosc 2018 49 1136 226. Zada, L.; Leslie, H. A.; Vethaak, A. D.; Tinnevelt, G. H.; Jansen, de Boer, J. F.; Ariese, F.; J. Raman Spectrosc. 2018, 49, 1136. 226. Zada, L.; Leslie, H. A.; Vethaak, A. D.; Tinnevelt, G. H.; Jansen, J. J.; de Boer, J. F.; Ariese, F.; J. Raman Spectrosc. 2018, 49, 1136. 191. Pasquini, C.; J. Braz. Chem. Soc. 2003, 14, 198. 192. Pasquini, C.; Anal. Chim. Acta 2018, 1026, 8. 227. Furstenberg, R.; Kendziora, C. A.; Papantonakis, M. R.; Nguyen, V.; McGill, R. A.; Next-Generation Spectrosc. Technol. V 2012, 8374, 837411. 193. Corradini, F.; Bartholomeus, H.; Huerta Lwanga, E.; Gertsen, H.; Geissen V; Sci Total Environ 2019 650 922 193. Corradini, F.; Bartholomeus, H.; Huerta Lwanga, E.; Gertsen, H.; Geissen, V.; Sci. Total Environ. 2019, 650, 922. 194. Pakhomova, S.; Zhdanov, I.; van Bavel, B.; Appl. Sci. 2020, 10, 1. 228. Marcott, C.; Kansiz, M.; Dillon, E.; Cook, D.; Mang, M. N.; Noda, I.; J. Mol. Struct. 2020, 1210, 128045. 195. Amigo, J. M.; Anal. Bioanal. Chem. 2010, 398, 93. 229. Hale, R. C.; Seeley, M. E.; La Guardia, M. J.; Mai, L.; Zeng, E. Y.; J. Geophys. Res. Ocean. 2020, 125, 1. 196. Vidal, C.; Pasquini, C.; Environ. Pollut. 2021, 285,117251. 197. Serranti, S.; Palmieri, R.; Bonifazi, G.; Cózar, A.; Waste Manage. 2018, 76, 117. 230. Primpke, S.; Wirth, M.; Lorenz, C.; Gerdts, G.; Anal. Bioanal. Chem. 2018, 410, 5131. 198. Zhang, Y.; Wang, X.; Shan, J.; Zhao, J.; Zhang, W.; Liu, L.; Wu, F.; Environ. Sci. Technol. 2019, 53, 5151. 231. Microplásticos: ocorrência ambiental e desafios analíticos Microplásticos: ocorrência ambiental e desafios analíticos Vol. 44, No. 10 1351 250. Sullivan, G. L.; Gallardo, J. D.; Jones, E. W.; Hollliman, P. J.; Watson, T. M.; Sarp, S.; Chemosphere 2020, 249, 126179. 283. Silva, M. M.; Maldonado, G. C.; Castro, R. O.; Felizardo, J. S.; Cardoso, R. P.; dos Anjos, R. M.; de Araújo, F. V.; Mar. Pollut. Bull. 2019, 141, 561. 251. Dierkes, G.; Lauschke, T.; Becher, S.; Schumacher, H.; Földi, C.; Ternes, 251. Dierkes, G.; Lauschke, T.; Becher, S.; Schumacher, H.; Földi, C.; Ternes, T.; Anal. Bioanal. Chem. 2019, 411, 6959. 251. Dierkes, G.; Lauschke, T.; Becher, S.; Schumacher, H.; Földi, C.; Ternes, T.; Anal. Bioanal. Chem. 2019, 411, 6959. T.; Anal. Bioanal. Chem. 2019, 411, 6959. 284. Figueiredo, G. M.; Vianna, T. M. P.; Mar. Pollut. Bull. 2018, 135, 256. 252. Okoffo, E. D.; Ribeiro, F.; O’Brien, J. W.; O’Brien, S.; Tscharke, B. J.; Gallen, M.; Samanipour, S.; Mueller, J. F.; Thomas, K. V.; Sci. Total Environ. 2020, 715, 136924. 285. Castro, R. O.; da Silva, M. L.; Marques, M. R. C.; de Araújo, F. V.; Mar. Pollut. Bull. 2020, 160, 111537. 286. Bertoldi, C.; Lara, L. Z.; Mizushima, F. A. de L.; Martins, F. C. G.; Battisti, M. A.; Hinrichs, R.; Fernandes, A. N.; Sci. Total Environ. 2020, 759, 143503. 253. Zhang, X.; Zhang, H.; Yu, K.; Li, N.; Liu, Y.; Liu, X.; Zhang, H.; Yang, B W W G J Ji J A l Ch 2020 92 4656 253. Zhang, X.; Zhang, H.; Yu, K.; Li, N.; Liu, Y.; Liu, X.; Zhang, H.; Yang, B.; Wu, W.; Gao, J.; Jiang, J.; Anal. Chem. 2020, 92, 4656. g, ; g, ; , ; , ; , ; , ; g, ; g, B.; Wu, W.; Gao, J.; Jiang, J.; Anal. Chem. 2020, 92, 4656. B.; Wu, W.; Gao, J.; Jiang, J.; Anal. Chem. 2020, 92, 4656. 254. Bannach, G.; Perpétuo, G. L.; Cavalheiro, É. T. G.; Cavalheiro, C. C. S.; Rocha, R. R.; Quim. Nova 2011, 34, 1825. 254. Bannach, G.; Perpétuo, G. L.; Cavalheiro, É. T. G.; Cavalheiro, C. C. S.; 287. Ferraz, M.; Bauer, A. L.; Valiati, V. H.; Schulz, U. H.; Water 2020, 12, 3115. Rocha, R. R.; Quim. Nova 2011, 34, 1825. 288. Lorenzi, L.; Reginato, B. C.; Mayer, D. G.; Dantas, D. V.; Environ. Sci. Pollut. Res. 2020, 27, 8797. 255. Microplásticos: ocorrência ambiental e desafios analíticos F.; Silva Cavalcanti, J. S.; Barbosa, C. C.; Portugal, J. L.; Barletta, M.; J. Coast. Res. 2011, 64, 339. Barletta, M.; J. Coast. Res. 2011, 64, 339. 264. Prata, J. C.; Alves, J. R.; da Costa, J. P.; Duarte, A. C.; Rocha-Santos, T.; Sci. Total Environ. 2020, 719, 137498. 296. do Sul, J. A. I.; Costa, M. F.; J. Coast. Res. 2013, 65, 48. 265. Gaston, E.; Woo, M.; Steele, C.; Sukumaran, S.; Anderson, S.; Appl. Spectrosc. 2020, 74, 1079. 297. Silva-Cavalcanti, J. S.; de Araújo, M. C. B.; da Costa, M. F.; Waste Manage. Res. 2009, 27, 93. 266. Iannilli, V.; Corami, F.; Grasso, P.; Lecce, F.; Buttinelli, M.; Setini, A.; Environ. Sci. Pollut. Res. 2020, 27, 14711. 298. Pinheiro, L. M.; Monteiro, R. C. P.; do Sul, J. A. I.; Costa, M. F.; Mar. Pollut. Bull. 2019, 141, 569. 267. Kankanige, D.; Babel, S.; Sci. Total Environ. 2020, 717, 137232. 299. Costa, M. F.; do Sul, J. A. I.; Silva-Cavalcanti, J. S.; Araújo, M. C. B.; Spengler, Â.; Tourinho, P. S.; Environ. Monit. Assess. 2010, 168, 299. 268. Scircle, A.; Cizdziel, J. V.; J. Chem. Educ. 2020, 97, 234. 269. Vermeiren, P.; Muñoz, C.; Ikejima, K.; Environ. Pollut. 2020, 262, 114198. 300. Araújo, M. C. B.; Silva-Cavalcanti, J. S.; Costa, M. F.; Front. Mar. Sci. 2018, 5, 1. 270. Sierra, I.; Chialanza, M. R.; Faccio, R.; Carrizo, D.; Fornaro, L.; Pérez- Parada, A.; Environ. Sci. Pollut. Res. 2020, 27, 7409. 301. Monteiro, R. C. P.; do Sul, J. A. I.; Costa, M. F.; Ocean Coast. Res. 2020, 68, e20235. 302. Santos, I. R.; Friedrich, A. C.; Barretto, F. P.; Mar. Pollut. Bull. 2005, 50, 778. 271. Bertoldi, C.; Lara, L. Z.; Gomes, A. A.; Fernandes, A. N.; Microchem. J. 2021, 160, 105690. 272. Gauci, A.; Deidun, A.; Montebello, J.; Abela, J.; Galgani, F.; Ocean & Coastal Management 2019, 182, 104950. 303. Santos, I. R.; Friedrich, A. C.; do Sul, J. A. I.; Environ. Monit. Assess. 2009, 148, 455. 273. Chetwynd, A.; Nel, H. A.; Krause, S.; Lynch, I. In SETAC Europe 30th Annual Meeting Abstract Book; 2020; p. 34. 304. do Sul, J. A. I.; Santos, I. R.; Friedrich, A. C.; Matthiensen, A.; Fillmann, G.; Estuaries and Coasts 2011, 34, 814. 274. Schirinzi, G. F.; Llorca, M.; Seró, R.; Moyano, E.; Barceló, D.; Abad, E.; Farré, M.; Chemosphere 2019, 236, 124321. 305. Leite, A. S.; Santos, L. Microplásticos: ocorrência ambiental e desafios analíticos L.; Costa, Y.; Hatje, V.; Mar. Pollut. Bull. 2014, 81, 242. 275. Zarfl, C.; Anal. Bioanal. Chem. 2019, 411, 3743. 306. Majer, A. P.; Vedolin, M. C.; Turra, A.; Mar. Pollut. Bull. 2012, 64, 1143. 307. Fernandino, G.; Elliff, C. I.; Silva, I. R.; Bittencourt, A. C. S. P.; Journal of Integrated Coastal Zone Management 2015, 15, 325. 276. Lima, A. R. A.; Barletta, M.; Costa, M. F.; Estuar. Coast. Shelf Sci. 2015, 165, 213. 277. Barletta, M.; Costa, M. F.; Dantas, D. V.; MethodsX 2020, 7, 100861. 308. de Souza, J. L.; Silva, I. R.; Soc. Nat. 2015, 27, 469. 278. Ramos, J. A. A.; Pessoa, W. V. N.; Mar. Pollut. Bull. 2019, 142, 428. 309. Santos, A. A.; Nobre, F. S. de M.; Ribeiro, F.; Nilin, J.; Mar. Pollut. Bull. 2020, 153, 111015. 279. Possatto, F. E.; Spach, H. L.; Cattani, A. P.; Lamour, M. R.; Santos, L. O.; Cordeiro, N. M. A.; Broadhurst, M. K.; Mar. Pollut. Bull. 2015, 91, 548. 310. Moruzzi, R. B.; Speranza, L. G.; Conceição, F. T.; Martins, S. T. de S.; 310. Moruzzi, R. B.; Speranza, L. G.; Conceição, F. T.; Martins, S. T. de S.; Busquets, R.; Campos, L. C.; Water 2020, 12, 1994. Busquets, R.; Campos, L. C.; Water 2020, 12, 1994. 311. Fisner, M.; Taniguchi, S.; Majer, A. P.; Bícego, M. C.; Turra, A.; Mar. Pollut. Bull. 2013, 76, 349. 280. Fernandino, G.; Elliff, C. I.; Frutuoso, G. A.; da Silva, E. V. N. M.; Gama, G. S.; Sousa, J. H. de O.; Silva, I. R.; Mar. Pollut. Bull. 2016, 110, 591. 312. Fisner, M.; Majer, A.; Taniguchi, S.; Bícego, M.; Turra, A.; Gorman, D.; Mar. Pollut. Bull. 2017, 122, 323. 281. Garcia, T. M.; Campos, C. C.; Mota, E. M. T.; Santos, N. M. O.; Campelo, R. P. de S.; Prado, L. C. G.; Melo Junior, M.; Soares, M. de O.; Mar. Pollut. Bull. 2020, 150, 110705. 313. Gorman, D.; Gutiérrez, A. R.; Turra, A.; Manzano, A. B.; Balthazar-Silva, D Oli i N R H i J F t E i S i 2020 8 559405 313. Gorman, D.; Gutiérrez, A. R.; Turra, A.; Manzano, A. B.; Balthazar-Silva, D.; Oliveira, N. R.; Harari, J.; Front. Environ. Sci. 2020, 8, 559405. 313. Gorman, D.; Gutiérrez, A. R.; Turra, A.; Manzano, A. B.; Balthazar Silva, D.; Oliveira, N. R.; Harari, J.; Front. Environ. Sci. 2020, 8, 559405. 282. Olivatto, G. REFERÊNCIAS Pan, Z.; Liu, Q.; Sun, Y.; Sun, X.; Lin, H.; Mar. Pollut. Bull. 2019, 146, 215. 240. Frère, L.; Paul-Pont, I.; Moreau, J.; Soudant, P.; Lambert, C.; Huvet, A.; Rinnert, E.; Mar. Pollut. Bull. 2016, 113, 461. Rinnert, E.; Mar. Pollut. Bull. 2016, 113, 461. 241. Hufnagl, B.; Lohninger, H.; Anal. Chim. Acta 2020, 1097, 37. 209. Sobhani, Z.; Zhang, X.; Gibson, C.; Naidu, R.; Megharaj, M.; Fang, C.; Water Res. 2020, 174, 115658. 242. Renner, G.; Sauerbier, P.; Schmidt, T. C.; Schram, J.; Anal. Chem. 2019, 91, 9656. 210. Sruthy, S.; Ramasamy, E. V.; Environ. Pollut. 2017, 222, 315. 243. Shim, W. J.; Hong, S. H.; Eo, S. E.; Anal. Methods 2017, 9, 1384. 211. Tong, H.; Jiang, Q.; Hu, X.; Zhong, X.; Chemosphere 2020, 252, 126493. 244. Yakovenko, N.; Carvalho, A.; ter Halle, A.; TrAC – Trends Anal. Chem. 2020, 131, 115979. 212. Xiong, X.; Zhang, K.; Chen, X.; Shi, H.; Luo, Z.; Wu, C.; Environ. Pollut. 2018, 235, 899. 245. Fischer, M.; Scholz-Böttcher, B. M.; Anal. Methods 2019, 11, 2489. 213. Yin, L.; Wen, X.; Du, C.; Jiang, J.; Wu, L.; Zhang, Y.; Hu, Z.; Hu, S.; Feng, Z.; Zhou, Z.; Long, Y.; Gu, Q.; Chemosphere 2020, 244, 125486. 246. Picó, Y.; Barceló, D.; TrAC – Trends Anal. Chem. 2020, 115964. 247. Zhang, Y.; Kang, S.; Allen, S.; Allen, D.; Gao, T.; Sillanpää, M.; Earth- Science Rev. 2020, 203, 103118. 214. Zhao, S.; Zhu, L.; Li, D.; Environ. Pollut. 2015, 206, 597. 248. Peñalver, R.; Arroyo-Manzanares, N.; López-García, I.; Hernández- Córdoba, M.; Chemosphere 2020, 242, 125170. 215. Cole, M.; Lindeque, P.; Fileman, E.; Halsband, C.; Goodhead, R.; Moger, J.; Galloway, T. S.; Environ. Sci. Technol. 2013, 47, 6646. 216. Ghosal, S.; Chen, M.; Wagner, J.; Wang, Z. M.; Wall, S.; Environ. Pollut. 2018, 233, 1113. 249. Steinmetz, Z.; Kintzi, A.; Muñoz, K.; Schaumann, G. E.; J. Anal. Appl. Pyrolysis 2020, 147, 104803. Microplásticos: ocorrência ambiental e desafios analíticos Duemichen, E.; Braun, U.; Senz, R.; Fabian, G.; Sturm, H.; J. Chromatogr. A 2014, 1354, 117. 289. Gerolin, C. R.; Pupim, F. N.; Sawakuchi, A. O.; Grohmann, C. H Labuto, G.; Semensatto, D.; Sci. Total Environ. 2020, 749, 141604. 256. Duemichen, E.; Eisentraut, P.; Celina, M.; Braun, U.; J. Chromatogr. A 2019, 1592, 133. 9. Gerolin, C. R.; Pupim, F. N.; Sawakuchi, A. O.; Grohma Labuto, G.; Semensatto, D.; Sci. Total Environ. 2020, 749, 141604. 290. Novaes, G. de O.; Monteiro, S. de M.; Rollnic, M.; J. Coast. Res. 2020, 95, 780. 257. Oßmann, B.; Schymanski, D.; Ivleva, N. P.; Fischer, D.; Fischer, F.; Dallmann, G.; Welle, F.; Water Res. 2019, 162, 516. 291. Martinelli Filho, J. E.; Monteiro, R. C. P.; Mar. Pollut. Bull. 2019, 145, 219. 258. Zuccarello, P.; Ferrante, M.; Cristaldi, A.; Copat, C.; Grasso, A.; Sangregorio, D.; Fiore, M.; Oliveri Conti, G.; Water Res. 2019, 166, 115077. 292. Palombini, F. L.; Demori, R.; Cidade, M. K.; Kindlein, W.; de Jacques, 292. Palombini, F. L.; Demori, R.; Cidade, M. K.; Kindlein, W.; de Jacques, J. J.; Environ. Sci. Pollut. Res. 2018, 25, 26218. J. J.; Environ. Sci. Pollut. Res. 2018, 25, 26218. 259. K. V., A.; C. C., H.; N. K., A.; Algal Res. 2020, 48, 101924. J. J.; Environ. Sci. Pollut. Res. 2018, 25, 26218. 293. Belarmino, P. H. P.; da Silva, S. M.; Rufener, M.-C.; de Araújo, M. C. B.; Rev. Gestão Costeira Integr. 2014, 14, 447. 260. Yan, R.; Moon, S.; Kenny, S. J.; Xu, K.; Acc. Chem. Res. 2018, 51, 697. 261. Cole, M.; Sci. Rep. 2016, 6, 1. 294. Cavalcante, R. M.; Pinheiro, L. S.; Teixeira, C. E. P.; Paiva, B. P.; 262. Shim, W. J.; Song, Y. K.; Hong, S. H.; Jang, M.; Mar. Pollut. Bull. 2016, 113, 469. 294. Cavalcante, R. M.; Pinheiro, L. S.; Teixeira, C. E. P.; Paiva, B. P.; Fernandes, G. M.; Brandão, D. B.; Frota, F. F.; Filho, F. J. N. S.; Schettini, C. A. F.; Waste Manage. 2020, 108, 13. Fernandes, G. M.; Brandão, D. B.; Frota, F. F.; Filho, F. J. N. S.; 263. Maes, T.; Jessop, R.; Wellner, N.; Haupt, K.; Mayes, A. G.; Sci. Rep. 2017, 7, 1. Schettini, C. A. F.; Waste Manage. 2020, 108, 13. 295. Costa, M. F.; Silva-Cavalcanti, J. S.; Barbosa, C. C.; Portugal, J. L.; Barletta, M.; J. Coast. Res. 2011, 64, 339. 295. Costa, M. This is an open-access article distributed under the terms of the Creative Commons Attribution License. Microplásticos: ocorrência ambiental e desafios analíticos P.; Martins, M. C. T.; Montagner, C. C.; Henry, T. B.; Carreira, R. S.; Mar. Pollut. Bull. 2019, 139, 157. 314. Cordeiro, T. C.; Barrella, W.; Butturi-Gomes, D.; Petrere Júnior, M.; Mar. Pollut. Bull. 2018, 128, 333. 1352 Montagner et al. Quim. Nova 315. Fisner, M.; Taniguchi, S.; Moreira, F.; Bícego, M. C.; Turra, A.; Mar. Pollut. Bull. 2013, 70, 219. 328. Suciu, M. C.; Tavares, D. C.; Costa, L. L.; Silva, M. C. L.; Zalmon, I. R.; Mar. Pollut. Bull. 2017, 119, 133. 316. Cordeiro, C. A. M. M.; Costa, T. M.; Mar. Pollut. Bull. 2010, 60, 1762. 329. da Silva, M. L.; Castro, R. O.; Sales, A. S.; de Araújo, F. V.; Mar. Pollut. Bull. 2018, 130, 153. 317. Balthazar-Silva, D.; Turra, A.; Moreira, F. T.; Camargo, R. M.; Oliveira, A. L.; Barbosa, L.; Gorman, D.; Front. Environ. Sci. 2020, 8, 123. 330. Oigman-Pszczol, S. S.; Creed, J. C.; J. Coast. Res. 2007, 23, 421. 318. Turra, A.; Manzano, A. B.; Dias, R. J. S.; Mahiques, M. M.; Barbosa, L.; Balthazar-Silva, D.; Moreira, F. T.; Sci. Rep. 2014, 4, 4435. 331. Andrades, R.; Martins, A. S.; Fardim, L. M.; Ferreira, J. S.; Santos, R. G.; Mar. Pollut. Bull. 2016, 109, 192. 332. Moreira, F. T.; Prantoni, A. L.; Martini, B.; de Abreu, M. A.; Stoiev, S. B T A M P ll t B ll 2016 102 114 332. Moreira, F. T.; Prantoni, A. L.; Martini, B.; de Abreu, M. A.; Stoie B.; Turra, A.; Mar. Pollut. Bull. 2016, 102, 114. 319. Moreira, F. T.; Balthazar-Silva, D.; Barbosa, L.; Turra, A.; Environ. Pollut. 2016, 218, 313. B.; Turra, A.; Mar. Pollut. Bull. 2016, 102, 114. 333. Krelling, A. P.; Souza, M. M.; Williams, A. T.; Turra, A.; Mar. Pollut. Bull. 2017, 119, 48. 320. Taniguchi, S.; Colabuono, F. I.; Dias, P. S.; Oliveira, R.; Fisner, M.; Turra, A.; Izar, G. M.; Abessa, D. M. S.; Saha, M.; Hosoda, J.; Yamashita, R.; Takada, H.; Lourenço, R. A.; Magalhães, C. A.; Bícego, M. C.; Montone, R. C.; Mar. Pollut. Bull. 2016, 106, 87. 334. Gorman, D.; Moreira, F. T.; Turra, A.; Fontenelle, F. R.; Combi, T.; Bícego, M. C.; de Castro Martins, C.; Chemosphere 2019, 223, 608. 334. Gorman, D.; Moreira, F. T.; Turra, A.; Fontenelle, F. R.; Combi, T.; Bí M C d C M i C Ch h 2019 223 608 M. C.; Montone, R. C.; Mar. Pollut. Microplásticos: ocorrência ambiental e desafios analíticos Bull. 2016, 106, 87. Bícego, M. C.; de Castro Martins, C.; Chemosphere 2019, 223, 608 35. Krelling, A. P.; Turra, A.; Mar. Pollut. Bull. 2019, 139, 282 321. Vedolin, M. C.; Teophilo, C. Y. S.; Turra, A.; Figueira, R. C. L.; Mar. Pollut. Bull. 2018, 129, 487. 336. Marin, C. B.; Niero, H.; Zinnke, I.; Pellizzetti, M. A.; Santos, P. H.; Rudolf, A. C.; Beltrão, M.; Waltrick, D. de S.; Polette, M.; Reg. Stud. Mar. Sci. 2019, 31, 100771. 322. Gimiliani, G. T.; Fornari, M.; Redígolo, M. M.; Bustillos, J. O. W. V.; Abessa, D. M. de S.; Pires, M. A. F.; Case Studies in Chemical and Environmental Engineering 2020, 2, 100020. 337. Corraini, N. R.; de Lima, A. S.; Bonetti, J.; Rangel-Buitrago, N.; Mar. Pollut. Bull. 2018, 131, 572. 323. da Silva, M. L.; de Araújo, F. V.; Castro, R. O.; Sales, A. S.; Mar. Pollut. Bull. 2015, 92, 233. 338. Widmer, W. M.; Hennemann, M. C.; J. Coast. Res. 2010, 26, 993. 324. da Silva, M. L.; Sales, A. S.; Martins, S.; Castro, R. de O.; de Araújo, F. V.; Mar. Pollut. Bull. 2016, 113, 36. 339. Farias, E. G. G.; Preichardt, P. R.; Dantas, D. V.; Environ. Sci. Pollut. Res. 2018, 25, 16246. 325. Costa, L. L.; Rangel, D. F.; Zalmon, I. R.; Mar. Pollut. Bull. 2018, 128, 438. 340. Santos, I. R.; Friedrich, A. C.; Wallner-Kersanach, M.; Fillmann, G.; Ocean & Coastal Management 2005, 48, 742. 326. Perez, L.; Soares-Gomes, A.; Bernardes, M. C.; J. Coast. Conserv. 2018, 22, 1085. 341. Wetzel, L.; Fillmann, G.; Niencheski, L. F. H.; Int. J. Environ. Pollut. 2004, 21, 153. 327. Alves, V. E. N.; Figueiredo, G. M.; Mar. Pollut. Bull. 2019, 146, 326.
https://openalex.org/W2950702831	https://www.biorxiv.org/content/biorxiv/early/2017/06/30/157842.full.pdf	English	null	Morphometrics of complex cell shapes: Lobe Contribution Elliptic Fourier Analysis (LOCO-EFA)	bioRxiv (Cold Spring Harbor Laboratory)	2,017	cc-by	34,549	Short title: Morphometrics of complex cell shapes Keywords: Cell shape \| Cellular Potts Model \| Image analysis \| Pavement cells \| Arabidopsis thaliana \| Drosophila \| . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint Morphometrics of complex cell shapes: Lobe Contribution Elliptic Fourier Analysis (LOCO-EFA) Morphometrics of complex cell shapes: Lobe Contribution Elliptic Fourier Analysis (LOCO-EFA) Yara E. S´anchez-Corrales1,3, Matthew Hartley1, Jop van Rooij1,2, Athanasius F. M. Mar´ee1,4, Verˆonica A. Grieneisen1,4 Yara E. S´anchez-Corrales1,3, Matthew Hartley1, Jop van Rooij1,2, Athanasius F. M. Mar´ee1,4, Verˆonica A. Grieneisen1,4 Yara E. S´anchez-Corrales1,3, Matthew Hartley1, Jop van Rooij1,2, Athanasius F. M. Mar´ee1,4, Verˆonica A. Grieneisen1,4 29th June 2017 Abstract Quantifying cell morphology is fundamental to the statistical study of cell popu- lations, and can help us unravel mechanisms underlying cell and tissue morphogen- esis. Current methods, however, require extensive human intervention, are highly sensitive to parameter choice, or produce metrics that are diﬃcult to interpret biologically. We therefore developed a novel method, Lobe Contribution Elliptical Fourier Analysis (LOCO-EFA), which generates from digitalised cell outlines mean- ingful descriptors that can be directly matched to morphological features. We show that LOCO-EFA provides a tool to phenotype eﬃciently and objectively popula- tions of cells by applying it to the complex shaped pavement cells of Arabidopsis thaliana wild type and speechless leaves. To further validate our method, we ana- lysed computer-generated tissues, where cell shape can be speciﬁed in a controlled manner. LOCO-EFA quantiﬁes deviations between the speciﬁed shape that an individual in silico cell takes up when in isolation and the resultant shape when they are allowed to interact within a conﬂuent tissue, thereby assessing the role of cell-cell interactions on population cell shape distributions. Summary statement Novel method (LOCO-EFA) quantiﬁes complex cell shapes, extracting meaningful biolo- gical features such as protrusion number and amplitude; here shown for plant pavement cells and validated on in silico tissues. 29th June 2017 1Computational and Systems Biology, John Innes Centre, Norwich NR4 7UH, UK. 2Theoretical Biology/Bioinformatics, Dept. of Biology, Utrecht University, Padualaan 8, 3584 CH Utrecht, The Netherlands. 1Computational and Systems Biology, John Innes Centre, Norwich NR4 7UH, UK. 2Theoretical Biology/Bioinformatics, Dept. of Biology, Utrecht University, Padualaan 8, 3584 CH Utrecht, The Netherlands. 3Current address: MRC-Laboratory of Molecular Biology, Cambridge Biomedical Campus, Francis Crick Avenue, Cambridge CB2 0QH, UK 4To whom correspondence should be addressed. E-mail: Stan.Maree@jic.ac.uk, Veronica.Grieneisen@jic.ac.uk Short title: Morphometrics of complex cell shapes Keywords: Cell shape \| Cellular Potts Model \| Image analysis \| Pavement cells \| Arabidopsis thaliana \| Drosophila \| Short title: Morphometrics of complex cell shapes 1 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint Introduction CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint To illustrate the issues involved in capturing quantitative aspects of complex cell shapes, consider the case of pavement cells in the plant epidermis (Figure 1A, B) or am- nioserosa cells in the Drosophila embryo (Figure 1C). Pavement cells present a striking development, requiring multiple locally divergent growth fronts within each cell, which are coordinated amongst neighbouring cells; amnioserosa cells present dynamical cell shape ﬂuctuations within a conﬂuent tissue. The resultant jigsaw-puzzle-like features of such cells illustrate the challenges in quantitatively characterising cell shape: 1) their com- plex non-holomorphic geometry cannot be captured in a meaningful way with traditional shape metrics (see below); and 2) the lack of recognisable landmarks (neither these plant nor animal cells have distinct features to serve as reference points), renders a myriad of shape statistics that have been developed in other ﬁelds (such as principal component and Procrustes analysis) eﬀectively useless (Klingenberg, 2010). y ) y ( g g, ) Traditional metrics for cell morphology include cell area, cell perimeter, aspect ratio and form factor. Useful as general descriptors, the shape information that can be extrac- ted from them is limited. Very diﬀerent shapes may yield a similar aspect ratio or form factor (Figure 1D–H). Besides not being unique, such descriptors tend to omit informa- tion regarding biologically relevant shape features (Fu et al., 2005). Several approaches classically used to quantify cells, including pavement cells, are summarised in Table 1. Some of those methods, such as the skeleton method, are highly sensitive to image noise as well as to the precise choice of parameters (for an example, see Le et al., 2006). Other metrics, such as lobe length and neck width assessments (Fu et al., 2005), are entirely dependent on humans to judge what a lobe is. Such decisions then strongly impact the average lobe length and neck width measurements (Figure 1 and Figure S1). It renders these metrics highly variable from cell to cell, from phenotype to phenotype and from human to human. In this respect, recently an automatic method was developed to count lobes and indentations (Wu et al., 2016), making this process more reliable. Introduction Cell geometry has long fascinated biologists (Thompson, 1917). This interest is driven by a wide range of underlying scientiﬁc questions. For instance, cell shape changes can be linked to physiological responses of cells, such as membrane protrusions during apoptosis and migration (Charras and Paluch, 2008); they are involved in controlling tissue homeo- stasis (Marinari et al., 2012; Veeman and Smith, 2013) and underlie cell behaviour such as chemotaxis to ensure survival (Keren et al., 2008; Driscoll et al., 2012). The cell shape it- self also inﬂuences intracellular processes such as microtubule organisation (Gomez et al., 2016) and the stress patterns in plant epithelia (Sampathkumar et al., 2014); it indirectly positions the plane of cell division (Minc et al., 2011) and can even determine the way a ﬂower attracts pollinators (Noda et al., 1994). Given the rich diversity of processes in which cell shape plays a decisive role, either actively or passively, it is not surprising that the qualitative and quantitative study of cell shape characteristics, cell morphometrics, has become important in developmental biology, enriching molecular and genetic tech- niques. In parallel, recent advances in imaging technology, including microscopy, usage of ﬂuorescent proteins and imaging software, allow us to collect and access remarkable amounts of cell morphological data, which in turn urgently calls for analytic tools for the extraction of meaningful cell shape information (Zhong et al., 2012). In stark contrast to the technological advances on the imaging front, there are relatively few automatic and quantitative tools available to analyse complex cell shapes (Rajaram et al., 2012; Ljosa et al., 2012; Ivakov and Persson, 2013). This gap reﬂects the non-trivial nature of this task: cell shape is often irregular and variable, making it very diﬃcult to establish universal criteria encompassing cell geometry. 2 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . Introduction As we will discuss further on, such a method, although already very useful, ﬁnds its limitations when the characteristics of the shape reside in the distribution and amplitude of lobes, not only in their number. For instance, some Arabidopsis mutants can present pavement cells with lobes that look more elongated or shallower, but which occur at a similar spatial frequency (Lin et al., 2013). ( ) Promising alternatives are methods that take into account the full cell outline in- formation and then simplify the contour of a cell into a series of coeﬃcients that can be employed as shape descriptors in a multivariate study (Ivakov and Persson, 2013; Pincus and Theriot, 2007). One example of these methods is Elliptical Fourier Analysis (EFA), that has been used to quantify two-dimensional complex shapes (Kuhl and Giardina, 1982; Diaz et al., 1989; Schmittbuhl et al., 2003). It takes the coordinates of the outline and decomposes them into a series of related ellipses (described by EFA coeﬃcients) that can be combined to reconstitute that given shape. Despite its wide usage in morphometric studies, EFA cannot retrieve information that directly relates to morphological features of a cell, obstructing biological interpretation. This is because the EFA coeﬃcients are highly dependent on how the cell outline is approximated (a detailed explanation is given in the Supplementary Materials and Methods). Here we present a new method based on EFA, termed Lobe-Contribution Elliptic Fourier Analysis (LOCO-EFA), that overcomes the common obstacles described above. Our method also uses the information of the whole cell contour but, unlike EFA, provides a set of metrics that directly relate to morphological features, permitting the assessment of cell shape complexity in an objective and automatic manner. Importantly, it is not 3 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint sensitive to cell orientation or imaging resolution. Introduction To validate the usage of our method on larger cellular datasets, we analyse pavement cell populations from confocal images of Arabidopsis thaliana. We then complement this study with the analysis of synthetic tissues generated using the Cellular Potts model (Graner and Glazier, 1992; Glazier and Graner, 1993), in which pavement-like cells have a parametrised speciﬁed shape. The latter approach also allows us to touch on a fun- damental question in developmental biology, by asking to what degree the resultant cell shape within a conﬂuent tissue context is shaped by cell-to-cell interactions and to what degree it can be explained by intracellular shape control mechanisms. Speciﬁcally, here we explore the inﬂuence of cell neighbours on the morphogenesis of individual cells by measuring the divergence of their speciﬁed cell shape within an isolated context to the shape taken up when being immersed within a tissue. This analysis, performed on in silico tissues, allows us to quantify such diﬀerences for a generic situation by combining CPM and LOCO-EFA . Quantitative characterisation of cell shape using LOCO-EF CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint while orbiting around the previous (n −1) harmonic ellipse, which in its turn is orbiting around its previous one (n −2), and so forth (Figure S3B). This summation results in an outline being ‘drawn’, as is visualised in Movie S1. Obviously, a cut-oﬀhas to be chosen for the number of modes that are taken into account. In general this will be at a value for which the cell contour description is reasonably close to the original cell outline. The fact that each ellipse represents a certain harmonic suggests that it captures dominant spatial frequencies within the original shape. This is why it has been considered to be a reasonable descriptor for cell outline properties (Schmittbuhl et al., 2003). When shapes are quantiﬁed using the EFA method, however, the pitch, i.e., the most basic cellular feature to quantify, is actually not directly retrieved, not even for the most simple shapes. For instance, one might expect a six-sided shape to present a strong contribution from the 6th mode. Instead, the EFA method represents such a shape as a mixed contribution from the two adjacent modes, the 5th and 7th mode (Figure 2D and Figure S3C). This mismatch is a consequence of how the individual ellipses obtained from decomposing the outline contribute to the outline. Each ellipse does so by rotating either clockwise or counterclockwise. The direction of this rotation in respect to the direction of rotation of the ﬁrst mode causes either an increase or a decrease in the actual number of features being drawn, always one oﬀfrom the actual mode (Figure S3D, E, Movie S2 and Movie S3). As a consequence, the ‘pitch’ obtained using EFA does not correspond to actual cell features, which hinders interpreting the resultant values of this method. Quantitative characterisation of cell shape using LOCO-EF Applying EFA to quantify cell shapes, we came across a number of speciﬁc shortcomings. We ﬁrst explain those issues to highlight our motivation and choices that led to the development of LOCO-EFA. Detailed mathematical implementation is provided in the Supplementary Materials and Methods. Here, we focus on explaining the analysis in terms of its biological relevance, emphasising how it can be applied and interpreted. The shape analysis proposed here is linked to harmonic compositions of digitalised cell shape outlines. We ﬁnd it useful to compare the decomposition of a complex cell shape, such as that of a pavement cell, to the way one decomposes the sounds of mu- sical instruments. This analogy will be used throughout the manuscript for purposes of illustrating the diﬀerent traits that our morphometric analysis is capturing. For ex- ample, when listening to a note generated by an instrument, a quantiﬁable observable is the pitch. Within the context of pavement cell shape this corresponds to the observed number of lobes or, as we will explain in detail, to the dominant spatial frequency of the pavement cell outline. Another quantiﬁable property of musical note is its volume, the amplitude. In terms of cell shape this corresponds to the extent to which lobes and indentations are protruded and retracted. Finally, the timbre of musical instruments is what essentially distinguishes, for example, a clarinet from an oboe, as both play the same note (same pitch) at the same volume. An analogous notion in cell shape studies would be the ability to capture additional aspects of cell shape morphology, as well as to recognise subtle diﬀerences between cells, even when the number of lobes and their level of protrusion is the same. As a starting point, Elliptical Fourier Analysis (EFA, Kuhl and Giardina, 1982) can be used to describe the contour of any complex two-dimensional shape, including non- holomorphic shapes such as pavement cells, which most other methods are unable to capture (see Figure S2 and Supplementary Materials and Methods). Using the coordinates of the two-dimensional (cell) outline (Figure S3A), EFA decomposes the shape into an inﬁnite series of ellipses (also referred to as “modes” or “harmonics”). This series of ellipses, n = 1 · · · ∞, can then be combined in the following way to exactly retrieve the original shape: each nth elliptic harmonic traces n revolutions around the ﬁrst ellipse 4 . Quantitative characterisation of cell shape using LOCO-EF CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint LOCO-EFA, thus consists of: eliminating multiple representations of a given outline; de- composing each nth elliptical EFA mode into two separate lobe contributions; and ﬁnally integrating those separate modes into single LOCO-EFA modes. Every Ln mode can then be regarded to represent two oppositely rotating circles, each with its own starting point for the rotation, the four Ln coeﬃcients providing the radii and starting angles of both circles. We next assign a scalar Ln value, to speciﬁcally capture the amplitude contri- bution (“volume”) of each mode. We found that in order to correctly capture this, both the radii of and the angular distance between the starting points of the two contributing circles, as well as the starting point of the main circle, L1, have to be taken into account (see Figure 2C, and Supplementary Materials and Methods for further details regarding its derivation). Calculating the Ln values results in a spectrum representing the relative contribution of each individual mode to the cell shape (Figure S3C). For instance, the spectrum of the six-lobed test shape used for Figure S3C indeed contains a pronounced peak at mode six, as well as a peak at mode one, which represents the overall circular shape of that outline. The original shape can be reconstructed using all modes up to a speciﬁc mode number. This allows us to visually appreciate the speciﬁc contribution of mode six to the six-lobed test shape (compare Figure 2E with Figure 2D). ( ) To further illustrate how LOCO-EFA characterises shapes, we apply it to simple geo- metrical shapes with variable numbers of protrusions (Figure 3A–I). Indeed, LOCO-EFA robustly determines the main Ln-mode of each shape corresponding to the correct num- ber of lobes (Figure 3J). Quantitative characterisation of cell shape using LOCO-EF Moreover, EFA coeﬃcients on their own are redundant, i.e., there are more parameters than needed to specify the same speciﬁc shape (Haines and Crampton, 2000). Therefore, comparison of cell shapes on the basis of their EFA coeﬃcients (for example, by means of principle component analysis) becomes nonsensical. Together, these traits make the EFA method unsuitable for reliable quantiﬁcation of cell morphology and renders prohibitive any meaningful comparisons between multiple cell outlines. Diaz et al. (1990) proposed a heuristic solution for the mismatch between actual shape features and EFA’s dominant harmonics, using the fact that the relative direction of rotation is a main determinant of the generated pitch (see Supplementary Materials and Methods for details). It turns out, however, that each ellipse is simultaneously contributing to two diﬀerent spatial frequencies, something this heuristic can not solve (see Figure S3F and Movie S4). As a consequence, although their method is often (but not always) able to correctly recapitulate the ‘pitch’, it is never able to correctly capture the amplitude or timbre of the cell shape. Further details regarding the range of issues when using EFA are discussed in depth in the Supplementary Materials and Methods. To overcome these limitations we essentially propose a new basis for the outline recon- struction, which we have coined Ln, after lobe number. In a similar manner as EFA, the modes can be summed up to recreate the original shape and each individual mode is cap- tured by a set of four unique parameters. However, there are two important distinctions. Firstly, any possible cell outline is now only represented by one unique combination of Ln coeﬃcients. In the Supplementary Materials and Methods we discuss in detail the steps required to eliminate all levels of redundancy for shape analysis. Secondly, cell shape fea- tures such as the protrusion number (the“pitch”), their amplitude (the“volume”) and the characteristic lobe distributions (“timbre”) are directly represented by the Ln coeﬃcients. The Ln coeﬃcients are obtained after decomposing each EFA harmonic into its exact, spe- ciﬁc contributions to two separate Ln modes (Figure 2A–C). In general, the EFA modes n −1 and n + 1 both partly contribute to mode Ln, although some speciﬁc exceptions apply (see Figure S4). The resulting method, which we coin Lobe Contribution-EFA or 5 . Quantitative characterisation of cell shape using LOCO-EF CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint each subsequent Ln mode is for explaining that speciﬁc shape. Figure 2E already illus- trated the importance of a speciﬁc mode for reconstructing the original shape (in that case, mode six). One can straightforwardly quantify the relative contribution of that mode for explaining the shape by calculating the total diﬀerence (in area, expressed as either number of grid points or µm2) between the original shape and the superimposed reconstructed shape when the ﬁrst N LOCO-EFA modes are used. To do so, we take the XOR (exclusive or) between the original and reconstructed cells (see Figure S5). In this context, a more “complex” shape is one in which more LOCO-EFA modes are needed to obtain a good match between the reconstruction and the real shape (i.e., when the XOR approaches zero only at higher N numbers). For this criterion, note that a circular cell can be perfectly reconstituted using only the contribution of the ﬁrst LOCO-EFA mode (N = 1), and is therefore the least complex shape with an XOR value of zero for the whole spectrum. On the other hand, cells presenting a high number of heterogeneous lobes require a high number of modes for XOR to approach zero (Figure 3). This can be observed in Figure 3K, U, which show the XOR diﬀerence proﬁles for the series of test shapes. This manner of deﬁning cellular complexity can be further compressed by taking the total area under the XOR diﬀerence curve, a scalar quantity which we call the cumulative diﬀerence (cd). It derives a single value from the XOR spectrum, with higher values for more complex shaped cells. The more similar the cell shape is to a circle (which can be perfectly described using the ﬁrst LOCO-EFA mode only), the more the cd value approximates zero. Figure 3M, W shows the cd values for the series of test shapes. From these, it becomes clear that cd becomes high for shapes that strongly deviate from circular, it is therefore not just correlated with having many lobes. Quantitative characterisation of cell shape using LOCO-EF Each Ln proﬁle provides information about the composition of all morphological periodicities contributing to the shape, reﬂecting the geometry being considered. We next tested if LOCO-EFA also correctly captures the “volume”, by ap- plying the method to generated shapes of the same “pitch” (i.e., having a same number of lobes), but with variable amplitudes (Figure 3N–Q). Indeed, the magnitude of the Ln components changes accordingly (Figure 3T), and their absolute value quantitatively cor- responds to the size of the extensions. Thus, LOCO-EFA not only retrieves the main number of morphological features of hypothetical cell shapes, but also a range of addi- tional characteristics. Following the analogy of sound decomposition, a more nuanced perception for quan- tiﬁcation is timbre. Timbre can be understood as residing in the additional amplitude spectrum. It is determined by which overtones are emphasised in relation to one another. For cell shape studies, we consider “timbre” analysis the ability to capture additional aspects of the cell shape complexity, besides the main number of protrusions/lobes which a cell presents. This additional information should enable distinction between diﬀerent cellular phenotypes, such as between wild type and mutants (Lin et al., 2013). To il- lustrate, Figure 3R, S show two additional six-sided shapes which diﬀer in “timbre” from Figure 3Q, with their accompanying Ln values (Figure 3T). For both shapes, there is a clear peak at L6, reﬂecting their six-lobedness, and additional peaks or overtones, such as L2, capturing the elongated nature of these shapes, and so forth. Thus, LOCO-EFA retrieves not only the main number of morphological features of an hypothetical cell (i.e., lobes), but also important ﬁne-grained characteristics. From the full characterisation of cell shape by Ln modes, additional objective metrics can be derived to help quantify diﬀerent aspects of“cell shape complexity”. We here deﬁne four metrics: XOR diﬀerence; marginal diﬀerence; cumulative diﬀerence; and entropy. Firstly, complementary to the magnitude of each individual Ln mode, the cell shape complexity can be estimated using the information of the approximation to the original shape by the ﬁrst N LOCO-EFA modes only. It addresses the question how relevant 6 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . LOCO-EFA applied to plant pavement cells To validate our method, and its derived measures introduced above, we analysed Ara- bidopsis thaliana leaf epidermal pavement cells, individual cells that were tracked in vivo over time as well as whole populations of cells at a single time point. Actual biological cells, such as pavement cells, can be highly asymmetrical, with their Ln landscape char- acterised by multiple peaks (Figure 3S, T). This is because the outline of an asymmetrical cell with a given number of protrusions placed quasi-periodically along its edge can be interpreted as multiple protrusion frequencies superimposed. In general, the total number of hand-counted lobes in a cell image matches to a peak at the corresponding Ln value (but note here that hand-counting is very subjective). For instance, for nine lobes a peak will be observed at L9. However, if the lobes are distributed in a more or less pentagonal clustered fashion, this would lead to an additional peak at L5, whilst superimposed on a triangular shaped cell basis a L3 contribution would be found, and so forth. Pavement cells acquire their characteristic jigsaw puzzle-like shape through multipolar growth patterns, such that relative simple shaped cells become highly complex over time (Figure 4A–G). Notably, the smooth shape changes over time are clearly reﬂected in the Ln proﬁles over time (Figure 4I). Its initial squarish shape and later nine- and thirteen- lobeness are well captured by the LOCO-EFA method, through peaks at mode L4, L9, and L13, and corresponding peaks in the marginal diﬀerence proﬁle. In contrast, when the EFA method is used (Kuhl and Giardina, 1982), the third and ﬁfth mode are erro- neously indicated to represent shape features, as well as other mismatches (Figure 4H). Importantly, the smooth transition in cell shape development over time is captured by a corresponding smooth change in the LOCO-EFA Ln proﬁle for the diﬀerent time points, but by highly irregular changes in the EFA proﬁle. This illustrates that very comparable shaped cells have very diﬀerent EFA proﬁles, making EFA further unsuitable to analyse real pavement cell populations. Figure S6 presents the robust time dynamics analysis for yet another pavement cell using LOCO-EFA. Quantitative characterisation of cell shape using LOCO-EF ; https://doi.org/10.1101/157842 doi: bioRxiv preprint plementary Materials and Methods). For many shapes, the entropy measure yields very similar results to the cumulative diﬀerence. They diverge for cell outlines having a strong contribution from the lower modes. We observe that in these cases the entropy values deliver more meaningful results in regard to what one considers being more “complex”. This is due to the fact that lower modes can have a large contribution to the cumulative diﬀerence. For example, there can be a high contribution from L2 in the case that a cell is very elongated. Simply being elongated, however, does not so much represent shape complexity in the way deﬁned above. For such a simple elongated shape, the cumulative diﬀerence (i.e., the integral or area under the curve of the XOR between the original and reconstructed proﬁle) can be very similar to a cell with contributions distributed among many modes. We would consider the latter outline, however, to be much more ‘complex’. The entropy measure is able to correctly capture this form of complexity. Next, we measured the Ln proﬁles and corresponding cell shape complexity measures for a population of real cell shapes, which are much less geometric than the idealised cases treated so far. Quantitative characterisation of cell shape using LOCO-EF In general, however, when morphological protrusions increase in number or become larger in amplitude, the cumulative diﬀerence also increases since higher order modes are required to approximate the shape. XOR proﬁles typically do not change smoothly. Instead, some modes peak as they strongly contribute in capturing the main features of the shape. Hence, the marginal decrease in the XOR value when an extra mode is added, coined marginal diﬀerence, provides additional information about the shape’s dominant modes (Figure 3L, V). This measure is comparable to the Ln values, also determining something akin to “pitch” and “amplitude”. We found, however, that it bears a higher discriminatory power for more complex and irregular cell shapes. Moreover, in the case that a cell shape has signiﬁc- ant contributions from a combination of diﬀerent modes, then high marginal diﬀerence levels can be directly linked to speciﬁc cellular features (see Figure S5). Thus, the mar- ginal diﬀerence can help identifying which modes are the most relevant for speciﬁc shape aspects. Finally, one can argue that shape complexity is not solely about high numbers and large amplitudes of the protrusions, but instead reﬂects overall irregularity of the protru- sions. For example, with the previous measures a highly regular star-shaped cell with ﬁve outspoken lobes could be quantiﬁed as being as complex as a highly distorted cell with diﬀerent amplitudes and distributions of ﬁve lobes, albeit less pronounced than the star- shaped case. One might therefore prefer to deﬁne cell shape complexity as the tendency of a cell to deviate from geometrically well-deﬁned periodic outlines. A useful measure for this alternative deﬁnition of “cell shape complexity” is calculating the Shannon en- tropy E of the Ln spectrum. The entropy measure is based upon the information content contained within the distribution over the whole Ln proﬁle (see Equation 47 in the Sup- 7 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. LOCO-EFA applied to plant pavement cells To visualise the shape characteristics of populations of pavement cells, we applied the analysis to leaves of the speechless mutant (MacAlister et al., 2007), which does not generate during the leaf development any other cell types such as meristemoids or stomata (Figure 1B, Figure 5A), as well as to wild type leaf epidermis, consisting of pavement cells, 8 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint stomata and other cells from the stomatal lineage (Figure 1A, Figure 5B). stomata and other cells from the stomatal lineage (Figure 1A, Figure 5B) Using LOCO-EFA, it is straightforwardly possible to dissect the precise contribution of each mode for each cell in the population. Figure 5A, B show the spatial distribution of cells within a tissue which are predominantly 4-, 5-, 6-, or 7-lobed, by colour-coding the cells according to those speciﬁc Ln values. Very few cell shapes present only a single high Ln value. Instead the majority of shapes are composed of signiﬁcant contributions stemming from multiple modes. Consequently, simply manually counting, or applying automatic counting algorithms, for such biological cells the number of lobes would lead to incomplete information regarding their shape. It would be diﬃcult to compare mutant phenotypes. Moreover, our data shows pavement cells do not have a population-wide preferential Ln with a highest contribution to the shape (Figure 5A, B). The heterogeneity in modes that composes real populations of pavement cells suggests that their resultant cell shapes cannot be explained solely by currently proposed intracel- lular molecular mechanisms underlying lobe and indentation patterning. This is because those mechanisms, based on the existence of two counteracting pathways (one for lobe formation and another for indentation formation, see details in Xu et al., 2010) should give rise to Turing-like instabilities which are therefore expected to generate symmetrical shapes (Vanag and Epstein, 2009). LOCO-EFA applied to plant pavement cells Moreover, these models would predict that equally sized cells exhibit equal lobe numbers. However, the cell shape patterning takes place within a conﬂuent tissue, which obviously complicates how individual cells generate their shape. In the experimental setting it is very hard to distinguish between the preferred shape of each individual cell due to its intracellular patterning, and the constraints im- posed by the tissue. It is well-known that if cells prefer to be round, they will take up an hexagonal shape within a tissue context (Thompson, 1917), but it is unclear what to expect for multi-lobed shapes. Therefore, to both better understand what should be our expectation when a population of cells with complex shape preferences form a conﬂuent tissue alike the leaf, and to further validate the LOCO-EFA method on populations of cells, we performed cell-based simulations of interacting cells with prespeciﬁed cell shape preferences, and employed LOCO-EFA on the resulting in silico tissue. Applying LOCO-EFA to in silico populations and the eﬀect of interactions between preferred cell shapes Applying LOCO-EFA to in silico populations and the eﬀect of interactions between preferred cell shapes We create in silico cells using the Cellular Potts Model (CPM), an energy-based frame- work that allows us to represent cells and their dynamics through small membrane exten- sions and retractions (see Methods section). In its basic form, CPM cell shapes emerge due to the interaction between interfacial tension (the strength of which can depend on the cell types involved), internal cellular pressure and cortical tension (Magno et al., 2015). Here we used an extension of the CPM which predeﬁnes intrinsic forces causing elongation and lobeness, resulting in more complex cell shapes. This extension consists in applying additional, cell-speciﬁc forces each time a cell extension or retraction is con- sidered, giving rise to elongated and/or multilobed preferred cell shapes (Equation 3). Three additional forces are used that capture (i) an intrinsic tendency to elongate; (ii) a tendency to form a speciﬁed number of lobes; and (iii) an additional force for the cell to round up (Figure 5E–G, van Rooij et al. (2017a) and Movie S5). The latter term robustly prevents cells from falling apart, especially within a conﬂuent tissue with many conﬂicting preferred cell shapes. A population of cells, all with a same preferred shape, were then allowed to interact with each other within a tissue context. In this way we can compare 9 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint the shape of a single cell in isolation with the shape cells attain within a tissue. p g p We here present the analysis on two distinct speciﬁed shapes (Figure 5C, D; see Table S1 for the speciﬁc parameters used). Both preferred shapes have six lobes, the cells shown in Figure 5D moreover tend to be elongated. Above the panels we plot the shape that the cells attain in isolation. Applying LOCO-EFA to in silico populations and the eﬀect of interactions between preferred cell shapes Although the same cell shape is speciﬁed for all cells within the population, cells acquire very diﬀerent shapes due to their local interactions. We quantiﬁed this diversity using LOCO-EFA. This was done by colour-coding the contri- bution of L4–L7 (as indicated for each panel), for both the isolated cells and for the resultant shapes of all the cells within the simulated tissue. As expected, the isolated cells present a very high contribution by L6, with marginal contributions from the other modes. In contrast, due to the cellular interactions within the tissue, other modes can also become prominent, highly varying from cell to cell, even though each cell has exactly the same speciﬁed shape. Thus, although a cell in isolation would be able to generate regular protrusions with speciﬁc amplitudes, the periodical lobe formation becomes in- hibited in a packed tissue environment, with symmetry and shape distortions being a direct consequence of tissue packing (Figure 5C, D). Such a tendency was observed for all simulations performed, irrespective of the speciﬁed cell shape, i.e., the number of lobes, their amplitude, and the level of overall cell elongation, as well as over a wide range of parameter values for the cell-cell interactions (Figure S7, Figure S8). Radially, well-spaced periodically lobed cell shapes are not likely space ﬁlling, hence resolving the competition between preferred shape and conﬂuency seems to be a universal driving force to complex cell shapes. Discussion The progress in microscopy and imaging techniques within the biological sciences gen- erates a need for adequate analytic tools to capture relevant information eﬃciently and objectively (Zhong et al., 2012). Image acquisition through high-throughput microscopy approaches generates amounts of data that are beyond the human ability (or patience) to be analysed manually, demanding computational automatic tools. We have developed a new analytic tool which takes as the input the contour of a 2D cell projection, and ex- tracts from it, in an eﬃcient and parameter-independent manner, quantitative meaningful cell shape information. Importantly, the pipeline can be integrated within other recent advances in segmentation techniques (Fernandez et al., 2010; van Rooij et al., 2017b), to fully automate shape analysis of series of images. Our method can intuitively be understood using the analogy of music perception. To quantify an instrument playing a certain note, say a violin playing the note A, one would ﬁrstly wish to have a device that can capture which note is played. We have indeed shown that LOCO-EFA, unlike EFA, is able to correctly determine the analogous feature for shapes, which is the number of protrusions. A recent method developed by Wu et al. (2016) can also be employed to count the number of protrusions. When the biological question asked requires not only the “pitch” to be measured, but also the “volume” and “timbre”, corresponding to lobe amplitude and other irregularities, such a method will be insuﬃcient. Indeed, LOCO-EFA provides a broader spectrum of the shape properties, a holistic set of measurements that allows complex morphologies to be quantiﬁed in a reproducible manner. We illustrated how the measurements obtained with LOCO-EFA can be interpreted 10 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. Discussion ; https://doi.org/10.1101/157842 doi: bioRxiv preprint ﬁrst using simple shapes (geometrical or symmetrical forms), then extending the analysis to highly asymmetric shapes, for which the pavement cells of Arabidopsis are a good paradigm. To assess the performance of our technique, we extended our study to confocal microscopy images of populations of pavement cells. Such a shape analysis is biologically relevant, because many of the players accounting for the lobe and indentation patterning are known (Xu et al., 2010; Grieneisen et al., 2017), enabling one to extend the study of cell shape control to mutants and experimental interferences in the future. We found that very few cells have a symmetrical shape, i.e., they can not be represented well by a single high Ln value. Such composition in real cell shapes of several Ln values is not likely to be explained only by the existence of two counteracting pathways specifying lobe and indentation identities. Our in silico approach instead suggests that the dynamics of space-ﬁlling complex shapes can dramatically increase the overall irregularity: even when the CPM cells are speciﬁcally programmed to develop well-deﬁned characteristic shapes, the interactions between them trigger the formation of highly deviating cell shape characteristics. As a result of being within the tissue, the main, speciﬁed mode decreases in strength while other modes become relevant. Altogether it leads to very asymmetrical resultant shapes, even though very symmetrical shapes were speciﬁed. Although our synthetic data is but a phenomenological reconstruction of real shapes, our results suggest that the local inﬂuence of neighbours during pavement cell develop- ment should be important for the acquisition of their ﬁnal shape. To assess this hypothesis in further studies it will be crucial to perform quantitative shape analysis on in vivo cell populations over time, combined with growth tensor analysis (i.e., determination of the spatial pattern in growth rate and its anisotropy). Such studies, combined with genetic or physical perturbations in cell growth and deformation, and with in silico cell growth models, should allow untangling of the relationship between the cell shape as speciﬁed at the cellular level and the resultant shape arising from the complex interactions between the cells at the tissue level. When LOCO-EFA was applied to cell tracking data, we observed that the proﬁles of those dynamically changing cells varied smoothly over time. Such trajectories are distinct from cell-to-cell and provide unique ﬁngerprints of each individual developing cell. Discussion This opens the possibility of using the Ln modes as cell identiﬁers within a temporal sequence of images, to help track populations of cells automatically. Although we have here applied LOCO-EFA to objectively measure cell shape diﬀer- ences in pavement cells, providing us a powerful tool to evaluate pavement cell shape dynamics and compare plant phenotypes, it can also be applied to diﬀerent cell types with complex cell shapes of any species, as well as to any other layouts, such as entire leaves, within or outside the realm of biology. Recently, Fourier analysis was applied to quantify myosin polarity (asymmetrical distribution of molecules) in animal epithelia (Tetley et al., 2016). In a similar way, the LOCO-EFA method could be applied to quantify the cellular distribution of molecules within complex cell shapes. Finally, our method could be integrated within recent image analysis pipelines, allowing one to ex- tract and analyse cell shape information in a high throughput manner (van Rooij et al., 2017b; Heller et al., 2016; Stegmaier et al., 2016). 11 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint XOR and other measurements All the grid points belonging to each individual real or synthetic pavement cell were compared to all the grid points captured by the subsequent series of LOCO-EFA recon- structions. A reconstruction of level N takes into account the ﬁrst N Ln modes. The in silico cells were generated using the Cellular Potts Model, which is a grid-based formal- ism, while for the experimental data the grid points were directly deﬁned by the imaging resolution. The scripts used to calculate the XOR and to colour-code the real and syn- thetic cells were written in the coding language C. Shape approximations, cumulative diﬀerence and entropy were calculated using the ﬁrst 50 Ln modes. To capture cell shape complexity linked to protrusions rather than mere anisotropy, cumulative diﬀerence is calculated from the second Ln mode onwards. This value turned out to be more than suﬃcient to capture any cell shape given the grid point resolution used for all cases here. Note, however, that very high resolution images might require additional modes to fully capture the shape. Confocal images and image processing Columbia wild type or speechless mutant (MacAlister et al., 2007) leaves expressing pmCherry-Aquaporin (Nelson et al., 2007) were imaged using a confocal microscope Leica SP5 at comparable stages and in comparable regions. Further image processing was done using ImageJ and images were segmented using in-house software (Segmentation Potts Model (van Rooij et al., 2017b)). Cells changing over time were imaged using a custom- made perfusion chamber (Robinson et al., 2011; Sauret-G¨ueto et al., 2012; Kuchen et al., 2012). Shape descriptors Average lobe lengths and neck widths were calculated using ImageJ (Analyse →Measure). The skeleton was calculated using Better Skeletonization by Nicholas Howe, available through MATLAB File Exchange. Geometric shapes All geometric shapes were generated by the “superformula” described in Gielis (2003), and were analysed in the same manner as the confocal images. Cellular Potts Model generating complex cell shapes The Cellular Potts Model (CPM) is an energy-based model formalism used to model cellular dynamics in terms of Cell Surface Mechanics (Magno et al., 2015). Individual cells are described by a set of grid points on a lattice. In this manuscript we used the CPM to generate in silico cells with relatively complex shape preferences that are allowed to interact within a conﬂuent setting. During each simulation step a grid point is chosen in a random fashion to evaluate whether its state changes into one of its neighbouring states, eﬀectively corresponding to a small cell shape modiﬁcation at that point. To evaluate if such state change will occur, the energy change is calculated that such a copy would 12 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint cause. This is done by calculating the change in the conﬁgurational energy as deﬁned by the following Hamiltonian, which sums up the energy contribution of each pixel within the entire ﬁeld as well as of all cells: cause. This is done by calculating the change in the conﬁgurational energy as deﬁned by the following Hamiltonian, which sums up the energy contribution of each pixel within the entire ﬁeld as well as of all cells: (1) H = X ij X i′j′ J 1 −δci,j,ci′,j′ + X c λa(ac −A)2 + λp X c (pc −P)2 , (1) J refers to the coupling energy, summed over all grid points (i, j) and their eight (2nd order) neighbours (i′, j′). Cellular Potts Model generating complex cell shapes The Kronecker delta term 1 −δci,j,ci′,j′ simply assures that neighbouring lattice sites of the same state (i.e., belonging to the same cell) do not contribute to the total energy of the system. The variables ac and pc denote respectively the actual cell area and the actual cell perimeter for each cell (c); the parameters A and P denote the target cell area and perimeter. The parameters λa and λp describe the resistance to deviate from the target area and perimeter, respectively. The probability a copying event is accepted depends on the change in the Hamiltonian, △H = Hafter − Hbefore, in the following way: (2) P = ( 1 if ∆H < −Y , e(−∆H+Y T ) if ∆H ≥−Y , (2) where Y corresponds to the yield or ability of a membrane to resist a force and T (simula- tion temperature) captures additional stochastic ﬂuctuations. Copying events which de- crease H by at least Y will always be accepted, otherwise acceptance follows a Boltzmann probability distribution (Equation 2). To generate cells with a particular number of preferred protrusions, we modify the change in the Hamiltonian as is calculated for every evaluated copying event, eﬀectively shortcutting intracellular biochemistry and biophysics, in the following way. Simulated cells are attributed with a speciﬁed preferred number of lobes, amplitude of lobes, overall elongation and roundness, implemented by modifying the change in the Hamiltonian for every evaluated copy event as follows (van Rooij et al., 2017a): (3) △H′ = △H −ν cos(nθ) −χ cos(2α) −µ( p A/π −r) . (3) Those three additional terms are evaluated for both cells involved in the copying event, so there are eﬀectively six additional terms. The ﬁrst term captures the tendency to form n lobes, with ν capturing the propensity to extend to form a lobe or to retract to form an indentation, thus giving rise to the amplitude or pointedness of the lobes. θ describes the angle between any of the n equally spread out target directions for outgrowth and the vector determined by the coordinates of the grid point under evaluation and the centre-of-mass of the cell (hereafter called the copy vector) (Figure 5E). Cellular Potts Model generating complex cell shapes To clarify, when a cell extension is considered right on top of one of the target directions, then nθ = 0, cos(nθ) = 1, and tendency to extend is maximally increased, while halfway two target directions, nθ = π, cos(nθ) = −1, and the tendency to extend is maximally suppressed. The second term in Equation 3 captures an overall elongation, implemented in a similar fashion. The parameter χ corresponds to the propensity to elongate and α is the angle between the elongation vector and the copy vector. If only these two terms are used, cells within tissue simulations can easily lose co- herence, i.e., fall apart. Therefore a third term was added, capturing a propensity to roundness. The parameter µ captures the resistance of a cell to deviate from a circle, 13 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint with r being the length of the copy vector, and p A/π being the preferred radius of cell, given its target area. For further details, see van Rooij et al. (2017a). with r being the length of the copy vector, and p A/π being the preferred radius of cell, given its target area. For further details, see van Rooij et al. (2017a). Importantly, the target lobe and elongation vectors are not ﬁxed during the simulation. At intervals of 100 simulation time steps they are dynamically updated, in order to attain the most favourable position, eﬀectively “accommodating” its lobe positions with respect to its neighbours. During a vector update step, the preferred directions of extension are matched to the set of directions the current shape of cells presents the strongest level of extension. The initial cell positions within the ﬁeld were randomly chosen. Simulations were run for 10000 time steps (see an example in Movie S5). Parameters used for each used speciﬁed cell shape are given in Table S1. Acknowledgements We thank Enrico Coen for stimulating discussions, and John Fozard for critical reading and helpful comments. This work has been supported by Consejo Nacional de Ciencia y Tecnolog´ıa (CONACYT) and by the UK Biological and Biotechnology Research Council (BBSRC) via grant BB/J004553/1 to the John Innes Centre. VAG acknowledges support from the Royal Society Dorothy Hodgkin fellowship. JAvR acknowledges support from the Netherlands Consortium for Systems Biology (NCSB), which is part of the Nether- lands Genomics Initiative/Netherlands Organization for Scientiﬁc Research (NGI/NWO). References Thompson DW (1917) On growth and form. Cambridge: Cambridge University Press. Charras G, Paluch E (2008) Blebs lead the way: how to migrate without lamellipodia. Nat Rev Mol Cell Biol 9:730–736. doi:10.1038/nrm2453. Marinari E, Mehonic A, Curran S, Gale J, Duke T, et al. (2012) Live-cell delamination counterbalances epithelial growth to limit tissue overcrowding. Nature 484:542–545. doi:10.1038/nature10984. Veeman MT, Smith WC (2013) Whole-organ cell shape analysis reveals the de- velopmental basis of ascidian notochord taper. Dev Biol 373:281–289. doi: 10.1016/j.ydbio.2012.11.009. Keren K, Pincus Z, Allen GM, Barnhart EL, Marriott G, et al. (2008) Mechanism of shape determination in motile cells. Nature 453:475–480. doi:10.1038/nature06952. Driscoll MK, McCann C, Kopace R, Homan T, Fourkas JT, et al. (2012) Cell shape dynamics: from waves to migration. PLoS Comput Biol 8:e1002392. doi: 10.1371/journal.pcbi.1002392. Gomez JM, Chumakova L, Bulgakova NA, Brown NH (2016) Microtubule organiza- tion is determined by the shape of epithelial cells. Nat Commun 7:13172. doi: 10.1038/ncomms13172. Sampathkumar A, Krupinski P, Wightman R, Milani P, Berquand A, et al. (2014) Sub- cellular and supracellular mechanical stress prescribes cytoskeleton behavior in Ara- bidopsis cotyledon pavement cells. Elife 3:e01967. doi:10.7554/eLife.01967. 14 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint Minc N, Burgess D, Chang F (2011) Inﬂuence of cell geometry on division-plane posi- tioning. Cell 144:414–426. doi:10.1016/j.cell.2011.01.016. Noda K, Glover BJ, Linstead P, Martin C (1994) Flower colour intensity depends on specialized cell shape controlled by a Myb-related transcription factor. Nature 369:661– 664. doi:10.1038/369661a0. References Zhong Q, Busetto AG, Fededa JP, Buhmann JM, Gerlich DW (2012) Unsupervised mod- eling of cell morphology dynamics for time-lapse microscopy. Nat Methods 9:711–713. doi:10.1038/nmeth.2046. Rajaram S, Pavie B, Wu LF, Altschuler SJ (2012) PhenoRipper: software for rapidly proﬁling microscopy images. Nat Methods 9:635–637. doi:10.1038/nmeth.2097. Ljosa V, Sokolnicki KL, Carpenter AE (2012) Annotated high-throughput microscopy image sets for validation. Nat Methods 9:637. doi:10.1038/nmeth.2083. Ivakov A, Persson S (2013) Plant cell shape: modulators and measurements. Front Plant Sci 4:439. doi:10.3389/fpls.2013.00439. Klingenberg CP (2010) Evolution and development of shape: integrating quantitative approaches. Nat Rev Genet 11:623–635. doi:10.1038/nrg2829. Fu Y, Gu Y, Zheng Z, Wasteneys G, Yang Z (2005) Arabidopsis interdigitating cell growth requires two antagonistic pathways with opposing action on cell morphogenesis. Cell 120:687–700. doi:10.1016/j.cell.2004.12.026. Le J, Mallery EL, Zhang C, Brankle S, Szymanski DB (2006) Arabidopsis BRICK1/HSPC300 is an essential WAVE-complex subunit that selectively stabilizes the Arp2/3 activator SCAR2. Curr Biol 16:895–901. doi:10.1016/j.cub.2006.03.061. Wu TC, Belteton SA, Pack J, Szymanski DB, Umulis DM (2016) Lobeﬁnder: A Convex Hull-Based Method for Quantitative Boundary Analyses of Lobed Plant Cells. Plant Physiol 171:2331–2342. doi:10.1104/pp.15.00972. Lin D, Cao L, Zhou Z, Zhu L, Ehrhardt D, et al. (2013) Rho GTPase signaling activ- ates microtubule severing to promote microtubule ordering in Arabidopsis. Curr Biol 23:290–297. doi:10.1016/j.cub.2013.01.022. Pincus Z, Theriot JA (2007) Comparison of quantitative methods for cell-shape analysis. J Microsc 227:140–156. doi:10.1111/j.1365-2818.2007.01799.x. Kuhl FP, Giardina CR (1982) Elliptic Fourier features of a closed contour. Comput Gr Image Process 18:236–258. Diaz G, Zuccarelli A, Pelligra I, Ghiani A (1989) Elliptic Fourier analysis of cell and nuclear shapes. Comput Biomed Res 22:405–414. Schmittbuhl M, Allenbach B, Le Minor JM, Schaaf A (2003) Elliptical descriptors: some simpliﬁed morphometric parameters for the quantiﬁcation of complex outlines. Math Geol 35:853–871. 15 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. References ; https://doi.org/10.1101/157842 doi: bioRxiv preprint Graner F, Glazier JA (1992) Simulation of biological cell sorting using a two-dimensional extended Potts model. Phys Rev Lett 69:2013–2016. Glazier JA, Graner F (1993) Simulation of the diﬀerential adhesion driven rearrangement of biological cells. Phys Rev E 47:2128–2154. Haines AJ, Crampton JS (2000) Improvements to the method of Fourier shape analysis as applied in morphometric studies. Palaeontology 43:765–783. Diaz G, Quacci D, Dell’Orbo C (1990) Recognition of cell surface modulation by elliptic Fourier analysis. Comput Methods Programs Biomed 31:57–62. MacAlister CA, Ohashi-Ito K, Bergmann DC (2007) Transcription factor control of asym- metric cell divisions that establish the stomatal lineage. Nature 445:537–540. doi: 10.1038/nature05491. Xu T, Wen M, Nagawa S, Fu Y, Chen JG, et al. (2010) Cell surface- and rho GTPase- based auxin signaling controls cellular interdigitation in Arabidopsis. Cell 143:99–110. doi:10.1016/j.cell.2010.09.003. Vanag VK, Epstein IR (2009) Pattern formation mechanisms in reaction-diﬀusion sys- tems. Int J Dev Biol 53:673–681. doi:10.1387/ijdb.072484vv. Magno R, Grieneisen VA, Mar´ee AFM (2015) The biophysical nature of cells: poten- tial cell behaviours revealed by analytical and computational studies of cell surface mechanics. BMC Biophys 8:8. doi:10.1186/s13628-015-0022-x. van Rooij JA, Magno R, Mar´ee AFM (2017a) Capturing complex cell shapes in the Cellular Potts Model framework. manuscript in preparation. Fernandez R, Das P, Mirabet V, Moscardi E, Traas J, et al. (2010) Imaging plant growth in 4D: robust tissue reconstruction and lineaging at cell resolution. Nat Methods 7:547– 553. doi:10.1038/nmeth.1472. van Rooij JA, Fozard JA, Carter R, Hartley M, Olsson TSG, et al. (2017b) The Segment- ation Potts Model: extracting quantitative cell information from 2d and 3d confocal microscopy images. manuscript in preparation. Grieneisen VA, van Berkel K, Xu T, Scheres B, Hogeweg P, et al. (2017) Intracellu- lar ROP partioning and auxin-mediated indirect cell-cell communication can explain interdigitation of pavement cells. under revision. Tetley RJ, Blanchard GB, Fletcher AG, Adams RJ, Sanson B (2016) Unipolar distribu- tions of junctional Myosin II identify cell stripe boundaries that drive cell intercalation throughout Drosophila axis extension. Elife 5:2331–2342. doi:10.7554/eLife.12094. Heller D, Hoppe A, Restrepo S, Gatti L, Tournier AL, et al. (2016) Epitools: An Open- Source Image Analysis Toolkit for Quantifying Epithelial Growth Dynamics. Dev Cell 36:103–116. doi:10.1016/j.devcel.2015.12.012. Stegmaier J, Amat F, Lemon WC, McDole K, Wan Y, et al. (2016) Real-Time Three- Dimensional Cell Segmentation in Large-Scale Microscopy Data of Developing Em- bryos. Dev Cell 36:225–240. doi:10.1016/j.devcel.2015.12.028. 16 . Figure 1: Complex cell shapes and the shortcomings of traditional shape quan- tiﬁers. (A–C) Complex cell shapes in both plant (A, B) and animal (C) tissues. (A, B) Pavement cells of Arabidopsis thaliana wild type (A) and speechless mutant (B), char- acterised by jigsaw-like shapes consisting of multiple alternating protrusions (lobes) and indentations. (C) Amnioserosa cells in the Drosophila embryo also present cell shapes with similar complexity. Scale bars 50 µm. (D–G) Individual cells from the imaged tissues (upper panel), and the corresponding segmented cell outlines (lower panel). Scale bars 10 µm. (H) Traditional metrics to quantify cell shape can lead to similar values for very diﬀerent shapes and are sensitive to parameter choices and imaging conditions. Here we compare the cells D–G. See also Figure S1. References CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint Nelson BK, Cai X, Nebenf¨uhr A (2007) A multicolored set of in vivo organelle markers for co-localization studies in Arabidopsis and other plants. Plant J 51:1126–1136. doi: 10.1111/j.1365-313X.2007.03212.x. Robinson S, Barbier de Reuille P, Chan J, Bergmann D, Prusinkiewicz P, et al. (2011) Generation of spatial patterns through cell polarity switching. Science 333:1436–1440. doi:10.1126/science.1202185. Sauret-G¨ueto S, Calder G, Harberd NP (2012) Transient gibberellin application promotes Arabidopsis thaliana hypocotyl cell elongation without maintaining transverse orienta- tion of microtubules on the outer tangential wall of epidermal cells. Plant J 69:628–639. doi:10.1111/j.1365-313X.2011.04817.x. Kuchen EE, Fox S, Barbier de Reuille P, Kennaway R, Bensmihen S, et al. (2012) Gen- eration of leaf shape through early patterns of growth and tissue polarity. Science 335:1092–1096. doi:10.1126/science.1214678. Gielis J (2003) A generic geometric transformation that uniﬁes a wide range of natural and abstract shapes. Am J Bot 90:333–338. doi:10.3732/ajb.90.3.333. Bai Y, Falk S, Schnittger A, Jakoby MJ, H¨ulskamp M (2010) Tissue layer speciﬁc regula- tion of leaf length and width in Arabidopsis as revealed by the cell autonomous action of ANGUSTIFOLIA. Plant J 61:191–199. doi:10.1111/j.1365-313X.2009.04050.x. Russ JC (2011) The Image Processing Handbook. Boca Raton: CRC Press, 6 edition. Andriankaja M, Dhondt S, De Bodt S, Vanhaeren H, Coppens F, et al. (2012) Exit from proliferation during leaf development in Arabidopsis thaliana: a not-so-gradual process. Dev Cell 22:64–78. doi:10.1016/j.devcel.2011.11.011. Sorek N, Gutman O, Bar E, Abu-Abied M, Feng X, et al. (2011) Diﬀerential eﬀects of prenylation and s-acylation on type I and II ROPs membrane interaction and function. Plant Physiol 155:706–720. doi:10.1104/pp.110.166850. 17 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. References ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint Figures Figures D E F G A B C H D E F G A B C H Descriptor D E F G Average length of lobes 6.1 µm 7.4 µm 6.8 µm 4.7 µm Average width of necks 7.5 µm 9.5 µm 14.3 µm 15.7 µm Form factor 4.2 4.2 1.9 1.4 Aspect ratio 0.80 0.60 0.53 0.52 Skeleton end points 4–5 6–8 5 4–7 Average polarity score 4.1–4.6 5.1–6.1 3.45 2.7–4.2 Complex cell shapes and the shortcomings of traditional shap A–C) Complex cell shapes in both plant (A, B) and animal (C) tissues cells of Arabidopsis thaliana wild type (A) and speechless mutant (B by jigsaw-like shapes consisting of multiple alternating protrusions (lob ons. (C) Amnioserosa cells in the Drosophila embryo also present cel ar complexity. Scale bars 50 µm. (D–G) Individual cells from the image anel), and the corresponding segmented cell outlines (lower panel). Sc H) Traditional metrics to quantify cell shape can lead to similar values shapes and are sensitive to parameter choices and imaging conditions. the cells D–G See also Figure S1 A B G C B B C A D E F G H G F D E Descriptor D E F G Average length of lobes 6.1 µm 7.4 µm 6.8 µm 4.7 µm Average width of necks 7.5 µm 9.5 µm 14.3 µm 15.7 µm Form factor 4.2 4.2 1.9 1.4 Aspect ratio 0.80 0.60 0.53 0.52 Skeleton end points 4–5 6–8 5 4–7 Average polarity score 4.1–4.6 5.1–6.1 3.45 2.7–4.2 Figure 1: Complex cell shapes and the shortcomings of traditional shape quan- tiﬁers. (A–C) Complex cell shapes in both plant (A, B) and animal (C) tissues. (A, B) Pavement cells of Arabidopsis thaliana wild type (A) and speechless mutant (B), char- acterised by jigsaw-like shapes consisting of multiple alternating protrusions (lobes) and indentations. (C) Amnioserosa cells in the Drosophila embryo also present cell shapes with similar complexity. Scale bars 50 µm. (D–G) Individual cells from the imaged tissues (upper panel), and the corresponding segmented cell outlines (lower panel). Scale bars 10 µm. (H) Traditional metrics to quantify cell shape can lead to similar values for very diﬀerent shapes and are sensitive to parameter choices and imaging conditions. Here we compare the cells D–G. See also Figure S1. 18 . Figures CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint n-1 n-1 λ−n−1 λ+n+1 Ln λ2n λ1n A B C C 1 2 3 4 6 7 6 7 5 5 1 2 4 3 Reconstruction using LOCO-EFA Reconstruction using EFA Original E λ+n+1 λ−n−1 Ln Original λ+n λ−n D Figure 2: To get LOCO. (A) Each EFA elliptic harmonic is decomposed into two counter-rotating circles. (B) Mode Ln is composed of the counter-clockwise rotating n+1th EFA harmonic circle and the clockwise rotating n−1th circle. (C) The combined amplitude contribution to Ln (yellow line) of the two counter-rotating circles with radii λ+n+1 and λ−n−1 is also dependent on the oﬀset in their starting points and the oﬀset of the overall (mode 1) starting point, which together determine the initial phase shift (green dots) in the amplitude contribution of each rotor. (D, E) Comparison of closed contour reconstruction through either EFA (D) or LOCO-EFA (E). Although both ap- proximations eventually converge to the original 6-lobed star shape (labelled ‘Original’), the reconstruction using EFA harmonics (D) generates a spurious shape after addition of the 5th harmonic and only recovers the original shape after addition of the 7th harmonic, n-1 n-1 λ−n−1 λ+n+1 Ln λ2n λ1n A B C λ+n+1 λ−n−1 Ln λ+n λ−n n-1 n-1 λ−n−1 λ+n+1 Ln B C λ+n+1 λ−n−1 Ln λ2n λ1n A λ+n λ−n B C A n-1 λ−n−1 Ln C 1 2 3 4 6 7 6 7 5 5 1 2 4 3 Reconstruction using LOCO-EFA Reconstruction using EFA Original E Ln Original λ+n λ−n D Figure 2: To get LOCO. (A) Each EFA elliptic harmonic is decomposed into two counter-rotating circles. (B) Mode Ln is composed of the counter-clockwise rotating n+1th EFA harmonic circle and the clockwise rotating n−1th circle. (C) The combined amplitude contribution to Ln (yellow line) of the two counter-rotating circles with radii λ+n+1 and λ−n−1 is also dependent on the oﬀset in their starting points and the oﬀset of the overall (mode 1) starting point, which together determine the initial phase shift (green dots) in the amplitude contribution of each rotor. Figures Ln mode 1 A B C D E F G H I 2 3 4 5 6 7 8 9 101112131415 0 0.05 0.1 0.15 0.2 1 2 3 4 5 6 7 8 9 101112131415 0 0.1 0.2 0.3 0.4 0.5 1 2 3 4 5 6 7 8 9101112131415 0 0.1 0.2 0.3 0.4 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1 2 3 4 5 6 7 0 0.5 1 A C D E G H I J K L M Ln mode Ln mode Ln mode Ln cd entropy B F XOR difference Marginal difference H I M B G L J K M Ln mode 1 112131415 1 2 3 4 5 6 7 8 9 101112131415 0 0.1 0.2 0.3 0.4 0.5 1 2 3 4 5 6 7 8 9101112131415 0 0.1 0.2 0.3 0.4 5 6 7 K L Ln mode Ln mode XOR difference Marginal difference 0 0.2 0.4 0.6 0.8 1 1.2 1.4 cd entropy N O P Ln mode Q R 1 N O Q S S P R 1 2 3 4 5 6 7 8 9 101112131415 0 0.5 1 1.5 U Ln mode XOR difference 1 2 3 4 5 6 7 8 9 101112131415 0 0.5 1 1.5 1 2 3 4 5 6 7 8 9 101112131415 0 0.1 0.2 0.3 0.4 0.5 V U Ln mode Ln mode Marginal difference 2 3 4 5 6 7 8 9 101112131415 0 0.05 0.1 0.15 0.2 0.25 1 2 3 4 5 6 7 0 0.5 1 O P Q R S N T Ln mode Ln 0 1 2 3 4 5 W cd entropy T U T V W 0 1 XOR difference Ln mode 3: Interpreting LOCO-EFA-derived measures for geometrical and asym- Figure 3: Interpreting LOCO-EFA-derived measures for geometrical and asym- metric shapes. (A–I) Symmetrical and well-deﬁned geometrical shapes with normalised area. (J–L) Ln, XOR, and marginal diﬀerence proﬁles for shapes (A–I). (J, L) For each geometric shape, a clear peak appears in the proﬁles, this main contributor to the shape always coinciding with the number of protrusions. (M) Cumulative diﬀerence (cd) and entropy for shapes (A–I). (N–Q) Symmetrical shapes with increasing protrusion amp- litude. (R, S) Asymmetrical shapes. (T–V) Ln, XOR, and marginal diﬀerence proﬁles for shapes (N–S). Increasing protrusion amplitude leads to increasing peak levels in the proﬁles. Figures It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint 2 3 4 5 6 7 8 9 101112131415 0 0.05 0.1 0.15 0.2 0.25 1 2 3 4 5 6 7 8 9 101112131415 0 0.5 1 1.5 1 2 3 4 5 6 7 8 9 101112131415 0 0.1 0.2 0.3 0.4 0.5 0 1 2 3 4 5 1 2 3 4 5 6 7 0 0.5 1 O P Q R S N N O P Q R S Ln mode 1 A B C D E F G H I 2 3 4 5 6 7 8 9 101112131415 0 0.05 0.1 0.15 0.2 1 2 3 4 5 6 7 8 9 101112131415 0 0.1 0.2 0.3 0.4 0.5 1 2 3 4 5 6 7 8 9101112131415 0 0.1 0.2 0.3 0.4 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1 2 3 4 5 6 7 0 0.5 1 A C D E G H I J K L M Ln mode Ln mode Ln mode Ln cd entropy B F XOR difference Marginal difference W V U T Ln mode Ln mode Ln mode cd entropy Ln XOR difference Marginal difference Figure 3: Interpreting LOCO-EFA-derived measures for geometrical and asym- metric shapes. (A–I) Symmetrical and well-deﬁned geometrical shapes with normalised area. (J–L) Ln, XOR, and marginal diﬀerence proﬁles for shapes (A–I). (J, L) For each geometric shape, a clear peak appears in the proﬁles, this main contributor to the shape always coinciding with the number of protrusions. (M) Cumulative diﬀerence (cd) and entropy for shapes (A–I). (N–Q) Symmetrical shapes with increasing protrusion amp- litude. (R, S) Asymmetrical shapes. (T–V) Ln, XOR, and marginal diﬀerence proﬁles for shapes (N–S). Increasing protrusion amplitude leads to increasing peak levels in the proﬁles. Asymmetric shapes present multiple peaks, indicating that multiple modes are needed to recapitulate the original shape. (W) Cumulative diﬀerence (cd) and entropy for shapes (N–S). Figures (D, E) Comparison of closed contour reconstruction through either EFA (D) or LOCO-EFA (E). Although both ap- proximations eventually converge to the original 6-lobed star shape (labelled ‘Original’), the reconstruction using EFA harmonics (D) generates a spurious shape after addition of the 5th harmonic and only recovers the original shape after addition of the 7th harmonic, whereas the reconstruction using LOCO-EFA (E) reconstitutes the original shape only at the 6th mode, matching the number of protrusions. The number of modes used for each i l i i i di d b l h h Reconstruction using EFA 1 2 3 4 Reconstruction using LOCO-EFA Reconstruction using EFA E D D Reconstruction using LOCO-EFA E 2 C 1 2 E 4 3 2 5 4 3 6 2 Figure 2: To get LOCO. (A) Each EFA elliptic harmonic is decomposed into two counter-rotating circles. (B) Mode Ln is composed of the counter-clockwise rotating n+1th EFA harmonic circle and the clockwise rotating n−1th circle. (C) The combined amplitude contribution to Ln (yellow line) of the two counter-rotating circles with radii λ+n+1 and λ−n−1 is also dependent on the oﬀset in their starting points and the oﬀset of the overall (mode 1) starting point, which together determine the initial phase shift (green dots) in the amplitude contribution of each rotor. (D, E) Comparison of closed contour reconstruction through either EFA (D) or LOCO-EFA (E). Although both ap- proximations eventually converge to the original 6-lobed star shape (labelled ‘Original’), the reconstruction using EFA harmonics (D) generates a spurious shape after addition of the 5th harmonic and only recovers the original shape after addition of the 7th harmonic, whereas the reconstruction using LOCO-EFA (E) reconstitutes the original shape only at the 6th mode, matching the number of protrusions. The number of modes used for each sequential reconstruction is indicated below each shape. 19 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. Figures Asymmetric shapes present multiple peaks, indicating that multiple modes are needed to recapitulate the original shape. (W) Cumulative diﬀerence (cd) and entropy for shapes (N–S). 20 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint 2 3 4 5 6 7 8 9 10 11 12 13 14 15 1 2 3 4 5 6 7 0 0.05 0.1 0.15 2 3 4 5 6 7 8 9 10 11 12 13 14 15 1 2 3 4 5 6 7 0 0.05 0.1 0.15 2 3 4 5 6 7 8 9 10 11 12 13 14 15 1 2 3 4 5 6 7 0 0.05 0.1 0.15 2 3 4 5 6 7 8 9 10 11 12 13 14 15 1 2 3 4 5 6 7 0 2 4 LOCO-EFA Ln Marginal difference Ln mode EFA A B C D E F G A B C D E F G Marginal difference Pn A B C D E F G H 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 time I EFA mode time time time Ln mode EFA mode Figure 4: LOCO-EFA metrics on a cell changing its shape over time. (A–G) Sequence of a tracked pavement cell growing over time with normalised area. (H) Pn and marginal diﬀerence proﬁles using EFA. Figures Applying EFA modes to approximate the cell shapes leads to erratic proﬁles that fail to recover the biological sequence of development, as can be observed in the Pn proﬁle and as spurious peaks at the third and ﬁfth har- monics in the marginal diﬀerence proﬁle. (I) Ln and marginal diﬀerence proﬁles using LOCO-EFA. The LOCO-EFA measurements recover the smooth transitions during the cell morphogenesis. The overall square symmetry of the cell is captured by a peak at L4, the formation of lobes by a smooth increase in L9, and later L13. A B C D E F G 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 A G B H 2 3 4 5 6 7 8 9 10 11 12 13 14 15 1 2 3 4 5 6 7 0 2 4 EFA A B C D E F G Pn H time EFA mode 2 3 4 5 6 7 8 9 10 11 12 13 14 15 1 2 3 4 5 6 7 0 0.05 0.1 0.15 Marginal difference time 0 0 Marginal difference 2 3 4 5 6 7 8 9 10 11 12 13 14 15 1 2 3 4 5 6 7 0 0.05 0.1 0.15 10 11 12 13 14 15 1 Marginal difference EFA mode time Ln mode 11 12 EFA mode 12 EFA mode 1 EFA mode 2 3 4 5 6 7 8 9 10 11 12 13 14 15 1 2 3 4 5 6 7 0 0.05 0.1 0.15 LOCO-EFA Ln Ln mode G A B C D E F G I time EFA mode Marginal difference 6 7 8 9 10 11 1 Ln mode 11 Ln mode Figure 4: LOCO-EFA metrics on a cell changing its shape over time. (A–G) Sequence of a tracked pavement cell growing over time with normalised area. (H) Pn and marginal diﬀerence proﬁles using EFA. Applying EFA modes to approximate the cell shapes leads to erratic proﬁles that fail to recover the biological sequence of development, as can be observed in the Pn proﬁle and as spurious peaks at the third and ﬁfth har- monics in the marginal diﬀerence proﬁle. Figures (I) Ln and marginal diﬀerence proﬁles using LOCO-EFA. The LOCO-EFA measurements recover the smooth transitions during the cell morphogenesis The overall square symmetry of the cell is captured by a peak at L4 Figure 4: LOCO-EFA metrics on a cell changing its shape over time Sequence of a tracked pavement cell growing over time with normalised area. (H) Pn and marginal diﬀerence proﬁles using EFA. Applying EFA modes to approximate the cell shapes leads to erratic proﬁles that fail to recover the biological sequence of development, as can be observed in the Pn proﬁle and as spurious peaks at the third and ﬁfth har- monics in the marginal diﬀerence proﬁle. (I) Ln and marginal diﬀerence proﬁles using LOCO-EFA. The LOCO-EFA measurements recover the smooth transitions during the cell morphogenesis. The overall square symmetry of the cell is captured by a peak at L4, the formation of lobes by a smooth increase in L9, and later L13. 21 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint L4 L5 L6 L7 C 0 0.2 0 0.16 0.12 0 0.08 L4 L5 L6 L7 D 0 0.15 0 0.12 0 0.09 0 0.06 polarity directions CPM 'copy' direction n=5 n=6 isolated CPM cell interacting CPM cell n=6 E L4 L5 L6 L7 A 0 0.1 0 0.08 0 0.06 0 0.04 L4 L5 L6 L7 B 0 0.1 0 0.08 0 0.06 0 0.04 F G 0 Figure 5: LOCO-EFA analysis on in vivo and in silico pavement cells. (A, B) LOCO-EFA applied to speechless mutant (A) and wild type (B) leaf tissue. Colour coding depicts the Ln values for 4 diﬀerent Ln modes (n = 4 . . . 7), as indicated above each panel, with the scale shown below each panel. Figures It shows that very few cell shapes can be reasonably captured through a single Ln mode, revealing the complexity of the cell shapes. (C, D) LOCO-EFA applied to in silico pavement cells reveals the degree of divergence from their speciﬁed shape that interacting cells within a tissue experience. Two diﬀerent speciﬁed cell shape populations are shown (SCS1 and SCS3, each with six lobes, see Table S1). The speciﬁed shapes are depicted above each panel. Colour coding within the panels and of the speciﬁed shapes above each panel again depicts the Ln values, with the scale shown below each panel. Because cells interact with each other within the tissue, strong deviations in Ln contributions can be observed. (E–G) Modelling framework to generate the in silico tissues (E) Standard Cellular Potts Model is modiﬁed to allow for L4 L5 L6 L7 C 0 0.2 0 0.16 0.12 0 0.08 polarity directions CPM 'copy' direction n=5 n=6 isolated CPM cell E L4 L5 L6 L7 A 0 0.1 0 0.08 0 0.06 0 0.04 L4 L5 L6 L7 B 0 0.1 0 0.08 0 0.06 0 0.04 F 0 A L5 L6 L7 0.04 B L7 E F D 0.12 0 0.06 0.09 0 0.15 0 Figure 5: LOCO-EFA analysis on in vivo and in silico pavement cells. (A, B) LOCO-EFA applied to speechless mutant (A) and wild type (B) leaf tissue. Colour coding depicts the Ln values for 4 diﬀerent Ln modes (n = 4 . . . 7), as indicated above each panel, with the scale shown below each panel. It shows that very few cell shapes can be reasonably captured through a single Ln mode, revealing the complexity of the cell shapes. (C, D) LOCO-EFA applied to in silico pavement cells reveals the degree of divergence from their speciﬁed shape that interacting cells within a tissue experience. Two diﬀerent speciﬁed cell shape populations are shown (SCS1 and SCS3, each with six lobes, see Table S1). The speciﬁed shapes are depicted above each panel. Colour coding within the panels and of the speciﬁed shapes above each panel again depicts the Ln values, with the scale shown below each panel. Because cells interact with each other within the tissue, strong deviations in Ln contributions can be observed. (E–G) Modelling framework to generate the in silico tissues. Figures It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint Table Table 1: Distinct shape descriptors have been used to quantify pavement cells. Measure Description Reference Average lobe length and neck width Length of each lobe and the distance between opposite indentations within a cell (called necks) are shown in Figure S1. The ﬁnal measure for a cell is the average of all lobe lengths and the average of all the neck widths. These measurements depend on human criteria to identify lobes and necks, and are given in absolute length units (thus, incomparable throughout growth stages). (Fu et al., 2005) Form factor (or circularity) Deﬁned as P 2 4πA, where p is the perimeter and A is cell area. A circle corresponds to a form factor 1, the lowest value possible. (Bai et al., 2010; Russ, 2011; Andriankaja et al., 2012) Skeleton This metric relies on the number of end points of a skeleton representation of the cell shape. The skeleton is formed by iteratively removing pixels from a grid-based cell shape representation, such that eventually a branched one-dimensional graph remains. There are diﬀerent variants of this algorithm to skeletonise shapes; the resulting branch patterns and length of branches highly depends on the parameters used and are very sensitive to the image resolution. (Le et al., 2006; Russ, 2011) Average polarity score Deﬁned as c+s 2 , where c is the circularity and s the number of skeleton end points. (Sorek et al., 2011) Table 1: Distinct shape descriptors have been used to quantify pavement cells. Measure Description Reference Average lobe length and neck width Length of each lobe and the distance between opposite indentations within a cell (called necks) are shown in Figure S1. The ﬁnal measure for a cell is the average of all lobe lengths and the average of all the neck widths. These measurements depend on human criteria to identify lobes and necks, and are given in absolute length units (thus, incomparable throughout growth stages). (Fu et al., 2005) Form factor (or circularity) Deﬁned as P 2 4πA, where p is the perimeter and A is cell area. A circle corresponds to a form factor 1, the lowest value possible. (Bai et al., 2010; Russ, 2011; Andriankaja et al., 2012) Skeleton This metric relies on the number of end points of a skeleton representation of the cell shape. . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint Figures (E) Standard Cellular Potts Model is modiﬁed to allow for a speciﬁed number of lobes (here, n = 5) to be formed at a regular radial spacing (α). (F) This gives rise to a symmetric, multi-lobed speciﬁed cell shape, shown in red. (G) Within a tissue context, however, the same speciﬁed shape deforms as to accompany the neighbouring cells. re 5: LOCO-EFA analysis on in vivo and in silico pavement cells. (A, B Figure 5: LOCO-EFA analysis on in vivo and in silico pavement cells. (A, B) LOCO-EFA applied to speechless mutant (A) and wild type (B) leaf tissue. Colour coding depicts the Ln values for 4 diﬀerent Ln modes (n = 4 . . . 7), as indicated above each panel, with the scale shown below each panel. It shows that very few cell shapes can be reasonably captured through a single Ln mode, revealing the complexity of the cell shapes. (C, D) LOCO-EFA applied to in silico pavement cells reveals the degree of divergence from their speciﬁed shape that interacting cells within a tissue experience. Two diﬀerent speciﬁed cell shape populations are shown (SCS1 and SCS3, each with six lobes, see Table S1). The speciﬁed shapes are depicted above each panel. Colour coding within the panels and of the speciﬁed shapes above each panel again depicts the Ln values, with the scale shown below each panel. Because cells interact with each other within the tissue, strong deviations in Ln contributions can be observed. (E–G) Modelling framework to generate the in silico tissues. (E) Standard Cellular Potts Model is modiﬁed to allow for a speciﬁed number of lobes (here, n = 5) to be formed at a regular radial spacing (α). (F) This gives rise to a symmetric, multi-lobed speciﬁed cell shape, shown in red. (G) Within a tissue context, however, the same speciﬁed shape deforms as to accompany the neighbouring cells. 22 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. Average polarity score Deﬁned as c+s 2 , where c is the circularity and s the number of skeleton end points. (Sorek et al., 2011) ndard Fourier Analysis cannot be used to quantify complex cell shapes Standard Fourier Analysis has been widely used to analyse cell morphology. It can, however, only be applied when cells present simple holomorphic shapes, i.e., when the radii emanating from the centroid of a cell intersect its outline only once (Figure S2A and Pincus and Theriot (2007)). When the geometry of a cell is more complex, as in the case of pavement cells, and radii emanating from the centroid can intersect the outline more than once, the shape cannot be decomposed using a Fourier expansion based on polar coordinates (Figure S2B and Schmittbuhl et al. (2003)). Table The skeleton is formed by iteratively removing pixels from a grid-based cell shape representation, such that eventually a branched one-dimensional graph remains. There are diﬀerent variants of this algorithm to skeletonise shapes; the resulting branch patterns and length of branches highly depends on the parameters used and are very sensitive to the image resolution. (Le et al., 2006; Russ, 2011) Average polarity score Deﬁned as c+s 2 , where c is the circularity and s the number of skeleton end points. (Sorek et al., 2011) 23 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint Supplementary Materials and Methods Decomposing shape: Lobe Contribution Elliptical Fourier Ana- lysis (LOCO-EFA) In this section, we ﬁrst summarise previous eﬀorts to make EFA coeﬃcients interpretable within a morphometrics perspective and explain why matching EFA coeﬃcients with shape features generally does not hold. We describe in detail our new method, Lobe Contribution Elliptical Fourier Analysis (LOCO-EFA). We show how it provides quant- itative and biologically interpretable measurements that are unique for a given shape, overcoming the shortfalls of the previous methods. ptical Fourier Analysis fails to align mode frequency with morphologica It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint point (xi, yi), is referred to as ti, i.e., ti = Pi p=1 △tp, with t0 = 0, and tK = T, the “total drawing time” or total perimeter length (see Figure 1 in the main text). Given that no equal spacing between the points is required, it is straightforward to deﬁne K observation points from any kind of cell contour. The only requirements are that the contour is closed and the coordinates form an ordered list that follows the contour. The EFA coeﬃcients are then given by: αn = T 2n2π2 K X i=1 △xi △ti cos 2nπti T −cos 2nπti−1 T , (5a) βn = T 2n2π2 K X i=1 △xi △ti sin 2nπti T −sin 2nπti−1 T , (5b) γn = T 2n2π2 K X i=1 △yi △ti cos 2nπti T −cos 2nπti−1 T , (5c) δn = T 2n2π2 K X i=1 △yi △ti sin 2nπti T −sin 2nπti−1 T . (5d) (5a) (5b) (5c) (5d) The oﬀset to the contour is given by: The oﬀset to the contour is given by: The oﬀset to the contour is given by: α0 = 1 T K X i=1 △xi 2△ti t2 i −t2 i−1 + ξi(ti −ti−1) + x0 , (6a) γ0 = 1 T K X i=1 △yi 2△ti t2 i −t2 i−1 + εi(ti −ti−1) + y0 . (6b) (6a) γ0 = 1 T K X i=1 △yi 2△ti t2 i −t2 i−1 + εi(ti −ti−1) + y0 . (6b) (6b) where ξi = Pi−1 j=1 △xj −△xi △ti Pi−1 j=1 △tj, εi = Pi−1 j=1 △yj −△yi △ti Pi−1 j=1 △tj, and ξ1 = ε1 = 0. Further details and full derivation can be found in Kuhl and Giardina (1982). ptical Fourier Analysis fails to align mode frequency with morphologica In 1982 Kuhl and Giardina proposed the Elliptical Fourier Analysis (EFA) to describe the contour of any two-dimensional shape (both holomorphic and non-holomorphic), derived from the coordinates of all the points along its outline. In short, EFA takes the x and y coordinates of a closed contour and decomposes it into an inﬁnite summation of related ellipses: x(t) =α0 + ∞ X n=1 αn cos 2nπt T + βn sin 2nπt T , (4a) y(t) =γ0 + ∞ X n=1 γn cos 2nπt T + δn sin 2nπt T , (4b) (4a) (4b) where αn, βn, γn and δn are the so-called EFA coeﬃcients and α0 and γ0 are the x- and y-oﬀset of the initial contour. The detailed derivation of the formulae for α0, γ0, αn, βn, γn and δn can be found in Kuhl and Giardina (1982). They are calculated from a discrete chain of contour points (xi, yi) with i = 1, . . . , K (see Figure S3A), K being the total number of points along the closed contour. We deﬁne (x0, y0) ≡(xK, yK), given that cell contours are closed. Now imagine drawing the contour of the cell, then △ti is the time spent drawing the line segment of the contour that links (xi−1, yi−1) to (xi, yi), i.e., q △ti = q (xi −xi−1)2 + (yi −yi−1)2 = p △x2 i + △y2 i . Note that △ti is not ﬁxed but can vary for each interval. Deﬁne T as the total time spent to draw the whole contour, i.e., T=PK i=1△ti. The “time” passed while drawing the contour, starting from contour point (x0, y0), or, equivalently, the distance passed along the contour to reach each contour △ti = q (xi −xi−1)2 + (yi −yi−1)2 = p △x2 i + △y2 i . Note that △ti is not ﬁxed but can vary for each interval. Deﬁne T as the total time spent to draw the whole contour, i.e., T=PK i=1△ti. The “time” passed while drawing the contour, starting from contour point (x0, y0), or, equivalently, the distance passed along the contour to reach each contour 24 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. ptical Fourier Analysis fails to align mode frequency with morphologica ; https://doi.org/10.1101/157842 doi: bioRxiv preprint The direction of rotation of the nth harmonic ellipse is given by the determinant of the EFA coeﬃcients matrix, det αn βn γn δn , i.e., the direction of rotation is given by the sign of (8) (8) rn = αnδn −βnγn . If rn < 0, the elliptic harmonic is rotating clockwise; if rn > 0 the elliptic harmonic is rotating counter-clockwise. When EFA is used for shape approximation, mode n con- tributes to shape features with an n + 1 or n −1 periodicity. This is in contrast to standard Fourier Analysis, in which mode n contributes to shape features with an n periodicity. Standard Fourier Analysis, however, is only possible for holomorphic shapes, and hence cannot be applied to, for example, pavement cells. Diaz et al. (1990) observed that whether mode n predominantly contributes to shape features with an n + 1 or with an n −1 periodicity strongly depends on whether the nth harmonic rotates together with or against the direction of the ﬁrst harmonic (see Movie S2 and Movie S3). This eﬀect of presenting contributions to the n + 1’th and n −1’th mode depending on the rotation direction of the ﬁrst and nth harmonic is a common phenomenon observed for objects orbiting around others (hereafter referred to as the relative direction eﬀect). A well-known example of the relative direction eﬀect is the rotation of the Earth and its movement around the sun. The actual number of rotations our planet makes per year (as observed from “star-rise to star-rise”, the so-called sidereal days) is one oﬀfrom the number of days we perceive in a year (from “sunrise to sunrise”, the so-called solar days). Because our planet rotates around its axis in the same direction as it moves around the Sun, the number of solar days per year is 365, one less than the number of sidereal days per year, which is 366. If the rotation of Earth would have been in the opposite direction as its movement around the sun, the number of solar days per year would instead have been 367. In light of exactly the same principle, Diaz et al. ptical Fourier Analysis fails to align mode frequency with morphologica where ξi = Pi−1 j=1 △xj −△xi △ti Pi−1 j=1 △tj, εi = Pi−1 j=1 △yj −△yi △ti Pi−1 j=1 △tj, and ξ1 = ε1 = 0. Further details and full derivation can be found in Kuhl and Giardina (1982). Each set of four coeﬃcients yields an ellipse (also referred to as the“nth mode”or“nth elliptic harmonic”), with a certain orientation and a certain starting point. The original cell outline can thus be expressed as an inﬁnite summation of ellipses. Note that x(t) and y(t) are periodic functions with period equal to T. A visual way to understand how the set of ellipses gives rise to the ﬁnal shape is as follows: the second elliptic harmonic traces two clockwise or counter-clockwise revolutions around the ﬁrst harmonic; the third harmonic traces three revolutions around the path drawn by the second harmonic; and the nth harmonic traces n revolutions around the path drawn by the previous harmonic (see Figure S3 and Movie S1). Diaz et al. (1990) proposed a heuristic measure regarding the contribution of each har- monic to the shape through an approximation of the perimeter of each ellipse multiplied by its harmonic number n: (7) Pn = 2πn r λ2 1n + λ2 2n 2 , (7) where λ1n and λ2n are the major and minor axis of the nth ellipse. Moreover, Diaz et al. (1990) introduced an additional correction to capture the complex relationship between EFA modes and shape feature periodicity. 25 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. Contouring the limitations: Lobe Contribution Elliptical Fourier Analysis (LOCO-EFA) To capture the biologically relevant cell shape features, overcoming the limitations of using Pn and rotation-dependent n + 1, n −1 adjustments, we have developed an alternative method coined Lobe Contribution Elliptical Fourier Analysis (LOCO-EFA). As the name indicates, it correctly maps the contribution of each mode/harmonic to the corresponding morphological features. This is done by separating each elliptic harmonic into two circular harmonics, each rotating in an opposite direction. First we rewrite the EFA (Equation 4) in matrix form: First we rewrite the EFA (Equation 4) in matrix form: (9) x(t) y(t) = α0 γ0 + N X n=1 αn βn γn δn cos 2nπt T sin 2nπt T , (9) with the inﬁnite sum being truncated at the Nth order harmonic. Equation 9 can concisely be expressed as (10) [X(t)] = [A0] + N X n=1 [An] [Mn(t)] , (10) in which [X(t)] corresponds to the drawn cell outline x(t) y(t) ; [A0] represents the spatial oﬀset α0 γ0 ; [An] corresponds to the EFA coeﬃcients matrix αn βn γn δn ; and [Mn(t)] refers to the rotor cos 2nπt T sin 2nπt T . (For clarity, we will use the notation [..] through- out to emphasise we are dealing with matrices; not to be confused with \|..\| that represents determinant, which we here only refer to as det [..].) The LOCO-EFA method consists of three steps: 1) eliminate multiple representations of the same outline; 2) decompose each nth elliptic harmonic into two circular harmonics, each rotating in an opposite direction; and 3) determine Ln and Ln for all N modes. Below we describe these steps in detail. ptical Fourier Analysis fails to align mode frequency with morphologica ; https://doi.org/10.1101/157842 doi: bioRxiv preprint new base. The details of how to do so are discussed below. We call the new base Ln, which, when summed up, can also reconstitute the original shape. new base. The details of how to do so are discussed below. We call the new base Ln, which, when summed up, can also reconstitute the original shape. Contouring the limitations: Lobe Contribution Elliptical Fourier Analysis (LOCO-EFA) ptical Fourier Analysis fails to align mode frequency with morphologica (1990) introduced that when the nth elliptic harmonic is moving in the same direction as the ﬁrst harmonic, its shape contribution Pn should be assigned to n−1; inversely, when the direction of a given mode is opposite to the ﬁrst harmonic, its shape contribution Pn should be assigned to n + 1. We will show below that this simple heuristic is reasonable as long as the ellipse marginally deviates from a circle, but is is not valid in general. When the aspect ratio of the ellipse (λ1n/λ2n) is large (i.e., the elliptical harmonic is very ﬂat, deviating signiﬁcantly from a circular shape), the proposed rule fails to apply. Figure S3F illustrates a situation when the rotation direction of the ﬁrst and third harmonic are opposite (and no other modes are used), yet instead of generating a contour with n −1 = 2 protrusions, as expected from the heuristic rule, a four-sided outline is generated, clearly illustrating that this method of Pn shifting does not work in general (see also Movie S4). Moreover, it is not possible to reconstruct the original shape using the Pn values, and therefore cannot be used for additional analysis based on shape reconstruction as presented in the main paper. This strongly limits usage of EFA for biological shape interpretations and statistical population analysis. Surprisingly, although EFA has been used to quantify morphology at the organ level, relative direction eﬀect has typically been ignored altogether (Yoshioka et al., 2005; Frieß and Baylac, 2003; Neto et al., 2006; Iwata et al., 1998, 2010; Chitwood et al., 2013). Realising that the source of the problem is linked to the eccentricity of the ellipses, it became clear to us that we could overcome this issue by essentially decomposing each ellipse into two counter-rotating circles. All circles can then be redistributed, forming a 26 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. (1) Eliminate multiple representations of the same outline The ﬁrst step is to determine where the new starting point should be positioned, as well as the scaled amount of time or temporal angle τ 1 = 2πt1 T required to reach the starting point (see Kuhl and Giardina, 1982). As stated above, we wish the starting point to coincide with one of the extremes of the semi-major axis of the ﬁrst harmonic. Points along the ﬁrst harmonic (x1, y1) can be described as : x1(τ) =α1 cos τ + β1 sin τ , (11a) y1(τ) =γ1 cos τ + δ1 sin τ , (11b) (11a) (11b) (11b) with τ = 2πt T being the scaled time or temporal angle. By diﬀerentiating the magnitude of the ﬁrst harmonic ellipse E(τ)= p x1(τ)2 + y1(τ)2 and setting its derivative to zero dE(τ) dτ = 0 , the temporal angles can be found at which the extremes along the semi- major and semi-minor axes of the ﬁrst harmonic are reached (Kuhl and Giardina, 1982): (12) τ1(ν) = 1 2 arctan 2(α1β1 + γ1δ1) α2 1 + γ2 1 −β2 1 −δ2 1 + ν 2π . (12) The values ν = 0, 1, 2, 3 give the four possible solutions along both the axes, after which the same points get repeated. For LOCO-EFA, it is required (see further below) to limit the starting point to the semi-major axis only. To satisfy this condition, the second derivative of E(τ), evaluated at the temporal angle, should be negative, i.e., d2E(τ) dτ 2 τ1 < 0. Substituting the found solutions into the second derivative results in ν = 0 and ν = 2 belonging to the points along the semi-major axis whenever the denominator of the arctan term is positive, and the solutions ν = 1 and ν = 3 belonging to the points along the semi-major axis whenever the denominator of the arctan term is negative. A very straightforward computational implementation of this result is to make use of the four-quadrant inverse tangent function (atan2) as provided by most programming languages (i.e., such that atan2(1, 1) = π/4 is diﬀerent from atan2(−1, −1) = −3π/4). Then, using (13) τ1 = 0.5atan2 2(α1β1 + γ1δ1), α2 1 + γ2 1 −β2 1 −δ2 1 (13) automatically and unambiguously ensures that the temporal angle τ1 is located at one of the extremes of the semi-major axis. (1) Eliminate multiple representations of the same outline It had already been noted that EFA coeﬃcients are redundant and therefore compromise statistical analysis and shape comparisons (Haines and Crampton, 2000). We found that there are three sources of degeneracy in the EFA coeﬃcients that therefore have to be eliminated. First, a contour can be drawn starting from any arbitrary initial point along the contour. While exactly the same outline is drawn, each starting point is represented by a completely diﬀerent set of EFA coeﬃcients for all modes [An]. Basically, whenever the starting point is changed, all elliptic harmonics take a diﬀerent orientation (Kuhl and Giardina, 1982). The ﬁrst step is therefore to transform the EFA coeﬃcients such that the starting point of the ﬁrst harmonic is always positioned at, for standardisation, the extreme of the semi-major axis (see further below). The second source of degeneracy, however, is that such a normalisation still allows for two possible representations of the 27 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint outline, since each of the two extremes along the semi-major axis can be chosen as the starting point. Moreover, a third source of degeneracy is due to the fact that the outline can be drawn clockwise or counter-clockwise. Clearly, all three sources of degeneracy have to be removed to make any comparison between cells sensible. outline, since each of the two extremes along the semi-major axis can be chosen as the starting point. Moreover, a third source of degeneracy is due to the fact that the outline can be drawn clockwise or counter-clockwise. Clearly, all three sources of degeneracy have to be removed to make any comparison between cells sensible. (1) Eliminate multiple representations of the same outline ; https://doi.org/10.1101/157842 doi: bioRxiv preprint [A1] [M1(t)] = [A1] [M1 (τ ′ + τ1)] , (14a) = [A1] cos (τ ′ + τ1) sin (τ ′ + τ1) , (14b) = [A1] cos (τ1) cos (τ ′) −sin (τ1) sin (τ ′) sin (τ1) cos (τ ′) + cos (τ1) sin (τ ′) , (14c) = [A1] cos (τ1) −sin (τ1) sin (τ1) + cos (τ1) cos (τ ′) sin (τ ′) , (14d) = [A1] [ψτ1] [M1(τ ′)] , (14e) = [A′ 1] [M1(τ ′)] , (14f) ] = [A1] [M1 (τ ′ + τ1)] , (14a) = [A1] cos (τ ′ + τ1) sin (τ ′ + τ1) , (14b) (14b) = [A1] [ψτ1] [M1(τ ′)] , (14e) ′ ′ = [A′ 1] [M1(τ ′)] , (14f) were [ψτ1] is the rotation operator, rotating by an angle τ1 and [A′ 1] = [A1] [ψτ1]. The spatial angle at the shifted starting point ϱ is given by (15) ϱ(ν) = arctan γ′ 1 α′ 1 + νπ . (15) Again, a single, unique and correct solution for ϱ can be obtained by using ϱ = atan2(γ′ 1, α′ 1) instead. The starting point lies in quadrant III or IV when ϱ < 0. In that case, τ1 is modiﬁed as follows: Again, a single, unique and correct solution for ϱ can be obtained by using ϱ = atan2(γ′ 1, α′ 1) instead. The starting point lies in quadrant III or IV when ϱ < 0. In that case, τ1 is modiﬁed as follows: (16) τ ⋆ 1 = (τ1 + π) . (16) Otherwise (when the starting point is already in quadrant I or II), τ ⋆ 1 = τ1. The new EFA coeﬃcients corrected for the starting point then become: Otherwise (when the starting point is already in quadrant I or II), τ ⋆ 1 = τ1. The new EFA coeﬃcients corrected for the starting point then become: (17) α⋆ n β⋆ n γ⋆ n δ⋆ n = αn βn γn δn cos (nτ ⋆ 1 ) −sin (nτ ⋆ 1 ) sin (nτ ⋆ 1 ) cos (nτ ⋆ 1 ) . (17) Finally, we ensure that the direction of contour approximation of the ﬁrst harmonic is always counter-clockwise (i.e., that r1 ≥0, Equation 8). (1) Eliminate multiple representations of the same outline This still leaves two ways to position the starting point (one for each of the extremes of the semi-major axis) and thereby two distinct representations of a same outline. We therefore further restrict τ1 to always lie within the ﬁrst or second quadrant (I or II in Figure S9A). This is achieved by testing if the obtained τ1 that shifts the starting point of the ﬁrst harmonic to the semi-major axis indeed positions it within quadrant I or II. To shift the starting point we ﬁrst introduce a time shift τ ′ = τ −τ1 such that at τ ′ = 0 the ﬁrst harmonic is positioned along its semi-major axis: 28 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint (1) Eliminate multiple representations of the same outline ; https://doi.org/10.1101/157842 doi: bioRxiv preprint the semi-major axis, for example, to be parallel to the x-axis (or in any other preferred orientation). The details on how to perform those rotations can be found in Kuhl and Giardina (1982). Please note that unlike in their study, our subsequent analysis does not require such a cell contour realignment. the semi-major axis, for example, to be parallel to the x-axis (or in any other preferred orientation). The details on how to perform those rotations can be found in Kuhl and Giardina (1982). Please note that unlike in their study, our subsequent analysis does not require such a cell contour realignment. For simplicity of notation in the rest of the Supplementary Materials and Methods we refer to the [An] matrix, which elements have been normalised regarding the starting point and direction of reconstruction of the ﬁrst harmonic: (19) [An] ≡ an bn cn dn ≡ α⋆ n β⋆ n γ⋆ n δ⋆ n . (19) After all possible sources of redundancy have been removed, the next step of the LOCO-EFA method is to split each elliptic harmonic into two counter-rotating circles. After all possible sources of redundancy have been removed, the next step of the LOCO-EFA method is to split each elliptic harmonic into two counter-rotating circles. (1) Eliminate multiple representations of the same outline Besides removing redundancy by restricting the freedom of choice regarding the overall direction of contour approximation, this transformation also guarantees a unique correspondence between the properties of each subsequent harmonic and its contribution to the morphological features. When the direction of the ﬁrst harmonic is clockwise (r1 < 0), we therefore invert the direction of motion of all ellipses, maintaining thereby their inter-relationships. This can be done by running “time” backwards: (18) x(−t) y(−t) = α⋆ n β⋆ n γ⋆ n δ⋆ n  cos 2nπ(−t) T sin 2nπ(−t) T   = α⋆ n β⋆ n γ⋆ n δ⋆ n cos 2nπt T −sin 2nπt T = α⋆ n −β⋆ n γ⋆ n −δ⋆ n cos 2nπt T sin 2nπt T . (18) x(−t) y(−t) = α⋆ n β⋆ n γ⋆ n δ⋆ n  cos 2nπ(−t) T sin 2nπ(−t) T   = α⋆ n β⋆ n γ⋆ n δ⋆ n cos 2nπt T −sin 2nπt T = α⋆ n −β⋆ n γ⋆ n −δ⋆ n cos 2nπt T sin 2nπt T . (18) In short, whenever r1 < 0, all indices β⋆ n and δ⋆ n should be negated. After these steps, each unique cell contour is represented by a unique set of EFA coeﬃcients. Note that the steps above do not alter the layout, nor do they rotate the shape. In certain study contexts, however, it might be desirable to rotate the contour itself, positioning 29 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. (2) Decompose each nth elliptic harmonic into two circles with opposite dir- ection of rotation It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint be equivalent, by combining Equation 22 and Equation 23, the [Λn] matrix can be iden- tiﬁed, with λ1n corresponding to the length of the semi-major axis, and the modulus of λ2n to the length of the semi-minor axis of the nth ellipse: be equivalent, by combining Equation 22 and Equation 23, the [Λn] matrix can be iden- tiﬁed, with λ1n corresponding to the length of the semi-major axis, and the modulus of λ2n to the length of the semi-minor axis of the nth ellipse: (24) [Λn] = [ψSn]−1 [An] [ψTn] = λ1n 0 0 λ2n . (24) Note that this process is similar to a singular value decomposition of [An], with the diﬀerence that here λ2n (but not λ1n) can be negative. The temporal angle φn corresponds, for each elliptic harmonic, to the scaled time τn to reach an extreme along the semi-major axis (Figure S9). Similarly to Equation 12, this corresponds to (25) φn(ν) = 1 2 arctan 2(anbn + cndn) a2 n + c2 n −b2 n −d2 n + ν 2π . (25) Once again, using atan2 ensures that φn corresponds to the semi-major axis (see above). For the next steps of the analysis (see below) it is essential that φn corresponds to the temporal angle to reach the semi-major axis, not the semi-minor axis, hence usage of atan2 or any equivalent function which determines the quadrant of the return value is essential. Once again, using atan2 ensures that φn corresponds to the semi-major axis (see above). For the next steps of the analysis (see below) it is essential that φn corresponds to the temporal angle to reach the semi-major axis, not the semi-minor axis, hence usage of atan2 or any equivalent function which determines the quadrant of the return value is essential. (2) Decompose each nth elliptic harmonic into two circles with opposite dir- ection of rotation In order to ﬁnd the contribution of nth harmonic to a given morphological feature, we rewrite the [An] matrices in Equation 9 such as to explicitly introduce the length of the semi-major and semi-minor axis of the nth ellipse (λ1n and λ2n). For this purpose, it is necessary to introduce both the temporal and the spatial rotation operator of each elliptic harmonic, given by [ψTn] and [ψSn], respectively. The temporal operator is deﬁned as (20) [ψTn] = cos φn −sin φn sin φn cos φn , (20) and the spatial operator is deﬁned as (21) [ψSn] = cos θn −sin θn sin θn cos θn , (21) where φn is the temporal angle (i.e., the time τn required to rotate to the semi-major axis) and θn the spatial angle (i.e., the angle of this position along the semi-major axis with the positive x-axis) (Figure S9A). where φn is the temporal angle (i.e., the time τn required to rotate to the semi-major axis) and θn the spatial angle (i.e., the angle of this position along the semi-major axis with the positive x-axis) (Figure S9A). Equation 10 can be written as: Equation 10 can be written as: (22) [X(t)] = [A0] + N X n=1 [ψSn] [ψSn]−1 [An] [ψTn] [ψTn]−1 [Mn(t)] , (22) given that [ψSn][ψSn]−1and [ψTn] [ψTn]−1 correspond to the identity matrix [I]. Equation 10 can also be written in a form which directly highlights the contribution of the semi-major and semi-minor axis: (23) [X(t)] = [A0] + N X n=1 [ψSn] [Λn] [ψTn]−1 [Mn(t)] . (23) This equation can be understood as follows: for each mode, correctly position the starting point relative to the semi-major axis, transform the original circle into an ellipse, its semi-major axis along the x-axis and semi-minor axis along the y-axis, and ﬁnally ro- tate the ellipse to its correct position (see Figure S9B–D). Since both descriptions should 30 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. (2) Decompose each nth elliptic harmonic into two circles with opposite dir- ection of rotation CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint where λ+n and λ−n are the radii of those circles (Figure 2). Summing up the diagonal matrices in Equation 28 yields where λ+n and λ−n are the radii of those circles (Figure 2). Summing up the diagonal matrices in Equation 28 yields x(t) y(t) = [A0] + N X n=1 cos θn −sin θn sin θn cos θn λ+n + λ−n 0 0 λ+n −λ−n cos φn sin φn −sin φn cos φn cos 2nπt T sin 2nπt T , (29) y( ) = [A0] + N X n=1 cos θn −sin θn sin θn cos θn λ+n + λ−n 0 0 λ+n −λ−n cos φn sin φn −sin φn cos φn cos 2nπt T sin 2nπt T , (29) (29) in which the major and minor axes of each elliptic harmonic are in which the major and minor axes of each elliptic harmonic are λ1n =λ+n + λ−n , (30a) λ2n =λ+n −λ−n , (30b) (30a) (30b) (30b) and hence the radii of each oppositely-rotating circle is given by λ+n =(λ1n + λ2n)/2 , (31a) λ−n =(λ1n −λ2n)/2 . (31b) (31a) (31b) (31a) (31a) (31b) (31b) Given that λ1n ≥0 and \|λ1n\| ≥\|λ2n\|, λ+n and λ−n are always positive. To approx- imate the cell contour (x(t), y(t)) using the circles λ+n and λ−n requires completing the transformations using the spatial (θn) and temporal angle (φn) as calculated before, most clearly seen through the expression (32) [X(t)] = [A0] + N X n=1 [ψSn] [Λ+n + Λ−n] [ψTn]−1 [Mn(t)] . (32) The term Λ+n presents the subset of the nth elliptic harmonic which is moving in the same direction as the ﬁrst harmonic, therefore purely contributing to n −1 “lobes”, i.e., shape features with periodicity n −1. In contrast, Λ−n presents the subset moving in the opposite direction, purely contributing to n + 1 “lobes” (shape features) only. (2) Decompose each nth elliptic harmonic into two circles with opposite dir- ection of rotation The spatial angle θn (see Figure S9) can be calculated after applying the temporal modiﬁcation: (26) θn(ν) = arctan c ′ n a ′ n + νπ , (26) where c ′ n and a ′ n, are the new coeﬃcients after the temporal transformation (i.e., after applying [An] [ψTn]). Again, it is essential to use atan2 to ensure that φn and θn are both relative to the same extreme of the semi-major axis. It does not matter, however, which of the two extremes is being used, hence for this step no check is required regarding the quadrants. Applying Equation 24 then provides λ1n and λ2n. Deriving a temporal and spatial angle relative to the semi-major axis and both being related to the same extreme guarantees that λ1n ≥0 (while λ2n can be positive or negative, depending on the rotation direction of the rotor) and \|λ1n\| ≥\|λ2n\|. Using the above, the EFA (Equation 10) can be rewritten as: Using the above, the EFA (Equation 10) can be rewritten as: x(t) y(t) = [A0] + N X n=1 cos θn −sin θn sin θn cos θn λ1n 0 0 λ2n cos φn sin φn −sin φn cos φn cos 2nπt T sin 2nπτt T . (27) x(t) y(t) = [A0] + N X n=1 cos θn −sin θn sin θn cos θn λ1n 0 0 λ2n cos φn sin φn −sin φn cos φn cos 2nπt T sin 2nπτt T . (27) (27) Written in this form, the contribution of each harmonic can easily be separated to correctly map to morphological feature number. The diagonal matrix containing the length of the semi-major and semi-minor axis of each nth mode can be decomposed into two diagonal matrices, each corresponding to circular orbits moving in opposite directions: (28) x(t) y(t) = [A0] + N X n=1 cos θn −sin θn sin θn cos θn λ+n 0 0 λ+n + λ−n 0 0 −λ−n cos φn sin φn −sin φn cos φn cos 2nπt T sin 2nπt T , (28) 31 . (2) Decompose each nth elliptic harmonic into two circles with opposite dir- ection of rotation (36b) Hence, introducing ζ+n = θn −φn and ζ−n = −θn −φn, Equation 34 can be written as (37) [Ajn] = [Λjn] ψζjn , for j = +, −, Hence, introducing ζ+n = θn −φn and ζ−n = −θn −φn, Equation 34 can be written a ng ζ+n = θn −φn and ζ−n = −θn −φn, Equation 34 can be written as (37) [Ajn] = [Λjn] ψζjn , for j = +, −, (37) which corresponds to which corresponds to which corresponds to (38) [Ajn] = ajn bjn cjn djn = λjn 0 0 jλjn cos ζjn −sin ζjn sin ζjn cos ζjn , for j = +, −. (38) We next label those matrices with respect to their lobe contribution instead of their EFA mode. To make the distinction, we here use subscript n to indicate the EFA mode, and subscript l to indicate the LOCO-EFA mode. In general, [A+l] = A+n+1 and [A−l] = A−n−1 . There are, however, a few exceptions, see Figure S4: [ l] n 1 1) The overall oﬀset of the contour is not solely given by [A0]. An additional contri- bution to the oﬀset is coming from [A+n=2]. Note, however, that the contribution from [A+n=2] is in fact not a perfect oﬀset to the contour, but also causes a kidney bean-shaped distortion to the contour, its deviation from a pure oﬀset becoming more pronounced when the contribution of this mode relative to the overall contour size is larger (see Figure S4B). 1) The overall oﬀset of the contour is not solely given by [A0]. An additional contri- bution to the oﬀset is coming from [A+n=2]. Note, however, that the contribution from [A+n=2] is in fact not a perfect oﬀset to the contour, but also causes a kidney bean-shaped distortion to the contour, its deviation from a pure oﬀset becoming more pronounced when the contribution of this mode relative to the overall contour size is larger (see Figure S4B). g ) 2) The overall circular shape of the contour (i.e., LOCO-mode 1) receives solely a contribution from [A+n=1] itself, i.e., [A+l=1] = [A+n=1]. 2) The overall circular shape of the contour (i.e., LOCO-mode 1) receives solely a contribution from [A+n=1] itself, i.e., [A+l=1] = [A+n=1]. . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint (2) Decompose each nth elliptic harmonic into two circles with opposite dir- ection of rotation Their contributions can be separated by writing: The term Λ+n presents the subset of the nth elliptic harmonic which is moving in the same direction as the ﬁrst harmonic, therefore purely contributing to n −1 “lobes”, i.e., shape features with periodicity n −1. In contrast, Λ−n presents the subset moving in the opposite direction, purely contributing to n + 1 “lobes” (shape features) only. Their contributions can be separated by writing: (33) [X(t)] = [A0] + N X n=1 [A+n] [Mn(t)] + N X n=1 [A−n] [Mn(t)] , (33) where where (34) [Ajn] = [ψSn] [Λjn] [ψTn]−1 , for j = +, −. (34) This can be further simpliﬁed. Straightforwardly, [ψSn] [Λ+n] = [ψSn] λ+n [I] = [Λ+n] [ψSn]. Regarding [Λ+n], This can be further simpliﬁed. Straightforwardly, [ψSn] [Λ+n] = [ψSn] λ+n [I] = [Λ+n] [ψSn] Regarding [Λ+n], [ψSn] [Λ−n] = cos θn −sin θn sin θn cos θn λ−n 0 0 −λ−n , (35a) = λ−n cos θn +λ−n sin θn λ−n sin θn −λ−n cos θn , (35b) = λ−n 0 0 −λ−n cos θn sin θn −sin θn cos θn , (35c) = λ−n 0 0 −λ−n cos (−θn) −sin (−θn) sin (−θn) cos (−θn) , (35d) = [Λ−n] [ψSn]−1 . (35e) (35a) (35b) 32 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint Thus, Thus, [A+n] = [Λ+n] [ψSn] [ψTn]−1 , (36a) [A−n] = [Λ−n] [ψSn]−1 [ψTn]−1 . (36b) Hence, introducing ζ+ = θn −φn and ζ−= −θn −φn, Equation 34 can be written as [A+n] = [Λ+n] [ψSn] [ψTn]−1 , (36a) [A−n] = [Λ−n] [ψSn]−1 [ψTn]−1 . (36b) [A+n] = [Λ+n] [ψSn] [ψTn]−1 , (36a) [A−n] = [Λ−n] [ψSn]−1 [ψTn]−1 . (36b) Hence, introducing ζ+n = θn −φn and ζ−n = −θn −φn, Equation 34 can be written as (37) [Ajn] = [Λjn] ψζjn , for j = +, −, (36a) (36b) [A−n] [Λ−n] [ψSn] [ψTn] . (2) Decompose each nth elliptic harmonic into two circles with opposite dir- ection of rotation [ n 1] [ l 1] [ n 1] 3) When N EFA modes are taken into account, then LOCO-mode N only receives a contribution from A−n=N−1 . 3) When N EFA modes are taken into account, then LOCO-mode N only receives a contribution from A−n=N−1 . 4) Likewise, when taking N EFA modes into account, there is still a contribution to LOCO-mode L = N + 1, solely coming from [A−n=N]. 4) Likewise, when taking N EFA modes into account, there is still a contribution to LOCO-mode L = N + 1, solely coming from [A−n=N]. Deﬁning [M+l] = [Mn+1] and [M−l] = [Mn−1], with the exceptions [M+l=0] = [Mn=2] and [M+l=1] = [Mn=1], then the same shape approximation can be achieved through LOCO-EFA as through EFA: (39) [X(t)] = [A0] + [A+l=0] [M+l=0(t)] + [A+l=1] [M+l=1(t)] + L−2 X l=2 [A+l] [M+l(t)] + L X l=2 [A−l] [M−l(t)] . (39) Equation 39 can be used to reconstruct the shape up to a certain L number, requiring EFA coeﬃcients up to mode L + 1: Equation 39 can be used to reconstruct the shape up to a certain L number, requiring EFA coeﬃcients up to mode L + 1: [XL=1(t)] = [A0] + [A+l=0] [M+l=0(t)] + [A+l=1] [M+l=1(t)] , (40a) [XL≥2(t)] = [A0] + [A+l=0] [M+l=0(t)] + [A+l=1] [M+l=1(t)] + L X l=2 [A+l] [M+l(t)] + L X l=2 [A−l] [M−l(t)] . (40b) (40a) (40b) 33 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint Each LOCO-EFA mode can be fully described by the combination of the radii and the starting positions of both circles. The radii are, as previously deﬁned, λ+l = λ+n+1 and λ−l = λ−n−1, with the exceptions λ+l=0 = λ+n=2 and λ−l=0 = 0, and λ+l=1 = λ+n=1 and λ−l=1 = 0. (3) Ll amplitude contributions In this section we ﬁrst show that the contribution to the shape of the two counter- rotating circles of a speciﬁc LOCO-EFA mode is not simply determined by the radii of those circles, but also depends on the relative position of the starting points. We then derive a heuristic which optimally captures its contribution, determining as well the limits of this approximation. To illustrate how a diﬀerence between the starting points ζ+l and ζ−l of mode l can aﬀect the generated amplitude, we ﬁrst plot the contribution of L3 = (λ+3, λ−3, ζ+3, ζ−3) superimposed on L1 = (1, 0, 0, 0), i.e., superimposed on the unit circle starting at zero degrees. As long as λ+3, λ−3 are not too large, a single amplitude value a (which is the amplitude of deviation from the unit circle) can be observed for any phase ω; peak amplitude values A occur for speciﬁc phases Ω(as illustrated in Figure S10I). Figure S10A–C depict the shape contribution of the negative rotor (green), positive rotor (red) and their summed contribution (orange), as a function of the phase ω, for L3 = (0.15, 0.15, 0, 0). This scenario illustrates that even though the negative rotor makes two sidereal rotations while the positive rotor makes four sidereal rotations, the amplitude pattern that they generate not only has a period three in both cases, as argued throughout the paper, but also perfectly match one another regarding the phases at which the peaks and troughs in the amplitude pattern are reached. The summed contribution (Figure S10C) therefore indeed yields a peak amplitude exactly equal to λ+3 + λ−3. In contrast, when both rotors are exactly out-of-phase (L3 = (0.15, 0.15, 0, π), Figure S10D–F), the peaks and troughs are exactly out-of-phase as well, and the patterns almost (but not totally) cancel each other out (Figure S10F). To determine the eﬀective contribution of a LOCO-EFA mode we therefore have to determine the phase at which each rotor reaches its peak amplitude. The angles Ω+, Ω−at which the positive and negative rotor reach their peak amplitude can be calculated straightforwardly. They occur when the phase of the rotor itself is equivalent to the overall phase generated by L1 (as illustrated in Figure S10G). The phase of L1 starts at ζ+1, while the phase of the positive rotor starts at ζ+l (illustrated in Figure S10K). (2) Decompose each nth elliptic harmonic into two circles with opposite dir- ection of rotation The starting points of both circles are deﬁned as ζ+l = ζ+n+1 and ζ−l = ζ−n−1, again with the exceptions ζ+l=0 = ζ+n=2 and ζ+l=0 = ζ+n=2. Together this gives a set of four coeﬃcients Ll = (λ+l, λ−l, ζ+l, ζ−l) that fully capture each LOCO-EFA mode and allow for a full reconstitution of the original shape: (41) [X(t)] = [A0] + [Λ+l=0] ψζ+l=0 [M+l=0(t)] + [Λ+l=1] ψζ+l=1 [M+l=1(t)] + L−2 X l=2 [Λ+l] ψζ+l [M+l(t)] + L X l=2 [Λ−l] ψζ−l [M−l(t)] . (41) In the next section we derive how from those values a single amplitude value can be found, the Ll contribution. (3) Ll amplitude contributions When the pattern is laid down, the phase of the positive rotor changes (l + 1) times faster than the phase of L1 (Equation 41). Thus, regarding the phase at peak amplitude it holds that 34 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint Ω+l = (l + 1) (Ω+l −ζ+1) + ζ+l + 2πν , (42a) −lΩ+l = ζ+l −(l + 1)ζ+1 + 2πν , (42b) Ω+l = ζ+1 + ζ+1 −ζ+l l + 2π l ν . (42c) (42a) (42b) (42c) where ν is any integer, and the values ν = 0, 1, . . . , l −1 provide the complete set of phases at which peak amplitude is reached. Likewise, the negative rotor rotates (l −1) times faster than L1 and in the opposite direction, hence starting at −ζ−l (see Figure S10J). Regarding the phase at peak amplitude it therefore holds that where ν is any integer, and the values ν = 0, 1, . . . , l −1 provide the complete set of phases at which peak amplitude is reached. Likewise, the negative rotor rotates (l −1) times faster than L1 and in the opposite direction, hence starting at −ζ−l (see Figure S10J). Regarding the phase at peak amplitude it therefore holds that Ω−l = −(l −1) (Ω+l −ζ+1) −ζ−l + 2πν , (43a) lΩ−l = −ζ−l + (l −1)ζ+1 + 2πν , (43b) Ω−l = ζ+1 −ζ+1 + ζ−l l + 2π l ν . (3) Ll amplitude contributions (43c) (43b) (43c) To assess the summed amplitude contribution we next ﬁt the amplitude pattern laid down by each rotor to a sine wave, exactly matching both the peak amplitude and the phase at which the peak amplitude is reached, and then sum those two contributions: To assess the summed amplitude contribution we next ﬁt the amplitude pattern laid down by each rotor to a sine wave, exactly matching both the peak amplitude and the phase at which the peak amplitude is reached, and then sum those two contributions: a+l = λ+l cos (lω + ζ+1 −(l + 1)ζ+1) , (44a) a−l = λ−l cos (lω + ζ−l −(l −1)ζ+1) , (44b) a+l + a−l = λ+l cos (lω + ζ+l −(l + 1)ζ+1) + λ−l cos (lω + ζ−l −(l −1)ζ+1) . (44c) (44a) ( ) (44b) (44c) a−l = λ−l cos (lω + ζ−l −(l −1)ζ+1) , (44b) (44 ) (44c) Equation 44a exactly matches the peak amplitude and its phase of the positive rotor, as derived in Equation 42, while Equation 44b exactly matches the peak amplitude and its phase of the negative rotor, as derived in Equation 43. While peak amplitude and phase match perfectly, Figure S10G, H illustrate that for not too large values of λ+l, λ−l also the rest of the pattern presents a close match. Using standard trigonometry, the summed amplitude can be written as a+l + a−l = al = q λ2 +l + λ2 −l + 2λ+lλ−l cos (ζ+l −ζ−l −2ζ+1) cos (lω + ζl) , (45a) (45a) where where ζl = atan2 (λ+l sin (ζ+1 −(l + 1)ζ+1) + λ−l sin (ζ−l −(l −1)ζ+1) , λ+l cos (ζ+1 −(l + 1)ζ+1) + λ−l cos (ζ−l −(l −1)ζ+1)) . (45b) ζl = atan2 (λ+l sin (ζ+1 −(l + 1)ζ+1) + λ−l sin (ζ−l −(l −1)ζ+1) , λ+l cos (ζ+1 −(l + 1)ζ+1) + λ−l cos (ζ−l −(l −1)ζ+1)) . (45b) 1)) (45b) Figure S10I illustrates that for not too large values of λ+l, λ−l this expression provides a close match to the pattern generated by the Ll mode, here illustrated using both a diﬀerent phase and a diﬀerent amplitude for both rotors. Using the equation above, the amplitude or contribution of mode l can hence be deﬁned as (46) Ll = q λ2 +l + λ2 −l + 2λ+lλ−l cos (ζ+l −ζ−l −2ζ+1) . . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint (3) Ll amplitude contributions (46) Ll = q λ2 +l + λ2 −l + 2λ+lλ−l cos (ζ+l −ζ−l −2ζ+1) . (46) 35 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint While above we have performed a more formal derivation, a more intuitive way to understand the amplitude contribution of the positive and negative rotor combined, is in terms of how much the amplitude contributions of both rotors are out-of-phase with each other. The most straightforward moment to determine their phase shift is at the initial point of drawing the cell’s outline. As illustrated in Figure S10J, K, the phase shift regarding the amplitude contribution of the negative rotor is given by ζ+1 +ζ−l, and of the positive rotor by ζ+l −ζ+1. To understand the graphical argument for the negative rotor case, one has to realise that the negative rotor rotates clockwise, and hence its starting angle is −ζ−l (for the positive rotor it is simply ζ+l), while for the same reason the phase shift between initial amplitude and maximum amplitude has to be calculated clockwise (instead of standard counterclockwise, as done for the positive rotor), again as shown in Figure S10J, K. Plotting the phase shift and strength of both amplitude contributions as vectors and then adding them up then provides the above equation for Ll (as is depicted in Figure 2C). Note that the pattern can never be perfectly described by a sine wave, which is why four parameters are needed to describe each mode, rather than the two required by Equation 45. Entropy measurement The entropy measure is deﬁned as (47) E = − L X l=1 fl ln fl , (47) where fl refers to the relative proportion of each Ll for a given L number of modes analysed, i.e., fl = Ll PL l=1 Ll . Final remarks where fl refers to the relative proportion of each Ll for a given L number of modes analysed, i.e., fl = Ll PL L . where fl refers to the relative proportion of each Ll for a given L number of modes analysed, i.e., fl = Ll PL l=1 Ll . where fl refers to the relative proportion of each Ll for a given L number of l d i f Ll analysed, i.e., fl = Ll PL l=1 Ll . analysed, i.e., fl = Ll PL l=1 Ll . (3) Ll amplitude contributions Even when following Equation 45 the two waves cancel each other out, this is not exactly the case, as seen in Figure S10F. There is therefore no additional level of redundancy, as the usage of the Ll numbers might suggest, even when the value of Ll = 0. When the amplitude of a rotor becomes larger, the generated pattern starts to deviate from a sine wave (Figure S10J, K), and their summed pattern from Equation 45 (Figure S10L). For simple closed contours (which cell outlines should always be), the devi- ations cannot be very large, since otherwise the contour becomes non-simple (i.e., crosses itself). This conﬁnement in deviations for cell outlines allows us to base a signiﬁcant part of our analysis on Ll values. Moreover, we observe that the positive rotor generates protrusions that are ﬂatter than a sine wave (Figure S10B, J), whereas the negative rotor generates lobes that are more pointy (Figure S10A, K). The extent of “ﬂatness” or “pointiness” of each mode can be determined by the proportion w+l = λ+l/(λ+l + λ−l) and w−l = 1 −w+l, respectively; a useful additional measure if needed. Final remarks ; https://doi.org/10.1101/157842 doi: bioRxiv preprint Final remarks To summarise, the LOCO-EFA method consists in: 1) eliminating degeneracies in the EFA coeﬃcients; 2) decomposing each elliptic harmonic into two circles rotating in op- posite directions (λ+n and λ−n) and therefore contributing to n −1 and n + 1 number of lobes (more generally, morphological features); and 3) calculating the oﬀset between 36 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint starting points of these two circles derived from each ellipse to estimate the amplitude of the Llth lobe contribution. starting points of these two circles derived from each ellipse to estimate the amplitude of the Llth lobe contribution. To eliminate cell area eﬀects (for example, when looking at shape diversity within a cell populations), it might also be desirable to normalise for cell size by dividing each Ln value by the square root of the cell area. This then provides a complete description of the number of lobes and their amplitude, which can be used to characterise and quantify intrinsic cell shape properties, irrespective of cell area, spacing between sampling points, and rotations or inversions of the cell (Figure S11A, B). Changes in the image resolu- tion, however can of course aﬀect the ﬁne-grained information retrieval, but only if the resolution becomes very low (Figure S11C, D). 37 37 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. Supporting Figures A B Lobe length Neck width p=45 p=65 p=85 Figure S1: Subjective choices involved in neck width and lobe length determin- ation and Skeletonisation. (A) Neck width and lobe length depend on human criteria for both identifying and quantifying such structures, as indicated by the question-marks. (B) Both the number of skeleton end-points and the length of the branches strongly de- pend on the parameter settings used for the skeletonisation algorithm. Here, parameter p (see material and Methods) was set to 45 (left), 65 (middle), or 85 (right). For those values, the number of predicted lobes varies between 6 and 8. Lobe length Neck width Neck width Neck width B B p=45 p=65 p=85 Figure S1: Subjective choices involved in neck width and lobe length determin- ation and Skeletonisation. (A) Neck width and lobe length depend on human criteria for both identifying and quantifying such structures, as indicated by the question-marks. (B) Both the number of skeleton end-points and the length of the branches strongly de- pend on the parameter settings used for the skeletonisation algorithm. Here, parameter p (see material and Methods) was set to 45 (left), 65 (middle), or 85 (right). For those values, the number of predicted lobes varies between 6 and 8. 38 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint A B Figure S2: Holomorphic and non-holomorphic shapes. (A) In holomorphic shapes all radii starting from the centroid intersect the outline only once. (B) In non-holomorphic shapes some radii intersect more than once. Very few pavement cells have a holomorphic shape, the majority presenting highly complex non-holomorphic outlines. The outline of non-holomorphic shapes cannot be represented as a function of the angle, precluding, for example, standard Fourier analysis. A B B B A Figure S2: Holomorphic and non-holomorphic shapes. Supporting Figures CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint Figure S3: Cell outline reconstruction using EFA and LOCO-EFA. (A) Calcu- lating the EFA coeﬃcients from a cell outline. The discrete chain of contour points can be positioned arbitrarily (for example, the points do not have to be associated to an un- derlying grid). Also the distances between points can be arbitrarily long. (B) Sequential approximation of the cell’s contour. The ﬁrst harmonic forms an elliptic shape (1, blue). The second harmonic describes an elliptic orbit (2–6, red), orbiting twice while moving around the ﬁrst harmonic (2, 3). Their summed trajectories are shown in green. The third elliptic harmonic (4–6, orange) orbits thrice while moving around this summed tra- jectory (4–6). The summation of the ﬁrst three elliptic harmonics is shown in blue. See Movie S1 for this dynamical reconstitution of the cell contour. (C) The EFA coeﬃcients cannot be directly linked to shape features, here shown through the power contribution of each harmonic (Pn) (upper panel): the main Pn contributions of a six-pointed star-like shape (shown as an inset) come from the 5th and 7th harmonic. To align the coeﬃcients to actual shape features, Diaz et al. (1990) proposed to shift the contribution of each harmonic to either n + 1 or n −1, depending on the rotation direction of each individual harmonic (middle panel). This brings the main shape contributor and the actual number of shape features in alignment to each other. This method, however, does not always hold (see F), and moreover generates a range of spurious contributions from a large set of diﬀerent modes, which is a notorious issue that hampers analysis of complex shapes using standard EFA (Haines and Crampton, 2000). The LOCO-EFA method correctly aligns the shape assessment and the real shape features (lower panel), without generating any additional spurious contributions. Supporting Figures (A) In holomorphic shapes all radii starting from the centroid intersect the outline only once. (B) In non-holomorphic shapes some radii intersect more than once. Very few pavement cells have a holomorphic shape, the majority presenting highly complex non-holomorphic outlines. The outline of non-holomorphic shapes cannot be represented as a function of the angle, precluding, for example, standard Fourier analysis. 39 39 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 2 4 6 8 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 2 4 6 8 E F D A B C 1 2 1 2 3 4 1 2 3 4 1 2 3 4 5 6 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 0.2 0.4 0.6 Pn without frequency aligment Pn with frequency aligment EFA mode EFA mode Ln mode Pn Ln Pn Ln from LOCO-EFA Figure S3 A A B 1 2 3 4 5 6 B A 6 5 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 2 4 6 8 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 2 4 6 8 C 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 0.2 0.4 0.6 Pn without frequency aligment Pn with frequency aligment EFA mode EFA mode Ln mode Pn Ln Pn Ln from LOCO-EFA Figure S3 C E F D 1 2 1 2 3 4 1 2 3 4 D E 1 F Figure S3 40 . Supporting Figures Moreover, unlike the other methods, the values correctly represent the amplitude of the shape features. (D–F) Contours (shown in blue), generated from the ﬁrst (green) and third EFA harmonic (orange) only. The number of morphological protrusions (lobes) speciﬁed by the nth EFA mode is aﬀected, but not fully determined, by its rotation direction. The heuristic rule proposed by Diaz et al. (1990) states that if the ﬁrst harmonic and the nth harmonic rotate in the same direction, a contour is generated with n −1 protrusions, while if their rotation direction is opposite, the contour will contain n + 1 lobes. This rule, however, is only correct when the elliptic harmonic has a circular shape (D, E), in which case indeed overall shapes are generated with either 2 (D) or 4 lobes (E), simply dependent on the rotation direction with respect to the ﬁrst mode. When, however, the elliptic harmonic has a higher eccentricity (F), the ﬁnal shape can have n + 1 protrusions (4 lobes) even though the rotation direction of the ﬁrst and third harmonic are opposite. This is a consequence of each single EFA mode actually contributing to two diﬀerent spatial modes. See also Movie S2–S4. 41 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint EFA mode LOCO-EFA mode 0 0 1 2 3 4 5 1 2 3 4 5 6 A B C D E Figure S4: Schematic mapping between EFA modes and LOCO-EFA modes. (A) Each nth EFA mode contributes to both n + 1 and n −1 morphological periodicities. The red arrows represent the contributions of the nth EFA mode to n+1 protrusions, due to the clockwise rotations of the circular harmonics λ−n+1. The blue arrows indicate the contributions to n −1 protrusions, due to the counter-clockwise rotations of the circular harmonics λ+n−1. Supporting Figures A few exceptions apply: The second EFA mode contributes to a shift in the positioning of the layout, i.e., to LOCO-EFA mode 0 (λ+0, blue and dashed line), rather than to the overall size of the layout, as might have been expected. The ﬁrst EFA mode contributes to the overall circular size of the layout (λ+1, blue and dashed line), rather than to a shift in the positioning of the layout, as might have been expected. The zeroth EFA mode only contributes to a shift in the positioning of the layout (yellow line). Finally, the two highest LOCO-EFA modes have incomplete contributions, given any cutoﬀin the number of EFA modes. (B–E) The contribution λ+0 is not simply an oﬀset of the contour, but also involves a kidney bean-shaped distortion, more pronounced for larger contributions. Mode λ+1 (the circular mode) is shown in blue; mode λ+0 in red; their summation in green; and the predicted shape if the contribution were purely an oﬀset in magenta. (B) λ+0 being 10% of λ+1, eﬀectively resulting in just an oﬀset to the shape outline (C). (D, E) λ+0 being 30% of λ+1, resulting in a clear kidney bean-shaped distortion of the shape. B D C E EFA mode LOCO-EFA mode A LOCO-EFA mode EFA mode B C A EFA mode LOCO-EFA mode 0 0 1 2 3 4 5 1 2 3 4 5 6 0 0 1 2 3 4 5 1 2 3 4 5 6 0 D E 6 Figure S4: Schematic mapping between EFA modes and LOCO-EFA modes. (A) Each nth EFA mode contributes to both n + 1 and n −1 morphological periodicities. The red arrows represent the contributions of the nth EFA mode to n+1 protrusions, due to the clockwise rotations of the circular harmonics λ−n+1. The blue arrows indicate the contributions to n −1 protrusions, due to the counter-clockwise rotations of the circular harmonics λ+n−1. A few exceptions apply: The second EFA mode contributes to a shift in the positioning of the layout, i.e., to LOCO-EFA mode 0 (λ+0, blue and dashed line), rather than to the overall size of the layout, as might have been expected. The ﬁrst EFA mode contributes to the overall circular size of the layout (λ+1, blue and dashed line), rather than to a shift in the positioning of the layout, as might have been expected. Supporting Figures The zeroth EFA mode only contributes to a shift in the positioning of the layout (yellow line). Finally, the two highest LOCO-EFA modes have incomplete contributions, given any cutoﬀin the number of EFA modes. (B–E) The contribution λ+0 is not simply an oﬀset of the contour, but also involves a kidney bean-shaped distortion, more pronounced for larger contributions. Mode λ+1 (the circular mode) is shown in blue; mode λ+0 in red; their summation in green; and the predicted shape if the contribution were purely an oﬀset in magenta. (B) λ+0 being 10% of λ+1, eﬀectively resulting in just an oﬀset to the shape outline (C). (D, E) λ+0 being 30% of λ+1, resulting in a clear kidney bean-shaped distortion of the shape. 42 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint C A B C 1 2 3 4 6 8 12 16 Original Reconstruction C A B C 1 2 3 4 6 8 12 16 Original Reconstruction XOR 1 2 3 4 6 8 12 16 Figure S5: XOR measure extracted from the LOCO-EFA coeﬃcients to provide additional information regarding the cell shape complexity. (A) Original cell contour. (B) LOCO-EFA reconstruction taking the ﬁrst n Ln modes into account, as indicated below the panels. (C) Determination of the level of mismatch between the original cell shape and its nth order truncated LOCO-EFA approximation, by applying an XOR (Exclusive OR) function. C A C Original B A CC 1 2 3 4 6 8 12 16 Figure S5: XOR measure extracted from the LOCO-EFA coeﬃcients to provide additional information regarding the cell shape complexity. (A) Original cell contour. (B) LOCO-EFA reconstruction taking the ﬁrst n Ln modes into account, as indicated below the panels. Supporting Figures (C) Determination of the level of mismatch between the original cell shape and its nth order truncated LOCO-EFA approximation, by applying an XOR (Exclusive OR) function. A B C D E F G A B C D E F G A B C D E F G H I 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 2 3 4 5 6 7 8 9 101112131415 0 0.5 1 1.5 2 EFA mode Pn 1 2 3 4 5 6 7 8 9 101112131415 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 EFA mode XOR Difference 1 2 3 4 5 6 7 8 9 101112131415 −0.01 0 0.01 0.02 0.03 0.04 0.05 0.06 EFA mode Marginal Difference cd entropy 0 0.5 1 1.5 2 2.5 3 1 2 3 4 5 6 7 0 5 2 3 4 5 6 7 8 9 101112131415 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 Ln mode Ln 1 2 3 4 5 6 7 8 9 101112131415 0 0.1 0.2 0.3 0.4 Ln mode XOR Difference 1 2 3 4 5 6 7 8 9 101112131415 −0.05 0 0.05 0.1 0.15 Ln mode Marginal Difference cd entropy 0 0.5 1 1.5 2 1 2 3 4 5 6 7 0 0.5 1 Figure S6: Additional example of LOCO-EFA metrics on a cell changing its shape over time. (A–G) Sequence of a tracked pavement cell growing over time with normalised area. (H) Pn, XOR diﬀerence and marginal diﬀerence proﬁles, cumulative diﬀerence and entropy using EFA. (I) Ln, XOR diﬀerence and marginal diﬀerence proﬁles, cumulative diﬀerence and entropy using LOCO-EFA. See also the ﬁrst example presented in Figure 4 in the main text. Supporting Figures A B C D E F G A B C D E F G H 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 2 3 4 5 6 7 8 9 101112131415 0 0.5 1 1.5 2 EFA mode Pn 1 2 3 4 5 6 7 8 9 101112131415 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 EFA mode XOR Difference 1 2 3 4 5 6 7 8 9 101112131415 −0.01 0 0.01 0.02 0.03 0.04 0.05 0.06 EFA mode Marginal Difference cd entropy 0 0.5 1 1.5 2 2.5 3 1 2 3 4 5 6 7 0 5 A B C D E F G H 2 3 4 5 6 7 8 9 101112131415 0 0.5 1 1.5 2 EFA mode Pn 1 2 3 4 5 6 7 0 5 H EFA mode EFA mode 1 2 3 4 5 6 7 8 9 101112131415 0 0.1 0.2 0.3 0.4 Ln mode XOR Difference 1 2 3 4 5 6 7 8 9 101112131415 −0.05 0 0.05 0.1 0.15 Ln mode Marginal Difference A B C D E F G I EFA mode 2 3 4 5 6 7 8 9 101112131415 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 Ln mode Ln 1 2 3 4 5 6 7 0 0.5 1 EFA mode 1 2 3 4 5 6 7 8 9 101112131415 0 0.1 0.2 0.3 0.4 Ln mode XOR Difference cd entropy 0 0.5 1 1.5 2 Figure S6: Additional example of LOCO-EFA metrics on a cell changing its shape over time. (A–G) Sequence of a tracked pavement cell growing over time with normalised area. (H) Pn, XOR diﬀerence and marginal diﬀerence proﬁles, cumulative diﬀerence and entropy using EFA. (I) Ln, XOR diﬀerence and marginal diﬀerence proﬁles, cumulative diﬀerence and entropy using LOCO-EFA. See also the ﬁrst example presented in Figure 4 in the main text. 43 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. Supporting Figures These eﬀects are present regardless of the number of speciﬁed lobes or the combination of parameters used, but becomes less pronounced at higher lobe numbers See also Figure 5 2 3 4 5 6 7 8 9 10 11 12 2 4 6 8 10 −0.1 0 0.1 SCS2 Ln mode <Ln>- Ln L specified shape comb2 B 2 3 4 5 6 7 8 9 10 11 12 2 4 6 8 10 −0.1 0 0.1 2 3 4 5 6 7 8 9 10 11 12 2 4 6 8 10 −0.1 0 0.1 2 3 4 5 6 7 8 9 10 11 12 2 4 6 8 10 −0.2 0 0.2 SCS1 SCS2 SCS3 Number of lobes=03 Number of lobes=04 Number of lobes=05 Number of lobes=06 Number of lobes=07 Number of lobes=08 Number of lobes=09 Number of lobes=10 Ln mode <Ln>- Ln <Ln>- Ln <Ln>- Ln Ln mode Ln mode specified shape comb1 specified shape comb2 specified shape comb3 Number of lobes=03 Number of lobes=04 N b f l b 05 A B C 2 3 4 5 6 7 8 9 10 11 12 2 4 6 8 10 −0.1 0 0.1 2 3 4 5 6 7 8 9 10 11 12 2 4 6 8 10 −0.1 0 0.1 SCS2 SCS3 Ln mode <Ln>- Ln <Ln>- Ln Ln mode specified shape comb2 specified shape comb3 B C 2 3 4 5 6 7 8 9 10 11 12 2 4 6 8 10 −0.1 0 0.1 Ln n Ln mode specified shape comb3 C B C A 2 3 4 5 6 7 8 9 10 11 12 2 4 6 8 10 −0.2 0 0.2 2 3 4 5 6 7 8 9 10 11 12 2 4 6 8 10 −0.2 0 0.2 specified shape comb3 specified shape comb2 Ln mode Ln mode <XOR> - XOR <XOR> - XOR E F 2 3 4 5 6 7 8 9 10 11 12 2 4 6 8 10 −0.2 0 0.2 specified shape comb2 Ln mode <XOR> - XOR E 2 3 4 5 6 7 8 9 10 11 12 2 4 6 8 10 −0.2 0 0.2 specified shape comb3 Ln mode <XOR> - XOR F E F D Ln mode 2 3 4 5 6 7 8 9 10 11 12 2 4 6 8 10 −0.2 0 0.2 2 3 4 5 6 7 8 9 10 11 12 2 4 6 8 10 −0.4 −0.2 0 0.2 specified shape comb3 specified shape comb2 Ln mode Ln mode <Md> - Md <Md> - Md H I 2 3 4 5 6 7 8 9 10 11 12 2 4 6 8 10 −0.2 0 0.2 specified shape comb3 Ln mode <Md> - Md I I G H 8 Ln mode 1 Ln mode Figure S8: Divergence between speciﬁed cell shapes and resultant population- level cell shape diversity within conﬂuent in silico cell populations. Supporting Figures ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint A B C D E F G H Q R S T U V W X M J K L N O P I SCS1 SCS2 SCS3 Figure S7 44 A B F E G H E H G L I J SC K J K P O M N N P M O R SC Q Q R U U V X W Figure S7 44 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint Figure S7: Conﬂuent in silico cell populations, simulated with three types of speciﬁed cell shapes (SCSs) and diﬀerent number of speciﬁed lobes. Paramet- ers for the diﬀerent speciﬁed cell shapes are given in Table S1. (A–H) Cells with large protrusions (Speciﬁed Cell Shapes 1, SCS1), with speciﬁed lobe number increasing from 3 (A) to 10 (H); (I–P) Cells with small protrusions (SCS2), with speciﬁed lobe number increasing from 3 (I) to 10 (P); (Q–X) Elongated cells (SCS3), with speciﬁed lobe num- ber increasing from 3 (Q) to 10 (X). Speciﬁed cell shapes, as resulting from single-cell simulations, are shown above each panel. Cells are randomly coloured. See also Figure 5 and Figure S8. 45 45 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. Supporting Figures ; https://doi.org/10.1101/157842 doi: bioRxiv preprint 2 3 4 5 6 7 8 9 10 11 12 2 4 6 8 10 −0.2 0 0.2 2 3 4 5 6 7 8 9 10 11 12 2 4 6 8 10 −0.4 −0.2 0 0.2 2 3 4 5 6 7 8 9 10 11 12 2 4 6 8 10 −0.5 0 0.5 2 3 4 5 6 7 8 9 10 11 12 2 4 6 8 10 −0.2 0 0.2 2 3 4 5 6 7 8 9 10 11 12 2 4 6 8 10 −0.2 0 0.2 2 3 4 5 6 7 8 9 10 11 12 2 4 6 8 10 −0.5 0 0.5 2 3 4 5 6 7 8 9 10 11 12 2 4 6 8 10 −0.1 0 0.1 2 3 4 5 6 7 8 9 10 11 12 2 4 6 8 10 −0.1 0 0.1 2 3 4 5 6 7 8 9 10 11 12 2 4 6 8 10 −0.2 0 0.2 SCS1 SCS2 SCS3 Number of lobes=03 Number of lobes=04 Number of lobes=05 Number of lobes=06 Number of lobes=07 Number of lobes=08 Number of lobes=09 Number of lobes=10 specified shape comb3 Ln mode Ln mode <Ln>- Ln <Ln>- Ln <Ln>- Ln Ln mode Ln mode specified shape comb1 specified shape comb2 specified shape comb3 specified shape comb1 specified shape comb2 specified shape comb3 specified shape comb2 specified shape comb1 Number of lobes=03 Number of lobes=04 Number of lobes=05 Number of lobes=06 Number of lobes=07 Number of lobes=08 Number of lobes=09 Number of lobes=10 Ln mode Ln mode Ln mode Ln mode Ln mode <XOR> - XOR <XOR> - XOR <XOR> - XOR Number of lobes=03 Number of lobes=04 Number of lobes=05 Number of lobes=06 Number of lobes=07 Number of lobes=08 Number of lobes=09 Number of lobes=10 <Md> - Md <Md> - Md <Md> - Md A D B C E G F H I Figure S8: Divergence between speciﬁed cell shapes and resultant population- level cell shape diversity within conﬂuent in silico cell populations. The mis- match is visualised as the diﬀerence in LOCO-EFA-derived measures between the average for conﬂuent population simulations and its value for single cell simulations. Three types of speciﬁed cell shapes (SCSs) were simulated, as indicated for each column, and paramet- rised in Table S1. Supporting Figures Lobe numbers vary from 3 to 10, as indicated for each row and depicted in Figure S7. (A–C) Diﬀerence between the average Ln (⟨Ln⟩) values of the cells of the simulated populations and that of a single simulated cell. (D–F) Diﬀerence between aver- age XOR (⟨XOR⟩) of the cells of the simulated population and that of a single simulated cell. (G–I) Same as in (A–C, D–F), but now for marginal diﬀerence (Md). In all cases the speciﬁed cell shape becomes less pronounced in the conﬂuent population simulations (indicated by negative values that all measures yield at the given speciﬁed lobe number), while the cells present an increased shape diversity and complexity (seen by the broad ﬂanking regions with positive values in the proﬁles, indicating a large range of modes that contribute to the shapes). Supporting Figures The mis- match is visualised as the diﬀerence in LOCO-EFA-derived measures between the average for conﬂuent population simulations and its value for single cell simulations. Three types of speciﬁed cell shapes (SCSs) were simulated, as indicated for each column, and paramet- rised in Table S1. Lobe numbers vary from 3 to 10, as indicated for each row and depicted in Figure S7. (A–C) Diﬀerence between the average Ln (⟨Ln⟩) values of the cells of the simulated populations and that of a single simulated cell. (D–F) Diﬀerence between aver- age XOR (⟨XOR⟩) of the cells of the simulated population and that of a single simulated cell. (G–I) Same as in (A–C, D–F), but now for marginal diﬀerence (Md). In all cases the speciﬁed cell shape becomes less pronounced in the conﬂuent population simulations (indicated by negative values that all measures yield at the given speciﬁed lobe number), while the cells present an increased shape diversity and complexity (seen by the broad ﬂanking regions with positive values in the proﬁles, indicating a large range of modes that contribute to the shapes). These eﬀects are present regardless of the number of speciﬁed lobes or the combination of parameters used, but becomes less pronounced at higher lobe numbers. See also Figure 5. 46 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint A B C D E A A A B C D E Figure S9 47 B C D E Figure S9 B C D E B C D E B C B D E E Figure S9 47 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . Supporting Figures CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint Figure S9: Temporal and spatial transformations required to calculate the pre- cise contribution of each EFA mode. (A) The blue ellipse depicts an elliptic har- monic, given by [X(t)] = [A] [M(t)], while the grey circle depicts the unit circle, given by [X(t)] = [M(t)]. The temporal angle φ is the scaled time required to move along an elliptic harmonic from the starting point at τ = 0 to one extreme along the semi-major axis (green ﬁlled circles). The angle φ cannot be trivially derived from the spatial po- sition at τ = 0, requiring ﬁrst an eﬀective projection upon the unit circle (green open circles). The spatial angle θ is the inclination of the elliptic harmonic. After apply- ing the appropriate spatial and temporal transformations to the EFA coeﬃcients using this geometrical interpretation, the semi-major and semi-minor axis, λ1 and λ2, can be straightforwardly obtained. To eliminate multiple representations of the same outline, the starting point of the ﬁrst harmonic is speciﬁcally positioned at the extreme along the semi-major axis which lies in either quadrant I or II, given by temporal angle τ ⋆ 1 (Equation 17). In contrast, all other temporal angles φn used, while also positioned along the semi-major axis, are not conﬁned to quadrant I or II. (B–E) Visual interpretation of Equation 23 as a step-wise construction of the elliptic harmonic. The matrix [M(t)] corresponds to the unit circle (B). For each mode, ﬁrst the starting point relative to the semi-major axis is correctly positioned (C); then the original circle is transformed into an ellipse, its semi-major axis along the x-axis and semi-minor axis along the y-axis (D); and ﬁnally the ellipse is rotated to its correct position (E). See further details in the Supplementary Materials and Methods. 48 48 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . Supporting Figures (J) L3 = 0, 0.2, 0, −5 12π , in green; (K) (L3 = 0.15, 0, 3 4π, 0 , in red; (L) L3 = 0.15, 0.2, 3 4π, −5 12π , in orange. Panel (J) and (K) also illustrate how the initial phase shifts with respect to the amp- litude contribution of the negative and positive rotor depend on ζ+1, and on ζ−l and ζ+l, respectively. They are given by ζ+1 + ζ−l for the negative rotor and by ζ+l −ζ+1 for the positive rotor. Figure S10: Impact of LOCO-EFA starting points ζ+l and ζ−l on the amp- litude of the reconstructed shape. All panels show L3 = (λ+3, λ−3, ζ+3, ζ−3) su- perimposed on L1 = (1, 0, ζ+1, 0), i.e., the unit circle. In (A–F), L1 = (1, 0, 0, 0). (A) Amplitude a as a function of phase ω (as depicted in I) for the negative rotor only (L3 = (0, 0.15, 0, 0)), shown in green. For several phases also the rotor (light blue) itself is drawn. Note that by plotting the rotor in the (ω, a) plane, the circles are slightly deformed. (B) Same for the positive rotor only (L3 = (0.15, 0, 0, 0)), in red. (C) Same for both rotors superimposed (L3 = (0.15, 0.15, 0, 0)), in orange. (D–F) Alike (A–C), but for an out-of-phase starting angle of the negative rotor. (D) L3 = (0, 0.15, 0, π), in green; (E) L3 = (0.15, 0, 0, 0), in red; (F) L3 = (0.15, 0.15, 0, π), in orange. In (G–L), L1 = 1, 0, 2 3π, 0 . (G) Pattern generated by the negative ro- tor only (L3 = 0, 0.1, 0, −5 12π ), shown in green, compared to the equation a−l = λ−l cos (lω + ζ−l −(l −1)ζ+1) (Equation 44b), in dark blue. They present a close match. (H) Same for the positive rotor only (L3 = 0.075, 0, 3 4π, 0 ), shown in red, compared to a+l = λ+l cos (lω + ζ+l −(l + 1)ζ+1) (Equation 44a), show in dark blue, again presenting a close match. Supporting Figures CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint G H I K J L 0 45 90 135 180 225 270 315 360 -0.2 0 0.2 A 0 45 90 135 180 225 270 315 360 -0.2 0 0.2 B 0 45 90 135 180 225 270 315 360 C -0.4 0.4 0 45 90 135 180 225 270 315 360 -0.2 0 0.2 D 0 45 90 135 180 225 270 315 360 -0.2 0 0.2 E 0 45 90 135 180 225 270 315 360 -0.4 0.4 F 0 0 (degrees) (degrees) (degrees) (degrees) (degrees) (degrees) Figure S10 . CC-BY 4.0 International license available under a 0 45 90 135 180 225 270 315 360 -0.2 0 0.2 A 0 45 90 135 180 225 270 315 360 -0.2 0 0.2 B 0 45 90 135 180 225 270 315 360 C -0.4 0.4 0 45 90 135 180 225 270 315 360 -0.2 0 0.2 D 0 45 90 135 180 225 270 315 360 -0.2 0 0.2 E 0 45 90 135 180 225 270 315 360 -0.4 0.4 F 0 0 (degrees) (degrees) (degrees) (degrees) (degrees) (degrees) 0 45 90 135 180 225 270 315 360 -0.2 0 0.2 A 0 45 90 135 180 225 270 315 360 -0.2 0 0.2 B 0 45 90 135 180 225 270 315 360 C -0.4 0.4 D 0 (degrees) (degrees) (degrees) A B C D E F G H I K J L (degrees) Figure S10 H I G L Figure S10 49 49 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. Supporting Figures ; https://doi.org/10.1101/157842 doi: bioRxiv preprint Figure S10: Impact of LOCO-EFA starting points ζ+l and ζ−l on the amp- litude of the reconstructed shape. All panels show L3 = (λ+3, λ−3, ζ+3, ζ−3) su- perimposed on L1 = (1, 0, ζ+1, 0), i.e., the unit circle. In (A–F), L1 = (1, 0, 0, 0). (A) Amplitude a as a function of phase ω (as depicted in I) for the negative rotor only (L3 = (0, 0.15, 0, 0)), shown in green. For several phases also the rotor (light blue) itself is drawn. Note that by plotting the rotor in the (ω, a) plane, the circles are slightly deformed. (B) Same for the positive rotor only (L3 = (0.15, 0, 0, 0)), in red. (C) Same for both rotors superimposed (L3 = (0.15, 0.15, 0, 0)), in orange. (D–F) Alike (A–C), but for an out-of-phase starting angle of the negative rotor. (D) L3 = (0, 0.15, 0, π), in green; (E) L3 = (0.15, 0, 0, 0), in red; (F) L3 = (0.15, 0.15, 0, π), in orange. In (G–L), L1 = 1, 0, 2 3π, 0 . (G) Pattern generated by the negative ro- tor only (L3 = 0, 0.1, 0, −5 12π ), shown in green, compared to the equation a−l = λ−l cos (lω + ζ−l −(l −1)ζ+1) (Equation 44b), in dark blue. They present a close match. (H) Same for the positive rotor only (L3 = 0.075, 0, 3 4π, 0 ), shown in red, compared to a+l = λ+l cos (lω + ζ+l −(l + 1)ζ+1) (Equation 44a), show in dark blue, again presenting a close match. (I) Same for both rotors superimposed (L3 = 0.075, 0.1, 3 4π, −5 12π ), shown in orange, and al = p λ2 +l + λ2 −l + 2λ+lλ−l cos (ζ+l −ζ−l −2ζ+1) cos (lω + ζl) (Equation 45), show in dark blue, again presenting a close match. Panel (G) also illus- trates that at peak amplitude the phase of the main and subrotor are equal; panel (I) also illustrates the concepts amplitude (a); peak amplitude (A); phase (ω); and phase at peak amplitude (Ω). (J–L) Alike (G–I), but for larger amplitude. . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint Supporting Figures (I) Same for both rotors superimposed (L3 = 0.075, 0.1, 3 4π, −5 12π ), shown in orange, and al = p λ2 +l + λ2 −l + 2λ+lλ−l cos (ζ+l −ζ−l −2ζ+1) cos (lω + ζl) (Equation 45), show in dark blue, again presenting a close match. Panel (G) also illus- trates that at peak amplitude the phase of the main and subrotor are equal; panel (I) also illustrates the concepts amplitude (a); peak amplitude (A); phase (ω); and phase at peak amplitude (Ω). (J–L) Alike (G–I), but for larger amplitude. (J) L3 = 0, 0.2, 0, −5 12π , in green; (K) (L3 = 0.15, 0, 3 4π, 0 , in red; (L) L3 = 0.15, 0.2, 3 4π, −5 12π , in orange. Panel (J) and (K) also illustrate how the initial phase shifts with respect to the amp- litude contribution of the negative and positive rotor depend on ζ+1, and on ζ−l and ζ+l, respectively. They are given by ζ+1 + ζ−l for the negative rotor and by ζ+l −ζ+1 for the positive rotor. 50 50 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. Supporting Figures (A) The cell outline of an experimentally observed cell was mirrored along the y-axis and/or rotated over diﬀerent angles, after which LOCO-EFA was applied to each image separately. (B) The Ln values and other derived metrics were invariant to those transformations. (C) The resolution of the original image was reduced, such that the number of contour points decreased from 1104 to 253, 108 and 27 (from left to right, respectively). (D) The Ln numbers and the associated metrics only deviated from the high-resolution values when the resolution was very low and the cell outline was clearly deviating from original cell outline. This in contrast to the skeletonisation method, for which even marginal resolution reductions can cause large deviations in the outcome. Same image with different resolution 0 1 2 3 0 1 2 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 C D Same image with d 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 2 3 Same image with different resolution Same image with different resolution 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 Same image with different resolution C D Same image with different resolution 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 C D 0 1 2 3 0 1 2 3 C D 0 1 2 3 4 5 6 7 8 9 101112131415 0 0.2 0.4 0.6 0.8 1 2 3 4 5 6 7 8 9 101112131415 0 0.5 1 1.5 2 Ln mode XOR Difference 0 0.2 0.4 Marginal Difference Ln mode Ln D 1 2 3 4 5 6 7 8 9 101112131415 0 0.2 0.4 Ln mode Marginal Difference 0 0 Marginal Difference Figure S11: LOCO-EFA is invariant to image rotation or mirroring, nor sens- itive to image resolution. (A) The cell outline of an experimentally observed cell was mirrored along the y-axis and/or rotated over diﬀerent angles, after which LOCO-EFA was applied to each image separately. (B) The Ln values and other derived metrics were invariant to those transformations. Supporting Figures ; https://doi.org/10.1101/157842 doi: bioRxiv preprint 0 1 2 3 4 5 6 7 8 9 101112131415 0 0.2 0.4 0.6 0.8 1 2 3 4 5 6 7 8 9 101112131415 0 0.5 1 1.5 2 Ln mode XOR Difference 1 2 3 4 5 6 7 8 9 101112131415 0 0.2 0.4 Ln mode Marginal Difference 0 1 2 3 4 5 6 7 8 9 101112131415 0 0.2 0.4 0.6 0.8 Ln mode Ln Ln mode Ln 1 2 3 4 5 6 7 8 9 101112131415 0 0.5 1 1.5 Ln mode XOR Difference 1 2 3 4 5 6 7 8 9 101112131415 0 0.1 0.2 0.3 0.4 Ln mode Marginal Difference A B C D Same image with different rotations and inversion Same image with different resolution 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 Figure S11: LOCO-EFA is invariant to image rotation or mirroring, nor sens- itive to image resolution. (A) The cell outline of an experimentally observed cell was mirrored along the y-axis and/or rotated over diﬀerent angles, after which LOCO-EFA was applied to each image separately. (B) The Ln values and other derived metrics were invariant to those transformations. (C) The resolution of the original image was reduced, such that the number of contour points decreased from 1104 to 253, 108 and 27 (from left to right, respectively). (D) The Ln numbers and the associated metrics only deviated from the high-resolution values when the resolution was very low and the cell outline was clearly deviating from original cell outline. Supporting Figures This in contrast to the skeletonisation method for which even marginal resolution reductions can cause large deviations in the 0 1 2 3 4 5 6 7 8 9 101112131415 0 0.2 0.4 0.6 0.8 Ln mode Ln 1 2 3 4 5 6 7 8 9 101112131415 0 0.5 1 1.5 Ln mode XOR Difference 1 2 3 4 5 6 7 8 9 101112131415 0 0.1 0.2 0.3 0.4 Ln mode Marginal Difference A B Same image with different rotations and inversion 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 A B Same image with different rotations and inversion 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 Same image with different rotations and inversion erent rotations and inversion 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 A B 0 1 2 3 4 5 6 7 8 9 101112131415 0 0.2 0.4 0.6 0.8 Ln mode Ln 1 2 3 4 5 6 7 8 9 101112131415 0 0.5 1 1.5 Ln mode XOR Difference 1 2 3 4 5 6 7 8 9 101112131415 0 0.1 0.2 0.3 0.4 Ln mode Marginal Difference B 0 1 2 3 4 5 6 7 8 9 101112131415 0 0.2 0.4 0.6 0.8 1 2 3 4 5 6 7 8 9 101112131415 0 0.5 1 1.5 2 Ln mode XOR Difference 1 2 3 4 5 6 7 8 9 101112131415 0 0.2 0.4 Ln mode Marginal Difference 0 1 2 3 4 5 6 7 8 9 101112131415 0 0.2 0.4 Ln mode Ln Ln mode Ln 1 2 3 4 5 6 7 8 9 101112131415 0 0.5 Ln mode XOR Diff 1 2 3 4 5 6 7 8 9 101112131415 0 0.1 0.2 Ln mode Marginal D C D Same image with different resolution 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 Figure S11: LOCO-EFA is invariant to image rotation or mirroring, nor sens- itive to image resolution. Supporting Figures (C) The resolution of the original image was reduced, such that the number of contour points decreased from 1104 to 253, 108 and 27 (from left to right, respectively). (D) The Ln numbers and the associated metrics only deviated from the high-resolution values when the resolution was very low and the cell outline was clearly deviating from original cell outline. This in contrast to the skeletonisation method, for which even marginal resolution reductions can cause large deviations in the outcome. 51 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint Supporting Table Table S1: Three types of speciﬁed cell shapes (SCS) as used in Figure S7 and Figure S8. Table S1: Three types of speciﬁed cell shapes (SCS) as used in Figure S7 and Figure S8. Table S1: Three types of speciﬁed cell shapes (SCS) as used in Figure S7 and Figure S8. SCS1 SCS2 SCS3 Target Area 858 1167 1197 Pointedness 6912 5328 5207 Number of lobes 3–10 3–10 3–10 Roundness 382 518 434 Elongation 4 28 5927 SCS1 SCS2 SCS3 Target Area 858 1167 1197 Pointedness 6912 5328 5207 Number of lobes 3–10 3–10 3–10 Roundness 382 518 434 Elongation 4 28 5927 52 52 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint Supporting Movies Movie S1: Approximation of a closed contour using Elliptic Fourier Analysis. A given two-dimensional shape can be approximated using EFA by summing n elliptic harmonics as follows: each nth elliptic harmonic traces n clockwise or counter-clockwise revolutions while moving around the previous elliptic harmonic. Movie S2: Direction of rotation opposite to the ﬁrst harmonic ellipse. EFA mode 3 generating a shape with 4 features, its rotation direction opposite to the rotation direction of the ﬁrst harmonic. Movie S3: Direction of rotation same as the ﬁrst elliptic harmonic. EFA mode 3 generating a shape with 2 features, its rotation direction the same as the rotation direction of the ﬁrst harmonic. Movie S4: Exception of the rule regarding rotation direction. If the eccentricity of an elliptical harmonic is very high, the number of generated lobes does not follow the rule-of-thumb proposed by Diaz et al. (1990). Here, EFA mode 3 generates a shape with 4 features, although its rotation direction is the same as the rotation direction of the ﬁrst harmonic. Movie S5: Example of an in silico simulation of a population of cells with more complex speciﬁed shapes. CPM simulation of multi-lobed cells take up pavement-like cell shapes when they are allowed to interact with their neighbours within a conﬂuent tissue. 53 53 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint Supporting Code loco-efa example.c Code written in C which applies the full procedure to calculate the Ll and Ll values to any ﬁle containing contour coordinates. For each computational step it is indicated where the relevant mathematical details can be found in the Supplementary Materials and Methods. We have intentionally kept the code as bare as possible, without for example any graphical interface, to allow it to be trivially compiled and run on any platform. Details regarding compilation and execution can be found in the header of the ﬁle. loco-efa example.c Code written in C which applies the full procedure to calculate the Ll and Ll values to any ﬁle containing contour coordinates. For each computational step it is indicated where the relevant mathematical details can be found in the Supplementary Materials and Methods. We have intentionally kept the code as bare as possible, without for example any graphical interface, to allow it to be trivially compiled and run on any platform. Details regarding compilation and execution can be found in the header of the ﬁle. cell outline.csv Contour data of the cell presented in Figure 3S, Figure S3A and Figure S11. This ﬁle, or any other ﬁle containing contour data, can be used as an input for the program. cell outline.csv Contour data of the cell presented in Figure 3S, Figure S3A and Figure S11. This ﬁle, or any other ﬁle containing contour data, can be used as an input for the program. 54 54 . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint . CC-BY 4.0 International license available under a not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which was this version posted June 30, 2017. ; https://doi.org/10.1101/157842 doi: bioRxiv preprint Supporting References Pincus Z, Theriot JA (2007) Comparison of quantitative methods for cell-shape analysis. J Microsc 227:140–156. doi:10.1111/j.1365-2818.2007.01799.x. Pincus Z, Theriot JA (2007) Comparison of quantitative methods for cell-shape analysis. J Microsc 227:140–156. doi:10.1111/j.1365-2818.2007.01799.x. Schmittbuhl M, Allenbach B, Le Minor JM, Schaaf A (2003) Elliptical descriptors: some simpliﬁed morphometric parameters for the quantiﬁcation of complex outlines. Math Geol 35:853–871. Kuhl FP, Giardina CR (1982) Elliptic Fourier features of a closed contour. Comput Gr Image Process 18:236–258. Diaz G, Quacci D, Dell’Orbo C (1990) Recognition of cell surface modulation by elliptic Fourier analysis. Comput Methods Programs Biomed 31:57–62. Yoshioka Y, Iwata H, Ohsawa R, Ninomiya S (2005) Quantitative evaluation of the petal shape variation in Primula sieboldii caused by breeding process in the last 300 years. Heredity (Edinb) 94:657–663. doi:10.1038/sj.hdy.6800678. Frieß M, Baylac M (2003) Exploring artiﬁcial cranial deformation using elliptic Four- ier analysis of Procrustes aligned outlines. Am J Phys Anthropol 122:11–22. doi: 10.1002/ajpa.10286. / Neto JC, Meyer GE, Jones DD, Samal AK (2006) Plant species identiﬁcation using El- liptic Fourier leaf shape analysis. Comput Electron Agric 50:121–134. Iwata H, Niikura S, Matsuura S, Takano Y, Ukai Y (1998) Evaluation of variation of root shape of Japanese radish (Raphanus sativus L.) based on image analysis using elliptic Fourier descriptors. Euphytica 102:143–149. Iwata H, Ebana K, Uga Y, Hayashi T, Jannink JL (2010) Genome-wide association study of grain shape variation among Oryza sativa L. germplasms based on elliptic Fourier analysis. Mol Breeding 25:203–215. Chitwood DH, Kumar R, Headland LR, Ranjan A, Covington MF, et al. (2013) A quantit- ative genetic basis for leaf morphology in a set of precisely deﬁned tomato introgression lines. Plant Cell 25:2465–2481. doi:10.1105/tpc.113.112391. / Haines AJ, Crampton JS (2000) Improvements to the method of Fourier shape analysis as applied in morphometric studies. Palaeontology 43:765–783. 55 55
https://openalex.org/W4380145461	https://zenodo.org/records/8023509/files/PRIMO%20Cristofanelli.pdf	Italian	null	La psicoterapia di gruppo negli stati mentali a rischio: una ipotesi di trattamento	Zenodo (CERN European Organization for Nuclear Research)	2,023	cc-by	5,413	Pubblicato: febbraio 2023 doi: 10.32069/pj.2021.2.182 ABSTRACT IN ITALIANO Direttore scientifico Raffaele Sperandeo Supervisore scientifico Valeria Cioffi Journal manager Enrico Moretto Contatta l’autore Silvia Cristofanelli cristofanelli@yahoo.it Ricevuto: 19 gennaio 2023 Accettato: 23 febbraio 2023 Pubblicato: 23 febbraio 2023 Direttore scientifico Raffaele Sperandeo Supervisore scientifico Valeria Cioffi Journal manager Enrico Moretto Contatta l’autore Silvia Cristofanelli cristofanelli@yahoo.it Ricevuto: 19 gennaio 2023 Accettato: 23 febbraio 2023 Pubblicato: 23 febbraio 2023 L’autore sostiene che intervenire precocemente sugli stati mentali a rischio, che si manifestano in età giovanile con disturbi che pregiudicano il funzionamento globale della persona, sia di fondamentale importanza al fine di evitare che il quadro clinico si aggravi ulteriormente. Intervenire sulla fase pro- dromica significa intercettare quei quadri clinici che rientrano nella definizione di disturbi emotivi comuni (DEC), molto diffusi nei Centri di Salute Mentale in questo momento post-pandemico. La psicoterapia di gruppo come modalità di intervento privilegiata, focalizza l’attenzione sulla dimensione dell’identità sociale ed ha come obiettivo ripristinare la percezione di appartenenza ad un contesto con il valore personale che ne consegue. KEYWORDS KEYWORDS Common emotional disorders; group cognitive-behavioural psychotherapy; intervention techniques. ABSTRACT The author claims that early intervention on at-risk mental states, which occur at a young age with di- sorders affecting the overall functioning of the person, is of fundamental importance in order to prevent the clinical picture from worsening further. Intervening on the prodromic phase means intercepting those clinical pictures that fall within the definition of common emotional disorders (DEC), which are widespread in the Mental Health Centres at this post-pandemic time. Group psychotherapy as a mode of privileged intervention, focusses attention on the dimension of social identity and aims to restore the perception of belonging to a context with the resulting personal value. Citation Cristofanelli S. (2023). La psicoterapia di gruppo negli stati mentali a rischio: una ipotesi di trattamento Phenomena Journal, 5, 1-12. https://doi.org/10.32069/pj.2021.2.182 Citation Cristofanelli S. (2023). La psicoterapia di gruppo negli stati mentali a rischio: una ipotesi di trattamento Phenomena Journal, 5, 1-12. https://doi.org/10.32069/pj.2021.2.182 IPOTESI E METODI DI STUDIO La psicoterapia di gruppo negli stati mentali a rischio: una ipotesi di trattamento IPOTESI E METODI DI STUDIO IPOTESI E METODI DI STUDIO La psicoterapia di gruppo negli stati mentali a rischio: una ipotesi di trattamento Silvia Cristofanelli1 1 Dipartimento di Salute Mentale ASL Roma 6 1. Introduzione Secondo Yung e McGorry [1] è ritenuto oramai opportuno intervenire sugli esordi della sintomatologia degli “stati mentali a rischio”, ARMS (at Risk Mental State), che si manifestano in: - cambiamenti comportamentali - cambiamenti emozionali - cambiamenti cognitivi Nelle linee guida introdotte dal PANMS (Piano di Azioni Nazionale per la Salute Mentale, 24 Gennaio 2013), vengono definiti i criteri operativi che ogni dipartimento di salute mentale deve considerare. Si sottolinea la necessità di adottare una meto- dologia incentrata sui percorsi di cura, che consideri la necessità di istituire un mo- dello specifico e differenziato sulla base dei bisogni di salute delle persone, trasversale e indipendentemente dal luogo di residenza, denominato Percorso Dia- gnostico Terapeutico Assistenziale (PDTA). Barlow e colleghi [2] sostengono che se i PDTA hanno prevalentemente impegnato in passato nei centri di salute mentale i disturbi gravi e cronici, tanto più ora in questo complesso momento di emergenza sanitaria post-pandemica, che ha visto un aumento del 50% di una serie di disturbi che non possono essere definiti in senso categoriale, i cosiddetti disturbi emotivi comuni (DEC), è necessario pensare a percorsi di cura coerenti e finalizzati a rispondere a specifici bisogni di salute, sulla base delle evi- denze scientifiche, come appunto i PDTA. Nel lavoro qui presentato, in una ottica di prevenzione del disagio psichiatrico grave e di intervento precoce, abbiamo avvertito la necessità di dare una risposta coerente e finalizzata a tutte quelle problematiche ascrivibili ai DEC che presentano un quadro eterogeneo, come stati depressivi, disturbi d’ansia e fobie sociali, disturbi somato- formi e condizioni di disagio esistenziale, problematiche che possono essere prodro- miche di un disagio psichiatrico grave, se non precocemente trattate. Il modello dei sintomi di base a cui ci si riferisce, è quello di Shultze-Lutter e Klosterkotter [3] che punta l’attenzione sui deficit lievi (substrate close) del funzionamento mentale, in particolare nella fase prodromica che precede lo sviluppo di un quadro psicopatolo- gico più conclamato, in cui il funzionamento sociale e attitudinale appare già com- promesso e pregiudicato. Secondo gli autori, è quindi necessario individuare la fase prodromica in uno stadio precoce, per prevenire non solo la psicosi od altri disturbi psicopatologici seri, ma anche il difetto funzionale correlato. Il modello psicotera- peutico di riferimento è quello della psicoterapia cognitiva post-razionalista, che pone le sue basi a partire dal padre fondatore Vittorio Guidano. PAROLE CHIAVE Disturbi emotivi comuni; psicoterapia cognitivo-comportamentale di gruppo; tecniche d‘intervento. pag. 1 Phenomena Journal \| www.phenomenajournal.it Gennaio-Giugno 2023 \| Volume 5 \| Numero 1 \| Ipotesi e metodi di studio Gennaio-Giugno 2023 \| Volume 5 \| Numero 1 \| Ipotesi e metodi di studio Gennaio-Giugno 2023 \| Volume 5 \| Numero 1 \| Ipotesi e metodi di studio pag. 1 Cristofanelli La psicoterapia di gruppo negli stati mentali a rischio: una ipotesi di trattamento Phenomena Journal \| www.phenomenajournal.it pag. 2 Phenomena Journal \| www.phenomenajournal.it Gennaio-Giugno 2023 \| Volume 5 \| Numero 1 \| Ipotesi e metodi di studio 1. Introduzione Nell’orientamento post- razionalista è posta grande enfasi sul funzionamento autoreferenziale del sé e sulla sua fondamentale unitarietà. Guidano [4] sottolinea la processualità e la dimensio- nalità nell’esprimersi delle diverse Organizzazioni di Significato Personale (OSP), che si dispiegano lungo un continuum, dalla normalità all’area nevrotica, fino a quella psicotica. L’intero ciclo di vita si muove attraverso il dispiegarsi di una propria trama narrativa interna, in cui lo stile di attaccamento e il proprio stile affettivo, sono la Gennaio-Giugno 2023 \| Volume 5 \| Numero 1 \| Ipotesi e metodi di studio pag. 2 Cristofanelli La psicoterapia di gruppo negli stati mentali a rischio: una ipotesi di trattamento cornice entro cui si struttura l’esperienza. Nell’ottica della processualità il modello teorico del post-razionalismo, fa riferimento al Modello Dinamico Maturativo (DMM) sull’attaccamento di Patricia Crittenden [5]. (DMM) sull attaccamento di Patricia Crittenden [5]. Il Modello Dinamico Maturativo, è più incline a disporre i modelli di attaccamento atipici lungo un continuum di integrazione del sé. Secondo la Crittenden gli aspetti fondamentali del funzionamento mentale, si muovono attraverso una continua rior- ganizzazione dell’esperienza e avvengono mediante la ricerca delle regolarità e delle discrepanze, tra ciò che era atteso e ciò che è avvenuto. Il dispiegarsi del proprio stile di attaccamento, determina il modo in cui strutturiamo le nostre relazioni affettive e si muove attraverso queste due costanti, che intervengono a determinare la “qualità” del proprio assetto esistenziale dell’intero ciclo di vita. La strategia terapeutica del modello post-razionalista, sempre secondo Guidano [6] si focalizza sulla progressiva consapevolezza della dialettica e dell’interfaccia tra l’esperienza immediata (Io), che sperimenta e agisce e l’immagine cosciente di sé (Me), che osserva e valuta. Lavorare in questa ottica, significa riconoscere come autoreferenziali i propri significati per- sonali di costruzione dell’esperienza, che nel modello post-razionalista prendono il nome di Organizzazioni di Significato Personale (OSP). Significa legittimare le di- screpanze ed i meccanismi di autoinganno che intervengono a mantenere il senso di sé in corso, talvolta in modo disfunzionale e maladattativo, così da favorire la co- struzione di una trama narrativa del sé più articolata e complessa, che è il fine ultimo del percorso terapeutico. Tuttavia nel progettare una psicoterapia di gruppo, che è il modello di intervento che abbiamo privilegiato, è necessario tenere presenti alcune variabili che possono interferire con il percorso terapeutico, variabili che nella nostra esperienza abbiamo riscontrato essere sempre presenti. Phenomena Journal \| www.phenomenajournal.it 1. Introduzione La difficoltà ad esporsi in gruppo connessa al timore del giudizio e di una possibile critica, la vergogna del sen- tirsi inadeguati e comunque non all’altezza, in particolare nella fase conoscitiva in cui il clima di appartenenza non si è ancora consolidato, sono elementi che possono interferire ed ostacolare la costruzione di un clima caldo ed empatico, che è la pre- messa per la costruzione di ogni gruppo. Inoltre, talvolta è necessario comprimere i tempi di esposizione di ognuno e focalizzare l’attenzione del gruppo sulla condivi- sione dei vissuti emotivi e cognitivi, stimolando l’esplorazione sulla risonanza emo- tiva e sul confronto con l’altro, che è il processo attraverso cui è possibile consolidare il senso di individuazione e di differenziazione del sé. Ciò può arrecare malcontento e delusione in alcuni, in particolare tra coloro che ne- cessitano e che richiedono una maggiore visibilità, tanto da innescare comportamenti oppositivi e di sfida, che se non adeguatamente fatti rientrare, possono determinare talvolta acting-out con abbandono repentino del percorso di cura. È necessario che la strategia terapeutica consideri queste variabili che inevitabilmente appartengono alla storia dei contesti umani e delle relazioni, cercando di favorire il più possibile un clima espositivo di totale accettazione dell’altro, libero dal giudizio ed empatico, tanto più nei confronti di chi invece desidera, nonostante tutto, distruggerlo. È utile in questi casi, affiancare un percorso di consultazione psicologica individuale, che rafforzi l’autostima ed integri gli aspetti disfunzionali e rigidi dei modelli operativi Gennaio-Giugno 2023 \| Volume 5 \| Numero 1 \| Ipotesi e metodi di studio pag. 3 Cristofanelli La psicoterapia di gruppo negli stati mentali a rischio: una ipotesi di trattamento interni, mitigando l’irrompere di aspetti disregolativi di emozioni disturbanti. Il mo- dello di trattamento qui proposto, ha cercato di considerare tutto questo, favorendo l’integrazione dei vissuti emotivi disturbanti attraverso la legittimazione ed il rico- noscimento del dolore condiviso in un contesto caldo ed accogliente, dove è possibile sentirsi comunque accolti. Phenomena Journal \| www.phenomenajournal.it 2. Terapia di gruppo: razionale della scelta Abbiamo ritenuto opportuno adottare la psicoterapia di gruppo per questa tipologia di problematiche psichiche, in quanto riteniamo necessario privilegiare una modalità di esplorazione dei vissuti esperienziali attraverso la graduale condivisione con gli altri partecipanti. Attraverso la dialogicità [7] intesa come una disposizione verso l’altro, fondata sulla reciprocità e sul riconoscimento dell’intersoggettività, è possibile riconoscere il proprio percorso individuale, che ha portato talvolta, all’attivazione e alla stabilizzazione di determinati assetti emozionali e cognitivi disfunzionali. Inoltre, secondo il modello elaborato da Frith [8] l’intervento basato sul gruppo mira a migliorare i processi di cognizione sociale, in quanto le problematiche psicopato- logiche connesse al ritiro sociale, convergerebbero in un deficit neuropsicologico a carico dei processi di mentalizzazione e di socializzazione, da cui deriva la difficoltà ad integrare in schemi conoscitivi più articolati e complessi, il significato ed il valore della relazione con l’altro. Infine scegliere un percorso terapeutico di gruppo, rende possibile costruire una cornice entro cui sperimentare la relazione con l’altro e raf- forzare il senso della propria identità sociale, che come dice Tajfel [9] è quella parte del concetto di sé di un individuo che deriva dalla consapevolezza di appartenere ad un gruppo, unito al valore emotivo attribuito a tale appartenenza. Abbiamo ritenuto opportuno adottare la psicoterapia di gruppo per questa tipologia di problematiche psichiche, in quanto riteniamo necessario privilegiare una modalità di esplorazione dei vissuti esperienziali attraverso la graduale condivisione con gli altri partecipanti. Attraverso la dialogicità [7] intesa come una disposizione verso l’altro, fondata sulla reciprocità e sul riconoscimento dell’intersoggettività, è possibile riconoscere il proprio percorso individuale, che ha portato talvolta, all’attivazione e alla stabilizzazione di determinati assetti emozionali e cognitivi disfunzionali. Inoltre, secondo il modello elaborato da Frith [8] l’intervento basato sul gruppo mira a migliorare i processi di cognizione sociale, in quanto le problematiche psicopato- logiche connesse al ritiro sociale, convergerebbero in un deficit neuropsicologico a carico dei processi di mentalizzazione e di socializzazione, da cui deriva la difficoltà ad integrare in schemi conoscitivi più articolati e complessi, il significato ed il valore della relazione con l’altro. Infine scegliere un percorso terapeutico di gruppo, rende possibile costruire una cornice entro cui sperimentare la relazione con l’altro e raf- forzare il senso della propria identità sociale, che come dice Tajfel [9] è quella parte del concetto di sé di un individuo che deriva dalla consapevolezza di appartenere ad un gruppo, unito al valore emotivo attribuito a tale appartenenza. 4.1 Caratteristiche del gruppo I partecipanti al gruppo sono stati di età compresa tra i 18 anni e i 24 anni, per la prima volta afferenti ad una struttura psichiatrica, non abusatori di sostanze od alcool in maniera abituale e continuativa, con Q.I. maggiore di 70 e problematiche ascrivibili ai DEC. La frequenza al gruppo è stata di una seduta a settimana, della durata di 2 ore ciascuna, per un totale di 24 incontri; la conduzione del gruppo è stata affidata a due co-terapeuti. Sono stati sottoposti al trattamento 8 pazienti di cui 3 maschi e 5 femmine. È Sono stati sottoposti al trattamento 8 pazienti di cui 3 maschi e 5 femmine. È stata garantita ad ognuno dei partecipanti l’opportunità di affiancare alla terapia di gruppo una terapia individuale, sollecitando la partecipazione al gruppo con una mo- dalità attiva ed interattiva, rispettando i tempi di ognuno per l’esposizione “in vivo”. 3. Obiettivi e finalità Costruire una cornice entro cui sperimentare il senso di appartenenza ad un contesto sociale entro cui è possibile sentirsi accolti, che ha come obiettivo rafforzare il senso della propria identità sociale, deve considerare tuttavia alcuni passaggi fondamen- tali. Le finalità dell’intervento hanno come esiti i seguenti traguardi: ❏miglioramento delle disfunzioni meta cognitive ❏miglioramento degli schemi interpersonali rigidi e coartati Gennaio-Giugno 2023 \| Volume 5 \| Numero 1 \| Ipotesi e metodi di studio pag. 4 Cristofanelli La psicoterapia di gruppo negli stati mentali a rischio: una ipotesi di trattamento ❏miglioramento della tendenza all’inibizione e all’appiattimento emotivo ❏miglioramento delle capacità di decentramento cognitivo ❏miglioramento della tendenza all’inibizione e all’appiattimento emotivo ❏miglioramento della tendenza all’inibizione e all’appiattimento emotivo ❏miglioramento delle capacità di decentramento cognitivo ❏miglioramento delle capacità di decentramento cognitivo ❏miglioramento delle capacità di decentramento cognitivo Phenomena Journal \| www.phenomenajournal.it pag. 5 Phenomena Journal \| www.phenomenajournal.it Gennaio-Giugno 2023 \| Volume 5 \| Numero 1 \| Ipotesi e metodi di studio 4.3 Consultazione psicologica individuale Successivamente alla fase diagnostica, intercettati i partecipanti al gruppo, è seguita una fase di consultazione psicologica individuale di 4/5 incontri. La fase di consul- tazione psicologica ha avuto la finalità di costruire una interpretazione dei sintomi a partire da un modello condiviso, che ha definito la cornice entro cui selezionare e adattare gli interventi, promuovere il cambiamento e prevenire le ricadute. La fase di consultazione segue il modello post-razionalista, che secondo Guidano [12] con- verge sulla comprensione del proprio stile di costruzione dell’esperienza e si articola nei seguenti passaggi: • Focalizzazione e riordinamento dell’esperienza immediata. • Differenziazione tra spiegazione ed esperienza immediata (passare dal perché… al come). • Riconoscere i propri ed altrui pensieri ed emozioni. • Favorire il passaggio di una spiegazione del sintomo da “esternalizzata” ad “internalizzata”. 4.2 Assessment L’assessment ha avuto come finalità quella di condividere fin dalle prime fasi un “patto di cura” coerente e finalizzato, entro cui tutti sono coinvolti nel raggiungimento di un traguardo ed ognuno dei partecipanti in qualche modo ne è stato il protagonista. La fase di valutazione ha previsto i seguenti passaggi: La fase di valutazione ha previsto i seguenti passaggi: ❏prima visita psichiatrica e psicologica ❏screening psicodiagnostico Nella valutazione psicodiagnostica abbiamo ritenuto opportuno considerare la valu- tazione del Funzionamento dei Livelli di Personalità, il livello di compromissione delle funzioni cognitive coinvolte e l’indice del Quoziente emozionale. Di seguito sono riportati i test somministrati all’inizio del trattamento (t0) e a con- clusione del trattamento (t1), al fine di monitorare l’andamento della sintomatologia e verificare l’effettivo miglioramento del quadro generale. ❏Spi-a (Schizofrenia Proneness Instrument- Adult) ❏SCL-90-R (Sympton Checklist-90-R) ❏EQ- i (Emotional Quotient Inventory) ❏Scala del Funzionamento dei Livelli di Personalità- DSM-5 La scelta di somministrare la Spi-a, è nata dall’esigenza di misurare la presenza e l’intensità dei cosiddetti “sintomi di base”, che sottostanno molto spesso ai DEC. L’esigenza di valutare i sintomi di base, al fine di intervenire precocemente prima che la sintomatologia degeneri in un quadro più complesso, prende origine dall’idea di Huber e Gross [10] e successivamente da Schultze-Lutter e Addington [11] che ne hanno elaborato lo strumento. Abbiamo pensato che utilizzare la Symptom Chec- Gennaio-Giugno 2023 \| Volume 5 \| Numero 1 \| Ipotesi e metodi di studio pag. 5 Cristofanelli La psicoterapia di gruppo negli stati mentali a rischio: una ipotesi di trattamento klist-90-R potesse offrire dei vantaggi, in quanto oltre ad essere uno strumento am- piamente validato e di facile somministrazione in ambito clinico per la valutazione della gravità dei sintomi del disagio psichico, può fornire una misura di outcome in psicoterapia. La scelta dell’EQ-i si è basata sulla peculiarità diagnostica dello stru- mento, in quanto consente di valutare variabili fondamentali come la competenza emotiva sociale e personale dell’individuo e permette di individuare cinque scale principali, (intrapersonale, interpersonale, adattabilità, gestione dello stress e umore generale) e quindici sottoscale, che forniscono un punteggio totale del quoziente emotivo. Inoltre la valutazione del Funzionamento dei Livelli di Personalità, ha po- tuto fornire indicazioni utili sia rispetto all’area del Sé che rispetto all’area Interper- sonale, delineando il grado di compromissione dell’identità e della capacità di autodeterminazione, oltre che della capacità di empatia e del desiderio di intimità. Phenomena Journal \| www.phenomenajournal.it pag. 6 Phenomena Journal \| www.phenomenajournal.it Gennaio-Giugno 2023 \| Volume 5 \| Numero 1 \| Ipotesi e metodi di studio ❏ Attività bilivellare, (distinguere tra l’immergersi nell’emozione e monitorare ciò che sta accadendo). ❏ Attività bilivellare, (distinguere tra l’immergersi nell’emozione e monitorare ciò che sta accadendo). ❏ Attività bilivellare, (distinguere tra l’immergersi nell’emozione e monitorare ciò che sta accadendo). ❏ Depotenziamento (condurre l’esplorazione dell’emozione temuta Successivamente, allorchè la capacità di esplorazione diventava più fluida e meno coartata, la terapia ha focalizzato l’attenzione sulla consapevolezza del proprio stile di attaccamento e di relazione affettiva. Attraverso i ricordi dei vissuti emozionali ed esperienziali connessi ai propri legami genitoriali, è stato possibile riordinare e ri- collocare in una trama narrativa del sé più articolata e complessa, scene di vita scar- samente integrate. Questa fase della terapia durata 4 incontri è stata di fondamentale importanza, in quanto ha reso possibile riconoscere emozioni e contenuti ideativi di- sturbanti, ricollocandoli in esperienze di vita passata e all’interno di una cornice af- fettivo-relazionale. Secondo il modello del “social referencing” di Fonagy e Target [15] la terapia si è centrata sul riconoscimento degli stili devianti genitoriali di ri- specchiamento. Ciò ha reso possibile accedere al dolore di disconferme e di rifiuti attraverso il ricordo di scene nucleari prototipiche. La fase centrale dell’intero per- corso terapeutico, durata 8 incontri, si è focalizzata sull’esposizione in vivo dei propri episodi critici, così come spontaneamente si andavano dispiegando nella quotidianità di ognuno. Questa fase della terapia, resa possibile dalla maggiore capacità di men- talizzazione e di differenziazione, ha consentito il consolidamento di un senso di identità più stabile e più differenziato. Come dice Guidano [16] riconoscere i processi che orientano il senso della propria esistenza e del proprio agire, riconoscere i mec- canismi di autoinganno e le discrepanze che intervengono come processi “taciti” nel mantenere il senso di una propria coerenza interna spesso disfunzionale e maladat- tativa, favorisce il processo di integrazione del sé verso un livello di consapevolezza più articolato e complesso. Le tecniche utilizzate in questa fase, ove ognuno dei par- tecipanti riusciva ad esprimere più liberamente il proprio disagio attraverso l’espo- sizione diretta, si sono intervallate a tecniche di esposizione più liberatorie e giocose come il role-playing, attraverso cui “mettere in scena” situazioni critiche nel ruolo interscambiabile di attore protagonista e di osservatore neutro. Abbiamo potuto ri- scontrare che intervallare le tecniche di esposizione in questa fase della terapia, ha avuto la finalità di risvegliare l’adesione al trattamento. Il dolore messo in scena, po- teva essere esperito e contenuto da un intero gruppo che poteva farsene carico. Phenomena Journal \| www.phenomenajournal.it ❏ Periferizzazione, (l’attenzione è posta su accadimenti che presentano gradienti emozionali accessibili). ❏ Periferizzazione, (l’attenzione è posta su accadimenti che presentano gradienti emozionali accessibili). 5. Strategia terapeutica di gruppo La strategia terapeutica nella prima fase di 4 incontri circa, ha cercato di favorire un primo livello di mentalizzazione, che secondo il modello di Fonagy [13] è quel pro- cesso attraverso cui apprendiamo che la nostra esperienza del mondo è mediata dal fatto che possediamo una mente, quindi intrinsecamente legata allo sviluppo del sé e all’elaborazione graduale della sua organizzazione. Per questa finalità si sono utilizzate tecniche immaginative, che potessero consentire un accesso graduale e neutro ai propri contenuti e costruire l’adesione al trattamento. La fase successiva di ulteriori 4 incontri, ha mantenuto un livello di esposizione gra- duale, attraverso l’utilizzo delle tecniche sotto elencate, che secondo il modello post- razionalista [14] potessero consentire un primo livello di internalizzazione e di riconoscimento del proprio stile di costruzione dell’esperienza: ❏Tecniche immaginative, (come il sogno guidato). ❏Tecniche immaginative, (come il sogno guidato). Gennaio-Giugno 2023 \| Volume 5 \| Numero 1 \| Ipotesi e metodi di studio pag. 6 Cristofanelli La psicoterapia di gruppo negli stati mentali a rischio: una ipotesi di trattamento ❏ Periferizzazione, (l’attenzione è posta su accadimenti che presentano gradienti emozionali accessibili). pag. 7 Phenomena Journal \| www.phenomenajournal.it Gennaio-Giugno 2023 \| Volume 5 \| Numero 1 \| Ipotesi e metodi di studio ❏ Attività bilivellare, (distinguere tra l’immergersi nell’emozione e monitorare ciò che sta accadendo). Se- condo Castefranchi [17] l’esperienza dell’appartenenza ad un gruppo è importante per tutti gli individui in funzione della costruzione della propria identità. Così l’espe- rienza del sentire di appartenere ad un contesto umano da cui potersi sentire nutriti, ha ulteriormente rafforzato in ognuno dei partecipanti il senso della propria identità, imprigionato dentro schemi rigidi e disfunzionali. L’ultima fase della terapia di 4 in- contri si è centrata sul consolidamento del riconoscimento precoce del disagio. La percezione di inadeguatezza poteva essere integrata ed associata ai pensieri disfun- Gennaio-Giugno 2023 \| Volume 5 \| Numero 1 \| Ipotesi e metodi di studio pag. 7 Cristofanelli La psicoterapia di gruppo negli stati mentali a rischio: una ipotesi di trattamento zionali ed automatici, “quando gli altri mi conosceranno scopriranno chi sono”. Il distacco ed il senso di estraneità veniva associato a “è inutile, non capisco gli altri e gli altri non capiscono me”. Il senso di costrizione e la perdita di controllo, “sono costretto a tollerare la presenza degli altri”. zionali ed automatici, “quando gli altri mi conosceranno scopriranno chi sono”. Il distacco ed il senso di estraneità veniva associato a “è inutile, non capisco gli altri e gli altri non capiscono me”. Il senso di costrizione e la perdita di controllo, “sono costretto a tollerare la presenza degli altri”. 6. Analisi dei dati L’esiguità del campione e l’assenza di un gruppo di controllo hanno consentito solo un’analisi descrittiva dei risultati, con confronto tra le medie del gruppo ad inizio trattamento (t0) e a fine trattamento (t1) ed analisi dei casi singoli. Phenomena Journal \| www.phenomenajournal.it Fig. 1: confronto medie indici globali SCL 90-R (t0) 7. Risultati Il confronto delle medie ottenuto da ognuno dei partecipanti al test SCL-90-R tra gli indici globali ad inizio trattamento (v.Fig.1), ha rilevato punteggi che si collocano ad un livello di gravità moderato (da 55 a 65 puntiT) per quanto riguarda l’indice GSI (Global Severity Index), indicatore globale del livello di intensità e di profondità del disagio. Mentre abbiamo riscontrato un livello di gravità elevato (da 65 a 75 pun- tiT) per quanto riguarda l’indice PSDI (Positive Sympton Distress Index), che rap- presenta l’intensità dei sintomi esperita dal soggetto. A fine trattamento i punteggi al test SCL-90-R (v.Fig.2), hanno evidenziato una so- stanziale riduzione dei punteggi dell’indice globale PSDI, che è risultato essere l’in- dice più significativo tra tutti, collocandosi ad un livello di intensità da moderato ed in alcuni casi ad un livello di normalità (da 45 a 55 punti T). Ciò a conferma che il percorso terapeutico aveva consolidato in ognuno dei partecipanti, una maggiore ca- pacità di accedere ed integrare i propri vissuti dolorosi senza restarne sopraffatto. Fig. 1: confronto medie indici globali SCL 90-R (t0) A fine trattamento i punteggi al test SCL-90-R (v.Fig.2), hanno evidenziato una so- stanziale riduzione dei punteggi dell’indice globale PSDI, che è risultato essere l’in- dice più significativo tra tutti, collocandosi ad un livello di intensità da moderato ed in alcuni casi ad un livello di normalità (da 45 a 55 punti T). Ciò a conferma che il percorso terapeutico aveva consolidato in ognuno dei partecipanti, una maggiore ca- pacità di accedere ed integrare i propri vissuti dolorosi senza restarne sopraffatto. A fine trattamento i punteggi al test SCL-90-R (v.Fig.2), hanno evidenziato una so- stanziale riduzione dei punteggi dell’indice globale PSDI, che è risultato essere l’in- dice più significativo tra tutti, collocandosi ad un livello di intensità da moderato ed in alcuni casi ad un livello di normalità (da 45 a 55 punti T). Ciò a conferma che il percorso terapeutico aveva consolidato in ognuno dei partecipanti, una maggiore ca- pacità di accedere ed integrare i propri vissuti dolorosi senza restarne sopraffatto. Gennaio-Giugno 2023 \| Volume 5 \| Numero 1 \| Ipotesi e metodi di studio pag. 8 Cristofanelli La psicoterapia di gruppo negli stati mentali a rischio: una ipotesi di trattamento Fig. 2: confronto medie indici globali SCL 90-R (t1) Fig. Phenomena Journal \| www.phenomenajournal.it Fig. 2: confronto medie indici globali SCL 90-R (t1) 7. Risultati 2: confronto medie indici globali SCL 90-R (t1) Al test EQ-i ad inizio trattamento (v.Fig.3), i punteggi si collocano ad un Quoziente emotivo Totale compreso tra un livello basso (da 70 a 90 puntiT) ad un livello molto basso (da 50 a 70 puntiT), indicatore questo di una compromissione generale in tutte e 4 le scale principali qui elencate: ❏ Intrapersonale: consapevolezza di sé e delle proprie capacità nell’espressione e comprensione dei propri stati d’animo, pensieri e idee. ❏ Interpersonale: capacità di comprendere ed apprezzare i sentimenti altrui instaurando e mantenendo relazioni interpersonali responsabili e soddisfacenti. ❏ Adattabilità: capacità di verificare oggettivamente i propri stati d’animo e quelli degli altri in funzione delle richieste dell’ambiente esterno, mostrando flessibilità nel gestire e modificare emozioni e pensieri al cambiare delle situazioni. ❏ Gestione dello Stress: capacità di gestire le situazioni stressanti tenendo sotto controllo emozioni ed impulsi. ❏ Umore Generale: capacità di essere ottimisti, saper provare ed esprimere sentimenti positivi e saper godere della presenza degli altri. Anche la media delle sottoscale si colloca ad un punteggio da basso a molto basso ed è composta da alcune dimensioni riferite ad aspetti diversi di ogni singola scala principale. Gennaio-Giugno 2023 \| Volume 5 \| Numero 1 \| Ipotesi e metodi di studio pag. 9 Cristofanelli La psicoterapia di gruppo negli stati mentali a rischio: una ipotesi di trattamento A fine trattamento i punteggi al test EQ-i si collocano ad un livello medio (da 90 a 110 punti T), sia per quanto riguarda il Quoziente emotivo Totale sia rispetto alla media delle sottoscale (v.Fig.4). Fig. 3: confronto Q.em. e M. scale (t0) (t0) A fine trattamento i punteggi al test EQ-i si collocano ad un livello medio (da 90 a 110 punti T), sia per quanto riguarda il Quoziente emotivo Totale sia rispetto alla media delle sottoscale (v.Fig.4). Fig. 4: confronto Q.em. e M. scale (t1) Ciò a conferma dell’ipotesi di partenza, che un percorso terapeutico di gruppo ha consolidato non solo la consapevolezza di sé, ma ha accresciuto la capacità di men- talizzazione di sé e dell’altro, deficitaria e compromessa in partenza. Attraverso il percorso terapeutico, ampliando la comprensione dei propri modelli di costruzione dell’esperienza e differenziandosi dagli altri, è stato possibile conoscere modalità re- lazionali intime ed accoglienti, che hanno restituito la dignità ad una immagine di sé incerta e svalorizzata da esperienze affettive pregresse. Phenomena Journal \| www.phenomenajournal.it Fig. 3: confronto Q.em. e M. scale (t0) 7. Risultati La valutazione espressa dai Gennaio-Giugno 2023 \| Volume 5 \| Numero 1 \| Ipotesi e metodi di studio pag. 10 Cristofanelli La psicoterapia di gruppo negli stati mentali a rischio: una ipotesi di trattamento partecipanti ad inizio trattamento circa il funzionamento della loro vita, Scala del Funzionamento dei Livelli di Personalità DSM-5, ha riscontrato in 6 di loro un livello di disfunzione da moderato (liv.2) a grave (liv.3) in particolare nell’area interperso- nale, che descrive la capacità di empatia e la capacità di costruire e mantenere rela- zioni intime. Nei restanti partecipanti non si sono riscontrate differenze significative sia rispetto all’area del sè sia rispetto all’area interpersonale. Infine, la valutazione fornita dalla Scala Spi-a, ha individuato 3 partecipanti positivi ai criteri codgis (Co- gnitive Disturbances), che sono i sintomi di base maggiormente predittivi di un pro- babile esordio psicotico nell’arco dei prossimi sei mesi se non trattato. La positività ai criteri codgis, prevede la presenza di almeno due dei nove sintomi di base sotto elencati, con un valore maggiore o uguale a 3, che equivale ad un livello di intensità moderata (sintomi presenti più volte a settimana) negli ultimi tre mesi. I punteggi ai sintomi di base codgis vanno da un range di 0 (sintomi assenti) ad un massimo di 6, che equivale ad un livello di intensità estrema (sintomi presenti quotidianamente e continuamente). Criteri per codgis: ❏ Incapacità di suddividere l’attenzione ❏ Interferenza di pensieri emotivamente neutri ❏ Affollamento, pressione dei pensieri, fuga delle idee ❏ Blocco del flusso del pensiero ❏ Disturbo recettivo del linguaggio ❏ Disturbo espressivo del linguaggio ❏ Tendenza all’autoreferenzialità, “ego-centrismo” ❏ Disturbo della comprensione simbolica ❏ Tendenza a fissare l’attenzione su dettagli percettivi Phenomena Journal \| www.phenomenajournal.it Phenomena Journal \| www.phenomenajournal.it BIBLIOGRAFIA 1. Yung, A. R., & McGorry, P.D. (1996). The prodromal phase of first-episode psychosis: Past and current conceptualizations. Schizofrenia Bulletin, 22(2), 353-370. https://doi.org/10.1093/schbul/22.2.353 https://doi.org/10.1093/schbul/22.2.353 2. Barlow, D. H. et al., Cavalletti, V. (a cura di) (2021). Il protocollo unificato per il trattamento tran- sdiagnostico dei disturbi emotivi. Milano: Franco Angeli. 3. Scultze-Lutter F., Klosterkotter, J., et al. (2007). Predicting firts-episode psychosis by basic sympton criteria.Clinical Neuropsychiatry: Journal of treatment Evaluation, 4(1), 11-22. 4. Guidano, V.F. (1987). La complessità del sé. Torino: Bollati Boringhieri Arnkil, T.E., 5. Crittenden, P.M. (1994). Nuove prospettive sull’attaccamento. Milano: Guerini Studio. 5. Crittenden, P.M. (1994). Nuove prospettive sull attaccamento. Milano: Guerini Studio. 6. Guidano, V.F., (1992). Il Sè nel suo divenire: verso una terapia cognitiva post-razionalista. Milano: Franco Angeli. 6. Guidano, V.F., (1992). Il Sè nel suo divenire: verso una terapia cognitiva post-razionalista. Milan Franco Angeli. 7. Seikkula, J. (2012). Metodi dialogici nel lavoro di rete. Trento: Erickson. 8. Frith, C. (1995). Neuropsicologia cognitiva della schizofrenia. Milano: Ra 8. Frith, C. (1995). Neuropsicologia cognitiva della schizofrenia. Milano: Raffaello Cortina Editore. 9. Tajfel, H. (1999). Gruppi umani e categorie sociali. Bologna: il Mulino. 9. Tajfel, H. (1999). Gruppi umani e categorie sociali. Bologna: il Mulino. j ( ) pp g g 10. Huber, G., Gros,s G. (2002). La psicopatologia di Kurt Schneider. Roma: La Feltrinelli. 11. Schultze-Lutter, F., Addington, J. and Rurhmann, S. (2016). Schizofrenia Proneness instrument adult version (SPI-A). Roma: Giovanni Fioriti Editore Srl. 12. Guidano, V. F., (2007). Psicoterapia cognitiva post-razionalista. Milano: Franco Angeli. 12. Guidano, V. F., (2007). Psicoterapia cognitiva post-razionalista. Milano: Franco Angeli. 13. Fonagy, P., Gergely G., Jurist E.L. (2004). Affect Regulation, Mentalization and the Developme 13. Fonagy, P., Gergely G., Jurist E.L. (2004). Affect Regulation, Mentalization and the Developm of the Self. Editore: Karnac Books. f f 14. Guidano, V.F., Cutolo G. (a cura di) (2008). La psicoterapia tra arte e scienza. Milano: Franco Angeli. 14. Guidano, V.F., Cutolo G. (a cura di) (2008). La psicoterapia tra arte e scienza. Milano: Franco Angeli. 15. Fonagy, P., Target M. (2001). Attaccamento e funzione riflessiva. Milano: Raffaello Cortina Edi- tore. 15. Fonagy, P., Target M. (2001). Attaccamento e funzione riflessiva. Milano: Raffaello Cortina Edi- tore. 16. Nardi, B. (2001). Processi psichici e psicopatologia nell’approccio cognitivo. Milano: Franco An- geli. 16. Nardi, B. (2001). Processi psichici e psicopatologia nell’approccio cognitivo. Milano: Franco An- geli. g 17. Castelfranchi, C. (2005). Che figura. 8. Conclusioni Il protocollo adottato ha prodotto sostanzialmente una riduzione della sintomatologia attraverso un percorso di gruppo finalizzato ad una maggiore consapevolezza delle proprie emozioni, oltre che attraverso una maggiore capacità di esplorazione dei pro- pri stati interni. Contestualmente la modalità gruppale ha favorito in tutti i parteci- panti, il recupero del desiderio di reciprocità e di intimità con l’altro, connesso al valore personale che ne consegue in termini di maggiore autostima e rafforzamento dell’immagine di sé. Gennaio-Giugno 2023 \| Volume 5 \| Numero 1 \| Ipotesi e metodi di studio pag. 11 Cristofanelli La psicoterapia di gruppo negli stati mentali a rischio: una ipotesi di trattamento BIBLIOGRAFIA BIBLIOGRAFIA Emozioni e immagine sociale. Bologna: Il Mulino. pag. 12 Gennaio-Giugno 2023 \| Volume 5 \| Numero 1 \| Ipotesi e metodi di studio
https://openalex.org/W4362609171	https://aacr.figshare.com/articles/journal_contribution/Supplementary_Figure_Legend_from_Treatment_with_Gefitinib_or_Lapatinib_Induces_Drug_Resistance_through_Downregulation_of_Topoisomerase_II_Expression/22498317/1/files/39956694.pdf	English	null	Supplementary Figure Legend from Treatment with Gefitinib or Lapatinib Induces Drug Resistance through Downregulation of Topoisomerase IIα Expression	null	2,023	cc-by	427	Supplementary Figure 1 Continuous treatment with gefitinib for 48 hours significantly inhibits the induction of γH2AX foci in response to doxorubicin, etoposide and m-AMSA but not IR. SK-Br-3 cells were treated with DFM or gefitinib (5 µM) for one or 48 hours, with gefitinib replacement at 23 and 47 hours. γ-H2AX foci induction was determined in response to (A) doxorubicin (0.5 µM), (B) etoposide (15 µM), (C) m-AMSA (0.5 µM) and (D) IR (2 Gy) either alone or following gefitinib treatment for one or 48 hours, at the time point indicated. Data are presented as the mean±SEM of three independent experiments. p<0.05 compared with cells treated with doxorubicin, etoposide, m-AMSA or IR alone. Supplementary Figure 1 Continuous treatment with gefitinib for 48 hours significantly inhibits the induction of γH2AX foci in response to doxorubicin, etoposide and m-AMSA but not IR. SK-Br-3 cells were treated with DFM or gefitinib (5 µM) for one or 48 hours, with gefitinib replacement at 23 and 47 hours. γ-H2AX foci induction was determined in response to (A) doxorubicin (0.5 µM), (B) etoposide (15 µM), (C) m-AMSA (0.5 µM) and (D) IR (2 Gy) either alone or following gefitinib treatment for one or 48 hours, at the time point indicated. Data are presented as the mean±SEM of three independent experiments. p<0.05 compared with cells treated with doxorubicin, etoposide, m-AMSA or IR alone. Supplementary Figure 1 Continuous treatment with gefitinib for 48 hours significantly inhibits the induction of γH2AX foci in response to doxorubicin, etoposide and m-AMSA but not IR. SK-Br-3 cells were treated with DFM or gefitinib (5 µM) for one or 48 hours, with gefitinib replacement at 23 and 47 hours. γ-H2AX foci induction was determined in response to (A) doxorubicin (0.5 µM), (B) etoposide (15 µM), (C) m-AMSA (0.5 µM) and (D) IR (2 Gy) either alone or following gefitinib treatment for one or 48 hours, at the time point indicated. Data are presented as the mean±SEM of three independent experiments. p<0.05 compared with cells treated with doxorubicin, etoposide, m-AMSA or IR alone. Supplementary Figure 2 Lapatinib also induce a G1 cell cycle arrest after 48 hours of treatment. SK-Br-3 cells were treated with DFM or gefitinib (5 µM) or lapatinib (2µM) for 48 hours, with TKI replacement every 24 hours. The effect on the cell cycle was investigated by collecting and fixing cells at 48 hours and staining DNA with propidium iodide. FACS was used to analyse the DNA content of 10, 000 cells and data are presented as the mean±SEM of three independent experiments. p<0.05 compared with untreated cells.
https://openalex.org/W2288015904	https://www.annalsofgeophysics.eu/index.php/annals/article/download/6714/6552	English	null	Characteristics of local earthquake seismograms of varying dislocation sources in a stratified upper crust and modeling for P and S velocity structure: comparison with observations in the Koyna-Warna region, India	Annals of geophysics	2,016	cc-by	11,842	ABSTRACT quake seismograms. Record sections displaying the seismograms in the reduced time versus epicentral dis- tance frame, similar to those obtained in seismic refrac- tion proﬁling, can be assembled for an average source depth using the processed seismograms of earthquakes of nearly equal source depths with similar source mech- anisms and recorded in a narrow azimuth range. Verti- cal component seismogram sections, assembled from a state-of-the-art Koyna-Warna digital seismograph net- work, illuminating the upper crustal structure demon- strate this approach. The Koyna-Warna seismic region located in the Western Ghats, in the south-western part of the Deccan Volcanic Province (DVP), in western India, reveals proliﬁc seismicity conﬁned to an area of 20×30 km2 and mostly originating in the upper crustal depths up to ~11 km in the north and 6-8 km in the south [Srinagesh and Sarma 2005]. Figure 1a shows the location map with the epicenters of some well-located earthquakes in the Koyna-Warna seismic region [Rai et al. 1999]. The seismicity pattern in this region starts as a single seismic tract in the north trending NE-SW, fur- ther branching into two distinct zones with NW-SE trend in the south (see inset in Figure 1a). Figure 1b shows the depth distribution of seismicity in these three zones [Rai et al. 1999]. Vertical component record sections of local earthquake seismograms from a state-of-the-art Koyna-Warna digital seismograph network are assem- bled in the reduced time versus epicentral distance frame, similar to those obtained in seismic refraction proﬁling. The record sections obtained for an average source depth display the processed seismograms from nearly equal source depths with similar source mechanisms and recorded in a narrow azimuth range, illuminating the upper crustal P and S velocity structure in the region. Further, the seismogram characteristics of the local earthquake sources are found to vary signiﬁcantly for different source mechanisms and the amplitude variations exceed those due to velocity model stratiﬁcation. In the present study a large number of reﬂectivity synthetic seismograms are obtained in near offset ranges for a stratiﬁed upper crustal model having sharp discontinuities with 7%-10% velocity contrasts. The synthetics are obtained for different source regimes (e.g., strike-slip, normal, reverse) and different sets of source parameters (strike, dip, and rake) within each regime. Seismogram sections with dominantly strike-slip mechanism are found to be clearly favorable in revealing the ve- locity stratiﬁcation for both P and S waves. ANNALS OF GEOPHYSICS, 58, 6, 2015, S0656; doi:10.4401/ag-6714 ANNALS OF GEOPHYSICS, 58, 6, 2015, S0656; doi:10.4401/ag-6714 ABSTRACT In contrast the seismogram sections for earthquakes of other source mechanisms seem to display the upper crustal P phases poorly with low amplitudes even in presence of sharp discontinuities of high velocity contrasts. The observed seismogram sections illustrated here for the earthquake sources with strike-slip and normal mechanisms from the Koyna-Warna seismic region substantiate these ﬁndings. Travel times and reﬂectivity synthetic seismograms are used for 1-D modeling of the observed virtual source local earthquake seis- mogram sections and inferring the upper crustal velocity structure in the Koyna-Warna region. Signiﬁcantly, the inferred upper crustal velocity model in the region reproduces the synthetic seismograms comparable to the observed sections for earthquake sources with differing mechanisms in the Koyna and Warna regions. The Koyna-Warna seismic activity has been at- tributed to both regional tectonism [Wadia 1968, Kr- ishna et al. 1973] and to reservoir-triggered seismicity [Gupta and Rastogi 1974, Gupta et al. 1980, Gupta 1992, Talwani 1997, Gupta 2002, Gupta 2005, Srinagesh and Sarma 2005, Shashidhar et al. 2011]. Rai et al. [1999] sug- gest that seismicity is related to lithology that extends into the lower crust. Gahalaut et al. [2004], Srinagesh and Sarma [2005], and Sarma and Srinagesh [2007] sug- Characteristics of local earthquake seismograms of varying dislocation sources in a stratified upper crust and modeling for P and S velocity structure: comparison with observations in the Koyna-Warna region, India V. G. Krishna Retd. CSIR – National Geophysical Research Institute, Hyderabad, India Article history Received December 24, 2014; accepted October 28, 2015. Subject classiﬁcation: Seismology, Earthquake source and dynamics, Waves and wave analysis, Computational geophysics, Data processing. 1. Introduction Mobile seismograph networks in seismogenic re- gions provide signiﬁcantly large database of local earth- S0656 KRISHNA mogram sources techniqu charact drastica Figure 1. (a) Location map of the Koyna-Warna seismic region show- ing some of the seismograph stations (ﬁlled triangles) of the digital network operated, and the well located epicenters (ﬁlled circles) of the local earthquakes. The inset shows the observed seismicity pat- tern in the region described in the text. (b) Depth distribution of seis- micity along three sections AB, CD, and EF observed in the region. (Figures 1a and 1b, after Rai et al. [1999]). (c) Composite source mech- anism solutions for different earthquake clusters CI C2 C3 and C8 to the presence of a large number of faults and lineaments, predominantly striking NNW-NE, inferred from LANDSAT images as well [Langston 1981]. The inferred faults and linea- ments tend to deﬁne a broad en-echelon zone that parallels the Western Ghats in this area. p It is well known that essentially two dif- ferent types of faulting mechanisms, strike-slip and normal dominate in the Koyna-Warna seismic region [Talwani 1997]. Further, there seems to be also a rapid variation of focal mech- anisms from strike-slip type in the Koyna re- gion earthquakes to normal type in the Warna region earthquakes [Talwani 1997, Sharma 2000, Srinagesh and Sarma 2005]. Figure 1c illus- trates the rapid variation of the earthquake focal mechanisms from a predominantly strike- slip type in the north in the clusters C1, C2, C3, C8 in the Koyna region to a predominantly normal type in the south in the clusters C4, C5, C6, C7 in the Warna region (compiled by Sharma [2000]). A recent study by inversion of broad-band waveform data of Warna earth- quakes also indicates predominantly normal type focal mechanism solutions, some having a component of strike-slip [Shashidhar et al. 2011]. p p Upper crustal P and S velocity structure can be inferred reliably from the observed seis- mogram sections assembled from local earthquake sources as well by the synthetic seismogram modeling techniques. It is important to note that the seismogram characteristics of the local earthquake sources vary drastically for different source mechanisms as will be shown here. The amplitude variations for varying source mechanisms may often exceed those expected from velocity model variations. 1. Introduction Therefore in seismo- genic regions of sharply varying source mechanisms (within a short distance range), it is necessary to as- semble the seismogram sections separately for earth- quakes of similar source mechanism. An understanding of local earthquake seismogram characteristics is im- portant for determining the velocity structure models of the upper crust as well as the character and ampli- tude of near-source ground motions. The rupture di- rectivity size, that controls the amplitude of near-source ground motion, depends on the dip angle and the di- rection of slip (rake angle) of the dislocation source [Aagaard et al. 2004]. Further, the local earthquake seis- mogram characteristics may vary signiﬁcantly for dif- ferent source regimes (e.g., strike-slip, normal or reverse) and even within each regime for different sets of source parameters (strike, dip, and rake) rather than depend- ing only on the seismic velocity and Q structure of the Figure 1. (a) Location map of the Koyna-Warna seismic region show- ing some of the seismograph stations (ﬁlled triangles) of the digital network operated, and the well located epicenters (ﬁlled circles) of the local earthquakes. The inset shows the observed seismicity pat- tern in the region described in the text. (b) Depth distribution of seis- micity along three sections AB, CD, and EF observed in the region. (Figures 1a and 1b, after Rai et al. [1999]). (c) Composite source mech- anism solutions for different earthquake clusters CI, C2, C3, and C8 in the Koyna, and C4, C5, C6, and C7 in the Warna seismic region (after Sharma [2000]). gest that sustained seismicity in Koyna-Warna seismic zone is related to the geometry of existing faults and their interaction through stress transfer. Most researchers classify the seismicity as resulting from reservoir induced high pore pressure on existing stressed tectonic faults. The structure and nature of these faults, however, are poorly known because the entire Koyna-Warna region is covered at the surface by up to ~1.5 km of Deccan Traps basalt. The ongoing seismic activity in the Koyna- Warna region is primarily considered as the most out- standing example of reservoir-triggered seismicity [e.g., Gupta et al. 1969, Gupta et al. 1972, Gupta and Rastogi 1976, Gupta 1985, Gupta 1992, Talwani 1997, Gupta et al. 2012] essentially due to its proximity to the two reser- voirs in the region, the Koyna (impoundment in 1962) and the Warna (impoundment in 1985). CHARACTERISTICS OF LOCAL EARTHQUAKE SEISMOGRAMS host region. Some combinations of the source param- eters of the upper crustal earthquakes may result in seismograms at local distances that do not reveal any signatures (like reﬂections and conversions) related to the vertically inhomogeneous seismic structure with presence of sharp interfaces of large velocity contrasts. A systematic study by numerical simulations for dif- ferent source mechanisms can provide the necessary insight to understand the characteristics of local earth- quake seismograms and their modeling for the upper crustal velocity structure. In the recent years, only a few studies [Oglesby et al. 2000, Aagaard et al. 2001a, Aa- gaard et al. 2001b, Aagaard et al. 2004] have systemati- cally explored how the source parameters affect the near-source ground motions. Similar efforts for study- ing feasibility of velocity modeling of record sections of local earthquake seismograms for varying source regimes are, however, yet to be initiated. Because it is essential that the seismograms should be revealing and prominent phases as reﬂections, conversions and any other energetic phases generated by the existing veloc- ity stratiﬁcation are well recognized. A source effect leading to apparently missing or extremely low ampli- tude phases in a seismogram section should not be in- terpreted in terms of the velocity model features alone that suggest absence of prominent discontinuities. The source effect can be properly predicted by a systematic study of the characteristics of the seismograms for dif- ferent source regimes. pose of the present study to examine how well the ex- isting velocity stratiﬁcation is revealed and can be mod- eled from the seismograms of local earthquakes for different source regimes considered. Seismogram sec- tions of local earthquakes with dominantly strike-slip mechanism are found to be clearly favorable in reveal- ing the velocity stratiﬁcation in the upper crust for both P and S waves. In contrast the seismogram sections for earthquakes of other source mechanisms, particularly with rake angles around ±90°, seem to display the upper crustal P and P-to-S converted phases poorly with low amplitudes even in presence of sharp discon- tinuities of high velocity contrasts. The signiﬁcance of this study has been realized as several local earthquake seismogram sections are assembled in the Koyna- Warna seismic region in the DVP and their modeling attempted for obtaining the upper crustal P and S ve- locity structure [Krishna, under review]. 1. Introduction The seismicity in the Koyna region is considered to be possibly related gest that sustained seismicity in Koyna-Warna seismic zone is related to the geometry of existing faults and their interaction through stress transfer. Most researchers classify the seismicity as resulting from reservoir induced high pore pressure on existing stressed tectonic faults. The structure and nature of these faults, however, are poorly known because the entire Koyna-Warna region is covered at the surface by up to ~1.5 km of Deccan Traps basalt. The ongoing seismic activity in the Koyna- Warna region is primarily considered as the most out- standing example of reservoir-triggered seismicity [e.g., Gupta et al. 1969, Gupta et al. 1972, Gupta and Rastogi 1976, Gupta 1985, Gupta 1992, Talwani 1997, Gupta et al. 2012] essentially due to its proximity to the two reser- voirs in the region, the Koyna (impoundment in 1962) and the Warna (impoundment in 1985). The seismicity in the Koyna region is considered to be possibly related 2 CHARACTERISTICS OF LOCAL EARTHQUAKE SEISMOGRAMS A few observed seismogram sections are illustrated here for the strike- slip and normal faulting regimes from the Koyna-Warna seismic region. The observed seismogram sections il- lustrated here for the earthquake sources with strike- slip and normal mechanisms from the Koyna-Warna seismic region in the Deccan volcanic province in the western Indian shield substantiate these ﬁndings. Travel times and synthetic seismogram computations by the reﬂectivity method, are used for 1-D modeling of the observed virtual source local earthquake seismogram sections and inferring the upper crustal velocity struc- ture in the Koyna-Warna region. Signiﬁcantly, the in- ferred velocity model for the upper crust in the region reproduces the synthetic seismograms comparable to the observed sections for earthquake sources with dif- fering mechanisms in the Koyna and the Warna re- gions. Earlier studies [e.g., Talwani 1997, Sharma 2000, Srinagesh and Sarma 2005] in this region reveal left-lat- eral strike-slip faulting in the north and rapidly chang- ing to normal faulting in the south escarpment zones of the Koyna seismic region, and normal faulting in the Warna seismic region as well. In the present study we investigate this problem by a large number of numerical simulations. Reﬂectiv- ity synthetic seismograms in near offset ranges are ob- tained for a stratiﬁed upper crustal model having sharp discontinuities with 7%-10% P and S velocity contrasts. The synthetics are obtained for different source regimes (e.g., strike-slip, normal, reverse) as well as for different sets of source parameters (strike, dip, and rake) within each regime. The characteristics of local earthquake seismograms generated for different dislocation sources in a stratiﬁed upper crust are systematically examined. The scenarios considered vary from strike-slip to nor- mal faulting regimes of the upper crustal seismogenic zones, besides a few dip-slip and reverse (thrust) fault- ing models. Various numerical simulations for different source regimes considered are obtained here for the same upper crustal P and S velocity models similar to those given by Krishna et al. [1999] for the 1993 Latur earthquake area in the Deccan Volcanic Province (DVP), in the western Indian shield. The P and S ve- locity models have sharp discontinuities with velocity contrasts of 7%-10% at the top and bottom boundaries of two prominent low-velocity layers (LVLs), i.e., with signiﬁcant stratiﬁcation in the upper crust. It is the pur- 2. Local earthquake data set As more reliable data sets became available over time in the Koyna-Warna seismic region, location esti- mates steadily improved. The CSIR - National Geophys- ical Research Institute (NGRI) deployed a temporary seismic network consisting of 20 stations during 1996- 1998, and Rai et al. [1999] used some of those stations to relocate more than 400 of the recorded events. Their events had estimated timing errors less than a few mil- liseconds, and 93% of their events had epicentral reso- lutions of 0.2 km or less and hypocentral resolutions of 1.0 km or less, but the locations were based on only about 3 KRISHNA KRISHNA EQ no. Cluster Epicenter Focal depth Magnitude Epicentral distances Lat.° Long.° (km) (Mcoda) (km) Record section CK-FR-SG (Figure 9a) 9 C3 17.32 73.74 8.9 1.6 18.1, 12.8, 13.7 23 C3 17.28 73.74 7.3 1.6 16.3, 14.4, 10.7 24 C3 17.28 73.73 7.4 2.4 25 C2 17.35 73.75 8.2 2.1 20.7, 13.9, 16.7 30 C3 17.30 73.73 8.0 1.9 16.5, 13.2, 11.5 36 C2 17.33 73.73 7.8 2.3 12.0, 14.1 Record section CK-FR (Figure 10a) 21 C4 17.23 73.72 8.5 1.9 17.0, 13.5 26 C5 17.20 73.76 7.4 2.6 22.1 39 C5 17.22 73.76 7.7 1.6 20.2 Table 1. Koyna (clusters C2, C3) and Warna (clusters C4, C5) earthquakes (EQ) data for the seismogram sections CK-FR-SG, CK-FR, for an average source depth of 8 km. Table 1. Koyna (clusters C2, C3) and Warna (clusters C4, C5) earthquakes (EQ) data for the seismogram sections CK-FR-SG, CK-FR, for an average source depth of 8 km. and the hypocentral depths for various events in the depth range to ~11 km of the seismogenic upper crust. Srinagesh and Sarma [2005] relocated more than 600 earthquakes in the Koyna-Warna seismic zone, sepa- rately for three spatially distinct clusters, namely the north escarpment zone (NEZ), south escarpment zone (SEZ) and Warna seismic zone (WSZ), based on the in- ferences by Rai et al. [1999] during the analysis of seis- micity patterns in this region. The constrained focal mechanisms show distinct styles of faulting amongst the various fault segments with NEZ predominantly showing the strike-slip faulting with left lateral motion on the NE-SW fault. In SEZ and WSZ the focal mech- anisms are basically normal on the steeply dipping NNW-SSE to N-S faults [Srinagesh and Sarma 2005]. The epicentral-hypocentral resolutions estimated by Rai et al. Table 2. Source parameters for various earthquake clusters in the Koyna-Warna seismic region (after Sharma [2000]). 3. Characteristics of seismograms of local earthquake sources with different source mechanisms in a strati- ﬁed upper crust 3. Characteristics of seismograms of local earthquake sources with different source mechanisms in a strati- ﬁed upper crust In order to examine the inﬂuence of different source mechanisms on the observed local earthquake seismo- grams for a stratiﬁed upper crustal structure with pres- ence of sharp velocity discontinuities (e.g., with P and S velocity contrasts up to 7%-10%), a simulation study is thus attempted as shown in the following. A good number of reﬂectivity synthetic seismogram sections are obtained for different source regimes (e.g., strike- slip, normal, reverse) as well as for different sets of source parameters (strike, dip, and rake) within each regime for the same model of P and S velocity structure given in Figure 2. The characteristics of the local earthquake seismograms, as will be shown here, vary signiﬁcantly for different source mechanisms. The amplitude varia- tions for varying source mechanisms may often exceed those expected from velocity model variations. There- fore in seismogenic regions of rapidly varying source mechanisms within a short distance range, speciﬁcally in the Koyna-Warna seismic region of the present study, it is indeed necessary to assemble the seismo- gram sections separately for earthquakes of similar source mechanism. Figure 2. Upper crustal P and S velocity models used to compute the synthetic seismogram gathers shown in Figures 3-7 to illustrate variations of the local earthquake seismogram characteristics for different source mechanisms. Note the signiﬁcant stratiﬁcation in the upper crust that is examined as to how well the P and S reﬂec- tions and the P-to-S conversions can be recognized in various syn- thetics for different source mechanisms. synthetics are obtained for several combinations. While a QP/QS ratio of 9/4 is used in all the computations other ratios, e.g., 1 and 1/2 are used for a set of strike-slip and normal mechanisms to examine the resulting seismo- grams. Various synthetics obtained as described above are displayed by using identical scaling of the true ampli- tudes, so that the seismogram sections shown in the fol- lowing may be directly compared. Figures 3-7 show the vertical component reﬂectiv- ity synthetic seismogram gathers (true-amplitude plots with reduction velocity 6 km/sec) for the common off- set range of 28 km. 2. Local earthquake data set [1999] are found to be good enough for vari- ous events considered for investigating the problem six stations. Srinagesh and Sarma [2005] and Sarma and Srinagesh [2007] applied double-difference methodol- ogy in more precisely relocating more than 600 earth- quakes from this dataset. Shashidhar et al. [2011] used the VELEST algorithm to jointly locate hypocenters and to determine the velocity structure in the Koyna- Warna area, and their data were recorded on 11 seis- mographs. In an attempt to better characterize the seismogenic crust in the Koyna-Warna region, Dixit et al. [2014] recently acquired seismic data using a portable net- work consisting of 97 Taurus® seismographs, which is claimed to be denser than the previously deployed net- works in the area. In the present study, we utilize the vertical com- ponent seismograms of local earthquakes well recorded by a state-of-the-art Koyna digital seismograph net- work deployed by CSIR-NGRI during 1996-1998 [Rai et al. 1999]. The events were recorded (at 100 samples/s) by various stations of the seismograph network equipped with 24 bit REFTEK/PASSCAL digital recorders of short period three-component seismometers and GPS timing system. Figure 1a shows the location map of the Koyna-Warna seismic region with the epicenters of the local earthquakes and the seismograph stations used in the present study. The inset shows the observed seis- micity pattern in this region. Table 1 gives the epicen- tral data of the local earthquakes in the Koyna (clusters C2, C3) and Warna (clusters C4, C5) regions considered here for creating the two vertical component seismic record sections CK-FR-SG, and CK-FR for an average source depth of 8 km. Cluster Focal depth (km) Strike° Dip° Rake° C1 4–7 40 90 10 C2 7–11 30 82 6 C3 7–11 30 82 6 C8 1–4 50 85 9 C4 1–11 340 50 -90 C5 1–11 340 50 -90 C6 1–11 340 50 -90 C7 1–11 340 50 -90 Table 2. Source parameters for various earthquake clusters in the Koyna-Warna seismic region (after Sharma [2000]). Statistical analysis of a dataset of about 400 local earthquakes of the region by Rai et al. [1999] reveals well-constrained estimates of the epicentral locations 4 4 CHARACTERISTICS OF LOCAL EARTHQUAKE SEISMOGRAMS CHARACTERISTICS OF LOCAL EARTHQUAKE SEISMOGRAMS Figure 2. Upper crustal P and S velocity models used to compute the synthetic seismogram gathers shown in Figures 3-7 to illustrate variations of the local earthquake seismogram characteristics for different source mechanisms. Note the signiﬁcant stratiﬁcation in the upper crust that is examined as to how well the P and S reﬂec- tions and the P-to-S conversions can be recognized in various syn- thetics for different source mechanisms. being addressed in the present study, and thus no at- tempts are made here to further relocate the hypocen- tral parameters. The results of a preliminary study of composite focal mechanism solutions obtained by Sharma [2000] are given in Table 2, at different depth ranges of the spatially different clusters shown in Figure 1c. As shown in Figure 1c, analysis of the composite fault-plane solutions in the Koyna-Warna seismic re- gion reveals two distinct faulting mechanisms, pre- dominantly left-lateral strike-slip faulting in the Koyna region in the north rapidly changing to predominantly normal faulting in the Warna region in the south. KRISHNA KRISHNA KRISHNA Figure 3. (a) Vertical-component reﬂectivity seismograms gather (true amplitudes plot with reduction velocity 6 km/sec) for a common off- set range of 28 km showing amplitude variations of different phases due to rake angle variations of various source mechanisms. All the seis- mograms are computed for the source depth of 6 km using the P and S velocity models shown in Figure 2. Onsets of various travel-time branches are given to the right of the seismograms. Note the P phase amplitudes are however recognizable for rake angles in the range of at least −20° to +20°. (b) Same as Figure 3a for the normal faulting mechanism. Note very low amplitude P phases for the intermediate rake angles near −90°. Figure 3. (a) Vertical-component reﬂectivity seismograms gather (true amplitudes plot with reduction velocity 6 km/sec) for a common off- set range of 28 km showing amplitude variations of different phases due to rake angle variations of various source mechanisms. All the seis- mograms are computed for the source depth of 6 km using the P and S velocity models shown in Figure 2. Onsets of various travel-time branches are given to the right of the seismograms. Note the P phase amplitudes are however recognizable for rake angles in the range of at least −20° to +20°. (b) Same as Figure 3a for the normal faulting mechanism. Note very low amplitude P phases for the intermediate rake angles near −90°. Figure 3. (a) Vertical-component reﬂectivity seismograms gather (true amplitudes plot with reduction velocity 6 km/sec) for a common off- set range of 28 km showing amplitude variations of different phases due to rake angle variations of various source mechanisms. All the seis- mograms are computed for the source depth of 6 km using the P and S velocity models shown in Figure 2. Onsets of various travel-time branches are given to the right of the seismograms. Note the P phase amplitudes are however recognizable for rake angles in the range of at least −20° to +20°. (b) Same as Figure 3a for the normal faulting mechanism. Note very low amplitude P phases for the intermediate rake angles near −90°. Figure 3. (a) Vertical-component reﬂectivity seismograms gather (true amplitudes plot with reduction velocity 6 km/sec) for a common off- set range of 28 km showing amplitude variations of different phases due to rake angle variations of various source mechanisms. 3. Characteristics of seismograms of local earthquake sources with different source mechanisms in a strati- ﬁed upper crust The theoretical travel-times com- puted for the VP and VS models (given in Figure 2) are also shown as indicated on various seismograms by their onsets (P and S: refractions; P1, P2, P3, P4, and S1, S2, S3, S4: primary reﬂections). The source parameters, for which the seismograms are generated, are indicated with each gather in Figures 3-7. Each of the gathers il- lustrates variation of the seismogram characteristics with variation of one of the source parameters, with the other two parameters remaining unchanged. 3.1. Synthetic seismogram gathers for varying source parameters - A simulation study A large number of reﬂectivity synthetic seismo- grams [Kind 1985, Müller 1985] are computed in near offset ranges out to about 40 km for an arbitrary 6 km deep earthquake source in the upper crust with the vertically inhomogeneous P and S velocity models shown in Figure 2. A double-couple point source with the source-time function of Brüstle and Müller [1983] is used for various computations. Source mechanisms sim- ilar to the strike-slip, normal, reverse, and dip-slip mod- els are considered for different computations. In each case the set of source parameters (strike, dip, and rake) are systematically varied (at 10°-20° intervals) and the Figures 3a and 3b and Figures 4a and 4b reveal the seismogram characteristics varying due to the rake 5 CHARACTERISTICS OF LOCAL EARTHQUAKE SEISMOGRAMS angle variations for the strike-slip, normal, reverse, and dip-slip mechanisms. It is clear that the amplitudes of the P phases (refractions and reﬂections) as well as the P-to-S conversions are signiﬁcantly low for the inter- mediate rake angles around ±90° and these phases are recognizable only for the extreme rake angles around ±180° and 0°. Seismograms for the strike-slip mecha- nisms generally appear to be favorable for revealing the phases of P velocity stratiﬁcation (see Figure 3a for the left-lateral strike-slip case, and the seismogram for rake angle 180° in Figure 4a for the right-lateral strike-slip case). The S phases do not reveal signiﬁcant amplitude variations as the P phases, although for intermediate rake angles around ±90° they are relatively low for the normal and reverse mechanisms. respond to those inferred for some upper crustal earth- quakes in the Koyna-Warna seismic region of the DVP in the western Indian shield [Sharma 2000]. QP/QS ra- tios in the upper crustal layers equal to or even less than 1 are not uncommon and similar ratios have been found in the 1993 Latur earthquake area of the DVP [Krishna et al. 1999], and the eastern Dharwar craton [Krishna and Ramesh 2000] in the South Indian shield. Similar smaller QP/QS ratios, if exist in the Koyna-Warna region, may further degenerate the amplitudes of the P phases in the seismograms of the events with normal mecha- nisms as can be seen from Figure 7b. The major result of interest in this simulation study is that, seismogram sections of local earthquakes with dominantly strike-slip mechanism are clearly favorable in revealing the velocity stratiﬁcation in the upper crust for both P and S waves. In contrast the seismogram sec- tions for earthquakes of other source mechanisms, par- ticularly with rake angles around ±90°, seem to display the upper crustal reﬂected P and P-to-S converted phases poorly with low amplitudes even in presence of sharp discontinuities of high velocity contrasts. Thus it is clearly necessary to assemble the seismogram sections from local earthquakes of similar source mechanisms for rec- ognizing prominent phases from potential velocity dis- continuities and their modeling leading to reliable upper crustal P and S velocity structure. Variations of the seismogram characteristics due to the dip angle variations are illustrated in Figures 5a and 5b for the strike-slip and the normal mechanisms. CHARACTERISTICS OF LOCAL EARTHQUAKE SEISMOGRAMS Amplitudes of the P phases and the P-to-S conversions are relatively low for the normal mechanisms as com- pared to the strike-slip mechanisms. However the am- plitudes of the S phases are stronger for the normal mechanisms. It is again clear that the seismograms for the strike-slip mechanisms appear to be more favorable to delineate both P and S velocity stratiﬁcation. Similar differences, especially concerning the P phase ampli- tudes, between the seismograms for the strike-slip and normal mechanisms are also evident from Figures 6a and 6b. In these ﬁgures the seismograms are shown for varying strike but the same dip and rake angles. KRISHNA All the seis- mograms are computed for the source depth of 6 km using the P and S velocity models shown in Figure 2. Onsets of various travel-time branches are given to the right of the seismograms. Note the P phase amplitudes are however recognizable for rake angles in the range of at least −20° to +20°. (b) Same as Figure 3a for the normal faulting mechanism. Note very low amplitude P phases for the intermediate rake angles near −90°. Figure 4. (a) Same as Figure 3a for the right-lateral strike-slip (rake 180°) and the dip-slip (rake ±90°) mechanism. Note very low amplitude P phases and relatively high amplitude S phases for the rake angles ±90°. (b) Same as Figure 3a for the reverse (thrust) mechanism. Note very low amplitude P phases for the intermediate rake angle 90°. Figure 4. (a) Same as Figure 3a for the right-lateral strike-slip (rake 180°) and the dip-slip (rake ±90°) mechanism. Note very low amplitude P phases and relatively high amplitude S phases for the rake angles ±90°. (b) Same as Figure 3a for the reverse (thrust) mechanism. Note very low amplitude P phases for the intermediate rake angle 90°. Figure 4. (a) Same as Figure 3a for the right-lateral strike-slip (rake 180°) and the dip-slip (rake ±90°) mechanism. Note very low amplitude P phases and relatively high amplitude S phases for the rake angles ±90°. (b) Same as Figure 3a for the reverse (thrust) mechanism. Note very low amplitude P phases for the intermediate rake angle 90°. Figure 4. (a) Same as Figure 3a for the right-lateral strike-slip (rake 180°) and the dip-slip (rake ±90°) mechanism. Note very low amplitude P phases and relatively high amplitude S phases for the rake angles ±90°. (b) Same as Figure 3a for the reverse (thrust) mechanism. Note very low amplitude P phases for the intermediate rake angle 90°. 6 6 4. Modeling of the Koyna-Warna local earthquake seismogram sections Figure 7 shows the differences between the seis- mograms for the strike-slip and normal mechanisms for varying QP/QS ratio. The source parameters (strike, dip, and rake) of the two source regimes considered here cor- The development of 1-D crustal velocity models, particularly in seismogenic regions as the Koyna-Warna region, is the basis for a range of applications in earth- quake seismology. A well-developed 1-D crustal model Figure 5. Same as Figure 3a showing amplitude variations due to dip angle variations for (a) strike-slip, and (b) normal mechanisms. Note higher amplitude P phases for the strike-slip mechanism. Figure 5. Same as Figure 3a showing amplitude variations due to dip angle variations for (a) strike-slip, and (b) normal mechanisms. Note higher amplitude P phases for the strike-slip mechanism. 7 KRISHNA Further P and S velocity-depth proﬁles modeled in the nearby 1993 Latur earthquake area of the DVP [Krishna et al. 1999] reveal signiﬁcant stratiﬁ- cation of the upper crust with presence of prominent discontinuities having velocity contrasts of about 7%. The VP and VS models similar to those given in Figure 2, used for computation of travel time curves in Fig- ures 9a and 10a also indicate similar order of the ve- locity contrasts at various interfaces in the upper crust in the Koyna region as well [Krishna 2006]. The high amplitude secondary phases in Figure 9a ﬁt the P2, P3, P4, and S2, S3, S4 travel-time curves which are primary reﬂections at the discontinuities with similar order of velocity contrasts. These travel-time curves shown in Figure 10a also ﬁt the lower amplitude secondary phases correlated in the record section. It is only the different source mechanism that causes the apparently low amplitudes of the secondary phases. Synthetic seismograms thus obtained clearly resolve this prob- lem by generating low amplitude secondary phases even in presence of discontinuities with similar high velocity contrasts. Therefore the local earthquake seis- mogram sections as shown in Figure 10a should be modeled only by considering the appropriate source mechanism, otherwise the existing stratiﬁcation in the upper crust may be misinterpreted as a vertically ho- model can be used as an initial model for developing more complex 2-D and 3-D velocity models. The tradi- tional way to develop a 1-D velocity model utilizes the traveltimes of seismic phases. Synthetic seismograms modeling is recognized as the more efﬁcient technique to determine the velocity structure mitigating non- uniqueness of the models obtained from the arrival times of a few recognized phases. g p By the approach given by Krishna et al. [1999] seis- mogram sections in near distance ranges are assem- bled for a few common source depths using well located earthquakes with hypocentral depths agreeing within 1 km. Figure 8 illustrates one such composite section for a common source depth of 8 km assembled for earth- quakes recorded by various stations in the region. This record section is considered as a composite section generated by a virtual source at 8 km depth, because this section includes the seismograms from various az- imuths and also those for different source mechanisms. It is clear that the direct and refracted P and S phases are well aligned in this section. KRISHNA KRISHNA KRISHNA Figure 6. (a) Same as Figure 3a showing amplitude variations as the strike varies for the strike-slip mechanism. (b) Same as Figure 6a for the normal mechanism. Note relatively low amplitude P phases as compared to those for the strike-slip mechanism. Figure 6. (a) Same as Figure 3a showing amplitude variations as the strike varies for the strike-slip mechanism. (b) Same as Figure 6a for the normal mechanism. Note relatively low amplitude P phases as compared to those for the strike-slip mechanism. Figure 7. Same as Figure 3a showing amplitude variations due to different QP/QS ratios in seismograms for (a) strike-slip, and (b) normal mechanism. Note higher amplitude P phases for the strike-slip mechanism. Figure 7. Same as Figure 3a showing amplitude variations due to different QP/QS ratios in seismograms for (a) strike-slip, and (b) normal mechanism. Note higher amplitude P phases for the strike-slip mechanism. 1-D model, which can be used in studies of the rupture processes of earthquakes [e.g., Hartzell and Heaton 1983, Dreger and Helmberger 1993]. Furthermore, the 1-D has been widely used to determine the source parame- ters of earthquakes. The computational efﬁciency to calculate Green’s functions is a primary advantage of the has been widely used to determine the source parame- ters of earthquakes. The computational efﬁciency to calculate Green’s functions is a primary advantage of the 8 CHARACTERISTICS OF LOCAL EARTHQUAKE SEISMOGRAMS Figure 8. Composite observed record section with amplitudes normalized, in the epicentral distance versus reduced time (reduction veloc- ity 6 km/s) frame, assembled from processed seismograms of local earthquakes with average source depth of 8 km. Figure 8. Composite observed record section with amplitudes normalized, in the epicentral distance versus reduced time (reduction veloc- ity 6 km/s) frame, assembled from processed seismograms of local earthquakes with average source depth of 8 km. the record sections for the strike-slip (Figure 9a) and normal events (Figure 10a). It should be considered here that the crustal volume sampled by the ray paths involved for the seismograms shown in Figures 9a and 10a is of limited dimensions only. Thus it is quite un- likely that the observed systematic variations of the seismogram characteristics can be caused by lateral ve- locity perturbations. Both P and S velocity tomo- graphic inversion using local earthquake arrival times revealed a maximum lateral variation of about 4% [Rai et al. 1999]. KRISHNA However, any later ar- riving phases, in the P and S windows, may not be ev- idently recognizable in this section. Therefore it is considered necessary to assemble only seismograms for a similar source depth as well as similar source mechanism and recorded in a narrow azimuth range from various seismograph stations. The earthquake epicentral data used for this purpose are given in Table 1. The record sections thus assembled are shown in Figure 9a for events with dominantly strike-slip mech- anism and in Figure 10a for events with normal mech- anism. These record sections are band-pass ﬁltered (5-20 Hz) and plotted with a reduction velocity of 6 km/sec and amplitudes trace-normalized. Signiﬁcant variations in the seismogram characteristics as de- scribed in the earlier sections are clearly observable in 9 KRISHNA (a) (b) 9. (a) Observed record section assembled from seismograms of local earthquakes with similar source mechanisms (the do ism is strike-slip), aligned to an average source depth of 8 km, and recorded at a combination of the seismograph stations C in the Koyna-Warna region. A reduction velocity of 6 km/sec is used for plotting. The amplitudes are trace-normalized. A ban 20 Hz is applied. Computed travel-time curves are shown for direct and refracted P and S phases, and primary reﬂection ph as well as for the P and S phases leaving the source and reﬂected at the free-surface followed by reﬂection at an upper crustal abeled dotted curves in the time-increasing-order for reﬂections PfPi and SfSi, [i=1-4]). The seismograph stations and their a re indicated on the top of the record section. (b) Synthetic seismogram section computed by the reﬂectivity method for the parameters and using the inferred models of the upper crustal P and S velocity structure in the region. The amplitudes are tra Th l f l h h l d l (a) (b) Figure 9. (a) Observed record section assembled from seismograms of local earthquakes with similar source mechanisms (the dominant mechanism is strike-slip), aligned to an average source depth of 8 km, and recorded at a combination of the seismograph stations CK, FR, and SG in the Koyna-Warna region. A reduction velocity of 6 km/sec is used for plotting. The amplitudes are trace-normalized. A band-pass ﬁlter 5-20 Hz is applied. CHARACTERISTICS OF LOCAL EARTHQUAKE SEISMOGRAMS (a) (b) 10. (a) Same as Figure 9a observed record section assembled from seismograms of local earthquakes with similar source m he dominant mechanism is normal), aligned to an average source depth of 8 km, and recorded at a combination of the seismo CK, and FR in the Koyna-Warna region. Note signiﬁcantly lower amplitudes of various phases, particularly P reﬂections and rsions, as compared to those in Figure 9a. (b) Same as Figure 9b synthetic seismogram section computed for a different set source parameters and using the same inferred models of the upper crustal P and S velocity structure in the region. The ampl e-normalized. The left panel shows the same P velocity model as in Figure 9b. (a) ( ) Figure 10. (a) Same as Figure 9a observed record section assembled from seismograms of local earthquakes with similar source mecha- nisms (the dominant mechanism is normal), aligned to an average source depth of 8 km, and recorded at a combination of the seismograph stations CK, and FR in the Koyna-Warna region. Note signiﬁcantly lower amplitudes of various phases, particularly P reﬂections and P-to- S conversions, as compared to those in Figure 9a. (b) Same as Figure 9b synthetic seismogram section computed for a different set of the virtual source parameters and using the same inferred models of the upper crustal P and S velocity structure in the region. The amplitudes are trace-normalized. The left panel shows the same P velocity model as in Figure 9b. KRISHNA Computed travel-time curves are shown for direct and refracted P and S phases, and primary reﬂection phases Pi and Si, as well as for the P and S phases leaving the source and reﬂected at the free-surface followed by reﬂection at an upper crustal bound- ary (unlabeled dotted curves in the time-increasing-order for reﬂections PfPi and SfSi, [i=1-4]). The seismograph stations and their azimuth range are indicated on the top of the record section. (b) Synthetic seismogram section computed by the reﬂectivity method for the virtual source parameters and using the inferred models of the upper crustal P and S velocity structure in the region. The amplitudes are trace-nor- malized. The left panel shows the P velocity model. mogeneous medium. This is particularly applicable for regions like the Koyna-Warna seismic zones where the source mechanisms sharply vary from strike-slip to nor- mal within a short distance range. ward problem runs are made by varying the source parameters to improve the synthetics in explaining various features of the seismogram. In the following, 1-D model computational results of the travel times and synthetic seismograms are presented for modeling the record sections of local earthquake seismograms avail- able out to a maximum range of about 22 km as shown in Figures 9 and 10. The virtual source parameters in- ferred from modeling of these record sections are fur- ther used to compute synthetics. By this modeling approach, plausible models of the P and S velocity structure of the upper crust are inferred. The methodology for modeling the record sec- tions in the present study is as follows. Preliminary de- terminations of layer parameters are obtained from travel times modeling for inferring Earth’s structure. Further, synthetic seismograms are computed using the inferred P and S velocity models and initial set of source parameters (strike, dip, and rake) appropriate for the source depth of the record section. Several for- 10 CHARACTERISTICS OF LOCAL EARTHQUAKE SEISMOGRAMS 4.1. Travel times modeling a source reﬂected on the free surface followed by a re- ﬂection on i-th interface and terminating on the free surface (travel time curves are shown by unlabeled dot- ted curves in the order for reﬂection on the i-th inter- face, i=1-4 for both P and S phases) The upper crustal P and S velocity models inferred from the INVSP (INverse Vertical Seismic Proﬁling) gather [Krishna 2006] for the station WR, are used as the initial models for travel times modeling computa- tions. The following wave codes are considered for the purpose of illustration and modeling. Figure 11 illustrates the ray paths of various wave types described above. It is clear that these wave codes are by no means exhaustive, but they certainly repre- sent a set of prominent P and S phases being consid- ered for modeling the upper crustal velocity structure using the travel times. Possibly a large number of other multiples as well as converted phases from potential in- terfaces in the stratiﬁed upper crust, may be generated in the full wave synthetic seismogram computations il- P, S: up-going direct waves from a source as well as down-going refracted waves bottoming in the upper crustal layers and terminating on the free surface; Pi, Si: down-going waves from a source, reﬂected on i-th interface and terminating on the free surface (travel time curves are shown by continuous lines); and also PfPi, SfSi, unlabeled and dotted: up-going waves from 11 KRISHNA Figure 11. Schematic ray diagram illustrating the ray paths of prominent wave types considered for computing the P and S travel time curves given in Figures 9a and 10a. A: illustrates the upper crustal model, B: illustrates the ray paths for P (or S), C: illustrates the ray paths for Pi (or Si), and D: illustrates the ray paths for PfPi (or SfSi). KRISHNA NA Figure 11. Schematic ray diagram illustrating the ray paths of prominent wave types considered for computing the P and S travel time curves given in Figures 9a and 10a. A: illustrates the upper crustal model, B: illustrates the ray paths for P (or S), C: illustrates the ray paths for Pi (or Si), and D: illustrates the ray paths for PfPi (or SfSi). lustrated. 4.1. Travel times modeling Travel time computations for the P and S as well as the later arriving Pi and Si and the unlabeled phases (PfPi and SfSi), consistent with the ray geometry (Figure 11) in the stratiﬁed upper crust as illustrated in Figure 2, for a good number of plausible velocity mod- els starting with the initial models, are obtained. The travel time curves computed for the ﬁnal models of upper crustal P and S velocity structure inferred in the study region are illustrated in the two observed record sections in Figures 9a and 10a. As can be seen from these ﬁgures, the travel times modeling substantiates qualitatively the viability of the upper crustal P and S velocity models and the stratiﬁed structure with pres- ence of low-velocity layers (LVLs) inferred from the INVSP gather for the station WR in the Koyna-Warna region [Krishna 2006]. The synthetic seismogram sec- tions, also illustrated in Figures 9b and 10b, further con- ﬁrm these models as discussed in the following. in the stratiﬁed upper crust for inferring the prelimi- nary models of the P and S velocity structure. The ve- locity models thus inferred are further substantiated with the aid of full wave synthetic seismogram com- putations for a set of source parameters (strike, dip, and rake) which are also progressively revised to generate the synthetic sections comparable to the observed sec- tions (Figures 9a and 10a). The synthetics are generated for the 1-D isotropic Earth models in the present study. Computation of synthetic seismograms in a lay- ered half-space is an efﬁcient technique for modeling the Earth’s structure as well as the dynamic process of earthquake sources from well recorded seismic data. Many numerical techniques are well developed for gen- erating synthetic seismograms for point sources in plane layered media in order to model observed datasets. The generalized ray and wavenumber integration techniques are however more commonly used in seismic model- ing studies. CHARACTERISTICS OF LOCAL EARTHQUAKE SEISMOGRAMS (assembled from earthquakes with different sets of the source parameters) being modeled. The azimuth used in the computations is the average azimuth for each of the observed sections in Figures 9 and 10. QP and QS in the upper crustal layers are initially used as those in- ferred in the nearby 1993 Latur earthquake area [Kr- ishna et al. 1999]. The QP/QS ratio is however suitably improved for the upper crust in order to obtain the syn- thetics that ﬁt the observed sections. The ﬁnal synthetic sections are selected based on the overall ﬁts with the prominent phases in the observed sections, and shown in Figures 9b and 10b. It may be noted here that the same set of the P and S velocity models, but different sets of the virtual source parameters (strike, dip, and rake), are used for modeling the sections in Figures 9a and 10a. It is signiﬁcant to note that the same set of ﬁnal P and S velocity models as shown in Figure 12 are able to reproduce the prominent features in both the seismogram sections, but different sets of the source parameters have to be used in order to generate the synthetics reasonably matching the observed sections. The viability of the inferred velocity models for the study region is thus established. integration method for the generation of synthetic seis- mograms gives a complete solution, but can be compu- tationally intensive. The complete solution, rather than individual rays, is considered in this full wave theory approach. This method can handle a larger number of plane layers, but requires considerable computational ef- fort, especially at high frequencies. The reﬂectivity method for computing synthetic seismograms for an earthquake source [Kind 1985] is used for modeling the observed seismogram sections shown in Figures 9a and 10a. A trial and error approach of forward computa- tions, initially for reﬁning the P and S velocity models and later for ﬁxing the virtual source parameters (strike, dip, and rake), yielded acceptable ﬁts (based on visual check) of the synthetic seismogram sections with the observed sections. Starting with the initial set of source parameters given in Table 2, each of the three param- eters are varied ﬁxing the other two, and in the process a good number of synthetics are computed for the same set of P and S velocity models obtained from travel times modeling. 4.2. Synthetic seismograms modeling 4.2. Synthetic seismograms modeling Fourier transform techniques require some form of wavenumber integration [e.g., Fuchs and Müller 1971, Kennett and Kerry 1979, Wang and Herrmann 1980, Bouchon 1981, Kind 1985, Müller 1985]. Wavenumber The travel times modeling of the Koyna-Warna local earthquake record sections given in the earlier sec- tion is limited to only a few prominent P and S phases 12 CHARACTERISTICS OF LOCAL EARTHQUAKE SEISMOGRAMS The virtual source parameters (strike, dip, and rake) are inferred for the record section 5. Results and discussion The P and S velocity models and the resulting travel time curves given in Figure 12 are inferred from modeling the seismogram sections (Figures 9a and 10a) recorded at stations CK, FR, and SG. The maximum Figure 12. Inferred set of ﬁnal P and S velocity models and the resulting travel time curves (for ray paths illustrated in Figure 11) reproduc- ing the prominent features in both the seismogram sections (Figures 9 and 10), but with different sets of the source parameters in order to generate the synthetics reasonably matching both the observed sections. Figure 12. Inferred set of ﬁnal P and S velocity models and the resulting travel time curves (for ray paths illustrated in Figure 11) reproduc- ing the prominent features in both the seismogram sections (Figures 9 and 10), but with different sets of the source parameters in order to generate the synthetics reasonably matching both the observed sections. 13 KRISHNA offset in the record sections modeled here is within 22 km. Thus the inferred models are applicable to a lim- ited area surrounding the epicentral region (see Figure 1a). These P and S velocity models are consistent with those obtained from modeling the 28-km constant off- set INVSP gather at the seismograph station WR in the study region [Krishna 2006]. Signiﬁcantly, the alternat- ing LVLs with an average velocity reduction of about 0.30 km/s (4%-5%) for P and about 0.18 km/s (4%-5%) for S waves at depths of 6.1-8.1 km and 10.8-12.6 km appear to be compatible with the travel times and syn- thetic seismograms generated (Figures 9 and 10 using different sets of virtual source parameters) in the pres- ent study. The ratio of P and S velocities VP/VS in the upper crust is generally found to be ~1.7 in the study region. The P and S velocities in the upper crust (below a 1 km thick Deccan Traps layer with VP 4.65 km/s and VS 2.80 km/s as used in the present study) in the Koyna-Warna region are found to be varying from 6.34-6.45 km/s and 3.70-3.75 km/s (1.0-6.1 km depth), 6.45-6.48 km/s and 3.78-3.83 km/s (below the LVLs up to 12.6 km depth), and 6.15 km/s and 3.57 km/s in the LVLs. Knowledge of the Deccan Traps thickness in the Koyna region and the nature of the underlying base- ment has been a subject of considerable interest. CHARACTERISTICS OF LOCAL EARTHQUAKE SEISMOGRAMS VP 5.96 km/s in their model. However, the lower crustal layers in their model could not be resolved unambigu- ously. A recent study by Kilaru et al. [2015], based on the digital data set acquired by a dense network of 97 seismograph stations in the region, inferred a 1km thick Deccan Traps layer with VP 4.81 km/s, followed by an upper crustal layer with VP 5.94 km/s down to 10 km, and a middle crustal layer with VP 6.47 km/s extend- ing to 25 km depth. A comparison of these models with the velocity model inferred in the present study reveals that the Deccan Traps layer thickness and the VP as well as the VP/VS ratio agree quite well, and also the upper crustal VP at ~10 km agrees well within 1%-2% in all the models. It may be noted here that all the 1-D velocity models inferred by Srinagesh and Sarma [2005], Shashid- har et al. [2011], and Kilaru et al. [2015] have been de- termined by VELEST software [Kissling 1988, Kissling et al. 1994]. It is generally recommended that, the VE- LEST runs have to begin without low-velocity layers (setting LOWVELOCLAY=0 by default) since they have strong effects on the ray paths and, thus, they increase the non-linearity of the problem. It is not clear whether presence of low-velocity layers in the upper crust in the study region is tested by any of these models. The in- ferred velocity models in the present study as well as those given by Krishna [2006] in this region signiﬁcantly differ and clearly suggest presence of alternating low- velocity layers in the upper crust based on interpretation of the coherent later arriving secondary P and S phases as plausible reﬂections from them. These model features are also consistently well revealed by at least more than 20 similar record sections compiled in the study region. It is thus believed that the upper crustal velocity mod- els inferred in the present study are quite compatible with majority of the seismic record sections compiled by using the local earthquake recordings revealing the wave-ﬁeld in the available range in the study region. Figure 14. Two record sections compiled in the region in the pres- ent study, using the recordings at DH-WR stations for a source depth of 4 km and at DH-KN stations for a source depth of 6 km. 5. Results and discussion Re- cent results of scientiﬁc drilling (reported in 2013) through the Deccan Traps near Koyna has revealed about a 933 m thick pile of basaltic ﬂows, which is found to underlain by granitic basement rocks. The Koyna Bore Hole-1 (KBH-1) located very near the Koyna Dam, in close proximity to the 1967 Koyna earthquake of magnitude 6.3, reached a depth of 951 m at the time of the report [Roy et al. 2013]. The transi- tion from basalt to granite is marked by the occurrence of quartz and pink feldspar, followed by typical coarse grained granite deeper down. The P velocity model considered by Srinagesh and Sarma [2005] for relocating the local earthquakes recorded by the Koyna-Warna digital network has a 1 km of the Deccan Traps layer with VP 4.70 km/s un- derlain by the layer with VP 6.04 km/s and increasing in steps to 6.56 km/s at 16 km depth. Shashidhar et al. [2011] have estimated a 1.2 km thickness for the Deccan Traps layer with VP 4.40 km/s underlain by a layer with Figure 13. VP iso-velocity plots given by Dixit et al. [2014] for a limited region (as shown by a square near the seismograph station DH in Figure 1a) along with the ray paths obtained in the present study for two source depths at ~4 km and ~6 km and the resulting travel time curves for the direct P waves in the available offset range −8 to 8 km. Figure 13. VP iso-velocity plots given by Dixit et al. [2014] for a limited region (as shown by a square near the seismograph station DH in Figure 1a) along with the ray paths obtained in the present study for two source depths at ~4 km and ~6 km and the resulting travel time curves for the direct P waves in the available offset range −8 to 8 km. 14 CHARACTERISTICS OF LOCAL EARTHQUAKE SEISMOGRAMS CHARACTERISTICS OF LOCAL EARTHQUAKE SEISMOGRAMS The P and S travel time curves for the ray paths as given in Figure 11 are also given in these two record sections. The direct P travel time curves computed for the model given by Dixit et al. [2014] are also indicated below these record sections. An attempt has also been made here to ray-trace through the complex velocity model recently given by Dixit et al. [2014] and generate the ﬁrst arrival travel times in the near-offset range available in their model to ~8 km distance from two sources at ~4 km and ~6 km depths. Figures 13a and 13b show their VP iso-velocity plots along with the ray paths from the two source depths, and the resulting travel time curves for the di- rect P waves in the offset range −8 to 8 km. The region of study by Dixit et al. [2014] is indicated as a square area in Figure 1a, near the seismograph station DH. Figures 14a and 14b show two record sections compiled in the region in the present study using the recordings at DH-WR stations for a source depth of 4 km and at DH-KN stations for a source depth of 6 km. The P and S travel time curves for the ray paths as given in Figure Figure 14. Two record sections compiled in the region in the pres- ent study, using the recordings at DH-WR stations for a source depth of 4 km and at DH-KN stations for a source depth of 6 km. The P and S travel time curves for the ray paths as given in Figure 11 are also given in these two record sections. The direct P travel time curves computed for the model given by Dixit et al. [2014] are also indicated below these record sections. Figure 14. Two record sections compiled in the region in the pres- ent study, using the recordings at DH-WR stations for a source depth of 4 km and at DH-KN stations for a source depth of 6 km. The P and S travel time curves for the ray paths as given in Figure 11 are also given in these two record sections. The direct P travel time curves computed for the model given by Dixit et al. [2014] are also indicated below these record sections. Figure 14. CHARACTERISTICS OF LOCAL EARTHQUAKE SEISMOGRAMS Two record sections compiled in the region in the pres- ent study, using the recordings at DH-WR stations for a source depth of 4 km and at DH-KN stations for a source depth of 6 km. The P and S travel time curves for the ray paths as given in Figure 11 are also given in these two record sections. The direct P travel time curves computed for the model given by Dixit et al. [2014] are also indicated below these record sections. 15 KRISHNA 11 are also given in the two record sections. The direct P travel time curves computed for the model given by Dixit et al. [2014] are also indicated below these record sections. These travel times are found to be late by 0.030 s as compared to those obtained in the present study. Further, it is clear that the two record sections given in Figure 14 are more revealing as regards the upper crustal velocity structure in the region brought out with the aid of coherent and energetic secondary (reﬂection) phases. This is the great advantage of compiling the record sections using the local earthquake recordings in order to visualize the entire wave-ﬁeld of interest and model the P and S velocity structure of the upper crust in the region of study. The recent and denser dataset collected by Dixit et al. [2014] may be more resourceful to generate a large number of record sections as shown in the present study. The presence or absence of the upper crustal low-velocity layers may also be investi- gated reliably by this approach. It is also well known that the presence of low-velocity layers is a challenge to ﬁrst-arrival tomography, because the ray paths stay above these layers. This point is generalized to a notion that it is challenging for seismic tomography, especially travel time tomography, to resolve low-velocity anom- alies because ray paths tend to stay away from them. Considering these problems, it is suggested compiling and modeling of a large number of local earthquake seismic record sections for several virtual sources in seismogenic regions as done in the present study may lead to reliable velocity models for the upper crust. for signiﬁcant stratiﬁcation and high velocity contrasts consistently display low P amplitudes well matching the observed sections. CHARACTERISTICS OF LOCAL EARTHQUAKE SEISMOGRAMS It is clear that quite often the stratiﬁcation resolved in the upper crust has important implications for the structure and physical state of the seismogenic crust. Especially the upper crustal low-ve- locity layers well resolved in seismogenic regions help understand the physical state of the earthquake source regions [Krishna et al. 1999]. 7. Data and resources f h d A state-of-the-art digital seismograph network was deployed in the Koyna region during 1996-1998 to pro- vide a reliable database for studying the Earth’s struc- ture and the seismicity in the region [Rai et al. 1999]. In the present study, we utilized the vertical component seismograms of local earthquakes well recorded by 3 stations selected from the Koyna digital seismograph network. The events were recorded (at 100 samples/s) by various stations of the seismograph network equipped with 24 bit REFTEK/PASSCAL digital recorders of short period three-component seismometers and GPS timing system. Local earthquake seismogram sections illustrated here are assembled by an approach discussed by Krishna et al. [1999]. Synthetic seismogram compu- tations are made by using the reﬂectivity software de- veloped by Kind [1985] for an earthquake source. Acknowledgements. I am grateful to Prof. Dr. Harsh K. Gupta, Panikkar Professor at the CSIR - National Geophysical Re- search Institute, Hyderabad, for inviting me to process and model the valuable dataset of local earthquake seismograms to derive the upper crustal velocity structure in the Koyna-Warna seismic region, as well as for his helpful discussions concerning this research. I grate- fully acknowledge the kind advice of Prof. Dr. R. Kind, GFZ, Ger- many, for successful implementation of his reﬂectivity software in the Windows 7-CYGWIN 64 system, extensively used for rapid com- putation of a large number of synthetic seismogram sections in the present study. Prof. Dr. Friedemann Wenzel of KIT, Universitaet Karlsruhe, Germany contributed by helpful discussions. Prof. Dr. S.S. Rai while at NGRI, made available the digital seismograms data of local earthquakes acquired by their research group in the Koyna- Warna seismic region illustrated here. Computations were made on a VAX-3100 system, and on a Windows 7-CYGWIN 64 system. 6. Conclusions The present study, by several numerical simulations for varying source parameters, investigates an impor- tant problem concerning local earthquake seismogram characteristics, and their constraints for modeling the upper crustal velocity structure. Seismogram sections of local earthquakes with strike-slip mechanism are clearly more favorable in revealing the velocity stratiﬁ- cation in the upper crust for both P and S waves. Specif- ically in seismogenic regions like the Koyna-Warna, where more than one type of source mechanisms are dominant within a short distance range, special care has to be taken to assemble the record sections of seismo- grams from events of similar source mechanisms. Seis- mogram sections of local earthquakes with normal, reverse, and dip-slip mechanisms particularly with rake angles around ±90° seem to display the upper crustal P and P-to-S converted phases poorly with low ampli- tudes, even in presence of sharp discontinuities of high velocity contrasts. Such record sections have to be mod- eled by considering the appropriate source mechanism and possibly by testing whether the synthetics obtained CHARACTERISTICS OF LOCAL EARTHQUAKE SEISMOGRAMS Earth-Sci. Rev., 58, 279-310. Earth-Sci. Rev., 58, 279-310. Earth-Sci. Rev., 58, 279-310. motions: why rupture directivity was minimal in the 1999 Chi-Chi, Taiwan earthquake, B. Seismol. Soc. Am., 94, 155-170. Gupta, H.K. (2005). Artiﬁcial water reservoir-triggered earthquakes with special emphasis at Koyna, Curr. Sci., 88 (10), 1628-1631. Bouchon, M. (1981), A simple method to calculate Green’s functions for elastic layered media, B. Seismol. Soc. Am., 71, 959-971. Gupta, H.K., N.P. Rao, D. Shashidhar, N. Mallika and S. Roy (2012). Deep scientiﬁc drilling to study Reser- voir Triggered Earthquakes at Koyna, India, in Deep Continental Studies in India, Newsletter, Depart- ment of Science and Technology, Government of India, New Delhi, 22, 2-8. Brüstle, W., and G. Müller (1983). Moment and dura- tion of shallow earthquakes from Love-wave mod- eling for regional distances, Phys. Earth Planet. Inter., 32, 312-324. Hartzell, S.H., and T.H. Heaton (1983). Inversion of strong ground motion and teleseismic waveform data for the fault rupture history of the 1979 Impe- rial Valley, California, earthquake, B. Seismol. Soc. Am., 73, 1553-1583. Dixit, M.M., S. Kumar, R.D. Catchings, K. Suman, D. Sarkar and M.K. Sen (2014). Seismicity, faulting, and structure of the Koyna-Warna seismic region, West- ern India from local earthquake tomography and hypocenter locations, J. Geophys. Res., Solid Earth, 119, 6372-6398; doi:10.1002/2014JB010950. Kennett, B.L.N., and N.J. Kerry (1979). Seismic waves in a stratiﬁed half-space, Geophys. J. Roy. Astr. S., 57, 557-583. Dreger, D.S., and D.V. Helmberger (1993). Determina- tion of source parameters at regional distances with three-component sparse network data, J. Geophys. Res., 98, 8107-8125. Kilaru, S., S. Kumar, M.M. Dixit and P. Rama Rao (2015). 1-D velocity model delineated using dense seismic network in the Koyna-Warna region, Maharashtra, India, J. Ind. Geophys. Union, 19 (2), 182-189. Fuchs, K., and G. Müller (1971). Computation of syn- thetic seismograms with the reﬂectivity method and comparison with observations, Geophys. J. Roy. Astr. S., 23, 417-433. Kind, R. (1985). The reﬂectivity method for different source and receiver structures and comparison with GRF data, J. Geophys., 58, 146-152. Gahalaut, V.K., Kalpna and S.K. Singh (2004). Fault in- teraction and earthquake triggering in the Koyna- Warna region, India, Geophys. Res. Lett., 31, L11614; doi:10.1029/2004GL019818. Kissling, E.W. (1988). Geotomography with local earth- quake data, Rev. Geophys., 26, 659-698. Kissling, E.W., W.L. Ellsworth, D. Eberhart-Phillips and U. Kradolfer (1994). Initial reference models in local earthquake tomography, J. Geophys. Res., 99, 19635- 19646. Gupta, H.K., H. Narain, B.K. References Aagaard, B.T., J.F. Hall and T.H. Heaton (2001a). Char- acterization of near-source ground motion with earth- quake simulations, Earthq. Spectra, 17, 177-207. Aagaard, B.T., T.H. Heaton and J.F. Hall (2001b). Dy- namic earthquake ruptures in the presence of litho static normal stresses: implications for friction mod- els and heat production, B. Seismol. Soc. Am., 91, 1765-1796. Aagaard, B.T., J.F. Hall and T.H. Heaton (2004). Effects of fault dip and slip rake angles on near-source ground Aagaard, B.T., J.F. Hall and T.H. Heaton (2004). Effects of fault dip and slip rake angles on near-source ground 16 CHARACTERISTICS OF LOCAL EARTHQUAKE SEISMOGRAMS Rastogi and I. Mohan (1969). A study of the Koyna Earthquake of Decem- ber 10, 1967, B. Seismol. Soc. Am., 59, 1149-1162. Krishna, B., N. Negi and J.G. Negi (1973). Rift valley be- neath Deccan Traps (India), Geophys. Res. Bull., 2, 207-237. Gupta, H.K., B.K. Rastogi and H. Narain (1972). Some discriminatory characteristics of Earthquakes near the Kariba, Kremasta and Koyna artiﬁcial lakes, B. Seismol. Soc. Am., 62, 493-507. Krishna, V.G., C.V.R.K. Rao, H.K. Gupta, D. Sarkar and M. Baumbach (1999). Crustal seismic velocity struc- ture in the epicentral region of the Latur earthquake (September 29, 1993), southern India: inferences from modeling of the aftershock seismograms, Tectono- physics, 304, 241-255. Gupta, H.K., and B.K. Rastogi (1974). Will another damaging earthquake occur in Koyna?, Nature, 248, 214-216. Gupta, H. K., and B.K. Rastogi (1976). Dams and Earth- quakes, 1st ed., Elsevier Scientiﬁc Publishing Com- pany, Amsterdam, 372 p. Krishna, V.G., and D.S. Ramesh (2000). Propagation of crustal-waveguide-trapped Pg and seismic velocity structure in the South Indian shield, B. Seismol. Soc. Am., 90, 1281-1294. Gupta, H.K., C.V.R.K. Rao and B.K. Rastogi (1980). An investigation of earthquakes in Koyna region, Ma- harashtra, for the period October 1973 through De- cember 1976, B. Seismol. Soc. Am., 70, 1833-1847. Krishna, V.G. (2006). INVSP gathers of local earthquake seismograms: an approach for modelling the upper crustal P and S velocity structure, Geophys. J. Int., 166, 148-154. Gupta, H.K. (1985). The present status of the reservoir- induced seismicity with the special emphasis on Koyna Earthquake, Tectonophysics, 118, 257-507. Gupta, H.K. (1992). Reservoir Induced Earthquakes, El- sevier Publishers, Amsterdam, 364 p. Krishna, V.G. (under review). An approach of refraction seismology for processing and modelling of local earthquake seismogram sections of virtual sources at multiple depths in seismogenic regions – Applica- tion to Koyna-Warna region, India for upper crustal Gupta, H.K. (2002). A review of recent studies of trig- gered earthquakes by artiﬁcial water reservoirs with special emphasis on earthquakes in Koyna, India, 17 KRISHNA P and S velocity structure; submitted to Annals of Geophysics, under review. Langston, C.A. (1981). Source inversion of seismic wave- forms: the Koyna, India, earthquakes of 13 Septem- ber 1967, B. Seismol. Soc. Am., 71, 1-24. Müller, G. (1985). The reﬂectivity method: a tutorial, J. Geophys., 58, 153-174. Oglesby, D.D., R.J. Archuleta and S.B. Nielsen (2000). The three-dimensional dynamics of dipping faults, B. Seismol. Soc. Am., 90, 616-628. B. Seismol. Soc. Am., 90, 616-628. *Corresponding author: V.G. Krishna, 10-2-267/2, Fashions’ Apts., West Marredpally, Secunderabad- 500026, India; email: v_gopalak@yahoo.com, gopalakrishna.velamakanni@gmail.com. © 2015 by the Istituto Nazionale di Geoﬁsica e Vulcanologia. All rights reserved. gopalakrishna.velamakanni@gmail.com. CHARACTERISTICS OF LOCAL EARTHQUAKE SEISMOGRAMS Rai, S.S., S.K. Singh, P.V.S.S. Rajagopal Sarma, D. Sri- Nagesh, K.N.S. Reddy, K.S. Prakasam and Y. Satya- narayana (1999). What triggers Koyna region earth- quakes? Preliminary results from seismic tomogra- phy digital array, Proc. Indian Acad. Sci. (Earth Planet. Sci.), 108, 1-14. Sarma, P.R., and D. Srinagesh (2007). Improved earth- quake locations in the Koyna-Warna seismic zone, Nat. Hazards, 40, 563-571. Sharma, J. (2000). Focal mechanism studies in Koyna- Warna seismic zone, M.Tech. dissertation, Kuruk- shetra Univ., India, 38 p. Shashidhar, D., N. Purnachandra Rao and H. Gupta (2011). Waveform inversion of broad-band data of local earthquakes in the Koyna-Warna region, West- ern India, Geophys. J. Int., 185, 292-304. Srinagesh, D., and P.R. Sarma (2005). High precision earthquake locations in Koyna-Warna seismic zone reveal depth variation in brittle-ductile transition zone, Geophys. Res. Lett., 32, L08310; doi:10.1029/ 2004GL022073. Roy, S., N.P. Rao, V.V. Akkiraju, D. Goswami, M. Sen, H. Gupta, B.K. Bansal and S. Nayak (2013). Granitic Basement below Deccan Traps Unearthed by Drilling in the Koyna Seismic Zone, Western India, NEWS AND NOTES, J. Geol. Soc. India, 81, 289-290. AND NOTES, J. Geol. Soc. India, 81, 289-290. Talwani, P. (1997). Seismotectonics of the Koyna-Warna area, India, Pure Appl. Geophys., 150, 511-550. Wadia, D.N. (1968). The Koyna earthquake, December 1967, special number on Koyna earthquake 11th De- cember 1967, J. Indian Geophys. Union, 5, 6-8. Wang, C.Y., and R.B. Herrmann (1980). A numerical study of P-, SV-, and SH-Wave generation in a plane layered medium, B. Seismol. Soc. Am., 70, 1015- 1036. 18
https://openalex.org/W4281676634	https://www.int-arch-photogramm-remote-sens-spatial-inf-sci.net/XLIII-B1-2022/213/2022/isprs-archives-XLIII-B1-2022-213-2022.pdf	English	null	AN INTEGRATED INDOOR POSITIONING ALGORITHM FOR SMARTPHONE USING PEDESTRIAN DEAD RECKONING WITH MAGNETIC FINGERPRINT AIDED	The international archives of the photogrammetry, remote sensing and spatial information sciences/International archives of the photogrammetry, remote sensing and spatial information sciences	2,022	cc-by	5,314	ABSTRACT: Indoor positioning has gained increasing attention in previous decades. There are many wireless communication technologies replacing Global Navigation Satellite System (GNSS) because of the sheltered GNSS signal in indoor environment. Most of them need to set up external transceiver with high spatial density to get user's position, such as Wi-Fi, Ultra-Wideband (UWB), Bluetooth and so on. In order to reach high positioning accuracy, the cost of external transceiver becomes higher. g p g y g This research focuses on low-cost pedestrian dead reckoning (PDR) without additional external equipment. Moreover, a magnetic fingerprint-based positioning is adopted to provide redundant observations of position and heading, using Extended Kalman Filter (EKF) to update the estimation. The proposed method reduces cumulative errors of PDR, achieving an improved algorithm. Since the geomagnetic field exists over the Earth, this technology doesn’t use external equipment either. The integration of PDR and magnetic fingerprint-based positioning, which only uses built-in sensors of a smartphone, should be a low-cost and wide coverage scheme. Chi-Hsin Huang 1, Yi-Feng Chang 2, Ya-Tang Tang2, Meng-Lun Tsai1, Kai-Wei Chiang 1 1 Department of Geomatics, National Cheng Kung University, Taiwan – (windstorm, taurusbryant, kwchiang) @geomatics.ncku.edu.tw 2 Department of Geomatics, National Cheng Kung University, Taiwan – (p66104166, p66084015) @gs.ncku.edu.tw Commission I, WG I/7 ndoor Positioning, inertial navigation, Pedestrian Dead Reckoning, Magnetic Fingerprint, Smartphone, Extended Kalman Filter. 1. INTRODUCTION 1. Develop the procedures of data collecting, magnetic fingerprint extraction and offline magnetic map generation in an efficient way. Recently, the number of smartphone users has continuously increased. There are 83.7% of the world’s population who own the smartphone in 2022, which means the navigation services on the smartphone become more accessible to people. Unfortunately, the GNSS signal would be sheltered in indoor environment, which leads to a challenge of indoor positioning. Since people spend about 90% of time in indoor environment, the requirement of indoor positioning occurs. The smartphone with built-in multiple sensors has high potential to achieve a stable seamless navigation system with desk-level (1-3meters) accuracy. 2. Propose a PDR based pattern recognition method, which divides the trajectory of pedestrian in several parts. The patterns and the corresponding magnetic series are used to be matched with offline magnetic map. 3. Propose a heading correction method using the magnetic orientation information in offline map to provide the reference and compensate the magnetic anomalies. 4. Develop the integration scheme using Extended Kalman Filter with the position and heading update from magnetic fingerprint-based algorithm. Pedestrian Dead Reckoning (PDR) is widely used in indoor positioning, which uses Micro Electro Mechanical Systems (MEMS) based accelerometers and gyroscopes to estimate the 2D trajectory of pedestrian (Chen, Pei, and Chen, 2011). This technology is a kind of self-contained relative positioning without wireless communication with external transceiver, which has the advantages of low cost and high convenience. Nevertheless, the characteristics of inertial navigation let errors cumulate and the trajectory drift with time. The external update from other sensers should be adopted to solve this problem. The rest of this paper is organized as follows: Section 2 provides the details of methodology, including offline magnetic map generation and online integrated positioning scheme. Section 3 describes the field and the architecture of experiment. In Section 4, the result of proposed method is presented. Finally, Section 5 comes up with the conclusion. Corresponding author This contribution has been peer-reviewed. https://doi.org/10.5194/isprs-archives-XLIII-B1-2022-213-2022 \| © Author(s) 2022. CC BY 4.0 License. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XLIII-B1-2022 XXIV ISPRS Congress (2022 edition), 6–11 June 2022, Nice, France The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XLIII-B1-2022 XXIV ISPRS Congress (2022 edition), 6–11 June 2022, Nice, France 2.1.2 Horizontal Component Transformation In offline phase, the extracted fingerprint is used to generate the magnetic feature map and orientation map. In online phase, the real time extracted fingerprint is matched with the offline feature map to get the absolute position; moreover, the heading correction is conducted by using orientation map to compensate the magnetic anomalies in indoor environment. The details of offline map generation are provided in Section 2.1. The online magnetic fingerprint-based positioning and the proposed integration scheme with PDR are organized in Section 2.2. For the purpose of magnetic heading estimation, it is necessary to get the orientation of horizontal component H. In other words, H should be transformed into a level frame so that the magnetic heading angle can be estimated. Ideally, a hand holding smartphone is supposed to be level, i.e., the roll and pitch angles are equal to zero. Based on this assumption, the component of H contains only x and y axis, the heading can be calculated by equation (1). Practically, the roll and pitch angles are not equal to zero, and the angles vary with walking. Figure 2(b) shows the roll and pitch angles of x axis and y axis. Using the gravity in body frame can not only extract the H vector but also determine roll and pitch angles as the equation (2) and equation (3) show. In this case, the H (𝐻𝐻𝑥𝑥𝑏𝑏, 𝐻𝐻𝑦𝑦𝑏𝑏, 𝐻𝐻𝑧𝑧𝑏𝑏) in body frame can be transformed into level frame H (𝐻𝐻𝑥𝑥′ 𝑙𝑙, 𝐻𝐻𝑦𝑦′ 𝑙𝑙, 0). After the transformation, the 𝐻𝐻𝑥𝑥𝑏𝑏, 𝐻𝐻𝑦𝑦𝑏𝑏 in equation (1) can be replaced by 𝐻𝐻𝑥𝑥′ 𝑙𝑙, 𝐻𝐻𝑦𝑦′ 𝑙𝑙; therefore, the magnetic heading angle, which is independent of gyroscope can be estimated. 𝜃𝑎𝑎𝑎𝑎𝐻𝑥𝑏𝐻𝑦𝑏 Figure 1. The overall flow-chart of magnetic fingerprint-based positioning 𝜃𝜃= 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎2൫𝐻𝐻𝑥𝑥𝑏𝑏, 𝐻𝐻𝑦𝑦𝑏𝑏൯, (1) 𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟= 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎൬ 𝑔𝑔𝑦𝑦𝑏𝑏 𝑔𝑔𝑧𝑧𝑏𝑏൰, (2) 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝ℎ= 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎ቌ −𝑔𝑔𝑥𝑥𝑏𝑏 ටቀ𝑔𝑔𝑦𝑦 𝑏𝑏2+𝑔𝑔𝑧𝑧𝑏𝑏2ቁ ቍ (3) 𝜃 𝜃𝜃= 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎2൫𝐻𝐻𝑥𝑥𝑏𝑏, 𝐻𝐻𝑦𝑦𝑏𝑏൯, (1) 𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟= 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎൬ 𝑔𝑔𝑦𝑦𝑏𝑏 𝑔𝑔𝑧𝑧𝑏𝑏൰, (2) 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝ℎ= 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎ቌ −𝑔𝑔𝑥𝑥𝑏𝑏 ටቀ𝑔𝑔𝑦𝑦 𝑏𝑏2+𝑔𝑔𝑧𝑧𝑏𝑏2ቁ ቍ (3) 𝜃 (1) Figure 1. The overall flow-chart of magnetic fingerprint-based positioning (2) (3) 2.1 Offline Magnetic Map Generation The procedures of vector projection, dimension reduction and transformation introduced in this Section is applied in offline phase as well as online phase. Conversely, the strategy of data collection and calibration is specific to offline phase. where 𝜃𝜃= magnetic heading angle 𝐻𝑥𝑏𝐻𝑦𝑏 2. METHODOLOGY The magnetic fingerprint-based positioning technology makes use of the magnetic anomalies caused by the magnetism materials in the buildings, e.g., reinforcing steel. The magnetic anomalies can become unique and stable “fingerprints” in the space, which can be used in matching algorithm. The magnetic fingerprint- based positioning is divided into two phases, i.e., offline and online phase (Gong et al., 2018). In offline phase, the magnetic field is collected, and the magnetic map is generated. In online phase, the real time measurement can be matched with the offline magnetic map; thus, the absolute positioning is achieved. This research makes the following contributions: The overall process of the magnetic fingerprint-based positioning is shown in Figure 1. The observed magnetic vector should be projected to the gravity direction and be decomposed into vertical and horizontal component (Li et al., 2012) so as to extract a rotational invariance, i.e., the fingerprint. In addition, the extracted horizontal component should be transformed into a level frame, that the orientation of horizontal component relative to the known heading of device can be obtained. As a side note, the gravity vector can be extracted from accelerometer by using low pass filter (Yun, Bachmann, and McGhee, 2008). 213 The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XLIII-B1-2022 XXIV ISPRS Congress (2022 edition), 6–11 June 2022, Nice, France The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XLIII-B1-2022 XXIV ISPRS Congress (2022 edition), 6–11 June 2022, Nice, France 2.1.1 Vector Projection and Dimension Reduction 𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟= roll angle determined by gravity 𝑝𝑝𝑝𝑝 The magnetic field M(x, y, z) sensed by triaxial magnetometer is defined in body frame of the sensor, which means the results change with the orientation of sensor. In order to extract the fingerprint, the roll, pitch, yaw (heading) angle of 3 axes should be known to transform the magnetic field from body frame to navigation frame, e.g., Local-Level Frame (ENU). Since it is unreliable for smartphone to estimate the orientation by using MEMS gyroscopes only; moreover, the heading angle estimated by magnetometer is distorted by the magnetic anomalies, the transformation is invalid. The magnetic heading can only be used after the procedures in Section 2.1.2 and Section 2.2.3. where 𝜃𝜃= magnetic heading angle 𝐻𝑥𝑏𝐻𝑦𝑏 𝐻𝐻𝑥𝑥𝑏𝑏, 𝐻𝐻𝑦𝑦𝑏𝑏= x and y component of H in body frame 𝑟𝑟𝑟𝑟 𝐻𝐻𝑥𝑥𝑏𝑏, 𝐻𝐻𝑦𝑦𝑏𝑏= x and y component of H in body frame 𝑟𝑟𝑟𝑟 This contribution has been peer-reviewed. https://doi.org/10.5194/isprs-archives-XLIII-B1-2022-213-2022 \| © Author(s) 2022. CC BY 4.0 License. This contribution has been peer-reviewed. s-archives-XLIII-B1-2022-213-2022 \| © Author(s) 2022. CC BY 4.0 License. 2.1.3 Strategy of Data Collection in Offline Phase 𝑀𝑀𝑚𝑚 𝑏𝑏= 𝑆𝑆𝑘𝑘𝑆𝑆𝑜𝑜𝑆𝑆𝑠𝑠𝑀𝑀𝑏𝑏+ 𝑏𝑏+ 𝑛𝑛= 𝑆𝑆𝑀𝑀𝑏𝑏+ 𝑏𝑏+ 𝑛𝑛, (4) 𝑀𝑀𝑥𝑥𝑏𝑏= 𝑀𝑀𝑚𝑚𝑚𝑚 𝑏𝑏+ −𝑀𝑀𝑚𝑚𝑚𝑚 𝑏𝑏− 2 = 𝑀𝑀𝑥𝑥𝑏𝑏+𝑏𝑏𝑥𝑥−൫−𝑀𝑀𝑥𝑥𝑏𝑏+𝑏𝑏𝑥𝑥൯ 2 , (5) 𝑀𝑚𝑏 (4) where 𝑀𝑀𝑚𝑚 𝑏𝑏= measured magnetic field in body frame 𝑀𝑏 where 𝑀𝑀𝑚𝑚 𝑏𝑏= measured magnetic field in body frame 𝑀𝑀𝑏𝑏= true magnetic field in body frame 𝑆𝑆𝑘𝑘= the sensitivities of 3 axis 𝑆𝑆𝑜𝑜= nonorthogonality and misalignment errors 𝑆𝑆𝑠𝑠= soft iron errors 𝑆𝑆= combination of 𝑆𝑆𝑘𝑘, 𝑆𝑆𝑜𝑜, 𝑆𝑆𝑠𝑠 𝑏𝑏= bias 𝑛𝑛= noise +, − denote the two opposite directions of sensor axis 𝑥𝑥 denotes the x axis 𝑀𝑀𝑏𝑏= true magnetic field in body frame 𝑆𝑘 g p p g g p Figure 5. The flow-chart of online positioning and integration 𝑆𝑆= combination of 𝑆𝑆𝑘𝑘, 𝑆𝑆𝑜𝑜, 𝑆𝑆𝑠𝑠 𝑏 𝑆𝑆= combination of 𝑆𝑆𝑘𝑘, 𝑆𝑆𝑜𝑜, 𝑆𝑆𝑠𝑠 𝑏 +, − denote the two opposite directions of sensor axis 𝑥 Figure 5. The flow-chart of online positioning and integration 𝑥𝑥 denotes the x axis 2.1.3 Strategy of Data Collection in Offline Phase 214 The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XLIII-B1-2022 XXIV ISPRS Congress (2022 edition), 6–11 June 2022, Nice, France The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XLIII-B1-2022 XXIV ISPRS Congress (2022 edition), 6–11 June 2022, Nice, France The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XLIII-B1-2022 XXIV ISPRS Congress (2022 edition), 6–11 June 2022, Nice, France Take the x axis for example, the formula can be expressed as equation (5). By collecting the data in two opposite directions respectively, e.g., rotating 180 degrees on z axis, the bias of x and y axis can be eliminated. 𝑀𝑆𝑆𝑆𝑀𝑏𝑛𝑆𝑀𝑏𝑛 Take the x axis for example, the formula can be expressed as equation (5). By collecting the data in two opposite directions respectively, e.g., rotating 180 degrees on z axis, the bias of x and y axis can be eliminated. 𝑀𝑆𝑆𝑆𝑀𝑏𝑛𝑆𝑀𝑏𝑛 the magnetic orientation map stores the heading angle of horizontal component of each grid, which is calculated by the procedure mention in Section 2.1.2 and is estimated with reference to the known heading of smartphone in offline phase. 2.2 Online Positioning and Integration Scheme Figure 5 shows the flow-chart of integration scheme based on PDR, which provides a continually relative positioning (Chen, Pei, and Chen, 2011) by using step detection, step length estimation and heading estimation. The PDR becomes the prediction in EKF, and the magnetic series matching (Gong et al., 2018) becomes a position update. A 2D matching scheme, which uses the predicted trajectory to generate a specific pattern is proposed in this research. By sliding the pattern in feature grid map, the 1D series matching is conducted and the 2D positioning is achieved. The magnetic heading estimated from online phase and the heading correction from offline phase become a heading update. The EKF thus works and integrates PDR and magnetic fingerprint-based positioning to get an optimal estimation. 2.2.1 Pedestrian Dead Reckoning PDR is a continually relative positioning, using the position of last step to estimate the position of next step. The model can be expressed as equation (6) and (7) (Chen, Pei, and Chen, 2011). The heading is calculated by integrating angular rate from gyroscopes. Accelerometers is used to detection the steps of pedestrian by recognizing periodic signal, which is the summation of 3 axis acceleration subtracts the gravity. Figure 6 shows the illustration of step detection. The crests and troughs of signal is observed, and the timing of closest zero after a pair of peak and valley is chosen as the end of one step. Finally, the step length is estimated by using the model as equation (8), which considers the user height and step frequency. By integrating heading estimation, step detection and step length estimation, the heading and displacement of each step is estimated. 𝑁𝐾𝑁𝐾𝑆𝐿𝐾𝑐𝑐𝑐𝜑𝐾 Figure 4 shows the effect of bias in map generation. Collecting the data in S shaped, i.e., collecting the data of adjacent line by reverse direction is the most efficient strategy. Unfortunately, the bias of x and y axis effect on the horizontal component oppositely in two reverse directions, which leads to discontinuous strips on the map. The uncalibrated bias of magnetometer is extremely significant for map generation. Therefore, the simplify calibration method as equation (5) is implemented in the strategy. Each line would be collected both back and forth to eliminate the bias of x and y axis. The calibrated horizontal component is suitable for heading correction, i.e., generates the orientation map. On the other hand, the bias of z axis is retained. Since the x axis nearly point upward in both offline phase and online phase, the trend and feature of vertical component would not change with heading. The difference between two phases is approximately systematic, that the matching scheme still can work. 𝑁𝑁𝐾𝐾+1 = 𝑁𝑁𝐾𝐾+ 𝑆𝑆𝐿𝐿𝐾𝐾× 𝑐𝑐𝑐𝑐𝑐𝑐𝜑𝜑𝐾𝐾, (6) 𝐸𝐸𝐾𝐾+1 = 𝐸𝐸𝐾𝐾+ 𝑆𝑆𝐿𝐿𝐾𝐾× 𝑠𝑠𝑠𝑠𝑠𝑠𝜑𝜑𝐾𝐾, (7) 𝑆𝑆𝑆𝑆= ቀ0.7 + 0.371 ∙(𝐻𝐻−1.75) + 0.227 ∙ (𝑆𝑆𝑆𝑆−1.79) ∙ 𝐻𝐻 1.75ቁ, (8) 𝐸𝑁 (6) Figure 4. The illustration of collecting data by reverse direction (8) where 𝐸𝐸, 𝑁𝑁= coordinated in Local-Level Frame 𝐾 𝐾𝐾= step counts 𝑆𝑆 𝐾𝐾= step counts 𝑆𝑆 Figure 4. The illustration of collecting data by reverse direction Figure 4. The illustration of collecting data by reverse direction 𝑆𝑆𝑆𝑆= step length φ = heading from gyroscope 𝐻 φ = heading from gyroscope 𝐻 2.1.3 Strategy of Data Collection in Offline Phase 2.1.3 Strategy of Data Collection in Offline Phase In this research, the experimental fields are divided into several regular grids. It is intuitive and accurate to collect the data in each center of grid, but this procedure is inefficient. Another common strategy is showed as Figure 3. Walking along the red lines with an assumption of constant speed, dividing data by time and averaging the data in each interval, the approximate results can be obtained. The feasibility of this strategy is evaluated in Section 3.1.1. Figure 3. The illustration of grid division and the proposed strategy of data collection In this research, the experimental fields are divided into several regular grids. It is intuitive and accurate to collect the data in each center of grid, but this procedure is inefficient. Another common strategy is showed as Figure 3. Walking along the red lines with an assumption of constant speed, dividing data by time and averaging the data in each interval, the approximate results can be obtained. The feasibility of this strategy is evaluated in Section 3.1.1. In this research, a vector projection is used to replace conventional transformation method. Figure 2(a) shows the projection from magnet vector M to gravity vector g. The magnet vector can be decomposed into vertical V component and horizontal component H, which is a rotationally invariant fingerprint. Even though the dimension of M is reduced, the 2D magnetic information is enough for matching, and the fingerprint becomes more reliable. Figure 3. The illustration of grid division and the proposed strategy of data collection Figure 3. The illustration of grid division and the proposed strategy of data collection Figure 2. The illustration of vector projection (a) and horizontal component transformation (b) Figure 3. The illustration of grid division and the proposed strategy of data collection Furthermore, the error model of magnetometer, which can be expressed as equation (4) (Fang et al., 2011), should be considered in the design of strategy. The 𝐒𝐒𝐤𝐤, 𝐒𝐒𝐨𝐨, 𝐒𝐒𝐬𝐬 matrix can be combined into a 𝐒𝐒 matrix, which would scale the output. Since the effect of 𝐒𝐒 generally smaller than the effect of bias. A simplify calibration method is proposed to reduce errors without pre-processed 8 shaped movement (Grand and Thrun, 2012). Figure 2. The illustration of vector projection (a) and horizontal component transformation (b) This contribution has been peer-reviewed. This contribution has been peer-reviewed. p https://doi.org/10.5194/isprs-archives-XLIII-B1-2022-213-2022 \| © Author(s) 2022. CC BY 4.0 License. 2.2.3 Magnetic Heading Correction Figure 8 describes the concept of magnetic heading correction, dealing with the issue that magnetic field does not point to North in indoor environment. The magnetic anomalies are recorded in offline phase, and be compensated in online phase after the positioning is done. In other words, the heading of magnetic field is stored in grids by the procedure in Section 2.1.2 so as to provide a corrected reference for measured magnetic heading. As the equation (9) shows, the heading of magnetic field 𝜑𝜑𝑀𝑀 is derived from the known heading of smartphone 𝜑𝜑𝑘𝑘 in offline. Therefore, the measured heading 𝜑𝜑𝑚𝑚 is corrected by using 𝜑𝜑𝑀𝑀 as the equation (10) shows. where φ = heading t denotes time (sampling rate = 10 Hz) Figure 8. The illustration of magnetic heading correction 2.2.4 Extended Kalman Filter The states matrix and motion model of EKF is designed as equation (11) shows. The heading change ∆φ is the integration result of gyroscope output, which is the change angle in 0.1 second. The step length S has value only if the step is detected. Otherwise, S is equal to 0. Linearizing the equation can get the transition matrix and estimate the quality of prediction, i.e., P matrix. As Figure 7 shows, the trajectory from PDR can be transformed into a specific pattern, such as pink grids. These grids can change the magnetic series into spatial domain by using the time of steps t and the time of intersection T to resample the series. The transformed magnetic series with spatial pattern is used to match with offline map. In practice, the spatial pattern is slid through the whole map to extract the corresponding grids in difference position. The extracted corresponding grids constitutes the offline series. Therefore, the DTW based series-to-series matching is conducted to the find out the optimal position. It is crucial that how to determine the quality of new measurement, i.e., R matrix so as to update the system. Since the heading update per step, but the position update per 3 steps, there are two R matrixes respectively. The R matrix of position is determined by using an index, cumulative distance of DTW, to evaluate the quality of matching. Multiply the cumulative distance by a fine-tuned ratio to adjust that index. On the other hand, the quality of heading can be derived from the error propagation of magnetometer. Nevertheless, the quality of heading correction should be considered. Using P matrix to estimate the positioning error, which means the uncertainty of the using of orientation map. The average gradient of orientation in that uncertainty area is adopted to multiplied by the positioning error so that the R matrix of heading can be estimated. 𝑥𝑡𝜑𝑡𝜑𝑡𝑋𝑡𝑌𝑡𝑆𝑡𝑇 g p p Figure 7. The illustration of PDR trajectory pattern Figure 7. The illustration of PDR trajectory pattern 𝑥𝑥𝑡𝑡+1 = [𝜑𝜑𝑡𝑡+1 ∆𝜑𝜑𝑡𝑡+1 𝑋𝑋𝑡𝑡+1 𝑌𝑌𝑡𝑡+1 𝑆𝑆𝑡𝑡+1]𝑇𝑇 = [𝜑𝜑𝑡𝑡 ∆𝜑𝜑𝑡𝑡 𝑋𝑋𝑡𝑡 𝑌𝑌𝑡𝑡 𝑆𝑆𝑡𝑡]𝑇𝑇 + [∆𝜃𝜃𝑡𝑡 0 𝑆𝑆𝑡𝑡𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃𝑡𝑡) 𝑆𝑆𝑡𝑡𝑠𝑠𝑠𝑠𝑠𝑠(𝜃𝜃𝑡𝑡) 0]𝑇𝑇 (11) (11) 2.1.4 Magnetic Feature Map and Orientation Map 2.1.4 Magnetic Feature Map and Orientation Map 𝐻𝐻= user height 𝑆𝑆 𝐻𝐻= user height 𝑆𝑆 g p p Adopt the strategy mention in Section 2.1.3, the data can be divided into grids and be calibrated. The magnetic feature map can be generated after the procedure mention in Section 2.1.1. the vertical and horizontal component of each grid is stored. Also, 𝑆𝑆𝑆𝑆= step frequency This contribution has been peer-reviewed. This contribution has been peer-reviewed. 215 The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XLIII-B1-2022 XXIV ISPRS Congress (2022 edition), 6–11 June 2022, Nice, France The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XLIII-B1-2022 XXIV ISPRS Congress (2022 edition), 6–11 June 2022, Nice, France Figure 6. The illustration of step detection 𝜑𝜑𝑀𝑀= 𝜑𝜑𝑘𝑘−𝜃𝜃= 𝜑𝜑𝑘𝑘−𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎( 𝐻𝐻𝑥𝑥𝑜𝑜𝑜𝑜𝑜𝑜 𝑏𝑏 𝐻𝐻𝑦𝑦𝑜𝑜𝑜𝑜𝑜𝑜 𝑏𝑏 ), (9) 𝜑𝜑𝑚𝑚= 𝜑𝜑𝑀𝑀−𝜗𝜗= 𝜑𝜑𝑀𝑀−𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎൬ 𝐻𝐻𝑥𝑥𝑜𝑜𝑜𝑜 𝑏𝑏 𝐻𝐻𝑦𝑦𝑜𝑜𝑜𝑜 𝑏𝑏൰, (10) 𝜑𝑀 (9) (10) where 𝜑𝜑𝑀𝑀= the heading of magnetic field 𝜑𝑘 where 𝜑𝜑𝑀𝑀= the heading of magnetic field 𝜑𝑘 𝜑𝜑𝑀𝑀 g g 𝜑𝜑𝑘𝑘= known heading of offline smartphone 𝜑𝜑𝑚𝑚= measured heading of online smartphone 𝜃𝜃, 𝜗𝜗= angle between magnetic field and smartphone 𝐻𝐻𝑥𝑥𝑏𝑏, 𝐻𝐻𝑦𝑦𝑏𝑏= x and y component of H in body frame 𝑜𝑜𝑜𝑜denotes the online phase Figure 6. The illustration of step detection This contribution has been peer-reviewed. https://doi.org/10.5194/isprs-archives-XLIII-B1-2022-213-2022 \| © Author(s) 2022. CC BY 4.0 License. 2.2.2 Magnetic Fingerprint Matching A Dynamic Time Warping (DTW) based series-to-series matching method is adopted in magnetic fingerprint matching because the point-to-point matching may mismatch in some featureless area. A prerequisite of DTW is that two series should have the same reference. In fact, the magnetic series is a time series, but magnetic field varies spatially. Since the offline map is in spatial domain, the online magnetic series has to be transformed from time domain into spatial domain (Gong et al., 2018). Based on this method, an extended 2D matching method is proposed. 𝑜𝑜𝑜𝑜 denotes the online phase off denotes the offline phase 3.1.4 Online Test with 5 Different Paths An online test of proposed integrated positioning algorithm is conducted with the designed 5 different paths. Path 1 and path 2 are tested in 2021/Jan./13, Path 3 is tested in 2021/Jan./21 and Path 4 and path 5 are tested in 2021/Mar./9. The range of test distance is about 37 to 78. The conventional PDR algorithm would be implemented to compare with proposed algorithm. Both of two maps generated in different time would be used and be compared. In Section 4, the detail of the result is presented and discussed. g μ Figure 10. The photo of experiment 3.1.3 Magnetic Map Generation The main experiment is conducted in the underground parking garage below the library of NCKU, which is an indoor environment with the length of 38 m and the width of 7 m as Figure 12 shows. This area is divided into 7 by 38 grids, which have the length of 1m. The smartphone Xiaomi Mi 8 is used, logging the data of accelerometers, gyroscopes and magnetometer by AndroSensor APP. The offline magnetic feature map and orientation map is generated by using the proposed strategy in 2020/Dec./10 and 2021/Aug./04, respectively as Figure 13 shows. There are 8 months between two maps, but they have high similarity. This result verifies the temporal stability of magnetic anomalies is enough for the propose of fingerprint matching. g Figure 9. The magnetic feature map of preliminary test Figure 12. The illustration of grid division and the photo of experimental field Figure 9. The magnetic feature map of preliminary test Figure 12. The illustration of grid division and the photo of experimental field 3. EXPERIMENT 3.1.1 Preliminary Test of Data Collection A preliminary test is conducted to compare the strategy of data collection. The experimental field is a small classroom with the length of 12 m and the width of 8 m in National Cheng Kung University (NCKU). This area is divided into 8 by 12 grids, which have the length of 1m. The magnetic feature map collected grid by grid is showed as Figure 9(a), and the map collected by Figure 8. The illustration of magnetic heading correction 216 The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XLIII-B1-2022 XXIV ISPRS Congress (2022 edition), 6–11 June 2022, Nice, France The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XLIII-B1-2022 XXIV ISPRS Congress (2022 edition), 6–11 June 2022, Nice, France The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XLIII-B1-2022 XXIV ISPRS Congress (2022 edition), 6–11 June 2022, Nice, France walking along the line is showed as Figure 9(b). Both of them are calibrated by collecting data back and forth. By using correlation coefficient (CC), the similarity of two maps can be checked. The CC of vertical component, horizontal component and magnitude is 0.98, 0.92, 0.94, respectively. The result descripts that two maps have high similarity. It is feasible to replace the conventional method with efficient strategy. Furthermore, the CC of vertical component is higher than horizontal component. Because the errors of x and y axis mostly effect on horizontal component, the next online test use vertical component to do the matching. 3.1.2 Preliminary Test of Electro-Magnetic Effect Since the electrical device would cause magnetic anomalies because of the magnetic effect of electric current. Different from the magnetic anomalies caused by building, the electrical devices, such as laptop can be moved around, which may lead to an unstable magnetic feature. A preliminary test is designed to evaluate the effect from laptop. The experimental field is a corridor. Use wood shelf to put the laptop in different distance as Figure 10 shows, and walk through the laptop holding the smartphone with a height of 90 cm. The path with a length of 7 m is divided into 14 grids, and the magnitude of magnetic field (μT) of each grid is showed as Figure 11. Figure 13. The offline magnetic map Figure 13. The offline magnetic map The result descripts that the trend and the feature of 5 different tests is similar to each other. Even if the small difference exists, it appears to be system errors, uncalibrated bias, that would not defeat magnetic fingerprint matching. The electro-magnetic effect should be insignificant. The effect becomes significant only if the smartphone contact laptop directly. The magnitude of magnetic field can increase tens of μT. This contribution has been peer reviewed. -archives-XLIII-B1-2022-213-2022 \| © Author(s) 2022. CC BY 4.0 License. This contribution has been peer-reviewed. https://doi.org/10.5194/isprs-archives-XLIII-B1-2022-213-2022 \| © Author(s) 2022. CC BY 4.0 License. 5. CONCLUSION Compared the trajectories to the designed ground truth, the overall positioning errors are smaller than about 2m. The errors of endpoints are calculated as Table 1 shows. The results using map (a) is smaller than 1m, and the mean error is 0.66m. The results using map (b) is smaller than 1.6 m, and the mean error is 0.87m. The performance of map (b) is inferior to map (a), but the overall improvement compared with original PDR is about 90%, and the mean ratio of error to path distance is about 1.3%. The reason why map (a) performs better is that the time of generation is closer to the time of online test. Nevertheless, the results show the stability of magnetic anomalies again. It is reasonable to update the magnetic map in an about half-year period. This research focus on the development of magnetic fingerprint- based positioning and the integration with PDR. A smartphone based self-contained positioning algorithm is proposed to reduce the cost and become user-friendly. The 2D magnetic fingerprint matching, magnetic heading correction and the integration with PDR is the main contribution. The results of online tests demonstrate that it is suitable for the indoor positioning to use this integrated positioning algorithm, and the positioning accuracy can reach desk-level (1-3meters). Furthermore, an efficient procedure of offline magnetic map generation is developed, and the accuracy can meet the requirements. In this case, the magnetic map is generated by smartphone. Since the magnetic map has a high potential for navigation application, the high-level magnetometer, such as HMR2300 can be used to generate the map. By using the complete error model of magnetometer, a high accuracy magnetic map can be achieved, which might be attached to the HD-map, providing more spatial information for autonomous applications. The proposed algorithm has a large improvement on account of the worst heading of original PDR. In other words, the main contribution of this algorithm is the reduction of the cumulative errors in original PDR, especially in heading. Table 1. The error of the end of trajectory result Table 1. The error of the end of trajectory result 2020/Dec./10 2021/Aug./04 Path 1 Path 2 Path 3 Path 4 Path 5 Figure 14. The online test with 5 different paths The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XLIII-B1-2022 XXIV ISPRS Congress (2022 edition), 6–11 June 2022, Nice, France The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XLIII-B1-2022 XXIV ISPRS Congress (2022 edition), 6–11 June 2022, Nice, France The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XLIII-B1-2022 XXIV ISPRS Congress (2022 edition), 6–11 June 2022, Nice, France ACKNOWLEDGEMENTS Table 1. The error of the end of trajectory result 2020/Dec./10 2021/Aug./04 Path 1 Path 2 Path 3 Path 4 Path 5 Figure 14. The online test with 5 different paths The authors would like to thank The Ministry of the Interior (MOI), ROC (Taiwan) for providing financial support. 4. RESULTS AND DISCUSSION Figure 14 displays the results of online test with 5 different paths, where the gray rectangle is the coverage of magnetic map, the reddish-brown grids are the ground truth of designed paths. The results in left hand side use the map generated in in 2020/Dec./10 (map (a)), and the results in right hand side use the map generated in in 2021/Aug./04 (map (b)). Figure 10. The photo of experiment Figure 11. The comparison of magnitude The green trajectories are the results of the original conventional PDR algorithm, which use accelerometers and gyroscopes only. The trajectory drift with time and distance significantly, which is caused by the low-cost MEMS gyroscopes. The blue trajectories are the results of EKF integrated algorithm, and red points are the position update provided by magnetic fingerprint matching. The improvement of proposed algorithm is significant; moreover, the results using two maps perform similarly. Figure 11. The comparison of magnitude This contribution has been peer-reviewed. 217 This contribution has been peer-reviewed. https://doi.org/10.5194/isprs-archives-XLIII-B1-2022-213-2022 \| © Author(s) 2022. CC BY 4.0 License. This contribution has been peer-reviewed. https://doi.org/10.5194/isprs-archives-XLIII-B1-2022-213-2022 \| © Author(s) 2022. CC BY 4.0 License. This contribution has been peer-reviewed. s-archives-XLIII-B1-2022-213-2022 \| © Author(s) 2022 CC BY 4 0 License This contribution has been peer-reviewed. s-archives-XLIII-B1-2022-213-2022 \| © Author(s) 2022. CC BY 4.0 License. REFERENCES Chen, Ruizhi, Ling Pei, and Yuwei Chen, 2011. A smart phone based PDR solution for indoor navigation, In Proceedings of the 24th international technical meeting of the satellite division of the institute of navigation (ION GNSS 2011), pp. 1404-08. Chen, Ruizhi, Ling Pei, and Yuwei Chen, 2011. A smart phone based PDR solution for indoor navigation, In Proceedings of the 24th international technical meeting of the satellite division of the institute of navigation (ION GNSS 2011), pp. 1404-08. Path 2 Fang, J. C., H. W. Sun, J. J. Cao, X. Zhang, and Y. Tao, 2011. A Novel Calibration Method of Magnetic Compass Based on Ellipsoid Fitting, IEEE Transactions on Instrumentation and Measurement, Vol. 60, No. 6, pp. 2053-61. Path 3 Gong, Peiwen, Dongyan Wei, Xinchun Ji, Wen Li, and Hong Yuan, 2018. Research on Geomagnetic Matching Localization for Pedestrian, In China Satellite Navigation Conference (CSNC) 2018 Proceedings, Singapore, 2018, pp. 537-49. Path 4 Grand, E. Le, and S. Thrun, 2012. 3-Axis magnetic field mapping and fusion for indoor localization, In 2012 IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI), 13-15 Sept. 2012, pp. 358-64. Li, B., T. Gallagher, A. G. Dempster, and C. Rizos, 2012. How feasible is the use of magnetic field alone for indoor positioning?, In 2012 International Conference on Indoor Positioning and Indoor Navigation (IPIN), 13-15 Nov. 2012, pp. 1-9. Path 5 Yun, X., E. R. Bachmann, and R. B. McGhee, 2008. A Simplified Quaternion-Based Algorithm for Orientation Estimation From Earth Gravity and Magnetic Field Measurements, IEEE Transactions on Instrumentation and Measurement, Vol. 57, No. 3, pp. 638-50. Figure 14. The online test with 5 different paths This contribution has been peer-reviewed. 218
https://openalex.org/W4206998822	http://ira.lib.polyu.edu.hk/bitstream/10397/94314/1/s40691-021-00283-4.pdf	English	null	The relevance of breast motions and gaits in running exercises	Fashion and textiles	2,022	cc-by	10,814	© The Author(s), 2021. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the mate- rial. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creativecommons.org/licenses/by/4.0/. The relevance of breast motions and gaits in running exercises Jie Zhou1 , Qian Mao1, Jun Zhang2, Newman M. L. Lau2 and Jianming Chen3 Correspondence: 17802592059@163.com; alice.zj.zhang@connect. polyu.hk 1 School of Apparel and Art Design, Xi’an Polytechnic University, No. 19 Jinhua South Road, Xi’an, Shaanxi, China 2 School of Design, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong Full list of author information is available at the end of the article Correspondence: 17802592059@163.com; alice.zj.zhang@connect. polyu.hk Abstract The control of breast motions is a critical indicator to evaluate the comfort and func- tion of sports bras. If the breast motions can be predicted based on the gait parameters detected by wearable sensors, it will more economical and convenient to evaluate the bras. Thirteen unmarried Chinese females with a breast cup of 75B were recruited in this study to investigate the regularity of breast motions and the relevance between breast motions and gaits during running exercises. The breast motion indicator is the distance alteration of breast regions. The gaits were described by the rotation angles of the hip, knee, ankle joints, and the foot height off the ground. Firstly, the Mann-Whit- ney U test and the Kruskal-Wallis H test were utilized to analyze the motion diversity among the eight breast regions. Then, the gray correlation analysis was applied to explore the relevance between breast motions and gaits. Finally, the back-propagation neural network, the genetic algorithm, and the particle swarm optimization algorithm were utilized to construct the prediction models for breast motions based on gait parameters. The results demonstrate that the same breast regions on the bilateral breasts and the different breast regions on the ipsilateral breasts present a significant motion diversity. There is a moderate correlation between breast motions and gait parameters, and the back-propagation neural network optimized by the particle swarm optimization algorithm performs better in breast motion prediction, which has a coef- ficient of determination of 84.58% and a mean absolute error of 0.2108. Keywords: Breast motions, Gaits, Running exercises, Back-propagation neural network, Optimization algorithms Open Access Open Access Correspondence: 17802592059@163.com; alice.zj.zhang@connect. polyu.hk 1 School of Apparel and Art Design, Xi’an Polytechnic University, No. 19 Jinhua South Road, Xi’an, Shaanxi, China 2 School of Design, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong Full list of author information is available at the end of the article Zhou et al. Fashion and Textiles (2022) 9:3 https://doi.org/10.1186/s40691-021-00283-4 Zhou et al. Fashion and Textiles (2022) 9:3 https://doi.org/10.1186/s40691-021-00283-4 Introduction The female breasts primarily consist of external skins and internal structures (fibro-adi- pose and fibro-glandular tissues) (McGhee & Steele, 2020a). The thickness and elasticity of human skins decrease as age increases (Coltman et al., 2017; Den Tonkelaar et al., 2004). Besides, the configuration of the internal structure is affected by age, race, weight, hormones, and so on (Boyd et al., 2009; Huang et al., 2011; Lee et al., 1997; McGhee & Steele, 2020a). Specifically, the proportion of fibro-glandular tissue decreases while the fibro-adipose percentage increases by the aging process (Boyd et al., 2009; Huang et al., 2011; Lee et al., 1997). It was well-documented that the lack of breast support causes women to experience frequent breast pain during exercises (McGhee et al., 2007). Zhou et al. Fashion and Textiles (2022) 9:3 Page 2 of 21 Zhou et al. Fashion and Textiles (2022) 9:3 Zhou et al. Fashion and Textiles Hence, enough external support for breasts is necessary for women. However, according to the relevant surveys, 56% of females have experienced breast pain during exercises (Lorentzen & Lawson, 1987). The larger breasts are, the more painful the sensations that women feel. As bras providing with sufficient support are capable of reducing the exces- sive breast motions and even the perceived pains (Scurr et al., 2011), numerous stud- ies have been done to observe the characteristics of the breast motions among different exercises to improve bra support (McGhee et al., 2013; Scurr et al., 2011; Zhou et al., 2012). Due to a lack of advanced measurement techniques and awareness of breast motions, Gerhlsen and Albohm only investigated the absolute displacement of the breasts in the coronal plane (Gehlsen & Albohm, 1980). The displacement trajectory tended to a fig- ure-of-eight pattern. With the development of motion capture techniques, researchers began to investigate the breast motions in three dimensions (Scurr et al., 2009; Wood et al., 2012; Zhou et al., 2012). Awareness of the relative property of breast motions in three dimensions also improves the development of a series of new coordinate sys- tems and breast motion indicators (Milligan et al., 2015; Scurr et al., 2009; Zhou et al., 2009, 2012). Zhou et. al. (2012) established a breast coordinate system based on four torso markers to eliminate the torso motions from the breast motions. The motions of six breast points presented a butterfly shape. Introduction In addition, it was speculated that relative breast motions were caused by the breast deformations and the stretching or contract- ing behavior of pectoralis major muscle (McGhee & Steele, 2020a). Lots of numerical models have also been constructed to study the constitutive material coefficients for breast tissues. Cai et. al. (2018) established a mass-spring-damper model for simulating breast motions. The abstract model verified the diversity of breast motions among dif- ference regions. Most of the existing researches focused on a series of point motion for the breasts, whereas the point motion cannot effectively reflect the deformation of the breasts. Thus, this study aims to investigate the extension and contraction of the breast surface based on the distance alterations of different breast regions. This awareness may help bra designers to improve the bra support in terms of the bra structure and fabric elasticity. During exercises, breast motions are influenced by the torso driving force, the vis- coelastic mechanical properties, and the bra’s supporting ability (Haake & Scurr, 2010). Previous literature reported that the females may change their body posture to reduce the breast motions and pains (White et al., 2009). Milligan et. al. (2015) compared the motion patterns of human torso, pelvis, and upper arm during a 5 km running exer- cise. The results demonstrated that the high supportive bras perform better in terms of decreasing arm extension than the lower one. In addition, the breast displacements increase as the stride length exaggerates (Eden et al., 1992). Meanwhile, the rhythm of running or walking for females wearing sports bras was significantly lower than those baring breasts (Li et al., 2018). There might be a relationship between breast motions and gait patterns. Such experimental motion analyses are very restricted by their compli- cated setup, large workload of data cleaning and processing. If the breast motions can be predicted based on a wider breadth of gait patterns, the bra’s support will be evaluated, thus more information on bra performance is bale to be provided efficiently. Besides, by deeper perception of how lower limb movements influence breast motions, the bra Zhou et al. Fashion and Textiles (2022) 9:3 Zhou et al. Fashion and Textiles Page 3 of 21 designers can optimize the application-centered bra design. Therefore, it is necessary to explore the patterns of breast motions under different gaits. The results will provide a scientific basis for the design of sports bras. Introduction i The back-propagation (BP) neural network is a kind of feed-forward neural network with a simple structure and wide applications. The neural network can approximate any continuous function and square productive function with arbitrary accuracy (Zhou & Ma, 2019). Thus, this method has been widely applied to deal with non-linear problems such as girdle pressure prediction (Zhou & Ma, 2020), fabric surface detection (Jin et al., 2020), and flow stress prediction (Ding et al., 2020). However, the BP neural network has great defects in threshold and weight initialization. To avoid the BP neural network falling into the local optimum, researchers (Ding et al., 2020; Zhang & Guo, 2019; Zhou & Ma, 2020) usually adopted the genetic algorithm (GA) and the particle swarm opti- mization (PSO) to optimize the BP neural network. The GA is a computational model of biological evolution. This method searches for the optimal solution by simulating the natural evolutionary process (Zhang & Guo, 2019). The PSO is a kind of optimization algorithm based on iteration, and it can search the optimal threshold and weight of the BP neural network in a large space (Ding et al., 2020). The distance alteration of the breast regions was applied to investigate the charac- teristics and diversities of breast motions in this study to address the above mentioned issues. Meanwhile, four gait parameters, including the hip, the knee, the ankle joint motion angles, and the foot height off the ground, were utilized to explore the correla- tion between breast motions and gaits. Moreover, the BP, the back-propagation neural network optimized by the genetic algorithm (GA-BP), and the back-propagation neu- ral network optimized by the particle swarm optimization (PSO-BP) were adopted to establish the prediction models for breast motions based on gait parameters. There- fore, the following are the research questions to be answered. (1) What characteristics do the breast motions present in different parts during running exercise, and whether the motion differs in various breast regions? (2) What kind of correlation exists between breast motions and gait parameters? (3) If there is a desirable correlation between breast motions and gaits, the BP neural network optimized by the GA and PSO algorithms will be utilized to forecast breast motions. The mean absolute error and the coefficient of determination were determined to evaluate the accuracy of the models. Methods Subjects Fifteen young Chinese females were recruited for measuring their breast sizes. B-cup breasts are the most prevalent in unmarried Chinese females (Liang et al., 2007; Mao et al., 2020). Thus, thirteen subjects with B-cup were chosen and two females with other cups were excluded in this study. All subjects were given consent form and had no pregnancy or breast disease experience. Ethical approval was obtained from the ethics committee of Xi’an Polytechnic University. The essential physical characteristics of the subjects are shown in Table 1. Page 4 of 21 Zhou et al. Fashion and Textiles (2022) 9:3 Zhou et al. Fashion and Textiles Table 1 The essential physical characteristics of the subjects Physical characteristics Mean ± SD Physical characteristics Mean ± SD Age (years) 21.50 ± 1.12 Waist circumference (cm) 65.87 ± 4.33 Height (cm) 159.92 ± 4.69 Buttocks circumference (cm) 88.42 ± 3.26 BMI (kg/m2) 19.53 ± 0.51 Bi-acromial breadth (cm) 36.76 ± 1.42 Arm length (cm) 51.87 ± 2.73 Bust breadth (cm) 28.09 ± 1.31 Neck circumference (cm) 31.23 ± 1.69 Waist breadth (cm) 24.18 ± 1.86 Bust circumference (cm) 85.60 ± 1.69 Buttocks breadth (cm) 33.35 ± 1.21 Table 1 The essential physical characteristics of the subjects Table 2 The definitions of the marker positions Marker Definition Left/right bust points (L0/R0) The nipple points of the left/right breast Left/right breast lateral midpoints (L1/R1) The midpoints of the left/right breast nipple points and the lateral edge points Left/right breast upper midpoints (L2/R2) The midpoints of the left/right breast nipple points and the upper edge points Left/right breast medial midpoints (L3/R3) The midpoints of the left/right breast nipple points and the medial edge points Left/right breast lower midpoints (L4/R4) The midpoints of the left/right breast nipple points and the lower edge points Left/right hip joint points (Ll1/Rl1) The anterior superior iliac spine points of the left/right leg Left/right knee joint points (Ll2/Rl2) The lateral epicondyle points of the left/right leg femur Left/right ankle joint points (Ll3/Rl3) The external ankle points of the left/right leg fibula Table 2 The definitions of the marker positions Table 2 The definitions of the marker positions Marker Definition Data processing The gait cycles were divided according to the vertical coordinate of the Ll3. A complete gait cycle was defined as a nadir to another nadir in the vertical coordinate. Three com- plete consecutive gait cycles were extracted for each subject. As a result, a total of 39 gait cycles were obtained to analyze the breast motions and gaits. Breast motion parameters The displacement, velocity, and acceleration were utilized to investigate the breast point’s motions (Scurr et al., 2010; Zhou et al., 2009), however, the motions of the breast areas were ignored in these reports. The distance alteration of the breast regions ( d ) was applied in this study to explore the extension and contraction of the breast surface during exercises as shown in Eq. (1). The eight breast regions are divided as shown in Fig. 1. (1) dij = dij(t) −d′ ij, dij = dij(t) −d′ ij, (1) where dij(t) is the euclidean distance between breast point i and j at moment t, d′ ij is the mean euclidean distance between breast point i and j in a natural standing state. To improve the quality of the data, the normality and outliers of d were checked. Measurements of d that smaller than the first quartile—(1.5 × interquartile range) or larger than the third quartile + (1.5 × interquartile range) were considered as outliers (Shen et al., 2018). The change of the data before and after deleting the abnormal values is shown in Fig. 2. As shown in Fig. 2, the d data maintain the same distribution after eliminating the outliers. Compared with the right breasts, there are more abnormal values in the left breasts, especially in the d01 regions. Data capture The experiment was carried out in a quiet, windless laboratory, with 25 ± 2 °C and relative humidity of 65 ± 3%. The subjects were asked to run on a treadmill (slope of 0°) at 7 km/h with nude upper body and tight-fitting athletic pants. Sixteen reflective markers were pasted on their bodies. The three-dimensional (3D) coordinates of the reflective markers were collected at 120 Hz frequency by a Vicon 612 (Oxford Met- ric, UK) motion analysis system. The system is comprised of 12 M-series cameras and its mean residual is 0.605 mm. The positions of ten breast markers and six leg markers were defined to explore breast motions and gaits (Ren et al., 2015; Wang et al., 2019; Zhou et al., 2012). The definitions of the marker positions are shown in Table 2. Firstly, the subjects carried out a 3-min exercise on the treadmill following their running habits. Then, the subjects were asked to stand naturally and a professional technician pasted the reflective markers on their body based on the Table 2. Finally, the Vicon workstation software (v5.2.9) was utilized to track and record the 3D coordinates of the reflective markers during natural standing and running. No less than ten seconds of natural standing and one minute of running were recorded for three times. The data without missing values for each marker at the same time were adopted for this study. Zhou et al. Fashion and Textiles (2022) 9:3 Page 5 of 21 Zhou et al. Fashion and Textiles Prediction models for breast motions Prediction models for breast motions Prediction models for breast motions Numerous researches for breast motions are based only on running and they lack other exercises (McGhee & Steele, 2020a). With the development of wearable sensors, it is convenient and easy to detect the gaits outdoor (Prakash et al., 2018). If the breast motions are able to be predicted by gait parameters, the bra comfort and support will be evaluated in any situation. The BP neural network has excellent nonlinear mapping capa- bility (Zhou & Ma, 2019), while the GA and the PSO algorithms can effectively optimize the BP neural network. The BP, the GA-BP, and the PSO-BP algorithm were applied in this study to establish the prediction models for breast motions. Gait parametersh The vertical displacement of the breasts is closely related to foot strikes during running (McGhee & Steele, 2020b). Therefore, the foot height off the ground ( h ), the hip, the knee, and the ankle joint rotation angles ( α , γ , β ) were determined as the gait parameters in this study. And the illustration of four parameters (left leg) is shown in Fig. 3. The Ll3′ denotes the coordinates of the ankle joint points in a natural standing state, and the four parameters were calculated in Eqs. (2) and (3). Fig. 1 The breast regions Page 6 of 21 Zhou et al. Fashion and Textiles Zhou et al. Fashion and Textiles (2022) 9:3 where a is the euclidean distance between Ll1 and Ll2, b is the euclidean distance (2) a = (x1 −x2)2 + y1 −y2 2 + (z1 −z2)2 b = (x1 −x3)2 + y1 −y3 2 + (z1 −z3)2 c = (x2 −x3)2 + y2 −y3 2 + (z2 −z3)2 Fig. 2 The outliers of the d Fig. 3 Gait parameters (left leg) Fig. 2 The outliers of the d Fig. 2 The outliers of the d Fig. 3 Gait parameters (left leg) Fig. 3 Gait parameters (left leg) (2) a = (x1 −x2)2 + y1 −y2 2 + (z1 −z2)2 b = (x1 −x3)2 + y1 −y3 2 + (z1 −z3)2 c = (x2 −x3)2 + y2 −y3 2 + (z2 −z3)2 (2) where a is the euclidean distance between Ll1 and Ll2, b is the euclidean distance between Ll1 and Ll3, and c is the euclidean distance between Ll2 and Ll3. where a is the euclidean distance between Ll1 and Ll2, b is the euclidean distance between Ll1 and Ll3, and c is the euclidean distance between Ll2 and Ll3. where a is the euclidean distance between Ll1 and Ll2, b is the euclidean distance between Ll1 and Ll3, and c is the euclidean distance between Ll2 and Ll3. Page 7 of 21 Page 7 of 21 (2022) 9:3 Zhou et al. Fashion and Textiles (3) α = arccos y1 −y2 a , β = arccos y3 −y2 c , γ = 180 ◦−arccos a2 + c2 −b2 2ac , h = z3−z′ 3. (3) α = arccos y1 −y2 a , β = arccos y3 −y2 c , γ = 180 ◦−arccos a2 + c2 −b2 2ac , h = z3−z′ 3. Gait parametersh (3) The principle for BP neural network Matlab 2017B version (MathWorks, USA) was utilized in this study to construct a BP neural network reflecting the nonlinear mapping relationship between breast motion indicators and gait parameters. Meanwhile, the gray correlation degree was adopted to calculate the weight of gait parameters ( ξ ) related to d . The ξ was calculated in Eq. (4). (4) ξij = δij 8 i=1 δij , ξij = δij 8 i=1 δij , (4) where i and j are the gait parameters and the d , respectively. δij denotes the gray cor- relation degree between i and j. The results of ξ are shown in Table 3. There were nine nodes in the input layer, including eight weighted gait parameters and breast region numbers (the breast regions d01(R), d02(R), d03(R), d04(R), d01(L), d02(L), d03(L), d04(L) were set as 1, 2, 3, 4, 5, 6, 7, 8, respectively). The output layer was the value of d . The optimal number of hidden layer nodes was mainly obtained by the trial-and-error method and the mean absolute error (MAE) was utilized as the evaluation index. The number was first set to five and then gradually increased to 50. When the number was 35, the MAE reached the smallest and the BP neural network performed better. Thus, Table 3 Weights of gait parameters related to d Gait parameters d01(R) d02(R) d03(R) d04(R) d01(L) d02(L) d03(L) d04(L) α(L) 0.13 0.12 0.12 0.12 0.13 0.13 0.14 0.13 β(L) 0.12 0.12 0.12 0.12 0.12 0.12 0.11 0.12 γ(L) 0.12 0.13 0.13 0.13 0.12 0.12 0.12 0.12 h(L) 0.12 0.12 0.13 0.12 0.13 0.13 0.12 0.13 α(R) 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.13 β(R) 0.13 0.13 0.13 0.13 0.12 0.12 0.12 0.12 γ(R) 0.12 0.12 0.12 0.12 0.12 0.12 0.12 0.12 h(R) 0.13 0.13 0.13 0.13 0.12 0.12 0.12 0.12 Table 3 Weights of gait parameters related to d Zhou et al. Fashion and Textiles (2022) 9:3 Zhou et al. Fashion and Textiles Page 8 of 21 the topology of the BP neural network was finally determined to be 9-35-1, as shown in Fig. 4. the topology of the BP neural network was finally determined to be 9-35-1, as shown in Fig. 4. There are 21,827 sets of data in the sample set. In this study, 21,500 sets of data were randomly selected as the training set of the models, and the rest were the test set (327). The principle for BP neural network In order to accelerate the convergence of the BP neural network, the input and output data were normalized to the range of [− 1, 1] based on the Eq. (5). (5) y = ymax −ymin (x −xmin) xmax −xmin + ymin, (5) where x and y represent the original data and normalized data, respectively. The transfer functions of the BP neural network include linear function (purelin) and sigmoid function (logsig and tansig). The output of the purelin function is an arbitrary value. The logsig function maps the neuron from (− ∞, + ∞) to (0, 1), while the tansig function maps the neuron from (− ∞, + ∞) to (− 1, 1). Therefore, the tansig and the purelin were determined as the transfer function of the hidden and output layer, respec- tively. Meanwhile, the training function of the models is trainlm in this study. The maxi- mum number of iteration steps, the training target, the learning rate, and the number of verification failures were set as 1000, 0.001, 0.01, and 6, respectively. i In order to avoid the BP neural network falling into the local optimum, we applied the GA and PSO algorithms to optimize the initial weights and thresholds for the BP neural network. Meanwhile, the topology of the BP neural network was maintained. The principle for GA‑BP neural network The j-th gene of the i-th individual is selected for mutation operation as shown in Eqs. (10) and (11). (10) aij = aij + aij −amax f g , r > 0.5 aij + amin −aij f g , r ≤0.5 , (10) (11) f g = r1 1 − g Gmax 2 , (11) where amax is the maximum value of aij and amin is the minimum value of aij; g is the number of iterations; Gmax is the maximum number of evolution; r1 is a random number, and r is a random number between 0 and 1. After the evolution of the GA algorithm, the optimal individuals were obtained. Finally, the prediction model was established by training the neural network. The spe- cific process is shown in Fig. 5. The principle for GA‑BP neural network The principle for the GA algorithm is to utilize individuals to represent the initial weights and thresholds of the BP neural network. Meanwhile, the GA algorithm takes the prediction error of the initial neural network as an individual fitness value to find the optimal initial parameters (Ding et al., 2020). The parameters of the GA algorithm were initialized firstly in this study. After repeated tests, the population size, the maxi- mum number of iteration, the crossover probability, and the mutation probability were determined to be 8, 50, 0.4, and 0.2, respectively. Then, the real coding was utilized for individual coding, and the length of coding was calculated in Eq. (6). Fig. 4 The topology of BP neural network Fig. 4 The topology of BP neural network Zhou et al. Fashion and Textiles (2022) 9:3 (2022) 9:3 Page 9 of 21 Zhou et al. Fashion and Textiles Page 9 of 21 (6) S = S1S2 + S1S3 + S1 + S3, S = S1S2 + S1S3 + S1 + S3, (6) where S1 is the number of hidden layer nodes, S2 and S3 are the number of the input layer and output layer nodes, respectively. The next step is to select the fitness function. The individual was utilized to represent the initial weights and thresholds of the neural network, and the absolute error of the neural network initialized by the individual was adopted as the fitness value F as shown in Eq. (7). (7) F = k S3 i=1 yi −oi , (7) where k is the coefficient, yi and oi are the expected output and the predictive output, respectively. The fitness ratio method was adopted to carry out the selection operation of the genetic algorithm. If an individual is set as Xi and its fitness is Fi, the selection probability is Pi as shown in Eq. (8). (8) Pi = Fi S i=1 Fi . Pi = Fi S i=1 Fi . (8) The process of gene recombination is known as cross operation. The cross mode of the k chromosome ak and the l chromosome al at the j position are shown in Eq. (9). (9) akj = akj(1 −b) + aljb alj = alj(1 −b) + akjb , (9) where b is a random number between 0 and 1.h where b is a random number between 0 and 1. The principle for PSO‑BP neural network The PSO algorithm is a parallel random search algorithm, which performs global searches under individual fitness conditions. In this process, the speed and position Page 10 of 21 Zhou et al. Fashion and Textiles (2022) 9:3 Zhou et al. Fashion and Textiles Fig. 5 The process of the GA-BP algorithm Fig. 5 The process of the GA-BP algorithm of each particle are constantly updated to accurately approach the target within a short period (Ding et al., 2020). The PSO algorithm also requires population initiali- zation like the GA algorithm, but it has no selection, crossover, and mutation opera- tions. Herein, the population size and the maximum number of iterations were also determined as 8 and 50, respectively. In the searching space, the PSO algorithm randomly initializes a group of particles. Each particle represents a potential optimal solution to the extreme value optimiza- tion problem. The fitness value of each particle is calculated according to the Eq. (7). The particle updates the individual position in the searching space by tracking the individual extreme value (Pbest) and the group extreme value (Gbest). In each itera- tion, the particle updates its speed and position through individual extreme value and group extreme value, as shown in Eq. (12). (12) vij(t + 1) = ωvij(t) + c1r2 Pbest(t) −xij(t) + c2r3 Gbest(t) −xij(t) xij(t + 1) = xij(t) + vij(t + 1) , (12) where vij represents the j-th dimension component of the i-th particle velocity. The par- ticle velocity was limited between [− 1, 1] in this study. t is the current iteration number, and ω is the inertia weight, herein, its range was set in [0.4, 09]. c1 and c2 are acceleration constants, herein, c1 = c2 = 1.5; r2 and r3 are random numbers between 0 and 1; xij is the j-th dimension component of the position of particle i. The motion range of the particles was set in [− 5, 5]. Page 11 of 21 Zhou et al. Fashion and Textiles Zhou et al. Fashion and Textiles (2022) 9:3 At the end of the PSO algorithm, the optimal initial weights and thresholds were out- put, and the prediction model was acquired by training the network. The specific pro- cess is shown in Fig. 6. Breast motions Some reports have found that there is a significant difference between the left and right breasts in terms of morphology and kinematics (Pei et al., 2019; Ren et al., 2016). The breast motions also exhibit considerable variability in different breast parts (Mason et al., 1999). The study aims to investigate the difference of breast extension and contrac- tion between breast regions, SPSS 20.0 version (IBM, USA) was applied for the statisti- cal analysis of d . Since the d were detected to be non-parametric data by the test for normal distribution, the Mann–Whitney U test was applied to analyze the motion diver- sity between the same regions on the bilateral breasts. Meanwhile, the Kruskal–Wallis H test was utilized to analyze the motion difference between different regions (d01, d02, d03, d04) on the ipsilateral breasts. All significance levels were predetermined at p ≤ 0.05. The results are shown in Table 4. As for the d , the positive and negative values represent the extension and contraction of breast regions, respectively. Combined with the d distribution in Fig. 2, most breast regions present both extension and contraction, except for the d01(L) region. According Fig. 6 The process of the PSO-BP algorithm Fig. 6 The process of the PSO-BP algorithm Fig. 6 The process of the PSO-BP algorithm Page 12 of 21 Zhou et al. Breast motions Fashion and Textiles Table 4 Statistical characteristics of d (unit: mm) Indicates a significant difference between the same regions on bilateral breasts (left and right breast) # Indicates a significant difference between the different regions (d01, d02, d03, d04) on ipsilateral breasts Breast region d01 d02 d03 d04 L R L R L R L R Min 0.08 − 1.18 − 0.86 − 1.81 − 1.22 − 0.77 − 1.12 − 1.59 Max 2.17 2.61 2.01 1.70 0.90 1.44 1.40 2.48 Range 2.09 3.79 2.87 3.51 2.12 2.21 2.52 4.07 Mean 1.18 0.61 0.58 − 0.01 − 0.23 0.26 0.14 0.42 SD 0.40# 0.80# 0.56# 0.69# 0.43# 0.42# 0.48# 0.94# Table 4 Statistical characteristics of d (unit: mm) *Indicates a significant difference between the same regions on bilateral breasts (left and right breast) # Indicates a significant difference between the different regions (d01, d02, d03, d04) on ipsilateral breasts to the average values, the motion pattern of the d01(L), the d01(R), the d02(L), the d03(R), the d04(L), and the d04(R) regions are mainly extension, while the d02(R) and d03(L) regions take contraction as the primary motion. In the d01 and d02 areas, the breast motions are more distinct in the left than in the right, and the reverse phenomenon occurs in the d03 and d04 breast regions. The phenomenon demonstrates that the motions of breast regions vary between the left and right breasts. As for the upper and outer breast regions, the motions are more intense in the left breasts, whereas the lower and inner breast regions move more intensely in the right breasts. In terms of the range values, the four regions on the right breasts have a wider range of stretch than the left breasts, and the maximum value (4.07 mm) occurs in the d04(R). The result indicates that the breast motions are more active in the right breasts. The squared deviation (SD) indicates the dispersion degree of the breast motion indicators. Except for the d03 areas where the SD values are similar in bilateral breasts, the other regions have different SD values between the right breasts and the left breasts. The SD values in the right breast regions were much greater than those in the left breasts. The result indi- cates that the right breast motions are more scattered and irregular than the left breasts. Breast motions As shown in Table 4, there are significant motion differences in the same regions on the bilateral breasts or the different regions on the ipsilateral breasts according to the two non-parametric tests. Numerous reports have verified the morphological asymmetry of bilateral breasts (Losken et al., 2005; Pei et al., 2019). Based on the investigation of local breast motions, the asymmetry of bilateral breast motions was found in this study. Therefore, the bra producers and designers are advised to improve the bra comfort and support based on the differences in both breast morphology and motions. Relevance of breast motions and gaits Due to the influence of trunk displacements, breast motions also present a periodic pattern (McGhee et al. 2013). Besides, the intensity of breast motions may lead to gait changes during exercises (Eden et al., 1992; Li et al., 2018). Thus, whether a cor- relation exists between breast motions and gaits is speculated. The mean values of parameters were calculated for correlation analysis. The duration of three cycles var- ies with individual subjects for the unavoidable changes in the arm swing amplitude (Zhou et al., 2012). Therefore, the gait cycles were reshaped by the reshape function in Pandas of Python. The maximum duration of cycles was determined as the new Page 13 of 21 Zhou et al. Fashion and Textiles Zhou et al. Fashion and Textiles (2022) 9:3 Fig. 7 Variations of gait parameters with time (left leg) Fig. 8 Variations of d with time Fig. 7 Variations of gait parameters with time (left leg) Fig. 7 Variations of gait parameters with time (left leg) Fig. 8 Variations of d with time Fig. 8 Variations of d with time duration, so the linear interpolation method was applied to increase the samples. The average gait and breast motion parameters for 13 subjects over time are shown in Figs. 7 and 8, respectively. As shown in Fig. 7, the curve of h with time is similar to the sine curve. The wave shapes of the lower limb rotation angles including the β , the γ , and the α are simi- lar to each other, and the curves all present a ‘W’ shape with a single peak and two troughs. Although the h curve is different from three other gait parameters, the four curves all present a stable periodic structure. Their peaks and the troughs also exhibit a period of synchronicity. On top of this, the hip rotation angle α has the smallest value, and the ankle rotation angle β has the relatively minimum range of change (26.89°). The knee rotation angle γ has the largest value and the maximum change range (42.98°). The result reveals that the knee rotation joint has the maximum motion amplitude while the ankle rotation joint has the minimum motion amplitude. Zhou et al. Fashion and Textiles (2022) 9:3 Page 14 of 21 Zhou et al. Fashion and Textiles (2022) 9:3 Zhou et al. Relevance of breast motions and gaits Fashion and Textiles Figure 8 shows that the d exhibits a continuous periodicity with time and the curves approximately present a periodic ‘V’ shape. There are two breast motion cycles in a gait cycle according to the gait cycle division. Thus, the period of breast motions is half of the gaits. The phenomenon demonstrates that breast motions are more frequent than trunk movements. The event is perhaps a critical factor to cause breast pain. Compared to the gait parameters, the d values include more noise and present a relatively large fluctua- tion. As a result, the main features of breast motions are obscured and it may be detri- mental to the correlation analysis between breast motion and gait. In order to reduce the noise of d , the wavelet threshold de-noising method was applied in this study to process the breast motion indicators. The result is shown in Fig. 8. In a single gait cycle, the patterns of the β , the γ , and the α are similar to the wave shape of the breast motion indicators. They all show a ‘W’ shape. Thus, the knee rotation angle γ (left leg) having the largest amplitude was determined to compare with the breast motions. The results are shown in Fig. 6. Meanwhile, the Mann–Whitney U test was applied to analyze the variability of two breast motion cycles in a gait cycle. The significance level was prede- termined at p ≤ 0.05. According to Fig. 9, although there are differences in the peaks and troughs of d and γ , the curves both show a ‘W’ shape within the same gait cycle. For the knee rotation angle γ , the peaks are more bulging while the troughs are a little concave. Meanwhile, the peaks and troughs have similar extents in the breast motion curves. In terms of the appearance times of the wave peaks and troughs, the peaks of d for the d03(R), the d01(L), the d02(L), and the d04(L) regions are consistent to the γ , respectively. It is interesting to observe that the peaks of d for the d01(R), the d04(R), and the d03(L) regions lag behind the γ , while the crests of d for the d02(R) are ahead of the γ . The appearance times of troughs within a gait cycle show another characteristic. Relevance of breast motions and gaits The first valley of d always lags behind the γ , whereas the second valley of d advances or synchronizes with the γ . The time lag of peaks and troughs may be due to the lack of bone and muscle in the breasts or the viscoelasticity variability between different breast parts (Cai et al., 2018; McGhee & Steele, 2020a). Although the expansion and contraction of the breasts show a V-shaped cycle in gen- eral, there are some phenomena of ’rebound’ in the wave troughs. The phenomena illus- trate that breast motions have a non-linear and complex character. The breast motion Fig. 9 Variations of d and γ with time Fig. 9 Variations of d and γ with time Page 15 of 21 Zhou et al. Fashion and Textiles (2022) 9:3 curves were divided into two cycles in a gait cycle, and the proportion of each breast motion cycle (breast motion cycle/gait cycle × 100%) was calculated. The results find that the first cycle of d is longer than the second cycle except for the d02(R) and the d03(R) regions. According to the results of the Mann–Whitney U test, breast regions have sig- nificant differences between the two cycles except for d02(L) and more significant differ- ences exist for d03(R), d01(L), and d04(L) regions. The results mean that the breast motions are more intense and complex than the gaits and they lead to the poor stability of the d curves. Gray relational analysis is a quantitative method to describe and compare the develop- mental dynamics of a system. The larger the gray correlation degree ( δ ), the more simi- lar the variation trend of the two variables, and vice versa (Xiao, 1997). The strength of the correlation is defined as weak δ ≤0.49 , moderate δ ≥0.50 to δ ≤0.74 , and strong δ ≥0.75 (Bridgman et al., 2010). The gray correlation degrees between parameters are shown in Fig. 10. From Fig. 10, the gait parameters have the smallest δ value between the left and right leg, and the maximum value among them is only 0.62. However, the δ values between the ipsilateral gait parameters are all above 0.75. The gray correlation degree between α and Fig. 10 Grey correlation degrees between breast motion indicators and gait parameters Zhou et al. Fashion and Textiles (2022) 9:3 Page 16 of 21 Zhou et al. Fashion and Textiles Fig. Relevance of breast motions and gaits 11 The predictive results of the neural network models Fig. 12 The R2 between experimental and predictive values: a PSO-BP, b GA-BP, c BP Fig. 11 The predictive results of the neural network models Fig. 11 The predictive results of the neural network models Fig. 12 The R2 between experimental and predictive values: a PSO-BP, b GA-BP, c BP Fig. 12 The R2 between experimental and predictive values: a PSO-BP, b GA-BP, c BP γ for the right leg reached the maximum (0.92). Moderate δ values are observed between the breast motion indicators, with a minimum value of 0.68. The δ values between d04(L) and other breast regions are the highest, while the gray correlation degree between d03(L), d04(R) and other breast regions are the lowest. The result indicates that the d03(L) and d04(R) areas have more complex and irregular motions. As the δ values between gait parameters and breast motion indicators are in the range of [0.6, 0.75], there is a moder- ate correlation between these two types of parameters. Therefore, the prediction models for breast motions based on gait parameters can be established in this study. Forecast of breast motions To compare the accuracy of the PSO-BP, the GA-BP, and the BP algorithms, the input and output data of the models were set up consistently in this study. The predictive values of the test set were output after training the three prediction models for breast motions. The predictive values and the experimental values of d are shown in Fig. 11. h Figure 11 shows that the values of d predicted by the PSO-BP, the GA-BP, and the BP neural network models differ lightly from the experimental data. Thus, the breast motion indicators in this study can be effectively predicted by the above three models based on the gait parameters. To quantify the performance of the models, we adopted the coefficient of determination (R2) and the MAE to evaluate the prediction accuracy of the algorithms, as shown in Eqs. (13) and (14). The results are shown in Fig. 12 and Table 5, respectively. Page 17 of 21 Zhou et al. Fashion and Textiles (2022) 9: Zhou et al. Fashion and Textiles Table 5 The MAE of PSO-BP, GA-BP, and BP models Table 5 The MAE of PSO-BP, GA-BP, and BP models The bold values are the top 20% of the MAE values d PSO-BP GA-BP BP d01(R) 0.1682 0.1888 0.1179 d02(R) 0.2330 0.2391 0.2316 d03(R) 0.1406 0.132 0.1354 d04(R) 0.2234 0.2335 0.2429 d01(L) 0.1251 0.1309 0.1465 d02(L) 0.1269 0.145 0.1501 d03(L) 0.1407 0.1549 0.1972 d04(L) 0.1429 0.188 0.1813 Overall 0.2108 0.2289 0.2371 (13) R2 = 1 − m i=1 d(i) −d′(i)2 m i=1 d′(i) −d′ 2 , (14) MAE = 1 m m i=1 d(i) −d′(i), (13) R2 = 1 − m i=1 d(i) −d′(i)2 m i=1 d′(i) −d′ 2 , (13) (14) MAE = 1 m m i=1 d(i) −d′(i), (14) where d and d′ are the experimental and predictive value of the models respectively; d′ is the average of the predictive values, and m indicates the sample size of the test set. Figure 12 shows that the R2 of the PSO-BP, the GA-BP, and the BP neural network models reaches more than 80%. The R2 of the GA-BP and the BP neural network models are similar to each other, while the R2 of the PSO-BP model has the maximum value of 84.58%. The bold values are the top 20% of the MAE values Discussion The awareness of breast motion patterns is significant to the optimization of bra comfort and support. A scientific and reasonable parameter utilized to quantify breast motions makes the results more comprehensive and reliable. Numerous reports have described the breast motions based on the points (Scurr et al., 2010; Zhou et al., 2009), but these studies were unable to reflect the motions of the breast regions. Herein, the distance alteration between breast points ( d ) was utilized to analyze the breast motions. The indicator aims to explore the line variation of different breast regions and it reflects the extension and contraction of breast regions. In this study, the results reveal that the main breast motions are the extension. Whether the same breast regions on the different sides of the breasts, or the various breast regions on the same side of the breasts, they all present a significant diversity in breast motions. The phenomenon may be related to the asymmetrical mass of the breasts. There are no consistent results for the motion inten- sity in different breast regions, and the motions in the right breasts show relatively larger fluctuation than that in the left breasts. The breast displacement of females varies mainly in the vertical direction during the two-step jumping exercise, and the vertical dis- placement accounts for 58% of the breast displacements (Bridgman et al., 2010). When females exercise on a treadmill at 10 km/h, the vertical, mediolateral, and anteroposte- rior components account for an average of 56%, 22%, and 19% of breast motions, respec- tively (Zhou et al., 2012). However, there is a dissimilar result for the breast motions in this study. The breast motions obtain maximum and minimum values in the d01 and d03 regions respectively, which demonstrates that the motions are more active in the lateral breast areas, followed by the vertical breast areas. Meanwhile, the medial breast areas present the slightest motions. The discrepancies with previous reports may be due to the different indicators of breast motions. f As for the breast motions and the trunk movements, some reports claimed that they all present a pattern of the sinusoidal curve (Haake & Scurr, 2010). However, the foot height off the ground is the only parameter founded in this study that shows a sinusoidal curve with time. Forecast of breast motions The result indicates that these three neural network models have the ability to predict the breast motions, and the PSO-BP model is more suitable for breast motion prediction in this study. As shown in Table 5, the overall MAE of the PSO-BP model is 0.2108, which is 11% lower than that of the BP neural network. Meanwhile, the overall MAE of the GA-BP model is 0.2289, which is 3% lower than that of the BP neural network. The phenomenon indicates that both the PSO and GA algorithms have the ability to improve the predic- tion accuracy of the BP neural network for breast motions. The PSO-BP model has the best prediction accuracy. Among the three prediction models for breast motions, the MAE of d02(R) and d04(R) are larger than other breast regions. The result demonstrates that the motion prediction in these two breast regions has poor performance. According to the MAE of the PSO-BP model, the left breasts have better prediction accuracy than the right breasts. Therefore, it is more effective to predict the left breast motions based on the gait parameters. The result might be related to the stretch ranges and the disper- sion of the breast motions. Page 18 of 21 Zhou et al. Fashion and Textiles (2022) 9:3 Zhou et al. Fashion and Textiles Discussion The rotation angles of the lower limb joints and the distance alterations of the breast regions all show a ‘W’ shape during a gait cycle. It is interesting to note that the breast regions exhibit a ‘rebound’ phenomenon during exercises. In research of dynamic model for the breasts, the ‘double bounce’ phenomenon was also observed at the mini- mum value of the breast motion curves (Haake & Scurr, 2010), and the researchers sug- gested that this phenomenon might be related to the dynamic forces of heel and forefoot strike during contact. Besides, there is a time delay between the curves of certain breast areas and that of limb rotation angles. The phenomenon is also discovered in the study investigating the absolute and relative displacement of breasts during walking and run- ning (Scurr et al., 2009). In the report, both the vertical and resultant breast displace- ments show a ‘bimodal distribution’ within a single gait cycle and the peaks and troughs of breast motion curves lag slightly behind the temporal division of the gaits. Once the subject’s heel hits the ground during running, that is, at the lowest point where the foot is high off the ground, the trunk suddenly decelerates and falls vertically. Nevertheless, the soft tissue of the breasts continues to move downward due to the inertia, and the breasts only begin to decelerate when the trunk starts to move upward (McGhee et al., Zhou et al. Fashion and Textiles (2022) 9:3 Zhou et al. Fashion and Textiles Page 19 of 21 2010). The time delay in trunk-breast motions causes the breasts to slap down against the chest wall and thus leads to the breast pain (Scurr et al., 2009; McGhee et al. 2013). According to this analysis, there is a moderate correlation between breast motions and gaits. Thus, the BP neural network, the PSO, and the GA algorithms were applied in this study to develop three prediction models for breast motions. The R2 of these three mod- els reaches more than 80%, and the PSO-BP model obtains the maximum R2 of 84.58%. Meanwhile, the PSO-BP model also has the smallest overall MAE of 0.2108. Therefore, the PSO-BP algorithm is considered to be more appropriate for breast motion prediction in this study. Discussion The validity of the prediction model allows researchers to predict breast motions with gait parameters, and the method can facilitate the investigation of breast motions in different environments lacking the access to measure the breast motions. f There are some limitations in this study, such as the correlation between breast motions and human gaits is only investigated under the bare breasts. For further research, it is vital to explore the difference of breast motions between the bare breasts and the bra-wearing breasts. Furthermore, the influence of bra structures, fabrics, and styles for the breast motions should also be analyzed to reduce breast expansion and contraction during exercises. Besides, the methods in this study are also applicable to investigate the breast motions for other subjects. The awareness of breast motions among different females in a series of exercises is helpful to improve bra support and comfort. Authors’ contributions Authors contributions ZHOU J, MQ were responsible for the experiment design and whole structure construction, while MQ, ZHANG J and L.M.L.N was responsible for the data collection, data cleaning, and MQ was responsible for data processing and model- ling. The manuscript drafted by MQ and ZHOU J, finally revised by CJM, ZHANG J, L.M.L.N, MQ, ZHOU J. All authors read and approved the final manuscript. Jie ZHOU, Professor, School of Apparel and Art Design, Xi’an Polytechnic University, No. 19 Jinhua South Road, Xi’an, Shaanxi, China. Qian MAO, Master Student, School of Apparel and Art Design, Xi’an Polytechnic University, No. 19 Jinhua South Road, Xi’an, Shaanxi, China. Jun ZHANG, PhD Student, School of Design, The Hong Kong Polytechnic University, Conclusions In this study, the distance alteration of the breast regions ( d ) was determined to evalu- ate breast motions. According to the analysis of the breast motions, a significant motion difference appears in the same regions on the bilateral breasts and the different regions on the ipsilateral breasts. Meanwhile, the d-time curves of eight breast regions show the same continuous ‘V’ pattern. Within a gait cycle, four gait parameters (the hip, the knee, the ankle joint rotation angles, and the foot height off the ground) demonstrate one period while breast motions present two periods. At the same time, the curves of the hip, the knee, the ankle joint rotation angles, and the d perform one peak and two troughs respectively. However, there is a slight difference in the occurred time of peaks and troughs. In this case, there is a moderate correlation between breast motion indica- tors and gait parameters in this study. Therefore, the BP neural network combined with optimization algorithms (the GA and PSO algorithms) is valid to predict breast motions based on gait parameters. Among these three models, the PSO-BP algorithm has the highest prediction accuracy which has an R2 of 84.58% and an MAE of 0.2108. Acknowledgements Not applicable. Competing interests p g The authors declare that they have no competing interests. Ethics declarations h h d This research was conducted under the exemption and supervision of The Hong Kong Polytechnic University Institu- tional Review Board(IRB Exemption No. HSEARS20200205001) regarding ethical issues. Author’s information Jie ZHOU, Professor, School of Apparel and Art Design, Xi’an Polytechnic University, No. 19 Jinhua South Road, Xi’an, Shaanxi, China. Qian MAO, Master Student, School of Apparel and Art Design, Xi’an Polytechnic University, No. 19 Jinhua South Road, Xi’an, Shaanxi, China. Jun ZHANG, PhD Student, School of Design, The Hong Kong Polytechnic University, Page 20 of 21 Zhou et al. Fashion and Textiles (2022) Hung Hom, Kowloon, Hong Kong. Newman M.L.LAU, Associate Professor, School of Design, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong. Jianming CHEN, Department of Biomedical Engineering, The Chinese University of Hong Kong, Shatin, New Territories, Hong Kong. Availability of data and materials a ab ty o data a d ate a s The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Received: 29 January 2021 Accepted: 27 October 2021 Received: 29 January 2021 Accepted: 27 October 2021 References Boyd, N., Martin, L., Chavez, S., Gunasekara, A., Salleh, A., Melnichouk, O., Yaffe, M., Friedenreich, C., Minkin, S., & Bronskill, M. (2009). Breast-tissue composition and other risk factors for breast cancer in young women: A cross-sectional study. The Lancet Oncology, 10(6), 569–580. https://doi.org/10.1016/S1470-2045(09)70078-6 Boyd, N., Martin, L., Chavez, S., Gunasekara, A., Salleh, A., Melnichouk, O., Yaffe, M., Friedenreich, C., Minkin, S., & Bronskill, M (2009). Breast-tissue composition and other risk factors for breast cancer in young women: A cross-sectional study. The Lancet Oncology, 10(6), 569–580. https://doi.org/10.1016/S1470-2045(09)70078-6 gy p g Bridgman, C., Scurr, J., White, J., Hedger, W., & Galbraith, H. (2010). Three-dimensional kinematics of the breast during a two-step star jump. Journal of Applied Biomechanics, 26(4), 465–472. https://doi.org/10.1123/jab.26.4.465 Cai, Y., Chen, L., Yu, W., Zhou, J., Wan, F., Suh, M., & Chow, D. H. K. (2018). A piecewise mass-spring-damper model of the human breast. Journal of Biomechanics, 67, 137–143. https://doi.org/10.1016/j.jbiomech.2017.11.027 Coltman, C. E., Steele, J. R., & McGhee, D. E. (2017). Effect of aging on breast skin thickness and elasticity: Imp breast support. Skin Research and Technology, 23(3), 303–311. https://doi.org/10.1111/srt.12335 Den Tonkelaar, I., Peeters, P. H. M., & Van Noord, P. A. H. (2004). Increase in breast size after menopause: Prevalence and determinants. Maturitas, 48(1), 51–57. https://doi.org/10.1016/j.maturitas.2003.10.002 Ding, F. J., Jia, X. D., Hong, T. J., & Xu, Y. L. (2020). Flow stress prediction model of 6061 aluminum alloy sheet based on GA-BP and PSO-BP neural networks. Rare Metal Materials and Engineering, 49(6), 1840–1853. https://chn.oversea.cnki. net/kcms/detail/detail.aspx?FileName=COSE202006003&DbName=CJFQ2020 Eden, K. B., Valiant, G. A., Lawson, L. J., & Himmelsbach, J. (1992). Three dimensional kinematic evaluation of sport bra design: 1121. Medicine & Science in Sports & Exercise, 24(5), 187. https://doi.org/10.1249/00005768-199205001-01122 Gehlsen, G., & Albohm, M. (1980). Evaluation of sports bras. The Physician and Sportsmedicine, 8(10), 88–97. https://doi. org/10 1080/00913847 1980 11948653 Eden, K. B., Valiant, G. A., Lawson, L. J., & Himmelsbach, J. (1992). Three dimensional kinema design: 1121. Medicine & Science in Sports & Exercise, 24(5), 187. https://doi.org/10.1249 Eden, K. B., Valiant, G. A., Lawson, L. J., & Himmelsbach, J. (1992). Three dimensional kinematic evaluation of sport bra design: 1121. Medicine & Science in Sports & Exercise, 24(5), 187. https://doi.org/10.1249/00005768-199205001-01122 g p p g Gehlsen, G., & Albohm, M. (1980). Evaluation of sports bras. The Physician and Sportsmedicine, 8(10), 88–97. https://doi. org/10.1080/00913847.1980.11948653 Haake, S., & Scurr, J. (2010). A dynamic model of the breast during exercise. Sports Engineering, 12(4), 189–197. Funding g This work was supported by Shaanxi Science and Technology Department International Science Technology Coopera- tion Funding (Grant Number 2018KW-056). Author details 1 1 School of Apparel and Art Design, Xi’an Polytechnic University, No. 19 Jinhua South Road, Xi’an, Shaanxi, China. 2 School of Design, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong. 3 Department of Biomedical Engi- neering, The Chinese University of Hong Kong, Shatin, New Territories, Hong Kong. Received: 29 January 2021 Accepted: 27 October 2021 doi.org/10.1152/physiol.00024.2019 McGhee, D. E., & Steele, J. R. (2020b). Biomechanics of breast support for active women. Exercise and Sport Sciences Reviews, 48(3), 99–109. https://doi.org/10.1249/JES.0000000000000221 McGhee, D. E., Steele, J. R., & Munro, B. J. (2010). Education improves bra knowledge and fit, and level of breast support in adolescent female athletes: A cluster-randomised trial. Journal of Physiotherapy, 56(1), 19–24. https://doi.org/10. 1016/s1836-9553(10)70050-3 McGhee, D. E., Steele, J. R., Zealey, W. J., & Takacs, G. J. (2013). Bra–breast forces generated in women with large breasts while standing and during treadmill running: Implications for sports bra design. Applied Ergonomics, 44(1), 112–118. https://doi.org/10.1016/j.apergo.2012.05.006 p g j p g Milligan, A., Mills, C., Corbett, J., & Scurr, J. (2015). The influence of breast support on torso, pelvis and arm kinematics during a five kilometer treadmill run. Human Movement Science, 42, 246–260. https://doi.org/10.1016/j.humov.2015. 05.008 Pei, J., Fan, J., & Ashdown, S. P. (2019). A novel method to assess breast shape and breast asymmetry. The Journal of the Textile Institute, 110(8), 1229–1240. https://doi.org/10.1080/00405000.2018.1555876 Prakash, C., Kumar, R., & Mittal, N. (2018). Recent developments in human gait research: Parameters, approaches, applica- tions, machine learning techniques, datasets and challenges. Artificial Intelligence Review, 49(1), 1–40. https://doi.org/ 10.1007/s10462-016-9514-6 Ren, J. P., Sun, M. Z., Zhou, X. L., & Zhang, J. (2016). Research on the asymmetry of the left and right breast during walking. Journal of Capital University of Physical Education and Sports, 28(04), 360–364. https://doi.org/10.14036/j.cnki.cn11- 4513.2016.04.016 Ren, J. P., Yan, Y., & Zhou, X. L. (2015). Research on the impact of sports bras on breast movements in walking and running with different speeds. Journal of Capital University of Physical Education and Sports, 27(02), 172–177. https://doi.org/ 10.14036/j.cnki.cn11-4513.2015.02.016 j Scurr, J., White, J., & Hedger, W. (2009). Breast displacement in three dimensions during the walking and running gait cycles. Journal of Applied Biomechanics, 25(4), 322–329. https://doi.org/10.1123/jab.25.4.322 Scurr, J. C., White, J. L., & Hedger, W. (2010). The effect of breast support on the kinematics of the breast during the run- i i l J l f S S i 28(10) 1103 1109 h //d i /10 1080/02640414 2010 497542 ning gait cycle. Journal of Sports Sciences, 28(10), 1103–1109. https://doi.org/10.1080/02640414.2010.497542 Scurr, J. C., White, J. L., & Hedger, W. (2011). Supported and unsupported breast displacement in three dimensions across treadmill activity levels. Journal of Sports Sciences, 29(1), 55–61. https://doi.org/10.1080/02640414.2010.521944 Shen, X. J., Fu, X. J., & Zhou, C. C. (2018). doi.org/10.1152/physiol.00024.2019 https://doi.org/10.13475/j.fzxb.20181001406 Zhou, J., & Ma, Q. R. (2020). Establishing a genetic algorithm-back propagation model to predict the pressure of girdles and to determine the model function. Textile Research Journal, 90(21–22), 2564–2578. https://doi.org/10.1177/00405 17520922947 Zhou, J., Yu, W., & Ng, S. P. (2012). Studies of three-dimensional trajectories of breast movement for better bra design. Textile Research Journal, 82(3), 242–254. https://doi.org/10.1177/0040517511435004 Zhou, J., Yu, W., & Ng, S. P. (2012). Studies of three-dimensional trajectories of breast movement for better bra design. Textile Research Journal, 82(3), 242–254. https://doi.org/10.1177/0040517511435004 Zhou, J., Yu, W., Ng, S. P., & Hale, J. (2009). Evaluation of shock absorbing performance of sports bras. Journal of Fiber Bioen- gineering and Informatics, 2(2), 108–113. https://doi.org/10.3993/jfbi09200906 Zhou, J., Yu, W., Ng, S. P., & Hale, J. (2009). Evaluation of shock absorbing performance of sports bras. Journal of Fiber Bioen gineering and Informatics, 2(2), 108–113. https://doi.org/10.3993/jfbi09200906 doi.org/10.1152/physiol.00024.2019 Characteristics of outliers in wind speed-power operation data of wind turbines and its cleaning method. Transactions of China Electrotechnical Society, 33(14), 3353–3361. https://doi.org/10.19595/j. cnki.1000-6753.tces.171129 Wang, H. L., Zhu, Y. A., Xu, W. W., Xu, R., Huang, Y. G., & Lu, W. (2019). Extraction and importance ranking of features for gait recognition. Chinese Journal of Medical Physics, 36(07), 811–817. https://oversea.cnki.net/kns/detail/detail.aspx?FileN ame=YXWZ201907013&DbName=CJFQ2019 p; Q White, J. L., Scurr, J. C., & Smith, N. A. (2009). The effect of breast support on kinetics during overground running perfor- mance. Ergonomics, 52(4), 492–498. https://doi.org/10.1080/00140130802707907 White, J. L., Scurr, J. C., & Smith, N. A. (2009). The effect of b White, J. L., Scurr, J. C., & Smith, N. A. (2009). The effect of breast support on kinetics during overground running perfor mance. Ergonomics, 52(4), 492–498. https://doi.org/10.1080/00140130802707907 g g Wood, L. E., White, J., Milligan, A., Ayres, B., Hedger, W., & Scurr, J. (2012). Predictors of three-dimensional breast kinematics during bare-breasted running. Medicine and Science in Sports and Exercise, 44(7), 1351–1357. https://doi.org/10.1249/ MSS.0b013e31824bd62c Wood, L. E., White, J., Milligan, A., Ayres, B., Hedger, W., & Scurr, J. (2012). Predictors of three-dimensional breast kinematics during bare-breasted running. Medicine and Science in Sports and Exercise, 44(7), 1351–1357. https://doi.org/10.1249/ MSS.0b013e31824bd62c Xiao, X. P. (1997). Theoretical study and reviews on the computation method of grey interconnet degree. Systems Engineering-Theory & Practice, 8, 77–82. https://doi.org/10.12011/1000-6788(1997)8-77 Xiao, X. P. (1997). Theoretical study and reviews on the computation method of grey interco Engineering-Theory & Practice, 8, 77–82. https://doi.org/10.12011/1000-6788(1997)8-77 Zhang, J., & Guo, W. H. (2019). Research on railway passenger flow prediction method based on GA improved BP neural network. Journal of Computer and Communications, 7(7), 283–292. https://chn.oversea.cnki.net/kcms/detail/detail. aspx?FileName=MGKY201907001002&DbName=IPFD2019 Zhang, J., & Guo, W. H. (2019). Research on railway passenger flow prediction method based on GA improved BP neural network. Journal of Computer and Communications, 7(7), 283–292. https://chn.oversea.cnki.net/kcms/detail/detail. aspx?FileName=MGKY201907001002&DbName=IPFD2019 Zhou, J., & Ma, Q. R. (2019). Prediction of relationship between shoulder strap attribute and breast amplitude of sports bra by BP neural network. Journal of Textile Research, 40(9), 186–191. https://doi.org/10.13475/j.fzxb.20181001406 Zhou, J., & Ma, Q. R. (2020). Establishing a genetic algorithm-back propagation model to predict the pressure of girdles and to determine the model function. Textile Research Journal, 90(21–22), 2564–2578. https://doi.org/10.1177/00405 17520922947 Zhou, J., & Ma, Q. R. (2019). Prediction of relationship between shoulder strap attribute and breast amplitude of sports bra by BP neural network. Journal of Textile Research, 40(9), 186–191. References https://doi. org/10.1007/s12283-010-0046-z Huang, S. Y., Boone, J. M., Yang, K., Packard, N. J., McKenney, S. E., Prionas, N. D., Lindfors, K. K., & Yaffe, M. J. (2011). The characterization of breast anatomical metrics using dedicated breast CT. Medical Physics, 38(4), 2180–2191. https:// doi.org/10.1118/1.3567147 Jin, S. F., Lin, Q. Q., Ma, Q. R., & Zhang, H. (2020). Method for detecting fluff quality of fabric surface based on BP neural network. Journal of Textile Research, 41(2), 69–76. https://doi.org/10.13475/j.fzxb.20181201008i Lee, N. A., Rusinek, H., Weinreb, J., Chandra, R., Toth, H., Singer, C., & Newstead, G. (1997). Fatty and fibroglandular tissue volumes in the breasts of women 20–83 years old: Comparison of X-ray mammography and computer-assisted MR imaging. American Journal of Roentgenology, 168(2), 501–506. https://doi.org/10.2214/ajr.168.2.9016235 g g g gy, ( ), p g j Li, S. X., Ren, J. P., & Zhou, X. L. (2018). Influence of sports bra on breast kinematic characteristics and the gait parameters under different stride frequencies. Journal of Beijing Sport University, 41(1), 82–88. https://doi.org/10.19582/j.cnki.11- 3785/g8.2018.01.012 Liang, S. Z., Zhang, X., & Zhou, J. (2007). Basic breast shapes of female undergraduate in the west of China based on 3-D body scanning. Journal of Textile Research, 28(8), 75–78. https://doi.org/10.13475/j.fzxb.2007.08.020 Liang, S. Z., Zhang, X., & Zhou, J. (2007). Basic breast shapes of female undergraduate in the west of China based on 3-D body scanning. Journal of Textile Research, 28(8), 75–78. https://doi.org/10.13475/j.fzxb.2007.08.020 Lorentzen, D., & Lawson, L. (1987). Selected sports bras: A biomechanical analysis of breast motion while jogging. The Physician and Sportsmedicine, 15(5), 128–139. https://doi.org/10.1080/00913847.1987.11709355 Lorentzen, D., & Lawson, L. (1987). Selected sports bras: A biomechanical analysis of breast motion while jogging. The Physician and Sportsmedicine, 15(5), 128–139. https://doi.org/10.1080/00913847.1987.11709355 Losken, A., Fishman, I., Denson, D. D., Moyer, H. R., & Carlson, G. W. (2005). An objective evaluation of breast symmetry and shape differences using 3-dimensional images. Annals of Plastic Surgery, 55(6), 571–575. https://doi.org/10.1097/01. sap.0000185459.49434.5f Page 21 of 21 Zhou et al. Fashion and Textiles (2022) 9:3 Zhou et al. Fashion and Textiles Mao, Q., Zhou, J., & Wang, Q. (2020). Breast shape recognition of young women in the west of China based on GRNN and PNN. Journal of Xi’an Polytechnic University, 34(04), 7–13. https://doi.org/10.13338/j.issn.1674-649x.2020.04.002 Mao, Q., Zhou, J., & Wang, Q. (2020). Breast shape recognition of young women in the west of China based on GRNN and PNN. Journal of Xi’an Polytechnic University, 34(04), 7–13. https://doi.org/10.13338/j.issn.1674-649x.2020.04.002 Mason, B. R., Page, K. References A., & Fallon, K. (1999). An analysis of movement and discomfort of the female breast during exercise and the effects of breast support in three cases. Journal of Science and Medicine in Sport, 2(2), 134–144. https://doi. org/10.1016/S1440-2440(99)80193-5 g McGhee, D. E., Power, B. M., & Steele, J. R. (2007). Does deep water running reduce exercise-induced breast discomfort? British Journal of Sports Medicine, 41(12), 879–883. https://doi.org/10.1136/bjsm.2007.036251 p p g j McGhee, D. E., & Steele, J. R. (2020a). Breast biomechanics: What do we really know? Physiology p p g j McGhee, D. E., & Steele, J. R. (2020a). Breast biomechanics: What do we really know? Physiology, 35(2), 144–156. https:/ p p g j McGhee, D. E., & Steele, J. R. (2020a). Breast biomechanics: What do we really know? Physiology, 35(2), 144–156. https:// doi.org/10.1152/physiol.00024.2019 Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
https://openalex.org/W1969309070	https://casesjournal.biomedcentral.com/counter/pdf/10.1186/1757-1626-2-10	English	null	Elective laparoscopic splenectomy for giant hemangioma: a case report	Cases journal	2,009	cc-by	2,850	Cases Journal Open Access Introduction nation revealed splenomegaly, although the patient was asymptomatic. No other clinical signs were detected. Ultrasonography and Magnetic resonance imaging (MRI) showed a hypoechogenic mass involving the spleen and the splenomegaly (Figure 1). Preoperative evaluation led to the possible diagnosis of hemangioma of the spleen. The patient was subjected to laparoscopy, which demon- strated a solid tumor of the spleen. Laparoscopic splenec- tomy was performed. Operative time was 65 minutes. Postoperative recovery was uneventful, the drain was removed on the second postoperative day and the patient was discharged on the same day. Histologic examination showed a capillary hemangioma of a spleen weighing 850 gr. The patient remains asymptomatic 6 months after the operation. Splenic hemangioma is a rare disorder, but remains the most common benign neoplasm of the spleen [1]. Splenectomy is indicated because spontaneous rupture with massive hemorrhage can occur [2]. Laparoscopic splenectomy (LS) can be a challenging procedure given the fragile, well-vascularized nature of the spleen and its proximity to the pancreas, stomach and colon. Since the mid-1990s LS has steadily replaced open splenectomy as the approach of choice for most elective splenectomies [3]. We present herein a case of a patient who underwent LS for a hemangioma of the spleen. p Elective laparoscopic splenectomy for giant hemangioma: a case t Address: 1Department of Surgery, Interbalkan European Medical Center, Thessaloniki, Greece, 23rd Surgical Clinic, AHEPA Hospital, Thessaloniki, Greece and 31st Propeudeutic Surgical Clinic of A.U.Th., AHEPA Hospital, Thessaloniki, Greece l: Christophoros Kosmidis* - dr.ckosmidis@gmail.com; Christopher Efthimiadis - info@efthimiadis.gr; i A hi idi @ d f M i G i i d i @ il K lli i V ili Email: Christophoros Kosmidis* - dr.ckosmidis@gmail.com; Christopher Efthimiadis - info@efthimiadis.gr; Georgios Anthimidis - georgeaggs@vodafone.net.gr; Marios Grigoriou - drgregoriou@gmail.com; Kalliopi Vasiliadou - kellyvas@gmail.com; Petros Sfikakis - pg_sfikakis@yahoo.gr; Nikolaos Tziris - kellyvas@gmail.com; Epaminondas Fahantidis - mennipus_cynic@yahoo.gr * C di h Email: Christophoros Kosmidis dr.ckosmidis@gmail.com; Christopher Efthimiadis info@efthimiadis.gr; Georgios Anthimidis - georgeaggs@vodafone.net.gr; Marios Grigoriou - drgregoriou@gmail.com; Kalliopi Vasiliadou - kellyvas@gmail.com; Petros Sfikakis - pg_sfikakis@yahoo.gr; Nikolaos Tziris - kellyvas@gmail.com; Epaminondas Fahantidis - mennipus_cynic@yahoo.gr * Corresponding author * Corresponding author Published: 5 January 2009 Cases Journal 2009, 2:10 doi:10.1186/1757-1626-2-10 Received: 1 December 2008 Accepted: 5 January 2009 This article is available from: http://www.casesjournal.com/content/2/1/10 © 2009 Kosmidis et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Received: 1 December 2008 Accepted: 5 January 2009 This article is available from: http://www.casesjournal.com/content/2/1/10 © 2009 Kosmidis et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Abstract Although unusual, hemangioma is the most common primary splenic neoplasm. Splenectomy is indicated when the tumor is large, with increased risk of hemorrhage. The laparoscopic approach is preferred for most elective splenectomies. Although technically feasible, laparoscopic splenectomy can be a challenge in the patient with splenomegaly. We present herein a case of an 18-year-old male asymptomatic patient who underwent laparoscopic splenectomy for the incidental finding of splenomegaly caused by a large splenic hemangioma. Laparoscopic splenectomy appears to be a safe and effective procedure, in appropriately experienced hands, for patients with splenomegaly, given the spleen's fragile anatomy and its relationship to other abdominal viscera. BioMed Central BioMed Central Technique Following induction of general anesthesia and endotra- cheal intubation, a nasogastric tube and a urinary catheter were inserted, and compression stockings were applied. The patient was placed on a beanbag in a 60-degree right lateral decubitus position. The right brachial plexus was protected with a pillow roll and the left arm was sup- ported by a splint. The patient was positioned so that the table could be flexed to create a wider working space in a reverse Trendelenburg position. The surgeon and the scrub nurse stood to the patient's right, and the assistants stood to the left. The video monitors were placed on each side of the table, above the level of the patient's shoulders. approached from the lower pole and dissection was con- tinued cephalad. With the spleen elevated, the short gas- tric vessels and main vascular pedicle were visualized. The tail of the pancreas was also visualized and avoided at this point as it approached the splenic hilum. The splenic pedicle was carefully dissected from the medial and lateral aspects. After the short gastric vessels had been divided with ultrasonic dissector, the splenic artery and vein were dissected. The vessels were divided by application of endoscopic vascular staplers (ENDOPATH®, ETS Flex 45 Endoscopic Articulating Linear Cutter 45 mm staple line, Pneumoperitoneum was established using a Veress needle to a pressure of 12 mm Hg. Four ports were used: the first 10-mm trocar was inserted at the left margin of the umbilicus. The other three 5-mm operating trocars were positioned as follows: the first trocar was on the median line, 8 cm above the umbilicus, a second operating trocar in a left subcostal position on the axillary line to enable gastroepiploic mobilization, and a left 5-mm operating trocar was positioned in the left lateral position, 10 cm from the umbilicus (Figure 2). Lift of the inferior pole of the spleen and division of the splenocolic ligament Figure 3 Lift of the inferior pole of the spleen and division of the splenocolic ligament. The operation is begun with a thorough search of the abdominal cavity. The inferior pole of the spleen was lifted superiorly. The splenocolic ligament was divided (Figure 3). This mobilized the inferior aspect of the spleen and allowed the spleen to be retracted cephalad. Great care was taken to avoid rupture of the splenic capsule dur- ing retraction. Case presentation An 18-year-old Greek male student presented for screen- ing test in order to get a health certificate. Physical exami- Page 1 of 4 (page number not for citation purposes) http://www.casesjournal.com/content/2/1/10 Cases Journal 2009, 2:10 Position of trocars Figure 2 Position of trocars. Magnetic resonance imaging (MRI) showing the splenomegaly and the hemangioms Figure 1 Magnetic resonance imaging (MRI) showing the splenomegaly and the hemangioms. Magnetic resonance imaging (MRI) showing the splenomegaly and the hemangioms Figure 1 Magnetic resonance imaging (MRI) showing the splenomegaly and the hemangioms. Magnetic resonance imaging (MRI) showing the splenome and the hemangioms Figure 1 Magnetic resonance imaging (MRI) showing the splenomegaly and the hemangioms. Position of trocars Figure 2 Position of trocars. Page 2 of 4 (page number not for citation purposes) Discussion Most splenic hemangiomas are discovered incidentally, and their clinical importance generally lies in differentiat- ing them from other conditions, particularly from metas- tases. Occasionally they may be associated with splenomegaly, abdominal pain, dyspnoea, diarrhea, or constipation. No potential for malignancy exists. Heman- giomas are not treated unless they are symptomatic or very large, with increased risk of hemorrhage; treatment is splenectomy [2]. In our case, the patient was asympto- matic and splenectomy was performed because of the large size of the spleen and the increased risk of rupture. Many surgeons now prefer to use the laparoscopic approach for most elective splenectomies. In experienced hands, LS can be performed as safely and effectively as open splenectomy [4]. Operative time is longer for LS, but the procedure offers more rapid postoperative recovery and shorter duration of hospital stay [5]. Laparoscopic splenectomy can be completed in about 90% of properly selected patients. The incidence of conversion to open splenectomy is between 0% and 20% [5]. Most of the con- versions are caused by intraoperative bleeding, but lack of surgical experience, extensive adhesions, large splenome- galy, and obesity are also involved [6-9]. A significant learning curve is observed with LS, and with increasing experience, the conversion rate has been reported to decrease dramatically [9-12]. The role of LS is now sup- ported, in appropriately expert and experienced hands, for patients with splenomegaly, multiple prior abdominal surgeries, morbid obesity and the need for concomitant procedures [3]. Though technically feasible, LS in the patient with splenomegaly can be a challenge. Hand- assisted laparoscopic surgery has been suggested as a means by which splenectomy can be more safely and expeditiously performed in case of a large spleen [11]. Incision of the lateral peritoneal attachments of the spleen Figure 4 Incision of the lateral peritoneal attachments of the spleen. 2.5 mm Staple Leg Length (Vascular/Thin). 45 MM Vascu- lar/Thin). Each jaw was positioned anterior and posterior to the splenic vessels (Figure 5). The instrument was fired two times in sequence. To remove the detached spleen, a nylon extraction bag was introduced through the left lat- eral trocar site. The bag was opened within the abdominal cavity. An incision adequate to enable removal of the bag containing the intact spleen was made at the left lateral site. The spleen was placed into the specimen retrieval bag. The drawstring was grasped, and the bag was closed and drawn out. Technique The lateral peritoneal attachments of the spleen, the splenorenal and splenophrenic ligaments were sequentially incised (Figure 4). The splenic hilum was Lift of the inferior pole of the spleen and division of the splenocolic ligament Figure 3 Lift of the inferior pole of the spleen and division of the splenocolic ligament. Lift of the inferior pole of the spleen and division of the splenocolic ligament Figure 3 Lift of the inferior pole of the spleen and division of the splenocolic ligament. Page 2 of 4 (page number not for citation purposes) Page 2 of 4 (page number not for citation purposes) Cases Journal 2009, 2:10 http://www.casesjournal.com/content/2/1/10 neal irrigation was carried out and a Rob drain tube N° 24 was placed at the residual cavity. Incision of the lateral peritoneal attachments of the spleen Figure 4 Incision of the lateral peritoneal attachments of the spleen. Page 3 of 4 (page number not for citation purposes) References 1. Bailez M, Elmo G: Laparoscopic Partial Splenectomy for a Hemangioma of the Spleen. Pediatric Endosurgery & Innovative Techniques 2004, 8(2):147-151. q ( ) 2. Disler DG, Chew FS: Splenic hemangioma. AJR Am J Roentgenol 1991, 157(1):44. ' q ( ) 2. Disler DG, Chew FS: Splenic hemangioma. AJR Am J Roentgenol 1991, 157(1):44. 3. Park AE, Mckinlay R: Spleen. In Schwartz's Principles of Surgery 8th edition. Edited by: Brunicardi FC, Andersen DK, Billiar TR, Dunn DL, Hunter JG, Pollock RE. New York: McGraw-Hill; 2005:1297-1315. 4. Katkhouda N, Hurwitz MB, Rivera RT, Chandra M, Waldrep DJ, Gugenheim J, Mouiel J: Laparoscopic splenectomy: Outcome and efficacy in 103 consecutive patients. Ann Surg 1998, 228:568-578. 5. Beauchamp RD, Holzman MD, Fabian TC, Weinberg JA: The Spleen. In Sabiston Textbook of Surgery, The Biological Basis of Modern Surgical Practice 14th edition. Edited by: Townsend CM, Beauchamp RD, Evers BM, Mattox KL. Philadelphia: Saunders; 2004:1679-1708. p 6. Flowers JL, Lefor AT, Steers J, Heyman M, Graham SM, Imbembo AL: Laparoscopic splenectomy in patients with hematologic dis- eases. Ann Surg 1996, 224:19-28. g 7. Friedman RL, Fallas MJ, Carroll BJ, Hiatt JR, Phillips EH: Laparo- scopic splenectomy for ITP: The gold standard. Surg Endosc 1996, 10:991-995. 8. Horowitz J, Smith JL, Weber TK: Postoperative complications after splenectomy for hematologic malignancies. Ann Surg 1996, 223:290-296. 9. Hansen K, Singer DB: Asplenic-hyposplenic overwhelming sep- sis: Postsplenectomy sepsis revisited. Pediatr Dev Pathol 2001, 4:105-121. 10. Glasgow RE, Yee LF, Mulvihill SJ: Laparoscopic splenectomy. The emerging standard. Surg Endosc 1997, 11:108-112. 11. Romanelli JR, Kelly JJ, Litwin DE: Hand-assisted laparoscopic sur- gery in the United States: An overview. Semin Laparosc Surg 2001, 8:96-103. 12. Shatz DV, Schinsky MF, Pais LB, Romero-Steiner S, Kirton OC, Car- lone GM: Immune responses of splenectomized trauma patients to the 23-valent pneumococcal polysaccharide vac- cine at 1 versus 7 versus 14 days after splenectomy. J Trauma 1998, 44:760-765. discussion 765–766 Consent Written informed consent was obtained from the patient for publication of this case report and accompanying images. A copy of the written consent is available for review by the Editor-in-Chief of his journal. http://www.casesjournal.com/content/2/1/10 http://www.casesjournal.com/content/2/1/10 Cases Journal 2009, 2:10 cholecystectomy. Regardless of the position, the patient is placed so that the kidney rest can be raised to maximize the space between the iliac crest and the costal margin. As with the anterior approach, the double access technique requires the placement of five or six trocars. On the con- trary, the lateral approach involves the use of three or four trocars and offers exposure of the anatomy in a way that allows for a more intuitive sequence of dissection, paral- leling that of open splenectomy; the spleen is suspended from its diaphragmatic attachments, and gravity retracts the stomach, transverse colon, and greater omentum, while placing the splenic hilum under tension; after the short gastric vessels have been divided, the splenic pedicle may be carefully dissected from both the medial and lat- eral aspects; the surgeon easily visualizes the tail of the pancreas and avoids injury when placing the endovascular stapler [3,5]. the challenge of splenectomy given the spleen's fragile anatomy and its relationship to other abdominal organs, especially the pancreas. Cautiousness is indicated in patients with portal hypertension and severe cardiopul- monary disease because they have a higher incidence of intraoperative bleeding and postoperative complications. Competing interests h h d l h h The authors declare that they have no competing interests. Authors' contributions CK, CE, MG, and GA analyzed and interpreted the data concerning the splenectomy operation. KV gathered the data concerning the patient case. PS, NT and EF contrib- uted in the reference data. All authors read and approved the final manuscript. The operation is begun with a thorough search of the abdominal cavity for the presence of accessory splenic tis- sue. Meticulous surgical technique to prevent splenosis secondary to splenic trauma during dissection is as impor- tant as a thorough search for accessory spleens [4]. A 1-cm cuff of peritoneum may be left along the lateral aspect of the spleen to act as a handle to manipulate the organ. The splenic vessels are divided following or prior to division of the short gastric vessels. The short gastric vessels can be divided by clips, endovascular stapling cartridges, ultra- sonic dissection, diathermy or radio frequency ablation. For the control of splenic vessels ultrasonic dissector, hemoclips, bipolar devices, Liga-Sure, or an endovascular stapling device can be used. In our case a laparoscopic endovascular stapler, fired two times in sequence, was used (ENDOPATH®, ETS Flex 45 Endoscopic Articulating Linear Cutter 45 mm staple line, 2.5 mm Staple Leg Length (Vascular/Thin). 45 MM Vascular/Thin). The use of hemoclips is minimized throughout the procedure and especially around the hilum because the clips may inter- fere with future applications of a stapling device. The sta- pler will not function if a clip is caught within its jaws, and this can result in significant bleeding from hilar vessels. When possible, the splenic artery and vein are divided separately. However, mass hilar stapling is increasingly practiced with good long-term results. This is especially true for the magistral configuration of the splenic artery. The placement of a drain is not routinely recommended because it may be associated with postoperative left sub- phrenic abscess [3,5]. In our case a Rob drain tube was placed and subsequently removed on the second postop- erative day. Our surgical technique has been described previously. Page 4 of 4 (page number not for citation purposes) Discussion The laparoscope was reinserted and the splenic bed was assessed for hemostasis. Finally, perito- Division of the vessels of the splenic hilum by application of endoscopic vascular staplers Figure 5 Division of the vessels of the splenic hilum by applica- tion of endoscopic vascular staplers. Although the indications for LS are the same as for open splenectomy, some cases require caution. Absolute con- traindications to the laparoscopic approach include severe cardiopulmonary disease and cirrhosis. Variceal short gas- tric vessels compounded by the coagulopathy of liver dis- ease present an unacceptable risk for operative hemorrhage in patients with portal hypertension [5]. The earliest reports of LS detailed an anterior approach with the patient in the supine or in the low lithotomy position. Most procedures are now performed in the right lateral decubitus position. Some authors advocate a mid- way «double access» technique in which the patient is in a 45° right lateral decubitus position, facilitating the per- formance of concomitant surgery, such as laparoscopic Division of the vessels of the splenic hilum by application of endoscopic vascular staplers Figure 5 Division of the vessels of the splenic hilum by applica- tion of endoscopic vascular staplers. Division of the vessels of the splenic hilum by application of endoscopic vascular staplers Figure 5 Division of the vessels of the splenic hilum by applica- tion of endoscopic vascular staplers. Page 3 of 4 (page number not for citation purposes) Page 3 of 4 (page number not for citation purposes) http://www.casesjournal.com/content/2/1/10 Conclusion LS is a safe procedure for patients with splenomegaly. Sur- geons require advanced laparoscopic skills to deal with Page 4 of 4 (page number not for citation purposes) Page 4 of 4 (page number not for citation purposes)
https://openalex.org/W2077497484	https://uobrep.openrepository.com/bitstream/10547/223778/1/834616.pdf	English	null	Strategic Team AI Path Plans: Probabilistic Pathfinding	International journal of computer games technology	2,008	cc-by	3,986	1School of Computer Engineering, Nanyang Technological University, Singapore 639798 2Department of Computing and Mathematics, Manchester Metropolitan University, Manchester M1 5GD, UK 1School of Computer Engineering, Nanyang Technological University, Singapore 639798 2Department of Computing and Mathematics, Manchester Metropolitan University, Manchester M1 5GD, UK Correspondence should be addressed to Edmond C. Prakash, e.prakash@mmu.ac.uk Received 29 September 2007; Accepted 13 December 2007 Recommended by Kok Wai Wong Recommended by Kok Wai Wong This paper proposes a novel method to generate strategic team AI pathﬁnding plans for computer games and simulations using probabilistic pathﬁnding. This method is inspired by genetic algorithms (Russell and Norvig, 2002), in that, a ﬁtness function is used to test the quality of the path plans. The method generates high-quality path plans by eliminating the low-quality ones. The path plans are generated by probabilistic pathﬁnding, and the elimination is done by a ﬁtness test of the path plans. This path plan generation method has the ability to generate variation or diﬀerent high-quality paths, which is desired for games to increase replay values. This work is an extension of our earlier work on team AI: probabilistic pathﬁnding (John et al., 2006). We explore ways to combine probabilistic pathﬁnding and genetic algorithm to create a new method to generate strategic team AI pathﬁnding plans. Copyright © 2008 Tng C. H. John et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Copyright © 2008 Tng C. H. John et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. 1. INTRODUCTION have diﬀerent plans, they move diﬀerently, thus the players cannot always predict the opponents’ locations. This work uses probabilistic pathﬁnding algorithm to obtain variations of team AI path plans. A popular game with heavy team AI is “Full Spectrum War- rior” (FSW), a console game on Xbox. This game is a down- sized version of “Full Spectrum Command” on the PC plat- form. The game “Full Spectrum Command” is actually a simulation of the real-world behavior of the US Army. This game was originally used in the military for leadership train- ing as well as decision making training. The game includes real-world army movements such as “bounding” and duck- ing. The main feature of the game is its team AI. The team AI of the game is actually derived from the real world and simulates how a real person will behave. The purpose of this game is to command two teams of soldiers to accomplish a mission. Section 2 talks about related work, existing problem, and a scenario that strategic team AI path planning can be ap- plied. Section 3 gives an introduction about team AI and probabilistic pathﬁnding. It explains how team AI is created in [1] by combining diﬀerent systems. Section 4 describes the main method of strategic team AI path plan generation. Section 5 suggests a method that can be used to optimize this work. Section 6 shows some interesting strategies gener- ated by this team AI path plan generation method. The ﬁnal section, Section 7, comprises conclusion, other applications, and future suggestions for this work. Even though it is a great game, one area it can improve on is the team AI plan. Players may feel that opponents al- ways appear at the same places after playing repeatedly, as most team AI plans use A∗algorithm or look up tables as their main pathﬁnding techniques. These techniques always produce the same path if the source and destination locations are the same. On the other hand, Probabilistic is able to pro- duce variations to the path even if the source and destination locations are the same. Hindawi Publishing Corporation International Journal of Computer Games Technology Volume 2008, Article ID 834616, 6 pages doi:10.1155/2008/834616 Hindawi Publishing Corporation International Journal of Computer Games Technology Volume 2008, Article ID 834616, 6 pages doi:10.1155/2008/834616 Hindawi Publishing Corporation International Journal of Computer Games Technology Volume 2008, Article ID 834616, 6 pages doi:10.1155/2008/834616 Tng C. H. John,1 Edmond C. Prakash,2 and Narendra S. Chaudhari1 Tng C. H. John,1 Edmond C. Prakash,2 and Narendra S. Chaudhari1 1School of Computer Engineering, Nanyang Technological University, Singapore 639798 2Department of Computing and Mathematics, Manchester Metropolitan University, Manchester M1 5GD, UK Figure 1: Situation without team AI. Figure 1: Situation without team AI. Figure 1: Situation without team AI. Figure 3: Diﬀerent paths generated by probabilistic pathﬁnding. In this work, we use and extend our earlier work on prob- abilistic A∗pathﬁnding algorithm [1]. Further readings on A∗pathﬁnding can be found in [2–4]. Bourg and Seeman [4] provide other data representation of the A∗pathﬁnding. A very useful article by Pinter [5] discusses methods to mod- ify a raw path generated from a path search to form a more convincing traveling path. An interesting work by Kamphuis et al. [6] attempts to simulate tactical pathﬁnding in urban environments for a small group of characters for games and simulations. The characters use tactical information such as road maps and special locations for pathﬁnding. This work uses common A∗data structure and a list of gateway points. The tactical pathﬁnding [6] algorithm is able to run in real time with the help of a preprocessing step. For this work to run in real time, no preprocessing is needed. With optimization, it can even run more eﬃciently. Figure 3: Diﬀerent paths generated by probabilistic pathﬁnding. 3. PREVIOUS WORK Notice that there are light grey squares in Figures 1, 2, and 3. They are actually the gateways of the map. Gateway is nar- row path in the environment that opens up to a bigger path before and after the gateway. That is, a gateway is a narrow link between two spaces. Figure 4 illustrates a gateway. This section is an introduction to team AI: Probabilistic pathﬁnding. The detail implementation and algorithm can be found in [1]. Team AI can be shown to exist if teammates coordinate to trap or capture an opposite team. Figure 1 shows a situation when team AI is not enabled. The enemy teammates ﬁnd the shortest path to capture the player. A blackboard messaging system is developed in [1] to facilitate communications between characters of a team. In short, blackboard is a place for characters to “write” use- ful information and let other teammates read it. After every teammate read, the message is deleted by the message writer. Figure 5 shows the concept of a blackboard. Figure 2 shows a situation with team AI enabled. The en- emy teammates surround, trap, and capture the player. Probabilistic is a modiﬁed version of A∗algorithm with an addition ability to generate diﬀerent paths controlled by a probability variable. The variable controls how diﬀerent the paths diﬀer from the shortest path. The paths may not be the shortest, but they are one of the shorter paths. In general, if the variable is 0, then probabilistic pathﬁnding will behave exactly the same as a usual A∗pathﬁnding. If the variable is set to 1, it will always produce a diﬀerent path that is not the shortest. For more details refer to [1]. Figure 3 below shows an example of an enemy character following diﬀerent paths to pursuit the player character. Putting together probabilistic pathﬁnding, gateways in- formation, and message system, we achieved what is shown in Figure 2, the complete working team AI in [1]. 2. RELATED WORK In this paper, team AI pathﬁnding and team AI plans are the same. Team AI pathﬁnding refers to the diﬀerent paths taken by teammates to reach a desired place. An example is a team of enemies enter a room from diﬀerent doors to trap and capture the player. Team AI pathﬁnding plans are popular in computer games. With good team plan, the game diﬃculty level increases, making the game more challenging and helps to showcase intelligent behavior of the game. Replay value can easily be added to the game by creating variations to the opponent team plans. When the opponents 2 International Journal of Computer Games Technology Figure 2: Situation with team AI. Figure 3: Diﬀerent paths generated by probabilistic pathﬁnding. 4. PATH PLAN GENERATION In this section, we illustrate the method used to generate high quality team AI path plans. The main idea is to test the team AI path plans with a ﬁtness function. The ﬁtness will Tng C. H. John et al. 3 Probabilistic pathﬁnding generates path plans Pass Fitness test Fail High quality path plans Figure 6: Path plan generation ﬂow chart. Unmovable area Movable area Gateway Figure 4: Illustration of a gateway. I want to inform B . . . Entity A A get message A post message Blackboard B can you please move . . .? Yes I can. On the way. B post message B get message I want to know A . . . Entity B Figure 5: Diﬀerent paths generated by probabilistic pathﬁnding. Figure 6: Path plan generation ﬂow chart. Figure 7: Path with overlapping regions. I want to inform B . . . Entity A A get message A post message Blackboard B can you please move . . .? Yes I can. On the way. B post message B get message I want to know A . . . Entity B Figure 5: Diﬀerent paths generated by probabilistic pathﬁnding. I want to inform B . A post message A get message Figure 7: Path with overlapping regions. is not acceptable for two teammates to get too near to each other or their paths overlap. An alternative could be for the team to explore a bigger area of the map to gain resources and familiarize with the terrain. It all depends on the game play. Therefore, the ﬁtness test for such a path plan will fail if two teammates get too close to each other. Fitness test can also be constraints of the team path plan. B get message B post message I want to know A . . Figure 7 shows a team path plans that are not acceptable because of overlap paths. Figure 5: Diﬀerent paths generated by probabilistic pathﬁnding. The path of the bottom-right enemy character follows the shortest path to the player. With team AI enabled, that bot- tom two characters have to trap the player through diﬀerent entrances. However, due to overlap paths constraint, the path plan is discarded. A new path plan is conducted and shown in Figure 8. 4. PATH PLAN GENERATION By comparing the plan shown in Figure 7 and Figure 8, the team path plan in Figure 8 is better than that of Figure 7 according to overlap constrains. determine whether the team plan is good enough. If the qual- ity is bad, a new team AI path plan search will be conducted. Figure 6 shows the ﬂow chart for path plan generation. This idea is mainly inspired by genetic algorithms [7], where a ﬁtness function is used to test the quality of the genes combination. The genes combination is formed based on its parents, some manipulation, and some randomness. The better the quality of the combination of genes is, the higher chance it will survive. The bad quality gene combina- tions get eliminated. Team AI path plan works the same way. Diﬀerent path plans are generated by probabilistic pathﬁnd- ing. Treat each path plan as a combination of genes. A ﬁtness function is used to test path plan. If it is not good enough, it will be eliminated. A new search will be conducted. The cycle repeats until a satisfactory quality path plan is obtained. The following are three ways to test the team paths with the ﬁtness function. They are illustrated below. 5. OPTIMIZATION Fitness testing should be cheap if it does not involve cal- culation of huge set of constraints and variables. The load of this system comes from A∗pathﬁnding. This means that all optimization techniques applicable to A∗pathﬁnding are here. A common optimization technique for A∗pathﬁnding Algorithm 1 search once a path plan passed the ﬁrst ﬁtness test. This method is eﬃcient if the ﬁtness test seldom fails on path plans. This method is the easiest and fastest to implement. No modiﬁcation is needed for probabilistic pathﬁnding gen- eration algorithm. No modiﬁcation is needed for team path planning. The exact algorithm is already shown in Figure 6. Algorithm 1 shows the pseudo code algorithm. Algorithm 3 4.2. Step test and the step test Table 1. It is based on each character. Af- ter each character has found its path, the ﬁtness test will be performed. If the test fails, the character will choose an- other path. This is the best method if the iterative test and step test always fail. Modiﬁcation needs to be made to team path planning. Fitness test is conducted per character (see Algorithm 3). Algorithm 2 shows the pseudo code of the step test. Algorithm 2 shows the pseudo code of the step test. The ﬁtness test is conducted at every segment of the path. This is the extreme opposite end of iterative test. On selection of the next node (using probability pathﬁnding search), if it does not pass the ﬁtness test, another node will be chosen in- stead. This method is good if iterative test always fails and the number of characters is small. As opposed to iterative test, modiﬁcation is needed for probabilistic pathﬁnding genera- tion algorithm. The ﬁtness test function has to be included into the probabilistic pathﬁnding algorithm. Path QualityPathPlanForWholeTeam () { Path pathOfWholeTeamSoFar = null; While (TeamPathPlanNotComplete ()) { pathOfWholeTeamSoFar + = ProbabilisticPathForOne (pathOfWholeTeamSoFar); } Return pathOfWholeTeamSoFar; } Path ProbabilisticPathForOne (Path pathOfWholeTeamSoFar) { Path currentMemberPath = null; Path temp = null; do { currentMemberPath += selectANextNode (); do { temp = pathOfWholeTeamSoFar + currentMemberPath.selectADiﬀerentNode (); } while (FitnessTest(temp) == fail); } while (currentMemberPath.SearchNotComplete ()) return currentMemberPath; } Figure 8: Path without overlap. Figure 8: Path without overlap. Path QualityPathPlanForWholeTeam() { Path pathOfWholeTeam; do { pathOfWholeTeam = null; While(TeamPathPlanNotComplete()) { pathOfWholeTeam + = ProbabilisticPathForOne(); } } while(FitnessTest(pathOfWholeTeam)==fail); Return pathOfWholeTeam; } Algorithm 1 Algorithm 2 Path QualityPathPlanForWholeTeam () { Path pathOfWholeTeamSoFar = null; Path temp = null; Path memberPath = null; do { do { memberPath = ProbabilisticPathForOne (); temp = pathOfWholeTeamSoFar + memberPath; } while (FitnessTest(temp) == fail) pathOfWholeTeamSoFar += memberPath; } while (TeamPathPlanNotComplete ()); Return pathOfWholeTeamSoFar; } Algorithm 3 4.1. Iterative test The ﬁtness test is conducted after the whole team found its path. If the ﬁtness test fails, a new path plan search will be conducted. However, the old path plan is saved. This is to prevent the system from doing too many searches and slow down the game. The user can specify a number of maximum searches to perform. If the maximum number of searches is reached, the path plan with the highest ﬁtness will be selected. The user may also choose to terminate the A ﬁtness test can be a simple function that calculates the distance between two characters. For example in a game, it International Journal of Computer Games Technology 4 4.3. Progressive test The test will be conducted after each character found its path. This is a middle solution between the iterative test Tng C. H. John et al. 5 Table 1: Summary ﬁtness function test. Methods Advantage Best for cases Iterative test Easiest and fastest to implement Many high quality solutions Fitness test mostly passes Progressive test A general solution that can solve most cases Average solution In the middle of iterative test and step test Step test Guaranteed to have a solution if it exists Few high quality solutions Fitness test mostly fails Figure 9: Combine force strategy. Figure 10: Trap strategy. Table 1: Summary ﬁtness function test. Methods Advantage Best for cases Iterative test Easiest and fastest to implement Many high quality solutions Fitness test mostly passes Progressive test A general solution that can solve most cases Average solution In the middle of iterative test and step test Step test Guaranteed to have a solution if it exists Few high quality solutions Fitness test mostly fails Figure 10: Trap strategy Figure 9: Combine force strategy. Figure 10: Trap strategy. Figure 10: Trap strategy. are able to meet before encountering the player, it is a good strategy. is search by parts so that the A∗search execution is spread out over many frames. Figure 10 shows a trap strategy. It is the reverse strategy of Figure 9. In this case, the team may have higher chance of success for killing the player. The trap strategy aims to trap the player at diﬀerent directions. As far as possible, this means that the enemy teammates should not have overlap paths. So there are two constrained ﬁtness functions for this example. The ﬁrst constraint means, as far as possible, that the teammates must not encounter the player in the same direction. In general, use the pigeonhole theorem [8]. Num- ber of teammates in same direction cannot be greater than teammates divided by number of directions. This is to ensure equal distribution of teammates in each direction. The sec- ond constraint is to avoid crossing paths of the teammates. This is to create higher chance of search space if the player escapes somewhere else. 6. ANALYSIS OF STRATEGIES One useful feature of this team AI path plan generation method is to generate various useful strategies. These strate- gies can be applied to ﬁrst person shooting team games as AI opponents or enemies. With such interesting strategies, the replay value of the game increases. The diﬃculty level in- creases and it becomes more challenging for player to defeat the enemies. This section analyzes interesting strategies gen- erated. Figure 9 shows the combined force strategy. In this par- ticular strategy, it is the reverse. Joining teammates together synergized the power of the enemy team and increased the chance of success for eliminating the player. From Figure 9, the enemies join forces along the way and attack the player together in a single path. The ﬁtness function to such a plan is to test the path before the destination (the player posi- tion) and ensure that before the destination all three team- mates must be together. This is the ﬁrst constraint. This ﬁt- ness function will eliminate plans that do not combine forces before they encounter the player. The second constraint con- trols can be how early the teammates must combine their forces before they encounter. The earlier they meet the higher chance of success in their mission. For this example, the sec- ond constrained ﬁtness function is not tested. As long they Figure 11 shows a similar strategy as in Figure 10. How- ever, a third ﬁtness constraint is applied. The third constraint is the minimum distance away from the path of teammates. This is the explore and trap strategy . No doubt, the main ob- jective of the teammates is to trap and capture the player in diﬀerent directions and paths. In addition, the enemy team- mates have a secondary objective to explore a wider area of the map. This will facilitate their future plans, actions, or op- erations. From Figures 10 and 11, with an additional ﬁtness function, the path generated is diﬀerent. A secondary objec- tive can be included with additional constraints. International Journal of Computer Games Technology 6 Figure 11: Explore and trap strategy. Figure 12: Combine and split trap strategy. Figure 12: Combine and split trap strategy. The path plan generated using this method is by trial and error. Generate a path, test it, and discard it if it is not good enough. A future step to go from here is to generate path plans by functions or heuristic. An example is to add a ﬁtness function as a heuristic function to the probability pathﬁnd- ing algorithm. With the ﬁtness heuristics function, the path plans generated will always be good. This will prevent all the wasteful discards of low quality path plans. Figure 12: Combine and split trap strategy. Figure 12: Combine and split trap strategy. This paper proposes a new method inspired by genetic algo- rithm to generate interesting and high quality path plans for team AI. Probabilistic pathﬁnding is used for path search and blackboard architecture is used for communication between teammates. The path generation algorithm runs in real time without any preprocessing. Only the standard graph data and a list of gateway points are needed for the A∗search. Control- lable randomness allows the path generation to be tuned eas- ily. The dynamic generation of path plans adds replay value to games. In addition, interesting path plans generated from this method are able to showcase the AI intelligences of the game which is a good selling point for games. g g g g This strategic path plan generation method can apply to other applications that involve path searching. It is best for applications that have many solutions. A good example is traﬃc control system or GPS system. Such systems can plan the path of vehicles to avoid congestion. Congestion condi- tion is used as constraints for ﬁtness test to fail. For exam- ple, a congestion condition can be that the number of vehicle traveling along a road must be less than a maximum number. Another good application is goal planning with many diﬀer- ent ways of achieving the goals. This path plan generation method can generate good quality path plans to achieve the goals. A real example is a mission-based game where there are many ways to solve a mission. Choosing a good plan to solve the mission can bring out the intelligence of game characters. Figure 12: Combine and split trap strategy. REFERENCES Figure 12 shows the most interesting strategy. One of the teammates follows the path of another teammate. To make analysis easier, the enemy teammate right of the player is known as E1 and the enemy teammate on top of E1 will be E2. In this situation, the map can be full of land mines. For every grid path that a character moves, it needs to remove all the landmines and make sure it is safe to travel before it can proceed to the next grid. E1 is ahead of E2 to the player. E2 tries to follow the path of E1 because in that case, E2 does not need to waste eﬀort removing all landmines. This is the ﬁrst constrained ﬁtness function, which is try to overlap paths if they are along the way. [1] T. C. H. John, E. C. Prakash, and N. S. Chaudhari, “Team AI: probabilistic pathﬁnding,” in Proceedings of the International Conference on Game Research and Development, vol. 223 of ACM International Conference Proceeding, pp. 191–198, Perth, Australia, December 2006. [2] S. Rabin, AI Game Programming Wisdom 2, Charles River Me- dia, Hingham, Mass, USA, 2004. [3] M. Buckland, Programming Game AI by Example, Wordware, Plano, Tex, USA, 2005. [4] D. M. Bourg and G. Seeman, AI for Game Developers, O’Reilly, Sebastopol, Calif, USA, 2004. [5] M. Pinter, “Gamasutra,” http://www.gamasutra.com/features/ 20010314/pinter 01.htm. This is also a trap strategy because all teammates trap the player from diﬀerent directions. This should be the sec- ond constraints. That is, all teammates should try to trap the player in diﬀerent directions. [6] A. Kamphuis, M. Rook, and M. H. Overmars, “Tactical path ﬁnding in urban environments,” 2005, http://www.cs.uu.nl/ centers/give/movie/index.php. g p p [7] S. J. Russell and P. Norvig, Artiﬁcial Intelligence: A Modern Ap- proach, Prentice Hall, Englewood Cliﬀs, NJ, USA, 2002. Using these two constrained ﬁtness functions, a very nice strategy is generated from the team AI path plan method. The teammates know what they are doing. They work to- gether and save eﬀort for removing landmines. When they are near to the player, they split their ways to trap the player. [8] D. B. West, Introduction to Graph Theory, Prentice Hall, Upper Saddle River, NJ, USA, 2nd edition, 2000.
https://openalex.org/W4360985063	https://link.springer.com/content/pdf/10.1007/JHEP03(2023)181.pdf	English	null	Dimensional Reduction of Cobordism and K-theory	The Journal of high energy physics/The journal of high energy physics	2,023	cc-by	28,048	Published for SISSA by Springer Published for SISSA by Springer Received: September 6, 2022 Revised: January 24, 2023 Accepted: March 6, 2023 Published: March 24, 2023 Received: September 6, 2022 Revised: January 24, 2023 Accepted: March 6, 2023 Published: March 24, 2023 JHEP03(2023)181 Open Access, c⃝The Authors. Article funded by SCOAP3. Dimensional Reduction of Cobordism and K-theory https://doi.org/10.1007/JHEP03(2023)181 Contents 1 Introduction 2 Cobordism and K-theory 2.1 Cobordism groups of pt and X 2.2 K-theory groups of pt and X 2.3 Gauging and tadpoles 2.4 Dimensional reduction of symmetries 3 Computing cobordism and K-theory on X 3.1 The Atiyah-Hirzebruch spectral sequence 3.1.1 Homological spectral sequence 3.1.2 Cohomological spectral sequence 3.1.3 Trivial ﬁbration and vanishing diﬀerentials on the 3.2 Application to cobordism 3.2.1 Computing Ωξ n(Sk) 3.2.2 Computing Ωξ n(T 2) 3.2.3 Computing ΩSpinc n (K3) 3.2.4 Computing ΩSpinc n (CY3) 3.3 Application to K-theory 3.3.1 Computing K−n(Sk) 3.3.2 Comment on Freed-Witten anomalies 3.3.3 Computing K−n(T k) 3.3.4 Computing K−n(K3) 3.3.5 Computing K−n(CY3) 3.3.6 KO-groups of spheres and tori 3.3.7 Computing KO−n(K3) 4 Physical interpretation 4.1 General aspects 4.2 Example of a Calabi-Yau threefold 4.3 Fate of low-dimensional ΩSpinc n (X) 5 Conclusion A Mathematical tools and results A.1 Short exact sequences, extensions and Ext A.2 Universal coeﬃcient theorem A.3 Properties of Steenrod squares A.4 Wedge sum, smash product and reduced suspension A.5 Cobordism groups of spheres and tori Contents 1 Introduction 1 2 Cobordism and K-theory 3 2.1 Cobordism groups of pt and X 3 2.2 K-theory groups of pt and X 6 2.3 Gauging and tadpoles 7 2.4 Dimensional reduction of symmetries 10 3 Computing cobordism and K-theory on X 12 3.1 The Atiyah-Hirzebruch spectral sequence 13 3.1.1 Homological spectral sequence 13 3.1.2 Cohomological spectral sequence 15 3.1.3 Trivial ﬁbration and vanishing diﬀerentials on the edge 16 3.2 Application to cobordism 17 3.2.1 Computing Ωξ n(Sk) 17 3.2.2 Computing Ωξ n(T 2) 19 3.2.3 Computing ΩSpinc n (K3) 21 3.2.4 Computing ΩSpinc n (CY3) 22 3.3 Application to K-theory 23 3.3.1 Computing K−n(Sk) 24 3.3.2 Comment on Freed-Witten anomalies 25 3.3.3 Computing K−n(T k) 26 3.3.4 Computing K−n(K3) 26 3.3.5 Computing K−n(CY3) 27 3.3.6 KO-groups of spheres and tori 28 3.3.7 Computing KO−n(K3) 28 4 Physical interpretation 29 4.1 General aspects 30 4.2 Example of a Calabi-Yau threefold 32 4.3 Fate of low-dimensional ΩSpinc n (X) 37 5 Conclusion 38 A Mathematical tools and results 39 A.1 Short exact sequences, extensions and Ext 39 A.2 Universal coeﬃcient theorem 40 A.3 Properties of Steenrod squares 41 A.4 Wedge sum, smash product and reduced suspension 42 A.5 Cobordism groups of spheres and tori 43 2.1 Cobordism groups of pt and X JHEP03(2023)181 Dimensional Reduction of Cobordism and K-theory Ralph Blumenhagen, Niccolò Cribiori, Christian Kneißl and Andriana Makridou Max-Planck-Institut für Physik (Werner-Heisenberg-Institut), Föhringer Ring 6, 80805 München, Germany E-mail: blumenha@mpp.mpg.de, cribiori@mpp.mpg.de, ckneissl@mpp.mpg.de, amakrido@mpp.mpg.de Abstract: It has been proposed that cobordism and K-theory groups, which can be mathematically related in certain cases, are physically associated to generalised higher-form symmetries. As a consequence, they should be broken or gauged in any consistent theory of quantum gravity, in accordance with swampland conjectures. We provide further support to this idea by showing that cobordism and K-theory groups of a general manifold X reproduce the pattern of symmetries expected from the dimensional reduction of the theory on X, as well as their breaking and gauging. To this end, we employ the Atiyah-Hirzebruch spectral sequence to compute such groups for common choices of X in string compactiﬁcations. Keywords: String and Brane Phenomenology, D-Branes, Global Symmetries ArXiv ePrint: 2208.01656 Open Access, c⃝The Authors. Article funded by SCOAP3. Open Access, c⃝The Authors. Article funded by SCOAP3. 3 Computing cobordism and K-theory on X 4 Physical interpretation 5 Conclusion 1 Introduction The absence of global symmetries is considered to be a fundamental property of quantum gravity. It is also one of the central and most tested conjectures in the swampland program [1, 2]. It is believed to hold true even when the deﬁnition of symmetry is extended, for example by including higher-form global symmetries [3], see e.g. [4, 5] for recent works. The cobordism conjecture [6] is yet another instance of this general feature. It associates a generalised global charge to any non-trivial cobordism group and thus it postulates its cancellation. In this work, we apply the cobordism conjecture to setups with a given, ﬁxed background X. JHEP03(2023)181 The cobordism charge is topological and should vanish on consistent quantum gravity backgrounds. These can, and in general have to, be endowed with defects and/or gauge ﬁelds, in order respectively to break or gauge the cobordism symmetry. As already discussed in [6], breaking the symmetry can lead to the prediction of new defects in quantum gravity. This idea has been further developed in several directions [7–13], in relation to the Hypothesis H [14] (see also [15–17]), and it has been given an eﬀective dynamical description [18– 22]. Generally, the introduction of new ingredients into the setup modiﬁes the structure (topological and geometric) of the given compact manifold. The aim is thus to get closer and closer to the postulated quantum gravity structure, even if a preferred path does not seem to be obvious. In this respect, in [23] the Whitehead tower construction has been identiﬁed as a possible organising principle for topological structures entering the cobordism conjecture. One can think of compact manifolds and defects in cobordism as being associated to closed and open strings respectively. Besides, it is well-known that open strings are related to K-theory [24]. At the mathematical level, a correspondence has been uncovered between certain versions of cobordism and K-theory. It is deeply rooted into the structure of such generalised (co)homology theories and relies on established theorems, such as the Conner-Floyd [25] and the Hopkins-Hovey [26] theorems. As proposed in [27], at the physical level it is tempting to interpret this correspondence as a manifestation of open-closed string duality. This would directly imply that if the cobordism charge must cancel, the K-theory charge should undergo the same fate. Indeed, cancellation of K-theory charge has already been proposed in the literature from diﬀerent perspectives [28–31]. B Tables of cobordism groups B Tables of cobordism groups 46 C Tables of K- and KO-theory groups 49 5 Conclusion – i – 1 Introduction Elaborating on this intuition, in [27] it has also been discussed how the structure behind certain tadpoles in string theory is precisely given by the interplay between cobordism and K-theory charges. This observation is pertinent to the case in which symmetries are gauged rather than broken, which has been little explored so far. While cobordism and K-theory groups considered in [6, 27] are those of the point, e.g. ΩSpinc n (pt) and K−n(pt), a more generic situation is to consider the case in which pt – 1 – is replaced by a topological space X. In fact, the Conner-Floyd and the Hopkins-Hovey theorems also hold in this more general case. At the mathematical level this requires that, given a compact n-dimensional manifold M representative of a cobordism class, there exists a continuous map f : M →X with certain properties at the boundary of the cobordism. As we are going to argue, at the physical level it means that one is ﬁxing the topology of X and studying global symmetries associated to ﬂuctuations in topology along the unﬁxed, complementary directions (if any) within M. On the other hand, ﬁxing a manifold X is most natural from a string theory point of view, where one is typically forced to perform a background-dependent analysis. Thus, this is yet another motivation for looking at cobordism and K-theory groups of a given X. JHEP03(2023)181 It is the purpose of this paper to work out in detail such cobordism and K-theory classes on X and to see how the Hopkins-Hovey isomorphism continues to hold beyond the trivial case, X = pt. For simple examples of X, such as spheres, tori and Calabi-Yau two- and three-folds, we can be very explicit and understand the results in terms of a straightforward dimensional reduction of global symmetries and their gauging and breaking. More precisely, we are going to show that, when performing a dimensional reduction of a given eﬀective theory on X, a lattice of charges arises which is reproduced by the evaluation of cobordism and K-theory groups ΩSpinc n (X) and K−n(X) (or ΩSpin n (X) and KO−n(X)). 1As already noticed in [27], the presence of 2-torsion classes can lead to a decoupling of the two charges in the tadpole. 1 Introduction In this way, by showing consistency upon compactiﬁcation, our analysis gives further support to the proposal of [27] that gauging of cobordism and K-theory symmetries related via the Hopkins-Hovey isomorphism generically occurs simultaneously, so that both charges appear in the same Bianchi identities and tadpole conditions.1 The technical tool we employ to compute cobordism and K-theory groups of X is the Atiyah-Hirzebruch spectral sequence. It is a standard technique in algebraic topology and we will refer mainly to [32, 33]; see also [34] for a recent review and application to particle physics. This spectral sequence has been applied in a string theoretic context already in [35–38] among others, and those works will be relevant for us. Aiming at giving a self-contained presentation for a physicist reader, we will introduce the aforementioned technique in a pedagogical way and collect known mathematical results, which are employed systematically in the computations. After computing the cobordism and K-theory groups of interest, we explain how to extract the information on the dimensional reduction of global symmetries (either broken or gauged) contained in them. We discuss both cobordism and K-theory groups and show how they cooperate in reproducing known tadpoles in string theory for compactiﬁcations on X, hence extending the analysis of [27]. This work is organised as follows. In section 2, we review the main properties of cobordism and K-theory needed for our analysis. In particular, we explain the diﬀerence between groups of the point and of a manifold X, and we recall the physical consequences of the correspondence between cobordism and K-theory, leading to a bottom-up construction of tadpole cancellation conditions. We also review how continuous higher-form symmetries (either broken or gauged) behave under dimensional reduction, when this is performed in – 2 – standard (co)homology. We expect that cobordism and K-theory should give a reﬁned description. In section 3, we compute cobordism and K-theory groups of X via the Atiyah- Hirzebruch spectral sequence, for X equal to spheres, tori, Calabi-Yau two- and three-folds. We explain in detail the main subtleties entering such computations and how they can be solved in our (simple) circumstances. In section 4, we propose a physical interpretation of these results in terms of the dimensional reduction of higher-form symmetries, when the theory is compactiﬁed on X. We show explicitly how the information of the reduction is encoded into cobordism and K-theory groups of X. 2 Cobordism and K-theory In this section, we review elements of cobordism and K-theory which are relevant for our analysis. We concentrate in particular on the diﬀerence between groups of the point and of a given compact manifold X. We recall their interpretation in terms of global symmetries and the correspondence between cobordism and K-theory, which is relevant when gauging such symmetries. 1 Introduction This formalism is more precise than standard (co)homology, since it also takes into account quantum mechanical eﬀects, such as cancellation of Freed-Witten anomalies. The latter are however absent for the simple backgrounds we consider. In section 5, we draw our conclusions and outline future research directions. Additional material is organised into three appendices: in appendix A, we summarise further useful notions of algebraic topology and we present a calculation of cobordism groups of spheres and tori by induction. In appendix B and C, we collect tables with all the main results from section 3. JHEP03(2023)181 2We recall that a generalised (co)homology theory can be deﬁned axiomatically. It satisﬁes the same axioms as ordinary (co)homology, but for the dimension axiom [32, 33]: Hn(pt) = 0 for n > 0. 2.1 Cobordism groups of pt and X Cobordism is a generalised homology theory classifying compact manifolds of the same dimension.2 Two compact n-dimensional manifolds M and N are cobordant, if there exists an (n+1)-dimensional compact manifold W called cobordism, such that ∂W = M ⊔N. The notion of cobordism can be reﬁned by considering tangential structures on M, N and W. An elementary example is requiring M, N to be oriented. The cobordism W is then such that ∂W = M ⊔¯N, where the bar refers to reversed orientation. In general, one can also consider manifolds with more reﬁned tangential structures, as for example Spin or Spinc. Being cobordant is an equivalence relation. Under disjoint union, the set of equivalence classes of compact n-dimensional manifolds with ξ-structure is an abelian group, called the cobordism group Ωξ n(pt). More precisely, this is the cobordism group of the point (pt). The cobordism conjecture of [6] is the statement that there exists a not necessarily unique quantum gravity structure QG such that the associated cobordism groups are trivial, ΩQG n (pt) = 0. (2.1) (2.1) The physical interpretation is the following. One should think of a compact n- dimensional manifold M representative of the cobordism class [M] as the compact part 2We recall that a generalised (co)homology theory can be deﬁned axiomatically. It satisﬁes the same axioms as ordinary (co)homology, but for the dimension axiom [32, 33]: Hn(pt) = 0 for n > 0. – 3 – of a d-dimensional background of a theory of quantum gravity, such as string theory. A non-vanishing cobordism group, Ωξ n(pt) ̸= 0, is then interpreted in [6] as the presence of a (d−n−1)-form global symmetry in the d-dimensional eﬀective ﬁeld theory with a conserved global current, dJn = 0. Via gluing a non-trivial element [M] ∈Ωξ n(pt) to Rn, one can construct a (d −n) dimensional defect. This is a gravitational soliton carrying charge under the (d −n −1)-form global symmetry.3 In order to avoid global symmetries in quantum gravity, the cobordism charge has either to be broken or gauged. As explained in [6], breaking the symmetry requires the introduction of (d −n −1)- dimensional defects, such that current is not conserved anymore, JHEP03(2023)181 0 ̸= dJn = X def j δ(n+1)(∆d−n−1,j) , (2.2) (2.2) where the δ-functions are the Poincaré dual of the (d −n −1)-cycles wrapped by the defects. 3We thank the referee for pointing this out to us. 2.1 Cobordism groups of pt and X The structure is modiﬁed, ξ →ξ + defects, in such a way that the cobordism group with the reﬁned structure is possibly vanishing, Ωξ+defects n (pt) = 0. Thus, the cobordism conjecture predicts additional objects in the eﬀective ﬁeld theory. Instead, if the background has vanishing charge [M] = 0 ∈Ωξ n(pt), even if the group is still non-trivial, the symmetry can be gauged. In this case, there exist appropriate gauge ﬁelds such that Jn = dFn−1 . (2.3) (2.3) One can now reﬁne the structure by introducing these gauge ﬁelds, in such a way that manifolds in the zero equivalence class, [M] = 0, lie in the cokernel of the map Ωξ+gauge n (pt) →Ωξ n(pt). The discussion can be generalised by going from the cobordism group of the point to that of a topological space X, denoted Ωξ n(X). Consider continuous maps f : M →X and g : N →X, which can be understood as deformations of M and N into X, even if the construction is more general. In particular, the dimension of X does not need to be the same as of M and N. Two pairs (M, f) and (N, g) are cobordant if there is a cobordism W, such that ∂W = M ⊔N, together with a map h : W →X appropriately restricting to f and g at the boundary ∂W. One can show that this is an equivalence relation and the cobordism group Ωξ n(X) is the set of equivalence classes of pairs (M, f). For X = pt we recover the deﬁnition of cobordism group of the point. We give a schematic representation in ﬁgure 1. In the present work, we are interested in the case in which X is a compact Spin or Spinc manifold without any additional structure, such as (higher-form) gauge ﬁelds. In particular, we will consider X = {Sk, T k, K3, CY3}, but the analysis can be repeated in principle for any other space. Our choice is motivated by the fact that these manifolds are commonly employed in string compactiﬁcations and they turn out to be simple enough to explicitly evaluate Ωξ n(X). When considering cobordism groups of the point, Ωξ n(pt), one is looking at global symmetries of the d-dimensional eﬀective theory by scanning through all possible topologies – 4 – M N W X h f g Figure 1. 2.1 Cobordism groups of pt and X Cobordism (W, h) between (M, f) and (N, g). JHEP03(2023)181 Figure 1. Cobordism (W, h) between (M, f) and (N, g). of n-dimensional compact manifolds. We will see that, when going from pt to X, the cobordism group is generically enlarged. In particular, the classes Ωξ n(pt) will also be present in Ωξ n(X), but new classes can appear, depending on the topology of X. Intuitively, a non-trivial topology carries a charge which is detected by the cobordism group and, in turn, it increases its rank. This intuitive picture on the relation between Ωξ n(pt) and Ωξ n(X) can be made precise by recalling a standard result which follows from the Splitting Lemma for abelian groups and which we will use systematically in our analysis. Consider the forgetful map φ : Ωξ n(X) →Ωξ n(pt) , (2.4) (2.4) ting as φ([M, f]) = [M]. Its kernel is called the reduced cobordism group, denoted as acting as φ([M, f]) = [M]. Its kernel is called the reduced cobordism group, denoted as ker φ ≡˜Ωξ n(X) , (2.5) (2.5) and one has ˜Ωξ n(pt) = 0 by deﬁnition. One can show that this map is surjective and that the short exact sequence 0 −→˜Ωξ n(X) −→Ωξ n(X) φ −→Ωξ n(pt) −→0 (2.6) (2.6) is split. Therefore, one ﬁnds Ωξ n(X) = Ωξ n(pt) ⊕˜Ωξ n(X), (2.7) (2.7) which is valid for any structure ξ. In accordance with our intuitive picture, when passing from Ωξ n(pt) to Ωξ n(X) the rank of the group is indeed increased. We interpret ˜Ωξ n(X) as the part of global symmetries genuinely stemming from having ﬁxed a manifold X. ˜ We will see explicitly in our examples that ˜Ωξ n(X) can be further decomposed into a direct sum of several pieces, but the details of the splitting depend on the topology of X. Each of these pieces will give rise to global symmetries. Therefore, for a given X, we will uncover a lattice of global symmetries organised according to the topology. These global symmetries are deemed pathological in a theory of quantum gravity, so one has to have ΩQG n (X) = 0, (2.8) (2.8) for X a consistent on-shell background of quantum gravity. This is implied by the cobor- dism conjecture. – 5 – 2.2 K-theory groups of pt and X K-theory is a generalised cohomology theory classifying vector bundles over a space X. Consider two vector bundles E and F over X, which can be of diﬀerent rank, and construct (E, F) = E −F. Then, one can introduce an equivalence relation (E ⊕H, F ⊕H) ∼(E, F), for any bundle H. The set of equivalence classes is a group, called the K-theory group K(X). To be precise, K(X) is the set of equivalence classes of complex vector bundles. We will also consider the set of equivalence classes of real vector bundles, denoted KO(X). Besides the simple choice X = pt, in this work we will focus on K-theory groups of X = {Sk, T k, K3, CY3}. JHEP03(2023)181 JHEP03(2023)181 Similarly to what we discussed for cobordism, the relation between K-theory groups of the point and of X can be made mathematically precise by employing the Splitting Lemma. One can indeed consider the map ϕ : K(X) →K(pt), (2.9) (2.9) which can be identiﬁed with the (virtual) rank which can be identiﬁed with the (virtual) rank which can be identiﬁed with the (virtual) rank ϕ[(E, F)] = rank(E) −rank(F), (E, F) ∈K(X) . (2.10) (2.10) The reduced K-theory group is then deﬁned as The reduced K-theory group is then deﬁned as The reduced K-theory group is then deﬁned as The reduced K-theory group is then deﬁned as ker ϕ ≡f K(X), (2.11) (2.11) namely it classiﬁes vector bundles on X with the same rank. One can show that the map ϕ is surjective and that the short exact sequence namely it classiﬁes vector bundles on X with the same rank. One can show that the map ϕ is surjective and that the short exact sequence 0 →f K(X) →K(X) ϕ→K(pt) →0 (2.12) (2.12) is split, K(X) = K(pt) ⊕˜K(X). is split, K(X) = K(pt) ⊕˜K(X). By introducing the reduced suspension Σ, the higher reduced K-theory groups are then deﬁned as ˜ ˜ ˜K−n(X) = ˜K(ΣnX) , g KO −n(X) = g KO(ΣnX), (2.13) K (X) = K(Σ X) , g KO −n(X) = g KO(ΣnX), (2.13) (2.13) for n ∈Z, n ≥0. By recalling the relations K(X) = f K(X ⊔pt) and KO(X) = g KO(X ⊔pt) and the properties of the reduced suspension (see appendix A.4), one has also for n ∈Z, n ≥0. 4In our conventions, the notation nG means Gn = G ⊕. . . ⊕G \| {z } n times . 2.2 K-theory groups of pt and X Let us ﬁnally note that, for a physical interpretation in terms of (wrapped) D-branes in critical type IIB and type I string theory, we should restrict to values of n such that (2.17) n + k ≤10 (2.17) n + k ≤10 where we deﬁned k = dim(X). 2.2 K-theory groups of pt and X By recalling the relations K(X) = f K(X ⊔pt) and KO(X) = g KO(X ⊔pt) and the properties of the reduced suspension (see appendix A.4), one has also f K(Sn) = f K(ΣnS0) = K−n(pt), g KO(Sn) = g KO(ΣnS0) = KO−n(pt). (2.14) (2.14) Then, the full generalised cohomology theories can be constructed as K−n(X) = K−n(pt) ⊕f K−n(X), KO−n(X) = KO−n(pt) ⊕g KO −n(X). (2.15) (2.15) – 6 – They satisfy the important property known as Bott periodicity They satisfy the important property known as Bott periodicity K−n(X) = K−n+2(X), KO−n(X) = KO−n+8(X) . (2.16) (2.16) For X = pt, the physical interpretation of these groups is well-known: they classify D-branes in string theory [24] (see [39, 40] for a review). In particular, KO−n(pt) classify p = 9 −n branes in type I, while K−n(pt) classify p = 9 −n and p = 10 −n branes in type IIB and IIA [24, 41] respectively. This can be understood directly from (2.14), meaning that K-theory groups of the point at degree n classify D-branes in ﬂat space which are point-like with respect to Sn. Note that this includes the type II and type I BPS D-branes charged with respect to the R-R forms, but it can also give stable non-BPS torsion branes. JHEP03(2023)181 As for cobordism, one can argue that a global symmetry is associated to each non- vanishing K-theory group, which must cancel in quantum gravity [28, 29]. It is known that K-theory charges are gauge charges [42, 43], so that their global symmetries will be gauged rather than broken. When passing from K-theory groups of the point to groups of X we are in fact introducing additional global symmetries. In particular, the reduced K-theory groups contain the information on symmetries genuinely associated to the choice of X. We will see explicitly in our examples that f K−n(X) and g KO −n(X) can be further split into a direct sum of several pieces, but the details of the splitting depend on the topology of X. In terms of D-branes, the ﬁrst term on the right hand side of (2.15) is a universal contribution, i.e. branes not depending on the choice of X and wrapping it completely. The second term contains branes wrapping internal cycles of X (and directions orthogonal to X). 2.3 Gauging and tadpoles As we brieﬂy review in this section, at the mathematical level, cobordism and K-theory are deeply related. For Spin/Spinc cobordism on one side and real/complex K-theory on the other side, this fact is based on rigorous theorems [25, 26]. A physical interpretation of this inherent relation in terms of the gauging of global symmetries has been provided in [27], which will also be recalled. For more details, we refer the reader to that paper. It is instructive to have a close look at the known cobordism and K-theory groups of the point. In the table 1 we list the cobordism groups4 and in table 2 the corresponding K-theory groups, where Bott periodicity is manifest. O th t th i hi t 8 f S i b di d l K th One can see that they are isomorphic up to n = 8, for Spin cobordism and real K-theory, and up to n = 4, for Spinc cobordism and complex K-theory. Indeed, one can introduce a – 7 – n 0 1 2 3 4 5 6 7 8 9 10 ΩSpin n (pt) Z Z2 Z2 0 Z 0 0 0 2Z 2Z2 3Z2 ΩSpinc n (pt) Z 0 Z 0 2Z 0 2Z 0 4Z 0 4Z ⊕Z2 Table 1. Spin and Spinc cobordism groups of the point up to n = 10. Table 1. Spin and Spinc cobordism groups of the point up to n = 10. n 0 1 2 3 4 5 6 7 8 9 10 KO−n(pt) Z Z2 Z2 0 Z 0 0 0 Z Z2 Z2 K−n(pt) Z 0 Z 0 Z 0 Z 0 Z 0 Z JHEP03(2023)181 JHEP03(2023)181 Table 2. K- and KO-groups of the point up to n = 10. Table 2. K- and KO-groups of the point up to n = 10. map known as the Atiyah-Bott-Shapiro (ABS) orientation [44] α : ΩSpin n (pt) →KO−n(pt) αc : ΩSpinc n (pt) →K−n(pt) (2.18) map known as the Atiyah-Bott-Shapiro (ABS) orientation [44] map known as the Atiyah-Bott-Shapiro (ABS) orientation [44] α : ΩSpin n (pt) →KO−n(pt) αc : ΩSpinc n (pt) →K−n(pt) (2.18) (2.18) and then show that there exist isomorphisms [45] and then show that there exist isomorphisms [45] ΩSpin n (pt)/ ker α ∼= KO−n(pt) , (2.19) ΩSpinc n (pt)/ ker αc ∼= K−n(pt) . 2.3 Gauging and tadpoles (2.20) (2.19) (2.20) Explicitly, the two ABS orientations at ﬁxed degree n are given by the Todd genus, i.e. the index of the Spinc Dirac operator Explicitly, the two ABS orientations at ﬁxed degree n are given by the Todd genus, i.e. the index of the Spinc Dirac operator αc n([M]) = Td(M) ≡ Z M tdn(M) , (2.21) (2.21) and by the index of the Dirac operator on M, respectively [46] and by the index of the Dirac operator on M, respectively [46] αn([M]) =                      ˆA(M) n = 8m, ˆA(M)/2 n = 8m + 4, dim H mod 2 n = 8m + 1, dim H+ mod 2 n = 8m + 2, 0 otherwise, (2.22) αn([M]) =                      ˆA(M) n = 8m, ˆA(M)/2 n = 8m + 4, dim H mod 2 n = 8m + 1, dim H+ mod 2 n = 8m + 2, 0 otherwise, (2.22) (2.22) where ˆA(M) is the ˆA genus and H (H+) the space of (positive) harmonic spinors. In [27], it has been proposed that whenever the ABS orientation exists, as for example in type IIB string theory on a Spinc manifold, it is not just the K-theory charge that is gauged but actually a combination of the K-theory and the respective cobordism charge. Hence, schematically the gauging can proceed like dFn−1 = JK n + a(n) Jcobord n (2.23) (2.23) (no sum over n in the second term on the right hand side), i.e. it is a linear combination of K-theory and cobordism global charges that is gauged. The constant a(n) is not a priori (no sum over n in the second term on the right hand side), i.e. it is a linear combination of K-theory and cobordism global charges that is gauged. The constant a(n) is not a priori – 8 – ﬁxed and in certain cases can also be vanishing. It would be interesting to uncover whether and how it can be ﬁxed just from mathematical data, without any reference to a concrete quantum gravity theory, like string theory. 2.3 Gauging and tadpoles Upon integration over a compact space, such a Bianchi identity implies a charge neutrality condition, which was shown in [27] to match some of the tadpole cancellation conditions known from string theory. More precisely, given non-vanishing groups ΩSpinc n (pt) and K−n(pt), then the current Jcobord n on the right hand side of (2.23) is given by the sum over the cobordism invariants of ΩSpinc n (pt), denoted as µj n. For 0 ≤n ≤6, a list of the independent cobordism invariants of ΩSpinc n (pt) is5 JHEP03(2023)181 µ0 = td0(M) = 1 , µ2 = td2(M) = 1 2c1(M) , µ1 4 = td4(M) = 1 12 c2(M) + c2 1(M) , µ2 4 = c2 1(M) , µ1 6 = td6(M) = 1 24c2(M) c1(M) , µ2 6 = 1 2c3 1(M) . (2.24) (2.24) Notice that the ABS orientation always provides one cobordism invariant, but in general there can be more and they all have to be taken into account since they all contribute as global charges in general. The K-theory theory current JK n is simply deﬁned by the delta functions for the localised branes classiﬁed by K−n(pt). Upon integration over a representative manifold M ∈[M], one thus obtains a tadpole constraint of the form 0 = Z M dFn−1 = Z M X i∈def Qi δ(n)(∆10−n,i) + Z M X j∈inv a(n) j µj n , (2.25) (2.25) where ∆10−n,i is the submanifold wrapped by the i-th Dp-brane (with p = 9 −n in type I/IIB) with charge Qi. Notice that (2.25) is valid oﬀ-shell, for all compact manifolds cobordant with M. Let us give an illustrative example taken from [27]. Consider ΩSpinc 6 (pt) = Z ⊕Z. This is a three-form global symmetry in ten dimensions and K−6(pt) classiﬁes D3-branes. Specialising (2.25) to this case, we get 0 = Z M X i∈def Qi δ(6)(∆4,i) + Z M a(6) 1 c1 c2 24 + a(6) 2 c3 1 2 ! . (2.26) (2.26) For the particular choice a(6) 1 = −12 and a(6) 2 = −30, this tadpole is realised by F-theory on a smooth Calabi-Yau fourfold which is elliptically ﬁbered over a base M. Many more examples were discussed in [27], supporting the above conjecture about the simultaneous gauging of K-theory and cobordism global symmetries. For the type I string, the Spin cobordism and the KO groups are relevant. 5For the Chern classes, we use the shorthand notation ci(M) ≡ci(TM). Similarly for Stiefel- Whitney classes. 2.3 Gauging and tadpoles For continuous global symmetries, corresponding to Z groups in tables 1 and 2, one similarly 5For the Chern classes, we use the shorthand notation ci(M) ≡ci(TM). Similarly for Stiefel- Whitney classes. – 9 – gets tadpole cancellation conditions known from the type I literature. For gauging torsion Z2 groups, even though there are no gauge ﬁelds that one can introduce, there is an in principle mixed Z2-valued charge neutrality condition. For instance, for the case ΩSpin 2 (pt)/KO2(pt) this reads 0 = Z M X i Qi δ(2)(∆8,i) −a(2) α2(M) mod 2 , (2.27) (2.27) where we have a contribution from non-BPS d D7-branes and from the cobordism group. However, for a(2) even, the latter contribution decouples in this equation and the global symmetry ΩSpin 2 = Z2 still needs to be broken by appropriate defects. Since the generator of ΩSpin 2 is M = S1 p × S1 p, a manifold that is a valid background of the type I string (without needing the presence of a d D7-brane), in this case a(2) should indeed be even. Whether such a decoupling holds generically for all Zn torsion charges remains to be understood. JHEP03(2023)181 Since the Hopkins-Hovey isomorphism holds also for cobordism and K-theory groups on generic backgrounds X, the gauging should carry over to this more general case as well, if the assertion of [27] is really correct. To check this quantitatively, we need to understand how such groups are actually computed, at least for some treatable classes of background manifolds X, and what kind of physical information they contain. The expectation is that, when specifying a background X, at least part of the cobordism and K-theory groups can be understood via dimensional reduction on X. 2.4 Dimensional reduction of symmetries Before we delve into the mathematically rather involved evaluation of cobordism and K-theory groups, it is instructive to review how dimensional reduction is usually performed in (co)homology. This will help us in appreciating what one really gains from using the description in terms of cobordism and K-theory. Let us consider an eﬀective theory in d dimensions, where of course we have d = 10 in mind. Recall that to a continuous global p-form symmetry, there exists an associated current Jn, with n = d −p −1, which is closed dJn = 0 . (2.28) (2.28) To break the symmetry, one introduces defects such that the current ceases to be closed anymore dJn = δ(n+1)(∆p) ̸= 0 , (2.29) (2.29) where ∆p is the cycle wrapped by the defect. To gauge the symmetry, one introduces (and further integrates over) gauge ﬁelds coupling minimally to the current S = Z −1 2Fp+2 ∧∗Fp+2 + Cp+1 ∧Jn + . . . , Fp+2 = dCp+1, (2.30) (2.30) such that from the equations of motion of Cp+1 it follows that the current is trivial in cohomology (i.e. exact) Jn = (−1)p d ∗Fp+2 . (2.31) (2.31) – 10 – – 10 – When performing a dimensional reduction over a compact space X, one typically expands the various objects, such as currents and gauge ﬁelds, in a cohomological basis of X. The expansion coeﬃcients are ﬁelds propagating along the external non-compact dimensions. Classically, the expansion can be performed in de Rham cohomology, Hp(X; R). Quantum mechanics typically imposes charges to be quantised, and thus one would rather consider singular cohomology with integer coeﬃcients, Hp(X; Z). Let us now compactify the theory on a k-dimensional space X. We get a D = d −k dimensional eﬀective theory with broken and gauged symmetries inherited from the parent theory. In general, a given p-form symmetry in D dimensions can receive contributions from diﬀerent (p + q)-form symmetries of the d-dimensional theory. To these contributions, we associate currents Jn+m, now with p = D −n −1 and q = k −m, wrapping m = 0, 1, . . . , k cycles in X and extending along n directions in the non-compact space. Let us consider a basis of cohomology, ω(m)a ∈Hm(X; Z), where a = 1, . . . , bm, with bm the Betti numbers. 2.4 Dimensional reduction of symmetries We can decompose the currents as JHEP03(2023)181 Jn+m = bm X a=1 j(m)a n ∧ω(m)a . (2.32) (2.32) Thus, p-form symmetries in D dimensions arise from the set of currents j(m)a n , for a = 1, . . . bm and m = 0, . . . , k. Since we are performing an expansion in cohomology, we see that if the Jn+m are closed, dJn+m = 0, they produce a lattice of global p-form symmetries in D dimensions, namely Thus, p-form symmetries in D dimensions arise from the set of currents j(m)a n , for a = 1, . . . bm and m = 0, . . . , k. Since we are performing an expansion in cohomology, we see that if the Jn+m are closed, dJn+m = 0, they produce a lattice of global p-form symmetries in D dimensions, namely dj(m)a n = 0, ∀a = 1, . . . bm, ∀m = 0, . . . k . (2.33) (2.33) Breaking or gauging global symmetries in d dimensions does not jeopardise this structure and one generically expects a lattice of broken or gauged symmetry in the lower dimensional theory to be produced. They arise from diﬀerent broken or gauged symmetries of the original theory. As we will discuss next, delta functions for the defects breaking the symmetry and gauge ﬁelds can be similarly expanded in cohomology. To break the currents Jn+m we need forms δ(n+m+1) in the d-dimensional theory such that dJn+m = δ(n+m+1)(∆p+q) ̸= 0, with p + q = d −n −m −1. These forms represent defects wrapping submanifolds ∆p+q = Πp × Σq of the d-dimensional space, where Πp is a p-dimensional submanifold of the non-compact space, while Σq is a q-dimensional cycle of X. For the global symmetry to be broken in the D-dimensional theory, we take p = D −n −1 and q = k −m, in such a way that the defect in the dimensionally reduced theory has codimension n + 1 (for n = 0, namely a (D −1)-form global symmetry, we get a domain wall in D dimensions). We can then formally expand in cohomology δ(n+m+1)(∆p+q) = bm X a=1 δ(n+1)(Πp)(m)a ∧ω(m)a. (2.34) (2.34) Thus, from any defect δ(n+m+1) in d dimensions we generate a lattice of codimension n + 1 defects in D dimensions, δ(n+1)(Πp)(m)a. 2.4 Dimensional reduction of symmetries They can be used to break the lattice of global currents (2.33), i.e. dj(m)a n = δ(n+1)(ΠD−n−1)(m)a ̸= 0, (2.35) (2.35) where again this is really a set of equations for a = 1, . . . bm and m = 0, . . . , k. – 11 – To gauge the currents Jn+m we need gauge ﬁeld strengths Fn+m−1 in the d-dimensional theory such that Jn+m = dFn+m−1 (here Fn+m−1 is the magnetic dual of the ﬁeld strength in (2.31)). The dimensional reduction of these Bianchi identities can be performed in analogy to what was previously done. In particular, one can replace Jn+m →Fn+m−1 in the above analysis and repeat the same steps. One thus ﬁnds a lattice of (n −1)-form ﬁeld strengths f(m)a n−1 in D dimensions which are gauging the n-form currents j(m)a n , thus giving the Bianchi identities ( ) j(m)a n = df(m)a n−1 . (2.36) (2.36) In general, the currents j(m)a n will also contain contributions from localised delta functions δ(n)(ΠD−n), arising from the reduction of D-branes in d dimensions. JHEP03(2023)181 After computing cobordism and K-theory groups of X in the next section, we will show that they exhibit exactly the same pattern explained here for the dimensional reduction of broken and gauged symmetries on X. We will see that the description in terms of cobordism and K-theory provides by itself an organising principle for the various symmetries in the dimensionally reduced theory, something which is not transparent from the above analysis. Indeed, contributions to a given (broken or gauged) p-form symmetry in D-dimensions and its corresponding charged objects will be encoded into K−n(X) and ΩSpinc k+n (X), for p = D −1 −n and n ≥0. We will see that for −k ≤n ≤0 the corresponding D-brane, respectively gravitational soliton, does not consistently ﬁt into the D-dimensional space so that there does not exist any obvious physical interpretation of the cobordism and K-theory groups. For type I, we have a similar story for ΩSpin k+n(X) and KO−n(X). This behavior under compactiﬁcation gives further support to the interpretation of K-theory and cobordism groups as higher-form charges in an eﬀective ﬁeld theory. As said, the above analysis was the classical dimensional reduction using (de Rham) singular (co)homology without torsion. 2.4 Dimensional reduction of symmetries Therefore, all objects in D-dimensions are the result of a naive dimensional reduction along homological cycles in X, nothing is lost and nothing new arises in D-dimensions. However, the appearance of torsion through the reﬁnement to generalised (co)homology theories can open up new decay channels of non-BPS branes, and it is known that new stable torsion branes can appear on X, even if they were not present in d dimensions. Moreover, for wrapped D-branes there can be quantum eﬀects that spoil these simple (classical) expectations. For instance, some wrapped branes can develop a Freed-Witten anomaly so that they should actually not be present in the D-dimensional theory. All these eﬀects are taken into account by the description in terms of cobordism and K-theory rather than (co)homology. 3.1 The Atiyah-Hirzebruch spectral sequence The Atiyah-Hirzebruch spectral sequence (AHSS) is a tool for calculating generalised (co)homology groups. We will use the homological AHSS to determine the cobordism groups Ωξ n(X) and the cohomological AHSS for K-groups K−n(X), with X a compact manifold of dimension up to ten. We will specialise to the choices of X mentioned in the previous section, namely X = {Sk, T k, K3, CY3}, and to ξ = Spin, Spinc. Standard references in the mathematical literature are for example [32, 33, 47], for introductory material, and [48] for a physics-motivated treatment. A nice, recent review with applications to anomaly cancellation in physics can be found e.g. in [34]. With the goal of providing a self-contained exposition of the subject, in the present section we will brieﬂy review the main steps of these techniques and some relevant mathematical results. Further background is presented in appendix A. JHEP03(2023)181 3 Computing cobordism and K-theory on X This section concerns the computation of cobordism and K-theory groups using a technique known as Atiyah-Hirzebruch spectral sequence. We ﬁrst introduce spectral sequences for (generalised) homology and cohomology and then we use them to compute physically relevant cobordism and K-theory groups, respectively. This material is well-known to experts, but it might be less familiar to physicists not directly working on the subject. Thus, we believe – 12 – that there is some pedagogical value in reviewing it. The reader interested in the physical interpretation might skip this section at ﬁrst reading and go directly to section 4. We collect all results of this section in appendices B and C. 6In the literature, the case with a trivial ﬁbration is sometimes referred to as the AHSS exclusively. Here, following [32–34, 47], we discuss the more general case of a not necessarily trivial ﬁbration and still refer to it as AHSS. 7In certain cases, it is also possible to use the spectral sequence “backwards” and compute for example Gn(F) from the Gn(E). 3.1.1 Homological spectral sequence 2 E2 0,2 E2 1,2 E2 2,2 E2 3,2 E2 4,2 . . . 1 E2 0,1 E2 1,1 E2 2,1 E2 3,1 E2 4,1 . . . 0 E2 0,0 E2 1,0 E2 2,0 E2 3,0 E2 4,0 . . . 0 1 2 3 4 5 Table 3. Example of a second page E2 of a ﬁrst quadrant homological spectral sequence and all possible d2 diﬀerentials. The non-vanishing d2 are shown by purple and blue arrows. Table 3. Example of a second page E2 of a ﬁrst quadrant homological spectral sequence and all possible d2 diﬀerentials. The non-vanishing d2 are shown by purple and blue arrows. JHEP03(2023)181 5 ... ... ... ... ... 4 E3 0,4 E3 1,4 0 E3 3,4 E3 4,4 . . . 3 0 E3 1,3 E3 2,3 E3 3,3 0 . . . 2 E3 0,2 0 0 E3 3,2 E3 4,2 . . . 1 E3 0,1 0 E3 2,1 0 E3 4,1 . . . 0 E3 0,0 E3 1,0 E3 2,0 0 E3 4,0 . . . 0 1 2 3 4 5 Table 4. Third page E3 of the same spectral sequence and all possible d3 diﬀerentials. The blue diﬀerentials have (co-)killed the page elements they were acting on, while the purple ones let them partially survive. The black elements, on which no diﬀerential acted, carried over intact to the next page, i.e. E3 p,q ∼= E2 p,q. Table 4. Third page E3 of the same spectral sequence and all possible d3 diﬀerentials. The blue diﬀerentials have (co-)killed the page elements they were acting on, while the purple ones let them partially survive. The black elements, on which no diﬀerential acted, carried over intact to the next page, i.e. E3 p,q ∼= E2 p,q. The page Er together with the diﬀerential dr fully determine the next page Er+1, but then additional input is necessary to determine the diﬀerentials dr+1. Intuitively, the spectral sequence calculates a generalised (co)homology by ﬁrst approximating it with ordinary (co)homology and then reﬁning the approximation by acting with diﬀerentials. In many circumstances, as it happens for the AHSS, there is more structure. Indeed, the pages can be bi-graded, i.e. Er = ⊕p,qEr p,q with p, q ∈Z, and the diﬀerential dr have a bi-degree (−r, r + 1), hence it maps between the bi-graded page elements as dr : Er p,q → Er p−r,q+r−1. [ ] 9Diﬀerent in this case means either vanishing, or possibly a subgroup of the original page element. 8This assumption is actually suﬃcient for the homological spectral sequence to terminate after a ﬁn number of steps [32]. 9 3.1.1 Homological spectral sequence To get some intuition on the framework we will be working in, consider a ﬁbration F → E →B, the three spaces F, E and B being ﬁber, total space and base respectively.6 The goal is to compute the (generalised) homology of E, denoted Gn(E), of which the cobordism groups Ωξ n(E) are a particular case. As a starting point, one typically needs some knowledge about the (generalised) homology of F or B. Then, to compute Gn(E), one can run the Atiyah-Hirzebruch spectral sequence for homology, that is based on a ﬁltration of Gn(E), i.e. a sequence of subspaces . . . ⊂Fp ⊂Fp+1 ⊂. . . whose union is Gn(E).7 This can be interpreted as an approximate method becoming more and more accurate with each iterative step and stabilising after a ﬁnite number of steps. However, this still does not give directly Gn(E), rather it produces an associated graded module, Gr(Gn(E)), which determines Gn(E) up to an extension problem. In general this has to be solved on a case by case basis, as it requires additional information beyond the AHSS. Let us describe the whole method in more detail below. By deﬁnition, a spectral sequence consists of a sequence of objects Er, called pages, together with endomorphisms dr, called diﬀerentials (since they square to zero), with r non-negative integers. The pairs (Er, dr) are such that the (r + 1)-st page Er+1 is given by the homology of the r-th page Er, Er+1 ∼= H(Er) = ker dr : Er →Er Im dr : Er →Er . (3.1) (3.1) 6In the literature, the case with a trivial ﬁbration is sometimes referred to as the AHSS exclusively. Here, following [32–34, 47], we discuss the more general case of a not necessarily trivial ﬁbration and still refer to it as AHSS. 6In the literature, the case with a trivial ﬁbration is sometimes referred to as the AHSS exclusively. Here, following [32–34, 47], we discuss the more general case of a not necessarily trivial ﬁbration and still refer to it as AHSS. 7In certain cases, it is also possible to use the spectral sequence “backwards” and compute for example Gn(F) from the Gn(E). – 13 – 5 ... ... ... ... ... 4 E2 0,4 E2 1,4 E2 2,4 E2 3,4 E2 4,4 . . . 3 E2 0,3 E2 1,3 E2 2,3 E2 3,3 E2 4,3 . . . p y g p q number of steps [32]. 9Diﬀerent in this case means either vanishing, or possibly a subgroup of the original page element. 3.1.1 Homological spectral sequence (3.2) (3.2) nd B simply connected, we have the Leray-Serre spectral sequence Second, for R a ring and B simply connected, we have the Leray-Serre spectral sequence E2 p,q ∼= Hp(B; R ⊗Hq(F; R)) ⇒Hp+q(E; R) . (3.3) (3.3) Finally, the Leray-Serre-Atiyah-Hirzebruch spectral sequence, or simply Atiyah-Hirzebruch Spectral Sequence is deﬁned for an additive homology theory G∗and a path-connected B Finally, the Leray-Serre-Atiyah-Hirzebruch spectral sequence, or simply Atiyah-Hirzebruch Spectral Sequence is deﬁned for an additive homology theory G∗and a path-connected B E2 p,q ∼= Hp(B; Gq(F)) ⇒Gp+q(E) . (3.4) (3.4) Note that Hp(B; Gq(F)) = 0 for p < 0. This spectral sequence can be used for the computation of cobordism groups. Note that Hp(B; Gq(F)) = 0 for p < 0. This spectral sequence can be used for the computation of cobordism groups. 3.1.1 Homological spectral sequence It is customary to have a pictorial representation of the pages and the relevant diﬀeren- tials, as in the tables 3 and 4 below, where we assume all entries outside the ﬁrst quadrant to vanish.8 Conventionally, the horizontal axis refers to the p-value and the vertical to the q-value of an element of the nth page, En p,q. As for the speciﬁc spectral sequences deﬁned later on, the starting point is usually taken to be the second page, E2. The procedure of acting on a page with the diﬀerential leading to the next page is often referred to as “turning the page”. The only elements that might be diﬀerent9 once we turn the page are those that non-vanishing diﬀerentials act on, while the rest carries – 14 – over intact to the next page. For a sequence conﬁned to the ﬁrst quadrant, such as a homological spectral sequence, after a ﬁnite number of iterations no non-trivial diﬀerential can act anymore. There, the sequence stabilises and we reach the so-called E∞-page. E∞is related to the desired (generalised) homology theory. In particular, one computes the generalised homology groups Gn(E) using all diagonal elements of E∞ p,q, with p + q = n. One says that the spectral sequence converges to Gn(E) and writes E2 p,q ⇒Gp+q. In the simplest case there is just one element on the diagonal of E∞, so a direct identiﬁcation is possible, but usually one has to deal with a non-trivial extension problem, especially when torsion is present. In the general case, one has Gr(Gn(E)) ∼= Ln p=0 E∞ p,n−p and to obtain Gn(E) one needs extra information. We will discuss speciﬁc examples and possible ways around the extension problem in the following sections 3.2 and 3.3. JHEP03(2023)181 Having explained the general idea of a spectral sequence, let us now apply it to our initial problem of computing (generalised) homologies of the total space E for a ﬁbration F →E →B. Depending on what kind of structure one has, one can distinguish three types of spectral sequences. First, let M be an abelian group and B path-connected. The homological Serre spectral sequence is a ﬁrst quadrant spectral sequence deﬁned as E2 p,q ∼= Hp(B; Hq(F; M)) ⇒Hp+q(E; M) . 3.1.2 Cohomological spectral sequence As mentioned, for the computation of the K-theory groups K−n(X) we will employ the cohomological version of the AHSS. Indeed, analogously to the discussion in the previous section one constructs spectral sequences to compute generalised cohomology groups. One starts with a ﬁbration fulﬁlling certain requirements and uses knowledge over cohomological groups of some space (such as ﬁber or base) to deduce what the generalised cohomology of the desired space is (such as the total space). Once again, we have a collection of objects (Er, dr), where now the bi-grading of the diﬀerential is (r, −r+1), i.e. dr : Ep,q r →Ep+r,q−r+1 r and the (r + 1)-st page Er+1 given by the cohomology of the Er page. In the pictorial representation, the pages of a cohomological spectral sequence look very similar to those of a homological one, with the exception that the diﬀerential arrows now point in the opposite direction. Another diﬀerence is that now the sequence possesses a cup product structure which may allow for a formal computation of the diﬀerentials. – 15 – The cohomological Serre spectral sequence is deﬁned similarly to the homological one. For the usual ﬁbration F →E →B, with B path-connected and R a ring, there is a ﬁrst quadrant cohomological spectral sequence of algebras, converging (as a graded algebra) as Ep,q 2 = Hp(B; Hq(F; R)) ⇒Hp+q(E; R) . (3.5) (3.5) π1(B) = 0 and R a ﬁeld, the previous equation simpliﬁes to Ep,q 2 = Hp(B) ⊗Hq(F; R) ⇒Hp+q(E; R) . (3.6) (3.6) Since K-theory is a generalised cohomology theory, a generalisation of the Serre spectral sequence is necessary. This is the Atiyah-Hirzebruch spectral sequence, deﬁned now for G∗a generalised cohomology theory and the ﬁbration as in (3.5). Namely, there is a half-plane cohomological spectral sequence Since K-theory is a generalised cohomology theory, a generalisation of the Serre spectral sequence is necessary. This is the Atiyah-Hirzebruch spectral sequence, deﬁned now for G∗a generalised cohomology theory and the ﬁbration as in (3.5). Namely, there is a half-plane cohomological spectral sequence JHEP03(2023)181 Ep,q 2 = Hp(B; Gq(F)) ⇒Gp+q(E) . (3.7) (3.7) 3.1.3 Trivial ﬁbration and vanishing diﬀerentials on the edge Besides the extension problem, computing the diﬀerentials in a spectral sequence can also be a tedious task. However, there are instances where one can generally show that they vanish. 10The choice of the trivial ﬁbration allows us to avoid a complication we have not discussed yet. Notice that we assumed B to be path connected, but in general not simply connected. When π1(B) ̸= 0, one deals with a system of local coeﬃcients over B with ﬁber Gq(F) [32]. As a consequence, in (3.4) one has to consider ordinary homology with local coeﬃcients. However, if the ﬁbration is trivial this complication can be ignored [32]. 3.1.2 Cohomological spectral sequence This is the case for diﬀerentials from/to the edge of a given page, when the AHSS involves particularly simple ﬁbrations. Consider the trivial ﬁbration pt ,−→X id →X. (3.8) (3.8) he inclusion pt ,−→X is split by the constant map X →pt, implying that The inclusion pt ,−→X is split by the constant map X →pt, implying that Gn(pt) →Gn(X) (3.9) (3.9) is a split injection (G∗being a generalised homology theory). On the other hand, this is also a special case of a map known as the edge homomorphism. Indeed, consider the ﬁbration F →E →B, which generalises (3.8). An edge homomorphism is deﬁned as is a split injection (G∗being a generalised homology theory). On the other hand, this is also a special case of a map known as the edge homomorphism. Indeed, consider the ﬁbration F →E →B, which generalises (3.8). An edge homomorphism is deﬁned as Gn(F) →H0(B; Gn(F)) = E2 0,n →E∞ 0,n →Gn(E), (3.10) (3.10) where the last arrow is an injection while the others are surjections. As stated e.g. in Theorem 9.10 of [32], this is equal to the map where the last arrow is an injection while the others are surjections. As stated e.g. in Theorem 9.10 of [32], this is equal to the map Gn(F) →Gn(E), (3.11) (3.11) induced by the inclusion F ,−→E. For F = pt and B = E = X, one should recover the split injection (3.9) and thus induced by the inclusion F ,−→E. For F = pt and B = E = X, one should recover the split injection (3.9) and thus E2 0,n ∼= E∞ 0,n. (3.12) (3.12) In other words, in this case the entries survive to the ﬁnal page and any diﬀerential acting on them, In other words, in this case the entries survive to the ﬁnal page and any diﬀerential acting on them, dr : Er r,q →Er 0,q+r−1, (3.13) (3.13) has to be zero. This observation greatly simpliﬁes the calculation of the related spectral se- quences and will have a direct application in the upcoming computation of cobordism groups. – 16 – 11As explained in appendix A.1, we denote by e(A, B) the extension of A by B. The opposite convention is also used in the literature, e.g. in [34]. 3.2 Application to cobordism In this section, we employ the homological version of the AHSS to compute cobordism groups Ωξ n(X) for non-trivial k-dimensional spaces X. Considering the trivial ﬁbration10 pt →X →X, the AHSS allows us to determine Ωξ n(X) from the known cobordism groups of the point given in table 1. Then, the second page of the AHSS is given by E2 p,q = Hp(X; Ωξ q) . (3.14) (3.14) To avoid cluttering the expressions, in the remainder of this section we use the shorthand notation Ωξ n(pt) ≡Ωξ n. Note that we will only show the parts of the pages with p, q ≤10, as this is suﬃcient to study the manifolds of interest for physical applications. To avoid cluttering the expressions, in the remainder of this section we use the shorthand notation Ωξ n(pt) ≡Ωξ n. Note that we will only show the parts of the pages with p, q ≤10, as this is suﬃcient to study the manifolds of interest for physical applications. JHEP03(2023)181 3.2.1 Computing Ωξ n(Sk) Before passing to higher-dimensional spheres, we start with the straightforward, yet illustra- tive, computation of Ωξ n(S2). We present here the case where ξ = Spin, while the similarly computed results for ξ = Spinc are relegated to the appendix B. While a direct computation of Hp(S2, ΩSpin q ) is straightforward for low q, in general one turns to the universal coeﬃcient theorem (see appendix A.2), according to which there is a short exact sequence 0 →Hn(S2; Z) ⊗ΩSpin q →Hn(S2; ΩSpin q ) →Tor1(Hn−1(S2; Z), ΩSpin q ) →0. (3.15) (3.15) Recalling the well known homology groups Hn(S2; Z) =    Z for n = 0, 2, 0 otherwise (3.16) (3.16) and the fact that Z is torsion-free, (3.14) can be directly evaluated as E2 p,q = Hp(S2; ΩSpin q ) ∼= Hp(S2; Z) ⊗ΩSpin q =    ΩSpin q for p = 0, 2, 0 otherwise. (3.17) (3.17) Hence, the second page of the AHSS takes the following form. We see that there exist four diﬀerentials that could kill some of the page entries. However, they all end on the ﬁrst column of the page and thus they vanish according to the edge homomorphism reviewed in section 3.1.3. Thus, one can immediately conclude that E2 p,q ∼= E3 p,q. From the third page, no diﬀerentials can act on the page elements, as its degree would be larger than any possible diﬀerence of degree between non-zero elements of the page. 3.2.1 Computing Ωξ n(Sk) • e(Z, Z2): we have Ext1(Z, Z2) = 0 and thus there is only the trivial extension, e(Z, Z2) = Z ⊕Z2. • e(Z, Z2): we have Ext1(Z, Z2) = 0 and thus there is only the trivial extension, e(Z, Z2) = Z ⊕Z2. • e(Z2, Z): we have from (A.8) that Ext1(Z2, Z) = Z2. The two possible extensions are Z and Z2 ⊕Z, so we need some additional input to select the appropriate one. One simple strategy would be to use the splitting lemma (2.7), which tells us that ΩSpin 4 (S2) should contain a factor ΩSpin 4 = Z. Unfortunately such factor is present in both extension options, so we cannot draw any conclusion. In appendix A.5, we show (indirectly) that for Ωξ n(Sk) the extension is always trivial, therefore even in this case e(Z2, Z) = Z ⊕Z2. • e(2Z, 3Z2): we have Ext1(2Z, 3Z2) = 2Ext1(Z, 3Z2) = 5Ext1(Z, Z2) = 0, so the trivial extension must be chosen, in accordance with the general proof of appendix A.5. • e(2Z, 3Z2): we have Ext1(2Z, 3Z2) = 2Ext1(Z, 3Z2) = 5Ext1(Z, Z2) = 0, so the trivial extension must be chosen, in accordance with the general proof of appendix A.5. • e(2Z, 3Z2): we have Ext1(2Z, 3Z2) = 2Ext1(Z, 3Z2) = 5Ext1(Z, Z2) = 0, so the trivial extension must be chosen, in accordance with the general proof of appendix A.5. We summarize our ﬁndings in the following table. 3.2.1 Computing Ωξ n(Sk) Therefore, E2 p,q ∼= E∞ p,q and we arrive at the results in table 6.11 – 17 – 10 ΩSpin 10 0 ΩSpin 10 0 0 0 9 ΩSpin 9 0 ΩSpin 9 0 0 0 8 ΩSpin 8 0 ΩSpin 8 0 0 0 7 ΩSpin 7 0 ΩSpin 7 0 0 0 6 ΩSpin 6 0 ΩSpin 6 0 0 0 5 ΩSpin 5 0 ΩSpin 5 0 0 0 4 ΩSpin 4 0 ΩSpin 4 0 0 0 3 ΩSpin 3 0 ΩSpin 3 0 0 0 2 ΩSpin 2 0 ΩSpin 2 0 0 0 1 ΩSpin 1 0 ΩSpin 1 0 0 0 0 ΩSpin 0 0 ΩSpin 0 0 0 0 0 1 2 3 4 5 = 10 3Z2 0 3Z2 0 0 0 9 2Z2 0 2Z2 0 0 0 8 2Z 0 2Z 0 0 0 7 0 0 0 0 0 0 6 0 0 0 0 0 0 5 0 0 0 0 0 0 4 Z 0 Z 0 0 0 3 0 0 0 0 0 0 2 Z2 0 Z2 0 0 0 1 Z2 0 Z2 0 0 0 0 Z 0 Z 0 0 0 0 1 2 3 4 5 Table 5. Second (and ﬁnal) page of AHSS for ΩSpin n (S2). n 0 1 2 3 4 5 6 7 8 9 10 ΩSpin n (S2) Z Z2 e(Z, Z2) Z2 e(Z2, Z) 0 Z 0 2Z 2Z2 e(2Z,3Z2) Table 6. Cobordism groups ΩSpin n (S2), n = 0, . . . , 10, up to extensions. n 0 1 2 3 4 5 6 7 8 9 10 ΩSpin n (S2) Z Z2 Z ⊕Z2 Z2 Z2 ⊕Z 0 Z 0 2Z 2Z2 2Z ⊕3Z2 Table 7. Cobordism groups ΩSpin n (S2). JHEP03(2023)181 Table 5. Second (and ﬁnal) page of AHSS for ΩSpin n (S2). Table 5. Second (and ﬁnal) page of AHSS for ΩSpin n (S2). Table 6. Cobordism groups ΩSpin n (S2), n = 0, . . . , 10, up to extensions. Table 6. Cobordism groups ΩSpin n (S2), n = 0, . . . , 10, up to extensions. Let us now tackle the extension problems one by one. Our main tools are brieﬂy reviewed in appendix A.1. We summarize our ﬁndings in the following table. The calculation of ΩSpin n (Sk) for higher k proceeds similarly. Since the only non- vanishing homology classes are H0(Sk; Z) = Hk(Sk; Z) = Z and the universal coeﬃcient – 18 – theorem applies, the second page for the trivial ﬁbration pt →Sk →Sk looks very similar to the one for S2, with the non-vanishing entries along the p = 0, k columns. The only possibly non-vanishing diﬀerentials are dk, but since they end on the ﬁrst column they vanish due to the edge homomorphism. Hence, the computation proceeds exactly as before. For S1 the computation is even simpler, since for degree reasons no diﬀerential can act. As explained at the beginning of the present section, the fact that π1(S1) ̸= 0 does not concern us since we are using a trivial ﬁbration. For the computation of the Spinc cobordism groups ΩSpinc n (Sk) one follows similar steps. Now the second page is JHEP03(2023)181 JHEP03(2023)181 E2 p,q = Hp(Sk; ΩSpinc q ) ∼= Hp(Sk; Z) ⊗ΩSpinc q =    ΩSpinc q for p = 0, k, 0 otherwise, (3.18) (3.18) and the same arguments as for the ΩSpin n (Sk) computation still go through. As proven in appendix A.5, for both structures ξ = Spin, Spinc the ﬁnal result can be compactly written as Ωξ n(Sk) = Ωξ n(pt) ⊕Ωξ n−k(pt) . (3.19) (3.19) Explicitly, the groups for n, k ≥10 are given in the appendix B. Explicitly, the groups for n, k ≥10 are given in the appendix B. 2.2 Computing Ωξ n(T 2) 3.2.2 Computing Ωξ n(T 2) For the two-torus, T 2 = S1 × S1, we present the computation for both ξ = Spin and ξ = Spinc in parallel. Starting from the known homology groups (recall the Betti numbers of the torus b0 = b2 = 1, b1 = 2) Hn(T 2; Z) =          Z for n = 0, 2, 2Z for n = 1, 0 otherwise, (3.20) (3.20) and using the universal coeﬃcient theorem again (with vanishing Tor1 group), one can compute the second page and using the universal coeﬃcient theorem again (with vanishing Tor1 group), one can compute the second page E2 p,q = Hp(T 2; Ωξ q) ∼= Hp(T 2; Z) ⊗Ωξ q =          Ωξ q for p = 0, 2, 2Ωξ q for p = 1, 0 otherwise . (3.21) (3.21) The second pages for the two structures ξ = Spin, Spinc are shown in table 8. For the Spin case we have four diﬀerentials which could be non-trivial, but they vanish due to the edge homomorphism for the trivial ﬁbration. For the Spinc case, no diﬀerential can act for degree reasons. Hence, the second pages above are in fact the ﬁnal pages and we have the results displayed in table 9, where we used the notation e(A, B, C) = e(A, e(B, C)). Two facts are crucial to solve the extension problem for these cobordism groups. First, the extensions of all free abelian groups are trivial. Second, e(mZ, nZk) = mZ ⊕nZk since Ext1(mZ, nZk) = 0. 3.2.2 Computing Ωξ n(T 2) However, since Ext1(Z2, Z2) = Z2, we cannot conclude anything about – 19 – 10 3Z2 6Z2 3Z2 0 0 0 9 2Z2 4Z2 2Z2 0 0 0 8 2Z 4Z 2Z 0 0 0 7 0 0 0 0 0 0 6 0 0 0 0 0 0 5 0 0 0 0 0 0 4 Z 2Z Z 0 0 0 3 0 0 0 0 0 0 2 Z2 2Z2 Z2 0 0 0 1 Z2 2Z2 Z2 0 0 0 0 Z 2Z Z 0 0 0 0 1 2 3 4 5 10 4Z ⊕Z2 8Z ⊕2Z2 4Z ⊕Z2 0 0 9 0 0 0 0 0 8 4Z 8Z 4Z 0 0 7 0 0 0 0 0 6 2Z 4Z 2Z 0 0 5 0 0 0 0 0 4 2Z 4Z 2Z 0 0 3 0 0 0 0 0 2 Z 2Z Z 0 0 1 0 0 0 0 0 0 Z 2Z Z 0 0 0 1 2 3 4 Table 8. Second (and ﬁnal) pages of AHSS for ΩSpin n (T 2) (left) and ΩSpinc n (T 2) (right). JHEP03(2023)181 Table 8. Second (and ﬁnal) pages of AHSS for ΩSpin n (T 2) (left) and ΩSpinc n (T 2) (right). n 0 1 2 3 4 ΩSpin n (T 2) Z e(2Z, Z2) e(Z, 2Z2, Z2) e(Z2, 2Z2) e(Z2, Z) ΩSpinc n (T 2) Z 2Z e(Z, Z) 2Z e(Z, 2Z) n 5 6 7 8 9 10 ΩSpin n (T 2) 2Z Z 0 2Z e(4Z, 2Z2) e(2Z, 4Z2, 3Z2) ΩSpinc n (T 2) 4Z e(2Z, 2Z) 4Z e(2Z, 4Z) 8Z e(4Z, 4Z ⊕Z2) Table 9. Cobordism groups ΩSpin n (T 2) and ΩSpinc n (T 2), n = 0, . . . , 10, up to extensions. n 0 1 2 3 4 ΩSpin n (T 2) Z e(2Z, Z2) e(Z, 2Z2, Z2) e(Z2, 2Z2) e(Z2, Z) ΩSpinc n (T 2) Z 2Z e(Z, Z) 2Z e(Z, 2Z) n 5 6 7 8 9 10 ΩSpin n (T 2) 2Z Z 0 2Z e(4Z, 2Z2) e(2Z, 4Z2, 3Z2) ΩSpinc n (T 2) 4Z e(2Z, 2Z) 4Z e(2Z, 4Z) 8Z e(4Z, 4Z ⊕Z2) Table 9. Cobordism groups ΩSpin n (T 2) and ΩSpinc n (T 2), n = 0, . . . , 10, up to extensions. Table 9. 3.2.2 Computing Ωξ n(T 2) Cobordism groups ΩSpin n (T 2) and ΩSpinc n (T 2), n = 0, . . . , 10, up to extensions. n 0 1 2 3 4 ΩSpin n (T 2) Z 2Z ⊕Z2 e(Z, 2Z2, Z2) e(Z2, 2Z2) e(Z2, Z) ΩSpinc n (T 2) Z 2Z 2Z 2Z 3Z n 5 6 7 8 9 10 ΩSpin n (T 2) 2Z Z 0 2Z 4Z ⊕2Z2 e(2Z, 4Z2, 3Z2) ΩSpinc n (T 2) 4Z 4Z 4Z 6Z 8Z 8Z ⊕Z2 Table 10. Cobordism groups ΩSpin n (T 2) and ΩSpinc n (T 2), n = 0, . . . , 10. n 0 1 2 3 4 ΩSpin n (T 2) Z 2Z ⊕Z2 e(Z, 2Z2, Z2) e(Z2, 2Z2) e(Z2, Z) ΩSpinc n (T 2) Z 2Z 2Z 2Z 3Z n 5 6 7 8 9 10 ΩSpin n (T 2) 2Z Z 0 2Z 4Z ⊕2Z2 e(2Z, 4Z2, 3Z2) ΩSpinc n (T 2) 4Z 4Z 4Z 6Z 8Z 8Z ⊕Z2 Table 10. Cobordism groups ΩSpin n (T 2) and ΩSpinc n (T 2), n = 0, . . . , 10. Table 10. Cobordism groups ΩSpin n (T 2) and ΩSpinc n (T 2), n = 0, . . . , 10. – 20 – 10 4Z ⊕Z2 0 88Z ⊕22Z2 0 4Z ⊕Z2 0 0 0 9 0 0 0 0 0 0 0 0 8 4Z 0 88Z 0 4Z 0 0 0 7 0 0 0 0 0 0 0 0 6 2Z 0 44Z 0 2Z 0 0 0 5 0 0 0 0 0 0 0 0 4 2Z 0 44Z 0 2Z 0 0 0 3 0 0 0 0 0 0 0 0 2 Z 0 22Z 0 Z 0 0 0 1 0 0 0 0 0 0 0 0 0 Z 0 22Z 0 Z 0 0 0 0 1 2 3 4 5 6 7 Table 11. Second (and ﬁnal) page of the AHSS for the computation of ΩSpinc n (K3). JHEP03(2023)181 Table 11. Second (and ﬁnal) page of the AHSS for the computation of ΩSpinc n (K3). able 11. Second (and ﬁnal) page of the AHSS for the computation of ΩSpinc n (K3). e(Z2, Z2), which is either 2Z2 or Z4. A similar story applies for e(Z2, Z). Up to this point, our results are shown in table 10. 3.2.2 Computing Ωξ n(T 2) According to the general proof given in appendix A.5, the remaining extension problems should be trivial. Indeed, there we generically show that the cobordism groups of k-dimensional tori have a simple decomposition, e(Z2, Z2), which is either 2Z2 or Z4. A similar story applies for e(Z2, Z). Up to this point, our results are shown in table 10. According to the general proof given in appendix A.5, the remaining extension problems should be trivial. Indeed, there we generically show that the cobordism groups of k-dimensional tori have a simple decomposition, Ωξ n(T k) = k M m=0 k m ! Ωξ n−m(pt), (3.22) (3.22) for a generic structure ξ, which can be taken to be Spin or Spinc. The binomial coeﬃcient can be interpreted as the number of m-cycles on T k. Explicit results with all extensions solved are reported in appendix B. for a generic structure ξ, which can be taken to be Spin or Spinc. The binomial coeﬃcient can be interpreted as the number of m-cycles on T k. Explicit results with all extensions solved are reported in appendix B. 3.2.3 Computing ΩSpinc n (K3) 3.2.4 Computing ΩSpinc n (CY3) The computation for the cobordism groups of a Calabi-Yau threefold are obtained similarly to those of K3. We start from the known result12 Hn(CY3; Z) =                Z for n = 0, 6, b2 Z for n = 2, 4, b3 Z for n = 3, 0 otherwise, (3.26) (3.26) where bp are the CY3 Betti numbers (recall that bp = b6−p). The second page is then given by where bp are the CY3 Betti numbers (recall that bp = b6−p). The second page is then given by given by E2 p,q = Hp(CY3; ΩSpinc q ) ∼= Hp(CY3; Z) ⊗ΩSpinc q =                ΩSpinc q for p = 0, 6, b2 ΩSpinc q for p = 2, 4, b3 ΩSpinc q for p = 3, 0 otherwise (3.27) (3.27) and shown explicitly in table 13. One realises that this time ﬁve non-vanishing columns E2 p,q exist in the second page, the elements of which are given by bpΩSpinc q . None of the diﬀerentials dr with even r can act for degree reasons. However, there are two kinds of third diﬀerentials that can be non-trivial. The ﬁrst class is and shown explicitly in table 13. One realises that this time ﬁve non-vanishing columns E2 p,q exist in the second page, the elements of which are given by bpΩSpinc q . p,q p q None of the diﬀerentials dr with even r can act for degree reasons. However, there are two kinds of third diﬀerentials that can be non-trivial. The ﬁrst class is d3 : E3 3,q →E3 0,q+2 , (3.28) (3.28) 12By assumption, the Calabi-Yau threefolds we consider in this work are such that π1(CY3) = 0. In general, there exist Calabi-Yau manifolds with π1(CY3) = Zn, for some integer n, i.e. with torsion in H1(CY3; Z). Typical examples are free quotient of Calabi-Yaus without torsion, such as the free quotient of the quintic P4[5]/Z5. They have been investigated, especially in a K-theory context, for instance in [36, 37]. For Calabi-Yau twofolds, one has π1(K3) = 0. Taking a free quotient by Zn reduces the Euler number to χ/24n, so that the quotient manifold is not K3 anymore. 3.2.3 Computing ΩSpinc n (K3) For the determination of the cobordism groups of K3 we again start with the known result for Hn(K3; Z). For the determination of the cobordism groups of K3 we again start with the known result for Hn(K3; Z). Hn(K3; Z) =          Z for n = 0, 4, 22Z for n = 2, 0 otherwise, (3.23) (3.23) where the non-vanishing Betti numbers of K3 are b0 = b4 = 1, b2 = 22. Once again using the trivial ﬁbration and the universal coeﬃcient theorem we compute the second page entries shown in table 11. For Spinc all diﬀerentials are trivial for degree reason, so that we can conclude E2 p,q = E∞ p,q with E2 p,q = Hp(K3; ΩSpinc q ) ∼= Hp(K3; Z) ⊗ΩSpinc q =          ΩSpinc q for p = 0, 4, 22 ΩSpinc q for p = 2, 0 otherwise . (3.24) (3.24) – 21 – – 21 – n 0 1 2 3 4 5 6 7 8 9 10 ΩSpinc n (K3) Z 0 23Z 0 25Z 0 47Z 0 50Z 0 94Z ⊕Z2 Table 12. Cobordism groups ΩSpinc n (K3), n = 0, . . . , 10. Table 12. Cobordism groups ΩSpinc n (K3), n = 0, . . . , 10. Table 12. Cobordism groups ΩSpinc n (K3), n = 0, . . . , 10. Up to n = 10 all extension problems are trivial, so that we can express the ﬁnal result as Up to n = 10 all extension problems are trivial, so that we can express the ﬁnal result a ΩSpinc n (K3) = ΩSpinc n (pt) ⊕˜ΩSpinc n (K3) = ΩSpinc n (pt) ⊕22 ΩSpinc n−2 (pt) ⊕ΩSpinc n−4 (pt) . (3.25) (3.25) JHEP03(2023)181 In this formula, it is understood that cobordism groups with negative index are set to zero. The explicit groups resulting from the formula above are reported in table 12. In this formula, it is understood that cobordism groups with negative index are set to zero. The explicit groups resulting from the formula above are reported in table 12. 2.4 Computing ΩSpinc n (CY3) 3.2.4 Computing ΩSpinc n (CY3) 13This diﬀerential is given by the homological dual of the cohomology operation Sq3 Z, the (integral) third Steenrod square (the operations Sqi are introduced brieﬂy later on; see also the appendix A.3). Interestingly, its triviality is the homological dual statement of the Freed-Witten anomaly cancellation [35, 38], which we are going to discuss later on in the K-theory calculations. 3.2.4 Computing ΩSpinc n (CY3) They eventually vanish for q ≤6. n 0 1 2 3 4 5 ΩSpinc n (CY3) Z 0 (b2 + 1)Z b3Z (2 + 2b2)Z b3Z n 6 7 8 9 10 ΩSpinc n (CY3) (3 + 3b2)Z 2b3Z (5 + 4b2)Z 2b3Z (6 + 6b2)Z ⊕Z2 Table 14. Cobordism groups ΩSpinc n (CY3), n = 0, . . . , 10. Table 14. Cobordism groups ΩSpinc n (CY3), n = 0, . . . , 10. which vanish due to the edge homomorphism (see section 3.1.3). The second class acts as d3 : E3 6,q →E3 3,q+2 , (3.29) (3.29) which is in principle non-vanishing.13 That this diﬀerential is trivial up to q = 6, too, follows from Lemma 3.1 of [49]. We thus get the results in table 14. which is in principle non-vanishing.13 That this diﬀerential is trivial up to q = 6, too, follows from Lemma 3.1 of [49]. We thus get the results in table 14. 3.2.4 Computing ΩSpinc n (CY3) 12By assumption, the Calabi-Yau threefolds we consider in this work are such that π1(CY3) = 0. In general, there exist Calabi-Yau manifolds with π1(CY3) = Zn, for some integer n, i.e. with torsion in H1(CY3; Z). Typical examples are free quotient of Calabi-Yaus without torsion, such as the free quotient of the quintic P4[5]/Z5. They have been investigated, especially in a K-theory context, for instance in [36, 37]. For Calabi-Yau twofolds, one has π1(K3) = 0. Taking a free quotient by Zn reduces the Euler number to χ/24n, so that the quotient manifold is not K3 anymore. – 22 – 10 4Z ⊕Z2 0 b2(4Z ⊕Z2) b3(4Z ⊕Z2) b2(4Z ⊕Z2) 0 4Z ⊕Z2 0 9 0 0 0 0 0 0 0 0 8 4Z 0 4b2Z 4b3Z 4b2Z 0 4Z 0 7 0 0 0 0 0 0 0 0 6 2Z 0 2b2Z 2b3Z 2b2Z 0 2Z 0 5 0 0 0 0 0 0 0 0 4 2Z 0 2b2Z 2b3Z 2b2Z 0 2Z 0 3 0 0 0 0 0 0 0 0 2 Z 0 b2Z b3Z b2Z 0 Z 0 1 0 0 0 0 0 0 0 0 0 Z 0 b2Z b3Z b2Z 0 Z 0 0 1 2 3 4 5 6 7 10 4Z ⊕Z2 0 b2(4Z ⊕Z2) b3(4Z ⊕Z2) b2(4Z ⊕Z2) 0 4Z ⊕Z2 0 9 0 0 0 0 0 0 0 0 8 4Z 0 4b2Z 4b3Z 4b2Z 0 4Z 0 7 0 0 0 0 0 0 0 0 6 2Z 0 2b2Z 2b3Z 2b2Z 0 2Z 0 5 0 0 0 0 0 0 0 0 4 2Z 0 2b2Z 2b3Z 2b2Z 0 2Z 0 3 0 0 0 0 0 0 0 0 2 Z 0 b2Z b3Z b2Z 0 Z 0 1 0 0 0 0 0 0 0 0 0 Z 0 b2Z b3Z b2Z 0 Z 0 0 1 2 3 4 5 6 7 JHEP03(2023)181 Table 13. Second (and ﬁnal) page of the AHSS for the computation of ΩSpinc n (CY3). One of the possibly non-vanishing diﬀerentials d3 : E3 6,q →E3 3,q+2 is displayed (for q = 0). They eventually vanish for q ≤6. Table 13. Second (and ﬁnal) page of the AHSS for the computation of ΩSpinc n (CY3). One of the possibly non-vanishing diﬀerentials d3 : E3 6,q →E3 3,q+2 is displayed (for q = 0). 3.3 Application to K-theory Next, we perform similar computations for the K- and KO-theory groups on spheres, tori and Calabi-Yau manifolds. For this purpose we employ the cohomological version of the AHSS. Real K-theory turns out to be more involved, but we report some results in sections 3.3.6 and 3.3.7. – 23 – 6 Z 0 0 Z 0 5 0 0 0 0 0 4 Z 0 0 Z 0 3 0 0 0 0 0 2 Z 0 0 Z 0 1 0 0 0 0 0 0 Z 0 0 Z 0 -1 0 0 0 0 0 -2 Z 0 0 Z 0 -3 0 0 0 0 0 -4 Z 0 0 Z 0 -5 0 0 0 0 0 -6 Z 0 0 Z 0 JHEP03(2023)181 JHEP03(2023)181 Table 15. Second (and ﬁnal) page of the AHSS for the computation of K−n(S3). One of the d3 diﬀerentials is shown explicitly. They all eventually vanish. 3.1 Computing K−n(Sk) 3.3.1 Computing K−n(Sk) (3.34) (3.34) Explicitly, it is given by the composition Explicitly, it is given by the composition d3 = Sq3 Z = β ◦Sq2 ◦ρ, (3.35) (3.35) where ρ is the reduction modulo 2 and β the Bockstein homomorphism, namely Sq3 Z : Hn(X; Z) ρ −→Hn(X; Z2) Sq2 −→Hn+2(X; Z2) β −→Hn+3(X; Z). (3.36) (3.36) JHEP03(2023)181 We refer the reader to the appendix A.3 for a more precise deﬁnition of Steenrod squares and of the Bockstein homomorphism, together with a short summary of their main properties. Fortunately, since no torsion is involved, according to Theorem 4.8 of [48] all diﬀerentials (including d3) vanish. This fact will be used systematically in the other computations of K−n(X) groups below.14 Moreover, the extension problem is always trivial, since only free abelian groups are present. Thus, for every odd value of k we recover K−n(S2k+1) = Z. The situation for even k is simpler as for degree reasons no diﬀerentials can act, so that Ep,q 2 = Ep,q ∞. We recover then K−2n−1(S2k) = 0 and K−2n(S2k) = 2Z. Notice that the ﬁnal result can be expressed as K−n(Sk) = K−n(pt) ⊕K−k−n(pt) . (3.37) (3.37) 14This is a consequence of the Chern isomorphism 3.3.1 Computing K−n(Sk) The K-theory groups of spheres Sk are known to be [33] The K-theory groups of spheres Sk are known to be [33] K−n(Sk) =          Z for k odd, 2Z for n, k even, 0 otherwise, (3.30) (3.30) but it is instructive to reproduce these results using the cohomological AHSS (3.7). As usual, we use the trivial ﬁbration pt →Sk →Sk and we do not have to worry about local coeﬃcients. Recalling that but it is instructive to reproduce these results using the cohomological AHSS (3.7). As usual, we use the trivial ﬁbration pt →Sk →Sk and we do not have to worry about local coeﬃcients. Recalling that K−n(pt) =    Z for n even, 0 otherwise, (3.31) (3.31) we have the second page Ep,q 2 = Hp(Sk; Kq(pt)) =    Z, for q even, p = 0, k, 0, otherwise . (3.32) (3.32) Note that it is essential to include the bottom quadrant (with q < 0) to arrive at reasonable results. Limiting our spectral sequence to the ﬁrst quadrant only, as in the homological case, is not consistent as it would violate Bott periodicity. Note that it is essential to include the bottom quadrant (with q < 0) to arrive at reasonable results. Limiting our spectral sequence to the ﬁrst quadrant only, as in the homological case, is not consistent as it would violate Bott periodicity. For concreteness, let us consider X = S3. We are interested in the groups K−n(X), with n > 0, so the relevant page elements lie on the p + q = −n bands of the ﬁnal page, which now intersect the axes only once. The d2 diﬀerential vanish so that Ep,q 3 = Ep,q 2 , but d3 may act non-trivially d3 : E0,q 3 →E3,q−2 3 , q even . (3.33) (3.33) – 24 – This diﬀerential was found by Atiyah and Hirzebruch [50] to be an instance of a cohomological operation known as (integral) Steenrod square (Sqi Z) This diﬀerential was found by Atiyah and Hirzebruch [50] to be an instance of a cohomological operation known as (integral) Steenrod square (Sqi Z) Sq3 Z : Hn(X; Z) →Hn+3(X; Z). 3.3.2 Comment on Freed-Witten anomalies Let us comment more on the role of d3 = Sq3 Z and on its physical consequences, beyond the computation of K−n(Sk). From [51], it is known that type II D-branes (in absence of B ﬁeld) must wrap a Spinc manifold Y , otherwise there is a global Freed-Witten anomaly. Given an element y ∈Hn(X; Z), one has (see appendix A.3) Sq3 Z(y) = W3(N) ∪y, (3.38) (3.38) where N is the normal bundle of the codimension n submanifold Poincaré dual to y, which we call Y below, while ∪is the cup product. Since W3(N) = 0, iﬀY is Spinc,15 one can relate a trivial action of d3 in the AHSS to the absence of Freed-Witten anomalies for a D-brane wrapping Y [35, 38]. Indeed, if E4 = ker d3/Im d3 is given in terms of the groups Hn(X; Z) without further restrictions, all cohomology classes (and their dual cycles) survive. Otherwise, either some are removed when passing from cohomology to K-theory or they change to a torsion group [38]. Physically, they would correspond to D-branes which are anomalous or unstable. 14This is a consequence of the Chern isomorphism K0(X) ⊗Z R ∼= M n H2n(X; R), K−1(X) ⊗Z R ∼= M n H2n+1(X; R), which implies that if there is no torsion in cohomology, the AHSS for K-theory terminates already at the second page. 15 15The obstruction to Spinc structure on Y is really W3(Y ) = β(w2(Y )). However, since in our case X is Spin and Y is oriented by assumption (in type II), one can show that w2(N) = w2(Y ), implying W3(N) = W3(Y ) [24, 51]. – 25 – 3.3.3 Computing K−n(T k) Next we consider the k-dimensional torus T k = (S1)k. To proceed, one can either compute the groups by using the AHSS in a similar manner as done for the sphere (extending also the page to include the fourth quadrant) or use the known results for the reduced K-theory groups f K−n(T k) and the decomposition (2.15). Starting with the second approach, we observe that according to [33] we have f K−n(T k) =    2k−1Z for n odd, (2k−1 −1)Z for n even. (3.39) (3.39) JHEP03(2023)181 Since K−2n(pt) = Z and K−2n−1(pt) = 0, it follows that K−n(T k) = 2k−1Z, (3.40) (3.40) for n any integer. For the trivial case k = 1, where the torus is just a circle, the above result coincides with the expected one from the sphere computation, i.e. K−n(T 1) = Z. Let us also comment on the calculation of K−n(T k) using the spectral sequence. One for n any integer. For the trivial case k = 1, where the torus is just a circle, the above result coincides with the expected one from the sphere computation, i.e. K−n(T 1) = Z. Let us also comment on the calculation of K−n(T k) using the spectral sequence. One has the second page Ep,q 2 = Hp(T k; Kq(pt)) . (3.41) (3.41) The computation using the AHSS for the trivial ﬁbration gives the same result (3.40), upon realising that once again all diﬀerentials vanish since there is no torsion, so Ep,q 2 = Ep,q ∞, and the extension problem is trivial. We note that the ﬁnal result can be elegantly written as K−n(T k) = k M m=0 k m ! K−m−n(pt) , (3.42) (3.42) where the binomial coeﬃcient can be interpreted as the number of m-cycles on T k. 3.3.4 Computing K−n(K3) 3.3.4 Computing K−n(K3) The AHSS also allows to straightforwardly compute the K-theory groups on K3. The second page of the sequence is given by Ep,q 2 = Hp(K3; Kq(pt)) =          Z for p = 0, 4, q even, 22 Z for p = 2, q even, 0 otherwise. (3.43) (3.43) This is explicitly shown in table 16. It is evident that no diﬀerentials can act non-trivially on the second page for degree reasons so that the sequence promptly terminates. Thus, the ﬁnal result reads  This is explicitly shown in table 16. It is evident that no diﬀerentials can act non-trivially on the second page for degree reasons so that the sequence promptly terminates. Thus, the ﬁnal result reads  K−n(K3) =    0 for n odd, 24Z for n even. (3.44) (3.44) Note that the factor 24 arises as b0 + b2 + b4 = 1 + 22 + 1 with bm being the Betti numbers of K3. Therefore, we can also express the K-theory groups on K3 as Note that the factor 24 arises as b0 + b2 + b4 = 1 + 22 + 1 with bm being the Betti numbers of K3. Therefore, we can also express the K-theory groups on K3 as K−n(K3) = 4 M m=0 b4−m(K3) K−m−n(pt) . (3.45) (3.45) – 26 – – 26 – 6 Z 0 22Z 0 Z 5 0 0 0 0 0 4 Z 0 22Z 0 Z 3 0 0 0 0 0 2 Z 0 22Z 0 Z 1 0 0 0 0 0 0 Z 0 22Z 0 Z -1 0 0 0 0 0 -2 Z 0 22Z 0 Z -3 0 0 0 0 0 -4 Z 0 22Z 0 Z -5 0 0 0 0 0 -6 Z 0 22Z 0 0Z Table 16. Second (and ﬁnal) page of the AHSS for the computation of K−n(K3). JHEP03(2023)181 Table 16. Second (and ﬁnal) page of the AHSS for the computation of K−n(K3). 3.3.4 Computing K−n(K3) 6 Z 0 b2Z b3Z b2Z 0 Z 5 0 0 0 0 0 0 0 4 Z 0 b2Z b3Z b2Z 0 Z 3 0 0 0 0 0 0 0 2 Z 0 b2Z b3Z b2Z 0 Z 1 0 0 0 0 0 0 0 0 Z 0 b2Z b3Z b2Z 0 Z -1 0 0 0 0 0 0 0 -2 Z 0 b2Z b3Z b2Z 0 Z -3 0 0 0 0 0 0 0 -4 Z 0 b2Z b3Z b2Z 0 Z -5 0 0 0 0 0 0 0 -6 Z 0 b2Z b3Z b2Z 0 Z Table 17. Second (and ﬁnal) page of AHSS for computation of K−n(CY3). 3.3.6 KO-groups of spheres and tori The KO groups can similarly be computed using the AHSS. However, in this case there is torsion, so the diﬀerentials can be non-vanishing. For spheres Sk, one can use the splitting lemma and determine the relevant groups as KO−n(Sk) = g KO(Sn+k) ⊕g KO(Sn) = KO−n−k(pt) ⊕KO−n(pt) . (3.48) (3.48) JHEP03(2023)181 The full results for KO−n(Sk) for n, k ≤10 are provided in appendix C. For tori, it was shown in [53] that KO−n(T k) = k M m=0 k m ! KO−m−n(pt) . (3.49) (3.49) 3.3.5 Computing K−n(CY3) The computation of K−n(CY3) is similar to that of K3. Omitting unnecessary details, we present directly the second page in table 17. The only possibly non-vanishing diﬀerential is d3 : E1,q 3 →E4,q−2 3 . However, due to lack of torsion it is in fact vanishing and, given also that the extension problem is trivial, we conclude that K−n(CY3) =    b3 Z if n odd, (2 + 2b2) Z if n even. (3.46) (3.46) Notice the factor (2 + 2b2) arises as b0 + b2 + b4 + b6, with b0 = b6 = 1 and b2 = b4, bp being the Betti numbers of the CY3. The result can also be found in Corollary 1.9 of [52]. Again, we can elegantly express the K-theory groups on (simply connected) Calabi-Yau threefolds Notice the factor (2 + 2b2) arises as b0 + b2 + b4 + b6, with b0 = b6 = 1 and b2 = b4, bp being the Betti numbers of the CY3. The result can also be found in Corollary 1.9 of [52]. Again, we can elegantly express the K-theory groups on (simply connected) Calabi-Yau threefolds – 27 – as K−n(CY3) = 6 M m=0 b6−m(CY3) K−m−n(pt) . (3.47) (3.47) 3.3.7 Computing KO−n(K3) For real K-theory, computations involving higher dimensional manifolds with a richer topology than the torus or the sphere can become potentially more complicated, due to more involved diﬀerentials and extension problems. Indeed, for Calabi-Yau threefolds the computation turned out to be fairly subtle, so that we postpone it to future work. However, in the case of K3, as we will show now, all diﬀerentials are vanishing and the computations can be performed, up to extensions. The second page of the spectral sequence is the following: Ep,q 2 = Hp(K3; KOq(pt)) =          KOq(pt) for p = 0, 4, 22 KOq(pt) for p = 2, 0 otherwise. (3.50) (3.50) We realise that the diﬀerentials d2 and d4 (depending on how d2 acts) can possibly be non-vanishing. At second degree, we have We realise that the diﬀerentials d2 and d4 (depending on how d2 acts) can possibly be non-vanishing. At second degree, we have d2 : Ep,q 2 →Ep+2,q−1 2 , (3.51) (3.51) for p = 0, 2 and q = 0, −1, together with all of its periodic copies. The explicit form of this diﬀerential is known to be [54, 55] for p = 0, 2 and q = 0, −1, together with all of its periodic copies. The explicit form of this diﬀerential is known to be [54, 55] d2 =    Sq2ρ : Hp(K3; KO0(pt)) →Hp+2(K3; KO−1(pt)) , Sq2 : Hp(K3; KO−1(pt)) →Hp+2(K3; KO−2(pt)) (3.52) (3.52) corresponding to q = 0, −1 respectively. Here, Sq2 : Hp(X; Z2) →Hp+2(X; Z2) is the second Steenrod square and ρ is the reduction modulo 2. We argue now that d2 is vanishing for X = K3. We discuss the case q = −1, but the analysis can be extended to q = 0 in a similar way. For any element y ∈Hp(X; Z2), we can represent Sq2(y) = ι∗(w2(N))∪y [56]. 3.3.7 Computing KO−n(K3) Here, N is the normal bundle of the submanifold Y ⊂X Poincaré dual to y and ι∗: Hp(Y ) →Hp(X) – 28 – 7 Z2 0 22Z2 0 Z2 6 Z2 0 22Z2 0 Z2 5 0 0 0 0 0 4 Z 0 22Z 0 Z 3 0 0 0 0 0 2 0 0 0 0 0 1 0 0 0 0 0 0 Z 0 22Z 0 Z -1 Z2 0 22Z2 0 Z2 -2 Z2 0 22Z2 0 Z2 -3 0 0 0 0 0 -4 Z 0 22Z 0 Z -5 0 0 0 0 0 -6 0 0 0 0 0 -7 0 0 0 0 0 Table 18. Second (and ﬁnal) page of the AHSS for the computation of KO−n(K3). JHEP03(2023)181 Table 18. Second (and ﬁnal) page of the AHSS for the computation of KO−n(K3). Table 18. Second (and ﬁnal) page of the AHSS for the computation of KO−n(K3). the cohomological push-forward. For p = 0, the diﬀerential d2 vanishes since y is dual to the whole four dimensional manifold X = K3 which is Spin, thus w1(N) = w2(N) = 0. Alternatively, it vanishes since Sq2(y) = 0 for y ∈H0(X; Z2) (see the properties of Sqi listed in appendix A.3). For p = 2, the diﬀerential vanishes as well, since from the condition w2(X) = w1(X) = 0 (i.e. X is Spin), one can then prove w2(N) = 0 for a two dimensional manifold Y not necessarily orientable [35]. Alternatively, for p = 2 we can also write Sq2(y) = ν2 ∪y (see equation (A.24)) and then the second Wu class, ν2 = w2(X) + w1(X)2, vanishes since X = K3 is Spin. Thus, d2 is trivial. At degree four, we have the diﬀerential At degree four, we have the diﬀerential d4 : E0,−1 4 →E4,−4 4 . (3.53) (3.53) However, since there cannot be non-trivial homomorphisms16 Zk →Z for k ≥2 also this diﬀerential must vanish and Ep,q 2 ∼= Ep,q ∞. Thus, one can read oﬀthe KO−n(K3) groups, which we present in table 19 up to extensions. Note that we have already made use of the splitting lemma (2.15) to simplify the results. 4 Physical interpretation In this section, we show that the cobordism and K-theory groups of X previously calculated with the AHSS can be interpreted in terms of the dimensional reduction of global symmetries, 16This can be seen directly as follows. Consider the case φ : Z2 →Z, the generalisation to k > 2 being straightforward. φ cannot be a non-trivial homomorphism since, choosing φ(0) = 0 and φ(1) = 1, one is lead to the contradiction 0 = φ(0) = φ(2) = φ(1) + φ(1) = 2. Thus, the only option is to set also φ(1) = 0 and φ is trivial. – 29 – n 0 1 2 3 KO−n(K3) Z ⊕e(22Z2, Z) Z2 Z2 ⊕22Z 0 n 4 5 6 7 KO−n(K3) 2Z Z2 Z2 ⊕22Z 22Z2 Table 19. KO-groups KO−n(K3), n = 0, . . . , 7, up to extensions. The result can be extrapolated to n ≥8 by Bott periodicity. Table 19. KO-groups KO−n(K3), n = 0, . . . , 7, up to extensions. The result can be extrapolated to n ≥8 by Bott periodicity. JHEP03(2023)181 thus making contact with section 2.4. For X ∈{Sk, T k, K3, CY3}, the analysis turns out to be particularly simple, since all diﬀerentials in the AHSS vanish and extensions are trivial, as we explicitly showed. For more complicated backgrounds, these simpliﬁcations might not occur, but the AHSS should still give the correct answer. First, we expect the story to become substantially more involved if one turns on ﬂuxes. For instance, allowing for non-trivial NS-NS three-form ﬂux H leads to the computation of H-twisted K-theory groups K−n H (X) and the corresponding cobordism groups ΩSpinc,H(X). In this case where W3 = 0, i.e. we have a Spinc-structure, the absence of Freed-Witten anomalies implies that the H-ﬂux through a D-brane must vanish. This will result in non-trivial maps dr : Ep,q r →Ep+r,q−r+1 in the evaluation of the AHSS. Second, even in the purely geometric case (no ﬂuxes), when computing say KO−n(X) there could be non-trivial diﬀerentials, indicating e.g. that certain cycles are not Spin. An explicitly veriﬁcation of these expectations is left for future work. 17In general this would be slightly inaccurate, since the groups of the point might be actually direct sums, so one of them could correspond to more sites in the lattice. Here, we neglect this complication for the sake of simplicity. The analysis can be directly adapted. 4.1 General aspects All of the analysed examples have in common that the ﬁnal results of the AHSS can be expressed in a convenient, compact manner. For the K-theory groups K−n(X) of a k-dimensional manifold X ∈{Sk, T k, K3, CY3}, we have in fact K−n(X) = k M m=0 bk−m(X) K−n−m(pt) , (4.1) (4.1) with n ≥0. The interpretation of this result in terms of D-branes is as follows. Say we are in d = 10 dimensions and compactify the theory on the k-dimensional manifold X, so that the total space is R1,d−k−1 × X. Then, K−n(X) classiﬁes all D-branes that are of codimension n in the ﬂat space R1,d−k−1. From the d-dimensional point of view, these are given by the set of all codimension n + m branes wrapping (k −m)-cycles on the compact space X. Hence, the result (4.1) just reﬂects that the dimensional reduction performed following this perhaps naive geometrical reasoning is already the correct answer on these manifolds. Nevertheless, thanks to the AHSS we also learn that none of the wrapped D-branes experiences a Freed-Witten anomaly nor that there is an instantonic decay-channel. The relation (4.1) has a nice connection to the completeness hypothesis. The right hand side of (4.1) is indeed a lattice of charges (q1, . . . , qk), where each entry qm is a charge – 30 – vector with bk−m components.17 The fact that they can and indeed they are populated independently of one another means that in general the full spectrum of charges (or rather stable states with that given charge) is complete. To understand the point, consider the simple two-dimensional situation in which the lattice is just Z ⊕Z. In this case, one not only has stable bound states of branes associated to say (1, 0) and (0, 1), but also to (1, 1). Thus, what the relation (4.1) is telling us is that to any non-vanishing element (q1, . . . , qk) must be associated a stable object and, in this sense, the spectrum is complete. In general, especially in the presence of multicharged or non-BPS branes, the situation might become highly involved, but K-theory should give the correct answer. For cobordism groups we found an analogous result, namely that for n ≥0 they can also be expressed as JHEP03(2023)181 ΩSpinc n+k (X) = k M m=0 bk−m(X) ΩSpinc n+m(pt) . 4.1 General aspects Via the relation K−n(X) = Kn+k(X), (4.4) (4.4) valid for X a k-dimensional Spinc manifold, the K-theory result (4.1) can be formally brought into the same form as (4.2), namely we can pass from generalised cohomology to homology.18 Therefore (for n ≥0) the ABS orientation can be extended to a map αc X : ΩSpinc n+k (X) →Kn+k(X), (4.5) (4.5) acting as αc in (2.22) on each term ΩSpinc n+k−m(pt). Dividing by the kernel of this map provides an isomorphism between cobordism and K-theory classes on X. Hence, at least for these simple cases, the latter isomorphism is directly inherited from the isomorphism between ΩSpinc n (pt) and Kn(pt). As we have shown, the AHSS gives analogous simple results for the Spin cobordism groups ΩSpin n+k(X) and the real K-theory classes KOn+k(X) for X ∈{Sk, T k}. This implies that the above structure carries over to such cases, as well. JHEP03(2023)181 We can also give an interpretation in terms of global symmetries. In our examples, the groups Kn+k(X) and ΩSpinc n+k (X) classify all global (D −1 −n)-form charges in the non-compact D = d−k dimensions. These can be thought of as arising from the dimensional reduction of global d−1−n, d−2−n, . . . , d−1−k−n form charges along the k, k −1, . . . , 0 cycles of X. Due to the simple underlying structure, it is now clear that the fate of these global symmetries will follow the standard rules of the dimensional reduction. As already laid out in section 2.4, if a global symmetry in D dimensions descends from a global symmetry in d dimensions, then its gauging involves the dimensionally reduced gauge ﬁeld in d dimensions and also the corresponding dimensionally reduced D-branes (defects). In fact, the whole tadpole cancellation condition in D dimensions arises from the dimensional reduction of the tadpole cancellation condition in d dimensions. We will provide more concrete examples in section 4.2. 4.1 General aspects (4.2) (4.2) The case −k ≤n < 0 will be discussed later. We propose the following intuitive interpreta- tion of this result. First, recall that in the deﬁnition of Ωn(X) one introduces continuous maps f : M →X, for every n-dimensional compact manifold M, such that [M, f] ∈Ωn(X). A non-vanishing term labelled by m in the sum on the right hand side indicates that the map f : M →X from the (n + k)-dimensional manifold M into the k-dimensional manifold X is such that it wraps M around a non-trivial (k −m)-cycle of X, while no other obstruction is introduced by the map in the remaining (n + m) directions of M. Since there are bk−m diﬀerent (k −m)-cycles on X, we get bk−m factors of ΩSpinc n+m(pt) in the total cobordism group ΩSpinc n+k (X). Taking into account that the objects charged under the cobordism groups Ωn(pt) are the (d−n)-dimensional gravitational solitons mentioned in section 2.1, one can provide a similar interpretation as for the K-theory groups. Accordingly, ΩSpinc n+k (X) classiﬁes all gravitational solitons that are of codimension n in the ﬂat space R1,d−k−1. From the d-dimensional point of view, they are given by the set of all codimension (n+m) objects wrapping (k −m)-cycles on the compact space X. Concretely, deﬁning a basis {Σa m} of m-cycles on X, with a = 1, . . . , bm(X), and taking into account that ΩSpinc even (pt) = Z, for a given m-charge vector qm = (q1 m, . . . , qbm m ) ∈Zbm, (4.3) (4.3) the map f is such the (n + k)-dimensional manifold Mn+k is wrapped qa m times around the m-cycle Σa m of X. Hence, one can think of such an m-cycle to be shared between M and X. For all values of the index n + k, our goal is to explain how to organise the information contained in K-theory and cobordism groups of X and then reconstruct tadpole cancellation conditions known from string theory. As we will see, for n ≥0 this is quite straightforward, whereas in the regime −k ≤n < 0 we will encounter some new issues. We thus assume n ≥0 for the time being. Given the previous results, we can understand how the Hopkins-Hovey – 31 – isomorphism applies to cobordism and K-theory groups of manifolds X which are not just a point. 18The relation (4.4) and the analogous one for real K-theory, namely 4.2 Example of a Calabi-Yau threefold Let us now focus on the case of the ten-dimensional type IIB superstring compactiﬁed on a Calabi-Yau threefold X. In the previous section, we computed K0(X) = b6 K0(pt) \| {z } Z ⊕b4 K−2(pt) \| {z } Z ⊕b2 K−4(pt) \| {z } Z ⊕b0 K−6(pt) \| {z } Z , (4.6) (4.6) with b0 = b6 = 1. The corresponding D-branes are all of codimension zero in the ﬂat R1,3 space. In particular, in subsequent order, the four types of (single charged) D-branes corresponding to the K−2n(pt) groups are: D9-branes wrapping the entire CY3, D7-branes with b0 = b6 = 1. The corresponding D-branes are all of codimension zero in the ﬂat R1,3 space. In particular, in subsequent order, the four types of (single charged) D-branes corresponding to the K−2n(pt) groups are: D9-branes wrapping the entire CY3, D7-branes 18The relation (4.4) and the analogous one for real K-theory, namely elation (4.4) and the analogous one for real K-theory, namely KO−n(X) = KOn+k(X), KO−n(X) = KOn+k(X), for X a k-dimensional Spin manifold, follow e.g. from Theorem 2.9 of section V of [57], after recalling that a manifold is K-oriented (resp. KO-oriented) iﬀit is Spinc (resp. Spin). See also [58]. for X a k-dimensional Spin manifold, follow e.g. from Theorem 2.9 of section V of [57], after recalling that a manifold is K-oriented (resp. KO-oriented) iﬀit is Spinc (resp. Spin). See also [58]. – 32 – wrapping the b4 4-cycles of the CY3, D5-branes wrapping the b2 2-cycles of the CY3 and ﬁnally D3-branes being point-like on the CY3. At the next level, we found K−1(X) = b3 K−4(pt) \| {z } Z , (4.7) (4.7) corresponding to a codimension one brane in R1,3 and given by D5-branes wrapping any of the b3 three-cycles on the CY3. As already explained, for all multi-charges there should exist corresponding bound states of the single charged states. This is consistent with the completeness hypothesis. JHEP03(2023)181 As we have inferred from the AHSS, the corresponding cobordism groups split in a very similar manner ΩSpinc 6 (X) = b6 ΩSpinc 0 (pt) \| {z } Z ⊕b4 ΩSpinc 2 (pt) \| {z } Z ⊕b2 ΩSpinc 4 (pt) \| {z } Z⊕Z ⊕b0 ΩSpinc 6 (pt) \| {z } Z⊕Z (4.8) (4.8) and and ΩSpinc 7 (X) = b3 ΩSpinc 4 (pt) \| {z } Z⊕Z . 4.2 Example of a Calabi-Yau threefold (4.9) (4.9) As mentioned, this pattern is related to dimensional reduction of global symmetries. In- deed, from ΩSpinc 6 (X) = Z ⊕b4Z ⊕b2(Z ⊕Z) ⊕(Z ⊕Z) (4.10) (4.10) we infer that there is a 3b2 + 3 dimensional lattice of Z-valued global 3-form charges in R1,3 (recall b4 = b2). These are the dimensional reduction of the ten-dimensional 9-form, 7-form, 5-form and 3-form global symmetries along the 6, 4, 2, 0-cycles of the CY3. We now explain how to organise the information in the groups above and reconstruct tadpole cancellation conditions. Consider a six-dimensional Spinc-manifold M6 that lies in the contribution bmΩSpinc 6−m (pt) to ΩSpinc 6 (X) but, contrary to the Calabi-Yau X, it is not necessarily a solution to the string theory equations of motion. Hence, in this sense, M6 can be oﬀ-shell. Since there must exist a continuous map f : M6 →X, the manifold M6 shares some of the m-cycles with the background space X. Which m-cycles are shared depends on the non-zero entries in the charge vector (4.3). Then, the magnetic (6 −m)-form currents are obtained from the cobordism invariants (2.24), which we repeat below for convenience ˜J0(M6) = td0(M6) = 1 , ˜J2(M6) = td2(M6) = 1 2c1(M6) , ˜J4,1(M6) = td4(M6) = 1 12 c2(M6) + c2 1(M6) , ˜J4,2(M6) = c2 1(M6) , ˜J6,1(M6) = td6(M6) = 1 24c2(M6) c1(M6) , ˜J6,2(M6) = 1 2c3 1(M6) . (4.11) (4.11) Concretely, we propose that the magnetic (6 −m)-form currents are deﬁned by expanding the right hand sides into a basis of those (6 −m)-forms in H6−m(M6; Z) that also lie in Concretely, we propose that the magnetic (6 −m)-form currents are deﬁned by expanding the right hand sides into a basis of those (6 −m)-forms in H6−m(M6; Z) that also lie in – 33 – H6−m(X; Z) (again depending on the entries in the charge vector). For the Poincaré dual of the currents, denoted with hat, this means that we expand H6−m(X; Z) (again depending on the entries in the charge vector). For the Poincaré dual of the currents, denoted with hat, this means that we expand ˆ˜Jm,i(M6) = bm X a=1 αa m,i qa m Σa m + . . . (4.12) (4.12) where the dots indicate more contributions along m-cycles of M6 that do not lie in X. 19Formally, this 0-form arises from the ten-dimensional delta δ(0)(R1,3 × M6) = δ(0)(R1,3) ∧δ(0)(M6). 4.2 Example of a Calabi-Yau threefold Note that the (co)homology of M6 can in principle be bigger than that of X. Since this expansion is also valid for M6 ̸= X, it allows us to go slightly oﬀ-shell. More in general, topological K-theory and cobordism groups classify all global charges that can be present in principle, irrespective of properties like supersymmetry or being on-shell. JHEP03(2023)181 Recall that the Todd classes also deﬁne the ABS orientation at ﬁxed degree, i.e. αc n(M6) = Td(M6). Due to this map and the fact that all K-theory global sym- metries are gauged, we can infer that at ﬁxed n = 0, 2, 4, 6 always one linear combination of the above currents is gauged. Such a combination is the one entering the tadpole cancellation conditions. We discuss this below for all four classes of global symmetries in turn. • First, we have ΩSpinc 0 (pt) and K0(pt). The factor ΩSpinc 0 (pt) = Z gives rise to a single global 3-form symmetry in four dimensions, with the trivial magnetic current ˜J0(M6) = td0(M6) = 1. In ten dimensions, the corresponding 9-form symmetry is gauged with the charged objects being D9-branes, classiﬁed by K0(pt) = Z. This leads to the tadpole constraint • First, we have ΩSpinc 0 (pt) and K0(pt). The factor ΩSpinc 0 (pt) = Z gives rise to a single global 3-form symmetry in four dimensions, with the trivial magnetic current ˜J0(M6) = td0(M6) = 1. In ten dimensions, the corresponding 9-form symmetry is gauged with the charged objects being D9-branes, classiﬁed by K0(pt) = Z. This leads to the tadpole constraint N δ(0)(M6) + a(0) td0(M6) = 0 (4.13) (4.13) where δ(0)(M6) denotes the 0-form Poincaré dual to the 6-cycle M6 wrapped by the stack of N D9-branes.19 For a(0) = 0, −32, this is the familiar D9-tadpole cancellation condition in type IIB/type I string theory. Thus, one linear combination of the initial ΩSpinc 0 (pt) ⊕K0(pt) = Z ⊕Z global symmetries is gauged, while the orthogonal one is in general broken. • Second, we have ΩSpinc 2 (pt) and K−2(pt). 4.2 Example of a Calabi-Yau threefold The factor b4ΩSpinc 2 (pt) = b4Z gives rise to b4 global 3-form symmetries in four dimensions, whose preserved magnetic 0-form currents ˜j(2)a 0 (again in D = 4 and with the notation of section 2.4) are given by the expansion of the ten-dimensional 2-form current ˜J2(M6) = td2(M6) in a cohomological basis ω(2)a ∈H2(X; Z), namely • Second, we have ΩSpinc 2 (pt) and K−2(pt). The factor b4ΩSpinc 2 (pt) = b4Z gives rise to b4 global 3-form symmetries in four dimensions, whose preserved magnetic 0-form currents ˜j(2)a 0 (again in D = 4 and with the notation of section 2.4) are given by the expansion of the ten-dimensional 2-form current ˜J2(M6) = td2(M6) in a cohomological basis ω(2)a ∈H2(X; Z), namely ˜J2(M6) = b4 X a=1 ˜j(2)a 0 ∧ω(2)a . (4.14) (4.14) Note that b4 = b2, so that this is the Poincaré dual to the expansion (4.12), where we included the charges qa 4 into the coeﬃcients. Similarly, for a D7-brane classiﬁed by K−2(pt) and wrapping 4-cycles Σ4 ∈H4(M6; Z) that are contained in X (times the 9Formally, this 0-form arises from the ten-dimensional delta δ(0)(R1,3 × M6) = δ(0)(R1,3) ∧δ(0)(M6). Note that b4 = b2, so that this is the Poincaré dual to the expansion (4.12), where we included the charges qa 4 into the coeﬃcients. Similarly, for a D7-brane classiﬁed by K−2(pt) and wrapping 4-cycles Σ4 ∈H4(M6; Z) that are contained in X (times the 9Formally, this 0-form arises from the ten-dimensional delta δ(0)(R1,3 × M6) = δ(0)(R1,3) ∧δ(0)(M6). – 34 – ﬂat space R1,3), we can expand its Poincaré dual 2-form as ﬂat space R1,3), we can expand its Poincaré dual 2-form as ﬂat space R1,3), we can expand its Poincaré dual 2-form as an expand its Poincaré dual 2-form as δ(2)(R1,3 × Σ4) = b4 X a=1 δ(0)(R1,3)(2)a ∧ω(2)a . (4.15) (4.15) In ten dimensions, the gauging of the corresponding 7-form global symmetry is associated to a tadpole constraint X j∈def Nj δ(2)(R1,3 × Σ4,j) + a(2) c1(M6) 2 = 0 (4.16) (4.16) which upon expansion in a cohomological basis of H2(X; Z) = b4Z leads to b4 = b2 tadpole cancellation conditions. Hence, a subgroup b4Z of the initially present global symmetry b4 ΩSpinc 2 (pt) ⊕K−2(pt) = b4 (Z ⊕Z) is gauged while the orthogonal b4Z group is broken. 4.2 Example of a Calabi-Yau threefold Of course, for M6 = X we have c1(X) = 0 and the tadpole cancellation condition simpliﬁes, but the power of K-theory and cobordism is that they allow us to go oﬀ-shell and see terms that could appear in principle, even if they are absent for the on-shell conﬁgurations. We note that, for a(2) = −24 and M6 being the base B3 of an elliptically ﬁbered Calabi-Yau fourfold, (4.16) is the well known 7-brane tadpole constraint of F-theory. JHEP03(2023)181 • Third, we have ΩSpinc 4 (pt) and K−4(pt). The factor b2ΩSpinc 4 (pt) = b2 (Z ⊕Z) gives rise to 2b2 global 3-form symmetries in four dimensions. Notice that this time the ABS orientation between K-theory and cobordism is not an isomorphism. The preserved magnetic 0-form currents ˜j(4)a 0,i , i = 1, 2, in D = 4 are given by the expansion of the ten-dimensional 4-form currents ˜J4,i(M6) in a cohomological basis ˆω(4)a of H4(X; Z) • Third, we have ΩSpinc 4 (pt) and K−4(pt). The factor b2ΩSpinc 4 (pt) = b2 (Z ⊕Z) gives rise to 2b2 global 3-form symmetries in four dimensions. Notice that this time the ABS orientation between K-theory and cobordism is not an isomorphism. The preserved magnetic 0-form currents ˜j(4)a 0,i , i = 1, 2, in D = 4 are given by the expansion of the ten-dimensional 4-form currents ˜J4,i(M6) in a cohomological basis ˆω(4)a of H4(X; Z) ˜J4,i(M6) = b2 X a=1 ˜j(4)a 0,i ∧ˆω(4)a . (4.17) (4.17) The defects classiﬁed by K−4(pt) are D5-branes wrapping 2-cycles ˆΣ2 on M6 that are shared with X (times the ﬂat space R1,3). Again their Poincaré duals can be expanded similarly to (4.17). The gauging of the ten-dimensional 5-form symmetry implies a tadpole condition of the form The defects classiﬁed by K−4(pt) are D5-branes wrapping 2-cycles ˆΣ2 on M6 that are shared with X (times the ﬂat space R1,3). Again their Poincaré duals can be expanded similarly to (4.17). The gauging of the ten-dimensional 5-form symmetry implies a tadpole condition of the form X j∈def Nj δ(4)(R1,3 × ˆΣ2,j) + a(4) 1 c2(M6) + c2 1(M6) 12 ! + a(4) 2 c2 1(M6) = 0, (4.18) (4.18) where we followed the general strategy reviewed in section 2.3, i.e. both cobordism invariants can in principle appear in the gauging procedure. 4.2 Example of a Calabi-Yau threefold The factor ΩSpinc 6 (pt) = Z ⊕Z gives rise to 2 global 3-form symmetries in four dimensions, whose preserved magnetic 0-form currents ˜j(6) 0,i , i = 1, 2 (again in D = 4) are given by the reduction of the ten-dimensional 6-form currents ˜J6,i(M6) along the volume 6-form of M6, • Fourth, we have ΩSpinc 6 (pt) and K−6(pt). The factor ΩSpinc 6 (pt) = Z ⊕Z gives rise to 2 global 3-form symmetries in four dimensions, whose preserved magnetic 0-form currents ˜j(6) 0,i , i = 1, 2 (again in D = 4) are given by the reduction of the ten-dimensional 6-form currents ˜J6,i(M6) along the volume 6-form of M6, ˜J6,i(M6) = ˜j(6) 0,i vol(M6) . (4.19) (4.19) The defects classiﬁed by K−6(pt) are D3-branes being point-like on M6. Then, the gauging of the ten-dimensional 3-form symmetry implies a tadpole condition of the general form The defects classiﬁed by K−6(pt) are D3-branes being point-like on M6. Then, the gauging of the ten-dimensional 3-form symmetry implies a tadpole condition of the general form JHEP03(2023)181 X j∈def Nj δ(6)(R1,3 × ptj) + a(6) 1 c2(M6) c1(M6) 24 + a(6) 2 c3 1(M6) 2 = 0 . (4.20) (4.20) Again, for a Calabi-Yau manifold, such as M6 = X, the two contributions from cobordism are vanishing but the oﬀ-shell nature of cobordism itself makes them visible in the general case. For a(6) 1 = −12 and a(6) 2 = −30, this tadpole condition is known to be realised in F-theory compactiﬁed on a smooth elliptically ﬁbered Calabi-Yau fourfold with base M6 = B3. Again, for a Calabi-Yau manifold, such as M6 = X, the two contributions from cobordism are vanishing but the oﬀ-shell nature of cobordism itself makes them visible in the general case. For a(6) 1 = −12 and a(6) 2 = −30, this tadpole condition is known to be realised in F-theory compactiﬁed on a smooth elliptically ﬁbered Calabi-Yau fourfold with base M6 = B3. Finally, let us discuss the four-dimensional global 2-form symmetries related to K−1(X) = b3K−4(pt) = b3Z and ΩSpinc 7 (X) = b3ΩSpinc 4 (pt) = b3(Z ⊕Z). From the ten-dimensional perspective, these arise from the reduction of the global 5-form symmetries along the b3 3-cycles of X. 4.2 Example of a Calabi-Yau threefold Upon expansion in a cohomological basis of H4(X; Z) = b2Z, one obtains b2 = b4 tadpole cancellation conditions. Hence, a subgroup b2Z of the initially present global symmetry is gauged while the orthogonal group is broken. The type I string on X = K3 × T 2 leads to such a D5-brane tadpole constraint for a(4) 1 = −12 and a(4) 2 = 3/2. Another setup is the Ωσ orientifold of type IIB on X = K3 × T 2 presented in [59, 60]. This example has only O5-planes and leads to the above tadpole constraint with a(4) 1 = −24 and a(4) 2 = 0. Thus, we see again that more information is needed to completely specify the gauging related to speciﬁc string models. where we followed the general strategy reviewed in section 2.3, i.e. both cobordism invariants can in principle appear in the gauging procedure. Upon expansion in a cohomological basis of H4(X; Z) = b2Z, one obtains b2 = b4 tadpole cancellation conditions. Hence, a subgroup b2Z of the initially present global symmetry is gauged while the orthogonal group is broken. The type I string on X = K3 × T 2 leads to such a D5-brane tadpole constraint for a(4) 1 = −12 and a(4) 2 = 3/2. Another setup is the Ωσ orientifold of type IIB on X = K3 × T 2 presented in [59, 60]. This example has only O5-planes and leads to the above tadpole constraint with a(4) 1 = −24 and a(4) 2 = 0. Thus, we see again that more information is needed to completely specify the gauging related to speciﬁc string models. – 35 – • Fourth, we have ΩSpinc 6 (pt) and K−6(pt). The factor ΩSpinc 6 (pt) = Z ⊕Z gives rise to 2 global 3-form symmetries in four dimensions, whose preserved magnetic 0-form currents ˜j(6) 0,i , i = 1, 2 (again in D = 4) are given by the reduction of the ten-dimensional 6-form currents ˜J6,i(M6) along the volume 6-form of M6, • Fourth, we have ΩSpinc 6 (pt) and K−6(pt). 4.2 Example of a Calabi-Yau threefold Concerning ΩSpinc 7 (X), the 2b3 preserved magnetic 1-form currents ˜j(3)a 1,i , with i = 1, 2, in D = 4 are given by the dimensional reduction of the ten-dimensional 4-form currents ˜J4,i(M6) along the basis 3-forms ω(3)a ∈H3(X; Z), ˜J4,i(M6) = b3 X a=1 ˜j(3)a 1,i ∧ω(3)a . (4.21) (4.21) Note that this is meant in principle and that the currents ˜j(3)a 1,i can also be vanishing. The D5-brane defects wrapping 3-cycles Σ3 on M6 shared with X times a three-dimensional submanifold Π3 of the ﬂat space R1,3 can be expanded in a similar fashion Note that this is meant in principle and that the currents ˜j(3)a 1,i can also be vanishing. The D5-brane defects wrapping 3-cycles Σ3 on M6 shared with X times a three-dimensional submanifold Π3 of the ﬂat space R1,3 can be expanded in a similar fashion δ(4)(Π3 × Σ3) = b3 X a=1 δ(1)(Π3)(3)a ∧ω(3)a. (4.22) (4.22) In ten dimensions, the global symmetry of K−4(pt) is gauged leading to a magnetic Bianchi identity d ˜F3 = X j∈def Nj δ(4)(Π3,j × Σ3,j) + a(4) 1 ˜J4,1(M6) + a(4) 2 ˜J4,2(M6) . (4.23) (4.23) Expanding now also the magnetic ﬁeld strength as Expanding now also the magnetic ﬁeld strength as ˜F3 = b3 X a=1 ˜f(3)a 0 ∧ω(3)a, (4.24) (4.24) – 36 – – 36 – we arrive at b3 Bianchi identities for the four-dimensional 0-forms d ˜f(3)a 0 = X j∈def Nj δ(1)(Π3,j)(3)a + a(4) 1 ˜j(3)a 1,1 + a(4) 2 ˜j(3)a 1,2 . (4.25) (4.25) Thus, everything ﬁts nicely together once more. The discussion for higher groups, such as ΩSpinc 8 (X) and ΩSpinc 9 (X), together with their K-theory counterparts, follows the same logic. Summary of results. We demonstrated that, for the example of a Calabi-Yau space X, the K-theory and cobordism classes on X for n ≥0 are to be interpreted from the point of view of global symmetries and their subsequent gauging. In this situation, the AHSS is simple in the sense that no non-trivial maps, i.e. diﬀerentials, appear and the outcomes reproduce the naive expectation from dimensional reduction. Of course, a more involved task is to compute K-theory and cobordism classes where maps can be non-trivial and D-branes become inconsistent or unstable. 4.2 Example of a Calabi-Yau threefold However, even if in these cases the D-brane spectrum for a background space X changes, the map between K-theory and cobordism is proven to be intact, so that the related global symmetries are guaranteed to disappear simultaneously. Therefore, we expect that an interpretation in terms of gauging will still be very similar to what we discussed above. JHEP03(2023)181 We also expect our results to carry over to the correspondence of KO-groups and Spin-cobordisms. A new aspect is the appearance of Z2 torsion groups related to non-BPS branes on the K-theory side. As discussed (see e.g. (2.27) or the examples in [27]), the corresponding cobordism groups can decouple from tadpoles or, more precisely, charge neutrality conditions, so they would need to be broken by some unknown defects. At this stage, we cannot exclude that a more thorough analysis reveals some subtle aspects, but this is beyond the scope of this paper. 5 Conclusion JHEP03(2023)181 The fact that there should be no global symmetries is believed to be a fundamental property of quantum gravity and is one of the best tested swampland conjectures. The cobordism conjecture [6] is a recent generalisation of this fact, which interprets a non-vanishing cobordism group as a higher-form global symmetry in an eﬀective ﬁeld theory of quantum gravity. Demanding its absence, one is either led to breaking or gauging the symmetry. While breaking can lead to the prediction of new objects (defects), gauging can be performed by exploiting the close relation between cobordism and K-theory. Indeed, K-theory charges are always gauged and this reﬂects to the fate of cobordism charges, as proposed in [27]. In this work, we gave further support to the idea that cobordism and K-theory groups are charges in quantum gravity by computing the groups associated to a compact manifold X, for typical choices employed in string compactiﬁcations. In particular, we showed that those groups contain precisely the information on how symmetries (broken or gauged) spread across the dimensional reduction of the eﬀective theory on X. One of the advantages of the cobordism and K-theory description, rather than standard (co)homology, is that it automatically takes into account quantum mechanical eﬀects, such as cancellation of Freed-Witten anomalies. This is most clear when performing the computation through the Atiyah-Hirzebruch spectral sequence, where absence of such anomalies is related to the vanishing of a certain diﬀerential. We reviewed this technique in detail, discussing and working out explicitly several examples. Then, we gave a physical interpretation of the results and show how the information of gaugings and tadpoles can be decoded from the cobordism and K-theory groups of X. The work here presented can be extended along various directions. It would be important to ﬁnd a ﬁrst principle derivation of the unﬁxed coeﬃcients (a(n) i ) in tadpole cancellation conditions without using any string theory input, as this could give support to the so called String Lamppost Principle. It would also be interesting to extend the analysis in order to systematically include torsional groups, both in K-theory [61] (see also [62]) and cobordism. Indeed, in most of the examples we discussed torsion was playing little role. However, it is unavoidable in more realistic and involved setups. 4.3 Fate of low-dimensional ΩSpinc n (X) It remains to discuss what happens in the regime −k ≤n < 0, for which the cobordism groups ΩSpinc n+k (X) are still non-vanishing. To get a better idea on what is diﬀerent with respect to the regime n ≥0, we start by asking what the corresponding K-theory groups are, namely Kn+k(X) = K−n(X) with −k ≤n < 0, and what they physically mean. For concreteness, consider e.g. the class K2(CY3). Extrapolating the relation (4.1) to n = −2, we would get K2(X) = 6 M m=2 b6−m(X) K2−m(pt) = b4(X) K0(pt) ⊕. . . , (4.26) (4.26) where from the sum we left out the term K2(pt), associated to m = 0. The latter could be deﬁned via Bott periodicity to be equal to Z, but it is not clear what it should represent physically. In addition, it does not appear on the cobordism side (4.2). The term written explicitly on the right hand side of (4.26) would correspond to a D9-brane wrapped on a 4-cycle of the CY3. However, for dimensional reasons the D9 should really wrap a 6-chain (which can be thought of as the 4-cycle times a 2-chain) on the CY3 and, as such, it is not a topologically non-trivial conﬁguration. This argument suggests that the groups – 37 – Kn+k(X) = K−n(X) with −k ≤n < 0 do not admit a clear physical interpretation in terms of wrapped D-branes. This is further supported by extrapolating the result from section 4.1, namely the fact that K−n(X) corresponds to codimension n branes in D = d−k dimensions. Clearly, for n negative the branes do not ﬁt. For the corresponding cobordism groups ΩSpinc n+k (X) with −k ≤n < 0, the reasoning is completely analogous with the D-branes exchanged by gravitational solitons. As also claimed in section 4.1, ΩSpinc n+k (X) classiﬁes all gravitational solitons that are of codimension n in the ﬂat non-compact space. Again, for n negative the gravitational solitons do not ﬁt. 5 Conclusion One can introduce torsion in the spectral sequence but also on the manifold X itself, for example by considering Calabi-Yau manifolds with torsion [36, 37]. The question is then whether torsion classes are killed by the diﬀerentials or if they survive until the end of the sequence, thus giving additional contributions, say, to tadpoles. Another possible direction to extend this work is – 38 – by including more structure on top of the considered manifold, such as gauge ﬁelds, which are again unavoidable in realistic settings. One should then turn to reﬁned version of the groups here considered, such as twisted and diﬀerential K-theory, see e.g. [55, 63–66]. We hope to come back to these questions in the future. Acknowledgments It is a pleasure to thank Arun Debray for useful discussions. The work of N.C. is supported by the Alexander-von-Humboldt foundation. JHEP03(2023)181 A.1 Short exact sequences, extensions and Ext A.1 Short exact sequences, extensions and Ext A Mathematical tools and results In this appendix, we collect mathematical tools and results used through the main part of the work. A.1 Short exact sequences, extensions and Ext A.1 Short exact sequences, extensions and Ext Two useful properties of these groups are Extn(⊕iAi, B) = ΠiExtn(Ai, B), (A.4) Extn(A, ΠiBi) = ΠiExtn(A, Bi), (A.5) (A.4) (A.5) (A.4) (A.5) – 39 – and we recall that for abelian groups direct product and direct sum coincide. For cyclic groups, we recall the results Ext1(Z, Z) = 0, (A.6) Ext1(Z, Zn) = 0, (A.7) Ext1(Zn, Z) = Zn, (A.8) Ext1(Zm, Zn) = Zk, (A.9) Ext1(Z, Z) = 0, (A.6) Ext1(Z, Zn) = 0, (A.7) Ext1(Zn, Z) = Zn, (A.8) Ext1(Zm, Zn) = Zk, (A.9) where k = GCD(m, n). All of this is used in the calculations of section 3 e k = GCD(m, n). All of this is used in the calculations of section 3 where k = GCD(m, n). All of this is used in the calculations of section 3 JHEP03(2023)181 Let us give two simple examples to illustrate how everything works in a combined way. Let us consider the short exact sequence 0 →Z2 →e(Z2, Z2) →Z2 →0. (A.10) (A.10) Since Ext1(Z2, Z2) = Z2, e(Z2, Z2) is not split, instead we have two possible extensions. Indeed, it is well-known that there are two short exact sequences Since Ext1(Z2, Z2) = Z2, e(Z2, Z2) is not split, instead we have two possible extensions. Indeed, it is well-known that there are two short exact sequences 0 →Z2 →Z4 →Z2 →0, (A.11) 0 →Z2 →Z2 ⊕Z2 →Z2 →0. (A.12) (A.11) (A.12) Instead, the short exact sequence Instead, the short exact sequence 0 →Z3 →Z6 →Z2 →0, (A.13) (A.13) is split, since Ext1(Z2, Z3) = 0. A.1 Short exact sequences, extensions and Ext Consider the abelian groups A, B and C. A short sequence onsider the abelian groups A, B and C. A short sequence 0 −→B β −→C α −→A −→0 (A.1) (A.1) is exact if the map β is injective and the map α surjective, i.e. if ker(α) = Im(β). In this case, we say that C is an extension of A by B and we denote it as is exact if the map β is injective and the map α surjective, i.e. if ker(α) = Im(β). In this case, we say that C is an extension of A by B and we denote it as C = e(A, B). (A.2) (A.2) The Splitting Lemma for abelian groups tells us that the extension is trivial, C = A ⊕B, (A.3) (A.3) iﬀthere is a left inverse to β iﬀthere is a right inverse to α. In this case, one says that the short exact sequence is split. In general, the extension might not be unique and there can be more extensions besides the trivial one. Equivalence classes of extensions of A by B are in one-to-one correspondence with elements of the group Ext1(A, B), with the trivial extension corresponding to 0 (see e.g. Theorem 3.4.3 of [67]). iﬀthere is a left inverse to β iﬀthere is a right inverse to α. In this case, one says that the short exact sequence is split. In general, the extension might not be unique and there can be more extensions besides the trivial one. Equivalence classes of extensions of A by B are in one-to-one correspondence with elements of the group Ext1(A, B), with the trivial extension corresponding to 0 (see e.g. Theorem 3.4.3 of [67]). The deﬁnition and main properties of the groups Extn(A, B) can be found e.g. in [67], chapter 3. We recall some of them below. As stated in Lemma 3.3.1, if A and B are abelian (as we assume) we have that Extn(A, B) = 0 for n ≥2. Therefore, only the groups associated to n = 0, 1 are relevant for us. We have that Ext0(A, B) = Hom(A, B), while Ext1(A, B) classiﬁes extensions of A by B, as anticipated above. A.3 Properties of Steenrod squares In this appendix, we collect some useful facts about Steenrod squares. For a nice, pedagogical review and for more information, we refer the reader to [68] and references therein. A standard textbook is [56]. We will work at prime 2, but it is possible to generalize the discussion to any prime p. A cohomology operation of degree i is a map A cohomology operation of degree i is a map Hn(X; Z2) →Hn+i(X; Z2). (A.16) (A.16) It is said to be stable if it commutes with the suspension isomorphism. Steenrod squares, Sqi, are stable cohomology operations of degree i satisfying the following deﬁning properties, for any i ≥0: JHEP03(2023)181 a) Sq0 = Id; b) Sqi(x) = x ∪x, for x ∈Hi(X; Z2); c) Sqi(x) = 0, for x ∈Hj(X; Z2) and j < i; d) Sqi(x ∪y) = X m+n=i Sqm(x) ∪Sqn(y) (Cartan formula). i(x) = x ∪x, for x ∈Hi(X; Z2); c) Sqi(x) = 0, for x ∈Hj(X; Z2) and j < i; c) Sqi(x) = 0, for x ∈Hj(X; Z2) and j < i; d) Sqi(x ∪y) = X m+n=i Sqm(x) ∪Sqn(y) (Cartan formula). d) Sqi(x ∪y) = X m+n=i Sqm(x) ∪Sqn(y) (Cartan formula). e) Sqi ◦Sqj = ⌊i/2⌋ X k=0 j −k −1 i −2k ! mod 2 Sqi+j−k ◦Sqk, for 0 < i < 2j (Adem relation) e) Sqi ◦Sqj = ⌊i/2⌋ X k=0 j −k −1 i −2k ! mod 2 Sqi+j−k ◦Sqk, for 0 < i < 2j (Adem relation). (Adem relation). The map Sq1 ≡˜β is an example of a Bockstein homomorphism. It is associated to the short exact sequence 2 0 →Z2 ×2 →Z4 ρ→Z2 →0, (A.17) (A.17) where the ﬁrst map is multiplication by 2 and the second (ρ) is the reduction modulo 2, which induces the long exact sequence where the ﬁrst map is multiplication by 2 and the second (ρ) is the reduction modulo 2, which induces the long exact sequence . . . ˜β→Hn(X; Z2) ×2 →Hn(X; Z4) ρ→Hn(X; Z2) ˜β→Hn+1(X; Z2) →. . . . (A.18) (A.18) Here, ˜β is the connecting homomorphism between cohomology groups of diﬀerent degree. Another Bockstein homomorphism, called β in the main text, can be constructed in association with the short exact sequence 0 →Z ×2 →Z ρ→Z2 →0, (A.19) (A.19) inducing in turn the long exact sequence inducing in turn the long exact sequence . . A.2 Universal coeﬃcient theorem The universal coeﬃcient theorem (see e.g. [67]) can be used to express (co)homology groups of a topological space X with coeﬃcients in a left Z-module A in terms of (co)homology groups with coeﬃcients in Z. It can be formulated both for homology and cohomology. The version for homology groups states that there is a short (noncanonically) split exact sequence (A.14) 0 →Hn(X) ⊗A →Hn(X; A) →Tor1(Hn−1(X), A) →0. (A.14) The deﬁnition of the groups Torn(A, B) can be found e.g. in [67], chapter 3. As stated in Proposition 3.1.2 and 3.1.4, if A and B are abelian, Torn(A, B) are torsion abelian groups and they vanish for n ≥2; if A is also torsion free, Tor1(A, B) = 0. The version for cohomology groups states that there is a short (noncanonically) split exact sequence (A.15) 0 →Ext1Hn−1(X; A) →Hn(X; A) →Hom(Hn−1(X), A) →0. (A.15) – 40 – A.3 Properties of Steenrod squares [35, 38]) Sqi(y) = ι∗(wi(N)) ∪y, (A.26) (A.26) where y ∈Hn(X; Z2), N is the normal bundle of the submanifold Y ⊂X Poincaré dual to y and ι : Y →X is the inclusion.21 This is most convenient for physical purposes, such as checking the absence of Freed-Witten anomalies for branes wrapping Y , on which we comment in section 3.3.2 (there, following [35, 38], we directly employ the integral lift W3(N) of w3(N) and omit the pushforward ι∗). where y ∈Hn(X; Z2), N is the normal bundle of the submanifold Y ⊂X Poincaré dual to y and ι : Y →X is the inclusion.21 This is most convenient for physical purposes, such as checking the absence of Freed-Witten anomalies for branes wrapping Y , on which we comment in section 3.3.2 (there, following [35, 38], we directly employ the integral lift W3(N) of w3(N) and omit the pushforward ι∗). A.3 Properties of Steenrod squares β→Hn(X; Z) ×2 →Hn(X; Z) ρ→Hn(X, Z2) β→Hn+1(X, Z) →. . . . (A.20) (A.20) The two Bocksteins are related by The two Bocksteins are related by ˜β = ρ ◦β. (A.21) ˜β = ρ ◦β. (A.21) – 41 – – 41 – At odd degree i = 2k + 1, one can deﬁne an integral lift of the Steenrod squares, Sq2m+1 Z = β ◦Sq2m, (A.22) (A.22) which is such that ρ ◦Sq2m+1 Z = Sq2m+1 and maps Hn(X; Z2) →Hn+i(X; Z). One further gets a map between integral cohomology by ﬁrst reducing modulo 2 and then acting with Sqi Z, which is such that ρ ◦Sq2m+1 Z = Sq2m+1 and maps Hn(X; Z2) →Hn+i(X; Z). One further gets a map between integral cohomology by ﬁrst reducing modulo 2 and then acting with Sqi Z, Sqi Z ◦ρ : Hn(X; Z) →Hn+i(X; Z). (A.23) (A.23) n integral lift of Sqi for even i = 2m does not exist.20 An integral lift of Sqi for even i = 2m does not exist.20 JHEP03(2023)181 Given an element x ∈Hk−i(X; Z2), with k = dim(X), the action of the Steenrod squares can be deﬁned as Sqi(x) = νi ∪x, (A.24) (A.24) where νi ∈Hi(X; Z2) is the i-th Wu class of X (more precisely, of a real vector bundle over X of rank k, which we generically take to be the tangent bundle), such that νi = 0 if i > k −i. Since the total Wu class is the Steenrod square of the total Stiefel-Whitney class, one can express each of the single Wu classes in terms of Stiefel-Whitney classes. At lower degree, one has ν1 = w1, ν2 = w2 + w1 ∪w1, (A.25) ν1 = w1, ν1 = w1, ν2 = w2 + w1 ∪w1, ν w ∪w (A.25) ν1 w1, ν2 = w2 + w1 ∪w1, (A.25) (A.25) ν3 = w1 ∪w2. n certain cases, one can give an alternative action of Sqi, namely (see e.g. [35, 38]) In certain cases, one can give an alternative action of Sqi, namely (see e.g. g qZ 21More in general [56], one can deﬁne an action Sqi(u) = π∗(wi(ξ)) ∪u, with u ∈Hk(E; Z2) and wi(ξ) ∈Hi(B; Z2), for any k-plane bundle ξ : F →E π→B of which the normal bundle N(B) is a particular case. A.4 Wedge sum, smash product and reduced suspension Consider two pointed topological spaces (X, x0) and (Y, y0). The wedge sum, X ∨Y , is deﬁned as X ∨Y = X ⊔Y/ ∼, (A.27) (A.27) 20This can be proven as follows. Suppose it exists an integral lift for the even case, Sq2m = ρ ◦Sq2m Z . Exactness of the sequence (A.20) means that ker β = Imρ, implying in turn β ◦Sq2m = β(ρ(Sq2m Z )) = 0. However, this is contradiction with the Adem relation Sq1 ◦Sq2m = Sq2m+1 ̸= 0 (recall Sq1 = ρ ◦β). Thus, such an integral lift Sq2m Z cannot exist. g qZ 21More in general [56], one can deﬁne an action Sqi(u) = π∗(wi(ξ)) ∪u, with u ∈Hk(E; Z2) and wi(ξ) ∈Hi(B; Z2), for any k-plane bundle ξ : F →E π→B of which the normal bundle N(B) is a particular case. – 42 – where the equivalence relation identiﬁes the two base points x0 and y0. The smash product, X ∧Y , is deﬁned as the quotient of the cartesian product by the wedge sum X ∧Y = X × Y X ∨Y (A.28) X ∧Y = X × Y X ∨Y (A.28) It satisﬁes the properties X ∧Y ∼= Y ∧X, (A.29) (X ∧Y ) ∧Z ∼= X ∧(Y ∧X), (A.30) ΣnX ∼= Sn ∧X. ΣnX ∼= Sn ∧X. (A.32) An important case is when X = Sk, thus giving ΣnSk ∼= Sn+k. (A.33) We also recall that Σ0 ∧X ∼= S0 ∧X ∼= X, (A.34) Σ0 ∧X ∼= S0 ∧X ∼= X, (A.34) where S0 ∼= pt ⊔pt. Another useful formula is Σ(X × Y ) ∼= ΣX ∨ΣY ∨Σ(X ∧Y ). (A.35) (A.35) .5 Cobordism groups of spheres and tori X ∧Y ∼= Y ∧X, (A.29) (X ∧Y ) ∧Z ∼= X ∧(Y ∧X), (A.30) (A.29) (A.30) JHEP03(2023)181 where the symbol ∼= means homeomorphic as topological spaces. Consider then the n-sphere Sn. The reduced suspension of X is deﬁned as ΣX ∼= S1 ∧X. (A.31) The construction can be iterated ΣX ∼= S1 ∧X. (A.31) (A.31) ΣnX ∼= Sn ∧X. (A.32) An important case is when X = Sk, thus giving ΣnSk ∼= Sn+k. (A.33) A.5 Cobordism groups of spheres and tori We can prove that for a generic structure ξ the cobordism groups of spheres (Sk) and tori (T k) have a simple decomposition in terms of the respective cobordism groups of the point, namely Ωξ n(Sk) = Ωξ n(pt) ⊕Ωξ n−k(pt), (A.36) Ωξ n(T k) = k M i=0 k i ! Ωξ n−i(pt), (A.37) (A.36) (A.37) where we implicitly assume the groups with negative index to be vanishing. This explains for example why, when computing Spin and Spinc cobordism groups of spheres in section 3.2, even if we found that in general where we implicitly assume the groups with negative index to be vanishing. This explains for example why, when computing Spin and Spinc cobordism groups of spheres in section 3.2, even if we found that in general ΩSpin n (Sk) =    ΩSpin n (pt) n < k, e(ΩSpin n−k(pt), ΩSpin n (pt)) n ≥k, (A.38) (A.38) – 43 – – 43 – (and similarly for Spinc cobordism), every time the information at our disposal was enough to solve the extension problem, it turned out to be trivial. We now prove (A.36) and (A.37) by induction. (and similarly for Spinc cobordism), every time the information at our disposal was enough to solve the extension problem, it turned out to be trivial. We now prove (A.36) and (A.37) by induction. We start from the cobordism groups of spheres, Sk. For S1, we have Ωξ n(S1) = Ωξ n(pt) ⊕˜Ωξ n(S1) = Ωξ n(pt) ⊕˜Ωξ n Σ(S0) = Ωξ n(pt) ⊕˜Ωξ n−1(S0) = Ωξ n(pt) ⊕Ωξ n−1(pt). (A.39) (A.39) JHEP03(2023)181 In passing from the second to the third line we used the suspension axiom ˜Ωξ n(ΣX) = ˜Ωξ n−1(X) [32], while in the last step we employed that ˜Ωξ n(S0) = Ωξ n(pt), which follows from In passing from the second to the third line we used the suspension axiom ˜Ωξ n(ΣX) = ˜Ωξ n−1(X) [32], while in the last step we employed that ˜Ωξ n(S0) = Ωξ n(pt), which follows from Ωξ n(S0) = Ωξ n(pt ⊔pt) = Ωξ n(pt) ⊕Ω∗ n(pt) = Ωξ n(pt) ⊕˜Ωξ n(S0). (A.40) (A.40) Then, we assume the formula to hold for Sk and we prove it for Sk+1. A.5 Cobordism groups of spheres and tori Using again the Splitting Lemma (2.7) and the suspension axiom, we have Ωξ n(Sk+1) = Ωξ n(pt) ⊕˜Ωξ n(Sk+1) = Ωξ n(pt) ⊕˜Ωξ n Σ(Sk) = Ωξ n(pt) ⊕˜Ωξ n−1(Sk) = Ωξ n(pt) ⊕Ωξ n−k−1(pt). (A.41) (A.41) This proves (A.36) by induction. This proves (A.36) by induction. Then, we look at the cobordism groups of tori, T k. The result for T 1 = S1 is already proven in (A.39). We thus assume the formula to hold for T k and we prove it for T k+1. To this purpose, using (A.35) we can write Σ(T k × S1) = Σ(T k) ∨Σ(S1) ∨Σ(T k ∧S1) (A.42) (A.42) and therefore Ωξ n(T k+1) = Ωξ n(pt) ⊕˜Ωξ n+1 Σ(T k+1) = Ωξ n(pt) ⊕˜Ωξ n+1 Σ(T k × S1) = Ωξ n(pt) ⊕˜Ωξ n+1 Σ(T k) ⊕˜Ωξ n+1(Σ(S1)) ⊕˜Ωξ n+1 Σ2(T k) = Ωξ n(pt) ⊕˜Ωξ n(T k) ⊕˜Ωξ n(S1) ⊕˜Ωξ n−1(T k) = Ωξ n(T k) ⊕Ωξ n−1(T k), (A.43) (A.43) where we used ˜Ω(X∨Y ) = ˜Ω(X)⊕˜Ω(Y ), valid for reduced generalized homology theories [32]. where we used ˜Ω(X∨Y ) = ˜Ω(X)⊕˜Ω(Y ), valid for reduced generalized homology theories [32]. – 44 – We can ﬁnally demonstrate that We can ﬁnally demonstrate that We can ﬁnally demonstrate that Ωξ n(T k+1) = Ωξ n−1(T k) ⊕Ωξ n(T k) = k M i=0 k i ! Ωξ n−1−i(pt) ⊕ k M i=0 k i ! Ωξ n−i(pt) = k+1 M i=1 k i −1 ! Ωξ n−i(pt) ⊕ k M i=0 k i ! Ωξ n−i(pt) = k+1 M i=0 k i −1 ! Ωξ n−i(pt) ⊕ k+1 M i=0 k i ! Ωξ n−i(pt) = k+1 M i=0 k + 1 i ! Ωξ n−i(pt). (A.44) (A.44) JHEP03(2023)181 In passing from the third to the fourth line we just added zero to both terms, while in the last step we used Pascal’s formula. This concludes our proof of (A.37) by induction. In passing from the third to the fourth line we just added zero to both terms, while in the last step we used Pascal’s formula. This concludes our proof of (A.37) by induction. An alternative proof can be given by exploiting some more advanced mathematical constructions. 22For any spectrum G, the reduced generalised homology is deﬁned as ˜Gn(X) = [S, G ∧X]n, where S = Σ∞is the sphere spectrum. The unreduced generalised homology is instead Gn(X) = [S, G ∧X+]n, where X+ = X ⊔pt. We refer e.g. to [32] for more details. A.5 Cobordism groups of spheres and tori In particular, one can use that Spin and Spinc cobordism are generalised homology theories classiﬁed by Thom spectra MSpin and MSpinc respectively. One can thus write22 ˜ΩSpin n (X) := [S, MSpin ∧X]n, (A.45) (A.45) where X is a generic topological space, and similarly for Spinc. Considering for example X = Sk, by exploiting the properties of the smash product and the suspension given in appendix A.4, we have where X is a generic topological space, and similarly for Spinc. Considering for example X = Sk, by exploiting the properties of the smash product and the suspension given in appendix A.4, we have ˜ΩSpin n (Sk) := [S, MSpin ∧Sk]n = [S, MSpin]n−k = πn−k(MSpin) = ΩSpin n−k(pt). (A.46) (A.46) In passing from the ﬁrst to the second line we used that [ΣX, Y ] = [X, ΩY ] and then that ΩΣX = X, with ΩX the loop space, while in the last step we used the Pontrjagin-Thom isomorphism. Combining this with the Splitting Lemma (2.7), one gets (A.36). – 45 – B Tables of cobordism groups n ΩSpin n (S1) ΩSpin n (S2) ΩSpin n (S3) ΩSpin n (S4) ΩSpin n (S5) 0 Z Z Z Z Z 1 Z2 ⊕Z Z2 Z2 Z2 Z2 2 2Z2 Z2 ⊕Z Z2 Z2 Z2 3 Z2 Z2 Z 0 0 4 Z Z2 ⊕Z Z2 ⊕Z 2Z Z 5 Z 0 Z2 Z2 Z 6 0 Z 0 Z2 Z2 7 0 0 Z 0 Z2 8 2Z 2Z 2Z 3Z 2Z 9 2Z2 ⊕2Z 2Z2 2Z2 2Z2 2Z2 ⊕Z 10 5Z2 3Z2 ⊕2Z 3Z2 3Z2 3Z2 Table 20. Spin cobordism groups of spheres ΩSpin n (Sk), k = 1, . . . , 5. JHEP03(2023)181 Table 20. Spin cobordism groups of spheres ΩSpin n (Sk), k = 1, . . . , 5. n ΩSpin n (S6) ΩSpin n (S7) ΩSpin n (S8) ΩSpin n (S9) ΩSpin n (S10) 0 Z Z Z Z Z 1 Z2 Z2 Z2 Z2 Z2 2 Z2 Z2 Z2 Z2 Z2 3 0 0 0 0 0 4 Z Z Z Z Z 5 0 0 0 0 0 6 Z 0 0 0 0 7 Z2 Z 0 0 0 8 Z2 ⊕2Z Z2 ⊕2Z 3Z 2Z 2Z 9 2Z2 3Z2 3Z2 2Z2 ⊕Z 2Z2 10 3Z2 ⊕Z 3Z2 4Z2 4Z2 3Z2 ⊕Z Table 21. Spin cobordism groups of spheres ΩSpin n (Sk), k = 6, . . . , 10. Table 21. Spin cobordism groups of spheres ΩSpin n (Sk), k = 6, . . . , 10. – 46 – n ΩSpinc n (S1) ΩSpinc n (S2) ΩSpinc n (S3) ΩSpinc n (S4) ΩSpinc n (S5) 0 Z Z Z Z Z 1 Z 0 0 0 0 2 Z 2Z Z Z Z 3 Z 0 Z 0 0 4 2Z 3Z 2Z 3Z 2Z 5 2Z 0 Z 0 Z 6 2Z 4Z 2Z 3Z 2Z 7 2Z 0 2Z 0 Z 8 4Z 6Z 4Z 6Z 4Z 9 4Z 0 2Z 0 2Z 10 4Z ⊕Z2 8Z ⊕Z2 4Z ⊕Z2 6Z ⊕Z2 4Z ⊕Z2 Table 22. Spinc cobordism groups of spheres ΩSpinc n (Sk), k = 1, . . . , 5. JHEP03(2023)181 Table 22. Spinc cobordism groups of spheres ΩSpinc n (Sk), k = 1, . . . , 5. B Tables of cobordism groups K-groups of spheres Kn(Sk), k = 1, . . . , 8. n KO−n(S1) KO−n(S2) KO−n(S3) KO−n(S4) 0 Z ⊕Z2 Z ⊕Z2 Z 2Z 1 2Z2 Z2 Z2 ⊕Z Z2 2 Z2 Z2 ⊕Z Z2 Z2 3 Z 0 0 0 4 Z Z Z 2Z 5 0 0 Z Z2 6 0 Z Z2 Z2 7 Z Z2 Z2 0 8 Z ⊕Z2 Z ⊕Z2 Z 2Z Table 27. KO-groups of spheres KO−n(Sk), k = 1, . . . , 4. n KO−n(S5) KO−n(S6) KO−n(S7) KO−n(S8) 0 Z Z Z 2Z 1 Z2 Z2 Z2 ⊕Z 2Z2 2 Z2 Z2 ⊕Z 2Z2 2Z2 3 Z Z2 Z2 0 4 Z ⊕Z2 Z ⊕Z2 Z 2Z 5 Z2 0 Z 0 6 0 Z 0 0 7 Z 0 0 0 8 Z Z Z 2Z . KO-groups of spheres KO−n(Sk), k = 6, . . . , 8. Note that Bott periodicity is respected old way, i.e. KO−n(Sk) = KO−n±8(Sk) = KO−n(Sk+8). −n(T 2) K−n(T 3) K−n(T 4) K−n(T 5) K−n(T 6) K−n(T 7) K−n(T 8) 2Z 4Z 8Z 16Z 32Z 64Z 128Z 2Z 4Z 8Z 16Z 32Z 64Z 128Z C Tables of K- and KO-theory groups C y g p n K−n(S1) K−n(S2) K−n(S3) K−n(S4) K−n(S5) K−n(S6) K−n(S7) K−n(S8) 0 Z 2Z Z 2Z Z 2Z Z 2Z 1 Z 0 Z 0 Z 0 Z 0 Table 26. K-groups of spheres Kn(Sk), k = 1, . . . , 8. n KO−n(S1) KO−n(S2) KO−n(S3) KO−n(S4) 0 Z ⊕Z2 Z ⊕Z2 Z 2Z 1 2Z2 Z2 Z2 ⊕Z Z2 2 Z2 Z2 ⊕Z Z2 Z2 3 Z 0 0 0 4 Z Z Z 2Z 5 0 0 Z Z2 6 0 Z Z2 Z2 7 Z Z2 Z2 0 8 Z ⊕Z2 Z ⊕Z2 Z 2Z Table 27. KO-groups of spheres KO−n(Sk), k = 1, . . . , 4. n K−n(S1) K−n(S2) K−n(S3) K−n(S4) K−n(S5) K−n(S6) K−n(S7) K−n(S8) 0 Z 2Z Z 2Z Z 2Z Z 2Z 1 Z 0 Z 0 Z 0 Z 0 Table 26. K-groups of spheres Kn(Sk), k = 1, . . . , 8. n K−n(S1) K−n(S2) K−n(S3) K−n(S4) K−n(S5) K−n(S6) K−n(S7) K−n(S8) 0 Z 2Z Z 2Z Z 2Z Z 2Z 1 Z 0 Z 0 Z 0 Z 0 Table 26. K-groups of spheres Kn(Sk), k = 1, . . . , 8. Table 26. B Tables of cobordism groups n ΩSpinc n (S6) ΩSpinc n (S7) ΩSpinc n (S8) ΩSpinc n (S9) ΩSpinc n (S10) 0 Z Z Z Z Z 1 0 0 0 0 0 2 Z Z Z Z Z 3 0 0 0 0 0 4 2Z 2Z 2Z 2Z 2Z 5 0 0 0 0 0 6 3Z 2Z 2Z 2Z 2Z 7 0 Z 0 0 0 8 5Z 4Z 5Z 4Z 4Z 9 0 Z 0 Z 0 10 6Z ⊕Z2 4Z ⊕Z2 5Z ⊕Z2 4Z ⊕Z2 5Z ⊕Z2 Table 23. Spinc cobordism groups of spheres ΩSpinc n (Sk), k = 6, . . . , 10. Table 23. Spinc cobordism groups of spheres ΩSpinc n (Sk), k = 6, . . . , 10. – 47 – n ΩSpin n (T 2) ΩSpinc n (T 2) 0 Z Z 1 2Z ⊕Z2 2Z 2 Z ⊕3Z2 2Z 3 3Z2 2Z 4 Z2 ⊕Z 3Z 5 2Z 4Z 6 Z 4Z 7 0 4Z 8 2Z 6Z 9 2Z2 ⊕4Z 8Z 10 7Z2 ⊕2Z 8Z ⊕Z2 Table 24. Cobordism groups of 2-torus, ΩSpin n (T 2), ΩSpinc n (T 2). n ΩSpin n (T 2) ΩSpinc n (T 2) 0 Z Z 1 2Z ⊕Z2 2Z 2 Z ⊕3Z2 2Z 3 3Z2 2Z 4 Z2 ⊕Z 3Z 5 2Z 4Z 6 Z 4Z 7 0 4Z 8 2Z 6Z 9 2Z2 ⊕4Z 8Z 10 7Z2 ⊕2Z 8Z ⊕Z2 Table 24. Cobordism groups of 2-torus, ΩSpin n (T 2), ΩSpinc n (T 2). JHEP03(2023)181 Table 24. Cobordism groups of 2-torus, ΩSpin n (T 2), ΩSpinc n (T 2). Table 24. Cobordism groups of 2-torus, ΩSpin n (T 2), ΩSpinc n (T 2). n ΩSpinc n (K3) ΩSpinc n (CY3) 0 Z Z 1 0 0 2 23Z (b2 + 1)Z 3 0 b3Z 4 25Z (2 + 2b2)Z 5 0 b3Z 6 47Z (3 + 3b2)Z 7 0 2b3Z 8 50Z (5 + 4b2)Z 9 0 2b3Z 10 94Z ⊕Z2 (6 + 6b2)Z ⊕Z2 Table 25. Spinc cobordism groups of CY manifolds, ΩSpinc n (K3), ΩSpinc n (CY3). Table 25. Spinc cobordism groups of CY manifolds, ΩSpinc n (K3), ΩSpinc n (CY3). – 48 – bles of K- and KO-theory groups n(S1) K−n(S2) K−n(S3) K−n(S4) K−n(S5) K−n(S6) K−n(S7) K−n(S8) Z 2Z Z 2Z Z 2Z Z 2Z Z 0 Z 0 Z 0 Z 0 Table 26. B Tables of cobordism groups K-groups of spheres Kn(Sk), k = 1, . . . , 8. n KO−n(S1) KO−n(S2) KO−n(S3) KO−n(S4) 0 Z ⊕Z2 Z ⊕Z2 Z 2Z 1 2Z2 Z2 Z2 ⊕Z Z2 2 Z2 Z2 ⊕Z Z2 Z2 3 Z 0 0 0 4 Z Z Z 2Z 5 0 0 Z Z2 6 0 Z Z2 Z2 7 Z Z2 Z2 0 8 Z ⊕Z2 Z ⊕Z2 Z 2Z Table 27. KO-groups of spheres KO−n(Sk), k = 1, . . . , 4. JHEP03(2023)181 Table 27. KO-groups of spheres KO−n(Sk), k = 1, . . . , 4. Table 27. KO-groups of spheres KO−n(Sk), k = 1, . . . , 4. n KO−n(S5) KO−n(S6) KO−n(S7) KO−n(S8) 0 Z Z Z 2Z 1 Z2 Z2 Z2 ⊕Z 2Z2 2 Z2 Z2 ⊕Z 2Z2 2Z2 3 Z Z2 Z2 0 4 Z ⊕Z2 Z ⊕Z2 Z 2Z 5 Z2 0 Z 0 6 0 Z 0 0 7 Z 0 0 0 8 Z Z Z 2Z Table 28. KO-groups of spheres KO−n(Sk), k = 6, . . . , 8. Note that Bott periodicity is respected in a two-fold way, i.e. KO−n(Sk) = KO−n±8(Sk) = KO−n(Sk+8). Table 28. KO-groups of spheres KO−n(Sk), k = 6, . . . , 8. Note that Bott periodicity is respected in a two-fold way, i.e. KO−n(Sk) = KO−n±8(Sk) = KO−n(Sk+8). Table 28. KO-groups of spheres KO−n(Sk), k = 6, . . . , 8. Note that Bott periodicity is respect in a two-fold way, i.e. KO−n(Sk) = KO−n±8(Sk) = KO−n(Sk+8). n K−n(T 2) K−n(T 3) K−n(T 4) K−n(T 5) K−n(T 6) K−n(T 7) K−n(T 8) 0 2Z 4Z 8Z 16Z 32Z 64Z 128Z 1 2Z 4Z 8Z 16Z 32Z 64Z 128Z Table 29. K-groups of tori Kn(T k), k = 2, . . . , 8. Table 29. K-groups of tori Kn(T k), k = 2, . . . , 8. – 49 – n KO−n(T 2) KO−n(T 3) KO−n(T 4) KO−n(T 5) 0 Z ⊕3Z2 Z ⊕6Z2 2Z ⊕10Z2 6Z ⊕15Z2 1 3Z2 Z ⊕4Z2 4Z ⊕5Z2 10Z ⊕6Z2 2 Z ⊕Z2 3Z ⊕Z2 6Z ⊕Z2 10Z ⊕Z2 3 2Z 3Z 4Z 6Z 4 Z Z 2Z 6Z ⊕Z2 5 0 Z 4Z ⊕Z2 10Z ⊕6Z2 6 Z 3Z ⊕Z2 6Z ⊕5Z2 10Z ⊕15Z2 7 2Z ⊕Z2 3Z ⊕4Z2 4Z ⊕10Z2 6Z ⊕20Z2 Table 30. KO-groups of tori KO−n(T k), k = 2, . B Tables of cobordism groups . . , 5. JHEP03(2023)181 Table 30. KO-groups of tori KO−n(T k), k = 2, . . . , 5. n KO−n(T 6) KO−n(T 7) KO−n(T 8) 0 16Z ⊕21Z2 36Z ⊕28Z2 72Z ⊕36Z2 1 20Z ⊕7Z2 36Z ⊕8Z2 64Z ⊕10Z2 2 16Z ⊕Z2 28Z ⊕2Z2 56Z ⊕10Z2 3 12Z ⊕Z2 28Z ⊕8Z2 64Z ⊕36Z2 4 16Z ⊕7Z2 36Z ⊕28Z2 72Z ⊕84Z2 5 20Z ⊕21Z2 36Z ⊕56Z2 64Z ⊕126Z2 6 16Z ⊕35Z2 28Z ⊕70Z2 56Z ⊕126Z2 7 12Z ⊕35Z2 28Z ⊕56Z2 64Z ⊕84Z2 Table 31. KO-groups of tori KO−n(T k), k = 6, 7, 8. Table 31. KO-groups of tori KO−n(T k), k = 6, 7, 8. n K−n(K3) K−n(CY3) 0 24Z (2 + 2b2)Z 1 0 b3Z Table 32. K-groups of CY manifolds, K−n(K3) and K−n(CY3). Table 32. K-groups of CY manifolds, K−n(K3) and K−n(CY3). n 0 1 2 3 4 5 6 7 KO−n(K3) Z ⊕e(22Z2, Z) Z2 Z2 ⊕22Z 0 2Z Z2 Z2 ⊕22Z 22Z2 Table 33. KO-groups of K3, KO−n(K3). Table 33. KO-groups of K3, KO−n(K3). Open Access. This article is distributed under the terms of the Creative Commons Attribution License (CC-BY 4.0), which permits any use, distribution and reproduction in any medium, provided the original author(s) and source are credited. SCOAP3 supports the goals of the International Year of Basic Sciences for Sustainable Development. – 50 – References [1] E. Palti, The Swampland: Introduction and Review, Fortsch. Phys. 67 (2019) 1900037 [arXiv:1903.06239] [INSPIRE]. [1] E. Palti, The Swampland: Introduction and Review, Fortsch. Phys. 67 (2019) 1900037 [arXiv:1903.06239] [INSPIRE]. [2] M. van Beest, J. Calderón-Infante, D. Mirfendereski and I. Valenzuela, Lectures on the Swampland Program in String Compactiﬁcations, Phys. Rept. 989 (2022) 1 [arXiv:2102.01111] [INSPIRE]. [3] D. Gaiotto, A. Kapustin, N. Seiberg and B. Willett, Generalized Global Symmetries, JHEP 02 (2015) 172 [arXiv:1412.5148] [INSPIRE]. [4] B. Heidenreich et al., Chern-Weil global symmetries and how quantum gravity avoids them, JHEP 11 (2021) 053 [arXiv:2012.00009] [INSPIRE]. JHEP03(2023)181 JHEP03(2023)181 [5] B. Heidenreich et al., Non-invertible global symmetries and completeness of the spectrum, JHEP 09 (2021) 203 [arXiv:2104.07036] [INSPIRE]. [6] J. McNamara and C. Vafa, Cobordism Classes and the Swampland, arXiv:1909.10355 [INSPIRE]. [7] I. García Etxebarria, M. Montero, K. Sousa and I. Valenzuela, Nothing is certain in string compactiﬁcations, JHEP 12 (2020) 032 [arXiv:2005.06494] [INSPIRE]. [8] M. Montero and C. Vafa, Cobordism Conjecture, Anomalies, and the String Lamppost Principle, JHEP 01 (2021) 063 [arXiv:2008.11729] [INSPIRE]. [9] M. Dierigl and J.J. Heckman, Swampland cobordism conjecture and non-Abelian duality groups, Phys. Rev. D 103 (2021) 066006 [arXiv:2012.00013] [INSPIRE]. [10] Y. Hamada and C. Vafa, 8d supergravity, reconstruction of internal geometry and the Swampland, JHEP 06 (2021) 178 [arXiv:2104.05724] [INSPIRE]. [11] A. Debray, M. Dierigl, J.J. Heckman and M. Montero, The anomaly that was not meant IIB, arXiv:2107.14227 [LMU-ASC 24/21] [DOI:10.1002/prop.202100168] [INSPIRE]. [12] J. McNamara, Gravitational Solitons and Completeness, arXiv:2108.02228 [INSPIRE]. [13] A. Bedroya, Y. Hamada, M. Montero and C. Vafa, Compactness of brane moduli and the String Lamppost Principle in d > 6, JHEP 02 (2022) 082 [arXiv:2110.10157] [INSPIRE]. [14] H. Sati and U. Schreiber, M/F-Theory as Mf-Theory, arXiv:2103.01877 [INSPIRE]. [15] D. Fiorenza, H. Sati and U. Schreiber, Twisted Cohomotopy implies M-theory anomaly cancellation on 8-manifolds, Commun. Math. Phys. 377 (2020) 1961 [arXiv:1904.10207] [INSPIRE]. [16] H. Sati and U. Schreiber, Equivariant Cohomotopy implies orientifold tadpole cancellation, J. Geom. Phys. 156 (2020) 103775 [arXiv:1909.12277] [INSPIRE]. [17] H. Sati and U. Schreiber, Twisted cohomotopy implies M5-brane anomaly cancellation, Lett. Math. Phys. 111 (2021) 120 [arXiv:2002.07737] [INSPIRE]. [18] G. Buratti, M. Delgado and A.M. Uranga, Dynamical tadpoles, stringy cobordism, and the SM from spontaneous compactiﬁcation, JHEP 06 (2021) 170 [arXiv:2104.02091] [INSPIRE]. [19] G. Buratti, J. Calderón-Infante, M. Delgado and A.M. Uranga, Dynamical Cobordism and Swampland Distance Conjectures, JHEP 10 (2021) 037 [arXiv:2107.09098] [INSPIRE]. [20] R. 0] R. Angius et al., At the end of the world: Local Dynamical Cobordism, JHEP 06 (2022) 142 [arXiv:2203.11240] [INSPIRE]. References Angius et al., At the end of the world: Local Dynamical Cobordism, JHEP 06 (2022) 142 [arXiv:2203.11240] [INSPIRE]. – 51 – [21] R. Blumenhagen, N. Cribiori, C. Kneissl and A. Makridou, Dynamical cobordism of a domain wall and its companion defect 7-brane, JHEP 08 (2022) 204 [arXiv:2205.09782] [INSPIRE]. [22] R. Angius, M. Delgado and A.M. Uranga, Dynamical Cobordism and the beginning of time: supercritical strings and tachyon condensation, JHEP 08 (2022) 285 [arXiv:2207.13108] [INSPIRE]. [23] D. Andriot, N. Carqueville and N. Cribiori, Looking for structure in the cobordism conjecture, SciPost Phys. 13 (2022) 071 [arXiv:2204.00021] [INSPIRE]. [24] E. Witten, D-branes and K-theory, JHEP 12 (1998) 019 [hep-th/9810188] [INSPIRE]. [25] P.E. Conner and E.E. Floyd, The relation of cobordism to k-theories, Lectures Notes in Mathematics 28, Springer, Berlin (1966) [DOI:10.1007/BFb0071091]. JHEP03(2023)181 JHEP03(2023)181 [26] M.J. Hopkins and M.A. Hovey, Spin cobordism determines real k-theory, Math. Z. 210 (1992) 181. [27] R. Blumenhagen and N. Cribiori, Open-closed correspondence of K-theory and cobordism, JHEP 08 (2022) 037 [arXiv:2112.07678] [INSPIRE]. [28] A.M. Uranga, D-brane probes, RR tadpole cancellation and K-theory charge, Nucl. Phys. B 598 (2001) 225 [hep-th/0011048] [INSPIRE]. [29] R. Blumenhagen, M. Brinkmann and A. Makridou, A Note on the dS Swampland Conjecture, Non-BPS Branes and K-Theory, Fortsch. Phys. 67 (2019) 1900068 [arXiv:1906.06078] [INSPIRE]. [30] C. Damian and O. Loaiza-Brito, Some remarks on the dS conjecture, ﬂuxes and K-theory in IIB toroidal compactiﬁcations, arXiv:1906.08766 [INSPIRE]. [31] I. García-Etxebarria, Anomalies and dimensional reduction, talk at String Phenomenology 2019: Anomalies in various dimensions, CERN, June 28, 2019. [32] J. Davis and P. Kirk, Lecture Notes in Algebraic Topology, American Mathematical Society, Providence, R.I. (2001) [DOI:10.1090/gsm/035] [INSPIRE]. [33] A. Hatcher, Algebraic topology, Cambridge University Press, Cambridge (2000). [34] I. García-Etxebarria and M. Montero, Dai-Freed anomalies in particle physics, JHEP 08 (2019) 003 [arXiv:1808.00009] [INSPIRE]. [35] D.-E. Diaconescu, G.W. Moore and E. Witten, E8 gauge theory, and a derivation of K theory from M theory, Adv. Theor. Math. Phys. 6 (2003) 1031 [hep-th/0005090] [INSPIRE]. [36] I. Brunner and J. Distler, Torsion D-branes in nongeometrical phases, Adv. Theor. Math. Phys. 5 (2002) 265 [hep-th/0102018] [INSPIRE]. [37] I. Brunner, J. Distler and R. Mahajan, Return of the torsion D-branes, Adv. Theor. Math. Phys. 5 (2002) 311 [hep-th/0106262] [INSPIRE]. [38] J.M. Maldacena, G.W. Moore and N. Seiberg, D-brane instantons and K theory charges, JHEP 11 (2001) 062 [hep-th/0108100] [INSPIRE]. [39] K. Olsen and R.J. Szabo, Constructing D-branes from K theory, Adv. References Theor. Math. Phys. 3 (1999) 889 [hep-th/9907140] [INSPIRE]. [40] J. Evslin, What does(n’t) K-theory classify?, hep-th/0610328 [INSPIRE]. [41] P. Horava, Type IIA D-branes, K theory, and matrix theory, Adv. Theor. Math. Phys. 2 (1999) 1373 [hep-th/9812135] [INSPIRE]. – 52 – [42] G.W. Moore and E. Witten, Selfduality, Ramond-Ramond ﬁelds, and K theory, JHEP 05 (2000) 032 [hep-th/9912279] [INSPIRE]. [43] D.S. Freed and M.J. Hopkins, On Ramond-Ramond ﬁelds and K theory, JHEP 05 (2000) 044 [hep-th/0002027] [INSPIRE]. [44] M.F. Atiyah, R. Bott and A. Shapiro, Cliﬀord modules, Topology 3 (1964) S3–S38 [INSPIRE]. 5] M. Kreck and S. Stolz, HP 2-bundles and elliptic homology, Acta Math. 171 (1993) 231. [46] N. Hitchin, Harmonic Spinors, Adv. Math. 14 (1974) 1 [INSPIRE]. [47] J. McCleary, A User's Guide to Spectral Sequences, Cambridge University Press (2000). JHEP03(2023)181 [48] D. Husemöller, M. Joachim, B. Jurčo and M. Schottenloher, The atiyah-hirzebruch spectral sequence in k-theory, in Basic Bundle Theory and K-Cohomology Invariants: With contributions by Siegfried Echterhoﬀ, Stefan Fredenhagen and Bernhard Krötz, Springer Berlin, Heidelberg (2008) [DOI:10.1007/978-3-540-74956-1]. [49] D. Arlettaz, The order of the diﬀerentials in the Atiyah-Hirzebruch spectral sequence, K-Theory 6 (1992) 347. [50] M.F. Atiyah and F. Hirzebruch, Vector bundles and homogeneous spaces, Proceedings of Symposia in Pure Mathematics 3, American Mathematical Society (1961), p. 7 [DOI:10.1090/pspum/003/0139181]. [51] D.S. Freed and E. Witten, Anomalies in string theory with D-branes, Asian J. Math. 3 (1999) 819 [hep-th/9907189] [INSPIRE]. [52] C. Doran and J.W. Morgan, Algebraic topology of Calabi-Yau threefolds in toric varieties, Geom. Topol. 11 (2007) 597. [53] S. Gukov, K-theory and D-branes on noncommutative tori, in the proceedings of NATO Advanced Study Institute: TMR Summer School on Progress in String Theory and M-Theory, Cargese France, May 24-June 5 1999, pp. 339–342 [INSPIRE]. [54] D.W. Anderson, E.H. Brown and F.P. Peterson, The structure of the spin cobordism ring, Annals Math. 86 (1967) 271. [55] D. Grady and H. Sati, Diﬀerential KO-theory: constructions, computations, and applications, Adv. Math. 384 (2021) 107671 [arXiv:1809.07059] [INSPIRE]. [56] J.W. Milnor and J.D. Stasheﬀ, Characteristic Classes, Princeton University Press and University of Tokyo Press, Princeton, New Jersey (1974) [DOI:10.1515/9781400881826]. [57] Y. Rudyak, On Thom Spectra, Orientability, and Cobordism, Springer Monographs in Mathematics, Springer (1998) [DOI:10.1007/978-3-540-77751-9]. [58] Y. Rudyak, Orientation of manifolds in generalised cohomology theories — deﬁnition, Bulletin of Manifold Atlass (2014). [59] A. Dabholkar and J. Park, An Orientifold of type IIB theory on K3, Nucl. Phys. References B 472 (1996) 207 [hep-th/9602030] [INSPIRE]. [60] E.G. Gimon and C.V. Johnson, K3 orientifolds, Nucl. Phys. B 477 (1996) 715 [hep-th/9604129] [INSPIRE]. [61] V. Braun, K theory torsion, hep-th/0005103 [HU-EP-00-24] [INSPIRE]. [62] F. Marchesano, D6-branes and torsion, JHEP 05 (2006) 019 [hep-th/0603210] [INSPIRE]. – 53 – [63] S. Fredenhagen and V. Schomerus, Branes on group manifolds, gluon condensates, and twisted K theory, JHEP 04 (2001) 007 [hep-th/0012164] [INSPIRE]. [64] U. Bunke and T. Schick, Diﬀerential K-theory. A survey, arXiv:1011.6663 [INSPIRE]. [65] U. Bunke, Diﬀerential cohomology, arXiv:1208.3961. [66] D. Freed, Lectures on twisted k-theory and orientifolds, in “K-Theory and Quantum Fields” workshop, Erwin Schrödinger International Institute for Mathematical Physics, Vienna, Austria (2012). [67] C. Weibel, An Introduction to Homological Algebra, Graduate Studies in Mathematics, Cambridge University Press (1994) [DOI:10.1017/cbo9781139644136]. JHEP03(2023)181 [68] A. Beaudry and J.A. Campbell, A Guide for Computing Stable Homotopy Groups, arXiv:1801.07530. – 54 –
https://openalex.org/W3037956187	https://discovery.ucl.ac.uk/10103546/1/acsomega.0c00413.pdf	English	null	Iron-Intercalated Zirconium Diselenide Thin Films from the Low-Pressure Chemical Vapor Deposition of [Fe(η<sup>5</sup>-C<sub>5</sub>H<sub>4</sub>Se)<sub>2</sub>Zr(η<sup>5</sup>-C<sub>5</sub>H<sub>5</sub>)<sub>2</sub>]<sub>2</sub>	ACS omega	2,020	cc-by	6,313	■INTRODUCTION post-transition20 metals have been intercalated to modulate electronic states and dramatically change the opto-electronic and magnetic properties of the host material. In some cases, a relatively low intercalation produces a signiﬁcant change while virtually retaining the host structure.13 Transition metal dichalcogenide (TMD) layered materials have attracted a great deal of interest due to their unique physical properties beyond those of graphene as they are generally non-toxic and abundant, have energy band gaps well- suited for solar energy conversion and high absorption coeﬃcients in the visible range.1 TMDs of group 4, 5, and 6 metals are diamagnetic semiconductors with band gap values in the range of 0.2−2.0 eV.2,3 They have received special interest due to their mechanical and electrical anisotropy2,4 and charge density wave transitions, related to their structures being composed of layers of TX2 (T = transition metal; X = chalcogen) stack along the c axis via van der Waals forces.2,5 Zirconium diselenide (ZrSe2) in particular has attracted considerable attention in the ﬁelds of physics, chemistry, and materials due to their unique structural, optical,6,7 and electronic properties. This is an open access article published under a Creative Commons Attribution (CC-BY) License, which permits unrestricted use, distribution and reproduction in any medium, provided the author and source are cited. This is an open access article published under a Creative Commons Attribution (CC-BY) License, which permits unrestricted use, distribution and reproduction in any medium, provided the author and source are cited. Downloaded via 151.231.178.69 on July 2, 2020 at 08:11:30 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles. ABSTRACT: Transition metal chalcogenide thin ﬁlms of the type FexZrSe2 have applications in electronic devices, but their use is limited by current synthetic techniques. Here, we demonstrate the synthesis and characterization of Fe-intercalated ZrSe2 thin ﬁlms on quartz substrates using the low-pressure chemical vapor deposition of the single-source precursor [Fe(η5-C5H4Se)2Zr(η5-C5H5)2]2. Powder X-ray diﬀraction of the ﬁlm scraping and subsequent Rietveld reﬁnement of the data showed the successful synthesis of the Fe0.14ZrSe2 phase, along with secondary phases of FeSe and ZrO2. Upon intercalation, a small optical band gap enhancement (Eg(direct) opt = 1.72 eV) is detected in comparison with that of the host material. © XXXX American Chemical Society © XXXX American Chemical Society Iron-Intercalated Zirconium Diselenide Thin Films from the Low- Pressure Chemical Vapor Deposition of [Fe(η5‑C5H4Se)2Zr(η5‑C5H5)2]2 Iron-Intercalated Zirconium Diselenide Thin Films from the Low- Pressure Chemical Vapor Deposition of [Fe(η5‑C5H4Se)2Zr(η5‑C5H5)2]2 Clara Sanchez-Perez, Caroline E. Knapp, Ross H. Colman, Carlos Sotelo-Vazquez, Sanjayan Sathasivam, Raija Oilunkaniemi, Risto S. Laitinen,* and Claire J. Carmalt* Article Recommendations * Supporting Information https://dx.doi.org/10.1021/acsomega.0c00413 ACS Omega XXXX, XXX, XXX−XXX Received: January 29, 2020 Accepted: June 10, 2020 ■RESULTS AND DISCUSSION However, small additional crystalline secondary phases of α- and γ-Fe, hexagonal FeSe, and ZrO2 were also observed. The secondary phase formation is likely due to a combination of post-deposition oxidation (semi-amorphous nature of ZrO2) during post-deposition ﬁlm handling and due to possible Fe saturation in between the layers (e.g., FeSe and α- and γ-Fe) (Figure 2). The parent compound ZrSe2 is known to crystallize in the CdI2 structure type with P3̅m1 symmetry, identical to that of TiSe2, as shown from the Rietveld analysis. The structure exists as sheets of Zr atoms, octahedrally coordinated by Se atoms, with each Se bonded to three Zr atoms. The two- dimensional ZrSe2 sheets are AA stacked and weakly bound, with the space between sheets commonly referred to as the van der Waals gap. Structural modiﬁcations to the parent ZrSe2 lattice upon the intercalation of variable amounts of Fe within the van der Waals gap have previously shown a change of symmetry due to intercalant site ordering at x ≥0.41 (in FexZrSe2), along with the onset of metallicity.13 In addition to the impurity phases, the two-dimensionality of the material (stacked layers) leads to the expected preferred orientation eﬀects that were modeled within the reﬁnement (see Supporting Information, Table S1 and Figure S1 for further reﬁnement details). The FexZrSe2 material shows no sign of superstructure and was best modeled using the parent P3̅m1 space group, indicating no site ordering of intercalated Fe Following our previous work on FexTiSe2 thin ﬁlms via LPCVD from the single-source precursor [Fe(η5-C5H4Se)2Ti- (η5-C5H5)2]2,34 FexZrSe2 thin ﬁlm synthesis from [Fe(η5- C5H4Se)2Zr(η5-C5H5)2]2 (1) (Figure 1a) was investigated. The highly reﬂective and black FexZrSe2 ﬁlms grown using LPCVD under a dynamic pressure of 10−1 Torr at 1000 °C were air/moisture-sensitive but highly robust with complete coverage of the quartz substrate. Scanning electron microscopy (SEM) analysis of the ﬁlms showed that the surface morphology consisted of irregular quasi-spherical grains uniformly distributed over the entire surface of the substrate (Figure 1b,c), in a similar way as previously reported for ZrSe2 thin ﬁlms.28 The ﬁlm is fairly densely packed both laterally and vertically, and the cross-sectional SEM imaging of the ﬁlm showed a thickness of ∼300 nm (Figure 1c, inset). ( g ) Glancing-angle X-ray diﬀraction (XRD) analysis of the FexZrSe2 ﬁlms was not possible due to ﬂuorescence of iron in the presence of the Cu Kα1 radiation. ■INTRODUCTION While the use of ZrSe2 in secondary lithium batteries confronts electrochemical challenges,8,9 it shows many potential applications as working materials in photovoltaic solar cells10,11 given that its band gap falls in the visible region of the electromagnetic spectrum (Eg(indirect) opt = 1.20 eV; Eg(direct) opt = 1.61 eV).6,12 Group 4 (d0) TMD such as ZrSe2 stabilize 1T structures under normal conditions, with a CdI2 (HCP packing) structure and P3̅m1 symmetry.13 This structure type allows the intercalation of small molecules or atoms (electron donors) between TX2 layers,5 which transfer electrons from the intercalate to the transition metal unoccupied d band.4 The intercalation of electron-donating atoms in TMD layered materials can be used to modulate the intrinsic physical properties of the host material4 that result into band gap changes,14 phase transition processes,15,16 and development of magnetic and superconducting behaviors.2,4,5 In the case of ZrSe2 diﬀerent alkali 17 transition15,16,18,19 and ZrSe2 is one of the few metal chalcogenide hosts able to accommodate Fe both into tetrahedral and octahedral sites between layers,15 with the Fe relative site distribution dependent on the total amount of Fe and the thermal treatment of the material.16 At intercalation levels x = 0−0.16, a semiconducting FexZrSe2 phase exists with a trigonal structure P3̅m1 derived from that of ZrSe2, intermediate between 1T-CdI2 and NiAs types, and at the critical iron concentration of x = 0.41, the material becomes metallic.13 Hybridization of the 3d bands of Zr and Fe, along with the overlap of their dz 2 electron shells along the c axis leads to a covalent interaction between ZrSe2 layers, ultimately aﬀecting the unit cell lattice constants.19,21 As the intercalation of transition metals in TMD 1T polymorphs donates free carriers to their empty conduction band, increasing their conductiv- ity,4,22 they can be selectively used to ﬁne-tune the electric, magnetic, optical, and structural properties of the host materials in a controllable way to ﬁt application requirements.5 ZrSe2 and FexZrSe2 materials have traditionally been synthesized from high-purity elements8,13,15,16,23 (long multi- step processes at high temperatures) or chemical vapor step processes at high temperatures) or chemical vapor Received: January 29, 2020 Accepted: June 10, 2020 development of magnetic and superconducting behaviors.2,4,5 In the case of ZrSe2 diﬀerent alkali,17 transition15,16,18,19 and © XXXX American Chemical Society © XXXX American Chemical Society © XXXX American Chemical Society https://dx.doi.org/10.1021/acsomega.0c00413 ACS Omega XXXX, XXX, XXX−XXX doi.org/10.1021/acsomega.0c00413 ACS Omega XXXX, XXX, XXX−XXX A ACS Omega http://pubs.acs.org/journal/acsodf Figure 1. ■INTRODUCTION This method allows growth of high-quality thin ﬁlms from volatile, temperature-sensitive reagents and has been used extensively in the past decade for the synthesis of thin ﬁlms of functional materials, many of them metal selenides.32−34 In the present paper, we investigate the ability of single-source precursor [Fe(η5-C5H4Se)2Zr(η5-C5H5)2]2 (1) to successfully generate thin ﬁlms of Fe-intercalated ZrSe2 (i.e., FexZrSe2) using a convenient single-step deposition process via LPCVD and record the optical band gap shift resultant from the intercalation. Figure 2. Rietveld reﬁned powder diﬀraction data (crosses), ﬁt (red line), diﬀerence curve (shifted blue line), and reﬂection markers for all phases (ticks).13,35−38 Structural analysis was performed using the GSAS package,39 and reference patterns from ICSD were used. Embedded in the right corner is an image of the FexZrSe2 structure drawn from structurally reﬁned PXRD data using VESTA,40 showing Zr and Se atoms (gray and black spheres) making up edge-linked layers of ZrSe2 octahedra in the ab plane, separated by Fe atoms (gold spheres) randomly occupying the octahedral vacancies within the van der Waals gap with a reﬁned occupancy of ∼14%. ■INTRODUCTION (a) Schematic representation of precursor [Fe(η5-C5H4Se)2Zr(η5-C5H5)2]2 used in LPCVD to grow FexZrSe2 thin ﬁlms. (b, c) SEM images of the thin ﬁlm at x10k and x40k magniﬁcation, respectively. The inset shows the cross-sectional image to reveal a ﬁlm thickness of ca. 300 nm. Figure 1. (a) Schematic representation of precursor [Fe(η5-C5H4Se)2Zr(η5-C5H5)2]2 used in LPCVD to grow FexZrSe2 thin ﬁlms. (b, c) SEM images of the thin ﬁlm at x10k and x40k magniﬁcation, respectively. The inset shows the cross-sectional image to reveal a ﬁlm thickness of ca. 300 nm. Figure 1. (a) Schematic representation of precursor [Fe(η5-C5H4Se)2Zr(η5-C5H5)2]2 used in LPCVD to grow F images of the thin ﬁlm at x10k and x40k magniﬁcation, respectively. The inset shows the cross-sectional image to nm Figure 2. Rietveld reﬁned powder diﬀraction data (crosses), ﬁt (red line), diﬀerence curve (shifted blue line), and reﬂection markers for all phases (ticks).13,35−38 Structural analysis was performed using the GSAS package,39 and reference patterns from ICSD were used. Embedded in the right corner is an image of the FexZrSe2 structure drawn from structurally reﬁned PXRD data using VESTA,40 showing Zr and Se atoms (gray and black spheres) making up edge-linked layers of ZrSe2 octahedra in the ab plane, separated by Fe atoms (gold spheres) randomly occupying the octahedral vacancies within the van der Waals gap with a reﬁned occupancy of ∼14%. transport reactions,2,6,8,17,24 mainly as single crystals and, in few cases, in nanoparticle form.25,26 The synthesis of ZrSe2 thin ﬁlms has been mainly reported via electrodeposition methods,11,27−29 although the synthesis using MBE30 and CVD31 has been reported as an alternative for the requirement of conductive substrates, which can be problematic for device applications. The absence of studies of FexZrSe2 in thin ﬁlm form is likely due to the lack of precursors for these ternary systems using established deposition methods. Low-pressure chemical vapor deposition (LPCVD) of organometallic precursors appears to be a suitable alternative due to its versatility toward the use of alternative air/moisture-sensitive precursors. https://dx.doi.org/10.1021/acsomega.0c00413 ACS Omega XXXX, XXX, XXX−XXX ■RESULTS AND DISCUSSION XPS data for a LPCVD grown FexZrSe2 ﬁlm showing the (a) Zr 3d and (b) Se 3d transitions. Zr 3d5/2 transition centered at 182.7 eV corresponding to Zr(IV) bound to Se or O and therefore providing evidence for the presence of ZrSe2 and/or ZrO2.30,50 It has been previously reported that the intercalation of ZrSe2 (Zr 3d3/2 = 183.05 eV) with a transition metal provokes a small shift of the Zr 3d environment to a lower energy (Zr 3d3/2 = 182.6 eV),19 overlapping with the Zr environment from ZrO2 (Zr 3d3/2 = 182.3 eV).32 A further two minor sub-oxide environments were also found with the Zr 3d5/2 transitions at 178.6 and 180.3 eV.50 The Se 3d spectrum (Figure 4b) was broad and best ﬁtted with three sets of doublets with a 3d5/2-3d3/2 separation of 0.86 eV. Se 3d5/2 peaks of almost equal intensity were found at 53.8 and 56.0 eV corresponding to Se2−bound to Zr and metallic Se (Se0), respectively, therefore providing evidence for the ZrSe2-based phase also seen from the XRD data as well as elemental selenium, a secondary product ZrSe2 oxidation.30,51 The third environment with the Se 3d5/2 peak at 59.5 eV matches with Se(IV) in the form of SeO2, a possible native oxide product.51 Fe was only detected with a low signal-to- noise ratio, and therefore the spectrum was not deconvoluted to determine oxidation states (see Supporting Information, Figure S3). The structural properties of the thin ﬁlms were further investigated using Raman spectroscopy (Figure 3). The Raman measurements were carried out after only minimal exposure to air (ca. 5 s) and after 5 min of exposure in an eﬀort to study Figure 3. Raman spectra of the LPCVD grown FexZrSe2 ﬁlms after minimal (ca. 5 s) and 5 min exposure to air. Bands corresponding to the contribution of FexZrSe2 (purple), iron oxide (α-Fe2O3, green), selenium (Se, orange), and amorphous zirconia (ZrO2, red) are detected in the spectra. Compositional analysis of the surface of the ﬁlm suggests a high excess of Se and O on the surface of the ﬁlms relative to Zr (and Fe), suggesting surface segregation of the Se and native oxide formation, respectively. Depth proﬁling studies on a layer of ∼100 nm into the FexZrSe2 ﬁlms revealed the total Fe-to-Zr ratio to be 0.12. It should be noted that the elemental composition calculations of O and Se were not carried out due Figure 3. ■RESULTS AND DISCUSSION Our observed interlayer lattice parameter, c = 6.1336(3) Å, lies between the values reported for x = 0.18 (c = 6.138(2) Å) and x = 0.465 (c/2 = 6.011(1) Å).13 Care must be taken when interpreting this eﬀect, however, due to the possibility of non- stoichiometric Zr:Se ratios. Non-stoichiometry of Zr:Se has been seen to signiﬁcantly inﬂuence both the lattice parameter and electron transport properties.41 Our observed c lattice parameter is larger than the reported value for near- stoichiometric ZrSe1.937 (c = 6.1297(9) Å) while being only slightly smaller than Se-deﬁcient ZrSe1.851 (c = 6.1510(9) Å), complicating interpretation as evidence of intercalation.41 After exposure of the samples to air for 5 min, characteristic bands of α-Fe2O3 were detected44,45 together with a small contribution from amorphous selenium46,47 (since the symmetric Se−Se stretching mode shows a very intensive Raman band, this spectrum indicates that the Se content is not very large). The uneven background in these spectra suggests the existence of amorphous ZrO2 (red), which was also detected from XRD;43 however, no FeSex phases could be observed.48,49 X-ray photoelectron spectroscopy (XPS) was used to study the surface composition of the FexZrSe2 ﬁlms. Zr, Se, Fe, and O were detected in the high-resolution XPS scans. The Zr 3d spectrum (Figure 4a) was ﬁtted with a major doublet with the While the reﬁnement of structural details has been attempted, reported errors on the values obtained may vary due to the data quality and the modeling of preferred orientation and additional phases. The reﬁnements were improved upon the introduction of Fe within the van der Waals gap modeled in an identical manner to that previously seen in FexZrSe2, placing Fe directly between the layers on the 1b Wyckoﬀsite (0, 0, and 1/2) and reﬁned to a value of x = 13.5(5)%, suggesting Fe0.14ZrSe2 (Table 1). Figure 4. XPS data for a LPCVD grown FexZrSe2 ﬁlm showing the (a) Zr 3d and (b) Se 3d transitions. Table 1. Selected Parameters from Rietveld Reﬁnements of Powder X-ray Diﬀraction Data target composition FexZrSe2 space group P3̅m1 lattice parameters (Å) a = 3.77239(6); c = 6.1336(3) Fe fractional occupancy, x 0.135(5) FexZrSe2 phase fraction (wt %) 61.9(10) impurity phase fractions: h-FeSe, α-Fe, γ-Fe, c-ZrO2 (wt %) 9.11(12), 3.55(6), 5.18(9), 20.25(19) goodness of ﬁt, wRp (%), Rp (%), χ2 1.04, 0.77, 4.167 Table 1. Selected Parameters from Rietveld Reﬁnements of Powder X-ray Diﬀraction Data Figure 4. https://dx.doi.org/10.1021/acsomega.0c00413 ACS Omega XXXX, XXX, XXX−XXX ■RESULTS AND DISCUSSION Therefore, the ﬁlms were scraped oﬀto give powders that were then analyzed via powder XRD (PXRD) using a MoKα1 source in order to obtain patterns with reduced ﬂuorescence eﬀects. Rietveld reﬁnement, carried out on the data to determine the crystallographic properties of the ﬁlms (Figure 2), showed that layered ZrSe2 with intercalated Fe had indeed been formed. B https://dx.doi.org/10.1021/acsomega.0c00413 ACS Omega XXXX, XXX, XXX−XXX http://pubs.acs.org/journal/acsodf Article ACS Omega any changes due to ﬁlm oxidation. The FexZrSe2 ﬁlms showed a minor scattering band at 173 cm−1 and a major scattering band at 197 cm−1, which were respectively consistent with the Eg and A1g bands for ZrSe2-like materials with P3̅m1 symmetry.30,42 No sign of the characteristic Raman bands for amorphous or crystalline ZrO2 43 or Fe2O3 44,45 were detected from the 633 nm laser excitation. atoms and a value of the intercalation of x < 0.41.13 Changes in lattice parameters are often used as an indication of intercalation in transition metal dichalcogenides. The interlayer distance is expectedly found to increase upon the intercalation of transition metals, resulting in an increase of the c lattice parameter (e.g., FexZrSe2 13 or CuxZrSe2 19). p ( g x 2 x 2 ) Our observed interlayer lattice parameter, c = 6.1336(3) Å, lies between the values reported for x = 0.18 (c = 6.138(2) Å) and x = 0.465 (c/2 = 6.011(1) Å).13 Care must be taken when interpreting this eﬀect, however, due to the possibility of non- stoichiometric Zr:Se ratios. Non-stoichiometry of Zr:Se has been seen to signiﬁcantly inﬂuence both the lattice parameter and electron transport properties.41 Our observed c lattice parameter is larger than the reported value for near- stoichiometric ZrSe1.937 (c = 6.1297(9) Å) while being only slightly smaller than Se-deﬁcient ZrSe1.851 (c = 6.1510(9) Å), complicating interpretation as evidence of intercalation.41 While the reﬁnement of structural details has been attempted, reported errors on the values obtained may vary due to the data quality and the modeling of preferred orientation and additional phases. The reﬁnements were improved upon the introduction of Fe within the van der Waals gap modeled in an identical manner to that previously seen in FexZrSe2, placing Fe directly between the layers on the 1b Wyckoﬀsite (0, 0, and 1/2) and reﬁned to a value of x = 13.5(5)%, suggesting Fe0.14ZrSe2 (Table 1). ■CONCLUSIONS In conclusion, the formation of the FexZrSe2 (x = 0.14) thin ﬁlms via LPCVD from [Fe(η5-C5H4Se)2Zr(η5-C5H5)2]2 was veriﬁed using PXRD, XPS, and Raman spectroscopy. The morphology of the thin ﬁlms was evaluated using SEM and shown to consist of densely packed domes. Rietveld reﬁnement analysis of PXRD data of the powder from ﬁlm scrapping indicates the formation of a major phase of FexZrSe2 (x = 0.14) and small impurity phases of Fe and FeSe presumably due to excess Fe plus post-deposition oxidation product ZrO2. Raman bands show the expected A1g and Eg modes for FexZrSe2 (x = 0.14) (CdI2 structure type, P3̅m1). No bands for oxidation products were detected in freshly deposited ﬁlms, whereas bands due to oxidation/decomposition products were found in ﬁlms exposed to air for longer times. A small shift in the direct optical band gap was detected for Fe0.14ZrSe2 thin ﬁlms with respect to the host material ZrSe2. The extreme sensitivity of the FexZrSe2 materials requires a carefully designed and sealed setup for its synthesis, and therefore a single precursor oﬀers better chances of homogeneity in the samples. Although utilization of the related precursor [(tBuSe)2Se]31 could possibly be used for the synthesis of FexZrSe2 ﬁlms in combination with an iron precursor, its higher toxicity and volatility makes its use far less desirable. Additionally, the selection of an oxygen-free iron source is not straightforward, as risk of carbon contamination can increase exponentially due to the increase of burned hydrocarbonated ligands and the use of alternative carbon-free sources (e.g., FeCl3) could involve addition of undesired halogen doping. Although the proposed single-source precursor does not lead to a pure FexZrSe2 ; , LPCVD experiments were carried out in a quartz tube under dynamic vacuum (10−1 Torr) embedded inside a furnace to allow uniform heating, which was controlled using Pt−Rh thermocouples. The polycrystalline powder precursor (ca. 0.20 g) was spread evenly in a glazed aluminosilicate ceramic boat (0.9 × 1.4 × 10.3 cm, VWR cat. no. 459-0224) and heated up at 20 °C/min to 1000 °C and kept at that temperature for 1 h with the goal of achieving its sublimation over quartz slides (2.5 cm × 1.0 cm × 2 mm, Multi-Lab), which prior to use were cleaned using acetone (99%), isopropanol (99%), and distilled water and dried at 180 °C overnight. ■EXPERIMENTAL SECTION Figure 5. Absorbance (black line) of the thin ﬁlm after minimal exposure to atmospheric conditions showing a FexZrSe2 absorption band centered at ∼1000 nm and a second absorption band centered at ∼450 nm due to the formation of α-Fe2O3 52 upon exposure of the thin ﬁlm to air prior to/during measurement. Graphical representa- tion of band gap calculation using a Tauc plot is depicted in the inset. Precursors were synthesized according to literature proce- dures.53 N,N,N′,N′-Tetramethylethylenediamine “TMEDA” (Aldrich) was distilled over sodium and stored over 3 Å molecular sieves (20% m/v) for 24 h. Ferrocene (Merck, 99%) was freeze-dried for 12 h prior to use. Selenium (shot, Aldrich), nBuLi (2.5 M in hexane, Aldrich), and bis- (cyclopentadienyl)zirconium(IV) dichloride (Aldrich) were used as purchased. Dry tetrahydrofuran (99.9% in Argon, Sigma) and dry toluene were stored over a sodium mirror for 24 h prior to use, and pre-dried dichloromethane was dried over 3 Å molecular sieves (20% m/v) for 48 h prior to use. All preparations were undertaken using Schlenk techniques, and all glassware was dried for 12 h at 150 °C prior to use. The synthesis and puriﬁcation of the precursor was performed under argon, which was passed over a drying column. After isolation, the polycrystalline powder precursor was stored in a glovebox under an argon atmosphere. Synthesis and puriﬁcation of the precursor were conﬁrmed by NMR and elemental analysis.53 [Fe(η5-C5H4Se)2Zr(η5-C5H5)2]2 (1): A bright red polycrystalline powder was isolated (1.25 g, 74%). 1H NMR (600 MHz) δ/ppm (C6D6): 5.67 (s, 10H, Cp); 4.35 (m, 8H, Fc); 13C{1H} NMR (600 MHz) δ/ppm (C6D6): 110.08 (m, Fc); 113.14 (m, Cp); 77Se NMR (400 MHz) δ/ ppm (tetrahydrofuran): 558.0; elemental analysis (C40H36Se4Fe2Zr2): Calc. C, 42.64; H, 3.22. Found: C, 42.58; H, 3.35. Figure 5. Absorbance (black line) of the thin ﬁlm after minimal exposure to atmospheric conditions showing a FexZrSe2 absorption band centered at ∼1000 nm and a second absorption band centered at ∼450 nm due to the formation of α-Fe2O3 52 upon exposure of the thin ﬁlm to air prior to/during measurement. Graphical representa- tion of band gap calculation using a Tauc plot is depicted in the inset. ﬁlm was determined plotting (αhν)2 against the photon energy (hν) (blue line) and extrapolating the linear part of the plot (broken red line). ACS Omega to the error associated with preferential sputtering of these lighter elements relative to Zr and Fe during the etching process. system using the described synthetic conditions, this study represents a proof of concept that the system can be synthesized from organometallic sources; the continuous development of organometallic sources with a variable ratio of Fe is extremely important for the future viability of these mixed metal selenides in industrial processes. p The optical absorption spectrum of the FexZrSe2 thin ﬁlm was calculated from sample transmittance and is shown in Figure 5 (black line). The direct optical band gap of the thin ■EXPERIMENTAL SECTION The obtained optical band gap value (1.72 eV) shows a small band gap increase (∼0.2 eV) with respect to previously published values for ZrSe2.12,27,28 ■RESULTS AND DISCUSSION Raman spectra of the LPCVD grown FexZrSe2 ﬁlms after minimal (ca. 5 s) and 5 min exposure to air. Bands corresponding to the contribution of FexZrSe2 (purple), iron oxide (α-Fe2O3, green), selenium (Se, orange), and amorphous zirconia (ZrO2, red) are detected in the spectra. C https://dx.doi.org/10.1021/acsomega.0c00413 ACS Omega XXXX, XXX, XXX−XXX http://pubs.acs.org/journal/acsodf https://dx.doi.org/10.1021/acsomega.0c00413 ACS Omega XXXX, XXX, XXX−XXX ■ASSOCIATED CONTENT * sı Supporting Information * pp g The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.0c00413. A table with the reﬁned parameters for FexZrSe2 powder data, a high-quality image of the Rietveld reﬁned powder data with the derived FexZrSe2 structure, and XPS survey and Fe surface scans of the thin ﬁlm samples (PDF) (2) Wilson, J. A.; Yoffe, A. D. The Transition Metal Dichalcogenides Discussion and Interpretation of the Observed Optical, Electrical and Structural Properties. Adv. Phys. 1969, 18, 193−335. y (3) Yoffe, A. D. Electronic Properties of Low Dimensional Solids: The Physics and Chemistry of Layer Type Transition Metal Dichalcogenides and Their Intercalate Complexes. Solid State Ionics 1990, 39, 1−7. Corresponding Authors (5) Friend, R. H.; Yoffe, A. D. Electronic Properties of Intercalation Complexes of the Transition Metal Dichalcogenides. Adv. Phys. 1987, 36, 1−94. Risto S. Laitinen −Laboratory of Inorganic Chemistry, Environmental and Chemical Engineering, University of Oulu, Oulu FI-90014, Finland; Email: risto.laitinen@oulu.ﬁ (6) Lee, P. A.; Said, G.; Davis, R.; Lim, T. H. On the Optical Properties of Some Layer Compounds. J. Phys. Chem. Solids 1969, 30, 2719−2729. Claire J. Carmalt −Materials Chemistry Centre, Department of Chemistry, University College London, London WC1H 0AJ, U.K.; orcid.org/0000-0003-1788-6971; Email: c.j.carmalt@ucl.ac.uk (7) Zheng, Y.; Wang, M.; Zhao, R.; Zhang, H.; Liu, D.; Li, D. Nonlinear Optical Absorption Properties of Zirconium Selenide in Generating Dark Soliton and Dark-Bright Soliton Pairs. Appl. Opt. 2020, 59, 396. ■AUTHOR INFORMATION (4) Marseglia, E. A. Transition Metal Dichalcogenides and Their Intercalates. Int. Rev. Phys. Chem. 2008, 3, 177−216. ■ABBREVIATIONS TMDs, transition metal dichalcogenides; ITMDs, intercalated transition metal dichalcogenides; LPCVD, low-pressure chemical vapor deposition ■CONCLUSIONS The method was followed by a gradual cooling process involving an initial cooling step of 13 °/min until reaching 355 °C followed by a second cooling step of 2 °C/min until reaching room temperature (switched oﬀ). The tube under vacuum was then transported inside a glovebox, where the ﬁlms on both the quartz substrate and glazed aluminosilicate ceramic boat were stored for character- ization. PXRD data were collected on a STOE diﬀractometer using monochromated MoKα1 radiation (0.70903 Å, 50 kV, 30 mA) and four scans per measurement over the 2θ range of 10− 40°, with a step size of 0.5° and a count time of 10 s per step. Polycrystalline powders were ground in a metal mortar within the glovebox, loaded into sealed 3 mm borosilicate capillaries, and analyzed using a STOE Stadi P diﬀractometer (MoKα1 radiation, 0.70903 Å, 50 kV, 30 mA), in which less ﬂuorescence was detected. Rietveld reﬁnements were per- formed using GSAS and EXPGUI.39 Raman spectroscopy of both powders and ﬁlms was carried out using a Renishaw 1000 spectrometer equipped with a 633 nm laser. The Raman D https://dx.doi.org/10.1021/acsomega.0c00413 ACS Omega XXXX, XXX, XXX−XXX Article ACS Omega http://pubs.acs.org/journal/acsodf https://pubs.acs.org/10.1021/acsomega.0c00413 system was calibrated using a silicon reference. X-ray photoelectron spectroscopy (XPS) of powders and ﬁlms was performed using a Thermo K alpha spectrometer with monochromated AlKα radiation (8.3418 Å), a dual beam charge compensation system, and a constant pass energy of 50 eV. Survey scans were collected in the range of 0−1200 eV. High-resolution peaks were used for the principal peaks of Zr (3d), Fe (2p), Se (3d), and C (1s). XPS data was analyzed using CasaXPS software, and the peaks were calibrated to adventitious carbon at 284.8 eV. Scanning electron microscopy (SEM) studies of ﬁlm surfaces were carried out for thin ﬁlm samples using JEOL 6301 (10 kV) and JEOL JSM-6700F ﬁeld emission instruments after sputtering of the samples with a thin layer of gold to reduce charging. UV/Vis/near-IR transmission and reﬂection spectra were recorded in the range of 300−2500 nm using a Perkin-Elmer Fourier Transform Lambda 950 UV/Vis/NIR spectrometer. The transmission spectra background was taken against an air background, and a Labsphere reﬂectance standard was used as a reference material. ■REFERENCES (1) Bhimanapati, G. R.; Lin, Z.; Meunier, V.; Jung, Y.; Cha, J.; Das, S.; Xiao, D.; Son, Y.; Strano, M. S.; Cooper, V. R.; Liang, L.; Louie, S. G.; Ringe, E.; Zhou, W.; Kim, S. S.; Naik, R. R.; Sumpter, B. G.; Terrones, H.; Xia, F.; Wang, Y.; Zhu, J.; Akinwande, D.; Alem, N.; Schuller, J. A.; Schaak, R. E.; Terrones, M.; Robinson, J. A. Recent Advances in Two-Dimensional Materials beyond Graphene. ACS Nano 2015, 9, 11509−11539. Funding h k Funding This work was supported by the Fortum Foundation, UCL, the Ramsay Memorial Trust, and EPSRC (grant EP/L017709). Notes ■ACKNOWLEDGMENTS Prof. Andrew L. Hector is thanked for his insightful comments. Dr. Steven Firth is thanked for useful discussions as well as training and assistance with the SEM instrument. Complete contact information is available at: https://dx.doi.org/10.1021/acsomega.0c00413 ACS Omega XXXX, XXX, XXX−XXX Authors Novel Route for Preparation of Zirconium Di-Selenide Thin Films. DAV Int. J. Sci. 2013, 2, 29. (48) Gnezdilov, V.; Pashkevich, Y. G.; Lemmens, P.; Wulferding, D.; Shevtsova, T.; Gusev, A.; Chareev, D.; Vasiliev, A. Interplay between Lattice and Spin States Degree of Freedom in the FeSe Super- conductor: Dynamic Spin State Instabilities. Phys. Rev. B 2013, 87, 144508. (28) Sargar, A. M.; Patil, N. S.; Mane, S. R.; Gawale, S. N.; Bhosale, P. N. Electrochemical Synthesis and Characterisation of ZrSe2 Thin Films. Int. J. Electrochem. Sci. 2009, 4, 87−894. (29) Rakkini, A. P. V.; Mohanraj, K. Effect of Different Combinations of Precursors of Zirconium and Selenium in the Electrodeposited ZrSe2 Thin Films. Ionics 2018, 24, 1243−1252. (49) Yuan, B.; Luan, W.; Tu, S.-t. One-Step Synthesis of Cubic FeS2 and Flower-like FeSe2 Particles by a Solvothermal Reduction Process. Dalton Trans. 2012, 41, 772−776. (30) Tsipas, P.; Tsoutsou, D.; Marquez-Velasco, J.; Aretouli, K. E.; Xenogiannopoulou, E.; Vassalou, E.; Kordas, G.; Dimoulas, A. Epitaxial ZrSe2/MoSe2 Semiconductor v. d. Waals Heterostructures on Wide Band Gap AlN Substrates. Microelectron. Eng. 2015, 147, 269−272. (50) Tsunekawa, S.; Asami, K.; Ito, S.; Yashima, M.; Sugimoto, T. XPS Study of the Phase Transition in Pure Zirconium Oxide Nanocrystallites. Appl. Surf. Sci. 2005, 252, 1651−1656. (51) Shenasa, M.; Sainkar, S.; Lichtman, D. XPS Study of Some Selected Selenium Compounds. J. Electron Spectrosc. Relat. Phenom. 1986, 40, 329−337. (31) Hector, A. L.; Levason, W.; Reid, G.; Reid, S. D.; Webster, M. Evaluation of Group 4 Metal Bis-Cyclopentadienyl Complexes with Selenolate and Tellurolate Ligands for CVD of ME2 Films (E = Se or Te). Chem. Mater. 2008, 20, 5100−5106. (52) Wickman, B.; Bastos Fanta, A.; Burrows, A.; Hellman, A.; Wagner, J. B.; Iandolo, B. Iron Oxide Films Prepared by Rapid Thermal Processing for Solar Energy Conversion. Sci. Rep. 2017, 7, 40500. (32) Boscher, N. D.; Carmalt, C. J.; Parkin, I. P. Atmospheric pressure chemical vapor deposition of WSe2 thin films on glass highly hydrophobic sticky surfaces. J. Mater. Chem. 2006, 16, 122− 127. (53) Sanchez-Perez, C.; Knapp, C. E.; Karjalainen, M. M.; Oilunkaniemi, R.; Carmalt, C. J.; Laitinen, R. S. Macrocycles Containing 1, 1’-Ferrocenyldiselenolato Ligands on Group 4 Metal- locenes. Dalton Trans. 2018, 5415. (33) Boscher, N. D.; Carmalt, C. J.; Parkin, I. P. Atmospheric Pressure CVD of TiSe2 Thin Films on Glass. Chem. Vap. Deposition 2006, 12, 54−58. (34) Sanchez-Perez, C.; Knapp, C. E.; Colman, R. Authors Clara Sanchez-Perez −Materials Chemistry Centre, Department of Chemistry, University College London, London WC1H 0AJ, U.K.; Laboratory of Inorganic Chemistry, Environmental and Chemical Engineering, University of Oulu, Oulu FI-90014, Finland; orcid.org/0000-0003-3291-8601 (8) O̅nuki, Y.; Inada, R.; Tanuma, S.-i.; Yamanaka, S.; Kamimura, H. Electrochemical Characteristics of TiS2, ZrSe2 and VSe2 in Secondary Lithium Battery. Jpn. J. Appl. Phys. 1981, 20, 1583−1588. ̅ (9) O̅nuki, Y.; Inada, R.; Tanuma, S.; Yamanaka, S.; Kamimura, H. Electrochemical Characteristics of ZrSe2 in a Secondary Lithium Battery. Solid State Ionics 1983, 8, 141−145. g Caroline E. Knapp −Materials Chemistry Centre, Department of Chemistry, University College London, London WC1H 0AJ, U.K.; orcid.org/0000-0003-4219-9313 (10) Jiang, H. Structural and Electronic Properties of ZrX 2 and HfX 2 (X = S and Se) from First Principles Calculations. J. Chem. Phys. 2011, 134, 204705. g Ross H. Colman −Faculty of Mathematics and Physics, Charles University, Prague 2 121 16, Czech Republic (11) Hankare, P. P.; Asabe, M. R.; Kokate, A. V.; Delekar, S. D.; Sathe, D. J.; Mulla, I. S.; Chougule, B. K. Effect of Annealing on Properties of ZrSe2 Thin Films. J. Cryst. Growth 2006, 294, 254−259. Carlos Sotelo-Vazquez −Materials Chemistry Centre, Department of Chemistry, University College London, London WC1H 0AJ, U.K.; orcid.org/0000-0003-0949-4548 (12) Moustafa, M.; Zandt, T.; Janowitz, C.; Manzke, R. Growth and Band Gap Determination of the ZrSxSe2−x Single Crystal Series. Phys. Rev. B 2009, 80, No. 035206. Sanjayan Sathasivam −Materials Chemistry Centre, Department of Chemistry, University College London, London WC1H 0AJ, U.K.; orcid.org/0000-0002-5206-9558 (13) Gleizes, A.; Revelli, J.; Ibers, J. A. Structures of a Semi- conducting and a Metallic Phase in the FeyZrSe2 System. J. Solid State Chem. 1976, 17, 363−372. Raija Oilunkaniemi −Laboratory of Inorganic Chemistry, Environmental and Chemical Engineering, University of Oulu, Oulu FI-90014, Finland (14) Kang, M.; Rathi, S.; Lee, I.; Li, L.; Khan, M. A.; Lim, D.; Lee, Y.; Park, J.; Yun, S. J.; Youn, D.-H.; Jun, C.; Kim, G.-H. Tunable Electrical Properties of Multilayer HfSe2 Field Effect Transistors by Oxygen Plasma Treatment. Nanoscale 2017, 9, 1645−1652. Complete contact information is available at: Complete contact information is available at: E https://dx.doi.org/10.1021/acsomega.0c00413 ACS Omega XXXX, XXX, XXX−XXX ACS Omega http://pubs.acs.org/journal/acsodf (35) Kumar, R. S.; Zhang, Y.; Sinogeikin, S.; Xiao, Y.; Kumar, S.; Chow, P.; Cornelius, A. L.; Chen, C. Crystal and Electronic Structure of FeSe at High Pressure and Low Temperature. J. Phys. Chem. B 2010, 114, 12597−12606. Authors (15) Buhannic, M.-A.; Ahouandjinou, A.; Danot, M.; Rouxel, J. Double coordinence du fer dans la phase FexZrSe2 (0 < x < 0,25): propriétés magnétiques et caractéristiques Mössbauer. J. Solid State Chem. 1983, 49, 77−84. (36) Owen, E. A.; Yates, E. L. XLI. Precision Measurements of Crystal Parameters. London Edinburgh Dublin Philos. Mag. J. Sci. 2009, 15, 472−488. (16) Buhannic, M.; Danot, M. Iron Migration in the Van Der Waals Gap of Zirconium Disulfide: Time Evolution of the FexZrS2 Compounds. Solid State Commun. 1990, 73, 739−742. (37) Basinski, Z. S.; Hume-Rothery, W.; Sutton, A. L. The Lattice Expansion of Iron. Proc. R. Soc. London, Ser. A 1955, 229, 459−467. ( ) p (17) Holgate, T. C.; Liu, Y.; Hitchcock, D.; Tritt, T. M.; He, J. Thermoelectric Properties of Li-Intercalated ZrSe2 Single Crystals. J. Electron. Mater. 2013, 42, 1751−1755. (37) Basinski, Z. S.; Hume-Rothery, W.; Sutton, A. L. The Lattice Expansion of Iron. Proc. R. Soc. London, Ser. A 1955, 229, 459−467. (38) Martin, U.; Boysen, H.; Frey, F. Neutron Powder Investigation of Tetragonal and Cubic Stabilized Zirconia, TZP and CSZ, at Temperatures up to 1400 K. Acta Crystallogr., Sect. B: Struct. Sci., Cryst. Eng. Mater. 1993, 49, 403−413. (38) Martin, U.; Boysen, H.; Frey, F. Neutron Powder Investigation of Tetragonal and Cubic Stabilized Zirconia, TZP and CSZ, at Temperatures up to 1400 K. Acta Crystallogr., Sect. B: Struct. Sci., Cryst. Eng. Mater. 1993, 49, 403−413. (18) Titov, A. N.; Yagafarova, Z. A.; Bikkulova, N. N. Investigation into the Structural and Electrical Properties of Silver-Intercalated Zirconium Diselenide. Phys. Solid State 2003, 45, 2067−2071. (39) Larson, A. C.; Von Dreele, R. B. GSAS; 1985, The University of California: Los Alamos, NM, 87545. (19) Muhammad, Z.; Mu, K.; Lv, H.; Wu, C.; ur Rehman, Z.; Habib, M.; Sun, Z.; Wu, X.; Song, L. Electron Doping Induced Semi- conductor to Metal Transitions in ZrSe2 Layers via Copper Atomic Intercalation. Nano Res. 2018, 4914. (40) Momma, K.; Izumi, F. VESTA 3 for Three-Dimensional Visualization of Crystal, Volumetric and Morphology Data. J. Appl. Crystallogr. 2011, 44, 1272−1276. (20) Ikhioya, L. I.; Chime, U. K.; Okoro, C. F.; Iroegbu, C.; Maaza, M.; Ezema, F. I. Influence of Dopant Concentration on the Electronic Band Gap Energy of Yb-ZrSe2 Thin Films for Photovoltaic Application via Electrochemical Deposition Technique. Mater. Res. Express 2020, 7, No. 026420. (41) Gleizes, A.; Jeannin, Y. Nonstoichiometry in the ZrSe2 Phase. J. Solid State Chem. Authors 1970, 1, 180−184. (42) Mleczko, M. J.; Zhang, C.; Lee, H. R.; Kuo, H.-H.; Magyari- Köpe, B.; Moore, R. G.; Shen, Z.-X.; Fisher, I. R.; Nishi, Y.; Pop, E. HfSe2 and ZrSe2: Two-Dimensional Semiconductors with Native High-κ Oxides. Sci. Adv. 2017, 3, No. e1700481. (21) Buhannic, M. A.; Colombet, P.; Danot, M. Spin-Glass Behavior of Iron-Intercalated Zirconium Disulfide. Solid State Commun. 1986, 59, 77−79. (43) Tiseanu, C.; Cojocaru, B.; Parvulescu, V. I.; Sanchez- Dominguez, M.; Primus, P. A.; Boutonnet, M. Order and Disorder Effects in Nano-ZrO2 Investigated by Micro-Raman and Spectrally and Temporarily Resolved Photoluminescence. Phys. Chem. Chem. Phys. 2012, 14, 12970. (22) Fatseas, G. A.; Goodenough, J. B. Mössbauer 57Fe Spectra Exhibiting “Ferrous Character”. J. Solid State Chem. 1980, 33, 219− 232. (23) McTaggart, F. K.; Wadsley, A. D. The Sulphides, Selenides, and Tellurides of Titanium, Zirconium, Hafnium, and Thorium. I. Preparation and Characterization. Aust. J. Chem. 1958, 11, 445. (44) de Faria, D. L. A.; Venâncio Silva, S.; de Oliveira, M. T. Raman Microspectroscopy of Some Iron Oxides and Oxyhydroxides. J. Raman Spectrosc. 1998, 28, 873−878. p J (24) Greenaway, D. L.; Nitsche, R. Preparation and Optical Properties of Group IV−VI2 Chalcogenides Having the CdI2 Structure. J. Phys. Chem. Solids 1965, 26, 1445−1458. (45) Li, Y.-S.; Church, J. S.; Woodhead, A. L. Infrared and Raman Spectroscopic Studies on Iron Oxide Magnetic Nano-Particles and Their Surface Modifications. J. Magn. Magn. Mater. 2012, 324, 1543− 1550. (25) Vollath, D.; Szabó, D. V. Nanoparticles from Compounds with Layered Structures. Acta Mater. 2000, 48, 953−967. (46) Yannopoulos, S. N.; Andrikopoulos, K. S. Raman Scattering Study on Structural and Dynamical Features of Noncrystalline Selenium. J. Chem. Phys. 2004, 121, 4747−4758. (26) Berthier, C.; Chabre, Y.; Segransan, P.; Chevalier, P.; Trichet, L.; Le Mehaute, A. Semiconductor-Metal Transition upon Inter- calation in LixZrSe2. Solid State Ionics 1981, 5, 379−382. (47) Goldan, A. H.; Li, C.; Pennycook, S. J.; Schneider, J.; Blom, A.; Zhao, W. Molecular Structure of Vapor-Deposited Amorphous Selenium. J. Appl. Phys. 2016, 120, 135101. x 2 (27) Asabe, M. R.; Ubale, V. P.; Rajmane, S. V.; Manukshete, A. H. Novel Route for Preparation of Zirconium Di-Selenide Thin Films. x 2 (27) Asabe, M. R.; Ubale, V. P.; Rajmane, S. V.; Manukshete, A. H. (27) Asabe, M. R.; Ubale, V. P.; Rajmane, S. V.; Manukshete, A. H. Novel Route for Preparation of Zirconium Di-Selenide Thin Films. DAV Int. J. Sci. 2013, 2, 29. https://dx.doi.org/10.1021/acsomega.0c00413 ACS Omega XXXX, XXX, XXX−XXX Authors H.; Sotelo- Vazquez, C.; Oilunkaniemi, R.; Laitinen, R. S.; Carmalt, C. J. Accessing New 2D Semiconductors with Optical Band Gap: Synthesis of Iron-Intercalated Titanium Diselenide Thin Films via LPCVD. RSC Adv. 2018, 8, 22552−22558. F ttps://dx.doi.org/10.1021/acsomega.0c00413 ACS Omega XXXX, XXX, XXX−XXX https://dx.doi.org/10.1021/acsomega.0c00413 ACS Omega XXXX, XXX, XXX−XXX
https://openalex.org/W3200896096	https://europepmc.org/articles/pmc8460001?pdf=render	English	null	Looking for more reliable biomarkers in breast cancer: Comparison between routine methods and RT-qPCR	PloS one	2,021	cc-by	9,967	PLOS ONE PLOS ONE RESEARCH ARTICLE Methods Herein, we investigated discrepancies in the assessment of estrogen receptor (ER), proges- terone receptor (PR), human epidermal growth factor receptor 2 (HER2) and marker of pro- liferation Ki67 comparing routinely obtained IHC (and FISH) data (ORI) with the results of manual (REV) and semi-automated (DIA) re-evaluation of the original IHC slides and then with RNA expression data from the same tissue block using the MammaTyper® (MT) gene expression assay. Published: September 23, 2021 Copyright: © 2021 Caselli et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Looking for more reliable biomarkers in breast cancer: Comparison between routine methods and RT-qPCR Emanuele CaselliID1☯, Cristina Pelliccia1☯, Valeria Teti1, Guido Bellezza1, Martina MandaranoID1, Ivana Ferri1, Kerstin Hartmann2, Mark Laible2, Ugur Sahin2, Zsuzsanna Varga3, Chiara LupiID4, Fabrizio Stracci4,5, Angelo SidoniID1 Emanuele CaselliID1☯, Cristina Pelliccia1☯, Valeria Teti1, Guido Bellezza1, Martina MandaranoID1, Ivana Ferri1, Kerstin Hartmann2, Mark Laible2, Ugur Sahin2, Zsuzsanna Varga3, Chiara LupiID4, Fabrizio Stracci4,5, Angelo SidoniID1 1 Department of Medicine and Surgery, Section of Anatomic Pathology and Histology, Medical School, University of Perugia, Perugia, Italy, 2 BioNTech Diagnostics GmbH, Mainz, Germany, 3 Institute for Pathology and Molecular Pathology, Universita¨tsspital Zu¨rich, Zu¨rich, Switzerland, 4 Umbria Cancer Registry, Perugia, Italy, 5 Department of Medicine and Surgery, Section of Public Health, University of Perugia, Perugia, Italy a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 ☯These authors contributed equally to this work. ☯These authors contributed equally to this work. * caselliemanuele1986@gmail.com ☯These authors contributed equally to this work. * caselliemanuele1986@gmail.com * caselliemanuele1986@gmail.com Purpose Decades of quality control efforts have raised the standards of immunohistochemistry (IHC), the principle method used for biomarker testing in breast cancer; however, computa- tional pathology and reverse transcription quantitative PCR (RT-qPCR) may also hold prom- ise for additional substantial improvements. Editor: Fernando Schmitt, Medical Faculty of Porto University and IPATIMUP, PORTUGAL Received: December 13, 2020 Accepted: July 19, 2021 Published: September 23, 2021 Received: December 13, 2020 Accepted: July 19, 2021 Published: September 23, 2021 Copyright: © 2021 Caselli et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the manuscript and its Supporting Information. Data Availability Statement: All relevant data are within the manuscript and its Supporting Information. OPEN ACCESS Citation: Caselli E, Pelliccia C, Teti V, Bellezza G, Mandarano M, Ferri I, et al. (2021) Looking for more reliable biomarkers in breast cancer: Comparison between routine methods and RT- qPCR. PLoS ONE 16(9): e0255580. https://doi.org/ 10.1371/journal.pone.0255580 Citation: Caselli E, Pelliccia C, Teti V, Bellezza G, Mandarano M, Ferri I, et al. (2021) Looking for more reliable biomarkers in breast cancer: ☯These authors contributed equally to this work. * caselliemanuele1986@gmail.com Introduction Analysis of breast cancer tissues by pathologists generates a wealth of information that can be used to classify disease for prognostic purposes and help guide therapeutic decisions. Recent molecular studies have lent further support to the historical importance of hormone receptors, growth factors and perturbations in proliferation as the key operators of breast cancer initiation and progression [1]. Thus, it is not surprising that assessing the expression of estrogen receptor (ER), progesterone receptor (PR), human epidermal growth factor two (HER2) and prolifera- tion remains the cornerstone of breast cancer diagnosis and treatment [2], although the well- known problems of accuracy and reproducibility which have necessitated the issuing of various guidelines [3–6], are still persistent [7]. In fact, IHC is methodologically error-prone, as it is a multi-step process, highly exposed to variabilities of various kinds, such as from the use of non- standardized local practices or from different, non-equally performing albeit approved reagents [8, 9]. However, the main source of variation is linked to the intrinsic limits of immunohisto- chemical stains, both for marked influences of the pre-analytical phase [7] and the subjectivity of interpretative criteria [10] as well as extreme intra- and inter-tumoral heterogeneity [11] Some IHC markers are more susceptible to methodological failures owing to these limita- tions. The proliferation marker Ki67 is the most prominent example, because it is expressed heterogeneously in tissues and is continuously distributed, which makes it difficult to identify a single, reproducible cut-off [10]. For HER2, IHC has been greatly improved by FISH. Unfor- tunately, FISH is a method that requires a specific expertise, a fluorescent microscope, and is more time-consuming. The complexities of FISH are exemplified in a recent guideline update, which aims to address an ever-expanding list of methodological intricacies [12]. However, there are other reliable methods of in situ hybridization for the assessment of HER2 amplifica- tion that surpass these drawbacks, such as CISH, SISH and DDISH. Even though modern scientific advances were immensely assisted by corresponding tech- nological breakthroughs, such as gene arrays and reverse transcription quantitative PCR (RT- qPCR), the landscape of breast cancer diagnostics remains dominated by conventional hema- toxylin and eosin (H&E) stains and by immunohistochemistry (IHC) [13]. Conclusions specific roles of these authors are articulated in the ‘author contributions’ section.” The remaining authors received no specific funding for this work. Determination of the breast cancer biomarkers ER, PR, HER2 and Ki67 at the mRNA level shows high degree of correlation with IHC and compares well with correlations between original with subsequent independent manual or semi-automated IHC assessments. The use of methods with wider dynamic range and higher reproducibility such as RT-qPCR may offer more precise assessment of endocrine responsiveness, improve Ki67 standardization and help resolve HER2 cases that remain equivocal or ambiguous by IHC/FISH. In sum- mary, our findings seem to configure RT-qPCR as a complementary method to be used in cases of either equivocal results or presenting, at the traditional determination assays, bio- markers expressions close to the cut-off values. Competing interests: Kerstin Hartmann, Mark Laible, Ugur Sahin: Salary and stock ownership BioNTech Diagnostics GmbH / BioNTech AG. These commercial affiliations do not alter our adherence to PLOS ONE policies on sharing data and materials. Mark Laible: Patent ownership of WO 2015/024942, Commercialized as MammaTyper (TM) Kit. The other authors declare that they have no conflict of interest. Results Correlation for ER and PR was high between ORI IHC and the other three study methods (REV, DIA and RT-qPCR). For HER2, 10 out of 96 discrepant cases can be detected between ORI and REV that involved at least one call in the equivocal category (except for one case). For Ki67, 22 (29.1%) cases were categorized differently by either REV alone (n = 17), DIA alone (n = 15) or both (n = 10) and 28 cases (29.2%) for RT-qPCR. Most of the dis- crepant Ki67 cases changed from low to high between the original and following assess- ment and belonged to the intermediate Ki67 expression range (between 9 and 30%). Funding: KH, ML and US received salary from and have stock ownership of BioNTech Diagnostics GmbH / BioNTech AG. However, BioNTech Diagnostics GmbH / BioNTech AG provided support in the form of salaries for authors KH, ML and US, but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. The PLOS ONE \| https://doi.org/10.1371/journal.pone.0255580 September 23, 2021 1 / 18 September 23, 2021 PLOS ONE RT-qPCR assessment of breast cancer biomarkers specific roles of these authors are articulated in the ‘author contributions’ section.” The remaining authors received no specific funding for this work. PLOS ONE \| https://doi.org/10.1371/journal.pone.0255580 September 23, 2021 Patient selection Between March 2010 and February 2012, 288 women underwent surgery for invasive breast carcinoma (Section of Anatomic Pathology and Histology, Perugia, Italy). The IHC results from patient histopathology reports were screened to identify cases with varying levels of diag- nostic difficulty (Fig 1). Available material from 116 cases, including FFPE blocks, H&E slides and IHC slides, were retrieved from the archive. The previously reported results for ER, PR, Ki67 and HER2 were registered as the original data (ORI) and consisted of assessments deliv- ered during diagnostic routine by 6 trained pathologists according to available guidelines at that time [3, 6]. An experienced pathologist and FISH expert (GB) performed the ORI FISH data. Informed consent The current study is a retrospective one, conducted on data and biological material already fully anonymized before the authors accessed them. The patients’ data and corresponding samples were accessed in the period between March 2010 and February 2012. According to the internal rules both of the University of Perugia and of the Hospital of Perugia, we have consid- ered only the biological material in respect of which the patients expressed his/her specific informed consent to use such a material for research purposes; we also have observed all rules concerning confidentiality and protection of person data, in accordance with European Union, International and National rules. Compliance with ethical standards All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Introduction Current consensus does not favor the use of alternative methods such as for example RT-qPCR for assessing the expression of individual markers, and even though reliable and reproducible solutions are commercially available, their role however being mostly complementary [14, 15]. In 2017, the US Food and Drug Administration (FDA) announced the approval of the first whole slide imaging (WSI) system for primary diagnosis in surgical pathology [16]. Aside PLOS ONE \| https://doi.org/10.1371/journal.pone.0255580 September 23, 2021 2 / 18 PLOS ONE RT-qPCR assessment of breast cancer biomarkers from RT-qPCR, automated digital image analysis (DIA) may also pave the way for more pre- cise biomarker assessment and more accurate histopathological breast cancer diagnosis and has the potential to provide the objectivity and reliability required to transform biomarker test- ing in breast cancer. The entire process, however, involves several steps, including image acquisition, storage and management, annotation and viewing or sharing, each one requiring stringent quality control procedures [17]. Therefore, considering the diversity of candidate analytical methods and ongoing chal- lenges with IHC/FISH, the decision for or against alternative methods for biomarker testing in breast cancer remains complex. In this respect, local attempts to benchmark new techniques against testing traditions may help estimate their feasibility and potential for routine use. In this study, we carried out parallel comparisons between 3 independent IHC assessments, including an image analysis solution and comparisons between these and RT-qPCR, using well-established local IHC methods, international scoring guidelines and the MammaTyper1 gene expression assay. The aim of the study was to detect discordance between methods and discuss their clinicopathological implications. Conventional histopathological assessments The re-evaluation dataset (REV) was compiled as follows: An experienced pathologist (EC) that was not involved in the initial screening and was blinded to clinicopathological informa- tion carried out tumor classifications and grading on the available original H&E slides 3 / 18 PLOS ONE \| https://doi.org/10.1371/journal.pone.0255580 September 23, 2021 PLOS ONE RT-qPCR assessment of breast cancer biomarkers Fig 1. Study design. https://doi.org/10.1371/journal.pone.0255580.g001 Fig 1. Study design. according to international standards in force at the time of the study [18, 19]. Twenty cases were excluded from further analyses due to inadequate fixation. In a second step the same pathologist re-evaluated ER, PR, Ki67 and HER2 on the existing IHC slides, but 2 cases for ER, 3 cases for PR, 2 cases for Ki67 and 12 cases for HER2 were re-stained, owing to missing origi- nal slides or low staining quality including fading. Standard international recommendations— in force when the research project was conducted, particularly according with the 2013 ASCO/ CAP guidelines [4]—were used for the assessment and quantification of staining reactions [4, 6, 20]. Of note, original assessments for Ki67 were based on quick eyeballing, whereas formal counting was used upon re-evaluation. Due to the impossibility of re-evaluation of the original FISH slides, all HER2 equivocal cases by IHC (2+ score) after REV, regardless of ORI FISH data when available, were subjected to FISH analysis (dual-probe, Leica HER2 FISH system). Furthermore, we performed FISH analysis in all discrepant HER2/ERBB2 cases between IHC/ FISH and MammaTyper1. Interpretation was performed by EC and an additional blinded evaluation was performed by GB in selected cases according to the ASCO/CAP 2013 guidelines (in force at the time of the study), either by counting signals in randomly chosen regions of interest with the assistance of a digital tool (Cytovision software, Leica Biosystems) (EC) or according to international standards in force at the time of the study [18, 19]. Twenty cases were excluded from further analyses due to inadequate fixation. In a second step the same pathologist re-evaluated ER, PR, Ki67 and HER2 on the existing IHC slides, but 2 cases for ER, 3 cases for PR, 2 cases for Ki67 and 12 cases for HER2 were re-stained, owing to missing origi- nal slides or low staining quality including fading. PLOS ONE \| https://doi.org/10.1371/journal.pone.0255580 September 23, 2021 Conventional histopathological assessments Standard international recommendations— in force when the research project was conducted, particularly according with the 2013 ASCO/ CAP guidelines [4]—were used for the assessment and quantification of staining reactions [4, 6, 20]. Of note, original assessments for Ki67 were based on quick eyeballing, whereas formal counting was used upon re-evaluation. Due to the impossibility of re-evaluation of the original FISH slides, all HER2 equivocal cases by IHC (2+ score) after REV, regardless of ORI FISH data when available, were subjected to FISH analysis (dual-probe, Leica HER2 FISH system). Furthermore, we performed FISH analysis in all discrepant HER2/ERBB2 cases between IHC/ FISH and MammaTyper1. Interpretation was performed by EC and an additional blinded evaluation was performed by GB in selected cases according to the ASCO/CAP 2013 guidelines (in force at the time of the study), either by counting signals in randomly chosen regions of interest with the assistance of a digital tool (Cytovision software, Leica Biosystems) (EC) or PLOS ONE \| https://doi.org/10.1371/journal.pone.0255580 September 23, 2021 4 / 18 September 23, 2021 PLOS ONE RT-qPCR assessment of breast cancer biomarkers manually (GB). Of note, there was no disagreement between the two pathologists. Three selected cases with discrepancies to MammaTyper1 ERBB2 determination were sent to an independent pathologist (ZV) for blinded FISH re-analysis (hybridization and evaluation). Computer-assisted IHC assessment The same pathologist (EC) independently analyzed the expression of hormone receptors and Ki67 with a digital semi-automated image analysis (DIA) system (Aperio ImageScope, Leica Biosystems). All slides were scanned on the Aperio AT2 Scanner (Leica Biosystems) with a 40x magnification lens, using single focus layer without Z-stacking and were displayed on a 4K (ultra-high definition) monitor. On the digital slide (40x magnification), the pathologist selected randomly several regions of interest for ER and PR, whereas for Ki67 areas were cho- sen in such a way as to be representative of heterogeneity when present. Digital image analysis for assessing the percentage of positive nuclei was performed according to the Leica nuclear v9 algorithm. As regards HER2, the digital image analysis was not performed because of an opti- mized and reliable algorithm for assessing the membrane immunolabeling was unavailable at the time of the study. Fluorescence in situ hybridization (FISH) FISH was carried out using the automated Leica HER2 FISH System on a BOND-III platform. Importantly, in the original determinations, FISH was performed using the HER-2 DNA Pro- beKit II (PathVysion) with the Vysis Paraffin Pretreatment Reagent Kit. The dual-color probes identified the centromere 17 (Cep 17) as green signal, whereas the HER2/Neu gene/locus as orange one. Positive and negative controls were always available. Immunohistochemistry (IHC) All reactions were carried out on a BOND-III fully automated immunohistochemistry stainer (Leica Biosystems) with antibodies against ER, PR, HER2 and Ki67 according to vendor proto- cols and the use of appropriate positive and negative controls (Table 1). Of note, all 19 stains that had to be repeated were performed on the same block as the origi- nals and without any reagent or protocol deviations compared to those available from the archive. Only for HER2, all slides prepared for re-evaluation were stained on the Autostainer- Link 48(Dako) using Hercept Test™. Dewaxing and rehydration were performed manually. Antigen retrieval (pH = 6) was performed by heating the slides for 40 minutes at 97˚C using the DakoPT-link. RNA isolation and mRNA quantification with RT-qPCR The MammaTyper1 test was performed on 10 μm sections. In cases with tumor cell content <20%, manual dissection was conducted. MammaTyper1 test was successfully performed in Table 1. Immunhistochemistry staining specifications. Antibody Clone Dilution Manufacturer Platform ER 6F11 Ready to use Leica Biosystems BOND-III fully automated IHC PR 16 Ready to use Leica Biosystems stainer (Leica Biosystems) Ki67 MIB1 1:100 Dako HER2 Rabbit Anti-Human HER2 protein Ready to use Dako BOND-III fully automated IHC stainer (Leica Biosystems) and AutostainerLink 48 Dako (for the 12 restained slides) https://doi.org/10.1371/journal.pone.0255580.t001 Table 1. Immunhistochemistry staining specifications. Antibody Clone Dilution Manufacturer Platform ER 6F11 Ready to use Leica Biosystems BOND-III fully automated IHC PR 16 Ready to use Leica Biosystems stainer (Leica Biosystems) Ki67 MIB1 1:100 Dako HER2 Rabbit Anti-Human HER2 protein Ready to use Dako BOND-III fully automated IHC stainer (Leica Biosystems) and AutostainerLink 48 Dako (for the 12 restained slides) https://doi.org/10.1371/journal.pone.0255580.t001 PLOS ONE \| https://doi.org/10.1371/journal.pone.0255580 September 23, 2021 5 / 18 Table 1. Immunhistochemistry staining specifications. PLOS ONE \| https://doi.org/10.1371/journal.pone.0255580 September 23, 2021 5 / 18 PLOS ONE RT-qPCR assessment of breast cancer biomarkers all 96 cases, applying manual dissection in 16 cases. Extraction of total RNA from FFPE sam- ples was performed using a CE-marked paramagnetic bead-based method (RNXtract1, BioN- Tech Diagnostics, Mainz, Germany) according to the manufacturers’ instructions, as previously described in detail [21]. The expression levels of ERBB2, ESR1, PGR and MKI67 were determined by RT-qPCR using the CE-marked MammaTyper1 IVD kit (BioNTech Diagnostics) on the CFX96TM (BIO-RAD1) platform according to the manufacturer’s instructions. Calculations of delta-delta-Cq values were carried out as described previously [21], however data was further transformed to the relative unit scale (RU) to allow easier inter- pretation. The RU values is calculated by centering the delta-delta-Cq around the marker spe- cific cutoff value. In this way, positive values correspond to gene expression higher than the cutoff and negative values represent an expression below the cutoff while the Cq (log2) scale is retained. Relative units (RU) = ((Median Cq target sample–Combined Reference sample)– (Median Cq target Positive Control–Combined Reference Positive Control))–marker specific cutoff. As cutoffs in this study, for ESR1 and PGR, a lower, diagnostic cutoff was applied (RU 0) which was established to fit the 1% IHC staining for ER and PR. For MKI67, the previously clinically validated cutoff was applied [22]. Rules for nominal categories (positive/negative) for all markers are summarized in Table 2. Statistical analysis The median difference, absolute median difference and standard deviation of the absolute median difference between ORI versus REV and ORI versus DIA for ER, PR and Ki67 were assessed. Calculation of intraclass correlation coefficient (ICC) which estimates the impreci- sion in relationship with the intersample variance was calculated using the formula: ICC = intersample variance/total variance [23, 24]. The Pearson correlation coefficient was used to measure the strength of the association between ORI versus REV, ORI versus DIA and ORI versus RT-qPCR for ER, PR and Ki67. Graphs were prepared using GraphPad Prism 6 (GraphPad Software, La Jolla, USA). A two-sided P value <0.05 was considered significant. PLOS ONE PLOS ONE RT-qPCR assessment of breast cancer biomarkers Table 3. Median proportion scores for IHC. ORI versus REV ORI versus DIA Median of Δ Median of abs. Δ SD of abs. Δ ICC Median of Δ Median of abs. Δ SD of abs. Δ ICC ER 0.0 3.0 8.6 0.949 0.0 4.0 7.4 0.955 PR 0.0 3.5 9.2 0.957 -2.5 5.0 9.3 0.945 Ki67 -3.5 5.0 5.6 0.903 -4.0 5.0 6.3 0.855 For ER, PR and Ki67 median and absolute (Abs.) median of difference between original and re-evaluation, as well as between original and digital image analysis was calculated. ICC: intraclass correlation. Table 3. Median proportion scores for IHC. For ER, PR and Ki67 median and absolute (Abs.) median of difference between original and re-evaluation, as well as between original and digital image analysis was calculated. ICC: intraclass correlation. For ER, PR and Ki67 median and absolute (Abs.) median of difference between original and re-evaluation, as well as betwee calculated. ICC: intraclass correlation. https://doi.org/10.1371/journal.pone.0255580.t003 2A and 2B) and confirmed the trend of larger variations observed for samples with intermedi- ate staining results (30–80% positive cells). Regarding PR, there were originally 18 negative and 78 positive cases with 2 discrepancies, both involving a difference between the original data and consecutive assessments, whereas the latter two were always concordant. Compared to original values, proportion scores differed in 57 cases upon REV (59.4%) and in 76 cases upon DIA (79.2%). In both cases, increase was more common than reduction (41 vs 16 for REV and 51 vs 25 for DIA), but in only 42 cases was the change in the same direction. Among 6 cases with a difference 25%, four belonged to the original group with positivity between 1–49%. Compared to ORI data, PR percentage positivity differed by more than 10% in 30 out of 96 cases (31.3%) for REV and in 34 out of 96 cases (35.4%) for DIA (Fig 2C and 2D). As seen for ER, the largest differences in the staining percent assessments are seen in the lower range (15– 80%). There were 58 low and 38 high Ki67 cases, based on the original evaluation, but 22 (29.1%) were categorized differently by either REV alone (n = 17), DIA alone (n = 15) or both (n = 10). PLOS ONE Most (twenty out of 22) discrepant cases changed from low to high between the original and following assessments (Fig 2G and 2H). Setting lower and upper Ki-67 cut-offs at 9% and 30% respectively to distinguish unequivocally low from high cases, we found that discrepant cases lied mostly between the two cut-offs, namely in the intermediate Ki67 expression range (median score of discrepant cases = 15%). In stark contrast, cases that were concordant between all 3 assessments, were either clearly low (median Ki67-low score = 5.5%) or clearly high (median Ki67-high score = 33.5%). As expected, discordance rates were significantly higher in the intermediate group (44.2%) compared with the low- and high groups (6.8% and 4.2% respectively). Compared to original values, proportion scores differed in 75 cases upon REV (78.1%) and in 94 cases upon DIA (97.9%). Upgrading was in both cases more common than downgrading (62 vs 21 for REV and 71 vs 23 for DIA) and was in the same direction in 63 cases. Discordant and concordant cases did not differ by any of the available clinic-patho- logical parameters. Compared to ORI data, Ki67 proportion scores differed by more than 10% in 26 out of 96 cases (27.1%) for REV and in 28 out of 96 cases (29.2%) for DIA (Fig 2E and 2F). With regard to absolute differences between the assessments, higher Ki67stainings (>15%) had clearly higher deviations than samples with a low Ki67 staining percentage (<15%). https://doi.org/10.1371/journal.pone.0255580.t003 Concordance between IHC assessments for ER, PR and Ki67 The values for ICC shown in Table 3 indicate the high agreement between the continuous per- centage scores of ORI IHC and the other two study methods (REV and DIA). There was no discordance for ER status between all three IHC assessments, with 10 ER-negative and 86 ER positive cases. Compared with original values, proportion scores differed in 58 cases upon REV (59.4%) and in 80 cases upon DIA (82.3%). In both cases, increase was more common than reduction (35 vs 23 for REV and 45 vs 35 for DIA), but in only 46 cases the change was in the same direction. Among 3 cases with a difference 25%, two belonged to the original group with positivity between 1–49%. Compared to ORI data, ER percentage positivity dif- fered by more than 10% in 17 out of 96 cases (17.7%) for REV and in 22 out of 96 cases (22.9%) for DIA. Variabilities were further graphically analyzed with Bland-Altman plots (Fig Table 2. Instrument-specific relative units (RU) for CFX96TM (BIO-RAD1) and the corresponding IHC equivalents. RU ERBB2 ESR1 PGR MKI67 Negative Equivocal Positive Negative Low Positive High Positive Negative Low Positive High Positive Low High MammaTyper < -0.7 -0.7 < 0 0 < 0 0 < 1.1 1.1 < 0 0 < 0.5 0.5 < 0 0 IHC equivalent 0 / 1+ 2+ 3+ 0 1–49% 50% 0 1–19% 20% <20% 20% https://doi.org/10.1371/journal.pone.0255580.t002 Table 2. Instrument-specific relative units (RU) for CFX96TM (BIO-RAD1) and the corresponding IHC equivalents. RU ERBB2 ESR1 PGR MKI67 Negative Equivocal Positive Negative Low Positive High Positive Negative Low Positive High Positive Low High MammaTyper < -0.7 -0.7 < 0 0 < 0 0 < 1.1 1.1 < 0 0 < 0.5 0.5 < 0 0 IHC equivalent 0 / 1+ 2+ 3+ 0 1–49% 50% 0 1–19% 20% <20% 20% https://doi.org/10.1371/journal.pone.0255580.t002 PLOS ONE \| https://doi.org/10.1371/journal.pone.0255580 September 23, 2021 6 / 18 le 2. Instrument-specific relative units (RU) for CFX96TM (BIO-RAD1) and the corresponding IHC equivalents. PLOS ONE \| https://doi.org/10.1371/journal.pone.0255580 September 23, 2021 6 / 18 PLOS ONE \| https://doi.org/10.1371/journal.pone.0255580 September 23, 2021 HER2 discrepancies Based on the original combined IHC/FISH data, there were 17 positive cases for HER2. Except for one case, all other discrepancies involved at least one call in the equivocal category (2013 ASCO/CAP guidelines, which were employed at the time of the current study). Here, the REV FISH result was negative (in line with the RT-qPCR result and different than the ORI assess- ment that was positive). As shown in Fig 4A, re-evaluation identified 6 potential false positives. All 13 ORI HER2 equivocal cases remained equivocal upon REV, but REV identified 9 addi- tional equivocal cases, 3 of which were ORI 3+, whereas 6 were ORI 0/1+. These 9 discrepant cases were subjected to FISH—evaluated according to the 2013 ASCO/CAP guidelines, which were in force at the time of the present study—, which agreed with the original IHC assessment in 6 cases. From the 14 cases with an ORI FISH result, FISH analysis was repeated for 13 sam- ples. All 9 negative cases and 1 equivocal case were confirmed by re-evaluation. However, from the 3 ORI FISH positive results only one was confirmed in re-evaluation, while one was identified as negative and one as equivocal during the second round of re-evaluation by ZV. Furthermore, the blinded FISH re-analysis by ZV differed also in the 2 other cases tested: none case has been found amplified, but both were defined equivocal. As shown in Fig 4B, discordance for HER2/ERBB2 was typically associated with cases being equivocal by at least one IHC assessment (ORI or REV). One ERBB2 potential false negative could be rescued by the introduction of an equivocal zone spanning 0.7 RUs to the left of the cut-off (Fig 3G and 3H). PLOS ONE PLOS ONE RT-qPCR assessment of breast cancer biomarkers Table 4. Agreement within IHC (ORI versus REV, ORI versus DIA) and between IHC (ORI) and RT-Qpcr. ER / ESR1 PR /PGR Ki67 / MKI67 IHC cutoff Positive if 1% Positive if 1% Positive if 20% RT-qPCR cutoff Positive if RU 0 Positive if RU 0 Positive if RU 0 ORI vs REV ORI vs DIA ORI vs RT-qPCR ORI vs REV ORI vs DIA ORI vs RT-qPCR ORI vs REV ORI vs DIA ORI vs RT-qPCR Pearson´s coefficient 0.811 0.835 0.826 0.892 0.911 0.912 0.692 0.708 0.709 https://doi.org/10.1371/journal.pone.0255580.t004 and positive at the RNA level (Fig 3A and 3B). By contrast, discordant cases for progesterone receptor were more often negative by RT-qPCR while being positive by REV or DIA IHC (7/ 9). The RNA expression of discordant cases took values close to the assay cut-off, while corre- sponding protein expression was low (Fig 3C and 3D). Two additional cases were discordant between MammaTyper1 and the original assessment only. The discordance rate between the two methods for the proliferation marker was dependent on the type of IHC assessment being highest for the ORI data (29.2%), followed by comparable rates for DIA and REV (17.7% and 15.6% respectively). Among 32 discordant cases affected by discrepancies between RT-qPCR and any IHC assessment, 23 also displayed discordances between the different IHC data and 29 showed high MKI67 expression (Fig 3E and 3F). Dis- crepant cases had an average median Ki67 score of 16.7% compared to 25% in the concordant group and their RNA expression showed a tendency to cluster around the MKI67cut-off. Dis- cordance was not associated with histology, grade or other available tumor characteristics. Correlation between IHC and RT-qPCR for ER, PR and Ki67 Correlation between original IHC and RT-qPCR was very good for ER, PR and HER2 and good for Ki67 and strikingly comparable to the quality of the correlation between different IHC methods (Table 4). Estrogen receptor status by IHC (all 3 independent assessments) and MammaTyper1 was discordant in 4 cases, with 3 out of 4 being negative at the protein level PLOS ONE \| https://doi.org/10.1371/journal.pone.0255580 September 23, 2021 7 / 18 PLOS ONE RT-qPCR assessment of breast cancer biomarkers Fig 2. Correlation between original (ORI) and paired re-evaluated (REV) or digital image analyzed (DIA) samples for IHC. Bland- Altman plots showing differences between original, re-evaluated and digital image analysis for ER (A, B), PR (C, D) and Ki67 (E, F). Cut-off value at ±10% is shown by red dotted lines. Scatterplots showing correlation between original, re-evaluated and digital image analysis for Ki67 (G, H). Cut-off value at 20% for Ki67 is shown by black dotted lines. Red dots highlight discordant cases between original and either re-evaluated or digital image analysis. https://doi.org/10.1371/journal.pone.0255580.g002 Fig 2. Correlation between original (ORI) and paired re-evaluated (REV) or digital image analyzed (DIA) samples for IHC. Bland- Altman plots showing differences between original, re-evaluated and digital image analysis for ER (A, B), PR (C, D) and Ki67 (E, F). Cut-off value at ±10% is shown by red dotted lines. Scatterplots showing correlation between original, re-evaluated and digital image analysis for Ki67 (G, H). Cut-off value at 20% for Ki67 is shown by black dotted lines. Red dots highlight discordant cases between original and either re-evaluated or digital image analysis. PLOS ONE \| https://doi.org/10.1371/journal.pone.0255580 September 23, 2021 8 / 18 Impact of discordance on subtyping Black dotted lines represent the respective cut-off for the RT-qPCR assay (ERBB2: 0.0) and for the positivity of the IHC staining (HER2: 2+). Second dotted line on the x-axis represents a second cut-off for ERBB2 (-0.7) after introduction of an equivocal zone. y-axis and MKI67 RU values on the x-axis. Cut-off values at 20% for IHC and 0.0 RU for RT-qPCR are shown by black dotted lines. Scatterplots showing correlation between IHC determination (ORI and REV) and RT-qPCR assessment for ERBB2/HER2 (G, H). Black dotted lines represent the respective cut-off for the RT-qPCR assay (ERBB2: 0.0) and for the positivity of the IHC staining (HER2: 2+). Second dotted line on the x-axis represents a second cut-off for ERBB2 (-0.7) after introduction of an equivocal zone. https://doi.org/10.1371/journal.pone.0255580.g003 https://doi.org/10.1371/journal.pone.0255580.g003 Luminal B-like upon re-evaluation (23.7% for manual, 18.4% for digital) or when MammaTy- per1 RNA data were used instead of IHC (34.2%) (Fig 5). Discrepant cases in the assessment of ER led to four luminal to non-luminal switches. As regards HER2, the discrepancies highlighted by Mammatyper1 allowed us to identify and better stratify 3 originally FISH-discrepant cases, overcoming the issues related to both laboratory quality assurance and inter-/intraobserver subjectivity in interpreting its immuno- histochemical and ISH assessment. Impact of discordance on subtyping Each tumor was assigned a molecular subtype according to the surrogate definitions of the St. Gallen expert panel [25, 26]. Not surprisingly considering the classification rules, the effect of discordance in the assessment of Ki67 was expressed as discrepancies mainly in the luminal categories with many originally Luminal A-like cases being subsequently categorized as PLOS ONE \| https://doi.org/10.1371/journal.pone.0255580 September 23, 2021 9 / 18 PLOS ONE RT-qPCR assessment of breast cancer biomarkers Fig 3. Comparison between IHC and RT-qPCR (MammaTyper1) for estrogen receptor (A, B), progesterone receptor (C, D), proliferation marker Ki67 (E, F) and human epidermal growth factor receptor 2 (G, H). Scatterplots with ORI IHC ER scores (A) or DIA IHC ER score (B) on the y-axis and ESR1 RU values on the x-axis. Cut-off values at 1% for IHC and 0.0 RU for RT-qPCR are shown by black dotted lines. Scatterplots with ORI IHC PR scores (C) or DIA IHC PR score (D) on the y-axis and PGR RU values on the x-axis. Cut-off values at 1% for IHC and 0.0 RU for RT-qPCR are shown by black dotted lines. Scatterplots with ORI IHC Ki67 scores (E) or DIA IHC Ki67 score (F) on th E RT-qPCR assessment of breast cancer biomarker Fig 3. Comparison between IHC and RT-qPCR (MammaTyper1) for estrogen receptor (A, B), progesterone receptor (C, D), proliferation marker Ki67 (E, F) and human epidermal growth factor receptor 2 (G, H). Scatterplots with ORI IHC ER scores (A) or DIA IHC ER score (B) on the y-axis and ESR1 RU values on the x-axis. Cut-off values at 1% for IHC and 0.0 RU for RT-qPCR are shown by black dotted lines. Scatterplots with ORI IHC PR scores (C) or DIA IHC PR score (D) on the y-axis and PGR RU values on the x-axis. Cut-off values at 1% for IHC and 0.0 RU for RT-qPCR are shown by black dotted lines. Scatterplots with ORI IHC Ki67 scores (E) or DIA IHC Ki67 score (F) on the PLOS ONE \| https://doi.org/10.1371/journal.pone.0255580 September 23, 2021 10 / 18 PLOS ONE RT-qPCR assessment of breast cancer biomarkers y-axis and MKI67 RU values on the x-axis. Cut-off values at 20% for IHC and 0.0 RU for RT-qPCR are shown by black dotted lines. Scatterplots showing correlation between IHC determination (ORI and REV) and RT-qPCR assessment for ERBB2/HER2 (G, H). PLOS ONE \| https://doi.org/10.1371/journal.pone.0255580 September 23, 2021 Discussion Immunohistochemistry showed a high degree of correlation for both ER and PR, even though the original data were generated by at least 6 different pathologists with variable experi- ence in data interpretation, indicating the critical contribution of the 2010 scoring guidelines, used in both cases, in raising the standards of ER and PR testing by harmonizing interpreta- tion. High correlation was also found between RT-qPCR and IHC; however, some cases were discordant as expected when comparing different methodologies and target molecules [27]. Notably, discrepant tumors expressed lower amounts of hormone receptor protein, while cor- responding RNA levels were also near the cut-off probably the result of the association between weak (but not negative) biological signals and uncertainty of measurements. Discor- dance was overall somewhat higher for progesterone receptor, consistent with this marker’s higher quantitative variability [28]. Obviously, the bimodal frequency distribution of estrogen receptor contributed to the high degree of correlation between RNA and protein data, whereas the more heterogeneous natural distribution of progesterone receptor probably allows for more inter-section variability (which may have contributed to some of the discordance between IHC and RT-qPCR) eventually inflating the dependency of IHC assessments on rather perceptual and cognitive functions [29]. In fact, anecdotal evidence from daily pathol- ogy practice suggests that the concern over missed treatment opportunities prompts many pathologists to categorize breast tumors as ER-positive, even when staining of nuclei does not formally meet the 1% criterion, despite the fact that low ER and/or PR expression has repeat- edly been linked to non-luminal biology and endocrine resistance [30–34]. Interestingly, most discrepancies for progesterone receptor in our study were scored low by IHC and were found negative at the RNA level, which is consistent with the view that, in tumors reacting weakly to antibodies against hormone receptors, gene expression profiles are indicative of more aggres- sive molecular subtypes [35]. These observations seem to confirm that the RNA testing utility better suited to quantitative biological continuity of hormone receptors, as advocated by some authors; however, the results about it were not always consistent [36–38]. Despite slight improvements, reliable assessment of Ki67 remains beyond reach in most parts of the world, owing mainly to the heterogeneity in the expression of this marker on tissue sections, which results in reduced reproducibility particularly in the mid-range of observations [39, 40]. Discussion Increasing the accuracy of breast cancer diagnosis to deliver the most precise and effective therapy is a continuous process involving on one end efforts to improve the performance of conventional methodologies and on the other end attempts to introduce technologically and operationally advanced analytical alternatives. The aim of our study was to explore discrepan- cies in the assessment of ER, PR, HER2 and Ki67 by comparing routinely obtained IHC (and Fig 4. HER2 IHC/FISH discrepancies. Schematic analysis of discrepancies in HER2: a. between the original assessment (ORI) and re-evaluation (REV) and b. between IHC (ORI) and RT-qPCR. Positive calls are indicated by red filled boxes. Negatives calls are presented in green and equivocal calls in grey, according to the 2013 ASCO/CAP guidelines in force at the time of the study. https://doi.org/10.1371/journal.pone.0255580.g004 Fig 4. HER2 IHC/FISH discrepancies. Schematic analysis of discrepancies in HER2: a. between the original assessment (ORI) and re-evaluation (REV) and b. between IHC (ORI) and RT-qPCR. Positive calls are indicated by red filled boxes. Negatives calls are presented in green and equivocal calls in grey, according to the 2013 ASCO/CAP guidelines in force at the time of the study. ncies. Schematic analysis of discrepancies in HER2: a. between the original assessment (ORI) and re-evaluation (REV) and b. R. Positive calls are indicated by red filled boxes. Negatives calls are presented in green and equivocal calls in grey, according s in force at the time of the study. PLOS ONE \| https://doi.org/10.1371/journal.pone.0255580 September 23, 2021 PLOS ONE \| https://doi.org/10.1371/journal.pone.0255580 September 23, 2021 11 / 18 PLOS ONE RT-qPCR assessment of breast cancer biomarkers Fig 5. Subtype redistribution after re-evaluation (REV) (A), semi-automated analysis (DIA), (B) and RT-qPCR measurement (C). Fig 5. Subtype redistribution after re-evaluation (REV) (A), semi-automated analysis (DIA), (B) and RT-qPCR Fig 5. Subtype redistribution after re-evaluation (REV) (A), semi-automated analysis (DIA), (B) and RT-qPCR measurement (C). https://doi.org/10.1371/journal.pone.0255580.g005 https://doi.org/10.1371/journal.pone.0255580.g005 https://doi.org/10.1371/journal.pone.0255580.g005 12 / 18 PLOS ONE \| https://doi.org/10.1371/journal.pone.0255580 September 23, 2021 PLOS ONE RT-qPCR assessment of breast cancer biomarkers FISH) data first with the results of manual and semi-automated re-evaluation of the original IHC slides and then with RNA expression data from the same tissue block using MammaTy- per1. PLOS ONE \| https://doi.org/10.1371/journal.pone.0255580 September 23, 2021 Discussion Meanwhile, experts remain reluctant to endorse new technologies, as intensive efforts for the standardization of existing methods have only recently started to payoff [8]. In our lab, the quality of HER2 testing improved upon the adoption of another analytical platform, which may explain discrepancies between original assessments and re-evaluation for cases that were re-stained with the current laboratory set-up that produces less background cytoplasmic staining. The use of the 2013 guidelines in the re- evaluation resulted in more equivocal cases, as has been already described [43]. Importantly, 3 FISH-positive cases with discrepant RT-qPCR results were tested independently and found to be equivocal, indicating that RNA may have helped identify potential false-positives by our own in-house FISH methodology. This finding is of importance, as it raises questions regard- ing the use of FISH as the “gold-standard” [44]. Although the present study presented a limita- tion having used outdated guidelines concerning the HER2 ISH interpretation [4] and the current recommendations [5, 12] do not longer provide an equivocal result for the ISH test, the RT-qPCR could be helpful in the interpretation of cases for which the second reader ISH analysis and the activation of the internal procedure to resolve the issue [5, 12] would never- theless not reach a confident result. Another solution for the standardization of breast cancer biomarkers and particularly for Ki67could be represented by automation with the help of image analysis platforms. In our study, performing DIA was labor-intensive, because regions of interest had to be manually selected due to insufficient performance of automatic detection of tumor nuclei; additional sig- nificant delays were caused by several software failures. In addition, we were not able to docu- ment any clear benefit from semi-automated scoring compared with manual re-evaluation. Nonetheless, recent studies have shown highly promising degree of reproducibility for compu- tational scoring of Ki67 on TMAs or biopsies that outperforms conventional manual analyses even when the latter are carried out under highly standardized conditions [45–47]. Therefore, computational techniques may hold real potential for improving the reliability of quantitative determinations compared with the human eye despite shortcomings that affect routine usabil- ity. On the other hand, automated quantification of staining is similarly limited by the short dynamic range of chromogen-based IHC and thus the information that is harvested for prog- nostic or predictive purposes is less than what can be obtained by RT-qPCR [48]. Discussion Our data clearly suggest that guideline-driven formal counting of nuclei, either per- formed manually or with the help of imaging software, may result in considerable differences in Ki67 scores compared to quick “eyeballing”. Upon re-evaluation, the percentage of positive nuclei were assessed across several high-power fields accounting in this way for potential het- erogeneity, especially from the so-called “hot-spots”, which may have been disregarded or preferentially targeted in the original assessment. Similarly, better averaging of RNA expres- sion throughout the whole section may explain why MammaTyper1 resulted in more MKI67-positive cases compared with IHC, a consistent finding in previous studies involving the assay [22]. Discrepancies may also be explained by the fact that the study pathologist was blinded to clinical data, whereas original assessments would be expected to have been far more biased, because, routinely, microscopic evidence is integrated with other available sources of information to reach an appropriate interpretation, including treatment preferences commu- nicated over tumor boards [41, 42]. As expected, differences in Ki67 between original IHC data and MammaTyper1 resulted in discrepancies in the St. Gallen classification, leading to a significant redistribution of the luminal subtypes in favour of the Luminal B-like category. Such redistributions were previously reported in the FinHer trial, in which MKI67 was supe- rior to IHC in predicting patient outcome and detecting interactions between an RNA-based St. Gallen molecular classification and treatment [22]. Owing to the clinical diversity of the PLOS ONE \| https://doi.org/10.1371/journal.pone.0255580 September 23, 2021 13 / 18 September 23, 2021 PLOS ONE RT-qPCR assessment of breast cancer biomarkers selected cases in this study, such effects were not investigated, as any associations between bio- marker groups and outcomes would have been extremely difficult to interpret due to multiple confounding factors and limited events. As opposed to controversies surrounding the use of Ki67, IHC for HER2 supplemented by FISH to resolve equivocal cases consistently receive strong votes of confidence across most jurisdictions [13]. HER2 is a critical biomarker, signal- ing for an aggressive form of breast cancer and, when present in the tissues of patients, it allows them to access effective, but also potentially toxic treatment with HER2 blocking agents. Due to its importance, interpretation rules for HER2 IHC and FISH results are periodically revised to address persisting or newly identified challenges [3–5]. PLOS ONE \| https://doi.org/10.1371/journal.pone.0255580 September 23, 2021 Discussion Another sig- nificant aspect of the routine or investigational use of DIA is the sample drop-outrate. In our study, no test was lost neither by MammaTyper1 nor by DIA; however, folding or twisting of tissue has been reported by others as a reason for considerable sample loss during automated analyses [49]. The decision between DIA and RT-qPCR for biomarker testing in breast cancer is further complicated by deployment conditions and pre-analytical requirements. At a first glance, RT-qPCR appears to represent a more realistic solution, due to savings in diagnostic time, broad accessibility of RT-qPCR instruments across laboratories, strong clinical validity and affordability. However, we are aware that the availability of these instruments is not wide- spread and, mainly for economic reasons, most of the pathology departments will continue to use immunohistochemistry to evaluate the expression of these biopathological parameters, obviously adhering carefully to external quality controls and specific guidelines. On the other PLOS ONE \| https://doi.org/10.1371/journal.pone.0255580 September 23, 2021 14 / 18 PLOS ONE RT-qPCR assessment of breast cancer biomarkers hand, the coarseness of the cutoffs currently used could benefit from more refined methods of analysis, such as gene expression profiles. Our study has some limitations. Although clinic-pathological data including follow-up data were available for all patients, we were unable to show meaningful associations between discor- dance and patient or tumor characteristics probably because of the highly heterogeneous nature of the cohort. For the same reason we reckoned that outcome statistics would be of lim- ited value. In addition, sample size was a limiting factor necessitating random selection of approximately 100 cases among all those that initially fulfilled our selection criteria. Moreover, we did not calculate agreement between methods due to the highly artificial constellation of challenging and/or ambiguous cases, which would prevent any attempt to extend such findings to other clinical/diagnostic settings. Finally, the study used outdated guidelines for the inter- pretation of IHC and FISH results concerning HER2 evaluation, according to the recommen- dations in force at the time of the analysis. In conclusion, the degree of correlation between IHC and RT-qPCR is high and compares well with the correlation between original with subsequent independent manual or semi-auto- mated IHC assessments. Intrinsic marker properties such as the type of protein and RNA fre- quency distributions or spatial heterogeneity in whole sections may interact with interpretation bias to shape the extent of inter-observer or inter-method variability. Acknowledgments We would like to thank Sotirios Lakis for providing medical writing services. S1 Data. (XLSX) S1 Data. (XLSX) S2 Data. (XLSX) Discussion The use of methods with wider dynamic range and higher reproducibility such as RT-qPCR may offer more precise assessment of endocrine responsiveness, improve Ki67 standardization and help resolve HER2 cases that present a difficult interpretation by IHC/FISH. In summary, our results seem to configure RT-qPCR as a complementary method to be used in all cases of equivocal results or close to the cut-off values. References 1. Weigelt B, Baehner FL, Reis-Filho JS. The contribution of gene expression profiling to breast cancer classification, prognostication and prediction: a retrospective of the last decade. J Pathol. 2010; 220 (2):263–80. https://doi.org/10.1002/path.2648 PMID: 19927298 2. Duffy MJ, Harbeck N, Nap M, Molina R, Nicolini A, Senkus E, et al. Clinical use of biomarkers in breast cancer: Updated guidelines from the European Group on Tumor Markers (EGTM). Eur J Cancer. 2017; 75:284–98. https://doi.org/10.1016/j.ejca.2017.01.017 PMID: 28259011 3. Wolff AC, Hammond ME, Schwartz JN, Hagerty KL, Allred DC, Cote RJ, et al. American Society of Clin- ical Oncology/College of American Pathologists guideline recommendations for human epidermal growth factor receptor 2 testing in breast cancer. J Clin Oncol. 2007; 25(1):118–45. https://doi.org/10. 1200/JCO.2006.09.2775 PMID: 17159189 4. Wolff AC, Hammond ME, Hicks DG, Dowsett M, McShane LM, Allison KH, et al. Recommendations for human epidermal growth factor receptor 2 testing in breast cancer: American Society of Clinical Oncol- ogy/College of American Pathologists clinical practice guideline update. J Clin Oncol. 2013; 31 (31):3997–4013. https://doi.org/10.1200/JCO.2013.50.9984 PMID: 24101045 5. Wolff AC, Hammond MEH, Allison KH, Harvey BE, Mangu PB, Bartlett JMS, et al. Human Epidermal Growth Factor Receptor 2 Testing in Breast Cancer: American Society of Clinical Oncology/College of American Pathologists Clinical Practice Guideline Focused Update. J Clin Oncol. 2018; 36(20):2105– 22. https://doi.org/10.1200/JCO.2018.77.8738 PMID: 29846122 6. Hammond ME, Hayes DF, Dowsett M, Allred DC, Hagerty KL, Badve S, et al. American Society of Clini- cal Oncology/College of American Pathologists guideline recommendations for immunohistochemical testing of estrogen and progesterone receptors in breast cancer (unabridged version). Arch Pathol Lab Med. 2010; 134(7):e48–72. https://doi.org/10.5858/134.7.e48 PMID: 20586616 7. Khoury T. Delay to Formalin Fixation (Cold Ischemia Time) Effect on Breast Cancer Molecules. Am J Clin Pathol. 2018; 149(4):275–92. https://doi.org/10.1093/ajcp/aqx164 PMID: 29471352 8. Pfitzner BM, Lederer B, Lindner J, Solbach C, Engels K, Rezai M, et al. Clinical relevance and concor- dance of HER2 status in local and central testing-an analysis of 1581 HER2-positive breast carcinomas over 12 years. Mod Pathol. 2018; 31(4):607–15. https://doi.org/10.1038/modpathol.2017.171 PMID: 29271415 9. Troxell ML, Long T, Hornick JL, Ambaye AB, Jensen KC. Comparison of Estrogen and Progesterone Receptor Antibody Reagents Using Proficiency Testing Data. Arch Pathol Lab Med. 2017; 141 (10):1402–12. https://doi.org/10.5858/arpa.2016-0497-OA PMID: 28714765 10. Denkert C, Budczies J, von Minckwitz G, Wienert S, Loibl S, Klauschen F. Strategies for developing Ki67 as a useful biomarker in breast cancer. Breast. 2015; 24 Suppl 2:S67–72. https://doi.org/10.1016/ j.breast.2015.07.017 PMID: 26283598 11. Visualization: Emanuele Caselli, Guido Bellezza. Visualization: Emanuele Caselli, Guido Bellezza. Writing – original draft: Emanuele Caselli. Writing – review & editing: Martina Mandarano, Kerstin Hartmann, Mark Laible. Author Contributions Conceptualization: Emanuele Caselli, Cristina Pelliccia, Valeria Teti, Ivana Ferri, Angelo Sidoni. Data curation: Emanuele Caselli, Ivana Ferri, Chiara Lupi, Fabrizio Stracci. Formal analysis: Emanuele Caselli, Guido Bellezza, Mark Laible. Funding acquisition: Ugur Sahin. Methodology: Cristina Pelliccia, Valeria Teti, Ivana Ferri. Project administration: Angelo Sidoni. Software: Emanuele Caselli, Martina Mandarano, Mark Laible. Supervision: Angelo Sidoni. Data curation: Emanuele Caselli, Ivana Ferri, Chiara Lupi, Fabrizio Stracci. Formal analysis: Emanuele Caselli, Guido Bellezza, Mark Laible. Formal analysis: Emanuele Caselli, Guido Bellezza, Mark Laible. Methodology: Cristina Pelliccia, Valeria Teti, Ivana Ferri. Project administration: Angelo Sidoni. Software: Emanuele Caselli, Martina Mandarano, Mark Laible. 15 / 18 PLOS ONE \| https://doi.org/10.1371/journal.pone.0255580 September 23, 2021 PLOS ONE \| https://doi.org/10.1371/journal.pone.0255580 September 23, 2021 PLOS ONE RT-qPCR assessment of breast cancer biomarkers Validation: Emanuele Caselli, Valeria Teti, Guido Bellezza, Ivana Ferri, Zsuzsanna Varga, Angelo Sidoni. Validation: Emanuele Caselli, Valeria Teti, Guido Bellezza, Ivana Ferri, Zsuzsanna Varga, Angelo Sidoni. References Dowsett M, Nielsen TO, A’Hern R, Bartlett J, Coombes RC, Cuzick J, et al. Assessment of Ki67 in breast cancer: recommendations from the International Ki67 in Breast Cancer working group. J Natl Cancer Inst. 2011; 103(22):1656–64. https://doi.org/10.1093/jnci/djr393 PMID: 21960707 21. Laible M, Schlombs K, Kaiser K, Veltrup E, Herlein S, Lakis S, et al. Technical validation of an RT- qPCR in vitro diagnostic test system for the determination of breast cancer molecular subtypes by quan- tification of ERBB2, ESR1, PGR and MKI67 mRNA levels from formalin-fixed paraffin-embedded breast tumor specimens. BMC cancer. 2016; 16:398. https://doi.org/10.1186/s12885-016-2476-x PMID: 27389414 22. Wirtz RM, Sihto H, Isola J, Heikkila P, Kellokumpu-Lehtinen PL, Auvinen P, et al. Biological subtyping of early breast cancer: a study comparing RT-qPCR with immunohistochemistry. Breast Cancer Res Treat. 2016; 157(3):437–46. https://doi.org/10.1007/s10549-016-3835-7 PMID: 27220750 23. Gisev N, Bell JS, Chen TF. Interrater agreement and interrater reliability: key concepts, approaches, and applications. Res Social Adm Pharm. 2013; 9(3):330–8. https://doi.org/10.1016/j.sapharm.2012. 04.004 PMID: 22695215 24. Eliasziw M, Young SL, Woodbury MG, Fryday-Field K. Statistical methodology for the concurrent assessment of interrater and intrarater reliability: using goniometric measurements as an example. Phys Ther. 1994; 74(8):777–88. https://doi.org/10.1093/ptj/74.8.777 PMID: 8047565 25. Goldhirsch A, Winer EP, Coates AS, Gelber RD, Piccart-Gebhart M, Thurlimann B, et al. Personalizing the treatment of women with early breast cancer: highlights of the St Gallen International Expert Con- sensus on the Primary Therapy of Early Breast Cancer 2013. Ann Oncol. 2013; 24(9):2206–23. https:// doi.org/10.1093/annonc/mdt303 PMID: 23917950 26. Curigliano G, Burstein HJ, Winer EP, Gnant M, Dubsky P, Loibl S, et al. De-escalating and escalating treatments for early-stage breast cancer: the St. Gallen International Expert Consensus Conference on the Primary Therapy of Early Breast Cancer 2017. Ann Oncol. 2017; 28(8):1700–12. https://doi.org/10. 1093/annonc/mdx308 PMID: 28838210 27. Liu Y, Beyer A, Aebersold R. On the Dependency of Cellular Protein Levels on mRNA Abundance. Cell. 2016; 165(3):535–50. https://doi.org/10.1016/j.cell.2016.03.014 PMID: 27104977 28. Nadji M, Gomez-Fernandez C, Ganjei-Azar P, Morales AR. Immunohistochemistry of estrogen and pro- gesterone receptors reconsidered: experience with 5,993 breast cancers. Am J Clin Pathol. 2005; 123 (1):21–7. https://doi.org/10.1309/4wv79n2ghj3x1841 PMID: 15762276 29. Muftah AA, Aleskandarany M, Sonbul SN, Nolan CC, Diez Rodriguez M, Caldas C, et al. Further evi- dence to support bimodality of oestrogen receptor expression in breast cancer. Histopathology. 2017; 70(3):456–65. https://doi.org/10.1111/his.13089 PMID: 27648723 30. Yi M, Huo L, Koenig KB, Mittendorf EA, Meric-Bernstam F, Kuerer HM, et al. Which threshold for ER positivity? References Koren S, Bentires-Alj M. Breast Tumor Heterogeneity: Source of Fitness, Hurdle for Therapy. Mol Cell. 2015; 60(4):537–46. https://doi.org/10.1016/j.molcel.2015.10.031 PMID: 26590713 12. Wolff AC, Hammond MEH, Allison KH, Harvey BE, Mangu PB, Bartlett JMS, et al. Human Epidermal Growth Factor Receptor 2 Testing in Breast Cancer: American Society of Clinical Oncology/College of American Pathologists Clinical Practice Guideline Focused Update. Arch Pathol Lab Med. 2018; 142 (11):1364–82. https://doi.org/10.5858/arpa.2018-0902-SA PMID: 29846104 13. Mittendorf EA, Bartlett JMS, Lichtensztajn DL, Chandarlapaty S. Incorporating Biology Into Breast Can- cer Staging: American Joint Committee on Cancer, Eighth Edition, Revisions and Beyond. Am Soc Clin Oncol Educ Book. 2018; 38:38–46. https://doi.org/10.1200/EDBK_200981 PMID: 30231409 14. Varga Z, Lebeau A, Bu H, Hartmann A, Penault-Llorca F, Guerini-Rocco E, et al. An international repro- ducibility study validating quantitative determination of ERBB2, ESR1, PGR, and MKI67 mRNA in breast cancer using MammaTyper(R). Breast Cancer Res. 2017; 19(1):55. https://doi.org/10.1186/ s13058-017-0848-z PMID: 28490348 15. Kos Z, Dabbs DJ. Biomarker assessment and molecular testing for prognostication in breast cancer. Histopathology. 2016; 68(1):70–85. https://doi.org/10.1111/his.12795 PMID: 26768030 PLOS ONE \| https://doi.org/10.1371/journal.pone.0255580 September 23, 2021 16 / 18 September 23, 2021 PLOS ONE RT-qPCR assessment of breast cancer biomarkers 16. Evans AJ, Bauer TW, Bui MM, Cornish TC, Duncan H, Glassy EF, et al. US Food and Drug Administra- tion Approval of Whole Slide Imaging for Primary Diagnosis: A Key Milestone Is Reached and New Questions Are Raised. Arch Pathol Lab Med. 2018; 142(11):1383–7. https://doi.org/10.5858/arpa. 2017-0496-CP PMID: 29708429 16. Evans AJ, Bauer TW, Bui MM, Cornish TC, Duncan H, Glassy EF, et al. US Food and Drug Administra- tion Approval of Whole Slide Imaging for Primary Diagnosis: A Key Milestone Is Reached and New Questions Are Raised. Arch Pathol Lab Med. 2018; 142(11):1383–7. https://doi.org/10.5858/arpa. 2017-0496-CP PMID: 29708429 17. Robertson S, Azizpour H, Smith K, Hartman J. Digital image analysis in breast pathology-from image processing techniques to artificial intelligence. Transl Res. 2018; 194:19–35. https://doi.org/10.1016/j. trsl.2017.10.010 PMID: 29175265 18. Lakhani SR, Ellis IO, Schnitt SJ, Tan PH, van de Vijver MJ. WHO Classification of Tumours of the Breast. 4th ed. Lyon: International Agency for Research on Cancer; 2012. 19. Lester SC BS, Chen Y-Y, Connolly JL, de Baca ME, Fitzgibbons PL, Hayes DF, et al. Protocol for Examination of Specimens From Patients With Invasive Carcinoma of the Breast. 2016 Jan [Cited 2019 Mar 01]. Available from: https://documents.cap.org/protocols/cp-breast-invasive-2016-v3300.pdf 20. References a retrospective study based on 9639 patients. Ann Oncol. 2014; 25(5):1004–11. https://doi. org/10.1093/annonc/mdu053 PMID: 24562447 31. Balduzzi A, Bagnardi V, Rotmensz N, Dellapasqua S, Montagna E, Cardillo A, et al. Survival outcomes in breast cancer patients with low estrogen/progesterone receptor expression. Clin Breast Cancer. 2014; 14(4):258–64. https://doi.org/10.1016/j.clbc.2013.10.019 PMID: 24325948 32. Sheffield BS, Kos Z, Asleh-Aburaya K, Wang XQ, Leung S, Gao D, et al. Molecular subtype profiling of invasive breast cancers weakly positive for estrogen receptor. Breast Cancer Res Treat. 2016; 155 (3):483–90. https://doi.org/10.1007/s10549-016-3689-z PMID: 26846986 33. Liao GS, Dai MS, Hsu HM, Chu CH, Hong ZJ, Fu CY, et al. Survival outcome of weak estrogen/proges- terone receptor expression in HER2 negative breast cancer is similar to triple negative breast cancer. Eur J Surg Oncol. 2017; 43(10):1855–61. https://doi.org/10.1016/j.ejso.2017.07.001 PMID: 28756019 34. Chen T, Zhang N, Moran MS, Su P, Haffty BG, Yang Q. Borderline ER-Positive Primary Breast Cancer Gains No Significant Survival Benefit From Endocrine Therapy: A Systematic Review and Meta-Analy- sis. Clin Breast Cancer. 2018; 18(1):1–8. https://doi.org/10.1016/j.clbc.2017.06.005 PMID: 28712925 PLOS ONE \| https://doi.org/10.1371/journal.pone.0255580 September 23, 2021 17 / 18 September 23, 2021 PLOS ONE RT-qPCR assessment of breast cancer biomarkers 35. Iwamoto T, Booser D, Valero V, Murray JL, Koenig K, Esteva FJ, et al. Estrogen receptor (ER) mRNA and ER-related gene expression in breast cancers that are 1% to 10% ER-positive by immunohis- tochemistry. J Clin Oncol. 2012; 30(7):729–34. https://doi.org/10.1200/JCO.2011.36.2574 PMID: 22291085 35. Iwamoto T, Booser D, Valero V, Murray JL, Koenig K, Esteva FJ, et al. Estrogen receptor (ER) mRNA and ER-related gene expression in breast cancers that are 1% to 10% ER-positive by immunohis- tochemistry. J Clin Oncol. 2012; 30(7):729–34. https://doi.org/10.1200/JCO.2011.36.2574 PMID: 22291085 36. Mackey JR. Can quantifying hormone receptor levels guide the choice of adjuvant endocrine therapy for breast cancer? J Clin Oncol. 2011; 29(12):1504–6. https://doi.org/10.1200/JCO.2010.34.3202 PMID: 21422423 37. Kim C, Tang G, Pogue-Geile KL, Costantino JP, Baehner FL, Baker J, et al. Estrogen receptor (ESR1) mRNA expression and benefit from tamoxifen in the treatment and prevention of estrogen receptor-pos- itive breast cancer. J Clin Oncol. 2011; 29(31):4160–7. https://doi.org/10.1200/JCO.2010.32.9615 PMID: 21947828 38. Dowsett M, Cuzick J, Wale C, Forbes J, Mallon EA, Salter J, et al. Prediction of risk of distant recurrence using the 21-gene recurrence score in node-negative and node-positive postmenopausal patients with breast cancer treated with anastrozole or tamoxifen: a TransATAC study. J Clin Oncol. 2010; 28 (11):1829–34. https://doi.org/10.1200/JCO.2009.24.4798 PMID: 20212256 39. References Laenkholm AV, Grabau D, Moller Talman ML, Balslev E, Bak Jylling AM, Tabor TP, et al. An inter- observer Ki67 reproducibility study applying two different assessment methods: on behalf of the Danish Scientific Committee of Pathology, Danish breast cancer cooperative group (DBCG). Acta Oncol. 2018; 57(1):83–9. https://doi.org/10.1080/0284186X.2017.1404127 PMID: 29202622 40. Varga Z, Cassoly E, Li Q, Oehlschlegel C, Tapia C, Lehr HA, et al. Standardization for Ki-67 assess- ment in moderately differentiated breast cancer. A retrospective analysis of the SAKK 28/12 study. PloS one. 2015; 10(4):e0123435. https://doi.org/10.1371/journal.pone.0123435 PMID: 25885288 41. Dunstan RW, Wharton KA Jr., Quigley C, Lowe A. The use of immunohistochemistry for biomarker assessment—can it compete with other technologies? Toxicol Pathol. 2011; 39(6):988–1002. https:// doi.org/10.1177/0192623311419163 PMID: 21971934 42. Fandel TM, Pfnur M, Schafer SC, Bacchetti P, Mast FW, Corinth C, et al. Do we truly see what we think we see? The role of cognitive bias in pathological interpretation. J Pathol. 2008; 216(2):193–200. https://doi.org/10.1002/path.2395 PMID: 18683853 43. Qian XL, Wen HY, Yang YL, Gu F, Guo XJ, Liu FF, et al. Assessment of dual-probe Her-2 fluorescent in situ hybridization in breast cancer by the 2013 ASCO/CAP guidelines produces more equivocal results than that by the 2007 ASCO/CAP guidelines. Breast Cancer Res Treat. 2016; 159(1):31–9. https://doi. org/10.1007/s10549-016-3917-6 PMID: 27455837 44. Bahreini F, Soltanian AR, Mehdipour P. A meta-analysis on concordance between immunohistochemis- try (IHC) and fluorescence in situ hybridization (FISH) to detect HER2 gene overexpression in breast cancer. Breast Cancer. 2015; 22(6):615–25. https://doi.org/10.1007/s12282-014-0528-0 PMID: 24718809 45. Acs B, Pelekanou V, Bai Y, Martinez-Morilla S, Toki M, Leung SCY, et al. Ki67 reproducibility using digi- tal image analysis: an inter-platform and inter-operator study. Lab Invest. 2019; 99(1):107–17. https:// doi.org/10.1038/s41374-018-0123-7 PMID: 30181553 46. Rimm DL, Leung SCY, McShane LM, Bai Y, Bane AL, Bartlett JMS, et al. An international multicenter study to evaluate reproducibility of automated scoring for assessment of Ki67 in breast cancer. Mod Pathol. 2019; 32(1):59–69. https://doi.org/10.1038/s41379-018-0109-4 PMID: 30143750 47. Klauschen F, Wienert S, Schmitt WD, Loibl S, Gerber B, Blohmer JU, et al. Standardized Ki67 Diagnos- tics Using Automated Scoring—Clinical Validation in the GeparTrio Breast Cancer Study. Clin Cancer Res. 2015; 21(16):3651–7. https://doi.org/10.1158/1078-0432.CCR-14-1283 PMID: 25501130 48. Meyerholz DK, Beck AP. Principles and approaches for reproducible scoring of tissue stains in research. Lab Invest. 2018; 98(7):844–55. https://doi.org/10.1038/s41374-018-0057-0 PMID: 29849125 49. Koopman T, Buikema HJ, Hollema H, de Bock GH, van der Vegt B. PLOS ONE \| https://doi.org/10.1371/journal.pone.0255580 September 23, 2021 49. Koopman T, Buikema HJ, Hollema H, de Bock GH, van der Vegt B. Digital image analysis of Ki67 prolif- eration index in breast cancer using virtual dual staining on whole tissue sections: clinical validation and inter-platform agreement. Breast Cancer Res Treat. 2018; 169(1):33–42. https://doi.org/10.1007/ s10549-018-4669-2 PMID: 29349710 References Digital image analysis of Ki67 prolif- eration index in breast cancer using virtual dual staining on whole tissue sections: clinical validation and inter-platform agreement. Breast Cancer Res Treat. 2018; 169(1):33–42. https://doi.org/10.1007/ s10549-018-4669-2 PMID: 29349710 18 / 18
https://openalex.org/W4383342991	https://umu.diva-portal.org/smash/get/diva2:1783435/FULLTEXT01	English	null	ASC- and caspase-1-deficient C57BL/6 mice do not develop demyelinating disease after infection with Theiler’s murine encephalomyelitis virus	Scientific reports	2,023	cc-by	9,670	Dandan Li 1, Melanie Bühler 1, Sandra Runft 1, Gisa Gerold 2,3,4,5, Katarzyna Marek 1, Wolfgang Baumgärtner 1, Till Strowig 6,7 & Ingo Gerhauser 1* Dandan Li 1, Melanie Bühler 1, Sandra Runft 1, Gisa Gerold 2,3,4,5, Katarzyna Marek 1 Wolfgang Baumgärtner 1, Till Strowig 6,7 & Ingo Gerhauser 1* Dandan Li 1, Melanie Bühler 1, Sandra Runft 1, Gisa Gerold 2,3,4,5, Katarzyna Marek 1, Wolfgang Baumgärtner 1, Till Strowig 6,7 & Ingo Gerhauser 1* Theiler’s murine encephalomyelitis virus (TMEV) induces an acute polioencephalomyelitis and a chronic demyelinating leukomyelitis in SJL mice. C57BL/6 (B6) mice generally do not develop TMEV- induced demyelinating disease (TMEV-IDD) due to virus elimination. However, TMEV can persist in specific immunodeficient B6 mice such as IFNβ−/− mice and induce a demyelinating process. The proinflammatory cytokines IL-1β and IL-18 are activated by the inflammasome pathway, which consists of a pattern recognition receptor molecule sensing microbial pathogens, the adaptor molecule Apoptosis-associated speck-like protein containing a CARD (ASC), and the executioner caspase-1. To analyze the contribution of the inflammasome pathway to the resistance of B6 mice to TMEV-IDD, ASC- and caspase-1-deficient mice and wild type littermates were infected with TMEV and investigated using histology, immunohistochemistry, RT-qPCR, and Western Blot. Despite the antiviral activity of the inflammasome pathway, ASC- and caspase-1-deficient mice eliminated the virus and did not develop TMEV-IDD. Moreover, a similar IFNβ and cytokine gene expression was found in the brain of immunodeficient mice and their wild type littermates. Most importantly, Western Blot showed cleavage of IL-1β and IL-18 in all investigated mice. Consequently, inflammasome- dependent activation of IL-1β and IL-18 does not play a major role in the resistance of B6 mice to TMEV-IDD. Theiler’s murine encephalomyelitis virus (TMEV) is a single-stranded RNA virus of the Picornaviridae family1. The intracerebral infection of SJL mice with TO subgroup strains of TMEV (BeAn, DA) induces a biphasic disease characterized by acute polioencephalomyelitis and chronic demyelinating leukomyelitis, which repre- sents an important animal model of human multiple sclerosis (MS)2–4. TMEV-induced demyelinating disease (TMEV-IDD) is initiated by oligodendrocyte apoptosis and axonal damage and aggravated by a delayed-type hypersensitivity response directed against viral antigens and later myelin components5–7. The virus mainly infects hippocampal and cortical neurons in the early phase of the disease but only persists in white matter glial cells and macrophages in advanced stages8. In contrast to SJL mice, C57BL/6 (B6) mice eliminate the virus from the central nervous system (CNS) within few weeks and do not develop TMEV-IDD9,10. www.nature.com/scientificreports www.nature.com/scientificreports www.nature.com/scientificreports Dandan Li 1, Melanie Bühler 1, Sandra Runft 1, Gisa Gerold 2,3,4,5, Katarzyna Marek 1, Wolfgang Baumgärtner 1, Till Strowig 6,7 & Ingo Gerhauser 1* Nevertheless, TMEV can persist in specific immunodeficient B6 mice such as interferon (IFN)-β knockout mice and even induce demy- elination in their spinal cord11. 1Department of Pathology, University of Veterinary Medicine Hannover, Foundation, Bünteweg 17, 30559 Hannover, Germany. 2Department of Biochemistry, University of Veterinary Medicine Hannover, Foundation, Bünteweg 17, 30559 Hannover, Germany. 3Research Center for Emerging Infections and Zoonoses (RIZ), University of Veterinary Medicine Hannover, Foundation, Bünteweg 17, 30559 Hannover, Germany. 4Wallenberg Centre for Molecular Medicine (WCMM), Umeå University, 90185 Umeå, Sweden. 5Department of Clinical Microbiology, Virology, Umeå University, 90185 Umeå, Sweden. 6Department for Microbial Immune Regulation, Helmholtz Centre for Infection Research, Inhoffenstraße 7, 38124 Braunschweig, Germany. 7Hannover Medical School, Carl‑Neuberg‑Straße 1, 30625 Hannover, Germany. email: Ingo.Gerhauser@tiho-hannover.de Scientific Reports \| (2023) 13:10960 \| https://doi.org/10.1038/s41598-023-38152-3 \| https://doi.org/10.1038/s41598-023-38152-3 Scientific Reports \| (2023) 13:10960 www.nature.com/scientificreports/ The proinflammatory cytokines IL-1β and IL-18 are produced as inactive precursors (Pro-IL-1β and Pro- IL-18), which are processed by caspase-1 and caspase-4/5/11 after activation of the inflammasome pathway12–17. Inflammasomes are vital players in innate immunity and involved in the complex host–pathogen-interactions. Typical inflammasomes represent multiprotein complexes composed of a pathogen sensor molecule such as nucleotide binding domain and leucine-rich repeat-containing receptor (NLR) or AIM2-like receptor (ALR), the adaptor molecule apoptosis-associated speck-like protein containing a caspase activation and recruitment domain (ASC) and the executioner caspase-118–21. These molecules are mainly expressed by microglia in the brain but astrocytes and neurons as well as myeloid cells infiltrating from the periphery can produce and activate inflamma- some pathway components22,23. Inflammasome activation prevents viral replication by inducing an inflammatory cell death program termed pyroptosis mediated by the pore-forming protein gasdermin D (GSDMD)18,24,25. More- over, the inflammasome pathway has an indirect antiviral effect due to the inhibition of the immunosuppressive effects of type I IFN (IFN-I) on T cells26,27. Correspondingly, the NLRP3 inflammasome pathway is critical for the control of West Nile virus (WNV) infections, because it inhibits WNV replication in neurons28,29. Nonetheless, wild type and caspase-1-deficient mice are equally susceptible to encephalomyocarditis virus (EMCV) and vesicu- lar stomatitis virus (VSV) infection despite the ability of the NLRP3 inflammasome to detect these viruses30. Interestingly, the promotion of T cell pathogenicity and CNS cell infiltration caused by inflammasome signaling contributes to the development of MS lesions31. Moreover, the inhibition of NLRP3 inflammasome activation by IFN-I plays a major role in the response of MS patients to IFNβ treatment32,33. Likewise, the NLRP3 inflamma- some can induce demyelinating lesions in experimental autoimmune encephalomyelitis (EAE), an animal model of MS, because it aggravates the infiltration of inflammatory immune cells into the CNS32. Nevertheless, EAE can even be induced in Nlrp3−/− and Asc−/− mice by an aggressive immunization using a myelin oligodendrocyte glycoprotein peptide in complete Freund’s adjuvant with 300 mg of heat-killed Mycobacteria. Similar to MS, the administration of IFNβ usually ameliorates EAE but in such NLRP3 inflammasome-independent EAE this treat- ment is not effective34. Asc−/− mice develop milder EAE lesions compared to Casp1−/− mice due to inflammasome- independent functions of ASC, which support the survival of CD4+ T cells including myelin oligodendrocyte glycoprotein–specific T cells35. \| https://doi.org/10.1038/s41598-023-38152-3 ASC also regulates the activity of the mitogen-activated protein kinase (MAPK) extracellular-signal regulated kinase (ERK) in murine and human monocytes/macrophages, antigen-specific IgG responses and chemokine and IFN-I expression independent of inflammasome activation28,36,37. Consequently, the different inflammasome pathway components can independently control inflammatory processes. fl p y p p yl y p ASC increases serum levels of IFNα, IFNγ, IL-1β, IL-6, CCL2, CCL5, CXCL1, and immunoglobulin M (IgM) during WNV infection thereby limiting viral replication. Nevertheless, elevated and not reduced levels of IFNγ, CCL2 and CCL5 were found in the brains from WNV-infected Asc−/− mice correlating with enhanced astrocyte activation, leukocyte infiltration and neuronal cell death28. Strong activation of the NLRP3 inflammasome and down-stream PGE2 signaling in dendritic cells and CD11b+ leukocytes, spleen cells and bone marrow cells of SJL mice also impairs early protective IFNγ-producing CD4+ and CD8+ T cell responses allowing viral persistence necessary for TMEV-IDD38. The exact role of specific inflammasome pathway components in the pathogenesis of TMEV-IDD has not been investigated so far. Therefore, the present study aimed to analyze the consequences of ASC or caspase-1 deficiency on virus elimination, CNS inflammation, and clinical outcome in TMEV-infected B6 mice. Moreover, the cytokine and IFN-I expression was analyzed in detail at 4 days post infection (dpi) due to the high number of T cells and B cells as well as peak of viral load in the brain of TMEV-infected B6 mice at this time point39–41. Results Cli i l Clinical and histological investigation. Clinical signs were absent in all investigated mice after TMEV infection. Moreover, mice continuously gained weight during the investigation period and their motor coordina- tion did not deteriorate (Fig. 1A,B). Asc−/− and Casp1−/− mice developed mild to moderate inflammatory brain and spinal cord lesions in the acute phase of the disease (4 and 14 dpi), which did not differ significantly from their wild type littermates (Fig. 1C–E). In the chronic phase of the disease (98 dpi) perivascular mononuclear cell infiltrates were also found in the spinal cord of Asc−/− and Casp1−/− mice as well as brain of Asc−/− mice but not in the brain of Casp1−/− mice (Fig. 1C,D). To illustrate a productive virus infection leading to demyelinating lesions in susceptible SJL but not in resistant B6 mice, clinical, histological, and RT-qPCR data of TMEV- and mock-infected SJL and B6 mice are summarized in Fig. 2. Immunohistochemistry. TMEV+ cells were mainly detected at 4 dpi in all investigated mouse groups (Fig. 3A,B). Virus antigen was only present in few cells of single animals at 14 dpi and not found at 98 dpi. The perivascular mononuclear cell infiltrates contained approximately 70% CD3+ T cells, 10% CD45R+ B cells and 20% Iba-1+ macrophages (Fig. 3C). No significant differences between Asc−/− and Casp1−/− mice and their respective wild type littermates were found in the composition of these leukocytes in the perivascular area. RT‑qPCR. RT-qPCR did not reveal significant differences in the amount of TMEV RNA as well as Ifnb1, Isg15, Eif2ak1 (PKR), Tnfa, Il1a, Il1b, Il6, Il10, Il12 (p40) and Ifng transcript numbers in the brain of Asc−/− and Casp1−/− mice and their respective wild type littermates at 4 days after TMEV infection (Fig. 4). No Il17 mRNA was found in the brain of these mice at this time point. Western Blot. Western blot analysis confirmed the expression of IL-1β and IL-18 proteins in the brain of TMEV-infected mice and wild type littermates at 4 dpi. Moreover, activation/cleavage of the precursor forms of these proinflammatory cytokines (Pro-IL-1β and Pro-IL-18) into the mature proteins was demonstrated in https://doi.org/10.1038/s41598-023-38152-3 Scientific Reports \| (2023) 13:10960 \| www.nature.com/scientificreports/ Figure 1. Clinical and histological data of Asc−/− and Casp1−/− mice (KO) and wild type littermates (WT) PFU of the BeAn strain of Theiler’s murine encephalomyelitis virus Weight analysis (A) and Rotarod perf Figure 1. Results Cli i l Clinical and histological data of Asc−/− and Casp1−/− mice (KO) and wild type littermates (WT) infected with 1 × 105 PFU of the BeAn strain of Theiler’s murine encephalomyelitis virus. Weight analysis (A) and Rotarod performance tests (B) showed a continuous increase in body weight and no deterioration in motor coordination in all mice. Semiquantitative scores of perivascular mononuclear cell infiltrates in the brain (C) and spinal cord (D) at 4, 14 and 98 days post infection (dpi). Cell infiltrates were analyzed in two complete cross sections of the brain at the levels of the hippocampus and cerebellum and complete cross sections of the cervical, thoracic and lumbar spinal cord. Mild to moderate inflammation was found at 4 and 14 dpi but not at 98 dpi in the brain, whereas mild inflammation was present in the spinal cord of at all investigated time points. No significant differences between KO and WT mice in the clinical and histological data were detected using Mann–Whitney tests. n = 7–10. Box plots with all data points. (E) Images demonstrate few perivascular mononuclear cells in the hippocampus and spinal cord of all mice at 4 dpi. Hematoxylin and eosin (HE) staining. Bars = 400 µm. Figure 1. Clinical and histological data of Asc−/− and Casp1−/− mice (KO) and wild type littermates PFU of the BeAn strain of Theiler’s murine encephalomyelitis virus Weight analysis (A) and Rotaro Figure 1. Clinical and histological data of Asc−/− and Casp1−/− mice (KO) and wild type littermates (WT) infected with 1 × 105 PFU of the BeAn strain of Theiler’s murine encephalomyelitis virus. Weight analysis (A) and Rotarod performance tests (B) showed a continuous increase in body weight and no deterioration in motor coordination in all mice. Semiquantitative scores of perivascular mononuclear cell infiltrates in the brain (C) and spinal cord (D) at 4, 14 and 98 days post infection (dpi). Cell infiltrates were analyzed in two complete cross sections of the brain at the levels of the hippocampus and cerebellum and complete cross sections of the cervical, thoracic and lumbar spinal cord. Mild to moderate inflammation was found at 4 and 14 dpi but not at 98 dpi in the brain, whereas mild inflammation was present in the spinal cord of at all investigated time points. No significant differences between KO and WT mice in the clinical and histological data were detected using Mann–Whitney tests. n = 7–10. Discussion ASC and caspase-1 are major components of the inflammasome pathway, which plays a critical role in several neuroinflammatory and neurodegenerative diseases including MS17,22,31,42. Inflammasome activation induces the maturation of the proinflammatory cytokines IL-1β, IL-18, and IL-33 by proteolytic cleavage of their precursor forms. IL-1β is involved in the acute phase response and stimulates the expression of adhesion molecules on endothelial cells and chemokines necessary for leukocyte recruitment. IL-1β also orchestrates the differentiation and function of innate and adaptive lymphoid cells43. For instance, the induction of eicosanoids such as PGE2 by IL-1β limits immunosuppressive effects of IFN-I on T cell functions26,44. IL-18 has been implicated in several autoimmune diseases including MS, because it stimulates γδ T cells to express IL-17 and promotes the produc- tion of the Th1 cytokine IFNγ from T cells and NK cells45. IL-33 is a potent inducer for a Th2 immune response, but this IL-1-like cytokine is inactivated after maturation by caspase-146,47.i yt y p However, the present study revealed no consequences of ASC or caspase-1 deficiency on virus elimination, CNS inflammation, cytokine and IFN-I expression and clinical outcome in TMEV-infected B6 mice. In contrast, the infection of immunodeficient IFN-β knockout mice (also B6 background) with the same virus strain and dosage resulted in virus persistence, increased cytokine expression and mild inflammatory demyelinating spinal cord lesions11. Lack of inflammasome-dependent IL-1β maturation might have been compensated by increased expression of IL-1α, which can be upregulated in TMEV-infected astrocytes and bind to the same receptor48–50. Nonetheless, no significant differences in Il1a mRNA levels were found between ASC- and caspase-1-deficient mice and their wild type littermates. Moreover, Tnfa, Il6, Il10, Il12, and Ifng mRNA levels were similar in all investigated mice at 4 dpi indicating that ASC- and caspase-1-deficiency does not influence cytokine expression in the brain of TMEV-infected B6 mice significantly. gi y Interestingly, Western blot analysis demonstrated the activation/cleavage of IL-1β and IL-18 proteins in both knockout mouse strains, which can be explained by alternative inflammasome-independent pathways of IL-1β and IL-18 maturation. In addition to microorganisms such as Candida albicans, Staphylococcus aureus and Streptococcus pyogenes, monocyte/macrophages, neutrophils, NK cells, mast cells and epithelial cells can produce proteases able to cleave pro-IL-1β including the serine proteases proteinase 3, cathepsin G, elastase, granzyme A, chymase, chymotrypsin as well as the metalloproteinases Meprin α and Meprin β16,51. Results Cli i l (E) Mononuclear cell infiltrates in the cervical spinal cord of a TMEV-infected SJL/J mouse at 98 dpi. Note perivascular mononuclear cells and gitter cells (arrows) in higher magnifications. Hematoxylin and eosin (HE) staining. Bars = 30 µm. (F) Demyelination in the inflamed ventrolateral funiculus of the spinal cord (same SJL/J mouse as in (F)). Note loss of luxol fast blue (LFB) staining due to severe inflammation (asterisk). Bar = 100 µm. (G) Immunohistochemistry was used to detect virus antigen in the spinal cord (same mouse as in (E) and (F)). Note TMEV+ cell (arrow) in the spinal cord white matter lesion. Avidin–biotin-complex (ABC) method using 3,3’-diaminobenzidine (DAB) as chromogen. Bar = 100 µm. (H) Semiquantitative scores of perivascular mononuclear cell infiltrates in the brain analyzed in a complete cross section of the cerebrum at the level of the hippocampus at 4 dpi. Moderate to strong inflammatory lesions were found in TMEV-infected SJL/J and C57BL/6 mice. (I) Immunohistochemistry was used to detect virus antigen in a complete cross section of the cerebrum at the level of the hippocampus at 4 dpi. No significant differences in the number of TMEV+ cells were found between SJL/J and C57BL/6 mice at 4 dpi using Mann–Whitney tests. n = 6. Box plots with all data points. ▸ Casp1+/+, Casp1−/−, Asc+/+ and Asc−/− mice as well as TMEV- and mock-infected SJL and B6 mice (Fig. 5, Suppl. Figure 1). Discussion Correspond- ingly, an increased expression of cathepsins and granzymes, which are released by monocytes/macrophages and NK cells respectively, has been described in the brain and spinal cord of TMEV-infected mice6,52. Consequently, the production of different serine proteases by NK cells and microglia/macrophages likely compensates lack of inflammasome-dependent regulation of IL-1-family cytokines during TMEV infection.ifl l g y y g In conclusion, ASC or caspase-1 deficiency had no effect on virus elimination, CNS inflammation, IFN-I and cytokine gene expression and clinical outcome in TMEV-infected B6 mice. The present results demonstrate that inflammasome-dependent pathways do not play a major role in the resistance of B6 mice to TMEV-IDD due to the activity of other proteases able to cleave IL-1β, IL-18, and IL-33. Nevertheless, future studies are needed to unravel the complex interactions of the inflammasome and other proinflammatory pathways in neuroinflam- matory diseases. Results Cli i l Box plots with all data points. (E) Images demonstrate few perivascular mononuclear cells in the hippocampus and spinal cord of all mice at 4 dpi. Hematoxylin and eosin (HE) staining. Bars = 400 µm. Figure 1. Clinical and histological data of Asc−/− and Casp1−/− mice (KO) and wild type littermates (WT) infected with 1 × 105 PFU of the BeAn strain of Theiler’s murine encephalomyelitis virus. Weight analysis (A) and Rotarod performance tests (B) showed a continuous increase in body weight and no deterioration in motor coordination in all mice. Semiquantitative scores of perivascular mononuclear cell infiltrates in the brain (C) and spinal cord (D) at 4, 14 and 98 days post infection (dpi). Cell infiltrates were analyzed in two complete cross sections of the brain at the levels of the hippocampus and cerebellum and complete cross sections of the cervical, thoracic and lumbar spinal cord. Mild to moderate inflammation was found at 4 and 14 dpi but not at 98 dpi in the brain, whereas mild inflammation was present in the spinal cord of at all investigated time points. No significant differences between KO and WT mice in the clinical and histological data were detected using Mann–Whitney tests. n = 7–10. Box plots with all data points. (E) Images demonstrate few perivascular mononuclear cells in the hippocampus and spinal cord of all mice at 4 dpi. Hematoxylin and eosin (HE) staining. Bars = 400 µm. https://doi.org/10.1038/s41598-023-38152-3 Scientific Reports \| (2023) 13:10960 \| www.nature.com/scientificreports/ Figure 2. Clinical disease and histological lesions in SJL/J and C57BL/6 mice mock-infected with cell culture medium or infected with 1.63 × 106 PFU of the BeAn strain of Theiler’s murine encephalomyelitis virus (TMEV). Weight analysis (A) and Rotarod performance tests (B) of TMEV-infected SJL/J showed a continuous increase in body weight but an impaired motor coordination at 28, 56, and 98 days post infection (dpi). No clinical signs were present in C57BL/6 mice. Mann–Whitney tests: P < 0.05; **P < 0.01. n = 6. (C) Semiquantitative scores of perivascular mononuclear cell infiltrates analyzed in complete cross sections of the cervical, thoracic and lumbar spinal cord. Severe inflammation only develops in SJL/J mice. n = 5–6. (D) TMEV RNA copy numbers in the spinal cord quantified with RT-qPCR. Note the logarithmic scale and high versus low/no virus replication in SJL/J and C57BL/6 mice at 28, 56 and 98 dpi, respectively. n = 5–6. Materials and methods Animal experiment. Asc−/− (B6.129S5-Pycardtm1Flv) mice (http://www.informatics.jax.org/allele/MGI: 3686870) and Casp1−/− (Casp1tm2.1Flv) were generated as described53–56. Briefly, Asc and Casp1 targeting vectors were electroporated into 129SvEvBrd Lex-1 and JM8 C57BL/6 embryonic stem cells, respectively. Offspring was backcrossed to B6 mice. Heterozygous knockout mice were used to breed 27 Asc−/− and 28 Casp1−/− mice Animal experiment. https://doi.org/10.1038/s41598-023-38152-3 Scientific Reports \| (2023) 13:10960 \| www.nature.com/scientificreports/ / p / and 60 wild type littermates. SJL/J and B6 mice were obtained from Harlan Winkelmann (now Envigo RMS GmbH, Düsseldorf, Germany). Mice were kept under controlled environmental conditions (22–24 °C; 50–60% humidity; 12/12 h light/dark cycle) with free access to standard rodent diet (R/M-H; sniff Spezialdiäten GmbH, i and 60 wild type littermates. SJL/J and B6 mice were obtained from Harlan Winkelmann (now Envigo RMS GmbH, Düsseldorf, Germany). Mice were kept under controlled environmental conditions (22–24 °C; 50–60% humidity; 12/12 h light/dark cycle) with free access to standard rodent diet (R/M-H; sniff Spezialdiäten GmbH, Soest, Germany) and tap water. At the age of five to six weeks, groups of seven to ten knockout mice and wild type littermates were inoculated into the right cerebral hemisphere with 1 × 105 plaque-forming units (PFU) per mouse of the BeAn strain of TMEV in 20 µl Dulbecco’s modified Eagle medium (DMEM; PAA Laboratories, Cölbe, Germany) with 2% fetal calf serum and 50 μg/kg gentamicin. This virus stock and dose has been used in several publications and is known to induce TMEV-IDD in SJL mice and in B6 mice11,39,57–60. To illustrate a productive viral infection and disease course, data of additional experiments using SJL and B6 mice infected with TMEV-BeAn (1.63 × 106 PFU) or mock-infected with DMEM were included in the study. Inoculation was and 60 wild type littermates. SJL/J and B6 mice were obtained from Harlan Winkelmann (now Envigo RMS GmbH, Düsseldorf, Germany). Mice were kept under controlled environmental conditions (22–24 °C; 50–60% humidity; 12/12 h light/dark cycle) with free access to standard rodent diet (R/M-H; sniff Spezialdiäten GmbH, Soest, Germany) and tap water. At the age of five to six weeks, groups of seven to ten knockout mice and wild type littermates were inoculated into the right cerebral hemisphere with 1 × 105 plaque-forming units (PFU) per mouse of the BeAn strain of TMEV in 20 µl Dulbecco’s modified Eagle medium (DMEM; PAA Laboratories, Cölbe, Germany) with 2% fetal calf serum and 50 μg/kg gentamicin. This virus stock and dose has been used in several publications and is known to induce TMEV-IDD in SJL mice and in B6 mice11,39,57–60. Animal experiment. Immunohistochemistry was used to detect virus antigen and determine percentages of perivascular immune cells in the brain of Asc−/− and Casp1−/− mice (KO) and wild type littermates (WT) infected with 1 × 105 PFU of the BeAn strain of Theiler’s murine encephalomyelitis virus (TMEV). (A) All TMEV+ cells were counted in a complete cross section of the cerebrum at the level of the hippocampus at 4, 14 and 98 days post infection (dpi). No significant differences in the number of TMEV+ cells were found between KO and WT mice using Mann–Whitney tests demonstrating rapid virus elimination. n = 7–10. (B) Images show TMEV+ cells in the CA1 and CA2 area of the hippocampus of KO and WT mice at 4 dpi. Avidin–biotin-complex (ABC) method using 3,3’-diaminobenzidine (DAB) as chromogen. Bars = 400 µm. (C) Percentages of perivascular CD3+ T cell, CD45R+ B cells and Iba-1+ macrophages at 4 and 7 dpi. No significant differences between KO and WT mice were detected using Mann–Whitney tests. n = 10 except Casp1−/− mice at 4 dpi (n = 9). Box plots with all data points. performed under general anesthesia using an intraperitoneal injection of medetomidine (0.5 mg/kg, Domitor; 50 µl for a mouse of 20 g; Pfizer, Karlsruhe, Germany) and ketamine (100 mg/kg, Ketamin 10%; 100 µl for a mouse of 20 g; WDT eG, Garbsen, Germany). General appearance, activity, and gait was evaluated weekly as pre- viously described61. Moreover, a rotarod assay (RotaRod Treadmill, TSE Technical & Scientific Equipment, Bad Homburg, Germany) was performed every week to test motor strength and control of TMEV-infected mice62. Animals were sacrificed at 4, 14 and 98 dpi using an intraperitoneal injection of an overdose of medetomidine (1 mg/kg, Domitor; 20 µl for a mouse of 20 g; Pfizer) and ketamine (200 mg/kg, Ketamin 10%; 40 µl for a mouse of 20 g; WDT eG). Segments of the brain and spinal cord of mice were removed immediately after death and either formalin-fixed and paraffin-embedded or snap-frozen in OCT embedding compound (Sakura Finetek Europe, Zoeterwoude, Netherlands) using liquid nitrogen63. Animal experiments were conducted in accordance with the German Animal Welfare Law and the ARRIVE guidelines64 and were authorized by the local govern- ment (Niedersächsisches Landesamt für Verbraucherschutz und Lebensmittelsicherheit, Oldenburg, Germany, permission numbers: 509c-42502-02/589, 509.6-42502-04/860, 33.12-42502-04-14/1656). Histological examination. Animal experiment. Immunohistochemistry was used to detect virus antigen and determine percentages of perivascular immune cells in the brain of Asc−/− and Casp1−/− mice (KO) and wild type littermates (WT) infected with 1 × 105 PFU of the BeAn strain of Theiler’s murine encephalomyelitis virus (TMEV). (A) All TMEV+ cells were counted in a complete cross section of the cerebrum at the level of the hippocampus at 4, 14 and 98 days post infection (dpi). No significant differences in the number of TMEV+ cells were found between KO and WT mice using Mann–Whitney tests demonstrating rapid virus elimination. n = 7–10. (B) Images show TMEV+ cells in the CA1 and CA2 area of the hippocampus of KO and WT mice at 4 dpi. Avidin–biotin-complex (ABC) method using 3,3’-diaminobenzidine (DAB) as chromogen. Bars = 400 µm. (C) Percentages of perivascular CD3+ T cell, CD45R+ B cells and Iba-1+ macrophages at 4 and 7 dpi. No significant differences between KO and WT mice were detected using Mann–Whitney tests. n = 10 except Casp1−/− mice at 4 dpi (n = 9). Box plots with all data points. Figure 3. Immunohistochemistry was used to detect virus antigen and determine percentages of perivascular immune cells in the brain of Asc−/− and Casp1−/− mice (KO) and wild type littermates (WT) infected with 1 × 105 PFU of the BeAn strain of Theiler’s murine encephalomyelitis virus (TMEV). (A) All TMEV+ cells were counted in a complete cross section of the cerebrum at the level of the hippocampus at 4, 14 and 98 days post infection (dpi). No significant differences in the number of TMEV+ cells were found between KO and WT mice using Mann–Whitney tests demonstrating rapid virus elimination. n = 7–10. (B) Images show TMEV+ cells in the CA1 and CA2 area of the hippocampus of KO and WT mice at 4 dpi. Avidin–biotin-complex (ABC) method using 3,3’-diaminobenzidine (DAB) as chromogen. Bars = 400 µm. (C) Percentages of perivascular CD3+ T cell, CD45R+ B cells and Iba-1+ macrophages at 4 and 7 dpi. No significant differences between KO and WT mice were detected using Mann–Whitney tests. n = 10 except Casp1−/− mice at 4 dpi (n = 9). Box plots with all data points. Figure 3. Animal experiment. To illustrate a productive viral infection and disease course, data of additional experiments using SJL and B6 mice infected with TMEV-BeAn (1.63 × 106 PFU) or mock-infected with DMEM were included in the study. Inoculation was and 60 wild type littermates. SJL/J and B6 mice were obtained from Harlan Winkelmann (now Envigo R https://doi.org/10.1038/s41598-023-38152-3 Scientific Reports \| (2023) 13:10960 \| www.nature.com/scientificreports/ w.nature.com/scientificreports/ performed under general anesthesia using an intraperitoneal injection of medetomidine (0.5 mg/kg, Domitor; 50 µl for a mouse of 20 g; Pfizer, Karlsruhe, Germany) and ketamine (100 mg/kg, Ketamin 10%; 100 µl for a mouse of 20 g; WDT eG, Garbsen, Germany). General appearance, activity, and gait was evaluated weekly as pre- viously described61. Moreover, a rotarod assay (RotaRod Treadmill, TSE Technical & Scientific Equipment, Bad Homburg, Germany) was performed every week to test motor strength and control of TMEV-infected mice62. Animals were sacrificed at 4, 14 and 98 dpi using an intraperitoneal injection of an overdose of medetomidine (1 mg/kg, Domitor; 20 µl for a mouse of 20 g; Pfizer) and ketamine (200 mg/kg, Ketamin 10%; 40 µl for a mouse of 20 g; WDT eG). Segments of the brain and spinal cord of mice were removed immediately after death and either formalin-fixed and paraffin-embedded or snap-frozen in OCT embedding compound (Sakura Finetek Europe, Zoeterwoude, Netherlands) using liquid nitrogen63. Animal experiments were conducted in accordance with the German Animal Welfare Law and the ARRIVE guidelines64 and were authorized by the local govern- ment (Niedersächsisches Landesamt für Verbraucherschutz und Lebensmittelsicherheit, Oldenburg, Germany, permission numbers: 509c-42502-02/589, 509.6-42502-04/860, 33.12-42502-04-14/1656). Histological examination. Two micrometers paraffin sections of brain (cerebrum and cerebellum) and spinal cord (cervical, thoracic and lumbar) were routinely stained with hematoxylin and eosin (HE). The degree of inflammation was evaluated in two complete cross sections of the brain at the levels of the hippocampus and cerebellum and complete cross sections of the cervical, thoracic and lumbar spinal cord using a semiquantitative scoring system of perivascular mononuclear cell infiltrates (0 = normal, 1 = single inflammatory cells, 2 = 2–3 layers of infiltrates, 3 = more than 3 layers of infiltrates surrounding meningeal and parenchymal vessels) after blinding6,10,61,65. Figure 3. Animal experiment. Moreover, all TMEV+ cells were counted in a serial section at this level. Polymerase chain reaction. RNA has been isolated from a complete cross section of the OCT-embedded cerebrum at the level of the hippocampus. Real-time quantitative polymerase chain reaction (RT-qPCR) was performed for TMEV, Ifnb1, Isg15, Eif2ak1, Tnfa, Il1a, Il1b, Il6, Il10, Il12 (p40), Il17, Ifng and three housekeeping Figure 4. The number of Theiler’s murine encephalomyelitis virus (TMEV) RNA copies/ng RNA as well as Ifnb1, Isg15, Eif2ak1 (PKR), Tnfa, Il1a, Il1b, Il6, Il10, Il12 (p40) and Ifng mRNA copies /ng RNA was quantified in the brain of Asc−/− and Casp1−/− mice (KO) and wild type littermates (WT) infected with 1 × 105 PFU of the BeAn strain of TMEV at 4 days post infection using RT-qPCR. No significant differences between Asc−/− and Casp1−/− mice and their respective wild type littermates were found using Mann–Whitney tests. n = 8 (Asc+/+, Casp1+/+ and Casp1−/− mice), n = 7 (Asc−/− mice). Box plots with all data points. Figure 4. The number of Theiler’s murine encephalomyelitis virus (TMEV) RNA copies/ng RNA as well as Ifnb1, Isg15, Eif2ak1 (PKR), Tnfa, Il1a, Il1b, Il6, Il10, Il12 (p40) and Ifng mRNA copies /ng RNA was quantified in the brain of Asc−/− and Casp1−/− mice (KO) and wild type littermates (WT) infected with 1 × 105 PFU of the BeAn strain of TMEV at 4 days post infection using RT-qPCR. No significant differences between Asc−/− and Casp1−/− mice and their respective wild type littermates were found using Mann–Whitney tests. n = 8 (Asc+/+, Casp1+/+ and Casp1−/− mice), n = 7 (Asc−/− mice). Box plots with all data points. omyelitis virus (TMEV) RNA copies/ng RNA as well as Immunohistochemistry. Immunohistochemistry was performed as previously described6,10,66–68. Briefly, paraffin sections were blocked with 20% goat serum and stained with rabbit polyclonal antibodies directed against TMEV antigen (VP1; 1:2000), CD3 (T lymphocytes; Agilent Technologies Deutschland GmbH, Wald- bronn, Germany; A045201; 1:500) and Iba-1 (microglia/macrophages; Wako Chemicals GmbH, Neuss, Ger- many; 019-19741; 1:1000) or rat monoclonal antibody directed against CD45R (B lymphocytes; BD Biosciences, Heidelberg, Germany; clone RA3-6B2; 553085; 1:1000) overnight at 4 °C. Primary antibodies were replaced by rabbit serum (R4505, Merck KGaA, Darmstadt, Germany) or rat serum (R9759; Merck KGaA) as negative control. Biotinylated goat-anti-rabbit IgG (BA-1000, Vector Laboratories, Burlingame, CA, USA; 1:200) and rabbit-anti rat IgG (BA-4001; Vector Laboratories; 1:200) were used as secondary antibodies. Animal experiment. Two micrometers paraffin sections of brain (cerebrum and cerebellum) and spinal cord (cervical, thoracic and lumbar) were routinely stained with hematoxylin and eosin (HE). The degree of inflammation was evaluated in two complete cross sections of the brain at the levels of the hippocampus and cerebellum and complete cross sections of the cervical, thoracic and lumbar spinal cord using a semiquantitative scoring system of perivascular mononuclear cell infiltrates (0 = normal, 1 = single inflammatory cells, 2 = 2–3 layers of infiltrates, 3 = more than 3 layers of infiltrates surrounding meningeal and parenchymal vessels) after blinding6,10,61,65. https://doi.org/10.1038/s41598-023-38152-3 Scientific Reports \| (2023) 13:10960 \| www.nature.com/scientificreports/ Figure 4. The number of Theiler’s murine encephalomyelitis virus (TMEV) RNA copies/ng RNA as well as Ifnb1, Isg15, Eif2ak1 (PKR), Tnfa, Il1a, Il1b, Il6, Il10, Il12 (p40) and Ifng mRNA copies /ng RNA was quantified in the brain of Asc−/− and Casp1−/− mice (KO) and wild type littermates (WT) infected with 1 × 105 PFU of the BeAn strain of TMEV at 4 days post infection using RT-qPCR. No significant differences between Asc−/− and Casp1−/− mice and their respective wild type littermates were found using Mann–Whitney tests. n = 8 (Asc+/+, Casp1+/+ and Casp1−/− mice), n = 7 (Asc−/− mice). Box plots with all data points. Immunohistochemistry. Immunohistochemistry was performed as previously described6,10,66–68. Briefly, paraffin sections were blocked with 20% goat serum and stained with rabbit polyclonal antibodies directed against TMEV antigen (VP1; 1:2000), CD3 (T lymphocytes; Agilent Technologies Deutschland GmbH, Wald- bronn, Germany; A045201; 1:500) and Iba-1 (microglia/macrophages; Wako Chemicals GmbH, Neuss, Ger- many; 019-19741; 1:1000) or rat monoclonal antibody directed against CD45R (B lymphocytes; BD Biosciences, Heidelberg, Germany; clone RA3-6B2; 553085; 1:1000) overnight at 4 °C. Primary antibodies were replaced by rabbit serum (R4505, Merck KGaA, Darmstadt, Germany) or rat serum (R9759; Merck KGaA) as negative control. Biotinylated goat-anti-rabbit IgG (BA-1000, Vector Laboratories, Burlingame, CA, USA; 1:200) and rabbit-anti rat IgG (BA-4001; Vector Laboratories; 1:200) were used as secondary antibodies. Immunolabeling was visualized by the avidin–biotin-peroxidase complex (ABC) method (PK-6100, Vector Laboratories) with 3,3-diaminobenzidine (DAB, Merck KGaA) as substrate and slight counterstaining was performed using Mayer’s hematoxylin. Percentages of perivascular CD3+ T cells, CD45R+ B cells and Iba-1+ macrophages were deter- mined by randomly counting 100 cells in the Virchow-Robin space of vessels present in one cross section of the cerebrum at the level of the hippocampus. Animal experiment. Western Blot was used to detect cleavage of the inactive precursors of the proinflammatory cytokines in the brain of Asc−/− and Casp1−/− mice (KO) and wild type littermates (WT) infected with 1 × 105 PFU of the BeAn strain of Theiler’s murine encephalomyelitis virus (TMEV). Brain samples of SJL/J and C57BL/6 mice mock-infected with cell culture medium or infected with 1.63 × 106 PFU of the TMEV-BeAn were also included. (A) A cleavage of the precursor Pro-IL-1β to the mature IL-1β protein was found in all brain samples. (B) Similarly, a cleavage of the precursor Pro-IL-18 to the mature IL-18 protein was demonstrated in all brain samples despite low protein levels in Asc−/− mice. Detection of Lamin B1 served as the loading control. genes (Actb, Gadph, Hprt1) using standard protocols, the AriaMx Real-Time PCR system (Agilent Technologies Deutschland GmbH), and Brilliant III Ultra-Fast SYBR® QPCR Master Mixes as described49,66,68,69. Tenfold serial dilution standards were used to quantify the results. Experimental data were normalized using a normalization factor calculated from the three housekeeping genes70. Specificity of each reaction was controlled by melting curve analysis. genes (Actb, Gadph, Hprt1) using standard protocols, the AriaMx Real-Time PCR system (Agilent Technologies Deutschland GmbH), and Brilliant III Ultra-Fast SYBR® QPCR Master Mixes as described49,66,68,69. Tenfold serial dilution standards were used to quantify the results. Experimental data were normalized using a normalization factor calculated from the three housekeeping genes70. Specificity of each reaction was controlled by melting curve analysis. Western blot. Proteins have been isolated from a complete cross section of the OCT-embedded cerebrum at the level of the hippocampus. Cells were lysed in radioimmunoprecipitation assay (RIPA) buffer (1 mM phe- nylmethylsulfonyl fluoride, 1% sodium deoxycholate, 50 mM Tris–HCl [pH 7.4], 1% Triton X-100, 0.1% SDS, 150 mM NaCl) with added protease inhibitor cocktail (cOmplete, EDTA free; Roche Diagnostics GmbH, Man- nheim, Germany). The total protein concentration of each sample was determined using the Pierce™ BCA™ Protein-Assay (Thermo Fisher, Rockford, IL, USA) according to the manufacture’s instructions. 20 µg of total protein were separated by SDS–polyacrylamide gel electrophoresis followed by transfer to polyvinylidene dif- luoride (PVDF) membranes (Novex™). Rabbit polyclonal antibodies were used to detect IL-1β (P420B; Thermo Fisher; 1:1000), IL-18 (PA5-79481; Thermo Fisher; 1:1000), and Laminin (PA5-19468; Thermo Fisher; 1:1000). Animal experiment. Immunolabeling was visualized by the avidin–biotin-peroxidase complex (ABC) method (PK-6100, Vector Laboratories) with 3,3-diaminobenzidine (DAB, Merck KGaA) as substrate and slight counterstaining was performed using Mayer’s hematoxylin. Percentages of perivascular CD3+ T cells, CD45R+ B cells and Iba-1+ macrophages were deter- mined by randomly counting 100 cells in the Virchow-Robin space of vessels present in one cross section of the cerebrum at the level of the hippocampus. Moreover, all TMEV+ cells were counted in a serial section at this level. Polymerase chain reaction. RNA has been isolated from a complete cross section of the OCT-embedded cerebrum at the level of the hippocampus. Real-time quantitative polymerase chain reaction (RT-qPCR) was performed for TMEV, Ifnb1, Isg15, Eif2ak1, Tnfa, Il1a, Il1b, Il6, Il10, Il12 (p40), Il17, Ifng and three housekeeping https://doi.org/10.1038/s41598-023-38152-3 Scientific Reports \| (2023) 13:10960 \| www.nature.com/scientificreports/ Figure 5. Western Blot was used to detect cleavage of the inactive precursors of the proinflammatory cytokines in the brain of Asc−/− and Casp1−/− mice (KO) and wild type littermates (WT) infected with 1 × 105 PFU of the BeAn strain of Theiler’s murine encephalomyelitis virus (TMEV). Brain samples of SJL/J and C57BL/6 mice mock-infected with cell culture medium or infected with 1.63 × 106 PFU of the TMEV-BeAn were also included. (A) A cleavage of the precursor Pro-IL-1β to the mature IL-1β protein was found in all brain samples. (B) Similarly, a cleavage of the precursor Pro-IL-18 to the mature IL-18 protein was demonstrated in all brain samples despite low protein levels in Asc−/− mice. Detection of Lamin B1 served as the loading control. Figure 5. Western Blot was used to detect cleavage of the inactive precursors of the proinflammatory cytokines in the brain of Asc−/− and Casp1−/− mice (KO) and wild type littermates (WT) infected with 1 × 105 PFU of the BeAn strain of Theiler’s murine encephalomyelitis virus (TMEV). Brain samples of SJL/J and C57BL/6 mice mock-infected with cell culture medium or infected with 1.63 × 106 PFU of the TMEV-BeAn were also included. (A) A cleavage of the precursor Pro-IL-1β to the mature IL-1β protein was found in all brain samples. (B) Similarly a cleavage of the precursor Pro-IL-18 to the mature IL-18 protein was demonstrated in all brain Figure 5. Animal experiment. Then, blots were incubated with a peroxidase-coupled goat anti-rabbit secondary antiserum (32460; Thermo Fisher; 1:500) and the binding was visualized using SuperSignal™ West Femto Maximum Sensitivity Substrate (34095; Thermo Fisher) and a ChemiDoc™ MP Imaging System (BioRad). For band size determination PageR- uler™ Plus Prestained Protein Ladder (10 to 250 kDa; 26620; Thermo Fisher) was used. Statistical analysis. Statistical analysis was performed using Prism 6 (GraphPad Software, La Jolla, CA, USA). Mann–Whitney tests were used to compare clinical, histological, immunohistochemical and PCR data (knockout vs. wild type mice). P < 0.05 was considered as statistical significant. Referencesh 1. Theiler, M. Spontaneous encephalomyelitis of mice, a new virus disease. J. Exp. Med. 65, 705–719 (1937). 1. Theiler, M. Spontaneous encephalomyelitis of mice, a new virus disease. J. Exp. Med. 65, 705–719 (1937). Theiler, M. Spontaneous encephalomyelitis of mice, a new virus dise h 2. Mecha, M., Carrillo-Salinas, F. J., Mestre, L., Feliu, A. & Guaza, C. Viral models of multiple sclerosis: Neurodegeneration and demyelination in mice infected with Theiler’s virus. Prog. Neurobiol. 101–102, 46–64. https://doi.org/10.1016/j.pneurobio.2012. 11.003 (2013). ( ) 3. Oleszak, E. L., Chang, J. R., Friedman, H., Katsetos, C. D. & Platsoucas, C. D. Theiler’s virus infection: A model for multiple sclerosis Clin. Microbiol. Rev. 17, 174–207 (2004).hh 4. Tsunoda, I. et al. Three immune-mediated disease models induced by Theiler’s virus: Multiple sclerosis, seizures and myocarditis Clin. Exp. Neuroimmunol. 7, 330–345. https://doi.org/10.1111/cen3.12341 (2016).h p p g 5. Tsunoda, I. & Fujinami, R. S. Neuropathogenesis of Theiler’s murine encephalomyelitis virus infection, an animal model for multiple l i J N i Ph l 5 355 369 h //d i /10 1007/ 11481 009 9179 (2010) 5. Tsunoda, I. & Fujinami, R. S. Neuropathogenesis of Theiler’s murine encephalomyelitis virus infection, an animal model for multiple sclerosis. J. Neuroimmune Pharmacol. 5, 355–369. https://doi.org/10.1007/s11481-009-9179-x (2010). 5. Tsunoda, I. & Fujinami, R. S. Neuropathogenesis of Theilers murine encephalomyelitis virus infection, an animal model for multiple sclerosis. J. Neuroimmune Pharmacol. 5, 355–369. https://doi.org/10.1007/s11481-009-9179-x (2010). h harmacol. 5, 355–369. https://doi.org/10.1007/s11481-009-9179-x p g 6. Gerhauser, I. et al. Dynamic changes and molecular analysis of cell death in the spinal cord of SJL mice in strain of Theiler’s murine encephalomyelitis virus. Apoptosis 23, 170–186. https://doi.org/10.1007/s10495- 6. Gerhauser, I. et al. Dynamic changes and molecular analysis of cell death in the spinal cord of SJL mice infected with the BeAn strain of Theiler’s murine encephalomyelitis virus Apoptosis 23 170–186 https://doi org/10 1007/s10495-018-1448-9 (2018) h p y p p , p g ( ) 7. Miller, S. D. et al. Persistent infection with Theiler’s virus leads to CNS autoimmunity via epitope spreading. Nat. Med. 3, 1133–1136 (1997). ( ) 8. Gerhauser, I., Hansmann, F., Ciurkiewicz, M., Löscher, W. & Beineke, A. Facets of Theiler’s murine encephalomyelitis virus-induced diseases: An update. Int. J. Mol. Sci. 20, E448. https://doi.org/10.3390/ijms20020448 (2019).h p p g j 9. Lipton, H. L. & Dal Canto, M. C. Susceptibility of inbred mice to chronic central nervous system infection by Theiler’s murine encephalomyelitis virus. Infect. Immun. Data availabilityh y The datasets generated and analyzed during the current study can be obtained from the corresponding author on reasonable request. Received: 3 November 2022; Accepted: 4 July 2023 Received: 3 November 2022; Accepted: 4 July 2023 https://doi.org/10.1038/s41598-023-38152-3 Scientific Reports \| (2023) 13:10960 \| www.nature.com/scientificreports/ Referencesh 26, 369–374 (1979).l p y f 10. Gerhauser, I., Alldinger, S. & Baumgärtner, W. Ets-1 represents a pivotal transcription factor for viral clearance, inflammation, and demyelination in a mouse model of multiple sclerosis. J. Neuroimmunol. 188, 86–94. https://doi.org/10.1016/j.jneuroim.2007.05. 019 (2007).ih ( ) 1. Bühler, M. et al. IFN-beta deficiency results in fatal or demyelinating disease in C57BL/6 mice infected with Theiler’s murine encephalomyelitis viruses. Front. Immunol. 13, 786940. https://doi.org/10.3389/fimmu.2022.786940 (2022).h i 2. Thornberry, N. A. et al. A novel heterodimeric cysteine protease is required for interleukin-1 beta processing in monocytes. Nature 356, 768–774. https://doi.org/10.1038/356768a0 (1992).ii p g 3. Kostura, M. J. et al. Identification of a monocyte specific pre-interleukin 1 beta convertase activity. Proc. Natl. Acad. Sci. U.S.A. 86 5227–5231. https://doi.org/10.1073/pnas.86.14.5227 (1989). p g p ( ) 4. Gu, Y. et al. Activation of interferon-gamma inducing factor mediated by interleukin-1beta converting enzyme. Science 275 206–209. https://doi.org/10.1126/science.275.5297.206 (1997). . Wang, S. et al. Murine caspase-11, an ICE-interacting protease, 5. Wang, S. et al. Murine caspase-11, an ICE-interacting protease, is essential for the activation of ICE. Cell 92, 501–509 (1998). 6. Netea, M. G., van de Veerdonk, F. L., van der Meer, J. W., Dinarello, C. A. & Joosten, L. A. Inflammasome-independent regulatio 15. Wang, S. et al. Murine caspase 11, an ICE interacting protease, is essential for the activation of ICE. Cell 92, 501 509 (1998). 16. Netea, M. G., van de Veerdonk, F. L., van der Meer, J. W., Dinarello, C. A. & Joosten, L. A. Inflammasome-independent regulation of IL-1-family cytokines. Annu. Rev. Immunol. 33, 49–77. https://doi.org/10.1146/annurev-immunol-032414-112306 (2015).ll 16. Netea, M. G., van de Veerdonk, F. L., van der Meer, J. W., Dinarello, C. A. & Joosten, L. A. Inflammasome-independent regulation of IL-1-family cytokines. Annu. Rev. Immunol. 33, 49–77. https://doi.org/10.1146/annurev-immunol-032414-112306 (2015).ll y y p g 17. Singhal, G., Jaehne, E. J., Corrigan, F., Toben, C. & Baune, B. T. Inflammasomes in neuroinflammation an tion: a focused review. Front. Neurosci. 8, 315. https://doi.org/10.3389/fnins.2014.00315 (2014).l y y p g ( 17. Singhal, G., Jaehne, E. J., Corrigan, F., Toben, C. & Baune, B. T. Inflammasomes in neuroinflammation and changes in brain f i f d i F N i 8 31 h //d i /10 3389/f i 2014 0031 (2014) 17. Singhal, G., Jaehne, E. J., Corrigan, F., Toben, C. & Baune, B. T. Inflammasomes in neuroinflammation and changes in brain tion: a focused review. Front. Neurosci. 8, 315. Referencesh https://doi.org/10.3389/fnins.2014.00315 (2014).l p g 18. Shrivastava, G., Leon-Juarez, M., Garcia-Cordero, J., Meza-Sanchez, D. E. & Cedillo-Barron, L. Inflammasomes and its importance in viral infections. Immunol. Res. 64, 1101–1117. https://doi.org/10.1007/s12026-016-8873-z (2016).l 19. Chakraborty, S., Kaushik, D. K., Gupta, M. & Basu, A. Inflammasome signaling at the heart of central nervous system patho J. Neurosci. Res. 88, 1615–1631. https://doi.org/10.1002/jnr.22343 (2010).f p g j ( ) 0. Ransohoff, R. M. & Brown, M. A. Innate immunity in the central nervous system. J. Clin. Invest. 122, 1164–1171. https://doi.org/ 10.1172/JCI58644 (2012).l 21. Platnich, J. M. & Muruve, D. A. NOD-like receptors and inflammasomes: A review of their canonical and non-canonical signaling pathways. Arch. Biochem. Biophys. 670, 4–14. https://doi.org/10.1016/j.abb.2019.02.008 (2019).ill 2. Voet, S., Srinivasan, S., Lamkanfi, M. & van Loo, G. Inflammasomes in neuroinflammatory and neurodegenerative diseases. EMBO Mol. Med. https://doi.org/10.15252/emmm.201810248 (2019).l p g ( ) 3. Olcum, M., Tastan, B., Kiser, C., Genc, S. & Genc, K. Microglial NLRP3 inflammasome activation in multiple sclerosis. Adv. Protein Chem. Struct. Biol. 119, 247–308. https://doi.org/10.1016/bs.apcsb.2019.08.007 (2020). p g p 4. Shi, J., Gao, W. & Shao, F. Pyroptosis: Gasdermin-mediated programmed necrotic cell death. Trends Biochem. Sci. 42, 245–254 https://doi.org/10.1016/j.tibs.2016.10.004 (2017).l p g j 5. Shi, J. et al. Cleavage of GSDMD by inflammatory caspases determines pyroptotic cell death. Nature 526, 660–665. https://doi.org/ 10.1038/nature15514 (2015).f 6. Kim, B. S. et al. IL-1 signal affects both protection and pathogenesis of virus-induced chronic CNS demyelinating disease. J Neuroinflammation 9, 217. https://doi.org/10.1186/1742-2094-9-217 (2012). fl p g 7. McRae, B. L., Semnani, R. T., Hayes, M. P. & van Seventer, G. A. Type I IFNs inhibit human dendritic cell IL-12 production and Th1 cell development. J. Immunol. 160, 4298–4304 (1998).l h 28. Kumar, M. et al. Inflammasome adaptor protein apoptosis-associated speck-like protein containing CARD (ASC) is critical for the immune response and survival in west nile virus encephalitis. J. Virol. 87, 3655–3667. https://doi.org/10.1128/JVI.02667-12 (2013). ( ) 29. Ramos, H. J. et al. IL-1beta signaling promotes CNS-intrinsic immune control of West Nile virus infection. PLoS Patho e1003039. https://doi.org/10.1371/journal.ppat.1003039 (2012).hl p g j pp 30. Rajan, J. V., Rodriguez, D., Miao, E. A. & Aderem, A. The NLRP3 inflammasome detects encephalomyocarditis virus and vesi stomatitis virus infection. J. Virol. 85, 4167–4172. https://doi.org/10.1128/JVI.01687-10 (2011).l p g 31. Barclay, W. & Shinohara, M. L. Inflammasome activation in multiple sclerosis and experimental autoimmune encephalomyelitis (EAE). Brain Pathol. 27, 213–219. https://doi.org/10.1111/bpa.12477 (2017).h p g p 2. Inoue, M. Referencesh & Shinohara, M. L. The role of interferon-beta in the treatment of multiple sclerosis and experimental autoimmune encephalomyelitis—In the perspective of inflammasomes. Immunology 139, 11–18. https://doi.org/10.1111/imm.12081 (2013).l 32. Inoue, M. & Shinohara, M. L. The role of interferon-beta in the treatment of multiple sclerosis and experimental autoimmune encephalomyelitis—In the perspective of inflammasomes. Immunology 139, 11–18. https://doi.org/10.1111/imm.12081 (2013). 33. Malhotra, S. et al. NLRP3 inflammasome is associated with the response to IFN-beta in patients with multiple sclerosis. Brain 138, 644–652 https://doi org/10 1093/brain/awu388 (2015) h encephalomyelitis—In the perspective of inflammasomes. Immunology 139, 11–18. https://doi.org/10.1111/imm.12081 (2013). 33. Malhotra, S. et al. NLRP3 inflammasome is associated with the response to IFN-beta in patients with multiple sclerosis. Brain 138, 644–652. https://doi.org/10.1093/brain/awu388 (2015).f l 3. Malhotra, S. et al. NLRP3 inflammasome is associated with the response to IFN-beta in patients with multiple sclerosis. Brain 138 644–652. https://doi.org/10.1093/brain/awu388 (2015).f p g ( ) 34. Inoue, M. et al. Interferon-beta therapy against EAE is effective only when development of the disease depends on the NL inflammasome. Sci. Signal 5, ra38. https://doi.org/10.1126/scisignal.2002767 (2012). l g p g g 5. Shaw, P. J. et al. Cutting edge: critical role for PYCARD/ASC in the development of experimental autoimmune encephalomyelitis J. Immunol. 184, 4610–4614. https://doi.org/10.4049/jimmunol.1000217 (2010).l p g j 36. Ellebedy, A. H. et al. Inflammasome-independent role of the apoptosis-associated speck-like protein containing CARD (ASC) in the adjuvant effect of MF59. Proc. Natl. Acad. Sci. U.S.A. 108, 2927–2932. https://doi.org/10.1073/pnas.1012455108 (2011). jf , p g p ( ) 7. Taxman, D. J. et al. The NLR adaptor ASC/PYCARD regulates DUSP10, mitogen-activated protein kinase (MAPK), and chemokine i d i i d d f h i fl J Bi l Ch 6 6 6 6 h //d i / /jb M ( ) jf p g p ( ) 37. Taxman, D. J. et al. The NLR adaptor ASC/PYCARD regulates DUSP10, mitogen-activated protein kinase (MAPK), and chemokine induction independent of the inflammasome. J. Biol. Chem. 286, 19605–19616. https://doi.org/10.1074/jbc.M111.221077 (2011).h jf p g p 37. Taxman, D. J. et al. The NLR adaptor ASC/PYCARD regulates DUSP10, mitogen-activated protein kinase (MAPK), an induction independent of the inflammasome J Biol Chem 286, 19605–19616 https://doi org/10 1074/jbc M111 2 f 37. Taxman, D. J. et al. The NLR adaptor ASC/PYCARD regulates DUSP10, mitogen-activated protein kinase (MAPK), and chemokine induction independent of the inflammasome. J. Biol. Chem. 286, 19605–19616. https://doi.org/10.1074/jbc.M111.221077 (2011). 38. Kim, S. J., Jin, Y. H. www.nature.com/scientificreports/ Induction of activator protein-1 and nuclear factor-kappaB as a prereq- uisite for disease development in susceptible SJL/J mice after theiler murine encephalomyelitis. J. Neuropathol. Exp. Neurol. 66, 809–818. https://doi.org/10.1097/nen.0b013e3181461f31 (2007). p g 66. Li, L., Ulrich, R., Baumgärtner, W. & Gerhauser, I. Interferon-stimulated genes-essential antiviral effectors implicated in resistance to Theiler’s virus-induced demyelinating disease. J. Neuroinflammation 12, 242. https://doi.org/10.1186/s12974-015-0462-x (2015). f ld k d h f l l l b d f h y g fl p g 67. Kummerfeld, M., Meens, J., Haas, L., Baumgärtner, W. & Beineke, A. Generation and characterization of a polyclonal antibody for the detection of Theiler’s murine encephalomyelitis virus by light and electron microscopy. J. Virol. Methods 160, 185–188. https:// doi.org/10.1016/j.jviromet.2009.04.030 (2009). g j j 8. Gerhauser, I., Alldinger, S., Ulrich, R. & Baumgärtner, W. Spatio-temporal expression of immediate early genes in the centra nervous system of SJL/J mice. Int. J. Dev. Neurosci. 23, 637–649. https://doi.org/10.1016/j.ijdevneu.2005.06.004 (2005). 9 W ltl I t l Mi li h t ti l i i l h liti i d d i d l t d hi l d B i 69. Waltl, I. et al. Microglia have a protective role in viral encephalitis-induced seizure deve Behav. Immun. 74, 186–204. https://doi.org/10.1016/j.bbi.2018.09.006 (2018). 9. Waltl, I. et al. Microglia have a protective role in viral encephalitis-induced seizure development and hippocampal damage. Brain Behav. Immun. 74, 186–204. https://doi.org/10.1016/j.bbi.2018.09.006 (2018). 70. Vandesompele, J. et al. Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes. Genome Biol. https://doi.org/10.1186/gb-2002-3-7-research0034 (2002).lh g g g 1. Li, D. Role of programmed cell death and the inflammasome pathway in Theiler’s murine encephalomyelitis. PhD thesis, https:// elib.tiho-hannover.de/servlets/MCRFileNodeServlet/tiho_derivate_00000806/Lid-ws00000820.pdf (2020). www.nature.com/scientificreports/ www.nature.com/scientificreports/ 40. Prajeeth, C. K. et al. Limited role of regulatory T cells during acute Theiler virus-induced encephalitis in resistant C57BL/6 mice. J. Neuroinflammation 11, 180. https://doi.org/10.1186/s12974-014-0180-9 (2014). fl p g 1. Kummerfeld, M. et al. Periventricular demyelination and axonal pathology is associated with subependymal virus spread in a murine model for multiple sclerosis. Intervirology 55, 401–416. https://doi.org/10.1159/000336563 (2012). y murine model for multiple sclerosis. Intervirology 55, 401–416. ht p gy p g 42. Noroozi, S., Meimand, H. A. E., Arababadi, M. K., Nakhaee, N. & Asadikaram, G. The effects of IFN-beta 1a on the expression of inflammasomes and apoptosis-associated speck-like proteins in multiple sclerosis patients. Mol. Neurobiol. 54, 3031–3037. https:// doi.org/10.1007/s12035-016-9864-8 (2017).h g ( ) 3. Garlanda, C., Dinarello, C. A. & Mantovani, A. The interleukin-1 family: Back to the future. Immunity 39, 1003–1018. https://doi org/10.1016/j.immuni.2013.11.010 (2013). g j 4. Mayer-Barber, K. D. et al. Host-directed therapy of tuberculosis based on interleukin-1 and type I interferon crosstalk. Nature 511 99–103. https://doi.org/10.1038/nature13489 (2014). p g 5. Dinarello, C. A., Novick, D., Kim, S. & Kaplanski, G. Interleukin-18 and IL-18 binding protein. Front. Immunol. 4, 289. https:// doi.org/10.3389/fimmu.2013.00289 (2013).ht gi 6. Cayrol, C. & Girard, J. P. The IL-1-like cytokine IL-33 is inactivated after maturation by caspase-1. Proc. Natl. Acad. Sci. U. S. A 106, 9021–9026. https://doi.org/10.1073/pnas.0812690106 (2009).l p g p 7. Abd Rachman Isnadi, M. F. et al. Critical roles of IL-33/ST2 pathway in neurological disorders. Mediators Inflamm. 2018, 5346413 https://doi.org/10.1155/2018/5346413 (2018).fh p g ( ) 48. Rubio, N. & Capa, L. Differential IL-1 synthesis by astrocytes from Theiler’s murine encephalomyelitis virus-susceptible and -resistant strains of mice. Cell. Immunol. 149, 237–247. https://doi.org/10.1006/cimm.1993.1151 (1993).h p g 49. Gerhauser, I. et al. Theiler’s murine encephalomyelitis virus induced phenotype switch of microglia in vitro. J. Neuroimmunol. 49–55. https://doi.org/10.1016/j.jneuroim.2012.07.018 (2012). g j j 0. Rubio, N. Demonstration of the presence of an interleukin-1 receptor on the surface of murine astrocytes and its regulation by cytokines and Theiler’s virus. Immunology 82, 178–183 (1994).li h 51. Fantuzzi, G. et al. Response to local inflammation of IL-1 beta-converting enzyme- deficient mice. J. Immunol. 158, 1818–1824 (1997). 52. Ciurkiewicz, M. et al. Transcriptome analysis following neurotropic virus infection reveals faulty innate immunity and delayed antigen presentation in mice susceptible to virus-induced demyelination. Brain Pathol. 31, e13000. https://doi.org/10.1111/bpa. 13000 (2021).fl 53. Mariathasan, S. et al. Differential activation of the inflammasome by caspase-1 adaptors ASC and Ipaf. Nature 430, 213–218. Referencesh & Kim, B. S. Prostaglandin E2 produced following infection with Theiler’s virus promotes the pathogenesis of demyelinating disease PLoS ONE 12 e0176406 https //doi org/10 1371/journal pone 0176406 (2017) 7. Taxman, D. J. et al. The NLR adaptor ASC/PYCARD regulates DUSP10, mitogen-activated protein kinase (MAPK), and chemokine induction independent of the inflammasome. J. Biol. Chem. 286, 19605–19616. https://doi.org/10.1074/jbc.M111.221077 (2011) 8. Kim, S. J., Jin, Y. H. & Kim, B. S. Prostaglandin E2 produced following infection with Theiler’s virus promotes the pathogenesis o demyelinating disease. PLoS ONE 12, e0176406. https://doi.org/10.1371/journal.pone.0176406 (2017). induction independent of the inflammasome. J. Biol. Chem. 286, 19605 19616. https://doi.org/10.1074/jbc.M111.221077 (2011) 8. Kim, S. J., Jin, Y. H. & Kim, B. S. Prostaglandin E2 produced following infection with Theiler’s virus promotes the pathogenesis o demyelinating disease. PLoS ONE 12, e0176406. https://doi.org/10.1371/journal.pone.0176406 (2017). y g p g j p ( ) 39. Ciurkiewicz, M. et al. Cytotoxic CD8(+) T cell ablation enhances the capacity of regulatory T cells to delay viral elimination in Theiler’s murine encephalomyelitis. Brain Pathol. 28, 349–368. https://doi.org/10.1111/bpa.12518 (2018). https://doi.org/10.1038/s41598-023-38152-3 Scientific Reports \| (2023) 13:10960 \| www.nature.com/scientificreports/ https:// doi.org/10.1038/nature02664 (2004). g 4. Sutterwala, F. S. et al. Critical role for NALP3/CIAS1/Cryopyrin in innate and adaptive immunity through its regulation of cas- pase-1. Immunity 24, 317–327. https://doi.org/10.1016/j.immuni.2006.02.004 (2006). 54. Sutterwala, F. S. et al. Critical role for NALP3/CIAS1/Cryopyrin in innate and adaptive immunity through its regulatio pase-1. Immunity 24, 317–327. https://doi.org/10.1016/j.immuni.2006.02.004 (2006). 55. Blazejewski, A. J. et al. Microbiota normalization reveals that canonical caspase-1 activation exacerbates chemically induced intestinal inflammation. Cell Rep. 19, 2319–2330. https://doi.org/10.1016/j.celrep.2017.05.058 (2017).l l 56. Case, C. L. et al. Caspase-11 stimulates rapid flagellin-independent pyroptosis in response to Legionella pneumophila. Proc. Natl. Acad. Sci. U.S.A. 110, 1851–1856. https://doi.org/10.1073/pnas.1211521110 (2013).l l U.S.A. 110, 1851–1856. https://doi.org/10.1073/pnas.1211521110 ( g 7. Allnoch, L., Baumgärtner, W. & Hansmann, F. Impact of astrocyte depletion upon inflammation and demyelination in a murine animal model of multiple sclerosis. Int. J. Mol. Sci. https://doi.org/10.3390/ijms20163922 (2019).f g j 58. Bröer, S. et al. Viral mouse models of multiple sclerosis and epilepsy: Marked differences in neuropathogenesis following infection with two naturally occurring variants of Theiler’s virus BeAn strain. Neurobiol. Dis. 99, 121–132. https://doi.org/10.1016/j.nbd. 2016.12.020 (2017).lf ( ) 59. Bröer, S. et al. Brain inflammation, neurodegeneration and seizure development following picornavirus infection markedly differ among virus and mouse strains and substrains. Exp. Neurol. 279, 57–74. https://doi.org/10.1016/j.expneurol.2016.02.011 (2016).i 60. Uhde, A. K. et al. Viral infection of the central nervous system exacerbates interleukin-10 receptor deficiency-mediated colitis in SJL mice. PLoS ONE 11, e0161883. https://doi.org/10.1371/journal.pone.0161883 (2016). 1. Ulrich, R. et al. MMP-12, MMP-3, and TIMP-1 are markedly upregulated in chronic demyelinating theiler murine encephalomy elitis. J. Neuropathol. Exp. Neurol. 65, 783–793. https://doi.org/10.1097/01.jnen.0000229990.32795.0d (2006).i p p p g j 62. Ulrich, R., Kalkuhl, A., Deschl, U. & Baumgärtner, W. Machine learning approach identifies new pathways associated with demy- elination in a viral model of multiple sclerosis. J. Cell. Mol. Med. 14, 434–448. https://doi.org/10.1111/j.1582-4934.2008.00646.x (2010).l ( ) 63. Markus, S., Failing, K. & Baumgärtner, W. Increased expression of pro-inflammatory cytokines and lack of up-regulation of anti- inflammatory cytokines in early distemper CNS lesions. J. Neuroimmunol. 125, 30–41 (2002). 64 Percie du Sert N et al The ARRIVE guidelines 2 0 Updated guidelines for reporting animal research J Cereb Blood Flow Metab 64. Percie du Sert, N. et al. The ARRIVE guidelines 2.0: Updated guidelines for reporting animal research. J. Cereb. Blood Flow M 40, 1769–1777. https://doi.org/10.1177/0271678X20943823 (2020). p g 65. Gerhauser, I., Ulrich, R., Alldinger, S. & Baumgärtner, W. Acknowledgements g Parts of this article have been included into the PhD thesis of D.L.71. The authors thank Julia Baskas, Petra Grünig, Heike Kanapin, Angela Karl, Christiane Namneck, Caroline Schütz, Friederike Walters, and Danuta Waschke for excellent technical assistance. Asc−/− mice were developed by Millennium Pharmaceuticals, Inc. (Cambridge, MA, USA) and kindly provided by Prof. Richard A. Flavell (Yale University, New Haven, CT, USA). Casp1−/− mice were also kindly provided by Prof. Richard A. Flavell. https://doi.org/10.1038/s41598-023-38152-3 Scientific Reports \| (2023) 13:10960 \| www.nature.com/scientificreports/ Author contributions Conceptualization: W.B., T.S. and I.G.; Data curation: D.L., M.B., S.R. and K.M.; Formal analysis: D.L. and I.G.; Funding acquisition: W.B. and I.G.; Investigation: D.L., M.B., S.R, K.M. and I.G.; Methodology: G.G., K.M., T.S. and I.G.; Project administration: W.B. and I.G.; Resources: G.G., T.S. and W.B.; Supervision: W.B. and I.G.; Vali- dation: G.G. and I.G.; Visualization: D.L. and I.G.; Writing – original draft: D.L.; Writing – review & editing: I.G. Funding g Open Access funding enabled and organized by Projekt DEAL. This work was supported by the Niedersachsen- Research Network on Neuroinfectiology (N-RENNT) of the Ministry of Science and Culture of Lower Saxony. D.L. was funded by the China Scholarship Council (File No. 201606170128). This Open Access publication was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation)-491094227 “Open Access Publication Funding” and the University of Veterinary Medicine Hannover, Foundation. © The Author(s) 2023 Additional informationh Additional information Supplementary Information The online version contains supplementary material available at https://doi.org/ 10.1038/s41598-023-38152-3. Correspondence and requests for materials should be addressed to I.G. Competing interests h The authors declare no competing interests. Reprints and permissions information is available at www.nature.com/reprints. Reprints and permissions information is available at www.nature.com/reprints. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and nstitutional affiliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. © The Author(s) 2023 https://doi.org/10.1038/s41598-023-38152-3 Scientific Reports \| (2023) 13:10960 \|
https://openalex.org/W4288038040	https://www.researchsquare.com/article/rs-1719969/latest.pdf	English	null	Time to use modern contraceptives and associated factors among women of reproductive age in Ethiopia: survival analysis of the 2016 Ethiopian demographic and health survey data	Research Square (Research Square)	2,022	cc-by	8,049	Time to use modern contraceptives and associated factors among women of reproductive age in Ethiopia: survival analysis of the 2016 Ethiopian demographic and health survey data Shibabaw Bekere Addis Ababa University Tilahun Asena (  feredetilahun14@gmail.com ) Arba Minch University Dejen Tesfaw Addis Ababa University Asmare Wube Ethiopian Public Health Institute Meseret Alemayehu Department of pediatrics, Pawi General Hospital, Time to use modern contraceptives and associated factors among women of reproductive age in Ethiopia: survival analysis of the 2016 Ethiopian demographic and health survey data Shibabaw Bekere1, Tilahun Asena2, Dejen Tesfaw1, Asmare Wube3 and Meseret Alemayehu4 1Department of Statistics, Addis Ababa University, Ethiopia 2Department of Statistics, Arba Minch University, Ethiopia 3Ethiopian Public Health Institute, Addis Ababa, Ethiopia 4Department of Pediatrics, Pawi General Hospital, Ethiopia Corresponding author name and email address: Tilahun Asena - feredetilahun14@gmail.com Shibabaw Bekere1, Tilahun Asena2, Dejen Tesfaw1, Asmare Wube3 and Meseret Alemayehu4 1Department of Statistics, Addis Ababa University, Ethiopia 2Department of Statistics, Arba Minch University, Ethiopia 3Ethiopian Public Health Institute, Addis Ababa, Ethiopia 4Department of Pediatrics, Pawi General Hospital, Ethiopia * *Corresponding author name and email address: Tilahun Asena - feredetilahun14@gmail.com Research Article Keywords: Contraceptive, Survival Model, Frailty, Log-Normal Posted Date: July 26th, 2022 DOI: https://doi.org/10.21203/rs.3.rs-1719969/v2 License:   This work is licensed under a Creative Commons Attribution 4.0 International License. Read Full License Time to use modern contraceptives and associated factors among women of reproductive age in Ethiopia: survival analysis of the 2016 Ethiopian demographic and health survey data Abstract Background: Contraceptive methods are one of the most important health interventions of the twenty-first century, and they are critical in slowing population growth and improving women's and family health. Contraceptive methods are now a global issue in general, and in developing countries in particular. Method: This study sought to identify the factors associated with the timing of the use of modern contraceptive methods among Ethiopian women. The study's data were based on EDHS (2016) obtained from CSA. To identify the risk factors associated with when to use modern contraceptive methods, a parametric survival model with frailty was used and compared using AIC and log-likelihood ratio test. Result: The study found that out of a total of 7,890 women, 6,829 (86.55 percent) of the 7,890 women did not use modern contraceptive methods or were censored, while 1,061 (13.45 percent) used modern contraceptive methods during the follow-up period. The minimum and maximum value of time for women to use modern contraceptive methods was one and thirty years, respectively. During the study period, the overall mean and median uptake time were 23.45 and 22 years, respectively. About 20.69 percent of women between the ages of 25 and 29 have used modern contraceptive methods. It is estimated that 20.52 percent of women aged 20 to 24 have used a modern contraceptive method. The modern contraceptive method was used by 19.83 percent of women aged 30-34 years. Modern contraceptive methods were used by 5.56 percent of women aged 45-49 years. In terms of location, 6.83 percent of women in urban areas used modern contraceptive methods, while 6.62 percent of women in rural areas did. there is a significant relationship between women's age group and the time they use modern contraceptive methods. When all other variables were held constant, the estimated time to modern contraceptive uptake for women aged 35-39 years, 40-44 years, and 45-49 years 1 increased by 21.6 percent, 28.5 percent, and 46.6 percent, respectively, when compared to women aged 15-19 years. In Ethiopia, the majority of women who do not receive medical care are those who live in rural areas, and the findings of this study revealed that residence has a significant impact, with women living in rural areas having a 15.5 percent shorter estimated time to use modern contraceptive methods than women living in urban areas. Abstract Table 4 further shows that the wealth index has a significant effect on the time to use modern contraceptive methods. When the effects of other variables were held constant, the relative risk of time to use modern contraceptive methods was reduced by 8.3 percent and 17.3 percent for women in the middle and rich wealth index categories, respectively, compared to women in the poor wealth index categories. The educational level of women had a significant impact on the time it took to use modern contraceptive methods. When other variables in the model were held constant, the estimated time to modern contraceptive uptake increased by 5.5 percent and 5.6 percent for primary, secondary, and higher education women, respectively, when compared to illiterate women. Conclusion: The Weibull regression survival model with Gamma frailty model was an appropriate fit of time to use modern contraceptive methods among the fitted frailty models. In comparison to the non-frailty survival models, the estimated frailty model, which accounts for unobservable regional heterogeneity, was found to be more appropriate for fitting time to use modern contraceptive methods across Ethiopian regions. Furthermore, using the selected frailty model results, the independent variables age, place of residence, wealth index, educational status, and television awareness information were identified as factors associated with the time to use modern contraceptive methods at a 5% level of significance. Key Word: Contraceptive, Survival Model, Frailty, Log-Normal. 1. Background Contraception is one of the most important health interventions of the twenty-first century, playing a critical role in slowing population growth and improving women's and family health. Family planning saves lives and improves the quality of life for women, children, and families as a whole. It also reduces pregnancy-related health risks in women, sexually transmitted infection (STI) and HIV/AIDS, and adolescent pregnancies, as well as slowing population growth [1]. By strengthening primary health care, the Ethiopian health sector aims to achieve the goal of the Health Sector Transformation Plan (HSTP), which is universal health coverage. Street women in informal settlements are likely to be denied access to formal public services, including contraception. Despite an increase in contraceptive prevalence from 29 percent in 2 2011 to 35 percent in 2016, Ethiopia's target of 65 percent by 2015 was not met [2]. The use of modern contraceptive methods results in the avoidance of unwanted pregnancy and subsequent abortions. If the population's use of contraception increases among sexually active Ethiopian men and women, there will be a significant reduction in unwanted pregnancies and abortions, resulting in lower maternal mortality. According to previous researches, more than 60% of women who have an unplanned pregnancy do not use any form of contraception [3]. A variety of factors contribute to unwanted pregnancy in Ethiopia, one of which is the country's low rate of contraceptive use. Furthermore, a desire to limit family size in order to provide a better education for children, increased female labor force participation, and urbanization all contribute to Ethiopian women's desire to have a set number of children. Contraceptive prevalence rates are related to maternal mortality rates, and countries with low contraceptive prevalence rates have very high maternal mortality ratios. Ethiopia has one of the highest maternal mortality rates in Sub-Saharan Africa, and it is one of six countries responsible for more than half of all maternal deaths globally [1 and 4]. A review of the literature on contraception use in Ethiopia identifies the reasons for low levels of contraceptive use, the factors responsible for this low utilization, and recommendations for interventions, programs, and policies to increase contraceptive utilization [5 and 6]. Unplanned pregnancies and abortions contribute to increased maternal mortality in Ethiopia, where more than 65 percent of women do not use any form of contraception. 1. Background It is critical for family planning workers to continue to meet the needs of existing contraceptive users while also addressing barriers for contraceptive users in society such as age, women's education level, wealth index, residence, religion, region, current marital status, family planning from radio, TV, newspaper, and so on. As a result, their efforts and approaches do not appear to be equally effective, equally served, or equally recognized in some areas Several studies have led many to the conclusion that it is preferable to begin contraception immediately after delivery to avoid the potential consequences [7]. The early start of family planning in the postpartum period has been linked to a variety of advantages. First and foremost, it is a reliable method of preventing unintended pregnancies. According to Singh et’al Unintended pregnancies are expected to fall by more than two-thirds, from 75 million in 2008 to 22 million per year, if both family planning and maternal and newborn services are available. As a result, women will no longer have to be concerned about becoming pregnant when they are not ready. Unsafe abortions will also be reduced because women will not become desperate and resort to all manner of risky methods 3 3 to end unwanted pregnancies. Maternal deaths are expected to fall dramatically, as are stillbirths, neonatal deaths, and infant deaths [8]. Women's contraceptive use was influenced by a variety of factors, including their age, place of residence, education of both the woman and her husband, occupation of both the woman and her husband, and the number of children. Women living in urban areas were more likely than women living in rural areas to make decisions on their own rather than jointly. Women whose husbands had no formal education or only a secondary education were more likely to make their own decisions than women whose husbands had a higher education. Women's education statuses or husbands' education statuses were the factors that influenced their decision-making. The findings also revealed that women with fewer than five children have more deciding power than those with more than five children [9]. Paul used a multiple logistic regression model in his study to assess the level of significance between the dependent and independent variables. 1. Background The findings show that a variety of factors, including partner approval, marital status, and employment status, are associated with the use of modern contraceptives, with religious beliefs and attitude being the only ones that are significantly associated with the use of modern contraceptives [10]. The findings revealed no statistically significant relationship between formal education level and current use of modern contraceptives. Tilahun (2013) [11] found that socio-demographic factors such as women's age, educational level, family size, economic status, and abortion history made a statistically significant difference in a study using logistic regression. Ethnicity, marital status, religion, and length of residence at the time of the survey had no statistically significant effect. Women aged 25-34 were 1.5 times more likely than women aged 15-24 to use modern contraception. Those over the age of 34 were also twice as likely to use modern contraception as those aged 15 to 24. Women with a high school diploma or higher had a significant effect, according to the findings. Those with a high school education or higher were twice as likely as illiterates to use modern contraception. Respondents with more than ten family members were nine times more likely to use modern contraception than those with one to five family members, based on the study population's family size. Women of medium to high socioeconomic status were less likely to use contraception than those of low socioeconomic status. Abortion survivors were less likely to use contraception. 4 4 According to a study using logistic regression conducted by Bogale et al. [12], the overall decision-making power in urban areas is joint 225 (67.06 percent), whereas in rural areas it is the husband without involving his wife 153. (45.83 percent). When their wishes did not coincide or their decisions were in conflict, the husband's decision was overruled by 331 (98.7 percent) of rural respondents and 305 (91.2 percent) of urban respondents. In both settings, 43 women (6.45%) said they can make independent decisions about their children's issues. Wives made a higher percentage of economic decisions in rural areas than in urban areas. Decisions about sociocultural and familial relationships, on the other hand, have the opposite effect. The majority of participants in focus groups agreed that mutual discussion is important, but in practice, men dominate domestic decisions, particularly in rural areas. 1. Background The current use of modern contraceptive methods was higher than in the national and regional data, and the urban-rural divide was smaller than in the regional and national data. In urban settings, gender equitable attitudes had a statistically significant association with decision making on modern contraceptive use, but not in rural settings. It was discovered that modern contraceptive decision-making power is greater in urban areas than in rural areas. Increased contraceptive knowledge empowers women in both urban and rural areas. In a multilevel logistic regression study, Tilahun discovered that place of residence, working status, exposure to mass media, educational status, and women's religion all had a significant association with the use of modern contraceptive methods. Women in rural areas used modern contraception at a lower rate (23.9 percent) than women in urban areas (20.7 percent) [11]. Tilahun's research also shows that a woman's religious beliefs have a significant impact on her use of modern contraception. Women of orthodox faith use modern contraception at a higher rate than women of other religions. Melash identified factors associated with modern contraceptive use using multivariable logistic regression. Modern contraceptive use was found to be associated with coming from a wealthy family, having a secondary or higher level of education, and wanting to space or have no more children [13]. Modern contraceptive use, on the other hand, was associated with a lower risk of breast cancer in women aged 35-49 years (AOR = 0.7; 95 percent CI: 0.5-0.9). In the Amhara region, the use of modern contraception was relatively high. Women with a high school diploma or higher were more likely to use modern contraception. Women from affluent families, as well as those who want to delay or avoid pregnancy, were more likely to use modern contraception. [14] investigated the relationship between husband approval for modern 5 family planning methods, couple discussion, male involvement in family planning decisions, desire for additional child, and prior use of contraception using logistic regression. Despite having a high level of knowledge about modern family planning methods, the study area used them infrequently. In a multilevel logistic regression study, Oluwafemi [15] discovered a significant relationship between educational level, marital status, parity, socioeconomic status, fertility intention, awareness of family planning methods, and use of modern contraceptives. Women who received family planning advice from a community member were more likely than those who did not to use modern contraception. 1. Background Kirui [16] studied the factors associated with the use of injectable contraceptives using multiple logistic regression. Education, marital status, wealth index, place of residence, and number of births were all significant predictors of contraceptive use among reproductive-age Kenyan women. Women with a postsecondary/vocational education were 54% less likely than those with no education to use contraception. George [17] examined associations between method- specific beliefs and choice of injectable, implants, or pills in women who did not use any method or were pregnant at baseline (round 1) but adopted these methods at the 12-month follow-up using the conditional logit model. Nonusers were generally skeptical of pills, injectables, and implants. Except for the pill, the majority of Nairobi residents believed that each method was likely to cause serious health problems, unpleasant side effects, menstrual disruption, and was unsafe for long-term use. As a result, the primary goal of this study was to identify the significant factors associated with the time for women in Ethiopia to use modern contraceptive methods. More specifically, to describe the socioeconomic and demographic factors that influence when Ethiopian women use modern contraception methods, as well as to fit the best-fitting survival model to the dataset. 2. Materials and Methods Source of data Secondary data for this study were obtained from the Central Statistics Agency (CSA) in the 2016 Ethiopia Demographic and Health Survey (2016EDHS). Individual interviews were available to all women aged 15 to 49 in the household. Statistical Methodology Statistical Methodology 6 Survival analysis is a set of statistical procedures for data analysis in which the outcome variable of interest is the amount of time until an event occurs. The survivor function and the hazard function are two commonly used Kaplan-Meier estimator used to display the graph of the survival time of the time to uptake of modern contraceptives among women in the reproductive age. The Kaplan-Meier estimate of the survival function at time t is given by:           k i j j j t R d t R t S 1 ) ( ) ( ) ( ) ( ) ( ) (ˆ , for t(j) < t < t(j+1) , k = 1, 2, … , r where, ) (ˆ t S is the survival function at time t where, ) (ˆ t S is the survival function at time t ) ( ) ( j t R is the risk set at time t ) ( ) ( j t R is the risk set at time t ) ( ) ( j t R is the risk set at time t ) ( j d is number death at time t. ) ( j d is number death at time t. ) ( j d is number death at time t. The log rank test, developed by Mantel and Haenszel, is a non-parametric test for comparing two or more independent survival functions. Since it is a non-parametric test, no assumption about the distributional form of the data is required. The log rank test statistic [18 and 20] for comparing two groups is given by: Q = [∑ wi(di −ê1i) m i=1 ]2 ∑ wi2v̂1i m i=1 where,  m is the number of rank ordered event times,  m is the number of rank ordered event times,  d1i is the observed number of events (use in group 1at event time,  d1i is the observed number of events (use in group 1at event time, e1i = n1i −d1i ni ⁄ is the expected no of events corresponding to d1i, number of individuals at risk in group1 just prior to event (use) time tj. v1i = n1i n2idi(ni−di) ni2 (ni − 1) is the variance of the number of events d1i at time tj. n2i is the number of individuals at risk in group1 just prior to event (use) time tj. ni and di are the number of individuals at risk and number of events in both groups (i.e., group 1 and group 2) just prior to event time ti respectively. Under the null hypothesis that two survival functions are equal, the log rank test statistic Q has an approximation of chi-square distribution with one degree of freedom for χ2(1) large samples. The null hypothesis of equality of survival functions will be rejected for large values of Q. The most frequently used 7 test is based on weights equal to one wi = 1 .Note that the log-rank test can be extended for comparing three or more groups of survival experience. Frailty Distributions There are various frailty models that have been developed and suggested in the literature and any distribution with a positive random variable can be used to model frailty [27]. Several authors have noted that unlike standard random effects models, inferential methods have been less developed in frailty models because of censoring and truncation. The frailty distributions most often applied are the gamma distribution [27-31], the inverse Gaussian distribution [32- 37], the compound Poisson distribution (Aalen, 1988) and the log-normal Distribution [38]. Gamma distribution Gamma frailty model belongs to the power variance function family [39] and can be expressed in terms of its Laplace transform [40] from which properties such as mean and variance are easily derived. Assuming a two-parameter gamma density with δ > 0 and γ > 0 as shape and scale parameters, respectively, the density function is given by: fu(u) = exp(−γu2)γδuδ−1 Γ(δ) ; with δ > 0 and γ > 0. fu(u) = exp(−γu2)γδuδ−1 Γ(δ) ; with δ > 0 and γ > 0. The distribution function of the frailty term u is therefore a one-parameter gamma distribution given by: The distribution function of the frailty term u is therefore a one-parameter gamma distribution given by: fu(u) = u1/θexp(−u θ ) θ1/θ Γθ fu(u) = u1/θexp(−u θ ) θ1/θ Γθ where θ > 0 and u > 1 indicates that individuals in group i are frail, whereas u < 1 indicates that individuals are strong and have lower risk. The corresponding Laplace transform is given by: L(s) = (1 + θs)𝟏/𝛉 where θ > 0 and u > 1 indicates that individuals in group i are frail, whereas u < 1 indicates that individuals are strong and have lower risk. The corresponding Laplace transform is given by: L(s) = (1 + θs)𝟏/𝛉 Once the frailty is integrated out, accounting for unobserved heterogeneity is reduced to estimating the variance of the frailty term. The variance θ of the frailty term represents the heterogeneity among clusters [40]. Weibull distribution The baseline hazard, h(t) can be chosen to follow a Weibull (λ; ρ) distribution which is more general and flexible than the exponential distribution. The Weibull baseline hazard allows for hazard rates that are non-constant but monotonic [42]. The probability density function is given by. by. f(t) = λρtρ−1exp(−λtρ) f(t) = λρtρ−1exp(−λtρ) f(t) = λρtρ−1exp(−λtρ) where, λ > 0 and ρ > 0 are shape and scale parameters, respectively. The corresponding survival function is given by; where, λ > 0 and ρ > 0 are shape and scale parameters, respectively. The corresponding survival function is given by; S(t) = exp(−λtρ) S(t) = exp(−λtρ) S(t) = exp(−λtρ) The corresponding hazard function is given by; h(t) = f(t)x2 s(t) = λρtρ−1exp(−λtρ) exp(−λtρ) = λρtρ−1 and the cumulative hazard is: H(t) = ∫ h(x)d ∞ 0 x =∫λρtρ−1d x 0 x = λtρ The hazard rises if ρ > 1, constant if ρ = 1 and decreases if ρ < 1. Exponential distribution is a special case of Weibull distribution when the shape parameter λ is 1. H(t) = ∫ h(x)d ∞ 0 x =∫λρtρ−1d x 0 x = λtρ The hazard rises if ρ > 1, constant if ρ = 1 and decreases if ρ < 1. Exponential distribution is a special case of Weibull distribution when the shape parameter λ is 1. The hazard rises if ρ > 1, constant if ρ = 1 and decreases if ρ < 1. Exponential distribution is a special case of Weibull distribution when the shape parameter λ is 1. The hazard rises if ρ > 1, constant if ρ = 1 and decreases if ρ < 1. Exponential distribution is a special case of Weibull distribution when the shape parameter λ is 1. Inverse Gaussian distribution The inverse Gaussian density function is given by: The inverse Gaussian density function is given by: fu(u) = ((a/2π )1/2u−3/2 exp ( −a 2uµ2) 2 with µ > 0 and α > 0. The corresponding Laplace transform is: The corresponding Laplace transform is: fu(u) = exp( a µ −(2 ( a2 µ2) + 2as) 1 2) 8 For µ = 1, θ = V ar(U) = 1/α (as α = ∞ corresponds with no heterogeneity (i. e. , θ = 0). 0). 0). Baseline hazard distributions for parametric frailty models Baseline hazard distributions for parametric frailty models Baseline hazard distributions for parametric frailty models The risk of an event occurring can be constant over time or with more complicated hazard rates that increase and decrease over time or that increase or decrease at faster or slower rates. Exactly how the hazard rate varies with time is generally referred to as time dependency. The logic of parametric duration models is that they assume a particular shape for the hazard rate. Below are some of the commonly used baseline hazards distributions. The risk of an event occurring can be constant over time or with more complicated hazard rates that increase and decrease over time or that increase or decrease at faster or slower rates. Exactly how the hazard rate varies with time is generally referred to as time dependency. The logic of parametric duration models is that they assume a particular shape for the hazard rate. Below are some of the commonly used baseline hazards distributions. Exponential distribution For the exponential model, the hazard rate is characterized by: For the exponential model, the hazard rate is characterized by: h(t) = λ This implies that the conditional probability of an event is constant over time. The corresponding cumulative hazard is given by; H(t) = λt. Weibull distribution Gompertz distribution 9 9 The Gompertz distribution is characterized by the fact that the log of the hazard function is linear in t(Jenkins, 2008)[40] : The Gompertz distribution is characterized by the fact that the log of the hazard function is linear in t(Jenkins, 2008)[40] : lnh(t) = ρt + c lnh(t) = ρt + c lnh(t) = ρt + c and after transformation, the hazard function for Gompertz distribution is given by: and after transformation, the hazard function for Gompertz distribution is given by: h(t) = λeρt h(t) = λeρt h(t) = λeρt where, λ = eXβ and ρ is the shape parameter. The corresponding survival function is where, λ = eXβ and ρ is the shape parameter. The corresponding survival function is where, λ = eXβ and ρ is the shape parameter. The corresponding survival function is S(t) = e−λρ−1(eρt−1) S(t) = e−λρ−1(eρt−1) S(t) = e−λρ−1(eρt−1) The corresponding cumulative hazard is given by; H(t) = ∫ h(x)dx ∞ 𝟎 = ꭍ λeρxdx = λρ (eρt − 1) Methods of Estimation in parametric frailty model Methods of Estimation in parametric frailty model Methods of Estimation in parametric frailty model When a parametric baseline hazard is assumed, maximum likelihood estimates can be obtained by maximizing the likelihood function. This not only makes estimation easier, but also describe explicitly the effect of the frailty on hazard ratios over time. Survival data consist of event times and censored observations and the likelihood function under right censoring is given by: L = ∏[(1 − Hj(t))fj(t)] δj [(1 − Fj(t)) hj(t)] 1−δj where, δj is the censoring indicator, h and H are the hazard function and the cumulative distribution function of the censoring time. f and F are the density function and the cumulative distribution function of the event time, respectively. The distribution of censoring times in the likelihood function can be ignored because it does not depend on the parameters of interest related to the survival function [43]. Therefore, the likelihood function for the jth subject assuming right censoring is of the form; L = ∏[(1 − Hj(t))fj(t)] δj [(1 − Fj(t)) hj(t)] 1−δj where, δj is the censoring indicator, h and H are the hazard function and the cumulative distribution function of the censoring time. f and F are the density function and the cumulative distribution function of the event time, respectively. Gompertz distribution The distribution of censoring times in the likelihood function can be ignored because it does not depend on the parameters of interest related to the survival function [43]. Therefore, the likelihood function for the jth subject assuming right censoring is of the form; L = ∏[(1 − Hj(t))fj(t)] δj [(1 − Fj(t)) hj(t)] 1−δ Comparison of Models Model comparison and selection are among the most common problems of statistical practice, with numerous procedures for choosing among a set of models [44 and 45]. There are several methods of model selection. The most commonly used methods include Akaike information and likelihood-based criteria. A data-driven model selection method such as an adapted version of Akaike's information criterion AIC [46] is used to find the truncation point of a series of models. In some circumstances, it might be useful to easily obtain AIC value for a series of candidate models [47]. In this study, we used the AIC criterion and log likelihood to compare three of parametric models. AIC is defined as AIC =−2log(L) + 2(K+C), Where log(L) is the 10 log-likelihood, K is the number of covariates in the model and C is the number of model- specific ancillary parameters. The addition of 2(K+C) can be thought of as a penalty if non predictive parameters are added to the model. Small values of AIC suggest a better model. Details of the survival and shared models is found on supplementary file. 3. Results and Discussion Table 1 depicts the descriptive results of the time to use modern contraceptive methods among all women in terms of socioeconomic and demographic factors. The study included 7,890 women in total. Approximately 6,829 (86.55 percent) of the 7,890 women did not use modern contraceptive methods or were censored, while 1,061 (13.45 percent) used modern contraceptive methods during the follow-up period. The minimum and maximum value of time for women to use modern contraceptive methods was one and thirty years, respectively. During the study period, the overall mean and median uptake time were 23.45006 and 22 years, respectively. Approximately 20.69 percent of women between the ages of 25 and 29 have used modern contraceptive methods. It is estimated that 20.52 percent of women aged 20 to 24 have used a modern contraceptive method. The modern contraceptive method was used by 19.83 percent of women aged 30-34 years. Modern contraceptive methods were used by 5.56 percent of women aged 45-49 years. In terms of location, 6.83 percent of women in urban areas used modern contraceptive methods, while 6.62 percent of women in rural areas did. The use of modern contraceptive methods by religious groups revealed that approximately 23.9 percent of orthodox religious followers’ women used modern contraceptive methods, 17.99 percent of protestant women used modern contraceptive methods, and 7.53 percent of Muslim women used modern contraceptive methods. Marital status of women who used modern contraceptive methods for married, never married, and others was 13.57 percent, 19.51 percent, and 11.07 percent, respectively. Modern contraceptive methods were used by 14.24 percent of illiterate women, 11.07 percent of women with primary education, and 10.97 percent of women with a secondary education or higher. In terms of region, approximately 25.44 percent of women in Addis Abeba have used modern contraceptive methods. In SNNPR, 29.20 percent of women have used a modern contraceptive method. In addition, 18% of women in Oromia have used a modern contraceptive method. In Somalia and Gambella, 0.67 percent and 6.20 percent of women, respectively, use modern contraception methods. In terms of wealth, approximately 23.10 percent, 14.07 percent, and 6.44 percent of women in the rich, middle, and poor classes, respectively, used modern contraceptive methods. Figure 2 here The log-rank test demonstrates that the two survival functions are equal, and the log-rank test of the result was use of modern contraceptive method in different categories of age, place of residence, wealth index, marital status, religion, region, and educational status, as well as heard family planning on radio, TV, and newspaper were significantly different at the 5% level. Furthermore, there were no statistically significant differences in survival function between work status categories (Table 2). Table 2 here Univariate Parametric, Multivariable Survival Analysis, and Model Comparisons Before proceeding to the multivariable analysis at a 25% level of significance, we used univariate analysis to see the effect of each covariate on the time-to-use modern contraceptive method. The results in Appendix 1 show that age, place of residence, wealth index, educational status, and heard family planning on TV were significant, whereas work status, religion, marital status, and heard family planning in the newspaper were not. The goal of model comparison is to find a model that accurately describes the dataset. In univariate analysis, all significant covariates were used to compare multivariable analysis of exponential, Weibull, and Gompertz parametric models. The Weibull distribution with the Gama shard frailty model fit the time to use modern contraceptive methods among Ethiopian women (Table 3). Table 3 here Test of heterogeneity Figure 1 here Women in urban areas, as shown in Figure 2, use modern contraceptive methods for a longer period of time after beginning sexual intercourse than women in rural areas. Figure 2 here 3. Results and Discussion 11 The Kaplan-Meier Estimator Figure 1 shows that the median time for the occurrence of the event is estimated from the Kaplan Meier curve as the X- axis (time) value at the point where a horizontal line at the 50% survival probability on the y -axis crosses the survival curve is approximately 22. Figure 1 here Test of heterogeneity The variance of the frailty term is significantly different from zero, indicating that there is subject heterogeneity. Using a Wald test, we can deduce this: Wm () = 0.426 /0.178= 2.40> 1.64, with 1.64 being the critical value for a standard one-sided test. And, with a p-value of 0.001, the likelihood ratio test for the hypothesis =0 was significant. This indicates that the frailty component contributed significantly to the model. Multivariable Analysis of Weibull Gamma Shared Frailty Model Multivariable Analysis of Weibull Gamma Shared Frailty Model 12 The p-value less than 5% significance level was used to determine whether or not a variable is significant. Time to use modern contraceptive method was significantly associated with age, type of residence, wealth index, education level, and hearing information on TV, as shown in Table 4. The estimated random effect (the frailty term), 0.426, was also significant, indicating differences in the use of modern contraceptive methods across regions and the presence of unobservable heterogeneity in the data. Time rationing can be used to interpret the estimated coefficients of the Weibull gamma shared frailty model. In the case of the Weibull regression survival model, the time ration represents the relative risk that was interpreted. According to Table 4, there is a significant relationship between women's age group and the time they use modern contraceptive methods. When all other variables were held constant, the estimated time to modern contraceptive uptake for women aged 35-39 years, 40-44 years, and 45-49 years increased by 21.6 percent, 28.5 percent, and 46.6 percent, respectively, when compared to women aged 15-19 years. In Ethiopia, the majority of women who do not receive medical care are those who live in rural areas, and the findings of this study revealed that residence has a significant impact, with women living in rural areas having a 15.5 percent shorter estimated time to use modern contraceptive methods than women living in urban areas. According to Table 4, there is a significant relationship between women's age group and the time they use modern contraceptive methods. When all other variables were held constant, the estimated time to modern contraceptive uptake for women aged 35-39 years, 40-44 years, and 45-49 years increased by 21.6 percent, 28.5 percent, and 46.6 percent, respectively, when compared to women aged 15-19 years. In Ethiopia, the majority of women who do not receive medical care are those who live in rural areas, and the findings of this study revealed that residence has a significant impact, with women living in rural areas having a 15.5 percent shorter estimated time to use modern contraceptive methods than women living in urban areas. Table 4 further shows that the wealth index has a significant effect on the time to use modern contraceptive methods. Multivariable Analysis of Weibull Gamma Shared Frailty Model When the effects of other variables were held constant, the relative risk of time to use modern contraceptive methods was reduced by 8.3 percent and 17.3 percent for women in the middle and rich wealth index categories, respectively, compared to women in the poor wealth index categories. The educational level of women had a significant impact on the time it took to use modern contraceptive methods. When other variables in the model were held constant, the estimated time to modern contraceptive uptake increased by 5.5 percent and 5.6 percent for primary, secondary, and higher education women, respectively, when compared to illiterate women. In terms of the significant effects of heard information on TV, the relative risk of time to use modern contraceptive method decreased by 5.2 percent when compared to no heard family planning by TV women when all other variables were held constant. Figure 3: here Discussion The primary goal of this study was to identify the main factors associated with the time for all Ethiopian women to use modern contraceptive methods. The study used an appropriate frailty survival model to account for the associated risk factors. In order to meet high levels of demand by addressing both women's and men's sexual and reproductive health needs, the provision of modern contraceptive methods that are safe, effective, and affordable for family planning is being emphasized. Women aged 35-39, 40-44, and 45-49 years were more likely to use modern contraception than women aged 15-19 years. [11,13,17, and 19] discovered that women aged 35 and under were less likely than those aged 35 and up to use modern contraceptive methods. Women in cities had used modern contraceptive methods for a longer period of time than women in rural areas, regardless of where they lived. This finding is consistent with the findings of [11], which show that women in cities use modern contraception more frequently than women in rural areas. Women with primary, secondary, and higher education are more likely than illiterate women to use modern contraception methods for a longer period of time. This finding is consistent with Susan's findings [7]. Women with a primary, secondary, or higher level of education were more likely to use modern contraception. Women who watched television used modern contraception methods. This finding was consistent with the findings of a study conducted by [7,8 and 9]. And variables such as marital status, partners' employment, and frequency of radio listening did not use modern contraceptive methods. [5,11 and 14] discovered that marital status, working partners, and frequency of radio listening all influenced the use of modern contraceptive methods. Model Diagnostics The Cox- Snell residuals had been obtained from fitting using the Weibull gamma shared frailty model. It can be seen that the plot of the cumulative hazard function against Cox-Snell residuals 13 is straight lines through the origin which indicates that the model best fits the data as shown in Figure 3. Availability of Data and Materials The datasets generated and analyzed during the current study are available upon reasonable request from the corresponding author. Competing interests The authors declare that they do not have any competing interests. Not applicable Not applicable Authors’ Contributions This project was collaboratively carried by all of the authors. SB and TA came up with the concept for the paper. SB analyzed the data and drafted the first draft of the paper. SB, TA, DT, AW, MA and MS revised the paper even more to improve it. The final manuscript was approved by all authors. 4. Conclusions The study found that the time it takes to use modern contraceptive methods based on age, place of residence, wealth index, marital status, education, and media sources such as newspapers, radio, and television. The Weibull gamma shared frailty model, among the fitted frailty models, provided an appropriate fit of time to use modern contraception methods. A multivariable analysis using the Weibull regression survival model and the gamma frailty model revealed that age group, place of residence, wealth index, educational level, and hearing family planning on TV were all significant factors in the time-to-use modern contraceptive method. 14 Acknowledgements The authors would like to express their gratitude to the Central Statistical Agency for allowing us to use the data analyzed in this study. We would like to express our gratitude to Addis Ababa University for their financial support of this research. Ethics Approval and Consent to Participate All survey participants, including minors' parents/legal guardians and illiterate respondents, provided written informed consent. The Central Statistical Agency's along with Addis Ababa University, Department of Statistics ethics committee and research advisory board have approved this consent. All procedures involving human participants in this study were carried out in accordance with the ethical standards of the institutional and national research committees. In its meeting on December 20, 2021, the institutional ethics committee of Addis Ababa University's department of statistics reviewed and discussed the research title "time to use modern contraceptives and associated factors among women of reproductive age in Ethiopia: survival analysis of the 2016 Ethiopian demographic and health survey data." The check list, executive summary, consent form, and record form were reviewed and approved: Authors' Information 15 Addis Ababa University, Department of statistics, Addis Ababa, Ethiopia Email: shibabawaddisu053@gmail.com Tilahun Asena, PhD Arba Minch University, Department of Statistics, 21, Arba Minch, Ethiopia Email: feredetilahun14@gmail.com Dejen Tesfaw, PhD Addis Ababa University, Department of statistics, Addis Ababa, Ethiopia Mr Asmare Wube, M.Sc Ethiopian Public Health Institute, Addis Ababa, Ethiopia Email: asmarewubee@gmail.com Meseret Alemayehu, MD Department of pediatrics, Pawi General Hospital, Pawi, Ethiopia Email: meseretmossisa@gmail.com Abbreviations AIC: Akaike information criterion AIDS: Acquired Immune Deficiency Syndrome BIC: Bayesian information criterion CI: Confidence Interval CSA: Central Statistical Agency DF: Degrees of Freedom EDHS: Ethiopia Demographic and Health Survey HIV: Human Immune Virus HR: Hazard Ratio HSTP: Health Sector Transformation Plan KM: Kaplan- Meier LE: Likelihood Estimation LL: Log Likelihood PH: Proportional hazard SE: Standard Error SNNPR: South Nation Nationality People Representative STATA: statistical software package STI: Sexually Transmitted Infection TV: Television WHO: World Health Organization? Addis Ababa University, Department of statistics, Addis Ababa, Ethiopia Email: shibabawaddisu053@gmail.com Tilahun Asena, PhD 16 References [1] World Health Organization (2018): Family Planning: A Global Handbook for Providers. Family Planning - A global handbook for providers 2018 edition. [2] Wondimnew et al. (2020): Modern Contraceptive Utilization and Determinant Factors among Street Reproductive-Aged Women in Amhara Regional State Zonal Towns, North West Ethiopia, 2019: Community-Based Study. [3] World Health Organization (2019): Contraception enables people to make informed choices about their sexual and reproductive health. Glob Health Sci Pract. 2016;4(2):191-210. [4] Berhan, Yifru, and Asres Berhan. “Causes of maternal mortality in Ethiopia: a significant decline in abortion related death.” Ethiopian journal of health sciences vol. 24 Suppl,0 Suppl (2014): 15-28. doi:10.4314/ejhs.v24i0.3s. [5] Birhanu, B.E., Kebede, D.L., Kahsay, A.B. et al. Predictors of teenage pregnancy in Ethiopia: a multilevel analysis. BMC Public Health 19, 601 (2019). https://doi.org/10.1186/s12889-019-6845-7. [6] Merhawi et al. (2018) Unsafe abortion and associated factors among reproductive aged women in Sub-Saharan Africa: a protocol for a systematic review and meta-analysis. [7] Ross et al (1989) International Journal of Reproductive Medicine. [8] Singh et al (2009). Journal article Unintended Pregnancy: Worldwide Levels, Trends, and Outcomes. [9] Edossa ZK, Debela TF, Mizana BA. (2020). Women’s Decision on Contraceptive Use in Ethiopia: Multinomial Analysis of Evidence from Ethiopian Demographic and Health Survey. Health Serv Res Manag Epidemiol. 2020 May 8; PMID: 32426421; PMCID: PMC7218467. [10] Paul (2018) using multiple logistic regression model Determinants and Predictors of SMEs' Financial Failure – MDPI. [11] Tilahun (2013) Married women's decision-making power on modern contraceptive use in urban and rural southern Ethiopia. (Journal of the American Statistical Association. [12] Bogale, B., Wondafrash, M., Tilahun, T. et al (2011). Married women's decision-making power on modern contraceptive use in urban and rural southern Ethiopia. BMC Public Health. [13] Melash B. (2020) using Bivariable and multivariable logistic regression Modern Contraceptive Use and Influencing Factors in Amhara Regional State: Further Analysis of Ethiopian Demographic Health Survey Data 2016 Advances in Public Health. 17 [14] Teshale (2020). Magnitude and associated factors of unintended pregnancy in Ethiopia: a multilevel analysis using 2016 EDHS data BMC Pregnancy and Childbirth. [15] Oluwafemi David (2020). using a multilevel logistic analysis Factors influencing the use of modern contraceptive in Nigeria: a multilevel logistic analysis using linked data from performance monitoring and accountability. [16] Kirui, Ernest Kiplangat (2021). Multiple Logistic Regression Model for Determinants of Injectable Contraceptive Uptake among women of reproductive age in Kenya. Journal of Family & Reproductive Health. References [17] George Odwe (2021) Method-specific beliefs and subsequent contraceptive method choice: Results from a longitudinal study in urban and rural Kenya. [18] Hosmer and Lemeshow, (1999) Applied Survival Analysis: Regression Modeling of Time to Event Data (Journal of the American Statistical Association) pp.132-168. [19] Kaplan and Meier (1958). Nonparametric Estimation from Incomplete Observations. Journal of the American Statistical Association [20] Kleinbaum and Klein, 2005 Applied Survival Analysis: Regression Modeling of Time to Event Data (Journal of the American Statistical Association) pp.132-168. [21] Li et al., (2007). Comparison between current and future environmental satellite imagers on cloud classification using MODIS. [22] Vaupel, J.W., Manton, K.G. and Stallard, E., 1979. The impact of heterogeneity in individual frailty on the dynamics of mortality. Demography, 16(3), pp.439-454. [23] Lancaster, T., 1979. Econometric methods for the duration of unemployment. Econometrica: Journal of the Econometric Society, pp.939-956. [24] Hougaard (1984), Survival Models for Heterogeneous Populations Derived from Stable Distributions. Journal article. [25]. Glidden and Vittingho (2004). Modeling Clustered Survival Data from Multicenter Clinical Trials taken under Statistics in Medicine. [26] Duchateau and Janssen (2008). Penalized partial likelihood for frailties and smoothing splines in time to first insemination models for dairy cows. Biometrics 60, 608-614. [27] Ulviya (2013). Frailty models for modeling heterogeneity. A Thesis Submitted to the School of Graduate Studies in Partial Fulfillment of the Requirement. McMaster University. [28] Clayton, 1978; Vaupel et al., 1979; Oakes,1982; Hougaard, 2000; Wienke et al., 2002; Parametric frailty and shared frailty survival models. The Stata Journal (2002). 18 [31] Vaupel, J.W., Manton, K.G. and Stallard, E., 1979. The impact of heterogeneity in individual frailty on the dynamics of mortality. Demography, 16(3), pp.439-454. [30] Susan Ontiri, (2020) Patterns and determinants of modern contraceptive discontinuation among women of reproductive age. PLoS One. 2020 [32] Hougaard, P. 1984. Life table methods for heterogeneous populations: distributions describing the heterogeneity. Biometrika, 71(1), pp.75-83 [33] Wienke et al., 2003a; Parametric frailty and shared frailty survival models. The Stata Journal (2002). [34] Hanagal and Sharma (2012), Modeling heterogeneity for bivariate survival data by shared gamma frailty regression model. Biometrics 58, 510-520. [35] Hougaard (1986b) positive stable family of distributions for the frailty, not available for estimation in current Stata. Biometrika 73, 387-396. [36] Hougaard (1986a), Frailty models for survival data using the inverse Gaussian distribution. Biometrika 73, 671-678. [37] Rao and Wu (2001). [47] Munda.et.al (2012). An Empirical Comparison between Partial Likelihood and Penalized Partial Likelihood Estimators for Semi-parametric Non-proportional Hazards Models with Frailty. University of Ilorin, Nigeria. References On model selection in Institute of Mathematical Statistics Lecture Notes - Monograph Series [38] McGilchrist and Aisbett (1991). Regression with Frailty in Survival Analysis. Journal article [39] Hougaard (1986b) positive stable family of distributions for the frailty, not available for estimation in current Stata. Biometrika 73, 387-396. [40] Aalen, (1988) Modeling Heterogeneity in Survival Analysis by the Compound Poisson Distribution Stat. Med. 7, 1121-1137. [41] Duchateau and Janssen (2008). The Frailty Model Probability Theory and Stochastic Processes. eBook Biometrics 60, 608-614. [42] Jenkins (2008). Weibull distribution-based model for prediction of wind potential in Enugu, Nigeria. Pelagia Research Library. [44] Kadane and Lazar (2001) Methods and Criteria for Model Selection. Journal of the American Statistical Association. [45] Rao and Wu (2001). On model selection in Institute of Mathematical Statistics Lecture Notes - Monograph Series. [45] Cox, D.R., 1972. Regression models and lifetables. Journal of the Royal Statistical Society: Series B (Methodological),34(2), pp.187-202 [46] Akaike H. (1974). A new look at the statistical model identification. IEEE Trans Automatic Control. Journal Article. 19 [47] Munda.et.al (2012). An Empirical Comparison between Partial Likelihood and Penalized Partial Likelihood Estimators for Semi-parametric Non-proportional Hazards Models with Frailty. University of Ilorin, Nigeria. [47] Munda.et.al (2012). An Empirical Comparison between Partial Likelihood and Penalized Partial Likelihood Estimators for Semi-parametric Non-proportional Hazards Models with Frailty. University of Ilorin, Nigeria. [47] Munda.et.al (2012). An Empirical Comparison between Partial Likelihood and Penalized Partial Likelihood Estimators for Semi-parametric Non-proportional Hazards Models with Frailty. University of Ilorin, Nigeria. [47] Munda.et.al (2012). An Empirical Comparison between Partial Likelihood and Penalized Partial Likelihood Estimators for Semi-parametric Non-proportional Hazards Models with Frailty. University of Ilorin, Nigeria. 20 Figures Figure 1 Overall estimate of Kaplan-Meier survivor function of the use of modern contraceptive method among women in Ethiopia Figure 3 Cox Snell residual for Weibull gamma shared frailty model Figure 1 Figure 1 Overall estimate of Kaplan-Meier survivor function of the use of modern contraceptive method among women in Ethiopia Figure 2 Figure 2 Figure 2 The plot of Kaplan-Meier survivor function by place of residence Figure 3 Figure 3 Figure 3 Supplementary Files This is a list of supplementary ¦les associated with this preprint. Click to download. supplementary¦le.docx supplementary¦le.docx
https://openalex.org/W2884756670	https://europepmc.org/articles/pmc6066243?pdf=render	English	null	Selenium maintains Ca2+ homeostasis in sheep lymphocytes challenged by oxidative stress	PloS one	2,018	cc-by	6,687	OPEN ACCESS Citation: Proietti P, Trabalza Marinucci M, Del Pino AM, D’Amato R, Regni L, Acuti G, et al. (2018) Selenium maintains Ca2+ homeostasis in sheep lymphocytes challenged by oxidative stress. PLoS ONE 13(7): e0201523. https://doi.org/10.1371/ journal.pone.0201523 Editor: Guo-Chang Fan, University of Cincinnati College of Medicine, UNITED STATES Editor: Guo-Chang Fan, University of Cincinnati College of Medicine, UNITED STATES Received: June 1, 2018 Accepted: July 17, 2018 Published: July 30, 2018 Received: June 1, 2018 Accepted: July 17, 2018 Published: July 30, 2018 Copyright: © 2018 Proietti et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the paper. Selenium maintains Ca2+ homeostasis in sheep lymphocytes challenged by oxidative stress Primo Proietti1☯, Massimo Trabalza Marinucci2☯, Alberto Marco Del Pino1☯, Roberto D’Amato1, Luca Regni1, Gabriele Acuti2, Elisabetta Chiaradia2, Carlo Alberto Palmerini1☯ Primo Proietti1☯, Massimo Trabalza Marinucci2☯, Alberto Marco Del Pino1☯, Roberto D’Amato1, Luca Regni1, Gabriele Acuti2, Elisabetta Chiaradia2, Carlo Alberto Palmerini1☯ 1 University of Study of Perugia, Department of Agricultural, Food and Environmental Sciences, Perugia, Italy, 2 University of Study of Perugia, Department of Veterinary Medicine, Perugia, Italy a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 ☯These authors contributed equally to this work. * carlo.palmerini@unipg.it (CP); massimo.trabalzamarinucci@unipg.it (MT) Data Availability Statement: All relevant data are within the paper. Funding: This work was supported by Fondazione Cassa di Risparmio di Perugia (www. fondazionecrpg.com): Project “SELENOLIVO” (2015.0347.021). The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. fondazionecrpg.com): Project “SELENOLIVO” (2015.0347.021). The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. RESEARCH ARTICLE Abstract Selenium (Se) is an essential element in human and animal diets, based upon a widespread range of beneficial effects that are primarily due to its antioxidant properties. While Se can be associated to anti-cancer and anti-diabetic activities, reproductive efficiency, and enhancement of the immune system, the mechanistic details of the corresponding biological processes are still largely elusive. To avoid deficiencies and increase bioavailability, Se it is generally supplied to livestock through Se-supplemented feeds or forage plants fertilized with inorganic Se. While the relationship between Ca2+ and ROS (reactive oxygen species) is well known, only a few studies have addressed the possible involvement of Se in the con- trol of cytosolic Ca2+ in oxidative stress. The results on Ca2+ homeostasis were obtained adding exogenous Se in the form of SeO4 2- to sheep lymphomonocytes cultured in vitro. In particular, Se strongly attenuated 1mM H2O2-induced alteration of intracellular [Ca2+]C as well as the entry of extracellular Ca2+ into the cells with comparable EC50 values for sodium selenate accounting to 1.72 and 2.28 mM, respectively. In an ex vivo trial, it was observed that Ca2+ homeostasis can effectively be rescued in sheep lymphomonocytes exposed in vivo to a Se concentration of approximately 1.9 mM, that was achieved by feeding sheep with olive leaves previously sprayed with 500 mg/plant Na-selenate. Thus the results obtained suggest that the mode of action of selenium markedly influenced Ca2+-related sig- naling events. Furthermore, results clearly reveal that the protective effect of Se on Ca2+ homeostasis under oxidative challenge can be clearly and effectively achieved through an appropriate dietary regimen obtained also in a circular economy logic using pruning of olive trees treated to reduce tree drought stress. Introduction Selenium (Se) is a trace chemical used as a dietary supplement in humans and animals. Initially regarded as a toxic molecule, it was subsequently shown to play a role in a number of biological Competing interests: The authors have declared that no competing interests exist. PLOS ONE \| https://doi.org/10.1371/journal.pone.0201523 July 30, 2018 1 / 12 Selenium maintains Ca2+ homeostasis in sheep lymphocytes challenged by oxidative stress processes across a variety of different experimental models [1, 2]. In this regard, Se has been described as an anticancer and cardio-protective element [3, 4], as well as an anti-diabetic agent based on regulatory implications within crucial events of the glucose metabolism, namely the glycolysis and gluconeogenesis pathways and the insulin response [5]. In ruminants Se deficiency has been mainly associated with nutritional muscular dystrophy (“white muscle disease”) (BIB) [6]. It was also shown to have beneficial effects towards repro- ductive efficiency in terms of placental size and time of delivery, incidence of metritis and ovarian cysts [7, 8], in vitro Leydig cell functions [9], prevalence of retained fetal membranes [10, 11] and calf mortality [12]. A number of studies have demonstrated that Se is involved in the physiological response to heat stress [13], immune system function [14, 15, 16] and antioxidant defence system [17, 18, 19]. Indeed, in the form of selenoproteins, Se can also be considered a powerful antioxidant, for protection against oxidative stress due to an excess of reactive oxygen species (ROS) and reactive nitrogen species (RNS) [20]. However, the functional implication of most selenopro- teins is still unknown. It must be noticed that, to date, details of Se-related benefits remain largely elusive, suggest- ing that an arbitrary and poorly controlled intake from the diet or alternative sources could compromise an expected beneficial effect and increase toxicity risks [21]. Reports have shown that biological implications of Se can be concentration-dependent. Excessive exposure to Se may cause anemia [22] due to Se binding to the β-subunit of haemoglobin and, consequently, alteration of carbamate formation resulting from the reaction between CO2 and primary amino groups [22]. Se toxicity has been also reported in animal species dependent on its chem- ical form and concentration [23, 24]. PLOS ONE \| https://doi.org/10.1371/journal.pone.0201523 July 30, 2018 Experimental study design Plant material. Olive leaves used in the sheep diet for experimental purposes were from 20 year old trees (cv Leccino), grown in central Italy and annually fertilized with 16.5 t ha-1 cow manure containing 0.018 mg kg-1 of total Se. Total Se concentration in the soil was 0.010 mg kg-1. Trees were sprayed in summer with 5 L/plant solution containing 100 mg L−1 Se obtained by dissolving Na-selenate (Sigma-Aldrich Corporation, St. Louis, Missouri, USA, cat. n. S0882-25g) in deionized water [43]. This treatment was enhanced by a wetting agent (Wet- ting PLUS 0.05%, Dow AgroSciences, Italy). For control purposes, a number of olive trees, grown in the same field and subjected to the same cultivation practices, were sprayed under the same conditions but in the absence of Na-selenate. Determination of total Se in olive leaves, hay and blood. Measurements of total Se content in sheep blood and olive leaves were performed using defrozen and dry samples, respectively. Samples of blood (0.5 g/sample), leaves (0.25 g/sample) and hay (0.25 g/sample) were microwave digested (ETHOS One high-performance microwave digestion system; Milestone Inc., Sorisole, Bergamo, Italy) with 8 mL of ultrapure concentrated nitric acid (65% w/w) and 2 mL of hydrogen peroxide (30% w/w). The heating program for the digestion procedure was 30 min with power of 1000 W and 200˚C. After cooling down, the digests were diluted with water up to 20 mL, then passed through 0.45 μm filters. The analysis were conduct using a graphite furnace atomic absorp- tion spectrophotometry, Shimadzu AA-6800 apparatus (GF-AAS; GFA-EX7, Shimadzu Corp., Tokyo, Japan) with deuterium lamp background correction and a matrix modifier (Pd(NO3)2, 0.5 mol L−1 in HNO3). All analyses were carried out in triplicate [44]. Animals and dietary treatment. The experiment was carried out at the “Azienda Zootec- nica Didattica”of the Department of Veterinary Medicine, University of Perugia. Twenty Sarda ewes in mid lactation (3rd-4th month after parturition) were randomly divided into two groups of equal size, balanced for body weight (41.3±0.9 kg) and body condition score (2.3 ±0.1), and fed one of the following isoenergetic and isonitrogenous concentrates: 1) a standard pelleted feed (OLIVE), which contained ground dehydrated untreated olive leaves (204.0 g kg- 1); 2) as in OLIVE, except that the olive leaves were obtained from sodium selenate-treated (OLIVE-Se) trees (Se content in leaves: 7.83±0.13 mg kg-1). All diets were administered for 60 days. Reagents FURA 2-AM (FURA-2-pentakis (acetoxymethyl) ester, Triton X-100 (t-octylphenoxypo- lyethoxyethanol), EGTA (ethylene glycol-bis (β-aminoethyl ether)-N,N,N’,N’-tetracetic acid), Trypan blue, and Iscove’s Modified Dulbecco Medium were purchased from Sigma-Aldrich Corporation (St. Louis, Missouri, USA). DMSO (dimethyl sulfoxide) All other reagents were of the highest available grade commercially available. Introduction Indeed, the amount of metabolically active Se may depend on the nature of the chemical entities ingested (organic compounds, such as selenomethionine and dimethylselenide, and inorganic selenites and selenates), which can be subject to differential bioavailability and tissue distribution [25]. In many areas of Europe, and particularly in Germany, Denmark, Scotland and Italy, Se concentration in forages is low due to its scarce presence in the soil [26, 27] which, in turn, is related to a number of factors such as high pH, large amount of oxygen, low organic matter and clay content [26]. To avoid deficiencies, Se is usually provided to livestock through mineral mixes, Se- enriched concentrate feeds, or forage plants fertilized with inorganic Se [27, 28, 29]. The bio- availability of the organic Se (mostly Se-methionine) incorporated into plants is known to be greater than the inorganic Na-selenite and Na-selenate usually included in livestock feeds [30, 31]. In addition, when the diet of dairy ruminants is supplemented with Se, milk Se concentra- tion is improved with potential benefits for the consumers [32]. While Se functionality appears to be most likely related to an antioxidant potential, and despite the well-established interplay between Ca2+ and ROS [33], only a few studies have addressed the possible involvement of Se in the control of cytosolic Ca2+ [34, 35]. Indeed, ROS were shown to cause a dynamic change of cytosolic Ca2+ in various types of cells [36], as a result of ion mobilization from Ca2+ stores and the subsequent entry of extracellular Ca2+, with consequences on the cellular processes in which the Ca2+ is involved (cell proliferation, differentiation, apoptosis, and regulation of signaling pathways) [37, 38, 39]. In this work, to clarify the relationship between Se, oxidative stress and Ca2+, the effect of Se on Ca2+ homeostasis in sheep lymphomonocytes exposed to hydrogen peroxide was investi- gated. The study was performed, in vitro, upon treatment of cells with exogenous Se, as well as in fresh lymphomonocytes isolated from sheep previously exposed to a diet that included PLOS ONE \| https://doi.org/10.1371/journal.pone.0201523 July 30, 2018 2 / 12 Selenium maintains Ca2+ homeostasis in sheep lymphocytes challenged by oxidative stress selenate-treated leaves resulting from pruning of olive trees treated with Se in order to reduce tree drought stress [40, 41, 42]. PLOS ONE \| https://doi.org/10.1371/journal.pone.0201523 July 30, 2018 Experimental study design In both groups, concentrate was fed at a rate of 350 g per head per day and was administered in two equal parts during the day. The ewes were adapted to the new feed progressively: during the first 15 days of the trial, the content of olive leaves in the concentrate was limited to 85.0 g kg-1. A third group of 10 ewes were subjected to the same dietary treatment, except that the concentrate did not contain olive leaves (CTR). The concentrate contained 167.8 g kg-1 crude protein and 290.9 g kg-1 neutral detergent fiber and was entirely consumed by the ewes during the whole length of the experiment. Lucerne hay (crude protein: 143.5 g kg-1; neutral detergent fiber: 461.3 g kg-1) was provided ad fiber and was entirely consumed by the ewes during the whole length of the experiment. Lucerne hay (crude protein: 143.5 g kg-1; neutral detergent fiber: 461.3 g kg-1) was provided ad PLOS ONE \| https://doi.org/10.1371/journal.pone.0201523 July 30, 2018 3 / 12 Selenium maintains Ca2+ homeostasis in sheep lymphocytes challenged by oxidative stress libitum. Estimated hay dry matter intake was 1.51, 1.45, and 1.48 kg/ewe/day for the CTR, OLIVE and OLIVE-Se group, respectively, and was not affected by the dietary treatment. libitum. Estimated hay dry matter intake was 1.51, 1.45, and 1.48 kg/ewe/day for the CTR, LIVE and OLIVE-Se group, respectively, and was not affected by the dietary treatment. Samples of the feeds were collected at 0, 15, 30, 45 and 60 days of the trial and analysed for chemical composition following AOAC methods [45, 46, 47]. The calcium content was deter- mined according to Julshamn et al. [48]. All ewes from the three dietary groups were blood sampled from the jugular vein before the morning meal at the end of the experiment. Blood samples were immediately transported to the laboratory on ice and processed upon arrival. Ethic rules. The study was conducted in accordance with Legislative Decree No. 146, implementing Directive 98/58/EC of 20 July 1998 concerning the protection of animals kept for farming purposes [49]. The Bioethics Committee of the University of Perugia approved the study protocol. Ethic rules. The study was conducted in accordance with Legislative Decree No. 146, implementing Directive 98/58/EC of 20 July 1998 concerning the protection of animals kept for farming purposes [49]. The Bioethics Committee of the University of Perugia approved the study protocol. Preparation of sheep lymphomonocytes. Experimental study design Whole blood samples were collected from sheep in heparinized Vacutainers tubes (BD). Peripheral blood lymphomonocytes were iso- lated by Ficoll1 Paque Plus (GE Healthcare Life Sciences) density gradient centrifugation. After centrifugation, the PBMC layer was collected and washed in Dulbecco’s phosphate-buff- ered saline (DPBS) (Sigma-Aldrich) and than re-suspended (106 cells/mL) in Hank’s Balanced Salt Sodium (HBSS) Ca2+ free (Sigma-Aldrich). Sheep blood and lymphomonocytes isolated from animals that were exposed to diets featur- ing different were so named. C1: a standard pelleted feed without olive leaves (CTR); C2: as in C1, except that with added olive leaves (OLIVE); C3: as in C2, except that with added olive leaves plus Se (OLIVE-Se). Measurement of cytosolic Ca2+. Intracellular calcium levels were determined spectro- fluorometrically using FURA-2AM the probe [50]. Sheep lymphomonocytes were harvested by centrifugation at 800 g for 10 min and then resuspended in 1 mL Ca2+-free HBSS buffer (120 mM NaCl, 5.0 mM KCl, MgCl2 1mM, 5 mM glucose, 25 mM Hepes, pH 7.4). Cell suspensions were incubated in the dark with FURA-2AM (2 μL of a 2 mM solution in DMSO) for 60 min, after which samples were centrifuged at 800 g for 5 min. Cells were then harvested and resuspended in 3 mL of Ca2+-free HBSS containing 0.1 mM EGTA, which was included to rule out or, at least, minimize a potential background due to contaminating ions. The Ca2+-FURA complex was excited by a dual-view wavelength splitter at 340nm and 380nm, and fluorescence subsequently detected at 510 nm. The ratio between 340-380nm to 510 nm emission was used to determine cytosolic calcium [Ca2+]c concentrations [50]. Based upon a 200s exposure to the detector, changes in cytosolic calcium were expressed as Δ[Ca2+]c (nM). In this regard, we employed a “Ca2+ add-back” approach to study the effects of H2O2 on the depletion of intracellular Ca2+ stores as well as Ca2+ entry into the cells. “Ca2+ add-back" implies an experimental procedure conducted in two steps: initially, Ca2+ is not included in the incubation medium; subsequently, CaCl2 is added to challenge the cellular system in order to obtain information about Ca2+ entry in cells [51]. Fluorescence was measured in a Perkin-Elmer LS 50 B spectrofluorometer (ex. 340 and 380 nm, em. 510 nm), set with a 10 nm and a 7.5 nm slit width in the excitation and emission win- dows, respectively. Fluorometric readings were normally taken after 300–350 s. PLOS ONE \| https://doi.org/10.1371/journal.pone.0201523 July 30, 2018 Experimental study design In all instances, cell-based assays implied cell viability rates that consistently exceeded 95%, as displayed by the Trypan Blue Exclusion method. Statistics. Results obtained from biological replicates are expressed as mean values ± stan- dard error of the mean (SEM). Statistical significance was determined by two-way ANOVA with Tukey’s post-hoc test for multiple comparisons. Statistical tests were performed using Graph Pad Prism 6.03 software for Windows (La Jolla, CA). 4 / 12 PLOS ONE \| https://doi.org/10.1371/journal.pone.0201523 July 30, 2018 Selenium maintains Ca2+ homeostasis in sheep lymphocytes challenged by oxidative stress PLOS ONE \| https://doi.org/10.1371/journal.pone.0201523 July 30, 2018 Results Levels of cytosolic calcium [Ca2+]c in sheep lymphomonocytes subject to oxidative stress by hydrogen peroxide. Basal [Ca2+]c levels in sheep lymphomonocytes were in the 4–5 nM range however, in the presence of hydrogen peroxide (H2O2), we observed a rapid(100s) spike in [Ca2+]c followed by a quick return to baseline values. Sharp rises in calcium levels were also found when the tissue culture medium was supplemented with increasing doses of H2O2, as shown in Fig 1. Both in the absence and presence of added Ca2+, the increase in [Ca2+]c was found to corre- late with the amount of H2O2 used to challenge the cells (Fig 1), thereby suggesting that, in sheep lymphomonocytes, the oxidative stress caused by H2O2 had an impact on Ca2+- stores depletion and deregulation of Ca2+-entry. Assessment of [Ca2+]c in sheep lymphomonocytes exposed to with selenium during H2O2- mediated oxidative stress. Above, we showed that H2O2-induced oxidative stress affects the levels of cytosolic Ca2+ and Ca2+- entry in sheep lymphomonocytes. Since Se must be considered an essential constitu- ent of the sheep diet [52], we decided to test whether this trace element would specifically be involved in the alteration of Ca2+ homeostasis. Fig 1. Oxidative stress by H2O2 in sheep lymphomonocytes. Time-course of sheep lymphomonocytes (1x106) after treatment with H2O2 (0.2, 0.5,1.0 mM) and, 200s later, 1mM CaCl2. Shown are levels (nM) of cytosolic calcium, measured at different time points. Each value calculated and expressed as the means from 5 experiments (biological replicates) ± SEM. https://doi.org/10.1371/journal.pone.0201523.g001 Fig 1. Oxidative stress by H2O2 in sheep lymphomonocytes. Time-course of sheep lymphomonocytes (1x106) after treatment with H2O2 (0.2, 0.5,1.0 mM) and, 200s later, 1mM CaCl2. Shown are levels (nM) of cytosolic calcium, measured at different time points. Each value calculated and expressed as the means from 5 experiments (biological replicates) ± SEM. https://doi.org/10.1371/journal.pone.0201523.g001 5 / 12 PLOS ONE \| https://doi.org/10.1371/journal.pone.0201523 July 30, 2018 Selenium maintains Ca2+ homeostasis in sheep lymphocytes challenged by oxidative stress Fig 2. SeO4 2- effect on calcium homeostasis during H2O2-induced oxidative stress. Increasing concentrations (0.01–10μM) of SeO4 2- were employed to pre-treat sheep lymphomonocytes prior to inducing oxidative stress with 1mM H2O2. Panel A shows the impact of selenium on [Ca2+]c, and Panel B on Ca2+-entry. In both instances, results are expressed in relation to samples that were not exposed to SeO4 2-. Each value corresponds to the mean calculated from 5 biological replicates ± SEM. PLOS ONE \| https://doi.org/10.1371/journal.pone.0201523 July 30, 2018 Results https://doi.org/10.1371/journal.pone.0201523.g002 Fig 2. SeO4 2- effect on calcium homeostasis during H2O2-induced oxidative stress. Increasing concentrations (0.01–10μM) of SeO4 2- were employed to pre-treat sheep lymphomonocytes prior to inducing oxidative stress with 1mM H2O2. Panel A shows the impact of selenium on [Ca2+]c, and Panel B on Ca2+-entry. In both instances, results are expressed in relation to samples that were not exposed to SeO4 2-. Each value corresponds to the mean calculated from 5 biological replicates ± SEM. Fig 2. SeO4 2- effect on calcium homeostasis during H2O2-induced oxidative stress. Increasing concentrations (0.01–10μM) of SeO4 2- were employed to pre-treat sheep lymphomonocytes prior to inducing oxidative stress with 1mM H2O2. Panel A shows the impact of selenium on [Ca2+]c, and Panel B on Ca2+-entry. In both instances, results are expressed in relation to samples that were not exposed to SeO4 2-. Each value corresponds to the mean calculated from 5 biological replicates ± SEM. Sheep lymphomonocytes labelled with FURA-2 were washed, transferred to Ca2+-free medium, and then treated with variable doses (0.01–10 μM) of SeO4 2-. Under these conditions, [Ca2+]c was not subject to changes. However, when cells, pre-exposed to 0.01–10 μM SeO4 2-, were subsequently challenged with 1.0 mM H2O2, the impact of oxidative stress appeared markedly reduced in relation to [Ca2+]c and, more emphatically, Ca2+ entry (Fig 2A and 2B). The effect of Se was dose-depen- dent, with EC50 of 1.72 and 2.28 μM, respectively. These results strongly suggest that selenium counteracts the effect of 1 mM H2O2 upon the depletion of Ca2+ stores and the resulting release of the ion into the cytosol. Similarly, it is shown that selenium can significantly offset H2O2-mediated Ca2+-entry into the cell. p g y y shown that selenium can significantly offset H2O2-mediated Ca2+-entry into the cell. Cytosolic calcium during H2O2-mediated oxidative stress in sheep lymphomonocytes exposed to Se in vivo. Cytosolic calcium during H2O2-mediated oxidative stress in sheep lymphomonocytes exposed to Se in vivo. Next, to better assess the nutritional role of Se in calcium homeostasis, we investigated sheep lymphomonocytes isolated from animals that were exposed to diets featuring different levels of the trace mineral. 6 / 12 PLOS ONE \| https://doi.org/10.1371/journal.pone.0201523 July 30, 2018 Selenium maintains Ca2+ homeostasis in sheep lymphocytes challenged by oxidative stress Table 1. Total Se contained in feeds. Total Se (mg kg -1) Olive leaves Concentrate Hay Control Se-enriched OLIVE OLIVE-Se 0.46±0.098 7.83±0.128 0.12±0.018 0.69±0.012 0.24 ±0.050 OLIVE: pelleted feed containing ground dehydrated olive leaves OLIVE-Se: pelleted feed containing ground dehydrated olive leaves obtained from sodium selenate-fertilized trees Table 1. Total Se contained in feeds. Table 1. Total Se contained in feeds. OLIVE-Se: pelleted feed containing ground dehydrated olive leaves obtained from sodium selenate-fertilized trees Table 1 shows the levels of total selenium in olive leaves and concentrate. Clearly, the con- tent of selenium in the leaves(Se-enriched) and pelleted food obtained from sodium selenate- treated trees (OLIVE-SE) was significantly higher than controls. As shown in Fig 3, higher levels of Se occurred in the C3, as compared to the C1 and C2 controls. Based on Se molecular mass of 78.96 Da, Se concentration in C3 blood accounted to approximately 1.9 μM. Within this context, C1 and C2 lymphomonocytes treated with different amounts of H2O2 (0.2, 0.5, 1mM), clearly displayed levels of cytosolic calcium that increased with increasing doses of H2O2. However, this trend was undoubtedly counteracted by 1.6 or 3.3 μM SeO4 2- dis- pensed in the tissue culture medium. On the other hand, under the same experimental condi- tions, C3 lymphomonocytes previously exposed in vivo to approximately 1.9 μM Se (Fig 3) showed markedly reduced [Ca2+]c that compared well with the results obtained in the presence of either 1.6 or 3.3 μM SeO4 2- (Fig 4A). These observations paralleled those obtained by mea- suring Ca2+ entry into the cells (Fig 4B). Taken together, the results unequivocally suggest that Se introduced in sheep diets can largely counteract the effects of oxidative stress on Ca2+ homeostasis. Discussion The reactive nature of reactive oxygen species (ROS) can cause irreparable damage to cellular components and eventually disrupt the integrity of the cell membrane. As an oxidant, Fig 3. Total Se content in sheep blood. Tests for total Se content were performed in the blood of sheep that were exposed to diets featuring different: C1, C2, C3. Each value corresponds to the mean calculated from 5 biological replicates ± SEM. https://doi.org/10.1371/journal.pone.0201523.g003 Fig 3. Total Se content in sheep blood. Tests for total Se content were performed in the blood of sheep that were exposed to diets featuring different: C1, C2, C3. Each value corresponds to the mean calculated from 5 biological replicates ± SEM. https://doi.org/10.1371/journal.pone.0201523.g003 Fig 3. Total Se content in sheep blood. Tests for total Se content were performed in the blood of sheep that were exposed to diets featuring different: C1, C2, C3. Each value corresponds to the mean calculated from 5 biological replicates ± SEM. https://doi.org/10.1371/journal.pone.0201523.g003 PLOS ONE \| https://doi.org/10.1371/journal.pone.0201523 July 30, 2018 7 / 12 Selenium maintains Ca2+ homeostasis in sheep lymphocytes challenged by oxidative stress Fig 4. Relevance of dietary Se intake to calcium homeostasis in sheep lymphomonocytes subject to H2O2-induced oxidative stress. Panel A (Ca2+ free): variations of Ca2+ cytosolic, and Panel B on Ca2+-entry.Each point corresponds to the average of five biological replicates (± SEM) measured on five preparations of lymphocytes obtained from the blood of five different animals included in the project. https://doi.org/10.1371/journal.pone.0201523.g004 Fig 4. Relevance of dietary Se intake to calcium homeostasis in sheep lymphomonocytes subject to H2O2-induced oxidative stress. Panel A (Ca2+ free): variations of Ca2+ cytosolic, and Panel B on Ca2+-entry.Each point corresponds to the average of five biological replicates (± SEM) measured on five preparations of lymphocytes obtained from the blood of five different animals included in the project. Fig 4. Relevance of dietary Se intake to calcium homeostasis in sheep lymphomonocytes subject to H2O2-induced oxidative stress. Panel A (Ca2+ free): variations of Ca2+ cytosolic, and Panel B on Ca2+-entry.Each point corresponds to the average of five biological replicates (± SEM) measured on five preparations of lymphocytes obtained from the blood of five different animals included in the project. PLOS ONE \| https://doi.org/10.1371/journal.pone.0201523 July 30, 2018 Author Contributions Conceptualization: Carlo Alberto Palmerini. Data curation: Alberto Marco Del Pino, Carlo Alberto Palmerini. Data curation: Alberto Marco Del Pino, Carlo Alberto Palmerini. Data curation: Alberto Marco Del Pino, Carlo Alberto Palmerini. Formal analysis: Alberto Marco Del Pino, Roberto D’Amato. Formal analysis: Alberto Marco Del Pino, Roberto D’Amato. Formal analysis: Alberto Marco Del Pino, Roberto D’Amato. Investigation: Carlo Alberto Palmerini. Methodology: Alberto Marco Del Pino, Gabriele Acuti, Carlo Alberto Palmerini. Resources: Primo Proietti, Carlo Alberto Palmerini. Supervision: Elisabetta Chiaradia, Carlo Alberto Palmerini. Validation: Carlo Alberto Palmerini. Visualization: Carlo Alberto Palmerini. Visualization: Carlo Alberto Palmerini. Writing – original draft: Carlo Alberto Palmerini. Writing – review & editing: Primo Proietti, Massimo Trabalza Marinucci, Alberto Marco Del Pino, Luca Regni, Carlo Alberto Palmerini. Discussion Thus, these observations point to a protecting impact of Se against deregulation of Ca2+ homeostasis prompted oxidative stress. Considering that treatment with Na-selenate is used to reduce the stress caused by drought in the olive tree [40, 41], in a circular economy logic, the leaves could been used as a Se vector in sheep feeding with the aim of providing an organic source of Se with greater bioavailability [30, 31] and at the same time contributing to the disposal of pruning by-products [54, 55]. In conclusion, we show that the dietary intake of Se can largely offset the effects of oxidative stress on Ca2+ homeostasis, thus elucidating, mechanistically, its beneficial role as an anti-oxi- dant. Thus, given the vital role of Ca2+ homeostasis in a plethora of physiological events [1, 2], and the likely occurrence of oxidative stress during sheep pregnancy [56] and early lactation [57], this paper further supports the notion that Se is a key element that should routinely be included in a healthy diet also using Se-treated tree by-products. Discussion https://doi.org/10.1371/journal.pone.0201523.g004 hydrogen peroxide can induce both apoptosis and necrosis of cells as a consequence of aber- rant Ca2+ homeostasis associated with the release of Ca2+ from intracellular Ca2+ stores and/or an abnormal entry of the ion from the extracellular space [36]. In this work, we employed hydrogen peroxide to experimentally induce oxidative stress in sheep lymphomonocytes and, consequently, assess whether marked alterations of Ca2+ homeostasis could be rescued by Se, which was either added to the lymphomonocytes as an exogenous reagent, SeO4 2-, or appropriately included in the animals’ diet in order to physio- logically impact lymphomonocytes. To better discriminate the implication of Ca2+- stores depletion in relation to Ca2+ entry from the outer space, we employed a "Ca2+ add-back" protocol based on the supplementation of 1mM CaCl2 in the tissue culture medium in order to enhance the detection of mechanisms related to an inflow of exogenous ion or, alternatively, a store-operated release of intracellular Ca2+ [51]. The results indicate that, in sheep lymphomonocytes, H2O2 elicit a dose-dependent increase of cytosolic Ca2+ under Ca2+ free conditions as well as a result of CaCl2 supplementation in the incubation medium, thereby suggesting that H2O2 can effect both Ca2+ mobilization from intracellular stores as well as its entry through the cell membrane. These observations essen- tially recapitulate what others had reported in different cell types [51, 53]. However, here we found that Se can clearly antagonize either effect of H2O2 in vitro with comparable EC50 values (Fig 2), suggesting a direct impact of Se on Ca2+ homeostasis. Furthermore, we observed that Ca2+ homeostasis could effectively be rescued by Se introduced as part of the animal’s diet, which we experimentally designed by treating olive leaves with sodium selenate as the only 8 / 12 PLOS ONE \| https://doi.org/10.1371/journal.pone.0201523 July 30, 2018 Selenium maintains Ca2+ homeostasis in sheep lymphocytes challenged by oxidative stress source of that particular trace element. Notably, a concentration of Se in blood sheep account- ing to 1.9 μM impacted Ca2+ homeostasis similarly to 1.6–3.3 μM SeO4 2- added exogenously to the incubation medium. Furthermore, we observed that two control groups (C1 and C2), exhibited identical patterns of altered, H2O2-induced, Ca2+ homeostasis, thereby suggesting that any protecting activity shown by Se-treated leaves did not imply any contribution from other anti-oxidants contained in food such as, for example, natural phenols and polyphenols. PLOS ONE \| https://doi.org/10.1371/journal.pone.0201523 July 30, 2018 References 1. Djanaguiraman M, Devi DD, Shanker AK, Sheeba JA, Bangarusamy U. Selenium—an antioxidative protectant in soybean during senescence. Plant and Soil, 2005, 272 (issue 1–2): 77–86. https://doi.org/ 10.1007/s11104-004-4039-1 2. Dolph LH, Petra AT, Bradley A.C, Gladyshev VN. Selenium and selenocysteine: roles in cancer, health and Development. Trends Biochem Sci, 2014, 39(3): 112–120. https://doi.org/10.1016/j.tibs.2013.12. 007 PMID: 24485058 3. Pallud S, Lennon AM, Ramauge MA, Gavaret JM, Croteau W, Pierre M, et al. Expression of the type II iodothyronine deiodinase in cultured rat astrocytes is selenium-dependent. J Biochem Chem, 1997, 272: 18104–18110. https://doi.org/10.1074/jbc.272.29.18104 PLOS ONE \| https://doi.org/10.1371/journal.pone.0201523 July 30, 2018 9 / 12 Selenium maintains Ca2+ homeostasis in sheep lymphocytes challenged by oxidative stress 4. Rayman PM. The importance of selenium to human health. The Lancet, 2000, 356 (9225): 233–241. https://doi.org/10.1016/S0140-6736(00)02490-9 4. Rayman PM. The importance of selenium to human health. The Lancet, 2000, 356 (9225): 233–241. https://doi.org/10.1016/S0140-6736(00)02490-9 5. Wang C, Yang S, Zhang N, Mu Y, Ren H, Wang Y, et al. Long-Term supra nutritional supplementation with selenate decreases hyperglycemia and promotes fatty liver degeneration by inducing hyperinsuli- nemia in diabetic db/db mice. PLoS One, 2014. 9 (7): e101315. https://doi.org/10.1371/journal.pone. 0101315 PMID: 24983750 6. Barragry T. Veterinary Drug Therapy. Lea and Febiger USA, 1994. 6. Barragry T. Veterinary Drug Therapy. Lea and Febiger USA, 1994. 7. Hemingway RG. The influences of dietary intakes and supplementation with selenium and vitamin E on reproduction diseases and reproductive efficiency in cattle and sheep. Vet Res Commun, 2003, 27(2): 159–174. https://doi.org/10.1023/A:102287140 PMID: 12718509 8. Lemley CO, Meyer AM, Neville TL, Hallford DM, Camacho LE, Maddock-Carlin KR, et al. Dietary sele- nium and nutritional plane alter specific aspects of maternal endocrine status during pregnancy and lac- tation. Domest Anim Endocrinol, 2014, 46:1–11. https://doi.org/10.1016/j.domaniend.2013.09.006 PMID: 24183757 9. Shi L, Song R, Yao X, Ren Y. Effects of selenium on the proliferation, apoptosis and testosterone pro- duction of sheep Leydig cells in vitro. Theriogenology, 2017, 93: 24–32. https://doi.org/10.1016/j. theriogenology.2017.01.022 PMID: 28257863 10. Allison RD, Laven RA. Effect of vitamin E supplementation on the health and fertility of dairy cows: a review. Vet Rec, 2000, 147(25):703–708. https://doi.org/10.1136/vr.147.25.703 PMID: 11140928 11. Spears JW, Weiss WP. Role of antioxidants and trace elements in health and immunity of transition dairy cows. Vet J, 2008, 176(1):70–76. https://doi.org/10.1016/j.tvjl.2007.12.015 PMID: 18325801 12. Maas JP. 1983. Diagnosis and management of selenium responsive diseases in cattle. Compend. Con- tin Educ Pract Vet, 1983, 5: S393–399. 13. References Alhidary IA, Shini S, Al Jassim RA, Abudabos AM, Gaughan JB. Effects of selenium and vitamin E on performance, physiological response, and selenium balance in heat-stressed sheep. J Anim Sci, 2015, 93(2):576–88. https://doi.org/10.2527/jas.2014-8419 PMID: 26020746 14. Elgendy R, Giantin M, Castellani F, Grotta L, Palazzo F, Dacasto M, et al. Transcriptomic signature of high dietary organic selenium supplementation in sheep: A nutrigenomic in sight using a custom micro- array platform and gene set enrichment analysis. J Anim Sci, 2016, 94(8):3169–3184. https://doi.org/ 10.2527/jas.2016-0363 PMID: 27695782 15. Ungva´ri E´ , Monori I, Megyeri A, Csiki Z, Prokisch J, Sztrik A, et al. Protective effects of meat from lambs on selenium nanoparticle supplemented diet in a mouse model of polycyclic aromatic hydrocarbon- induced immune toxicity. Food ChemToxicol, 2014, 64: 298–306. https://doi.org/10.1016/j.fct.2013.12. 004 PMID: 24315870 16. Alimohamady R, Aliarabi H, Bahari A, Dezfoulian AH. Influence of Different Amounts and Sources of Selenium Supplementation on Performance, Some Blood Parameters, and Nutrient Digestibility in Lambs. Biol Trace Elem Res, 2013, 154 (1): 45–54. https://doi.org/10.1007/s12011-013-9698-4 PMID: 23677850 17. Kumar N, Garg AK, Mudgal V. Effect of different levels of selenium supplementation on growth rate, nutrient utilization, blood metabolic profile, and immune response in lambs. Biol Trace Elem Res, 2008, 126 (suppl. 1): S44–S56. https://doi.org/10.1007/s12011-008-8214-8 PMID: 18759066 18. Kumar M, Garg AK, Dass RS, Chaturvedi VK, Mudgal V, Varshney VP. Selenium supplementation influ- ences growth performance, antioxidant status and immune response in lambs. Anim Feed Sci Tech, 2009, 153: 77–87. https://doi.org/10.1016/j.anifeedsci.2009.06.007 19. Chung JY, Kim JH, Ko YH, Jang IS. Effects of Dietary Supplemented Inorganic and Organic Selenium on Antioxidant Defense Systems in the Intestine, Serum, Liver and Muscle of Korean Native Goats. Asian-Aust J Anim Sci, 2007, 20 (1): 52–59. https://doi.org/10.5713/ajas.2007.52 20. Tinggi U. Selenium: its role as antioxidant in human health. Environ Health Prev, 2008, 13(2), 102–108. https://doi.org/10.1007/s12199-007-0019-4 PMID: 19568888 21. Navarro-Alarcon M, Cabrera-Vique C. Selenium in food and the human body: A review. Sci total Envi- ron, 2008, 400: 115–141. https://doi.org/10.1016/j.scitotenv.2008.06.024 PMID: 18657851 22. Sopjani M, Fo¨ller M, Gulbins E, Lang F. Suicidal Death of Erythrocytes Due to Selenium-Compounds. Cell Physiol Biochem, 2008, 22: 387–394. https://doi.org/10.1159/000185452 PMID: 19088420 23. Olson OE. Selenium Toxicity in Animals with Emphasis on Man. Int J Toxicol, 1986, 5 (1): 45–70. https://doi.org/10.3109/10915818609140736 24. Zwolak I and Zaporowska H. Selenium interactions and toxicity: a review Cell Biology and Toxicology, 2012, 28 (1): 31–46. https://doi.org/10.1007/s10565-011-9203-9 PMID: 21913064 25. Thomson CD. PLOS ONE \| https://doi.org/10.1371/journal.pone.0201523 July 30, 2018 References Selenium speciation in human body fluids. Analyst, 1998, 123: 827–831. https://doi.org/ 10.1039/a707292i PMID: 9709477 10 / 12 PLOS ONE \| https://doi.org/10.1371/journal.pone.0201523 July 30, 2018 Selenium maintains Ca2+ homeostasis in sheep lymphocytes challenged by oxidative stress 26. Jones GD, Droz B, Greve P, Gottschalk P, Poffet D, McGrath SP, et al. Selenium deficiency risk pre- dicted to increase under future climate change. PNAS, 2017, 114 (11): 2848–2853. https://doi.org/10. 1073/pnas.1611576114 PMID: 28223487 27. Filley SJ, Peters A, Bouska C, Pirelli G, Oldfield J. Selenium Fertilization of Pastures for Improved For- age Selenium Content. The Professional Animal Scientist, 2007, 23(2): 144–147. https://doi.org/10. 15232/S1080-7446(15)30954-2 28. Se´boussi R, Tremblay GF, Ouellet V, Chouinard PY, Chorfi Y, Be´langer G, et al. Selenium-fertilized for- age as a way to supplement lactating dairy cows. J Dairy Sci, 2016, 99(7): 5358–5369. https://doi.org/ 10.3168/jds.2015-10758 PMID: 27085399 29. Tremblay GF, Belanger G, Lajeunesse J, Chouinard PY, and E. Charbonneau E´ . Timothy response to increasing rates of selenium fertilizer in Eastern Canada. Agron J, 2015, 107(1): 211–220. https://doi. org/10.2134/agronj14.0397 30. Hall JA, Van Saun RJ., Bobe G, Stewart WC, Voracheck WR, Mosher WD, et al. Organic and inorganic selenium: I. Oral bioavailability in ewes. J Anim Sci, 2012, 90(2): 568–576. https://doi.org/10.2527/jas. 2011-4075 PMID: 21965451 31. Cun GS, Robinson PH, Benes SE. Bioavailability of selenium in ‘Jose’ tall wheatgrass (Thinopyrum ponticum var ‘Jose’) hay as a substitute for sodium selenite in the diets of dairy cattle. Sci Total Environ, 2015, 518–519: 159–167. https://doi.org/10.1016/j.scitotenv.2015.02.091 PMID: 25747375 32. Givens DI, Allison R, Cottrill B, Blake JS. Enhancing the selenium content of bovine milk through alter- ation of the form and concentration of selenium in the diet of the dairy cow. J Sci Food Agric, 2004, 84 (8): 811–817. https://doi.org/10.1002/jsfa.1737 33. Go¨rlach A. Bertram K. Hudecova S. Krizanova O. Calcium and ROS: A mutual interplay. Redox Biol, 2015, 6: 260–271. https://doi.org/10.1016/j.redox.2015.08.010 PMID: 26296072 34. Reeves MA, Bellinger FP, Berry MJ. The neuro protective functions of Seleno protein M and its role in cytosolic calcium regulation. Antioxid Redox Signal, 2010, 12(7): 809–818. https://doi.org/10.1089/ars. 2009.2883 PMID: 19769485 35. Demirci S, Kutluhan S, Naziroğlu M, Uğuz AC, Yu¨rekli VA, Demirci K. Effects of selenium and topira- mate on cytosolic Ca2+ influx and oxidative stress in neuronal PC12 Cells. Neurochem Res, 2013, 38 (1): 90–97. https://doi.org/10.1007/s11064-012-0893-z PMID: 23011208 36. Yan Y, Wei CL, Zhang WR, Cheng HP, Liu J. Cross-talk between calcium and reactive oxygen species signaling. PLOS ONE \| https://doi.org/10.1371/journal.pone.0201523 July 30, 2018 References Acta Pharmacol Sin, 2006, 27(7): 821–826. https://doi.org/10.1111/j.1745-7254.2006. 00390.x PMID: 16787564 37. Brini M, Calı` T, Ottolini D, Carofoli E. Intracellular calcium homeostasis and signaling. In Banci L. editor. Metallomics and the Cell. Metal Ions in Life Sciences, 2013, 12: 119–168. https://doi.org/10.1007/978- 94-007-5561-1_5 PMID: 23595672 38. Clapham DE. Calcium signaling. Cell, 2007, 131(6): 1047–1058. https://doi.org/10.1016/j.cell.2007.11. 028 PMID: 18083096 39. Orrenius S, Gogvadze V, Zhivotovsky B. Calcium and mitochondria in the regulation of cell death. Bio- chem Biophys Res Commun, 2015, 460(1): 72–81. https://doi.org/10.1016/j.bbrc.2015.01.137 PMID: 25998735 40. Proietti P, Nasini L, Del Buono D, D’Amato R, Tedeschini E, Businelli D. Selenium protects olive (Olea europaea L.) from drought stress. Scientia Horticulturae, 2013, 164: 165–171. https://doi.org/10.1016/j. scienta.2013.09.034 41. D’Amato R, Proietti P, Nasini L, Del Buono D, Tedeschini E, Businelli D. Increase in the selenium con- tent of extra virgin olive oil: Quantitative and qualitative implications. Grasas Aceites, 2014, 65(2): 1–9. https://doi.org/10.3989/gya.097313 42. D’Amato R, Proietti P, Onofri A, Regni L, Esposto S, Servili M, et al. Biofortification (Se): Does it increase the content of phenolic compounds in virgin olive oil (VOO)? PLoS One, 2017, 12(4): e0176580. https://doi.org/10.1371/journal.pone.0176580 PMID: 28448631 43. Malagoli M, Schiavon M, Dall’Acqua S, Pilon-Smits EH. Effects of selenium biofortification on crop nutri- tional quality. Front Plant Sci, 2015, 6, 280. https://doi.org/10.3389/fpls.2015.00280 PMID: 25954299 44. SW-846 EPA Method 3051, Microwave assisted acid digestion of sediments, sludges, soils and oils, in: Test Methods for Evaluating Solid Waste, 3rd Edition, 3rd Update, US Environmental Protection Agency, Washington DC,1995. 45. Helrich K. Official method of analysis ( 15th ed). Association of Official Analytical Chemists (AOAC International), Arlington, VA, USA, 1990. 46. Cunniff P. Official method of analysis ( 16th ed). Association of Official Analytical Chemists (AOAC International), Arlington, VA, USA, 1996. 11 / 12 PLOS ONE \| https://doi.org/10.1371/journal.pone.0201523 July 30, 2018 Selenium maintains Ca2+ homeostasis in sheep lymphocytes challenged by oxidative stress 47. Horwitz W. Official method of analysis ( 17th ed). Association of Official Analytical Chemists (AOAC International), Gaithersburg, MD, USA, 2000. 48. Julshamn K, Maage A, Wallin HC. Determination of magnesium and calcium in foods by atomic absorp- tion spectrometry after microwave digestion: NKML collaborative study. J AOAC Int, 1998, 81(6):1202– 1208. 49. Guillen J, Prins JB, Smith D, Degryse AD. 2014. The European Framework on Research Animal Wel- fare Regulations and Guidelines In: Guillen J, ed. Laboratory Animals—Regulations and Recommenda- tions for Global Collaborative Research. San Diego: Academic Press, 2014. PLOS ONE \| https://doi.org/10.1371/journal.pone.0201523 July 30, 2018 References pp.117–202. https://doi. org/10.1016/B978-0-12-849880-4.00005–2 50. Grynkiewicz G, Poenie M, Tsien RY. A new generation of Ca2+ indicators with greatly improved fluores- cence properties. J Biol Chem, 1985, 260: 3440–3450. PMID: 3838314 51. Hecquet CM, Ahmmed GU, Vogel SM, Malik AB. Role of TRPM2 channel in mediating H2O2-induced Ca2+ entry and endothelial hyper permeability. Circ Res, 2008, 102(3): 347–355. https://doi.org/10. 1161/CIRCRESAHA.107.160176 PMID: 18048770 52. Van Metre DC, Callan RJ. Selenium and vitamin E. Vet Clin North Am Food Anim Pract, 2001 Jul; 17 (2):373–402, vii-viii. Vet Clin North Am Food Anim Pract, 2001, 17(2): 373–402. PMID: 11515407 53. Trebak M, Ginnan R, Singer HA, Jourd’heuil D. Inter play between calcium and reactive oxygen /nitro- gen species: an essential paradigm for vascular smooth muscle signalling. Antioxyd Redox Signal, 2010, 12(5): 657–674. https://doi.org/10.1089/ars.2009.2842 PMID: 19719386 54. Naziri E, Nenadis N, Mantzouridou FT, Tsimidou MZ. Valorization of the major agrifood industrial by- products and waste from Central Macedonia (Greece) for the recovery of compounds for food applica- tions. Food Res Int, 2014, 65 (C): 350–358. https://doi.org/10.1016/j.foodres.2014.09.013 55. Winans K, Kendall A, Deng H. The history and current applications of the circular economy concept. Renew Sust Energ Rev, 2017, 68(Part 1): 825–33. https://doi.org/10.1016/j.rser.2016.09.123 56. Mutinati M, Piccinno M, Roncetti M, Campanile D, Rizzo A, Sciorsci R. Oxidative stress during preg- nancy in the sheep. Reprod Domest Anim, 2013, 48(3): 353–357. https://doi.org/10.1111/rda.12141 PMID: 23346938 57. Celi P. Biomarkers of oxidative stress in ruminant medicine. Immunopharmacol Immunotoxicol, 2011, 33(2): 233–240. https://doi.org/10.3109/08923973.2010.514917 PMID: 20849293 PLOS ONE \| https://doi.org/10.1371/journal.pone.0201523 July 30, 2018 12 / 12
https://openalex.org/W4288490374	https://zenodo.org/records/6918399/files/ZDIF1924.pdf	Russian	null	ЗНАЧЕНИЕ УСТОЙЧИВОГО ЭКОНОМИЧЕСКОГО РОСТА В УСЛОВИЯХ НОВЫХ ГЕОЭКОНОМИЧЕСКИХ РЕАЛИЙ	Zenodo (CERN European Organization for Nuclear Research)	2,022	cc-by	2,079	«Zamonaviy dunyoda ijtimoiy fanlar: nazariy va amaliy izlanishlar» nomli ilmiy, masofaviy, onlayn konferensiya ЗНАЧЕНИЕ УСТОЙЧИВОГО ЭКОНОМИЧЕСКОГО РОСТА В УСЛОВИЯХ НОВЫХ ГЕОЭКОНОМИЧЕСКИХ РЕАЛИЙ Қодиров Фозил Баходир ўғли УМЭД факультет Международная экономика и менеджмент Магистратура 1 курс https://doi.org/10.5281/zenodo.6918399 «Zamonaviy dunyoda ijtimoiy fanlar: nazariy va amaliy izlanishlar» nomli ilmiy, masofaviy, onlayn konferensiya Аннотация Работа посвящена обоснованию необходимости переходе от моделей экономического роста, абстрагирующихся от эколого-экономических противоречий, к устойчивому развитию. Раскрываются сущностные характеристики устойчивого развития. Ключевые слова: экономический рост, расширенное воспроизводство, экономическое развитие, устойчивое развитие, эколого-экономическое равновесие. Экономическая политика ведущих стран мира за последние два столетия базировалась на теории экономического роста, направленного на безудержное наращивание темпов и масштабов производства на основе все возрастающего использования невоспроизводимых природных ресурсов. На ранних этапах развития вмешательство человека в природу носило несистемный, локальный характер и поэтому оно не сопровождалось комплексным разрушением природной среды. Это объясняет позицию многих известных представителей экономической науки, которые придерживались мнения о неисчерпаемости природных ресурсов. Так, например, Д. Рикардо утверждал, что «ничего не платится за включение природных агентов, поскольку они неисчерпаемы и доступны всем». Ж. Сэй полагал, что природные ресурсы в силу их неограниченности и того, что мы их получаем даром «не представляют собой объекта экономической науки». К. Маркс, развивая теорию трудовой стоимости, в свою очередь, писал: «силы природы не стоят ничего; они входят в процесс труда, не входя в процесс образования стоимости». Подобные взгляды были присущи также А. Смиту, А. Маршаллу и другим ученым. В эпоху бурного развития капитализма и промышленного прогресса человечество стало ориентироваться на изучении природы, но, прежде всего, с целью интенсивного использованию в производстве ее ресурсов. Популярной стала идея Ф. Бэкона: «Знание – сила», которая фактически была трансформирована в лозунг покорения человеком природы. В 137 «Zamonaviy dunyo izlanishlar» nomli «Zamonaviy duny «Zamonaviy dunyoda ijtimoiy fanlar: nazariy va amaliy izlanishlar» nomli ilmiy, masofaviy, onlayn konferensiya соответствии с этим считалось, что человек может быть либо ее рабом, либо непримиримым врагом, завоевателем, самовластным диктатором. На опасность таких взглядов все же обращали внимание некоторые мыслители. Например, Гегель констатировал, что человек, желающий познать мир, стремится «завладеть им и подчинить его себе. Он пользуется природой против нее же самой. «Потребность стремиться к тому, чтобы употребить природу в свою пользу, стирать ее грани, истощать ее, короче говоря, уничтожать ее». Последние десятилетия минувшего века в развитых странах ознаменовались принципиальным переосмыслением моделей экономического развития. Концепции, основанные на абсолютизации количественных аспектов общественного производства, начали терять свое значение. Они пришли в противоречие с новыми условиями хозяйствования, накладывающие серьезные ресурсные и экологические ограничения на дальнейший экстенсивный рост производства. Одной из распространенных взглядов является противопоставление понятий «экономическое развитие» и «экономический рост». Экономический рост трактуется как простое наращивание производственных результатов национального хозяйства за определенный период, за рамками которого остаются качественные изменения. Экономическое развитие, напротив, определяется с позиции качественных характеристик, в отрыве от количественной стороны. Например, Й. Аннотация Приверженность новой концепции экономического развития означает учет следующих императивов: человечество способно придать развитию устойчивый характер, отвечающий потребностям нынешних и будущих поколений; имеющиеся ограничения в области эксплуатации природных ресурсов относительны, связаны с современным уровнем техники и социальной организации; общество обязано удовлетворять базовые потребности людей и предоставить им право на достойную жизнь; необходимо согласовать образ жизни стран и людей, которые располагают большими денежными средствами с экологическими возможностями Земли; темпы роста населения должны быть согласованы с производительным потенциалом экосистемы планеты; обеспечение преимущественно интенсивного типа экономического роста; сохранение и укрепление природно- экологической среды; представление потомкам достаточных условий по удовлетворению их потребностей в природных ресурсах; поиск качественно новых возможностей использования ресурсосберегающих и энергосберегающих технологий, альтернативных и воспроизводимых источников производства; создание благоприятных эколого-экономических и социальных условий для накопления и реализации человеческого капитала; определение человека в качестве главного мерила и цели общественного развития. Сущность устойчивого развитие мы определяем как институционализированную форму экономического развития, связанную «Zamonaviy dunyo izlanishlar» nomli й «Zamonaviy duny «Zamonaviy dunyoda ijtimoiy fanlar: nazariy va amaliy izlanishlar» nomli ilmiy, masofaviy, onlayn konferensiya «Zamonaviy dunyoda ijtimoiy fanlar: nazariy va amaliy izlanishlar» nomli ilmiy, masofaviy, onlayn konferensiya всеобъемлющего характера, который носит понятие «экономическое развитие». Экономический рост составляет частную сторону развития, которое, кроме того, включает любые прогрессивные изменения, происходящие в области экономической, социальной и экологической деятельности общества. всеобъемлющего характера, который носит понятие «экономическое развитие». Экономический рост составляет частную сторону развития, которое, кроме того, включает любые прогрессивные изменения, происходящие в области экономической, социальной и экологической деятельности общества. Важнейшим свойством экономического развития является её сбалансированность, нарушение которой означает выход в зону, порождающая экономические, социальные и экологические диспропорции. Можно обеспечить высокие темпы макроэкономической динамики и не создавать при этом необходимых условий для последующего развития. Поэтому диапазон экономического развития представляет собой долгосрочную или устойчивую тенденцию, когда позитивные изменения не нарушают необходимых условий воспроизводства за пределами краткосрочного периода. Аннотация Шумпетер под экономическим развитием подразумевал качественные имманентные изменения хозяйственного кругооборота, под вторым - обычное увеличение населения и богатства. Другими словами, экономический рост отражает простое увеличение реального ВВП и выражает значительно более узкий круг связей и отношений, чем экономическое развитие. На наш взгляд, экономический рост, реализуется в единстве количественных и качественных характеристик, диалектическое взаимодействие которых обусловливает переход к более высокому, качественно новому уровню экономического состояния, что и отражается понятием «экономическое развитие». В условиях перехода к постиндустриальной стадии развития, когда возрастает значение соответствии с этим считалось, что человек может быть либо ее рабом, либо непримиримым врагом, завоевателем, самовластным диктатором. На опасность таких взглядов все же обращали внимание некоторые мыслители. Например, Гегель констатировал, что человек, желающий познать мир, стремится «завладеть им и подчинить его себе. Он пользуется природой против нее же самой. «Потребность стремиться к тому, чтобы употребить природу в свою пользу, стирать ее грани, истощать ее, короче говоря, уничтожать ее». Последние десятилетия минувшего века в развитых странах ознаменовались принципиальным переосмыслением моделей экономического развития. Концепции, основанные на абсолютизации количественных аспектов общественного производства, начали терять свое значение. Они пришли в противоречие с новыми условиями хозяйствования, накладывающие серьезные ресурсные и экологические ограничения на дальнейший экстенсивный рост производства. Одной из распространенных взглядов является противопоставление понятий «экономическое развитие» и «экономический рост». Экономический рост трактуется как простое наращивание производственных результатов национального хозяйства за определенный период, за рамками которого остаются качественные изменения. Экономическое развитие, напротив, определяется с позиции качественных характеристик, в отрыве от количественной стороны. Например, Й. Шумпетер под экономическим развитием подразумевал качественные имманентные изменения хозяйственного кругооборота, под вторым - обычное увеличение населения и богатства. Другими словами, экономический рост отражает простое увеличение реального ВВП и выражает значительно более узкий круг связей и отношений, чем экономическое развитие. На наш взгляд, экономический рост, реализуется в единстве количественных и качественных характеристик, диалектическое взаимодействие которых обусловливает переход к более высокому, качественно новому уровню экономического состояния, что и отражается понятием «экономическое развитие». В условиях перехода к постиндустриальной стадии развития, когда возрастает значение факторов инновации, человеческого капитала, информации, содержание этих категорий становятся все более близкими. Вместе с тем, их не следует отождествлять. При всем широком рассмотрении содержания экономического роста, он не имеет того «Zamonaviy dunyoda ijtimoiy fanlar: nazariy va amaliy izlanishlar» nomli ilmiy, masofaviy, onlayn konferensiya всеобъемлющего характера, который носит понятие «экономическое развитие». Экономический рост составляет частную сторону развития, которое, кроме того, включает любые прогрессивные изменения, происходящие в области экономической, социальной и экологической деятельности общества. Аннотация Важнейшим свойством экономического развития является её сбалансированность, нарушение которой означает выход в зону, порождающая экономические, социальные и экологические диспропорции. Можно обеспечить высокие темпы макроэкономической динамики и не создавать при этом необходимых условий для последующего развития. Поэтому диапазон экономического развития представляет собой долгосрочную или устойчивую тенденцию, когда позитивные изменения не нарушают необходимых условий воспроизводства за пределами краткосрочного периода. Приверженность новой концепции экономического развития означает учет следующих императивов: человечество способно придать развитию устойчивый характер, отвечающий потребностям нынешних и будущих поколений; имеющиеся ограничения в области эксплуатации природных ресурсов относительны, связаны с современным уровнем техники и социальной организации; общество обязано удовлетворять базовые потребности людей и предоставить им право на достойную жизнь; необходимо согласовать образ жизни стран и людей, которые располагают большими денежными средствами с экологическими возможностями Земли; темпы роста населения должны быть согласованы с производительным потенциалом экосистемы планеты; обеспечение преимущественно интенсивного типа экономического роста; сохранение и укрепление природно- экологической среды; представление потомкам достаточных условий по удовлетворению их потребностей в природных ресурсах; поиск качественно новых возможностей использования ресурсосберегающих и энергосберегающих технологий, альтернативных и воспроизводимых источников производства; создание благоприятных эколого-экономических и социальных условий для накопления и реализации человеческого капитала; определение человека в качестве главного мерила и цели общественного развития. Сущность устойчивого развитие мы определяем как институционализированную форму экономического развития, связанную «Zamonaviy dunyoda ijtimoiy fanlar: nazariy va amaliy izlanishlar» nomli ilmiy, masofaviy, onlayn konferensiya всеобъемлющего характера, который носит понятие «экономическое развитие». Экономический рост составляет частную сторону развития, которое, кроме того, включает любые прогрессивные изменения, происходящие в области экономической, социальной и экологической деятельности общества. Важнейшим свойством экономического развития является её сбалансированность, нарушение которой означает выход в зону, порождающая экономические, социальные и экологические диспропорции. Можно обеспечить высокие темпы макроэкономической динамики и не создавать при этом необходимых условий для последующего развития. Поэтому диапазон экономического развития представляет собой долгосрочную или устойчивую тенденцию, когда позитивные изменения не нарушают необходимых условий воспроизводства за пределами краткосрочного периода. Аннотация 139 Приверженность новой концепции экономического развития означает учет следующих императивов: человечество способно придать развитию устойчивый характер, отвечающий потребностям нынешних и будущих поколений; имеющиеся ограничения в области эксплуатации природных ресурсов относительны, связаны с современным уровнем техники и социальной организации; общество обязано удовлетворять базовые потребности людей и предоставить им право на достойную жизнь; необходимо согласовать образ жизни стран и людей, которые располагают большими денежными средствами с экологическими возможностями Земли; темпы роста населения должны быть согласованы с производительным потенциалом экосистемы планеты; обеспечение преимущественно интенсивного типа экономического роста; сохранение и укрепление природно- экологической среды; представление потомкам достаточных условий по удовлетворению их потребностей в природных ресурсах; поиск качественно новых возможностей использования ресурсосберегающих и энергосберегающих технологий, альтернативных и воспроизводимых источников производства; создание благоприятных эколого-экономических и социальных условий для накопления и реализации человеческого капитала; определение человека в качестве главного мерила и цели общественного развития. Сущность устойчивого развитие мы определяем как институционализированную форму экономического развития, связанную с поступательным движением общества в целях достижения объективно прогрессивной системы общественных ценностей, на основе Сущность устойчивого развитие мы определяем как институционализированную форму экономического развития, связанную с поступательным движением общества в целях достижения объективно прогрессивной системы общественных ценностей, на основе Сущность устойчивого развитие мы определяем как институционализированную форму экономического развития, связанную с поступательным движением общества в целях достижения объективно прогрессивной системы общественных ценностей, на основе «Zamonaviy dunyo izlanishlar» nomli «Zamonaviy duny «Zamonaviy dunyoda ijtimoiy fanlar: nazariy va amaliy izlanishlar» nomli ilmiy, masofaviy, onlayn konferensiya гармоничного сочетании экономических, экологических и социальных факторов. Оно представляет собой смену общественных приоритетов от безудержного экономического роста к эффективному использованию и сохранению природного и человеческого потенциала. Устойчивое развитие связано с формированием принципиально нового отношения к человеку, с одной стороны, как субъекта эколого- экономической системы, с другой - как главной цели его развития, более полной реализации его потенциала. Устойчивое развитие есть относительно самостоятельная категория, имеющая собственное субстанциональное бытие, не сводимое к другим экономическим категориям. В то же время устойчивое развитие находится в сфере взаимовлияния со всей совокупностью социально-экономических, организационно-экономических и институциональных отношений. Зак ю е е Таким образом, принципиально важным видится положение, отражающее необходимость не только увеличения и качественного совершенствование национального продукта, но и обеспечение последующей его реализации. Прежде всего, речь идет о проблеме формирования эффективных агентов, осуществляющих его присвоение. В связи с этим неизмеримо возрастают требования непосредственно к институциональной форме воспроизводства и присвоения национального продукта. Институциональная система должна не только фиксировать факт расширенного воспроизводства и присвоения общественного продукта, но и определять прогрессивную, социально-ориентированную динамику его использования. Список использованной литературы: 1. Цит. по: Митякова О.Н. Устойчивое развитие как экономическая категория// Экономическая теория. – 2009. - № 3. – С. 99. 2. Маркс К. Экономическая рукопись 1861-1863 годов/ Маркс К., Энгельс Ф. Соч. 2-е изд. – М.: Госполитиздат, 1962. т.47. – С. 498. 3. Митякова О.Н. Устойчивое развитие как экономическая категория // Экономическая теория. – 2009. - № 3. – С. 99. 4. Гегель Г. Философия права. – М.: Мир книги, 2007. – С. 464. 5. Цит. по: Митякова О.Н. Устойчивое развитие как экономическая категория // Экономическая теория. – 2009. - № 3. – С. 100. «Zamonaviy dunyoda ijtimoiy fanlar: nazariy va amaliy izlanishlar» nomli ilmiy, masofaviy, onlayn konferensiya 6. Маркс К., Энгельс Ф. Соч. 2-е изд., т. 20 – М.: Госполитиздат, 1962. – С. 495-496. 6. Маркс К., Энгельс Ф. Соч. 2-е изд., т. 20 – М.: Госполитиздат, 1962. – С. 495-496. 6. Маркс К., Энгельс Ф. Соч. 2-е изд., т. 20 – М.: Госполитиздат, 1962. – С. 495-496. 7. Цит. по: Губайдуллина Т.Н. Устойчивое развитие эколого- экономической системы: вопросы теории и методологии. – Казань: Изд-во КГУ, 2000. – С. 83. 7. Цит. по: Губайдуллина Т.Н. Устойчивое развитие эколого- экономической системы: вопросы теории и методологии. – Казань: Изд-во КГУ, 2000. – С. 83. 8. Бухарин Н.И. К постановке проблем теории исторического материализма, 1923 г. // Избранные произведения. М.: Политиздат, 1988. 8. Бухарин Н.И. К постановке проблем теории исторического материализма, 1923 г. // Избранные произведения. М.: Политиздат, 1988.
https://openalex.org/W4226227792	https://scholarlypublications.universiteitleiden.nl/access/item%3A3444029/view	English	null	Prevalence, risk factors, and long-term outcomes of cerebral ischemia in hospitalized COVID-19 patients – study rationale and protocol of the CORONIS study: A multicentre prospective cohort study	European stroke journal	2,022	cc-by	5,169	1Department of Neurology, Donders Center for Medical Neuroscience, Radboud University Medical Center, Nijmegen, The Netherlands 2Department of Neurology and Neurosurgery, Brain Center, University Medical Center Utrecht, Utrecht, The Netherlands 3Department of Neurology, Leiden University Medical Center, Leiden, The Netherlands 4Department of Medical Imaging, Radboud University Medical Center, Nijmegen, The Netherlands 5Department of Radiology, Leiden University Medical Center, Leiden, the Netherlands 6Department of Radiology and Nuclear Medicine, University Medical Center, Utrecht University, Utrecht, The Netherlands 7Department of Cardiology, Radboud University Medical Center, Nijmegen, The Netherlands 8Department of Cardiology, Heart and Lung Centre, Leiden University Medical Centre, Leiden, The Netherlands 9Department of Internal Medicine, Radboud University Nijmegen Medical Center, Nijmegen, The Netherland 10Department of Internal Medicine and Radboud Center for Infectious Diseases, Radboud University Medical Center, Nijmegen, The Netherlands 11Department of Medicine – Thrombosis and Hemostasis, Leiden University Medical Centre, Leiden, The Netherlands 12Department of Clinical Epidemiology, Leiden University Medical Centre, Leiden, The Netherlands Prevalence, risk factors, and long-term outcomes of cerebral ischemia in hospitalized COVID-19 patients - study rationale and protocol of the CORONIS study: a multicentre prospective cohort study Lith, T.J. van; Sluis, W.M.; Wijers, N.T.; Meijer, F.J.A.; Kamphuis-van Ulzen, K.; Bresser, J. de; ... ; Leeuw, F.E. de Citation Lith, T. J. van, Sluis, W. M., Wijers, N. T., Meijer, F. J. A., Kamphuis-van Ulzen, K., Bresser, J. de, … Leeuw, F. E. de. (2022). Prevalence, risk factors, and long-term outcomes of cerebral ischemia in hospitalized COVID-19 patients - study rationale and protocol of the CORONIS study: a multicentre prospective cohort study. European Stroke Journal, 7(2), 180-187. doi:10.1177/23969873221092538 Citation Lith, T. J. van, Sluis, W. M., Wijers, N. T., Meijer, F. J. A., Kamphuis-van Ulzen, K., Bresser, J. de, … Leeuw, F. E. de. (2022). Prevalence, risk factors, and long-term outcomes of cerebral ischemia in hospitalized COVID-19 patients - study rationale and protocol of the CORONIS study: a multicentre prospective cohort study. European Stroke Journal, 7(2), 180-187. doi:10.1177/23969873221092538 Citation Lith, T. J. van, Sluis, W. M., Wijers, N. T., Meijer, F. J. A., Kamphuis-van Ulzen, K., Bresser, J. de, … Leeuw, F. E. de. (2022). Prevalence, risk factors, and long-term outcomes of cerebral ischemia in hospitalized COVID-19 patients - study rationale and protocol of the CORONIS study: a multicentre prospective cohort study. European Stroke Journal, 7(2), 180-187. doi:10.1177/23969873221092538 Version: Publisher's Version License: Creative Commons CC BY 4.0 license Downloaded from: https://hdl.handle.net/1887/3307285 Version: Publisher's Version License: Creative Commons CC BY 4.0 license Downloaded from: https://hdl.handle.net/1887/3307285 License: Note: To cite this publication please use the final published version (if applicable). 92538 ESO0010.1177/23969873221092538European Stroke Journalvan Lith et al. 92538 ESO0010.1177/23969873221092538European Stroke Journalvan Lith et al. Study Protocol European Stroke Journal 2022, Vol. 7(2) 180–187 European Stroke Journal 2022, Vol. 7(2) 180–187 © European Stroke Organisation 2022 Prevalence, risk factors, and long-term outcomes of cerebral ischemia in hospitalized COVID-19 patients – study rationale and protocol of the CORONIS study: A multicentre prospective cohort study Prevalence, risk factors, and long-term outcomes of cerebral ischemia in hospitalized COVID-19 patients – study rationale and protocol of the CORONIS study: A multicentre prospective cohort study © European Stroke Organisation 2022 https://doi.org/10.1177/23969873221092538 Article reuse guidelines: sagepub.com/journals-permissions DOI: 10.1177/23969873221092538 journals.sagepub.com/home/eso https://doi.org/10.1177/2396987322109253 Article reuse guidelines: sagepub.com/journals-permissions DOI: 10.1177/23969873221092538 journals.sagepub.com/home/eso Theresa J van Lith1* , Wouter M Sluis2, Naomi T Wijers3, Frederick JA Meijer4 , Karin Kamphuis-van Ulzen4, Jeroen de Bresser5, Jan Willem Dankbaar6, Frederik MA van den Heuvel7, M Louisa Antoni8, Catharina M Mulders-Manders9, Quirijn de Mast10, Frank L van de Veerdonk10, Frederikus A Klok11 , Anil M Tuladhar1, Suzanne C Cannegieter11,12, Marieke JH Wermer3, H Bart van der Worp2 , Menno V Huisman11* and Frank-Erik de Leeuw1* Co espo d g aut o : Frank-Erik de Leeuw, Department of Neurology (935), Radboud University Medical Centre, PO Box 9101, 6500 HB Nijmegen, The Netherlands. Email: FrankErik.deLeeuw@radboudumc.nl Corresponding author: Frank-Erik de Leeuw, Department of Neurology (935), Radboud University Medical Centre, PO Box 9101, 6500 HB Nijmegen, The Netherlands. Email: FrankErik.deLeeuw@radboudumc.nl Abstract Background: COVID-19 is often complicated by thrombo-embolic events including ischemic stroke. The underlying mechanisms of COVID-19-associated ischemic stroke, the incidence and risk factors of silent cerebral ischemia, and the long-term functional outcome in these patients are currently unknown. Patients and methods: CORONavirus and Ischemic Stroke (CORONIS) is a multicentre prospective cohort study investigating the prevalence, risk factors and long-term incidence of (silent) cerebral ischemia, and the long-term functional outcome among patients with COVID-19. We aim to include 200 adult patients hospitalized with COVID-19 without symptomatic ischemic stroke to investigate the prevalence of silent cerebral ischemia compared with 60 (matched) controls with MRI. In addition, we will identify potential risk factors and/or causes of cerebral ischemia in COVID-19 patients with (n = 70) or without symptomatic stroke (n = 200) by means of blood sampling, cardiac workup and brain MRI. We will measure functional outcome and cognitive function after 3 and 12 months with standardized questionnaires in all patients with COVID-19. Finally, the long-term incidence of (new) silent cerebral ischemia in patients with COVID-19 will be assessed with follow up MRI (n = 120). p p ( ) Summary: The CORONIS study is designed to add further insight into the prevalence, long-term incidence and risk factors of cerebral ischemia, and the long-term functional outcome in hospitalized adult patients with COVID-19. Corresponding author: p g Frank-Erik de Leeuw, Department of Neurology (935), Radboud University Medical Centre, PO Box 9101, 6500 HB Nijmegen, The Netherlands. Email: FrankErik.deLeeuw@radboudumc.nl 181 van Lith et al. Keywords COVID-19, SARS-CoV-2, silent cerebral ischemia, ischemic stroke Date received: 11 February 2022; accepted: 16 March 2022 Controls Age- and sex-matched adult (⩾18 years) controls without previous COVID-19 infection from the general population will be recruited among the patients’ next of kin or social environment. Previous studies have mainly investigated symptomatic ischemic stroke. However, it may very well be that “clini- cally silent” ischemic brain lesions occur due to a proco- agulant or proinflammatory COVID-19 response which can impair recovery.8,13 Insight into the magnitude, causes and long-term outcomes of cerebral ischemia in hospital- ized patients with COVID-19 is crucial for patient care to provide optimal diagnostic strategies and prophylactic and therapeutic treatment. Introduction and rationale another apparent cause. “TIA” must be diagnosed based on transient focal neurological symptoms lasting <24 h pre- sumed to be due to focal brain, spinal cord, or retinal ischemia without evidence of acute infarction by neuroim- aging or pathology (or in the absence of imaging).14 The clinical course of coronavirus disease 2019 (COVID- 19) is complicated by a high risk of thrombo-embolic com- plications, with a higher incidence in patients admitted to the intensive care unit (ICU) compared to the general ward.1–3 The occurrence of these complications is associ- ated with a poor clinical outcome and a higher risk of mor- tality.2 The majority of these events consist of venous thrombosis and pulmonary embolism.4,5 However, arterial ischemic events, such as ischemic stroke have also been described.6–10 The study will be performed in three Dutch academical hospitals: Radboud University Medical Center, Leiden University Medical Center (LUMC), and the University Medical Center Utrecht (UMCU). Patients from other hos- pitals in the Netherlands can be referred to the participating hospitals for participation in the study. Table 1 summarizes the in- and exclusion criteria. Possible mechanisms of COVID-19-associated ischemic stroke include generalized coagulopathy, systemic embo- lism secondary to atrial fibrillation, paradoxical (venous) emboli due to a patent foramen ovale (PFO), arterial throm- bosis, and arterial wall inflammation of the cerebral or cer- vical arteries.11,12 Design The CORONavirus and Ischemic Stroke (CORONIS) study is a multicentre prospective observational cohort study. The study was approved by the medical ethics committee region Arnhem–Nijmegen and all patients will provide written informed consent. Study procedures and follow-up All eligible patients will be recruited during admission or shortly after discharge. Baseline measurements will be executed during admission or during a visit in the outpa- tient department (T0). Follow-up at 3 (T1) and 12 (T2) months after inclusion consists of a telephone inter- view using standardized questionnaires including cogni- tive assessment. A follow-up brain MRI will be performed 3 months after baseline MRI in a random sample of the patients with COVID-19 (n = 120) (T1). In control subjects we will only perform baseline questionnaires and brain MRI. Study procedures are described in Table 2. Study objectives The main study objectives are: 1. To determine the prevalence of asymptomatic (silent) cerebral ischemia on MRI in patients with COVID-19 compared to controls. The CORONIS study was designed to investigate the prevalence, risk factors, and the long-term effects of (silent) cerebral ischemia in hospitalized patients with COVID-19. 2. To assess causes of cerebral ischemia in patients with COVID-19. 3. To measure functional outcome and cognitive function in patients with COVID-19 after 3 and 12 months. Methods 4. To determine the incidence of new cerebral ischemia on MRI after 3 months of follow-up in patients with COVID-19. Patient population Two hundred hospitalized patients with laboratory-con- firmed COVID-19 infection, without a symptomatic ischemic stroke, will be included. In addition, 70 patients with a symptomatic ischemic stroke or transient ischemic attack (TIA) will be included. Ischemic stroke’ must be confirmed with neuroimaging demonstrating either infarc- tion in the corresponding vascular territory or absence of European Stroke Journal 7(2) 182 Table 1. Inclusion and exclusion criteria of the CORONIS study. Inclusion and exclusion criteria: Inclusion criteria Age ⩾18 years Admitted to the hospital because of COVID-19 Exclusion criteria MRI contraindication and/or post COVID-19 disability interfering with MRI acquisition (e.g. severe delirium) eGFR ⩽ 30 ml/min Pregnancy Limited life expectancy (<3 months) Major disease interfering with study participation or follow-up Not able to give informed consent COVID-19: coronavirus disease 2019; MRI: magnetic resonance imaging; eGFR: estimated glomerular filtration rate. Table 2. Study assessments. Assessment COVID-19 patients Controls Study phase: Baseline (T0) 3 months follow-up (T1) 1 year follow-up (T2) Baseline Medical history + vascular risk factors x x x x Medication use x x x x Recurrent events x x Demographics x x Questionnaires (education, lifestyle) x x Functional outcome: mRS x x x x Functional outcome post-COVID: PCFS x x – Mood questionnaire: HADS x x – Cognitive assessment x x x – Blood chemistry x – Biobanking x – Contrast transthoracic echocardiography x – 48–72 h heart rhythm monitoring x – Brain MRI x x x mRS: modified Rankin Scale; HADS: hospital and anxiety depression score; PCFS: post-COVID functional scale; MRI: magnetic resonance imaging. Baseline questionnaires Medical history. For each patient a history of diabetes mel- litus, hypertension, hypercholesterolemia, atrial fibrilla- tion, TIA, ischemic and hemorrhagic stroke, myocardial infarction, peripheral arterial disease, venous thromboem- bolism, lung diseases, autoimmune diseases, or malignancy will be collected. A past medical history of other neurologi- cal disease than the above will be recorded if applicable. The presence of a family history of all of the above, current medication use and vaccination status will be recorded. Demographics, lifestyle, and functional outcome. A struc- tured questionnaire will be used at baseline to assess demographic data (age, sex, body mass index (BMI), eth- nicity, and education) and lifestyle behavior. Education will be classified using seven categories: one being less than primary school and seven reflecting an academic degree.15 Questions regarding lifestyle include current or past nicotine, alcohol, and illicit drug use. Alcohol con- sumption is defined as units per day and the age alcohol consumption started (and if applicable stopped). Smok- ing behavior is defined as the number of pack years, cal- culated as the number of packs of cigarettes smoked per day multiplied by the number of years a patient has smoked. Cognitive assessment At baseline, the patients with COVID-19 will undergo a short cognitive screening with the Montreal Cognitive Assessment (MoCA). This 10-min test covers various cog- nitive domains, including memory, visuoconstruction, attention, executive functioning, and language.21 During follow-up, patients will undergo the Telephone Interview for Cognitive Status (TICS), which covers verbal and work- ing memory, orientation, language, and attention.22 MRI abnormalities. Brain MRIs will be evaluated for acute or previous ischemic lesions, markers for cerebral small vessel disease, and intracranial vessel wall abnormalities. An acute ischemic lesion is defined by the presence of restricted diffusion on DWI. Markers of cerebral small ves- sel disease are defined as recent small (sub)cortical infarcts, white matter hyperintensities of presumed vascular origin, and cerebral microbleeds. The markers of small vessel dis- ease are assessed in concordance with the STandards for ReportIng Vascular changes on nEuroimaging (STRIVE) criteria for cerebral small vessel disease.20 Follow-up questionnaires Through a telephone interview by one of the researchers, patients will undergo structured questionnaires at 3 and 12 months after baseline testing on the occurrence of new cardiovascular events, lung diseases, and persistence of COVID-19 symptoms such as fatigue and dyspnea. Presence of actual depressive or anxiety symptoms will be assessed Functional performance before hospital admission will be assessed by the modified Rankin Scale (mRS). 183 van Lith et al. Table 3. Scanning protocol MRI. Participating hospital Radboudumc Nijmegen LUMC UMCU Type of scanner Siemens 3T Prisma Philips 3T Ingenia Ingenia Elition 3T X Contrast agent 15 ml Dotarem® (0.5 mmol/ml) Clariscan (0.2 ml/kg) 0.1 ml Gadovist/kg Duration (in min) 40 35 25 T1-weighted Orientation 3D space fatsat 3D T1 Axial Voxels 0.9 mm isotropic 1.15 mm isotropic 0.5 × 0.5 × 0.5 mm FLAIR Orientation 3D space flair fatsat 2D FLAIR Axial Voxel + resolution 1 mm isotropic 0.7 × 0.7 × 5.00 mm 0.6 × 0.6 × 4 mm Diffusion weighted imaging (DWI) Orientation Axial Resolve Axial Axial Target-slice thickness + resolution 5 mm 5 mm 4 mm Susceptibility weighted imaging (SWI) Orientation Axial 3D Axial Target-slice thickness + resolution 3 mm 2 mm 2 mm Intracranial vessel wall imaging with and without contrast Orientation 3D space fatsat 3D Axial Target-slice thickness + resolution 0.9 mm isotropic 0.6 × 0.6 × 1.0 mm 0.5 mm isotropic Diffusion tensor image (DTI) 2 mm B0, 1000, 2000 64 directions – – Radboudumc: Radboud University Medical Center; LUMC: Leiden University Medical Center; UMCU: University Medical Center Utrecht; FLAIR: fluid-attenuated inversion recovery. ud University Medical Center; LUMC: Leiden University Medical Center; UMCU: University Medical Center Utrecht; FLAIR i using the HADS scale.16 Patients will also be asked about current medication use. Functional outcome will be deter- mined using the Post-COVID-19 Functional Status (PCFS) scale and the modified Rankin Scale (mRS).17–19 Intracranial vessel wall abnormalities are defined as major vessel wall changes, such as dissections, occlusions, and stenoses. In addition, images will be assessed for enhancing foci specified for the vessel segment. Vessel wall enhancement is classified as concentric or eccentric type enhancement. The MRI characteristics of interest are showed in Table 4. Brain MRI Brain MRI scans will be rated qualitatively following a standardized, structured protocol by experienced neurora- diologists blinded to clinical data. The MRI protocol is designed to detect acute and chronic cerebral ischemia, markers of cerebral small vessel disease and vessel wall abnormalities. Table 3 presents the MRI scanning protocol in each participating center. Contrast transthoracic echocardiography To assess presence of PFO, patients will undergo agitated saline contrast transthoracic echocardiography. All echocar- diography examinations will be performed by an experi- enced cardiologist. The interatrial septum will be assessed in multiple views. In addition shunting will be evaluated with European Stroke Journal 7(2) 184 Table 4. MRI characteristics of interest. MRI-sequence Outcome Additional information FLAIR White matter hyperintensities (WMH) Fazekas (0/1/2/3) Previous cerebral infarction Location: Local, multifocal Cortical, lacunar Signs of delayed cerebral hypoxia DWI Acute ischemic lesions, DWI + lesion Location: Local, multifocal Lacunar, territorial SWI Cerebral hemorrhage Location Cerebral microbleeds Location: lobar, deep Number of lesions: <5, 5–10, >10 T1 intracranial vessel wall imaging Vasculopathy Mural hematoma (pre-contrast) Location: MCA, ACA, PCA, BA, or VA Concentric versus eccentric Vessel wall abnormalities Stenosis Dissections Occlusions Enhancement Location: MCA, ACA, PCA, BA, or VA Concentric versus eccentric T1 post-contrast Meningeal contrast enhancement Location: Leptomeningeal/pachymeningeal Cranial nerve enhancement Location Coincidental findings Presence/absence MCA: middle cerebral artery; ACA: anterior cerebral artery; PCA: posterior cerebral artery; BA: basilar artery; VA: vertebral artery; DWI: diffusion weighted imaging; SWI: susceptibility weighted imaging. MCA: middle cerebral artery; ACA: anterior cerebral artery; PCA: posterior cerebral artery; BA: basilar artery; VA: vertebral artery; DWI: diffusion weighted imaging; SWI: susceptibility weighted imaging. according to the hospital’s regulations or in the affiliated biobank. color flow Doppler and first-generation contrast. The appearance of microbubbles in the left atrium within three to six cardiac beats after opacification of the right atrium is considered positive for the presence of an intracardiac shunt such as a PFO. Valsalva maneuver will be performed to pro- mote right-to-left shunting of microbubbles to identify a PFO when no shunting is present without provocation.23 Heart rhythm monitoring Patients will receive an ambulatory Holter to monitor heart rhythm for a period of 48–72 h. Holter monitoring will be performed according to standard procedures. If patients have received rhythm monitoring during admission for standard medical practice (e.g. telemetry during ICU admission ⩾48 h), these data will be used for the current study and no additional Holter monitoring will take place to reduce the burden for the patients. Statistical analysis Sample size calculation. To determine the prevalence of asymptomatic (silent) cerebral ischemia on MRI in patients with COVID-19 compared to controls, we based our sample size calculation on the currently available lit- erature. We expect an incidence of about 1%–3% of symptomatic ischemic stroke in hospitalized patients with COVID-19.2,6,8,10 Extrapolating from existing literature we expect a six- to nine-fold increased prevalence of asymptomatic (silent) cerebral ischemia (i.e. 18%–25% assuming a 3% prevalence of symptomatic events) as compared to symptomatic ischemic stroke.24 In 200 patients undergoing MRI scanning this would lead to identification of about 40 cases with asymptomatic (silent) cerebral ischemia and at least 160 controls without asymp- tomatic cerebral ischemia. The observed prevalence will be compared with that in controls. Assuming a prevalence of silent cerebral ischemia in these subjects of max 1%25 and of 20% in the patients with COVID-19, we will have 95% power to detect a significant difference at the signifi- cance level of 0.05 (95% confidence level). Blood sampling For objective 3 and 4, measuring functional outcome and cognitive function after 3 and 12 months and determin- ing the incidence of new cerebral ischemia on MRI after 3 months of follow-up in patients with COVID-19, the sam- ple size will be equal to that of the population of patients with COVID-19 in the previous sub studies. The precision of the descriptive results for this study will be determined by this study size (no statistical comparisons are made here). A loss to follow-up rate of 10% is taken into account for all the sample size calculations. For objective 3 and 4, measuring functional outcome and cognitive function after 3 and 12 months and determin- ing the incidence of new cerebral ischemia on MRI after 3 months of follow-up in patients with COVID-19, the sam- ple size will be equal to that of the population of patients with COVID-19 in the previous sub studies. The precision of the descriptive results for this study will be determined by this study size (no statistical comparisons are made here). A loss to follow-up rate of 10% is taken into account for all the sample size calculations. Analysis of primary outcomes. For objective 1 we will deter- mine the prevalence of silent cerebral ischemia among patients with COVID-19 admitted to or discharged from the hospital and in age and sex matched controls without (pre- vious) COVID-19 infection from the general population including corresponding 95% confidence intervals. Our study is the first to prospectively conduct a brain MRI in hospitalized patients with COVID-19 during the acute phase of their infection. This study will therefore pro- vide more knowledge about the possible effects of COVID- 19 on the brain and on cerebrovascular damage in this patient population. Combined with the follow-up MRI we will gain knowledge on dynamics of cerebral ischemia in patients with COVID-19 as well. This can help clinicians to understand mechanisms/causes of COVID-19 related func- tional loss. For objective 2, cases, patients with COVID-19 with cerebral ischemia (symptomatic and asymptomatic) and controls (without cerebral ischemia) will be compared with respect to the prevalence of possible risk factors using logistic regression models. Odds ratios will be estimated as measures for the relative risks associated with each possible risk factor. Each risk factor will be analyzed in univariable and multivariable analysis to correct for confounders as age, sex, and comorbidities. Blood sampling Among the strengths of this study is the multicentre design, leading to a large sample size of patients included from multiple regions throughout the Netherlands. Moreover, the prospective design with two follow-up assessments allows us to collect longitudinal and detailed standardized information, including demographics, vascu- lar risk factors, cognitive tests, and imaging measurements of the patients. Due to our limited exclusion criteria, we will be able to include a patient group with high external validity. Adding matched controls enables us to compare the prevalence of silent cerebral ischemia and other cere- brovascular lesions in both groups of patients. In a follow-up study after 3 and 12 months, we will describe functional performance and cognitive function in COVID-19 patients. We will stratify for subgroups in anal- ysis (symptomatic ischemic stroke, asymptomatic cerebral ischemia, no cerebral ischemia). After 3 months we will investigate the long-term inci- dence of asymptomatic (silent) cerebral ischemia in patients with COVID-19. The incidence rate will be determined as the number of new (or first) silent cerebral ischemia divided by the total amount of person-time. Discussion The CORONIS study is a multicentre prospective cohort study investigating the prevalence, risk factors and long- term incidence of (silent) cerebral ischemia in patients hos- pitalized with COVID-19, and to determine long-term functional outcome. Summary and conclusions In conclusion, CORONIS is a pivotal study to investigate the prevalence, long-term incidence and risk factors of silent cerebral ischemia in hospitalized COVID-19 patients, and will determine long-term functional outcome in this population. Blood sampling Fifty-four ml blood (18 ml citrated plasma, 20 ml serum, and 10–16 ml EDTA) will be sampled to assess biomarkers of inflammation and coagulation, including genetic vari- ants of these factors. The samples will be stored locally 185 van Lith et al. with COVID-19. Research on ischemic stroke as a complica- tion of COVID-19 showed a prevalence ranging from 1% to 3%.2,6,8,10 The rate of ischemic stroke as complication of COVID-19 seems to be higher than in other respiratory viruses such as influenza.7,26 Recent neuroradiologic studies described multiple brain MRI abnormalities such as (micro) hemorrhage, ischemic lesions, and signs of encephalitis in patients with COVID-19.27 Post-mortem pathology studies showed vascular damage, including hypoxic damage, ischemic lesions and (micro)hemorrhages, and inflammatory infiltrates in brain tissue.28,29 However, most of our knowl- edge on these cerebrovascular complications of COVID-19 is derived from retrospective data. Several studies have sug- gested coagulopathy and endotheliopathy both to be as pos- sible mechanisms of COVID-related ischemic stroke, however no prospective risk factor analysis has been done. To anticipate on the possible consequences of both sympto- matic and silent ischemic brain lesions, prospective studies in patients with COVID-19 investigating the effects of COVID-19 in the brain are urgently needed. Regarding objective 2, to assess causes of cerebral ischemia in patients with COVID-19, based on the expected 110 cases with symptomatic (n = 70) or asymptomatic (expected n = 40) cerebral ischemia and (expected) 160 con- trols without (a)symptomatic cerebral ischemia we have a power of 95% to demonstrate a relative risk of 3 (alpha 0.05). We will still have 80% power to identify less frequent exposures, for example for exposures with a prevalence of 5% in the controls, we can demonstrate a relative risk of 3.4. Regarding objective 2, to assess causes of cerebral ischemia in patients with COVID-19, based on the expected 110 cases with symptomatic (n = 70) or asymptomatic (expected n = 40) cerebral ischemia and (expected) 160 con- trols without (a)symptomatic cerebral ischemia we have a power of 95% to demonstrate a relative risk of 3 (alpha 0.05). We will still have 80% power to identify less frequent exposures, for example for exposures with a prevalence of 5% in the controls, we can demonstrate a relative risk of 3.4. FE. 13. de Havenon A, Ney JP, Callaghan B, et al. Impact of COVID-19 on outcomes in ischemic stroke patients in the United States. J Stroke Cerebrovasc Dis 2021; 30: 105535. Informed consent Written informed consent was obtained from all subjects before the study. 10. Sluis WM, Linschoten M, Buijs JE, et al. Risk, clinical course, and outcome of ischemic stroke in patients hospi- talized with COVID-19: a multicenter cohort study. Stroke 2021; 52: 3978–3986. ORCID iDs Theresa J van Lith https://orcid.org/0000-0003-2814-8368 Frederick JA Meijer https://orcid.org/0000-0001-5921-639X Frederikus A Klok https://orcid.org/0000-0001-9961-0754 H Bart van der Worp https://orcid.org/0000-0001-9891-2136 Theresa J van Lith https://orcid.org/0000-0003-2814-8368 Frederick JA Meijer https://orcid.org/0000-0001-5921-639X Frederikus A Klok https://orcid.org/0000-0001-9961-0754 H Bart van der Worp https://orcid.org/0000-0001-9891-2136 17. Wilson JT, Hareendran A, Hendry A, et al. Reliability of the modified Rankin Scale across multiple raters: benefits of a structured interview. Stroke 2005; 36: 777–781. Theresa J van Lith https://orcid.org/0000-0003-2814-8368 Frederick JA Meijer https://orcid.org/0000-0001-5921-639X Frederikus A Klok https://orcid.org/0000-0001-9961-0754 H B t d W htt // id /0000 0001 9891 2136 18. Klok FA, Boon GJAM, Barco S, et al. The Post- COVID-19 functional status scale: a tool to measure functional status over time after COVID-19. Eur Respir J 2020; 56: 2001494. H Bart van der Worp Declaration of conflicting interests The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Little is known about the occurrence and consequences of clinically “silent” cerebral ischemia in patients hospitalized 186 European Stroke Journal 7(2) 7(2) 7. Merkler AE, Parikh NS, Mir S, et al. Risk of ischemic stroke in patients with coronavirus disease 2019 (COVID- 19) vs patients with influenza. JAMA Neurol 2020; 77: 1–7. Funding The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work is supported by The Netherlands Organisation for Health Research and Development (ZonMw) and The Heart Council (grant number 10430012010014). 8. Nannoni S, de Groot R, Bell S, et al. Stroke in COVID-19: a systematic review and meta-analysis. Int J Stroke 2021; 16: 137–149. 9. Stals M, Grootenboers M, van Guldener C, et al. Risk of thrombotic complications in influenza versus COVID-19 hospitalized patients. Res Pract Thromb Haemost 2021; 5: 412–420. Author Contributions TL, WS, and NW wrote the first draft of this manuscript. TL, WS, and NW are involved in patient recruitment and data analysis. FEdl, MH, MW, SC, BvdW, AT, FK, and FvdV were involved in protocol development and gaining ethical approval. All authors were involved in designing aspect of the study related to his/her field or assessment of the data. All authors revised the manuscript and approved the final version before publication. 14. Lansky AJ, Messé SR, Brickman AM, et al. Proposed stand- ardized neurological endpoints for cardiovascular clinical trials: an academic research consortium initiative. J Am Coll Cardiol 2017; 69: 679–691. 15. Verhage F. Intelligentie en leeftijd: onderzoek bij nederland- ers van twaalf tot zevenenzeventig jaar. Assen: Van Gorcum, 1964. 16. Zigmond AS and Snaith RP. The hospital anxiety and depres- sion scale. Acta Psychiatr Scand 1983; 67: 361–370. Ethical approval The study is conducted according to the Declaration of Helsinki and the local ethics committee of Arnhem-Nijmegen approved this study (NL75780.091.20). 11. Ellul MA, Benjamin L, Singh B, et al. Neurological associa- tions of COVID-19. Lancet Neurol 2020; 19: 767–783. 12. Oxley TJ, Mocco J, Majidi S, et al. Large-vessel stroke as a presenting feature of Covid-19 in the young. New Engl J Med 2020; 382: e60. 24. Lopez OL, Jagust WJ, Dulberg C, et al. Risk factors for mild cognitive impairment in the Cardiovascular Health Study Cognition Study: part 2. Arch Neurol 2003; 60: 1394–1399. 26. Taquet M, Geddes JR, Husain M, et al. 6-month neurological and psychiatric outcomes in 236 379 survivors of COVID-19: a retrospective cohort study using electronic health records. Lancet Psychiatry 2021; 8: 416–427. g y p 25. Yamada K, Nagakane Y, Sasajima H, et al. Incidental acute infarcts identified on diffusion-weighted images: a university hos- pital-based study. AJNR Am J Neuroradiol 2008; 29: 937–940. References 19. Machado FVC, Meys R, Delbressine JM, et al. Construct validity of the Post-COVID-19 functional status scale in adult subjects with COVID-19. Health Qual Life Outcomes 2021; 19: 40. 1. Connors JM and Levy JH. COVID-19 and its implica- tions for thrombosis and anticoagulation. Blood 2020; 135: 2033–2040. 2. Klok FA, Kruip MJHA, van der Meer NJM, et al. Confirmation of the high cumulative incidence of thrombotic complications in critically ill ICU patients with COVID-19: an updated analysis. Thromb Res 2020; 191: 148–150. 20. Wardlaw JM, Smith EE, Biessels GJ, et al. Neuroimaging standards for research into small vessel disease and its con- tribution to ageing and neurodegeneration. Lancet Neurol 2013; 12: 822–838. 3. Kaptein FHJ, Stals MAM, Grootenboers M, et al. Incidence of thrombotic complications and overall survival in hospital- ized patients with COVID-19 in the second and first wave. Thromb Res 2021; 199: 143–148. 21. Nasreddine ZS, Phillips NA, Bédirian V, et al. The Montreal Cognitive Assessment, MoCA: a brief screening tool for mild cognitive impairment. J Am Geriatr Soc 2005; 53: 695–699. 4. Malas MB, Naazie IN, Elsayed N, et al. Thromboembolism risk of COVID-19 is high and associated with a higher risk of mortality: a systematic review and meta-analysis. EClinicalMedicine 2020; 29: 100639. 22. van den Berg E, Ruis C, Biessels GJ, et al. The telephone interview for cognitive status (Modified): relation with a comprehensive neuropsychological assessment. J Clin Exp Neuropsychol 2012; 34: 598–605. 5. Tan BK, Mainbourg S, Friggeri A, et al. Arterial and venous thromboembolism in COVID-19: a study-level meta-analy- sis. Thorax 2021; 76: 970–979. 23. Silvestry FE, Cohen MS, Armsby LB, et al. Guidelines for the echocardiographic assessment of atrial septal defect and patent foramen ovale: from the American Society of Echocardiography and society for cardiac angiography and interventions. J Am Soc Echocardiogr 2015; 28: 910–958. 6. Qureshi AI, Baskett WI, Huang W, et al. Acute ischemic stroke and COVID-19: an analysis of 27,676 patients.. Stroke 2021; 52: 905–912. 187 van Lith et al. 27. Choi Y and Lee MK. Neuroimaging findings of brain MRI and CT in patients with COVID-19: A systematic review and meta-analysis. Eur J Radiol 2020; 133: 109393. 28. Maiese A, Manetti AC, Bosetti C, et al. SARS-CoV-2 and the brain: A review of the current knowledge on neuropathol- ogy in COVID-19. Brain Pathol 2021; 31: e13013. 29. Mondello C, Roccuzzo S, Malfa O, et al. 29. Mondello C, Roccuzzo S, Malfa O, et al. Pathological find- ings in COVID-19 as a tool to define SARS-CoV-2 patho- genesis. A systematic review. Front Pharmacol 2021; 12: 614586. 27. Choi Y and Lee MK. Neuroimaging findings of brain MRI and CT in patients with COVID-19: A systematic review and meta-analysis. Eur J Radiol 2020; 133: 109393. References Pathological find- ings in COVID-19 as a tool to define SARS-CoV-2 patho- genesis. A systematic review. Front Pharmacol 2021; 12: 614586.
https://openalex.org/W3196546189	https://durham-repository.worktribe.com/file/1215135/1/Published%20Journal%20Article	English	null	Provision of social-norms feedback to general practices whose antibiotic prescribing is increasing: a national randomized controlled trial	Zeitschrift für Gesundheitswissenschaften/Journal of public health	2,021	cc-by	5,457	Abstract Aim The Chief Medical Officer of England writes an annual social-norms-feedback letter to the highest antibiotic-prescribing GP practices. We investigated whether sending a social-norms-feedback letter to practices whose prescribing was increasing would reduce prescribing. p g Subject and methods We conducted a two-armed randomised controlled trial amongst practices whose STAR-PU-adjusted prescribing was in the 20th–95th percentiles and had increased by > 4% year-on-year in the 2 previous financial years. Intervention practices received a letter on 1st March 2018 stating ‘The great majority (80%) of practices in England reduced or stabilised their antibiotic prescribing rates in 2016/17. However, your practice is in the minority that have increased their prescribing by more than 4%.’. Control practices received no letter. The primary outcome was the STAR-PU-adjusted rate of antibiotic prescribing in the months from March to September 2018. Results We randomly assigned 930 practices; ten closed or merged pre-trial, leaving 920 practices — 448 in the intervention and 472 in the control. An autoregressive and moving average model of first order ARMA(1,1) correlation structure showed no effect of the intervention (β < −0.01, z = −0.50, p = 0.565). Prescribing reduced over time in both arms (β < −0.01, z = −36.36, p < 0.001). Conclusions A social-norms-feedback letter to practices whose prescribing was increasing did not decrease prescribing compared to no letter. Trial registration NCT03582072. Trial registration NCT03582072. Keywords Antibiotics . Antimicrobial resistance . Feedback . Primary care . Social norms * Natalie Gold n.gold@lse.ac.uk Provision of social-norms feedback to general practices whose antibiotic prescribing is increasing: a national randomized controlled trial Natalie Gold1,2,3 & Michael Ratajczak1,4 & Anna Sallis1 & Ayoub Saei1 & Robin Watson1,5 & Paul van Schaik1,6 & Sarah Bowen1,7 & Tim Chadborn1 Received: 16 December 2020 /Accepted: 13 August 2021 # The Author(s) 2021 Journal of Public Health: From Theory to Practice https://doi.org/10.1007/s10389-021-01645-4 Journal of Public Health: From Theory to Practice https://doi.org/10.1007/s10389-021-01645-4 ORIGINAL ARTICLE 7 School of Economics, Sir Clive Granger Building University Park, Nottingham NG7 2RD, UK Background Statistical modelling suggests that between 8.8% and 23.1% of all antibiotic prescriptions could be classed as ‘inappropriate’, meaning that the antibiotic is likely to have zero or marginal benefit, which is likely to be outweighed by the potential risks of prescribing to the patient (Smieszek et al. 2018). prescribing frequency by more than 4%, an increase that we judged to be greater than what might be expected due to nat- ural variation alone. At the time of our trial, there was no evidence on the effec- tiveness of providing GPs with social-norm feedback telling them that their practice is among a minority whose antibiotic prescribing is increasing. To address this knowledge gap, we conducted a randomized controlled trial in March 2018, of social-norms-feedback letters delivered to practices that were outside the top 20% of prescribers, but whose prescribing had increased by more than 4% between the financial year 2015/ 16 and financial year 2016/17. The intervention letter included a message telling GPs that they were in the minority of prac- tices whose prescribing was increasing. We excluded the low- est prescribers from the trial (those whose prescribing was below the 95th percentile) because, after consulting with pub- lic health professionals, we decided it was unsuitable to send the lowest prescribers letters about overprescribing. Researchers at Public Health England (PHE) conducted a comprehensive review and analysis of the behaviours that support antibiotic stewardship, and which of the drivers of these behaviours are amenable to change (Pinder et al. 2015). Based on this analysis, they conducted a national randomised controlled trial in 2014/15, testing a social- norms-feedback letter, sent from England’s Chief Medical Officer (CMO) to 791 GP practices whose prescribing rate for antibiotics was in the top 20% in their local area (Hallsworth et al. 2016). The letter said that the practice was in the top 20% of prescribers in their area and listed three simple steps that prescribers could take to reduce inappropri- ate prescribing. The letter led to a 3.3% reduction in prescrib- ing amongst GP practices that received it, with an estimated 73,406 fewer antibiotic items dispensed. Based on this evi- dence, similar letters were sent out to all practices whose pre- scribing was in the top 20% nationally in 2015/16 and 2016/ 17 (the NHS Local Area Teams, which were used in the orig- inal letter, had been disbanded). Background The use of antibiotics is outpacing the discovery of new drugs, leading to antimicrobial resistance (AMR) as bacteria evolve to become ‘superbugs’, which are resistant to the drugs we use to treat them. In the UK alone in 2015, there were an estimated 52,971 cases of infection with antibiotic-resistant bacteria, resulting in 2172 deaths, which is similar to the burden of influenza, tuberculosis, and HIV combined (Cassini et al. 2019). A review of the problems posed by antimicrobial re- sistance commissioned by the UK Prime Minister in 2014 estimates a continued rise in resistance by 2050, leading to the deaths of up to 10 million people worldwide each year (O'Neill 2016). Increased illness also has an economic cost, and the review estimates that the cost of AMR could be $100 trillion worldwide by 2050. 1 Public Health England, Wellington House, 133-155 Waterloo Road, London SE1 8UG, UK 1 Public Health England, Wellington House, 133-155 Waterloo Road, London SE1 8UG, UK 2 Centre for Philosophy of Natural and Social Sciences, London School of Economics, Houghton Street, London WC2A 2AE, UK 2 Centre for Philosophy of Natural and Social Sciences, London School of Economics, Houghton Street, London WC2A 2AE, UK 3 Behavioural Practice, Kantar Public, 4 Millbank, Westminster, London SW1P 3JA, UK 4 Department of Mathematics and Statistics, Lancaster University, Lancaster LA1 4YF, UK 5 Department of Anthropology, Durham University, Dawson Building, South Road, Durham DH1 3LE, UK 6 Centre for Applied Psychological Science, Teeside University, Middlesbrough, Tees Valley TS1 3BX, UK 6 Centre for Applied Psychological Science, Teeside University, Middlesbrough, Tees Valley TS1 3BX, UK 7 School of Economics, Sir Clive Granger Building University Park, Nottingham NG7 2RD, UK 7 School of Economics, Sir Clive Granger Building University Park, Nottingham NG7 2RD, UK J Public Health (Berl.): From Theory to Practice A substantial driver of antibiotic resistance is the over- prescription of antibiotics in primary care (Her Majesty’s Government 2014). General practitioners (GPs) are often the first port-of-call for patients seeking medical help, and ap- proximately 80% of antibiotics are prescribed in primary care, making this a suitable target for an intervention (Her Majesty’s Government 2014). Trial design We conducted a two-armed randomised controlled trial. In February 2018, a letter and a ‘Treating your infection’ leaflet were sent to half of the practices between the 20th and 95th percentiles of practices whose antibiotic prescribing (adjusted for practice demographics) was increasing, informing them that they were in the minority of practices whose prescribing was increasing (see the Interventions section for more details, and Appendix 1 for examples of the letter and the leaflet). The control group did not receive a letter, which was standard practice. However, evidence suggests that there is overprescribing in all practices, not just those in the top 20% of prescribers (Smieszek et al. 2018). Given the success of the social- norms-feedback letter, we wanted to devise another social- norms message that we could send to prescribers outside of the top 20%. Practices are tasked with taking action to ensure that their prescribing behaviour is appropriate and to monitor judicious use of antimicrobials (NICE 2015). Antibiotic pre- scribing in primary care settings, measured in terms of antibi- otic items being prescribed, has fallen by 16.7% from 2014 to 2018 (Public Health England 2018), which suggests that this message is having an effect and many practices are decreasing their prescribing. In the period of October 2016 to September 2017 (the most recent 12-month period of data available be- fore the study’s launch), only 20% of practices increased their Background An evaluation of the 2016/17 letter using a regression discontinuity design showed that the letter continued to have an effect, causing a 3.69% reduction in prescribing, with an estimated 124,952 fewer antibiotic items dispensed (Ratajczak et al. 2019). The intervention has also been successfully implemented in Ireland and Australia (Behavioural Economics Team of the Australian Government 2018; HSC Public Health Agency 2017). The aim of this trial was to determine whether a social- norms-feedback letter reduces prescribing rates in GP prac- tices prescribing between 20th and 95th percentiles whose prescribing had increased by more than 4% in 12 months be- fore the intervention letter was sent, compared to a control group of the same characteristics who were not sent a letter. 2 https://digital.nhs.uk/services/organisation-data-service/data-downloads/gp- and-gp-practice-related-data 3 https://www.prescqipp.info/our-resources/webkits/antimicrobial- stewardship/ 1 https://fingertips.phe.org.uk/profile/amr-local-indicators/data#page/0/gid/ 1938132909/pat/46/par/E39000030/ati/19/are/E38000010 Randomisation and masking A study investigator (MR) randomly assigned GP practices to intervention or control group, grouped by sampling a hypo- thetical binary distribution of control (0) and treatment (1) groups, where the probability of treatment group membership was 0.5. This was done using R (version 3.4.3) (Team 2017). GP names and addresses were matched to practices using GPs practice codes.2 Participants in intervention groups are likely to have been aware of the interventions they were assigned to, but may have been unaware that they were involved in a trial. However, information was published on the PrescQIPP AMS Hub,3 which supports Clinical Commissioning Group antimicrobial stewardship activity, including which GP practices had re- ceived letters, so we cannot be sure they were unaware. Since the prescribing dataset had to be matched with the dataset of GP names using GP practice codes, it was not prac- tical to blind the study team to group assignment. Participants There were 930 eligible GP practices in the trial. The prescrib- ing indicator was the rate of dispensed antibiotic items per 1000 population for the practice, after applying the specific therapeutic group age-sex related prescribing units (STAR- PU) controls for age and sex. Practices were eligible if their prescribing fell between the 20th and 95th percentiles of pre- scribers for October 2016 to September 2017 (the most recent 12-month period of data available before the study’s launch) and their antibiotic prescribing had increased by more than 4% in the financial year 2016/17 compared to the previous J Public Health (Berl.): From Theory to Practice financial year, 2015/16. Practices above the 20th percentile were excluded because they were sent the standard practice social-norms letter, and practices between the 95th and 100th percentiles were excluded as outliers. The prescribing data are collected for each GP practice (individual prescribers’ data is not available) on a monthly basis by the NHS Business Services Authority. Public Health England makes these data available on Fingertips.1 We used BNF code 5.1 and we di- vided the number of items prescribed by STAR-PU, in order to get the number of items dispensed per 1000 patients in the practice population when adjusted for age and sex. previously, and as smaller effect sizes would not represent a meaningful effect. Interventions Letters from the CMO were sent to GPs at the end of February 2018, timed to land on March 1st, 2018, in an envelope that said on the outside that it was ‘From the Chief Medical Officer, Professor Dame Sally C Davies’, with a Department of Health and Social Care logo. The header stated that ‘Antibiotic prescribing rates in your practice have increased.’ Inside the text of the letter, there was a banner that informed GPs that ‘The great majority (80%) of practices in England reduced or stabilised their antibiotic prescribing rates in 2016/17. However, your practice is in the minority that have increased their prescribing by more than 4%.’ Ethical approval Ethical approval was granted by the NHS Health Research Authority, REC reference: 14/LO/1544. Like the letter sent to the top 20% in previous years, the letter listed three simple actions that the recipient could take to reduce unnecessary prescriptions of antibiotics: giving pa- tients advice on self-care, offering a delayed prescription, and talking about the issue with other prescribers in the prac- tice. All letters were accompanied by a copy of the patient- focused “Treating your infection” leaflet developed for the TARGET programme. A sample letter and the leaflet are reproduced in full in Appendix 1. Statistical analysis Monthly prescribing rates by different GP practices were con- sidered to be independent, while those from the same GP practice are potentially correlated. Therefore, we used a statis- tical modelling approach that allows the monthly prescribing from the same GP practice to be correlated, by allowing the model residuals to be correlated. An autoregressive and mov- ing average of first-order ARMA(1,1) correlation structure was adopted here for the correlated residuals. The model in- cluded a trend in addition to the intervention. The data analy- sis was done in SAS using [SAS/STAT] software, Version 9.4 of the SAS 64 BIT WIN (SAS 2016). Outcomes and sample size The primary outcome for the trial is the STAR-PU-adjusted rate of antibiotic prescribing in the 7 months following the intervention, March 2018 to September 2018. The trials were powered to detect a 2% change in antibiotic prescribing rate between arms of the trial, for each of March, April individu- ally, and May to August combined (since prescribing is lower in those months), controlling for baseline prescribing behav- iour — these were pre-registered outcomes, but by the time we analysed the trial, data was available for September 2018 as well, so we included that in our analyses; it does not affect the outcome. A 2% effect size was chosen as it was similar to the effects observed in comparable interventions carried out Discussion The social-norms-feedback letter to practices outside of the top 20% whose prescribing was increasing did not have a statistically significant effect (β < −0.01, p = 0.565). Prescribing was decreasing in both trial arms over the out- come period from March 2018 to September 2018 (β < −0.01, p < 0.001). The lack of effect of the intervention was surprising, given the evidence of effectiveness of previous social-norms-feedback letters to GPs in the UK (Hallsworth et al. 2016; Ratajczak et al. 2019) and internationally (Behavioural Economics Team of the Australian Government 2018; HSC Public Health Agency 2017). >In the model, as expected, the monthly measurements (items per STAR-PU prescribed) from the same GP practice were highly correlated, with estimated autocorrelation ρ = 0.970, SE < 0.01 and moving average parameter γ = 0.77, SE < 0.01 (for model coefficients see Table 2). The declin- ing item rates trend in Fig. 2 was strongly supported by the results from the model, which showed a statistically signif- icant negative trend in prescribing (β < −0.01, z = −36.36, p < 0.001). There are various possible explanations why this letter was not effective. Previous letters, which have been effective, targeted the highest 20% of prescribers (Behavioural Economics Team of the Australian Government 2018; HSC Public Health Agency 2017; Ratajczak et al. 2019), whereas these letters targeted practices whose prescribing was between the 20th and 95th percentiles. There may be more scope for reducing inappropriate prescribing amongst higher prescribers. There was no evidence of an effect of the intervention: the correlated adjusted estimate of the intervention effect was negative but it was not statistically significant (β < −0.01, z = −0.50, p = 0.565) — see Table 2. Model diagnostics (not reported here) did not indicate any serious model failure. Harms and unintended effects March 1st, 2018 and prescribing data from March to September 2018 inclusive was included in the analyses. By March 2018 there were 920 GP practices with pre- scribing data (data ceases when a practice merges or closes), leaving 448 (48.17%) practices who received the intervention and 472 (50.75%) in the control. See the trial flowchart in Fig. 1. A further 22 GP practices had their last month of prescribing between March and September 2018 and therefore had some missing data for the outcome measure. These practices were included in the analysis. The number of practices included in the analysis each month is shown in Table 1, along with the 7 months of prescribing data from March 2018 to September 2018 inclusive. The monthly mean for each arm is graphed in Fig. 2. The CMO’s address was included in the letter, and she re- ceived correspondence about the letter, but no harms or unin- tended effects were reported. Results There were 930 practices randomly assigned in the trial. Feedback letters were sent to 451 (48.49%) practices and 479 (51.51%) were not sent a letter. The letter landed on J Public Health (Berl.): From Theory to Practice Table 1 Antibiotic prescribing rates per STAR-PU for the overall sample, intervention, and control groups from March until September 2018 inclusive Discussion In previous years, the letter header reinforced the social-norms message, saying ‘Your practice is amongst the 20% highest prescribers of antibiotics’ (Hallsworth et al. 2016; Ratajczak et al. 2019). Antibiotic prescribing rates in your practice have increased, but most other practices reduced or stabilised their prescribing’ or ‘Compared to other prac- tices, your prescribing rates have increased’. The other hypothesised active ingredient of these letters is the messenger — the CMO at the time Dame Professor Sally Davies. The fact that the present letter was not effective sug- gests it cannot solely be the messenger effect which has made previous letters effective. Although social-norms interventions have been widely found to be effective (Sheeran et al. 2016), this is not the first social- norms study to have returned a null result. Social-norms mes- sages have failed to work in other healthcare interventions, in- cluding immunization (Leight and Safran 2019) and alcohol mis- use (Foxcroft et al. 2015; Moreira et al. 2009), and there have been other null results in areas where they have previously been used successfully, such as energy usage (Harries et al. 2013). But it is hard to know why some interventions have been unsuccess- ful. Social-norms interventions are heterogeneous, and it may be that some types of social-norm interventions are not effective. Simply adding a photograph to a social-norms message aimed at increasing organ donation registration proved to be worse than an identical message with no image or an identical message with a logo, and less effective than the control arm using no persuasive message at all (Sallis et al. 2018). Message content and mode of delivery may be important (Dempsey et al. 2018). In this study, the mode of delivery and a very similar message have both been successful in the past (Behavioural Economics Team of the Australian Government 2018; Hallsworth et al. 2016; HSC Public Health Agency 2017; Ratajczak et al. 2019). Future re- search might investigate the effectiveness of different types of content of social-norms messages. We found that prescribing was decreasing over the trial period, with a negative trend across both arms. Possibly the letter has ceased to be effective as prescribing has come down. However, evidence shows that there is still plenty of scope for practices to reduce prescribing (Smieszek et al. 2018). Discussion 2 Monthly trend of prescribing mean over 8 months J Public Health (Berl.): From Theory to Practice Table 2 Parameter estimates (Est) and their associated standard error (SE), z-value and p value Parameter Est SE Z-value P value Intercept 0.07 < 0.01 117.52 < 0.001 Received-letter < −0.01 < 0.01 −0.50 0.565 Trend < −0.01 < 0.01 −36.36 < 0.001 Table 2 Parameter estimates (Est) and their associated standard error (SE), z-value and p value as the intervention letter. An evaluation of its effectiveness, for instance using a regression discontinuity design, could help discriminate between some of the explanations for the lack of effectiveness of the letter in the present study. For instance, if the top 20% letter was effective, then that is consistent with the explanations that the message that prescribing was ‘in- creasing’ caused the lack of effect or that targeting prescribers outside of the top 20% is less effective. However, it is incon- sistent with the explanation that the national campaigns have already led GPs to implement all the actions listed in our letters, as the same actions were included in the standard letter to the highest 20% of prescribers. Alternatively, if the top 20% letter was not effective, then that would support the explana- tions that reductions in prescribing have left less room for improvement, or that the actions listed are already widely implemented, and it would not support explanations that refer to differences in message content. practices whose prescribing was increasing. The active ingre- dient of the message was intended to be the social norms part, about being in a minority: social norms have been effective in a variety of areas (Sheeran et al. 2016), including tax evasion (Hallsworth et al. 2017), energy usage (Ayres et al. 2013), and charitable giving (Agerström et al. 2016), and there is nothing in theory to suggest that the exact content of the message or what group they were the minority of would have an effect (Bicchieri 2016; Cialdini 2007). Possibly the social-norms message was less salient than in our previous letters, since although the red font in the body of the letter told GPs that their practice was in a minority, the large font header at the top of the letter said ‘Antibiotic prescribing rates in your practice have increased’. Discussion There were many other national interventions happening at the same time, such as the Quality Premium, Commissioning for Quality and Innovation (CQUINs) quality improvement goals, national guidelines launched by the National Institute for Health and Care Excellence (NICE), and the extension of the TARGET (Treat Antibiotics Responsibly, Guidance, Education, Tools) toolkit resources via a number of work- shops (Public Health England 2018). Many of these interven- tions were also happening in previous years, when the letter to the highest prescribers was shown to be effective (Hallsworth et al. 2016; Ratajczak et al. 2019). However, our letter listed three actions that GPs could take to reduce inappropriate pre- scribing and, if these actions were already widely known and implemented, maybe as a result of other campaigns, then the letter would simply have reinforced actions to reduce prescrib- ing that were already happening. Discussion Table 1 Antibiotic prescribing rates per STAR-PU for the overall sample, intervention, and control groups from March until September 2018 inclusive Month Control Feedback Intervention Overall Mean (SD) N Mean (SD) N Mean (SD) N March 0.08 (0.02) 472 0.08 (0.02) 448 0.08 (0.02) 920 April 0.06 (0.01) 469 0.07 (0.02) 446 0.07 (0.02) 915 May 0.06 (0.01) 468 0.07 (0.02) 444 0.06 (0.01) 912 June 0.06 (0.02) 468 0.06 (0.01) 444 0.06 (0.01) 912 July 0.06 (0.02) 465 0.06 (0.01) 443 0.06 (0.02) 908 August 0.06 (0.02) 461 0.06 (0.01) 441 0.06 (0.02) 902 September 0.06 (0.01) 461 0.06 (0.01) 439 0.06 (0.01) 900 Average over the outcome period 0.06 (0.01) 3264 0.06 (0.02) 3105 0.06 (0.02) 6369 J Public Health (Berl.): From Theory to Practice Fig. 1 Trial profile Fig. 1 Trial profile Fig. 1 Trial profile However, statistical modelling shows that all practices have scope for decreasing inappropriate prescribing (Smieszek et al. 2018). Nevertheless, the GPs who received the letters in this trial may have known that they were not in the highest prescribing practices and therefore not have been very con- cerned about their antibiotic prescribing rates. In a discussion at a workshop about giving prescribing feedback to primary However, statistical modelling shows that all practices have scope for decreasing inappropriate prescribing (Smieszek et al. 2018). Nevertheless, the GPs who received the letters in this trial may have known that they were not in the highest prescribing practices and therefore not have been very con- cerned about their antibiotic prescribing rates. In a discussion at a workshop about giving prescribing feedback to primary care, participants said that GPs care about being outliers on prescribing indicators (i.e. very high or very low) and are more comfortable being in the middle (Steels et al. 2021). Because the target group was different, the message was different from previous letters: instead of telling practices that they were in a minority of practices whose overall prescribing was high, we told them that they were in a minority of Fig. References Agerström J, Carlsson R, Nicklasson L, Guntell L (2016) Using descrip- tive social norms to increase charitable giving: the power of local norms. J Econ Psychol 52:147–153. https://doi.org/10.1016/j.joep. 2015.12.007 Author contributions NG jointly formulated the trial hypothesis, contrib- uted to the development of the trial materials, managed the implementa- tion and analysis of the trial, contributed to the interpretation of the re- sults, and wrote the first draft of the paper. Ayres I, Raseman S, Shih A (2013) Evidence from two large field exper- iments that peer comparison feedback can reduce residential energy usage. J Law Econ Org 29(5):992–1022. https://doi.org/10.1093/ jleo/ews020 MR jointly formulated the trial hypothesis, developed the trial mate- rials, downloaded the data on which the trial implementation was based, conducted the power analysis, conducted the randomization, and imple- mented the trial. Behavioural Economics Team of the Australian Government (2018) Nudge vs superbugs: a behavioural economics trial to reduce the overprescribing of antibiotics. Australian Government, Australia ASallis jointly formulated the trial hypothesis, oversaw the develop- ment of the trial materials, and contributed to the interpretation of the results. Bicchieri C (2016) Norms in the wild: how to diagnose, measure, and change social norms. Oxford University Press, Oxford ASaei oversaw the data analysis and provided the final model. RW downloaded the data for the results analysis and did preliminary data analysis. Cassini A, Högberg LD, Plachouras D, Quattrocchi A, Hoxha A, Simonsen GS et al (2019) Attributable deaths and disability- adjusted life-years caused by infections with antibiotic-resistant bac- teria in the EU and the European Economic Area in 2015: a population-level modelling analysis. Lancet 19(1):56–66. https:// doi.org/10.1016/S1473-3099(18)30605-4 y PvS did preliminary data analysis. SB created the trial profile and did preliminary data analysis. SB created the trial profile and did preliminary data analysis. TC jointly formulated the trial hypothesis and was the Senior Responsible Officer for the project. All authors read and approved the final manuscript. All authors read and approved the final manuscript. Cialdini RB (2007) Descriptive social norms as underappreciated sources of social control. Psychometrika 72(2):263. https://doi.org/10.1007/ s11336-006-1560-6 Funding Public Health England; As part of its standard responsibilities, employees of the funder of the study had a role in the study design, data interpretation, and writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit to publication. Conclusion A social-norms-feedback letter to practices whose prescribing was increasing did not decrease prescribing compared to con- trol practices who did not receive a letter. Prescribing de- creased in both trial arms over the outcome period. There are As well as running the trial, we sent the standard letter to the highest 20% of prescribers, which landed on the same date J Public Health (Berl.): From Theory to Practice various possible explanations why this letter was not effective in this instance. Further research could investigate them. adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. List of abbreviations AMR, Antimicrobial resistance; BNF, British National Formulary; CMO, Chief Medical Officer; GP, General practi- tioner; PHE, Public Health England; RCT, Randomized controlled trial; STAR-PU, Specific therapeutic group age-sex related prescribing units Supplementary Information The online version contains supplementary material available at https://doi.org/10.1007/s10389-021-01645-4. Supplementary Information The online version contains supplementary material available at https://doi.org/10.1007/s10389-021-01645-4. Acknowledgments We would like to thank Professor Dame Sally Davies GCB DBE FRS FMedSci, UK Special Envoy on AMR. Acknowledgments We would like to thank Professor Dame Sally Davies GCB DBE FRS FMedSci, UK Special Envoy on AMR. References Dempsey RC, McAlaney J, Bewick BM (2018) A critical appraisal of the Social Norms Approach as an interventional strategy for health- related behaviour and attitude change. Front Psychol 9:2180. https://doi.org/10.3389/fpsyg.2018.02180 Foxcroft DR, Moreira MT, Almeida Santimano NML, Smith LA (2015) Social norms information for alcohol misuse in university and col- lege students. Cochrane Database Syst Rev 12:CD006748. https:// doi.org/10.1002/14651858.CD006748.pub4 Availability of data and materials The datasets generated and analysed during the current study are available on Fingertips, https://fingertips.phe. org.uk/profile/amr-local-indicators/data#page/0/gid/1938132909/pat/46/ par/E39000030/ati/19/are/E38000010 Hallsworth M, Chadborn T, Sallis A, Sanders M, Berry D, Greaves F et al (2016) Provision of social norm feedback to high prescribers of antibiotics in general practice: a pragmatic national randomised con- trolled trial. Lancet 387(10029):1743–1752. https://doi.org/10. 1016/s0140-6736(16)00215-4 Declarations Ethical approval was granted by the NHS Health Research Authority, REC reference: 14/LO/1544. Competing interests The authors declare that they have no competing interests. Hallsworth M, List JA, Metcalfe RD, Vlaev I (2017) The behavioralist as tax collector: using natural field experiments to enhance tax compli- ance. J Public Econ 148:14–31. https://doi.org/10.1016/j.jpubeco. 2017.02.003 Ethical approval and consent to participate Ethical approval was given by the Health Research Authority, London — Harrow Research Ethics Committee, REC Reference 14/LO/1544; this included a waiver on GPs or GP practices having to consent to participate since that would have undermined the trial results, particularly from the control group, and these effects would be disproportionate to the intervention. Harries T, Rettie R, Studley M, Burchell K, Chambers S (2013) Is social norms marketing effective? A case study in domestic electricity consumption. Eur J Mark 47(9):1458–1475. https://doi.org/10. 1108/EJM-10-2011-0568 Her Majesty’s Government (2014) UK five-year antimicrobial resistance (AMR) strategy 2013–2018: annual progress report and implemen- tation plan. Retrieved from https://www.gov.uk/government/ uploads/system/uploads/attachment_data/file/385733/UK_AMR_ annual_report.pdf Consent for publication Not applicable. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, HSC Public Health Agency (2017) Surveillance of antimicrobial use and resistance in Northern Ireland, Annual Report, 2017. Retrieved from HSC Public Health Agency (2017) Surveillance of antimicrobial use and resistance in Northern Ireland, Annual Report, 2017. Retrieved from J Public Health (Berl.): From Theory to Practice design. J Antimicrob Chemother 74(12):3603–3610. https://doi.org/ 10.1093/jac/dkz392 http://www.publichealth.hscni.net/sites/default/files/AMR_annual_ report_final_0.pdf Leight J, Safran E (2019) Increasing immunization compliance among schools and day care centers: evidence from a randomized con- trolled trial. J Behav Public Adm 2(2):1–15. https://doi.org/10. References 30636/jbpa.22.55 Sallis A, Harper H, Sanders M (2018) Effect of persuasive messages on National Health Service Organ Donor Registrations: a pragmatic quasi-randomised controlled trial with one million UK road tax- payers. J Trials 19(1):513. https://doi.org/10.1186/s13063-018- 2855-5 Moreira MT, Smith LA, Foxcroft D (2009) Social norms interventions to reduce alcohol misuse in university or college students. Cochrane Database Syst Rev 3:CD006748. https://doi.org/10.1002/14651858. CD006748.pub2 SAS (2016) SAS/STAT, Version 9.4. SAS Institute Inc, Cary, IN Sheeran P, Maki A, Montanaro E, Avishai-Yitshak A, Bryan A, Klein WMP, Miles E, Rothman AJ (2016) The impact of changing atti- tudes, norms, and self-efficacy on health-related intentions and be- havior: a meta-analysis. Health Psychol 35(11):1178. https://doi. org/10.1037/hea0000387 NICE (2015) Antimicrobial stewardship: systems and processes for ef- fective antimicrobial medicine use. In: NICE guideline [NG15]. National Institute for Health and Care Excellence, London O'Neill J (2016) Review on antimicrobial resistance tackling drug- resistant infections globally. Retrieved from https://amr-review. org/sites/default/files/160518_Final%20paper_with%20cover.pdf. Accessed 22 Dec 2021 Smieszek T, Pouwels KB, Dolk FCK, Smith DR, Hopkins S, Sharland M et al (2018) Potential for reducing inappropriate antibiotic prescrib- ing in English primary care. J Antimicrob Chemother 73(suppl_2): ii36–ii43. https://doi.org/10.1093/jac/dkx500 Pinder R, Berry D, Sallis A, Chadborn T (2015) Behaviour change and antibiotic prescribing in healthcare settings Literature review and behavioural analysis. Public Health England, London https://doi. org/10.25561/22194 Steels S, Gold N, Palin V, Chadborn T, van Staa TP (2021) Improving our understanding and practice of antibiotic prescribing: a study on the use of social norms feedback letters in primary care. Int J Environ Res Public Health 18(5):2602. https://doi.org/10.3390/ ijerph18052602 Public Health England (2018) English Surveillance Programme for Antimicrobial Utilisation and Resistance (ESPAUR). Retrieved from https://assets.publishing.service.gov.uk/government/uploads/ system/uploads/attachment_data/file/759975/ESPAUR_2018_ report.pdf Team RC (2017) R: a language and environment for statistical comput- ing. version 3.4.3. RC Team, Vienna Publisher’s note Springer Nature remains neutral with regard to jurisdic- tional claims in published maps and institutional affiliations. Publisher’s note Springer Nature remains neutral with regard to jurisdic- tional claims in published maps and institutional affiliations. Ratajczak M, Gold N, Hailstone S, Chadborn T (2019) The effectiveness of repeating a social norm feedback intervention to high prescribers of antibiotics in general practice: a national regression discontinuity
https://openalex.org/W850992531	https://www.nature.com/articles/srep11733.pdf	English	null	Validation of a criteria-specific long-term survival prediction model for hepatocellular carcinoma patients after liver transplantation	Scientific reports	2,015	cc-by	5,596	Validation of a criteria-specific long-term survival prediction model for hepatocellular carcinoma patients after liver transplantation received: 24 December 2014 accepted: 03 June 2015 Published: 22 June 2015 Fei Teng, Qiu-Cheng Han, Guo-Shan Ding, Zhi-Jia Ni, Hong Fu, Wen-Yuan Guo, Xiao-Min Shi, Xiao-Gang Gao, Jun Ma & Zhi-Ren Fu The aim of this study was to validate a criteria-specific long-term survival prediction model (MHCAT) in a large cohort of hepatocellular carcinoma (HCC) patients after liver transplantation (LT) in China. Independent risk factors in MHCAT were retrospectively analysed for HCC patients recorded in the China Liver Transplant Registry. Survival predictions for each patient were calculated using MHCAT scores and the Metroticket formula separately, and the prediction efficacy of MHCAT and Metroticket was compared using the area under ROC curve (c-statistic). A total of 1371 LTs for HCC were analysed in the study, with a median follow-up of 22.2 months (IQR 6.1–72.4 months). The proportions meeting the Milan, UCSF, Fudan and Hangzhou criteria were 34.4%, 39.7%, 44.2% and 51.9%, respectively. The c-statistics for MHCAT predictions of 3- and 5-year survival rates of HCC recipients were 0.712–0.727 and 0.726–0.741, respectively. Among these patients, 1298 LTs for HCC were ultimately selected for the comparison analysis for prediction efficacy. The c-statistic of MHCAT for predictions of 3-year survival with reference to the Milan, UCSF and Fudan criteria was significantly increased compared with that for Metroticket (p < 0.05). In conclusion, MHCAT can effectively predict long-term survival for HCC recipients after LT. Hepatocellular carcinoma (HCC) is the fifth most frequent and the third most deadly cancer in the world1. Over the past 15 years, the occurrence of HCC has more than doubled. Each year 500,000 new cases and 360,000 deaths are reported in the Asia-Pacific region, over 60% of which occur in China2. Since liver transplantation (LT) was first utilized, it was widely recognized as an effective treatment for HCC given that it could cure both the tumour and underlying liver diseases. However, access to trans- plantation is a balance between maximizing a patient’s chances of cure and overall survival due to the scarcity of liver donations. Criteria for LT and organ allocation systems have led to various controversies over LT for HCC patients in terms of whether and to what extent the criteria should be less restrictive. www.nature.com/scientificreports www.nature.com/scientificreports www.nature.com/scientificreports received: 24 December 2014 accepted: 03 June 2015 Published: 22 June 2015 Results P i C Patient Characteristics. All LTs were ABO-type compatible. Among the 1371 patients, 1244 (90.8%) were male, and 126 (9.2%) were female; the mean age was 49.8 ± 9.1 years. A total of 1309 (95.5%) patients received a LT from a deceased donor, whereas 62 (4.5%) patients received a LT from a living donor. The median follow-up time was 22.2 months (IQR, 6.1–72.4 months). The 1-, 3-, and 5-year cumulative survival rates were 63.1%, 43.8%, and 37.6%, respectively, with a total number of 748 deaths from any cause, and tumour recurrence was noted in 247 patients in total. The 1-, 3- and 5-year HCC recurrence-free survival rates after LT were 55.8%, 40.4% and 35.7%, respectively. Table 1 presents patient characteristics regarding MHCAT. Compared with the Milan criteria, the UCSF, Fudan and Hangzhou criteria expanded the population meeting the criteria by 15.53%, 28.51% and 51.28%, respectively. Population expansions among different criteria are presented in Supplement 1, 2, and 3. Validation of MHCAT in patients with HCC after LT in China. Each of the 1371 patients was evaluated using MHCAT with reference to the Milan, UCSF, Fudan, and Hangzhou criteria, and the individual 3- and 5-year predicted survival rates were calculated using MHCAT scores. By comparing the predicted rates with the actual rates, ROC curves were generated for the 3- and 5-year survival predic- tions of MHCAT system (Supplement 4 and Fig. 1). The areas under the ROC curves were 0.712–0.727 for 3-year survival predictions and 0.726–0.741 for 5-year survival predictions. The high area under the ROC values, of which no significant differences were noted using the Kruskal-Wallis tests (p > 0.05), indicated that MHCAT exhibited a good performance in predicting the long-term post-transplant sur- vival of HCC patients.hit The primary endpoint of our study on MHCAT was the survival status at five years after liver trans- plantation. In addition, the area under the ROC curve for 5-year survival was considerably increased compared with the 3-year survival curve. Therefore, the best MHCAT score cut-off values were obtained from 5-year survival ROC curves. The best MHCAT score cut-off values to stratify low and high risk for post-transplant mortality were 2.085, 1.689, 1.479, and 1.331 for the Milan, UCSF, Fudan, and Hangzhou criteria, respectively. Regardless of the criteria adopted, Kaplan–Meier analysis revealed sig- nificantly increased overall survival and recurrence-free survival in low-risk patients compared with high-risk patients (Fig. 2 and Supplement 5, p < 0.001). www.nature.com/scientificreports/ Carefully considering peri-operative risk factors, we developed a criteria-specific long-term survival prediction model for HCC patients after LT (MHCAT) based on the Milan UCSF Fudan and Hangzhou criteria (similar to the UCSF criteria, extending the diameter to 9.0 cm for a solitary nodule, 5.0 cm for the largest nodule, and 9.0 cm as the total diameter when multiple nodules are present) and Hangzhou criteria (a total tumour diameter 8 cm or less or a total tumour diameter of greater than 8 cm with an Edmondson grade I or II and pre-operative alpha-fetoprotein level of 400 ng/mL or less simultaneously). Carefully considering peri-operative risk factors, we developed a criteria-specific long-term survival prediction model for HCC patients after LT (MHCAT) based on the Milan, UCSF, Fudan and Hangzhou criteria8. Our objectives in this study were to compare the efficacy of survival prediction between the MHCAT and Metroticket systems and to validate MHCAT in LT for Chinese HCC patients using the large cohort of HCC cases with adequate follow-up recorded in the China Liver Transplant Registry (CLTR). g p p p p g y Carefully considering peri-operative risk factors, we developed a criteria-specific long-term survival prediction model for HCC patients after LT (MHCAT) based on the Milan, UCSF, Fudan and Hangzhou criteria8. Our objectives in this study were to compare the efficacy of survival prediction between the MHCAT and Metroticket systems and to validate MHCAT in LT for Chinese HCC patients using the large cohort of HCC cases with adequate follow-up recorded in the China Liver Transplant Registry (CLTR). Validation of a criteria-specific long-term survival prediction model for hepatocellular carcinoma patients after liver transplantation A number of extended criteria were proposed based on the Milan criteria (a solitary HCC nodule 5.0 cm or less in diameter or no more than three tumour nodules with the largest lesion 3.0 cm or less in diam- eter without tumour invasion of blood vessels or lymph nodes), such as the University of California San Francisco (UCSF) criteria (similar to the Milan criteria, extending the diameter to 6.5 cm for a solitary nodule, 4.5 cm for the largest nodule, and 8.0 cm as the total diameter when multiple nodules are pres- ent) and Up-to-Seven criteria3–7. Transplant scientists in China have also established the Shanghai Fudan Organ Transplantation Institute of Changzheng Hospital, Second Military Medical University, Shanghai 200003, China. *These authors contributed equally to this work. Correspondence and requests for materials should be addressed to Z.F. (email: zhirenf@vip.sina.com) 1 Scientific Reports \| 5:11733 \| DOI: 10.1038/srep11733 www.nature.com/scientificreports/ www.nature.com/scientificreports/ Variables Value Meeting Milan Criteria 472 (34.4%) Meeting UCSF Criteria 544 (39.7%) Meeting Shanghai Fudan Criteria 606 (44.2%) Meeting Hangzhou Criteria 712 (51.9%) Re-transplantation 49 (3.6%) AFP (ng/ml) mean ± SD 12828.0 ± 249523.4 median (q1, q3) 148.1 (15.0, 1235.0) TB (ng/ml) mean ± SD 60.6 ± 89.1 median (q1, q3) 30.9 (18.3, 58.9) Intraoperative blood loss (IU) mean ± SD 12.4 ± 15.7 median (q1, q3) 7.5 (4.0, 15.0) Table 1. Characteristics of hepatocellular carcinoma patients based on the MHCAT. TB, total bilirubin; AFP, alpha-fetoprotein; UCSF, University of California San Francisco. Table 1. Characteristics of hepatocellular carcinoma patients based on the MHCAT. TB, total bilirubin; AFP, alpha-fetoprotein; UCSF, University of California San Francisco. Table 1. Characteristics of hepatocellular carcinoma patients based on the MHCAT. TB, total bilirubin; AFP, alpha-fetoprotein; UCSF, University of California San Francisco. criteria (similar to the UCSF criteria, extending the diameter to 9.0 cm for a solitary nodule, 5.0 cm for the largest nodule, and 9.0 cm as the total diameter when multiple nodules are present) and Hangzhou criteria (a total tumour diameter 8 cm or less or a total tumour diameter of greater than 8 cm with an Edmondson grade I or II and pre-operative alpha-fetoprotein level of 400 ng/mL or less simultaneously).i criteria (similar to the UCSF criteria, extending the diameter to 9.0 cm for a solitary nodule, 5.0 cm for the largest nodule, and 9.0 cm as the total diameter when multiple nodules are present) and Hangzhou criteria (a total tumour diameter 8 cm or less or a total tumour diameter of greater than 8 cm with an Edmondson grade I or II and pre-operative alpha-fetoprotein level of 400 ng/mL or less simultaneously).i criteria (similar to the UCSF criteria, extending the diameter to 9.0 cm for a solitary nodule, 5.0 cm for the largest nodule, and 9.0 cm as the total diameter when multiple nodules are present) and Hangzhou criteria (a total tumour diameter 8 cm or less or a total tumour diameter of greater than 8 cm with an Edmondson grade I or II and pre-operative alpha-fetoprotein level of 400 ng/mL or less simultaneously). Results P i C Supplement 6 and 7 present the overall survival and recurrence-free survival for 1, 3, and 5 years in different subgroups. MHCAT and Metroticket comparisons. According to the Metroticket formula, a predicted 3-year and 5-year survival of each HCC recipient can be calculated based on the known number of tumours, size of the largest tumour and vascular invasion. Among the 1371 patients, 73 patients had missing Scientific Reports \| 5:11733 \| DOI: 10.1038/srep11733 2 www.nature.com/scientificreports/ Figure 1. ROC curve for MHCAT predicting 5-year survival after LT. Figure 1. ROC curve for MHCAT predicting 5-year survival after LT. or abnormal data on the number of tumours or the largest tumour size. Therefore, 1298 patients were ultimately selected for the MHCAT and Metroticket comparative analysis. ROC curves were generated for the three- and five-year survival predictions of MHCAT and the Metroticket system (Figs 3 and 4). The c-statistics of the Metroticket system for 3- and 5-year survival predictions were 0.699 and 0.738, respectively. Kruskal-Wallis analysis showed that the c-statistic of MHCAT based on the Milan, UCSF and Fudan criteria was significantly increased compared with the Metroticket 3-year survival predic- tion (p < 0.05), whereas no differences in 5-year survival predictions were noted between Metroticket and MHCAT, regardless of any one of the four criteria (p > 0.05). We measured the number of true 3-year and 5-year survival predictions using different MHCAT with reference to the four criteria as well as Metroticket system. Compared with the Metroticket system, the accuracy of predicting patients the number of patients dead at 3 or 5 years using MHCAT was considerably higher, whereas the accuracy of predicting the number of living patients was lower. Furthermore, the overall accuracy of predicting survival status using MHCAT was substantially increased (Tables 2 and 3). Scientific Reports \| 5:11733 \| DOI: 10.1038/srep11733 Discussion With over 26,000 LT cases, China ranks second only to the USA. However, the five-year survival after LT in China is significantly reduced compared with the USA and Europe (60.5% vs. 73.7% and 73.0%, respectively)9–11. However, the survival gap of patients with benign end-stage liver diseases is small: 73.2% in China, 74.1% in America and 73.2% in Europe. A low curative effect is mainly responsible for the low survival rate of HCC patients in China compared with those in the USA and Europe (49.7% vs. 67.5% and 64.0%, respectively). HCC accounts for 16% and 20.9% of all LT cases in Europe and the USA, respectively9,11. In China, the rate is as high as 40%10. Moreover, 65.6% of HCC patients in China exceeded the Milan criteria before transplantation, which inevitably affected their long-term survival. The MHCAT was built with reference to the four most representative HCC LT criteria using accurate HCC patient data from a single centre in China with a follow-up of at least six years. By validating its prediction efficacy in a large cohort of HCC patients in China, this model may help clinicians determine what type of HCC patient should receive LT as well as determine individual therapies administered to patients. p HCC patients meeting the Milan criteria have achieved satisfactory 5-year post-transplant survival over the past two decades of liver transplantation, which is similar to that observed for benign liver diseases. A number of other extended criteria were proposed to explore the ceiling of tumour load with acceptable long-term survival. However, controversies arose concerning the criteria adopted as prior- itization tools for organ allocation or transplant programs. To some HCC patients, the yes-no question might indicate the chance to receive a liver transplantation and survive. In this sense, MHCAT might be a more rational and equitable quantification tool. i Increasing evidence demonstrates that additional risks, such as AFP, microvascular invasion, or response to TACE, independently affect post-transplant survival or HCC recurrence in addition to conventionally accepted factors, such as tumour size, number and major vascular invasion12–14. Among the independent predictors of MHCAT, the criteria for LT primarily represent tumour morphological Scientific Reports \| 5:11733 \| DOI: 10.1038/srep11733 3 www.nature.com/scientificreports/ Figure 2. Overall survival analysis based on cut-off values for the MHCAT scoring system (p < 0.001 for all). (A) Milan criteria; (B) University of California San Francisco criteria; (C) Shanghai Fudan criteria; (D) Hangzhou criteria. Discussion n cut off values for the MHCAT scoring system (p<0 001 for 4 ific Reports \| 5:11733 \| DOI: 10.1038/srep11733 Figure 2. Overall survival analysis based on cut-off values for the MHCAT scoring system (p < 0.001 for all). (A) Milan criteria; (B) University of California San Francisco criteria; (C) Shanghai Fudan criteria; (D) Hangzhou criteria. Figure 3. Comparison between MHCAT and Metroticket 3-year survival predictions. Figure 2. Overall survival analysis based on cut-off values for the MHCAT scoring system (p < 0.001 for all). (A) Milan criteria; (B) University of California San Francisco criteria; (C) Shanghai Fudan criteria; (D) Hangzhou criteria. Figure 3. Comparison between MHCAT and Metroticket 3-year survival predictions. Figure 3. Comparison between MHCAT and Metroticket 3-year survival predictions. Scientific Reports \| 5:11733 \| DOI: 10.1038/srep11733 Scientific Reports \| 5:11733 \| DOI: 10.1038/srep11733 www.nature.com/scientificreports/ Figure 4. Comparison between MHCAT and Metroticket 5-year survival predictions. Figure 4. Comparison between MHCAT and Metroticket 5-year survival predictions. Survival status at 3 years Alive (n) Dead (n) Accuracy (%) Actual 582 716 – Predicted MHCAT (Milan) 363 416 60.0% MHCAT (UCSF) 369 415 60.4% MHCAT (Fudan) 359 423 60.2% MHCAT (Hangzhou) 344 435 60.0% Metroticket 414 255 51.5% Table 2. True 3-year survival predictions using different MHCAT systems based on the four criteria and the Metroticket system. Survival status at 3 years Alive (n) Dead (n) Accuracy (%) Actual 582 716 – Predicted MHCAT (Milan) 363 416 60.0% MHCAT (UCSF) 369 415 60.4% MHCAT (Fudan) 359 423 60.2% MHCAT (Hangzhou) 344 435 60.0% Metroticket 414 255 51.5% Table 2. True 3-year survival predictions using different MHCAT systems based on the four criteria and the Metroticket system. Table 2. True 3-year survival predictions using different MHCAT systems based on the four criteria and the Metroticket system. Table 2. True 3-year survival predictions using different MHCAT systems based on the four criteria and the Metroticket system. Survival status at 5 years Alive (n) Dead (n) Accuracy (%) Actual 502 796 – Predicted MHCAT (Milan) 363 416 61.7% MHCAT (UCSF) 369 415 61.9% MHCAT (Fudan) 359 423 61.9% MHCAT (Hangzhou) 344 435 61.8% Metroticket 414 255 53.8% Table 3. True 5-year survival status predictions using different MHCAT systems with reference to the four criteria and the Metroticket system. Table 3. True 5-year survival status predictions using different MHCAT systems with reference to the four criteria and the Metroticket system. Table 3. Discussion True 5-year survival status predictions using different MHCAT systems with reference to the four criteria and the Metroticket system. Table 3. True 5-year survival status predictions using different MHCAT systems with reference to the four criteria and the Metroticket system. characteristics, whereas AFP reflects the tumour biological activity at some level. In this post-transplant survival prediction model, we included intraoperative blood loss for the first time because circulating tumour cells likely plat a significant role in HCC recurrence. Based on all of these risk factors, MHCAT demonstrated increased efficacy compared with Metroticket for 3-year survival predictions and a com- parable 5-year survival prediction efficacy. fi Although other criteria have demonstrated satisfactory performances, the Milan criteria, which was integrated into the Union for International Cancer Control (UICC) TMN and United Network for Organ Sharing (UNOS) staging systems for HCC, continues to play a leading role among all types of criteria. Patients meeting the Milan criteria receive extra MELD scores in the current organ allocation system; Scientific Reports \| 5:11733 \| DOI: 10.1038/srep11733 5 www.nature.com/scientificreports/ thus, it is inappropriate to replace the Milan criteria with other criteria. However, we can provide HCC patients a MHCAT score on the basis of the Milan criteria in combination with other predictors. Thus, the coefficients for the Milan criteria and other predictors were established, which helped us to optimize the allocation system using a modified extra MELD score.fi y gi In addition to the increased prediction efficacy, the most important advantage for MHCAT is the capacity to process the unavoidable discrepancies between pre-transplant imaging interpretation and ex-plant pathology examination. When the decision for LT is made for an HCC patient, predictions of size and the number of nodules are based upon imaging. Pre-transplant staging fails to predict the num- ber of hepatocellular tumours at pathology in approximately 25 to 35% of patients15–17. Approximately 28% of patients within the Milan criteria were underestimated by pre-transplant imaging15. Our study indicates that the use of the UCSF, Fudan, and Hangzhou criteria resulted in an obvious expansion of the number of eligible patients by 15.53%, 11.65%, and 19.66%, respectively, compared with the Milan criteria, whereas the efficacy of MHCAT with reference to the four different criteria remained fairly con- sistent. Methods Patients Patients. The study population consisted of 5740 adult patients with HCC who underwent transplan- tation in 81 centres in China from January 1980 to June 2008. The inclusion criteria were as follows: (1) patients older than 18 years of age, and (2) histopathological proof of HCC on the explanted liver. The exclusion criteria were as follows: (1) combined transplantation, (2) incidental HCC, and (3) mixed carcinoma and other malignant hepatic tumours. Among the initial 5740 patients, 280 (4.9%) patients for whom LTs were performed at Changzheng Hospital, Shanghai were excluded as the MHCAT was constructed using this cohort of single-centre patients. In addition 3436 (59.9%) patients were excluded because at least one datum involved in MHCAT (four criteria for HCC, pre-transplant AFP and total bilirubin value, re-transplantation, and intraoperative blood loss) was missing; the individual MHCAT score for each HCC patient enrolled cannot be obtained if any datum involved in MHCAT was missing. Third, 646 (11.3%) patients were excluded because they were lost to follow-up within 5 years. Fourth, 7 (0.1%) patients were excluded because their survival status or follow-up dates were missing. Finally, the MHCAT was used for long-term survival predictions in a cohort of 1371 (23.9%) patients with HCC after LT in China. Figure 5 illustrates the patient selection process. Data collection. In May 2008, the CLTR was authorized by the Ministry of Health of the People’s Republic of China as the national registry system for liver transplantation. Pre-transplant data at enrol- ment and post-transplant events were collected retrospectively by independent clinical research assistants in each liver transplantation centre and entered into the CLTR database. Data on the main tumour size, number of tumours, vascular invasion, and conditions of satisfying or dissatisfying different criteria were collected based on the final pathology review of the explanted liver. The study proposal, which was consistent with the ethical guidelines of the 1975 Declaration of Helsinki, was presented to CLTR and approved by Scientific Committee of CLTR in 2013. According to CLTR policy, researchers cannot obtain original data, and the statistical analysis is independently conducted by the CLTR system and statisti- cians complying with the study proposal. All methods were performed in accordance with the approved guidelines for CLTR clinical studies. MHCAT and Metroticket. MHCAT with reference to the Milan, UCSF, Shanghai Fudan, and Hangzhou criteria have been reported elsewhere. Discussion This consistency suggests that we can use the next expanded criteria for MHCAT scoring when an HCC patient demonstrates a borderline score on one criterion.h p The major limitation of the current model is that intraoperative blood loss cannot be predicted before transplantation, hence weakening the role of MHCAT on HCC candidate selection. It is generally agreed that a rational approach that considers markers of tumour behaviour might be useful to select HCC patients for transplantation. Yet, no potential biological, histologic, or molecular markers have been implemented to date in routine practice for selecting HCC candidates. As liver transplantation com- pletely removes primary tumours and potential intrahepatic metastases, HCC recurrence is largely due to circulating tumour cells (CTCs) that existed prior to transplantation18. Intraoperative blood loss may facilitate tumour cell entry into the circulation. Therefore, the counting and culturing of CTCs acts as a promising substitution for intraoperative blood loss or possibly a more valuable predictor1,19,20. A second potential limitation of our study is that the 5-year survival was approximately 54.1 to 59.0% even in low-risk groups. Given that only 34.4% of HCC patients met the Milan criteria in our cohort, the overall long-term survival was quite low. In future studies, we should determine cut-off values that distinguish HCC subgroups with a 5-year survival of greater than 65%, which is comparable with that observed in patients with benign liver diseases. p g In conclusion, we validated the reliability of MHCAT in long-term survival predictions using a large cohort of HCC patients after liver transplantation. Compared with Metroticket, MHCAT exhibited enhanced prediction efficacy for 3-year survival and similar efficacy for 5-year survival, thus making MHCAT a useful tool for selecting HCC transplant candidates. Methods Patients Supplement 8 shows the prediction formula for 3-year and 5-year survival of HCC patients after LT according to MHCAT score. The Metroticket Calculator is available at http://www.hcc-olt-metroticket.org/calculator/. 6 Scientific Reports \| 5:11733 \| DOI: 10.1038/srep11733 www.nature.com/scientificreports/ Figure 5. An illustration of the patient selection process. Figure 5. An illustration of the patient selection process. Statistical Analysis. Statistical analyses were conducted independently by a professional statistician of CLTR using SAS V9.2 (SAS institute, Cary, NC, USA) and checked by another senior statistician. Continuous variables were reported as the mean ± SD or as median (q1, q3) if the variable was not normally distributed. Categorical variables were reported as frequencies (%). The primary outcome was death or re-transplantation of patients. Time was censored at the date of last follow-up assessment for patients who were still alive. The receiver operating characteristic (ROC) curve was used to determine the efficacy of MHCAT and Metroticket, and the area under the ROC curve (c-statistic) was compared using Kruskal-Wallis analysis. To differentiate two groups of patients with significantly different risks of mortality, a cut-off value was derived from the area under the ROC curve of the score based on the highest Youden index. The Kaplan-Meier method with the log-rank test was used to analyse long-term survival rates. P-values were estimated in a two-tailed manner. Differences were considered statistically significant at p < 0.05. References p p p gy j 2. Bertino, G. et al. Systemic therapies in hepatocellular carcinoma: present and future. Future Oncol 9, 1533–1548, doi: 10.2217/ fon.13.171 (2013).h p p p gy j 2. Bertino, G. et al. Systemic therapies in hepatocellular carcinoma: present and future. Future Oncol 9, 1533–1548, doi: 10.2217 fon.13.171 (2013).h 3. Mazzaferro, V. et al. Liver transplantation for the treatment of small hepatocellular carcinomas in patients with cirrhosis. The New England journal of medicine 334, 693–699, doi: 10.1056/NEJM199603143341104 (1996). 3. Mazzaferro, V. et al. Liver transplantation for the treatment of small hepatocellular carcinomas in patients with cirrhosis. The New England journal of medicine 334, 693–699, doi: 10.1056/NEJM199603143341104 (1996). g j f 4. Clavien, P. A. et al. Recommendations for liver transplantation for hepatocellular carcinoma: an international con conference report. The lancet oncology 13, e11–22, doi: 10.1016/S1470-2045(11)70175-9 (2012). h 5. Yao, F. Y. et al. Liver transplantation for hepatocellular carcinoma: validation of the UCSF-expanded criteria based on preoperative imaging. American journal of transplantation : official journal of the American Society of Transplantation and the American Society of Transplant Surgeons 7, 2587–2596, doi: 10.1111/j.1600-6143.2007.01965.x (2007). h 5. Yao, F. Y. et al. Liver transplantation for hepatocellular carcinoma: validation of the UCSF-expanded criteria based on preoperative imaging. American journal of transplantation : official journal of the American Society of Transplantation and the American Society of Transplant Surgeons 7, 2587–2596, doi: 10.1111/j.1600-6143.2007.01965.x (2007). f p g j ( ) 6. Yao, F. Y. et al. Liver transplantation for hepatocellular carcinoma: expansion of the tumor size limits does not adversely impact survival. Hepatology 33, 1394–1403, doi: 10.1053/jhep.2001.24563 (2001). f p g j 6. Yao, F. Y. et al. Liver transplantation for hepatocellular carcinoma: expansion of the tumor size limits does not adversely impac survival. Hepatology 33, 1394–1403, doi: 10.1053/jhep.2001.24563 (2001).t p gy j p 7. Mazzaferro, V. et al. Predicting survival after liver transplantation in patients with hepatocellular carcinoma beyond the Milan criteria: a retrospective, exploratory analysis. The lancet oncology 10, 35–43, doi: 10.1016/S1470-2045(08)70284-5 (2009).it 7. Mazzaferro, V. et al. Predicting survival after liver transplantation in patients with hepatocellular carcinoma beyond the Milan criteria: a retrospective, exploratory analysis. The lancet oncology 10, 35–43, doi: 10.1016/S1470-2045(08)70284-5 (2009).it j f g gy jg 9. Adam, R. et al. Evolution of indications and results of liver transplantation in Europe. A report from the European Liver Transplant Registry (ELTR). implications, and future perspectives. Seminars in oncology 39, 449–460, doi: 10.1053/j.seminoncol.2012.05.012 (2012). 2. Bertino, G. et al. Systemic therapies in hepatocellular carcinoma: present and future. Future Oncol 9, 1533–1548, doi: 10.2217/ fon.13.171 (2013).h 3. Mazzaferro, V. et al. Liver transplantation for the treatment of small hepatocellular carcinomas in patients with cirrhosis. The New England journal of medicine 334, 693–699, doi: 10.1056/NEJM199603143341104 (1996). p g y f p gy j j p 0. China Liver Transplant Registry. CLTR 2011 Annual Scientific Report. (2012). Available at: http://www.cltr.org/pages/statistics statistics_report.jsp. (Accessed: 15th December 2014). p gy j p ( ) 7. Mazzaferro, V. et al. Predicting survival after liver transplantation in patients with hepatocellular carcinoma beyond the Milan criteria: a retrospective, exploratory analysis. The lancet oncology 10, 35–43, doi: 10.1016/S1470-2045(08)70284-5 (2009).it ph gy ( ) ( ) 5. Yao, F. Y. et al. Liver transplantation for hepatocellular carcinoma: validation of the UCSF-expanded criteria based on preoperative imaging. American journal of transplantation : official journal of the American Society of Transplantation and the American Society of Transplant Surgeons 7, 2587–2596, doi: 10.1111/j.1600-6143.2007.01965.x (2007). www.nature.com/scientificreports/ Outcome of small (10-20 mm) arterial phase-enhancing nodules seen on triphasic liver CT in patients with cirrhosis or chronic liver disease. The American journal of gastroenterology 100, 1523–1528, doi: 10.1111/j.1572-0241.2005.41814.x (2005).i j 18. Toso, C., Mentha, G. & Majno, P. Liver transplantation for hepatocellular carcinoma: five steps to prevent recurrence. American journal of transplantation : official journal of the American Society of Transplantation and the American Society of Transplant Surgeons 11, 2031–2035, doi: 10.1111/j.1600-6143.2011.03689.x (2011). 18. Toso, C., Mentha, G. & Majno, P. Liver transplantation for hepatocellular carcinoma: five steps to prevent recurrence. American journal of transplantation : official journal of the American Society of Transplantation and the American Society of Transplant Surgeons 11, 2031–2035, doi: 10.1111/j.1600-6143.2011.03689.x (2011). 18. Toso, C., Mentha, G. & Majno, P. Liver transplantation for hepatocellular carcinoma: five steps to prevent recurrence. American journal of transplantation : official journal of the American Society of Transplantation and the American Society of Transplant Surgeons 11, 2031–2035, doi: 10.1111/j.1600-6143.2011.03689.x (2011). g j 19. Fan, S. T. et al. Prediction of posthepatectomy recurrence of hepatocellular carcinoma by circulating cancer stem ce prospective study. Annals of surgery 254, 569–576, doi: 10.1097/SLA.0b013e3182300a1d (2011). g j 19. Fan, S. T. et al. Prediction of posthepatectomy recurrence of hepatocellular carcinoma by circulating cancer stem cells: a prospective study. Annals of surgery 254, 569–576, doi: 10.1097/SLA.0b013e3182300a1d (2011). g j 19. Fan, S. T. et al. Prediction of posthepatectomy recurrence of hepatocellular carcinoma by circulating cancer stem cells: a prospective study. Annals of surgery 254, 569–576, doi: 10.1097/SLA.0b013e3182300a1d (2011). 20. Sun, Y. F. et al. Circulating stem cell-like epithelial cell adhesion molecule-positive tumor cells indicate poor prognosis of hepatocellular carcinoma after curative resection. Hepatology 57, 1458–1468, doi: 10.1002/hep.26151 (2013). 0. Sun, Y. F. et al. Circulating stem cell-like epithelial cell adhesion molecule-positive tumor cells indicate poor prognosis o hepatocellular carcinoma after curative resection. Hepatology 57, 1458–1468, doi: 10.1002/hep.26151 (2013). 20. Sun, Y. F. et al. Circulating stem cell-like epithelial cell adhesion molecule-positive tumor cells indicate poor prognosis of hepatocellular carcinoma after curative resection. Hepatology 57, 1458–1468, doi: 10.1002/hep.26151 (2013). Acknowledgmenthi g This work was financially supported by the Foundation of Shanghai Science and Technology Commission(no.134119a7300). We highly appreciate Jian-Min Zhang, Professor of English at Zhejiang University, China, for his English editing of the paper. Author Contributions T.F. and H.Q. contributed equally to this work. T.F. and H.Q. wrote the main manuscript text. D.G. and N.Z. acquired data. G.W., S.X. and F.H. analysed and interpreted data. G.X. and M.J. prepared the figures and tables. F.Z. designed and supervised the study. All authors reviewed the manuscript. References Journal of hepatology 57, 675–688, doi: 10.1016/j.jhep.2012.04.015 (2012).i j f g gy jg 9. Adam, R. et al. Evolution of indications and results of liver transplantation in Europe. A report from the European Liver Transplant Registry (ELTR). Journal of hepatology 57, 675–688, doi: 10.1016/j.jhep.2012.04.015 (2012).i 10. China Liver Transplant Registry. CLTR 2011 Annual Scientific Report. (2012). Available at: http://www.cltr.org/pages/statistics/ statistics_report.jsp. (Accessed: 15th December 2014). 10. China Liver Transplant Registry. CLTR 2011 Annual Scientific Report. (2012). Available at: http://www.cltr.org/pages/statistics/ statistics_report.jsp. (Accessed: 15th December 2014). Scientific Reports \| 5:11733 \| DOI: 10.1038/srep11733 7 www.nature.com/scientificreports/ www.nature.com/scientificreports/ 11. Kim, W. R. et al. OPTN/SRTR 2011 Annual Data Report: liver. American journal of transplantation : official journal of the American Society of Transplantation and the American Society of Transplant Surgeons 13 Suppl 1, 73–102, doi: 10.1111/ajt.12021 (2013). ( ) 12. Melloul, E., Lesurtel, M., Carr, B. I. & Clavien, P. A. Developments in liver transplantation for hepatocellular carcinoma. Sem in oncology 39, 510–521, doi: 10.1053/j.seminoncol.2012.05.008 (2012). 13. Duvoux, C. et al. Liver transplantation for hepatocellular carcinoma: a model including alpha-fetoprotein improves performance of Milan criteria. Gastroenterology 143, 986–994 e983; quiz e914-985, doi: 10.1053/j.gastro.2012.05.052 (2012 performance of Milan criteria. Gastroenterology 143, 986 994 e983; quiz e914 985, doi: 10.1053/j.gastro.2012.05.052 (2012 14. Otto, G. et al. How to decide about liver transplantation in patients with hepatocellular carcinoma: size and number of le p gy q j g 14. Otto, G. et al. How to decide about liver transplantation in patients with hepatocellular carcinoma or response to TACE? Journal of hepatology 59, 279–284, doi: 10.1016/j.jhep.2013.04.006 (2013). 14. Otto, G. et al. How to decide about liver transplantation in patients with hepatocellular carcinoma: size and number of lesions or response to TACE? Journal of hepatology 59, 279–284, doi: 10.1016/j.jhep.2013.04.006 (2013). 14. Otto, G. et al. How to decide about liver transplantation in patients with hepatocellular carcinoma: size and or response to TACE? Journal of hepatology 59, 279–284, doi: 10.1016/j.jhep.2013.04.006 (2013). 15. Grasso, A. et al. Liver transplantation and recurrent hepatocellular carcinoma: predictive value of nodule size in a retros and explant study. Transplantation 81, 1532–1541, doi: 10.1097/01.tp.0000209641.88912.15 (2006).t 16. Shah, S. A. et al. Accuracy of staging as a predictor for recurrence after liver transplantation for hepatocellular carcin Transplantation 81, 1633–1639, doi: 10.1097/01.tp.0000226069.66819.7e (2006). y g gt tion 81, 1633–1639, doi: 10.1097/01.tp.0000226069.66819.7e (200 17. O’Malley, M. E., Takayama, Y. & Sherman, M. Outcome of small (10-20 mm) arterial phase-enhancing nodules seen on triphasic liver CT in patients with cirrhosis or chronic liver disease. The American journal of gastroenterology 100, 1523–1528, doi: 10.1111/j.1572-0241.2005.41814.x (2005).i 17. O’Malley, M. E., Takayama, Y. & Sherman, M. Outcome of small (10-20 mm) arterial phase-enhancing nodules seen on triphasic liver CT in patients with cirrhosis or chronic liver disease. The American journal of gastroenterology 100, 1523–1528, doi: 10.1111/j.1572-0241.2005.41814.x (2005).i 17. O’Malley, M. E., Takayama, Y. & Sherman, M. Additional Information Supplementary information accompanies this paper at http://www.nature.com/srep Supplementary information accompanies this paper at http://www.nature.com/srep Competing financial interests: The authors declare no competing financial interests. Competing financial interests: The authors declare no competing financial interests. How to cite this article: Teng, F. et al. Validation of a criteria-specific long-term survival prediction model for hepatocellular carcinoma patients after liver transplantation. Sci. Rep. 5, 11733; doi: 10.1038/srep11733 (2015). How to cite this article: Teng, F. et al. Validation of a criteria-specific long-term survival prediction model for hepatocellular carcinoma patients after liver transplantation. Sci. Rep. 5, 11733; doi: 10.1038/srep11733 (2015). This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Com- mons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/ Scientific Reports \| 5:11733 \| DOI: 10.1038/srep11733 8
https://openalex.org/W2417081056	https://revistadaanpoll.emnuvens.com.br/revista/article/download/147/157	Portuguese	null	O PAPEL DA INTER-RELAÇÃO DO VERBO-VISUAL NO GÊNERO ENTREVISTA PINGUE-PONGUE	Revista da ANPOLL	2,009	cc-by	9,697	p g g ( ) *Doutoranda em Linguística pela universidade Federal de Santa Catarina, Florianópolis, SC. Bolsista CAPES. niveajoi@yaho.com.br. Este trabalho é parte integrante do grupo de pesquisa “Os gêneros do discurso: práticas pedagógicas e análise de gêneros” e do projeto “Estudo dos gêneros do discurso jornalísticos: análises na perspectiva bakhtiniana da linguagem”, ambos coordenados pela Profª. Drª. Rosângela Hammes Rodrigues (uFSC). p j y *Professora do Departamento de Língua e Literaturas Vernáculas – DLLV – uFSC, Florianópolis, SC. Doutora em Linguística Aplicada e Estudos da Linguagem pela Pontifícia universidade Católica de São Paulo. hammes@ cce.ufsc.br 1 De acordo com o tradutor do livro Estética da criação verbal, Paulo Bezerra, com relação ao enunciado, Bakhtin utiliza um só termo – viskázivanie – para designar tanto a enunciação (emissão do discurso) quanto o enunciado (discurso já pronunciado). INTRODUçãO Este artigo tem por objetivo apresentar uma análise de aspectos relativos à dimensão linguageira do gênero entrevista pingue-pongue, do jornalismo de revista, tendo como enfoque a articulação das modalidades verbal e pictórica (ou visual) nesse gênero. Para tanto, inicialmente, delinearemos o quadro teórico em que nos inserimos, a teoria de gêneros do discurso do Círculo de Bakhtin, apresentando conceitos centrais para a consecução do objetivo proposto, em especial, os de enunciado e de gêneros do discurso. Em seguida, faremos a descrição dos dados de pesquisa, precedida de breve exposição e justificativa do percurso metodológico escolhido para a análise. Finalmente, apresentaremos aspectos mais gerais da análise do gênero entrevista pingue-pongue e a análise da articulação do verbo-visual, em especial a modalidade pictórica, destacando o papel discursivo que desempenha no gênero analisado. THE ROLE Of THE VERb-VISUAL INTERRELATION IN THE PING- PONG INTERVIEW GENRE Nívea Rohling da Silva Rosângela Hammes Rodrigues* Nívea Rohling da Silva Rosângela Hammes Rodrigues*** RESumO: Este artigo analisa a inter-relação das modalidades verbal e visual no gênero entrevista pingue-pongue, tendo como embasamento a teoria dialógica do discurso. Observamos que a articulação entre as modalidadesverbalevisualcontribuisignificativamenteparaaconstituição do gênero e do sentido nas entrevistas publicadas. Dentre os elementos pictóricos (fotografia, cor, disposição gráfica dos elementos), a fotografia do entrevistado mostrou-se o mais “saliente” no gênero. Ela é responsável por “atrair” a atenção do leitor, concretizando a interação entre autor e leitor, além de atribuir valoração ao entrevistado e seu discurso. PALAVRAS-CHAVE: entrevista pingue-pongue, linguagem verbo-visual, gêneros do discurso, Bakhtin. ABSTRACT: This article analyzes the inter-relation of the verb-visual modalities in the ping-pong interview genre, based on the dialogic theory of discourse. We observed that the articulation between the verbal and visual modalities significantly contributes to the construction of the genre and the sense in the published interviews. Among the pictorial elements (photography, color, graphic disposal of elements), the interviewee’s photograph was shown to be the most “remarkable” in the genre. It is responsible for “attracting” the reader’s attention, concretizing the interaction between author and reader, as well as valuing the interviewee and his speech. KEYWORDS: ping-pong interview, verbal-visual language, speech genres, Bakhtin. 2 Sobre a referenciação da autoria dos textos “disputados”, optamos por fazer a referência seguindo as indicações dos autores das obras consultadas. 1. A CONCEPçãO DE ENUNCIADO EM bAkHTIN Oenunciado1éumdosconceitoscentraisnateoriadoCírculodeBakhtin, tendo em vista a própria afirmação de Bakhtin a respeito da importância do enunciado nas interações sociais: “Aprender a falar significa aprender a construir enunciados (porque falamos por meio de enunciados e não por orações isoladas e, evidentemente, não por palavras isoladas).” (2003: 283). Segundo Volochinov e Bakhtin2 (1926: 9), “o enunciado concreto (e não a abstração lingüística) nasce, vive e morre no processo da interação social entre os participantes da enunciação. Sua forma e significado são determi- nados basicamente pela forma e caráter desta interação.” Para Volochinov e Bakhtin (1926), o enunciado é a expressão e o produto da interação de três participantes, a saber: o falante (autor), o interlocutor (leitor, contem- plador, espectador, ouvinte) e o tópico (herói, tema, o que ou quem da fala), constituídos e sulcados pelas relações dialógicas (BAKHTIN, 2003). Com relação a sua constituição, o enunciado compõe-se de três elementos intrinsecamente ligados: conteúdo temático, estilo e construção composicional. O conteúdo temático, definido pelas atividades humanas, relaciona-seàsescolhasdoobjetododiscursofeitaspelofalanteparacompor seu discurso. Tais escolhas são realizadas na interlocução; o objeto pode também surgir de outros enunciados, em forma de reação-resposta ativa. A construção composicional, por sua vez, está relacionada à organização textual-discursiva do enunciado e à relação entre os interlocutores, propi- ciando a noção de acabamento do enunciado. Segundo Bakhtin (2003), uma determinada construção composicional permite que o falante tenha a percepção de conjunto do discurso e até faz com que ele antecipe seu (do enunciado do outro) fim. i Quanto ao estilo, Bakhtin (2003) enfatiza que todo enunciado é individual e, por isso, pode refletir a individualidade do falante na linguagem, mas ressalta que nem todos os gêneros possibilitam a expressão do estilo individualizado; alguns gêneros têm como característica principal a padronização e a eliminação de marcas individuais nos enunciados, como, por exemplo, os gêneros das esferas administrativa, militar e acadêmica. Em resumo, de acordo com Bakhtin (2003: 262), o estilo do enunciado, que se molda ao estilo do gênero, caracteriza-se pela “seleção dos recursos léxicos, fraseológicos e gramaticais da língua”. Aindanoquetangeàcaracterizaçãodoenunciado,podemoscitaralgumas características suas como unidade da comunicação discursiva, a saber: o fato de ter contato com a realidade (situação extraverbal [ou extralinguageira]); Nívea Rohling da Silva, Rosângela Hammes Rodrigues 162 a possibilidade de ser delimitado pela alternância dos sujeitos do discurso; a expressividade e a conclusibilidade (BAKHTIN, 2003). SegundoconcepçõesdoCírculodeBakhtin,oenunciado,comoumtodo de sentido, não se limita apenas a sua dimensão verbal (ou linguageira). 1. A CONCEPçãO DE ENUNCIADO EM bAkHTIN O papel da inter-relação do verbo-visual no gênero entrevista pingue-pongue O papel da inter-relação do verbo-visual no gênero entrevista pingue-pongue 163 Na explanação sobre a conclusibilidade, Bakhtin afirma que ela é uma espécie de aspecto interno da alternância dos sujeitos do discurso; essa alternância pode ocorrer precisamente porque o falante disse (ou escreve) tudo o que quis dizer em dado momento ou sob dadas condições. Quando ouvimos ou vemos, percebemos nitidamente o fim do enunciado, como se ouvíssemos o “dixi” conclusivo do falante (2003: 280). Tal noção de conclusibilidade, nos termos bakhtinianos, é facilmente verificada em uma conversa cotidiana, em que o falante percebe o fim do enunciado do interlocutor antes mesmo de sua conclusão e, muitas vezes, antecipa sua resposta. Sob essa ótica, o primeiro critério da conclusibilidade é“apossibilidadederesponderaele[aoenunciado],emtermosmaisprecisos e amplos, de ocupar em relação a ele uma posição responsiva” (2003: 208). O autor aponta, como exemplos de posição responsiva, situações como: cumprir uma ordem, o discurso científico com o qual se pode concordar ou não (inteiramente ou em parte) e as repostas a questionamentos feitos no cotidiano, tais como, “Que horas são?” (BAKHTIN, 2003). O enunciado, na perspectiva bakhtiniana, apresenta duas facetas: o enunciado na condição de evento (acontecimento) e o enunciado como historicidade. Na condição de evento, o enunciado é “acabado”, há um início e um fim que se marca pela alternância dos sujeitos e pela possibilidade de resposta. Enfim, do ponto de vista do acontecimento, o enunciado é único e irrepetível e é um elo na cadeia da comunicação discursiva. Já o caráter histórico do enunciado, por sua vez, está relacionado à noção de “não-acabamento”. Aliás, essa noção está presente no pensamento bakhtiniano não somente no que se refere ao conceito de enunciado, mas, também,noquetangeàprópriaconstruçãodasubjetividadedosujeito,pois, este é, para Bakhtin, também inacabado. Em outros termos, o enunciado, em sua historicidade, está ligado dialogicamente a outros enunciados, uma vez que, segundo Bakhtin, “[n]ão pode haver enunciado isolado. Ele sempre pressupõe enunciados que o antecedem e o sucedem. Nenhum enunciado pode ser primeiro ou último. Ele é apenas o elo na cadeia e fora dessa cadeia não pode ser estudado.” (2003: 371) Assim, o conceito bakhtiniano de enunciado mostra-se complexo, podendo somente ser compreendido no plano do discurso e não no plano da língua (sistema de formas), uma vez que é uma unidade do discurso. 1. A CONCEPçãO DE ENUNCIADO EM bAkHTIN Para além de uma parte verbal, fazem parte do enunciado, como elementos necessários a sua constituição e a sua compreensão total, outros aspectos constitutivos, que compõem sua dimensão extraverbal, ou a sua dimensão social constitutiva. Essa situação extraverbal do enunciado, de acordo com Volochinov (1926), compreende três fatores: 1) o horizonte espacial e temporal comum dos interlocutores (onde e quando do enunciado); 2) o conhecimento e a compreensão comum da situação por parte dos interlocu- tores (aquilo de que se fala); e 3) sua avaliação comum dessa situação (atitude dos participantes do discurso face ao objeto do discurso). O aspecto da expressividade, resultado dessa atitude avaliativa, igual- mente, é considerado uma característica inerente ao enunciado, dado a não existência de enunciados neutros. Portanto, a expressividade está ligada à posição axiológica do autor frente ao objeto do seu discurso e aos interlocutores da interação. A alternância dos sujeitos do discurso, que termina com a transmissão da palavra ao outro, caracteriza-se por fixar “os limites de cada enunciado concreto como unidade da comunicação discursiva” (BAKHTIN, 2003: 175). Compreendemos que essa alternância dos falantes se relaciona às fronteiras do enunciado, em outras palavras, o enunciado do outro serve de limite para o meu próprio enunciado. Na interação verbal, os sujeitos se intercalam na enunciação, produzindo enunciados que requerem reações-respostas ativas. Nessa troca, evidencia-se o caráter dialógico dos enunciados. A alternância dos sujeitos do discurso (limite do enunciado) constitui-se em um dos primeiros critérios que emprestam ao enunciado o status de unidade real da comunicação discursiva e que o caracterizam como um elo na cadeia de muitos outros enunciados, produzindo uma teia de significações. “É esse limite que ‘emoldura’ o enunciado e cria para ele a massa firme.” (BAKHTIN, 2003: 279). i Já a conclusibilidade, por sua vez, confere ao enunciado o caráter de acabamento; ela parte, primeiramente, da possibilidade responsiva do inter- locutor. Esse fechamento, que faz com que o enunciado seja compreendido e passível de resposta, não se apresenta gramaticalmente, mas sim através de interações reais entre os sujeitos do discurso (BAKHTIN, 2003). Nívea Rohling da Silva, Rosângela Hammes Rodrigues 3 Em Rojo (2005), no texto intitulado “Gêneros do discurso e gêneros textuais: questões teóricas e aplicadas”, há uma discussão sobre a diferença teórico-metodológica envolvida no uso das terminologias: gêneros do discurso ou discursivos e gêneros de texto ou textuais. Para a autora, o uso de uma ou outra terminologia reflete duas vertentes de pesquisas ancoradasemdiferentesleituras(interpretações)daobradoCírculodeBakhtinediferentes concepções de gênero. Assim, a distinção está no fato de que a primeira centra seu estudo nas situações de produção dos enunciados ou textos e em seus aspectos sócio-históricos, e a segunda, na descrição da materialidade do texto. 1. A CONCEPçãO DE ENUNCIADO EM bAkHTIN Até aqui expusemos o conceito de enunciado, que se constitui em um elemento importante para a apreensão da noção de gêneros do discurso, que será o tema da seção seguinte. Nívea Rohling da Silva, Rosângela Hammes Rodrigues 164 2. GÊNEROS DO DISCURSO: AS fORMAS DE DISCURSO SOCIAL A partir do conceito de enunciado podemos chegar ao conceito de gêneros do discurso postulado pelo Círculo de Bakhtin. Esse conceito é apresentado em muitos textos do Círculo, e sua terminologia oscila entre formas de discurso social, formas de um todo, tipos de interação verbal e tipos de enunciados (RODRIGuES, 2005). Entretanto, no texto intitulado “Os gêneros do discurso” (2003), onde a questão dos gêneros é detalhada, Bakhtin opta pelo termo gêneros do discurso, que acabou sendo também o termo mais utilizado nas pesquisas em Linguística Aplicada hoje em dia.3 No referido texto, Bakhtin define os gêneros do discurso como “tipos relativamente estáveis de enunciados”, ou seja, através de enunciados individuais, de uma dada situação social de interação de uma determinada esfera social (jornalismo, ciência, escola etc.), que se movimentam em direção a uma regularidade (que é resultado da estabilização dessa situação social de interação), surge o gênero. uma vez constituído, ele funciona como referência (sua normatividade) para a produção de novos enunciados.i Definidos como tipos relativamente estáveis de enunciados, observamos que os gêneros carregam em si um caráter flexível e plástico e de não acabamento. Nessa definição, a palavra relativamente é fundamental, pois ela acena para a flexibilidade do gênero, a qual está diretamente ligada às interações sociais e para o seu não acabamento. Se as relações humanas são complexas, as mudanças são ininterruptas e os gêneros constituem- se a partir das atividades humanas, consequentemente, eles irão refletir as mudanças histórico-sociais. uma vez que as interações humanas estão em constante constituição, os gêneros possuem, então, a mesma característica de “não-acabamento” das atividades humanas e dos enunciados. Ainda sobre a conceituação de gêneros, de acordo com Rodrigues, Bakhtin concebe os gêneros como uma “tipificação social dos enunciados O papel da inter-relação do verbo-visual no gênero entrevista pingue-pongue 165 que apresentam certos traços (regularidades) comuns, que se constituíram historicamente nas atividades humanas, em uma situação de interação relativamente estável” (RODRIGuES, 2005: 164). 4 Blog pode ser traduzido por diário na rede. Sobre esse assunto ver: Komesu (2005). 2. GÊNEROS DO DISCURSO: AS fORMAS DE DISCURSO SOCIAL Assim, segundo Rodrigues, é necessário olhar os gêneros a partir de sua historicidade, já que não são unidades convencionais, mas tipos históricos de enunciados, possuindo a mesma natureza do enunciado (natureza social, discursiva e dialógica).i De acordo com Bakhtin (2003), a diversidade dos gêneros é infinita porque são inesgotáveis as possibilidades das atividades humanas e porque cada esfera comporta um repertório de gêneros do discurso que vai se diferenciando e se ampliando à medida que a própria esfera se desenvolve e torna-se mais complexa. Ao estabelecer a noção de gênero, o autor apresenta uma distinção entre os gêneros primários e os secundários. Os gêneros primários são aqueles ligados à esfera cotidiana, como, por exemplo, o gênero bilheteea conversa cotidiana etc.Jáosgêneros secundários, segundo Bakhtin, “surgem nas condições de um convívio cultural mais complexo e relativamente desenvolvido e organizado (predominantemente escrito)” (2003: 263). Em seu processo de constituição, eles reelaboram diversos gêneros primários. São exemplos de gêneros secundários: o romance, os gêneros científicos, como o artigo e a tese, os gêneros jornalísticos, como a notícia, o editorial e a resenha, entre outros. Já com relação ao surgimento de novos gêneros, podemos dizer que esse processo se dá a partir das demandas sociais de interação, pois um gênero surge ou desaparece em função das condições sócio-discursivas. Nessa mesma perspectiva, Geraldi (2006) propõe que a emergência de novos gêneros está associada às atividades sociais, e que, quanto mais complexa é uma sociedade, mais complexos e em maior número são seus gêneros. Na contemporaneidade, temos observado o surgimento e o desapareci- mentodediversosgêneros.Rodrigues(2005)cita,comoexemplodegêneros que desapareceram de circulação social, a conversa de salão e o romance- folhetim. Já outros gêneros surgem de uma espécie de transmutação de outros gêneros, como, por exemplo, muitos dos gêneros encontrados no suporte digital: o blog4, que lembra o diário, o e-mail, que possui caracte- rísticas da carta. Todavia, os novos gêneros não substituem os já estabe- lecidos: o telefonema não substituiu a conversa e o artigo assinado não excluiu o editorial (RODRIGuES, 2005). Não se trata de uma relação de Nívea Rohling da Silva, Rosângela Hammes Rodrigues Nívea Rohling da Silva, Rosângela Hammes Rodrigues 166 substituição entre gêneros, mas do aparecimento de gêneros a partir das novas necessidades de interlocução, o que ocorre através das mudanças sócio-históricas, repercutindo nas relações de subjetividade e alteridade dos sujeitos. 5 É importante ressaltar que Tzvetan Todorov foi um leitor de Bakhtin, embasando-se nesse autor para elaborar sua construção teórica sobre os gêneros do discurso. Contudo, no texto citado, Todorov não faz alusão à teoria bakhtiniana de gêneros. 2. GÊNEROS DO DISCURSO: AS fORMAS DE DISCURSO SOCIAL A origem dos gêneros, bem como seu processo de mudança, também é discutido por Todorov (1980)5 no âmbito dos gêneros da esfera da literatura, porém essa reflexão também é válida para o presente estudo, uma vez que está em consonância com a teoria dos gêneros do discurso do Círculo de Bakhtin. Na perspectiva de Todorov (1980: 46), um “novo gênero é sempre a transformação de um ou de vários gêneros antigos”. Ainda dentro do escopo literário, mas que pode ser ampliado para uma teoria geral dos gêneros, o autor pontua que, ao se pensar em temporalidade, não há nada “anterior aos gêneros”, pois nunca houve literatura sem gênero. Como diz Bakhtin (2003), sempre falamos por meio de gêneros, pois todos os nossos enunciados são moldados segundo um gênero; por isso, segundo o autor, adquirimos conjuntamente as formas da língua e as formas do discurso (os gêneros). g Todorov propõe que tanto os gêneros literários quanto os gêneros das demais esferas sócio-discursivas têm por origem o discurso humano, e que sua constituição se dá a partir de práticas sociais. A mola propulsora do apa- recimento ou desaparecimento de um gênero são as necessidades comuni- cativas intrínsecas às práticas de interação social. Somente as necessidades discursivas, surgidas nas interações humanas, propiciam o surgimento e modalizam os modos sociais de dizer, a saber, os gêneros do discurso. Em resumo, os gêneros norteiam as interações sociais e, ao mesmo tempo, são por elas norteados; apresentam normatividade (mas também flexibilidade) para as organizações dos enunciados; servem como baliza para o dizer social; trazem, intrinsecamente, todo um universo axiológico. Isso porque os sujeitos se enunciam por meio dos enunciados, construídos dentrodeumdeterminadogêneroe,pormeiodoenunciadoedeseugênero, expressamsuasconcepções demundo,suas crenças, seus valores,revelando, não raras vezes, a voz do outro que compõe seu discurso, formando, desse modo, uma cadeia ininterrupta de sentidos. Dessa maneira, compreende- se a posição também central que os gêneros ocupam na arquitetura do pensamento do Círculo de Bakhtin. O papel da inter-relação do verbo-visual no gênero entrevista pingue-pongue 167 6 Como este artigo apresenta uma parte da totalidade da pesquisa realizada, as questões mais ligadas à dimensão social do gênero não serão discutidas aqui, embora o que apresentamos sobre a dimensão linguageira do gênero esteja assentado nos resultados de análise dos aspectos da dimensão social. 7 As revistas são referenciadas pelo nome tal como ele aparece em suas capas: CartaCapital, sem espaço entre as palavras; ISTOÉ, em letra maiúscula e sem espaço entre as palavras; e Veja, com inicial maiúscula. sobre a dimensão linguageira do gênero esteja assentado nos resultados de análise dos aspectos da dimensão social. 7 As revistas são referenciadas pelo nome tal como ele aparece em suas capas: CartaCapital, sem espaço entre as palavras; ISTOÉ, em letra maiúscula e sem espaço entre as palavras; e Veja, com inicial maiúscula. 6 Como este artigo apresenta uma parte da totalidade da pesquisa realizada, as questões mais ligadas à dimensão social do gênero não serão discutidas aqui, embora o que apresentamos sobre a dimensão linguageira do gênero esteja assentado nos resultados de análise dos aspectos da dimensão social. 7 As revistas são referenciadas pelo nome tal como ele aparece em suas capas: CartaCapital, sem espaço entre as palavras; ISTOÉ, em letra maiúscula e sem espaço entre as palavras; e Veja, com inicial maiúscula. 8 Em seu artigo, Bonini (2000) discute as peculiaridades da interação que se estabelece entre entrevistador e entrevistado através da entrevista por e-mail, propondo estudar uma mídia que impõe um padrão diferenciado de interação, e que, consequentemente, apresenta uma textualidade distinta da entrevista clássica (face a face). 3. PERCURSO METODOLÓGICO Como esta pesquisa insere-se nos estudos analíticos de gêneros e de discurso do Círculo de Bakhtin, também tomamos como fundamento metodológico a ordem metodológica de cunho sócio-histórico proposta por Bakhtin e Volochinov (2004) para o estudo da linguagem. Além disso, também optamos pela proposta metodológica de estudo de gêneros do discurso de Rodrigues (2001), cujas ideias estão ancoradas em Bakhtin. Seguindo a proposta da autora, partimos da análise da “dimensão social do gênero para posteriormente analisarmos sua dimensão verbal [dimensãolinguageira]”.Essapropostaapontaparaanecessidadedeestudar, de antemão, os aspectos sócio-discursivos do gênero, ou seja, o auditório social, as condições de produção, a esfera social em que circula,6 para, só então, proceder à análise de suas (relativas) regularidades linguageiras (no nosso caso, do verbo-visual), correlacionadas com as regularidades da situação social de interação. Entretanto, é importante salientar que esse procedimento de análise (a ordem de análise das dimensões do gênero, que, em dados momentos, são analisadas e apresentadas separadamente) justifica-se apenas por questões metodológicas, tendo em vista que as duas dimensões são indissociáveis na concretização do enunciado e, portanto, também do gênero. Segundo Rodrigues (2001: 248), “tem-se uma relação inextricável entre as dimensões social e verbal do enunciado, que formam a sua unidade, e do enunciado singular e o seu gênero”. Os dados da pesquisa constituem-se de todas as entrevistas pingue- pongues,52entrevistas,publicadasemtrêsrevistassemanaisdeinformação, de circulação nacional: CartaCapital, ISTOÉ e Veja,7 publicadas no período de 4 de outubro de 2006 a 8 de novembro de 2006, período de cobertura do segundo turno das eleições presidenciais no Brasil. Para este artigo, apresen- taremos especificamente algumas regularidades da dimensão linguageira do gênero pesquisado, conforme mencionado. Nívea Rohling da Silva, Rosângela Hammes Rodrigues 168 9 A interação discursiva entre jornalista e entrevistado, que ocorre no momento da entrevista face a face, tanto pode ser reenunciada como entrevista pingue-pongue (objeto deste estudo), enunciado publicado em que a há a sequência de pergunta e resposta; como pode ser reenunciada em forma de discurso citado (reportado) do entrevistado que é inserido em outro gênero (por exemplo, notícia, reportagem, etc.). No caso de inserção do discurso citado do entrevistado em outro gênero, esse discurso passa a fazer parte de outra situação de interação discursiva, com objetivos discursivo-axiológicos diferenciados dos da entrevista pingue-pongue. 4. O GÊNERO ENTREVISTA PINGUE-PONGUE A opção pelo uso do termo entrevista pingue-pongue, ou seja, da nomeação do gênero objeto de pesquisa como entrevista pingue-pongue, ocorre em função da polissemia do termo entrevista, que pode indicar uma variedade de gêneros nomeados como tal (entrevista de emprego, entrevista médica,entrevistafaceafaceetc.);alémdisso,esseéotermomaisrecorrente na esfera de trabalho do jornalismo e, sobretudo, no jornalismo de revista, para identificar as entrevistas que apresentam textualmente a sequência de perguntas e respostas, que são resultado da reenunciação da entrevista face a face. No jornalismo, segundo Oliveira (2002), a entrevista concretiza-se tanto na modalidade oral quanto na modalidade escrita. Na modalidade oral, é um importante gênero constitutivo da mídia radiofônica e televisiva. É comum, tanto no rádio quanto na TV, aparecer em forma de pergunta e resposta (pingue-pongue) entre entrevistador e entrevistado. No suporte jornal ou revista, também se realiza em forma de pergunta-resposta (pingue-pongue) (OLIVEIRA, 2002). A entrevista pingue-pongue apresenta uma complexa relação discursiva entre entrevistador, editoria, entrevistado e leitor. Os interlocutores ocupam lugares diferenciados: quem pergunta (entrevistador), quem responde (entrevistado) e quem edita o texto. mais que posições diferenciadas, tem-se um complexo processo de coautoria, pois editoria, entrevistador e entrevistado constroem o texto. Contudo, vale ressaltar que a definição do conteúdo,composiçãoeestilodogêneroédefinidapelaesferadojornalismo, a partir de uma linha editorial. Para Bonini (2000), a função interativa que se estabelece em qualquer entrevista jornalística é uma publicação em coautoria que deve trazer algo de novo sobre o entrevistado. Assim, a entrevista pingue-pongue constitui-se a partir da edição/ reenunciação da interação direta (face a face) entre entrevistador e entre- vistado, que foi gravada ou registrada em forma de anotações, e, mais recentemente, realizada através de e-mail.8 A partir dessa interação, no processo de reenunciação e retextualização final da entrevista face a face, há uma modalização da fala do entrevistado, isto é, a sua fala é um discurso O papel da inter-relação do verbo-visual no gênero entrevista pingue-pongue 169 citado dentro da fala do entrevistador, que dá o acabamento ao enunciado (a entrevista). O que dizem Bakhtin e Volochinov (2004) para o discurso relatado pode ser aqui aplicado, pois a fala do entrevistado, na reenunciação da entrevista pingue-pongue, é o discurso no discurso. Desse modo, podemos dizer que o gênero entrevista pingue-pongue constitui-se como um discurso citado da entrevista face a face,9 ou seja, há um enquadramento do discurso do entrevistado a partir de uma reenun- ciação da entrevista face a face. 4. O GÊNERO ENTREVISTA PINGUE-PONGUE Esse discurso citado pode ser considerado como “o discurso de outrem [do entrevistado] na linguagem de outrem [do autor], que serve para refratar a expressão das intenções do autor (BAKHTIN, 1998: 127). Ainda sobre a análise da entrevista pingue-pongue, queremos fazer menção ao que Bakhtin diz acerca do discurso do outro no gênero romance: No romance, o homem que fala e sua palavra são objeto tanto de representação verbal como literária. O discurso do sujeito falante no romance não é apenas transmitido ou reproduzido, mas representado artisticamente e, à diferença do drama, representado pelo próprio discurso (do autor). Porém a pessoa que fala e seu discurso constituem um objeto específico enquanto objeto do discurso: não se pode falar do discurso como se fala dos outros objetos [...] os objetos inanimados, os fenômenos, os acontecimentos, etc. (1998: 135). Essa passagem do texto de Bakhtin se refere à “pessoa que fala no romance”, em que há uma representação estética do discurso do outro. Já nos gêneros da “prosa extra-artística” (caso do nosso objeto de pesquisa), Bakhtin afirma haver um processo de transmissão do discurso do outro. Contudo, observamos que o discurso do “sujeito falante” na entrevista pingue-pongue não é apenas transmitido; há uma espécie de “encenação” da entrevista face a face, que, no entanto, é reenunciada e valorada pelo autor da entrevista pingue-pongue. Nívea Rohling da Silva, Rosângela Hammes Rodrigues 170 E tendo em vista que, na entrevista pingue-pongue, o entrevistado e seu discurso constituem-se como objeto do discurso (quem é o entrevistado e o que ele diz), a valoração axiológica se sobressai, uma vez que esse entrevistado já vem envolvido em um “fundo” aperceptivo dos discursos alheios. Segundo Bakhtin (1998: 86), o objeto do discurso (no nosso caso, o entrevistado e seu discurso) “está amarrado e penetrado por idéias gerais, por pontos de vista, por apreciações de outros e por entonações”. Contudo,nãoéqualquerpessoaquepodeassumiroethosdeentrevistado; essa “posição” é “outorgada” aos leitores que possuem uma relação assimétrica com os demais leitores; eles são “selecionados” a partir de seu papel social, que os qualifica para assumir a “posição de entrevistado”.i i Dessa forma, o “fio condutor” na entrevista é o entrevistado e seu discurso, em outras palavras, o que interessa, de fato, nesse gênero, é o entrevistado, cujo discurso, através do enquadramento feito pelo autor da entrevista, é “encharcado” de valoração, é “desacreditado”, é “contestado” ou “enaltecido”. 4. O GÊNERO ENTREVISTA PINGUE-PONGUE Essa valoração dada ao entrevistado concretiza-se através de seu papel social, que é validado pelas “opiniões” sociais e pelos já-ditos sobre esse objeto (entrevistado). Isso nos remete à metáfora de “discurso-raio”, utilizada por Bakhtin para explicar a orientação sobre o objeto do discurso: Se representarmos a intenção, isto é, a orientação sobre o objeto de tal discurso pela forma de um raio, então nós explicaremos o jogo vivo e inimitável de cores e luzes nas facetas da imagem que é construída por elas, devido à refração do “discurso-raio” não no próprio objeto (como o jogo de imagem-tropo do discurso poético no sentido restrito, na “palavra isolada”), mas pela sua refração naquele meio de discursos alheios, de apreciações e de entonações através do qual passa o raio, dirigindo-se para o objeto. A atmosfera social do discurso que envolve o objeto faz brilhar as facetas de sua imagem (1998: 87). Semelhantemente ao efeito de um raio, a atmosfera social do discurso que envolve o objeto faz brilhar as facetas de sua imagem, da mesma forma, os discursos que envolvem o entrevistado fazem com que os já-ditos sobre ele se intensifiquem, tenham ressonâncias ideológicas, constituindo a imagem da “personalidade fútil”, do “político ético ou inescrupuloso”, do “especialista” que é voz de autoridade etc. Em virtude disso, a valoração axiológica do jornalista na entrevista face a face, do autor da entrevista pingue-pongue e do próprio leitor sobre o entrevistado e seu discurso afloram de forma saliente no gênero. Trata-se de discursos “outros” sobre o objeto do discurso, cujas ressonâncias ideológicas se tornam intensas e “tensas” na textualização da entrevista. O papel da inter-relação do verbo-visual no gênero entrevista pingue-pongue O papel da inter-relação do verbo-visual no gênero entrevista pingue-pongue 171 Essa reenunciação da entrevista face a face pode ser compreendida como um processo de intercalação de gênero, tendo em vista que os temas, o auditório social, a concepção de autoria sinalizam para a “presença” da interação face a face, que é reenunciada em forma de entrevista pingue- pongue (enunciado publicado). Contudo, não se trata de uma presença explícita, pelo contrário, é como se fosse uma “encenação”, demonstrando umarelaçãodeconstutividadegenérica10entreessesdoisgêneros(entrevista face a face e entrevista pingue-pongue).i Por fim, evidenciamos, ainda, que o gênero entrevista pingue-pongue se materializa através de dois materiais semióticos: o verbal e o pictórico, tema que será analisado na seção seguinte. 11 A noção de organização textual está relacionada à composição do gênero de Bakhtin. Porém a composição do gênero nos termos bakhtinianos é mais ampla que a noção de organização (ou composição) textual, uma vez que inclui os aspectos da dimensão social, como a com- posição dos participantes da interação. usamos o termo organização textual por estarmos apresentando apenas uma faceta, a textual, da noção de composição do gênero. 10 “Capturar” o processo axiológico-discursivo de reenunciação da entrevista face a face, discutindo a relação de dependência constitutiva entre esta e a entrevista pingue-pongue, é um dos objetivos de pesquisa de doutorado em andamento de Nívea Rohling da Silva. 12 As entrevistas pingue-pongues satélites não são publicadas pela revista CartaCapital. 10 “Capturar” o processo axiológico-discursivo de reenunciação da entrevista face a face, discutindo a relação de dependência constitutiva entre esta e a entrevista pingue-pongue, é um dos objetivos de pesquisa de doutorado em andamento de Nívea Rohling da Silva. 11 A noção de organização textual está relacionada à composição do gênero de Bakhtin. Porém a composição do gênero nos termos bakhtinianos é mais ampla que a noção de organização (ou composição) textual, uma vez que inclui os aspectos da dimensão social, como a com- posição dos participantes da interação. usamos o termo organização textual por estarmos apresentando apenas uma faceta, a textual, da noção de composição do gênero. 12 As entrevistas pingue-pongues satélites não são publicadas pela revista CartaCapital. 5. A ORGANIZAçãO TEXTUAL DA ENTREVISTA PINGUE-PONGUE11 A organização textual, como o nome já explicita, diz respeito aos elementos que compõem textualmente o gênero entrevista pingue-pongue, como título, subtítulo, introdução, sequência de perguntas e de respostas, “olho”, fotografia etc. Nesta seção, abordaremos esse conjunto de elementos, bem como sua sequenciação na organização da dimensão linguageira do gênero (materialidade do gênero). Inicialmente, vale salientar que durante a análise dos dados percebemos manifestações diferenciadas do gênero nas revistas pesquisadas (no que se refere ao seu lugar na revista e seu papel no todo da revista), que refletiam a relevância e a valoração dadas ao entrevistado e seu discurso. Em virtude disso, fizemos o seguinte agrupamento das entrevistas: i g g 1. Entrevistas pingue-pongues nucleares: publicadas nas páginas vermelhas e nas seções principais da revista ISTOÉ; nas páginas amarelas e seções principais da revista da Veja; e em qualquer seção na CartaCapital. ç p p j q q ç p 2. Entrevistas pingue-pongues satélites:12 publicadas em seções diversas, como, por exemplo, nas seções destinadas a colunismo social (Holofote 12 As entrevistas pingue-pongues satélites não são publicadas pela revista CartaCapital. Nívea Rohling da Silva, Rosângela Hammes Rodrigues 172 (Veja) e Gente (Veja)); nas seções dedicadas a discutir “acontecimentos” políticos da semana (Brasil Confidencial (ISTOÉ)); e nas seções que discutem comportamento, profissão, saúde etc. (seção GuIA Veja (Veja)). Nessas seções, a entrevista cumpre a tarefa de “completar” e/ou “reafirmar” outros enunciados, como reportagens. O critério para esse agrupamento das entrevistas, como já mencionado, foi estabelecido a partir da percepção de que, nas variadas manifestações do gênero nas edições pesquisadas, algumas entrevistas se encontravam em uma posição de “independência” em relação aos demais gêneros e ocu- pavam, nesses casos, uma página ou até mesmo uma seção da revista, o que nos levou a nomear esse conjunto de entrevistas como entrevista nuclear. Já em outros casos, o gênero se encontrava em posição de “subordinação” a outros gêneros em uma mesma página da revista e, consequentemente, em uma mesma seção; a esse conjunto, atribuímos o nome de entrevista satélite. Além disso, após analisar os elementos constitutivos do gênero e sua organização na materialidade do texto, agrupamos as entrevistas em duas tabelas,deacordocomsuassimilaridades:aTabela1apresentaaorganização textual da entrevista nuclear nas diferentes revistas e a Tabela 2 apresenta a organização textual da entrevista satélite (ISTOÉ e Veja). 5. A ORGANIZAçãO TEXTUAL DA ENTREVISTA PINGUE-PONGUE11 CartaCapital istoÉ Veja Nome da seção (variável) Nome da seção (no caso: Entrevista) e nome do entrevistado (na mesma linha) Nome da seção (no caso: Entrevista) e nome do entrevistado (na mesma linha) Título Título Título Subtítulo e foto do entrevistado Subtítulo e foto do entrevistado Subtítulo e foto do entrevistado Nome do jornalista Nome do jornalista Nome do jornalista Introdução Introdução Introdução Sequência de perguntas e res- postas: as perguntas, primeira- mente, introduzidas com o nome darevistae,depois,somentepelas iniciais da revista; e respostas (introduzidas com o nome do entrevistado). Há intercalação de “olho”e fotografias. A entrevista é apresentada em colunas. Não há inserção de gêneros da esfera da propaganda. Sequência de perguntas e res- postas: perguntas (introduzidas comonomedarevista)erespostas (introduzidas com o nome do entrevistado). Há intercalação de “olho” e fotografias. A entrevista se apresenta em colunas. As bordas das páginas são impressas na cor vermelha em um fundo branco. Há inserção de gêneros da esfera da propaganda, que ocupam um espaço de destaque dentro da entrevista. Sequência de perguntas e res- postas: perguntas (introduzidas comonomedarevista)erespostas (introduzidas através do nome do entrevistado). Há intercalação de “olho” e fotografias. A entrevista apresenta-se em colunas. A entrevista é publicada em folha de cor amarela. Há inserção de gêneros da esfera da propaganda, que ocupam um espaço de destaque dentro da entrevista. Sequência de perguntas e res- postas: as perguntas, primeira- mente, introduzidas com o nome darevistae,depois,somentepelas iniciais da revista; e respostas (introduzidas com o nome do entrevistado). Há intercalação de “olho”e fotografias. A entrevista é apresentada em colunas. Não há inserção de gêneros da esfera da propaganda. Sequência de perguntas e res- postas: perguntas (introduzidas comonomedarevista)erespostas (introduzidas através do nome do entrevistado). Há intercalação de “olho” e fotografias. A entrevista apresenta-se em colunas. A entrevista é publicada em folha de cor amarela. Há inserção de gêneros da esfera da propaganda, que ocupam um espaço de destaque dentro da entrevista. Tabela 1 - Organização textual da entrevista pingue-pongue nuclear. O papel da inter-relação do verbo-visual no gênero entrevista pingue-pongue 173 Os dados apresentados na Tabela 1 revelam que a entrevista nuclear apresenta uma relativa estabilidade em sua organização textual, a qual não se altera significativamente de uma edição para outra, ou de revista para revista. Nívea Rohling da Silva, Rosângela Hammes Rodrigues 5. A ORGANIZAçãO TEXTUAL DA ENTREVISTA PINGUE-PONGUE11 Essa organização textual compreende a seguinte sequência de elementos: a) Inserção do título da entrevista – constitui-se de um “olho”; b) Inserção de introdução – a introdução apresenta o entrevistado e o “teor” da entrevista; também faz menção ao nome do jornalista, que é, na maioria das vezes, identificado como “repórter”; i c) Inserção de sequência de perguntas e respostas – em alguns casos, as perguntas e respostas são introduzidas, respectivamente, pelo nome da revista e pelo nome do entrevistado; contudo, em outras situações, inexiste a inserção do nome da revista e do nome do entrevistado;i d) Inserção de foto do entrevistado – a fotografia do entrevistado geralmente se situa na lateral direita do texto ou na parte inferior da entrevista; e) Inserção de uma frase de “fechamento”. Ainda, na organização textual (cf. Tabela 1), constatamos a inserção, ou melhor, a “invasão” de gêneros da propaganda na entrevista pingue- pongue, que consideramos um gênero “parasitário” nessa situação, uma vez que impõe sua presença no fluxo da leitura da entrevista. Há também a presença de outros materiais semióticos além do verbal, que nomeamos como material semiótico pictórico ou visual (cores, diagramação das letras, fotografia etc.). Tendo em vista a importância da questão verbo-visual na constituição do nosso objeto de estudo, apresentaremos, na seção seguinte, a análise de seu papel na entrevista pingue-pongue. Ainda, na organização textual (cf. Tabela 1), constatamos a inserção, ou melhor, a “invasão” de gêneros da propaganda na entrevista pingue- pongue, que consideramos um gênero “parasitário” nessa situação, uma vez que impõe sua presença no fluxo da leitura da entrevista. Há também a presença de outros materiais semióticos além do verbal, que nomeamos como material semiótico pictórico ou visual (cores, diagramação das letras, fotografia etc.). Tendo em vista a importância da questão verbo-visual na constituição do nosso objeto de estudo, apresentaremos, na seção seguinte, a análise de seu papel na entrevista pingue-pongue. 5. A ORGANIZAçãO TEXTUAL DA ENTREVISTA PINGUE-PONGUE11 Entretanto, podem ocorrer algumas alterações com relação à diagramação, ou seja, elementos como a fotografia do entrevistado, o título, o subtítulo, dentre outros, podem receber diagramações diferenciadas, mas a presença desses elementos, bem como sua “sequência” no texto, sofrem pouca variação. Portanto, há uma relativa estabilidade na organização textual do gênero, que obedece à seguinte sequenciação textual: a) Inserção do nome da seção e do nome do entrevistado – o nome da seção em que está inserida a entrevista aparece na parte superior da página e, na mesma linha, há a inserção do nome do entrevistado; b) Inserção do título – o título está interligado ao conteúdo semântico- objetal da entrevista; em algumas situações, ele constitui-se em um “olho”; algo que foi dito pelo entrevistado e que está inserido na entrevista; c) Inserção do nome do jornalista – aparece com menos destaque; essa “assinatura” do jornalista sinaliza a autoria; d) Inserção de introdução – a introdução contextualiza o entrevistado, destacando seu papel social, suas realizações e o assunto/tema sobre o qual ele se pronuncia; e) Inserção de sequência de perguntas e respostas – introduzidas, respectivamente, pelo nome da revista e pelo nome do entrevistado. Há, nessa sequência, a intercalação de “olhos” e fotografia(s) do entrevistado. q gi Já a entrevista satélite apresenta a seguinte organização textual: Elementos da organização textual das entrevistas pingue-pongues satélites Título Introdução (reduzida) Pergunta e resposta (introduzidas com o nome da revista e do entrevistado) Foto do entrevistado (ao final da entrevista ou nas margens laterais) Frase de fechamento Tabela 2 - Organização textual da entrevista pingue-pongue satélite. Tabela 2 - Organização textual da entrevista pingue-pongue satélite. A Tabela 2, diferentemente do que aconteceu com a Tabela 1, não foi subdivida por revistas porque nas publicações das revistas ISTOÉ e Veja a organização apresenta a mesma sequência de elementos. A organização textual da entrevista satélite também se caracteriza fundamentalmente pela Nívea Rohling da Silva, Rosângela Hammes Rodrigues 174 “estrutura” pergunta-resposta, porém não contempla todos os elementos observados na entrevista nuclear. Issoocorre em virtude do espaço reduzido que ocupa na revista, o que, consequentemente, constitui-se em um índice valorativo. Essa organização textual compreende a seguinte sequência de elementos: “estrutura” pergunta-resposta, porém não contempla todos os elementos observados na entrevista nuclear. Issoocorre em virtude do espaço reduzido que ocupa na revista, o que, consequentemente, constitui-se em um índice valorativo. 6. A QUESTãO DOVERbO-VISUAL NA ENTREVISTA PINGUE-PONGUE Aentrevistapingue-pongue,comomencionado,materializa-sepormeio de duas modalidades semióticas: a verbal e a pictórica, cuja articulação na organizaçãotextualtambéméresponsávelpelaconstruçãodossentidosdas/ nas entrevistas. O material semiótico pictórico é constituído por elementos como as fotografias, as cores das páginas e também a disposição gráfica dos elementos no texto/enunciado. Dentre os elementos mencionados, o que mais se destaca nesse gênero é a fotografia, que, na esfera do jornalismo, normalmente integra também os gêneros notícia e reportagem e é chamada de fotojornalismo. Brait (2004: 47) observa que “a técnica fotográfica assegura o simulacro visual O papel da inter-relação do verbo-visual no gênero entrevista pingue-pongue O papel da inter-relação do verbo-visual no gênero entrevista pingue-pongue 175 do acontecimento, [pois] cria-se o efeito de objetividade, de transparência, como se não houvesse um enunciador”. mas, apesar desse apagamento do enunciador, a autora salienta que, na verdade, a construção da cena enunciativa revela a existência de um sujeito da enunciação e isso ocorre, segundo a autora, porque o enquadramento, o dimensionamento da luz e outros recursos da linguagem fotográfica funcionam discursivamente, isto é, não têm um valor em si, enquanto signos de um sistema de comunicação e significação, mas assinalam escolhas de um sujeito, tendo em vista o discurso a ser construído e os efeitos de sentido que devem ser produzidos no enunciatário (BRAIT, 2004: 47). Já a fotografia que está incorporada à entrevista pingue-pongue tem uma função diferente da dos gêneros notícia e reportagem, pois ela não tem por objetivo “capturar” e valorar um “fato” propriamente dito, mas apresentar, na maioria dos casos, foto(s) do entrevistado e de outros elementos relacionados ao horizonte temático da entrevista em particular (embora também valoradas discursivamente). A fotografia do entrevistado é um elemento constitutivo do gênero, haja vista que em todas as incidências dos dados de pesquisa constatamos a inserção de uma ou mais fotografias da pessoa entrevistada. Isso ocorre porque é ela (a fotografia) que reforça e impulsiona o leitor para a leitura das entrevistas, uma vez que, ao visualizar uma fotografia “destacada” como objeto discursivo (o entrevistado), o leitor pode se sentir impelido a ler (ou não) a entrevista.i Além de atrair a atenção do leitor, a fotografia também é um lugar da materialização da valoração axiológica, pois, dependendo da foto publicada, ela ou “depõe contra” o entrevistado, ou o exalta, tendo em vista seu papel social e a intenção interlocutiva do autor da foto e da entrevista (e da instância jornalística). Nívea Rohling da Silva, Rosângela Hammes Rodrigues 6. A QUESTãO DOVERbO-VISUAL NA ENTREVISTA PINGUE-PONGUE Portanto, a escolha desse elemento pictórico, que ajuda a compor a dimensão linguageira do gênero, corrobora com o projeto discursivo do autor da entrevista, o que equivale a dizer que não se trata de uma escolha “neutra” ou “aleatória”, mas de um trabalho estilístico- composicional pertencente ao domínio da autoria. Segundo Vannuchi (2007a), na parte final de “produção”, após a edição gráfica, a entrevista retorna à editoria para que sejam feitos os últimos ajustes necessários para a apresentação visual do texto, ou seja, algum corte que ainda seja necessário, a composição das legendas das fotos etc. Assim, a articulação entre os elementos verbais e pictóricos (principalmente os fotográficos) faz parte do acabamento estilístico-composicional do gênero. Nívea Rohling da Silva, Rosângela Hammes Rodrigues 176 Entretanto, é preciso salientar que a fotografia não é apenas um dos materiais semióticos da entrevista pingue-pongue; ela também é a materialização de um enunciado citado, de um já-dito, que é reenunciado e enquadrado na entrevista para fazer parte do projeto discursivo do gênero. Assim, consideramos a fotografia inserida em uma entrevista como um enunciado dentro de outro enunciado, ou seja, como um discurso citado, tendo em vista que ela também possui seu próprio projeto discursivo, com sua concepção de autor e de interlocutor; ela é um enunciado. Na entrevista, há a inserção tanto de fotografias de arquivo, ou seja, de fotografias “tiradas” em outra situação social de interação para cumprir outros propósitos discursivos, quanto de fotografias “tiradas” no momento da entrevista face a face.13 Na sequência, apresentaremos a articulação do verbo-visual no gênero entrevista pingue-pongue por meio de análise de uma entrevista em particular, cujo entrevistado é o escritor João ubaldo Ribeiro. Por questões de ordem prática, tendo em vista a impossibilidade de reproduzir uma entrevista de cinco páginas no corpo deste artigo, apresentaremos cada página dessa entrevista separadamente, com exceção das páginas em que se insere a propaganda; neste caso, juntamos duas páginas em um só espaço, uma vez que, neste artigo, não analisaremos o papel da intercalação de propagandas na entrevista. A entrevista com o escritor João ubaldo Ribeiro (ISTOÉ, n. 1930, 2006) foi publicada na seção de entrevistas (páginas vermelhas) em uma edição anterioraosegundoturnodaseleiçõespresidenciais.Tendoemvistaaépoca da publicação, o assunto que se sobressai é a posição do entrevistado sobre a eleição presidencial. 13 Isso é uma prática recorrente no jornalismo, haja vista a inserção de fotografias “assinadas” por diferentes fotógrafos em uma mesma entrevista. 6. A QUESTãO DOVERbO-VISUAL NA ENTREVISTA PINGUE-PONGUE A entrevista é “percebida” pelo leitor como um todo, ou seja, ela é considerada como um único enunciado, que é constituído pelo material verbal e pelo material pictórico, articulados através da diagramação, quer dizer, pela forma como os elementos verbais e pictóricos estão distribuídos nas páginas da entrevista, que, por sua vez, é uma das facetas do acabamento do enunciado. Toda a entrevista é “emoldurada” por bordas vermelhas, o que justifica o “apelido” de “páginas vermelhas” que essa seção recebe na redação da revista (VANNuCHI, 2006). Os elementos que integram a entrevista são (em sequência): O papel da inter-relação do verbo-visual no gênero entrevista pingue-pongue 177 Figura 1 - Primeira página da entrevista com João ubaldo Ribeiro (ISTOÉ). Figura 1 - Primeira página da entrevista com João ubaldo Ribeiro (ISTOÉ). Figura 1 - Primeira página da entrevista com João ubaldo Ribeiro (ISTOÉ). Figura 1 - Primeira página da entrevista com João ubaldo Ribeiro (ISTOÉ). a) Inserção do nome da seção da revista e do nome do entrevistado (Entrevista – João Ubaldo Ribeiro) – ocupa a parte superior da entrevista e se destaca pelo tamanho e pela forma das letras;i b) Inserção de uma fotografia colorida do entrevistado – é o elemento de maior destaque na primeira página. A fotografia insere o entrevistado em seu “ambiente” de trabalho (escritório), pois, ao fundo, percebemos a presença de livros e de um computador. Como a imagem “mostra” o ambiente de trabalho do entrevistado, ela conduz o leitor a relacionar essa fotografia com o papel social de intelectual (escritor) do entrevistado. Na extremidade direita, a fotografia recebe a “assinatura” do fotógrafo (Hélio Nagamine); b) Inserção de uma fotografia colorida do entrevistado – é o elemento de maior destaque na primeira página. A fotografia insere o entrevistado em seu “ambiente” de trabalho (escritório), pois, ao fundo, percebemos a presença de livros e de um computador. Como a imagem “mostra” o ambiente de trabalho do entrevistado, ela conduz o leitor a relacionar essa fotografia com o papel social de intelectual (escritor) do entrevistado. Na extremidade direita, a fotografia recebe a “assinatura” do fotógrafo (Hélio Nagamine); Nívea Rohling da Silva, Rosângela Hammes Rodrigues 178 c) Inserção de título (“NÃO AGÜENTO A CARA DELES”) – o título aparecebemdestacado,comletrasmaiúsculas,centralizado;eleéconstruído a partir de um “olho”, que, segundo Vannuchi (2007b), constitui-se em uma frase dita pelo entrevistado e que é colocada em destaque, pelo autor da entrevista. Ao “retirar” do corpo do texto um trecho de uma “fala” e inseri- lo no título, temos um caso de bivocalidade, uma vez que percebemos duas vozes que se materializam nesse novo enunciado. gura 3 - Quarta página da entrevista com João ubaldo Ribeiro (ISTOÉ). Há a voz do entrevistado, que efetivamente diz “Não agüento a cara deles” e a voz do autor, que quer ressaltar essa posição valorativa do entrevistado; esse destacamento de um trecho da fala do entrevistado também é um ato valorativo, pois a escolha demanda destaque para certas partes da fala do entrevistado e exclusão de outras; d) Inserção do nome do jornalista (Por ELIANE LOBATO) – em letra maiúscula, mas com menor destaque; d) Inserção do nome do jornalista (Por ELIANE LOBATO) – em letra maiúscula, mas com menor destaque; e) Inserção de introdução e subtítulo – em três colunas insere-se a introdução, que apresenta o entrevistado (“escritor baiano João ubaldo Ribeiro, 65 anos”) e, na sequência, os “assuntos” abordados na entrevista: 1. O descontentamento do entrevistado para com os políticos brasileiros (“ainda não sabe o que vai fazer quando estiver frente a frente com a urna eletrônica”); 2. A criminalidade (“realça a angústia da criminalidade carioca”); 3. A relação do entrevistado com o álcool (“revela como se livrou do alcoolismo”). Como podemos perceber, a introdução faz uma espécie de apresentação e resumo da entrevista e, de certa forma, “simula” a apresentação que geralmente o jornalista faz do entrevistado na entrevista da mídia televisiva ou radiofônica. “Encaixada” na introdução, há a inserção de uma frase do entrevistado, que se constitui em um subtítulo: “Totalmente desencantado com os políticos brasileiros, o escritor baiano não consegue sequer assistir aos debates eleitorais na tevê”. O subtítulo acentua a temática do segundo turno das eleições presidenciais e “expõe” a valoração do entrevistado em relação ao referido tema. O papel da inter-relação do verbo-visual no gênero entrevista pingue-pongue 179 Figura 2 - A inserção de gênero da esfera da propaganda na entrevista com João ubaldo Ribeiro (ISTOÉ). Inserção de propaganda – ocupa página dupla, ou seja, a segunda e a terceira páginas da entrevista. Trata-se de uma propaganda do provedor de Internet Terra. Figura 2 - A inserção de gênero da esfera da propaganda na entrevista com João ubaldo Ribeiro (ISTOÉ). Inserção de propaganda – ocupa página dupla, ou seja, a segunda e a terceira páginas da entrevista. Trata-se de uma propaganda do provedor de Internet Terra. Inserção de propaganda – ocupa página dupla, ou seja, a segunda e a terceira páginas da entrevista. Trata-se de uma propaganda do provedor de Internet Terra. Nívea Rohling da Silva, Rosângela Hammes Rodrigues 180 Figura 3 - Quarta página da entrevista com João ubaldo Ribeiro (ISTOÉ). a) Inserção do nome da seção (ENTREVISTA) – indicação do nome da seção em letra maiúscula sobre fundo preto e vermelho; a) Inserção do nome da seção (ENTREVISTA) – indicação do nome da seção em letra maiúscula sobre fundo preto e vermelho; a) Inserção do nome da seção (ENTREVISTA) – indicação do nome da seção em letra maiúscula sobre fundo preto e vermelho; b) Inserção da sequência de perguntas e respostas – é justamente dessa sequência que se originou o termo entrevista pingue-pongue, pois ela “simula” as perguntas e respostas da entrevista face a face. A pergunta é introduzida pelo nome da revista, que aparece em letras maiúsculas e destacado em negrito (ISTOÉ). A resposta é introduzida pelo nome completo do entrevistado, em letras minúsculas e destacado em negrito; somente na primeira inserção do nome do entrevistado é que é utilizado o O papel da inter-relação do verbo-visual no gênero entrevista pingue-pongue 181 nome completo (João Ubaldo Ribeiro), pois, da segunda inserção em diante, aparece somente o pré-nome (João Ubaldo), o que sugere um “ar” de já-dito, de conhecido e de intimidade com o entrevistado, uma vez que a escolha poderia ter recaído pela retomada do sobrenome do autor;i c) Inserção de “olho” (“Eu fico de queixo caído quando alguém me escreve defendendo Lula, dizendo que ele fez um governo extraordinário”) – as primeiras linhas de abertura, as quais indicam que é uma “fala” do entrevistado que está integrando o “conteúdo” da entrevista, são destacadas através de ampliação da fonte e estão impressas na cor preta; as aspas de fechamento são apresentadas em fonte menor, na cor branca;i p d) Inserção de fotografia do candidato à Presidência Luiz Inácio Lula da Silva – o autor da entrevista insere, logo abaixo do “olho”, a fotografia de uma pessoa citada na entrevista (o candidato à Presidência Luiz Inácio Lula da Silva). Trata-se de uma fotografia colorida e pequena, em que o candidato aparece em tom “sério”; e, ao contrário da foto anterior, esta não explicita a “assinatura”. A intercalação de outra fotografia, que não a do entrevistado, é uma particularidade da revista ISTOÉ e, em alguns casos, da revista CartaCapital. Figura 4 - Quinta página da entrevista com João ubaldo Ribeiro (ISTOÉ). O papel da inter-relação do verbo-visual no gênero entrevista pingue-pongue Figura 3 - Quarta página da entrevista com João ubaldo Ribeiro (ISTOÉ). Já nas entrevistas das páginas amarelas da revista Veja, constatamossomenteapresençadeumaúnicafotografia,doentrevistado.A maneira como a fotografia do presidente e candidato Lula está diagramada, em que ele é mostrado de forma a exprimir um “tom” oblíquo, confere a ele uma expressão de dúvida, de fragilidade, de defesa. A forma como o projeto gráfico seleciona essa foto, e não outra, e a articula com os demais elementos verbais (por exemplo, o título “Não agüento a cara deles”) e com o “olho” (“eu fico de queixo caído quando alguém me escreve defendendo Lula, dizendo que ele fez um governo extraordinário”), possibilita ao leitor- eleitor capturar efeitos de sentido que vão além do objetivo discursivo de “ilustrar” a entrevista, adentrando caminhos em direção à materialização ideológica de posições políticas. A relação entre o material linguístico e o materialpictóricocriaefeitosdesentidos,resultadosdeposiçõesaxiológicas da autoria do gênero. Nívea Rohling da Silva, Rosângela Hammes Rodrigues 182 Figura 4 - Quinta página da entrevista com João ubaldo Ribeiro (ISTOÉ). a) Inserção da sequência de perguntas e respostas; b) Inserção de imagem – é inserida a imagem de Nossa Senhora do Perpétuo Socorro, mas, nesse caso, não se trata de uma fotografia, mas de uma réplica de pintura. A imagem da Santa é inserida na entrevista, uma vez que o entrevistado a cita, atribuindo a ela sua decisão de ter “se livrado do alcoolismo”. Da mesma maneira que a fotografia do presidente Lula, a imagem da Santa se refere a um “outro” inserido na entrevista; b) Inserção de imagem – é inserida a imagem de Nossa Senhora do Perpétuo Socorro, mas, nesse caso, não se trata de uma fotografia, mas de uma réplica de pintura. A imagem da Santa é inserida na entrevista, uma vez que o entrevistado a cita, atribuindo a ela sua decisão de ter “se livrado do alcoolismo”. Da mesma maneira que a fotografia do presidente Lula, a imagem da Santa se refere a um “outro” inserido na entrevista; O papel da inter-relação do verbo-visual no gênero entrevista pingue-pongue 183 c) Inserção de “olho” (“Prometi para Nossa Senhora do Perpétuo Socorro que pararia de beber. Continuo indo ao boteco, mas só tomo guaraná.”) – logo abaixo da imagem da Santa. A análise da articulação entre elementos verbais e visuais na entrevista pingue-pongue mostrou que a inter-relação desses elementos contribui significativamente para a constituição do gênero, e que, dentre os elementos pictóricos supracitados (cor, disposição gráfica dos elementos etc.), a fotografia do entrevistado é o mais “saliente” no gênero. Isso se comprova pelo fato de que em todas as incidências do gênero há pelo menos uma fotografia do entrevistado. É ela a responsável por “atrair” a atenção do leitor para que ele leia a entrevista, concretizando assim a interação discursiva entre autor e leitor, além de atribuir valoração axiológica à pessoa do entrevistado.i Pudemos observar que a fotografia (bem como os demais elementos pictóricos)nãoseconstituiemumelementomeramenteilustrativo.Trata-se, pois, de um importante aspecto que é responsável pelas relações dialógicas no enunciado e pela materialização dos acentos de valor atribuídos ao entrevistado e ao tema de seu discurso, conforme apresentado na seção de análise. Assim, a inserção de foto(s) do entrevistado e de outros elementos relacionados ao horizonte temático da entrevista em particular constitui o enunciado (e o gênero) e corrobora para a produção de efeitos de sentido na comunicação discursiva que se realiza por meio da entrevista. CONSIDERAçõES fINAIS Procuramos, neste trabalho, apresentar algumas regularidades observadas na análise da dimensão linguageira do gênero entrevista pingue-pongue, do jornalismo de revista, em especial, a inter-relação dos elementos verbais e pictóricos (verbo-visual) na entrevista. Demonstramos que o gênero entrevista pingue-pongue se materializa através de dois materiais semióticos: o verbal e o pictórico. O elemento pictórico que se sobressaiu na análise dos dados foi a fotografia, tendo em vista que esta se faz presente em todos os enunciados do gênero que compuseram os dados de pesquisa. mostramos também que esse caráter multimodal do gênero amplia os sentidos discursivo-axiológicos materializados nos enunciados e pode se constituir em uma área fértil para novas pesquisas. Nívea Rohling da Silva, Rosângela Hammes Rodrigues 184 REfERÊNCIAS bIbLIOGRÁfICAS BAKHTIN, m. m. Estética da criação verbal. Trad. do russo por Paulo Bezerra. 4. ed. São Paulo: martins Fontes, 2003. BAKHTIN, m.; VOLOCHINOV, V. N. Marxismo e filosofia da linguagem: problemas fundamentais do método sociológico na ciência da linguagem. Trad. do francês por michel Lahud e Yara F. Vieira. 11. ed. São Paulo: Hucitec, 2004. BONINI, Adair. Entrevista por e-mail: pragmática de um gênero (des) conhecido ou problemas comunicativos na variação do gênero. Revista de Letras, Fortaleza - CE, v. 22, n. 1/2, p. 5-13, 2000. BRAIT, Beth. A construção do sentido: exemplo fotográfico persuasivo. Líbero, São Paulo, v. 6, n. 11, p. 44-49, 2004. GERALDI, João Wanderley. [Curso ministrado] Seminário Bakhtin: linguagem e sujeito, entre a ética e a estética. Florianópolis: universidade Federal de Santa Catarina, 2006. KOmESu, F. C. Blogs e as práticas de escrita sobre si na Internet. In: mARCuSCHI, L. A.; xAVIER, A. C. (Org.). Hipertexto e gêneros digitais: novas formas de construção do sentido. Rio de Janeiro: Lucerna, 2005. p. 110-119. OLIVEIRA, Ana Tereza Pinto de. O gênero entrevista na imprensa escrita e sua relação com as modalidades da língua. 2002. Disponível em: <http://www.fiamfaam.br/comunicacao/projetos/inovacoes/idademidia/ pdfs/art_111-16_im1.pdf>. Acesso em: 15 jul. 2006. RIBEIRO, João ubaldo. Não agüento a cara deles. ISTOÉ, São Paulo: Três Editorial, n. 1930, 18 out. 2006. RODRIGuES. Rosângela Hammes. A constituição e o funcionamento do gênero jornalístico artigo: cronotopo e dialogismo. Tese (Doutorado em LinguísticaAplicadaeEstudosdaLinguagem)–Pontifíciauniversidade Católica de São Paulo, 2001. O papel da inter-relação do verbo-visual no gênero entrevista pingue-pongue O papel da inter-relação do verbo-visual no gênero entrevista pingue-pongue 185 RODRIGuES.RosângelaHammes.Osgênerosdodiscursonaperspectiva dialógica da linguagem: a abordagem de Bakhtin. In: mEuRER, José Luiz; BONINI, Adair; mOTA-ROTH, Désirée (Org.). Gêneros: teorias, métodos e debates. São Paulo: Parábola Editorial, 2005. p. 152-183. RODRIGuES.RosângelaHammes.Osgênerosdodiscursonaperspectiva dialógica da linguagem: a abordagem de Bakhtin. In: mEuRER, José Luiz; BONINI, Adair; mOTA-ROTH, Désirée (Org.). Gêneros: teorias, métodos e debates. São Paulo: Parábola Editorial, 2005. p. 152-183. ROJO, Roxane. Gêneros do discurso e gêneros textuais: questões teóricas e aplicadas. In: mEuRER, José Luiz; BONINI, Adair; mOTA-ROTH, Désirée (Org.). Gêneros: teorias, métodos e debates. São Paulo: Parábola Editorial, 2005. p. 184-207. SANT’ANNA, A. Propaganda: teoria-técnica-prática. 7. ed. São Paulo: Pioneira, 2001. SILVA, Nívea Rohling da. O gênero entrevista pingue-pongue: reenunciação, enquadramento e valoração do discurso do outro. Dissertação (mestrado) – universidade Federal de Santa Catarina, Florianópolis, 2007. TODOROV, Tzvetan. Os gêneros do discurso. São Paulo: martins Fontes, 1980. VANNuCHI, Camilo. REfERÊNCIAS bIbLIOGRÁfICAS A entrevista pingue-pongue no jornalismo de revista. Entrevista concedida via e-mail em 12 set. 2006. VANNuCHI, Camilo. Condições de produção de uma revista semanal. Entrevista concedida via e-mail em 02 fev. 2007a. VANNuCHI, Camilo. Processo de produção da entrevista pingue-pongue. Entrevista concedida via e-mail em 03 maio 2007b. VOLOCHINOV, V. N.; BAKHTIN, m. m. Discurso na vida e discurso na arte (sobre a poética sociológica). Trad. inédita de Carlos Alberto Faraco e Cristóvão Tezza [para fins didáticos]. Versão da língua inglesa de I. R. Titunik a partir do original russo, 1926. Nívea Rohling da Silva, Rosângela Hammes Rodrigues 186
https://openalex.org/W2340029238	https://iris.uniroma1.it/bitstream/11573/960180/1/Gallo_Campylobacter_2016.pdf	English	null	Campylobacter jejuni Fatal Sepsis in a Patient with Non-Hodgkin’s Lymphoma: Case Report and Literature Review of a Difficult Diagnosis	International journal of molecular sciences	2,016	cc-by	6,010	Campylobacter jejuni Fatal Sepsis in a Patient wit Non-Hodgkin’s Lymphoma: Case Report and Literature Review of a Difﬁcult Diagnosis Maria Teresa Gallo 1,†, Enea Gino Di Domenico 1,,†, Luigi Toma 2, Francesco Marchesi 3, Lorella Pelagalli 4, Nicola Manghisi 1, Fiorentina Ascenzioni 5, Grazia Prignano 1, Andrea Mengarelli 3 and Fabrizio Ensoli 1 1 Department of Clinical Pathology and Microbiology, San Gallicano Institute, IRCCS, Rome 00144, Italy gallo@ifo.it (M.T.G.); n.manghisi@gmail.com (N.M.); prignano@ifo.it (G.P.); ensoli@ifo.it (F.E.) 1 Department of Clinical Pathology and Microbiology, San Gallicano Institute, IRCCS, Rome 00144, Italy gallo@ifo.it (M.T.G.); n.manghisi@gmail.com (N.M.); prignano@ifo.it (G.P.); ensoli@ifo.it (F.E.) 1 Department of Clinical Pathology and Microbiology, San Gallicano Institute, IRCCS, Rome 00144, Italy; gallo@ifo.it (M.T.G.); n.manghisi@gmail.com (N.M.); prignano@ifo.it (G.P.); ensoli@ifo.it (F.E.) 2 Department of Infectious Disease, San Gallicano Institute, IRCCS, Rome 00144, Italy; toma@ifo.it 3 Department of Hematology, Regina Elena National Cancer Institute IRCCS, Rome 00144, Italy; marchesi.francesco@tiscali.it (F.M.); mengarelli@ifo.it (A.M.) p gy gy y gallo@ifo.it (M.T.G.); n.manghisi@gmail.com (N.M.); prignano@ifo.it (G.P.); ensoli@ifo.it (F.E.) 2 Department of Infectious Disease, San Gallicano Institute, IRCCS, Rome 00144, Italy; toma@ifo.it 3 Department of Hematology, Regina Elena National Cancer Institute IRCCS, Rome 00144, Italy; marchesi.francesco@tiscali.it (F.M.); mengarelli@ifo.it (A.M.) p , , , , y; 3 Department of Hematology, Regina Elena National Cancer Institute IRCCS, Rome 00144, Italy; marchesi.francesco@tiscali.it (F.M.); mengarelli@ifo.it (A.M.) 4 Intensive Care Medicine, Regina Elena National Cancer Institute IRCCS, Rome 00144, Italy; pelagalli@ifo.it 5 Department of Biology and Biotechnology “Charles Darwin”, University of Rome Sapienza, Rome 00185, 4 Intensive Care Medicine, Regina Elena National Cancer Institute IRCCS, Rome 00144, Italy; pelagalli@ifo.it 5 4 Intensive Care Medicine, Regina Elena National Cancer Institute IRCCS, Rome 00144, Italy; pelagalli@ifo.it 5 Department of Biology and Biotechnology “Charles Darwin”, University of Rome Sapienza, Rome 00185, Italy; ﬁorentina.ascenzioni@uniroma1.it 4 Intensive Care Medicine, Regina Elena National Cancer Institute IRCCS, Rome 00144, Italy; pelagalli@ifo.it 5 Department of Biology and Biotechnology “Charles Darwin”, University of Rome Sapienza, Rome 00185, Italy; ﬁorentina.ascenzioni@uniroma1.it , g , , y; p g 5 Department of Biology and Biotechnology “Charles Darwin”, University of Rome Sapienza, Rome 00185 Italy; ﬁorentina.ascenzioni@uniroma1.it y Correspondence: e.didomenico@ifo.it; Tel.: +39-06-5266-2956; Fax: +39-06-5266-5396 † These authors contributed equally to this work. Academic Editor: Susanna Esposito Received: 25 January 2016; Accepted: 7 April 2016; Published: 12 April 2016 Abstract: Campylobacter jejuni (C. jejuni) bacteremia is difﬁcult to diagnose in individuals with hematological disorders undergoing chemotherapy. The cause can be attributed to the rarity of this infection, to the variable clinical presentation, and to the partial overlapping symptoms underlying the disease. Campylobacter jejuni Fatal Sepsis in a Patient wit Non-Hodgkin’s Lymphoma: Case Report and Literature Review of a Difﬁcult Diagnosis Here, we report a case of a fatal sepsis caused by C. jejuni in a 76-year-old Caucasian man with non-Hodgkin’s lymphoma. After chemotherapeutic treatment, the patient experienced fever associated with severe neutropenia and thrombocytopenia without hemodynamic instability, abdominal pain, and diarrhea. The slow growth of C. jejuni in the blood culture systems and the difﬁculty in identifying it with conventional biochemical phenotyping methods contributed to the delay of administering a targeted antimicrobial treatment, leading to a fatal outcome. Early recognition and timely intervention are critical for the successful management of C. jejuni infection. Symptoms may be difﬁcult to recognize in immunocompromised patients undergoing chemotherapy. Thus, it is important to increase physician awareness regarding the clinical manifestations of C. jejuni to improve therapeutic efﬁcacy. Moreover, the use of more aggressive empirical antimicrobial treatments with aminoglycosides and/or carbapenems should be considered in immunosuppressed patients, in comparison to those currently indicated in the guidelines for cancer-related infections supporting the use of cephalosporins as monotherapy. Keywords: Campylobacter jejuni; non-Hodgkin’s lymphoma; chemotherapy; skin lesion International Journal of Molecular Sciences International Journal of Molecular Sciences 2. Case Presentation A 76-year-old ma A 76-year-old man was hospitalized in our Department of Hematology of the “Regina Elena” National Cancer Institute in Rome on 13 March, 2014. He suffered from a Diffuse Large B-Cell Lymphoma that had evolved from a previously diagnosed indolent non-Hodgkin Lymphoma (NHL) which was refractory to three chemo-immunotherapeutic lines of treatment and was characterized by cerebral and meningeal involvement at the time of last progression. National Cancer Institute in Rome on 13 March, 2014. He suffered from a Diffuse Large B-Cell Lymphoma that had evolved from a previously diagnosed indolent non-Hodgkin Lymphoma (NHL) which was refractory to three chemo-immunotherapeutic lines of treatment and was characterized by cerebral and meningeal involvement at the time of last progression. Upon admission, the patient had evening fever and severe dysarthria (Figure 1). On March 14, Upon admission, the patient had evening fever and severe dysarthria (Figure 1). On March 14, he received an urgent salvage treatment based on a chemo-immunotherapeutic regimen containing Rituximab 375 mg/m2 on day 1, Methotrexate 1 g/m2 on day 2, and Cytarabine 1 g total dose twice daily, for days 3 and 4. Given the presence of evening fevers and a moderate increase in procalcitonin levels (mini VIDAS system, bioMérieux, Florence, Italy) to 2.62 ng/mL (normal, <0.5 ng/mL), an empirical antibiotic therapy was administered including Ceftriaxone (2 g daily) at the beginning of the salvage chemo-immunotherapy, even in the absence of any microbiological evidence from the blood cultures and surveillance swabs. After 48 h, a complete regression of fever and a decrease in procalcitonin levels to 1.69 ng/mL were observed. Serial blood cultures, taken on March 18, were incubated in an automated, noninvasive culture system (BacT/ALERT, bioMérieux, Florence, Italy). he received an urgent salvage treatment based on a chemo-immunotherapeutic regimen containing Rituximab 375 mg/m2 on day 1, Methotrexate 1 g/m2 on day 2, and Cytarabine 1 g total dose twice daily, for days 3 and 4. Given the presence of evening fevers and a moderate increase in procalcitonin levels (mini VIDAS system, bioMérieux, Florence, Italy) to 2.62 ng/mL (normal, <0.5 ng/mL), an empirical antibiotic therapy was administered including Ceftriaxone (2 g daily) at the beginning of the salvage chemo-immunotherapy, even in the absence of any microbiological evidence from the blood cultures and surveillance swabs. After 48 h, a complete regression of fever and a decrease in procalcitonin levels to 1.69 ng/mL were observed. 1. Introduction Campylobacter jejuni represents one of the most common worldwide causes of bacterial gastroenteritis with over 190,000 cases occurring annually in the 27 member states of the European Union (www.efsa.europa.eu/efsajournal). Clinical manifestations include abdominal pain, fever, and diarrhea [1]. Unlike other enteric infections, C. jejuni is only rarely associated with extraintestinal localization and systemic invasive illness [1,2]. Bacteremia caused by C. jejuni has been detected in less than 1% of Int. J. Mol. Sci. 2016, 17, 544; doi:10.3390/ijms17040544 www.mdpi.com/journal/ijms Int. J. Mol. Sci. 2016, 17, 544 Unlike other ente localization and system 2 of 9 stinal ed in patients with gastroenteritis and it has been mainly reported in elderly and in immunocompromised patients [1,2]. immunocompromised patients [1,2]. In this study, we describe a case of C. jejuni sepsis in a patient with non-Hodgkin’s lymphoma h l d i f l Th l i id f C j j i b i d h i f In this study, we describe a case of C. jejuni sepsis in a patient with non-Hodgkin’s lymphoma that resulted in a fatal outcome. The low incidence of C. jejuni bacteremia and the paucity of associated symptoms make this infection difﬁcult to detect in patients with hematological disorders where selecting the appropriate antibiotic treatment is crucial, and at present, early and distinctive clinical features have not yet been fully elucidated. that resulted in a fatal outcome. The low incidence of C. jejuni bacteremia and the paucity of associated symptoms make this infection difficult to detect in patients with hematological disorders where selecting the appropriate antibiotic treatment is crucial, and at present, early and distinctive clinical features have not yet been fully elucidated. 2 Case Presentation 2. Case Presentation A 76-year-old ma Serial blood cultures, taken on March 18, were incubated in an automated, noninvasive culture system (BacT/ALERT, bioMérieux, Florence, Italy). Figure 1. The patient’s clinical course. Procalcitonin (PCT-Q) levels were expressed as ng/mL. Antimicrobial susceptibility testing (AST) was performed by Etest®, according to the Clinical and Laboratory Standards Institute (CLSI) breakpoints for non-Enterobacteriaceae. Figure 1. The patient’s clinical course. Procalcitonin (PCT-Q) levels were expressed as ng/mL. Antimicrobial susceptibility testing (AST) was performed by Etest®, according to the Clinical and Laboratory Standards Institute (CLSI) breakpoints for non-Enterobacteriaceae. Figure 1. The patient’s clinical course. Procalcitonin (PCT-Q) levels were expressed as ng/mL. Antimicrobial susceptibility testing (AST) was performed by Etest®, according to the Clinical and Laboratory Standards Institute (CLSI) breakpoints for non-Enterobacteriaceae. Figure 1. The patient’s clinical course. Procalcitonin (PCT-Q) levels were expressed as ng/mL. Antimicrobial susceptibility testing (AST) was performed by Etest®, according to the Clinical and Laboratory Standards Institute (CLSI) breakpoints for non-Enterobacteriaceae. On March 19, the hemocytometric assessment showed severe neutropenia and thrombocytopenia (hemoglobin 75 g/liter, platelet count 6 × 109/liter, white blood cell count 0.06 × 109/liter). On March 20, the stool culture exam gave negative results. Nevertheless, on March 21, the patient had a relapse (fever > 39 °C) in the absence of symptoms indicating hemodynamic instability as well as abdominal pain or diarrhea. Based on the assumption that the patient was undergoing a sepsis, the patient was On March 19, the hemocytometric assessment showed severe neutropenia and thrombocytopenia (hemoglobin 75 g/liter, platelet count 6 ˆ 109/liter, white blood cell count 0.06 ˆ 109/liter). On March 20, the stool culture exam gave negative results. Nevertheless, on March 21, the patient had a relapse (fever > 39 ˝C) in the absence of symptoms indicating hemodynamic instability as well as abdominal pain or diarrhea. Based on the assumption that the patient was undergoing a sepsis, the patient was empirically treated with intravenous Piperacillin-Tazobactam (4.5 g three times a day), without 3 of 9 Int. J. Mol. Sci. 2016, 17, 544 clinical improvement. The abdominal echography revealed a severe circumferential thickening of the cecum wall with submucosal edema, whereas procalcitonin levels increased to 3.64 ng/mL. Meanwhile, on March 22, the blood cultures were positive revealing curved gram-negative rods at the microscopic analysis. The organism was subcultured onto chocolate agar (bioMérieux, Florence, Italy) and then incubated at 36 ˝C in a microaerophilic environment with 5% CO2. 2. Case Presentation A 76-year-old ma Thus, on March 23, based on the abdominal echography (suggestive for ileotiphlitis), and the patient’s general clinical conditions and increased procalcitonin levels, even in the absence of microbiological data (blood cultures were negative, so far), a different antibiotic therapeutic regimen was implemented. The patient was administered Tygeciclin 50 mg intravenously twice a day after a loading dose of 100 mg, Metronidazole 500 mg four times a day, and Caspofungin 50 mg daily after a loading dose of 70 mg. Despite implementing this type of antibiotic treatment, a rapid clinical deterioration in the patient was observed. Additionally, on March 23, cellulitis in the patient’s left leg was observed during a dermatological consultation. However, a skin biopsy was not advised due to the general health condition of the patient. p After 48 h of incubation, on March 24, irregular shaped grey and ﬂat colonies appeared on the chocolate agar plates. The isolate was initially identiﬁed as C. jejuni by distinct colony morphology and by conventional biochemical tests resulting in oxidase- and hippurate-positive results. Despite the microbiology laboratory promptly notifying the possible or likely infection of C. jejuni and the immediate implementation of empirical intravenous treatment with Gentamicin 6 mg/kg/Die, a further worsening of the patient’s clinical condition was observed on March 24. Surprisingly, microbiological testing by VITEK 2 system (bioMérieux, Florence, Italy) initially identiﬁed the microorganism as Francisella Tularensis (96% of identiﬁcation conﬁdence) whereas repeated testing yielded Moraxella spp. (95% of identiﬁcation conﬁdence), thereby creating uncertainty in the identiﬁcation of the microorganism present. Thus, the poor health condition of the patient and severe cytopenia (hemoglobin 69 g/liter, platelet count 3 ˆ 109/liter, and white blood cell count 0.5 ˆ 109/liter) contributed to a rapid fatal outcome on March 26. Further identiﬁcation of the microorganism was performed by sequence analysis (ABI PRISM 3130xl Genetic Analyzer) of the 16S rRNA gene [3]. The sequence showed 99.9% similarity and 100% coverage for the strains of C. jejuni subsp. jejuni ATCC 700819. The sequences were deposited in the European Nucleotide Archive (ENA) with accession number LN864495. Antimicrobial susceptibility testing (AST) was performed by Etest®, according to the Clinical and Laboratory Standards Institute (CLSI) breakpoints for non-Enterobacteriaceae as follows: ciproﬂoxacin, §1 µg/mL (Sensitive); doxycycline, §4 µg/mL (Sensitive); gentamicin, §4 µg/mL (Sensitive); meropenem, §4 µg/mL (Sensitive) (Table 1). Table 1. Antibiotic susceptibility testing of the isolated bacteria. 3. Discussion Infections caused by C. jejuni are only rarely complicated by extraintestinal localization or bacteremia [1]. In immunocompetent patients, C. jejuni bacteremia can be transient and resolved without antimicrobial therapy [1]. Conversely, individuals with immune deﬁciency or another serious underlying condition (cardiovascular disorders, hematological malignancies, liver disease, hypogammaglobulinemia, and human immunodeﬁciency virus infection) are exposed to an increased risk of bacteremia due to C. jejuni [2,4]. In these individuals, an effective antimicrobial treatment has been signiﬁcantly associated with an improved outcome [2]. In a large number of cases, a timely identiﬁcation of the pathogen and appropriate empirical antimicrobial therapy are hampered by the atypical presentation of the symptoms caused by C. jejuni [1,2]. The clinical signs of Campylobacter bacteraemia are generally accompanied by an acute-onset febrile illness of a transient nature with self-limiting enteritis. Nevertheless, in a large percentage of cases the clinical presentation of Campylobacter bacteraemia may show a febrile illness without gastrointestinal symptoms [4]. Other typical manifestations observed in severe sepsis caused by Campylobacter may include, skin lesions, cytopenia, and diarrhea, however, these symptoms also occur frequently in patients with aggressive lymphomas undergoing chemotherapy [2]. Moreover, the absence of consensus on the optimal antibiotic regimen and the lack of studies comparing different empirical treatments for C. jejuni bacteraemia make it difﬁcult for the clinician to select an appropriate antimicrobial therapy. Different strategies were adopted, including ﬂuoroquinolones (ciproﬂoxacin), macrolides (erythromycin), and aminoglycosides (gentamicin) [5]. Fluoroquinolones (e.g., ciproﬂoxacin) were largely used for the treatment of Campylobacter infection and, in general, are considered the drugs of choice for the empirical treatment of diarrheal illnesses [6–8]. Campylobacter and other organisms, such as Salmonella or Shigella species, were generally susceptible to ﬂuoroquinolones, thus empirical treatment with these drugs is used without waiting for the stool culture results. However, since the early 1990s a growing number of ﬂuoroquinolone-resistant Campylobacter strains have been registered in Asia as well as in several European countries. This increase of resistant strains is not only the result of the excessive use of these antimicrobials in clinical practice, but it is also the consequence of the use of ﬂuoroquinolones in food producing animals and in veterinary species [9–11]. Thus, the possibility of ﬂuoroquinolone-resistant strains must be considered in all cases of Campylobacter bacteraemia. In the presence of conﬁrmed Campylobacter infections, macrolides (erythromycin, or alternatively clarithromycin or azithromycin) represent the frontline agents [12], with tetracycline, doxycycline, and chloramphenicol considered alternative drugs [8]. 2. Case Presentation A 76-year-old ma Antibiotic Tested MIC Test Result Ciproﬂoxacin §1 µg/mL Sensitive Doxycycline §4 mcg/mL Sensitive Gentamicin §4 mcg/mL Sensitive Meropenem §4 mcg/mL Sensitive MIC: Minimal Inhibitory Concentration performed by Etest® (bioMérieux, Florence, Italy), according to the Clinical and Laboratory Standards Institute (CLSI) breakpoints for non-Enterobacteriaceae. Table 1. Antibiotic susceptibility testing of the isolated bacteria. The Central Ethics Committee I.R.C.C.S. Lazio, section of the Istituti Fisioterapici Ospitalieri in Rome, in compliance with the Helsinki Declaration, approved this case report (Prot. CE/1016/15—4 December 2015). Data and relevant scientiﬁc articles were identiﬁed via speciﬁc PubMed database searches from January 1980 and December 2015. The terms included in the search comprised: “Campylobacter jejuni” 4 of 9 Int. J. Mol. Sci. 2016, 17, 544 and “bacteremia” or “Campylobacter” and “bacteremia” or “non-Hodgkin’s lymphoma”. Research was restricted to English language articles. and “bacteremia” or “Campylobacter” and “bacteremia” or “non-Hodgkin’s lymphoma”. Research was restricted to English language articles. 3. Discussion However, recent evidence suggests that Campylobacter is also becoming increasingly resistant to macrolides, which represents a rising concern for public health [13]. The use of macrolides at subtherapeutic levels in chickens is considered a major factor inﬂuencing the emergence of resistant strains [13–15]. Thus, for serious systemic infections it has been demonstrated that aminoglycoside, gentamicin, or carbapenems are the most efﬁcient antimicrobials [2,4,8,16,17]. In our case, the absence of clear clinical signs of a possible infection with enteric pathogens suggested that the patient be treated with ceftriaxone in accordance with the guidelines for cancer-related infections in immunosuppressed patients that support the use of cephalosporins in monotherapy [18]. Third-generation cephalosporins are largely used for the empirical treatment of community-acquired infectious diarrhea. However, these antimicrobial agents have not been proven effective for treating bacteremia due to Campylobacter species other than Campylobacter fetus [2,19]. Moreover, the use of third-generation cephalosporins and ﬂuoroquinolones in the treatment of Campylobacter bacteraemia has shown poor prognosis and a high frequency of resistant strains has resulted in a general discouragement towards using this class of antibiotics [4,20], particularly Int. J. Mol. Sci. 2016, 17, 544 5 of 9 in hospitals and communities with a high prevalence of extended-spectrum beta-lactamases (ESBLs)-producing bacteria. After the ﬁrst antimicrobial treatment, the patient presented neutropenia and fever, and therapy was then subsequently changed. In the absence of relevant microbiological data, the guidelines for the empirical therapy of febrile neutropenic cancer patients receiving chemotherapy recommend the use of pipercillin-tazobactam as ﬁrst line monotherapy for the treatment of bloodstream infections [21]. However, Campylobacter isolates are not regularly susceptible to penicillins [22–24] and the b-lactamase enzyme found in C. jejuni is preferentially inhibited by clavulanic acid, but not by tazobactam or sulbactam [22,24]. In our case, only after having diagnosed sepsis caused by C. jejuni, the patient was empirically treated with gentamicin and the subsequent susceptibility drug proﬁle indicated that this strain was in fact susceptible to this antimicrobial (Table 1). Nevertheless, the patient died because of complications due to a septic status and multiorgan failure. It is important to note that the treatment with gentamicin in this patient had started long after the appearance of initial enteric symptoms (diarrhea) and the ﬁrst signs of sepsis. The delayed start of the targeted antimicrobial treatment was due to the difﬁculty in identifying C. 3. Discussion jejuni bacteraemia, which, in turn, was the consequence of the very slow growth of this bacterium in standard automatic blood culture systems [25]. In fact, blood cultures are only rarely performed in patients presenting an apparently simple diarrhea symptom. This, as well as the slow growth of the characteristics of C. jejuni and the self-limited nature of this infection may represent a contributing cause to underestimating the real incidence of C. jejuni bacteraemia [26]. Additionally, the inability of the automated biochemical phenotyping system to promptly and correctly identify C. jejuni further deferred the recognition of the pathogen. The slow growth of C. jejuni in the BacT/ALERT and the repeated unsuccessful attempts in identifying the bacteria reported for the VITEK 2 system made the recognition of this pathogen particularly elusive. Indeed, previous studies have demonstrated that despite the Neisseria-Haemophilus (NH) identiﬁcation card for VITEK 2 correctly identifying most C. jejuni ssp. Jejuni, misclassiﬁcations occur at a rate of more than 10% [27]. In this case, the diagnosis, and consequently the start of an appropriate therapy, was further delayed by the negative result of the stool cultures after the ﬁrst episodes of diarrhea. Diarrheal illnesses in patients with neoplasia and immunosuppressive therapy are rarely perceived as a necessity to perform blood cultures, even when there is a fever present. On the other hand, it should be considered that blood stream infections caused by C. jejuni might occur without evidence of diarrhea, suggesting that this bacterium can access the intestinal mucosa without causing local inﬂammation [28]. A retrospective study suggested that a diagnostic clue for the presence of C. jejuni infection might be represented by leukopenia or thrombocytopenia, particularly when associated with an acute febrile diarrheal illness [29]. However, in neoplastic and immune suppressed patients, such as in our case, the marked cytopenia might be interpreted as a result of the immunosuppressed status of the patient who underwent a chemo-immunotherapeutic program. In addition, three days before the fatal outcome, the patient also experienced the occurrence of cellulitis of the left leg. It has been reported that, although less recognized, skin lesions may represent a complication of Campylobacter bacteraemia that occurs particularly in patients with immune-related problems [30]. Again, the presence of NHL and chemotherapy made it difﬁcult to recognize cellulitis as a sign of C. jejuni infection since lymphomas can be also characterized by an initial skin presentation [31]. 4. Conclusions In summary, although C. jejuni bacteraemia is uncommon, it may develop either primarily or secondarily from gastroenteritis, and thereby may represent a severe disease for immunocompromised individuals [30]. Occasionally, both NHLs and C. jejuni sepsis may intertwine; in cases such as this, it may create difﬁculties in being able to make a plain distinction between the root cause(s) of a patient’s symptoms. Many hematological disorders, especially lymphoid neoplasms, have a high risk for infection, thus when dealing with immunocompromised patients a septic disease should be 6 of 9 Int. J. Mol. Sci. 2016, 17, 544 suspected even in the presence of mild symptoms. In aggressive lymphoma, and in patients undergoing chemotherapy, fatigue, fever, diarrhea, as well as skin lesions and cytopenia may occur frequently, but these symptoms may occur also in severe sepsis caused by C. jejuni. Recognizing the early symptoms of a C. jejuni bacteraemia in hematological patients is key to initiate an effective antimicrobial therapy. From our experience, and from the data reported in the literature [2,4], blood cultures should always be performed in febrile patients with gastroenteritis. Therapy with appropriate antimicrobial agents is an important component in the management of immunocompromised patients with C. jejuni bacteraemia. Guidelines for cancer-related infections in immunosuppressed patients support the use of cephalosporins in monotherapy [18], whereas for the treatment of febrile neutropenic cancer patients receiving chemotherapy the use of pipercillin-tazobactam is recommended [21]. From our study, and from the data reported in the literature, it emerged that immunosuppressed patients with suspected Campylobacter sepsis should receive a more aggressive antimicrobial treatment—possibly combining aminoglycosides and/or carbapenems with cephalosporins in the ﬁrst line antimicrobial empirical treatment. Nevertheless, the risk caused by the rise in antibiotic resistance among bacteria, particularly with Campylobacter spp. should also be considered where an increase in the administration of multiple antibiotics is likely to lead to colonization and infection with antibiotic-resistant organisms [32]. antibiotics is likely to lead to colonization and infection with antibiotic-resistant organisms [32]. In this case, the unequivocal identiﬁcation of C. jejuni was not obtained in time, and only by sequence analysis of the 16S rRNA gene. This further suggested that diagnostic systems, other than those based on the biochemical identiﬁcation (i.e., molecular techniques and Mass Spectrometry—MS) should be preferred for a prompt and unequivocal laboratory identiﬁcation of C. jejuni. References 1. Young, K.T.; Davis, L.M.; Dirita, V.J. Campylobacter jejuni: Molecular biology and pathogenesis. Nat. Rev. Microbiol. 2007, 5, 665–679. [CrossRef] [PubMed] 2. Pacanowski, J.; Lalande, V.; Lacombe, K.; Boudraa, C.; Lesprit, P.; Legrand, P.; Trystram, D.; Kassis, N.; Arlet, G.; Mainardi, J.L.; et al. Campylobacter bacteremia: Clinical features and factors associated with fatal outcome. Clin. Infect. Dis. 2008, 47, 790–796. [CrossRef] [PubMed] 3. Di Domenico, E.G.; Toma, L.; Prignano, G.; Pelagalli, L.; Police, A.; Cavallotti, C.; Torelli, R.; Sanguinetti, M.; Ensoli, F. Misidentiﬁcation of Streptococcus uberis as a human pathogen: A case report and literature review. Int. J. Infect. Dis. 2015, 33, 79–81. [CrossRef] [PubMed] 4. Nielsen, H.; Hansen, K.K.; Gradel, K.O.; Kristensen, B.; Ejlertsen, T.; Østergaard, C.; Schønheyder, H.C. Bacteraemia as a result of Campylobacter species: A population-based study of epidemiology and clinical risk factors. Clin. Microbiol. Infect. 2010, 16, 57–61. [CrossRef] [PubMed] 5. Hagensee, M.E.; Benyunes, M.; Miller, J.A.; Spach, D.H. Campylobacter jejuni bacteremia and Guillain-Barre´ syndrome in a patient with GVHD after allogeneic BMT. Bone Marrow. Transplant. 1994, 13, 349–351. [PubMed] 6. Guerrant, R.L.; van Gilder, T.; Steiner, T.S.; Thielman, N.M.; Slutsker, L.; Tauxe, R.V.; Hennessy, T.; Grifﬁn, P.M.; DuPont, H.; Sack, R.B.; et al. Practice guidelines for the management of infectious diarrhea. Clin. Infect. Dis. 2001, 32, 331–351. [CrossRef] [PubMed] 7. Aarestrup, F.M.; McDermott, P.F.; Wegener, H.C. Transmission of antibiotic resistance from food animals to humans. In Campylobacter; Nachamkin, I., Szymanski, C.M., Blaser, M.J., Eds.; ASM Press: Washington, DC, USA, 2008; pp. 645–665. 8. Ge, B.; Wang, F.; Sjölund-Karlsson, M.; McDermott, P.F. Antimicrobial resistance in campylobacter: Susceptibility testing methods and resistance trends. J. Microbiol. Methods. 2013, 95, 57–67. [CrossRef] [PubMed] 9. Endtz, H.P.; Ruijs, G.J.; van Klingeren, B.; Jansen, W.H.; van der Reyden, T.; Mouton, R.P. Quinolone resistance in Campylobacter isolated from man and poultry following the introduction of ﬂuoroquinolones in veterinary medicine. J. Antimicrob. Chemother. 1991, 27, 199–208. [CrossRef] [PubMed] y 10. Sam, W.I.C.; Lyons, M.M.; Waghorn, D.J. Increasing rates of ciproﬂoxacin resistant Campylobacter. J. Clin. Pathol. 1999, 52, 709–710. [CrossRef] [PubMed] 11. Luangtongkum, T.; Jeon, B.; Han, J.; Plummer, P.; Logue, C.M.; Zhang, Q. Antibiotic resistance in Campylobacter: Emergence, transmission and persistence. Future Microbiol. 2009, 4, 189–200. [CrossRef] [PubMed] 12. Blaser, M.J.; Engberg, J. Clinical aspects of Campylobacter jejuni and Campylobacter coli infections. In Campylobacter; Nachamkin, I., Szymanski, C.M., Blaser, M.J., Eds.; ASM Press: Washington, DC, USA, 2008; pp. 99–121. pp 13. 4. Conclusions Combined molecular protocols (such as 16S rRNA PCR, DNA sequencing, and Multilocus Sequence Typing (MLST) analysis) revealed the successful identiﬁcation of C. jejuni strains from stool and from blood cultures, even in patients where traditional culture protocols failed [33–36]. These results demonstrate the potential of molecular methods in improving the diagnosis of bacterial infections caused by C. jejuni. Numerous PCR-based techniques (real-time PCR and pyrosequencing) have also been developed for the rapid detection and identiﬁcation of bacteria in clinical blood specimens [37]. Commercially available real-time PCR for the direct detection of bacteria in blood has been introduced [38], but the use of these tools has not become routine in clinical microbiology laboratories. Indeed, molecular techniques are rather costly, and require people with high levels of technical expertise, and therefore these techniques are consequently not suitable for routine identiﬁcation, particularly in institutes with limited ﬁnancial resources or in developing countries. Moreover, the high sensitivity of PCR-based methods and DNA sequencing that have the potential to detect all bacterial DNA present in a clinical sample may cause serious problems in clinical interpretation. Background levels of bacterial DNA might be detected in the blood of patients in the absence of any signs of bacteremia [39]. Matrix Assisted Laser Desorption Ionization Time-Of-Flight (MALDI-TOF) MS is a reliable tool for a rapid, precise, and cost-effective classiﬁcation of a broad spectrum of bacteria and yeast [40]. MALDI-TOF MS analysis was in complete agreement with molecular tests identifying C. jejuni and C. coli [41]. Moreover, changes in protein biomarkers, such as those caused by an amino acid substitution, have been used to differentiate between C. jejuni ssp. jejuni and subsp. doylei, and to assess phylogenetic relationships in different isolates [42]. Several studies have evaluated the contribution of MALDI-TOF MS towards identifying microorganisms in positive blood culture [40,43]. Results showed that MALDI-TOF MS accurately identiﬁed blood-borne organisms in more than 80% of cases. Nevertheless, the ability of MALDI-TOF MS to correctly identify microorganisms in blood cultures clearly depends on the bacteria concentration [44,45]. Novel application of MALDI-TOF MS has increased its potential in the detection of blood-borne organisms and thus may allow faster bacterial identiﬁcation than the conventional automated blood cultures systems in the near future [46]. However, the efﬁcacy of MALDI-TOS MS technology in reducing the time for identifying positive blood cultures, particularly for slow growing bacteria such as C. jejuni, remains to be evaluated. Int. J. 4. Conclusions Mol. Sci. 2016, 17, 544 7 of 9 Since individuals with hematological disorders, especially lymphoid neoplasms, have a high risk for infection, the close cooperation between the hematologist, infectious disease specialist, and microbiologist can be of primary importance in providing a timely and effective intervention. Acknowledgments: This work was supported by L’Associazione Nazionale Contro le Infezioni Ospedaliere (L’ANCIO). We would like to thank Tania Merlino, who kindly edited the English language used in our manuscript. Author Contributions: Maria Teresa Gallo; Enea Gino Di Domenico; Luigi Toma; Francesco Marchesi; Lorella Pelagalli; Nicola Manghisi; Fiorentina Ascenzioni; Grazia Prignano; Andrea Mengarelli; Fabrizio Ensoli participated in the study conception and design; Maria Teresa Gallo, Grazia Prignano, Nicola Manghisi contributed to the acquisition of literature data; Enea Gino Di Domenico, Luigi Toma, Francesco Marchesi, Andrea Mengarelli, Fabrizio Ensoli drafted the manuscript. All authors read and approved the ﬁnal manuscript. Conﬂicts of Interest: The authors declare no conﬂict of interest. [CrossRef] [PubMed] 16. Blaser, M.J. Campylobacter jejuni and related species. In Mandell, Douglas and Bennett’s Principles and Practice of Infectious Diseases, 6th ed.; Mandell, G.L., Bennett, J.E., Dolin, R., Eds.; Churchill Livingstone: New York, NY, USA, 2006; pp. 2548–2557. 17. Okada, H.; Kitazawa, T.; Harada, S.; Itoyama, S.; Hatakeyama, S.; Ota, Y.; Koike, K. Combined treatment with oral kanamycin and parenteral antibiotics for a case of persistent bacteremia and intestinal carriage with Campylobacter coli. Intern. Med. 2008, 47, 1363–1366. [CrossRef] [PubMed] 18. Baden, L.R.; Bensinger, W.; Angarone, M.; Casper, C.; Dubberke, E.R.; Freifeld, A.G.; Garzon, R.; Greene, J.N.; Greer, J.P.; Ito, J.I.; et al. Prevention and treatment of cancer-related infections. J. Natl. Compr. Cancer Netw. 2012, 10, 1412–1445. 19. Morroka, T.; Oda, T. In vitro evaluation of antibiotics for treatment of meningitis caused by Campylobacter fetus subsp fetus. Pediatr. Infect. Dis. J. 1989, 8, 653–654. [CrossRef] 20. Engberg, J.; Neimann, J.; Nielsen, E.M.; Aerestrup, F.M.; Fussing, V. Quinolone-resistant Campylobacter infections: Risk factors and clinical consequences. Emerg. Infect. Dis. 2004, 10, 1056–1063. [CrossRef] [PubMed] 21. Phillips, R.; Hancock, B.; Graham, J.; Bromham, N.; Jin, H.; Berendse, S. Prevention and management of neutropenic sepsis in patients with cancer: Summary of NICE guidance. BMJ 2012, 345, e5368. [CrossRef] [PubMed] 22. Lachance, N.; Gaudreau, C.; Lamothe, F.; Lariviere, L. Role of the b-lactamase of Campylobacter jejuni in resistance to b-lactam agents. Antimicrob. Agents Chemother. 1991, 35, 813–818. [CrossRef] [PubMed] 23. Tajada, P.; Gomez-Garces, J.L.; Alos, J.I.; Balas, D.; Cogollos, R. Antimicrobial susceptibilities of Campylobacter jejuni and Campylobacter coli to 12 b-lactam agents and combination with b-lactamase inhibitors. Antimicrob. Agents Chemother. 1996, 40, 1924–1925. [PubMed] g 24. Tremblay, C.; Gaudreau, C.; Lorange, M. Epidemiology and antimicrobial susceptibilities of 111 Campylobacter fetus subsp. fetus strains isolated in Quebec, Canada, from 1983 to 2000. J. Clin. Microbiol. 2003, 41, 463–466. [PubMed] 25. Wang, W.L.; Blaser, M.J. Detection of pathogenic Campylobacter species in blood culture systems. J. Clin. Microbiol. 1986, 23, 709–714. [PubMed] 26. Louwen, R.; van Baarlen, P.; van Vliet, A.H.; van Belkum, A.; Hays, J.P.; Endtz, H.P. Campylobacter bacteremia: A rare and under-reported event? Eur. J. Microbiol. Immunol. (Bp). 2012, 2, 76–87. [CrossRef] [PubMed] 27. Martiny, D.; Dediste, A.; Debruyne, L.; Vlaes, L.; Haddou, N.B.; Vandamme, P.; Vandenberg, O. Accuracy of the API Campy system, the Vitek 2 Neisseria-Haemophilus card and matrix-assisted laser desorption ionization time-of-ﬂight mass spectrometry for the identiﬁcation of Campylobacter and related organisms. Clin. Microbiol. References Luangtongkum, T.; Shen, Z.; Seng, V.W.; Sahin, O.; Jeon, B.; Liu, P.; Zhang, Q. Impaired ﬁtness and transmission of macrolide-resistant Campylobacter jejuni in its natural host. Antimicrob. Agents Chemother. 2012, 56, 1300–1308. [CrossRef] [PubMed] Int. J. Mol. Sci. 2016, 17, 544 8 of 9 14. Ladely, S.R.; Harrison, M.A.; Fedorka-Cray, P.J.; Berrang, M.E.; Englen, M.D.; Meinersmann, R.J. Development of macrolide-resistant Campylobacter in broilers administered subtherapeutic or therapeutic concentrations of tylosin. J. Food Prot. 2007, 70, 1945–1951. [PubMed] 15. Lin, J.; Yan, M.; Sahin, O.; Pereira, S.; Chang, Y.J.; Zhang, Q. Effect of macrolide usage on emergence of erythromycin-resistant Campylobacter isolates in chickens. Antimicrob. Agents Chemother. 2007, 51, 1678–1686. [CrossRef] [PubMed] [CrossRef] [PubMed] Infect. 2011, 17, 1001–1006. [CrossRef] [PubMed] 28. Callahan, C.; Greene, J.N.; Sandin, R.L.; Ruge, D.; Johnson, J. Campylobacter Jejuni Bacteremia in an HIV-Positive Patient with Non-Hodgkin’s Lymphoma. Cancer Control. 1998, 5, 357–360. [PubMed] 9. Schattner, A. Campylobacter jejuni and cytopenias. Am. J. Med. 2013, 126, 1020–1021. [CrossRef] [PubM 30. Fernández-Cruz, A.; Muñoz, P.; Mohedano, R.; Valerio, M.; Marín, M.; Alcalá, L.; Rodriguez-Créixems, M.; Cercenado, E.; Bouza, E. Campylobacter bacteremia: Clinical characteristics, incidence, and outcome over 23 years. Medicine (Baltimore) 2010, 89, 319–330. [CrossRef] [PubMed] 31. Dummer, R.; Asagoe, K.; Cozzio, A.; Burg, G.; Doebbeling, U.; Golling, P.; Fujii, K.; Urosevic, M. Recent advances in cutaneous lymphomas. J. Dermatol. Sci. 2007, 48, 157–167. [CrossRef] [PubMed] 32. Steele, R.W. Managing infection in cancer patients and other immunocompromised children. Ochsner J. 2012, 12, 202–210. [PubMed] 33. Dingle, K.E.; Colles, F.M.; Wareing, D.R.A.; Ure, R.; Fox, A.J.; Bolton, F.E.; Bootsma, H.J.; Willems, R.J.; Urwin, R.; Maiden, M.C. Multilocus sequence typing system for Campylobacter jejuni. J. Clin. Microbiol. 2001, 39, 14–23. [CrossRef] [PubMed] 9 of 9 Int. J. Mol. Sci. 2016, 17, 544 34. Dingle, K.E.; Colles, F.M.; Ure, R.; Wagenaar, J.A.; Duim, B.; Bolton, F.J.; Fox, A.J.; Wareing, D.R.; Maiden, M.C. Molecular characterization of Campylobacter jejuni clones: A basis for epidemiologic investigation. Emerg. Infect. Dis. 2002, 8, 949–955. [CrossRef] [PubMed] 35. Morris, G.A.; Ikumapayi, U.N.; Antonio, M.; Howie, S.R.; Adegbola, R.A. A novel Campylobacter jejuni sequence type from a culture-negative patient in the Gambia. PLoS ONE 2008, 12, e1773. [CrossRef] [PubMed] 36. Bessede, E.; Delcamp, A.; Sifre, E.; Buissonniere, A.; Megraud, F. New methods for detection of Campylobacters in stool samples in comparison to culture. J. Clin. Microbiol. 2011, 49, 941–944. [CrossRef] [PubMed] 37. Jordan, J.A.; Jones-Laughner, J.; Durso, M.B. Utility of pyrosequencing in identifying bacteria directly from positive blood culture bottles. J. Clin. Microbiol. 2009, 47, 368–372. [CrossRef] [PubMed] 38. Gaibani, P.; Rossini, G.; Ambretti, S.; Gelsomino, F.; Pierro, A.M.; Varani, S.; Paolucci, M.; Landini, M.P.; Sambri, V. Blood culture systems: Rapid detection—How and why? Int. J. Antimicrob. Agents 2009, 34, S13–S15. [CrossRef] 39. McLaughlin, R.W.; Vali, H.; Lau, P.C.; Palfree, R.G.; de Ciccio, A.; Sirois, M.; Ahmad, D.; Villemur, R.; Desrosiers, M.; Chan, E.C. Are there naturally occurring pleomorphic bacteria in the blood of healthy humans? J. Clin. Microbiol. 2002, 40, 4771–4775. [CrossRef] [PubMed] 40. Drancourt, M. Detection of microorganisms in blood specimens using MALDI-TOF mass spectrometry: A review. Clin. Microbiol. Infect. [CrossRef] [PubMed] 2010, 16, 1620–1625. [CrossRef] [PubMed] 41. Kolinska, R.; Drevinek, M.; Jakubu, V.; Zemlickova, H. Species identiﬁcation of Campylobacter jejuni ssp. jejuni and C. coli by matrix-assisted laser desorption/ionization time-of-ﬂight mass spectrometry and PCR. Folia Microbiol. 2008, 53, 403–409. 42. Fagerquist, C.; Bates, A.; Heath, S.; King, B.C.; Garbus, B.R.; Harden, L.A.; Miller, W.G. Sub-speciating Campylobacter jejuni by proteomic analysis of its protein biomarkers and their posttranslational modiﬁcations. J. Proteome Res. 2006, 5, 2527–2538. [CrossRef] [PubMed] 43. Carbonnelle, E.; Mesquita, C.; Bille, E.; Day, N.; Dauphin, B.; Beretti, J.L.; Ferroni, A.; Gutmann, L.; Nassif, X. MALDI-TOF mass spectrometry tools for bacterial identiﬁcation in clinical microbiology laboratory. Clin. Biochem. 2011, 44, 104–109. [CrossRef] [PubMed] 44. Ferroni, A.; Suarez, S.; Beretti, J.L.; Dauphin, B.; Bille, E.; Meyer, J.; Bougnoux, M.E.; Alanio, A.; Berche, P.; Nassif, X. Real time identiﬁcation of bacteria and yeast in positive blood culture broths by MALDI-TOF-mass spectrometry. J. Clin. Microbiol. 2010, 48, 1542–1548. [CrossRef] [PubMed] 45. Christner, M.; Rohde, H.; Wolters, M.; Sobottka, I.; Wegscheider, K.; Aepfelbacher, M. Rapid identiﬁcation of bacteria from positive blood culture bottles using MALDI-TOF mass spectrometry ﬁngerprinting. J. Clin. Microbiol. 2010, 48, 1584–1591. [CrossRef] [PubMed] 46. Idelevich, E.A.; Grünastel, B.; Peters, G.; Becker, K. Direct blood culturing on solid medium outperforms an automated continuously monitored broth-based blood culture system in terms of time to identiﬁcation and susceptibility testing. New Microbes New Infect. 2016, 10, 19–24. [CrossRef] [PubMed] © 2016 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC-BY) license (http://creativecommons.org/licenses/by/4.0/). © 2016 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC-BY) license (http://creativecommons.org/licenses/by/4.0/). © 2016 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC-BY) license (http://creativecommons.org/licenses/by/4.0/).
https://openalex.org/W3158344372	https://art.torvergata.it/bitstream/2108/275769/1/Richards.pdf	English	null	Nowcasting COVID‐19 incidence indicators during the Italian first outbreak	Statistics in medicine	2,021	cc-by	15,819	Correspondence Correspondence Pierfrancesco Alaimo Di Loro, Department of Statistical Sciences, University of Rome “La Sapienza”, Rome, Lazio 00185, Italy. Email: pierfrancesco.alaimodiloro@uniroma1.it K E Y W O R D S 6Department of GEPLI, Libera Universitá Maria Ss Assunta, Rome, Italy COVID-19, growth curves, Richards’ equation, SARS-CoV-2 COVID-19, growth curves, Richards’ equation, SARS-CoV-2 7Department of Mathematics, University of Bergen, Bergen, Norway 8IAC - CNR, Institute of Applied Computing “M. Picone”, Rome, Italy Received: 24 October 2020 Revised: 8 March 2021 Accepted: 8 April 2021 Received: 24 October 2020 Revised: 8 March 2021 Accepted: 8 April 2021 DOI: 10.1002/sim.9004 Revised: 8 March 2021 Accepted: 8 April 2021 Received: 24 October 2020 DOI: 10.1002/sim.9004 R E S E A R C H A R T I C L E Nowcasting COVID-19 incidence indicators during the Italian first outbreak Pierfrancesco Alaimo Di Loro1 Fabio Divino2 Alessio Farcomeni3 Giovanna Jona Lasinio1 Gianfranco Lovison4,5 Antonello Maruotti6,7 Marco Mingione1,8 Pierfrancesco Alaimo Di Loro1 Fabio Divino2 Alessio Farcomeni3 Giovanna Jona Lasinio1 Gianfranco Lovison4,5 Antonello Maruotti6,7 Marco Mingione1,8 1Department of Statistical Sciences, University of Rome “La Sapienza”, Rome, Italy 2Department of Bio-Sciences, University of Molise, Campobasso, Italy 3Department of Economics and Finance, University of Rome “Tor Vergata”, Rome, Italy 4Department of Economics, Management and Statistics, University of Palermo, Palermo, Italy 5Department of Epidemiology and Public Health, Swiss TPH Basel, Basel, Switzerland 6Department of GEPLI, Libera Universitá Maria Ss Assunta, Rome, Italy 7Department of Mathematics, University of Bergen, Bergen, Norway 8IAC - CNR, Institute of Applied Computing “M. Picone”, Rome, Italy Correspondence Pierfrancesco Alaimo Di Loro, Department of Statistical Sciences, University of Rome “La Sapienza”, Rome, Lazio 00185, Italy. Email: pierfrancesco.alaimodiloro@uniroma1.it 1Department of Statistical Sciences, University of Rome “La Sapienza”, Rome, Italy 1Department of Statistical Sciences, University of Rome “La Sapienza”, Rome, Italy A novel parametric regression model is proposed to fit incidence data typically collected during epidemics. The proposal is motivated by real-time monitoring and short-term forecasting of the main epidemiological indicators within the first outbreak of COVID-19 in Italy. Accurate short-term predictions, includ- ing the potential effect of exogenous or external variables are provided. This ensures to accurately predict important characteristics of the epidemic (e.g., peak time and height), allowing for a better allocation of health resources over time. Parameter estimation is carried out in a maximum likelihood framework. All computational details required to reproduce the approach and replicate the results are provided. 2Department of Bio-Sciences, University of Molise, Campobasso, Italy 3Department of Economics and Finance, University of Rome “Tor Vergata”, Rome, Italy 4Department of Economics, Management and Statistics, University of Palermo, Palermo, Italy K E Y W O R D S COVID-19, growth curves, Richards’ equation, SARS-CoV-2 K E Y W O R D S COVID-19, growth curves, Richards’ equation, SARS-CoV-2 © 2021 The Authors. Statistics in Medicine published by John Wiley & Sons Ltd. 2 10% of those infected.3 During the outbreak, it was crucial to set up appropriate data collection and modeling systems quickly. Both were necessary for monitoring, evaluation of policy interventions, and prediction. 10% of those infected.3 During the outbreak, it was crucial to set up appropriate data collection and modeling systems quickly. Both were necessary for monitoring, evaluation of policy interventions, and prediction. Generally speaking, the nature of epidemics’ spread has nearly always followed the same scenario: first, the growth in the number of infected people is (close to) exponential; in a second moment, this growth gradually but consistently slows down as an effect, for instance, of various containment measures. This pattern can cyclically recur until the outbreak is tamed. So far, in order to explain the spread of epidemics and predict their consequences, a number of mathematical and sta- tistical models of different complexity levels have been used. The starting point is often the Verhulst logistic equation,4 which can easily capture both the exponential increase in the number of infected people at the initial stage of the epi- demic development, and the tendency towards a constant value by its ending. In more complex models, people are divided into different groups: (S) the susceptible class, namely those individuals who are capable of contracting the dis- ease and becoming infected; (I) the infected class, namely those individuals who are capable of transmitting the disease to others; (R) the removed class, namely infected individuals who are deceased or have recovered, who are either perma- nently immune or isolated. This group of mathematical models are called SIR (or compartmental) models.5 References include,6-9 and several more. However, whilst being potentially very appropriate to model the dynamics underlying any epidemic, SIR-based models rely on accurate initial estimates of several quantities governing its spreading mechanism (which are unknown). Poor data input on key features of the pandemic can heavily bias these estimates, jeopardizing the reliability of any theory-based forecasting effort. SIR models are microsimulation models and we believe that, gen- erally speaking, they should be used mostly for “scenario evaluation” rather than predicting future outcomes. Indeed, they rely on several speculations and strict theoretical assumptions, not necessarily met by the analyzed data and, espe- cially during the first stage of the outbreak, failed in predicting various COVID-19 related outcomes.10 Such specifics lead the choice of coefficients in the equations defining the SIR model and define its initial conditions. It is well known that even a slight change in those can lead to large differences in the final results. For instance, at the beginning of the epi- demic, early data providing estimates for case fatality rate, infection fatality rate, basic reproductive number, and other key numbers that are essential for the modeling, are often inflated and may cause potentially large overestimation of the epidemic severity. Similar criticism to using compartmental modeling for nowcasting can also be found in Reference 11, and references therein. Hence, we have preferred to follow an alternative approach, which involved direct modeling of the observed counts.12 This encompasses the use of phenomenological models without detailed mechanistic foundations, but which have the advantage of allowing simple calibrations to the empirical reported data. Such approaches are partic- ularly suitable when substantial uncertainty tarnishes the epidemiology of an infectious disease, including the potential contribution of multiple transmission pathways. In these situations, phenomenological models provide a starting point for obtaining early estimates of the transmission potential and short-term forecasts of the epidemic evolution.13 g We propose a parametric regression model for the modeling of incidence indicators (defined in Section 2.1) based on the use of the Richards’ curve (a generalized logistic function) in place of the widely used exponential or polynomial trends. Furthermore, we replace the generally entrenched Gaussian assumption for the distribution of log-counts14,15 by the more appropriate Poisson or Negative Binomial distributions for counts. In this way we avoid the implausible assumptions stemming from the more common alternatives: the former allows the underlying counts to potentially grow indefinitely; the latter neglects the proper specification of dependence between mean and variance under the log-normal distribution. We further propose different ways of including the effect of exogenous information on the response function of counts, in an extended generalized linear model framework. These models have been implemented during the outbreak with the aim of modeling the medium to long term evolution of the epidemic wave. 1 INTRODUCTION Italy has been the first European country to be severely hit by the first epidemic wave due to the spread of the SARS-CoV-2 virus. COVID-19 syndrome emerged in northern Italy in February 2020, with a basic reproduction number R0 between 2.5 and 4.1 In its most severe form, COVID-19 has two challenging characteristics:2 it is highly infectious and, despite having a benign course in the vast majority of patients, it requires hospital admission and even intensive care for about This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2021 The Authors Statistics in Medicine published by John Wiley & Sons Ltd Italy has been the first European country to be severely hit by the first epidemic wave due to the spread of the SARS-CoV-2 virus. COVID-19 syndrome emerged in northern Italy in February 2020, with a basic reproduction number R0 between 2.5 and 4.1 In its most severe form, COVID-19 has two challenging characteristics:2 it is highly infectious and, despite having a benign course in the vast majority of patients, it requires hospital admission and even intensive care for about wileyonlinelibrary.com/journal/sim Statistics in Medicine. 2021;1–22. 1 ALAIMO DI LORO et al. 2.1 Incidence and prevalence indicators: different mathematical features The epidemiological data provided by CPD can be distinguished into two basic types: 1. incidence indicators (flows) 1. incidence indicators (flows) 2. prevalence indicators (stocks) The use of logistic-based curves is also widely discussed in the literature.16-18 Logistic growth curves can be seen as a flexible formulation for approximating a large variety of growth phenomena, especially in biology and in epidemiology.19-23 In particular, highly flexible parametric models such as Gompertz curves and the unified Richards’ family24 have been proposed in the study of organisms’ growth, for a review see Reference 25. The article is organized as follows: Section 2 gives a detailed description of the Italian situation and provides a brief account of the Italian public data made available daily, with some remarks on limitations and flaws in the data collection process; Section 3 contains a description of our approach to modeling incidence indicators, including remarks on how to obtain standard errors for parameters and predictions performances; Section 4 illustrates results of our approach applied to the incidence indicators recorded during first wave of the Italian outbreak of COVID-19. Finally, results are discussed and commented along with some concluding remarks in Section 5. The methods discussed in this article have also been implemented in a Shiny app, publicly available at https:// statgroup19.shinyapps.io/StatGroup19-Eng/. in this article have also been implemented in a Shiny app, publicly available at https:// tatGroup19-Eng/. ALAIMO DI LORO et al. 3 2 AVAILABLE DATA AND THEIR LIMITATIONS The Italian Civil Protection Department (CPD), starting from February 24th, 2020, has been gathering data at the regional level every day and making these public in a GitHub repository. During most of the Italian epidemic, data were com- mented by the department head in an official press release at about 6 pm. The daily updated data are currently stored at https://github.com/pcm-dpc/COVID-19. For public health service purposes, Italy is divided into 21 regions. There are 19 administrative regions, plus two autonomous provinces (Trento and Bolzano) that form the administrative region of Trentino-Alto-Adige. In the sequel, we focus on modeling the indicators aggregated at the national level. Nevertheless, the supplementary material contains graphical and quantitative performances of our model on the 21 single regions. 2.1.1 Incidence indicators (B) F I G U R E 1 Time series the Italian daily incidence indicators: daily positives, A an daily deceased, B (B) F I G U R E 1 Time series of the Italian daily incidence indicators: daily positives, A and daily deceased, B F I G U R E 1 Time series of the Italian daily incidence indicators: daily positives, A and daily deceased, B (A) (B) (B) F I G U R E 2 Time serie the Italian cumulative incide indicators: cumulative positiv A and cumulative deceased, B (B) (A) F I G U R E 2 Time series of the Italian cumulative incidence indicators: cumulative positives, A and cumulative deceased, B F I G U R E 2 Time series of the Italian cumulative incidence indicators: cumulative positives, A and cumulative deceased, B (A) (B) By their nature of cumulative counts, these data series are necessarily monotonically nondecreasing (see Figure 2 for the positives and deceased example). By their nature of cumulative counts, these data series are necessarily monotonically nondecreasing (see Figure 2 for the positives and deceased example). 2.1.1 Incidence indicators Incidence indicators measure the number of individuals with a particular condition, related with the epidemic, recorded during a given period. They can be referred to different time periods; in particular, in the CPD dataset, daily incidence counts are available for the following indicators: • positives, which are subclassified into two subconditions: – hospitalized (either in regular wards or in ICU) – isolated-at-home • deceased • recovered/discharged • recovered/discharged These indicators can be considered, by analogy with the terminology used in econometrics, as flow data, quantifying the daily input (positives) and output (deceased and recovered/discharged) of the system. The time series of daily positives and daily deceased aggregated at the national level are shown in Figure 1. From the viewpoint of the following modeling effort, one important feature of these indicators is that they can be referred to longer time intervals, simply cumulating them over time. The most interesting cumulative incidence indicators are those referring to the whole history of the pan- demic, computed from a conventional date of “beginning of the pandemic” (typically, the day the systematic recording of daily positives began) to the current day: • cumulative positives • cumulative deceased • cumulative recovered/discharged In particular, given Y 0 = 0, we can build the whole series of cumulative counts conditionally on the value of the cumulative indicator at time (t −1), and the incidence indicators at time t, for each t = 1, … , T: Y c t = Y c t−1 + It, where Y c t represents the cumulative indicator and It represents the inputs in the system, for example: cumulative positives at time t are the cumulative positives at time (t −1) plus the daily positives at day t. 4 ALAIMO DI LORO et al. (A) (B) F I G U R E 1 Time series of the Italian daily incidence indicators: daily positives, A and daily deceased, B (A) (B) F I G U R E 2 Time series of the Italian cumulative incidence indicators: cumulative positives, A and cumulative deceased, B By their nature of cumulative counts, these data series are necessarily monotonically nondecreasing (see Figure 2 for the positives and deceased example). 4 (A) ALAIMO DI LORO et al. 2.1.2 Prevalence indicators Prevalence indicators measure the number of individuals with a particular condition, related with the epidemic, at a given instant in time (or at a given short interval of time, eg, a day). They are typically obtained from simple algebra from other indicators; in particular, in the CPD dataset, the following indicators are available daily: • current positives; • current intensive care units (ICU) occupancy. These indicators result from the balance between total inputs and outputs of the system, for example: current positives are the difference between cumulative positives and cumulative deceased plus recovered/discharged. Again, by analogy with the terminology used in econometrics, they can be considered as stock data. In particular, given Y 0 = 0, we can build the whole series conditionally on the value of the prevalence indicator at time (t −1), and the incidence indicators at time t, for each t = 1, … , T: Y p t = Y p t−1 + It −Ot, ALAIMO DI LORO et al. 5 F I G U R E 3 Time series of Italian daily prevalence indicators: current positive, A and ICU occupancy, B (A) (B) where Y p t represents the prevalence indicator, It represents the inputs in the system and Ot represents the outputs, for example: current positives at time t are the current positives at time (t −1) plus the daily positives at day t and minus the sum of deceased and discharged recovered at day t. However, given the different delay in reporting the various information by the regional agencies, there exists a relevant temporal misalignment among all the quantities reported at the daily scale. Therefore, the simultaneous consideration of all these flows may be significantly flawed and we rather prefer modeling the indicators individually. Two important features of these indicators are that: 5 (B) ALAIMO DI LORO et al. 2.1.2 Prevalence indicators by their own nature, these indicators are not monotone, since they can increase or decrease as a result of different trends of the component series. Typically, we expect the series of current positives and ICU occupancy to increase in the rising phase of an epidemic, reach a peak and then decrease to a lower asymptote (see Figure 3), although more complex patterns due to resurgence of the epidemic are also plausible. Prevalence indicators are characterized by a strong and tangled dependence structure which is cumbersome to simplify into a manageable and useful statistical model on the short run. For this reason, the focus of this work concerns only incidence indicators. Our model proposal, from a strictly mathe- matical point of view, could potentially applied also on prevalence indicators. However, from the statistical point of view, the modeling assumptions which are assumed to hold (with good approximation) considering the incidence indicators, are likely to be strongly violated by prevalence indicators and the resulting outcome cannot be considered reliable. A brief discussion about some possible approaches for the analysis of prevalence indicators is given in Section 5. 2.1.2 Prevalence indicators 5 Time series of evalence rent positive, upancy, B (A) F I G U R E 3 Time series of Italian daily prevalence indicators: current positive, A and ICU occupancy, B F I G U R E 3 Time series of Italian daily prevalence indicators: current positive, A and ICU occupancy, B (B) (A) where Y p t represents the prevalence indicator, It represents the inputs in the system and Ot represents the outputs, for example: current positives at time t are the current positives at time (t −1) plus the daily positives at day t and minus the sum of deceased and discharged recovered at day t. However, given the different delay in reporting the various information by the regional agencies, there exists a relevant temporal misalignment among all the quantities reported at the daily scale. Therefore, the simultaneous consideration of all these flows may be significantly flawed and we rather prefer modeling the indicators individually. Two important features of these indicators are that: where Y p t represents the prevalence indicator, It represents the inputs in the system and Ot represents the outputs, for example: current positives at time t are the current positives at time (t −1) plus the daily positives at day t and minus the sum of deceased and discharged recovered at day t. However, given the different delay in reporting the various information by the regional agencies, there exists a relevant temporal misalignment among all the quantities reported at the daily scale. Therefore, the simultaneous consideration of all these flows may be significantly flawed and we rather prefer modeling the indicators individually. Two important features of these indicators are that: 1. given their stock nature, they cannot be aggregated (eg,: it does not make sense to compute “cumulative current positives”); 1. given their stock nature, they cannot be aggregated (eg,: it does not make sense to compute “cumulative current positives”); 2. by their own nature, these indicators are not monotone, since they can increase or decrease as a result of different trends of the component series. Typically, we expect the series of current positives and ICU occupancy to increase in the rising phase of an epidemic, reach a peak and then decrease to a lower asymptote (see Figure 3), although more complex patterns due to resurgence of the epidemic are also plausible. 2. could be even weeks before. Positive status is also counted on the day that test results are received, with swabs being processed from one day to weeks after symptoms’ onset. No distinction between actively symptomatic and asymptomatic patients was made. Swabs and positive cases are not time-aligned. For example, in countries like Singapore (https://www.moh.gov.sg/ covid-19), daily data include information on total swabs tested, total unique persons swabbed as well as total swabs per 1 000 000 total population and total unique persons swabbed per 1 000 000 total population. In Italy, up to April 19th, 2020, only the total number of daily swabs is available, and no linkage between swabs and tested individuals was kept in the data repository. Hence, it is impossible to make statistically sound use of swabs’ count to model the whole first pandemic wave. Finally, it is crucial to recall that people diagnosed with COVID-19 disease are only a small fraction of the people infected by the virus. Moreover, since the tracking was highly symptoms driven, especially in the first phase of the out- break, the detected number of positives cases can provide only a partial estimate of the true incidence of COVID-19 in the Italian population. Eventually, we expect this detected fraction to vary wildly over space and time. In our opinion, the most reliable indicator is the count of ICU occupancy. The reason is that the Italian Society for Emergency Care issued national guidelines (that did not change substantially during the epidemic) for testing patients with a suspected infection by SARS-CoV-2, who also had top priority for swab access and reporting; and ICU admissions can be expected to depend on the proportion of infected population susceptible to severe infection, rather than on the regional strategy for testing and contact tracing. However, while probably reliable, this indicator also presents some draw- backs. First of all, it provides only a partial snapshot of the epidemic’s current stage, which concerns the most severe cases of the disease. The latter is a critical issue, especially in the COVID-19 case, which is known to present severe symptoms only in a small percentage of the currently affected individuals. Second, this snapshot is affected by a constant delay (ie, the time between catching the disease and manifesting severe symptoms). As mentioned in Section 2.1.2, its daily vari- ation is obtained as a combination of new incoming patients (+) and the deceased or recovered ones (-), whose effects blend and are hard to disentangle. As a consequence, incidence indicators, such as daily positives and daily deceased, while being measured with some error and even more delay in the case of deaths, still represent the critical indicators for timely and appropriate monitoring of the pandemic. 3 MODEL SPECIFICATION The time series of any of the observed indicators, denoted by z = {zt}T t=t0, is modeled separately and considered as the realization of the stochastic process Z = {Zt}T t=t0. The idea behind this article is to model any of the mentioned indicators through a Generalized Model with a response function E[Zt] = 𝜇(t) = g−1(t; 𝜽), where g(⋅) is a known link function and 𝜽is a parameter vector, that is appropriate for the specific mathematical features of the epidemic process. This must be coupled with a response distribution f(Zt; 𝛉) coherent with the domain of such indicators, which are counts and therefore Natural numbers. 2.2 Data issues COVID-19 public Italian data present several issues that severely affect their quality. The information has been gathered and reported at a regional level, and each regional healthcare organization has a different transmission and data collection system. Measurement errors, and errors in data entry, are expected to be often present. Delays in reporting has been, sometimes, substantial. Some patients were transferred (eg, from Lombardia to Puglia, and even to Germany) without notification, and they were counted as hospital patients of the receiving region (or not at all when sent abroad) and posi- tive cases of the region of residence. Most importantly, counts were updated on the notification day rather than aligned to a more appropriate date. For example, death is counted on the day of the reporting, not on the day of the outcome, which see https://www.epiprev.it/materiali/2020/EP2-3/112_edit1.pdf for further details. ALAIMO DI LORO et al. 3.1 Response function for incidence indicators F I G U R E 4 Example of Richards’ curve, A and derivative of the Richards’ curve, B 0.0 0.3 0.6 λ 0.9 0 1 2 3 4 5 t (A) t F I G U R E 4 Example of Richards’ curve, A and derivative of the Richards’ curve, B F I G U R E 4 Example of Richards’ curve, A and derivative of the Richards’ curve, B (B) can be expressed as: can be expressed as: E[Y c t ] = g−1(t; 𝜸) = 𝜆𝜸(t) = b + r (1 + 10h(p−t))s . (1) E[Y c t ] = g−1(t; 𝜸) = 𝜆𝜸(t) = b + r (1 + 10h(p−t))s . (1) b ∈R+ represents a lower asymptote and r > 0 is the distance between the upper and the lower asymptote, hence b + r would be the final epidemic size; h is known as the hill, and represents the infection/growth rate; p ∈R represents a lag-phase of the trajectory and determines the peak position (it tells when the curve growth speed slows down); s ∈R is an asymmetry parameter regulating differences in the behavior of the ascending and descending phase of the outbreak. In our context, since cumulative incidences are always monotone increasing indicators, it is reasonable to assume h, s > 0 †. An extensive review of the Richards’ curve and other logistic growth models, together with discussion on the proper interpretation of the parameters, is given in Reference 24. An extensive review of the Richards’ curve and other logistic growth models, together with discussion on the proper interpretation of the parameters, is given in Reference 24. An Extended Generalized Linear Model with (1) as response function seems to be a natural choice for modeling time series of cumulative counts, whose monotonically nondecreasing average behaves as the Richards’ curve. Unfortunately, there is a significant drawback to this choice. As it will be better clarified in Section 3.2, a very useful working assumption would be that all these counts were stochastically independent, given their mean function 𝜆𝜸(t). However, we cannot consider this assumption as realistic in the case of cumulative counts, since the constraint on the domain of definition on subsequent counts (ie, yc t ≥yc 𝜏, ∀𝜏< t) is not guaranteed to be satisfied. †Conversely, we may assume h, s < 0 3.1 Response function for incidence indicators Let us denote by {yc t}T t=0 the time-series of cumulative incidence indicators since the start of the epidemic (t0 = 0, first day of systematic data recording). Visual inspection of these indicators in Figure 2 suggests that their expected values follow a logistic-type growth curve. Different example of logistic curves have been proposed in the literature, all representing solutions to specific differential equations that model the spread of epidemics.26-28 Differently from the more standard exponential models, these are able to describe the slowdown of the outbreak associated with a decaying transmission rate just after the number of cases approaches its inflection point. They have been already widely used to describe the evolution of the COVID-19 pandemic in different states during its early to medium stage.29,30 Here, for all the incidence indicators, we consider the Generalized Logistic Function, also known as Richards’ curve (see Figure 4 as an example), as response function for the mean of the process.31 This curve was widely used to describe various biological processes,32 but has been recently adapted also in epidemiology for real-time prediction of outbreak of diseases.33-35 The specialty of the Richards’ curve lies in its ability to describe a great variety of growing processes, endowed with strong flexibility, that includes as special cases the standard logistic growth curve,36 the Gompertz growth curve37 and others. It can be expressed in different forms.38-41 One of its most general formulation depends on the vector of five parameters 𝜸⊤= [b, r, h, p, s] and ALAIMO DI LORO et al. 7 F I G U R E 4 Example of Richards’ curve, A and derivative of the Richards’ curve, B 0.0 0.3 0.6 λ λ 0.9 0 1 2 3 4 5 t (A) (B) t 0.00 0.02 0.04 0.06 0 1 2 3 4 5 t t ~ can be expressed as: E[Y c t ] = g−1(t; 𝜸) = 𝜆𝜸(t) = b + r (1 + 10h(p−t))s . (1) 7 λ (B) 0.00 0.02 0.04 0.06 0 1 2 3 4 5 t t ~ ALAIMO DI LORO et al. 8 ALAIMO DI LORO et al. In addition, we may also consider adding a kink effect/baseline 𝛼to the first differences ̃𝜆𝜸(⋅), which is to say assuming the following functional form for the mean of the daily counts: ̃𝜇𝜽(t) = 𝛼+ ̃𝜆𝜸(t), 𝛼≥0, (2) ̃𝜇𝜽(t) = 𝛼+ ̃𝜆𝜸(t), 𝛼≥0, (2) ̃𝜇𝜽(t) = 𝛼+ ̃𝜆𝜸(t), 𝛼≥0, (2) where 𝜽= (𝛼, 𝜸). This would correspond to the following mean function for the cumulative counts: orrespond to the following mean function for the cumulative counts: 𝜇𝜽(t) = 𝛼⋅(t −1) + 𝜆𝜸(t). 𝜇𝜽(t) = 𝛼⋅(t −1) + 𝜆𝜸(t). In practice, the parameter 𝛼includes the possibility of having a strictly positive baseline rate, which can be interpreted as the endemic steady state incidence rate. This is in line with the current perspective that SARS-CoV-2 might not be completely eradicated within the next few years.42 On the other hand, the first differences of the Richards’ curve ̃𝜆𝜸(t) are (by construction) forced to decrease asymptotically to the value of 0. However, this asymptotic result is not necessarily observed in real data. In particular, Figure 1 highlights that both time-series do not attain the 0 value, but settle to a low, constant level. This situation may, potentially, continue indefinitely: new cases will be found as long as people will be tested. Consequently, the model without a baseline lacks the ability to catch this tail and, because of the curve parametric form, this may indirectly affect the fit on the whole series. In practice, the parameter 𝛼includes the possibility of having a strictly positive baseline rate, which can be interpreted as the endemic steady state incidence rate. This is in line with the current perspective that SARS-CoV-2 might not be completely eradicated within the next few years.42 On the other hand, the first differences of the Richards’ curve ̃𝜆𝜸(t) are (by construction) forced to decrease asymptotically to the value of 0. However, this asymptotic result is not necessarily observed in real data. In particular, Figure 1 highlights that both time-series do not attain the 0 value, but settle to a low, constant level. This situation may, potentially, continue indefinitely: new cases will be found as long as people will be tested. Consequently, the model without a baseline lacks the ability to catch this tail and, because of the curve parametric form, this may indirectly affect the fit on the whole series. 8 In the first instance, one solution would be to fit the model, including the kink effect 𝛼. Afterward, if it is estimated not to be sensibly different from 0, the model without 𝛼can be fitted again to stabilize the estimation procedure and decrease the uncertainty on the other parameters. 3.2 Response distribution for incidence indicators Before introducing the distributions for the daily incidence counts, we must make some assumptions about the time dependence structure. In particular, we assume that given the mean function ̃𝜇𝜽(t), the daily incidence counts Y t are stochastically independent from the previous cumulative counts: Yt⊥Y c 𝜏∀𝜏< t. We denote this hypothesis of indepen- dence by HI. We also assume the value of the first cumulative count Y c 0 = yc 0 to be known and fixed. Exploiting HI, we can express the joint density of all the subsequent cumulative counts conditional on Y c 0 = yc 0 as the product of the univariate densities of the corresponding daily counts {Yt}T t=1. The equivalence follows from the following conditional argument: fY c 1, … ,Y c T(yc 1, … , yc T\|yc 0; 𝜃) = T ∏ t=1 fY c t (yc t\|yc 0, … , yc t−1; 𝜃) = T ∏ t=1 fY c t (yt + yc t−1\|yc 0, … , yc t−1; 𝜃) = = T ∏ t=1 fYt(yt\|yc 0, … , yc t−1; 𝜃) HI= T ∏ t=1 fYt(yt\|𝜃), where the second identity is justified in the light of Y c t = Yt + Y c t−1, t = 1, … , T, which is true by definition. From a practical point of view, this also implies a first-order Markov property for the cumulative counts: where the second identity is justified in the light of Y c t = Yt + Y c t−1, t = 1, … , T, which is true by definition. From a practical point of view, this also implies a first-order Markov property for the cumulative counts: Y c t \|Y c t−1⊥Y c 1, … , Y c t−2, t = 1, … , T Y c t \|Y c t−1⊥Y c 1, … , Y c t−2, t = 1, … , T and mutual independence between the daily counts: and mutual independence between the daily counts: Yt⊥Y𝜏, ∀t, 𝜏, t ≠𝜏. We remark that although these independence structure is just an approximation in the present case, this kind of approach has provided valid inference for all the available Italian incidence indicators. We remark that although these independence structure is just an approximation in the present case, this kind of approach has provided valid inference for all the available Italian incidence indicators. 3.1 Response function for incidence indicators On the other hand, the stochastic independence assumption sounds more reasonable, albeit not necessarily true, for the daily incidence counts {yt}T t=1, that is, the addenda of the cumulative counts excluding the starting point y0, which can be defined as: yc t = t∑ 𝜏=0 y𝜏 ⇒ yt = yc t −yc t−1, t = 1, … , T, where y0 = 0 by definition. where y0 = 0 by definition. y0 y Using Equation (1), and exploiting the additive properties of the expected value, we have: Using Equation (1), and exploiting the additive properties of the expected value, we have: ̃𝜇(t) = E[Yt] = E[Y c t ] −E[Y c t−1] = 𝜆𝜸(t) −𝜆𝜸(t −1) = = r ⋅[(1 + 10h(p−t))−s −(1 + 10h[p−(t−1)])−s] = ̃𝜆𝜸(t) which, in particular, does not depend on the baseline b. Therefore, we shall adopt an extended Generalized Model with response function given by the first differences of the Richards’ curve ̃𝝀𝜸= { ̃𝜆𝜸(t)}T t=1 to model the daily expected values 𝝁= {𝜇(t)}T t=1 of the observed incidence counts y = {yt}T t=1 (see example in Figure 4). which, in particular, does not depend on the baseline b. Therefore, we shall adopt an extended Generalized Model with response function given by the first differences of the Richards’ curve ̃𝝀𝜸= { ̃𝜆𝜸(t)}T t=1 to model the daily expected values 𝝁= {𝜇(t)}T t=1 of the observed incidence counts y = {yt}T t=1 (see example in Figure 4). †Conversely, we may assume h, s < 0 3.2.1 Poisson distribution or of daily incidence counts, y = {y1, … , yt}, is composed of independent Poisson realizations Let us assume that the vector of daily incidence counts, y = {y1, … , yt}, is composed of independent Poisson realizations with expected value ̃𝜇𝜽(t): Yt\|𝜽∼Pois( ̃𝜇𝜽(t)), t = 1, … , T. Hence, the likelihood can be written as: Hence, the likelihood can be written as: Hence, the likelihood can be written as: (𝜽\|y) = T ∏ t=1 Pois(yt\| ̃𝜇𝜽(t)) ∝ ∝̃𝜇𝜽(t) ∑T t=1yt ⋅exp { − T ∑ t=1 ̃𝜇𝜽(t) } and the log-likelihood is given by: l(𝜃\|y) = log (𝜽\|y) ∝ T ∑ t=1 yt log( ̃𝜇𝜽(t)) − T ∑ t=1 ̃𝜇𝜽(t). (𝜽\|y) = T ∏ t=1 Pois(yt\| ̃𝜇𝜽(t)) ∝ ∝̃𝜇𝜽(t) ∑T t=1yt ⋅exp { − T ∑ t=1 ̃𝜇𝜽(t) } and the log-likelihood is given by: and the log-likelihood is given by: l(𝜃\|y) = log (𝜽\|y) ∝ T ∑ t=1 yt log( ̃𝜇𝜽(t)) − T ∑ t=1 ̃𝜇𝜽(t). Remark that, under the assumption of Poisson distribution and the baseline 𝛼= 0 (ie, ̃𝜇(𝛼,𝜸) = ̃𝜆𝜸(⋅)), we can exploit the well-known Poisson’s additive property ‡ to conclude that each cumulative count Y c t is still marginally distributed according to a Poisson, parameterized by the original Richards’ curve function 𝜆𝜸(⋅): Y c t \|𝜸∼Pois ( t∑ 𝜏=1 ̃𝜆𝜸(𝜏) ) = Pois(𝜆𝜸(t)). 3.2 Response distribution for incidence indicators For communication purposes, it can be of interest to report the results of analyses and predictions in terms of cumu- lative, rather than daily, incidence indicators. Clearly, it is possible to model and predict the daily incidence indicators and, from these estimates and predictions, obtain the relevant cumulative incidence indicators. ALAIMO DI LORO et al. 9 3.3 Response function depending on covariates The trend of any of the considered indicators may also depend on additional exogenous information, which we may assume to be known a priori either because it is immutable (ie, the day of the week), or because policymakers fixed it (daily number of tested cases/swabs set by the government). For instance: one might want to correct for possible weekly seasonality, which is known to affect the daily positives series since many laboratories are closed during the weekend and cannot evaluate swabs. The latter can be used to disentangle the underlying trend of the epidemic from the obvious positive correlation between tested cases and daily positives. In general, we may want to include the effect of any set of k time-varying covariates XT×(k+1) = [x(t)]T t=1 in the Richards’ framework through the usual linear predictor 𝜂(X) = X𝜷, where 𝜷is a k + 1-dimensional vector of real valued parameters (including intercept). Let us denote the mean function of the considered indicator as ̃𝜇𝜃(t) = E[Yt], where 𝜽= (𝛼, 𝜸, 𝜷). In order to respect the positivity of the mean parameter (which is necessary both in the Poisson and in the Negative Binomial case), we consider the link function g(⋅) = log(⋅), so that the effect on the mean is expressed as: 𝜇𝜷(X) = exp{𝜂(X)} = exp{X𝜷}. Considering a single time point t, we would get the following functional form: Considering a single time point t, we would get the following functional form: 𝜇𝜷(x(t)) = exp{x(t)𝜷}. The mean term of our model shall take into account both the effect of the covariates through 𝜇𝜷(⋅) and the temporal behavior induced by the Richards’ curve 𝜆𝜸(⋅). As a matter of fact, these two components may be combined in different ways. We considered two alternative specifications denoted in the sequel as: additive and multiplicative. 3.3.1 Additive inclusion of covariates The inclusion of an additive effect of covariates implies that the effect of every covariate is constant through-out the pandemic, notwithstanding the current contagion level: for instance, one may think that an increase of daily tested cases will always produce the same increase of daily daily positives. If that is the case, we may just express the baseline parameter 𝛼at each time-point t as the linked linear combination of covariates 𝜇𝜷(x(t)) = exp{x(t)𝜷}, which would produce the following mean function: ̃𝜇𝜃(t) = 𝜇𝜷(x(t)) + ̃𝜆𝜸(t). On the whole vector of observations, this can be expressed as ̃𝝁𝜽= 𝜇𝜷(X) + ̃𝝀𝜸. 10 The Negative Binomial does not satisfy the same additive property as the Poisson, hence we cannot draw the same conclusion reached in the Poisson case about the marginal distribution of the cumulative count Y c t when 𝛼= 0. In gen- eral, the cumulative count in the NB case will follow the distribution stemming from the sum of independent Negative Binomial r.v. with common dispersion parameter 𝜈but different means ̃𝝁= { ̃𝜆𝜸(t)}T t=1. 3.2.2 Negative Binomial distribution When counts are overdispersed the Poisson distribution is not a suitable choice. We can model the observed daily inci- dence counts y = {y1, … , yt} as independent realizations from a Negative Binomial with mean ̃𝜆𝜸(t) and dispersion parameter 𝜈∈R+: Yt\|𝜽∼NB( ̃𝜇𝜽(t), 𝜈), t = 1, … , T. Hence, the likelihood can be written as: Hence, the likelihood can be written as: (𝜽, 𝜈\|d) = T ∏ t=1 NB(yt\| ̃𝜇𝜽(t), 𝜈) ∝ ∝ T ∏ t=1 [Γ(𝜈+ yt) Γ(𝜈) ( 𝜈 𝜈+ ̃𝜇𝜽(t) )𝜈( ̃𝜇𝜽(t) 𝜈+ ̃𝜇𝜽(t) )yt] (𝜽, 𝜈\|d) = T ∏ t=1 NB(yt\| ̃𝜇𝜽(t), 𝜈) ∝ ∝ T ∏ t=1 [Γ(𝜈+ yt) Γ(𝜈) ( 𝜈 𝜈+ ̃𝜇𝜽(t) )𝜈( ̃𝜇𝜽(t) 𝜈+ ̃𝜇𝜽(t) )yt] and the log-likelihood is: l(𝜃, 𝜈\|y) = log (𝜽, 𝜈\|y) ∝ T ∑ t=1 log (Γ(𝜈+ yt) Γ(𝜈) ) + 𝜈 T ∑ t=1 log ( 𝜈 𝜈+ ̃𝜇𝜽(t) ) T ( ) (𝜽, 𝜈\|d) = T ∏ t=1 NB(yt\| ̃𝜇𝜽(t), 𝜈) ∝ ∝ T ∏ t=1 [Γ(𝜈+ yt) Γ(𝜈) ( 𝜈 𝜈+ ̃𝜇𝜽(t) )𝜈( ̃𝜇𝜽(t) 𝜈+ ̃𝜇𝜽(t) )yt] and the log-likelihood is: l(𝜃, 𝜈\|y) = log (𝜽, 𝜈\|y) ∝ T ∑ t=1 log (Γ(𝜈+ yt) Γ(𝜈) ) + 𝜈 T ∑ t=1 log ( 𝜈 𝜈+ ̃𝜇𝜽(t) ) + T ∑ t=1 yt log ( ̃𝜇𝜽(t) ̃𝜇𝜽(t) + 𝜈 ) . ‡the sum of independent Poissons is still a Poisson with parameter the sum of the parameters ‡the sum of independent Poissons is still a Poisson with parameter the sum of the parameters 10 ALAIMO DI LORO et al. 3.3.2 Multiplicative inclusion of covariates The inclusion of a multiplicative effect of covariates would imply that the more serious the pandemic situation, the more severe the impact of any covariate on the indicators’ daily rate. p y y First, let us recall that in Section 2.1.1 we computed the first differences of the Richards’ curve function as: ̃𝜆𝜸(t) = r ⋅[(1 + 10h(p−t))−s −(1 + 10h[p−(t−1)])−s] = r ⋅̃𝜆𝜸,−r(t). On the log-scale, it would return the more familiar: log( ̃𝜆𝜸(t)) = log(r) + log( ̃𝜆𝜸,−r(t)). (3) (3) (3) log( ̃𝜆𝜸(t)) = log(r) + log( ̃𝜆𝜸,−r(t)). ALAIMO DI LORO et al. 11 From Equation (3), it comes natural the idea of expressing log(r) at each time-point t as the linear combination of covariates 𝜂(x(t)) as in a Generalized Poisson model with log link function. Indeed this provides a multiplicative effect of the covariates, where the parameter r can be expressed as 𝜇𝜷(⋅) in the following way: r𝜷(x(t)) = 𝜇𝜷(x(t)) = exp{x(t)𝜷}. (4) r𝜷(x(t)) = 𝜇𝜷(x(t)) = exp{x(t)𝜷}. (4) (4) r𝜷(x(t)) = 𝜇𝜷(x(t)) = exp{x(t)𝜷}. Note that the constant r is still present and included in Equation (4) through the intercept 𝛽0. Therefore, the mean at time t is expressed as: ̃𝜇𝜃(t) = 𝛼+ r𝜷(x(t)) ⋅̃𝜆𝜸,−r(t). Considering the whole vector of observations, we would have the following vector of means ̃𝝁𝜽= 𝜶+ r𝜷(X) ⋅̃𝝀𝜸,−r, where 𝜶= 𝛼⋅1T. Considering the whole vector of observations, we would have the following vector of means ̃𝝁𝜽= 𝜶+ r𝜷(X) ⋅̃𝝀𝜸,−r, where 𝜶= 𝛼⋅1T. 3.4 Model estimation Parameters can be estimated by maximizing the log-likelihood l(𝜽\|y), where 𝜽in this case includes all the parameters the likelihood depends on (eg, includes 𝜈in the Negative Binomial case). This optimization problem does not have an analytical solution, and numerical maximization must be used. To improve computation, we derived analytical expres- sions for the gradient and Hessian of the two possible log-likelihoods (ie, Poisson or Negative Binomial counts), making Fisher-scoring iteration very fast. The expressions are reported in the supplementary material. Given the nonsmooth shape of the objective function, we are at risk of being trapped by local maxima of the log-likelihood, depending on the initial conditions. Therefore, in order to strengthen the optimization procedure, a multistart procedure based on a combination of genetic and gradient descent algorithms has been used.43,44 Once an approximate point of maximum ̂𝜽has been obtained, we could theoretically obtain an estimate of the asymp- totic variance-covariance matrix of the estimated parameters through inverse of the negative log-likelihood Hessian in ̂𝜽 (which corresponds to the Observed Fisher Information): ̂V𝜽= −H(l(̂𝜽\|y))−1, where H denotes the Hessian matrix. Nevertheless, we may want to account for the potential misspecification of our model potentially arising from the independence assumption HI in Section 3.2. Therefore, we resort to a robust approach for estimating the standard errors and covariance structure associated with the parameter vector 𝜽. In particular, we consider the Huber Sandwich Estimator of the variance-covariance matrix,45,46 that can be computed as: where H denotes the Hessian matrix. Nevertheless, we may want to account for the potential misspecification of our model potentially arising from the independence assumption HI in Section 3.2. Therefore, we resort to a robust approach for estimating the standard errors and covariance structure associated with the parameter vector 𝜽. In particular, we consider the Huber Sandwich Estimator of the variance-covariance matrix,45,46 that can be computed as: ̂V R 𝜽= (−H(l(̂𝜽\|y))−1)∇l(̂𝜽\|y)∇l(̂𝜽\|y)⊤(−H(l(̂𝜽\|y))−1), ̂V R 𝜽= (−H(l(̂𝜽\|y))−1)∇l(̂𝜽\|y)∇l(̂𝜽\|y)⊤(−H(l(̂𝜽\|y))−1), ̂V R 𝜽= (−H(l(̂𝜽\|y))−1)∇l(̂𝜽\|y)∇l(̂𝜽\|y)⊤(−H(l(̂𝜽\|y))−1), where ∇l(̂𝜽\|y) represents the gradient of the log-likelihood in the point of maximum. Interval estimates for the parameters are directly derived through the asymptotic distribution of the Maximum Likelihood Estimator, with the corresponding robust covariance matrix ̂𝜽∼(𝜽, ̂V R 𝜽). A similar theoretical result for predictions is not as straightforward. Therefore, these are derived through a parametric double bootstrap procedure,47-49 which accounts for both the uncertainty of parameter estimation and the randomness of the observations. In practice, resampled trajectories {Yi}B i=1 are obtained by simulating B sets of parameters from their asymptotic distribution and computing B mean functions trajectories {𝜇𝜽i(t)}B i=1. An artificial time series of counts is then simulated for each of the B trajectories and 95% confidence intervals are obtained by computing the pointwise 2.5% and 97.5% quantiles. The dispersion parameter 𝜈, being a poorly identifi- able nuisance parameter of no impact on the mean curve behavior, has been excluded from the bootstrapping procedure and kept fixed at its estimated value ̂𝜈. Diagnostic check on the model has been performed through the Pearson residuals and the Deviance residuals. Computation of the former is trivial, where we recall their definition as: ̂𝜌t = yt −̂yt ̂ Var[Yt] , t = 1, … , T. ALAIMO DI LORO et al. 12 and Negative Binomial assumptions we have: Under the Poisson and Negative Binomial assumptions we have: Under the Poisson and Negative Binomial assumptions we have: ̂ VarPoi[Yt] = 𝜇̂𝜽(t), ̂ VarNB[Yt] = 𝜇̂𝜽(t) + 𝜇̂𝜽(t)2 ̂𝜈 , (5) (5) (5) respectively. The Deviance Residuals are instead defined as the individual contributions of each observation to the Deviance of the model, that is, the discrepancy between the proposed model and the full model (perfect fit) fits in terms of log-likelihood: ̂dt = 2 ⋅[log(f(yt\|̂𝜽s) −log(f(yt\|̂𝜽)], where f(⋅\|⋅) is the chosen distribution function and ̂𝜽s is the parameter vector of the saturated model. For the Poisson and Negative Binomial this can be computed as: where f(⋅\|⋅) is the chosen distribution function and ̂𝜽s is the parameter vector of the saturated model. ̂V R 𝜽= (−H(l(̂𝜽\|y))−1)∇l(̂𝜽\|y)∇l(̂𝜽\|y)⊤(−H(l(̂𝜽\|y))−1), For the Poisson and Negative Binomial this can be computed as: ̂d Poi t = sgn(yt −𝜇̂𝜽(t)) ⋅ √ 2yt log ( yt 𝜇̂𝜽 ) −(yt −𝜇̂𝜽(t)), ̂d NB t = sgn(yt −𝜇̂𝜽(t)) ⋅ √ 2 [ yt log ( yt 𝜇̂𝜽(t) ) −(yt + 𝜈) ⋅log ( yt + 𝜈 𝜇̂𝜽(t) + 𝜈 )] , respectively.50 If the model correctly describes the variability in the data, then both the Pearson residuals and the Deviance residuals are expected to be Normally distributed and independent, with the latter being generally more robust to outliers. 3.5 Validation Fitting performances are further evaluated through numerical metrics such as the pseudo-R2 and coverage of the 95% prediction intervals: R2 = 1 −MSE 𝜎2 y = 1 − ∑T t=1 (yt −̂yt)2 ∑T t=1 (yt −y)2 , Cov95% = 1 T ⋅ T ∑ t=1 I(̂yl t;̂yu t )(yt), where MSE is the Mean Squared Error, y = 1 T ∑T t=1 yt and I(⋅) denotes the indicator function over the set . where MSE is the Mean Squared Error, y = 1 T ∑T t=1 yt and I(⋅) denotes the indicator function over the set . 4 NOWCASTING THE ITALIAN OUTBREAK OF COVID-19 For the sake of brevity, here we present results referred to the proposed Richards’ growth model only for daily positives aggregated at the national level. Further results of the model performances for daily deceased are included in the supple- mentary material. We only present results obtained adopting the Negative Binomial distribution because of the substantial overdispersion present at all levels for these indicators (spatially and temporally heterogeneous data collection process, varying containment measures, and so on). We first show the fitted curve for each indicator, and compare its shape with the observed time series. We also calculate the residuals and check if model assumptions under the proposed framework hold. According to the results drawn from the residual analysis, we modify the empirical setting, keeping the theoretical one fixed, to better capture specific data features. Later on, we show the performance for two fundamental issues: (i) predicting the epidemic trend in advance and (ii) predicting the date of the peak of the epidemic. 3.6 Step-ahead predictions We test our model’s ability to predict the evolution of the epidemic (at least its first wave) from the short to the medium term. Indeed, while the choice of a rigid parametric form for the mean function is penalizing in terms of flexibility and fitting ability, it allows for extrapolation outside the observed domain and is supposed to provide robust forecasts (at least in the short/medium term). Therefore, using the best model for the two indicators (ie, baseline + week-day additive effect), we calculated the out-of-sample root mean squared prediction error (RMSPE) for: • different fitting windows t = 1, … , ̃t; • different forecast horizons, say K ∈{1, 5, 10, 15}. We recall that, given the fitting window set 1, … , ̃t: RMSPẼt,K = √ √ √ √1 K K ∑ j=1 (ỹt+j −̂ỹt+j)2. 13 ALAIMO DI LORO et al. 13 Index Model without baseline Model with baseline log-likelihood −1081.4 −982.8 AIC 2152.7 1953.6 BIC 2162.3 1965 AICc 2137.8 1935.7 13 TA B L E 1 Log-likelihood, AIC, BIC, and AICc for the model without baseline and the model with baseline, on daily positives TA B L E 2 Parameters’ points estimates and 95% confidence intervals for the model with baseline on daily positives TA B L E 2 Parameters’ points estimates and 95% confidence intervals for the model with baseline on daily positives Parameter Point estimate 95% Interval 𝛼 173.17 (103.2, 290.54) r 222.95 × 103 (220.56 × 103, 225.36 × 103) h 0.0288 (0.0285, 0.0291) p −31.18 (−32.75, −29.62) s 72.54 (48.29, 96.79) 𝜈 18.73 (17.77, 19.73) 4.1 Model on daily positives 14 (B) Cumulative 0K 50K 100K 150K 200K 250K 01−Mar 01−Apr 01−May 01−Jun 01−Jul 01−Aug µt F I G U R E 5 Bootstrapped trajectories corresponding to the Huber Sandwich covariance matrix in the point of maximum for the model with baseline on daily positives viceversa). Finally, 𝜈is an overdispersion parameter and does not present any evident communicable interpretation. The larger it is and the lower the overdispersion, according to the formula in Equation (5). viceversa). Finally, 𝜈is an overdispersion parameter and does not present any evident communicable interpretation. The larger it is and the lower the overdispersion, according to the formula in Equation (5). We here want to stress the fact that the uncertainty characterizing some of the parameters (like s) is not alarming. In particular, variations of s at values distant from 1 have very little effect on the curve shape. Furthermore, the parameter vector presents a covariance structure that highlights how different combination of parameters can yield similar curves. Indeed, simulating M = 5000 set of parameters from the Normal distribution with variance corresponding to the covari- ance underlying the Huber Sandwich covariance matrix, we get the set of difference and cumulative curves represented in Figure 5. g We can also directly obtain point predictions {̂yt}T t=1 as: point predictions {̂yt}T t=1 as: We can also directly obtain point predictions {̂yt}T t=1 as: (6) ̂yt = 𝜇̂𝜽(t), t = 1, … , T, (6) and prediction intervals {(̂yl t; ̂yu t )}T t=1 through the same set of bootstrapped trajectories, whose statistical validity relies on the asymptotic properties introduced in Section 3.4. Figure 6 shows the model fit on the whole available time series of counts: the former on the daily series, the latter on the cumulative one. We can see how the estimated curve does catch the observed general behavior, providing a smooth approximation only marginally influenced by extreme values. Our model produces an R2 = 0.941 and coverage Cov95% = 0.945, meaning that the percentage of observed daily counts falling inside the estimated bounds is perfectly coherent with the specified confidence level. Looking at Figure 6, we notice how daily counts boundaries get smaller as time passes, due to the implicit relationship between mean and variance that characterizes count distributions. At the same time, the opposite happens to the bounds on the cumulative counts. 4.1 Model on daily positives To decide whether or not to include the kink effect, we fitted the model with and without the baseline 𝛼and compared the two fits in terms of log-likelihood, AIC, BIC, and corrected AIC (AICc). The values are presented in Table 1 and pro- vide clear evidence in favor of the model with baseline (ie, with mean 𝜇𝜽(⋅) as in Equation (2)). Parameters’ estimates of the model ̂𝜽and the respective 95% confidence intervals are shown in Table 2, where the baseline 𝛼is estimated to be ̂𝛼= 173.17, with interval (103.2, 290.54), which confirms that the baseline is estimated to be significantly different from 0, and it should be included in the model. As explained in Section 3.1, this parameter represents the long-term endemic incidence rate that may (possibly indefinitely) follow the end of the main outbreaks. This obviously would hold exactly with constant social interactions, containment measures, control of cases, and so on. Hence, in the considered time hori- zon, we expect this endemic level to be of ≈173 daily positives per day. When the baseline is included, the parameter r does not indicate anymore the final epidemic size, but only the final outbreak size. This is the number of positive cases due to the uncontrolled outbreak, additional to what would have been observed in the steady endemic state. This amount is estimated to be ≈222 950, an amount that would have been reached in ≈1288 days at the endemic state level. The parame- ters h, p, and s do not have an easily quantifiable and absolute interpretation, but are useful for comparison. As explained in Section 3.1, the first indicates how fast the infection spreads, the second how soon it starts descending (lag-phase) and the last asymmetries between the ascending and descending phase (s < 1 the ascending is slower than the descending and 14 ALAIMO DI LORO et al. 0K 2K 4K 01−Mar 01−Apr 01−May 01−Ju (A) First differences (B) Cumulative n 01−Jul 01−Aug µt 0K 50K 100K 150K 200K 250K 01−Mar 01−Apr 01−May 01−Jun 01−Jul 01−Aug µt F I G U R E 5 Bootstrapped trajectories corresponding to the Huber Sandwich covariance matrix in the point of maximum for the model with baseline on daily positives ALAIMO DI LORO et al. 4.1 Model on daily positives The latter is not surprising: indeed, they are built marginally on all the epidemic’s possible scenarios. Therefore, they give us a clear sight of what we could have currently observed, keeping into account and aggregating the uncertainty at each stage of the epidemic. We performed a diagnostic check on both the Pearson and the Deviance residuals. The plots in Figure 7 show the Deviance residuals behavior: histogram (A), including the P-value from the Shapiro test; Normal qq-plot (B); autocorrelation plot (C); plot of the residuals vs fitted values (D). The first two check the (approximated) Normality assumption on the residuals, while the second two control for the correlation of the residuals (among them and with the observed values). 4.1.1 Weekly seasonality The diagnostic check on both type of residuals showed that the Normality assumption is not rejected, but the correla- tion plot manifests undesirable patterns (see Figure 7). In particular, the autocorrelation between errors is larger at lag 7 ALAIMO DI LORO et al. 15 F I G U R E 6 Observed (black dots) and fitted values (gray solid lines) with 95% confidence intervals (gray dashed lines) for the model with baseline on daily positives (A) (B) F I G U R E 7 Deviance residuals for the model with baseline on daily positives (A) (B) (C) (D) (and multiples of this). We can interpret this outcome as the presence of an intense weekly seasonality (especially dur- ing/after the weekend). This suggests people would rather not come forward for testing on the weekend or, alternatively, ALAIMO DI LORO et al. 15 (B) (A) (A) (B) F I G U R E 7 Deviance residuals for the model with baseline on daily positives (A) (B) (C) (D) (and multiples of this). We can interpret this outcome as the presence of an intense weekly seasonality (especially dur- ing/after the weekend). This suggests people would rather not come forward for testing on the weekend or, alternatively, the system has less capacity at the weekend, meaning it is more challenging to get a test. Undeniably, viruses work 7 days a week. Looking at the low numbers during/after the weekend might give you a false sense of reassurance. This is because it seems cases are down, and therefore elimination of the virus is possible. But, unlike our population, viruses do work on weekends. This may be adjusted by simply adding a weekday effect in our model as a covariate, using the approach in Section 3.3. Such effect may be included either in an additive or a multiplicative fashion. At first, we considered effects for each day of the week, taking Monday as a corner point. Preliminary results showed that not all week-days present a significant deviation from the common mean. On the other hand, the distribution of the Deviance residuals ̂dt of the stan- dard model aggregated by week-day (see Figure 8) shows that an evident overestimation pattern (ie, negative deviations) is taking place on Monday and Tuesday. 4.1.1 Weekly seasonality −2 0 2 Monday Tuesday Wednesday Thursday Friday Saturday Sunday d^ F I G U R E 8 Deviance residuals distribution aggregated by day of the week for daily positives Index Additive effect Multiplicative effect log-likelihood −971.74 −974.1 AIC 1929.48 1934.3 AICc 1942.67 1947.5 BIC 1908.60 1913.4 TA B L E 3 Log-likelihood, AIC, BIC, and AICc for the models with baseline including additive or multiplicative week-day effect on daily positives (A) (B) F I G U R E 9 Observed (black dots) and fitted values (gray solid lines) with 95% confidence intervals (gray dashed lines) for the model with baseline and week-day additive effect, estimated on the daily positives may suffer from week seasonality. The additive option is chosen over its alternative because of its lower/improved AIC, BIC, and AICc score (see Table 3). The resulting fit of the model with week seasonality on the observed data are shown in Figure 9, where, on the left, we show the fitted curve and the 95% confidence intervals; on the right, we can observe the fit on the cumulative indicator. Estimated parameters are shown in Table 4 This model estimates the baseline to be at e ̂𝛽0 = 192 48 on Wednesday to 16 ALAIMO DI LORO et al. −2 0 2 Monday Tuesday Wednesday Thursday Friday Saturday Sunday d^ F I G U R E 8 Deviance residuals distribution aggregated by day of the week for daily positives Index Additive effect Multiplicative effect log-likelihood −971.74 −974.1 TA B L E 3 Log-likelihood, AIC, BIC, and AICc for the models with baseline including additive or multiplicative week-day effect on daily positives 16 −2 0 2 Monday Tuesday Wednesday Thursday Friday Saturday Sunday d^ ALAIMO DI LORO et al. 4.1.1 Weekly seasonality 16 F I G U R E 8 Deviance residuals distribution aggregated by day of the week for daily positives F I G U R E 8 Deviance residuals distribution aggregated by day of the week for daily positives TA B L E 3 Log-likelihood, AIC, BIC, and AICc for the models with baseline including additive or multiplicative week-day effect on daily positives Index Additive effect Multiplicative effect log-likelihood −971.74 −974.1 AIC 1929.48 1934.3 AICc 1942.67 1947.5 BIC 1908.60 1913.4 (A) (B) F I G U R E 9 Observed (black dots) and fitted values (gray solid lines) with 95% confidence intervals (gray dashed lines) for the model with baseline and week-day additive effect, estimated on the daily positives may suffer from week seasonality. The additive option is chosen over its alternative because of its lower/improved AIC, BIC, and AICc score (see Table 3). (B) F I G U R E 9 Observed (black dots) and fitted values (gray solid lines) with 95% confidence intervals (gray dashed lines) for the model with baseline and week-day additive effect, estimated on the daily positives (A) (A) (B) may suffer from week seasonality. The additive option is chosen over its alternative because of its lower/improved AIC, BIC, and AICc score (see Table 3). h l i fi f h d l i h k li h b d d h i i h h l f may suffer from week seasonality. The additive option is chosen over its alternative because of its lower/improved AIC, BIC, and AICc score (see Table 3). The resulting fit of the model with week seasonality on the observed data are shown in Figure 9, where, on the left, we show the fitted curve and the 95% confidence intervals; on the right, we can observe the fit on the cumulative indicator. ̂ Estimated parameters are shown in Table 4. This model estimates the baseline to be at e ̂𝛽0 = 192.48 on Wednesday to Sunday and at e ̂𝛽0+ ̂𝛽wd = 121.51 on Mondays and Tuesdays. Deriving the corresponding 95% intervals, these two baselines result significantly different from the estimate of the overall baseline ̂𝛼in the model without covariates. 4.1.1 Weekly seasonality (A) (B) F I G U R E 7 Deviance residuals for the model with baseline on daily positives F I G U R E 7 Deviance residuals for the model with baseline on daily positives (B) (A) (A) (B) (D) (C) (C) (D) (and multiples of this). We can interpret this outcome as the presence of an intense weekly seasonality (especially dur- ing/after the weekend). This suggests people would rather not come forward for testing on the weekend or, alternatively, the system has less capacity at the weekend, meaning it is more challenging to get a test. Undeniably, viruses work 7 days a week. Looking at the low numbers during/after the weekend might give you a false sense of reassurance. This is because it seems cases are down, and therefore elimination of the virus is possible. But, unlike our population, viruses do work on weekends. This may be adjusted by simply adding a weekday effect in our model as a covariate, using the approach in Section 3.3. Such effect may be included either in an additive or a multiplicative fashion. At first, we considered effects for each day of the week, taking Monday as a corner point. Preliminary results showed that not all week-days present a significant deviation from the common mean. On the other hand, the distribution of the Deviance residuals ̂dt of the stan- dard model aggregated by week-day (see Figure 8) shows that an evident overestimation pattern (ie, negative deviations) is taking place on Monday and Tuesday. f Therefore, in the sequel, we will present only results obtained with the dichotomous variable that is equal to 1 when- ever the week-day is Monday or Tuesday (0 vice versa). Note that lower tests effort during the weekend shows in the data on Monday and Tuesday, since daily reports involve mostly results received the day before, with swabs therefore dating back 48 hours on the day of publication. This confirms that working with daily data require special care as the cases reporting 16 ALAIMO DI LORO et al. 4.1.1 Weekly seasonality The diagnostic check of the Pearson’s and Deviance residuals showed that adherence to Gaussianity improved and the correlation pattern at lag 7 is still present but mitigated (see Figure 10 for the Deviance F I G U R E 10 Deviance residuals for the model with baseline and week-day additive effect estimated on daily positives (A) (B) (A) (D) (C) (C) (D) covariates, denoting less overdispersion with respect to the equivariance hypothesis. This is completely reasonable since the week-day effect is able to explain some of the previously unaccounted heterogeneity. In terms of model validation, the inclusion of this effect improves sensibly the R2 (0.956), while the average cover- age Cov95% is constant (0.950). The diagnostic check of the Pearson’s and Deviance residuals showed that adherence to Gaussianity improved and the correlation pattern at lag 7 is still present but mitigated (see Figure 10 for the Deviance residuals). 4.1.1 Weekly seasonality The estimates of the outbreak size ̂r and of the infection rate ̂h of the two models are in agreement, while the point estimates of the asymmetry parameter ̂s are different but both large and mutually included in the corresponding 95% intervals. This is reasonable since we would not expect the outbreak size, rate and symmetry to vary after accounting for week-day het- erogeneity. On the other hand, the new estimate ̂p of p detects a shorter lag-phase and hence a slightly faster approach to the descending phase. Finally, the estimate of the dispersion parameter ̂𝜈is slightly larger than in the model without ALAIMO DI LORO et al. 17 ALAIMO DI LORO et al. 17 TA B L E 4 Parameters’ point estimates and 95% confidence intervals for the additive model on daily positives Parameter Point estimate 95% Interval 𝛽0 5.26 (5.18, 5.34) 𝛽wd −0.46 (−0.53, −0.38) r 224.57 × 103 (224.13 × 103, 225.01 × 103) h 0.0289 (0.0287, 0.0291) p −23.26 (−29.64, −16.88) s 44.42 (−35.67, 124.51) 𝜈 22.01 (21.35, 22.70) F I G U R E 10 Deviance residuals for the model with baseline and week-day additive effect estimated on daily positives (A) (B) (C) (D) TA B L E 4 Parameters’ point estimates and 95% confidence intervals for the additive model on daily positives Parameter Point estimate 95% Interval 𝛽0 5.26 (5.18, 5.34) 𝛽wd −0.46 (−0.53, −0.38) r 224.57 × 103 (224.13 × 103, 225.01 × 103) h 0.0289 (0.0287, 0.0291) p −23.26 (−29.64, −16.88) s 44.42 (−35.67, 124.51) 𝜈 22.01 (21.35, 22.70) s 44.42 (−35.67, 124.51) 𝜈 22.01 (21.35, 22.70) F I G U R E 10 Deviance residuals for the model with baseline and week-day additive effect estimated on daily positives (A) (B) (C) (D) covariates, denoting less overdispersion with respect to the equivariance hypothesis. This is completely reasonable since the week-day effect is able to explain some of the previously unaccounted heterogeneity. In terms of model validation, the inclusion of this effect improves sensibly the R2 (0.956), while the average cover- age Cov95% is constant (0.950). 4.1.2 Prediction of future cases and of the peak date For the latter empirical model, the RMSPEs for each steps-ahead are presented in Figure 11. Results match the expecta- tions as: (i) the error decreases with the length of the fitting window; (ii) the error trend is more stable on larger testing windows (10-15 steps ahead vs 1-5 steps ahead); (iii) larger errors are made around the day of the peak. It can be seen nevertheless that predictions are always reasonable at these time horizons. This is a good point for our theoretical frame- work, as it can be used as a guidance tool to plan nonpharmaceutical-interventions due to its capability to predict future scenarios with reasonable accuracy. Of course, with more detailed data and including confounding factors, the accuracy may be further improved. Unfortunately, the aggregated available data do not contain important information which may strongly improve the prediction, for example, stratifications of cases by age, gender, comorbidities, and so on. Finally, we evaluate the model’s ability to predict the date of the peak. The approximate dates and heights of the peak have important epidemiological implications. This becomes possible under the assumption that sensible modifications ALAIMO DI LORO et al. 18 18 ALAIMO DI LORO et al. 4.1.2 Prediction of future cases and of the peak date kk k k k k k k k k k kk k k k k k k k k k k k k k kk k kk kk k kkk k k k k k k k k k kk kk k k k k k kk k k k k kk k k k kkkk k kkk k k kk kkk kkkkkkkkkkkkkk k kkk k kkkkkkkkk 0 500 1000 1500 01−Apr 01−May 01−Jun 01−Jul RMSPE 1 step−ahead k kkk k k k k k k k k k k k k k k k k k k k kkkkk kkkkkkkk kkkkk k kkkkkk kkkkkkkkkkkkkkkkk kkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkk 0 500 1000 1500 01−Apr 01−May 01−Jun 01−Jul RMSPE 5 step−ahead kk k k k k kk k k k kk k k k k kk kk k k kkkkkkkk kkkkkkkkkk kkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkk 0 500 1000 1500 01−Apr 01−May 01−Jun 01−Jul RMSPE 10 step−ahead kkk k k k k k k k k kkk k k k k kk k k k kkk kkkkkkkkkkkkkkk kkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkk 0 500 1000 1500 01−Apr 01−May 01−Jun 01−Jul RMSPE 15 step−ahead F I G U R E 11 Root mean squared prediction error for daily positives at different steps-ahead (A) (B) (C) (D) F I G U R E 12 Estimation of the date of the peak for daily positives at different steps-before of the adopted epidemiological strategies do not emerge. However, if exogenous events, for example, efficient treatments or vaccines, arise at a certain point in time, our framework allows to include it to predict the peak, in a similar manner as we did for the week seasonality effect. To do so, we estimate the model without covariates, using all available data until K ∈{15, 10, 5, 3, 2, 1} days before the observed peak. For the sake of conciseness, we only report results for K ∈{10, 5, 2, 1} as shown in Figure 12. When s = 1, the peak ̂t is directly expressed by the parameter p. 4.1.2 Prediction of future cases and of the peak date When s ≠1, after some algebra it can be seen that the peak can still be computed analytically as: k kkk k k k k k k k k k k k k k k k k k k k kkkkk kkkkkkkk kkkkk k kkkkkk kkkkkkkkkkkkkkkkk kkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkk 0 500 1000 1500 01−Apr 01−May 01−Jun 01−Jul RMSPE 5 step−ahead F I G U R E 11 Root mean squared prediction error for daily positives at different steps-ahead kk k k k k k k k k k kk k k k k k k k k k k k k k kk k kk kk k kkk k k k k k k k k k kk kk k k k k k kk k k k k kk k k k kkkk k kkk k k kk kkk kkkkkkkkkkkkkk k kkk k kkkkkkkkk 0 500 1000 1500 01−Apr 01−May 01−Jun 01−Jul RMSPE 1 step−ahead kkk k k k k k k k k kkk k k k k kk k k k kkk kkkkkkkkkkkkkkk kkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkk 0 500 1000 1500 01−Apr 01−May 01−Jun 01−Jul RMSPE 15 step−ahead kk k k k k kk k k k kk k k k k kk kk k k kkkkkkkk kkkkkkkkkk kkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkk 0 500 1000 1500 01−Apr 01−May 01−Jun 01−Jul RMSPE 10 step−ahead RMSPE RMSPE (B) (A) F I G U R E 12 Estimation of the date of the peak for daily positives at different steps-before F I G U R E 12 Estimation of the date of the peak for daily positives at different steps-before (A) (B) (D) (C) (C) (D) of the adopted epidemiological strategies do not emerge. However, if exogenous events, for example, efficient treatments or vaccines, arise at a certain point in time, our framework allows to include it to predict the peak, in a similar manner as we did for the week seasonality effect. To do so, we estimate the model without covariates, using all available data until K ∈{15, 10, 5, 3, 2, 1} days before the observed peak. For the sake of conciseness, we only report results for K ∈{10, 5, 2, 1} as shown in Figure 12. of the adopted epidemiological strategies do not emerge. 4.1.2 Prediction of future cases and of the peak date The dashed gray vertical lines represent the bounds of the confidence interval and the predicted date of the peak (confidence area is shaded with the same gray). The solid vertical black line represents the “true” date of the peak (ie, obtained via smoothing of the observed counts through nonparametric polynomial approximations). The observed time-series is represented through point and lines, where the black section is referred to the training window while the gray section is referred to the testing (out-of-sample) window. Confidence intervals are obtained through the same bootstrap procedure introduced in Section 3.4. The dashed gray vertical lines represent the bounds of the confidence interval and the predicted date of the peak (confidence area is shaded with the same gray). The solid vertical black line represents the “true” date of the peak (ie, obtained via smoothing of the observed counts through nonparametric polynomial approximations). The observed time-series is represented through point and lines, where the black section is referred to the training window while the gray section is referred to the testing (out-of-sample) window. As expected, as we approach the real date of the peak, we predict it more accurately. Point predictions are very accu- rate since 5 days before the actual peak. At the same time, interval bounds get tighter and tighter as the fitting interval approached the day of the peak and, in general, the day of the peak is always included in such bounds (see Table 5 for exact numerical evaluation). All the aforementioned results have been calculated also for the national aggregated daily deceased. Exposition and discussion of these results, which are in fact very similar to the daily positivesones, are included in the supplementary material. Here, we just want to highlight how the peak is accurately predicted with a shorter delay and generally smaller uncertainty for the daily deceased than for the daily positives (see Table 5). This is probably related to the more regular behavior of the series, due to a likely more homogeneous collection process of the records. Finally, we here want to stress the point that we are introducing a framework with the highly desirable goal to formu- late a model which would predict an evolution curve. 4.1.2 Prediction of future cases and of the peak date To be more precise, a great variety of epidemiological models have been proposed in the literature, but most standard versions of SIR-like models typically yield an increase before the peak that is quite similar to the decrease after the peak. The proposed framework, based on more complex evolution dynamics, is robust enough to be fitted successfully on the (poor quality) available data while explaining and forecasting different increasing and decreasing behavior before and after the peak. We emphasize that such increase-decrease quantitative behaviors appear to satisfactorily conform to reality. 4.1.2 Prediction of future cases and of the peak date However, if exogenous events, for example, efficient treatments or vaccines, arise at a certain point in time, our framework allows to include it to predict the peak, in a similar manner as we did for the week seasonality effect. To do so, we estimate the model without covariates, using all available data until K ∈{15, 10, 5, 3, 2, 1} days before the observed peak. For the sake of conciseness, we only report results for K ∈{10, 5, 2, 1} as shown in Figure 12. When s = 1, the peak ̂t is directly expressed by the parameter p. When s ≠1, after some algebra it can be seen that the peak can still be computed analytically as: ̂t ̂𝜸= ̂p + log10(̂s) ̂h . ALAIMO DI LORO et al. 19 TA B L E 5 Delay (days) in point estimation of the peak 10 5 2 1 Days before Delay Width Delay Width Delay Width Delay Width daily deceased −1 37 –3 25 –4 22 –3 21 daily positives 20 106 17 69 1 37 2 37 Confidence intervals are obtained through the same bootstrap procedure introduced in Section 3.4. The dashed gray vertical lines represent the bounds of the confidence interval and the predicted date of the peak (confidence area is shaded with the same gray). The solid vertical black line represents the “true” date of the peak (ie, obtained via smoothing of the observed counts through nonparametric polynomial approximations). The observed time-series is represented through point and lines, where the black section is referred to the training window while the gray section is referred to the testing (out-of-sample) window. As expected, as we approach the real date of the peak, we predict it more accurately. Point predictions are very accu- rate since 5 days before the actual peak. At the same time, interval bounds get tighter and tighter as the fitting interval ALAIMO DI LORO et al. 19 TA B L E 5 Delay (days) in point estimation of the peak 10 5 2 1 Days before Delay Width Delay Width Delay Width Delay Width daily deceased −1 37 –3 25 –4 22 –3 21 daily positives 20 106 17 69 1 37 2 37 TA B L E 5 Delay (days) in point estimation of the peak Confidence intervals are obtained through the same bootstrap procedure introduced in Section 3.4. ORCID Pierfrancesco Alaimo Di Loro https://orcid.org/0000-0002-6075-3659 Alessio Farcomeni https://orcid.org/0000-0002-7104-5826 Giovanna Jona Lasinio https://orcid.org/0000-0001-8912-5018 Gianfranco Lovison https://orcid.org/0000-0003-3861-8204 Antonello Maruotti https://orcid.org/0000-0001-8377-9950 h // id / Pierfrancesco Alaimo Di Loro https://orcid.org/0000-0002-6075-3659 Alessio Farcomeni https://orcid.org/0000-0002-7104-5826 Giovanna Jona Lasinio https://orcid.org/0000-0001-8912-5018 Gianfranco Lovison https://orcid.org/0000-0003-3861-8204 Antonello Maruotti https://orcid.org/0000-0001-8377-9950 Marco Mingione https://orcid.org/0000-0002-5662-3499 20 20 A limitation of our approach is that logistic growth curves are constrained so that only one wave at a time can be successfully modeled. This implies that initial (and possibly final) dates shall be set by the user to identify a wave. This is rather simple empirically (eg, the initial date can be the last day with zero incidence, and the final date can be the first day with incidence above (or under) a prespecified threshold). On the other hand, multiple waves could be modeled by modification of our nonlinear model as a weighted average of multiple Richards’ curves (one for each wave), in which weights of the noncurrent wave are forced to decay to zero with the distance from the wave-specific peak. We leave this as grounds for further work. A Bayesian approach will also be experimented in order to overcome possible issues with the asymptotic properties of the maximum likelihood estimator. Notably, implementation of the no-U-turn sampler algorithm for the estimation of nonlinear models might be a valid working solution. In addition, a Bayesian approach may also be used to include spatial dependence into the modeling framework and also to relax the first-order Markov assumption for taking into account more complex temporal dependence. In particular, the latter may be key in order to adapt the introduced Richards’ curve model for the nowcasting of prevalence indicators, for example, current positives and current intensive care units occupancy. Indeed, any modeling effort shall account for the strong temporal dependence between subsequent counts stemming from the fact that daily counts at time t potentially include units which are in stock since times 𝜏< t. Furthermore, as specified in Section 2.1.2, prevalence indicators are nonmonotonic and their value is the result of the combination of the incidence components building up each of those. These two last issues may be addressed by adapting the Richards’ response function to accommodate nonmonotonicity and/or by hierarchically specifying a model for the prevalence indicators through the combination of models for their incidence components. A successful attempt in accurately nowcasting the ICU occupancy is given in Reference 51. SOFTWARE Software in the form of R code, together with a sample input dataset and complete documentation is available on request from the corresponding author. DATA AVAILABILITY STATEMENT The data that support the findings of this study are openly available in “Protezione Civile Italiana GitHub Repository” at https://github.com/pcm-dpc/COVID-19. 20 ALAIMO DI LORO et al. 5 DISCUSSION AND FURTHER WORK We presented an approach to modeling and prediction of epidemic indicators that has proven useful during the first outbreak of COVID-19 in Italy. The model has been validated on publicly available data, and has proved flexible enough to adapt to different indicators. It is important to underline up front that the available data are clearly biased. Incidence depends on testing and trac- ing efforts, whose indications have varied wildly over time and space. Comparability of indicators over time and space might in part be achieved by including the daily number of swabs as a predictor, which anyway would make predictions cumbersome. Different definitions of COVID-19 related death make it also very hard to compare mortality across coun- tries. This problem does not apply to our data, that refer only to Italy. However, while this definition has been constant over time in Italy, it shall be remarked that also deaths might be underestimated, with the degree of undercount positively associated with incidence. Correcting for this bias is not trivial, and would require corrections based on individual-level data and/or reliable statistics about excess mortality. Summarizing the results, we would like to emphasize that the proposed Richards’ curve model describes properly the growth in the number of COVID-19 daily positives and daily deceased, despite its simplicity. Indeed, it is able to reflect properly the trend of the daily incidence indicators; and also allows for the straightforward inclusion of exogenous infor- mation. Basic covariates such as the week-day effect proved to sensibly enhance model fitting and prediction accuracy. While we have illustrated results at the national level, the model can be used also at the regional/local level (perhaps including specific local effects). The resulting fits are included in the supplementary material. The maximum likelihood approach so far considered is rather stable, as long as reasonable starting values are found to initialize the algorithm. Robust inference from robust Tsallis score: application to COVID 18. Ritz C, Baty F, Streibig JC, Gerhard D. Dose-response analysis using R. PLoS One. 2015;10:e0146021. 19. Hsu F, Nelson C, Chow W. A mathematical model to utilize the logistic function in germination and seedling growth. J Exp Bot. 1984;35(11):1629-1640. 19. Hsu F, Nelson C, Chow W. A mathematical model to utilize the logistic function in germination and seedling growth. J Exp Bot. 1984;35(11):1629-1640. 20. Grossman M, Bohren B. Logistic growth curve of chickens: heritability of parameters. J Hered. 1985;76(6):459-462. 20. Grossman M, Bohren B. Logistic growth curve of chickens: heritability of parameters. J Hered. 1985;76(6):459-462. erkson J. Application of the logistic function to bio-assay. J Am Stat Assoc. 1944;39(227):357-365. 22. Berkson J. Application of the logistic function to bio-assay. J Am Stat Assoc. 1944;39(227):357-365. 22. Berkson J. Application of the logistic function to bio-assay. J Am Stat Assoc. 1944;39(227):357-365. 23. Wachenheim DE, Patterson JA, Ladisch MR. Analysis of the logistic function model: derivation and applic microorganisms. Bioresour Technol. 2003;86(2):157-164. Tjørve E, Tjørve KMC. A unified approach to the Richards-model family for use in growth analyses: Why we J Theoret Biol 2010;267(3):417 425 http://www sciencedirect com/science/article/pii/S00225193100%04741 24. Tjørve E, Tjørve KMC. A unified approach to the Richards-model family for use in growth analyses: Why J Theoret Biol. 2010;267(3):417-425. http://www.sciencedirect.com/science/article/pii/S00225193100%047 Tjørve E, Tjørve KMC. A unified approach to the Richards model family for use in growth analyses: Why we J Theoret Biol. 2010;267(3):417-425. http://www.sciencedirect.com/science/article/pii/S00225193100%04741. oret Biol. 2010;267(3):417-425. http://www.sciencedirect.com/science/article/pii/S00225193100%04741. 25. Tjørve K, Tjørve E. The use of Gompertz models in growth analyses, and new Gompertz-model approach: an addition to the Unified-Richards family. PLoS One. 2017;12(6):e0178691. 25. Tjørve K, Tjørve E. The use of Gompertz models in growth analyses, and new Gompertz-model approach: an addition to the Unified-Richards family. PLoS One. 2017;12(6):e0178691. 26. Banks RB. Growth and Diffusion Phenomena: Mathematical Frameworks and Applications. Vol 14. Berlin, Germany: Springer Science & Business Media; 1993. 26. Banks RB. Growth and Diffusion Phenomena: Mathematical Frameworks and Applications. Vol 14. Berlin, Germany: Springer Science & Business Media; 1993. 27. Hsieh YH, Fisman DN, Wu J. On epidemic modeling in real time: an application to the 2009 novel a (H1N1) influenza outbreak in Canada. BMC Res Notes. 2010;3(1):1-8. 27. Hsieh YH, Fisman DN, Wu J. On epidemic modeling in real time: an application to the 2009 novel a (H1N1) influenza outbreak in Canada. BMC Res Notes. 2010;3(1):1-8. 28. 21 9. Dehning J, Zierenberg J, Spitzner FP, et al. Inferring change points in the spread of COVID-19 reveals the effectiveness of interventions. Science. 2020. 9. Dehning J, Zierenberg J, Spitzner FP, et al. Inferring change points in the spread of COVID-19 reveals the effectiveness of interventions. Science. 2020. 10. Ioannidis JPA, Cripps S, Tanner MA. Forecasting for COVID-19 has failed. Int J Forecast. 2020. http://www.sciencedirect.com/science/ article/pii/S016920%7020301199. 10. Ioannidis JPA, Cripps S, Tanner MA. Forecasting for COVID-19 has failed. Int J Forecast. 2020. http://www.sciencedirect.com/science/ article/pii/S016920%7020301199. 11. Baek J, Farias VF, Georgescu A, et al. The limits to learning an SIR process: granular forecasting for COV arias VF, Georgescu A, et al. The limits to learning an SIR process: granular forecasting for COVID 19. 2020. ar Tran Kiem C, Lefrancq N, et al. Estimating the burden of SARS-CoV-2 in France. Science. 2020. 12. Salje H, Tran Kiem C, Lefrancq N, et al. Estimating the burden of SARS-CoV-2 in France. Science. 2020. j , , q , g 13. Chowell G, Hincapie-Palacio D, Ospina J, et al. Using phenomenological models to characterize transmissibility and forecast patterns and final burden of Zika epidemics PLoS Currents 2016;8 https://www ncbi nlm nih gov/pmc/articles/PMC4922743/# ffn sectitle 13. Chowell G, Hincapie-Palacio D, Ospina J, et al. Using phenomenological models to characterize transmiss final burden of Zika epidemics. PLoS Currents. 2016;8. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC G, Hincapie-Palacio D, Ospina J, et al. Using phenomenological models to characterize transmissibility and fore den of Zika epidemics. PLoS Currents. 2016;8. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4922743/#__ff 14. Grasselli G, Pesenti A, Cecconi M. Critical care utilization for the COVID-19 outbreak in Lombardy, Italy: early experience and forecast during an emergency response. J Am Med Assoc. 2020;323:1545-1546. 14. Grasselli G, Pesenti A, Cecconi M. Critical care utilization for the COVID-19 outbreak in Lombardy, Italy: early experience and forecast during an emergency response. J Am Med Assoc. 2020;323:1545-1546. 15. Sebastiani G, Massa M, Riboli E. COVID-19 epidemic in Italy: evolution, projections and impact of government measures. Eur J Epidemiol. 2020;35:341-345. 15. Sebastiani G, Massa M, Riboli E. COVID-19 epidemic in Italy: evolution, projections and impact of government measures. Eur J Epidemiol. 2020;35:341-345. 16. Cabras S. A Bayesian deep learning model for estimating COVID-19 evolution in Spain; 2020. arXiv20051 16. Cabras S. A Bayesian deep learning model for estimating COVID-19 evolution in Spain; 2020. arXiv200510335. 17. Girardi P, Greco L, Mameli V, et al. Ma J, Dushoff J, Bolker BM, Earn DJ. Estimating initial epidemic growth rates. Bull Math Biol. 2014;76(1 29. Wu K, Darcet D, Wang Q, Sornette D. Generalized logistic growth modeling of the COVID-19 outbreak: comparing the dynamics in the 29 provinces in China and in the rest of the world. Nonlinear Dyn. 2020;101(3):1561-1581. Wu K, Darcet D, Wang Q, Sornette D. Generalized logistic growth modeling of the COVID-19 outbreak: com 29 i i Chi d i th t f th ld N li D 2020 101(3) 1561 1581 29. Wu K, Darcet D, Wang Q, Sornette D. Generalized logistic growth modeling of the COVID-19 outbrea 29 provinces in China and in the rest of the world. Nonlinear Dyn. 2020;101(3):1561-1581. 29 provinces in China and in the rest of the world. Nonlinear Dyn. 2020;101(3):1561-1581. 30. Lee SY, Lei B, Mallick B. Estimation of COVID-19 spread curves integrating global data and borrowing information. PLoS One. 2020;15(7):e0236860. 30. Lee SY, Lei B, Mallick B. Estimation of COVID-19 spread curves integrating global data and borrowing information. PLoS One. 2020;15(7):e0236860. 31. Richards F. A flexible growth function for empirical use. J Exp Bot. 1959;10(2):290-301. 31. Richards F. A flexible growth function for empirical use. J Exp Bot. 1959;10(2):290-301. 31. Richards F. A flexible growth function for empirical use. J Exp Bot. 1959;10(2):290-301. 32. Werker A, Jaggard K. Modelling asymmetrical growth curves that rise and then fall: applications to foliage dynamics of sugar beet (Beta vulgarisL). Ann Bot. 1997;79(6):657-665. 32. Werker A, Jaggard K. Modelling asymmetrical growth curves that rise and then fall: applications to folia vulgarisL). Ann Bot. 1997;79(6):657-665. 33. Hsieh YH. Richards model: a simple procedure for real-time prediction of outbreak severity. Modeling and Singapore: World Scientific Publishing Co Pte Ltd; 2009:216-236. 34. Hsieh YH, Chen C. Turning points, reproduction number, and impact of climatological events for multi-wave dengue outbreaks. Tropical Med Int Health. 2009;14(6):628-638. 34. Hsieh YH, Chen C. Turning points, reproduction number, and impact of climatological events for multi-wave dengue outbreaks. Tropical Med Int Health. 2009;14(6):628-638. ( ) 35. Hsieh YH. Pandemic influenza A (H1N1) during winter influenza season in the southern hemisphere. Influenza Other Respir Viruses. 2010;4(4):187-197. 35. Hsieh YH. Pandemic influenza A (H1N1) during winter influenza season in the southern hemisphere. Influenza Other Respir Viruses. 2010;4(4):187-197. 36. Tsoularis A, Wallace J. Analysis of logistic growth models. Math Biosci. 2002;179(1):21-55. 36. Tsoularis A, Wallace J. Analysis of logistic growth models. REFERENCES 1. Flaxman S, Mishra S, Gandy A, et al. Estimating the number of infections and the impact of non-pharmaceutical interventions on COVID-19 in European countries: technical description update; 2020. arXiv:200411342. 1. Flaxman S, Mishra S, Gandy A, et al. Estimating the number of infections and the impact of non-pharmaceutical interventions on COVID-19 in European countries: technical description update; 2020. arXiv:200411342. 2. Peeri NC, Shrestha N, Rahman S, et al. The SARS, MERS and novel coronavirus (COVID-19) epidemics, the newest and biggest global health threats: what lessons have we learned? Int J Epidemiol. 2020. 2. Peeri NC, Shrestha N, Rahman S, et al. The SARS, MERS and novel co health threats: what lessons have we learned? Int J Epidemiol. 2020. p Hu B, Guo H, Zhou P, Shi ZL. Characteristics of SARS-CoV-2 and COVID-19. Nat Rev Microbiol. 2020;1-14. 3. Hu B, Guo H, Zhou P, Shi ZL. Characteristics of SARS-CoV-2 and CO Liang K. Mathematical model of infection kinetics and its analysis for COVID-19, SARS and MERS. Infect G http://www.sciencedirect.com/science/article/pii/S15671348203%01374. 5. Diekmann O, Heesterbeek H, Britton T. Mathematical Tools for Understanding Infectious Disease Dynamics. Princeton University Press: Princeton, NJ; 2013. 6. Chen YC, Lu PE, Chang CS. A Time-dependent SIR model for COVID-19; 2020. arXiv:200300122. 6. Chen YC, Lu PE, Chang CS. A Time-dependent SIR model for COVID-19; 2020. arXiv:200300122. 7. Giordano G, Blanchini F, Bruno R, et al. Modelling the COVID-19 epidemic and implementation of population-wide interventions in Italy. Nature Med. 2020. 8. Gatto M, Bertuzzo E, Mari L, et al. Spread and dynamics of the COVID-19 epidemic in Italy: effects of emergency containment measures. Proc Natl Acad Sci. 2020;117(19):10484-10491. https://www.pnas.org/content/117/19/10484. 8. Gatto M, Bertuzzo E, Mari L, et al. Spread and dynamics of the COVID-19 epidemic in Italy: effec Proc Natl Acad Sci. 2020;117(19):10484-10491. https://www.pnas.org/content/117/19/10484. ALAIMO DI LORO et al. Math Biosci. 2002;179(1):21-55. 36. Tsoularis A, Wallace J. Analysis of logistic growth models. Math Biosci. 2002;179(1):21-55. 37. Gompertz B. XXIV. On the nature of the function expressive of the law of human mortality, and on a new of life contingencies. in a letter to Francis Baily, Esq. FRS C Philos Trans R Soc Lond. 1825;115:513-583. 38. Causton D. A computer program for fitting the Richards function. Biometrics. 1969;25(2):401-409. 39. Birch CP. A new generalized logistic sigmoid growth equation compared with the Richards growth equatio 40. Kahm M, Hasenbrink G, Lichtenberg-Fraté H, Ludwig J, Kschischo M. Grofit: fitting biological growth curves. Journal of Statistical Software. 2010;33(7):1–21. https://doi.org/10.18637/jss.v033.i07. 40. Kahm M, Hasenbrink G, Lichtenberg-Fraté H, Ludwig J, Kschischo M. Grofit: fitting biological growth curves. Journal of Statistical Software. 2010;33(7):1–21. https://doi.org/10.18637/jss.v033.i07. f ( ) p g j 41. Cao L, Shi PJ, Li L, Chen G. A new flexible sigmoidal growth model. Symmetry. 2019;11(2):204. f g j 41. Cao L, Shi PJ, Li L, Chen G. A new flexible sigmoidal growth model. Symmetry. 2019;11(2):204. 41. Cao L, Shi PJ, Li L, Chen G. A new flexible sigmoidal growth model. Symmetry. 2019;11(2):204. 42. Shaman J, Galanti M. Will SARS-CoV-2 become endemic? Science. 2020;370:527-529. 43. Scrucca L. GA: a package for genetic algorithms in R. J Stat Softw. 2013;53:1-37. Nash JC, Varadhan R, Grothendieck G, Nash MJC, Yes L. Package ’optimx’; 2020. 44. Nash JC, Varadhan R, Grothendieck G, Nash MJC, Yes L. Package ’optimx’; 2020. 45. Hardin JW. The sandwich estimate of variance. maximum likelihood estimation of misspecified models: twenty years later; Emerald Group Publishing Limited; 2003:45-73. 45. Hardin JW. The sandwich estimate of variance. maximum likelihood estimation of misspecified models: twenty years later; Emerald Group Publishing Limited; 2003:45-73. A. On the so-called εHuber sandwich estimatorε and εrobust standard errorsε. Am Stat. 2006;60(4):299-302. 46. Freedman DA. On the so-called εHuber sandwich estimatorε and εrobust standard errorsε. Am Stat. 2006 tion of prediction error: covariance penalties and cross-validation. J Am Stat Assoc. 2004;99(467):619-632. Efron B. The estimation of prediction error: covariance penalties and cross-validation. J Am Stat Assoc. 2004; 47. Efron B. The estimation of prediction error: covariance penalties and cross-validation. J Am Stat Assoc. 2004;99(467):619-632. 47. Efron B. The estimation of prediction error: covariance penalties and cross-validation. J Am Stat Assoc. 2004;99(467):619-632. The estimation of prediction error: covariance penalties and cross-validation. J Am Stat Assoc. 2004;99(467):61 aiti T On parametric bootstrap methods for small area prediction J Royal Stat Soc Ser B (Stat Methodol) 2006 22 ALAIMO DI LORO et al. 22 49. Efron B. Bayesian inference and the parametric bootstrap. Ann Appl Stat. 2012;6(4):1971. 50. Svetliza CF, Paula GA. Diagnostics in nonlinear negative binomial models. Commun Stat Theory Methods. 2003;32(6):1227-1250. 51. Farcomeni A, Maruotti A, Divino F, Jona-Lasinio G, Lovison G. An ensemble approach to short-term forecast of COVID-19 intensive care occupancy in Italian regions. Biom J. 2020;63:503-513. SUPPORTING INFORMATION y p p pp ( ) 50. Svetliza CF, Paula GA. Diagnostics in nonlinear negative binomial models. Commun Stat Theory Methods. 2003;32(6):1227-1250. 51 Farcomeni A Maruotti A Divino F Jona-Lasinio G Lovison G An ensemble approach to short-term forecast of COVID-19 intensive care 9. o . ayes a e e ce a d t e pa a et c bootst ap. nn ppl Stat. 0 ;6( ): 97 . 50. Svetliza CF, Paula GA. Diagnostics in nonlinear negative binomial models. Commun Stat Theory Methods. 2003;32(6):1227-1250. 51. Farcomeni A, Maruotti A, Divino F, Jona-Lasinio G, Lovison G. An ensemble approach to short-term forecast of COVID-19 intensive care occupancy in Italian regions. Biom J. 2020;63:503-513. 51. Farcomeni A, Maruotti A, Divino F, Jona-Lasinio G, Lovison G. An ensemble approach to short-term forecast of COVID-19 intensive care occupancy in Italian regions. Biom J. 2020;63:503-513. SUPPORTING INFORMATION How to cite this article: Alaimo Di Loro P, Divino F, Farcomeni A, et al. Nowcasting COVID-19 incidence indicators during the Italian first outbreak. Statistics in Medicine. 2021;1–22. https://doi.org/10.1002/sim.9004
https://openalex.org/W4387677667	https://periodicos.ufrn.br/turismocontemporaneo/article/download/30789/17424	Portuguese	null	análise da execução dos roteiros afro do Olha! Recife	Revista de Turismo Contemporâneo	2,023	cc-by	6,612	Artigo recebido em: 10-11-2022 Artigo aprovado em: 29-03-2023 ISSN 2357-821 Uma análise da execução dos roteiros afro do Olha! Recife Isabela Andrade de Lima Morais Professora no Departamento de Hotelaria e Turismo da Universidade Federal de Pernambuco - UFPE, Recife/PE, Brasil E-mail: isabela.morais@ufpe.br Camila de Lima Graduada em Turismo pelo Departamento de Hotelaria e Turismo da Universidade Federal de Pernambuco - UFPE, Recife/PE, Brasil E-mail: ufpecamilalima@gmail.com Izabelle Cristine Pereira de Andrade Graduada em Turismo pelo Departamento de Hotelaria e Turismo da Universidade Federal de Pernambuco - UFPE, Recife/PE, Brasil E-mail: izabelle.pereira.andrade@gmail.com ISSN 2357-8211 Uma análise da execução dos roteiros afro do Olha! Recife Isabela Andrade de Lima Morais Professora no Departamento de Hotelaria e Turismo da Universidade Federal de Pernambuco - UFPE, Recife/PE, Brasil E-mail: isabela.morais@ufpe.br Camila de Lima Graduada em Turismo pelo Departamento de Hotelaria e Turismo da Universidade Federal de Pernambuco - UFPE, Recife/PE, Brasil E-mail: ufpecamilalima@gmail.com Izabelle Cristine Pereira de Andrade Graduada em Turismo pelo Departamento de Hotelaria e Turismo da Universidade Federal de Pernambuco - UFPE, Recife/PE, Brasil E-mail: izabelle.pereira.andrade@gmail.com ISSN 2357-8211 Revista de Turismo Contemporâneo, Natal, v. 11, n. 3, p. 389-404 set/dez.2023 ISSN 2357-821 Uma análise da execução dos roteiros afro do Olha! Recife Isabela Andrade de Lima Morais Professora no Departamento de Hotelaria e Turismo da Universidade Federal de Pernambuco - UFPE, Recife/PE, Brasil E-mail: isabela.morais@ufpe.br Camila de Lima Graduada em Turismo pelo Departamento de Hotelaria e Turismo da Universidade Federal de Pernambuco - UFPE, Recife/PE, Brasil E-mail: ufpecamilalima@gmail.com Izabelle Cristine Pereira de Andrade Graduada em Turismo pelo Departamento de Hotelaria e Turismo da Universidade Federal de Pernambuco - UFPE, Recife/PE, Brasil E-mail: izabelle.pereira.andrade@gmail.com ISSN 2357-821 Uma análise da execução dos roteiros afro do Olha! Recife Isabela Andrade de Lima Morais Professora no Departamento de Hotelaria e Turismo da Universidade Federal de Pernambuco - UFPE, Recife/PE, Brasil E-mail: isabela.morais@ufpe.br Camila de Lima Graduada em Turismo pelo Departamento de Hotelaria e Turismo da Universidade Federal de Pernambuco - UFPE, Recife/PE, Brasil E-mail: ufpecamilalima@gmail.com Izabelle Cristine Pereira de Andrade Graduada em Turismo pelo Departamento de Hotelaria e Turismo da Universidade Federal de Pernambuco - UFPE, Recife/PE, Brasil E-mail: izabelle.pereira.andrade@gmail.com RESUMO O turismo étnico afro endossa a cultura vinculada às etnias dos povos afrodescendentes. O afroturismo é uma atividade turística afrocentrada onde a população negra narra sua própria história. No Brasil, nos âmbitos público e privado, ambos segmentos turísticos vêm ganhando visibilidade nos últimos anos e impactando positivamente a vida de vários brasileiros afrodescendentes que planejam o turismo ou que consomem estes segmentos. Este estudo teve como objetivo fazer uma análise da execução dos roteiros afro do Projeto de Sensibilização Turística “Olha! Recife”, idealizado pela Secretaria de Turismo e Lazer da cidade do Recife. Foi realizado o mapeamento dos roteiros afro idealizados pelo projeto e uma entrevista com o responsável pelo projeto a fim obter informações sobre como são executados, além da realização de duas observações participantes. A pesquisa visou contribuir para uma maior visibilidade acadêmica acerca do turismo étnico afro e afroturismo, temas ainda pouco explorados na área acadêmica. Palavras-chave: Turismo étnico afro. Afrodescendente. Afroturismo. Roteiro afro. Olha! Revista de Turismo Contemporâneo, Natal, v. 11, n. 3, p. 389-404 set/dez.2023 Afrotourism. Afro scripts. Olha! Recife. Camila de Lima Revista de Turismo Contemporâneo, Natal, v. 11, n. 3, p. 389-404 set/dez.2023 389 Uma análise da execução dos roteiros afro do Olha! Recife. 1. INTRODUÇÃO A escravidão no Brasil durou mais de 300 anos e mesmo depois da abolição, ainda existem muitas questões que repercutem negativamente na vida da população afrodescendente. Pernambuco, de acordo com o Instituto Brasileiro de Geografia e Estatística (IBGE), foi um dos estados com maior número de africanos escravizados, constituindo-se, assim, uma das áreas com maior população escravizada até a segunda metade do século XIX (IBGE, 2007). Ainda de acordo com o censo do IBGE (2021), a população do Recife, capital de Pernambuco, é de aproximadamente 1.661.017 habitantes, sendo 66% desta população negra. 390 Uma análise da execução dos roteiros afro do Olha! Recife. A Prefeitura da Cidade do Recife conta com dois órgãos que trabalham ativamente no combate ao racismo e fomento da cultura afrodescendente: a Gerência de Igualdade Racial, que tem como objetivo fortalecer o trabalho de enfrentamento ao racismo institucional, reforçando a necessidade do diálogo entre o poder municipal e a sociedade civil; e o Núcleo de Cultura Afro-brasileira, que trabalha com projetos de valorização cultural da população afro do Recife. As esferas públicas municipais, estaduais e federal estão reconhecendo que é preciso discutir e apoiar pautas raciais também no turismo, tanto para combater o racismo e seus desdobramentos, quanto para reconhecer a importância e contribuição da população negra para o desenvolvimento do país. Ainda no campo do turismo se observa no Brasil um crescimento do turismo étnico afro e do afroturismo nos últimos anos. Iniciativas já foram tomadas pelo poder público em outras cidades como Salvador e Rio de Janeiro, que abrigam projetos que tem como objetivo reparar e apoiar comunidades afrodescendentes (em contextos urbanos ou rurais), evidenciando os aportes culturais da população afro e trazendo benefícios econômicos para essas comunidades. Com isso, estão surgindo cada vez mais experiências turísticas afrocentradas, que percebem os africanos ou os afro-brasileiros como sujeitos e agentes de fenômenos atuando sobre sua própria imagem cultural e de acordo com seus próprios interesses (Asante, 2009, p. 93). Essas experiências tem o objetivo de colocar as comunidades negras como protagonistas de sua história, gerando renda ao mesmo tempo em que combatem o racismo no turismo. No campo público, o Salvador Capital Afro é um projeto lançado pela Prefeitura de Salvador em 2022 que visa valorizar manifestações culturais, endossar a força das tradições, da ancestralidade e incentivar o empreendedorismo feito por pessoas negras (https://mundonegro.inf.br/salvador-lanca-projeto-de-fortalecimento-do-turismo-negro-e- valorizacao-da-heranca-ancestral/). Revista de Turismo Contemporâneo, Natal, v. 11, n. 3, p. 389-404 set/dez.2023 1. INTRODUÇÃO Em Recife há o Projeto de Sensibilização Turística da Secretaria de Turismo e Lazer do Recife, criado em 2014, que disponibiliza mais de 300 roteiros gratuitos espalhados pela cidade e adjacências. Os roteiros são divididos em quatro modalidades: 1) “Olha! Recife de ônibus”; 2) “Olha! Recife a Pé”; 3) “Olha! Recife Pedalando e 4) “Olha! Recife no Rio”. O presente artigo tem o objetivo de analisar e descrever de que forma os roteiros afro foram executados pelo Projeto de Sensibilização Turística da cidade do Recife, o Olha! Recife. A fins de critérios avaliativos, foi considerado roteiro afro todos os que têm a palavra “afro” no título ou na descrição dos roteiros encontrados no site oficial do projeto. Para realizar a análise, foram identificados os conceitos de turismo étnico afro e afroturismo e mapeou-se os roteiros afro executados pelo Olha! Recife através do site oficial 391 Uma análise da execução dos roteiros afro do Olha! Recife. do projeto. Visando obter dados importantes para o mapeamento e enriquecer a pesquisa, também foi realizada uma entrevista online com o Gerente de Inovação Turística responsável pela iniciativa, além da observação participante em dois roteiros afro do projeto: um na modalidade “Olha! Recife pedalando” e outro na modalidade “Olha! Recife a pé”. A escolha da cidade do Recife se deu pelo seu expressivo patrimônio cultural material, imaterial e histórico reconhecido internacionalmente, por ser de população majoritariamente negra e também por oferecer passeios gratuitos para a comunidade local e visitantes. Recife também vem estimulando e aprimorando o setor turístico levando em conta a demanda. Analisar apenas os roteiros afro partiu de uma inquietação das pesquisadoras ao perceberem a pouca representatividade negra no turismo, por ser um tema invisibilizado e ainda pouco estudado na área acadêmica, que ultimamente vem ganhando espaço em eventos e periódicos com o debate e produção de alguns pesquisadores brasileiros. Ademais, como estudantes de turismo e pesquisadoras afrodescendentes, investigar a relação entre turismo e pessoas negras através de uma análise da execução de roteiros afro é imprescindível para entender como a história desta população vem sendo contada pelos órgãos públicos responsáveis pelas atividades turísticas e de lazer na cidade, além de contribuir academicamente para a discussão. Revista de Turismo Contemporâneo, Natal, v. 11, n. 3, p. 389-404 set/dez.2023 2. TURISMO CULTURAL E TURISMO ÉTNICO Conforme o Ministério do Turismo do Brasil (MTur) (2010a, p. 15) o turismo cultural compreende “as atividades turísticas relacionadas à vivência do conjunto de elementos significativos do patrimônio histórico e cultural e dos eventos culturais, valorizando e promovendo os bens materiais e imateriais da cultura”. O turismo cultural, ainda segundo o MTur (2010), pode apresentar-se de forma material ou imaterial e tem como característica a manutenção e valorização do patrimônio cultural, materializado através da experiência sustentável do viajante por meio da atividade turística. Aqui, a cultura é vista como um produto a ser comercializado pelo turismo. Pensando no turismo cultural como precursor de outras possibilidades de turismo, e alavancado pela segmentação turística, observa-se dentre suas vertentes o turismo étnico, que se baseia no conceito de etnia - que são as características culturais que remetem a identidade, origem de um povo, ancestralidade e práticas sociais, físicas e religiosas. A etnia está ligada à “preservação de características, manifestações e produtos culturais próprios de um povo ou de uma comunidade, percebidos em situações sociais que diferem de 392 Uma análise da execução dos roteiros afro do Olha! Recife. suas próprias (contextos de diferença)” (Cardozo, 2004, p. 144 como citado em Souza & Pinheiro, 2018, p. 28). A etnia também influencia as formas de interação social internas e externas (Souza & Pinheiro, 2018, p. 27). A geografia cultural também tem um papel importante na manutenção do significado de etnia, já que a manifestação cultural étnica dos povos também está vinculada ao território, às outras formas de vida que habitam nele e às formas de vivenciá-lo. Assim, vários elementos convergem possibilitando uma experiência transversal e unificadora dos seres que ali vivem (Souza & Pinheiro, 2018). Portanto, o turismo étnico usufrui destas características identitárias que compõem um grupo e o transforma em um produto, gerando interações sociais entre o turista e a comunidade. A comunidade abre espaço para o turista, guiando-o pelos seus costumes, crenças, modo de vida e rituais, visando uma impulsão econômica e social sustentável. O turista conhece novos costumes, aprende mais sobre as crenças, expressões culturais e modo de vida daquela comunidade, consumindo uma experiência. Revista de Turismo Contemporâneo, Natal, v. 11, n. 3, p. 389-404 set/dez.2023 2.1 Turismo étnico afro e afroturismo Partindo da reflexão sobre o que significa o turismo étnico e de como ele se materializa nas relações entre cultura e turismo, neste trabalho focaremos no turismo étnico afro - que potencializa os aspectos culturais, as práticas sociais, territoriais e ancestrais dos povos afrodescendentes. Segundo Pinho (2005, p. 27), desde a década de 70 afrodescendentes provenientes dos Estados Unidos vem ao Brasil buscando uma conexão com a cultura africana presente. Entretanto, a discussão sobre o turismo étnico afro no Brasil começou a ganhar visibilidade no início do século XXI. No ano de 2006, em São Paulo, foi criada uma agência de turismo privada pioneira no ramo do turismo étnico afro, a Rota da Liberdade, comandada por Solange Barbosa, que tem como principal objetivo elaborar roteiros focados na valorização da comunidade negra quilombola, visando uma expansão econômica sustentável das comunidades (Guia Negro, 2020). Em 2007 o Estado da Bahia criou um programa em conjunto com a Secretaria de Turismo focado no segmento do turismo étnico afro. O programa visa reparar as consequências negativas causadas pela ausência de políticas públicas e projetos sociais direcionados às comunidades afrodescendentes, que historicamente viveram à margem dos lucros gerados pelo 393 Uma análise da execução dos roteiros afro do Olha! Recife. turismo, através do reconhecimento cultural e do desenvolvimento econômico sustentável (Pinheiro, 2015). Ainda em 2010, promovido pela Bahiatursa em parceria com a Secretaria Estadual do Turismo (Setur), a cidade de Salvador sediou o I Seminário Nacional de Turismo Étnico Afro. O encontro teve como objetivo expandir o debate sobre turismo étnico afro, abraçar as contribuições das diversas comunidades que já praticavam esse tipo de atividade turística, como também apontar as estratégias usadas para a expansão econômica da prática a nível nacional e internacional (Governo da Bahia, 2010). Em Pernambuco houve algumas discussões sobre turismo étnico afro feitas em 2011 em parceria entre a Empresa de Turismo de Pernambuco, a Latour e o Instituto Walmart, que trouxe experts em turismo étnico afro ao Centro de Convenções de Pernambuco para promover o segmento e divulgar o potencial turístico afro do Estado (Reigada, 2011). No entanto, nos tempos atuais, novas investigações realizadas sobre pessoas afrodescendentes e turismo trazem uma crítica à definição de turismo étnico afro. 2.1 Turismo étnico afro e afroturismo Apesar de o termo ser amplamente utilizado dentro dos aparelhos turísticos estatais e demais organizações relacionadas ao turismo, o que vemos acontecer é a execução deste turismo associado somente às comunidades quilombolas, o que contribui para a valorização da cultura negra, porém termina delimitando a discussão apenas a estas comunidades. Dada esta limitação, pesquisadores vêm utilizando o termo do afroturismo ou turismo afrocentrado para se referir a cultura negra como potencializadora do turismo, com um olhar mais crítico voltado às questões identitárias rurais e urbanas. De acordo com Oliveira (2020, p. 308), o turismo afrocentrado pode ser pensado como “o turismo pautado por narrativas afroreferenciadas, sejam elas urbanas, rurais etc., de maneira a deslocar o olhar do turismo tradicional, pautado por uma visão branca e eurocêntrica para uma perspectiva negra dos fenômenos sociais envoltos no fazer turismo”. O turismo afrocentrado ou afroturismo busca retirar as lentes eurocêntricas que fundamentaram o turismo, posicionando a pessoa afro como sujeita, dona de sua própria narrativa, seus processos epistemológicos, socioculturais e históricos, em contextos urbanos ou rurais. Revista de Turismo Contemporâneo, Natal, v. 11, n. 3, p. 389-404 set/dez.2023 Ainda conforme o Ministério do Turismo (2007, p. 16), a roteirização auxilia no processo de identificação, elaboração e consolidação de novos roteiros turísticos e, além disso, tem como função apontar a necessidade de aumento dos investimentos em projetos já existentes, seja na melhoria da estrutura atual, seja na qualificação dos serviços turísticos oferecidos. O roteiro turístico é um itinerário que tem como objetivo planejar e organizar o percurso da viagem para que o turista possa desfrutar do passeio e visitar os lugares que agregam valor à experiência. Para Silva e Costa Novo (2010, p. 16), “os roteiros não se resumem a uma visita a determinados atrativos, mas representam uma importante ferramenta para a leitura da realidade existente e da situação sociocultural vigente na localidade”. O roteiro pode existir em qualquer parte onde o turismo é praticado e pode ser realizado em diferentes ambientes, são importantes pois contribuem para uma contextualização dos atrativos existentes e da realidade de uma determinada localidade. Podem ser comercializados ou não e serem feitos de forma organizada ou espontânea. Para Souza e Corrêa (2000, p. 130), roteiro é “o itinerário escolhido pelo turista. Pode ser organizado por agência (roteiro programado) ou pode ser criado pelo próprio turista (roteiro espontâneo)”. Estando, portanto, o roteiro diretamente ligado ao itinerário. Tomando como referência as definições de roteiro turístico - e por não encontrar referências sobre roteiro afro – se pode afirmar que roteiro afro é um tipo de roteiro que resgata a história dos afrodescendentes, sendo protagonizado pela população afro na região em que está inserido, entregando também ao turista um contexto sociocultural dos lugares visitados. 2.3 Roteiro turístico e roteiro afro Segundo o Ministério do Turismo (2007, p. 13), roteiro turístico pode ser entendido como “um itinerário caracterizado por um ou mais elementos que lhe conferem identidade, definido e estruturado para fins de planejamento, gestão, promoção e comercialização turística das localidades que formam o roteiro”. 394 Uma análise da execução dos roteiros afro do Olha! Recife. Ainda conforme o Ministério do Turismo (2007, p. 16), a roteirização auxilia Revista de Turismo Contemporâneo, Natal, v. 11, n. 3, p. 389-404 set/dez.2023 3.METODOLOGIA Esta é uma pesquisa de caráter descritiva qualitativa básica. Primeiro foi realizado um levantamento bibliográfico e de dados para encontrar informações sobre turismo étnico afro e afroturismo. Foram consultadas as bases de dados google acadêmico e scielo, onde foram encontrados os artigos utilizados para a base conceitual da pesquisa. Além disso, foi consultado o site do Guia Negro (https://guianegro.com.br/), onde é possível encontrar notícias sobre afroturismo, afroemprendedorismo, tendências de mercados, etc. O site do Panrotas (https://www.panrotas.com.br/) e do Ministério do Turismo (https://www.gov.br/turismo/pt-br) também foram consultados para obtenção de dados relevantes para embasar a pesquisa. Em seguida, foi efetuado o mapeamento e descrição dos roteiros afro oferecidos pelo Projeto de Sensibilização Turística da Secretaria de Turismo e Lazer do Recife, “Olha! Recife” através do site oficial, salienta-se que foi contabilizado como roteiro afro do Olha! Recife aqueles que tivessem a palavra afro no nome ou na descrição do roteiro. 395 Uma análise da execução dos roteiros afro do Olha! Recife. Também foi realizada uma entrevista semiestruturada online, através do google meet, no dia 26 de agosto de 2022 com o Gerente de Inovação e Roteiros Turísticos, pessoa responsável pela gerência do projeto. A entrevista de caráter semiestruturada, com um total de 12 perguntas, teve como objetivo coletar informações específicas sobre os roteiros afro, tais como: quem idealiza os roteiros afro, a frequência em que cada roteiro ocorre, quantas pessoas trabalham na equipe do projeto, a estrutura dos roteiros, os locais visitados, entre outras informações que não foram encontradas no site oficial. Além do mais, realizou-se uma pesquisa de campo com a observação participante em dois momentos, o primeiro aconteceu na modalidade “Olha! Recife Pedalando” no dia 15 de maio de 2022, próximo ao dia 13 de maio, dia em que a lei Áurea foi assinada; o segundo aconteceu na modalidade “Olha! Recife a Pé” no dia 30 de julho de 2022, o passeio foi executado em comemoração ao dia da Mulher Negra Latino-Americana e Caribenha, celebrado em 25 de julho. As observações foram realizadas nestas datas já que durante a elaboração deste trabalho, foram os únicos momentos onde se abriram inscrições para esses roteiros afro. 4. ANÁLISE E DISCUSSÃO DOS RESULTADOS 4. RESULTADO E DISCUSSÃO Revista de Turismo Contemporâneo, Natal, v. 11, n. 3, p. 389-404 set/dez.2023 4.1 Projeto Olha! Recife Entre os objetivos do projeto Olha!Recife estão “elevar o sentimento de pertencimento da sociedade e autoestima, promover novas ideias e posturas sociais de valorização das tradições socioculturais e da vocação turística do Recife” (Olha! Recife, 2022), além da preservação do patrimônio histórico e cultural. No site oficial do projeto, para fins de pesquisa, foi contabilizado como roteiro afro do Olha! Recife aqueles que tivessem a palavra afro no nome ou na descrição. Foram encontrados doze roteiros com a característica mencionada, além do roteiro intitulado Circuito da Abolição, que faz menção a abolição da escravatura no Brasil. A seguir serão apresentadas, segundo suas modalidades, as descrições de cada roteiro afro identificado no projeto na seguinte ordem: nome registrado e descrição dos pontos visitados de acordo com o site do projeto Olha! Recife (Olha Recife, 2022). Vale salientar que os nomes e descrições dos roteiros foram retirados do site oficial, portanto, alguns atrativos não estão descritos. Segundo o Gerente de Inovação, estes atrativos só são informados pelo público participante no momento da execução do passeio, pois podem ser modificados: 396 Uma análise da execução dos roteiros afro do Olha! Recife. 1) “Olha! Recife de ônibus” 1) “Olha! Recife de ônibus” ● Circuito Afro: passeio por locais de referência à memória afro no Recife. São visitados a Igreja do Rosário dos Homens Pretos, a estátua de Zumbi, o monumento aos maracatus e o terreiro Oxum Opará do Tata Raminho de Oxossi, em Olinda. ● Circuito Afro - Literário: em homenagem ao mês da consciência negra, o Olha! Recife Literário é um circuito sobre a influência afro na literatura. ● Circuito Afro - Literário: em homenagem ao mês da consciência negra, o Olha! Recife Literário é um circuito sobre a influência afro na literatura. ● Folhas e Solos Sagrados: passeio pelo Recife com visita a alguns dos pontos de maior representatividade para a cultura africana ou afro-brasileira em nossa cultura. Circuito da Abolição: passeio de ônibus pelos principais pontos turísticos do Recife. 2) “Olha! Recife a Pé” ● Circuito Afro: Caminhada pelos principais atrativos, com acompanhamento de guia de turismo, apontando a história e curiosidades dos locais visitados. ● Recife Afro: Poesias, Lutas e Canções: Caminhada pelos principais atrativos, com acompanhamento de guia de turismo, apontando a história e curiosidades dos locais visitados. Revista de Turismo Contemporâneo, Natal, v. 11, n. 3, p. 389-404 set/dez.2023 4.1 Projeto Olha! Recife ● 13 de maio não é dia de negro: Passeio a pé que passa por locais emblemáticos da memória afro do Recife e a história de luta e resistência dos negros no estado. ● Pátio de São Pedro: Visita ao Pátio de São Pedro e seus equipamentos culturais como o Memorial Luiz Gonzaga, Memorial Chico Science e Núcleo Afro do Recife. ● Curiocidades - Recife Afro: Roteiro a pé que passa por locais emblemáticos da memória afro do Recife e a história de luta e resistência dos negros no estado. O roteiro conta com a presença de projeções da Eletrobike (bicicletas elétricas). ● Roteiro Fotográfico Afro: Roteiro a pé especial focado em fotografia pelas ruas do centro do Recife com apresentações artísticas. 3) “Olha! Recife Pedalando” 4) “Olha! Recife no Rio” 4) “Olha! Recife no Rio” 4) “Olha! Recife no Rio” Não foram encontrados roteiros afro nessa modalidade. Não foram encontrados roteiros afro nessa modalidade. De acordo com o Gerente de Inovação e Roteiros Turísticos, o público-alvo do projeto é a população recifense, apesar de haver sempre turistas presentes nas atividades, pois foi observado através de pesquisas feitas pelos responsáveis do projeto que atualmente um total de aproximadamente 20% dos participantes são turistas, fato que fez criar alternativas também para os viajantes, como o Recife Walking Tour. O Gerente afirmou que em 2013, quando o projeto foi lançado, contava com uma equipe maior (não foi informada a quantidade exata de membros da equipe), mas que atualmente dispõem de quatro colaboradores, responsáveis pela idealização dos roteiros, gerenciamento e execução das atividades. As inscrições para os roteiros são realizadas geralmente às sextas-feiras a partir das 09h00 no site do Olha! Recife. A duração média dos passeios é de 2h30 para as modalidades “Olha! Recife a Pé” e “Olha! Recife Pedalando”, e de 3h30 para a modalidade “Olha! Recife de ônibus”. Não foi informada a duração média da modalidade “Olha! Recife no Rio”. Ainda conforme a Gerência de Inovação, todos os roteiros afro foram elaborados pela própria gerência, além de sugestões da sociedade civil que são acatadas pela gerência através de um formulário de pesquisa que é enviado após a realização dos roteiros. Para a elaboração dos roteiros, é feita uma relação entre lugares visitados, tempo e distância média, também se considera a aceitação do público através do mesmo formulário de pesquisa. A execução desses roteiros é seccionada por épocas do ano, ou seja, a razão principal de suas realizações é vinculada a datas emblemáticas - a fim de contextualizá-las. O projeto busca a colaboração de alguns atores da sociedade civil envolvidos com a história afro na execução dos roteiros, que são consultados com antecedência. O Gerente de Inovação citou parcerias com o Sítio de Pai Adão, localizado no bairro de Água Fria, no Recife; o Cores do Amanhã, um projeto social localizado no bairro do Totó, também na cidade do Recife; e o Palácio de Iemanjá, localizado no Alto da Sé, em Olinda. A colaboração pode ser orgânica, para fins de divulgação das partes, ou envolver contribuição monetária. Revista de Turismo Contemporâneo, Natal, v. 11, n. 3, p. 389-404 set/dez.2023 3) “Olha! Recife Pedalando” 3) “Olha! Recife Pedalando” ● Circuito Afro: passeio por locais representativos da cultura e memória afro no centro do Recife, percorrendo a Praça da República e seu baobá, o Pátio do Carmo e a estátua em homenagem a Zumbi dos Palmares, Igreja do Rosário dos Homens Pretos, Pátio de São Pedro e estátua de Solano Trindade, Pátio do Terço e Monumento aos maracatus. Revista de Turismo Contemporâneo, Natal, v. 11, n. 3, p. 389-404 set/dez.2023 397 Uma análise da execução dos roteiros afro do Olha! Recife. ● Rainhas, Santos e Maracatus: é uma pedalada por locais e monumentos históricos que fazem referência a ícones da cultura afro. ● Rainhas, Santos e Maracatus: é uma pedalada por locais e monumentos históricos que fazem referência a ícones da cultura afro. ● Rainhas, Santos e Maracatus: é uma pedalada por locais e monumentos históricos que fazem referência a ícones da cultura afro. 4) “Olha! Recife no Rio” Ainda sobre a execução dos roteiros, o Gerente de Inovação afirma que se preocupa com a questão da representatividade, portanto todos os guias de turismo designados são negros. 398 Uma análise da execução dos roteiros afro do Olha! Recife. Após a execução do roteiro é enviada uma pesquisa via e-mail, no formato de um formulário estruturado no google forms, que visa obter informações mais detalhadas sobre as pessoas que participaram do roteiro, além de avaliações sobre o guia e o roteiro. O Gerente não soube informar se no formulário existe alguma pergunta referente a raça dos indivíduos que realizaram os roteiros. O Olha! Recife realiza, cerca de duas a três vezes por ano, o Ciclo de Atualizações dos Guias de Turismo, que abrange temas diversos; sua finalidade é capacitar os guias de turismo responsáveis pela execução de todos os roteiros disponibilizados pelo projeto. O Gerente de Inovação ressaltou que os guias de turismo que trabalham para o Olha! Recife não são guias exclusivos do projeto. Por último, o Gerente de Inovação afirmou que o projeto tem uma parceria sólida com a Gerência de Igualdade Racial. Quando solicitado pela mesma, o Olha! Recife disponibiliza roteiros voltados para a temática afro e/ou racismo, além de guias de turismo para atender as demandas da Gerência para realização de turismo pedagógico, que geralmente provém de escolas na cidade do Recife e adjacências. Revista de Turismo Contemporâneo, Natal, v. 11, n. 3, p. 389-404 set/dez.2023 4.2 Observação participante Nos dias 15 de maio de 2022 e 30 de julho de 2022 foram feitas observações participantes durante os Circuitos Afro do projeto de sensibilização turística do Olha! Recife, em duas modalidades: a pé e de bicicleta (quadro 01). De acordo com o Instagram oficial do Olha! Recife e o guia responsável pelo roteiro, o Circuito Afro realizado no dia 30 de julho de 2022 visava homenagear as mulheres negras pelo dia 25 de julho, no qual é comemorado o dia da Mulher Afrolatina, Afrocaribenha e da Diáspora. Quadro 01 - Cronograma dos passeios observados DATA HORÁRIO MODALIDADE TEMA 15/05/2022 9:00 - 11:30 De Bicicleta Circuito Afro 30/07/2022 9:00 - 11:30 A pé Circuito Afro Fonte: Elaborado pela equipe Fonte: Elaborado pela equipe O Circuito Afro de bicicleta teve seu início às 9:00 da manhã no dia 15 de maio de 2022 em frente ao Centro de Atendimento ao Turista (CAT) na Praça do Arsenal, sendo finalizado aproximadamente às 11:30 em frente ao baobá da Praça da República. Este circuito passou por pontos como: estátua de Naná Vasconcelos no Marco Zero; a Igreja do Rosário dos Homens Pretos na Praça do Diário; A estátua de Solano Trindade no Pátio de São Pedro; Monumento 399 Uma análise da execução dos roteiros afro do Olha! Recife. ao Maracatu próximo ao Pátio do Terço; Estátua de Joaquim Nabuco e pintura de Abelardo da Hora, ambas próximas à ponte de ferro do Recife. A última parada foi no baobá da Praça da República. O guia responsável pelo grupo de aproximadamente 15 pessoas era um homem negro, e com ele havia também uma funcionária da Secretaria de Turismo. O grupo saiu da Praça do Arsenal em direção ao Marco Zero onde paramos em frente a estátua de Naná Vasconcelos, lá foi falado sobre a importância do percussionista Naná Vasconcelos para a comunidade negra e sobre a visibilidade dos grupos de maracatus na abertura do carnaval do Recife. De lá seguimos para a Praça do Diário onde o guia mencionou sobre a história do Diário de Pernambuco, minutos depois fomos para a Igreja do Rosário dos Homens Pretos, onde o guia relatou sobre a coroação do congo e sobre a relação entre a igreja católica e os maracatus. Revista de Turismo Contemporâneo, Natal, v. 11, n. 3, p. 389-404 set/dez.2023 4.2 Observação participante Em seguida o grupo foi guiado até o Pátio de São Pedro, em frente a estátua de Solano Trindade, lá foi contada a história do poeta e seus poemas. Do Pátio de São Pedro seguimos para o Monumento ao Maracatu, próximo ao Pátio do Terço, onde foi abordado sobre os maracatus Nação e sobre o maracatu Elefante ou maracatu de Dona Santa. Ainda neste mesmo ponto o guia mencionou o Pátio do Terço e a casa de Badia, porém não paramos em frente, assim como a estátua de Zumbi dos Palmares na Praça do Carmo, que foi apenas mencionada de longe. Do Pátio do Terço fomos pedalando até a ponte de ferro, lá o guia discorreu sobre a estátua de Joaquim Nabuco. Ainda próximo a ponte de ferro foi mencionada a pintura do Abelardo da Hora, a qual retrata a chegada dos colonizadores e do povo negro escravizado ao Brasil, técnicas de monocultura e libertação. O passeio foi finalizado em frente ao baobá na Praça da República, onde foi destacada a importância da árvore para o povo negro e seus significados. O Circuito Afro a pé teve seu início às 9:00 da manhã no dia 30 de julho de 2022 em frente ao Centro de Atendimento ao Turista (CAT) na Praça do Arsenal e foi finalizado aproximadamente às 11:30 no Monumento ao Maracatu próximo ao Pátio do Terço. Este circuito passou pelos seguintes pontos: Rua do Bom Jesus, onde está localizada a Sinagoga Kahal Zur Israel; estátua de Naná Vasconcelos no Marco Zero; estátua de Chico Science na Rua da Moeda; Baobá da Praça da República; Igreja do Rosário dos Homens Pretos na Praça do Diário; Monumento ao Maracatu no Pátio do Terço. O tempo médio em cada lugar foi de aproximadamente 10 minutos. O circuito contou com aproximadamente 15 pessoas e a presença do guia, um homem negro e de religião de matriz africana, além de um funcionário da Secretária de Turismo. Saímos da Praça do Arsenal em direção a Rua do Bom Jesus e paramos em frente ao Sindicato dos Trabalhadores Portuários e da Sinagoga Judaica Kahal Zur Israel, lá foi abordado como o povo 400 Uma análise da execução dos roteiros afro do Olha! Recife. negro foi trazido e como eram tratados como mercadorias. 4.2 Observação participante Neste ponto do roteiro o guia fazia uso repetitivo do termo “escravo” e algumas das mulheres presentes no circuito chamaram a atenção para o uso equivocado deste termo. O grupo seguiu para o Marco Zero em frente a estátua de Naná Vasconcelos e lá ficamos aproximadamente dez minutos, o guia falou sobre a importância do percussionista Naná Vasconcelos para a cultura negra na cidade e no final mencionou o grupo Voz Nagô, composto por cantoras negras que participavam da abertura do carnaval de Recife, juntamente com Naná Vasconcelos e as nações de maracatu. De la, o grupo seguiu para a Praça do Arsenal em frente a estátua de Chico Science, onde foi dado ênfase para a história dos homens negros, nesse momento houve uma interrupção por parte de algumas mulheres negras presentes no circuito questionando sobre ausência de referências à participação das mulheres negras já que era um passeio voltado para comemorar o dia da mulher afro-latina. Então o guia citou a importância da cantora Isaar para o movimento negro de Recfe e reforçou que existiam muitas mulheres no movimento, porém que não tiveram muita visibilidade. Da Praça do Arsenal, passamos em frente ao Museu de Arte Afro Brasil Rolando Toro (Muafro Recife), porém o museu não foi mencionado pelo guia. Então o grupo seguiu para a Praça da República onde ficamos esperando liberação da nossa entrada na rua. Ao entrar, paramos em frente à estátua de Maurício de Nassau e falou-se muito sobre ele e seus feitos. Neste momento, o grupo mostrou mais uma vez certa inquietação, já que o roteiro tinha como objetivo contar a história de pessoas negras, sobretudo das mulheres negras. Ao pararmos em frente ao baobá na Praça da República ouvimos um poema da Inaldete Pinheiro; cada participante precisou abrir o poema no YouTube para poder ouvir, já que não havia caixas de som disponíveis para este momento. Então seguimos para a Igreja do Rosário dos Homens Pretos, onde foi falado sobre a coroação do congo e outros temas pertinentes à religiosidade dos grupos de maracatu. Depois visitamos o Pátio do Terço, local onde acontece a noite dos tambores silenciosos, encontro dos afoxés e encontro dos maracatus no carnaval da cidade do Recife. Paramos em frente a casa de Badia e neste momento o guia falou um pouco mais sobre a história de Badia, da sua relação com o carnaval e com o candomblé. 4.2 Observação participante Finalizamos o circuito aproximadamente às 11:30 em frente ao Monumento ao Maracatu onde se comentou sobre Dona Santa e um pouco da história dos maracatus de baque virado da cidade do Recife. Revista de Turismo Contemporâneo, Natal, v. 11, n. 3, p. 389-404 set/dez.2023 5. CONSIDERAÇÕES FINAIS A presente pesquisa alcançou seus objetivos mapeando os roteiros afro idealizados pelo Projeto Olha! Recife através do filtro selecionado, analisando a execução dos roteiros e 401 Uma análise da execução dos roteiros afro do Olha! Recife. entendendo como a história da população negra vem sendo contada pelos órgãos públicos responsáveis pelo turismo e lazer da cidade do Recife, assim contribuindo para uma maior visibilidade acadêmica sobre as questões aqui abordadas. Ao alinhar a teoria apresentada neste trabalho e as análises realizadas através da metodologia escolhida, podemos concluir que os roteiros afro executados pelo Olha! Recife não atendem aos requisitos teóricos do afroturismo. Embora resgatem histórias afrodescendentes e colaborem para a construção de uma narrativa antirracista, não é possível observar algum desenvolvimento econômico sustentável gerado pela execução dos roteiros. Não são roteiros idealizados ou protagonizados pela comunidade negra que habita a cidade, portanto, não são pautados por narrativas afrocentradas. É importante citar o espaço de destaque que os colonizadores receberam durante os circuitos, com comentários positivos feitos pelos guias das observações participantes. Consideramos que essa reverência é algo que não deve acontecer em roteiros que contam a história dos afrodescendentes. Houve superficialidade na hora de abordar temas importantes para a comunidade negra. Também é importante observar a falta de informação no site oficial sobre os roteiros afro aqui destacados, já que a descrição dos roteiros cita lugares emblemáticos, mas não especificam quais são esses lugares. Também não é possível encontrar informações sobre a frequência dos roteiros no site. O Olha! Recife é uma iniciativa gratuita que busca fomentar a sensação de pertencimento por parte dos moradores do Recife, realizando passeios que contam a história da cidade e endossam a valorização da cultura e, apesar de ter como público alvo os moradores, o projeto reverbera no turismo pois conta com a participação de turistas, uma vez que o morador local se transforma em um dos divulgadores da iniciativa, agregando os turistas que visitam a cidade de Recife. Contar a história da comunidade negra é mais um ato para reforçar esses aspectos, portanto, consideramos que o projeto atinge seus objetivos. Revista de Turismo Contemporâneo, Natal, v. 11, n. 3, p. 389-404 set/dez.2023 5. CONSIDERAÇÕES FINAIS As atividades realizadas entre o projeto - vinculado a Secretaria de Turismo e Lazer do Recife - e a Gerência de Igualdade Racial ampliam a discussão sobre pessoas negras e racismo e promovem o turismo pedagógico, portanto, acreditamos que uma colaboração com a Gerência de Igualdade Racial na elaboração dos roteiros afro do Olha! Recife traria outras perspectivas para os roteiros. É essencial que os órgãos públicos e seus gestores dimensionem a importância da comunidade negra para a história, economia e cultura brasileira. Endossar o turismo étnico afro e o afroturismo é necessário para afirmar essa importância, permitindo que as narrativas sobre a história negra sejam contadas pela população negra, visando propagar a valorização da cultura negra por todos os brasileiros, o que ajuda a combater o racismo. Ademais, as parcerias com 402 Uma análise da execução dos roteiros afro do Olha! Recife. atores da sociedade civil - pessoas negras - para a elaboração e execução do turismo étnico afro e do afroturismo geram renda para essa população, estimulando o empreendedorismo negro. A existência de mais pessoas negras em espaços públicos de poder proporciona ambientes mais críticos e plurais dentro do turismo. REFERÊNCIAS Asante, K. (2009). Afrocentricidade: notas sobre uma posição disciplinar. In Nascimento, E. L. (org.). Afrocentricidade: uma abordagem epistemológica inovadora. (pp. 93-110). São Paulo: Selo Negro. Ministério do Turismo. (2010a). Turismo Cultural: Orientações Básicas. Brasília: Ministério do Turismo. Recuperado de: http://antigo.turismo.gov.br/sites/default/turismo/o_ministerio/publicacoes/downloads _publicacoes/Turismo_Cultural_Versxo_Final_IMPRESSxO_.pdf. Ministério do Turismo. (2010). Segmentação do Turismo e Mercado. Recuperado de: https://www.gov.br/turismo/pt-br/centrais-de-cnteudo-/publicacoes/segmentacao-do- turismo/segmentacao-do-turismo-e-o-mercado.pdf. Ministério do Turismo. (2007). Programa de Regionalização do Turismo - Roteiros do Brasil: Módulo Operacional 7 - Roteirização Turística. Secretaria Nacional de Políticas de Turismo. Departamento de Estruturação, Articulação e Ordenamento Turístico. Coordenação Geral de Regionalização. Governo da Bahia. (2010). Salvador sedia primeiro Seminário Nacional de Turismo Étnico Afro. Recuperado de: http://www.bahia.ba.gov.br/2010/08/noticias/turismo/salvador- sedia-primeiro-seminario-nacional-de-turismo-etnico-afro/. Guia Negro. (2020). Rota da Liberdade organiza passeios para quilombos e tem trabalho reconhecido mundialmente. Recuperado de: https://guianegro.com.br/rota-da- liberdade-organiza-passeios-para-quilombos-e-tem-trabalho-reconhecido- mundialmente/. Instituto Brasileiro de Geografia e Estatística. (2021). Cidades e estados - Estudos e pesquisas: informação demográfica e socioeconômica. Recuperado de: https://www.ibge.gov.br/cidades-e-estados/pe/recife.html. Instituto Brasileiro de Geografia e Estatística. (2007). Brasil: 500 anos de povoamento. Centro de Documentação e Disseminação de Informações. Recuperado de: https://biblioteca.ibge.gov.br/visualizacao/livros/liv6687.pdf Olha Recife. (2022). O Projeto. Recuperado de: http://www.olharecife.com.br/projeto.php. Oliveira, N. A. (2020). Turismo afrocentrado: debates iniciais. In Mello, R. G., & Freitas, P. G. (orgs.). Novos olhares sobre Turismo, Patrimônio e Cultura, 305-315. Rio de Janeiro: Editora e-Publicar. Recuperado de: https://editorapublicar.com.br/novos- olhares-sobre-turismo-patrimonio-e-cultura Pinheiro, T. R. (2015). Turismo étnico e a construção das fronteiras étnicas: o caso do Quilombo do Campinho da Independência, Paraty (RJ). Dissertação de Mestrado em Memória Social. Programa de Pós-graduação em Memória Social, Universidade Federal do Estado do Rio de Janeiro, Rio de Janeiro. Revista de Turismo Contemporâneo, Natal, v. 11, n. 3, p. 389-404 set/dez.2023 403 Uma análise da execução dos roteiros afro do Olha! Recife. Pinho, P. S. (2005). Descentrando os Estados Unidos nos estudos sobre negritude no Brasil. Revista Brasileira de Ciências Sociais (São Paulo), 20(59), 40-43. https://doi.org/10.1590/S0102-69092005000300003 Pinho, P. S. (2005). Descentrando os Estados Unidos nos estudos sobre negritude no Brasil. Revista Brasileira de Ciências Sociais (São Paulo), 20(59), 40-43. https://doi.org/10.1590/S0102-69092005000300003 Reigada, M. I. (2011). Pernambuco promove turismo étnico em famtour. Recuperado de: https://www.panrotas.com.br/noticia-turismo/destinos/2011/11/pernambuco-promove- turismo-etnico-em-famtour_72921.html. Reigada, M. I. (2011). Pernambuco promove turismo étnico em famtour. Recuperado de: https://www.panrotas.com.br/noticia-turismo/destinos/2011/11/pernambuco-promove- turismo-etnico-em-famtour_72921.html. Silva, G. T. & Costa Novo, C. B. M. (2010). Roteiro turístico. Manaus: Centro de Educação Tecnológica do Amazonas. Silva, G. T. & Costa Novo, C. B. M. (2010). Roteiro turístico. Manaus: Centro de Educação Tecnológica do Amazonas. Souza, A. M. & Corrêa, M. FORMATO PARA CITAÇÃO DESTE ARTIGO Morais, I. A. L., Lima, C., & Andrade, I. C. P. (2023). Uma análise da execução dos roteiros afro do Olha! Recife. Revista de Turismo Contemporâneo, 11(3), 389-404. https://doi.org/10.21680/2357-8211.2023v11n3ID30789 Revista de Turismo Contemporâneo, Natal, v. 11, n. 3, p. 389-404 set/dez.2023 REFERÊNCIAS V. M. (2000). Turismo – Conceitos, definições e siglas. Manaus: Editora Valer. Souza, N. N. S. & Pinheiro, T. R. (2018). Turismo étnico. Rio de Janeiro: Fundação Cecierj. FORMATO PARA CITAÇÃO DESTE ARTIGO Morais, I. A. L., Lima, C., & Andrade, I. C. P. (2023). Uma análise da execução dos roteiros afro do Olha! Recife. Revista de Turismo Contemporâneo, 11(3), 389-404. https://doi.org/10.21680/2357-8211.2023v11n3ID30789 Revista de Turismo Contemporâneo, Natal, v. 11, n. 3, p. 389-404 set/dez.2023 404
https://openalex.org/W2914836117	https://aquila.usm.edu/cgi/viewcontent.cgi?article=17149&context=fac_pubs	English	null	Deep Learning Based Analysis of Histopathological Images of Breast Cancer	Frontiers in genetics	2,019	cc-by	14,532	Deep Learning Based Analysis of Histopathological Images of Deep Learning Based Analysis of Histopathological Images of Breast Cancer Breast Cancer Chaoyang Zhang University of Southern Mississippi, Chaoyang.Zhang@usm.edu Follow this and additional works at: https://aquila.usm.edu/fac_pubs Part of the Analytical, Diagnostic and Therapeutic Techniques and Equipment Commons The University of Southern Mississippi The University of Southern Mississippi The Aquila Digital Community The Aquila Digital Community Faculty Publications 1-28-2019 Deep Learning Based Analysis of Histopathological Images of Deep Learning Based Analysis of Histopathological Images of Breast Cancer Breast Cancer Juanying Xie Shaanxi Normal University Ran Liu Shaanxi Normal University Joseph Luttrell IV University of Southern Mississippi Chaoyang Zhang University of Southern Mississippi, Chaoyang.Zhang@usm.edu The University of Southern Mississippi The University of Southern Mississippi The Aquila Digital Community The Aquila Digital Community Faculty Publications Recommended Citation Recommended Citation Recommended Citation Recommended Citation Xie, J., Liu, R., Luttrell, J., Zhang, C. (2019). Deep Learning Based Analysis of Histopathological Images of Breast Cancer. Frontiers in Genetics, 10, 1-19. Available at: https://aquila.usm.edu/fac_pubs/15838 Available at: https://aquila.usm.edu/fac_pubs/15838 This Article is brought to you for free and open access by The Aquila Digital Community. It has been accepted for inclusion in Faculty Publications by an authorized administrator of The Aquila Digital Community. For more information, please contact aquilastaff@usm.edu. ORIGINAL RESEARCH published: 19 February 2019 doi: 10.3389/fgene.2019.00080 Deep Learning Based Analysis of Histopathological Images of Breast Cancer Juanying Xie 1, Ran Liu 1, Joseph Luttrell IV 2 and Chaoyang Zhang 2 1 School of Computer Science, Shaanxi Normal University, Xi’an, China, 2 School of Computing Sciences and Computer Engineering, University of Southern Mississippi, Hattiesburg, MS, United States Juanying Xie 1, Ran Liu 1, Joseph Luttrell IV 2 and Chaoyang Zhang 2 1 School of Computer Science, Shaanxi Normal University, Xi’an, China, 2 School of Computing Sciences and Computer Engineering, University of Southern Mississippi, Hattiesburg, MS, United States Breast cancer is associated with the highest morbidity rates for cancer diagnoses in the world and has become a major public health issue. Early diagnosis can increase the chance of successful treatment and survival. However, it is a very challenging and time-consuming task that relies on the experience of pathologists. The automatic diagnosis of breast cancer by analyzing histopathological images plays a signiﬁcant role for patients and their prognosis. However, traditional feature extraction methods can only extract some low-level features of images, and prior knowledge is necessary to select useful features, which can be greatly affected by humans. Deep learning techniques can extract high-level abstract features from images automatically. Therefore, we introduce it to analyze histopathological images of breast cancer via supervised and unsupervised deep convolutional neural networks. First, we adapted Inception_V3 and Inception_ResNet_V2 architectures to the binary and multi-class issues of breast cancer histopathological image classiﬁcation by utilizing transfer learning techniques. Then, to overcome the inﬂuence from the imbalanced histopathological images in subclasses, we balanced the subclasses with Ductal Carcinoma as the baseline by turning images up and down, right and left, and rotating them counterclockwise by 90 and 180 degrees. Our experimental results of the supervised histopathological image classiﬁcation of breast cancer and the comparison to the results from other studies demonstrate that Inception_V3 and Inception_ResNet_V2 based histopathological image classiﬁcation of breast cancer is superior to the existing methods. Furthermore, these ﬁndings show that Inception_ResNet_V2 network is the best deep learning architecture so far for diagnosing breast cancers by analyzing histopathological images. Therefore, we used Inception_ResNet_V2 to extract features from breast cancer histopathological images to perform unsupervised analysis of the images. We also constructed a new autoencoder network to transform the features extracted by Inception_ResNet_V2 to a low dimensional space to do clustering analysis of the images. The experimental results demonstrate that using our proposed autoencoder network results in better clustering results than those based on features extracted only by Inception_ResNet_V2 network. All of our experimental results demonstrate that Inception_ResNet_V2 network based deep transfer learning provides a new means of performing analysis of histopathological images of breast cancer. Keywords: histopathological images, breast cancer, deep convolutional neural networks, autoencoder, transfer learning, classiﬁcation, clustering Edited by: Alfredo Pulvirenti, Università degli Studi di Catania, Italy Reviewed by: Omar Al-Kadi, Yale University, United States Hamid Reza Marateb, Universitat Politecnica de Catalunya, Spain Correspondence: Juanying Xie xiejuany@snnu.edu.cn Chaoyang Zhang chaoyang.zhang@usm.edu Specialty section: This article was submitted to Bioinformatics and Computational Biology, a section of the journal Frontiers in Genetics Received: 26 September 2018 Accepted: 28 January 2019 Published: 19 February 2019 Citation: Xie J, Liu R, Luttrell IV J and Zhang C (2019) Deep Learning Based Analysis of Histopathological Images of Breast Cancer. Front. Genet. 10:80. doi: 10.3389/fgene.2019.00080 Specialty section: This article was submitted to Bioinformatics and Computational Biology, a section of the journal Frontiers in Genetics Received: 26 September 2018 Accepted: 28 January 2019 Published: 19 February 2019 RELATED WORKS However, the analysis of the histopathological images is a diﬃcult and time-consuming task that requires the knowledge of professionals. Furthermore, the outcome of the analysis may be aﬀected by the level of experience of the pathologists involved. Therefore, computer-aided (Aswathy and Jagannath, 2017) analysis of histopathological images plays a signiﬁcant role in the diagnosis of breast cancer and its prognosis. However, the process of developing tools for performing this analysis is impeded by the following challenges. First, histopathological images of breast cancer are ﬁne-grained, high-resolution images that depict rich geometric structures and complex textures. The variability within a class and the consistency between classes can make classiﬁcation extremely diﬃcult, especially when dealing with multiple classes. The second challenge is the limitations of feature extraction methods for histopathological images of breast cancer. Traditional feature extraction methods, such as scale-invariant feature transform (SIFT) (Lowe, 1999) and gray- level co-occurrence matrix (GLCM) (Haralick et al., 1973), all rely on supervised information. Furthermore, prior knowledge of data is needed to select useful features, which makes the feature extraction eﬃciency very low and the computational load very heavy. In the end, the ﬁnal extracted features are only some low- level and unrepresentative features of histopathological images. Consequently, this can lead to the ﬁnal model producing poor classiﬁcation results. Breast cancer diagnosis based on image analysis has been studied for more than 40 years, and there have been several notable research achievements in the area. These studies can be divided into two categories according to their methods: one is based on traditional machine learning methods, and the other is based on deep learning methods. The former category is mainly focused on small datasets of breast cancer images and is based on labor intensive and comparatively low-performing, abstract features. The latter category can deal with big data and can also extract much more abstract features from data automatically. y For example, Zhang et al. (2013) proposed a new cascade random subspace ensemble scheme with rejection options for microscopic biopsy image classiﬁcation in 2012. This classiﬁcation system consists of two random subspace classiﬁer ensembles. The ﬁrst ensemble consists of a set of support vector machines which correspond to the K binary classiﬁcation problems transformed from the original K-class classiﬁcation problem (K = 3). The second ensemble consists of a Multi-Layer Perceptron ensemble which focuses on rejected samples from the ﬁrst ensemble. Citation: Xie J, Liu R, Luttrell IV J and Zhang C (2019) Deep Learning Based Analysis of Histopathological Images of Breast Cancer. Front. Genet. 10:80. doi: 10.3389/fgene.2019.00080 February 2019 \| Volume 10 \| Article 80 Frontiers in Genetics \| www.frontiersin.org 1 Breast Cancer Histopathological Images Analysis Xie et al. INTRODUCTION of breast cancer, this paper analyzes histopathological images of breast cancer using deep learning techniques. On one hand, we use advanced deep convolutional neural networks, including Inception_V3 (Szegedy et al., 2016) and Inception_ResNet_V2 (Szegedy et al., 2017), combined with transfer learning techniques to classify the histopathological images of breast cancer (Pan and Yang, 2010). On the other hand, by combining deep learning with clustering and utilizing the dimension-reduction functionality of the autoencoder network (Hinton and Salakhutdinov, 2006), we propose a new autoencoder network structure to apply non-linear transformations to features in histopathological images of breast cancer extracted by the Inception_ResNet_V2 network. This eﬀectively maps the extracted features to a lower dimensional space. The newly obtained features are then used as input for the classical clustering algorithm known as K- means (MacQueen, 1967) to perform clustering analysis on histopathological images of breast cancer. Also, we designed a number of comparable experiments to verify the validity of our proposed method of histopathological image analysis of breast cancer images based on deep learning techniques. Cancers have become one of the major public health issues. According to statistics by the IARC (International Agency for Research on Cancer) from the WHO (World Health Organization), and GBD (Global Burden of Disease Cancer Collaboration), cancer cases increased by 28% between 2006 and 2016, and there will be 2.7 million new cancer cases emerging in 2030 (Boyle and Levin, 2008; Moraga-Serrano, 2018). Among the various types of cancer, breast cancer is one of the most common and deadly in women (1.7 million incident cases, 535,000 deaths, and 14.9 million disability-adjusted life years) (Moraga-Serrano, 2018). Therefore, the diagnosis of breast cancer has become very important. Although the diagnosis of breast cancers has been performed for more than 40 years using X-ray, MRI (Magnetic Resonance Imaging), and ultrasound etc. (Stenkvist et al., 1978), biopsy techniques are still the main methods relied on to diagnose breast cancer correctly. Common biopsy techniques include ﬁne- needle aspiration, vacuum-assisted biopsy and surgical biopsy. The process involves collecting samples of cells or tissues, ﬁxing them on the microscope slide, and then staining them (Veta et al., 2014). After that, the histopathological images are analyzed and the diagnosis is made by pathologists (Spanhol et al., 2016a). Frontiers in Genetics \| www.frontiersin.org RELATED WORKS This system was tested on a database composed of 361 images, of which 119 were normal tissue, 102 were carcinoma in situ, and 140 were lobular carcinoma or invasive ductal. The authors randomly split the images into training and testing sets, with 20% of each class’ images used for testing and the rest used for training. It obtained a high classiﬁcation accuracy of 99.25% and a high classiﬁcation reliability of 97.65% with a small rejection rate of 1.94%. In 2013, Kowal et al. (2013) used four clustering algorithms to perform nuclei segmentation for 500 images from 50 patients with breast cancer. Then, they used three diﬀerent classiﬁcation approaches to classify these images into benign and malignant tumors. Among 500 images, there were 25 benign and 25 malignant cases with 10 images per case. They achieved classiﬁcation accuracy between 96 and Deep learning techniques have the power to automatically extract features, retrieve information from data automatically, and learn advanced abstract representations of data. They can solve the problems of traditional feature extraction and have been successfully applied in computer vision (He et al., 2015; Xie et al., 2018), biomedical science (Gulshan et al., 2016; Esteva et al., 2017) and many other ﬁelds. In view of the powerful feature extraction advantages of deep learning and the challenges in histopathological image analysis February 2019 \| Volume 10 \| Article 80 2 Breast Cancer Histopathological Images Analysis Xie et al. 100% using a 50-fold cross-validation technique. In the same year, Filipczuk et al. (2013) presented a breast cancer diagnosis system based on the analysis of cytological images of ﬁne needle biopsies to discriminate between benign or malignant biopsies. Four traditional machine learning methods including KNN (K- nearest neighbor with K = 5), NB (naive Bayes classiﬁer with kernel density estimate), DT (decision tree) and SVM (support vector machine with Gaussian radial basis function kernel and scaling factor σ = 0.9) were used to build the classiﬁers of the biopsies with 25 features of the nuclei. These classiﬁers were tested on a set of 737 microscopic images of ﬁne needle biopsies obtained from 67 patients, which contained 25 benign (275 images) and 42 malignant (462 images) cases. The best reported eﬀectiveness is up to 98.51%. In 2014, George et al. (2014) proposed a diagnosis system for breast cancer using nuclear segmentation based on cytological images. RELATED WORKS Four classiﬁcation models were used, including MLP (multilayer perceptron using the backpropagation algorithm), PNN (probabilistic neural network), LVQ (learning vector quantization), and SVM. The parameters for each model can be found in Table 5 in George et al. (2014). The classiﬁcation accuracy using 10-fold cross- validation is 76∼94% with only 92 images, including 45 images of benign tumors and 47 images of malignant tumors. In 2016, a performance comparison was conducted by Asri et al. (2016) between four machine learning algorithms, including SVM, DT, NB and KNN, on the Wisconsin Breast Cancer dataset, which contains 699 instances (including 458 benign and 241 malignant cases). Experimental results demonstrated that SVM achieved the highest accuracy of 97.13% with 10-fold cross-validation. requiring the manual design of features by a domain expert (Spanhol et al., 2016b). As a consequence, Spanhol et al. (2016b) classiﬁed histopathological images of breast cancer from BreaKHis using a variation of the AlexNet (Krizhevsky et al., 2012) convolutional neural network that improved classiﬁcation accuracy by 4–6%. Bayramoglu et al. (2016) proposed to classify breast cancer histopathological images independently of their magniﬁcations using CNN (convolutional neural networks). They proposed two diﬀerent architectures: the single task CNN used to predict malignancy, and the multi-task CNN used to predict both malignancy and image magniﬁcation level simultaneously. Evaluations were carried out on the BreaKHis dataset, and the experimental results were competitive with the state-of-the-art results obtained from traditional machine learning methods. However, the above studies on the BreaKHis dataset only focus on the binary classiﬁcation problem. Multi-class classiﬁcation studies on histopathological images of breast cancer can provide more reliable information for diagnosis and prognosis. As a result, Araújo et al. (2017) proposed a CNN based method to classify the hematoxylin and eosin stained breast biopsy images from a dataset composed of 269 images into four classes (normal tissue, benign lesion, in situ carcinoma and invasive carcinoma), and into two classes (carcinoma and non- carcinoma), respectively. An SVM classiﬁer with the radial basis kernel function was trained using the features extracted by CNN. The accuracies of the SVM for the four-class and two- class classiﬁcation problems are 77.8–83.3%, respectively. To realize the development of a system for diagnosing breast cancer using multi-class classiﬁcation on BreaKHis, Han et al. (2017) proposed a class structure-based deep convolutional network to provide an accurate and reliable solution for breast cancer multi- class classiﬁcation by using hierarchical feature representation. RELATED WORKS Using these techniques, they were able to achieve multi-class classiﬁcation of breast cancer with a maximum accuracy of 95.9%. This study is important for precise treatment of breast cancer. In addition, Nawaz et al. (2018) presented a DenseNet based model for multi-class breast cancer classiﬁcation to predict the subclass of the tumors. The experimental results on BreaKHis achieved the accuracy of 95.4%. After that, Motlagh et al. (2018) used the pre-trained model of ResNet_V1_152 (He et al., 2016) to perform diagnosis of benign and malignant tumors as well as diagnosis based on multi-class classiﬁcation of various subtypes of histopathological images of breast cancer in BreaKHis. They were able to achieve an accuracy of 98.7–96.4% for binary classiﬁcation and multi-class classiﬁcation, respectively. g y However, the above breast cancer diagnosis studies focused on Whole-Slide Imaging (Zhang et al., 2013, 2014). Since the operation of Whole-Slide Imaging is complex and expensive, many studies based on this technique use small datasets and achieve poor generalization performance. To solve these problems, Spanhol et al. (2016a) published a breast cancer dataset called BreaKHis in 2016. BreaKHis contains 7,909 histopathological images of breast cancer from 82 patients. The authors used 6 diﬀerent feature descriptors and 4 diﬀerent traditional machine learning methods, including 1-NN (1 Nearest Neighbor), QDA (Quadratic Discriminant Analysis), RF (Random Forest), and SVM with the Gaussian kernel function, to perform binary diagnosis of benign and malignant tumors. The classiﬁcation accuracy is between 80 and 85% using 5-fold cross-validation. Although traditional machine learning methods have made great achievements in analyzing histopathological images of breast cancer and even in dealing with relatively large datasets, their performance is heavily dependent on the choice of data representation (or features) for the task they are trained to perform. Furthermore, they are unable to extract and organize discriminative information from data (Bengio et al., 2013). Deep learning methods typically are neural network based learning machines with much more layers than the usual neural network. They have been widely used in the medical ﬁeld since they can automatically yield more abstract—and ultimately more useful— representations (Bengio et al., 2013). That is, they can extract the discriminative information or features from data without Although there are 7,909 histopathological images from 82 patients in BreaKHis, the number of images is far from enough for eﬀectively using deep learning techniques. Frontiers in Genetics \| www.frontiersin.org RELATED WORKS Therefore, we proposed to combine transfer learning techniques with deep learning to perform breast cancer diagnosis using the relatively small number of histopathological images (7,909) from the BreaKHis dataset. The Inception_V3 (Szegedy et al., 2016) and Inception_ResNet_V2 (Szegedy et al., 2017) networks were proposed by Szegedy et al. (2016, 2017), respectively. In the 2012 February 2019 \| Volume 10 \| Article 80 3 Breast Cancer Histopathological Images Analysis Xie et al. Xie et al. ImageNet Large Scale Visual Recognition Challenge (ILSVRC) competition, the Inception_V3 network achieved 78.0–93.9% accuracy in top-1 and top-5 metrics, respectively, while the Inception_ResNet_V2 achieved 80.4–95.3% accuracy in the same evaluation. (PT), and Tubular Adenoma (TA). Malignant tumors include Ductal Carcinoma (DC), Lobular Carcinoma (LC), Mucinous Carcinoma (MC), and Papillary Carcinoma (PC). Sample descriptions for the BreaKHis dataset are shown in Table 1. Since the input sizes of Inception_V3 and Inception_ResNet_V2 networks used in this paper are both 299 × 299, each of the histopathological images of breast cancer must be transformed into a 299 × 299 image to match the required input size of the network structure. Some image preprocessing methods in the TensorFlow framework were used in the transforming process, including cutting the border box, adjusting image size, and adjusting saturation, etc. In this way, a 3-channel image conforming to the input size of the model was generated, and the pixel values of each channel were normalized to the interval of [−1, 1]. In order to ensure the universality of the experimental results in the classiﬁcation task, the datasets of the four magniﬁcation factors were randomly partitioned into training and testing subsets according to the proportion of 7:3. One common method for performing transfer learning (Pan and Yang, 2010) involves obtaining the basic parameters for training a deep learning model by pre-training on large data sets, such as ImageNet, and then using the data set of the new target task to retrain the last fully-connected layer of the model. This process can achieve good results even on small data sets. Therefore, we adopt two deep convolutional neural networks, speciﬁcally Inception_V3 and Inception_Resnet_V2, to study the diagnosis of breast cancer in the BreaKHis dataset via transfer learning techniques. To solve the unbalanced distribution of samples of histopathological images of breast cancer, the BreaKHis dataset was expanded by rotation, inversion, and several other data augmentation techniques. Classiﬁcation Analysis This subsection will discuss our experiments of classifying histopathological images of breast cancer using the deep learning models of Inception_V3 (Szegedy et al., 2016) and Inception_ResNet_V2 (Szegedy et al., 2017) as well as the analyses of our experimental results. Network Structures for Classiﬁcation The Inception_V3 (Szegedy et al., 2016) and Inception_ResNet_V2 (Szegedy et al., 2017) networks, proposed by Szegedy et al. in 2016 and 2017, respectively, were adopted in our experiments. It was demonstrated in the ILSVRC competition that Inception_ResNet_V2 could defeat the Inception_V3 network when applied to big data. An important diﬀerence between the Inception_V3 and Inception_ResNet_V2 networks is that the latter is equipped with residual connections. To test whether the experimental results from Inception_ResNet_V2 are superior to those from Inception_V3 on small datasets or not, these two networks are adopted in this paper to perform classiﬁcation of the histopathological images of breast cancer. The network structures are shown in Figure 1. RELATED WORKS The Inception_ResNet_V2 network was chosen to conduct binary and multi-class classiﬁcation diagnosis on the expanded set of histopathological breast cancer images for its better performance on the original dataset of BreaKHis compared to that of Inception_V3. The powerful feature extraction capability of the Inception_ResNet_V2 network was used to extract features of the histopathological images of breast cancer for the linear kernel SVM and 1-NN classiﬁers. The image features extracted by the Inception_ResNet_V2 network are also used as the input of the K-means algorithm to do clustering analysis for the BreaKHis dataset. Furthermore, a new autoencoder deep learning model is constructed to apply a non-linear transformation to the image features extracted by Inception_ResNet_V2 network in order to get the low-dimensional features of the image, and to do clustering analysis for BreaKHis dataset using the K- means algorithm. Datasets The dataset named BreaKHis used in this article was published by Spanhol et al. (2016a) in 2016. It is composed of 7,909 histopathological images from 82 clinical breast cancer patients. The database can be accessed through the link http:// web.inf.ufpr.br/vri/breast-cancer-database. To save the original organization structure and molecular composition, each image was taken by a pathologist from a patient’s breast tissue section using a surgical biopsy. Then, the images were collected via haematoxylin and eosin staining. Finally, the real class label was given to each image by pathologists via their observations of the images from a microscope. All the histopathological images of breast cancer are 3 channel RGB micrographs with a size of 700 × 460. Since objective lenses of diﬀerent multiples were used in collecting these histopathological images of breast cancer, the entire dataset comprised four diﬀerent sub-datasets, namely 40, 100, 200, and 400X. All of these sub-datasets are classiﬁed into benign and malignant tumors. Therefore, both benign and malignant tumors have four diﬀerent subsets. Benign tumors include Adenosis (A), Fibroadenoma (F), Phyllodes Tumor It can be seen from Figure 1 that the structures of the two networks are very similar. The ﬁrst several layers are characteristic transformation via the traditional convolutional layers and the pooling layers, and the middle part is composed of multiple Inception modules stacked together. The results are ﬁnally output through the fully-connected layer using the Softmax function. One of the main diﬀerences between the Inception_V3 and Inception_ResNet_V2 networks lies in the diﬀering composition of the two networks’ Inception modules. To enhance the network’s adaptability to diﬀerent convolution kernels, each Inception module of the Inception_V3 network is composed of ﬁlters with diﬀerent sizes including 1 × 1, 1 × 3, 3 × 1. For the Inception_ResNet_V2 network, to avoid the deterioration of the network gradient that is often associated with an increase in the number of layers, a residual unit is added to each Inception module. Besides using ﬁlters of diﬀerent February 2019 \| Volume 10 \| Article 80 Frontiers in Genetics \| www.frontiersin.org 4 Breast Cancer Histopathological Images Analysis Xie et al. TABLE 1 \| Image distribution of different subclasses in different magniﬁcation factors. Datasets Magniﬁcation Benign Malignant Total A F PT TA DC LC MC PC 40X 114 253 109 149 864 156 205 145 1,995 100X 113 260 121 150 903 170 222 142 2,081 200X 111 264 108 140 896 163 196 135 2,013 400X 106 237 115 130 788 137 169 138 1,820 Total 444 1,014 453 569 3,451 626 792 560 7,909 #Patients 4 10 3 7 38 5 9 6 82 FIGURE 1 \| The network structures, (A) Inception_V3, (B) Inception_ResNet_V2. TABLE 1 \| Image distribution of different subclasses in different magniﬁcation factors. FIGURE 1 \| The network structures, (A) Inception_V3, (B) Inception_ResNet_V2. FIGURE 1 \| The network structures, (A) Inception_V3, (B) Inception_ResNet_V2. train a complex deep network from scratch with only a small dataset. Furthermore, there are not any existing principles to design a network structure for a speciﬁc task. What we can do is adopt the model and the parameters obtained by other researchers via time-consuming and computationally intensive training on the very large image dataset of ImageNet and use the knowledge it has gained as pre-training for our speciﬁc research task. Then, we can retrain the last deﬁned fully-connected layer of the model using only a relatively small amount of data to achieve good results for our target task. sizes in the network, the deterioration caused by increasing layers can also be solved by jumping layers as allowed by the use of residual connections. Figure 2 displays the diﬀerences in the construction of the Inception module with a size of 8 × 8 between Inception_V3 and Inception_ResNet_V2. The other details can be found in the original references (Szegedy et al., 2016, 2017). Frontiers in Genetics \| www.frontiersin.org decayed_learning_rate = learning_rate decayed_learning_rate = learning_rate ×decay_rate(global_step/decay_steps) (1) (1) Transfer Learning Transfer learning (Pan and Yang, 2010) emerges from deep learning. It is well-known that it is typically impossible to February 2019 \| Volume 10 \| Article 80 Frontiers in Genetics \| www.frontiersin.org 5 Breast Cancer Histopathological Images Analysis Xie et al. Xie et al. FIGURE 2 \| The inception module of size 8 × 8 in two networks, (A) Inception_V3, (B) Inception_ResNet_V2. FIGURE 2 \| The inception module of size 8 × 8 in two networks, (A) Inception_V3, (B) Inception_ResNet_V2. FIGURE 2 \| The inception module of size 8 × 8 in two networks, (A) Inception_V3, (B) Inception_ResNet_V2. FIGURE 3 \| The Inception_ResNet_V2 network structure for transfer learning. ption ResNet V2. FIGURE 2 \| The inception module of size 8 × 8 in two networks, (A) Inception_V3, (B) Inception_ResNet_V2. FIGURE 3 \| The Inception_ResNet_V2 network structure for transfer learning. FIGURE 3 \| The Inception_ResNet_V2 network structure for transfer learning. Transfer learning is adopted in this paper to classify the histopathological images of breast cancer using Inception_V3 and Inception_ResNet_V2 networks. We ﬁrst downloaded the models and parameters of Inception_V3 and Inception_ResNet_V2 networks trained on the ImageNet dataset. The dataset is composed of about 1.2 million training images, 50,000 validation images, and 100,000 testing images. This comprises a total of 1,000 diﬀerent categories. Then, we froze all of the parameters before the last layer of the networks. We modiﬁed the number of neurons of the last fully-connected layer as 2 for binary classiﬁcation and 8 for multi-class classiﬁcation. After that, the parameters of the fully-connected layer are trained on the histopathological images of breast cancer. The modiﬁed network structure of the Inception_ResNet_V2 network is shown in Figure 3. The modiﬁed Inception_V3 network structure is similar, so it is omitted. provide more stability at the later stage and makes it easier to obtain the optimal solution. The decay coeﬃcient is set as 0.7 (Bergstra and Bengio, 2012), and the decay speed is set so that the decay occurs every two epochs. The speciﬁc decay process is shown in (1), where decayed_learning_rate is the current learning rate, learning_rate is the initial learning rate, decay_rate is the decay coeﬃcient, global_step is the current iteration step, and decay_steps is the decay speed. decayed_learning_rate = learning_rate ×decay_rate(global_step/decay_steps) (1) Evaluation Criteria for Classiﬁcation Results Evaluation Criteria for Classiﬁcation Results To evaluate the performance of the classiﬁcation model more accurately and comprehensively, the classiﬁcation results are evaluated by some popular benchmark metrics, including sensitivity (Se), speciﬁcity (Sp), positive predictive value (PPV), diagnostic odds ratio (DOR), F1 measure (F1), area under the receiver operating characteristic curve (AUC), Kappa criteria (Kappa), Macro-F1, Micro-F1, image level test accuracy (ACC_IL), and patient level test accuracy (ACC_PL). The latter two criteria were proposed in (5). The Macro-F1 and Micro-F1 are two variations of F1 for multi-class classiﬁcation problems. Macro-F1 is the average of F1 for each class. Micro-F1 is deﬁned as F1 but depending on the precision and recall deﬁned by the sum of TP (true positive), FP (false positive), and FN (false Our classiﬁcation process was developed based on the TensorFlow deep learning framework. The Adam (adaptive moment estimation) (Kingma and Ba, 2014) algorithm was used in the training process to perform optimization by iterating through 70 epochs using the histopathological image dataset of breast cancer. The batch_size is set to 32 in the experiments, and the initial learning rate is 0.0002 (Bergstra and Bengio, 2012). Then, the exponential decay method is adopted to reduce the learning rate and ensure that the model moves through iterations quickly at the initial training stage. This also helps to February 2019 \| Volume 10 \| Article 80 6 Breast Cancer Histopathological Images Analysis Xie et al. negative) for all classes. The deﬁnitions of the criteria are shown in Equations (2–9). under the ROC curve, which is another widely used metric for evaluating binary classiﬁcation models. The range of AUC is [0, 1] (Bradley, 1997), with higher values representing better model performance. We calculate AUC in our experiments by calling the roc_auc_score function from the Scikit-learn library that is available as a Python package (sklearn). Equation (9) is the Kappa coeﬃcient, where P0 is the image level test accuracy deﬁned in (6), and Pe is the ratio of the sum of the product of the number of real images in each category and the predicted number of images in that category to the square of the total samples. The calculation of the Kappa coeﬃcient is based on the confusion matrix. Kappa is used for consistency checking, and its value is in the range of [−1, 1]. Clustering Analysis The classiﬁcation analysis of histopathological images of breast cancer based on deep convolutional neural networks is introduced in the previous section. However, this type of classiﬁcation is supervised learning and requires experienced pathologists to examine the histopathological images of breast cancer and assign labels to them that identify the data as coming from patients or normal people. This is very diﬃcult, time- consuming, and expensive work, especially with the increasing number of samples in the dataset. On the contrary, unsupervised learning, speciﬁcally clustering, does not need any labels for samples. It only uses the similarities between samples to group them into diﬀerent clusters, such that the samples in the same cluster are similar to each other and dissimilar to those from other clusters. Therefore, we adopt clustering techniques to study the histopathological images of breast cancer. (9) The value of TP in the equations above is the number of images correctly recognized as malignant tumor in the testing subset. The value of TP in the equations above is the number of images correctly recognized as malignant tumor in the testing subset. correctly recognized as malignant tumor in the testing subset. FP is the number of images that were incorrectly recognized as malignant tumor in the testing subset. FN is the number of images incorrectly recognized as benign tumor in the testing subset. TN is the number of images correctly recognized as benign tumor in the testing subset. Therefore, Se in (2) deﬁnes the ratio of the recognized malignant tumor images to all malignant tumor images in the testing subset. Sp in (3) expresses the ratio of the recognized benign tumor images to all benign tumor images. That is, Se and Sp are the accuracy of the positive and negative class, respectively. PPV in (4) is the ratio of correctly recognized malignant tumor images to all recognized malignant tumor images in the testing subset. In fact, it is the precision in (8). DOR expresses the ratio of the product of TP and TN to the product of FP and FN. It is clear that DOR will become inﬁnity when the related classiﬁer is perfect. It is reported that a diagnosis system is reliable if Se> = 80%, Sp> = 95%, PPV> = 95%, and DOR> = 100 (Ellis, 2010; Colquhoun, 2014). Evaluation Criteria for Classiﬁcation Results It can be divided into six groups representing the following consistency levels: −1∼0.0 (poor), 0.0∼0.20 (slight), 0.21∼0.40 (fair), 0.41∼0.60 (moderate), 0.61∼0.80 (substantial), and 0.81∼1 (almost perfect) (Landis and Koch, 1977). Se = TP TP + FN (2) Sp = TN TN + FP (3) PPV = TP TP + FP (4) DOR = TP × TN FP × FN (5) ACC_IL = Nrec Nall (6) Se = TP TP + FN (2) Sp = TN TN + FP (3) PPV = TP TP + FP (4) DOR = TP × TN FP × FN (5) ACC_IL = Nrec Nall (6) (2) (3) (6) ACC_PL = P Patient Score Total Number of Patients, Patient Score = Nrec NP (7) ( ) F1 = 2 × precision × recall precision + recall , recall = TP TP + FN (8) Kappa = p0 −pe 1 −pe , p0 = Nrec Nall , Pe = P Ntrue_i × Npre_i Nall × Nall (9) Clustering Analysis Equation (6) deﬁnes image level test accuracy (ACC_IL) by the ratio of Nrec (the number of the histopathological images of breast cancer correctly identiﬁed in the testing subset), to Nall (the total number the histopathological images of breast cancers in the testing subset). Equation (7) deﬁnes patient level test accuracy (ACC_PL), that is, the ratio of the sum of patient score to the total number of patients in the testing subset. Here, the patient score is the ratio of Nrec to NP, that is, the ratio of correctly identiﬁed images of patient P to all the images of patient P in the testing subset. Equation (8) describes a popular metric known as the harmonic mean of precision and recall. Here, precision is the same as PPV deﬁned as the ratio of correctly recognized malignant tumor images to all recognized malignant tumor images in the testing subset, and recall is the ratio of correctly recognized malignant tumor images to the true number of malignant tumor images in the testing subset. AUC is the area Frontiers in Genetics \| www.frontiersin.org February 2019 \| Volume 10 \| Article 80 Network Structures for Clustering The Inception_ResNet_V2 network is adopted to extract features for performing clustering analysis of the histopathological images of breast cancer because of its excellent performance when classifying these images using its advantage of extracting features automatically. Each histopathological image of breast cancer can be well-expressed by the extracted features of the 1,536-dimension vector produced by the Inception_ResNet_V2 network before its ﬁnal classiﬁcation layer. The extracted feature vectors are used as input to a clustering algorithm in order to perform clustering analysis on the histopathological images of breast cancer. The very simple and fast, typical clustering algorithm K- means is adopted in this paper to perform this clustering analysis. To determine the proper value of K for the K-means algorithm, the internal criterion metric SSE (Silhouette Score) (Rousseeuw, 1987) is adopted to search for the optimal K. The features extracted by the Inception_ResNet_V2 network for each breast cancer histopathological image are thought of as a representation of the images, and the K-means clustering algorithm is adopted to cluster the breast cancer histopathological images into clusters. Also, in order to get better clustering results and to visualize the clustering results, we constructed a new autoencoder network February 2019 \| Volume 10 \| Article 80 7 Xie et al. Breast Cancer Histopathological Images Analysis Xie et al. FIGURE 4 \| The network structures of our proposed autoencoder and its combination with Inception_ResNet_V2, (A) Autoencoder network, (B) Inception_ResNet_V2 and autoencoder network. FIGURE 4 \| The network structures of our proposed autoencoder and its combination with Inception_ResNet_V2, (A) Autoencoder network, (B) Inception_ResNet_V2 and autoencoder network. FIGURE 4 \| The network structures of our proposed autoencoder and its combination with Inception_ResNet_V2, (A) Autoencoder network, (B) Inception_ResNet_V2 and autoencoder network. number of clusters of the histopathological images of breast cancer. Then, after the clustering results have been obtained by K-means, it is used to evaluate the clustering results together with the aforementioned external metrics. Equation (10) gives the Silhouette value of sample i. to map the 1,536-dimension vector to a 2-dimension vector via a non-linear transformation. In this way, the breast cancer histopathological images can be represented in a very low dimensional space. Figure 4A displays the autoencoder network we constructed in our experiments. There are 2 encode layers with neuron sizes of 500 and 2, respectively, and there are 2 corresponding decode layers to reconstruct the original input. Network Structures for Clustering Using this autoencoder, the 1,536-dimension feature vector extracted by the Inception_ResNet_V2 network for a breast cancer histopathological image will be transformed to 2-dimenision feature vector via training the layers depicted in Figure 4A. Then, the 2-dimension feature vector is used as input for K-means which performs the clustering analysis for histopathological images of breast cancer. The entire network is shown in Figure 4B. s (i) = b (i) −a (i) max a (i) , b (i) (10) (10) Here, b (i) is the smallest average distance of sample i to all samples in any other cluster to which sample i does not belong. a (i) is the mean distance from sample i to all other samples within the same cluster, and s (i) is the Silhouette value of sample i. The average s (i) of all samples in a cluster is a measure of how tightly grouped all the samples in the cluster are. Therefore, the average s (i) over all samples in an entire dataset is a measure of how appropriately the samples have been clustered; that is what is called the SSE metric. Here, b (i) is the smallest average distance of sample i to all samples in any other cluster to which sample i does not belong. a (i) is the mean distance from sample i to all other samples within the same cluster, and s (i) is the Silhouette value of sample i. The average s (i) of all samples in a cluster is a measure of how tightly grouped all the samples in the cluster are. Therefore, the average s (i) over all samples in an entire dataset is a measure of how appropriately the samples have been clustered; that is what is called the SSE metric. Frontiers in Genetics \| www.frontiersin.org February 2019 \| Volume 10 \| Article 80 Evaluation Criteria of Clustering Results The evaluation criteria of clustering results comprise internal and external metrics. The internal metrics are independent of the external information, so they are always used to ﬁnd the true number of clusters in a dataset. The external metrics depend on the true pattern of the dataset. Some of the most common external metrics are clustering accuracy (ACC), adjusted rand index (ARI) (Hubert and Arabie, 1985), and adjusted mutual information (AMI) (Vinh et al., 2010). The external metrics used in this paper are ACC, ARI (Hubert and Arabie, 1985) and AMI (Vinh et al., 2010). It was reported that ARI is one of the best external metrics (Hubert and Arabie, 1985). ARI is deﬁned in (11) and uses the following variables: a (the number of pairs of samples in the same cluster before and after clustering), b (the pairs of samples in the same cluster while partitioned into diﬀerent clusters by the clustering algorithm), c (the pairs of samples that are from diﬀerent clusters but are grouped into the same cluster The internal metric SSE (Silhouette Score) (Rousseeuw, 1987) is used in our experiments. It is ﬁrst used to ﬁnd the most proper February 2019 \| Volume 10 \| Article 80 Frontiers in Genetics \| www.frontiersin.org 8 Breast Cancer Histopathological Images Analysis Xie et al. incorrectly by the clustering algorithm), and d (the number of pairs of samples from diﬀerent clusters that are still in diﬀerent clusters after clustering). The AMI is deﬁned in (12), where U is the original partition and V is the clustering of a clustering algorithm. Here, MI (U, V) denotes the mutual information between two partitions U and V, and E {MI (U, V)} represents the expected mutual information between the original partition U and the clustering V. H (U) , H (V) are the entropy of the original partition U and the clustering V, respectively. AMI is a variation of mutual information and can be used to compare the clustering V of a clustering algorithm and the true pattern U of the dataset. It corrects the eﬀect of agreement solely due to chance between the clustering and the original pattern. This is similar to the way that the adjusted Rand index corrects the Rand index. there are 8 subclasses in total, including 4 benign tumors (A, F, PT, and TA) and 4 malignant tumors (DC, LC, MC, and PC). EXPERIMENTAL RESULTS AND ANALYSIS The experimental results in Table 2 show that Se>95%, Sp>90%, PPV>95%, and DOR>100 on each dataset regardless of magniﬁcation factor and network structure (Inception_V3 or Inception_ResNet_V2). The results from the Inception_ResNet_V2 network show that Se>98%, Sp>92%, PPV>96%, and DOR>100, especially on the 40X dataset where Se >98%, Sp>96%, PPV>98%, and DOR>100. Considering research which suggests that a diagnosis system is reliable when Se> = 80%, Sp> = 95%, PPV> = 95%, and DOR> = 100 (Ellis, 2010; Colquhoun, 2014), we can say that our breast cancer diagnosis system based on the 40X dataset and the Inception_ResNet_V2 network is very reliable. The diagnosis system based on the Incepiton_V3 network is also comparatively reliable. The section will present our classiﬁcation and clustering experimental results on the 7,909 histopathological images of breast cancer from the BreaKHis dataset and provide some analyses and discussions of the results. Evaluation Criteria of Clustering Results The available studies for the histopathological images of breast cancer only focus on binary classiﬁcation of the images. However, multi-class classiﬁcation is more signiﬁcant than binary classiﬁcation for providing accurate treatment and prognosis for breast cancer patients. Therefore, we did a multi-class classiﬁcation diagnosis study on the histopathological images of breast cancer by using Inception_V3 and Inception_ResNet_V2 with transfer learning techniques. The experimental results of our multi-class classiﬁcation of histopathological images of breast cancer are shown in the bottom half in Table 2 in terms of ACC_IL, ACC_PL, Macro-F1, Micro-F1 and Kappa. The experimental results in Table 2 show that the Inception_ResNet_V2 network can get better results in all evaluation metrics compared to the Inception_V3 network, regardless of binary or multi-class classiﬁcation (which is indicated by the red underline). One reason for this is that residual connections are added to the Inception_ResNet_V2 network, which avoids the vanishing gradient problem typically caused by increasing the number of layers in a network. This also improves the network performance and allows it to extract more informative features from images than Incepiton_V3 can. ARI = 2(ad −bc) (a + b)(b + d) + (a + c)(c + d) (11) (11) AMI (U, V) = MI (U, V) −E {MI (U, V)} max {H (U) , H (V)} −E {MI (U, V)} (12) (12) We calculate the criteria listed above in our experiments by calling functions embedded in the sklearn library (available as a Python package), such as silhouette_score (SSE), linear_assignment (ACC), adjusted_rand_score (ARI), and adjusted_mutual _info_score (AMI). Furthermore, the experimental results show that all metrics on the 40X dataset are better than those on the other datasets with any other magniﬁcation factors, which is shown in black font. These results are in agreement with those reported in (5). The reason for this should be the 40X dataset containing more signiﬁcant characteristics of breast cancer. Classiﬁcation Results This subsection will present and discuss all of the classiﬁcation results of histopathological images of breast cancer from BreaKHis dataset provided by Spanhol (Spanhol et al., 2016a). The experimental results include those conducted on the raw dataset and on the augmented dataset. In addition to this, we provide a comparison between our results and the results produced by other researchers. Frontiers in Genetics \| www.frontiersin.org February 2019 \| Volume 10 \| Article 80 Experiments on the Raw Dataset In addition, the values of AUC and Kappa in Table 2 tell us that our models are perfect and have obtained almost perfect agreement for binary classiﬁcation of histopathological images of breast cancer. The values of Kappa in Table 2 reveal that our models for multi-class classiﬁcation are also perfect. The models based on the Inception_ResNet_V2 network can get perfect agreement for multi-class classiﬁcation of breast cancer histopathological images, except when applied to the 400X dataset (which still achieves substantial agreement). We used the Inception_V3 and Inception_ResNet_V2 networks to perform binary classiﬁcation of histopathological images of breast cancer into benign and malignant tumors via transfer learning. Table 2’s upper part gives the experimental results using Inception_V3 and Inception_ResNet_V2 networks to perform binary classiﬁcation on the histopathological images of breast cancer in terms of Se, Sp, PPV, DOR, ACC_IL, ACC_PL, F1, AUC and Kappa. In the table, INV3 is the abbreviation for the Inception_V3 network, and IRV2 is the abbreviation for the Inception_ResNet_V2 network. Besides the above analysis, we further verify the power of our approaches for analyzing the breast cancer histopathological images using the p-value of AUC and Kappa. The p-value is a probability that measures the statistical signiﬁcance of evidence against the null hypothesis. A lower p-value provides stronger According to the description of the histopathological image dataset of breast cancer, the benign and malignant tumors can be classiﬁed into four diﬀerent subclasses, respectively. So, February 2019 \| Volume 10 \| Article 80 Frontiers in Genetics \| www.frontiersin.org 9 Breast Cancer Histopathological Images Analysis Xie et al. TABLE 2 \| Results of binary and multi-class classiﬁcation using Inception_V3 (INV3) and Inception_ResNet_V2 (IRV2)/%. Experiments on the Raw Dataset Classiﬁcation Network Criteria Magniﬁcation factors 40X 100X 200X 400X Binary INV3 Se 98.00 98.48 99.01 96.41 Sp 94.31 93.46 91.40 90.99 PPV 97.41 96.67 95.88 95.89 DOR 81,233 92,303 106,700 27,105 ACC_IL 96.84 96.76 96.49 94.71 ACC_PL 97.74 94.19 87.21 96.67 F1 97.70 97.56 97.42 96.15 AUC 99.47 99.03 99.29 97.91 Kappa 92.64 92.74 91.95 87.68 IRV2 Se 98.48 98.90 99.13 98.06 Sp 96.63 92.95 92.80 92.10 PPV 98.46 96.45 96.39 96.51 DOR 185,774 118,782 147,138 58,835 ACC_IL 97.90 96.88 96.98 96.98 ACC_PL 98.03 97.07 82.74 88.12 F1 98.47 97.66 97.74 97.28 AUC 99.57 98.84 99.61 98.81 Kappa 95.12 92.96 93.18 91.05 Multi-class INV3 ACC_IL 90.28 85.35 83.99 82.08 ACC_PL 90.44 89.05 80.63 81.08 Macro-F1 88.55 82.59 79.64 77.98 Micro-F1 90.28 85.35 83.99 82.08 Kappa 87.37 80.26 77.91 76.39 IRV2 ACC_IL 92.07 88.06 87.62 84.50 ACC_PL 89.11 88.45 86.07 71.42 Macro-F1 90.89 85.67 84.08 80.13 Micro-F1 92.07 88.06 87.62 84.50 Kappa 89.74 84.03 82.84 79.70 †For each magniﬁcation factor, the underline shows the best result of each evaluation index between the two network structures of INV3 and IRV2. Bold font shows the best result of each evaluation index with respect to the different magniﬁcation factors. TABLE 2 \| Results of binary and multi-class classiﬁcation using Inception_V3 (INV3) and Inception_ResNet_V2 (IRV2)/%. †For each magniﬁcation factor, the underline shows the best result of each evaluation index between the two network structures of INV3 and IRV2. Bold font shows the best result of each evaluation index with respect to the different magniﬁcation factors. †For each magniﬁcation factor, the underline shows the best result of each evaluation index between the two network structures of INV3 and IRV2. Bold font shows the best result of each evaluation index with respect to the different magniﬁcation factors. SEKappa = q p0 × 1 −p0 √ N × 1 −pe (15) ZKappa = Kappa SEKappa (16) SEKappa = q p0 × 1 −p0 √ N × 1 −pe (15) evidence to reject the null hypothesis. Therefore, to determine whether the predictions are due to chance, we calculate the p- values for AUC and Kappa and compare the p-value to the signiﬁcance level α . It is usually set as α = 0.05. We consider both binary and multi-class classiﬁcation of breast cancer histopathological images with Inception_ResNet_V2 when calculating the p-value for AUC and Kappa. Frontiers in Genetics \| www.frontiersin.org Experiments on the Raw Dataset Magniﬁcation Benign Malignant Total A F PT TA DC LC MC PC 40X 798 759 763 894 864 936 1,025 870 6,909 100X 791 780 847 900 903 1,020 1,110 852 7,203 200X 777 792 756 840 896 978 980 810 6,829 400X 742 711 805 780 788 822 845 828 6,321 Total 3,108 3,042 3,171 3,414 3,451 3,756 3,960 3,360 27,262 #Patients 4 10 3 7 38 5 9 6 82 FIGURE 5 \| The change in the loss function during the training of Inception_ResNet_V2 on raw and augmented data with 40 factor magniﬁcation, (A) binary classiﬁcation, (B) multi-class classiﬁcation. TABLE 3 \| The augmented image distribution of different subclasses in different magniﬁcation factors. Magniﬁcation Benign Malignant Total A F PT TA DC LC MC PC 40X 798 759 763 894 864 936 1,025 870 6,909 100X 791 780 847 900 903 1,020 1,110 852 7,203 200X 777 792 756 840 896 978 980 810 6,829 400X 742 711 805 780 788 822 845 828 6,321 Total 3,108 3,042 3,171 3,414 3,451 3,756 3,960 3,360 27,262 #Patients 4 10 3 7 38 5 9 6 82 TABLE 3 \| The augmented image distribution of different subclasses in different magniﬁcation factors. FIGURE 5 \| The change in the loss function during the training of Inception_ResNet_V2 on raw and augmented data with 40 factor magniﬁcation, (A) binary classiﬁcation, (B) multi-class classiﬁcation. of binary or multi-class classiﬁcation. All the p-values for AUC and Kappa are much <0.05. This means that we can reject the null hypothesis (that the predictive result is a random guess) and accept that our prediction is statistically signiﬁcant and not random. This holds true for both our binary and multi-class image classiﬁcation results. experimental results. For each magniﬁcation factor dataset, we chose the DC subclass as the baseline, and ampliﬁed each of the remaining subclasses by turning images up and down, left and right, and using counterclockwise rotation of 90◦and 180◦. After doing this, the sample number of each subclass was approximately the same. The extended datasets are shown in Table 3. Experiments on the Augmented Dataset We randomly partitioned the extended datasets into training and testing subsets in a 7:3 ratio as we did with the original datasets. Then, we used transfer learning to retrain the Inception_ResNet_V2 network to perform eﬀective diagnosis of breast cancer based on histopathological images of breast cancer. Here, we only retrained the Inception_ResNet_V2 network because it performed better than the Incepiton_V3 network on the raw datasets. To compare the diﬀerences in the loss function on the original datasets and on the expansion datasets during the training process, we plotted the value of the loss function changing with the number of epochs in the raw and extended datasets. Here, we only compared the loss function from the Inception_ResNet_V2 network on the 40X dataset in order to observe the changing trend of the loss function. The trends of the other magniﬁcation factor datasets are similar. Figure 5 compared the loss function of the Inception_ResNet_V2 network on the raw and extended datasets, respectively, for binary and multi-class classiﬁcation of histopathological images of breast cancer. Table 4 shows the experimental results on the original and expanded datasets for binary and multi-class classiﬁcation, respectively. The deep learning parameters for both binary and multi-class classiﬁcation remain the same. Experiments on the Augmented Dataset Comparing the results in Table 2 for binary and multi-class classiﬁcation, we can see that the performance of multi-class classiﬁcation is worse than that of the binary classiﬁcation. So, we output the confusion matrix of multi-class classiﬁcation for further analysis. The confusion matrix can be found in the Supplementary Material. From observing this confusion matrix, we can see that many benign tumors are incorrectly classiﬁed as malignant tumors. This causes a high false positive rate. For example, some samples from F are erroneously recognized as being from DC. Also, the diﬀerent subclasses in the same class are often misclassiﬁed, such as samples from LC being recognized as samples from DC. One reason leading to the poor classiﬁcation results for multi-class classiﬁcation is the imbalance in sample distribution. This makes the extracted features unable to thoroughly represent the subclasses with fewer samples. As a result, the samples from the subclass with fewer samples are erroneously classiﬁed into the categories with more samples. Frontiers in Genetics \| www.frontiersin.org Experiments on the Raw Dataset The null hypothesis is “the prediction is a random guess.” The p-values for AUC and Kappa are calculated in Equations (13–16) and the pnorm function in R. It should be noted that for multi-class classiﬁcation, there is only the p-value of Kappa to be calculated. (15) ZKappa = Kappa SEKappa (16) (16) Here, na and nn in (13) are, respectively, the number of abnormal (malignant tumor) and normal (benign tumor) samples (breast cancer histopathological images) in the testing subset. A in (14) is the value of AUC. p0 and pe in (15) are the same as those in (9), and N in (15) is the total number of samples. We convert the z value for AUC in (14) and for Kappa in (16) to the corresponding p-value by using the pnorm function in R. SEAUC = r 0.25 + (na + nn −2) na × nn × 12 (13) ZAUC = A −0.5 SEAUC (14) SEAUC = r 0.25 + (na + nn −2) na × nn × 12 (13) (13) Except for in binary classiﬁcation, the p-values for AUC are p = 6.88e-85 (40X), p = 2.24e-89 (100X), p = 3.73e-89 (200X), and p = 9.20e-75 (400X). P-values for Kappa are all 0.0, regardless (14) February 2019 \| Volume 10 \| Article 80 Frontiers in Genetics \| www.frontiersin.org 10 February 2019 \| Volume 10 \| Article 80 Frontiers in Genetics \| www.frontiersin.org Frontiers in Genetics \| www.frontiersin.org 10 Breast Cancer Histopathological Images Analysis Xie et al. TABLE 3 \| The augmented image distribution of different subclasses in different magniﬁcation factors. Magniﬁcation Benign Malignant Total A F PT TA DC LC MC PC 40X 798 759 763 894 864 936 1,025 870 6,909 100X 791 780 847 900 903 1,020 1,110 852 7,203 200X 777 792 756 840 896 978 980 810 6,829 400X 742 711 805 780 788 822 845 828 6,321 Total 3,108 3,042 3,171 3,414 3,451 3,756 3,960 3,360 27,262 #Patients 4 10 3 7 38 5 9 6 82 FIGURE 5 \| The change in the loss function during the training of Inception_ResNet_V2 on raw and augmented data with 40 factor magniﬁcation, (A) binary classiﬁcation, (B) multi-class classiﬁcation. TABLE 3 \| The augmented image distribution of different subclasses in different magniﬁcation factors. Experiments on the Augmented Dataset To avoid the high false positive rate in multi-class classiﬁcation, we expanded the original samples of the dataset to suppress the inﬂuence that sample imbalance has on the February 2019 \| Volume 10 \| Article 80 11 Breast Cancer Histopathological Images Analysis Xie et al. TABLE 4 \| Results of binary and multi-class classiﬁcation on raw and augmented data using Inception_ResNet_V2/%. Classiﬁcation Datasets Criteria Magniﬁcation factors 40X 100X 200X 400X Binary Raw_data Se 98.48 98.90 99.13 98.06 Sp 96.63 92.95 92.80 92.10 PPV 98.46 96.45 96.39 96.51 DOR 185,774 118,782 147,138 58,835 ACC_IL 97.90 96.88 96.98 96.98 ACC_PL 98.03 97.07 82.74 88.12 F1 98.47 97.66 97.74 97.28 AUC 99.57 98.84 99.61 98.81 Kappa 95.12 92.96 93.18 91.05 Aug_data Se 99.95 99.45 99.65 98.88 Sp 99.61 99.26 99.18 99.34 PPV 99.66 99.39 99.31 99.42 DOR 56122,884 2440,736 3427,114 1342,245 ACC_IL 99.79 99.37 99.43 99.10 ACC_PL 99.93 99.96 100.0 99.90 F1 99.81 99.42 99.48 99.15 AUC 100.0 99.99 99.95 99.97 Kappa 99.59 98.72 98.86 98.19 Multi-class Raw_data ACC_IL 92.07 88.06 87.62 84.50 ACC_PL 89.11 88.45 86.07 71.42 Macro-F1 90.89 85.67 84.08 80.13 Micro-F1 92.07 88.06 87.62 84.50 Kappa 89.74 84.03 82.84 79.70 Aug_data ACC_IL 97.63 97.00 96.89 97.49 ACC_PL 98.42 98.07 97.85 97.40 Macro-F1 97.68 97.06 97.02 97.48 Micro-F1 97.63 97.00 96.89 97.49 Kappa 97.28 96.55 96.44 97.13 † The underline shows the best result of each metric between the two network structures being compared (INV3 and IRV2). The bold font shows the best result of each metric for each magniﬁcation level. TABLE 4 \| Results of binary and multi-class classiﬁcation on raw and augmented data using Inception_ResNet_V2/%. Multi-class Multi-class † The underline shows the best result of each metric between the two network structures being compared (INV3 and IRV2). The bold font shows the best result of each metric for each magniﬁcation level. each dataset regardless of magniﬁcation factor or the eﬀects of augmentation (raw or augmented). This is especially true for the results on the augmented datasets where Se>98%, Sp>99%, PPV>99%, and DOR>100. This tells us that the breast cancer diagnosis system based on the augmented dataset and the Inception_ResNet_V2 network is very reliable. Compared to the results in Table 2, we can say that augmenting raw imbalanced breast cancer histopathological image datasets can greatly improve the reliability of the diagnosis system. Frontiers in Genetics \| www.frontiersin.org Experiments on the Augmented Dataset The results in Figure 5 show that the value of the loss function decreases much faster and more smoothly converges to a much smaller value on the extended datasets than on the raw datasets. This is true for both experiments on binary and multi-class classiﬁcation of histopathological images of breast cancer. each dataset regardless of magniﬁcation factor or the eﬀects of augmentation (raw or augmented). This is especially true for the results on the augmented datasets where Se>98%, Sp>99%, PPV>99%, and DOR>100. This tells us that the breast cancer diagnosis system based on the augmented dataset and the Inception_ResNet_V2 network is very reliable. Compared to the results in Table 2, we can say that augmenting raw imbalanced breast cancer histopathological image datasets can greatly improve the reliability of the diagnosis system. The experimental results in Table 4 show that the experiments on extended datasets have produced much better results than those performed on the raw datasets. This is reﬂected by the data marked with red underlines, especially the results of multi- class classiﬁcation on the expanded datasets. These results are a signiﬁcant improvement compared to those from the original datasets. In addition, the experimental results in Table 4 tell us that the evaluation metrics of experimental results on 40X datasets are much better than those on any other datasets with diﬀerent magniﬁcation factors, which can also be seen from the values with black fonts in Table 4. The results further demonstrate that the 40X dataset should contain more signiﬁcant characteristics of breast cancer. In addition, the values of AUC in Table 4 show that our models are excellent. One even achieved the maximum value of AUC (1.0) on the augmented 40X dataset. The values of Kappa in Table 4 show that our models have obtained perfect agreement for binary classiﬁcation of histopathological images of breast cancer. The values of Kappa in Table 4 show that our models are perfect when applied to augmented datasets for multi-class classiﬁcation. Furthermore, we calculated the p-values for AUC and Kappa on all augmented datasets for binary and multi-class classiﬁcation. The p-values for AUC and Kappa are both 0.0, The experimental results in Table 4 for binary classiﬁcation show that Se>98%, Sp>92%, PPV>96%, and DOR>100 on February 2019 \| Volume 10 \| Article 80 Frontiers in Genetics \| www.frontiersin.org 12 Breast Cancer Histopathological Images Analysis Xie et al. Xie et al. Experiments on the Augmented Dataset TABLE 5 \| Binary and multi-class classiﬁcation comparison between our experimental results and the ones available from other studies /%. Classiﬁcation Criteria Methods Magniﬁcation factors 40X 100X 200X 400X Binary ACC_IL AlexNet_Raw(25) 85.6 ± 4.8 83.5± 3.9 83.1± 1.9 80.8± 3.0 CSDCNN_Raw(29) 95.8± 3.1 96.9± 1.9 96.7± 2.0 94.9± 2.8 INV3_Raw 96.84 96.76 96.49 94.71 IRV2_Raw 97.90 96.88 96.98 96.21 IRV2_Aug 99.79 99.37 99.43 99.10 ACC_PL PFTAS+QDA_Raw(5) 83.8± 4.1 82.1± 4.9 84.2± 4.1 82.0± 5.9 PFTAS+SVM_Raw(5) 81.6± 3.0 79.9± 5.4 85.1± 3.1 82.3± 3.8 AlexNet_Raw(25) 90.0± 6.7 88.4± 4.8 84.6± 4.2 86.1± 6.2 CSDCNN_Raw(29) 97.1± 1.5 95.7± 2.8 96.5± 2.1 95.7± 2.2 INV3_Raw 97.74 94.19 87.23 96.67 IRV2_Raw 98.03 97.07 82.74 88.12 IRV2_Aug 99.93 99.96 100.0 99.90 Multi-class ACC_IL LeNet_Raw(29) 40.1± 7.1 37.5± 6.7 40.1± 3.4 38.2± 5.9 LeNet_Aug(29) 46.4± 4.5 47.3± 4.9 46.5± 5.6 45.2± 9.1 AlexNet_Raw(29) 70.1± 7.4 68.1± 7.6 67.6± 4.8 67.3± 3.4 AlexNet_Aug(29) 86.4± 3.1 75.8± 5.4 72.6± 4.8 84.6± 3.6 CSDCNN_Raw(29) 89.4± 5.4 90.8± 2.5 88.6± 4.7 87.6± 4.1 CSDCNN_Aug(29) 92.8± 2.1 93.9± 1.9 93.7± 2.2 92.9± 1.8 INV3_Raw 90.28 85.35 83.99 82.08 IRV2_Raw 92.07 88.06 87.62 84.50 IRV2_Aug 97.63 97.00 96.89 97.49 ACC_PL LeNet_Raw(29) 38.1± 9.3 37.5± 3.4 38.5± 4.3 37.2± 3.6 LeNet_Aug(29) 48.2± 4.5 47.6± 7.5 45.5± 3.2 45.2± 8.2 AlexNet_Raw(29) 70.4± 6.2 68.7± 5.3 66.4± 4.3 67.2± 5.6 AlexNet_Aug(29) 74.6± 7.1 73.8± 4.5 76.4± 7.4 79.2± 7.6 CSDCNN_Raw(29) 88.3± 3.4 89.8± 4.7 87.6± 6.4 87.0± 5.2 CSDCNN_Aug(29) 94.1± 2.1 93.2± 1.4 94.7± 3.6 93.5± 2.7 INV3_Raw 90.44 89.05 80.63 81.08 IRV2_Raw 89.11 88.45 86.07 71.42 IRV2_Aug 98.42 98.07 97.85 97.40 † Bold fonts represent the best results among compared approaches with the same classiﬁer. extended datasets, respectively. The bold fonts denote the best results. which is much <0.05. This fact tells us that we can reject the null hypothesis (that the prediction result is a random guess), and accept the fact that our prediction is statistically signiﬁcant and not random. The experimental results in Table 5 tell us that both the evaluation criteria of ACC_IL and ACC_PL applied to the results obtained from the Inception_ResNet_V2 network have the best value among all of the available studies we found in the literature concerning the classiﬁcation of histopathological images of breast cancer on the expanded datasets for both binary and multi-class classiﬁcation. The results on the raw datasets produced by the Inception_ResNet_V2 network are better than those produced by other networks. Experiments on the Augmented Dataset Therefore, the deep learning network of Inception_ResNet_V2 with residual connections is very suitable for classifying the histopathological images of breast cancer. Also, using the expanded histopathological image datasets of breast cancer can obtain better classiﬁcation and diagnosis results. Experimental Comparisons This subsection will compare the experimental results of classifying histopathological images of breast cancer using the Inception_V3 and Inception_ResNet_V2 networks in addition to a selection of methods from the available studies carried by other research teams. The experimental results will be compared in terms of ACC_IL and ACC_PL, because the available studies only used these two evaluation criteria. The binary and the multi-class classiﬁcation experimental results are displayed in Table 5. Here, INV3_Raw denotes the results obtained by using Inception_V3 on original dataset. IRV2_Raw and IRV2_Aug represent the results produced by Inception_ResNet_V2 on the original and To judge whether or not our approaches are statistically signiﬁcant, we adopted the Friedman’s test (Borg et al., 2013) to discover the signiﬁcant diﬀerence between the compared algorithms. If a signiﬁcant diﬀerence has been detected by February 2019 \| Volume 10 \| Article 80 Frontiers in Genetics \| www.frontiersin.org 13 Breast Cancer Histopathological Images Analysis Xie et al. TABLE 6 \| Results of Friedman’s test between our approaches and the compared algorithms atα = 0.05. Binary classiﬁcation Multi-class classiﬁcation χ2 df p χ2 df p ACC_IL 14.6 4 0.0056 30.6667 8 0.0002 ACC_PL 18.1071 6 0.0060 30.8667 8 0.00015 Friedman’s test, then the multiple comparison test is used as a post hoc test to detect the signiﬁcant diﬀerence between pairs of the compared algorithms. Friedman’s test is considered preferable for comparing algorithms over several datasets without any normal distribution assumption (Borg et al., 2013). We conducted Friedman’s test at α = 0.05 using the results of algorithms on all datasets in terms of ACC_IL and ACC_PL for binary and multi-class classiﬁcation shown in Table 5. The Friedman’s test results are shown in Table 6. Here, χ2 is chi-square, df is the degree of freedom, and p is p-value. The results in the tables in the Supplementary Material show that each classiﬁer gets its best experimental results on the extended datasets of histopathological images of breast cancer, regardless of using binary or multi-class classiﬁcation. The experimental results of the Inception_ResNet_V2 network on the expanded datasets of histopathological images of breast cancer are the best ones among the results from all of the listed classiﬁers in the tables in the Supplementary Material. The experimental results of the SVM and 1-NN classiﬁers are not better than that of the Softmax classiﬁer, even though the features are extracted by the Inception_ResNet_V2 network. Experimental Comparisons Therefore, it is very appropriate to use the Inception_ResNet_V2 network to classify histopathological images of breast cancer. The Friedman’s test results in Table 6 tell us that there is a strong signiﬁcant diﬀerence between our approaches and the compared algorithms because any p in Table 6 supports p ≺ 0.05. Therefore, we conduct a multiple comparison test between each pair of algorithms at the conﬁdence level of 0.95 and show these statistical test results in Table 7. The mean rank diﬀerence between algorithms is shown in the upper triangle of the table. The statistical signiﬁcance between pairs of algorithms is displayed in the lower triangle using “∗.” The results in Table 8 reveal that even when using the same classiﬁers, such as SVM or 1-NN, the experimental results are diﬀerent. The results based on the extracted features from the Inception_ResNet_V2 network are much better than those in (5) based on the features extracted by other networks. The best results were also obtained using the extended datasets. This analysis further demonstrates that the deep learning network Inception_ResNet_V2 has a powerful ability to extract informative features automatically. The multiple comparison tests in Table 7 reveal that our breast cancer diagnosis model which uses Inception_ResNet_V2 on the augmented dataset is very powerful. It oﬀers a statistically signiﬁcant improvement compared to the results from available references that we can ﬁnd. Clustering Results This subsection will describe the great advantages of Inception_ResNet_V2 network when it is used for automatically extracting informative features from histopathological images of breast cancer. The 1,536-dimension features are extracted by using Inception_ResNet_V2 to process histopathological images of breast cancer, and the K-means clustering algorithm is adopted to group these images into proper clusters. In addition, a new AE (Autoencoder) network with a shape of [1536, 500, 2] is constructed to perform a non-linear transformation to the 1,536-dimension feature vectors produced by Inception_ResNet_V2. In this way, the 2-dimension features of the histopathological images of breast cancer can be obtained for K-means in low dimensional space. Here, IRV2+Kmeans represents the clustering results of K-means with the features extracted by Inception_ResNet_V2, while IRV2+AE+Kmeans represents the clustering results of K-means based on the features transformed by our proposed AE using the features extracted by Inception_ResNet_V2. This subsection will further compare the experimental results of Inception_ResNet_V2 on histopathological images of breast cancer to those of SVM and 1-NN classiﬁers with the 1,536-dimension features extracted by the Inception_ResNet_V2 network. Also, it will compare the experimental results of the SVM and 1-NN classiﬁers with features extracted by other networks. The experimental results of binary classiﬁcation of histopathological images of breast cancer with features extracted by Inception_ResNet_V2 are shown in Table S1 in terms of Se, Sp, PPV, DOR, ACC_IL, ACC_PL, F1, AUC and Kappa. Table S2 shows the experimental results of multi-class classiﬁcation of histopathological images of breast cancer with features extracted by Inception_ResNet_V2 in terms of ACC_IL, ACC_PL, Macro-F1, Micro-F1, and Kappa. Table 8 compared the studies in (5) and ours in terms of ACC_PL, the only evaluation criterion used in (5), when the experimental results are all from SVM and 1-NN classiﬁers. The diﬀerences between our methods and those in (5) are the features. We adopted the Inception_ResNet_V2 network to extract features of histopathological images of breast cancer while those in (5) used other networks to extract features. Frontiers in Genetics \| www.frontiersin.org Experiments to Find the Number of Clusters in the Dataset To ﬁnd the proper K for K-means, we adopt the internal criterion SSE (Silhouette Score) to search for it. The SSE index combines February 2019 \| Volume 10 \| Article 80 Frontiers in Genetics \| www.frontiersin.org 14 Breast Cancer Histopathological Images Analysis Xie et al. TABLE 7 \| Paired rank comparison of algorithms in ACC_IL and AII_PL for binary and multi-class classiﬁcation. ACC_IL for binary IRV2_Aug IRV2_Raw INV3_Raw CSDCNN_Raw(29) AlexNet_Raw(25) IRV2_Aug 1.25 2.75 2.0 4.0 IRV2_Raw 1.5 0.75 2.75 INV3_Raw −0.75 1.25 CSDCNN_Raw(29) 2.0 AlexNet_Raw(25) ACC_PL for binary IRV2 _Aug IRV2 _Raw INV3 _Raw CSDCNN _Raw(29) AlexNet _Raw(25) PFTAS+SVM _Raw(5) PFTAS+QDA _Raw(5) IRV2_Aug 2.75 2.0 2.0 4.0 5.0 5.25 IRV2_Raw −0.75 −0.75 1.25 2.25 2.5 INV3_Raw 0.0 2.0 3.0 3.25 CSDCNN_Raw(29) 2.0 3.0 3.25 AlexNet_Raw(25) 1.0 1.25 PFTAS+SVM_Raw(5) * 0.25 PFTAS+QDA_Raw(5) * ACC_IL for multi-class IRV2 _Aug IRV2 _Raw INV3 _Raw CSDCNN _Aug(29) CSDCNN _Raw(29) AlexNet _Aug(29) AlexNet _Raw(29) LetNet _Aug(29) LeNet _Raw(29) IRV2_Aug 3.0 4.0 1.0 2.5 4.5 6.0 7.0 8.0 IRV2_Raw 1.0 −2.0 −0.5 1.5 3.0 4.0 5.0 INV3_Raw −3.0 −1.5 0.5 2.0 3.0 4.0 CSDCNN_Aug(29) 1.5 3.5 5.0 6.0 7.0 CSDCNN_Raw(29) 2.0 3.5 4.5 5.5 AlexNet_Aug(29) 1.5 2.5 3.5 AlexNet_Raw(29) 1.0 2.0 LeNet_Aug(29) * 1.0 LeNet_Raw(29) * ACC_PL for multi-class IRV2 _Aug IRV2 _Raw INV3 _Raw CSDCNN _Aug(29) CSDCNN _Raw(29) AlexNet _Aug(29) AlexNet _Raw(29) LetNet _Aug(29) LeNet _Raw(29) IRV2_Aug 3.75 3.0 1.0 2.5 4.75 6.0 7.0 8.0 IRV2_Raw −0.75 −2.75 −1.25 1.0 2.25 3.25 4.25 INV3_Raw −2.0 −0.5 1.75 3.0 4.0 5.0 CSDCNN_Aug(29) 1.5 3.75 5.0 6.0 7.0 CSDCNN_Raw(29) 2.25 3.5 4.5 5.5 AlexNet_Aug(29) 1.25 2.25 3.25 AlexNet_Raw(29) 1.0 2.0 LeNet_Aug(29) * 1.0 LeNet_Raw(29) * †The upper triangle shows the difference between algorithms. The lower triangle shows pairs with statistical signiﬁcance. Asterisks indicate signiﬁcant difference between the pairs of algorithms in the table. 7 \| Paired rank comparison of algorithms in ACC_IL and AII_PL for binary and multi-class classiﬁcation. The SSE value of clustering of the histopathological images of breast cancer is variable with the number of clusters. Figure 6 plots the curves of SSE with the number of clusters on the 40X original dataset of histopathological images of breast cancer. The SSE curves of other magniﬁcation factor datasets are similar to those in Figure 6. the degree of condensation and separation and can be used in cases without any label information. The interval of SSE is [−1, 1]. Result Evaluation FIGURE 6 \| The silhouette score value with different numbers of clusters. This subsection will compare the clustering results of IRV2+AE+Kmeans and IRV2+Kmeans in terms of external criteria, including ACC, ARI, AMI, and the internal metric SSE. Figure 7 displays the clustering results in terms of the aforementioned four evaluation criteria on datasets with diﬀerent magniﬁcation factors. The experimental results in Figure 7 reveal the following facts: (1) the clustering results of IRV2+AE+Kmeans are better than those of IRV2+Kmeans in terms of ARI, AMI, SSE, and ACC on each dataset with diﬀerent magniﬁcation factors. This means that our proposed AE network can produce much more abstract and expressive features by encoding the features extracted by the Inception_ResNet_V2 network. (2) The values of ARI, AMI, SSE, and ACC for the same clustering are ascending, regardless of whether or not any transformation has been applied to the features that were extracted by Inception_ResNet_V2. (3) The best clustering accuracy (ACC) with features produced by the Inception_ResNet_V2 network is 59.3% on the 40X dataset, whereas the best ACC with features transformed by the proposed AE network using extracted features from the Inception_ResNet_V2 network is 76.4% on the 200X dataset. In summary, the best ACC scores of IRV2+AE+Kmeans and IRV2+Kmeans are 76.4 and 59.3%, respectively. FIGURE 6 \| The silhouette score value with different numbers of clusters. The results in Figure 6 show the best SSE score was achieved when the number of clusters is 2, regardless of how the features were extracted. This suggests that the histopathological images of breast cancer should be grouped into 2 categories of benign and malignant tumors, which is consistent with the real case. The results in Figure 6 also reveal that the clustering results of IRV2+AE+Kmeans are better than those from IRV2+Kmeans. This means that the proposed AE network can transform the features extracted by the Inception_ResNet_V2 network into much more informative ones, such that a better clustering of histopathological images of breast cancer can be detected. Frontiers in Genetics \| www.frontiersin.org Experiments to Find the Number of Clusters in the Dataset Higher SSE values are associated with samples belonging to the same cluster being closer together and samples belonging to diﬀerent groups being farther apart. SSE values closer to 1 indicate better clustering. the degree of condensation and separation and can be used in cases without any label information. The interval of SSE is [−1, 1]. Higher SSE values are associated with samples belonging to the same cluster being closer together and samples belonging to diﬀerent groups being farther apart. SSE values closer to 1 indicate better clustering. February 2019 \| Volume 10 \| Article 80 Frontiers in Genetics \| www.frontiersin.org 15 Breast Cancer Histopathological Images Analysis Xie et al. Xie et al. TABLE 8 \| Comparison between different networks extracting features for binary classiﬁcation/%. Criteria Methods Magniﬁcation factors 40X 100X 200X 400X ACC_PL CLBP+SVM_Raw(5) 77.4± 3.8 76.4± 4.5 70.2± 3.6 72.8± 4.9 GLCM+SVM_Raw(5) 74.0± 1.3 78.6± 2.6 81.9± 4.9 81.1± 3.2 LBP+SVM_Raw(5) 74.2± 5.0 73.2± 3.5 71.3± 4.0 73.1± 5.7 LPQ+SVM_Raw(5) 73.7± 5.5 72.8± 5.0 73.0± 6.6 73.7± 5.7 ORB+SVM_Raw(5) 71.9± 2.3 69.4± 0.4 68.7± 0.8 67.3± 3.1 PFTAS+SVM_Raw(5) 81.6± 3.0 79.9± 5.4 85.1± 3.1 82.3± 3.8 IRV2+SVM_Raw 97.93 96.58 97.07 96.62 IRV2+SVM_Aug 99.27 98.97 98.90 98.74 CLBP+1-NN_Raw(5) 73.6± 2.5 71.0± 2.8 69.4± 1.5 70.1± 1.3 GLCM+1-NN_Raw(5) 74.7± 1.0 76.8± 2.1 83.4± 3.3 81.7± 3.3 LBP+1-NN_Raw(5) 75.6± 2.4 73.0± 2.4 72.9± 2.3 71.2± 3.6 LPQ+1-NN_Raw(5) 72.8± 4.9 71.1± 6.4 74.3± 6.3 71.4± 5.2 ORB+1-NN_Raw(5) 71.6± 2.0 69.3± 2.0 69.6± 3.0 66.1± 3.5 PFTAS+1-NN_Raw(5) 80.9± 2.0 80.7± 2.4 81.5± 2.7 79.4± 3.9 IRV2+1-NN_Raw 97.32 95.91 96.12 95.88 IRV2+1-NN_Aug 98.04 97.50 97.85 97.48 †Bold fonts represent the best results among compared approaches with the same classiﬁer. TABLE 8 \| Comparison between different networks extracting features for binary classiﬁcation/%. †Bold fonts represent the best results among compared approaches with the same classiﬁer. CONCLUSIONS AND FUTURE WORK This paper proposed our methods for the analysis of histopathological images of breast cancer based on the deep convolutional neural networks of Inception_V3 and Inception_ResNet_V2 trained with transfer learning techniques. The aforementioned two networks are pre-trained on the large image dataset of ImageNet. Then, their learned structure and February 2019 \| Volume 10 \| Article 80 Frontiers in Genetics \| www.frontiersin.org 16 Breast Cancer Histopathological Images Analysis Xie et al. Xie et al. FIGURE 7 \| Clustering results in terms of ARI, AMI, SSE, and ACC for datasets with different magniﬁcation factors (A) 40X, (B) 100X, (C) 200X, (D) 400X. esults in terms of ARI, AMI, SSE, and ACC for datasets with different magniﬁcation factors (A) 40X, (B) 100X, (C) 200X, (D) 400X. not as good as classiﬁcation accuracies because the latter used label information. parameters are frozen. The number of neurons in the last fully-connected layer is set according to our speciﬁc task, and the parameters of the fully-connected layer are re-trained. In this way, the model can be used to perform binary or multi-class classiﬁcation of the histopathological images of breast cancer. We demonstrate that our experimental results are superior to the ones available in other studies that we have found, and that the Inception_ResNet_V2 network is more suitable for performing analysis of the histopathological images of breast cancer than the Inception_V3 network. Finding ways that we can improve the clustering accuracy will require further study. In addition to this, ﬁnding the number of clusters of histopathological images of breast cancer in both cases of 8 classes and 2 classes is another task that needs to be addressed. Noise is a prevalent issue in medical imaging and can have a signiﬁcant eﬀect on results. Some common sources of noise include white patches on slides after deparaﬃnization, visible patches on tissue after hydrating, and uneven staining. It was reported that batch eﬀects can lead to huge dissimilarities in features extracted from images (Mathews et al., 2016). For the histopathological images used in this paper, it is a fact that the diﬀerences of the resolution, contrast and appearance between images from same class are much more apparent than those from diﬀerent classes. The variance of the ﬁne-grained histopathological images of breast cancer results in diﬃculties when trying to classify an image as benign, malignant, or another speciﬁc category. CONCLUSIONS AND FUTURE WORK How we can avoid or reduce the inﬂuence on the analysis of histopathological images of breast cancer from these issues will be the focus of our future work. Also, our experimental results from the augmented datasets are much better than those from the original datasets. This is especially true when doing multi-class classiﬁcation with the histopathological images of breast cancer that we used. Our comparison of the experimental results demonstrates that the Inception_ResNet_V2 network is able to extract much more informative features than the other networks we referenced. The clustering analysis of the histopathological images of breast cancer using the typical clustering algorithm K-means demonstrates that the proper K value for K-means can be found by using the internal criterion SSE. The proposed AE network can detect much more informative, low dimensional features present in histopathological images of breast cancer. Furthermore, the clustering results produced by K-means using features extracted by Inception_ResNet_V2 and transformed by the proposed AE are much better, in terms of ARI, AMI, SSE, and ACC, than the results produced with features extracted only by Inception_ResNet_V2. Frontiers in Genetics \| www.frontiersin.org REFERENCES Imagenet classiﬁcation with deep convolutional neural networks. Adv. Neural Inform. Proce. Syst. 60, 84–90. doi: 10.1145/3065386 Boyle, P., and Levin, B. (2008). World Cancer report 2008: IARC Press. International Agency for Research on Cancer. Bradley, A. P. (1997). The use of the area under the ROC curve in the evaluation of machine learning algorithms. Pattern Recog. 30, 1145–1159. doi: 10.1016/S0031-3203(96)00142-2 Landis, J. R., and Koch, G. G. (1977). The measurement of observer agreement for categorical data. Biometrics 33, 159–174. Colquhoun, D. (2014). An investigation of the false discovery rate and the misinterpretation of p-values. R. Soc. Open Sci. 1:140216. doi: 10.1098/rsos.140216 Lowe, D. G. (ed) (1999). “Object recognition from local scale-invariant features.” in The Proceedings of the Seventh IEEE International Conference on Computer Vision. (Kerkyra: IEEE). MacQueen, J. (eds) (1967). “Some methods for classiﬁcation and analysis of multivariate observations,” in Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability (Oakland, CA). Ellis, P. D. (2010). The Essential Guide to Eﬀect Sizes: Statistical Power, Meta- Analysis, and the Interpretation of Research Results. New York, NY: Cambridge University Press. Mathews, A., Simi, I., and Kizhakkethottam, J. J. (2016). Eﬃcient diagnosis of cancer from histopathological images by eliminating batch eﬀects. Procedia Technol. 24, 1415–1422. doi: 10.1016/j.protcy.2016.05.165 Esteva, A., Kuprel, B., Novoa, R. A., Ko, J., Swetter, S. M., Blau, H. M., et al. (2017). Dermatologist-level classiﬁcation of skin cancer with deep neural networks. Nature 542, 115–118. doi: 10.1038/nature21056 Filipczuk, P., Fevens, T., Krzyzak, A., and Monczak, R. (2013). Computer- aided breast cancer diagnosis based on the analysis of cytological images of ﬁne needle biopsies. IEEE Trans. Med. Imaging 32, 2169–2178. doi: 10.1109/TMI.2013.2275151 Moraga-Serrano, P. E. (2018). Global, regional, and national cancer incidence, mortality, years of life lost, years lived with disability, and disability- adjusted life-years for 29 cancer groups, 1990 to 2016: a systematic analysis for the global burden of disease study. JAMA Oncol. 4, 1553–1568. doi: 10.1001/jamaoncol.2018.2706 George, Y. M., Zayed, H. H., Roushdy, M. I., and Elbagoury, B. M. (2014). Remote computer-aided breast cancer detection and diagnosis system based on cytological images. IEEE Syst. J. 8, 949–964. doi: 10.1109/JSYST.2013. 2279415 Motlagh, N. H., Jannesary, M., Aboulkheyr, H., Khosravi, P., Elemento, O., Totonchi, M., et al. (2018). Breast Cancer Histopathological Image Classiﬁcation: a Deep Learning Approach. bioRxiv 242818. doi: 10.1101/242818 Gulshan, V., Peng, L., Coram, M., Stumpe, M. C., Wu, D., Narayanaswamy, A., et al. (2016). REFERENCES Han, Z., Wei, B., Zheng, Y., Yin, Y., Li, K., and Li, S. (2017). Breast cancer multi- classiﬁcation from histopathological images with structured deep learning model. Sci. Rep. 7:4172. doi: 10.1038/s41598-017-04075-z Araújo, T., Aresta, G., Castro, E., Rouco, J., Aguiar, P., Eloy, C., et al. (2017). Classiﬁcation of breast cancer histology images using convolutional neural networks. PLoS ONE 12:e0177544. doi: 10.1371/journal.pone.0177544 Haralick, R. M., Shanmugam, K., and Dinstein, I. H. (1973). Textural features for image classiﬁcation. IEEE Transac. Syst. Man Cybernet. 3, 610–621. doi: 10.1109/TSMC.1973.4309314 networks. PLoS ONE 12:e0177544. doi: 10.1371/journal.pone.0 Asri, H., Mousannif, H., Al Moatassime, H., and Noel, T. (2016). Using machine learning algorithms for breast cancer risk prediction and diagnosis. Procedia Comput. Sci. 83, 1064–1069. doi: 10.1016/j.procs.2016.04.224 He, K., Zhang, X., Ren, S., and Sun, J. (eds) (2016).“Deep residual learning for image recognition,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (Las Vegas, NV). Aswathy, M., and Jagannath, M. (2017). Detection of breast cancer on digital histopathology images: present status and future possibilities. Inform. Med. Unlocked 8, 74–79. doi: 10.1016/j.imu.2016.11.001 He, K., Zhang, X., Ren, S., and Sun, J. (eds) (2015). “Delving deep into rectiﬁers: Surpassing human-level performance on imagenet classiﬁcation,” in Proceedings of the IEEE international conference on computer vision (Santiago). Bayramoglu, N., Kannala, J., and Heikkilä, J. (eds) (2016). “Deep learning for magniﬁcation independent breast cancer histopathology image classiﬁcation,” in 23rd International Conference on Pattern Recognition (ICPR), 2016. (Cancun: IEEE). Hinton, G. E., and Salakhutdinov, R. R. (2006). Reducing the dimensionality of data with neural networks. Science 313, 504–507. doi: 10.1126/science.1127647 Hubert, L., and Arabie, P. (1985). Comparing partitions. J. Classiﬁcation 2, 193–218. doi: 10.1007/BF01908075 Bengio, Y., Courville, A., and Vincent, P. (2013). “Representation learning: A review and new perspectives,” in IEEE Transactions on Pattern Analysis and Machine Intelligence 35, 1798–1828. doi: 10.1109/TPAMI.2013.50 Kingma, D. P., and Ba, J. (2014). Adam: a Method for Stochastic Optimization. arXiv preprint arXiv:14126980. Bergstra, J., and Bengio, Y. (2012). Random search for hyper-parameter optimization. J. Machine Learn. Res. 13, 281–305. Kowal, M., Filipczuk, P., Obuchowicz, A., Korbicz, J., and Monczak, R. (2013). Computer-aided diagnosis of breast cancer based on ﬁne needle biopsy microscopic images. Comput. Biol. Med. 43, 1563–1572. doi: 10.1016/j.compbiomed.2013.08.003 Borg, A., Lavesson, N., and Boeva, V. (eds) (2013). Comparison of Clustering Approaches for Gene Expression Data. Aalborg: SCAI. Krizhevsky, A., Sutskever, I., and Hinton, G. E. (eds) (2012). AUTHOR CONTRIBUTIONS JX made substantial contributions to the conception and design of the work, drafted the work, and revised it critically for important intellectual content by discussing with CZ, JL, and RL. RL also made substantial contributions to the conception and design of the work. She wrote the code for the algorithms, analyzed the experimental results, and wrote the experimental report. CZ and JL discussed with JX and RL about the technique details, then CZ and JL revised the paper critically for important All of the work in this paper demonstrates that the deep convolutional neural network Inception_ResNet_V2 has the advantage when it comes to extracting expressive features from histopathological images of breast cancer. The clustering accuracies of histopathological images of breast cancers are February 2019 \| Volume 10 \| Article 80 17 Breast Cancer Histopathological Images Analysis Xie et al. and Development Program of China under Grant No. 2016YFC0901900 and the Fundamental Research Funds for the Central Universities under Grant Nos. GK201701006 and GK201806013. At the same time, it is supported by the Innovation Funds of Graduate Programs at Shaanxi Normal University under Grant Nos. 2015CXS028 and 2016CSY009 as well. intellectual content. JX and CZ gave approval for publication of the content. JX and CZ agree to be accountable for all aspects of the work and will ensure that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. SUPPLEMENTARY MATERIAL The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fgene. 2019.00080/full#supplementary-material ACKNOWLEDGMENTS The authors would like to thank Professor Spanhol et al. who provided the database of BreaKHis for us to use, and the other involved researchers for their work to stimulate this research. This work is supported in part by the National Natural Science Foundation of China under Grant No. 61673251. It is also supported by the National Key Research REFERENCES Development and validation of a deep learning algorithm for detection of diabetic retinopathy in retinal fundus photographs. JAMA 316, 2402–2410. doi: 10.1001/jama.2016.17216 Nawaz, M., Sewissy, A. A., and Soliman, T. H. A. (2018). Multi-class breast cancer classiﬁcation using deep learning convolutional neural network. Int. J. Adv. Comput. Sci. Appl. 9, 316–322. doi: 10.14569/IJACSA.2018.0 90645 February 2019 \| Volume 10 \| Article 80 Frontiers in Genetics \| www.frontiersin.org 18 Breast Cancer Histopathological Images Analysis Xie et al. Pan, S. J., and Yang, Q.,(2010). A survey on transfer learning. IEEE Transac. Knowledge Data Eng. 22, 1345–1359. doi: 10.1109/TKDE.2009.191 Vinh, N. X., Epps, J., and Bailey, J. (2010). Information theoretic measures for clusterings comparison: variants, properties, normalization and correction for chance. J. Machine Learn. Res. 11, 2837–2854. Rousseeuw, P. J. (1987). Silhouettes: a graphical aid to the interpretation and validation of cluster analysis. J. Comput. Appl. Math. 20, 53–65. doi: 10.1016/0377-0427(87)90125-7 Xie, J., Hou, Q., Shi, Y., Peng, L., Jing, L., Zhuang, F., et al. (2018). The Automatic Identiﬁcation of Butterﬂy Species. arXiv [preprint]. arXiv:180 306626. Spanhol, F. A., Oliveira, L. S., Petitjean, C., and Heutte, L. (2016a). A dataset for breast cancer histopathological image classiﬁcation. IEEE Transac. Biomed. Eng. 63, 1455–1462. doi: 10.1109/TBME.2015.2496264 Zhang, Y., Zhang, B., Coenen, F., and Lu, W. (2013). Breast cancer diagnosis from biopsy images with highly reliable random subspace classiﬁer ensembles. Machine Vision Appl. 24, 1405–1420. doi: 10.1007/s00138-012- 0459-8 Spanhol, F. A., Oliveira, L. S., Petitjean, C., and Heutte, L. (eds) (2016b). “Breast cancer histopathological image classiﬁcation using convolutional neural networks,” in 2016 International Joint Conference on Neural Networks (IJCNN). (Vancouver, BC: IEEE). ensembles. Machine Vision Appl. 24, 1405–1420. doi: 10.1007/s00138-012- 0459-8 Zhang, Y., Zhang, B., Coenen, F., Xiao, J., and Lu, W. (2014). One-class kernel subspace ensemble for medical image classiﬁcation. EURASIP J. Adv. Signal Proces. 2014:17. doi: 10.1186/1687-6180-2014-17 Stenkvist, B., Westman-Naeser, S., Holmquist, J., Nordin, B., Bengtsson, E., Vegelius, J., et al. (1978). Computerized nuclear morphometry as an objective method for characterizing human cancer cell populations. Cancer Res. 38, 4688–4697. Conﬂict of Interest Statement: The authors declare that the research was conducted in the absence of any commercial or ﬁnancial relationships that could be construed as a potential conﬂict of interest. Szegedy, C., Ioﬀe, S., Vanhoucke, V., and Alemi, A. A. (eds) (2017). Inception-v4, Inception-Resnet and the Impact of Residual Connections on Learning. AAAI Copyright © 2019 Xie, Liu, Luttrell and Zhang. Frontiers in Genetics \| www.frontiersin.org REFERENCES This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. Szegedy, C., Vanhoucke, V., Ioﬀe, S., Shlens, J., and Wojna, Z. (eds) (2016). “Rethinking the inception architecture for computer vision,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (Las Vegas, NV). Veta, M., Pluim, J. P., van Diest, P. J., and Viergever, M. A. (2014). Breast cancer histopathology image analysis: a review. IEEE Transac. Biomed. Eng. 61, 1400–1411. doi: 10.1109/TBME.2014.2303852 February 2019 \| Volume 10 \| Article 80 Frontiers in Genetics \| www.frontiersin.org 19
https://openalex.org/W2962909855	https://escholarship.org/content/qt6w39h4r7/qt6w39h4r7.pdf?t=qb1npp	English	null	Understanding Android Obfuscation Techniques: A Large-Scale Investigation in the Wild	Springer eBooks	2,018	cc-by	9,950	Peer reviewed Peer reviewed UC Irvine UC Irvine Previously Published Works Title Understanding Android Obfuscation Techniques: A Large-Scale Investigation in the Wild Permalink https://escholarship.org/uc/item/6w39h4r7 Authors Dong, Shuaike Li, Menghao Diao, Wenrui et al. Publication Date 2018-01-04 Copyright Information This work is made available under the terms of a Creative Commons Attribution License, availalbe at https://creativecommons.org/licenses/by/4.0/ UC Irvine UC Irvine Previously Published Works Title Understanding Android Obfuscation Techniques: A Large-Scale Investigation in the Wild Permalink https://escholarship.org/uc/item/6w39h4r7 Authors Dong, Shuaike Li, Menghao Diao, Wenrui et al. Publication Date 2018-01-04 Copyright Information This work is made available under the terms of a Creative Commons Attribution License, availalbe at https://creativecommons.org/licenses/by/4.0/ Copyright Information This work is made available under the terms of a Creative Commons Attribution License, availalbe at https://creativecommons.org/licenses/by/4.0/ ABSTRACT platform, obfuscation is particularly useful, given that the task of disassembling or decompiling Android app is substantially easier than doing so for other sorts of binary code, like X86 executables. platform, obfuscation is particularly useful, given that the task of disassembling or decompiling Android app is substantially easier than doing so for other sorts of binary code, like X86 executables. Program code is a precious asset to its owner. Due to the easy- to-reverse nature of Java, code protection for Android apps is of particular importance. To this end, code obfuscation is widely uti- lized by both legitimate app developers and malware authors, which complicates the representation of source code or machine code in order to hinder the manual investigation and code analysis. Despite many previous studies focusing on the obfuscation techniques, however, our knowledge on how obfuscation is applied by real- world developers is still limited. arXiv:1801.01633v1 [cs.CR] 5 Jan 2 Android obfuscation arguably is pervasive. On the one hand, there are already more than 2.8 million apps available for down- loading just in one app market, Google Play, up to March 2017 [18]. On the other hand, many off-the-shelf obfuscators are developed, and some authors claim their tools are used by more than 300,000 apps [2]. Consequently, the issues around app obfuscation attract many researchers. So far, most of the studies focus on the topics like what obfuscation techniques can be used [26], how they can be improved [48], how well they can be handled by state- of-art code analysis tools [47], and how to deobfuscate the code automatically [28]. While these studies provide solid ground for understanding the obfuscation techniques and its implications, there is a still an unfilled gap in this domain: how obfuscation is actually used by the vast amount of developers? arXiv:1801.01633v1 [cs.CR] 5 In this paper, we seek to better understand Android obfuscation and depict a holistic view of the usage of obfuscation through a large-scale investigation in the wild. In particular, we focus on four popular obfuscation approaches: identifier renaming, string encryption, Java reflection, and packing. To obtain the meaningful statistical results, we designed efficient and lightweight detection models for each obfuscation technique and applied them to our massive APK datasets (collected from Google Play, multiple third- party markets, and malware databases). We have learned several interesting facts from the result. Android, obfuscation, static analysis, code protection Android, obfuscation, static analysis, code protection ABSTRACT For example, malware authors use string encryption more frequently, and more apps on third-party markets than Google Play are packed. We are also interested in the explanation of each finding. Therefore we carry out in-depth code analysis on some Android apps after sampling. We believe our study will help developers select the most suitable obfuscation approach, and in the meantime help researchers improve code analysis systems in the right direction. We believe this topic needs to be studied, and the answer could enlighten new research opportunities. To name a few, for developers, learning which obfuscation techniques should be used is quite important. Not all obfuscation techniques are equally effective, and using some might even bring the incompatibility issue. Plenty of code analysis approaches were proposed, but their effects are usually hampered by obfuscation and the impact greatly differs based on the specific obfuscation technique in use, e.g., identifier renaming is much less of an issue comparing to string encryption. Knowing the distribution of obfuscation techniques can better assist the design of code analysis tools and prioritize the challenges need to be tackled. All roads paving to the correct conclusions call for measurement on real-world apps, and only the result coming from a comprehensive study covering a diverse portfolio of apps (published in different markets, in different countries, from both malware authors and legitimate companies) is meaningful.i Understanding Android Obfuscation Techniques: A Large-Scale Investigation in the Wild v:1801.01633v1 [cs.CR] 5 Jan 2018 platform, obfuscation is particularly useful, given that the task of disassembling or decompiling Android app is substantially easier than doing so for other sorts of binary code, like X86 executables. Android obfuscation arguably is pervasive. On the one hand, there are already more than 2.8 million apps available for down- loading just in one app market, Google Play, up to March 2017 [18]. On the other hand, many off-the-shelf obfuscators are developed, and some authors claim their tools are used by more than 300,000 apps [2]. Consequently, the issues around app obfuscation attract many researchers. So far, most of the studies focus on the topics like what obfuscation techniques can be used [26], how they can be improved [48], how well they can be handled by state- of-art code analysis tools [47], and how to deobfuscate the code automatically [28]. While these studies provide solid ground for understanding the obfuscation techniques and its implications, there is a still an unfilled gap in this domain: how obfuscation is actually used by the vast amount of developers? Powered by the California Digital Library University of California eScholarship.org eScholarship.org 1 INTRODUCTION Code is a very important intellectual property to its developers, no matter if they work as individuals or for a large corporation. To protect this property, obfuscation is frequently used by developers, which is also considered as a double-edged sword by the security community. To a legitimate software company, obfuscation keeps its competitors away from copying the code and quickly building their own products in an unfair way. To a malware author, obfuscation raises the bar for automated code analysis and manual investigation, two approaches adopted by nearly every security company. For a mobile app, especially the one targeting Android Our Work. As the first step, in this paper, we systematically study the obfuscation techniques used in Android apps and carry out a large-scale investigation for apps in the wild. We focus on four most popular Android obfuscation techniques (identifier renaming, string encryption, Java reflection, and packing) and measure the base and popular implementation of each technique. To notice, the existing tools, like deobfuscators, cannot solve our problem here, since they either work well against a specific technique or a specific off-the-shelf obfuscator (e.g., ProGuard). As such, they cannot be 1 assets This directory is similar to the res directory and used to store static files in the APK. However, unlike res directory, developers can create subdirectories in any depth with the arbitrary file structure.i used to provide a holistic view. Our key insight to this end is that instead of mapping the obfuscated code to its original version, a challenge not yet fully addressed, we only need to cluster them based on their code patterns or statistical features. Therefore, we built a set of lightweight detectors for all studied techniques, based on machine learning and signature matching. Our tools are quite effective and efficient, suggested by the validation result on ground- truth datasets. We then applied them on a real-world APK dataset with 114,560 apps coming from three different sources, including Google Play set, third-party markets set, and malware set, for the large-scale study. lib The code compiled for specific platforms (usually library files, like .so) are stored in this directory. Subdirectories can be created according to the type of processors, like armeabi, armeabi-v7a, x86, x86_64, mips. 2.2 Android Obfuscation Characterization In general, obfuscation attempts to garble a program and makes the source or machine code more difficult for humans to understand. Programmers can deliberately obfuscate code to conceal its purpose or logic, in order to prevent tampering, deter reverse engineering, or behave as a puzzle for someone reading the code. Specifically, there are several common obfuscation techniques used by Android apps, including identifier renaming, string encryption, excessive overloading, reflection, and so forth. • New Techniques. We propose several techniques for detecting different obfuscation techniques accurately, such as n-gram -based renaming detection model and backward slicing-based reflection detection algorithm. • Large-scale Evaluation. We carried out large-scale ex- periments and applied our detection techniques on over 100K APK files collected from three different sources. We listed our findings and provided explanations based on in-depth analysis of obfuscated code. Identifier Renaming. In software development, for good read- ability, code identifiers’ names are usually meaningful, though developers may follow different naming rules (like CamelCase, Hungarian Notation). However, these meaningful names also accommodate reverse-engineers to understand the code logic and locate the target functions rapidly. Therefore, to reduce the potential information leakage, the identifier’s name could be replaced by a meaningless string. The following code snippet gives an example, in which all identifiers in class Account are renamed. Roadmap. The rest of this paper is organized as follows: We systematically summarize popular Android obfuscation techniques in Section 2. Section 3 overviews the high-level architecture of our detection framework. The detailed detection strategies and statistical results on large-scale datasets are provided in Section 4. Also, we discuss some limitations and future plans in Section 5. Section 6 reviews the previous research on Android obfuscation, and Section 7 concludes this paper. 1 public class a{ 2 private Integer a; 3 private Float = b; 4 public void a(Integer a, Float b){ 5 this.a = a + Integer.valueOf(b) 6 } 7 } 2 BACKGROUNDl In this section, we briefly introduce the structure of APK file and overview some common Android obfuscation techniques. 1 INTRODUCTION META-INF This directory is responsible for saving the signa- ture information of a specific app, which is used to validate the integrity of an APK file.ii AndroidManifest.xml This XML file is the configuration of an APK, declaring its basic information, like name, version, required permissions and components. Each APK has an AndroidManifest file, and the only one.i Discoveries. Our study reveals several interesting facts, with some confirming people’s intuition but some contradicting to common beliefs: for example, as an obfuscation approach, identifier renaming is more widely-used in third-party apps than in malware. Also, though basic obfuscation is prevalently applied in benign apps, the utilization rate of other advanced obfuscation techniques is much lower than that of malware. We believe these insights coming from “big code” are valuable in guiding developers and researchers in building, counteracting or using obfuscation techniques. classes.dex The dex file contains all the information of the classes in an app. The data is organized in a way the Dalvik virtual machine can understand and execute.i resources.arsc This file is used to record the relationship between the resource files and related resource ID and can be leveraged to locate specific resources. Contributions. We summarize this paper’s contributions as below: • Systematic Study. We systematically study the current mainstream Android obfuscation techniques used by app developers. 3.1 System Overview To detect the usage of obfuscation techniques, we propose an architecture to analyze APK files automatically, as illustrated in Figure 1. After the APK files collected from several channels (details are provided in Section 3.2) are stored in our server, this detection framework will try to unpack them for the primary testing. Some damaged APK files failing to pass this step will be discarded. Then this framework applies four targeted detection methods to identify obfuscated Smali code blocks. These detection methods could be classified into two categories: signature-based and ma- chine learning-based. For the obfuscation techniques with specific features, we search the corresponding signatures in Smali code to determine the existence. For example, the reflective calls which implicitly invoke another function can be located by searching the sequence pattern [Class.forName()→getMethod()→invoke()]. However, it is difficult to extract fixed features for some techniques (e.g., encrypted strings), so we utilize machine learning algorithms 4 OBFUSCATION DETECTIONS AND LARGE-SCALE INVESTIGATION In this section, we introduce the detection approaches for each obfuscation technique and summarize our findings based on large- scale experiments. 3 SYSTEM DESIGN Our target is to systematically study the Android obfuscation techniques and carry out a large-scale investigation. As the first step, we design an efficient Android code analysis framework to identify the obfuscation techniques used by developers. Here we overview the high-level design of this framework and introduce the datasets prepared for the subsequent large-scale investigations. 3.2 APK Dataset We are interested in the obfuscation usage status of apps in different types, so three representative APK datasets were used in our experiment: Google Play set (26,614 samples), third-party market set (65,666 samples), and malware set (22,280 samples). These samples were collected during 2016 and 2017. In total, our experiment dataset contains 114,560 sample with the size of around 1.521TB. More details are given in Table 1.fi Packing. Packing is a widely-used code protection technique. The packed APK file is composed of an encrypted origin APK and a wrapper APK. When the user launches the APK, the wrapper will run first, decrypt the original APK and load it into the memory, and then the execution will be handed to the decrypted APK. Due to the cryptographic procedure and runtime release, it becomes hard to get the original code through static analysis. We regard packing as an obfuscation skill in a broad sense because its goal is to hinder the reverse-engineering as well. g As the official app store for Android, Google Play is the main Android app distribution channel. Thus, its sample set could reflect the deployment status of obfuscation used by mainstream developers. Also, due to the policy restriction, in some countries (such as China), Google Play is not available, and users have to install apps from third-party markets. Therefore, in the second dataset, we select six popular app markets from China (say Anzhi [5], Xiaomi [25], Wandoujia [24], 360 [1], Huawei [13], and AppChina [6]) and developed the corresponding crawlers to collect their apps. Note that the replicated samples from different markets have been excluded. Lastly, except for legitimate app samples, we are also curious about whether malware authors heavily use obfuscation skills to hide their malicious intentions. So, the last dataset contains the malware samples coming from VirusShare [22] and VirusTotal [23, 37]. 2.1 APK File Structure to classify automatically. The training set comes from F-Droid [12], an open source Android app repository. to classify automatically. The training set comes from F-Droid [12], an open source Android app repository. As an obfuscation technique, reflection is a good choice of hiding program behaviors because it can transfer the control to a certain function implicitly, which can not be well handled by state-of-the- art static analysis tools. Therefore, malware developers usually heavily employ reflection to hide malicious actions. 2.1 APK File Structure String Encryption. Strings are very common-used data structures in software development. In an obfuscated app, strings could be encrypted to prevent information leakage. Based on cryptographic functions, the original plaintexts are replaced by random strings and restore at runtime. As a result, string encryption could effectively hinder hard-coded static scanning. The following code block shows an example. String Encryption. Strings are very common-used data structures in software development. In an obfuscated app, strings could be encrypted to prevent information leakage. Based on cryptographic functions, the original plaintexts are replaced by random strings and restore at runtime. As a result, string encryption could effectively hinder hard-coded static scanning. The following code block shows an example. An APK (Android application package) file is a zip compressed file containing all the content of an Android app, in general, including four directories (res, assets, lib, and META-INF) and three files (AndroidManifest.xml, classes.dex, and resources.arsc). The purposes of these directories and files are listed as below. res This directory stores Android resource files which will be mapped into the .R file in Android and allocated the corresponding ID. res This directory stores Android resource files which will be mapped into the .R file in Android and allocated the corresponding ID. 1 String option = "@^@#\x `1 m7 %9_!v"; String option = "@^@#\x `1 m7 %*9_!v"; 2 Table 1: APK Dataset for Investigation Table 1: APK Dataset for Investigation Type Source Number Official Market Google Play 26,614 3rd-party Market Wandoujia 8,979 360 18,724 Huawei 22,048 Anzhi 7,121 Xiaomi 4,649 AppChina 4,145 Malware VirusShare 19,004 VirusTotal 3,267 to classify automatically. The training set comes from F-Droid [12], 2 this.execute(decrypt(option)); Java Reflection. Reflection is an advanced feature of Java [21], which provides developers with a flexible approach to interact with the program, e.g., creating new object instances and invoking methods dynamically. One common legitimate usage is to invoke nonpublic APIs in the SDK (with the annotation @hide). The following code block gives an example of reflection that invokes a hidden API batteryinfo. 1 Object object = new Object (); 2 Method getService = Class.forName("android.os. ServiceManager").getMethod("getService", String.class); 3 Object obj = getService.invoke(object , new Object []{ new String("batteryinfo")}); 1 Object object = new Object (); 2 Method getService = Class.forName("android.os. ServiceManager").getMethod("getService", String.class); 3 Object obj = getService.invoke(object , new Object []{ new String("batteryinfo")}); to classify automatically. The training set comes from F-Droid [12], an open source Android app repository. 4.1 Identifier Renaming Generally, in the software development, the names of identifiers (variable names, function names, and so forth) are usually meaning- ful, which could provide good code readability and maintainability. However, such clear names may leak much information due to the easy-to-reverse feature of Java. As a solution, identifier renaming is proposed and widely used in practice.f The renaming operation can be appended at different stages of APK file packaging. For example, ProGuard [19] and Allatori [3] 3 Official Market Third-party Markets Signature Searching Reflection Packing Unpack APK Repository Malware Format Check Statistical Result Machine Learning-based Detection Signature-based Detection VirusShare.com Training Phase Success Fail Testing Phase Identifier Renaming String Encryption Figure 1: Android App Obfuscation Detection Framework Format Check Figure 1: Android App Obfuscation Detection Framework deployed by developers proactively. Therefore, we have pre-removed over 12,000 common third-party libraries using the approach of Li et al. [39].i work at the source-code level, mapping the original names to mangled ones based on the user’s configuration. The other ob- fuscators, like DashO [9], DexProtector[11], and Shield4J [20], can work directly on APK files, modifying .class and .dex files.i (2) Feature Generation. The amount of identifiers varies among different apps. To build a uniform expression, we apply the n-gram algorithm [17] to generate a fixed-length feature vector for each app. An n-gram is a contiguous sequence of n items from a given sequence of text or speech. In our implementation, we apply 3-gram2 to traverse each name string in extracted raw name set to form a fixed-length3 feature vector. The feature vector records the frequency of each three continuous characters and will be normalized.i Given an identifier, we can easily tell whether some obfuscator has renamed it based on the information it contains. In other words, if an identifier name is obscure and meaningless, it can be regarded as obfuscated because it tries to hide the actual purpose and intention. A typical renaming operation is changing the original name to a single character (like "a", "b") or some kind of puzzling string (like "IlllIlII", "oO00O0oo") [26]. However, the manual check is obviously not qualified for our large-scale scanning goal. Moreover, we focus on the whole APK contents rather than a single identifier. Therefore, we need to design a robust and systematic detection method for identifier renaming.i (3) Classification. The training set is based on an open-source Android app repository – F-Droid [12]. 2For example, if there is a string "abcdefgh", all of the 3-gram sequences it contains are {abc, bcd, cde, def, efg, fgh}. 3The length is restricted by the legal characters sets used for contracting a name in Java: ["a-z", " A-Z", "0-9", "_", "$", "\"]. 4.1 Identifier Renaming We apply different obfuscators on these Android source code to generate obfuscated apps as the ground truth. Lastly, we choose Support Vector Machine (SVM) as the classification algo- rithm. Beyond that, as a special case of identifier renaming, the excessive overloading technique utilizes the overloading feature of Java and could map irrelevant identifier names to the same one, making the code more confusing to analysts [27]. For example, in the sample idfhn1, more than 46 functions are named as idfhn (the same as the package name). Though the compiler could distinguish these variables with the same name, security analysts have to face more troubles. In our research, we also paid attention to the application of overloading feature and its impact on code analysis. Experiment Settings. We implemented a prototype of our de- tection model based on Androguard [4] with more than 1,500 Python lines of code. For training, we downloaded 3,147 apps and their corresponding source code from F-Droid. Two obfuscators, ProGuard and DashO, were used to generate the obfuscated samples because they have different renaming policies. Note that, due to the diversity of apps’ project configurations, not all of them can be processed by both ProGuard (2,107 successful samples) and DashO (654 successful samples). Among them, we randomly chose 500 original apps and 500 obfuscated apps (250 for ProGuard and 250 for DashO) as the training set.f Identifier Renaming Detection. To the above challenges and targets, we combine the computational linguistics and machine learning techniques for accurate renaming detection. The high- level idea is based on the probabilistic language model. The insight is that identifier renaming will lead to the abnormal distribution of characters and character combinations, which can be used to distinguish from normal ones (non-obfuscated). Here we give our three-step approach: We then conducted three steps to validate the effectiveness of our renaming detection model. First, we randomly selected 1,000 original apps and did manual check to make sure that they were non-obfuscated. Our classifier completely correctly labeled these apps as "original", which means the false positive rate is 0%. We then tested our model on 1,000 obfuscated apps(500 obfuscated by Proguard and 500 by DashO) and our model mis-classified 6 samples(5 from Proguard, 1 from DashO), reaching a 0.6% false (1) Data Pre-processing. 1MD5: 7d9eb791c09b9998336ef00bf6d43387 3The length is restricted by the legal characters sets used for contracting a name in Java: ["a-z", " A-Z", "0-9", "_", "$", "\"]. 4.1 Identifier Renaming Based on the result of excessive overloading detection, we find: ⇒1. The deployment rate of excessive overloading approximates that of identifier renaming. ⇒2. Malware may use irrelevant names to hide the true intention. Based on the result of excessive overloading detection, we find: Based on the result of excessive overloading detection, we find: ⇒1. The deployment rate of excessive overloading approximates that of identifier renaming. ⇒2. Malware may use irrelevant names to hide the true intention. Large-scale Investigation and Findings. The purpose of our study is to plot the current usage status of Android obfuscation in the wild. Therefore, we carried out a large-scale detection on the three typical datasets (Google Play, third-party markets, and malware) mentioned in Section 3.2. The obfuscation detection result by dataset is given in Figure 2. According to such statistics, we have two immediate findings: Our statistics show that most of the excessive overloading cases appear along with identifier renaming. The reason may derive from that many obfuscators configure the excessive over- loading by default. For example, Proguard provides the option "-overloadaggressively" for convenient deployment.ii To the second finding, we find there are also some non-name- obfuscated samples applying overloading to confuse analysts. In sample tw.org.ncsist.mdm4, the name of overloaded function attachBaseContext (A protected method in class android.app. Application) will mislead security analysts because the logic of this function is implemented for encryption. ⇒1. Compared with the apps on Google Play, the ones from third-party markets apply more renaming operations. ⇒2. Over one third of malware don’t apply identifier renaming. ⇒1. Compared with the apps on Google Play, the ones from third-party markets apply more renaming operations. ⇒2. Over one third of malware don’t apply identifier renaming. To the first finding, we ascribe it to the discrepancy between app market environments. The piracy issue in Chinese app markets are quite severe [43], say nearly 20% apps are repacked or cloned [30]. Such situation urges developers to put more effort into protecting their apps. On the other hand, Google Play provides more strict and timely supervision, which mitigates the severity of software piracy largely. The better application ecosystem makes many developers believe obfuscation is just an optional protection approach.i 4MD5: 01a93f7e94531e067310c1ee0f083c07 4.1 Identifier Renaming As the most frequently used three identifiers, the names of all classes, methods, and fields of the target APK sample are extracted as the training candi- dates. Note that, software developers often introduce third- party libraries into their apps instead of redevelopment. However, those third-party libraries may also contain obfuscated code, which can not reflect the protection 4 43.0% 57.0% Google Play 73.0% 27.0% Third-party Markets 63.5% 36.5% Malware Obfuscated (Renaming) Non-obfuscated (Renaming) Figure 2: Ratio of Identifier Renaming in Three Datasets In addition, we explored the difference in renaming implementa- tion between malware and benign apps. The result reflects: Malware Google Play ⇒Malware authors prefer to use more complex renaming policies. We find that, in benign apps (the samples on Google Play and third-party markets), most identifier names are mapped to {a, b, aa, ab, aaa, . . . } and so on, in lexicographic order. In fact, such renaming rules accord with the default configurations of many obfuscators (such as ProGuard). That is to say, app developers do not intend to change the renaming rules to more ingenious ones. However, malware authors usually put more effort into configuring the renaming policies. For example, some malware samples utilize special characters (encoded in Unicode) as obfuscated names (e.g., È, ô), which seems very odd but still be regarded as legal by Java compilers. Also, some dazzling weird names (like {IlllIlII, oO00O0oo, . . . }) could be found. Figure 2: Ratio of Identifier Renaming in Three Datasets Figure 2: Ratio of Identifier Renaming in Three Datasets negative rate in total. Due to identifier renaming will lead to an abnormal distribution of character combinations, we consider our model can be generalized to other obfuscators even if they have different implementing policies. To verify this, we conducted a third experiment. We collected another testing set consisting of 200 samples obfuscated by another obfuscator Allatori. The completely successful classification results showed our model’s good attribute of generalization. negative rate in total. Due to identifier renaming will lead to an abnormal distribution of character combinations, we consider our model can be generalized to other obfuscators even if they have different implementing policies. To verify this, we conducted a third experiment. We collected another testing set consisting of 200 samples obfuscated by another obfuscator Allatori. The completely successful classification results showed our model’s good attribute of generalization. 4.2 String Encryption 5 0.0% 100.0% Google Play 0.1% 99.9% Third-party Markets 5.3% 94.7% Malware Obfuscated (Encryption) Non-obfuscated (Encryption) Figure 3: Ratio of String Encryption in Three Datasets ⇒1. Nearly all benign apps don’t use string encryption. ⇒2. String encryption is more popular in malware. Malware Google Play This statistical result complies with our perception, and we could understand it from two perspectives. (1) String encryption is not a common feature provided by off-the-shelf obfuscators. For example, ProGuard [19], as the default obfuscator integrated into Android Studio, does not provide such option. The obfuscators offering the string encryption feature are either expensive (DexGuard [10], DexProtector [11]) or difficult to configure (Allatori [3]). (2) Many developers may lack the knowledge or awareness of deploying more advanced obfuscation techniques. They may believe the default identifier renaming is enough for code protection and it is not necessary to consider other techniques. (3) String encryption can help malware evade the signature scanning of some anti-virus software and hidden the intention effectively, leading to a higher rate of utilization than benign apps. Figure 3: Ratio of String Encryption in Three Datasets In the implementation, we reused most code of identifier renaming detection model. Since string encryption is not a common function provided by off-the-shelf obfuscators, we chose DashO and DexProtector to generate the ground truth and finally obtained 737 string-encrypted samples for training. To avoid the overfitting caused by unbalanced data, we randomly selected 500 original apps and 500 string-encrypted apps to train our model. To verify the effectiveness, we randomly selected another 100 original apps and 100 string-encrypted apps for testing. The result shows our model could achieve 98.5% success rate with FP 1% and FN 2%. In addition, we also conducted an experiment targeting at the implementations of cryptographic functions for obfuscation. In this analysis, we focused on the malware set because the other two benign datasets can not provide enough string-encrypted samples. Finally, we obtained 1,190 cryptographic functions. Base on the further reviews, we get the following findings. ⇒The cryptographic functions usually disguise its true inten- tion by changing to an irrelevant name. Cryptographic Function Analysis. Previous work has proposed various approaches to identify cryptographic functions in a pro- gram, like [29, 34, 41]. Those methods were specifically designed for the identification of the standard, modern cryptographic algorithms in binary code, like AES, DES, and RC4. 5MD5: fab2711b0b55eb980f44bfebc2c17f1f 6MD5: 95f7d37a60ef6d83ae7443a3893bb246 4.2 String Encryption The strings in a .dex (Dalvik executable) file may leak a lot of private information about the program. As security protection, those hard-coded texts can be stored in an encrypted form to prevent reverse analysis. In this section, we take a deep insight into the string encryption and focus on two aspects: To the second finding, the percentage of malware utilizing identifier renaming is only 63.5%, slightly less than third-party apps, which is opposite our traditional opinion. After manually checking the code of malware without renaming-obfuscation, we conclude that two aspects contribute to such phenomenon. (1) Detect whether an app uses the string encryption. (2) Analyze the cryptographic functions invoked by apps. String Encryption Detection. Similar to the approach for identi- fier renaming detection (Section 4.1), we trained a machine-learning based model to classify encrypted strings and plain-text strings. We reused the n-gram algorithm, SVM algorithm, and the open-source apps from F-Droid. Here we only describe the different steps. At first, all strings appeared in an app are extracted. Next, a vector was generated for each app via 3-gram algorithm. Distinct from the setting for identifier renaming detection, there is no restriction on the content of a string. Therefore, we extended the acceptable character set to all ASCII codes. String Encryption Detection. Similar to the approach for identi- fier renaming detection (Section 4.1), we trained a machine-learning based model to classify encrypted strings and plain-text strings. We reused the n-gram algorithm, SVM algorithm, and the open-source apps from F-Droid. Here we only describe the different steps. At first, all strings appeared in an app are extracted. Next, a vector was generated for each app via 3-gram algorithm. Distinct from the setting for identifier renaming detection, there is no restriction on the content of a string. Therefore, we extended the acceptable character set to all ASCII codes. • Script Kiddies. Many entry-level malware authors only could develop simple malicious apps and lack the knowl- edge of how to disguise malicious behaviors through obfuscation. • False Alarmed "Malware". For some apps, their main bodies are benign and non-obfuscated, while the imported third-party libraries contain some kinds of sensitive and suspicious behaviors which are recognized as malicious by some anti-virus software. A common example is the advertising library. 5MD5: fab2711b0b55eb980f44bfebc2c17f1f 6 4.2 String Encryption The features used by the previous commonly include entropy analysis, searchable constant patterns, excessive use of bitwise arithmetic operations, memory fetch patterns and so on, besides, the dynamic binary instrument is also widely-used by analysts to better locate and identify the cryptographic primitives. However, previous approaches do not fit android platform very well due to three reasons: (1) Smali instructions have different representations from the x86 assembly language, especially for memory access. (2) Java provides the complete implementations of standard cryptographic algorithms through Java Cryptography Extension [15]. Therefore, in most cases, developers do not need to implement cryptographic related functions again. (3) Java provides a series of string & character operations, like concat(), substring(), getChars(), strim() and so on, which can be used to build an encrypted string.i For instance, in sample com.solodroid.materialwallpaper5, the decryption function is disguised to a common legitimate API NavigationItem;->getDrawable() which should be used for retrieving a drawable object. ⇒About 17.6% of string-encrypted malware implement multiple cryptographic functions and take turns to use them in a single app. In sample com.yandex.metrica6, four different cryptographic functions were implemented. All of them share similar code structures – first initializing the key, then doing the encryption/de- cryption. However, the key initialization procedures are quite different from each other. As a result, the workload of restoring rises significantly for analysts. 1 // In class com.yandex.metrica.impl.ad; 2 static final String a(String str){ 3 if (c == null){ 4 a13840 (); // key initialization function 5 } 6 Continue ... 7 } ⇒The secret keys used in cryptographic functions can be statically defined or dynamically generated. To better handle the identification in Android apps, we extended the previous approaches with more empirical features, shown as below. • The ratio of bit and loop operations. ⇒The secret keys used in cryptographic functions can be statically defined or dynamically generated. • The usage of Java Cryptography Extension API invoking. • The amount of operations on string & character variables. • The frequency of encrypted strings as function parameters (for decryption function). In the static case, the key is either hard-coded or directly imported as the parameter, which can be easily located and obtained. On the other hand, the dynamic key is usually generated at runtime Large-scale Investigation and Findings. We applied our string encryption detection model on the testing datasets. The results are presented in Figure 3. 6MD5: 95f7d37a60ef6d83ae7443a3893bb246 4.2 String Encryption The direct findings are that: Large-scale Investigation and Findings. We applied our string encryption detection model on the testing datasets. The results are presented in Figure 3. The direct findings are that: 6 48.3% 51.7% Google Play 49.7% 50.3% Third-party Markets 51.0% 49.0% Malware Reflection No Reflection Google Play Malware and even could be fluctuating in different runtime context, which is nearly impossible to be handled by static analysis. The following code snippet shows an example of dynamic key generation, in which elements[3] is not a fixed value because of the uncertain stack trace at runtime. 1 StackTraceElement [] elements = Thread. currentThread ().getStackTrace (); 2 int hashCode = elements [3]. getClassName ()+ elements [3]. getMethodName ().hashCode (); 4.3 Reflectionl Figure 4: Ratio of Reflection in Three Datasets Figure 4: Ratio of Reflection in Three Datasets Reflection allows programs to create, modify and access an object at runtime, which brings many flexibilities. However, such dynamic feature also impedes static analysis due to those reflective invoca- tions, especially those invoking other functions. Such uncertain behaviors could result in that the static analysis cannot capture the real intention.l Dataset Google Play 3rd-p Markets Malware Recovery 65.7% 50.2% 27.1% 2 const-string v0,'android.os.SystemProperties ' 3 invoke-static v0,Ljava/lang/Class;->forName(Ljava/ lang/String ;) Ljava/lang/Class; 4 const-string v2, 'get' 5 . . . 6 invoke-virtual v0, v2, v3, Ljava/lang/Class;-> getMethod(Ljava/lang/String; [Ljava/lang/ Class ;) Ljava/lang/reflect/Method; In this section, we explore two questions on reflection:l (1) How widespread the reflection is used in the wild? (2) Among these use cases, how many of them are used for the obfuscation purpose?lf Reflection provides diverse APIs targeting at different objects like Class, Method and Field. In practice, particular APIs are often executed in sequence to achieve specific functionalities. In our study, we focus on the sequence pattern [Class.forName() → getMethod() →invoke()] which is the most frequent pattern for reflective calls mentioned by Li et al. [38]. Also, in this sequence, the execution of program is implicitly transfered to another function (the parameter of getMethod()), which has an obvious influence on program status, especially the control flow.il Large-scale Investigation and Findings. The implementation of our detection models (reflection usage and invoked functions in reflection) is still based on Androguard with around 1600 Python lines of code. After experiments on our APK dataset, the reflection statistics are shown in Figure 4. We could find: Reflection Detection. The first target is fast reflection detection, which could be achieved through signature searching, say the se- quence pattern [Class.forName() →getMethod() →invoke()].l quence pattern [Class.forName() →getMethod() →invoke()]. Another target is to discover the invoked function in reflection, that is the input parameter of reflective calls. In theory, dynamic analysis is the best way to find the input parameter. However, its low path coverage and efficiency issues are not suitable for large-scale scanning. To balance the efficiency and coverage, we developed a light-weight tool to trace the input parameters of Class.forName() and getMethod(). The high-level idea is to find the real content of the parameters through backward slicing.i ⇒The proportions of reflection deployment in benign apps and malware are similar. 4.3 Reflectionl ⇒The proportions of reflection deployment in benign apps and malware are similar. We are also interested in the purposes of applying reflec- tion in apps. Since our detection model does not work at the dynamic level, part of the invoked targets cannot be precisely acquired. To some complex invoking cases, our model will try to record relevant information as much as possible. For example, if the real target is delivered as the return value of another function, our tool will record the information of this function. The percentage of recovered targets is shown in Table 2, which indicates malware hold the least recovery rate among the three datasets. Furthermore, we checked the results of our backward slicing prototype and found that most of the strings delivered to reflection calls in malware are the return values of certain cryptographic functions, like Ltp5x/WGt12/StringDecoder; → decode(Ljava/lang/String;)Ljava/lang/String;. More details, first our tool scans the function body and locates two reflection calls – Class.forName() and getMethod(). The parameter registers will be set as slicing criterion. Then it traces back from the locations, analyzing each instruction to find the corresponding slices. After that, this tool parses and simulates each instruction in slices, and calculates the final value of the slicing criterion. Note that, to reduce the maintenance complexity, we do not carry out recursive function invoking resolution. Here, we use a real-world example (see the below code block) to illustrate such work flow. In this case, our tool will mark the positions of blue-highlighted reflective calls and trace the data flow of red-highlighted registers. The final output would be {"android.os.SystemProperties", "get"}. To the successfully recovered functions, we further explore why these reflection implementations are necessary. According to different APK dataset, the most frequently invoked functions are listed in Table 3, Table 4, and Table 5 respectively. These lists reflect: const/4 v1, 0 7 Table 3: Functions Invoked via Reflection (Google Play) Frequency Recovered Function 2,275 android.support.v4.content. LocalBroadcastManager.getInstance 1,297 android.webkit.WebView.onPause 1,250 android.os.SystemProperties.get 821 org.apache.harmony.xnet.provider.jsse .NativeCrypto.RAND_seed 523 com.google.android.gms.common.GooglePlay- ServicesUtil.isGooglePlayServicesAvailable Table 3: Functions Invoked via Reflection (Google Play) As one example, the following code block is extracted from an obfuscated malware7. After analysis, the function invoked by reflection could be restored as: 1 if (!ò.trim().toLowerCase ().contains(ˆ0("G))OCH" ))) {Function Body} 1 if (!ò.trim().toLowerCase ().contains(ˆ0("G))OCH" ))) {Function Body} 1 if (!ò.trim().toLowerCase ().contains(ˆ0("G))OCH" ))) {Function Body} As comparison, the original code is shown below. 4.3 Reflectionl In this case, all string operations can be written in non-reflection forms. We could find such reflection usage makes the code structure more complicated and confusing, which enhances the effect of code obfuscation. if (!(( Boolean) Class.forName("java.lang.String ").getMethod("contains", new Class ({ CharSequence.class }).invoke(Class.forName(" java.lang.String").getMethod("toLowerCase", null).invoke(Class.forName("java.lang. String").getMethod("trim", null).invoke(ò, null), null), new Object []{ˆ0("G))OCH")})). booleanValue ()) { Function Body } if (!(( Boolean) Class.forName("java.lang.String ").getMethod("contains", new Class ({ CharSequence.class }).invoke(Class.forName(" java.lang.String").getMethod("toLowerCase", null).invoke(Class.forName("java.lang. String").getMethod("trim", null).invoke(ò, null), null), new Object []{ˆ0("G))OCH")})). booleanValue ()) { Function Body } Table 4: Functions Invoked via Reflection (3rd-p Market) Frequency Recovered Function 3,859 android.os.SystemProperties.get 1,800 android.support.v4.content. LocalBroadcastManager.getInstance 1,158 org.apache.harmony.xnet.provider.jsse .NativeCrypto.RAND_seed 721 android.os.ServiceManager.getService 613 android.os.Build.hasSmartBar Table 4: Functions Invoked via Reflection (3rd-p Market) Frequency Recovered Function 3,859 android.os.SystemProperties.get 1,800 android.support.v4.content. LocalBroadcastManager.getInstance 1,158 org.apache.harmony.xnet.provider.jsse .NativeCrypto.RAND_seed 721 android.os.ServiceManager.getService 613 android.os.Build.hasSmartBar Table 4: Functions Invoked via Reflection (3rd-p Market) 7MD5: 7ff1b8afd22c1ed77ed70bfc04635315 Table 5: Functions Invoked via Reflection (Malware) Table 5: Functions Invoked via Reflection (Malware) Frequency Recovered Function 2,977 java.lang.String.valueOf 2,142 android.telephony.gsm.SmsManager.getDefault 687 android.os.SystemProperties.get 518 java.lang.String.charAt 352 java.lang.String.equals Table 5: Functions Invoked via Reflection (Malware) Frequency Recovered Function 2,977 java.lang.String.valueOf 2,142 android.telephony.gsm.SmsManager.getDefault 687 android.os.SystemProperties.get 518 java.lang.String.charAt 352 java.lang.String.equals Packing Detection. Our study shows the apps using packing usually have the following heuristic features: (1) Derived Application Class. android.app.Application is the base class maintaining the global app state. When launching an app, this class (or its subclass) will be instantiated first. The operation of packing apps usually needs a derived Application class acting as the wrapper, preparing for the subsequent APK loading. (2) Encrypted Data File. The real APK is usually encrypted and stored in the lib or assets folder. ⇒Most of the reflection cases are used to invoke hidden functions or to support backward compatibility. (3) Thin Wrapper Class. In general, the wrapper class only performs the bootstrap function, and the core work is performed by native functions based on Java Native Interface (JNI) [16]. In Android system, the functions related to the Android frame- work and OS itself are usually annotated with the label "@hide", which can only be called through reflection. In above three tables, all functions starting with android.os. and android.webkit.* are hidden-annotated. Also, the packing tools always introduce new files (such as ijiami.data and baiduproduct.jar) or code into the original APK file. These modifications usually differ from one packing service to another and can be the fingerprints of service providers. Those certain modifications could be treated as a detection feature as well. To further study, we tested six popular packing services and analyzed the corresponding packed APK files. The extracted signatures are listed in Table 6. Noted that, such signatures may be changed with the update of packing service. We also manually checked the use case of android.v4.content. LocalBroadcastManager.getInstance. We found that the corre- sponding reflective calls are usually enclosed in a try-catch block, aiming to check the existence of particular class and handle the not-found exception. Such pattern is a programming standard recommended by Android official documents [7].i Large-scale Investigation and Findings. We applied our pack- ing detection prototype (300 Python lines of code) to the three APK datasets. The statistical results are shown in Figure 5. The direct finding is: To malware samples, we find: ⇒Compared with benign apps, malware prefers to use more complex reflection invoking patterns to hide its intentions. 4.4 Packingf Different from previous three obfuscation techniques, packing is a kind of whole-APK-reinforcing protection, which does not aim at preventing others from understanding the code, but preventing the code from being obtained. Currently, many packing services are provided as online services and free for individual users, such as Qihoo [2], ijiami [14], and Bangcle [8]. Table 5: Functions Invoked via Reflection (Malware) 8 Table 6: Signatures of Packing Services Table 6: Signatures of Packing Services Table 6: Signatures of Packing Services Packer File Signature(s) Code Signature(s) Ali lib/armeabi/libmobisec.so \| aliprotect.dat com.ali.fixHelper \| com.ali.mobisecenhance.StubApplication Tencent lib/armeabi/libmain.so \| lib/armeabi/libshell.so \| lib/armeabi/mix.dex com.tencent.StubShell Qihoo assets/libjiagu.so com.qihoo.util.StubApplication iJiami assets/ijiami.dat \| /armeabi/libexec.so \| /armeabi/libexecmain.so com.shell.SuperApplication Bangcle assets/bangcle_classes.jar \| lib/armeabi/libsecexe.so \| lib/armeabi/libsecmain.so com.secshell.shellwrapper.SecAppWrapper \| com.bangcle.protect.ApplicationWrapper Baidu assets/baiduprotect.jar \| lib/armeabi/libbaiduprotect.so com.baidu.protect.StubApplication 0.0% 100.0% Google Play 10.4% 89.6% Third-party Markets 2.1% 97.9% Malware Obfuscated (Packing) Non-obfuscated (Packing) Figure 5: Ratio of Packing in Three Datasets 0.0% 100.0% Google Play 10.4% 89.6% Third-party Markets 2.1% 97.9% Malware Obfuscated (Packing) Non-obfuscated (Packing) Figure 5: Ratio of Packing in Three Datasets Through our observation, we find that control flow obfuscation is non-universal and only provided by a minority of obfuscators, like DashO and Allatori. However, based on our analysis, these tools do not provide a strong control flow obfuscation method as they claimed. For example, given an app, only very few methods’ control flows are obfuscated, and the others remain unchanged. Therefore, at this stage, we cannot capture enough control-flow obfuscated samples for investigation. Another uncovered topic is native code obfuscation which could bring more protection to an app’s binary code. However, native code programming requires more advanced skills for developers, which makes it still not a mainstream technique in Android app development. Also, the implementation of native code obfuscation is quite different from other Java-level techniques, which could be treated as an independent research topic. Therefore, we leave it as our future study. Figure 5: Ratio of Packing in Three Datasets ⇒Third-party apps and malware held a higher deployment rate of packing services. 6.1 Obfuscation Measurement and Assessment Obfuscation techniques have been widely used in the Android app development. Naturally, in academia, researchers are interested in whether these techniques do work. An early attempt is [33] which empirically evaluates a set of 7 obfuscation methods on 240 APKs. Also, Park et al. [44] empirically analyzed the effects of code obfuscation on Android app similarity analysis. Recently, Faruki et al. [32] conducted a survey to review the mainstream Android code obfuscation and protection techniques. However, they concentrated on the technical analysis to evaluate different techniques, not like our work based on a large-scale dataset. They show that many obfuscation methods are idempotent or monotonous. Wang et al. [51] defined the obfuscator identification problem for Android and proposed a solution based on machine learning techniques. The experiments indicated that their approach could achieve about 97% accuracy to identify ProGuard, Allatori, DashO, Legu, and Bangcle. On the aspect of deobfuscation research, Bichsel et al. [28] proposed a structured prediction approach for According to our observation, packing is a practical approach to code protection for ordinary developers. Its basic functionality has been able to impede entry-level reverse-engineers from peeping into the original code. However, the protection may be not strong enough to prevent an adept analyst from obtaining the code. 6 RELATED WORK Obfuscation is always a hot research topic in Android ecosystem, and there are several studies performed on how to obfuscate Android apps effectively and how to measure the obfuscation effectiveness. As an one-stop approach of code protection, the popularity of online packing service is reasonable. Currently, the research on packing and unpacking has become a hot topic, and researchers have proposed several tools targeting at unpacking apps automati- cally, like Zhang et al. [54] and Yang et al. [52]. Most of these tools rely on dumping the code from memory through customized Dalvik virtual machine (DVM) or Android Runtime (ART). As arm races, packing providers enhance their services time to time to prevent cracking. 6.2 Security Impact of Android Obfuscation [3] Accessed: September 2017. Allatori. http://www.allatori.com/. (Accessed: September 2017). As discussed earlier, the obfuscation will create barriers for Android program analysis. Works on clone / repackage detection [35, 42, 50, 53, 55] find that obfuscations can impair detection results. [4] Accessed: September 2017. androguard. https://github.com/androguard/ androguard. (Accessed: September 2017). [5] Accessed: September 2017. Anzhi. http://www.anzhi.com/. (Accessed: September 2017). i p Studies of malware detection also showed that obfuscation is an obstacle to malware analysis. Rastogi et al. [47] evaluated several commercial mobile anti-malware products for Android and tested how resistant they are against various common obfuscation techniques. Their experiment result showed anti-malware tools make little effort to provide transformation-resilient detection (in the year 2013). After that, Maiorca et al. [40] conducted a large-scale experiment in which the detection performance of anti-malware solutions are tested against malware samples under different obfuscation strategies. Their results showed the improvement of anti-malware engines in recent years. Recently, Hoffmann et al. [36] developed a framework for automated obfuscation, which implemented fine-grained obfuscation strategies and could be used as test benches for evaluating analysis tools. Similar works are also completed by Preda et al. [46], Pomilia [45], and Faruki et al. [31]. To handle obfuscated samples, Suarez-Tangil et al. [49] propose DroidSieve, an Android malware classifier based on static analysis and deep inspection that is resilient to obfuscation. ) [6] Accessed: September 2017. AppChina. http://www.appchina.com/. (Accessed: September 2017). [7] Accessed: September 2017. Backward compatibility for Android applications. https://android-developers.googleblog.com/2009/04/ backward-compatibility-for-android.html. (Accessed: September 2017). [7] Accessed: September 2017. Backward compatibility for Android applications. https://android-developers.googleblog.com/2009/04/ backward-compatibility-for-android.html. (Accessed: September 20 [8] Accessed: September 2017. bangcle. http://www.bangcle.com/. (Accessed: September 2017). [9] Accessed: September 2017. DashO. https://www.preemptive.com/products/ dasho/overview. (Accessed: September 2017). [10] Accessed: September 2017. DexGuard. https://www.guardsquare.com/en/ dexguard. (Accessed: September 2017). [11] Accessed: September 2017. DexProtector. https://dexprotector.com/. (Accessed: September 2017). [12] Accessed: September 2017. F-Droid. https://f-droid.org/. (Accessed: September 2017). [13] Accessed: September 2017. Huawei Appstore. http://appstore.huawei.com/. (Accessed: September 2017). [14] Accessed: September 2017. ijiami. http://www.ijiami.cn/. (Accessed: September 2017). [15] Accessed: September 2017. Java Cryptography Extension. http://www.oracle. com/technetwork/java/javase/downloads/jce8-download-2133166.html. (Ac- cessed: September 2017). p ) [16] Accessed: September 2017. Java Native Interface. http://docs.oracle.com/javase/ 7/docs/technotes/guides/jni/. (Accessed: September 2017). g p [17] Accessed: September 2017. n-gram. https://en.wikipedia.org/wiki/N-gram. (Accessed: September 2017). For malware detection, researchers mainly discussed arms race between obfuscation and malware detection. 5 DISCUSSION In this section, we discuss some limitations of our study and then describe the future plan. Though we have conducted a large-scale investigation of mainstream obfuscation techniques used in Android apps, we should point out there are still some existing techniques not involved in our research, say control flow obfuscation and native code obfuscation. 9 performing probabilistic layout deobfuscation of Android APKs and implemented a scalable probabilistic system called DeGuard. Different from above research, our work is based on a large Android app datasets which cover official Google play store, third- party Android markets, and update-to-date malware families. We attempt to understand the distribution of Android obfuscation techniques and provide the up-to-date knowledge about app protection. put great efforts on more advanced code protection skills. Also, we provide the corresponding illustrations to enlighten developers to select the most suitable code protection methodologies and help researchers improve code analysis systems in the right direction. performing probabilistic layout deobfuscation of Android APKs and implemented a scalable probabilistic system called DeGuard.f Different from above research, our work is based on a large Android app datasets which cover official Google play store, third- party Android markets, and update-to-date malware families. We attempt to understand the distribution of Android obfuscation techniques and provide the up-to-date knowledge about app protection. 6.2 Security Impact of Android Obfuscation Although some malware detection tools claim to still work well in the presence of obfuscation, none could eliminate the obfuscation effects in their experimental evaluation. Our study focuses on the empirical study of security impacts of obfuscation in the wild from different views, which are complementary to existing works. That is, we statistically evaluate the distribution of obfuscation methods from views of different markets, hardening capability of obfuscations and temporal evolution, with a light-weight and scalable obfuscation detection framework. We believe some of our findings would be useful for developers and researchers to better understand the usage of obfuscation, for example, keeping pace with the development of obfuscation technique. [18] Accessed: September 2017. Number of available applications in the Google Play Store from December 2009 to March 2017. http://www.statista.com/statistics/ 266210/number-of-available-applications-in-the-google-play-store/. (Accessed: September 2017). [19] Accessed: September 2017. ProGuard. https://www.guardsquare.com/en/ proguard. (Accessed: September 2017). [20] Accessed: September 2017. Shield4J. http://shield4j.com/. (Accessed: September 2017).l ) [21] Accessed: September 2017. Trail: The Reflection API. http://docs.oracle.com/ javase/tutorial/reflect/index.html. (Accessed: September 2017). l javase/tutorial/reflect/index.html. (Accessed: September 2017). l [22] Accessed: September 2017. VirusShare. https://virusshare.com/. (Accessed: September 2017). p [23] Accessed: September 2017. VirusTotal. https://www.virustotal.com/. (Accessed: September 2017). [24] Accessed: September 2017. Wandoujia. https://www.wandoujia.com/. (Accessed: September 2017). [25] Accessed: September 2017. Xiaomi Application Store. http://app.mi.com/. (Accessed: September 2017). p [26] Axelle Apvrille and Ruchna Nigam. 2014. Obfuscation in Android malware, and how to fight back. Virus Bulletin (2014), 1–10. REFERENCES [1] Accessed: September 2017. 360 Smartphone Assistant. http://zhushou.360.cn/. (Accessed: September 2017). [1] Accessed: September 2017. 360 Smartphone Assistant. http://zhushou.360.cn/. (Accessed: September 2017). [ ] p (Accessed: September 2017). [2] Accessed: September 2017. 360jiagu. http://jiagu.360.cn/. (Accessed: September 2017). 7 CONCLUSION Sapienza University of Rome. Master’s thesis. Sapienza University of Rome. [33] Felix C. Freiling, Mykola Protsenko, and Yan Zhuang. 2014. An Empirical Evaluation of Software Obfuscation Techniques Applied to Android APKs. In International Conference on Security and Privacy in Communication Networks - 10th International ICST Conference, SecureComm 2014, Beijing, China, September 24-26, 2014, Revised Selected Papers, Part II.i [46] Mila Dalla Preda and Federico Maggi. 2017. Testing android malware detectors against code obfuscation: a systematization of knowledge and unified methodol- ogy. J. Computer Virology and Hacking Techniques 13, 3 (2017), 209–232. [47] Vaibhav Rastogi, Yan Chen, and Xuxian Jiang. 2013. DroidChameleon: Evaluating Android Anti-malware against Transformation Attacks. In Proceedings of the 8th ACM Symposium on Information, Computer and Communications Security (ASIACCS), Hangzhou, China, May 8-10, 2013. p [34] Felix Gröbert, Carsten Willems, and Thorsten Holz. 2011. Automated Identifi- cation of Cryptographic Primitives in Binary Programs. In Recent Advances in Intrusion Detection - 14th International Symposium, RAID 2011, Menlo Park, CA, USA, September 20-21, 2011. Proceedings. ( ) g y [48] Junliang Shu, Juanru Li, Yuanyuan Zhang, and Dawu Gu. 2014. Android App Protection via Interpretation Obfuscation. In Proceeding of the 12th IEEE International Conference on Dependable, Autonomic and Secure Computing (DASC), Dalian, China, August 24-27, 2014. [35] Quanlong Guan, Heqing Huang, Weiqi Luo, and Sencun Zhu. 2016. Semantics- Based Repackaging Detection for Mobile Apps. In Proceedings of the 8th International Symposium Engineering Secure Software and Systems (ESSoS), London, UK, April 6-8, 2016.f g [49] Guillermo Suarez-Tangil, Santanu Kumar Dash, Mansour Ahmadi, Johannes Kinder, Giorgio Giacinto, and Lorenzo Cavallaro. 2017. DroidSieve: Fast and Accurate Classification of Obfuscated Android Malware. In Proceedings of the Seventh ACM on Conference on Data and Application Security and Privacy (CODASPY), Scottsdale, AZ, USA, March 22-24, 2017. [36] Johannes Hoffmann, Teemu Rytilahti, Davide Maiorca, Marcel Winandy, Giorgio Giacinto, and Thorsten Holz. 2016. Evaluating Analysis Tools for Android Apps: Status Quo and Robustness Against Obfuscation. In Proceedings of the Sixth ACM on Conference on Data and Application Security and Privacy (CODASPY), New Orleans, LA, USA, March 9-11, 2016. [50] Haoyu Wang, Yao Guo, Ziang Ma, and Xiangqun Chen. 2015. WuKong: A Scalable and Accurate Two-Phase Approach to Android App Clone Detection. In Proceedings of the 2015 International Symposium on Software Testing and Analysis (ISSTA), Baltimore, MD, USA, July 12-17, 2015.i [37] Heqing Huang, Cong Zheng, Junyuan Zeng, Wu Zhou, Sencun Zhu, Peng Liu, Suresh Chari, and Ce Zhang. 7 CONCLUSION 2016. Android Malware Development on Public Malware Scanning Platforms: A Large-scale Date-driven Study. In Proceeding of the 2016 IEEE International Conference on Big Data (BigData), Washington DC, USA, December 5-8, 2016. y [51] Yan Wang and Atanas Rountev. 2017. Who Changed You? Obfuscator Identifica- tion for Android. In Proceedings of the 4th IEEE/ACM International Conference on Mobile Software Engineering and Systems (MOBILESoft), Buenos Aires, Argentina, May 22-23, 2017. [38] Li Li, Tegawendé F. Bissyandé, Damien Octeau, and Jacques Klein. 2016. DroidRA: Taming Reflection to Support Whole-Program Analysis of Android Apps. In Proceedings of the 25th International Symposium on Software Testing and Analysis (ISSTA), Saarbrücken, Germany, July 18-20, 2016. [52] Wenbo Yang, Yuanyuan Zhang, Juanru Li, Junliang Shu, Bodong Li, Wenjun Hu, and Dawu Gu. 2015. AppSpear: Bytecode Decrypting and DEX Reassembling for Packed Android Malware. In Research in Attacks, Intrusions, and Defenses - 18th International Symposium, RAID 2015, Kyoto, Japan, November 2-4, 2015, Proceedings. [39] Menghao Li, Wei Wang, Pei Wang, Shuai Wang, Dinghao Wu, Jian Liu, Rui Xue, and Wei Huo. 2017. LibD: Scalable and Precise Third-party Library Detection in Android Markets. In Proceedings of the 39th International Conference on Software Engineering (ICSE), Buenos Aires, Argentina, May 20-28, 2017. [53] Fangfang Zhang, Heqing Huang, Sencun Zhu, Dinghao Wu, and Peng Liu. 2014. ViewDroid: towards obfuscation-resilient mobile application repackaging detection. In Proceedings of 7th ACM Conference on Security & Privacy in Wireless and Mobile Networks (WiSec), Oxford, United Kingdom, July 23-25, 2014. Engineering (ICSE), Buenos Aires, Argentina, May 20-28, 2017. [40] Davide Maiorca, Davide Ariu, Igino Corona, Marco Aresu, and Giorgio Giacinto. 2015. Stealth Attacks: An Extended Insight into the Obfuscation Effects on Android Malware. Computers & Security 51 (2015), 16–31. [54] Yueqian Zhang, Xiapu Luo, and Haoyang Yin. 2015. DexHunter: Toward Extracting Hidden Code from Packed Android Applications. In Computer Security - ESORICS 2015 - 20th European Symposium on Research in Computer Security, Vienna, Austria, September 21-25, 2015, Proceedings, Part II. 293–311. [41] Felix Matenaar, André Wichmann, Felix Leder, and Elmar Gerhards-Padilla. 2012. CIS: The Crypto Intelligence System for automatic detection and localization of cryptographic functions in current malware. In Proceeding of the 7th International Conference on Malicious and Unwanted Software (MALWARE), Fajardo, PR, USA, October 16-18, 2012. [55] Wu Zhou, Yajin Zhou, Xuxian Jiang, and Peng Ning. 2012. Detecting repackaged smartphone applications in third-party android marketplaces. 7 CONCLUSION [27] Vivek Balachandran, Sufatrio, Darell J. J. Tan, and Vrizlynn L. L. Thing. 2016. Control flow obfuscation for Android applications. Computers & Security 61 (2016), 72–93. In this paper, we concentrate on exploring the current deployment status of Android code obfuscation in the wild. For this target, we developed specific detection tools for four common obfuscation techniques and performed a large-scale scanning on three represen- tative APK datasets. The results show that, to different techniques and app categories, the status of code obfuscation differs in many aspects. For example, the basic renaming obfuscation has become widely-used among Chinese third-party market developers, while still not pervasive in Google Play market. Besides, malware authors [28] Benjamin Bichsel, Veselin Raychev, Petar Tsankov, and Martin T. Vechev. 2016. Statistical Deobfuscation of Android Applications. In Proceedings of the 2016 ACM SIGSAC Conference on Computer and Communications Security (CCS), Vienna, Austria, October 24-28, 2016. [29] Joan Calvet, José M. Fernandez, and Jean-Yves Marion. 2012. Aligot: Crypto- graphic Function Identification in Obfuscated Binary Programs. In Proceedings of the 19th ACM Conference on Computer and Communications Security (CCS), Raleigh, NC, USA, October 16-18, 2012. 169–182. [30] Kai Chen, Peng Liu, and Yingjun Zhang. 2014. Achieving accuracy and scalability simultaneously in detecting application clones on Android markets. In Proceeding 10 of the 36th International Conference on Software Engineering (ICSE), Hyderabad, India, May 31 - June 07, 2014. [43] Maggie Nazarenus. Jul 2015. Chinese app piracy: Why it exists and what you can do about it. http://www.techinasia.com/talk/chinese-app-piracy-exists. (Jul 2015). y J [31] Parvez Faruki, Ammar Bharmal, Vijay Laxmi, Manoj Singh Gaur, Mauro Conti, and Muttukrishnan Rajarajan. 2014. Evaluation of Android Anti-malware Techniques against Dalvik Bytecode Obfuscation. In Proceedings of the 13th IEEE International Conference on Trust, Security and Privacy in Computing and Communications (TrustCom), Beijing, China, September 24-26, 2014. [44] Jonghwa Park, Hyojung Kim, Younsik Jeong, Seong-je Cho, Sangchul Han, and Minkyu Park. 2015. Effects of Code Obfuscation on Android App Similarity Analysis. Journal of Wireless Mobile Networks, Ubiquitous Computing, and Dependable Applications 6, 4 (2015), 86–98. [32] Parvez Faruki, Hossein Fereidooni, Vijay Laxmi, Mauro Conti, and Manoj Singh Gaur. 2016. Android Code Protection via Obfuscation Techniques: Past, Present and Future Directions. CoRR abs/1611.10231 (2016). p pp ( ) [45] Matteo Pomilia. 2016. A Study on Obfuscation Techniques for Android Malware. Master’s thesis. Sapienza University of Rome. p pp [45] Matteo Pomilia. 2016. A Study on Obfuscation T Master’s thesis. 7 CONCLUSION In Proceedings of the 2nd ACM Conference on Data and Application Security and Privacy (CODASPY), San Antonio, TX, USA, February 7-9, 2012. [42] Jiang Ming, Fangfang Zhang, Dinghao Wu, Peng Liu, and Sencun Zhu. 2016. Deviation-Based Obfuscation-Resilient Program Equivalence Checking With Application to Software Plagiarism Detection. IEEE Trans. Reliability 65, 4 (2016), 11 11
W4311850735.txt	https://www.frontiersin.org/articles/10.3389/fdgth.2022.1065581/pdf	en		Enhanced neurologic concept recognition using a named entity recognition model based on transformers	Frontiers in digital health	2,022	cc-by	6,779	\| EDITED BY Harry Hochheiser, University of Pittsburgh, United States REVIEWED BY Terri Elizabeth Workman, George Washington University, United States, Md Adnanul Islam, Monash University, Australia CORRESPONDENCE Sima Azizi azizi.sima.67@gmail.com TYPE Brief Research Report PUBLISHED 08 December 2022 DOI 10.3389/fdgth.2022.1065581 Enhanced neurologic concept recognition using a named entity recognition model based on transformers Sima Azizi1, Daniel B. Hier1,2 and Donald C. Wunsch II1,3 1 Applied Computational Intelligence Laboratory, Department of Electrical & Computer Engineering, Missouri University of Science & Technology, Rolla, MO, United States, Department of Neurology and Rehabilitation, University of Illinois at Chicago, Chicago, IL, United States, National Science Foundation, ECCS Division, Arlington, VA, United States 2 3 SPECIALTY SECTION This article was submitted to Health Informatics, a section of the journal Frontiers in Digital Health 09 October 2022 21 November 2022 PUBLISHED 08 December 2022 RECEIVED ACCEPTED CITATION Azizi S, Hier DB and Wunsch II DC (2022) Enhanced neurologic concept recognition using a named entity recognition model based on transformers. Front. Digit. Health 4:1065581. doi: 10.3389/fdgth.2022.1065581 COPYRIGHT © 2022 Azizi, Hier and Wunsch II. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. Although deep learning has been applied to the recognition of diseases and drugs in electronic health records and the biomedical literature, relatively little study has been devoted to the utility of deep learning for the recognition of signs and symptoms. The recognition of signs and symptoms is critical to the success of deep phenotyping and precision medicine. We have developed a named entity recognition model that uses deep learning to identify text spans containing neurological signs and symptoms and then maps these text spans to the clinical concepts of a neuro-ontology. We compared a model based on convolutional neural networks to one based on bidirectional encoder representation from transformers. Models were evaluated for accuracy of text span identiﬁcation on three text corpora: physician notes from an electronic health record, case histories from neurologic textbooks, and clinical synopses from an online database of genetic diseases. Both models performed best on the professionally-written clinical synopses and worst on the physicianwritten clinical notes. Both models performed better when signs and symptoms were represented as shorter text spans. Consistent with prior studies that examined the recognition of diseases and drugs, the model based on bidirectional encoder representations from transformers outperformed the model based on convolutional neural networks for recognizing signs and symptoms. Recall for signs and symptoms ranged from 59.5% to 82.0% and precision ranged from 61.7% to 80.4%. With further advances in NLP, fully automated recognition of signs and symptoms in electronic health records and the medical literature should be feasible. KEYWORDS named entity recognition, clinical concepts, concept extraction, phenotype, transformers, natural language processing, annotation I. Introduction Several factors have accelerated interest in the automated recognition of clinical concepts in unstructured text held in electronic health records and electronic publications (1). First, most paper medical records have been converted to electronic health records (EHRs) (2) with as much as 80% of the data held as unstructured text Frontiers in Digital Health 01 frontiersin.org Azizi et al. 10.3389/fdgth.2022.1065581 abbreviations, acronyms, and negation, can generate word variants, and can perform word sense disambiguation (27). In a preliminary study, we found that MetaMap can identify signs and symptoms in neurological case histories with an accuracy of 55–84% (30). Most MetaMap errors were false negatives due to a failure to recognize neurological concepts that had been expressed as descriptions (e.g., reﬂexes were absent) as opposed to those expressed as discrete lexical items (e.g., hyporeﬂexia). In their 2017 literature review of automated information extraction, Wang et al. (31) reviewed 263 information extraction studies and found most centered on identifying diseases or drugs. The most common systems used were MetaMap, MedLEE, and cTAKES (32–36) followed by traditional machine learning algorithms (conditional random ﬁelds, support vector machines, random forests, decision trees, and naive Bayes). Third-generation systems for NER are built on deep learning (37–40). Lample et al. suggested a model for named entity recognition based on an RNN (recurrent neural network) with bidirectional LSTM (long short term memory) and conditional random ﬁelds (CRFs). Vani et al. (41) proposed a “grounded” RNN to predict medical diagnoses based on text from patient discharge summaries. Liu et al. (42) found that on a task to label protected health information in medical records that RNNs based on bidirectional LSTM outperformed those that used CRFs. An LSTM NER model with conditional random ﬁelds (CRFs) has been used to identify ﬁve classes of chemicals, species, genes/ proteins, cell lines, and diseases (43). Hybrid methods that combine rule-based and machine learning-based methods have been proposed to identify protected health information (PHI) in clinical discharge summaries (44). Liu et al. (42) developed a hybrid system to identify clinical information by ensemble learning that combined the instances predicted from a bidirectional LSTM, a CRF model, and a rule-based system (45,46). Gehrmann et al. (47) used a convolutional neural network (CNN) for ten phenotyping tasks and compared it with other common NLP models. Arbabi et al. (48) have created a neural concept recognizer (NCR) that uses CNNs and word embedding to recognize clinical concepts in unstructured text. The NCR uses an encoder to convert input phrases to word vectors and word embedding to convert entries in the target ontology into word vectors. The similarity between the input phrases and concepts in the target ontology is calculated by the dot product. For concept recognition in PubMed abstracts or clinical notes, the NCR outperformed the NCBO Annotator and BioLark (49). RNNs and variants can handle long-term dependency in text, but only for a limited span length. The deep learning architecture transformers can process longer text spans and has shown improved performance on NLP tasks (50). Bidirectional encoder representations from transformers (BERT) have outperformed other neural network architectures on named (3). Second, most medical journals are available electronically (4). Third, the deep phenotyping and precision medicine initiatives have made the detailed description of patient signs and symptoms a key piece of data (5,6). Fourth, automated clinical concept recognition is an important area of natural language processing (NLP) research. Automated concept recognition is closely related to the NLP problems of text mining and named entity recognition. Other important NLP research areas include machine translation, text classiﬁcation, text clustering, speech recognition, question answering, text summarization, sentiment analysis, picture captioning, and natural language understanding (7–14). Krauthammer and Nenadic (1) have divided concept recognition (variously called term identiﬁcation, concept extraction, and information extraction) into three steps: term recognition (identiﬁcation of the text span corresponding to the clinical concept), term classiﬁcation (identiﬁcation of the class membership of the term, i.e., drug, disease, sign, symptom, etc.), and term mapping (linking of the term to an entry in a standard vocabulary with an identiﬁcation code which is also known as “concept normalization” (15)). Clinical concept recognition is closely related to the NLP problem of named entity recognition (NER) in which text spans referring to named entities (people, places, organizations, etc.) are tagged and mapped to dictionaries, gazetteers, or other registries (16). Text spans that encode clinical concepts (diseases, drugs, signs, symptoms, etc.) can be mapped (normalized) to hierarchical ontologies that include SNOMED CT with 352,000 concepts, the Human Phenotype Ontology (HPO) with 20,000 concepts, the Online Mendelian Inheritance in Man ontology (OMIM) with 97,000 concepts, or the UMLS Metathesaurus with 4.6 million concepts (17–20). The NLM UMLS Metathesaurus maintains interchangeable machinereadable codes for SNOMED CT, UMLS, HPO, and the OMIM. Initial NER systems for clinical concept recognition were either dictionary-based, or rule-based (1,21,22). Some secondgeneration NER systems were based on machine learning algorithms such as conditional random ﬁelds, support vector machines, and hidden Markov models (23,24). Other secondgeneration NER systems developed as an outgrowth of advances in semantic and syntactic analysis (25,26). MetaMap utilizes linguistic analysis and statistical algorithms to identify clinical concepts in unstructured text and maps them to machine-readable codes in the UMLS (27,28). The UMLS has grown from 900,000 concepts, and 2 million names in 2004 (29) to 4.6 million concepts and 17 million names in 2022 (20). MetaMap tokenizes text input, ﬁnds sentence boundaries, and uses lexical and syntactic analysis to identify candidate phrases for mapping to concepts in the UMLS. Candidate phrases are compared to target strings in the UMLS, lists of potential clinical concepts are generated, and scored by statistical algorithms. MetaMap can recognize Frontiers in Digital Health 02 frontiersin.org Azizi et al. 10.3389/fdgth.2022.1065581 Since neurologic signs and symptoms can be extracted from both the medical literature and electronic health records, we have tested the concept identiﬁcation pipeline on three corpora: case histories from neurological textbooks, neurological clinical synopses from the Online Mendelian Inheritance of Man (OMIM), and physician neurological notes from an electronic health record. With this work, we propose to address four questions: entity recognition (51,50). For clinical concept recognition, BERT models that are pre-trained on the medical literature (BioBERT) or clinical notes (ClinicalBERT) outperform BERT models pre-trained on general corpora by at least 1% (52–55). A. Proposed approach 1. Does writing style differ by corpus? 2. Does the accuracy of concept recognition differ by corpus? 3. Is the accuracy of clinical concept recognition reduced with longer text spans? 4. Does concept recognition based on BERT outperforms concept recognition based on CNNs? Although considerable work has been done on automated concept identiﬁcation of drugs and diseases, less work has been done on the automated identiﬁcation of signs and symptoms (52). Identifying signs and symptoms is critical to precision medicine and deep phenotyping (56). To make the problem tractable, we limited the signs and symptoms to the specialty of neurology and restricted the target ontology to a neuro-ontology with 1,600 concepts (57). Automating the recognition of signs and symptoms is more challenging than automating the recognition of diseases or drugs for three reasons. First, many neurological signs and symptoms have multiple synonyms; something that is not typical with diseases or drugs. For example, an expressionless face may be described as a “masked face,” or “hypomimia.” Second, physicians variably choose to record signs and symptoms as descriptions or as names. For example, a patient with diplopia can be described as “seeing double” or a patient with nausea can be described as “sick to their stomach.” In contrast, physicians uniformly identify drugs and diseases by name and not by description. Third, the meaning of a term may depend on context. For example, to a neurologist ptosis is a droopy eyelid, but to a gynecologist, ptosis is a prolapsed uterus. We propose to identify and normalize the neurological signs and symptoms found in the unstructured text in two steps: ﬁrst, we have trained a neural network-based named entity recognition model to identify text spans that contain clinical concepts (signs and symptoms). Second, we have normalized identiﬁed text spans by mapping them to clinical concepts in a neuro-ontology using a look-up table and similarity metric (Figure 1). Although the superiority of BERT over other neural networks for concept identiﬁcation is well-established, the contribution of this work is to demonstrate that the accuracy of concept identiﬁcation depends upon text span length and corpus writing style. II. Methods A. Corpora We identiﬁed signs and symptoms (clinical concept identiﬁcation) in three corpora: neurological case histories from ﬁve neurological textbooks (referred to as Textbook Corpus (58–62), clinical synopses of neurological disease from the Online Mendelian Inheritance of Man (referred to as OMIM Corpus) (18), and neurology physician notes from the electronic health record of the University of Illinois at Chicago (referred to as EHR Corpus). The use of de-identiﬁed physician notes was approved by the Institutional Review Board of the University of Illinois at Chicago. Corpora were converted to plain text ﬁles and pre-processed using python. Email addresses, URLs, HTML, special characters, and unnecessary punctuation were removed using regular expressions in python. Contractions were replaced with the expanded form. Misspelled words, separated words, and FIGURE 1 Overview of the pipeline that recognizes text spans that are clinical concepts in three corpora: Textbook neurology case histories, EHR physician notes, and Clinical synopses in the OMIM. Text spans are normalized by mapping to clinical concepts in a neuro-ontology. Frontiers in Digital Health 03 frontiersin.org Azizi et al. 10.3389/fdgth.2022.1065581 hyphenated words were corrected manually using the spelling correction tool in Microsoft Word. Abbreviations were not edited. The pre-processed ﬁles were manually inspected for errors and converted to JSONL ﬁles. C. NN model training and evaluation Two neural network models were trained to recognize text spans that encoded clinical concepts in text corpora. Both models were based on NER pipelines. NER pipelines identify a named entity in a text span and assign the named entity to a predeﬁned category. Each NN model assigned text spans to one of the seven deﬁned categories of clinical concepts (unigram, bigram, trigram, tetragram, extended, compound, and tabular). For each corpus, 80% of the instances were used for training and 20% for evaluation. The baseline NN was the default spaCy named entity recognition model based on a four-layer convolutional neural network (CNN) that looks at four words on either side of each token using the NER pipeline and tok2vec with an initial learning rate 1 103 . The standard word vectors included with spaCy were used for word embedding. The second named entity recognition model was based on BERT (51). The BERT base model was implemented in spaCY (66) and consisted of 12 layers of transformer encoder, 12 attention heads, 786 hidden size, and 100 M parameters. The BERT model was pre-trained with publicly available weights and ﬁne-tuned using our training set. We used the Adam optimizer with a learning rate of 5 105 , b1 ¼ 0:9, b2 ¼ 0:99, a learning rate warm-up over the ﬁrst 500 steps, and a linear decay learning rate. The dynamic batch size was set according to the longest sequence in the batch. The training was conducted over 20,000 steps. The mini-batch size dynamically changed according to the longest sequence in the batch. The largest padded size for batch sequences was 2,000, and the buffer was 256. A GELU activation function was used. For each corpus and each model, the F score, precision, and recall were computed (Table I). B. Text annotation Signs and symptoms in JSONL ﬁles were annotated by a neurologist using the Prodigy annotation tool (63,64). An interrater reliability study with two other raters based on ﬁfteen neurology notes showed an unadjusted agreement rate for text span annotation of 89% and a kappa statistic of 0.85 (65). Each sign or symptom was tagged as a unigram, bigram, trigram, tetragram, extended, compound, or tabular concept. Unigrams were signs and symptoms of length one-word such as alexia, hyperreﬂexia, or bradykinesia. Bigrams were signs and symptoms of length two-words such as double vision, facial weakness, and poor balance. Trigrams were signs of symptoms of length three-words such as absent ankle reﬂex, impaired hand dexterity, or weak ankle dorsiﬂexors. Tetragrams were four-word signs and symptoms such as relative afferent pupil defect and Hoffman sign was present. Text spans were tagged as extended when signs and symptoms were more than four words, such as hand grip was very weak and barely able to lift his legs off the bed. Text spans were tagged as compound when more than one sign or symptom was combined in a single text span such as decreased vibratory sensation, joint position, and pinprick below the knees. Tabular concepts with separate columns for the right and left sides of the body were found only in the EHR notes. Examples of concepts in table form included biceps weakness represented as [biceps strength 3 3] (meaning that biceps strength was 3/5 on both right and left sides) or knee hyperreﬂexia represented as [knee reﬂexes 4+ 4+] (meaning that the knee reﬂex was 4+ on both right and left sides). Text span annotations were stored in an SQLite database and exported in JSONL format for further processing in the spaCy (Explosion, Berlin, Germany) python programming environment. D. Mapping text spans to concepts in the neuro-ontology (normalization) Candidate text spans identiﬁed by the CNN and BERT models were mapped to neurological concepts in the target neuro-ontology. The neuro-ontology (57) is a hierarchical ontology with 1,600 concepts constructed with the Protégé ontology editor (67). All concepts map to terms and CUIs (unique concept identiﬁers) from the UMLS (20). The highest levels of neuro-ontology correspond to the main elements of the neurological examination: mental status, cranial nerves, motor, sensory, reﬂexes, and symptoms. The neuro-ontology is available for download in CSV or OWL format at the National Center for Biomedical Ontologies BioPortal (https:// bioportal.bioontology.org/ontologies/NEO). We manually created a look-up table by mapping 3,500 potential target phrases to concepts in the neuro-ontology. Similarities between the candidate text spans (from either the TABLE I. Performance of CNN and BERT neural networks on concept extraction task. Corpus EHR Textbook OMIM NN F Precision Recall CNN 57.5 65.6 51.2 BERT 61.7 64.0 59.5 CNN 69.0 70.1 67.9 BERT 73.0 73.6 72.3 CNN 76.2 78.8 73.7 BERT 80.4 79.0 82.0 Frontiers in Digital Health 04 frontiersin.org Azizi et al. 10.3389/fdgth.2022.1065581 CNN or BERT models) and target phrases in the lookup table were calculated using the doc.similarity method from spaCy (66). Both the candidate text span and the target phrase were converted to doc objects using the spaCy NLP pipeline (https:// spacy.io/api/doc/#similarity), which converts each token in the phrase into a word vector. The similarity is the cosine distance between the word vectors from the two phrases and ranges between 0.0 (least similar) and 1.0 (most similar). We mapped the candidate text span to its most similar target text span in the look-up table and retrieved the corresponding concept name and UMLS CUI from the neuro-ontology (57). III. Results A. Writing style and accuracy varied by corpus The OMIM corpus used more unigrams and digrams to encode signs and symptoms and had shorter spans of text annotations than the EHR corpus or the Textbook corpus (Figure 2). The length of annotations (histogram insets, Figure 2) was longer for the EHR corpus. Extended annotations were more frequent in the EHR corpus and Textbook corpus. Only the EHR corpus had tabular annotation (clinical concepts expressed in table format). Performance on the concept identiﬁcation task differed by corpus; F, precision, and recall were highest for the OMIM corpus and lowest for the EHR corpus (Table 1). B. Performance of NER model decreased with the increasing text span length For all three corpora, the recognition of clinical concepts as measured by F scores was better for shorter text spans (Figures 3A,B). This applied to both the CNN and the BERT models for concept identiﬁcation (Table 1). F was highest for unigrams (one-word concepts like ataxia, diplopia, aphasia) for all three corpora. In general, performance on bigrams was better than trigrams, and performance on trigrams was better than tetragrams. Performance tended to be worse for text spans greater than four words (extended), or text spans with compound constructions such as weakness of the biceps, triceps, and deltoids. FIGURE 2 Text spans that identiﬁed clinical concepts were longer in the EHR corpus and shortest in the OMIM corpus (see blue inset histograms). Proportionately, the OMIM corpus used the most unigrams and bigrams as compared to the EHR corpus and the Textbook corpus (see red bar charts). IV. Discussion C. Performance varied by neural network model Named entity recognition models based on deep learning can recognize neurologic signs and symptoms in the biomedical literature and electronic health records (Table 1). Previous work has shown that BERT outperforms CNNs on recognizing drugs and diseases in annotated test corpora (52,55). We For all three corpora, BERT outperformed the CNN neural network for the recall of clinical concepts. Precision in clinical concept identiﬁcation was about the same for all three corpora when BERT was compared to the CNN model (Table 1). Frontiers in Digital Health 05 frontiersin.org Azizi et al. 10.3389/fdgth.2022.1065581 FIGURE 3 (A) F values for the BERT NER model sorted by text span label type. The BERT NER model performs similarly on the three corpora for unigrams and bigrams, but F values lag for the EHR corpora for the tetragrams, extended text spans, and compound text spans. Note that the BERT NER model performs signiﬁcantly worse on the EHR corpus for tetragrams, extended text spans, and compound text spans when compared to the Textbook or OMIM corpus. (B) F values for the BERT NER model sorted by corpus. All three corpora show the same pattern with declining F values with the increasing length of the text span. irregular abbreviations, spelling, grammatical, and other writing errors in the clinical notes created by physicians (68– 72) The general approach of the writers of the OMIM corpus was brevity. OMIM writers tended to use lists of clinical concepts such as “the patient had optic disk pallor, miosis, anisocoria, and a relative afferent pupil defect.” The general approach of the writers of the Textbook corpus was didactic and explanatory so that a relative afferent pupil defect might be described as “the swinging ﬂashlight test was abnormal and the pupil dilated when the light was placed over the abnormal pupil and the pupil constricted when the light was moved to the normal pupil.” The EHR corpus was characterized by brevity but irregular spellings, abbreviations, and syntax so that the same patient might be described as “RAPD present on R.” The lower accuracy for recognition of signs and symptoms in the EHR corpus (physician notes) deserves further comment. One way to improve automated recognition of signs and symptoms in physician notes is to encourage them to use structured rather than unstructured documentation (73). However, given physician burnout associated with clinical documentation (74), and physician distaste for structured documentation (75), it seems unlikely that physicians will adopt structured documentation for recording signs and symptoms. Furthermore, given that by training, physicians are often asked to describe ﬁndings rather than name ﬁndings, it seems unlikely that physicians can be converted to using short names instead of lengthy descriptions of signs and symptoms. Rather, improvements in NLP are needed to identify better extend these observations to demonstrate the superiority of BERT over CNNs for recognizing neurological signs and symptoms in electronic health records and biomedical literature. A signiﬁcant ﬁnding was that the accuracy of recognition of signs and symptoms fell with increasing text span length (Figures 3A,B). Increased variability in longer text spans likely poses greater difﬁculty for NER pipelines, regardless of whether they are based on linguistic/symbolic methods like MetaMap or deep learning like BERT or CNNs. Longer text spans are more likely to be descriptions of named entities (e.g., “the patient fell to the left when standing with eyes closed” rather than more concise named entities themselves (e.g., “Romberg sign positive”). Normalization of longer text spans (mapping to suitable concepts in the ontology) may pose additional challenges. The successful mapping (normalization) of “wavering with eyes closed” to “Romberg sign positive” may require vectorization (word embedding) of terms in an ontology, as well as the synonyms and deﬁnitions of these terms (48,55). Another signiﬁcant observation was that recall of neurologic signs and symptoms was lower in the EHR corpus than in the OMIM corpus or Textbook corpus. The Textbook and the OMIM corpus were written by professional writers and had undergone careful editing and correction. The EHR corpus was written by physicians who were not professional writers. The EHR corpus was marred by irregular spelling, irregular abbreviations, typographical errors, grammatical errors, and other irregularities absent from the OMIM corpus and the Textbook corpus. Others have noted the high frequency of Frontiers in Digital Health 06 frontiersin.org Azizi et al. 10.3389/fdgth.2022.1065581 clinical concepts held as lengthier texts spans or represented as descriptions of named entities rather than as the named entity itself. NLP models that extract clinical concepts from free text must recognize negation successfully. The sentence “the patient has ataxia” has a clinical concept whereas the sentence “ataxia is absent” denies ataxia (76–78). Negation makes it difﬁcult to determine if a sign or symptom is present and suggests that strategies based on regular expressions (REGEX) will fail. The patient who complains of tremor, who is tremulous, or is observed to have a tremor must be distinguished from the patient who denies tremor, is not tremulous, or has no tremor. MetaMap uses the NEGEX algorithm to recognize negation (27). We relied on examples to train the neural networks to recognize negated concepts for our BERT and CNN models. Further work is needed on handling negated concepts accurately and efﬁciently (77). Another challenge is word disambiguation (79). The sentence the “patient has had a fall” may contain a valid neurological concept, whereas the sentence “the patient was seen in the Fall” does not. Word disambiguation is another area of continuing research in NLP (79). This study has several limitations. The study was limited to the domain of neurology (neurological signs and symptoms). Furthermore, the text span annotations were done by a single annotator. We have planned an inter-rater agreement study (65). We limited the target ontology to 1,600 neurological concepts. Whether our methods can be generalized to more complex domains and larger ontologies is uncertain. Although we achieved a recall of 80% to 90% with shorter text span lengths, the recall was lower for longer text span lengths. To make automated high throughput neuro-phenotyping practical, we estimate that a recall of at least 90% is needed depending on the application (i.e., research versus patient care). Identifying clinical concepts in complex grammatical structures remains challenging for even the best NLP algorithms. For example, identifying the concepts biceps weakness, triceps weakness, and hand weakness in the sentence the patient had 3+/5 strength in the biceps, 2+/5 strength in the triceps, and 1/5 hand grip strength remains problematic. Efﬁcient NLP algorithms that simplify grammar and syntax are an area of evolving research (80,81). Another limitation of the study is the small corpus used for training. Our NER models would likely have improved with more training annotations. In conclusion, given the burden of physician documentation (74), patient signs and symptoms will likely continue in electronic health records as unstructured text. The automated identiﬁcation of these signs and symptoms is critical to the success of deep phenotyping, and precision medicine initiatives (5,6). Advances in NLP based on word embedding and deep learning make the automated identiﬁcation of signs and symptoms in unstructured text increasingly feasible. Frontiers in Digital Health Data availability statement The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author/s. Ethics statement The studies involving human participants were reviewed and approved by Institutional Review Board of the University of Illinois at Chicago. The patients/participants provided their written informed consent to participate in this study. Author’s contributions Concept and design by SA and DBH. Model parameters and computations by SA. Data interpretation, drafting, revising, and ﬁnal approval by SA, DBH, and DCW II.All authors contributed to the article and approved the submitted version. Funding The research was partially sponsored by the Mary K. Finley Missouri Endowment, the Missouri S&T Intelligent Systems Center, the National Science Foundation, and the Leonard Wood Institute in cooperation with the U.S. Army Research Laboratory. It was accomplished under Cooperative Agreement Number W911NF-14-2-0034. The views, opinions, ﬁndings, recommendations, or conclusions contained in this document are those of the authors. They should not be interpreted as representing the views or ofﬁcial policies expressed or implied by the Leonard Wood Institute, the Army Research Laboratory, the National Science Foundation, or the U.S. Government. The U.S. Government is authorized to reproduce and distribute reprints for Government purposes notwithstanding any copyright notation hereon. Acknowledgments The authors thank Dr. Michael D. Carrithers for providing the physician EHR notes. Conﬂict of interest The authors declare that the research was conducted in the absence of any commercial or ﬁnancial relationships that could be construed as a potential conﬂict of interest. 07 frontiersin.org Azizi et al. 10.3389/fdgth.2022.1065581 afﬁliated organizations, or those of the publisher, the editors Publisher’s note and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not All claims expressed in this article are solely those of the authors and do not necessarily represent those of their guaranteed or endorsed by the publisher. References 1. Krauthammer M, Nenadic G. Term identiﬁcation in the biomedical literature. J Biomed Inform. (2004) 37:512–26. doi: 10.1016/j.jbi.2004.08.004 21. Eltyeb S, Salim N. Chemical named entities recognition: a review on approaches, applications. J Cheminform. (2014) 6:1–12. doi: 10.1186/1758-2946-6-17 2. Ofﬁce of the National Coordinator for Health Information Technology. Adoption of electronic health records by hospital service type 2019–2021, Health IT Quick Stat #60 (2022). Available from: https://www.healthit.gov/data/ quickstats/adoption-electronic-health-records-hospital-service-type-2019-2021. 22. Quimbaya AP, Múnera AS, Rivera RAG, Rodríguez JCD, Velandia OMM, Peña AAG, et al. Named entity recognition over electronic health records through a combined dictionary-based approach. Procedia Comput Sci. (2016) 100:55–61. doi: 10.1016/j.procs.2016.09.123 3. Banda JM, Seneviratne M, Hernandez-Boussard T, Shah NH. Advances in electronic phenotyping: from rule-based deﬁnitions to machine learning models. Annu Rev Biomed Data Sci. (2018) 1:53. doi: 10.1146/annurev-biodatasci080917-013315 23. Hirschman L, Morgan AA, Yeh AS. Rutabaga by any other name: extracting biological names. J Biomed Inform. (2002) 35:247–59. doi: 10.1016/S1532-0464 (03)00014-5 24. Uzuner Ö, South BR, Shen S, DuVall SL. 2010 i2b2/VA challenge on concepts, assertions,, relations in clinical text. J Am Med Inform Assoc. (2011) 18:552–6. doi: 10.1136/amiajnl-2011-000203 4. Tenopir C, Grayson M, Zhang Y, Ebuen M, King DW, Boyce PB. Patterns of journal use by scientists through three evolutionary phases. D-Lib (2003) 9:1–15. doi: 10.1045/may2003-king 25. Funk C, Baumgartner W, Garcia B, Roeder C, Bada M, Cohen KB, et al. Large-scale biomedical concept recognition: an evaluation of current automatic annotators and their parameters. BMC Bioinf. (2014) 15:1–29. doi: 10.1186/ 1471-2105-15-59 5. Collins FS, Varmus H. A new initiative on precision medicine. N Engl J Med. (2015) 372:793–5. doi: 10.1056/NEJMp1500523 6. Robinson PN. Deep phenotyping for precision medicine. Hum Mutat. (2012) 33:777–80. doi: 10.1002/humu.22080 26. Shah NH, Bhatia N, Jonquet C, Rubin D, Chiang AP, Musen MA. Comparison of concept recognizers for building the open biomedical annotator. BMC Bioinf. (2009) 10:1–9. doi: 10.1186/1471-2105-10-S2-S1 7. Fu S, Chen D, He H, Liu S, Moon S, Peterson KJ, et al. Clinical concept extraction: a methodology review. J Biomed Inform. (2020) 109:103526. doi: 10. 1016/j.jbi.2020.103526 27. Aronson AR, Lang FM. An overview of MetaMap: historical perspective and recent advances. J Am Med Inform Assoc. (2010) 17:229–36. doi: 10.1136/jamia. 2009.002733 8. Esteva A, Robicquet A, Ramsundar B, Kuleshov V, DePristo M, Chou K, et al. A guide to deep learning in healthcare. Nat Med. (2019) 25:24–9. doi: 10.1038/ s41591-018-0316-z 28. Lindberg DA, Humphreys BL, McCray AT. The uniﬁed medical language system. Yearb Med Inform. (1993) 2:41–51. doi: 10.1055/s-0038-1637976 9. Chowdhary K. Natural language processing. Fundam Artif Intell. (2020):603–49. 29. Bodenreider O. The uniﬁed medical language system (UMLS): integrating biomedical terminology. Nucleic Acids Res. (2004) 32:D267–70. doi: 10.1093/ nar/gkh061 10. Hirschberg J, Manning CD. Advances in natural language processing. Science. (2015) 349:261–6. doi: 10.1126/science.aaa8685 30. Hier DB, Yelugam R, Azizi S, Carrithers MD, Wunsch II DC. High throughput neurological phenotyping with MetaMap. Eur Sci J. (2022) 18:37–49. doi: 10.19044/esj.2022.v18n4p37 11. Islam MA, Anik MSH, Islam ABMAA. Towards achieving a delicate blending between rule-based translator, neural machine translator. Neural Comput Appl. (2021) 33:12141–67. doi: 10.1007/s00521-021-05895-x 31. Wang Y, Wang L, Rastegar-Mojarad M, Moon S, Shen F, Afzal N, et al. Clinical information extraction applications: a literature review. J Biomed Inform. (2018) 77:34–49. doi: 10.1016/j.jbi.2017.11.011 12. Islam MA, Mukta MSH, Olivier P, Rahman MM. Comprehensive guidelines for emotion annotation. Proceedings of the 22nd ACM International Conference on Intelligent Virtual Agents, 2022 Sep. New York, NY, USA: Association for Computing Machinery (2022). p. 1–8. 32. Sevenster M, Van Ommering R, Qian Y. Automatically correlating clinical ﬁndings and body locations in radiology reports using MedLEE. J Digit Imaging. (2012) 25:240–9. doi: 10.1007/s10278-011-9411-0 13. Mohammad S. A practical guide to sentiment annotation: challenges, solutions. Proceedings of the 7th Workshop on Computational Approaches to Subjectivity, Sentiment, Social MEDIA Analysis, 2016 Jun. San Diego, California: Association for Computational Linguistics (2016). p. 174–9. 33. Savova GK, Masanz JJ, Ogren PV, Zheng J, Sohn S, Kipper-Schuler KC, et al. Mayo clinical text analysis and knowledge extraction system (cTAKES): architecture, component evaluation and applications. J Am Med Inform Assoc. (2010) 17:507–13. doi: 10.1136/jamia.2009.001560 14. Hasan HM, Islam MA, Hasan MT, Hasan MA, Rumman SI, Shakib MN. A spell-checker integrated machine learning based solution for speech to text conversion. 2020 Third International Conference on Smart Systems and Inventive Technology (ICSSIT) (2020). p. 1124–30. 34. Friedman C, Shagina L, Lussier Y, Hripcsak G. Automated encoding of clinical documents based on natural language processing. J Am Med Inform Assoc. (2004) 11:392–402. doi: 10.1197/jamia.M1552 15. Gonzalez-Hernandez G, Sarker A, O’Connor K, Savova G. Capturing the patient’s perspective: a review of advances in natural language processing of health-related text. Yearb Med Inform. (2017) 26:214–27. doi: 10.15265/IY2017-029 35. Friedman C, Shagina L, Socratous SA, Zeng X. A web-based version of MedLEE: a medical language extraction and encoding system. Proceedings of the AMIA Annual Fall Symposium. American Medical Informatics Association (1996). p. 938. 16. Bird S, Klein E, Loper E. Natural language processing with python. Sebastopol, CA: O’Reilly Media (2009). Available from: https://www.nltk.org/ book/. 36. Friedman C. A broad-coverage natural language processing system. Proceedings of the AMIA Symposium. American Medical Informatics Association. (2000). p. 270–4. 17. SNOMED CT. NCBO BioPortal (2022). Available from: https://bioportal. bioontology.org/ontologies/SNOMEDCT/ (Accessed October 5, 2022). 37. Huang Z, Xu W, Yu K. Bidirectional LSTM-CRF models for sequence tagging [Preprint] (2015). Available at: http://arxiv.org/1508.01991. 18. Online Mendelian Inheritance in Man. NCBO BioPortal (2022). Available from: https://bioportal.bioontology.org/ontologies/OMIM (Accessed October 5, 2022). 38. Lample G, Ballesteros M, Subramanian S, Kawakami K, Dyer C. Neural architectures for named entity recognition [Preprint] (2016). Available at: http://arxiv.org/1603.01360. 19. Human Phenotype Ontology. NCBO BioPortal (2022). Available from: https://bioportal.bioontology.org/ontologies/HP (Accessed October 5, 2022). 39. Chiu JP, Nichols E. Named entity recognition with bidirectional LSTM-CNNs. Trans Assoc Comput Linguist. (2016) 4:357–70. doi: 10.1162/tacl_ a_00104 20. UMLS Metathesaurus Browser. National Library of Medicine (2022). Available from: https://uts.nlm.nih.gov/uts/umls/home (Accessed October 5, 2021). Frontiers in Digital Health 08 frontiersin.org Azizi et al. 10.3389/fdgth.2022.1065581 61. Kung D, Nguyen T. Absolute case-based neurology review. Oxford UK: Springer (2019). 40. Peters ME, Ammar W, Bhagavatula C, Power R. Semi-supervised sequence tagging with bidirectional language models [Preprint] (2017). Available at: http:// arxiv.org/1705.00108. 62. Macleod M, Pal S, Simpson M. Neurology clinical cases uncovered. San Francisco CA: Wiley-Blackwell (2011). 41. Vani A, Jernite Y, Sontag D. Grounded recurrent neural networks [Preprint] (2017). Available at: http://arxiv.org/1705.08557. 63. Neves M, Ševa J. An extensive review of tools for manual annotation of documents. Brief Bioinformatics. (2021) 22:146–63. doi: 10.1093/bib/bbz130 42. Liu Z, Tang B, Wang X, Chen Q. De-identiﬁcation of clinical notes via recurrent neural network and conditional random ﬁeld. J Biomed Inform. (2017) 75:S34–S42. doi: 10.1016/j.jbi.2017.05.023 64. Montani I, Honnibal M. Prodigy: a new annotation tool for radically efﬁcient machine teaching. Artif Intell. (2018). Available from: https://explosion.ai/blog/ prodigy-annotation-tool-active-learning. 43. Habibi M, Weber L, Neves M, Wiegandt DL, Leser U. Deep learning with word embeddings improves biomedical named entity recognition. Bioinformatics. (2017) 33:i37–i48. doi: 10.1093/bioinformatics/btx228 65. Oommen C, Howlett-Prieto Q, Carrithers MD, Hier DB. Inter-Rater Agreement for the Annotation of Neurologic Concepts in Electronic Health Records. medRxiv (2022). Available from: http://doi.org/10.1101/2022.11.16. 22282384. 44. Dehghan A, Kovacevic A, Karystianis G, Keane JA, Nenadic G. Combining knowledge-and data-driven methods for de-identiﬁcation of clinical narratives. J Biomed Inform. (2015) 58:S53–9. doi: 10.1016/j.jbi.2015.06.029 66. Vasiliev Y. Natural language processing with Python and Spacy. San Francisco CA: No Starch Press (2020). 45. Hochreiter S, Schmidhuber J. Long short-term memory. Neural Comput. (1997) 9:1735–80. doi: 10.1162/neco.1997.9.8.1735 67. Noy NF, McGuinness DL. Ontology development 101: a guide to creating your ﬁrst ontology. Stanford Knowledge Systems Laboratory Technical Report KSL-01-05 (2001). 46. Lafferty J, McCallum A, Pereira FC. Conditional random ﬁelds: probabilistic models for segmenting and labeling sequence data. Proceedings of the 18th International Conference on Machine Learning 2001. San Francisco, CA, USA: Morgan Kaufmann Publishers Inc. (2001). 68. Assale M, Dui LG, Cina A, Seveso A, Cabitza F. The revival of the notes ﬁeld: leveraging the unstructured content in electronic health records. Front Med. (2019) 6:66. doi: 10.3389/fmed.2019.00066 47. Gehrmann S, Dernoncourt F, Li Y, Carlson ET, Wu JT, Welt J, et al. Comparing deep learning and concept extraction based methods for patient phenotyping from clinical narratives. PLoS ONE. (2018) 13:e0192360. doi: 10. 1371/journal.pone.0192360 69. Shilo G, Shilo L. Writing style of young physicians in the computer and internet era. Int J Med Educ. (2014) 5:82. doi: 10.5116/ijme.534a.a3e2 70. Pagano MP, Mair D. Writing medical records. J Tech Writ Commun. (1986) 16:331–41. doi: 10.2190/WY9T-634E-V2JT-JDVQ 48. Arbabi A, Adams DR, Fidler S, Brudno M, et al. Identifying clinical terms in medical text using ontology-guided machine learning. JMIR Med Inform. (2019) 7: e12596. doi: 10.2196/12596 71. Zisowitz ML. Teaching medical students and physicians to write. Acad Med. (1964) 39:481–4. 49. Groza T, Köhler S, Doelken S, Collier N, Oellrich A, Smedley D, et al. Automatic concept recognition using the human phenotype ontology reference and test suite corpora. Database. (2015) 2015:1–13. doi: 10.1093/database/bav005 72. Hamiel U, Hecht I, Nemet A, Pe’er L, Man V, Hilely A, et al. Frequency, comprehension and attitudes of physicians towards abbreviations in the medical record. Postgrad Med J. (2018) 94:254–8. doi: 10.1136/postgradmedj2017-135515 50. Vaswani A, Shazeer N, Parmar N, Uszkoreit J, Jones L, Gomez AN, et al. Attention is all you need. Adv Neural Inf Process Syst. (2017) 30:5998–6008. 73. Rosenbloom ST, Denny JC, Xu H, Lorenzi N, Stead WW, Johnson KB. Data from clinical notes: a perspective on the tension between structure and ﬂexible documentation. J Am Med Inform Assoc. (2011) 18:181–6. doi: 10.1136/jamia. 2010.007237 51. Devlin J, Chang MW, Lee K, Toutanova K. BERT: pre-training of deep bidirectional transformers for language understanding [Preprint] (2018). Available at: http://arxiv.org/1810.04805. 52. Zhu R, Tu X, Huang JX. Utilizing BERT for biomedical, clinical text mining. Data Analytics in Biomedical Engineering, Healthcare. Elsevier (2021). p. 73–103. Available from: https://doi.org/10.1016/B978-0-12-819314-3.00005-7 74. Thomas Craig KJ, Willis VC, Gruen D, Rhee K, Jackson GP. The burden of the digital environment: a systematic review on organization-directed workplace interventions to mitigate physician burnout. J Am Med Inform Assoc. (2021) 28:985–97. doi: 10.1093/jamia/ocaa301 53. Yu X, Hu W, Lu S, Sun X, Yuan Z. Biobert based named entity recognition in electronic medical record. 2019 10th international conference on information technology in medicine and education (ITME). New York NY: IEEE (2019). p. 49–52. 75. Han H, Lopp L. Writing and reading in the electronic health record: an entirely new world. Med Educ Online. (2013) 18:18634. doi: 10.3402/meo.v18i0. 18634 54. Lee J, Yoon W, Kim S, Kim D, Kim S, So CH, et al. Biobert: a pre-trained biomedical language representation model for biomedical text mining. Bioinformatics. (2020) 36:1234–40. 76. Shivade C, de Marneffe MC, Fosler-Lussier E, Lai AM. Extending negex with kernel methods for negation detection in clinical text. Proceedings of the Second Workshop on Extra-Propositional Aspects of Meaning in Computational Semantics (ExProM 2015) (2015). p. 41–46. 55. Ji Z, Wei Q, Xu H. Bert-based ranking for biomedical entity normalization. AMIA Summits Transl Sci Proc. (2020) 2020:269. 77. Wu S, Miller T, Masanz J, Coarr M, Halgrim S, Carrell D, et al. Negation’s not solved: generalizability versus optimizability in clinical natural language processing. PLoS ONE. (2014) 9:e112774. doi: 10.1371/journal.pone.0112774 56. Weng C, Shah NH, Hripcsak G. Deep phenotyping: embracing complexity and temporality-towards scalability, portability, and interoperability. J Biomed Inform. (2020) 105:103433. doi: 10.1016/j.jbi.2020.103433 57. Hier DB, Brint SU. A neuro-ontology for the neurological examination. BMC Med Inform Decis Mak. (2020) 20:1–9. doi: 10.1186/s12911-020-1066-7 78. Elkin PL, Brown SH, Bauer BA, Husser CS, Carruth W, Bergstrom LR, et al. A controlled trial of automated classiﬁcation of negation from clinical notes. BMC Med Inform Decis Mak. (2005) 5:1–7. doi: 10.1186/1472-6947-5-13 58. Gondolo T. Neurology study guide: oral board examination review. Cham Switzerland: Springer Nature (2005). 79. Navigli R. Word sense disambiguation: a survey. ACM Comput Surv. (2009) 41:1–69. doi: 10.1145/1459352.1459355 59. Ubogu EE. Neurology oral boards review. New York NY: Humana Press (2005). 80. Shardlow M. A survey of automated text simpliﬁcation. Int J Adv Comput Sci Appl. (2014) 4:58–70. 60. Alpert JN. The neurologic diagnosis: a practical bedside approach. Cham Switzerland: Springer (2018). 81. Al-Thanyyan SS, Azmi AM. Automated text simpliﬁcation: a survey. ACM Comput Surv. (2021) 54:1–36. doi: 10.1145/3442695 Frontiers in Digital Health 09 frontiersin.org
https://openalex.org/W21511935	https://www.scienceopen.com/document_file/bd865aca-1f22-4447-a630-6eead7f899cc/ScienceOpen/020_Terrenghi.pdf	English	null	Evolving the Role and Agenda of HCI for Sustainable Development	Electronic workshops in computing	2,009	cc-by	2,436	1. INTRODUCTION Lately, sustainability has become a popular term within the Human-Computer Interaction community. In 2007, Blevis won the best paper award at CHI [1], presenting a claim for Sustainable Interaction Design (SID), and several interest groups have emerged in this space. In this context, sustainability has mostly been associated with the environmental implications of design. At the same time, ICT for development has become a relevant topic in the HCI field, tackling several aspects concerned with sustainability [2]: In this context, sustainability has rather been associated with the design implications on cultural and financial development. So what do we intend with sustainability after all, and how does it relate to HCI and interaction design? In this paper we try to tackle this question, and to consider the relationship and potential differences and convergences between HCI agendas in the so-called ‘developed’ and ‘developing’ economies. We claim that HCI has become an essential component of designing for sustainability and development. Any project embarking on a sustainable design agenda should involve a person well versed both in the theory and practice of HCI. To this end, we argue that HCI needs to re-consider some of its assumptions and borrow from some of the principles of sustainable design. From this point of view, we reflect on the differences between local and global perspectives of HCI, and how developed and developing economies can benefit from each-other in leveraging the HCI knowledge. Evolving the Role and Agenda of HCI for Sustainable Development Lucia Terrenghi Lucia Terrenghi Vodafone Group Services R&D, 81549 Munich, Germany lucia.terrenghi@vodafone.com Vodafone Group Services R&D, 81549 Munich, Germany lucia.terrenghi@vodafone.com Sriram Subramanian Department of Computer Science, University of Bristol, UK sriram@cs.bris.ac.uk Department of Computer Science, University of Bristol, UK sriram@cs.bris.ac.uk Department of Computer Science, University of Bristol, UK sriram@cs.bris.ac.uk The paper examines the evolving role of HCI for sustainable development. We claim that HCI is central to the idea of designing for development and sustainability. We therefore propose the idea that researchers in HCI have to strive for a greater balance between attempting to produce global and local knowledge. HCI designs for local knowledge can create case-studies that can be drive and be driven by theories and models that could lead to global knowledge. 2.3 HCI for Access to Digital Economy The reason why HCI has become an essential aspect of sustainable development is quite simple: The access to digital information and communication has become an indispensable resource for survival in a global economy and society. Still, whereas this is a disposal resource in some geographies or populations, similar to water, it is a scarce resource in many others. The reasons are not only related to the infrastructure. Because of the regions where computing technology has grown, interfaces have been designed with the values, mental models, stereotypes of the western industrialized economies in mind. Those often prevent the access to other types of users, who possibly have literacy, linguistic, physical barriers, or simply have different cultures and mindsets. As technologies reach new regions and populations, those assumptions need to be challenged to guarantee inclusiveness and democratization of access. How does this impact the HCI agenda? Despite the fact that there is an everlasting conflict between fundamental and applied research, in this paper we argue that HCI for sustainability cannot be agnostic of its impact. 1.1 Defining Sustainability g y In the 1987 United Nations Commission report on Environment and Development (Bruntland Commission) entitled “Our Common Future”, sustainable is defined as “…development that meets the needs of the present without compromising the ability of future generations to meet their needs.” There are different ‘lenses’ and frameworks through which one can look at sustainability: We like to consider the synthesis suggested by Shedroff [5], which considers human, natural and financial capitals. Shedroff presents several existing frameworks (e.g., cradle to cradle [7], sustainable return on investment, Natural Capitalism [6]), which are focusing on some of those three capitals in particular, in the attempt to assess or measure sustainability. In this paper our main focus is on sustainability of social capital, being aware that the environmental and financial ones are connected to that, in respect of the holistic complexity of sustainability. 20 2.2 Technological Reasons: The Diffusion of Computing Technology in Everyday Life 2.2 Technological Reasons: The Diffusion of Computing Technology in Everyday Life The inclusion of a more ‘human centric’ perspective has been co-responsible of the evolution of interaction techniques: Such interaction techniques, building on technological evolutions (accelerometers in small devices, capacitive sensing, and camera vision for example) have started to become more common-place, and utilize a broader bandwidth of human input capabilities (e.g., 3D gestures, multi-touch). In other words, interaction techniques have evolved not only through technology driving innovation but also because there is more concern on human capabilities and the related business opportunities. In the industrialized economies, computing in the form of desktop computers and laptops has become a central part of everyday life, thus pushing the HCI agenda towards the design of interfaces that are accessible to a broader range of users in a broader set of contexts. At the same time, emerging economies such as India are communication and technology enriched through televisions, radio-sets and mobile technology (see [10] for example). Thus interaction with digital technology is radically different in industrialized and emerging economies. This fact poses new challenges for HCI and its implications for sustainability. 2. THE EVOLVING RELATIONSHIP BETWEEN HCI AND SUSTAINABILITY While sustainability has actually been a design topic for a long time already (e.g., [4]), the relationship between HCI and sustainability is more recent, and somewhat still in evolution. We speculate that there are some main reasons for this: 2.1 Historic Reasons: The Inclusion of Design in the HCI Field of Research g HCI has its origins in the ‘80s in human factors, and was first motivated by enhancing interaction in the military domain, where technology was first implemented. But it has evolved as an interdisciplinary field of research over the last 30 years. Its interdisciplinary nature has influenced the creation and evolution of user centred design methods; furthermore the inclusion of human sciences in the field has triggered an agenda which considers human values and social implications, i.e. its impact on social capital. In the transition from human factors and cognitive psychology to a more interdisciplinary area of research, the inclusion of design disciplines has raised differences in opinion about practice vs. research, and utility vs. knowledge. Nevertheless, design has brought to evidence the value implication of HCI, and ethical questions about technology design and implementation. We claim that HCI for sustainability cannot be agnostic of the values it promotes, and of its impact on social, as well as on environmental and financial capital. 3. IMPACT VS SCIENTIFIC CONTRIBUTION: AN ONGOING DILEMMA The scientific contribution of work in the field of HCI is often questioned when it derives from case studies: What are the lessons that the specific case study teaches, and that can be generalized so as to inform other cases? The researcher’s approach to case studies is often one of problem- solving with the goal of having a beneficial impact on a specific problem at hand, and in a specific context (sees Fig. 1). For example, the design of a mobile application for community digital library access in an Indian village [10]. On the other hand, empirical studies aim at answering more fundamental questions, which can be more easily generalized to form predictive theories such as Keystroke-Level Model and Fitts’ Law [9]. It is often neglected, still, that even those empirical studies and measures - that have informed the HCI body of knowledge – are mostly based on mental models and measure parameters that are conventionally agreed upon in some cultures, but 21 can be different in others. This often means those values and mental models are mostly based on western cultures. As computing technology is reaching new regions and cultures, the HCI community needs to become more inclusive: Case studies from cultures and contexts that so far have been marginalized from digital technology can inform the HCI body of knowledge by revealing issues about accessibility and methodology that are sometimes taken for granted. At the same time, the existing HCI body of knowledge can inform the approach to case studies in terms of theory and methodology. To achieve such a balance, it is our taking that the HCI community needs to embark into intercultural collaborations. Addressing cultural sustainability is, in our eyes, an indispensible condition for any design striving for sustainability (and all the aspects thereof). Figure 1: HCI for sustainability must consider both its contextual, local impact, and its universal, global contribution: Thus, case studies and theories of HCI must iteratively inform each-other. HCI for Sustainability Impact Utility Knowledge Local Global Case studies Theory and methods Focus on Acts Produces Focus on Acts Produces Contribution HCI for Sustainability Contribution Case studies Figure 1: HCI for sustainability must consider both its contextual, local impact, and its universal, global contribution: Thus, case studies and theories of HCI must iteratively inform each-other. 4. SUSTAINABLE DEVELOPMENT THROUGH COLLABORATION REFERENCES [1] Blevis, E. 2007. Sustainable interaction design: invention & disposal, renewal & reuse. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (San Jose, California, USA, April 28 - May 03, 2007). CHI '07. ACM, New York, NY, 503-512. p y ) [2] Dearden, A., Light, A., Dray, S., Thomas, J., Best, M., Buckhalter, C., Greenblatt, D., Krishnan, G., and Sambasivan, N. 2007. User centered design and international development. In CHI '07 Extended Abstracts on Human Factors in Computing Systems (San Jose, CA, USA, April 28 - May 03, 2007). CHI '07. ACM, New York, NY, 2825-2828. [2] Dearden, A., Light, A., Dray, S., Thomas, J., Best, M., Buckhalter, C., Greenblatt, D., Krishnan, G., and Sambasivan, N. 2007. User centered design and international development. In CHI '07 Extended Abstracts on Human Factors in Computing Systems (San Jose, CA, USA, April 28 - May 03, 2007). CHI '07. ACM, New York, NY, 2825-2828. y ) [3] Network on Interactive Technologies for the End Users. Official website http://www.ukinit.org [4] Papanek, Victor (1971). Design for the Real World: Human Ecology and Social Change, New York, Pantheon Books . [3] Network on Interactive Technologies for the End Users. Official website http://www.ukinit.org [4] Papanek, Victor (1971). Design for the Real World: Human Ecology and Social Change, New [3] Network on Interactive Technologies for the End Users. Official website http://www.ukinit.org [4] Papanek, Victor (1971). Design for the Real World: Human Ecology and Social Change, New York, Pantheon Books . 5] Shedroff, N. 2009. Design is the Problem: The Future of Design Must be Sustainable. Rosenfield Media Ed. [6] Stuart L. Hart, (2005) Capitalism at the Crossroads: The Unlimited Business Opportunities in Solving the World's Most Difficult Problems. Wharton School Publishing. [6] Stuart L. Hart, (2005) Capitalism at the Crossroads: The Unlimited Business Opportunities in Solving the World's Most Difficult Problems. Wharton School Publishing. g g [7] William McDonough & Michael Braungart (2003) Cradle to Cradle: Remaking the Way We Make Things, Rodale Press. g g [7] William McDonough & Michael Braungart (2003) Cradle to Cradle: Remaking the Way We Make Things, Rodale Press. [8] Muller, M. J. and Wharton, C. 1997. Workshop on HCI research and practice agenda based on human needs and social responsibility. In CHI '97 Extended Abstracts on Human Factors in Computing Systems: Looking To the Future (Atlanta, Georgia, March 22 - 27, 1997). CHI '97. 4. SUSTAINABLE DEVELOPMENT THROUGH COLLABORATION ad hoc solutions ad hoc solutions ad hoc solutions Figure 2: HCI is challenged to find a balance on the continuum between the maintenance of cultural identity (i.e. maintenance of cultural diversity, left pole in the schema) and the leveraging of standard solutions (i.e., definition of conventions for scalability and interoperability, the right pole in the schema). 4. SUSTAINABLE DEVELOPMENT THROUGH COLLABORATION The inclusion of a growing and diverse population in the digital economy will challenge the HCI community to find a balance on the continuum between the maintenance of cultural identity (i.e., maintenance of cultural diversity) and the leveraging of standard solutions (i.e., definition of conventions for scalability and interoperability). To this end, we believe it is important that the HCI discipline is taught and an HCI culture is developed across the world with a focus on inclusivity, ethics and sustainability. HCI, indeed, can become a catalyst for sustainable development if it succeeds in informing the design of solutions that facilitate access to the digital economy. In order to create universal knowledge and specific designs that can maintain the balance between diversity and intercommunication, and create value in context, it is important that HCI scholars and practitioners from different cultures assume a design with, rather than a design for attitude, engaging in cross-cultural collaborations. 22 An example for this is the Network on Interactive Technologies for the End Users, which we have initiated in early 2009 and involves UK and Indian universities as well as multi-national industries [[3]]. Our experience so far is that creating opportunities for cross-cultural and interdisciplinary relationships can stimulate synergies both within and across geographies. From this sort of collaboration we expect to leverage the HCI knowledge in several ways. For example, by stimulating novel methods and approaches for co-design with users that are fundamentally different from designers (e.g., neophytes, children, elderly, physically challenged, and illiterate users); by designing interfaces that are adaptable to users’ interaction capabilities and context; and by designing expressive, multimodal interfaces (e.g., using gestures, touch, speech) that can enhance the communication between people and people mediated by the system. As a result, we hope to learn how to design for a more inclusive access to digital communication. HCI cutural differences standard of quality life local context ad hoc solutions scalable solutions HCI cutural differences standard of quality life local context ad hoc solutions HCI cutural differences standard of quality life local context ad hoc solutions global communication Figure 2: HCI is challenged to find a balance on the continuum between the maintenance of cultural identity (i.e. maintenance of cultural diversity, left pole in the schema) and the leveraging of standard solutions (i.e., definition of conventions for scalability and interoperability, the right pole in the schema). REFERENCES ACM, New York, NY, 234-234. [9] John M. Carroll (Editor), (2003) HCI Models, Theories, and Frameworks: Toward a Multidisciplinary Science, Chapter 3. Morgan Kaufmann Press. [10] Jones, M., Harwood, W., Buchanan, G., and Lalmas, M. 2007. “StoryBank: an indian village community digital library”. In Proceedings of the 2007 Conference on Digital Libraries (Vancouver, BC, Canada, June 18 - 23, 2007). JCDL '07. ACM, New York, NY, 257-258. 23
https://openalex.org/W2292585069	https://link.springer.com/content/pdf/10.1140%2Fepjc%2Fs10052-016-4059-z.pdf	English	null	Exact solutions, energy, and charge of stable Q-balls	European physical journal. C, Particles and fields	2,016	cc-by	9,004	a e-mail: bazeia@ﬁsica.ufpb.br Exact solutions, energy, and charge of stable Q-balls D. Bazeia1,a, M. A. Marques1, R. Menezes2,3 D. Bazeia1,a, M. A. Marques1, R. Menezes2,3 1 Departamento de Física, Universidade Federal da Paraíba, João Pessoa, PB 58051-970, Brazil 2 Departamento de Ciências Exatas, Universidade Federal da Paraíba, Rio Tinto, PB 58297-000, Brazil 3 Departamento de Física, Universidade Federal de Campina Grande, Campina Grande, PB 58109-970, Brazil D. Bazeia1,a, M. A. Marques1, R. Menezes2,3 1 Departamento de Física, Universidade Federal da Paraíba, João Pessoa, PB 58051-970, Brazil 2 Departamento de Ciências Exatas, Universidade Federal da Paraíba, Rio Tinto, PB 58297-000, Brazil 3 Departamento de Física, Universidade Federal de Campina Grande, Campina Grande, PB 58109-970, Brazil Received: 4 February 2016 / Accepted: 3 April 2016 / Published online: 29 April 2016 © The Author(s) 2016. This article is published with open access at Springerlink.com Abstract In this work we deal with nontopological solu- tions of the Q-ball type in two spacetime dimensions. We study models of current interest, described by a Higgs-like and other, similar potentials which unveil the presence of exact solutions. We use the analytic results to investigate how to control the energy and charge to make the Q-balls stable. baryon asymmetry of the universe appears via the Afﬂeck– Dine mechanism [24], as a feature of the ﬂat direction inﬂa- tion, with the ﬂat direction condensate giving rise to Q-balls [12,13,16,17], which can later decay under reheating [25]. Numerical simulations were used to study stability under small ﬂuctuations, interactions, and the scattering of Q-balls. We have searched the literature and found some exact, analyt- ical solutions for Q-balls for Higgs-like potentials and other similar potentials, and this motivated us to investigate exact solutions for Q-balls, with focus on the study of properties such as shape, energy, charge, stability, and splitting, without relying on numerical solutions. baryon asymmetry of the universe appears via the Afﬂeck– Dine mechanism [24], as a feature of the ﬂat direction inﬂa- tion, with the ﬂat direction condensate giving rise to Q-balls [12,13,16,17], which can later decay under reheating [25]. Numerical simulations were used to study stability under small ﬂuctuations, interactions, and the scattering of Q-balls. We have searched the literature and found some exact, analyt- ical solutions for Q-balls for Higgs-like potentials and other similar potentials, and this motivated us to investigate exact solutions for Q-balls, with focus on the study of properties such as shape, energy, charge, stability, and splitting, without relying on numerical solutions. 1 Introduction In high energy physics, defect structures can engender topo- logical or nontopological proﬁles. Topological structures are stable thanks to topological arguments, because one can asso- ciate conserved currents to them, which are conserved by construction, due to the topological properties of the conﬁg- urations [1]. However, nontopological structures [2,3] do not attain topological charges to make them stable, but they can be stabilized in a diversity of cases, in particular as Q-balls [4,5]. We start the investigation in Sect. 2, where we describe the models and review some basic facts about Q-balls. We continue the study in Sect. 3, where we investigate several speciﬁc models and study stability and other related features. We summarize the results and add some comments to end the work in Sect. 4. The basic properties of Q-balls have been largely stud- ied in the literature [3–23], and the investigations usually require a numerical approach, since it is hard to ﬁnd ana- lytical solutions for the nonlinear equations that govern the system. These methods allow us to understand the basic prop- erties [14] such as its existence and stability in certain (thin [4] and thick wall [9]) limits, but this usually requires a ﬁne tuning of parameters. Eur. Phys. J. C (2016) 76:241 DOI 10.1140/epjc/s10052-016-4059-z Regular Article - Theoretical Physics Regular Article - Theoretical Physics Regular Article - Theoretical Physics 2 The models (5) (5) To search for Q-balls we take the usual ansatz ϕ(x, t) = σ(x) eiωt. (6) ϕ(x, t) = σ(x) eiωt. (6) The conserved Noether charge is Q = 1 2i ∞ ∞ dx ¯ϕ ˙ϕ −ϕ ˙¯ϕ , (7) Fig. 1 The effective potential (12) depicted for n = 1, a = 4/9, and ω2 = 0.6, 0.7, and 0.8. The thickness of the line increases with ω2. In the inset, we show the behavior of the effective potential for σ ∈[0, 1] (7) or better or better Q = ω ∞ ∞ dx σ 2(x). Q = ω ∞ ∞ dx σ 2(x). (8) (8) larger values. We then see that there is a large range of values for the parameter a that can, in principle, give rise to Q-balls. In Fig. 1 we depict the potential for a = 4/9 and n = 1, for several values of ω2, that satisﬁes the condition (13); in the inset we illustrate how the potential vanishes near the origin for nonvanishing σ. The equation of motion (9) admits the solution The equation of motion becomes σ ′′ = (1 −ω2)σ −2 + n 3 σ 1+n + a(1 + n) 2 σ 1+2n. (9) As usual, we consider the boundary conditions (10) (10) σ ′(0) = 0; σ(∞) = 0. (10) σ(x) = 1 −ω2 2a 1 2n tanh n √ 1 −ω2 2 x + b −tanh n √ 1 −ω2 2 x −b 1 n , (14) The above equation of motion (9) can be cast in the form The above equation of motion (9) can be cast in the form σ ′′ = dU dσ , (11) σ ′′ = dU dσ , (11) (14) with U = U(σ) being a kind of effective potential for the ﬁeld σ. It has the form where U(σ) = 1 2(1 −ω2)σ 2 −1 3σ 2+n + 1 4a σ 2+2n. (12) b = 1 2 arctanh ⎛ ⎝3 (1 −ω2) a 2 ⎞ ⎠. 2 The models In order to investigate Q-balls, we consider the Lagrange density In order to investigate Q-balls, we consider the Lagrange density L = 1 2∂μ ¯ϕ∂μϕ −V (\|ϕ\|), (1) (1) where ϕ is a complex scalar ﬁeld and In the simplest case, the presence of Q-balls is related to the existence of global U(1) symmetries. However, in the standard model the presence of global U(1) symme- tries can be related to baryonic and leptonic charges, and in extended supersymmetric models the scalar superpartners of baryons and leptons can condensate and give rise to Q-balls. In this sense, Q-balls are of current interest to baryogene- sis, for instance. As is well known, one can suggest that the V (\|ϕ\|) = 1 2m2 \|ϕ\|2 −1 3α \|ϕ\|2+n + 1 4β \|ϕ\|2+2n (2) (2) is the potential, with n = 1, 2, 3 . . . This model was ﬁrst considered in [7], and for n = 1 it reproduces the model investigated before in [6,8], and for n = 2 it gives the model studied in [3]. Here, however, we explore the several val- ues of n, and study the shape, energy, charge, stability and splitting of the corresponding Q-balls. In the potential, we consider m > 0 as the mass parameter, with α and β being 12 3 241 Page 2 of 13 Eur. Phys. J. C (2016) 76 :241 241 Page 2 of 13 241 real parameters. Using the rescaling real parameters. Using the rescaling Fig. 1 The effective potential (12) depicted for n = 1, a = 4/9, and ω2 = 0.6, 0.7, and 0.8. The thickness of the line increases with ω2. In the inset, we show the behavior of the effective potential for σ ∈[0, 1] real parameters. Using the rescaling real parameters. Using the rescaling ϕ → m2 α 1 n ϕ, x →x m , L →m2 m2 α 2 n L, (3) we get the Lagrange density L = 1 2∂μϕ∗∂μϕ −1 2\|ϕ\|2 + 1 3\|ϕ\|2+n −1 4a \|ϕ\|2+2n, (4) (4) with a = βm2/α2. We are working in (1, 1) spacetime dimensions, so the equation of motion has the form with a = βm2/α2. We are working in (1, 1) spacetime dimensions, so the equation of motion has the form ¨ϕ −ϕ′′ + ϕ −2 + n 3 \|ϕ\|nϕ + a(1 + n) 2 \|ϕ\|2nϕ = 0. 2 The models (8); the general result has the form L+ = 4 arccosh 2 n 2 δ−1 2 + 9a √ 1 −2−n 4 δ 1 2 + O δ 3 2 , (24) (24) Q = 2 1 n −1√π a 1 n ω 1 −ω2 1 n −1 2 2 + n tanh 2 n b sech2b 2 + 2 n 3 2 + 2 n × 2(2 + n)2F1 −1 2, 2 n ; 3 2 + 2 n ; tanh2 b −n(1 + tanh2 b)2F1 1 2, 2 n ; 3 2 + 2 n ; tanh2 b , (18) which, in the limit ω →ω+, diverges faster as we increase n. The amplitude (16) behaves as A+ = 3δ 2 1 n 1 + 9a 8n δ + O δ2 . (25) (25) Then, as ω approaches ω+, we see that the width becomes increasingly larger, while the amplitude becomes smaller and smaller. (18) where 2F1(a, b; c; z) is the hypergeometric function and (z) is the Gamma function. We see from (15) that b increases to larger and larger values, as ω decreases toward its lower bound, ω−, which depends on a. Thus, one can rewrite the width (17) in terms of the parameters n, a, and b, to study its asymptotic behavior with respect to b, for a given a and n; we can write It is possible to obtain the behavior of the solution for ω ≈ω+; it is σ+(x) ≈A sech 2 n n √ 1 −ω2 2 x , (19) (19) L−= 6 √ 2a 1 2 ln 2n −1 + b + e−2b 2n −1 + O e−4b . (26) where A is the amplitude, as it appears in Eq. (16), which, for ω ≈ω+, behaves as A ≈ 3 (ω2 + −ω2)/2 1/n. We also have studied how the effective potential (12) behaves in the limit ω →ω+, when calculated with the solution (14) at x = 0, which is the amplitude (16); we deﬁne δ = ω2 + −ω2 to get (26) Then we see that the width increases linearly with b, as b increases to larger values. 2 The models As a ≥2/9, and we see that the maximum amplitude starts at A = 31/n fora = 2/9 and decreases up to zero as we increase a, for a given n. We deﬁne the width L of the solution as the width at half height: Q+ = √π 2(2 + n) 2 + 2 n 3 2 + 2 n 3 2 2 n δ 2 n −1 2 × 4 + n + 9a(2 + n) −n(4 + n) 2n δ + O δ 3 2 . (23) L ≡ 4 √ 1 −ω2 arcsech ⎛ ⎝ 1 −tanh2 b 2n −tanh2 b ⎞ ⎠. (17) (23) (17) Then, when ω →ω+, the charge tends to zero for n < 4. For n = 4, it tends to a positive constant. For n > 4, it diverges, and this modiﬁes the stability of the system. The width (17) behaves as Then, when ω →ω+, the charge tends to zero for n < 4. For n = 4, it tends to a positive constant. For n > 4, it diverges, and this modiﬁes the stability of the system. The width (17) behaves as We see from the above expression that the width of the solu- tion increases as ω decreases toward ω−, making the solution to develop a plateau of height A−. We see from the above expression that the width of the solu- tion increases as ω decreases toward ω−, making the solution to develop a plateau of height A−. It is possible to use the exact solution (14) to calculate the charge from Eq. 2 The models (15) (12) (15) As is well known, in order to have solutions obeying the boundary conditions (10), the effective potential U(σ) has to have: Since a ≥2/9 and ω2 are bounded according to (13), b does not vanish, so the solution (14) is bell-shaped, with amplitude given by the value of σ that identiﬁes a zero of the effective potential. We call this the point of return of the solution, A. It is given by • symmetry breaking; • zeros at points in which σ is nonvanishing. • zeros at points in which σ is nonvanishing. The ﬁrst condition gives a superior bound for ω, that is, ω+ = V ′′(0) = 1. The second condition gives the infe- rior bound ω−= 2V (σ0)/σ 2 0 = √1 −2/(9a), where σ0 is the minimum of V (σ)/σ 2. Then ω is such that The ﬁrst condition gives a superior bound for ω, that is, ω+ = V ′′(0) = 1. The second condition gives the infe- rior bound ω−= 2V (σ0)/σ 2 0 = √1 −2/(9a), where σ0 is the minimum of V (σ)/σ 2. Then ω is such that A = 2 − 4 −18a(1 −ω2) 3a 1 n , (16) (16) and it obeys U(A) = 0. The point of return controls the amplitude of the solution, and it depends on a and ω, such that for a given a it diminishes as ω2 increases. We deﬁne A−≡[2/(3a)]1/n as the limit of the amplitude in the case ω →ω−and A+ ≡0 as the limit of ω →ω+. Then the and it obeys U(A) = 0. The point of return controls the amplitude of the solution, and it depends on a and ω, such that for a given a it diminishes as ω2 increases. We deﬁne A−≡[2/(3a)]1/n as the limit of the amplitude in the case ω →ω−and A+ ≡0 as the limit of ω →ω+. Then the ω−< ω < ω+. (13) (13) ω−< ω < ω+. We take a ≥2/9 to ensure that ω is real. Thus, ω−varies in the interval [0, 1), as a increases from 2/9 to larger and 123 Page 3 of 13 241 Eur. Phys. J. C (2016) 76 :241 regime, for the charge (18), we have regime, for the charge (18), we have amplitude of our solution is such that A+ < A < A−. 2 The models (41) Thi b i t t d t i E E + E i il t (30) (31) ϵp = ϵk + ϵg. (41) (41) This can be integrated to give E p = Ek + Eg, similar to the virial theorem. The same expression can be obtained from scaling arguments [26,27]. However, the condition that appears from the energy densities is stronger, because it is valid locally, for any x. are the kinetic, gradient, and potential portions of the energy density. After using the ansatz (6), the energy density becomes ϵ = 1 2σ ′2 + ω2 + 1 2 σ 2 −1 3σ 2+n + 1 4a σ 2+2n. (33) We can integrate the energy density to ﬁnd the total energy (33) E = 2 (Ek + Eg). (42) (42) We can substitute Eq. (14) in the above equation to get the explicit form of the energy density; however, the full expres- sion is cumbersome, and so we omit it here. We have made a closer inspection on the energy density, searching for any possible change of behavior. For a given a, one can show that it starts to split, the splitting appearing for ω in the interval ω−< ω < ωc, where The potential energy does not appear in the above expression because of the constraint given by Eq. (41). We note that the kinetic energy can be written in terms of the charge (18) The potential energy does not appear in the above expression because of the constraint given by Eq. (41). We note that the kinetic energy can be written in terms of the charge (18) Ek = ω Q 2 . (43) (43) The gradient energy is given by The gradient energy is given by ωc = n a (n+2)2−18a(n+1)+(9a−1)(n+1)−1 1/2 3 (n+1) . (34) Eg = √π 2 1 n −1 n a1/n 2 n 3 2 + 2 n 1 −ω2 1 n + 1 2 tanh 2 n b × 2F1 −1 2, 2 n ; 3 2 + 2 n ; tanh2 b . (44) (34) (34) Since ωc must be real, we see that the splitting appears if (44) Since ωc must be real, we see that the splitting appears if 2 9 ≤a < ac = 1 18 (n + 2)2 n + 1 . 2 The models We can implement a similar inves- tigation, to show that the charge (8) can be written as Q−= 2 3a 2 n √18a −4 n × n 2 −H 2 n + 2b −2e2b + O e−4b , (27) 1 2(1 −ω2)σ 2 = 1 2 3δ 2 2 n δ + 9a 4n δ2 + O δ3 , (20) 1 3σ 2+n = 1 2 3δ 2 2 n δ + 9a(2 + n) 8n δ2 + O δ3 , (21) 1 4a σ 2+2n = 9 16 3δ 2 2 n aδ2 + O δ3 . (22) (27) which shows a behavior similar to the width. In this case, Hm is the mth harmonic number. Thus, both the charge and the width vary linearly with b, for larger values of b. (21) (22) The model under investigation has the energy-momentum tensor We see that the terms with σ 2 and σ 2+n are proportional to δ1+2/n and the term with σ 2+2n is proportional to δ2+2/n. Then, when ω →ω+, we have δ →0, and the term σ 2+2n can be neglected and the solution in this approximation is given by Eq. (19); this result ﬁts within the thick wall approx- imation, as it is considered in [9] for small Q-balls. In this Tμν = 1 2∂μ ¯ϕ∂νϕ + 1 2∂μϕ∂ν ¯ϕ −ημνL. (28) (28) The energy density can be calculated from (28) with the Lagrange density (4). It is given by The energy density can be calculated from (28) with the Lagrange density (4). It is given by Lagrange density (4). It is given by ϵ = ϵk + ϵg + ϵp, (29) ϵ = ϵk + ϵg + ϵp, (29) 123 (29) ϵ = ϵk + ϵg + ϵp, 123 3 241 Page 4 of 13 Eur. Phys. J. C (2016) 76 :241 σ ′2 = 1 −ω2 −2 3σ n + 1 2a σ 2n σ 2, (40) where (40) ϵk = 1 2\| ˙ϕ\|2, (30) ϵg = 1 2\|ϕ′\|2, (31) ϵp = 1 2\|ϕ\|2 −1 3\|ϕ\|2+n + 1 4a \|ϕ\|2+2n, (32) 3 2 and it is solved by the solution (14). Furthermore, the condi- tion T11 = 0 lead us to ϵp = ϵk + ϵg. 2 The models (35) Let us now turn attention to the stability of the Q-ball; see, e.g., Refs. [4,22,23] and references therein. To make the investigation complete, we highlight the possibilities: (35) The other components of the energy-momentum tensor (28) are The other components of the energy-momentum tensor (28) are • Quantum mechanical stability, which concerns stability against decay into free particles. As stated in Eq. (13), Q-ball solutions exist for ω in a speciﬁc range of values. The Q-ball is stable if the ratio between the energy and the charge satisﬁes E/Q < ω+. T01 = Re ˙¯ϕϕ′ , (36) T11 = 1 2\|ϕ′\|2+ 1 2\| ˙ϕ\|2−1 2\|ϕ\|2+ 1 3\|ϕ\|2+n−1 4a\|ϕ\|2+2n. (37) (36) (37) • Classical stability, which concerns stability under small perturbations of the ﬁeld conﬁguration. The Q-ball is classically stable if dQ/dω < 0. This means that the charge Q is monotonically decreasing with ω. With the ansatz (6), they become T01 = 0, (38) T11 = 1 2σ ′2 −1 −ω2 2 σ 2 + 1 3σ 2+n −1 4a σ 2+2n. (39) (38) There is another type of stability, against ﬁssion, which requires that d2E/dQ2 < 0. However, as we know that ∂E/∂Q = ω, it is straightforward to show that classi- cally stable Q-balls are also stable against ﬁssion. In our model, Eq. (23) allows us to see that the charge is inﬁnite for ω →ω+, when n > 4. Thus, models with n > 4 are classi- cally unstable. In the next section we study some particular cases of the general potential (12). There is another type of stability, against ﬁssion, which requires that d2E/dQ2 < 0. However, as we know that ∂E/∂Q = ω, it is straightforward to show that classi- cally stable Q-balls are also stable against ﬁssion. In our model, Eq. (23) allows us to see that the charge is inﬁnite for ω →ω+, when n > 4. Thus, models with n > 4 are classi- cally unstable. In the next section we study some particular cases of the general potential (12). Since the energy-momentum tensor is conserved, i.e., ∂μT μν Since the energy-momentum tensor is conserved, i.e., ∂μT μν μ = 0, we see that T11 is constant. Although T11 is not zero in general, we see that it vanishes on shell, that is, if we take for σ the solution (14). 2 The models We also note that the equation of motion (11) can be written in ﬁrst-order form, compatible with T11 = 0, as 3 3 Page 5 of 13 241 Eur. Phys. J. C (2016) 76 :241 3 Illustration Let us now illustrate our results, investigating several distinct possibilities, controlled by the integer n, which we choose to be n = 1, n = 2, and n = 3. Fig. 3 The behavior of the width (17) for n = 1 as a function of the charge (46) for a = 4/9 3 Illustration Let us now illustrate our results, investigating several distinct possibilities, controlled by the integer n, which we choose to be n = 1, n = 2, and n = 3. 3.1 The case n = 1 3.1 The case n = 1 3.1 The case n = 1 We take n = 1 in Eq. (12) to get U(σ) = 1 2(1 −ω2)σ 2 −1 3σ 3 + 1 4a σ 4. (45) U(σ) = 1 2(1 −ω2)σ 2 −1 3σ 3 + 1 4a σ 4. (45) (45) This potential contains up to the fourth-order power in the scalar ﬁeld, so it is of current interest to high energy phe- nomenology. It was studied before in [6,8], but here we go further to add some new features, not seen before. The poten- tial is plotted in Fig. 1. Fig. 3 The behavior of the width (17) for n = 1 as a function of the charge (46) for a = 4/9 The solution can be found setting n = 1 in Eq. (14), which is depicted in Fig. 2 for a = 4/9 (ω2 −= 0.5), for several values of ω obeying Eq. (13). In Fig. 2, in the left panel we can see the plateau for ω ≈ω−and in the right panel it is shown that the amplitude of the solution decreases as ω increases toward ω+. The kinetic (43) and gradient (44) energies simplify to The kinetic (43) and gradient (44) energies simplify to Ek = 2ω2√ 1 −ω2 a (2b coth(2b) −1) , (47) Eg = (1 −ω2)3/2 3a × 1 + 3(8b + 3)e4b + 3(8b −3)e8b −e12b 1 −e4b3 . (48) (47) The charge (18) simpliﬁes for n = 1, becoming Q = 4ω √ 1 −ω2 a (2b coth(2b) −1) . (46) (46) We see that Q →0 as ω →ω+, for any a. Also, speciﬁcally for a = 2/9 we have ω−= 0, which makes Q →0 for ω →ω−= 0. For a > 2/9 we have Q →∞for ω →ω−. The width can easily be obtained from Eq. (17). In Fig. 3, we display the width as a function of the charge for a = 4/9. The minimum of this curve can be calculated numerically for each value of a. In particular, for a = 4/9, the minimum appears for ωm ≈0.7757294, Q ≈2.6728904, and L ≈6.7119190. 3 Illustration We deﬁne this as the point that separates small Q-balls from large Q-balls. (48) To study stability we depict in Figs. 4 and 5 the ratio E/Q and Q as a function of ω2, respectively, for three distinct val- ues of a, as we now explain. We start with the lowest value for the parameter a, that is, a = a0 = 2/9 ≈0.2222222. For 0 = ω−< ω < ˜ω ≈0.0752748 one can see that E/Q > ω+, which is out of the interval in which the Q- ball is stable. Furthermore, the charge is not monotonically decreasing with ω. Thus, the case a = a0 is unstable clas- sically and quantum mechanically. We continue the inves- tigation taking values of a higher than a0. The ratio E/Q has its peak above ω+ but, for a increasing, it goes down until we get to a1 = 2/9 + 0.0004596 ≈0.2226819, where the peak in E/Q is approximately equal to ω+. The inter- esting fact in this case is that the ratio E/Q now is in the allowed range that ensures quantum mechanical stability. Nevertheless, the model is yet classically unstable, because the charge is not monotonically decreasing with ω. We go further on, increasing a, and the peak of E/Q in the small ω region goes down and down, until the concavity of the curve changes at a2 = 2/9 + 0.0031751 ≈0.2253973. As we increase a up to a2 the charge tends to inﬁnity for ω →ω− and has a local minimum in the small ω region, which Fig. 2 The solution (14) depicted for n = 1 and a = 4/9, with ω2 = 0.5 + 5ϵ, ϵ = 10−9, 10−8, 10−7, 10−6, and 10−5 (left), and with ω2 = 0.6, 0.7, 0.8 and 0.9 (right). The plateau in the left panel increases as ω approaches ω−. The thickness of the line in the right panel increases as ω2 increases Fig. 2 The solution (14) depicted for n = 1 and a = 4/9, with ω2 = 0.5 + 5ϵ, ϵ = 10−9, 10−8, 10−7, 10−6, and 10−5 (left), and with ω2 = 0.6, 0.7, 0.8 and 0.9 (right). The plateau in the left panel increases as ω approaches ω−. The thickness of the line in the right panel increases as ω2 increases 12 3 Eur. Phys. J. 3 Illustration In each plot, we take ω2 = ω2 −+ 10−3, and the thickness of the line increases as a increases Fig. 7 The effective potential (49) depicted for a = 4/9 and ω2 = 0.6, 0.7, and 0.8. The thickness of the line increases with ω2. In the inset, we show the behavior of the effective potential for σ ∈[0, 1] Fig. 8 The solution (14) depicted for n = 2 and a = 4/9, with ω2 = 0.5 + 5ϵ, ϵ = 10−9, 10−8, 10−7, 10−6, and 10−5 (left), and with ω2 = 0.6, 0.7, 0.8 and 0.9 (right). The plateau in the left panel increases as ω approaches ω−. The thickness of the line in the right panel increases as ω2 increases becomes an inﬂection point when a = a2. Therefore, for a > a2, the solution is stable, both classically and quantum mechanically. As we see, in Fig. 4 we depict the ratio E/Q for the three values of a, a = a0, a = a1, and a = a2. We note that quan- tum mechanical instability appears only in the top panel, because E/Q may overcome ω+. Moreover, classical stabil- ity only appears in the bottom panel of the ﬁgure. We also depict in Fig. 5 the charge as a function of ω2, for the same three values of a (a0, a1, a2), and there we see that the charge only becomes a monotonically decreasing function of ω2 for a > a2. Fig. 8 The solution (14) depicted for n = 2 and a = 4/9, with ω2 = 0.5 + 5ϵ, ϵ = 10−9, 10−8, 10−7, 10−6, and 10−5 (left), and with ω2 = 0.6, 0.7, 0.8 and 0.9 (right). The plateau in the left panel increases as ω approaches ω−. The thickness of the line in the right panel increases as ω2 increases We have done a closer inspection on the energy den- sity, searching for any possible change of behavior. Setting n = 1 in Eq. (35), we see that the energy density tends to split when a is in the interval a2 < a < a3 = 1/4, with ω in the range of Eq. (34). We start with a = a2, and as we increase a, the central well in the energy den- sity becomes a hill, making the splitting to vanish. We illus- trate this in Fig. 3 Illustration C (2016) 76 :241 241 Page 6 of 13 g y ( ) Fig. 4 The ratio E/Q as a function of ω2, for the parameter a as a0 (top panel), a1 (center panel), and a2 (bottom panel). The region in between the two dashed horizontal lines ensures quantum mechanical Fig. 5 The charge as a function of ω2, for the case n = 1, with th parameter a as a0 (top panel) a1 (center panel) and a2 (bottom pane Fig. 4 The ratio E/Q as a function of ω2, for the parameter a a (top panel), a1 (center panel), and a2 (bottom panel). The regio between the two dashed horizontal lines ensures quantum mechan stability of the Q-ball Fig. 5 The charge as a function of ω2, for the case n = 1, with parameter a as a0 (top panel), a1 (center panel), and a2 (bottom pa Fig. 4 The ratio E/Q as a function of ω2, for the parameter a as a0 (top panel), a1 (center panel), and a2 (bottom panel). The region in between the two dashed horizontal lines ensures quantum mechanical stability of the Q-ball Fig. 5 The charge as a function of ω2, for the case n = 1, with the parameter a as a0 (top panel), a1 (center panel), and a2 (bottom panel) 12 3 Eur. Phys. J. C (2016) 76 :241 Page 7 of 13 241 Fig. 6 The energy density for a = 9/40, 1/4, and 11/40. In each plot, we take ω2 = ω2 −+ 10−3, and the thickness of the line increases as a increases Fig. 7 The effective potential (49) depicted for a = 4/9 and ω2 = 0.6, 0.7, and 0.8. The thickness of the line increases with ω2. In the inset, we show the behavior of the effective potential for σ ∈[0, 1] Fig. 6 The energy density for a = 9/40, 1/4, and 11/40. In each plot, we take ω2 = ω2 −+ 10−3, and the thickness of the line increases as a increases Fig. 7 The effective potential (49) depicted for a = 4/9 and ω2 = 0.6, 0.7, and 0.8. The thickness of the line increases with ω2. In the inset, we show the behavior of the effective potential for σ ∈[0, 1] 2 Fig. 6 The energy density for a = 9/40, 1/4, and 11/40. 3 Illustration 6, where we depict the energy density for a = 9/40, 1/4, and 11/40, using ω2 = ω2 −+10−3. It is inter- esting to note that the tendency to split starts to appear for a ∈(a2, a3), for ω in the interval (34), so it is inside the range where the Q-ball is stable, both quantum mechanically and classically. This potential is plotted in Fig. 7. With the right scaling of parameters, one can show that this is the same case studied in [3], but here we go further and add new features to the model. The solution can be found setting n = 2 in Eq. (14), which is depicted in Fig. 8 for a = 4/9 (ω2 −= 0.5), for several values of ω obeying Eq. (13). In Fig. 8, in the left panel we can see the plateau for ω ≈ω−and in the right panel it is shown that the amplitude of the solution decreases as ω increases toward ω+. 3.2 The case n = 2 3.2 The case n = 2 3.2 The case n = 2 The charge (18) simpliﬁes for n = 2, becoming We take n = 2 in Eq. (12) to get to Q = 2 a ω arctanh ⎛ ⎝3 (1 −ω2) a 2 ⎞ ⎠. (50) (50) (49) 12 241 Page 8 of 13 Eur. Phys. J. C (2016) 76 :241 g y Fig. 9 The behavior of the width (17) for n = 2 as a function of the charge (50) for a = 4/9 Fig. 9 The behavior of the width (17) for n = 2 as a function of the charge (50) for a = 4/9 Fig. 10 The ratio E/Q as a function of ω2, for the case n = 2, the parameter a as a0 (top panel), a1 (center panel), and a2 (bo panel). The region in between the two dashed horizontal lines ens quantum mechanical stability of the Q-ball Fig. 9 The behavior of the width (17) for n = 2 as a function of the charge (50) for a = 4/9 We see that Q →0 as ω →ω+, for any a. Also, speciﬁcally for a = 2/9 we have ω−= 0, which makes Q →0 for ω →ω−= 0. For a > 2/9 we have Q →∞for ω →ω−. The width can easily be obtained from Eq. (17). In Fig. 9, we display the width as a function of the charge for a = 4/9. The minimum of this curve can be calculated numerically for each value of a. In particular, for a = 4/9, the minimum appears for ωm ≈0.7655362, Q ≈2.4796264, and L ≈9.6671940. We deﬁne this as the point that separates small Q-balls from large Q-balls. The kinetic (43) and gradient (44) energies simplify to Ek = ω2 √ 2a arctanh ⎛ ⎝3 (1 −ω2) a 2 ⎞ ⎠ (51) Eg = 1 4 2 a e8b −8b e4b −1 e4b −1 2 1 −ω2 . (52) (51) (52) To study stability as before. We depict in Figs. 10 and 11 the ratio E/Q and Q as a function of ω2, respectively, for three distinct values of a, as we now explain. We start with a = a0 = 2/9 ≈0.2222222. For 0 = ω−< ω < ˜ω ≈ 0.2985278 one can see that E/Q > ω+, which is out of the interval in which the Q-ball is stable. 3.2 The case n = 2 Also, the charge is not monotonically decreasing with ω, and so the case a = a0 is unstable classically and quantum mechanically. We con- tinue the investigation, and for a higher than a0, the ratio E/Q has its peak above ω+ but, for a increasing, it goes down until we get to a1 = 2/9 + 0.0067204 ≈0.2289426, where the peak in E/Q is approximately equal to ω+. The interesting fact in this case is that the ratio E/Q now is in the allowed range that ensures quantum mechanical stability. Nevertheless, the model is yet classically unstable, because Fig. 10 The ratio E/Q as a function of ω2, for the case n = 2, with the parameter a as a0 (top panel), a1 (center panel), and a2 (bottom panel). The region in between the two dashed horizontal lines ensures quantum mechanical stability of the Q-ball 12 123 123 Page 9 of 13 241 Eur. Phys. J. C (2016) 76 :241 g. 11 The charge as a function of ω2, for the parameter a as a0 (top l) ( l) d (b l) Fig. 12 The energy density for a = we take ω2 = ω2 −+ 10−3, and the increases the charge is not monotonica further, increasing a, and th ω region goes down and dow curve changes at a2 = 2/9 As we increase a up to a2 th ω →ω−and has a local mi which becomes an inﬂection a > a2, the solution is stable, mechanically. In Fig. 10 we depict the ratio a = a0, a = a1, and a = a2. W ical instability appears only in may overcome ω+. Also, clas the bottom panel of the ﬁgure charge as a function of ω2, f (a0, a1, a2), and there we see a monotonically decreasing fu As in the previous model, of the energy density. Setting the energy density tends to s a2 < a < a3 = 8/27, with o We start with a = a2, and as that appears in the energy de the splitting to vanish. We ill we depict the energy density using ω2 = ω2 −+ 10−3. We starts to appear for a ∈(a2, where the Q-ball is stable, bo classically Fig. 12 The energy density for a = 7/27, 8/27, and 1/3. 3.2 The case n = 2 In each plot, we take ω2 = ω2 −+ 10−3, and the thickness of the line increases as a increases Fig. 11 The charge as a function of ω2, for the parameter a as a0 ( panel), a1 (center panel), and a2 (bottom panel) Fig. 12 The energy density for a = 7/27, 8/27, and 1/3. In each plot, we take ω2 = ω2 −+ 10−3, and the thickness of the line increases as a increases the charge is not monotonically decreasing with ω. We go further, increasing a, and the peak of E/Q in the small ω region goes down and down, until the concavity of the curve changes at a2 = 2/9 + 0.0164064 ≈0.2386286. As we increase a up to a2 the charge tends to inﬁnity for ω →ω−and has a local minimum in the small ω region, which becomes an inﬂection point when a = a2. Thus, for a > a2, the solution is stable, both classically and quantum mechanically. In Fig. 10 we depict the ratio E/Q for the three values of a, a = a0, a = a1, and a = a2. We note that quantum mechan- ical instability appears only in the top panel, because E/Q may overcome ω+. Also, classical stability only appears in the bottom panel of the ﬁgure. We also depict in Fig. 11 the charge as a function of ω2, for the same three values of a (a0, a1, a2), and there we see that the charge only becomes a monotonically decreasing function of ω2 for a > a2. As in the previous model, we have studied the behavior of the energy density. Setting n = 2 in Eq. (35), we see that the energy density tends to split when a is in the interval a2 < a < a3 = 8/27, with omega in the range of Eq. (34). We start with a = a2, and as a increases, the central well that appears in the energy density becomes a hill, making the splitting to vanish. We illustrate this in Fig. 12, where we depict the energy density for a = 7/27, 8/27, and 1/3, using ω2 = ω2 −+ 10−3. We note that the tendency to split starts to appear for a ∈(a2, a3), so it is inside the range where the Q-ball is stable, both quantum mechanically and classically. Fig. 3.2 The case n = 2 11 The charge as a function of ω2, for the parameter a as a0 (top panel), a1 (center panel), and a2 (bottom panel) 12 123 Eur. Phys. J. C (2016) 76 :241 241 Page 10 of 13 Fig. 13 The effective potential (53) depicted for a = 4/9 and ω2 = 0.6, 0.7, and 0.8. The thickness of the line increases with ω2. In the inset, we show the behavior of the effective potential for σ ∈[0, 1] Fig. 15 The behavior of the width (17) for n = 2 as a function of the charge (50) for a = 4/9 Fig. 15 The behavior of the width (17) for n = 2 as a function of the charge (50) for a = 4/9 Fig. 13 The effective potential (53) depicted for a = 4/9 and ω2 = 0.6, 0.7, and 0.8. The thickness of the line increases with ω2. In the inset, we show the behavior of the effective potential for σ ∈[0, 1] a > 2/9 we have Q →∞for ω →ω−. The width can easily be obtained from Eq. (17). In Fig. 15, we display the width as a function of the charge for a = 4/9. The investigation is similar to the previous one, and for a = 4/9, the minimum appears for ωm ≈0.7580576, Q ≈2.2272465, and L ≈ 12.1294293. This is the point that separates small Q-balls from large Q-balls in this case. Fig. 14 The solution (14) depicted for n = 3 and a = 4/9, with ω2 = 0.5 + 5ϵ, ϵ = 10−9, 10−8, 10−7, 10−6, and 10−5 (left), and with ω2 = 0.6, 0.7, 0.8 and 0.9 (right). The plateau in the left panel increases as ω approaches ω−. The thickness of the line in the right panel increases as ω2 increases To study stability we depict in Figs. 16 and 17 the ratio E/Q and Q asafunctionofω2,respectively,forthreedistinct values of a. We start with the lowest value for the parameter a, that is, a = a0 = 2/9 ≈0.2222222. For 0 = ω−< ω < ˜ω ≈0.6202147 one can see that E/Q > ω+, which is out of the interval in which the Q-ball is stable. Also, the charge is not monotonically decreasing with ω, thus the case a = a0 is unstable classically and quantum mechanically. We then consider a to be higher than a0. 3.2 The case n = 2 The ratio E/Q has its peak above ω+ but, for a increasing, it goes down until we get to a1 = 2/9 + 0.0351702 ≈0.2573924, where the peak in E/Q is approximately equal to ω+. The ratio E/Q is now in the allowed range to ensure quantum mechanical stability. Nevertheless, the model is yet classically unstable, since the charge is not monotonically decreasing with ω. We keep increasing a, and the peak of E/Q in the small ω region goes down and down, until the concavity of the curve changes at a2 = 2/9 + 0.0534856 ≈0.2757078. As we increase a up to a2 the charge goes to inﬁnity for ω →ω−, and it has a local minimum in the small ω region, which becomes an inﬂection point when a = a2. For a > a2, the solution is then stable, both classically and quantum mechanically. Fig. 14 The solution (14) depicted for n = 3 and a = 4/9, with ω2 = 0.5 + 5ϵ, ϵ = 10−9, 10−8, 10−7, 10−6, and 10−5 (left), and with ω2 = 0.6, 0.7, 0.8 and 0.9 (right). The plateau in the left panel increases as ω approaches ω−. The thickness of the line in the right panel increases as ω2 increases 3.3 The case of n = 3 We take n = 3 in Eq. (12) to get the potential U(σ) = 1 2(1 −ω2)σ 2 −1 3σ 5 + 1 4a σ 8. (53) 3.3 The case of n = 3 We take n = 3 in Eq. (12) to get the potential U(σ) = 1 2(1 −ω2)σ 2 −1 3σ 5 + 1 4a σ 8. (53) We take n = 3 in Eq. (12) to get the potential U(σ) = 1 2(1 −ω2)σ 2 −1 3σ 5 + 1 4a σ 8. (53) (53) This potential is plotted in Fig. 13. This potential is plotted in Fig. 13. This potential is plotted in Fig. 13. The solution can be found setting n = 3 in Eq. (14), which is depicted in Fig. 14 for a = 4/9 (ω2 −= 0.5), for several values of ω obeying Eq. (13). In this Fig. 14, we see the plateau for ω ≈ω−in the left panel, and it also shows that the amplitude of the solution decreases as ω increases toward ω+, in the right panel. In Fig. 10 we depict the ratio E/Q for the three values of a, a = a0, a = a1, and a = a2. We note that quantum mechanical instability appears only in the top panel, because E/Q may overcome ω+. However, classical stability only We have not been able to ﬁnd a simpler expression for (18) in the case n = 3. However, it is possible to see that Q →0 as ω →ω+, for any a. Also, speciﬁcally for a = 2/9 we have ω−= 0, which makes Q →0 for ω →ω−= 0. For 123 Page 11 of 13 241 Eur. Phys. J. C (2016) 76 :241 Eur. Phys. J. C (2016) 76 :241 Fig. 16 The ratio E/Q as a function of ω2, for the case n = 3, he parameter a as a0 (top panel), a1 (center panel), and a2 (bo anel). The region in between the two dashed horizontal lines en quantum mechanical stability of the Q-ball ur. Phys. J. C (2016) 76 :241 Page 11 of 13 241 ig. 16 The ratio E/Q as a function of ω2, for the case n = 3, with he parameter a as a0 (top panel), a1 (center panel), and a2 (bottom anel). The region in between the two dashed horizontal lines ensures uantum mechanical stability of the Q-ball Fig. 17 The charge as a function of ω2, for the parameter a as a0 (top panel), a1 (center panel), and a2 (bottom panel) Fig. 3.3 The case of n = 3 16 The ratio E/Q as a function of ω2, for the case n = 3, with the parameter a as a0 (top panel), a1 (center panel), and a2 (bottom panel). The region in between the two dashed horizontal lines ensures quantum mechanical stability of the Q-ball Fig. 17 The charge as a function of ω2, for the parameter a as a0 (top panel), a1 (center panel), and a2 (bottom panel) 12 3 241 Page 12 of 13 Eur. Phys. J. C (2016) 76 :241 Fig. 18 The energy density for a = 5/18, 25/72, and 5/12. In each plot, we take ω2 = ω2 −+ 10−3, and the thickness of the line increases as a increases sponding dynamics [18] and modify the scenario for charge- swapping Q-ball interactions [21]. An interesting issue con- cerns the tendency to split, which should be further investi- gated. Another related issue concerns duality, as suggested in [20], where the (Noether) charge of a stationary Q-ball may be dual to the (topological) charge of a static, kinklike structure. The exact results that we constructed above are of direct importance for Q-balls, and they can be used to improve the numerical simulations that appear in the related literature. Perhaps, they may shed light on the present understand- ing of issues such as baryogenesis, dark matter, and other related areas of phenomenology, as we ﬁnd, for instance, in Refs. [28–30] and in the references therein. Another issue of current interest is related to string theory and concerns scenarios involving extra dimensions. A recent work on Q-branes [31] shows that dielectric D-brane systems admit non-abelian Q-ball solutions on their world-volume, and we are now investigating other possible scenarios. Fig. 18 The energy density for a = 5/18, 25/72, and 5/12. In each plot, we take ω2 = ω2 −+ 10−3, and the thickness of the line increases as a increases Acknowledgments We would like to acknowledge the Brazilian agency CNPq for partial ﬁnancial support. DB thanks support from contracts 455931/2014-3 and 06614/2014-6, MAM thanks support from contract 140735/2015-1, and RM thanks support from contracts 508177/2010-3 and 455619/2014-0. appears in the bottom panel of the ﬁgure. We also depict in Fig. 11 the charge as a function of ω2, for the same three values of a (a0, a1, a2), and there we see that the charge only becomes a monotonically decreasing function of ω2 for a > a2. 3.3 The case of n = 3 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecomm ons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. We have investigated the behavior of the energy density, to see if it splits, as it appeared in the previous models. Taking n = 3 in Eq. (35), we see that the energy density tends to split when a is in the interval a2 < a < a3 = 25/72, with omega in the range of Eq. (34). We start with a = a2, and as we increase a, the well that appears in the central region of the energy density becomes a hill, making the splitting to vanish. We illustrate this in Fig. 18, where we depict the energy density for a = 5/18, 25/72, and 5/12, using ω2 = ω2 −+ 10−3. It is interesting to note that the tendency to split starts to appear for a ∈(a2, a3), so it is inside the range where the Q-ball is stable, both quantum mechanically and classically. Funded by SCOAP3. References 1. N. Manton, P. Sutcliffe, Topological solitons (Cambridge Univer- sity Press, Cambridge, 2004) 2. L. Wilets, Nontopological solitons (World Scientiﬁc, 1989) 3. T.D. Lee, Y. Pang, Phys. Rep. 221, 251 (1992) 4. S. Coleman, Nucl. Phys. B 262, 263 (1985) [erratum 269, 744 (1986)] Also, we investigated the case with n = 4, and we found similar results, so we omit it here. 5. J.A. Frieman, G.B. Gelmini, M. Gleiser, E.W. Kolb, Phys. Rev. Lett. 60, 2101 (1988) 6. J.M. Cerveró, P.G. Estévez, Phys. Lett. B 176, 139 (1986) 7. C.N. Kumar, A. Khare, J. Phys. A 20, L1219 (1987) 8. T.I. Belova, A.E. Kudryavtsev, Sov. Phys. JETP 68 4 Ending comments 9. A. Kusenko, Phys. Lett. B 404, 285 (1997) 10. A. Kusenko, V. Kuzmin, M. Shaposhnikov, P.G. Tinyakov, Phys. Rev. Lett. 80, 3185 (1998) 10. A. Kusenko, V. Kuzmin, M. Shaposhnikov, P.G. Tinyakov, Phys. 10. A. Kusenko, V. Kuzmin, M. Shaposhnikov Rev. Lett. 80, 3185 (1998) In this work we studied Q-balls in models engendering global U(1) symmetry. We found exact solutions for several distinct values of the parameters that describe the models. In par- ticular, we found that the Q-ball engender the tendency to split, and we also found regions in parameter space where the Q-ball is unstable, quantum mechanically stable, and quantum mechanically and classically stable. The results unveil abehavior whichmaycontributeandenlargethecorre- ( ) 11. A. Kusenko, M. Shaposhnikov, Phys Lett. B 417, 99 (1998) 12. K. Enqvist, J. McDonald, Phys. Lett. B 425, 309 (1998) 13. K. Enqvist, J. McDonald, Nucl. Phys. B 538, 321 (1999) 14. T. Multamäki, I. Vilja, Nucl. Phys. B 574, 130 (2000 15. M. Axenides, S. Komineas, L. Perivolaropoulos, M. Floratos, Phys. Rev. D 61, 085006 (2000) 15. M. Axenides, S. Komineas, L. Perivolaropoulos, M. F 26. R. Hobart, Proc. Phys. Soc. Lond. 82, 201 (1963) 27. G.H. Derrick, J. Math. Phys. 5, 1252 (1964) 28. F. Doddato, J. MacDonald, JCAP 1206, 031 (1012) 29. F. Doddato, J. MacDonald, JCAP 1307, 004 (2013) 30. S. Kasuya, M. Kawasaki, Phys. Rev. D 89, 103534 (2014) 31. S. Abel, A. Kehagias, JHEP 1511, 096 (2015) 19. A. Kusenko, P.J. Steinhardt, Phys. Rev. Lett. 87, 141301 (2001) 20. P. Bowcock, D. Foster, P. Sutcliffe, J. Phys. A 42, 085403 (2009) 21. E.J. Copeland, P.M. Safﬁn, S.-Y. Zhou, Phys. Rev. Lett. 113, 231603 (2014) 22. A. Tranberg, D.J. Weir, JHEP 1404, 184 (2014) 23. EYa. Nugaev, M.N. Smolyakov, JHEP 1407, 009 (2014) 24. I. Afﬂeck, M. Dine, Nucl. Phys. B 249, 361 (1985) 25. K. Enqvist, S. Kasuya, A. Mazumdar, Phys. Rev. Lett. 89, 091301 (2002) 25. K. Enqvist, S. Kasuya, A. Mazumdar, Phys. Rev. Lett. 89, 091301 (2002) 20. P. Bowcock, D. Foster, P. Sutcliffe, J. Phys. A 42, 085403 (2009) 24. I. Afﬂeck, M. Dine, Nucl. Phys. B 249, 361 (1985) 19. A. Kusenko, P.J. Steinhardt, Phys. Rev. Lett. 87, 141301 (2001) 19. A. Kusenko, P.J. Steinhardt, Phys. Rev. Lett. 87, 141301 (2001) 20. P. Bowcock, D. Foster, P. Sutcliffe, J. Phys. A 42, 085403 (2009) 21. E.J. Copeland, P.M. Safﬁn, S.-Y. Zhou, Phys. Rev. Lett. 113, 231603 (2014) 22. A. Tranberg, D.J. Weir, JHEP 1404, 184 (2014) 23 EYa Nugaev M N Smolyakov JHEP 1407 009 (2014) Rev. D 61, 085006 (2000) 16. S. Kasuya, M. Kawasaki, Phys. Rev. D 62, 023512 (2000) 17. S. Kasuya, M. Kawasaki, Phys. Rev. D 64, 123515 (2001) 18. R.A. Battye, P.M. Sutcliffe, Nucl. Phys. B 590, 329 (20 123 Eur. Phys. J. C (2016) 76 :241 Page 13 of 13 241 20. P. Bowcock, D. Foster, P. Sutcliffe, J. Phys. A 42, 085403 (2009) 21. E.J. Copeland, P.M. Safﬁn, S.-Y. Zhou, Phys. Rev. Lett. 113, 231603 (2014) 21. E.J. Copeland, P.M. Safﬁn, S.-Y. Zhou, Phys. Rev. Lett. 113, 231603 (2014) 22. A. Tranberg, D.J. Weir, JHEP 1404, 184 (2014) 23. EYa. Nugaev, M.N. Smolyakov, JHEP 1407, 009 (2014) 24. I. Afﬂeck, M. Dine, Nucl. Phys. B 249, 361 (1985) 25. K. Enqvist, S. Kasuya, A. Mazumdar, Phys. Rev. Lett. 89, 091301 (2002) 12 123
https://openalex.org/W2807334545	https://www1.racgp.org.au/getattachment/5025a4ff-af3f-4fea-8ad7-81bda90a8883/Queensland-tick-typhus.aspx	English	null	Queensland tick typhus (Rickettsia australis) in a man after hiking in rural Queensland	null	2,018	cc-by-sa	1,357	CLINICAL CLINICAL Answer 3 Serological R. rickettsii indirect immunofluorescence assay detects the infection at 1:128 dilution, with cross- reactivity between R. rickettsii (Rocky Mountain spotted fever) and R. australis (Queensland tick typhus).3 Positive cases that are detected at 1:128 dilution have repeated titrations of sera, providing a final titre result.4 Question 4 What is the treatment for this disease? Question 1 Where and how do people contract Queensland tick typhus? Where and how do people contract Queensland tick typhus? Question 3 Question 3 What testing is done for this disease? What testing is done for this disease? Question 2 How does the disease typically present? Further questioning revealed a history of visiting the Bunya Mountains National Park, Queensland, for a hiking trip two weeks prior to presentation. The patient was not aware of being bitten by ticks or other insects; however, Queensland tick typhus was suspected clinically, given this is a common presentation in patients with tick bites who have visited that area, while Sweet syndrome was considered as a remote differential. Initial management included doxycycline 100 mg twice daily and follow-up. Answer 2 A man aged 51 years presented with a one-week history of fever, lethargy, anorexia and generalised arthralgias, with an erythematous eruption (Figure 1) and a large plaque with eschar noted on examination (Figure 2). The patient was afebrile and haemodynamically stable, with an otherwise unremarkable examination and no organomegaly or lymphadenopathy. There is an incubation period of two to 14 days from the time of bite. Initial clinical presentation typically includes eschar at the bite site, fever, headache and a confluent erythematous eruption. If left untreated for more than one to two weeks, the disease poses some risk of pneumonitis, encephalitis, septic shock or death.2 Prolonged lethargy or fatigue, even after rash clearance, is a common symptom reported with rickettsial infection. Blood tests may reveal an elevated white cell count, acute transaminitis and elevated C-reactive protein. © The Royal Australian College of General Practitioners 2018 Queensland tick typhus (Rickettsia australis) in a man after hiking in rural Queensland his rash. Some ongoing minor lethargy persisted. Rickettsial serology was repeated at this time (four weeks after likely bite exposure) and showed an elevated R. rickettsii titre. survive by feeding off other wildlife (Figure 4), including marsupials and rodents. Stephen A Thomas, Jason Wu Stephen A Thomas, Jason Wu References 1. Appuhamy RD, Tent J, Mackenzie JS. Toponymous diseases of Australia. Med J Aust 2010;193(11−12):642–46. Figure 1. Initial clinical presentation of non-confluent erythematous plaques on upper back Figure 3. Distribution map of Queensland Tick Typhus in Australia.4 Figure 1. Initial clinical presentation of non-confluent erythematous plaques on upper back 2. Expert Group for Antibiotic. Rickettsial infections. In: eTG complete [Internet]. Melbourne: Therapeutic Guidelines Limited, 2014. Adapted with permission from Graves S. Update on Australian rickettsial infections. Surry Hills, NSW: Australasian Society for Infectious Diseases, 2017 3. health.vic. Rickettsial infections. Melbourne: Department of Health and Human Services, 2017. Available at www2.health.vic.gov.au/public-health/ infectious-diseases/disease-information-advice/ rickettsial-infection [Accessed 11 August 2017]. Figure 2. Initial clinical presentation of central eschar on broad erythematous plaque on back (suggestive of original bite site) Figure 4. Photograph of an unfed and fed tick. ‘Tick before and after feeding’ by Bjørn Christian Tørrissen, www.pvv.org/~bct/nsw/imagepages/ image53.htm. Licence at https://creativecommons. org/licenses/by-sa/3.0/ 4. Graves S. Update on Australian rickettsial infections. Surry Hills, NSW: Australasian Society for Infectious Diseases, 2017. Available at www. asid.net.au/documents/item/415 [Accessed 11 August 2017]. 4. Graves S. Update on Australian rickettsial infections. Surry Hills, NSW: Australasian Society for Infectious Diseases, 2017. Available at www. asid.net.au/documents/item/415 [Accessed 11 August 2017]. 5. Wang JM, Hudson BJ, Watts MR, et al. Diagnosis of Queensland tick typhus and African tick bite fever by PCR of lesion swabs. Emerg Infect Dis 2009;15(6):963–65. doi: 10.3201/eid1506.080855. 5. Wang JM, Hudson BJ, Watts MR, et al. Diagnosis of Queensland tick typhus and African tick bite fever by PCR of lesion swabs. Emerg Infect Dis 2009;15(6):963–65. doi: 10.3201/eid1506.080855. Figure 4. Photograph of an unfed and fed tick. ‘Tick before and after feeding’ by Bjørn Christian Tørrissen, www.pvv.org/~bct/nsw/imagepages/ image53.htm. Licence at https://creativecommons. org/licenses/by-sa/3.0/ Figure 2. Initial clinical presentation of central eschar on broad erythematous plaque on back (suggestive of original bite site) Authors Stephen A Thomas MBBS, Department of Dermatology, Princess Alexandra Hospital, Brisbane, Qld. stephenanthonythomas@hotmail.com Jason Wu FACD, Department of Dermatology, Princess Alexandra Hospital, Brisbane, Qld Competing interests: None. Provenance and peer review: Not commissioned, externally peer reviewed. Stephen A Thomas MBBS, Department of Dermatology, Princess Alexandra Hospital, Brisbane, Qld. stephenanthonythomas@hotmail.com Provenance and peer review: Not commissioned, externally peer reviewed. Provenance and peer review: Not commissioned, externally peer reviewed. Key points suspected exposure but is unlikely to alter management. • R. australis is an infection typically presenting with eschar at bite site, fevers, lethargy and arthralgias in patients who have visited endemic areas (typically rural or bush areas along the eastern seaboard of Australia). An alternative diagnostic method by polymerase chain reaction (PCR) is available nationally from the Australian Rickettsial Reference Laboratory. PCR looks for rickettsial DNA, although this has inferior diagnostic sensitivity. PCR of a skin biopsy or swab from an eschar lesion, or from the patient’s blood, may be more sensitive during days one to five of the infection.4,5 • Clinical presentations of rickettsial infection in the context of recent exposure to endemic areas should cause the physician to have a low threshold for treatment, with or without patient knowledge of a tick bite. knowledge of a tick bite. Answer 1 Rickettsia refers to a group of Gram- negative bacterial infections. Australian tick typhus, otherwise known as Queensland tick typhus (R. australis), is a commonly encountered rickettsial infection in Australia. The name, Queensland tick typhus, arose after soldiers training in Queensland during World War II were affected.1 R. australis may occur from tick exposure, typically along the east coast of Australia (Wilsons Promontory in Victoria to tropical north Queensland; Figure 3). Infection can result from the bite of an infected tick or from exposure to the faeces of infected hosts. Ticks in the affected areas generally Lesional biopsy for histopathology was performed at presentation, confirming a mononuclear vasculitis consistent with rickettsial infection, and no features of Sweet syndrome. Initial serology immunofluorescence assay was negative (<1:128 titre) for Rickettsia rickettsii (the serological test that is cross-reactive for Rickettsia australis). Other blood tests, including liver and renal function tests, were normal. Serology takes at least six days before obtaining a positive result and is currently the gold standard for testing.4 Initial negative serological studies do not rule out rickettsial infection and should not alter treatment completion in potentially infected patients. Testing could subsequently be repeated in more suspicious cases four weeks after At review two weeks later, the patient reported that most of his symptoms had resolved and demonstrated clearing of 3 REPRINTED FROM AJGP VOL. 47, NO. 6, JUNE 2018 \| 359 QUEENSLAND TICK TYPHUS CLINICAL Carins Queensland Tick Typhus R. australis Brisbane Sydney Hobart Darwin Broome Perth Adelaide Figure 1. Initial clinical presentation of non-confluent erythematous plaques on upper back Figure 3. Distribution map of Queensland Tick Typhus in Australia.4 Adapted with permission from Graves S. Update on Australian rickettsial infections. Surry Hills, NSW: Australasian Society for Infectious Diseases, 2017 Figure 1. Initial clinical presentation of non-confluent erythematous plaques on upper back Carins Queensland Tick Typhus R. australis Brisbane Sydney Hobart Darwin Broome Perth Adelaide Carins Queensland Tick Typhus R. australis Brisbane Sydney Hobart Darwin Broome Perth Adelaide correspondence ajgp@racgp.org.au © The Royal Australian College of General Practitioners 2018 Answer 4 • Serological R. rickettsii immunofluorescence is the current gold standard for testing; however, typically, this will not test positive until at least one week following tick bite exposure. PCR is a more appropriate test on days one to five after bite exposure but has inferior overall sensitivity. Initial management of suspected cases should include a seven-day course of oral doxycycline 100 mg twice daily. If allergy or contraindications to doxycycline exist, oral azithromycin 250 mg may be used once daily for seven days. Treatment response is usually rapid. correspondence ajgp@racgp.org.au 0 \| REPRINTED FROM AJGP VOL. 47, NO. 6, JUNE 2018 360 © The Royal Australian College of General Practitioners 2018
https://openalex.org/W4287748285	https://zenodo.org/records/3948386/files/08-Baker-edited-TESOL.pdf	English	null	Going Beyond Readability Formula: How Do Titles Contribute to the Readability of Essays?	Zenodo (CERN European Organization for Nuclear Research)	2,020	cc-by	7,145	Keywords y Text selection, readability, titles, Lexile, rhetorics, writing center administration Going Beyond Readability Formula: How Do Titles Contribute to the Readability of Essays? John R. Baker National Quemoy University, Taiwan Abstract The belief that genre specific reading provides numerous benefits for apprenticing writers is something we in the field of writing education value as an underlying constant. Accepting this, writing center directors stock their self-access library shelves with a variety of composition texts, to include rhetorics (and the essays therein). To select these materials, readability formulae (e.g., the Lexile Readability Formula) are often employed. However, such formulae only measure two of the many features that make up the readability of an essay (i.e., semantic, syntactic). Other important features such as the title are not considered. To address this, this article reports the results of a sequential, mixed-methods study conducted in an Asian postsecondary setting. The study found that titles influence readability both as (a) a primary (i.e., an isolated feature) and (b) a conjoined feature (i.e., consisting of two or more associated entities where the second impacts the first). The article also makes a recommendation for teachers, writing center staff, and the publishing industry that readability formulae be administered in a hybrid fashion to explore additional features such as the title when considering the difficulty of exemplars. International Journal of TESOL Studies (2020) Vol. 2 (1) 119-132 https://doi.org/10.46451/ijts.2020.06.08 International Journal of TESOL Studies (2020) Vol. 2 (1) 119-132 https://doi.org/10.46451/ijts.2020.06.08 Address: 1 University Rd. Jinning Township, Kinmen County, 892, Taiwan E-mail: drjohnrbakerr@yahoo.com Address: 1 University Rd. Jinning Township, Kinmen County, 892, Taiwan E-mail: drjohnrbakerr@yahoo.com © 2020 International TESOL Union. 1 Introduction The belief that genre specific reading provides numerous benefits for the preparation of apprenticing writers is something we in the field of writing education value as an underlying constant (Thaiss & Zawacki, 2006). That is, those who read more tend to write better (Krashen, 2004), demonstrating as much as a .50 to .70 correlation (Grabe, 2003). Following this, directors stock their self-access library shelves with a variety of composition texts, to include rhetorics (Baker, 2019), rhetorically organized anthologies of paragraphs and essays which explicate major rhetorical forms, present sample texts exemplifying major rhetorical patterns, and offer procedures to show student writers how to reproduce genre in their own writing (Ferris & Hedgcock, 2005). And like all those who select texts for prospective © 2020 International TESOL Union. All rights reserved. 120 International Journal of TESOL Studies 2 (1) (1) readers, writing center staff need to consider whether the selected texts will be a good match for potential readers (Gardner & Miller, 1999), “the study of which ... has come to be called readability” (Gilliand, 1972, p. 12). Historically, readability has been defined as the consideration of the text, the reader, and the interaction between the two (Dale & Chall, 1949; Gilliand, 1972; Harrison, 1980; Klare, 1963; Kintsch & Vipond, 1979; Kintsch & Miller, 1981; Schirmer & Lockman 2001). Chall, Bissex, Conrad, and Harris-Sharples’ (1996) metaphor-based definition is especially elucidating because they liken readability to an iceberg: beneath the water level there are “various sources of difficulty. The more difficult the passage, the greater the ice beneath” (p. 6). This definition is so appropriate because it points to the fact that readability, like an iceberg, is not one solid, homogeneously intransparent entity. Instead, it is a heterogeneous mix of features that make up what Goodman (1967) called the psycholinguistic guessing game of reading, an activity that requires a consideration of the many features that contribute to the complex phenomenon known as reading. Early explorations of this phenomenon date as far back as 900 AD (Lorge, 1944) and became more frequent in the early 20 th century (Dubay, 2007). These early explorations were mainly quantitative (focusing on semantic features, syntactic features, or a combination of the two) and laid the theoretical base for the first readability formula (Lively & Pressey, 1923), a predictive device which provides quantitative, objective estimates of difficulty (Klare, 1963). 1 Introduction Since then, formulae have employed these two features, as such features have been found to be reliable predictors of readability (Dubay, 2007). However, it was soon acknowledged that other features needed to be considered during a comprehensive assessment of the text/reader combination (see Ojemann, 1934), a concern which continues to be echoed today (Armbruster, 2016; Chall & Dale, 1995; Gunning, 2003; Fry, 2002; Kintsch & Vipond, 1979; Lexile, 2010; Meyer, 2003). To address this concern, many researchers have supported a paradigm shift toward a hybrid approach, one where a readability formula is first employed and then other factors are considered qualitatively (Fry, 2002). This is because using a readability formula and subjective criteria together reduces the risk of presenting prospective readers “a seemingly appropriate book but one they cannot read due to format, language, structure, or content” (Weaver, 2000, p. 33). The popular readability formula employed in this study (i.e., the Lexile Readability Formula) has received similar attention (Gunning, 2003; Lexile, 2010). Gunning (2003), for instance, suggested that “although teachers might use Lexiles ... , they need to go beyond the numbers ... [and] complement the objective data yielded by the formula with subjective judgment” (pp. 182-186). Revisiting Gunning’s advice, Chall et al’s (1996) iceberg reference is remarkably appropriate, especially when thinking about one feature of the essay, the title (the essay’s tip), as this is the initial signaling device readers encounter and often utilize to begin processing a text (Noor, 2006), and this strategy has been found to provide the reader with valuable insight into understanding the text (Fan & Liu, 2008). 2.2.1 Gender and age Sjogren & Timpson (1979) looking at the question of titles and schema through the lens of gender, utilized short, 145-word narratives and alternating gender biasing titles. They reported that gender specific titles helped readers identify the content of a passage that might otherwise be ambiguously read due to gender related schema. More recently, Miller, Cohen, and Wingfield (2006), adapting short materials from others (Bransford & Johnson, 1972; Dooling & Lachman, 1971) presented material with (a) a title and (b) with no title. They too confirmed that the presence of titles improves reading comprehension and added that age and working memory were related to this effect: Younger readers (ages 18–34 years) did better than middle aged readers (ages 35–59 years) who did better than older readers (ages 60–85 years). 2.1 Titles and schema Discussions regarding the effects titles have on reading comprehension in native English speaking (NS) postsecondary contexts often begin by referencing Bartlett’s (1932) pioneering work with schema. Following this, Dooling and Lachman (1971) began what would be a limited series of discussions on the topic. Extending Bartlett’s idea of schema, they presented students with a vague, short (77 word) crafted narrative in two conditions--(a) with a thematic title and (b) without a thematic title -- and reported that titles improve “students’ overall recall of texts which are vague and metaphorical” (p. 216). 121 John R. Baker John R. Baker Drawing on Dooling and Lachman (1971), Bransford and Johnson (1973) utilized a 180-word prose paragraph regarding how to wash clothes and added a third condition (providing readers access to the title after reading the text). They found that students who had access to the title before encountering the passage had higher recall whereas those who received the title after engaging the passage or not at all demonstrated lower performance. Several studies have extended Bransford and Johnson’s (1973) work in different directions. Alba, Alexander, Hasher, and Caniglia (1981), for instance, arguing that a test of comprehension was lacking, replicated Brandon and Johnsons’ study and explained that having access to the title prior to reading produced higher comprehension than in the other two conditions. Similarly, Schallert (1976) investigated the meaningfulness of titles with texts from a different procedural perspective. Using a text that could be read with alternative meanings (e.g., a baseball game/ a glass factory), he too presented the texts in each of the three conditions. However, the first condition was presented with alternate biasing titles regarding the two topics. He concluded that students recalled more as a result of receiving a stronger biasing title. He also maintained that having access to a title produced more recall than the other two conditions. 2.2 Titles and other variables Presenting a title in alternate conditions or using biasing titles has been used to explore a number of areas: titles’ relationship with (a) gender and age and (b) titles and their relationship to the text 2.3 Titles and their relationship to the text Work with titles has also focused on how titles affect NSs’ understanding of a text’s organization (Bock, 1980; Schwartz & Flammer; 1981), sentence level concepts (Smith & Swinney, 1992), and structural areas (proposition, sentence, and word level) (Wiley & Rayner, 2000). 2.3 Research with English language learners Following the work with NS undergraduates, research on the effects of titles has also been done with ELL university students, and although generally more recent, it is much less abundant. Studies with ELLs have demonstrated that NNSs, like NSs, place a great deal of importance on titles and that they too look to titles first when approaching a text (Noor, 2006), but that student proficiency (Carrell, 1983; Fan & Liu, 2008) and the meaningfulness of the title (Fan & Liu; 2008; Zhang & Hoosain, 2001) both play a part. An early work that included ELLs is Carrell (1983). Carrell adapted the materials used in Bransford and Johnson (1972) to examine the effects of titles in a North American setting with NSs (as well as Arabic, Chinese, Greek, Japanese, Korean, and Spanish-speaking ELLs) at two proficiency levels: higher and lower proficiency. She found that for both the NSs and advanced ELLs, “the presence of a title is the single best predictor of readers’ perceptions of relative ease or difficulty of comprehension” (Carrell, 1983, p. 192), as it activates students’ schema so that they can engage the text. Carrel (1983), noted, however, the same effect was not found for the lower proficient students. That is, the lower proficiency students remained linguistically tied to the text and thus were less able to take advantage of the benefits titles provide. In a more recent study, Zhang and Hoosain (2001), working with Chinese ELLS in an Asian context, also acknowledged the value of titles, but they, like Bock (1980) and Schallert (1976), looked at the content of the titles themselves. Working with short narratives, they reported that a title, “particularly a meaningful one incorporating a substantive verb ... triggers the relevant schema, and this schema generates expectations about what is likely to happen” (p. 17), the result of which is faster reading speeds and higher recall. In the most recent study found to date, Fan and Liu (2008), studied Chinese ELLs and short narratives and offered similar results about the importance of titles being meaningful (i.e., incorporating a verb), but they, like Carrell (1983), added that more proficient readers make better use of titles than less proficient ones. 2.3.2 Propositions, sentence level, and word level concepts In a second experiment with a newspaper story, Bock (1980) explored the effects of biasing titles on NSs’ identification of content words and the propositions that titles point to and found that titles can influence students’ interpretation of which content words are important. He also observed that propositions marked as important in titles are better recalled. Smith and Swinney (1992), adapting Bransford and Johnson’s (1972) materials, investigated the role titles (and the resulting schema the titles activate) play in students’ processing of sentence level concepts. They discovered that the presence of a title can improve comprehension of sentence level concepts as well as reading speed. Likewise, Wiley and Rayner (2000), using materials from other studies (Bransford & Johnson, 1972; Dooling & Lachman, 1971) found that titles can help students process texts at various structural levels: (a) proposition, (b) sentence, and (c) word levels (i.e., ambiguous words). Additionally, they reiterated that when a title is present, “the text is generally read faster, rated as more comprehensible, and recalled better” (Wiley & Rayner, 2000, p. 1,011). 2.3.1 Organization Bock (1980), working with biasing titles, presented two different biasing titles and a newspaper story cut into strips. Drawing on students’ organization of the texts, Bock concluded “titles provide the starting point for setting up text structures” (p. 308). Similarly, Schwartz and Flammer (1981) investigated the effect of the presence or absence of titles with (a) organized, (b) slightly disorganized, and (c) very disorganized 247-word narratives (fairy tales). They found that titles can help readers understand the structure of organized and slightly disorganized texts but have no impact on unstructured texts. 122 International Journal of TESOL Studies 2 (1) 3 Methods This article reports the results of a larger sequential, mixed-methods study which explored the effects 16 features have on postsecondary ELLs’ perceptions of difficulty when reading exemplars from rhetorics. Separating the study and publishing separate articles regarding each feature was done in the interest of length so as to give each unique feature’s literature review, data set, and discussion within the length of one article. The purpose of this article is to explore what benefits and difficulties titles pose for postsecondary ELL students in an Asian context (e.g. Taiwan) when they read exemplars (i.e., essays) excepted from rhetorics. 2.3 Research with English language learners The examination of the extant literature illustrates that early studies done with NSs and more recent work with ELLs point to the effects titles may have on the reading experiences of postsecondary ELL apprenticing writers in the Asian context with short texts (i.e., short crafted paragraphs, fairy tales, newspaper stories); however, empirical investigations with ELLs and longer academic texts (essays), were noticeably absent during the literature review conducted for this study. This article is intended to 123 John R. Baker fill this gap. To address this under investigated area, one research question was posed: What benefits and difficulties do titles pose for postsecondary ELLs students in an Asian context (e.g., Taiwan) when they read exemplars (i.e., essays) excepted from rhetorics? fill this gap. To address this under investigated area, one research question was posed: What benefits and difficulties do titles pose for postsecondary ELLs students in an Asian context (e.g., Taiwan) when they read exemplars (i.e., essays) excepted from rhetorics? 3.2.1 Identifying participants In keeping with the nature of qualitative theory, the informants were purposively selected to best help the researcher understand the problem (Creswell, 2013). Following Kvale’s (1996) suggestion that the num ber of informants tends to be 15 ± 10 in interview studies, a smaller cluster sample (n=14) was identified from the larger sample. The informants were identified because they received SRI scores at the top of their class (828-928L, the top 15%), which allowed them to examine a wide range of exemplars and help the researcher to holistically explore the research question (Merriam, 1991).i After the informants were identified, they were asked by e-mail if they would be willing to participate in follow-up, post-course interview. Twelve informants assented and were provided with pseudonyms to protect their anonymity: seven females (mean age 20.14 years) and five males (mean age 20.8 years). The makeup of the sample was indicated by the students’ Lexile measures which identified them as appropriate participants rather than any purposeful intent of the sampling procedure (e.g., ethnicity, gender, age). 3.2 Identifying texts and participants To identify potential texts and participants for study, three stages were performed: (a) an examination of the reading levels of the exemplars in the rhetorics, (b) an examination of the reading levels of the target sample, and (c) a comparison of the two. To identify the exemplars for study, the reading levels of exemplars (N=893) from 12 rhetorics available on the local market were examined. This was done using a non-fee based computerized version of the Lexile Analyzer available to researchers. To determine the reading Lexile level of potential participants, the SRI (Scholastic Reading Inventory) was administered to a purposive sample (N=91), students enrolled in five of the seven sections of sophomore composition at the university, as this group makes up the majority of visitors to the writing center. And to determine which exemplars are accessible to the reading levels of the selected participants, a comparison of the readability levels of the exemplars in the texts and target students’ reading levels was performed. 3.1 Overall design To collect and analyze the resulting data to answer the research question, a sequential mixed-methods approach was employed (Johnson & Onwuegbuzie, 2004). To order this mix, an adaption of Creswell’s (2013) design was employed (Figure 1). To collect and analyze the resulting data to answer the research question, a sequential mixed-methods approach was employed (Johnson & Onwuegbuzie, 2004). To order this mix, an adaption of Creswell’s (2013) design was employed (Figure 1). Figure 1. Sequential Mixed-Methods Research Design Step 1 A quantitative comparison of the texts’ readability levels (using the Lexile Readability Formula) and stu dents’ reading levels (using the Scho lastic Reading Inventory) to identify texts and participants Step 2 A qualitative exploration of the ben efits and difficulties titles pose to the students who read exemplars from rhetorics Qualitative A cline-questionnaire procedure. Qualitative A further exploration of the results of the cline-questionnaire procedure via semi-structured retrospective inter views to add scope and breadth to the findings. Step 2 A qualitative exploration of the ben efits and difficulties titles pose to the students who read exemplars from rhetorics Qualitative A further exploration of the results of the cline-questionnaire procedure via semi-structured retrospective inter views to add scope and breadth to the findings. Figure 1. Sequential Mixed-Methods Research Design Following the sequential mixed methods design, two sections (steps) and their relevant subsections are reported here: (1) Identifying Text and Participants for Study and (2) An Exploration of the Benefits and Difficulties Titles Provide Students When They Read Essays Excerpted from Rhetorics. These steps were conducted at a university in Taiwan. The university maintains a writing center and offers its 9,000 124 International Journal of TESOL Studies 2 (1) students undergraduate and graduate degrees in 16 majors, including a Bachelor of Arts in Applied English (the degree the population of this study is seeking). students undergraduate and graduate degrees in 16 majors, including a Bachelor of Arts in Applied English (the degree the population of this study is seeking). Exemplars Exemplars Hughes, L. Salvation. In S. V. Buscemi, & C. Smith (Eds.), 75 readings plus (pp. 10-14). New York, NY: McGraw-Hill. 740L McDonald, C. P. A view from the bridge. In T. Cooley (Ed.), The Norton sampler: Short essays for composition (pp. 37-41). New York, NY: Norton & Company. 810L Harris, S. Freedom and security. In G. Levin (Ed.), Prose models (pp. 389-392). Belmont, CA: Wadsworth. 910L Harris, S. Freedom and security. In G. Levin (Ed.), Prose models (pp. 389-392). Belmont, CA: Wadsworth. 910L Dalfonos, D. Grammy rewards. In T. Cooley (Ed.), The Norton sampler: Short essays for composition (pp. 206-208). New York, NY: Norton & Company. 1010L 3.3.1 The cline phase The purpose of the cline phase was to have the students read the five essays and put them in a cline of difficulty (from easiest to most difficult) so that the students would be able to reflect on this activity while completing a closed-response, Likert questionnaire. The exemplars were presented to the informants in random order and criteria for ranking was withheld in order to ensure the informants engaged in the type of decision-making process “normally used when making such judgments” (Chall, et al., 1996, p. 77). 3.3 An exploration of the benefits and difficulties titles provide students when they read essays excerpted from rhetorics Once the texts and the participants were identified, the effects of titles were explored. This was done via an untimed two-stage process: (a) a quantitative cline-questionnaire procedure and (b) qualitative semi- structured retrospective interviews. 3.2.2 Identifying exemplars Five exemplars ranging from 610-1010L (Table 1) were purposively chosen to be below, within, and slightly above the informants’ Lexile range (i.e., 828-928L). This number was chosen to allow enough variety for the informants to engage in thoughtful comparisons and small enough to be examined and discussed within a reasonable time via the cline questionnaire and interviews, reasonable in that valuable data could be gleaned but informant fatigue could be avoided. 125 John R. Baker Table 1 Exemplars Chosen for the Study Exemplars Chosen for the Study Exemplars Lexile Measures Traig, J. A Guide to proper hand-washing technique. In M. L. Conlin (Ed.), Patterns plus: A short prose reader with argumentation (pp. 176-178). New York, NY: Cengage. 610L Hughes, L. Salvation. In S. V. Buscemi, & C. Smith (Eds.), 75 readings plus (pp. 10-14). New York, NY: McGraw-Hill. 740L McDonald, C. P. A view from the bridge. In T. Cooley (Ed.), The Norton sampler: Short essays for composition (pp. 37-41). New York, NY: Norton & Company. 810L Harris, S. Freedom and security. In G. Levin (Ed.), Prose models (pp. 389-392). Belmont, CA: Wadsworth. 910L Dalfonos, D. Grammy rewards. In T. Cooley (Ed.), The Norton sampler: Short essays for composition (pp. 206-208). New York, NY: Norton & Company. 1010L 3.3.3 Interview To help the researcher triangulate the data from the questionnaire, the informants (after creating the cline and completing the questionnaire) participated in semi-structured retrospective interviews. Each interview, in accordance with Creswell (1994), began with structured questions from the questionnaire and was followed up with semi-structured prompts which later became open-ended (Nunan, 1996). A bilingual research assistant (translator) was present to assist with any language difficulties, and an observational protocol, which included both video and audio taping, was utilized to record the interviews. The interviews lasted for an average of 32.5 minutes (range 19.3 - 57.4 minutes). Variation was dependent on how much each informant had to offer and how much translation was required. After the interviews were completed, the audio recordings were transcribed, and the informants checked their transcripts. Once these steps were completed, the informants’ responses were explored using Erlandson, Harris, Skipper, and Allen’s (1993) emergent category analysis procedure, a procedure which allows “categories to follow data rather than precede them” (p. 112). “To add strength and fertility to the entire analysis” (Erlandson, Harris, Skipper, & Allen’s (1993, pp. 128-129), a second-level group debate procedure was also included 3.3.2 The questionnaire phase After the informants ordered their clines, they completed the closed-response questionnaire. This phase was administered to help the informants reflect on why they created the cline the way they did and to relate it in such a way that would provide insight into what other factors beyond readability formulae (e.g., the Lexile Readability Formula), i.e., titles, they feel influence their perceptions of difficulty when reading exemplars excerpted from rhetorics. The questionnaire addressed 16 features related to comprehension, one of which was the title. The question related to titles is listed in Table 2. International Journal of TESOL Studies 2 (1) 126 Excerpt from the Questionnaire Instructions: Now that you have ordered the texts from 1 (easiest) to 5 (most difficult), please explain your reasons for arranging the texts in the way you did by completing the questionnaire below. Titles: How well the title of each text described each text influenced my decision about how to arrange the texts in the way that I did. a. Strongly Agree b. Agree c. Neither Agree nor Disagree d. Disagree e. Strongly Disagree a. Strongly Agree b. Agree c. Neither Agree nor Disagree d. Disagree e. Strongly Disagree a. Strongly Agree b. Agree c. Neither Agree nor Disagree d. Disagree e. Strongly Disagree To ensure reliability of the questionnaire, it was translated into the students’ L1 (i.e., Mandarin) using a back-translation procedure, checked with a second translator for accuracy, and a pretest was conducted with a small number of respondents who were not part of the sample in the study (n = 2). 4.2.1 Titles and their relationship with vocabulary and vocabulary in context As a conjoined feature, two features were reported to impact the amount of influence titles have on an informant’s perception of difficulty: (a) vocabulary and (b) vocabulary in context. Two of the informants (20%) (Ben, Linda) explained that whether they could make out the meaning of the title of an essay helped them to determine how difficult an essay would be. Specifically, they explained that how difficult they found a title was influenced by how understandable the vocabulary in the title was. Ben offered an illustration, explaining that the vocabulary in the title of the essay he placed as easiest (“A Guide to Proper Hand-washing Technique”) was easier to understand than the vocabulary in the title of the other essays. He further illustrated this by explaining that he had trouble with the conceptual meaning of the phrase freedom and security which made up the title of the essay he placed last, most difficult (“Freedom and Security”). Ben’s report is in line with studies that have suggested that the meaningfulness of titles is a factor in how helpful they are (Bock, 1980; Fan & Liu, 2008; Schallert, 1976; Zhang & Hoosain, 2001). Ben also noted that he could not initially make out the title of the “Freedom and Security” essay, but that he was later able to infer its meaning from context because of the meaning related to the terms in the text. His report is associated with literature that has purported that ELL readers rely on context clues to catch the meaning of unknown words (Cooper, 1999; Fraser, 1999; Nassaji, 2003; Paribakht & Wesche, 1999). 4 Results The purpose of this article is to report the benefits and difficulties titles pose to ELL apprenticing writers when they read exemplars (i.e., essays) taken from rhetorics. The first step helped to quantitatively identify potential texts (N=5) and participants for study (N=14). Twelve assented, 11 reported to the test site, and 10 successfully completed the procedures and thus provided useful data. The second step helped to answer the research question: What benefits and difficulties do titles pose for postsecondary ELLs students in an Asian context (e.g. Taiwan) when they read exemplars (i.e., essays) excepted from rhetorics? Examining the informants’ responses, it was found that the informants, as a group, perceived titles to be both a primary (i.e., an isolated feature) and (b) a conjoined feature (i.e., consisting of two or more 127 John R. Baker associated entities where the second impacts the first). The latter being that the respondents felt that titles had a relationship with vocabulary, vocabulary in context, and logical organization. associated entities where the second impacts the first). The latter being that the respondents felt that titles had a relationship with vocabulary, vocabulary in context, and logical organization. 4.2 Conjoined features The results of the interviews further evidenced that the informants found titles to be one of 10 features conjoined with other features. Here too titles played an important but smaller role than vocabulary (70%) and sentence length (50%). Only 30% percent of the respondents perceived tiles to be influential as a conjoined feature. 4.1 Primary features The results of the interviews indicated that the informants reported the title be one of 15 primary features to affect the readability of the exemplars in the rhetorics, albeit the title took a less prominent role in comparison to the first feature measured by the Lexile Readability Formula (i.e., semantic). Nine of the 10 informants (90%), for example, cited vocabulary and vocabulary in context as influential whereas 40% reported titles to be so. Conversely, the title took a more prominent role than the second feature utilized by readability formulae (i.e., syntactic—sentence length). Forty of the informants reported the title to be influential as a primary feature whereas only 10% found sentence length to be so. This surprising result can be accounted for in that the overall impact of sentence length cannot be considered only as a primary feature. Its relevance is also in relation to its conjoined properties (e.g., grammar, vocabulary, punctuation) (Dubay, 2007). 5 Conclusions and Discussion Accepting that readability is an important and current field of study (Mesmer, 2008), yet the readability of essay exemplars found in rhetorics is an understudied area in both L1 and L2 literature (Baker, 2019), this article explored what effects titles have on ELL apprenticing writers’ perceptions of text difficulty when reading exemplars excerpted from rhetorics. The results indicated that titles impact readability as (a) a primary (isolated feature) and (b) a conjoined feature (consisting of two or more associated entities where the second effects the first, i.e., a feature influences the impact of titles or titles influence the impact of another feature). These findings further literature which has reported the effects titles have on the readability of other types of materials (e.g., short crafted narratives, fairy tales, newspaper articles). As a primary factor, for instance, the findings corroborate Noor (2006) who reported that NNS, like NSs place a great deal of importance on the title when approaching a text. The findings also validate the views of Dubay (2007) and others who argue that the relevance of the title is related to conjoined features. The results regarding the conjoining feature vocabulary, for instance, are consistent with explorations done in both L1 (Miller, Cohen, & Wingfield, 2006) and L2 settings (Carrell, 1983; Fan & Liu; 2008; Zhang & Hoosain, 2001) that found that whether the reader can make out the meaning of the title is a good predictor of a reader’s perception of text difficulty. Work with ELLs and vocabulary in context is also furthered (Cooper, 1999; Fraser, 1999; Nassaji, 2003; Paribakht & Wesche, 1999), as this study demonstrates that ELL readers rely on context clues to infer the meaning of unknown words in titles. The findings regarding the influence of titles on logical organization additionally advances the findings of Bock (1980) who reported that that titles can help readers anticipate a text’s logical organization. Taken together, the data presented in this article provides a unique contribution to the literature, as the study demonstrates how titles contribute to the readability of essays taken from rhetorics when read by ELL apprenticing writers. Specifically, it was found that that the title is an important contributing feature but one that is less influential than the two features employed in quantitative readability formulae (i.e., semantic, syntactic), and therefore the title needs to be considered when holistically thinking about readability and selecting essays from rhetorics. 4.2.2 Titles and their relationship with logical organization Titles were also reported to be related to the amount of difficulty caused by logical organization. Dan explained why he felt that the title of the “A Guide to Proper Hand-washing Technique” essay (the one he positioned as easiest) was helpful. He reported that the essay’s title assisted him in understanding the essay’s logical organization because he could anticipate the essay would be reporting a procedure. Dan’s 128 International Journal of TESOL Studies 2 (1) report is in accordance with Bock’s (1980) work which explained that titles can help informants set up ideas about forthcoming logical organization. 5 Conclusions and Discussion its relation to readability has received a moderate amount of historically relevant attention in the North American context, attention which prematurely ceased in the late 1980s, and a limited amount of work in L2 studies thereafter, yet it is still an under-researched area in Asian and non-North American contexts. Thus, as this empirical study explored only one Asian context (Taiwan), but purposively provides an exhaustive literature review and detailed methodology section, it marks a starting point for replication studies and further discussions of the importance of the title outside the North American context. 5 Conclusions and Discussion This result extends the work of Weaver (2000) who explains that using readability formulae and subjective criteria together reduces the risk of presenting students a seemingly appropriate text (based on the results of readability formulae) but one that they cannot read due to intervening effects of other features. The results further extend the stance taken by Gunning (2003) and others (Chall & Dale, 1995; Fry, 2002; 2003; Lexile, 2010; Meyer, 2003; Zakaluk & Samuels, 1988) that quantitative readability formulae are a good starting point for readability investigations, but that the other features formulae do not measure (e.g., titles) need to be considered in a sequential fashion: First employ quantitative readability and then qualitatively consider the influence of subject features readability formulae do not measure.f Regarding implications, the results presented in this article offer practical guidance for instructors, writing center staff, and the research community as a whole, as the teaching of writing (see Devanadera, 2018), writing centers (Paiz, 2017), and writing center self-access libraries (Baker, 2018) are becoming a welcoming commonplace in L2 settings. The data is also of use to members of the publishing industry as text selection is an ongoing concern (Baker, 2019) and reading difficulty has been cited as a factor to be considered when including exemplars in the 200 plus rhetorics published each year (Bloom, 1999). The resulting data, however, also raises additional questions which merit investigation, one of which is that in-depth discussions of other features that contribute to readability of essays found in rhetorics still need to be had (Baker, 2019). Another question related to this is a broader potential focus. That is, the title and 129 John R. Baker its relation to readability has received a moderate amount of historically relevant attention in the North American context, attention which prematurely ceased in the late 1980s, and a limited amount of work in L2 studies thereafter, yet it is still an under-researched area in Asian and non-North American contexts. Thus, as this empirical study explored only one Asian context (Taiwan), but purposively provides an exhaustive literature review and detailed methodology section, it marks a starting point for replication studies and further discussions of the importance of the title outside the North American context. References Alba, J. W., Alexander, S. G., Hasher, L., & Caniglia, K. (1981). The role of context in the encoding of information. Journal of Experimental Psychology: Human Learning and Memory, 7(4), 283-292. doi:10.1037/02787393.7.4.283 Armbruster, B.B. (2016). Matching readers and texts: The continuing quest. In Content area reading and learning (pp. 47-64). New York, NY: Routledge. Baker, J. R. (2018). Writing about the writing center: Exploring what factors motivate writing center usage outside the North American context. Asian ESP Journal, 14(5), 7-56. Retrieved http://asian- esp-journal.com/ Baker, J. (2019). Writing about the writing center in the Asian context: Exploring the mis/match between the reading levels of self-access materials and the students who visit the center. Asian ESP Journal, 15(3), 256-285. Bartlett, F. C. (1932). Remembering: A study in experimental and social psychology. Cambridge, England: Cambridge University Press. Bloom, L. (1999). The essay canon. College English, 61(4), 401-430. doi:10.1016/S8755- 4615(89)80016-7 Bock, M. (1980). Some effects of titles on building and recalling text structures. Discourse Processes, 3(4), 301-311. doi:1080/01638538009544494 Bransford, J. D., & Johnson, M. K. (1972). Contextual prerequisites for understanding: Some investigations of comprehension and recall. Journal of Verbal Learning and Verbal behavior, 11(6), 717-726. doi:10.1016/S00225371(72)80006-9 Bransford, J. D., & Johnson, M. K. (1973). Considerations of some problems of comprehension. In W. G. Chase (Ed.), Visual information processing. (pp. 383-439), New York, NY: Academic Press. Carrell, P. L. (1983). Three components of background knowledge in reading comprehension. Language learning, 33(2), 183-203. doi10.1111/j.1467-1770.1983.tb00534.x Chall, J. S., Bissex, G. L., Conrad, S. S., & Harris-Sharples, S. (1996). Qualitative assessment of text difficulty. Cambridge, MA: Brookline Books. Chall, J.S., & Dale, E. (1995). Readability revisited: The new Dale-Chall readability formula. Cambridge, MA: Brookline Books. Cooper, T. C. (1999). Processing of idioms by L2 learners of English. TESOL Quarterly, 33(2), 233-262. doi:10.2307/3587719 Creswell, J. W. (1994). Research design: Qualitative and quantitative approaches. Thousand Oaks, CA: Sage. Creswell, J. W. (2013). Research design: Qualitative, quantitative, and mixed methods approaches (4th ed.). Thousand Oaks, CA: Sage. Dale, E., & Chall, J. S. (1949). The concept of readability. Elementary English, 26, 19-26. doi:10.1111/ j.1467-9345.1968.tb00749.x International Journal of TESOL Studies 2 (1) 130 Dalfonos, D. (2003). Grammy rewards. In T. Cooley (Ed.). The Norton sampler: Short essays for composition (pp. 206-208). New York, NY: Norton & Company. Devanadera, A. C. (2018) Assessing Vietnamese EFL students’ writing in the light of World Englishes The Journal of English as and International Language, 13(2), 88-105. References Retrieved from https://www. elejournals.com Dooling, D. J., & Lachman, R. (1971). Effects of comprehension on retention of prose. Journal of Experimental Psychology, 88(2), 216-222. doi:10.1037/h0030904 Erlandson, D. A., Harris, E. L., Skipper, B. L., & Allen, S. D. (1993). Doing naturalistic inquiry: A guide to methods. Newbury Park, CA: Sage. Fan, L., & Liu, Z. Q. (2008). The effects of the title of a narrative text on Chinese EFL learners online thematic inferential processing. Foreign Language Research, 4(143), 820-827. Retrieved from http://en.cnki.com.cn/ p http://en.cnki.com.cn/ Fraser, C. A. (1999). Lexical processing strategy use and vocabulary learning through reading. Studies in Second Language Acquisition, 21(2), 225-241. doi:10.1017/S0272263199002041 Fry, E. (2002). Readability versus leveling: Both of these procedures can help teachers select books for readers at different stages. The Reading Teacher, 56(3), 286-292. Retrieved from http://www.jstor. org/stable/20205195 Gardner, D., & Miller. L. (1999). Establishing self-access: From theory to practice. New York, NY: Cambridge University Press. Gilliand, G. (1972). Readability. London, England: Hodder and Stoughton. Goodman, K. S. (1967). Reading: A psycholinguistic guessing game. Journal of the Reading Specialist, 6(4), 126-135. doi:10.1080/19388076709556976 Grabe, W. (2003). Reading and writing relations: Second language perspectives on research and practice. In B. Kroll (Ed.), Exploring the dynamics of second language writing (pp.242-262). Cambridge: Cambridge University Press. Gunning, T. G. (2003). The role of readability in today’s classrooms. Top Lang Disorders, 23(3), 175- 189. Retrieved from http://journals.lww.com/topicsinlanguagedisorders Harris, S. (2001). Freedom and security. In G. Levin (Ed.), Prose models (4th ed.) (pp. 389-392). Belmont, CA: Wadsworth. Harrison, C. (1980). Readability in the classroom. London, England: Cambridge Universit Hughes, L. (2007). Salvation. In S. V. Buscemi, & C. Smith (Eds.), 75 readings plus (pp. 10-14). New York, NY: McGraw-Hill. Johnson, R. B., & Onwuegbuzie, A. J. (2004). Mixed methods research: A research paradigm whose time has come. Educational Researcher, 33(7), 14-26. doi:10.3102/0013189X033007014 Kintsch, W., & Miller, J. R. (1981). Readability: A view from cognitive psychology. In J. Flood (Ed.), Understanding reading comprehension (pp. 220-232). Newark, DE: International Reading Association. Kintsch, W., & Vipond, D. (1979). Reading comprehension and readability in educational practice and psychological theory. In L. G. Nilsson (Ed.), Perspectives on memory research (pp. 329-366). Hillsdale, NJ: Lawrence Erlbaum Associates. Klare, G. (1963). The measurement of readability. Ames, Iowa: Iowa State Press. Krashen, S. (2004). The power of reading. Portsmouth, NH: Heinemann. Krashen, S. (2004). The power of reading. Portsmouth, NH: Heinemann. Kvale, S. (1996). Interviews: An introduction to qualitative research interviewing. Thousand Oaks, CA: Sage. Lexile. (2010). Lexile Professional Anaylzer. Retrieved from http://lexile.com/analyzer 131 John R. Baker Lively, B., & Pressey, S. (1923). A method for measuring the vocabulary burden of textbooks. Educational Administration and Supervision, 9, 389-398. Lorge, I. (1944). Word lists as background for communication. Teachers College Record, 45(8), 543-552. Retrieved from http://www.tcrecord.org McDonald, C. P. (2003). A view from the bridge. In T. Cooley (Ed.). The Norton sampler: Short essays for composition (pp. 37-41). p http://en.cnki.com.cn/ New York, NY: Norton & Company. Merriam, S. B. (1991). Case study research in education: A qualitative approach. San Francisco, CA: Jossey-Bass. Mesmer, H. (2008). Tools for matching readers to texts. New York, NY: Guilford Press. Meyer, B. J. F. (2003). Text coherence and readability. Topics in Language Disorders, 23(3), 204-224. do i:10.1097/0001136320030700000007 Miller, L. M. S., Cohen, J. A., & Wingfield, A. (2006). Contextual knowledge reduces demands on working memory during reading. Memory & Cognition, 34(6), 1355-1367. doi:10.3758/ BF03193277 Nassaji, H. (2003). L2 vocabulary learning from context: Strategies, knowledge sources, and their relationship with success in l2 lexical inferencing. TESOL Quarterly, 37(4), 645-670. Retrieved from http://www.jstor.org/stable/3588216 Noor, N. M. (2006). Reading academic text: Awareness and experiences among university ESL learners. GEMA online journal of language studies, 6(2), 65-78. Retrieved from http://journalarticle.ukm. my/754/ Nunan, D. (1996). Towards autonomous learning: Some theoretical, empirical and practical issues. In R. Pemberton, E. S. L., Li, W. W. F. Or, & H. D. Pierson (Eds.), Taking control: Autonomy in language learning (pp. 13-26). Hong Kong: Hong Kong University Press. Ojemann, R. (1934). The reading ability of parents and factors associated with reading difficulty of parent education materials. Researches in Parent Education, 8, 11-32. Paiz, J.M. (2017). Uses of and attitudes towards owls as L2 writing support tools. The Asian EFL Journal Quarterly, 19(1), 56-80. Retrieved from www.researchgate.net/publication/313994344 Paribakht, T. S., & Wesche, M. (1999). Reading and “incidental” L2 vocabulary acquisition: An introspective study of lexical inferencing. Studies in Second Language Acquisition, 21(2), 195-224. https://doi.org/10.1017/S027226319900203X Phillip G., & Queroda, P. G. (2018) Theme writing skills of Pangasinan State University (PSU) education. TESOL International Journal, 13(3), 31-44. Retrieved from https://www.elejournals.com Schallert, D. L. (1976). Improving memory for prose: The relationship between depth of processing and context. Journal of Verbal Learning and Verbal Behavior, 15(6), 621-632. doi. org/10.1016/00225371(76)900554 Schirmer, A., & Lockman, B. R. (2001). How do I find a book to read? Middle and high school students use a rubric for self-selecting material for independent reading. Teaching Exceptional Children, 34(1), 36-42. doi:10.1177/004005990103400105 Schwartz, M., & Flammer, A. (1981). Text structure and title-effects on comprehension and recall. Journal of Verbal Learning and Verbal Behavior, 20(3), 61-66. doi:10.1016/S00225371(81)90301- 7 Sjogren, D., & Timpson, W. (1979). Frameworks for comprehending discourse: A replication study. American Educational Research Journal, 16(4), 341-346. doi:10.2307/1162827 Smith, E. E., & Swinney, D. A. (1992). The role of schemas in reading text: A real-time examination. p http://en.cnki.com.cn/ Discourse Processes, 15(3), 303-316. doi:10.1080/01638539209544814 International Journal of TESOL Studies 2 (1) 132 Thaiss, C., & Zawacki, M. (2006). Engaged writers and academic writing life. Portsmouth, NH: Boynton/Cook. Thaiss, C., & Zawacki, M. (2006). Engaged writers and academic writing life. Portsmouth, NH: Boynton/Cook. Traig, J. (2008). A guide to proper hand-washing technique. In M. L. Conlin (Ed.). Patterns plus: A short prose reader with argumentation (9th ed.) (pp. 176-178). New York, NY: Houghton Mifflin. Weaver, B. M. (2000). Leveling books K–6: Matching readers to text. Newark, DE: International Reading Association. Wiley, J., & Rayner, K. (2000). Effects of titles on the processing of text and lexically ambiguous words: Evidence from eye movements. Memory & Cognition, 28(6), 1011-1021. doi:10.3758/BF03209349 Zakaluk, B. L., & Samuels, S. J. (1988). Toward a new approach to predicting text comprehensibility. In B. L. Zakaluk & S. J. Samuels (Eds.), Readability: Its past, present, and future (pp. 121-140). Newark, DE: International Reading Association. Zhang, H. and Hoosain, R., 2001. The influence of narrative text characteristics on thematic inference during reading. Journal of Research in Reading, 24(2), 173-186. doi:10.1111/1467-9817.00140 John R. Baker has taught in the U.S.A. and Asia (Korea, Taiwan, and Vietnam) and has been an editor and reviewer for several journals and publishers. His research interests include second language writing and reading, self-access and writing center administration, various literature interests, and how these come together in an interdisciplinary nature.
https://openalex.org/W4200036993	https://discovery.ucl.ac.uk/id/eprint/10159254/1/Soft%20Self-Healing%20Fluidic%20Tactile%20Sensors%20with%20Damage%20Detection%20and%20Localization%20Abilities.%20.pdf	English	null	Soft Self-Healing Fluidic Tactile Sensors with Damage Detection and Localization Abilities	Sensors	2,021	cc-by	7,354	Keywords: soft robotic sensors; self-healing sensors; ﬂuidic sensing; damage detection Citation: George Thuruthel, T.; Bosman, A.; Hughes, J.; Iida, F. Soft Self-Healing Fluidic Tactile Sensors with Damage Detection and Localization Abilities. Sensors 2021, 21, 8284. https://doi.org/10.3390/ s21248284 Article Soft Self-Healing Fluidic Tactile Sensors with Damage Detection and Localization Abilities Thomas George Thuruthel 1,, Anton W. Bosman 2 , Josie Hughes 1 and Fumiya Iida 1 1 Bio-Inspired Robotics Laboratory, Department of Engineering, University of Cambridge, Cambridge CB2 1PZ, UK; jaeh2@cam.ac.uk (J.H.); ﬁ224@cam.ac.uk (F.I.) 2 SupraPolix BV, Horsten 1, 5612 AX Eindhoven, The Netherlands; bosman@suprapolix.com Correspondence: tg444@cam.ac.uk 1 Bio-Inspired Robotics Laboratory, Department of Engineering, University of Cambridge, Cambridge CB2 1PZ, UK; jaeh2@cam.ac.uk (J.H.); ﬁ224@cam.ac.uk (F.I.) 2 SupraPolix BV, Horsten 1, 5612 AX Eindhoven, The Netherlands; bosman@suprapolix.com * Correspondence: tg444@cam.ac.uk Abstract: Self-healing sensors have the potential to increase the lifespan of existing sensing tech- nologies, especially in soft robotic and wearable applications. Furthermore, they could bestow additional functionality to the sensing system because of their self-healing ability. This paper presents the design for a self-healing sensor that can be used for damage detection and localization in a continuous manner. The soft sensor can recover full functionality almost instantaneously at room temperature, making the healing process fully autonomous. The working principle of the sensor is based on the measurement of air pressure inside enclosed chambers, making the fabrication and the modeling of the sensors easy. We characterize the force sensing abilities of the proposed sensor and perform damage detection and localization over a one-dimensional and two-dimensional surface using multilateration techniques. The proposed solution is highly scalable, easy-to-build, cheap and even applicable for multi-damage detection. Citation: George Thuruthel, T.; Bosman, A.; Hughes, J.; Iida, F. Soft Self-Healing Fluidic Tactile Sensors with Damage Detection and Localization Abilities. Sensors 2021, 21, 8284. https://doi.org/10.3390/ s21248284 sensors sensors 1. Introduction Self-healing elastomeric polymers can provide improved performances and novel functionalities to existing soft robotic and wearable systems [1–3]. They are primarily used for functional recovery of physical [4–6] and electrical properties [7–9]. Among them, self- healing soft sensory systems are of particular interest due to their potential applications [10]. However, there are numerous challenges in developing smart self-healing materials with the desired sensing properties while being able to repair-and-recover their functionality after a damage cycle in a fast and autonomous manner. This work presents a novel self- healing elastomer with rapid and autonomous self-healing capabilities combined with an intelligent ﬂuidic sensing architecture. Our methodology allows us to develop fast- recovering self-healing soft sensors with tunable sensing properties and the ability to detect and localize damage in a continuous manner. Academic Editor: Seokheun Choi Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional afﬁl- iations. Soft robotic sensors are vital for obtaining information about internal physical states in a non-intrusive manner [11,12]. Typically, they are designed with materials that change their electrical properties in response to strain, stress, pressure, temperature, etc. [13]. Currently, self-healing soft sensors are made with composites of self-healing polymers and the addition of conductive metallic [14] or carbon particles [15–17]. There have been impressive demonstrations where 90% functional recovery was obtained after 15 s of healing at ambient temperatures [14]. Recent works have looked into improved sens- ing/healing properties [18–20], physical/optical properties [21], 3D printability [7] and bio-compatibility [22,23]. Nonetheless, the addition of conductive particles to the SH polymer leads to suboptimal mechanical and healing performances. Moreover, certain manual effort is required to bring damaged components together and align them. Here, we present a soft sensor design based on encapsulated ﬂuid bodies that allows us to de- velop self-healing sensors with only the non-conductive SH polymers (see Figure 1a). This Copyright: © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). https://www.mdpi.com/journal/sensors Sensors 2021, 21, 8284. https://doi.org/10.3390/s21248284 Sensors 2021, 21, 8284 2 of 12 allows us to maintain the physical properties of the base SH polymer. Additionally, with the help of a supramolecular self-healing material, we can obtain almost instantaneous self-healing even at room temperature. 1. Introduction This autonomous self-healing material consists of supramolecular polymers that comprise speciﬁc hydrogen-bonding arrays based on ureidopyrimidones [24], which combine high strength with a highly dynamic nature and therefore give self-healing properties to the materials [25,26]. Their ease of synthesis, processing and biocompatible nature make these supramolecular self-healing materials eminently suitable for their application in soft robotics [27]. With the appropriate design of the ﬂuidic volume (highly concave internal surfaces), we can ensure that internal stresses created after damage will autonomously align and heal the soft sensor. As the sensors are made with only one material, fabrication also becomes easier. Similar design concepts have been used to develop soft sensors, albeit without the self-healing capabilities [28–30]. Due to the unique spectral characteristics of pressure signals in enclosed volumes that undergo damage and the relatively slow speed of pressure signals (speed of sound in the medium), our SH sensor design can be adapted to detect and localize damage in a continuous manner. Damage detection abilities are vital for monitoring the integrity of the surrounding system and for safer interactions. Conventional methods have used specialized tools for damage detection and localization. They include techniques, such as visual inspection, X-ray radiography, ultrasonics, etc. [31–34]. However, these devices are specially designed for damage detection without additional sensing capabilities. Additionally, these methods are difﬁcult to be transferred to soft-bodied systems in a cheap and practical way. Markvicka et al. have recently demonstrated a soft sensory skin that can detect and localize damage based on soft composite material with liquid metal droplets [35]. They demonstrate how such capabilities can facilitate an intelligent response to mechanical damages. However, their sensory skin did not have self-healing capabilities and requires a 2D array of parallel sensory ﬁbers for damage detection and localization, which restricts their applicability and scalability. More recent works have looked into incorporating self-healing with damage detection and localization [36]; however, they still have all the disadvantages of composites of SH polymers and conductive particles. Moreover, the damage localization is discrete, requiring one sensor for detection of damage for each pre-deﬁned location. Continuous localization capabilities can be obtained by extending the proposed sys- tem by measuring the pressure signals inside the enclosed chamber through different pathways. 2.1. SH-Material and Fabrication The self-healing supramolecular elastomer was obtained in a process described pre- viously from 2 to amino-4-hydroxy-6-methyl-pyrimidine, 4,4‚-methylenebis(cyclohexyl isocyanate) and poly(tetramethylene oxide) (Mn = 1000) [37], resulting in a telechelic polyurethane with ureidopyrimidone end groups and having a number average molar mass (Mn) of 20 kDa and a mass average molar mass (Mw) of 40 kDa (SEC in THF against PSt standards). The isolated polymer was subsequently processed into a clear ﬁlm with thicknesses in the range of 0.5–1.0 mm by using a hydraulic laboratory press from Fontijne Press (Delft, the Netherlands) at 120 °C and 150 N. Once the SH material is formed into ﬁlms, the sensors can be developed using com- pression molding techniques (Figure 1a). The inverse mold of the sensor is 3D printed using heat resistance ABS. The SH ﬁlms are placed on the mold, locally heated using a heat gun set at 100 degree Celsius and molded to the desired shape by applying pressure. Once the material is cooled down, the parts are removed from the mold. The connecting tube is added at this stage, and complementary parts can be attached by heating the open surfaces for a short duration and bring the surfaces together. This creates a leak-proof binding. The interface between the SH material and the non-SH connecting tube can have leaks due to impedance mismatch, in which case, the leaks are plugged using silicone glue. For both the damage detecting sensors, the non-SH connecting tube passes through the entire chamber for ease of fabrication and structural stability of the inner chamber. This also prevents the SH material from adhering onto itself during fabrication. As the connecting tube is thin, it does not affect the performance of the sensor after each damage-and-heal cycle. (a) 0 381 718 1250 1450 Strain (%) 0 0.5 1 1.5 2 2.5 3 Stress (MPa) Original 5 Minutes 1 Hour 20 Hours Fracture at new location (b) m N N O N N O H H H N H O O N N O N N O H H H N H O O O n (c) A B D C (d) Figure 1. Working principle of the self-healing ﬂuidic soft sensor. (a) Design of the ﬂuidic sensor. The curvature of the chamber and the elasticity of the SH material ensures that the structural integrity of the ﬂuidic chamber is maintained. 1. Introduction Since pressure signals travel at the speed of sound in the medium, we can employ multilateration techniques for localizing the source of the pressure signal based on the time-of-ﬂight of the pressure signal. As the pressure signals are generated by the contact or damage event itself, an additional energy source is not required for localization, unlike typical multilateration systems. Due to the impedance mismatch between the ﬂuidic chamber and the surrounding self-healing material, the localization sensor can be scaled to any complex geometry. This work presents a soft sensor design based on encapsulated ﬂuid bodies that al- lows us to develop self-healing sensors with only the non-conductive supramolecular SH polymers. This allows us to obtain almost instantaneous self-healing even at room temperature. Due to the compliance of the material and its biocompatible nature, these sensors are well suited for soft robotic applications. We investigate the self-healing charac- teristics of the material experimentally and the sensing properties using a ﬁnite element model. Finally, we demonstrate the applicability of the sensor for damage detection and localization in a one-dimensional and two-dimensional sensing surface. This work is the ﬁrst to demonstrate a soft self-healing sensor that can detect and localize damage on a continuous surface. 3 of 12 Sensors 2021, 21, 8284 2.3. Experimental Setup For measuring the pressure inside the chamber, we use the MPXH6400A Absolute, Integrated Pressure Sensor. The analog pressure values from the sensor are read using a National Instruments (NI) USB 6212 data acquisition system. The analog signals are then sampled at 200 KHz with a 16-bit resolution. For measuring the applied forces, we use the ATI Nano43 6-axis force sensor. The analog signals from the force sensors are ampliﬁed using the NI FTIFPS1 ampliﬁer and then read by the NI USB-6212 ADC. The data from the data acquisition system is read through the serial port and processed in MATLAB. The indentation probe is controlled using a UR5 robotic manipulator. 2.4. Data Processing All the data processing of the pressure signals is conducted on MATLAB. For character- izing the sensor properties, the sensor signals are read through the serial port on-demand. For detecting and localizing damage, the sensor signals are read continuously at 200 KHz for a total duration of 10 s. The raw signals are then ﬁltered using a bandpass ﬁlter (using MATLAB bandpass function) with a passband frequency range of 150–1000 Hz for both the sensor morphologies. The steepness of the ﬁlter was set at 0.8, and the stopband attenuation was set to 50 dB. The ﬁltered signal is further smoothened using a simple thresholding method to remove low decibel noise. The MATLAB function ﬁnddelay is used for measuring the lag between the two pressure signals. For obtaining the pathway of the 2D sensor morphology, we use some standard image processing tools. First, a picture of the sensor is taken, and the pathway is manually traced on picture. A binary image of this is obtained after converting the picture to an HSV colormap. The MATLAB functions imregionalmin and bwskel are then used to trace the thinnest connected pathway in the binary image. This pathway in the image space is then calibrated with respect to the real system by using the total length of the ﬂuidic tube and its corresponding pixel length. 2.1. SH-Material and Fabrication (b) Healing characteristics of the SH elastomeric polymer at ambient conditions. Note that complete healing is not required for full functional recovery of the soft sensor. (c) Reversible hydrogen-bonding arrays responsible for the self-healing of the supramolecular polymer together with its molecular structure. (d) Instantaneous healing of the SH polymer under ambient conditions. (a) 0 381 718 1250 1450 Strain (%) 0 0.5 1 1.5 2 2.5 3 Stress (MPa) Original 5 Minutes 1 Hour 20 Hours Fracture at new location (b) (a) (b) m N N O N N O H H H N H O O N N O N N O H H H N H O O O n (c) A B D C (d) (c) (d) Figure 1. Working principle of the self-healing ﬂuidic soft sensor. (a) Design of the ﬂuidic sensor. The curvature of the chamber and the elasticity of the SH material ensures that the structural integrity of the ﬂuidic chamber is maintained. (b) Healing characteristics of the SH elastomeric polymer at ambient conditions. Note that complete healing is not required for full functional recovery of the soft sensor. (c) Reversible hydrogen-bonding arrays responsible for the self-healing of the supramolecular polymer together with its molecular structure. (d) Instantaneous healing of the SH polymer under ambient conditions. 4 of 12 Sensors 2021, 21, 8284 2.2. Working Principle 2.2. Working Principle The working principle of the sensor is based on the transmission of external forces through pressure waves inside the enclosed ﬂudic chamber (Figure 1a). These pressure signals can be measured away from the contact location using a commercial pressure sensor. The morphology of the chamber determines the sensitivity of the sensor to the external stimulation, which is investigated in Section 3.1. The self-healing of the material happens because of the reversible hydrogen-bonds present in the polymer. Damage detection in the ﬂudic chamber is feasible because of its characteristic frequency response. As damage is associated with a sharp drop is pressure, this creates pressure waves with a particular high frequency component. This component is a function of the chamber morphology. With high enough sampling frequnecy this component can be easily detected. Damage localization is conducted with multi-lateration techniques. As pressure waves travel at the speed of sound inside the chamber, by looking at the time-of-ﬂight differences at two different pressure sensors, the location of the pressure source can be localized. We assume that the pressure waves travel along the shortest path with the effects of reﬂections neglected. 3.1. Sensor Characterization The single-output hemispherical sensor schematically represented in Figure 1a can be used as a simple soft force sensor. It must, however, be pointed out that being a soft sensor, the ﬂuidic sensor will respond to multiple physical cues. Hence, discerning the applied force in one direction using a single sensor is not possible without simpli- fying assumptions [12,38]. For practical applications, a large array of these sensors are required to decouple and estimate applied forces without constraints. The setup used for measuring the response of the soft sensor is shown in Figure 2a. The ﬂuidic chamber is a hemisphere of 6 mm diameter, and the enclosing surface is a cuboid of dimension 10 mm × 10 mm × 5 mm. The pressure response of the sensor to a periodic indentation to constant height is shown in Figure 2c along with the measured vertical forces. We can see that the sensor response to the applied force is highly repeatable and the pressure response is in sync with applied force, at least during the application of force. Upon removal of the load, there appears to be a slow return to the baseline pressure due to the viscous effects of the material, which delays the return of the sensor to its original geometry. There is also a drift in the baseline pressure possibly due to strain relaxation in the material. These temporal nonlinearites can however be compensated by recent advancements in learning-based techniques [38,39]. By tuning the geometric parameters of the sensor, the response of the sensor to the applied force can be tuned. The relation between applied force to the measure pressure is analyzed using an Ansys model (Figure 2d), and the results are shown in Figure 2e. As the diameter of the hemisphere increases with respect to the enclosing surface, we can obtain higher sensitivity to applied vertical forces. This sensitivity is independent of the material stiffness, viscosity and Poisson’s ratio as veriﬁed through the simulation. Finally, the resilience of the sensor to damage is studied. For this, we apply constant and impulsive indentation to the sensor. Impulsive forces are used to ensure that the enclosure is tightly sealed after damage. The results from this test are shown in Figure 2b. Upon ﬁrst contact, we can see a sudden spike in the pressure and force value due to the impulse forces. 3.1. Sensor Characterization Small oscillations can also be observed due to the oscillations of the robotic arm to which the indenter is attached to. For this step signal, we can see that the internal pressure quickly settles to a constant, while the applied force on the force sensor slowly settles to a constant. This indicates that the stress relaxation occurring in the material while the force is applied does not change the geometry of the deformed chamber; hence, the pressure remains constant. The sensor is damaged using a surgical knife around the 140 s mark. We can observe that on the next indentation, the pressure is still maintained at a constant value even though the material had only few seconds to heal. No indication of a leak is observed here. There is however a small shift in the peak pressure, indicating that an additional calibration process might be required after damage. 3. Results We study the capabilities of our proposed sensing system using three experimen- tal scenarios. In the ﬁrst scenario, we characterize the performance of the single-output ﬂuidic sensor and investigate its potential as a force sensor. In the second and third experi- mental scenarios, we investigate and study the performance of the multi-output ﬂuidic sensor for damage detection and localization in a one-dimensional and two-dimensional surface, respectively. 5 of 12 Sensors 2021, 21, 8284 3.1. Sensor Characterization 3.2. One-Dimensional Damage Detection and Localization The damage detection and localization capabilities of the multi-output self-healing sensor is investigated along one dimension ﬁrst. The sensory system and its schematic is shown in Figure 3a. Any external contact on the sensor creates pressure waves that travel at the speed of sound, originating from the point of contact. The pressure waves will travel along the path of the least resistance. The signals are severely attenuated when it travels from one medium to the other. Hence, by measuring the time difference between the arrival of the pressure signal at the two pressure sensors and knowing the path of the pressure signal, we can triangulate the location of contact. The time-of-arrival difference (t2 −t1) can be obtained as: t2 −t1 = L/2 + x vel −L/2 −x vel t2 −t1 = 2x vel t2 −t1 = L/2 + x vel −L/2 −x vel Sensors 2021, 21, 8284 6 of 12 where L is the total length of the air cavity between the ends of the two pressure sensors, x is the distance of contact from the centre of the air cavity, and vel is the speed of sound in the medium. For all our experiments, we use the speed of sound as 33,100 cm/s, which is the speed of sound at zero degrees Celsius. For the resolution of our setup (0.165 cm), a 100% error in the estimate of the sound speed would only lead to an error bias of 0.165 cm. where L is the total length of the air cavity between the ends of the two pressure sensors, x is the distance of contact from the centre of the air cavity, and vel is the speed of sound in the medium. For all our experiments, we use the speed of sound as 33,100 cm/s, which is the speed of sound at zero degrees Celsius. For the resolution of our setup (0.165 cm), a 100% error in the estimate of the sound speed would only lead to an error bias of 0.165 cm. 3.2. One-Dimensional Damage Detection and Localization (a) 100 110 120 130 140 150 160 170 180 190 200 Time(sec) 0 2 4 6 8 10 12 Force(N) 100 105 110 115 Pressure(KPa) Sensor Damage (b) 0 50 100 150 200 250 300 Time (sec) 0 1 2 3 4 5 6 7 8 9 10 Force (N) 98 100 102 104 106 108 110 Pressure (KPa) (c) CONTPRES Subject: Author: Prepared For: Date Thursday, February 25, 2021 Comments: Page 1 of 1 4/8/2021 e:///D:/Program%20Files/ANSYS%20Inc/v201/aisol/DesignSpace/DSPages/html/PrintPr... (d) 4 6 8 10 Diameter 2 4 6 8 Average Pressure 10-3 (e) Figure 2. The single output hemisphere sensor characterization. (a) Experimental setup for character- izing the single output self-healing soft sensor. (b) Sensor response to damage during a periodic step indentation. (c) Sensor response to a periodic truncated sine signal. (d) Ansys model for investigating the relation between sensor geometry and force sensitivity. (e) Exponential relation between the diameter of the hemisphere and the internal pressure for a constant force input. (a) 100 110 120 130 140 150 160 170 180 190 200 Time(sec) 0 2 4 6 8 10 12 Force(N) 100 105 110 115 Pressure(KPa) Sensor Damage (b) (b) (a) 0 50 100 150 200 250 300 Time (sec) 0 1 2 3 4 5 6 7 8 9 10 Force (N) 98 100 102 104 106 108 110 Pressure (KPa) (c) CONTPRES Subject: Author: Prepared For: Date Thursday, February 25, 2021 Comments: Page 1 of 1 4 6 8 10 Diameter 2 4 6 8 Average Pressure 10-3 (e) (d) (e) (d) 4/8/2021 /D:/Program%20Files/ANSYS%20Inc/v201/aisol/DesignSpace/DSPages/html/PrintPr... Figure 2. The single output hemisphere sensor characterization. (a) Experimental setup for character- izing the single output self-healing soft sensor. (b) Sensor response to damage during a periodic step indentation. (c) Sensor response to a periodic truncated sine signal. (d) Ansys model for investigating the relation between sensor geometry and force sensitivity. (e) Exponential relation between the diameter of the hemisphere and the internal pressure for a constant force input. 7 of 12 Sensors 2021, 21, 8284 1 cm 12 cm Centerline Type equation here. 𝑥 𝐿/2 𝑡1 = 𝐿 2 −𝑥/𝑣𝑒𝑙 𝑡2 = 𝐿 2 + 𝑥/𝑣𝑒𝑙 To Pressure Sensor To Pressure Sensor Location 1 Location 6 Location 11 (a) Bandpass Filter (b) 1 cm 12 cm Centerline Type equation here. 3.2. One-Dimensional Damage Detection and Localization 𝑥 𝐿/2 𝑡1 = 𝐿 2 −𝑥/𝑣𝑒𝑙 𝑡2 = 𝐿 2 + 𝑥/𝑣𝑒𝑙 Sensor Sensor Location 1 Location 6 Location 11 (a) Bandpass Filter (b) 1 2 3 4 5 6 7 8 9 10 11 Location -10 -5 0 5 10 Distance from Centre (cm) Estimated Damage Location True Location (c) Figure 3. One-dimensional damage detection and localization using the multi-output ﬂuidic sensor. (a) One-dimensional damage detection and localization sensor. The time-of-ﬂight model for the 1D model is shown below. (b) Processing of the raw pressure signals for damage detection and localization. (c) Damage localization accuracy of the one-dimensional damage sensor. Bandpass Filter (b) 1 2 3 4 5 6 7 8 9 10 11 Location -10 -5 0 5 10 Distance from Centre (cm) Estimated Damage Location True Location (c) (c) Figure 3. One-dimensional damage detection and localization using the multi-output ﬂuidic sensor. (a) One-dimensional damage detection and localization sensor. The time-of-ﬂight model for the 1D model is shown below. (b) Processing of the raw pressure signals for damage detection and localization. (c) Damage localization accuracy of the one-dimensional damage sensor. Sensors 2021, 21, 8284 8 of 12 Measuring the time-of-arrival difference requires the identiﬁcation of discernible features in the signal for the particular sampling frequency. As the sampling frequency decreases, the features must lie at a higher frequency spectrum for accuracy. Hence, normal contacts are difﬁcult to be localized. However, damage events have high-frequency components that are characteristic of their internal geometry. This makes the detection and localization easier even at a relatively low sampling frequency of 200 KHz. For our sampling rate, the highest localization resolution we can obtain is 0.165 cm. The raw pressure signals and its spectrogram are shown in Figure 3b. When the sensor is damaged, a high-frequency component can be observed. Due to noise in the raw pressure signal, detecting features for the time-of-arrival difference measurement is not possible. Therefore, the raw signal is ﬁltered with a bandpass ﬁlter whose range is manually estimated for each sensor geometry. Once the ﬁlter parameters are ﬁxed, damage can be reliably detected, irrespective of the damage location. Detecting features in the ﬁltered signal is now easier (see Figure 3b). To test the performance of our damage detection and localization setup, we perform an experiment where the sensors are damaged in known locations along the sensor length. 3.2. One-Dimensional Damage Detection and Localization There were a total of 11 locations, and each location was damaged thrice. The results are shown in Figure 4b. Of the 33 damages, 31 of the damages were detected, amounting to a detection accuracy of 93%. The localization accuracy was 1.13 ± 1.41 cm. Non-detection of damage happens when the ﬁltered signals are too weak. Using more sensitive pressure sensors with a low signal-to-noise ratio will greatly improve the damage sensing abilities and also allow normal contact localization abilities. Alternatively, having a thicker SH matrix will improve the damage detection ability, as higher power is required to induce damage in that case. 3.3. Two-Dimensional Damage Detection and Localization Conclusions This article presents the design and fabrication of a self-healing The sensor works on the principle of information transfer from the p To Pressure Sensor To Pressure Sensor 7cm 7cm 1 cm Location 1 Location 42 (a) 7cm 7cm 1 cm (a) 5 10 15 20 25 30 35 40 Damage Location 0 2 4 6 8 10 Localization Error (cm) 2.87 cm Distance from Actual Location Average Damage Localization Error (b) True Damage Location Predicted Damage Location (c) Figure 4. Two-dimensional damage detection and localization using the multi-output ﬂuidic sensor. (a) The 2D sensory structure for continuous damage detection and localization along a surface area. (b) Localization error at each location of damage. (c) Examples of damage detection and location for the 2D sensor morphology. (a) 5 10 15 20 25 30 35 40 Damage Location 0 2 4 6 8 10 Localization Error (cm) 2.87 cm Distance from Actual Location Average Damage Localization Error (b) (b) True Damage Location Predicted Damage Location (c) (c) Figure 4. Two-dimensional damage detection and localization using the multi-output ﬂuidic sensor. (a) The 2D sensory structure for continuous damage detection and localization along a surface area. (b) Localization error at each location of damage. (c) Examples of damage detection and location for the 2D sensor morphology. 3.3. Two-Dimensional Damage Detection and Localization As mentioned before, due to the high impedance among different mediums, the path taken by the pressure signal is almost always along the ﬂuidic chamber, even if the chamber is not straight. This allows us to easily scale the proposed system to arbitrarily complex surfaces with the same number of output pressure sensors. Figure 4a shows the design of a SH sensory structure for 2D damage detection and localization. The total length of the ﬂuidic chamber is around 42 cm. Similar to the previous subsection, the sensory skin is manually damaged at marked locations, and the damage localization error is measured. Due to the complex morphology of the ﬂuidic path, the 2D path of the chamber is estimated using computer vision techniques. Each marked location was damaged twice. Due to the higher thickness of the current system, all the damage instances were detected. The localization error was slightly higher at 2.87 ± 2.26 cm. This amounts to a localization error within 15% of the length of the sensory pathway. The absolute error at each location is shown in Figure 4b. Examples of damage localization on the 2D morphology for nine random points are shown in Figure 4c. Although we assume that the pressure waves travel along the ﬂuidic chamber due to the curvature of the shape, there are possible reﬂections that might affect the time-of-arrival difference estimation. 9 of 12 Sensors 2021, 21, 8284 To Pressure Sensor To Pressure Sensor 7cm 7cm 1 cm Location 1 Location 42 (a) 5 10 15 20 25 30 35 40 Damage Location 0 2 4 6 8 10 Localization Error (cm) 2.87 cm Distance from Actual Location Average Damage Localization Error (b) True Damage Location Predicted Damage Location (c) Figure 4. Two-dimensional damage detection and localization using the multi (a) The 2D sensory structure for continuous damage detection and localization (b) Localization error at each location of damage. (c) Examples of damage dete the 2D sensor morphology. 4. 4. Conclusions This article presents the design and fabrication of a self-healing soft ﬂuidic sensor. The sensor works on the principle of information transfer from the physical stimuli via Sensors 2021, 21, 8284 10 of 12 10 of 12 pressure waves traveling in an enclosed ﬂuidic chamber. This allows us to obtain tactile information remotely and unobtrusively. More importantly, this enables us to fabricate these soft sensors without the addition of any functional materials. By using self-healing supramolecular polymers as the enclosing matrix, we obtain exemplary self-healing prop- erties characterized by instantaneous functional recovery at ambient conditions without external inputs. We perform extensive characterization of the soft sensor and its application as a force sensor. Furthermore, using ﬁnite element analysis, we demonstrate how the chamber morphology can be designed to obtain higher sensitivity to applied forces. By leveraging acoustic characteristics of damage and the relatively slow speed of pressure waves, we ﬁnally show how our sensing principle can be used to even detect and localize damage in a continuous manner indeﬁnitely. To the best of our knowledge, this is the ﬁrst demonstration of such a use of a soft sensor. We demonstrate highly reliable damage detec- tion and accurate localization abilities on our modiﬁed multi-output sensor morphologies using commonly available hardware. The detection and localization performance can be improved and conducted in real-time by executing all the computation using electronic components. This would also facilitate highly scalable contact localization abilities using just two output pressure sensors. The ability to detect and localize damage repeatedly could be an important tool for soft robots and soft wearable devices for monitoring struc- tural integrity and as a feedback mechanism to adapt and react to damage-causing actions. These systems can also be devised for structural health monitoring as a replacement to conventional Non-Destructive Testing techniques due to low compliance, ease of man- ufacturing and deployment. The self-healing capabilities also increase the life-span of the sensors. Author Contributions: Conceptualization, T.G.T. and J.H.; data curation, T.G.T.; formal analysis, T.G.T.; investigation, T.G.T. and J.H.; methodology, T.G.T. and J.H.; project administration, F.I.; resources, A.W.B. and F.I.; software, T.G.T.; supervision, F.I.; visualization, T.G.T.; writing—original draft, T.G.T. and A.W.B.; writing—review and editing, T.G.T., A.W.B. and F.I. All authors have read and agreed to the published version of the manuscript. 7. Darabi, M.A.; Khosrozadeh, A.; Mbeleck, R.; Liu, Y.; Chang, Q.; Jiang, J.; Cai, J.; Wang, Q.; Luo, G.; Xing, M. Skin-inspired multifunctional autonomic-intrinsic conductive self-healing hydrogels with pressure sensitivity, stretchability, and 3D printability. Adv. Mater. 2017, 29, 1700533. [CrossRef] Conﬂicts of Interest: The authors declare no conﬂict of interest. Conﬂicts of Interest: The authors declare no conﬂict of interest. f 6. Wallin, T.; Pikul, J.; Bodkhe, S.; Peele, B.; Mac Murray, B.; Therriault, D.; McEnerney, B.; Dillon, R.; Giannelis, E.; Shepherd, R. Click chemistry stereolithography for soft robots that self-heal. J. Mater. Chem. B 2017, 5, 6249–6255. [CrossRef] 4. Terryn, S.; Brancart, J.; Lefeber, D.; Van Assche, G.; Vanderborght, B. Self-healing soft pneumatic robots. Sci. Robot. 2017, 2, 9. [CrossRef] [PubMed] 2. Bilodeau, R.A.; Kramer, R.K. Self-healing and damage resilience for soft robotics: A review. Front. Robot. AI 2017, 4, 48. [CrossRef] 3. Utrera-Barrios, S.; Verdejo, R.; López-Manchado, M.A.; Santana, M.H. Evolution of self-healing elastomers, from extrinsic to combined intrinsic mechanisms: A review. Mater. Horizons 2020, 7, 2882–2902. [CrossRef] 5. Roels, E.; Terryn, S.; Brancart, J.; Verhelle, R.; Van Assche, G.; Vanderborght, B. Additive manufactu robots. Soft Robot. 2020, 7, 711–723. [CrossRef] [PubMed] 4. Conclusions Funding: This work was supported by the SHERO project, a Future and Emerging Technologies (FET) program of the European Commission (grant agreement ID 828818). Institutional Review Board Statement: Not applicable. Institutional Review Board Statement: Not applicable. Institutional Review Board Statement: Not applicable. Institutional Review Board Statement: Not applicable. Informed Consent Statement: Not applicable. Informed Consent Statement: Not applicable. Data Availability Statement: Supporting Information will be available upon request. Acknowledgments: The authors also would like to acknowledge Leone Costi for providing ideas for multi-lateration. Acknowledgments: The authors also would like to acknowledge Leone Costi for providing ideas for multi-lateration. Conﬂicts of Interest: The authors declare no conﬂict of interest. References Highly transparent, self-healable, and bio-inspired intelligent ionic skins. ACS Appl. Mater. Interfaces 2020, 12, 15657–15666. [CrossRef] hang, Z.; Wang, L.; Yu, H.; Zhang, F.; Tang, L.; Feng, Y.; Feng, W. Highly transparent, self-healable, and adhes o-inspired intelligent ionic skins. ACS Appl. Mater. Interfaces 2020, 12, 15657–15666. [CrossRef] red intelligent ionic skins. ACS Appl. Mater. Interfaces 2020, 12, 15657–15666. [CrossRef] 22. Zhang, J.; Wan, L.; Gao, Y.; Fang, X.; Lu, T.; Pan, L.; Xuan, F. Highly stretchable and self-healable MXene/polyvinyl alcohol hydrogel electrode for wearable capacitive electronic skin. Adv. Electron. Mater. 2019, 5, 1900285. [CrossRef] 23. Chakraborty, P.; Guterman, T.; Adadi, N.; Yadid, M.; Brosh, T.; Adler-Abramovich, L.; Dvir, T.; Gazit, E. A self-healing, all-organic, conducting, composite peptide hydrogel as pressure sensor and electrogenic cell soft substrate. ACS Nano 2018, 13, 163–175. [CrossRef] [PubMed] [ ] [ ] 24. Sijbesma, R.P.; Beijer, F.H.; Brunsveld, L.; Folmer, B.J.; Hirschberg, J.K.; Lange, R.F.; Lowe, J.K.; Meijer, E. Reversible polymers formed from self-complementary monomers using quadruple hydrogen bonding. Science 1997, 278, 1601–1604. [CrossRef] 24. Sijbesma, R.P.; Beijer, F.H.; Brunsveld, L.; Folmer, B.J.; Hirschberg, J.K.; Lange, R.F.; Lowe, J.K.; Meijer, E. Reversible polymers formed from self-complementary monomers using quadruple hydrogen bonding. Science 1997, 278, 1601–1604. [CrossRef] 25. van Gemert, G.M.; Peeters, J.W.; Söntjens, S.H.; Janssen, H.M.; Bosman, A.W. Self-healing supramolecular polymers in action. Macromol. Chem. Phys. 2012, 213, 234–242. [CrossRef] 25. van Gemert, G.M.; Peeters, J.W.; Söntjens, S.H.; Janssen, H.M.; Bosman, A.W. Self-healing supramolecular polymers in action. Macromol. Chem. Phys. 2012, 213, 234–242. [CrossRef] 26. Söntjens, S.H.; Renken, R.A.; van Gemert, G.M.; Engels, T.A.; Bosman, A.W.; Janssen, H.M.; Govaert, L.E.; Baaijens, F.P. Thermoplastic elastomers based on strong and well-deﬁned hydrogen-bonding interactions. Macromolecules 2008, 41, 5703–5708. [CrossRef] 27. Besseling, P.J.; Mes, T.; Bosman, A.W.; Peeters, J.W.; Janssen, H.M.; Bakker, M.H.; Fledderus, J.O.; Teraa, M.; Verhaar, M.C.; Gremmels, H.; et al. The in-vitro biocompatibility of ureido-pyrimidinone compounds and polymer degradation products. J. Polym. Sci. 2021, 59, 1267–1277. [CrossRef] 28. Fishel, J.A.; Santos, V.J.; Loeb, G.E. A robust micro-vibration sensor for biomimetic ﬁngertips. In Proceedings of the 2008 2nd IEEE RAS & EMBS International Conference on Biomedical Robotics and Biomechatronics, Scottsdale, AZ, USA, 19–22 October 2008; pp. 659–663. pp 29. Soter, G.; Garrad, M.; Conn, A.T.; Hauser, H.; Rossiter, J. Skinﬂow: A soft robotic skin based on ﬂuidic transmission. In Proceedings of the 2019 2nd IEEE International Conference on Soft Robotics (RoboSoft), Seoul, Korea, 14–18 April 2019; pp. References Electronic skins and machine Wang, H.; Totaro, M.; Beccai, L. Toward perceptive 12. Shih, B.; Shah, D.; Li, J.; Thuruthel, T.G.; Park, Y.L.; Iida, F.; Bao, Z.; Kramer-Bottiglio, R.; Tolley, M.T. Electronic skins and machine learning for intelligent soft robots. Sci. Robot. 2020, 5, 41. [CrossRef] g, K.U.; Park, I.; Sitti, M. Stretchable, skin-mountable, and wearable strain sensors and their potential ew. Adv. Funct. Mater. 2016, 26, 1678–1698. [CrossRef] 13. Amjadi, M.; Kyung, K.U.; Park, I.; Sitti, M. Stretchable, skin-mountable, and wearable strain sen applications: A review. Adv. Funct. Mater. 2016, 26, 1678–1698. [CrossRef] 14. Tee, B.C.; Wang, C.; Allen, R.; Bao, Z. An electrically and mechanically self-healing composite with pressure-and ﬂexion-sensitive properties for electronic skin applications. Nat. Nanotechnol. 2012, 7, 825–832. [CrossRef] 15. D’Elia, E.; Barg, S.; Ni, N.; Rocha, V.G.; Saiz, E. Self-healing graphene-based composites with sensing capabilities. Adv. Mater. 2015, 27, 4788–4794. [CrossRef] [PubMed] 16. Yang, Y.; Zhu, B.; Yin, D.; Wei, J.; Wang, Z.; Xiong, R.; Shi, J.; Liu, Z.; Lei, Q. Flexible self-healing nanocomposites for recoverable motion sensor. Nano Energy 2015, 17, 1–9. [CrossRef] ang, J.; Tok, J.B.H.; Bao, Z. Self-healing soft electronics. Nat. Electron. 2019, 2, 144–150. [CrossRef] T Zh Y Li Q Ch W W X P L X B X H A lf h l bl hi hl Kang, J.; Tok, J.B.H.; Bao, Z. Self-healing soft electro 18. Wang, T.; Zhang, Y.; Liu, Q.; Cheng, W.; Wang, X.; Pan, L.; Xu, B.; Xu, H. A self-healable, highly stretchable, and solution processable conductive polymer composite for ultrasensitive strain and pressure sensing. Adv. Funct. Mater. 2018, 28, 1705551. [CrossRef] 19. Liu, Y.J.; Cao, W.T.; Ma, M.G.; Wan, P. Ultrasensitive wearable soft strain sensors of conductive, self-healing, and elastic hydrogels with synergistic “soft and hard” hybrid networks. ACS Appl. Mater. Interfaces 2017, 9, 25559–25570. [CrossRef] 20. Wu, M.; Chen, J.; Ma, Y.; Yan, B.; Pan, M.; Peng, Q.; Wang, W.; Han, L.; Liu, J.; Zeng, H. Ultra elastic, stretchable, self-healing conductive hydrogels with tunable optical properties for highly sensitive soft electronic sensors. J. Mater. Chem. A 2020, 8, 24718–24733. [CrossRef] , [ ] 21. Zhang, Z.; Wang, L.; Yu, H.; Zhang, F.; Tang, L.; Feng, Y.; Feng, W. Highly transparent, self-healable, and adhesive organogels for bio-inspired intelligent ionic skins. ACS Appl. Mater. Interfaces 2020, 12, 15657–15666. [CrossRef] 21. Zhang, Z.; Wang, L.; Yu, H.; Zhang, F.; Tang, L.; Feng, Y.; Feng, W. 34. Bigoni, C.; Zhang, Z.; Hesthaven, J.S. Systematic sensor placement for structural anomaly detection in the absence of damaged states. Comput. Methods Appl. Mech. Eng. 2020, 371, 113315. [CrossRef] References 1. Terryn, S.; Langenbach, J.; Roels, E.; Brancart, J.; Bakkali-Hassani, C.; Poutrel, Q.A.; Georgopoulou, A.; Thuruthel, T.G.; Safaei, A.; Ferrentino, P.; et al. A review on self-healing polymers for soft robotics. Mater. Today 2021, 47, 187–205. [CrossRef] 2. Bilodeau, R.A.; Kramer, R.K. Self-healing and damage resilience for soft robotics: A review. Front. Robot. AI 2017, 4, 48. [CrossRef] 3. Utrera-Barrios, S.; Verdejo, R.; López-Manchado, M.A.; Santana, M.H. Evolution of self-healing elastomers, from extrinsic to combined intrinsic mechanisms: A review. Mater. Horizons 2020, 7, 2882–2902. [CrossRef] 4. Terryn, S.; Brancart, J.; Lefeber, D.; Van Assche, G.; Vanderborght, B. Self-healing soft pneumatic robots. Sci. Robot. 2017, 2, 9. [CrossRef] [PubMed] ; Brancart, J.; Verhelle, R.; Van Assche, G.; Vanderborght, B. Additive manufacturing for self-healing soft 020, 7, 711–723. [CrossRef] [PubMed] 6. Wallin, T.; Pikul, J.; Bodkhe, S.; Peele, B.; Mac Murray, B.; Therriault, D.; McEnerney, B.; Dillon, R.; Giannelis, E.; Shepherd, R. Click chemistry stereolithography for soft robots that self-heal. J. Mater. Chem. B 2017, 5, 6249–6255. [CrossRef] 11 of 12 11 of 12 Sensors 2021, 21, 8284 8. Lei, Z.; Wang, Q.; Sun, S.; Zhu, W.; Wu, P. A bioinspired mineral hydrogel as a self-healable, mechanically adaptable ionic skin for highly sensitive pressure sensing. Adv. Mater. 2017, 29, 1700321. [CrossRef] [PubMed] g y p g 9. Baumgartner, M.; Hartmann, F.; Drack, M.; Preninger, D.; Wirthl, D.; Gerstmayr, R.; Lehner, L.; Mao, G.; Pruckner, R.; Demchyshyn, S.; et al. Resilient yet entirely degradable gelatin-based biogels for soft robots and electronics. Nat. Mater. 2020, 19, 1102–1109. [CrossRef] [PubMed] [ ] [ ] 10. Khatib, M.; Zohar, O.; Haick, H. Self-Healing Soft Sensors: From Material Design to Implementation. Adv. Mater. 2021, 33, 2004190. [CrossRef] 11. Wang, H.; Totaro, M.; Beccai, L. Toward perceptive soft robots: Progress and challenges. Adv. Sci. 2018, 5, 1800541. [CrossRef] 12. Shih, B.; Shah, D.; Li, J.; Thuruthel, T.G.; Park, Y.L.; Iida, F.; Bao, Z.; Kramer-Bottiglio, R.; Tolley, M.T. Electronic skins and machine learning for intelligent soft robots. Sci. Robot. 2020, 5, 41. [CrossRef] 11. Wang, H.; Totaro, M.; Beccai, L. Toward perceptive soft robots: Progress and challenges. Adv. Sci. 2018, 5, 1800541. [CrossRef] 11. Wang, H.; Totaro, M.; Beccai, L. Toward perceptive soft robots: Progress and challenges. Adv. Sci. 2018, 5, 1800541. [CrossRef] 12. Shih, B.; Shah, D.; Li, J.; Thuruthel, T.G.; Park, Y.L.; Iida, F.; Bao, Z.; Kramer-Bottiglio, R.; Tolley, M.T. 35. Markvicka, E.J.; Tutika, R.; Bartlett, M.D.; Majidi, C. Soft electronic skin for multi-site damage detection and localization. Adv. Funct. Mater. 2019, 29, 1900160. [CrossRef] 39. Thuruthel, T.G.; Hughes, J.; Georgopoulou, A.; Clemens, F.; Iida, F. Using Redundant and Disjoint Time-Variant Soft Robotic Sensors for Accurate Static State Estimation. IEEE Robot. Autom. Lett. 2021, 6, 2099–2105. [CrossRef] y 38. Thuruthel, T.G.; Shih, B.; Laschi, C.; Tolley, M.T. Soft robot perception using embedded soft sensors and recurrent neural networks. Sci. Robot. 2019, 4, eaav1488. [CrossRef] [PubMed] 36. Khatib, M.; Zohar, O.; Saliba, W.; Haick, H. A Multifunctional Electronic Skin Empowered with Damage Mapping and Autonomic Acceleration of Self-Healing in Designated Locations. Adv. Mater. 2020, 32, 2000246. [CrossRef] [PubMed] 37. Janssen, H.M.; Van Gemert, G.M.L.; Bosman, A.W. A Process for the Preparation of a Supramolecular Polymer. EP 2450394B1, 9 May 2012. , , [ ] 36. Khatib, M.; Zohar, O.; Saliba, W.; Haick, H. A Multifunctional Electronic Skin Empowered with Damage Mapping and Autonomic Acceleration of Self-Healing in Designated Locations. Adv. Mater. 2020, 32, 2000246. [CrossRef] [PubMed] 37. Janssen, H.M.; Van Gemert, G.M.L.; Bosman, A.W. A Process for the Preparation of a Supramolecular Polymer. EP 2450394B1, 9 M 2012 g g [ ] [ ] 37. Janssen, H.M.; Van Gemert, G.M.L.; Bosman, A.W. A Process for the Preparation of a Supramolecular Polymer. EP 2450394B1, 9 May 2012. , , [ ] 36. Khatib, M.; Zohar, O.; Saliba, W.; Haick, H. A Multifunctional Electronic Skin Empowered with Damage Mapping and Autonomic Acceleration of Self-Healing in Designated Locations. Adv. Mater. 2020, 32, 2000246. [CrossRef] [PubMed] References 355–360. 30. Hughes, J.; Spielberg, A.; Chounlakone, M.; Chang, G.; Matusik, W.; Rus, D. A Simple, Inexpensive, Wearable Glove with Hybrid Resistive Pressure Sensors for Computational Sensing Proprioception and Task Identiﬁcation Adv Intell Syst 2020 2 2000002 29. Soter, G.; Garrad, M.; Conn, A.T.; Hauser, H.; Rossiter, J. Skinﬂow: A soft robotic skin based on ﬂuidic transmission. In Proceedings of the 2019 2nd IEEE International Conference on Soft Robotics (RoboSoft), Seoul, Korea, 14–18 April 2019; pp. 355–360. 30 H h J S i lb A Ch l k M Ch G M ik W R D A Si l I i W bl Gl i h H b id of the 2019 2nd IEEE International Conference on Soft Robotics (RoboSoft), Seoul, Korea, 14–18 April 2019; pp. 355–360. 30. Hughes, J.; Spielberg, A.; Chounlakone, M.; Chang, G.; Matusik, W.; Rus, D. A Simple, Inexpensive, Wearable Glove with Hybrid Resistive-Pressure Sensors for Computational Sensing, Proprioception, and Task Identiﬁcation. Adv. Intell. Syst. 2020, 2, 2000002. [CrossRef] 31. Cartz, L. Nondestructive. Testing. 1995. Available online: https://www.osti.gov/biblio/260617 (accessed on 9 December 2021). 32. Salawu, O. Detection of structural damage through changes in frequency: A review. Eng. Struct. 1997, 19, 718–723. [CrossRef] 31. Cartz, L. Nondestructive. Testing. 1995. Available online: https://www.osti.gov/biblio/260617 (acces 32. Salawu, O. Detection of structural damage through changes in frequency: A review. Eng. Struct. 1997 , g p // g / / ( ) 32. Salawu, O. Detection of structural damage through changes in frequency: A review. Eng. Struct. 1997, 19, 718–723. [CrossRef] 33. Fedele, R.; Praticò, F.G.; Carotenuto, R.; Della Corte, F.G. Instrumented infrastructures for damage detection and management. In Proceedings of the 2017 5th IEEE International Conference on Models and Technologies for Intelligent Transportation Systems (MT-ITS), Naples, Italy, 26–28 June 2017; pp. 526–531. p y pp 34. Bigoni, C.; Zhang, Z.; Hesthaven, J.S. Systematic sensor placement for structural anomaly detection in the absence of damaged states. Comput. Methods Appl. Mech. Eng. 2020, 371, 113315. [CrossRef] 12 of 12 12 of 12 Sensors 2021, 21, 8284 y 38. Thuruthel, T.G.; Shih, B.; Laschi, C.; Tolley, M.T. Soft robot perception using embedded soft sensors and recurrent neural networks. Sci. Robot. 2019, 4, eaav1488. [CrossRef] [PubMed]
https://openalex.org/W4392355146	https://intjem.biomedcentral.com/counter/pdf/10.1186/s12245-024-00609-1	English	null	The leading role of expert safety knowledge in supporting the mission of caring for patients during man-made and natural disasters: state of emergency medicine in Ethiopia, Myanmar, and Ukraine	International journal of emergency medicine	2,024	cc-by	5,372	STATE OF INTERNATIONAL EMERGENCY MEDICINE Open Access The leading role of expert safety knowledge in supporting the mission of caring for patients during man-made and natural disasters: state of emergency medicine in Ethiopia, Myanmar, and Ukraine Ralph C. Miles1, Vivian I. Avelino-Silva1, Wilfred Odoke1, Jan van den Hombergh1, Fernanda F. Fonseca1, Mengistu GebreMichael2, Yaroslava Lopatina3, Win Oo4 and Adele Schwartz Benzaken1 Abstract Abstract Preparedness to endure extreme situations such as natural disasters or military conflicts is not commonplace in healthcare training programs. Moreover, multidisciplinary teams in health services rarely (if ever) include experts in security. However, when emergency situations occur, prevailing healthcare demands do not cease to exist, and unexpected demands often surge due to the shortage of other services and supplies or as a consequence of the emergency condition itself.l g y With services in 45 countries, AIDS Healthcare Foundation (AHF) has operated in several conflict zones, facing broad and challenging security demands. Since 2017 AHF has implemented the Global Department of Safety and Security (GDSS), a dedicated intelligence and safety program that had a key role in the security monitoring, preparedness, and defense responses, assisting staff members and clients during recent conflicts. p p p gf gl In this manuscript, we describe the experience of AHF’s GDSS in three recent military conflicts in Ethiopia, Myanmar, and Ukraine, and provide insights into steps that can be taken to assure staff safety and support the mission of caring for patients throughout catastrophic events. Keywords Disaster Planning, Security measures, Civil Defense, Health Care Facilities, Manpower, and services Correspondence: Vivian I. Avelino-Silva viviansilva87@gmail.com 1Aids Healthcare Foundation Global Program, 6255 Sunset Blvd., 21s Los Angeles, CA 90028, USA 2Aids Healthcare Foundation Ethiopia, Addis Ababa, Ethiopia 3Aids Healthcare Foundation Ukraine, Kyiv, Ukraine 4Aids Healthcare Foundation Myanmar, Yangon, Myanmar © The Author(s) 2024. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. *Correspondence: Vivian I. Avelino-Silva viviansilva87@gmail.com 1Aids Healthcare Foundation Global Program, 6255 Sunset Blvd. Los Angeles, CA 90028, USA 2Aids Healthcare Foundation Ethiopia, Addis Ababa, Ethiopia 3Aids Healthcare Foundation Ukraine, Kyiv, Ukraine 4Aids Healthcare Foundation Myanmar, Yangon, Myanmar (2024) 17:31 (2024) 17:31 Miles et al. International Journal of Emergency Medicine (2024) 17:31 https://doi.org/10.1186/s12245-024-00609-1 International Journal of Emergency Medicine Miles et al. International Journal of Emergency Medicine https://doi.org/10.1186/s12245-024-00609-1 Background security specialists from a staggering amount of data comprising both useful evidence and lots of noise. Once impending risks are identified, knowledge and back ground safety expertise are crucial in guiding actions that will determine if a healthcare facility will be able to continue providing services. In other words, security and intelligence are grounded on specific skills and tech niques that are unknown to healthcare providers, yet necessary to continue the mission of caring for patients when disasters occur. g In December 2021, the AIDS Healthcare Foundation (AHF) Global Department of Safety and Security (GDSS) recommended to the overall AHF leadership, Europe Bureau, and country program members that preparations be made for a potential Russian invasion in Ukraine. Such recommendations were not trivial - they affected the daily routine of staff members and more than 50,000 cli ents in HIV care in the Ukraine program and represented a significant additional expenditure to the institution. The initiative was implemented based on a rigorous mon itoring and analysis process of the disintegrating diplo matic ties as well as the observation that Russian military forces were initiating unequivocal offensive movements. After almost two years of unremitting military con flict, the impact of the projected risk assessment and implementation of a preparedness plan is now reflected on the fact that AHF staff and their families have been able to protect themselves and remain in the country, providing services. Moreover, HIV prevention and care indicators highlight some of the concrete effects of this intervention: the AHF program in Ukraine experienced an increase in new enrollments in care, new antiretroviral therapy initiations, and number of clinics in active opera tion, reflecting the continuity of testing, focus on linkage to care, and an efficient approach to open new facilities supporting both internally displaced clients and patients relocated from other services. AHF is a global non-governmental organization spe cializing in HIV prevention and care of since 1987. Cur rently, AHF supports more than 1.8 million clients in care in 45 countries including low, middle, and high-income states. To accomplish the goal of providing the best avail able care to patients, AHF collaborates with local insti tutions and governments, combining existing resources with additional services to achieve a tailored, optimized care strategy to each country’s fight against HIV. Background Work ing in a wide range of places and conditions means that AHF must have good relationships with all levels of gov ernment, maintaining a neutral position towards political conflicts. Simultaneously, a global operation also entails broad and challenging security demands. Although several international armistice settlements have been established, particularly after World War II, challenges such as social inequalities and deteriorat ing political relations in the past years indicate with no uncertainty that the civil society continues to face secu rity risks. This led AHF to implement the GDSS in 2017, aiming to enhance the wellbeing of patients, staff mem bers and their families, and to support continuity of care to the best possible extent during extreme situations. This decision quickly proved to be crucial. AHF’s GDSS has been a protagonist in three recent military conflicts in Ethiopia, Myanmar, and Ukraine. Here we describe the challenges and achievements of the program and discuss insights into actions that can be done to guarantee the safety of clients and staff members, and support continu ity of care throughout catastrophic events. Retrospectively, the decision to prepare the organiza tion for the chaotic impact of a potential conflict seems obvious. Yet, its implementation was only possible due to the expertise of a qualified security and intelligence team with previously built credibility across the institu tion. Ordinarily, experts in security are not envisioned as part of multidisciplinary teams in healthcare services or health institutions. Similarly, most healthcare train ing programs do not comprise preparedness for extreme situations such as natural disasters or military conflicts. However, the importance of an intelligence and security strategy cannot be underestimated; a well-developed and thought-out intelligence program has the ability to provide early warnings about threats that can impact critical service operations. Predicting risks even a few days before a catastrophic event happens is crucial to determine one’s ability to protect staff members, clients, assets, and to allow the continuity of health services. Ultimately, when emergency situations occur, prevailing healthcare demands do not cease to exist, and organiza tions are further challenged with unexpected demands resulting from the shortage of other services and sup plies, or from the emergence of novel ailments caused by the disaster itself.h Tigray war (Ethiopia, November 2020 - November 2022) Ethiopia is the second most populous country in Africa with approximately 114 million inhabitants, of whom 610,000 are estimated to live with HIV. Miles et al. International Journal of Emergency Medicine (2024) 17:31 Miles et al. International Journal of Emergency Medicine (2024) 17:31 Miles et al. International Journal of Emergency Medicine Page 2 of 8 Page 2 of 8 (2024) 17:31 Background As seen in many countries in Sub-Saharan Africa, HIV prevalence is higher among adult women (1.0%) compared to men (0.6%), and more than 300,000 children and adolescents aged 0–17 are orphans due to Aids. In the past years, Ethiopia registered significant improvements in key HIV indicators, with 70% reduction in new HIV infections and 57% reduction in Aids-related deaths since 2010 [1]. The analysis of risk mostly relies on information avail able through open sources, decoded by experienced AHF started operations in Ethiopia in 2012, providing comprehensive care services for people living with HIV, Miles et al. International Journal of Emergency Medicine (2024) 17:31 Page 3 of 8 Miles et al. International Journal of Emergency Medicine Page 3 of 8 (2024) 17:31 exacerbated food insecurity, sparked outbreaks of infec tious diseases such as malaria, measles, and cholera, and collapsed immunization programs and health services [7, 8]. including counseling and testing, antiretroviral therapy, treatment of opportunistic infections, diagnosis and management of other sexually transmitted infections, and condom distribution and promotion.h The 1,104,300 km2 Ethiopian territory is divided in 11 administrative regions that are home to long-standing ethno-nationalist tensions. In November 2020, the fed eral defense force started a military offensive against the local dominant group Tigray People’s Liberation Front (TPLF) in the northern region of the country. Over two years, the conflict led to thousands dead and millions dis placed [2, 3]. AHF became concerned about the poten tial impact of the conflict and its harmful effects on staff members and clients in September 2021, following the escalation of Tigrayan forces towards Addis Ababa metropolitan area, where AHF concentrated most of its operations. The preparedness plan included the iden tification of evacuation routes for staff members, and the development of a communication plan less likely to be interrupted by cyberattacks. AHF conducted train ing sessions with employees, anticipating the impact of a potential invasion, and prioritizing strategies to shelter in place or prepare for evacuation if needed. AHF operations were significantly affected during the war in Ethiopia. Although there was no closure of ser vices, the program reported a significant reduction in the number of condoms distributed in the country, and a small reduction in newly enrolled clients one year after the conflict onset compared to the same period one year before. Background Despite these difficulties, a few indica tors demonstrate that the preparedness plan allowed the continuation of key services in AHF facilities through out this challenging period (Fig. 1). Immediately before the conflict onset in Ethiopia, AHF had 20,476 clients in care; this number increased to 31,035 in October 2021, one year into the conflict, and recent data from July 2023 shows 43,494 clients in care in AHF facilities across the country. The number of facilities under the auspices of AHF also increased since 2020, from eight to twelve units. Myanmar coup d’état (Myanmar, February 2021) Myanmar coup d état (Myanmar, February 2021) Previously known as Burma, Myanmar is a 57 million- inhabitant southeast Asian country, with independence from United Kingdom since 1948 [9]. The country’s his tory is characterized by intense struggles for ethnic and territorial autonomy. Myanmar’s first free election in 2015 was followed by a coup d’état started on February 1st, 2021, on the day the Parliament was to endorse the election results [10]. Mass demonstrations and conflicts between civil protesters and the military junta arose. So far, the conflict displaced at least 1.4 million people and left a third of the country’s population in need of support [11]. p p p According to testimonials from our local staff, health care providers have reported frequent interruptions of HIV services in various facilities located in conflict areas. Patients also reported that they were frequently unable to get their antiretroviral medications, either because they couldn’t visit the facilities due to fear of the ongo ing conflicts or because the facilities where they had fol low up were destroyed. A number of patients were unable to travel to receive care at higher level health institutions even though they were given referrals; this happened because roads were blocked, means of transportation were unavailable, patients and their family members were unable to afford the travel costs due to the economic impact of the civil war, or because they were not allowed to cross inland borders due to their ethnicity. Hospitals and health units were occupied, looted, and destroyed. Besides the incidents affecting healthcare facilities, pro viders were victims of violence including episodes of physical injury, sexual assault, and assassinations [4, 5].h A distinct aspect that further complicated matters in Myanmar was the fact that many healthcare work ers were in the front line of the country’s civil disobedi ence demonstrations [12]. Background Consequently, security forces arrested providers at their homes, clinics and hospitals, and dozens were killed [13]; Myanmar has been consid ered one of the most dangerous places in the world to be a healthcare worker [14]. The inability to access services or scarcity of medi cal supplies in functioning facilities has forced Ethiopi ans in several regions to interrupt treatment and care of multiple chronic conditions, including diabetes, hyper tension, heart disease, tuberculosis, and cancer. People who needed tertiary level services experienced delays and many developed complications that could have been avoided had treatment been timely implemented [6].lfi According to UNAIDS data, approximately 280,000 adults and children are living with HIV in Myanmar, with higher prevalence among men (1.0%) compared to women (0.7%). Despite the political adversities, since 2010 the percentage of new HIV infections dropped 35%, while the percentage of Aids-related deaths declined 47% [15]. AHF has been providing overarching and free-of- charge HIV prevention and care services in Myanmar since 2015. AHF’s intelligence and safety program was surprised by the coup despite the historical political Although the military conflict officially ended in November 2022, the results of massive displacement and destruction of civilian structures and essential social services over 24 months were catastrophic. The conflict Miles et al. International Journal of Emergency Medicine (2024) 17:31 (2024) 17:31 Miles et al. International Journal of Emergency Medicine Page 4 of 8 Fig. 1 AHF’s HIV prevention and care indicators in Ethiopia, Myanmar, and Ukraine before/after the conflict onset. Figure developed by the authors using MapChart (Version 2.0, Quadratyx, Estonia). Fig. 1 AHF’s HIV prevention and care indicators in Ethiopia, Myanmar, and Ukraine before/after the conflict onset. Figure developed by the authors using Fig. 1 AHF’s HIV prevention and care indicators in Ethiopia, Myanmar, and Ukraine before/after the conflict onset. Figure developed by the authors using MapChart (Version 2.0, Quadratyx, Estonia). instability in Myanmar. The intensification of the con flict prompted concerns about AHF’s staff members and more than 18,000 clients in care. Past interactions with Asia Bureau and country program teams supported the establishment of AHF’s GDSS program’s credibility in conducting the Myanmar crisis response. The GDSS led periodic meetings with key country personnel explaining the events that could be anticipated, the potential impact of attacks, how to avoid them, and how to ensure security of AHF’s assets. instability in Myanmar. Conclusion Emergency situations will inevitably trigger evacuations, service and business closures, economic constraints, and supply shortages, challenging the safety of patients, healthcare workers, and their families. Moreover, extreme circumstances may impact health services oper ations and continuity of care. Our manuscript describes the experience of AHF’s GDSS program in three recent military conflicts, bringing insights into provisions that can be taken to support providers and clients during emergency situations. Several months before the Russian invasion, AHF’s GDSS program had identified Ukraine as an incontest able Targeted Area of Interest – a location of specific concern that requires being monitored by a special ist, using an intelligence data collection and assessment cycle. Based on this analysis, the security program was able to predict that the possibility of Russian invasion has very high, even when most people believed that an offen sive intervention would not occur. The risk assessment triggered the implementation of safety provisions across all Ukrainian facilities. The overarching plan included staff training, communication and cyber security, basic supplies and financial provisions for staff and their fam ily members, and medical supplies/equipment to prevent interruptions of healthcare services. Preparedness has enabled AHF employees to take care of themselves and their families, allowing the continuation of direct patient care in most sites and even the expansion to new loca tions despite the challenges imposed by the military conflict. Effective preparedness depends on the existence of an experienced intelligence team with established cred ibility, working closely with leaders and stakeholders to make assertive decisions. The anticipation of risks through the continuous intelligence monitoring and the prompt establishment of an emergency response plan are key to operationalize the timely provision of resources. The components of the response vary according with the type and scale of the anticipated warfare, as well as the characteristics of the conflict zone: at the lower end of the spectrum, warfare may be limited to protests and civil disobedience demonstrations; at the higher end, conflicts may be generalized, with mechanized war fare involving multiple countries. Understanding of the spectrum of actions allows reasonable forecasting of the characteristics of potential attacks, including the types of weaponry, speed and number of combatants, and pos sible level of destruction. Background The intensification of the con flict prompted concerns about AHF’s staff members and more than 18,000 clients in care. Past interactions with Asia Bureau and country program teams supported the establishment of AHF’s GDSS program’s credibility in conducting the Myanmar crisis response. The GDSS led periodic meetings with key country personnel explaining the events that could be anticipated, the potential impact of attacks, how to avoid them, and how to ensure security of AHF’s assets. for HIV and other sexually transmitted infections and community-based education campaigns had to be can celled due to restrictions on public gatherings. AHF Myanmar focused on sustaining access to HIV testing and treatment during the conflict. Because of workforce shortages, AHF relied on different categories of health personnel and technicians to ensure sustained provision of services, including peer volunteers trained to conduct targeted recruitment for HIV counseling and testing among key affected populations. f Before the military coup d’état, AHF registered 18,784 clients in care in six facilities; of those, five were located in central Myanmar (four in Yangon, one in Bago region), and one was located in Kayin State, at the frontier with Thailand. Over 12 months before the conflict set place, AHF provided diagnostic testing for more than 24,000 people and enrolled more than 2,300 new clients in care. One year after the coup, the program registered draw backs in the number of condoms distributed in the coun try, the number of people tested for HIV, and the number Many healthcare facilities endured shortage of human resources since the military coup, including AHF partner sites. During the first months, hospitals and clinics were temporarily shut down; civil unrest and violence, mobil ity restrictions, and limited health supplies hampered access to HIV care services. Facilities gradually resumed activities with limited staff and resources. Besides the incidents involving healthcare providers, local facilities reported that some patients were lost to follow-up after being arrested or fleeing to border areas. Mobile testing Miles et al. International Journal of Emergency Medicine (2024) 17:31 Miles et al. International Journal of Emergency Medicine Page 5 of 8 Page 5 of 8 (2024) 17:31 of newly enrolled clients; despite these challenges, the total number of clients in care increased 18%, whereas the number of facilities operated by AHF increased from six to eight units (Fig. 1). Russian invasion in Ukraine (Ukraine, February 2021) Russian invasion in Ukraine (Ukraine, February 2021) With independence from former Soviet Union in 1991, Ukraine is the second largest country in eastern Europe after Russia. Most of its 43 million inhabitants are Ukrai nians, but a significant minority are Russians, reflecting a long history of occupation and dispute, particularly over the eastern parts of the country and the Crimean Penin sula [16, 17]. l Finally, besides internal displacements, millions of Ukrainians sought refuge in neighboring countries since the beginning of the conflict. Taking advantage of AHF’s global operations, AHF Ukraine and AHF Europe orga nized a network to support and facilitate linkage to care of Ukrainians living with HIV who needed integration into health services in countries of transition or arrival. Ukraine is home to approximately 240,000 people liv ing with HIV, with key populations including people who inject drugs, prisoners, men who have sex with men, and sex workers being disproportionately affected [18]. AHF started operations in Ukraine in 2010 in collabora tion with the Ministry of Health and with regional health departments, offering comprehensive HIV care services at no cost for its clients. In the past 12 years, Ukraine reg istered a 47% reduction in new HIV infections and 81% reduction in Aids-related deaths [18]. Background destruction of the Mariupol AIDS Center, AHF Ukraine established new partnerships with three governmental institutions located in western regions, aiming to support people living with HIV who were displaced from east ern occupied regions. Remarkably, the number of newly enrolled clients went from 8,359 in the 12 months before the invasion, up to 26,067 in the year following the con flict onset; in 2023, AHF registered 26 facilities and more than 65,000 clients were under the responsibility of the Ukraine program. (Fig. 1). This striking increase in new enrollments resulted largely from admissions of displaced persons and patients whose original health services col lapsed during the conflict. Conclusion The emergency plan should then match the anticipated risks, including staff training on what to expect from the conflict, how to shelter in place, how to prepare for evacuation; defense infrastruc ture such as air raid shelters; streamlined communication l Since early 2022, health facilities and providers in Ukraine have been facing major challenges, with multiple reports documenting direct impact of the military con flict on healthcare infrastructure and personnel [19–21]. AHF-supported Mariupol AIDS Center was destroyed and became non-operational along with many other civil facilities located in southeast regions of Ukraine occu pied by Russian forces over the first months of conflict.l il Before the conflict onset, AHF oversaw 24 facilities and had more than 52,000 registered clients in care in the country. After the onset of the military conflict and the Miles et al. International Journal of Emergency Medicine (2024) 17:31 Miles et al. International Journal of Emergency Medicine Page 6 of 8 Page 6 of 8 (2024) 17:31 difficult to anticipate the occurrence of conflicts when scarce evidence is available for the intelligence analysis, or when events leading to the conflict occur in a very fast pace, as was the case in Myanmar. Moreover, the implementation of the emergency plan requires a solid relationship and credibility built with the institution’s leadership and staff; financial flexibility; and the ability to make adaptations as the conflict evolves. Of note, the structure and operating methods of the institution have central relevance for both the characteristics of the pre paredness plan and the ability to maintain local services. In the case of AHF, we work primarily through direct hir ing of local human resources, and hiring staff for deploy ment in foreign countries is not part of the institution’s routine operations. Therefore, during emergency situa tions, our staff generally desires to stay near their fami lies and communities, and the GDSS contingency plans focus on providing the resources to allow their continua tion on site, as long as safety conditions can be provided; evacuations are therefore restricted to extreme situations when measures are seen as insufficient to guarantee the safety of our collaborators. The staff’s safety and desire to stay are the first conditions to allow continuation of ser vices, and expansions very often occur naturally for those that are able to remain functioning throughout extreme events, resulting from the closure of other facilities and enrollment of displaced patients. Conclusion systems and cyber security; definition and periodic revi sion of evacuation routes; basic supplies for staff and their family members; and medical supplies/equipment to prevent discontinuation of patient care. On each of these steps, our team has experienced broad and diverse challenges, developing pragmatic solutions (Fig. 2). For instance, attacks to communication resources during extreme situations may block vital messaging systems that are crucially needed to establish and guide evacu ation routes; moreover, cyberattacks may compromise access to medical charts and patient confidentiality. Therefore, provisions must be adopted to guarantee alternative communication tools and improve data safety; additionally, a check-in system with daily updates about secure evacuation routes and triggering signs should be implemented during critical periods. Although AHF’s GDSS program focuses primarily on AHF personnel and assets, multiple elements of the pre paredness plan also benefit other local collaborators and partner institutions. For instance, all AHF contractors in Ukraine, including government-hired staff working in AHF-supported facilities, were connected to the commu nication system that allowed daily reporting AHF’s GDSS about their physical condition, basic supplies availability, and need to evacuate. As another example, AHF sup ported the renovation of Kyiv Hospital’s basement into a fully equipped 70-person bomb shelter. Even a well prepared and experienced security pro gram will face challenges and limitations. It may be very Despite unequivocal advantages, most health services and institutions cannot dedicate resources to constitute a Fig 2 S mmar of strategies to be prioriti ed in the preparedness plan healthcare organi ations Miles et al. International Journal of Emergency Medicine (2024) 17:31 Page 7 of 8 Page 7 of 8 Miles et al. International Journal of Emergency Medicine (2024) 17:31 (2024) 17:31 Funding Thi Funding This report has been developed with no specific funding. y p 14. The New York Times. Myanmar’s Health System Is in Collapse, ‘Obliterated’ by the Regime. [cited 2023 Sep 10]; Available from: https://www.nytimes. com/2022/04/19/world/asia/myanmars-coup-doctors.html#:~:text=After%20 14. The New York Times. Myanmar’s Health System Is in Collapse, ‘Obliterated’ by the Regime. [cited 2023 Sep 10]; Available from: https://www.nytimes. com/2022/04/19/world/asia/myanmars-coup-doctors.html#:~:text=After%20 Declarations formal Intelligence Program, and sometimes the value of these programs is underestimated as catastrophic events are perceived as extremely unlikely. Yet, it is still a good idea to consult expert knowledge and develop a pre paredness strategy. Think of it as a health insurance; you don’t plan to get sick, but in case you do, it’s better to be safe than sorry. Preparedness trainings and drills are rou tinely implemented to protect civilians from the damages caused by fire, earthquakes, and armed intruders; simi larly, a minimal set of safety strategies could be proposed to mitigate the impact of catastrophic situations such as military conflicts in healthcare facilities. A few safety ele ments could be broadly available in health institutions worldwide, including emergency communication tools, basic survival supplies, and an overall plan for sheltering and evacuation with clear trigger indicators. Recent con flicts have also revealed an appalling reality: healthcare services and providers are unfortunately not spared from violence during military conflicts. In fact, the presump tion of safety could lead to suboptimal preparedness and even negligence of risk indicators. Ethics approval and consent to participatei This manuscript includes reports of field experience and aggregated monitoring data; no individual-patient information was assessed in this study; no ethics approval or consent for participation was required for the development of this manuscript. References Other potential challenge that may interfere with healthcare operations during military or civil conflicts is the interference of ethnic or nationality contentions in the provision of care. For instance, although AHF did not experience situations where local health staff refused to treat people of certain ethnicities or nationalities, there were reported occasions when healthcare workers were forced not to provide health services for casualties who were from certain armed groups. These are dramatic sit uations with practical and ethical implications that also require dedicated discussions and preparedness. 1. UNAIDS. Ethiopia [Internet]. Available from: https://www.unaids.org/en/ regionscountries/countries/ethiopia. 2. BBC. Ethiopia country profile. [cited 2023 Sep 10]; Available from: https:// www.bbc.com/news/world-africa-13349398. 3. CNN. Tigray war fast facts. [cited 2023 Sep 10]; Available from: https://www. cnn.com/2022/11/11/world/tigray-war-fast-facts/index.html#:~:text=The%20 Tigray%20conflict%2C%20which%20began,and%20has%20fueled%20a%20 famine. 4. Insecurity Insight. Ethiopia. Violence against health care in conflict. 2021. [Internet]. Available from: https://insecurityinsight.org/wp-content/ uploads/2022/05/2021-Ethiopia-SHCC-Factsheet.pdf. 5. Doctors Without Borders. Ethiopia: MSF demands investigation into kill ing of three staff members in Tigray. [Internet]. 2021 Jul. Available from: https://www.doctorswithoutborders.org/latest/ethiopia-msf-demands- investigation-killing-three-staff-members-tigray#:~:text=The%20MSF%20 team%20members%20killed%20were%20Mar%C3%ADa%20Hern%C3%A 1ndez%2C,were%20traveling%20in%20a%20clearly%20marked%20MSF%20 vehicle. While man-made and natural disasters are expected to impact health services, there are steps that can be taken to improve staff safety and preserve patient care. Our experience suggests that a comprehensive and flexible preparedness program founded on expert safety knowl edge with continuous support for staff members and patients can mitigate the detrimental impact of military conflicts on healthcare. 6. Doctors Without Borders. People left with few healthcare options in Tigray as facilities looted, destroyed. [Internet]. 2021 Mar. Available from: https://www. msf.org/health-facilities-targeted-tigray-region-ethiopia. g g g y g p 7. World Health Organization. Crisis in Northern Ethiopia [Internet]. Available from: https://www.bbc.com/news/world-asia-pacific-12990563. 8. Gesesew H, Berhane K, Siraj ES, Siraj D, Gebregziabher M, Gebre YG, et al. The impact of war on the health system of the Tigray region in Ethiopia: an assessment. BMJ Glob Health. 2021;6:e007328. 9. BBC. Myanmar country profile. [cited 2023 Sep 10]; Available from: https:// www.bbc.com/news/world-asia-pacific-12990563. 7. World Health Organization. Crisis in Northern Ethiopia [Internet]. Available from: https://www.bbc.com/news/world-asia-pacific-12990563. 8. Gesesew H, Berhane K, Siraj ES, Siraj D, Gebregziabher M, Gebre YG, et al. The impact of war on the health system of the Tigray region in Ethiopia: an assessment. BMJ Glob Health. 2021;6:e007328. 9. BBC. Myanmar country profile. [cited 2023 Sep 10]; Available from: https:// www.bbc.com/news/world-asia-pacific-12990563. Author contributions RCM, VAS and ASB conceived the manuscript. WO and JH provided and interpreted data concerning HIV prevention and care indicators. VAS and RCM drafted the first version of the manuscript. FFF contributed with manuscript conception and development of Fig. 2. MG, YL, and WO provided data on local experiences and perceptions. All authors revised and approved the final version of the manuscript. 11. BBC. Myanmar coup: Thousands of Burmese flee to Thailand after intense fighting. [cited 2023 Sep 10]; Available from: https://www.bbc.com/news/ world-asia-65210336. 12. Insecurity Insight. Myanmar. [Internet]. Available from: https://insecurityin sight.org/country-pages/myanmar. 13. Insecurity Insight. Violence or Obstruction of Health Care in Myanmar. Janu ary 2023 update. [Internet]. Available from: https://insecurityinsight.org/wp- content/uploads/2023/02/January-2023-Violence-Against-or-Obstruction-of- Health-Care-in-Myanmar.pdf. Acknowledgements Acknowledgements We thank Thiago Avelino-Silva for developing Fig. 1. We thank Thiago Avelino-Silva for developing Fig. 1. 10. The New York Times. Myanmar’s Coup and Its Aftermath, Explained. [cited 2023 Sep 10]; Available from: https://www.nytimes.com/article/myanmar- news-protests-coup.html.l Competing interests Th h d l The authors declare no competing interests. Received: 10 October 2023 / Accepted: 22 February 2024 Received: 10 October 2023 / Accepted: 22 February 2024 Received: 10 October 2023 / Accepted: 22 February 2024 Author expertise Ralph C. Miles is the director of AHF’s Global Department of Safety and Security. He spent nine years serving as an officer in the U.S. Army in various missions around the world before initiating his civilian career. He has accrued more than 25 years of experience in the private sector working with the design, development, and implementation of security and intelligence strategies in a variety of businesses and organizations. Data availability All data generated or analyzed during this study are included in this published article. Miles et al. International Journal of Emergency Medicine (2024) 17:31 (2024) 17:31 Miles et al. International Journal of Emergency Medicine Page 8 of 8 the%20coup%2C%20thousands%20of,private%20hospitals%20and%20 underground%20clinics. 15. UNAIDS. Myanmar [Internet]. Available from: https://www.unaids.org/en/ regionscountries/countries/myanmar. 16. Britannica. Ukraine summary [Internet]. Available from: https://www.britan nica.com/summary/Ukraine. 17. BBC. Ukraine Country Profile. [cited 2023 Sep 10]; Available from: https:// www.bbc.com/news/world-europe-18018002. 18. UNAIDS. UNAIDS data 2022 [Internet]. Available from: https://www.unaids. org/en/resources/documents/2023/2022_unaids_data. 19. eyeWitness to Atrocities (eyeWitness), Insecurity Insight, the Media Initia tive for Human Rights (MIHR), Physicians for Human Rights (PHR), and the Ukrainian Healthcare Center (UHC). Destruction and Devastation. One Year of Russia’s Assault on Ukraine’s Health Care System. [Internet]. 2023 Feb. Avail able from: https://www.attacksonhealthukraine.org/. 20. Insecurity Insight. Ukraine. Violence Against Health Care in Conflict. 2022. [Internet]. Available from: https://insecurityinsight.org/wp-content/ uploads/2023/05/2022-SHCC-Ukraine.pdf. 21. Physicians for Human Rights. Coercion and Control: Ukraine’s Health Care System under Russian Occupation. [Internet]. 2023 Dec. Available from: https://phr.org/our-work/resources/coercion-and-control-ukraines-health- care-system-under-russian-occupation/. Publisher’s Note S i N i Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
https://openalex.org/W4240953488	https://zenodo.org/records/1448413/files/article.pdf	English	null	The Croonian Lectures on Degeneration of the Neurone: Delivered at the Royal College of Physicians of London	BMJ. British medical journal	1,900	public-domain	16,522	THE EFFECTS OF INJURY OF THE NEURONE. i li d d t ti f th THE EFFECTS OF INJURY OF THE NEURONE. d i f h ll Section of the axis cylinder or destruction of the nerve cell (speaking collectively), whether produced by direct trauma or by morbid processes occurring in the vascular and supporting structures of the nervous system, causes a simultaneous death change of a characteristic chemical nature affecting the myelin sheath and axon extending from the point of injury to its terminal fibrils. This is Wallerian degeneration, or secondary degeneration of Waller and Turck. Ranvier showedc that the sheath of Schwann consists of a series of tubular mesoblastic cells wrapped round the axon, and that when a nerve ia cut there is degeneration of the myelin and axon above the lesion as far as the next node. Each internode, therefore, represents a cell; and the fact that the degeneration does not extend upwards beyond the node allows the pre- sumption that these tubular cells have some metabolic inter- action with the contained axon and its surrounding myelin. But whatever that reciprocal metabolic interaction of the axon and its ensheathing structures may be, secondary Wallerian degeneration conclusively proves that it is con- trolled and determined by the nerve cell and its contained nucleus; indeed, it is possible to conceive that the metabolic activity of the nucleus is the mainspring of this trophic- influence, by the production of some bio-chemical or bio- physical stimulus which pervades the whole complex neurone even to its far distant prolongations. The Nissl method has shown that section of an axon causes changes in the nucleus and cell protoplasm; these may be temporary and recoverable or permanent and irrecoverable, the result apparently depends very much upon the degree and extent of the injury. When a peripheral nerve fibre is cut through, the trophic influence of the cell is removed from the axon and myelin sheath which it generated. These die; not so, however, the independent tubular cells of the sheath of Schwann, for these, soon after the death change has commenced in the myelin, show signs of active proliferation; whether, as some authorities would have us believe, they can lead to regenera- tion of the nerve, or whether, as seems much more probable, they prepare the way for the down-growing axon, I shall not here enter into; but that they have an important function in regeneration is undoubted, seeing that regeneration in the central nervous system. [JUNE 30, 1900. [JUNE 30, 1900. CROONIAN LECTURES. i58z MUsE:kLBRltzuN1ou ticular industry to which the laboratory belongs, but that original research of every kind in chemistry is encouraged. The cost of such laboratories is extremely large, for the salaries of the workers are said to be on a very liberal scale. And some of the researches (which occupy weeks or months of the time of the workers) and of the space and accessories of the laboratory, are of no direct use to the industry. Yet the industry, which is proprietary, and not the property of the State, or founded for the advancement of chemical science, derives so much advantage from a certain percentage of the discoveries and applications made in the laboratories, that the funds expended on them are well laid out. Those Englishmen who are familiar with these establishments tell me that this investment of funds in scientific research has served to place the German coloar industry far ahead of our own. ticular industry to which the laboratory belongs, but that original research of every kind in chemistry is encouraged. The cost of such laboratories is extremely large, for the salaries of the workers are said to be on a very liberal scale. And some of the researches (which occupy weeks or months of the time of the workers) and of the space and accessories of the laboratory, are of no direct use to the industry. Yet the industry, which is proprietary, and not the property of the State, or founded for the advancement of chemical science, derives so much advantage from a certain percentage of the discoveries and applications made in the laboratories, that the funds expended on them are well laid out. Those Englishmen who are familiar with these establishments tell me that this investment of funds in scientific research has served to place the German coloar industry far ahead of our own. LECTURE If. GENTLEMEN MR. PRESIDENT AND GENTLEMEN,---In my last lecture I brie fly referred to the present position of the neurone theory. I gave some examples of its practical applica- tion, and I dwelt at some length upon a method which has been used for studying the bio-chemical changes which occur in the protoplasm of the nerve cell. Moreover, I pointed out the important evidence afforded by the develop- ment of the myelin sheath in proving correlation of structure and function: and that the formation of myelin depends upon three fundamental biological principles: l i l y We want something of the same kind for the advancement of medicine and surgery, a direct and organised scheme for research and the application of pathology to our art. Of the advantage which may be derived from it I have no doubt. And if we fail to obtain it is because our forces are too divided and our views are not sufficiently in harmony. To those who think that the necessary funds would not be forthcoming, the answer must be that very considerable sums of money have, during the last few years, been devoted to medical research, and that the public mind is gradually becoming alive to the advantage to the nation of good health and the preservation of life. three g p p i. An acquired inherent tendency the result of evolutional differentiation of structure and function for certain systems, groups, and communities of neurones at successive periods of time to ensheath their axis-cylinders with an insulating and protecting structure, the myelin sheath, which in the case of the peripheral nerves is contained within, and partially formed by a chain of tubular mesoblastic cells. i ll i l f NOTES AND REFERENCES. d Cl i i B l 1708 1 Cours d'Op6ration8 de Cltirurgie, Brussels, 1708. 2 Traitd des Hernies, etc., par Pierre Franco de Turriers en Provence, demeurant & present a6 Orenge. A Lyon, par Thibauld Payan. 156I. 3 London, I779. 4London, I782. 5M'moires de l'Acadfmie Royale de Chirurgie, t. i, 1743; Pr6faee, p. xlv; "Le plan que se propose l'Acadlmie est d'hlever a chirurgie sur les observations, sur les rechercheR physiques et sur les expdriences." (Address on the Jubilee of the Patnologlcal Society of London, October 20th, I896, Pathological Transactions, vol. xlviii. 7Faraday as a Discoverer, by John Tyndall. London, I870. 8 The Chirurgical Works of Percivall Pott, a new edition, London, I779. 9 Hey began to practise in Leeds il I759. 10 Triumphs of Invention and Discovery in Art and Science, London, x88T. 11 The Collected Works of Sir Humphry Davy, Bart., LL.D., F.R.S., vol. vi, London, I840. 12 Evidence before the Committee of the House of Commons in 1833, Smiles's Life of George Stephenson, fourth edition, 1857. 13The rushing of the fire-damp from fissures in the strata of coal is called a ' blower." y 2. Phylogenetically and ontogenetically stimulus from without is the determiining factor in the bio-chemical change associated with the deposition of myelin. di i p y 3. Absence of stimulus as occurs from long disuse occasions a regressive metamorphosis back to the embryonic type. THE CROONIAN LECTURES ON DEGENERAT [ON OF THE NEURONE. Delivered at the Royal College of Physicians of London By FREDERICK W. MOTT, M.D.LoND., F.R.S., Pathologist to the London County Asylums; Physician in Charge of Out- patients, Charing Cross Hospital. Pathologist to the London County Asylums; Physician in Charge of Out- patients, Charing Cross Hospital. JUNIM 30, 1goo'l JUNIM 30, 1goo'l tions of the nuclei of the two sides make very evident changes, which occur as the result of section of the nerve on one side. like structures remain unaltered, while the axon alone de- generates; it would thus appear that the degenerating myelin may act as a formative stimulus, causing nuclear mitosis and cellular proliferation. Marinesco divides these changes which occur in the nucleus into three phases. The reaction phase occurs during the first month, and is characterised by a dissolution of the chromatophil elements, and a displacement of the nucleus towards the periphery of the cell. This change has also been recognised by many other observers- namely, Nissl, Ballet et Dutil, Lugaro, Flatau, Van Gehuchten, Sano, Erlanger, and others. The condition is not necessarily one of degeneration, as I sball show later on. It is important to bear this in mind, for a similar condition may be found affecting the anterior horn cells in peripheral neur- itis, where disease has led to a destruction of the axis- cylinder process. In the reparation phase, the cell body increases in volume, the nucleus resume3 its original position, and the cell shows a distinct pyknomorphie condition (abundance of stainable substance). About ninety days after the injury, the process of repara- tion has proceeded so far that the nuclear gTOUp of the injured side is only distinguished from the uninju:ed side by the increase in size of the cells and their procesmes. A little later the cells return to their normal appearance, in point of time corresponding to the regeneration of the nerve. We may consider this cell-bypertrophy as evidence of the in- creased physiological activity necessary for regeneration. If now the hvpoglossal nerve be torn out, according to the method of Von Gudden, or if a large piece of the nerve be re- sected, or any other condition arise so severe that the re- generation process in the nerve is prevented, then the reaction phase proceeds rapidly, and instead of a reparation phase of hypertrophy, the cells undergo atrophy, and a degenerative phase ensues. The accompanying diagram (Fig. A) shows the changes in the cells of the bypoglossal nucleas which result from section of the nerve in which regeneration is possible. It differs from tearing out the hypoglossal nerve, in which regeneration is impossible. JUNIM 30, 1goo'l In the former case the return of the cells to normal is coincident with regenera- tion and return of function; in the latter. the degenera- tion and subsequent atrophy of the cell group is brought about by such an injury as to render repair impos!ible, and re- sembles, therefore, the effect s ofamputation of a limab. Recently, Schafer has shown that hemisection of the spinal cord pro- duceschromolytic changes in the cells of Clarke's column and subsequent atrophy, substantiating tbereby the theory that these cells give origin to the dorEal and ventral cere- bellar tracts; and Lloyd, a pupil of his, has also shown that the cells of Deiter's nucleus undergo similar changes, thus tending to support the observations of Risien Russell, Ferrier and Turner that the cells of this nucleus give origin to descending fibres in the antero-lateral region. We can thus see that this method has proved of the greatest value in determining the cells of origin of tracts of fibres. Some observers, Bregmann and Darkschewitz, have even found, after tearing out cranial nerves, a degeneration recognisable by the Marchi method, and demonstrating the course of the nerve in the medulla or pons by the black-stained myelin. I have, by the kindness of Mr. Victor Horsley, the opportunity of exhibiting a lantern slide of the medulla showing degenera- tion back to the nucleus after an injury made just at the point of emergence of the nerve. Marinesco in his admirable work on the effects of amputa- THE EFFECTS OF INJURY OF A NERVE UPON THE CELLS OF ORIGIN. Only since the introduction of the Nissl method of staining has i,t become known that changes occur in the ganglion cells almost immediately after section of the nerve. Nissl, who first studied this, called it "primlare Beizung"; and Marinesco, who has made a special study of this subjett, has shown that section of the hypoglossal nerve on one side is fol- lowed by three stages of change in the corresponding nucleus of origin. g Fi.A.-Hypoglossal nuclei of two sides. In the upper part of the figure are shown the swollen cells of the nucleus on the side of section during the phase of reparation (after Marinesco). The hypogloasal nucleus is especially favourable for th study of this question, the nerve being purely motor. THE EFFECTS OF INJURY OF THE NEURONE. i li d d t ti f th where the sheath of Schwann is absent, does not occur. Tackett has shown that when a non- melullated sympathetic nerve is cut, the nuclei and sheath THE LIFE-SAVING SOCIETY.-We have received the annual report of the Life-Saving Society, wbich is now in the ninth year of its existence, and are pleased to sce that it is in a flourishing condition, and extending its useful work. It is, we believe, the only Society that organises systematic teach- ing of the various methods of rescuing the drowning and resuscitating those who are apparently drowned, and it has been the means of inducing many swimming organisations to make life saving an important element in their courses of instruction. By its instrumentality the subject has been introduced and successfully carried on by the aid of honorary teachers into many public and elementary schools and other educational bodies, and classes have also beenformed in connec- tion with the police, dock labourers, and Her Majesty's forces, the instruction given being theoretical and practical, on land and in the water. Another important branch of its work is arranging public demonstrations in various baths and docks, when large numbers of people, sometimes amounting to thousands, witness the graceful swimming and diving exer- cises practised, as well as the various modes of succouring and bringing safely to land the apparently drowning. This has perhaps done more than anything else to awaken a general in- terest in the work, and increase the demandfor information and instruction. The Society, which has its central office at 8, Bayley Street, Bedford Square, London, W.C, is supported by voluntary subscriptions, and is well worthy of being brought into greater prominence. REGISTERED MEDICAL PRACTITIONERS IN THE UNITED STATES. -According to the Medical Record, the average number of registered physicians in the United States to the whole popu- lation is about I to 647. The proportion in the States varies greatly; California appears to be the State which is most liberally supplied, the proportion there being i physician to 420 inhabitants, while in Iowa nearly the same conditions prevail. On the other hand, in North and South Dakota and New Mexico, the proportion was respectively I to 1,285, 8,296, and 1,391 inhabitants. TIM BR,2= IMEMIE"L JOUILNAL 1583 DEGENERATION OF THE NEURONF. Fi.A.-Hypoglossal nuclei of two sides. In the upper part of the figure are shown the swollen cells of the nucleus on the side of section during the phase of reparation (after Marinesco). The hypogloasal nucleus is especially favourable for the study of this question, the nerve being purely motor. From the close proximity of the two nuclei, comparative observa- JUNIM 30, 1goo'l The changes which occur in the cells of origin of a nerve after section undoubtedly prove that the whole neurone had suffered by the injury, but there is no evi- dence to show that a degeneration occurs (in the sense of the chemical change to be referred to later) except in the peri- pheral portions of the cut nerve, and in the central portion as far up as the next node of Ranvier, unless the injury is s0 severe as to cause such a shock to the neurone that it dies as a whole. DISEASES OF THE NERVOUS SYSTEM AS DISTINGUISHED FROM DISEASES WITHIN THE NERVOUS SYSTEM. DISEASES WITHIN THE NERVOUS SYSTEM. The histological elements which make up the nervous eystem may be divided into two groups; (i) the nervous units or neurones; (2) the supporting, protecting, and nutrient tissues. Organic diseases may start in a primary degeneration of the nervous units or neurones, or the neurones may be affected secondarily by diseases starting in the sup- porting, protecting, and nutrient tissues; such are essen- tially diseases within the nervous system, and include dis- eases of the blood vessels, lymphatics, membranes, and special nerve connective tissue (neuroglia). These give rise to secondary degeneration, either by direct injury, inflamma- tory compression, or by cutting off the blood supply. The causes of pathological processes occurring in the nervous system may be considered under two headings, external and internal, but it may be remarked that in all cases -except direct injury the two groups are more or less com- bined. The external causes depend upon the condition of the blood and lymph, by which the neurones are nourished, and the excess or deficiency of normal stimulation, or existence of abnormal stimulation. The internal causes depend upon the inherent vitality of the neurones themselves. In considering, therefore, the causes of degeneration of the neurone, it will be necessary to point out the result of (i) failure of the blood supply; (2) toxic conditions of the blood; (3) the effect of excess or deficiency of stimulation; (4) inherited defects in the nervous system as a whole, or in some particular groups or systems of neurones. To do this, however, in a satisfactory manner would necessitate my referring in a very superficial manner to all the diseases affecting the nervous system. JUNIM 30, 1goo'l Fro the close proximity of the two nuclei, comparative observa emergence Marinesco, in his admirable work on the effects of amputa- tion, describes slow progressive changes in the myelin sheath, and sal s the normal refraction of the fat disappears, the& colour reaction of the same tends to disappear, and following the changes in the myelin there is a charge in theaxis- cylinder. In addition, there appears a considerable nuclear proliferation of the sheath when the myelin has disappeared, the atrophied fibres bearing resemblance to non-medullated fibres. A recent observation by Noll, which 1 shall refer to later on, showed that the protagon diminished in the centram stump of a nerve after section, though of course, not to such an extent as in the peripheral. b i t ll peripheral. Barker, after referring to the above experiments, as well as to the studies of Von Gudden and his pupils, and the observ- ations upon the nervous system following amputation, re- marks: "We have seen that these observations bave partially at least annulled the validity of Waller's doctrine of the trophic relations of the nerve cells, for after injury to an Fi.A.-Hypoglossal nuclei of two sides. In the upper part of the figure are shown the swollen cells of the nucleus on the side of section during the phase of reparation (after Marinesco). h h l l l i i ll f bl f th LCUE.[uE3,1 CROONIAN LECTURES. Xnaa1 TM jftmn 1 1584 MMM" IOVM"-l Duwz 30, 19M Duwz 30, 19M to prevent destructive changes occurring in the nervous elements. With two carotids and two vertebrals tied the dogs are for some days paretic and demented, very much in the condition of the dogs which Goltz showed after having removed the cerebral hemispheres; however, as Sir Astley Cooper and numerous other observers have shown, these animals recover completely. Dr. Hill has been good enough to send me portions of the brains of twelve such animals ope- rated upon, which, after ligation of the four arteries, were al- lowed to live for varying periods of time.-I have examined the nervous elements with aviewof determiningthe conditionofthe nerve cells when the dog is in a state of paresis and dementia, and when it hasrecovered so completely as to pass for a normal animal. JUNIM 30, 1goo'l The methods which I have adopted have been for this and all other experiments and observations referred to, the Nissl method, the Marchi, the Marchi-Pal, and rapid Golgi methods of staining. The changes observed in the brain by the Nissl method, are shown in the accompanying photomicrographs of the sections (Figs. 2, 4, 5). Before, however, speaxing of the pathological changes, I will again call your attention to the appearances of the pyramidal cells of the oortex, medulla, and spinal cord of a normal animal, when examined by this method, for it is the most reliable we have at present for studying patho- logical changes by comparing the appearances presented with the normal cells (Figs. I and 3). You will observe that, first of all, the cells have changed their shape, they are swollen up, their edges have not straight or incurved sides, but are curved outwards, the nuclei are swollen and often eccentric, and in some few of the cells may actually be extruded; when this has happened, of course the cellhas practicallybeen destroyed, and with it the nerve fibre which contains its axis-cylinder projection. Another obvious sign of change, although not necessarily a destructive change, is the appearance presented by the chromophilous (stainable) substance. It no longer presents those definite figures which have received the name of Nissl bodies; sometimes it appears as a fine dust, some- times arranged in long threads separated from one another. One of the earliest changes to be observed is the absence or diminution of the chromophilous substance of the dendrons; moreover, the edges of the cells and the processes present a ragged instead of a clear-cut outline. Generally speaking, the changes in the chromophilous substanec only do not denote destruction, but should the swelling be so great as to produce an extrusion of the nucleus, the cell must necessarily die; and we can, if the animal be kept alive over ten days, determine how many psycho- motor cells have been destroyed by the temporary an emia occasioned by the ligation of the arteries. We can do this by examining sections of the pons, medulla, and spinal cord by the Marchi method; the number of black degenerated fibres will indicate the number of pyramidal motor neurones which have perished as a result of the anaemia. JUNIM 30, 1goo'l These are usually comparatively few; we may consequently say that mere swelling of the cells, with very marked chromolytic changes, does not mean more than functional disturbance. Mere hydration-for thus I explain the swelling-without dis- location of the nucleus from the cell, even though the chromo- lytic changes are very intense, does not mean death of the neurone. This is an important fact in view of the association of pathological symptoms with changes in the nerve cell. Such microscopical changes are, for the most part, illus- trated in the special plate by Figs. 2, 4, 5, 12. I would add in connection with this series of experiments that the tangential fibres were found in most cases intact in animals that had completely recovered. We must suppose, therefore, that collateral circulation in dogs is restored very shortly after ligation of four arteries. The experiments of Ehrlich and Brieger in I884 proved that ansmia of the lower portion of the spinal cord could be produced by ligature of the abdominal aorta, and that if the circulation was cut off for not longer than a quarter to three-quarters of an hour, only a temporary paralysis of the hinder extremities resulted; but if for one hour permanent paralysis and destruction of the nervous elements were invariably the result. Numbers of other experimenters have produced anaemia of the lumbar spinal cord by this method; and Sarbo has given a full account of the changes which occur in the cells as a result of l t i l ti f h t hour and half axon, in addition to the degeneration in the axon, peripheral to the lesion there are demonstrable alterations in the cyto- proximal end of the axon, and especially in the cell body of the neurone itself." The changes which occur after the ampu- tation of limbs are probably, however, due to regressive atrophy from lack of stimulus from the structures over which the neurones presided. JUNIM 30, 1goo'l The disease itself is, I consider, primarily a progressive decay of the nervous elements, but owing to the establishment of a vicious circle numbers of neurones are destroyed by circulatory disturbances. i i after ligation; and the changes which he describes as occur- ring in the motor cells of the spinal cord as the result of com- plete occlusion of the circulation coincide very closely with the changes which I shall describe in other animals, especially monkeys and cats; also in dogs, which have died as a result of ligation of cerebral arteries, and in which collateral circu- lation was not soon enough established to prevent destructive changes. Sarbo found that ligature of the aorta for one hour in the rabbit caused a cell degeneration, which was mani- fested by a granular destruction of the stainable sub- stance of the cell protoplasm, and a progressive homo- geneous atrophic process, affecting the cell nucleus. The appearances presented by the degenerating cells, depended upon the time which was allowed to elapse between the liga- tion of the aorta and the killing of the animal-the longer the time the more evident was the cell destruction. He found no vascular changes indicating inflammations and no heemor- rhages. in the pyramidal sysLems. Consequently we may assume that epileptiform seizures in this disease are associated with death of the cortical pyramidal motor neurones, due in some instances (I believe in great measure) to vascular disturb- ances of the cortex (Fig. 9). The disease itself is, I consider, primarily a progressive decay of the nervous elements, but owing to the establishment of a vicious circle numbers of neurones are destroyed by circulatory disturbances. i i y I will now pass on to the effects of experimental aniemia, produced by ligature of arteries in which collateral circulation was not restored soon enough to prevent destructive changes in the nerve cells. If cats or monkeys have all four arteries ligatured, or if instead of ligaturing both vertebrals and both carotids in dogs, a subelavian be tied instead of one of the vertebrals, a sufficient collateral circulation cannot be re- stored soon enough, and the animal dies in a period varying from a quarter to twenty-four hours. The time variation no doubt depends upon the effect produced upon the medulla by the anemia. JUNIM 30, 1goo'l The changes observed in the cells of animals which have thus lived some hours are different entirely to those which have been previously described. There is not merely a physical change due to hydration; this may or may not be present, for sometimes the cells are not swollen, but are even shrunken. The staining reaction is also different, showing a bio-chemical as well as a bio-physical change- sometimes the whole cell stains uniformly, but not with a brilliant coloration (Fig. 7). If a double stain has been used, for example, methyl blue and saffranine, the whole cell may be stained a uniform dull purple, the processes, as well as the body of the cell, having a homogeneous instead of a differenti- ated action to the dyes, a condition which is similar to that met with in hyperpyrexia, and to the appearances described by Sarbo in the motor spinal neurones of the lumbar-eacral region after clamping the abdominal aorta. These animals prior to death, according to Dr. Hill's notes, generally had epileptiform convulsions. One animal-a monkey-presented the most instructive changes, because after ligature of both carotids and one vertebral it was paretic and demented, took no notice of anything, and behaved exactly like an animal with its higher cortical centres destroyed, which, indeed, they were. It was killed on the fifth day, and examination of the brain of this animal exhibited the following changes, whichl are shown in the photomicrograph (Fig. 8). The nerve cells and all their processes were uniformly stained a diffuse dull purple, and were readily discerned on account of the dilatation of the lymph space in which the neurone lies; scattered through the protoplasm of the cells was a fine dust of coloured particles; the apical processes of the cells were either destroyed or twisted like a corkscrew; in many the dendrites had disappeared, but in some of the cells the axis-cylinder could be traced with unusual distinct- ness, probably due to some swelling; and one cell was observed which is of interest, because the axis-cylinder process could be traced some distance, giving off lateral twigs that seemed to merge into a general network, supporting therefore the diffuse nerve network theory of Golgi (Fig. 6). Many of the cells were swollen up, as shown in Fig. 8, others were shrunken. Some could be seen with phagocytes sticking to them and devouring the dead cells. JUNIM 30, 1goo'l Although one observed these very marked changes in the cells by the Nissl and Weigert iron methods, yet by the chrome-silver method the cells appeared normal, showing the gemmules on their processes with unusual distinctness (photomicrographs 1o and ii). Thebrainsofother animals, dogs and cats, were examined by similar methods with the same results. It may be that the chrome silver only stained those cells which had not undergone disorganisation; but none of the cells when examined by the methods I bave alluded to exhibited normal appearances as regards their bio- chemical condition, nor was the animal physiologically normal. I have, therefore, oome to the conclusion that the rapid silver method of Golgi, which stains only a few cells, although un- doubtedly of the greatest value for anatomical purposes, yet for acute pathological findings is unreliable. Many of the statements which are made with regard to the effects of drugs, aniesthetics, and acute disease as represented by the appear- ances of the gemnmules on the processes of the cells, as made evident by the chrome-silver method, are therefore open to serious objections. It is very probable that the method of Cox might be found of value in determining changes in the cells of g The subject of cerebral anaemia is of clinical importance, as it is well known that in man, ligature of a carotid for aneurysm has been followed by hemiplegia, and I have examined the brain in one such case. Dr. Hill sent me a piece of the cortex from similar situations in both hemispheres, and asked me to decide by microscopical examination upon which side the patient was paralysed. Tiais I correctly determined by the difference in the appearances of the cells of the two sides. We can understand also, how temporary loss of motor or psy- chical function, for example, transitory aphasia, monoplegia, hemiparesis, stupor, and dementia, or even hemiplegia, may occur in syphiiitic disease of the arteries; for if collateral cir- culation be restored within an hour or so after the blocking of the artery, providing a thrombus does not spread to the small branches, the circulation may be restored in the part, and function may return in the course of a few hours or days. JUNIM 30, 1goo'l I ti ith th i t l i f th i l and function may return t e cou se hours y In connection with the experimental ansemia of the spinal cord, produced by clamping the abdominal arteries, it is of interest to note two points. The exogenous fibres of the pos- terior columns, together with the pyramidal tracts-tracts, the fibres of which have their trophic and genetic centres re- spectively in the spinal ganglia and the cerebral cortex-do not undergo degeneration, whereas tracts which have for their trophic and genetic centres cells in the grey matter of the spinal cord, undergo degeneration. Munzer's and Wiener's ex- periments showed this, and recently I had the opportunity of examining a case of acute anterior poliomyelitis in an infant fourteen days after the onset of the disease; in this case also the degeneration was limited entirely to those tracts, the fibres of which arise from cells in the spinal cord, and I can- not help thinking that this case was of vascular origin. [A full account of this case is given in the Arch. of Neurol., vol. i, p. 365.] Ballet and Dutil found from their experiments that complete anwemia of the lower end of the spinal cord for a few minutes (by clamping the aorta) produced chromolytic changes in the anterior horn cells. I have also found that swelling of the cells previously described with chromolytic appearances may come on within less than ten minutes after ligation of the four cerebral arteries, for occasionally one of Dr. Hill's dogs or cats has died from the effect of the anaesthetic very shortly after the four arteries have been liga- tured. It maybe said that this was due to the anwesthetic ; It was not, however, so, for occasionally I have had the oppor- tunity of examining the brain tissues of animals which have died from the anaesthetic before any of the arteries were tied, and these showed no changes (Fig. 3). We may therefore consi- der that the changes noted are due to the cessation of the circu- lation. Dr. JUNIM 30, 1goo'l I shall, therefore, limit myself especially to personal observa- tions relating to some few causes of the degeneration of the neurone in each of those groups which I have studied. EFFECTS OF TEMPORARY OR PERMANENT FAILURE OF THE BLOOD SUPPLY UPON THE NEURONE. h ff t d d b l t i f i The effects produced by complete anaemia for varying periods of time upon the nervous elements of the spinal cord have been the subject of numerous experimental inquiries, but I am unaware of any systematic series of observations upon the effects produced by ligation of the cerebral arteries. Many clinical facts show that this is a subject worthy of atten- tion, for the sudden loss of consciousness which results from syncope is explained by a failure of activity on the part of the nervous elements of the cortex cerebri, in consequence of default in the circulation in the hemispneres. Again, the sudden loss of consciousness in epilepsy, followed imme- diately by tonic spasms and then clonic spasms, has recently found a possible explanation in the series of very valuable experimental observations by Dr. Leonard Hill. He has found that artificial cerebral anaemia in cats and monkeys produced by ligation of the four cerebral arteries produces tonic spasm. If absinthe be injected, instead of causing clonic spasm, it increases the tonic spasm. Then, if the clamp or ligature be loosened on a carotid so that blood flows back to the hemisphere, clonic spasms almost immediately occur. This will again give place to tonic spasm on closing the artery: and again, on removing the clamp and allowing the blood to flow to the hemisphere, the clonic spasms supervene. He has, moreover, shown that if all four arteries in an animal be ligatured, in the dog recovery takes place; the ex- planation is that collateral circulation is restored soon enough DEGENERATION OF THE NEURONE. T BRrr 55 L&i ]DTC ,L JOlUNAL 5 T BRrr L&i ]DTC ,L JOlUNAL JUNE 30, 1900I. JUNE 30, 1900I. in the pyramidal sysLems. Consequently we may assume that epileptiform seizures in this disease are associated with death of the cortical pyramidal motor neurones, due in some instances (I believe in great measure) to vascular disturb- ances of the cortex (Fig. 9). JUNIM 30, 1goo'l Watson in my laboratory has carefully compared the sections of experimental anaemic brains with a large number of sections of general paralysis; he has made drawings of cells which he considered presented similar appearances in the two conditions, and the result of his observations appears to show that whilst a large number of the cells exhibit changes indicating chronic atrophy, a considerable number of cells in general paralysis present acute destructive changes similar to those met with in cases of experimental anremia from arterial occlusion, but they are not due to the same cause, for the arteries in this disease are seldom occluded, but congestive stasis in the arterioles, capillaries, and veins, even thrombosis in the latter is common, and this congestive stasis probably takes part in the destruction of the nervous tissue. i takes part in the destruction t e It is curious how general it is to find the brain atrophy in this disease limited particularly to the frontal and central convolutions, whereas the occipital and temporal lobes escape in great measure. Over the atrophied regions the pia-arachnoid membrane is thickened and adherent. Many observers, including Flechsig and Mickle, consider that the region of atrophy corresponds particularly with the area of distribution of the internal carotid arteries. I think, how- ever, the area of atrophy and pia-arachnoid thickening more closely corresponds with the distribution of veins which open into the longitudinal sinus, and from a large number' of ob- servations made by myself and my assistants, I have come to the conclusion that venous congestive stasis in that portion of the brain plays an important part in the symptomatology and pathology of general paralysis. I shall, however, have occasion to refer to this more fully later on. h i i b li th t l l i de d o s o t e spinal oto Effects of Hceemorrkage.-Large quantities of blood may be lost to the body from hwemorrhage without producing any appreciable changes in the nervous elements. Voss has done a number of experiments on animals, endeavouring to pro- duce degeneration of the spinal cord, by Injections of pyridine eufficient to produce a severe anaemia, but without results. Likewise, numbers of cases of pernicious anremia, leucocyt- hemia, and exhausting blood disease, are not found associated with any noticeable change in the neurones. JUNIM 30, 1goo'l Such collateral circulation when established (and if the animal lives it must have been established) would be accompanied by blood supply to the choroid plexus; and the cerebro-spinal fluid, which functions as the lymph of the brain, would con- sequently be secreted in abundance and soon compensate the diminution caused by the failure of the arterial circulation; thus the increase of cerebro-spinal flaid would account for the .ceiema of the brain, the dropsical condition of the cells, and the dilatation of the perivascular lymphatics. p y p Some experiments which I made five years ago with Professor C. S. Sherrington, and which we have never found opp:rtunity to continue and develop, are (although only few in number), I think, of sufficient importance to mention with this subject of ansamia. We found that compression of the spinal cord in the dorsal region, even enough to produce indentation, would not stop the passage of the impulses, generated by faradic excitation of the cerebral cortex, to the lower spinal motor neurones. Whereas in two animals- monkeys-upon compression with an eye removed from a rat immediaWely alter death, fixed at the end of a glass tube attached to a horizontal lever in such a way that compression of the lumbar spinal cord could be male by this living tissue, and the pressure measured by means of a sliding weight-we found that extremely ligbt pres- sure upon the lumbar spinal cord was sufficient to abolish conduction in one minute to a minute and a-half after the pressure was applied; and upon removing the pressure conduction, in a minute or two returned, because faradic excitation of the cortex was responded to by an appropriate muscular contraction. The experiment was repeated a con- siderable number of times, and I have d) doubt that in these two animals we managed to produce an anaemia of the grey :rmtter of the lumbo-sacral spinal cord by compression of the large artery that comes in about the third lumbar root, causing a physiological block in the synapsis formed by the terminal arborisatious of the axons of the cortical pyramidal neurones with the dendrons of the spinal motor neuronee. i i the cell body a d ts nucleus. Dr. JUNIM 30, 1goo'l f ll d th individuals who suffer from this nervous affection have an in- herent defective vitality or lowered durability of the nervous system, which, in combination with the defect of the blood supply in quantity and quality, leads to degeneration. The fact that inanition, even extreme inanition, produces little or no loss of weight in the nervous system, and hardly any changes of importance in the microscopical appearances of the nervous elements, is of considerable importance, for it seems to show that the metabolic exchange that goes on in the nervous system, is either especially protected by the metabolism of the whole body being subservient to it, or that, however important the metabolic exchange in the nervous system may be, in amount it is not considerable. Those primary wasting di- seases affecting systems, communities, and groups of neurones, associated one with another by functional relationship rather than by anatomical or vascular supply, insidious in origin and progressive in course, cannot be ostensibly due to anything else than a metabolic failure on the part of the neurones them- selves (which, after all, are but complex differentiated cells) to assimilate from the blood the necessary materials to maintain their specific vital energy. They die in a manner the reverse of their evolution; the most distant parts of the tree, namely, the terminal twigs (collaterals) being the first to go, then the branches, and last of all the trophic and genetic centre itself, the cell body and its nucleus. f ll d th the cortex in chronic disease, and possibly also in acute disease. I have not mentioned that in moss of these experimental brains a striking naked-eye feature was the distension of the veins of the cortex with blood, and frequently there was subpial hsemorrhage; also the perivascular lymphatics were usually distended with flaid; no doubt both the in- crease of blood in the superficial veins, the increase of fluid in the perivascular lymphatics, and the general cedema of the brain itself were proportional and compensatory to the diminution of the arterial blood; the arteries were generally empty, likewise the capillaries. In some cases the perivas- cular lymphatics and membranes were the seat of a marked cellular infiltration, resembling the condition found in general paralysis. Collateral circulation to the medulla, by which the respiratory centre is kept going, is essential to life. JUNIM 30, 1goo'l Hill has shown that the cortex of these animals is readily excitable, and in some cases even hyperexcitable to electrical stimulations, epilepsybeing produced; consequently this change in the appearance of the cells can be associated with increased electrical excitability, or perhaps more likely, increased tendency of stimulus to spread, a fact of consider- able importance when put beside another fact, namely, that I have found in status epilepticus the oellspresentingin manyin- stances this swollen dropsicalappearance, with marked changes in the chromophilous substance (Fig. 13). In general paralysis of the insane epileptiform seizures are of frequent occurrence, and when they do occur in the progress of the disease I have found invariably that in proportion to their frequency and severity, there is a large number of recent degenerated fibres CROONIAN LEOTURES. LJUNE 30, 1900.. LJUNE 30, 1900.. 1 586 rM JBP individuals who suffer from this nervous affection have an in- herent defective vitality or lowered durability of the nervous system, which, in combination with the defect of the blood supply in quantity and quality, leads to degeneration. The fact that inanition, even extreme inanition, produces little or no loss of weight in the nervous system, and hardly any changes of importance in the microscopical appearances of the nervous elements, is of considerable importance, for it seems to show that the metabolic exchange that goes on in the nervous system, is either especially protected by the metabolism of the whole body being subservient to it, or that, however important the metabolic exchange in the nervous system may be, in amount it is not considerable. Those primary wasting di- seases affecting systems, communities, and groups of neurones, associated one with another by functional relationship rather than by anatomical or vascular supply, insidious in origin and progressive in course, cannot be ostensibly due to anything else than a metabolic failure on the part of the neurones them- selves (which, after all, are but complex differentiated cells) to assimilate from the blood the necessary materials to maintain their specific vital energy. They die in a manner the reverse of their evolution; the most distant parts of the tree, namely, the terminal twigs (collaterals) being the first to go, then the branches, and last of all the trophic and genetic centre itself, the cell body and its nucleus. JUNIM 30, 1goo'l I have examined the nervous system in several cases of pernicious anremia, in one of which the corpuscles had sunk to 500,000 per c.mm.; cases of very severe anaemia from hremorrhage and exhaustiDg disease; leucocythemia, in which the red corpuscles were re- duced to one-fourth, and in which the whole of the brain and vessels showed congestive stasis and haemorrhages ; and yet the nerve cells of these cases of extreme ansemia showed little or no bio-ebemical change, the Nissl granules appearing quite normal (Fig. I); nor did sections of the nervous system in these case3 show any degenerative changes by the Marchi method. It is therefore more than probable that those cases of combined sclerosis of the spinal cord, sometimes met with in pernicious ansemia, but more often accompanied bv grave anreinia, are due to some other cause than the deficiency of red corpuscles; probably in the blood there is some neuro. toxin which produces this primary systemic degeneration of neurones, with long axis cylinder projections, forming the long tracts in the spinal cord-or it possibly may be that those occasion to fully Mendel and many authorities believe that general paralysis is a primary inflammatory condition of the vessels and meninges, with secondary destruction of the nervous ele- ments. In i884, Mendel, who believed that the necessary factors were (i) diseased walls of vessels allowing transuda- tion of plasma and leucocytes; and (2) hyperremia of the brain, claimed to have produced this condition artificially by rotating dogs upon a table with the heads outwards. Fiirstner repeated these experiments over a considerable period of time, causing degenerative changes in the cortex, with secondary descending changes in the cord. There are many reasons, however, which will be alluded to later, for consider- ing general paralysis to be a primary progressive decay of the nervous elements, with secondary changes in the vessels and membranes. ALTERED STATES OF THE BLOOD. Hyperpyrexia. f th l i f t d or more, I have found the same bio-chemical change revealed by the Nissl method in all the cells of the central nervous sys- tem. I have, however, met with cases of status epilepticus in which the temperature was 1070 to 107.50 F. fora short time without showing this change. Again, I have seen in one case a moderate degree of chromatolysis and diffuse staining in prolonged pyrexia, varying from io50 to I06.50 F. This was a case of typhoid fever, and it might very well be that the change was due to the influence of the toxin; in fact, in most of the human cases the changes could be ascribed to the con- ditions which produced the fever and not to the influence of the increased temperature of the blood; but some of the cases which I have examined were so sudden, for example, one was probable sunstroke, another was a case of hb emorrhage into the spinal canal and the base of the brain in general paralysis, another was a case of Congo sickness. y It is matter of speculation whether structures later de- veloped and functionally more highly differentiated, as the cells of the cerebral cortex, are more highly susceptible to fever, and succumb more readily under it. PoEsibly high temperature of the surrounding blood and lymph is moie readily felt by the small cells of the superficial layers of the cortex on account of the increased surface exposed, and de- lirium followed by coma may be the result of this; although the vital centres in the medulla may still be able to perform their functions. diff i i diff i f h The diffuse staining of the cells indicates a diffusion of the nucleo-proteid through the substance of the cell body and its processes. We may suppose, therefore, that the essential achromatic fibrillary substance is killed, and that the nucleo- proteid which is normally contained in the reticulum of the cell in solution has soaked into the achromatic substance, and given the protoplasm the uniform staining, which when once general is quite characteristic of hyperp3rexial death, as I have been able to verify in several instances. These facts, pointing to an actual bio-chemical change involving the death of the protoplasm, when the temperature remains for some hours above IO9 F., support Dr. ALTERED STATES OF THE BLOOD. Hyperpyrexia. f th l i f t d If the temperature exceeded I09.50 F., (430 C.) the lesions of the nerve cells of the spinal cord were very definite, extend- ing throughout the whole of the grey matter. g g g y 3. The duration of the experimental hyperthermia they showed to be a very important factor in the production of the lesions ; for if the animal be kept for some hours between o70 and so8O F., the same effect on the nerve cells is produced as occurs in a much more rapid manner when the temperature is raised to I09.50 F.-namely, swelling of the cell and its pro- cesses, diffuse staining of the whole neurone with disintegra- tion of the chromophilous substance, and around the nucleus the chromophilous elements may still persist to some extent as if the outer portions of the cell were more affected than the central. The nucleus has an irregular and often an angular appearance. changes by regards staining. Goldscheider and Marinesco have shown experimentally that a certain degree of this coagulation change may take place without death of the cells, for if they produce artificially hyperthermia in animals for only a short time and then kill them, they find a commencement of this coagulation process to have taken place; but these animals would have lived had they not been killed for the purpose of examination, conse- quently we may suppose that a certain degree of bio-chemical change, associated with coagulation of neuroglobulin, may take place without destroying the protoplasm to such a degree as to render it incapable of recovery. I think this entirely coincides with and explains the remission of symptoms, the return of consciousness and the recovery, which often takes place in some forms of hyperpyrexia, when the cold-bath method is resorted to without delay. pp Marinesco has also done some valuable work on this subject. He found that if artificial hyperthermia in animals be produced, so that the temperature is raised to i 6.60 F. in the rectum, the animal died in thirty minutes with characteristic changes in the nerve cells. He also remarked that these changes occur at a lower temperature when the duration is increased. In five cases of hyperpyrexia occurring in the human subject, which I have examined, and in which the temperature reached I090 F. ALTERED STATES OF THE BLOOD. Hyperpyrexia. f th l i f t d Hyperpyrexia. The study of the lesions of nervous centres due to hyper- pyrexia is of great interest from a theoretical as well as from a practical point of view; from the former, because the exact knowledge of the nature of the lesion might be able to throw some light upon the nature of the pathological process and the clinical symptoms which accompany hyperpyrexia; from the latter, because it would indicate the direction of treat- DEGENERATION OF THE NEURONE, VUNE 30, 1900.] THz RITIO8X LMEDICAL JOURNAL THz RITIO8X LMEDICAL JOURNAL _ 1l FF~~~~~~~~~~~~~~~~~~~ s_ 1l'aai! Fis2. 2. ~~~~~~~~~~~~~~~~~~~~~~~~~ Fig. 3. _ 1l FF~~~~~~~~~~~~~~~~~~~ s_ 1l'aai! Fis2. 2. ~~~~~~~~~~~~~~~~~~~~~~~~~~ Fig. 3. Fis2*. 2. Fig. 3. Fig. 5. Fis2 . 2. Fig. 5. s . 2. Fig. 4. Fig. 6t Fig. 6t Fig. 4. Fig. 4. Fig. 5. Fig. 6t Fig. 4. &. -.. P%. I o Fig. 7. Fig. 4. &. -.. P%. I o Fig. 7. Fig. 4. &. -.. P%. I o Fig. 7. Fig. B. Fig. 7. Fig. B. DEGENERATION OF THE NEURONE. Tug BRiTisn 1 MEDICAL JOURNALJ [JUNE 30, I1900 DEGENERATION OF THE NEURONE. [JUNE 30, I1900 Fig. 11. Fig. 14. Fig. 15. Tug BRiTisn 1 MEDICAL JOURNALJ Fig. 10. Flg. 12. Fig. 13. Fig. 11. Fig. 10. Fig. 10. Flg. 12. Fig. 10. Fig. 11. Flg. 12. Fig. 11. Fig 15 g Fig. 14. Fig. 14. Fig. 13. Fig. 13. Fig. 13. Fig. 14. Fig. 15. - Fig. 16. Fig. 13. Fig. 15. Fig. 17. g 15. Fig. 17. - Fig. 16. Fig. 17. DEGENERAuION O3F THE NE5RONE. Tmu BILLWJA. IMIRDIC&L JOU&MAL JUNR 30, .1900-1 JUNR 30, .1900-1 1587 lated at 132.80 F. (560 C.); (3) a neuroglobulin which is pre- cipitated at I670 F. (750 C.); and that these three bodies can be separated out by fractional coagulation. We see, then, how it is that Marinesco found that animals in which the temperature is raised to I I6.60 F. (470 C.) died almost immediately with changes in the cells indicating co- agulation in the protoplasbm. No doubt this neuroglobulin is precipitated at once, and it occurred to me that if a solution of neuroglobu,in were subjected to long heating, it would coagulate at a much lower temperature. I mentioned the matter to Dr. ALTERED STATES OF THE BLOOD. Hyperpyrexia. f th l i f t d Halliburton, who was of the same opinion, and we have therefore made experiments upon this subject, and we find that prolonged heating for four hours, for example, will cause coagulation of the neuroglobulin at a temperature of between IO7.60 and 109.40 F. (420 and 430 C.) Brodie and Richardson have shown that in frog's muscle the tissue loses it irritability at that temperature, at which coagulation is induced first in one of its constituent proteids. We bave thus a chemical explanation of the cause of death of the protoplasm, also of the changes manifested by it as regards staining. ld h id d i h i t ll lated at 132.80 F. (560 C.); (3) a neuroglobulin which is pre- cipitated at I670 F. (750 C.); and that these three bodies can be separated out by fractional coagulation. We see, then, how it is that Marinesco found that animals in which the temperature is raised to I I6.60 F. (470 C.) died almost immediately with changes in the cells indicating co- agulation in the protoplasbm. No doubt this neuroglobulin is precipitated at once, and it occurred to me that if a solution of neuroglobu,in were subjected to long heating, it would coagulate at a much lower temperature. I mentioned the matter to Dr. Halliburton, who was of the same opinion, and we have therefore made experiments upon this subject, and we find that prolonged heating for four hours, for example, will cause coagulation of the neuroglobulin at a temperature of between IO7.60 and 109.40 F. (420 and 430 C.) Brodie and Richardson have shown that in frog's muscle the tissue loses it irritability at that temperature, at which coagulation is induced first in one of its constituent proteids. We bave thus a chemical explanation of the cause of death of the protoplasm, also of the changes manifested by it as regards staining. i d i h i t ll ment. Goldscheider and Flatau, who for the first time de- scribed lesions of the nerve cells in the rabbit, caused by experimental hyperthermia, came to the following conclu- eions: eions: I. If the temperature remained about I06.70 F. (420 C.), the cells of the spinal cord examined by Nissl's method presented no appreciable modifications. 09 0 pp 2. ALTERED STATES OF THE BLOOD. Hyperpyrexia. f th l i f t d Osler's view that cases of paradoxical temperature occurring in women, and termed hysterical hyperpyrexia, are frauds as a rule, although he states that other cases have to be accepted, the explanation of which is impossible under known rules; in fact, it is quite conclusive that temperature above 1140 F. is incompatible with life, even for a short time. g The photomicrograph shows three anterior horn cells from the last case. You will observe the diffuse staining of the processes and of the body of the cell (Fig. 14). Another case of Congo sickness, in which the temperature never went above 103 80 F., showed no change in the Nissl granules, and the cells presented a fairly normal appearance. have examined the nerve cells in many cases of septic poisoning and other diseases, but I have never seen this change unless there was eitherhyper- pyrexia or prolonged high fever. The fact that experimental ilyperthermia in animals produces this bio-chemical change in the protoplasm of the cell indicate3 that it is the altered temperature of the blood which is the cause in the human subject of these changes. Another point is that if the see- tions are stained with methylene blue and saffranine, the pro- toplasm is stained a uniform dull purple, instead of being differentiated. Now we have seen that the healthy cell shows a differentiation of staining; the Nissl granules, which have been shown to be a nucleo-albumen, are no longer visible in hyperpyrexia; and there is no longer the differentiation of a stainable and an unstainable substance with basic dyes ; there must, therefore, have been produced by the fever a pro- found modification of the protoplasm of the cell. There are two points to be considered in this-namely: 109 50 Toxic CONDITIONS OF THE BLOOD AND LYMPH. h d i h i i th The primary degenerative changes occurring in the neurone are usually the result of poisonous conditions of the blood and lymph. The poisons may be produced: i h lymph. poisons may produced: i. Within the body by perverted functions of the-organs or tissues (auto-intoxication). i f i i th li i fl id p y i. The fact that a temperature of 109.50 F. can bring about rapidly this coagulation process and death of the protoplasm, also that it affects first and most markedly the periphery of the cell. ( ) 2. ALTERED STATES OF THE BLOOD. Hyperpyrexia. f th l i f t d Perhaps one of the most striking demonstrations of this selective affinity of protoplasm for particular chemical com- pounds isshown by the intra-vitam methylene-blue method introduced by Professor Ehrlich, and which, in the hands many investigators, has produced such marvellous revelations into the minute structure of the nervous system. found that only those aniline dyes which contained sulphur in their composition were suitable for yielding the action. g Marinesco maintains that in nerve cells subjected to the action of the tetanus toxin, two independent phenomena can be distinguished; one, the result of the chemical combination of the poison with the cellular protoplasm, and the other the wearing out of the cell from excessive functional activity. He agrees with the conclusion of Goldscheider that there is a chemical transformation exalting the excitability of the motor neurones. There is a general consensus of opinion among authors who have worked at this subject. including those named, that the injection of tetanus antitoxin retarded these cell changes, and caused a more rapid return to normal; it is only right to say that Courmon, Doyen, Poirot (ref. Ford Robertson), from experimental results of their own, deny that tetanus toxin produces any alteration in the cells of the anterior horns, and certainly, specimens of the medulla and pons from two cases of tetanus in the human subject, which I have had the opportunity of examining, showed no certain noteworthy change. A th i b bl f i bi l i i bi h y g Another poison, probably of microbial origin, rabies, has a special selective influence on the medulla, although the whole nervous system is charged with the poison; pointing, there- fore, to the fact that there is some special affinity of the neu- ronesinthemedullaforthepoison. I have had the opportunity, through the kindness of Dr. H. Durham, of examining the me- dulla of a child, who died of hydrophobia in Guy's Hospital six weeks after having been bitten by a mad dog. The greater number of the cells of the motor nuclei of the medulla showed markeddegenerativechanges indicating a slow process of coagu- lation necrosis, namely, uniform diffuse staining with absence of Nisslgranules ; there wasalso agreat deal of nuclearprolifera- tion inthe glia substance, denoting an irritative as well as a degenerative process. ALTERED STATES OF THE BLOOD. Hyperpyrexia. f th l i f t d He was also kind enough to give me the nervous systems of three animals which had been poisoned with the toxin of" bacillus botulinus," a disease first described by Ermenghen, who discovered the organism in unsound meat and fish eaten by people who had died with a characteristic group of symptoms. Such a condition in this country would probably have been called" ptomaine poisoning." The changes which I have found in the nerve cells are similar to those de- scribed by Marinesco, who has specially studied this subject. There is a marked chromolytic change, some cells being much more affected than others; the poison is extremely virulentand, if a minute dose is given, so that the animal does not die before the end of a week or so, the changes in the nerve cells are not more marked than in an animal which died within 24 hours (Fig. i6). The most noteworthy change, however,which I have found in ananimal which survived seven days, was the ex- treme fatty degeneration of the heart and striped muscles of the body (Fig. 17). The liver also had undergone extreme fatty degeneration. Seeing that the peripheral nerves showed no degenerative changes, one may conclude that the poison acted more particularly upon the muscle, or the nerve endings in muscle. As I have already had occasion to point out, chromolytic changes of themselves do not indicate death of the nerve cell, but rather functional depression; seeing that there were no changes in the peripheral nerves or in the cord, observable by the Marchi method, it is probable that the changes in the nerve cells were mostly related to functional depression rather than actual destruction, and might have arisen from circulatory failure in the central nervous system, due to paralysis of the heart. Marinesco noticed that the Examples of the special selective influence of the poisons which produce disease are numerous. There is whole class of poisons which affect specially the peripheral nerves; and it is a debatable point whether in peripheral neuritis the whole neurone suffers, or whether the changes which have been found in the cells of origin of the peripheral nerves, in the spinal cord and ganglia, are due primarily the action of the poison; or whether they are secondary to the destructive effects of the poison upon the outgrowths of cells-namely, the axis-cylinder processes of the peripheral nerves. ALTERED STATES OF THE BLOOD. Hyperpyrexia. f th l i f t d We know that strych- nine, for example, has a special selective influence upon the spinal cord, increasing reflex excitability. Gold- sheider and Flatau have shown that strychnine poisoning produces changes in the motor anterior horn cells, closely corresponding to those observed in tetanus poisoning; these changes may be observed as commencing three minutes injection of the poison, if a sufficient dose be given. As tetanus poisoning, there is no close relationship between morphological change and the degree of functional disturb- ance. They have also shown that acute cell changes occur as result of malonitril poisoning (NC-CH2-CN); injection sodium hyposulphite neutralises the toxic effects of this poison, and prevents the changes in the nerve cells. poisons which produce degeneration and disease a special selective influence. We know that strych- nine, for example, has a special selective influence upon the spinal cord, increasing reflex excitability. Gold- sheider and Flatau have shown that strychnine poisoning produces changes in the motor anterior horn cells, closely corresponding to those observed in tetanus poisoning; these changes may be observed as commencing three minutes injection of the poison, if a sufficient dose be given. As tetanus poisoning, there is no close relationship between morphological change and the degree of functional disturb- ance. They have also shown that acute cell changes occur as result of malonitril poisoning (NC-CH2-CN); injection sodium hyposulphite neutralises the toxic effects of this poison, and prevents the changes in the nerve cells. Experiment has shown that, curare poisons the motor plates, and recently, in an interesting communication, Waller showed how a very slight difference-a molecale of water-will alter the action of an alkaloid. Thus he has shown that veratrin has a special selective toxic action upon muscle, while proto-veratrin acts especially upon These facts, together with others which I will cite connection with degenerative changes produced by selective action of poisons in the production of disease, cate a special chemical affinity of certain protoplasmic struc- tures for certain particular poisons, and the corollary is that probably the protoplasm of every nerve structure having different function varies in some slight degree. There is difference which is not to be detected by chemical methods, but the varieties of which are clearly shown by the difference in their physiological reaction to an altered environment. ALTERED STATES OF THE BLOOD. Hyperpyrexia. f th l i f t d By the action of micro-organisms upon the living fluids and tissues of the body, whereby various toxic substances are produced either locally or generally, and by their escape into the blood cause degenerative changes.i d d i h b d the degenerative changes. 3. The poisonous substance may be introduced into the body from without. i lk l id d i di i 2. That prolonged high temperature of 1070 to io8° F. will pi oduce this same coagulation process. dege e at e g 3. The poisonous substance may be introduced into the body from without. h i lk l id d i di i Just as we know that various alkaloids used in medicine have a specific affinity for particular portions of the nervous and muscular systems, eo we find certain p g p Halliburton has shown that there are three proteids in nervous matter-namely: (I) a neuroglobulin which is coagu- .ated at I6.60 F. (470 C.); (2) a nucleo-proteid which is coagu- Just as we know that various alkaloids used in medicine have a specific affinity for particular portions of the nervous and muscular systems, eo we find certain 6 6 CROONIAN LECTURES. [JUNE 30, 1900. 1583 TM.D BEAU I5o u IIOmmaJ nucleus of the fifth nerve. This, however, is not the case in animals when poisoned by the tetanus toxin. We do not know whether this selection is due to some anatomical condition which favours the absorption of the poison, or whether it is the result of a bio-chemical affinity of that, particular group of neurones for the poison. It is remark- able that experiment has shown that the tetanus toxin, if mixed with an emulsion of nervous matter before injection into an animal, loses its toxicity, showing thereby its affinity for nervous matter. Goldscheider and Flatau, who have studied the changes in the nerve cells of animals that have been injected with the tetanus toxin, describe conditions which they consider characteristic of the action of this par- ticular poison. Examination of the cells of the anterior horns of the spinal cord shows enlargement and pallor of the nucleus, swelling and crumbling of the Nissl bodies, and fine granular disintegration of the Nissl bodies, associated with swelling of the whole cell. i i i h i ll poisons which produce degeneration and disease a special selective influence. ALTERED STATES OF THE BLOOD. Hyperpyrexia. f th l i f t d Again, there are a number of poisons-diphtheria, botulismus, etc., in which the fatty degeration of the muscle in excess of what could be reasonably attributed to changes found in the nerves. Probably the vulnerable point of the neuro-muscular mechanism is at the junction of the nerve with the muscle. Comparative studies, however, would show degenerative changes in the motor end plates, are extremely difficult; and it is rather by inference than direct observation that we must believe the poison to act upon this structure. The effects of a poison in producing degenerative changes may be immediate or remote, even distantly remote. some poisons on entering the system may produce sudden, or almost sudden, effects-for example, strychnine and absinthe. Others produce more or less remote effects-for example, rabies, tetanus, diphtheria; others produce distantly remote effects-for example, syphilis. Again, many poisons, in order to be effective, must accumulate in the system- namely, the mineral poisons lead and arsenic. Dixon states that lead may exist in the brain in appreciable quanti- ties in encephalitis saturnina. SELECTIVE INFLUECNCE0 POISONS. Th t k bl SELECTIVE INFLUECNCE0 POISONS. Warden and Waddell, Non-bacillar Nature of Abrus Poisoning, Calcutta, 1884. chromatolysis generally occurred first, at the periphery of the cells; as the morbid process advanced there was destruction of the achromatic fibrils and formation of vacuoles. I have not observed vacuolisation, only extreme chromatolysis. Kempner and Pollak have found similar changes in the nerve cells, and have further observed the slow restoration of the cells to normal under the action of specific antitoxin. Somewhat similar changes have been described by several authorities in the anterior horn cells of the spinal cord and medulla of animals injected with diph- theria toxin. Sidney Martin separated the chemical toxin of diphtheria, and produced by its injection into animals marked fatty degeneration of the muscles, which he attributed to an associated change in the nerves. In 4 cases of diphtherial paralysis in man which I have examined I was able to find fatty degeneration of the heart muscle in all; in only i case could I find Wallerian degeneration of the peripheral nerves, and in only this one case could I find any chromolytic changes in the motor cells of the medulla. chromatolysis generally occurred first, at the periphery of the cells; as the morbid process advanced there was destruction of the achromatic fibrils and formation of vacuoles. I have not observed vacuolisation, only extreme chromatolysis. Kempner and Pollak have found similar changes in the nerve cells, and have further observed the slow restoration of the cells to normal under the action of specific antitoxin. Somewhat similar changes have been described by several authorities in the anterior horn cells of the spinal cord and medulla of animals injected with diph- theria toxin. Sidney Martin separated the chemical toxin of diphtheria, and produced by its injection into animals marked fatty degeneration of the muscles, which he attributed to an associated change in the nerves. In 4 cases of diphtherial paralysis in man which I have examined I was able to find fatty degeneration of the heart muscle in all; in only i case could I find Wallerian degeneration of the peripheral nerves, and in only this one case could I find any chromolytic changes in the motor cells of the medulla. Abrin and Rcin- Toxic Proteid Bodies.-Sidney Martin long ago showed that the seeds of jequirity contained a proteid body, which, if injected into an animal, profoundly modified the blood and caused its death. SELECTIVE INFLUECNCE0 POISONS. SELECTIVE INFLUECNCE The most remarkable example of selective influence I can cite is tetanus. The bacilli are found only in the wound; they must therefore be comparatively few in number, they elaborate a virulent poison which affects particular groups of neurones. The fact that lockjaw in man always occurs first shows that the poison selects the DEGENERATION OF THE NEURONE. JUNE 30, 1900.] LM bJo.&L 1589 p. 915. Bregmann, ref. Barker's Nervous System, p. 23I. Brodie and Ricnardson, Phil. Trans., vol. CXCiB. Sir Astley Cooper, ret. Hill, Thte Cere- bral Circulation. Darkschewitz, ref. Barker's Nervous System, p. 231. Ehrlich and Brieger, Ueber die Auschaltung des Lendenmarks, etc. ; Zeitschriftf. klin. Med., I884, Bd. viii, Suppl. H. Ehrlich, Deut. med. Woch., I89I, 32 and 34. Erlanger, ref. Barker'sNlVervous Spstem, p. 207 Ferrier and Turner, A Record of Experiments illustrative of the Symptomatology and Degenerations following Lesions of the Cerebellum aud its Peduncles and Related Structures in Monkeys, Phil. Trans., vol. clxxxv, x894. Flatau, Neue experimentelle Arbeiten uber die Pathologie der Nervenzelle, Fortschritte der Medecin, I897, No. 6. Flechsig, Gehirn und Seele. Van Gehuchten, Anat. due Syste ne Nerveux. Golclscheider und Flatau, Normnale und pathologische Anatomie der lVervenzellen, Berlin, I898. Leonard Hill, The Cerebral Circulation, Arris and Gale Lectures and Proc. Roy. Soc., June, igoo. W. D. Halliburton, Jouurn. of Physiol., vol. xii. I893, p. Ioo. Kempner and Pollak, Deut. med. lVToch, I897, No. 32. Lloyd, Journ. oJ Physiol, vol. 25, No. 3. Lugaro, ref. Ford Robertson, Brain, Sum. No., I899, p. 259. Marinesco, Verainderungen der Nervencentren nach ausreisaung der Nerven mit einigen Erwagungen betreffs ihrer Natur, NYeurol. Centtralbl., I898, No. I9. Dixon Mann, Forensic Medicine. Marinesco, Recherches sur les Lesions des Centres Nerveux conshcutives a l'Hyper- thermie Exp6rimentale et ai la Fi6vre, Revue Neurol., I899, S. Martin, BRIT. MED. JOUR-., Ii8g, vol. ii, p. I84. Martin, LocalGovernmentBoardReport, 1893. Von. Monakow, Archiv f. Psych., xiv., p. i, I883. F. Nissl, Allegem. Zeit- schrift f. Psych., td. xlviii, S. I97, I892. Risien Russell, Proc. Roy. Soc., lvi, p. 303. I894. Sarbo, Neurolog. Gentralbl., I895. Schifer. Proceedings Phlsiol. Soc., Journt. of Phys., I899. Munzer and Wiener, Archiv f. exper. Path. u. Pharmakol., Leipzig., Band. xxxv, I895. Tuckett, Journ. of Physiol., vol. xix, No. 4, 1896 Voss, Deut. Archiv .f khtn. Med, I897. Heft 6, p. 489. Waller, Effets de la Veratrine et de la Protov6ratrine sur les Nerfs de la Grenouille. SELECTIVE INFLUECNCE0 POISONS. 4 P id l ll f d ft li ti f t tid t b l py , y g , 4.-Pyramidal cell of a dog after ligation of two carotids, one vertebral and one subelavian. Great swelling of the nucleus, advanced chromota- lysis most muarked at the periphery of the cell. Magnification 700. i f d f li i f i h il py , y g , 4.-Pyramidal cell of a dog after ligation of two carotids, one vertebral and one subelavian. Great swelling of the nucleus, advanced chromota- lysis most muarked at the periphery of the cell. Magnification 700. i ll f d f li ti f t i h il t y p p y g 5.-Pyramidal cell of a dog after ligation of arteries showilng extreme chromatolysis with commencing extrusion of the nucleus. Magnification 700.6 i li d f l id l ll h l h y p p y g 5.-Pyramidal cell of a dog after ligation of arteries showilng extreme chromatolysis with commencing extrusion of the nucleus. Magnification 700. i ll 700.6.-Axis-cylinder process from a large pyramidal cell the lymph space around which is distended and showing the collateral side branches apparently forming a diffuse nerve network. Magnification Iooo. i i f f stud ed changes produced t e nerve cells by c . Absinthe.-Dr. Hill, while making his experiments upon the cerebral circulation, had occasion to inject animals with absinthe. He sent me the nervous tissues of these animals, and I was greatly surprised to find that changes were dis- coverable in the cortical pyramidal cells of animals that had died in fits within ten minutes of the injection of 5 minims of the poison. The cells were, however, not all equally affected; some groups of cells showed marked chromolytic change, while others showed no change. Animals that had lived two hours after the injection of a smaller dose (2 minims) like- wise showed unequal effects. The cortical pyramidal cells were, however, always more affected than the motor cells of the anterior horns of the medulla and spinal cord, and one therefore naturally asks, Can we account for this by supposing that the cells have each a specific vital resistance? BIBLIOGRAPHY. L. Barker, The Nervous System. Ballet et Dutil Sur Quelques LUsions Expdrimentales de la Cellule Nerveuse, Abstract Neurol Centralb., 1897, SELECTIVE INFLUECNCE0 POISONS. That this may be so is perfectly clear from the fact that such poiaons as abrin, botulin, ricin, etc., produce when injected much more marked effects on some cells of the same function than on others. In the case of absinthe, it must be remembered that we have to do with an oily fluid, and it is possible that really the fits may have been due to irritation caused by minute capillary emboli of an acrid and poisonous nature becoming lodged in the minute vessels of the cerebral cortex. The capillaries in the cortex are extremely small, and would favour such lodgment. Such an explanation seems possible, for if absinthe be injected subcutaneously, even in large quantities, fits are not pro- duced (Hill). I am also indebted to Professor Horsley for the nervous system of a cat, which had received 2 minims of absinthe intravenously injected. The animal had three fits and survived two hours. The changes in the motor cells of the brain and cord corresponded to those above described. apparently o g g 7.-Pyramidal cell with diffuse staining from a cat after ligation of four cerebral arteries. Magnification 5co. 8 id l ll f k fi d ft li t f t tid g 8.-Pyramidal cell from monkey five days after ligature of two carotids and one vertebral, referred to in lecture, showing swelling in pyramidal cell, diffuse homogeneous staining owing to stainable substance being scattered through the protoplasiii of the cell as a fine dust. l i h i scattered through t e protoplasiii 9. Section of the cortex of a case of general paralysis, showing acute cell changes. Magnification 250. C ll ki j b ll f d l d f li changes. Magnification 250. mo.-Cell of Purkinje from the cerebellum of a dog twelve days after liga- tion of four arteries slhowing absolutely normal appearances. id l ll f t f k f d t i l t t i d g y pp xI.-Pyramidal cell from cortex of monkey referred to in lecture stained by rapid Golgi method, showing gemmules on the dendrons, and all the external appearances of a normal cell. Magnification I50. f i d f k d ft pp g I2. Section of pyramidal tract of spinal cord of monkey ten days after ligation of two carotids and one vertebral. A few scattered degenerated fibres are revealed by the Marchi method. SELECTIVE INFLUECNCE0 POISONS. This body, abrin, together with a similar body, ricin, obtained from castor-oil seeds, were made the subject of a most valuable research by Ehrlich. Very small quantities of those poisons, injected into animals, cause profound modifications in the appearance of the nerve cells. The Nissl bodies disappear, the cells are stained uni- formly, and show a fine, coloured dust all through the proto- plasm of the cell and processes. All the nerve cells are affected, and the cause of these altered appearances would indicate a changed condition of the blood and lymph. By the kindness of Dr. Durham I have had the opportunity of examining the nervous system of two guinea-pigs, one which died forty-five hours after injection of 2 milligrams of abrus- globulin; the other after immunisation received in the course of a week an amount equal to 6o fatal doses. The former showed a marked change (shown in photomicrograph Fig. I6) in all the cells of the spinal cord and medulla; the latter showed marked chromolytic changes, but many of the cells showed Nissl granules and much less abnormal appearance. Although two observations are not of much valuie, it would point to the poison acting upon the nerve cells. Berkeley has studied the changes produced in the nerve cells by ricin. DESCRIPION OF PHOT0MICROGRAPHS. ll b i i d i.-Normal pyramidal Bett's cell cortex cerebri, stained by Nissl method, preparation made from the brain of a man who died from an obscure septic affection aft er eight weeks' illness. Towards the end of life he suffered with severe haemorrhages producing the most extreme an&emia, the blood containing a large number of nucleated corpuscles. There was a remittent pyrexia during the whole period; the fever, however, never surmounted 103.50 F. The body was extremely emaciated at death. He remainied conscious to the end, and it is interesting that this combination of factors was insufficient to produce bio-chemical changes in the nerve cells. Magnification, 7oo diameters. ce s. Magnification, oo 2.-Section of the motor cortex of a dog after ligation of both carotids and both vertebrals. Magnification, 250. f h di d d h i fl f g , 3.-Section of the motor cortex of a cat that died under the influence of an antesthetic prior to ligation of arteries. Normal appearance of the cortical pyramidal cells, stained by Nissl method. Magnification, 300. pe s ed, I3.-Swollen odematous cell from the top of the ascending frontal con- volution, with diffuse staining and absence of Nissl granules froin a case of status epilepticus. Magnification 700. 4 i f h i l d f f C i k ith h 700.6.-Axis-cylinder process from a large pyramidal cell the lymph space around which is distended and showing the collateral side branches apparently forming a diffuse nerve network. Magnification Iooo. i i f , p y g I5.-Anterior horn cell of spinal cord from a guinea-pig which died 45 hours after injection of 0.2 mg. of abrus-globulin. All the cells showet this diffuse staining and absence of Nissl granules. Magnification 4cO. i 0 h i l d f bbit py , y g , 4.-Pyramidal cell of a dog after ligation of two carotids, one vertebral and one subelavian. Great swelling of the nucleus, advanced chromota- lysis most muarked at the periphery of the cell. Magnification 700. i f d f li i f i h il SELECTIVE INFLUECNCE0 POISONS. A trace of albumen was found four times in the urine, the daily output of which in- creased under the use of digitalis from 30 to 50 ounces, while its specific gravity decreased to Ioo6 to 1o0o, with urea only I per cent. The blue line was not quite gone when he left the hospital. , g yp After-History.-During the first fortnight after admission she was fre- quently sick independently of food, and paresis of the left external rectus muscle was observed. For a month she suffered from intense headacbe, and remained drowsy and apathetic. On August 28th the optic neuritis had subsided, and the discs were atrophic. She was blind, and hardly able to point to the direction of the window. The paresis of the left externa) rectus had gone, and the general muscular tone and balancing power were much better. On September ioth she could walk well, and had the complete use of all her limbs, was free from headache, sickness, or abdominal pain, andwasplump and rosy. The eyesight was, however, un- altered. I visited this patient at her home in March, I90. and found her in excellent general health, but still blind. The medicinal and genera} treatment was the same as in the last case. i h l f l d q g p In this case the brunt of the disease fell on the brain, but clearly the kidneys, alimentary tract, and heart also suffered. I am sorry not to have been able to trace the further history of this patient, who was under my care eight years ago. d p , y g y g CASE II.-In April, I899, I saw in consultation a married woman, aged 39, for brain disease. Hi t Sh id t h h d t d These threecases illustrate most of the usual symptoms of lead encephalopathy, and show how closely this affection resembles cerebral tumour. One symptom only of this disease was not prominent in these cases, namely, epileptiform attacks, which Dr. Oliver states are a frequent and early symptom. The woman had had three fits, but the other cases had none. One satisfactory feature of the cases was that all recovered, only the poor girl being left seriously injured. SELECTIVE INFLUECNCE0 POISONS. The period soon came on, and her menses had been regular since, but she began to feel weak and dizzy within a week of taking the pills. i f d i i h f d b l d th thi t set of symptoms: -A youth of 17 was sent into hospital for cerebral tumour. different set of symptoms: CASE I.-A youth of 17 was sent into hospital for cerebral tumour. y p History.-For three weeks he had suffered from headache, vomiting- chiefly afterfood-colicky pains in theabdomen and dimness of vision. The patient was a carriage painter who for 2 years had been engaged on fine work, but for the last three months had taken to coarse work in which much red lead was used. uc Condition on Acdmission.-He was found to be well nourished butanaemic, lethargic, but complained of intense frontal headache. No paralysis, spasm, or anaesthesLa, pupils and reflexes normal. Well marked double optic neuritis and a large retinal haimorrhage in the left eye. Heart somewhat dilated, blowing murmur at the apex; soft, regular pulse. Urine: speciflc gravity IOI to IOI6, free from albumen and sugar. Bowels open, tongue moist, furred and tremulous. Slight blue line on the gums. h b l k d i i did d b g y g p Anamnesis.-After admission she was found to be a pale and rather thin girl with features drawn with constant pain in the head. A distinct blue line on the gums. She is drowsy, and occasionally a little delirious. She can sit up, but not stand alone, the knees doubling under her, and the feet being dragged. Knee-jerks absent; plantar reflexes well marked. Arms rather weak. but not paralysed. Pupils, eye and face movements normal; no antesthesia; well-marked double optic neuritis. Her vision was unfor- tunately not properly tested at this time, but she could recognise the doctors and nurses, and I think read large type. i d i i f g Treatment-The bowels were kept open and potassium iodide and bro- mide, with an occasional dose of antipyrin, ordered. A fortnight later he was no better, and then digitalis was added to his medicine. Within a week he was better, and all his symptoms left him five weeks after admis- sion. The vision improved and the optic neuritis became perceptiblyless. but some yellow striue developed round each macula lutea. SELECTIVE INFLUECNCE0 POISONS. The fact that the married woman was said to have shown severe hysterical sym- ptoms a year before is of interest, for a "toxic hysteria" has been found in other cases to sometimes precede the severe and even fatal form of the disease, and to have obscured the diagnosis. Possibly some cases diagnosed as hysteria which have ended fatally may have been due to undiscovered lead or other poisoning. d l d History.-She was said to have had severe nervous symptoms and a mysterious abdominal pain a year ago, but these abated when it was pro- posed to send her into hospital. She had had two live children, the youngest IS months old, and six miscarriages, but there was no history of a syphilitic rash. Apparently she had not been really well for a year, but had for four or five months suffered severely from headache and vomiting without relation to food. She had had three fits within the previous three months. A month ago she began to be unsteady on her feet. and had been confined to bed for three weeks. For a week she had been semi-delirious, having visual hallucinations, and for two days had had diplopia. No in- formation as to her menstrual history could be obtained, either then or during her stay in hospital. i l i i i g y p Anamnesis.-Only an imperfect examination was possible in the small dark room of her home, but we noted no paralysis, ataxv resembliog that of cerebellar disease, and double optic neuritis. A week later she came into hospital, where in the well-lit ward I at once observed on her gums the most intense blue line I lhave ever seen. The notes say : " Patient is a stout, florid woman; intellect and memory poor, but she can answer some questions fairly; has some visual hallucinations at night; complailns of severe headache on the vertex. She can sit up in bed, but can hardly stand or walk alone. tottering with the head craned forward and swaying from side to side. The ataxy is not worse when the eyes are shut. No muscular wasting or tremors, and no local paralysis of the limbs, though all are weak. No anaesthesia. Knee-jerks well marked. The pupils are of medium size, the left a shade the larger; they react nor- mally. SELECTIVE INFLUECNCE0 POISONS. The dense black masses are fat cells; but the degenerated muscle fibres are indicated by the diffuse black staining. The musculi papillares of the heart showed even more marked fatty degeneration. 2 8 i solids. The muscle was stained by Marchi method, and shows that quite one half of the fibres have undergone acutefatty degeneration. The dense black masses are fat cells; but the degenerated muscle fibres are indicated by the diffuse black staining. The musculi papillares of the heart showed even more marked fatty degeneration. i 2 3 8 13 14 5 6 h i h f y g Figures I, 2. 3, 4, 5, 7, 8, q. 13, 14, I5, ,6 are photomicrographs of prepara- tions stained by the Nissi method. Figure 6 was stained by the Weigert iron method. Figures Io and xI were stained by the rapid Golgi method. Figures 12 and 17 were stained by the Marchi method. ON LEAD ENCEPHALOPATHY AND THE USE OF DIACHYLON AS AN ABORTIFACIENT. By W. B. R&NSOYI, M.A., M.D., F.R.C.P., Physician to the General Hospital, Nottingham. IN the following paper I propose to record three cases of cere- bral disease due to lead poisoning, two of which were caused by diachylon taken to procure abortion, and some other cases In which diachylon taken with the same object produced a different set of symptoms: g CASE III.-H. W., an unmarried girl, aged 22, was admitted in June, 2899, for pains in the stomach and head, of tlhree or four months' duratlon. Ri t Sh h d b t t d t id f t i l and kept on CASE III. H. W., an unmarried girl, aged 22, was admitted in June, 2899, for pains in the stomach and head, of tlhree or four months' duratlon. Ristory.-She had been treated outside for gastric ulcer and kept on milk diet, as solid food caused acute pain. There had also been some vomiting and some severe colicky pains. For the last month she had suffered from intense headache chiefly in tlle frontal and right occipital regions On interrogation, the patient admitted that in the March pre- vious, having gone a week beyond her proper menstrual period, she nad taken one day a half-pennyworth of diachylon, which she had made intoc two pills. She was positive that she had never taken any more. SELECTIVE INFLUECNCE0 POISONS. of 1590 1590 LLEAD ENCEPHALOPATHY. [JUNE 30, 1900. however, to see Mrs. C. on March 20th, igoo, and found her in perfect health, and the proud mother of a three weeks old baby. There was neither blue line nor squint, she could read the local newspaper with ease, and the ophthalmoscope showed nothing abnormal but the faintest haze over the origins of the vessels in the optic discs. She must have con- ceived immediately on leaving the hospital, and did not repeat her experiments on eebolics. The baby is, however, a puny creature, said to have " consumption of the bowels," but is not hydrocephalic, as the children of lead workers are at times. On interrogation, she now asserted that she had taken diachy- lon for a fortnight a year before I saw her, being then six weeks pregnant. Abortion took place at the end of the third month. She adhered to the statement that she had taken none since, although the symptoms of poisoning steadily in- creased up to her admission into hospital. I have no means of testing the truth of this statement. however, to see Mrs. C. on March 20th, igoo, and found her in perfect health, and the proud mother of a three weeks old baby. There was neither blue line nor squint, she could read the local newspaper with ease, and the ophthalmoscope showed nothing abnormal but the faintest haze over the origins of the vessels in the optic discs. She must have con- ceived immediately on leaving the hospital, and did not repeat her experiments on eebolics. The baby is, however, a puny creature, said to have " consumption of the bowels," but is not hydrocephalic, as the children of lead workers are at times. On interrogation, she now asserted that she had taken diachy- lon for a fortnight a year before I saw her, being then six weeks pregnant. Abortion took place at the end of the third month. She adhered to the statement that she had taken none since, although the symptoms of poisoning steadily in- creased up to her admission into hospital. I have no means of testing the truth of this statement. solids. The muscle was stained by Marchi method, and shows that quite one half of the fibres have undergone acutefatty degeneration. SELECTIVE INFLUECNCE0 POISONS. There is some convergent strabismus, as the left eye cannot be moved completely outward. There is occasional diplopia ; no facial or lingual paralysis; speech normal. She can read ordinary type with ease; hears a watch ats inches from rom right ear. The oph- thalmoscope shows double optic neuritis, swelling of disc = I.5 D. in each eye. There are hoemorrliages on and around each disc; no choroiditis or sign of syphilis. Theurine contained a trace Mf albumen for the first two days only. The uterus showed no distinct sign of a recent miscarriage and was not enlarged. h i i i i ( d il ) d p g It is very remarkable that such acute, severe, and prolonged disease should have been produced in the girl by a single dose of lead-namely, one halfpennyworth of diachylon taken at once. To get some idea of the amount of lead taken by this girl I bought a pennyworth of diachylon at a chemist's, and had it analysed by Mr. Trotman, the city analyst. The diachylon, which was in the form of a ball about the size of a walnut, contained 86 per cent. of oleate of lead and 46.7 grs. of lead, so that we may suppose the.girl to have taken about 23 grs. of lead. b g This case proves that the excretion of the metal must be very slow, as does also that of Mrs. C., if her story be true. Th t t t f th i dditi t t ilk di t p very slow, as does also that of Mrs. C., if her story be true. The treatment of these cases in addition to rest, milk diet, and good hygienic conditions, consisted mainly of moderate doses of potassium iodide with Epsom salts. It is curious that the iodide should have obtained such a reputation in the treatment of lead poisoning, for Dr. Dixon Mann's researches seem to prove that it does not increase the elimination of the metal, the greater part of which escapes by the freces and not by the urine. It should be noted, however, that in the first ca2e improvement did not occur until the heart had been g Treatment.-The patient was put upon potassium iodide (gr. xv daily) and magnesium su}phate, and improved with wonderful rapidity. SELECTIVE INFLUECNCE0 POISONS. These were more numerous on the side opposite to the hemisphere on which the vertebral was liga- tured, but altogether not more than 6o in number, so that only an incon- siderable number of the psycho-motor cells had perished as a result of the anaemia. This quite conforms with the fact that the animal had returned to the normal physiological condition when it was killed with chloroform. Dogs with four arteries ligatured and kept alive over ten days frequently showed that a certain number of the psychomotor neurones had perished, but the numbers were inconsiderabfe. 3 ll d t ll f h f h di f t l perished, but t e I3.-Swollen odematous cell from the top of the ascending frontal con- volution, with diffuse staining and absence of Nissl granules froin a case of status epilepticus. Magnification 700. 4 i f h i l d f f C i k ith h p p g I4.-Section of the spinal cord of a case of Congo sickness with hyper pyrexia in which the temperature reached I09° F. prior to deatlh. The whole of the cells throughout the central nervous system showed a diffuse homogeneous dull staining. The Nissl granules had entirely disappeared from the processes and the body of the cell, and the stainable substance had a fine dust-like appearance. The processes are unusually distinct on account of the diffuse stofliing, the nucleus is clear and swollen, faintly stained, and the nucleolus deeply stained. Magnification 400. stained, and the nucleolus deeply stained. Magnification 400. I5.-Anterior horn cell of spinal cord from a guinea-pig which died 45 hours after injection of 0.2 mg. of abrus-globulin. All the cells showet this diffuse staining and absence of Nissl granules. Magnification 4cO. 0 i l d f bbi this diffuse staining and absence of ss granules. Magnification 4cO. I6.-A section of the anterior horn o0 the spinal cord of a rabbit, 21 hours after injection of o.oI c.cm. of botulismus toxin, which killed the animal, the whole of the cells throughout the central nervous system showed more or less a marked chromolytic change as indicated in the photomicrograph. Magnification 250. i di h f i i hi h di d th i hth d photomicrograph. Magnification 250. I5.-Section of the diaphram of guinea-pig, which died on the eighth day after injection of o.ooo5 c.cm. of botulismus toxin, containing 2 per cent. SELECTIVE INFLUECNCE0 POISONS. Within three weeks the headache had gone, she was quite rational, and could walk well and steadily. She left the hospital a month after admission with the weakness of the left rectus externus muscle nearly gone and the optic neuritis diminishing. The blue line had also nearly gone. Thi i d i d h i k di h l i ill f p g y g This patient admitted having taken diachylon in pills for the purpose of checking the course of pregnancy, but we could not ascertain from her exactly when or what quantity she had taken, or what the effect had been on the uterus. I called,
https://openalex.org/W3133841844	https://www.researchsquare.com/article/rs-10205/v1.pdf	English	null	Effect of different restorative crown design and materials on stress distribution in endodontically treated molars and peripheral bone: A finite element analysis study	Research Square (Research Square)	2,019	cc-by	5,408	Effect of different restorative crown design and materials on stress distribution in endodontically treated molars and peripheral bone: A finite element analysis study Jie Lin (  linjie.dds@gmail.com ) Fujian Medical University Zhenxiang Lin Hospital of Fujian Provincial Authorities Zhiqiang Zheng Fujian Medical University Research article Research article Keywords: finite element analysis, endocrown, post-core crown, fiber post, zirconia Posted Date: December 30th, 2019 DOI: https://doi.org/10.21203/rs.2.19592/v1 License:   This work is licensed under a Creative Commons Attribution 4.0 International License. Read Full License Page 1/13 Abstract Background: The purposes of this simulation study were to evaluate the stresses in the roots of endodontically treated molars with extensive coronal tissue loss which were restored by endocrowns (all- in-one core and crown) and traditional crowns with post-cores, during masticatory simulation using finite element analysis. Background: The purposes of this simulation study were to evaluate the stresses in the roots of Methods: A mesio-distal cross-section of a lower right first molar was digitized and used to create 2- dimensional models of the teeth and supporting tissue; different crown designs, viz ., endocrown with 2 mm occlusal clearance, endocrown with 4 mm occlusal clearance and post-core crown; different crown materials, viz ., zirconia (Zr) and lithia-disilicate reinforced glass ceramic (LDRGC), and different post and core materials, viz. , glass fiber (GF), stainless steel (SS) and metal cast (MC). A simulated 100 N vertical occlusal load was applied to the distal marginal ridge of the crown. Results: The root canal inner wall stresses of SS post (maximum 33.7 MPa) and MC post (maximum 36.3 MPa) were higher than that of GF post (maximum 19.1 MPa) and endocrown (maximum 8.9 MPa). Conclusion: Endocrown showed reduced stresses at its root canal inner wall but increased stresses at the coronal cavity inner wall when compared to post-core crowns. Background The decision of how to rehabilitate endodontically treated molars (ETM) with extensive loss of coronal structure is a challenge in restorative dentistry. Coronal tooth tissue is often significantly damaged after endodontic treatment and are traditionally restored with metal posts and cores and prosthetic crowns (1−3). Initially, it was believed that this procedure would provide better reinforcement of the remaining dental structure (4, 5). However, it has been observed that the use of intracanal retainers only help in the retention of the prosthetic crown. As a result of removing a dental structure to enable the placement of rigid dental materials devoid of mechanical behaviors similar to those of the tooth, the remaining tooth is weakened (6). The preparation of a molar for a post in relatively narrow root canals also involves a risk of accidental root perforation and fracture (7). In fact, minimally invasive preparations, with maximal tissue conservation, are now considered ‘the gold standard’ for restoring ETM (8). In 1995, the endocrown was described by Pissis who is the forerunner of the endocrown technique, as the ‘mono-block porcelain technique’ (9). Currently, due to the advances in adhesive methods and materials, endocrown type of intracoronal restorations were suggested for damaged posterior teeth as an alternative to post and core retained ones (10). It is a method particularly indicated in cases in which there is excessive loss of hard tissues of the crown, interproximal space is limited, and traditional post-core crown is not possible because of inadequate ceramic thickness (11). Their advantages include the fact that tooth structures require little preparation, ease of preparation, demand less clinical time when compared with conventional crowns(12,13). Knowledge of the stress distribution within and around the roots is a key factor for understanding root fracture, which are well-known problems with ETMs. It has been proposed that molars restored with endocrowns are less prone to fracture than those with posts (14,15). Nevertheless, so far there was no clear evidence to prove it. Dejak et al. (16) compared equivalent stresses in molars restored with endocrowns as well as posts and cores during masticatory simulation using finite element analysis (FEA) and found the tensile stresses achieved were 4 times higher values than under endocrowns. These tensile stresses occurred in the dentin under the crown shoulder, rather than in the root. A similar study by Lin et al. Background (15) showed that the stress values on the dentin and luting cement for the endocrown restoration were lower than those for the crown. However, these studies made no attempt to compare endocrowns and post-core crowns. Moreover, relatively little is known about the differences of stress distribution in the roots. The type of restoration (endocrown or post-core crown, different crown and post materials) will provide rational stress distribution and reduce a risk of fracture in molars? Because of the absence of information about the biomechanical behavior of endocrowns and the expectation that this type of restoration would behave similarly or superiorly to post-core crowns, the present study has aims to evaluate the von Mises stresses in the roots of ETMs with extensive coronal loss, restored by endocrowns and post-core crowns, during masticatory simulation using FEA, and simulate stresses at the first molar made with different crown and post materials. Endocrown and post-core crown designs The Endocrown-2 mm designs were created with 2.0 occlusal clearance, 7.0 mm cavity depth, and 5.3 mm base width. The prepared cavity walls tapered with 2-5 degrees from the cavity base to the cavosurface (Fig. 1b). The Endocrown-4 mm designs were created with 4.0 occlusal clearance, 5.0 mm cavity depth, 5.3 mm base width and 2-5 degrees cavity walls taper (Fig. 1c). Jacket crown preparations were created with 2.0 mm occlusal clearance, 0.5-1.5 mm cervical clearance and shoulder margin, 2-5 degrees tapering angle for first molars, 14.0 mm post lengths. Rounded shoulder margins and anatomic occlusal reduction were incorporated in model (Fig. 1d). The surrounding bone was modeled as cortical bone (1.5 mm thickness) and cancellous bone, which were assumed to be isotropic, homogeneous, and linearly elastic. A 0.2 mm periodontal ligamentwas modeled around the roots. A 0.1 mm thick cement-imitating layer was formed around the root part of the created post and under the crown. Perfect bonding was assumed at all the interfaces, including those between the teeth, the cores, the crowns, the posts and bones. Structures and geometric conditions of the computer aided design (CAD) model A mesio-distal cross-section of a lower right first molar was digitized and used to create 2-dimensional (2- D) models. There were three different model designs (Fig 1), viz., endocrown with 2 mm occlusal clearance, endocrown with 4 mm occlusal clearance and post-core crown. The restorations used two different crown materials, viz., zirconia (Zr) and lithia-disilicate reinforced glass ceramic (LDRGC), and three different post and core materials, viz., glass fiber (GF), stainless steel (SS) and metal cast (MC). There were ten kinds of combination in this study. Endocrown-2 mm-Zr: full zirconia endocrown with 2 mm occlusal clearance; Endocrown-2 mm-Zr: full zirconia endocrown with 2 mm occlusal clearance; Endocrown-2 mm- LDRGC: full lithia-disilicate reinforced glass ceramic endocrown with 2 mm occlusal clearance; Endocrown-4 mm-Zr: full zirconia endocrown with 4 mm occlusal clearance; ndocrown-4 mm-Zr: full zirconia endocrown with 4 mm occlusal clearance; ocrown-4 mm-Zr: full zirconia endocrown with 4 mm occlusal clearance; Page 3/13 Endocrown-4 mm- LDRGC: full lithia-disilicate reinforced glass ceramic endocrown with 4 mm occlusal clearance; GF post-Zr: glass fiber post + full zirconia crown; GF post- LDRGC: glass fiber post + full lithia-disilicate reinforced glass ceramic crown; Endocrown-4 mm- LDRGC: full lithia-disilicate reinforced glass ceramic endocrown with 4 mm occlusal clearance; GF post-Zr: glass fiber post + full zirconia crown; Endocrown-4 mm- LDRGC: full lithia-disilicate reinforced glass ceramic endocrown with 4 mm occlusal clearance; SS post-Zr: stainless steel post + full zirconia crown; SS post- LDRGC: stainless steel post + full lithia-disilicate reinforced glass ceramic crown; MC post-Zr: metal cast post + full zirconia crown; MC post- LDRGC: metal cast post + full lithia-disilicate reinforced glass ceramic crown. In the GF posts and SS posts, the cores were made of composite, while in the MC posts they were made of metal. The model was created from radiographic image of a real tooth (Fig. 1a) using a FEA software (ANSYS v. 10; ANSYS Inc., Canonsburg, PA, USA) (17). Material properties, mesh generation and boundary conditions The elastic moduli and Poisson’s ratios of the materials used are shown in Table 1. Material properties were assumed to be isotropic, homogenous, and linear-elastic, except the GF post. The material of GF post was anisotropic (Young’s modulus along its long axis was 38.5 GPa, and 12.0 GPa perpendicular to that axis). For calculation purposes, each tooth model was divided into 2-D 4-node structural solid elements (PLANE42). This element is defined by four nodes having two degrees of freedom at each node: translations in the nodal x and y directions. In model with endocrown-2 mm, 4,596 elements joined at 4,701 nodes were used. In model with endocrown-4 mm, 4,582 elements joined at 4,693 nodes were used. In model with post-core crown, 4,657 elements joined at 4,759 nodes were used. The aim of this preliminary FEA was to identify the most highly stressed regions within the restoration, especially those along the distal root inner and outer surface. These would be the regions to which shape optimization Page 4/13 Page 4/13 would be applied. Thus, the mesh around the distal root inner and outer surface was made much finer than those in the other areas, with an average element edge length of 0.2 mm. Fixed zero-displacement in both the horizontal and vertical directions was defined at the horizontal and vertical cut-planes of the supporting bone. A load was applied that corresponded to static loading, assuming no vibrational or dynamic effects in the structure. To reflect the stress distribution at the moment of equilibrium, a simulated 100 N vertical occlusal load was applied to the distal marginal ridge. The von Mises stress values were calculated by FEA along the distal root canal inner wall and the root outer surface (Fig 1: A→B→C). We focused on the distal root canal inner wall because the post was set in the distal root canal, from preliminary analysis the distal root canal inner wall was analyzed in greater detail. The stress distribution within the tooth/restoration cross-section was solved with the FEA software (ANSYS). Discussion The numeric FEA modeling is able to reveal the otherwise inaccessible stress distribution within the tooth- restoration complex. It has proven to be an important tool in the design process for the understanding of tooth biomechanics and the biomimetic approach (27, 28). Although teeth are 3-D structures, important mechanical events in endocrown and post-core crown appear within the mesio-distal plane (27, 29). These events support the use of the 2-D plane-strain model for numerical analyses. Volume meshing of 3-D teeth structures is usually restricted to tetrahedral elements. The tetrahedral element has a good ability to model irregular shapes; however, its accuracy is poor for bending and shearing dominated problems. The use of a 2-D model is also valuable because of its improved performance in terms of element number and simulation quality. Consequently, 2-D analysis was chosen in this study. Inversely to anterior teeth, posterior cusps do not deform under load as simple cantilever beams (30, 31). The deformation mode is complicated by the numerous possibilities in the application of loads. It is inevitable for the molar to receive non-axial force in the process of occlusion. The load configuration applied in this study was selected because it creates a maximum challenge for distal root flexure, which seems to represent an important biomechanical feature of post-core crown. Mean masticatory forces have been reported by Anderson (32) to be in the range of 70.6 to 146.1 N. Thus, the applied 100 N load lies within the range of these values in this study, and it has been used in a number of FE model validation studies(33). Despite great variations in crown material properties, there were only minor differences in the alveolar bone. For a given load configuration, it appears that overall stress distribution within the tooth-bone complex was more influenced by geometry design of restoration (endocrown vs. post-core crown) than by composition (e.g. crown, post and core restorative material type). The endocrown showed a relatively smooth stress distribution in the root and the periodontal support tissue. This is largely due to two reasons: firstly, post-core crown applied extra-coronal retainer, while endocrown used intra-coronal retainer. Endocrown with intra-coronal retainer was more conducive to transfer the force to the wall of the pulp chamber and the periodontal tissue, rather than to the root canal wall. Secondly, endocrown geometrically reduced the rotation center of the crown restoration in comparison with the full crown (Fig. Results The von Mises stress analysis for the 2 crown designs, 2 crown materials, 3 post materials tested is presented in Fig. 2. The stress distributions were similar for 2 different crown materials (Zr and LDRGC) and 2 different occlusal clearance (2 mm and 4 mm). The crown stresses of Zr were more concentrated. For the endocrown-2 mm-Zr and endocrown-4 mm-Zr (Fig. 2), a stress peak was found at the distal cavity inner wall (maximum 27.7 MPa and 17.8 MPa). Compared with the endocrown-2 mm, the endocrown- 4 mm reduced the stress concentration in crown. For the post-core crown, the main difference was found at the distal root. The low elastic modulus of the GF post showed reduced the von Mises stresses (maximum 23.6 MPa) (Fig. 2 GF post); both the SS post (maximum 62.9 MPa) and MC post (maximum 67.5 MPa) ensure a stress concentration at the post and distal root. Differences also were found in the stresses at the alveolar bones. The endocrown showed increased and smooth stresses in the periodontal support tissue (maximum 24.2 MPa). (Fig. 2). The plots of the post- core crowns displayed only minor differences between post materials, 20.1 MPa for GF post, 17.4 MPa for SS post and16.6 MPa for MC post. Figure 3 shows the path plots of the interfacial von Mises stresses along the distal root canal inner wall and the root outer surface (A→B→C) in endocrown and post-core crown. No difference was found in the Zr crowns and the LDRGC crowns in the path plots. In the endocrown covered tooth, maximum von Mises stress values of 9 MPa were recorded in the distal root canal inner wall. The root canal inner wall stresses of SS post (maximum 33.7 MPa) and MC post (maximum 36.3 MPa) were higher than that of GF post (maximum 19.1 MPa) and endocrown (maximum 8.9 MPa). The behavior of the endocrown clearly differed from that of the post-core crowns. The endocrown only had one peak, nevertheless the post-core crowns had two peaks. The peak 1 was caused by the apical foramen, and the peak 2 was caused by the tip of the post. von Mises stress values around the SS and MC post tip (Peak 2) reached 12.7 MPa and 14.1 MPa. They were higher than that of GF post (7.5 MPa). Results In the same region, von Mises stress of Page 5/13 Page 5/13 endocrown achieved a value of 5 MPa. The endocrown showed higher stresses occurred in the root outer surface than that of post-core crowns. Discussion 1, Ring arrow). This also contributed to transfer the occlusal force to periodontal tissue. The root canal after instrumentation (root canal or post preparation) is thinner and weaker than the rest of the tooth. Stress concentration at the tip of the post must therefore be regarded as most harmful. It is precisely in the area of concentrated stresses where differences were found (Fig. 3, Peak 2 of SS post and MC post). e.g. stresses in the tip of the post around MC posts achieved 3 times higher values than under endocrowns and twice higher values than under GF posts. In this specific zone, low–elastic modulus restorative materials showed reduced stresses, which can be explained by the stress redistribution into Page 6/13 Page 6/13 the more flexible GF post and composite core. Stress concentration of endocrown in the root canal was relatively small, it is good to avoid the weak tip of the post. The molar post-core crown increased the risk of accidental root fracture. The highest von Mises stresses in the distal root canal occurred in molar restored with SS and MC posts. These types of restorations seem to be the least beneficial in molar teeth. Biac-chi and Basting (34) found that molars with endocrowns are more fracture resistant than teeth restored with GF posts and cores and ceramic crowns. Taking into consideration the suitable stress distribution of endocrowns, minimal invasive preparation of tooth structures and no roots damage, these restorations can be recommended to use in molars. Although von Mises stress levels in the root and the periodontal tissue of molars restored with post-core crowns were higher than stress levels in the tooth with the endocrown. Zr endocrown represented the one condition with a slightly greater amount of stresses concentration in the distal cavity inner wall when compared to the post-core crowns. Thus, this is regarded as a potential threat, knowing that remaining coronal tooth structure fractured. On the other hand, experimental strength study by Forberger and Although von Mises stress levels in the root and the periodontal tissue of molars restored with post-core crowns were higher than stress levels in the tooth with the endocrown. Zr endocrown represented the one condition with a slightly greater amount of stresses concentration in the distal cavity inner wall when compared to the post-core crowns. Thus, this is regarded as a potential threat, knowing that remaining coronal tooth structure fractured. Discussion On the other hand, experimental strength study by Forberger and Göhring (35), have shown no significant differences between teeth restored with posts and endocrowns in terms of fracture resistance. Under the analytical conditions of this study, the results were largely dependent on the Young's modulus and Poisson ratio of the materials. However in reality there will be other dominating factors such as the bond strength, potential for micro-crack at the interface, fatigue damage potentials both for the hard tissues and the restorative materials. Further experimental studies and clinical trials are needed to validate the results of this FEA study. Abbreviations Zr: zirconia, LDRGC: lithia-disilicate reinforced glass ceramic, GF: glass fiber, SS: stainless steel, MC: metal cast, ETM: endodontically treated molars, FEA: finite element analysis, CAD: computer aided design Conclusions In conclusion, posterior ceramic endocrowns bonded to the tooth substrate is ajudicious way to reduce excessive tooth tissue removal and surgical crown lengthening. Within the limitations of this FEA experiment, it can be concluded that: Endocrown showed reduced stresses at its root canal inner wall but increased stresses at the coronal cavity inner wall when compared to post-core crowns. Endocrown exhibited higher stresses occurred at the root outer surface, while post-core crowns showed increased stresses at the root canal inner wall, especially SS post and MC post. The stress distributions in root were similar for 2 different crown materials (zirconia and lithia-disilicate reinforced glass ceramic) and 2 different occlusal clearance (2 mm and 4 mm). Acknowledgements The study design and the operation of the finite element analysis were supported by Prof. Shinya Akikazu. The study design and the operation of the finite element analysis were supported by Prof. Shinya Akikazu. Funding This work was supported by the Fujian province science and technology innovation joint fund project (Grant No. 2017Y9095) and the Health and family planning commission of the Fujian province, youth research project (Grant No. 2016-1-25). Availability of data and material The complete data and materials described in the research article are freely available from the corresponding author on reasonable request. Authors' contributions JL and ZZ conceived and designed the study. JL performed the finite element analysis. JL and ZL wrote the paper. ZZ reviewed and edited the manuscript. All authors read and approved the final manuscript. JL and ZZ conceived and designed the study. JL performed the finite element analysis. JL and ZL wrote the paper. ZZ reviewed and edited the manuscript. All authors read and approved the final manuscript. Consent for publication Not applicable Declarations Ethics approval and consent to participate Page 7/13 Not applicable Competing interests The authors have no conflicts of interest relevant to this article. References 1. Dartora NR, de Conto Ferreira MB, Moris ICM, et al. Effect of Intracoronal Depth of Teeth Restored with Endocrowns on Fracture Resistance: In Vitro and 3-dimensional Finite Element Analysis. J Endod. 2018;44;1179-85. 1. Dartora NR, de Conto Ferreira MB, Moris ICM, et al. Effect of Intracoronal Depth of Teeth Restored with Endocrowns on Fracture Resistance: In Vitro and 3-dimensional Finite Element Analysis. J Endod. 2018;44;1179-85. 2. Schlichting LH, Resende TH, Reis KR, Magne P. Simplified treatment of severe dental erosion with ultrathin CAD/CAM composite occlusal veneers and anterior bilaminar veneers. J Prosthet Dent 2016;116;474-82. 2. Schlichting LH, Resende TH, Reis KR, Magne P. Simplified treatment of severe dental erosion with ultrathin CAD/CAM composite occlusal veneers and anterior bilaminar veneers. J Prosthet Dent 2016;116;474-82. 3. Rocca GT, Rizcalla N, Krejci I. Fiber-reinforced resin coating for endocrown preparations: a technical report. Oper Dent 2013;38;242-8. 3. Rocca GT, Rizcalla N, Krejci I. Fiber-reinforced resin coating for endocrown preparations: a technical report. Oper Dent 2013;38;242-8. Page 8/13 4. Valdivia ADCM, Rodrigues MP, Bicalho AA, et al. Biomechanical Effect of Ferrule on Incisors Restored with a Fiberglass Post and Lithium-Disilicate Ceramic Crown after Thermal Cycling and Fatigue Loading. J Adhes Dent. 2018;20;133-42. Page 8/13 4. Valdivia ADCM, Rodrigues MP, Bicalho AA, et al. Biomechanical Effect of Ferrule on Incisors Restored with a Fiberglass Post and Lithium-Disilicate Ceramic Crown after Thermal Cycling and Fatigue Loading. J Adhes Dent. 2018;20;133-42. Page 8/13 5. Ferrari M, Sorrentino R, Juloski J, et al. Post-Retained Single Crowns versus Fixed Dental Prostheses: A 7-Year Prospective Clinical Study. J Dent Res. 2017;96(13):1490-7. 6. Soares CJ, Santana FR, Silva NR, Preira JC, Pereira CA. Influence of the endodontic treatment on mechanical properties of root dentin. J Endod 2007;33;603-6. 6. Soares CJ, Santana FR, Silva NR, Preira JC, Pereira CA. Influence of the endodontic treatment on mechanical properties of root dentin. J Endod 2007;33;603-6. 7. Ferrari M, Vichi A, Grandini S. Efficacy of different adhesive techniques on bonding to root canal walls: an SEM investigation. Dent Mater 2001;17;422-9. 7. Ferrari M, Vichi A, Grandini S. Efficacy of different adhesive techniques on bonding to root canal walls: an SEM investigation. Dent Mater 2001;17;422-9. 8. Dietschi D, Duc O, Krejci I, Sadan A. Biomechanical considerations for the restoration of endodontically treated teeth: A systematic review of the literature, part ii (evaluation of fatigue behavior, interfaces, and in vivo studies). Quintessence Int 2008;39;117-29. 8. References Dietschi D, Duc O, Krejci I, Sadan A. Biomechanical considerations for the restoration of endodontically treated teeth: A systematic review of the literature, part ii (evaluation of fatigue behavior, interfaces, and in vivo studies). Quintessence Int 2008;39;117-29. 9. Pissis P. Fabrication of a metal-free ceramic restoration utilizing the monobloc technique. Pract Periodontics Aesthet Dent 1995;7;83-94. 9. Pissis P. Fabrication of a metal-free ceramic restoration utilizing the monobloc technique. Pract Periodontics Aesthet Dent 1995;7;83-94. 10. Shin Y, Park S, Park JW, Kim KM, Park YB, Roh BD. Evaluation of the marginal and internal discrepancies of CAD/CAM endocrowns with different cavity depths: An in vitro study. J Prosthet Dent 2016;117;109-15. 10. Shin Y, Park S, Park JW, Kim KM, Park YB, Roh BD. Evaluation of the marginal and internal discrepancies of CAD/CAM endocrowns with different cavity depths: An in vitro study. J Prosthet Dent 2016;117;109-15. 11. Gaintantzopoulou MD, El-Damanhoury HM. Effect of Preparation Depth on the Marginal and Internal Adaptation of Computer-aided Design/Computer-assisted Manufacture Endocrowns. Oper Dent 2016;41;607-16. 11. Gaintantzopoulou MD, El-Damanhoury HM. Effect of Preparation Depth on the Marginal and Internal Adaptation of Computer-aided Design/Computer-assisted Manufacture Endocrowns. Oper Dent 2016;41;607-16. 12. Gresnigt MM, Özcan M, van den Houten ML, Schipper L, Cune MS. Fracture strength, failure type and Weibull characteristics of lithium disilicate and multiphase resin composite endocrowns under axial and lateral forces. Dent Mater 2016;32;607-14. 13. Rocca GT, Saratti CM, Poncet A, Feilzer AJ, Krejci I. The influence of FRCs reinforcement on marginal adaptation of CAD/CAM composite resin endocrowns after simulated fatigue loading. Odontology 2016;104;220-32. 14. El-Damanhoury HM, Haj-Ali RN, Platt JA. Fracture resistance and microleakage of endocrowns utilizing three CAD-CAM blocks. Oper Dent 2015;40;201-10. 15. Lin CL, Chang YH, Chang CY, Pai CA, Huang SF. Finite element and Weibull analyses to estimate failure risks in the ceramic endocrown and classical crown for endodontically treated maxillary premolar. Eur J Oral Sci 2015;118;87-93. 16. Dejak B, Młotkowski A. 3D-Finite element analysis of molars restored with endocrowns and posts during masticatory simulation. Dent Mater 2013;29;e309-17. 17. Zienkiewicz O, Tylor R, editors. Finite element method. The Basis, 1, 5th ed. Oxford: Butterworth– Heinemann 2000;87-110. 18. Habelitz S, Marshall S, Marshall G, Balooch M. Mechanical properties of human dental enamel on the nanometre scale. Arch Oral Biol 2001;46;173-83. 19. Craig R, Peyton F. Elastic and mechanical properties of human dentin. J Dent Res 1958;37;710-8. 20. Friedman CM, Sandrik JL, Heuer MA, Rapp GW. References Comparison of methods for measuring cuspal deformation in teeth. J Dent 2001;229;75-82. 30. Jantarat J, Panitvisai P, Palamara JE, Messer HH. Comparison of methods for measuring cuspal deformation in teeth. J Dent 2001;229;75-82. 31. Magne M, Belser U. Rationalization of shape and related stress distribution in posterior teeth: A finite element study using nonlinear contact analysis. Int J Periodontics Restorative Dent 2002;22;425-33. d f 31. Magne M, Belser U. Rationalization of shape and related stress distribution in posterior teeth: A finite element study using nonlinear contact analysis. Int J Periodontics Restorative Dent 2002;22;425-33. element study using nonlinear contact analysis. Int J Periodontics Restorative Dent 2002;22;425 33. 32. Anderson DJ. Measurement of stress in mastication. I. J Dent Res 1956;35, 664-70. 32. Anderson DJ. Measurement of stress in mastication. I. J Dent Res 1956;35, 664-70. 32. Anderson DJ. Measurement of stress in mastication. I. J Dent Res 1956;35, 664-70. 33. Lin J, Matinlinna JP, Shinya A, Botelho MG, Zheng Z. Effect of fiber post length and abutment height on fracture resistance of endodontically treated premolars prepared for zirconia crowns. Odontology. 2018;106;215-22. 33. Lin J, Matinlinna JP, Shinya A, Botelho MG, Zheng Z. Effect of fiber post length and abutment height on fracture resistance of endodontically treated premolars prepared for zirconia crowns. Odontology. 2018;106;215-22. 34. Biacchi GR, Basting RT. Comparison of fracture strength of endocrowns and glass fiber post-retained conventional crowns. Oper Dent 2012;37;130-6. 34. Biacchi GR, Basting RT. Comparison of fracture strength of endocrowns and glass fiber post-retained conventional crowns. Oper Dent 2012;37;130-6. 35. Forberger N, Göhring TN. Influence of the type of post andcore on in vitro marginal continuity, 35. Forberger N, Göhring TN. Influence of the type of post andcore on in vitro marginal continuity, fracture resistance, and fracture mode of lithia disilicate-based all-ceramic crowns. J Prosthet Dent 2008;100;264-73 35. Forberger N, Göhring TN. Influence of the type of post andcore on in vitro marginal continuity, fracture resistance, and fracture mode of lithia disilicate-based all-ceramic crowns. J Prosthet Dent 2008;100;264-73. g g yp p g y fracture resistance, and fracture mode of lithia disilicate-based all-ceramic crowns. J Prosthet Dent 2008;100;264-73. References Composition and mechanical properties of gutta- percha endodontic points. J Dent Res 1975;54;921-5. Page 9/13 Page 9/13 21. Borchers L, Reichart P. Three-dimensional stress distribution around a dental implant at different stages of interface development. J Dent Res 1983;62;155-9. 22. Lin J, Shinya A, Gomi H, Shinya A. Finite element analysis to compare stress distribution of connector of lithia-disilicate reinforced glass ceramic and zirconia based fixed partial denture. Odontology 2012;100;96-9. 23. Jie L, Shinya A, Lassila LV, Vallittu PK. Composite resin reinforced with pre-tensioned fibers: a three- dimensional finite element study on stress distribution. Odontology 2013;101;29-33. 24. Morris HF. The mechanical properties of metal ceramic alloys as cast and after simulated porcelain firing. J Prosthet Dent 1989;61;160-9. 25. Lanza A, Aversa R, Rengo S, Apicella D, Apicella A. 3D FEA of cemented steel, glass and carbon posts in a maxillary incisor. Dent Mater 2005;21;709-15. 26. Pegoretti A, Fambri L, Zappini G, Bianchetti M. Finite element analysis of a glass fibre reinforced composite endodontic post. Biomaterials 2002;23;2667-82. 26. Pegoretti A, Fambri L, Zappini G, Bianchetti M. Finite element analysis of a glass fibre reinforced composite endodontic post. Biomaterials 2002;23;2667-82. 27. Zheng Z, Lin J, Shinya A, Matinlinna JP, Botelho MG, Shinya A. Finite element analysis to compare stress distribution of gold alloy, lithium-disilicate reinforced glass ceramic and zirconia based fixed partial denture. J Investig Clin Dent 2012;3;291-7. 27. Zheng Z, Lin J, Shinya A, Matinlinna JP, Botelho MG, Shinya A. Finite element analysis to compare stress distribution of gold alloy, lithium-disilicate reinforced glass ceramic and zirconia based fixed partial denture. J Investig Clin Dent 2012;3;291-7. 28. Lin J, Zheng Z, Shinya A, Matinlinna JP, Botelho MG, Shinya A. Structural stability of posterior retainer design for resin-bonded prostheses: a 3D finite element study. Odontology 2015;103;333-8. 28. Lin J, Zheng Z, Shinya A, Matinlinna JP, Botelho MG, Shinya A. Structural stability of posterior retainer design for resin-bonded prostheses: a 3D finite element study. Odontology 2015;103;333-8. 29. Chuang SF, Yaman P, Herrero A, Dennison JB, Chang CH. Influence of post material and length on endodontically treated incisors: an in vitro and finite element study. J Prosthet Dent 2010;104;379-88. 29. Chuang SF, Yaman P, Herrero A, Dennison JB, Chang CH. Influence of post material and length on endodontically treated incisors: an in vitro and finite element study. J Prosthet Dent 2010;104;379-88. 30. Jantarat J, Panitvisai P, Palamara JE, Messer HH. Table 1 Properties of restorative materials and tooth tissue. Table 1 Properties of restorative materials and tooth tissue. Page 10/13 Young’s modulus (GPa) Poisson ratio Enamel 84.1(18) 0.33(18) Dentin 18.6(19) 0.31(19) Gutta-percha 0.14(20) 0.45(20) Periodontium 0.05(16) 0.45(16) Cortical bone/ Cancellous bone 13.7/ 1.37(21) 0.3(21) Luting resin cement 8.3(16) 0.35(16) LDRGC 96(22) 0.23(22) Zirconia 205(22) 0.19(22) Glass fiber post: Fiber_longitudinal/ Fiber_transverse 38.5/ 12(23) 0.35/0.11(23) Metal cast post （Ni-Cr） 188(24) 0.33(24) Stainless steel post 210(25) 0.3(25) Composite resin core 7(26) 0.3(26) LDRGC: lithia-disilicate reinforced glass ceramic LDRGC: lithia-disilicate reinforced glass ceramic Figures Figures Page 11/13 Figure 1 Figure 1 Page 11/13 2-D models of first mandibular molar tooth with roots and periodontium (bucco -lingual side view). Original contours developed from radiographic image of first mandibular molar tooth. (a) radiographic image of first mandibular molar tooth, (b) endocrown, (c) post-core crown. Thick arrow: simulated 100 N vertical occlusal load. AB and BC: distal root outer surface and distal root canal inner wall. Ring arrow: rotation center. Endocrown geometrically reduced the rotation center of the crown restoration in comparison with the full crown. 2-D models of first mandibular molar tooth with roots and periodontium (bucco -lingual side view). Original contours developed from radiographic image of first mandibular molar tooth. (a) radiographic image of first mandibular molar tooth, (b) endocrown, (c) post-core crown. Thick arrow: simulated 100 N vertical occlusal load. AB and BC: distal root outer surface and distal root canal inner wall. Ring arrow: rotation center. Endocrown geometrically reduced the rotation center of the crown restoration in comparison with the full crown. Figure 2 von Mises stress distribution maps in FE analysis results. Color bar indicates range of 0 to 50 MPa. Figure 2 von Mises stress distribution maps in FE analysis results. Color bar indicates range of 0 to 50 MPa. von Mises stress distribution maps in FE analysis results. Color bar indicates range of 0 to 50 MPa. Page 12/13 Page 12/13 Figure 3 The path plots of the interfacial von Mises stress along the distal root canal inner wall and the root outer surface (A→B→C). Blue line: SS post-Zr and SS post-LDRGC; Pink line: MC post-Zr and MC post-LDRGC; Red line: GF post-Zr and GF post-LDRGC; Yellow line: endocrown-2mm-Zr, endocrown-2mm-LDRGC, endocrown-4mm-Zr and endocrown-4mm-LDRGC. Peak 1: apical foramen; Peak 2: tip of post. Figure 3 The path plots of the interfacial von Mises stress along the distal root canal inner wall and the root outer surface (A→B→C). Blue line: SS post-Zr and SS post-LDRGC; Pink line: MC post-Zr and MC post-LDRGC; Red line: GF post-Zr and GF post-LDRGC; Yellow line: endocrown-2mm-Zr, endocrown-2mm-LDRGC, endocrown-4mm-Zr and endocrown-4mm-LDRGC Peak 1: apical foramen; Peak 2: tip of post Figure 3 The path plots of the interfacial von Mises stress along the distal root canal inner wall and the root outer surface (A→B→C). Figures Blue line: SS post-Zr and SS post-LDRGC; Pink line: MC post-Zr and MC post-LDRGC; Red line: GF post-Zr and GF post-LDRGC; Yellow line: endocrown-2mm-Zr, endocrown-2mm-LDRGC, endocrown-4mm-Zr and endocrown-4mm-LDRGC. Peak 1: apical foramen; Peak 2: tip of post. Page 13/13 Page 13/13
https://openalex.org/W1939528315	https://revistas.upch.edu.pe/index.php/RMH/article/download/614/581	es		Tratamiento de la anemia aplásica severa con gama globulina antilinfocítica.	Revista médica herediana	2,013	cc-by	277	Tratamiento de la anemia aplásica severa con gama globulina antilinfocítica. En este número de la Revista Médica Herediana se publica el trabajo de Ulloa, Ruiz, Tokumura y colaboradores, titulado gama globulina antilinfocítica en el tratamiento de la anemia aplásica severa. Este trabajo representa para los autores un gran logro, por haberlo realizado con profesionalismo en una población de escasos recursos económicos. En el se señala que la gama globulina antilinfocítica (GAL) es efectiva en ésta forma de anemia aplásica en un significativo número de pacientes; similar al obtenido por otros autores en trabajos con un mayor universo y en pacientes tratados con transplante de médula ósea alogénica (1) (2). Es necesario enfatizar que el tratamiento con transplante de médula ósea alogénica es el de preferencia en la opinión de algunos autores en la población muy joven afectada con este tipo de anemia; pero esta población también responde al tratamiento con GAL. Sin embargo es digno de remarcarse que es difícil encontrar donantes de médula ósea compatibles. Creo que en nuestro país así como en países semejantes, el tratamiento con gama globulina antilinfocítica es el recomendable: por ser efectivo, de menor costo que el transplante de médula ósea y por estar éste último limitado a ser realizado, aún en algunos hospitales con apropiada tecnología. Armando Silicani Della Pina* BIBLIOGRAFÍA 1. Issaragrisil, Kauffman DW Anderson TE et al. Aplastic Anemia study group.Br J Haematology. 1995; 91:80-4 * 2. Speck B, Gratwohl A, Nissen C, et al. Treatment of severe aplastic anemia with antilymphocyte globulin or bonemarrow transplantation. Br Med J 1981; 282:860 Profesor Principal de Medicina de la Universidad Peruana Cayetano Heredia. Rev Med Hered 10 (4), 1999 131
https://openalex.org/W3117684333	https://www.research-collection.ethz.ch/bitstream/20.500.11850/482481/3/Hansen2021_Article_GeometrizingNon-relativisticBi.pdf	English	null	Geometrizing non-relativistic bilinear deformations	The Journal of high energy physics/The journal of high energy physics	2,021	cc-by	19,577	ETH Library Journal Article Author(s): Hansen, Dennis; Jiang, Yunfeng; Xu, Jiuci Publication date: Creative Commons Attribution 4.0 International Open Access, c⃝The Authors. Article funded by SCOAP3. Originally published in: Originally published in: g y p Journal of High Energy Physics 2021(4), https://doi.org/10.1007/JHEP04(2021)186 This page was generated automatically upon download from the ETH Zurich Research Collection. For more information, please consult the Terms of use. This page was generated automatically upon download from the ETH Zurich Research Collection. For more information, please consult the Terms of use. Published for SISSA by Springer Published for SISSA by Springer Received: January 30, 2021 Revised: March 9, 2021 Accepted: March 21, 2021 Published: April 20, 2021 Received: January 30, 2021 Revised: March 9, 2021 Accepted: March 21, 2021 Published: April 20, 2021 Keywords: 2D Gravity, Field Theories in Lower Dimensions, Integrable Field Theories Geometrizing non-relativistic bilinear deformations JHEP04(2021)186 Dennis Hansen,a Yunfeng Jiangb,c and Jiuci Xud,e Dennis Hansen,a Yunfeng Jiangb,c and Jiuci Xud,e aInstitut für Theoretische Physik, Eidgenössische Technische Hochschule Zürich, Wolfgang-Pauli-Strasse 27, 8093 Zürich, Switzerland bDepartment of Theoretical Physics, CERN, 1 Esplanade des Particules, Geneva 23, CH-1211, Switzerland cShing-Tung Yau Center and School of physics, Southeast University, Nanjing 210096, China dDepartment of Physics, University of California, Santa Barbara, CA 93106, U.S.A. eUniversity of Science and Technology of China, 96 Jinzhai Road 230026, Hefei, Anhui, China E-mail: dehansen@phys.ethz.ch, jinagyf2008@gmail.com, Jiuci_Xu@umail.ucsb.edu Dennis Hansen,a Yunfeng Jiangb,c and Jiuci Xud,e aInstitut für Theoretische Physik, Eidgenössische Technische Hochschule Zürich, Wolfgang-Pauli-Strasse 27, 8093 Zürich, Switzerland bDepartment of Theoretical Physics, CERN, 1 Esplanade des Particules, Geneva 23, CH-1211, Switzerland cShing-Tung Yau Center and School of physics, Southeast University, Nanjing 210096, China dDepartment of Physics, University of California, Santa Barbara, CA 93106, U.S.A. eUniversity of Science and Technology of China, 96 Jinzhai Road 230026, Hefei, Anhui, China E-mail: dehansen@phys.ethz.ch, jinagyf2008@gmail.com, Jiuci_Xu@umail.ucsb.edu Abstract: We deﬁne three fundamental solvable bilinear deformations for any massive non-relativistic 2d quantum ﬁeld theory (QFT). They include the TT deformation and the recently introduced hard rod deformation. We show that all three deformations can be interpreted as coupling the non-relativistic QFT to a speciﬁc Newton-Cartan geometry, similar to the Jackiw-Teitelboim-like gravity in the relativistic case. Using the gravity for- mulations, we derive closed-form deformed classical Lagrangians of the Schrödinger model with a generic potential. We also extend the dynamical change of coordinate interpretation to the non-relativistic case for all three deformations. The dynamical coordinates are then used to derive the deformed classical Lagrangians and deformed quantum S-matrices. Keywords: 2D Gravity, Field Theories in Lower Dimensions, Integrable Field Theories ArXiv ePrint: 2012.12290 Open Access, c⃝The Authors. Article funded by SCOAP3. Open Access, c⃝The Authors. Article funded by SCOAP3. https://doi.org/10.1007/JHEP04(2021)186 1 3 3 5 5 6 7 8 9 10 10 11 14 16 16 17 1 Introduction 1 2 Newton-Cartan gravity in two dimensions 3 2.1 Geometric content 3 2.2 Flat Newton-Cartan spacetime 5 2.3 Covariant derivatives and matter actions 5 2.4 Matter currents 6 3 Bilinear deformations and Newton-Cartan gravity 7 3.1 Simple examples 8 3.2 The bilinear deformations 9 4 Deformed classical Lagrangians 10 4.1 Schrödinger model and conserved currents 10 4.2 Deformed Lagrangian I. The direct approach 11 4.3 Deformed Lagrangian II. 1 Introduction Recent studies of solvable irrelevant deformations of relativistic quantum ﬁeld theories have extended our understanding of them. The most studied example of such deformations is the TT deformation [1, 2], which can be deﬁned for any Lorentz invariant 2d QFT with a local stress energy tensor. For theories with additional symmetries, similar deformations such as the J ¯T [3], JTa deformations [4] and the ones constructed by higher conserved currents [5, 6] have been studied. All these deformations share similar features. They all modify the UV behavior of the QFTs and lead to non-local theories, yet they are under good analytical control due to the deformations’ solvability. JHEP04(2021)186 Surprisingly, such deformations can be deﬁned not only for 2d relativistic QFTs but also for a much broader class of theories. These include non-Lorentz invariant QFTs [7, 8], non-relativistic quantum many body systems such as the Bose gas [9] and lattice models like quantum spin chains [10–12]. Furthermore, it was recently shown [8, 9] that the de- formed 1d Bose gas share many qualitative features with TT deformed relativistic QFTs, such as the break down of UV physics for the spectrum and the Hagedorn behavior for thermodynamics. These ﬁndings hint that what we have seen so far from solvable deforma- tions of relativistic QFTs is only the tip of the iceberg — the structure and main features of such deformations can be extended to a much wider setting. One of the important lessons we learned from relativistic QFT is that TT deformation can be reformulated as coupling the QFT to certain special 2d topological gravity theory, both in ﬂat [13, 14] and curved space [15, 18], at least classically. The topological gravity one encounters in the relativistic case is similar to Jackiw-Teitelboim gravity [16, 17] on ﬂat spacetime, but it is not exactly the same gravity theory. Similar interpretations also holds for JTa deformation [4] where the gravity theory also involves an additional U(1) gauge ﬁeld. The 2d gravity formulation is tightly related to the random geometry interpre- tation of TT deformation [19] and the dynamical change of coordinates [18, 20–22]. These formulations oﬀer us a more geometrical understanding of the TT deformation. It is natural to ask whether similar geometrical interpretations exist for other settings. This paper is the ﬁrst step towards such an interesting goal by giving an aﬃrmative answer in the context of non-relativistic QFTs. Geometrizing non-relativistic bilinear deformations The gravity approach 14 5 Dynamical coordinates and gauge ﬁelds 16 5.1 The dynamical coordinates 16 5.2 Deformed Lagrangian III. Dynamical coordinates 17 5.3 Deformed quantum S-matrices 19 6 Conclusions and discussions 23 A Review of D-dimensional Newton-Cartan geometry 24 A.1 Newton-Cartan geometries 25 A.1.1 Type I 26 A.1.2 Type II 26 A.2 Connections and curvatures 27 A.3 Matter currents of ﬁeld theories on Newton-Cartan backgrounds 29 B Non-relativistic groups 30 C Null reductions 31 C.1 Lorentzian spacetimes with a null Killing vector 31 C.2 Klein-Gordon theory 32 JHEP04(2021)186 – i – 1 Introduction It is probably not too surprising that geometrical interpretations exist for non-Lorentz invariant QFTs, as the random geometry interpreta- tion was already pointed out in [7]. Nevertheless, notice that the random geometry picture follows directly from a Hubbard-Stratonovich transformation of the TT operator. It is not at all obvious from this what the gravity theory the QFT couples to is. Besides, given that the seed and deformed theories are not Lorentz invariant, the gravity theory cannot be a usual Einstein-Hilbert type gravity. A more natural candidate, in this case, is Newton- Cartan geometry, which is manifestly covariant under non-relativistic symmetry [23–27]. We will show that for non-relativistic QFTs, an elegant gravity interpretation for T¯T- and two other solvable deformations to be deﬁned shortly indeed exists in the framework of Type I Newton-Cartan geometry. For the TT deformation, the gravity theory written in the ﬁrst-order formalism takes the same form as its relativistic counterpart [14, 15]. The gravity theory for the other two deformations is similar to the ones for JTa deformed – 1 – relativistic QFTs [4]. From the gravity formulation, the dynamical change of coordinates interpretation naturally follows [15, 18]. The dynamical coordinates provide a powerful tool to compute several important quantities, such as the deformed classical Lagrangian and quantum S-matrices. Apart from extending what we have learned from relativistic QFTs to the non-relativistic settings, the gravity formulation for the non-relativistic QFTs is also in- teresting in its own right. To start with, most ﬁeld theories one encounters in condensed matter physics are not Lorentz invariant. TT deformations for Galilean or Bargmann (the central extension of the Galilei group [28]) invariant QFTs are deﬁnitely interesting to study. In particular, such QFTs are closely related to condensed matter systems that can be realized in experiments, one may gain more physical intuitions about TT deformation by studying such systems. Notice that Newton-Cartan gravity has already played a role in condensed matter systems such as quantum Hall eﬀect [29] and unitary Fermi gas [30]. It would be fascinating to make connections to these ﬁelds. JHEP04(2021)186 From the gravity point of view, Newton-Cartan geometry is diﬀerent from Einstein- Hilbert gravity and exhibits new features. One important diﬀerence is that there is a built- in U(1) symmetry in massive non-relativistic QFTs, which corresponds to the conservation of mass or particle number. 1To be more precise, this is true for one sign of the deformation parameter. For the other sign, the hard rod deformation increases the distance between particles. 1 Introduction Correspondingly, an U(1) gauge ﬁeld is an essential ingredient in so-called Type I Newton-Cartan geometry, which displays local Bargmann symmetry. Therefore, any Bargmann invariant 2d QFTs have at least three fundamental local sym- metry currents, which correspond to mass, momentum, and energy conservation. This implies, apart from the TT deformation, we can always deﬁne two other solvable bilinear deformations constructed from the mass current with the momentum or energy currents. We shall call these two deformations the hard rod deformation and the JE deformation, re- spectively. As we alluded before, these two deformations also allow a gravity interpretation in Newton-Cartan geometry. We study these three fundamental deformations in parallel. The hard rod and JE deformations are new in non-relativistic QFTs. They are similar to the JTa deformation of relativistic QFTs. The diﬀerence is that in the non-relativistic setting such U(1) symmetry is built-in and does not require further assumptions. Interestingly, the hard rod deformation was constructed very recently in [8, 9]. It is found that this deformation has the eﬀect of deforming the point particles to ﬁnite size hard rods.1 We will conﬁrm this intuition and show that it can be formulated equivalently as coupling the undeformed theory to a speciﬁc Newton-Cartan geometry. Intuitively, this relation can be understood as follows. Traditionally, the hard rod model can be solved by performing a change of coordinates which eliminates the sizes of the rods and leaves only the free space (see for example [31]). In the new coordinate, the rods become point particles and the model can be solved readily. Therefore, the crucial point here is the change of coordinates. Since this change amounts to a redeﬁnition of the length, it can be formulated as putting the theory on a diﬀerent geometry where the metric is deﬁned diﬀerently. A covariant way to formulate this intuition is precisely coupling the original – 2 – theory to the proper Newton-Cartan geometry. We will see that the change of coordinate we mentioned above appear naturally from the 2d gravity formulation in section 5. The TT deformation can be understood in the same way. Under TT deformation, point particles also become hard rods.2 The only diﬀerence from the hard rod deformation is that now the size of each rod is no longer a ﬁxed number, but depends on the energy of the particle (or the rod), which needs to be determined self-consistently. 1 Introduction Therefore we obtain certain ‘dynamical’ hard rod model. This new kind of hard rod model can also be solved by a change of coordinate, but now the new coordinates depend on the stress-energy tensor of the theory and thus become ﬁeld-dependent, or dynamical. This gives a clear physical understanding of the dynamical change of coordinates, both in the non-relativistic case and the Lorentz invariant case. This also explains why the TT deformed theory is non-local. It is simply because we are describing ﬁnite size objects such as strings or hard rods in the deformed theory. JHEP04(2021)186 The paper is structured as follows: In section 2, we give a pedagogical review of Type I Newton-Cartan gravity in 2d. In section 3, we derive the gravity formulations for the three bilinear deformations. This is achieved by viewing these deformations’ deﬁnitions as equations for the classical action and solving them formally by a heat-kernel-like approach. In section 4, we apply and test our gravity formulation by deriving closed-form classical Lagrangians. We compare the results with the ones obtained from a direct approach and ﬁnd a perfect match. In section 5, we derive the dynamical change of coordinates from the gravity formulation. The dynamical coordinates provide yet another approach to derive the deformed Lagrangian. In addition, we derive the deformed quantum S-matrices using the dynamical coordinates. We conclude in section 6 and discuss future directions. Appendix A is dedicated to a more detailed introduction to Newton-Cartan geometry. 2Again for the sign where the eﬀective length of the system becomes smaller. For the other sign, the space between the particles is increased. 2 Newton-Cartan gravity in two dimensions We give a pedagogical and minimal review on Newton-Cartan geometry suitable for our aims in the subsequent sections. Only the formalism required for that is introduced. A general review can be found in appendix A. 2.1 Geometric content Newton-Cartan (NC) geometry is the natural framework to study non-relativistic ﬁeld the- ories and their coupling to gravity. It is a covariant formulation where the Galilean group is a subgroup of the local symmetry group. It was ﬁrst obtained as a geometrization of Newtonian spacetime by Cartan [23, 24], but has seen a revival in recent years. There is extensive literature on the subject and its applications to a wide selection of areas such as generalized holography [26, 32–42], condensed matter theory [29, 43–47], hydrodynam- ics [48–54], string theory [55–63] and more [64–68]. We shall consider the 2-dimensional gravity relevant to our aims. – 3 – In 2d NC geometry, there are two fundamental tensors: • the clock-form τµ, giving the local ﬂow of time and • the clock-form τµ, giving the local ﬂow of time and • the spatial vector eµ, giving the local space direction. • the spatial vector eµ, giving the local space direction. Newton-Cartan geometry is deﬁned by requiring that they satisfy the fundamental orthog- onality relation (2.1) τµeµ = 0. (2.1) This implies that there is indeed a preferred direction of time in NC geometry, and the structure is implemented covariantly. However, we may only deﬁne projective inverses −vµ, eµ that preserves (2.1). The ambiguity in selecting the inverses is exactly the Galilean boost freedom, which is also referred to as Milne boosts in the literature [43]. JHEP04(2021)186 JHEP04(2021)186 When grouped together, the ﬁelds naturally form Galilean zweibeine deﬁned as EA µ = (τµ, eµ) , Eµ A = (−vµ, eµ) , (2.2) (2.2) and satisfying the completeness relations EA µ Eµ B = δA B, EA µ Eν A = δν µ. (2.3) (2.3) Written out in components these are Written out in components these are τµeµ = 0, τµvµ = −1, eµeµ = 1, eµvµ = 0, −τµvν + eµeν = δν µ. (2.4) The zweibeine transform under diﬀeomorphisms and local Galilean transformations as δτµ = Lξτµ, δeµ = Lξeµ + η τµ, (2.5) δvµ = Lξvµ + η eµ, δeµ = Lξeµ, (2.6) (2.5) (2.6) (2.6) where δxµ := ξµ is a vector ﬁeld so that Lξ is the Lie derivative LξXµν = ξρ∂ρXµν −∂ρξµXρν + ∂νξρXµρ, (2.7) (2.7) that generates inﬁnitesimal diﬀeomorphisms of the zweibeine and η = η(x0, x1) is a local Galilean boost. 2.2 Flat Newton-Cartan spacetime JHEP04(2021)186 An important special case is, of course, ﬂat Newton-Cartan spacetime [44, 71]. For suitable coordinates x0 = t, x1 = x it is given by τµ = δt µ, eµ = δx µ, vµ = −δµ t , eµ = δµ x, mµ = ∂µθ, (2.10) (2.10) where θ is an arbitrary function, i.e. mµ is a pure gauge. The residual coordinate transfor- mations are exactly global Galilean transformations t′ = t + a, (2.11) x′ = x + vt + b, (2.12) θ′(t′, x′) = θ(t, x) −1 2v2t + vx, (2.13) (2.11) (2.13) where we have the parameters v for the Galilean boost, and a, b for the translations. In particular we may choose θ = constant, but it will regain spacetime dependence after doing a boost. where we have the parameters v for the Galilean boost, and a, b for the translations. In particular we may choose θ = constant, but it will regain spacetime dependence after doing a boost. 2.1 Geometric content Contrary to the relativistic case, the boost does not aﬀect all the compo- nents in the same way — the hallmark of non-relativistic physics. We see that τµ, eµ indeed transform as tensors since they are invariant under local Galilean transformations. The general dimensional case where spatial rotations also enter can be found in appendix A.1. A generalization of the Newtonian potential Φ also arises. The principle of covariance tells us that it must be obtained as a projection of a gauge ﬁeld. Indeed one ﬁnds that it is given by Φ = −vµmµ, (2.8) (2.8) where mµ is a U(1) gauge ﬁeld known as the mass (or particle number) gauge ﬁeld. mµ transforms as δmµ = Lξmµ + η eµ + ∂µσ, (2.9) (2.9) – 4 – where σ is a U(1) transformation parameter. From a more group theoretical perspective, mµ is the gauge connection associated with the central charge of the Bargmann group. The Bargmann group is the non-trivial central extension of the Galilean group, and the extra generator corresponds to mass or particle number conservation [28]. We review these groups in appendix B. In other words, the geometry we have described above is the result of gauging the Bargmann group, a useful approach that has been studied in for example [25, 69, 70]. This also makes it clear that mµ is an integral part of the geometry. 2.3 Covariant derivatives and matter actions If we want to geometrize a given non-relativistic ﬁeld theory, we must use Galilean covariant derivatives for the theory to be compatible with local Galilei transformations. Analogous to the relativistic case, we may introduce a covariant derivative ∇µ with a Galilean or Newton-Cartan aﬃne connection Γρ µν and a spin connection ωµAB. The construction is similar to Lorentzian geometry, with the main diﬀerence being that the symmetry of the tangent and frame bundles is Galilean [72–75]. For this paper, it is suﬃcient to restrict to scalar ﬁelds φ where it, of course, reduces to the partial derivative, ∇µφ = ∂µφ. The general case is discussed further in appendix A.2. Besides the local Galilean symmetry, the scalar ﬁelds also transform under U(1) mass or particle number symmetry. To not break the particle number symmetry, we must use a U(1) covariant derivative. This is naturally deﬁned by taking Dµ = ∂µ + mmµ, (2.14) (2.14) where m is the mass of the ﬁeld. A temporal derivative is then formed in a covariant way as ∼vµDµ and a spatial derivative as ∼eµDµ. Notice that the ﬁrst transform under a local Galilean boost, which must be compensated by other terms. – 5 – Given a matter theory with ﬁeld ϕ coupled to a Newton-Cartan geometry, we can write its Bargmann invariant action S and Lagrangian L as S [ϕ, τ, e, m] = Z M d2xe L [ϕ, τ, e, m] , (2.15) (2.15) where we have the Galilean invariant measure e := det(τµ, eµ). (2.16) (2.16) All dependence on mµ must be through the covariant derivative. JHEP04(2021)186 In general, it is not an easy task to form Lagrangians that respect both boost and particle number symmetries [43, 45, 76]. One way where this is guaranteed is to consider null reductions of relativistic theories: Any 3d ﬁeld theory on a Lorentzian background with a null Killing vector can be null-reduced to a 2d ﬁeld theory on the type of Newton- Cartan background we have considered here. In particular, Schrödinger-type models are obtained by null reductions of Klein-Gordon-type actions. In appendix C, we review this procedure in more detail. 2.4 Matter currents We can deﬁne three covariantly conserved currents of the action (2.15) as the response to the variation of the background geometry: δbgdS [ϕ, τ, e, m] := Z M d2x e (Eµδτµ + Pµδeµ + J µδmµ) (2.17) (2.17) or equivalently or equivalently Eµ := e−1 δS δτµ , Pµ := e−1 δS δeµ , J µ := e−1 δS δmµ , (2.18) (2.18) where here Eµ is the energy current, Pµ the momentum current and J µ the mass current. These three currents are of fundamental importance to our constructions below. We stress again that the mass current is always present in a Bargmann invariant theory. In fact, breaking the U(1) symmetry would lead to drastic consequences. The particles’ mass would no longer be well-deﬁned, and the non-relativistic dispersion relation E = P 2/(2m) would not hold. Such a situation is rather unphysical. In quantum mechanics, we would furthermore have that the states are not localizable [28]. In conclusion, we better not break the Bargmann symmetry to Galilei if we want to avoid odd physics. The diﬀeomorphisms of the background ﬁelds (2.5)–(2.6) give the covariant conserva- on laws of the currents: 0 = ∂µEµ+ e−1∂µe Eµ, 0 = ∂µPµ+ e−1∂µe Pµ, 0 = ∂µJ µ+ e−1∂µe J µ. (2.19) (2.19) These can be rewritten in terms of the covariant derivative ∇µ using e−1∂µe = Γρ µρ for any Newton-Cartan connection Γρ µν. We then obtain for Cµ = {Eµ, Pµ, J µ} 0 = ∇µCµ + 2Γρ [µρ]Cµ, (2.20) (2.20) – 6 – – 6 – where 2Γρ [µρ] is the torsion of the chosen connection. Unlike the Levi-Civita connection, Newton-Cartan connections are naturally torsionful and non-metric compatible [75]. More discussion can be found in appendix A.2. where 2Γρ [µρ] is the torsion of the chosen connection. Unlike the Levi-Civita connection, Newton-Cartan connections are naturally torsionful and non-metric compatible [75]. More discussion can be found in appendix A.2. Finally, as the ﬁelds are not manifestly boost-invariant, a Galilean boost relates Pµ and J µ through the on-shell Ward identity Pµτµ = −J µeµ, (2.21) (2.21) which shows that Pµ should be thought of as the stress-mass current. This relation is well-known for non-relativistic theories on ﬂat spacetimes. Here it shows up as relations between the Noether currents after a simplifying redeﬁnition [77]. JHEP04(2021)186 3 Bilinear deformations and Newton-Cartan gravity This section deﬁnes three fundamental bilinear deformations for non-relativistic QFTs and derives the corresponding gravity formulations. As reviewed in the previous section, any 2d non-relativistic Bargmann QFT has three fundamental symmetries, corresponding to mass, momentum, and energy conservation. The conserved currents are denoted by J µ, Pµ and Eµ, respectively, as reviewed in the previous section. We can choose any two of them and construct a bilinear operator. For instance, the TT operator corresponds to the choice O2,1 = e ϵµνEµPν = e det(Eµ, Pν), (3.1) (3.1) where ϵµν is the Levi-Civita symbol. The bilinear operator which triggers the hard rod deformation is O0,1 = e ϵµνJ µPν. The remaining bilinear operator is O0,2 = e ϵµνJ µEν. We shall call the corresponding deformation the JE deformation. The three bilinear defor- mations are deﬁned by dSλ dλ = − Z d2x e Oa,b(x), (3.2) (3.2) here Sλ is the deformed action and (a, b) = (0, 1), (0, 2), (2, 1). where Sλ is the deformed action and (a, b) = (0, 1), (0, 2), (2, 1). where Sλ is the deformed action and (a, b) = (0, 1), (0, 2), (2, 1). To derive the gravity formulation, we use the fact that the three conserved currents can be written as the variations of the action as deﬁned through (2.18). Using these relations, the deﬁnition (3.2) can be seen as equations for the classical action. Let us explain this point more explicitly by the TT deformation. Consider the deformed partition function Zλ = Z Dφ e−Sλ[φ]. (3.3) (3.3) From (3.2), we have dZλ dλ = Z Dφ −dSλ[φ] dλ e−Sλ[φ] = bD2,1Zλ, (3.4) (3.4) where where bD2,1 = Z d2x ϵµν : δ δτµ(x) δ δeν(x) : . (3.5) (3.5) – 7 – Here the normal ordering means we subtract the contribution of the term proportional to δ2S/δτµδeν.3 Similar equations can be derived for the other two deformations. Notice that (3.4) is similar to a diﬀusion equation. We can solve it formally by a heat kernel-like approach, which will be discussed shortly. The formal solution gives us the gravity action immediately. This approach has been applied in the relativistic QFTs in [78–80]. 3.1 Simple examples 3.2 The bilinear deformations 3.2 The bilinear deformations The bilinear deformations We can apply the same method to solve the ﬂow equation of the bilinear deformations (3.2). Let us consider the TT deformed partition function We can apply the same method to solve the ﬂow equation of the bilinear deformations (3.2). Let us consider the TT deformed partition function We can apply the same method to solve the ﬂow equation of the bilinear deformations (3.2). Let us consider the TT deformed partition function ∂λZλ[τµ, eµ] = bD2,1Zλ[τµ, eµ]. (3.14) (3.14) Following the same steps, we arrive at the following formal solution Zλ[τµ, eµ] = Z D˜τµD˜eµ exp −1 λ Z d2xϵµν(τµ −˜τµ)(eν −˜eν) Z0[˜τµ, ˜eµ], (3.15) (3.15) JHEP04(2021)186 where as before we neglected the prefactors from the Gaussian integral. From (3.15), we can extract the deformed classical action STT λ [τµ, eµ; ˜τµ, ˜eµ\|φ] = 1 λ Z d2x ϵµν(τµ −˜τµ)(eν −˜eν) + S0[˜τµ, ˜eµ\|φ]. (3.16) (3.16) Here S0 is the undeformed action on the background described by the auxiliary zweibein (˜τµ, ˜eµ). The other term can be interpreted as the non-relativistic gravity action, which couples to the undeformed theory. Notice that the gravity action takes the same form as its relativistic counterpart [14], but it is also manifestly Galilean invariant under (2.5)–(2.6). Similarly, for the hard rod and JE deformations, we ﬁnd the deformed classical actions SHR λ [mµ, eµ; ˜mµ, ˜eµ\|φ] = 1 λ Z d2x ϵµν(mµ −˜mµ)(eν −˜eν) + S0[ ˜mµ, ˜eµ\|φ] (3.17) (3.17) and SJE λ [mµ, τµ; ˜mµ, ˜τµ\|φ] = 1 λ Z d2x ϵµν(mµ −˜mµ)(τν −˜τν) + S0[ ˜mµ, ˜τµ\|φ]. (3.18) (3.18) Notice that for these two deformations, the gravity action involve both the zweibein and the U(1) gauge ﬁeld, similar to those of JTa deformed QFTs [4, 80]. Since our derivation of the deformed action is somewhat heuristic, let us now verify that the deformed actions indeed satisfy the deﬁnition of the bilinear deformations (3.2). We consider the TT deformation as an example. The proof for the other two cases is similar. The action (3.16) depends on both the zweibein (τµ, eµ) and the auxiliary zweibein (˜τµ, ˜eµ). To obtain the TT deformed classical action, we need to integrate out the auxiliary zweibein (˜τµ, ˜eµ). In practice, this means ﬁnding the saddle-point solution for (˜τµ, ˜eµ) and plug it back in. We denote the saddle-point solution by ˜τ ⋆ µ and ˜e⋆ µ. 3.1 Simple examples As a warm-up, we consider two simple examples. The ﬁrst one is the 1d heat equation ∂tf(t, x) = ∂2 xf(t, x), (3.6) (3.6) JHEP04(2021)186 JHEP04(2021)186 which can be solved in a few steps: First, we can write which can be solved in a few steps: First, we can write f(t, x) = et∂2 xf(0, x). (3.7) (3.7) Secondly, rewrite Secondly, rewrite f(0, x) = Z dy δ(x −y)f(0, y) = 1 2π Z dy Z dp eip(x−y)f(0, y). (3.8) (3.8) Finally, plug (3.8) into the right-hand side of (3.7) and integrate out p. We obtain f(t, x) = 1 2 √ πt Z dy e−1 4t(x−y)2f(0, y). (3.9) (3.9) This is nothing but the heat kernel solution. This is nothing but the heat kernel solution. g Next, we consider a slightly more non-trivial equation which involves functio- nal derivatives4 Next, we consider a slightly more non-trivial equation which involves functio- nal derivatives4 ∂tZt[φ] = Z d2x : δ δφ(x) δ δφ(x) : Zt[φ]. (3.10) (3.10) This equation can be solved similarly. First, we have This equation can be solved similarly. First, we have Zt[φ] = exp t Z d2x : δ δφ(x) δ δφ(x) : Z0[φ]. (3.11) (3.11) Secondly, rewrite Z0[φ] = Z Dϕ δ(φ −ϕ)Z0[ϕ] ∝ Z Dϕ Z DJ e R d2xJ(x)[φ(x)−ϕ(x)]Z0[ϕ], (3.12) Z0[φ] = Z Dϕ δ(φ −ϕ)Z0[ϕ] ∝ Z Dϕ Z DJ e R d2xJ(x)[φ(x)−ϕ(x)]Z0[ϕ], (3.12) (3.12) where δ(φ −ϕ) is the functional delta-function in the proper sense and ϕ is an auxiliary scalar ﬁeld to be integrated over. Plugging into (3.11) and integrating out J(x), we obtain Zt[φ] ∝ Z Dϕ e−1 4t R d2x[φ(x)−ϕ(x)]2Z0[ϕ], (3.13) (3.13) where we have neglected the prefactors which are not important for us. 3More explicitly, we have : δ2 δτµ(x)δeν(x)e−Sλ := δ2 δτµ(x)δeν(x)e−Sλ + δ2Sλ δτµ(x)δeν(x)e−Sλ. 4The interpretation of normal ordering is the same as (3.5). – 8 – – 8 – 3.2 The bilinear deformations We will show that STT[τµ, eµ\|φ] := STT λ [τµ, eµ; ˜τ ⋆ µ, ˜e⋆ µ\|φ] satisﬁes the deﬁnition of the TT deformation. Taking variation of (3.16) with respect to ˜τµ and ˜eµ, we obtain the saddle-point equations τµ = ˜τ ⋆ µ + λ ˜e ϵµν Pν 0 , eµ = ˜e⋆ µ −λ ˜e ϵµνEν 0 , (3.19) (3.19) where P0 and E0 are the undeformed currents. On the other hand, taking variations with respect to τµ and eµ lead to the deﬁnitions of the deformed currents δSTT[τµ, eµ\|φ] δτµ = e Eµ, δSTT[τµ, eµ\|φ] δeµ = e Pµ, (3.20) (3.20) – 9 – – 9 – which can be written as which can be written as which can be written as τµ = ˜τµ + λ e ϵµν Pν, eµ = ˜eµ −λ e ϵµνEν. (3.21) (3.21) Here Eµ and Pµ are the deformed currents. Taking derivative of STT with respect to λ, we have Here Eµ and Pµ are the deformed currents. Taking derivative of STT with respect to λ, we have dSTT dλ = −1 λ2 Z d2x ϵµν(τµ −˜τ ⋆ µ)(eν −˜e⋆ ν) + δSTT δ˜τ ⋆µ ∂˜τ ⋆ µ ∂λ + δSTT δ˜e⋆µ ∂˜e⋆ µ ∂λ . (3.22) (3.22) From the deﬁnition of saddle-point equation, JHEP04(2021)186 δSTT δ˜τ ⋆µ = δSTT δ˜e⋆µ = 0, (3.23) (3.23) and therefore the last two terms in (3.22) vanish. Using (3.21), we have dSTT dλ = − Z d2x e2 ϵµνEµPν = − Z d2x e O2,1, (3.24) (3.24) which is precisely (3.2). The generalization to the other two cases is straightforward. 4 Deformed classical Lagrangians In this section, we derive the deformed classical Lagrangian for the Schrödinger model with a generic potential for all three bilinear deformations. The same result can be derived from two diﬀerent approaches. The ﬁrst one is the direct approach where one starts from the deﬁnition (3.2) and work out the deformed Lagrangian order by order in λ; the other approach exploits the gravity interpretation we developed in the previous section and gives closed-form expressions for the deformed Lagrangians. The fact that these two quite dif- ferent methods lead to the same deformed Lagrangians can be seen as a non-trivial test of our proposal. 4.1 Schrödinger model and conserved currents (4.5) Eµ = i 2δµ t φ∂tφ† −φ†∂tφ −2iL + δµ x ∂xφ†∂tφ + ∂xφ∂tφ† , Pµ = i 2δµ t φ∂xφ† −φ†∂xφ + δµ x L + 2∂xφ∂xφ† , 1 i J µ = −1 2δµ t φφ† + i 2δµ x φ†∂xφ −φ∂xφ† . (4.5) (4.5) In general curved background, the currents can be calculated via (2.18), leading to Eµ = i 2vµvρ φDρφ† −φ†Dρφ −eµeρvσ Dρφ†Dσφ + Dσφ†Dρφ −vµ L, Pµ = i 2vµeρ φ†Dρφ −φDρφ† + eµeρeσ Dρφ†Dσφ + Dσφ†Dρφ + eµ L, J µ = 1 2vµ φφ† + i 2eµeν φ†Dνφ −φDνφ† . (4 Eµ = i 2vµvρ φDρφ† −φ†Dρφ −eµeρvσ Dρφ†Dσφ + Dσφ†Dρφ −vµ L, Pµ = i 2vµeρ φ†Dρφ −φDρφ† + eµeρeσ Dρφ†Dσφ + Dσφ†Dρφ + eµ L, (4.6) J µ = 1 2vµ φφ† + i 2eµeν φ†Dνφ −φDνφ† . (4.6) It is easy to check that on ﬂat spacetime as described in section 2.2, (4.6) reduces to (4.5). 4.1 Schrödinger model and conserved currents Let us ﬁrst deﬁne the Schrödinger model and its conserved currents. In curved space, the action is given by SSch = Z d2x e i 2vµ(φDµφ† −φ†Dµφ) −eµeνDµφ†Dνφ −V (\|φ\|) , (4.1) (4.1) where φ is a complex scalar ﬁeld and V (\|φ\|) is any potential that does not depend on the metric nor on the covariant derivative (2.14) with m = 1/2. Furthermore, we require that the potential is invariant under Hermite conjugation. Taking V (\|φ\|) = 0 leads to the non- relativistic free boson. One slightly more non-trivial example is the Lieb-Liniger model, or the non-linear Schrödinger model where we take V (\|φ\|) = c φ†φ†φφ, with c being the coupling constant. One nice feature of the Schrödinger model is that it can be obtained from a relativistic 3d Klein-Gordon type theory. This is described in detail in appendix C. – 10 – In ﬂat spacetime, we recover the familiar Schrödinger action In ﬂat spacetime, we recover the familiar Schrödinger action S = Z d2x L = Z d2x i 2(φ†∂tφ −φ∂tφ†) −∂xφ†∂xφ −V (\|φ\|) . (4.2) (4.2) Now we discuss the symmetries of the action (4.2). Spacetime translation invariance leads to a conserved local Noether stress-energy tensor T µν. In terms of the Lagrangian, T µν = ∂L ∂(∂µφ)∂νφ + ∂L ∂(∂µφ†)∂νφ† −L δµ ν . (4.3) (4.3) The energy and momentum currents are identiﬁed with Eµ = T µt and Pµ = T µx, respec- tively. In addition, there is a global U(1) symmetry φ →eiθφ which is related to the conservation of mass. The Noether current is given by The energy and momentum currents are identiﬁed with Eµ = T µt and Pµ = T µx, respec- tively. In addition, there is a global U(1) symmetry φ →eiθφ which is related to the conservation of mass. The Noether current is given by JHEP04(2021)186 J µ = i 2 ∂L ∂(∂µφ)φ − ∂L ∂(∂µφ†)φ† ! . (4.4) (4.4) The corresponding conserved charge is the total mass of the system. The three conserved currents written explicitly are Eµ = i 2δµ t φ∂tφ† −φ†∂tφ −2iL + δµ x ∂xφ†∂tφ + ∂xφ∂tφ† , Pµ = i 2δµ t φ∂xφ† −φ†∂xφ + δµ x L + 2∂xφ∂xφ† , J µ = −1 2δµ t φφ† + i 2δµ x φ†∂xφ −φ∂xφ† . 4.2 Deformed Lagrangian I. The direct approach We are now ready to derive the deformed Lagrangians for the three bilinear deformations. In this subsection, we perform the calculation using the direct approach. This was the ﬁrst approach to derive the deformed Lagrangian in the relativistic case (see for example [2, 81]). In this approach, one performs a formal expansion of the deformed Lagrangian Lλ = ∞ X k=0 Lk λk (4.7) (4.7) and calculates Lk order by order using the deﬁnition. This can always be worked out explicitly up to certain orders in λ. By observing the patterns of Lk, one can usually make – 11 – an ansatz for the deformed Lagrangian and turn the deﬁnition into a diﬀerential equation, which can be solved and gives the deformed Lagrangian. This strategy works fairly well for the hard rod deformation because the deformed Lagrangian takes a compact form, and it is relatively easy to guess the general pattern. However, for TT and JE deformations, the closed-form deformed Lagrangians are quite complicated, as we will see shortly. Therefore it is hard to see the patterns in these cases. Nevertheless, we can work out the results up to relatively high orders as perturbative data, which can be checked against the closed-form expressions obtained from other approaches. The hard rod deformation. Let us start with the simplest case, namely the hard rod deformation. We work in a ﬂat spacetime. The deﬁnition is JHEP04(2021)186 JHEP04(2021)186 dLHR λ dλ = −ϵµνJ µPν. (4.8) (4.8) We expand both the Lagrangian and current densities in λ We expand both the Lagrangian and current densities in λ Lλ = ∞ X n=0 Lnλn, J µ = ∞ X n=0 Jµ nλn, Pµ = ∞ X n=0 P µ n λn. (4.9) (4.9) Plugging into (4.8), we obtain the following recursion relation Ln+1 = − 1 n + 1 n X k=0 ϵµνJµ k P ν n−k, (4.10) (4.10) where Jµ k and T µ k are deﬁned via Lk: where Jµ k and T µ k are deﬁned via Lk: where Jµ k and T µ k are deﬁned via Lk: Jµ k := i ∂Lk ∂(∂µφ)φ − ∂Lk ∂(∂µφ†)φ† ! , P µ k := ∂Lk ∂(∂µφ)∂xφ + ∂Lk ∂(∂µφ†)∂xφ† −Lk δµ x. (4.11) (4.11) Using (4.11), (4.10) and the initial condition L0 = L in (4.2), we can calculate Lk order by order. 4.2 Deformed Lagrangian I. The direct approach The ﬁrst few orders are given by L1 = −φφ† 2 ! L + φxφ† x −1 4 φ2(φ† x)2 + φ2 x(φ†)2 , L2 = −φφ† 2 !2 L + φxφ† x + 1 16φφ†(φ† xφ −φ†φx)2, L3 = −φφ† 2 !3 L + φxφ† x −1 32(φφ†)2(φ† xφ −φ†φx)2. (4.12) (4.12) where we have deﬁned the shorthand notations where we have deﬁned the shorthand notations φt := ∂tφ, φx := ∂xφ, φ† t := ∂tφ†, φ† x := ∂xφ†. (4.13) (4.13) – 12 – – 12 – Working out a few more orders, we can ﬁnd the pattern Ln = −φφ† 2 !n (L + φxφ† x) + (−1)n 2n+2 (φφ†)n−1(φ† xφ −φ†φx), n ≥2. (4.14) (4.14) The full deformed Lagrangian is then given by The full deformed Lagrangian is then given by LHR λ = 1 2 + λφφ† 2L −λ 8 (4 + λφφ†)(φ† xφ + φxφ†)2 . (4.15) (4.15) The TT deformation. Now we consider the TT deformation whose deﬁnition is JHEP04(2021)186 JHEP04(2021)186 dLλ dλ = −ϵµνEµPν. (4.16) (4.16) Similarly, we expand the Lagrangian and the currents in λ Similarly, we expand the Lagrangian and the currents in λ Lλ = ∞ X n=0 Ln λn, Eµ = ∞ X n=0 Eµ nλn, Pµ = ∞ X n=0 P µ n λn, (4.17) (4.17) which leads to a similar recursion relation which leads to a similar recursion relation Ln+1 = − 1 n + 1 n X k=0 ϵµνEµ k P ν n−k, (4.18) Ln+1 = − 1 n + 1 n X k=0 ϵµνEµ k P ν n−k, (4.18) (4.18) where where where ∂Lk ∂Lk † where Eµ k := ∂Lk ∂(∂φ)∂tφ + ∂Lk ∂(∂φ†)∂tφ† −Lkδµ t (4.19) where Eµ k := ∂Lk ∂(∂µφ)∂tφ + ∂Lk ∂(∂µφ†)∂tφ† −Lkδµ t (4.19) Eµ k := ∂Lk ∂(∂µφ)∂tφ + ∂Lk ∂(∂µφ†)∂tφ† −Lkδµ t (4.19) (4.19) and P µ k is given in (4.11). The ﬁrst few orders are L0 = L and and P µ k is given in (4.11). 4.2 Deformed Lagrangian I. The direct approach The ﬁrst few orders are L0 = L and L1 = 1 2 iφφtφ†2 x −iφ†φ† tφ2 x + φ†2 x φ2 x + i 2 φ†φt −φ† tφ V −V2, L1 = 1 2 iφφtφ†2 x −iφ†φ† tφ2 x + φ†2 x φ2 x + i 2 φ†φt −φ† tφ V −V2, L2 = −V3 + i 2 φ†φt −φφ† t V2 −φ†2 x φ2 x V −2φ3 xφ†3 x + iφxφ† x φ† tφ2 xφ† −φtφ†2 x φ −1 2φtφ† tφxφ† xφφ† + i 2φ2 xφ†2 x φtφ† −φ† tφ + 1 4φtφ† t φ†2 x φ2 + φ2 xφ†2 . (4.20) 2 2 L2 = −V3 + i 2 φ†φt −φφ† t V2 −φ†2 x φ2 x V −2φ3 xφ†3 x + iφxφ† x φ† tφ2 xφ† −φtφ†2 x φ −1 2φtφ† tφxφ† xφφ† + i 2φ2 xφ†2 x φtφ† −φ† tφ + 1 4φtφ† t φ†2 x φ2 + φ2 xφ†2 . (4.2 (4.20) We see that the second-order result is already quite lengthy. Higher-order terms are more complicated, and it is hard to see the general pattern. Nevertheless, we can derive a closed-form result from the gravity approach. he JE deformation. This deformation is deﬁned using the mass and energy currents The JE deformation. This deformation is deﬁned using the mass and energy currents dLJE λ dλ = −ϵµνJ µEν. (4.21) (4.21) We expand the Lagrangian and currents in λ as before, which leads to the following recur- sion relation n We expand the Lagrangian and currents in λ as before, which leads to the following recur- sion relation LJE n+1 = − 1 n + 1 n X k=0 ϵµνJµ k Eν n−k, (4.22) (4.22) – 13 – – 13 – where Jµ k and Eν k are deﬁned via Lk in (4.11) and (4.19). The ﬁrst few Lk are given by L0 = L and where Jµ k and Eν k are deﬁned via Lk in (4.11) and (4.19). 4.2 Deformed Lagrangian I. The direct approach The ﬁrst few Lk are given by L0 = L and L1 = i 2 φxφ† −φ† xφ V + i 2φxφ† x φxφ† −φ† xφ −1 2φφ† φtφ† x + φ† tφx , L2 = −1 4φφ†V2 + 1 4 φ2φ†2 x + φ†2φ2 x −4φφ†φxφ† x + iφφ† φtφ† −φ† tφ V + 1 4φxφ† x φ2φ†2 x + φ†2φ2 x −1 4φtφ† tφ2φ†2 −3 4φ2 xφ†2 x φφ† + i 2φxφ† xφφ† φtφ† −φ† tφ −i 4φφ† φtφ†2 x φ −φ† tφ2 xφ† . (4.23) (4.23) JHEP04(2021)186 Like the TT deformation, the results get more involved at higher orders, and the pattern for a closed-form expression for Ln is not obvious. 4.3 Deformed Lagrangian II. The gravity approach In this subsection, we present another approach to compute the deformed classical La- grangian. This approach exploits the gravity actions (3.16), (3.17) and (3.18). We take the TT deformation as an example. Starting from (3.16), we ﬁx the zweibein τµ, eµ in the ﬂat gauge τµ = δt µ and eµ = δx µ described in section 2.10. Then we ﬁnd the saddle points ˜τ ⋆ µ and ˜e⋆ µ and plug back in the action. The deformed action is given by STT[δt µ, δx µ; ˜τ ⋆ µ, ˜e⋆ µ\|φ]. This approach has been applied to the relativistic case in [18]. The deformed actions for the other two deformations can be obtained similarly. The main diﬀerence is that the ﬂat limit of the U(1) gauge ﬁeld is a pure gauge mµ = ∂µθ. In this subsection, we can simply take mµ = 0 by gauge ﬁxing. The hard rod deformation. We consider hard rod deformation ﬁrst. The saddle-point equation obtained from (3.17) is 0 = ˜mµ + λ ˜e ϵµν Pν 0 , δx µ = ˜eµ −λ ˜e ϵµν J ν 0 . (4.24) (4.24) where Pν 0 and J ν 0 are given in (4.6). There is a unique solution to this equation, which is given by ˜m⋆ t = −λ 2 + λφφ† 1 8λ φφ† x + φ†φx 2 4 + λφ†φ + 2 L0 + 2φxφ† x , ˜m⋆ x = iλ 2 + λφφ† φ† xφ −φxφ† , ˜e⋆ t = iλ 2 + λφφ† φ† xφ −φxφ† , ˜e⋆ x = 2 2 + λφφ† . (4.25) (4.25) Plugging these into SHR λ [0, δx µ; ˜m⋆ µ, ˜e⋆ µ\|φ] reproduces precisely the deformed Lagrangian given in (4.15). Notice that the deformation gives a non-trivial Newtonian potential Φ⋆= ˜m⋆ t . – 14 – – 14 – The TT deformation. Now we consider the TT deformation. The saddle-point equa- tion reads δt µ = ˜τµ + λ ˜e ϵµν Pν 0 , δx µ = ˜eµ −λ ˜e ϵµν Eν 0 . (4.26) (4.26) These equations can also be solved explicitly. There are several solutions to this equation; we choose the one that is regular in the λ →0 limit. However, the solution is too involved to be presented here explicitly. It can be found in the ancillary notebook. 4.3 Deformed Lagrangian II. The gravity approach Plugging in STT λ [δt µ, δx µ; ˜τ ⋆ µ, ˜e⋆ µ\|φ], we obtain the following expression for the deformed Lagrangian These equations can also be solved explicitly. There are several solutions to this equation; we choose the one that is regular in the λ →0 limit. However, the solution is too involved to be presented here explicitly. It can be found in the ancillary notebook. Plugging in STT λ [δt µ, δx µ; ˜τ ⋆ µ, ˜e⋆ µ\|φ], we obtain the following expression for the deformed Lagrangian LTT λ = −F1 + 2√F2 4λ(1 −λV), (4.27) (4.27) JHEP04(2021)186 where F1 = iλ φ† tφ−φtφ† +4λV−2, F2 = λ4φφ† φ† tφx−φtφ† x V−λ3 φ† tφx−φtφ† x 2iV φ† xφ+φxφ† +φφ† φ† tφx−φtφ† x +λ2 1 2φtφ† tφφ†−4φxφ† xV−1 4 φ†2 t φ2+φ2 t φ†2 +2i(φ† tφx−φtφ† x) φ† xφ+φxφ† +λ 4φ† xφx+i φ† tφ−φtφ† +1. (4.28) (4.28) As we can see, the result is highly non-trivial. Performing a perturbative expansion in λ to high powers, we can check that the results match what we obtained from the direct approach (4.20). The JE deformation. Finally, we consider the JE deformation. The saddle-point equa- tions read tions read 0 = ˜mµ + λ ˜e ϵµν Eν 0 , δt µ = ˜τµ −λ ˜e ϵµν J ν 0 . (4.29) (4.29) The solution can be found but is again rather involved to be written down. There is a unique solution which is regular in the λ →0 limit. Plugging this solution to SJE λ [0, δt µ; ˜m⋆ µ, ˜τ ⋆ µ\|φ], we obtain the deformed Lagrangian The solution can be found but is again rather involved to be written down. There is a unique solution which is regular in the λ →0 limit. Plugging this solution to SJE λ [0, δt µ; ˜m⋆ µ, ˜τ ⋆ µ\|φ], we obtain the deformed Lagrangian LJE λ = G1 + √G2 2λ2φ†φ , (4.30) (4.30) where where G1 = 2iλ(φ† xφ −φxφ†) −4, (4.31) G2 = −λ4φ2φ†2 φ† tφ + φtφ† −4λ3φφ†(φ† tφ + φtφ†) −λ2 16φφ†V −4[φ†2 x φ2 + φ2 xφ†2 + 2φxφ† xφφ† + 2iφφ†(φ† tφ −φtφ†)] + 16i(φxφ† −φφ† x) + 16. (4.32) (4.31) (4.32) + 16i(φxφ† −φφ† x) + 16. 4.3 Deformed Lagrangian II. The gravity approach (4.32) We can check explicitly that the perturbative expansion of (4.30) match the results from the direct approach. – 15 – – 15 – 5 Dynamical coordinates and gauge ﬁelds 5 In the relativistic case, TT deformation can be seen as a dynamical or ﬁeld-dependent change of coordinates [13, 20, 21]. However, such an interpretation is only valid on-shell (for a more detailed discussion, see [18]), it has several important applications. At the classical level, the dynamical change of coordinates gives yet another way to derive the classical deformed Lagrangian [82] as well as ﬁnding solutions to the deformed equation of motion and analyze the deformed classical symmetries [83]. At the quantum level, it can be used to derive the deformed S-matrix [13], at least in ﬂat spacetime. In this section, we will show that TT deformation of non-relativistic QFTs also has such an interpretation. For the other two bilinear deformations, which involve the current J µ, the interpretation is also interesting. In addition to the change of coordinates, one also needs to make a dynamical change of the U(1) gauge ﬁeld mµ. More explicitly, in ﬂat spacetime, the undeformed gauge ﬁeld is a pure gauge mµ = ∂µθ. Under the bilinear deformations, we have θ 7→Θ where Θ is ﬁeld dependent. Such interpretations ﬁrst appear in the J ¯T and JTa deformations of relativistic QFTs [3, 4].5 In what follows, we will ﬁrst give the proposals of the dynamical change of coordinates and gauge ﬁelds and then apply them to ﬁnd the deformed Lagrangians, which match our previous results. Finally, we apply them to ﬁnd the deformed quantum S-matrix. JHEP04(2021)186 5Here we mean the undeformed theory is relativistic. The deformed theory is, of course, no longer Lorentz invariant. 5.1 The dynamical coordinates Again we consider the saddle-point equation of the gravity action and take the auxiliary ﬁelds in the ﬂat gauge ˜mµ = ∂µθ and ˜eµ = δx µ, which leads to JHEP04(2021)186 mµ = ∂µθ + λ ϵµνPν 0 , eµ = δx µ −λϵµνJ ν 0 , (5.4) (5.4) where Pν 0 and J ν 0 are the ﬂat space currents (4.5) with the modiﬁcation where Pν 0 and J ν 0 are the ﬂat space currents (4.5) with the modiﬁcation ∂µφ 7→Dµφ = ∂µ + i 2∂µθ φ, ∂µφ† 7→Dµφ† = ∂µ −i 2∂µθ φ†. (5.5) (5.5) In the TT deformation, we can simply take θ = 0 by gauge ﬁxing. For the hard rod and JE deformations, we keep the pure gauge term explicitly for later convenience. Now we need to make a proper interpretation of (5.4). We propose that it deﬁnes a change of coordinate (t, x) 7→(t, X) together with a change of gauge θ 7→Θ as follows ∂µΘ = ∂µθ + λ ϵµνPν 0 , ∂µX = δx µ −λϵµνJ ν 0 . (5.6) (5.6) Notice that here only the spatial coordinate is transformed; the temporal direction is left invariant. This is very natural since eµ is related to the spatial direction. Similarly, to make such interpretations we need to check the consistency relations ∂µ∂νΘ = ∂ν∂µΘ and ∂µ∂νT = ∂ν∂µT which follow from the conservation equations of the currents. Notice that here only the spatial coordinate is transformed; the temporal direction is left invariant. This is very natural since eµ is related to the spatial direction. Similarly, to make such interpretations we need to check the consistency relations ∂µ∂νΘ = ∂ν∂µΘ and ∂µ∂νT = ∂ν∂µT which follow from the conservation equations of the currents. The JE deformation. This case is similar to the hard rod deformation. The relevant equations are mµ = ∂µθ + λ ϵµνEν, τµ = δt µ −λϵµνJ ν. (5.7) (5.7) We propose that this corresponds to the coordinate transformation (t, x) 7→(T, x) together with the gauge transformation θ 7→Θ We propose that this corresponds to the coordinate transformation (t, x) 7→(T, x) together with the gauge transformation θ 7→Θ ∂µΘ = ∂µθ + λ ϵµνEν, ∂µT = δt µ −λϵµνJ ν. (5.8) (5.8) 5.1 The dynamical coordinates There are diﬀerent ways to ﬁnd the dynamical coordinates and gauge ﬁelds [3, 13, 18, 20, 22], one of which is provided by the gravity formulation. The TT deformation. Let us ﬁrst discuss TT deformation. From the gravity formu- lation, we derive the saddle-point equations for the auxiliary ﬁelds ˜τµ, ˜eµ (3.21). In the previous section, we ﬁx τµ = δt µ, eµ = δx µ in the ﬂat gauge and solve for ˜τµ, ˜eµ. Alternatively, we could ﬁx ˜τµ = δt µ, ˜eµ = δx µ in (3.21), which leads to the following equations τµ = δt µ + λϵµνPν 0 , eµ = δx µ −λϵµνEν 0 , (5.1) (5.1) where Pν 0 and Eν 0 are the currents in ﬂat space (4.5). These equations are no longer saddle-point equations for ˜τµ and ˜eµ. Instead, following the intuition from the relativistic case, we can interpret them as deﬁning a change of coordinates from (x1, x2) = (t, x) to (X1, X2) = (T, X) by setting τµ = ∂µX1 and eµ = ∂µX2 in (5.1). Therefore, the dynamical change of coordinates (t, x) 7→(T, X) is deﬁned by ∂µT = δt µ + λϵµνPν 0 , ∂µX = δx µ −λϵµνEν 0 . (5.2) (5.2) One important comment is that, to interpret τµ and eµ as derivatives of the new coordinates, they need to satisfy the consistency condition ∂µ∂νXa = ∂ν∂µXa. This is guaranteed by 5Here we mean the undeformed theory is relativistic. The deformed theory is, of course, no longer Lorentz invariant. – 16 – the conservation of the currents. For example, we need to check that ∂µ∂νT = ∂ν∂µT. Using (5.2), we have ∂µ∂νT −∂ν∂µT = ∂µϵναPα 0 −∂νϵµαPα 0 = −ϵµν∂αPα 0 , (5.3) (5.3) which is zero due to the conservation equation ∂αPα 0 = 0. Notice that the conservation is valid only on-shell, namely when the fundamental ﬁelds satisfy equations of motion. This implies the dynamical change of coordinate interpretation is an on-shell statement. The hard rod deformation. Now, we consider the hard rod deformation. 5.2 Deformed Lagrangian III. Dynamical coordinates In this subsection, we derive the deformed Lagrangians using the dynamical coordinates and gauge ﬁelds as a non-trivial check of our proposals (5.2), (5.6), (5.8). The derivation for the TT deformation is similar to the relativistic case. The generalizations to the other two deformations are new results. The calculation involves two steps. In the ﬁrst step, we ﬁnd the quantities ∂µXa and ∂µΘ in terms of fundamental ﬁelds and their derivatives. In the second step, we perform a change of coordinates/gauge ﬁeld of the original Lagrangian to the new coordinates/gauge ﬁeld and plug in the quantities that we found in the ﬁrst step. This leads to the deformed Lagrangian in the new coordinates/gauge ﬁeld. – 17 – The TT deformation. We ﬁrst discuss how to obtain the Jacobian ∂µXa. Notice that the right hand side of (5.2) is given in terms of fundamental ﬁelds and their derivatives (4.5). We rewrite the derivatives of the ﬁelds by the chain rule ∂µφ = ∂Xaφ ∂µXa and ∂µφ† = ∂Xaφ† ∂µXa. Then (5.2) becomes an equation for ∂µXa, which can be solved explicitly in terms of φ, φ† and ∂Xaφ, ∂Xaφ†. The solution is considerably more complicated than the relativistic case and can be found in the ancillary notebook. After obtaining the Jacobian ∂µXa, we plug it into After obtaining the Jacobian ∂µXa, we plug it into LTT λ = 1 det(∂µXa) (L(T, X) + λ ϵµνEµ 0 Pν 0 ) . (5.9) (5.9) JHEP04(2021)186 The determinant det(∂µXa) comes from the change of the integration measure d2x → det(∂µXa)d2X in the action. The quantities in the bracket are written in terms of ∂Xaφ, ∂Xaφ† by the chain rule. After plugging in the explicit forms of the Jacobian, we obtain the deformed Lagrangian (4.27) in the new coordinates. The hard rod deformation. Let us ﬁrst explain how to obtain the quantities ∂µΘ and ∂µX. Writing out the ﬁrst equation of (5.6) explicitly ∂tΘ = iλ 2 Dtφ φ† −φ Dtφ† + λ DxφDxφ† −λ V, (5.10) ∂xΘ = iλ 2 Dxφ φ† −φ Dxφ† , (5.10) where where Dµφ = ∂µφ + i 2∂µΘ φ, Dµφ† = ∂µφ† −i 2∂µΘ φ†. (5.11) (5.11) Plugging into (5.10), we can solve for ∂µΘ in terms of the fundamental ﬁelds and their derivatives. The explicit results can be found in the ancillary ﬁle. 5.2 Deformed Lagrangian III. Dynamical coordinates To rewrite the second equation (5.13) explicitly, we perform the change of coordinate (t, x) 7→(t, X) where the new spacial coordinate X depends on t and x. We have the following chain rule ∂tφ(t, x) = ∂tφ(t, X) + ∂xX ∂Xφ(t, X), ∂xφ(t, x) = ∂xX ∂Xφ(t, X). (5.12) (5.12) The second equation then can be written as ∂xX = 1 + λ 2 φφ†, ∂tX = iλ 2 ∂xX DXφφ† −φDXφ† , (5.13) (5.13) where DXφ = ∂Xφ + i 2∂XΘ φ, DXφ† = ∂Xφ† −i 2∂XΘ φ†. (5.14) (5.14) Here ∂XΘ = ∂xΘ/∂xX. We can solve (5.13) explicitly, which leads to Here ∂XΘ = ∂xΘ/∂xX. We can solve (5.13) explicitly, which leads to ∂tX = iλ 2 ∂Xφ φ† −φ ∂Xφ† , ∂xX = 1 + λ 2 φφ†. (5.15) (5.15) – 18 – After obtaining ∂µΘ, ∂µX in terms of φ, ∂tφ, ∂Xφ and their conjugates, we plug into After obtaining ∂µΘ, ∂µX in terms of φ, ∂tφ, ∂Xφ and their conjugates, we plug into 1 ∂xX (L(Θ, X) + λϵµνJ µ 0 Pν 0 ) , (5.16) (5.16) where ∂xX comes from the change of integration measure dtdx 7→dtdX. In the bracket, we replace the partial derivative by the covariant counterparts (5.11) and perform the change of coordinate from (t, x) to (t, X). Going through these steps, we ﬁnd the deformed Lagrangian (4.15). The JE deformation. This case is similar to the hard rod deformation, and thus we will be brief. The chain rule for the change of coordinate (t, x) 7→(T, x) is now JHEP04(2021)186 ∂tφ(t, x) = ∂tT ∂T φ(T, x), ∂xφ(t, x) = ∂xφ(T, x) + ∂xT ∂T φ(T, x). (5.17) (5.17) After ﬁnding the solution of ∂µΘ, ∂µT, we plug into After ﬁnding the solution of ∂µΘ, ∂µT, we plug into 1 ∂tT (L0(Θ, T) + λϵµνJ µ 0 Eν 0 ) . (5.18) (5.18) This reproduces (4.30). 5.3 Deformed quantum S-matrices As another application for the dynamical coordinates/gauge ﬁeld, we derive the deformed S- matrix in this section. The main steps parallel the derivations for the relativistic case [4, 13], adapted to the non-relativistic settings. The deformed S-matrix for non-relativistic QFTs have been derived using other methods, see [8, 9]. It is shown that the eﬀect of solvable bilinear deformations on the S-matrix is by multiplying a CDD like phase factor. We will conﬁrm these results from the dynamical coordinate point of view. TT deformation. To deﬁne the S-matrix, we need the notion of asymptotic states. Let us consider the asymptotic in-states. In the far past t →−∞, the ﬁelds are free and allow the following mode expansion φin = Z dp √ 2πain(p) e−itωp+ixp−imθ, φ† in = Z dp √ 2πa† in(p) eitωp−ixp+imθ, (5.19) (5.19) where ωp = p2. Several remarks are in order. Firstly, notice that the mode expansion of each ﬁeld involves only one type of ladder operators instead of both. This is due to the non-relativistic nature of the QFTs under consideration. Secondly, we included a background gauge potential θ in the mode expansion, which can in principle be absorbed as a normalization of the ﬁeld. Here we put it explicitly because, under the hard rod and JE deformations, we have the additional transformation θ 7→Θ, which has non-trivial eﬀects. It is, therefore, convenient to include them in the ﬁrst place. The Fourier modes in the expansion of φin satisﬁes the free Schrödinger equation iDtφ = −D2 xφ, Dµφ = (∂µ + im∂µθ)φ, (5.20) (5.20) – 19 – where m can be seen as the mass of the particle. In the previous sections, we have put m = 1/2, here it is more convenient to leave it generic. where m can be seen as the mass of the particle. In the previous sections, we have put m = 1/2, here it is more convenient to leave it generic. Under TT deformation, we are equivalently putting the theory on the new coordi- nates (T, X). Therefore it is more natural to perform the mode expansion in terms of the new coordinates φin = Z dp √ 2πAin(p) e−iωpT+ipX+imθ, φ† in = Z dp √ 2πA† in(p) eiωpT−ipX−imθ. (5.21) (5.21) Notice that for the TT deformation, the gauge ﬁeld is left untouched. 6There is a relative sign between the two terms in the exponent since in our deﬁnition of Eµ, the eigenvalue of E0 is −P i p2 i . 5.3 Deformed quantum S-matrices Comparing the two expansions (5.19) and (5.21), we ﬁnd that the two sets of modes are related by Comparing the two expansions (5.19) and (5.21), we ﬁnd that the two sets of modes are related by JHEP04(2021)186 JHEP04(2021)186 A† in(p) = a† in(p) eiωp∆T−ip∆X, (5.22) (5.22) where ∆T = t−T and ∆X = x−X. From the deﬁnition of the dynamical coordinates (5.2), we have where ∆T = t−T and ∆X = x−X. From the deﬁnition of the dynamical coordinates (5.2), we have ∂µ(∆T) = −λϵµνPν 0 , ∂µ(∆X) = +λϵµνEν 0 . (5.23) (5.23) In the far past t →−∞we can integrate along the spacial direction and obtain In the far past t →−∞we can integrate along the spacial direction and obtain ∆T(x) = const1 + λ Z x −∞ P0(x′)dx′, ∆X(x) = const2 −λ Z x −∞ E0(x′)dx′. (5.24) (5.24) We introduce the notations P<(x) = Z x −∞ P0(x′)dx′, P>(x) = Z ∞ x P0(x′)dx′, (5.25) (5.25) where P < (x) and P>(x) measure the total momentum to the left and right of x. We deﬁne E<(x) and E>(x) similarly. The integration constants in (5.24) can be chosen arbitrarily. For convenience, we follow [4, 13] and choose the constants in a parity symmetric way const1 = −λ 2 Z ∞ −∞ P0(x) dx, const2 = +λ 2 Z ∞ −∞ E0(x) dx. (5.26) (5.26) We then have ∆T(x) = −λ 2 (P>(x) −P<(x)) , ∆X(x) = −λ 2 (E<(x) −E>(x)). (5.27) (5.27) In the inﬁnite past, the spatial ordering of the particles is equivalent to their momentum ordering, which is special to 1+1 dimensional physics. Keeping this in mind, we have6 A† in(pk) = a† in(pk) × exp  −iλ 2   k−1 X j=1 (ejpk −pjek) + N X j=k+1 (ekpj −ejpk)    , (5.28) (5.28) 6There is a relative sign between the two terms in the exponent since in our deﬁnition of Eµ, the eigenvalue of E0 is −P i p2 i . – 20 – – 20 – where ek = ωpk = p2 k is the energy of the k-th particle. As a result, the deformed and undeformed in-states are related by a phase factor where ek = ωpk = p2 k is the energy of the k-th particle. 5.3 Deformed quantum S-matrices As a result, the deformed and undeformed in-states are related by a phase factor \|{pj}in⟩λ = exp  −iλ X j<k (ejpk −pjek)  \|{pj}in⟩0, (5.29) (5.29) which takes the same form as in the relativistic case. A similar analysis can be done for the out-states. Therefore, we conclude under TT deformation, the S-matrix is deformed in the same way as in relativistic QFT STT λ ({pi}, {¯pj}) = e−iλP j<k(ejpk−pjek)e−iλP j<k(¯ej ¯pk−¯pj¯ek) S0 ({pi}, {¯pj}) , (5.30) JHEP04(2021)186 JHEP04(2021)186 (5.30) where {pi} and {¯pj} are the momenta for the in- and out-states. where {pi} and {¯pj} are the momenta for the in- and out-states. The hard rod deformation. Now we consider the hard rod deformation. The mode expansion in the dynamical coordinates and gauge ﬁeld is given by The hard rod deformation. Now we consider the hard rod deformation. The mode expansion in the dynamical coordinates and gauge ﬁeld is given by φin = Z dp √ 2πAin(p)e−iωpt+ipX+imΘ, φ† in = Z dp √ 2πA† in(p)eiωpt−ipX−imΘ. (5.31) (5.31) Notice that the time coordinate is unchanged in this case, but the gauge θ is changed. The two sets of modes are thus related by A† in(p) = a† in(p)e−ip∆X−im∆Θ, (5.32) (5.32) where ∆Θ = θ −Θ. From (5.6), we ﬁnd where ∆Θ = θ −Θ. From (5.6), we ﬁnd where ∆Θ = θ −Θ. From (5.6), we ﬁnd ∂µ(∆X) = +λϵµνJ ν, ∂µ(∆Θ) = −λϵµνPν. (5.33) (5.33) Integrating these equations along the spacial direction in the asymptotic past as before, we obtain ∆X(x) = −λ 2 (M<(x) −M>(x)), ∆Θ(x) = −λ 2 (P>(x) −P<(x)), (5.34) (5.34) where M<(x) and M>(x) are the total mass to the left and right of x, respectively. Similar considerations like before lead to the following deformed S-matrix SHR λ ({pi}, {¯pj}) = e+iλP j<k(mjpk−pjmk)e+iλP j<k( ¯mj ¯pk−¯pj ¯mk) S0 ({pi}, {¯pj}) , (5.3 (5.35) where {mk} and { ¯mk} are the masses of the particles of the in- and out-states. For simple theories with only one type of particle, we have mk = ¯mk = m. Let us comment on an important diﬀerence between the hard rod and TT deformation. In the hard rod case, the temporal direction is not modiﬁed. Therefore the non-locality only occurs in the spatial direction. Alternatively, we can integrate the ﬁrst equation of (5.33) by taking the integration constant to be zero. 5.3 Deformed quantum S-matrices This leads to X(x) = x −λ Z x −∞ J 0(x′)dx′. (5.36) (5.36) – 21 – – 21 – Let us assume for simplicity that there is only one type of particle with mass m and take λ > 0. In this case, we have X(x) = x −(λm) × {number of particles to the left of x}. (5.37) X(x) = x −(λm) × {number of particles to the left of x}. (5.37) X(x) = x −(λm) × {number of particles to the left of x}. (5.37) (5.37) As alluded in the introduction, this new coordinate has an intuitive physical interpreta- tion [8, 9]. Suppose instead of considering a collection of point particles that we consider hard rods of size mλ. The new coordinate X(x) is measuring the free space between the rods to the left of x. The phase factor in the deformed S-matrix precisely takes into account the fact that the ‘particle’ now has a ﬁnite size. Therefore, the hard rod deformation makes the point particles to ﬁnite-sized hard rods, which is the origin of its name. For λ < 0, the interpretation is that the distance between the particles is increased. JHEP04(2021)186 For TT deformation, similar interpretation applies to the change of coordinates in the spacial direction x →X. Namely, particles become hard rods under the deformation. The size of each rod is proportional to the energy of the rod. On the other hand, for TT deformation, the temporal coordinate is also transformed t →T. Therefore the non- locality is also in the time direction. For JE deformation, only the temporal coordinate is transformed. The physical interpretation for the change of coordinates in the temporal direction seems to be more subtle. This is also reﬂected in the fact that the deformed Lagrangians of the TT and JE deformations are much more complicated than the hard rod deformed one. The JE deformation. Finally, we consider the JE deformation. The mode expansion in the new coordinates and background gauge ﬁeld is φ† in = Z dp √ 2πA† ineiωpT−ipx−imΘ. (5.38) φin = Z dp √ 2πAine−iωpT+ipx+imΘ, φ† in = Z dp √ 2πA† ineiωpT−ipx−imΘ. (5.38) (5.38) The two sets of modes are related by A† in(p) = ain(p) eiωp∆T−im∆Θ. (5.39) (5.39) Using (5.8), we have Using (5.8), we have ∂µ(∆T) = λϵµνJ ν, ∂µ(∆Θ) = −λϵµνEν. (5.40) ∂µ(∆T) = λϵµνJ ν, ∂µ(∆Θ) = −λϵµνEν. 6 Conclusions and discussions We have shown that for non-relativistic QFTs, three fundamental solvable bilinear defor- mations can be deﬁned: TT, hard rod, and JE. These deformations can be interpreted as coupling the undeformed QFT to speciﬁc Newton-Cartan geometries. The gravity formula- tions oﬀer us a geometrical perspective for such deformations and provide us with powerful tools to compute important physical quantities. Classically, we computed the deformed Lagrangians in closed forms for the Schrödinger model with a generic potential. Quantum mechanically, we derived the deformed quantum S-matrices for the three deformations. There are many future directions one can pursue based on the current work. One immediate question is the quantization of the Newton-Cartan gravities in this paper. This question is of great conceptual importance and may shed light on the relativistic case as well. Technically quantizing Newton-Cartan gravity can be anticipated to be more tractable because of a preferred time foliation [61, 62]. Furthermore, it gives a new route to reach relativistic quantum gravity by considering relativistic corrections after quantization [84, 85]. One exciting and concrete question in the non-relativistic context is the hard rod deformation of the free boson. On the one hand, it is shown that quantum mechanically, the deformed theory describes a collection of free hard rods [8, 9], interacting only when they touch each other. This quantum mechanical model has been known for a few decades and has been studied extensively (see for example [86–88]). On the other hand, our work gives a rather diﬀerent, although not completely unexpected, formulation of the model — coupling the free boson to a non-relativistic gravity theory. This paper has mainly focused on the classical aspects of the gravity formulation, although we also derived the deformed S-matrix. It is interesting to study the quantum aspects of the gravity theory and make more direct contacts to the quantum hard rod model. JHEP04(2021)186 In the relativistic case, the topological gravity theory that appears is similar to Jackiw- Teitelboim gravity. Recently non-relativistic formulations of Jackiw-Teitelboim (JT) grav- ity have been studied [67, 68]. It would be interesting to explore the connection with the gravity approach further. Through the JT/SYK correspondence, this relates to the non- relativistic regime of the Sachdev-Ye-Kitaev (SYK) model on the boundary. This could also provide a path to quantizing the geometry through the relation to the BF formalism. Another interesting direction is to explore the relations with TT deformations of rel- ativistic theories. 5.3 Deformed quantum S-matrices (5.40) (5.40) Integrating along the spacial direction and choosing the parity symmetric integration con- stants, we ﬁnd in the asymptotic past ∆T(x) = −λ 2 (M<x −M>x), ∆Θ(x) = −λ 2 (E>x −E<x). (5.41) (5.41) It follows that the deformed S-matrix is given by SJE λ ({pi}, {¯pj}) = e−iλP j<k(mjek−mkej)e−iλP j<k( ¯mj¯ek−¯mk¯ej) S0 ({pi}, {¯pj}) . (5.42) SJE λ ({pi}, {¯pj}) = e−iλP j<k(mjek−mkej)e−iλP j<k( ¯mj¯ek−¯mk¯ej) S0 ({pi}, {¯pj}) . (5.42) (5.42) We found that the deformed S-matrices for the three bilinear deformations have similar structures. In fact, this structure holds in more general situations. For integrable theories, one can construct similar solvable bilinear deformations using the higher conserved currents. They lead to similar CDD factors, as has been shown in [9]. We found that the deformed S-matrices for the three bilinear deformations have similar structures. In fact, this structure holds in more general situations. For integrable theories, one can construct similar solvable bilinear deformations using the higher conserved currents. They lead to similar CDD factors, as has been shown in [9]. – 22 – – 22 – 6 Conclusions and discussions It is well-known that a non-relativistic QFT in D dimension can be obtained from relativistic QFTs in at least two ways: The ﬁrst one is by a null reduction of a D + 1-dimensional QFT, which is discussed in more detail in appendix C; the other is by performing a 1/c expansion of a D-dimensional relativistic QFT where c is the speed of light. In the context of solvable deformations, both relations are interesting to explore. From the null reduction perspective, the Schrödinger model we considered in this paper can be obtained by a 3d Klein-Gordon model with a generic potential. The three fundamental currents J µ, Pµ and Eµ are diﬀerent components of the stress-energy tensor of the 3d theory. It would be fascinating to see whether we can ‘uplift’ the deformations we deﬁned in this paper to the 3d theory in some proper sense. This might give a concrete clue for deﬁning a TT like deformation for QFTs in higher dimensions, at least in 3d. – 23 – From the 1/c expansion perspective, we can start with a relativistic QFT, TT deform it, and perform the 1/c expansion. If such a procedure is well-deﬁned, it gives another way to deﬁne a TT deformed theory. As an interesting example, it is known that the Lieb- Liniger model can be obtained from the Sinh-Gordon theory by taking the non-relativistic limit [89, 90]. The TT deformation of the Sinh-Gordon model has been studied in [2, 6]. Therefore, it is interesting to take the 1/c expansion of the deformed Sinh-Gordon model and compare it with what we obtained in the current work. We expect the results to be rather diﬀerent since it is known that the 1/c expansion is typically related to the Type II Newton-Cartan gravity, which is brieﬂy studied in appendix A.1.2. It would be interesting to clarify the details and generalize to other theories. JHEP04(2021)186 Yet another interesting direction is to study the holographic dual of the deformed non-relativistic theories. In the relativistic case, there have been several proposals for the holographic dictionary [21, 91–93]. It would be interesting to see how these proposals are generalized to the non-relativistic cases. We emphasize that the hard rod and JE deformations are new in the non-relativistic contexts, and it would be exciting to see their holographic interpretations. Acknowledgments YJ would like to thank Yang Zhang for initiating the collaboration with JX. The work of DH is supported by the Swiss National Science Foundation through the NCCR SwissMAP. We would like to thank Paolo Ceschin, Riccardo Conti and especially Roberto Tateo for helpful correspondences. 6 Conclusions and discussions To explore the holographic dictionary, it is necessary ﬁrst to study the deformed theories coupled to non-relativistic conformal Type I Newton-Cartan geometry, which has local Schrödinger symmetry [32, 40, 44, 94]. Such theories exhibit a larger symmetry and should be under more analytical control. As a result, more physical quantities can be computed. It would be interesting to compute physical quantities such as the spectrum [1, 2], partition functions [95–100], and correlation functions [22, 101–107] more explicitly in these cases. Finally, it is known that the TT deformation has a deep connection with string theory, see for example [108–116]. It would be interesting to explore similar connections between our three bilinear deformations and the non-relativistic string theories [56–63]. Some of these questions have been investigated recently in [117] for the TT deformation. We believe our results in this paper will be helpful to pursue this direction further. Note added. While ﬁnishing this paper, we became aware of the paper [118], which has some partial overlap with our work. In particular, the dynamical change of coordinates and the deformed classical Lagrangian of the TT deformation are derived independently. A Review of D-dimensional Newton-Cartan geometry In this appendix, we review Newton-Cartan geometry in general D = d + 1 dimensions, of which there are actually two related but distinct types. We will study the signiﬁcance of torsion and non-metricity in the connections, where the situation is fundamentally diﬀerent – 24 – from Lorentzian geometry. Finally, we will study matter theories on these geometries, their currents, and on-shell Ward identities. from Lorentzian geometry. Finally, we will study matter theories on these geometries, their currents, and on-shell Ward identities. 7In section 2 we considered eµ as the fundamental object instead of hµν, but this was only because in D = 2 we have that since hµν is of rank 1, it factorises as hµν = eµeν. D = 2 we have that since hµν is of rank 1, it factorises as hµν = eµeν. 7In section 2 we considered eµ as the fundamental object instead of hµ A.1 Newton-Cartan geometries – 25 – – 25 – Flat spatial indices can be raised and lowered at will with δab and δab, but the same is not true for the zero indices. The vielbeine transforms under local Galilean transformations as Flat spatial indices can be raised and lowered at will with δab and δab, but the same is not true for the zero indices. The vielbeine transforms under local Galilean transformations as δτµ = Lξτµ, (A.11) δea µ = Lξea µ + λabeb µ + λaτµ, (A.12) δvµ = Lξvµ + eµ aλa, (A.13) δeµ b = Lξeµ b + λbaeµ a, (A.14) (A.14) where λab is a local spatial rotation and λa = eµaλµ is a local Galilean boost. We review the Galilei algebra and other non-relativistic algebras in appendix B. The elemental Newton- Cartan geometry and its curvature tensors can also elegantly be obtained as the gauging of the Galilei algebra [70]. JHEP04(2021)186 Depending on what extra ﬁelds there are present, we can deﬁne two distinct types of Newton-Cartan geometry, whose properties we review below.8 9One can also do the more general 1/c expansion, which introduces several new ﬁelds, see [127, 128]. A.1.1 Type I Type I Newton-Cartan geometry is the type we have considered in the main text, where the local frame bundle symmetry is the Bargmann group [25, 69, 121, 122]. We have here that the extra ﬁeld is the U(1) gauge ﬁeld mµ. It is closely related to the Newtonian potential with Φ = −vµmµ being the Newtonian potential. It transforms as δmµ = Lξmµ + λaea µ + ∂µσ, (A.15) (A.15) where σ is an arbitrary function. The geometry can likewise be obtained by gauging the Bargmann algebra [25]. It can also conveniently be embedded in a D + 1 dimensional Lorentzian spacetime with a null Killing vector as we will review in appendix C. It is not the geometry that describes Newtonian gravity in a covariant formulation unless dτ = 0, but it has other applications as advertised in the main text. See [27, 123] for more details. A.1 Newton-Cartan geometries Newton-Cartan geometry essentially consists of the metrics7 τµ, hµν satisfying τµhµν = 0 (A.1) (A.1) and their projective inverses −vµ, hµν [23, 24]. τµ is called the clock-form and gives the local direction of time. The total time elapsed for an observer following her world-line γ between two events A and B is JHEP04(2021)186 ∆tAB = Z γ τ = Z tB tA τµ (x(t)) ˙xµ(t)dt, (A.2) (A.2) which need not be the same as that of another observer with worldline γ′ between the same events because of local time-dilation. On the other hand, hµν is the inverse spatial metric, which is degenerate and of corank 1 because of τµhµν = 0. The metrics satisfy the completeness relations −τνvµ + hµρhρν = δµ ν , (A.3) (A.3) and they transform as and they transform as δτµ = Lξτµ, (A.4) δhµν = Lξhµν + τµλν + τνλµ, (A.5) δvµ = Lξvµ + hµνλν, (A.6) δhµν = Lξhµν, (A.7) (A.7) where λµ satisfying vµλµ = 0 is the Galilean boost parameter and ξµ is a diﬀeomorphism generating vector ﬁeld as in (2.7). τµ, hµν are thus tensorial, while vµ, hµν transform under Galilean boosts, corresponding to ambiguity in deﬁning them as proper inverses. We can deﬁne the (inverse) vielbeine with Galilean frame bundle covariance as EA µ = τµ, ea µ , Eµ A = (−vµ, eµ a) , (A.8) (A.8) where the spatial vielbeine now gets a spatial frame index a, b, . . . = {1, . . . , d = D −1}. Written out in components the completeness relations are where the spatial vielbeine now gets a spatial frame index a, b, . . . = {1, . . . , d = D −1}. Written out in components the completeness relations are τµeµ b = 0, τµvµ = −1, ea µeµ b = δa b , ea µvµ = 0, −τµvν + ea µeν a = δν µ. (A.9) The spatial metrics are related to the vielbeine as hµν := δabeµ aeν b, hµν := δabea µeb ν. (A.10) (A.10) 7In section 2 we considered eµ as the fundamental object instead of hµν, but this was only because in D = 2 we have that since hµν is of rank 1, it factorises as hµν = eµeν. y g g , p [ , ] 9One can also do the more general 1/c expansion, which introduces several new ﬁelds, see [127, 128]. 9One can also do the more general 1/c expansion, which introduces several new ﬁe A.1.2 Type II Type II Newton-Cartan geometry is the geometry that arises in the large speed of light c expansion of general relativity when expanded in 1/c2 in a covariant fashion [27, 124– 126]. In addition to the fundamental Newton-Cartan metrics τµ, vµ, hµν, hµν, which are leading order, there are also two next-to-leading order gauge ﬁelds mµ and ¯Φµν. The mµ here is not a U(1) gauge ﬁeld as it has additional terms proportional to dτ compared to (A.15); it is a so-called torsional-U(1) gauge ﬁeld. The transformation properties of mµ only coincides with its Type I cousin when dτ = 0. The ﬁelds all arise from the large speed of light expansion9 in 1/c2 of any D-dimensional Lorentzian metric, which can be expanded systematically as gµν = −c2τµτν + ¯hµν + c−2 ¯Φµν + O c−4 , gµν = hµν −c−2 ˆvµˆvν + hµρhνσ ¯Φρσ + O c−4 , (A.16) (A.16) 8There is also many other interesting extended geometries, see for example [119, 120]. 9One can also do the more general 1/c expansion, which introduces several new ﬁelds, see [127, 128]. 8There is also many other interesting extended geometries, see for example [119, 120]. 9One can also do the more general 1/c expansion, which introduces several new ﬁelds, see [127, 128]. – 26 – where10 ¯hµν := hµν −2τ(µmν), (A.17) ˆΦ := −vµmµ + 1 2hµνmµmν, (A.18) ˆvµ := vµ −hµνmν, (A.19) ˆvµ := vµ −hµνmν, (A.19) (A.19) all are invariant under local Galilean boosts and satisfy the completeness relation [70, 71] all are invariant under local Galilean boosts and satisfy the completeness relation [70, 71] ˆµ + hµρ¯h δµ (A 20) (A.20) (A.20) −τνˆvµ + hµρ¯hρν = δµ ν . (A.20) these, it is easy to form objects that are guaranteed to be Galilean boost invariant. ith these, it is easy to form objects that are guaranteed to be Galilean boost invariant. JHEP04(2021)186 The Lorentzian metric can, in principle, be expanded to any order. The gauge transfor- mations of the next-to-leading order ﬁelds mµ and ¯Φµν can be traced back to be the result of subleading diﬀeomorphisms. This is the geometry that describes Newtonian gravity in a covariant formulation. When dτ = 0, ¯Φµν decouples and the two geometries coincides, but they are distinct with diﬀerent couplings to matter. Relativistic matter ﬁelds can be expanded in 1/c2 in a similar spirit. A.1.2 Type II One can then also expand the Lagrangian of the matter ﬁelds coupled to Lorentz geometry systematically order-by-order. The higher the order, the more relativistic eﬀects are taken into account. From now on, we focus on Type I Newton-Cartan geometry, although many statements are identical or have closely related equivalents in Type II. We refer the reader to for example [27, 123, 126, 127, 129, 130] for more information and applications. 10These ﬁelds can likewise be deﬁned in Type I Newton-Cartan geometry, where they are also boost invariant, but not U(1) invariant. A.2 Connections and curvatures (A.24) (A.24) As can be seen from (A.2), all observers agree on the time interval between two events independent of their worldline as the closedness of τµ implies H τ = 0. As can be seen from (A.2), all observers agree on the time interval between two events independent of their worldline as the closedness of τµ implies H τ = 0. τ ∧dτ = 0 τ ∧dτ = 0 τ ∧dτ = 0. This is known as twistless torsional Newton-Cartan geometry (TTNC). In general, we have H τ ̸= 0, so observers experience local time dilation. However, there is a foliation of spacetime into spatial hypersurfaces of simultaneity as is guaranteed locally by the Frobenius theorem. We can, without loss of generality, write τ ∧dτ = 0 τ ∧dτ = 0 τ ∧dτ = 0. This is known as twistless torsional Newton-Cartan geometry (TTNC). In general, we have H τ ̸= 0, so observers experience local time dilation. However, there is a foliation of spacetime into spatial hypersurfaces of simultaneity as is guaranteed locally by the Frobenius theorem. We can, without loss of generality, write JHEP04(2021)186 τµ = e−Ψ∂µT, (A.25) (A.25) where Ψ = Ψ(x) is known as the Luttinger potential measuring the local time dilation and T = T(x) is the time-function, which can also be taken as a coordinate. where Ψ = Ψ(x) is known as the Luttinger potential measuring the local time dilation and T = T(x) is the time-function, which can also be taken as a coordinate. τ ∧dτ ̸= 0 τ ∧dτ ̸= 0 τ ∧dτ ̸= 0. This is known as general torsional Newton-Cartan geometry (TNC). It is acausal because locally any two points can be connected with space-like curves, i.e., one with tangent vectors τµ ˙xµ = 0 [131]. For our purposes of coupling a ﬁeld theory to a background geometry, this is the relevant one: Only in this case can we do completely arbitrary variations as τµ is unconstrained. τ ∧dτ ̸= 0 τ ∧dτ ̸= 0 τ ∧dτ ̸= 0. This is known as general torsional Newton-Cartan geometry (TNC). It is acausal because locally any two points can be connected with space-like curves, i.e., one with tangent vectors τµ ˙xµ = 0 [131]. A.2 Connections and curvatures To form diﬀeomorphic invariant actions, we must introduce covariant derivatives ∇µ. It is natural to require the covariant conservation of the metrics through 0 = ∇ρτµ = ∂ρτµ −Γλ ρµτλ, (A.21) 0 = ∇ρhµν = ∂ρhµν + Γµ ρλhλν + Γν ρλhµλ, (A.22) (A.21) (A.22) (A.21) ( ) (A.22) where Γρ µν is the aﬃne connection. Any connection that satisﬁes this is a Newton-Cartan metric-compatible connection [72, 73, 75]. Notice that this does not imply that vµ and hµν are covariantly constant because of the degenerate metric structure. Equation (A.21) implies that the temporal part of torsion of a Newton-Cartan connection is always ﬁxed to be (dτ)µν = 2∂[µτν] = τρΓρ [µν]. (A.23) (A.23) There is no equivalent of the Levi-Civita connection because there is no unique solution to requiring torsionlessness and metricity. Contrary to Lorentzian connections, we can see from (A.21) that requiring a torsionless connection actually puts a constraint on the temporal vielbein. This serves as another indication that both torsion and non-metricity are natural features of Newton-Cartan geometry, a statement that can be made precise by studying Galilean frame bundles [75]. Depending on the properties of (dτ)µν we may subdivide Newton-Cartan geometry into three diﬀerent classes: There is no equivalent of the Levi-Civita connection because there is no unique solution to requiring torsionlessness and metricity. Contrary to Lorentzian connections, we can see from (A.21) that requiring a torsionless connection actually puts a constraint on the temporal vielbein. This serves as another indication that both torsion and non-metricity are natural features of Newton-Cartan geometry, a statement that can be made precise by studying Galilean frame bundles [75]. Depending on the properties of (dτ)µν we may subdivide Newton-Cartan geometry into three diﬀerent classes: – 27 – dτ = 0 dτ = 0 dτ = 0. This is torsionless Newton-Cartan geometry. In this class, there exists a notion of absolute time t as we have (up to topological obstructions) dτ = 0 dτ = 0 dτ = 0. This is torsionless Newton-Cartan geometry. In this class, there exists a notion of absolute time t as we have (up to topological obstructions) dτ = 0 dτ = 0 dτ = 0. This is torsionless Newton-Cartan geometry. In this class, there exists a notion of absolute time t as we have (up to topological obstructions) (dτ)µν = 0 =⇒ τµ = ∂µt. A.2 Connections and curvatures In components, this gives Rµνσρ := −∂µΓρ νσ + ∂νΓρ µσ −Γρ µλΓλ νσ + Γρ νλΓλ µσ. (A.35) The Ricci tensor can also always be deﬁned as The Ricci tensor can also always be deﬁned as Rµν := Rµρνρ, (A.36) (A.36) and as the Newton-Cartan connections are both torsionful and non-metric, the antisym- metric part is in general non-zero. and as the Newton-Cartan connections are both torsionful and non-metric, the antisym- metric part is in general non-zero. A.2 Connections and curvatures For our purposes of coupling a ﬁeld theory to a background geometry, this is the relevant one: Only in this case can we do completely arbitrary variations as τµ is unconstrained. The canonical choice for a connection is ˇΓλ µν := −vλ∂µτν + 1 2hλσ (∂µhνσ + ∂νhµσ −∂σhµν) , (A.26) (A.26) which has both torsion 2ˇΓλ [µν] = −2vλ∂[µτν] (A.27) 2ˇΓλ [µν] = −2vλ∂[µτν] (A.27) τν = ˇ∇µhνρ = 0 we have (A.27) and non-metricity, as besides ˇ∇µτν = ˇ∇µhνρ = 0 we have ˇ∇µvν = 1 2hνρLvhρµ, ˇ∇µhνρ = τ(νLvhρ)µ, (A.28) (A.28) where Lv is the Lie derivative along the ﬂow of vµ. The connection (A.26) is U(1) invariant as it is only built from the vielbeine, but it transforms under local Galilean transforma- tions. This is in contrast with the Levi-Civita connection, which is invariant under local Lorentzian transformations. It is still, in a sense, the closest we get to a “Levi-Civita connection" in NC geometry: It has the minimal torsion allowed as the spatial torsion 2ea λˇΓλ [µν] = 0 is zero. [µ ] Another natural connection to work with is the manifestly boost invariant connection ¯Γλ µν := −ˆvλ∂µτν + 1 2hλσ ∂µ¯hνσ + ∂ν¯hµσ −∂σ¯hµν , (A.29) (A.29) where the boost invariant ﬁelds ˆvλ, ¯hµν are given by (A.17), (A.19). The connection is torsionful with 2¯Γλ [µν] = −2ˆvλ∂[µτν] (A.30) (A.30) – 28 – – 28 – and non-metricity, as besides ¯∇µτν = ¯∇µhνρ = 0 we have and non-metricity, as besides ¯∇µτν = ¯∇µhνρ = 0 we have and non-metricity, as besides ¯∇µτν = ¯∇µhνρ = 0 we have ¯∇ρvµ = 1 2hµλLˆv¯hρλ −hµλ τρ∂λ ˆΦ −ˆΦ(dτ)ρλ , (A.31) ¯∇ρ¯hµν = τ(µLˆv¯hν)ρ + 2ˆΦ(dτ)ρ(µτν) −2τµτν∂ρ ˆΦ −2τρτ(µ∂ν) ˆΦ. (A.32) (A.31) (A.32) The connection depends on mµ linearly through (A.17)–(A.19) and is thus not U(1) in- variant: It is not possible to write down a form where both symmetries are manifest simultaneously using just Type I ﬁelds. One thus needs to work harder to guarantee the Bargmann symmetry of ﬁeld theories; see [71, 76] for more comments. JHEP04(2021)186 Whatever choice of connection one makes, we can deﬁne a Riemann curvature the usual way through the commutator of the covariant derivatives [123]: [∇µ, ∇ν] Xσ = RµνσρXρ −2Γρ [µν]∇ρXσ, (A.33) [∇µ, ∇ν] Xρ = −RµνσρXσ −2Γσ [µν]∇σXρ, (A.34) where Xµ as a vector and Xµ a covector. A.3 Matter currents of ﬁeld theories on Newton-Cartan backgrounds Diﬀerent choices for what set of ﬁelds one varies exist and each choice deﬁnes a diﬀer- ent set of currents as the response to said variations. Using the boost invariant ﬁelds ˆvµ, hµν, ˆΦ one will get a set of currents with all indices down, with manifest boost invari- ance [43, 45, 71, 132]. The canonical set of currents, in the sense that they in ﬂat gauge are the Noether cur- rents and all spacetime indices are raised, is deﬁned as the response to varying τµ, ea µ, mµ: δbgdS [ϕ, τ, e, m] := Z M dDx e Eµδτµ + Pµaδea µ + J µδmµ , (A.37) (A.37) where e := det(τµ, ea µ) is the measure, or equivalently where e := det(τµ, ea µ) is the measure, or equivalently Eµ := e−1 δS δτµ , Pµa := e−1 δS δeaµ , J µ := e−1 δS δmµ , (A.38) (A.38) where here Eµ is the energy current, Pµa the momentum current and J µ the mass current. A Galilean boost relates Pµa and J µ through the on-shell Ward identity Pµaτµ = −J µeµa, (A.39) (A.39) – 29 – – 29 – which shows that Pµa should be thought of as the stress-mass current. In D > 2, we also have a rotational on-shell Ward identity which shows that Pµa should be thought of as the stress-mass current. In D > 2, we also have a rotational on-shell Ward identity 0 = eµ[aPµ b], (A.40) (A.40) telling us that the spatial components of the momentum current are symmetric, also well- known from studying the Noether currents in ﬂat spacetime. Diﬀeomorphism invariance of the Lagrangian guarantees that the currents satisﬁes the following covariant conservation equations: JHEP04(2021)186 = ∂µEµ + e−1∂µe Eµ, 0 = ∂µPµa + e−1∂µe Pµa, 0 = ∂µJ µ + e−1∂µe J µ, (A.41) (A.41) which can be written as a covariant Newton-Cartan derivative using for example the con- nections (A.26) or (A.29) if one wants to make covariance explicit, using the formula e−1∂µ (eXµ) = ∇µXµ + 2Γρ [µρ]Xµ, (A.42) (A.42) for any vector ﬁeld Xµ. for any vector ﬁeld Xµ. B Non-relativistic groups The hallmark of non-relativistic groups is that there is a notion of absolute time in the sense that there is no boost rescaling the time coordinate [49, 69]. This is exactly not the case for the Poincaré group ISO(1, d) = SO(1, d) ⋉R1,d because of its Lie algebra commutator [J0j, Pk] = −δjkH, (B.1) (B.1) here J0j is a Lorentz boost, Pk a spatial momentum and H the Hamiltonian generator. The simplest non-relativistic group is the d dimensional Euclidean group ISO(d) since there is no time direction. Its Lie algebra generators consist of spatial momenta Pi and spatial rotation generators Jij = −Jji. The non-zero commutation relations are [Jij, Pk] = δikPj −δjkPi, (B.2) [Jij, Jkl] = δikJjl −δjkJil −δilJjk + δjlJik. (B.3) (B.2) (B.3) (B.2) (B.3) When including the Hamiltonian H and non-relativistic Galilean boosts generated by Gi, we get the Lie algebra of the Galilei group Gal(d) = SO(d) ⋉Rd ⋉R1,d, which has some additional non-zero commutators given by [H, Gi] = Pi, (B.4) [Jij, Gk] = δikGj −δjkGi. (B.5) (B.4) (B.5) (B.4) (B.5) The Galilei algebra can be obtained by İnönü-Wigner contracting the Poincaré algebra [133]. The Galilei algebra can be obtained by İnönü-Wigner contracting the Poincaré algebra [133]. – 30 – – 30 – The Galilei group’s central extension is non-trivial and of special interest and is called the Bargmann group Barg(d) = SO(d) ⋉Rd ⋉ R1,d ⊗U(1) [28]. Its Lie algebra has, in addition to the above non-zero commutation relations, a new non-zero one given by (B.6) [Pi, Gj] = Nδij, (B.6) where N is a central charge corresponding to particle number or mass. The algebra can be obtained from a null reduction of the Poincaré algebra in D + 1 dimensions as well, which lays the foundations for the next section. where N is a central charge corresponding to particle number or mass. The algebra can be obtained from a null reduction of the Poincaré algebra in D + 1 dimensions as well, which lays the foundations for the next section. There are also conformal extensions like the Galilean conformal algebra [134–136] and the Schrödinger algebra, which has an anisotropic scaling [94, 137, 138]. Both allow for inﬁnite-dimensional Virasoro-like extensions [139]. JHEP04(2021)186 C Null reductions In this appendix, we review how to deﬁne the null reduction of D+1 dimensional Lorentzian geometries with a null Killing vector to obtain a D dimensional Type I Newton-Cartan geometry. We also study how to derive Bargmann invariant matter theories from the null reduction of relativistic ones. Lorentzian spacetimes with a null Killing vector C.1 Lorentzian spacetimes with a null Killing vector The Type I NC ﬁelds ﬁt into the null reduction of D + 1-dimensional Lorentzian metric gMN with M = (µ, u) and N = (ν, u) [39, 140, 141]. Any Lorentzian metric with a null isometry ∂u can be written in adapted coordinates as gMN = gµν gµu guν guu ! = ¯hµν τµ τν 0 ! , (C.1) gMN = gµν gµu guν guu ! = hµν −ˆvµ −ˆvν 2ˆΦ ! , (C.2) (C.1) (C.2) where the Galilean boost invariant objects ¯hµν, ˆvν, ˆΦ are deﬁned in (A.17)–(A.19). Notice that we have the constraint guu = 0, which means that we cannot do variations with respect to guu without leaving the null hypersurface. Thus, we can not obtain the current corresponding to guu in the null reduced theory, but all remaining ones are available to us. We can, of course, also decompose the Lorentzian metric in terms of vielbeine as gMN = η ˆ A ˆBE ˆ A ME ˆB N, (C.3) gMN = η ˆ A ˆBE ˆ A ME ˆB N, (C.3) where the D + 1-dimensional Lorentzian vielbeine with ˆA = (A, u) = (0, a, u) and ˆB = (B, u) = (0, b, u) can be written as where the D + 1-dimensional Lorentzian vielbeine with ˆA = (A, u) = (0, a, u) and ˆB = (B, u) = (0, b, u) can be written as E ˆ A M = τµ eµa −mµ 0 0 1 ! , EM ˆ A = −vµ eµa 0 −mµvµ mµeµa 1 ! , (C.4) (C.4) where the inverse vielbein is solved such that the completeness relation holds. – 31 – These formulae are useful for null reducing relativistic theories on Lorentzian back- grounds to obtain non-relativistic theories on Type I NC backgrounds, as we shall now see an example of. C.2 Klein-Gordon theory We here want to obtain the Schrödinger model on a general Newton-Cartan background by performing a null reduction of the Klein-Gordon model [40, 43]. Our stating point is the complex Klein-Gordon scalar ﬁeld Ψ (t, x, u) with potential V (\|Ψ\|) coupled to Lorentzian geometry as JHEP04(2021)186 JHEP04(2021)186 ˆSKG [Ψ, g] = Z dD+1x√−g −gMN∂MΨ†∂NΨ −V (\|Ψ\|) . (C.5) (C.5) Since we want a Bargmann scalar of mass m, the higher-dimensional ﬁeld must be taken to be of the form Ψ (t, x, u) = e+imuφ (t, x) . (C.6) (C.6) Using the null reduction of the metric (C.2), we can easily perform the null reduction decomposing the metric and taking derivatives of the scalar ﬁeld. The result is Using the null reduction of the metric (C.2), we can easily perform the null reduction decomposing the metric and taking derivatives of the scalar ﬁeld. The result is SSchr = Z dDxe imvµφDµφ† −imvνφ†Dνφ −hµνDµφ†Dνφ −V (\|φ\|) , (C.7) (C.7) where we have deﬁned the U(1)-covariant derivative Dµφ := ∂µφ + immµφ. (C.8) (C.8) Upon setting m = 1/2 and D = 2 one obtains the action (4.1) considered in the main text. In addition to local Galilean symmetry, we have U(1) symmetry under Upon setting m = 1/2 and D = 2 one obtains the action (4.1) considered in the main text. In addition to local Galilean symmetry, we have U(1) symmetry under φ (x) →e−imσ(x)φ (x) (C.9) mµ →mµ + ∂µσ (x) . (C.10) (C.9) (C 10) (C.9) (C.10) The relation to the relativistic Hilbert energy-momentum tensor found by varying the D + 1-dimensional metric gMN T MN Hil := 1 2 √−g δS δgMN , (C.11) (C.11) is Eµ = T µu Hil −T µν Hilmν, (C.12) Pµa = T µν Hileµa, (C.13) J µ = −T µν Hilτν. (C.14) J µ = −T µν Hilτν. (C.14) (C.14) This decomposition holds in general and can also be used to show how the lower-dimensional currents can be used to assembly the relativistic energy-momentum tensor. – 32 – Open Access. This article is distributed under the terms of the Creative Commons Attribution License (CC-BY 4.0), which permits any use, distribution and reproduction in any medium, provided the original author(s) and source are credited. References [1] F.A. Smirnov and A.B. Zamolodchikov, On space of integrable quantum ﬁeld theories, Nucl. Phys. B 915 (2017) 363 [arXiv:1608.05499] [INSPIRE]. [2] A. Cavaglià, S. Negro, I.M. Szécsényi and R. Tateo, T ¯T-deformed 2D Quantum Field Theories, JHEP 10 (2016) 112 [arXiv:1608.05534] [INSPIRE]. JHEP04(2021)186 JHEP04(2021)186 [3] M. Guica, An integrable Lorentz-breaking deformation of two-dimensional CFTs, SciPost Phys. 5 (2018) 048 [arXiv:1710.08415] [INSPIRE]. [4] T. Anous and M. Guica, A general deﬁnition of JTa-deformed QFTs, arXiv:1911.02031 [INSPIRE]. [5] B. Le Floch and M. Mezei, Solving a family of T ¯T-like theories, arXiv:1903.07606 [INSPIRE]. [6] R. Conti, S. Negro and R. Tateo, Conserved currents and T¯Ts irrelevant deformations of 2D integrable ﬁeld theories, JHEP 11 (2019) 120 [arXiv:1904.09141] [INSPIRE]. [7] J. Cardy, TT deformations of non-Lorentz invariant ﬁeld theories, arXiv:1809.07849 [INSPIRE]. [8] J. Cardy and B. Doyon, TT deformations and the width of fundamental particles, arXiv:2010.15733 [INSPIRE]. [9] Y. Jiang, TT-deformed 1d Bose gas, arXiv:2011.00637 [INSPIRE]. [10] T. Bargheer, N. Beisert and F. Loebbert, Long-Range Deformations for Integrable Spin Chains, J. Phys. A 42 (2009) 285205 [arXiv:0902.0956] [INSPIRE]. [11] B. Pozsgay, Y. Jiang and G. Takács, T ¯T-deformation and long range spin chains, JHEP 03 (2020) 092 [arXiv:1911.11118] [INSPIRE]. [12] E. Marchetto, A. Sfondrini and Z. Yang, T ¯T Deformations and Integrable Spin Chains, Phys. Rev. Lett. 124 (2020) 100601 [arXiv:1911.12315] [INSPIRE]. [13] S. Dubovsky, V. Gorbenko and M. Mirbabayi, Asymptotic fragility, near AdS2 holography and TT, JHEP 09 (2017) 136 [arXiv:1706.06604] [INSPIRE]. [14] S. Dubovsky, V. Gorbenko and G. Hernández-Chiﬄet, TT partition function from topological gravity, JHEP 09 (2018) 158 [arXiv:1805.07386] [INSPIRE]. [15] A.J. Tolley, TT deformations, massive gravity and non-critical strings, JHEP 06 (2020) 050 [arXiv:1911.06142] [INSPIRE]. [16] R. Jackiw, Lower Dimensional Gravity, Nucl. Phys. B 252 (1985) 343 [INSPIRE]. [17] C. Teitelboim, Gravitation and Hamiltonian Structure in Two Space-Time Dimensions, Phys. Lett. B 126 (1983) 41 [INSPIRE]. [18] P. Caputa, S. Datta, Y. Jiang and P. Kraus, Geometrizing TT, JHEP 03 (2021) 140 [arXiv:2011.04664] [INSPIRE]. – 33 – [19] J. Cardy, The TT deformation of quantum ﬁeld theory as random geometry, JHEP 10 (2018) 186 [arXiv:1801.06895] [INSPIRE]. [20] R. Conti, S. Negro and R. Tateo, The TT perturbation and its geometric interpretation, JHEP 02 (2019) 085 [arXiv:1809.09593] [INSPIRE]. [21] M. Guica and R. Monten, T ¯T and the mirage of a bulk cutoﬀ, SciPost Phys. 10 (2021) 024 [arXiv:1906.11251] [INSPIRE]. [22] J. References Cardy, T ¯T deformation of correlation functions, JHEP 12 (2019) 160 [arXiv:1907.03394] [INSPIRE]. [23] E. Cartan, Sur les variétés à connexion aﬃne et la théorie de la rélativité généralisée (première partie), Ann. Éc. Norm. Super. 40 (1923) 325. JHEP04(2021)186 JHEP04(2021)186 [24] E. Cartan, Sur les variétés à connexion aﬃne et la théorie de la rélativité généralisée (première partie)(suite), Ann. Éc. Norm. Super. 41 (1924) 1. [25] R. Andringa, E. Bergshoeﬀ, S. Panda and M. de Roo, Newtonian Gravity and the Bargmann Algebra, Class. Quant. Grav. 28 (2011) 105011 [arXiv:1011.1145] [INSPIRE]. [26] M.H. Christensen, J. Hartong, N.A. Obers and B. Rollier, Torsional Newton-Cartan Geometry and Lifshitz Holography, Phys. Rev. D 89 (2014) 061901 [arXiv:1311.4794] [INSPIRE]. [27] D. Hansen, J. Hartong and N.A. Obers, Action Principle for Newtonian Gravity, Phys. Rev. Lett. 122 (2019) 061106 [arXiv:1807.04765] [INSPIRE]. [28] V. Bargmann, On Unitary ray representations of continuous groups, Annals Math. 59 (1954) 1. [29] D.T. Son, Newton-Cartan Geometry and the Quantum Hall Eﬀect, arXiv:1306.0638 [INSPIRE]. [30] D.T. Son and M. Wingate, General coordinate invariance and conformal invariance in nonrelativistic physics: Unitary Fermi gas, Annals Phys. 321 (2006) 197 [cond-mat/0509786] [INSPIRE]. [31] M. Motta, E. Vitali, M. Rossi, D.E. Galli and G. Bertaina, Dynamical structure factor of one-dimensional hard rods, Phys. Rev. A 94 (2016) 043627 [arXiv:1608.07722]. [32] D.T. Son, Toward an AdS/cold atoms correspondence: A Geometric realization of the Schrödinger symmetry, Phys. Rev. D 78 (2008) 046003 [arXiv:0804.3972] [INSPIRE]. [33] M. Taylor, Non-relativistic holography, arXiv:0812.0530 [INSPIRE]. [34] C.P. Herzog, M. Rangamani and S.F. Ross, Heating up Galilean holography, JHEP 11 (2008) 080 [arXiv:0807.1099] [INSPIRE]. [35] K. Balasubramanian and J. McGreevy, Gravity duals for non-relativistic CFTs, Phys. Rev Lett. 101 (2008) 061601 [arXiv:0804.4053] [INSPIRE]. [36] M. Alishahiha, A. Davody and A. Vahedi, On AdS/CFT of Galilean Conformal Field Theories, JHEP 08 (2009) 022 [arXiv:0903.3953] [INSPIRE]. [37] S. Janiszewski and A. Karch, Non-relativistic holography from Hořava gravity, JHEP 02 (2013) 123 [arXiv:1211.0005] [INSPIRE]. [38] T. Griﬃn, P. Hořava and C.M. Melby-Thompson, Lifshitz Gravity for Lifshitz Holography, Phys. Rev. Lett. 110 (2013) 081602 [arXiv:1211.4872] [INSPIRE]. – 34 – [39] M.H. Christensen, J. Hartong, N.A. Obers and B. Rollier, Boundary Stress-Energy Tensor and Newton-Cartan Geometry in Lifshitz Holography, JHEP 01 (2014) 057 [arXiv:1311.6471] [INSPIRE]. [40] J. Hartong, E. Kiritsis and N.A. Obers, Schrödinger Invariance from Lifshitz Isometries in Holography and Field Theory, Phys. Rev. D 92 (2015) 066003 [arXiv:1409.1522] [INSPIRE]. [41] M. Taylor, Lifshitz holography, Class. Quant. [57] A. Bagchi, Tensionless Strings and Galilean Conformal Algebra, JHEP 05 (2013) 141 [arXiv:1303.0291] [INSPIRE]. References Grav. 33 (2016) 033001 [arXiv:1512.03554] [INSPIRE]. [42] T. Harmark, J. Hartong, L. Menculini, N.A. Obers and Z. Yan, Strings with Non-Relativistic Conformal Symmetry and Limits of the AdS/CFT Correspondence, JHEP 11 (2018) 190 [arXiv:1810.05560] [INSPIRE]. JHEP04(2021)186 [43] K. Jensen, On the coupling of Galilean-invariant ﬁeld theories to curved spacetime, SciPost Phys. 5 (2018) 011 [arXiv:1408.6855] [INSPIRE]. [44] J. Hartong, E. Kiritsis and N.A. Obers, Lifshitz space-times for Schrödinger holography, Phys. Lett. B 746 (2015) 318 [arXiv:1409.1519] [INSPIRE]. [45] M. Geracie, D.T. Son, C. Wu and S.-F. Wu, Spacetime Symmetries of the Quantum Hall Eﬀect, Phys. Rev. D 91 (2015) 045030 [arXiv:1407.1252] [INSPIRE]. [46] A. Gromov and A.G. Abanov, Thermal Hall Eﬀect and Geometry with Torsion, Phys. Rev. Lett. 114 (2015) 016802 [arXiv:1407.2908] [INSPIRE]. [47] S. Laurila and J. Nissinen, Torsional Landau levels and geometric anomalies in condensed matter Weyl systems, Phys. Rev. B 102 (2020) 235163 [arXiv:2007.10682] [INSPIRE]. [48] K. Jensen, Aspects of hot Galilean ﬁeld theory, JHEP 04 (2015) 123 [arXiv:1411.7024] [INSPIRE]. [49] M. Geracie, K. Prabhu and M.M. Roberts, Fields and ﬂuids on curved non-relativistic spacetimes, JHEP 08 (2015) 042 [arXiv:1503.02680] [INSPIRE]. [50] J. Hartong, N.A. Obers and M. Sanchioni, Lifshitz Hydrodynamics from Lifshitz Black Branes with Linear Momentum, JHEP 10 (2016) 120 [arXiv:1606.09543] [INSPIRE]. [51] J. de Boer, J. Hartong, N.A. Obers, W. Sybesma and S. Vandoren, Perfect Fluids, SciPost Phys. 5 (2018) 003 [arXiv:1710.04708] [INSPIRE]. [52] J. de Boer, J. Hartong, N.A. Obers, W. Sybesma and S. Vandoren, Hydrodynamic Modes of Homogeneous and Isotropic Fluids, SciPost Phys. 5 (2018) 014 [arXiv:1710.06885] [INSPIRE]. [53] J. Armas, J. Hartong, E. Have, B.F. Nielsen and N.A. Obers, Newton-Cartan Submanifolds and Fluid Membranes, Phys. Rev. E 101 (2020) 062803 [arXiv:1912.01613] [INSPIRE]. [54] J. de Boer, J. Hartong, E. Have, N.A. Obers and W. Sybesma, Non-Boost Invariant Fluid Dynamics, SciPost Phys. 9 (2020) 018 [arXiv:2004.10759] [INSPIRE]. [55] J. Gomis and H. Ooguri, Nonrelativistic closed string theory, J. Math. Phys. 42 (2001) 3127 [hep-th/0009181] [INSPIRE]. [56] R. Andringa, E. Bergshoeﬀ, J. Gomis and M. de Roo, ‘Stringy’ Newton-Cartan Gravity, Class. Quant. Grav. 29 (2012) 235020 [arXiv:1206.5176] [INSPIRE]. [57] A. Bagchi, Tensionless Strings and Galilean Conformal Algebra, JHEP 05 (2013) 141 [arXiv:1303.0291] [INSPIRE]. – 35 – [58] T. Harmark, J. Hartong and N.A. Obers, Nonrelativistic strings and limits of the AdS/CFT correspondence, Phys. Rev. D 96 (2017) 086019 [arXiv:1705.03535] [INSPIRE]. [59] E. Bergshoeﬀ, J. Gomis and Z. References Yan, Nonrelativistic String Theory and T-duality, JHEP 11 (2018) 133 [arXiv:1806.06071] [INSPIRE]. [60] C.D.A. Blair, A worldsheet supersymmetric Newton-Cartan string, JHEP 10 (2019) 266 [arXiv:1908.00074] [INSPIRE]. [61] A.D. Gallegos, U. Gürsoy and N. Zinnato, Torsional Newton Cartan gravity from non-relativistic strings, JHEP 09 (2020) 172 [arXiv:1906.01607] [INSPIRE]. [62] J. Gomis, J. Oh and Z. Yan, Nonrelativistic String Theory in Background Fields, JHEP 10 (2019) 101 [arXiv:1905.07315] [INSPIRE]. JHEP04(2021)186 JHEP04(2021)186 [63] D. Roychowdhury, Nonrelativistic giant magnons from Newton Cartan strings, JHEP 02 (2020) 109 [arXiv:2001.01061] [INSPIRE]. [64] K. Morand and J.-H. Park, Classiﬁcation of non-Riemannian doubled-yet-gauged spacetime, Eur. Phys. J. C 77 (2017) 685 [Erratum ibid. 78 (2018) 901] [arXiv:1707.03713] [INSPIRE]. [65] K. Cho and J.-H. Park, Remarks on the non-Riemannian sector in Double Field Theory, Eur. Phys. J. C 80 (2020) 101 [arXiv:1909.10711] [INSPIRE]. [66] D.S. Berman, C.D.A. Blair and R. Otsuki, Non-Riemannian geometry of M-theory, JHEP 07 (2019) 175 [arXiv:1902.01867] [INSPIRE]. [67] D. Grumiller, J. Hartong, S. Prohazka and J. Salzer, Limits of JT gravity, JHEP 02 (2021) 134 [arXiv:2011.13870] [INSPIRE]. [68] J. Gomis, D. Hidalgo and P. Salgado-ReboLledó, Non-relativistic and Carrollian limits of Jackiw-Teitelboim gravity, arXiv:2011.15053 [INSPIRE]. [69] C. Duval, G. Burdet, H.P. Kunzle and M. Perrin, Bargmann Structures and Newton-cartan Theory, Phys. Rev. D 31 (1985) 1841 [INSPIRE]. [70] J. Hartong and N.A. Obers, Hořava-Lifshitz gravity from dynamical Newton-Cartan geometry, JHEP 07 (2015) 155 [arXiv:1504.07461] [INSPIRE]. [71] J. Hartong, E. Kiritsis and N.A. Obers, Field Theory on Newton-Cartan Backgrounds and Symmetries of the Lifshitz Vacuum, JHEP 08 (2015) 006 [arXiv:1502.00228] [INSPIRE]. [72] X. Bekaert and K. Morand, Connections and dynamical trajectories in generalised Newton-Cartan gravity I. An intrinsic view, J. Math. Phys. 57 (2016) 022507 [arXiv:1412.8212] [INSPIRE]. [73] X. Bekaert and K. Morand, Connections and dynamical trajectories in generalised Newton-Cartan gravity II. An ambient perspective, J. Math. Phys. 59 (2018) 072503 [arXiv:1505.03739] [INSPIRE]. [74] J. Figueroa-O’Farrill, On the intrinsic torsion of spacetime structures, arXiv:2009.01948 [INSPIRE]. [75] D. Hansen, J. Hartong, N.A. Obers and G. Oling, Galilean ﬁrst-order formulation for the non-relativistic expansion of general relativity, arXiv:2012.01518 [INSPIRE]. [76] M. Geracie, K. Prabhu and M.M. Roberts, Covariant eﬀective action for a Galilean invariant quantum Hall system, JHEP 09 (2016) 092 [arXiv:1603.08934] [INSPIRE]. – 36 – [77] G. Festuccia, D. Hansen, J. Hartong and N.A. Obers, Torsional Newton-Cartan Geometry from the Noether Procedure, Phys. Rev. D 94 (2016) 105023 [arXiv:1607.01926] [INSPIRE]. [78] W. Cottrell and A. References Hashimoto, Comments on T ¯T double trace deformations and boundary conditions, Phys. Lett. B 789 (2019) 251 [arXiv:1801.09708] [INSPIRE]. [79] E.A. Mazenc, V. Shyam and R.M. Soni, A T ¯T Deformation for Curved Spacetimes from 3d Gravity, arXiv:1912.09179 [INSPIRE]. [80] J. Aguilera-Damia, V.I. Giraldo-Rivera, E.A. Mazenc, I. Salazar Landea and R.M. Soni, A path integral realization of joint JT, TJ and TT ﬂows, JHEP 07 (2020) 085 [arXiv:1910.06675] [INSPIRE]. JHEP04(2021)186 JHEP04(2021)186 [81] G. Bonelli, N. Doroud and M. Zhu, T ¯T-deformations in closed form, JHEP 06 (2018) 149 [arXiv:1804.10967] [INSPIRE]. [82] E.A. Coleman, J. Aguilera-Damia, D.Z. Freedman and R.M. Soni, TT -deformed actions and (1,1) supersymmetry, JHEP 10 (2019) 080 [arXiv:1906.05439] [INSPIRE]. [83] M. Guica and R. Monten, Inﬁnite pseudo-conformal symmetries of classical T ¯T, J ¯T and JTa - deformed CFTs, arXiv:2011.05445 [INSPIRE]. [84] K. Kuchar, Gravitation, Geometry and Nonrelativistic Quantum Theory, Phys. Rev. D 22 (1980) 1285 [INSPIRE]. [85] C. Duval, G.W. Gibbons and P. Horvathy, Celestial mechanics, conformal structures and gravitational waves, Phys. Rev. D 43 (1991) 3907 [hep-th/0512188] [INSPIRE]. [86] T. Nagamiya, Statistical mechanics of one-dimensional substances i, in Proceedings of the Physico-Mathematical Society of Japan. 3rd Series 22 (1940) 705. [87] B. Sutherland, Quantum many body problem in one-dimension: Ground state, J. Math. Phys. 12 (1971) 246 [INSPIRE]. [88] M. Wadati and G. Kato, One-dimensional hard-core bose gas, Chaos, Solitons & Fractals 14 (2002) 23. [89] M. Kormos, G. Mussardo and A. Trombettoni, Expectation Values in the Lieb-Liniger Bose Gas, Phys. Rev. Lett. 103 (2009) 210404 [arXiv:0909.1336] [INSPIRE]. [90] M. Kormos, G. Mussardo and A. Trombettoni, 1D Lieb-Liniger Bose Gas as Non-Relativistic Limit of the Sinh-Gordon Model, Phys. Rev. A 81 (2010) 043606 [arXiv:0912.3502] [INSPIRE]. [91] L. McGough, M. Mezei and H. Verlinde, Moving the CFT into the bulk with TT, JHEP 04 (2018) 010 [arXiv:1611.03470] [INSPIRE]. [92] P. Kraus, J. Liu and D. Marolf, CutoﬀAdS3 versus the TT deformation, JHEP 07 (2018) 027 [arXiv:1801.02714] [INSPIRE]. [93] S. Hirano and M. Shigemori, Random boundary geometry and gravity dual of TT deformation, JHEP 11 (2020) 108 [arXiv:2003.06300] [INSPIRE]. [94] E.A. Bergshoeﬀ, J. Hartong and J. Rosseel, Torsional Newton-Cartan geometry and the Schrödinger algebra, Class. Quant. Grav. 32 (2015) 135017 [arXiv:1409.5555] [INSPIRE]. [95] S. Datta and Y. Jiang, T ¯T deformed partition functions, JHEP 08 (2018) 106 [arXiv:1806.07426] [INSPIRE]. – 37 – [96] O. Aharony, S. Datta, A. Giveon, Y. Jiang and D. References Kutasov, Modular invariance and uniqueness of T ¯T deformed CFT, JHEP 01 (2019) 086 [arXiv:1808.02492] [INSPIRE]. [97] O. Aharony, S. Datta, A. Giveon, Y. Jiang and D. Kutasov, Modular covariance and uniqueness of J ¯T deformed CFTs, JHEP 01 (2019) 085 [arXiv:1808.08978] [INSPIRE]. [98] A. Hashimoto and D. Kutasov, Strings, symmetric products, T ¯T deformations and Hecke operators, Phys. Lett. B 806 (2020) 135479 [arXiv:1909.11118] [INSPIRE]. [99] A. Hashimoto and D. Kutasov, TT, JT, TJ partition sums from string theory, JHEP 02 (2020) 080 [arXiv:1907.07221] [INSPIRE]. [100] P. Caputa, S. Datta and V. Shyam, Sphere partition functions & cut-oﬀAdS, JHEP 05 (2019) 112 [arXiv:1902.10893] [INSPIRE]. JHEP04(2021)186 JHEP04(2021)186 [101] O. Aharony and T. Vaknin, The TT* deformation at large central charge, JHEP 05 (2018) 166 [arXiv:1803.00100] [INSPIRE]. [102] M. Guica, On correlation functions in J ¯T-deformed CFTs, J. Phys. A 52 (2019) 184003 [arXiv:1902.01434] [INSPIRE]. [103] S. He, J.-R. Sun and Y. Sun, The correlation function of (1,1) and (2,2) supersymmetric theories with T ¯T deformation, JHEP 04 (2020) 100 [arXiv:1912.11461] [INSPIRE]. [104] S. He and Y. Sun, Correlation functions of CFTs on a torus with a TT deformation, Phys. Rev. D 102 (2020) 026023 [arXiv:2004.07486] [INSPIRE]. [105] Y. Li and Y. Zhou, CutoﬀAdS3 versus TT CFT2 in the large central charge sector: correlators of energy-momentum tensor, JHEP 12 (2020) 168 [arXiv:2005.01693] [INSPIRE]. [106] J. Kruthoﬀand O. Parrikar, On the ﬂow of states under TT, arXiv:2006.03054 [INSPIRE [107] S. Hirano, T. Nakajima and M. Shigemori, T ¯T Deformation of Stress-Tensor Correlators from Random Geometry, arXiv:2012.03972 [INSPIRE]. [108] A. Giveon, N. Itzhaki and D. Kutasov, TT and LST, JHEP 07 (2017) 122 [arXiv:1701.05576] [INSPIRE]. [109] M. Baggio and A. Sfondrini, Strings on NS-NS Backgrounds as Integrable Deformations, Phys. Rev. D 98 (2018) 021902 [arXiv:1804.01998] [INSPIRE]. [110] L. Apolo and W. Song, Strings on warped AdS3 via T¯J deformations, JHEP 10 (2018) 165 [arXiv:1806.10127] [INSPIRE]. [111] S. Chakraborty, A. Giveon and D. Kutasov, JT deformed CFT2 and string theory, JHEP 10 (2018) 057 [arXiv:1806.09667] [INSPIRE]. [112] L. Apolo, S. Detournay and W. Song, TsT, T ¯T and black strings, JHEP 06 (2020) 109 [arXiv:1911.12359] [INSPIRE]. [113] A. Sfondrini and S.J. van Tongeren, T ¯T deformations as TsT transformations, Phys. Rev. D 101 (2020) 066022 [arXiv:1908.09299] [INSPIRE]. [114] N. Callebaut, J. Kruthoﬀand H. Verlinde, TT deformed CFT as a non-critical string, JHEP 04 (2020) 084 [arXiv:1910.13578] [INSPIRE]. [115] S. References Frolov, TT Deformation and the Light-Cone Gauge, Proc. Steklov Inst. Math. 309 (2020) 107 [arXiv:1905.07946] [INSPIRE]. – 38 – [116] S. Frolov, TT, eJJ, JT and eJT deformations, J. Phys. A 53 (2020) 025401 [arXiv:1907.12117] [INSPIRE]. [117] C.D.A. Blair, Non-relativistic duality and T ¯T deformations, JHEP 07 (2020) 069 [arXiv:2002.12413] [INSPIRE]. [118] P. Ceschin, R. Conti and R. Tateo, T¯T-deformed Nonlinear Schrödinge arXiv:2012.12760 [INSPIRE]. [119] E.A. Bergshoeﬀ, K.T. Grosvenor, C. Simsek and Z. Yan, An Action for Extended String Newton-Cartan Gravity, JHEP 01 (2019) 178 [arXiv:1810.09387] [INSPIRE]. [120] E. Bergshoeﬀ, J.M. Izquierdo, T. Ortín and L. Romano, Lie Algebra Expansions and Actions for Non-Relativistic Gravity, JHEP 08 (2019) 048 [arXiv:1904.08304] [INSPIRE]. JHEP04(2021)186 JHEP04(2021)186 [121] P. HAVAS, Four-Dimensional Formulations of Newtonian Mechanics and Their Relation to the Special and the General Theory of Relativity, Rev. Mod. Phys. 36 (1964) 938 [INSPIRE]. [122] R. De Pietri, L. Lusanna and M. Pauri, Standard and generalized Newtonian gravities as ‘gauge’ theories of the extended Galilei group. I. The standard theory, Class. Quant. Grav. 12 (1995) 219 [gr-qc/9405046] [INSPIRE]. [123] D. Hansen, J. Hartong and N.A. Obers, Non-Relativistic Gravity and its Coupling to Matter, JHEP 06 (2020) 145 [arXiv:2001.10277] [INSPIRE]. [124] G. Dautcourt, PostNewtonian extension of the Newton-Cartan theory, Class. Quant. Grav. 14 (1997) A109 [gr-qc/9610036] [INSPIRE]. [125] W. Tichy and E.E. Flanagan, Covariant formulation of the post-1-Newtonian approximation to General Relativity, Phys. Rev. D 84 (2011) 044038 [arXiv:1101.0588] [INSPIRE]. [126] D. Van den Bleeken, Torsional Newton-Cartan gravity from the large c expansion of general relativity, Class. Quant. Grav. 34 (2017) 185004 [arXiv:1703.03459] [INSPIRE]. [127] D. Van den Bleeken, Torsional Newton-Cartan gravity and strong gravitational ﬁelds, in 15th Marcel Grossmann Meeting on Recent Developments in Theoretical and Experimental General Relativity, Astrophysics, and Relativistic Field Theories, (2019) [arXiv:1903.10682] [INSPIRE]. [128] M. Ergen, E. Hamamci and D. Van den Bleeken, Oddity in nonrelativistic, strong gravity, Eur. Phys. J. C 80 (2020) 563 [Erratum ibid. 80 (2020) 657] [arXiv:2002.02688] [INSPIRE]. [129] D. Hansen, J. Hartong and N.A. Obers, Non-relativistic expansion of the Einstein-Hilbert Lagrangian, in 15th Marcel Grossmann Meeting on Recent Developments in Theoretical and Experimental General Relativity, Astrophysics, and Relativistic Field Theories, (2019) [arXiv:1905.13723] [INSPIRE]. [130] D. Hansen, J. Hartong and N.A. Obers, Gravity between Newton and Einstein, Int. J. Mod. Phys. D 28 (2019) 1944010 [arXiv:1904.05706] [INSPIRE]. [131] M. Geracie, K. Prabhu and M.M. Roberts, Curved non-relativistic spacetimes, Newtonian gravitation and massive matter, J. Math. Phys. References 56 (2015) 103505 [arXiv:1503.02682] [INSPIRE]. [132] E.A. Bergshoeﬀand J. Rosseel, Three-Dimensional Extended Bargmann Supergravity, Phys. Rev. Lett. 116 (2016) 251601 [arXiv:1604.08042] [INSPIRE]. – 39 – [133] E. Inonu and E.P. Wigner, On the Contraction of groups and their represenations, Proc. Nat. Acad. Sci. 39 (1953) 510 [INSPIRE]. [134] C. Duval and P.A. Horvathy, Non-relativistic conformal symmetries and Newton-Cartan structures, J. Phys. A 42 (2009) 465206 [arXiv:0904.0531] [INSPIRE]. [135] A. Bagchi and R. Gopakumar, Galilean Conformal Algebras and AdS/CFT, JHEP 07 (2009) 037 [arXiv:0902.1385] [INSPIRE]. [136] A. Bagchi, S. Detournay, R. Fareghbal and J. Simón, Holography of 3D Flat Cosmological Horizons, Phys. Rev. Lett. 110 (2013) 141302 [arXiv:1208.4372] [INSPIRE]. [137] U. Niederer, The maximal kinematical invariance group of the free Schrödinger equation., Helv. Phys. Acta 45 (1972) 802 [INSPIRE]. JHEP04(2021)186 [138] M. Henkel, Schrödinger invariance in strongly anisotropic critical systems, J. Statist. Phys. 75 (1994) 1023 [hep-th/9310081] [INSPIRE]. [139] A. Bagchi and I. Mandal, On Representations and Correlation Functions of Galilean Conformal Algebras, Phys. Lett. B 675 (2009) 393 [arXiv:0903.4524] [INSPIRE]. [139] A. Bagchi and I. Mandal, On Representations and Correlation Functions of Galilean Conformal Algebras, Phys. Lett. B 675 (2009) 393 [arXiv:0903.4524] [INSPIRE]. [140] B. Julia and H. Nicolai, Null Killing vector dimensional reduction and Galilean geometrodynamics, Nucl. Phys. B 439 (1995) 291 [hep-th/9412002] [INSPIRE]. [140] B. Julia and H. Nicolai, Null Killing vector dimensional reduction and Galilean geometrodynamics, Nucl. Phys. B 439 (1995) 291 [hep-th/9412002] [INSPIRE]. [141] K. Balasubramanian and K. Narayan, Lifshitz spacetimes from AdS null and cosmological solutions, JHEP 08 (2010) 014 [arXiv:1005.3291] [INSPIRE]. [141] K. Balasubramanian and K. Narayan, Lifshitz spacetimes from AdS null and cosmological solutions, JHEP 08 (2010) 014 [arXiv:1005.3291] [INSPIRE]. – 40 –
https://openalex.org/W2343033229	https://europepmc.org/articles/pmc4847226?pdf=render	English	null	An improved procedure for isolation of high-quality RNA from nematode-infected Arabidopsis roots through laser capture microdissection	Plant methods	2,016	cc-by	7,209	© 2016 Anjam et al. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/ publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Abstract Background: Cyst nematodes are biotrophs that form specialized feeding structures in the roots of host plants, which consist of a syncytial fusion of hypertrophied cells. The formation of syncytium is accompanied by profound transcriptional changes and active metabolism in infected tissues. The challenge in gene expression studies for syncy- tium has always been the isolation of pure syncytial material and subsequent extraction of intact RNA. Root fragments containing syncytium had been used for microarray analyses. However, the inclusion of neighbouring cells dilutes the syncytium-specific mRNA population. Micro-sectioning coupled with laser capture microdissection (LCM) offers an opportunity for the isolation of feeding sites from heterogeneous cell populations. But recovery of intact RNA from syncytium dissected by LCM is complicated due to extended steps of fixation, tissue preparation, embedding and sectioning. Results: In the present study, we have optimized the procedure of sample preparation for LCM to isolate high quality of RNA from cyst nematode induced syncytia in Arabidopsis roots which can be used for transcriptomic studies. We investigated the effect of various sucrose concentrations as cryoprotectant on RNA quality and morphology of syncy- tial sections. We also compared various types of microscopic slides for strong adherence of sections while removing embedding material. Conclusion: The use of optimal sucrose concentrations as cryoprotection plays a key role in RNA stability and mor- phology of sections. Treatment with higher sucrose concentrations minimizes the risk of RNA degradation, whereas longer incubation times help maintaining the morphology of tissue sections. Our method allows isolating high-qual- ity RNA from nematode feeding sites that is suitable for downstream applications such as microarray experiments. Keywords: Syncytium, LCM, Root, Arabidopsis, RNA degradation, Cyst nematode, Nematode Keywords: Syncytium, LCM, Root, Arabidopsis, RNA degradation, Cyst nematode, Nematode endoparasites. The endoparasitic root-knot (Meloidogyne spp.) and cyst nematodes (Globodera spp. and Heterodera spp.) are sedentary parasites of roots and the primary nematode pathogen of a wide range of crops. Infective stage juveniles (J2s) of cyst nematodes invade the host root and migrate intracellularly until they reach the vas- cular cylinder. There, these nematodes select an initial syncytial cell (ISC) and become sedentary. Within 24 h of the ISC selection the cells adjacent to the ISC appear hypertrophied and fused together due to local dissolution Plant Methods Plant Methods Anjam et al. Plant Methods (2016) 12:25 DOI 10.1186/s13007-016-0123-9 Open Access An improved procedure for isolation of high‑quality RNA from nematode‑infected Arabidopsis roots through laser capture microdissection Muhammad Shahzad Anjam1,4, Yvonne Ludwig2, Frank Hochholdinger2, Chisato Miyaura3, Masaki Inada3, Shahid Siddique1 and Florian M. W. Grundler1* © 2016 Anjam et al. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/ publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Correspondence: grundler@uni‑bonn.de 1 INRES ‑ Molecular Phytomedicine, Rheinische Friedrich-Wilhelms- Universitaet Bonn, Karlrobert‑Kreiten‑Straße 13, 53115 Bonn, Germany Full list of author information is available at the end of the article Background Three days after ISC selection, the cells that have been incorporated into the syncytium are enlarged and exhibit the features of a typical syncy- tial cell such as condensed cytoplasm and enrichment of ribosomes, endoplasmic reticulum, mitochondria and plastids. The nuclei have become enlarged, and several smaller vacuoles are present in the cytoplasm. Moreover the outer cell walls thicken to withstand the high osmotic pressure within the syncytium [5]. The syncytium serves as the sole source of nutrients and water for the rest of the nematode’s successive growth and development. A cocktail of secretions, which are synthesised in the oesophageal glands of the nematodes, are responsible for the modulation of the plant’s defence and the develop- mental pathways that lead to the formation of a syncyt- ium [8, 9, 12, 29]. Therefore understanding the function of host genes that are required for syncytium develop- ment is crucial for identifying novel targets for nematode resistance. of the cell walls [32]. Three days after ISC selection, the cells that have been incorporated into the syncytium are enlarged and exhibit the features of a typical syncy- tial cell such as condensed cytoplasm and enrichment of ribosomes, endoplasmic reticulum, mitochondria and plastids. The nuclei have become enlarged, and several smaller vacuoles are present in the cytoplasm. Moreover the outer cell walls thicken to withstand the high osmotic pressure within the syncytium [5]. The syncytium serves as the sole source of nutrients and water for the rest of the nematode’s successive growth and development. A cocktail of secretions, which are synthesised in the oesophageal glands of the nematodes, are responsible for the modulation of the plant’s defence and the develop- mental pathways that lead to the formation of a syncyt- ium [8, 9, 12, 29]. Therefore understanding the function of host genes that are required for syncytium develop- ment is crucial for identifying novel targets for nematode resistance. Compatible interaction is best studied between cyst nematode Heterodera schachtii and the model plant Arabidopsis thaliana. In recent years a lot of work with this particular pathosystem has shown that the develop- ment of the syncytium is accompanied by massive tran- scriptomic, metabolomic and proteomic changes [11, 13, 31]. Nevertheless isolating pure syncytial material and extracting good quality RNA from Arabidopsis roots for downstream application, such as microarray analyses, qPCR and RNA-seq is challenging. Puthoff et al. Background Plant-parasitic nematodes (PPNs) are obligate bio- trophs that cause significant damage to almost every economically important crop [27]. PPNs are classified based on their feeding habits as either ectoparasites or Correspondence: grundler@uni‑bonn.de 1 INRES ‑ Molecular Phytomedicine, Rheinische Friedrich-Wilhelms- Universitaet Bonn, Karlrobert‑Kreiten‑Straße 13, 53115 Bonn, Germany Full list of author information is available at the end of the article Anjam et al. Plant Methods (2016) 12:25 Page 2 of 9 Page 2 of 9 Successful LCM requires good morphological pres- ervations of tissues. Furthermore, isolated nucleic acids (RNA or DNA), proteins or metabolites must be of high quality to ensure their utility for downstream applica- tions. Therefore, tissue preparation is the crucial step towards successful RNA isolation of samples generated by LCM. Since LCM requires micro-sections of 2–20 µm, samples must be fixed and embedded prior to section- ing and mounting them on the slides. During this mul- tistep procedure, RNA integrity remains at high risk of degradation due to the presence of RNases. Samples for immuno-histological analyses are generally fixed in para- formaldehyde and embedded in paraffin to ensure well- preserved tissue morphology. In contrast, ethanol:acetic acid (EAA) solution is preferred for fixation in experi- ments involving RNA extraction because paraformalde- hyde cross links with RNA and thus makes its recovery difficult [14, 17, 18, 30]. After fixation the tissues are embedded either in paraffin at room temperature or in an optimum cutting temperature (OCT) medium at −20 °C (cryosectioning). While OCT-embedded samples are sectioned through a cryomicrotome at −20 °C, the paraffin-embedded samples are sectioned at room tem- perature. Cryosectioning is often preferred over section- ing of paraffin-embedded samples because processing the tissue at a low temperature reduces the activity of RNases and increase yield [1, 23]. Nevertheless keeping the RNA intact during the various steps of the cryosectioning for LCM poses a challenge and entails fixing and processing tissues. For example, plant cells contain big vacuoles for water storage, which may form crystals because of fast freezing of samples during cryosectioning leading to the destruction of cellular structures. Similarly, it is challeng- ing to maintain the histological preservation during cry- osectioning so that the target cells can be distinguished from others cell types. Therefore a number of tissue fixa- tion and embedding procedures have been developed to ensure optimised handling of tissues with different physi- cal rigidity [4]. of the cell walls [32]. Background [25] used the entire root systems of cyst nematode-infected Arabidopsis plants to perform the first transcriptomic studies. However including non-infected cells dilutes the feeding cell specific mRNA expression profile [10, 31]. In a step forward, Szakasits et al. [31] tried to overcome this limitation by collecting the pure cytoplasm of feeding cells through micro-aspiration. But producing a sufficient amount of material for analysis using this technique is laborious. Moreover it works better on larger, well-devel- oped syncytial cells as compared to syncytial tissues dur- ing the early stages of development. An alternative and promising technique for collecting pure cells is laser cap- ture microdissection (LCM). Nakazono et al. [23] were the first to report application of LCM in plant samples by isolating the epidermis and vascular tissue cells of maize plants, which were subse- quently used in microarray studies. Soon after, Kerk et al. [17] presented an RNA isolation technique using the paraffin-embedded sections of a variety of plant tissues. Further the LCM procedure has been applied to isolate syncytial material from soybean roots infected with cyst nematode H. glycines [15, 18, 19]. These authors applied paraffin embedding to process the syncytial samples and showed that a number of genes are differentially regu- lated in these plants upon nematode infection. But the utility of this protocol for nematode-infected Arabidop- sis roots remains questionable as it may lead to poor LCM is becoming a popular tool among cell biologists because it allows isolation of RNA, DNA, proteins and metabolites from a heterogeneous cell population. Cur- rent LCM techniques are based on two principles. In the “touch-free” approach, cells identified under a micro- scope are excised by a laser beam and removed from their tissue context by gravity or by catapulting them away with a photonic pulse. Alternatively, cells are fused to a plastic membrane and are subsequently removed mechanically from the remaining tissue (reviewed in Ludwig and Hochholdinger [22]). Anjam et al. Plant Methods (2016) 12:25 Page 3 of 9 Page 3 of 9 34 % overnight) during cryoprotection; we modified the tissue preparation procedure by reducing the sucrose incubation steps. For this, root segments containing infected tissues were cut and fixed in EAA solution fol- lowed by a direct incubation in 15 % sucrose solution for 3 h (low concentration sucrose-treated samples). These samples were then left overnight in 34 % sucrose solu- tion. Results To assess the potential of a previously described giant cell RNA isolation protocol in Arabidopsis for its utility for syncytial samples, we grew Arabidopsis plants in KNOP medium and infected them with J2 nematodes of H. schachtii. Root segments containing syncytia were cut at 5 days post inoculation (dpi). The infected root segments were subsequently fixed in EAA (3:1) and processed for LCM as described previously [1, 23, 24]. However, RNA was heavily degraded in these samples and was unusable for downstream applications (data not shown). Fi 1 I fl f ti fi ti d b ddi RNA lit Fig. 1 Influence of tissue fixation and embedding on RNA quality. Electropherograms indicating the quality of RNA extracted from Arabidopsis root segments infected with nematodes at 5 dpi. a RNA was extracted from tissues following all steps of fixation, cryoprotec- tion and embedding, however before LCM. Number of root pieces (containing syncytium) used for extraction: 10; RNA concentration: 3407 pg/μL. b Tissue samples after fixation and cryoprotection but before embedding. There is a splice between Ladder and RNA samples; however, both runs are from the same chip and were put together to facilitate visualization. Number of root pieces (contain- ing syncytium) used for extraction: 15; RNA concentration: 4693 pg/ μL. c Control samples with fixation but without cryoprotection or embedding. Number of root pieces (containing syncytium) used for extraction: approx. 25 RNA concentration: 1440 pg/μL. a The experi- ment was performed once. b, c The experiments were repeated three times and data from one representative experiment is provided. S seconds, bp base pair To identify the crucial step in which the RNA was degraded, we hand-dissected root segments contain- ing syncytia at 5 dpi and isolated RNA at three different steps (Table 1). The subsequent quality analysis revealed that the RNA was already degraded before embedding in OCT. Hence sections mounted on the slides already con- tained deteriorated RNA (Fig. 1a, b). Conversely the RNA isolated from syncytial tissues that were fixed but not treated for cryoprotection produced high-quality RNA (Fig. 1c). We concluded that the cryoprotection proce- dure needs to be modified to obtain good quality RNA. Background Alternatively fixed tissues were directly incubated in 34 % for 5 h (high concentration sucrose-treated sam- ples). Subsequently, tissues from low- and high-concen- tration sucrose-treated samples were embedded in OCT, cryosectioned and used for RNA isolation. While the RNA from low-concentration sucrose-treated tissues was quality RNA extraction. LCM was also used to capture giant cells and isolate RNA induced by root-knot nema- todes in Arabidopsis, rice and tomato during different stages of infection [1, 3, 7, 16, 24, 26]. However no proto- col describing the isolation of syncytial cells from Arabi- dopsis have been described so far. In the present study we optimised the LCM procedure for the isolation of syncytial cells from Arabidopsis roots in such a manner that their morphology can be preserved and high-quality RNA can be isolated for downstream applications such as microarray analyses. Modification of cryoprotection steps yielded good quality RNA To investigate whether RNA degradation might have occurred because of multiple incubations in different sucrose concentrations (10 % for 3 h, 15 % for 3 h and Fig. 1 Influence of tissue fixation and embedding on RNA quality. Electropherograms indicating the quality of RNA extracted from Arabidopsis root segments infected with nematodes at 5 dpi. a RNA was extracted from tissues following all steps of fixation, cryoprotec- tion and embedding, however before LCM. Number of root pieces (containing syncytium) used for extraction: 10; RNA concentration: 3407 pg/μL. b Tissue samples after fixation and cryoprotection but before embedding. There is a splice between Ladder and RNA samples; however, both runs are from the same chip and were put together to facilitate visualization. Number of root pieces (contain- ing syncytium) used for extraction: 15; RNA concentration: 4693 pg/ μL. c Control samples with fixation but without cryoprotection or embedding. Number of root pieces (containing syncytium) used for extraction: approx. 25 RNA concentration: 1440 pg/μL. a The experi- ment was performed once. b, c The experiments were repeated three times and data from one representative experiment is provided. S seconds, bp base pair Fig. 1 Influence of tissue fixation and embedding on RNA quality. Electropherograms indicating the quality of RNA extracted from Arabidopsis root segments infected with nematodes at 5 dpi. a RNA was extracted from tissues following all steps of fixation, cryoprotec- tion and embedding, however before LCM. Number of root pieces (containing syncytium) used for extraction: 10; RNA concentration: 3407 pg/μL. b Tissue samples after fixation and cryoprotection but before embedding. There is a splice between Ladder and RNA samples; however, both runs are from the same chip and were put together to facilitate visualization. Number of root pieces (contain- ing syncytium) used for extraction: 15; RNA concentration: 4693 pg/ μL. c Control samples with fixation but without cryoprotection or embedding. Number of root pieces (containing syncytium) used for extraction: approx. 25 RNA concentration: 1440 pg/μL. a The experi- ment was performed once. b, c The experiments were repeated three times and data from one representative experiment is provided. S seconds, bp base pair Fig. 1 Influence of tissue fixation and embedding on RNA quality. Electropherograms indicating the quality of RNA extracted from Arabidopsis root segments infected with nematodes at 5 dpi. a RNA was extracted from tissues following all steps of fixation, cryoprotec- tion and embedding, however before LCM. Modification of cryoprotection steps yielded good quality RNA Number of root pieces (containing syncytium) used for extraction: 10; RNA concentration: 3407 pg/μL. b Tissue samples after fixation and cryoprotection but before embedding. There is a splice between Ladder and RNA samples; however, both runs are from the same chip and were put together to facilitate visualization. Number of root pieces (contain- ing syncytium) used for extraction: 15; RNA concentration: 4693 pg/ μL. c Control samples with fixation but without cryoprotection or embedding. Number of root pieces (containing syncytium) used for extraction: approx. 25 RNA concentration: 1440 pg/μL. a The experi- ment was performed once. b, c The experiments were repeated three times and data from one representative experiment is provided. S seconds, bp base pair Table 1 An overview of influence of tissue fixation and embedding on RNA quality A, B and C refers to three different steps from which RNA was isolated. (A) RNA was extracted from tissues following all steps of fixation, cryoprotection and embedding, however before LCM. (B) Tissue samples after fixation and cryoprotection but before embedding. (C) Control samples with fixation but without cryoprotection or embedding Method Fixation Cryoprotection in sucrose Embedding RNA quality A Yes Yes Yes (OCT) Degraded B Yes Yes No Degraded C Yes No No Good Table 1 An overview of influence of tissue fixation and embedding on RNA quality Table 1 An overview of influence of tissue fixation and embedding on RNA quality 3407 pg/μL. b Tissue samples after fixation and cryoprotection but before embedding. There is a splice between Ladder and RNA samples; however, both runs are from the same chip and were put together to facilitate visualization. Number of root pieces (contain- ing syncytium) used for extraction: 15; RNA concentration: 4693 pg/ μL. c Control samples with fixation but without cryoprotection or embedding. Number of root pieces (containing syncytium) used for extraction: approx. 25 RNA concentration: 1440 pg/μL. a The experi- ment was performed once. b, c The experiments were repeated three times and data from one representative experiment is provided. S seconds, bp base pair Anjam et al. Plant Methods (2016) 12:25 Page 4 of 9 heavily degraded (Fig. 2a), the sections from high-con- centration sucrose-treated tissues showed high-quality RNA (Fig. 2b). Therefore these samples were further pro- cessed (mounting on slides, and removing OCT) to iso- late syncytial sections through LCM. Subsequent RNA isolation showed a slight degradation in quality of RNA; however, both the quantity and quality of RNA was still reasonable to perform downstream applications (Fig. 2c). We concluded that incubating fixed samples directly in a 34 % sucrose solution for 5 h improves the RNA quality considerably. this might be due to shorter and direct incubation in high concentration sucrose, which may lead to sudden loss of tissue pressure. Therefore we increased the vacuum infiltration of samples in 34 % sucrose from 15 to 45 min followed by an overnight incubation at 4 °C. As a con- sequence, tissue morphology was improved to normal (Fig. 3c). The RNA quality of the samples was not affected by increasing the incubation time. Poly‑l‑lysine‑coated glass slides strongly adhere to micro‑sectionsf Different types of glass slides with diverse features are available. In this study, four types of slides were tested to obtain a strong tissue adherence: (a) polyethylene teraphthalate (PET)-membrane-coated slides, (b) poly- ethylene napthalate (PEN)-membrane-coated slides, (c) Superfrost® positively charged slides, and (d) poly- l-lysine-coated glass slides. We mounted 10 µm sec- tions on PET- and PEN-membrane-coated glass slides and washed them with PBS buffer and 70 % ethanol to remove the OCT medium. The complete removal of the OCT medium is necessary for laser cutting. However during washing, all sections were washed away from the slides. We pre-treated the slides with UV light for 30 min to 4 h to enhance adherence, but the majority of the sec- tions was still washed off during the washing step. Next we used Superfrost® positively-charged glass slides, on which tissue retention was greatly improved, while the use of poly-l-lysine-coated glass slides proved robust in holding the tissues during the washing step. Hence, in all subsequent experiments, we used poly-l-lysine-coated glass slides, and the laser capture machine was also cali- brated to glass slides. Incubation in higher sucrose concentrations affected tissue morphologyi Although incubating fixed samples directly in 34 % sucrose for 5 h resulted in improved RNA quality, the tis- sue morphology was poor (Fig. 3a, b). We reasoned that Fig. 2 Quality assessment of RNA isolation from samples after modi- fying the tissue preparation steps. a RNA isolated after direct incuba- tion in 15 % sucrose solution (low sucrose concentration treated samples). Number of root pieces (containing syncytium) used for extraction: 10; RNA Concentration: 2514 pg/μL. b RNA isolated after direct incubation in 34 % sucrose (high sucrose concentration treated samples). Number of root pieces (containing syncytium) used for extraction: 10; RNA concentration: 2253 pg/μL. c RNA isolation from samples B after LCM. Number of syncytium used for cryosection- ing: 30; RNA concentration: 6297 pg/μL. a–c The experiments were repeated three times and data from one representative experiment is provided. S seconds, bp base pair In vitro amplification of cDNA To assess the quality of RNA isolated through our proto- col for downstream applications, cDNA was synthesised from 5-days-old syncytial using NuGEN’s Ovation Pico WTA System according to the manufacturers’ instruc- tions. Subsequent analysis showed that enough high quality cDNA was amplified (Fig. 4a–c). The amplified cDNA was used to perform further steps of hybridization with the GeneChip® Arabidopsis ATH1 Genome (Affy- metrix UK Ltd). The data analysis showed that amplified cDNA was suitable for microarray experiments (data not shown). Whereas the detailed results from microar- ray will be presented elsewhere, we analysed the expres- sion of six different genes (BGLU28, PGIP1, PDF1.4, WRKY76, Xylanase, and ß tubulin4) in 5-days-old syncy- tial samples by RT-PCR (see methods for details). These genes were selected based on intensity of expression in syncytium in our microarray data (data not shown). Under optimized PCR conditions, five out of six genes Fig. 2 Quality assessment of RNA isolation from samples after modi- fying the tissue preparation steps. a RNA isolated after direct incuba- tion in 15 % sucrose solution (low sucrose concentration treated samples). Number of root pieces (containing syncytium) used for extraction: 10; RNA Concentration: 2514 pg/μL. b RNA isolated after direct incubation in 34 % sucrose (high sucrose concentration treated samples). Number of root pieces (containing syncytium) used for extraction: 10; RNA concentration: 2253 pg/μL. c RNA isolation from samples B after LCM. Number of syncytium used for cryosection- ing: 30; RNA concentration: 6297 pg/μL. a–c The experiments were repeated three times and data from one representative experiment is provided. S seconds, bp base pair Anjam et al. Plant Methods (2016) 12:25 Page 5 of 9 Fig. 3 Morphology of longitudinal syncytial samples (10 µm thin)upon different sucrose treatment. a Infected root segments upon direct incuba- tion in 15 % sucrose solution (low sucrose concentration treated samples). b Infected root segments upon direct incubation in 34 % sucrose (high sucrose concentration treated samples). c Infected root segments upon direct overnight incubation in 34 % sucrose. Asterisk () nematode, S syncytium, scale bar 100 µM Fig. 3 Morphology of longitudinal syncytial samples (10 µm thin)upon different sucrose treatment. a Infected root segments upon direct incuba- tion in 15 % sucrose solution (low sucrose concentration treated samples). b Infected root segments upon direct incubation in 34 % sucrose (high sucrose concentration treated samples). c Infected root segments upon direct overnight incubation in 34 % sucrose. Discussion After the successful use of LCM on animal and human tissues, it has become increasingly popular among plant scientists because of its ability to isolate a single cell-type from a heterogeneous population. The collected material can be used for a broad range of downstream molecular studies, for example, qPCR, RNAseq, microarray and proteomics. In contrast to animal cells the presence of a cell wall and large vacuole in plants necessitates modi- fications to preserve cell morphology and macromol- ecules, such as RNA and proteins. As the use of LCM in plant sciences expands, a variety of optimized protocols for specific cell-type harvest have been established. How- ever in contrast to a typical plant cell with a large cen- tral vacuole, the cyst-nematode-induced syncytium is a distinctive structure surrounded by a thick cell wall that contains several smaller vacuoles. Therefore performing LCM on syncytial samples may necessitate certain modi- fications in existing protocols. Although the 34 % sucrose concentration significantly increased RNA quality, the morphology of the sections was strongly compromised. We reasoned that this may be due to direct tissue incubation in a higher concentration sucrose solution (34 %), which may result in a rapid loss of water and thus shrunk tissues. Indeed tissue morphol- ogy was maintained through longer vacuum infiltration and an overnight incubation in a 34 % sucrose solution (see “Methods”). The slow intake of solution might have helped the cells maintain their shape. After the morphology of the sections was satisfying, the embedding medium was removed completely by washing which is a perquisite for the laser beam to func- tion properly. The commercially available OCT medium is water soluble and can be removed easily by washing with PBS buffer. However the root sections can also be washed away unless they adhere strongly to the slides. We tried membrane-covered slides (PET and PEN), which permit very fine dissection of target cells, and the targeted area can be catapulted as a single intact piece, but they proved very poor in holding root sections dur- ing the washing step. In contrast, Ramsay et al., [26] used PEN slides for the LCM of giant cells from tomato roots, and Barcala et al. [1] used Superfrost® positively- charged glass slides for giant cells in Arabidopsis roots. However we found that poly-l-lysine-coated glass slides robustly anchored the frozen sections of 5-day-old syncytia. In vitro amplification of cDNA Asterisk () nematode, S syncytium, scale bar 100 µM Fig. 4 Amplification of cDNA, where a–c represents three replications of syncytial samples processed after LCM. The virtual gel generated by an Agilent 2100 Bioanalyzer is shown in d. nt nucelotide, bp base pair Fig. 4 Amplification of cDNA, where a–c represents three replications of syncytial samples processed after LCM. The virtual gel generated by an Agilent 2100 Bioanalyzer is shown in d. nt nucelotide, bp base pair binding of primers (Fig. 5). Nevertheless, these results confirm that RNA isolated through LCM protocol as described in this manuscript can be used to study gene produced a single band of the expected size from LCM- derived syncytial RNA. However, WRKY76 showed an extra band in syncytial samples suggesting unspecific Anjam et al. Plant Methods (2016) 12:25 Page 6 of 9 Fig. 5 Analysis of gene expression in uninfected leaf tissues and in 5-days-old syncytium by RT-PCR. Negative control is without template The use of an RNase inhibitor, for example, RNase later®, maintains high-quality RNA; however RNase later® severely compromised the morphology of the sections in some reports [2]. expression in syncytium during early stages of nematode infection. Discussion i In this study we used a previously described protocol for LCM on giant cells induced by root-knot nematodes in Arabidopsis as a starting point [1]. However when we incubated the samples in various sucrose concentrations, the RNA in the samples was highly degraded (Fig. 1a). In contrast incubating samples directly in a 34 % sucrose solution preserved the RNA integrity, indicating that RNase activity may not completely stop at lower sucrose concentrations. A higher concentration of sugar mini- mise the availability of water in the tissues, which may reduce enzymatic activities including that of RNase. The cryoprotective use of higher sucrose concentrations (20 or 30 %) has already been reported in animal tissues when performing LCM and immunohistochemistry [6, 20, 21]. In conclusion, we have shown that fixation with EAA and a direct infiltration with 34 % sucrose, prior to embedding in OCT medium generates syncytial sam- ples with well-preserved morphology that are suitable Page 7 of 9 Page 7 of 9 Anjam et al. Plant Methods (2016) 12:25 to distinguish syncytial cells from uninfected surround- ing tissue during LCM. Moreover, high quality RNA was extracted from LCM-processed syncytial samples, which was suitable for downstream applications such as microarrays analysis and RT-PCR. Importantly, we were able to analyse the RNA quality prior to amplifi- cation, which has not been shown previously [24]. Our protocol will help analysing gene expression during early stages of cyst nematode infection, which may help understanding genes and pathways that are important for nematode development. In addition, development of analytical tools such as syncytium-specific promot- ers will be greatly facilitated by application of this methodology. at RT was suddenly placed over the sections to adhere them. The slides with sections were stored in 70 % etha- nol, which was already pre-chilled in the cryomicrotome. For OCT removal and dehydration, the slides were treated twice with 70 % ethanol for 30 s each at RT, 95 % ethanol overnight at −20 °C and 99.8 % etha- nol for 2 min at RT. After air drying the slides at RT for 2 min, the sections were immediately processed for LCM, and the remaining slides were stored in a box con- taining silica gel. The LCM captured syncytium cells were collected in RNAse-free adhesive caps (Cat. No. 415190-9211-001 Zeiss, Germany) using the follow- ing settings for PALM Microbeam laser capture (Zeiss, Germany). Discussion Energy: 55–57, Focus: 79–80, Cutting Pro- gram: “CloseCut+AutoLPC” RNA Extraction, Qual- ity Assessment and Gene expression: 10 µl extraction buffer from the PicoPure® RNA isolation kit (Cat no. KIT0204, ThermoFisher Scientific, Germany) was added to an adhesive cap, incubated at 42 °C for 20 min. The extraction mixture was briefly spun down in a table centrifuge at 8000 rpm and was stored at −80 °C until RNA extraction. The later steps of RNA extraction fol- lowed the instructions provided in the kit by the man- ufacturer. The quality of the isolated RNA was tested with an Agilent 2100 Bioanalyzer using the Agilent RNA 6000 Pico kit (Cat. No. 5067-1513. Agilent, Germany). Cryoprotectionhi Table 2 Primer sequences used in this study Gene Primer Sequence Product size Locus BGLU28 Forward GCTACGACACT GGCAACAAA 501 AT2G44460 Reverse TGTGATTTGTTA CTCGCCATTG PGIP1 Forward CCATTCCAA- GTTCTCTCTCT ACG 221 AT5G06860 Reverse AGCATCACCTT GGAGCTTGT PDF1.4 Forward CTTCCT- TATAGCTT CCACTGAGAT 130 AT1G19610 Reverse AGCACGTTCCCA TCTCTTAC WRKY75 Forward ATGGAGGGATAT- GATAATGGGTC 418 AT5g13080 Reverse GCATTTGAGTG AGAATAT- GCTCG Xylanase Forward CTGTTCTTGGT CGTCCCATT 360 AT1G10050 Reverse CGACAACGA ACGTTTTGAGA Beta tubulin 4 Yes TTTCCGTACCC TCAAGCTCG 160 AT5G44340 GTGAAGCCTTG CGAATGGGA Table 2 Primer sequences used in this study The fixative solution was replaced with a 34 % sucrose solution prepared in 0.01 M PBS buffer (treated with 0.1 % DEPC), pH 7.4. Safranin-O (0.01 % in water; Cat. No. S2255. Sigma-Aldrich, USA) was added to visualise the otherwise transparent tissues. Moreover, to facilitate solution penetration, a vacuum was applied for 45 min at 4 °C. Subsequently, samples were incubated in an undu- lating shaker at 4 °C for 24 h. Tissue preparation and sample collection p p p 10-day-old Arabidopsis thaliana plants were grown on KNOP medium and inoculated with 60–70 juveniles of Heterodera schachtii as described previously [28]. Sub- sequently, 5-day-old infected root samples were hand- dissected and vacuum infiltrated in EAA solution (3:1) for 20 min. Afterwards, EAA solution was replaced with fresh solution and samples were incubated at 70 rpm on an undulating shaker at 4 °C for 1 h. Embeddingh The tissues were horizontally embedded in a tissue freez- ing medium, OCT (Polyfreeze®, Cat No. 25112. Poly Sci- ences, Germany) using Cryomold Tissue-Tek® (Cat No. E62534. EMS, USA). The embedded tissues were frozen by touching the back of cryomolds to liquid nitrogen. The embedded samples (OCT blocks) were stored at −80 °C until cryosectioning. Author details 1 y 14. Inada N, Wildermuth MC. Novel tissue preparation method and cell- specific marker for laser microdissection of Arabidopsis mature leaf. Planta. 2005;221:9–16. 1 INRES ‑ Molecular Phytomedicine, Rheinische Friedrich-Wilhelms-Universitaet Bonn, Karlrobert‑Kreiten‑Straße 13, 53115 Bonn, Germany. 2 INRES ‑ Crop Functional Genomics, Rheinische Friedrich-Wilhelms-Universitaet Bonn, Friedrich‑Ebert‑Allee 144, 53113 Bonn, Germany. 3 Department of Biotechnol- ogy and Life Science, and Global Innovation Research Organization, Tokyo University of Agriculture and Technology, 2‑24‑16 Nakacho, Koganei, Tokyo 184‑8588, Japan. 4 Institute of Molecular Biology and Bio‑technology, Bahaud- din Zakariya University, Multan, Pakistan. 15. Ithal N, Recknor J, Nettleton D, Maier T, Baum TJ, Mitchum MG. Develop- mental transcript profiling of cyst nematode feeding cells in soybean roots. Mol Plant Microbe Interact. 2007;20:510–25. 16. Ji HL, Gheysen G, Denil S, Lindsey K, Topping JF, Nahar K, Haegeman A, De Vos WH, Trooskens G, Van Criekinge W, De Meyer T, Kyndt T. Tran- scriptional analysis through RNA sequencing of giant cells induced by Meloidogyne graminicola in rice roots. J Exp Bot. 2013;64:3885–98. Received: 7 January 2016 Accepted: 19 April 2016 20. Krol J, Busskamp V, Markiewicz I, Stadler MB, Ribi S, Richter J, Duebel J, Bicker S, Fehling HJ, Schubeler D, Oertner TG, Schratt G, Bibel M, Roska B, Filipowicz W. Characterizing light-regulated retinal microRNAs reveals rapid turnover as a common property of neuronal microRNAs. Cell. 2010;141:618–31. 21. Kwon B, Houpt TA. A combined method of laser capture microdissection and X-Gal histology to analyze gene expression in c-Fos-specific neurons. J Neurosci Methods. 2010;186:155–64. Cryosectioning and laser capture microdissectionh The OCT blocks containing tissues were left inside a cryomicrotome (CM 1850 UV, Leica, Germany) pre- cooled to −20 °C for 15 min for acclimatisation before making sections. Longitudinal sections of 10 µm were mounted on poly-l-lysine-coated glass slides (Cat no. J2800AMNZ. Manzel Glaser, Germany). After collecting 10–15 sections on the cryomicrotome platform, a slide Anjam et al. Plant Methods (2016) 12:25 Page 8 of 9 7. He B, Magill C, Starr JL. Laser capture microdissection and real-time PCR for measuring mRNA in giant cells induced by Meloidogyne javanica. J Nematol. 2005;37:308–12. cDNA synthesis was done with NuGEN’s Ovation Pico WTA System (Cat. No. 3302-12. Nugen, USA) accord- ing to the manufacturers’ instructions, starting with 1–50 ng of total RNA. NuGEN’s Encore Biotin Module (Cat. No. 4200-12. Nugen, USA) was used to fragment 3.95 µg cDNA followed by Biotin-labelling according to the manufacturers’ instructions. The gene expression was analysed through RT-PCR. Each sample contained 6 μL Master Mix, 1 μL template cDNA (1:20 dilution), 0.5 μL of forward and 0.5 μL of reverse primers (10 μM) and water in 20 μL of total reaction volume. Primers are given in Table 2. 8. Hewezi T, Baum TJ. Manipulation of plant cells by cyst and root-knot nematode effectors. Mol Plant Microbe Interact. 2013;26:9–16. 9. Hewezi T, Juvale PS, Piya S, Maier TR, Rambani A, Rice JH, Mitchum MG, Davis EL, Hussey RS, Baum TJ. The cyst nematode effector protein 10A07 targets and recruits host posttranslational machinery to mediate its nuclear trafficking and to promote parasitism in Arabidopsis. Plant Cell. 2015;27:891–907. 10. Hofmann J, Hess PH, Szakasits D, Blöchl A, Wieczorek K, Daxböck-Horvarth S, Bohlmann H, van Bel AJE, Grundler FMW. Diversity and activity of sugar transporters in nematode-induced root syncytia. J Exp Bot. 2009;60:3085–95. 11. Hofmann J, El Ashry A, Anwar S, Erban A, Kopka J, Grundler F. Metabolic profiling reveals local and systemic responses of host plants to nematode parasitism. Plant J. 2010;62:1058–71. Acknowledgements 17. Kerk NM, Ceserani T, Tausta SL, Sussex IM, Nelson TM. Laser capture micro- dissection of cells from plant tissues. Plant Physiol. 2003;132:27–35. We thank Jennifer Topping and Keith Lindsey for technical training of Muham- mad Shahzad Anjam. 18. Klink VP, Alkharouf N, MacDonald M, Matthews B. Laser capture microdis- section (LCM) and expression analyses of Glycine max (soybean) syncy- tium containing root regions formed by the plant pathogen Heterodera glycines (soybean cyst nematode). Plant Mol Biol. 2005;59:965–79. Competing interests h h d l h The authors declare that they have no competing interests. 19. Klink VP, Overall CC, Alkharouf NW, MacDonald MH, Matthews BF. Laser capture microdissection (LCM) and comparative microarray expression analysis of syncytial cells isolated from incompatible and compatible soybean (Glycine max) roots infected by the soybean cyst nematode (Heterodera glycines). Planta. 2007;226:1389–409. Authors’ contributions f p 12. Holbein J, Grundler FMW, Siddique S. Plant basal defence to nematodes: an update. J Exp Bot. 2016. doi:10.1093/jxb/erw005. 12. Holbein J, Grundler FMW, Siddique S. Plant basal defence to nematodes an update. J Exp Bot. 2016. doi:10.1093/jxb/erw005. MSA and YL performed experiments. MSA, SS and FMWG designed the research and wrote the paper. FH, CM, and MI provided technical support and helped editing the manuscript. All authors read and approved the final manuscript. 13. Hütten M, Geukes M, Misas-Villamil JC, van der Hoorn RAL, Grundler FMW, Siddique S. Activity profiling reveals changes in the diversity and activity of proteins in Arabidopsis roots in response to nematode infection. Plant Physiol Biochem. 2015;14:36–43. References 1. Barcala M, Garcia A, Cabrera J, Casson S, Lindsey K, Favery B, Garcia- Casado G, Solano R, Fenoll C, Escobar C. Early transcriptomic events in microdissected Arabidopsis nematode-induced giant cells. Plant J. 2010;61:698–712. 22. Ludwig Y, Hochholdinger F. Laser microdissection of plant Cells. In: Plant cell morphogenesis: methods and protocols. Methods Mol Biol. 2014;1080, 249–258. in microdissected Arabidopsis nematode-induced giant cells. Plant J. 2010;61:698–712. 2. Bevilacqua C, Makhzami S, Helbling JC, Defrenaix P, Martin P. Maintaining RNA integrity in a homogeneous population of mammary epithelial cells isolated by laser capture microdissection. BMC Cell Biol. 2010;11:95. 3. Fosu-Nyarko J, Jones MGK, Wang Z. Functional characterization of tran- scripts expressed in early-stage Meloidogyne javanica-induced giant cells isolated by laser microdissection. Mol Plant Pathol. 2009;10:237–48. 4. Gautam V, Sarkar AK. Laser assisted microdissection, an efficient technique to understand tissue specific gene expression patterns and functional genomics in plants. Mol Biotechnol. 2015;57:299–308. 5. Golinowski W, Grundler FMW, Sobczak M. Changes in the structure of Arabidopsis thaliana during female development of the plant-parasitic nematode Heterodera schachtii. Protoplasma. 1996;194:103–16. 6. Haynes T, Luz-Madrigal A, Reis ES, Ruiz NPE, Grajales-Esquivel E, Tzekou A, Tsonis PA, Lambris JD, Del Rio-Tsonis K. Complement anaphylatoxin C3a is a potent inducer of embryonic chick retina regeneration. Nat Commun. 2013;4:2312. 2. Bevilacqua C, Makhzami S, Helbling JC, Defrenaix P, Martin P. Maintaining RNA integrity in a homogeneous population of mammary epithelial cells isolated by laser capture microdissection. BMC Cell Biol. 2010;11:95. 2. Bevilacqua C, Makhzami S, Helbling JC, Defrenaix P, Martin P. Maintaining RNA integrity in a homogeneous population of mammary epithelial cells isolated by laser capture microdissection. BMC Cell Biol. 2010;11:95. 23. Nakazono M, Qiu F, Borsuk LA, Schnable PS. Laser-capture microdissec- tion, a tool for the global analysis of gene expression in specific plant cell types: identification of genes expressed differentially in epidermal cells or vascular tissues of maize. Plant Cell. 2003;15:583–96. 3. Fosu-Nyarko J, Jones MGK, Wang Z. Functional characterization of tran- scripts expressed in early-stage Meloidogyne javanica-induced giant cells isolated by laser microdissection. Mol Plant Pathol. 2009;10:237–48. 24. Portillo M, Lindsey K, Casson S, Garcia-Casado G, Solano R, Fenoll C, Esco- bar C. Isolation of RNA from laser-capture-microdissected giant cells at early differentiation stages suitable for differential transcriptome analysis. Mol Plant Pathol. 2009;10:523–35. 4. Gautam V, Sarkar AK. Laser assisted microdissection, an efficient technique to understand tissue specific gene expression patterns and functional genomics in plants. Funding Muhammad Shahzad Anjam was supported by a grant from German Exchange Service (DAAD). Muhammad Shahzad Anjam was supported by a grant from German Exchange Service (DAAD). Received: 7 January 2016 Accepted: 19 April 2016 Received: 7 January 2016 Accepted: 19 April 2016 Anjam et al. Plant Methods (2016) 12:25 29. Siddique S, Radakovic ZS, De La Torre CM, Chronis D, Novak O, Ramireddy E, Holbein J, Matera C, Hutten M, Gutbrod P, Anjam MS, Rozanska E, Habash S, Elashry A, Sobczak M, Kakimoto T, Strnad M, Schmulling T, Mitchum MG, Grundler FMW. A parasitic nematode releases cytokinin that controls cell division and orchestrates feeding site formation in host plants. Proc Natl Acad Sci USA. 2015;112:12669–74. References Mol Biotechnol. 2015;57:299–308. 25. Puthoff DP, Nettleton D, Rodermel SR, Baum TJ. Arabidopsis gene expres- sion changes during cyst nematode parasitism revealed by statistical analyses of microarray expression profiles. Plant J. 2003;33:911–21. 5. Golinowski W, Grundler FMW, Sobczak M. Changes in the structure of Arabidopsis thaliana during female development of the plant-parasitic nematode Heterodera schachtii. Protoplasma. 1996;194:103–16. 26. Ramsay K, Wang ZH, Jones MGK. Using laser capture microdissection to study gene expression in early stages of giant cells induced by root-knot nematodes. Mol Plant Pathol. 2004;5:587–92. 6. Haynes T, Luz-Madrigal A, Reis ES, Ruiz NPE, Grajales-Esquivel E, Tzekou A, Tsonis PA, Lambris JD, Del Rio-Tsonis K. Complement anaphylatoxin C3a is a potent inducer of embryonic chick retina regeneration. Nat Commun. 2013;4:2312. 27. Sasser JN, Freckman DW. A world perspective on nematology—the role of the society. J Nematol. 1986;18:596. Page 9 of 9 28. Siddique S, Matera C, Radakovic ZS, Hasan MS, Gutbrod P, Rozanska E, Sobczak M, Torres MA, Grundler FMW. Parasitic worms stimulate host NADPH oxidases to produce reactive oxygen species that limit plant cell death and promote infection. Sci Signal. 2014;7:ra33. 30. Srinivasan M, Sedmak D, Jewell S. Effect of fixatives and tissue pro- cessing on the content and integrity of nucleic acids. Am J Pathol. 2002;161:1961–71. 28. Siddique S, Matera C, Radakovic ZS, Hasan MS, Gutbrod P, Rozanska E, Sobczak M, Torres MA, Grundler FMW. Parasitic worms stimulate host NADPH oxidases to produce reactive oxygen species that limit plant cell death and promote infection. Sci Signal. 2014;7:ra33. 29. Siddique S, Radakovic ZS, De La Torre CM, Chronis D, Novak O, Ramireddy E, Holbein J, Matera C, Hutten M, Gutbrod P, Anjam MS, Rozanska E, Habash S, Elashry A, Sobczak M, Kakimoto T, Strnad M, Schmulling T, Mitchum MG, Grundler FMW. A parasitic nematode releases cytokinin that controls cell division and orchestrates feeding site formation in host plants. Proc Natl Acad Sci USA. 2015;112:12669–74. 31. Szakasits D, Heinen P, Wieczorek K, Hofmann J, Wagner F, Kreil DP, Sykacek P, Grundler FMW, Bohlmann H. The transcriptome of syncytia induced by the cyst nematode Heterodera schachtii in Arabidopsis roots. Plant J. 2009;57:771–84. 32. Wyss U, Grundler FMW. Heterodera schachtii and Arabidopsis thaliana, a model host-parasite interaction. Nematologica. 1992;38:488–93. 30. Srinivasan M, Sedmak D, Jewell S. Effect of fixatives and tissue pro- cessing on the content and integrity of nucleic acids. Am J Pathol. 2002;161:1961–71. 31. Szakasits D, Heinen P, Wieczorek K, Hofmann J, Wagner F, Kreil DP, Sykacek P, Grundler FMW, Bohlmann H. The transcriptome of syncytia induced by the cyst nematode Heterodera schachtii in Arabidopsis roots. Plant J. 2009;57:771–84. 32. Wyss U, Grundler FMW. Heterodera schachtii and Arabidopsis thaliana, a model host-parasite interaction. Nematologica. 1992;38:488–93. Submit your next manuscript to BioMed Central and we will help you at every step: • We accept pre-submission inquiries • Our selector tool helps you to ﬁnd the most relevant journal • We provide round the clock customer support • Convenient online submission • Thorough peer review • Inclusion in PubMed and all major indexing services • Maximum visibility for your research Submit your manuscript at www.biomedcentral.com/submit Submit your next manuscript to BioMed Central and we will help you at every step:
https://openalex.org/W4237381919	https://zenodo.org/record/5402593/files/source.pdf	English	null	Notes on the genus Beronium (Acari, Eutrombidiinae) enlightened by new captures of Beronium laemostenis in Spain	Acarologia	2,013	cc-by	2,040	Acarologia is proudly non-profit, with no page charges and free open access Please help us maintain this system by encouraging your institutes to subscribe to the print version of the journal and by sending us your high quality research on the Acari. Subscriptions: Year 2021 (Volume 61): 450 € http://www1.montpellier.inra.fr/CBGP/acarologia/subscribe.php Previous volumes (2010-2020): 250 € / year (4 issues) Acarologia, CBGP, CS 30016, 34988 MONTFERRIER-sur-LEZ Cedex, France ISSN 0044-586X (print), ISSN 2107-7207 (electronic) Acarologia A quarterly journal of acarology, since 1959 Publishing on all aspects of the Acari All information: http://www1.montpellier.inra.fr/CBGP/acarologia/ acarologia-contact@supagro.fr Acarologia is proudly non-profit, with no page charges and free open access Please help us maintain this system by encouraging your institutes to subscribe to the print version of the journal and by sending us your high quality research on the Acari. Subscriptions: Year 2021 (Volume 61): 450 € http://www1.montpellier.inra.fr/CBGP/acarologia/subscribe.php Previous volumes (2010-2020): 250 € / year (4 issues) Acarologia, CBGP, CS 30016, 34988 MONTFERRIER-sur-LEZ Cedex, France ISSN 0044-586X (print), ISSN 2107-7207 (electronic) Acarologia A quarterly journal of acarology, since 1959 Publishing on all aspects of the Acari All information: http://www1.montpellier.inra.fr/CBGP/acarologia/ acarologia-contact@supagro.fr Acarologia is proudly non-profit, with no page charges and free open access Please help us maintain this system by encouraging your institutes to subscribe to the print version of the journal and by sending us your high quality research on the Acari. Subscriptions: Year 2021 (Volume 61): 450 € http://www1.montpellier.inra.fr/CBGP/acarologia/subscribe.php Previous volumes (2010-2020): 250 € / year (4 issues) Acarologia, CBGP, CS 30016, 34988 MONTFERRIER-sur-LEZ Cedex, France ISSN 0044-586X (print), ISSN 2107-7207 (electronic) A quarterly journal of acarology, since 1959 Publishing on all aspects of the Acari All information: http://www1.montpellier.inra.fr/CBGP/acarologia/ acarologia-contact@supagro.fr Acarologia is under free license and distributed under the terms of the Creative Commons-BY. The digitalization of Acarologia papers prior to 2000 was supported by Agropolis Fondation under the reference ID 1500-024 through the « Investissements d’avenir » programme (Labex Agro: ANR-10-LABX-0001-01) Please help us maintain this system by encouraging your institutes to subscribe to the print version of the journal and by sending us your high quality research on the Acari. Subscriptions: Year 2021 (Volume 61): 450 € http://www1.montpellier.inra.fr/CBGP/acarologia/subscribe.php Previous volumes (2010-2020): 250 € / year (4 issues) Acarologia, CBGP, CS 30016, 34988 MONTFERRIER-sur-LEZ Cedex, France ISSN 0044-586X (print), ISSN 2107-7207 (electronic) Acarologia is under free license and distributed under the terms of the Creative Commons-BY. Acarologia 53(4): 425–427 (2013) DOI: 10.1051/acarologia/20132106 Acarologia 53(4): 425–427 (2013) DOI: 10.1051/acarologia/20132106 Acarologia 53(4): 425–427 (2013) DOI: 10.1051/acarologia/20132106 INTRODUCTION of a grant subsidized by the Environmental Protec- tion Agency of Andalusia (Southern Spain) which aims to characterize the arthropod-fauna inhabiting the caves of the south of Spain in order to protect those enigmatic and threatened species. The genus Beronium was erected by Southcott (1986) in order to accommodate the species Hoplothrom- bium coiffaiti (Beron, 1973); this species was captured from the cave Jebel El Ouad in Morocco as a par- asite of the carabid beetle Laemostenus kolbi (Coif- fait, 1972). Two additional species have been de- scribed as belonging to the genus, Beronium veron- icae Haitlinger (1994) from Tenerife (Canary Is- lands) captured parasitizing Licinopsis alternans (De- jean,1828), and Beronium laemostenis Mayoral and Barranco, 2005 described from Cueva Larga cave in Almeria (Spain) also a parasite of a carabid beetle, Laemostenis barrancoi Mateu, 1996. All these species have been described only in their larval stages. NOTES ON THE GENUS BERONIUM (ACARI, EUTROMBIDIINAE) ENLIGHTENED BY NEW CAPTURES OF BERONIUM LAEMOSTENIS IN SPAIN Jaime G. MAYORAL (Received 14 May 2013; accepted 11 July 2013; published online 19 December 2013) Dpto. Biología y Geología. CITE II-B. Universidad de Almería, 04120, Almería, Spain. jgmayoral@hotmail.com ABSTRACT — The genus Beronium encompasses only three species, one described from Morocco, one from the Canary Islands and another from the south-east of the Iberian Peninsula. Each of these species was reported from a single location, their type localities. Captures in six new locations, all caves, are reported for one of the species, Beronium laemostenis Mayoral and Barranco 2005, previously considered endemic to a single cave in the south of Spain. Of the ﬁfty larvae studied, a single teratological case in the second dorsal sclerite is reported and illustrated. Comments on the morphology, host speciﬁcity and distribution in caves for the genus Beronium are provided. KEYWORDS — Beronium laemostenis; Spain; Andalusia; teratology; cave http://www1.montpellier.inra.fr/CBGP/acarologia/ ISSN 0044-586-X (print). ISSN 2107-7207 (electronic) Acarologia 53(4): 425–427 (2013) DOI: 10.1051/acarologia/20132106 MATERIAL STUDIED of Andalusia. The morphometric data and chaeto- taxia are identical to those 14 specimens previously described as been parasites of the carabid L. barran- coi from Cueva Larga (prov. Almeria) (Mayoral and Barranco, 2005). The key characters which separate this species from the other two in the genus are the presence of seven normal setae on tibia I and the presence of one solenidion on genua III (Mayoral and Barranco, 2005). These distinctive characters are conserved in the B. laemostenis specimens from the new locations. In addition to those aforemen- tioned, other characters are also well conserved in all specimens. We have found only one abnormal specimen out of the ﬁfty larvae studied; this ab- normality was present on the second dorsal sclerite. This is normally a single and continuous plate that bears two setae (Fig.1A); in this specimen, the scle- rite has been reduced and it is medially split in two, each part bearing a single seta (Fig.1B); however, its overall chateotaxia and measurements are identical to the other 49 specimens. Cadiz: 5 larvae. Cueva de Berrueco. T.M. Ubrique- Cortes de la Frontera, Cadiz, Spain. 12-IX-2009. GIEX leg. On Carabidae (R296A-E). 1 larva. Cueva de Motilla Paralejo. Complejo Motillas. T.M. Jerez- Cortes de la Frontera, Cadiz, Spain. 8-XI-2009. GIEX leg. (R297). 1 larva. Cueva Berrueco. Ubrique-Cortes de la Frontera, Cadiz, Spain. 7-XI- 2009. GIEX leg. (R298). Jaen: 12 larvae. Sima Curva del Espino. Pozo Romero. T.M. Siles, Jaen, Spain. 7-IX-2009. GEV leg. (R299A-L). 20 larvae. Sima Curva del Espino. Pozo Romero. T.M. Siles, Jaen, Spain. 7-IX-2009. GEV leg. On Laemostenis (Antisphodrus) sp. (R300A-P). 1 larva. Sima de los 30 años. Cerro de Bucentaina. T.M. Siles, Jaen, Spain. 8-IX-2009. GEV leg. (R301). 4 larvae. Sima Curva del Espino. Pozo Romero. T.M. Siles, Jaen, Spain. 16-I-2010. GEV leg. (R302A- D). Almeria: 3 larvae. Cueva del Lago. Sierra Cabr- era, T.M. Turre, Almeria, Spain. 11-XII-2009. ECA leg. On Laemostenis (Antisphodrus) sp. (R303A-C). 3 larvae. Covadura, T.M. Sorbas (Almeria). Spain. Ruiz-Portero, C. leg. 20-III-2002. (R324A-C). Almeria: 3 larvae. Cueva del Lago. Sierra Cabr- era, T.M. Turre, Almeria, Spain. 11-XII-2009. ECA leg. On Laemostenis (Antisphodrus) sp. (R303A-C). 3 larvae. Covadura, T.M. Sorbas (Almeria). Spain. Ruiz-Portero, C. leg. 20-III-2002. (R324A-C). These new captures expand the known distribu- tion of B. MATERIAL AND METHODS All the carabids hosting parasitic mites were cap- tured using pit-fall type traps in different caves in the south of Spain. Parasites were removed in the laboratory under the microscope. Specimens were cleared in 50 % lactic acid and mounted on slides with PVA medium using a Nikon Optiphot-2 com- pound microscope. In the following report, mites parasitizing bee- tles from six caves in three provinces of Andalusia (Spain) are studied. These captures are in the frame 425 http://www1.montpellier.inra.fr/CBGP/acarologia/ ISSN 0044-586-X (print). ISSN 2107-7207 (electronic) Mayoral J.G. FIGURE 1: Beronium laemostenis Mayoral and Barranco 2005, with abnormal second dorsal sclerite. A – normal, B – teratological. Scale bar: 60 µm. FIGURE 1: Beronium laemostenis Mayoral and Barranco 2005, with abnormal second dorsal sclerite. A – normal, B – teratological. Scale bar: 60 µm. MATERIAL STUDIED laemostenis from a unique cave in Almeria (the type locality, Cueva Larga) to a total of seven lo- cations; two additional caves in Almeria, two new locations in Jaen, and also two new locations in Cadiz. The new data shows a wide distribution of B. laemostenis throughout Andalusia. Previously, May- oral and Barranco (2005) pointed out the absence of eyes in this species, which is a character typically REFERENCES The three species of the genus show a high host-speciﬁcity. The specimens of Beronium spp. found parasiizing a host were captured on beetles of the family Carabidae, subfamily Platyninae Bonelli, 1810, tribe Sphodrini Laporte, 1834, subtribe Spho- drina Laporte, 1834. Beronium coiffaiti and B. lae- mostenis were found on members of the genus Lae- mostenus (but in a different subgenera, Ceuthosthenes Antoine, 1957 and Antisphodrus Schaufuss, 1865), and B. veronicae was parasitizing a member of the genus Licinopsis Bedel, 1899. The distribution of the hosts Laemostenus barrancoi and Laemostenus kolbi is also singular and restricted to caves; L. barran- coi is known from its type locality, Cueva Larga, and two nearby caves BC-4 and Sima de Pedro Navarro. L. kolbi was known from a single cave in Morocco, but recently reported from two other Mo- roccan caves (Casale and Comas, 2012). Unfortu- nately, no parasites were found on these new spec- imens (Dr. Casale, personal communication). The restricted distribution of these two species of Lae- mostenus suggests that they could have searched for refuge in caves; and if this is the case, it is likely that the host and the parasite arrived simultaneously at the caves. Interestingly, the host of B. laemostenis in "Cueva del Lago", Almeria (one of the new lo- cations reported here for the species) represents a new species of the genus Laemostenis (Antisphodrus) and it is currently being described (Drs. Ortuño and Barranco, personal communication). In the Canary Islands, L. alternans, host of B. veronicae, was not Beron P. 1973 — Une nouvelle larve d’Acarien (Ho- plothrombium coiffaiti sp. n., Trombidiidae), parasite d’un coléoptère cavernicole du Maroc — Annales de Speleology, 28: 413-416. Casale A., Comas J. 2012 — New or little known Lae- mostenus species from southern Spain and Morocco (Coleoptera: Carabidae: Sphodrini) — Heteropterus, Revista de Entomologia, 12: 173-182. Haitlinger R. 1994 — Four new larval mites (Acari: Trom- bidiidae: Eutrombidiidae) ectoparasitic on carabids (Insecta: Coleoptera: Carabidae) — Revista Chilena de Entomologia, 21: 47-56. Haitlinger R. 1997 — A new species of Hexathrombium Cooreman, 1944 based on parasitic larva on Erotylids from Brazil (Acari: Eutrombidiidae) — Memórias do Instituto Oswaldo Cruz, 92: 81-83. doi:10.1590/S0074- 02761997000100015 Mayoral J.G., Barranco P. 2005 — A new larval mite Bero- nium laemostenis sp. nov (Acari: Eutrombidiidae host on a troglobius beetle from Spain — Biología, 60: 113- 119. Southcott R.V. RESULTS AND DISCUSSION Fifty larval mites of Beronium laemostenis were found parasitizing Carabidae or free-living in caves 426 Acarologia 53(4): 425–427 (2013) found in troglobite species. found in troglobite species. found in caves and as it was highlighted above, fur- ther data is needed to understand the morphology and life history of this scarce genus of parasites. Beronium coiffaiti is another species solely found in caves. Southcott (1986) stated about this species, "from its lack of eyes one may deduce that it is prob- ably a well-adapted cavernicolous species". In fact, all of the three members of the genus lack eyes. Anophthalmia is a character found in cave-dwelling arthropods (troglobites) as an adaptation to low- light habitats. However, it is unclear whether this is the case for this genus of mites since the third species B. veronicae was captured outside of caves, despite the fact that Haitlinger (2004) described this species with "eyes absent". The description of further hypogeal and epigeal species in the genus will clarify whether the presence of eyes is either a generic character or indicative of an adaptation to live in low-light environments. REFERENCES 1986 — Studies on the taxonomy and bi- ology of the subfamily Trombidiinae (Acarina: Trom- bidiinae) with a critical revision of the genera — Aus- tralian Journal of Zoology: 123: 1-116. ACKNOWLEDGEMENTS I thank Dr. Pablo Barranco and Solomon Osei-Amo for their comments on the ﬁnal version of the MS. I would like to thank the three speleology clubs that sampled the caves: ECA (from Almeria), GEV (from Jaen) and GIEX (Cadiz). Also, I thank the institu- tions that provided funding to do the ﬁeld work: FEADER (EEC) and "Consejeria de Medio Ambi- ente de la Junta de Andalucia" (Environmental Pro- tection Agency in Andalusia). COPYRIGHT Mayoral J.G. Acarologia is under free license. This open-access article is distributed under the terms of the Creative Commons-BY-NC-ND which permits unre- stricted non-commercial use, distribution, and reproduc- tion in any medium, provided the original author and source are credited. 427
https://openalex.org/W1628252009	https://europepmc.org/articles/pmc2977806?pdf=render	Sinhala, Sinhalese	null	(<i>E</i>)-3-[4-(Dodecyloxy)phenyl]-1-(2-hydroxyphenyl)prop-2-en-1-one	Acta crystallographica. Section E	2,009	cc-by	6,704	‡ Thomson Reuters ResearcherID: A-5599-2009. § Additional correspondence author, e-mail: hkfun@usm.my; Thomson Reuters ResearcherID: A-3561-2009. organic compounds organic compounds Experimental Crystal data C27H36O3 Mr = 408.56 Triclinic, P1 a = 7.4953 (6) A˚ b = 13.4714 (9) A˚ c = 23.7874 (18) A˚ = 75.116 (4) = 83.876 (5) = 84.669 (5) V = 2302.7 (3) A˚ 3 Z = 4 Mo K radiation = 0.08 mm1 T = 100 K 0.55 0.13 0.06 mm Data collection Bruker APEXII diffractometer Absorption correction: multi-scan (SADABS; Bruker, 2005) Tmin = 0.960, Tmax = 0.996 8571 measured reﬂections 8571 independent reﬂections 4737 reﬂections with I > 2(I) Reﬁnement R[F 2 > 2(F 2)] = 0.069 wR(F 2) = 0.206 S = 1.03 8571 reﬂections 546 parameters H-atom parameters constrained max = 0.29 e A˚ 3 min = 0.30 e A˚ 3 Table 1 Hydrogen-bond geometry (A˚ , ). D—H A D—H H A D A D—H A O1A—H1A O2A 0.82 1.79 2.513 (4) 146 O1B—H1B O2B 0.82 1.81 2.530 (4) 146 C22B—H22C Cg1i 0.97 2.77 3.654 (4) 151 C17B—H17D Cg2 0.97 2.82 3.612 (4) 139 C22A—H22B Cg3 0.97 2.93 3.765 (4) 145 Symmetry code: (i) x 1; y; z. Cg1, Cg2 and Cg3 are the centroids of the C1A--C6A, C10A–C15A and C1B–C6B rings, respectively. Data collection: APEX2 (Bruker, 2005); cell reﬁnement: SAINT organic compounds Acta Crystallographica Section E Structure Reports Online ISSN 1600-5368 Acta Crystallographica Section E Structure Reports Online ISSN 1600-5368 ISSN 1600-5368 Ibrahim Abdul Razak,a‡ Hoong-Kun Fun,a§ Zainab Ngaini,b Siti Muhaini Haris Fadzillahb and Hasnain Hussainc aX-ray Crystallography Unit, School of Physics, Universiti Sains Malaysia, 11800 USM, Penang, Malaysia, bDepartment of Chemistry, Faculty of Resource Science and Technology, Universiti Malaysia Sarawak, 94300 Kota Samarahan, Sarawak, Malaysia, and cDepartment of Molecular Biology, Faculty of Resource Science and Technology, Universiti Malaysia Sarawak, 94300 Kota Samarahan, Sarawak, Malaysia Correspondence e-mail: arazaki@usm.my Received 1 April 2009; accepted 22 April 2009 Received 1 April 2009; accepted 22 April 2009 Key indicators: single-crystal X-ray study; T = 100 K; mean (C–C) = 0.005 A˚; R factor = 0.069; wR factor = 0.206; data-to-parameter ratio = 15.7. In the title compound, C27H36O3, the asymmetric unit consists of two crystallographically independent molecules. The aromatic rings form dihedral angles of 17.1 (2) and 17.6 (2) in the two molecules. In both molecules, the enone groups adopt an s–cis conformation and the alkoxyl chains are in trans conformations curving out of the zigzag plane. Intramolecular O—H O hydrogen bonds involving the keto and hydroxy groups generate S(6) ring motifs. The molecules are stacked alternately in a head-to-tail fashion along the a axis and the crystal structure is stabilized by weak C—H interactions. The crystal studied was a non-merohedral twin, the ratio of components being 0.788 (2):0.212 (2). y g g y ( , ) D—H A D—H H A D A D—H A O1A—H1A O2A 0.82 1.79 2.513 (4) 146 O1B—H1B O2B 0.82 1.81 2.530 (4) 146 C22B—H22C Cg1i 0.97 2.77 3.654 (4) 151 C17B—H17D Cg2 0.97 2.82 3.612 (4) 139 C22A—H22B Cg3 0.97 2.93 3.765 (4) 145 Symmetry code: (i) x 1; y; z. Cg1, Cg2 and Cg3 are the centroids of the C1A--C6A, C10A–C15A and C1B–C6B rings, respectively. Data collection: APEX2 (Bruker, 2005); cell reﬁnement: SAINT (Bruker, 2005); data reduction: SAINT (Bruker, 2005); program(s) used to solve structure: SHELXTL (Sheldrick, 2008); program(s) used to reﬁne structure: SHELXTL (Sheldrick, 2008); molecular graphics: SHELXTL (Sheldrick, 2008); software used to prepare material for publication: SHELXTL (Sheldrick, 2008). Acta Cryst. (2009). E65, o1133–o1134 (E)-3-[4-(Dodecyloxy)phenyl]-1-(2- hydroxyphenyl)prop-2-en-1-one Experimental Crystal data C27H36O3 Mr = 408.56 Triclinic, P1 a = 7.4953 (6) A˚ b = 13.4714 (9) A˚ c = 23.7874 (18) A˚ = 75.116 (4) = 83.876 (5) = 84.669 (5) V = 2302.7 (3) A˚ 3 Z = 4 Mo K radiation = 0.08 mm1 T = 100 K 0.55 0.13 0.06 mm Data collection Bruker APEXII diffractometer Absorption correction: multi-scan (SADABS; Bruker, 2005) Tmin = 0.960, Tmax = 0.996 8571 measured reﬂections 8571 independent reﬂections 4737 reﬂections with I > 2(I) Reﬁnement R[F 2 > 2(F 2)] = 0.069 wR(F 2) = 0.206 S = 1.03 8571 reﬂections 546 parameters H-atom parameters constrained max = 0.29 e A˚ 3 min = 0.30 e A˚ 3 Ibrahim Abdul Razak,a‡ Hoong-Kun Fun,a§ Zainab Ngaini,b Siti Muhaini Haris Fadzillahb and Hasnain Hussainc Related literature For general background to the biological activity of chalcone derivatives, see: Bhat et al. (2005); Xue et al. (2004); Satya- narayana et al. (2004); Zhao et al. (2005); Lee et al. (2006). For related structures, see: Ng et al. (2006); Razak et al. (2009); Ngaini, Fadzillah et al. (2009); Ngaini, Rahman et al. (2009). For details of hydrogen-bond motifs, see: Bernstein et al. (1995). For stability of the temperature controller used for the data collection, see: Cosier & Glazer (1986). HKF and IAR thank the Malaysian Government and Universiti Sains Malaysia for the Science Fund grant No. 305/ PFIZIK/613312 and for the Research University Golden Goose grant No. 1001/PFIZIK/811012. ZN and HH thank Universiti Malaysia Sarawak for the Geran Penyelidikan Dana Khas Inovasi, grant No. DI/01/2007(01). SMHF thanks Malaysian Government and Universiti Malaysia Sarawak for providing a scholarship for postgraduate studies. Supplementary data and ﬁgures for this paper are available from the IUCr electronic archives (Reference: LH2802). Razak et al. o1133 doi:10.1107/S1600536809014925 organic compounds Ngaini, Z., Fadzillah, S. M. H., Rahman, N. I. A., Hussain, H., Razak, I. A. & Fun, H.-K. (2009). Acta Cryst. E65, o879–o880. Zhao, L. M., Jin, H. S., Sun, L. P., Piao, H. R. & Quan, Z. S. (2005). Chem. Lett. 15, 5027–5029. Acta Cryst. (2009). E65, o1133–o1134 Ngaini, Z., Fadzillah, S. M. H., Rahman, N. I. A., Hussain, H., Razak, I. A. & Fun, H.-K. (2009). Acta Cryst. E65, o879–o880. Ngaini, Z., Fadzillah, S. M. H., Rahman, N. I. A., Hussain, H., Razak, I. A. & Fun, H.-K. (2009). Acta Cryst. E65, o879–o880. Ngaini, Z., Rahman, N. I. A., Hussain, H., Razak, I. A. & Fun, H.-K. (2009). Acta Cryst. E65, o889–o890. Razak, I. A., Fun, H.-K., Ngaini, Z., Fadzillah, S. M. H. & Hussain, H. (2009). Acta Cryst. E65, o881–o882. Satyanarayana, M., Tiwari, P., Tripathi, B. K., Srivastava, A. K. & Pratap, R. (2004). Bioorg. Med. Chem. Lett. 12, 883–889. Sheldrick, G. M. (2008). Acta Cryst. A64, 112–122. Xue, C. X., Cui, S. Y., Liu, M. C., Hu, Z. D. & Fan, B. T. (2004). Eur. J. Med. Chem. 39, 745–753. Zhao, L. M., Jin, H. S., Sun, L. P., Piao, H. R. & Quan, Z. S. (2005). Chem. Lett. 15, 5027–5029. ( ) Ngaini, Z., Rahman, N. I. A., Hussain, H., Razak, I. A. & Fun, H.-K. (2009). Acta Cryst. E65, o889–o890. Xue, C. X., Cui, S. Y., Liu, M. C., Hu, Z. D. & Fan, B. T. (2004). Eur. J. Med. Chem. 39, 745–753. Razak, I. A., Fun, H.-K., Ngaini, Z., Fadzillah, S. M. H. & Hussain, H. (2009). Acta Cryst. E65, o881–o882. ( ) Sheldrick, G. M. (2008). Acta Cryst. A64, 112–122. References Bernstein, J., Davis, R. E., Shimoni, L. & Chang, N.-L. (1995). Angew. Chem. Int. Ed. Engl. 34, 1555–1573. Ngaini, Z., Rahman, N. I. A., Hussain, H., Razak, I. A. & Fun, H.-K. (2009). Acta Cryst. E65, o889–o890. Bhat, B. A., Dhar, K. L., Puri, S. C., Saxena, A. K., Shanmugavel, M. & Qazi, G. N. (2005). Bioorg. Med. Chem. Lett. 15, 3177–3180. Razak, I. A., Fun, H.-K., Ngaini, Z., Fadzillah, S. M. H. & Hussain, H. (2009). Acta Cryst. E65, o881–o882. N. (2005). Bioorg. Med. Chem. Lett. 15, 3177–3180. Bruker (2005). APEX2, SAINT and SADABS. Bruker AXS Inc., Madison, ( ) g , Bruker (2005). APEX2, SAINT and SADABS. Bruker AXS Inc., Madison, Wi i USA Bruker (2005). APEX2, SAINT and SADABS. Bruker AXS Inc., Madison, Wisconsin, USA. Satyanarayana, M., Tiwari, P., Tripathi, B. K., Srivastava, A. K. & Pratap, R. (2004). Bioorg. Med. Chem. Lett. 12, 883–889. Wisconsin, USA. Cosier, J. & Glazer, A. M. (1986). J. Appl. Cryst. 19, 105–107. , Cosier, J. & Glazer, A. M. (1986). J. Appl. Cryst. 19, 105–107. Cosier, J. & Glazer, A. M. (1986). J. Appl. Cryst. 19, 105–107 Sheldrick, G. M. (2008). Acta Cryst. A64, 112–122. Lee, Y. S., Lim, S. S., Shin, K. H., Kim, Y. S., Ohuchi, K. & Jung, S. H. (2006). Biol. Pharm. Bull. 29, 1028–1031. Xue, C. X., Cui, S. Y., Liu, M. C., Hu, Z. D. & Fan, B. T. (2004). Eur. J. Med. Chem. 39, 745–753. Biol. Pharm. Bull. 29, 1028–1031. Ng, S.-L., Razak, I. A., Fun, H.-K., Shettigar, V., Patil, P. S. & Dharmaprakash, S. M. (2006). Acta Cryst. E62, o2175–o2177. Zhao, L. M., Jin, H. S., Sun, L. P., Piao, H. R. & Quan, Z. S. (2005). Chem. Lett. 15, 5027–5029. o1134 Razak et al. C27H36O3 Acta Cryst. (2009). E65, o1133–o1134 Comment Chalcone derivatives are of interest because of their biological properties such as anticancer (Bhat et al., 2005), antimalarial (Xue et al., 2004), antiangiogenic and antitumour (Lee et al., 2006), antiplatelet activity (Zhao et al., 2005) and antihyper- glycemic activity (Satyanarayana et al., 2004). As part of our studies on their biological properties, we have synthesized the title chalcone derivative, (I). Its antibacterial activities were tested against E. coli ATCC 8739 and showed antimicrobial activity. The structure determination of (I) was carried out and is reported in this paper. The asymmetric unit of (I) consists of two crystallographically independent molecules, A and B (Fig.1). The bond lengths show normal values (Allen et al., 1987). The mean plane through the enone moiety (O2/C7/C8/C9) makes dihedral angles with the two benzene rings with values of 3.4 (2)° (C1—C6) and 16.0 (2)° (C10—C15) in molecule A and 7.8 (2)° (C1—C6) and 15.7 (2)° (C10—C15) in molecule B. The two benzene rings form dihedral angles with each other of 17.1 (2)° and 17.6 (2)° in molecules A and B, respectively. The enone moieties adopt s-cis conformation with the O2—C7—C8—C9 torsion angle being 6.5 (5)° for molecule A and 8.8 (5)° for B. In what follows, the distortion of the angles is relative to what is expected in terms of hybridization principles. In molecule A, the slight enlargement of the C5A—C6A—C7A (122.4 (3)°) and C6A—C7A—C8A (121.9 (3)°) angles may be the result of the short H5AA···H8AA (2.14 Å) contact whereas the short H8AA···H11A (2.35 Å) contact may widen the C8A—C9A—C10A (129.0 (3)°) and C9A—C10A—C11A (123.0 (3)°) angles. The short H14A···H16A (2.26 Å) contact may result in the opening of the O3A—C13A—C14A (124.9 (3)°) angle. Likewise, in molecule B, a close interatomic contact between H5BA and H8BA (2.13 Å) may result in the widening of the C5B—C6B—C7B (123.1 (3)°) and C6B—C7B—C8B (121.2 (3)°) angles whereas the opening of C8B—C9B—C10B and C9B—C10B—C11B angles to 128.6 (3)° and 123.0 (3)°, respectively, may be the result of the close H8BA···H11B (2.30 Å) contact. Similar strain induced by a short H14B···H16C (2.26 Å) contact may result in the opening of the O3B—C13B—C14B (125.2 (3)°) angle. These features were also reported in related structures (Ng et al., 2006; Razak et al., 2009; Ngaini, Fadzillah et al., 2009; Ngaini, Rahman et al., 2009). supplementary materials supplementary materials Acta Cryst. (2009). E65, o1133-o1134 [ doi:10.1107/S1600536809014925 ] Acta Cryst. (2009). E65, o1133-o1134 [ doi:10.1107/S1600536809014925 ] Experimental A mixture of 2-hydroxyacetophenone (2.72 ml, 20 mmol) and 4-dodecyloxybenzaldehyde (5.81 ml, 20 mmol) and KOH (4.04 g, 72 mmol) in 60 ml of methanol was heated at reflux for 10 h. The reaction was cooled to room temperature and acidified with cold diluted HCl (2 N). The resulting precipitate was filtered, washed and dried. After redissolving in hexane, followed by a few days of slow evaporation, crystals were collected. Comment The conformation throughout the zigzag alkoxyl tails in both molecules is trans with the largest deviation from the ideal value being -174.4 (3)° for the C19A—C20A—C21A—C22A torsion angle in molecule A and -173.5 (3)° for the C17B—C18B—C19B—C20B torsion angle in B. Even though the torsion angle C16—O3—C13—C14 in each molecule is -6.8 (5)° for A and -12.2 (5)° for B, the alkoxyl chains curve out of the zigzag plane with the least-squares plane through the chain making dihedral angle with the attached benzene ring of 17.02 (19)° [maximum deviation of -0.302 (4)Å at C21A] and 16.73 (19)° [maximum deviation of -0.256 (4)Å at C21B], for molecules A and B, respectively. An intramolecular O-H···O interaction involving the keto and hydroxy groups (Table 1) in both molecules generates S(6) ring motifs (Bernstein et al., 1995). In the crystal structure, the molecules are stacked alternately along the a axis in a head-to-tail manner (Fig. 2). In the absence of conventional hydrogen bonds, the crystal structure is stabilized by weak C—H···π interactions (Table 1). sup-1 supplementary materials Figures Fig. 1. The asymmetric unit of the title compound, showing 50% probability displacement el- lipsoids and the atom numbering scheme. Intramolecular interactions were shown as dashed line. Fig. 2. Part of the crystal structure of (I) viewed along the a axis. Fig. 2. Part of the crystal structure of (I) viewed along the a axis. (E)-3-[4-(Dodecyloxy)phenyl]-1-(2-hydroxyphenyl)prop-2-en-1-one (E)-3-[4-(Dodecyloxy)phenyl]-1-(2-hydroxyphenyl)prop-2-en-1-one Refinement All H atoms were positioned geometrically and refined using a riding model. The Uiso(H) values were constrained to be 1.5Ueq(C,O) (methyl H and hydroxyl H atoms) and 1.2Ueq(C) (other H atoms). The rotating model group was considered for the methyl group. The crystal is a twin with a refined BASF = 0.212 (2). Figures Fig. 1. The asymmetric unit of the title compound, showing 50 lipsoids and the atom numbering scheme. Intramolecular inter line. Fig. 2. Part of the crystal structure of (I) viewed along the a ax (E)-3-[4-(Dodecyloxy)phenyl]-1-(2-hydroxyphenyl)prop-2-en-1-one Refinement Secondary atom site location: difference Fourier map Hydrogen site location: inferred from neighbouring sites H-atom parameters constrained H-atom parameters constrained H-atom parameters constrained w = 1/[σ2(Fo 2) + (0.0924P)2 + 0.763P] where P = (Fo 2 + 2Fc 2)/3 (Δ/σ)max < 0.001 Δρmax = 0.29 e Å−3 Δρmin = −0.30 e Å−3 w = 1/[σ2(Fo 2) + (0.0924P)2 + 0.763P] where P = (Fo 2 + 2Fc 2)/3 (Δ/σ)max < 0.001 Δρmax = 0.29 e Å−3 Δρmin = −0.30 e Å−3 546 parameters Extinction correction: none Extinction correction: none (E)-3-[4-(Dodecyloxy)phenyl]-1-(2-hydroxyphenyl)prop-2-en-1-one Crystal data C27H36O3 Z = 4 Mr = 408.56 F000 = 888 Triclinic, P1 Dx = 1.179 Mg m−3 Hall symbol: -P 1 Mo Kα radiation λ = 0.71073 Å a = 7.4953 (6) Å Cell parameters from 3668 reflections b = 13.4714 (9) Å θ = 2.7–28.0º c = 23.7874 (18) Å µ = 0.08 mm−1 α = 75.116 (4)º T = 100 K β = 83.876 (5)º Needle, yellow γ = 84.669 (5)º 0.55 × 0.13 × 0.06 mm V = 2302.7 (3) Å3 sup-2 supplementary materials Data collection Bruker APEXII diffractometer 8571 independent reflections Radiation source: sealed tube 4737 reflections with I > 2σ(I) Monochromator: graphite Rint = 0.0000 T = 100 K θmax = 25.5º π and ω scans θmin = 0.9º Absorption correction: multi-scan (SADABS; Bruker, 2005) h = −8→9 Tmin = 0.960, Tmax = 0.996 k = −16→16 8571 measured reflections l = −28→28 Data collection Bruker APEXII diffractometer 8571 independent reflections Radiation source: sealed tube 4737 reflections with I > 2σ(I) Monochromator: graphite Rint = 0.0000 T = 100 K θmax = 25.5º π and ω scans θmin = 0.9º Absorption correction: multi-scan (SADABS; Bruker, 2005) h = −8→9 Tmin = 0.960, Tmax = 0.996 k = −16→16 8571 measured reflections l = −28→28 Data collection Bruker APEXII diffractometer Radiation source: sealed tube Monochromator: graphite T = 100 K π and ω scans Absorption correction: multi-scan (SADABS; Bruker, 2005) Tmin = 0.960, Tmax = 0.996 8571 measured reflections Absorption correction: multi-scan (SADABS; Bruker, 2005) Tmin = 0.960, Tmax = 0.996 8571 measured reflections Special details Experimental. The crystal was placed in the cold stream of an Oxford Cryosystems Cobra open-flow nitrogen cryostat (Cosier & Glazer, 1986) operating at 100.0 (1) K. Geometry. All e.s.d.'s (except the e.s.d. in the dihedral angle between two l.s. planes) are estimated using the full covariance mat- rix. The cell e.s.d.'s are taken into account individually in the estimation of e.s.d.'s in distances, angles and torsion angles; correlations between e.s.d.'s in cell parameters are only used when they are defined by crystal symmetry. An approximate (isotropic) treatment of cell e.s.d.'s is used for estimating e.s.d.'s involving l.s. planes. Refinement. Refinement of F2 against ALL reflections. The weighted R-factor wR and goodness of fit S are based on F2, convention- al R-factors R are based on F, with F set to zero for negative F2. The threshold expression of F2 > σ(F2) is used only for calculating R- factors(gt) etc. and is not relevant to the choice of reflections for refinement. R-factors based on F2 are statistically about twice as large as those based on F, and R- factors based on ALL data will be even larger. 3 Fractional atomic coordinates and isotropic or equivalent isotropic displacement parameters (Å2) x y z Uiso/Ueq O1A 0.6765 (4) 0.69790 (18) 0.78918 (11) 0.0307 (6) H1A 0.6609 0.6985 0.7555 0.046 O2A 0.6335 (3) 0.62359 (18) 0.70527 (11) 0.0290 (6) Fractional atomic coordinates and isotropic or equivalent isotropic displacement parameters (Å2) sup-3 supplementary materials supplementary materials supplementary materials sup-6 C18B 0.1366 (5) 0.3554 (3) 0.64911 (15) 0.0245 (8) H18C 0.2292 0.3050 0.6663 0.029* H18D 0.0206 0.3289 0.6646 0.029* C19B 0.1520 (5) 0.4559 (3) 0.66580 (15) 0.0249 (8) H19C 0.2613 0.4858 0.6457 0.030* H19D 0.0517 0.5031 0.6515 0.030* C20B 0.1549 (5) 0.4484 (3) 0.73017 (15) 0.0231 (8) H20C 0.2600 0.4055 0.7444 0.028* H20D 0.0492 0.4153 0.7510 0.028* C21B 0.1587 (5) 0.5525 (3) 0.74315 (15) 0.0248 (8) H21C 0.0480 0.5927 0.7321 0.030* H21D 0.2572 0.5882 0.7189 0.030* C22B 0.1799 (5) 0.5491 (3) 0.80653 (15) 0.0255 (8) H22C 0.0780 0.5170 0.8307 0.031* H22D 0.2875 0.5063 0.8184 0.031* C23B 0.1928 (5) 0.6542 (3) 0.81750 (15) 0.0262 (8) H23C 0.0837 0.6963 0.8067 0.031* H23D 0.2924 0.6871 0.7924 0.031* C24B 0.2195 (5) 0.6512 (3) 0.88035 (15) 0.0262 (8) H24C 0.3210 0.6032 0.8925 0.031* H24D 0.1137 0.6252 0.9049 0.031* C25B 0.2525 (5) 0.7555 (3) 0.89035 (15) 0.0276 (9) H25C 0.3601 0.7808 0.8666 0.033* H25D 0.1524 0.8041 0.8774 0.033* C26B 0.2743 (5) 0.7519 (3) 0.95361 (15) 0.0281 (9) H26C 0.3818 0.7090 0.9654 0.034* H26D 0.1724 0.7200 0.9779 0.034* C27B 0.2879 (6) 0.8576 (3) 0.96430 (17) 0.0361 (10) H27D 0.3122 0.8498 1.0041 0.054* H27E 0.1764 0.8979 0.9570 0.054* H27F 0.3835 0.8915 0.9386 0.054* Atomic displacement parameters (Å2) U11 U22 U33 U12 U13 U23 O1A 0.0346 (16) 0.0267 (14) 0.0343 (16) −0.0020 (11) −0.0117 (13) −0.0102 (12) O2A 0.0320 (15) 0.0243 (14) 0.0325 (15) −0.0041 (11) −0.0084 (12) −0.0074 (11) O3A 0.0314 (15) 0.0256 (14) 0.0220 (14) −0.0052 (11) −0.0045 (11) −0.0081 (11) C1A 0.0151 (19) 0.027 (2) 0.033 (2) −0.0025 (15) −0.0031 (16) −0.0072 (17) C2A 0.022 (2) 0.031 (2) 0.034 (2) −0.0024 (16) −0.0058 (17) −0.0160 (17) C3A 0.025 (2) 0.043 (2) 0.023 (2) −0.0046 (17) −0.0076 (17) −0.0121 (18) C4A 0.032 (2) 0.029 (2) 0.025 (2) −0.0052 (17) −0.0047 (17) −0.0054 (16) C5A 0.025 (2) 0.027 (2) 0.027 (2) −0.0053 (16) −0.0017 (16) −0.0102 (16) C6A 0.0143 (18) 0.027 (2) 0.024 (2) −0.0053 (15) −0.0010 (15) −0.0078 (16) C7A 0.0136 (18) 0.025 (2) 0.027 (2) −0.0029 (14) −0.0030 (15) −0.0055 (16) C8A 0.020 (2) 0.024 (2) 0.027 (2) −0.0006 (15) −0.0037 (16) −0.0048 (16) C9A 0.0148 (19) 0.025 (2) 0.029 (2) −0.0038 (14) −0.0057 (16) −0.0032 (16) C10A 0.0111 (18) 0.028 (2) 0.028 (2) −0.0022 (14) −0.0037 (15) −0.0102 (16) supplementary materials sup-4 O3A 0.6291 (3) 0.19687 (18) 0.50595 (10) 0.0255 (6) C1A 0.6917 (5) 0.5994 (3) 0.82144 (16) 0.0249 (8) C2A 0.7137 (5) 0.5837 (3) 0.88068 (16) 0.0273 (9) H2AA 0.7158 0.6397 0.8966 0.033* C3A 0.7324 (5) 0.4852 (3) 0.91544 (16) 0.0294 (9) H3AA 0.7456 0.4751 0.9550 0.035* C4A 0.7320 (5) 0.4007 (3) 0.89221 (15) 0.0288 (9) H4AA 0.7469 0.3344 0.9159 0.035* C5A 0.7094 (5) 0.4154 (3) 0.83407 (15) 0.0256 (8) H5AA 0.7088 0.3585 0.8187 0.031* C6A 0.6871 (4) 0.5145 (3) 0.79750 (15) 0.0210 (8) C7A 0.6563 (4) 0.5329 (3) 0.73494 (15) 0.0220 (8) C8A 0.6552 (5) 0.4480 (3) 0.70683 (15) 0.0235 (8) H8AA 0.6597 0.3802 0.7291 0.028* C9A 0.6477 (4) 0.4687 (3) 0.64905 (15) 0.0231 (8) H9AA 0.6432 0.5381 0.6297 0.028* C10A 0.6455 (4) 0.3974 (3) 0.61220 (15) 0.0214 (8) C11A 0.6156 (5) 0.2928 (3) 0.63470 (15) 0.0230 (8) H11A 0.5988 0.2660 0.6749 0.028* C12A 0.6109 (4) 0.2297 (3) 0.59822 (15) 0.0221 (8) H12A 0.5905 0.1606 0.6139 0.027* C13A 0.6362 (4) 0.2674 (3) 0.53777 (15) 0.0203 (8) C14A 0.6673 (5) 0.3704 (3) 0.51432 (15) 0.0229 (8) H14A 0.6862 0.3961 0.4741 0.028* C15A 0.6698 (5) 0.4344 (3) 0.55132 (15) 0.0233 (8) H15A 0.6882 0.5037 0.5354 0.028* C16A 0.6352 (5) 0.2302 (3) 0.44325 (15) 0.0236 (8) H16A 0.7410 0.2680 0.4277 0.028* H16B 0.5292 0.2746 0.4313 0.028* C17A 0.6417 (5) 0.1341 (3) 0.42151 (15) 0.0252 (8) H17A 0.5401 0.0951 0.4406 0.030* H17B 0.7502 0.0923 0.4334 0.030* C18A 0.6386 (5) 0.1513 (3) 0.35641 (15) 0.0233 (8) H18A 0.7441 0.1861 0.3368 0.028* H18B 0.5331 0.1953 0.3437 0.028* C19A 0.6354 (5) 0.0500 (3) 0.33918 (15) 0.0240 (8) H19A 0.7360 0.0046 0.3550 0.029* H19B 0.5261 0.0178 0.3575 0.029* C20A 0.6443 (5) 0.0575 (3) 0.27412 (15) 0.0255 (8) H20A 0.7542 0.0885 0.2554 0.031* H20B 0.5437 0.1023 0.2578 0.031* C21A 0.6396 (5) −0.0469 (3) 0.26067 (15) 0.0252 (8) H21A 0.7325 −0.0935 0.2806 0.030* H21B 0.5246 −0.0746 0.2764 0.030* C22A 0.6664 (5) −0.0442 (3) 0.19601 (15) 0.0246 (8) H22A 0.7767 −0.0116 0.1794 0.029* H22B 0.5678 −0.0025 0.1765 0.029* C23A 0.6765 (5) −0.1509 (3) 0.18409 (15) 0.0251 (8) H23A 0.7717 −0.1936 0.2051 0.030* sup-4 supplementary materials H23B 0.5641 −0.1823 0.1992 0.030* C24A 0.7112 (5) −0.1494 (3) 0.11987 (15) 0.0265 (9) H24A 0.8193 −0.1139 0.1041 0.032* H24B 0.6119 −0.1105 0.0993 0.032* C25A 0.7332 (5) −0.2558 (3) 0.10808 (15) 0.0276 (9) H25A 0.8331 −0.2946 0.1284 0.033* H25B 0.6254 −0.2915 0.1240 0.033* C26A 0.7664 (5) −0.2538 (3) 0.04410 (15) 0.0288 (9) H26A 0.8748 −0.2187 0.0282 0.035* H26B 0.6671 −0.2145 0.0237 0.035* C27A 0.7868 (6) −0.3608 (3) 0.03238 (16) 0.0336 (10) H27A 0.8210 −0.3548 −0.0084 0.050* H27B 0.6745 −0.3925 0.0431 0.050* H27C 0.8778 −0.4023 0.0551 0.050* O1B 0.1478 (4) −0.19486 (18) 0.21550 (11) 0.0324 (7) H1B 0.1245 −0.1949 0.2500 0.049* O2B 0.0945 (3) −0.11900 (18) 0.30324 (11) 0.0290 (6) O3B 0.1502 (3) 0.30711 (17) 0.50023 (10) 0.0251 (6) C1B 0.1699 (5) −0.0980 (3) 0.18370 (16) 0.0248 (8) C2B 0.2029 (5) −0.0831 (3) 0.12353 (17) 0.0296 (9) H2BA 0.2081 −0.1390 0.1071 0.036* C3B 0.2280 (5) 0.0137 (3) 0.08821 (17) 0.0328 (10) H3BA 0.2487 0.0230 0.0480 0.039* C4B 0.2224 (5) 0.0976 (3) 0.11249 (16) 0.0310 (9) H4BA 0.2395 0.1631 0.0886 0.037* C5B 0.1917 (5) 0.0837 (3) 0.17155 (16) 0.0252 (8) H5BA 0.1892 0.1402 0.1874 0.030* C6B 0.1638 (4) −0.0137 (3) 0.20895 (16) 0.0224 (8) C7B 0.1286 (4) −0.0313 (3) 0.27275 (15) 0.0224 (8) C8B 0.1362 (4) 0.0522 (3) 0.30173 (15) 0.0230 (8) H8BA 0.1448 0.1195 0.2794 0.028* C9B 0.1308 (4) 0.0320 (3) 0.35977 (15) 0.0217 (8) H9BA 0.1227 −0.0367 0.3796 0.026* C10B 0.1359 (5) 0.1032 (3) 0.39596 (15) 0.0231 (8) C11B 0.1184 (5) 0.2102 (3) 0.37436 (15) 0.0234 (8) H11B 0.1037 0.2379 0.3350 0.028* C12B 0.1225 (5) 0.2748 (3) 0.41016 (16) 0.0258 (9) H12B 0.1100 0.3457 0.3949 0.031* C13B 0.1452 (5) 0.2351 (3) 0.46948 (16) 0.0230 (8) C14B 0.1622 (5) 0.1297 (3) 0.49237 (16) 0.0249 (8) H14B 0.1760 0.1024 0.5319 0.030* C15B 0.1585 (5) 0.0651 (3) 0.45546 (16) 0.0259 (9) H15B 0.1713 −0.0057 0.4708 0.031* C16B 0.1390 (5) 0.2755 (3) 0.56305 (15) 0.0237 (8) H16C 0.2351 0.2240 0.5761 0.028* H16D 0.0247 0.2466 0.5783 0.028* C17B 0.1564 (5) 0.3710 (3) 0.58335 (15) 0.0247 (8) H17C 0.0658 0.4232 0.5666 0.030* H17D 0.2732 0.3969 0.5684 0.030* sup-5 supplementary materials Atomic displacement parameters (Å2) sup-6 supplementary materials C11A 0.0179 (19) 0.029 (2) 0.023 (2) −0.0024 (15) −0.0054 (15) −0.0049 (16) C12A 0.0167 (19) 0.0210 (19) 0.028 (2) −0.0016 (14) −0.0059 (15) −0.0042 (16) C13A 0.0132 (18) 0.0224 (19) 0.029 (2) 0.0012 (14) −0.0062 (15) −0.0130 (16) C14A 0.022 (2) 0.027 (2) 0.021 (2) −0.0014 (15) −0.0082 (16) −0.0062 (16) C15A 0.0210 (19) 0.0197 (19) 0.029 (2) −0.0052 (15) −0.0059 (16) −0.0035 (15) C16A 0.022 (2) 0.026 (2) 0.025 (2) −0.0033 (15) −0.0034 (16) −0.0098 (16) C17A 0.022 (2) 0.030 (2) 0.025 (2) −0.0019 (16) −0.0032 (16) −0.0084 (16) C18A 0.021 (2) 0.025 (2) 0.026 (2) −0.0010 (15) −0.0059 (16) −0.0073 (16) C19A 0.021 (2) 0.026 (2) 0.025 (2) −0.0035 (15) −0.0063 (16) −0.0045 (16) C20A 0.020 (2) 0.030 (2) 0.028 (2) −0.0011 (16) −0.0066 (16) −0.0077 (16) C21A 0.024 (2) 0.0225 (19) 0.031 (2) −0.0004 (15) −0.0100 (17) −0.0057 (16) C22A 0.022 (2) 0.028 (2) 0.026 (2) −0.0028 (15) −0.0058 (16) −0.0092 (16) C23A 0.021 (2) 0.027 (2) 0.027 (2) 0.0012 (15) −0.0087 (16) −0.0061 (16) C24A 0.026 (2) 0.029 (2) 0.025 (2) −0.0020 (16) −0.0028 (16) −0.0075 (16) C25A 0.026 (2) 0.030 (2) 0.028 (2) −0.0037 (16) −0.0076 (17) −0.0067 (16) C26A 0.032 (2) 0.029 (2) 0.026 (2) −0.0017 (17) −0.0053 (17) −0.0075 (16) C27A 0.040 (3) 0.036 (2) 0.028 (2) −0.0003 (18) −0.0104 (19) −0.0130 (18) O1B 0.0386 (17) 0.0279 (15) 0.0329 (16) −0.0060 (12) −0.0052 (14) −0.0094 (12) O2B 0.0291 (15) 0.0265 (14) 0.0324 (15) −0.0084 (11) −0.0100 (12) −0.0040 (12) O3B 0.0280 (15) 0.0236 (13) 0.0250 (14) −0.0037 (11) −0.0058 (11) −0.0064 (11) C1B 0.0171 (19) 0.027 (2) 0.032 (2) −0.0024 (15) −0.0081 (16) −0.0070 (17) C2B 0.029 (2) 0.029 (2) 0.036 (2) −0.0018 (17) −0.0062 (18) −0.0156 (18) C3B 0.034 (2) 0.040 (2) 0.026 (2) 0.0029 (18) −0.0060 (18) −0.0129 (18) C4B 0.034 (2) 0.027 (2) 0.030 (2) −0.0030 (17) −0.0050 (18) −0.0033 (17) C5B 0.025 (2) 0.023 (2) 0.030 (2) 0.0008 (15) −0.0060 (17) −0.0105 (16) C6B 0.0127 (18) 0.0245 (19) 0.032 (2) 0.0008 (14) −0.0059 (16) −0.0104 (16) C7B 0.0121 (18) 0.027 (2) 0.029 (2) −0.0004 (14) −0.0098 (15) −0.0060 (16) C8B 0.0172 (19) 0.0221 (19) 0.031 (2) −0.0010 (15) −0.0056 (16) −0.0068 (16) C9B 0.0161 (19) 0.0212 (19) 0.028 (2) −0.0028 (14) −0.0046 (15) −0.0053 (15) C10B 0.0159 (19) 0.025 (2) 0.030 (2) −0.0054 (15) −0.0028 (16) −0.0078 (16) C11B 0.0180 (19) 0.030 (2) 0.022 (2) −0.0036 (15) −0.0038 (15) −0.0049 (16) C12B 0.019 (2) 0.026 (2) 0.033 (2) −0.0031 (15) −0.0057 (16) −0.0059 (17) C13B 0.0152 (19) 0.027 (2) 0.028 (2) −0.0052 (15) −0.0021 (15) −0.0082 (16) C14B 0.025 (2) 0.029 (2) 0.021 (2) −0.0026 (16) −0.0044 (16) −0.0067 (16) C15B 0.023 (2) 0.024 (2) 0.030 (2) −0.0020 (15) −0.0055 (17) −0.0046 (16) C16B 0.0204 (19) 0.026 (2) 0.024 (2) −0.0050 (15) −0.0049 (15) −0.0031 (16) C17B 0.0174 (19) 0.028 (2) 0.029 (2) −0.0025 (15) −0.0041 (16) −0.0058 (16) C18B 0.020 (2) 0.027 (2) 0.026 (2) −0.0006 (15) −0.0043 (16) −0.0061 (16) C19B 0.022 (2) 0.025 (2) 0.027 (2) −0.0014 (15) −0.0054 (16) −0.0034 (16) C20B 0.0196 (19) 0.025 (2) 0.024 (2) −0.0023 (15) −0.0044 (15) −0.0046 (15) C21B 0.019 (2) 0.027 (2) 0.028 (2) −0.0025 (15) −0.0023 (16) −0.0061 (16) C22B 0.021 (2) 0.028 (2) 0.028 (2) 0.0007 (15) −0.0052 (16) −0.0069 (16) C23B 0.024 (2) 0.025 (2) 0.030 (2) −0.0040 (16) −0.0052 (17) −0.0072 (16) C24B 0.025 (2) 0.027 (2) 0.028 (2) −0.0034 (16) −0.0079 (17) −0.0068 (16) C25B 0.027 (2) 0.029 (2) 0.028 (2) −0.0027 (16) −0.0070 (17) −0.0079 (16) C26B 0.030 (2) 0.031 (2) 0.024 (2) −0.0031 (17) −0.0047 (17) −0.0084 (16) C27B 0.044 (3) 0.035 (2) 0.033 (2) −0.0091 (19) −0.006 (2) −0.0123 (18) sup-7 supplementary materials supplementary materials Geometric parameters (Å, °) O1A—C1A 1.353 (4) O1B—C1B 1.345 (4) O1A—H1A 0.8200 O1B—H1B 0.8200 O2A—C7A 1.251 (4) O2B—C7B 1.251 (4) O3A—C13A 1.366 (4) O3B—C13B 1.361 (4) O3A—C16A 1.440 (4) O3B—C16B 1.440 (4) C1A—C2A 1.396 (5) C1B—C2B 1.392 (5) C1A—C6A 1.407 (5) C1B—C6B 1.410 (5) C2A—C3A 1.374 (5) C2B—C3B 1.375 (5) C2A—H2AA 0.9300 C2B—H2BA 0.9300 C3A—C4A 1.388 (5) C3B—C4B 1.392 (5) C3A—H3AA 0.9300 C3B—H3BA 0.9300 C4A—C5A 1.372 (5) C4B—C5B 1.366 (5) C4A—H4AA 0.9300 C4B—H4BA 0.9300 C5A—C6A 1.400 (5) C5B—C6B 1.403 (5) C5A—H5AA 0.9300 C5B—H5BA 0.9300 C6A—C7A 1.484 (5) C6B—C7B 1.473 (5) C7A—C8A 1.467 (5) C7B—C8B 1.470 (5) C8A—C9A 1.337 (5) C8B—C9B 1.334 (5) C8A—H8AA 0.9300 C8B—H8BA 0.9300 C9A—C10A 1.459 (5) C9B—C10B 1.449 (5) C9A—H9AA 0.9300 C9B—H9BA 0.9300 C10A—C15A 1.403 (5) C10B—C15B 1.399 (5) C10A—C11A 1.404 (5) C10B—C11B 1.399 (5) C11A—C12A 1.366 (5) C11B—C12B 1.370 (5) C11A—H11A 0.9300 C11B—H11B 0.9300 C12A—C13A 1.395 (5) C12B—C13B 1.397 (5) C12A—H12A 0.9300 C12B—H12B 0.9300 C13A—C14A 1.387 (5) C13B—C14B 1.384 (5) C14A—C15A 1.384 (5) C14B—C15B 1.389 (5) C14A—H14A 0.9300 C14B—H14B 0.9300 C15A—H15A 0.9300 C15B—H15B 0.9300 C16A—C17A 1.508 (5) C16B—C17B 1.507 (5) C16A—H16A 0.9700 C16B—H16C 0.9700 C16A—H16B 0.9700 C16B—H16D 0.9700 C17A—C18A 1.508 (5) C17B—C18B 1.517 (5) C17A—H17A 0.9700 C17B—H17C 0.9700 C17A—H17B 0.9700 C17B—H17D 0.9700 C18A—C19A 1.525 (5) C18B—C19B 1.522 (5) C18A—H18A 0.9700 C18B—H18C 0.9700 C18A—H18B 0.9700 C18B—H18D 0.9700 C19A—C20A 1.519 (5) C19B—C20B 1.511 (5) C19A—H19A 0.9700 C19B—H19C 0.9700 C19A—H19B 0.9700 C19B—H19D 0.9700 C20A—C21A 1.525 (5) C20B—C21B 1.515 (5) C20A—H20A 0.9700 C20B—H20C 0.9700 C20A—H20B 0.9700 C20B—H20D 0.9700 Geometric parameters (Å, °) O1A—C1A 1.353 (4) O1B—C1B O1A—H1A 0.8200 O1B—H1B O2A—C7A 1.251 (4) O2B—C7B O3A—C13A 1.366 (4) O3B—C13B O3A—C16A 1.440 (4) O3B—C16B C1A—C2A 1.396 (5) C1B—C2B C1A—C6A 1.407 (5) C1B—C6B C2A—C3A 1.374 (5) C2B—C3B C2A—H2AA 0.9300 C2B—H2BA C3A—C4A 1.388 (5) C3B—C4B C3A—H3AA 0.9300 C3B—H3BA C4A—C5A 1.372 (5) C4B—C5B C4A—H4AA 0.9300 C4B—H4BA C5A—C6A 1.400 (5) C5B—C6B C5A—H5AA 0.9300 C5B—H5BA C6A—C7A 1.484 (5) C6B—C7B C7A—C8A 1.467 (5) C7B—C8B C8A—C9A 1.337 (5) C8B—C9B C8A—H8AA 0.9300 C8B—H8BA C9A—C10A 1.459 (5) C9B—C10B C9A—H9AA 0.9300 C9B—H9BA C10A—C15A 1.403 (5) C10B—C15B C10A—C11A 1.404 (5) C10B—C11B C11A—C12A 1.366 (5) C11B—C12B C11A—H11A 0.9300 C11B—H11B C12A—C13A 1.395 (5) C12B—C13B C12A—H12A 0.9300 C12B—H12B C13A—C14A 1.387 (5) C13B—C14B C14A—C15A 1.384 (5) C14B—C15B C14A—H14A 0.9300 C14B—H14B C15A—H15A 0.9300 C15B—H15B C16A—C17A 1.508 (5) C16B—C17B C16A—H16A 0.9700 C16B—H16C C16A—H16B 0.9700 C16B—H16D C17A—C18A 1.508 (5) C17B—C18B C17A—H17A 0.9700 C17B—H17C C17A—H17B 0.9700 C17B—H17D C18A—C19A 1.525 (5) C18B—C19B C18A—H18A 0.9700 C18B—H18C C18A—H18B 0.9700 C18B—H18D C19A—C20A 1.519 (5) C19B—C20B C19A—H19A 0.9700 C19B—H19C C19A—H19B 0.9700 C19B—H19D C20A—C21A 1.525 (5) C20B—C21B C20A—H20A 0.9700 C20B—H20C C20A—H20B 0.9700 C20B—H20D °) 1.353 (4) O1B—C1B 1.345 (4) 0.8200 O1B—H1B 0.8200 1.251 (4) O2B—C7B 1.251 (4) 1.366 (4) O3B—C13B 1.361 (4) 1.440 (4) O3B—C16B 1.440 (4) 1.396 (5) C1B—C2B 1.392 (5) 1.407 (5) C1B—C6B 1.410 (5) 1.374 (5) C2B—C3B 1.375 (5) 0.9300 C2B—H2BA 0.9300 1.388 (5) C3B—C4B 1.392 (5) 0.9300 C3B—H3BA 0.9300 1.372 (5) C4B—C5B 1.366 (5) 0.9300 C4B—H4BA 0.9300 1.400 (5) C5B—C6B 1.403 (5) 0.9300 C5B—H5BA 0.9300 1.484 (5) C6B—C7B 1.473 (5) 1.467 (5) C7B—C8B 1.470 (5) 1.337 (5) C8B—C9B 1.334 (5) 0.9300 C8B—H8BA 0.9300 1.459 (5) C9B—C10B 1.449 (5) 0.9300 C9B—H9BA 0.9300 1.403 (5) C10B—C15B 1.399 (5) 1.404 (5) C10B—C11B 1.399 (5) 1.366 (5) C11B—C12B 1.370 (5) 0.9300 C11B—H11B 0.9300 1.395 (5) C12B—C13B 1.397 (5) 0.9300 C12B—H12B 0.9300 1.387 (5) C13B—C14B 1.384 (5) 1.384 (5) C14B—C15B 1.389 (5) 0.9300 C14B—H14B 0.9300 0.9300 C15B—H15B 0.9300 1.508 (5) C16B—C17B 1.507 (5) 0.9700 C16B—H16C 0.9700 0.9700 C16B—H16D 0.9700 1.508 (5) C17B—C18B 1.517 (5) 0.9700 C17B—H17C 0.9700 0.9700 C17B—H17D 0.9700 1.525 (5) C18B—C19B 1.522 (5) 0.9700 C18B—H18C 0.9700 0.9700 C18B—H18D 0.9700 1.519 (5) C19B—C20B 1.511 (5) 0.9700 C19B—H19C 0.9700 0.9700 C19B—H19D 0.9700 1.525 (5) C20B—C21B 1.515 (5) 0.9700 C20B—H20C 0.9700 0.9700 C20B—H20D 0.9700 sup-8 supplementary materials supplementary materials 1.522 (5) C21B—C22B 1.521 (5 0.9700 C21B—H21C 0.9700 0.9700 C21B—H21D 0.9700 1.529 (5) C22B—C23B 1.518 (5 0.9700 C22B—H22C 0.9700 0.9700 C22B—H22D 0.9700 1.517 (5) C23B—C24B 1.519 (5 0.9700 C23B—H23C 0.9700 0.9700 C23B—H23D 0.9700 1.520 (5) C24B—C25B 1.531 (5 0.9700 C24B—H24C 0.9700 0.9700 C24B—H24D 0.9700 1.510 (5) C25B—C26B 1.519 (5 0.9700 C25B—H25C 0.9700 0.9700 C25B—H25D 0.9700 1.528 (5) C26B—C27B 1.523 (5 0.9700 C26B—H26C 0.9700 0.9700 C26B—H26D 0.9700 0.9600 C27B—H27D 0.9600 0.9600 C27B—H27E 0.9600 0.9600 C27B—H27F 0.9600 109.5 C1B—O1B—H1B 109.5 119.7 (3) C13B—O3B—C16B 119.3 (3 117.3 (3) O1B—C1B—C2B 117.2 (3 122.7 (3) O1B—C1B—C6B 122.8 (3 120.0 (3) C2B—C1B—C6B 120.0 (3 119.9 (3) C3B—C2B—C1B 120.5 (3 120.0 C3B—C2B—H2BA 119.7 120.0 C1B—C2B—H2BA 119.7 120.8 (3) C2B—C3B—C4B 120.1 (4 119.6 C2B—C3B—H3BA 119.9 119.6 C4B—C3B—H3BA 119.9 119.7 (3) C5B—C4B—C3B 119.8 (3 120.1 C5B—C4B—H4BA 120.1 120.1 C3B—C4B—H4BA 120.1 121.1 (3) C4B—C5B—C6B 121.7 (3 119.5 C4B—C5B—H5BA 119.2 119.5 C6B—C5B—H5BA 119.2 118.5 (3) C5B—C6B—C1B 117.8 (3 122.4 (3) C5B—C6B—C7B 123.1 (3 119.1 (3) C1B—C6B—C7B 119.1 (3 119.3 (3) O2B—C7B—C8B 119.0 (3 118.8 (3) O2B—C7B—C6B 119.8 (3 121.9 (3) C8B—C7B—C6B 121.2 (3 119.6 (3) C9B—C8B—C7B 120.5 (3 120.2 C9B—C8B—H8BA 119.7 120.2 C7B—C8B—H8BA 119.7 129.0 (3) C8B—C9B—C10B 128.6 (3 115.5 C8B—C9B—H9BA 115.7 C21A—C22A 1.522 (5) C21B—C22B 1.521 (5) C21A—H21A 0.9700 C21B—H21C 0.9700 C21A—H21B 0.9700 C21B—H21D 0.9700 C22A—C23A 1.529 (5) C22B—C23B 1.518 (5) C22A—H22A 0.9700 C22B—H22C 0.9700 C22A—H22B 0.9700 C22B—H22D 0.9700 C23A—C24A 1.517 (5) C23B—C24B 1.519 (5) C23A—H23A 0.9700 C23B—H23C 0.9700 C23A—H23B 0.9700 C23B—H23D 0.9700 C24A—C25A 1.520 (5) C24B—C25B 1.531 (5) C24A—H24A 0.9700 C24B—H24C 0.9700 C24A—H24B 0.9700 C24B—H24D 0.9700 C25A—C26A 1.510 (5) C25B—C26B 1.519 (5) C25A—H25A 0.9700 C25B—H25C 0.9700 C25A—H25B 0.9700 C25B—H25D 0.9700 C26A—C27A 1.528 (5) C26B—C27B 1.523 (5) C26A—H26A 0.9700 C26B—H26C 0.9700 C26A—H26B 0.9700 C26B—H26D 0.9700 C27A—H27A 0.9600 C27B—H27D 0.9600 C27A—H27B 0.9600 C27B—H27E 0.9600 C27A—H27C 0.9600 C27B—H27F 0.9600 C1A—O1A—H1A 109.5 C1B—O1B—H1B 109.5 C13A—O3A—C16A 119.7 (3) C13B—O3B—C16B 119.3 (3) O1A—C1A—C2A 117.3 (3) O1B—C1B—C2B 117.2 (3) O1A—C1A—C6A 122.7 (3) O1B—C1B—C6B 122.8 (3) C2A—C1A—C6A 120.0 (3) C2B—C1B—C6B 120.0 (3) C3A—C2A—C1A 119.9 (3) C3B—C2B—C1B 120.5 (3) C3A—C2A—H2AA 120.0 C3B—C2B—H2BA 119.7 C1A—C2A—H2AA 120.0 C1B—C2B—H2BA 119.7 C2A—C3A—C4A 120.8 (3) C2B—C3B—C4B 120.1 (4) C2A—C3A—H3AA 119.6 C2B—C3B—H3BA 119.9 C4A—C3A—H3AA 119.6 C4B—C3B—H3BA 119.9 C5A—C4A—C3A 119.7 (3) C5B—C4B—C3B 119.8 (3) C5A—C4A—H4AA 120.1 C5B—C4B—H4BA 120.1 C3A—C4A—H4AA 120.1 C3B—C4B—H4BA 120.1 C4A—C5A—C6A 121.1 (3) C4B—C5B—C6B 121.7 (3) C4A—C5A—H5AA 119.5 C4B—C5B—H5BA 119.2 C6A—C5A—H5AA 119.5 C6B—C5B—H5BA 119.2 C5A—C6A—C1A 118.5 (3) C5B—C6B—C1B 117.8 (3) C5A—C6A—C7A 122.4 (3) C5B—C6B—C7B 123.1 (3) C1A—C6A—C7A 119.1 (3) C1B—C6B—C7B 119.1 (3) O2A—C7A—C8A 119.3 (3) O2B—C7B—C8B 119.0 (3) O2A—C7A—C6A 118.8 (3) O2B—C7B—C6B 119.8 (3) C8A—C7A—C6A 121.9 (3) C8B—C7B—C6B 121.2 (3) C9A—C8A—C7A 119.6 (3) C9B—C8B—C7B 120.5 (3) C9A—C8A—H8AA 120.2 C9B—C8B—H8BA 119.7 C7A—C8A—H8AA 120.2 C7B—C8B—H8BA 119.7 C8A—C9A—C10A 129.0 (3) C8B—C9B—C10B 128.6 (3) C8A—C9A—H9AA 115.5 C8B—C9B—H9BA 115.7 sup-9 supplementary materials supplementary materials sup-10 C10A—C9A—H9AA 115.5 C10B—C9B—H9BA 115.7 C15A—C10A—C11A 117.7 (3) C15B—C10B—C11B 117.3 (3) C15A—C10A—C9A 119.3 (3) C15B—C10B—C9B 119.7 (3) C11A—C10A—C9A 123.0 (3) C11B—C10B—C9B 123.0 (3) C12A—C11A—C10A 120.7 (3) C12B—C11B—C10B 121.3 (3) C12A—C11A—H11A 119.6 C12B—C11B—H11B 119.4 C10A—C11A—H11A 119.6 C10B—C11B—H11B 119.4 C11A—C12A—C13A 121.0 (3) C11B—C12B—C13B 120.5 (3) C11A—C12A—H12A 119.5 C11B—C12B—H12B 119.8 C13A—C12A—H12A 119.5 C13B—C12B—H12B 119.8 O3A—C13A—C14A 124.9 (3) O3B—C13B—C14B 125.2 (3) O3A—C13A—C12A 115.5 (3) O3B—C13B—C12B 115.0 (3) C14A—C13A—C12A 119.6 (3) C14B—C13B—C12B 119.8 (3) C15A—C14A—C13A 119.4 (3) C13B—C14B—C15B 119.0 (3) C15A—C14A—H14A 120.3 C13B—C14B—H14B 120.5 C13A—C14A—H14A 120.3 C15B—C14B—H14B 120.5 C14A—C15A—C10A 121.7 (3) C14B—C15B—C10B 122.2 (3) C14A—C15A—H15A 119.1 C14B—C15B—H15B 118.9 C10A—C15A—H15A 119.1 C10B—C15B—H15B 118.9 O3A—C16A—C17A 106.4 (3) O3B—C16B—C17B 106.2 (3) O3A—C16A—H16A 110.4 O3B—C16B—H16C 110.5 C17A—C16A—H16A 110.4 C17B—C16B—H16C 110.5 O3A—C16A—H16B 110.4 O3B—C16B—H16D 110.5 C17A—C16A—H16B 110.4 C17B—C16B—H16D 110.5 H16A—C16A—H16B 108.6 H16C—C16B—H16D 108.7 C16A—C17A—C18A 115.6 (3) C16B—C17B—C18B 114.6 (3) C16A—C17A—H17A 108.4 C16B—C17B—H17C 108.6 C18A—C17A—H17A 108.4 C18B—C17B—H17C 108.6 C16A—C17A—H17B 108.4 C16B—C17B—H17D 108.6 C18A—C17A—H17B 108.4 C18B—C17B—H17D 108.6 H17A—C17A—H17B 107.4 H17C—C17B—H17D 107.6 C17A—C18A—C19A 111.6 (3) C17B—C18B—C19B 111.2 (3) C17A—C18A—H18A 109.3 C17B—C18B—H18C 109.4 C19A—C18A—H18A 109.3 C19B—C18B—H18C 109.4 C17A—C18A—H18B 109.3 C17B—C18B—H18D 109.4 C19A—C18A—H18B 109.3 C19B—C18B—H18D 109.4 H18A—C18A—H18B 108.0 H18C—C18B—H18D 108.0 C20A—C19A—C18A 116.1 (3) C20B—C19B—C18B 116.2 (3) C20A—C19A—H19A 108.3 C20B—C19B—H19C 108.2 C18A—C19A—H19A 108.3 C18B—C19B—H19C 108.2 C20A—C19A—H19B 108.3 C20B—C19B—H19D 108.2 C18A—C19A—H19B 108.3 C18B—C19B—H19D 108.2 H19A—C19A—H19B 107.4 H19C—C19B—H19D 107.4 C19A—C20A—C21A 112.7 (3) C19B—C20B—C21B 112.7 (3) C19A—C20A—H20A 109.0 C19B—C20B—H20C 109.1 C21A—C20A—H20A 109.0 C21B—C20B—H20C 109.1 C19A—C20A—H20B 109.0 C19B—C20B—H20D 109.1 C21A—C20A—H20B 109.0 C21B—C20B—H20D 109.1 H20A—C20A—H20B 107.8 H20C—C20B—H20D 107.8 115.5 C10B—C9B—H9BA 115.7 117.7 (3) C15B—C10B—C11B 117.3 (3) 119.3 (3) C15B—C10B—C9B 119.7 (3) 123.0 (3) C11B—C10B—C9B 123.0 (3) 120.7 (3) C12B—C11B—C10B 121.3 (3) 119.6 C12B—C11B—H11B 119.4 119.6 C10B—C11B—H11B 119.4 121.0 (3) C11B—C12B—C13B 120.5 (3) 119.5 C11B—C12B—H12B 119.8 119.5 C13B—C12B—H12B 119.8 124.9 (3) O3B—C13B—C14B 125.2 (3) 115.5 (3) O3B—C13B—C12B 115.0 (3) 119.6 (3) C14B—C13B—C12B 119.8 (3) 119.4 (3) C13B—C14B—C15B 119.0 (3) 120.3 C13B—C14B—H14B 120.5 120.3 C15B—C14B—H14B 120.5 121.7 (3) C14B—C15B—C10B 122.2 (3) 119.1 C14B—C15B—H15B 118.9 119.1 C10B—C15B—H15B 118.9 106.4 (3) O3B—C16B—C17B 106.2 (3) 110.4 O3B—C16B—H16C 110.5 110.4 C17B—C16B—H16C 110.5 110.4 O3B—C16B—H16D 110.5 110.4 C17B—C16B—H16D 110.5 108.6 H16C—C16B—H16D 108.7 115.6 (3) C16B—C17B—C18B 114.6 (3) 108.4 C16B—C17B—H17C 108.6 108.4 C18B—C17B—H17C 108.6 108.4 C16B—C17B—H17D 108.6 108.4 C18B—C17B—H17D 108.6 107.4 H17C—C17B—H17D 107.6 111.6 (3) C17B—C18B—C19B 111.2 (3) 109.3 C17B—C18B—H18C 109.4 109.3 C19B—C18B—H18C 109.4 109.3 C17B—C18B—H18D 109.4 109.3 C19B—C18B—H18D 109.4 108.0 H18C—C18B—H18D 108.0 116.1 (3) C20B—C19B—C18B 116.2 (3) 108.3 C20B—C19B—H19C 108.2 108.3 C18B—C19B—H19C 108.2 108.3 C20B—C19B—H19D 108.2 108.3 C18B—C19B—H19D 108.2 107.4 H19C—C19B—H19D 107.4 112.7 (3) C19B—C20B—C21B 112.7 (3) 109.0 C19B—C20B—H20C 109.1 109.0 C21B—C20B—H20C 109.1 109.0 C19B—C20B—H20D 109.1 109.0 C21B—C20B—H20D 109.1 107.8 H20C—C20B—H20D 107.8 sup-10 supplementary materials supplementary materials C22A—C21A—C20A 114.4 (3) C20B—C21B—C22B 115.0 (3) C22A—C21A—H21A 108.6 C20B—C21B—H21C 108.5 C20A—C21A—H21A 108.6 C22B—C21B—H21C 108.5 C22A—C21A—H21B 108.6 C20B—C21B—H21D 108.5 C20A—C21A—H21B 108.6 C22B—C21B—H21D 108.5 H21A—C21A—H21B 107.6 H21C—C21B—H21D 107.5 C21A—C22A—C23A 113.3 (3) C23B—C22B—C21B 113.7 (3) C21A—C22A—H22A 108.9 C23B—C22B—H22C 108.8 C23A—C22A—H22A 108.9 C21B—C22B—H22C 108.8 C21A—C22A—H22B 108.9 C23B—C22B—H22D 108.8 C23A—C22A—H22B 108.9 C21B—C22B—H22D 108.8 H22A—C22A—H22B 107.7 H22C—C22B—H22D 107.7 C24A—C23A—C22A 113.7 (3) C22B—C23B—C24B 114.0 (3) C24A—C23A—H23A 108.8 C22B—C23B—H23C 108.7 C22A—C23A—H23A 108.8 C24B—C23B—H23C 108.7 C24A—C23A—H23B 108.8 C22B—C23B—H23D 108.7 C22A—C23A—H23B 108.8 C24B—C23B—H23D 108.7 H23A—C23A—H23B 107.7 H23C—C23B—H23D 107.6 C23A—C24A—C25A 113.9 (3) C23B—C24B—C25B 114.2 (3) C23A—C24A—H24A 108.8 C23B—C24B—H24C 108.7 C25A—C24A—H24A 108.8 C25B—C24B—H24C 108.7 C23A—C24A—H24B 108.8 C23B—C24B—H24D 108.7 C25A—C24A—H24B 108.8 C25B—C24B—H24D 108.7 H24A—C24A—H24B 107.7 H24C—C24B—H24D 107.6 C26A—C25A—C24A 113.7 (3) C26B—C25B—C24B 113.7 (3) C26A—C25A—H25A 108.8 C26B—C25B—H25C 108.8 C24A—C25A—H25A 108.8 C24B—C25B—H25C 108.8 C26A—C25A—H25B 108.8 C26B—C25B—H25D 108.8 C24A—C25A—H25B 108.8 C24B—C25B—H25D 108.8 H25A—C25A—H25B 107.7 H25C—C25B—H25D 107.7 C25A—C26A—C27A 113.5 (3) C25B—C26B—C27B 113.4 (3) C25A—C26A—H26A 108.9 C25B—C26B—H26C 108.9 C27A—C26A—H26A 108.9 C27B—C26B—H26C 108.9 C25A—C26A—H26B 108.9 C25B—C26B—H26D 108.9 C27A—C26A—H26B 108.9 C27B—C26B—H26D 108.9 H26A—C26A—H26B 107.7 H26C—C26B—H26D 107.7 C26A—C27A—H27A 109.5 C26B—C27B—H27D 109.5 C26A—C27A—H27B 109.5 C26B—C27B—H27E 109.5 H27A—C27A—H27B 109.5 H27D—C27B—H27E 109.5 C26A—C27A—H27C 109.5 C26B—C27B—H27F 109.5 H27A—C27A—H27C 109.5 H27D—C27B—H27F 109.5 H27B—C27A—H27C 109.5 H27E—C27B—H27F 109.5 O1A—C1A—C2A—C3A −179.1 (3) O1B—C1B—C2B—C3B −179.6 (3) C6A—C1A—C2A—C3A 0.6 (5) C6B—C1B—C2B—C3B −0.7 (5) C1A—C2A—C3A—C4A 0.8 (6) C1B—C2B—C3B—C4B 0.7 (6) C2A—C3A—C4A—C5A −1.2 (6) C2B—C3B—C4B—C5B −0.1 (6) C3A—C4A—C5A—C6A 0.2 (6) C3B—C4B—C5B—C6B −0.6 (6) C4A—C5A—C6A—C1A 1.2 (5) C4B—C5B—C6B—C1B 0.6 (5) C4A—C5A—C6A—C7A −177.9 (3) C4B—C5B—C6B—C7B −179.4 (3) sup-11 114.4 (3) C20B—C21B—C22B 115.0 (3) 108.6 C20B—C21B—H21C 108.5 108.6 C22B—C21B—H21C 108.5 108.6 C20B—C21B—H21D 108.5 108.6 C22B—C21B—H21D 108.5 107.6 H21C—C21B—H21D 107.5 113.3 (3) C23B—C22B—C21B 113.7 (3) 108.9 C23B—C22B—H22C 108.8 108.9 C21B—C22B—H22C 108.8 108.9 C23B—C22B—H22D 108.8 108.9 C21B—C22B—H22D 108.8 107.7 H22C—C22B—H22D 107.7 113.7 (3) C22B—C23B—C24B 114.0 (3) 108.8 C22B—C23B—H23C 108.7 108.8 C24B—C23B—H23C 108.7 108.8 C22B—C23B—H23D 108.7 108.8 C24B—C23B—H23D 108.7 107.7 H23C—C23B—H23D 107.6 113.9 (3) C23B—C24B—C25B 114.2 (3) 108.8 C23B—C24B—H24C 108.7 108.8 C25B—C24B—H24C 108.7 108.8 C23B—C24B—H24D 108.7 108.8 C25B—C24B—H24D 108.7 107.7 H24C—C24B—H24D 107.6 113.7 (3) C26B—C25B—C24B 113.7 (3) 108.8 C26B—C25B—H25C 108.8 108.8 C24B—C25B—H25C 108.8 108.8 C26B—C25B—H25D 108.8 108.8 C24B—C25B—H25D 108.8 107.7 H25C—C25B—H25D 107.7 113.5 (3) C25B—C26B—C27B 113.4 (3) 108.9 C25B—C26B—H26C 108.9 108.9 C27B—C26B—H26C 108.9 108.9 C25B—C26B—H26D 108.9 108.9 C27B—C26B—H26D 108.9 107.7 H26C—C26B—H26D 107.7 109.5 C26B—C27B—H27D 109.5 109.5 C26B—C27B—H27E 109.5 109.5 H27D—C27B—H27E 109.5 109.5 C26B—C27B—H27F 109.5 109.5 H27D—C27B—H27F 109.5 109.5 H27E—C27B—H27F 109.5 −179.1 (3) O1B—C1B—C2B—C3B −179.6 (3) 0.6 (5) C6B—C1B—C2B—C3B −0.7 (5) 0.8 (6) C1B—C2B—C3B—C4B 0.7 (6) −1.2 (6) C2B—C3B—C4B—C5B −0.1 (6) 0.2 (6) C3B—C4B—C5B—C6B −0.6 (6) 1.2 (5) C4B—C5B—C6B—C1B 0.6 (5) −177.9 (3) C4B—C5B—C6B—C7B −179.4 (3) sup-11 supplementary materials supplementary materials sup-12 O1A—C1A—C6A—C5A 178.1 (3) O1B—C1B—C6B—C5B 178.9 (3) C2A—C1A—C6A—C5A −1.5 (5) C2B—C1B—C6B—C5B 0.1 (5) O1A—C1A—C6A—C7A −2.8 (5) O1B—C1B—C6B—C7B −1.1 (5) C2A—C1A—C6A—C7A 177.6 (3) C2B—C1B—C6B—C7B 180.0 (3) C5A—C6A—C7A—O2A 178.4 (3) C5B—C6B—C7B—O2B 175.9 (3) C1A—C6A—C7A—O2A −0.7 (5) C1B—C6B—C7B—O2B −4.1 (5) C5A—C6A—C7A—C8A −2.7 (5) C5B—C6B—C7B—C8B −5.4 (5) C1A—C6A—C7A—C8A 178.2 (3) C1B—C6B—C7B—C8B 174.7 (3) O2A—C7A—C8A—C9A 6.5 (5) O2B—C7B—C8B—C9B 8.8 (5) C6A—C7A—C8A—C9A −172.3 (3) C6B—C7B—C8B—C9B −169.9 (3) C7A—C8A—C9A—C10A 179.9 (3) C7B—C8B—C9B—C10B −179.7 (3) C8A—C9A—C10A—C15A −168.6 (4) C8B—C9B—C10B—C15B −169.5 (4) C8A—C9A—C10A—C11A 13.0 (6) C8B—C9B—C10B—C11B 10.8 (6) C15A—C10A—C11A—C12A 0.1 (5) C15B—C10B—C11B—C12B −0.2 (5) C9A—C10A—C11A—C12A 178.5 (3) C9B—C10B—C11B—C12B 179.5 (3) C10A—C11A—C12A—C13A 0.2 (5) C10B—C11B—C12B—C13B 0.3 (5) C16A—O3A—C13A—C14A −6.8 (5) C16B—O3B—C13B—C14B −12.2 (5) C16A—O3A—C13A—C12A 173.7 (3) C16B—O3B—C13B—C12B 168.4 (3) C11A—C12A—C13A—O3A 179.8 (3) C11B—C12B—C13B—O3B 178.9 (3) C11A—C12A—C13A—C14A 0.3 (5) C11B—C12B—C13B—C14B −0.5 (5) O3A—C13A—C14A—C15A 179.5 (3) O3B—C13B—C14B—C15B −178.6 (3) C12A—C13A—C14A—C15A −1.0 (5) C12B—C13B—C14B—C15B 0.8 (5) C13A—C14A—C15A—C10A 1.4 (5) C13B—C14B—C15B—C10B −0.7 (5) C11A—C10A—C15A—C14A −0.9 (5) C11B—C10B—C15B—C14B 0.5 (5) C9A—C10A—C15A—C14A −179.4 (3) C9B—C10B—C15B—C14B −179.2 (3) C13A—O3A—C16A—C17A 175.0 (3) C13B—O3B—C16B—C17B 177.5 (3) O3A—C16A—C17A—C18A 176.8 (3) O3B—C16B—C17B—C18B 176.5 (3) C16A—C17A—C18A—C19A −176.9 (3) C16B—C17B—C18B—C19B −178.8 (3) C17A—C18A—C19A—C20A −176.2 (3) C17B—C18B—C19B—C20B −173.5 (3) C18A—C19A—C20A—C21A −179.7 (3) C18B—C19B—C20B—C21B −176.4 (3) C19A—C20A—C21A—C22A −174.4 (3) C19B—C20B—C21B—C22B −174.4 (3) C20A—C21A—C22A—C23A 175.3 (3) C20B—C21B—C22B—C23B 177.0 (3) C21A—C22A—C23A—C24A −177.5 (3) C21B—C22B—C23B—C24B −178.3 (3) C22A—C23A—C24A—C25A 176.3 (3) C22B—C23B—C24B—C25B 173.9 (3) C23A—C24A—C25A—C26A 179.6 (3) C23B—C24B—C25B—C26B 178.6 (3) C24A—C25A—C26A—C27A −179.5 (3) C24B—C25B—C26B—C27B −174.1 (3) Hydrogen-bond geometry (Å, °) D—H···A D—H H···A D···A D—H···A O1A—H1A···O2A 0.82 1.79 2.513 (4) 146 O1B—H1B···O2B 0.82 1.81 2.530 (4) 146 C22B—H22C···Cg1i 0.97 2.77 3.654 (4) 151 C16B—H16D···Cg2i 0.97 3.00 3.736 (4) 134 C17B—H17D···Cg2 0.97 2.82 3.612 (4) 139 C22A—H22B···Cg3 0.97 2.93 3.765 (4) 145 Symmetry codes: (i) x−1, y, z. sup-12 supplementary materials Fig. 1 Fig. 1 Fig. 1 sup-13 supplementary materials Fig. 2 Fig. 2 sup-14
https://openalex.org/W3007702218	https://europepmc.org/articles/pmc7034851?pdf=render	English	null	Adaptive trust calibration for human-AI collaboration	PloS one	2,020	cc-by	11,045	a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 RESEARCH ARTICLE OPEN ACCESS OPEN ACCESS Citation: Okamura K, Yamada S (2020) Adaptive trust calibration for human-AI collaboration. PLoS ONE 15(2): e0229132. https://doi.org/10.1371/ journal.pone.0229132 Citation: Okamura K, Yamada S (2020) Adaptive trust calibration for human-AI collaboration. PLoS ONE 15(2): e0229132. https://doi.org/10.1371/ journal.pone.0229132 Citation: Okamura K, Yamada S (2020) Adaptive trust calibration for human-AI collaboration. PLoS ONE 15(2): e0229132. https://doi.org/10.1371/ journal.pone.0229132 Editor: Chen Lv, Nanyang Technological University, SINGAPORE Received: July 17, 2019 Accepted: January 30, 2020 Published: February 21, 2020 Copyright: © 2020 Okamura, Yamada. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. RESEARCH ARTICLE Adaptive trust calibration for human-AI collaboration Kazuo OkamuraID1, Seiji YamadaID1,2 Kazuo OkamuraI 1 Department of Informatics, School of Multidisciplinary Sciences, The Graduate University for Advanced Studies (SOKENDAI), Tokyo, Japan, 2 Digital Content and Media Sciences Research Division, National Institute of Informatics, Tokyo, Japan ok@nii.ac.jp a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 Abstract Safety and efficiency of human-AI collaboration often depend on how humans could appro- priately calibrate their trust towards the AI agents. Over-trusting the autonomous system sometimes causes serious safety issues. Although many studies focused on the importance of system transparency in keeping proper trust calibration, the research in detecting and mit- igating improper trust calibration remains very limited. To fill these research gaps, we pro- pose a method of adaptive trust calibration that consists of a framework for detecting the inappropriate calibration status by monitoring the user’s reliance behavior and cognitive cues called “trust calibration cues” to prompt the user to reinitiate trust calibration. We evalu- ated our framework and four types of trust calibration cues in an online experiment using a drone simulator. A total of 116 participants performed pothole inspection tasks by using the drone’s automatic inspection, the reliability of which could fluctuate depending upon the weather conditions. The participants needed to decide whether to rely on automatic inspec- tion or to do the inspection manually. The results showed that adaptively presenting simple cues could significantly promote trust calibration during over-trust. Introduction There are growing interests in automation and autonomous AI technologies in many fields of applications. One highly anticipated application is the unmanned autonomous vehicle, such as driverless shuttles that transport people in rural areas and unmanned aerial vehicles for aerial images, deliveries, and also military purposes. Other expanding application areas such as robotics, autonomous web-based systems, and decision aids are changing all aspects of our daily life. Data Availability Statement: All data is available from the figshare service: https://doi.org/10.6084/ m9.figshare.11538792.v1. Funding: SY No. 26118005 Grant-in-Aid for Scientific Research on Innovative Areas (KAKENHI) Japan Society for the Promotion of Science The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Collaboration between human users and autonomous AI agents is always essential as such technologies are never perfect. One key aspect of such collaborations is that users trust the agents. Trust is an attitudinal judgment of the degree to which a user can rely on an agent to achieve their goals under conditions of uncertainty [1]. Successful collaborations between users and agents would require the users to appropriately adjust their level of trust with the actual reliability of the agents. This process is called trust calibration [1, 2]. While an agent’s Competing interests: The authors have declared that no competing interests exist. PLOS ONE \| https://doi.org/10.1371/journal.pone.0229132 February 21, 2020 1 / 20 Adaptive trust calibration for human-AI collaboration reliability changes for various reasons in an environment, the users often fail to calibrate their trust in the agent and end up in the status called over-trust or under-trust. Over-trust is poorly calibrated trust in which the user overestimates the reliability of the agent; it can result in mis- use of an agent that performs outside of its designed capability. Under-trust is poorly cali- brated trust in which the user underestimates the agent’s capability; it can result in disuse of the agent, excessive user workload, and/or deterioration of the total system performance. Poor trust calibration sometimes causes serious safety issues [3, 4]. In the current study, we focused on the problem of over-trust or under-trust and propose a novel method of adaptive trust calibration that consists of a framework for detecting the inap- propriate status of calibration and cognitive cues called “trust calibration cues” (TCCs) to prompt the user to reinitiate trust calibration. Introduction Considering the challenge of measuring trust, given that it is a complex psychological con- struct, our framework focuses on trust as observable choice behavior [5] so that we can esti- mate calibration status by monitoring user behavior. When over-trust or under-trust is detected, a TCC is presented to the user to alert them to recalibrate their trust. Unlike existing studies [6, 7] that emphasize the importance of continuously displaying system information to maintain appropriate trust calibration, our method adaptively presents information that trig- gers only when it is deemed necessary for users to recalibrate their trust. To our knowledge, most of the existing studies on trust calibration have been about how to prevent inappropriate calibration status; very few have investigated how to mitigate it. The initial idea of the proposed method was discussed in the work [8]. In the current study, we defined a framework and con- ducted an online experiment with a web-based drone simulator to evaluate the effectiveness of our method in an over-trust scenario. The participants of the experiment performed a pothole inspection task [9] to check if there were any holes or cracks in the road images from the drone. Participants chose to use the drone’s automatic inspection or to check the road image manually. By observing the participants’ choice behavior, the framework judged the trust cali- bration status and presented TCCs when over-trust was detected. We measured behavioral changes to see if our adaptive method could effectively restore an appropriate status of trust calibration. We found that adaptively presented TCCs significantly affected the choice behavior of the participants. The participants of the group without TCC did not change their behavior when they were in over-trust status, even if the system reliability information was continuously pre- sented. One of the groups with TCCs had a significantly higher value of sensitivity d’ than the group without TCC, showing that adaptive trust calibration was effective in situations where continuous trust calibration was not effective. Despite several limitations in our framework and our experiment, our study demonstrates the effects of adaptively presenting cognitive cues in instances of over-trust. In the following subsections, we review related work on trust calibration and propose our adaptive trust calibration approach. Next, we cover our experiment and the results. Finally, we discuss the results and describe our future work. Measuring trust Trust is a latent construct and cannot be directly measured; measuring trust experimentally is difficult in general. Most previous literature used self-report measures such as subjective rating with questionnaires [18] or rating scales [19, 20]; however, self-report measures are too intru- sive to be viable in applied settings. Physiological and neural measures have been proposed, such as gaze behavior [21], facial and voice tracking [22], heart-rate, and EEG. Although these measures can be used for dynamic tracking of trust, they usually require special hardware. Behavioral measures are useful for applications in real-world situations. Behavior used as trust measures includes choosing manual or automatic tasks [23], choosing an automation level [24], and accepting advice (reliance behavior). Behaviors might not be observable if there is no interaction; nevertheless, behavioral measures are practical and can easily be used as a basis for modeling and prediction [25]. Trust is a latent construct and cannot be directly measured; measuring trust experimentally is difficult in general. Most previous literature used self-report measures such as subjective rating with questionnaires [18] or rating scales [19, 20]; however, self-report measures are too intru- sive to be viable in applied settings. Physiological and neural measures have been proposed, such as gaze behavior [21], facial and voice tracking [22], heart-rate, and EEG. Although these measures can be used for dynamic tracking of trust, they usually require special hardware. Behavioral measures are useful for applications in real-world situations. Behavior used as trust measures includes choosing manual or automatic tasks [23], choosing an automation level [24], and accepting advice (reliance behavior). Behaviors might not be observable if there is no interaction; nevertheless, behavioral measures are practical and can easily be used as a basis for modeling and prediction [25]. Factors influencing trust Extensive research has been conducted to examine the factors that influence a human’s trust in autonomous agents such as automation. Hancock et al. [10] examined factors that affect trust by applying meta-analysis methods to existing empirical studies. They identified three major trust factors: robot-related factors (performance-based and attribute-based), human-related factors (ability-based and human characteristics), and environment-related factors (team col- laboration and task-based factors). They found that the factors related to the robot and its PLOS ONE \| https://doi.org/10.1371/journal.pone.0229132 February 21, 2020 2 / 20 Adaptive trust calibration for human-AI collaboration performance had the greatest current association with trust. Hoff and Bashir [11] proposed a three-layered trust model to categorize factors that influence automation trust: dispositional trust representing an individual’s overall tendency to trust, situational trust based on the exter- nal environment and context-dependent human characteristics, and learned trust, which is knowledge of a system drawn from past experiences or a current interaction. Learned trust is further divided into two types: initial learned trust, trust prior to interacting with a system, and dynamic learned trust, trust formed during an interaction. In this study, we investigate trust calibration by focusing on dynamic learned trust factors and performance-based factors such as agent reliability. System transparency and continuous trust calibration Humans require a user interface that captures the state of an entire system in order to interact appropriately with the agent. Studies such as [2, 12, 13] have emphasized the importance of system transparency in maintaining proper trust calibration. System transparency has been defined as “the quality of an interface pertaining to its ability to afford an operator’s compre- hension about an intelligent agent’s intent, performance, future plans, and reasoning process [14].” This definition is essentially in accordance with the factors that influence a human’s trust. McGuirl and Sarter [15] showed how continually updated system confidence information can improve trust calibration and increase the performance of the human-machine team. Studies on visualizing car uncertainty during autonomous driving [16, 17] have indicated that providing good transparency by constantly presenting the system information helps maintain appropriate trust of the vehicles. Seppet [7] demonstrated that continuous feedback on auto- mation behavior viably promotes calibrated trust and reliance. Most of the studies investigated how to maintain appropriate trust calibration by continu- ously presenting system information to prevent over-trust or under-trust. Very few studies have focused on detecting poor trust calibration or how to recover from over-trust or under- trust swiftly. Our approach: Adaptive trust calibration Once users fall into the categories of over-trust or under-trust, it might not be easy for them to escape them. Calibration can only occur in response to new evidence that may change the users’ prevailing recognition, while no new evidence can be learned without changing the cur- rent behavior first [5]. To solve this dilemma that could perpetuate an inappropriate status of PLOS ONE \| https://doi.org/10.1371/journal.pone.0229132 February 21, 2020 3 / 20 Adaptive trust calibration for human-AI collaboration trust calibration, a new trigger would be necessary in instances of over-trust or under-trust. To realize this idea, we propose a new framework to detect inappropriate trust calibration. trust calibration, a new trigger would be necessary in instances of over-trust or under-trust. To realize this idea, we propose a new framework to detect inappropriate trust calibration. Framework for detecting over-trust and under-trust. We propose a framework for detecting the inappropriate status of the trust calibration with a behavior based approach. Sup- pose a user and an AI agent are jointly working on a set of tasks. The user should decide whether to rely on the agent or do each task manually. In our framework, three parameters, Pauto, Ptrust, and Pman, are defined as follows: • Pauto: Probability that a task done by an agent will be successful. This is called the “reliability of the agent.” • Ptrust: User’s estimation of Pauto. This is a user’s trust in the agent. • Pman: Probability that a task done manually by a user will be successful. This is called the “capability of the user.” Note that “man” means “manual.” ility of the user.” Note that “man” means “manual. Pauto varies depending on the conditions of the agent. Ptrust also changes accordingly and becomes equal to Pauto if trust is appropriately calibrated. Over-trust occurs if Ptrust > Pauto, and under-trust occurs if Ptrust < Pauto. Since measuring Ptrust is difficult, we modified the defi- nitions of over-trust and under-trust by using a third parameter Pman in addition to Ptrust and Pauto as follows: • Over-trust: the user estimates that the agent is better at the task than the user even though the actual reliability of the agent is lower than the user’s capability. Our approach: Adaptive trust calibration ðPtrust > ^PmanÞ ^ ðPman > PautoÞ ð1Þ ð1Þ ð1Þ • Under-trust: the user estimates that they are better at the task than the agent even though the actual reliability of the agent is higher than the user’s capability. • Under-trust: the user estimates that they are better at the task than the agent even though the actual reliability of the agent is higher than the user’s capability. ðPtrust < ^PmanÞ ^ ðPman < PautoÞ ð2Þ ð2Þ ^Pman is a user’s self estimation of Pman, which corresponds to the user’s self-confidence. These two parameters were not clearly distinguished in [8]. Several studies [13, 23, 26] have demon- strated that reliance behavior can be explained by the relationship between a user’s trust in the agents and the user’s self-confidence. When a user decides to rely on an agent, it is reasonable to say that this behavior indicates Ptrust > ^Pman. If the user decides to do the task manually rather than rely on the agent, it indicates Ptrust < ^Pman. Instead of directly measuring Ptrust or ^Pman, the first inequalities of (1) and (2) can be estimated by observing the user’s reliance behavior; thus, the trust calibration status can be detected, if the second inequalities of Pman and Pauto can be estimated. Trust calibration cue as a new trigger. In this study, we explore the novel idea of giving cognitive cues to users when over-trust or under-trust is detected. This cue is expected to trig- ger the user to promptly notice what has been happening in the environment and to calibrate the trust based on the new findings. We call this cognitive cue a “trust calibration cue” (TCC). Visser et. al [27] proposed a design guideline for trust cues, which are information elements Trust calibration cue as a new trigger. In this study, we explore the novel idea of giving cognitive cues to users when over-trust or under-trust is detected. This cue is expected to trig- ger the user to promptly notice what has been happening in the environment and to calibrate the trust based on the new findings. We call this cognitive cue a “trust calibration cue” (TCC). Visser et. al [27] proposed a design guideline for trust cues, which are information elements used to make a trust assessment about an agent. PLOS ONE \| https://doi.org/10.1371/journal.pone.0229132 February 21, 2020 end if end loop The purpose of this method is to adaptively prompt a user to calibrate her/his trust by pre- senting a TCC only when our framework detects over-trust or under-trust by observing the user’s choice behavior. This approach is to mitigate over-trust or under-trust, in contrast with the traditional approach of trying to maintain appropriate trust calibration with continuous system transparency. Hypothesis We expected users to change their choice behavior if TCCs were adaptively presented when the framework detected inappropriate trust calibration. If our method could effectively miti- gate the over-trust or under-trust, the following are hypothesized: [H0] the manual choice rates decrease in cases of over-trust or increase in cases of under-trust. [H1] the users with TCCs perform better than the users without TCCs. [H0] the manual choice rates decrease in cases of over-trust or increase in cases of under-trust. Materials and methods All studies were carried out in accordance with the recommendations of the Ethical Guidelines for Medical and Health Research Involving Human Subjects provided by the Ministry of Edu- cation, Culture, Sports, Science and Technology and Ministry of Health, Labor, and Welfare in Japan with written informed consent from all participants. All participants gave written informed consent in accordance with the Declaration of Helsinki. The protocol was approved by the ethics committee of the National Institute of Informatics. Our approach: Adaptive trust calibration They classified the cues in terms of trust dimensions (intent, performance, process, expressiveness, and origin) and the trust processing stages (perception, comprehension, projection, decision, and execution). Unlike our TCC, Visser et. al [27] proposed a design guideline for trust cues, which are information elements used to make a trust assessment about an agent. They classified the cues in terms of trust dimensions (intent, performance, process, expressiveness, and origin) and the trust processing stages (perception, comprehension, projection, decision, and execution). Unlike our TCC, PLOS ONE \| https://doi.org/10.1371/journal.pone.0229132 February 21, 2020 4 / 20 Adaptive trust calibration for human-AI collaboration their trust cues were to bring the information specific to the dimensions and stages. Komatsu et.al [28] proposed an intuitive notification methodology called “artificial subtle expressions” (ASE). One of the design requirements is “complementary,” which means that notifications should not interfere with the main communication protocol. Cowell et.al [29] discussed the five non-verbal behaviors of an embodied conversational agent. Waytz et.al [30] demonstrated that anthropomorphism increases trust in an autonomous vehicle. Laughert et. al [31] exam- ined three important objectives in effective warnings: attract attention, elicit knowledge, and enable compliance behavior. Based on these pieces of literature, we designed four types of TCCs (Visual/Audio/Verbal/Anthropomorphic) and evaluated them in the experiment. Details will be explained in the next section. Adaptive trust calibration. With the framework and TCCs described above, we propose a method of adaptive trust calibration as follows. (Details of the detection algorithm will be described later). Method Adaptive trust calibration Method Adaptive trust calibration loop Method Adaptive trust calibration loop Observe a user’s reliance behavior on an agent. Evaluate expression (1) and (2) of the framework. if the over-trust or under-trust is detected then Present a TCC to the user. Present a TCC to the user. end if end loop Participants We recruited participants online through a crowdsourcing service provided by Macromill, Inc. Regarding online experiments in general, Crump et al. [32] showed that the data collected online using a web-browser seemed mostly in line with laboratory results, so long as the exper- iment methods were solid. 5 / 20 PLOS ONE \| https://doi.org/10.1371/journal.pone.0229132 February 21, 2020 Adaptive trust calibration for human-AI collaboration 194 participants joined the experiment online. They were between 20 to 69 years old (M = 44.35, SD = 14.10). 96 participants were male and 98 were female. Apparatus and materials We developed a 3D drone simulator based on an open-source JavaScript WebGL library CesiumJS [33] and the Bing Map API [34]. A screenshot of the simulator running on a Chrome browser is shown in Fig 1. Pothole inspection tasks. A route with 30 checkpoints (CKPs) was defined in the simu- lated environment. Each CKP was located in the center of a rectangular area that was to be inspected to see if there were any potholes in it. Out of the 30 CKPs, 10 had potholes in the cor- responding areas while the other 20 did not. CKPs on the route were shown as small yellow circles on the screen. When the drone came close enough to one of the CKPs on the route, a message popped up (Fig 2) in which the drone asked the participants to make a choice. The indicator at the bottom left area of the screen always showed the reliability of the auto- matic pothole inspection (Fig 3). This continuous display helped to increase the system trans- parency in terms of the reliability. If the participants selected the “Auto” button, an automatic-inspection result was shown for three seconds with a road image of the area around the CKP. This feedback information helped the participants understand how well the automatic inspection performed, thereby increasing the system transparency [2, 14]. If the “Manual” button was selected, a road image Fig 1. Online drone simulator. Operating the simulator was relatively easy, with two cursor keys for controlling the direction of the drone and mouse buttons for making choices. Geospatial Information Authority of Japan (https://maps.gsi.go.jp) CC BY 4.0. The image is similar but not identical to the original one used in the experiment due to a copyright reason. Fig 1. Online drone simulator. Operating the simulator was relatively easy, with two cursor keys for controlling the direction of the drone and mouse buttons for making choices. Geospatial Information Authority of Japan (https://maps.gsi.go.jp) CC BY 4.0. The image is similar but not identical to the original one used in the experiment due to a copyright reason. https://doi.org/10.1371/journal.pone.0229132.g001 6 / 20 PLOS ONE \| https://doi.org/10.1371/journal.pone.0229132 February 21, 2020 Adaptive trust calibration for human-AI collaboration Fig 2. Popup message asking the participants for choice. Geospatial Information Authority of Japan (https://maps.gsi.go.jp) CC BY 4.0 The image is similar but not identical to the original one used in the experiment due to a copyright reason. Apparatus and materials https://doi.org/10.1371/journal.pone.0229132.g002 Fig 2. Popup message asking the participants for choice. Geospatial Information Authority of Japan (https://maps.gsi.go.jp) CC BY 4.0 The image is similar but not identical to the original one used in the experiment due to a copyright reason. https://doi.org/10.1371/journal.pone.0229132.g002 https://doi.org/10.1371/journal.pone.0229132.g002 was displayed, and the participants had to make a pothole report manually. Popup windows of both cases are as shown in Fig 4. Trust calibration cues (TCCs) in experiment. We designed and evaluated four types of TCCs (Fig 5) in our experiment. TCC is similar but not identical to the original one used in the experiment due to a copyright reason. Fig 5. Four types of TCCs. The image of the anthro. TCC is similar but not identical to the original one used in the experiment due to a copyright reason Fig 5. Four types of TCCs. The image of the anthro. TCC is similar but not identical to the original one used in the experim reason https://doi.org/10.1371/journal.pone.0229132.g005 Trust calibration cues (TCCs) in experiment. We designed and evaluated four types of TCCs (Fig 5) in our experiment. Trust calibration cues (TCCs) in experiment. We designed and evaluated four types of TCCs (Fig 5) in our experiment. The visual TCC was designed as a red warning sign in the shape of an upside-down triangle, which is considered to be one of the most common alerting signs according to [31, 35]. The audio TCC uses a sound with a frequency that decreases from 400 Hz to 250 Hz and can con- vey that the agent has a low confidence level [28]. Verbal TCC is a tooltip balloon with the warning message “This choice might not be a good idea.” The anthropomorphic TCC is an animated drone image with cartoon-like eyes that show the agent’s state. TCCs were presented to the participants when the framework detected the over-trust status: the audio TCC was played once, and other TCCs were displayed on the screen close to the “Auto” button for two seconds. Fig 3. Reliability indicator at the bottom left area of the screen. (A) Showing a higher reliability. (B) Showing a deteriorated reliability. https://doi.org/10.1371/journal.pone.0229132.g003 Fig 3. Reliability indicator at the bottom left area of the screen. (A) Showing a higher reliability. (B) Showing a deteriorated reliability. https://doi.org/10.1371/journal.pone.0229132.g003 PLOS ONE \| https://doi.org/10.1371/journal.pone.0229132 February 21, 2020 7 / 20 Adaptive trust calibration for human-AI collaboration Fig 4. Popup windows of the pothole inspections. (A) Automatic inspection result window. (B) Manual inspection window. Both images contain potholes as dark brown spots in the upper road areas. Geospatial Information Authority of Japan (https://maps.gsi.go.jp) CC BY 4.0. The images are similar but not identical to the original ones used in the experiment due to a copyright reason. Potholes were artificially rendered as irregular shapes in a dark brown color on a road image in the popup window. Fig 4. Popup windows of the pothole inspections. (A) Automatic inspection result window. (B) Manual inspection window. Both images contain potholes as dark brown spots in the upper road areas. Geospatial Information Authority of Japan (https://maps.gsi.go.jp) CC BY 4.0. The images are similar but not identical to the original ones used in the experiment due to a copyright reason. Potholes were artificially rendered as irregular shapes in a dark brown color on a road image in the popup window. https://doi.org/10.1371/journal.pone.0229132.g004 Fig 5. Four types of TCCs. The image of the anthro. Over-trust detection based on the proposed framework The following algorithm was used to detect over-trust. Note that a simple moving average of three CKP window was used here. Algorithm Over-trust detection Initialize: Total number of check points(CKPs): M = the number of CKPs.; Over-trust flag list: OT[1], . . ., OT[M] are initialized with zero; The number of current CKP: i ( 1; while i ≦M not time-over do if the drone reached a CKP then if choice behavior is AUTO and Pman > Pauto then PLOS ONE \| https://doi.org/10.1371/journal.pone.0229132 February 21, 2020 8 / 20 Adaptive trust calibration for human-AI collaboration OT[i] ( 1; if i ≧3 and (OT[i −2] + OT[i −1]) ≧1 then Over-trust is detected and TCC is presented to the user; end if end if i ( i + 1; end if end while Manipulation of Pauto The parameter Pauto was manipulated to evaluate our method in an over-trust scenario. The performance of the automatic pothole inspection was configured based on signal detection theory(SDT) [36]. The SDT describes the detection of signals in noisy environments. The noise and the signal are represented as two overlapping density distributions. The distance between the two curves represents the sensitivity d0 of the system. In this experiment, the underlying base rate of potholes was 0.3. Under good weather con- ditions, Pauto and the corresponding sensitivity d0 were 90% and 1.8 respectively, indicating a pretty good discriminating ability of the agent. Under bad weather conditions, Pauto dropped to 50% and the corresponding sensitivity d0 became 0.1, meaning the reliability of the auto- matic pothole inspection had greatly deteriorated. Fig 6 illustrates the manipulation of Pauto and its relationship with Ptrust and Pman. Participants were expected to reach at least the 15th CKP in this experiment. Procedure Participants were randomly assigned to one of five groups with the corresponding TCCs: NoTCC group (without TCCs), visual group (with the visual sign TCC), audio group (with the audio TCC), verbal group (with the verbal TCC), and anthro. group (with the anthropomor- phic TCC). The NoTCC group was the control group in this experiment. p g p g p p The experiment was completed online in three phases. In the instruction phase, the par- ticipants were given instructions stating the goal of the experiment was to inspect 22 CKPs out of 30 CKPs on a test route within a time limit. The participants learned they could inspect CKPs by checking the road image manually or by relying on the drone’s automatic pothole inspection capability. They were told that the average success rate of manual pothole inspec- tion was 75% so that they could adjust their initial self-confidence ^Pman accordingly. They also learned that the reliability of the automatic pothole inspection was very high, although it could fluctuate depending on the conditions of the weather and sunshine. At the end of the instructions, the participants were guided to adjust the sound volume level by listening to a 400 Hz beep sound. Next, in the training phase, the participants started to fly the simulated drone in the training mode. They learned how to operate the drone and how to inspect the CKPs with on-screen guides. When the first three CKPs were inspected, the training mode was finished and the main phase was started. The reliability of the drone’s pothole inspection Pauto was artificially manipulated by changing the conditions of the weather and sunshine in the simulated environment. Initially, the weather was good, and Pauto was set to 90%. The fine weather continued until the drone visited six CKPs in the main phase. This period of six CKPs was intended for the participants to calibrate their trust toward the drone with a higher reliability of automatic inspection under the good weather conditions. Immediately after the 6th CKP was inspected, sounds of a thunderstorm began. The visibility of the field also became very low, and the Pauto was decreased from 90% to 50%, which changed the sign of Pman −Pauto. During this period, the participants were expected to over-trust the drone due to carry-over from the previous weather condition. The proposed framework was evaluated in this period. Assumptions The images of the potholes were carefully designed so that the average success rate of manual inspection would be more than 75%. Although machine image recognition has been remark- ably advanced with deep neural networks [37], Geirhos et al. [38] and Dodge et al. [39] demon- strated that human object recognition outperforms the top-performing deep neural networks under image degradation, such as Gaussian blur and additive Gaussian noise. These findings could be applied to the estimation of the inequation of Pman and Pauto in the experiment because the pothole inspection became an image recognition task with dark and foggy road images when the weather conditions turned worse. Therefore, we assumed that Pauto would Fig 6. Relationship between the three parameters under changing weather conditions. https://doi.org/10.1371/journal.pone.0229132.g006 Fig 6. Relationship between the three parameters under changing weather conditions. PLOS ONE \| https://doi.org/10.1371/journal.pone.0229132 February 21, 2020 9 / 20 Adaptive trust calibration for human-AI collaboration fluctuate more widely than Pman under changing weather conditions. On the basis of this assumption, we calculated the inequality relationship between Pauto and Pman in the experi- ment; the inequality Pauto > Pman was true during the good weather period and false during the bad one. PLOS ONE \| https://doi.org/10.1371/journal.pone.0229132 February 21, 2020 Procedure When the participants clicked the “Auto” button for automatic inspection and the framework detected the over-trust status using the over-trust detection algorithm described above, the corresponding TCC was presented at the timing right after the button was selected. The following information was available to the participants during the experiment: • Simulated weather information with visibility changes, brightness changes, and the sounds of a thunderstorm. • Simulated weather information with visibility changes, brightness changes, and the sounds of a thunderstorm. • A reliability indicator to continuously show the reliability of the automatic inspection. • An enlarged photo images of each CKP. • An enlarged photo images of each CKP. • The result of the drone’s automatic inspection when AUTO was chosen. • A TCC when the proposed framework detected over-trust. The experiment ended if the 22nd CKP was inspected or the time limit was reached. We established a time limit of 8.5 minutes (510 seconds) based on pre-trials with this test route, PLOS ONE \| https://doi.org/10.1371/journal.pone.0229132 February 21, 2020 10 / 20 Adaptive trust calibration for human-AI collaboration and we expected a single automatic inspection to take 10 seconds, one manual inspection to take 15 seconds, and reaching the next CKP to take 10 seconds. and we expected a single automatic inspection to take 10 seconds, one manual inspection to take 15 seconds, and reaching the next CKP to take 10 seconds. Dependent measures The dependent variables of interest in this experiment were a TCC presentation rate(hereinaf- ter, called TCC rate), a manual choice rate(hereinafter, called manual rate), sensitivity d0, and accuracy of the task results. TCC rates mean how often the proposed framework detected over-trust. Changes in manual choice rates indicate how the participants changed their behav- iors as a result of trust calibration. Both sensitivity d0 and accuracy indicate the performance of the human-AI collaboration. All keyboard inputs and mouse clicks were recorded and used to calculate these variables. Results One hundred sixteen participants successfully inspected 15 CKPs or more within the time limit. Seventy eight participants unintentionally moved the drone far from the area where the CKPs were located, and they failed to complete the tasks within the time limit. As for the suc- cessful participants, their ages ranged from 20 to 69 years old (M = 43.25, SD = 14.01), 66 par- ticipants were male and 50 were females. 28 were in the NoTCC group, 18 in the visual group, 22 in the audio group, 29 in the verbal group, and 19 in the anthro. group. They inspected the total of 1,740 CKPs from the 1st CKP to the 15th CKP, and the results of 1,282 inspections were correct, making the correct answer rate 0.74. Automatic inspection was selected 1,236 times (the choice rate = 0.71). The participants did the manual inspection 504 times (the choice rate = 0.29). Table 1 shows the TCC rates at each CKP. Note that TCCs were not pre- sented in the period from the 7th CKP to the 9th CKP, since the sliding window of three CKPs was used in the detection algorithm. Means and standard errors of the other dependent mea- sures can be found in Table 2. Hereinafter, we call the period from the 1st visited CKP to 6th Table 1. Means of TCC rates at each CKP. group CKP9 CKP10 CKP11 CKP12 CKP13 CKP14 CKP15 Visual TCC 0.78 (0.10) 0.67 (0.11) 0.56 (0.12) 0.67 (0.11) 0.67 (0.11) 0.50 (0.12) 0.56 (0.12) Audio TCC 0.55 (0.11) 0.64 (0.10) 0.45 (0.11) 0.50 (0.11) 0.50 (0.11) 0.50 (0.11) 0.50 (0.11) Verbal TCC 0.48 (0.09) 0.45 (0.09) 0.28 (0.08) 0.31 (0.09) 0.34 (0.09) 0.38 (0.09) 0.07 (0.05) Anthro TCC 0.53 (0.11) 0.47 (0.11) 0.37 (0.11) 0.53 (0.11) 0.47 (0.11) 0.74 (0.10) 0.63 (0.11) Standard errors in parentheses. Table 1. Means of TCC rates at each CKP. group CKP9 CKP10 CKP11 CKP12 CKP13 CKP14 CKP15 Visual TCC 0.78 (0.10) 0.67 (0.11) 0.56 (0.12) 0.67 (0.11) 0.67 (0.11) 0.50 (0.12) 0.56 (0.12) Audio TCC 0.55 (0.11) 0.64 (0.10) 0.45 (0.11) 0.50 (0.11) 0.50 (0.11) 0.50 (0.11) 0.50 (0.11) Verbal TCC 0.48 (0.09) 0.45 (0.09) 0.28 (0.08) 0.31 (0.09) 0.34 (0.09) 0.38 (0.09) 0.07 (0.05) Anthro TCC 0.53 (0.11) 0.47 (0.11) 0.37 (0.11) 0.53 (0.11) 0.47 (0.11) 0.74 (0.10) 0.63 (0.11) Standard errors in parentheses. https://doi.org/10.1371/journal.pone.0229132.t001 Table 2. Means of the other dependent measures. “Good weather” means the good weather period and “Bad weather” means the bad weather period. Standard errors in parentheses. p “Good weather” means the good weather period and “Bad weather” means the bad weather period. https://doi.org/10.1371/journal.pone.0229132.t002 https://doi.org/10.1371/journal.pone.0229132.t002 Manual rates TCCs were presented multiple times per participant in most cases. The effects of presenting TCCs might be accumulated and did not always appear immediately after presentation. In the current study, we evaluated the TCC effects by comparing the six-CKP mean values of the manual rates both for the good and bad weather periods, so that we could also capture the accumulated effects in each period. We conducted a two factor mixed ANOVA with the TCC groups (NoTCC, visual, audio, verbal, and anthro.) as between subjects and CKP periods (the good weather period and the bad weather period) as within subjects. The analysis revealed a significant main effect for the CKP periods [F(1, 111) = 51.69, p < 0.01, Z2 p ¼ 0:32]. The partic- ipants changed their choice behavior as the weather conditions deteriorated. A significant interaction was found between the two factors [F(4, 111) = 4.86, p < 0.01, Z2 p ¼ 0:15]. In the good weather period, there was no simple effect for the TCC groups, meaning that the manual rates of each TCC group were not significantly different from each other [Fð4; 111Þ ¼ 0:47; p ¼ 0:76; Z2 p ¼ 0:02]. The NoTCC group did not show a simple effect for the CKP periods [Fð1; 27Þ ¼ 1:23; p ¼ 0:28; Z2 p ¼ 0:04] indicating the manual rates of the NoTCC group were not significantly different between the two CKP periods. In contrast with this, all of the groups with TCCs showed significantly higher manual rates in the bad weather period than in the good weather period [Visual TCC, F(1, 17) = 9.20, p < 0.01, Z2 p ¼ 0:35; Audio TCC, F(1, 21) = 5.54, p = 0.03, Z2 p ¼ 0:21; Verbal TCC, F(1, 28) = 62.9, p < 0.001, Z2 p ¼ 0:69; Anthro TCC, F(1, 18) = 8.55, p < 0.01, Z2 p ¼ 0:32]. Fig 7 shows how the manual rates changed over two CKP periods. Holm–Bonferroni-adjusted post hoc comparisons were also conducted to investigate the effects of TCCs. For the bad weather period, the verbal group showed a significantly higher manual rate than both the NoTCC group (t(111) = 4.77, adj.p < 0.01) and the anthro. group (t(111) = 2.74, adj.p = 0.04). No other differences between the groups were found to be signifi- cant. Although the effectiveness among TCCs differ, these results support H0. TCC rates TCC rates in the verbal TCC group were higher in the early part of the period and gradually decreased. This indicates that over-trust decreased during this period. The visual and audio TCC groups also showed a similar trend, while TCC rates in the anthro. TCC group did not follow the decreasing trend. Results Group Manual rate Sensitivity d’ Accuracy Good weather Bad weather Good weather Bad weather Good weather Bad weather NoTCC 0.15 (0.04) 0.22 (0.06) 1.36 (0.09) 0.38 (0.14) 0.86 (0.04) 0.60 (0.04) Visual 0.11 (0.06) 0.37 (0.09) 1.35 (0.13) 0.52 (0.15) 0.87 (0.05) 0.62 (0.04) Audio 0.20 (0.06) 0.42 (0.08) 1.62 (0.10) 0.61 (0.15) 0.94 (0.03) 0.65 (0.04) Verbal 0.17 (0.05) 0.63 (0.04) 1.39 (0.10) 0.92 (0.14) 0.87 (0.04) 0.72 (0.04) Anthro. 0.20 (0.06) 0.37 (0.07) 1.40 (0.15) 0.54 (0.15) 0.88 (0.05) 0.62 (0.04) Standard errors in parentheses. Table 1. Means of TCC rates at each CKP. https://doi.org/10.1371/journal.pone.0229132.t001 https://doi.org/10.1371/journal.pone.0229132.t001 Table 2. Means of the other dependent measures. Group Manual rate Sensitivity d’ Accuracy Good weather Bad weather Good weather Bad weather Good weather Bad weather NoTCC 0.15 (0.04) 0.22 (0.06) 1.36 (0.09) 0.38 (0.14) 0.86 (0.04) 0.60 (0.04) Visual 0.11 (0.06) 0.37 (0.09) 1.35 (0.13) 0.52 (0.15) 0.87 (0.05) 0.62 (0.04) Audio 0.20 (0.06) 0.42 (0.08) 1.62 (0.10) 0.61 (0.15) 0.94 (0.03) 0.65 (0.04) Verbal 0.17 (0.05) 0.63 (0.04) 1.39 (0.10) 0.92 (0.14) 0.87 (0.04) 0.72 (0.04) Anthro. 0.20 (0.06) 0.37 (0.07) 1.40 (0.15) 0.54 (0.15) 0.88 (0.05) 0.62 (0.04) Table 2. Means of the other dependent measures. PLOS ONE \| https://doi.org/10.1371/journal.pone.0229132 February 21, 2020 11 / 20 Adaptive trust calibration for human-AI collaboration CKP “the good weather period” and the period with possible TCC presentations from the 10th CKP to 15th CKP “the bad weather period”. CKP “the good weather period” and the period with possible TCC presentations from the 10th CKP to 15th CKP “the bad weather period”. PLOS ONE \| https://doi.org/10.1371/journal.pone.0229132 February 21, 2020 Adaptive trust calibration for human-AI collaboration Fig 7. Manual rates over time. https://doi.org/10.1371/journal.pone.0229132.g007 Table 3. 3-CKP mean values of Pman. CKPs Good weather period Bad weather period 1 to 3 4 to 6 7 to 9 10 to 12 13 to 15 Pman 0.81 (0.05) 0.80 (0.06) 0.85 (0.04) 0.86 (0.03) 0.90 (0.03) Standard errors in parentheses. https://doi.org/10.1371/journal.pone.0229132.t003 Fig 7. Manual rates over time. https://doi.org/10.1371/journal.pone.0229132.g007 Fig 7. Manual rates over time. https://doi.org/10.1371/journal.pone.0229132.g007 Fig 7. Manual rates over time. Fig 7. Manual rates over time. Table 3. 3-CKP mean values of Pman. CKPs Good weather period Bad weather period 1 to 3 4 to 6 7 to 9 10 to 12 13 to 15 Pman 0.81 (0.05) 0.80 (0.06) 0.85 (0.04) 0.86 (0.03) 0.90 (0.03) Standard errors in parentheses. https://doi.org/10.1371/journal.pone.0229132.t003 https://doi.org/10.1371/journal.pone.0229132.t003 among the five groups[Fð4; 111Þ ¼ 1:62; p ¼ 0:18; Z2 p ¼ 0:06]. Hypothesis H1 is not sup- ported regarding accuracy; however, the verbal group showed the highest mean value (0.72 (SE 0.04)), and other groups with TCCs also had better accuracy values (0.6 (SE 0.04)) than the NoTCC group. Regarding the accuracy of the manual inspections which corresponds to Pman, Table 3 shows the 3-CKP mean values of Pman. Although the mean values of Pman slightly increased, a Welch’s t-test indicated that there was no significant difference between Pman in the good weather period and in the bad one [t(114) = −1.08, p = 0.28, Cohen’s d = 0.16]. This result indi- cates that Pman did not degrade under the change in weather conditions. One-sample t-tests showed that Pman in the good weather period was significantly smaller than Pauto = 0.90 [Mean = 0.81, t(71) = −2.17, p = 0.03, Cohen’s d = 0.26] and that Pman in the bad weather period was significantly larger than Pauto = 0.50 [Mean = 0.86, t(200) = 17.79, p < 0.01, Cohen’s d = 1.25]. Performance: Sensitivity d0 and accuracy The same ANOVA with the TCC groups and the CKP periods revealed that the sensitivity d0 in the bad weather period was significantly lower than in the good weather period [Fð1; 111Þ ¼ 107:22; p < 0:01; Z2 p ¼ 0:49]. In the bad weather period, post-hoc comparisons indicated that the sensitivity d0 of the verbal group, which was the highest (0.92 (SE 0.14) among all the groups, was significantly higher than the sensitivity d0 of the NoTCC group [t(111) = 2.97, adj.p = 0.04, Cohen’s d = 0.75]. In terms of sensitivity d0, hypothesis H1 is supported. Accuracy, the rate of the correct inspection, also significantly declined in the bad weather period[Fð1; 111Þ ¼ 87:58; p < 0:01; Z2 p ¼ 0:46], but there was no significant difference PLOS ONE \| https://doi.org/10.1371/journal.pone.0229132 February 21, 2020 12 / 20 Effects of TCCs to change the participants’ behavior The groups with visual, verbal, and anthro. TCCs showed significantly higher manual rates for the bad weather period than for the good weather period (Fig 7). The audio TCC that used an audio ASE that decreased in frequency seemed promising, as it could convey the low level of confidence in the system [28]. The result, however, showed that its manual choice rate was not significantly larger than the one of NoTCc group. One possible reason is that other sounds were also being played in the simulated environment, such as the drone flying and the thun- derstorm, which may have reduced the effect of this audio TCC. The effect of the anthro. TCC was not as large as we originally expected; it was significantly smaller than the effect of the ver- bal TCC (Fig 8). The manual choice rate of the visual TCC were also smaller than other TCCs. It was obvious that the participants recognized these visually impressive TCCs; however, the results suggested that just being salient on the screen was not enough for some of the partici- pants to change their choice behavior [31, 40]. The verbal group had the highest manual rate (Fig 8). Only the verbal TCC referred to the purpose by using the word “choice,” while the other TCCs were implemented just as a caution or warning. Reading this word might have helped the participants proceed more easily to the latter stages of the trust calibration process. Based on these results, our tentative guideline for designing TCCs is that TCCs should be reasonably noticeable in the task environment and should contain connotations that can link Based on these results, our tentative guideline for designing TCCs is that TCCs should be reasonably noticeable in the task environment and should contain connotations that can link the user to the next possible actions in the collaborative task. If the participants wanted to complete the scan tasks quickly, they could have used the auto- matic inspection, which was faster. However, the participants promptly increased the manual choices after recognizing TCCs despite the longer completion time. This result indicates that the possible automation bias caused by the difference in the task completion times did not crit- ically impact the decision-making of the participants in the experiment. Fig 8. Manual rates for each TCC during the bad weather period. https://doi.org/10.1371/journal.pone.0229132.g008 Fig 8. Manual rates for each TCC during the bad weather period. Discussion However, the participants promptly increased the manual choices after recognizing TCCs despite the longer completion time. This result indicates that the possible automation bias caused by the difference in the task completion times did not crit- ically impact the decision-making of the participants in the experiment. Fig 8. Manual rates for each TCC during the bad weather period. https://doi.org/10.1371/journal.pone.0229132.g008 https://doi.org/10.1371/journal.pone.0229132 February 21, 2020 14 / 20 Discussion The overall results demonstrated that adaptively presenting TCCs strongly affected whether the choice behavior of the participants would change, while continuously presenting the reli- ability information did not help the participants change their bias to rely on the automation. PLOS ONE \| https://doi.org/10.1371/journal.pone.0229132 February 21, 2020 13 / 20 Adaptive trust calibration for human-AI collaboration The better task performances were also achieved with the behavior changes triggered by TCCs, whose presentation timing was decided by the proposed framework. whose presentation timing was decided by the proposed framework. Effects of TCCs to change the participants’ behavior The groups with visual, verbal, and anthro. TCCs showed significantly higher manual rates for the bad weather period than for the good weather period (Fig 7). The audio TCC that used an audio ASE that decreased in frequency seemed promising, as it could convey the low level of confidence in the system [28]. The result, however, showed that its manual choice rate was not significantly larger than the one of NoTCc group. One possible reason is that other sounds were also being played in the simulated environment, such as the drone flying and the thun- derstorm, which may have reduced the effect of this audio TCC. The effect of the anthro. TCC was not as large as we originally expected; it was significantly smaller than the effect of the ver- bal TCC (Fig 8). The manual choice rate of the visual TCC were also smaller than other TCCs. It was obvious that the participants recognized these visually impressive TCCs; however, the results suggested that just being salient on the screen was not enough for some of the partici- pants to change their choice behavior [31, 40]. The verbal group had the highest manual rate (Fig 8). Only the verbal TCC referred to the purpose by using the word “choice,” while the other TCCs were implemented just as a caution or warning. Reading this word might have helped the participants proceed more easily to the latter stages of the trust calibration process. Based on these results, our tentative guideline for designing TCCs is that TCCs should be reasonably noticeable in the task environment and should contain connotations that can link the user to the next possible actions in the collaborative task. If the participants wanted to complete the scan tasks quickly, they could have used the auto- matic inspection, which was faster. Adaptive method vs. continuous method The manual rate of the NoTCC group did not significantly change over the two periods. When the participants were exposed to the bad weather, the weather change was made very notice- able with the screen visibility and the sound effects. The reliability indicator showed a big per- formance degradation of the system due to the poor visibility. Nevertheless, the participants of the NoTCC group continued to rely on the drone’s automatic pothole inspection, which had less reliability than the actual manual success rate. Thus, the participants over-trusted the auto- matic inspection despite the system information indicating the reliability becoming worse. This result is not in line with the previous studies [6, 7, 15] that emphasized the effectiveness of the continuous trust calibration with system transparency. A possible explanation for the result could be made by discussing models for the trust process [1, 27, 41]. Miring et al.[42] defined a model with four stages: perception, understanding, prediction, and adaption. Although the reliability indicator continuously displayed the deterioration of the reliability, the participants in the NoTCC group might not fully acquire the knowledge to move beyond the perception stage. They would have behave differently if the experiment had continued longer enough for them to understand the relationship between the indicator change and the performance of the system. In contrast to this, the participants in other groups with TCCs could successfully reached the adaptation stage and change their behaviors in this experiment. We believe that the results demonstrated the effectiveness of the adaptive method. TCCs were given right after the behavior only if the participants were judged to over-trust, so that it would be easier for the participants to understand the implication of the cues and to move forward in the trust calibra- tion process. The manual rate of the NoTCC group did not significantly change over the two periods. When the participants were exposed to the bad weather, the weather change was made very notice- able with the screen visibility and the sound effects. The reliability indicator showed a big per- formance degradation of the system due to the poor visibility. Nevertheless, the participants of the NoTCC group continued to rely on the drone’s automatic pothole inspection, which had less reliability than the actual manual success rate. Thus, the participants over-trusted the auto- matic inspection despite the system information indicating the reliability becoming worse. Performance The results of the manual accuracy were consistent with our assumption to estimate Pman −Pauto in this experiment. While the mean values of sensitivity d0 in each group were dropped in the bad weather period, the verbal group showed a significantly higher sensitivity d0 than the NoTCC group, and three other groups with TCCs also had better values than the NoTCC group. The results showed that all the groups with TCCs showed higher discriminating performance in the bad weather period than the NoTCC group. Though there was no statistical difference in accuracy among the groups, all the groups with TCCs also showed better accuracy than the NoTCC group. These results indicated that adaptively presenting TCCs promoted appropriate trust cali- bration leading to the better performance in the bad weather period. Effects of TCCs to change the participants’ behavior https://doi.org/10.1371/journal.pone.0229132.g008 Fig 8. Manual rates for each TCC during the bad weather period. https://doi.org/10.1371/journal.pone.0229132.g008 https://doi.org/10.1371/journal.pone.0229132.g008 PLOS ONE \| https://doi.org/10.1371/journal.pone.0229132 February 21, 2020 14 / 20 Adaptive trust calibration for human-AI collaboration These results indicated that adaptively presenting TCCs strongly affected the choice behav- ior of the participants who otherwise failed to find opportunities to change their tendency to rely on the automation. Limitations and future research The several limitations of our study suggest the need for further experiments and future research. The current framework focuses on performance-related factors to detect over-trust and under-trust. The capabilities of humans and agents are compared to identify the possible choice behavior that might lead to a better performance. However, automation could be beneficial beyond a better performance, such as for faster task completion, lighter workloads, fewer risks, etc. [3] Therefore, trust as the observable and rational choice behavior can be a product of believ- ing that these benefits will outweigh the costs [5]. For example, Naujoks et.al [43] discussed the desire to engage the non-driving-related tasks during autonomous driving, which requires the workload of the driving task to be lighter and makes the driver select the autonomous mode. Our framework could be integrated with such factors by adjusting the effect of the term Pman with a coefficient γ, which represents a degree of human intention to do the main task manually. P0 man ¼ g Pman ð3Þ ð3Þ P0 man ¼ g Pman For example, if human would not do a task manually because they want to continue a sec- ondary task, a value of γ < 1 indicates a negative bias toward the manual choice. Similarly, a value of γ > 1 represents a positive bias in doing a task manually. γ can be decided in a practical manner by experimenting with several levels of pre-defined factors that influence auto-manual choice decisions. Dual-task paradigm [44] could be applied to determine the values of γ. The feedback information given for each inspection was very important for the participants to make decisions. The pothole inspection task in the experiment is a remote sensing task, and it would be quite difficult for the system to know the correct answer (ground truth) at the time of each inspection in practical situations because the only information available is the image data and the results of automatic recognition. Therefore the correct answer for each inspection was not presented to the participants in the experiment. The result of the automatic inspection was shown to the participants when they selected automatic inspection, not when they did the inspections manually. Although this was to simplify the conditions and focus on evaluating the effect of presenting TCCs, further study should consider possible combinations of feedback information and evaluate their effects. Applicability in real-world situations Although providing a model to estimate the second inequalities of Pman and Pauto in the pro- posed framework is beyond the scope of this paper, we believe that they could be estimated as follows. Pauto, which represents the reliability of an AI agent, could be calculated with the sen- sor models and algorithms used to implement the AI agent. Pman, which is a human capability index, could be estimated by using the parameters of a target task and environmental condi- tions. The results of the previous studies [38, 39] are such examples that provide a basis for esti- mating the second inequalities. If an appropriate estimation model for Pman is not available, PLOS ONE \| https://doi.org/10.1371/journal.pone.0229132 February 21, 2020 15 / 20 Adaptive trust calibration for human-AI collaboration trial operations can be performed to collect the necessary data to estimate Pman empirically. In practical situations, it is quite common for users of a system to practice how to operate the sys- tem in advance. The second inequalities could be estimated even in a real-time situation. The first inequalities in the proposed framework could be estimated by observing users’ choice behaviors, without measuring Ptrust and ^Pman directly. Although a pop-up dialogue was used to observe the behaviors in the current experiment, continuous measurements of the behaviors could also be used with the proposed framework. For example, a driver’s intention to use auto- matic driving could be inferred with a touch sensor on a steering wheel to check if the driver’s hands are on the wheel. Similarly, a switch button to turn automation on and off at any time could provide necessary information on humans’ reliance on the automation. The first inequality in the framework could be calculated with the information from these continuous methods that could work well with real-time tasks. Therefore, we believe that our proposed framework can be applied to real-time applications that require human-agent collaboration. PLOS ONE \| https://doi.org/10.1371/journal.pone.0229132 February 21, 2020 Supporting information S1 File. Complete data set. https://doi.org/10.6084/m9.figshare.11538792.v1. (XLSX) Limitations and future research In this experiment, we focused on evaluating an over-trust case, which often has more seri- ous adverse effects in actual situations [45, 46]. Experiments are currently being prepared with bi-directional trust change scenarios to evaluate cases of over-trust and under-trust. The PLOS ONE \| https://doi.org/10.1371/journal.pone.0229132 February 21, 2020 16 / 20 Adaptive trust calibration for human-AI collaboration proposed method deals with users’ behaviors to choose Auto or Manual and does not guaran- tee the convergence of the behaviors, which is the result of trust calibration by users. Future research could investigate a better way to develop the algorithm in light of control theory, where the trust status would be represented as a non-binary variable, and TCCs could also be presented in a different way. We used a pothole inspection task in the experiment, which is often categorized as a recon- naissance task in the trust research literature [13]. Future research should explore different types of tasks, such as autonomous driving, decision aids, and interactive games. Further experiments should be to evaluate our approach with the different types of TCCs to investigate the requirements of effective cues. The current study mainly dealt with dynamic learned trust [11]. Future studies should investigate other factors of trust such as dispositional trust [1] and situational trust to gain a deeper understanding of trust calibration. Conclusion To our knowledge, this is the first study to show how to adaptively prompt the users to cali- brate their trust toward an automated agent based on their trust calibration status. Previous studies emphasized the importance of the system transparency for proper trust calibration. Our results indicated that they are not always sufficient to recover from over-trust, and our method of adaptive trust calibration significantly helped the participants change their behavior and recover from the over-trust. Despite several limitations, this study has demonstrated the effectiveness of presenting cognitive cues at the time of over-trust. We strongly believe that the findings of this study will contribute to better user interface designs for collaborative systems with autonomous AI agents. Author Contributions Conceptualization: Kazuo Okamura, Seiji Yamada. Formal analysis: Kazuo Okamura. Investigation: Kazuo Okamura. Methodology: Kazuo Okamura, Seiji Yamada. Project administration: Kazuo Okamura. References 1. Lee JD, See KA. Trust in automation: Designing for appropriate reliance. Human Factors. 2004; 46(1):50–80. https://doi.org/10.1518/hfes.46.1.50_30392 PMID: 15151155 2. Muir BM. Trust in automation: Part I. Theoretical issues in the study of trust and human intervention in automated systems. Ergonomics. 1994; 37(11):1905–1922. https://doi.org/10.1080/ 00140139408964957. 3. Parasuraman R, Riley Victor. Humans and automation: Use, misuse, disuse, abuse. Human Factors. 1997; 39(2):230–253. https://doi.org/10.1518/001872097778543886. 4. Robinette P, Li W, Allen R, Howard AM, Wagner AR. Overtrust of robots in emergency evacuation sce- narios. In: Proceedings of the ACM/IEEE International Conference on Human-Robot Interaction; 2016. p. 101–108. https://doi.org/10.1109/HRI.2016.7451740. 5. Adams BD, Bruyn LE, Houde S. Trust in automated systems literature review. Defence Research and Development Canada; 2003. Report No. CR-2003-096. 6. Merritt SM, Lachapell-unnerstall J, Louis S. Continuous calibration of trust in automated systems. Air Force Research Laboratory Technical Report; 2014. AFRL-RH-WP-TR-2014-0026. 7. Seppelt BD. Supporting operator reliance on automation through continuous feedback [dissertation]. University of Iowa; 2009. https://doi.org/10.17077/etd.90t9uqnh. 8. Okamura K, Yamada S. Adaptive trust calibration for supervised autonomous vehicles. In: Adjunct Pro- ceedings of the 10th International ACM Conference on Automotive User Interfaces and Interactive Vehicular Applications; 2018. p. 92–97. https://doi.org/10.1145/3239092.3265948. 9. Lee SW, Kim S, Han J, An KE, Ryu SK, Seo D. Experiment of image processing algorithm for efficient pothole detection. In: Proceedings of the 2019 IEEE International Conference on Consumer Electronics (ICCE). IEEE; 2018. p. 1–2. https://doi.org/10.1109/ICCE.2019.8661973. 10. Hancock PA, Billings DR, Schaefer KE, Chen JYC, De Visser EJ, Parasuraman R. A meta-analysis of factors affecting trust in human-robot interaction. Human Factors. 2011; 53(5):517–527. https://doi.org/ 10.1177/0018720811417254.PMID: 22046724 11. Hoff K, Bashir M. Trust in automation: integrating empirical evidence on factors that influence trust. Human Factors. 2015; 57(3):407–434. https://doi.org/10.1109/MIS.2013.24. PMID: 25875432 12. Mercado JE, Rupp MA, Chen JYC, Barnes MJ, Barber D, Procci K. Intelligent agent transparency in human–agent teaming for multi-UxV management. Human Factors. 2016; 58(3):401–415. https://doi. org/10.1177/0018720815621206 PMID: 26867556 13. Yang XJ, Unhelkar VV, Li K, Shah JA. Evaluating effects of user experience and system transparency on trust in automation. In: Proceedings of the ACM/IEEE International Conference on Human-Robot Interaction; 2017. p. 408–416. https://doi.org/10.1145/2909824.3020230. 14. Chen JYC, Barnes MJ. Human—Agent teaming for multirobot control: A review of human factors issues. IEEE Transactions on Human-Machine Systems. 2014; 44(1):13–29. https://doi.org/10.1109/ THMS.2013.2293535. 15. McGuirl JM, Sarter NB. Supporting trust calibration and the effective use of decision aids by presenting dynamic system confidence information. Human Factors. 2006; 48(4):656–665. https://doi.org/10. 1518/001872006779166334. Project administration: Kazuo Okamura. Resources: Kazuo Okamura. Software: Kazuo Okamura. Supervision: Kazuo Okamura, Seiji Yamada. Validation: Kazuo Okamura. Visualization: Kazuo Okamura. Writing – original draft: Kazuo Okamura. Writing – review & editing: Kazuo Okamura. 17 / 20 PLOS ONE \| https://doi.org/10.1371/journal.pone.0229132 February 21, 2020 PLOS ONE \| https://doi.org/10.1371/journal.pone.0229132 February 21, 2020 17 / 20 Adaptive trust calibration for human-AI collaboration References PMID: 17240714 16. Helldin T, Falkman G, Riveiro M, Davidsson S. Presenting system uncertainty in automotive UIs for sup- porting trust calibration in autonomous driving. In: Proceedings of the 5th International Conference on Automotive User Interfaces and Interactive Vehicular Applications; 2013. p. 210–217. https://doi.org/ 10.1145/2516540.2516554. 17. Haeuslschmid R, Buelow MV, Pfleging B, Butz A. Supporting trust in autonomous driving. In: Proceed- ings of the 22nd International Conference on Intelligent User Interfaces; 2017. p. 319–329. https://doi. org/10.1145/3025171.3025198. 18. Jian Jy, Bisantz AM, Drury CG. Foundations for an empirically determined scale of trust in automated systems. International Journal of Cognitive Ergonomics. 2000; 4(1):53–71. https://doi.org/10.1207/ S15327566IJCE0401. 19. Brown RD, Galster SM. Effects of reliable and unreliable automation on subjective measures of mental workload, situation awareness, trust, and confidence in a dynamic flight task. In: Proceedings of the 48th Annual Meeting of Human Factors and Ergonomics Society; 2004. p. 147–151. https://doi.org/10. 1177/154193120404800132. 20. Master R, Jiang X, Khasawneh MT, Bowling SR, Grimes L, Gramopadhye AK, et al. Measurement of trust over time in hybrid inspection systems. Human Factors and Ergonomics in Manufacturing. 2005; 15(2):177–196. https://doi.org/10.1002/hfm.20021. 18 / 20 PLOS ONE \| https://doi.org/10.1371/journal.pone.0229132 February 21, 2020 Adaptive trust calibration for human-AI collaboration 21. Hergeth S, Lorenz L, Vilimek R, Krems JF. Keep your scanners peeled: Gaze behavior as a measure of automation trust during highly automated driving. Human Factors. 2016; 58(3):509–519. https://doi.org/ 10.1177/0018720815625744. PMID: 26843570 22. Nooralishahi P, Chu Kiong L, Chin AV, Khalid HM, Wei Shiung L, Rasool Z, et al. Exploring psycho- physiological correlates to trust: Implications for human-robot-human interaction. In: Proceedings of the Human Factors and Ergonomics Society 2016 Annual Meeting; 2016. p. 697–701. https://doi.org/10. 1177/1541931213601160. 23. de Vries P, Midden C, Bouwhuis D. The effects of errors on system trust, self-confidence, and the allo- cation of control in route planning. International Journal of Human Computer Studies. 2003; 58(6):719– 735. https://doi.org/10.1016/S1071-5819(03)00039-9 24. Desai M, Kaniarasu P, Medvedev M, Steinfeld A, Yanco H. Impact of robot failures and feedback on real-time trust. In: Proceedings of the ACM/IEEE International Conference on Human-Robot Interac- tion; 2013. p. 251–258. https://doi.org/10.1109/HRI.2013.6483596. 25. Drnec K, Marathe AR, Lukos JR, Metcalfe JS, Gramann K, Wykowska A, et al. From trust in automation to decision neuroscience: applying cognitive neuroscience methods to understand and improve interac- tion decisions involved in human automation interaction. Frontires In Human Neuroscience. 2016; 10. https://doi.org/10.3389/fnhum.2016.00290. 26. Gac J, Lee JD. References Extending the decision field theory to model operators’ reliance on automation in super- visory control situations. IEEE Transactions on Systems, Man, and Cybernetics Part A: Systems and Humans. 2006; 36(5):943–959. https://doi.org/10.1109/TSMCA.2005.855783. 27. de Visser EJ, Cohen M, Freedy A, Parasuraman R. A design methodology for trust cue calibration in cognitive agents. In: Proceedings of the International Conference on Virtual, Augmented and Mixed Reality. Springer International Publishing; 2014. p. 251–262. https://doi.org/10.1007/978-3-319-07458- 0_24. 28. Komatsu T, Yamada S, Kobayashi K, Funakoshi K, Nakano M. Artificial subtle expressions: intuitive notification methodology of artifacts. In: Proceedings of the SIGCHI Conference on Human Factors in Computing Systems; 2010. p. 1941–1944. https://doi.org/10.1145/1753326.1753619. 29. Cowell AJ, Stanney KM. Manipulation of non-verbal interaction style and demographic embodiment to increase anthropomorphic computer character credibility. International Journal of Human Computer Studies. 2005; 62(2):281–306. https://doi.org/10.1016/j.ijhcs.2004.11.008 30. Waytz A, Heafner J, Epley N. The mind in the machine: Anthropomorphism increases trust in an autono- mous vehicle. Journal of Experimental Social Psychology. 2014; 52:113–117. https://doi.org/10.1016/j. jesp.2014.01.005. 31. Laughery KR, Wogalter MS. A three-stage model summarizes product warning and environmental sign research. Safety Science. 2014; 61:3–10. https://doi.org/10.1016/j.ssci.2011.02.012. 32. Crump MJC, Mcdonnell JV, Gureckis TM. Evaluating amazon’s mechanical turk as a tool for experi- mental behavioral research. PLoS ONE. 2013; 8(3):57410. https://doi.org/10.1371/journal.pone. 0057410. 33. The Cesium Consortium. CesiumJS—Geospatial 3D mapping and virtual globe platform; 2018. Avail- able from: http://cesiumjs.org. 34. Microsoft. Bing Maps API documentation; 2018. Available from: https://www.microsoft.com/en-us/ maps/. 35. Changizi MA, Brucksch M, Kotecha R, McDonald K, Rio K. Ecological warnings. Safety Science. 2014; 61:36–42. https://doi.org/10.1016/j.ssci.2013.07.012. 36. Stanislaw H. Calculation of signal detection theory measures. Behavior Research Methods, Instru- ments, & Computers. 1999; 31(1):137–149. https://doi.org/10.3758/BF03207704. 37. Krizhevsky A, Sutskever I, Hinton GE. Imagenet classification with deep convolutional neural networks. In: Advances in Neural Information Processing Systems 25 (NIPS 2012); 2012. p. 1097–1105. https:// doi.org/DOI:10.1145/3065386. 38. Geirhos R, Medina Temme CR, Rauber J, Schu¨tt HH, Bethge M, Wichmann FA. Generalisation in humans and deep neural networks. In: Proceedings of the 32nd Conference on Neural Information Pro- cessing Systems; 2018. p. 7549–7561. https://dl.acm.org/doi/10.5555/3327757.3327854. 39. Dodge S, Karam L. A study and comparison of human and deep learning recognition performance under visual distortions. 2017 26th International Conference on Computer Communications and Net- works, ICCCN 2017. 2017;https://doi.org/10.1109/ICCCN.2017.8038465. 40. Conzola VC, Wogalter MS. A Communication—Human Information Processing (C–HIP) approach to warning effectiveness in the workplace. Journal of Risk Research. 2001; 4(4):309–322. https://doi.org/ 10.1080/1366987011006271. PLOS ONE \| https://doi.org/10.1371/journal.pone.0229132 February 21, 2020 PLOS ONE \| https://doi.org/10.1371/journal.pone.0229132 February 21, 2020 References 19 / 20 PLOS ONE \| https://doi.org/10.1371/journal.pone.0229132 February 21, 2020 Adaptive trust calibration for human-AI collaboration 41. Parasuraman R, Sheridan TB, Wickens CD. A model for types and levels of human interaction with automation. IEEE Transactions on Systems, Man, and Cybernetics—Part A: Systems and Humans. 2000; 30(3):286–297. https://doi.org/10.1109/3468.844354. 42. Mirnig AG, Wintersberger P, Sutter C, Ziegler J. A framework for analyzing and calibrating trust in auto- mated vehicles. In: Adjunct Proceedings of the 8th International Conference on Automotive User Inter- faces and Interactive Vehicular Applications; 2017. p. 33–38. https://doi.org/10.1145/3004323. 3004331. 43. Naujoks F, Wiedemann K, Scho¨mig N. The importance of interruption management for usefulness and acceptance of automated driving. In: Proceedings of the 9th International Conference on Automotive User Interfaces and Interactive Vehicular Applications; 2017. p. 254–263. https://doi.org/10.1145/ 3122986.3123000. 44. Strayer DL, Johnston WA. Driven to distraction: Dual-task studies of simulated driving and conversing on a cellular telephone. Psychological Science. 2001; 12(6):462–466. https://doi.org/https://doi.org/10. 1111/1467-9280.00386. PMID: 11760132 45. National Highway Traffic Safety Administration(NHTSA). Automatic vehicle control systems—investiga- tion of Tesla accident; 2017. National Highway Traffic Safety Administration. Available from: https:// static.nhtsa.gov/odi/inv/2016/INCLA-PE16007-7876.PDF. 46. National Transportation Safety Board(NTSB). Preliminary report—crash involving pedestrian, Uber Technologies, Inc., test vehicle; 2018. Available from: https://www.ntsb.gov/investigations/ AccidentReports/Reports/HWY18MH010-prelim.pdf. 20 / 20
https://openalex.org/W4285504007	https://upcommons.upc.edu/bitstream/2117/374059/1/IPDPS_2022.pdf	English	null	OmpSs@cloudFPGA: An FPGA Task-Based Programming Model with Message Passing	2022 IEEE International Parallel and Distributed Processing Symposium (IPDPS)	2,022	public-domain	11,273	OmpSs@cloudFPGA: An FPGA Task-Based Programming Model with Message Passing Juan Miguel de Haro∗†, Rub´en Cano∗, Carlos ´Alvarez∗†, Daniel Jim´enez-Gonz´alez∗†, Xavier Martorell∗†, Eduard Ayguad´e∗†, Jes´us Labarta∗† ∗Barcelona Supercomputing Center †Universitat Polit`ecnica de Catalunya {juan.deharoruiz, ruben.cano, carlos.alvarez, djimenez, xavier.martorell, eduard.ayguade, jesus.labarta}@bsc.es Francois Abel, Burkhard Ringlein, Beat Weiss IBM Research Europe {fab, ngl, wei}@zurich.ibm.com In these systems, part of the work traditionally done by a CPU is ofﬂoaded to a specialized device. This device, e.g., a GPU, FPGA, or ASIC, is more efﬁcient than the CPU in some way, like performance or power. The most common heterogeneous architectures for High-Performance Computing (HPC) use GPUs. Recently, more and more FPGAs are making their way into data centers (DC) [1] to help optimize the compute as well as the data movement of cloud workloads [2]. An FPGA is a reconﬁgurable device that can host any hardware design that ﬁts within its available resources. Therefore, FPGAs can be tuned and optimized for a wide variety of workloads. The vast majority of FPGAs deployed in DCs operate with a traditional CPU-FPGA bus attachment such as the Periph- eral Component Interconnect Express (PCIe). In recent data centers, we observe the emergence of a new interconnection pattern in which FPGAs are directly connected to the DC network fabric. Examples of such a pattern include the FPGAs placed in between the CPU and the network in a so-called ”bump-in-the-wire” conﬁguration [3] and the FPGAs operated as standalone network-attached accelerators [2]. This is a complete change of paradigm in CPU-to-FPGA and FPGA-to- FPGA inter-communications. It opens new perspectives for the use and the deployment of clusters of FPGAs in heterogeneous cloud DCs. However, the speciﬁcities of network-attached FPGAs require a new framework to proﬁt from the potential provided by the clustering of FPGAs over high-speed and low- latency networks. Abstract—Nowadays, a new parallel paradigm for energy- efﬁcient heterogeneous hardware infrastructures is required to achieve better performance at a reasonable cost on high- performance computing applications. Under this new paradigm, some application parts are ofﬂoaded to specialized accelerators that run faster or are more energy-efﬁcient than CPUs. Field- Programmable Gate Arrays (FPGA) are one of those types of accelerators that are becoming widely available in data centers. g y This paper proposes OmpSs@cloudFPGA, which includes novel extensions to parallel task-based programming models that enable easy and efﬁcient programming of heterogeneous clusters with FPGAs. OmpSs@cloudFPGA: An FPGA Task-Based Programming Model with Message Passing The programmer only needs to annotate, with OpenMP-like pragmas, the tasks of the application that should be accelerated in the cluster of FPGAs. Next, the proposed program- ming model framework automatically extracts parts annotated with High-Level Synthesis (HLS) pragmas and synthesizes them into hardware accelerator cores for FPGAs. Additionally, our extensions include and support two novel features: 1) FPGA-to- FPGA direct communication using a Message Passing Interface (MPI) similar Application Programming Interface (API) with one-to-one and collective communications to alleviate host com- munication channel bottleneck, and 2) creating and spawning work from inside the FPGAs to their own accelerator cores based on an MPI rank-like identiﬁcation. These features break the classical host-accelerator model, where the host (typically the CPU) generates all the work and distributes it to each accelerator. We also present an evaluation of OmpSs@cloudFPGA for different parallel strategies of the N-Body application on the IBM cloudFPGA research platform. Results show that for cluster sizes up to 56 FPGAs, the performance scales linearly. To the best of our knowledge, this is the best performance obtained for N-body over FPGA platforms, reaching 344 Gpairs/s with 56 FPGAs. Finally, we compare the performance and power consumption of the proposed approach with the ones obtained by a classical execution on the MareNostrum 4 supercomputer, demonstrating that our FPGA approach reduces power consumption by an order of magnitude. In this paper, we introduce a programming model frame- work called OmpSs@cloudFPGA that automatically takes care of all the low-level infrastructure of such clusters. With it, the user can exploit the parallelism of large FPGA clusters and scale out an application with minimal changes to the code. The main objective of OmpSs@cloudFPGA is to scale with decent performance on big clusters. This paper presents the following contributions: Index Terms—FPGA, MPI, OpenMP, programming models, network-attached FPGA, stand-alone FPGA, High-Level Synthe- sis, heterogeneous programming, High-performance computing ©2022 IEEE Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribu-tion to servers or lists, or reuse of any copyrighted component of this work in other works. This is the accepted version of an article published by IEEE. Accepted for IEEE IPDPS 2022 Accepted for IEEE IPDPS 2022 B. OmpSs OmpSs is a task-based programming model [4] developed at the Barcelona Supercomputing Center. In OmpSs, the paral- lelism of a C/C++ application is expressed mainly with tasks, similar to OpenMP tasks. These tasks are pieces of code that may require or produce data, in the form of single memory addresses or regions of memory. The runtime of OmpSs is called Nanos5 [5]. It dynamically checks for dependencies among tasks, based on the data requirement and production, and executes all possible tasks in parallel on Symmetric Multiprocessors (SMP). The programming model of OmpSs is ﬂexible and was extended to execute tasks in devices such as GPUs and FPGAs. In the case of FPGAs, the extension is referred to as OmpSs@FPGA. I. INTRODUCTION • OmpSs@cloudFPGA, a parallel task-based programming model that allows easy and efﬁcient programming of FPGA clusters. As applications require more performance, the underlying hardware has to evolve to satisfy their needs. This is why het- erogeneous architectures are becoming more popular recently. • A method to distribute program task control over the • A method to distribute program task control over the FPGA nodes on the cluster to improve system scalability. FPGA nodes on the cluster to improve system scalability. dynamic memory allocation (e.g., malloc) and recursive calls are not supported. Moreover, OmpSs@FPGA introduces some limitations too. The bit width of task parameters is limited to 64-bit because the communication protocol with the FPGA uses a single 64-bit word per parameter. Also, at the time of writing, the framework does not provide a mechanism to support return types other than void. • Direct MPI-like FPGA-to-FPGA communication to alle- viate host communication channel bottlenecks. • A data directory that dynamically performs data copies between a CPU host and other types of hardware devices attached to it. • An evaluation of the presented model on a cluster of 56 FPGAs with an N-body application, delivering the highest reported performance of N-body implemented over FPGA to the best of our knowledge. The compilation process is shown in Fig. 1. It consists of a source-to-source compiler, Mercurium, the Accelerator Integrator Tool (AIT), and the native compiler (e.g., GCC). Mercurium reads the original code and separates the code to be executed by CPUs from the code targeted at FPGAs. However, it does not check language compatibility with the vendor HLS tool, hence any compilation error related to this matter is reported by the tool itself. AIT receives the FPGA code and generates the ﬁnal bitstream. The native host compiler links the user application with the Nanos5 runtime, linked to the Xtasks library. The latter implements the low-level communication between the speciﬁc host and FPGA in the system. • A comparison of the performance and power consump- tion of the presented model with the one obtained by a CPU cluster. A. FPGA programming Traditionally, the complete design implemented on an FPGA had to be programmed in a Hardware Description Lan- guage (HDL), e.g., Verilog or VHDL. The Register Transfer Level (RTL) engineer would have to code a hardware module and connect it directly to the FPGA pins or use Intellectual Property (IP) libraries that already implement the necessary protocols. With the rise of High-Level Synthesis and automatic tools, the gap between RTL and software engineers has been dramatically reduced. Nowadays, the FPGA user can program a custom IP with high-level languages and build and test the design through an automatic framework. For example, the Xilinx Vitis platform allows synthesizing accelerators using C/C++ on the FPGA, and communicating with the software application with a provided API. OmpSs@FPGA introduces new clauses to the #pragma target to specify that the code associated with the task is to be executed on the FPGA, along with other device-speciﬁc clauses. For example, the user can declare local arrays, which are buffers stored in the FPGA local memory, e.g., SRAMs embedded in the FPGA fabric. Mercurium identiﬁes FPGA tasks, extracts the necessary code, applies some transforma- tions, and generates new source code that includes the code of the tasks. An HLS tool synthesizes this code into one or more hardware IP accelerators. The user code is instantiated into a wrapper that commu- nicates with a hardware runtime inside the FPGA bitstream, called Picos OmpSs Manager (POM). AIT automatically con- nects POM with the user-deﬁned accelerators and with the CPU at compile time. At runtime, POM directly manages the accelerators and communicates with the CPU host of the device. Each time a CPU thread creates an FPGA task and is ready for execution, the Xtasks library sends the task information directly to POM through a hardware queue. Accelerators can also create CPU and FPGA tasks, and POM can handle the dependencies they may have. Therefore, POM is the counterpart of the Nanos5 runtime located in the host. The tasks spawned by POM remain local to the FPGA and are not visible by the runtime in the CPU. As a result, the number of communications between the host and the FPGA is reduced. D. OmpSs-2 The OmpSs@FPGA framework [6] is the extension of OmpSs that enables executing FPGA tasks on heterogeneous CPU+FPGA-based systems. It uses FPGA-speciﬁc vendor tools to automate the generation of the FPGA bitstream from the original user source code written in C/C++. The supported HLS tool at the time of writing is Vivado HLS. Therefore, the language subset in FPGA tasks is limited by Vivado HLS, which does not support all features usually available in a reg- ular C/C++ program due to the nature of FPGAs. For example, OmpSs-2 is an improved version of OmpSs. Although the programming model itself is very similar, OmpSs-2 provides a slightly different syntax, such as the use of pragma oss in- stead of omp. The signiﬁcant differences relate to the Nanos6 runtime library [5], which comes with better performance, more and richer features than Nanos5. Therefore, we decided to work directly with OmpSs-2 and Nanos6 to develop our work. (fpgacc, fpgacxx) Native compiler AIT FPGA specific vendor tools Linker Fig. 1. Compilation process for OmpSs@FPGA To solve this issue, we propose an interface similar to the well-known MPI for FPGAs and a corresponding API that can be called from the user code. We called this FPGA version of MPI OmpSs MPI for FPGAs (OMPIF). The implemented API provides basic calls to send and receive messages. Each FPGA gets assigned a rank ranging from zero to the size of the cluster minus one. The CPU does not have a rank associated with the initial implementation as we have not implemented a software OMPIF runtime. I.e., all ranks are FPGA nodes, and the cluster size counts only the number of FPGAs. In future implementations, we plan to provide a software runtime that can communicate with the FPGAs through OMPIF. FPGA specific vendor tools Native compiler Fig. 1. Compilation process for OmpSs@FPGA Along with the rank and the size of the cluster, an FPGA accelerator can use the API to communicate with other de- vices. Like in classical MPI, a message comprises user data, and extra information called the envelope. The latter contains the source and destination ranks of the message and a user- deﬁned tag. This envelope is used to route the message to its corresponding destination and identify it so a receive call can match it. The equivalent of an MPI communicator is ﬁxed to the total number of available FPGAs. A. Modiﬁcations to the Nanos6 runtime To perform the work presented we have extended the Nanos6 runtime with support for multiple FPGAs. Our main contribution in Nanos6 that is not present in Nanos5 is the data directory, used to keep coherency between devices connected to the same CPU host. Before a task is executed on a device, the directory registers the memory regions accessed by this task, extracted from the dependencies. Initially, Nanos6 only supported GPUs with CUDA uniﬁed memory. The GPU automatically performs copies when needed in this model, but only between CPU-GPU and GPU-GPU. Instead, the directory handles both GPU and FPGA devices even when connected to the same host, and performs device memory allocations, address translation, and data copies. If data accessed by a device task is present only in the host, the directory issues a memory copy from the host to the device. It also handles copies from device to host if a host task needs data modiﬁed by a device and copies between devices if a device task needs data from another device. When the host requests a taskwait, all device data regions are invalidated and thus copied back to the CPU. The directory transparently takes care of all the data copies without any explicit allocation and memory copies in the user code. • void OMPIF_Send(const void* data, int count, OMPIF_Datatype datatype, int destination, uint8_t tag, OMPIF_COMM communicator) • void OMPIF_Recv(const void* data, int count, OMPIF_Datatype datatype, int source, uint8_t tag, OMPIF_COMM communicator) All parameters are equivalent to their MPI counterparts except for a few modiﬁcations. tag is restricted to an 8-bit unsigned integer. Currently, the communicator is only allowed to have the OMPIF_COMM_WORLD value that involves all the FPGAs in the cluster. The API calls and speciﬁcation are proposals and, therefore, subject to change. Both send and receive operations are blocking: OMPIF_Send returns when the send buffer is safe to be modiﬁed, and OMPIF_Recv returns when the buffer contains the matching message data. D. OmpSs-2 However, a distributed task could target a subset of the cluster and use communicators to communicate in those subsets. The prototypes of the send and receive calls are: B. Extensions to OmpSs-2 With the directory approach, the host has to manage all tasks and data movements between all devices. Although this feature is valuable and easy for the programmer to use, as the number of FPGAs increases, the host management of data movements and tasks may become a bottleneck. This is further discussed in section VI. 2) Distributed task spawn: Up to this point, FPGA accel- erators in a cluster can create tasks and communicate with each other through the OMPIF API. However, there is no way to easily start executing an application on the whole cluster from the host. The main objective is to provide the user with a simple way to execute a distributed application in a cluster of any size. The classical MPI approach is to execute the same binary in the cluster, and depending on the rank, each node decides what to do. However, there are some limitations for FPGAs. One of the main problems is that the cluster is heterogeneous, with at least two types of devices: a CPU and many FPGAs. Both have different characteristics, features, and roles in the application. Thus they require very different codes. For example, FPGA devices may not have a disk attached, so the CPU must perform all OmpSs@cloudFPGA extends the OmpSs-2 programming model and framework with two main features to solve the problem mentioned above. 1) FPGA point-to-point communication: To scale out appli- cations on FPGA clusters, we ﬁrst need to distribute the work spawn on each FPGA. This is achieved by the POM hardware runtime, which allows an FPGA to create tasks independently, without the interaction of a CPU. Next, we need a way to distribute the data movement between FPGAs. POM CPU Nanos6 DC Network DATA (nanos6_distributed_memcpy) + TASKS DATA (OMPIF_send/recv) FPGA Main App FPGA POM Rank 0 Rank 1 Rank 2 Rank 3 Rank n-2 Rank n-1 FPGA POM App FPGA POM App FPGA POM App FPGA POM App App Fig. 2. High-level view of an OmpSs@cloudFPGA cluster disk operations required by the application. Another problem is memory allocation. User accelerators are not supported by an operating system or virtual memory in the FPGA. Instead of using an external allocator, it is more straightforward if the CPU does all the allocations and communicates the addresses to the accelerators. With the proposed model and use case, there is no immediate beneﬁt in allocating memory on the FPGA. B. Extensions to OmpSs-2 High-level view of an OmpSs@cloudFPGA cluster #pragma oss task device(fpga) in([4]a) out([4]b) void fpga_comp_code(int* a, int* b) {...} #pragma oss task device(fpga) distributed \ scatter([nblocks4]a) gather([nblocks4]b) void fpga_creator_code(int* a, int* b, int nblocks) { int rank = OMPIF_comm_Rank(OMPIF_COMM_WORLD); int size = OMPIF_comm_Size(OMPIF_COMM_WORLD); for (int i = 0; i < nblocks; ++i) fpga_comp_code(a + i4, b + i4); #pragma oss taskwait OMPIF_Send(b, nblocks4, OMPIF_INT, (rank+1)%size, 0, OMPIF_COMM_WORLD); OMPIF_Recv(a, nblocks4, OMPIF_INT, (size+rank-1)%size, 0, OMPIF_COMM_WORLD); for (int i = 0; i < nblocks; ++i); fpga_comp_code(a + i4, b + i4); #pragma oss taskwait } int main() { int a, b; int nblocks; ...//Initialize input with nblocks4nnodes elements fpga_creator_code(a, b, nblocks); #pragma oss taskwait ...//Process output } Listing 1: Example of OmpSs@cloudFPGA C code #pragma oss task device(fpga) in([4]a) out([4]b) void fpga_comp_code(int* a, int* b) {...} #pragma oss task device(fpga) distributed \ scatter([nblocks4]a) gather([nblocks4]b) void fpga_creator_code(int* a, int* b, int nblocks) { int rank = OMPIF_comm_Rank(OMPIF_COMM_WORLD); int size = OMPIF_comm_Size(OMPIF_COMM_WORLD); for (int i = 0; i < nblocks; ++i) fpga_comp_code(a + i4, b + i4); #pragma oss taskwait OMPIF_Send(b, nblocks4, OMPIF_INT, (rank+1)%size, 0, OMPIF_COMM_WORLD); OMPIF_Recv(a, nblocks4, OMPIF_INT, (size+rank-1)%size, 0, OMPIF_COMM_WORLD); for (int i = 0; i < nblocks; ++i); fpga_comp_code(a + i4, b + i4); #pragma oss taskwait } int main() { int a, b; int nblocks; ...//Initialize input with nblocks4nnodes elements fpga_creator_code(a, b, nblocks); #pragma oss taskwait ...//Process output } #pragma oss task device(fpga) in([4]a) out([4]b) void fpga_comp_code(int* a, int* b) {...} #pragma oss task device(fpga) distributed \ scatter([nblocks4]a) gather([nblocks4]b) void fpga_creator_code(int* a, int* b, int nblocks) { int rank = OMPIF_comm_Rank(OMPIF_COMM_WORLD); int size = OMPIF_comm_Size(OMPIF_COMM_WORLD); for (int i = 0; i < nblocks; ++i) fpga_comp_code(a + i4, b + i4); #pragma oss taskwait OMPIF_Send(b, nblocks4, OMPIF_INT, (rank+1)%size, 0, OMPIF_COMM_WORLD); OMPIF_Recv(a, nblocks4, OMPIF_INT, (size+rank-1)%size, 0, OMPIF_COMM_WORLD); for (int i = 0; i < nblocks; ++i); fpga_comp_code(a + i4, b + i4); #pragma oss taskwait The ﬁnal problem to consider is distributing the initial data in the FPGAs and retrieving it back. The CPU does not implement OMPIF, so we can not use send/receive calls. Instead, we propose different methods, including well-known communication patterns as pragma clauses, only valid for distributed tasks. • Broadcast: The speciﬁed range is broadcasted to all FPGAs in the cluster before executing the task. B. Extensions to OmpSs-2 • Scatter: The speciﬁed buffer is divided into equal chunks of the speciﬁed size, and each chunk is sent to a different FPGA before executing the task. } int main() { int a, b; int nblocks; ...//Initialize input with nblocks4nnodes elements fpga_creator_code(a, b, nblocks); #pragma oss taskwait ...//Process output } Listing 1: Example of OmpSs@cloudFPGA C code • Gather: When the execution ﬁnishes, the speciﬁed buffer is divided into equal chunks of the speciﬁed size, and each chunk is read from a different FPGA. • Send/receive: The speciﬁed range is sent to the speciﬁed rank before executing the task, or received after the execution ﬁnishes. Furthermore, we provide a blocking Nanos6 API, nanos6_distributed_memcpy that enables the func- tionality of the clauses above at the application level. Listing 1: Example of OmpSs@cloudFPGA C code C. Execution ﬂow the nanos6_distributed_memcpy API before calling the distributed task and after the taskwait. The CPU then creates a distributed task that Nanos6 broadcasts to all FPGAs. In the example, they execute the fpga_creator_code (App box in Fig. 2) function. This code communicates with its neighbor ranks and spawns fpga_comp_code tasks that POM handles. In Fig. 2, these tasks are represented by the colored circles in each FPGA node. When all FPGAs ﬁnish, the CPU program retrieves data from the cluster and processes the output. To summarize the OmpSs@cloudFPGA model, Fig. 2 shows a high-level view of an OmpSs@cloudFPGA cluster and Listing 1 shows a code example. In Listing 1, vectors a and b are integer vectors divided in blocks of four elements. The total size of a and b is proportional to the number of nodes in the system since they are distributed with a scatter/gather collective. Each device computes some code over a chunk of b with a task per block. The result is sent to the next FPGA that receives this data, stores it in its part of a and computes another chunk of b. The CPU executes the main function (Main box in Fig. 2), initializes the data, and distributes it to the whole cluster with the scatter/gather clauses in the task pragma. Alternatively, the user could use B. Extensions to OmpSs-2 In general, the CPU provides more features, and the FPGA is mainly used to perform calculations only. DATA (nanos6_distributed_memcpy) + TASKS DATA (OMPIF_send/recv) In the OmpSs@FPGA model, introduced in section II-C, the FPGA code is already isolated inside a task, which is speciﬁed as a C/C++ function. We extended the FPGA task declaration pragma with a distributed clause. Distributed tasks can only be called from the application code running in the CPU. When invoked, a single instance is replicated for each device in the cluster. This task instance becomes the entry point of the application execution. A distributed task is implemented as a regular accelerator, and it can use OMPIF and create tasks based on its rank and cluster size. These properties can be read at runtime by calls to the OMPIF API, equivalent to MPI_comm_rank and MPI_comm_size. Fig. 2. A. The cloudFPGA system • Application messages: Packets generated by the use of the OMPIF API. There are two subtypes: cloudFPGA is a research cluster that aims to demonstrate new concepts and techniques for deploying FPGAs at a large scale in DCs. The system is built on three mains pillars: 1) the use of standalone network-attached FPGAs cards, 2) a hyperscale infrastructure for deploying such FPGA cards at large scale and in a cost-effective way [7] and, 3) an accelerator service that integrates and manages the standalone network-attached FPGAs in the Cloud [8]. – Data messages: Contain the actual data that one FPGA sends to another through a send operation. – Ack messages: Do not contain any relevant data, but are used to conﬁrm to the sender that one or more messages have been successfully received. This is needed because UDP is not a reliable protocol and thus some packets can be dropped in the network. The main difference with other research projects and com- mercial products is the absence of a PCIe bus for the host or another device to interact with the FPGA. Instead, this classical communication channel and its associated card driver are replaced with two UDP/IP and TCP/IP network stack interfaces and their afﬁliated socket programming models. 1) Packet decoder and encoder: These modules are the entry and exit points of the Role for external communication. The decoder captures all incoming messages and forwards them to the corresponding module depending on the packet type. The decoder is also responsible for replying to data messages with an ack. Similar to many other FPGA cloud offerings, cloudFPGA standalone network-attached FPGA builds on the common Shell-Role Architecture (SRA) design pattern [9]. This design separates the platform-speciﬁc parts (i.e., the Shell) from the application-speciﬁc parts (i.e., the Role) to increase the re- usability and isolate the two parts. The Shell contains all necessary I/O components, and the network stack that hooks the FPGA to the DC network, as shown in the leftmost part of Fig. 3. It further abstracts all these hardware components by exposing standard AXI-stream interfaces to the user. From a computer operating system perspective, the Shell can be seen as the conceptual counterpart of the kernel space. The Role is the application-speciﬁc part of the FPGA logic. It embeds the user’s custom application and can be assimilated to a CPU application executed in user space. A. The cloudFPGA system The packet encoder forwards messages sent by any module of the Role to the Shell. It also sets the destination node and port and adds a header for responses to CPU commands. Both modules include debugging counters that track the number of packets sent/received, classiﬁed by the aforemen- tioned types and subtypes. There are also registers that track the number of application messages exchanged with each individual node in the cluster. The host can access them with a CPU command. 2) OMPIF message passing runtime: As shown in Fig. 3, the runtime is a collection of two modules to handle message emission and reception, managed by POM. Thus internally, user accelerators issue send/receive petitions with the same interface used to create tasks. The message sender handles send operations only. It reads the memory in the speciﬁed address and transmits the data to the packet encoder. The message receiver reads a temporary buffer where the messages are stored just after arriving at the Role. When a message matching a receive request is found, it copies the data to the speciﬁed address. IV. IMPLEMENTATION OF OMPSS@CLOUDFPGA This section describes the implementation details of our framework concerning a speciﬁc cluster of standalone done using UDP protocol because it has lower latency than TCP, and packet loss rate stays in acceptable bounds. There are many types of packets transported with UDP, used by the FPGA and the host. They can be classiﬁed according to the format of the header and the destination port: network-attached FPGAs. Such a cluster is accessible at the IBM Research laboratory in Zurich, Switzerland [7], [8], [9], and was made available for us to carry out this work. In the remainder of this article, we will refer to this cluster as cloudFPGA. Despite the fact that the presented implementation targets a speciﬁc platform, most of its components and the program- ming model can target other clusters. The programming model is agnostic of the underlying platform where the tasks are executed. • CPU commands: These commands are used by the host to control the Role, for example, to execute a task or read/write FPGA memory. • CPU command responses: Most CPU commands expect one or more responses from the FPGA. The data con- tained in the packets depends on the command. D. Host runtime implementation The broadcaster is also responsible for starting memory copies since the directory is unaware of the data locality once it belongs to the broadcaster. The underlying network that moves a message between two FPGAs has a speciﬁed Maximum Transmission Unit (MTU) of 1450 bytes, but the API supports sending more than this quantity. To overcome this limitation, the sender engine splits a message in frames of 1408 bytes. This size is the maximum multiple of 64 bytes that the payload can carry. The size is limited by a restriction in the memory address alignment. It has to be aligned to 64 bytes because the AXI4 data channels of the memory controller are 512-bit wide. Using unaligned addresses increases the resource usage and the complexity of the code. Therefore we decided to keep all addresses aligned. B. cloudFPGA Role architecture With OmpSs@cloudFPGA, we add a new abstraction layer to the user in the Role. From the programmer’s perspective, the application is built with hardware accelerators, usually in HLS C/C++. These accelerators are invoked from tasks created by other C/C++ code in the host or the FPGA and can communicate with other nodes in the cluster through the OMPIF API. In our framework, the cloudFPGA Role provides all the necessary components to make this possible. It contains mainly the OmpSs@FPGA hardware runtime (POM), the ap- plication accelerators, the message passing runtime (message sender/receiver), the packet decoder/encoder, and the memory manager. All these components are interconnected, as shown in the rightmost part of Fig. 3. External communication is 3) Memory manager: This module handles memory read/write requests directly from the host and write requests to store data messages in a temporary buffer. Mainly it translates an AXI-stream codiﬁed with a speciﬁc format to multiple AXI-stream interfaces with another format. These interfaces communicate with a data mover IP, which receives read/write commands and access actual memory through an AXI4 interface. Packet Decoder Memory manager Message receiver Message sender Packet encoder POM X Simple AXI-stream AXI4 UDP AXI-stream Data mover AXI-stream User Accelerator User application OMPIF runtime From Shell To Shell To memory To data mover OmpSs@cloudFPGA infrastructure FPGA Module DRAM DC Network Shell DDR4 I/F Integrated NIC (iNIC) Role Fig. 3. cloudFPGA Shell-Role architecture, and detailed view of the OmpSs@cloudFPGA Role. Memory manager Fig. 3. cloudFPGA Shell-Role architecture, and detailed view of the OmpSs@cloudFPGA Role. D. Host runtime implementation 1) Xtasks backend: The ﬁrst step is to build a new backend for the Xtasks library, modifying the API speciﬁcation. The old API assumes that only one device is attached, so there is no means to specify a target FPGA index. With this issue solved, we implemented a backend that uses UDP sockets to do all communication. They use two ports, one to send/receive tasks and another to read/write device memory. The library is thread safe and supports concurrent communication between different devices. Nevertheless, packets targeting the same device are serialized due to the protocol used to exchange information. The designed protocol is used to control send/receive com- munication between two nodes. All FPGAs reserve a region of their memory address space to store temporarily data messages. When the user code issues a matching receive call, the data message is moved to its ﬁnal address. In the proposed implementation, the reserved region is a circular queue stored in the board DRAM. To simplify the protocol, it is assumed that there is enough space in the reserved buffer to store all possible messages. This way, the sender can assume that all packets that arrive at the destination can be stored, and the receiver does not have to check if the buffer is full. We plan to remove this limitation in the future. p g 2) Nanos6 runtime: To support distributed tasks, we added a new type of virtual device in Nanos6 called broadcaster. This device is in fact the host, which receives a single distributed task and sends it to all devices in the cluster. This implementation is helpful because it uses an abstract class representing a device, and therefore can potentially work with any device, not only FPGAs. The broadcaster is also responsible for starting memory copies since the directory is unaware of the data locality once it belongs to the broadcaster. 2) Nanos6 runtime: To support distributed tasks, we added a new type of virtual device in Nanos6 called broadcaster. This device is in fact the host, which receives a single distributed task and sends it to all devices in the cluster. This implementation is helpful because it uses an abstract class representing a device, and therefore can potentially work with any device, not only FPGAs. V. TEST APPLICATION: N-BODY The N-body application simulates the dynamic interaction of particles inﬂuenced by the force of gravity. Each particle has a position, velocity and mass associated, and the force between each pair of particles is calculated following the Newton’s law of gravity: The message passing protocol also takes into account packet loss. Because of UDP, data messages can be dropped before reaching their destination. The proposed solution is to use an acknowledge message to conﬁrm that the data reached the destination. If this ack message is not received in a ﬁxed amount of time, the sender retransmits the data message. To increase the bandwidth, a ﬁxed window of four packets can be sent before waiting for an ack. If any of the four messages or the ack is dropped, the sender transmits the whole window again. Fij = G × mi × mj × (pj −pi) \|\|pj −pi\|\|3 Where Fij is a 3-dimensional vector with the forces between particles i and j, mi is the mass of particle i, pi is a 3- dimensional vector with the position of particle i and G is the gravitational constant. V. TEST APPLICATION: N-BODY The simulation is an iterative process that calculates the forces between all pairs of particles on each iteration, ac- cumulates the forces for each particle, and then updates the Algorithm 1 Pseudocode of the FPGA accelerator that creates tasks and communicates with the cluster start ←(n/size) ∗rank end ←start + n/size for t in 0..timesteps do for i in 0..n do for j in start..end do calculate forces(parti, partj, forcesj) end for end for taskwait() OMPIF Allgather(forces0..n) for i in 0..n do update particles(parti, forcesi) end for end for Algorithm 1 Pseudocode of the FPGA accelerator that creates tasks and communicates with the cluster start ←(n/size) ∗rank end ←start + n/size for t in 0..timesteps do for i in 0..n do for j in start..end do calculate forces(parti, partj, forcesj) end for end for taskwait() OMPIF Allgather(forces0..n) for i in 0..n do update particles(parti, forcesi) end for end for Algorithm 1 Pseudocode of the FPGA accelerator that creates tasks and communicates with the cluster start ←(n/size) ∗rank end ←start + n/size for t in 0..timesteps do for i in 0..n do for j in start..end do calculate forces(parti, partj, forcesj) end for end for taskwait() OMPIF Allgather(forces0..n) for i in 0..n do update particles(parti, forcesi) end for end for position and velocities using the Euler method. Therefore, the algorithm has O(n2) time complexity. Algorithm 1 Pseudocode of the FPGA accelerator that creates tasks and communicates with the cluster tasks and communicates with the cluster start ←(n/size) ∗rank end ←start + n/size for t in 0..timesteps do for i in 0..n do for j in start..end do calculate forces(parti, partj, forcesj) end for end for taskwait() OMPIF Allgather(forces0..n) for i in 0..n do update particles(parti, forcesi) end for end for B. FPGA bitstream conﬁguration However, particle update tasks from one timestep are overlapped with the execution of the force calculation tasks from the next timestep. 3) Final version, distributed communication and control: The main improvement of this implementation is that the FP- GAs distributively handle both tasks and data movements. The bitstream includes an accelerator devoted to create tasks and call the OMPIF API, following algorithm 1. The function calls calculate forces and update particles are non-blocking task spawn functions, whereas OMPIF ALLGATHER is a blocking call. In the pseudocode, n is the number of blocks, rank is a device index starting at 0, and size is the size of the cluster. Each device spawns only the tasks for its own accelerators and handles the dependencies using POM. Forces are sent after all force calculation tasks have ﬁnished. There is also no overlapping with the tasks to update particles. However, particle update tasks from one timestep are overlapped with the execution of the force calculation tasks from the next timestep. In the tested bitstream, we ﬁt four instances of the force calculation accelerator and one instance to update the forces. The block size is 2048 particles, and the force calcula- tion accelerator calculates 8 forces per cycle. The resulting throughput is 32 forces per cycle at 200MHz. A. Parallelization strategy From the described application, we can extract two task types: one to calculate the interaction of particles and another to update the positions and velocities. The ﬁrst step to paral- lelize the N-body is to distribute the particles in blocks of a ﬁxed size. Then, each task operates on the block granularity. For example, to calculate the force interactions of four blocks, we need to execute the ﬁrst task 16 times. To update the positions and velocities, we only need to execute the second four times. Each task has dependencies on the blocks that it reads or writes. data movements because to copy a block from one FPGA to another, the data must pass through the CPU memory ﬁrst. B. FPGA bitstream conﬁguration Although this approach is easier to implement for the user, it has some limitations. The scalability depends heavily on the host throughput for both sending tasks and moving data between devices. Our implementation of the N-body has been tested on Xilinx Kintex Ultrascale FPGAs of the cloudFPGA cluster. More details on the boards are discussed in section VI-A. Regarding the application, each task type can be implemented as an accelerator that can be instantiated multiple times in the FPGA fabric. Local memory, mainly Xilinx Block RAMs (BRAM), can be exploited because the block size is ﬁxed. The accelerator ﬁrst loads the block data in local memory then executes the algorithm using this memory, and ﬁnally stores the result back in the main memory. The force calculation accelerator is also optimized to take advantage of the FPGA properties and thus increase performance. The algorithm loop is unrolled by a factor of n and then pipelined with an initiation interval of one. That implies that n forces are calculated and accumulated on each cycle. 2) Second version, host-centric with distributed communi- cation: This version takes advantage of the OMPIF runtime available in the bitstream. To move a block of data from one FPGA to another, Nanos6 calls a send operation on the device with the most recent data and a receive operation on the device that needs to update. With this method, the host throughput does not affect the application and the memory movements are distributed. However, the host still has to process all tasks of the application. 3) Final version, distributed communication and control: The main improvement of this implementation is that the FP- GAs distributively handle both tasks and data movements. The bitstream includes an accelerator devoted to create tasks and call the OMPIF API, following algorithm 1. The function calls calculate forces and update particles are non-blocking task spawn functions, whereas OMPIF ALLGATHER is a blocking call. In the pseudocode, n is the number of blocks, rank is a device index starting at 0, and size is the size of the cluster. Each device spawns only the tasks for its own accelerators and handles the dependencies using POM. Forces are sent after all force calculation tasks have ﬁnished. There is also no overlapping with the tasks to update particles. C. Task creation and communication implementation 1) First version, host-centric: The critical task of the N- body is force calculation. Because of its quadratic growth, most of the time is spent calculating pairs of forces. Therefore, our objective is to distribute this work between all FPGAs in the cluster. In our implementation, each device calculates only a subset of the forces. E.g., with 10 blocks of particles and 2 FPGAs, each one calculates the forces of 5 blocks against all the 10 blocks. That is a total of 50 tasks per device out of 100. However, both FPGAs need all the particle’s positions to calculate the forces. Despite that in previous versions, only positions were ex- changed between devices, we decided to move forces instead for this version. The N-body data set is stored in two buffers, one with forces and another with particle properties, including positions. Therefore, positions are not stored in consecutive memory locations between blocks. The host runtime is able to copy the necessary regions between FPGAs, but the OMPIF API requires one consecutive buffer to send/receive data. With the initial OmpSs-2 FPGA support, we can achieve this distribution of work with host task creation. Because the CPU handles all tasks, this one has a global view of the data locations. Thus the directory is able to maintain all accesses in the FPGAs coherent. To achieve coherency, the Nanos6 runtime has to move particle positions from host to FPGA and vice-versa. Nevertheless, this approach doubles the amount of We use an equivalent operation to an MPI Allgather with the force buffer. After the device has sent and received all necessary blocks, it updates all the particles’ positions and Block 0 Block 1 Block 2 Rank 0 Rank 1 Rank 2 Fig. 4. Exchange of forces in N-body with the OMPIF allgather collective. TABLE I TABLE I KINTEX ULTRASCALE RESOURCE COUNT, RELATIVE USAGE AND MAX FREQUENCY OF THE CLOUDFPGA SHELL AND ROLE FOR THE N-BODY LUT FF DSP BRAM LUTRAM Fmax (MHz) Resources 331K 663K 2760 1080 146K Usage (%) Bitstream 75.4 47.7 57.9 51.5 13.8 Shell 34.4 19 0.65 27.8 7.5 Accs 39.3 25.3 57.1 22.2 6.3 266 POM 0.62 0.35 0 1.1 0.08 219 OMPIF 0.2 0.2 0.18 0 0.01 304 Enc/dec 0.16 0.06 0 0.19 0.01 - Fig. 4. Exchange of forces in N-body with the OMPIF allgather collective. Fig. 4. Exchange of forces in N-body with the OMPIF allgather collective. C. Results The FPGA cards of the cloudFPGA platform that we used implement a Kintex Ultrascale FPGA (xcku060-ffva-1156- 2-i) and 16GB of DDR4 memory. The Kintex is a mid- range FPGA chosen for its excellent performance-per-dollar ratio, a metric that is particularly relevant for a large scale deployment in the Cloud. The CPU host is an Intel Xeon E5-2640 v4 @ 2.40GHz processor with 24GB of RAM. The MareNostrum 4 supercomputer uses high-performance Intel Xeon Platinum 8160 CPUs @ 2.1GHz. Each node has 48 cores and 96GB of memory. For the experiments, we use the same task creation and communication as algorithm 1 but using OmpSs-2 instead of OmpSs@cloudFPGA and Intel MPI runtime (MPI_allgather) instead of OMPIF. The force calculation task is vectorized using Intel intrinsics of the AVX- 512 extension. We built two versions of the application, the ﬁrst using all node resources, with 512-bit vectors (16 forces) and 48 cores per node. The block size is 512 since it proved to achieve the best performance. Each CPU computes 768 parallel forces, which is 24 times more than the Kintex FPGAs. Therefore, we decided also to evaluate a more limited version using four cores with 256-bit vectors (8 forces), which is equivalent in vector length and number of computing units to the FPGA bitstream. We also used the same block size. In all our tests, we ﬁx the number of iterations to 16 and only change the number of particles to limit the number of possible input parameters. We also measure performance in Gpairs/s, which represents the number of force interactions calculated per second. The total number of particle pairs is n2 ×t, where n is the number of particles and t is the number of iterations. In our ﬁrst experiment, we compared the performance of the three presented N-body versions in section V. With 16 FPGAs in the cluster, we ran the applications with the same bitstream, increasing the number of particles. Results are shown in Fig. 5. Host Tasks & Copies is the host-centric implementation in which the directory copies data from FPGA to host and again to another FPGA (section V-C1). Host Tasks+FPGA copies is the other host-centric implementation that uses the OMPIF runtime to copy data directly between FPGAs (section V-C2). FPGA Tasks+OMPIF refers to using the distributed task with FPGA task creation and the use of OMPIF from user code (section V-C3). VI. EVALUATION We demonstrate the scalability of the OmpSs@cloudFPGA framework with the N-body application described in section V on a cluster of 56 FPGAs. We also perform a similar study with CPU nodes in the MareNostrum 4 supercomputer. B. FPGA runtime and accelerator usage The allgather collective is already supported by OMPIF, but only when the send buffer is a part of the receive buffer (equivalent in MPI to use the parameter MPI_IN_PLACE). Internally, each rank sends its part of the data to the next rank and increases the destination rank, going to 0 when overﬂowing until all ranks have been served. This process is illustrated in Fig. 4, in which there are three ranks and three blocks. Each rank has a reserved buffer to store all force blocks but calculates a single force block. Then, each node sends its own block to the other two, following the order of Fig. 4 to ﬁll the missing blocks. 1/1 Table I shows the number of resources for each primitive of the Kintex FPGA and also the relative usage of the tested bitstream. This resource usage is split between the Shell and the Role. For the latter, table I shows resource usage of the OmpSs@cloudFPGA components and the user accelerators. These are the POM and OMPIF hardware runtimes, packet encoder, and decoder. The table also reports the max fre- quency calculated after hardware synthesis. It is an estimation because the routing delay is not accurate. It depends on the place&route, which at the same time depends on the whole design. However, we can conclude that POM contains the critical path. Therefore we set the accelerator clock to 200MHz, giving a margin to the place&route algorithm for extra route overhead. Increasing the frequency requires a more detailed study of the internal design of POM and a change to its implementation with a more restrictive frequency target. The packet encoder and decoder have a ﬁxed frequency of 156.25MHz because it is connected directly to a Shell clock. C. Task creation and communication implementation velocities. Although each FPGA does extra work, this part does not signiﬁcantly affect performance due to the linear against quadratic time relation between the two task types. B. FPGA runtime and accelerator usage C. Results As we increase the number of FPGAs or the problem size, the runtime overhead becomes more critical for the host-centric versions. As it can be observed at the rightmost part of Fig. 5, the runtime overhead (suffered by both host-centric implementations but more acutely by the Host Tasks+FPGA copies version) becomes so large that it overcomes the data transfer overhead of the host-centric copies version. distributed version is 4x faster. The host can only reach 26% of the peak performance with both copy strategies. This bottleneck worsens with the number of FPGAs used, leading to even worst comparable results with bigger clusters. Most of the performance loss is due to the communication latency between the host and FPGAs, there is a higher penalty since the data packets have to pass through more switches to reach the FPGAs. Even with only 16, the host cannot give work to each of the accelerators inside each FPGA, limiting its performance. For the data transfer, the protocol used by the host is similar to the one used by OMPIF. A ﬁxed number of transfers is sent before waiting for ﬁnishing ack messages and sending the subsequent transfers. However, sending tasks from the host to the accelerators is more complex because the runtime does not know how many tasks it will send. Therefore, the runtime sends only one outstanding task per FPGA before waiting for the ack. Although it could be improved, this solution will not have signiﬁcant effect in all applications because if tasks depend on each other, the runtime can not send outstanding tasks without breaking the dependence model. As we increase the number of FPGAs or the problem size, the runtime overhead becomes more critical for the host-centric versions. As it can be observed at the rightmost part of Fig. 5, the runtime overhead (suffered by both host-centric implementations but more acutely by the Host Tasks+FPGA copies version) becomes so large that it overcomes the data transfer overhead of the host-centric copies version. number of tasks per FPGA also increases as well as the number of messages to send/receive. As seen in Fig. 6, with 56 FPGAs, we reach 344.22 Gpairs/s. If we look at the single FPGA performance, we get 6.27 Gpairs/s. This is equivalent to a 97% efﬁciency over the theo- retical peak performance of 6.4 Gpairs/s (32 forces per cycle at 200MHz). C. Results This 3% difference is caused by the memory movements of the forces between local and main memory, the update part of the particles, and the runtime overhead. We measured a more precise estimation of the overhead added by OmpSs@cloudFPGA with a different experiment by running the application with zero computation time. The results indicate that, on average, 60 cycles per task are spent on the hardware runtime (mainly dependence management), equivalent to 0.01% of a single force calculation task execution time. Hence, we can conclude that the runtime overhead is negligible. We also measured the overhead introduced by OMPIF in a similar way. We ran the application without spawning computation tasks, only executing OMPIF_Allgather. The results show that we achieve a bandwidth of 4.2GB/s across 16 FPGAs out of 20GB/s (10Gbit/s ethernet per FPGA). The overhead added by data transfers is 0.32% of the total execution time. With the host-centric implementations, we measured an overhead of 54% with host copies and 50% with FPGA copies, caused by the bottleneck introduced when centralizing all transfers in a single device. Using FPGA copies reduces this overhead because the FPGAs perform the actual transfers, but the host still needs to communicate to start the copies. The host-FPGA latency is ≈0.5ms, sending a single message and receiving an ack is equivalent to 38% of the force calculation task time. Contrary to the FPGA runtime overhead, this time is not negligible and is one of the leading causes of the limited performance shown in Fig. 5. Next, we investigated the scalability of the FPGA tasks+OMPIF version when increasing the number of FPGAs. Fig. 6 illustrates the performance of the application when varying the number of FPGAs from 1 to 56. The ideal plot represents the performance of a single FPGA multiplied by the cluster size. Our distributed implementation efﬁciency (actual performance compared to the theoretical maximum) is 98% for 56 FPGA nodes. The number of particles on each run also increases to give enough work to each FPGA. There are 30 blocks of 2048 particles per node (61440 particles). Although the number of particles is constant per device, the amount of work is not. As there are more FPGAs in the cluster, the Finally, we compare the best performance we have (FPGA tasks+OMPIF version and 56 nodes) with the equivalent execution in the MareNostrum 4 supercomputer. The results are shown in Table II. C. Results We can see that both host-centric implementations do not scale at the same rate as the distributed one. Also, the performance starts to stabilize at a much lower point. The 0 10 20 30 40 50 #FPGAs 0 50 100 150 200 250 300 350 Gpairs/s ideal FPGA Tasks+OMPIF Fig. 6. Performance of the N-body application distributed over a cluster of FPGAs (61440 particles per FPGA) 0 10 20 30 40 50 #FPGAs 0 50 100 150 200 250 300 350 Gpairs/s ideal FPGA Tasks+OMPIF 0.0 0.2 0.4 0.6 0.8 1.0 1.2 #particles 1e6 0 20 40 60 80 100 Gpairs/s FPGA Tasks+OMPIF Host Tasks+FPGA copies Host Tasks & Copies 0.0 0.2 0.4 0.6 0.8 1.0 1.2 #particles 1e6 0 20 40 60 80 100 Gpairs/s FPGA Tasks+OMPIF Host Tasks+FPGA copies Host Tasks & Copies Fig. 5. Performance of the three FPGA versions of the N-body in a cluster of 16 FPGAs. Fig. 6. Performance of the N-body application distributed over a cluster of FPGAs (61440 particles per FPGA) Fig. 5. Performance of the three FPGA versions of the N-body in a cluster of 16 FPGAs. distributed version is 4x faster. The host can only reach 26% of the peak performance with both copy strategies. This bottleneck worsens with the number of FPGAs used, leading to even worst comparable results with bigger clusters. Most of the performance loss is due to the communication latency between the host and FPGAs, there is a higher penalty since the data packets have to pass through more switches to reach the FPGAs. Even with only 16, the host cannot give work to each of the accelerators inside each FPGA, limiting its performance. For the data transfer, the protocol used by the host is similar to the one used by OMPIF. A ﬁxed number of transfers is sent before waiting for ﬁnishing ack messages and sending the subsequent transfers. However, sending tasks from the host to the accelerators is more complex because the runtime does not know how many tasks it will send. Therefore, the runtime sends only one outstanding task per FPGA before waiting for the ack. Although it could be improved, this solution will not have signiﬁcant effect in all applications because if tasks depend on each other, the runtime can not send outstanding tasks without breaking the dependence model. C. Results The MareNostrum vec256 row refers to TABLE II PERFORMANCE AND POWER COMPARISON OF THE N-BODY APPLICATION ON A CLOUDFPGA CLUSTER AND ON THE MARENOSTRUM 4 SUPERCOMPUTER, WITH 56 NODES AND 3440640 PARTICLES. Performance (Gpairs/s) Efﬁciency Watts Perf./W OmpSs@cloudFPGA 344.22 98% 840 0.41 MareNostrum vec256 211.81 94% 8673 0.024 MareNostrum vec512 2409.49 90% 15504 0.16 the limited resources implementation, it uses 256-bit vectors, 4 cores per node, and 2048 block size. The MareNostrum vec512 row is the version with full resources and uses 512- bits vector lengths, 48 cores per node, and 512 block size. With similar resources, the FPGA cluster proves to have 1.6 better performance than the CPU. However, the vec512 version is 7 times faster than the FPGA. Although in terms of efﬁciency, the FPGA version is above the other two. We also have to take into account that we are comparing a mid-range FPGA with a high-performance CPU. Moreover, the Kintex Ultrascale technology is four years older than the Intel Xeon Platinum CPUs. The Kintex uses 20nm TSMC lithography, while the Xeon uses Intel 14nm. Even with this disadvantage, the FPGA consumes one order of magnitude less power than the vec256 version and 18 times less than vec512. Looking at the performance per watt, the FPGA version is above both MareNostrum implementations. TABLE II PERFORMANCE AND POWER COMPARISON OF THE N-BODY APPLICATION ON A CLOUDFPGA CLUSTER AND ON THE MARENOSTRUM 4 SUPERCOMPUTER, WITH 56 NODES AND 3440640 PARTICLES. Performance (Gpairs/s) Efﬁciency Watts Perf./W OmpSs@cloudFPGA 344.22 98% 840 0.41 MareNostrum vec256 211.81 94% 8673 0.024 MareNostrum vec512 2409.49 90% 15504 0.16 all other implementations. Applying our OmpSs@cloudFPGA framework to bigger or more modern FPGAs will result in even better results. TABLE II TABLE II PERFORMANCE AND POWER COMPARISON OF THE N-BODY APPLICATION ON A CLOUDFPGA CLUSTER AND ON THE MARENOSTRUM 4 SUPERCOMPUTER, WITH 56 NODES AND 3440640 PARTICLES. OpenMP also includes a mechanism to ofﬂoad program regions to other devices with the target pragma. This feature has been used to accelerate OpenMP code with FPGAs. Knaus et al. [13] use the Clang compiler to extract the LLVM Intermediate Representation (IR) code of the target regions. This code is transformed to an RTL IP with the OpenCL HLS toolﬂow that FPGA vendors (e.g. Xilinx and Intel) provide. They feed the vendor tools directly with a custom IR instead of the original C/C++ code. C. Results The Kintex uses 20nm TSMC lithography, while the Xeon uses Intel 14nm. Even with this disadvantage, the FPGA consumes one order of magnitude less power than the vec256 version and 18 times less than vec512. Looking at the performance per watt, the FPGA version is above both MareNostrum implementations. p g IBM has worked on an MPI implementation with FPGAs using the cloudFPGA platform [16]. Ringlein et al. imple- mented an MPI runtime, called ZRLMPI (with part of the MPI API), in both the FPGA and CPU. The ZRLMPI execution model is similar to MPI: all nodes have the same entry point to start execution. Although the same code targets both CPU and FPGA, it is optimized at compile-time depending on the executing device. Their proposal is not limited to only one CPU in the cluster, which can also participate in the application. Nevertheless, there is no task-level parallelism inside the FPGA. It can be seen as a single accelerator that executes HLS code. In the ZRLMPI approach, the cluster conﬁguration must be known at compile-time, so the bitstream must be recompiled every time the conﬁguration changes. With OmpSs@FPGA, the same bitstream and binary executable can be used for any number of FPGAs. The power consumption numbers for the FPGAs, 15 watts per FPGA, were measured in real-time by the cluster hardware. We can report that these numbers validate the power estimation generated by the Vivado tool for the considered bitstream. The MareNostrum power is taken from the Slurm manager reports, with 276 watts per node. C. Results The host code is modiﬁed to include calls to the OpenCL API. Sommer el al. [14] also modify the Clang compiler to extract the code in the OpenMP target regions to compile them with an HLS toolﬂow. However, instead OpenCL they use Xilinx Vivado HLS to synthesize the FPGA code. Then they use a custom toolchain, called ThreadPoolComposer (TPC), which similarly to AIT, generates automatically the whole FPGA design. To com- municate with the host, they implement a libomptarget plugin that is used by OpenMP for the target regions. This library is linked to the TPC API that implements the low- level communication with the device. More related work for OpenMP FPGA ofﬂoading can be found in [15]. the limited resources implementation, it uses 256-bit vectors, 4 cores per node, and 2048 block size. The MareNostrum vec512 row is the version with full resources and uses 512- bits vector lengths, 48 cores per node, and 512 block size. With similar resources, the FPGA cluster proves to have 1.6 better performance than the CPU. However, the vec512 version is 7 times faster than the FPGA. Although in terms of efﬁciency, the FPGA version is above the other two. We also have to take into account that we are comparing a mid-range FPGA with a high-performance CPU. Moreover, the Kintex Ultrascale technology is four years older than the Intel Xeon Platinum CPUs. The Kintex uses 20nm TSMC lithography, while the Xeon uses Intel 14nm. Even with this disadvantage, the FPGA consumes one order of magnitude less power than the vec256 version and 18 times less than vec512. Looking at the performance per watt, the FPGA version is above both MareNostrum implementations. the limited resources implementation, it uses 256-bit vectors, 4 cores per node, and 2048 block size. The MareNostrum vec512 row is the version with full resources and uses 512- bits vector lengths, 48 cores per node, and 512 block size. With similar resources, the FPGA cluster proves to have 1.6 better performance than the CPU. However, the vec512 version is 7 times faster than the FPGA. Although in terms of efﬁciency, the FPGA version is above the other two. We also have to take into account that we are comparing a mid-range FPGA with a high-performance CPU. Moreover, the Kintex Ultrascale technology is four years older than the Intel Xeon Platinum CPUs. VIII. CONCLUSION [8] B. Ringlein, F. Abel, A. Ditter, B. Weiss, C. Hagleitner, and D. Fey, “System architecture for network-attached fpgas in the cloud using partial reconﬁguration,” in 2019 29th International Conference on Field Programmable Logic and Applications (FPL), 2019, pp. 293–300. With the introduction of FPGAs in data centers, a new paradigm has appeared for distributed parallel applications. In this work, we present extensions to an existing task- based programming model and a framework that can take advantage of the internal parallelism of an FPGA, as well as its inter and intra-communication capabilities to scale out parallel applications. With a single C/C++ code annotated with pragmas and a single bitstream compilation, the programmer can run an application in an FPGA cluster of any size. We have introduced an implementation of OmpSs@cloudFPGA, with support to an MPI-like API. We evaluated our framework on the IBM cloudFPGA research project equipped with mid-range network-attached FPGAs. The results show that the N-body application scales linearly to 56 nodes when implemented with OmpSs@cloudFPGA with near-perfect scalability. Fur- thermore, we have compared its performance and power consumption with the MareNostrum 4 supercomputer. The results show that even when using all the CPU resources, FPGAs still outperform CPUs’ performance per watt. Fur- thermore, when comparing similar CPU and FPGA resources, the FPGA also provides better absolute performance than the CPU. In addition, the close to perfect linear scaling of FPGA accelerators shows the general advantage of the distributed task-programming approach to accelerate future high-performance applications at a low energy cost and with the same programmability as CPUs. pp [9] IBM (2021) ”cloudFPGA”. [Online]. Available: https://github.com/ cloudfpga [10] K. Sano, S. Abiko, and T. Ueno, “Fpga-based stream computing for high-performance n-body simulation using ﬂoating-point dsp blocks,” in Proceedings of the 8th International Symposium on Highly Efﬁcient Accelerators and Reconﬁgurable Technologies, ser. HEART2017. As- sociation for Computing Machinery, 2017. [11] E. Del Sozzo, M. Rabozzi, L. Di Tucci, D. Sciuto, and M. D. Santam- brogio, “A scalable fpga design for cloud n-body simulation,” in 2018 IEEE 29th International Conference on Application-speciﬁc Systems, Architectures and Processors (ASAP), 2018, pp. 1–8. [12] J. Makino and H. Daisaka, “Grape-8 – an accelerator for gravitational n-body simulation with 20.5gﬂops/w performance,” in SC ’12: Proceed- ings of the International Conference on High Performance Computing, Networking, Storage and Analysis, 2012, pp. 1–10. [13] M. Knaust, F. Mayer, and T. VII. RELATED WORK There have been other efforts to implement the N-body on FPGAs and even ASICS. In [10], Sano et al. implement a full custom FPGA design that solves the N-body problem on a single Intel Arria10 FPGA, achieving 10.944 Gpairs/s. In [6], de Haro et al. uses OmpSs@FPGA to execute the N-body on a Xilinx Alveo U200 board reaching 37.62 Gpairs/s with a performance per watt of 0.58. Del Sozzo et al. [11] also presented a custom N-body implementation on a Xilinx Virtex Ultrascale+ board (VU9P). However, they only accelerate the force calculation and update the positions and velocities with software. They reach 13.441 Gpairs/s with a performance per watt of 0.672. ASICs have also been developed to accelerate the computation of the N-body. GRAvity PipE (GRAPE) [12] is an ASIC that computes the force interaction between particles, with 48 pipelines at 250MHz. The update part is still done in the CPU, which communicates with the chip through PCIe and an FPGA that works as a bridge. Its theoretical peak performance is 480 Gﬂops which translates to 24Gpairs/s, with a performance per watt of 0.5. Even though single-node performance of the Kintex is the lowest at 6.2 Gpairs/s our implementation with 56 nodes outperforms De Matteis et al. [17] also present an MPI-like model for FPGAs, more focused on streaming data from point-to-point rather than sending bulk messages. Nevertheless, they target a set of FPGAs connected with a speciﬁc topology, and message routing is part of their framework. They do not parallelize the application inside one FPGA using multiple accelerators. Naylor et al. [18] introduce a RISCV-based multithreaded CPU for FPGAs and an FPGA cluster of 12 Stratix V. They present a message passing API similar to MPI to communicate data between cores of different FPGAs. Like De Matteis et al., they rely on a custom and ﬁxed topology and hardware. They do not use hardware acceleration to execute the application, as it runs on the CPUs, but they accelerate global synchronization of nodes with hardware-assisted primitives. There have been other extensions to the OmpSs program- ming model to support cluster ofﬂoading with tasks [19]. Sainz et al. present a way to create MPI clusters dynamically and ofﬂoad created tasks to a speciﬁc rank in a speciﬁc cluster. To achieve this, they use MPI_Comm_spawn to create and isolate each cluster and provide channels to communicate between clusters. VII. RELATED WORK However, they test their system on Intel Xeon Phi clusters, with up to 128 nodes. [2] J. Weerasinghe, F. Abel, C. Hagleitner, and A. Herkersdorf, “Enabling fpgas in hyperscale data centers,” in 2015 IEEE 12th Intl Conf on Ubiquitous Intelligence and Computing and 2015 IEEE 12th Intl Conf on Autonomic and Trusted Computing and 2015 IEEE 15th Intl Conf on Scalable Computing and Communications and Its Associated Workshops (UIC-ATC-ScalCom), Beijing, China, August 10-14, 2015. IEEE Computer Society, 2015, pp. 1078–1086. [3] A. M. Caulﬁeld et al., “A cloud-scale acceleration architecture,” in 2016 49th Annual IEEE/ACM International Symposium on Microarchitecture (MICRO), 2016, pp. 1–13. Xiong et al. [20] introduce an FPGA implementation of the MPI_Irecv operation. They ofﬂoad only the message matching part of the MPI runtime to the FPGA to accelerate MPI applications on CPUs. The architecture they propose does a similar job to the message receiver of OmpSs@cloudFPGA. Their implementation supports many outstanding receive re- quests using a two-level queue. However, we are more limited in resources because a big part of the FPGA is used for the user application, limiting the message receiver’s capabilities. [4] A. Fern´andez, V. Beltran, X. Martorell, R. M. Badia, E. Ayguad´e, and J. Labarta, “Task-based programming with ompss and its application,” in Euro-Par 2014: Parallel Processing Workshops, 2014, pp. 601–612. [5] (2022, February) BSC Programming Models, OmpSs and OmpSs-2 documentation and code. [Online]. Available: https://pm.bsc.es [6] J. M. de Haro et al., “Ompss@fpga framework for high performance fpga computing,” IEEE Transactions on Computers, vol. 70, no. 12, pp. 2029–2042, 2021. [7] F. Abel, J. Weerasinghe, C. Hagleitner, B. Weiss, and S. Paredes, “An fpga platform for hyperscalers,” in 2017 IEEE 25th Annual Symposium on High-Performance Interconnects (HOTI), 2017, pp. 29–32. VIII. CONCLUSION Steinke, “Openmp to fpga ofﬂoading prototype using opencl sdk,” in 2019 IEEE International Parallel and Distributed Processing Symposium Workshops (IPDPSW), 2019, pp. 387–390. [14] L. Sommer, J. Korinth, and A. Koch, “Openmp device ofﬂoading to fpga accelerators,” in 2017 IEEE 28th International Conference on Application-speciﬁc Systems, Architectures and Processors (ASAP), 2017, pp. 201–205. [15] F. Mayer, M. Knaust, and M. Philippsen, “Openmp on fpgas—a survey,” in OpenMP: Conquering the Full Hardware Spectrum. Cham: Springer International Publishing, 2019, pp. 94–108. [16] B. Ringlein, F. Abel, A. Ditter, B. Weiss, C. Hagleitner, and D. Fey, “Programming reconﬁgurable heterogeneous computing clusters using mpi with transpilation,” in 2020 IEEE/ACM International Workshop on Heterogeneous High-performance Reconﬁgurable Computing (H2RC), 2020, pp. 1–9. [17] T. De Matteis, J. de Fine Licht, J. Ber´anek, and T. Hoeﬂer, “Streaming message interface: High-performance distributed memory programming on reconﬁgurable hardware,” in Proceedings of the International Con- ference for High Performance Computing, Networking, Storage and Analysis, ser. SC ’19. Association for Computing Machinery, 2019. ACKNOWLEDGMENTS This work has been done in the context of the IBM/BSC Deep Learning Center initiative. This work has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 754337 (EuroEXA), from Spanish Government (PID2019-107255GB- C21/AEI/10.13039/501100011033), and from Generalitat de Catalunya (2017-SGR-1414 and 2017-SGR-1328). [18] M. Naylor, S. W. Moore, and D. Thomas, “Tinsel: A manythread overlay for fpga clusters,” in 2019 29th International Conference on Field Programmable Logic and Applications (FPL), 2019, pp. 375–383. [19] F. Sainz, J. Bell´on, V. Beltran, and J. Labarta, “Collective ofﬂoad for heterogeneous clusters,” in 2015 IEEE 22nd International Conference on High Performance Computing (HiPC), 2015, pp. 376–385. [20] Q. Xiong, A. Skjellum, and M. C. Herbordt, “Accelerating mpi message matching through fpga ofﬂoad,” in 2018 28th International Conference on Field Programmable Logic and Applications (FPL), 2018, pp. 191– 1914. [1] A. Putnam et al., “A reconﬁgurable fabric for accelerating large-scale datacenter services,” in 2014 ACM/IEEE 41st International Symposium on Computer Architecture (ISCA), 2014, pp. 13–24. REFERENCES [1] A. Putnam et al., “A reconﬁgurable fabric for accelerating large-scale datacenter services,” in 2014 ACM/IEEE 41st International Symposium on Computer Architecture (ISCA), 2014, pp. 13–24.
https://openalex.org/W3195300530	https://www.esaim-ps.org/articles/ps/pdf/2022/01/ps210102.pdf	English	null	Bayesian Sequential Composite Hypothesis Testing in Discrete Time	ESAIM. Probability and statistics	2,022	cc-by	9,683	Mathematics Subject Classiﬁcation. 62L10, 60G40, 62C10 Mathematics Subject Classiﬁcation. 62L10, 60G40, 62C10. Received October 14, 2021. Accepted April 11, 2022. ESAIM: Probability and Statistics www.esaim-ps.org ESAIM: Probability and Statistics www.esaim-ps.org ESAIM: PS 26 (2022) 265–282 https://doi.org/ 10.1051/ps/2022005 1. Introduction Assume that a sequence of random variables X1, X2, . . . is observed sequentially, and that the sequence is drawn from a one-parameter family of distributions depending on a real-valued random variable Θ in such a way that X1, X2, . . . are independent (conditional on Θ). Consider a tester who wants to test the two alternative hypotheses H0 : Θ ≤θ0, H1 : Θ > θ0, H0 : Θ ≤θ0, where θ0 ∈R is a given constant (the ’threshold’). In the presence of an observation cost, a tradeoﬀbetween statistical precision and costly observation arises. where θ0 ∈R is a given constant (the ’threshold’). In the presence of an observation cost, a tradeoﬀbetween statistical precision and costly observation arises. In a Bayesian formulation of the problem, the tester’s initial belief is described by a prior distribution µ for the unknown parameter Θ. Denote by T the set of FX-stopping times with values in N0 = {0, 1, 2...}, where FX = {FX n }∞ n=1 is the ﬁltration generated by the observation process X = {Xn, n ≥0}. Given a stopping time τ ∈T , let Dτ be the set of FX τ -measurable random variables D with values in {0, 1}. The random variable D here represents the decision of the tester, with ’D = d’ representing that hypothesis Hd is accepted, d = 0, 1. ∗The ﬁrst author gratefully acknowledges support from Vetenskapsr˚adet (The Swedish Research C The ﬁrst author gratefully acknowledges support from Vetenskapsr˚adet (The Swedish Research Council Keywords and phrases: Sequential analysis, hypothesis testing, exponential family, optimal stoppin words and phrases: Sequential analysis, hypothesis testing, exponential family, optimal stopping. Department of Mathematics, Uppsala University, Box 256, 75105 Uppsala, Sweden. Corresponding author: ekstrom@math.uu.se This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Corresponding author: ekstrom@math.uu.se BAYESIAN SEQUENTIAL COMPOSITE HYPOTHESIS TESTING IN DISCRETE TIME∗ Erik Ekstr¨om** and Yuqiong Wang Abstract. We study the sequential testing problem of two alternative hypotheses regarding an unknown parameter in an exponential family when observations are costly. In a Bayesian setting, the problem can be embedded in a Markovian framework. Using the conditional probability of one of the hypotheses as the underlying spatial variable, we show that the cost function is concave and that the posterior distribution becomes more concentrated as time goes on. Moreover, we study time monotonic- ity of the value function. For a large class of model speciﬁcations, the cost function is non-decreasing in time, and the optimal stopping boundaries are thus monotone. Mathematics Subject Classiﬁcation. 62L10, 60G40, 62C10. © The authors. Published by EDP Sciences, SMAI 2022 © The authors. Published by EDP Sciences, SMAI 2022 This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. E. EKSTR ¨OM AND Y. WANG 266 We deﬁne the cost V := inf τ∈T inf D∈Dτ {P(D = 1, Θ ≤θ0) + P(D = 0, Θ > θ0) + cE[τ]} , (1.1 (1.1) where c > 0 is a given and ﬁxed cost of each observation. where c > 0 is a given and ﬁxed cost of each observation. The case when µ is a two-point distribution with P(Θ = θ2) = π, P(Θ = θ1) = 1 −π, where π ∈(0, 1) and θ1 ≤θ0 < θ2 was studied in the classical reference [24], see also [21, Chapter 4.1]. It turns out that the statistical problem (1.1) can be reduced to an optimal stopping problem in terms of the posterior probability process Πn := P(Θ = θ2\|Fn), and since Π in this case is a (time-homogeneous) Markov process, the stopping problem can be embedded in a Markovian framework. It is shown in [21] that the cost function is concave in the prior belief Π0 = π; as a consequence, the continuation region is an interval, and the optimal stopping time is the ﬁrst exit time from this interval (the latter property was also obtained in [24]). where π ∈(0, 1) and θ1 ≤θ0 < θ2 was studied in the classical reference [24], see also [21, Chapter 4.1]. It turns out that the statistical problem (1.1) can be reduced to an optimal stopping problem in terms of the posterior probability process Πn := P(Θ = θ2\|Fn), and since Π in this case is a (time-homogeneous) Markov process, the stopping problem can be embedded in a Markovian framework. It is shown in [21] that the cost function is concave in the prior belief Π0 = π; as a consequence, the continuation region is an interval, and the optimal stopping time is the ﬁrst exit time from this interval (the latter property was also obtained in [24]). In the current article we relax the assumption about a two-point prior distribution and study the sequential analysis problem (1.1) in a Bayesian set-up for general prior distributions µ. © The authors. Published by EDP Sciences, SMAI 2022 To do that, we impose a one- dimensional exponential structure on the distribution of Xk. As in [21], the conditional probability process Π is then still Markovian; however, Π is in general time-inhomogenous, which leads to time-dependence in the cost function, and the study of optimal strategies is more involved. In the absence of explicit solutions for the cost and the optimal strategy, we focus on structural properties of the solution. In particular, we prove that spatial concavity of the cost function holds regardless of the prior distribution. We also show a concentration result for the posterior distribution, which combined with the concavity result has implications for the monotonicity of the cost with respect to the time parameter. While the analysis in the classical studies [21, 24] is restricted to the case with a two-point distribution for the unknown parameter, it allows for an arbitrary distribution of the observation sequence. In contrast, our set-up allows for an arbitrary distribution of the unknown parameter, which then forces us to restrict our attention to the exponential class of distributions for the observations. We also note that the key complication when abondoning the two-point distribution is the time-dependencies that are intrinsic in the problem; by exploiting the fact of the exponential class that the possible posterior distributions at each time point belongs to a one- parameter family of distributions, we manage to overcome irregularities with respect to time. In fact, we use a concentration result for the posterior distribution to prove that a large class of models exhibits monotonicity with respect to time. 1.1. Literature review The problem of sequential testing of an unknown parameter has attracted much attention in the statistical literature, with [24] as an early reference covering the case of two simple hypotheses and independent and identically distributed observations. Sequential testing of composite hypotheses in a discrete time setting with Bernoulli distributed observations is studied in [17, 18], with linear penalty for wrong decisions and relying on a conjugate prior for the unknown parameter. In [23], Sobel studies sequential testing of composite hypotheses for an arbitrary class of distributions in the exponential family and with a general prior distribution of the unknown parameter. In a key result, he establishes the existence of two stopping boundaries beyond which it is optimal to stop. Related literature in discrete time, but more focused on the case of sequential estimation, includes [1, 6]. Another strand of literature has focused on continuous time approximations of sequential testing problems and their connections with free boundary problems. For the sequential testing of two simple hypotheses, [22] solved the problem of determining the unknown drift of a Brownian motion, and [19] solved the corresponding sequential testing problem of determining an unknown intensity of a Poisson process. In [2], a problem with 267 BAYESIAN SEQUENTIAL COMPOSITE HYPOTHESIS TESTING IN DISCRETE TIME composite hypotheses was studied in continuous time and for a normal prior distribution, with a ‘0-1’ loss function for wrong decisions (as in (1.1)), and in a series of papers (see [7] and the references therein), Chernoﬀ studied the same problem but with linear penalty functions. In the case of sequential composite hypothesis testing, explicit solutions are rare, and a main focus in this literature is on deriving asymptotics of the problem as the cost of observation tends to zero, as well as asymptotically optimal solutions (e.g. [3, 16, 20]) and deriving bounds for the stopping boundaries. More recent literature has focused on diﬀerent variants of these continuous-time problems. To mention a few, [12] studies a version with ﬁnite horizon, [8] studies a setting with combined learning from several Brownian motions and compound Poisson processes, and [11] studies Wiener sequential testing in a multi-dimensional set-up. All these papers study simple hypotheses, i.e. set-ups where the unknown parameters can take only two possible values. 1.1. Literature review In [25], a hypothesis testing problem for a case with three possible drifts is examined, and in [10] a composite hypothesis problem for the drift of a Wiener process is studied with a general prior distribution. Moreover, [9] study a sequential estimation problem for a Wiener process in the same set-up. Key to the analysis in [9, 10] is the choice of appropriate variables. In fact, in [10] it is shown that if instead of the observation process one uses the conditional probability Π as state variable, then the corresponding continuation region is shrinking in time; a similar result holds for sequential least-square estimation if one uses the conditional expectation as state variable. 1.2. Our contribution Even though the sequential composite hypothesis testing problem is very natural, it seems that qualitative studies of properties of the stopping boundaries are missing in the literature, even in the fundamental case of the exponential family. In the current article, we study the sequential composite hypothesis testing problem (1.1) using a Markovian approach. Our analysis is general in the sense that we treat the whole one-parameter exponential family with arbitrary prior distribution, and we thus do not rely on conjugate priors. Following [10], we use the conditional probability process as the underlying state variable, and we show that a concavity result holds in these coordinates. An immediate consequence of concavity is that each time section of the continuation region is an interval. Another reason to study concavity is the close connection between spatial concavity and monotonicity results with respect to time; this connection is well-known in continuous time models of option pricing (cf. [14, 15]), but still unexplored in the current setting. Our analysis shows that spatial concavity implies monotonicity results in time also in the current setting, provided that an assumption of stochastic domination holds. Consequently, under this assumption, the continuation region (in terms of the conditional probability) shrinks in time; in addition to being a fundamental property in itself, time monotonicity also is useful for the derivation of upper bounds on the growth of the stopping boundaries when translating back to the observation coordinates. Moreover, we provide a concentration result for the posterior distribution, showing that the posterior becomes more concentrated along level curves of the conditional probability; this can then be used to identify large classes of problem speciﬁcations for which the assumption of stochastic dominance is fulﬁlled, thereby guaranteeing time monotonicity. , y g g y The paper is organised as follows. In Section 2 we recall some basic properties of statistical inference in the exponential family, and we introduce the notion of π-level curves along which the value of the conditional probability Π is constant. In Section 3, we provide a Markovian embedding of (1.1), and we prove that the embedded cost function is spatially concave. In Section 4 we prove that the posterior distribution becomes more concentrated about the threshold θ0 along level curves. Sections 5 and 6 deal with the question whether the value function is monotone with respect to the time parameter. 2. Preliminaries on the exponential family In this article, we will consider the case of a one-dimensional exponential family of distributions for Xk, k ≥1. More precisely, let ν be a σ-ﬁnite measure ν on R, and deﬁne 268 E. EKSTR ¨OM AND Y. WANG B(u) := log Z R exp{ux}ν(dx) and N = u ∈R : Z R exp{ux}ν(dx) < ∞ h B(u) := log Z R exp{ux}ν(dx) B(u) := log Z R exp{ux}ν(dx) B(u) := log Z R exp{ux}ν(dx) N = u ∈R : Z R exp{ux}ν(dx) < ∞ B(u) < ∞ for u ∈N. For u ∈N, let for u ∈N. For u ∈N, let pu(x) := exp{ux −B(u)} (2.1) for u ∈N. For u ∈N, let for u ∈N. For u ∈N, let pu(x) := exp{ux −B(u)} (2.1) pu(x) := exp{ux −B(u)} (2.1) (2.1) so that R R puν(dx) = 1. We assume that the distribution of Xk, conditional on Θ = u, is so that R R puν(dx) = 1. We assume that the distribution of Xk, conditional on Θ = u, is P(Xk ∈A\|Θ = u) = Z A pu(x)ν(dx). (2.2) (2.2) Remark 2.1. In some literature, the notion of an exponential family allows for densities on the form pu(x) = exp{η(u)T(x) −B(u)}, and the case (2.2) in which η(u) = u and T(x) = x is then refered to as a natural exponential family. Using the transformed variables η = η(u) and T = T(x), an exponential form can be transformed into a natural form, so we may consider the natural form (as above) without loss of generality. Remark 2.1. In some literature, the notion of an exponential family allows for densities on the form pu(x) = exp{η(u)T(x) −B(u)}, and the case (2.2) in which η(u) = u and T(x) = x is then refered to as a natural exponential family. Using the transformed variables η = η(u) and T = T(x), an exponential form can be transformed into a natural form, so we may consider the natural form (as above) without loss of generality. We start with some well-known results. Lemma 2.2. We have that (i) B is convex, and N is an interval. (i) B is convex, and N is an interval. Denote by N ◦the interior of N. Then Denote by N ◦the interior of N. 2. Preliminaries on the exponential family Then (ii) all derivatives of B exist on N ◦, and they are given by the expressions obtained by formally diﬀerentiating inside the integral. In particular, B′(u) = R R x exp{ux}ν(dx) R R exp{ux}ν(dx) = E[X1\|Θ = u]; (iii) the function u 7→E[G(X1)\|Θ = u] is non-decreasing for any non-decreasing function G : R →R. Proof. For (i) and (ii) we refer to ([5], Thm. 1.13) and ([5], Thm. 2.2), respectively. For (iii), we have (iii) the function u 7→E[G(X1)\|Θ = u] is non-decreasing for any non-decreasing function G : R →R. Proof. For (i) and (ii) we refer to ([5], Thm. 1.13) and ([5], Thm. 2.2), respectively. For (iii), we have ∂ ∂uE[G(X1)\|Θ = u] = ∂ ∂u Z R G(x)pu(x) ν(dx) = Z R G(x)(x −B′(u))pu(x) ν(dx) = E[G(X1)X1\|Θ = u] −E[G(X1)\|Θ = u]E[X1\|Θ = u] ≥0, where the ﬁnal inequality is due to the fact the covariance of two non-decreasing functions evaluated at the same random variable is non-negative. where the ﬁnal inequality is due to the fact the covariance of two non-decreasing functions evaluated at the same random variable is non-negative. BAYESIAN SEQUENTIAL COMPOSITE HYPOTHESIS TESTING IN DISCRETE TIME 269 We use a Bayesian set-up in which the unknown parameter Θ has a given prior distribution µ; we assume that µ is a measure on N ◦, and we denote the support of µ by S. Moreover, denote S+ = S ∩(θ0, ∞) & S−= S ∩(−∞, θ0] = S \ S+. Naturally, to avoid degenerate cases we assume that 0 < µ(S+) < 1. Next, by standard means, the optimization problem (1.1) can be reduced to an optimal stopping problem, i.e. a problem in which only one optimization (namely over τ) takes place. In fact, given a stopping time τ ∈T , an optimal decision rule D ∈Dτ is given by D = ( 0 if Πτ ≤1/2 1 if Πτ > 1/2, where the posterior probability process Π is given by Πn := P(Θ > θ0\|FX n ). Consequently, Consequently, V = inf τ∈T E [Πτ ∧(1 −Πτ) + cτ] , V = inf τ∈T E [Πτ ∧(1 −Πτ) + cτ] , where a ∧b = min{a, b}. To derive an expression for Π1, note that P(Θ > θ0\|X1 = x1) = R S+ pu(x1)µ(du) R S pu(x1)µ(du) , so Π1 = R S+ pu(X1)µ(du) R S pu(X1)µ(du) . 2. Preliminaries on the exponential family Moreover, R S+ euy−nB(u)µ(du) R S−euy−nB(u)µ(du) ≥ R S+ e(u−θ0)y−nB(u)µ(du) R S−e−nB(u)µ(du) →∞ R S+ euy−nB(u)µ(du) R S−euy−nB(u)µ(du) ≥ R S+ e(u−θ0)y−nB(u)µ(du) R S−e−nB(u)µ(du) →∞ as y →∞, so q(n, y) →1 as y →∞. A similar argument shows that q(n, y) →0 as y →−∞, so q(n, ·) is surjective. For each ﬁxed value π ∈(0, 1), denote by y(n, π) the unique value such that q(n, y(n, π)) = π. The set {(n, y(n, π), n ≥0} consists of all points (n, y) with q(n, y) = π, and is refered to as the π−level curve. Since the function y 7→q(n, y) is a bijection, two π−level curves with diﬀerent π-values never intersect. Furthermore, they are ordered so that if π1 < π2, then y(n, π1) < y(n, π2). 2. Preliminaries on the exponential family Π1 = R S+ pu(X1)µ(du) R S pu(X1)µ(du) . R S pu( 1)µ( ) More generally, at time n, given observations X1 = x1, X2 = x2, . . . , Xn = xn, we have by independence R n More generally, at time n, given observations X1 = x1, X2 = x2, . . . , Xn = xn, we have by independence P(Θ > θ0\|X1 = x1, . . . , Xn = xn) = R S+ Qn i=1 pu(xi)µ(du) R S Qn i=1 pu(xi)µ(du) = R S+ exp{u Pn i=1 xi −nB(u)}µ(du) R S exp{u Pn i=1 xi −nB(u)}µ(du) . Thus, denoting Thus, denoting Yn := n X i=1 Xi we have Πn = q(n, Yn), we have Πn = q(n, Yn), Πn = q(n, Yn), E. EKSTR ¨OM AND Y. WANG 270 where q(n, y) := R S+ euy−nB(u)µ(du) R S euy−nB(u)µ(du) . Remark 2.3. The fact that Yn is a suﬃcient statistic in any exponential family is well-known. Moreover, also a converse holds: under some mild conditions it holds that any family of distributions that admits a real-valued suﬃcient statistic for sample size larger than one is a one-parameter exponential family, see e.g. [4, 13]. We denote by µn,y(du) := euy−nB(u)µ(du) R S euy−nB(u)µ(du) µn,y(du) := euy−nB(u)µ(du) R S euy−nB(u)µ(du) the posterior distribution of Θ at time n conditional on Yn = y. Note that the prior distribution satisﬁes µ = µ0,0; however, for reasons of Markovian embedding, below we will consider simultaneously the whole family {µ0,y, y ∈R} of alternative prior distributions. the posterior distribution of Θ at time n conditional on Yn = y. Note that the prior distribution satisﬁes µ = µ0,0; however, for reasons of Markovian embedding, below we will consider simultaneously the whole family {µ0,y, y ∈R} of alternative prior distributions. Lemma 2.4. The function y 7→q(n, y) : R →(0, 1) is an increasing bijection for each ﬁxed n. Proof. We have Proof. We have ∂q(n, y) ∂y = R S+ ueuy−nB(u)µ(du) R S euy−nB(u)µ(du) − R S ueuy−nB(u)µ(du) R S+ euy−nB(u)µ(du) ( R S euy−nB(u)µ(du))2 (2.3) = E[Θ1{Θ>θ0}\|Yn = y] −P(Θ > θ0\|Yn = y)E[Θ\|Yn = y]. (2.3) Since µn,y assigns positive mass on each side of the threshold θ0, the above covariance is strictly positive. Thus ∂q(n,y) ∂y > 0, so q(n, ·) is strictly increasing. 3. Markovian embedding It follows from Lemma 2.4 that the process Π is a (time-inhomogeneous) Markov process, and we can write the Π-process in terms of Y as Πn = Z S+ µn,Yn(du) = R S+ pu(Xn)µn−1,Yn−1(du) R S pu(Xn)µn−1,Yn−1(du) . BAYESIAN SEQUENTIAL COMPOSITE HYPOTHESIS TESTING IN DISCRETE TIME 271 Furthermore, this allows us to embed the optimal stopping problem (1.1) as a time-dependent problem in terms of the Markov process Π as Furthermore, this allows us to embed the optimal stopping problem (1.1) as a time-dependent problem in terms of the Markov process Π as V (n, π) = inf τ∈T En,π[Πτ+n ∧(1 −Πτ+n) + cτ]. (3.1) (3.1) Here Pn,π(·) := P(·\|Πn = π) is the probability measure under which Θ has distribution µn,y(n,π). We emphasize that V : N0 × (0, 1) →[0, ∞), i.e. π can take any value in (0, 1). Here Pn,π(·) := P(·\|Πn = π) is the probability measure under which Θ has distribution µn,y(n,π). We emphasize hat V : N0 × (0, 1) →[0, ∞), i.e. π can take any value in (0, 1). Lemma 3.1. The value function V (n, π) satisﬁes Lemma 3.1. The value function V (n, π) satisﬁes V (n −1, π) = min{π ∧(1 −π), c + En−1,π[V (n, Πn)]}. V (n −1, π) = min{π ∧(1 −π), c + En−1,π[V (n, Πn)]}. oof. This follows directly from the Markovian structure of the process Π. Proof. This follows directly from the Markovian structure of the process Π. Proof. This follows directly from the Markovian structure of the process Π. Proof. This follows directly from the Markovian structure of the process Π Lemma 3.2. Let f : [0, 1] →[0, ∞) be a concave function. Then π 7→En,π[f(Πn+1)] is concave on (0, 1). Lemma 3.2. Let f : [0, 1] →[0, ∞) be a concave function. Then π 7→En,π[f(Πn+1)] is concave on (0, 1). Proof. To simplify the notation, we prove the statement for n = 0. Moreover, we will assume that f is twice continuously diﬀerentiable; the general case follows readily by approximation. F h First note that First note that First note that E0,π[f(Π1)] = Z R f α(x, π) β(x, π) β(x, π) dx, where α(x, π) = Z S+ pu(x)µ0,y(0,π)(du), β(x, π) = Z S pu(x)µ0,y(0,π)(du). α(x, π) = Z S+ pu(x)µ0,y(0,π)(du), β(x, π) = Z S pu(x)µ0,y(0,π)(du). 3. Markovian embedding Deﬁne H1(z) := f(z) + (1 −z)f ′(z) H1(z) := f(z) + (1 −z)f ′(z) and H2(z) := f(z) −zf ′(z). H2(z) := f(z) −zf ′(z). Straightforward diﬀerentiation yields Straightforward diﬀerentiation yields Straightforward diﬀerentiation yields Straightforward diﬀerentiation yields Straightforward diﬀerentiation yields ∂2E0,π[f(Π1)] ∂π2 = Z R f(α β )βππ + f ′(α β )(βαππ −αβππ)2 β + f ′′(α β )(βαπ −αβπ)2 β3 dx ≤ Z R f(α β )βππ + f ′(α β )(βαππ −αβππ)2 β dx = Z R αππH1 α β + (β −α)ππH2 α β dx = I1 + I2, E. EKSTR ¨OM AND Y. WANG 272 where where I1 := Z R αππH1 α β dx & I2 := Z R (β −α)ππH2 α β dx Note that H1 is decreasing on (0, 1), and H2 is increasing. Furthermore, by Lemma 2.4, α(x,π) β(x,π) increases in x. We will show that Note that H1 is decreasing on (0, 1), and H2 is increasing. Furthermore, by Lemma 2.4, α(x,π) β(x,π) increases in x. We will show that I1 ≤0 & I2 ≤0. I1 ≤0 & I2 ≤0. I1 ≤0 & I2 ≤0. I1 ≤0 & I2 ≤0. To do that, ﬁrst note that To do that, ﬁrst note that To do that, ﬁrst note that To do that, ﬁrst note that α(x, π) = Z S+ pu(x) euy(0,π) R R euy(0,π)µ(du)µ(du), α(x, π) = Z S+ pu(x) euy(0,π) R R euy(0,π)µ(du)µ(du), so I1 = Z S+ euy(0,π) R R euy(0,π)µ(du) ππ Z S pu(x)H1 α(x, π) β(x, π) dxµ(du) = Z S+ euy(0,π) R R euy(0,π)µ(du) ππ E H1 α(X1, π) β(X1, π) \|Θ = u µ(du). (3.2) (3.2) By Lemma 2.2, the function By Lemma 2.2, the function By Lemma 2.2, the function u 7→E H1 α(X1, π) β(X1, π) \|Θ = u (3.3) (3.3) is non-increasing. is non-increasing. is non increasing. 3. Markovian embedding To study the ﬁrst factor of the integrand in (3.2), denote g(y) := R R euyµ(du) and note that g To study the ﬁrst factor of the integrand in (3.2), denote g(y) := R R euyµ(du) and note that To study the ﬁrst factor of the integrand in (3.2), denote g(y) := R R euyµ(d ∂ ∂π euy(0,π) R S euy(0,π)µ(du) = ∂ ∂y( euy g(y)) π′(y) y=y(0,π) , where where π(y) := R S+ euyµ(du) R S euyµ(du) . Consequently, Using BAYESIAN SEQUENTIAL COMPOSITE HYPOTHESIS TESTING IN DISCRETE TIME 273 and and π′ = g R S+ ueuyµ(du) −g′ R S+ euyµ(du) g2 , straightforward calculations show that ∂2 ∂π2 euy(0,π) R S euy(0,π)µ(du) = euy(0,π)F(u) (π′)3g3 , where F(u) = u2 g Z S+ ueuyµ(du) −g′ Z S+ euyµ(du) + u g′′ Z S+ euyµ(du) −g Z S+ u2euyµ(du) + g′ Z S+ u2euyµ(du) −g′′ Z S+ ueuyµ(du). π′ = g R S+ ueuyµ(du) −g′ R S+ euyµ(du) g2 , straightforward calculations show that ∂2 ∂π2 euy(0,π) R S euy(0,π)µ(du) = euy(0,π)F(u) (π′)3g3 , where F(u) = u2 g Z S+ ueuyµ(du) −g′ Z S+ euyµ(du) + u g′′ Z S+ euyµ(du) −g Z S+ u2euyµ(du) + g′ Z S+ u2euyµ(du) −g′′ Z S+ ueuyµ(du). straightforward calculations show that ∂2 ∂π2 euy(0,π) R S euy(0,π)µ(du) = euy(0,π)F(u) (π′)3g3 , where F(u) = u2 g Z S+ ueuyµ(du) −g′ Z S+ euyµ(du) + u g′′ Z S+ euyµ(du) −g Z S+ u2euyµ(du) + g′ Z S+ u2euyµ(du) −g′′ Z S+ ueuyµ(du). Note that F is a quadratic function in u, and that the coeﬃcient of u2 is positive since Note that F is a quadratic function in u, and that the coeﬃcient of u2 is positive since Note that F is a quadratic function in u, and that the coeﬃcient of u2 is positive since g Z + ueuyµ(du) −g′ Z + euyµ(du) = g2Cov0,π(Θ, 1{Θ>θ0}) > 0. g Z S+ ueuyµ(du) −g′ Z S+ euyµ(du) = g2Cov0,π(Θ, 1{Θ>θ0}) > 0. Consequently, the set {F < 0} is a bounded interval (possibly empty). Moreover, since Consequently, the set {F < 0} is a bounded interval (possibly empty). Thus π 7→E0,π[f(Π1)] is concave. Thus π 7→E0,π[f(Π1)] is concave. Thus π 7→E0,π[f(Π1)] is concave. Theorem 3.3. The function π 7→V (n, π) is concave for each ﬁxed n ≥0. Theorem 3.3. The function π 7→V (n, π) is concave for each ﬁxed n ≥0. Proof. Deﬁne the cost function V N(n, π) as in (3.1), but with the inﬁmum being taken over stopping times τ ≤N −n (V N is then the value function in a problem with a ﬁnite horizon). By an iterated use of Lemma 3.1 and Lemma 3.2 and the fact that the minimum of two concave functions is concave, π 7→V N(n, π) is concave. Moreover, it is straightforward to check that V N(n, π) →V (n, π) as N →∞, and since the pointwise limit of concave functions is concave, the result follows. So far we have been working under the assumption that π ∈(0, 1). One can further extend the value function V to the boundary points π ∈{0, 1} by setting V (n, 0) = V (n, 1) = 0 for all n. In this way, V is deﬁned for every π ∈[0, 1] and the concavity is preserved. y [ , ] y p In accordance with standard stopping theory, we introduce the continuation region C by C := {(n, π) ∈N0 × [0, 1] : V (n, π) < π ∧(1 −π)}, and the stopping region D by D := {(n, π) ∈N0 × [0, 1] : V (n, π) = π ∧(1 −π)}. The stopping time τ ∗:= inf{k ≥0 : (n + k, Πn+k) ∈D} τ ∗:= inf{k ≥0 : (n + k, Πn+k) ∈D} is an optimal strategy for our testing problem. is an optimal strategy for our testing problem. is an optimal strategy for our testing problem. p gy g p The concavity of the value function has important implications for the structure of the continuation region. The concavity of the value function has important implications for the structure of the Corollary 3.4. There exist functions b1 : N0 →[0, 1 2] and b2 : N0 →[ 1 2, 1] such that C = {(n, π) ∈N0 × [0, 1] : b1(n) < π < b2(n)}. C = {(n, π) ∈N0 × [0, 1] : b1(n) < π < b2(n)}. Proof. Since V (n, 0) = V (n, 1) = 0, we have {(n, 0)} ∪{(n, 1)} ⊆D. 3. Markovian embedding Moreover, since π = R S+ euy(0,π)µ(du) g(y(0, π)) = 1 − R S−euy(0,π)µ(du) g(y(0, π)) , we have Z S+ ∂2 ∂π2 euy(0,π) R R euy(0,π)µ(du) µ(du) = Z S− ∂2 ∂π2 euy(0,π) R R euy(0,π)µ(du) µ(du) = 0. (3.4) (3.4) Therefore we must have Therefore we must have Therefore we must have F(θ0) < 0, so the interval {F < 0} ̸= ∅. Denote the end-points of this interval by u0 and u1, respectively, so that {F < 0} = (u0, u1), with θ0 ∈(u0, u1). Then, using (3.3) we ﬁnd that so the interval {F < 0} ̸= ∅. Denote the end-points of this interval by u0 and u1, respectively, so that {F < 0} = (u0, u1), with θ0 ∈(u0, u1). Then, using (3.3) we ﬁnd that I1 = Z S∩(−∞,u1) euy(0,π)F(u) (π′(y(0, π)))3g3(y(0, π))E H1 α(X1, π) β(X1, π) Θ = u µ(du) + Z S∩[u1,∞) euy(0,π)F(u) (π′(y(0, π)))3g3(y(0, π))E H1 α(X1, π) β(X1, π) Θ = u µ(du) ≤E H1 α(X1, π) β(X1, π) Θ = u1 Z (θ0,u1) euy(0,π)F(u) (π′(y(0, π)))3g3(y(0, π))µ(du) 274 E. EKSTR ¨OM AND Y. WANG +E H1 α(X1, π) β(X1, π) Θ = u1 Z [u1,∞) euy(0,π)F(u) (π′(y(0, π)))3g3(y(0, π))µ(du) = 0, where we used (3.4) in the last equality. Similarly, E h H2 α(X1,π) β(X1,π) \|Θ = u i increases in u, so Similarly, E h H2 α(X1,π) β(X1,π) \|Θ = u i increases in u, so Similarly, E h H2 α(X1,π) β(X1,π) \|Θ = u i increases in u, so I2 = Z S+∩(−∞,u0] euy(0,π)F(u) (π′(y(0, π)))3g3(y(0, π))E H2 α(X1, π) β(X1, π) Θ = u µ(du) + Z S−∩(u0,∞) euy(0,π)F(u) (π′(y(0, π)))3g3(y(0, π))E H2 α(X1, π) β(X1, π) Θ = u µ(du) ≤E H2 α(X1, π) β(X1, π) Θ = u0 Z S− euy(0,π)F(u) (π′(y(0, π)))3g3(y(0, π))µ(du) = 0. Thus π 7→E0,π[f(Π1)] is concave. The result then follows from concavity of π 7→V (n, π) and the piecewise linearity of π 7→π ∧(1 −π). Proof. Since V (n, 0) = V (n, 1) = 0, we have {(n, 0)} ∪{(n, 1)} ⊆D. The result then follows from concavity of π 7→V (n, π) and the piecewise linearity of π 7→π ∧(1 −π). 275 BAYESIAN SEQUENTIAL COMPOSITE HYPOTHESIS TESTING IN DISCRETE TIME Remark 3.5. In view of the bijection in Lemma 2.4, the fact that time sections of the continuation region are intervals in the (n, π)-coordinates implies that also time sections of the continuation region expressed in (n, y)-coordinates are intervals. This is a well-known result, see [23] (under somewhat diﬀerent assumptions). Remark 3.5. In view of the bijection in Lemma 2.4, the fact that time sections of the continuation region are intervals in the (n, π)-coordinates implies that also time sections of the continuation region expressed in (n, y)-coordinates are intervals. This is a well-known result, see [23] (under somewhat diﬀerent assumptions). 4. Concentration of the posterior distribution Recall that the mass above θ0 of the posterior distribution remains constantly equal to π along a π-level curve. In this section we show that the posterior distribution becomes more concentrated around θ0 along a level curve. This result, however natural it appears, seems to be new in the literature; for related results showing that the conditional variance of the mean-square estimate is a supermartingale, see [9]. Theorem 4.1. If a < θ0 < b, then Theorem 4.1. If a < θ0 < b, then heorem 4.1. If a < θ0 < b, then n 7→Pn,π(Θ ≤a) & n 7→Pn,π(Θ > b) are decreasing. are decreasing. Proof. For the ﬁrst claim, it suﬃces to show that P0,π(Θ ≤a) ≥P1,π(Θ ≤a) (4.1) P0,π(Θ ≤a) ≥P1,π(Θ ≤a) (4.1) ver, without loss of generality, we may assume that y(0, π) = 0 so that µ0,y(0,π) = µ. Let where a < θ0. Moreover, without loss of generality, we may assume that y(0, π) = 0 so where a < θ0. Moreover, without loss of generality, we may assume that y(0, π) = 0 so that µ0,y(0,π) = µ. Let f(u) := euy(1,π)−B(u), and let Sa := S ∩(−∞, a]. Note that and let Sa := S ∩(−∞, a]. Note that P0,π (Θ ≤a) = Z Sa µ(du) P0,π (Θ ≤a) = Z Sa µ(du) and P1 π (Θ ≤a) = R Sa f(u)µ(du) R ( ) ( ) . and and 1,π (Θ ≤a) = R Sa f(u)µ(du) R S f(u)µ(du) . P1,π (Θ ≤a) = R Sa f(u)µ(du) R S f(u)µ(du) . Also note that ∂f(u) f(u)(y(1 π) B′(u)) Also note that ∂f(u) ∂u = f(u)(y(1, π) −B′(u)). ∂f(u) ∂u = f(u)(y(1, π) −B′(u)). Therefore, since B is convex, we have that f changes its monotonicity (from increasing to decreasing) at most once. Now we consider two separate cases: (i) f(a) ≤f(θ) and (ii) f(a) > f(θ). (i) f(a) ≤f(θ) (i) f(a) ≤f(θ) and (ii) f(a) > f(θ). ( ) ( ) ( ) If (i) holds, then (f(u) −f(a))(u −a) ≥0 for u ≤θ0 (since f changes its monotonicity at most once). Consequently, if µ(S−\ Sa) ̸= 0, then R Sa f(u)µ(du) R S−\Sa f(u)µ(du) ≤ f(a) R Sa µ(du) f(a) R S−\Sa µ(du) = R Sa µ(du) R S−\Sa µ(du). (4.2) (4.2) E. EKSTR ¨OM AND Y. WANG 276 (if µ(S−\ Sa) = 0, then (4.1) holds trivially with equality). Since (if µ(S−\ Sa) = 0, then (4.1) holds trivially with equality). Since (if µ(S−\ Sa) = 0, then (4.1) holds trivially with equality). are decreasing. Since R S−f(u)µ(du) R S f(u)µ(du) = 1 −π = Z S−µ(du), R S f(u)µ(du) = 1 π = Z S−µ(du), (4.2) implies that (4.2) implies that (4.2) implies that P1,π (Θ ≤a) = R Sa f(u)µ(du) R S f(u)µ(du) = R Sa f(u)µ(du) R Sa µ(du) R S−f(u)µ(du) = R Sa f(u)µ(du) R S−\Sa f(u)µ(du) R S−µ(du) 1 + R Sa f(u)µ(du) R S−\Sa f(u)µ(du) ≤ Z Sa µ(du) = P0,π(Θ ≤a), ( ) On the other hand, if (ii) holds, then the fact that f changes its monotonicity at most once gives that (f(u) −f(θ0))(u −θ0) ≤0 for all u ≥a. Consequently, R S−\Sa f(u)µ(du) R S+ f(u)µ(du) ≥ f(θ0) R S−\Sa µ(du) f(θ0) R S+ µ(du) = R S−\Sa µ(du) R S+ µ(du) . (4.3) (4.3) Since R S+ f(u)µ(du) R S f(u)µ(du) = π = Z S+ µ(du), R S+ f(u)µ(du) R S f(u)µ(du) = π = Z S+ µ(du), the inequality (4.3) yields P1,π (Θ > a) = R S\Sa f(u)µ(du) R S f(u)µ(du) = R S\Sa f(u)µ(du) R S+ µ(du) R S+ f(u)µ(du) = 1 + R S−\Sa f(u)µ(du) R S+ f(u)µ(du) ! Z S+ µ(du) ≥ Z S\Sa µ(du) = P0,π(Θ > a), from which (4.1) follows. Finally, the second inequality (for b > θ0) follows by a similar argument. As a consequence, we can show that the level curves are spreading out along the time axis. Corollary 4.2. Let 0 < π1 < π2 < 1. Then n 7→y(n, π2) −y(n, π1) is non-decreasing. Proof. Recall from (2.3) that a consequence, we can show that the level curves are spreading out along the time axis. As a consequence, we can show that the level curves are spreading out along the time axis. ∂q ∂y (n, y(n, π)) = En,π[Θ1{Θ>θ0}] −Pn,π(Θ > θ0)En,π[Θ] = (1 −π)En,π[Θ1{Θ>θ0}] −πEn,π[Θ1{Θ≤θ0}] By Theorem 4.1, this covariance (with respect to the posterior distrbution) is non-increasing in n. By this, the level curves are spreading out. BAYESIAN SEQUENTIAL COMPOSITE HYPOTHESIS TESTING IN DISCRETE TIME 277 5. Conditions for monotonicity in time In this section we investigate whether n 7→V (n, π) is non-decreasing. If this monotonicity holds, then the stopping boundaries b1 and b2 will be non-decreasing and non-increasing, respectively. To prove the monotonicity of V , we will use the following assumption. Assumption 5.1. For any π ∈(0, 1) and n ≥m ≥0, the random variable Πm+1\|{Πm = π} dominates Πn+1\|{Πn = π} in convex order. ssumption 5.1. For any π ∈(0, 1) and n ≥m ≥0, the random variable Πm+1 n+1\|{Πn = π} in convex order. Theorem 5.2. Assume that Assumption 5.1 holds. Then V (n, π) is non-decreasing in n, and the boundaries b1 and b2 are thus non-decreasing and non-increasing, respectively. Proof. For any concave function f, we have Em,π[f(Πm+1)] ≤En,π[f(Πn+1)]. It thus follows from Lemma 3.1 that V (n, π) is non-decreasing in n; the monotonicity of the boundaries b1 and b2 is a direct consequence. Since Πm+1\|{Πm = π} and Πn+1\|{Πn = π} have the same expected value π, a suﬃcient condition for stochas- tic domination in convex order is that there exists a point π0 around which the distribution of Πn+1\|{Πn = π} is more concentrated compared to the distribution of Πm+1\|{Πm = π} in the sense that Pm,π(Πm+1 ≤α) ≥Pn,π(Πn+1 ≤α) Pm,π(Πm+1 ≤α) ≥Pn,π(Πn+1 ≤α) Pm,π(Πm+1 ≤α) ≥Pn,π(Πn+1 ≤α) and Pm,π(Πm+1 > β) ≥Pn,π(Πn+1β) Pm,π(Πm+1 > β) ≥Pn,π(Πn+1β) Pm,π(Πm+1 > β) ≥Pn,π(Πn+1β) for α < π0 < β. Using Theorem 4.1, we now state a suﬃcient condition under which the above concentration property holds. for α < π0 < β. Using Theorem 4.1, we now state a suﬃcient condition under which the above concentration property holds. Theorem 5.3. Assume that the observations are continuously distributed with density of the form m 5.3. Assume that the observations are continuously distributed with density of the form h(x)pu(x), h(x)pu(x), with ν(dx) = h(x)dx for some nonnegative continuous function h such that I := {h > 0} is an interval. Assume that either with ν(dx) = h(x)dx for some nonnegative continuous function h such that I := {h > 0} is an interval. Assume that either (i) h(x)pu(x) is increasing in x on I, and S+ is a singleton (i) h(x)pu(x) is increasing in x on I, and S+ is a singleton (ii) h(x)pu(x) is decreasing in x on I, and S−is a singleton (ii) h(x)pu(x) is decreasing in x on I, and S−is a singleton holds. Then Assumption 5.1 holds, so n 7→V (n, π) is non-decreasing. holds. Then Assumption 5.1 holds, so n 7→V (n, π) is non-decreasing. Then Assumption 5.1 holds, so n 7→V (n, π) is non-decreasing. holds. Then Assumption 5.1 holds, so n 7→V (n, π) is non-decreasing. Proof. We will verify Assumption 5.1 for m = 0 and π = µ(S ∩[0, ∞)); the general case follows by translation. Thus we consider two distributions µ and µ′ := µn,y(n,π). Let F(a) = P0,π(Π1 ≤a) and G(a) = Pn,π(Πn+1 ≤a). Since F and G are continuous distribution functions with the same expected value π, there exists π0 ∈(0, 1) with F(π0) = G(π0). We claim that ′(π0) < G′(π0) (5.1) F ′(π0) < G′(π0) (5.1) 278 E. EKSTR ¨OM AND Y. WANG at such a point (unless µ is a two-point distribution), which implies that there is a single intersection point π0 and that G is more concentrated about π0 in comparison with F. To prove the claim, let π0 be such that F(π0) = G(π0), and denote by x1 and xn+1 the unique values such that q(1, x1) = π0 = q(n + 1, y(n, π) + xn+1), so that (1, x1) and (n + 1, y(n, π) + xn+1) are both on the π0-level curve. We then have F(π0) = Z x1 −∞ Z S h(x)pu(x)µ(du)dx and G(π0) = Z xn+1 −∞ Z S h(x)pu(x)µ′(du)dx, so F ′(π0) = R S h(x1)pu(x1)µ(du) ∂q ∂y(1, x1) and G′(π0) = R S h(xn+1)pu(xn+1)µ′(du) ∂q ∂y(n + 1, y(n, π) + xn+1) . so and G′(π0) = R S h(xn+1)pu(xn+1)µ′(du) ∂q ∂y(n + 1, y(n, π) + xn+1) . h(x)pu(x), Note that Note that ∂q ∂y (1, x1) = Cov1,π0(Θ, 1{Θ≤θ0}) ≥Covn+1,π0(Θ, 1{Θ≤θ0}) = ∂q ∂y (n + 1, y(n, π) + xn+1) ∂q ∂y (1, x1) = Cov1,π0(Θ, 1{Θ≤θ0}) ≥Covn+1,π0(Θ, 1{Θ≤θ0}) = ∂q ∂y (n + 1, y(n, π) + xn+1) since the covariance decreases along the π0-level curve. Furthermore, the covariance is strictly decreas- ing along the level curve (unless µ has support on only two points). Therefore, it suﬃces to show that R S h(x1)pu(x1)µ(du) ≤ R S h(xn+1)pu(xn+1)µ′(du). + since the covariance decreases along the π0-level curve. Furthermore, the covariance is strictly decreas- ing along the level curve (unless µ has support on only two points). Therefore, it suﬃces to show that R S h(x1)pu(x1)µ(du) ≤ R S h(xn+1)pu(xn+1)µ′(du). S S To do that, assume that (i) holds so that supp µ ∩S+ = {θ1} for some θ1 > θ0. Then µ and both µ′ are identical on S+, so it follows from Theorem 4.1 that µ′ stochastically dominates µ. Consequently, Z x −∞ Z S h(y)pu(y)µ(du)dy ≥ Z x −∞ Z S h(y)pu(y)µ′(du)dy Z x −∞ Z S h(y)pu(y)µ(du)dy ≥ Z x −∞ Z S h(y)pu(y)µ′(du)dy for any x, so F(π0) = G(π0) implies x1 ≤xn+1. Moreover, the relation p(1, x1) = p(n+1, y(n, π)+xn+1) implies that for any x, so F(π0) = G(π0) implies x1 ≤xn+1. Moreover, the relation p(1, x1) = p(n+1, y(n, π)+xn+1) implies that R S−h(x1)pu(x1)µ(du) h(x1)pθ1(x1) = R S−h(xn+1)pu(xn+1)µ′(du) h(xn+1)pθ1(xn+1) . (5.2) (5.2) BAYESIAN SEQUENTIAL COMPOSITE HYPOTHESIS TESTING IN DISCRETE TIME 279 Since the density function is increasing in x, we have h(x1)pθ1(x1) ≤h(xn+1)pθ1(xn+1). Equation (5.2) thus implies Z S h(x1)pu(x1)µ(du) ≤ Z S h(xn+1)pu(xn+1)µ′(du). (5.3) (5.3) If instead (ii) holds, then x1 ≥xn+1, and (5.3) is derived in a similar way. It follows that F and G have a unique intersection point π0 (unless the support of µ has only two points, in which F ≡G), and that G is more concentrated about π0 compared to F. Consequently, the random variable Π1\|{Π0 = π} dominates Πn+1\|{Πn = π} in convex order. Remark 5.4. At ﬁrst it may seem counterintuitive that the cost V (π, n) is increasing in n, despite the fact that more observations should enhance statistical precision. h(x)pu(x), The explanation is that when moving along a level curve, the posterior distribution concentrates about the level curve, compare Theorem 4.1, and a concentrated distribution is disadvantageous from the hypothesis testing perspective. Example (Exponential observations.) Assume that {Xk, k ≥1} are exponentially distributed with unknown intensity Θ and independent (conditional on Θ) so that P(X1 ≤x\|Θ = u) = 1 −e−ux. This is not of the exponential form (2.2), but it is straightforward to check that if one instead considers X′ k := −Xk, then X′ k has density h(x)pu(x) = exp{ux + log u} x < 0 0 x ≥0 with respect to Lebesgue measure. For u > 0, this density is increasing in x on I = (−∞, 0). Consequently, if S+ is a singleton, then (i) in Theorem 5.3 gives that V (n, π) is increasing in n, which leads to the monotonicity of the stopping boundaries. with respect to Lebesgue measure. For u > 0, this density is increasing in x on I = (−∞, 0). Consequently, if S+ is a singleton, then (i) in Theorem 5.3 gives that V (n, π) is increasing in n, which leads to the monotonicity of the stopping boundaries. Example (Gaussian observations with unknown variance.) If {Xk, k ≥1} are normally distributed with mean 0 and unknown standard deviation Θ and independent (conditional on Θ), then the random variables X′ k = −1 2(Xk)2 are of the exponential form (2.2) with respect to the unknown variable Θ′ := Θ−2, with density h(x)pu(x) = 2 √−πx exp ux + 1 2 log u x < 0 0 x ≥0 with respect to Lebesgue measure. Note that this density is increasing in x. Also note that H0 holds precisely when Θ′ ≥θ−2 0 . Therefore, the value function is decreasing in time for any prior distribution µ such that S−is a singleton. with respect to Lebesgue measure. Note that this density is increasing in x. Also note that H0 holds precisely when Θ′ ≥θ−2 0 . Therefore, the value function is decreasing in time for any prior distribution µ such that S−is a singleton. Remark 5.5. Theorem 5.3 considers random variables that are continuously distributed on an interval. However, a closer inspection of the proof reveals that one can relax this assumption and instead assume that H := supp(h) is the union of disjoint intervals. 6.1. Gaussian observations with unknown mean Assume that the Gaussian sequence {Xk, k ≥1} has unknown mean Θ and known standard deviation 1 so that the density (conditional on Θ = u) is 1 √ 2π e−x2 2 exp{ux −B(u)}, where B(u) = u2/2. In this case, the discrete time Π-process can be embedded in the corresponding continuous time process, as studied in [10]. Moreover, the discrete time problem then corresponds to the continuous time problem in [10] but with the restriction that stopping is only allowed at integer times. For such a problem, the techniques used in [10] (in particular, preservation of concavity for martingale diﬀusions coupled with time-decay of the diﬀusion coeﬃcient of Π) show that V (n, π) is non-decreasing in n. where B(u) = u2/2. In this case, the discrete time Π-process can be embedded in the corresponding continuous time process, as studied in [10]. Moreover, the discrete time problem then corresponds to the continuous time problem in [10] but with the restriction that stopping is only allowed at integer times. For such a problem, the techniques used in [10] (in particular, preservation of concavity for martingale diﬀusions coupled with time-decay of the diﬀusion coeﬃcient of Π) show that V (n, π) is non-decreasing in n. Conjecture 6.1. The function V (n, π) in (3.1) is non-decreasing in n for any prior distribution µ and any ν. Conjecture 6.1. The function V (n, π) in (3.1) is non-decreasing in n for any prior dist h(x)pu(x), Moreover, when these intervals become small, using 280 E. EKSTR ¨OM AND Y. WANG approximation arguments one would expect a similar result for discrete distributions; we leave the details, as well as the precise formulation, of such a result. approximation arguments one would expect a similar result for discrete distributions; we leave the details, as well as the precise formulation, of such a result. ents one would expect a similar result for discrete distributions; we leave the details formulation, of such a result. 6. Further discussion on time monotonicity While the conditions in Theorem 5.3 may appear somewhat restrictive, we have failed to remove the con- ditions. On the other hand, we have also been unable to produce examples within the exponential family for which the asserted time monotonicity fail. In this ﬁnal section, we show that time monotonicity holds for arbi- trary prior distributions µ in a few particular examples, see Sections 6.1-3 below. Based on these ﬁndings, we formulate the following conjecture. Conjecture 6.1. The function V (n, π) in (3.1) is non-decreasing in n for any prior distribution µ and any ν. 6.3. Binomial observations Consider a general prior distribution µ on [0, 1] for Θ, and observations {Xk, k ≥1} that are Bin(N, Θ). As in Section 6.2, this is on the exponential form if one uses Θ′ = log Θ 1−Θ as the unknown parameter. 1 Θ Now consider a sequential testing problem for Bernoulli observations with the same unknown parameter Θ, but where the cost c is only imposed on the (Nk + 1)-th observation, for all k ∈N0. Denoting the value function of that sequential problem by V Ber(n, π), arguments similar to those in Section 6.2 imply that the function k 7→V Ber(kN, π) is non-decreasing. However, the value function V (n, π) for Binomial observations clearly coincide with V Ber(nN, π), so time monotonicity holds also for V (n, π). 6.2. Bernoulli observations Consider a sequence {Xk, k ≥1} which is Bernoulli distributed with parameter Θ so that P(Xk = 1\|Θ = u) = 1 −P(Xk = 0\|Θ = u) = u. (This is on the exponential form if one instead uses Θ′ := log Θ 1−Θ as the unknown parameter, because then (This is on the exponential form if one instead uses Θ′ := log Θ 1−Θ as the unknown par P(Xk = x\|Θ′ = u) = P(Xk = x\|Θ = eu 1 + eu ) = eux−log(1+eu) for x ∈{0, 1}.) Also assume that the distribution of Θ is µ, which is a given measure on [0, 1]. Note that since observations are binary, Πn+1 can only take two diﬀerent values if started at a given poin (n, π). For m ≤n we have o assume that the distribution of Θ is µ, which is a given measure on [0, 1]. Note that since observations are binary, Πn+1 can only take two diﬀerent values if started at a given point (n, π). For m ≤n we have Pm,π(Θ > θ0\|Xm+1 = 1) = R S+ uµm,y(m,π)(du) R S uµm,y(m,π)(du) = 1 1 + Em,π[θ1{θ≤θ0}] Em,π[θ1{θ>θ0}] ≥ 1 1 + En,π[θ1{θ≤θ0}] En,π[θ1{θ>θ0}] = Pn,π(Θ > θ0\|Xn+1 = 1), BAYESIAN SEQUENTIAL COMPOSITE HYPOTHESIS TESTING IN DISCRETE TIME 281 where the inequaility is a consequence of Theorem 4.1. Similarly, where the inequaility is a consequence of Theorem 4.1. Similarly, Pm,π(Θ > θ0\|Xm+1 = 0) = 1 1 + Em,π[(1−θ)1{θ≤θ0}] Em,π[(1−θ)1{θ>θ0}] ≤ 1 1 + En,π[(1−θ)1{θ≤θ0}] En,π[(1−θ)1{θ>θ0}] = Pn(Θ > θ0\|Xn+1 = 0). Since two two-point distributions with the same mean and with mass on {a, b} and {a′, b′}, where a ≤a′ < b′ ≤b, are ordered in convex order, it follows that the distribution of Πm+1 under Pm,π dominates the distribution of Πn+1 under Pn,π in convex order. Thus Assumption 5.1 is satisﬁed, so time monotonicity for arbitrary priors µ holds by Theorem 5.2. Since two two-point distributions with the same mean and with mass on {a, b} and {a′, b′}, where a ≤a′ < b′ ≤b, are ordered in convex order, it follows that the distribution of Πm+1 under Pm,π dominates the distribution of Πn+1 under Pn,π in convex order. Thus Assumption 5.1 is satisﬁed, so time monotonicity for arbitrary priors µ holds by Theorem 5.2. References M. Alvo, Bayesian sequential estimation. Ann. Statist. 5 (1977) 955–968. [1] M. Alvo, Bayesian sequential estimation. Ann. Statist. 5 (1977) 955–968. [1] M. Alvo, Bayesian sequential estimation. Ann. Statist. 5 (1977) 955–968. [2] J.A. Bather, Bayes procedures for deciding the sign of a normal mean. Proc. Cambridge Philos. So [ ] [3] P.J. Bickel, On the asymptotic shape of Bayesian sequential tests of θ ≤0 versus θ > 0 for exponential families. Ann. Statist. 1 (1973) 231–240. [3] P.J. Bickel, On the asymptotic shape of Bayesian sequential tests of θ ≤0 versus θ > 0 for exponential families. Ann. Statist. 1 (1973) 231–240. [4] L. Brown, Suﬃcient statistics in the case of independent random variables. Ann. Math. Statist. 35 [ ] , p ( ) [5] L. Brown, Fundamentals of statistical exponential families with applications in statistical decision theory. Institute of Mathematical Statistics, Hayward, CA (1986). [5] L. Brown, Fundamentals of statistical exponential families with applications in statistical decision theory. Institute of Mathematical Statistics, Hayward, CA (1986). ] P. Cabilio, Sequential estimation in Bernoulli tria [7] H. Chernoﬀ, Sequential tests for the mean of a normal distribution. IV. (Discrete case). Ann. Math. Statist. 36 (1965) 55–68. [8] S. Dayanik and S. Sezer, Multisource Bayesian sequential binary hypothesis testing problem. Ann. Oper. Res. 201 (2012) 99–130. [7] H. Chernoﬀ, Sequential tests for the mean of a normal distribution. IV. (Discrete case). Ann. Math. Statist. 36 (1965) 55–68. [8] S Dayanik and S Sezer Multisource Bayesian sequential binary hypothesis testing problem Ann Oper Res 201 (2012) [7] H. Chernoﬀ, Sequential tests for the mean of a normal distribution. IV. (Discrete case). Ann. Math. Statist. 36 (1965) 55–68. [ ] , q ( ) [8] S. Dayanik and S. Sezer, Multisource Bayesian sequential binary hypothesis testing problem. Ann 99–130. [9] E. Ekstr¨om, I. Karatzas and J. Vaicenavicius, Bayesian sequential least-squares estimation for the drift of a Wiener process. Stoch. Process. Appl. 145 (2022) 335–352. [10] E. Ekstr¨om and J. Vaicenavicius, Bayesian sequential testing of the drift of a Brownian motion. ESAIM: PS. 19 (2015) 626–648. E. Ekstr¨om and Y. Wang, Multi-dimensional sequent [12] P.V. Gapeev and G. Peskir, The Wiener sequential testing problem with ﬁnite horizon. Stoch. Stoch. Rep. 76 (2004) 59–75. [13] C. Hipp, Suﬃcient statistics and exponential families. Ann. Statist. 2 (1974) 1283–1292. [12] P.V. Gapeev and G. Peskir, The Wiener sequential testing problem with ﬁnite horizon. Stoch. Stoch. Rep. [1] M. Alvo, Bayesian sequential estimation. Ann. Statist. 5 (1977) 955–968. References 76 (2004) 59–75. [13] C Hi S ﬃi t t ti ti d ti l f ili A St ti t 2 (1974) 1283 1292 V. Gapeev and G. Peskir, The Wiener sequential testing problem with ﬁnite horizon. Stoch. Stoch. Rep Hipp, Suﬃcient statistics and exponential families. Ann. Statist. 2 (1974) 1283–1292. [13] C. Hipp, Suﬃcient statistics and exponential families. Ann. Statist. 2 (1974) 1283–1292. [ ] pp, p ( ) [14] D. Hobson, Volatility misspeciﬁcation, option pricing and superreplication via coupling. Ann. Appl. Probab. 8 (1998) 193–205. [15] S. Janson and J. Tysk, Volatility time and properties of option prices. Ann. Appl. Probab. 13 (2003) 890–913. n, Volatility misspeciﬁcation, option pricing and superreplication via coupling. Ann. Appl. Probab. 8 (199 ( ) [14] D. Hobson, Volatility misspeciﬁcation, option pricing and superreplication via coupling. Ann. Ap [14] D. Hobson, Volatility misspeciﬁcation, option pricing and superreplication via coupling. Ann. Appl. Probab. 8 (1998) 193–205. [15] S. Janson and J. Tysk, Volatility time and properties of option prices. Ann. Appl. Probab. 13 (2003) 890–913. [15] S. Janson and J. Tysk, Volatility time and properties of option prices. Ann. Appl. Probab. 13 (2003) 890–913. [15] S. Janson and J. Tysk, Volatility time and properties of option prices. Ann. Appl. Probab. 13 ( composite hypotheses. Ann. Statist. 16 (1988) 856–8 [17] D. Lindley and B. Barnett, Sequential sampling: Two decision problems with linear losses for binomial and normal random variables. Biometrika 52 (1965) 507–532. [17] D. Lindley and B. Barnett, Sequential sampling: Two decision problems with linear losses for b variables. Biometrika 52 (1965) 507–532. [18] S. Moriguti and H. Robbins, A Bayes test of “p ≤1/2” versus “p > 1/2”. Rep. Statist. Appl. Res. Un. Jpn. Sci. Eng. 9 (1962) 39–60. [19] G. Peskir and A. Shiryaev, Sequential testing problems for Poisson processes. Ann. Statist. 28 (200 [20] G. Schwarz, Asymptotic shapes of Bayes sequential testing regions. Ann. Math. Statist. 33 (1962) [21] A. Shiryayev, Optimal stopping rules. Applications of Mathematics, Vol. 8. Springer-Verlag, New York-Heidelberg (1978). [22] A. Shiryaev, Two problems of sequential analysis. Cybernetics 3 (1967) 63-69 (1969). [21] A. Shiryayev, Optimal stopping rules. Applications of Mathematics, Vol. 8. Springer-Verlag, New York-Heidelberg (1978). [21] A. Shiryayev, Optimal stopping rules. Applications of Mathematics, Vol. 8. Springer-Ver 22] A. Shiryaev, Two problems of sequential analysi [ ] y , p q y y ( ) ( ) [23] M. [ ] y , p q y y ( ) ( ) [23] M. Sobel, An essentially complete class of decision functions for certain standard sequential problems. Ann. Math. Stat. 24 (1953) 319–337. arz, Asymptotic shapes of Bayes sequential testing regions. Ann. Math. Statist. 33 (1962) 224–236. r and A. Shiryaev, Sequential testing problems for Poisson processes. Ann. Statist. 28 (2000) 837–859. References Sobel, An essentially complete class of decision functions for certain standard sequential problems. Ann. Math. Stat. 24 (1953) 319–337 [23] M. Sobel, An essentially complete class of decision functions for certain standard sequential problems. Ann. Math. Stat. 24 (1953) 319–337. 282 E. EKSTR ¨OM AND Y. WANG A. Wald and J. Wolfowitz, Bayes solutions of sequential decision problems. Ann. Math. Stat. 21 (1950 [25] M. Zhitlukhin and A.N. Shiryaev, A Bayesian sequential testing problem of three hypotheses for Brownian motion. Stat. Risk Model. 28 (2011) 227–249. This journal is currently published in open access under a Subscribe-to-Open model (S2O). S2O is a transformative model that aims to move subscription journals to open access. Open access is the free, immediate, online availability of research articles combined with the rights to use these articles fully in the digital environment. We are thankful to our subscribers and sponsors for making it possible to publish this journal in open access, free of charge for authors. Please help to maintain this journal in open access! Please help to maintain this journal in open access! Check that your library subscribes to the journal, or make a personal donation to the S2O programme, by contacting subscribers@edpsciences.org Check that your library subscribes to the journal, or make a personal donation to the S2O programme, by contacting subscribers@edpsciences.org More information, including a list of sponsors and a ﬁnancial transparency report, available at: https://www.edpsciences.org/en/maths-s2o-programme More information, including a list of sponsors and a ﬁnancial transparency report, available at: https://www.edpsciences.org/en/maths-s2o-programme
W2471546819.txt	https://www.shs-conferences.org/articles/shsconf/pdf/2016/05/shsconf_cmlf2016_05005.pdf	fr		Le processus de reconnaissance verbale à l’épreuve des amorces	SHS web of conferences	2,016	cc-by	6,539	SHS Web of Conferences 27, 05005 (2016) DOI: 10.1051/ shsconf/20162705005 Congrès Mondial de Linguistique Française - CMLF 2016 Le processus de reconnaissance verbale à l’épreuve des amorces Gabriel Bergounioux Laboratoire Ligérien de Linguistique (LLL, UMR 72 70) – Université d’Orléans gabriel.bergounioux@univ-orleans.fr Résumé. Cette étude se propose, à partir de l'exploitation d'un millier d'amorces dans le corpus des entretiens de l'Enquête Sociolinguistique à Orléans (ESLO2), de produire une typologie des amorces et d'en faire l'analyse. Après une présentation concernant la constitution des données, à partir des occurrences, sont examinées les différentes modalités de réalisation. Les amorces sont utilisées comme un observatoire de la conduite des auditeurs dans le cas d'informations lacunaires dans le signal. La réflexion est prolongée par un questionnement concernant le statut de la Forme Sous-Jacente (FSJ) lorsque l'unité, telle qu'elle est déclarée en morpho-phonologie, s'avère absente dans le discours, posant la question de son identification. Abstract. How can we know what a false start means? Based on a French oral corpus – ESLO –, this paper discusses how we proceed to assign an interpretation when we hear a false start. It presents its special properties according to the spoken data on which this study is grounded. The discussion concludes, from the interpretation of the false starts, to the hearer's behaviour when he is confronted with that kind of riddles in the discourse. Introduction L’une des opérations les plus délicates à appréhender dans la réception des messages concerne les modalités de compréhension par l’auditeur. Comment des énoncés prononcés dans des environnements bruités, avec de fortes variations sociales, inter- et intra-individuelles, parfois réalisés de façon incomplète ou fautive, peuvent-ils être néanmoins identifiés par le destinataire ? Cette question de l’interprétation – celle du locuteur, si le terme prend sa valeur musicale, et celle de l’auditeur en se rapprochant de l’acception psychanalytique – est le plus souvent effectuée à partir d’une approche d’inspiration sémantique. On se propose dans cette contribution d’en considérer la présence à partir de la traduction phonologique, quand le signal – les éléments acoustiques – en vient à être identifié mentalement à une (ou plusieurs) unité(s) lexicale(s), à un (ou plusieurs) signe(s). Dans un premier temps, on présentera les données utilisées pour l’analyse, un ensemble d’amorces recueillies dans un corpus dont on exposera les caractéristiques. On indiquera ensuite de quelle façon on peut en réaliser le classement avant d’interroger la façon dont se réalise la reconnaissance du mot interrompu. Cette dernière étape a la valeur d’un test puisqu’elle met à l’épreuve la reconnaissance verbale appréhendée dans les conditions extrêmes d’une information incomplète. Autrement dit, cette étude propose d’expliciter ce qui se trouve engagé dans une telle opération en conjecturant qu’elle s’apparente à la conduite ordinaire de l’auditeur dont elle révèlerait, par contraste, ce que seraient les mécanismes de son fonctionnement. Sans rien exclure des propositions de la psychologie cognitive ou des théories de l’acquisition du langage, à l’articulation de la phonologie et de la morphologie, cette étude privilégie une approche phonologique © The Authors, published by EDP Sciences. This is an open access article distributed under the terms of the Creative Commons Attribution License 4.0 (http://creativecommons.org/licenses/by/4.0/). SHS Web of Conferences 27, 05005 (2016) DOI: 10.1051/ shsconf/20162705005 Congrès Mondial de Linguistique Française - CMLF 2016 guidée par l’exploitation d’un corpus de français parlé. La compatibilité des résultats avec les travaux conduits dans d’autres champs d’investigation exige un décentrement de la perspective qui permettra de formuler des validations (ou des réfutations) indépendantes des propositions avancées dans cet article. Cette complémentarité est l’un des apports du repérage effectué dans les données. 1 Qu’est-ce qu’une amorce ? A la différence du bégaiement pour lequel une interprétation physiologique est souvent latente et qui est de l’ordre du handicap, l’amorce, dont la phénoménologie semble proche, n’appelle pas d’explication neurologique. Même si un parallèle serait possible entre les variations de sa fréquence en discours et l’aggravation du bégaiement dans des circonstances homologues, elle n’est pas imputable à une constante du comportement expressif. On la retrouve chez tout un chacun en oral spontané, en situation de dialogue (trouverait-on des amorces dans le discours endophasique ?). 1.1 Description du phénomène Le français a retenu, au moment d’assigner un nom au phénomène, son aspect positif : l’unité lexicale est présente par son (ou ses) premier(s) phonème(s). La terminologie est empruntée à la psychologie où l’« amorce » désigne le début d’un mot employé comme stimulus pour aider le sujet de l’expérience à en deviner la suite. A l’inverse, l’anglais utilise « false start » en mettant l’accent sur l’échec de la réalisation complète, un terme emprunté aux compétitions sportives, la course en particulier. Comme si le français visait la possibilité d’une compréhension par l’auditeur – on va montrer qu’elle est souvent effective – et l’anglais l’incapacité du locuteur à réaliser la production complète. Les amorces relèvent de la catégorie des universaux de langage. Elles se retrouvent dans toutes les langues, ou plutôt dans leur actualisation en discours en sorte que le phénomène a échappé à des descriptions centrées sur le système et non sur la parole, quand bien même elles étaient amenées à se fonder sur l’examen de données orales. Les linguistes ne proposent pas d’hypothèse particulière concernant leur émergence, les catégories concernées ou la forme que devrait prendre un mot quand il se réduit à ce segment. Les études sur le sujet viennent plutôt de la psychologie ou de la théorie de la communication, éventuellement du TAL (les disfluences) voire de la linguistique clinique dans la perspective ouverte au XIXe siècle par John Hughlings Jackson. Invisibles pour une linguistique des langues, les amorces se présentent d’ordinaire comme de simples éléments de l’échange verbal qu’il n’est pas nécessaire de restituer dans la communication pour accéder à la compréhension. Ce sont des interférences, des parasites. Elles interpolent des séquences qui ralentissent le flux discursif et le fourvoient en sorte que, lors du processus d’audition, elles seront spontanément éliminées. Pas de meilleur témoignage du tri sélectif opéré en amont de l’interprétation du message que l’expérience du transcripteur qui s’astreint à les consigner et qui éprouve les pires difficultés à en conserver la trace en mémoire immédiate. Plusieurs explications ont été proposées pour en raisonner l’apparition, se référant : - tantôt à quelque particularité linguistique (séquence phonotactique complexe, mot rare, construction syntaxique dérivée) si l’accent est mis sur le fonctionnement du système, - tantôt à un dysfonctionnement des capacités cognitives (en lien avec une conception de la grammaire comme système de règles computationnelles manipulant des symboles qui correspondent à des classes d’équivalence), qu’on mette en avant - la difficulté de la tâche en linguistique cognitive ou - des défaillances neurologiques en linguistique clinique, - tantôt à une perturbation liée aux conditions sociales de l’échange, au contexte de l’expression, résultant - soit d’interférences intrasubjectives – comme dans le lapsus –, 2 SHS Web of Conferences 27, 05005 (2016) DOI: 10.1051/ shsconf/20162705005 Congrès Mondial de Linguistique Française - CMLF 2016 - soit d’une insécurité dans l’adresse au moment d’entrer en communication. L’analyse emprunte ici une voie différente. Elle s’intéresse à l’audition – et non à la production – afin de déterminer de quelle façon une amorce en vient à être identifiée à un mot à partir des données extraites d’un corpus. 1.2 Notation : du signal à la transcription Une fois le cas du bégaiement éliminé, l’amorce, telle qu’elle est conçue dans cette étude, doit répondre à trois propriétés : phonologique, cognitive et graphique. Phonologiquement, elle apparaît sous la forme d’une trace phonétique, un fragment de signal qui peut être décomposé en un ou plusieurs phonèmes dont l’épellation ne coïncide pas avec un mot sans proscrire la possibilité de compléter ce qui est rétroactivement considéré comme son ou ses premiers éléments. Trois questions apparaissent alors : (i) la nécessité de distinguer l’amorce de l’abréviation (prof n’est pas une amorce), [1] vous me racontez euh depuis quand vous habitez dans l’agglo- de (2460) « l’AgglO » est l’abréviation officielle pour la Communauté d’Agglomération d’Orléans Val de Loire. (ii) la pertinence d’une insertion du mot conjecturé à telle place dans l’énoncé, [2] venir les les les visit leur rendre (2411) où « visiter » est confirmé par la reprise « rendre (visite) ». (iii) la résolution d’une homonymie qu’accroît de façon inversement proportionnelle la réduction du nombre de phonèmes. Il y a beaucoup plus de mots commençant par r- que de mots commençant par re-, de mots commençant par re- que de mots commençant par rev- etc. [3] avec euh y- l’aménagem- (3797) ne laisse guère d’autre alternative que « aménagement » pour l’assignation lexicale. Cognitivement, les sons entendus doivent être identifiés avec un mot, ce qui aboutit par exemple à l’absence, dans le corpus, de formes qui seraient notées « eu- », « heu- » et qui ont été systématiquement réinterprétées comme des « euh » par le transcripteur. Il est probable qu’il y ait aussi une sous-estimation d’amorces réduites à « m- » qui sont entendues comme une forme d’approbation et qui sont notées « mh ». Ainsi, pour que l’amorce se développe en mot, deux conditions doivent être réunies : une séquence de phonèmes présente dans un ou plusieurs mots et une ratification conditionnée par sa probabilité contextuelle, « actuelle » en [4] : [4] Regardez la jeunesse actue- gamin (2321) le cas le plus simple étant la répétition du mot dans sa forme complète. [5] parce que y a y a tellement de gens qui viennent de di- régions différentes (2217) Graphiquement, les amorces recensées sont celles consignées par la transcription, c’est-à-dire celles pour lesquelles ont été surmontées non seulement les habitudes du « sujet écrivant » dans la prise de notes par exemple, ou de « nettoyage » des corpus (en TAL notamment), mais, plus difficiles à neutraliser, ces éliminations qui sont effectuées sans intention, à un niveau subconscient. Cette réduction est le produit d’un apprentissage initial du tri qu’il faut effectuer dans le signal : on n’entend pas ce qu’on sait n’être pas nécessaire à la compréhension du message. C’est une cause d’oublis d’une importance comparable à celle qui résulte de la mauvaise qualité sonore ou de l’inattention du transcripteur. 3 SHS Web of Conferences 27, 05005 (2016) DOI: 10.1051/ shsconf/20162705005 Congrès Mondial de Linguistique Française - CMLF 2016 On remarquera que le transcripteur à son tour anticipe l’interprétation par la façon dont il restitue orthographiquement les propos. Dans l’exemple suivant, la notation par « mai-» (et non « mé ») prescrit la lecture « mairie » : [6] Avec les mai- mairies de gauche (364) 1.3 Les données : le corpus ESLO Les données sont empruntées aux transcriptions du corpus de l’Enquête Sociolinguistique à Orléans entreprise depuis 2008 et sténographiée ESLO2 pour la distinguer d’ESLO1 qui correspondait à une phase précédente de collecte réalisée entre 1968 et 1971. Les deux corpus sont accessibles sur http://eslo.huma-num.fr/ où se trouvent toutes les indications concernant le mode de collecte et les caractéristiques de la ressource. ESLO2 est subdivisé en différents modules. Le plus important quantitativement (environ 40% du temps total) comprend une centaine d’entretiens en face à face, d’une durée moyenne d’une heure. Les propos sont recueillis à partir d’un questionnaire ouvert centré sur la trajectoire biographique et la vie des témoins dans leur environnement quotidien. Cinq éléments à souligner concernant la façon dont les analyses ultérieures sont préparées par le mode de constitution du corpus : 1. ESLO2 n’est pas conçue comme une étape préparatoire au travail linguistique mais comme une phase de la recherche de plein exercice impliquant une analyse de la constitution des données et un retour sur leur élaboration. 2. On ne s’y propose pas de définir un usage du français standard mais, comme cela avait été le cas pour ESLO1, on privilégie l’enregistrement de la variation dans toutes ses dimensions, diachronique (avec ESLO1), diastratique (l’étude est sociolinguistique), dialectale (tous les témoins ne sont pas natifs d’Orléans et une attention particulière est portée au contact de langues) et diaphasique (d’où la diversité des modules). 3. ESLO2 s’organise autour d’un programme, « le portrait sonore d’une ville », qui se focalise sur ce qu’un Orléanais capte en fait de réalisations langagières autant et plus que sur ce qu’il est à même de produire face à un micro. 4. La transcription est alignée sur le signal : il est possible de les séparer pour une analyse en phonétique instrumentale ou pour l’étude des fréquences lexicales mais le corpus est conçu comme la réunion de l’enregistrement et de sa restitution orthographique (pour les conventions adoptées, se reporter au site). 4. On intègre la variation aussi au niveau des transcriptions puisque pour chaque interview (et plus généralement pour chaque enregistrement), il existe au minimum trois versions désignées comme A, B et C : - A est le premier état, comprenant le formatage numérique, la segmentation du discours, une transcription « brute » et un lien aux métadonnées à des fins de catalogage et d’indexation. - B est une reprise de A afin d’en compléter et d’en affiner la restitution, de vérifier la pertinence ces choix opérés, de corriger des erreurs éventuelles et d’assurer l’anonymisation, - C est une reprise de B par un linguiste qui doit réviser l’ensemble, en particulier l’orthographe et le respect des conventions, et contrôler la qualité de l’anonymisation avant la mise en ligne. Les trois versions sont conservées. Elles permettent d’avoir accès aux processus de compréhension à travers les alternatives d’interprétation survenant à l’audition, ce qui recoupe pour partie le type de travail présenté ici. Par ailleurs, ce processus est indéfiniment cumulatif. Tout chercheur travaillant sur ces données, par exemple pour les annoter (mais il peut s’agir aussi bien d’anthropologues, de statisticiens voire d’artistes…) a la possibilité de corriger, ou du moins de 4 SHS Web of Conferences 27, 05005 (2016) DOI: 10.1051/ shsconf/20162705005 Congrès Mondial de Linguistique Française - CMLF 2016 compléter ou de modifier à son tour la transcription, selon ce qu’il aura entendu ou ce qu’il veut en faire, créant ainsi de nouvelles versions : D, E ad. lib. L’enrichissement est également réalisé par l’outillage du corpus, un programme mené en partenariat avec l’Equipex Ortolang dirigé par J.-M. Pierrel et sa pérennisation relève d’un archivage effectué en lien avec la BnF, au département des documents sonores sous la supervision de P. Cordereix et A. Viault et avec COCOON (Collection de Corpus Oraux Numériques, dont les responsables sont M. Jacobson et S. Guillaume) en lien avec la TGIR Huma-Num, présidée par O. Baude. 2 Le traitement effectué : de la requête au tri des données Pour recenser les amorces, une requête a été lancée par Layal Kanaan-Caillol, IR au LLL sur l’ensemble du module « entretiens » d’ESLO2, en exploitant les conventions de transcription. La demande était formulée de la façon suivante : « extraire toutes les chaînes de caractères ayant pour trois derniers éléments {lettre + trait d’union + espace} ». Il en est résulté une liste de 9504 réponses dont la présentation en colonnes, telle qu’elle a été décidée, livre les informations sous la forme d’un tableau Excel ayant les valeurs suivantes (en italiques, une explication des indications) : Col. 1 ESLO2_ENT_1006_C Dans le corpus « ESLO2 », il s’agit du module « entretien » avec un enregistrement codé « 1006 » dont est utilisée la version de transcription C Col. 2 0S6 Identifiant anonymisé du témoin (le code est fourni par un programme aléatoire) Col. 3 00:50:35 Moment de la réalisation en heure/minutes/secondes dans l’enregistrement (début) Col. 5 00:50:43 Fin de la réalisation en heure/minutes/secondes dans l’enregistrement (fin) Col. 5 personnellement ça m’a jam- ça me touchait pas Séquence découpée par le transcripteur Col. 6 jamais Interprétation de l’amorce Col. 7 <commentaire éventuel> e.g. lapsus, abréviation… Les colonnes 1 à 5 correspondant aux résultats de la requête, les colonnes 6 et 7 ont été ajoutées pour cette étude. La référence des exemples portée à la fin entre parenthèses correspond à la numérotation des lignes proposée par Excel. On obtient, à partir de : [7] voyez là j’ai décrété lundi je travaille pas je dis je vais aller voir les Chèvres euh du pres- du Pentagone (1872) Soit : Col. 1 Col. 2 Col. 3 Col. 4 Col. 5 Col. 6 Col. 7 ESLO2_ENT_1016_C VB16 00:28:28 00:28:40 (exemple 1872) Président lapsus A partir des résultats bruts, la première tâche à réaliser concerne l’élimination des erreurs de transcription telles que : 5 SHS Web of Conferences 27, 05005 (2016) DOI: 10.1051/ shsconf/20162705005 Congrès Mondial de Linguistique Française - CMLF 2016 - l’insertion erronée d’une espace, e.g. « peut- être », ou une mauvaise interprétation d’un énoncé : [8] j- jazz oh f- tout ça oui ça ça oui ça je j’allais (1052) où le transcripteur n’a pas identifié le syntagme festival « jazz off », substituant une amorce en « f- ». - les abréviations (triso-, asso-…) - les amorces qu’une écoute attentive ne permet pas de retrouver dans l’enregistrement. Inversement est prise en compte la multiplicité des amorces dans une même séquence : [9] au niveau de le d’Oliv- d’Oliv- O- la S- mh la Source non je pense pas (1875) Dans cet exemple, on compte quatre amorces – trois fois « Olivet » (une commune au sud d’Orléans) et une fois « la Source » – à l’intérieur d’une seule séquence. On n’a pas complété le tableau en ajoutant de nouvelles amorces qui n’avaient pas été repérées par les transcripteurs et qui ont été relevées lors des vérifications sur les enregistrements. Il aurait fallu, pour donner cohérence à ce travail, reprendre l’ensemble des entretiens et créer une version D qui soit conforme aux conventions pour la mise en ligne. Il a paru préférable de conserver un état identifié de la ressource dans sa globalité. Les résultats comprennent donc 9504 lignes pour une centaine d’heures d’enregistrement, soit en première analyse une amorce et demie par minute environ, l’élimination d’une partie d’entre elles étant partiellement compensée par la prise en compte des répétitions. L’essentiel des soustractions est conditionné par l’information fournie, ou non, dans la sixième colonne. Si les lettres suivies du tiret et de l’espace ne peuvent être assignées à un mot, on considère qu’il ne s’agit pas d’une amorce mais seulement d’un élément parasite, qui n’accède pas au statut d’objet linguistique. Ont été également retirées de la liste : - les troncations morphologiques : [10] et mais ça l’a pas empêché de se représenter et d’être re- re- réélu (1469) - les abréviations : [11] donc euh voilà pour organiser des des expo- voilà (1447) - les lapsus : [12] explicite ici pas forcément ailleurs j’ai rémarch- remarqué (1504) On a retenu pour une analyse détaillée les lignes 1000 à 1999 sur lesquelles : - 45 unités ont été ôtées (abréviations…), - 106 amorces ont été ajoutées (multiplicité des amorces dans une seule séquence), - 243 amorces ont dû être éliminées faute de correspondre à un mot identifiable (pas d’interprétation plausible dans la colonne 6). Ainsi, en lieu et place des 1000 amorces potentielles fournies par la requête, il en demeure : 1000 + 106 – (45 + 243) = 818. On a choisi de conserver, en les considérant comme une variété particulière d’amorces, les 22 lapsus. Chacun des exemples sera cité en référence au numéro de la ligne et le fichier sera joint à la version électronique de l’article. 3 Comment identifier une amorce ? 6 SHS Web of Conferences 27, 05005 (2016) DOI: 10.1051/ shsconf/20162705005 Congrès Mondial de Linguistique Française - CMLF 2016 A présent, il devient possible de répondre à la question posée dès l’introduction en référence aux données recueillies : quels sont les moyens à la disposition de l’auditeur qui lui permettent d’assigner une interprétation lexicale à une séquence sonore d’un ou plusieurs phonèmes correspondant à un début de mot ? 3.1 La reprise Le cas le plus simple est celui où s’effectue une reprise. Le mot interrompu est complété immédiatement : [13] je me suis fait no- nommer ici à l’o- à l’occasion d’un changement de de grade euh (1217) A partir de cet exemple on constate que les reprises concernent : - soit le mot entier : « de de », - soit la première syllabe d’un mot : « no- nommer », - soit la première syllabe d’un mot compris à l’intérieur d’une locution prépositionnelle « à l’o- à l’occasion ». Il y a une différence fonctionnelle entre la reprise d’un mot, qui reproduit une découpe grammaticale, et l’amorce. Celle-ci se conforme à un patron strictement phonologique, indépendamment de toute considération morphologique (la restitution du préfixe dans occasion n’est pas accessible en synchronie pour les locuteurs) et, comme le montre l’extension des deux amorces, la séquence à considérer peut être extensible, du mot au syntagme. La taille de l’extrait à examiner pour identifier l’amorce est également à prendre en compte dès lors que la reprise peut intervenir à un ou plusieurs mots de distance : [14] quand je suis arrivée à Orléans j’ai laissé tomber la m- j’avais laissé tom- enfin ch- (1371) Reconstituer « tomber » à partir de « tom- » s’impose : la structure de la phrase est réitérée à l’identique et la modification concerne seulement l’aspect verbal (un plus-que-parfait au lieu d’un passé composé) et non le contenu de l’information. Le « ch- » final serait plus conjectural. Il est presque sûr, à l’audition, qu’il correspond à une réitération du « je suis » au terme d’une assimilation très fréquente à l’oral. On relèvera la permutation de l’ordre : du cas considéré jusqu’à présent, c’est-à-dire de l’amorce reformulée dans le mot entier, on passe à une succession inverse où le mot (tomber) précède son amorce (tom-), ce qui exclut une explication qui se fonderait sur des difficultés d’accès lexical. 3.2 La substitution Un autre cas de figure correspond à la substitution d’un terme suffisamment proche par la forme et par le sens pour qu’il soit possible de restituer la forme enclenchée : [15] y a une dizai- oh à peine dix ans (1927) De « dix ans » à « une dizaine (d’années) », l’équivalence est explicite, la correction portant sur la paraphrase de l’énoncé qui, de « à peu près dix ans », passe à « un peu moins de dix ans ». La synonymie a la particularité de ménager une certaine variation dans l’interprétation. L’assignation peut se faire sans qu’il y ait d’apparentement entre les termes, la disposition syntaxique et l’énoncé d’un terme de valeur proche dans l’environnement immédiat permettant de compléter l’amorce selon le principe de permutation paradigmatique. En voici deux exemples : [16] je laisse couler comme j’ai dit j’aurais obli- j’aurais forcément (1828) Il est probable que « obli- » s’explicite en « obligatoirement » sans qu’il soit possible de fournir une explication aux raisons qui ont décidé d’une modification lexicale tardive. Il en va de même pour : [17] vous avez eu en- vous vous aimez changer (1819) 7 SHS Web of Conferences 27, 05005 (2016) DOI: 10.1051/ shsconf/20162705005 Congrès Mondial de Linguistique Française - CMLF 2016 où la locution « avoir envie » semble un substitut envisageable dont il serait difficile de dire en quoi elle est mieux adaptée à l’intention communicative. Dans une approche paradigmatique de substitution se retrouvent des occurrences telles que : [18] je crois que c’est au Quai bran- je suis pas sûre ou au musée d’Orsay je sais plus (1866) où le contexte élargi et la référence à un second musée parisien également établi sur les quais rive gauche dans les trente dernières années incitent à compléter par « quai Branly » (en intervertissant les majuscules du transcripteur). La restitution se rapprocherait du cas des anaphores associatives (Kleiber 2009) lorsque, dans un échange où il est fait mention du travail dans une banque, on est amené à supputer « agence » : [19] à Toury euh on était une agen- on était cinq hein (1704) 3.3 La rectification La modification peut être rendue nécessaire par la correction d’un lapsus : [20] de bateaux les baga- les gabares les (1744) Un mot spécialisé, de faible fréquence, « gabares », au terme d’une permutation de consonnes, appelle « bagarres » et l’élimine avant même d’être parvenu au bout de sa réalisation. Parfois, en lieu et place d’une inversion, on trouve une anticipation : [21] comme ça m'av- bl- ça m'avait bien plu (1522) où la combinaison /occlusive + liquide/ de « plu » est réalisée en avance sur « bien », ou au contraire une persévération : [22] on lui avait attribué un lun- un logement (1582) quand la nasale du déterminant se substitue à la première syllabe du mot suivant. Il se rencontre aussi des cas où une construction syntaxique doit être remaniée afin d’éviter une tournure agrammaticale, comme le recours à la forme pronominale dans : [23] donc ça complique la tâche d’adap- de s’adapter enfin moi je pensais que tu en avais (1159) ou le respect d’un accord en genre (ou en nombre) : [24] de par son s- sa région (1074) voire une liaison ou, plus généralement, toutes les formes d’enchaînement, en particulier pour des monosyllabes à vocalisme en schwa qui requièrent une élision : [25] la vieille l- l’hôtel de la Vieille Intendance (1400) 3.4 L’interruption Une dernière condition permet le rétablissement du mot interrompu, c’est la réalisation d’un nombre suffisant de phonèmes pour que l’incertitude soit levée, surtout quand le terme supposé s’ajuste bien au contexte : [26] donc là c'est pas vraiment une répartis- c'est plus une question de place (1340) où « répartition » s’impose de lui-même, la transcription reproduisant la prononciation de la constrictive /s/ sans égard à la convention orthographique. Si on récapitule les quatre cas de figure qui paraissent à même d’expliquer l’ensemble des occurrences pour lesquelles l’amorce peut être élucidée, on aboutit aux subdivisions suivantes : 8 SHS Web of Conferences 27, 05005 (2016) DOI: 10.1051/ shsconf/20162705005 Congrès Mondial de Linguistique Française - CMLF 2016 a) Les reprises (qui s’apparentent à la répétition) comprennent : - la reprise par un même mot - qu’il s’agisse de l’énoncer seul ou de rétablir l’ensemble d’un groupe syntagmatique (locution, SN, SV…), - que l’explicitation se fasse immédiatement ou après (voire avant) que d’autres mots se sont intercalés ; - la substitution, que le mot soit de la même famille ou non (substitution paradigmatique), voire qu’il relève de l’anaphore associative. En général, l’amorce précède le mot entier. b) Les rectifications sont distinguées selon qu’elles concernent : - un lapsus, - le remaniement d’un enchaînement phonétique (élision, liaison…), - la mise en conformité d’un accord grammatical. A chacun des cas correspond une opération différente, autant lorsque l’amorce est préservée : - en étant complétée dans le cas de la reprise, [27] j’ai pas fait hein maleu- enfin pas malheureusement mais (1604) - ou corrigée pour les rectifications, [28] euh moi je- j’aime beaucoup le théâtre j’aime beaucoup le cinéma (1631) que dans les cas où : - elle est remplacée avec la sélection d’un autre élément dans le même paradigme, [29] et alors le conservat- enfin le directeur à l’époque s’était (1620) - ou abandonnée quand la phrase est reprise à son point de départ, l’amorce restant inachevée. [30] Oui deux g- deux cousines n- (1692) On a d’un côté un ajout (le mot est complété) ou une réduction (l’amorce reste la seule occurrence), de l’autre une modification, qu’elle soit phonologique, morphologique ou lexicale. 4. Que restitue-t-on ? On laissera le cas particulier des segments qui n’ont pas d’équivalent, sans représentant dans le lexique mental : mot inconnu du locuteur – un terme technique par exemple –, nom propre, emprunt à une langue étrangère… Dans tous ces cas, seul le signal informe l’interprétation. Dès lors que la cible est connue par l’auditeur, sa stratégie pour restituer les éléments bruités (mot incomplet, prononciation défectueuse, conditions d’écoute difficile) ne paraît pas différer sensiblement de celle exposée à propos des amorces. Le partage tiendrait à la nécessité éprouvée de reconstituer celles-là quand l’élimination de celles-ci est partie intégrante des conditions d’une bonne réception des messages. On mentionnera cependant l’exploitation qui peut en être faite dans un second temps pour interpréter ce que seraient les intentions de celui dont la langue a fourché. Depuis Freud, le lapsus est investi d’une valeur emblématique Sans faire entrer en ligne de compte les cas d’indécision, on examinera trois cas de figure que les amorces ont permis de prendre en considération, (i) la correspondance avec la graphie, (ii) la pluralité d’identification et (iii) la détermination de la forme sous-jacente (FSJ) telle qu’elle doit être restituée. 9 SHS Web of Conferences 27, 05005 (2016) DOI: 10.1051/ shsconf/20162705005 Congrès Mondial de Linguistique Française - CMLF 2016 4.1 Correspondance Comment parvient-on à déduire d’une séquence sonore un mot dont les phonèmes ne s’ajustent pas au signal phonétique ? Les conventions de transcription spécifient que les mots doivent être écrits conformément à leur orthographe sans que soient rétablis des mots absents. Par suite, une séquence phonétique entendue comme « chépa » sera reproduite sous la forme « je sais pas », le premier élément de la négation, « ne », n’apparaissant pas. Ainsi, les conventions résolvent par avance de nombreux cas dont il ne demeure de traces sporadiques que dans des amorces où le transcripteur, faute d’une interprétation, s’est résigné à conserver l’impression phonétique. Dans l’exemple suivant, il pourrait s’agir d’un « je » suivi d’un mot commençant par une consonne sourde (tels que « je pense » ou « je crois »), jusqu’à ce qu’une construction impersonnelle conduise à l’abandon de la proposition commencée : [31] donc euh donc ch- ça serait bien (1101) Dans l’exemple [32], il est difficile de décider s’il s’agit d’un cas du même ordre (« chuis gênée ») ou bien d’un lapsus, la persévération d’une réalisation palatale de la constrictive sourde en lieu et place de la dentale (* « je chuis (gênée) ») : [32] une fois <Nom de personne> elle dit oh je ch- je suis gênée ben je dis tu as pas (2571). Si ces cas sont moins nombreux qu’on ne s’y attendrait, ce n’est pas qu’ils manqueraient dans les propos enregistrés, c’est qu’ils sont tranchés par avance dans la phase de traitement initial en sorte que les occurrences restantes ne soulèvent pas moins de questions d’interprétation pour l’analyse que pour la transcription. 4.2 Pluralité d’identification Autre difficulté : le choix de la forme à sélectionner dès lors que plusieurs paraissent également possibles dans le même environnement, comme dans le cas du déterminant où, à partir de « l- », l’amorce peut référer aussi bien à le, la ou les. [33] je vais plutôt demander si l- si le produit sort euh (1151) Ici, la reprise au masculin singulier ratifie rétroactivement « le » mais rien ne permet d’affirmer qu’au moment où l’énoncé a été interrompu, le locuteur avait déjà tranché. De même pour le pluriel : [34] et et euh l- les les grandes manifestations orléanaises (1653) Le choix opéré a posteriori laisserait supposer que la forme retenue était celle attendue mais on rencontre des cas de rectification : [35] Par par cet- ce ce refus de rentrer dans (1081) L’indécision s’accroît de façon inversement proportionnelle au nombre de phonèmes disponible. Une réduction à un seul phonème, la nasale labiale par exemple, sera le plus souvent interprétée comme une marque d’approbation ou d’hésitation notée « mh », ou reportée en amorce comme dans : [36] non m- parce que c’est c’est euh (1029) où le « m -» laisse conjecturer un « mais ». Si on examine le cas du « s- » qui est l’amorce la plus fréquente – elle compte à elle seule pour environ 10 % des occurrences tous cas confondus –, on constate qu’elle équivaut à « ce » (1007), « ça » (1016), « c’est » (1030), « c’était » (1051), « son » (1073), « sa » (1074), « se » (1083), « si » (1111), « suis » (1142) etc. [37] mh alors s- ce que vous avez envie de faire c’est plutôt quoi ? (1007) 10 SHS Web of Conferences 27, 05005 (2016) DOI: 10.1051/ shsconf/20162705005 Congrès Mondial de Linguistique Française - CMLF 2016 [38] donc je fais s- je je fais ça euh (1016) [39] c’est s- c’est un outil intéressant pour euh (1030) [40] dans le jardin là ça s- je trouvais très (1051) [41] de s- de par son origine sociale et de par euh (1073) [42] de par son s- sa région (1074) [43] euh un lieu où y s- y se se mélange beaucoup de de milieux (1083) [44] et servir oui parce que si s- avec votre accord donc (1111) [45] c’est pour ça oui alors je te disais Paris tout à l’heure mais c’est vrai que je s- je suis parti d'Orléans aussi (1142) Dans chacun des cas, la suite de la proposition permet de lever l’incertitude pour autant que l’énoncé n’est pas remanié complètement. L’amorce apparaît comme un arrêt au moment de choisir un mot qui dépend moins de la mise à contribution du stock lexical que de l’arrangement morpho-syntaxique dont l’enchaînement présente une difficulté au moment de son énonciation. Dans les neuf exemples, il s’agit de mots-outils (pronoms, déterminants, présentatif, conjonction, auxiliaire…) en sorte que l’amorce n’est pas liée à la difficulté de le réaliser en tant que tel mais à la capacité du locuteur à décider ce qui doit suivre et qui concerne le contenu du message plus que sa phraséologie. 5 La « Forme Sous-Jacente » Un dernier cas doit être considéré qui implique la forme sous-jacente ou FSJ (Brandão de Carvalho, Nguyen & Wauquier 2010). La FSJ correspond à la séquence de phonèmes (associée dans d’autres langues à des traits prosodiques) qui caractérisent structuralement un morphème ou une unité lexicale. Par sa consignation mentale, elle sert de référence pour l’identification du mot par l’auditeur, dans la superposition postulée entre l’I-Language et l’E-Language. Une équivalence globale, assez peu interrogée, postule une certaine équivalence entre la transcription en API donnée par les dictionnaires, augmentée des transformations morphologiques attendues (dans le cas d’un verbe conjugué, d’un pluriel irrégulier, d’une modification liée au genre…) et cette FSJ. Les amorces rendent possibles un réexamen de ce point de vue. Au nombre des difficultés de circonscription de la FSJ, relevées depuis longtemps, a été pointée la pertinence de la notation de la consonne de liaison (Encrevé 1988) ou de certains schwas. Quels sont aujourd’hui les cas d’emploi réels du schwa dans un mot comme « bouleverser » en français central ? Faut-il indiquer un schwa à la fin d’un mot comme « arc » qui apparaîtra dans un syntagme comme « arc brisé » ? La prise en compte des amorces soulève une interrogation d’une autre nature. La convention de transcription stipule que tout mot, identifiable ou non, qui n’aura pas été réalisé pleinement constitue une amorce et doit être consigné comme tel. Faisant retour des réalisations orales vers la structure de la langue, on voudrait interroger ce qu’il en est de la FSJ concernant les pronoms « il(s) » et « elle(s) » – pour celui-ci en restreignant l’observation à sa fonction de sujet, en excluant les emplois en apposition : [46] elle était dans quel d- domaine elle ? (1595) ou en complément : [47] parce que bon i- elle est elle avait sympathisé avec elle (1655) L’hypothèse sera formulée de la façon suivante : Devant consonne, la forme standard de il/ils est i-, de elle/elles est e- (prononcé « è »). Devant voyelle, la forme standard est il, el et, au pluriel, iz, èz. 11 SHS Web of Conferences 27, 05005 (2016) DOI: 10.1051/ shsconf/20162705005 Congrès Mondial de Linguistique Française - CMLF 2016 Un allomorphisme phonétique, d’ordre contextuel, est observable, marqué par l’apparition d’une consonne. En cas de liaison obligatoire au pluriel (« ils ont », « ils étaient ») le /l/ reste élidé. Cela concerne une consonne qui ne figure pas dans la liste de celles qui sont impliquées, en français, dans la liaison mais surtout, de façon inverse à ce qui se produit avec les liaisons facultatives, le /l/ peut se maintenir devant une consonne, comme cela a pu être relevé pour d’autres mots tels que but, fait ou quand. De même, l’absence de réduction dans des fonctions autres que le sujet rejoint partiellement le comportement du schwa qui ne peut intervenir dans les positions emphatiques, soit par substitution de forme (« je » > « moi »), soit par accentuation (« dis-le » > *« dis-l »). Signe de l’embarras de certains transcripteurs qui ne perçoivent pas le /i/ isolé comme une amorce, la tentation de lui substituer graphiquement un « y » qui ne se justifie pas grammaticalement : [48] Ou alors y- des fois ils font des euh des créneaux ce qu’ils appellent des réceptions en libre (1164) Du fait que sur les mille premières phrases comprenant une amorce, aucun des {il/ils/elle/elles + consonne} en position sujet ne se présente avec sa forme complète, on devrait conclure que la FSJ ne comporte pas le /l/ comme un élément obligatoire mais plutôt comme un phonème alternant de liaison, au même titre que le /z/ du pluriel. Il est possible que le phénomène ait une certaine profondeur diachronique que la graphie a masquée. 6 Conclusion Au-delà du phénomène de disfluence et de ses effets, l’amorce permet d’observer comment les locuteurs interprètent les éléments phonologiques disponibles pour interpoler les informations manquant dans le signal. Parce que cette situation peut être généralisée à l’ensemble des cas où l’auditeur doit établir une identification lexicale (de la trace phonétique à la FSJ) en dépit d’une disponibilité lacunaire des informations contenues dans le message, l’amorce est qualifiable en tant qu’expérience et acquiert une portée supérieure à ce que le terme négatif de disfluence pourrait laisser croire. A côté des modes compensatoires syntagmatiques (contextualisation et reprise) et paradigmatiques, qui ratifient la FSJ, on a pu noter un effet en retour de l’analyse des productions sur la FSJ. Le pronom personnel sujet de troisième personne ne se présente pas avec la forme attendue non seulement phonétiquement, comme il en va pour n’importe quel mot, mais aussi phonologiquement. La consignation d’une consonne qui a déjà disparu en cette position (de la même façon qu’elle a disparu dans le mot outil), au croisement de la morpho-syntaxe (seule la position sujet est concernée) et de la phonologie pose la question de la représentation abstraite de cette unité lexicale. La réduction à une prothèse vocalique (hors liaison) du pronom sujet de la 3e personne (la non personne) pourrait transformer en préfixe la marque devenue verbale, l’amuïssement de la consonne conduisant à un appui sur le mot suivant comme dans « il me dit » prononcé /im-di/ ou « il se croit » prononcé /is-kRwa/, ce qui laisse entrevoir des possibilités d’infixation (celle des pronoms réfléchis), voire l’éventualité d’un effacement complet déjà effectif dans un syntagme figé comme « (il) faut ». Références bibliographiques Brandão de Carvalho, J., Nguyen, N., Wauquier-Gravelines S. (2010) Comprendre la phonologie. Paris : PUF. Chomsky, N. (1986) Knowledge of Language. New York : Praeger. Encrevé, P. (1988) La Liaison avec et sans enchaînement. Paris : Seuil. Jackson, J. H. (1932) Selected Writings, édition par J. Taylor (2 vol.). Londres : Hodder & Stoughton. Kleiber, G. (2001) L’Anaphore associative. Paris : PUF. 12
https://openalex.org/W4361884111	https://figshare.com/articles/journal_contribution/Supplementary_Figure_S2_from_Enhanced_Delivery_of_SN-38_to_Human_Tumor_Xenografts_with_an_Anti-Trop-2_SN-38_Antibody_Conjugate_Sacituzumab_Govitecan_/22456001/1/files/39907157.pdf	English	null	Supplementary Figure S2 from Enhanced Delivery of SN-38 to Human Tumor Xenografts with an Anti-Trop-2–SN-38 Antibody Conjugate (Sacituzumab Govitecan)	null	2,023	cc-by	668	Figure S2. Clearance of IMMU-132 from mouse serum. (A) Measurement of SN-38 release from IMMU- 132 held in mouse serum in vitro (37 oC). (B) Determination of IMMU-132 clearance rate in mice based on ELISA measurements of IMMU-132 (SN-38 capture/anti- hRS7 idiotype antibody probe) or hRS7 IgG, as well as measurement of SN-38 [TOTAL] in acid-hydrolyzed samples assayed by reversed-phase HPLC. Figure S2. Clearance of IMMU-132 from mouse serum. (A) Measurement of SN-38 release from IMMU- 132 held in mouse serum in vitro (37 oC). (B) Determination of IMMU-132 clearance rate in mice based on ELISA measurements of IMMU-132 (SN-38 capture/anti- hRS7 idiotype antibody probe) or hRS7 IgG, as well as measurement of SN-38 [TOTAL] in acid-hydrolyzed samples assayed by reversed-phase HPLC. Figure S2. Clearance of IMMU-132 from mouse serum. (A) Measurement of SN-38 release from IMMU- 132 held in mouse serum in vitro (37 oC). (B) Determination of IMMU-132 clearance rate in mice based on ELISA measurements of IMMU-132 (SN-38 capture/anti- hRS7 idiotype antibody probe) or hRS7 IgG, as well as measurement of SN-38 [TOTAL] in acid-hydrolyzed samples assayed by reversed-phase HPLC. ( C)Pharmacokinetic parameters after IV bolus dosing of IMMU-132 (1.0 mg), monitoring the clearance of IMMU-132 and hRS7 IgG detected by ELISA Drug Tmax (h) T½ (h) Cmax (µg/mL) AUC(0-last) (hµg/mL) V (mL) Cl (mL/h) MRT (h) IMMU-132 1 14.0 400.7 7,163 2.75 0.136 19.0 hRS7 IgG 1 67.1 468.2 13,368 4.19 0.043 86.7 Parameters estimated using a non-compartmental analysis (linear-log trapezoidal method). Figure S2. Clearance of IMMU-132 from mouse serum. (A) Measurement of SN-38 release from IMMU- 132 held in mouse serum in vitro (37 oC). (B) Determination of IMMU-132 clearance rate in mice based on ELISA measurements of IMMU-132 (SN-38 capture/anti- hRS7 idiotype antibody probe) or hRS7 IgG, as well as measurement of SN-38 [TOTAL] in acid-hydrolyzed samples assayed by reversed-phase HPLC. Half-life 17.45 h C 0 2 4 4 8 7 2 9 6 1 2 0 0 1 0 0 2 0 0 H o u rs S N -3 8 x 1 0 6 0 2 4 4 8 7 2 1 0 1 1 0 2 1 0 3 1 0 2 1 0 3 1 0 4 H o u rs E L IS A (g /m L ) h R S 7 Ig G a n d IM M U -1 3 2 H P L C (n g /m L ) S N -3 8 [T o ta l] h R S 7 Ig G IM M U -13 2 (E L IS A ) S N -3 8 [T O T A L ] A B Half-life 17.45 h 0 2 4 4 8 7 2 9 6 1 2 0 0 1 0 0 2 0 0 H o u rs S N -3 8 x 1 0 6 A 0 2 4 4 8 7 2 1 0 1 1 0 2 1 0 3 1 0 2 1 0 3 1 0 4 H o u rs E L IS A (g /m L ) h R S 7 Ig G a n d IM M U -1 3 2 H P L C (n g /m L ) S N -3 8 [T o ta l] h R S 7 Ig G IM M U -13 2 (E L IS A ) S N -3 8 [T O T A L ] B B A ( C)Pharmacokinetic parameters after IV bolus dosing of IMMU-132 (1.0 mg), monitoring the clearance of IMMU-132 and hRS7 IgG detected by ELISA C C ( C)Pharmacokinetic parameters after IV bolus dosing of IMMU-132 (1.0 mg), monitoring the clearance of IMMU-132 and hRS7 IgG detected by ELISA Drug Tmax (h) T½ (h) Cmax (µg/mL) AUC(0-last) (hµg/mL) V (mL) Cl (mL/h) MRT (h) IMMU-132 1 14.0 400.7 7,163 2.75 0.136 19.0 hRS7 IgG 1 67.1 468.2 13,368 4.19 0.043 86.7 Parameters estimated using a non-compartmental analysis (linear-log trapezoidal method). ameters estimated using a non-compartmental analysis (linear-log trapezoidal method).
https://openalex.org/W2019441307	https://ntrs.nasa.gov/api/citations/19730011434/downloads/19730011434.pdf	English	null	A New Technique for Auger Analysis of Surface Species Subject to Electron-Induced Desorption	Review of scientific instruments online/Review of scientific instruments	1,973	public-domain	5,558	NASA TECHNICAL NOTE 10 cs NASA TN D-7257 FIL C O P V ^d? tl i A NEW TECHNIQUE FOR AUGER ANALYSIS OF SURFACE SPECIES SUBJECT TO ELECTRON-INDUCED DESORPTION by Stephen V. Pepper Lewis Research Center Cleveland, Ohio 44135 NATIONAL AERONAUTICS AND SPACE ADMINISTRATION WASHINGTON D C APRIL 1973 10 cs 10 cs A NEW TECHNIQUE FOR AUGER ANALYSIS OF SURFACE SPECIES SUBJECT TO ELECTRON-INDUCED DESORPTION by Stephen V. Pepper Lewis Research Center Cleveland, Ohio 44135 by Stephen V. Pepper Lewis Research Center Cleveland, Ohio 44135 A NEW TECHNIQUE FOR AUGER ANALYSIS OF SURFACE SPECIES SUBJECT TO ELECTRON-INDUCED DESORPTION by Stephen V. Pepper Lewis Research Center SUMMARY A method is presented to observe surface species subject to electron-induced de- sorption by Auger electron spectroscopy. The surface to be examined is moved under the electron beam at constant velocity, establishing a time-independent condition and eliminating the time response of the electron spectrometer as a limiting factor. The dependence of the Auger signal on the sample velocity, incident electron current, beam diameter and desorption cross section are analyzed. It is shown that it is advantageous to analyze the moving sample with a high beam current, in contrast to the usual practice of using a low beam current to minimize desorption from a stationary sample. The method is illustrated by the analysis of a friction transfer film of PTFE, in which the fluorine is removed by electron-induced desorption. The method is relevant to surface studies in the field of lubrication and catalysis. NATIONAL AERONAUTICS AND SPACE ADMINISTRATION • WASHINGTON, D. C. • APRIL 1973 NATIONAL AERONAUTICS AND SPACE ADMINISTRATION • WASHINGTON, D. C. • APRIL 1973 Auger spectrum. It is, therefore, advantageous to work with an electron spectrometer with the fast- est possible response time when detecting desorbable surface species. In this sense the cylindrical mirror analyzer (ref. 7), with its fast tracing time of approximately 100 mil- liseconds, is superior to the retarding potential analyzer (ref. 8) with its slower tracing time of minutes. However, the fast response time of the cylindrical mirror analyzer is useful only for detecting major Auger peaks. Detecting minor peaks requires an in- crease in lock-in amplifier sensitivity resulting in greater noise. Removing the noise by filtering results in a slower time response. Thus no matter which method is used, there is always the suspicion that a particular species might be desorbed before its Auger signal can be detected. It is the purpose of this report to present a method for obtaining the Auger spectra of surface species subject to electron-induced desorption that does not depend on the time response of the electron spectrometer. The method consists of moving the sample under the electron beam at a constant velocity. Thus, fresh sample is continuously being supplied for Auger analysis, and the system reaches a steady state, allowing great flexibility in the choice of electron spectrometers and lock-in amplifier time constants. The uncertainty as to whether a species has been desorbed before its Auger signal has been detected is eliminated and the Auger analyses of desorbable and nondesorbable species are placed on a more equivalent basis. In the next section the dependence of the Auger signal of the desorbable species on the surface velocity and electron beam current is analyzed and discussed. An experi- mental illustration of the technique is then given in the last section. INTRODUCTION The technique of Auger electron spectroscopy has been widely used to identify ele- ments present on solid surfaces (refs. 1 to 4). This technique, however, can be a de- structive method of surface analysis. The electron beam (500 to 5000 V, 1 to 100 fiamp) that ionizes the atoms in the surface region can promote desorption of surface species (refs. 5 and 6). The Auger signal of this element then decreases exponentially in time. The desorption of particular elements from the surface can thus lead to an incorrect assessment of the elemental composition of the surface by Auger electron spectroscopy. This problem has usually been dealt with by either decreasing the incident beam current or by defocusing the beam spot while maintaining the same total electron cur- rent. The effect of these changes is to decrease the desorption rate so that there is, hopefully, negligible change in the concentration during the time necessary to obtain the Auger spectrum. ANALYSIS The basic relation that describes the depletion of a surface species under bombard- ment by monoenergetic electrons is -lot/jrrjl N(t) - N(0)e U (1) (1) where N(t) is the surface number density of the particular species at time t, I is the incident electron current in the beam spot of radius TQ, and a is the desorption cross section (ref. 9). Since the magnitude of the Auger signal S is proportional to the sur- face density of the species, we have where N(t) is the surface number density of the particular species at time t, I is the incident electron current in the beam spot of radius TQ, and a is the desorption cross section (ref. 9). Since the magnitude of the Auger signal S is proportional to the sur- face density of the species, we have -Iot/jrr« S(t) = S(0)e ° (2) (2) I S(t) = S(0)e where S(O) is the magnitude of the Auger signal in the absence of electron-induced de- sorption, conventionally measured as the peak-to-peak height in the dN/dE spectrum. The exponential function in equation (2) is an attenuation factor (<1) that describes the loss of Auger signal due to desorption and is a function only of the dimensionless param- 2 eter lot/TIT Q. We now seek a relation analogous to equation (2) for the magnitude of the Auger sig- nal from a sample moving under the electron beam with velocity v. For this purpose consider the geometry depicted in figure 1. The sample moves to the left under the Sample ^Circumference of beam Figure 1. - Schematic of electron beam spot bombarding a surface moving to the left with velocity v. The spot radius is TQ, and the electron current I is uniform within the spot. Sample Figure 1. - Schematic of electron beam spot bombarding a surface moving to the left with velocity v. The spot radius is TQ, and the electron current I is uniform within the spot. Figure 1. - Schematic of electron beam spot bombarding a surface moving to the left with velocity v. The spot radius is TQ, and the electron current I is uniform within the spot. electron beam of radius rQ; current I is assumed to be uniform within the spot. The desorbing species is assumed to be uniformly distributed on the surface. ANALYSIS Then the Auger signal from an elemental area of the moving sample within the beam spot is given by dS = S(O) exp _ Iot(x, y) dx dy (3) (3) •is the time that the element of sample at (x, y) has spent under electron bombardment. The total Auger signal is then given by integration over the area under bombardment; that is, S - dS =_S(0) jrr, ^exp ^V r / \~I Ia(yr2 - y2 - x) Jrrov dx dy (4) (4) The integration over x is performed first. Then a change of variable for the y integration, z = Y / T , gives the result S = S(O)—^1 - (5) (5) (6) Defining exp (7) exp (7) (7) we finally have we finally have S = (8) (8) Equation (8) is the expression for the Auger signal that is analogous to equation (2). The function tp (/3) is the attenuation factor that describes the loss of Auger signal due to electron-induced desorption from a moving sample and is a function only of the lump- ed dimensionless parameter /3. The function <p ()3) has been numerically evaluated and is plotted as curve A in figure 2(a). Consider the dependence of the Auger signal on the physical parameters that make up the dimensionless parameter /3. Since the surface velocity v appears in the denom- inator of ]3, the velocity dependence of S is best understood in terms of ]3~ . Thus (f> Curve A f(p) B wp4>(p)/4 (a) Attenuation function $ and function Tt0$((5)/4 that describes Auger signal as function of incident beam current (v constant) plotted against dimensionless parameter p = 2Io/Tirgv. 10 P'fl rl (b) Attenuation function $ plotted against p . This curve describes behavior of Auger signal as a func- tion of sample velocity. Figure 2. - Attenuation functions that describe Auger signal of electron-desorbable species on surface moving with velocity v. Curve A f(p) B wp4>(p)/4 (p) B wp4>(p)/4 (a) Attenuation function $ and function Tt0$((5)/4 that describes Auger signal as function of incident beam current (v constant) plotted against dimensionless parameter p = 2Io/Tirgv. 10 P'fl rl (b) Attenuation function $ plotted against p . This curve describes behavior of Auger signal as a func- tion of sample velocity. Figure 2. - Attenuation functions that describe Auger signal of electron-desorbable species on surface moving with velocity v. has been plotted in figure 2(b) as a function of ;3~ . The signal vanishes for zero veloc- ity since the measurements considered here are time independent and the relevant spe- rl (b) Attenuation function $ plotted against p . This curve describes behavior of Auger signal as a func- tion of sample velocity. rl (b) Attenuation function $ plotted against p . This curve describes behavior of Auger signal as a func- tion of sample velocity. Figure 2. - Attenuation functions that describe Auger signal of electron-desorbable species on surface moving with velocity v. Figure 2. - Attenuation functions that describe Auger signal of electron-desorbable species on surface moving with velocity v. Defining has been plotted in figure 2(b) as a function of ;3~ . The signal vanishes for zero veloc- ity since the measurements considered here are time independent and the relevant spe- cies on a stationary surface is eventually totally desorbed by the electron beam in accordance with equation (2). For large )3~ (large v) the Auger signal asymptotically approaches the value it would have in the absence of desorption. In this velocity region, fresh sample is being fed into the beam spot at a rate high enough to maintain the aver- age coverage in the spot close to the coverage outside the spot. It is therefore advan- tageous to operate with the highest practical velocity. There are certain points of similarity between the method presented here and the usual time-dependent method. In the first place, higher desorption cross sections lead to smaller signals in both cases. Secondly, a smaller beam radius (for the same total beam current) leads to a higher desorption rate and a smaller signal. It is thus advan- tageous to defocus the incident beam in both methods. However, the dependence of the signal on the incident electron beam current is quite different in the two methods. The current dependence of the Auger signal here is contained not only in (p(P) but also in S(O), since the Auger signal in the absence of de- sorption is itself proportional to the current. The behavior of the signal with current is therefore the result of a trade-off; larger I yields a larger signal from those atoms present on the surface (S(O) ~ )3), but it also means a higher desorption rate of these atoms (<p(/3)). Since P ~ I, the functional form of the dependence of the signal S(/3) on current is given by /3^(/3). Multiplying by a factor IT/4 yields the function plotted as curve B in figure 2(a). The curve is the Auger signal, relative to its maximum value for a given velocity, as a function of the incident current. From the monotonic increase it is seen that the increase in signal due to the increase in S(O) dominates the decrease in signal due to enhanced desorption. An upper limit on the Auger signal, for a given v, arises when the species is completely desorbed before it leaves the beam spot and represents the most efficient use of the desorbable species for the purpose of Auger electron spectroscopy. Defining Thus, in contrast to the usual practice of reducing the beam current to minimize the desorption rate, here it is advantageous to work with a high beam current to maximize the time independent Auger signal. As a numerical example of this analysis, consider a fast desorption time of 5 sec- 2 1 onds (Ia/7rrQ = 0.2 sec~ ) and rQ = 0. 05 centimeter. Then /3 = 0. 01/v and a velocity of 0. 001 centimeter per second yields /3 - 10 and (p - 0.13 (from fig. 2(a), curve A). This reduction in signal strength is quite modest when it is remembered that this method permits the use of high sensitivity and appreciable noise filtering on the lock-in ampli- fier. Finally, it should be noted that the velocity and current dependence of the signal depicted in figure 2 may be used to experimentally determine the desorption cross sec- tion a. Such a procedure requires obtaining the signal over a wide range of either velocity or beam current and searching for the best fit of the relevant curve to the data. Such a procedure would be analogous to plotting the time-dependent signal from a sta- tionary surface and using equation (2) to obtain <j. However, just as the time response of the electron spectrometer imposes a limitation on the use of equation (2), there may be practical difficulties to the use of the velocity or current dependence of the signal for this purpose. For example, increasing the beam current may result in an enlarged beam spot, leading to a signal that increases instead of saturating as in curve B, fig- ure 2(a). The particular experimental situation will no doubt dictate the particular method to be used, and therefore no general procedure can be given. ILLUSTRATION An illustration of the essential features of the method described herein is provided by the examination of a mechanically applied thin lubricant film of polytetrafluoroethyl- ene (PTFE) on a tungsten surface. It has recently been shown that when PTFE slides on a surface, a transfer film only a few mono layers thick is developed on the surface (refs. 10 and 11). Such a film was developed here by sliding the tip of a PTFE bullet (radius, 0.476 cm) on a sputter-cleaned tungsten disk in vacuum at a load of 100 grams and a velocity of 0.1 centimeter per second (ref. 11). The experimental arrangement is depicted in figure 3. Cylindrical mirror analyzer -, Integral electron Electron beam Tungsten disk - PTFE bullet • 11 Magnetic rotary feedth rough Figure 3. - Experimental arrangement used to apply PTFE transfer film to disk in vacuum. PTFE bullet was raised from disk after film was applied. Cylindrical mirror analyzer -, Magnetic rotary feedth rough Figure 3. - Experimental arrangement used to apply PTFE transfer film to disk in vacuum. PTFE bullet was raised from disk after film was applied. The Auger spectra from this surface are presented in figure 4. In figure 4(a) the spectrum from the stationary surface under bombardment by the 2000-volt, 10- microampere beam for 2 minutes indicates the major surface species are tungsten and carbon from the PTFE. Fluorine is present only as a very minor peak. The Auger spectrum from the moving surface presented in figure 4(b) indicates the presence of 100 o "o "o on o> 7 oil 15 -O f I 6° \|\| 40 u >•> -S en > c 20 Tungsten J FluorineJ Carbon J/- Carbon - Tungsten Secondary electron energy, E-- (a) Spectrum for stationary (b) Spectrum from same sample after electron bom- sample moving with bardment for 2 minutes. velocity v • 0.018 centi- meter per second. Figure 4. - Auger spectrum of FTFE transfer film on tung- sten surface. Electron beam current, 10 microamperes; beam voltage, 2000 volts. 100 o "o "o on o> 7 oil 15 -O f I 6° \|\| 40 u >•> -S en > c 20 Tungsten J FluorineJ Carbon J/- Carbon - Tungsten Secondary electron energy, E-- (a) Spectrum for stationary (b) Spectrum from same sample after electron bom- sample moving with bardment for 2 minutes. velocity v • 0.018 centi- meter per second. Figure 4. ILLUSTRATION - Auger spectrum of FTFE transfer film on tung- sten surface. Electron beam current, 10 microamperes; beam voltage, 2000 volts. Figure 4. - Auger spectrum of FTFE transfer film on tung- sten surface. Electron beam current, 10 microamperes; beam voltage, 2000 volts. much more fluorine on the surface than is indicated by figure 4(a). Note also that the size of the tungsten and carbon peaks from the moving surface are smaller than those from the stationary surface, indicating that the fluorine in the PTFE chain attenuates the Auger electrons of the tungsten and carbon atoms. Thus all the major peaks in Auger spectrum are affected by electron-induced desorption of the fluorine and the con- venience of exhibiting the entire Auger spectrum on a time-independent basis is appar- ent. The velocity dependence of the magnitude of the fluorine Auger signal is presented in figure 5. There are three main features to this data. First, note that the signal does not vanish for zero velocity indicating that some of the fluorine is not subject to electron-induced desorption. Secondly, the signal increases rapidly for small velocities and then starts to level off. The general behavior is thus similar to that depicted in fig- ure 2(b) for the velocity dependence of the signal. Finally, note that, instead of reaching S(v) - A Experimental data — Theoretical curve based on o = 5xlO"18 cm2 S(0)= 100 rn - 0.05 cm 0 .005 .01 .015 .04 .08 .12 .16 .18 .22 Velocity of sample, v, cm/sec Figure 5. - Fluorine Auger signal plotted against velocity of sample. Electron beam current, 10 microamperes. S(v) - A Experimental data — Theoretical curve based on o = 5xlO"18 cm2 S(0)= 100 rn - 0.05 cm 0 .005 .01 .015 .04 .08 .12 .16 .18 .22 Velocity of sample, v, cm/sec Figure 5. - Fluorine Auger signal plotted against velocity of sample. Electron beam current, 10 microamperes. a maximum and leveling off, the fluorine signal is still increasing for the largest veloc- ity used. Thus the results in figure 5 indicate that the fluorine in the film exhibits a wide range of binding states as far as electron-induced desorption is concerned and this is probably a consequence of the multilayer structure of this mechanically applied film. ILLUSTRATION The fluorine in the interior of the film adjacent to the substrate has a high probability of reforming the bond broken by the incident electrons and is thus subject to desorption at a very low rate. This fluorine may be detected by Auger spectrometers that have the poorest time response such as the retarding potential analyzer. The bulk of the fluorine in the film has approximately the same cross section for electron-induced desorption since the velocity dependence is generally in accord with that predicted on the basis of a single desorption cross section (fig. 2(b)). The desorption time for this fluorine is about 40 seconds and may be detected on a stationary surface with a cylindrical mirror analyzer, but not with a retarding potential analyzer with any accuracy. A choice of 18 a = 5x10 and S(O) = 100 in the theoretical expression for the velocity dependence provides a curve in fair accord with the data in figure 5. This cross section is in good agreement with that obtained from the time dependence of the Auger signal from a sta- tionary surface (ref. 11). This is also a physically reasonable value for atoms in such a multilayer structure since it is larger than that for chemisorbed species on metals 18 2 (a < 10" cm ) and smaller than the ionization cross section for free atoms (~10~16cm2) (ref. 12). 18 ( ) ( ) 18 Since the theoretical curve based on a = 5xlO~ square centimeters is essentially flat for v > 0. 04 centimeter per second, the fact that the signal is still increasing for these velocities implies that there is some fluorine in the film with desorption cross 18 section much greater than 5x10 square centimeters. This fluorine is probably at the PTFE-vacuum interface and has a cross section approaching the ionization cross sec- tion for atoms in the free state. Detection of such species in the usual way on a station- ary surface is a difficult, if not impossible, task especially if the species is a minor constituent of the surface. In contrast, the technique presented here relaxes the strin- gent requirements on the time response of the spectrometer system and makes the de- tection of species with such high desorption cross sections a rather straightforward task. CONCLUDING REMARKS It has been shown here that it is possible to observe the Auger spectra of electron- desorbable surface species on a time-independent basis by moving the surface at a con- stant velocity with respect to the electron beam. A uniform coverage of the desorbable species in the region probed by the electron beam is required and the technique is time- independent to the extent that the Auger spectrum can be obtained before this region of the sample is traversed. The technique presented here will be most useful in those situations for which the rate of electron-induced desorption is greatest. High de- sorption rates are expected for those species in least electronic contact with the sub- strate and this is the case for fluorine in a PTFE film. In addition, the fluorine has only a single valence bond to the polymer chain. High electron-induced desorption rates are thus expected for polymeric lubricant films. In this connection, it has re- cently been pointed out that the weakly bound surface species most active in catalysis are also sensitive to electron-induced desorption (ref. 9). The technique presented here should thus find application in the investigation of both polymeric lubricant films and catalytic processes by Auger electron spectroscopy. The experimental arrangement used here (fig. 3) moves the sample with respect to the stationary beam, but there is also the possiblity of moving the electron beam over the stationary sample. However, the amount of sample that can be traversed with such an arrangement is limited, due to the necessity of having the source of the Auger elec- trons quite close to the axis of the electron spectrometer. Thus most effective use of the technique presented here requies a motor-driven sample holder and the use of somewhat extended samples. Finally we point out that the technique presented here for dealing with electron- induced desorption is a rather general way of alleviating the effects of any undesirable phenomena associated with the electron beam. The charging of insulators (especially under normal incidence), heating of the sample, and electron-induced adsorption (ref. 13) all occur within the beam spot. Moving fresh sample into the beam tends to reduce the degree to which these processes occur and thus provide a more desirable surface for Auger analysis. Lewis Research Center, National Aeronautics and Space Administration, Cleveland, Ohio, February 2, 1973, 502-01. 10 APPENDIX - SYMBOLS I electron beam current, electrons/sec -2 N surface number density of desorbable species, cm rQ electron beam radius, cm S Auger signal, arbitrary units S(O) Auger signal in the absence of electron-induced desorption t time that the sample has spent under electron beam, sec V beam voltage, V v velocity of sample, cm/sec £ 2la/7rrQV, dimensionless a cross section for electron-induced desorption (p function that describes the attenuation of Auger signal due to electron-induced desorption from moving sample, dimensionless 11 11 REFERENCES 1. Weber, R. E.; and Peria, W. T.: Use of LEED Apparatus for the Detection and Identification of Surface Containments. J. Appl. Phys., vol. 38, no. 11, Oct. 1967, pp. 4355-4358. 2. Harris, L. A.: Analysis of Materials by Electron-Excited Auger Electrons. J. Appl. Phys., vol. 39, no. 3, Feb. 15, 1968, pp. 1419-1427. 3. Harris, R. A.: Some Observations of Surface Segregation by Auger Electron Emis- sion. J. Appl. Phys., vol. 39, no. 3, Feb. 15, 1968, pp. 1428-1431. 4. Chang, Chuan C.: Auger Electron Spectroscopy. Surface Sci., vol. 25, no. 1, Mar. 1971, pp. 53-79. 5. Florio, J. V.; and Robertson, W. D.: Chlorine Reactions on the Si(m) Surface. Surface Sci., vol. 18, 1969, pp. 398-427. 6. Tracy, J. C.; and Palm berg, P. W.: Structural Influences on Adsorbate Binding Energy. I. Carbon Monoxide on (100) Palladium. J. Chem. Phys., vol. 51, no. 11, Dec. 1, 1969, pp. 4852-4862. 7. Palmberg, P. W.; Bonn, G. K.; and Tracy, J. C.: High Sensitivity Auger Elec- tron Spectrometer. Appl. Phys. Letters, vol. 15, no. 8, Oct. 15, 1969, pp. 254-255. 8. Taylor, N. J.: Resolution and Sensitivity Considerations of an Auger Electron Spectrometer Based on Display LEED Optics. Rev. Sci. Instr., vol. 40, no. 6, June 1969, pp. 792-804. 9. Madey, Theodore E.; and Yates, John T., Jr.: Electron-Stimulated Desorption as a Tool for Studies of Chemisorption: A Review. J. Vac. Sci. Tech., vol. 8, no. 4, 1971, pp. 525-555. 10. Pooley, Christine M.; and Tabor, D.: Friction and Molecular Structure: The Behaviour of Some Thermoplastics. Proc. Roy. Soc. (London) Ser. A, vol. 329, no. 1578, Aug. 22, 1972, pp. 251-274. 11. Pepper, Stephen V.; and Buckley, Donald H.: Adhesion and Transfer of Polytetra- fluoroethylene to Metals Studied by Auger Emission Spectroscopy. NASA TN D-6983, 1972. 12. Kieffer, L. J.; and Dunn, Gordon H.: Electron Impact lonization Cross-Section Data for Atoms, Atomic Ions, and Diatomic Molecules: I. APPENDIX - SYMBOLS Experimental Data. Rev. Mod. Phys., vol. 38, no. 1, Jan. 1966, pp. 1-35. 13. Coad, J. P.; Bishop, H. E.; and Riviere, J. C.: E lectr on-Beam Assisted Ad- sorption on the Si (m) Surface. Surface Sci., vol. 21, 1970, pp. 253-264. NASA-Langley, 1973 6 E-7206 12 E-7206 1. Report No. NASA TN D-7257 2. Government Accession No. 3. Recipient's Catalog No. 4. Title and Subtitle A NEW TECHNIQUE FOR AUGER ANALYSIS OF SURFACE SPECIES SUBJECT TO ELECTRON-INDUCED DESORPTION 5. Report Date April 1973 6. Performing Organization Code 7. Author(s) Stephen V. Pepper 8. Performing Organization Report No. E-7206 9. Performing Organization Name and Address Lewis Research Center National Aeronautics and Space Administration Cleveland, Ohio 44135 10. Work Unit No. 502-01 11. Contract or Grant No. 12. Sponsoring Agency Name and Address National Aeronautics and Space Administration Washington, D. C. 20546 13. Type of Report and Period Covered Technical Note 14. Sponsoring Agency Code 15. Supplementary Notes 16. Abstract A method is presented to observe surface species subject to electron-induced desorption by Auger electron spectroscopy. The surface to be examined is moved under the electron beam at constant velocity, establishing a time-independent condition and eliminating the time response of the electron spectrometer as a limiting factor. The dependence of the Auger signal on the surface velocity, incident electron current, beam diameter, and desorption cross section are analyzed. The method is illustrated by the Auger analysis of PTFE, in which the fluorine is removed by electron-induced desorption. 1. Report No. NASA TN D-7257 2. Government Accession No. 3. Recipient's Catalog No. 4. Title and Subtitle A NEW TECHNIQUE FOR AUGER ANALYSIS OF SURFACE SPECIES SUBJECT TO ELECTRON-INDUCED DESORPTION 5. Report Date April 1973 6. Performing Organization Code 7. Author(s) Stephen V. Pepper 8. Performing Organization Report No. E-7206 9. Performing Organization Name and Address Lewis Research Center National Aeronautics and Space Administration Cleveland, Ohio 44135 10. Work Unit No. 502-01 11. Contract or Grant No. 12. Sponsoring Agency Name and Address National Aeronautics and Space Administration Washington, D. C. 20546 13. Type of Report and Period Covered Technical Note 14. Sponsoring Agency Code 15. Supplementary Notes 16. Abstract A method is presented to observe surface species subject to electron-induced desorption by Auger electron spectroscopy. APPENDIX - SYMBOLS The surface to be examined is moved under the electron beam at constant velocity, establishing a time-independent condition and eliminating the time response of the electron spectrometer as a limiting factor. The dependence of the Auger signal on the surface velocity, incident electron current, beam diameter, and desorption cross section are analyzed. The method is illustrated by the Auger analysis of PTFE, in which the fluorine is removed by electron-induced desorption. 7. Author(s) Stephen V. Pepper 8. Performing Organization Report No. E-7206 9. Performing Organization Name and Address Lewis Research Center National Aeronautics and Space Administration Cleveland, Ohio 44135 10. Work Unit No. 502-01 11. Contract or Grant No. 12. Sponsoring Agency Name and Address National Aeronautics and Space Administration Washington, D. C. 20546 13. Type of Report and Period Covered Technical Note 14. Sponsoring Agency Code 15. Supplementary Notes 16. Abstract A method is presented to observe surface species subject to electron-induced desorption by Auger electron spectroscopy. The surface to be examined is moved under the electron beam at constant velocity, establishing a time-independent condition and eliminating the time response of the electron spectrometer as a limiting factor. The dependence of the Auger signal on the surface velocity, incident electron current, beam diameter, and desorption cross section are analyzed. The method is illustrated by the Auger analysis of PTFE, in which the fluorine is removed by electron-induced desorption. 17. Key Words (Suggested by Author(s)) Auger electron spectroscopy Electron-induced desorption Surface analysis Polymers 18. Distribution Statement Unclassified - unlimited 19. Security Classif. (of this report) Unclassified 20. Security Classif. (of this page) Unclassified 21. No. of Pages IS 22. Price* $3.00 * For sale by the National Technical Information Service, Springfield, Virginia 22151 A method is presented to observe surface species subject to electron-induced desorption by Auger electron spectroscopy. The surface to be examined is moved under the electron beam at constant velocity, establishing a time-independent condition and eliminating the time response of the electron spectrometer as a limiting factor. The dependence of the Auger signal on the surface velocity, incident electron current, beam diameter, and desorption cross section are analyzed. The method is illustrated by the Auger analysis of PTFE, in which the fluorine is removed by electron-induced desorption. A method is presented to observe surface species subject to electron-induced desorption by Auger electron spectroscopy. NATIONAL AERONAUTICS AND SPACE ADMINISTRATION WASHINGTON. D.C. 2O546 OFFICIAL BUSINESS PENALTY FOR PRIVATE USE $3OO SPECIAL FOURTH-CLASS RATE BOOK POSTAGE AND FEES PAID NATIONAL AERONAUTICS AND SPACE ADMINISTRATION 451 POSTMASTER : If Undeliverable (Section 158 Postal Manual) Do Not Return POSTAGE AND FEES PAID NATIONAL AERONAUTICS AND SPACE ADMINISTRATION 451 POSTMASTER : If Undeliverable (Section 158 Postal Manual) Do Not Return "The aeronautical and space activities of the United States shall be conducted so as to contribute . . . to the expansion of human knowl- edge of phenomena in the atmosphere and space. The Administration shall provide for the widest practicable and appropriate dissemination of information concerning its activities and the results thereof." g —NATIONAL AERONAUTICS AND SPACE ACT OF 1958 APPENDIX - SYMBOLS The surface to be examined is moved under the electron beam at constant velocity, establishing a time-independent condition and eliminating the time response of the electron spectrometer as a limiting factor. The dependence of the Auger signal on the surface velocity, incident electron current, beam diameter, and desorption cross section are analyzed. The method is illustrated by the Auger analysis of PTFE, in which the fluorine is removed by electron-induced desorption. . Distribution Statement Unclassified - unlimited * For sale by the National Technical Information Service, Springfield, Virginia 22151 NASA SCIENTIFIC AND TECHNICAL PUBLICATIONS TECHNICAL REPORTS: Scientific and technical information considered important, complete, and a lasting contribution to existing knowledge. TECHNICAL TRANSLATIONS: Information published in a foreign language considered to merit NASA distribution in English. TECHNICAL TRANSLATIONS: Information published in a foreign language considered to merit NASA distribution in English. SPECIAL PUBLICATIONS: Information derived from or of value to NASA activities. Publications include final reports of major . projects, monographs, data compilations, ; handbooks, sourcebooks, and special bibliographies. TECHNICAL NOTES: Information less broad in scope but nevertheless of importance as a contribution to existing knowledge. TECHNICAL NOTES: Information less broad in scope but nevertheless of importance as a contribution to existing knowledge. TECHNICAL MEMORANDUMS: Information receiving limited distribution because of preliminary data, security classifica- tion, or other reasons. Also includes conference proceedings with either limited or unlimited distribution. TECHNOLOGY UTILIZATION PUBLICATIONS: Information on technology used by NASA that may be of particular interest in commercial and other non-aerospace applications. Publications include Tech Briefs, Technology Utilization Reports and Technology Surveys. CONTRACTOR REPORTS: Scientific and technical information generated under a NASA contract or grant and considered an important contribution to existing knowledge. POSTMASTER : If Undeliverable (Section 158 Postal Manual) Do Not Return Details on the availability of these publications may be obtained from: SCIENTIFIC AND TECHNICAL INFORMATION OFFICE N A T I O N A L A E R O N A U T I C S A N D S P A C E A D M I N I S T R A T I O N Washington, D.C. 20546 Details on the availability of these publications may be obtained from: SCIENTIFIC AND TECHNICAL INFORMATION OFFICE N A T I O N A L A E R O N A U T I C S A N D S P A C E A D M I N I S T R A T I O N Washington, D.C. 20546
https://openalex.org/W4362536580	https://figshare.com/articles/journal_contribution/Figure_S3_from_G-Protein-Coupled_Estrogen_Receptor_1_Promotes_Gender_Disparities_in_Hepatocellular_Carcinoma_via_Modulation_of_SIN1_and_mTOR_Complex_2_Activity/22514784/1/files/39977283.pdf	unk	null	Figure S2 from G-Protein-Coupled Estrogen Receptor 1 Promotes Gender Disparities in Hepatocellular Carcinoma via Modulation of SIN1 and mTOR Complex 2 Activity	null	2,023	cc-by	18	Figure S3 Figure S3 DMSO+si-Scramble DMSO+si-GPER1 E2+si-Scramble E2+si-GPER1 Figure S3 DMSO+si-GPER1 DMSO+si-Scramble g DMSO+si-GPER1 DMSO+si-Scramble E2+si-GPER1 E2+si-Scramble E2+si-Scramble
https://openalex.org/W2914932343	https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0211139&type=printable	English	null	Use of shotgun metagenomics for the identification of protozoa in the gut microbiota of healthy individuals from worldwide populations with various industrialization levels	PloS one	2,019	cc-by	12,104	RESEARCH ARTICLE Use of shotgun metagenomics for the identification of protozoa in the gut microbiota of healthy individuals from worldwide populations with various industrialization levels Ana LokmerID1, Amandine Cian2, Alain Froment1, Nausicaa Gantois2, Eric Viscogliosi2, Magali Chabe´2☯, Laure Se´gurel1☯* 1 UMR7206 Eco-anthropologie et Ethnobiologie, CNRS—MNHN—Univ Paris Diderot—Sorbonne Paris Cite´, Paris, France, 2 Univ. Lille, CNRS, Inserm, CHU Lille, Institut Pasteur de Lille, U1019 –UMR 8204 –CIIL– Centre d’Infection et d’Immunite´ de Lille, Lille, France 1 UMR7206 Eco-anthropologie et Ethnobiologie, CNRS—MNHN—Univ Paris Diderot—Sorbonne Paris Cite´, Paris, France, 2 Univ. Lille, CNRS, Inserm, CHU Lille, Institut Pasteur de Lille, U1019 –UMR 8204 –CIIL– Centre d’Infection et d’Immunite´ de Lille, Lille, France a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 ☯These authors contributed equally to this work. * laure.segurel@mnhn.fr OPEN ACCESS Protozoa have long been considered undesirable residents of the human gut, but recent findings suggest that some of them may positively affect the gut ecosystem. To better understand the role and ecological dynamics of these commensal and potentially beneficial protozoan symbionts, we need efficient methods to detect them, as well as accurate esti- mates of their prevalence across human populations. Metagenomics provides such an opportunity, allowing simultaneous detection of multiple symbionts in a single analytical pro- cedure. In this study, we collected fecal samples of 68 individuals from three Cameroonian populations with different subsistence modes and compared metagenomics-based and tar- geted methods of detection for two common protozoan genera: Blastocystis and Ent- amoeba. In addition, we analyzed our data along with publicly available fecal metagenomes from various worldwide populations to explore the prevalence and association patterns of ten protozoan genera. Regarding the detection method, microscopy was much less sensi- tive than metagenomics for Entamoeba, whereas qPCR was at least as sensitive as meta- genomics for Blastocystis sp. However, metagenomics was more likely to detect co- colonizations by multiple subtypes. Out of the ten examined genera in 127 individuals from Cameroon, Tanzania, Peru, Italy or USA, only three (Blastocystis, Entamoeba and Entero- monas) had an overall prevalence exceeding 10%. All three genera were more common in less industrialized populations and their prevalence differed between continents and subsis- tence modes, albeit not in a straightforward manner. The majority (72.5%) of colonized indi- viduals carried at least two protozoan species, indicating that mixed-species colonizations are common. In addition, we detected only positive and no negative association patterns between different protozoa. Despite the pitfalls of the metagenomic approach, ranging from the availability of good-quality sequencing data to the lack of standard analytical procedures, we demonstrated its utility in simultaneous detection of multiple protozoan genera, and Citation: Lokmer A, Cian A, Froment A, Gantois N, Viscogliosi E, Chabe´ M, et al. (2019) Use of shotgun metagenomics for the identification of protozoa in the gut microbiota of healthy individuals from worldwide populations with various industrialization levels. PLoS ONE 14(2): e0211139. https://doi.org/10.1371/journal. pone.0211139 Editor: Tiffany L. Weir, Colorado State University, UNITED STATES Introduction Gut protozoa have long been exclusively a topic of parasitological research, although they may actually be more often commensal than pathogenic [1, 2]. Recent findings even suggest that some protozoa, such as Blastocystis sp. and non-pathogenic Entamoeba spp., could benefit their hosts, as they are associated with an increased gut microbiome diversity and a higher fre- quency of potentially beneficial bacterial taxa [2–6]. It has thus been proposed that the virtual eradication of gut protozoa in industrialized countries might negatively influence human health due to the disruption of ecological interactions in the gut ecosystem [2, 6]. In addition, there is a number of other protozoan species that may affect the gut ecosystem, but whose pathogenicity and prevalence are even less known [2]. In order to study the ecological and clin- ical impact of gut protozoa, we first need reliable estimates of their prevalence. g To date, the prevalence of gut protozoa is often underestimated due to the lack of adequate detection and surveillance systems [7]. Furthermore, these organisms are often studied sepa- rately, impeding the understanding of their ecological dynamics. Finally, traditional diagnostic procedures based on microbiological methods including cultivation, microscopy and antigen- based tests, have significant drawbacks such as a high limit of detection and are available only for a limited number of pathogens. However, a recent shift towards molecular procedures has increased diagnostic sensitivity and specificity, and improved quantification [7]. Notably, it is now recognized that non-pathogenic Entamoeba spp. have a non-negligible prevalence in many African populations (0–40%, [8, 9]). Moreover, genotyping of Entamoeba spp. revealed that, in Africa, E. dispar is far more common than E. histolytica [10]. Differentiating between the pathogenic E. histolytica and morphologically undistinguishable non-pathogenic Ent- amoeba species, E. dispar and E. moshkovskii [10], is of great importance for public health, as E. histolytica is among the major causative agents of diarrheal disease, especially in developing regions such as sub-Saharan Africa [11]. However, reliable data on the worldwide distribution of Entamoeba spp. remain scarce. Similarly, molecular methods unveiled high genetic diversity of Blastocystis sp., the most common single-celled eukaryote in fecal samples of healthy humans [12–19]: so far, 10 genetic subtypes (ST1–9 and ST12) have been found in humans, with ST3 being the most common [16]. Apart from Entamoeba and Blastocystis, other presum- ably non-pathogenic intestinal protozoa have been studied only exceptionally by molecular methods [1]. Gut protozoa identification and prevalence determination using metagenomics especially its ability to efficiently detect mixed-species colonizations. Our study corroborates and expands prevalence results previously obtained for Blastocystis sp. and provides novel data for Entamoeba spp. and several other protozoan genera. Furthermore, it indicates that multiple protozoa are common residents of the healthy human gut worldwide. Funding: This work was supported by the ANR MICROREGAL (ANR-15-CE02-0003), the Centre National de la Recherche Scientifique (LS) and the Institut Pasteur of Lille. AC was supported by a PhD fellowship from the University of Lille 2 and the Institut Pasteur of Lille. The funding bodies had no role in the study design, data collection and analysis, decision to publish or preparation of the manuscript. Competing interests: The authors have declared that no competing interests exist. Editor: Tiffany L. Weir, Colorado State University, UNITED STATES Copyright: © 2019 Lokmer et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: SSU rRNA gene sequences obtained in this study were deposited in GenBank and are available under accession numbers MG907124-MG907188. Raw shotgun data have been deposited in the European Nucleotide Archive (ENA) with the BioProject ID PRJEB27005, under the accession numbers ERS2539904-ERS2539960. 1 / 20 PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019 PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019 Ethics statement The research protocol was approved by the CPP (Comite´ de Protection des Personnes) French ethical committee (N˚2010-avril-12276) and samples were authorized to be collected and con- served by the French Ministry of Higher Education and Research (N˚DC 2009–1068). Research permits were obtained by the “Institut de Recherche pour le De´veloppement “(IRD) in agreement with the “Ministère de la Recherche Scientifique et de l’Innovation” (MINRESI) of Cameroon. In addition, an ethical clearance was obtained from the Cameroonese National Committee (CNERSH, Central National d’Ethique de la Recherche pour la Sante´ Humaine, Approval N 2017/05/900) and a regional research permit was obtained from the Health Minis- ter (Centre region, Approval N˚0061). The CPP (Comite´ de Protection des Personnes) clear- ance was required the Cameroon national ethical committee for the application submission. Written consent was obtained from all participants. The research protocol was approved by the CPP (Comite´ de Protection des Personnes) French ethical committee (N˚2010-avril-12276) and samples were authorized to be collected and con- served by the French Ministry of Higher Education and Research (N˚DC 2009–1068). Gut protozoa identification and prevalence determination using metagenomics extremely low prevalence of other gut protozoa. To our knowledge, no study has evaluated the potential of MG in the detection of gut protozoa (pathogenic or non-pathogenic) other than Blastocystis sp so far. In our study, we aimed to address these gaps in the field in the following way: i) by assessing the potential of MG for diagnosis of protozoan (co-)colonization; ii) by estimating the preva- lence of various gut protozoa in healthy individuals and testing whether they are common resi- dents of a healthy human gut; iii) by exploring the influence of industrialization and sanitary conditions on the diversity and prevalence of gut protozoa; and iv) by examining association patterns (co-occurrence and co-exclusion) among the studied protozoa in order to make pre- dictions about the ecological factors shaping the gut protozoan community. In order to do so, we first compared the MG-based approach with targeted diagnostic methods (microscopy for Entamoeba spp. detection and qPCR for Blastocystis sp. STs detec- tion) in 68 Cameroonian adults. We then analyzed the metagenomes generated from these Cameroonian samples together with publicly available metagenomes from Tanzanian, Peru- vian, USA and Italian individuals [25, 26], summing up to 127 worldwide fecal metagenomes of healthy individuals. We systematically looked for multiple pathogenic and non-pathogenic protozoa: Blastocystis sp., Chilomastix mesnili, Cryptosporidium spp., Cystoisospora belli, Dien- tamoeba fragilis, Endolimax nana, Entamoeba spp., Enteromonas hominis, Giardia intestinalis and Iodamoeba bütschlii. We took advantage of this combined dataset to assess the prevalence of gut protozoa in populations with varying levels of industrialization (rural populations with traditional modes of subsistence, further referred to as non-industrialized populations, and urban industrialized populations). We also examined the effect of subsistence mode (hunter- gatherer, farmer or fisher) on gut protozoa prevalence in non-industrialized populations. Although we do not believe that diet is a major determinant of gut eukaryotes diversity, we hypothesize that there are likely systematic sanitary differences between these subsistence modes that result in varying environmental exposure to gut eukaryotes. Finally, as we simulta- neously determined the presence of multiple non-pathogenic and pathogenic gut protozoa in the studied individuals, we were able to investigate (to our knowledge for the first time) associ- ation patterns between them. Introduction Recent accumulation of shotgun metagenomic data from human fecal samples provides an unprecedented opportunity to simultaneously obtain detailed taxonomical and genetic infor- mation for a broad range of gut protozoan species, as metagenomics (MG) allows joint identi- fication and genomic characterization of multiple species without the need for species-specific procedures [20, 21]. Another advantage of this method is the ability to identify mixed-species colonizations. Indeed, MG bypasses one of the disadvantages of targeted PCR or qPCR tech- niques—competitive amplification of differentially abundant strains in a sample. To date, the few studies that used metagenomics to detect gut protozoa investigated only Blastocystis sp. [22–24], likely because they focused on highly industrialized populations with PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019 2 / 20 Metagenomic data Metagenomic data were obtained for 57 of the 68 sampled individuals from Cameroon. The DNA Libraries were prepared using Nextera XT DNA Library Prep Kit (Illumina, San Diego, CA, USA) and sequenced as 100 bp paired-end reads on a HiSeq2000 sequencer (Illumina) at the University of Minnesota Genomics Center (Minneapolis, MN, USA). Raw data have been deposited in the European Nucleotide Archive (ENA) with the BioProject ID PRJEB27005, under the accession numbers ERS2539904-ERS2539960. We obtained an average of 25.6 mil- lion read pairs per sample. The reads from low-quality areas of the flowcell were filtered out using FilterByTile tool from BBMap package [27]. We further explored metagenomic datasets focusing on non-industrialized populations published by Obregon-Tito et al., Rampelli et al. and Yatsunenko et al. [25, 26, 28], which we downloaded from SRA NCBI and MG-RAST in June 2017. However, we did not pursue with the Yatsunenko dataset (rural individuals from Venezuela and Malawi, and urban individuals from USA) due to their low sequencing depth (average of 155,890 350bp reads per sample) [28]. The Rampelli dataset (study accession: PRJNA278393; average of 10.9 million 100bp read pairs per sample) [26] consists of 27 healthy rural Hadza of Tanzania and 11 healthy urban Italians. The Obregon-Tito dataset (study accession: PRJNA268964, a mix of 75bp, 100 bp and 150 bp paired-end reads with an average of 22.3 million read pairs per sample) [25] consists of 12 rural Tunapuco and 24 rural Matses healthy individuals from Peru and 22 urban healthy subjects from Norman, USA. As all Cameroonian subjects were adults, we removed the indi- viduals younger than 18 years of age from the two downloaded datasets to increase compara- bility, resulting in 22 Hadza, 11 Italians, 8 Tunapuco, 10 Matses, and 19 adults from Norman. Details about the samples used, together with the number of reads per sample, can be found in S1 Table. For all datasets, we used the bbduk tool [27] to trim the adapters, very low-quality bases (Phred Q<3) and to remove reads shorter than 36 bp. The quality of the metagenomic data was assessed by FastQC [29]. Gut protozoa identification and prevalence determination using metagenomics Informed consent for this research project was obtained from all volunteers sampled in this study. The 16S-rRNA-gene based characterization of these samples including the contextual data has been previously published [3]. PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019 Sample collection Fecal samples were collected from 68 healthy individuals (23 hunter-gatherers, 24 farmers and 21 individuals from a fishing population; 40 males and 28 females) aged 26 to 78 years (median of 50 years) and living in seven different villages situated in Southwest Cameroon (see [3]). 3 / 20 PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019 Gut protozoa identification and prevalence determination using metagenomics house script to sequentially filter out the alignments with more than 0.1 of the read length soft- clipped, with edit distance exceeding 0.01 of the read length and finally with the alignment length shorter than 75 bp. As genomes from public databases can contain vectors and sequenc- ing artifacts, we performed an additional step to remove the alignments to such regions. In detail, we extracted the sequences of the hit genomic regions with BEDTools [32] and excluded those that mapped with high confidence (utilizing the similar procedure as above with adjusted cut-offs: 0.02 of the read length for edit distance, and alignment length > 35 bp) to any of 120 bacterial reference genomes [33], UniVec database [34] or to the "sequencingartifacts" file pro- vided by BBmap [27]. To further decrease the rate of false positives due to similarity between the genomes, we removed the reads that mapped to multiple genomes from further analysis. However, this step had no effect on the results (results not shown). For the species with SSU rRNA gene sequences, we performed a similar procedure with the following adjustments: we did not require both reads to map and we applied a 0.01 read length cutoff for soft-clipping (instead of 0.1). We additionally removed all read pairs where each mate mapped to different species to reduce the false positive rate. As no reads mapped uniquely to E. histolytica, we excluded it from final analysis in order to correctly estimate the abundance of E. dispar (which has a highly similar SSU rRNA sequence). In case of SSU rRNA gene sequences, we assumed that any hit passing the above filtering criteria could be considered reliable and we therefore used the presence of hits as our diagnos- tic. For whole genomes, we additionally inspected the distribution of hits across the scaffolds and calculated the breadth of coverage. Namely, shotgun reads may map to contaminating sequences (such as vectors or sequences originating from other organisms) that publicly avail- able genomes often contain, which may result in a high number of hits concentrated in few regions and very low breadth of coverage. This is indeed what we observed in few cases, pri- marily for Blastocystis sp. ST6 (S5 Fig). PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019 Detection of various gut protozoa using metagenomic data We searched for multiple gut protozoa species, i.e., Blastocystis sp., Chilomastix mesnili, Cryp- tosporidium hominis, Cryptosporidium parvum, Cystoisospora belli, Dientamoeba fragilis, Endolimax nana, Entamoeba spp., Enteromonas hominis, Giardia intestinalis and Iodamoeba bütschlii, by aligning quality-controlled data to publicly available data using BWA-MEM [30]. The SSU rRNA gene sequences of Retortamonas and Pentatrichomonas available in public databases were either very short or not of human origin and therefore they were not included in the analysis. To detect Blastocystis sp., Giardia intestinalis, Cryptosporidium hominis and C. parvum, we used whole-genome data, while, for the other species (including all Entamoeba spp.), we used complete SSU rRNA gene sequences (S2 Table). For E. histolytica and E. dispar (for which both genomes and complete SSU sequences are available), we decided to use the SSU rRNA gene sequences, in order to obtain comparable data for a larger number of Ent- amoeba species. In order to retain only high-confidence alignments for the species with available genomes, we applied the following filtering procedure. We used SAMTools [31] to remove the reads with unmapped mates (-F8) and supplementary alignments (-F2048). We further used an in- PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019 4 / 20 qPCR and molecular subtyping of Blastocystis sp. in Cameroon For 68 samples from Cameroonian subjects, 1 μl of extracted DNA was subjected to a qPCR assay using the Blastocystis sp.-specific primers BL18SPPF1 (5’-AGTAGTCATACGCTCGT CTCAAA-3’) and BL18SR2PP (5’-TCTTCGTTACCCGTTACTGC-3’) targeting the SSU rRNA gene as previously described [35]. The positive qPCR products were purified and both strands were directly sequenced at Genoscreen, Lille, France. Direct sequencing of several qPCR products generated mixed signals that could reflect colonization by different STs. These samples were thus re-analyzed by non-qPCR using the same primer pair as for qPCR, followed by cloning. In detail, end-point PCR amplifications were performed in 50 μl according to stan- dard conditions for Platinum Taq High-Fidelity DNA polymerase (Invitrogen, Groningen, the Netherlands). After denaturation at 94˚C for 5 min, 40 cycles of amplification were performed with a Bioer LifeECO apparatus (Binjiang District, China) as follows: 30 s at 94˚C, 35 s at 60˚C, and 50 s at 68˚C. The final extension was continued for 2 min. End-point PCR products were separated by agarose gel electrophoresis and bands of the expected size (approximately 320 bp) were purified using the Wizard SV Gel and PCR clean-up system (Promega, Madison, WI, USA). Purified PCR products were cloned in the T-vector, pCR 2.1-TOPO (Invitrogen) and amplified in Escherichia coli One Shot TOP10 competent cells. Minipreparations of plas- mid DNA were done using the NucleoSpin Plasmid kit (Macherey-Nagel, Du¨ren, Germany). Five positive clones containing inserts of approximately the expected size were arbitrarily selected for each sample and sequenced in both directions. The sequences obtained were com- pared with all Blastocystis sp. homologous sequences available from the National Centre for Biotechnology Information (NCBI) using the nucleotide BLAST program. The STs were PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019 5 / 20 Gut protozoa identification and prevalence determination using metagenomics identified by finding the exact or closest match against all known mammalian and avian Blas- tocystis sp. STs according to the last classification by Alfellani et al. [36]. All sequences shared 99–100% identity with the publicly available sequences, allowing the direct subtyping of the corresponding isolates. The sequences were deposited in GenBank and are available under accession numbers MG907124-MG907188. Statistical tests All statistical analyses were conducted in R [37]. We compared different methods of detection with McNemar’s test and we calculated a corresponding odds ratio with 95% confidence inter- vals to estimate the effect size if applicable. To examine the differences between populations, we performed Fisher’s exact tests, first grouping non-industrialized and industrialized popula- tions together, and then examining differences among non-industrialized populations only. If the latter was significant, we fitted a logistic regression model to examine if these differences were due to the country of origin and/or to subsistence mode. The particular contrasts we were interested in were between: Africa and South America, Eastern and Western Africa (i.e. Tanzania and Cameroon), hunter-gatherers and farmers/fishers, and between farmers and fishers. As logistic regressions do not allow for reliable estimation of coefficients and their con- fidence intervals if prevalence is 0%, we excluded the contrasts where all populations of one group had 0% prevalence. We tested if Entamoeba spp. and Blastocystis sp. preferentially infect the same individuals with a Chi-square test. We additionally calculated a Pearson’s phi coefficient to provide an esti- mate of effect size where appropriate. We further looked for significant association patterns between each pair of protozoan species/STs by conducting co-occurrence analysis imple- mented in the R-package cooccur [38]. Observed and expected frequencies of co-occurrence between each pair of species are calculated with the expected frequencies assuming random and independent distribution of each ST/species. The analysis returns the probabilities that a more extreme (either low or high) value of co-occurrence could have been obtained by chance. Coinfection plots were created with the R-package UpSetR [39]. Gut protozoa identification and prevalence determination using metagenomics thresholds. In addition, the number of positive individuals was fully concordant between these thresholds for ST1 and did not significantly differ for ST2 and ST3 (McNemar’s test: ST2: chi- square (1) = 2.25, p = 0.134; ST3: chi-square (1) = 0.5, p = 0.480). We decided to be stringent and use 10% positive contigs as our diagnostic threshold in the rest of the work. For the analysis of SSU rRNA gene sequences, we used the presence of hits (after the align- ment filtering step, see details in Methods) as a diagnostic measure. We did not detect Ent- amoeba histolytica, Entamoeba chattoni, Cryptosporidium hominis, C. parvum or Blastocystis ST6, ST7 and ST9 in any of the 127 samples. All other species had at least one positive individ- ual (S3 Table). Although 71.7% of individuals carried at least one protozoan, the only genera with an overall prevalence exceeding 10% were Blastocystis (60.6%), Entamoeba (45.7%) and Enteromonas (11.0%), and these were thus the only ones analyzed in detail. Comparison of detection methods for Blastocystis sp. and Entamoeba spp. in Cameroon To compare the MG-based approach with targeted diagnostic methods, the Cameroonian samples were analyzed by qPCR (Blastocystis sp.) and microscopy (Entamoeba spp.) in addi- tion to MG. The overall prevalence of Blastocystis sp. was 88.2% according to qPCR (N = 68) and 77.2% according to MG (N = 57). We obtained congruent results among the 57 samples analyzed by both methods (McNemar’s chi-square (1) = 2.286, p = 0.131), with 75.4% of indi- viduals identified as positive and 12.3% as negative by both of them (Fig 1). Regarding the remaining cases, 10.5% were positive only by qPCR and 1.8% only by MG. Therefore, although we found no statistical difference between qPCR and MG methods for detection of Blastocystis sp., qPCR seems a bit more sensitive. Regarding the distribution of Blastocystis sp. STs, both methods detected single colonization by ST3 as the most prevalent (42.6% and 31.6% by qPCR and MG, respectively), followed by ST1 (27.9% and 21.1%) and ST2 (10.3% and 5.3%) single colonizations (Fig 1). Mixed-subtype colonizations were found in 7.4% (qPCR) and 19.3% (MG) of individuals, indicating that MG is more likely to detect these. The combination of ST1/ST3 accounted for the majority of mixed colonizations (60% and 73% by qPCR and MG, respectively), followed by ST2/ST3 (20% and 18%) and ST1/ST2 (20% and 9%). The results were concordant between the methods for each of the three STs (Fig 1, McNemar’s test: ST1: chi-square (1) = 0.8, p = 0.371; ST2: chi- square (1) = 0, p = 1; ST3: chi-square (1) = 0, p = 1). For the prevalence of Entamoeba spp., microscopy yielded a considerably lower estimate (37.9%, N = 66) than MG (63.2%, N = 57). The inspection of the 55 samples analyzed by both methods revealed that no sample was positive by microscopy only (Fig 1) and confirmed a sig- nificantly higher sensitivity of MG (McNemar’s chi-square (1) = 13.067, p < 10−3, odds ratio (OR) with 95% confidence intervals (CI) = 0 [0, 0.34]). As microscopy does not allow for finer taxonomical resolution, we could not compare the distribution of individual Entamoeba spp. species between the methods. Detection of gut protozoa from metagenomic data In order to detect gut protozoan species belonging to ten different genera and estimate their prevalence in eight populations across four continents, we mapped 127 gut metagenomes (from 57 individuals living in Cameroon, 22 in Tanzania, 18 in Peru, 19 in USA and 11 in Italy) either to available protozoan genomes or to SSU rRNA gene sequences (listed in S2 Table). To keep the false positive rate as low as possible, we filtered the alignments in a step- wise manner, applying strict filtering criteria (see Methods section for details). Briefly, our diagnostic measure for genome-based detection relied on two criteria: we first selected poten- tial positives based on the percentage of positive contigs and subsequently excluded cases with breadth of coverage < 0.001. We chose the percentage of positive contigs over the breadth of coverage as a primary criterion because the latter may vary more widely depending on the size of contaminating genome regions (i.e. not originating from the protozoan itself) than the dis- tribution of hits across the contigs. Still, the number and size of contigs/scaffolds depends on the assembly quality, and the assemblies with very few contigs could easily result in false posi- tives without the second step. We observed a uniform-like hit distribution in the samples with at least 5% positive contigs. Applying a 5% or 10% detection threshold yielded highly concordant results for Blastocystis sp. (McNemar’s chi-square (1) = 1.33, p = 0.248, N = 127), with the same STs detected using both PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019 6 / 20 Prevalence of Blastocystis sp. in culturally diverse populations Using MG as the detection method, we found that at least one Blastocystis sp. ST was present in 77 out of 127 (60.6%) individuals in the eight studied populations (rural populations: hunter-gatherers, farmers and fishers in Cameroon, hunter-gatherers and farmers in Peru, hunter-gatherers in Tanzania; urban populations: USA and Italy), with the frequency per pop- ulation ranging from 9.1% in Italy to 95.5% in Cameroonian farmers (Fig 2, S3 Table). Regard- ing the distribution of Blastocystis sp. STs (S1 Fig), the three most common types of colonization were: a single colonization by ST3 (18.1% overall, found in all populations except 7 / 20 PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019 Gut protozoa identification and prevalence determination using metagenomics qPCR/Micro + MG MG qPCR/Micro Negative 0 20 40 60 80 100 % individuals B. sp. ST1 ST2 ST3 E. sp. Fig 1. Concordance between detection methods for Blastocystis sp. (overall and by subtype, N = 57) and Entamoeba spp. (N = 55) in Cameroon. Shown are the proportions of individuals positive for both methods (MG- based and targeted approach), only one of them, or none. Blastocystis sp. (B. sp.) was assessed by MG and qPCR, while Entamoeba spp. (E. sp.) was assessed by MG and microscopy (Micro). https://doi.org/10.1371/journal.pone.0211139.g001 qPCR/Micro + MG MG qPCR/Micro Negative 0 20 40 60 80 100 % individuals B. sp. ST1 ST2 ST3 E. sp. E. sp. E. sp. Fig 1. Concordance between detection methods for Blastocystis sp. (overall and by subtype, N = 57) and Entamoeba spp. (N = 55) in Cameroon. Shown are the proportions of individuals positive for both methods (MG- based and targeted approach), only one of them, or none. Blastocystis sp. (B. sp.) was assessed by MG and qPCR, while Entamoeba spp. (E. sp.) was assessed by MG and microscopy (Micro). https://doi.org/10.1371/journal.pone.0211139.g001 https://doi.org/10.1371/journal.pone.0211139.g001 Peruvian and Tanzanian hunter-gatherers), a single colonization by ST1 (15%, in all African and South-American populations), and a combination of these two STs (9.5%, in Tanzanian hunter-gatherers, Peruvian farmers, Cameroonian farmers and fishers). ST2 was found more often in mixed (11%) than in single colonization (5.5%). Whereas single ST2 colonization was found in African hunter-gatherers as well as in Cameroonian and Peruvian farmers, ST2/ST3 (5.5%) and ST1/ST2 (3.9%) combinations were restricted to African hunter-gatherers. More- over, mixed-subtype colonization by ST1/ST2/ST3 (1.6%) were detected only in Tanzanian population. ST4 and ST8 were found only in USA, each in one individual. PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019 Prevalence of Blastocystis sp. in culturally diverse populations Peruvian and Tanzanian hunter-gatherers), a single colonization by ST1 (15%, in all African and South-American populations), and a combination of these two STs (9.5%, in Tanzanian hunter-gatherers, Peruvian farmers, Cameroonian farmers and fishers). ST2 was found more often in mixed (11%) than in single colonization (5.5%). Whereas single ST2 colonization was found in African hunter-gatherers as well as in Cameroonian and Peruvian farmers, ST2/ST3 (5.5%) and ST1/ST2 (3.9%) combinations were restricted to African hunter-gatherers. More- over, mixed-subtype colonization by ST1/ST2/ST3 (1.6%) were detected only in Tanzanian population. ST4 and ST8 were found only in USA, each in one individual. Considering the influence of subsistence mode on the prevalence of Blastocystis sp. (Fig 2, S3 Table), we found a significantly higher prevalence in non-industrialized than in industrial- ized populations (Fisher’s exact test, p < 10−3), as well as significant differences between non- industrialized populations (Fisher’s exact test, p = 0.004). A logistic regression model including only non-industrialized populations revealed significant effects of both subsistence mode (i.e., farmer, fisher or hunter-gatherer) and country (Likelihood Ratio Test (LRT) for single term deletion, p = 0.013 and 0.001, S4 Table). Specifically, we detected a higher prevalence in Africa than in South America (p = 0.004, OR with 95% CI = 1.89 [1.24, 2.39]) and a higher prevalence in farmers compared with fishers (p = 0.001, OR = 4.28 [1.89, 7.5]). Regarding the Blastocystis sp. STs (Fig 2), ST1 was found only (and in all) non-industrial- ized countries (Fisher’s exact test comparing non-industrialized and industrialized popula- tions, p < 10−3), with prevalence ranging from 25% in Peruvian farmers to 50% in Tanzanian hunter-gatherers, but without significant differences between non-industrialized populations (Fisher’s exact test, p = 0.836). ST2 was also found only in non-industrialized countries (Fish- er’s exact test comparing non-industrialized and industrialized populations, p = 0.004), again with the highest prevalence in Tanzanian hunter-gatherers (59.1%). The prevalence of ST2 fur- ther significantly differed among non-industrialized populations (Fisher’s exact test, p < 10−3). Logistic regression revealed a significant effect of country (LRT = 8.65, p = 0.013), 8 / 20 PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019 Gut protozoa identification and prevalence determination using metagenomics HG_Cam Farm_Cam Fish_Cam HG_Tanz HG_Peru Farm_Peru Indstr_USA Indstr_It 20 40 60 80 100 ST1 ST2 ST3 ST4 ST8 Blastocystis sp. 0 81.3% 95.5% 52.6% 81.8% 40% 75% 21.1% 9.1% Prevalence (%) Fig 2. Prevalence of Blastocystis sp. Prevalence of Blastocystis sp. in culturally diverse populations and its subtypes across populations according to metagenomic analysis. Subtypes with zero prevalence (ST6, ST7 and ST9) are not shown. Population labels are created as subsistence_country, with following abbreviations: HG = hunter-gatherers, Farm = farmers, Fish = fishers, Indstr = industrialized, Cam = Cameroon, Tanz = Tanzania, It = Italy. https://doi.org/10.1371/journal.pone.0211139.g002 HG_Cam Farm_Cam Fish_Cam HG_Tanz HG_Peru Farm_Peru Indstr_USA Indstr_It 20 40 60 80 100 ST1 ST2 ST3 ST4 ST8 Blastocystis sp. 0 81.3% 95.5% 52.6% 81.8% 40% 75% 21.1% 9.1% Prevalence (%) Fig 2. Prevalence of Blastocystis sp. and its subtypes across populations according to metagenomic analysis. Subtypes with zero prevalence (ST6, ST7 and ST9) are not shown. Population labels are created as subsistence_country, with following abbreviations: HG = hunter-gatherers, Farm = farmers, Fish = fishers, Indstr = industrialized, Cam = Cameroon, Tanz = Tanzania, It = Italy. https://doi.org/10.1371/journal.pone.0211139.g002 https://doi.org/10.1371/journal.pone.0211139.g002 with a higher prevalence in Tanzania than in Cameroon (p = 0.024, OR = 0.44 [0.21, 0.87]), whereas subsistence mode was marginally significant (LRT = 6.33, p = 0.042), likely due to a complete absence of ST2 in fishers (S4 Table). Finally, the most prevalent Blastocystis subtype— ST3—was found in all populations except Peruvian hunter-gatherers, with a prevalence ranging from 9.1% in Italy to 63.6% in Cameroonian farmers (Fig 2). Fisher’s exact tests and logistic regression matched the results obtained for overall Blastocystis prevalence, reflecting the higher prevalence in non-industrialized countries (Fisher’s exact test, p < 10−3) as well as the effect of country (LRT = 9.29, p = 0.010) and subsistence mode (LRT = 6.57, p = 0.037), with significant differences between Africa and South America (p = 0.008, OR = 1.91 [1.23, 3.27]) and farmers and fishers (p = 0.019, OR = 2.16 [1.16, 4.26], S4 Table). PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019 Prevalence of Entamoeba spp. in culturally diverse populations MG-based detection revealed the presence of at least one Entamoeba species in 58 out of 127 (45.7%) individuals worldwide, with overall frequencies per population ranging from 0% in Italy and USA to a minimum of 37.5% in non-industrialized populations and reaching up to 90% in Peruvian hunter-gatherers (Fig 3, S3 Table). Regarding the distribution of different Entamoeba species within individuals (S2 Fig), although single colonization by either E. coli or E. hartmanni were the most common types of colonization (both 10.2%), both species were more often co-occurring with other species: 62.9% and 64.9% of individuals carrying E. coli PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019 9 / 20 Gut protozoa identification and prevalence determination using metagenomics E. coli E. dispar E. hartmanni E. polecki E. moshkovskii Entamoeba sp. 68.8% 77.3% 42.1% 45.5% 90% 37.5% 0% 0% 20 40 60 80 100 0 Prevalence (%) HG_Cam Farm_Cam Fish_Cam HG_Tanz HG_Peru Farm_Peru Indstr_USA Indstr_It Fig 3. Prevalence of Entamoeba spp. and its species across populations according to metagenomic analysis. Species with zero prevalence (E. chattoni and E. histolytica) are not shown. The population abbreviations are the same as in Fig 2. https://doi.org/10.1371/journal.pone.0211139.g003 Prevalence (%) HG_Tanz HG_Peru Fig 3. Prevalence of Entamoeba spp. and its species across populations according to metagenomic analysis. Species with zero prevalence (E. chattoni and E. histolytica) are not shown. The population abbreviations are the same as in Fig 2. https://doi.org/10.1371/journal.pone.0211139.g003 https://doi.org/10.1371/journal.pone.0211139.g003 resp. E. hartmanni also carried at least one additional species of Entamoeba. Similarly, E. dispar was found alone only in 2.3% of cases, compared to 12.6% in mixed-species colonizations. Regarding the mixed-species colonizations, the most common one was E. coli/E. hartmanni (8.7%), followed by E. dispar/E. hartmanni and E. coli/E. dispar/E. hartmanni (3.1% each), and E. coli/E. dispar (2.4%). The rest of the cases included either E. moshkovskii (4.7%) or E. polecki (a single individual, co-infected by E. coli and E. hartmanni), both of them found exclusively in mixed-species colonization. Overall, single colonization by E. coli and E. hartmanni were found in all non-industrialized populations, whereas all other combinations were restricted to different subsets of these populations. Considering the prevalence of Entamoeba according to subsistence mode (Fig 3, S3 Table), both the difference between non-industrialized and industrialized countries (p < 10−3) and the difference among the non-industrialized populations (p = 0.019) were supported by Fish- er’s exact test. PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019 Prevalence of Entamoeba spp. in culturally diverse populations Logistic regression showed a marginally significant effect of subsistence mode (LRT = 6.23, p = 0.044, S5 Table), reflecting a trend towards higher prevalence in hunter-gath- erers compared with farmers and fishers (p = 0.054, OR = 1.46 [1.01, 2.19]), but also in farmers compared with fishers (p = 0.075, OR = 1.76 [0.95, 3.34]). Although the main effect of country was not significant (LRT = 5.53, p = 0.063), regression coefficients suggested higher prevalence in Cameroon than in Tanzania (p = 0.024, OR = 2.18 [1.13, 4.45]). Regarding the prevalence of individual Entamoeba species (Fig 3), Fisher’s exact test revealed significantly higher frequencies in non-industrialized than in industrialized popula- tions for E. coli (p < 10−3), E. hartmanni (p < 10−3) and E. dispar (p = 0.004), but not for E. PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019 10 / 20 Gut protozoa identification and prevalence determination using metagenomics moshkovskii nor E. polecki (p = 0.334 and 1). Only the prevalence of E. coli significantly varied among non-industrialized populations (Fisher’s exact test, p = 0.013, other species p > 0.188), with differences related to both country and subsistence mode according to the logistic regres- sion model (LRT = 8.48 and 6.57, p = 0.014 and 0.037, respectively; S5 Table). In particular, whereas there was no difference between continents (p = 0.276), the prevalence was higher in Cameroon than in Tanzania (p = 0.007, OR = 2.64 [1.35, 5.65]), and in hunter-gatherers com- pared with farmers and fishers (p = 0.021, OR = 1.52 [1.07, 2.21]). Association patterns (co-occurrence and co-exclusion) between protozoan species Overall, 72.5% of all colonized individuals carried at least two protozoan species and only Blas- tocystis sp. ST1, ST2, ST3 and ST8, E. hartmanni, D. fragilis and C. belli were found alone (S4 Fig). The most prevalent protozoa identified in this study were Blastocystis sp. (60.6%) and Entamoeba spp. (45.7%). We found that the individuals carrying one of these two genera were more likely to carry the other as well (Chi-square test: chi-square (1) = 17.080, p < 10−3, Pear- son’s phi with 95% CI: 0.383 [0.224, 0.522], df = 125). However, this positive correlation disap- pears if low-prevalence industrialized populations are excluded (Chi-square test: chi-square (1) = 2.666, p = 0.103), indicating that the above result is mainly due to double-negatives observed in these populations. To examine association patterns between all 17 protozoan species/STs detected in this study, we performed a co-occurrence analysis within non-industrialized populations only. Out of possible 40 pairwise combinations, we detected seven significant associations in six species (Fig 4), all of them positive. Specifically, the most common Blastocystis subtype, ST3, was posi- tively correlated with E. dispar, E. hartmanni and E. moshkovskii. Whereas there were no sig- nificant associations between the Blastocystis sp. STs, Entamoeba spp. tended to preferentially occur together. E. coli was additionally positively associated with C. mesnili. Prevalence of Enteromonas hominis in culturally diverse populations Using MG, we detected Enteromonas hominis in all hunter-gatherer populations (6.2% - 50%), as well as in Cameroonian fishers (15.8%), whereas all other populations were negative (S3 Fig, S3 Table). Significant differences in the prevalence of E. hominis between non-industrialized and industrialized populations as well as within non-industrialized populations were sup- ported by Fisher’s exact test (p = 0.039 and 0.004, respectively). Logistic regression revealed a significant effect of country (LRT = 6.75, p = 0.034), with a higher prevalence in South America than in Africa (p = 0.022, OR = 0.52 [0.28, 0.89]), as well as a significant effect of subsistence mode (LRT = 12.32, p = 0.002), likely reflecting the absence of E. hominis in farmers (S6 Table). It is noteworthy that E. hominis was only identified in mixed-species colonization with other protozoa (S4 Fig)). MG: A method for detection and diagnosis of gut protozoa? Gut protozoa can strongly affect human gut ecosystem, and consequently human health, in both beneficial and detrimental ways. In order to understand the role of individual protozoan species in the gut ecosystem, we first need to develop efficient and reliable methods of detec- tion. Here we used metagenomic data to search for ten protozoan genera in eight human pop- ulations inhabiting four continents and characterized by contrasted subsistence modes. Our work shows that metagenomics is a promising method for detection and characterization of PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019 11 / 20 Gut protozoa identification and prevalence determination using metagenomics Chilomastix mesnili E. coli E. moshkovskii Blastocystis ST3 E. dispar E. hartmanni negative random positive Fig 4. Association patterns between protozoan species in non-industrialized populations. Only species with at least one significant correlation are shown. E. = Entamoeba. https://doi.org/10.1371/journal.pone.0211139.g004 E. hartmanni Fig 4. Association patterns between protozoan species in non-industrialized populations. Only species with at least one significant correlation are shown. E. = Entamoeba. https://doi.org/10.1371/journal.pone.0211139.g004 https://doi.org/10.1371/journal.pone.0211139.g004 protozoan assemblages in the gut ecosystem. Comparing different methods of detection (qPCR versus MG for Blastocystis sp. and microscopy versus MG for Entamoeba spp.), we found no significant difference in sensitivity between qPCR and MG, but a clear improvement in sensitivity between microscopy and MG. However, although qPCR and MG are not signifi- cantly different with respect to sensitivity, the biodiversity of Blastocystis sp. is better captured by metagenomics. This is quite expected, as qPCR is known to underestimate the prevalence of mixed-subtype colonization because of competitive amplification between differentially abundant STs. The main advantage of MG is the ability to simultaneously detect a potentially unlimited number of eukaryotic symbionts within a single sampling and analysis protocol. Still, precise and accurate detection of protozoa depends on many factors, among others: sampling and library preparation procedure [40], sequencing depth and coverage [41], community structure [42] as well as the bioinformatic approach [43–45]. Recently, Beghini et al. [24] analyzed the prevalence of Blastocystis sp. from published metagenomic data in the same cohort of Tanza- nian [26] and Peruvian [25] individuals as we did. Their estimates were considerably lower than ours: 55.6% vs. 81.5% in Tanzania and 16.7% vs. 55.6% in Peru (including all individuals, adults and children). Unlike Beghini et al. PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019 Gut protozoa identification and prevalence determination using metagenomics alignment with subsequent filtering of short and low-quality mappings. In addition, Beghini et al. [24] based their prevalence estimates on the breadth of coverage (with a 10% cutoff), while we used a combination of the hit distribution (10% positive contigs) and the breadth of coverage (with a 0.1% cutoff) as our detection criterion. It is noteworthy that applying such a stringent 10% cutoff for breadth of coverage would have resulted in very few positive samples in our case, reflecting our higher stringency in the alignment step. Still, our approach seems to be more sensitive for detection of mixed-subtype colonization, as minor STs are much less likely to be detected with Beghini’s et al. strict cutoff. Other published MG-based approaches for Blastocystis sp. detection include that of Ander- sen et al. [22], who first constructed co-abundance genes groups (CAGs), and then assigned them to a ST which the majority of the genes in the group aligned to, and that of Forsell et al. [23], who used SSU rRNA gene sequences instead of genomes. In both cases, the power to detect rare STs or mixed colonization of closely related species is considerably reduced. To summarize, intriguingly, none of the previous studies utilizing MG to detect Blastocystis sp. found mixed-subtype colonizations [22–24], while these are known to occur [46, 47]. We are thus the first to report mixed Blastocystis ST colonization by a metagenomic approach. DNA-based detection methods, be it PCR or MG, suffer from additional drawbacks in the applied context of detection of enteric pathogens. Namely, a positive result might be due to residual DNA from past infections or to asymptomatic carriage, which commonly occurs in endemic parasite regions [48]. Therefore, careful interpretation is necessary for both diagnos- tic routine and epidemiological evaluations. Among other challenges for the integration of MG into public health are: high cost, difficulties in comparison of the results from different NGS platforms, rapidly evolving technology and the need for high level of specialized exper- tise. Although universal bioinformatic approach in the context of public health is yet to be developed [48], the ability of MG to simultaneously address the variation in protozoan and bacterial assemblages will likely enable testing of relevant theoretical and applied questions concerning gut microbiome and human health. PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019 MG: A method for detection and diagnosis of gut protozoa? [24], we have also found ST8 in USA and ST3 in Italy, as well as a considerable portion of mixed-subtype colonization, with an especially high prevalence in the Tanzanian population (S1, S2 and S4 Figs). There is a number of differences between our and Beghini et al. [24] analytical procedure that might account for the differences in the prevalence estimates. Firstly, we handled the problem of contamination of publicly available genome assemblies differently. Beghini et al. [24] mapped the available Blastocystis genomes to 55 archaeal and bacterial genomes and esti- mated that 1–30% of nucleotides therein were of bacterial/archaeal origin. They removed these contaminated contigs prior to mapping, whereas we removed the contamination after the alignment step. Secondly, they performed global alignment, whereas we performed local 12 / 20 PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019 Prevalence of protoza in the human gut Out of ten genera we looked for, we detected all of them except Cryptosporidium spp., but only three had appreciable (>10%) frequencies: Blastocystis sp., Entamoeba spp. and Enteromonas hominis. As could be expected, we found few pathogenic species (absence of Cryptosporidium hominis, C. parvum and Entamoeba histolytica, one individual carrying Giardia intestinalis and three individuals carrying Cystoisospora belli). Furthermore, we found low prevalence of non-pathogenic species such as Iodamoeba bütschlii, Endolimax nana and Chilomastix mesnili. Similarly, Parfrey et al. 2014 [1], in their SSU rRNA gene amplicon-based study of gut eukary- otic communities, found Blastocystis and Entamoeba, but no Chilomastix, Cryptosporidium or Dientamoeba. However, we must consider that the DNA extraction methods used for the generation of the analyzed metagenomic datasets had not been optimized for protozoa, and especially cystic forms of these parasites are known to be often difficult to break (e.g. [49]). Notably, we detected Cystoisospora belli and Dientamoeba fragilis only in US samples, which were frozen within 24h after collection [25] unlike other samples that were stored in ethanol or ice for a longer time. However, it is known that the storage of stool at 4˚C for more than 3 days or at room temperature for one week seriously impedes detection of D. fragilis [50]. Furthermore, the excretion of vegetative or cystic forms is unpredictable in many proto- zoan species and intermittent excretion occurs regularly. In addition, SSU rRNA genes can exhibit intraspecific heterogeneity that would result in false negatives given our PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019 13 / 20 Gut protozoa identification and prevalence determination using metagenomics stringent detection criteria [51]. Nevertheless, despite our stringent cutoffs, we identified more than one protozoan species in 72.5% of infected individuals (S4 Fig). Co-coloniza- tion thus seems to be common and often accounts for a large proportion (>50%) of proto- zoan colonizations detected by traditional methods [18, 52–54]. It is likely that including additional species into the analysis would result in even higher estimates. Similar to other studies [53, 55, 56], we found even the relatively abundant species, such as Entamoeba coli or Enteromonas hominis, only in mixed colonizations. Therefore, the assembly of the non- pathogenic protozoan community in the gut does not seem to be driven by competition, but rather–as further indicated by the absence of co-exclusions (Fig 4)—by the presence of these protozoa in the environment and the general susceptibility of an individual. Prevalence of Blastocystis sp in worldwide populations Our results on the prevalence of Blastocystis sp. are in line with previous findings, showing that it frequently exceeds 50% in Western, Middle and Eastern African countries [15–19, 57], whereas it ranges from around 35% in urban [58] to over 50% in rural areas of South America [59]. The relatively low prevalence found among the fishers from Cameroon (52.6%) com- pared with the neighboring hunter-gatherer (81.3%) and farmer populations (95.5%) could reflect the difference in sanitary and environmental conditions. Namely, this group represents a coastal population that lives close to the main road (thus having better access to medical care) and in a less tropical environment than the neighboring hunter-gatherer and farmer pop- ulations. A corresponding difference in prevalence has also been observed for Entamoeba and intestinal worms [3]. Regarding the individual STs, the absence of ST6, ST7 and ST9, the broad geographic distri- bution of ST3, ST1 and ST2, as well as the narrow ST4 and ST8 distribution corroborate the results of previous studies [15–19, 24]. Our results provide further evidence for the virtual absence of ST4 from Africa and South America and the sporadic prevalence of ST8 in humans in general [16, 18, 24, 59, 60]. Although we found the prevalence of ST1, ST2 and ST3 to exceed 10% in each of the non-industrialized populations, their distribution varied considerably across countries and lifestyles (Fig 2). Beghini et al. [24] argued that ST2 was strongly affected by industri- alization based on its dominance in non-industrialized countries (Peru and Tanzania) and its absence from Europe and USA. Our results corroborate the dominance of ST2 in Tanzanian population, whereas for Peruvian population, we found ST2 to be only the third most prevalent ST. It is noteworthy that other, PCR-based studies of Blastocystis sp. STs in South America found a co-dominance of ST1, ST2 and ST3 in rural populations [59], whereas the prevalence of ST2 was considerably lower in urban communities [58]. Similar to Peru, ST2 is only the third most prevalent ST in Cameroon, with considerable (31.3%) prevalence only among hunter-gatherers. The difference in prevalence of ST2 between hunter-gatherers and farmers and fishers from Cameroon (31.3% versus 4.5% and 0%, resp.) is actually quite impressive given that these groups live in the same envi- ronment and in very close proximity. Prevalence of protoza in the human gut The ability of MG to detect mixed infections/colonization in a single analysis could signifi- cantly improve worldwide prevalence estimates, but also improve our knowledge about the clinical importance of different protozoa. PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019 Prevalence of Entamoeba spp. in worldwide populations MG revealed a very high prevalence of Entamoeba spp. in non-industrialized populations, as well as very frequent co-colonization of the four detected species of Entamoeba (Fig 3, S2 Fig). Prevalence of Entamoeba is often studied in children (e.g. [52, 54, 61], as the infection with E. histolytica considerably contributes to diarrheal burden and childhood mortality [62]. Whereas some studies report little variation in prevalence of Entamoeba with age [63], others find the contrary [59, 64, 65] and thus the results of these studies may not be directly compara- ble with ours. However, with few exceptions [52, 54], the reported prevalence is usually much lower (< 25%, [59, 61, 63–66]) than what we have found. Although the discrepancy could be explained by lower sensitivity of microscopy in some cases [63, 65], Ouattara et al. [54] found E. coli in 62% of examined children in Cote d’Ivoire using this method, possibly reflecting high parasite burden. Regarding the PCR-based studies, the most common species in our study— non-pathogenic E. hartmanni is less frequently searched for [54, 64] than the E. dispar/histoly- tica/moshkovskii complex [52, 54, 59, 64, 66], which could partially explain the observed differ- ence. Still, both Ouattara et al. [54] and Calegar et al. [64] report E. hartmanni prevalence < 5%, much lower than detected here. The prevalence of another common non-pathogenic Ent- amoeba in our study, E. coli, seems to vary widely both with geography and subsistence mode, as illustrated by the difference between Tanzania and Cameroon here (Fig 3, S3 Table), a very high prevalence in Cote d’Ivoire [54] and a 5%-25% prevalence in rural South American popu- lations [59, 64, 65]. E. coli was also a dominant Entamoeba (23%) in poor Indian communities [63]. Although our estimate of E. coli prevalence in Peru (44.4%) is much higher than usually reported from South America [59, 64, 65], a closer look reveals that this is due to a very high prevalence in hunter-gatherers (Fig 3), who were not included in the previous studies. On the other hand, the prevalence in Peruvian farming population corresponds well with the previous reports [59, 64, 65]. Regarding the E. dispar/histolytica/moshkovskii complex, the absence of pathogenic E. histo- lytica is expected given that the individuals were overall healthy. In all populations, we found more E. dispar than E. moshkovskii, which we detected only in Africa and never alone (Fig 3). Similarly, Ali et al. Prevalence of Blastocystis sp in worldwide populations Although the hunter-gatherer subsistence mode might seem conducive to colonization by ST2 if we look at Africa only, the absence of ST2 in Peruvian hunter-gatherers does not support this hypothesis. It seems that the factors influencing ST2 distribution are more complex and cannot be reduced to the large-scale effects of either industrialization or subsistence mode. PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019 14 / 20 Gut protozoa identification and prevalence determination using metagenomics Prevalence of Entamoeba spp. in worldwide populations [61] reported that the majority (74%) of E. moshkovskii cases in children in Bangladesh were mixed colonization. Conversely, in rural Columbia, Lo´pez et al. [52] found app. 25% prevalence of each E. dispar and E. moshkovskii and only 5% mixed colonization. HIV-positive Tanzanians actually carried more E. moshkovskii (13%) than E. coli (5%, [66]). In the arid areas of Brazil, Calegar et al. [64] found only E.dispar, but no E.moshkovskii. These contrasting results highlight the need for better understanding of the ecology of individual Ent- amoeba species and of the complex geographical and cultural factors influencing their distribution. It is important to note that grouping according to subsistence mode does not reflect fine- scale cultural and environmental differences that may be crucial for understanding of gut pro- tozoa colonization and prevalence in the studied populations. For example, most protozoa are transmitted through fecal-oral route and detailed knowledge about sanitary conditions would likely improve our understanding of the observed prevalence patterns. However, the subsis- tence mode is likely to influence sanitary conditions to a certain extent, and we therefore think it is a valuable information to account for in the absence of the more detailed data. Conclusion and perspectives Our study is the first to use a metagenomic approach to determine the prevalence of gut proto- zoa other than Blastocystis sp. Specifically, we used MG data to examine the prevalence of ten protozoan genera in eight populations across the globe. The particular approach we applied 15 / 20 PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019 Gut protozoa identification and prevalence determination using metagenomics enabled us to detect high prevalence of mixed colonization in healthy individuals from non- industrialized populations, including both closely and distantly related protozoa. Our results support the hypothesis that protozoa could be common residents of the healthy human gut [2], but also highlight the pitfalls of MG-based diagnostics due to variety of strategies that can be applied to the analysis of MG data. The role of MG in diagnostics will likely become more important with the increasing number of good-quality genomes in the public databases and the development of well-defined diagnostic protocols. Unveiling the factors that influence the assembly of gut protozoan communities and eluci- dating their interactions with the human host and its bacterial microbiota are an important step in understanding the gut protozoa role in human health and disease. We first need to bet- ter understand drivers of protozoa (co-)colonization by collecting detailed sanitary, medical, anthropological, dietary and other cultural and biological information about the individuals in addition to the sequencing data, as well as describe correlations between these factors. Further, MG-data provide an excellent opportunity to study the effect of gut protozoa on the gut bacte- rial microbiome diversity as well as its effect on human health. Few such studies exist, but accumulation of metagenomes from non-industrialized populations with higher gut protozoa prevalence should increase their number in the future. Finally, controlled experiments with vertebrate model organisms will be necessary in order to infer causality and fully elucidate the mechanisms and factors that affect ecological dynam- ics of individual gut protozoa as well as their communities and interactions with bacterial microbiota and the human host. Supporting information pp g S1 Table. Information about the samples used in the study including per-sample results after filtering of the multi-mapped reads. (XLSX) S2 Table. List of protozoan genomes and SSU sequences used for mapping of metagenomic data. (XLSX) S3 Table. Prevalence of detected protozoan species in different populations. (XLSX) S4 Table. Logistic regression models: effect of country and subsistence mode on Blastocys- tis prevalence in non-industrialized countries. (XLSX) S5 Table. Logistic regression models: effect of country and subsistence mode on Ent- amoeba prevalence in non-industrialized countries. (XLSX) S6 Table. Logistic regression models: effect of country and subsistence mode on Enteromo- nas prevalence in non-industrialized countries. (XLSX) S1 Fig. Distribution of Blastocystis sp. subtypes within individuals. (PDF) S2 Fig. Distribution of Entamoeba species within individuals. (PDF) S1 Table. Information about the samples used in the study including per-sample results after filtering of the multi-mapped reads. (XLSX) References 1. Parfrey LW, Walters WA, Lauber CL, Clemente JC, Berg-Lyons D, Teiling C, et al. Communities of microbial eukaryotes in the mammalian gut within the context of environmental eukaryotic diversity. Front Microbiol. 2014; 5: 298. https://doi.org/10.3389/fmicb.2014.00298 PMID: 24995004 2. Lukesˇ J, Stensvold CR, Jirků-Pomajbı´kova´ K, Wegener Parfrey L. Are Human Intestinal Eukaryotes Beneficial or Commensals? PLoS pathogens. 2015; 11: e1005039. https://doi.org/10.1371/journal. ppat.1005039 PMID: 26270819 3. Morton ER, Lynch J, Froment A, Lafosse S, Heyer E, Przeworski M, et al. Variation in Rural African Gut Microbiota Is Strongly Correlated with Colonization by Entamoeba and Subsistence. PLoS Genet. 2015; 11(11): e1005658. https://doi.org/10.1371/journal.pgen.1005658 PMID: 26619199 4. Andersen LOB, Stensvold CR. Blastocystis in Health and Disease: Are We Moving from a Clinical to a Public Health Perspective? Journal of Clinical Microbiology. 2016; 54: 524–8. https://doi.org/10.1128/ JCM.02520-15 PMID: 26677249 5. Audebert C, Even G, Cian A, Blastocystis Investigation G, Loywick A, Merlin S, et al. Colonization with the enteric protozoa Blastocystis is associated with increased diversity of human gut bacterial micro- biota. Sci Rep. 2016; 6: 25255. https://doi.org/10.1038/srep25255 PMID: 27147260 6. Chabe´ M, Lokmer A, Se´gurel L. Gut Protozoa: Friends or Foes of the Human Gut Microbiota? Trends in Parasitology. 2017; https://doi.org/10.1016/j.pt.2017.08.005 7. Ryan U, Paparini A, Oskam C. New Technologies for Detection of Enteric Parasites. Trends Parasitol. 2017; 33(7): 532–46. https://doi.org/10.1016/j.pt.2017.03.005 PMID: 28385423 8. Verweij JJ, Oostvogel F, Brienen EAT, Nang-Beifubah A, Ziem J, Polderman AM. Short communication Prevalence of Entamoeba histolytica and Entamoeba dispar in northern Ghana. Tropical Medicine and 1. Parfrey LW, Walters WA, Lauber CL, Clemente JC, Berg-Lyons D, Teiling C, et al. Communities of microbial eukaryotes in the mammalian gut within the context of environmental eukaryotic diversity. Front Microbiol. 2014; 5: 298. https://doi.org/10.3389/fmicb.2014.00298 PMID: 24995004 2. Lukesˇ J, Stensvold CR, Jirků-Pomajbı´kova´ K, Wegener Parfrey L. Are Human Intestinal Eukaryotes Beneficial or Commensals? PLoS pathogens. 2015; 11: e1005039. https://doi.org/10.1371/journal. ppat.1005039 PMID: 26270819 3. Morton ER, Lynch J, Froment A, Lafosse S, Heyer E, Przeworski M, et al. Variation in Rural African Gut Microbiota Is Strongly Correlated with Colonization by Entamoeba and Subsistence. PLoS Genet. 2015; 11(11): e1005658. https://doi.org/10.1371/journal.pgen.1005658 PMID: 26619199 4. Andersen LOB, Stensvold CR. Blastocystis in Health and Disease: Are We Moving from a Clinical to a Public Health Perspective? Journal of Clinical Microbiology. 2016; 54: 524–8. https://doi.org/10.1128/ JCM.02520-15 PMID: 26677249 5. Audebert C, Even G, Cian A, Blastocystis Investigation G, Loywick A, Merlin S, et al. Acknowledgments We wish to thank the “Plateau Technique de Pale´oge´nomique et Ge´ne´tique Mole´culaire (MNHN, site du Muse´e de l’Homme) for their technical help. We wish to thank the “Plateau Technique de Pale´oge´nomique et Ge´ne´tique Mole´culaire (MNHN, site du Muse´e de l’Homme) for their technical help. Project administration: Laure Se´gurel. Visualization: Ana Lokmer, Laure Se´gurel. Visualization: Ana Lokmer, Laure Se´gurel. Writing – original draft: Ana Lokmer, Eric Viscogliosi, Magali Chabe´, Laure Se´gurel. Writing – review & editing: Ana Lokmer, Eric Viscogliosi, Magali Chabe´, Laure Se´gurel. S3 Fig. Prevalence of Enteromonas hominis across populations. (PDF) S4 Fig. Coinfection patterns including all protozoan species. (PDF) S3 Table. Prevalence of detected protozoan species in different populations. (XLSX) S4 Table. Logistic regression models: effect of country and subsistence mode on Blastocys- tis prevalence in non-industrialized countries. (XLSX) S5 Table. Logistic regression models: effect of country and subsistence mode on Ent- amoeba prevalence in non-industrialized countries. (XLSX) S6 Table. Logistic regression models: effect of country and subsistence mode on Enteromo- nas prevalence in non-industrialized countries. (XLSX) S1 Fig. Distribution of Blastocystis sp. subtypes within individuals. (PDF) S2 Fig. Distribution of Entamoeba species within individuals. (PDF) 16 / 20 PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019 Gut protozoa identification and prevalence determination using metagenomics Conceptualization: Ana Lokmer, Magali Chabe´, Laure Se´gurel. Conceptualization: Ana Lokmer, Magali Chabe´, Laure Se´gurel. Conceptualization: Ana Lokmer, Magali Chabe´, Laure Se´gurel. Data curation: Ana Lokmer, Eric Viscogliosi, Magali Chabe´, Laure Se´gurel. Formal analysis: Ana Lokmer. Data curation: Ana Lokmer, Eric Viscogliosi, Magali Chabe´, Laure Se´gurel. Investigation: Amandine Cian, Alain Froment, Nausicaa Gantois, Laure Se´gurel. Methodology: Ana Lokmer, Magali Chabe´, Laure Se´gurel. Project administration: Laure Se´gurel. International Health. 2003; 8: 1153–6. https://doi.org/10.1046/j.1360-2276.2003.01145.x PMID: 14641852 9. Stauffer W, Abd-Alla M, Ravdin JI. Prevalence and incidence of Entamoeba histolytica infection in South Africa and Egypt. Arch Med Res. 2006; 37: 266–9. S0188-4409(05)00369-3 [pii]r10.1016/j. arcmed.2005.10.006. https://doi.org/10.1016/j.arcmed.2005.10.006 PMID: 16380329 10. Xime´nez C, Mora´n P, Rojas L, Valadez A, Go´mez A. Reassessment of the epidemiology of amebiasis: State of the art. Infection, Genetics and Evolution. 2009; 9: 1023–32. https://doi.org/10.1016/j.meegid. 2009.06.008 PMID: 19540361 11. Kirk MD, Pires SM, Black RE, Caipo M, Crump JA, Devleesschauwer B, et al. World Health Organiza- tion Estimates of the Global and Regional Disease Burden of 22 Foodborne Bacterial, Protozoal, and Viral Diseases, 2010: A Data Synthesis. PLoS Med. 2015; 12(12): e1001921. https://doi.org/10.1371/ journal.pmed.1001921 PMID: 26633831 12. Bart A, Wentink-Bonnema EMS, Gilis H, Verhaar N, Wassenaar CJa, van Vugt M, et al. Diagnosis and subtype analysis of Blastocystis sp. in 442 patients in a hospital setting in the Netherlands. BMC infec- tious diseases. 2013; 13: 389. https://doi.org/10.1186/1471-2334-13-389 PMID: 23972160 13. Scanlan PD, Hill CJ, Ross RP, Ryan CA, Stanton C, Cotter PD. The intestinal protist Blastocystis is not a common member of the healthy infant gut microbiota in a Westernized country (Ireland). Parasitology. 2018;10.1017/S0031182018000033: 1–5. https://doi.org/10.1017/S0031182018000033 14. El Safadi D, Cian A, Nourrisson C, Pereira B, Morelle C, Bastien P, et al. Prevalence, risk factors for infection and subtype distribution of the intestinal parasite Blastocystis sp. from a large-scale multi-cen- ter study in France. BMC Infectious Diseases. 2016; 16. https://doi.org/10.1186/s12879-016-1776-8 15. El Safadi D, Gaayeb L, Meloni D, Cian A, Poirier P, Wawrzyniak I, et al. Children of Senegal River Basin show the highest prevalence of Blastocystis sp. ever observed worldwide. BMC infectious diseases. 2014; 14: 164. https://doi.org/10.1186/1471-2334-14-164 PMID: 24666632 16. Alfellani MA, Stensvold CR, Vidal-Lapiedra A, Onuoha ESU, Fagbenro-Beyioku AF, Clark CG. Variable geographic distribution of Blastocystis subtypes and its potential implications. Acta Tropica. 2013; 126: 11–8. https://doi.org/10.1016/j.actatropica.2012.12.011 PMID: 23290980 17. Poulsen CS, Efunshile AM, Nelson JA, Stensvold CR. Epidemiological aspects of Blastocystis coloniza- tion in children in Ilero, Nigeria. American Journal of Tropical Medicine and Hygiene. 2016; 95: 175–9. https://doi.org/10.4269/ajtmh.16-0074 PMID: 27139454 18. Forsell J, Granlund M, Samuelsson L, Koskiniemi S, Edebro H, Evengård B. High occurrence of Blasto- cystis sp. subtypes 1–3 and Giardia intestinalis assemblage B among patients in Zanzibar, Tanzania. Parasites & Vectors. 2016; 9: 370. https://doi.org/10.1186/s13071-016-1637-8 19. D’Alfonso R, Santoro M, Essi D, Monsia A, Kabore´ Y, Gle´ C, et al. PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019 References Colonization with the enteric protozoa Blastocystis is associated with increased diversity of human gut bacterial micro- biota. Sci Rep. 2016; 6: 25255. https://doi.org/10.1038/srep25255 PMID: 27147260 6. Chabe´ M, Lokmer A, Se´gurel L. Gut Protozoa: Friends or Foes of the Human Gut Microbiota? Trends in Parasitology. 2017; https://doi.org/10.1016/j.pt.2017.08.005 7. Ryan U, Paparini A, Oskam C. New Technologies for Detection of Enteric Parasites. Trends Parasitol. 2017; 33(7): 532–46. https://doi.org/10.1016/j.pt.2017.03.005 PMID: 28385423 8. Verweij JJ, Oostvogel F, Brienen EAT, Nang-Beifubah A, Ziem J, Polderman AM. Short communication: Prevalence of Entamoeba histolytica and Entamoeba dispar in northern Ghana. Tropical Medicine and 17 / 20 PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019 Gut protozoa identification and prevalence determination using metagenomics International Health. 2003; 8: 1153–6. https://doi.org/10.1046/j.1360-2276.2003.01145.x PMID: 14641852 International Health. 2003; 8: 1153–6. https://doi.org/10.1046/j.1360-2276.2003.01145.x PMID: 14641852 International Health. 2003; 8: 1153–6. https://doi.org/10.1046/j.1360-2276.2003.01145.x PMID: 14641852 Blastocystis in Coˆte d’Ivoire: molecu- lar identification and epidemiological data. European Journal of Clinical Microbiology & Infectious Dis- eases. 2017;10.1007/s10096-017-3053-1. https://doi.org/10.1007/s10096-017-3053-1 20. Pallen MJ. Diagnostic metagenomics: potential applications to bacterial, viral and parasitic infections. Parasitology. 2014; 141: 1856–62. https://doi.org/10.1017/S0031182014000134 PMID: 24576467 21. Simner PJ, Miller S, Carroll KC. Understanding the Promises and Hurdles of Metagenomic Next-Gener- ation Sequencing as a Diagnostic Tool for Infectious Diseases. Clinical Infectious Diseases. 2018; 66: 778–88. https://doi.org/10.1093/cid/cix881 PMID: 29040428 22. Andersen LO, Bonde I, Nielsen HB, Stensvold CR. A retrospective metagenomics approach to studying Blastocystis. FEMS Microbiol Ecol. 2015; 91(7). https://doi.org/10.1093/femsec/fiv072 23. Forsell J, Bengtsson-Palme J, Angelin M, Johansson A, Evengård B, Granlund M. The relation between Blastocystis and the intestinal microbiota in Swedish travellers. BMC Microbiology. 2017; 17. https://doi. org/10.1186/s12866-017-1139-7 24. Beghini F, Pasolli E, Truong TD, Putignani L, Cacciò SM, Segata N. Large-scale comparative metage- nomics of Blastocystis, a common member of the human gut microbiome. The ISME journal. 2017; 11: 2848. https://doi.org/10.1038/ismej.2017.139 PMID: 28837129 25. Obregon-Tito AJ, Tito RY, Metcalf J, Sankaranarayanan K, Clemente JC, Ursell LK, et al. Subsistence strategies in traditional societies distinguish gut microbiomes. Nature Communications. 2015; 6: 1–9. https://doi.org/10.1038/ncomms7505 26. Rampelli S, Schnorr S, Consolandi C, Turroni S, Severgnini M, Peano C, et al. Metagenome Sequenc- ing of the Hadza Hunter-Gatherer Gut Microbiota. Current Biology. 2015;10.1016/j.cub.2015.04.055: 1–24. https://doi.org/10.1016/j.cub.2015.04.055 27. Bushnell B. BBMap short read aligner, and other bioinformatic tools2017. Available from: https:// sourceforge.net/projects/bbmap/. 18 / 20 PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019 Gut protozoa identification and prevalence determination using metagenomics 28. Yatsunenko T, Rey F, Manary M, Trehan I, Dominguez-Bello M, Contreras M, et al. Human gut micro- biome viewed across age and geography. Nature. 2012;10.1038/nature11053: 1–7. https://doi.org/10. 1038/nature11053 29. Andrews S. FastQC: a quality control tool for high throughput sequence data.2010. Available from: http://www.bioinformatics.babraham.ac.uk/projects/fastqc. 30. Li H. Aligning sequence reads, clone sequences and assembly contigs with BWA-MEM. arXiv preprint arXiv:13033997. 2013. 31. Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, et al. The Sequence Alignment/Map format and SAMtools. Bioinformatics. 2009; 25(16): 2078–9. https://doi.org/10.1093/bioinformatics/btp352 PMID: 19505943 32. Quinlan AR, Hall IM. BEDTools: a flexible suite of utilities for comparing genomic features. Bioinformat- ics. 2010; 26: 841–2. https://doi.org/10.1093/bioinformatics/btq033 PMID: 20110278 33. O’Leary NA, Wright MW, Brister JR, Ciufo S, Haddad D, McVeigh R, et al. Reference sequence (RefSeq) database at NCBI: current status, taxonomic expansion, and functional annotation. Nucleic Acids Research. 2016; 44: D733–D45. International Health. 2003; 8: 1153–6. https://doi.org/10.1046/j.1360-2276.2003.01145.x PMID: 14641852 https://doi.org/10.1093/nar/gkv1189 PMID: 26553804 34. Coordinators NR. Database resources of the National Center for Biotechnology Information. Nucleic Acids Res. 2016; 44(D1): D7–19. https://doi.org/10.1093/nar/gkv1290 PMID: 26615191 35. Poirier P, Wawrzyniak I, Albert A, El Alaoui H, Delbac F, Livrelli V. Development and evaluation of a real-time PCR assay for detection and quantification of blastocystis parasites in human stool samples: prospective study of patients with hematological malignancies. Journal of clinical microbiology. 2011; 49: 975–83. https://doi.org/10.1128/JCM.01392-10 PMID: 21177897 36. Alfellani MA, Taner-Mulla D, Jacob AS, Imeede CA, Yoshikawa H, Stensvold CR, et al. Genetic diversity of blastocystis in livestock and zoo animals. Protist. 2013; 164(4): 497–509. https://doi.org/10.1016/j. protis.2013.05.003 PMID: 23770574 37. R Core Team. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2017. 38. Griffith DM, Veech JA, Marsh CJ. cooccur: Probabilistic Species Co-Occurrence Analysis in R. 2016. 2016; 69(Code Snippet 2): 17. https://doi.org/10.18637/jss.v069.c02 39. Lex A, Gehlenborg N, Strobelt H, Vuillemot R, Pfister H. UpSet: Visualization of Intersecting Sets. IEEE Transactions on Visualization and Computer Graphics. 2014; 20: 1983–92. https://doi.org/10.1109/ TVCG.2014.2346248 PMID: 26356912 40. Escobar-Zepeda A, Vera-Ponce de Leon A, Sanchez-Flores A. The Road to Metagenomics: From Microbiology to DNA Sequencing Technologies and Bioinformatics. Front Genet. 2015; 6: 348. https:// doi.org/10.3389/fgene.2015.00348 PMID: 26734060 41. Rodriguez-r LM, Konstantinidis KT. Estimating coverage in metagenomic data sets and why it matters. The ISME journal. 2014; 8: 2349. https://doi.org/10.1038/ismej.2014.76 PMID: 24824669 42. Peabody MA, Van Rossum T, Lo R, Brinkman FSL. Evaluation of shotgun metagenomics sequence classification methods using in silico and in vitro simulated communities. BMC Bioinformatics. 2015; 16. https://doi.org/10.1186/s12859-015-0788-5 43. Sharpton TJ. An introduction to the analysis of shotgun metagenomic data. Frontiers in Plant Science. 2014; 5. https://doi.org/10.3389/fpls.2014.00209 44. Breitwieser FP, Lu J, Salzberg SL. A review of methods and databases for metagenomic classification and assembly. Briefings in Bioinformatics. 2017; https://doi.org/10.1093/bib/bbx120 45. Sczyrba A, Hofmann P, Belmann P, Koslicki D, Janssen S, Dro¨ge J, et al. Critical Assessment of Meta- genome Interpretation—a benchmark of metagenomics software. Nature Methods. 2017; 14: 1063–71. https://doi.org/10.1038/nmeth.4458 PMID: 28967888 46. Meloni D, Poirier P, Mantini C, Noe¨l C, Gantois N, Wawrzyniak I, et al. Mixed human intra- and inter-sub- type infections with the parasite Blastocystis sp. Parasitology International. 2012; 61: 719–22. https:// doi.org/10.1016/j.parint.2012.05.012 PMID: 22659011 47. Scanlan PD, Stensvold CR, Cotter PD. Development and application of a Blastocystis subtype-specific PCR assay reveals that mixed-subtype infections are common in a healthy human population. International Health. 2003; 8: 1153–6. https://doi.org/10.1046/j.1360-2276.2003.01145.x PMID: 14641852 Applied and Environmental Microbiology. 2015; 81: 4071–6. https://doi.org/10.1128/AEM.00520-15 PMID: 25841010 48. Frickmann H, Schwarz NG, Rakotozandrindrainy R, May J, Hagen RM. PCR for enteric pathogens in high-prevalence settings. What does a positive signal tell us? Infectious Diseases. 2015; 47: 491–8. https://doi.org/10.3109/23744235.2015.1022212 PMID: 25761823 49. Hawash Y. DNA extraction from protozoan oocysts/cysts in feces for diagnostic PCR. Korean J Parasi- tol. 2014; 52(3): 263–71. https://doi.org/10.3347/kjp.2014.52.3.263 PMID: 25031466 19 / 20 PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019 Gut protozoa identification and prevalence determination using metagenomics 50. Stark D, Beebe N, Marriott D, Ellis J, Harkness J. Detection of Dientamoeba fragilis in fresh stool speci- mens using PCR. International Journal for Parasitology. 2005; 35: 57–62. https://doi.org/10.1016/j. ijpara.2004.09.003 PMID: 15619516 51. Xiao L, Escalante L, Yang C, Sulaiman I, Escalante AA, Montali RJ, et al. Phylogenetic analysis of Cryp- tosporidium parasites based on the small-subunit rRNA gene locus. Applied and environmental microbi- ology. 1999; 65: 1578–83. PMID: 10103253 52. Lo´pez MC, Leo´n CM, Fonseca J, Reyes P, Moncada L, Olivera MJ, et al. Molecular Epidemiology of Entamoeba: First Description of Entamoeba moshkovskii in a Rural Area from Central Colombia. PLOS ONE. 2015; 10: e0140302. https://doi.org/10.1371/journal.pone.0140302 PMID: 26465744 53. Cancrini G, Bartoloni A, Nuñez L, Paradisi F. Intestinal parasites in the Camiri, Gutierrez and Boyuibe areas, Santa Cruz Department, Bolivia. Parassitologia. 30: 263–9. PMID: 3271990 54. Ouattara M, N’Guessan N A, Yapi A, N’Goran E K. Prevalence and spatial distribution of Entamoeba histolytica/dispar and Giardia lamblia among schoolchildren in Agboville area (Cote d’Ivoire). PLoS Negl Trop Dis. 2010; 4(1): e574. https://doi.org/10.1371/journal.pntd.0000574 PMID: 20087416 55. Goldsmid JM, Rogers SUE, Mahomed K. Observations on the Intestinal Protozoa Infecting Man In Rho- desia. 1976: 4. 56. El Shazly AM, Awad SE, Sultan DM, Sadek GS, Khalil HHM, Morsy TA. Intestinal parasites in Dakahlia governorate, with different techniques in diagnosing protozoa. Journal of the Egyptian Society of Parasi- tology. 2006; 36: 1023–34. PMID: 17153711 57. Iebba V, Santangelo F, Totino V, Pantanella F, Monsia A, Di Cristanziano V, et al. Gut microbiota related to Giardia duodenalis, Entamoeba spp. and Blastocystis hominis infections in humans from Cote d’Ivoire. J Infect Dev Ctries. 2016; 10(9): 1035–41. https://doi.org/10.3855/jidc.8179 PMID: 27694739 58. Valenc¸a Barbosa C, de Jesus Batista R, Pereira Igreja R, d’Avila Levy CM, Werneck de Macedo H, Car- neiro Santos HL. Distribution of Blastocystis subtypes isolated from humans from an urban community in Rio de Janeiro, Brazil. Parasites & Vectors. PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019 International Health. 2003; 8: 1153–6. https://doi.org/10.1046/j.1360-2276.2003.01145.x PMID: 14641852 2017; 10. https://doi.org/10.1186/s13071-017-2458-0 59. Barbosa CV, Barreto MM, Andrade RdJ, Sodre´ F, d’Avila-Levy CM, Peralta JM, et al. Intestinal parasite infections in a rural community of Rio de Janeiro (Brazil): Prevalence and genetic diversity of Blastocys- tis subtypes. PLoS ONE. 2018; 13: e0193860. https://doi.org/10.1371/journal.pone.0193860 PMID: 29522552 60. Ramı´rez JD, Sa´nchez A, Herna´ndez C, Flo´rez C, Bernal MC, Giraldo JC, et al. Geographic distribution of human Blastocystis subtypes in South America. Infection, Genetics and Evolution. 2016; 41: 32–5. https://doi.org/10.1016/j.meegid.2016.03.017 PMID: 27034056 61. Ali IKM, Hossain MB, Roy S, Ayeh-Kumi PF, Petri WA, Haque R, et al. Entamoeba moshkovskii infec- tions in children, Bangladesh. Emerging infectious diseases. 2003; 9: 580–4. https://doi.org/10.3201/ eid0905.020548 PMID: 12737742 62. Haque R, Mondal D, Duggal P, Kabir M, Roy S, Farr BM, et al. Entamoeba histolytica Infection in Chil- dren and Protection from Subsequent Amebiasis. Infection and Immunity. 2006; 74: 904–9. https://doi. org/10.1128/IAI.74.2.904-909.2006 PMID: 16428733 63. Dhanabal J, Selvadoss PP, Muthuswamy K. Comparative study of the prevalence of intestinal parasites in low socioeconomic areas from South chennai, India. J Parasitol Res. 2014; 2014: 630968. https://doi. org/10.1155/2014/630968 PMID: 24587897 64. Calegar DA, Nunes BC, Monteiro KJL, dos Santos JP, Toma HK, Gomes TF, et al. Frequency and molecular characterisation of Entamoeba histolytica, Entamoeba dispar, Entamoeba moshkovskii, and Entamoeba hartmanni in the context of water scarcity in northeastern Brazil. Memo´rias do Instituto Oswaldo Cruz. 2016; 111: 114–9. https://doi.org/10.1590/0074-02760150383 PMID: 26841049 65. Chacin-Bonilla L, Bonilla E, Parra AM, Estevez J, Morales LM, Suarez H. Prevalence of Entamoeba his- tolytica and other intestinal parasites in a community from Maracaibo, Venezuela. Annals of Tropical Medicine & Parasitology. 1992; 86: 373–80. https://doi.org/10.1080/00034983.1992.11812680 66. Beck DL, Doğan N, Maro V, Sam NE, Shao J, Houpt ER. High prevalence of Entamoeba moshkovskii in a Tanzanian HIV population. Acta tropica. 2008; 107: 48–9. https://doi.org/10.1016/j.actatropica. 2008.03.013 PMID: 18471796 20 / 20 PLOS ONE \| https://doi.org/10.1371/journal.pone.0211139 February 6, 2019
https://openalex.org/W2953299812	https://zenodo.org/records/5885946/files/422-423.pdf	English	null	Synthesis of some New 2-Thioxoquinazolin-4-ones	Zenodo (CERN European Organization for Nuclear Research)	1,997	cc-by	1,188	Experimental All m.ps. were determined in open capillary tubes and are uncorrected. Ir spectra (KBr) were recorded on a Shimadt.u IR-437 spectrophotometer and pmr spectra (TFA/ DMSO-d6) on a Perkin-Elmer R-32 spectrometer using TMS as an internal standard. Purity of compounds was checked by tlc (silica gel G). Dithiocarbamate salt was pre- pared by reacting substituted anilines with carbon disulphide in presence of triethy1amine2. 3-Aryl-2-thioxo-4(3H)-quinazolin-4-one (la): A mix· ture of dithiocarbamate salt of substituted aniline in etha· nol (10 g, 0.1 mol) and anthranilic acid (15 g. 0.1 mol) was refluxed on a steam-bath for 5 h. The excess solvent was then removed under reduced pressure. The separated !>Olid was redissolved in ethanolic NaOH, filtered and the fil- trate on neutralisation with dilute HCI gave a solid which was crystallised from ethanol, (88.7%), m.p. 240" (Found: C, 67.16; H, 4.48, N,10.45. C 15H120N2S requires: C, 67.10; H, 4.5; N, 10.30%); ~·max 3 250 (NH), 1 660 (C=O), 1 620 cm-1(C=C); 6 2.3 (3H, s, ArCH3), 6.8-8 (8H. m. ArH). 13.0 ( lH, sbr, NH). Simi! arty, other quinazolinones (yields 84-97%) were prepared: lb, m.p. 26::?.•; c, 286°; d. 248"; e, 2&8•; f, 220", g, 253•; h, 256°. ( 2 ) Methyl [ 3, 4-dihydro-3-( o-methylpheuyl)-4-oxo-qwna· zolinyl]thioacetate (2a 1) : To a solution of quinazolinone 1 (0.1 mol) and methyl chloroacetate (0.1 mol) in acetone (50 ml), K2C03 (0.5 g) was added and the reaction mix· ture was refluxed for -18 h on an oil-bath. The sol vent was then removed under reduced pressure and cooled to get a solid which was crystallised from ethanol, (70.13%), m.p. 106• (Found : C, 62.42; H, 4.70; N, 8.2. C 18H 160 3N2S requires : C, 62.2; H, 4.8; N, 8.4%); Vmax l 750 (C=O). 1 700-l 690 (cyclic C=O), 1 620 (C=N) cm-1: 82.15 (3H. s, ArCH3), 3.7 (3H, S, O-CH3). 4.0 (2H, S, S-C~). 7-8.1 (8H, m, ArH). Similarly, other compounds (yields 61-85%) were prepared : 2b)' m.p.l32°; c)' 122°; dl, 100°; ep 68": f1, 78•; g1• 140"; h 1, 242•; ~· too•; h2, 242•; d~. 1ss•; h2• 246·. ( 4} R= R a: o-CH3, b; m-CH3• c; p-CH3: d; m-Cl, e; p-CI. C; p-OCH3. g; o-OCH3. h; m-OCH3. Scheme I. Reagents: (ia) ClCHzCOOCHf1<2C03, acetone, (ib) CHr CHCl·COOCHiK2C03, acetone. (rll NH1NH2 H20/ethanol and (iii) Ph-N=C=S/ethanol. Scheme I. Reagents: (ia) ClCHzCOOCHf1<2C03, acetone, (ib) CHr CHCl·COOCHiK2C03, acetone. (rll NH1NH2 H20/ethanol and (iii) Ph-N=C=S/ethanol. Synthesis of some New 2-Thioxoquinazolin-4-ones Department of Chemistry, Shivaji University, Kolhapur-416 004 Manuscnpt recerved 19 May 1995, revrsed 16 November1995, accepted 19 January 1996 Quinazolines exhibit important physiological activities1• Here we report the synthesis of some new 3-substituted- thiocarbamate derivatives of 2-thioquinazolin-4-one. reaction with hydrazine hydrate furnished the correspond- ing 3-aryl-2-thioacetylhydrazidoquinazolin-4-one (3a 1-h1) and 3-aryl-2-thiopropionylhydrazidoquinazolin-4-one (3a2, 3b2, 3d2, 3h2). Compounds 3a 1-h 1 and 3a2, 3b2, 3d2, 3h2 on reaction with phenyl isothiocyanate gave the desired thiocarbarnates (4a1-h1) and (4a2, 4b2, 4~. 4~) respec- tively. reaction with hydrazine hydrate furnished the correspond- ing 3-aryl-2-thioacetylhydrazidoquinazolin-4-one (3a 1-h1) and 3-aryl-2-thiopropionylhydrazidoquinazolin-4-one (3a2, 3b2, 3d2, 3h2). Compounds 3a 1-h 1 and 3a2, 3b2, 3d2, 3h2 on reaction with phenyl isothiocyanate gave the desired thiocarbarnates (4a1-h1) and (4a2, 4b2, 4~. 4~) respec- tively. thiocarbamate derivatives of 2 thioquinazolin 4 one. For i a X • - CH 2- ~ 0 R ((;-Q S-X-COOCH3 ( 2 ) ( ii l~ 0 R ((:~--0 N S-X-CONHNH2 ( 3) (iii)~ 0 R ~N-o ~N~ - ~ 5- X- CONHN H- C- NHPh ( 4} R= a: o-CH3, b; m-CH3• c; p-CH3: d; m-Cl, e; p-CI. C; p-OCH3. g; o-OCH3. h; m-OCH3. Scheme I. Reagents: (ia) ClCHzCOOCHf1<2C03, acetone, (ib) CHr CHCl·COOCHiK2C03, acetone. (rll NH1NH2 H20/ethanol and (iii) Ph-N=C=S/ethanol. Rc•mlts and Discussion NOTE NOTE Rc•mlts and Discussion 3-Aryl-2-thioxo-4(3H}-quinazolinone (l) on reaction with methyl chloroacetate and methyl 2-chloropropionate gave 3,4-dihydro-3-(substituted-phenyl)-4-oxo-2-quina- zoline thioacetic/thiopropionic acid alkyl esters (2a 1-h 1) and (2a2• 2b2, 2d2, 2h2) respectively, which on separate 422 NOTE nol, (63.7%), m.p.175" (Found: C, 65; H, 4.74; N, 15.8. C24H210 2N5S requires : C, 65.2; H, 4.74; N, 14.5); l'max 3 300 (NH), I 680, (C=O), I 620 cm-1 (C=N); b2.35 (3H, s, ArCH3), 3.7 (2H, s, S-CH2), 4.8 (br, s, 2xNH), 7-7.65 (13H, m, ArH), 10.5 (IH, s, CONH). Similarly, other thiocarbamates (yields 50-68%) were prepared : 4g 1, m.p. 58"; h1, 56"; b2 , 62"; d2 , 55"; h2 , 59". nol, (63.7%), m.p.175" (Found: C, 65; H, 4.74; N, 15.8. C24H210 2N5S requires : C, 65.2; H, 4.74; N, 14.5); l'max 3 300 (NH), I 680, (C=O), I 620 cm-1 (C=N); b2.35 (3H, s, ArCH3), 3.7 (2H, s, S-CH2), 4.8 (br, s, 2xNH), 7-7.65 (13H, m, ArH), 10.5 (IH, s, CONH). Similarly, other thiocarbamates (yields 50-68%) were prepared : 4g 1, m.p. 58"; h1, 56"; b2 , 62"; d2 , 55"; h2 , 59". nol, (63.7%), m.p.175" (Found: C, 65; H, 4.74; N, 15.8. C24H210 2N5S requires : C, 65.2; H, 4.74; N, 14.5); l'max 3 300 (NH), I 680, (C=O), I 620 cm-1 (C=N); b2.35 (3H, s, ArCH3), 3.7 (2H, s, S-CH2), 4.8 (br, s, 2xNH), 7-7.65 (13H, m, ArH), 10.5 (IH, s, CONH). Similarly, other thiocarbamates (yields 50-68%) were prepared : 4g 1, m.p. 58"; h1, 56"; b2 , 62"; d2 , 55"; h2 , 59". 3, 4-Dihydro-3 -( o-meth ylphenyl )-4-oxo-2 -qu inazo- linylthioacetic acid hydrazide (3a) : A mixture of 2 (0.1 mol) and hydrazine hydrate (80%; 0. 1 mol) in ethanol (25 ml) was refluxed on a steam-bath for 3 hat 70". The sol- vent was then removed under reduced pressure to get a solid which was crystallised from ethanol, (83.7%), m.p.l38"(Found: C, 60.0; H, 4.71, N. 16.47. C 17H 160 2N4S requires : C, 60.2; H, 4.8; N, I6.2%); Vmax 3 250 (NHNH2), I 680 (C=O), 1 660 (C=O), I 600 cm-1: b 2.3 (3H, s, ArCH3), 3.7 (2H, s, S-CH2), 4.8 (2H, s, NH2), 7-7.5 (8H, m, ArH),l0.5 ( lH, s, CONH). Other hydrazides (yields 52- 94%) were prepared similarly: 3b1, m.p.135"; cl' 148"; dp I95"; cl' 208"; fp 196"; g1, 170"; h1, 255"; ~· ll2"; b2, 203"; d2, 280"; h2, 230•. 2. R. LAKIIAJio and M. SRIVASTAV, Proc Tndran Acad Scr (Chern Set), 19Q3, 105, II. References I. H. Y AMMAMOTO, Jap. Pat. 6917136/1969 (Chem Abstr 191\9,71, 123505); H. YAMMAMOTO.M. NAKAO and E.SCOTIONE, US Pat. 3479348/1970(Chem. Abstr.,1970, 73, 117~8); H. ZEUENER. M PAllER and PRACKMA\R, Au•t. Pat. 228204/1963 (Chem Abs11., 1963, 59, 1251); T. 0. YELUN, US Pat, 3635971/1972 (Chem Abstr., 1972,76, 99708); I. A HAMED, J Tndran Chem Soc., 1988, 65, 362; A. KUMAR, B. SINGH, A. K. SAXENA and K. SHANKAR, Indian J Chem, See/ B. 1988, 27, 443; S. SAXI:..'>IA. M. BHALLA. M. VERMA, A. K. SAXENA and K. SHA 'IKAR, J lndtmr Chem Soc., 1991, 68, 142. The thiocarbamate derivatives (4a 1) : A mixture of 3 (0.1 mol) and phenyl isothiocyanate (O.I mol) in ethanol (20 ml) was refluxed for 3 h on a steam-bath and the ex- cess of solvent was then removed under reduced pressure and cooled. The resulting solid was crystallised from etha- 2. R. LAKIIAJio and M. SRIVASTAV, Proc Tndran Acad Scr (Chern Set), 19Q3, 105, II. 423 J/CS -11
https://openalex.org/W2093361724	https://europepmc.org/articles/pmc3317846?pdf=render	English	null	Generic phylogeny, historical biogeography and character evolution of the cosmopolitan aquatic plant family Hydrocharitaceae	BMC evolutionary biology	2,012	cc-by	10,521	* Correspondence: qfwang@wbgcas.cn 1CAS Key Laboratory of Aquatic Botany and Watershed Ecology, Chinese Academy of Sciences, Wuhan 430074, Hubei, P. R. China Full list of author information is available at the end of the article RESEARCH ARTICLE Open Access Open Access Generic phylogeny, historical biogeography and character evolution of the cosmopolitan aquatic plant family Hydrocharitaceae Ling-Yun Chen1,2,3, Jin-Ming Chen1,2, Robert Wahiti Gituru4 and Qing-Feng Wang1,2* © 2012 Chen et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Chen et al. BMC Evolutionary Biology 2012, 12:30 http://www.biomedcentral.com/1471-2148/12/30 Chen et al. BMC Evolutionary Biology 2012, 12:30 http://www.biomedcentral.com/1471-2148/12/30 * Correspondence: qfwang@wbgcas.cn 1CAS Key Laboratory of Aquatic Botany and Watershed Ecology, Chinese Academy of Sciences, Wuhan 430074, Hubei, P. R. China Full list of author information is available at the end of the article © 2012 Chen et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Background Hydrocharitaceae is a fully aquatic monocot family, con- sists of 18 genera and approximately 120 species [1,2] with a cosmopolitan distribution. The group is an important component of aquatic ecosystems and occurs in both freshwater and marine habitats. It also exhibits great diversity in form and habit including annual and perennial life histories; submersed, partially submersed and floating leaf habits, and linear to orbicular leaf shapes [1]. It includes a number of valuable aquarium plants and some species (e.g., Hydrilla verticillata and Elodea canadensis) provide fish breeding sites and fod- der for the poultry raising industry [3]. Owing to habitat destruction and unprecedented increase in commercial shipping activities, several species (e.g., Ottelia alis- moides and Blyxa japonica) are threatened, while some other species (e.g., Hydrilla verticillata) have become invasive weeds of great concern [4]. Similar to many aquatic plants, the group displays considerable conver- gent adaptations and morphological reduction, which make the study of phylogenetics and taxonomy of the group difficult [1,5-7]. There exists little consensus in classification systems of the group based on morphologi- cal characters [1,8]. The geographic origin of Hydrocharitaceae remains unresolved. The diversity centre of the family has been suggested to be in tropical Asia [20]. However, the diversity centre of a taxon does not necessarily corre- spond to its centre of origin. Numerous fossils of Hydrocharitaceae have been found in Europe [17], implying a possible European origin of the family. A biogeographic analysis is required to elucidate the geo- graphical origin of the family. Although molecular phylogenetic studies of Hydro- charitaceae have created consensus on the relationships between certain genera, disagreements on the relation- ships of other genera still exist. The seagrass subclade which includes Halophila, Enhalus and Thalassia was resolved as sister to the subclade comprising Najas, Hydrilla and Vallisneria by analyses using rbcL+ matK [8] and rbcL+ cob + atp1 [9]. However, the seagrasses was resolved as sister to the subclade comprising Necha- mandra, Vallisneria and Maidenia by analyses using morphological + rbcL+ matK+ trnK intron + ITS [1]. Similarly, Hydrilla, which was reported to be closely related to Vallisneria by analyses using rbcL, matK [2,8], rbcL, cob and atp1 [9], was reported to be closely related to Najas by analyses using morphological char- acter + rbcL+ matK+ trnK intron + ITS [1]. Abstract Background: Hydrocharitaceae is a fully aquatic monocot family, consists of 18 genera with approximately 120 species. The family includes both fresh and marine aquatics and exhibits great diversity in form and habit including annual and perennial life histories; submersed, partially submersed and floating leaf habits and linear to orbicular leaf shapes. The family has a cosmopolitan distribution and is well represented in the Tertiary fossil record in Europe. At present, the historical biogeography of the family is not well understood and the generic relationships remain controversial. In this study we investigated the phylogeny and biogeography of Hydrocharitaceae by integrating fossils and DNA sequences from eight genes. We also conducted ancestral state reconstruction for three morphological characters. Results: Phylogenetic analyses produced a phylogeny with most branches strongly supported by bootstrap values greater than 95 and Bayesian posterior probability values of 1.0. Stratiotes is the first diverging lineage with the remaining genera in two clades, one clade consists of Lagarosiphon, Ottelia, Blyxa, Apalanthe, Elodea and Egeria; and the other consists of Hydrocharis-Limnobium, Thalassia, Enhalus, Halophila, Najas, Hydrilla, Vallisneria, Nechamandra and Maidenia. Biogeographic analyses (DIVA, Mesquite) and divergence time estimates (BEAST) resolved the most recent common ancestor of Hydrocharitaceae as being in Asia during the Late Cretaceous and Palaeocene (54.7-72.6 Ma). Dispersals (including long-distance dispersal and migrations through Tethys seaway and land bridges) probably played major roles in the intercontinental distribution of this family. Ancestral state reconstruction suggested that in Hydrocharitaceae evolution of dioecy is bidirectional, viz., from dioecy to hermaphroditism, and from hermaphroditism to dioecy, and that the aerial-submerged leaf habit and short-linear leaf shape are the ancestral states. Conclusions: Our study has shed light on the previously controversial generic phylogeny of Hydrocharitaceae. The study has resolved the historical biogeography of this family and supported dispersal as the most likely explanation for the intercontinental distribution. We have also provided valuable information for understanding the evolution of breeding system and leaf phenotype in aquatic monocots. ocharitaceae, Phylogeny, Historical biogeography, Dispersal, Vicariance, Morphological character Keywords: Hydrocharitaceae, Phylogeny, Historical biogeography, Dispersal, Vicariance, Morphological character evolution Chen et al. BMC Evolutionary Biology 2012, 12:30 http://www.biomedcentral.com/1471-2148/12/30 Chen et al. BMC Evolutionary Biology 2012, 12:30 http://www.biomedcentral.com/1471-2148/12/30 Chen et al. BMC Evolutionary Biology 2012, 12:30 http://www.biomedcentral.com/1471-2148/12/30 Page 2 of 12 rbcL and fossil calibrations. Notably, in that study the family was represented by only 16 terminals and there was no internal calibration point. Abstract This could be improved by better sampling and by incorporating inter- nal fossil calibration points [12]. Kato et al (2003) [13] dated the seagrasses within Hydrocharitaceae at 119 ± 11 Ma by analyses using the substitution rates of rbcL and matK. However, this time overlaps with the gener- ally accepted age of the order Alismatales thus putting the validity of the results of that study into doubt [14]. He et al. (1991) [15] proposed that Ottelia had origi- nated no later than the Cretaceous based on the distri- bution of the genus and the predictions of the continental drift theory. Fossils of Hydrocharitaceae have been found in Europe including those of the extant genera Vallisneria, Hydrilla, Ottelia, Thalassia, Stra- tiotes, Hydrocharis and Najas from the Eocene, Oligo- cene and Miocene [16-18]. The oldest fossil of Hydrocharitacae (genus Stratiotes) is from the Late Paleocene [19]. These fossils have prior to the present study not been incorporated in divergence time estimates. Background Further- more, Stratiotes has been resolved as sister to the rest of this family by analyses using rbcL [9] and mitochondrial genes (nad5, cob, ccmB, mtt2, atp1) [9,10]. The position is not in agreement with the results of two other studies using chloroplast and nuclear sequences [1,2]. In addi- tion, a single species was selected for each genus in pre- vious studies. The phylogeny of this family could be improved by a better sampling of taxon and DNA sequence, and perhaps a careful outgroup selection [1]. The transoceanic distribution of angiosperms has long intrigued biologists. Two competing hypotheses have been proposed to explain this phenomenon: the first attributing it to dispersal [21-23] and the second to vicariance (continental drift) [24]. Les et al. (2003) [21] proposed that dispersal is the major factor accounting for the disjunctive distribution of aquatic plants. This is contrary to the traditional viewpoint which considered vicariance [24] as the major cause of the disjunctive dis- tribution in aquatic taxa such as Limnocharitaceae [25], Ottelia (Hydrocharitaceae) [15] and Sagittaria (Alisma- taceae) [26]. Hydrocharitaceae exhibits a wide transocea- nic distribution at genera and species levels. The family can serve as a suitable model to investigate transoceanic distribution in aquatic monocots. About 10% of all flowering plants have unisexual flow- ers which have traditionally been regarded as a derived state in angiosperms [27]. Most species of Hydrocharita- ceae are unisexual while some are hermaphrodite. Her- maphroditism is regarded as the ancestral condition which gave rise to unisexual flowers [28]. However, her- maphroditic flowers also occur in more recently evolved genera such as Ottelia [1]. This suggests that the view The divergence time of Hydrocharitaceae is still a sub- ject of debate, and two competing ages (one much more recent than the other) have been proposed. Janssen and Bremer (2004) [11] placed the crown node age of this family in the Late Cretaceous (75 Ma) by analyses using Chen et al. BMC Evolutionary Biology 2012, 12:30 http://www.biomedcentral.com/1471-2148/12/30 Page 3 of 12 (BS = 96, PP = 1.0) formed by Najas, Hydrilla, Necha- mandra, Vallisneria and Maidenia (Figure 1). Clade B (BS = 100, PP = 1.0) consisted of Lagarosiphon, Ottelia, Blyxa, Apalanthe, Elodea and Egeria (Figure 1). The two outgroups selected resulted in slightly different support values, but identical topologies of Hydrocharitaceae. that hermaphroditism is the ancestral state needs to be re-examined. Dating analysis and ancestral area reconstruction Dating analysis and ancestral area reconstruction All BEAST MCMC runs yielded sufficient effective sam- ple sizes (> 200) for all relevant parameters and con- verged on topologies identical to the tree in Figure 1. The crown node age of Hydrocharitaceae was dated at 65.2 Ma (95% HPD: 54.6-79.6 Ma). The mean diver- gence between clade A and clade B, estimated to be 63.1 Ma. For the three calibration nodes, mean posterior estimates were consistent with prior node ages, suggest- ing that the calibration points were sufficiently concor- dant [29,30]. In this study, we sampled 17 genera of Hydrocharita- ceae, using DNA sequences from 8 genes. The aims of our study were: 1) to reappraise the generic relation- ships of Hydrocharitaceae; 2) to estimate the divergence times; 3) to investigate the area of origin and the major factor(s) shaping the global distribution; and 4) to inves- tigate the evolution of reproductive system and leaf morphology in this family. Sequence characteristics Seven biogeographic areas were recognized according to Morse [31] (Figure 2a) namely A, Nearctic area; B, Neotropical area; C, West Palearctic area; D, Afrotropi- cal area; E, Oriental area; F, Australasian area; and G, East Palearctic area. Two strategies were applied in the biogeographic analyses. One used genera as terminal taxa, the other used species as terminal taxa. Results of the analyses using genera as terminal taxa suggested that the most recent common ancestors (MRCAs) of both Hydrocharitaceae and clade A occurred in Oriental area. The MRCA of clade B occurred in Oriental, Afro- tropical and Neotropical areas (Figure 2b). Details of the voucher and DNA sequence information are provided in Additional file 1. We successfully gener- ated 117 sequences. Other sequences used in this study were downloaded from GenBank. Seventeen of the 18 genera in Hydrocharitaceae were sampled. Genus Apper- tiella was not included because of unavailability of molecular data. The supermatrix dataset, which resulted from assembling the DNA sequences of the 8 genes, was 8,086 nt in length (2,035 mt, 4,406 chl, 1,645 nu). The dataset consisted of 38 terminals of Hydrocharita- ceae and was submitted to TreeBASE (accession number S12110). The dataset comprised about 29% missing characters mainly due to unavailability of some sequences. Results of the biogeographic analyses using species as terminal taxa suggested that Hydrocharitaceae origi- nated in Orient (Figure 2c). A minimum of 76 dispersal events was inferred from DIVA to explain the current distribution of Hydrocharitaceae. The ancestor of Stra- tiotes was suggested to have been in Orient and dis- persed to Europe during the Late Cretaceous and Paleocene (Figure 2c, d, arrow 1). This route is similar to that which has been reported for Alangiopollis (Alan- giaceae) [32]. The analyses using genera or species as terminal taxa yielded comparable results on the ances- tral area of Hydrocharitaceae. This indicates that incom- plete sampling may have little effect on the accuracy of investigation into the geographical origin of Hydrochari- taceae. However, the analyses using species as terminal successfully resolved the ancestral areas for more genera than those using genera as terminal. Background Leaf habit in Hydrocharitaceae varies from aerial, aerial-submerged to fully submerged, and leaf shape varies from circular, linear, to ribbon like [1]. Ancestral state reconstruction is useful in understanding the evolutionary history of reproductive system and leaf morphology and their significance in adapting the plants to the aquatic environment. Phylogenetic analyses Butomus and (Butomus + Alisma (Alismataceae) + Cym- odocea (Cymodoceaceae) + Hydrocleys (Limnocharita- ceae) + Potamogeton (Potamogetonaceae)) were independently selected as outgroup. Both Maximum likelihood (ML) and Bayesian analysis using the super- matrix dataset resulted in completely identical relation- ships and strong support (bootstrap value (BS)> 95, Bayesian posterior probability (PP) = 1.0) for most branches (Figure 1). ML analysis involving partitioning the supermatrix dataset into eight genes and no parti- tion resulted in slightly different support values, but identical topologies. Stratiotes was resolved as the first diverging lineage of Hydrocharitaceae with strong sup- port (BS = 100, PP = 1.0). Other genera formed two clades (BS = 99, PP = 1.0; Figure 1). Clade A (BS = 95, PP = 1.0) included 10 genera. Hydrocharis-Limnobium was resolved as the first diverging lineage of this clade; the seagrasses formed a well supported clade (BS = 100, PP = 1.0) which was resolved as sister to the subclade The MRCAs of clade A, the seagrass genera and the sea- grass subclade were shown to be of Oriental origin. The taxa then diversified in Oriental region (DIVA, Mesquite; Figure 2c). DIVA suggested that Vallisneria originated in Oriental and Australasian regions. Mesquite suggested an Oriental origin for Vallisneria, followed by diversification and dispersal to other continents (Figure 2c). Long Page 4 of 12 Chen et al. BMC Evolutionary Biology 2012, 12:30 http://www.biomedcentral.com/1471-2148/12/30 Figure 1 Phylogeny and divergence time estimates of Hydrocharitaceae based on combined 18S + rbcL+ matK+ trnK 5’ intron + rpoB + rpoC1 + cob + atp1 data set. Numbers above branches refer to the maximum likelihood bootstrap values (BS, left) and the posterior probabilities (PP, right). Numbers in blue refer to the branches with BS> 95 and PP = 1.0. Gray bars indicate 95% highest posterior distributions, and nodes labelled with stars refer to the positions of fossil calibration points. Figure 1 Phylogeny and divergence time estimates of Hydrocharitaceae based on combined 18S + rbcL+ matK+ trnK 5’ intron + rpoB + rpoC1 + cob + atp1 data set. Numbers above branches refer to the maximum likelihood bootstrap values (BS, left) and the posterior probabilities (PP, right). Numbers in blue refer to the branches with BS> 95 and PP = 1.0. Gray bars indicate 95% highest posterior distributions, and nodes labelled with stars refer to the positions of fossil calibration points. Phylogenetic analyses The ancestor of clade B was inferred to have dispersed from Orient to Afrotropical region (Mesquite) or South- ern hemisphere (DIVA) during the Eocene (Figure 2c, d, arrow 2), followed by diversification in Southern hemi- sphere during the Tertiary. MRCA of Ottelia was sug- gested to have lived in Oriental and Afrotropical regions (Figure 2c). LDD from West Africa into South America for this genus was inferred from O. brasiliensis (Figure 2c, e, arrow 9). This route is similar to the one sug- gested for the dispersal of Gossypium (Malvaceae) [35] and is further supported by the fact that Ottelia in South America is confined to the southeastern area [36]. LDD from S.E. Asia to Australasia was inferred from O. ovalifolia (Figure 2c, e, arrow 5). Dispersal from Africa into Asia was also inferred from O. cordata and O. mesenterium (Figure 2c, e, arrow 3). This route is simi- lar to that envisaged for the two genera Coccinia and distance dispersal (LDD) for Vallisneria from Oriental area to Africa and Europe was inferred from V. spinulosa, V. spiralis and V. denseserrulata (Figure 2c, e, arrow 1 & 2). LDD from Australasia to North America for this genus was inferred from V. neotropicalis and V. americana (Fig- ure 2c, e, arrow 7). This route is similar to that envisaged for the taxon Scaevola (Goodeniaceae) [33]. LDD from Australasia to Asia similar to that which has been recorded for the plant family Cucurbitaceae [34] was inferred from taxa including V. asiatica and V. natans (Figure 2c, e, arrow 4). The genus Najas was inferred to have originated in Oriental area during the Oligocene. LDD from Asia to North America (inferred from N. gracil- lima), to Europe (inferred from N. minor), to Africa (inferred from N. minor) and to Australia (inferred from subclade N. browniana + N. tenuifolia) was suggested (Fig- ure 2c, e, arrow 6, 1, 2 & 5). distance dispersal (LDD) for Vallisneria from Oriental area to Africa and Europe was inferred from V. spinulosa, V. spiralis and V. denseserrulata (Figure 2c, e, arrow 1 & 2). LDD from Australasia to North America for this genus was inferred from V. neotropicalis and V. americana (Fig- ure 2c, e, arrow 7). This route is similar to that envisaged for the taxon Scaevola (Goodeniaceae) [33]. Phylogenetic analyses LDD from Australasia to Asia similar to that which has been recorded for the plant family Cucurbitaceae [34] was inferred from taxa including V. asiatica and V. natans (Figure 2c, e, arrow 4). The genus Najas was inferred to have originated in Oriental area during the Oligocene. LDD from Asia to North America (inferred from N. gracil- lima), to Europe (inferred from N. minor), to Africa (inferred from N. minor) and to Australia (inferred from subclade N. browniana + N. tenuifolia) was suggested (Fig- ure 2c, e, arrow 6, 1, 2 & 5). distance dispersal (LDD) for Vallisneria from Oriental area to Africa and Europe was inferred from V. spinulosa, V. spiralis and V. denseserrulata (Figure 2c, e, arrow 1 & 2). LDD from Australasia to North America for this genus was inferred from V. neotropicalis and V. americana (Fig- ure 2c, e, arrow 7). This route is similar to that envisaged for the taxon Scaevola (Goodeniaceae) [33]. LDD from Australasia to Asia similar to that which has been recorded for the plant family Cucurbitaceae [34] was inferred from taxa including V. asiatica and V. natans (Figure 2c, e, arrow 4). The genus Najas was inferred to have originated in Oriental area during the Oligocene. LDD from Asia to North America (inferred from N. gracil- lima), to Europe (inferred from N. minor), to Africa (inferred from N. minor) and to Australia (inferred from subclade N. browniana + N. tenuifolia) was suggested (Fig- ure 2c, e, arrow 6, 1, 2 & 5). Chen et al. BMC Evolutionary Biology 2012, 12:30 http://www.biomedcentral.com/1471-2148/12/30 Page 5 of 12 Figure 2 Historical biogeography of Hydrocharitaceae. (a) Area delimitation in biogeographic analyses. (b) Analyses using genera as terminal taxa, and (c) analyses using species as terminal taxa. Distribution of each genus or species is indicated along with taxon names. Results of Mesquite are indicated by coloured circles; results of dispersal-vicariance analysis (DIVA) are indicated by capital letters (only the results of major clades are shown); equally optimal ancestral distributions are indicated by pie-charts (Mesquite) or slashes (DIVA). The times inferred from divergence time estimates were marked on the major clades, numbers represent millions of years before present. (d) & (e) Possible origin, differentiation centres and dispersal routes of Hydrocharitaceae. mitation in biogeographic analyses. (b) Analyses using genera as term h genus or species is indicated along with taxon names. Phylogenetic analyses Results of e analysis (DIVA) are indicated by capital letters (only the results of m by pie-charts (Mesquite) or slashes (DIVA). The times inferred from Figure 2 Historical biogeography of Hydrocharitaceae. (a) Area delimitation in biogeographic analyses. (b) Analyses using genera as terminal taxa, and (c) analyses using species as terminal taxa. Distribution of each genus or species is indicated along with taxon names. Results of Mesquite are indicated by coloured circles; results of dispersal-vicariance analysis (DIVA) are indicated by capital letters (only the results of major clades are shown); equally optimal ancestral distributions are indicated by pie-charts (Mesquite) or slashes (DIVA). The times inferred from divergence time estimates were marked on the major clades, numbers represent millions of years before present. (d) & (e) Possible origin, differentiation centres and dispersal routes of Hydrocharitaceae. Chen et al. BMC Evolutionary Biology 2012, 12:30 http://www.biomedcentral.com/1471-2148/12/30 Page 6 of 12 Page 6 of 12 Page 6 of 12 Chen et al. BMC Evolutionary Biology 2012, 12:30 http://www.biomedcentral.com/1471-2148/12/30 Page 6 of 12 Momordica in family Cucurbitaceae [34]. The MRCA of the subclade comprising Apalanthe, Egeria and Elodea was suggested to have arisen in Neotropical region, while the ancestor of this subclade may have come from Afrotropical region (Figure 2c, e, arrow 9). Dispersal from South America to North America during the Mio- cene was inferred for the ancestor of Elodea, probably via the island chains of Central America (Figure 2c, e, arrow 10). Discussion Systematics By merging diverse sequences into a supermatix data set, we obtained a well-resolved phylogeny with most branches strongly supported by BT values greater than 95% and PP values of 1.0. This indicates that increasing the number of taxa and the number of molecular mar- kers improved the resultant phylogeny, and it further supports the notion that a supermatrix can be used to obtain a well-resolved and strongly supported phylogeny in cases where some data are missing [37-39]. The phy- logenetic relationships of Egeria, Elodea, Ottelia, Blyxa, Apalanthe and Lagarosiphon have remained unchanged in all the earlier molecular phylogenetic studies [1,2,8,9,40,41]. Our analyses resolved the generic rela- tionships that are largely similar to those reported in those studies. However, incongruences existed for the other genera, which we briefly address here below. Ancestral character state reconstructions Morphological characters of each species used in analy- sis were indicated in Figure 3. Dioecy was suggested as the progenitorial state, monoecy and hermaphroditism were derived from it (Figure 3a); Ottelia emersa and O. acuminata experienced reverse evolution from her- maphroditism to dioecy (Figure 3a). The aerial-sub- merged leaf habit was suggested as the progenitorial state, which gave rise to aerial leaf and submerged leaf (Figure 3b); the aerial-submerged leaf in O. ovalifolia and O. emersa was originated from the submerged leaf due to reverse evolution (Figure 3b). The short-linear leaf shape was suggested as the ancestral state, ribbon like and broad-ribbon like leafs were derived from it (Figure 3c); the broad-circular leaf evolved indepen- dently in three lineages, viz. Hydrocharis-Limnobium, Ottelia and Halophila. The genus Hydrilla comprises only one species, H. verticillata. Based on rbcL, matK, trnK intron, ITS and morphological data, Les et al. (2006) [1] suggested that Hydrilla was most closely related to Najas, despite their being quite divergent at the phenotypic level. Our phy- logenetic analyses suggested that Hydrilla is most clo- sely related to the subclade comprising Nechamandra, Vallisneria and Maidenia (Figure 1). This position is consistent with all previous phylogenetic studies (except Figure 3 Reconstruction of ancestral character states by Mesquite. Figure 3 Reconstruction of ancestral character states by Mesquite. Chen et al. BMC Evolutionary Biology 2012, 12:30 http://www.biomedcentral.com/1471-2148/12/30 Page 7 of 12 Chen et al. BMC Evolutionary Biology 2012, 12:30 http://www.biomedcentral.com/1471-2148/12/30 Les et al., 2006) based on molecular and morphological data [2,8,9,40,41]. A close relationship between the sub- clade (Hydrilla (Nechamandra (Vallisneria + Maidenia)) and Najas was strongly supported (BS = 96, PP = 1.0, Figure 1). This is in agreement with earlier results from rbcL+ matK+ trnK intron analyses with ML approach [1], rbcL [9], and rbcL+ matK [8]. However, the results did not support the close affinity between the subclade and seagrasses which had been inferred from rbcL [2] and rbcL+ matK+ trnK intron + ITS analysis [1]. the 100 Ma inferred from analyses using the substitution rates of rbcL and matK [13]. These discrepancies indi- cate that for estimating divergence times in aquatic plants, incorporating fossil calibration point would be more reliable. Ancestral character state reconstructions The Oriental origin of Hydrocharitaceae inferred from our analysis is supported by the known existence of regions with humid and warm conditions in southeast- ern Asia during the late Cretaceous and Palaeocene [46,47] and the fact that the genetic diversity centre of this family is in tropical Asia [20]. The ancestor of clade A was inferred to have originated and diversified in the Orient, while that of clade B dispersed from the Orient to the Southern Hemisphere during the Late Cretaceous and Paleocene (Figure 2c, d, arrow 2). Different environ- ments and oceanic barriers among the major continents (vicariance mechanism) during the Tertiary probably contributed to the diversification of this family resulting in taxa such as the African endemic Lagarosiphon. Stratiotes was resolved as the first diverging lineage within Hydrocharitaceae (BS = 100, PP = 1.0; Figure 1). This is in agreement with analyses based on rbcL [9], mtt2 and nad5 [10] and the fossil records of Stratiotes which include the most abundant and the oldest fossils of’ Hydrocharitaceae [19,42]. However, the phylogenetic position of Stratiotes seems to be mainly derived from the mitochondrial sequences (cob, atp1) which are prone to flaws in plant phylogenetic analysis [43]. Therefore, further studies are required to confirm the position of the genus obtained in this study. Most fossils of Hydrocharitaceae and its close relatives Butomaceae and Alismataceae have been found in Eur- ope (Butomaceae in the Neogene of south Aral region, Miocene of northwest and east Caucasus [18,48]; Alis- mataceae in the Tertiary of Europe, a few in North America). The fossil records seem to be inconsistent with the Oriental origin of this family. However, the absence of reports of fossils from Asia most likely reflects a bias in paleobotany, rather than an indication of the origin and past distribution of Hydrocharitaceae. A similar situation exists in Rhinolophus (Rhinolophi- dae), for which, although the genus is thought to have originated in Asia, fossils have only been reported from Europe and Africa but not from Asia [49,50]. Origin Results of divergence time estimates are in agreement with the fossil records of Hydrocharitaceae. The 95% HPD of Najas was 11.9-34.3 Ma, consistent with the oldest fossil of this genus in the Oligocene [17] (Figure 1). The stem node age of Hydrocharis-Limnobium was dated around 54.7 Ma. However, the crown node age of this subclade was dated around 15.9 Ma, younger than the oldest fossil of Hydrocharis from the Upper Eocene [17,44,45]. This could be interpreted as an indication that Limnobium had split from the relatively ancient Hydrocharis in the Miocene (Figure 1), and the great morphological similarity between the two genera is probably due to the short evolutionary history of Limno- bium. Although the present study has yielded improved divergence time estimates, it is possible that the esti- mates of the time of origin for some genera such as Ottelia, Vallisneria, Najas and Blyxa may have been affected by under-representation in sampling. Does the origin fit with dispersal? Does the origin fit with dispersal? The modern continents viz. South America, Africa, Eur- asia, Australia and North America have been separated by oceans since at least ca. 90 Ma [46,51,52], earlier than the origin of Hydrocharitaceae. Therefore, dispersal must have played a dominant role in the transoceanic distribution of this family. This contradicts the view that the transoceanic distribution of Ottelia mainly resulted from vicariance [15]. The role of dispersal in transocea- nic distribution has been supported by evidences from the studies of geological events and land plant families. Ocean currents are a viable means of dispersal of plants [53], and a tropical westward-flowing ocean current had spanned the world from the Cretaceous to Paleocene [54,55]. Island chains existed in the Tethys from Cretac- eous to Eocene, which served as a stepping-stone in bio- tic dispersal between S.E. Asia, Africa and southern Europe [32,56,57]. The Malay Archipelago probably facilitated biotic dispersal between S.E. Asia and Austra- lia during the Miocene [58]. The North Atlantic Land The modern continents viz. South America, Africa, Eur- asia, Australia and North America have been separated by oceans since at least ca. 90 Ma [46,51,52], earlier than the origin of Hydrocharitaceae. Therefore, dispersal must have played a dominant role in the transoceanic distribution of this family. This contradicts the view that the transoceanic distribution of Ottelia mainly resulted from vicariance [15]. The role of dispersal in transocea- nic distribution has been supported by evidences from the studies of geological events and land plant families. The age of Hydrocharitaceae estimated in this study (mean: 65.5 Ma, 95% HPD: 54.6-79.6 Ma) is in agree- ment with that based on rbcL analysis and external fossil calibration points (crown node age = 75 Ma) [11]. How- ever, the stem node age of seagrasses estimated in this study (15.9-41.3 Ma) (Figure 1), is more recent than the 119 ± 11 Ma suggested from analysis using the substitu- tion rates of rbcL and matK [13]. Similarly, our esti- mates of the stem node age of Ottelia (8.1-33.3 Ma) is more recent than the Cretaceous origin suggested by He et al. (1991) [15]. The split between Zosteraceae and Potamogetonaceae has been dated at 47 Ma by rbcL and fossil calibration [11]. The time is also more recent than Chen et al. Evolution of morphological characters Ancestral state reconstruction of reproductive system in Hydrocharitaceae provides empirical evidence that evo- lution of dioecy in plants has been a bidirectional, viz. from dioecy to hermaphroditism, and from hermaphro- ditism to dioecy (Figure 3a). This view is supported by Delph (2009) [81] and Canovas et al. (2011) [82], but rejects the view that hermaphroditism is the ancestral state in Hydrocharitaceae [28]. Does the origin fit with dispersal? BMC Evolutionary Biology 2012, 12:30 http://www.biomedcentral.com/1471-2148/12/30 Page 8 of 12 Bridge (NALB) aided plant migration between North America and Europe during the late Cretaceous and early Tertiary [59-61]. The Bering Land Bridge (BLB) was open from at least the early Paleocene until its clo- sure ca. 7.4-4.8 Ma [62]. Several recent studies of angiosperms based on molecular and fossil data have supported dispersal as the dominant factor responsible for transoceanic distribution, e.g., in Cucurbitaceae [34], Sapindaceae [63], Chrysophylloideae (Sapotaceae) [64], Burseraceae [65] and Malphigiaceae [52]. It is probable that Hydrocharitaceae have dispersed to all continents of the world via island chains, land bridges and ocean currents. were suggested to have been dispersed from Oriental to other regions (Figure 2c), probably by ocean currents [73,77]. For example, the warm northward Kuroshio Current carried seagrasses from the equatorial region to the Nansei Islands [77]. Seagrasses are capable of surviv- ing during the LDD between major ocean systems [78]. Vallisneria has a world-wide distribution, with the highest number of species in Australia [79,80]. Les et al. (2008) [80] resolved the phylogeny of this genus, but they conceded that the geographical origin is difficult to pinpoint. In this study by DIVA analysis, Oriental and Australasian areas were suggested as the co-existed ancestral areas of Vallisneria. However, Oriental area is more likely the centre of origin considering the follow- ing facts: the closest relative of Vallisneria namely Nechamandra is confined to Asia [67]; the ancestral species in Vallisneria namely Vallisneria spinulosa, V. spiralis and V. denseserrulata are confined to the Old World [67,80]. Biogeographic studies have suggested that the sub-cos- mopolitan distribution of the aquatic plant family Alis- mataceae has mainly resulted from dispersals (the work will be reported in a separate paper). It is probable that dispersal is the dominant factor, accounting for transo- ceanic distribution of aquatic angiosperms. However, more studies on aquatic angiosperms are required to investigate this idea further. Historical biogeography of some genera of Hydrocharitaceae The ancestor of Stratiotes was suggested to have dis- persed from Orient into Europe during the late Cretac- eous and Palaeocene (Figure 2b &2c, 2d, arrow 1), which coincided with the existence of the Tethys seaway (TESW) [54]. Alternatively, the ancestor may have migrated from Orient to Europe across Eurasia. Abun- dant fossils (15 fossil species) of this genus in Europe [42] suggested that the genus had diversified widely in this region adapting to wet swamps in the Late Cretac- eous [66]. The evolution of leaf habit and leaf shape in Hydro- charitaceae provides several cases of evolutionary adap- tation to diverse habitats. The evolution from aerial- submerged leaf to submerged leaf is probably due to change in habitat from shallow to deep waters [83]. The reverse evolution from submerged leaf to aerial-sub- merged leaf in Ottelia is probably an adaptation to change in habitat from deep to shallow water or some other disadvantageous habitat(s). Taxa with broad-circu- lar leaves (e.g., Ottelia and Hydrocharis) usually occur in still water, while those with ribbon like leaves such as Enhalus and Thalassia occur in coastal waters with strong waves [67]. The genus Hydrilla is native to Eurasia and Australia [67], and introduced to Americas [68] and parts of Africa [69]. The centre of differentiation of the genus was thought by Cook and Luond (1982) to lie in tropical Asia [70]. This idea got support from genetic diversity analysis which revealed that the highest diversity is located in China and with lower albeit similar genetic types occurring in Africa, India and USA [71]. Hydrilla might have arisen in the Orient dispersing to Europe and Australia (Figure 2, e, arrow 1 & 5). Sampling and molecular protocols Most materials used for DNA sequencing was collected from Wuhan Botanical Garden. Some were collected from natural populations in China. Eight genes were used, among which 18S is from nuclear; rbcL, matK, trnK5’ intron, rpoB and rpoC1 are from chloroplasts; cob and atp1 are from mitochondria. A detailed list including the voucher information and GenBank acces- sion numbers is provided in Additional file 1. Genomic DNA was extracted from silica-dried leaves using the Plant Genomic DNA Isolation Kit (Dingguo Biotech, Beijing, China). All polymerase chain reactions (PCR) were conducted in the ABI 2720 Thermal Cycler (Applied Biosystems) in 40 μl volume containing 4 μl of 10 × amplification buffer (200 mM Tris-HCl (pH 8.4), 200 mM KCl, 100 mM (NH4)2SO4, 20 mM MgSO4), 0.8 μl of each primer (10 μM), 0.8 μl of dNTPs (10 mM), 2 U of Taq DNA polymerase (TransGen Biotech Co., Beij- ing, China) and 60 ng of DNA template. For cob, atp1, 18S, trnK 5’ intron, rpoB and rpoC1, the following PCR profile was adopted: 94°C for 3 min, 35 cycles of 30 s at 94°C, 30 s at 50°C, 1 min at 72°C and a final step for 10 min at 70°C. For matK, 55°C Tm and 1 min 30 s exten- sion times were used. Primer sequences were obtained from previous studies: cob (COB1F, COB1R) and atp1 (atpAF1.5, atpAl137r) [9]; rbcL (1 F, 1204R) [84]; matK [7]; trnK 5’ intron (3914-F, TRANK2-R) [85]; 18S (N- NS1, C-18 L) [86]; rpoB (1f, 4r) and rpoC1 (2f, 4r) [87]. Purified PCR products were double direction sequenced using an automated DNA sequencer (ABI 3730, Applied Biosystems). All newly generated sequences were depos- ited in GenBank (Additional file 1). Conclusions The MRCA of the seagrasses within Hydrocharitaceae were suggested to have lived in Oriental area during the Oligocene and Miocene. The result is in agreement with the view that seagrasses possibly originated in the S.E. Asia [72,73]. The result is supported by the environ- ments of S.E. Asia which was characterized by abundant islands, spacious shallow-seas, warm temperature and plenty of isolated seas [74]. However, the result denied the Cretaceous origin of the group which has been sug- gested in previous studies [13,75,76]. The seagrasses In summary, this study has reconstructed the phylogeny of Hydrocharitaceae. The family was suggested to origi- nate in Oriental area during the Late Cretaceous and Paleocene (54.7-72.6 Ma). Dispersal is the most likely factor shaping the transoceanic distribution of this family. Ancestral character state reconstruction of gen- der and leaf morphology offered valuable information for understanding adaptive evolution in aquatic mono- cots. However, the historical biogeography for some genera (e.g., Ottelia, Vallisneria) may suffer from under- Chen et al. BMC Evolutionary Biology 2012, 12:30 http://www.biomedcentral.com/1471-2148/12/30 Page 9 of 12 representation in sampling, and require re-evaluation in future studies. partition the data set. GTRCAT and GTRGAMMA options were used, 1000 rapid bootstrap replicates were conducted to assess bootstrap values. Phylogenetic analyses All sequences were aligned individually using Clustal X v2.0 [88]. The output was manually inspected, and ambiguously aligned parts were excluded. Inspired by other studies (e.g. [37,39,89,90]), we assembled all the aligned sequences into a supermatrix data set (combined data set), which was used in phylogenetic analyses. Buto- mus (Butomaceae) was used as outgroup according to Les et al. (2006) [1]. We also selected (Butomus + Alisma + Cymodocea + Hydrocleys + Potamogeton) as outgroup in order to investigate the influence of out- group in topology of Hydrocharitaceae. ML analysis was conducted using RAxML v7.2.5 [91] via the Cyberinfrastructure for Phylogenetic Research (CIPRES) Portal http://www.phylo.org. Two strategies were employed, one involved partitioning the superma- trix data set into eight genes, while another did not Divergence time estimates Divergence time estimates were conducted in BEAST v1.5.4 [95] using the supermatrix data set. Butomus, Alisma, Cymodocea, Hydrocleys and Potamogeton were selected as outgroup. To prevent the negative effects from heterogeneity of substitution rates and uncertainty of fossil data, we used a relaxed clock and Uncorrelated Lognormal (UCLN) model [96,97]. GTR + I + G model with Gamma Categories set to 6 was adopted. The start- ing tree was randomly generated with a Yule process prior. More than 90,000,000 generations of MCMC were implemented of which every 1,000 generations were sampled. Tracer v1.5 [94] was used to check the parameters and the first 10% generations were discarded as burn-in. Lognormal distribution was selected for each calibra- tion point according to Adamson et al. (2010) [98]. This distribution defined the minimum ages for calibrated nodes but allowed the maximum ages to be estimated following a lognormal distribution without hard limit [99]. Three calibration points were incorporated. The oldest reliable fossils of Hydrilla and Vallisneria were reported from the Upper Eocene (33.7-55.8 Ma) [17,44]. Therefore, the split between Hydilla and (Vallisneria + Nechamandra) was constrained to a minimum of 33.7 Ma (offset = 33.7, mean = 1.1, SD = 1.2). In addition, the oldest reliable fossil of Ottelia was from the Upper Eocene (33.7-55.8 Ma) [17,44]. Therefore, the split between Blyxa and Ottelia was constrained to a mini- mum of 33.7 Ma (offset = 33.7, mean = 1.1, SD = 1.2). Lastly, the oldest fossil of this family (genus Stratiotes) was 0.1 Ma younger than the Paleocene-Eocene bound- ary (54.6 Ma) [19]). Therefore, the fossil was used to set the split between Stratiotes and the remaining genera of this family not later than 54.5 Ma (offset = 54.5, mean = 1.0, SD = 1.0). Methods Bayesian analysis was conducted in MrBayes v3.1.2 [92]. The best-fit model of nucleotide substitution was chosen by MrModeltest v2.3 [93] according to the Akaike Information Criterion (combined data set: GTR + I + G model). Two separate runs of four concurrent runs (one cold, three heated each) of 16,000,000 genera- tions were employed with sampling at every 1,000 gen- erations. The stationarity of the likelihood scores of sampled trees was evaluated in Tracer v1.5 [94], and the first 10% generations were discarded as burn-in. Appendix A Appendix A Additional file 1 Taxa included in this study with vou- cher information and GenBank accession numbers (DOC 99 kb) 10. Cuenca A, Petersen G, Seberg O, Davis JI, Stevenson DW: Are substitution rates and RNA editing correlated? BMC Evol Biol 2010, 10:349. 10. Cuenca A, Petersen G, Seberg O, Davis JI, Stevenson DW: Are substitution rates and RNA editing correlated? BMC Evol Biol 2010, 10:349. 11. Janssen T, Bremer K: The age of major monocot groups inferred from 800 + rbcL sequences. Bot J Linn Soc 2004, 146:385-398. 11. Janssen T, Bremer K: The age of major monocot groups inferred from 800 + rbcL sequences. Bot J Linn Soc 2004, 146:385-398. Competing interests The authors declare that they have no competing interests. The authors declare that they have no competing interests. Received: 4 January 2012 Accepted: 10 March 2012 Published: 10 March 2012 Received: 4 January 2012 Accepted: 10 March 2012 Published: 10 March 2012 Received: 4 January 2012 Accepted: 10 March 2012 Published: 10 March 2012 Biogeographic analyses Seven biogeographic areas were recognized according to Dr Morse [31] (Figure 2a). Biogeographic distribution of Page 10 of 12 Page 10 of 12 Chen et al. BMC Evolutionary Biology 2012, 12:30 http://www.biomedcentral.com/1471-2148/12/30 Hydrocharitaceae was mainly compiled from literature [36,67,72,78,80,100-106]. Fossil was not considered in the area coding as no fossil has been found outside the natural distribution of any genus. Distribution that is known to have been caused by human activities was not included in the analyses. Two methods were used in the analyses: dispersal-vicariance analysis implemented in DIVA v1.2 [107] with the maximum number of ances- tral areas at each node constrained to four, and parsi- mony ancestral state reconstruction implemented in Mesquite [108]. References 1. Les DH, Moody ML, Soros CL: A reappraisal of phylogenetic relationships in the monocotyledon family Hydrocharitaceae (Alismatidae). Aliso 2006, 22:211-230. 2. Les DH, Cleland MA, Waycott M: Phylogenetic studies in Alismatidae, II: Evolution of marine angiosperms (seagrasses) and hydrophily. Syst Bot 1997, 22:443-463. 3. He JB: Systematic Botanical and Biosystematic Studies on Ottelia in China. 1 edition. Wuhan: Wuhan University Press; 1991. 3. He JB: Systematic Botanical and Biosystematic Studies on Ottelia in China. 1 edition. Wuhan: Wuhan University Press; 1991. y 4. Chambers PA, Lacoul P, Murphy KJ, Thomaz SM: Global diversity of aquatic macrophytes in freshwater. Hydrobiologia 2008, 595:9-26. 4. Chambers PA, Lacoul P, Murphy KJ, Thomaz SM: Global diversity of aquatic macrophytes in freshwater. Hydrobiologia 2008, 595:9-26. 5. Les DH, Moody ML, Jacobs SWL: Phylogeny and systematics of Aponogeton (Aponogetonaceae): The Australian species. Syst Bot 2005, 30:503-519. 5. Les DH, Moody ML, Jacobs SWL: Phylogeny and systematics of Aponogeton (Aponogetonaceae): The Australian species. Syst Bot 2005, 30:503-519. Author details 1 1CAS Key Laboratory of Aquatic Botany and Watershed Ecology, Chinese Academy of Sciences, Wuhan 430074, Hubei, P. R. China. 2Wuhan Botanical Garden, Chinese Academy of Sciences, Wuhan 430074, Hubei, P. R. China. 3Graduate University of Chinese Academy of Sciences, 100049 Beijing, China. 4Botany Department, Jomo Kenyatta University of Agriculture and Technology, P. O. Box 62000-00200, Nairobi, Kenya. Two strategies were applied in the biogeographic ana- lyses. One strategy used genera as terminal taxa in the analyses. A tree (Figure 2b) that represented the generic topologies inferred from the phylogenetic analysis using the supermatrix (Figure 1) was constructed, and it was used in the analyses. Each genus was coded based on the current distribution. Details of the distribution are provided in Figure 2b. This strategy followed the meth- ods in the biogeography study of Ranunculaceae [23] and the suggestion from Yan Yu (one of the authors of S-DIVA). Another strategy used species as terminal taxa and a tree including 72 species of Hydrocharitaceae was used (Figure 2c). Species belonging to Ottelia [15], Halophila [78] and Vallisneria [80] with known phylo- genetic relationships were manually added to the tree which resulted from the phylogenetic analysis (Figure 1). In addition, seven species of Najas with the topology from a ML analysis based on rbcL were also added to the tree. Each species was coded based on the current distribution. Details of the distribution are provided in Figure 2c. The purpose of using species as terminal taxa was to reconstruct the ancestral areas at the family and genus levels. Acknowledgements We thank Mao-Xian He, Chun-Feng Yang, Shu-Ying Zhao, Fan Liu, Zhi-Yuan Du, Kuo Liao, Zhong-Ming Ye and Wen-Kui Dai for help in the field work. Ji Yang, Yu-Guo Wang and Yong-Qing Zhu assisted in data analyses. We thank Can Dai and Jun Wen for revising the manuscript. We thank Yang-Zhou Wang and Zhao-Bo Li for comments on this manuscript. We are grateful to Prof Miguel Vences and the two anonymous reviewers for valuable suggestions on this manuscript. This work received financial support from the Foundation of the Chinese Academy of Sciences granted to Qing-Feng Wang (KSCX2-YW-Z-0805) and the National Natural Science Foundation of China (No. 30970202 and 30970195). Authors’ contributions LYC participated in design of the study, carried out the experiment work, performed data analyses and drafted the manuscript. JMC participated in design of the study and helped to draft the manuscript. GWR revised the manuscript. QFW conceived the study, revised the manuscript and gave final approval of the version to be published. All authors read and approved the final manuscript. Ancestral character state reconstructions Information on the reproductive system, leaf habit and leaf shape was mainly compiled from literature [1,36,67,103]. Details of the phenotypic data are pro- vided in Figure 3. Parsimony ancestral state reconstruc- tion was performed using the Mesquite [108] and the tree inferred from the phylogenetic analyses using the supermatrix dataset. 6. Les DH, Haynes RR: Systematics of subclass Alismatidae: a synthesis of approaches. In Monocotyledons: Systematics and Evolution. Edited by: Rudall PJ, Cribb PJ, Cutler DF, Humphries CJ. Kew: Royal Botanic Gardens; 1995:353-377. 6. Les DH, Haynes RR: Systematics of subclass Alismatidae: a synthesis of approaches. In Monocotyledons: Systematics and Evolution. Edited by: Rudall PJ, Cribb PJ, Cutler DF, Humphries CJ. Kew: Royal Botanic Gardens; 1995:353-377. 7. Lehtonen S, Myllys L: Cladistic analysis of Echinodorus (Alismataceae): simultaneous analysis of molecular and morphological data. Cladistics 2008, 24:218-239. 7. Lehtonen S, Myllys L: Cladistic analysis of Echinodorus (Alismataceae): simultaneous analysis of molecular and morphological data. Cladistics 2008, 24:218-239. 8. Tanaka N, Setoguchi H, Murata J: Phylogeny of the family Hydrocharitaceae inferred from rbcL and matK gene sequence data. J Plant Res 1997, 110:329-337. 8. Tanaka N, Setoguchi H, Murata J: Phylogeny of the family Hydrocharitaceae inferred from rbcL and matK gene sequence data. J Plant Res 1997, 110:329-337. 9. Petersen G, Seberg O, Davis JI, Stevenson DW: RNA editing and phylogenetic reconstruction in two monocot mitochondrial genes. Taxon 2006, 55:871-886. 9. Petersen G, Seberg O, Davis JI, Stevenson DW: RNA editing and phylogenetic reconstruction in two monocot mitochondrial genes. Taxon 2006, 55:871-886. Additional material 12. Ramirez SR, Gravendeel B, Singer RB, Marshall CR, Pierce NE: Dating the origin of the Orchidaceae from a fossil orchid with its pollinator. Nature 2007, 448:1042-1045. 12. Ramirez SR, Gravendeel B, Singer RB, Marshall CR, Pierce NE: Dating the origin of the Orchidaceae from a fossil orchid with its pollinator. Nature 2007, 448:1042-1045. Additional file 1: Taxa included in this study with voucher information and GenBank accession numbers (DOC 99 kb). 13. Kato Y, Aioi K, Omori Y, Takahata N, Satta Y: Phylogenetic analyses of Zostera species based on rbcL and matK nucleotide sequences: Page 11 of 12 Chen et al. BMC Evolutionary Biology 2012, 12:30 http://www.biomedcentral.com/1471-2148/12/30 41. Chen JM, Chen D, Gituru WR, Wang QF, Guo YH: Evolution of apocarpy in Alismatidae using phylogenetic evidence from chloroplast rbcL gene sequence data. Bot Bull Acad Sin 2004, 45:33-40. Implications for the origin and diversification of seagrasses in Japanese waters. Genes Genet Syst 2003, 78:329-342. 14. Magallon S, Castillo A: Angiosperm Diversification through Time. Am J Bot 2009, 96:349-365. 42. Cook CDK, Urmikonig K: A Revision of the Genus Stratiot 42. Cook CDK, Urmikonig K: A Revision of the Genus St (Hydrocharitaceae). Aquat Bot 1983, 16:213-249. 15. He JB, Sun XZ, Zhong Y, Huang DS: Cladistic studies on the genus Ottelia (Hydrocharitaceae). J Wuhan Bot Res 1991, 9:121-129. (Hydrocharitaceae). Aquat Bot 1983, 16:213-249. 43. Hiesel R, von Haeseler A, Brennicke A: Plant mitochondrial nucleic acid sequences as a tool for phylogenetic analysis. Proc Natl Acad Sci USA 1994, 91:634-638. 16. Kvacek Z: Bilina: a window on Early Miocene marshland environments. Rev Palaeobot Palynol 1998, 101:111-123. y 17. Stockey RA: The fossil record of basal monocots. Aliso 2006, 22:91-106. 44. Mai DH, Walther H: Die Floren der Haselbacher Serie im Weisselster- Becken (Bezirk Leipzig, DDR). Abhand-lungen des Staatlichen Museums für Mineralogie und Geologie zu Dresden 1978, 28:1-101. 18. Daghlian CP: A review of the fossil record of monocotyledons. Bot Rev 1981, 47:517-555. 19. Sille NP, Collinson ME, Kucera M, Hooker JJ: Morphological evolution of Stratiotes through the Paleogene in England: An example of microevolution in flowering plants. Palaios 2006, 21:272-288. 45. Mai DH, Walther H: Die obereozänen Floren des Weisselsterbeckens und seiner Randgebiete. Abhandlungen des Staatlichen Museums für Mineralogie und Geologie zu Dresden 1985, 33:1-260. g 46. Markwick PJ: Paul’s Palaeo Pages 2011 [http://www.palaeogeography.net]. 20. Cook CDK: Hydrocharitaceae. In Flowering Plants, Monocotyledons: Alismatanae and Commelinanae (except Gramineae). Additional material Edited by: Kubitzki K, Huber H. Berlin: Springer; 1998:234-247. 47. Okada H: Mateer NJ: Cretaceous Environments of Asia. Netherlands: Elsevier Science B.V; 2000. p g 21. Les DH, Crawford DJ, Kimball RT, Moody ML, Landolt E: Biogeography of discontinuously distributed hydrophytes: A molecular appraisal of intercontinental disjunctions. Int J Plant Sci 2003, 164:917-932. 48. Takhtajan A, Vakrameev V: Radchenko GP: Fundamentals of paleontology: Gymnosperms and angiosperm. Moscow, Russia: Akadamiia Nauk S.S.S.R; 1963. 22. Emadzade K, Horandl E: Northern Hemisphere origin, transoceanic dispersal, and diversification of Ranunculeae DC. (Ranunculaceae) in the Cenozoic. J Biogeogr 2011, 38:517-530. 49. Stoffberg S, Jacobs DS, Mackie IJ, Matthee CA: Molecular phylogenetics and historical biogeography of Rhinolophus bats. Mol Phylogenet Evol 2010, 54:1-9. 23. Emadzade K, Gehrke B, Linder HP, Horandl E: The biogeographical history of the cosmopolitan genus Ranunculus L. (Ranunculaceae) in the temperate to meridional zones. Mol Phylogenet Evol 2011, 58:4-21. 50. Stoffberg S: Rhinolophus capensis (Chiroptera: Rhinolophidae). Mamm Species 2008, 810:1-4. 51. Lomolino MV, Riddle BR, Brown JH: Biogeography. 3 edition. Sunderland: Sinauer Associates; 2006. 24. Raven PH, Axelrod DI: Angiosperm biogeography and past continental movements. Ann Mo Bot Gard 1974, 61:539-673. 52. Davis CC, Bell CD, Mathews S, Donoghue MJ: Laurasian migration explains Gondwanan disjunctions: Evidence from Malpighiaceae. Proc Natl Acad Sci USA 2002, 99:6833-6837. 5. Haynes RR, Holm-Nielsen LB: The Limnocharitaceae. In Flora Neo 25. Haynes RR, Holm-Nielsen LB: The Limnocharitaceae. In Flora Neotropica. Volume 56. New York: New York Botanical Garden Press; 1992:1-32. 53. Renner S: Plant dispersal across the tropical Atlantic by wind and sea currents. Int J Plant Sci 2004, 165:23-33. 26. Chen JK: Systematic and Evolutionary Botanical Studies on Chinese Sagittaria. 1 edition. Wuhan: Wuhan University Press; 1989. 1 edition. Wuhan: Wuhan University Press; 1989. 54. Bush ABG: Numerical simulation of the Cretaceous Tethys circumglobal current. Science 1997, 275:807-810. 27. Barrett SCH: The evolution of plant sexual diversity. Nat Rev Genet 2002, 3:274-284. 55. Gordon WA: Marine life and ocean surface currents in the Cretaceous. J Geol 1973, 81:269-284. 28. Sculthorpe CD: The Biology of Aquatic Vascular Plants London: Edward Arnold; 1967. 29. Sanders KL, Mumpuni , Hamidy A, Head JJ, Gower DJ: Phylogeny and divergence times of filesnakes (Acrochordus): Inferences from morphology, fossils and three molecular loci. Mol Phylogenet Evol 2010, 56:857-867. 56. Tiffney BH, Manchester SR: The use of geological and paleontological evidence in evaluating plant phylogeographic hypotheses in the Northern Hemisphere tertiary. Additional material Int J Plant Sci 2001, 162:S3-S17. 57. Heaney LR: Conservation biogeography in oceanic archipelagoes. In Frontiers of Biogeography: New directions in the Geography of Nature. Edited by: Lomolino MV. Sunderland: Sinauer Associates; 2004:345-360. 30. Sanders KL, Lee MSY: Evaluating molecular clock calibrations using Bayesian analyses with soft and hard bounds. Biol Lett 2007, 3:275-279. 30. Sanders KL, Lee MSY: Evaluating molecular clock calibrations using Bayesian analyses with soft and hard bounds. Biol Lett 2007, 3:275-279. 31. Morse JC: Trichoptera World Checklist 2011 [http://entweb.clemson.edu/ database/trichopt]. Bayesian analyses with soft and hard bounds. Biol Lett 2007, 3:275-279. 31. Morse JC: Trichoptera World Checklist 2011 [http://entweb.clemson.edu/ database/trichopt]. 58. Van Welzen PC, Slik JWF, Alahuhta J: Plant distribution patterns and plate tectonics in Malesia. Biol Skr 2005, 55:199-217. 31. Morse JC: Trichoptera World Checklist 2011 [http://entweb.clemson.edu/ database/trichopt]. 59. Tiffney BH: The Eocene North Atlantic land bridge: its importance in Tertiary and modern phytogeography of the Northern Hemisphere. J Arnold Arbor 1985, 66:243-273. 32. Feng CM, Manchester SR, Xiang QY: Phylogeny and biogeography of Alangiaceae (Cornales) inferred from DNA sequences, morphology, and fossils. Mol Phylogenet Evol 2009, 51:201-214. 60. Denk T, Grimsson F, Zetter R: Episodic migration of oaks to Iceland: Evidence for a North Atlantic “land bridge” in the latest Miocene. Am J Bot 2010, 97:276-287. 33. Howarth DG, Gustafsson MHG, Baum DA, Motley TJ: Phylogenetics of the genus Scaevola (Goodeniaceae): implication for dispersal patterns across the Pacific Basin and colonization of the Hawaiian Islands. Am J Bot 2003, 90:915-923. 61. Mao K, Hao G, Liu J, Adams RP, Milne RI: Diversification and biogeography of Juniperus (Cupressaceae): variable diversification rates and multiple intercontinental dispersals. New Phytol 2010, 188:254-272. 34. Schaefer H, Heibl C, Renner SS: Gourds afloat: a dated phylogeny reveals an Asian origin of the gourd family (Cucurbitaceae) and numerous oversea dispersal events. Proc R Soc B Biol Sci 2009, 276:843-851. 62. Marincovich L, Gladenkov AY: Evidence for an early opening of the Bering Strait. Nature 1999, 397:149-151. 35. Cronn R, Wendel JF: Cryptic trysts, genomic mergers, and plant speciation. New Phytol 2004, 161:133-142. 63. Buerki S, Forest F, Alvarez N, Nylander JAA, Arrigo N, Sanmartin I: An evaluation of new parsimony-based versus parametric inference methods in biogeography: a case study using the globally distributed plant family Sapindaceae. J Biogeogr 2011, 38:531-550. 36. Cook CDK, Urmikonig K: A revision of the genus Ottelia (Hydrocharitaceae). 2. The species of Eurasia, Australasia and America. 66. Col J: Enchanted Learning 2011 [http://www.enchantedlearning.com/ subjects/dinosaurs/mesozoic/cretaceous/lc.shtml]. Chen et al. BMC Evolutionary Biology 2012, 12:30 http://www.biomedcentral.com/1471-2148/12/30 Additional material Madeira PT, Coetzee JA, Center TD, White EE, Tipping PW: The origin of 69. Madeira PT, Coetzee JA, Center TD, White EE, Tipping PW: The origin of Hydrilla verticillata recently discovered at a South African dam. Aquat Bot 2007, 87:176-180. 95. Drummond AJ, Rambaut A: BEAST: Bayesian evolutionary analysis by sampling trees. BMC Evol Biol 2007, 7:214. Hydrilla verticillata recently discovered at a South African dam. Aquat Bot 2007, 87:176-180. 70. Cook CDK, Luond R: A revision of the genus Hydrilla (Hydrocharitaceae). Aquat Bot 1982, 13:485-504. 96. Drummond AJ, Ho SY, Phillips MJ, Rambaut A: Relaxed phylogenetics and dating with confidence. PLoS Biol 2006, 4:e88. 97. Smith SA, Beaulieu JM, Donoghue MJ: An uncorrelated relaxed-clock analysis suggests an earlier origin for flowering plants. Proc Natl Acad Sci USA 2010, 107:5897-5902. 71. Overholt WA, Copeland R, Williams D, Cuda J, Nzigidahera B, Nkubaye E, Wanda F, Gidudu B: Exploration for natural enemies of Hydrilla verticillata in East/Central Africa and genetic characterization of worldwide populations. Report to the Florida Department of Environmental Protection Tallahassee: Bureau of Invasive Plant Management; 2008. analysis suggests an earlier origin for flowering plants. Proc Natl Acad Sci USA 2010, 107:5897-5902. 98. Adamson EAS, Hurwood DA, Mather PB: A reappraisal of the evolution of Asian snakehead fishes (Pisces, Channidae) using molecular data from multiple genes and fossil calibration. Mol Phylogenet Evol 2010, 56:707-717. 72. Short F, Carruthers T, Dennison W, Waycott M: Global seagrass distribution and diversity: A bioregional model. J Exp Mar Biol Ecol 2007, 350:3-20. 73. Mukai H: Biogeography of the tropical seagrasses in the western Pacific. Mar Freshw Res 1993, 44:1-17. 99. Popp M, Mirre V, Brochmann C: A single Mid-Pleistocene long-distance dispersal by a bird can explain the extreme bipolar disjunction in crowberries (Empetrum). Proc Natl Acad Sci USA 2011, 108:6520-6525. 74. Benzie JAH: Genetic structure of marine organisms and SE Asian biogeography. In Biogeography and geological Evolution of SE Asia. Edited by: Hall R, Holloway JD. Netherlands: Backhuys Publishers; 1998:197-209. 100. Eiseman NJ, Mcmillan C: A new species of seagrass, Halophila johnsonii from the Atlantic coast of Florida. Aquat Bot 1980, 9:15-19. 75. McCoy ED, Heck KL: Biogeography of corals, seagrasses, and mangroves; an alternative to the center of origin concept. Syst Biol 1976, 25:201-210. 101. Johnstone IM: Papua-New-Guinea seagrasses and aspects of the biology and growth of Enhalus-acoroides (Lf) Royle. Aquat Bot 1979, 7:197-208. 76. Additional material Aquat Bot 1984, 20:131-177. 36. Cook CDK, Urmikonig K: A revision of the genus Ottelia 36. Cook CDK, Urmikonig K: A revision of the genus Ottelia (Hydrocharitaceae). 2. The species of Eurasia, Australasia and America. Aquat Bot 1984, 20:131-177. 37. McGowen MR, Spaulding M, Gatesy J: Divergence date estimation and a comprehensive molecular tree of extant cetaceans. Mol Phylogenet Evol 2009, 53:891-906. 37. McGowen MR, Spaulding M, Gatesy J: Divergence date estimation and a comprehensive molecular tree of extant cetaceans. Mol Phylogenet Evol 2009, 53:891-906. 64. Bartish IV, Antonelli A, Richardson JE, Swenson U: Vicariance or long- distance dispersal: historical biogeography of the pantropical subfamily Chrysophylloideae (Sapotaceae). J Biogeogr 2011, 38:177-190. 38. Wiens JJ, Moen DS: Missing data and the accuracy of Bayesian phylogenetics. J Syst Evol 2008, 46:307-314. 65. Weeks A, Daly DC, Simpson BB: The phylogenetic history and biogeography of the frankincense and myrrh family (Burseraceae) based on nuclear and chloroplast sequence data. Mol Phylogenet Evol 2005, 35:85-101. 39. Thomson RC, Shaffer HB: Sparse supermatrices for phylogenetic inference: taxonomy, alignment, rogue taxa, and the phylogeny of living turtles. Syst Biol 2010, 59:42-58. 39. Thomson RC, Shaffer HB: Sparse supermatrices for phylogenetic inference: taxonomy, alignment, rogue taxa, and the phylogeny of living turtles. Syst Biol 2010, 59:42-58. 66. Col J: Enchanted Learning 2011 [http://www.enchantedlearning.com/ subjects/dinosaurs/mesozoic/cretaceous/lc.shtml]. 66. Col J: Enchanted Learning 2011 [http://www.enchantedlearning.com/ subjects/dinosaurs/mesozoic/cretaceous/lc.shtml]. 40. Li XX, Zhou ZK: Phylogenetic studies of the core Alismatales inferred from morphology and rbcL sequences. Prog Nat Sci 2009, 19:931-945. 40. Li XX, Zhou ZK: Phylogenetic studies of the core Alismatales inferred from morphology and rbcL sequences. Prog Nat Sci 2009, 19:931-945. Page 12 of 12 Chen et al. BMC Evolutionary Biology 2012, 12:30 http://www.biomedcentral.com/1471-2148/12/30 67. Wang QF, Guo YH, Haynes RR, Hellquist CB: Hydrocharitaceae. In Flora of China. Volume 23. Edited by: Wu ZY, Peter HR. Beijing and St. Louis: Science Press and Missouri Botanical Garden Press; 2010:91-102. 92. Cronquist A: An integrated system of classification of flowering plants New York: Columbia University Press; 1981. 93. Nylander JAA: MrModeltest v2. Program distributed by the author. Evolutionary Biology Centre, Uppsala University, Sweden; 2004. 68. Haller WT, Sutton DL: Community structure and competition between Hydrilla and Vallisneria. Hyacinth Control J 1975, 13:48-50. Evolutionary Biology Centre, Uppsala University, Sweden; 2004 94. Rambaut A, Drummond AJ: Tracer v1.4 2007, Available from http://beast.bio. ed.ac.uk/Tracer. y y 69. Additional material Den HC: The sea-grasses of the world Amsterdam: North Holland Publishing Co; 1970. 102. Kenworthy WJ, Wyllie-Echeverria S, Coles R, Pergent G, Pergent-Martini C: Seagrass conservation biology: An interdisciplinary science for protection of the seagrass biome. In Seagrasses: Biology, Ecology and Conservation. Edited by: Larkum AWD, Orth RJ, Duarte CM. Netherlands: Springer; 2006:595-623. 77. Kuo J, Kanamoto Z, Iizumi H, Aioi K, Mukai H: Seagrasses from the Nansei Islands, Southern Japanese Archipelago: species composition, distribution and biogeography. Mar Ecol 2006, 27:290-298. g 103. Cook CDK: Aquatic plant book Hague: SPB Academic Publishing; 1990. 78. Waycott M, Freshwater DW, York RA, Calladine A, Kenworthy WJ: Evolutionary trends in the seagrass genus Halophila (thouars): Insights from molecular phylogeny. Bull Mar Sci 2002, 71:1299-1308. 104. Wingfield RA, Murphy KJ, Hollingsworth P, Gaywood MJ: The Ecology of Najas Flexilis. Scottish Natural Heritage Commissioned Report No. 017; 2004. 79. Jacobs SWL, Frank KA: Notes on Vallisneria (Hydrocharitaceae) in Australia, with descriptions of two new species. Telopea 1997, 7:111-118. 105. Gambi MC, Barbieri F, Bianchi CN: New record of the alien seagrass Halophila stipulacea (Hydrocharitaceae) in the western Mediterranean: A further clue to changing Mediterranean Sea biogeography. JMBA2 Biodiv Rec 2009, 2:e84. 80. Les DH, Jacobs SWL, Tippery NP, Chen L, Moody ML, Wilstermann- Hildebrand M: Systematics of Vallisneria (Hydrocharitaceae). Syst Bot 2008, 33:49-65. 81. Delph LF: Sex allocation: evolution to and from dioecy. Curr Biol 2009, 19: R249-R251. 106. Wang QF, Haynes RR, Hellquist CB: Alismataceae and Butomaceae. In Flora of China. Volume 23. Edited by: Wu ZY, Peter HR. Beijing and St. Louis: Science Press and Missouri Botanical Garden Press; 2010:84-90. 82. Canovas FG, Mota CF, Serrao EA, Pearson GA: Driving south: a multi-gene phylogeny of the brown algal family Fucaceae reveals relationships and recent drivers of a marine radiation. BMC Evol Biol 2011, 11:371. 107. Ronquist F: Dispersal-vicariance analysis: A new approach to the quantification of historical biogeography. Syst Biol 1997, 46:195-203. 83. Chen JM, Robert GW, Wang QF: Evolution of aquatic life-forms in Alismatidae: Phylogenetic estimation from chloroplast rbcL gene sequence data. Isr J Plant Sci 2004, 52:323-329. 108. Maddison WP, Maddison DR: Mesquite: a modular system for evolutionary analysis. Version 2.74 2010 [http://mesquiteproject.org]. doi:10.1186/1471-2148-12-30 Cite this article as: Chen et al.: Generic phylogeny, historical biogeography and character evolution of the cosmopolitan aquatic plant family Hydrocharitaceae. BMC Evolutionary Biology 2012 12:30. doi:10.1186/1471-2148-12-30 Cite this article as: Chen et al.: Generic phylogeny, historical biogeography and character evolution of the cosmopolitan aquatic plant family Hydrocharitaceae. BMC Evolutionary Biology 2012 12:30. Chen et al. BMC Evolutionary Biology 2012, 12:30 http://www.biomedcentral.com/1471-2148/12/30 Additional material doi:10.1186/1471-2148-12-30 Cite this article as: Chen et al.: Generic phylogeny, historical biogeography and character evolution of the cosmopolitan aquatic plant family Hydrocharitaceae. BMC Evolutionary Biology 2012 12:30. doi:10.1186/1471-2148-12-30 Cite this article as: Chen et al.: Generic phylogeny, historical biogeography and character evolution of the cosmopolitan aquatic plant family Hydrocharitaceae. BMC Evolutionary Biology 2012 12:30. 84. Zurawski G, Clegg M: Evolution of higher-plant chloroplast DNA-encoded genes: Implications for structure-function and phylogenetic studies. Annu Rev Plant Physiol 1987, 38:391-418. 85. Les DH, Schneider EL, Padgett DJ, Soltis PS, Soltis DE, Zanis M: Phylogeny, classification and floral evolution of water lilies (Nymphaeaceae; Nymphaeales): A synthesis of non-molecular, rbcL, matK, and 18S rDNA data. Syst Bot 1999, 24:28-46. y 86. Bult C, Kallersjo M, Suh Y: Amplification and sequencing of 16/18S rDNA from gel-purified total plant DNA. Plant Mol Biol Rep 1992, 10:273-284. 87. Sass C, Little DP, Stevenson DW, Specht CD: DNA barcoding in the cycadales: Testing the potential of proposed barcoding markers for species identification of cycads. PLoS One 2007, 2:e1154. Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color ﬁgure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit 88. Thompson JD, Gibson TJ, Plewniak F, Jeanmougin F, Higgins DG: The CLUSTAL_X windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res 1997, 25:4876-4882. Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color ﬁgure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit Submit your next manuscript to BioMed Central and take full advantage of: 89. Escobar JS, Scornavacca C, Cenci A, Guilhaumon C, Santoni S, Douzery EJP, Ranwez V, Glemin S, David J: Multigenic phylogeny and analysis of tree incongruences in Triticeae (Poaceae). BMC Evol Biol 2011, 11:181. 90. Geisler JH, McGowen MR, Yang G, Gatesy J: A supermatrix analysis of genomic, morphological, and paleontological data from crown Cetacea. BMC Evol Biol 2011, 11:112. 91. Stamatakis A: RAxML-VI-HPC: maximum likelihood-based phylogenetic analyses with thousands of taxa and mixed models. Bioinformatics 2006, 22:2688-2690.
W4205849962.txt	https://www.frontiersin.org/articles/10.3389/fvets.2021.782149/pdf	en		Lesion Material From Treponema-Associated Hoof Disease of Wild Elk Induces Disease Pathology in the Sheep Digital Dermatitis Model	Frontiers in veterinary science	2,022	cc-by	10,112	ORIGINAL RESEARCH published: 12 January 2022 doi: 10.3389/fvets.2021.782149 Lesion Material From Treponema-Associated Hoof Disease of Wild Elk Induces Disease Pathology in the Sheep Digital Dermatitis Model Jennifer H. Wilson-Welder 1, Kristin Mansfield 2 , Sushan Han 3 , Darrell O. Bayles 1 , David P. Alt 1 and Steven C. Olsen 1 1 Infectious Bacterial Diseases of Livestock Research Unit, National Animal Disease Center, Agricultural Research Service, United States Department of Agriculture (USDA), Ames, IA, United States, 2 Washington Department of Fish and Wildlife, Spokane Valley, WA, United States, 3 Colorado State University Diagnostic Medicine Center, Fort Collins, CO, United States Edited by: Georgios Oikonomou, University of Liverpool, United Kingdom Reviewed by: Nicholas James Evans, University of Liverpool, United Kingdom Jennifer Sarah Duncan, University of Liverpool, United Kingdom Correspondence: Jennifer H. Wilson-Welder jennifer.wilson-welder@usda.gov Specialty section: This article was submitted to Zoological Medicine, a section of the journal Frontiers in Veterinary Science Received: 23 September 2021 Accepted: 29 November 2021 Published: 12 January 2022 Citation: Wilson-Welder JH, Mansfield K, Han S, Bayles DO, Alt DP and Olsen SC (2022) Lesion Material From Treponema-Associated Hoof Disease of Wild Elk Induces Disease Pathology in the Sheep Digital Dermatitis Model. Front. Vet. Sci. 8:782149. doi: 10.3389/fvets.2021.782149 A hoof disease among wild elk (Cervus elaphus) in the western United States has been reported since 2008. Now present in Washington, Oregon, Idaho, and California, this hoof disease continues to spread among elk herds suggesting an infectious etiology. Causing severe lesions at the hoof-skin junction, lesions can penetrate the hoof-horn structure causing severe lameness, misshapen hooves, and in some cases, sloughed hooves leaving the elk prone to infection, malnutrition, and premature death. Isolated to the feet, this disease has been termed treponeme-associated hoof disease due to the numerous Treponema spp. found within lesions. In addition to the Treponema spp., treponeme-associated hoof disease shares many similarities with digital dermatitis of cattle and livestock including association with several groups of anaerobic bacteria such as Bacteroides, Clostridia, and Fusobacterium, neutrophilic inflammatory infiltrate, and restriction of the disease to the foot and hoof tissues. To determine if there was a transmissible infectious component to this disease syndrome, elk lesion homogenate was used in a sheep model of digital dermatitis. Ten animals were inoculated with lesion material and lesion development was followed over 7 weeks. Most inoculated feet developed moderate to severe lesions at 2- or 4-weeks post-inoculation timepoints, with 16 of 18 feet at 4 weeks also had spirochetes associated within the lesions. Histopathology demonstrated spirochetes at the invading edge of the lesions along with other hallmarks of elk hoof disease, neutrophilic inflammatory infiltrates, and keratinocyte erosion. Treponema-specific PCR demonstrated three phylotypes associated with elk hoof disease and digital dermatitis were present. Serum of infected sheep had increased anti-Treponema IgG when compared to negative control sheep and pre-exposure samples. Analysis of the bacterial microbiome by sequencing of the bacterial 16S rRNA gene showed a community structure in sheep lesions that was highly similar to the elk lesion homogenate used as inoculum. Bacteroidies, Fusobacterium, and Clostridia were Frontiers in Veterinary Science \| www.frontiersin.org 1 January 2022 \| Volume 8 \| Article 782149 Wilson-Welder et al. TAHD in Sheep Model among the bacterial taxa overrepresented in infected samples as compared to negative control samples. In conclusion, there is a highly transmissible, infectious bacterial component to elk treponeme-associated hoof disease which includes several species of Treponema as well as other bacteria previously associated with digital dermatitis. Keywords: Treponema, hoof disease, disease transmission, model, wildlife, elk, sheep, Cervus elaphus INTRODUCTION be necessary to fulfill or demonstrate causality of an etiologic factor to onset of disease. To study hoof diseases in controlled settings, various models of naturally occurring hoof diseases have been used. Most involve housing animals with wet bedding, using boots, bandages, or a system of wraps to macerate the skin and hold bacterial cultures or lesion material in place (11–15). Recently, cell-based models have even been proposed but appear to be limited to evaluating pathogenesis of a single bacterial species at a time (16, 17). Models for DD of cattle have been developed using lengthy macerations (12) or initial abrasion of the skin to provide bacterial inoculum for a site for penetration and lesion production (13, 15). All models require extensive handling of the animal and their feet. However, under research settings the safety of care staff and animals must be considered. Large cervids, including elk (Cervus elaphus) and white-tailed deer (Odocoileus virginianus), can be raised in captivity and have been used in infectious disease research (18–22). Although stress and injury can occur in animals not acclimated to human interaction, confined spaces, or handling facilities (23), cervids, even captive bred animals, appear especially susceptible. Injury remains a leading cause of mortality among captive cervids (24– 26). Large animals also pose an injury risk to human handlers (27). Even though our facility excels at handling cervids under research conditions (19, 20), the cumulative risk to animals and people is high when doing frequent manipulations of potential fractious elk for development of a disease model. Fortunately, a sheep model of bovine DD provides a viable alternative to large animal models (15). In comparison to larger animal models, sheep are easier to handle, are less expensive to maintain, and previous work has demonstrated that Treponemadriven bovine DD can be reproduced in a sheep model (15). The primary objective of the current study was to determine if material from TAHD lesions in free-roaming wild elk of the western U.S. would induce similar lesions in a sheep model. In our attempt to fulfill Koch’s postulates, we used elk lesion tissue homogenate in the ovine model, characterized primary lesions, and examined lesion material for evidence of Treponema species as previously reported in elk TAHD lesions. In 2008 increased numbers of limping elk (Cervus elaphus) and an increased presence of deformed hoofs were observed in southwestern Washington State. Throughout the following decade, cases were confirmed all along the western range of the Cascade mountains, further north into Washington, throughout neighboring Oregon state, and as far south as northern California. Recently, cases have been confirmed in far eastern Washington state, and adjacent Idaho, representing movement of this disease across to the eastern side of the Cascade mountains. Animals presented as severely lame, often with overgrown, misshaped, or completely sloughed hooves. Reports rapidly increased over a short period of time radiating from the region where lesions were first observed (1). Preliminary investigations indicated an infectious agent as the etiology as other systemic, metabolic, or environmental causes could not be identified (1). Bacterial species isolated from the hoof lesions were similar to those present in polymicrobial digital dermatitis (DD) of livestock (2, 3). Histologic examination of affected tissues revealed spirochete bacteria at the leading/invading edge of the lesions. Treponema isolates from elk hoof lesions were determined to be the same three phylotypes as isolated from DD lesions of cattle and sheep (2, 4). The disease spread outward rapidly from a single area of origin, but there was a lack of other etiologic factors. The current hypothesis is that this elk hoof disease, called Treponemeassociated hoof disease (TAHD), is a transmissible bacterial disease, and most likely poly-bacterial (2, 3). Koch’s postulates, greatly simplified, would be that utilization of a disease-causing organism in healthy naïve hosts will reproduce similar clinical signs and subsequently allow re-isolation of the infectious agent from lesions. The Bradford Hill criteria, more widely used in epidemiology fields, contains five criteria for examining a disease and causation: (1) strength of association, (2) consistency, (3) specificity, (4) temporality, and (5) plausibility (5). In both TAHD and DD of livestock, several bacteria have been suspected to be included in the disease pathogenesis, but demonstration of a precise etiology remains elusive. To date, several species of Treponema and other anaerobic bacteria are consistently associated with DD in livestock lesions (6). Other variables, such as wet hoof environment, seasonality, high stocking rates or population density, and nutritional or metabolic stress brought on by parturition, early lactation, or poor forage contribute to susceptibility of cattle to DD, small ruminants to contagious ovine digital dermatitis, or elk to TAHD (1, 3, 7–10). The ability to model or reproduce all cofactors in an animal model may not Frontiers in Veterinary Science \| www.frontiersin.org METHODS Ethical Statement All animal procedures were approved by National Animal Disease Center Institutional Animal Care and Use Committee in accordance with the standards established by: Public Health Service Policy “US Government Principles for the Utilization and Care of Vertebrate Animals Used in Testing, Research, 2 January 2022 \| Volume 8 \| Article 782149 Wilson-Welder et al. TAHD in Sheep Model All experimental groups (inoculated and mock-inoculated) were housed together for the duration of the study. and Training,” the Guide for the Care and Use of Laboratory Animals (National Research Council), the Animal Welfare Act (1966), and the Guide for the Care and Use of Agricultural Animals in Research and Teaching (USDA, Federation of Animal Science Societies). Experimental Overview The experiment used 15 sheep total, two groups of 5 inoculated/infected sheep elk TAHD lesion-derived material and 5 mock-inoculated, negative controls. Two hind feet were used on each sheep and lesion development was evaluated at 2-, 4-, and 7weeks post-inoculation. Group sizes and infectivity time (weeks PI) were based on previous data using bovine DD lesion material in which a < 90% infectivity rate was observed, and average time of lesion development was ∼4 weeks PI (15). Lesion Material Collection Lesion material was collected from four feet (three hind and one front foot) of three adult animals with clinical signs of TAHD within the known endemic area (Hoquiam and Oakville, Washington) by Washington Department of Fish and Wildlife biologists as part of humane removal programs. The four hooves were characterized as lesion grades 3, 2, 4, and 4 in accordance with a previously described scoring system (3) (Figure 1). Hooves were shipped on ice packs overnight to our laboratory in Ames, IA, where lesion tissue homogenates were prepared from active lesion sites as previously described (15, 28). Briefly, debris was removed from hooves using sterile phosphate buffered saline solution (PBS, pH 7.2), and sections of lesions were removed using sterile instruments. After transfer to an anaerobic chamber (Coy Laboratory Products, Grass Lake, MI), tissue was homogenized in Oral Treponeme Enrichment Broth (OTEB, Anaerobe Systems, Morgan Hill, CA). Sterile glycerol and serum (bovine, horse, and rabbit serum mixed in 2:1:1 ratio, Gibco) were added to tissue homogenate and filtered through several layers of sterile gauze. Spirochetes were numerated by direct count method and adjusted to approximately 1 × 107 per ml. Lesion homogenate was frozen at −80◦ C in 5 ml aliquots for later use. Lesion material from a front and rear hoof of one animal collected in June 2017 was used to prepare material designated as lesion homogenate WA I. Material collected from the hind feet of two different animals in October 2017 was used and the combined lesion homogenates derived from a single hoof from each animal was designated as lesion homogenate WA II. Preparation and Wrapping of Inoculation Site Feet were prepared and wrapped as previously described (15). Three days prior to inoculation, hair and wool on both hind legs were clipped to above the hock. Sheep were restrained and inverted using a sheep-specific tilt-table (Sydell Inc., Burbank, SD), allowing access to feet and legs. Two regions, each approximately 1.5 cm2 , on each side of the pastern area, were abraded using a rotary tool with tungsten bit. Cotton cast padding, ∼40 x 40 x 10 mm, was saturated with a mixture of 50% phosphate buffered saline (PBS) (pH 7.2) and 50% clarified sterile bovine rumen fluid and placed over the abraded area. A piece of small tubing with syringe port (SureFlo R Winged Infusion Set with needle portion removed, Terumo Corporation, Tokyo, Japan) was sandwiched in duct tape and placed so that the tip was over the cotton padding. Cotton and tubing were secured with a layer of brown “cling” gauze (Jorgensen Laboratories, Loveland, CO) followed by plastic film applied to retain moisture and exclude air. Next, a layer of Vetwrap R (3M Animal Care Products, St. Paul, MN) was applied for cushioning followed by an overlapping layer of Gorilla Tape R (Gorilla Glue, Cincinnati, OH). An additional round of Vetwrap R was used to secure and cover the injection port of the tubing. Wraps were checked twice per week for integrity. Gorilla Tape R was reapplied as needed. Every 7 days ± 2 days, 5 ml of 50% Anaerobe Basal Broth (Oxoid, Basinstoke, UK), 25% clarified bovine rumen fluid, and 25% PBS was added via the catheter port to maintain the wet, anaerobic environment in the pastern region. Three days following initial application of wet wraps, inoculation with lesion material or sterile OTEB media was performed once through a small incision made at the bottom of the wrap. One ml of prepared inoculum was placed under the cotton padding via 1 mL slip-tip syringe and the wrap was re-sealed with tape. Study Design Fifteen white-faced crossbred sheep (Ovis aries) ranging in age from 6 to 12 months were obtained from a local source. Sheep included both castrated males and virgin females. Before initiation of the study, hooves were trimmed and examined for signs of disease or other abnormalities and sheep were evaluated for soundness. Sheep were housed in an outdoor concrete paddock with deep straw bedding under the covered half of the paddock, with free access to water and grass hay. In order to emulate the low nutritional status of elk in the greater Mount Saint Helens area (1), no concentrate or grain was fed during the study. Body condition was monitored by visual inspection or by palpation of the backbone and assessment using a five-point scale (http://smallfarms.oregonstate.edu; accessed 12/28/15). Animals were maintained at body condition scores ≥ 2. Sheep were nonrandomly assigned to three treatment groups: those receiving lesion homogenate WA I and those receiving lesion homogenate WA II, and a mock process control group. To ensure variances in body size/ages were equally distributed across treatment groups, sheep were sorted largest to smallest by weight and every third sheep was selected for the mock-inoculated group. Frontiers in Veterinary Science \| www.frontiersin.org Clinical Observations Sheep were observed twice daily, and feet were monitored no less than twice each week for development of clinical signs including lameness (altered gait or non-weight bearing posture), evidence of heat or tenderness on palpation, and the presence of atypical odor associated with the bandaged foot. Frequency of lameness observations increased as clinical signs developed. Meloxicam (1-1.25 mg/kg, orally, every 24 h) was given for analgesia once animals demonstrated non-weight bearing lameness in at least one hind limb and was continued as long as lameness was 3 January 2022 \| Volume 8 \| Article 782149 Wilson-Welder et al. TAHD in Sheep Model FIGURE 1 \| Elk TAHD lesions used to generate lesion homogenate inoculum. (A) Lesion grade 3 from hind foot of adult elk cow. (B) Lesion grade 2 from front hoof of adult elk cow, same animal as in (C). (C) Lesion grade 4 from hind foot of adult elk cow. (D) Lesion grade 4 from hind foot of adult elk cow. processed, paraffin embedded, and 5 µm thick sections were stained with hematoxylin and eosin (H&E). Adjacent 5 µm sections were stained using Steiner and Steiner silver stain for visualization of bacteria. Slides were examined by a pathologist blinded to the identity of the study groups. Observations were made for each individual foot, with both feet from the same animal considered to be not independent observations. observed or until euthanasia. Personnel assessing clinical signs and administering analgesics were not blinded to the treatment group of the individual animals. Both feet from all animals were unwrapped at 2 and 4 weeks PI and observed for lesion development. Photographs were taken of feet for documentation of lesion development. Photographs of feet observed at 2, 4, and 7 weeks PI were scored as no lesions, mild, moderate, or severe. Mild lesions were defined as a lesion area that was less than the original scarified area, often with a decrease in erythema and evidence of healing. Moderate lesions were defined as lesions or erosions that were equivalent to the original scarified area and extending into an adjacent pastern area, dewclaw area, coronary band, or heel bulb or hoof sole but not covering > 30% of the total surface area of pastern/dewclaw area, heel bulb, or hoof sole. Severe lesions were defined as being inclusive of the above and encompassing two or more areas and >30% of the total surface area of pastern/dewclaw area, heel bulb, or sole covered in lesion or erosions. Lesion development observations were made on the foot level, with the assumption that 2 feet on the same animal were not independent observations. Wraps were reapplied. Animals were euthanized at 7 weeks PI and hoof samples collected. PCR To test lesions for the presence of treponemes associated with DD and TAHD (3), swabs of lesion sites were collected at 2, 4, and 7 weeks PI. After removal of all bandages, a sterile, dry cotton swab (Puritan Medical Products Company, Guilford, ME) was rubbed over the abrasion or lesion for 15 s. Resulting swabs were placed inside individual sterile polystyrene tubes and transported to the laboratory. The tip of the cotton swab was cut off using sterile scissors and DNA extracted using Qiagen DNeasy Blood and Tissue kit (Qiagen, Germantown, MD) following the manufacturer’s instructions for bacteria. Due to the absorption by the cotton, an additional 200 µl of lysing buffer (ATL) and 20 µl of proteinase K were added during the initial lysis step in the first incubation. DNA from swabs from each foot were maintained separately and evaluated on the individual foot level. DNA was quantified using QubitTM Fluorometric Quantitation (ThermoFisher, Waltham, MA). PCR for Treponema medium, Treponema pedis, Treponema phagedenis, Fusobacterium necrophorum, and Dichelobacter nodosus was performed in a two-stage nested PCR assay using primers and conditions as previously described (15) and presented in Supplementary Table S1. Positive controls were genomic DNA from each of the respective bacteria, details of strains used are given in Supplementary Table S1, and negative control was PCR reaction mixture with sterile water. PCR reaction products were visualized by electrophoresis on 1.5% agarose gel stained with SYBR Safe DNA gel stain (Invitrogen Life Technologies, Carlsbad, CA). Histology One strip per foot, ∼10 mm wide from the base of the dewclaw, through the center heel bulb to the center sole, was excised, encompassing lesions and adjacent healthy skin, and were preserved in 10% neutral buffered formalin. After fixation, the entire tissue sample (skin and attached keratinized hoof horn) was de-calcified with Kristensen’s decalcifying formic acid solution (50% 1 N sodium formate, 50% 8 N formic acid). Decalcification was monitored through chemical testing (29). Briefly, decalcifying solution was tested each day for completion as follows. The decalcifying solution (5 ml) was neutralized with 0.5 N sodium hydroxide, and then 1 ml of 5% ammonium oxalate was added. If the resulting solution became turbid, decalcification was not complete, and tissues were placed in fresh Kristensen’s solution. This was repeated each day until the chemical test was no longer turbid, indicating decalcification was complete. Once decalcification was complete, hoof/tissues were soaked in Fibrous Tissue Soaking Solution (100 ml Tween 80 in 1,000 ml 1N hydrochloric acid) (30). Tissues were tested daily for softness (pliability). Depending on the size of the tissues, this process took between 3 and 10 days. At this time, tissues were routinely Frontiers in Veterinary Science \| www.frontiersin.org Bacterial Observation Microscopic observations: At 2, 4, or 7 weeks PI, sterile cotton swabs of the lesion sites as described above were placed in sterile 5-ml polystyrene tubes and transported to the laboratory. Swabs from each foot were maintained separately and evaluated on the individual foot level. Approximately 0.5 ml sterile PBS was added to each tube and incubated at room temperature with periodic 4 January 2022 \| Volume 8 \| Article 782149 Wilson-Welder et al. TAHD in Sheep Model al. and https://mothur.org/wiki/miseq_sop/ (accessed January 2021) (32, 33). The 16S rRNA gene v4 sequences were aligned to the Silva bacteria reference database (release 132) and taxonomy was assigned based on the mothur-formatted version of the RDPtraining set (trainset16_022016). Sequences were clustered into OTUs based on the pairwise distances between the unique sequences. Further data analyses and statistical comparisons were made using R and the following R-packages “DESeq2,” “metacoder,” “microbial,” “OTUbase,” “phyloseq,” “taxa,” and “vegan.” The analyses also employed the following general Rpackages: “dplyr,” “ggplot2,” “gplots,” “magrittr,” “RColorBrewer,” “reshape,” “scales,” “tidyverse,” and “VennDiagram” (https:// CRAN.R-project.org/package=microbial) (https://CRAN.Rproject.org/package=vegan) (32, 34–37). The data were rarefied before comparative analyses except for the DESeq2 analysis that was performed with the unrarefied data. vortex for 1 h, then 10 µl was placed on a clean glass slide and viewed using dark-field microscopy to examine for the presence of spirochete-shaped bacteria. Enzyme-Linked Immunosorbent Assay Whole blood (8-10 ml) was collected via jugular venipuncture into serum separator vacutainer tubes (BD Vacutainer SST, Franklin Lakes, NJ) prior to inoculation (week 0) and 6 weeks PI. Blood was allowed to clot, and serum was separated by centrifugation (Beckman Coulter Avanti J-E with JS5.3 rotor, 700 × G, 20 min, 4◦ C), then stored at −20◦ C. T. phagedenis, T. medium, and T. pedis were grown in OTEB to confluence, bacterial cells harvested and washed by centrifugation, and antigen prepared from whole-cell sonicates as described previously (15). Mixed treponeme antigen (6 µg whole cell dry weight/ml in PBS) was bound overnight to a 96-well plate (Corning R 96-well Clear Flat Bottom Polystyrene High Bind Microplate #9018). Plates were blocked with 45 µg/ml porcine IgG (Sigma-Aldrich, St. Louis, MO) in PBS with 0.05% (v/v) Tween 20 (Sigma-Aldrich, St. Louis, MO) (PBST) overnight. Serum was serially diluted, added to plates, and incubated for 1 h at 37◦ C then at 4◦ C overnight. Plates were washed three times with PBST. Bound antibody was detected by horseradish peroxidase-conjugated rabbit anti-sheep IgG (γchain) (KPL, Gaithersburg, MD), incubated 1 h at 37◦ C. Substrate was SureBlue Reserve TMB Microwell Peroxidase Substrate (KPL, Gaithersburg, MD) and reaction stopped with TMB stop solution (R&D systems, Minneapolis, MN). Plates were read at 455 nm using SpectraMax M2 plate reader (Molecular Devices, San Jose, CA). Titer was reported as the lowest dilution with an optical density equal or greater than the mean plus two standard deviations of wells containing only PBS. Assay was performed in duplicate by a technician blinded to treatment groups and lesion outcomes for individual animals. ELISA results were analyzed using Prism 6 (GraphPad Software, San Diego, CA) statistical software fitting a two-way ANOVA for repeated measures to log transformed data with and Tukey’s multiple comparisons for simple effects within columns (between inoculum groups) and between rows (within inoculum groups, between timepoints). Data Management Raw data collected from histology report, ELISA, PCR, gross and microscopic observation, and representative photographs of each foot at each timepoint accompany this manuscript in the Supplementary Data Files. 16S rRNA gene sequencing data (fastq files) can be accessed under NCBI Bioproject accession number PRJNA781217 (https://www.ncbi.nlm.nih. gov/bioproject). RESULTS Clinical Presentation and Histology Five sheep were inoculated with elk hoof lesion batch 1 (WA I) and five sheep were inoculated with elk hoof lesion batch 2 (WA II). Two animals in WA I group were removed from the study, one due to injury unrelated to the experiment at 3 weeks PI and one due to penetration of severe lesions into tendon and muscle structures beneath the dermis occurring at 5 weeks PI. There were no observed differences in lesion development between WA I and WA II inoculums. For this reason, data from both inoculums were combined and data are presented as elk inoculum (n = 10 sheep with 8 surviving to 7 weeks, 16 feet) or mock inoculated (n = 5 sheep, 10 feet) groups. At 2 weeks PI all mock inoculated control animals had healed from the abrasion procedure and no lesions or spirochetes were observed for the remainder of the trial (Figures 2, 3; Table 1). Representative photographs of lesion progression are given in Figure 2. Photographs of all feet are provided in Supplementary Figure S2. Photographs of all feet were scored categorically and proportions of each category per inoculum at each timepoint are presented in Figure 3. At 2 weeks PI, almost all (85%) of the infected animals had developed lesions, with most classified as either moderate or severe (42 and 21%, respectively) (Figure 3). The lesion severity increased at 4 weeks PI with 29% observed feet with moderate lesions and 29% with severe lesions. However, this trend reversed at week 7 with many feet demonstrating resolution of lesions. At this time, 31% of observed feet had no lesions and 44% were scored as having mild or resolving lesions. In general, observation of spirochetes 16S rRNA Gene Sequencing and Data Analysis Using DNA extracted from lesion swabs collected prior to infection (week 0), and at 2, 4, and 7 weeks PI, amplicon library preparation for sequencing of the bacterial 16S rRNA gene v4 region was completed according to standard methodology at Argonne National Laboratory (https://www.anl.gov/bio/ environmental-sample-preparation-and-sequencing-facility). Amplicons were sequenced on an Illumina HiSeq 2500. Universal 16S rRNA gene forward primer (515F) and unique Golay barcoded reverse primers (806R) were used, and amplification was as previously described (31). Reads for each individual sample were demultiplexed and binned according to barcode by idemp (https://github.com/ yhwu/idemp). Reads were checked for quality and processed into operational taxonomic units (OTUs) using mothur analysis software (v.1.43.0) following the strategy outlined by Kozich et Frontiers in Veterinary Science \| www.frontiersin.org 5 January 2022 \| Volume 8 \| Article 782149 Wilson-Welder et al. TAHD in Sheep Model TABLE 1 \| Number of observed feet positive for spirochetes in swab fluid. Number of feet positive/number of feet observed* 2 week PI 4 week PI 7 week PI Mock 0/10 0/10 0/10 Inoculated 9/18 16/18 11/16 Minimum of 10 microscope fields observed. FIGURE 4 \| ELISA for serum antibody against mixed Treponema antigen measured prior to and 7 weeks PI. Indicates statistical significance from both week 0 as well as mock inoculated group (P < 0.05). FIGURE 2 \| Representative photographs of mock inoculated and inoculated feet showing lesion progression from scarification to 7 weeks PI. (A) Mock inoculated, (B) Elk TAHD lesion inoculated which maintained lesions through 7 weeks PI, (C) elk TAHD lesion inoculated spontaneously resolved before 7 weeks PI. associated with TAHC lesion material included mild erosions, mild hyperkeratosis, mild to moderate proliferative acanthosis, mild to moderate superficial lymphoplasmacytic inflammation, moderate dermal edema and alopecia, and numerous superficial coccobacilli and plump rods extending into dermal erosions. The full histopathology report is in the Supplementary Data File. ELISA Antibodies to Treponema are usually an indication of exposure and titers tend to correlate well with lesion severity (38). Serum antibody titers increased (P < 0.05) in the inoculated group but not in the mock infected group between week 0 (pre-infection) and week 7 (Figure 4). PCR T. medium, T. pedis, and T. phagedenis have consistently been found within TAHD lesions (2, 3) as well as within bovine DD lesions (2, 3, 15, 31, 39–47). Using species specific primers, one or more of these species were identified on inoculated foot swabs in at least one sampling timepoint (Table 2). All foot swabs were negative for Treponema species prior to inoculation (week 0). Most feet were positive for Fusobacterium at multiple timepoints, including samples from mock inoculated and from samples obtained prior to inoculation with lesion material (week 0). While Fusobacterium is a known contributor to foot rot of cattle and small ruminants, its role in DD is not well defined and is controversial (11, 41, 48–54). Our data supports an ambiguous role for Fusobacterium in TAHD as most sheep feet, even in the mock inoculated treatment group, FIGURE 3 \| Percentage of observed feet in each lesion severity category at 2, 4, and 7 weekS PI. All mock inoculated feet were non-lesioned or healthy looking at all timepoints (not shown), TAHD inoculated feet were scored as lesions (white bars/open bars), mild lesions (yellow bars/diagonal lines), moderate (orange bars/cross-hatched), or severe lesions (red bars/dotted). in swab fluid correlated with the presence of lesions (Table 1, Supplementary Data File). Histopathological examination of the wrapped feet identified changes in the skin environment caused by moist wraps were exhibited by samples from mock infected controls. Lesions Frontiers in Veterinary Science \| www.frontiersin.org 6 January 2022 \| Volume 8 \| Article 782149 Wilson-Welder et al. TAHD in Sheep Model theoretically, all sheep feet were the same. High similarity was also observed between inoculum and inoculated feet at week 2 (Supplementary Figure S3). Beta diversity analysis indicates overlap of clustering of samples at week 0, but divergence at week 2 between elk lesion inoculated and mock inoculated samples (Figure 5). Further divergence of the bacterial community was noted at week 4 as TAHD-inoculated samples more closely resembled the original elk lesion material and clustered less with week 0 or mock-inoculated sheep samples. Most abundant phyla in each inoculum type and timepoint are depicted in Figure 6. Week 0 samples are dominated by phyla Proteobacteria, order Pseudomonadales, and phlya Firmicutes, order Clostridales, with a strong component of Bacteroidetes (Bacteroidales) represented across all samples. Spirochetes and Spirochaetales were a dominant phylum in the elk lesion homogenate, and they remain in the top 10 Phyla of the inoculated samples at post-inoculation timepoints (2, 4, and 7 weeks PI) (Figure 6). Using DESeq2, pairwise analysis was made of timepoints between the inoculated and mock samples. There were 86 OTUs found to be positively significantly overrepresented in inoculated samples at 2, 4, or 7 weeks PI and have a base mean count > 20. Figure 7 shows the distribution of these samples across the timepoints. There were 43 OTUs present at all three timepoints meeting these criteria (significantly different between mock and inoculated, increased in inoculated with base mean count > 20) including 3 Treponema, 1 Actinobacteria (Actinomyces), 4 unclassified Bacteria, 10 Bacteroidetes (including 4 Porphyromonas and 1 Prevotella), 21 Firmicutes (Clostridia, Peptostreptococcus, Filifactor, and Ruminococcaceae), 1 Fusobacteria, 1 Proteobacteria (Campylobacter), SR1, and 1 Tenericutes (Mycolpasma). As represented by the treponemes (Figure 8) OTUs Otu0049, Otu0121, and Otu0156, the OTUs are present in the elk lesion material and inoculated samples as lesions develop (2, 4, and 7 weeks PI) but not with samples obtained prior to inoculation (week 0) or mock samples at 2, 4, or 7 weeks. Plots showing the representation of individual OTUs by sample and relative abundance for the other 40 OTUs can be found in Supplementary Figures S4-S6. TABLE 2 \| Number of PCR positive reactions over the number of sampled lesion swabs prior to scarification (0 week) and at 2, 4, and 7 weeks PI. T. pedis 0 week 2 week 4 week 7 week Mock 0/10 1/10 1/10 0/10 Inoculated 0/20 9/20 16/18 4/16 T. medium 0 week 2 week 4 week 7 week Mock 0/10 0/10 3/10 0/10 Inoculated 0/20 1/20 15/18 9/16 7 week T. phagedenis 0 week 2 week 4 week Mock 0/10 0/10 4/10 1/10 Inoculated 0/10 14/20 15/18 16/16 7 week Fusobacter 0 week 2 week 4 week Mock 8/10 0/10 4/10 4/10 Inoculated 15/20 18/20 14/18 16/16 Dichelobacter 0 week 2 week 4 week 7 week Mock 0/10 0/10 0/10 0/10 Inoculated 0/20 0/20 0/15 2/14 were positive for Fusobacterium prior to inoculation and many in the control group did not develop lesions. Animals were sorted into treatment/infection groups prior to Fusobacterium presence on feet was known. Only two samples, from the same inoculated animal at 7 weeks PI, was positive for Dichelobacter, indicating that it was not a key factor in this disease. Several mock inoculated feet were PCR positive for T. medium and T. phagedenis, most commonly at 7 weeks PI. No lesions were observed on mock-inoculated feet at week 7, nor spirochetes in silver-stained histopathology slides. These data should be interpreted carefully considering the observation that PCR techniques used in the current study are sensitive to crosscontamination which may have been influenced by housing all sheep in the same pen. Research staff does handle animals in a manner to reduce any cross-contamination (i.e., working noninfected controls first or on separate days, disinfection of gloves and equipment between groups); however, sometimes excessive rain or adverse conditions work against us (55, 56). DISCUSSION The goal of this experiment was to evaluate the ability of TAHD lesion material from elk to induce disease in a sheep model. Using lesion material collected from wild elk within the endemic disease area exhibiting classic TAHD lesions, we were able to produce lesions in most sheep by week 4 PI. Lesions contained spirochete shaped bacteria and molecular analysis determined that Treponema species or phylotypes typically associated with TAHD and livestock DD were associated with lesions. Spirochete shaped bacteria were at the leading edge of the lesions and accompanied by neutrophilic inflammatory infiltrates, and keratinocyte erosion and proliferation in accordance with typical histological presentation of TAHD (3). However, gross appearance of lesions 16S rRNA Gene Sequencing For 16S rRNA gene analysis, 119 samples representing four timepoints, two inoculum types (either elk lesion homogenate or mock), and 15 sheep (both right and left feet) were sequenced with 54,615 average number of read pairs per sample (16,137 min, 75,369 max). Prior to comparative analyses, the data were subsampled to 14,113 sequences, which was the smallest number of processed 16S rRNA gene v4 sequences among the samples. The DESeq2 analysis used the full (i.e., not subsampled) dataset. Data were analyzed by inoculum type: elk lesion material, infected, or mock inoculated, and by timepoints (0, 2, 4, and 7 weeks PI). Alpha diversity was most similar at week 0 when, Frontiers in Veterinary Science \| www.frontiersin.org 7 January 2022 \| Volume 8 \| Article 782149 Wilson-Welder et al. TAHD in Sheep Model FIGURE 5 \| Beta diversity ordination plot. Bray-Curtis Principal Component Analysis of bacterial 16S rRNA gene sequencing OTU grouped by inoculum and weeks. Elk lesion homogenate (i1c0), lesion inoculated sheep feet at 0 weeks (i1w0), 2 weeks (i1w2), 4 weeks (i1w4), and 7 weeks (i1w7) post-inoculation and mock inoculated 0 weesk (i2w0), 2 weeks (i2w2), 4 weeks (i2w4), and 7 weeks (i2w7) post-mock inoculation. in the sheep model differed from natural infection in elk, as lesions originated in the interdigital area on the front of the foot and then extended into or under the hoof horn wall. Inoculation of a sheep model with TAHD material resulted in a lesion pattern similar to that produced by bovine DD material, with lesions originating at the scarification sites in the pastern area. In this study, sheep appeared to be naïve to the treponeme spirochetes, as serum antibody titers increased in THAD infected sheep but not in mock inoculated controls. It should be emphasized that the sheep model is an acute model for studying induction and early pathogenesis of lesion development but is probably not adequate for characterizing chronic stages of disease when more extreme gross lesions such as hoof capsule sloughing are observed. However, the containment provided by the extensive wraps in the sheep model allows study of early lesion development including evaluation of invading bacteria without confounding effects of soil or manure contamination. Frontiers in Veterinary Science \| www.frontiersin.org In addition to the presence of Treponema, sequencing of bacterial 16S rRNA genes from lesions produced in the sheep model demonstrated the consistent presence of other bacterial phylotypes associated with TAHD and DD. These included bacteria in Porphyromonas, Prevotella, Bacterioidetes, Actinomyces, Campylobacter, and Mycoplasma genuses. Comparing mock to inoculated tissues at three timepoints, 86 OTUs were increased in inoculated samples with the majority found in all three sampling (43 OTUs) timepoints but the highest number was found at week 2 (Figure 7). Others have reported that microbial diversity of DD lesions decreases with time, with treponemes eventually becoming the dominant organism (31, 41, 44, 57). This was not observed in the current study. TAHD may differ from DD in this. Future studies of TAHD are needed to characterize microbial diversity or dominance with different lesion grades or time after infection with TAHD lesion material. Previous studies have identified bacteria in the complex community of DD lesions and have suggested roles 8 January 2022 \| Volume 8 \| Article 782149 Wilson-Welder et al. TAHD in Sheep Model FIGURE 6 \| Relative abundance of (A) top 10 Phyla taxonomic bacterial groups and (B) top 15 order in elk lesion homogenate (i1c0) and swabs from inoculated (i1) and mock-inoculated sheep (i2) feet at 0, 2, 4, and 7 weeks post-inoculation. most commonly Dichelobacter nodosus (58). D. nodosus can be transmitted between species (sheep and cattle), especially when common grazing areas are used (59). However, we did not find D. nodosus on the wrapped feet using bacterial 16S rRNA gene sequencing analysis or targeted PCR (Table 2). Proteobacter quantity and diversity decreased the longer the feet were wrapped, regardless of mock or TAHD inoculation. As others have identified a non-typable Campylobacter associated with bovine DD lesions (31, 60–62), the observation in the current study that a Campylobacter isolate, Out0128, is overrepresented in inoculated feet and increased with time was of interest (Supplementary Figure S4G). SR1 is an interesting taxon, found in sequencing projects from environmental samples to human microbiome (http://bytesizebio.net/2013/03/29/the- for different bacteria in lesion development. Actinomyces are known for anaerobic infection of compromised skin, forming branched, fusiform, fungal-like mats. Actinomyces have been observed in TAHD tissues and may be mistaken for spirochetes on observation (1, 3) (Supplementary Figure S4B). Fusobacteria are a causative agent of foot rot in ungulates and have been strongly associated with DD. In our experiment, we detected Fusobacterium in nearly all inoculated feet and just less than half of the mock inoculated samples prior to inoculation (Table 2). Otu0002, identified as a Fusobacterium species, was significantly over-represented in TAHD inoculated feet (Supplementary Figure S4F), indicating a probable positive correlation with lesion development. In foot rot of livestock, Fusobacterium is found in conjunction with other bacteria, Frontiers in Veterinary Science \| www.frontiersin.org 9 January 2022 \| Volume 8 \| Article 782149 Wilson-Welder et al. TAHD in Sheep Model and that treponemes are often invasive, found deep within the lesions, and often located at the leading edge of the lesion. Our results and data from others may suggest that, when compared to other bacterial populations, treponemes are not present in large numbers in lesions or readily identified by surface swabs (11). While individual species resolution using only the 16S rRNA gene V4 region has reservations, BLAST search of the three significant Treponema OTUs and the three non-significant OTUs (Otu0216, Otu0267, and Otu0276) all gave 98% or greater homology to previously recognized bovine DD Treponema species, Otu0049 and Otu0216 with 99 and 97% homology to T. pedis (NCBI GenBank ID: KR025849.1), Otu121 with 99 and 98% homology with T. medium and T. vincentii (NCBI GenBank ID: KP750179.1; KT192153.1; which cannot be resolved using only V4), Otu0156 with 99% homology to T. phagedenis (NCBI GenBank ID: CP054692.1), and Otu267 and Otu276 having 98% homology to T. putidum and Treponema sp. PT8 (which again has 97% homology to T. pedis). Therefore, it is very possible as was suggested by the close genetic relatedness of the elk, sheep, and bovine isolates as studied by Clegg et al. that DD regardless of host (bovine, sheep, or elk) the disease is being driven by the same organisms (4). Much like Fusobacterium and D. nodosus, hoof-associated Treponema does not appear to be host species restricted, in fact, the presence of cattle may be a risk factor on sheep farms for CODD (64). Recent publications showing that Mediterranean Buffalo (housed in milking herds) and European Bison (in zoological parks) are also susceptible to DD (65–67) further suggesting that any ungulate species is susceptible given the proper environmental conditions and the presence of these pathogenic bacteria. Characterizing the role of various bacteria and associated phylotypes was not a primary goal of this study. We did find that bacterial 16S rRNA gene sequencing was a useful method for contrasting microbial communities between mock and inoculated samples. By evaluating the bacterial community both as a whole and using individual OTU contribution, our data suggest a core transmissible bacterial element to TAHD. We cannot explain why TAHD lesions did not persist in the ovine model. Despite attempting to mimic environmental stressors placed on elk under field conditions such as a lower nutritional plane, our results suggest that not all elements of the elk-pathogen relationship were replicated in the ovine model. Although elk hoof samples were shipped and processed quickly, the possibility cannot be eliminated that a key member of the bacterial consortium necessary for chronicity of the lesions was eliminated. Additionally, we recognize the limitations of the sheep model. We utilized a method and inoculation steps that had previously resulted in success, putting the emphasis on testing the transmission and “pathogenesis” of the TAHD bacteria. Alternate inoculation sites (the dorsal aspect of the interdigital space in contrast to the pastern) might have eventually led to gross or clinical lesions similar to those seen in the elk. However, for welfare reasons we were not going to develop the long-term chronic lesions that resulted in the gross pathology associated with TAHD, namely the extreme overly grown, cavitated, and eventual sloughed FIGURE 7 \| Ven diagram of OTUs that are significantly positively overrepresented in infected samples with mean count > 20. power-of-single-cell-genomics-the-mysterious-sr1-bacteriahave-a-unique-genetic-code/). Associated with oral cavity and periodontal disease, it is not surprising to find it associated with TAHD (Supplementary Figure S4H), as there is an interesting overlap of bacterial phylotypes between periodontal disease and TAHD or DD of ruminants. Clostridia are a very large, varied class and are known for wide distribution across environments, including soil, and human and animal intestinal tracts, with many species being opportunistic pathogens. In the current study, 18 OTUs belonging to the Clostridiales cluster including unclassified Clostridiales, Filifactor, Peptostreptococcus, Helcoccus, Lachnospiraceae, and Clostridiales Incertae Sendis XI increased in TAHD inoculated samples as compared to mock samples (Supplementary Figure S5). Bacterioidetes are the largest phyla found in the rumen of sheep and cattle. Bacterioidetes phyla includes the genera Porphyromonas, Prevotella, and Odoribacter. Otus identified as these genera were increased in the TAHD infected samples (Supplementary Figure S6). These bacteria and closely related species have been identified in periodontal disease of humans and dogs, soft tissue abscesses, foot rot, and DD (63). However, specific roles for these bacteria have not yet been identified in TAHD or DD lesion development (63). The role of many of these bacteria in lesion development or continuance is unclear as a few phylotypes identified as Porphyromonas increase under a wet foot environment alone (Figure 6, i2w7). Although spirochete bacteria are usually dominant in DD and can be observed microscopically (Table 1) or identified by targeted PCR (Table 2), 16S rRNA gene sequencing suggests they are a minor portion of TAHD lesions. In the current study we observed that three Treponema OTUs, Otu0049, Otu0121, and Otu0156, were overrepresented in infected feet samples after inoculation. This may be influenced by differences in when swabs were collected and the use of swabs for sampling rather than biopsy samples. It should be noted that the size of sheep’s feet makes taking repeat biopsies inadvisable (11) Frontiers in Veterinary Science \| www.frontiersin.org 10 January 2022 \| Volume 8 \| Article 782149 Wilson-Welder et al. TAHD in Sheep Model FIGURE 8 \| Three Treponema genus OTUs which are significantly more abundant in two or more timepoints (weeks 2, 4, and/or 7) in inoculated feet as compared to mock feet. (A) Otu0049, (B) Otu0121, and (C) Otu0156. Frontiers in Veterinary Science \| www.frontiersin.org 11 January 2022 \| Volume 8 \| Article 782149 Wilson-Welder et al. TAHD in Sheep Model hooves. Further work characterizing the bacterial populations in TAHD lesions, further development of animal models, and/or continuation of efforts to characterize the epidemiologic characteristics of the disease will be required to develop intervention strategies for reducing disease prevalence in elk in the endemic area. In this study we were able to experimentally induce lesions with hallmarks of TAHD in a sheep model using lesion material derived from wild elk with TAHD. Histopathology was analyzed by a pathologist who had done the original description of the elk disease. Sheep lesions had strong histopathologic similarity to TAHD in elk, suggesting the same cellular process despite varying in location on the foot. The molecular assays revealed that the bacterial population was transmitted. While imperfect, the model allowed for initiation of earlystage lesions, and demonstrated the transmissible state of the bacterial consortium in a relevant, readily available animal model that has a lower risk of injury to both animals and human workers. Further study on the bacterial community and how individual bacterial members contribute to lesion development is warranted. AUTHOR CONTRIBUTIONS DATA AVAILABILITY STATEMENT ACKNOWLEDGMENTS The datasets presented in this study can be found in online repositories. The sequence data files are available at https://www. ncbi.nlm.nih.gov/ under Bioproject TAHD lesion material in sheep model Accession: PRJNA781217 ID: 781217. The authors wish to acknowledge Scott Harris for collecting and shipping infected elk hooves, Ami Frank and Elizabeth Weyer for excellence in laboratory assistance, and the NCAH Animal Care Unit for excellence in animal care. ETHICS STATEMENT SUPPLEMENTARY MATERIAL The animal study was reviewed and approved by National Animal Disease Center Institutional Animal Care and Use Committee. The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fvets. 2021.782149/full#supplementary-material JW-W, KM, SH, and DA contributed to conception and design of the study. KM provided the elk lesion material. JW-W, DA, and SO performed the study and collected the data. SH performed the pathology analysis. DB performed the 16S rRNA gene sequencing analysis. JW-W wrote the first draft of the manuscript. SH, DB, SO, and DA provided sections of the manuscript, editing, and critical review. All authors contributed to manuscript revision, read, and approved the submitted version. FUNDING This work was conducted without grants or funds from public or private entities. Work was completed by US Department of Agriculture employees in the course of their assigned duties in relation to project number 5030-32000-223-00-D. Mention of trade names or commercial products in this study is solely for providing specific information and does not imply recommendation or endorsement by the U.S. Department of Agriculture. REFERENCES 7. Angell JW, Cripps PJ, Grove-White DH, Duncan JS. A practical tool for locomotion scoring in sheep: reliability when used by veterinary surgeons and sheep farmers. Vet Rec. (2015) 176:521. doi: 10.1136/vr.102882 8. Cortes JA, Thomas A, Hendrick S, Janzen E, Pajor EA, Orsel K. Risk factors of digital dermatitis in feedlot cattle. Transl Anim Sci. (2021) 5:txab075. doi: 10.1093/tas/txab075 9. Hesseling J, Legione AR, Stevenson MA, McCowan CI, Pyman MF, Finochio C, Nguyen D, et al. Bovine digital dermatitis in Victoria, Australia. Aust Vet J. (2019) 97:404–13. doi: 10.1111/avj.12859 10. Relun A, Lehebel A, Bruggink M, Bareille N, Guatteo R. Estimation of the relative impact of treatment and herd management practices on prevention of digital dermatitis in French dairy herds. Prev Vet Med. (2013) 110:558–62. doi: 10.1016/j.prevetmed.2012.12.015 11. Duncan JS, Angell JW, Richards P, Lenzi L, Staton GJ, Grove-White D, et al. The dysbiosis of ovine foot microbiome during the development and treatment of contagious ovine digital dermatitis. Anim Microbiome. (2021) 3:19. doi: 10.1186/s42523-021-00078-4 12. Gomez A, Cook NB, Bernardoni ND, Rieman J, Dusick AF, Hartshorn R, et al. An experimental infection model to induce digital dermatitis infection in cattle. J Dairy Sci. (2012) 95:1821–30. doi: 10.3168/jds.2011-4754 13. Krull AC, Shearer JK, Gorden PJ, Scott HM, Plummer PJ. Digital dermatitis: natural lesion progression and regression in Holstein dairy cattle over 3 years. J Dairy Sci. (2016) 99:3718–31. doi: 10.3168/jds.2015-10535 1. Han S, Mansfield KG. Severe hoof disease in free-ranging Roosevelt elk (Cervus elaphus roosevelti) in southwestern Washington, USA. J Wildl Dis. (2014) 50:259–70. doi: 10.7589/2013-07-163 2. Clegg SR, Mansfield KG, Newbrook K, Sullivan LE, Blowey RW, Carter SD, Evans NJ. Isolation of digital dermatitis treponemes from hoof lesions in Wild North American Elk (Cervus elaphus) in Washington State, USA. J Clin Microbiol. (2015) 53:88–94. doi: 10.1128/JCM.02276-14 3. Han S, Mansfield KG, Bradway DS, Besser TE, Read DH, Haldorson GJ, et al. Treponeme-associated hoof disease of free-ranging Elk (Cervus elaphus) in Southwestern Washington State, USA. Vet Pathol. (2019) 56:118–32. doi: 10.1177/0300985818798108 4. Clegg SR, Carter SD, Birtles RJ, Brown JM, Hart CA, Evans NJ. Multilocus sequence typing of pathogenic treponemes isolated from cloven-hoofed animals and comparison to treponemes isolated from humans. Appl Environ Microbiol. (2016) 82:4523–36. doi: 10.1128/AEM.00025-16 5. Hill AB. The environment and disease: association or causation? Proc R Soc Med. (1965) 58:295–300. doi: 10.1177/0035915765058 00503 6. Wilson-Welder JH, Alt DP, Nally JE. Digital dermatitis in cattle: current bacterial and immunological findings. Animals. (2015) 5:1114–35. doi: 10.3390/ani5040400 Frontiers in Veterinary Science \| www.frontiersin.org 12 January 2022 \| Volume 8 \| Article 782149 Wilson-Welder et al. TAHD in Sheep Model 34. Beck D, Settles M, Foster JA. OTUbase: an R infrastructure package for operational taxonomic unit data. Bioinformatics. (2011) 27:1700–1. doi: 10.1093/bioinformatics/btr196 35. Foster ZS, Sharpton TJ, Grunwald NJ. Metacoder: an R package for visualization and manipulation of community taxonomic diversity data. PLoS Comput Biol. (2017) 13:e1005404. doi: 10.1371/journal.pcbi.1005404 36. Foster ZSL, Chamberlain S, Grunwald NJ. Taxa: an R package implementing data standards and methods for taxonomic data. F1000Res. (2018) 7:272. doi: 10.12688/f1000research.14013.2 37. McMurdie PJ, Holmes S. phyloseq: an R package for reproducible interactive analysis and graphics of microbiome census data. PLoS ONE. (2013) 8:e61217. doi: 10.1371/journal.pone.0061217 38. Gomez A, Anklam KS, Cook NB, Rieman J, Dunbar KA, Cooley KE, et al. Immune response against Treponema spp. and ELISA detection of digital dermatitis. J Dairy Sci. (2014) 97:4864–75. doi: 10.3168/jds.2013-7616 39. Beninger C, Naqvi SA, Naushad S, Orsel K, Luby C, Derakhshani H, et al. Associations between digital dermatitis lesion grades in dairy cattle and the quantities of four Treponema species. Vet Res. (2018) 49:111. doi: 10.1186/s13567-018-0605-z 40. Blanchard AM, Staley CE, Shaw L, Wattegedera SR, Baumbach CM, Michler JK, et al. A Trifecta of new insights into ovine footrot for infection drivers, immune response and host pathogen interactions. Infect Immun. (2021) 89:IAI0027021. doi: 10.1128/IAI.00270-21 41. Caddey B, De Buck J. Meta-analysis of bovine digital dermatitis microbiota reveals distinct microbial community structures associated with lesions. Front Cell Infect Microbiol. (2021) 11:685861. doi: 10.3389/fcimb.2021.685861 42. Caddey B, Orsel K, Naushad S, Derakhshani H, De Buck J. Identification and quantification of bovine digital dermatitis-associated microbiota across lesion stages in feedlot beef cattle. mSystems. (2021) 6:e0070821. doi: 10.1128/mSystems.00708-21 43. Calvo-Bado LA, Oakley BB, Dowd SE, Green LE, Medley GF, Ul-Hassan A, et al. Ovine pedomics: the first study of the ovine foot 16S rRNA-based microbiome. ISME J. (2011) 5:1426–37. doi: 10.1038/ismej.2011.25 44. Gotoh Y, Chiba K, Sekiyama Y, Okada K, Hayashi T, Misawa N. 16S rRNA-based amplicon analysis of changes in the bacterial population in the lesions of papillomatous digital dermatitis in dairy cattle after topical treatment with allyl isothiocyanate. Microbiol Immunol. (2020) 64:416–23. doi: 10.1111/1348-0421.12786 45. Klitgaard K, Boye M, Capion N, Jensen TK. Evidence of multiple Treponema phylotypes involved in bovine digital dermatitis as shown by 16S rRNA gene analysis and fluorescence in situ hybridization. J Clin Microbiol. (2008) 46:3012–20. doi: 10.1128/JCM.00670-08 46. Yano T, Yamagami R, Misumi K, Kubota C, Moe KK, Hayashi T, et al. Genetic heterogeneity among strains of Treponema phagedenis-like spirochetes isolated from dairy cattle with papillomatous digital dermatitis in Japan. J Clin Microbiol. (2009) 47:727–33. doi: 10.1128/JCM.01574-08 47. Zinicola M, Higgins H, Lima S, Machado V, Guard C, Bicalho R. Shotgun metagenomic sequencing reveals functional genes and microbiome associated with bovine digital dermatitis. PLoS ONE. (2015) 10:e0133674. doi: 10.1371/journal.pone.0133674 48. Bay V, Griffiths B, Carter S, Evans NJ, Lenzi L, Bicalho RC, et al. 16S rRNA amplicon sequencing reveals a polymicrobial nature of complicated claw horn disruption lesions and interdigital phlegmon in dairy cattle. Sci Rep. (2018) 8:15529. doi: 10.1038/s41598-018-33993-9 49. Gelasakis AI, Kalogianni AI, Bossis I. Aetiology risk factors, diagnosis and control of foot-related lameness in dairy sheep. Animals. (2019) 9:509. doi: 10.3390/ani9080509 50. Handeland K, Boye M, Bergsjo B, Bondal H, Isaksen K, Agerholm JS. Digital necrobacillosis in Norwegian wild tundra reindeer (Rangifer tarandus tarandus). J Comp Pathol. (2010) 143:29–38. doi: 10.1016/j.jcpa.2009.12.018 51. Maboni G, Frosth S, Aspan A, Totemeyer S. Ovine footrot: new insights into bacterial colonisation. Vet Rec. (2016) 179:228. doi: 10.1136/vr.103610 52. Moreira TF, Facury Filho EJ, Carvalho AU, Strube ML, Nielsen MW, Klitgaard K, et al. Pathology and bacteria related to digital dermatitis in dairy cattle in all year round grazing system in Brazil. PLoS ONE. (2018) 13:e0193870. doi: 10.1371/journal.pone.0193870 53. Nielsen MW, Strube ML, Isbrand A, Al-Medrasi WD, Boye M, Jensen TK, et al. Potential bacterial core species associated with digital dermatitis in 14. Morck DW, Gard MS, Olson ME. Experimental evaluation of a commercial footrot vaccine against native Canadian strains of Dichelobacter nodosus. Can J Vet Res. (1994) 58:122–6. 15. Wilson-Welder JH, Nally JE, Alt DP, Palmer MV, Coatney J, Plummer P. Experimental transmission of bovine digital dermatitis to sheep: development of an infection model. Vet Pathol. (2018) 55:245–57. doi: 10.1177/0300985817736572 16. Maboni G, Davenport R, Sessford K, Baiker K, Jensen TK, Blanchard AM, et al. A novel 3D skin explant model to study anaerobic bacterial infection. Front Cell Infect Microbiol. (2017) 7:404. doi: 10.3389/fcimb.2017.00404 17. Newbrook K, Carter SD, Crosby-Durrani H, Evans NJ. Challenge of bovine foot skin fibroblasts with digital dermatitis treponemes identifies distinct pathogenic mechanisms. Front Cell Infect Microbiol. (2020) 10:538591. doi: 10.3389/fcimb.2020.538591 18. Boggiatto PM, Crawford LS, Kanipe C, Palmer MV, Olsen SC. Case report: fading Elk syndrome in a herd of captive Elk (Cervus elaphus) in the North American Midwest. Front Vet Sci. (2020) 7:497. doi: 10.3389/fvets.2020.00497 19. Cox RJ, Nol P, Ellis CK, Palmer MV. Research with Agricultural animals and wildlife. ILAR J. (2019) 60:66–73. doi: 10.1093/ilar/ilz006 20. Palmer MV, Cox RJ, Waters WR, Thacker TC, Whipple DL. Using white-tailed deer (Odocoileus virginianus) in infectious disease research. J Am Assoc Lab Anim Sci. (2017) 56:350–60. Available online at: https://www.ingentaconnect. com/content/aalas/jaalas/2017/00000056/00000004/art00001;jsessionid= 637lcbfch3ra1.x-ic-live-03# 21. Palmer MV, Whipple DL, Payeur JB, Bolin CA. Use of the intradermal tuberculin test in a herd of captive elk (Cervus elaphus nelsoni) naturally infected with Mycobacterium bovis. J Vet Diagn Invest. (2011) 23:363–6. doi: 10.1177/104063871102300230 22. Waters WR, Sacco RE, Fach SJ, Palmer MV, Olsen SC, Kreeger TJ. Analysis of mitogen-stimulated lymphocyte subset proliferation and nitric oxide production by peripheral blood mononuclear cells of captive elk (Cervus elaphus). J Wildl Dis. (2002) 38:344–51. doi: 10.7589/0090-3558-38.2.344 23. Valkova L, Vecerek V, Voslarova E, Kaluza M, Takacova D. The welfare of cattle, sheep, goats and pigs from the perspective of traumatic injuries detected at slaughterhouse postmortem inspection. Animals. (2021) 11:1406. doi: 10.3390/ani11051406 24. Bruning-Fann CS, Shank KL, Kaneene JB. Descriptive epidemiology of captive cervid herds in Michigan, USA. Vet Res. (1997) 28:295–302. 25. Hattel AL, Shaw DP, Fisher JS, Brooks JW, Love BC, Drake TR, et al. Mortality in Pennsylvania captive elk (Cervus elaphus): 1998–2006. J Vet Diagn Invest. (2007) 19:334–7. doi: 10.1177/104063870701900322 26. Woodbury MR, Berezowski J, Haigh J. A retrospective study of the causes of morbidity and mortality in farmed elk (Cervus elaphus). Can Vet J. (2005) 46:1108-13, 1120-1. 27. Tosswill M, Roskruge M, Smith A, Christey G. Livestock-related injuries in the Midland region of New Zealand. N Z Med J. (2018) 131:13–20. 28. Krull AC, Cooper VL, Coatney JW, Shearer JK, Gorden PJ, Plummer PJ. A highly effective protocol for the rapid and consistent induction of digital dermatitis in holstein calves. PLoS ONE. (2016) 11:e0154481. doi: 10.1371/journal.pone.0154481 29. Sheehan DC, Hrapchak BB. Theory and Practice of Histotechnology. Columbus, OH: Battelle Press (1987). 30. Callis G, Sterchi D. In: Bowie M, Veterinary RC, editors. Animal Processing Manual. Ellicott City, MD: National Sociesty of Histotechnology (2002). 31. Krull AC, Shearer JK, Gorden PJ, Cooper VL, Phillips GJ, Plummer PJ. Deep sequencing analysis reveals temporal microbiota changes associated with development of bovine digital dermatitis. Infect Immun. (2014) 82:3359–73. doi: 10.1128/IAI.02077-14 32. Kozich JJ, Westcott SL, Baxter NT, Highlander SK, Schloss PD. Development of a dual-index sequencing strategy and curation pipeline for analyzing amplicon sequence data on the MiSeq Illumina sequencing platform. Appl Environ Microbiol. (2013) 79:5112–20. doi: 10.1128/AEM. 01043-13 33. Schloss PD, Westcott SL, Ryabin T, Hall JR, Hartmann M, Hollister EB, et al. Introducing mothur: open-source, platform-independent, community-supported software for describing and comparing microbial communities. Appl Environ Microbiol. (2009) 75:7537–41. doi: 10.1128/AEM. 01541-09 Frontiers in Veterinary Science \| www.frontiersin.org 13 January 2022 \| Volume 8 \| Article 782149 Wilson-Welder et al. 54. 55. 56. 57. 58. 59. 60. 61. 62. TAHD in Sheep Model 63. Flint HJ, Duncan SH. Bacteroides and Prevotella, Encyclopedia of Food Microbiology. 2nd ed. Amsterdam: Elsevier/Academic Press (2014). p. 203–8. 64. Angell JW, Duncan JS, Carter SD, Grove-White DH. Farmer reported prevalence and factors associated with contagious ovine digital dermatitis in Wales: a questionnaire of 511 sheep farmers. Prev Vet Med. (2014) 113:132–8. doi: 10.1016/j.prevetmed.2013.09.014 65. Guccione J, Carcasole C, Alsaaod M, D’Andrea L, Di Loria A, De Rosa A, et al. Assessment of foot health and animal welfare: clinical findings in 229 dairy Mediterranean Buffaloes (Bubalus bubalis) affected by foot disorders. BMC Vet Res. (2016) 12:107. doi: 10.1186/s12917-016-0726-4 66. Hoby S, Jensen TK, Brodard I, Gurtner C, Eicher R, Steiner A, et al. Detection of treponemes in digital dermatitis lesions of captive European bison (Bison bonasus). PLoS ONE. (2021) 16:e0255921. doi: 10.1371/journal.pone.0255921 67. Hoby S, Steiner A, Kuhnert P, Furtado Jost R, Guthruf S, Schonbachler K, et al. Foot health and prevalence of Dichelobacter nodosus in 11 ungulate species at Berne Animal Park. Schweiz Arch Tierheilkd. (2020) 162:675–81. doi: 10.17236/sat00277 cattle herds identified by molecular profiling of interdigital skin samples. Vet Microbiol. (2016) 186:139–49. doi: 10.1016/j.vetmic.2016.03.003 Witcomb LA, Green LE, Kaler J, Ul-Hassan A, Calvo-Bado LA, Medley GF, et al. A longitudinal study of the role of Dichelobacter nodosus and Fusobacterium necrophorum load in initiation and severity of footrot in sheep. Prev Vet Med. (2014) 115:48–55. doi: 10.1016/j.prevetmed.2014.03.004 Angell JW, Clegg SR, Grove-White DH, Blowey RW, Carter SD, Duncan JS, et al. Survival of contagious ovine digital dermatitis (CODD)-associated treponemes on disposable gloves after handling CODD-affected feet. Vet Rec. (2017) 181:89. doi: 10.1136/vr.104228 Gillespie AV, Carter SD, Blowey RW, Staton GJ, Evans NJ. Removal of bovine digital dermatitis-associated treponemes from hoof knives after foot-trimming: a disinfection field study. BMC Vet Res. (2020) 16:330. doi: 10.1186/s12917-020-02552-8 Klitgaard K, Strube ML, Isbrand A, Jensen TK, Nielsen MW. Microbiota Analysis of an environmental slurry and its potential role as a reservoir of bovine digital dermatitis pathogens. Appl Environ Microbiol. (2017) 83:e00244–17. doi: 10.1128/AEM.00244-17 Zhou H, Bennett G, Hickford JG. Variation in Fusobacterium necrophorum strains present on the hooves of footrot infected sheep, goats and cattle. Vet Microbiol. (2009) 135:363–7. doi: 10.1016/j.vetmic.2008.09.084 Knappe-Poindecker M, Gilhuus M, Jensen TK, Vatn S, Jørgensen HJ, Fjeldaas T. Cross-infection of virulent Dichelobacter nodosus between sheep and co-grazing cattle. Vet Microbiol. (2014) 170:375–82. doi: 10.1016/j.vetmic.2014.02.044 Döpfer D, Koopmans A, Meijer FA, Szakáll I, Schukken YH, Klee W, et al. Histological and bacteriological evaluation of digital dermatitis in cattle, with special reference to spirochaetes and Campylobacter faecalis. Vet Rec. (1997) 140:620–3. doi: 10.1136/vr.140.24.620 Schroeder CM, Parlor KW, Marsh TL, Ames NK, Goeman AK, Walker RD. Characterization of the predominant anaerobic bacterium recovered from digital dermatitis lesions in three Michigan dairy cows. Anaerobe. (2003) 9:151–5. doi: 10.1016/S1075-9964(03)00084-2 Shibahara T, Ohya T, Ishii R, Ogihara Y, Maeda T, Ishikawa Y, et al. Concurrent spirochaetal infections of the feet and colon of cattle in Japan. Aust Vet J. (2002) 80:497–502. doi: 10.1111/j.1751-0813.2002.tb12474.x Frontiers in Veterinary Science \| www.frontiersin.org Conflict of Interest: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. Publisher’s Note: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. Copyright © 2022 Wilson-Welder, Mansfield, Han, Bayles, Alt and Olsen. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. 14 January 2022 \| Volume 8 \| Article 782149
https://openalex.org/W2616367663	https://ojs.amhinternational.com/index.php/jebs/article/download/1649/1418	English	null	Personality, Attitude and Behavioural Components of Financial Literacy: A Comparative Analysis	Journal of economics and behavioral studies	2,017	cc-by	7,494	1. Introduction The need for the development of financial literacy has been continually present in the past decades. However, it came to the centre of attention only in connection with the financial crisis unfolding in 2008 (Botos et al., 2012). The notion of financial literacy has been defined by many and in many ways. Related research in Hungary and abroad has often been initiated and financed by the largest financial services providers. In this context the view that the more familiar the population is with banking products and the more financial risks they are willing to take, the more advanced their financial literacy is has been widely accepted and held to be axiomatic by researchers. According to the most accepted definition in Hungary, "financial literacy is the level of financial knowledge and skills that enables individuals to identify and subsequently interpret basic financial information in order to be able to make conscious and prudent decisions and to be able to assess the potential future financial and other consequences of their decision" (Magyar Nemzeti Bank, 2014). The State Audit Office of Hungary's definition equally emphasizes the appropriate levels of financial knowledge, and the ability to manage money and states that financial literacy is above all about having a realistic self-image of one's own financial literacy and making adjusted decisions (Huzdik, Béres & Németh, 2014). The present research examines the personality, attitude, and behavioural components of financial literacy. It is based on the premise that financial literacy is not only determined by knowledge and skills but also by factors such as attitudes towards money, behavioural patterns, financial planning, an ability to prolong one's satisfaction of needs, and whether an individual is capable of "keeping their finances in order" (Zsótér et al., 2016). The main aim of the present study is to compare the results of a study about financial personalities conducted in Hungary in 2015 on behalf of the Financial Compass Foundation (Pénziránytű Alapítvány) with those of a study about financial literacy conducted in 2015 by market research company GfK and the Financial Compass Foundation. A 36-item financial personality test was developed by Erzsébet Németh that was completed, during the summer of 2015, by 3088 individuals (Németh et al., 2016; Béres et al., 2015). The test examines the economic and psychological aspects of individuals' relationship with money. A factor analysis revealed 9 factors that cover financial personality types. Keywords: Personality, attitude, behaviour, financial literacy Keywords: Personality, attitude, behaviour, financial literacy Personality, Attitude and Behavioural Components of Financial Literacy: A Comparative Erzsébet Németh1, Boglárka Zsótér2 1Budapest Metropolitan University, Hungary 2Corvinus University of Budapest, Hungary enemeth@metropolitan.hu, boglarka.zsoter@uni-corvinus. Erzsébet Németh1, Boglárka Zsótér2 1Budapest Metropolitan University, Hungary 2Corvinus University of Budapest, Hungary enemeth@metropolitan.hu, boglarka.zsoter@uni-corvinus.hu Abstract: Since the financial crisis in 2008 the investigation of financial literacy–especially its components (personality, attitudes, behaviour etc.) - is in the limelight. Modern economics have recognized that in order to effectively forecast financial and economic processes it is primordial to understand the attitudes of the members of society toward finances, as well as the characteristics of various social group sharing the same views and behaviours. In 2015 two relevant pieces of research were conducted in this topic in Hungary. One focuses on the financial personality types, while the other investigates Hungarians’ financial culture in general based on the research methodology of the OECD. Based on these two databases our comparative study highlights the main characteristics of financial personality types. The three clusters based on the OECD research cover the nine personality types from the results of the other Hungarian research. Our findings show that the cluster of “anxious unsatisfied” encapsulates the “economizers with little money”, the “price sensitive” and the “collector” personality types. Furthermore, the “satisfied conscious” covers the “order creates value”, the “diligent” and the “planner” personality types. Finally, the “moderately anxious unconsidered” involves the “ups and downs”, the “money-devouring” and the “cannot control finances” personality types. The clusters identified during the research show idiosyncratic financial and psychological vulnerability and/or protection. Journal of Economics and Behavioral Studies (ISSN: 2220-6140) Vol. 9, No. 2, pp. 46-57, April 2017 Journal of Economics and Behavioral Studies (ISSN: 2220-6140) Vol. 9, No. 2, pp. 46-57, April 2017 2. Literature Review Money does not only have intrinsic value but also outstanding features and strong motivating power (Opsahl and Dunnette, 1966). Furthermore it is a highly subjective concept, meaning something different for each person (Wernimont and Fitzpatrick, 1972). For these reasons, in the following, the most prominent research directions and results related to financial personality and financial attitudes are synthesized. The extensive research in the field goes back to the beginning of the 1970s; the following theoretical review gives an overview of the most prominent results since then. Yamauchi and Templer (1982) developed a standard measure of financial attitude called the Money Attitude Scale (MAS) and identified four dimensions of money attitude. The first dimension is power-prestige, in which money is the symbol of success and power in the individuals’ attitudes the second dimension is retention-time. In the case of individuals in this group, the main focus is on preparation for the future and keeping the financial situation under continuous control. For them, saving and amassing are of primary importance, and they regularly record the situation of their finances. The third dimension is distrust. The common feature of individuals in this category is that they look at money with suspicion, almost with fear. Individuals that have no trust in money and finances, usually do not trust themselves enough either. The fourth and last dimension of the authors is anxiety, and includes individuals prone to worrying and distress over money matters. Chan (2003) compares financial approaches with the consumer types described by Sproles and Kendall (1986) applying Yamauchi and Templer’s (1982) theory. People who regard money as a symbol of power are mainly quality-oriented and novelty seeking consumers. People who have a distrustful attitude to money tend to be uncertain about and frustrated with their consumer decisions. Consumers with an anxious attitude usually suffer from ambivalent feelings in their consumer decision-making as they are seeking both pleasure and price-quality balance concurrently. Furnham (1984) developed a standard measure of money-related beliefs and behaviour patterns called the Money Beliefs and Behaviours Scale (MBBS). The 60-item scale is further reduced to six factors, namely: obsession, power/spending, retention, security/conservative, inadequacy, and effort/stability. In an investigation among teenagers, Furnham (1999) defined money attitudes as attitudes to spending and saving, and with respect to the age characteristics of the target group, applied a scale different from the MBBS scale (Furnham, 1984). Journal of Economics and Behavioral Studies (ISSN: 2220-6140) Vol. 9, No. 2, pp. 46-57, April 2017 averages. The thus obtained three factors and three clusters were compared with the results of the financial personality study. Each cluster obtained in the OECD study covers three factors of the financial personality study. A review of the concepts of financial attitude, financial personality, financial decisions, and financial behaviour served as the theoretical foundations of the study. averages. The thus obtained three factors and three clusters were compared with the results of the financial personality study. Each cluster obtained in the OECD study covers three factors of the financial personality study. A review of the concepts of financial attitude, financial personality, financial decisions, and financial behaviour served as the theoretical foundations of the study. 1. Introduction A 12-item scale examining respondents' financial attitudes and behaviour was part of the study of financial literacy based on the OECD questionnaire. 1000 individuals were surveyed, representing the population of Hungary by age, gender and type of settlement. A factor analysis was performed on the financial attitude and behaviour scale, followed by a cluster analysis based on factor 46 Journal of Economics and Behavioral Studies (ISSN: 2220-6140) Vol. 9, No. 2, pp. 46-57, April 2017 averages. The thus obtained three factors and three clusters were compared with the results of the financial personality study. Each cluster obtained in the OECD study covers three factors of the financial personality study. A review of the concepts of financial attitude, financial personality, financial decisions, and financial behaviour served as the theoretical foundations of the study. Journal of Economics and Behavioral Studies (ISSN: 2220-6140) Vol. 9, No. 2, pp. 46-57, April 2017 Journal of Economics and Behavioral Studies (ISSN: 2220-6140) Vol. 9, No. 2, pp. 46-57, April 2017 are not characteristic features of this personality type. Money monks feel bad when in possession of a lot of money. Such situations bring about a certain sense of guilt in them, especially when they come into a large sum of money. They are convinced that money spoils everything. Money avoiders try to avoid daily tasks about money. Individuals belonging to this group do not like to deal with their finances, so they usually also do not produce budgets. Money amassers consider the amount of money available to them – or rather the increase of it – one of their main objectives, as they consider that ultimately this is the way they can also prove their power. Mellan (1994) refers to a combination of the hoarder and spender personality types as a binger. Bingers tend to economize for a while (e.g. for the achievement of a major objective), but if affected by an external stimulus (impulse), they are susceptible to shop without consideration. The common characteristic feature of money worriers is that they lack self-confidence, they are afraid to lose control and therefore they keep control over their finances. They tend to continuously monitor their financial situation. Risk-takers perceive money as a source of means adventure, excitement and freedom. They enjoy risking their money as they enjoy the shivering and adrenaline that come with it. Mellan's (1994) last category is the group of risk-avoiders. For them, money equals security; therefore, they prefer keeping their money at home, if they can. Table 1 provides a summary of the results of the studies presented beforehand. Table 1: A summary on research on financial attitude and financial personality types Author(s) Dimensions of financial personality and attitude Goldberg and Lewis (1978) Forman (1987) miser autonomy worshipper power grabber gambler Yamauchi and Templer (1982) power-prestige retention-time distrust anxiety Furnham (1984) obsession power retention security inadequacy effort/ability Tang (1992) budget negative feelings about money money as a token of success Mellan (1994) hoarder spender money monk money avoider money amasser binger money worrier risk-taker risk-avoider Table 1: A summary on research on financial attitude and financial personality types Based on the above, one can state that a considerable amount of research has been devoted to financial personalities, but, at the same time, one can equally note that certain personality types cannot be clearly distinguished from each other: there are some overlaps between them. 2. Literature Review In the end, the 20-statement-scale resulted in five factors, labelled as spending money, saving money, mechanics of banking, work ethic and indifference to money. Wilhelm Stern was the first psychologist to study personality. Since Stern, the recognition of “self” has been considered as the essential condition of a personality. This is the ability of the individual to distinguish themselves from their environment. Only humans possess a personality. According to Stern, personality is a manifold, dynamic unit. Personality psychology is a branch of psychology that studies how internal and external factors affect the development of personality. Personality is the ensemble of traits that distinguish an individual from other people and that an individual has by nature and gains later in their life. Several studies have shown the existence of a strong relationship between personality and making wrong financial decisions (see e.g. Jureviciene and Jermakova, 2012, Iqbal et al., 2012; Brozynski et al., 2004). The most common model of trait research is the five-factor personality model (McCrae, 2009). The five factors, referred to as the “Big Five” since Goldberg (1971), are Openness to Experience, Conscientiousness, Extraversion, Agreeableness and Neuroticism or Emotional Stability (Borghans et al., 2008). According to the results by Kübilay and Bayrakdaroğlu (2016), each personality type faces different biases and each investor has different risk tolerance. Mellan (1994) identifies nine personality types on the basis of their attitude to money: hoarder, spender, money monk, money avoider, money amasser, binger, money worrier, risk-taker, and risk-avoider. A hoarder is an individual who sticks to their money, finds it hard to buy things that would cause momentary pleasure to themselves or their beloved ones. Money represents a certain security to them, thereby being estranged from all kinds of hedonistic behaviour. Spenders find pleasure in spending their money when and on what they feel necessary – this status is usually related to an external stimulus. Saving money and making budgets 47 3. Methodology To identify financial personalities, the authors of the present study used a personality test containing 36 statements (Németh et al., 2016; Béres et al., 2015). To test their financial personality and to study what behavioural patterns, habits and attitudes characterize them; respondents who visited the site http://penziranytu.hu/penzugyi-szemelyisegteszt were asked to complete an online questionnaire. Respondents had to decide the extent to which they were characterized by each statement of the questionnaire using a five-point Likert-scale where 1 indicated "strongly disagree", and 5 indicated "strongly agree". During the summer of 2015, 3,139 respondents filled in the questionnaire and following data cleaning the sample consisted of a total of 3,088 responses. The low dropout rate is the result of a number of preliminary methodological considerations. First, respondents typically did not interrupt the process of questionnaire response. Interruption usually happens when respondents fail to understand the statements and/or questions, or if they consider the questionnaire boring or too intrusive. Second, very few straight-line or pattern responses (i.e. responding without thought by answering the same for all statements or by recording a pattern [e.g. zigzag or Christmas-tree]) were recorded. None of the above issues occurred, thanks to the following considerations: 1. Phrasing: Statements were phrased in a way that ordinary people could understand them, without perceiving them as too scientific or the topic as remote. This allowed for minimizing the risk of misunderstanding. 2. Sensitive issues: Sensitive issues were completely omitted from the present questionnaire, i.e. beyond the test containing the 36 statements, respondents were not asked for any additional socio-demographic data (e.g. to state their income). The lack of these former can also be interpreted as a limitation of the research, however, the resulting increase in confidence allowed for a considerable increase in the willingness to reply. 3. Motivation: Following the completion of the personality test, respondents were immediately provided with the evaluation of their profile. The inclusion of feedback equally facilitated the willingness to participate. The main aim of the financial culture research conducted by market research company GfK and the Financial Compass Foundation (Pénziránytű Alapítvány) was to assess the financial awareness and literacy of Hungarian adults. The OECD produced a unified international methodology that was joined by Hungary – along 13 other countries – in 2010. The Financial Compass Foundation once again participated in the study in 2015, with, this time, about 30 countries conducting the research at the same time. Journal of Economics and Behavioral Studies (ISSN: 2220-6140) Vol. 9, No. 2, pp. 46-57, April 2017 It is important to note that all dimensions of financial personality or financial behaviour do not always appear in each related study. The dimensions of retention and time, as well as negative feelings about money are recurring elements of these studies. The widest profile range was offered by the works of Furnham (1984) and Mellan (1994). Similar personality types were found in the empirical part of the present research. 48 48 4. Results Results of the financial personality survey: Descriptive statistics of the 36-item scale developed by Erzsébet Németh are first examined. Table 2 gives an overview of the mean and standard deviation values for the 36 statements in an ascending order based on means. The lowest mean value (1.51) was obtained for the statement "I often have to borrow at the end the month", while the highest (4.27) for "I know exactly how much money I have in cash and on my bank account". This latter equally has the lowest standard deviation value (1.031), suggesting a convergence of respondents' answers in this regard. The highest standard deviation value (1.465) was observed for the "Bills are killing me" item. Approximately 45 percent of respondents indicated that economizing was totally true for them when they had little money, and also, that they did not like to throw out still usable things, which we also evaluated in a positive way, as the former reflects an economizing character, while the latter reflects the proper assessment of values. In addition, more than 30 percent of respondents thought it was completely true for them that they controlled their spending, always had enough savings, compared the prices in shops where they took shopping lists compiled with proper consideration. Table 2: Questionnaire item scores (1 – strongly disagree, 5 – strongly agree) Descriptive statistics Mean St.dev. I often have to borrow at the end the month. 1.51 1.056 I enjoy going shopping with friends. 1.70 1.060 I am in a desperate fix with debts. 1.82 1.284 I am puzzled about where your money goes. 1.94 1.218 If I pay in cash I never ask for the change. 1.95 1.114 Only when I clean up I realize the amount of my unnecessary purchases. 1.99 1.100 I sometimes end up paying a few bills late. 2.03 1.368 I enjoy trying my luck. 2.08 1.141 I often surprise my loved ones with self-made gifts. 2.15 1.151 I don’t enjoy cooking, we rather eat ready meals. 2.16 1.268 Bills are killing me. 2.43 1.465 If I like something, I buy it. 2.53 1.109 When grocery shopping, I am often surprised how much I have to pay at the end. 2.56 1.202 I love trendy things. 2.58 1.250 I have a hard time resisting when I am offered something at a great price. 3. Methodology The methodology based on a standard questionnaire allows for comparable results among countries as well as for an examination of time series data. Data was collected with CAPI (Computer-Aided Personal Interviewing). As numerous demographic data were queried, one can state that respondents' age varied between 18 and 79 years. 1000 individuals were surveyed, representing the population of Hungary by age, gender and type of settlement. The advantage of personal interviews is that respondents are likely to complete the survey and without the possibility of "running through" the questionnaire (even without reading it) as the interviewer is in control of the pace by reading out loud the questions and statements of the questionnaire. Its disadvantage, however, is that respondents may be prone to provide answers that they assume are expected of them, thereby trying to meet social expectations, potentially reducing the proportion of honest answers (Malhotra & Simon, 2009; Atkinson & Messy, 2012). Financial attitudes, time orientation, and money-related emotions are measured through a 12-item part of the questionnaire. Respondents had to decide the extent to which they agreed with each statement using a five- point Likert-scale where 1 indicated "strongly agree", and 5 indicated "strongly disagree". This direction of this scaling is the opposite of the one used for the financial personality test presented beforehand. Thus, in order to facilitate the analysis and provide comparability, responses gathered through the OECD questionnaire were reverse-coded in order for higher scores to represent higher respondent agreement. As these statements overlap with certain statements of the 36-item personality test in some respect, the present study examines and then compares the results of the two queries. The 12 relevant items of the OECD questionnaire were grouped into three factors as a result of a factor analysis (main component analysis with varimax rotation). K-means clustering was then used based on factor averages and resulted in the identification of 3 respondent segments. K-means clustering was chosen for its stability for a large sample size (Sajtos & Mitev, 2007). 49 49 Journal of Economics and Behavioral Studies (ISSN: 2220-6140) Vol. 9, No. 2, pp. 46-57, April 2017 Results of the OECD financial literacy study: The 12 items of the OECD pertaining to personal finances were first examined in terms of the resulting mean values (Table 3). 4. Results 2.67 1.216 I have a few bad habits that cost me a lot of money. 2.69 1.279 I often reward myself. 2.69 1.083 I enjoy going out with my friends. 2.79 1.333 I spend a lot on healthy food and mineral water. 2.81 1.178 Sometimes, when shopping, I spend more than I previously expected. 3.03 1.114 I like it when it’s warm at the apartment. 3.10 1.156 When I need more money, I take up extra work. 3.13 1.417 I prefer preparing sandwiches rather than shopping at the cafeteria. 3.20 1.411 I always have enough savings for unexpected expenses. 3.39 1.461 I know exactly the price of everything. 3.44 1.140 I tend to browse a lot before purchasing a product. 3.52 1.243 I want to provide everything for my children. 3.60 1.271 Table 2: Questionnaire item scores (1 – strongly disagree, 5 – strongly agree) 50 Journal of Economics and Behavioral Studies (ISSN: 2220-6140) Vol. 9, No. 2, pp. 46-57, April 2017 8. I keep my household in order. 3.61 1.102 9. I am good at rationing my money. 3.61 1.131 0. I keep good track of my expenses. 3.64 1.263 1. I always make a shopping list. 3.65 1.360 2. Before going shopping, I always carefully think through of what I need. 3.83 1.088 3. I always compare prices before purchasing anything. 3.93 1.096 4. I only save on my expenses when I am short of money. 3.98 1.176 5. I don’t like throwing out things that still can be used. 4.08 1.066 6. I know exactly how much money I have in cash and on my bank account. 4.27 1.031 A questionnaire item can be considered divisive when all possible answers (1, 2, 3, 4, or 5) are provided by at least 14% of respondents each. Among the 36 statements the three following turned out to be most divisive:  I prefer preparing sandwiches rather than shopping at the cafeteria.  I always have enough savings for unexpected expenses.  I always have enough savings for unexpected expenses. 4. Results 51 Journal of Economics and Behavioral Studies (ISSN: 2220-6140) Vol. 9, No. 2, pp. 46-57, April 2017 Table 3: Questionnaire item scores (1 – strongly disagree, 5 – strongly agree) Descriptive statistics Mean St.dev. I am willing to risk some of my money when it comes to savings or investment 1.6653 1.01695 I currently have too many debts 1.9186 1.26650 I enjoy spending money more than saving it for later 2.1903 1.18502 I rather live a day at a time – I’ll manage tomorrow somehow 2.2432 1.20116 I am satisfied with my current financial situation 2.4050 1.23633 I am often worried about my ordinary living expenses 3.0700 1.36050 Money is meant to be spent 3.1240 1.19501 I set up long-term financial goals and strive to achieve them 3.2265 1.31301 I am restricted by my financial situation in doing things I consider important 3.5180 1.29080 0. I personally and carefully monitor my finances 3.5681 1.31912 1. Before purchasing something I carefully consider whether I can afford it 4.1782 .99108 2. I pay my bills in time 4.2863 1.01776 A factor analysis (main component analysis with varimax rotation) was conducted for the 12 items. The analysis yielded the three factors presented in Table 4. The worries and dissatisfaction with regards to finances appear in the first factor, which thus regrouping variables in connection with negative feelings about finances. It is important to note that the item related to satisfaction loads negatively in the factor structure, hence indicating dissatisfaction. The second factor gathers statements that represent a certain present- hedonistic (Zimbardo & Boyd, 1999) attitude. A "carpe diem" attitude characterizes this factor, with recklessness, a certain negligence towards obligations (e.g. paying bills), and spending prevailing over building up reserves. The third factor includes items on consciousness, i.e. having goals and monitoring finances. The willingness to take risks equally appears here, only in connection with savings and investment. The three factors are suitable for a K-means cluster analysis to be performed on the averages on the related scale item scores in order to divide respondents into groups. 4. Results Financial personality types: The factor analysis (main component analysis with varimax rotation) conducted on the 36 statements yielded 9 factors, namely: nality types: The factor analysis (main component analysis with varimax rotation) conducted Financial personality types: The factor analysis (main component analysis with varimax rota  Economizer with little money  Economizer with little money  Money-devourer (opposite of Moderate)  Order creates value  Price sensitive  Collector  Planner  Ups and downs  Diligent  Cannot control finances The dimension of economizer with little money includes people who have trouble managing their finances, most of them struggle with debts, but at the same time and as opposed to it, it may happen that they also have some savings. Among the characteristic features of money-devourers, it is primarily the short-term features that dominate – they love to have fun, they immediately buy what they like, they love shopping and often reward themselves. Moreover risk-taking is also present among them. Respondents performing well in the order creates value factor keeps track of his expenses, knows exactly when and how much money he has, and from this, it partly comes that he keeps his home and household tidy, and before shopping, always thinks over what he needs. The price-sensitive dimension contains people for whom it is most typical that they compare prices before shopping, and as a result, are able to take their time in selecting the articles. Collectors take advantage of sales and try to amass everything. They do not necessarily keep their environment tidy, but when they do, they realize how many unnecessary things they have. Planners make lists before going shopping, i.e. they tend to carefully plan their purchases. In the ups and downs dimension, savers and spenders appear alongside each other. The central organizing principle of the diligent dimension is work, in connection with which individuals assess the acquired income, and as a consequence, they are able to appreciate it. Individuals who cannot control finances are not able to appreciate the real value of the goods they wish to consume, and over the short term it means that they are surprised at the amount they have to pay at the cash- desk. Results of the OECD financial literacy study: The 12 items of the OECD pertaining to personal finances were first examined in terms of the resulting mean values (Table 3). 4. Results Table 4: The three factors and item factor scores as identified by the factor analysis Results of the factor analysis Component 1 2 3 I am often worried about my ordinary living expenses .840 -.005 -.086 I am restricted by my financial situation in doing things I consider important .810 .022 -.155 I am satisfied with my current financial situation (R) .765 -.162 -.205 I currently have too many debts .592 .309 .336 I enjoy spending money more than saving it for later -.033 .774 -.062 I rather live a day at a time – I’ll manage tomorrow somehow .027 .725 -.168 Money is meant to be spent .100 .567 -.112 Before purchasing something I carefully consider whether I can afford it (R) -.379 .532 .026 I pay my bills in time etc. (R) .462 .520 .060 I set up long-term financial goals and strive to achieve them -.154 -.271 .739 I personally and carefully monitor my finances .004 -.304 .635 I am willing to risk some of my money when it comes to savings or investment -.138 .410 .597 R: reverse-coded Table 4: The three factors and item factor scores as identified by the factor analysis Results of the factor analysis Compo 4: The three factors and item factor scores as identified by the factor analysis ts of the factor analysis three factors and item factor scores as identified by the factor analysis R: reverse-coded 52 Journal of Economics and Behavioral Studies (ISSN: 2220-6140) Vol. 9, No. 2, pp. 46-57, April 2017 Cluster analysis: The clusters obtained by the K-means cluster analysis provide a deeper insight and a more detailed picture of the results by the help of the combination of factors. The cluster analysis was performed using the averages of the item scores of each factor. Sizes and centres of the three resulting clusters are shown in Table 5. Journal of Economics and Behavioral Studies (ISSN: 2220-6140) Vol. 9, No. 2, pp. 46-57, April 2017 Cluster analysis: The clusters obtained by the K-means cluster analysis provide a deeper insight and a more detailed picture of the results by the help of the combination of factors. The cluster analysis was performed using the averages of the item scores of each factor. Sizes and centres of the three resulting clusters are shown in Table 5. 4. Results Table 5: Cluster centres in the three clusters based on factors Final cluster centres Cluster 1 (N=305) 2 (N=359) 3 (N=300) Carpe Diem factor 8.74 9.47 15.17 Self-conscious factor 8.06 9.26 8.05 Worried, dissatisfied factor 15.52 8.21 13.28 Table 5: Cluster centres in the three clusters based on factors The first cluster is least likely to reach a high score at the "carpe diem" factor, while at the same time they are most prone to providing high scores for the items of the worried-dissatisfied factor. As a consequence, this group of respondents can be characterized as being anxious about their finances, who dare not live for the day or spend money irresponsibly. This, however, does not entail a higher level of consciousness. The second cluster performs highest in consciousness, while in terms of negative feelings, they are the least anxious or dissatisfied about their finances. Compared to the other clusters, the third cluster performs remarkably high in the "carpe diem" factor. These respondents are thus most likely to live a day at a time, but also to worry about their finances, even though not as much as those respondents who belong in the first cluster. Next, the average scores in each factor of the focal 12 items were examined. Table 7 contains the item scores within the full sample and within each factor. Based on the mean values, Table 6 gives an overview of the main attributes of each cluster. Table 6: Description of clusters Cluster 1 Anxious unsatisfied Cluster 2 Satisfied conscious Cluster 3 Moderately anxious unconsidered • careful in their spending • least likely to live a day at a time • do not prefer spending over savings • lowest risk-taking willingness • most unsatisfied • most anxious • most perceiving their financial situation as prohibitive • most likely to pay their obligations in time • most likely to set up financial goals and control their finances • least anxious • least likely to have debt and to perceive their financial situation as prohibitive • most satisfied • live a day at a time • least likely to formulate and set up financial goals • least likely to monitor and control their finances • prefer spending over savings • least likely to pay their obligations (e.g. 4. Results bills) in time • highest risk-taking willingness • feel crushed by their debts Table 7: Item scores within the full sample and within factors Descriptives N Mean Std. Dev. Before purchasing something I carefully consider whether I can afford it 1 305 4.7574 .59062 2 359 4.1253 .93856 3 300 3.6567 1.05633 Total 964 4.1795 .98796 I rather live a day at a time – I’ll manage tomorrow somehow 1 305 1.6590 .92931 2 359 1.8301 .92549 3 300 3.2567 1.01697 Table 6: Description of clusters Table 6: Description of clusters Cluster 2 Satisfied conscious Table 7: Item scores within the full sample and within factors Descriptives Table 7: Item scores within the full sample and within factors Table 7: Item scores within the full sample and within factors Descriptives N Mean Std. Dev. Before purchasing something I carefully consider whether I can afford it 1 305 4.7574 .59062 2 359 4.1253 .93856 3 300 3.6567 1.05633 Total 964 4.1795 .98796 I rather live a day at a time – I’ll manage tomorrow somehow 1 305 1.6590 .92931 2 359 1.8301 .92549 3 300 3.2567 1.01697 53 Journal of Economics and Behavioral Studies (ISSN: 2220-6140) Vol. 9, No. 2, pp. 46-57, April 2017 Journal of Economics and Behavioral Studies (ISSN: 2220-6140) Vol. 9, No. 2, pp. 46-57, April 2017 Total 964 2.2199 1.18461 I enjoy spending money more than saving it for later 1 305 1.5213 .76113 2 359 1.8357 .92347 3 300 3.2467 1.05334 Total 964 2.1753 1.17493 I pay my bills in time, etc. 1 305 4.3902 .92582 2 359 4.8106 .45193 3 300 3.6033 1.14178 Total 964 4.3019 1.00112 I am willing to risk some of my money when it comes to savings or investment 1 305 1.3246 .74088 2 359 1.6880 1.02899 3 300 2.0033 1.14061 Total 964 1.6712 1.02145 I personally and carefully monitor my finances 1 305 3.6328 1.39651 2 359 3.8942 1.10860 3 300 3.1867 1.29773 Total 964 3.5913 1.29660 I set up long-term financial goals and strive to achieve them 1 305 3.0984 1.34635 2 359 3.6769 1.19182 3 300 2.8600 1.22695 Total 964 3.2396 1.30007 Money is meant to be spent 1 305 2.7115 1.24949 2 359 2.7409 1.03420 3 300 3.9233 .86407 Total 964 3.0996 1.19554 I am restricted by my financial situation in doing things I consider important 1 305 4.4492 .81002 2 359 2.3760 1.00312 3 300 3.8867 .95754 Total 964 3.5021 1.29180 I am often worried about my ordinary living expenses 1 305 4.1279 .99673 2 359 1.9694 .91947 3 300 3.3367 1.13178 Total 964 3.0778 1.36153 I currently have too many debts 1 305 2.3574 1.44646 2 359 1.2006 .54301 3 300 2.3733 1.31137 Total 964 1.9315 1.27337 I am satisfied with my current financial situation 1 305 1.4164 .71662 2 359 3.3398 .87571 3 300 2.3133 1.17747 Total 964 2.4118 1.23109 54 5. Conclusion Modern economics have recognized that in order to effectively forecast financial and economic processes it is primordial to understand the attitudes of the members of society toward finances, as well as the characteristics of various social group sharing the same views and behaviours. The aim of the present study was to compare the relevant parts of two studies conducted in 2015 focusing on financial attitudes and behaviour, and therefore on financial personality types. The theoretical framework reviews various scales developed since as early as the 1970s. Both the study of financial personality types and the OECD study verify and overlap, to varying degrees, with the results of this framework. The nine financial personality profiles can be matched with the three clusters identified in the OECD study. The clusters identified during the research show idiosyncratic financial and psychological vulnerability and/or protection. The main conclusion is that self-consciousness in finances is associated with an emphasis on order, planning and diligence. Individuals in the group that does so are also most satisfied with their current financial situation. A combination of these attributes can, therefore, provide a protection of sorts both financially and mentally. The literature review hints that concern and negative feelings are recurring dimensions of individual finances. The group of anxious identified during the cluster analysis covers individuals who are characterized by usually having little money, price sensitivity and a collecting behaviour. Moreover, they consider that they economize well; still they are dissatisfied with their financial situation. The source of their dissatisfaction can be traced to low income and to the feeling of vanity in trying to make do with their finances. Here, amassing and anxiety are a cause of financial and mental vulnerability, respectively. Individuals in the third group, unconsidered, cannot control their finances and have a lot of purposeless expenditures. This group is in fact the opposite of Yamauchi and Templer's (1982) retention-time dimension. In the financial personality study the "order creates value" factor was identified to be the best indicator of one's financial awareness. This analogy, emphasized in the study, equally supports our earlier finding, as this factor is precisely most in line with the self-conscious factor. The second cluster, satisfied conscious, includes the order creates value, diligent and planner personalities. These dimensions were most highly regarded in the study in terms of financial personality. This is also the group with the highest levels of income. 5. Conclusion An interesting avenue for further research is the study of whether high levels of income necessarily lead to higher financial awareness and satisfaction, or, on the contrary, is it a positive and conscious attitude that leads to higher levels of income. Identifying the personality, attitude, and behavioural components of financial culture can help contribute to a better understanding of the causes and aspects of financial behaviours beyond rationality. The present study provides a synthesis of two studies performed in Hungary as well as of relevant parts of the literature. It contributes to the research on the broader concept of financial literacy and thereby can serve as an input for further studies as well as for the development of programs on improving financial literacy. Brozynski, T., Lukas, M. & Ulrich, S. (2004). The Impact of Experience on Risk Taking, Overconfidence and Herding of Fund Managers: Complementary Survey Evidence. Discussion Paper No: 292. University of Hannover School of Economics and Management. Botos, K., Botos, J., Béres, D., Csernák, J. & Németh, E. (2012). Financial literacy and Risk-Taking of Households in the Hungarian Central Great Plain. Public Finance Quarterly, 57(3), 267-285 Borghans, L., Angela, L. D., James, J. H. & Baster, W. (2008). The Economics and Psychology of Personality Traits, Discussion Paper No: 3333. Forschungsinstitut zur Zukunft der Arbeit Institute for the Study of Labor. Journal of Economics and Behavioral Studies (ISSN: 2220-6140) Vol. 9, No. 2, pp. 46-57, April 2017 Comparison of the results: The first factor based on the items of the OECD questionnaire (anxious unsatisfied) echoes with the economizer with little money factor yielded by the study on financial personalities. This is the dimension of money avoiders that points out that respondents performing highly here are likely to have problems related to the handling of their finances and also to struggle with debt. The second factor, "carpe diem" shows similarities with the money-devouring factor. Results of the research on financial personalities suggest a certain short-sightedness, as well as shopping and self-rewarding as core values for this factor. Thus, here, spending money dominates over saving it. The third factor regrouping the dimensions of objectives and control shows resemblance to the order creates value factor. This is a dimension with positive views and pertains to respondent's level of prudence by revealing how they keep track of their finances. The similarities between the two factor analyses are illustrated in Figure 1. Figure 1: Relationships between the factors of the two studies (source: own elaboration) Figure 1: Relationships between the factors of the two studies (source: own elaboration) The three clusters identified based on the OECD database cover – both in contents and in their characteristics – all nine identified financial personality types. The first cluster, anxious unsatisfied, regroups the economizers with little money, price sensitive, and collector personality types. The second cluster, satisfied conscious, includes the order creates value, diligent and planner personalities. The third cluster, moderately anxious unconsidered, covers the money-devouring, ups and downs and cannot control finances personalities. Figure 2: Financial personality types covered by the clusters of the OECD study (source: own elaboration) Figure 2: Financial personality types covered by the clusters of the OECD study (source: own elaboration) 55 Journal of Economics and Behavioral Studies (ISSN: 2220-6140) Vol. 9, No. 2, pp. 46-57, April 2017 References A Critical Review of Capital Information Management and Business Review, 4(11), 553-557 Jureviciene, D. & Jermakova, K. (2012). The Impact of Individuals’ Financial Behaviour on Investment Decisions. Annual meeting fort he Society of 1 st. Electronic International Interdisciplinary Conference, Slovakia, September 3-7. Kübilay, B. & Bayrakdaroğlu, A. (2016). An Empirical Research on Investor Biases in Financial Decision- Making, Financial Risk Tolerance and Financial Personality. International Journal of Financial Research, 7(2), 171-182. Malhotra, N. K. & Simon, J. (2009). Marketingkutatás [Marketing research]. Akadémiai Kiadó Zrt., Budapest. Mellan, O. (1994). Money Harmony: Resolving Money Conflicts In Your Life and Relationships. Walker & Company, New York Magyar Nemzeti Bank [the Central Bank of Hungary]. (2014). Tudás és érték, a Magyar Nemzeti Bank társadalmi felelősségvállalási stratégiája [Knowledge and values, corporate social responsibility strategy of the central bank of Hungary]. Magyar Nemzeti Bank, June 2014, https://www.mnb.hu/letoltes/mnb-csr-kiadvany-eng-final-1.pdf McCrae, R. R. (2009). The Five-Factor Model of Personality Traits: Consensus and Controversy. In Philip J. Corr and Gerald Matthews (Eds.), The Cambridge Handbook of Personality Psychology (pp. 148-162). Cambridge: Cambridge University Press. Németh, E., Béres, D., Huzdik, K. & Zsótér, B. (2016). Financial personality types in H methods and results. Financial and Economic Review, 15(2), 153-172. Opsahl, R. L. & Dunnette, M. D. (1966). The role of financial incentives in industrial motivation. Psychological Bulletin, 66, 95–116. Sajtos, L. & Mitev, A. (2007). SPSS kutatási és adatelemzési kézikönyv [SPSS Research and Data Analysis Manual]. AlineaKiadó, Budapest. Sproles, G. B. & Kendall, E. L. (1986). A methodology for profiling consumers' decision-making styles. Journal of Consumer Affairs, 20(2), 267-279. Tang, T. L. P. (1992). The Meaning of Money Revisited. Journal of Organizational Behaviour, 13(2), 197-202. Wernimont, P. F. & Fitzpatrick, S. (1972). The meaning of money. Journal of Applied Psychology, 56 (3), 218- Tang, T. L. P. (1992). The Meaning of Money Revisited. Journal of Organizational Behaviour, (1992). The Meaning of Money Revisited. Journal of Organizational Behaviour, 13(2), 197-202 Tang, T. L. P. (1992). The Meaning of Money Revisited. Journal of Organizational Behaviour, 13(2), 197-202. Wernimont, P. F. & Fitzpatrick, S. (1972). The meaning of money. Journal of Applied Psychology, 56 (3), 218- 226. Wernimont, P. F. & Fitzpatrick, S. (1972). The meaning of money. Journal of Applied Psycholo 226. Yamauchi, K. T. & Templer, D. J. (1982).The Development of a Money Attitude Scale. Journal of Personality Assessment, 46(5), 522-528. ( ) Zimbardo, P. G. & Boyd, J. N. (1999). References Atkinson, A. & Messy, F. A. (2012). Measuring Financial Literacy: Results of the OECD / International Network on Financial Education (INFE) Pilot Study, OECD Working Papers on Finance, Insurance and Private Pensions, No. 15, OECD Publishing, Paris. Atkinson, A. & Messy, F. A. (2012). Measuring Financial Literacy: Results of the OECD / International Network on Financial Education (INFE) Pilot Study, OECD Working Papers on Finance, Insurance and Private Pensions, No. 15, OECD Publishing, Paris. Béres, D., Huzdik, K., Németh, E. & Zsótér, B. (2015.) Pénzügyi személyiség. A magyar lakosság pénzügyi magatartása, szokásai és attitűdjei [Financial personality. Financial behaviour, habits and attitudes of the Hungarian population]. Research report. Financial Compass Foundation (Pénziránytű Alapítvány), Budapest. Borghans, L., Angela, L. D., James, J. H. & Baster, W. (2008). The Economics and Psychology of Personality Traits, Discussion Paper No: 3333. Forschungsinstitut zur Zukunft der Arbeit Institute for the Study of Labor. Botos, K., Botos, J., Béres, D., Csernák, J. & Németh, E. (2012). Financial literacy and Risk-Taking of Households in the Hungarian Central Great Plain. Public Finance Quarterly, 57(3), 267-285 Brozynski, T., Lukas, M. & Ulrich, S. (2004). The Impact of Experience on Risk Taking, Overconfidence and Herding of Fund Managers: Complementary Survey Evidence. Discussion Paper No: 292. University of Hannover School of Economics and Management. 56 Journal of Economics and Behavioral Studies (ISSN: 2220-6140) Vol. 9, No. 2, pp. 46-57, April 2017 Chan, S. Y. F. (2003). The exploratory relationship between money attitude and consumer style. Australian Journal of Psychology, 55, 119-119. f y gy Forman, N. (1987). Mind over money. Toronto: Doubleday. Forman, N. (1987). Mind over money. Toronto: Doubleday. Furnham, A. F. (1984). Many sides of the coin: The psychology of money usage. Personality and Individual Differences, 5(5), 501-509. Furnham, A. (1999). The saving and spending habits of young people. Journal of Economic Psychology, 20 (6), 677-697. Goldberg, L. R. (1971). A Historical Survey of Personality Scales and Inventories. In Paul McReynolds (Ed.), Advances in Psychological Assessment, 2, 293-336. USA: Science and Behavior Books Inc. Goldberg, H. & Lewis, R. T. (1978). Money Madness: The Psychology of Saving, Spending, Loving and Hating Money. William Morrow and Co., New York. Huzdik, K., Béres, D. & Németh, E. (2014). An Empirical Study of Financial Literacy versus Risk Tolerance Among Higher Education Students. Public Finance Quarterly, 59(4), 476-488. Iqbal, S. M. J., Muneer, S., Jahanzeb, A. & Rehman, S. U. (2012). References Putting Time in Perspective: A Valid, Reliable Individual-Differences Metric. Journal of Personality and Social Psychology, 77(6), 1271-88. Zsótér, B., Németh, E. & Béres D. (2016). A pénzügyi kultúra személyiség-, attitűd- és viselkedésbeli komponensei – Összehasonlító elemzés. [Components of financial literacy from the perspective of personality, attitudes and behaviour. Comparative study]. Kultúra és Közösség / Culture and Community, 41(3)89-102. 57
https://openalex.org/W4293124894	https://link.springer.com/content/pdf/10.1007/s12518-022-00457-9.pdf	English	null	GNSS-SDR pseudorange quality and single point positioning performance assessment	Applied geomatics	2,022	cc-by	10,158	Abstract In recent years, we have witnessed a growing demand for GNSS receiver customization in terms of modification of signal acquisition, tracking, and processing strategies. Such demands may be addressed by software-defined receivers (SDRs) which refers to an ensemble of hardware and software technologies and allows re-configurable radio communication architectures. The crux of the SDRs is the replacement of the hardware components through software modules. In this paper, we assess the quality of GNSS observables acquired by SDR against the selected u-blox low-cost receiver. In the following, we investigate the performance level of single point positioning that may be reached with an ultra-low-cost SDR and compare it to that of the low-cost GNSS receiver. The signal quality assessment revealed a comparable performance in terms of carrier-to-noise density ratio and a significant out-performance of the u-blox over SDR in terms of code pseudorange noise. The experimenta- tion in the positioning domain proved that software-defined receivers may offer a position solution with three-dimensional standard deviation error at the level of 5.2 m in a single point positioning mode that is noticeably poorer accuracy as com- pared to the low-cost receiver. Such results demonstrate that there is still room for SDR positioning accuracy improvement. Keywords GNSS · Software-defined receivers · Low-cost receivers · Signal assessment · Single poin GNSS‑SDR pseudorange quality and single point positioning performance assessment Umberto Robustelli1 · Matteo Cutugno1 · Jacek Paziewski2 · Giovanni Pugliano3 Received: 17 May 2021 / Accepted: 1 August 2022 © The Author(s) 2022 / Published online: 26 August 2023 * Umberto Robustelli umberto.robustelli@uniparthenope.it * Umberto Robustelli umberto.robustelli@uniparthenope.it 1 Department of Engineering, Parthenope University of Naples, 80133 Naples, Italy 2 Department of Geodesy, Faculty of Geoengineering, University of Warmia and Mazury in Olsztyn, Oczapowskiego 1, 10‑719 Olsztyn, Poland 3 Department of Civil, Architectural and Environmental Engineering, University of Naples Federico II, via Claudio 21, 80125 Naples, Italy https://doi.org/10.1007/s12518-022-00457-9 Applied Geomatics (2023) 15:583–594 https://doi.org/10.1007/s12518-022-00457-9 Applied Geomatics (2023) 15:583–594 ORIGINAL PAPER Data collection and experiment design To assess the performance of single point positioning and to evaluate the quality of the code pseudoranges acquired by the software-defined receiver, we employed the GPS L1 band observations collected on March 1, 2021. The test site with known ground-truth coordinates was located in Naples (Italy), as shown in panel d of Fig. 1. The site is expected to be an open-sky and low-multipath environment. The ground-truth coordinates of the site were determined in a static positioning using GNSS observations collected by a surveying-grade receiver and antenna. p p g p Regarding processing enhancements exploiting GPU- based techniques, Knežević et al. (2010) developed an 8-channel GPS SDR capable of processing 40 Msps and 8-bit resolution data in real time using a single-core 3.0- GHz CPU and an NVIDIA GeForce 8800 GTX GPU, while in Hobiger et al. (2009) a real-time GPS SDR supporting 12 channels with 8 Msps and 4-bit resolution data using an Intel Core 2 Q9450 CPU and an NVIDIA GeForce GTX 280 GPU has been developed. In the FOSS scenario, few solutions are made available; among them, GNSS-SDR, developed by the CTTC (Centre Tecnològic Telecomunicacions Catalunya), seems to be the best FOSS for GNSS signal processing in terms of approachability to the code and activity of devel- oper’s community; it is hosted on the free platform github. com where anyone can fork the repository and create a copy to test and develop the algorithms. GNSS-SDR is capable of acquiring, processing, and computing navigation solutions for different kinds of constellations (GPS, GLONASS, and Galileo) as reported in Fernández-Prades et al. (2011). To assess the software-defined receiver performance against a low-cost COTS receiver, the UBX zed-f9 shar- ing the same GNSS active antenna was employed simulta- neously. The UBX data logging was accomplished via the u-center evaluation software, which stores data in the propri- etary .ubx format. This was later converted in post-mission with open-source RTKconv utility. In the analysis, we used a common period and the same sampling interval of 1 s for observations collected by both receivers. In specific, the time window starts at 16:59:21 UTC and ends at 17:56:35 UTC. The hardware needed to log data with SDR reduces to an active antenna, a bias-tee, an RF front-end, and a comput- ing platform. The former, namely a UBX ANN-MB active GNSS antenna shown in panel a of Fig. Introduction (2003b) developed a 12-channel real- time GPS software receiver and then the same authors deter- mine its accuracy and tracking performance under dynamic conditions in Ledvina et al. (2003a). In Seo et al. (2011), a GPS software-defined radio with adaptive beam steering capability for anti-jam applications with massively parallel processor computing has been presented. “Signal quality assessment” section, we comprehensively evaluate the quality of code GNSS observations collected by SDR against that of the low-cost receiver. In the “Code observation noise” section, we provide the theoretic back- ground of the code-minus-carrier analysis. In the “Single point positioning performance assessment” section, we investigate the performance of single point positioning (SPP) with SDR and low-cost receiver observations. Lastly, the “Discussion” section provides the discussion of the results. Introduction boxes. A few years ago, the scenario suddenly changed with the advent of two decisive developments: software-defined radio technology and the concept of Free-and-Open-Source- Software (FOSS). The latter consists of distributed develop- ment processes where everyone can contribute to the pro- ject and modify and redistribute it. In this perspective, the software-defined approach can define useful tools both for research purposes and low-cost industrialized commoditi- zation. Software-defined receivers (SDRs) are capable of extreme customization, allowing users to access, visualize, and modify signal acquisition, tracking, and processing strategies. SDR refers to an ensemble of hardware, software technologies, and design choices allowing re-configurable radio communication architectures. The main idea behind this concept is the replacement of the dedicated hardware components through software modules. Nowadays, there are some software solutions in GNSS positioning that follow the SDR approach, among all: PLAN research group from Calgary University has proposed the GSNRx (GNSS soft- ware navigation receiver) in Petovello et al. (2008), Munich University has created ipexSR in Anghileri et al. (2001), the ISMB in collaboration with Polytechnic University of Turin have proposed N-Gene in Fantino et al. (2009). Moreover, A GNSS receiver suitable to face GPS modernization should be re-configurable and flexible in design so that the possibil- ities of new specifications and algorithms can be exploited, and the price should be low enough to enable mass-market interests (Paziewski (2020)). For the latter, global industri- alization provides an awesomely wide range of hardware keeping costs low. The downside of industrialized tech- nology is the fact that, in the last years, it has somehow constrained academic research raising legal and economic barriers; this results in GNSS receivers being seen as black * Umberto Robustelli umberto.robustelli@uniparthenope.it 3 Department of Civil, Architectural and Environmental Engineering, University of Naples Federico II, via Claudio 21, 80125 Naples, Italy (0123 1 3456789) 3 Applied Geomatics (2023) 15:583–594 584 Abbasiannik (2009) and Kang et al. (2002) have developed a combined GPS and GLONASS software receiver capable of providing a position solution. Lin et al. (2011) proposed a vector-based high sensitivity software receiver and its ultra- tight version. Schmidt et al. (2018) presented a LabVIEW (LV) and C/C++-based GPS L1 receiver platform with real-time capabilities. Hurskainen et al. (2009) described a multicore software-defined radio architecture for Global Navigation Satellite System (GNSS) receiver implementa- tion. Ledvina et al. Data collection and experiment design Panel a shows the UBX ANN-MB GNSS active antenna; panel b shows the u-blox zed-f9p; panel c shows the RTL-dongle, namely the Rafael micro-sdr; and panel d shows the site of acquisition (Eremo ai Camaldoli, Naples, Italy) Table 1 Main features of the ANN-MB active GNSS antenna ANN-MB L1 band Frequency (MHz) 1559–1606 Impedance (Ohm) 50 Peak gain (dBic) Typ.3.5 Gain (no cable) (dB) 25–31 Noise figure (dB_ 2.8 DC voltage (V) 3–5 Polarization RHCP Price (€) 55 All settings for the software-defined receiver were defined in the configuration file. Receiver general settings, such as acquisition rate and signal source, were defined first. More specialized settings for each block of the processing chain, such as acquisition and tracking settings, were specified thereafter. The settings file also defines which signals the receiver will attempt to acquire and track, how many chan- nels will be created, and which output will be stored. The output consists in several files of different format, such as .nav and .obs RINEX 3.03, and .kml. Even raw data from each block can be stored. The official documentation is pro- vided by the developers in CTTC (2021). For the details of the parameters settings in a similar application, one can refer to Cutugno et al. (2019) and Cutugno et al. (2020). temperature drift. It has lower RF insertion losses and low power-supply needs; it seems also durable, indeed, as shown in panel a of Fig. 1, and it is enclosed in an alu- minium case. The poor quality of the front-end forced us to limit the duration of the experiment to 1 h. In the past, several tests with RTL-dongle have been carried out and we experienced that it starts to overheat after about 30 min and, after about 1 h, it begins to lose the lock of GPS sig- nals. This is probably due to the high temperatures reached by the clock, even if it is a TCXO and it should be compen- sated for temperature effects. Under these conditions, the oscillator starts to drift too much, and the receiver cannot track satellite signals anymore. Data collection and experiment design 1, which character- istics are reported in Table 1, was mounted on a 10-cm metal ground plane; it was connected via a SubMiniature version A (SMA) splitter to deliver the signal to both receivers: the UBX z-f9p (UBX) and RF front-end driven by GNSS-SDR, shown in panel b and panel c of Fig. 1, respectively. Taking into account the gain-loss introduced by signal splitting, a proper gain was provided by 4.5 V bias-tee. Following the work done in Cutugno et al. (2019) and Cutugno et al. (2020), this paper aims to assess the quality of the pseudoranges (PR) generated by an ultra-low-cost front- end driven by a software-defined receiver through the code- minus-carrier (CMC) analysis. For this reason, a low-cost commercial-off-the-shelf (COTS) receiver, namely the UBX z-f9p with a patch antenna (ANN-MB), has been employed as a term of comparison. The analysis carried out also aims to investigate the potential correlations between pseudorange errors and other parameters, e.g. the satellite elevation angle and the carrier-to-noise density ratio ( C∕N0 ). The analysis can be prodromic to enhance the positioning capabilities of ultra-low-cost SDR. The information on the SDR signal quality is useful for the development of the stochastic model for positioning with ultra-low-cost SDR. The RF front-end used for experimentation was the Rafael micro-SDR equipped with a Realtek RTL2832U chip; it is a cost-effective RF front-end available on the market at just 30 $. The RTL-dongle has limited connec- tivity and poor hardware capabilities (e.g. sampling rate and tuning ranges) compared to more expensive boards. It can perform tuning from 25 to 1700 MHz and it is equipped with a 1-PPM temperature compensated crys- tal oscillator (TCXO), accurate tuning, and 0.5–1-PPM This paper is organized as follows. In the “Data collec- tion and experiment design” section, we provide the descrip- tion of the data collection and the experiment design. In the 1 3 3 585 Applied Geomatics (2023) 15:583–594 Fig. 1 Hardware and site of acquisition. Panel a shows the UBX ANN-MB GNSS active antenna; panel b shows the u-blox zed-f9p; panel c shows the RTL-dongle, namely the Rafael micro-sdr; and panel d shows the site of acquisition (Eremo ai Camaldoli, Naples, Italy) Fig. 1 Hardware and site of acquisition. Signal quality assessment Analysis has been conducted in the same time window PRN n◦ obs SDR n◦ obs UBX 5 204 7 635 8 1366 10 3289 3435 15 357 16 3435 3435 18 3435 20 3435 3435 23 3435 3435 25 22 26 3435 3435 27 3435 3435 29 590 3094 31 1611 2988 Fig. 2 Number of tracked satel- lites and PDOP of the SDR; in blue on left axes the number of satellites while in red on right axes the PDOP. In the first, second, third, and fourth panels, the results for 5 ◦ , 10◦ , 15◦ , and 30◦ mask angles are shown Fig. 3 Mean signal-to-noise ratio values for the SDR and the UBX receivers in blue and red bars, respectively Table 2 Number of observations for each tracked satellite. The left column denotes the PRN, the central column indicates the number of observation for the SDR, and the right column the number of observation for the UBX. Analysis has been conducted in the same time window PRN n◦ obs SDR n◦ obs UBX 5 204 7 635 8 1366 10 3289 3435 15 357 16 3435 3435 18 3435 20 3435 3435 23 3435 3435 25 22 26 3435 3435 27 3435 3435 29 590 3094 31 1611 2988 hardware, cannot exploit more than 8 channels, according to the software developer’s recommendation. This configura- tion prevents the hardware to suffer from an extreme com- putational burden. Therefore, using 8 channels, it happens that some satellites cannot be received as all the channels are busy.f In Fig. 2, the number of tracked satellites for the different mask angles used is shown. It can be noticed that the number of satellites does not vary going from 5 to 10◦ mask angles. A small decrease can be seen with a mask angle of 15◦ while the decrease is clear with a mask angle of 30◦ . The reduction in the number of satellites has an effect on the PDOP which remains constant when 5 ◦ and 10◦ mask angles are set, wors- ens slightly for 15◦ , whereas it reaches values greater than 15 in specific time intervals when 30◦ mask angle is set. i Afterward, the signal-to-noise ratio has been investi- gated. Figure 3 reports a mean ( C∕N0 ) for GPS observa- tions acquired by the employed receivers. Signal quality assessment The analysis starts with the comparison of the tracked sat- ellites recorded in the observation file. Table 2 depicts the number of observations stored for each satellite. As one can notice, the SDR was unable to track satellites PRN 5, 7, 8, 15, 18, and 25. Furthermore, as reported in Table 2, the analysis of the number of observations suggests that only satellites PRN 16, 20, 23, 26, and 27 were tracked continu- ously by both receivers. Another considerable difference lies in the number of the observations of the satellites PRN 29 and 31. We note that the SDR manages to acquire only one-sixth and half of that of the low-cost UBX receiver, respectively. These results are limited by the capabilities of the front-end; indeed, the SDR, when coupled with this The minimum required software to operate with the RTL dongle is the osmoSDR library needed to interface with GNSS-SDR. Other useful utilities can be built along with the Osmocom driver. 1 1 3 3 Applied Geomatics (2023) 15:583–594 Applied Geomatics (2023) 15:583–594 hardware, cannot exploit more than 8 channels, according to the software developer’s recommendation. This configura- tion prevents the hardware to suffer from an extreme com- putational burden. Therefore, using 8 channels, it happens that some satellites cannot be received as all the channels are busy. In Fig. 2, the number of tracked satellites for the different mask angles used is shown. It can be noticed that the number of satellites does not vary going from 5 to 10◦ mask angles. A small decrease can be seen with a mask angle of 15◦ while the decrease is clear with a mask angle of 30◦ . The reduction in the number of satellites has an effect on the PDOP which remains constant when 5 ◦ and 10◦ mask angles are set, wors- ens slightly for 15◦ , whereas it reaches values greater than 15 in specific time intervals when 30◦ mask angle is set. Afterward, the signal-to-noise ratio has been investi- gated. Figure 3 reports a mean ( C∕N0 ) for GPS observa- tions acquired by the employed receivers. Comparable per- formance of both receivers in terms of C∕N0 for most of the satellites, with slightly higher values for the UBX receiver Table 2 Number of observations for each tracked satellite. Signal quality assessment The left column denotes the PRN, the central column indicates the number of observation for the SDR, and the right column the number of observation for the UBX. Analysis has been conducted in the same time window PRN n◦ obs SDR n◦ obs UBX 5 204 7 635 8 1366 10 3289 3435 15 357 16 3435 3435 18 3435 20 3435 3435 23 3435 3435 25 22 26 3435 3435 27 3435 3435 29 590 3094 31 1611 2988 Fig. 2 Number of tracked satel- lites and PDOP of the SDR; in blue on left axes the number of satellites while in red on right axes the PDOP. In the first, second, third, and fourth panels, the results for 5 ◦ , 10◦ , 15◦ , and 30◦ mask angles are shown Fig. 3 Mean signal-to-noise ratio values for the SDR and the UBX receivers in blue and red bars, respectively 586 586 hardware, cannot exploit more than 8 channels, according to the software developer’s recommendation. This configura- tion prevents the hardware to suffer from an extreme com- putational burden. Therefore, using 8 channels, it happens that some satellites cannot be received as all the channels are busy. In Fig. 2, the number of tracked satellites for the different mask angles used is shown. It can be noticed that the number of satellites does not vary going from 5 to 10◦ mask angles. A small decrease can be seen with a mask angle of 15◦ while the decrease is clear with a mask angle of 30◦ . The reduction in the number of satellites has an effect on the PDOP which remains constant when 5 ◦ and 10◦ mask angles are set, wors- ens slightly for 15◦ , whereas it reaches values greater than 15 in specific time intervals when 30◦ mask angle is set. Afterward, the signal-to-noise ratio has been investi- gated. Figure 3 reports a mean ( C∕N0 ) for GPS observa- tions acquired by the employed receivers. Comparable per- formance of both receivers in terms of C∕N0 for most of the satellites, with slightly higher values for the UBX receiver Table 2 Number of observations for each tracked satellite. The left column denotes the PRN, the central column indicates the number of observation for the SDR, and the right column the number of observation for the UBX. Code observation noise (3) CMCres ≃Ms r,P + 휖s r,P + 휖s ﬁlter (3) We employ the code-minus-carrier phase (CMC) linear com- bination (LC) to assess the combined impact of the noise and multipath effect on code observations and, therefore, high- light the differences in signal quality between the employed receivers. CMC LC reads as follows Therefore, “CMC residuals”, as expressed in (3), contain only code multipath, noise, and residual error of filtering. Since the antenna used is the same and the measurements have been made at the same time, we can state that the substantial difference is in the residual noise of the observables. Given that the acquisition of the SDR observables did not occur continuously, to correctly evaluate the CMCs, we must select the epochs where the satel- lites are seen continuously; for this reason, we will focus our attention on the satellites 16, 20, 23, 26, and 27. (1) Ps r −Φs r = 2Is r,i −휆Ns r,i + Ms r,P −Ms r,Φ + br,i −bs i −Br,i + Bs i + 휖s r,P −휖s r,Φ (1) where s states for the satellite; r is the receiver; Is r,i is the slant ionospheric delay in metres; Ns r,i states for the inte- ger ambiguity of phase observations in cycles; Ms r,Φ and Ms r,P denote the multipath effect in the unit of metres for phase and code observations, respectively; 휆i is the signal wavelength in metres on selected frequency i; Bs i and Br,i are the satellite and receiver phase delays in metres; bs i and br,i stand for the satellite and receiver code PR delays in metres, respectively; and finally, 휖s r,P and 휖s r,Φ are the obser- vation noise of code and phase observations, respectively. CMC LC exposes the noise and multipath of code observa- tions. All other unwanted effects such as phase ambigui- ties, satellite and receiver code, and phase biases, which are considered constant, as well as phase multipath and a doubled ionospheric delay, that are time-variant parameters, should be carefully handled. This may be done by making use of a moving average polynomial fitting or other filter- ing methods (see Peter de Bakker et al. (2012); Paziewski et al. (2021)). These terms occupy different bands in the fre- quency domain; the frequency spectrum of the ionospheric delay is lower than 0.1 mHz, so it can be considered bias (see Zhang and Bartone (2005)). Code observation noise If the cycle slips and clock jumps are detected and repaired, the ambiguity term can also be considered a bias (Aram et al. (2007); Delgado and Haag (2011)). Since the first two terms have a lower center frequency than the code multipath and the residual error, we can estimate them simply with a low-pass filter. The values Figure 4 shows the CMC residuals for both receiver and time; in particular, the SDR-related values are shown in the left column while the UBX residuals are in the right one. Each subfigure refers to a different satellite, namely satellites 16, 20, 23, 26, and 27 from the top. To ensure proper readability, two different scales have been employed for the Y-axis. In the fig- ure, the low-frequency component is of low magnitude; thus, we may attribute the time-series variability explicitly to the code noise. The CMC analysis highlights how the observables generated by the SDR under the same acquisition conditions (same antenna, same place) are affected by a noise two orders of magnitude greater than that afflicting the observables of the UBX, being the first in the interval [−20 20] m and the second in that [−0.25 0.25] m. Therefore, CMC analysis highlights that the observables generated by the SDR are much noisier than those from the UBX. It is clear that this noise directly propagates into the positioning domain. Signal quality assessment Comparable per- formance of both receivers in terms of C∕N0 for most of the satellites, with slightly higher values for the UBX receiver 1 3 Fig. 2 Number of tracked satel- lites and PDOP of the SDR; in blue on left axes the number of satellites while in red on right axes the PDOP. In the first, second, third, and fourth panels, the results for 5 ◦ , 10◦ , 15◦ , and 30◦ mask angles are shown Fig. 3 Mean signal-to-noise ratio values for the SDR and the UBX receivers in blue and red bars, respectively Fig. 2 Number of tracked satel- lites and PDOP of the SDR; in blue on left axes the number of satellites while in red on right axes the PDOP. In the first, second, third, and fourth panels, the results for 5 ◦ , 10◦ , 15◦ , and 30◦ mask angles are shown Fig. 3 Mean signal-to-noise ratio values for the SDR and the UBX receivers in blue and red bars, respectively Fig. 3 Mean signal-to-noise ratio values for the SDR and the UBX receivers in blue and red bars, respectively 1 3 Applied Geomatics (2023) 15:583–594 587 obtained can be subtracted from the CMCs to obtain the “CMC residuals”. After filtering unwanted terms, we have against the SDR one, has been discovered. Conversely, for satellites PRN 29 and 31, SDR shows higher C∕N0 as com- pared to the UBX ones; even though, this metric cannot be properly taken into account since it refers to means over different numbers of measurements; indeed, SDR meas- urements for satellites PRN 29 and 31 are 590 and 1611, respectively; on the contrary, the UBX achieved more the 3000 measurements for both satellites. Reviewing Fig. 3 and Table 2, one can notice that the satellites tracked only by the UBX receiver (PRN 5, 7, 8, 15, and 25, see Table 2) are characterized by the lowest signal gains. (2) CMCres = (Ps r −Φs r)ﬁl. = Ms r,P −Ms r,Φ + 휖s r,P −휖s r,Φ + 휖s ﬁlter (2) The carrier phase multipath error can reach a maximum value of a quarter of a cycle (approximately 4.8 cm for the L1), while the pseudorange multipath error can reach several metres for the C/A-code measurements in a highly reflective scenario, so that the term Ms r,Φ is negligible with respect to Ms r,P (El-Rabbany (2002)). Signal quality assessment In addition, the observation noise of code measurements is on the order of centimetres, while that of carrier phase measurements is on the order of millimetres, so this term is also negligible (Misra and Enge (2011)); thus, Single point positioning performance assessment As one can notice, the dominant error is the Up component for both receivers while planar errors are smaller for the UBX solutions.i f To compare the performances of the SDR and the UBX, we were obliged to use the single point positioning technique with the GPS-only signals given the limited capabilities of the ultra-low-cost front-end used to feed the SDR. There- fore, the performances are influenced by the limited number of satellites in view and by their non-optimal geometry, as shown in Fig. 2. The figures highlight that the SDR errors are higher than the UBX ones since they suffer from high noise as evidenced in the CMC analysis carried out in the previous section. The higher noise seen in the SDR measurements depends mainly on the limited capabilities of the ultra-low-cost front-end. The hardware components such as the oscillator are very cheap even if compared to the zed-f9p. Finally, regardless of the elevation mask angle adopted, the point clouds of the SDR solutions are always wider than those from the UBX.i Single point solutions were obtained with RTKLIB ver- sion 2.4.2. The ionosphere correction applied is the Klobu- char model while for the troposphere the Saastamoinen model has been adopted and, lastly, broadcast ephemeris were used. We can note that to properly compare the SDR and the UBX performances, SPP solutions were obtained by using the same satellites tracked by the SDR. To better assess the error distribution, the figures below show the 3D positioning error histograms for both receivers. Figure 10 shows the three-dimensional positioning error histogram for both the SDR and the UBX solutions. The results achieved by using the SDR observables are repre- sented in blue while those obtained with the UBX are in red. Each bar is 0.5 m wide. This figure confirms that the accuracy obtained by the UBX is higher than that obtained by the SDR; the centre of the SDR histogram has a greater abscissa than that of the UBX revealing a higher mean of the error. Moreo- ver, it emerges that the SDR histogram is smoother and wider with respect to the UBX one; this last indicates that the errors in the UBX solution are mainly included in the range of 1–3 The analysis starts from the comparison of coordinate errors in the time domain between the two receivers. Single point positioning performance assessment The analysis carried out in the previous section suggests that the SDR should be less performing than the UBX due to the higher presence of noise in the pseudoranges caused by the 1 3 3 Applied Geomatics (2023) 15:583–594 588 Fig. 4 Time evolution of CMC residuals. In the left panel, the plots of CMC residuals for the SDR are shown while on the right panel the CMC residu- als for the UBX are shown. Please note that two different scales have been employed for the Y-axis due to the different orders of magnitude between CMCs of the two receivers Fig. 4 Time evolution of CMC residuals. In the left panel, the plots of CMC residuals for the SDR are shown while on the right panel the CMC residu- als for the UBX are shown. Please note that two different scales have been employed for the Y-axis due to the different orders of magnitude between CMCs of the two receivers limited capabilities of the ultra-low-cost front-end used to feed it. To assess the SDR performance in the positioning domain, different operational conditions have been simulated by applying different elevation cut-off angles. The elevation cut-off angles considered for the analysis are 5 ◦ , 10◦ , 15◦ , and 30◦ . In this way, we assess the impact of low-elevated and low-gain signals that are more prone to multipath, as well as the positioning performance in an obstructed envi- ronment that is simulated by a 30◦ cut-off angle. satellites since there is only one satellite less after epoch 149500: 9 satellites before epoch 149500 and 8 satellites after. Figures 6, 7, 8, and 9 depict the three-dimensional scatter plots of the positioning errors for the SDR and the UBX for mask angles of 5 ◦ , 10◦ , 15◦ , and 30◦ , respectively. In par- ticular, in the left panel of Fig. 6, blue markers represent the positioning error of the SDR while red markers in the right panel are referred to the UBX receiver. Both panels refer to an elevation cut-off angle of 5 ◦ . 3 Single point positioning performance assessment Blue markers in the left panel represent errors of the SDR solutions; red markers in the right panel refer to errors obtained with the UBX. Both solutions refer to an elevation cut-off angle of 15◦ Fig. 9 3D positioning error scatter plots for single point positioning. Blue markers in the left panel represent errors of the SDR solutions; red markers in the right panel refer to errors obtained with the UBX. Both solutions refer to an elevation cut-off angle of 30◦ Fig. 10 3D positioning error histogram for the SDR and the UBX solutions. The results obtained with the SDR are represented in blue while those obtained with the UBX are represented in red. Each bar is 0.5 m wide sitioning error or single point Blue markers in the resent errors of the s; red markers in l refer to errors the UBX. Both r to an elevation of 15◦ 1 3 tough degradation of the results: the mean error is twice the mean error for the SDR solution of the previous cases (with elevation cut-off angles of 5 ◦ , 10◦ , and 15◦ ). On the other hand, the elimination of the low-elevated satellites provides an enhancement of the positioning precision. In specific, the UBX statistics show better performance when a 15◦ elevation cut-off angle is adopted as compared to the statistics that cor- respond to 5 ◦ and 10◦ elevation masks. left panel represent errors of the SDR solutions; red markers in the right panel refer to errors obtained with the UBX. Both solutions refer to an elevation cut-off angle of 15◦ Fig. 9 3D positioning error scatter plots for single point positioning. Blue markers in the left panel represent errors of the SDR solutions; red markers in the right panel refer to errors obtained with the UBX. Both solutions refer to an elevation cut-off angle of 30◦ Fig. 10 3D positioning error histogram for the SDR and the UBX solutions. The results obtained with the SDR are represented in blue while those obtained with the UBX are represented in red. Each bar is 0.5 m wide Fig. 10 3D positioning error histogram for the SDR and the UBX solutions. The results obtained with the SDR are represented in blue while those obtained with the UBX are represented in red. Single point positioning performance assessment Fig- ure 5 depicts East, North, and Up components errors con- sidering a mask angle of 10◦ ; in particular, the left panel of the figure shows the coordinate components errors versus the time for the SDR while the right panel refers to the UBX ones. Figure 5 reveals that the SDR is stable over time show- ing high noise in the coordinate component errors during the entire duration of the experiment. Conversely, the UBX experiences a degradation from epoch 149500. This behav- iour cannot be addressed to the decrease in the number of 1 3 3 Applied Geomatics (2023) 15:583–594 589 Fig. 5 Time series of East, North, and Up coordinate component error for single point positioning. The left panel refers to the SDR solutions while the UBX ones are shown in the right panel. The top row indicates the East error compo- nents. The middle row depicts the North component error while the bottom row refers to the Up component Fig. 6 3D positioning error scatter plots for single point positioning. Blue markers in the left panel represent errors of the SDR solutions; red markers in the right panel refer to errors obtained with the UBX. Both solutions refer to an elevation cut-off angle of 5 ◦ Fig. 7 3D positioning error scatter plots for single point positioning. Blue markers in the left panel represent errors of the SDR solutions; red markers in the right panel refer to errors obtained with the UBX. Both solutions refer to an elevation cut-off angle of 10◦ rror point ers in the ors of the kers in errors . Both vation Fig. 7 3D positioning error scatter plots for single point positioning. Blue markers in the left panel represent errors of the SDR solutions; red markers in the right panel refer to errors obtained with the UBX. Both solutions refer to an elevation cut-off angle of 10◦ m. This is confirmed in Table 3; indeed, comparing the SDR and the UBX statistics, it can be noticed that the UBX mean errors are the half part of the respective SDR indicators while the UBX standard deviation is even the third part. Moreo- ver, referring to the 30◦ elevation cut-off angle case that aims to simulate the obstructed scenario, the SDR experiences a 1 3 Applied Geomatics (2023) 15:583–594 590 Fig. 8 3D positioning error scatter plots for single point positioning. Single point positioning performance assessment Each bar is 0.5 m wide tough degradation of the results: the mean error is twice the mean error for the SDR solution of the previous cases (with elevation cut-off angles of 5 ◦ , 10◦ , and 15◦ ). On the other hand, the elimination of the low-elevated satellites provides an enhancement of the positioning precision. In specific, the UBX statistics show better performance when a 15◦ elevation cut-off angle is adopted as compared to the statistics that cor- respond to 5 ◦ and 10◦ elevation masks. 1 3 591 Applied Geomatics (2023) 15:583–594 Table 3 Three-dimensional coordinate error statistics (mean and STD) for the SDR and the UBX, considering elevation cut-off angles of 5 ◦ , 10◦ , 15◦ , and 30◦ SDR UBX Elevation cut-off angle ( ◦) Mean error (m) STD (m) Mean error (m) STD (m) 5 8.56 5.17 3.54 4.39 10 8.56 5.17 3.56 4.47 15 8.66 5.16 2.79 3.20 30 17.40 15.46 5.97 5.86 Fig. 12 Horizontal coordinate component error scatter plots for single point positioning: top-left panel refers to the solution obtained with an elevation cut-off angle of 5 ◦ ; top-right, bottom-left, and bottom- right panels refer to the solutions obtained with elevation cut-off angles of 10◦ , 15◦ , and 30◦ , respectively In the simulated obstructed scenario (30◦ mask angle), the worse performance can be found for both receivers; nev- ertheless, the UBX mean error and standard deviation are about the third part of the relative SDR metrics. Figure 11 represents three-dimensional coordinate error box-plots for the SDR and the UBX solutions with 5 ◦ , 10◦ , 15◦ , and 30◦ elevation mask angles, respectively. Blue boxes represent the SDR SPP solutions while red boxes are referred to the UBX ones. Some errors for the SDR are very high (occasionally greater than 30 m). To ensure a better view of the figure, a threshold of 40 m has been adopted. The figure shows that red boxes are always tighter and placed further down than the SDR ones. Fig. Single point positioning performance assessment The results obtained with the SDR are represented in blue while those obtained with the UBX are represented in red. Each bar is 0.5 m wide Table 4 Two-dimensional coordinate error statistics (mean and STD) for the SDR and the UBX considering elevation cut-off angles of 5 ◦ , 10◦ , 15◦ , and 30◦ SDR UBX Elevation cut-off angle ( ◦) Mean error (m) STD (m) Mean error (m) STD (m) 5 4.60 3.15 2.22 2.04 10 4.60 3.15 2.11 1.83 15 4.64 3.20 1.89 1.74 30 10.58 11.90 4.60 4.62 Table 5 Vertical error statistics (mean and STD) for the SDR and the UBX considering elevation cut-off angles of 5 ◦ , 10◦ , 15◦ , and 30◦ SDR UBX Elevation cut-off angle ( ◦) Mean error (m) STD (m) Mean error (m) STD (m) 5 0.40 8.30 2.06 4.29 10 0.40 8.30 2.25 4.45 15 0.44 8.35 1.45 3.06 30 −1.39 16.93 −2.73 4.47 Fig. 13 2D positioning error histogram for the SDR and the UBX SPP solutions. The results obtained with the SDR are represented in blue while those obtained with the UBX are represented in red. Each bar is 0.5 m wide ig. 13 2D positioning error istogram for the SDR and he UBX SPP solutions. The esults obtained with the SDR re represented in blue while hose obtained with the UBX re represented in red. Single point positioning performance assessment Each bar s 0.5 m wide Table 4 Two-dimensional coordinate error statistics (mean and STD) for the SDR and the UBX considering elevation cut-off angles of 5 ◦ , 10◦ , 15◦ , and 30◦ SDR UBX Elevation cut-off angle ( ◦) Mean error (m) STD (m) Mean error (m) STD (m) 5 4.60 3.15 2.22 2.04 10 4.60 3.15 2.11 1.83 15 4.64 3.20 1.89 1.74 30 10.58 11.90 4.60 4.62 Table 4 Two-dimensional coordinate error statistics (mean and STD) for the SDR and the UBX considering elevation cut-off angles of 5 ◦ , 10◦ , 15◦ , and 30◦ Table 5 Vertical error statistics (mean and STD) for the SDR and the UBX considering elevation cut-off angles of 5 ◦ , 10◦ , 15◦ , and 30◦ SDR UBX Elevation cut-off angle ( ◦) Mean error (m) STD (m) Mean error (m) STD (m) 5 0.40 8.30 2.06 4.29 10 0.40 8.30 2.25 4.45 15 0.44 8.35 1.45 3.06 30 −1.39 16.93 −2.73 4.47 higher number of outliers (represented by red crosses). The high number of outliers (also for UBX) is due to the single point and to the GPS-only approach used, which limited the number of satellites available and their geometry, also considering that to make a comparison we used only satellites received both by SDR and UBX. distribution is slightly different from that of 3D error hav- ing a more recognizable peak around 2.5 m while the 3D distribution showed a smoother peak around higher errors, e.g. moved on the right of the x-axes. This confirms that the Up component degrades the whole solution. In Table 4, the statistics relative to the two-dimensional errors are shown. The results confirm those shown in Table 3. Figure 15 depicts the horizontal component error scatter plots for the SDR (blue markers), the UBX with the same satellites of the SDR (red markers), and the UBX with all the satellites tracked (yellow markers). All the solutions here are obtained with an eleva- tion cut-off angle of 10◦ . The figure shows that the error point cloud obtained by the UBX with all satellites is tighter than that from the UBX with the SDR satellites (red markers). This was expected since the number of tracked satellites by the UBX is higher than those tracked by the SDR and there is also an improve- ment in their geometric distribution. Single point positioning performance assessment i In Table 5, the statistics relative to the vertical errors are shown. The results confirm that the SDR performance is worse than that of the UBX; indeed, despite the SDR vertical mean errors being lower than those of the UBX, standard deviations are twice the UBX ones for 5 ◦ , 10◦ , and 15◦ cut-off angles and even four times the UBX ones in the 30◦ cut-off angle case. f Figure 14 represents two-dimensional coordinate error box- plots for the SDR and the UBX SPP solutions with 5 ◦ , 10◦ , 15◦ , and 30◦ elevation cut-off angles; like before, blue boxes represent the SDR solutions and red boxes are referred to the UBX ones. From the comparison of the boxes in the figures, the better positioning performance obtained by the UBX receiver is evident: the UBX boxes are always tighter than those of the SDR and also placed further down. The worse performance of the SDR in the positioning domain is highlighted also by the Single point positioning performance assessment 12 Horizontal coordinate component error scatter plots for single point positioning: top-left panel refers to the solution obtained with an elevation cut-off angle of 5 ◦ ; top-right, bottom-left, and bottom- right panels refer to the solutions obtained with elevation cut-off angles of 10◦ , 15◦ , and 30◦ , respectively Figure 13 shows the two-dimensional positioning error histogram for both the SDR and the UBX solutions. As previously, the results achieved by using the SDR are rep- resented in blue while in red are those obtained with the UBX. Each bar is 0.5 m wide. This figure confirms that the accuracy obtained by the UBX is higher than that obtained by the SDR; the centre of the SDR histogram is placed fur- ther to the right concerning to that of the UBX confirming that the SDR has a higher mean of the positioning error. Similarly to Fig. 10, it emerges that the SDR histogram is smoother and wider with respect to the UBX; indeed, the worse performance of the SDR is reflected in the higher standard deviation. In Fig. 13, the two-dimensional error Figure 12 shows the horizontal component error scatter plots for the SDR (blue markers) and the UBX (red mark- ers) for elevation cut-off angles of 5 ◦ , 10◦ , 15◦ , and 30◦ . The figure shows the horizontal error comparison between the SDR and the UBX. It can be noticed that the UBX point cloud is denser than the SDR one regardless of the elevation cut-off angle considered. To ensure a proper com- parison between each cut-off angle, common axes limits have been adopted. 1 3 g. 11 3D positioning error ox-plots for the SDR and e UBX SPP solutions with lected elevation cut-off gles. Red crosses represent e outliers 1 3 Fig. 11 3D positioning error box-plots for the SDR and the UBX SPP solutions with selected elevation cut-off angles. Red crosses represent the outliers Fig. 11 3D positioning error box-plots for the SDR and the UBX SPP solutions with selected elevation cut-off angles. Red crosses represent the outliers 1 3 Applied Geomatics (2023) 15:583–594 592 distribution is slightly different from that of 3D error hav- ing a more recognizable peak around 2.5 m while the 3D distribution showed a smoother peak around higher errors, e.g. moved on the right of the x-axes. This confirms that the Up component degrades the whole solution. 3 Single point positioning performance assessment In Table 4, the statistics relative to the two-dimensional errors are shown. The results confirm those shown in Table 3. In Table 5, the statistics relative to the vertical errors are shown. The results confirm that the SDR performance is worse than that of the UBX; indeed, despite the SDR vertical mean errors being lower than those of the UBX, standard deviations are twice the UBX ones for 5 ◦ , 10◦ , and 15◦ cut-off angles and higher number of outliers (represented by red crosses). The high number of outliers (also for UBX) is due to the single point and to the GPS-only approach used, which limited the number of satellites available and their geometry, also considering that to make a comparison we used only satellites received both by SDR and UBX. Figure 15 depicts the horizontal component error scatter plots for the SDR (blue markers), the UBX with the same satellites of the SDR (red markers), and the UBX with all the satellites tracked (yellow markers). All the solutions here are obtained with an eleva- tion cut-off angle of 10◦ . The figure shows that the error point cloud obtained by the UBX with all satellites is tighter than that from the UBX with the SDR satellites (red markers). This was Fig. 13 2D positioning error histogram for the SDR and the UBX SPP solutions. The results obtained with the SDR are represented in blue while those obtained with the UBX are represented in red. Single point positioning performance assessment Each bar is 0.5 m wide Table 4 Two-dimensional coordinate error statistics (mean and STD) for the SDR and the UBX considering elevation cut-off angles of 5 ◦ , 10◦ , 15◦ , and 30◦ SDR UBX Elevation cut-off angle ( ◦) Mean error (m) STD (m) Mean error (m) STD (m) 5 4.60 3.15 2.22 2.04 10 4.60 3.15 2.11 1.83 15 4.64 3.20 1.89 1.74 30 10.58 11.90 4.60 4.62 Table 5 Vertical error statistics (mean and STD) for the SDR and the UBX considering elevation cut-off angles of 5 ◦ , 10◦ , 15◦ , and 30◦ SDR UBX Elevation cut-off angle ( ◦) Mean error (m) STD (m) Mean error (m) STD (m) 5 0.40 8.30 2.06 4.29 10 0.40 8.30 2.25 4.45 15 0.44 8.35 1.45 3.06 30 −1.39 16.93 −2.73 4.47 distribution is slightly different from that of 3D error hav- ing a more recognizable peak around 2.5 m while the 3D distribution showed a smoother peak around higher errors, e.g. moved on the right of the x-axes. This confirms that the Up component degrades the whole solution. In Table 4, the statistics relative to the two-dimensional errors are shown. The results confirm those shown in Table 3. In Table 5, the statistics relative to the vertical errors are shown. The results confirm that the SDR performance is worse than that of the UBX; indeed, despite the SDR vertical mean b i l h h f h UBX d d d i i higher number of outliers (represented by red crosses). The high number of outliers (also for UBX) is due to the single point and to the GPS-only approach used, which limited the number of satellites available and their geometry, also considering that to make a comparison we used only satellites received both by SDR and UBX. Figure 15 depicts the horizontal component error scatter plots for the SDR (blue markers), the UBX with the same satellites of the SDR (red markers), and the UBX with all the satellites tracked (yellow markers). All the solutions here are obtained with an eleva- tion cut-off angle of 10◦ . The figure shows that the error point l d b i d b h UBX i h ll lli i i h h h Fig. 13 2D positioning error histogram for the SDR and the UBX SPP solutions. Discussion The first will involve a de-noising step of the observables to inject them into the PVT algorithm once “cleaned”. The second will consider the testing of several RF front ends Fig. 14 2D positioning error box-plots for the SDR and the UBX SPP solutions depending on the cut-off angles consid- ered. Red crosses represent the outliers Fig. 15 Horizontal coordinate component error scatter plots for sin- gle point positioning: the SDR solution (represented in blue marker) versus the UBX with the SDR ephemeris (represented in red marker) and the UBX with all available satellites (represented in yellow mark- ers) 593 Fig. 14 2D positioning error box-plots for the SDR and the UBX SPP solutions depending on the cut-off angles consid- ered. Red crosses represent the outliers The results showed that the real issue with this configu- ration, namely Rafael micro-SDR driven by GNSS-SDR, is the limited capabilities of the ultra-low-cost front-end used which generates noisy measurements, as illustrated by CMC analysis conducted in the “Code observation noise” sec- tion and by the results shown in Fig. 4. Fig. 15 Horizontal coordinate component error scatter plots for sin- gle point positioning: the SDR solution (represented in blue marker) versus the UBX with the SDR ephemeris (represented in red marker) and the UBX with all available satellites (represented in yellow mark- ers) Then, we assessed the single point positioning perfor- mance of the SDR against the UBX receiver. The analysis has been carried out using the same satellites for both receiv- ers. Different operational conditions have been simulated by applying different elevation cut-off angles of 5 ◦ , 10◦ , 15◦ , and 30◦ . Low cut-off angles of 5 ◦ and 10◦ allowed for assessing the impact of low-elevated and low-gain signals that are prone to multipath, on the positioning performance. A high elevation cut-off angle of 30◦ , in turn, made it pos- sible to simulate an obstructed environment. The analysis showed that the positioning performance obtained by the SDR is worse than that of the UBX in all scenarios. In detail, we found that the mean and standard deviation of the UBX horizontal error are about half of those of the SDR going from 4.60 to about 2 m for mean and 3.15 to about 1.8 m for standard deviation for 5 ◦,10◦ , and 15◦ mask angles. Discussion When a 30◦ mask angle is applied, performance worsens for both receivers with a mean horizontal error of about 10 m for the SDR with respect to 4.60 m for the UBX. Fig. 15 Horizontal coordinate component error scatter plots for sin- gle point positioning: the SDR solution (represented in blue marker) versus the UBX with the SDR ephemeris (represented in red marker) and the UBX with all available satellites (represented in yellow mark- ers) p Concerning the three-dimensional error, the deterioration in the performance of the SDR compared to the UBX is even more evident, and the SDR error is about three times greater than that of the UBX both for mean and standard deviation. The future works will mainly follow two lines of research. The first will involve a de-noising step of the observables to inject them into the PVT algorithm once “cleaned”. The second will consider the testing of several RF front-ends to assess the influence of the hardware on the overall posi- tioning performance. The actual hardware implementation is not capable of sampling at more than 2.4 Msps without Concerning the three-dimensional error, the deterioration in the performance of the SDR compared to the UBX is even more evident, and the SDR error is about three times greater than that of the UBX both for mean and standard deviation. the SDR was not able to acquire all the satellites that were tracked by the UBX. These satellites were characterized by a low signal gain. Bearing in mind that the experiment was con- ducted with the same conditions for both receivers, e.g. simul- taneous acquisition and shared antenna, we estimated the noise of the observables exploiting the code-minus-carrier analysis; such analysis has confirmed the presence of more noise in the observables generated by the SDR. the SDR was not able to acquire all the satellites that were tracked by the UBX. These satellites were characterized by a low signal gain. Bearing in mind that the experiment was con- ducted with the same conditions for both receivers, e.g. simul- taneous acquisition and shared antenna, we estimated the noise of the observables exploiting the code-minus-carrier analysis; such analysis has confirmed the presence of more noise in the observables generated by the SDR. The future works will mainly follow two lines of research. Discussion The present research led to the following conclusions. From the comparison between the SDR and the UBX, we can state that the carrier-to-noise density ratios are comparable. Nonetheless, 3 Applied Geomatics (2023) 15:583–594 593 Applied Geomatics (2023) 15:583–594 the SDR was not able to acquire all the satellites that were tracked by the UBX. These satellites were characterized by a low signal gain. Bearing in mind that the experiment was con- ducted with the same conditions for both receivers, e.g. simul- taneous acquisition and shared antenna, we estimated the noise The results showed that the real issue with this configu- ration, namely Rafael micro-SDR driven by GNSS-SDR, is the limited capabilities of the ultra-low-cost front-end used which generates noisy measurements, as illustrated by CMC analysis conducted in the “Code observation noise” sec- tion and by the results shown in Fig. 4. Then, we assessed the single point positioning perfor- mance of the SDR against the UBX receiver. The analysis has been carried out using the same satellites for both receiv- ers. Different operational conditions have been simulated by applying different elevation cut-off angles of 5 ◦ , 10◦ , 15◦ , and 30◦ . Low cut-off angles of 5 ◦ and 10◦ allowed for assessing the impact of low-elevated and low-gain signals that are prone to multipath, on the positioning performance. A high elevation cut-off angle of 30◦ , in turn, made it pos- sible to simulate an obstructed environment. The analysis showed that the positioning performance obtained by the SDR is worse than that of the UBX in all scenarios. In detail, we found that the mean and standard deviation of the UBX horizontal error are about half of those of the SDR going from 4.60 to about 2 m for mean and 3.15 to about 1.8 m for standard deviation for 5 ◦,10◦ , and 15◦ mask angles. When a 30◦ mask angle is applied, performance worsens for both receivers with a mean horizontal error of about 10 m for the SDR with respect to 4.60 m for the UBX. Concerning the three-dimensional error, the deterioration in the performance of the SDR compared to the UBX is even more evident, and the SDR error is about three times greater than that of the UBX both for mean and standard deviation. The future works will mainly follow two lines of research. Conflict of interest The authors declare no competing interests. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Com- mons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. Lin T, O’Driscoll C, Lachapelle G (2011) Development of a context- aware vector-based high-sensitivity gnss software receiver. In: Proceedings of the 2011 international technical meeting of the institute of navigation, pp 1043–1055 Misra P, Enge P (2011) Global Positioning System: signals, measure- ments, and performance. Ganga-Jamuna Press, Lincoln Paziewski J (2020) Recent advances and perspectives for positioning and applications with smartphone GNSS observations. Meas Sci Technol 31(9). https://doi.org/10.1088/1361-6501/ab8a7d Paziewski J, Fortunato M, Mazzoni A, Odolinski R (2021) An analysis of multi-GNSS observations tracked by recent android smartphones and smartphone-only relative positioning results. Measurement, 175:109162. ISSN 0263-2241. https://doi.org/ 10.1016/j.measurement.2021.109162. Discussion The first will involve a de-noising step of the observables to inject them into the PVT algorithm once “cleaned”. The second will consider the testing of several RF front-ends to assess the influence of the hardware on the overall posi- tioning performance. The actual hardware implementation is not capable of sampling at more than 2.4 Msps without 1 3 Applied Geomatics (2023) 15:583–594 594 losing samples; this is enough for GPS but not for other constellations such as Galileo. Moreover, an in-depth pseu- dorange errors analysis could be conducted to investigate pseudorange quality and a potential correlation between psue- dorange errors and C∕N0 as already happens with smartphone GNSS receivers (see Robustelli et al. (2021)). Fernández-Prades C, Arribas J, Closas P, Avilés C, Esteve L (2011) GNSS- SDR: an open source tool for researchers and developers. In: proc. of the 24th international technical meeting of the satellite division of the institute of navigation 2011, ION GNSS 2011, volume 2, 09 Hobiger T, Gotoh T, Amagai J, Koyama Y, Kondo T (2009) A GPU based real-time GPS software receiver. GPS Solutions 14(207–216):03. https://doi.org/10.1007/s10291-009-0135-2 Hurskainen H, Jussi R, Ahonen T, Nurmi J (2009) Multicore software- defined radio architecture for gnss receiver signal processing. EURA- SIP J Embed Syst 2009:01. https://doi.org/10.1155/2009/543720 Funding Open access funding provided by Università Parthenope di Napoli within the CRUI-CARE Agreement. This study was supported by the pro- ject “Innovative precise monitoring system based on integration of low-cost GNSS and IMU MEMS sensors” POIR 01.01.01-00-0753/21, co-financed by the European Regional Development Fund within the Sub-measure 1.1.1 of the Smart Growth Operational Program 2014–2020. Publication co- financed from the state budget under the program of the Minister of Edu- cation and Science (Poland) called “Science for Society” project number NdS/536964/2021/2021 amount of funding PLN 1,557,100 total value of the project PLN 1,557,100. Kang J, Lee Y, Park J (2002) Enhancement of continuity and accuracy by GPS/GLONASS combination, and software development. Korean J Geomatics 2(1):65–73 Knežević A, O’Driscoll C, Lachapelle G (2010) Co-processor aiding for real-time software gnss receivers. In: International Technical Meeting of the Institute of Navigation (ION) Ledvina BM, Cerruti AP, Psiaki ML, Powell SP, Kintner PM (2003a) Per- formance tests of a 12-channel real-time GPS L1 software receiver. In: Proceedings of the 16th international technical meeting of the sat- ellite division of the institute of navigation (ION GPS/GNSS 2003), pp 679–688 Declarations Ledvina BM, Powell SP, Kintner PM, Psiaki ML (2003b) A 12-channel real-time GPS L1 software receiver1. In: Proceedings of the 2003 national technical meeting of the institute of navigation, pp 767–782 Conflict of interest The authors declare no competing interests. References Peter de Bakker CCJM, Marel TH, Bree R (2012) Short and zero baseline analysis of GPS L1 C/A, L5Q, GIOVE E1B, and E5aQ signals. GPS Solutions 16(53–64):01. https://doi.org/10.1007/ s10291-011-0202-3 Abbasiannik S (2009) Multichannel Dual Frequency GLONASS Software Receiver in Combination with GPS L1 C/A. PhD thesis, University of Calgary, Calgary, Canada, 04 Petovello M, O’Driscoll C, Lachapelle G, Borio D, Murtaza H (2008) Architecture and benefits of an advanced GNSS soft- ware receiver. Positioning 1(66–78):12. https://doi.org/10.5081/ jgps.7.2.156 Anghileri M, Pany T, Güixens DS,Won JH, Sicramaz A, Stöber C, Krämer I, Dötterböck D, Hein G, Eissfeller B (2001) Performance evaluation of a multi–frequency GPS/Galileo/SBAS software receiver. In: proc. of ION GNSS’07, pp 2749–2761, 01 Robustelli U, Paziewski J, Pugliano G (2021) Observation quality assessment and performance of gnss standalone positioning with code pseudoranges of dual-frequency android smartphones. Sen- sors 21(6):2125 Aram M, El-Rabbany A, Krishnan S, Anpalagan A (2007) Single frequency multipath mitigation based on wavelet analysis. J Navig 60(281– 290):05. https://doi.org/10.1017/S0373463307004146 CTTC (2021) GNSS-SDR website, https://gnss-sdr.org/docs/. Accessed: 2020 Jan 21 Schmidt E, Akopian D, Pack D (2018) Development of a real-time software-defined GPS receiver in a LabVIEW-based instrumen- tation environment. IEEE Transactions on Instrumentation and Measurement, PP:1–15, 03. https://doi.org/10.1109/TIM.2018. 2811446. Cutugno M, Robustelli U, Pugliano G (2019) Testing a GNSS software receiver for end-user utilization. In: Proc. of 2019 International Work- shop IMEKO-TC19 Metrosea, pages 91–96, 10 Cutugno M, Robustelli U, Pugliano G (2020) Low-cost GNSS software receiver performance assessment. Geosciences, 10(2). ISSN 2076-3263. https://doi.org/10.3390/geosciences10020079. Seo J, Chen Y, De Lorenzo DS, Lo S, Enge P, Akos D, Lee J (2011) A real-time capable software-defined receiver using GPU for adap- tive anti-jam GPS sensors. Sensors, 11(9):8966–8991. ISSN 1424- 8220. https://www.mdpi.com/1424-8220/11/9/8966 Delgado N, Haag M (2011) Multipath analysis using code-minus-carrier for dynamic testing of gnss receivers. In: 2011 International Conference on Localization and GNSS, pages 25–30, 06 Zhang Y, Bartone C (2005) A real-time meteorological-based tropo- sphere (RMT) correction with integrity bound for long baseline DGPS. GPS Solutions 9(255–272):11. https://doi.org/10.1007/ s10291-005-0132-z El-Rabbany A (2002) Introduction to GPS: the global positioning system. Artech housei Fantino M, Molino A, Nicola M (2009) N–gene GNSS receiver: benefits of software radio in navigation. proc. of ENC-GNSS’09, 05 1 3
https://openalex.org/W4383815459	https://researchinvolvement.biomedcentral.com/counter/pdf/10.1186/s40900-023-00459-w	English	null	Positioning patients to partner: exploring ways to better integrate patient involvement in the learning health systems	Research involvement and engagement	2,023	cc-by	4,656	COMMENT Open Access © The Author(s) 2023. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativeco mmons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Positioning patients to partner: exploring ways to better integrate patient involvement in the learning health systems Nakia K. Lee‑Foon1, Maureen Smith2, Sarah M. Greene3, Kerry Kuluski1,4 and Robert J. Reid1 Abstract Globally, health systems are increasingly striving to deliver evidence based care that improves patients’, caregivers’ and communities’ health outcomes. To deliver this care, more systems are engaging these groups to help inform healthcare service design and delivery. Their lived experiences—experiences accessing and/or supporting some‑ one who accesses healthcare services—are now viewed by many systems as expertise and an important part of understanding and improving care quality. Patients’, caregivers’ and communities’ participation in health systems can range from healthcare organizational design to being members of research teams. Unfortunately, this involve‑ ment greatly varies and these groups are often sidelined to the start of research projects, with little to no role in later project stages. Additionally, some systems may forgo direct engagement, focusing solely on patient data collection and analysis. Given the benefits of active patient, caregiver and community participation in health systems on patient health outcomes, systems have begun identifying different approaches to studying and applying findings of patient, caregiver and community informed care initiatives in a rapid and consistent fashion. The learning health system (LHS) is one approach that can foster deeper and continuous engagement of these groups in health systems change. This approach embeds research into health systems, continuously learning from data and translating findings into health‑ care practices in real time. Here, ongoing patient, caregiver and community involvement is considered vital for a well functioning LHS. Despite their importance, great variability exists as to what their involvement means in practice. This commentary examines the current state of patient, caregiver and community participation in the LHS. In particular, gaps in and need for resources to support their knowledge of the LHS are discussed. We conclude by recommending several factors health systems must consider in order to increase participation in their LHS. Systems must: (1) assess patients’, caregivers and community understanding of how their feedback are used in the LHS and how collected data are used to inform patient care; (2) review the level and extent of these groups’ participation in health system improvement activities; and (3) examine whether health systems have the workforce, capacity and infrastructure to nurture continuous and impactful engagement. Keywords Learning health system, Patient engagement, Stakeholder engagement, Community based, Patient involvement, Research communities of practice, Patient partners Correspondence: Nakia K. Lee‑Foon nakia.lee-foon@thp.ca Full list of author information is available at the end of the article *Correspondence: Nakia K. Research Involvement and Engagement Research Involvement and Engagement Research Involvement and Engagement Lee‑Foon et al. Research Involvement and Engagement (2023) 9:51 https://doi.org/10.1186/s40900-023-00459-w Lee‑Foon et al. Research Involvement and Engagement (2023) 9:51 https://doi.org/10.1186/s40900-023-00459-w Abstract Lee‑Foon nakia.lee-foon@thp.ca Full list of author information is available at the end of the article Background example, they support the creation of quality improve- ment initiatives and staff hiring and training [7]. The PCC’s input in designing services greatly increases the chance that services reflect the values and meanings of its current and future users [8, 9]. Additionally, their involvement can lead to improved policy making along with better patient health outcomes [2]. Despite their positive impact on health systems, the level of PCCs’ involvement in and the incorporation of their perspec- tives into healthcare redevelopment can fluctuate or be limited in scope [3, 6]. For instance, PCC’s involve- ment often occurs at the beginning of a project and then gradually subsides [6, 10]. They tend to have less of a role in the later stages of a project such as helping inform the uptake and dissemination of a healthcare innovation [6]. I see patients’, caregivers’ and citizens’ roles in the Learning Health System as one leg of a wob- bly stool. There is no doubt that the lived experi- ences of patients, caregivers, and citizens are a pil- lar of the Learning Health System. We often hear the term “embedded in the system.” That needs to be explained to us in terms that are understandable to us so that we can see concrete examples of how it is operationalized throughout the process and pro- vide further input in how we can move forward and provide our input on the process. –Maureen Smith, Patient Partner Globally, health systems are increasingly striving to deliver evidence based care that improves patients’, car- egivers’ and communities’ (PCC) health outcomes [1]. Over the past few years, PCCs have taken more active roles in health systems, partnering with clinicians (e.g. doctors, nurses, social workers) and researchers to shape the design and delivery of healthcare services [2, 3]. The PCC’s lived experiences—individuals’ experiences access- ing and/or supporting someone who accesses healthcare services—are seen by many health systems as an impor- tant part of understanding and improving care quality [3–6]. These experiences can be used to support health- care improvements [7]. Many health systems state that they work to involve PCC in improving healthcare [11]. However, this ‘work’ often focuses on patient data collection and analysis and quality improvement measures without sufficiently acting on the analyzed findings in a mean- ingful and rapid way [12]. © The Author(s) 2023. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativeco mmons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Lee‑Foon et al. Research Involvement and Engagement (2023) 9:51 Page 2 of 5 Lee‑Foon et al. Research Involvement and Engagement Plain English summary Patients, caregivers and communities have started taking more hands on roles in health systems, partnering with healthcare providers and researchers to impact the ways healthcare services are made and delivered. Their input has been shown to improve patient health. While many systems are working to include patients, caregivers and communities in helping improve healthcare, this work often focuses on collecting and analyzing patient data without using it in a timely way. Also, the level of their input can vary and is often limited to the start of a research project. As more health systems recognize the importance of their input in creating better healthcare, some are using different approaches to make this feedback a constant part of their systems. The learning health system (LHS) is one approach that can support deeper and ongoing patient, caregiver and community involvement in health system change. In the LHS, projects are frequently reviewed and feedback used to help health systems make changes as they go. While their involvement is critical to a well functioning LHS, it is unclear what this involvement looks like. This commentary reviews the current state of this involvement. We offer readers a way forward and suggestions to help them determine if they are actively including patients, caregivers and communities in their LHS. Suggestions include reviewing: (1) the ways data are collected and used; (2) how patients, caregivers and communities are involved in health system improvement efforts; and (3) whether or not systems have the tools needed to frequently partner with these groups. Background This slow pace may be due to the challenges of change management and evidence mobilization in healthcare, as well as competing clini- cal priorities [13]. Additionally, grant funding is often time limited, preventing sustainable PCC involve- ment. As more institutions recognize that most of what influences patients’ health and wellbeing (e.g. social determinants of health) occurs outside their clinical walls, some have begun implementing differ- ent approaches to studying and applying the findings of PCC informed care initiatives in a timely fashion [10]. Across health systems, PCC’s involvement and impact can take many forms. They can be active members of and share decision making with their care teams. The PCCs’ may participate in healthcare organizational design and governance and research teams who con- duct patient partnered research. They may also be part of patient and family advisory councils where, for Page 3 of 5 Lee‑Foon et al. Research Involvement and Engagement (2023) 9:51 Page 3 of 5 Lee‑Foon et al. Research Involvement and Engagement (2023) 9:51 LHS healthcare outcomes and quality improvement are lacking. The learning health system (LHS) is one such frame- work that can support deeper and ongoing PCC engage- ment in healthcare change initiatives. The framework calls for continuous PCC involvement across all stages of healthcare initiatives’ development. Their involve- ment is considered vital to the conceptualization and successful operationalization of initiatives [1]. The PCC’s involvement is seen as the impetus for health- care change [14]. The LHS helps health systems marry research with quality improvement through a continu- ous cycle of data collection and analysis. Pre-existing clinical and sociodemographic data from sources such as electronic health records and diverse patients, pro- grams and healthcare settings are used. This allows the LHS framework to bypass the commonly used approach for generating healthcare data in a research context and avoid delays typical of a protracted research process [15]. Analyzed data are then given back to healthcare providers and decision makers (e.g., healthcare lead- ers, policy makers), adjustments made using this “real- time feedback” and the healthcare improvement cycle is continued [15, 16]. When PCCs are actively involved in applied health research and quality improvement [12], tensions can often arise when research significance, communi- ties’ and health systems’ key issues and capabilities are misaligned [18]. Background This may be prompted by the contin- ued, traditional healthcare research funding approach where proposals are selected based on their scientific significance, funding agency priorities, investigator knowledge and the potential scientific impact of newly created, generalizable information. This approach counters local systems and community prioritized research that can be multi-faceted, narrower in scope [18] and may have not yet garnered sufficient attention from academia to foster its own body of peer reviewed literature. The resulting production and use of evi- dence from this traditional research approach may not actively include PCC. Lack of guidance to support PCC’s understanding of the LHS Considering that many health systems centre their work on PCC healthcare and the LHS framework empha- sizes PCC involvement, the LHS is well positioned to ensure that patient voices continuously inform research and, in turn, practice in real-time. However, great vari- ability exists as to what their involvement in the LHS means in practice [12]. As such, this commentary delves into the current state of PCC involvement in the LHS and ways to ensure PCCs are actively involved along every step of the LHS. The commentary is informed by the literature and experiences of an academic-community collaborative comprised of a patient partner and public health, community based research and patient and car- egiver scholars. For many, the LHS remains an abstract concept and, for some, a buzzword. Although literature on frame- works, and examples of LHS-informed interventions exist [12], articles focused on how patients learn about and understand LHS are lacking. This is a curious over- sight, given that LHS literature underscores the need for PCC engagement and their active participation in shap- ing healthcare [12, 16]. A scoping review of LHS articles from 2016 to 2020 found articles discussing the level of patient involvement in LHS were scarce [12]. As such, it should come as no surprise that anecdotal evidence indicates many patients are unfamiliar with the LHS, an unfamiliarity that results from a variety of factors. For instance, many scholars may be more focused on devel- oping and revising LHS frameworks than on examining LHS’s impact on various healthcare issues and PCCs’ understanding of LHS. Furthermore, many funders have not emphasized the development of methods, language and approaches to help integrate PCCs into the LHS. PCCs’ limited involvement in the LHS Despite health systems’ increased interest in LHS, PCCs are limited in their involvement in the LHS. No com- mon language, tools or frameworks for discussing and operationalizing LHS exist, making it likely that many healthcare institutions are using this approach without explicitly naming it as such [12]. Tools that exist often focus on the ‘average patient’ failing to engage and reflect diverse voices and needs, particularly those from equity deserving groups who are marginalized due to their socio-economic status, gender identity, racialization, sexual orientation and other categories of difference. Limited literature discusses the creation of a practical, equitable LHS framework co-designed by PCCs [12]. The lack of commonly used language and LHS frame- works makes it difficult to explain the benefits of obtain- ing care in and the importance of their role in shaping the LHS. Furthermore, there is a lack of research focused on understanding PCCs’ experience and perspective of LHS. Many PCCs remain unaware of how their involvement in LHS informs healthcare practice. At the time of writ- ing, courses dedicated to teaching the LHS approach and methods are minimal, making it difficult for some health system leaders to understand how best to introduce and teach various components of the LHS to PCCs. Second, healthcare institutions must ask themselves: “how are patients and caregivers involved in the learning activities (e.g., research prioritization, evidence collec- tion, data synthesis, dissemination) of the health system’s planning, improvement and knowledge dissemination efforts, and at what level (e.g., consultations, involve- ment, collaboration, lead/support)?” As previously noted, PCCs are often not included in every stage of health system work and only consulted after data analy- sis or implementation activities occur. They are not told how analyzed data and research fit into the care delivery and improvement approach, nor how it will be used to inform healthcare changes and future research. True LHS informed patient-centered care approaches require PCCs to actively participate at every step of the research pro- cess from research design to then applying study findings to patient-centered care initiatives. They must be made aware of their work’s impact on healthcare systems and whether these systems are supporting their needs. Given the dearth of research on PCC’s engagement in LHS, case studies, learning communities, or other exemplars that show effective PCC engagement in learning activities are crucial. PCCs’ limited involvement in the LHS At the same time, health systems must examine whether they have the capacity to build the core values and infrastructure that foster continuing and substan- tive PCC engagement and whether its workforce have the skills needed to collaborate with PCCs to develop more effective healthcare [20]. This examination and follow- up aligns with LHS and ensures that care truly reflects patient and caregiver needs. A way forward Despite the aforesaid gaps in PCC involvement in the LHS, we believe there are several ways to begin mending these gaps. When seeking to involve patients in LHS, sev- eral key items must be considered. Health systems must first ask themselves: “exactly what is being done with the data we collect and how can we use it to fulfill our mission of caring for patients?” This question is vital to enhancing patient centered-care strategies at the local/ organizational level as systems are prompted to reflect on why data are being collected and how the findings will inform care. This reflection is lacking in many systems. In order to maintain PCC’s trust of health systems, systems that have historically failed equity deserving groups [13], PCCs must be shown how their experiences and input are used in LHS. Additionally, they must be provided with accessible information and training to help them better participate and co-determine their role and impact in LHS. Patients as more than data donorsi Over a decade after patients were identified as critical for catalyzing change in the LHS [17], much of the recently published LHS literature have focused on generating and translating knowledge and enhancing patient care through data [16]. There continues to be a gap in under- standing patients’ roles in LHS beyond being ‘donors’ of data. Quality performance measures and healthcare out- come data continue to supplant direct PCC feedback. As noted by Kuluski and Guilcher [16], measurement tools often do not provide researchers the opportunity to understand PCCs, their abilities and the social contexts (e.g. knowledge of caregivers’ needs) they experience. Lit- erature exploring PCCs’ perspectives and understanding of their specific role in LHS is sparse. Research examin- ing the impact of varying levels of PCC involvement on The literature provides limited insight on the role cli- nicians and healthcare institutions play in shaping PCCs’ understanding of and ways they can contribute to the LHS. As the LHS is a merging of healthcare delivery and research systems, institutions–particularly those without an embedded research unit or team–may find it difficult to determine best practices for PCCs’ involvement. This is unfortunate as opportunities for synergies and mutual understanding between PCCs involved in research and quality improvement initiatives exist. Further, low health Lee‑Foon et al. Research Involvement and Engagement (2023) 9:51 Page 4 of 5 this concept and how and where it is operationalized to include their lived experiences and insights is vital. literacy—the limited ability to access, comprehend, pro- cess and apply health information—can act as a barrier to participating in health system improvement for some PCCs [19]. Finally, healthcare institutions may be failing to engage PCCs as collaborators in advancing high value healthcare delivery that best fits patients’ and caregivers’ needs [14]. literacy—the limited ability to access, comprehend, pro- cess and apply health information—can act as a barrier to participating in health system improvement for some PCCs [19]. Finally, healthcare institutions may be failing to engage PCCs as collaborators in advancing high value healthcare delivery that best fits patients’ and caregivers’ needs [14]. Availability of data and materials Availability of data and materials Not applicable. 14. Olsen L, Saunders RS, McGinnis JM. Patients charting the course: citizen engagement and the learning health system: workshop summary. In: Olsen L, Saunders RS, McGinnis JM, editors. Patients charting the course: citizen engagement and the learning health system: workshop summary Washington, DC: The National Academies Collection: Reports funded by National Institutes of Health; 2011. 14. Olsen L, Saunders RS, McGinnis JM. Patients charting the course: citizen engagement and the learning health system: workshop summary. In: Olsen L, Saunders RS, McGinnis JM, editors. Patients charting the course: citizen engagement and the learning health system: workshop summary Washington, DC: The National Academies Collection: Reports funded by National Institutes of Health; 2011. Author details 1 18. Reid RJ, Greene SM. Gathering speed and countering tensions in the rapid learning health system. Learning Health Syst. 2023. https://doi.org/ 10.1002/lrh2.10358. 18. Reid RJ, Greene SM. Gathering speed and countering tensions in the rapid learning health system. Learning Health Syst. 2023. https://doi.org/ 10.1002/lrh2.10358. 1 Institute for Better Health, Trillium Health Partners, 100 Queensway West, Clin‑ ical and Administrative Building, 6th Floor, Mississauga, ON L5B 1B8, Canada. 2 Ottawa, Canada. 3 SG Strategies, 1249 NE 89th Street, Seattle, WA 98115, USA. 4 Institute of Health Policy, Management and Evaluation, University of Toronto, 155 College St 4th Floor, Toronto, ON M5T 3M6, Canada. 19. Palumbo R, Annarumma C, Manna R, Musella M, Adinolfi P. Improving quality by involving patient. The role of health literacy in influencing patients’ behaviors. Int J Healthc Manag. 2019;14(1):144–52. 19. Palumbo R, Annarumma C, Manna R, Musella M, Adinolfi P. Improving quality by involving patient. The role of health literacy in influencing patients’ behaviors. Int J Healthc Manag. 2019;14(1):144–52. 20. Baker R, Judd M, Fancott C, Maika C. Creating “engagement-capable environments” in healthcare. In: Patient Engagement—CatalyzingIm‑ provement and Innovation in Healthcare. 2016. pp. 11–34. https://www. longwoods.com/content/24908/books%2B/creating-engagement-capab le-environments-in-healthcare. Accessed 10 Nov 2022. Received: 21 March 2023 Accepted: 22 June 2023 Received: 21 March 2023 Accepted: 22 June 2023 Received: 21 March 2023 Accepted: 22 June 2023 Acknowledgements 11. Bodenheimer T, Sinsky C. From triple to quadruple aim: care of the patient requires care of the provider. Ann Fam Med. 2014;12(6):573–6. 11. Bodenheimer T, Sinsky C. From triple to quadruple aim: care of the patient requires care of the provider. Ann Fam Med. 2014;12(6):573–6. Author contributions All authors contributed to the conception, design and drafting of this com‑ mentary. All authors read and approved the final manuscript. 12. Zurynski Y, Smith CL, Vedovi A, Ellis LA, Knaggs G, Meulenbroeks I, Warwick M, Gul H, Pomare C, Braithwaite J. Mapping the learning health system: a scoping review of current evidence. Sydney: Australian Institute of Health Innovation and the NHMRC Partnership Centre for Health System Sustainability; 2020. Competing interests 17. Institute of Medicine (US) Roundtable on Evidence-Based Medicine, Olsen L, Aisner D, McGinnis JM, editors. The learning healthcare system: workshop summary. Washington, DC: National Academies Press; 2007. The authors declare that they have no competing interests. Consent for publication Not applicable. 16. Kuluski K, Guilcher SJT. Toward a person-centred learning health system: understanding value from the perspectives of patients and caregivers. Healthc Pap. 2019;18(4):36–46. Ethics approval and consent to participate Not applicable. Ethics approval and consent to participate Not applicable. 15. Reid RJ. Embedding research in the learning health system. Healthc Pap. 2016;16:30–5. 15. Reid RJ. Embedding research in the learning health system. Healthc Pap. 2016;16:30–5. Acknowledgements Not applicable. Acknowledgements Not applicable. Missed opportunities for PCC involvement in the LHS PCCs’ engagement in health systems may stem from a diagnosis or condition and their perceived ability to trigger change. This engagement can occur via roles in different spheres like patient-partnered research and patient and family advisory councils. Although syner- gies exist between these two roles, PCCs may not inter- act and differ in their approach to supporting healthcare issues and challenges. It may be difficult for them to dis- tinguish between LHS informed versus non-informed healthcare institutions and no specific guidelines exist on how best to integrate PCCs into LHS [10]. This gap is a missed opportunity for health systems to learn directly from PCCs how to increase patient satisfaction, health- care service delivery and, ultimately, health outcomes. As PCC involvement is vital to moving the gears of LHS along, fostering PCC’s awareness and understanding of Lee‑Foon et al. Research Involvement and Engagement (2023) 9:51 Lee‑Foon et al. Research Involvement and Engagement (2023) 9:51 Page 5 of 5 Page 5 of 5 Finally, the LHS offers a tremendous opportunity to dismantle the silos of patient-partnered research and healthcare quality improvement research, both of which are vital components of a successful LHS. Bringing these groups together would enhance their respective strengths, accelerate the pace of care improvement, and likely enhance the types of evidence that can be shared and meaningfully synthesized. We believe these aforesaid recommendations will give rise to healthcare institutions who truly partner with a diverse array of PCCs and ulti- mately care that responds to their healthcare priorities and needs. 6. Cluley V, Ziemann A, Feeley C, Olander EK, Shamah S, Stavropoulou C. Mapping the role of patient and public involvement during the differ‑ ent stages of healthcare innovation: a scoping review. Health Expect. 2022;25(3):840–55. 7. Carman KL, Dardess P, Maurer M, Sofaer S, Adams K, Bechtel C, et al. Patient and family engagement: a framework for understanding the ele‑ ments and developing interventions and policies. Health Aff (Millwood). 2013;32(2):223–31. 8. Institute of Medicine of the National Academies, Olsen L, Aisner D, McGinnis JM. The learning healthcare system: workshop summary (IOM roundtable on evidence-based medicine). Washington, DC: The National Academies Press; 2007. 8. Institute of Medicine of the National Academies, Olsen L, Aisner D, McGinnis JM. The learning healthcare system: workshop summary (IOM roundtable on evidence-based medicine). Washington, DC: The National Academies Press; 2007. 9. Van der Velden M, Mörtberg C. Participatory design and design for values. Funding g We wish to acknowledge financial support from The Ontario SPOR SUPPORT Unit (OSSU) and philanthropic financial support from the Hazel McCallion Research Chair in Learning Health Systems at Trillium Health Partners. 13. Bokhour BG, Fix GM, Mueller NM, Barker AM, Lavela SL, Hill JN, et al. How can healthcare organizations implement patient-centered care? Examining a large-scale cultural transformation. BMC Health Serv Res. 2018;18(1):168. 13. Bokhour BG, Fix GM, Mueller NM, Barker AM, Lavela SL, Hill JN, et al. How can healthcare organizations implement patient-centered care? Examining a large-scale cultural transformation. BMC Health Serv Res. 2018;18(1):168. Availability of data and materials Not applicable. Missed opportunities for PCC involvement in the LHS In: Handbook of ethics, values, and technological design: sources, theory, values and application domains; 2015. p. 41–66. 9. Van der Velden M, Mörtberg C. Participatory design and design for values. In: Handbook of ethics, values, and technological design: sources, theory, values and application domains; 2015. p. 41–66. 10. Ponte PR, Conlin G, Conway JB, Grant S, Medeiros C, Nies J, Shulman L, Branowicki P, Conley K. Making patient-centered care come alive: achieving full integration of the patient’s perspective. JONA J Nurs Adm. 2003;33(2):82–90. 10. Ponte PR, Conlin G, Conway JB, Grant S, Medeiros C, Nies J, Shulman L, Branowicki P, Conley K. Making patient-centered care come alive: achieving full integration of the patient’s perspective. JONA J Nurs Adm. 2003;33(2):82–90. Publisher’s Note S i N i 1. Wolfe A. Institute of Medicine Report: crossing the quality chasm: a new health care system for the 21st century. Policy Polit Nurs Pract. 2001;2(3):233–5. 2. Duffett L. Patient engagement: what partnering with patient in research is all about. Thromb Res. 2017;150:113–20. 3. Donetto S, Pierri P, Tsianakas V, Robert G. Experience-based co-design and healthcare improvement: realizing participatory design in the public sector. Des J. 2015;18(2):227–48. 4. Lukyanenko R, Wiggins A, Rosser HK. Citizen science: an information qual‑ ity research frontier. Inf Syst Front. 2020;22:961–83. 5. McCarthy S, O’Raghallaigh P, Woodworth S, Lim YY, Kenny LC, Adam F. Embedding the pillars of quality in health information technology solu‑ tions using “Integrated Patient Journey Mapping”(IPJM): case study. JMIR Hum Factors. 2020;7(3):e17416. 1. Wolfe A. Institute of Medicine Report: crossing the quality chasm: a new health care system for the 21st century. Policy Polit Nurs Pract. 2001;2(3):233–5. Springer Nature remains neutral with regard to jurisdictional claims in pub‑ lished maps and institutional affiliations. 2. Duffett L. Patient engagement: what partnering with patient in research is all about. Thromb Res. 2017;150:113–20. 3. Donetto S, Pierri P, Tsianakas V, Robert G. Experience-based co-design and healthcare improvement: realizing participatory design in the public sector. Des J. 2015;18(2):227–48. 4. Lukyanenko R, Wiggins A, Rosser HK. Citizen science: an information qual‑ ity research frontier. Inf Syst Front. 2020;22:961–83. 5. McCarthy S, O’Raghallaigh P, Woodworth S, Lim YY, Kenny LC, Adam F. Embedding the pillars of quality in health information technology solu‑ tions using “Integrated Patient Journey Mapping”(IPJM): case study. JMIR Hum Factors. 2020;7(3):e17416.
https://openalex.org/W3126259941	https://researchonline.lshtm.ac.uk/id/eprint/4665928/1/Regassa_etal_2021_Prevalance-of-cardiovascular-disease-and.pdf	English	null	Prevalence of Cardiovascular Disease and Associated Factors Among Type 2 Diabetes Patients in Selected Hospitals of Harari Region, Eastern Ethiopia	Frontiers in public health	2,021	cc-by	7,336	ORIGINAL RESEARCH published: 05 February 2021 doi: 10.3389/fpubh.2020.532719 ORIGINAL RESEARCH published: 05 February 2021 doi: 10.3389/fpubh.2020.532719 Lemma Demissie Regassa 1, Assefa Tola 1 and Yohanes Ayele 2 Lemma Demissie Regassa 1, Assefa Tola 1 and Yohanes Ayele 2 1 Department of Epidemiology and Biostatistics, School of Public Health, College of Health and Medical Sciences, Haramaya University, Harar, Ethiopia, 2 Department of Clinical Pharmacy, School of Pharmacy, College of Health and Medical Sciences, Haramaya University, Harar, Ethiopia Background: Cardiovascular disease (CVD) is the most prevalent complication and the leading cause of death among patients with diabetes mellitus (DM). Type 2 diabetes mellitus (T2DM) patients have a 2- to 4-fold increased risk of CVD. There is a scarcity of data about the magnitude of CVD among patients with diabetes in Ethiopia. This study aimed to assess the prevalence and associated factors of CVD among T2DM patients at selected hospitals of Harari regional state of Ethiopia. Edited by: Charumathi Sabanayagam, Singapore Eye Research Institute (SERI), Singapore Reviewed by: Avinash Ravipati, University of Oklahoma Health Sciences Center, United States Abdurezak Ahmed Abdela, Addis Ababa University, Ethiopia Methods: This hospital-based retrospective data review was conducted among T2DM patients on follow-up in the diabetes clinics of selected hospitals of Harari regional state. The records of T2DM patients who have been diagnosed between January 1, 2013, and December 31, 2017, were reviewed from March to April 2018. Data were collected by using structured checklists from all necessary documents of T2DM patients. Statistical analysis was done using STATA 14.1. Bivariate and multivariate logistic regressions were used to identify factors associated with CVD. Correspondence: Lemma Demissie Regassa esraeldemiss@gmail.com Specialty section: This article was submitted to Clinical Diabetes, a section of the journal Frontiers in Public Health Result: The records of 454 T2DM patients were extracted from three government hospitals in Harari regional state. Their age was ranging from 15 to 86 years with a mean age (±SD) of 45.39 (14.76). The overall prevalence of CVD among T2DM patients was 42.51%, composed of hypertensive heart diseases (38.99%), heart failure (6.83%), and stroke (2.20%). The ﬁnal multivariate logistic regression model revealed that age older than 60 years [adjusted odds ratio (AOR) = 3.22; 95% CI: 1.71–6.09], being physically inactive (AOR = 1.45; 95 CI: 1.06–2.38), drinking alcohol (AOR = 2.39; 95% CI: 1.17–6.06), hypertension (AOR = 2.41; 95% CI: 1.52–3.83), body mass index >24.9 kg/m2 (AOR = 1.81; 95% CI: 1.07–3.07), and experiencing microvascular diabetic complications (AOR = 3.62; 95% CI: 2.01–6.53) were signiﬁcantly associated with the odds of having CVD. Received: 05 February 2020 Accepted: 21 December 2020 Published: 05 February 2021 ORIGINAL RESEARCH published: 05 February 2021 doi: 10.3389/fpubh.2020.532719 Population and Selection Criteria Records of T2DM patients who were diagnosed after January 1, 2013, and before December 31, 2017, were included, but patients with no baseline records were excluded. Additionally, patients with body mass index (BMI) < 18.5 kg/m2, end-stage renal diseases, transplanted organs or on dialysis, and/or other diagnosed chronic diseases like human immunodeﬁciency virus (HIV)/acquired immunodeﬁciency syndrome (AIDS), chronic obstructive pulmonary disease (COPD), or chronic liver disease (cirrhosis) were also excluded, as these factors lead to the immune deﬁciency. Although microvascular complications have a signiﬁcant role in the prognosis of T2DM, CVDs are the leading cause of morbidity and mortality among patients living with T2DM (6, 20). More than 70% of hospitalizations for chronic complications of diabetes are attributable to CVD (21, 22). The risk of morbidity and mortality caused by CVD in diabetes patients increases with the long duration of the diabetes (23–25). Even though epidemiological studies have demonstrated an association between CVD and blood glucose levels, studies that indicate the magnitude of CVD and associated factors among diabetes patients in Harari region are limited. Thus, we aimed to assess the prevalence and associated factors of CVD among T2DM patients in hospitals of Harari regional state of Ethiopia. Citation: Regassa LD, Tola A and Ayele Y (2021) Prevalence of Cardiovascular Disease and Associated Factors Among Type 2 Diabetes Patients in Selected Hospitals of Harari Region, Eastern Ethiopia. Front. Public Health 8:532719. doi: 10.3389/fpubh.2020.532719 February 2021 \| Volume 8 \| Article 532719 Frontiers in Public Health \| www.frontiersin.org Regassa et al. Cardiovascular Disease Among Type-2 Diabetes Conclusion: The prevalence of CVD was high and associated with advanced age, physical inactivity, drinking alcohol, higher body mass index, hypertension, and having microvascular complications. Health care workers should educate T2DM patients about healthy lifestyles like physical activity, weight reduction, blood pressure control, and alcohol secession, which can reduce the risk of CVD. Conclusion: The prevalence of CVD was high and associated with advanced age, physical inactivity, drinking alcohol, higher body mass index, hypertension, and having microvascular complications. Health care workers should educate T2DM patients about healthy lifestyles like physical activity, weight reduction, blood pressure control, and alcohol secession, which can reduce the risk of CVD. Keywords: type 2 diabetes mellitus, prevalence, cardiovascular diseases, Ethiopia, hospitals, Harar Study Area and Period Another reason for the occurrence of CVD is inﬂammation, as immune response occasionally resulted in a detrimental eﬀect. Even though DM is characterized by low-level inﬂammation, there is evidence showing that the immune activation preceding insulin resistance in diabetic and pre-diabetic states increases cardiovascular risk in T2DM processes (17). In addition to the impact of DM, modiﬁable and non-modiﬁable factors are contributing to the causation of CVD (18, 19). Study Design Hospital-based retrospective data review was conducted on records of T2DM patients at government hospitals of Harari regional state of Ethiopia. Study Area and Period Type 2 diabetes mellitus (T2DM) is a progressive and chronic metabolic disorder that is characterized by insulin resistance and functional failure of pancreatic beta cells (1). The prevalence of T2DM has been increasing intensely over the past few decades, with the highest rates of growth being seen in Sub-Saharan Africa (2, 3). This study was conducted among T2DM patients on follow- up in the diabetes clinics of government hospitals of Harari regional state. Harari is the smallest of the nine states of Ethiopia located in the eastern part of the country and surrounded by the east Hararghe zone of Oromia regional state. In the region, there are two public hospitals, one federal police hospital, two private hospitals, and eight health centers. This study was conducted in two public hospitals and one federal police hospital, namely, Hiwot Fana Specialized University Hospital (HFSUH), Jugal General Hospital (JGH), and Federal Harar Police Hospital (FHPH). Cardiovascular disease (CVD), which involves heart and blood vessels, includes coronary heart disease (CHD), cerebrovascular disease, peripheral arterial disease, deep vein thrombosis, and pulmonary embolism (4, 5). It is the main cause of complications and morbidity among patients with T2DM globally (6, 7). Among T2DM patients, CVD risk was estimated to be 2- to 4-fold higher than the non-diabetic population (8, 9). HFSUH is the teaching hospital for Haramaya University and comprehensive hospital for East Ethiopia (including Harari region, some parts of Somali region, and eastern Hararghe zone of Oromia) that is expected to serve about 5.8 million people in the eastern part of Ethiopia (26). Similarly, JGH is the oldest hospital in the country. FHPH is the government hospital serving police communities and their families. Hospitals are serving diabetes patients under the established chronic follow-up clinics. Patients’ records from January 2013 to December 2017 were extracted in March to April 2018. g p p ( , ) The natural history of T2DM is a slow process and may last even a decade; it might be initially presented with macroangiopathy, particularly CHD (10–13). The eﬀect of T2DM on cardiovascular manifestations varies based on the speciﬁc cardiovascular outcome. Atherosclerosis, the major cause of macrovasculature, is the result of metabolic syndrome in diabetes patients (14–16). Similarly, alterations of small vessels in the brain, heart, and peripheral vasculature are contributing to the development of CVD and mortality (15). Frontiers in Public Health \| www.frontiersin.org Hypertension (HTN) deﬁned as systolic blood pressure >139 mmHg and/or diastolic blood pressure >89 mmHg. Body Mass Index The BMI is reclassiﬁed as normal if BMI is between 18.5 and 24.9 kg/m2 (18.5–24.9 kg/m2) and above normal if BMI is above 24.9 kg/m2 (33). RESULTS Sociodemographic Characteristics Total records of 454 T2DM patients were extracted from three government hospitals: 230 (50.66%) HFSUH, 111 (24.45%) JGH, and 113 (24.89%) FHPH. Two hundred ﬁfty-nine (57%) were males. The age ranges from 15 to 86 years with mean (±SD) of 45.39 (14.76). The median age was 50 years (q1:30, q3:60). The majority (74.67%) of the patients where older than 40 years, and only 44 (9.69%) were younger than 30 years. The majority (73.84%) of the patients were urban dwellers and 389 (85.68%) were currently married. Regarding the occupation, 186 (40.97%) were civil servants, while 177 (39%) had private work (Table 1). Controlled Blood Pressure Most (93.17%) patients started their DM follow-up immediately after diagnosis, but 6.83% delayed from 1 month to a year. Metformin alone (38.55%) and metformin in combination with glibenclamide (49.56%) were the commonly utilized therapies. From the total, 115 (25.39%) had a family history of DM. Majority (68.81%) of the patients’ FBS level was above 130 mg/dl, while only 128 patients (28.19%) have a controlled FBS level. deﬁned as systolic blood pressure between 120 and 139 mmHg and diastolic blood pressure between 65 and 85 mmHg (30, 31). p Cardiovascular Disease CVD comprises the major disorders of the heart and the arterial circulation supplying the heart, brain, and peripheral tissues. Thus, CVD will be considered if the patient had at least one but not limited to hypertensive heart diseases, heart failure, or stroke (4, 28). Controlled Blood Glucose Glycemic status was categorized as good glycemic control if average (3-months average) fasting blood glucose (FBG) is 80– 130 mg/dl (4.4–7.2 mmol/L) and poor control if FBG was >130 mg/dl (>7.2 mmol/L) (32). Regarding the complications of DM, 75 (16.52%) patients have experienced acute complications of DM, either diabetic ketoacidosis or hyperglycemic hyperosmolar state. Microvascular complications were seen among 84 (18.50%) patients. Retinopathy was the major (47.62%) microvascular complication followed by nephropathy (27.38%) and neuropathy (25%). Other complications were developing foot complications (21%) and non-diabetic kidney diseases (15.20%). Physical Activity classiﬁed based on the occupational status of the patients (29). Physical activities were merged and regrouped into three categories: both moderate and high physical activity are leveled as “physically active” and those whose score is within the range of sedentary life are termed as “physically inactive.” Data Management and Analysis conﬁdence level, 80% power, and 5% precision. For the ﬁrst objective, the calculated sample size was 376, which was based on the 42.6% prevalence of CVD in North India during 2011– 2014 (27). For the second objective, the maximum sample was found for the obesity (≥30 kg/m2) (N = 334). By adding the 20% for incomplete records for CVD, the minimum sample size calculated was 454. Patients were selected by simple random sampling from the registry of the follow-up using computer- generated numbers. Patient identiﬁcation number (medical record identiﬁer) was randomized by using Excel to select an individual patient. Data Management and Analysis Statistical analysis was done using STATA 14.1. The risks of CVD and sociodemographic characteristics were summarized using proportion and mean with standard deviations. Outcome variable is determined if a patient experienced at least one type of CVD (coronary artery diseases, hypertensive heart diseases, stroke, heart failure, or any other else). To determine the factors for CVD, bivariate and multivariate logistic regressions were ﬁtted, and variables were selected using the acyclic graph model selection (Figure 1). The ﬁnal optimal model was selected based on Akaike information criterion (AIC) (36). Hosmer and Lemeshow test was ﬁtted to test model ﬁtness, and appropriate methods of multicollinearity test between independent variables were applied. Sampling Technique and Sample Size Determination The sample size was calculated by Epi Info version 7 using single and double population proportion formula taking 95% February 2021 \| Volume 8 \| Article 532719 Frontiers in Public Health \| www.frontiersin.org 2 Cardiovascular Disease Among Type-2 Diabetes Regassa et al. Data Collection Methods The data were collected by using structured checklists from T2DM patients’ documents including DM registration book, electronic information databases, patient card, and follow-up records. Data were collected by health oﬃcers and nurses working in the respective hospitals but not at diabetes clinics. All the ﬁlled extraction sheets were checked for completeness and consistency by supervisors and investigators to ensure the quality of data. We also cross-checked the data entry and clariﬁed any missing data. Overall, 193 (42.51%, 95% CI: 38.02–47.13) patients were diagnosed with CVD. Majority of them were diagnosed with hypertensive heart diseases (38.99%) and heart failure (6.83%), and the remaining experienced stroke (2.20%). The prevalence of hypertensive heart diseases was 28% (95% CI: 21–35) for younger than 40 years, 40% (95% CI: 33–46) for 40–60 years, and 49% (95% CI: 52–64) for older than 60 years patients. The magnitude of heart failure was 2.5, 8.5, and 4.4% for age <40, 40–60, and above 60 years patients, respectively. At least one February 2021 \| Volume 8 \| Article 532719 Frontiers in Public Health \| www.frontiersin.org 3 Cardiovascular Disease Among Type-2 Diabetes Regassa et al. FIGURE 1 \| Conceptual framework for the causation and risk factors of cardiovascular disease in type 2 diabetes. Source [different kinds of literature (28, 34, 35)]. TABLE 1 \| Sociodemographic characteristics of type 2 diabetes patients in government hospitals of the Harari region, Eastern Ethiopia. Sociodemographic characteristics Categories Frequency Percentage (%) Sex Male 259 57.05 Female 195 42.95 Age Mean (±SD) 48.39 (14.76) Residence Urban 333 73.84 Rural 118 26.16 Marital status Married 389 85.68 Non-marrieda 65 14.32 Occupation Civil servants 186 40.97 Private workersb 177 38.99 Unemployed 64 14.10 Retiredc 27 5.95 aCould be single, widowed, divorced, or separated. bAre farmers, merchants, and other self-employments. cRetired individuals used to work in the government or non-governmental organization and terminates their job due to reaching the age of retirement. TABLE 1 \| Sociodemographic characteristics of type 2 diabetes patients in government hospitals of the Harari region, Eastern Ethiopia. TABLE 1 \| Sociodemographic characteristics of type 2 diabetes patients in government hospitals of the Harari region, Eastern Ethiopia. physically inactive, 28 (6.17%) were active smokers, 34 (7.49%) were alcoholics, and 83 (18.28%) were obese (Table 2). In bivariate logistic regression, age, place of residence, physical activity, drinking alcohol, smoking, level of blood pressure, BMI, history of infection, and microvascular complications are signiﬁcantly associated with CVD. Data Collection Methods Similarly, the likelihood of acquiring CVD among hypertensive patients is more than two times (AOR = 2.41; 95% CI: 1.52–3.83) higher than those with normal blood pressure. Patients who had a history of microvascular DM complications had three times (AOR = 3.62; 95% CI: 2.01–6.53) the likelihood of developing CVD compared with T2DM patients who did not experience microvascular complications of DM (Table 3). (BMI > 24.9 kg/m2) patients compared to patients who had normal weight (BMI 18.5–24.9 kg/m2). Similarly, the likelihood of acquiring CVD among hypertensive patients is more than two times (AOR = 2.41; 95% CI: 1.52–3.83) higher than those with normal blood pressure. Patients who had a history of microvascular DM complications had three times (AOR = 3.62; 95% CI: 2.01–6.53) the likelihood of developing CVD compared with T2DM patients who did not experience microvascular complications of DM (Table 3). Physical inactivity is positively associated with the development of CVD among T2DM patients in this study. It is in line with the study conducted in Ethiopia among T2DM patients that indicated physical inactivity is signiﬁcantly associated with the chronic complications of T2DM (48). Previous studies have consistently showed the protective eﬀect of physical activity in diabetic patients at any level of other risk factors for CVD in diabetic patients (29, 40, 49, 50). This might be the reason for the conclusion of lifestyle modiﬁcation that includes the exercise as one of the milestones of diabetes treatment and prevention of its complications (51). Data Collection Methods TABLE 2 \| Treatment categories and magnitude of CVD risk factors among T2DM patients in government hospitals of Ethiopia, from 2013 to 2017. CVD with Malik et al. (39) and Einarson et al. (37) might be due to the wide range of age composition (younger population) in this study. In this study, the prevalence of the CVD among T2DM is associated with the presence of older age, being physically inactive, drinking alcohol, BMI > 24.9 kg/m2, HTN, and presence of microvascular complications. Patients older than 60 years have more than three times the likelihood of developing CVD than those who are younger than 40 years. The prevalence of HTN is also seen increased with age. This result is comparable to the result of studies conducted in Pakistan (40), Taiwan (41), and Sweden (42), as all of these studies revealed older age is the important predictor of a cardiovascular event in T2DM. Aging can cause changes in the heart and blood vessels that may increase a person’s risk of developing CVD. Moreover, there is a high prevalence of atherosclerosis and arteriosclerosis due to the progression of diabetes in advanced age (43). The likelihood of CVD in the female is lowered by 10% compared to male T2DM patients, though it is not statistically signiﬁcant. There is heterogeneity between studies regarding the diﬀerence of risk of CVD among the male and female. Many studies indicated that the females are more likely to develop CVD (44–46). To the opposite of the former evidence from a review by Al-Salameh et al. (47), which showed that regardless of CVD being more prevalent in female in the absence of T2DM, the disparity disappears for the T2DM patients and the CVD is not barely associated with the sex diﬀerence but it is inﬂuenced by BMI above normal (overweight or obesity) and high prevalence of HTN after the age of 60–65 years in women. Hence, the reason for the non-signiﬁcant association of sex and CVD in this study might be due to the higher levels of risk factors (obesity, microvascular diabetes complications, and age). BP, blood pressure; BMI, body mass index; CVD, cardiovascular disease; kg/m2, kilograms per square meter; T2DM, type 2 diabetes mellitus. (BMI > 24.9 kg/m2) patients compared to patients who had normal weight (BMI 18.5–24.9 kg/m2). Frontiers in Public Health \| www.frontiersin.org Data Collection Methods However, in multivariate logistic regression, the signiﬁcant association was not seen for age between 40 and 60 years, place of residence, smoking, and history of infection. In the ﬁnal multivariate logistic regression model, age older than 60, being physically inactive, drinking alcohol, HTN, BMI > 24.9 kg/m2, and experiencing microvascular DM complications were signiﬁcantly and positively associated with CVD. Patients with age older than 60 years were having three times [adjusted odds ratio (AOR) = 3.22; 95% CI: 1.71–6.09] higher chance of experiencing CVD as compared with those aged <40 years. The odds of developing CVD was more than two times (AOR = 2.39; 95% CI: 1.17–6.06) higher among adults who consumed alcohol compared with T2DM patients who did not drink alcohol at all. Moreover, physically inactive T2DM patients were having 45% higher odds of developing CVD (AOR =1.45; 95 CI: 1.06–2.38) when compared to the physically active patients. risk factor for CVD was recorded among 272 (59.91%). From the total study participants, 274 (60.35%) were hypertensive. From 274 hypertensive patients, 166 (60.6%) of them were with uncontrolled blood pressure. Moreover, 124 (27.31%) were On the other hand, the odds of developing CVD was 1.81 times (AOR = 1.81; 95% CI: 1.07–3.07) higher among overweight February 2021 \| Volume 8 \| Article 532719 Frontiers in Public Health \| www.frontiersin.org 4 Cardiovascular Disease Among Type-2 Diabetes Regassa et al. TABLE 2 \| Treatment categories and magnitude of CVD risk factors among T2DM patients in government hospitals of Ethiopia, from 2013 to 2017. Variables Category Frequency Percentage (%) Treatment Insulin 54 11.89 Metformin only 175 38.55 Combined (metformin and glibenclamide) 225 49.56 Physical activities Active 330 72.69 Inactive 124 27.31 Cigarette smoking Never smoked 426 93.83 Currently smoking 28 6.17 Alcohol drinking Never drink 420 92.51 Currently drinking 34 7.49 Blood pressure Normal BP 180 39.65 Hypertensive 274 60.35 Hypertension (n = 274) Controlled 108 39.4 Uncontrolled 166 60.6 BMI 18.5–24.9 kg/m2 371 81.72 Above 24.9 kg/m2 83 18.28 Systemic infection No 312 68.72 Yes 142 31.28 Blood glucose level Good control 128 28.19 Poor control 326 71.81 Acute complication No 379 83.48 Yes 75 16.52 Microvascular complications No 370 81.50 Yes 84 18.50 BP, blood pressure; BMI, body mass index; CVD, cardiovascular disease; kg/m2, kilograms per square meter; T2DM, type 2 diabetes mellitus. DISCUSSION Identiﬁcation of potentially modiﬁable associated factors of CVD is an initial step to prevent and control CVD and its outcome among T2DM patients. In this study, the overall prevalence of CVD was 42.5% (95% CI: 38%, 47%). Speciﬁc complications account for 39% hypertensive heart diseases, 7% heart failure, and 2.2% stroke. This prevalence is higher than the pooled prevalence (32%) reported in a systematic literature review of studies in the period of 2007–2017, though the speciﬁc complications are lower in the current study (37). It is also higher than those of studies conducted in Iraq (38) and Scotland (39). This discrepancy might be due to the diﬀerence in sample size, follow-up, and treatment protocol between the countries. The diﬀerence in subtypes of Drinking alcohol is positively associated with the prevalence of CVD. The likelihood of CVD is increased by about 3-fold among alcohol consumers. Similar evidence was reported in a systematic review and meta-analysis of 20 studies (52) and individual study in southwest Ethiopia (48). Alcoholism (heavy or moderate drinking) might accelerate the development of coronary arterial diseases (CADs), as it causes systemic HTN, valvular diseases, cardiomyopathies, rhythm disturbances, and many non-cardiac problems, such as anemia, infection, and tumors (53–55). Though the risk is increased with the dose of intake and types of beverage, alcohol drinking is the risk for CVD in any individual (56). February 2021 \| Volume 8 \| Article 532719 Frontiers in Public Health \| www.frontiersin.org 5 Cardiovascular Disease Among Type-2 Diabetes Regassa et al. TABLE 3 \| Factors associated with cardiovascular disease among type 2 diabetes patients in government hospitals of the Harari region, Eastern Ethiopia, from 2013 to 2017 (N = 454). DISCUSSION Variables Categories CVD COR AOR Yes No Baseline age in years <40 46 109 1 1 40–60 87 113 1.82 [1.17–2.84] 1.53 [0.93–2.54] Above 60 60 30 4.74 [2.71–8.28] 3.22 [1.71–6.09]* Sex Male 114 145 1 1 Female 79 116 0.86 [0.59–1.26] 0.90 [0.71–1.83] Place of residence Urban 163 200 1 1 Rural 41 87 0.58 [0.38–0.88] 0.60 [0.36–1.01] Drinking alcohol No 169 251 1 1 Yes 24 10 3.51 [1.69–7.31] 2.39 [1.17–6.06]* Smoking No 174 252 1 1 Yes 19 9 3.18 [1.46–6.92] 2.36 [0.87–6.33] HTN No 55 125 1 1 Yes 138 136 2.31 [1.55–3.43] 2.41 [1.52–3.83]** Physical activity Active 119 211 1 1 Inactive 74 50 2.56 [1.70–3.86] 1.45 [1.06–2.38]* BMI 18.5–24.9 kg/m2 145 226 1 1 >24.9 kg/m2 48 35 2.33 [1.45–3.73] 1.81 [1.07–3.07]* Acute complication No 134 236 1 Yes 59 25 0.61 [0.36–1.01] 0.49 [0.27–1.92] History of infection No 118 194 1 1 Yes 75 67 1.96 [1.33–2.90] 1.86 [0.9–2.38] Microvascular complication No 134 236 1 1 Yes 59 25 4.39 [2.66–7.24] 3.62 [2.01–6.53]** Controlled blood glucose Yes 60 68 1 1 No 133 193 1.28 [0.88–2.01] 1.14 [0.70–1.82] BMI, body mass index; CVD, cardiovascular disease; SD, standard deviation; AOR, adjusted odds ratio; COR, crude odds ratio; HTN, hypertension. p < 0.05, p < 0.001. BMI, body mass index; CVD, cardiovascular disease; SD, standard deviation; AOR, adjusted odds ratio; COR, crude odds ratio; HTN, hypertension. p < 0.05, **p < 0.001. was evidence of strong association between hyperglycemia and CVD (40, 63, 65–67). Hyperglycemia is the principal cause of microvasculopathy but also appears to play an important role in causation of macrovasculopathy (40, 68). The risk of cardiovascular complications might be increased in the long term (more than 10 years’ duration); hyperglycemia as long-standing hyperglycemia induces the toxicity of endothelial cell and alters its function (11, 69, 70). The discrepancy might be due to the short period of follow-up in our study. Since the follow-up was limited to 5 years, it is rare to ﬁnd the association between the uncontrolled blood glucose and CVD. The association between smoking and CVD is not signiﬁcant in the current study. This ﬁnding is in line with a similar study conducted in Saidu Teaching Hospital, Pakistan (40). But most previous studies were reporting that smoking habit is the risk factor for the CVD, as it alters the process of controlling blood glucose level (57, 58). Frontiers in Public Health \| www.frontiersin.org REFERENCES 8. Gu K, Cowie CC, Harris MI. Diabetes and decline in heart disease mortality in Us adults. JAMA. (1999) 281:1291–7. doi: 10.1001/jama.281.14.1291 1. Nolan CJ, Ruderman NB, Kahn SE, Pedersen O, Prentki MJD. Insulin resistance as a physiological defense against metabolic stress: implications for the management of subsets of type 2 diabetes. Diabetes. (2015) 64:673–86. doi: 10.2337/db14-0694 1. Nolan CJ, Ruderman NB, Kahn SE, Pedersen O, Prentki MJD. Insulin resistance as a physiological defense against metabolic stress: implications for the management of subsets of type 2 diabetes. Diabetes. (2015) 64:673–86. doi: 10.2337/db14-0694 9. Haﬀner SJ, Cassells H. Hyperglycemia as a cardiovascular risk factor. Am J Med. (2003) 115:6–11. doi: 10.1016/j.amjmed.2003.09.009 10. Laakso M. Cardiovascular disease in type 2 diabetes: challenge for treatment and prevention. J Intern Med. (2001) 249:225–35. doi: 10.1046/j.1365-2796.2001.00789.x 2. Bandello F, Zarbin MA, Lattanzio R, Zucchiatti I (editors). Management of diabetic retinopathy. Dev Ophthalmol. (2017) 60:1–5. doi: 10.1159/0004 59641 11. Kitada M, Zhang Z, Mima A, King GL. Molecular mechanisms of diabetic vascular complications. J Diabetes Invest. (2010) 1:77–89. doi: 10.1111/j.2040-1124.2010.00018.x 3. Mensah GA. Descriptive epidemiology of cardiovascular risk factors and diabetes in Sub-Saharan Africa. Prog Cardiovasc Dis. (2013) 56:240–50. doi: 10.1016/j.pcad.2013.10.014 12. Kosiborod M, Gomes MB, Nicolucci A, Pocock S, Rathmann W, Shestakova MV, et al. Vascular complications in patients with type 2 diabetes: prevalence and associated factors in 38 countries (the Discover Study Program). Cardiovasc Diabetol. (2018) 17:150. doi: 10.1186/s12933-018-0787-8 4. WHO. Deﬁnition of Cardiovascular Diseases. (2017). Available online at: http://www.euro.who.int/en/health-topics/noncommunicable- diseases/cardiovascular-diseases/cardiovascular-diseases2/deﬁnition-of- cardiovascular-diseases 13. Mbanya J-C, Sobngwi E. Diabetes in Africa. Diabetes microvascular and macrovascular disease in Africa. J Cardiovasc Risk. (2003) 10:97–102. doi: 10.1177/174182670301000204 5. Birkeland KI, Jørgensen ME, Carstensen B, Persson F, Gulseth HL, Thuresson M, et al. Cardiovascular mortality and morbidity in patients with type 2 diabetes following initiation of sodium-glucose co-transporter- 2 inhibitors versus other glucose-lowering drugs (Cvd-real nordic): a multinational observational analysis. Lancet Diabetes Endocrinol. (2017) 5:709–17. doi: 10.1016/S2213-8587(17)30258-9 14. Dinesh Shah A, Langenberg C, Rapsomaniki E, Denaxas S, Pujades-Rodriguez M, Gale CP, et al. Type 2 diabetes and incidence of a wide range of cardiovascular diseases: a cohort study in 1.9 million people. Lancet. (2015) 385(Suppl.1):S86. doi: 10.1016/S0140-6736(15)60401-9 15. Dokken BB. The pathophysiology of cardiovascular disease and diabetes: beyond blood pressure and lipids. Diabetes Spectrum. (2008) 21:160–5. doi: 10.2337/diaspect.21.3.160 6. Laakso M. Hyperglycemia and cardiovascular disease in type 2 diabetes. Diabetes Care. (1999) 48:937–42. doi: 10.2337/diabetes.48.5.937 7. RECOMMENDATION To reduce the risk of CVD in T2DM, helping the patients to attain a healthy lifestyle by encouraging physical activity, weight reduction, and cessation of alcohol drinking during the follow- up is the essential step. In addition to this, we recommend a multifactorial intervention aimed at achieving recommended levels of critical indicators (blood pressure, blood cholesterol, microvascular complications, and treatment at early stage). CONCLUSION The prevalence of CVD was high and associated with advanced age, being physically inactive, drinking alcohol, BMI higher than 24.9 kg/m2, being hypertensive, and having microvascular complications. DISCUSSION In INTERHEART study from 52 countries, smoking was among the reported potentially modiﬁable risk factors of heart diseases (59). The discrepancy might be due to the low proportion of smokers among the study population of the current study. In the current study, there was an increment of CVD among hypertensive patients by more than two times. This study was in line with the study conducted in Sweden, as it showed that the risk of CVD is increased with the high blood pressure (31). The result from CVD research using linked bespoke studies and electronic health records (CALIBER) indicated that people with HTN had a higher lifetime risk of overall CVD and developed CVD 5 years earlier than those with normal blood pressure (30). HTN is also reported as a risk factor for CVD in published articles (31, 40–42, 60–63) through causing hypertensive heart diseases (64). In the present study, the likelihood of developing CVD is increased by 81% among patients whose BMI is >24.9 kg/m2 compared to those whose BMI is between 18.5 and 24.9 kg/m2. A similar result was reported by a systematic review of 57 individual studies—the positive relationship between obesity and increased prevalence rates of CVD (37). This might be due to the nature of T2DM, as the majority of individuals suﬀering from T2DM are obese and suﬀer from metabolic syndrome, which could be a risk factor for the major cardiovascular events (71). Patients who had a history of microvascular complications had an increased likelihood of developing CVD than their counterparts. Mainly recorded microvascular complications were retinopathy and nephropathy. Similarly, published studies reported the association between microvascular complication The likelihood of CVD is increased by 14% among individuals with uncontrolled blood glucose level compared to patients with controlled blood glucose. But this association was not signiﬁcant statistically. To the opposite of our result, there February 2021 \| Volume 8 \| Article 532719 Frontiers in Public Health \| www.frontiersin.org 6 Regassa et al. Cardiovascular Disease Among Type-2 Diabetes ACKNOWLEDGMENTS We are grateful to the respective hospital administrators, hospital staﬀworking at chronic follow-up clinics, and the data collectors for their willingness and unreserved contribution in this study. AUTHOR CONTRIBUTIONS LR contributed to the proposal development, data curation, investigation, formal analysis, methodology, project administration, writing the original draft, and writing, review, and editing. AT contributed to the proposal development, investigation, methodology, writing the original draft, and writing, review, and editing. YA contributed to the proposal development, investigation, methodology, writing the original draft, and writing, review, and editing. The manuscript was also developed through the active participation of all authors. All authors read and approved the manuscript. ETHICS STATEMENT Finally, this study has a number of limitations. Since this study was a retrospective document review in a resource-limited setting, there was no recorded HbA1c, as it is not routinely available, blood cholesterol level, diet, family history, and time at CVD diagnoses for most of the patients. The coronary artery diseases, the most common types of heart disease, were not reported in the records of the patients. Moreover, the criteria used to diagnose the type of CVD are not speciﬁcally registered, and the lifetime burden of risk factors was diﬃcult to determine. The studies involving human participants were reviewed and approved by University of Gondar College of Medical and Health Science Ethical Review Board. The patients/participants provided their written informed consent to participate in this study. DATA AVAILABILITY STATEMENT and CVD (70, 72–74). Diabetic vascular complications are a continuum and depend on each other (74). In addition to endothelial cell-dependent vascular injury mechanisms, endothelial cell-independent vascular dysfunction is leading to BK channelopathy and vascular complications in T2DM (75). All datasets generated for this study are included in the article/Supplementary Material. REFERENCES (2017) 317:912–24. doi: 10.1001/jama.2017.0947 42. Rawshani A, Rawshani A, Franzén S, Sattar N, Eliasson B, Svensson A-M, et al. Risk factors, mortality, and cardiovascular outcomes in patients with type 2 diabetes. N Engl J Med. (2018) 379:633–44. doi: 10.1056/NEJMoa1800256 22. Wingard DL, Barrett-Connor E. Heart disease and diabetes. Diabetes Am. (1995) 2:429–48. 43. Zhou K, Donnelly LA, Morris AD, Franks PW, Jennison C, Palmer CN, et al. Clinical and genetic determinants of progression of type 2 diabetes: a direct study. Diabetes Care. (2014) 37:718–24. doi: 10.2337/dc13-1995 23. Branch M, German C, Bertoni A, Yeboah J. Incremental risk of cardiovascular disease and/or chronic kidney disease for future Ascvd and mortality in patients with type 2 diabetes mellitus: accord trial. J Diabetes Complications. (2019) 33:468–72. doi: 10.1016/j.jdiacomp.2019.04.004 44. Ballotari P, Venturelli F, Greci M, Giorgi Rossi P, Manicardi V. Sex diﬀerences in the eﬀect of type 2 diabetes on major cardiovascular diseases: results from a population-based study in Italy. Int J Endocrinol. (2017) 2017: 6039356. doi: 10.1155/2017/6039356 (2019) 33:468–72. doi: 10.1016/j.jdiacomp.2019.04.004 24. Petrie JR, Guzik TJ, Touyz RM. Diabetes, hypertension, and cardiovascular disease: clinical insights and vascular mechanisms. Can J Cardiol. (2018) 34:575–84. doi: 10.1016/j.cjca.2017.12.005 45. Gouni-Berthold I, Berthold HK, Mantzoros CS, Böhm M, Krone WJDc. Sex disparities in the treatment and control of cardiovascular risk factors in type 2 diabetes. Diabetes Care. (2008) 31:1389-91. doi: 10.2337/dc08-0194 25. Hayward RA, Reaven PD, Wiitala WL, Bahn GD, Reda DJ, Ge L, et al. Follow- up of glycemic control and cardiovascular outcomes in type 2 diabetes. N Engl J Med. (2015) 372:2197–206. doi: 10.1056/NEJMoa1414266 46. Eeg-Olofsson K, Cederholm J, Nilsson P, Zethelius B, Nunez L, Gudbjörnsdóttir S, et al. Risk of cardiovascular disease and mortality in overweight and obese patients with type 2 diabetes: an observational study in 13,087 patients. Diabetologia. (2009) 52:65. doi: 10.1007/s00125-008-1190-x 26. Sisay H. Nursing Process and Factors Aﬀecting Its Utilization in Governmental Hospitals in Harari and Dire Dawa regions (Doctoral dissertation), Haramaya University, Dire Dawa. 47. Al-Salameh A, Chanson P, Bucher S, Ringa V, Becquemont L, editors. Cardiovascular disease in type 2 diabetes: a review of sex-related diﬀerences in predisposition and prevention. Mayo Clin Proc. (2019) 94:287–308. doi: 10.1016/j.mayocp.2018.08.007 27. Bhatti GK, Bhadada SK, Vijayvergiya R, Mastana SS, Bhatti JS. Metabolic syndrome and risk of major coronary events among the urban diabetic patients: North Indian diabetes and cardiovascular disease study—Nidcvd-2. J Diabetes Complications. (2016) 30:72–8. doi: 10.1016/j.jdiacomp.2015.07.008 28. REFERENCES Roger VL, Sidney S, Fairchild AL, Howard VJ, Labarthe DR, Shay CM, et al. Recommendations for cardiovascular health and disease surveillance for 2030 and beyond: a policy statement from the American Heart Association. Circulation. (2020) 141:e104–19. doi: 10.1161/CIR.0000000000000756 48. Tamiru S, Alemseged F. A risk factors for cardiovascular diseases among diabetic patients in Southwest Ethiopia. Ethiop J Health Sci. (2010) 20:2. doi: 10.4314/ejhs.v20i2.69438 49. Selvin E, Marinopoulos S, Berkenblit G, Rami T, Brancati FL, Powe NR, et al. Meta-analysis: glycosylated hemoglobin and cardiovascular disease in diabetes mellitus. Ann Intern Med. (2004) 141:421–31. doi: 10.7326/0003-4819-141-6-200409210-00007 29. Hu G, Jousilahti P, Barengo NC, Qiao Q, Lakka TA, Tuomilehto J. Physical activity, cardiovascular risk factors, and mortality among ﬁnnish adults with diabetes. Diabetes Care. (2005) 28:799–805. doi: 10.2337/diacare.28.4.799 50. Tanasescu M, Leitzmann MF, Rimm EB, Hu FB. Physical activity in relation to cardiovascular disease and total mortality among men with type 2 diabetes. Circulation. (2003) 107:2435–9. doi: 10.1161/01.CIR.0000066906.11 109.1F 30. Rapsomaniki E, Timmis A, George J, Pujades-Rodriguez M, Shah AD, Denaxas S, et al. Blood pressure and incidence of 12 cardiovascular diseases: lifetime risks, healthy life-years lost, and age-speciﬁc associations in 1.25 million people. Lancet Diabetes. (2014) 383:1899–911. doi: 10.1016/S0140-6736(14)60685-1 51. Praet S, Van Loon L. Exercise: the brittle cornerstone of type 2 diabetes treatment. Diabetologia. (2008) 51:398–401. doi: 10.1007/s00125-007-0910-y 31. Afghahi H, Svensson MK, Pirouzifard M, Eliasson B, Svensson A-M. Blood pressure level and risk of major cardiovascular events and all- cause of mortality in patients with type 2 diabetes and renal impairment: an observational study from the Swedish National Diabetes Register. Diabetologia. (2015) 58:1203–11. doi: 10.1007/s00125-015-3548-1 52. Baliunas DO, Taylor BJ, Irving H, Roerecke M, Patra J, Mohapatra S, et al. Alcohol as a risk factor for type 2 diabetes: a systematic review and meta- analysis. J Diabetes Care. (2009) 32:2123–32. doi: 10.2337/dc09-0227 53. Klatsky AL. Alcohol, cardiovascular diseases and diabetes mellitus. Pharmacol Res. (2007) 55:237–47. doi: 10.1016/j.phrs.2007.01.011 32. Ivanov TD, Ivanova MJK. American Diabetes Association. Standards of medical care in diabetes-−2017. Kidneys. (2017) 6:47–63. doi: 10.22141/2307-1257.6.1.2017.93784 54. Polsky S, Akturk HK. Alcohol consumption, diabetes risk, and cardiovascular disease within diabetes. Curr Diab Rep. (2017) 17:136. doi: 10.1007/s11892-017-0950-8 33. World Health Organization. Bmi Classiﬁcation, 2014. c2020–02. Available online at: http://apps.who.int/bmi/index.jsp?introPage=intro_3.html. 55. Wahid A, Manek N, Nichols M, Kelly P, Foster C, Webster P, et al. Quantifying the association between physical activity and cardiovascular disease and diabetes: a systematic review and meta-analysis. REFERENCES O’Donnell MJ, Chin SL, Rangarajan S, Xavier D, Liu L, Zhang H, et al. Global and regional eﬀects of potentially modiﬁable risk factors associated with acute stroke in 32 countries (interstroke): a case-control study. Lancet Diabetes. (2016) 388:761–75. doi: 10.1016/S0140-6736(16)30506-2 16. Meigs JB, Mittleman MA, Nathan DM, Toﬂer GH, Singer DE, Murphy- Sheehy PM, et al. Hyperinsulinemia, hyperglycemia, and impaired hemostasis: the framingham oﬀspring study. JAMA. (2000) 283:221–8. doi: 10.1001/jama.283.2.221 Frontiers in Public Health \| www.frontiersin.org February 2021 \| Volume 8 \| Article 532719 7 Cardiovascular Disease Among Type-2 Diabetes Regassa et al. 17. Festa A, D’Agostino R Jr, Howard G, Mykkanen L, Tracy RP, Haﬀner SM. Chronic subclinical inﬂammation as part of the insulin resistance syndrome: the insulin resistance atherosclerosis study (Iras). Circulation. (2000) 102:42– 7. doi: 10.1161/01.CIR.102.1.42 37. Einarson TR, Acs A, Ludwig C, Panton UH. Prevalence of cardiovascular disease in type 2 diabetes: a systematic literature review of scientiﬁc evidence from across the world in 2007–2017. Cardiovasc Diabetol. (2018) 17:83. doi: 10.1186/s12933-018-0728-6 18. Lin JS, Evans CV, Johnson E, Redmond N, Burda BU, Coppola EL, et al. Nontraditional Risk Factors in Cardiovascular Disease Risk Assessment: A Systematic Evidence Report for the U.S. Preventive Services Task Force. Rockville, MD: Agency for Healthcare Research and Quality (US) (2018). doi: 10.1001/jama.2018.4242 38. Mansour AA, Ajeel NA. Atherosclerotic cardiovascular disease among patients with type 2 diabetes in Basrah. World J Diabetes. (2013) 4:82–7. doi: 10.4239/wjd.v4.i3.82 39. Malik MO, Govan L, Petrie JR, Ghouri N, Leese G, Fischbacher C, et al. Ethnicity and risk of cardiovascular disease (Cvd): 4.8 years follow-up of patients with type 2 diabetes living in Scotland. Diabetologia. (2015) 58:716– 25. doi: 10.1007/s00125-015-3492-0 19. Singh P, Singh S, Pandi-Jain GS. Eﬀective heart disease prediction system using data mining techniques. Int J Nanomed. (2018) 13:121–4. doi: 10.2147/IJN.S124998 40. Akhtar S, Asghar N. Risk factors of cardiovascular disease in district Swat. J Pak Med Assoc. (2015) 65:1001–4. 20. Kannel WB, McGee DL. Diabetes and cardiovascular disease: the framingham study. JAMA. (1979) 241:2035–8. doi: 10.1001/jama.1979.03290450033020 41. Kornelius E, Chiou J-Y, Yang Y-S, Lu Y-L, Peng C-H, Huang C-N. The diabetes shared care program and risks of cardiovascular events in type 2 diabetes. Am J Med. (2015) 128:977–85. doi: 10.1016/j.amjmed.2015.03.025 21. Micha R, Peñalvo JL, Cudhea F, Imamura F, Rehm CD, Mozaﬀarian D. Association between dietary factors and mortality from heart disease, stroke, and type 2 diabetes in the United States. JAMA. REFERENCES J Am Heart Assoc. (2016) 5:e002495. doi: 10.1161/JAHA.115.002495 34. Martín-Timón I, Sevillano-Collantes C, Segura-Galindo A, Del Cañizo- Gómez FJ. Type 2 diabetes and cardiovascular disease: have all risk factors the same strength? World J Diabetes. (2014) 5:444–70. doi: 10.4239/wjd.v5. i4.444 56. Piano MR, Mazzuco A, Kang M, Phillips SA. Cardiovascular consequences of binge drinking: an integrative review with implications for advocacy, policy, and research. Alcohol Clin Exp Res. (2017) 41:487–96. doi: 10.1111/acer.13329 35. Stern MPJD. Diabetes and cardiovascular disease: The “Common Soil” hypothesis. Perspect Diabetes. (1995) 44:369–74. doi: 10.2337/diab.44.4.369 36. Bozdogan H. Akaike’s information criterion and recent developments in information complexity. J Math Psychol. (2000) 44:62–91. doi: 10.1006/jmps.1999.1277 57. Balakumar P, Maung-U K, Jagadeesh G. Prevalence and prevention of cardiovascular disease and diabetes mellitus. Pharmacol Res. (2016) 113:600– 9. doi: 10.1016/j.phrs.2016.09.040 February 2021 \| Volume 8 \| Article 532719 Frontiers in Public Health \| www.frontiersin.org 8 Cardiovascular Disease Among Type-2 Diabetes Regassa et al. 58. Sasso FC, Chiodini P, Carbonara O, De Nicola L, Conte G, Salvatore T, et al. High cardiovascular risk in patients with type 2 diabetic nephropathy: the predictive role of albuminuria and glomerular ﬁltration rate. The Nid- 2 prospective cohort study. Nephrol Dial Transplant. (2012) 27:2269–74. doi: 10.1093/ndt/gfr644 68. Cardoso CRL, Leite NC, Moram CBM, Salles GF. Long-term visit-to-visit glycemic variability as predictor of micro- and macrovascular complications in patients with type 2 diabetes: the Rio de Janeiro type 2 diabetes cohort study. Cardiovasc Diabetol. (2018) 17:33. doi: 10.1186/s12933-018-0677-0 69. Giacco F, Brownlee M. Oxidative stress and diabetic complications. Circulation. (2010) 107:1058–70. doi: 10.1161/CIRCRESAHA.110.223545 59. Yusuf S, Hawken S, Ôunpuu S, Dans T, Avezum A, Lanas F, et al. Eﬀect of potentially modiﬁable risk factors associated with myocardial infarction in 52 countries (the interheart study): case-control study. Lancet. (2004) 364:937–52. doi: 10.1016/S0140-6736(04)17018-9 70. van Wijngaarden RPT, Overbeek JA, Heintjes EM, Schubert A, Diels J, Straatman H, et al. Relation between diﬀerent measures of glycemic exposure and microvascular and macrovascular complications in patients with type 2 diabetes mellitus: an observational cohort study. Diabetes Ther. (2017) 8:1097–109. doi: 10.1007/s13300-017-0301-4 60. Miller IM, Ellervik C, Yazdanyar S, Jemec GB. Meta-analysis of psoriasis, cardiovascular disease, and associated risk factors. J Am Acad Dermatol. (2013) 69:1014–24. doi: 10.1016/j.jaad.2013.06.053 71. Scheen AJ. Pathophysiology of type 2 diabetes. Acta Clinica Belgica. (2003) 6:58. doi: 10.1179/acb.2003.58.6.001 61. Keech A, Simes RJ, Barter P, Best J, Scott R, Taskinen MR, et al. REFERENCES For the ﬁeld study investigators. Eﬀects of long-term fenoﬁbrate therapy on cardiovascular events in 9,795 people with type 2 diabetes mellitus (the ﬁeld [fenoﬁbrate intervention and event lowering in diabetes] study): randomised controlled trial. Lancet. (2005) 366:1849–61. doi: 10.1016/S0140-6736(05)67667-2 72. Gimeno-Orna JA, Faure-Nogueras E, Castro-Alonso FJ, Boned-Juliani B. Ability of retinopathy to predict cardiovascular disease in patients with type 2 diabetes mellitus. Am J Cardiol. (2009) 103:1364–7. doi: 10.1016/j.amjcard.2009.01.345 62. Shah AD, Langenberg C, Rapsomaniki E, Denaxas S, Pujades-Rodriguez M, Gale CP, et al. Type 2 diabetes and incidence of cardiovascular diseases: a cohort study in 1.9 million people. Lancet Diabetes Endocrinol. (2015) 3:105–13. doi: 10.1016/S2213-8587(14)70219-0 73. Seferovic JP, Bentley-Lewis R, Claggett B, Diaz R, Gerstein HC, Køber LV, et al. Retinopathy, neuropathy, and subsequent cardiovascular events in patients with type 2 diabetes and acute coronary syndrome: analysis from elixa and importance of duration of diabetes. Circulation. (2016) 134(Suppl.1):A19191. doi: 10.1155/2018/1631263 63. Vazquez-Benitez G, Desai JR, Xu S, Goodrich GK, Schroeder EB, Nichols GA, et al. Preventable major cardiovascular events associated with uncontrolled glucose, blood pressure, and lipids and active smoking in adults with diabetes with and without cardiovascular disease: a contemporary analysis. Diabetes Care. (2015) 38:905–12. doi: 10.2337/dc14-1877 74. Chawla A, Chawla R, Jaggi S. Microvasular and macrovascular complications in diabetes mellitus: distinct or continuum? Indian J Endocrinol Metabol. (2016) 20:546. doi: 10.4103/2230-8210.183480 75. Lu T, Lee H-C. Impaired Vascular BK Channel Function in Type 2 Diabetes Mellitus, Medical Complications of Type 2 Diabetes, Croniger C, editor. IntechOpen. (2011). doi: 10.5772/22077. Available online at: https:// www.intechopen.com/books/medical-complications-of-type-2-diabetes/ impaired-vascular-bk-channel-function-in-type-2-diabetes-mellitus 64. Tefera YG, Abegaz TM, Abebe TB, Mekuria AB. The changing trend of cardiovascular disease and its clinical characteristics in Ethiopia: hospital- based observational study. Vasc Health Risk Manage. (2017) 13:143. doi: 10.2147/VHRM.S131259 65. Reaven PD, Emanuele NV, Wiitala WL, Bahn GD, Reda DJ, McCarren M, et al. Intensive glucose control in patients with type 2 diabetes-−15-years follow-up. N Engl J Med. (2019) 380:2215–24. doi: 10.1056/NEJMoa1806802 Conﬂict of Interest: The authors declare that the research was conducted in the absence of any commercial or ﬁnancial relationships that could be construed as a potential conﬂict of interest. 66. Freemantle N, Danchin N, Calvi-Gries F, Vincent M, Home PD. Relationship of glycaemic control and hypoglycaemic episodes to 4-years cardiovascular outcomes in people with type 2 diabetes starting insulin. Diabetes Obes Metab. (2016) 18:152–8. doi: 10.1111/dom.12598 Copyright © 2021 Regassa, Tola and Ayele. Frontiers in Public Health \| www.frontiersin.org February 2021 \| Volume 8 \| Article 532719 REFERENCES This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. 67. Kamuhabwa AR, Charles E. Predictors of poor glycemic control in type 2 diabetic patients attending public hospitals in Dar Es Salaam. Drug Healthcare Patient Saf. (2014) 6:155–65. doi: 10.2147/DHPS.S 68786 February 2021 \| Volume 8 \| Article 532719 Frontiers in Public Health \| www.frontiersin.org 9
https://openalex.org/W3029899877	https://europepmc.org/articles/pmc7270138?pdf=render	English	null	Dynamics of porous and amorphous magnesium borohydride to understand solid state Mg-ion-conductors	Scientific reports	2,020	cc-by	9,828	Dynamics of porous and amorphous magnesium borohydride to understand solid state Mg-ion- conductors Michael Heere 1,2 ✉, Anna-Lena Hansen1, SeyedHosein Payandeh 3, Neslihan Aslan 4, Gökhan Gizer 5, Magnus H. Sørby6, Bjørn C. Hauback6, Claudio Pistidda5, Martin Dornheim5 & Wiebke Lohstroh2 ✉ Rechargeable solid-state magnesium batteries are considered for high energy density storage and usage in mobile applications as well as to store energy from intermittent energy sources, triggering intense research for suitable electrode and electrolyte materials. Recently, magnesium borohydride, Mg(BH4)2, was found to be an effective precursor for solid-state Mg-ion conductors. During the mechanochemical synthesis of these Mg-ion conductors, amorphous Mg(BH4)2 is typically formed and it was postulated that this amorphous phase promotes the conductivity. Here, electrochemical impedance spectroscopy of as-received γ-Mg(BH4)2 and ball milled, amorphous Mg(BH4)2 confirmed that the conductivity of the latter is ~2 orders of magnitude higher than in as-received γ-Mg(BH4)2 at 353 K. Pair distribution function (PDF) analysis of the local structure shows striking similarities up to a length scale of 5.1 Å, suggesting similar conduction pathways in both the crystalline and amorphous sample. Up to 12.27 Å the PDF indicates that a 3D net of interpenetrating channels might still be present in the amorphous phase although less ordered compared to the as-received γ-phase. However, quasi elastic neutron scattering experiments (QENS) were used to study the rotational mobility of the [BH4] units, revealing a much larger fraction of activated [BH4] rotations in amorphous Mg(BH4)2. These findings suggest that the conduction process in amorphous Mg(BH4)2 is supported by stronger rotational mobility, which is proposed to be the so-called “paddle-wheel” mechanism. Energy storage is one of the grand challenges for present and future generations. In recent years, intermittent renewable energy production has increased worldwide resulting in a high demand for energy storage systems. “Beyond Li-batteries”, which are all-solid-state batteries with alternative working ions, including Na and Mg, are considered a promising alternative as they are cheaper and with respect to their natural abundancy more sus- tainable. However, the transport properties of larger Na+ or double-charged Mg2+ are challenging and directly correlated to the underlying crystal structure and dynamics. Understanding the accompanying structural and dynamic changes as well as finding high-performance cathode materials remain bottlenecks for the improvement of Mg-ion batteries1. g Mg-ion batteries (MIBs) have several advantages compared to Li-ion technology2. For instance, the low elec- trochemical potential of −2.4 V (vs. www.nature.com/scientificreports www.nature.com/scientificreports Scientific Reports \| (2020) 10:9080 \| https://doi.org/10.1038/s41598-020-65857-6 Results and Discussionsh The characterization of as-received and ball milled Mg(BH4)2 by synchrotron radiation powder X-ray diffrac- tion (SR-PXD) and synchrotron X-ray total scattering experiments with corresponding PDF analysis is shown in Fig. 1. Figure 1a confirms the highly symmetric cubic structure with space group Id–3a for as-received γ-Mg(BH4)2, which we refer to as “crystalline”19. A single Mg atom is coordinated by the edges of four tetrahedral [BH4] groups. γ-Mg(BH4)2 has a 3D net of interpenetrating channels of ~12.3 Å outer diameter giving a poros- ity of ~33%. The ball milled Mg(BH4)2 shows diffuse scattering apart from minor Bragg peaks of γ-Mg(BH4)2, indicating a mostly X-ray amorphous phase, which we refer to as “amorphous”. Therefore notations such as “as-received” (ar) for the crystalline γ-phase as well as “ball milled” (bm) for the amorphous-Mg(BH4)2 will be used interchangeably.hh g y The PDFs of the amorphous and crystalline γ-Mg(BH4)2 are depicted in Fig. 1b. The PDF of the amorphous sample can be divided in two main parts, A and B, with rA < 5.1 Å representing the internal local structure of the main constituents or building blocks and the region 5.1 Å < rB < 12.3 Å, revealing insights into the interplay (coherence) between the building blocks. Besides the first Mg – H bond, which will be discussed in more detail, the local structure of the amorphous sample agrees well with the crystalline one up to ~5.1 Å. Above ~5.1 Å, a slight oscillation is still observable, and with the aforementioned 3D net of interpenetrating channels of 12.27 Å diameter, this could indicate that the fundamental structure of the amorphous sample is still formed by these channels, even though less well-ordered. Only above ~12.3 Å the structure is completely random resulting in a featureless PDF. The good agreement of the local structure up to the first Mg – Mg distance at 4.82 Å supports spectroscopic results by Filinchuk et al. that the main building blocks of the structure are Mg – BH4 – Mg units (Fig. 1c)19 and that their reported X-ray amorphous phase revealed similarities to the local structure of the γ- and δ-Mg(BH4)2 19. The B-B distances are less pronounced in the PDF of the amorphous samples, indicating a higher amount of disorder and less correlation between the tetrahedra. www.nature.com/scientificreports/ more than 2% in the earth crust (compared to 0.006% for lithium). Mg seems beneficial due to its non-toxicity, easy machining and handling properties3,4. Although a recent study suggests the appearance of Mg dendrites5, Mg is much less prone to form dendrites than Li6, where hazardous Li plating is a major safety concern. Thus, ‘pure’ Mg metal, with a very high capacity, could be used as safe and reliable anode material. g y g p y In 2017, Mohtadi and Orimo stated that the present research on complex metal hydrides is experiencing a “renaissance as energy materials”7. Nevertheless, for the success of future complex metal hydride-based research, the development of highly conductive electrolytes and electrodes is one of the main requirements for a succes- sor of the Li-ion battery8. A first Mg-ion electrolyte based on complex metal hydrides was reported in 2012. Mohtadi et al. demonstrated the possibility to employ magnesium tetrahydroborate, colloquially referred to as magnesium borohydride Mg(BH4)2, dissolved in dimethoxy ethane (DME) in a rechargeable magnesium bat- tery9. Shortly after, Zhao-Karger et al. published an ionic Mg electrolyte based on the precursor Mg(BH4)2 with the reported reaction product being Mg(BR4)2 in DME (R = –OCH(CF3)2). This electrolyte showed the highest, so far, reported electrochemical stability window of 4.3 V while being stable in air, and an ionic conductivity of 0.011 S·cm−1 in a 0.3 M DME solution10. Recently, a new compound synthesized from Mg(BH4)2 and ethylenedi- amine (C2H8N2, ‘en’) was reported to have a high magnesium ion conductivity of up to 6·10−5 S·cm−1 at 343 K in the solid state11, while a follow up with different organic complexes reported details on conductivities with focus on Mg(BH4)2-diglyme0.5 12. The authors reported that the amorphous Mg(BH4)2 phase has a beneficial influence on the conductivities. Amorphization has also been reported to be beneficial for ionic conductivity in other material classes such as glassy solid electrolytes based on Li3PS4 13,14. Returning to complex metal hydrides: In 2014, Mg(BH4)2 and Mg(BH4)(NH4) were proven to be “solid state” Mg ion conductors15 while recently, Le Ruyet et al. reported a correlation between an amorphous phase found in a Mg−B − N − H system by NMR with a high conductivity of 3·10−6 S·cm−1 at 373 K for a solid-state Mg conductor based on Mg(BH4)(NH2)16. www.nature.com/scientificreports/ In general, the work on Mg−B − N − H systems and the exploration of for instance dihydrogen bonds17 seems to increase with a recent study even showing a conductivity of σ = 3.3·10−4 S cm–1 at T = 353 K for Mg(BH4)2·NH3 18. y g y g 4 2 3 Mg(BH4)2 seems very suitable as a precursor for Mg-ion electrolytes and hence the correlation between amor- phous Mg(BH4)2 and a high ion conductivity deserves further attention. The present work addresses the structure and dynamics of porous (γ–phase) and amorphous Mg(BH4)2. Hydrogen dynamics in both materials are studied on the picosecond timescale using quasi-elastic neutron scattering. Moreover, the materials are investigated by synchrotron X-ray powder diffraction and total scattering, which were employed to elaborate the local structures despite the lack of long-range order through pair distribution function (PDF) analysis. Especially, PDF analysis is very well suited to give insights into the local structure of amorphous materials. The conductivity is determined and the results are discussed in view of the structure and dynamics in porous and amorphous Mg(BH4)2. Scientific Reports \| (2020) 10:9080 \| https://doi.org/10.1038/s41598-020-65857-6 Dynamics of porous and amorphous magnesium borohydride to understand solid state Mg-ion- conductors standard hydrogen electrode (SHE)) is close to the one of Li with −3.0 V of Li/Li+, which allows for high cell voltages. Furthermore, Mg metal has a higher volumetric capacity of 3833 mA·cm−3 compared to 2036 mAh·cm−3 of Li metal, and magnesium has a higher natural abundancy of 1Institute for Applied Materials—Energy Storage Systems (IAM-ESS), Karlsruhe Institute of Technology (KIT), 76344, Eggenstein, Germany. 2Heinz Maier-Leibnitz Zentrum (MLZ), Technische Universität München, Lichtenbergstr. 1, 85748, Garching bei München, Germany. 3Empa, Swiss Federal Laboratories for Materials Science and Technology, 8600, Dübendorf, Switzerland. 4German Engineering Materials Science Centre (GEMS) at Heinz Maier-Leibnitz Zentrum (MLZ), Helmholtz-Zentrum Geesthacht GmbH, Lichtenbergstr. 1, 85748, Garching, Germany. 5Institute of Materials Research, Materials Technology, Helmholtz-Zentrum Geesthacht, D-21502, Geesthacht, Germany. 6Department for Neutron Materials Characterization, Institute for Energy Technology, NO-2027, Kjeller, Norway. ✉e-mail: Michael.Heere@kit.edu; Wiebke.Lohstroh@frm2.tum.de www.nature.com/scientificreports/ Results and Discussionsh It has to be noted, that even hydrogen bonds can be observed in the PDF, due to the fact that H has a formal oxidation state of –1 and therefore a notable electron density. However, to understand the different intensities of the Mg – H peak of the amorphous and the crystalline sample, two peculiarities have to be kept in mind. First, just as in B2H6, Mg – BH4 – Mg are interconnected via three-centre-two-electron bonds (tc-te). This means that the two electrons are “smeared out” over three atoms as depicted in the inset of Fig. 1d. Second, only information about electron densities are accessible using X-ray scattering methods. Consequently, in case of an X-ray PDF, the peak of corresponding tc-te bonds is expected to have only half of the intensity and being broadened, due to the delocalization of electrons. The fact that the Mg – H peak of the amorphous sample is more intense and narrower, reveals the presence of fewer tc-te bonds and therefore, more terminating Mg – H bonds.i To account for the bonding situation, the occupancy of H atoms was reduced to 0.5 in the models to fit the PDF. To refine the structural model of γ-Mg(BH4)2 the so called real space Rietveld approach was used20. It was not possible to fit the PDF over the whole range, therefore, it was divided into a short range ordering (SRO) and a long range ordering (LRO) part, being r < 12.3 Å ( = diameter of a interpenetrating channel) and r > 12.3 Å, Scientific Reports \| (2020) 10:9080 \| https://doi.org/10.1038/s41598-020-65857-6 www.nature.com/scientificreports/ Figure 1. (a) SR-PXD of γ-Mg(BH4)2 (blue curve) and amorphous-Mg(BH4)2 (red curve). λ = 0.2072 Å. Inset in the upper left is showing three [BH4] tetrahedra in their respective Mg setting and a magnification into one tetrahedron and its rotational axes are shown. C3 is the 3-fold 120° axis and C2∥ and C2⊥ are the 2-fold 180° axis. The inset image shows the crystal structure of γ-Mg(BH4)2 with one interpenetrating channel as reported in ref. 19. Spheres in orange: Mg-, green: B- and grey: H-atoms. (b) PDF obtained from total scattering data collected at P02.1 (DESY) of amorphous and crystalline γ-Mg(BH4)2. λ = 0.20723 Å, inset: One 1D channel of the structure with 12.27 Å diameter. (c) Peaks of the local structure of the amorphous PDF agree well with the crystalline one up to ~5.1 Å. Results and Discussionsh The last coinciding peak is the Mg – Mg distance which is marked in orange in the figure and the structural model inset. The most intense peak corresponds to the Mg-B bond (green). (d) Real space Rietveld fitting of the PDF of γ-Mg(BH4)2. The PDF was fitting using two different models. Details can be found in the text. Inset: Sketch of a three-centre-two-electron-bond. Figure 1. (a) SR-PXD of γ-Mg(BH4)2 (blue curve) and amorphous-Mg(BH4)2 (red curve). λ = 0.2072 Å. Inset in the upper left is showing three [BH4] tetrahedra in their respective Mg setting and a magnification into one tetrahedron and its rotational axes are shown. C3 is the 3-fold 120° axis and C2∥ and C2⊥ are the 2-fold 180° axis. The inset image shows the crystal structure of γ-Mg(BH4)2 with one interpenetrating channel as reported in ref. 19. Spheres in orange: Mg-, green: B- and grey: H-atoms. (b) PDF obtained from total scattering data collected at P02.1 (DESY) of amorphous and crystalline γ-Mg(BH4)2. λ = 0.20723 Å, inset: One 1D channel of the structure with 12.27 Å diameter. (c) Peaks of the local structure of the amorphous PDF agree well with the crystalline one up to ~5.1 Å. The last coinciding peak is the Mg – Mg distance which is marked in orange in the figure and the structural model inset. The most intense peak corresponds to the Mg-B bond (green). (d) Real space Rietveld fitting of the PDF of γ-Mg(BH4)2. The PDF was fitting using two different models. Details can be found in the text. Inset: Sketch of a three-centre-two-electron-bond. respectively. No difference was observed if H was included in the model to fit the LRO. The modelled PDF agrees reasonably well up to 60 Å, but fitting the SRO certainly needs more attention. yi g y Various models were used to describe the SRO features in the PDF (Fig. A1, supplementary information SI). All the models are based on the cubic γ-Mg(BH4)2 structure. All peak positions agree well, as already indicated in Fig. 1d. Nevertheless, especially the Mg – Mg peaks (4.82 Å, 7.40 Å, 8.85 Å) differ in their relative intensities. More precisely, either the intensity of the higher orders or the intensity of the first peaks can be fitted by using large or small anisotropic temperature factors, model 1 or 2, respectively. Results and Discussionsh temperature to 432 K and 461 K compared to the as-received sample, at the same heating rate, which is in good agreement with literature22. The decomposition reactions are not reported here but generally start from ~470 K with the release of hydrogen and the formation of non-crystalline, intermediate Mg–B–H species of different stoichiometry23,24. In order to investigate the internal dynamics of γ- and amorphous Mg(BH4)2, quasi-elastic neutron scatter- ing (QENS) experiments have been conducted. QENS is a powerful technique to investigate stochastic motions such as diffusion or jump rotations in condensed matter, which result in characteristic broadening of the elas- tic line at zero energy transfer (∆E = 0 meV). In general, a good overview of dynamic investigation of complex metal hydrides can be found in ref. 25. Figure 2 shows the obtained scattering function S(Q, ∆E) at λ1 = 5 Å for as-received and ball milled Mg(BH4)2 at various temperatures. The results at 310 K (Fig. 2a) indicate that the low energy spectra – both the quasi-elastic and the inelastic contribution – are strongly dependent on the local structure as can be seen by the comparison of as-received γ-Mg(BH4)2 and ball milled amorphous material. While as-received Mg(BH4)2 at 310 K shows almost no quasi-elastic scattering around the elastic peak, the ball milled material exhibits significant broadening of the elastic peak indicating a higher rotational mobility of the [BH4] units (for details see Figs. 3 and 4). Furthermore, distinct low energy (vibrational) inelastic peaks are observed in the as-received γ-species, which cannot be resolved in the ball milled sample. The energy of the inelastic excita- tions is constant over all probed Q-values.hi p Both samples underwent two heating cycles. The first heating went up to temperatures below the ε-phase transition (419 K for ar and 410 K for bm), the second one above the phase transformation temperature (435 K for ar and 431 K for bm). After the first heating a crystallization reaction towards γ-phase occurred in the ball milled sample, which resulted in a very similar scattering function compared to the as-received sample (Fig. 2b). The subsequent cooling step to 310 K was essential to compare the two samples near room temperature (after crystallization of bm phase to the γ-phase). The quasi-elastic signal and inelastic scattering in Fig. 2c, measured at 310 K, seems to be an average of those depicted in Fig. 2a. Results and Discussionsh This reveals, that the Mg – Mg correlation within the interpenetrating channels cannot be described in accordance with the cubic symmetry. We therefore, carefully allowed the Mg ions to move individually into the direction the temperature factors were pointing, e.g. into the channels (model 3), leaving their positions given by the cubic symmetry. To simplify this model, the H atoms were not included and isotropic temperature factors were utilized. Comparing all three models in Fig. A1, model 3 seems the best to describe the local Mg disorder for the SRO. In future, neutron powder diffraction meas- urements are planned especially within the frame of the “Energy research with Neutrons (ErwiN)” instrument at the MLZ, Germany21.f y In situ SR-PXD and thermogravimetric and differential thermal analysis (TG-DTA) are shown in Fig. A2 and A3 in the SI. TG-DTA data are presented in Fig. A3, showing the various phase evolutions in the investigated samples until 500 K, which all correspond to the literature22. The γ-Mg(BH4)2 transforms into ε-Mg(BH4)2 at Tpeak = 439 K and into β’-Mg(BH4)2 at Tpeak = 470 K. Both are endothermic events (blue curve in Fig. A3). The ball milled amorphous-Mg(BH4)2 transforms via an exothermic event first into γ-Mg(BH4)2 at 372 K (crystallization), followed by the endothermic transitions towards the ε- and β′-phase. The latter events are slightly reduced in peak Scientific Reports \| (2020) 10:9080 \| https://doi.org/10.1038/s41598-020-65857-6 www.nature.com/scientificreports/ Figure 2. Mg(BH4)2 in crystalline γ-modification (as-received (ar), blue circles), and amorphous modification (ball milled (bm), red squares). S(Q, ∆E) measured at λ1 = 5 Å, Q = 1.35 Å−1. (a) T = 310 K, the solid grey curve shows the measured resolution (res) function at 3.5 K. (b) T(ar) = 419 K, T(bm) = 410 K, (c) 310 K, (d) T(ar) = 435 K, T(bm) = 431 K. The solid black curves represent the fit to the data. Figure 2. Mg(BH4)2 in crystalline γ-modification (as-received (ar), blue circles), and amorphous modification (ball milled (bm), red squares). S(Q, ∆E) measured at λ1 = 5 Å, Q = 1.35 Å−1. (a) T = 310 K, the solid grey curve shows the measured resolution (res) function at 3.5 K. (b) T(ar) = 419 K, T(bm) = 410 K, (c) 310 K, (d) T(ar) = 435 K, T(bm) = 431 K. The solid black curves represent the fit to the data. Results and Discussionsh The strong quasi-elastic signal of the ball milled sample decreased while the as-received quasi-elastic signal increased. Simultaneously, the inelastic scattering increased for the ball milled sample and decreased for the as-received sample. p p Figure 3 presents the imaginary (or dissipative) part of the dynamical susceptibility –χ” as a function of energy transfer −∆E [meV] at the same experimental conditions as reported in Fig. 2. The dissipative part of the dynamical susceptibility is connected to the scattering function S(Q, ΔE) through the dissipation - fluctuation theorem by Eq. 1: S Q E n 1 ( , ) (1) B π χ″ = −Δ S Q E n 1 ( , ) B π χ″ = −Δ (1) nB Scientific Reports \| (2020) 10:9080 \| https://doi.org/10.1038/s41598-020-65857-6 www.nature.com/scientificreports/ Figure 3. Dissipative part of the dynamical susceptibility (−χ”) as a function of energy transfer –∆E [meV] and temperatures of (a) 310 K, (b) the temperature was below the ε-phase transition, T < Tε in detail T(ar) = 419 K, T(bm) = 410 K. (c) Followed by a cooling to T=310 K and (d) a second heating procedure above the ε-phase transition temperature, T > Tε in detail T(ar) = 435 K, T(bm) = 431 K. The solid black curves represent the fit to the data. Blue circles: as-received γ-Mg(BH4)2. Red squares: ball milled Mg(BH4)2. Figure 3. Dissipative part of the dynamical susceptibility (−χ”) as a function of energy transfer –∆E [meV] and temperatures of (a) 310 K, (b) the temperature was below the ε-phase transition, T < Tε in detail T(ar) = 419 K, T(bm) = 410 K. (c) Followed by a cooling to T=310 K and (d) a second heating procedure above the ε-phase transition temperature, T > Tε in detail T(ar) = 435 K, T(bm) = 431 K. The solid black curves represent the fit to the data. Blue circles: as-received γ-Mg(BH4)2. Red squares: ball milled Mg(BH4)2. Figure 4. Elastic incoherent scattering function (EISF) and quasi-elastic incoherent scattering function (QISF) as a function of four different temperatures for as-received γ-Mg(BH4)2 and ball milled Mg(BH4)2. (a) 310 K, (b) the temperature was below the ε-phase transition, T < Tε in detail T(ar) = 419 K, T(bm) = 410 K. (c) Followed by a cooling to T=310 K and (d) a second heating procedure above the ε-phase transition temperature, T > Tε in detail T(ar) = 435 K, T(bm) = 431 K. Results and Discussionsh A8a, the dissipative part of the dynamical susceptibility of as-received γ-Mg(BH4)2 at 310 K is com- pared to the one of α-Mg(BH4)2 at 300 K30, illustrating that the external modes in γ-Mg(BH4)2 are even softer, but clearly as rich or even richer in features as α-Mg(BH4)2. The plot highlights the dependency of the excitations on the long range structure of the respective Mg(BH4)2 polymorph. (Note, that the two spectra have been meas- ured at different instrument settings, thus, the decreased half-width at half maximum (HWHM) of γ-Mg(BH4)2 compared to the α-Mg(BH4)2 is partially due to the better energy resolution in the present work). After heating to 460 K, the spectra depicted in Fig. A8b resembles very much the one obtained for β-Mg(BH4)2 from ref. 31, meaning that the β-phase transition is completed as shown in SI Fig. A4.hi The black curves in Fig. 3 are the fitting of χ −′′ and these values were employed to determine the characteris- tic frequencies of the underlying inelastic features. The results are given in Fig. A7 (SI). Remarkably, the values of the characteristic energies ED,i for both samples are fairly constant over the entire temperature range. During the first heating, at 310 K, the amorphous sample exhibits a larger damping γD,i as compared to the as-received one, while no such differences are observed for the second heating procedure. f g p Using ED,i as fixed parameters, S(Q, ΔE) for both samples have been analyzed according to Eq. 2. E Res S S Res D Q A Q E D Q A Q L E S ) [ ( ) ( ) ( ) ( ) ( ) ( , ) ] (2) rot vib inel 0 1 1 1 δ = ⊗ ⊗ = ⊗ Δ + Γ Δ + Res D Q A Q E D Q A Q L E S [ ( ) ( ) ( ) ( ) ( ) ( , ) ] (2) inel 0 1 1 1 δ = ⊗ Δ + Γ Δ + (2) Two Lorentzian functions, L1 and L2 with HWHM Γ1 and Γ2, and the elastic line are needed in addition to the damped harmonic oscillators, DHO (with ED,i (i=1-4)) to describe the data satisfactorily. The results of the fits are shown in Fig. 2 as solid black curves. For both Lorentzians, the HWHM is independent of Q, which indicates localized motions. Results and Discussionsh 3a (ar) has not been identified yet (and it is not subject of this study), the features are resembling rigid unit motions as observed in various modifications of SiO2, which results from small scale rotations of interconnected SiO4 tetrahedra27. Note that the librational frequency of the BH4 tetrahedra is at much higher energies around 65 meV28,29. The existence of rigid unit motions suggests a coupling of adjacent BH4 tetrahedra via a connecting Mg ion, as there is no direct BH4 – BH4 bond in the structure (Fig. 1).   −χ” emphasizes weak features in the inelastic regime, and the analysis of the position and width of the vibrational peaks was used as input for the in depth - analysis of the quasielastic contribution. As there is no Q-dependence in the inelastic spectra, the data have been summed over all Q values to increase statistics. In Fig. 3a as-received γ-Mg(BH4)2 shows distinct contributions at –∆E ~ 3, 4.2, 5.4 and 9.4 meV. In contrast, –χ” of ball milled Mg(BH4)2 has less pronounced features and exhibits an increased intensity at low energy transfer due to the strong quasi-elastic contribution. In Fig. 3b, after first heating, the contributions in both phases become quite similar, suggesting the transformation of the amorphous phase into the γ-phase, which is in good agreement with data presented before (DTA, SR-PXD). In Fig. 3c,d, showing the imaginary part of the dynamical susceptibility –χ” after the first heating and during the second heating and that there are no noticeable differences observed between the two samples. Although the nature of distinct contributions (or modes) in Fig. 3a (ar) has not been identified yet (and it is not subject of this study), the features are resembling rigid unit motions as observed in various modifications of SiO2, which results from small scale rotations of interconnected SiO4 tetrahedra27. Note that the librational frequency of the BH4 tetrahedra is at much higher energies around 65 meV28,29. The existence of rigid unit motions suggests a coupling of adjacent BH4 tetrahedra via a connecting Mg ion, as there is no direct BH4 – BH4 bond in the structure (Fig. 1). g In Fig. Results and Discussionsh For the analysis, L1 (smaller HWHM) was ascribed to the quasi-elastic contribution originat- ing from rotational motions of the BH4 units while L2 was considered part of the inelastic scattering contribution originating from an (additional) overdamped harmonic oscillator. Since the intensity of the second Lorentzian is much weaker compared to L1 its origin cannot be a faster rotation around a different axis, in this case, one would expect the contribution to be more intense than L1. There is only one crystallographic site of the BH4 units in γ-Mg(BH4)2. Nevertheless, characteristic times τ1,2 = ħ/Γ1,2 were obtained from the fit of the two Lorentzian functions as a function of temperature which are shown in Fig. A9. While Fig. A9a shows the results of the first heating cycle for both as-received γ-Mg(BH4)2 and ball milled amorphous-Mg(BH4)2, Fig. A9b depicts the results of the second heating cycle. At 310 K, τ1 and τ2 are lower in γ-Mg(BH4)2 compared to amorphous-Mg(BH4)2. During the first heat, the amorphous sample first shows a decrease of τ1 and τ2 when heating from 310 K to 355 K, indicating that jump rates become higher. Upon further heating and the onset of crystallization, both τ1 and τ2 steadily increase and at 420 K both samples show comparable (slower) jump rates. During the second heating, no such differences are observed and both samples follow the same trend: Up to 420 K, τ1 and τ2 are almost constant and only towards the ε-phase transition at 423 K, a continuous decrease of τ1,2 is observed. The solid line in Fig. A9b illustrates an Arrhenius type fit to τ1 (ascribed to rotational motions), i.e. τ = τ0 exp(Ea/RT), which gives an apparent activation energy in the ε- and β’-phase of Ea = 3.2 ± 0.5 kJ/mol (33.2 ± 5.2 meV) and τ0 = 1.4 ps. In contrast, in γ-Mg(BH4)2 τ1 remained almost constant with increasing temperature, and no activation energy could be determined. This observation is in agreement with NMR data on the γ-phase32, which showed that the apparent activation energies for rotation are much higher in the γ-phase compared to the β-modification. pp g g p pi In general, the temperature dependence of the jump frequency τ1 was used to calculate the activation energy Ea in reported literature as well. Blanchard et al. described three thermally activated processes in the β-Mg(BH4)2 polymorph33. Results and Discussionsh Blue circles: as-received γ-Mg(BH4)2. Red squares: ball milled Mg(BH4)2. EISF top curves, QISF bottom curves. λ = 5 Å. Figure 4. Elastic incoherent scattering function (EISF) and quasi-elastic incoherent scattering function (QISF) as a function of four different temperatures for as-received γ-Mg(BH4)2 and ball milled Mg(BH4)2. (a) 310 K, (b) the temperature was below the ε-phase transition, T < Tε in detail T(ar) = 419 K, T(bm) = 410 K. (c) Followed by a cooling to T=310 K and (d) a second heating procedure above the ε-phase transition temperature, T > Tε in detail T(ar) = 435 K, T(bm) = 431 K. Blue circles: as-received γ-Mg(BH4)2. Red squares: ball milled Mg(BH4)2. EISF top curves, QISF bottom curves. λ = 5 Å. Scientific Reports \| (2020) 10:9080 \| https://doi.org/10.1038/s41598-020-65857-6 www.nature.com/scientificreports/ where nB is the Bose occupation factor, ( ) n exp 1 B E kT 1 =   −   Δ − , and k is the Boltzmann constant26. A plot of where nB is the Bose occupation factor, ( ) n exp 1 B E kT 1 =   −   Δ − , an ( )   −χ” emphasizes weak features in the inelastic regime, and the analysis of the position and width of the vibrational peaks was used as input for the in depth - analysis of the quasielastic contribution. As there is no Q-dependence in the inelastic spectra, the data have been summed over all Q values to increase statistics. In Fig. 3a as-received γ-Mg(BH4)2 shows distinct contributions at –∆E ~ 3, 4.2, 5.4 and 9.4 meV. In contrast, –χ” of ball milled Mg(BH4)2 has less pronounced features and exhibits an increased intensity at low energy transfer due to the strong quasi-elastic contribution. In Fig. 3b, after first heating, the contributions in both phases become quite similar, suggesting the transformation of the amorphous phase into the γ-phase, which is in good agreement with data presented before (DTA, SR-PXD). In Fig. 3c,d, showing the imaginary part of the dynamical susceptibility –χ” after the first heating and during the second heating and that there are no noticeable differences observed between the two samples. Although the nature of distinct contributions (or modes) in Fig. Scientific Reports \| (2020) 10:9080 \| https://doi.org/10.1038/s41598-020-65857-6 Results and Discussionsh The first one with an activation energy of Ea = 39 ± 0.5 meV was confirmed by Silvi et al. who reported Ea = 36 ± 3 meV31. This mode had been assigned to the rotation around the C2\|\| axis, as it is the energeti- cally more favourable rotation of the [BH4] bidentate configuration. Furthermore, in ref. 33, two modes have been reported at Ea = 76 ± 5 meV and Ea = 214 ± 4 meV for β-Mg(BH4)2 with an assignment to the C2\|\| and C3 rotation axis. In ref. 29 the energy of rotation have been reported to peak around 65 meV and have been assigned to rigid liberations of the [BH4]– tetrahedra after ref. 28 and been confirmed as well in ref. 34 with the same trend observed. ti Interestingly, our relaxation times of the amorphous-Mg(BH4)2 can be compared to those of the solid state Mg conductor Mg(BH4)2-diglyme0.5 12, where it was reported that an amorphous component was present. The authors hypothesized that it was amorphous-Mg(BH4)2. It can be observed that the decrease in relaxation times reported here for both τ1 and τ2 up to 360 K can be found in that particular study as well. This means that there is Scientific Reports \| (2020) 10:9080 \| https://doi.org/10.1038/s41598-020-65857-6 www.nature.com/scientificreports/ Figure 5. Arrhenius plot of the magnesium ion conductivity of as-received γ-Mg(BH4)2 and ball milled- Mg(BH4)2. Figure 5. Arrhenius plot of the magnesium ion conductivity of as-received γ-Mg(BH4)2 and ball milled- Mg(BH4)2. evidence of an amorphous Mg(BH4)2 in their sample, although no attempt on heating towards crystallization was reported. If that would have happened, it would most likely have shown a decrease in the temperature dependent characteristic relaxation times.i evidence of an amorphous Mg(BH4)2 in their sample, although no attempt on heating towards crystallization was reported. If that would have happened, it would most likely have shown a decrease in the temperature dependent characteristic relaxation times. From the fit of S(Q, ΔE), the intensities Iel, and Iqe of the elastic (el) and quasi-elastic (qe) contribution to the scattering have been obtained. From these, the Elastic (A0) and Quasi-elastic (A1) Incoherent Structure Factors can be calculated (with A0 = 1 – A1) according to Eq. Results and Discussionsh 3: = = + = + = − = − EISF A I I I D Q A Q D Q A Q D Q A Q QISF A ( ) ( ) ( ) ( ) ( ) ( ) 1 1 (3) el el qe 0 0 0 1 1 (3) The results are shown in Fig. 4. As aforementioned, there is a marked difference between amorphous Mg(BH4)2 and γ-Mg(BH4)2 at 310 K. The latter shows almost no quasi-elastic intensity, and accordingly, the EISF is close to one. Once the amorphous sample is crystallized, the observed structure factors are very similar for both samples. The EISF and QISF have been analyzed assuming hindered rotations around the C2 or C3 axis of the BH4 tetrahedra, which can be modelled as described in Eq. 4: = + −    +          − EISF p p j Q d (1 )1 2 1 2 2 3 (4) B H 0 (4) The solid lines in Fig. 4 are fits according to Eq. 4 with the free parameter p, the fraction of hindered rotations, and dB–H, the boron–hydrogen distance. Values obtained for dB–H vary between 1.10–1.28 Å, thus, well within the expected range for the BH4 tetrahedra19 (Note, that the sensitivity of the fit on dB–H is limited). For crystalline Mg(BH4)2, the fraction of hindered rotations p exhibit a continuous increase of activated BH4 rotations until all C2/C3 modes are active (i.e. p = 0) at a temperature that coincides with the ε-phase transition temperature. In contrast, amorphous Mg(BH4)2 initially exhibits a larger rotational activity at 310 K, however the number of activated rotations decreases upon heating towards crystallization (see Fig. A10). One might speculate that the larger number of terminating Mg-H bonds in amorphous Mg(BH4)2 (effectively interrupting the interpenetrating channels and their ..-Mg-H2BH2-Mg-.. chains) favors stochastic rotations of the [BH4] units. After the sample has crystallized into the γ-phase, the dynamic properties are almost identical for both samples. The mean square displacement (<u2>) of the hydrogen atoms has also been obtained from the analysis: γ-Mg(BH4)2 shows an continuous increase in <u2> with temperature while <u2> of the ball milled phase (at 310 K)) is larger, and remains almost constant upon heating to 380 K (crystallization). Both samples appear to be quite similar in the temperature range 380–455 K (Fig. Results and Discussionsh A11).h g g The temperature dependence of the magnesium ion conductivity for γ-Mg(BH4)2 and ball milled-Mg(BH4)2 was determined by electrochemical impedance spectroscopy (EIS) and is presented in Fig. 5. γ-Mg(BH4)2 shows the conductivity of 5.3·10−14 S·cm−1 at 313 K which increases to 6.9·10−13 S·cm−1 and 1.2·10−11 S·cm−1 at 353 K and 393 K, respectively, with the activation energy of 0.68 eV. Ball milled-Mg(BH4)2 shows a higher conductiv- ity of 2.5·10−13 S·cm−1 already at 313 K and 1.8·10−11 S·cm−1 at 353 K with an activation energy of 0.95 eV. The conductivity values of the ball milled sample are higher in the entire temperature range. At 373 K, a drop in the conductivity is observed which corresponds to the aforementioned crystallization to γ-Mg(BH4)2 also shown in PXD data depicted in Fig. A12. However, our measured conductivity is low compared to earlier work were a conductivity of <10−12 S·cm−1 at 303 K was reported for pristine Mg(BH4)2 11. Scientific Reports \| (2020) 10:9080 \| https://doi.org/10.1038/s41598-020-65857-6 www.nature.com/scientificreports/ The exploration of dynamics of the host lattice goes back to the 1970s, where it was suggested that the ionic conductivity, which is a thermally activated process35, is proportional to the phonon spectrum36. The latter is caused by lattice vibrations which correlate to the elastic stiffness of the lattice. In Kraft et al. the authors suggest two contradicting processes for an increase in conductivity in lithium argyrodites based materials37. The first is the lowering in activation energy caused by a softer lattice while simultaneously a softer lattice decreases the attempt frequency of the ion jump as well as the migration entropy37. A softer lattice is thus not necessarily better for a solid ionic conductor37. In our work, we show that the number of rotating [BH4] units in Mg(BH4)2 increases by mechanochemical treatment and that ionic conductivity is almost 2 orders of magnitudes higher at 353 K. y y g g The increase of the number of rotating [BH4] units is, what we believe, direct evidence of the so-called “paddle-wheel” mechanism38, which most recently has also been shown by QENS measurements in carba-borohydrides39,40. This “paddle-wheel” mechanism has been suggested by computational results for the complex [BH4] anion41. Therefore, it was suggested to be aiding the conduction process in complex metal hydrides42–44. Experimentally, solid-state Nuclear Magnetic Resonance and QENS measurements45,46 indicated that the high rotational mobility promotes super-conductivity in these materials47,48. Results and Discussionsh Thus, the observed cor- relation between rotational mobility and conductivity lead to effects which are possibly flattening the energy landscape by dynamic frustration49.h p y y The activation energy increased for the ball milled sample, which is contra intuitive, leaving us puzzled what the dominating process of the ion conduction process is. A decrease of the attempt frequency of the ion jump and the migration entropy will be investigated further in future. One might speculate, that the former is correlated to the local [BH4] rotations. During heating, the number of activated rotations in amorphous Mg(BH4)2 slowly decreases already at temperatures well below the crystallization temperature, thus potentially counteracting the thermal activation of the ion jumps (compare Fig. A10). As a remark, in our work we do change the local atomic structure possibly by misplacing Mg ions slightly (as shown in comparison of model 3 in Fig. A1 and Fig. 1d), therefore it is hard to judge if the increased conductivity stems from the change in structure or from the softer lattice. Probably a combination of both, while we are not changing any anions, therefore not changing the charge carriers. Conclusions Quasi-elastic neutron scattering (QENS) studies were employed to investigate the dynamics of porous (γ-) and amorphous Mg(BH4)2. The corresponding PDFs show that the local structure of the amorphous sample agrees reasonable well with the crystalline one up to ~5.1 Å, meaning that main building blocks of the structure remain Mg – BH4 – Mg units. Above 5.1 Å, a slight oscillation is still observable up to ~12.3 Å, which is in good agree- ment with the diameter of a one dimensional channel and thus indicates that the fundamental structure of the amorphous sample is still formed by these channels, even though less well-ordered. QENS studies found a corre- lation in the relaxation times of the ball milled (amorphous) Mg(BH4)2 and Mg(BH4)2-diglyme0.5. Additionally, for Mg(BH4)2-diglyme0.5, a step-like increase of the mean square displacement has been reported. In contrast, the mean square displacement in our study was found to be completely linear over a broad temperature spectrum12. q p y p y p p The mechanochemical synthesis method of the recently reported solid-state Mg-ion conductors, Mg(en) (BH4)2 and Mg(BH4)2-diglyme0.5, tends to form amorphous Mg(BH4)2 as a byproduct11,12. Its influence on the conduction properties is unknown, but it was postulated that the amorphous phase is helping to increase the con- ductivity. Electrochemical impedance spectroscopy for ionic conductivities measurements were employed here and it was found that the conductivity of the amorphous phase is indeed ~2 orders of magnitude higher than the as-received γ-Mg(BH4)2 at 353 K. PDF analysis found similar local building blocks, thus suggesting also similar conduction pathways. QENS data showed a higher fraction of activated rotations in the amorphous sample. Thus it is postulated that the conduction process in amorphous Mg(BH4)2 is supported by rotating [BH4] units. Upon crystallization at 373 K, the number of rotations decreases as well as conductivity values.hi y y This study confirms that the amorphous phase of Mg(BH4)2 has an important contribution in future Mg conductors and therefore its presence needs to be taken into account for follow-up investigations. Experimental Methods Synchrotron radiation powder X-ray diffraction (SR-PXD). data were collected at DESY at beam- line P02.153. For in situ SR-PXD experiments the samples were contained in 0.8 mm sealed sapphire capillaries. The sample-to-detector distances and the wavelength were calibrated from a NIST silicon standard. Data were collected using a Perkin Elmer XRD1621 area detector. The exposure time was set to 10 s giving a temperature resolution of 0.83 K per pattern. The data were integrated to 1D diffraction patterns in DAWN 254. Synchrotron X-ray total scattering. experiments were executed at room temperature. Data were col- lected at the high energy powder diffraction and total scattering beamline P02.1 with photon energies of 60 keV (λ = 0.20723 Å)53. Sample-detector-distance was 220 mm. The scattering data was acquired using a Perkin Elmer XRD1621 (200 × 200 µm2 pixels) area detector. The total exposure time was 30 minutes. The integration of 2D pattern was performed using DAWNscience54. An empty 1 mm glass (Hilgenberg Glass no. 10) capillary was measured under the same conditions and subtracted from the measured data. To account for instrumental contri- bution silicon standard (NIST 640a) was measured. The corresponding pair distribution function was calculated using the program PDFgetX3 with a Qmax = 23.8 Å−155. In future, neutron powder diffraction measurements are planned especially within the frame of the “Energy research with Neutrons (ErwiN)” instrument at the MLZ, Germany21. Attenuated total reflection IR (ATR-IR). measurements were performed using a -Agilent Technologies Cary 630 infrared spectrometer with a diamond crystal inside an Ar-filled glove box. The spectra were obtained in the wavenumber range of 4000-650 cm–1 with a resolution of 4 cm−1 at RT. 300 scans were averaged for each spectrum and the background. IR spectra were ATR corrected using commercial spectroscopic software OPUS. The samples were measured without any dilution. Mg ionic conductivity measurements. were measured by electrochemical impedance spectroscopy (EIS) using a Novocontrol potentiostat with a voltage amplitude of 50 mV in the frequency range from 1 MHz to 0.01 Hz. Powder samples (150, 112 mg) were compressed in pellets with thicknesses of 1.21 and 1.34 mm and diameters of 15.5 and 12.5 mm using an axial hydraulic press with a pressure of 282 and 80 MPa for γ-Mg(BH4)2 and ball milled Mg(BH4)2 respectively. The pellets were sandwiched between indium foils to improve the contact between the sample and the brass electrodes and were mounted in an airtight cell (BDS1308). Experimental Methods p Sample preparation. Mg(BH4)2 powder was purchased from Sigma–Aldrich in γ–modification (>95%) (space group Id–3a19) and was used as-received (ar). Amorphization of the Mg(BH4)2 was achieved by ball milled (bm) in a P6 planetary ball mill for a total of 1 h, divided in 4 times 15 min and 5 min breaks in between to avoid overheating50. Stainless steel vials with stainless steel–balls were used with a ball-to-powder ratio 40:1. All sample manipulations were performed under Argon atmosphere in an MBraun Unilab glove box (O2/H2O sensors were kept under 1ppm). Quasi-elastic neutron scattering (QENS). were conducted at the time-of-flight spectrometer TOFTOF operated by the Technische Universität München at the Heinz Maier-Leibnitz Zentrum (MLZ) in Garching, Germany. TOFTOF has a direct geometry and employs cold neutrons51. The raw neutron data were normal- ized to the incoming flux and vanadium, corrected for background and self-shielding absorption effects. The time-of-flight data were transformed to energy transfer, and the momentum transfer Q was calculated. The obtained dynamic structure factor S(Q, ΔE) was binned into a regular grid in energy transfer (ΔE) and momen- tum transfer (ΔQ). Measurements were taken at λ = 5 Å incident wavelength, which gave an instrument resolu- tion at the elastic line of 0.065 meV (FWHM = full width half maximum) and an accessible momentum transfer range of Q = 0.3–2.2 Å−1. All measurements were performed in transmission mode with the sample containers oriented at an angle of 135° with respect to the incoming beam. Measurements were performed in a cryofurnace at discrete temperatures of 3.5 K (resolution), and between 310 K and 460 K. Scientific Reports \| (2020) 10:9080 \| https://doi.org/10.1038/s41598-020-65857-6 www.nature.com/scientificreports/ ~100 mg of material was employed for each measurement with an estimated neutron transmission of ~85% (using samples with natural boron). A broad range of temperatures was chosen with two motivations: First to observe the exothermic phase transition from amorphous to the crystalline γ-modification, secondly to observe the endothermic phase transitions during heating above 423 K (ε- and β′-Mg(BH4)2 structures24,52). Each temper- ature was measured for 3–5 h, with additional 30 minutes of equilibration time. Thermogravimetric and differential thermal analysis (TG-DTA). experiments were conducted using a Netzsch STA 409 C/CD analyzer. The experiments were conducted from room temperature (RT) to 563 K at 5 K min–1. All samples were measured within Al2O3 crucibles. The Ar flow (protective and purge gas) was 20 and 50 ml min–1, respectively. References 1. Zhao-Karger, Z. & Fichtner, M. Beyond Intercalation Chemistry for Rechargeable Mg Batteries: A Short Review and Perspective Front. Chem. 6, 656, https://doi.org/10.3389/fchem.2018.00656 (2018).h p g 2. Payandeh, S., Remhof, A. & Battaglia, C. In Magnesium Batteries: Research and Applications 60–78 (The Royal Society of Chemistry 2020).f 3. Gregory, T. D., Hoffman, R. J. & Winterton, R. C. Nonaqueous Electrochemistry of Magnesium: Applications to Energy Storage. J Electrochem. Soc. 137, 775–780, https://doi.org/10.1149/1.2086553 (1990).h p g 4. Aurbach, D., Cohen, Y. & Moshkovich, M. The study of reversible magnesium deposition by in situ scanning tunneling microscopy Electrochem. Solid-State Lett. 4, A113–A116 (2001). ( ) 5. Davidson, R. et al. Formation of magnesium dendrites during electrodeposition. ACS Energy Lett. 4, 375–376 (2018). 6. Jäckle, M. & Groß, A. Microscopic properties of lithium, sodium, and magnesium battery anode materials related to possible dendrite growth. J. Chem. Phys. 141, 174710 (2014).h 7. Mohtadi, R. & Orimo, S.-I. The renaissance of hydrides as energy materials. Nat. Rev. Mater. 2, 16091, https://doi.org/10.1038 natrevmats.2016.91 (2016).fi 8. Hadjixenophontos, E. et al. A Review of the MSCA ITN ECOSTORE—Novel Complex Metal Hydrides for Efficient and Compac S f R bl E H d d El h //d / / ( ) 8. Hadjixenophontos, E. et al. A Review of the MSCA ITN ECOSTORE—Novel Complex Metal Hydrides for Efficient and Compact Storage of Renewable Energy as Hydrogen and Electricity. Inorganics 8, 17, https://doi.org/10.3390/inorganics8030017 (2020). 8. Hadjixenophontos, E. et al. A Review of the MSCA ITN ECOSTORE Novel Complex Metal Hydrides for Efficient and Compact Storage of Renewable Energy as Hydrogen and Electricity. Inorganics 8, 17, https://doi.org/10.3390/inorganics8030017 (2020). Storage of Renewable Energy as Hydrogen and Electricity. Inorganics 8, 17, https://doi.org/10.3390/inorganics8030017 (2020). 9. Mohtadi, R., Matsui, M., Arthur, T. S. & Hwang, S.-J. Magnesium Borohydride: From Hydrogen Storage to Magnesium Battery A Ch d h //d / / ( ) g gy y g y g g g 9. Mohtadi, R., Matsui, M., Arthur, T. S. & Hwang, S.-J. Magnesium Borohydride: From Hydrogen Storage to Magnesium Battery Angew. Chem. Int. Ed. 51, 9780–9783, https://doi.org/10.1002/anie.201204913 (2012).fi 10. Zhao-Karger, Z., Bardaji, M. E. G., Fuhr, O. & Fichtner, M. A new class of non-corrosive, highly efficient electrolytes for rechargeable magnesium batteries. J. Mater. Chem. A 5, 10815–10820 (2017). g 11. Roedern, E., Kuhnel, R. S., Remhof, A. & Battaglia, C. Magnesium Ethylenediamine Borohydride as Solid-State Electrolyte for Magnesium Batteries. Sci. Rep. Scientific Reports \| (2020) 10:9080 \| https://doi.org/10.1038/s41598-020-65857-6 Experimental Methods The samples were measured from 20 to 120 °C. A R1/Q1 model was used to fit the conductivity data (Fig. A13). The relative densi- ties of the pellets are presented in Tab. S1. Received: 17 January 2020; Accepted: 12 May 2020; Published: xx xx xxxx References 7, 46189, https://doi.org/10.1038/srep46189 (2017). g p p g p 12. Burankova, T. et al. Dynamics of the Coordination Complexes in a Solid-State Mg Electrolyte. J. Phys. Chem. Lett. 9, 6450– https://doi.org/10.1021/acs.jpclett.8b02965 (2018). p g jp 13. Liu, Z. et al. Anomalous high ionic conductivity of nanoporous β-Li3PS4. J. Am. Chem. Soc. 135, 975–978 (2013). 14. Mizuno, F., Hayashi, A., Tadanaga, K. & Tatsumisago, M. High lithium ion conducting glass-ceramics in the system Li2S–P2S5. Solid. State Ion. 177, 2721–2725 (2006). p g jp 13. Liu, Z. et al. Anomalous high ionic conductivity of nanoporous g j 13. Liu, Z. et al. Anomalous high ionic conductivity of nanoporous β-Li3PS4. J. Am. Chem. Soc. 135, 975–978 (2013). 14 Mi F H hi A T d K & T t i M Hi h lithi i d ti l i i th t Li2S P2S5 9 Scientific Reports \| (2020) 10:9080 \| https://doi.org/10.1038/s41598-020-65857-6 www.nature.com/scientificreports/ 5. Higashi, S., Miwa, K., Aoki, M. & Takechi, K. A novel inorganic solid state ion conductor for rechargeable Mg batteries. Chem Commun. 50, 1320–1322, https://doi.org/10.1039/C3CC47097K (2014). p g 16. Le Ruyet, R. et al. Investigation of Mg(BH4)(NH2)-Based Composite Materials with Enhanced Mg2+ Ionic Conductivity. J. Chem. C. 123, 10756–10763, https://doi.org/10.1021/acs.jpcc.9b00616 (2019). p g jp ( ) 7. Filippov, S. et al. Analysis of Dihydrogen Bonding in Ammonium Borohydride. J. Phys. Chem. C. 123, 28631–28639, https://doi org/10.1021/acs.jpcc.9b08968 (2019).h g jp 8. Yan, Y. et al. The mechanism of Mg2+ conduction in ammine magnesium borohydride promoted by a neutral molecule. Phys Chem. Chem. Phys.https://doi.org/10.1039/D0CP00158A (2020). y p g ( ) 19. Filinchuk, Y. et al. Porous and dense magnesium borohydride frameworks: synthesis, stability, and reversible absorption o y p g 19. Filinchuk, Y. et al. Porous and dense magnesium borohydride frameworks: synthesis, stability, and reversible absorption of guest species Angew Chem Int Ed Engl 50 11162 11166 https://doi org/10 1002/anie 201100675 (2011) p g g p g 20. Billinge, S. J. L. In Local Structure from Diffraction (eds. S. J. L. Billinge & M. F. Thorpe) 137–156 (Springer US, 2002).f fh 21. Heere, M. et al. Energy research with neutrons (ErwiN) and installation of a fast neutron powder diffraction option at Germany. Journal of Applied Crystallography 51, https://doi.org/10.1107/S1600576718004223 (2018).f 2. Heere, M. et al. Effect of additives, ball milling and isotopic exchange in porous magnesium borohydride. Rsc Adv. 8, 27645–27653 https://doi.org/10.1039/c8ra05146a (2018). 23. Zavorotynska, O., El-Kharbachi, A., Deledda, S. References & Hauback, B. C. Recent progress in magnesium borohydride Mg(BH4)(2): Fundamentals and applications for energy storage. Int. J. Hydrog. Energy 41, 14387–14403, https://doi.org/10.1016/j. ijhydene.2016.02.015 (2016). j y ( ) 4. David, W. I. F. et al. The structure, thermal properties and phase transformations of the cubic polymorph of magnesium tetrahydroborate. Phys. Chem. Chem. Phys. 14, 11800–11807, https://doi.org/10.1039/C2CP23439D (2012). h h d f h d d d d b h h 5. Lohstroh, W. & Heere, M. Structure and Dynamics of Borohydrides Studied by Neutron Scattering Techniques: A Review. J. Phys Soc. Jpn. 89, 1–12, https://doi.org/10.7566/JPSJ.89.051011 (2020). g 6. Schober, H. An introduction to the theory of nuclear neutron scattering in condensed matter. J. Neutron. Res. 17, 109–357, https:/ doi.org/10.3233/JNR-140016 (2014). g 7. Wells, S. A., Dove, M. T., Tucker, M. G. & Trachenko, K. Real-space rigid-unit-mode analysis of dynamic disorder in quartz cristobalite and amorphous silica. J. Physics: Condens. Matter 14, 4645 (2002).t p y 8. Allis, D. G. & Hudson, B. S. Inelastic neutron scattering spectra of NaBH4 and KBH4: reproduction of anion mode shifts via periodic DFT. Chem. Phys. Lett. 385, 166–172 (2004). 29. Giannasi, A. et al. High resolution Raman and neutron investigation of Mg(BH4)2 in an extensive temperature range. J. Phys. Chem. A 114, 2788–2793 (2010). 0. Silvi, L. et al. A quasielastic and inelastic neutron scattering study of the alkaline and alkaline-earth borohydrides LiBH4 and Mg(BH4)2 and the mixture LiBH4+ Mg(BH4). Phys. Chem. Chem. Phys. 21(2), 718–728 (2019). 31. Silvi, L., Rohm, E., Fichtner, M., Petry, W. & Lohstroh, W. Hydrogen dynamics in [small beta]-Mg(BH4)2 on the picosecond timescale. Phys. Chem. Chem. Phys. 18, 14323–14332, https://doi.org/10.1039/C6CP00995F (2016). 32. Soloninin, A. V. et al. NMR Study of Reorientational Motion in Alkaline-Earth Borohydrides: β and γ Phases of Mg (BH4) 2 and α and β Phases of Ca (BH4) 2. J. Phys. Chem. C. 116, 4913–4920 (2012). y 3. Blanchard, D. et al. Hindered Rotational Energy Barriers of BH4–Tetrahedra in β-Mg(BH4)2 from Quasielastic Neutron Scattering and DFT Calculations. J. Phys. Chem. C. 116, 2013–2023 (2012). 33. Blanchard, D. et al. Hindered Rotational Energy Barriers of BH4–Tetrahedra in β-Mg(BH4)2 from and DFT Calculations. J. Phys. Chem. C. 116, 2013–2023 (2012). y 4. Hagemann, H. et al. New fundamental experimental studies on α-Mg(BH4)2 and other borohydrides. J. Alloy. Compd. 509 S688–S690 (2011). 35. Bührer, W., Nicklow, R. M. & Brüesch, P. Lattice dynamics of $\ensuremath{\beta}$-(silver iodide) by neutron scattering. References Phys B 17, 3362–3370, https://doi.org/10.1103/PhysRevB.17.3362 (1978). 6. Zeller, H. R., Brüesch, P., Pietronero, L. & Strässler, S. In Superionic Conductors (eds Gerald D. Mahan & Walter L. Roth) 201–215 (Springer US, 1976).tl ( p g ) 37. Kraft, M. A. et al. Influence of Lattice Polarizability on the Ionic Conductivity in the Lithium Superionic Argyrodites Li6PS5X (X = Cl, Br, I). J. Am. Chem. Soc. 139, 10909–10918, https://doi.org/10.1021/jacs.7b06327 (2017).f 38. Jansen, M. Volume Effect or Paddle-Wheel Mechanism—Fast Alkali‐Metal Ionic Co Complex Anions. Angew. Chem. Int. Ed. Engl. 30, 1547–1558 (1991). 8. Jansen, M. Volume Effect or Paddle-Wheel Mechanism—Fast Alkali‐Metal Ionic Conduction in Solids with Rotationally Disordered Complex Anions. Angew. Chem. Int. Ed. Engl. 30, 1547–1558 (1991). 9. Tang, W. S. et al. Unparalleled lithium and sodium superionic conduction in solid electrolytes with large monovalent cage-like anions. Energy Environ. Sci. 8, 3637–3645 (2015). gy 0. Dimitrievska, M. et al. Carbon incorporation and anion dynamics as synergistic drivers for ultrafast diffusion in superionic LiCB11H12 and NaCB11H12. Adv. Energy Mater. 8, 1703422 (2018). gy 1. Fang, H. & Jena, P. Li-rich antiperovskite superionic conductors based on cluster ions. Proc. Natl Acad. Sci. 114, 11046–11051 (2017). ( ) 42. Martelli, P. et al. Rotational motion in LiBH4/LiI solid solutions. J. Phys. Chem. A 115, 5329–5334 (2011). y 43. Verdal, N., Udovic, T. J., Rush, J. J., Wu, H. & Skripov, A. V. Evolution of the reorientational motions of the tetrahydroborate in hexagonal LiBH4–LiI solid solution by high-Q quasielastic neutron scattering. J. Phys. Chem. C. 117, 12010–12018 (2013) 44. Skripov, A. V., Soloninin, A. V., Ley, M. B., Jensen, T. R. & Filinchuk, Y. Nuclear Magnetic Resonance Studies of BH4 Reorientations and Li Diffusion in LiLa(BH4)3Cl. J. Phys. Chem. C. 117, 14965–14972, https://doi.org/10.1021/jp403746m (2013). f 45. Skripov, A. V. et al. Nuclear Magnetic Resonance Study of Atomic Motion in A2B12H12 (A = Na, K, Rb, Cs): Anion Reorient and Na+ Mobility. J. Phys. Chem. C. 117, 25961–25968, https://doi.org/10.1021/jp4106585 (2013). 6. Verdal, N. et al. Anion Reorientations in the Superionic Conducting Phase of Na2B12H12. J. Phys. Chem. C. 118, 17483–17489 https://doi.org/10.1021/jp506252c (2014). 7. Famprikis, T., Canepa, P., Dawson, J. A., Islam, M. S. & Masquelier, C. Fundamentals of inorganic solid-state electrolytes fo batteries. Nature materials, 1–14 (2019). 8. Paskevicius, M. et al. Metal borohydrides and derivatives - synthesis, structure and properties. Chem. Soc. Rev. 46, 1565–1634 https://doi.org/10.1039/c6cs00705h (2017). g 9. Kweon, K. E. et al. References Structural, Chemical, and Dynamical Frustration: Origins of Superionic Conductivity in closo-Borate Solid Electrolytes. Chem. Mat. 29, 9142–9153, https://doi.org/10.1021/acs.chemmater.7b02902 (2017). y p g 50. Huot, J. et al. Mechanochemistry of metal hydrides: recent advances. (2019).l t al. Mechanochemistry of metal hydrides: recent advances. (2019). y y 51. Lohstroh, W. & Evenson, Z. TOFTOF: Cold neutron time-of-flight spectrometer. J. large-scale Res. facilities JLSRF 1, 15 (2015) k l ff d f ( ) h h h 51. Lohstroh, W. & Evenson, Z. TOFTOF: Cold neutron time-of-flight spectrometer 51. Lohstroh, W. & Evenson, Z. TOFTOF: Cold neutron time-of-flight spectrometer. J. large-scale Res. facilities JLSRF 1, 15 (2015). 52. Paskevicius, M. et al. In-Situ X-ray Diffraction Study of γ-Mg(BH4)2 Decomposition. J. Phys. Chem. C. 116, 15231–15240, https:// doi.org/10.1021/jp302898k (2012).f g jp 3. Dippel, A.-C. et al. Beamline P02. 1 at PETRA III for high-resolution and high-energy powder diffraction. J. synchrotron Radiat. 22 675–687 (2015).f 4. Filik, J. et al. Processing two-dimensional X-ray diffraction and small-angle scattering data in DAWN 2. J. Appl. Crystallography 50 959–966, https://doi.org/10.1107/S1600576717004708 (2017). g 55. Juhás, P., Davis, T., Farrow, C. L. & Billinge, S. J. PDFgetX3: a rapid and highly automatable program for processing powder diffraction data into total scattering pair distribution functions. J. Appl. Crystallography 46, 560–566 (2013). Scientific Reports \| (2020) 10:9080 \| https://doi.org/10.1038/s41598-020-65857-6 10 www.nature.com/scientificreports/ Acknowledgementsh g This work contributes to the research performed at CELEST (Center for Electrochemical Energy Storage Ulm- Karlsruhe) and was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany´s Excellence Strategy – EXC-2154 – 390874152. We acknowledge DESY (Hamburg, Germany), a member of the Helmholtz Association HGF, for the provision of experimental facilities. Parts of this research were carried out at PETRA III and we would like to thank Jo–Chi Tseng and Martin Etter for assistance in using beamline P02.1. M. Heere acknowledges the funding from the project 05K16VK2/ 05K19VK3 “Energy research with Neutrons (ErwiN)” by the German Federal Ministry of Education and Research (BMBF). S. Payandeh acknowledges the EMPAPOSTDOCS-II program funding from the European Union’s Horizon 2020 research and innovation program under the Marie Skłodowska-Curie grant agreement number 754364. Competing interestsh p g The authors declare no competing interests. Author contributions Conceptualization, writing—original draft preparation, Michael Heere & Wiebke Lohstroh; writing—review and editing Michael Heere, Anna-Lena Hansen, SeyedHosein Payandeh, Neslihan Aslan, Gökhan Gizer, Magnus H. Sørby, Bjørn C. Hauback, Claudio Pistidda, Martin Dornheim, Wiebke Lohstroh; sample provision, synthesis and analysis shown in Figure 1a: Michael Heere, Magnus H. Sørby, Bjørn C. Hauback; performing QENS experiments and analysis shown in Figures 2–4: Michael Heere, Neslihan Aslan and Wiebke Lohstroh; performing PDF and in situ SR-PXD experiments and analysis: Anna-Lena Hansen (Figure 1b–d), Neslihan Aslan, Gökhan Gizer, Claudio Pistidda, Martin Dornheim; performing EIS experiments, TG-DTA and ATR-IR and analysis: SeyedHosein Payandeh (Figure 5) and Neslihan Aslan. Additional information Supplementary information is available for this paper at https://doi.org/10.1038/s41598-020-65857-6. Supplementary information is available for this paper at https://doi.org/10.10 Correspondence and requests for materials should be addressed to M.H. or W.L. Reprints and permissions information is available at www.nature.com/reprints. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Cre- ative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not per- mitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. © The Author(s) 2020 Scientific Reports \| (2020) 10:9080 \| https://doi.org/10.1038/s41598-020-65857-6
https://openalex.org/W4317353370	https://zenodo.org/records/7545127/files/1-1%20PAVOS%20BIG6%20ALIMENTADOS%20CON%20BALANCEADO.pdf	es		PAVOS BIG6 ALIMENTADOS CON BALANCEADO UTEQ + NABO EN FASE ENGORDE PARA MEJORAR LAS CARACTERÍSTICAS ORGANOLÉPTICAS DE LA CARNE	Zenodo (CERN European Organization for Nuclear Research)	2,022	cc-by	2,490	JOURNAL OF SCIENCE AND RESEARCH E-ISSN: 2528-8083 PAVOS BIG6 ALIMENTADOS CON BALANCEADO UTEQ + NABO EN FASE ENGORDE PARA MEJORAR LAS CARACTERÍSTICAS ORGANOLÉPTICAS DE LA CARNE BIG6 TURKEYS FED WITH BALANCED UTEQ + TURNIP IN THE FATTENING PHASE TO IMPROVE THE ORGANOLEPTIC CHARACTERISTICS OF THE MEAT DOI….. AUTORES: Piedad Yepez Macias1* Wellington Arévalo Briones2 Luis Vásquez Cortez3 Kerly Alvarado Vásquez4 DIRECCIÓN PARA CORRESPONDENCIA: pyepez@uteq.edu.ec* Fecha de recepción: 01 / 09 / 2022 Fecha de aceptación: 21 / 11 / 2022 RESUMEN. La crianza de pavos constituye un importante apoyo económico-alimenticio para las poblaciones rurales y suburbanas, así como un recurso genético pecuario propio de las granjas avícolas, el interés económico de la explotación actual del pavo se apoya en su enorme rendimiento en carne, su bajo costo de producción, y la calidad nutritiva por el bajo 1 * Ingeniera Zootecnista, Magister en Producción Animal, Universidad Técnica Estatal de Quevedo, pyepez@uteq.edu.ec. 2 Ingeniero Zootecnista, Universidad Técnica Estatal de Quevedo, cv0505542@gmail.com 3 Ingeniero en Alimentos, Maestrante de Agroindustria, Instituto de Posgrados Universidad Técnica de Manabí, lvasquez7265@utm.edu.ec 4 Ingeniera en Alimentos, Maestrante de Agroindustria, Instituto de Posgrados Universidad Técnica de Manabí, kalvarado6940@utm.edu.ec Vol. 7, N°. CININGEC II - 2022 JOURNAL OF SCIENCE AND RESEARCH E-ISSN: 2528-8083 contenido de grasa de su carne, la presente investigación se llevó a cabo en los predios de la Finca Experimental “La María”, perteneciente a la Universidad Técnica Estatal de Quevedo (UTEQ), ubicada en el km 7 ½ Vía Quevedo- El Empalme, Cantón Mocache. Se realizó un ensayo de tipo experimental para conocer el comportamiento productivo de pavos americanos BIG-6 (consumo de alimento: CA; ganancia de peso: GP; conversión alimenticia: CAL y el rendimiento a la canal: RC), alimentados con una dieta balanceada y suplementados con nabo (Brassica rapa L.), además las características organolépticas y la rentabilidad. Los tratamientos evaluados fueron: T1 (testigo). T2 (balanceado + 10,0% de nabo) y T3 (balanceado + 15,0% de nabo). Se aplicó un diseño completamente al azar (DCA) con cinco repeticiones. La unidad experimental estuvo conformada por cuatro pavos. Para determinar las diferencias entre medias se aplicó la prueba de Tukey (P≤0.05) y para el análisis económico la Relación Beneficio Costo. Para el análisis de los datos se empleó el paquete estadístico S.A.S. El mayor (P<0.05) CA, GP, CAL y la mayor rentabilidad lo registró el tratamiento T3 (160,75 g animal-1 día-1; 78,51 g animal-1 día- 1, 2.0 y 87,87%, respectivamente). Además, demostró las mejores características organolépticas (textura, sabor, color olor y jugosidad). Palabras clave: Pavos Americanos, Dieta, Degustar. ABSTRACT. Turkey breeding is an important economic and nutritional support for rural and suburban populations, as well as a livestock genetic resource for poultry farms, the economic interest of the current exploitation of turkeys is based on its enormous meat yield, its low production cost, This research was carried out on the premises of the "La María" Experimental Farm, which belongs to the Quevedo State Technical University (UTEQ), located at km 7 ½ of the Quevedo-El Empalme road, Cantón Mocache. An experimental trial was carried out to determine the productive behavior of American BIG-6 turkeys (feed consumption: CA; weight gain: GP; feed conversion: CAL and carcass yield: RC), fed a balanced diet and supplemented with turnip (Brassica rapa L.), as well as the organoleptic characteristics and profitability. The treatments evaluated were: T1 (control). T2 (balanced + 10,0% turnip) and T3 (balanced + 15,0% turnip). A completely randomized design (CRD) with five replications was applied. The experimental unit consisted of four turkeys. Tukey's test (P≤0.05) was used Vol. 7, N°. CININGEC II - 2022 JOURNAL OF SCIENCE AND RESEARCH E-ISSN: 2528-8083 to determine the differences between means and the Benefit-Cost Ratio was used for the economic analysis. The statistical package S.A.S. was used for data analysis. The highest (P<0.05) CA, GP, CAL and the highest profitability were recorded by treatment T3 (160.75 g animal-1 day-1; 78.51 g animal-1 day-1, 2.0 and 87,87%, respectively). In addition, it showed the best organoleptic characteristics (texture, flavor, color, odor and juiciness). Translated with www.DeepL.com/Translator (free version) Keywords: American turkeys, Diet, Tasting. INTRODUCCIÓN. El Codex Alimentarius define la carne como “todas las partes de un animal que han sido dictaminadas como inocuas y aptas para el consumo humano o se destinan para este fin. Sin embargo, normalmente se denomina carne al músculo esquelético de los animales de sangre caliente, producidos principalmente por las técnicas ganaderas modernas y en parte por la caza (Gallo & Tadich, 2008). El presente trabajo pretende evaluar las características organolépticas de la carne de pavo americano BIG-6 con dieta alimenticia (balanceado + nabo (Brassica rapa L.)) para determinar la calidad de la carne y el grado de aceptación por parte de los consumidores, como una alternativa que permitirá mejorar la producción de carne aviar (pavos) inclusive como un recurso para dar un valor agregado a la avicultura (Sasongko, 2015). Objetivos 1. Evaluar las características organolépticas de la carne de pavo Americano BIG-6 con dieta alimenticia (Balanceado + Nabo (Brassica rapa L.)). 2. Examinar la incidencia de la dieta alimenticia de balanceado + nabo (Brassica rapa L.) en pavos Americanos BIG-6 en la obtención de una carne de calidad. 3. Identificar la característica sensorial con mayor influencia sobre la aceptación de la carne de pavo por parte del consumidor. Comprobar los resultados mediante el método estadístico Chi Cuadrado. Vol. 7, N°. CININGEC II - 2022 JOURNAL OF SCIENCE AND RESEARCH E-ISSN: 2528-8083 METODOLOGÍA La presente investigación se realizó en la Finca Experimental “La María”, propiedad de la Universidad Técnica Estatal de Quevedo (UTEQ), específicamente en el Programa Avícola, ubicada en el km 7 ½ de la vía Quevedo – El Empalme. La investigación tuvo una duración de 15 días, distribuidos en: Recolección de información, elaboración del producto terminado (carne cocida), y la valoración organoléptica aplicando la Prueba de Intensidad de Atributo citado por la autora (Vásquez & Vera, 2021). Diseño de investigación En la presente investigación se evaluó la calidad organoléptica (sabor, olor, color y jugosidad) de la carne de pavo Americano BIG-6 con dos dietas alimenticias balanceado + nabo (Brassica rapa L.) y (balanceado puro), a través de la prueba intensiva de atributos, esta prueba permitió evaluar los atributos del producto terminado, el trabajo experimental tuvo una duración de 15 días, distribuidos en: Recolección de información, elaboración del producto terminado (carne cocida), y la valoración organoléptica aplicando la Prueba de Intensidad de Atributo citado por la autora. Población, muestra o grupo de estudio Las unidades experimentales consistieron en 5000 g. de pechuga de pavo por cada tratamiento y 100 g. por cada catador con cinco repeticiones para análisis organolépticos, el cual se realizó con un grupo académicos de la institución. La valoración fue con respecto a los ITEMS siendo: Nada, Poco, Moderada, Mucho, con respecto se midió las variables de intensidad de color, intensidad de sabor, intensidad de olor, intensidad de textura Vol. 7, N°. CININGEC II - 2022 JOURNAL OF SCIENCE AND RESEARCH E-ISSN: 2528-8083 RESULTADOS. Intensidad de sabor Tabla 1. Intensidad de sabor de la carne de pavos americanos BIG-6 alimentados con balanceado + nabo (Brassica rapa L.). Tratamientos Balanceado + nabo Ítems (Brassica rapa L.) Balanceado Nada 2 48 Poco 7 2 Moderado 18 0 Mucho 23 0 Total 50 50 Interpretación: La prueba de chi cuadrado indicó que existen diferencias de sabor en la carne de pavos alimentados con los dos tratamientos con un valor de chi (86,10) altamente significativo referente a la tabla chi. En el cuadro 1 se observó un elevado valor, los datos de los encuestados indicaron que el tratamiento a base de balanceado + nabo (Brassica rapa L.) incidió en la intensidad de sabor en una escala de Mucho, mientras que en el tratamiento a base de balanceado solo, casi la totalidad de los panelistas indicó que no poseía ninguna intensidad de sabor ajena al sabor regular del pavo. Vol. 7, N°. CININGEC II - 2022 JOURNAL OF SCIENCE AND RESEARCH E-ISSN: 2528-8083 Intensidad de olor Tabla 2. Intensidad de olor de la carne de pavos alimentados con balanceado + nabo (Brassica rapa L.). Tratamientos Balanceado + nabo Ítems (Brassica rapa L.) Balanceado Nada 0 41 Poco 0 9 Moderado 36 0 Mucho 14 0 Total 50 50 Interpretación: La prueba de chi cuadrado indicó que existen diferencias de olor en la carne de pavos alimentados con los dos tratamientos con un valor de chi (100) altamente significativo referente a la tabla chi. En el cuadro 2 se observó un ponderado valor, los datos de los encuestados indicaron que el tratamiento a base de balanceado + nabo (Brassica rapa L.) incidió en la intensidad de olor en una escala de Moderado, mientras que en el tratamiento a base de balanceado solo, casi la totalidad de los panelistas indicó que no poseía ninguna intensidad de olor ajeno al olor regular del pavo. Vol. 7, N°. CININGEC II - 2022 JOURNAL OF SCIENCE AND RESEARCH E-ISSN: 2528-8083 Intensidad de color Tabla 3. Intensidad de color de la carne de pavos alimentados con balanceado + nabo (Brassica rapa). Tratamientos Balanceado + nabo Ítems (Brassica rapa L.) Balanceado Nada 1 0 Poco 11 8 Moderado 37 42 Mucho 1 0 Total 50 50 Interpretación: La prueba de chi cuadrado indicó que existen mínimas diferencias de color en la carne de pavos alimentados con los dos tratamientos con un valor de chi (2,79), porcentaje bajo referente a la tabla chi. En el cuadro 3 se observó un moderado valor, los datos de los encuestados indicaron que el tratamiento a base de balanceado + nabo (Brassica rapa L.) incidió insignificantemente en la intensidad de color en una escala de Moderado, mientras que en el tratamiento a base de balanceado solo, casi la totalidad de los panelistas indicaron que no poseía ninguna intensidad de color ajeno al color regular del pavo. Vol. 7, N°. CININGEC II - 2022 JOURNAL OF SCIENCE AND RESEARCH E-ISSN: 2528-8083 Intensidad de textura Tabla 4. Intensidad de textura de pavos alimentados con balanceado y nabo (Brassica rapa L.). Tratamientos Balanceado + nabo Ítems (Brassica rapa L.) Balanceado Nada 0 1 Poco 15 42 Moderado 34 7 Mucho 1 0 Total 50 50 Interpretación: La prueba de chi cuadrado indicó que existen diferencias de textura (jugosidad) en la carne de pavos alimentados con los dos tratamientos con un valor de chi (32.57), porcentaje significativo referente a la tabla chi. En el cuadro 4 se observó un moderado valor, los datos de los encuestados indicaron que el tratamiento a base de balanceado + nabo (Brassica rapa L.) incidió moderadamente en la intensidad de textura (jugosidad) en una escala de Moderado, mientras que en el tratamiento a base de balanceado solo, casi la totalidad de los panelistas indicaron que no poseía ninguna intensidad de jugosidad ajena a la regular del pavo. DISCUSIÓN. Los compuestos del balanceado para las aves es una respuesta efectiva por aplicación de forrajes de nabo a la alimentación dietética para los pavos Big-6 además de menorar costo de producción de acuerdo a Gonzales et al., ( 2019) la dieta corresponde alrededor del 70% de los costos de producción además de que exista buenas practica agrícolas en sanidad de la parvada y el aumento de peso para la comercialización de la misma. En contexto, estos datos tienen concordancia con los obtenidos por Cruz & Andrade, ( 2021) donde encontraron un mejor aumento propiedades y calidad sensorial en alimentos de carne de pavos y aceptado al consumidor. Vol. 7, N°. CININGEC II - 2022 JOURNAL OF SCIENCE AND RESEARCH E-ISSN: 2528-8083 El panel de cata mencionó que el determinar la textura es complejo establecer una diferencia precisa, la utilización de la dieta de origen vegetal en este caso nabo tiene una cualidad de mejorar el metabolismo del animal por lo que conlleva a lo mencionado por Suárez & Zambrano, ( 2016) no obstante tiene que ver con la variabilidad de propiedades físicas de modo que será percibida por medio de receptores táctiles tanto piel y músculos. Los catadores concuerdan que la aplicación de alimento de origen vegetal como el nabo en la dieta de los pavos genera intensidad de sabor de manera regular con lo que concuerda de (Suárez & Zambrano, 2016) el cual menciona que la ayuda a aumentar la calidad sensorial “sabor” por respuesta a tener un aumento de las cualidades en palatabilidad. En cuanto al olor guarda relación con la investigación de Suárez & Zambrano, ( 2016) en el cual define que la alimentación con forraje de nabo en pavos aumento la incidencia de olor en una escala de moderado en comparación al testigo de alimentación de balanceado tradicional. No obstante para el color tiene semejanza por lo mencionado de los autores (USDA, 2020) quien describe que la coloración de la carne cambia su tonalidad con la dieta que se le esté suministrando a las aves por lo que cambia su aspecto de tal modo que existe la influencia en la aceptación de los consumidores. CONCLUSIONES. Una vez aplicada la evaluación de las características organolépticas de la carne de pavo americano BIG-6 mediante la prueba de intensidad de atributos se concluye en que: * La dieta alimenticia de balanceado + nabo (Brassica rapa L.) en pavos Americanos BIG-6 incidió en la calidad organoléptica produciendo una carne de mejor calidad. * Las características sensoriales con mayor influencia sobre la aceptación de la carne de pavo por parte del consumidor son el sabor, olor y jugosidad. * Los resultados obtenidos mediante el método estadístico Chi Cuadrado fueron los esperados por lo que se aprueba la hipótesis. Vol. 7, N°. CININGEC II - 2022 JOURNAL OF SCIENCE AND RESEARCH E-ISSN: 2528-8083 REFERENCIAS BIBLIOGRÁFICAS Cruz, J., & Andrade, V. (2021). EVALUACIÓN DEL COMPORTAMIENTO PRODUCTIVO DE UN SISTEMA DE CRIANZA DE PAVOS BLANCOS (Meleagridis gallopavo) EN LA COMUNA SINCHAL, PROVINCIA DE SANTA ELENA [Universidad Estatal Península de Santa Elena]. https://repositorio.upse.edu.ec/bitstream/46000/6414/1/UPSE-TIA-2021-0053.pdf Gallo, C., & Tadich, N. (2008). Bienestar animal y calidad de carne durante los manejos previos al faenamiento en bovinos. Revista Electrónica Veterinaria REDVET, 9(10B), 1–19. https://www.redalyc.org/pdf/636/63617111001.pdf Gonzales, M., Cámara, L., Fernández, Á., Aguirre, L., & Fondevila, G. (2019). Factores antinutricionales de los ingredientes y su impacto en alimentación de aves y porcino. Politécnica, 113–132. https://oa.upm.es/64962/ Sasongko, A. (2015). Evaluación de dos sistemas de crianza para mejorar los parametros productivos en pavos blnacos (Meleagridis Pavipollo) [Universidad Técnica de Ambato]. In Khatulistiwa Informatika. https://repositorio.uta.edu.ec/bitstream/123456789/24744/1/Tesis 76 Medicina Veterinaria y Zootecnia -CD 454.pdf Suárez, G., & Zambrano, S. (2016). ALIMENTACIÓN DE PAVOS AMERICANOS BIG-6 CON UNA DIETA BALANCEADA MÁS LA SUPLEMENTACIÓN DE NABO (Brassica rapa L.) EN LA FASE DE ENGORDE [Universidad Técnica Estatal de Quevedo]. https://repositorio.uteq.edu.ec/bitstream/43000/1873/1/T-UTEQ-0035.pdf USDA. (2020). El Color de las Carnes y de las Aves Acabo. Servicio de Inocuidad e Inspección de Alimentos, 2(1), 1–3. https://www.produccionanimal.com.ar/informacion_tecnica/carne_y_subproductos/192-Color_Carnes.pdf Vásquez, L. H., & Vera, J. F. (2021). INDUCCIÓN DE Rhizobium japonicum EN LA MASA FERMENTATIVA DE DOS VARIEDADES DE CACAO (Theobroma cacao L.) NACIONAL Y TRINITARIO COMO ESTRATEGIA PARA LA DISMINUCIÓN DE CADMIO. Universidad Técnica Estatal de Quevedo. Vol. 7, N°. CININGEC II - 2022
https://openalex.org/W2989297097	https://www.repo.uni-hannover.de/bitstream/123456789/9444/1/Kauth_2019_IOP_Conf._Ser.__Mater._Sci._Eng._643_012142.pdf	English	null	Experimental validation of a compact mixed-flow compressor for an active high-lift system	IOP conference series. Materials science and engineering	2,019	cc-by	5,543	IOP Conference Series: Materials Science and Engineering IOP Conference Series: Materials Science and Engineering PAPER • OPEN ACCESS PAPER • OPEN ACCESS This content was downloaded from IP address 194.95.159.52 on 18/02/2020 at 12:35 Experimental validation of a compact mixed-flow compressor for an active high-lift system To cite this article: Felix Kauth et al 2019 IOP Conf. Ser.: Mater. Sci. Eng. 643 012142 To cite this article: Felix Kauth et al 2019 IOP Conf. Ser.: Mater. Sci. Eng. 643 012142 View the article online for updates and enhancements. This content was downloaded from IP address 194.95.159.52 on 18/02/2020 at 12:35 International Scientific Electric Power Conference – 2019 IOP Conf. Series: Materials Science and Engineering 643 (2019) 012142 IOP Publishing doi:10.1088/1757-899X/643/1/012142 International Scientific Electric Power Conference – 2019 Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. Published under licence by IOP Publishing Ltd 1 * E-mail: kauth@tfd.uni-hannover.de Abstract. Compact, electrically-driven compressors are a core component of a novel active high- lift system for future commercial aircraft. A newly-developed aeromechanical optimization process was used to design the compressor stage. The optimization resulted in an unusual mixed- flow compressor design with very low aspect ratio blades and a high rotational speed of up to 60,000 rpm. Due to the unusual design, experimental validation of the performance predictions by means of CFD is necessary. This paper presents the first experimental results obtained using a preliminary prototype at part-speed, i.e. rotational speeds from 20,000 to 30,000 rpm. The experimentally-determined pressure ratios deviate up to 1.5 %, the polytropic efficiencies up to 4 percentage points from the CFD predictions. Besides the deficiencies of available turbulence models, the underestimation of overall losses is presumably due to the omission of the volute in the CFD model. An experimental validation of the CFD predictions at full-speed is under way. Experimental validation of a compact mixed-flow compressor for an active high-lift system Felix Kauth1,* , Niklas Maroldt1, and Joerg R Seume1 1 Leibniz University, Institute of Turbomachinery and Fluid Dynamics, 30167 Hannover, Germany Felix Kauth1,* , Niklas Maroldt1, and Joerg R Seume1 1 Leibniz University, Institute of Turbomachinery and Fluid Dynamics, 30167 Hannover, Germany * E-mail: kauth@tfd.uni-hannover.de 1. Introduction Improved high-lift systems have the potential to reduce noise and carbon dioxide emissions of modern commercial aircraft. An efficient active high-lift system using a flexible leading-edge device, together with a combination of boundary layer suction and blowing over a Coanda flap was presented by Radespiel and Heinze [1]. The required air supply for this system is provided by compact electrical compressor units, which are integrated into the wings. A design process for the components of these compressor units was developed by Teichel et al. [2]. Components with high specific power are needed to reduce mass and volume of the system. The use of automatic optimization methods as described by Teichel et al. [3] resulted in an unusual, transonic, mixed-flow compressor design with a high axial flow component. According to CFD predictions, the aerodynamic design point (total pressure ratio of 2.33 and corrected mass flow rate of 1.11 kg/s) is reached at a corrected rotational speed of 60,125 rpm with a polytropic efficiency of 83.3 %. The impact of inlet flow distortion on the stage performance resulting from an integration of the electrical components into the compressor inlet duct was investigated by Kauth et al. [4]. Figure 1 shows the compressor test rig, which was used for the experimental part-speed validation of the present study [5]. International Scientific Electric Power Conference – 2019 IOP Conf. Series: Materials Science and Engineering 643 (2019) 012142 doi:10.1088/1757-899X/643/1/012142 OP Conf. Series: Materials Science and Engineering 6 Figure 1. Left: longitudinal section of the compressor as tested, with performance measurement stations ‘1’ in inlet duct and ‘2’ in radial duct upstream of the volute; right: compressor test facility [5]. Figure 1. Left: longitudinal section of the compressor as tested, with performance measurement stations ‘1’ in inlet duct and ‘2’ in radial duct upstream of the volute; right: compressor test facility [5]. Figure 1. Left: longitudinal section of the compressor as tested, with performance measurement stations ‘1’ in inlet duct and ‘2’ in radial duct upstream of the volute; right: compressor test facility [5]. Mixed-flow compressors combine advantages of both axial and radial compressors, resulting in compact stages with high specific power. Most known mixed-flow compressor designs are similar to centrifugal compressors with a high impeller hub cone angle. Watanabe & Sakai [6] showed an improvement of the total and static pressure ratio for decreasing cone angles. Diener et al. 1. Introduction [7] combined a neural network with CFD computations in order to optimize the aerodynamic design of a mixed-flow compressor, followed by mechanical analyses. For initial design considerations Hazby et al. [8] and Nassar et al. [9] made use of a simpler streamline curvature solver, before analysing the design using CFD. Additionally, Hazby et al. [8] compared the predicted performance to experimental results, showing good agreements for low-speed operation. However, in particular for higher mass flow rates, lower total pressure ratios were predicted. In contrast, mixed-flow compressors with a high axial flow component have rarely been discussed in open literature. A detailed design description and experimental characterisation of such a transonic mixed-flow compressor was given by Musgrave & Plehn [10]. The stage with a pressure ratio of 3:1 was designed to run downstream of an axial compressor stage. Experimental and analytical investigations of a supersonic mixed-flow compressor were done by Mönig et al. [10]. The performance and stability of any type of compressor can be strongly influenced by the tip leakage flow. For axial compressors the tip clearance effects are mainly determined by the clearance-to-blade- height ratio according to Storer & Cumpsty [12]. Due to the small blade heights, the magnitude of this effect is relatively high for mixed-flow compressors. However, only few works were published on this subject. An experimental study of tip clearance effects for a mixed-flow compressor with a high proportion of radial flow was performed by Rajakuma et al. [13] for constant and variable tip clearance gaps. Numakura et al. [14] conducted a detailed numerical and experimental investigation of the casing treatment of two mixed flow compressors. For the present compressor an axial slot casing treatment has been designed by Du & Seume [15] and recently experimentally evaluated by Du et al. [16]. 2.1. Computational fluid dynamics Additionally, the inlet and outlet ducts of the stage, where pressure and temperature were measured in the experiment, were included into the model. Structured grids were generated for rotor, stator and inlet duct using NUMECA AutoGrid5. The structured grid of the radial outlet duct was generated with ANSYS ICEM CFD. as initial condition. A solution was considered as converged if the root mean square residuals decreased to 10-5, and the maximum residuals to 10-3. Additionally, mass flow, pressure ratio, efficiency and maximum eddy viscosity were monitored and not allowed to vary more than 0.01 % over the last 100 iterations. The CFD simulations during the optimized design process described in [3] had been done using only a model of the rotor and stator row without any connecting ducts [2]. To validate the CFD simulations, the numerical model should match the experimental setup as well as possible. Consequently, new grids were generated for the blade rows, which had to be manufactured with larger leading and trailing edge radii than anticipated. Different rotor tip clearances between 0.1 and 0.4 mm were modelled, while the stator tip clearance was fixed to 0.1 mm. Additionally, the inlet and outlet ducts of the stage, where pressure and temperature were measured in the experiment, were included into the model. Structured grids were generated for rotor, stator and inlet duct using NUMECA AutoGrid5. The structured grid of the radial outlet duct was generated with ANSYS ICEM CFD. as initial condition. A solution was considered as converged if the root mean square residuals decreased to 10-5, and the maximum residuals to 10-3. Additionally, mass flow, pressure ratio, efficiency and maximum eddy viscosity were monitored and not allowed to vary more than 0.01 % over the last 100 iterations. The CFD simulations during the optimized design process described in [3] had been as initial condition. A solution was considered as converged if the root mean square residuals decreased to 10-5, and the maximum residuals to 10-3. Additionally, mass flow, pressure ratio, efficiency and maximum eddy viscosity were monitored and not allowed to vary more than 0.01 % over the last 100 iterations. The CFD simulations during the optimized design process described in [3] had been done using only a model of the rotor and stator row without any connecting ducts [2]. To validate the CFD simulations, the numerical model should match the experimental setup as well as possible. 2.1. Computational fluid dynamics Consequently, new grids were generated for the blade rows, which had to be manufactured with larger leading and trailing edge radii than anticipated. Different rotor tip clearances between 0.1 and 0.4 mm were modelled, while the stator tip clearance was fixed to 0.1 mm. Additionally, the inlet and outlet ducts of the stage, where pressure and temperature were measured in the experiment, were included into the model. Structured grids were generated for rotor, stator and inlet duct using NUMECA AutoGrid5. The structured grid of the radial outlet duct was generated with ANSYS ICEM CFD. A grid convergence study according to [19] was performed for the key variable polytropic efficiency on three grids of increasing resolution at a high Reynolds number operating point. The fine grid convergence index GCI1 was calculated using Eq. 1 with the approximated relative error according to Eq. 2, with fi being the solution on grid level i. The grid refinement factors were 𝑟21 = 1.45 and 𝑟32 = 1.36. An empirical safety factor of 𝐹𝑠= 1.25 was used, as suggested by [20]. The apparent order was calculated using Eq. 3, where 𝑠= 𝑠𝑔𝑛((𝑓3 −𝑓2)/(𝑓2 −𝑓1)). Due to its reasonable trade-off between computational cost and accuracy, the medium grid (𝐺𝐶𝐼2 = 0.0011) was chosen for the numerical performance analysis. 𝐺𝐶𝐼1 = 𝐹𝑠𝑒21 𝑟21 𝑝−1 (1) 𝑒21 = \| 𝑓1−𝑓2 𝑓1 \| (2) 𝑝= ln\|𝑓3−𝑓2 𝑓2−𝑓1\|+ln( 𝑟21 𝑝−𝑠 𝑟32 𝑝−𝑠) ln(𝑟21) (3) 2.2. Experiment 𝐺𝐶𝐼1 = 𝐹𝑠𝑒21 𝑟21 𝑝−1 (1) 𝑒21 = \| 𝑓1−𝑓2 𝑓1 \| (2) 𝑝= ln\|𝑓3−𝑓2 𝑓2−𝑓1\|+ln( 𝑟21 𝑝−𝑠 𝑟32 𝑝−𝑠) ln(𝑟21) (3) (1) (2) (3) 2.1. Computational fluid dynamics 2.1. Computational fluid dynamics The industrial CFD code TRACE (version 8) was used for the numerical performance predictions. The governing equations in TRACE are discretized by means of the finite volume method, time discretization uses an implicit first-order Euler method and the solver uses a predictor-corrector algorithm [17]. TRACE was verified for turbomachinery applications e.g. in [18]. The SST turbulence model was applied with the assumption of a fully-turbulent flow. All walls were set as adiabatic with a non-slip condition. A low-Reynolds number approach was used, and consequently all meshes used a dimensionless wall distance of y+ ≈ 1 to ensure that the first grid node was situated within the viscous sublayer of the boundary layer. All simulations used the same inlet boundary conditions: total pressure 101,350 Pa, total temperature 288.15 K, radial and tangential flow angle 0°, turbulent intensity 5 %, and turbulent length scale 18.8 µm. At the outlet, an average static pressure between 102,000 and 130,000 Pa was imposed at mid-span. Interfaces between rotating and stationary domains were modelled using the mixing-plane approach. A rotational speed between 20,000 and 30,000 rpm was imposed on the fluid in the rotor domain. To compute the characteristic compressor speed lines, the static pressure at the outlet was increased incrementally for each rotational speed, using the previous converged solution 2 International Scientific Electric Power Conference – 2019 IOP Publishing IOP Conf. Series: Materials Science and Engineering 643 (2019) 012142 doi:10.1088/1757-899X/6 as initial condition. A solution was considered as converged if the root mean square residuals decreased to 10-5, and the maximum residuals to 10-3. Additionally, mass flow, pressure ratio, efficiency and maximum eddy viscosity were monitored and not allowed to vary more than 0.01 % over the last 100 iterations. The CFD simulations during the optimized design process described in [3] had been done using only a model of the rotor and stator row without any connecting ducts [2]. To validate the CFD simulations, the numerical model should match the experimental setup as well as possible. Consequently, new grids were generated for the blade rows, which had to be manufactured with larger leading and trailing edge radii than anticipated. Different rotor tip clearances between 0.1 and 0.4 mm were modelled, while the stator tip clearance was fixed to 0.1 mm. 2.2. Experiment p Experimental validation of the numerical performance predictions is necessary in particular because of the unconventional compressor design. As the whole design process of the electrically-powered high- lift system relies on these performance predictions, first results were required before the newly developed full-speed electrical machine was available. Therefore as a first step, part-speed tests were conducted using a commercially available electric motor with a maximum rotational speed of 30,000 rpm. The inlet flow direction was axial and downstream of the stator the flow was guided to the radial direction and into a volute, due to the bigger diameter of this motor, compared to the future full- speed motor. p The rotor tip clearance was varied by using distance washers of varying thicknesses between the rotor and the stator casing. The nominal tip clearances used during the experiments were the “Design clearance” of 0.1 mm and the “Large clearance” of 0.2 mm. Due to manufacturing tolerances a constant stator tip clearance of 0.1 mm was unavoidable. To evaluate the compressor performance, static pressure taps and total temperature probes were used at the inlet and outlet of the compressor rig (see Figure 1), as well as in the radial duct between stator and volute. Inflow dynamic pressure was measured using a Prandtl tube. Pressures were measured relative to the ambient pressure, which was recorded separately. An ultrasonic gas flow meter was used to measure the air volume flow. All variables were recorded using a sampling rate of 5 Hz with 20 samples for every operating point. Three parallel throttle valves were used to control the compressor back pressure. Table 1 summarizes the information about all measurement devices. 3 3 International Scientific Electric Power Conference – 2019 IOP Publishing IOP Conf. Series: Materials Science and Engineering 643 (2019) 012142 doi:10.1088/1757-899X/643/1/012142 Conf. Series: Materials Science and Engineering 643 (2019) 012142 doi:10.1088/1757-899X/643/1/01214 Table 1. 2.2. Experiment Instrumentation of the test rig Location Transducer Range Accuracy Measuring cabin Mensor CPT6100 55.2 to 117.2 kPa 0.01 % Air intake FlowSIC 600 20 to 30000 m³/h 1 % Inlet duct NetScanner 9116 15 psi 0.05 % FS PT100 -100 to 450 °C 0.15 + 0.002 \|t\| °C Radial duct NetScanner 9816 45 psi 0.05 % FS Type K thermocouple -40 to 1000 °C 1.5 °C Outlet duct NetScanner 9116 45 psi 0.05 % FS PT100 -100 to 450 °C 0.15 + 0.002 \|t\| °C For each performance parameter 𝑦= 𝑓(𝑥1, … , 𝑥𝑚), the resulting measurement uncertainty was quantified according to [21] using Eq. 4. Assuming a Student’s t-distribution, the random uncertainty of each measurand xi was calculated based on its experimental standard deviation for a 90 % confidence interval (CI). The systematic uncertainty of each measurand was calculated from the accuracy of the transducer, assuming a uniform distribution. 𝑢(𝑦) = √∑ ( 𝛿𝑓 𝛿𝑥𝑖) 2 𝑢2(𝑥𝑖) 𝑚 𝑖=1 (4) (4) In the following, measurement results are given as arithmetic mean values together with the calculated standard uncertainty. The main compressor performance parameters used for the validation are static and total pressure ratio, as well as polytropic efficiency. While the static pressure ratio Π = 𝑝𝑜𝑢𝑡/𝑝𝑖𝑛 can be calculated directly from the measured static pressures at inlet and outlet, the total pressure at the outlet has to be calculated using Eq. 5 to 7. The polytropic compressor efficiency is calculated according to Eq. 8. 𝑝𝑡= 𝑝( 𝑇𝑡 𝑇) 𝜅 𝜅−1 (5) 𝑇= 𝑇𝑡− 𝑐𝑥2 2𝑐𝑝 (6) 𝑐𝑥= 𝑚̇ 𝑅𝑇 𝑝𝐴 (7) 𝜂𝑝= 𝜅−1 𝜅 𝑙𝑛( 𝑝𝑡,2 𝑝𝑡,1) 𝑙𝑛( 𝑇𝑡,2 𝑇𝑡,1) (8) (5) (6) (8) For the performance comparison, the rotational speed N and the measured mass flow rate 𝑚̇ are corrected using inlet pressure and temperature, as well as the international standard atmosphere (𝑝𝑟𝑒𝑓 = 101,325 Pa and 𝑇𝑟𝑒𝑓 = 288.15 K) [22]. 𝑚̇ 𝑐𝑜𝑟𝑟= 𝑚̇ 𝑝𝑟𝑒𝑓 𝑝𝑡,1 √ 𝑇𝑡,1 𝑇𝑟𝑒𝑓 (9) N𝑐𝑜𝑟𝑟= N√ 𝑇𝑟𝑒𝑓 𝑇𝑡,1 (10) (9) (9) (10) 2.3. Finite Element Analysis 2.3. Finite Element Analysis y The compressor geometry changes during operation due to centrifugal forces and thermal expansion. To estimate the appropriate “hot” geometry, a finite element analysis (FEA) was performed using ANSYS Workbench. For this purpose, wall temperatures and pressures at different operating points were extracted from the CFD results and mapped on the FEA grids of rotor and casing. The maximum calculated overall deformation of the rotor blades for the operating point with the highest temperatures and pressures at 30,000 rpm is 0.06 mm. An evaluation of the radial component of rotor tip and casing 4 International Scientific Electric Power Conference – 2019 IOP Conf. Series: Materials Science and Engineering 643 (2019) 012142 doi:10.1088/1757-899X/6 displacement yields a mean rotor tip clearance increase of 0.009 mm at that operating point. It is assumed that the effect on the subsonic compressor performance due to those small changes in clearance and blade profile shape can be neglected. For this reason, the “cold” geometry was used for the CFD analysis presented in this paper. However, this assumption will not be valid at higher rotational speeds for two reasons. First, the deformations will become more severe due to higher centrifugal forces and temperatures. Moreover, the locally supersonic flow regime on the suction side at high-speed operating points is more sensitive to changes in blade profile geometry [22]. 3. Results As described before, compressor outlet pressures and temperatures were measured at two stations – upstream and downstream of the volute. Performance parameters calculated with the measurements upstream of the volute are denoted with “stage”, those calculated with the downstream measurements with “overall rig”. Figure 2 presents a comparison of both performance calculations at design tip clearance. Due to the use of thermocouples instead of PT100 at the measurement station upstream of the volute, the resulting uncertainties are higher for the “stage” performance. As can be seen in Figure 2, the volute seems to cause substantial pressure losses, most severely at high mass flow rates. The reason for this is probably a mismatch of stage and volute, as the latter is part of the test rig and has not been designed for the mixed-flow compressor. As the performance of the volute is not of interest for the present study, all following evaluations are done using the “stage” performance. \ Figure 2. Compressor performance (design tip clearance) for different outlet measurement positions. Figure 2. Compressor performance (design tip clearance) for different outlet measurement positions. Figure 2. Compressor performance (design tip clearance) for different outlet measurement positions. The predicted stage performance at design tip clearance is also shown in Figure 2. While the measured static pressure ratios (not shown) match the prediction within less than 1 %, the total pressure ratios deviate up to 1.5 %, and the polytropic efficiencies up to 4 percentage points. The total pressure ratio offset is almost constant over most of the operating range and gets smaller near peak efficiency. Moreover, the deviations increase at higher rotational speeds. The experimentally determined stability limit was observed at lower corrected mass flow rates compared to the last converged CFD solution for every rotational speed. Deviations in the same order of magnitude between CFD predictions and performance measurements of a mixed-flow compressor were observed by [8]. They used the k-ε turbulence model for their CFD predictions, and presumed that the deviations were due to turbulence modelling and geometrical differences between numerical model and experiment. A h i l i h d h lik l f h diff Figure 2. Compressor performance (design tip clearance) for different outlet measurement positions. Figure 2. Compressor performance (design tip clearance) for different outlet measurement positions. The predicted stage performance at design tip clearance is also shown in Figure 2. 3. Results 5 nternational Scientific Electric Power Conference – 2019 IOP Conf. Series: Materials Science and Engineering 643 (2019) 012142 g doi:10.1088/1757-899X/643/1/012142 Figure 3. CFD performance prediction using different turbulence models for two tip clearance settings. The mixed-flow compressor investigated here has a unique design and due to its compactness, “secondary” flow phenomena and associated losses are expected to be more dominant than in bigger axial compressors. An important source of modelling errors in RANS computations is the use of turbulence models [23]. To demonstrate the sensitivity of the CFD results to the choice of turbulence model, Figure 3 presents results obtained using two of the most commonly used turbulence models for compressor simulations, the SST and the k-ω model. As explained by [23], turbulence models usually give good results for fully turbulent and attached boundary layers. However, especially for compressors near stall, where separating and separated flows occur, big differences between turbulence models can be expected. Figure 3 confirms this hypothesis, as the biggest differences between the two models occur near the stability limit. Figure 3. CFD performance prediction using different turbulence models for two tip clearance settings. The mixed-flow compressor investigated here has a unique design and due to its compactness, “secondary” flow phenomena and associated losses are expected to be more dominant than in bigger axial compressors. An important source of modelling errors in RANS computations is the use of turbulence models [23]. To demonstrate the sensitivity of the CFD results to the choice of turbulence model, Figure 3 presents results obtained using two of the most commonly used turbulence models for compressor simulations, the SST and the k-ω model. As explained by [23], turbulence models usually give good results for fully turbulent and attached boundary layers. However, especially for compressors near stall, where separating and separated flows occur, big differences between turbulence models can be expected. Figure 3 confirms this hypothesis, as the biggest differences between the two models occur near the stability limit. For the design tip clearance, both turbulence models give similar results, although the efficiency from the k-ω model is slightly closer to the experimental results. With increasing tip clearance the total pressure ratio decreases, while the impact increases with higher rotor speeds. As mentioned by [13], the effect of tip clearance is influenced by factors like geometry and flow field, and therefore not linearly correlated with the speed. 3. Results When comparing the numerical results of both tip clearance settings, the k-ω model seems to be less sensitive to a change in tip clearance compared to both experimental as well as SST results. A parameter study was performed with different tip clearances to further evaluate this phenomenon. Figure 4 shows the results for the maximum total pressure ratio and maximum polytropic efficiency at two different rotational speeds. When using the SST model, the detrimental effect of an increased rotor tip clearance on compressor performance is more severe compared to the predictions using the k-ω model. Figure 4. Effect of turbulence model, rotational speed, and rotor tip clearance on maximum performance. 4. Conclusions An experimental part-speed validation was performed for a newly designed mixed-flow compressor for an active high-lift system at 20,000 to 30,000 rpm. For this speed range, it was demonstrated that using the “cold” geometry for the CFD analysis is acceptable. An upstream effect of the volute, which was omitted in the current CFD model, is the most likely reason for the deviations of the measured stage performance to the CFD predictions in terms of total pressure ratio and polytropic efficiency. The Figure 4. Effect of turbulence model, rotational speed, and rotor tip clearance on maximum performance. Figure 4. Effect of turbulence model, rotational speed, and rotor tip clearance on maximum performance. Figure 4. Effect of turbulence model, rotational speed, and rotor tip clearance on maximum performance. 3. Results While the measured static pressure ratios (not shown) match the prediction within less than 1 %, the total pressure ratios deviate up to 1.5 %, and the polytropic efficiencies up to 4 percentage points. The total pressure ratio offset is almost constant over most of the operating range and gets smaller near peak efficiency. Moreover, the deviations increase at higher rotational speeds. The experimentally determined stability limit was observed at lower corrected mass flow rates compared to the last converged CFD solution for every rotational speed. Deviations in the same order of magnitude between CFD predictions and performance measurements of a mixed-flow compressor were observed by [8]. They used the k-ε turbulence model for their CFD predictions, and presumed that the deviations were due to turbulence modelling and geometrical differences between numerical model and experiment. The predicted stage performance at design tip clearance is also shown in Figure 2. While the measured static pressure ratios (not shown) match the prediction within less than 1 %, the total pressure ratios deviate up to 1.5 %, and the polytropic efficiencies up to 4 percentage points. The total pressure ratio offset is almost constant over most of the operating range and gets smaller near peak efficiency. Moreover, the deviations increase at higher rotational speeds. The experimentally determined stability limit was observed at lower corrected mass flow rates compared to the last converged CFD solution for every rotational speed. Deviations in the same order of magnitude between CFD predictions and performance measurements of a mixed-flow compressor were observed by [8]. They used the k-ε turbulence model for their CFD predictions, and presumed that the deviations were due to turbulence modelling and geometrical differences between numerical model and experiment. g g p As there is no clear reason to mistrust the measurement data, the most likely reason for the difference between numerical and experimental results is omitting the volute of the test rig in the CFD model. This could cause a deviation in the results, as a volute imposes a non-axisymmetrical pressure field on the stage upstream [22]. Another feature of the test rig, which could not be modelled and has a potentially detrimental effect on the performance, is the axial thrust compensation system of the compressor [5]. A labyrinth seal is used to minimize the leakage through the rotor hub, but its effectiveness could not be verified yet. References [1] Radespiel R and Heinze W 2014 SFB 880: fundamentals of high lift for future commercial aircraftCEAS Aeronautical J. 5 [1] Radespiel R and Heinze W 2014 SFB 880: fundamentals of high lift for future commercial aircraftCEAS Aeronautical J. 5 [2] Teichel S H, Dörbaum M, Misir O, Merkert A, Mertens A, Seume J.R, Ponick B 2015 Design Considerations for the Components of Electrically Powered Active High-lift Systems in Civil Aircraft, CEAS Aeronautical J. 6 [3] Teichel S H, Verstraete T, Seume J R 2017 Optimized Multidisciplinary Design of a Small Transonic Compressor for Active High-Lift Systems JGPP 9 [4] Kauth F, Narjes G, Müller J K, Mertens A, Ponick B, Seume J R 2018 Electrically driven, Compact, Transonic Mixed-Flow Compressor for Active High-Lift Systems in Future Aircraft, Proceedings of GPPS Forum 18 [5] Kauth F, Narjes G, Müller J, Ponick B, Mertens A, Seume J.R 2017 Compact Electrical Compressors for Active Flow Control in Autonomous High-Lift Systems, SFB 880 – Fundamentals of High-Lift for Future Civil Aircraft: Biennial Report (TU Braunschweig – Campus Forschungsflughafen, Braunschweig, 2017) [6] Watanabe I, Sakai T 1965 Effect of the Cone Angle of the Impeller Hub of the Mixed Flow Compressor upon Performance Characteristics SAE Technical Paper 650037 [7] Diener O H F, van der Spuy S J, von Backström T.W, Hildebrandt T 2016 Multi-Disciplinary Optimization of a Mixed-Flow Compressor Impeller Proceedings of ASME Turbo Expo 2016 Optimization of a Mixed-Flow Compressor Impeller Proceedings of ASME Turbo Expo 2016 [8] Hazby H, Casey M, Numakura R, Tamaki H 2015 A Transonic Mixed Flow Compressor for an Extreme Duty, ASME. J. Turbomach. [8] Hazby H, Casey M, Numakura R, Tamaki H 2015 A Transonic Mixed Flow Compressor for an Extreme Duty, ASME. J. Turbomach. [9] Nassar A, Giri G, Moroz L, Sherbina A and Klimov I 2016 Design and Analysis of a High Pressure Ratio Mixed Flow Compressor Stage, 52nd AIAA/SAE/ASEE Joint Propulsion Conference, AIAA Propulsion and Energy Forum (2016) [10] Musgrave D S, Plehn N J 1987 Mixed-Flow Compressor Stage Design and Test Results With a Pressure Ratio of 3:1 ASME. J. Turbomach [11] Mönig R, Elmendorf W, Gallus H E 1993 Design and Rotor Performance of a 5:1 Mixed-Flow Supersonic Compressor ASME. J. Turbomach 115, 3 p p [12] Storer J A, Cumpsty N A 1994 An Approximate Analysis and Prediction Method for Tip Clearance Loss in Axial Compressors ASME. J. 4. Conclusions 4. Conclusions An experimental part-speed validation was performed for a newly designed mixed-flow compressor for an active high-lift system at 20,000 to 30,000 rpm. For this speed range, it was demonstrated that using the “cold” geometry for the CFD analysis is acceptable. An upstream effect of the volute, which was omitted in the current CFD model, is the most likely reason for the deviations of the measured stage performance to the CFD predictions in terms of total pressure ratio and polytropic efficiency. The An experimental part-speed validation was performed for a newly designed mixed-flow compressor for an active high-lift system at 20,000 to 30,000 rpm. For this speed range, it was demonstrated that using the “cold” geometry for the CFD analysis is acceptable. An upstream effect of the volute, which was omitted in the current CFD model, is the most likely reason for the deviations of the measured stage performance to the CFD predictions in terms of total pressure ratio and polytropic efficiency. The 6 International Scientific Electric Power Conference – 2019 IOP Publishing IOP Conf. Series: Materials Science and Engineering 643 (2019) 012142 doi:10.1088/1757-899X/6 numerical stability limit was demonstrated to be a conservative criterion for the compressor stall limit. The next step will be to include the volute into the CFD model and to estimate the leakage flow through the labyrinth seal of the axial thrust compensation system. Moreover, the appropriate “hot” geometry will be used for the CFD analysis in the context of the planned full-speed validation. y p p It was shown that the CFD results of this mixed-flow compressor are sensible to the choice of turbulence model, depending on tip clearance and rotational speed. As no detailed flow measurements were done yet, it was not possible to evaluate whether local flow phenomena, e.g. separation, were correctly modelled. For this reason, it is planned to investigate the detailed flow structure in the future, e.g. by means of particle image velocimetry. Further experiments using a newly developed high-speed electrical machine to validate the performance predictions and the correction method in the transonic range from 40,000 to 60,000 rpm are under way. Acknowledgements The authors would like to thank the German Research Foundation (DFG) for supporting this fundamental research in active high-lift systems for future aircraft as part of the Collaborative Research Centre 880 (Sonderforschungsbereich SFB 880). Moreover, the authors thank the German Aerospace Center (DLR) for the permission to use the solver TRACE. References References Turbomach 116, 4 [13] Rajakumar D R, Ramamurthy S, Govardhan M 2015 Experimental investigations on effects of tip clearance in mixed-flow compressor performance Proceedings of the Institution of Mechanical Engineers, Part G: Journal of Aerospace Engineering 229 5 g f p g g [14] Numakura R, Tamaki H, Hazby H, Casey M 2014 Effect of a Recirculation Device on the Performance of Transonic Mixed Flow Compressors Proceedings of ASME Turbo Expo 2014 7 International Scientific Electric Power Conference – 2019 [15] Du J, Seume J R 2017 Design of Casing Treatment on a Mixed-Flow Compressor. Proceedings of ASME Turbo Expo 2017 [15] Du J, Seume J R 2017 Design of Casing Treatment on a Mixed-Flow Compressor. Proceedings of ASME Turbo Expo 2017 [16] Du J, Li J, Zhang Q, Kauth F, Seume J R 2019 Experimental Study on the Influence of Casing Treatment on Near-Stall Unsteady Behavior of a Mixed-Flow Compressor. Proceedings of ASME Turbo Expo 2019, to be published p p [17] Deutsches Zentrum für Luft- und Raumfahrt, TRACE User Guide, http://www.trace- portal.de/userguide/trace/index.html (accessed March 2019) [18] Kügeler E, Nürnberger D, Weber A, Engel K 2008 Influence of blade fillets on the performance of a 15 stage gas turbine compressor, Proceedings of ASME Turbo Expo 2008 [19] American Society of Mechanical Engineers, Standard for verification and validation in computational fluid dynamics and heat transfer (American Society of Mechanical Engineers, New York, 2009) ) [20] Roache P J 1998 Verification and Validation in Computational Science and Engineering (Hermosapublishers, Albuquerque, 1998) [21] DIN 1319-3:1996-05 1996Fundamentals of metrology - Part 3: Evaluation of measurements of a single measurand, measurement uncertainty g , y [22] Cumpsty N A 2004 Compressor aerodynamics (Krieger Pub, Malabar, Fla, 2nd ed, 2004 [23] Denton J D 2010 Some limitations of turbomachinery CFD, Proceedings of ASME Turbo Expo 2010 8 8
https://openalex.org/W2969166116	https://zenodo.org/records/3369504/files/BISS_article_38554.pdf	English	null	Facing e-Biodiversity Challenges Together: GBIO framework-based synergies between DiSSCo and LifeWatch ERIC	Biodiversity Information Science and Standards	2,019	cc-by	1,032	Facing e-Biodiversity Challenges Together: GBIO framework-based synergies between DiSSCo and LifeWatch ERIC Juan Miguel González-Aranda , Dimitrios Koureas , Wouter Addink , Tim Hirsch , Christos Arvanitidis , Antonio José Sáenz Albanés , Peter Schalk ‡ § § \| ‡ ‡ ¶ ‡ LifeWatch ERIC, Seville, Spain § Naturalis Biodiversity Center, Leiden, Netherlands \| GBIF, Copenhagen, Denmark ¶ Naturalis, Leiden, Netherlands Corresponding author: Juan Miguel González-Aranda (cto@lifewatch.eu) Received: 25 Jul 2019 \| Published: 08 Aug 2019 Citation: González-Aranda JM, Koureas D, Addink W, Hirsch T, Arvanitidis C, Albanés AJS, Schalk P (2019) Facing e-Biodiversity Challenges Together: GBIO framework-based synergies between DiSSCo and LifeWatch ERIC. Biodiversity Information Science and Standards 3: e38554. https://doi.org/10.3897/biss.3.38554 Corresponding author: Juan Miguel González-Aranda (cto@lifewatch.eu) Received: 25 Jul 2019 \| Published: 08 Aug 2019 Citation: González-Aranda JM, Koureas D, Addink W, Hirsch T, Arvanitidis C, Albanés AJS, Schalk P (2019) Facing e-Biodiversity Challenges Together: GBIO framework-based synergies between DiSSCo and LifeWatch ERIC. Biodiversity Information Science and Standards 3: e38554. https://doi.org/10.3897/biss.3.38554 Corresponding author: Juan Miguel González-Aranda (cto@lifewatch.eu) Received: 25 Jul 2019 \| Published: 08 Aug 2019 © González-Aranda J et al. This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Presenting author Juan Miguel González-Aranda Abstract 3 Participation in the co-design and co-implementation of relevant e-Services in LifeBlock (LifeWatch ERIC blockchain-based technology platform); • Participation in the co-design and co-implementation of relevant e-Services in LifeBlock (LifeWatch ERIC blockchain-based technology platform); • The active participation of DiSSCo for integrating collections data: DiSSCo is one of the main resources needed for the integration of GLOBIS-B GLOBal Infrastructures for Supporting Biodiversity work on Essential Biodiversity Variables (EBVs) (Kissling et al. 2018). Thus, EBVs together with species traits will be integrated into LifeBlock platform in order to feed Ecosystem Services needed to further support Biodiversity Ecosystem Services VRE provided by LifeWatch ERIC distributed e- Infrastructure. Presented at Biodiversity_Next 2019 Biodiversity_Next 2019 Abstract The collaboration between LifeWatch ERIC and DiSSCo (Distributed System of Scientific Collections), both pan-European research infrastructures focusing on biodiversity, can be achieved in a number of ways. The direct initiation of this collaboration can be carried out through their joint support to GBIF (Global Biodiversity Information Facility). This approach will facilitate meeting GBIF’s overall objective stated as: “Connecting data and expertise: a new alliance for biodiversity knowledge” (Hobern and Miller 2019). LifeWatch ERIC supports GBIF in a collaborative way by integrating and providing e- Services according to Global Biodiversity Informatics Outlook (GBIO) Framework objectives (Fig. 1), particularly suitable for the Understanding focus area. This concentrates on building modeled representations of biodiversity patterns and properties, based on any possible evidence, using the following components: © González-Aranda J et al. This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. González-Aranda J et al González-Aranda J et al 2 Figure 1. GBIO Framework (Hobern et al. 2012) identifying 20 components as essential elements of biodiversity informatics and organized as four layers of focus: Culture, Data, Evidence and Understanding. GBIO Framework (Hobern et al. 2012) identifying 20 components as essential elements of biodiversity informatics and organized as four layers of focus: Culture, Data, Evidence and Understanding. In this regard, and during the 2nd Global Biodiversity Information Conference, LifeWatch ERIC actively participated in one of the four parallel working groups reviewing different components from the GBIO framework. Each component was selected to capture information on a broad range of different challenges and opportunities. At the same event, DiSSCo mainly focused on the Data layer, as the main provider of data and other types of collections resources in Europe. The Evidence layer is the fertile interface to develop sound synergies for collaboration by both research infrastructures in order to support GBIF through the development of 3 concrete activities: • Participation in the co-design, development and deployment of a multi-purpose Virtual Research Environment (VRE) to support DiSSCo, specifically by integrating the collections e-Services and by engaging the various communities of practice; Facing e-Biodiversity Challenges Together: GBIO framework-based synergies ... References References • Hobern D, Apostolico A, Arnaud E, Bello JC, Canhos D, Dubois G, Field D, Alonso García E, Hardisty A, Harrison J, Heidorn B, Krishtalka L, Mata E, Page R, Parr C, Price J, Willoughby S (2012) Global Biodiversity Informatics Outlook: Delivering biodiversity knowledge in the information age. Global Biodiversity Information Facility https:// doi.org/10.15468/6JXA-YB44 • Hobern D, Miller J (2019) An alliance for biodiversity knowledge: Rethinking international collaboration in biodiversity informatics. Biodiversity Information Science and Standards 3 https://doi.org/10.3897/biss.3.37324 • Kissling WD, Walls R, Bowser A, Jones M, Kattge J, Agosti D, Amengual J, Basset A, van Bodegom P, Cornelissen JC, Denny E, Deudero S, Egloff W, Elmendorf S, Alonso García E, Jones K, Jones O, Lavorel S, Lear D, Navarro L, Pawar S, Pirzl R, Rüger N, Sal S, Salguero-Gómez R, Schigel D, Schulz K, Skidmore A, Guralnick R (2018) Towards global data products of Essential Biodiversity Variables on species traits. Nature Ecology & Evolution 2 (10): 1531‑1540. https://doi.org/10.1038/s41559-018-0667-3 • Hobern D, Apostolico A, Arnaud E, Bello JC, Canhos D, Dubois G, Field D, Alonso García E, Hardisty A, Harrison J, Heidorn B, Krishtalka L, Mata E, Page R, Parr C, Price J, Willoughby S (2012) Global Biodiversity Informatics Outlook: Delivering biodiversity knowledge in the information age. Global Biodiversity Information Facility https:// doi.org/10.15468/6JXA-YB44 • Hobern D, Miller J (2019) An alliance for biodiversity knowledge: Rethinking international collaboration in biodiversity informatics. Biodiversity Information Science and Standards 3 https://doi org/10 3897/biss 3 37324 • Kissling WD, Walls R, Bowser A, Jones M, Kattge J, Agosti D, Amengual J, Basset A, van Bodegom P, Cornelissen JC, Denny E, Deudero S, Egloff W, Elmendorf S, Alonso García E, Jones K, Jones O, Lavorel S, Lear D, Navarro L, Pawar S, Pirzl R, Rüger N, Sal S, Salguero-Gómez R, Schigel D, Schulz K, Skidmore A, Guralnick R (2018) Towards global data products of Essential Biodiversity Variables on species traits. Nature Ecology & Evolution 2 (10): 1531‑1540. https://doi.org/10.1038/s41559-018-0667-3
https://openalex.org/W4390206434	https://jppres.com/jppres/pdf/vol12/jppres23.1737_12.2.286.pdf	English	null	Effect of Centella asiatica (L.) Urb. extracts through expression of SIRT1 and Per2 on zebrafish (Danio rerio) larvae insomnia model	Journal of pharmacy & pharmacognosy research	2,024	cc-by	8,148	Abstract Context: Insomnia is difficulty falling asleep or maintaining sleep or sleeping conditions that cannot adequately restore the body and mental state. However, insomnia therapy has not been satisfying for improving the condition. Several alternative therapy options for treating insomnia include Centella asiatica (CA). Many pathways lead to sleep regulation, including circadian rhythms and cellular reactions. Further research is necessary to treat insomnia through circadian rhythms by administering CA with observations on zebrafish models of insomnia. Aims: To evaluate the influence of CA extract administration on deacetylase sirtuin-1 (SIRT1) and period circadian regulator-2 (Per2) expression in zebrafish (Danio rerio) larvae in an insomnia model. Methods: A laboratory experimental study with a randomized control group post-test only was used, with a sample of D. rerio larvae insomnia model that was given treatment with CA ethanolic extract (2.5, 5, and 10 µg/mL). The D. rerio larvae were prepared for reverse transcription-quantitative polymerase chain reaction (RT-qPCR) to evaluate SIRT1 and Per2 levels. Results: This study found a significant effect of CA ethanolic extract on the expression of SIRT1 (p=0.001), with CA extract 10 µg/mL elevating SIRT1 expression in the group of D. rerio larvae with the insomnia model. There was no significant effect of CA extract on Per2 expression in the group of D. rerio larvae with the insomnia model (p=0.051). Conclusions: Administration of CA extract at a concentration of 10 µg/mL can elevate the expression of SIRT1 and have no significant effect on Per2 expression in D. rerio larvae with insomnia model. Conclusions: Administration of CA extract at a concentration of 10 µg/mL can elevate the expression of SIRT1 and have no s in D. rerio larvae with insomnia model. Keywords: Centella asiatica; circadian rhythm; deacetylase sirtuin-1; insomnia; period circadian regulator-2; zebrafish. Keywords: Centella asiatica; circadian rhythm; deacetylase sirtuin-1; insomnia; period circadian regulator-2; zebrafish. © 2024 Journal of Pharmacy & Pharmacognosy Research, 12 (2), 286-295, 2024 ISSN 0719-4250 https://jppres.com © 2024 Journal of Pharmacy & Pharmacognosy Research, 12 (2), 286-295, 2024 ISSN 0719-4250 https://jppres.com © 2024 Journal of Pharmacy & Pharmacognosy Research, 12 (2 Original Article Zamroni Afif1,2, Nurvia Andriani2, Rodhiyan Rakhmatiar2, Shahdevi Kurniawan2, Holipah3, Husnul Khotimah4, Nurdiana4, Irawan Satriotomo5 1Doctoral Program in Medical Science, Faculty of Medicine, Universitas Brawijaya, Malang, Indonesia. 2Department of Neurology Faculty of Medicine, Universitas Brawijaya, Saiful Anwar General Hospital, Malang, Indonesia. 3Department of Public Health, Faculty of Medicine, Universitas Brawijaya, Malang, Indonesia. 4Departement of Pharmacology, Faculty of Medicine, Universitas Brawijaya, Malang, Indonesia. 5Indonesian Neuroscience Institute, Malang, Indonesia. E mail: husnul farmako fk@ub ac id g E-mail: husnul_farmako.fk@ub.ac.id Resumen Contexto: El insomnio es la dificultad para conciliar o mantener el sueño o las condiciones de sueño que no pueden restaurar adecuadamente el cuerpo y el estado mental. Sin embargo, la terapia del insomnio no ha sido satisfactoria para mejorar la condición. Varias opciones de terapia alternativa para tratar el insomnio incluyen Centella asiatica (CA). Muchas vías conducen a la regulación del sueño, incluidos los ritmos circadianos y las reacciones celulares. Es necesario seguir investigando para tratar el insomnio a través de los ritmos circadianos mediante la administración de CA con observaciones en modelos de insomnio con pez cebra. Objetivos: Evaluar la influencia de la administración de extracto de CA en la expresión de la deacetilasa sirtuina-1 (SIRT1) y del regulador circadiano del periodo-2 (Per2) en larvas de pez cebra (Danio rerio) en un modelo de insomnio. Métodos: Se utilizó un estudio experimental de laboratorio con un grupo de control aleatorizado post-test únicamente, con una muestra de larvas de D. rerio modelo de insomnio a las que se administró tratamiento con extracto etanólico de CA (2,5; 5 y 10 µg/mL). Las larvas de D. rerio se prepararon para la reacción en cadena de la polimerasa de transcripción inversa-cuantitativa (RT-qPCR) para evaluar los niveles de SIRT1 y Per2. Resultados: Este estudio encontró un efecto significativo del extracto etanólico de CA sobre la expresión de SIRT1 (p=0,001), con el extracto de CA 10 µg/mL elevando la expresión de SIRT1 en el grupo de larvas de D. rerio con el modelo de insomnio. No hubo un efecto significativo del extracto de CA sobre la expresión de Per2 en el grupo de larvas de D. rerio con el modelo de insomnio (p=0,051). Conclusiones: La administración de extracto de CA a una concentración de 10 µg/mL puede elevar la expresión de SIRT1 y no tener un efecto significativo sobre la expresión de Per2 en larvas de D. rerio con modelo de insomnio. Palabras Clave: Centella asiática; deacetilasa sirtuin-1; insomnio; pez cebra; regulador circadiano del periodo-2; ritmo circadiano. ARTICLE INFO AUTHOR INFO Received: July 5 3, 2023. ORCID: 0000-0002-2736-0941 (ZA) 0000-0002-6143-8239 (H) Accepted: October 30, 2023. 0009-0007-8091-6834 (NA) 0000-0002-2374-4358 (HK) Available Online: December 25, 2023. INTRODUCTION The light treatment used in the preliminary study refers to the research method of Pinheiro-da-Silva et al. (2017), included the negative control group (expo- sure to 12 hours light:dark), the positive control group with 24 hours light, 16 hours light 8 hours dark flash (2 minutes light:2 minutes dark), and 16 hours light 8 hours dark flash (4 minutes light:1 minute dark). This preliminary study obtained the 24-hour light group as an insomnia induction model based on observations of determining the insomnia model using light treat- ment induction of the light and dark phases on days 5, 6, and 7. The research obtained prolonged latency to first or sleep latency in the larval model zebrafish exposed to light for 24 hours. Continuous exposure to light can cause this condition, which aligns with the theory that zebrafish have endogenously controlled circadian rhythm behaviors that light can influence. Light regulation of sleep-wake cycles is related to the regulation of melatonin and the hypocretin/orexin (Hcrtr) system (Afif et al., 2022). Insomnia is the disintegration of molecules that regulate the rhythm of waking and sleeping in the brain. It is critical to distinguish between normal cy- clical and dysregulated rhythms that cause circadian rhythm oscillators to be disrupted. Circadian rhythm deviations can lead to poor rhythms and disruption of cellular core oscillator components. Circadian rhythms are regulated by the core circadian locomo- tor output cycles kaput (CLOCK) gene, which in- cludes three per genes (Per1, Per2, and Per3), CLOCK, ARNTL (also called BMAL1), and two homologous cryptochrome genes (CRY1 and CRY2). These genes control a proportion of the genome. Approximately 10% of all expressed genes are thought to be regulat- ed by clock genes (Qiu et al., 2016). In controlling circadian rhythms, deacetylase sirtuin-1 (SIRT1) is an accessory component of cellular circadian oscillators. It can increase the amplitude of circadian rhythms, influencing melatonin through circadian rhythms (Hardeland, 2021) and period circadian regulator-2 (Per2), a CLOCK gene that regulates circadian rhythms. Per2 and SIRT1 have a reciprocal relation- ship as SIRT1 binds to CLOCK-BMAL1 circadian and promotes deacetylation and degradation of Per2 (Asher et al., 2008). CA acts as a neuroprotector against various neuro- logical disorders. The potency of this species is at- tributed to its antioxidant, anti-inflammatory, anxio- lytic, and anti-stress properties (Gohil et al., 2010). INTRODUCTION CA administration in rats with sleep deprivation for 72 hours significantly improved locomotor activity, anti- anxiety effects, lowered cortisol levels, and improved neuronal inflammation and apoptosis response (Chanana and Kumar, 2017). The asiatic acid (AA) contained in CA can affect the expression of SIRT1 through peroxisome proliferator-activated receptor-γ- coactivator-1 α (PGC-1α) in cells. From one study, it was stated that cells preincubated with AA showed upregulation of SIRT1 and PGC-1α compared to those without AA so that SIRT1 activated NAD+ deacetylat- ed PGC-1α, which binds to the accessory rhythm component of retinoic acid receptor-related orphan receptor α (RORα). RORα responds to signaling to promote BMAL1 and CLOCK. BMAL1 and CLOCK are activated by rhythm-controlling genes PPARα. PPARα can directly activate the expression of REV- ERBα and Per2. BMAL1 and CLOCK activate the expression of the Per2 gene. When Per2 protein is accumulated maximally, it forms a complex with the BMAL1-CLOCK heterodimer, which causes suppres- sion of Per2 transcription. SIRT1 binds CLOCK- BMAL1 and promotes deacetylation and degradation of Per2 (Borrás et al., 2021). Currently, pharmacological therapy for insomnia has several therapeutic targets, one of which is ben- zodiazepines. It works by increasing the activity of the neurotransmitter gamma-aminobutyric acid (GABA) through modulation of the type A GABA receptor complex. However, these drugs have side effects in long-term use (more than four weeks), in- cluding dependence, discontinuation syndrome, psy- chomotor retardation, learning difficulties, memory impairment and potential for cognitive impairment, delirium, accidents/falls, abuse, and a death potential for patients with chronic pulmonary insufficiency or sleep apnea. Consequently, the clinical use of off-label drugs and new drugs that do not target the GABAer- gic system is increasing (Atkin et al., 2018). Adverse long-term effects of insomnia therapy require alterna- tive therapies from natural ingredients that have min- imal side effects. One of which is by developing this study, whose effect of Centella asiatica (L.) Urb. (family Apiaceae) (CA) on zebrafish models of insomnia, with observations primarily through SIRT1 and Per2 in improving sleep activity in zebrafish with models of insomnia that have not been discussed in previous studies. The research on the effect of CA extract on the expression of SIRT1 and Per2 in zebrafish larvae insomnia model used the previous preliminary re- search by Afif et al. (2022), which observed zebrafish exposed to light to induce an insomnia model. Resumen 0009-0005-5353-5134 (RR) 0000-0002-7953-1124 (N) 0000-0002-9351-7505 (SNK) 0000-0003-0519-9509 (IS) Palabras Clave: Centella asiática; deacetilasa sirtuin-1; insomnio; pez cebra; regulador circadiano del periodo-2; ritmo c AUTHOR INFO ORCID: 0000-0002-2736-0941 (ZA) 0000-0002-6143-8239 (H) 0009-0007-8091-6834 (NA) 0000-0002-2374-4358 (HK) 0009-0005-5353-5134 (RR) 0000-0002-7953-1124 (N) 0000-0002-9351-7505 (SNK) 0000-0003-0519-9509 (IS) ARTICLE INFO Received: July 5 3, 2023. Accepted: October 30, 2023. Available Online: December 25, 2023. ARTICLE INFO Received: July 5 3, 2023. Accepted: October 30, 2023. Available Online: December 25, 2023. Afif et al. Effect of Centella asiatica in zebrafish insomnia model Effect of Centella asiatica in zebrafish insomnia model https://jppres.com Induction of insomnia The light exposure used in this study to cause in- somnia conditions in zebrafish larvae refers to the research method of Pinheiro-da-Silva et al. (2017). with modification of light exposure as follows: nega- tive control group with light exposure 12 hours light and 12 hours dark (normal group), and positive con- trol group includes light treatment 24 hours light (24 hours) (Pinheiro-da-Silva et al., 2017). Light is given at 200 lux from 0 dpf until the age of 7 dpf. Modifica- tions were made to the design of lamps and tools used during the study (Afif et al., 2022). Phytochemicals of CA protective effect on CA with experimental animal models of zebrafish larvae (Khotimah et al., 2015). This concentration has been proven optimal to show the pharmacological effect of CA extract. This study aimed to prove that the administration of CA ethanol- ic extract influences SIRT1 and Per2 expression in the zebrafish (Danio rerio) larvae insomnia model. Scientific studies have revealed that CA contains more than 70 phytochemicals. CA is rich in triter- penes, flavonoids, essential oils, alkaloids, and amino acids. CA has diverse and complex components of chemically active compounds in which the main groups include terpenes (monoterpenes, sesquiter- penes, diterpenes, triterpenes, tetraterpenes), phenolic compounds (flavonoids, tannins), alkaloids, carbohy- drates, vitamins, minerals and amino acids (Saba- ragamuwa et al., 2018). INTRODUCTION CA extract concentrations (2.5, 5, and 10 µg/mL) used in this study have been referred to by Khotimah et al. (2015). Meanwhile, the concentration of the eth- anol extract of CA has been widely used in other re- search, for example, the study of Parkinson's models, stunting, and also research to determine the neuro- https://jppres.com J Pharm Pharmacogn Res (2024) 12(2): 287 Effect of Centella asiatica in zebrafish insomnia model Effect of Centella asiatica in zebrafish insomnia model Afif et al. Research design The Faculty of Medicine at the University of Brawijaya's Health Research Ethics Committee has approved this project. Ethics Permit No: 237/EC/KEPK/11/2022. The research design used in this study was a true laboratory experiment with a randomized control group post-test approach. Zebrafish larvae were divided into 5 groups: control negative group (exposed to 12 hours of light and 12 hours of darkness); control positive group or insom- nia group (24 hours light exposure); insomnia group with 2.5 µg/mL of CA (CA 2.5 µg/mL); insomnia group with 5 µg/mL of CA (CA 5 µg/mL), and in- somnia group with 10 µg/mL of CA (CA 10 µg/mL). Exposure to CA extract The administration of CA ethanol extract with concentrations of 2.5, 5, and 10 μg/mL (Khotimah et al., 2015) was sequentially introduced into the treat- ment group with 24 hours of light exposure group. CA extract was induced once a day from the third day to the seventh day of dpf (day post-fertilization). Maintenance of zebrafish embryos CA was mashed with a blender until smooth. The dry sample weighed 100 g and was put into a 1-liter Erlenmeyer glass. Then, it was soaked with 98% etha- nol to a volume of 900 mL. Shake until thoroughly mixed (± 30 minutes). Let stand one night until set- tled. The top layer of ethanol (solvent) mixture with the active substance has been mixed (can be filtered using filter paper). The soaking process was carried out thrice, followed by evaporation. The solution from the extraction process was put into a 1-liter evaporation flask. The evaporating flask was installed on the evaporator with the water bath set to 70°C. Wild-type zebrafish at age 0–7 days post fertiliza- tion (dpf) was used in this study, obtained from the Reproductive Laboratory of the Faculty of Fisheries and Marine Sciences, Universitas Brawijaya. Embryos were maintained at 28 ± 10°C. Maintenance of zebrafish embryos with dark:light condition for 12:12 hours and feeding plankton is given 3 times a day, conductivity 350-600 S and salinity 0-0.6 ppt (parts per thousand) zebrafish larvae were taken on the 7 dpf and placed in an ice pack. After the zebrafish larvae were unconscious (showing no spontaneous movements), the bottles were tightly closed and stored at -80°C and tissue analysis was performed. RNA was isolated using the MEGAshortscript T7 kit (Ambion. Inc., Austin, Texas). Drying CA extract process The stems and leaves that were above the ground were washed thoroughly and then cut into pieces of 1-2 cm to speed up the drying process. Next, put the oven at 40°C to dry (free of water content). Geographical location of CA specimen collection C. asiatica specimen collection was obtained and identified by Laboratory Materia Medica Batu with coordinate points -7.867086,112.519319. https://jppres.com J Pharm Pharmacogn Res (2024) 12(2): 288 https://jppres.com used in this study were for Per2: Forward primer (5’– 3’) ACGAGGACAAGCCAGAGGAACG, reverse primer (5’-3’) GCACTGGCTGGTGATGGAGA (Ren et al., 2017). most optimum concentration of 10 µg/mL compared to others. The lowest of increased SIRT1 expression was at a concentration of 2.5 µg/mL compared to others. In the negative control, i.e., in normal zebrafish larvae (light treatment 12 hours light: 12 hours dark), the average SIRT1 expression was close to the mean SIRT1 expression in zebrafish larvae with insomnia model given exposure to CA extract 10 µg/mL, which showed that the higher the CA extract concentration, the more it can increase SIRT1 expression in insomnia model zebrafish larvae. In the positive control, the zebrafish larvae that were exposed to light for 24 hours, the mean expression results showed higher results than the insomnia model zebrafish larvae that were exposed to the lowest concentration of CA extract, 2.5 µg/mL. RESULTS The results of quantitative SIRT1 expression data were tested for normality and homogeneity. Normality testing was carried out using the Shapiro- Wilk test. The normality test is fulfilled if p>0.05. Using SPSS software, the normality test results are obtained as in Table 1. Statistical analysis The research data were analyzed using IBM SPSS Statistics 27 software and expressed in terms of mean ± standard deviation (mean ± SD). The normality test used the Shapiro-Wilk Test of normality, and the ho- mogeneity test used the Homogeneity of Variance test; then, if the data was normally distributed and homogeneous, it was continued with the One-Way ANOVA difference test. Significant difference data with p<0.05 was followed by Tukey's post-hoc test. Examination of real-time quantitative chain reaction (RT-qPCR) BDNF was measured on samples taken from the tissue of zebrafish larvae by the qRT-PCR method. Primers used in this study for SIRT1: forward primer CAA GGA AAT CTA CCC CGG ACA GT reverse primer CAG TGT GTC GAT ATT CTG CGT GT. Per2 was measured on samples taken from the tissue of zebrafish larvae by the qRT-PCR method. The primers https://jppres.com J Pharm Pharmacogn Res (2024) 12(2): 288 Afif et al. Effect of Centella asiatica in zebrafish insomnia model Figure 1. SIRT1 in zebrafish larvae model of insomnia after exposure to CA extract. Data are expressed as mean ± SD (n = 15).p<0.05; p<0.01; p<0.001. One-Way ANOVA difference test followed by Tukey's post-hoc test. Figure 1. SIRT1 in zebrafish larvae model of insomnia after exposure to CA extract. Data are expressed as mean ± SD (n = 15).p<0.05; p<0.01; p<0.001. One-Way ANOVA difference test followed by Tukey's post-hoc test. Table 1. Quantitative data normality test and homogeneity test results for SIRT1 expression in zebrafish larvae insomnia model. Variable Group p-value normality test Description P-value homogeneity test Description SIRT1 Control (-) 0.402 Normal 0.064 Homogeneous Control (+) 0.205 Normal CA 2.5 µg/mL 0.899 Normal CA 5 µg/mL 0.586 Normal CA 10 µg/mL 0.411 Normal Table 1. Quantitative data normality test and homogeneity test results for SIRT1 expression in zebrafish larvae insomnia model. used in this study were for Per2: Forward primer (5’– 3’) ACGAGGACAAGCCAGAGGAACG, reverse primer (5’-3’) GCACTGGCTGGTGATGGAGA (Ren et al., 2017). J Pharm Pharmacogn Res (2024) 12(2): 289 SIRT1 examination results with RT-qPCR Treatment p-value Description Control (-) Control (+) 0.549 Not significantly different CA 2.5 µg/mL 0.002* Significantly different CA 5 µg/mL 0.031* Significantly different CA 10 µg/mL 0.956 Not significantly different Control (+) CA 2.5 µg/mL 0.017 Significantly different CA 5 µg/mL 0.316 Not significantly different CA 10 µg/mL 0.897 Not significantly different CA 2.5 µg/mL CA 5 µg/mL 0.349 Not significantly different CA 10 µg/mL 0.005* Significantly different CA 5 µg/mL CA 10 µg/mL 0.089 Not significantly different Figure 2. Tukey post-hoc test results effect of treatment on SIRT1 expression. Data are expressed as mean ± SD (n = 15).p<0.05; p<0.01; p<0.001. One- Way ANOVA followed by Tukey's post- hoc test. Table 3. Tukey post-hoc test results effect of treatment on SIRT1 expression. Treatment p-value Description Control (-) Control (+) 0.549 Not significantly different CA 2.5 µg/mL 0.002 Significantly different CA 5 µg/mL 0.031* Significantly different CA 10 µg/mL 0.956 Not significantly different Control (+) CA 2.5 µg/mL 0.017 Significantly different CA 5 µg/mL 0.316 Not significantly different CA 10 µg/mL 0.897 Not significantly different CA 2.5 µg/mL CA 5 µg/mL 0.349 Not significantly different CA 10 µg/mL 0.005* Significantly different CA 5 µg/mL CA 10 µg/mL 0.089 Not significantly different Table 3. Tukey post-hoc test results effect of treatment on SIRT1 expression. Figure 2. Tukey post-hoc test results effect of treatment on SIRT1 expression. Data are expressed as mean ± SD (n = 15).p<0.05; p<0.01; *p<0.001. One- Way ANOVA followed by Tukey's post- hoc test. Figure 2. Tukey post-hoc test results effect of treatment on SIRT1 expression. Data are expressed as mean ± SD (n = 15).p<0.05; p<0.01; p<0.001. One- Way ANOVA followed by Tukey's post- hoc test. Figure 2. Tukey post-hoc test results effect of treatment on SIRT1 expression. insomnia model. In addition, a posthoc Tukey test was carried out on the expression of SIRT in zebrafish larvae with the insomnia model to find out the differences in each treatment (Table 3, Fig. 2). In Tukey's posthoc test, the group exposed to CA at 2.5 µg/mL concentrations and 5 µg/mL significantly differed from the negative control group. In comparison, exposure to CA at 10 µg/mL concentration was not significantly different. It indicates that the values were close to normal. data was homogeneous. https://jppres.com SIRT1 examination results with RT-qPCR The results of the SIRT1 examination using the RT- qPCR method in the form of mean SIRT1 in the zebrafish larvae model of insomnia after exposure to CA extract are presented in Fig. 1. From Tables 1 and 2, it can be seen that the SIRT1 variable data in each group was normally distributed. Therefore, the following variant homogeneity test was performed with the Levene test. The homogeneity test results were fulfilled with p>0.05. Statistical analysis results showed that the SIRT1 expression variable The diagram shows that the group exposed to CA extract exhibited an increased SIRT1 expression in insomnia model zebrafish larvae along with an increase in concentrations of CA extract, with the J Pharm Pharmacogn Res (2024) 12(2): 289 Afif et al. Effect of Centella asiatica in zebrafish insomnia model Table 2. Average SIRT1 expression in insomnia model zebrafish larvae. No. Treatment SIRT1 expression p-value 1 Control (-) 30.39 ± 0.38 0.001 2 Control (+) 29.9 ± 0.34 3 CA 2.5 µg/mL 28.65 ± 0.11 4 CA 5 µg/mL 29.27 ± 0.66 5 CA 10 µg/mL 30.18 ± 0.14 Data are expressed as mean ± SD (n = 15). Table 2. Average SIRT1 expression in insomnia model zebrafish larvae. Data are expressed as mean ± SD (n = 15). Table 3. Tukey post-hoc test results effect of treatment on SIRT1 expression. Treatment p-value Description Control (-) Control (+) 0.549 Not significantly different CA 2.5 µg/mL 0.002* Significantly different CA 5 µg/mL 0.031* Significantly different CA 10 µg/mL 0.956 Not significantly different Control (+) CA 2.5 µg/mL 0.017 Significantly different CA 5 µg/mL 0.316 Not significantly different CA 10 µg/mL 0.897 Not significantly different CA 2.5 µg/mL CA 5 µg/mL 0.349 Not significantly different CA 10 µg/mL 0.005* Significantly different CA 5 µg/mL CA 10 µg/mL 0.089 Not significantly different Figure 2. Tukey post-hoc test results effect of treatment on SIRT1 expression. Data are expressed as mean ± SD (n = 15).p<0.05; p<0.01; *p<0.001. One- Way ANOVA followed by Tukey's post- hoc test. was homogeneous. Based on the results of the ality and homogeneity tests of the data it can be insomnia model. In addition, a posthoc was carried out on the expression of SIRT i Table 3. Tukey post-hoc test results effect of treatment on SIRT1 expression. J Pharm Pharmacogn Res (2024) 12(2): 290 SIRT1 examination results with RT-qPCR Based on the results of the normality and homogeneity tests of the data, it can be concluded that the statistical testing process was carried out using a parametric statistical approach using the One-way ANOVA test because the data was normally distributed and homogeneous. insomnia model. In addition, a posthoc Tukey test was carried out on the expression of SIRT in zebrafish larvae with the insomnia model to find out the differences in each treatment (Table 3, Fig. 2). In Tukey's posthoc test, the group exposed to CA at 2.5 µg/mL concentrations and 5 µg/mL significantly differed from the negative control group. In comparison, exposure to CA at 10 µg/mL concentration was not significantly different. It indicates that the values were close to normal. Based on the analysis results using the One-way ANOVA test, a p<0.001 was obtained. Thus, this test obtained a significant effect of exposure to CA on SIRT1 expression in zebrafish larvae with the J Pharm Pharmacogn Res (2024) 12(2): 290 Afif et al. Effect of Centella asiatica in zebrafish insomnia model Figure 3. Mean Per2 expression in zebrafish larvae model of insomnia after exposure to CA extract. Data are expressed as mean ± SD (n = 15).p<0.05; p<0.01; p<0.001. One-Way ANOVA difference test followed by Tukey's post-hoc test. Table 4. Quantitative data normality test results for Per2 Expression in zebrafish larvae model of insomnia. Variable Group p-value normality test Description P-value homogeneity test Description Per2 Control (-) 0.205 Normal 0.001 Inhomogeneous Control (+) 0.105 Normal CA 2.5 µg/mL 0.6 Normal CA 5 µg/mL 0.11 Normal CA 10 µg/mL 0.025 Abnormal Table 4. Quantitative data normality test results for Per2 Expression in zebrafish larvae model of insomnia. Table 5. Mean Per2 expression in zebrafish larvae insomnia model. No Treatment Mean ± SD p-value 1 Control (-) 37.91 ± 0.53 0.051 2 Control (+) 30.68 ± 0.39 3 CA 2.5 µg/mL 28.55 ± 0.41 4 CA 5 µg/mL 28.23 ± 0.11 5 CA 10 µg/mL 30.38 ± 4.36 This shows that exposure to CA extract with a concentration of 10 µg/mL was the most optimum in increasing SIRT1 expression in zebrafish larvae with the insomnia model. From the quantitative data of Per2 expression, normality, and homogeneity tests were conducted. Normality testing was performed using the Shapiro- Wilk test. The assumption of normality if p>0.05. Using the help of SPSS software, the normality test results are showed in Table 4. https://jppres.com J Pharm Pharmacogn Res (2024) 12(2): 291 Per2 examination results with RT-qPCR Tukey post-hoc test results effect of treatment on expression of Per2. Data are expressed as mean ± SD (n = 15).p<0.05; p<0.01; p<0.001. One-Way ANOVA followed by Tukey's post-hoc test. Figure 4. Tukey post-hoc test results effect of treatment on expression of Per2. Data are expressed as mean ± SD (n = 15).p<0.05; p<0.01; p<0.001. One-Way ANOVA followed by Tukey's post-hoc test. Kruskall Wallis because the data was not normally distributed and was inhomogeneous (Table 6). autosomal dominance. Other evidence supporting the genetic basis of insomnia is research reports of polymorphisms in circadian genes such as hPer3, arylakylamin N-acetyltransferase, human leukocyte antigen, and Jam, which are associated with diurnal preferences and insomnia (Riemann et al., 2010). Based on the Kruskall Wallis analysis results, a p- value was 0.051 (p>0.05). Therefore, it can be concluded that there is no significant effect of the concentration of CA extract on the expression of Per2 in insomnia model zebrafish. Thus, this test shows that there was no significant effect of concentration of CA extract on Per2 expression, or there was no significant difference in Per2 levels due to treatment (Fig. 4). Centella asiatica (CA) can improve the regulation of SIRT1 gene expression From the results of this study, the mean expression of SIRT1 in the insomnia model of zebrafish larvae in the exposure group CA was found. It can be explained that the expression level of the SIRT1 gene increased with increasing concentrations of CA extract as evidenced by the results of Tukey's posthoc test, i.e., the exposure group of CA with concentration 2.5 µg/mL and 5 µg/mL significantly different from the negative control group. In comparison, exposure to a CA concentration of 10 µg/mL did not differ significantly, indicating that the values were close to normal. This shows that exposure to CA extract with a concentration of 10 µg/mL is the most optimum in increasing SIRT1 expression in zebrafish larvae with the insomnia model. Per2 examination results with RT-qPCR The results of the Per2 expression examination using the RT-qPCR method to obtain data on average Per2 expression in zebrafish larvae model insomnia after exposure to CA extract are presented in the following Fig. 3. Table 5 shows that the Per2 variable data in each group is normally distributed, except for the 10 µg/mL group. Furthermore, the variance homogeneity test was carried out using the Levene test. The assumption of homogeneity of variance is fulfilled if p>0.05. The results of statistical analysis show that the Per2 variable data is inhomogeneous. Based on the results of the normality and homogeneity tests of the data, it can be concluded that the statistical testing process was carried out using a non-parametric statistical approach using Based on the figure above, it can be seen that the group exposed to 10 µg/mL CA extract had a higher average Per2 expression value compared to other concentrations. However, the highest Per2 expression average value was in the negative control. The lowest mean value of Per2 expression was in the group exposed to CA extract of 5 µg/mL. J Pharm Pharmacogn Res (2024) 12(2): 291 Afif et al. Effect of Centella asiatica in zebrafish insomnia model Table 6. Tukey post-hoc test results effect of treatment on expression of Per2. Treatment p-value Description CA 2.5 µg/mL Control (+) 0.248 Not significantly different Control (-) 0.248 Not significantly different CA 5 µg/mL CA 10 µg/mL 1.000 Not significantly different CA 2.5 µg/mL 1.000 Not significantly different Control (+) 0.248 Not significantly different Control (-) 0.248 Not significantly different CA 10 µg/mL CA 2.5 µg/mL 1.000 Not significantly different Control (+) 1.000 Not significantly different Control (-) 0.248 Not significantly different Figure 4. Tukey post-hoc test results effect of treatment on expression of Per2. Data are expressed as mean ± SD (n = 15).p<0.05; p<0.01; *p<0.001. One-Way ANOVA followed by Tukey's post-hoc test. Table 6. Tukey post-hoc test results effect of treatment on expression of Per2. Treatment p-value Description CA 2.5 µg/mL Control (+) 0.248 Not significantly different Control (-) 0.248 Not significantly different CA 5 µg/mL CA 10 µg/mL 1.000 Not significantly different CA 2.5 µg/mL 1.000 Not significantly different Control (+) 0.248 Not significantly different Control (-) 0.248 Not significantly different CA 10 µg/mL CA 2.5 µg/mL 1.000 Not significantly different Control (+) 1.000 Not significantly different Control (-) 0.248 Not significantly different Figure 4. https://jppres.com DISCUSSION The role of SIRT1 in the oscillation process provides a systemic role of sirtuins throughout the body, influencing a large number of functions controlled by circadian rhythms, ranging from ups and downs of regulation of genes and their respective products to regulation of metabolism, endocrine, excitatory, immunological, and many other physiological functions. This is a crucial aspect of circadian gene expression concerning epigenetic modulation by the interaction of core components and accessory oscillators, chromatin remodeling, and modulation of noncoding RNA. Therefore, the increase in the positive control can affect the ups and downs of the regulation of genes and their respective products to regulate the metabolism, endocrine, excitatory, immunological, and many physiological functions of the zebrafish larvae (Mendelsohn and Larrick, 2017). Based on other studies stated that the circadian system is a complex multioscillator machine that is useful in avoiding many diseases and disorders. The circadian amplitude decreases with age in central and peripheral oscillators, so increased amplitude through SIRT1 is of high value for healthy aging supported mitochondrial function by SIRT1, which can contrib- ute to the avoidance of excessive oxidative stress. Sirtuins from various studies have relevance to mito- chondrial function, which is useful in therapeutic aspects related to its role in metabolic homeostasis (Yamazaki et al., 2002). Kazimi and Cahill (199) found that environmental conditions affect the circadian rhythm of zebrafish larvae through the influence of light. In this study, it was stated that the optimum temperature that best influences circadian rhythms, one of which is an increase in melatonin, is a temperature of 28.58°C. These environmental conditions may be affecting SIRT1 expression in positive controls (Kazimi and Cahill, 1999). In addition, SIRT1 is also known to be involved in mitochondrial proliferation, especially in central nervous system cells. Neuron cells subjected to oxida- tive stress, excitotoxic, or other cell stress, mitophagy can cause a reduction in mitochondria in the periph- eral parts of the cell, leading to decreased connectivity leading to loss of function, as in neurodegenerative diseases such as Alzheimer's disease and occurs to some degree in normal aging. Thus, mitochondrial proliferation by SIRT1 may maintain functions, in- cluding regulation of PGC-1α and PPARγ factors, as both influence circadian rhythms through antioxidant effects associated with traumatic or ischemic brain injury and damage by mitochondrial toxins, such as doxorubicin. In ischemic and traumatic brain injury, the neuroprotective function of SIRT1 is anti- inflammatory and antiapoptotic. DISCUSSION Insomnia is the result of disruption of the endogenous circadian system. For example, many physiological markers of the circadian phase exhibit sleep-wake inhibiting patterns. There is also evidence that individuals with insomnia have hypersensitivity to nocturnal melatonin suppression due to bright light. They also have decreased oscillator sensitivity to alignment and long circadian cycles. Furthermore, the duration and timing of environmental light and dark exposure may be essential in expressing the insomnia phenotype. Insomnia has also been reported in cases of mild head trauma and has a genetic basis. Some cases of this syndrome can be hereditary due to This study found that SIRT1 gene expression increased with increasing concentrations of CA J Pharm Pharmacogn Res (2024) 12(2): 292 https://jppres.com Afif et al. Effect of Centella asiatica in zebrafish insomnia model extract. In theory, the SIRT1 cascade through the active substance of CA extract, and the triterpene content, asiatic acid (AA), can increase SIRT1 gene expression. AA content in CA can influence SIRT1 expression through PGC-1α in cells; preincubation of SIRT1 with AA shows upregulation of SIRT1 and PGC-1α compared to those without AA (Xu et al., 2013). SIRT1 interacts with the core oscillator, which increases the amplitude of circadian oscillators in the center and peripherally. SIRT1 expression in the central circadian rhythm through the expression of CLOCK (circadian locomotor output cycles kaput), which depends on the NAD+ sirtuin substrate which oscillates with the expression of the NAD enzyme, nicotinamide phosphoribosyltransferase (NAMPT) through the promoter of the BMAL1 core oscillator component (brain and muscle aryl hydrocarbon receptor nuclear translocator-like 1), which binds CLOCK. NAD+-activated SIRT1 deacetylates PGC-1α, which binds to the accessory clock component RORα (retinoic acid receptor-related orphan receptor α) (Mendelsohn and Larrick, 2017). expression results showed higher results than the insomnia model zebrafish larvae that were exposed to the lowest concentration of CA extract, 2.5 µg/mL. Theoretically, SIRT1 decreases in insomnia conditions, but various regulatory conditions can influence the increase and decrease of SIRT1. SIRT1 is an accessory component of a cellular circadian oscillator that is part of a circadian system whose complexity, aside from the main clock cycle activity of the hypothalamus, is the suprachiasmatic nucleus (SCN). However, it also comprises a peripheral oscillator in every tissue and countless nucleated cells. DISCUSSION Research publica- tions mention that neuroprotective effects can be me- diated by SIRT1 (Hernández-Jiménez et al., 2013). https://jppres.com Centella asiatica (CA) did not affect the regulation of Per2 gene expression This study found that the average Per2 expression in insomnia model zebrafish larvae in the CA exposure group showed no significant effect of concentration of CA extracts on Per2 expression. This is evidenced by the analysis using Kruskall Wallis, where there was no significant effect of the concentration of CA extract on Per2 expression. The content of AA in CA is an indirect pathway Per2 from the AA, and SIRT1 cascades through BMAL1 and CLOCK. BMAL1 and CLOCK activate the expression of the Per2 gene. When Per2 protein accumulates maximally, it will form a complex with the BMAL1-CLOCK heterodimer, which causes suppression of Per2 transcription. SIRT1 binds to In the positive control, the zebrafish larvae that were exposed to light for 24 hours, the mean J Pharm Pharmacogn Res (2024) 12(2): 293 https://jppres.com Effect of Centella asiatica in zebrafish insomnia model Afif et al. CLOCK-BMAL1 circadian and promotes deacetylation and degradation of Per2 (Xu et al., 2013). results of this study, which state that there is no significant effect of concentrations of CA extract on the expression of Per2 due to the influence of environmental factors (light, food) with nuclear receptors in cells related to various metabolisms outside the circadian rhythm including the dopaminergic system, inflammation, glucose metabolism, temperature, and fatty acid metabolism (Kim et al., 2019). Other research mentions that the regulation of wake and sleep rhythms in the brain can also be regulated by remodeling neuronal connectivity, either strengthening or weakening synaptic connections. The study mentioned that Per2 is a CLOCK gene that regulates circadian rhythms in entering light and dark information in vesicular glutamate transporter 1 (vGLUT1) in synaptic vesicles. Light can impact through Per2, so changes in light and dark, such as those experienced in jet lag and shift work, can affect behavior (Cox and Takahashi, 2019). This study has a limitation in that the gene expression preparation sample uses the whole body or whole body of zebrafish larvae, so it is not specific to target organs or tissues such as SIRT1 and Per2 gene expression in the brain of zebrafish larvae. This research also does not explore specific active compounds from CA extract that improve sleep activity in insomnia events. CONCLUSION Administration of Centella asiatica extract at a concentration of 10 µg/mL increased the expression of SIRT1 and had no effect on the expression of Per2 in zebrafish larvae of the insomnia model. Synchronization of circadian oscillators with environmental inputs and outputs of physiological circadian rhythm pathways requires complex transcriptional feedback loops. Per2 mediates the main output of the molecular oscillator. Direct interaction with nuclear receptors (NR) homo or heterodimers can affect promoters of the appropriate target genes and metabolic or physiological processes. Per2 activity can be modulated by input (e.g., light or food) to the circadian oscillator (Asher et al., 2008). Protein-protein interactions in circadian rhythms require coupling these proteins to circadian rhythm oscillators in nuclear receptor target genes. The high flexibility of this interaction allows Per2 to act as a coactivator or corepressor, depending on the balance of nuclear receptor activity. Several studies stated that the potential for Per2 interaction and its regulation with post-translational modification requires further experimentation. These interactions may form the basis of many of Per2's additional functions in metabolism and the brain. Other core circadian oscillators can interact with various other proteins and regulators. Per2 can directly interact with more transcriptional regulators (Di Rosa et al., 2015). https://jppres.com Centella asiatica (CA) did not affect the regulation of Per2 gene expression Therefore, further research is needed to see the specific location of the SIRT1 and Per2 genes in the organs or brain tissue of zebrafish larvae in insomnia models, and it is necessary to measure levels of SIRT1 and Per2 and other genes involved in insomnia conditions. Further research is needed to determine which compounds from CA are the most effective for improving sleep activity in insomnia conditions. The mechanism of circadian rhythm generation relies on feedback loops that involve transcription factors positively and negatively. BMAL1 and CLOCK activate the expression of the Per and cryptochrome (Cry) genes when the Per and Cry proteins accumulate maximally, they form complexes with BMAL1-CLOCK heterodimers that cause suppression of their own gene transcription. Per2 and SIRT1 have a reciprocal relationship. It is said that SIRT1 binds CLOCK-BMAL1 circadianly and promotes deacetylation and degradation of Per2. The potential interaction of Per2 with other proteins is regulated by the accumulation of Per2 proteins during the circadian cycle (Asher et al., 2008). ACKNOWLEDGMENTS The authors thank the Faculty of Medicine Universitas Brawijaya for funding this research with research grant No. 3969.2/58/UN10.f08/PN/2021. The authors also would like to thank Prof. Dr. Moch and Istiadjid Eddy Santoso, Sp.S, Sp.BS(K), M.Hum, for their suggestions during the research. discovery. Pharmacol Rev 70(2): 197–245. https://doi.org/10.1124/PR.117.014381 discovery. Pharmacol Rev 70(2): 197–245 https://doi.org/10.1124/PR.117.014381 Borrás S, Martínez-Solís I, Ríos JL (2021) Medicinal plants for insomnia related to anxiety: An updated review. Planta Med 87(10–11): 738–753. https://doi.org/10.1055/A-1510-9826 Mendelsohn AR, Larrick JW (2017) The NAD+/PARP1/SIRT1 axis in aging. Rejuvenation Res 20(3): 244–247. https://doi.org/10.1089/REJ.2017.1980 Chanana P, Kumar A (2017) Further investigations on the neuroprotective potential of Centella asiatica against sleep deprivation induced anxiety like behaviour: Possible implications of mitoprotective and anti-stress pathways. J Sleep Disord Treat Care 6: 2. https://doi.org/10.4172/2325- 9639.1000193 Pinheiro-da-silva J, Silva PF, Nogueira MB, Luchiari AC (2017) Sleep deprivation effects on object discrimination task in zebrafish (Danio rerio) sleep deprivation effects on object discrimination task in zebrafish (Danio rerio). Anim Cogn 20: 159–169. https://doi.org/10.1007/s10071-016-1034-x Cox KH, Takahashi JS (2019) Circadian clock genes and the transcriptional architecture of the clock mechanism. J Mol Endocrinol 63(4): R93–R102. https://doi.org/10.1530/JME-19- 0153 Qiu MH, Yao QL, Vetrivelan R, Chen MC, Lu J (2016) Nigrostriatal dopamine acting on Globus pallidus regulates sleep. Cereb Cortex 26(4): 1430–1439. https://doi.org/10.1093/CERCOR/BHU241 Di Rosa V, Frigato E, López-Olmeda JF, Sánchez-Vázquez FJ, Bertolucci C (2015) The light wavelength affects the ontogeny of clock gene expression and activity rhythms in zebrafish larvae. PLoS One 10(7): e0132235. https://doi.org/10.1371/journal.pone.0132235 Ren DL, Ji C, Wang XB, Wang H, Hu B (2017) Endogenous melatonin promotes rhythmic recruitment of neutrophils toward an injury in zebrafish. Sci Rep 7: 4696. https://doi.org/10.1038/s41598-017-05074-w Gohil, KJ, Patel, JA, Gajjar, AK (2010) Pharmacological review on Centella asiatica: A potential herbal cure-all. Indian J Pharm Sci 72(5): 546–556. https://doi.org/10.4103/0250-474X.78519 Riemann D, Spiegelhalder K, Feige B, Voderholzer U, Berger M, Perlis M, Nissen C (2010) The hyperarousal model of insomnia: A review of the concept and its evidence. Sleep Med Rev 14(1): 19–31. https://doi.org/10.1016/J.SMRV.2009.04.002 Hernández-Jiménez M, Hurtado O, Cuartero MI, Ballesteros I, Moraga A, Pradillo JM, McBurney MW, Lizasoain I, Moro MA (2013) Silent information regulator 1 protects the brain against cerebral ischemic damage. Stroke 44(8): 2333–2337. https://doi.org/10.1161/STROKEAHA.113.001715 Sabaragamuwa R, Perera CO, Fedrizzi B (2018) Centella asiatica (Gotu kola) as a neuroprotectant and its potential role in healthy ageing. Trends Food Sci Technol 79: 88–97. https://doi.org/10.1016/J.TIFS.2018.07.024 Kazimi N, Cahill GM (1999) Development of a circadian melatonin rhythm in embryonic zebrafish. Dev Brain Res 117(1): 47–52. https://doi.org/10.1016/S0165-3806(99)00096-6 Xu CL, Qu R, Zhang J, Li LF, Ma SP (2013) Neuroprotective effects of madecassoside in early stage of Parkinson’s disease induced by MPTP in rats. Fitoterapia 90: 112–118. discovery. Pharmacol Rev 70(2): 197–245. https://doi.org/10.1124/PR.117.014381 https://doi.org/10.1016/J.FITOTE.2013.07.009 Khotimah H, Ali M, Sumitro SB, Widodo MA (2015) Decreasing α- synuclein aggregation by methanolic extract of Centella asiatica in zebrafish Parkinson’s model. Asian Pac J Trop Biomed 5(11): 948–954. https://doi.org/10.1016/J.APJTB.2015.07.024 Yamazaki S, Straume M, Tei H, Sakaki Y, Menaker M, Block GD (2002) Effects of aging on central and peripheral mammalian clocks. Proc Nat Acad Sci USA 99(16): 10801–10806. https://doi.org/10.1073/PNAS.152318499 Kim DW, Chang C, Chen X, Doran AC, Gaudreault F, Wager T, AUTHOR CONTRIBUTION: Contribution Afif Z Andriani N Rakhmatiar R Holipah Kurniawan S Khotimah H Nurdiana Satriotomo I Concepts or ideas x x x x x x x x Design x x x x x x Definition of intellectual content x x x Literature search x x x x x x x x Experimental studies x x x x Data acquisition x x Data analysis x x x x Statistical analysis x x x x x Manuscript preparation x x x x Manuscript editing x x x x Manuscript review x x x x x x x x Citation Format: Afif Z, Andriani N, Rakhmatiar R, Holipah, Kurniawan S, Khotimah H, Nurdiana, Satriotomo I (2024) Effect of Centella asiatica (L.) Urb. extracts through expression of SIRT1 and Per2 on zebrafish (Danio rerio) larvae insomnia model. J Pharm Pharmacogn Res 12(2): 286–295. https://doi.org/10.56499/jppres23.1737_12.2.286 Publisher’s Note: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. Open Access: This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/ licenses/by/4.0/), which permits use, duplication, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made. discovery. Pharmacol Rev 70(2): 197–245. https://doi.org/10.1124/PR.117.014381 AUTHOR CONTRIBUTION: Contribution Afif Z Andriani N Rakhmatiar R Holipah Kurniawan S Khotimah H Nurdiana Satriotomo I Concepts or ideas x x x x x x x x Design x x x x x x Definition of intellectual content x x x Literature search x x x x x x x x Experimental studies x x x x Data acquisition x x Data analysis x x x x Statistical analysis x x x x x Manuscript preparation x x x x Manuscript editing x x x x Manuscript review x x x x x x x x Citation Format: Afif Z, Andriani N, Rakhmatiar R, Holipah, Kurniawan S, Khotimah H, Nurdiana, Satriotomo I (2024) Effect of Centella asiatica (L.) Urb. extracts through expression of SIRT1 and Per2 on zebrafish (Danio rerio) larvae insomnia model. J Pharm Pharmacogn Res 12(2): 286–295. https://doi.org/10.56499/jppres23.1737_12.2.286 Publisher’s Note: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. Open Access: This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/ licenses/by/4.0/), which permits use, duplication, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made. Citation Format: Afif Z, Andriani N, Rakhmatiar R, Holipah, Kurniawan S, Khotimah H, Nurdiana, Satriotomo I (2024) Effect of Centella asiatica (L.) Urb. extracts through expression of SIRT1 and Per2 on zebrafish (Danio rerio) larvae insomnia model. J Pharm Pharmacogn Res 12(2): 286–295. https://doi.org/10.56499/jppres23.1737_12.2.286 Publisher’s Note: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the bli h th dit d th i A d t th t b l t d i thi ti l l i th t b d b it f t i t t d d d Citation Format: Afif Z, Andriani N, Rakhmatiar R, Holipah, Kurniawan S, Khotimah H, Nurdiana, Satriotomo I (2024) Effect of Centella asiatica (L.) Urb. REFERENCES Afif Z, Santoso MIE, Khotimah H, Satriotomo I, Widjajanto E, Rahayu M, Kurniawan SN, Iskandar DS, Hakimah A, Azizah S, Andriani N, Agustina K (2022) Light exposure effects on inactive state duration and sleep latency in zebrafish (Danio rerio) larvae insomnia model. Malang Neurol J 8(2): 129–134. https://doi.org/10.21776/ub.mnj.2022.008.02.11 Asher G, Gatfield D, Stratmann M, Reinke H, Dibner C, Kreppel F, Mostoslavsky R, Alt FW, Schibler U (2008) SIRT1 regulates circadian clock gene expression through PER2 deacetylation. Cell 134(2): 317–328. https://doi.org/10.1016/J.CELL.2008.06.050 https://doi.org/10.1016/J.CELL.2008.06.050 The interaction of Per2 with various proteins and circadian rhythm oscillator regulators supports the Atkin T, Comai S, Gobbi G (2018) Drugs for insomnia beyond benzodiazepines: Pharmacology, clinical applications, and J Pharm Pharmacogn Res (2024) 12(2): 294 Afif et al. Effect of Centella asiatica in zebrafish insomnia model DeMarco GJ, Kim JK (2019) Systems approach reveals photosensitivity and PER2 level as determinants of clock- modulator efficacy. Mol Syst Biol 15(7): e8838. https://doi.org/10.15252/MSB.20198838 DeMarco GJ, Kim JK (2019) Systems approach reveals photosensitivity and PER2 level as determinants of clock- modulator efficacy. Mol Syst Biol 15(7): e8838. https://doi.org/10.15252/MSB.20198838 discovery. Pharmacol Rev https://doi.org/10.1124/PR.117.014381 discovery. Pharmacol Rev 70(2): 197–245. https://doi.org/10.1124/PR.117.014381 extracts through expression of SIRT1 and Per2 on zebrafish (Danio rerio) larvae insomnia model. J Pharm Pharmacogn Res 12(2): 286–295. https://doi.org/10.56499/jppres23.1737_12.2.286 Citation Format: Afif Z, Andriani N, Rakhmatiar R, Holipah, Kurniawan S, Khotimah H, Nurdiana, Satriotomo I (2024) Effect of Centella asiatica (L.) Urb. extracts through expression of SIRT1 and Per2 on zebrafish (Danio rerio) larvae insomnia model. J Pharm Pharmacogn Res 12(2): 286–295. https://doi.org/10.56499/jppres23.1737_12.2.286 Publisher’s Note: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. Open Access: This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/ licenses/by/4.0/), which permits use, duplication, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made Publisher’s Note: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. Open Access: This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/ licenses/by/4.0/), which permits use, duplication, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made. Open Access: This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/ licenses/by/4.0/), which permits use, duplication, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made. J Pharm Pharmacogn Res (2024) 12(2): 295 J Pharm Pharmacogn Res (2024) 12(2): 295 https://jppres.com
https://openalex.org/W2591972047	https://europepmc.org/articles/pmc5387652?pdf=render	English	null	Effect of food processing on degradation of hexachlorocyclohexane and its isomers in milk	Veterinary world/Veterinary World	2,017	cc-by	4,388	Copyright: Singh and Nelapati. Open Access. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/ by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http:// creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Abstract Aim: To study the effect of different food processing techniques on the degradation of organochlorine compounds (α, β, ɣ and δ hexachlorocyclohexane isomers (HCH)) residues in both natural and fortified samples of milk. Materials and Methods: Raw milk samples are collected from the local areas of Hyderabad, India. Naturally and fortified milk samples (HCH) were subjected to various food processing techniques, pasteurization (63ºC for ½ h), sterilization (121ºC for 15 min) and boiling for 5 min and analyzed by gas chromatography with electron capture detector using quick, easy, cheap, effective, rugged and safe method for multiresidue analysis of pesticides in milk with slight modification. Results: The final mean residual concentration of pesticide in milk after heat processing and percentage of degradation were calculated with respective treatments. Conclusion: Heat treatments are highly effective on reduction of mean residual concentration of HCH in milk. In which Sterilization and boiling proved to be more effective in degradation of HCH isomers. Keywords: gas chromatography-electron capture detector, hexachlorocyclohexane isomers, pesticide residues, quick; easy; cheap; effective; rugged and safe method. directly or indirectly. The OC insecticides known to be extremely persistent compounds are either banned or their uses are severely restricted in most of the developed countries in the world [5]. In the recent past, γ hexachlorocyclohexane (HCH) (lindane) and technical benzene hexachloride (BHC) (a mixture of HCH isomers) have been used extensively, particu larly for the control of agricultural pests and mosqui toes. The use of the technical grade of HCH has been banned in India for April 1997, and only lindane can be used for field application on crops. Among various HCH isomers, only lindane has the insecticidal prop erty and is the only isomer of HCH which is permitted for use in agriculture [6]. The range of HCH isomer concentration (as sum of each isomer) is extremely wide (0.001-4.0 mg/L) in milk and milk products in different countries [7-9]. In India, dairy milk and milk products are highly contaminated with dichlorodi phenyltrichloroethane (DDT) and HCH isomers for decades [10,11]. The OC compounds are known for inducing or aggravating certain health problems in humans such as cancer, immune systems suppression, and the disruption of hormonal functions [12]. Due to lipophilic and relative stable properties of these res idues, may easily metabolized in different products and accumulate in fatty tissues of animal (meat, milk, etc.). RESEARCH ARTICLE Open Access RESEARCH ARTICLE Open Access Veterinary World, EISSN: 2231-0916 A il bl t t i ld / Veterinary World, EISSN: 2231-0916 Available at www.veterinaryworld.org/Vol.10/March-2017/1.pdf Abstract Continuous intake of contaminated milk and milk products may lead to biomagnification of these residues in the human body, causing chronic toxicity after long-term exposure. To ensure food safety, it is necessary to find simple and cost-effective strategies Sujatha Singh and Krishnaiah Nelapati Department of Veterinary Public Health and Epidemiology, College of Veterinary Science, P. V. Narsim Veterinary University, Rajendranagar, Hyderabad - 500 030, Telangana, India. Veterinary Public Health and Epidemiology, College of Veterinary Science, P. V. Narsimha Rao Telangana Veterinary University, Rajendranagar, Hyderabad - 500 030, Telangana, India. y y, j g , y , g , Corresponding author: Sujatha Singh, e-mail: sujathasingh.hem@gmail.com, Co-author: KN: drnkrishnaiah@gmail.com y y j g y g Corresponding author: Sujatha Singh, e-mail: sujathasingh.hem@gmail.com, Co-author: KN: drnkrishnaiah@gmail.com g Received: 02-09-2016, Accepted: 24-01-2017, Published online: 03-03-2017 doi: 10.14202/vetworld.2017.270-275 How to cite this article: Singh S, Nelapati K (2017) Effect of food processing on degradation of hexachlorocyclohexane and its isomers in milk, Veterinary World, 10(3): 270-275. Collection of milk samples Raw milk samples were collected from local markets of Hyderabad. Later both natural and spiked milk samples (α, β, γ and δ HCH at 1 parts per mil lion [ppm]) were processed by adopting quick, easy, cheap, effective, rugged and safe (QuEChERS), AOAC official method 2007.01 with slight modifica tion for analysis of pesticide residues and method was validated with recovery of 70-120% is the acceptable limit for the analysis of pesticide residues in milk sam ples on gas chromatography-electron capture detector (GC-ECD). The pattern of elution of beta-endosulfan standards (Figure-1) was analyzed on the basis of specific reten tion time for GC-ECD. The limit of detection and limit of quantification of specific OC compounds were 0.01 ppm and 0.05 ppm, respectively. The recov ery value was calculated from the calibration curves constructed from the concentration and peak areas of the obtained chromatograms from the milk samples (Figure-2) with standards of OC pesticide. Blank anal ysis of milk samples was also performed to check the different matrix interferences. Ethical approval No animal was harmed or given stress during collection of milk samples. Method validation The required quantity of OC compounds of International Standards was prepared from certified reference materials obtained from Dr. Erhenstofer, Germany, and stock standards were obtained from All India Network Project on Pesticide Residue Lab located at Professor Jayashankar Telangana State Agricultural University, Rajendranagar, Hyderabad, Telangana, India. The beta-endosulfan standard was fortified in the representative samples of milk at the rate of 0.1 ppm. The recovery of pesticide residues above 70-120% is considered, as the acceptable limit. The pattern of elution of beta-endosulfan standards (Figure-1) was analyzed on the basis of specific reten tion time for GC-ECD. The limit of detection and limit of quantification of specific OC compounds were 0.01 ppm and 0.05 ppm, respectively. The recov ery value was calculated from the calibration curves constructed from the concentration and peak areas of the obtained chromatograms from the milk samples (Figure-2) with standards of OC pesticide. Blank anal ysis of milk samples was also performed to check the different matrix interferences. Chemicals and reagents Acetonitrile (ACN), acetone, n-hexane, anhy drous sodium sulfate, sodium acetate, primary sec ondary amines (PSA) of high-performance liquid chromatography residue grade were obtained from Qualigens and Merck Specialities Private Limited. Analytical standards with >99% purity were obtained from Dr. Ehrenstosfer, Germany and stored in deep freeze maintained at −40°C. Introduction Milk is highly nutritious diet for all age groups including adolescents and patients. The contamination of milk is considered as one of the major public health problems; mainly arise due to biological agents and residues of pesticides, antibiotics, and heavy metals. Injudicious and indiscriminate usage of pesticides, not only contaminate the ecosystem, but also bio-accumu late in the food chain and can be traced in plant and animal tissues causing serious health hazards [1,2]. As per World Health Organization (WHO) approximately 20,000 deaths annually were recorded, till now due to pesticide exposures in human population [3], there fore, there is strong need for discussion, on causes of environmental contamination, pesticide residues in milk, meat and other dairy products. The organo chlorine (OC) compounds usage has been restricted in agriculture but permitted to use in limited quantities successively for public health activities in controlling vector-borne diseases in most of the developing coun tries and also in India [4]. In India, 58% of the population engaged in agri culture, so they are more exposed to the pesticides Veterinary World, EISSN: 2231-0916 270 Available at www.veterinaryworld.org/Vol.10/March-2017/1.pdf using Turbovac LV set at 52°C until near dryness up to 20 min. The dried residue content was recon stituted in 1 ml of n-hexane and further analyzed in GC-ECD set under the standard operating conditions (Table-1). In control samples, pure and fortified milk samples with standard mixtures were analyzed using the same protocol. The recovered residue levels from both natural and spiked samples of milk were calcu lated using following formula. using Turbovac LV set at 52°C until near dryness up to 20 min. The dried residue content was recon stituted in 1 ml of n-hexane and further analyzed in GC-ECD set under the standard operating conditions (Table-1). In control samples, pure and fortified milk samples with standard mixtures were analyzed using the same protocol. The recovered residue levels from both natural and spiked samples of milk were calcu lated using following formula. to reduce pesticide residue concentration in the food commodities. Food processing is the best alternative, at domestic and industry level to tackle the current scenario of unsafe food. The adventitious removal of residues by processing is influenced by the type of food, location of pesticide, nature of pesticide, and processing method [13,14]. Introduction Food processing of live stock products, i.e., cooking, boiling, sterilization, microwave oven cooking, drying, fermentation, and storage processes exhibited large percentage of reduc tion in residue levels [14-16]. Residues in ppm = Sample peak area × concentratedof standard (ppm) Standard peak area Finalvolumeof l standardinjected the sample (2 ml) × × l of sampleinjected Weightof thesample µ µ Residues in ppm = Sample peak area × concentratedof standard (ppm) Standard peak area Finalvolumeof l standardinjected the sample (2 ml) × × l of sampleinjected Weightof thesample µ µ Residues in ppm = This study was undertaken to study the effect of different food processing methods on degradation of HCH isomers (α, β, γ and δ HCH) in both natural and fortified samples of milk. Finalvolumeof l standardinjected the sample (2 ml) × × l of sampleinjected Weightof thesample µ µ Statistical analysis Multiresidue pesticide analysis was performed using QuEChERS method with slight modifica tion [17,18]. Weigh 15 g of milk samples with 15 ml of ACN containing 1% acetic acid followed by addition of 6 g of anhydrous magnesium sulfate (MgSO4) and 1.5 g of sodium acetate. Shake vigorously for 1 min by hand. Later centrifuge the tubes at 5000 rpm for 1 min. After centrifugation, 1 ml of ACN extract will be transferred to mini centrifuge tubes for dispersive solid phase extraction, i.e. (d SPE) in which 50 mg of PSA, 50 mg of C18 (octadecylsilane) and 150 mg of anhydrous MgSO4 were added and mixed the extracts for 20 s. The tube was mixed for 30 s using a vortex mixer and centrifuged again for 5 min at 3000 rpm. Finally, 2 ml of clear extract was collected and evap orated under the gentle stream of nitrogen (15 psi) The procedure was run thrice with each sample 10 in numbers were analyzed statistically through one-way ANOVA using SPSS software. 271 Table-1: GC operating conditions. GC column Zebron‑ZB ‑ 50, length – 30 m, 0.25 µm film thickness, internal diameter 0.25 mm Column oven (°C) 280°C Detector temperature 300°C Injector temperature 260°C Injector status Font injector type split, split ratio 1:10 Carrier gas flow (ml/min) 12.4 ml/min Total run time (min) 60 min GC=Gas chromotography Table-1: GC operating conditions. Veterinary World, EISSN: 2231-0916 271 Available at www.veterinaryworld.org/Vol.10/March-2017/1.pdf Figure-1: Chromatogram representing elution pattern of beta endosulfan standard. Figure-2: Chromatogram representing the elution pattern of beta endosulfan in fortified milk. was differed significantly (p<0.01) among the treat ments. The β HCH content in natural raw milk was 0.0634 ppm, degraded during heat processing reach ing 0.0532 ppm in pasteurized milk, 0.0398 ppm in boiled and 0.0235 ppm in sterilized milk, accounting 16.08%, 37.22% and 62.93%, respectively. In the spiked (1 ppm) raw milk, the initial concentration of β HCH was 0.785 ppm, degraded to 0.6712 in pas teurized, 0.4444 in boiled and 0.2745 ppm in steril ized milk accounting to 14.49%, 43.38% and 65.03% reductions, respectively. The content of β HCH in raw and spiked milk was differed significantly (p<0.01) among the treatments in both natural and spiked milk samples. Figure-1: Chromatogram representing elution pattern of beta endosulfan standard. Results The mean residual levels and percentage of deg radation of OC compounds after heat processing in both natural and spiked milk samples were presented in Table-2. Residual levels of α HCH content in natural raw milk was 0.0460 ppm, degraded to 0.0375, 0.0150 and 0.0275 ppm accounting 18.47%, 45.65% and 40.21% of degradation after pasteurization, boiling and sterilization, respectively. In the spiked (1 ppm) raw milk, the initial concentration was 1.11 ppm which has been degraded to 0.895, 0.3587 and 0.4377 ppm accounting 19.36%, 76.69% and 60.56% reductions, respectively. The content of α HCH in raw and spiked milk was differed significantly among the treatments. The processing methods, i.e., boiling and sterilization did not differed significantly in natural samples, whereas in spiked milk, the residual content Statistical analysis p The γ HCH content in natural raw milk was 0.0158 ppm, degraded during heat processing reach ing 0.0140 ppm in pasteurized milk and 0.0115 ppm in boiled, accounting 11.39 and 27.21% of degradation, respectively, whereas it was below detectable level after sterilization. In the spiked (1 ppm) raw milk, it was 0.756 ppm, degraded to 0.5676 in pasteurized milk, 0.3605 ppm in boiled milk and 0.2883 ppm in sterilized milk accounting 24.92%, 52.31% and 61.86% reductions, respectively. The content of γ HCH in raw and spiked milk differed significantly with the treatments, and all the treatments differed significantly (p<0.01) among themselves. Figure-1: Chromatogram representing elution pattern of beta endosulfan standard. Figure-1: Chromatogram representing elution pattern of beta endosulfan standard. Figure-2: Chromatogram representing the elution pattern of beta endosulfan in fortified milk. The δ HCH content in natural raw milk was 0.4732 ppm, reduced after heat processing reach ing 0.4327 ppm in pasteurized milk, 0.312 ppm in boiled, 0.256 ppm in sterilized milk accounting 8.5%, 34.06% and 45.90% reductions, respectively. In the spiked (1 ppm) raw milk, it was 0.854 ppm, degraded to 0.745 ppm in pasteurized milk, 0.355 ppm in boiled milk and 0.254 ppm in sterilized milk accounting 11.70%, 58.43% and 70.25% reductions, respectively. In natural and spiked milk samples, the residue lev els differed significantly (p<0.01) and also among the three treatments. Figure-2: Chromatogram representing the elution pattern of beta endosulfan in fortified milk. The total HCH content in natural raw milk was 0.5984 ppm, reduced to 0.5374 ppm in pasteurized milk, 0.3737 ppm in boiled, 0.307 ppm in sterilized milk accounting 10.19%, 37.55% and 48.69%, respec tively. In the spiked (1 ppm) raw milk, the total HCH was 3.505 ppm, degraded to 2.878 ppm in pasteur ized milk, 1.618 ppm in boiled milk and 1.221 ppm in sterilized milk accounting 17.88%, 66.85% and 65.16% reductions, respectively. In natural and spiked milk samples, the residue levels differed significantly (p<0.01) within treatments, whereas in spiked milk boiling and sterilization treatments did not differ significantly. Veterinary World, EISSN: 2231-0916 Discussion During food processing the pesticide on the basis of individual nature, may degrade in different ways, such as it may evaporate, thermally degrade or co-distillate accordingly [16]. Food processing stud ies are very important, to relate the levels of residue Veterinary World, EISSN: 2231-0916 272 Available at www.veterinaryworld.org/Vol.10/March-2017/1.pdf Table-2: Residual levels of organochlorine compounds after heat processing in milk. Table-2: Residual levels of organochlorine compounds after heat processing in milk. Name of the pesticide Raw milk (ppm) Pasteurization (ppm) Deg (%) Boiling (ppm) Deg (%) Sterilization (ppm) Deg (%) α HCH Natural 0.0460±0.005a 0.0375±0.003b 18.47 0.0150±0.0025c 45.65 0.0275±0.002c 40.21 Spiked 1.11±0.023a 0.895±0.019b 19.36 0.3587±0.002d 76.69 0.4377±0.013c 60.56 β HCH Natural 0.0634±0.003a 0.0532±0.025b 16.08 0.0398±0.021c 37.22 0.0235±0.012d 62.93 Spiked 0.785±0.05a 0.6712±0.039b 14.49 0.4444±0.006c 43.38 0.2745±0.010d 65.03 γ HCH Natural 0.0158±0.002a 0.0140±0.002b 11.39 0.0115±0.03c 27.21 BDL ‑ Spiked 0.756±0.05a 0.5676±0.024b 24.92 0.3605±0.016c 52.31 0.2883±0.019d 61.86 δ HCH Natural 0.4732±0.04a 0.4327±0.014b 8.5 0.312±0.015c 34.06 0.256±0.012d 45.90 Spiked 0.854±0.06a 0.745±0.058b 11.70 0.355±0.014c 58.43 0.254±0.014d 70.25 Total HCH Natural 0.5984±0.032a 0.5374±0.017b 10.19 0.3737±0.016c 37.55 0.307±0.018d 48.69 Spiked 3.505±0.65a 2.878±0.51b 17.88 1.618±0.22c 66.85 1.221±0.21c 65.16 Residual levels bearing different superscripts (a, b, c, and d) horizontally differed significantly (p<0.01). HCH=Hexachlorocyclohexane reduced the lindane levels by 75.0%, 79.6% and 85.4%, respectively [26]. Spiked samples of milk with lindane subjected to boiling for 2 min and reported that boiling was more effective in dissipation of pesticide residues [27]. The percentage of degrada tions in both natural and spiked samples of milk on sterilization were 40.21 and 60.56 for α HCH, 62.93 and 65.03 for β HCH, 61.86 for γ HCH, 45.90 and 70.25 for δ HCH and 48.69 and 65.16 for total HCH, respectively. Sterilization of milk at 12ºC for 15 min. showed 83.25%, 91.67% and 68.70% loss in β BHC, lindane and p,p’ DDT content, respectively [28]. Skim milk recombined with butter oil fortified with 2 ppm HCH on sterilization, there was 19% of degra dation in the HCH content [29]. The effect of steril ization on the residues of lindane and its metabolites contaminated at the level of 1 mg/kg fat, observed to reduce the residual content by 84.4% and 76.6%, respectively [26]. concentration in raw food commodities to the level of processed food products, to evaluate the strength of processing factor in degradation of various pesticide in the products [19]. Discussion Most of the OC compounds and its metabolites are lipid soluble, and thus they found more concentrated in the fatty portions of foods espe cially in butter, cream ghee and cheeses, etc. [10,20]. The percentage of degradation in both natural and spiked samples of milk was (18.47% and 19.36% for α HCH), (16.08% and 14.49% for β HCH), (11.39% and 24.92% for γ HCH), (8.5% and 11.70% for δ HCH), and (total HCH 10.19% and 17.88%), respectively, on pasteurization. Pasteurization could not eliminate the residues in the milk, but degraded to some extent which was similar to 15.38% of degra dation [7] whereas very little effect of pasteurization seen on the degradation levels of hexachlorobenzene (HCB) and lindane content in milk samples [21] whereas some scientist reported that HCB pesticide residue remained unaffected by pasteurization [22]. The percentages of degradation in both natural and spiked samples of milk were 45.65% and 76.69%, 37.22% and 43.38%, 27.21% and 52.31%, 34.06% and 58.43%, 37.55% and 66.85%, respectively, for α, β, γ and δ HCH. During boiling, the milk samples were subjected to higher temperature where consid erable decrease in residue levels in all isomers of HCH were recorded, while in β HCH degradation was less compared to other isomers of HCH in spiked samples, as by the physiochemical nature of β HCH have lower volatility and high melting point, makes more stable compound than rest of the isomers of HCH [23]. The percentage of degradation of 24.49% was seen in β HCH content on boiling, which was less than the present study [24]. Skim milk recom bined with butter oil fortified with 2 ppm HCH shown very little impact of boiling and boiling with malai removal reduced the HCH isomers by 11.54-26.78% and 35.86-50.88% [25]. Boiling for 5, 10, and 15 min References 1. Coat, S., Monti, D., Legendre, P., Bouchon, C., Massat, F. and Lepoint, G. (2011) Organochlorine pollution in tropical rivers (Guadeloupe): Role of ecological factors in food web bioaccumulation. Environ. Pollut., 159(6): 1692-1701. 19. Gonzalez-Rodriguez, R.M., Rial-Otero, R., Cancho-Grande, B., Gonzalez-Barreiro, C. and Simal- Gándara, J. (2011) A review on the fate of pesticides during the processes within the food-production chain. Crit. Rev. Food Sci. Nutr., 51(2): 99-114. ( ) 2. Borgå, K., Gabrielsen, G.W. and Skaare, J.U. (2001) Biomagnification of organochlorines along a Barents Sea food chain. Environ. Pollut., 113(2): 187-198. 20. Bayat, S., Sari, A.E., Bahramifar, N., Younesi, H. and Behrooz, R.D. (2011) Survey of organochlorine pesticides and polychlorinated biphenyls in commercial pasteurized milk in Iran. Environ. Monit. Assess., 175(1-4): 469-474. 3. WHO, Joint, and FAO Expert Consultation. (1990) Diet, Nutrition and the Prevention of Chronic Diseases. WHO, Geneva. 3. WHO, Joint, and FAO Expert Consultation. (1990) Diet, Nutrition and the Prevention of Chronic Diseases. WHO, Geneva. 21. Donia, M.A., Abou-Arab, A.A.K., Enb, A., El-Senaity, M.H. and Abd-Rabou, N.S. (2010) Chemical composition of raw milk and the accumulation of pesticide residues in milk products. Glob. Vet., 4(1): 6-14. 4. Jadhav, V.J. and Waskar, V.S (2011) Public health impli cations of pesticide residues in meat. Vet. World, 4(4): 178-182. 5. Battu, R.S., Singh, B. and Kang, B.K. (2004) Contamination of liquid milk and butter with pesticide residues in the Ludhiana district of Punjab state, India. Ectotoxicol. Environ. Saf., 59(3): 324-331. 22. Pietrino, D. (1991) Pesticide residues and milk processing. Ind. Latte, 27: 3-32. 23. POPRC. (2007) UNEP/POPS/POPRC.3/20/Add.5; Stockholm Convention on Persistent Organic Pollutants. 6. Walker, K., Vallero, D.A. and Lewis, R.G. (1999) Factors influencing the distribution of lindane and other hexachlo rocyclohexanes in the environment. Environ. Sci. Technol., 33: 4373-4378. 24. Ismail, T. and Elkassas, W.M. (2016) Prevalence of some pesticides residues in buffalo’s milk with refer to impact of heating. Alex. J. Vet. Sci., 48(2): 113-123. g ( ) 25. Madan, V.K. and Kathpal, T.S. (2001) Decontamination of Pesticide Residues in Milk by Different Kitchen Processes. International Conference Pesticide, Environment, Food Security. IARI, New Delhi. p267-268. 7. Kampire, E., Kiremire, B.T., Nyanzi, S.A. and Kishimba, M. (2011) Organochlorine pesticide in fresh and pasteurized cow’s milk from Kampala markets. Chemosphere, 84(7): 923-927. 26. Abou-Arab, A.A.K. (1999) Effects of processing and stor age of dairy products on lindane residues and metabolites. Food Chem., 64: 467-473. 8. Competing Interests The authors declare that they have no competing interests. 18. Jeong, I.S., Kwak, B.M., Ahn, J.H. and Jeong, S.H. (2012) Determination of pesticide residues in milk using a QuEChERS-based method developed by response surface methodology. Food Chem., 133(2): 473-481. Conclusion The HCH residues were detected in the milk sam ples, which exceeded the maximum residue level lim its (0.001 ppm), recommended by FAO/WHO. Effect of different food processing plays an efficient role on degradation of OC compound residues in milk. Most of the residues, which are highly lipophilic nature, are hardly get metabolized, hence it may easily concen trate in milk products leading to bioaccumulation and biomagnification [30]. Prolonged consumption of con taminated milk will severely affect the human health and responsible for producing chronic effects like can cer and risk of Alzheimer’s disease [31]. Application pesticide in enhancing agricultural productivity must be balanced against possible health hazards arising from toxic pesticide residues in food. First and fore most thing is application pesticide should be in com pliance with Good Agricultural Practices and shifting from chemical farming to organic farming. However, Veterinary World, EISSN: 2231-0916 273 Available at www.veterinaryworld.org/Vol.10/March-2017/1.pdf (2002) Persistent organochlorine pesticide residues in milk and dairy products in India. Food Addit. Contam., 19(2): 153-157. (2002) Persistent organochlorine pesticide residues in milk and dairy products in India. Food Addit. Contam., 19(2): 153-157. in developing countries like India, the process of adoption of good agricultural practices and organic farming is very slow; meanwhile, it is highly practi cal to adopt some simple processing techniques and acquire significance for reducing the harmful pesti cide residues in food. 11. Negi, R.K. and Rani, S. (2015) Contamination profile of DDT and HCH in packaged milk samples collected from Haridwar, India. Int. J. Pure Appl. Biosci., 3(5): 121-127. 12. Vincenzo, R.M., Campanella, L. and Avino, P. (2006) Determination of organophosphorus pesticide residues in human tissues by capillary gas chromatography - Negative chemical ionization mass spectrometry analysis. J. Chromatogr. B, 780: 431-441. Authors’ Contributions The study was part of SS’s research work during the Ph.D. program. SS prepared, processed and analyzed the samples. KN designed and planned the research programme and approved the final manuscript. 13. Keikotlhaile, B.M., Spanoghe, P. and Steurbaut, W. (2010) Effects of food processing on pesticide residues in fruits and vegetables: A meta-analysis approach. Food Chem. Toxicol., 48(1): 1-6. 14. Bajwa, U. and Sandhu, K.S. (2014) Effect of handling and processing on pesticide residues in food - A review. J. Food Sci. Technol., 51(2): 201-220. Acknowledgments The authors are highly thankful to All India Network Project on Pesticide Residues (ICAR-RES/2009-10/1) for financial help. Professor Jayashankar, Telangana State Agricultural University, Rajendranagar, and the Dean, College of Veterinary Science, Rajendranagar, for providing necessary infra structure facilities in carrying out the research work. 15. Regueiro, J., López-Fernández, O., Rial-Otero, R., Cancho-Grande, B. and Simal-Gándara, J. (2015) A review on the fermentation of foods and the residues of pesti cides - Biotransformation of pesticides and effects on fermentation and food quality. Crit. Rev. Food Sci. Nutr., 55(6): 839-863. 16. Kaushik, G., Satya, S. and Naik, S.N. (2009) Food process ing a tool to pesticide residue dissipation - A review. Food Res. Int., 42(1): 26-40. 17. Lehotay, S.J., Mastovska, K. and Yun, S.J. (2005) Evaluation of two fast and easy methods for pesticide residue analysis in fatty food matrixes. JAOAC Int., 88(2): 630-638. p y , 30. Darko, G. and Acquaah, S.O. (2008) Levels of organochlo rine pesticides residues in dairy products in Kumasi, Ghana. Chemosphere, 71(2): 294-298. References Luzardo, O.P., Almeida-González, M., Henríquez- Hernández, L.A., Zumbado, M., Álvarez-León, E.E. and Boada, L.D. (2012) Polychlorobiphenyls and organochlo rine pesticides in conventional and organic brands of milk: Occurrence and dietary intake in the population of the Canary Islands (Spain). Chemosphere, 88: 307-315. 27. Dhiman, N., Bakhshi, A.K., Singh, B. and Gagan, J. (2005) Effect of conventional household processes on chlorpyrios and lindane residues in milk. Pestic. Res. J., 17(2): 104-106. 28. Abou-Arab, A.A.K. (1991) Microbiological and Compositional Quality of Dairy Products in Relation to Some Pollutants, Ph.D. Faculty of Agriculture Ain Shams University. 9. Georgescu, B., Georgescu, C., Daraban, S. and Mihaiescu, T. (2011) Assessment of persistent organic pollutants acting as endocrine disruptor chemicals in animal fat, cow milk and lacteous sub-products from Cluj Country, Romania. Anim. Biol. Anim. Husb., 3: 1-9. y 29. Nath, B.S., Unnikrishnan, V., Bhavadasan, M.K., Chitra, P.S. and Murthy, M.K.R. (1997) Effect of processing on some organochlorine pesticide contents of milk and milk 10. Pandit, G.G., Sharma, S., Srivastava, P.K. and Sahu, S.K. Veterinary World, EISSN: 2231-0916 274 Available at www.veterinaryworld.org/Vol.10/March-2017/1.pdf 31. Singh, N.K., Chhillar, N., Banerjee, B.D., Bala, K., Basu, M. and Mustafa, M. (2013) Organochlorine pesticide levels and risk of Alzheimer’s disease in North Indian pop ulation. Hum. Exp. Toxicol., 32(1): 24-30. Veterinary World, EISSN: 2231-0916 products. Indian J. Dairy Biosci., 8: 6-9. 30. Darko, G. and Acquaah, S.O. (2008) Levels of organochlo rine pesticides residues in dairy products in Kumasi, Ghana. Chemosphere, 71(2): 294-298. 31. Singh, N.K., Chhillar, N., Banerjee, B.D., Bala, K., Basu, M. and Mustafa, M. (2013) Organochlorine pesticide levels and risk of Alzheimer’s disease in North Indian pop ulation. Hum. Exp. Toxicol., 32(1): 24-30. ******** products. Indian J. Dairy Biosci., 8: 6-9. Available at www.veterinaryworld.org/Vol.10/March-2017/1.pdf 275 Veterinary World, EISSN: 2231-0916 275
https://openalex.org/W2809933609	https://europepmc.org/articles/pmc6423314?pdf=render	English	null	Vertical Redistribution of Soil Organic Carbon Pools After Twenty Years of Nitrogen Addition in Two Temperate Coniferous Forests	Ecosystems	2,018	cc-by	17,950	ABSTRACT SOC andsoil total N responded to 20 years of regular, low-dose N addition in two European coniferous forests in Switzerland and Denmark. At the Swiss site, the addition of 22 kg N ha-1 y-1 (or 1.3 times throughfall deposition) stimulated tree growth, but decreased soil pH and exchangeable calcium. At the Danish site, the addition of 35 kg N ha-1 y-1 (1.5 times throughfall deposition) impaired tree growth, increased ﬁne root biomass and led to an accumu- lation of N in several belowground pools. At both sites, elevated N inputs increased SOC pools in the moderately decomposed organic horizons, but de- creased them in the mineral topsoil. Hence, long- term N addition led to a vertical redistribution of SOC pools, whereas overall SOC storage within 30 cm depth was unaffected. Our results imply that an N- induced shift of SOC from older, mineral-associated pools to younger, unprotected pools might foster the vulnerability of SOC in temperate coniferous forest soils. SOC andsoil total N responded to 20 years of regular, low-dose N addition in two European coniferous forests in Switzerland and Denmark. At the Swiss site, the addition of 22 kg N ha-1 y-1 (or 1.3 times throughfall deposition) stimulated tree growth, but decreased soil pH and exchangeable calcium. At the Danish site, the addition of 35 kg N ha-1 y-1 (1.5 times throughfall deposition) impaired tree growth, increased ﬁne root biomass and led to an accumu- lation of N in several belowground pools. At both sites, elevated N inputs increased SOC pools in the moderately decomposed organic horizons, but de- creased them in the mineral topsoil. Hence, long- term N addition led to a vertical redistribution of SOC pools, whereas overall SOC storage within 30 cm depth was unaffected. Our results imply that an N- induced shift of SOC from older, mineral-associated pools to younger, unprotected pools might foster the vulnerability of SOC in temperate coniferous forest soils. Nitrogen (N) inputs from atmospheric deposition can increase soil organic carbon (SOC) storage in temperate and boreal forests, thereby mitigating the adverse effects of anthropogenic CO2 emissions on global climate. However, direct evidence of N-in- duced SOC sequestration from low-dose, long-term N addition experiments (that is, addition of < 50 kg N ha-1 y-1 for > 10 years) is scarce worldwide and virtually absent for European temperate forests. Vertical Redistribution of Soil Organic Carbon Pools After Twenty Years of Nitrogen Addition in Two Temperate Coniferous Forests 1Institute of Soil Research, Department of Forest and Soil Sciences, University of Natural Resources and Life Sciences (BOKU), 1190 Vienna, Austria; 2Department of Microbiology and Ecosystem Science, University of Vienna, 1090 Vienna, Austria; 3Swiss Federal Institute for Forest, Snow and Landscape Research (WSL), 8903 Birmensdorf, Switzerland; 4Department of Geosciences and Natural Resource Management, University of Copenhagen, 1958 Frederiksberg C, Denmark; 5Division of Radiation Protection, Department of Radiation Protection and Radiochemistry, Austrian Agency for Health and Food Safety (AGES), 1220 Vienna, Austria 2018 The Author(s) 2018 The Author(s) 2018 The Author(s) Ecosystems (2019) 22: 379–400 https://doi.org/10.1007/s10021-018-0275-8 Received 11 September 2017; accepted 7 June 2018; published online 26 June 2018 Electronic supplementary material: The online version of this article (https://doi.org/10.1007/s10021-018-0275-8) contains supplementary material, which is available to authorized users. Authors’ Contribution SJF contributed to study design, performed research, analyzed data and wrote the manuscript; VW and SM per- formed research; KMK, ED-P, FH and WW co-wrote the manuscript; PS and PG maintained the N addition experiments, provided data and co- wrote the manuscript; MT conceived the study and performed research; MHG conceived the study and co-wrote the manuscript; SZ-B conceived the study, co-wrote the manuscript, was the ﬁrst supervisor of students and led the project. Electronic supplementary material: The online version of this article (https://doi.org/10.1007/s10021-018-0275-8) contains supplementary material, which is available to authorized users. Authors’ Contribution SJF contributed to study design, performed research, analyzed data and wrote the manuscript; VW and SM per- formed research; KMK, ED-P, FH and WW co-wrote the manuscript; PS and PG maintained the N addition experiments, provided data and co- wrote the manuscript; MT conceived the study and performed research; MHG conceived the study and co-wrote the manuscript; SZ-B conceived the study, co-wrote the manuscript, was the ﬁrst supervisor of students and led the project. INTRODUCTION Reactive nitrogen (NR) inputs from atmospheric deposition to terrestrial ecosystems have more than doubled since the onset of the industrial revolu- tion, mainly due to agricultural intensiﬁcation and fossil fuel burning (Galloway and others 2008). Further increases in global NR deposition rates are predicted up to 2050 (Galloway and others 2004; Simpson and others 2014). Temperate forests, which covered 7.67 9 106 km2 globally in 2007 (Pan and others 2011), are particularly affected by high NR deposition in the vicinity of the densely populated areas in North America, Europe and Asia (Townsend and others 1996; Holland and others 1997). At the same time, these forests were a net sink for atmospheric CO2 of up to 0.8 Pg of carbon (C) per year during the early 2000s (Pan and others 2011) corresponding to about 1/3 of the ‘residual land sink’ (Ciais and others 2014). ( ) The net uptake of C by temperate forests has been linked to anthropogenic NR inputs (Oren and others 2001; Reay and others 2008; Ferna´ndez- Martı´nez and others 2014), which help to alleviate N limitation of tree growth (Tamm 1991; LeBauer and Treseder 2008; Thomas and others 2010). Hence, most of the additionally sequestered C is stored in tree biomass (Pregitzer and others 2008; De Vries and others 2009; Frey and others 2014). However, the vegetation contains only 40% of C stored in temperate forest ecosystems, while soils down to 1 m depth account for the remaining 60%, mainly in the form of soil organic carbon (SOC; Dixon and others 1994; Lal 2005). It is therefore vital to quantify N-induced changes in SOC in or- der to assess the overall effects of NR deposition on temperate forest C (De Vries and others 2006, 2014). Fine roots represent a relatively small fraction of total plant biomass in temperate forests (4–7%; Vogt and others 1995). However, ﬁne root pro- duction is a large component of belowground net primary production (BNPP) in these systems (Nadelhoffer 2000), which in turn can comprise up to 45% of the total belowground C ﬂux (TBCF; Litton and Giardina 2008). Furthermore, root-de- rived C has increasingly been recognized as an important contributor to SOC (Rasse and others 2005; Crow and others 2009; Tefs and Gleixner 2012; Angst and others 2018). ABSTRACT Here, we examine how tree growth, ﬁne roots, physicochemical soil properties as well as pools of Electronic supplementary material: The online version of this article (https://doi.org/10.1007/s10021-018-0275-8) contains supplementary material, which is available to authorized users. Authors’ Contribution SJF contributed to study design, performed research, analyzed data and wrote the manuscript; VW and SM per- formed research; KMK, ED-P, FH and WW co-wrote the manuscript; PS and PG maintained the N addition experiments, provided data and co- wrote the manuscript; MT conceived the study and performed research; MHG conceived the study and co-wrote the manuscript; SZ-B conceived the study, co-wrote the manuscript, was the ﬁrst supervisor of students and led the project. Key words: carbon sequestration; carbon miner- alization; nitrogen deposition; nitrogen saturation; ﬁne roots; exchangeable cations; soil carbon; soil pH; Norway spruce. Corresponding author; e-mail: stefan.forstner@boku.ac.at 379 S. J. Forstner and others 380 2016). Part of these inconsistencies might arise from the fact that few studies evaluated changes in SOC pools, as opposed to SOC concentrations, in experiments explicitly simulating the long-term effects of low-dose N addition (for example, Pre- gitzer and others 2008). In such experiments, it is crucial to distinguish between N-induced changes in SOC concentrations and SOC pools, respectively, as only the latter represent the actual amount of C stored in a deﬁned soil volume/area. This is even more important as N has been shown to affect concentrations and pools of SOC differently (Maarouﬁand others 2015; Boot and others 2016). We are aware of only two studies that quantiﬁed SOC pools in temperate forests after more than 10 years of experimental N addition. Pregitzer and colleagues (2008) found that SOC pools in the top 10 cm of mineral soil in a hardwood forest in- creased by 25% on average after 10 years of adding N at 30 kg ha-1 y-1. More recently, Frey and col- leagues (2014) reported that organic horizon SOC pools increased by 33 and 52% in hardwood and pine stands, respectively, whereas in mineral horizons SOC pools did not respond to 20 years of N addition treatment (50 kg ha-1 y-1). INTRODUCTION Thus, even small changes in ﬁne root biomass (FRB) in response to altered N inputs may feed back on ecosystem C storage (Xia and Wan 2008; Janssens and others 2010; Li and others 2015; Peng and others 2017). ) Several meta-analyses suggest that experimental N addition increases the amount of SOC in tem- perate forests (Hyvo¨nen and others 2008; Nave and others 2009; Janssens and others 2010; Liu and Greaver 2010; Yue and others 2016), although this response is not universal (Lu and others 2011b). The main mechanisms behind the observed in- creases in SOC appear to be higher litter inputs to soil via stimulated tree productivity and/or sup- pressed decomposition of recalcitrant soil organic matter (SOM; Franklin and others 2003; Janssens and others 2010; Hagedorn and others 2012). However, amount and location of the additional C sequestered in SOM may vary considerably be- tween ecosystems and soil horizons (Nave and others 2009; Liu and Greaver 2010; Yue and others Typically, FRB decreases with increasing N availability, especially relative to aboveground biomass (Nadelhoffer and others 1985; Gundersen and others 1998; Yuan and Chen 2010). A recent review found a 13.5% reduction of FRB in forests exposed to experimentally increased N deposition for up to 14 years (Li and others 2015). However, results from the few long-term experiments (‡ 15 years of N addition) are more ambiguous: While higher N inputs reduced FRB in a temperate coniferous forest stand (Frey and others 2014), FRB was not affected by enhanced N inputs in a tem- perate deciduous (Burton and others 2012; Frey Long-Term N Addition and Soil Organic C 381 and others 2014) or in a boreal coniferous forest (Maarouﬁand others 2015). the atmosphere consists equally of ammonium (NH4 +) and nitrate (NO3 -) and approximates to 12 kg N ha-1 y-1 (Schleppi and others 1999a). Here, we studied two temperate coniferous for- ests located in Switzerland and Denmark, which have received low doses of additional N in monthly or more frequent intervals for two decades. We quantiﬁed the amount of C and N stored in soil and ﬁne root pools after 19 and 20 years of treatment, respectively, to test if experimental N addition in- creased SOC and soil total N (STN) pools (Hy- pothesis 1) and decreased ﬁne root C (and N) pools (Hypothesis 2). Study Sites We studied two temperate coniferous forest sites at Alptal (Switzerland) and Klosterhede (Denmark), which have received low-dose N addition for approximately two decades (see Table 1 for site characteristics). Alptal is located in a valley on the northern edge of the Alps in central Switzerland (4702¢N, 843¢E). The local climate is wet and cool with a mean annual precipitation of 2300 mm and a mean annual temperature of 6C. The experi- mental site is located on a 20% slope with a western aspect. The soils are clay-rich Gleysols that have developed from Flysch, a sedimentary rock formation consisting of alternating calcareous sandstones with argillite and bentonite schists (Leupold 1942; Hagedorn and others 2001b). The heterogeneous microtopography of mounds and depressions results in a patchy distribution of soil types (Schleppi and others 1998, 2017; Hagedorn and others 2001b; Krause and others 2013). On mounds, Umbric Gleysols dominate with a thick organic layer (Oi, Oe and Oa horizons) on top of an Ah, an oxidized Bl and a partly oxidized Blr hori- zon. In depressions, the organic layer consists of an Oi horizon, whereas Oe and Oa horizons are mostly lacking, so that the soil classiﬁes as Mollic Gleysol (Schleppi and others 1998; Hagedorn and others 2001b; Krause and others 2013). Ground vegeta- tion is well developed, and different botanical associations are found on mounds and in depres- sions, respectively (Schleppi and others 1999b). The naturally regenerating forest stand is domi- nated by up to 260-year-old Norway spruce trees (Picea abies L. Karst.) with a 15% contribution of silver ﬁr (Abies alba Mill.). Bulk N deposition from INTRODUCTION In addition, we combine measure- ments of soil pH, exchangeable cations and extractable N pools with long-term monitoring data on tree growth to assess the effects of N addition on an ecosystem level. The Klosterhede experimental site is located in Western Jutland, Denmark (5629¢N, 824¢E), with a mean annual temperature of 9C and a mean annual precipitation of 860 mm. The site is ﬂat and microtopography is more uniform compared to Alptal. The coarse-textured, nutrient-poor soil is classiﬁed as Haplic Podzol and has developed from glacio-ﬂuvial sands (Gundersen and Rasmussen 1995). A thick organic layer consisting of Oi (not sampled in this study), Oe and Oa horizons overlays a humic, partly eluvial AE horizon and a bleached E horizon, followed by Bh and Bs horizons. Ground vegetation is dominated by Deschampsia ﬂexuosa L. Trin. and mosses (Gundersen and Rasmussen 1995). The managed forest stand is the second generation after heathland conversion and dominated by Nor- way spruce (Gundersen 1998). Trees were 97 years old in 2014. Bulk atmospheric N deposition aver- aged 9.4 kg N ha-1 y-1 from 1988 to 1992. Experimental Design Stand age (years) Up to 260 97 Density of stems > 10 cm DBH (ha-1) 430 860 Dominant tree height (m) 30 20 Basal area (m2 ha-1) 41 30 Microtopography Mounds, depressions Homogenous Soil type Umbric/Mollic Gleysols Haplic Podzol Soil horizon sequence Oi/Oe/Oa/Ah/Bl/Blr (Oia)/Oe/Oa/AE/E/Bh/Bs Parent material Flysch Glacio-ﬂuvial sands Bulk N deposition (kg N ha-1 y-1) 12.3b 9.4c Throughfall N deposition (kg N ha-1 y-1) 16.8b 23.0c Experimental N addition (kg N ha-1 y-1) 22 35 Form of added N NH4NO3 NH4NO3 Frequency of N addition In each rain event Monthly Start (duration) of N addition 1995 (19 years) 1992 (20 years)d Data for Alptal were compiled from Schleppi and others (1998), Hagedorn and others (2001a, b) and Krause and others (2012a, b, 2013). Data for Klosterhede were compiled from Gundersen and Rasmussen (1995) and Gundersen (1998). aThe Oi horizon is not included in this study. bAverage of two experimental catchments from April 1993 to March 1995. cAverage of control plots from 1988 to 1992. dNo treatment in 1998–1999. Table 1. Selected Site Characteristics of Alptal (CH) and Klosterhede (DK) using a backpack sprayer. Nitrogen addition started in April 1995 and varied annually with the local precipitation regime. At the time of sampling, 21.6 ± 4.6 kg N ha-1 y-1 (mean ± SD) have been added to N addition plots (Krause and others 2012a). using a backpack sprayer. Nitrogen addition started in April 1995 and varied annually with the local precipitation regime. At the time of sampling, 21.6 ± 4.6 kg N ha-1 y-1 (mean ± SD) have been added to N addition plots (Krause and others 2012a). (Figure S7.1). Due to the original layout of the experiment, N addition plots were spatially segre- gated. However, the arrangement of control plots allowed us to test whether spatial gradients were present across the experimental area using multi- variate ordination techniques (see Appendix S7 for details). As we did not ﬁnd such gradients, we analyzed the experiment as a split-block design. At Klosterhede, the original experimental design involved one N addition area (15 m 9 15 m) and two control areas (15 m 9 15 m and 15 m 9 10 m) located side by side to the N addition area (Gundersen and Rasmussen 1995). Experimental Design At Alptal, we sampled four pairs of circular plots (20 m2 each). These plots were used previously to study N transformations, soil collembola and soil trace gas ﬂuxes (Mohn and others 2000; Hagedorn and others 2001b; Xu and others 2009; Krause and others 2013). Each pair consisting of one control plot and one N addition plot in close vicinity to each other (< 30 m) was selected based on com- parable microtopography and vegetation cover. The N addition treatment was assigned randomly to one plot within each pair. Thus, each pair is con- sidered as a block and the experiment was analyzed as a split-block design with four replicate blocks. Nitrogen was added as NH4NO3 to rainwater col- lected on a polyethylene sheet spread outside the forest (300 m2), which was then directed into a water tank and applied automatically by sprinklers (Schleppi and others 2017). Sprinklers were mounted 1.5 m above ground, so that N was added below canopy level but on top of ground vegetation (Krause and others 2012a). The N treatment was applied during precipitation events (that is, approximately 200 times per year) to mimic natural atmospheric N deposition as realistically as possible without changing the water regime of the plots (Krause and others 2013; Schleppi and others 2017). Control plots received the same amount of unaltered rainwater. In winter, automatic irriga- tion was replaced by the occasional application of concentrated NH4NO3 solution on top of snow S. J. Forstner and others 382 Table 1. Selected Site Characteristics of Alptal (CH) and Klosterhede (DK) Alptal Klosterhede Latitude 4702¢N 5629¢N Longitude 843¢E 824¢E MAT (C) 6 9 MAP (mm) 2300 860 Elevation (masl) 1200 27 Slope (%) 20 0 Dominant tree species Picea abies (L.) Karst. Picea abies (L.) Karst. Experimental Design The treat- ment area had received N in the form of NH4NO3 at a rate of 35 kg N ha-1 y-1 since February 1992 by hand-spraying of monthly aliquots (except for 2 years 1998–1999 and in a few drought periods). Water added to the N treated area was less than 1% of the throughfall volume, while control areas were subjected to natural precipitation (Gundersen and Rasmussen 1995). For the present study, we di- vided each area (N addition and combined controls) into four plots (7.5 m 9 7.5 m), which are re- garded as experimental units in the statistical analysis (detailed below). Then, each N addition plot was paired with the control plot in closest juxtaposition to create four blocks of two plots each Data for Alptal were compiled from Schleppi and others (1998), Hagedorn and others (2001a, b) and Krause and others (2012a, b, 2013). Data for Klosterhede were compiled from Gundersen and Rasmussen (1995) and Gundersen (1998). aThe Oi horizon is not included in this study. bAverage of two experimental catchments from April 1993 to March 1995. cAverage of control plots from 1988 to 1992. dNo treatment in 1998–1999. Soil Sampling Concentrations of non-purgeable organic carbon (NPOC) and total dissolved nitrogen (TDN) were determined in K2SO4 extracts using a TOC/TN analyzer (Shimadzu TOC-L/TNM-L, Shi- madzu, Korneuburg, Austria). Non-purgeable or- ganic C was taken as a measure of extractable organic carbon (EOC). Extractable or- ganic nitrogen (EON) was calculated by subtracting the sum of extractable NH4 +-N and NO3 --N from TDN. Upon arrival in the laboratory, small roots that were not visible in the ﬁeld were separated from organic soil samples by hand. Then, root-free or- ganic soil material was ground in a polytron blen- der (7000 rpm for 3 min; Retsch Grindomax GM2000, Retsch, Haan, Germany). Although grinding does not affect bulk chemical properties, salt-extractable C and N pools and potential C mineralization rates might have been affected. Samples from N addition and control plots, how- ever, were treated in the exact same manner allowing for comparisons between treatments. Mineral soil samples were sieved through a 2 mm mesh to retain roots, stones and particulate organic matter that remained on the sieve. All roots were sorted by hand under a binocular into ﬁne and coarse roots using a diameter cutoff of 2 mm. Masses of roots, particulate organic matter and stones were determined before and after oven- drying (70C, 24 h). Gravimetric water content and soil dry mass were determined by drying soil sub- samples at 105C to constant mass. Another set of subsamples was air-dried to constant mass for the analyses of soil texture, exchangeable cations, car- bonate and concentrations of C and N. The remaining ﬁeld-moist soil was stored at 4C up to two weeks for analysis of K2SO4-extractable C and N pools, soil pH and potential C mineralization rates. Carbon and Nitrogen Pools of Soil and Fine Roots Subsamples of air-dried soil were ground (1500 rpm for 2.5 min, MM2000, Retsch, Haan, Germany) and analyzed for total C and N concen- trations by dry combustion at 1050C (Carlo Erba NA 1500, Milan, Italy) according to standard O¨ NORM procedures (www.austrian-standards.at/ en). Inorganic C was measured as CO2 after treat- ment with 10% HCl by the Scheibler method and subtracted from total C to obtain SOC concentra- tions. Soil organic C pools (SOCP; kg m-2) were calculated for each horizon i as ð1Þ SOCPi ¼ qB;i di SOCi 1 hi=100 ð Þ ð Þ; ð1Þ Soil Sampling We collected genetic soil horizons from three sampling locations within each plot (that is, a total of 12 samples for each horizon/treatment combi- nation). Sampling locations were selected in the ﬁeld to cover topographic variation within each plot. As microtopography varies strongly over short distances at Alptal, we collected samples from both mounds (17 locations) and depressions (7 loca- tions) and included microtopography as a random factor in statistical analysis. At each sampling location, organic horizon material was quantita- tively removed from within a 25 cm 9 25 cm me- tal frame and horizon depth was measured at each side of the pit. Visible roots were separated by hand and weighed, before the fresh mass of organic Long-Term N Addition and Soil Organic C 383 material was determined on site. It is important to note that we did not separate live and dead roots, which has implications for our deﬁnition of FRB (see ‘Discussion’). Subsequently, four cores of mineral soil were sampled with a steel corer (length 30 cm, diameter 4.5 cm). Depths of mineral hori- zons were determined on each retrieved core be- fore individual horizons were pooled to create one composite sample per horizon for each sampling location. Fresh mass of each composite sample was determined on site. Soil and root samples were stored in airtight plastic bags, placed in cooled boxes and immediately transported to the labora- tory. Sampling was conducted from April 23 to 24, 2014 (Klosterhede), and from June 10 to 12, 2014 (Alptal). oxidized with H2O2. Data from the pipette method are shown for Klosterhede and from the sedigraph method for Alptal. Soil pH was electrochemically measured in 1:10 w/v slurries of soil and Milli-Q water (WTW 196, WTW, Bayern, Germany). Exchangeable cations were determined in un- buffered BaCl2 extracts (O¨ NORM L 1086-1) via atomic absorption spectroscopy (for Ca2+, Mg2+, K+, Na+, Fe3+, Mn2+; PinAAcle 900T, Perkin Elmer, MA, USA) or ICP-MS (for Al3+; 77009, Agilent Technologies O¨ sterreich, Vienna, Austria). Effec- tive cation exchange capacity (CECeff) was calcu- lated as the sum of the above cations, while base saturation (BSeff) represents the fractional contri- bution of base cations (Ca2+, Mg2+, K+ and Na+) to CECeff. Ammonium-N and NO3 --N were deter- mined photometrically (Perkin Elmer 2300 En- Spire, USA) in 0.5 M K2SO4 extracts (1:10 w/v for mineral horizons, 1:20 w/v for organic horizons) using published methods (Hood-Nowotny and others 2010). Physicochemical Soil Properties where qB,i is ﬁne earth bulk density (kg m-3), di is horizon thickness (m), SOCi is SOC concentration (kg kg-1), and hi is volume fraction of roots and stones. Soil total N pools (STNP) were calculated by substituting total N concentration (STNi) for SOCi in equation (1). Our method for calculating SOC and STN pools essentially equals method 4 of Poe- The texture of mineral horizons was determined using the pipette method according to standard O¨ NORM procedures (www.austrian-standards.at/ en). For selected horizons, particle size distribution was additionally quantiﬁed with a sedigraph (Se- diGraph III, micromeritics Germany GmbH, Aa- chen, Germany) after organic matter had been S. J. Forstner and others 384 plau and others (2017) as we corrected estimates of soil volume for stone volume. In addition, we also corrected soil volume for the volume of roots (see Appendix S4 for details on calculations of qB,i and hi). CO2 efﬂux from published and unpublished sour- ces (see Tables S6.1, S6.2 for complete list of parameters including references). Data Analysis In these cases, we built simple linear models (package ‘stats’) and tested for effects of N treatment, soil horizon and their interaction with Type III ANOVA as implemented in the ‘car’ package. Differences between soil horizons were assessed using pairwise contrasts with Tukey’s HSD tests (package ‘lsmeans’). Assumptions were checked visually for all models by plots of residuals vs. ﬁtted values and qq-plots. For some variables, model residuals severely violated normality or heteroscedasticity even after data transformation (concentrations of SOC, STN, NH4 +-N, NO3 --N, EON at Klosterhede). Consequently, we split the data set and reran the analyses separately for or- ganic and mineral horizons (package ‘lmerTest’; Kuznetsova and others 2017). Simple linear regression was used to test for linear relationships between calendar year and N-induced changes in basal area increment (BAI), average needle bio- mass, needle Mg/N and litter N (Figure 3). In addition, median responses of N concentrations of ﬁne roots and soil were added to Figure 3. We did FRCPi ¼ Mi Ai FRCi 103; ð4Þ ð4Þ where M is ﬁne root dry mass (kg), A is sampling area (m2), and FRC is ﬁne root C concentration (kg kg-1). Fine root N pools (FRNP) were calcu- lated accordingly. Data Analysis We recalculated mineral horizon SOC pools in 10-cm increments as We used linear (mixed) models in R (v3.2.5) to test for the ﬁxed effects of N treatment, soil horizon and their interaction on soil and ﬁne root pools as well as on physicochemical soil properties (R Core Team 2016). Data were tested for normality and homo- geneity of variances using Shapiro–Wilk and Le- vene’s tests, respectively. In case of non-normality or heteroscedasticity, data were log- or box-cox- transformed using the ‘powerTransform’ function from the ‘car’ package (Fox and Weisberg 2011). When sufﬁcient replicates were available to allow for the inclusion of random effects (that is, for pools and most physicochemical soil properties), we built linear mixed models using the packages ‘lme4’ (Bates and others 2015) and ‘afex’ (Singmann and others 2015). Block and replicate within block and treatment were included as random effects in the models to account for spatial autocorrelation. Microtopographic position (mounds/depressions) was included as random effect for Alptal. Post hoc differences between soil horizons were assessed by Tukey’s HSD tests, and Dunnett’s tests were used to test for differences between treatments within horizons using the ‘lsmeans’ package (Lenth 2016). Low replication precluded the inclusion of ran- dom effects for some variables (sand/silt/clay, CE- Ceff, BSeff). In these cases, we built simple linear models (package ‘stats’) and tested for effects of N treatment, soil horizon and their interaction with Type III ANOVA as implemented in the ‘car’ package. Differences between soil horizons were assessed using pairwise contrasts with Tukey’s HSD tests (package ‘lsmeans’). Assumptions were checked visually for all models by plots of residuals vs. ﬁtted values and qq-plots. For some variables, model residuals severely violated normality or heteroscedasticity even after data transformation (concentrations of SOC, STN, NH4 +-N, NO3 --N, EON at Klosterhede). Consequently, we split the data set and reran the analyses separately for or- ganic and mineral horizons (package ‘lmerTest’; Kuznetsova and others 2017). Simple linear regression was used to test for linear relationships between calendar year and N-induced changes in basal area increment (BAI), average needle bio- mass, needle Mg/N and litter N (Figure 3). In addition, median responses of N concentrations of ﬁne roots and soil were added to Figure 3. Data Analysis We did We used linear (mixed) models in R (v3.2.5) to test for the ﬁxed effects of N treatment, soil horizon and their interaction on soil and ﬁne root pools as well as on physicochemical soil properties (R Core Team 2016). Data were tested for normality and homo- geneity of variances using Shapiro–Wilk and Le- vene’s tests, respectively. In case of non-normality or heteroscedasticity, data were log- or box-cox- transformed using the ‘powerTransform’ function from the ‘car’ package (Fox and Weisberg 2011). When sufﬁcient replicates were available to allow for the inclusion of random effects (that is, for pools and most physicochemical soil properties), we built linear mixed models using the packages ‘lme4’ (Bates and others 2015) and ‘afex’ (Singmann and others 2015). Block and replicate within block and treatment were included as random effects in the models to account for spatial autocorrelation. Microtopographic position (mounds/depressions) was included as random effect for Alptal. Post hoc differences between soil horizons were assessed by Tukey’s HSD tests, and Dunnett’s tests were used to test for differences between treatments within horizons using the ‘lsmeans’ package (Lenth 2016). Low replication precluded the inclusion of ran SOCPincrement ¼ X k i¼1 SOCPi Fi ð Þ; ð2Þ ð2Þ where k is the number of genetic mineral soil horizons within the respective depth increment, SOCPi is the C pool of genetic mineral soil horizon i (kg m-2), and Fi is the fraction of genetic mineral soil horizon i within the respective depth incre- ment. Mineral horizon STN pools were recalculated accordingly. To account for the inﬂuence of hori- zon thickness on pools, we also calculated pools sizes of modeled, 1-cm-thick layers for each hori- zon (kg m-2 cm-1; Mu¨ ller and Ko¨gel-Knabner 2009): SOCPmodeled;i ¼ SOCPi di 100 ð Þ : ð3Þ ð3Þ Modeled STN pools were calculated accordingly. Modeled STN pools were calculated accordingly. Subsamples of dried ﬁne roots were milled (MM2000, Retsch, Haan, Germany) before C and N concentrations were determined by dry combus- tion at 1250C (LECO TruSpec CN, LECO, Mo¨nchengladbach, Germany). Fine root C pools (FRCP) for each horizon i (g m-2) were calculated as Low replication precluded the inclusion of ran- dom effects for some variables (sand/silt/clay, CE- Ceff, BSeff). Physicochemical Soil Properties Two decades of N addition consistently reduced soil pH at Alptal by 0.28 across all horizons (P = 0.032, Figure 1A, Table S1.1). Extractable organic carbon (EOC) concentrations were marginally lower in N addition plots by 38% on average across all hori- zons (P = 0.089, Figure 1B, Table S1.1). Similarly, N addition decreased CECeff in mineral horizons by 38% (P = 0.010, Figure 1C), mainly due to loss of exchangeable calcium (P = 0.008, Figure 1D, Nitrogen addition had even more pronounced effects on extractable inorganic N pools at Kloster- hede. Ammonium was consistently higher across the whole soil proﬁle in N addition plots (P < 0.001 in both organic and mineral horizons; Figure 2A, B, Table S1.3). In comparison, NO3 - was often not detectable in control plots, but low concentrations were measured in N addition plots 67 62 Organic and mineral horizons A 4 5 6 7 8 Control + Nitrogen pH 67 62 Organic and mineral horizons B 0 3500 7000 10500 14000 Control + Nitrogen EOC (µg g−1 soil) 12 12 Mineral horizons C 0 200 400 600 800 1000 Control + Nitrogen CECeff (mmolc kg−1 soil) 11 12 Mineral horizons D 0 100 200 300 400 500 600 700 Control + Nitrogen Ca2+ (mmolc kg−1 soil) Alptal Figure 1. Soil pH (A), extractable organic C (B), cation exchange capacity (C) and exchangeable calcium (D) across all horizons (A, B) and in mineral horizons (C, D) at Alptal. Box–whisker plots are shown for each treatment. The black line is the median, lower and upper boundaries correspond to the ﬁrst and third quartiles, respectively, and whiskers span 1.5 times the interquartile range. Gray lines around the boxes depict Gaussian kernel estimates of probability densities. Sample numbers are depicted above the x-axes. 67 62 Organic and mineral horizons A 4 5 6 7 8 Control + Nitrogen pH 67 62 Organic and mineral horizons B 0 3500 7000 10500 14000 Control + Nitrogen EOC (µg g−1 soil) A B 12 12 Mineral horizons C 0 200 400 600 800 1000 Control + Nitrogen CECeff (mmolc kg−1 soil) 11 12 Mineral horizons D 0 100 200 300 400 500 600 700 Control + Nitrogen Ca2+ (mmolc kg−1 soil) C D Figure 1. Potential C Mineralization Rates and Ecosystem-Level Response to N Addition Potential rates of C mineralization were measured in short-term laboratory incubations to evaluate if long-term N addition affected C availability in the soil (see Appendix S6 for details). To assess the ecosystem-level response to N addition, we further compiled data on tree growth, needle biomass, needle chemistry and litter N concentrations (see Figure 3 legend for references). Nitrogen effects on ecosystem C cycling were assessed by summarizing related parameters such as foliar litterfall or soil Long-Term N Addition and Soil Organic C 385 not statistically evaluate the effects of N addition on ecosystem C cycling due to limited data availability. Table S1.2). Nitrogen addition also increased extractable NH4 + concentrations in the Oe horizon (P = 0.043), as well as extractable NO3 - concen- trations in Oi and Oa horizons (P = 0.021 and P = 0.050, respectively), whereas extractable inor- ganic N pools in mineral soil horizons were not affected (Table S1.1). Physicochemical Soil Properties Soil pH (A), extractable organic C (B), cation exchange capacity (C) and exchangeable calcium (D) across all horizons (A, B) and in mineral horizons (C, D) at Alptal. Box–whisker plots are shown for each treatment. The black line is the median, lower and upper boundaries correspond to the ﬁrst and third quartiles, respectively, and whiskers span 1.5 times the interquartile range. Gray lines around the boxes depict Gaussian kernel estimates of probability densities. Sample numbers are depicted above the x-axes. S. J. Forstner and others 386 (P < 0.001 and P = 0.002 in organic and mineral horizons, respectively; Figure 2C, D, Table S1.3). Likewise, EON was signiﬁcantly higher in organic horizons (P < 0.001, Figure 2E) as well as in AE and E horizons of N addition plots (P = 0.001 and P = 0.009, respectively; Table S1.3). There was a signiﬁcant interaction of N addition treatment and horizon on EOC/EON (P = 0.011), with smaller ewise, EON was signiﬁcantly higher in organic izons (P < 0.001, Figure 2E) as well as in AE signiﬁcant interaction of N addition trea horizon on EOC/EON (P = 0.011), wi 24 24 Organic horizons A 0 100 200 300 400 500 600 Control + Nitrogen NH4 + −N (µg g−1 soil) 44 48 Mineral horizons B 0 5 10 15 20 Control + Nitrogen NH4 + −N (µg g−1 soil) 24 24 Organic horizons C 0 2 4 6 8 10 Control + Nitrogen NO3 −N (µg g− − 1 soil) 44 48 Mineral horizons D 0.00 0.35 0.70 1.05 1.40 Control + Nitrogen NO3 −N (µg g− − 1 soil) 24 24 Organic horizons E 0 100 200 300 400 500 600 Control + Nitrogen EON (µg g−1 soil) 24 24 Organic horizons F 0 2 4 6 8 Control + Nitrogen EOC:EON Klosterhede ure 2. Salt-extractable pools of NH4 +-N (A, B), NO3 --N (C, D), organic N (E) and salt-extractable EOC anic horizons (A, C, E, F) and mineral horizons (B, D) at Klosterhede. Box–whisker plots are shown for each black line is the median, lower and upper boundaries correspond to the ﬁrst and third quartiles, respe skers span 1.5 times the interquartile range. Gray lines around the boxes depict Gaussian kernel estimates o sities. Sample numbers are depicted above the x-axes. Tree Growth and N Accumulation Patterns Long-term effects of N addition on tree growth, needle biomass, needle chemistry and litter N concentrations were assessed using time series data (Figure 3). Tree growth was differently affected by N additions at the two sites. At Alptal, basal area increment (BAI) increased by 1.2% per year in response to N between 1995 and 2008 (Figure 3A; At Alptal, N addition induced a relative accu- mulation of N in litter (+ 1.7% per year over con- trols between 1995 and 2008, Figure 3C); response of needle N concentrations, however, was only transient (Figure 3C, Krause and others 2012a). At A B C D Figure 3. Relative response of tree growth and N accumulation in various ecosystem compartments to two decades of N addition at Alptal (A, C) and Klosterhede (B, D). A, B show tree growth (basal area increment), average needle biomass and ﬁne root biomass. C, D show N concentrations in needles, litter, ﬁne roots and soil, and needle Mg/N. The insert in (D) shows needle N concentrations at Klosterhede from 1992 to 1995. Symbols for ﬁne root biomass, ﬁne root N and soil N concentrations represent median responses, a more robust measure of location than means. Error bars are the adjusted median absolute deviations (MAD), a robust measure of dispersion analogous to standard deviation (Huber and Ronchetti 2009; Leys and others 2013). Assuming normal distribution of the data, the adjusted MAD encompasses 50% of the observations (Rousseeuw and Croux 1993). Fine root biomass and N concentrations were summarized for organic (‘Org.’) and mineral (‘Min.’) horizons. For soil horizon designation see Table 1. Data on basal area increment, needle biomass, needle N concentrations and needle Mg/N and, in part, litter N concentrations have been obtained from Ginzburg (2014), Gundersen (1998), Krause and others (2012a, 2013), P. Gundersen (unpublished results) and P. Schleppi (unpublished results). For data on ﬁne root biomass and ﬁne root N concentrations see Appendix S3. For data on soil N concentrations see Appendix S4. Note that ﬁne root biomass, ﬁne root N concentrations and soil N concentrations were measured on two occasions in 2014 and are separated graphically in plots to enhance readability. The linear regression equations for basal area increment, needle biomass, needle Mg/N and litter N concentrations are shown in Table S2.1. Open symbols depict points excluded from regression analyses. B A A A B C D D Figure 3. Physicochemical Soil Properties 44 48 Mineral horizons B 0 5 10 15 20 Control + Nitrogen NH4 + −N (µg g−1 soil) erhede 24 24 Organic horizons A 0 100 200 300 400 500 600 Control + Nitrogen NH4 + −N (µg g−1 soil) Klos A 44 48 Mineral horizons D 0.00 0.35 0.70 1.05 1.40 Control + Nitrogen NO3 −N (µg g− − 1 soil) 24 24 Organic horizons C 0 2 4 6 8 10 Control + Nitrogen NO3 −N (µg g− − 1 soil) C D 24 24 Organic horizons E 0 100 200 300 400 500 600 Control + Nitrogen EON (µg g−1 soil) 24 24 Organic horizons F 0 2 4 6 8 Control + Nitrogen EOC:EON E F Figure 2. Salt-extractable pools of NH4 +-N (A, B), NO3 --N (C, D), organic N (E) and salt-extractable EOC/EON (F) in organic horizons (A, C, E, F) and mineral horizons (B, D) at Klosterhede. Box–whisker plots are shown for each treatment. The black line is the median, lower and upper boundaries correspond to the ﬁrst and third quartiles, respectively, and whiskers span 1.5 times the interquartile range. Gray lines around the boxes depict Gaussian kernel estimates of probability densities. Sample numbers are depicted above the x-axes. Long-Term N Addition and Soil Organic C 387 ratios under high N inputs in the organic layer (Figure 2F) and AE horizon and similar ratios in mineral subsoil (Table S1.3). Nitrogen addition did not affect soil pH or CECeff at Klosterhede (Tables S1.3, S1.4). see Table S2.1 for regression statistics). Similarly, N addition increased average needle biomass by 1.2% per year relative to controls between 1995 and 2009 (Figure 3A, Table S2.1). At Klosterhede, BAI declined by 1.6% per year in response to high N inputs between 1993 and 2013 (Figure 3B, Table S2.1), while average needle biomass was not affected (Figure 3B). Along with the reduction in BAI, tree mortality was higher in N addition plots at Klosterhede (P. Gundersen, pers. comm.). Tree Growth and N Accumulation Patterns Fine Roots The horizon-speciﬁc analysis revealed that N addi- tion had little effect on ﬁne root traits at Alptal. Fine root biomass did not respond to N addition except for the Blr horizon (horizon/treatment interaction P = 0.022), where we observed a sig- niﬁcant, but small increase in FRB (P < 0.001, Figure S3.1). Also, N addition increased FRN con- centrations only in the Ah horizon (+ 26%, P = 0.016) where ﬁne root C/N was reduced accordingly (- 21%, P = 0.011, Table S3.1). Overall, FRC and FRN pools were not affected by N at Alptal (Figure S3.2). In contrast, ﬁne roots were strongly affected by N addition at Klosterhede. Fine root biomass and FRN concentrations increased in response to N across all horizons (FRB: + 12 to + 86%, P = 0.016; FRN concentrations: + 19 to + 46%, P < 0.001; ﬁne root C/N: - 16 to - 32%, P < 0.001; Figure 4A, B, Figure S3.3, Table S3.2). As a result, N addition signiﬁcantly increased whole-proﬁle FRC pools (P = 0.038, Figure 5A, C, Table S5.1). Fine root N 78 78 Organic and mineral horizons 0.00 0.25 0.50 0.75 1.00 1.25 1.50 Control + Nitrogen Fine root N (mass%) 78 78 Organic and mineral horizons 0 20 40 60 80 100 Control + Nitrogen Fine root C : N Klosterhede A B Figure 4. Fine root N concentrations (A) and ﬁne root C/N (B) across all horizons at Klosterhede. Box–whisker plots are shown for each treatment. The black line is the median, lower and upper boundaries correspond to the ﬁrst and third quartiles, respectively, and whiskers span 1.5 times the interquartile range. Gray lines around the boxes depict Gaussian kernel estimates of probability densities. Sample numbers are depicted above the x-axes. 78 78 Organic and mineral horizons 0.00 0.25 0.50 0.75 1.00 1.25 1.50 Control + Nitrogen Fine root N (mass%) Kloste A 78 78 Organic and mineral horizons 0 20 40 60 80 100 Control + Nitrogen Fine root C : N hede B A B Figure 4. Fine root N concentrations (A) and ﬁne root C/N (B) across all horizons at Klosterhede. Box–whisker plots are shown for each treatment. The black line is the median, lower and upper boundaries correspond to the ﬁrst and third quartiles, respectively, and whiskers span 1.5 times the interquartile range. Soil Organic C and Soil Total N Pools We found signiﬁcant interactions between horizon and N addition for SOC and STN pools at both sites (Alptal: P = 0.007 for both SOC and STN pools; Klosterhede: SOCP P = 0.028, STNP P = 0.099). At Alptal, SOC pools in the Oe horizon were 120% larger in N addition plots (1.01 vs. 0.46 kg m-2 in controls, P = 0.039, Figure 6A). In contrast, Ah horizon SOC pools were 47% smaller in N addition plots compared to controls (2.35 vs. 4.42 kg m-2, P = 0.002, Figure 6A). Similarly, STN pools were 150% larger in the Oe horizon (0.05 vs. 0.02 kg m-2, P = 0.021, Figure 6B) and 44% smaller in Ah horizons in the N addition plots (0.12 vs. 0.21 kg m-2, P = 0.0002, Figure 6B). The effect of N addition on SOC and STN pools was, however, smaller when ﬁxed-depth increments were com- pared at both sites (increment/treatment interac- tions for Alptal: SOCP P = 0.102, STNP P = 0.129; for Klosterhede: SOCP P = 0.159, STNP P = 0.456). At Alptal, SOC and STN pools of combined O horizons still tended to be larger in N addition plots (+ 22 and + 40%, respectively), but the effects of N addition treatment were no longer signiﬁcant (P = 0.298 and P = 0.184, respectively; Figure 6C, D). In the 0–10 cm depth increment, SOC and STN pools were 26 and 22% smaller in N addition than in control plots, respectively (P = 0.031 and Tree Growth and N Accumulation Patterns Soil Organic C and Soil Total N Pools Tree Growth and N Accumulation Patterns Relative response of tree growth and N accumulation in various ecosystem compartments to two decades of N addition at Alptal (A, C) and Klosterhede (B, D). A, B show tree growth (basal area increment), average needle biomass and ﬁne root biomass. C, D show N concentrations in needles, litter, ﬁne roots and soil, and needle Mg/N. The insert in (D) shows needle N concentrations at Klosterhede from 1992 to 1995. Symbols for ﬁne root biomass, ﬁne root N and soil N concentrations represent median responses, a more robust measure of location than means. Error bars are the adjusted median absolute deviations (MAD), a robust measure of dispersion analogous to standard deviation (Huber and Ronchetti 2009; Leys and others 2013). Assuming normal distribution of the data, the adjusted MAD encompasses 50% of the observations (Rousseeuw and Croux 1993). Fine root biomass and N concentrations were summarized for organic (‘Org.’) and mineral (‘Min.’) horizons. For soil horizon designation see Table 1. Data on basal area increment, needle biomass, needle N concentrations and needle Mg/N and, in part, litter N concentrations have been obtained from Ginzburg (2014), Gundersen (1998), Krause and others (2012a, 2013), P. Gundersen (unpublished results) and P. Schleppi (unpublished results). For data on ﬁne root biomass and ﬁne root N concentrations see Appendix S3. For data on soil N concentrations see Appendix S4. Note that ﬁne root biomass, ﬁne root N concentrations and soil N concentrations were measured on two occasions in 2014 and are separated graphically in plots to enhance readability. The linear regression equations for basal area increment, needle biomass, needle Mg/N and litter N concentrations are shown in Table S2.1. Open symbols depict points excluded from regression analyses. S. J. Forstner and others 388 Klosterhede, needle N concentrations in N addition plots increased by 1.1% per year relative to controls between 1992 and 2009 (Figure 3D). Similarly, litter N concentrations increased by 1.9% per un- der high N inputs between 1992 and 2010 (Fig- ure 3D). Median N concentrations of ﬁne roots were higher with N in organic and mineral hori- zons (+ 33 and + 37%, respectively) and so were median STN concentrations of Oa, AE and Bs horizons (Figure 3D). pools were consistently higher in N addition plots across all soil horizons (P < 0.001, Figure 5B, D, Table S5.1), due to of both higher FRB and higher FRN concentrations. Fine Roots Gray lines around the boxes depict Gaussian kernel estimates of probability densities. Sample numbers are depicted above the x-axes. Long-Term N Addition and Soil Organic C 389 A B C D Figure 5. Fine root C pools (A, C) and ﬁne root N pools (B, D) in genetic horizons (upper panels) and depth increments (lower panels) at Klosterhede. Means (± SE) were derived from 12 samples except for the following combinations of horizon/increment and treatment: Bh Control, Bs Control, 10–20 cm Control and 20–30 cm Control (10 samples). Lower- case letters right to each plate originate from pairwise comparison of horizon/increment means. Means with no letter in common are signiﬁcantly different (Tukey’s HSD, a = 0.05). The N addition main effects for ﬁne root N pools were signiﬁcant at P = 0.010 and P = 0.006 for genetic horizons (B) and depth increments (D), respectively. A B D C D D Figure 5. Fine root C pools (A, C) and ﬁne root N pools (B, D) in genetic horizons (upper panels) and depth increments (lower panels) at Klosterhede. Means (± SE) were derived from 12 samples except for the following combinations of horizon/increment and treatment: Bh Control, Bs Control, 10–20 cm Control and 20–30 cm Control (10 samples). Lower- case letters right to each plate originate from pairwise comparison of horizon/increment means. Means with no letter in common are signiﬁcantly different (Tukey’s HSD, a = 0.05). The N addition main effects for ﬁne root N pools were signiﬁcant at P = 0.010 and P = 0.006 for genetic horizons (B) and depth increments (D), respectively. Horizon Thickness, Bulk Density and Elemental Concentrations of Genetic Soil Horizons Horizon Thickness, Bulk Density and Elemental Concentrations of Genetic Soil Horizons P = 0.055, respectively; Figure 6C, D). These N- induced shifts in SOC and STN pools at Alptal were also apparent when sampling sites located in depressions were excluded from the analysis (data not shown). Higher SOC pools in the Oe horizon of N addition plots at Alptal were due to greater soil masses ra- ther than higher SOC concentrations. Oe horizons in N addition plots were insigniﬁcantly thicker (+ 55%, P = 0.179) and had a higher bulk density (+ 73%, P = 0.033) compared to control plots, resulting in greater masses of Oe horizons (P = 0.014), while neither SOC nor STN concen- trations were signiﬁcantly affected by N (Table S4.1). However, N addition signiﬁcantly re- duced SOC/STN of Oe horizon material from 27 to 23 (P = 0.033, Table S4.1). Conversely, thickness and bulk density of the Ah horizon tended to be lower in N addition plots compared to controls (- 19 and - 17%, respectively; P = 0.119 and P = 0.163). Soil organic C concentrations of the Ah horizon were lower in N addition plots (- 24%, P = 0.039), and STN concentrations tended to de- At Klosterhede, the SOC pool in the Oe horizon was 32% larger in N addition plots (3.23 vs. 2.45 kg m-2 in controls, P = 0.043), whereas N inputs reduced the AE horizon SOC pool by 25% (2.64 vs. 3.51 kg m-2 in controls, P = 0.024, Fig- ure 7A). Moreover, the STN pool in the Oe horizon was 37% larger in N addition plots (0.10 vs. 0.08 kg m-2 in controls, P = 0.025, Figure 7B). Pools of SOC and STN summed across organic horizons were higher in N addition plots compared to controls (SOCP: 4.89 vs. 3.93 kg m-2, P = 0.037; STNP: 0.15 vs. 0.11 kg m-2, P = 0.063; Figure 7C, D). Soil organic C and STN pools in subsoil horizons below 10 cm depth were not affected by the N addition treatment neither at Alptal nor at Klosterhede (Figures 6, 7). S. J. Forstner and others 390 A B C D Figure 6. Soil organic C pools (A, C) and soil total N pools (B, D) in genetic horizons (upper panels) and depth increments (lower panels) at Alptal. Horizon Thickness, Bulk Density and Elemental Concentrations of Genetic Soil Horizons Means (± SE) were derived from 12 samples except for the following combinations of horizon and treatment: Oi + N (11 samples), Oe Control (7), Oe + N (8), Oa Control (11), Oa + N (6), Ah Control and Ah + N (13). Lower-case letters right to each plate originate from pairwise comparison of horizon/increment means. Means with no letter in common are signiﬁcantly different (Tukey’s HSD, a = 0.05). Symbols next to bars indicate signiﬁcant post hoc differences between treatments within a given horizon/increment (nP < 0.1; P < 0.05; P < 0.01; P < 0.001). Note that post hoc differences were found for pools of 0–10 cm increments (C, D) despite nonsigniﬁcant interactions. B C C D D Figure 6. Soil organic C pools (A, C) and soil total N pools (B, D) in genetic horizons (upper panels) and depth increments (lower panels) at Alptal. Means (± SE) were derived from 12 samples except for the following combinations of horizon and treatment: Oi + N (11 samples), Oe Control (7), Oe + N (8), Oa Control (11), Oa + N (6), Ah Control and Ah + N (13). Lower-case letters right to each plate originate from pairwise comparison of horizon/increment means. Means with no letter in common are signiﬁcantly different (Tukey’s HSD, a = 0.05). Symbols next to bars indicate signiﬁcant post hoc differences between treatments within a given horizon/increment (nP < 0.1; P < 0.05; P < 0.01; P < 0.001). Note that post hoc differences were found for pools of 0–10 cm increments (C, D) despite nonsigniﬁcant interactions. crease in N addition plots as well (- 20%, P = 0.095, Table S4.1). increases in Oe horizon SOC pools as well as de- creases in Ah horizon SOC pools prevailed after accounting for differences in horizon thickness. Thus, changes in pool sizes were mainly controlled by corresponding changes bulk density (Oe hori- zon) or SOC concentrations (Ah horizon, Table S4.3). At Klosterhede pool sizes of modeled, 1-cm-thick layers did not differ between N treat- ments (Table 3). ) At Klosterhede, both horizon thickness and bulk density of the Oe horizon tended to be higher in N addition plots (+ 8 and + 20%, respectively; Table S4.2). In contrast, N addition signiﬁcantly reduced thickness and bulk density in the AE horizon by 24% (P = 0.016) and 12% (P = 0.011), respectively (Table S4.2). Horizon Thickness, Bulk Density and Elemental Concentrations of Genetic Soil Horizons Nitrogen addition further tended to increase SOC concentrations of the Oa horizon (+ 5%, P = 0.062) and increased STN concentrations in the Oa horizon (+ 9%, P = 0.034, Table S4.2). Similarly, STN concentra- tions of the AE horizon tended to be higher in N addition plots (+ 25%, P = 0.098), resulting in a marginally signiﬁcant decline in SOC/STN from 57 to 52 (P = 0.094, Table S4.2). Total Belowground C and N Pools Nitrogen addition had no effects on soil, ﬁne root or total belowground (soil plus ﬁne root) pools of C or N at Alptal (Table S5.1). In contrast, N addition marginally increased the STN pool down to 30 cm at Klosterhede by 15% (P = 0.055). Consequently, total belowground C pools did not change with N at Klosterhede, while total belowground N pools were increased by 13% on average (P = 0.044, Table S5.1). We further estimated the relative importance of horizon thickness and bulk density for pool sizes by calculating horizon-speciﬁc pools for modeled, 1- cm-thick layers (Tables 2, 3). At Alptal, N-induced Long-Term N Addition and Soil Organic C 391 A B C D Soil profile Figure 7. Soil organic C pools (A, C) and soil total N pools (B, D) in genetic horizons (upper panels) and depth increments (lower panels) at Klosterhede. Means (± SE) were derived from 12 samples except for the following combinations of horizon/increment and treatment: Bh Control, Bs Control, 10–20 cm Control and 20–30 cm Control (10 samples each). Lower-case letters right to each plate originate from pairwise comparison of horizon/increment means. Means with no letter in common are signiﬁcantly different (Tukey’s HSD, a = 0.05). Symbols next to bars indicate signiﬁcant post hoc differences between treatments within a given horizon/increment (nP < 0.1; P < 0.05; P < 0.01; P < 0.001). Note that post hoc differences were found for pools of combined organic horizons (C, D) despite nonsigniﬁcant interactions. C Soil profile D D C Figure 7. Soil organic C pools (A, C) and soil total N pools (B, D) in genetic horizons (upper panels) and depth increments (lower panels) at Klosterhede. Means (± SE) were derived from 12 samples except for the following combinations of horizon/increment and treatment: Bh Control, Bs Control, 10–20 cm Control and 20–30 cm Control (10 samples each). Lower-case letters right to each plate originate from pairwise comparison of horizon/increment means. Means with no letter in common are signiﬁcantly different (Tukey’s HSD, a = 0.05). Symbols next to bars indicate signiﬁcant post hoc differences between treatments within a given horizon/increment (nP < 0.1; P < 0.05; P < 0.01; *P < 0.001). Note that post hoc differences were found for pools of combined organic horizons (C, D) despite nonsigniﬁcant interactions. Potential C Mineralization Rates and Ecosystem-Level C Cycling of trees and mosses with N addition (- 12 and - 78%, respectively; Table S6.2). Belowground C inputs from roots and ectomycorrhiza tended to be lower in N addition plots while C outputs via litter decomposition or soil CO2 efﬂux were not affected whether measured in situ or the laboratory (see Table S6.2 for references). Carbon mineralization rates did not differ between N addition and control plots in laboratory incuba- tions, except for the Oe horizon where potential C mineralization per g SOC was increased by 144% (P = 0.001, Figure S6.1). The compilation of long- term monitoring data revealed site-speciﬁc patterns of ecosystem-level C cycling in response to N (see Figure S6.2, Tables S6.1, S6.2). At Alptal, above- ground C inputs to soil generally increased with N addition as trees grew better and produced more needle litter (+ 22%) over 13 years of N treatment. Conversely, C outputs via in situ soil CO2 efﬂux tended to decrease (- 18%), although high be- tween-plots variability precluded the detection of a signiﬁcant N effect (Krause and others 2013). At Klosterhede, aboveground C inputs via litterfall showed little response to increased N inputs. There was, however, a tendency toward reduced growth DISCUSSION Experimental N addition was observed to increase SOC pools in temperate and boreal forests (Ho¨gberg and others 2006; Pregitzer and others 2008; Lovett and others 2013; Frey and others 2014; Maarouﬁ and others 2015). In contrast to these ﬁndings and our ﬁrst hypothesis, we found that whole-proﬁle SOC pools were not affected by two decades of regular, low-dose N addition at the investigated sites. Although higher N inputs led to a SOC accumulation in Oe horizons, this was offset by a concomitant reduction of SOC pools in A horizons. S. J. Forstner and others 392 Table 2. Soil Organic C and Soil Total N Pools of Modeled, 1-cm-Thick Layers at Alptal SOC pool (kg m-2 cm-1) STN pool (kg m-2 cm-1) Oi a a Control 0.27 ± 0.04 0.008 ± 0.001 + Nitrogen 0.39 ± 0.09 0.013 ± 0.004 Oe ab ab Control 0.27 ± 0.06 0.010 ± 0.002 + Nitrogen 0.40 ± 0.07 0.018 ± 0.004 Oa ab ab Control 0.22 ± 0.04 0.009 ± 0.002 + Nitrogen 0.40 ± 0.06 0.017 ± 0.004 Ah b c Control 0.55 ± 0.07 0.026 ± 0.003 + Nitrogen 0.36 ± 0.04 0.018 ± 0.002 Bl a bc Control 0.28 ± 0.02 0.015 ± 0.001 + Nitrogen 0.33 ± 0.04 0.017 ± 0.002 Blra a b Control 0.28 ± 0.03 0.015 ± 0.001 + Nitrogen 0.23 ± 0.03 0.013 ± 0.001 Sign. effects Horizon, + N 9 horizon + Nb, horizon, + N 9 horizon Bold and bold italic values indicate signiﬁcant (P < 0.05) and marginally signiﬁcant (P < 0.1) differences between treatments, respectively. Means (± SE) were derived from 12 samples except for the following horizon/treatment combinations: Oi + N (11 samples), Oe Control (7), Oe + N (8), Oa Control (11), Oa + N (6), Ah Control and Ah + N (13). Lower-case letters within columns indicate post hoc differences between horizons (P < 0.05). aBlr horizons were sampled to an average depth of 30.0 ± 2.1 and 26.3 ± 1.0 cm from top of mineral soil in control and N addition plots, respectively. bMarginally signiﬁcant main effect (P = 0.088). Table 2. Soil Organic C and Soil Total N Pools of Modeled, 1-cm-Thick Layers at Alptal SOC pool (kg m-2 cm-1) STN pool (kg Bold and bold italic values indicate signiﬁcant (P < 0.05) and marginally signiﬁcant (P < 0.1) differences between treatments, respectively. Means (± SE) were derived from 12 samples except for the following horizon/treatment combinations: Bh Control, Bs Control, 10–20 cm Control and 20–30 cm Control (10 samples). Lower-case letters within columns indicate post hoc differences between horizons (P < 0.05). aBs horizons were sampled to an average depth of 31.6 ± 1.3 and 29.7 ± 0.8 cm from top of mineral soil in control and N addition plots, respectively. DISCUSSION Means (± SE) were derived from 12 samples except for the following horizon/treatment combinations: Oi + N (11 samples), Oe Control (7), Oe + N (8), Oa Control (11), Oa + N (6), Ah Control and Ah + N (13). Lower-case letters within columns indicate post hoc differences between horizons (P < 0.05). aBlr horizons were sampled to an average depth of 30.0 ± 2.1 and 26.3 ± 1.0 cm from top of mineral soil in control and N addition plots, respectively. bMarginally signiﬁcant main effect (P = 0.088). Table 3. Soil Organic C and Soil Total N Pools of Modeled, 1-cm-Thick Layers at Klosterhede SOC pool (kg m-2 cm-1) STN pool (kg m-2 cm-1) Oe a a Control 0.49 ± 0.04 0.015 ± 0.001 + Nitrogen 0.59 ± 0.04 0.019 ± 0.001 Oa a ab Control 0.60 ± 0.09 0.016 ± 0.003 + Nitrogen 0.51 ± 0.06 0.014 ± 0.002 AE a c Control 0.42 ± 0.02 0.007 ± 0.000 + Nitrogen 0.41 ± 0.04 0.008 ± 0.001 E b d Control 0.29 ± 0.03 0.006 ± 0.000 + Nitrogen 0.24 ± 0.01 0.005 ± 0.000 Bh a b Control 0.45 ± 0.02 0.012 ± 0.001 + Nitrogen 0.44 ± 0.02 0.011 ± 0.001 Bsa b c Control 0.27 ± 0.02 0.007 ± 0.001 + Nitrogen 0.28 ± 0.01 0.007 ± 0.000 Sign. effects Horizon Horizon Means (± SE) were derived from 12 samples except for the following horizon/treatment combinations: Bh Control, Bs Control, 10–20 cm Control and 20–30 cm Control (10 samples). Lower-case letters within columns indicate post hoc differences between horizons (P < 0.05). aBs horizons were sampled to an average depth of 31.6 ± 1.3 and 29.7 ± 0.8 cm from top of mineral soil in control and N addition plots, respectively. Soil Organic C and Soil Total N Pools of Modeled, 1-cm-Thick Layers at Klosterhede Table 3. Soil Organic C and Soil Total N Pools of Modeled, 1-cm-Thick Layers at Klos SOC pool (kg m-2 cm-1) STN Long-Term N Addition and Soil Organic C 393 were highly site-speciﬁc, emphasizing the impor- tance of inherent site characteristics (Lu and others 2015) in addition to the aforementioned differ- ences in N application and deposition rates. Experimental N Addition in Relation to Background N Deposition One of the reasons for the relatively small re- sponses of belowground C pools to long-term N addition might be the modest relative increase in N inputs. At both sites, the amount of experimentally added N was relatively low compared to annual background N deposition (N addition/throughfall N deposition: 1.3 at Alptal, 1.5 at Klosterhede). Two reasons can be given for these low ratios. First, rates of experimental N addition were chosen to be well below doses applied for forest fertilization (for example, Hyvo¨nen and others 2008) to realistically mimic elevated inputs from the atmosphere (Gal- loway and others 2008; Simpson and others 2014). Second, measured throughfall N deposition rates were already relatively high at the both sites, the late 1980s and early 1990s (17 and 23 kg N ha- 1 y-1 at Alptal and Klosterhede, respectively; Ta- ble 1). Since then, N deposition has slightly de- clined throughout Europe, but atmospheric input rates are still higher than in large parts of North America (Vet and others 2014; Jia and others 2016). Notwithstanding the similarities in N addi- tion/throughfall, Klosterhede received about 60% more additional N than Alptal during the study period. This might have contributed to the diver- gent responses observed at Klosterhede (for example, reduced tree vigor or belowground N accumulation) than at Alptal. Declines in soil pH and loss of base cations are well-known consequences of N addition (Tamm 1991; McNulty and Aber 1993; Ho¨gberg and others 2006; Lucas and others 2011; Tian and Niu 2015). Both effects were observed at Alptal and have been shown to impede nutrient uptake by trees (Schro¨- der and others 1988; Schulze 1989), and cause elemental imbalances in foliage (Schaberg and others 1997; Gundersen 1998; Jonard and others 2015), which ultimately might limit plant produc- tivity and C inputs to soil. At Alptal, however, tree growth positively responded to N until 2008 (Fig- ure 3). Thus, N-induced declines in soil pH and exchangeable cations at this site with carbonate- containing parent material (Hagedorn and others 2001b) were apparently not severe enough to outweigh the fertilizing effect of N on trees. At Klosterhede, increases in extractable N and ﬁne root N after two decades of experimental N addition point to a different response trajectory (Figures 2, 4). Experimental N Addition in Relation to Background N Deposition Increases in extractable inorganic N pools in response to N addition have been com- monly observed (Lu and others 2011a) and are in line with results from Klosterhede after four treat- ment years (Gundersen 1998). In addition, sub- stantial amounts of added N were taken up by trees at this site (Gundersen 1998), which increased lit- ter and ﬁne root N concentrations in turn (Fig- ures 3, 4). Thus, the decomposition of plant inputs high in N might have contributed to the observed increases in extractable N pools in organic horizons and mineral topsoil. DISCUSSION Al- though experimental N addition reduced both soil pH and exchangeable cations (mainly calcium) in the pedogenetically ‘younger’ soil at Alptal (Fig- ure 1; Xu and others 2009), these properties were not affected by N in the pedogenetically ‘older’ soil at Klosterhede. This suggests that the Gleysol at Alptal was able to physicochemically buffer the effects of N addition more effectively than the Podzol at Klosterhede where the soil was already strongly acidic in the control plots. Even though Klosterhede is situated close to the sea and receive cations via sea salt spray, CECeff in the control plots was two orders of magnitude lower than at Alptal, which might have limited further cation losses in response to N addition. Consequently, N addition did not affect total SOC pools down to 30 cm depth, but resulted in a ver- tical redistribution of SOC within the soil proﬁle, with gains in organic horizons and losses in mineral topsoil horizons. Further, FRB was either unre- sponsive to or increased with additional N at Alptal and Klosterhede, respectively, which contradicts our second hypothesis based on negative N re- sponses of FRB observed in gradient studies and N manipulation experiments (for example, Li and others 2015). Nitrogen Addition Induced Either Soil Acidiﬁcation or Belowground N Accumulation, Depending on Site Third, additional N may indirectly slow SOM decomposition by decreasing the mining of soil microorganisms for N (Craine and others 2007; Talbot and others 2008; Sinsabaugh 2010). While litter N concentrations increased as a con- sequence of N addition, the relative importance of abiotic N incorporation and reduced N mining for C accumulation in the Oe horizon at Klosterhede remains elusive. (Gundersen and others 1998). At Alptal, for which longer measurements are available, leaching of mainly DON and NO3 - led to a continuous loss of about one-third of the added N during 14 years of treatment (Schleppi and others 2017), indicating that N supply exceeded the retention capabilities of the plant–soil system. Despite high leaching losses, additional N also accumulated in vegetation and soil at both sites (Gundersen 1998; Krause and others 2012a, b), which progressively saturated their N retention capacity. However, the plant–soil system responded differently to additional N at each site (Figure 3). ( g ) At Alptal, the sustained, gradual increase in tree growth and needle biomass in the absence of con- comitant increases in needle N concentrations suggests that trees accumulated N in parallel with C (Figure 3A, C; Krause and others 2012a). This re- sponse corresponds to the C accumulation compo- nent of the vegetation sink (Lovett and Goodale 2011) and was linked to increases in leaf area ra- ther than higher rates of leaf-level photosynthesis at Alptal (Krause and others 2012a). Carbon accumulation in trees occurred despite progressive imbalances in tree mineral nutrition, as indicated by a relative decrease in needle Mg/N by 1.6% per year (Figure 3; McNulty and others 1996; Boxman and others 1998; Minocha and others 2000). Also, N-induced losses of base cations (Figure 1C, D) did not impair tree growth at Alptal. Nitrogen-induced decreases in C/N (stoichiometric N sink; Lovett and Goodale 2011) also contributed to N accumulation at Alptal as N addition decreased C/N of Oe horizon material (Table S4.1), ﬁne roots of the Ah horizon (Table S3.1) and litter (Figure 3C). This is in line with a decrease in soil C/N at Alptal from 1997 to 2009 reported by Schleppi and others (2017). Still, the overall contribution of altered C/N stoichiom- etry to the N sink was comparatively small and tended to decrease over time (Schleppi and others 2017). Nitrogen Addition Induced Either Soil Acidiﬁcation or Belowground N Accumulation, Depending on Site Experimentally added N affected soil pH, exchangeable cations and belowground N pools, all of which can alter SOC storage via effects on tree productivity (Oren and others 2001; Ho¨gberg and others 2006; Goll and others 2012) and decompo- sition of SOM (Sollins and others 1996; Hobbie and others 2007). Our results show that these changes In addition, the accumulation of N in labile soil and root pools at Klosterhede may have con- tributed to increased SOC storage in the Oe horizon by decelerating SOM decomposition through sev- S. J. Forstner and others 394 eral mechanisms. First, inorganic N can be abioti- cally incorporated into SOM, which can reduce SOM decomposability (No¨mmik and Vahtras 1982; Thorn and Mikita 1992; Berg and Matzner 1997; Compton and Boone 2002). Second, N addition can slow down the decomposition of lignin-rich litter such as spruce needles by increasing litter N con- centrations (Figure 3; Knorr and others 2005; Hobbie and others 2012; van Diepen and others 2015). The exact mechanisms behind this slow down are still unclear (van Diepen and others 2015), but might include N-induced changes in efﬁciency and growth rates of decomposers (Agren and others 2001), in particular of saprotrophic fungi (van Diepen and others 2016), and/or the suppression of lignolytic enzymes (Berg and Matzner 1997). Third, additional N may indirectly slow SOM decomposition by decreasing the mining of soil microorganisms for N (Craine and others 2007; Talbot and others 2008; Sinsabaugh 2010). While litter N concentrations increased as a con- sequence of N addition, the relative importance of abiotic N incorporation and reduced N mining for C accumulation in the Oe horizon at Klosterhede remains elusive. eral mechanisms. First, inorganic N can be abioti- cally incorporated into SOM, which can reduce SOM decomposability (No¨mmik and Vahtras 1982; Thorn and Mikita 1992; Berg and Matzner 1997; Compton and Boone 2002). Second, N addition can slow down the decomposition of lignin-rich litter such as spruce needles by increasing litter N con- centrations (Figure 3; Knorr and others 2005; Hobbie and others 2012; van Diepen and others 2015). The exact mechanisms behind this slow down are still unclear (van Diepen and others 2015), but might include N-induced changes in efﬁciency and growth rates of decomposers (Agren and others 2001), in particular of saprotrophic fungi (van Diepen and others 2016), and/or the suppression of lignolytic enzymes (Berg and Matzner 1997). Tree Growth, N Accumulation and the N Saturation Concept About three decades ago, Aber and others (1989) ﬁrst proposed that several plant and soil parameters change when temperate forest stands sequentially progress from N limitation to N saturation. While observed plant responses generally ﬁt to this con- cept of N saturation (Aber and others 1998; Niu and others 2016), its wider applicability beyond plant traits has been questioned (Lovett and Goodale 2011; Niu and others 2016). As a consequence, it has been proposed that a focus on the multiple fates of N (for example, sequestration, denitriﬁcation, leaching) might be better suited to address the manifold responses of ecosystems to additional N (Lovett and Goodale 2011). Importantly, this ap- proach distinguishes capacity N saturation, where ecosystem N sinks are zero or negative, from kinetic N saturation, where ecosystem sinks retain N but at lower rates compared to N addition (Lovett and Goodale 2011). In contrast, changes in C/N stoichiometry appear to dominate ecosystem N retention at Klosterhede. Nitrogen accumulated relative to C in needles, litter (Figure 3D), ﬁne roots (Table S3.2), salt-ex- tractable pools (Table S1.3) and AE soil horizon (Table S4.2). At the same time, trees clearly suf- fered in N treated plots from 2009 onwards (that is, 17 years after N addition began; Figure 3) in accordance with some results from boreal (Ho¨gberg and others 2006) and temperate forests (Magill and others 2004; McNulty and others 2005; Thomas and others 2010; Frey and others 2014). This, in turn, limited the sequestration of added N in new biomass at Klosterhede. Thus, it appears that N By this deﬁnition, both Alptal and Klosterhede quickly approached kinetic saturation as N leaching increased already within the ﬁrst year of N addi- tions (Gundersen 1998; Schleppi and others 2017). At Klosterhede, NO3 - leaching was enhanced for 3 years after initiation of N additions although the site had initially been characterized as ‘N-limited’ Long-Term N Addition and Soil Organic C 395 addition has saturated the C accumulation com- ponent of the vegetation N sink at Klosterhede. addition has saturated the C accumulation com- ponent of the vegetation N sink at Klosterhede. hede) to explain the observed responses of FRB (Nadelhoffer 2000). Indeed, it has been shown re- cently that N addition can slow down FRD (Sun and others 2016), which in another experiment contributed 5–51% to observed increases in O horizon SOC pools (Xia and others 2018). Shifts in Soil C Pools Were Mainly Driven by Changes in Horizon Thickness and Bulk Density Soil organic C and STN pools are directly dependent on horizon thickness, bulk density and the ele- mental concentrations [equation (1)]. However, SOC concentrations only (negatively) responded to N in the Ah horizon of Alptal. Thus, the observed changes in Oe and A horizon SOC and STN pools at the two sites were primarily caused by variations in horizon thickness and bulk density (Tables S4.1, S4.2, S4.3). As we sampled soil based on its pedo- genetic horizons, these variations could in part re- sult from our sampling procedure or from intrinsically high spatial heterogeneity rather than from the N addition treatment per se. In contrast to a ﬁxed-depth sampling scheme, the sampling of pedogenetic horizon considers soil development and is commonly used for estimating SOC and STN pools (Mu¨ ller and Ko¨gel-Knabner 2009; Gosheva and others 2017; Poeplau and others 2017). How- ever, to facilitate cross-study comparisons we also calculated: (i) SOC and STN pools per 10-cm increments (Figures 6C, D, 7C, D); and (ii) SOC and STN pools of modeled, 1-cm-thick layers (Ta- bles 2, 3). These corrections either reduce (method i) or eliminate (method ii) the direct inﬂuence of horizon thickness on pool sizes while retaining any effects of bulk density. ( ) Neither FRP nor FRD has been measured at the two sites. However, as FRP is fueled by plant C allocated belowground, an upper boundary can be estimated by calculating the total amount of C potentially available for FRP as difference between soil CO2 efﬂux and aboveground litterfall (Raich and Nadelhoffer 1989). This approach showed that estimated total belowground carbon ﬂux (TBCF) was on average lower in N addition plots at both sites (- 28 and - 6% at Alptal and Klosterhede, respectively; Figure S6.2, Tables S6.1 and S6.2), which suggests that less C was available for FRP in N addition plots. This is in line with observations that higher soil N availability can alleviate N limi- tation of trees, which in turn reduce TBCF and, eventually, FRP (Litton and others 2007; Peng and others 2017). However, inferring FRP from TBCF is subjected to at least two limitations. First, our estimates of TBCF assume that soil, litter and root C pool sizes did not change between treatments over time and that C losses other than due to soil CO2 efﬂux were negligible (Giardina and Ryan 2002). Tree Growth, N Accumulation and the N Saturation Concept Reduced FRD thus offers a likely alternative explanation for the observed increases in FRB pools in response to N. Nevertheless, measurements of FRP and FRD are needed to clarify the involved mechanisms and to estimate the contribution of ﬁne root inputs to SOC pools (Xia and others 2018). No Reduction of Fine Root Biomass After N Addition No Reduction of Fine Root Biomass After N Addition Nitrogen addition did not affect FRB at Alptal and increased FRB across all horizons at Klosterhede. This was contrary to what we anticipated in our second hypothesis, as FRB typically decreases along natural gradients of N availability (Nadelhoffer and others 1985; Vogt and others 1986; Gundersen and others 1998; Yuan and Chen 2010) and with experimental N addition (Haynes and Gower 1995; Magill and others 2004; Wang and others 2012; Li and others 2015). To reconcile these unexpected results, it is useful to recall that measures of FRB pools integrate ﬁne root production (FRP) and mortality followed by ﬁne root decomposition (FRD). Shifts in Soil C Pools Were Mainly Driven by Changes in Horizon Thickness and Bulk Density Shifts in Soil C Pools Were Mainly Driven by Changes in Horizon Thickness and Bulk Density Shifts in Soil C Pools Were Mainly Driven by Changes in Horizon Thickness and Bulk Density Increases in organic horizon thickness (and mass) in response to N can result from larger aboveground litter in- puts, reduced decomposition or both. Although N generally slowed down organic matter decomposi- tion in similar experiments (Franklin and others 2003; Burton and others 2004; DeForest and others 2004; Maarouﬁand others 2015), evidence for N- induced reductions of Oe horizon decomposition at Klosterhede is unclear. Although N did not signif- icantly alter soil CO2 efﬂux (measured in situ dur- ing 2002/2003, Figure S6.2, Table S6.2), potential C mineralization rates per g SOC from Oe material was 1.4-fold higher in N addition plots (Fig- ure S6.1). However, rates of potential C mineral- ization and soil CO2 efﬂux are not directly comparable as they integrate different sources of CO2 and are measured under different conditions. forest ﬂoor SOC pools (Ma¨kipa¨a¨ 1995; Olsson and others 2005; Maarouﬁand others 2015). Increases in organic horizon thickness (and mass) in response to N can result from larger aboveground litter in- puts, reduced decomposition or both. Although N generally slowed down organic matter decomposi- tion in similar experiments (Franklin and others 2003; Burton and others 2004; DeForest and others 2004; Maarouﬁand others 2015), evidence for N- induced reductions of Oe horizon decomposition at Klosterhede is unclear. Although N did not signif- icantly alter soil CO2 efﬂux (measured in situ dur- ing 2002/2003, Figure S6.2, Table S6.2), potential C mineralization rates per g SOC from Oe material was 1.4-fold higher in N addition plots (Fig- ure S6.1). However, rates of potential C mineral- ization and soil CO2 efﬂux are not directly comparable as they integrate different sources of CO2 and are measured under different conditions. Estimating Oe horizon turnover by dividing Oe mass (+ 35% with N addition in 2014) by above- ground litter input (+ 11% with N addition on average from 1992 to 2010) showed that organic horizon turnover based on mass balance increased from 22 to 26 years on average with N addition (Figure S6.2, Table S6.2; compare Zak and others 2008 for the mass balance approach). Thus, it ap- pears that both slower decomposition of Oe horizon material and higher litter inputs contributed to the observed increase in Oe horizon thickness at Klosterhede. Shifts in Soil C Pools Were Mainly Driven by Changes in Horizon Thickness and Bulk Density We suspect, however, that slowed decomposition will be more important for possible future accumulations of Oe material, as higher lit- ter inputs in N addition plots resulted from in- creased tree mortality rather than from higher NPP. Shifts in Soil C Pools Were Mainly Driven by Changes in Horizon Thickness and Bulk Density Second, even if TBCF was reduced by N, trees might allocate relatively more C toward FRP at the expense of coarse roots, mycorrhiza and/or root exudates (Giardina and others 2005). Therefore, we cannot exclude higher FRP despite lower esti- mated TBCF in N addition plots. y Calculating SOC pools of modeled, 1-cm-thick layers revealed that direct controls of SOC pools were differently affected by N at the two sites. At Klosterhede, differences in SOC pool sizes were entirely driven by differences in horizon thickness as pools of modeled, 1-cm-thick layers did not differ between treatments (Table 3). For the Oe horizon, this is in line with observations that organic hori- zon mass often increases with N addition (Ma¨kipa¨a¨ 1995; Blackwood and others 2007; Pregitzer and others 2008; Zak and others 2008; Lovett and others 2013), which in some studies led to larger However, we deem it more likely that N-induced reductions in TBCF did indeed result in lower FRP at the investigated sites as total BNPP (with FRP as an important component) generally scales with TBCF (Litton and Giardina 2008). An N-induced reduction in FRP thus would require a similar (Alptal) or even larger decrease in FRD (Kloster- S. J. Forstner and others 396 pends on soil porosity and the density of soil par- ticles (Blume and others 2010). However, if N increased the bulk density of the Oe horizon by reducing pore space, by increasing particle density or by both remains elusive. In the Ah horizon, N- induced decreases in SOC pools sizes were mainly driven by reductions of SOC concentrations (Tables S4.1, S4.3). These could have resulted from decreased inputs via roots or dissolved organic C (DOC) and/or increased outputs via SOC mineral- ization or DOC leaching. Although DOC concen- trations in 5 cm depth were not affected after 3 years of N addition (Hagedorn and others 2001a), we found a trend toward lower ﬁne root C pools, indicating that belowground inputs to the Ah horizon might have been reduced (Figures S3.2, S6.2, Table S6.1). As we found no differences in horizon-speciﬁc C output (potential C mineraliza- tion rates), these slight reductions in root inputs potentially contributed to lower SOC concentra- tions in the Ah horizon of N treated plots at Alptal. forest ﬂoor SOC pools (Ma¨kipa¨a¨ 1995; Olsson and others 2005; Maarouﬁand others 2015). CONCLUSIONS Our results, drawn from two decadelong N addition experiments, indicate that N addition induced a range of responses that were site-speciﬁc. Nitrogen enhanced tree growth despite reductions in soil pH and extractable base cations at the pedogenetically ‘younger’ Swiss site. At the pedogenetically ‘older’, nutrient-poor Danish site, however, N addition increased tree mortality and led to N accumulation in several above- and belowground pools. In the soil N addition resulted in vertical shift of SOC pools within the proﬁle that was consistent across sites. Although N induced an accumulation of SOC in the organic layers, there was a concomitant de- cline in mineral horizon SOC pools. This could have far-reaching implications for long-term C stabilization in temperate forests. Because SOC in the organic layer is subjected to a lower degree of physicochemical protection than in the mineral soil, the former is more likely to be released as CO2 in case of disturbances or changing environmental conditions. Thus, a vertical redistribution of SOC from mineral to organic layer pools as a result of increased N deposition may lead to a greater vul- nerability of SOC and reduce long-term C seques- tration in temperate forest soils. y g The N-induced accumulation of SOC in the Oe horizon at Klosterhede was compensated by lower SOC pools due to thinner AE horizons in N addition plots (Tables S4.2, S4.3). Reductions in mineral horizon thickness in response to N are generally not reported in the literature as many studies avoid potential problems with horizon identiﬁcation by sampling mineral soil in ﬁxed-depth increments (for example, Pregitzer and others 2008). In the long run, however, acidiﬁcation associated with chronic N addition may enhance the eluviation of organic matter complexed by Al and Fe ions from (A)E horizons of podzols (Funakawa and others 1993; Lundstro¨m and others 2000), which is hardly captured with a ﬁxed-depth soil sampling strategy. OPEN ACCESS This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/ 4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. Compton JE, Boone RD. 2002. Soil nitrogen transformations and the role of light fraction organic matter in forest soils. Soil Biol Biochem 34:933–43. Craine JM, Morrow C, Fierer N. 2007. Microbial nitrogen limi- tation increases decomposition. Ecology 88:2105–13. Crow SE, Lajtha K, Filley TR, Swanston CW, Bowden RD, Caldwell BA. 2009. Sources of plant-derived carbon and sta- bility of organic matter in soil: implications for global change. Glob Change Biol 15:2003–19. Compliance with Ethical Standards Burton AJ, Pregitzer KS, Crawford JN, Zogg GP, Zak DR. 2004. Simulated chronic NO3 - deposition reduces soil res- piration in northern hardwood forests. Glob Change Biol 10:1080–91. Conflict of interest The authors declare that they have no conﬂict of interest. Conflict of interest The authors declare that they have no conﬂict of interest. Ciais P, Sabine C, Bala G, Bopp L, Brovkin V, Canadell J, Chhabra A, DeFries R, Galloway J, Heimann M, others. 2014. Carbon and other biogeochemical cycles. In: Climate change 2013: the physical science basis. Contribution of working group I to the ﬁfth assessment report of the intergovernmental panel on climate change. Cambridge University Press. pp 465– 570. FUNDING This study was funded by the Austrian Science Fund (FWF; Project Number P 25438). Burton AJ, Jarvey JC, Jarvi MP, Zak DR, Pregitzer KS. 2012. Chronic N deposition alters root respiration-tissue N rela- tionship in northern hardwood forests. Glob Change Biol 18:258–66. ACKNOWLEDGEMENTS At Alptal, Oe horizon SOC pools were still larger in N addition plots after depth normalization (Ta- ble 2), emphasizing that increases in bulk density were responsible for the observed increases in SOC pools (Tables S4.1, S4.3). Bulk density itself de- Open access funding provided by Austrian Science Fund (FWF). We thank Ewald Brauner, Karin Hackl, Astrid Hobel, Angelika Hromka, Elisabeth Long-Term N Addition and Soil Organic C 397 no impact of nitrogen deposition. Environ Microbiol 9:1306– 16. Kopecky and Axel Mentler for their support with laboratory work, Eugenie Fink, Marcel Hirsch and Barbara Kitzler for analysis of ﬁne roots, and Karin Wrissnig for sedigraph analysis. We are grateful to Sonja Leitner, David Ramler and two anonymous reviewers for helpful comments on earlier versions of the manuscript. Jo¨rg Schnecker, Florian Hof- hansl and Prof. Karl Moder are acknowledged for sharing their statistical insight. We are especially thankful to Editor Dr. Christian Giardina whose comments greatly improved this contribution. Blume H-P, Bru¨ mmer GW, Horn R, Kandeler E, Ko¨gel-Knabner I, Kretzschmar R, Stahr K, Wilke B-M, Eds. 2010. Physikalische Eigenschaften und Prozesse. In: Scheffer/ Schachtschabel: Lehrbuch der Bodenkunde. Berlin: Springer. pp 171–271. Boot CM, Hall EK, Denef K, Baron JS. 2016. Long-term reactive nitrogen loading alters soil carbon and microbial community properties in a subalpine forest ecosystem. Soil Biol Biochem 92:211–20. Boxman AW, Blanck K, Brandrud T-E, Emmett BA, Gundersen P, Hogervorst RF, Kjønaas OJ, Persson H, Timmermann V. 1998. Vegetation and soil biota response to experimentally- changed nitrogen inputs in coniferous forest ecosystems of the NITREX project. For Ecol Manage 101:65–79. REFERENCES Hagedorn F, Kammer A, Schmidt MWI, Goodale CL. 2012. Nitrogen addition alters mineralization dynamics of 13C-de- pleted leaf and twig litter and reduces leaching of older DOC from mineral soil. Glob Change Biol 18:1412–27. Fox J, Weisberg S. 2011. An R companion to applied regression. Thousand Oaks, CA: Sage. Hagedorn F, Schleppi P, Bucher J, Flu¨ hler H. 2001b. Retention and leaching of elevated N deposition in a forest ecosystem with Gleysols. Water Air Soil Pollut 129:119–42. Franklin O, Ho¨gberg P, Ekblad A, A˚ gren GI. 2003. Pine forest ﬂoor carbon accumulation in response to N and PK additions: bomb 14C modelling and respiration studies. Ecosystems 6:644–58. Haynes BE, Gower ST. 1995. Belowground carbon allocation in unfertilized and fertilized red pine plantations in northern Wisconsin. Tree Physiol 15:317–25. Frey SD, Ollinger S, Nadelhoffer K, Bowden R, Brzostek E, Burton A, Caldwell BA, Crow S, Goodale CL, Grandy AS, Finzi A, Kramer MG, Lajtha K, LeMoine J, Martin M, McDowell WH, Minocha R, Sadowsky JJ, Templer PH, Wickings K. 2014. Chronic nitrogen additions suppress decomposition and se- quester soil carbon in temperate forests. Biogeochemistry 121:305–16. Hobbie SE, Eddy WC, Buyarski CR, Adair EC, Ogdahl ML, Weisenhorn P. 2012. Response of decomposing litter and its microbial community to multiple forms of nitrogen enrich- ment. Ecol Monogr 82:389–405. Hobbie SE, Ogdahl M, Chorover J, Chadwick OA, Oleksyn J, Zytkowiak R, Reich PB. 2007. Tree species effects on soil or- ganic matter dynamics: the role of soil cation composition. Ecosystems 10:999–1018. Funakawa S, Mambu K, Hirai H, Kyuma K. 1993. Pedogenetic acidiﬁcation process of forest soils in Northern Kyoto. Soil Science and Plant Nutrition 39:677–90. Ho¨gberg P, Fan H, Quist M, Binkley D, Tamm CO. 2006. Tree growth and soil acidiﬁcation in response to 30 years of experimental nitrogen loading on boreal forest. Glob Change Biol 12:489–99. Galloway JN, Dentener FJ, Capone DG, Boyer EW, Howarth RW, Seitzinger SP, Asner GP, Cleveland CC, Green PA, Hol- land EA, Karl DM, Michaels AF, Porter JH, Townsend AR, Vo¨osmarty CJ. 2004. Nitrogen cycles: past, present, and fu- ture. Biogeochemistry 70:153–226. Holland EA, Braswell BH, Lamarque J-F, Townsend A, Sulzman J, Mu¨ ller J-F, Dentener F, Brasseur G, Levy H, Penner JE, Roelofs G-J. 1997. Variations in the predicted spatial distri- bution of atmospheric nitrogen deposition and their impact on carbon uptake by terrestrial ecosystems. Journal of Geophys- ical Research: Atmospheres 102:15849–66. REFERENCES Galloway JN, Townsend AR, Erisman JW, Bekunda M, Cai Z, Freney JR, Martinelli LA, Seitzinger SP, Sutton MA. 2008. Transformation of the nitrogen cycle: recent trends, questions, and potential solutions. Science 320:889–92. Giardina CP, Coleman MD, Hancock JE, King JS, Lilleskov EA, Loya WM, Pregitzer KS, Ryan MG, Trettin CC. 2005. The re- sponse of belowground carbon allocation in forests to global change. In: Binkley D, Menyailo OV, Eds. Tree species effects on soils: implications for global change. NATO science series IV: earth and environmental sciences. Dordrecht: Springer. pp 119–54. Hood-Nowotny R, Umana NH-N, Inselbacher E, Oswald- Lachouani P, Wanek W. 2010. Alternative methods for mea- suring inorganic, organic, and total dissolved nitrogen in soil. Soil Sci Soc Am J 74:1018. Huber PJ, Ronchetti E. 2009. Robust statistics. 2nd edn. Hobo- ken, NJ: Wiley. Giardina CP, Ryan MG. 2002. Total belowground carbon allo- cation in a fast-growing eucalyptus plantation estimated using a carbon balance approach. Ecosystems 5:487–99. Hyvo¨nen R, Persson T, Andersson S, Olsson B, A˚ gren GI, Linder S. 2008. Impact of long-term nitrogen addition on carbon stocks in trees and soils in northern Europe. Biogeochemistry 89:121–37. Ginzburg SO. 2014. Nitrogen deposition effects on soil carbon dynamics in temperate forests. Janssens IA, Dieleman W, Luyssaert S, Subke J-A, Reichstein M, Ceulemans R, Ciais P, Dolman AJ, Grace J, Matteucci G, Pa- pale D, Piao SL, Schulze E-D, Tang J, Law BE. 2010. Reduction of forest soil respiration in response to nitrogen deposition. Nat Geosci 3:315–22. Goll DS, Brovkin V, Parida BR, Reick CH, Kattge J, Reich PB, van Bodegom PM, Niinemets U¨ . 2012. Nutrient limitation reduces land carbon uptake in simulations with a model of combined carbon, nitrogen and phosphorus cycling. Biogeosciences 9:3547–69. Jia Y, Yu G, Gao Y, He N, Wang Q, Jiao C, Zuo Y. 2016. Global inorganic nitrogen dry deposition inferred from ground- and space-based measurements. Scientiﬁc Reports 6:19810. Gosheva S, Walthert L, Niklaus PA, Zimmermann S, Gimmi U, Hagedorn F. 2017. Reconstruction of historic forest cover changes indicates minor effects on carbon stocks in Swiss forest soils. Ecosystems 20:1512–28. Jonard M, Fu¨ rst A, Verstraeten A, Thimonier A, Timmermann V, Potocˇic´ N, Waldner P, Benham S, Hansen K, Merila¨ P, Ponette Q, de la Cruz AC, Roskams P, Nicolas M, Croise´ L, Ingerslev M, Matteucci G, Decinti B, Bascietto M, Rautio P. 2015. Tree mineral nutrition is deteriorating in Europe. Glob Change Biol 21:418–30. REFERENCES Aber J, McDowell W, Nadelhoffer K, Magill A, Berntson G, Kamakea M, McNulty S, Currie W, Rustad L, Fernandez I. 1998. Nitrogen saturation in temperate forest ecosystems. Bioscience 48:921–34. De Vries W, Du E, Butterbach-Bahl K. 2014. Short and long- term impacts of nitrogen deposition on carbon sequestration by forest ecosystems. Current Opinion in Environmental Sustainability 9–10:90–104. Aber JD, Nadelhoffer KJ, Steudler P, Melillo JM. 1989. Nitrogen saturation in northern forest ecosystems. Bioscience 39:378– 86. De Vries W, Reinds GJ, Gundersen P, Sterba H. 2006. The impact of nitrogen deposition on carbon sequestration in European forests and forest soils. Glob Change Biol 12:1151–73. Agren GI, Bosatta E, Magill A. 2001. Combining theory and experiment to understand effects of inorganic nitrogen on litter decomposition. Oecologia 128:94–8. De Vries W, Solberg S, Dobbertin M, Sterba H, Laubhann D, van Oijen M, Evans C, Gundersen P, Kros J, Wamelink GWW, Reinds GJ, Sutton MA. 2009. The impact of nitrogen deposi- tion on carbon sequestration by European forests and heath- lands. For Ecol Manage 258:1814–23. Angst G, Messinger J, Greiner M, Ha¨usler W, Hertel D, Kirfel K, Ko¨gel-Knabner I, Leuschner C, Rethemeyer J, Mueller CW. 2018. Soil organic carbon stocks in topsoil and subsoil con- trolled by parent material, carbon input in the rhizosphere, and microbial-derived compounds. Soil Biol Biochem 122:19–30. DeForest JL, Zak DR, Pregitzer KS, Burton AJ. 2004. Atmo- spheric nitrate deposition, microbial community composition, and enzyme activity in northern hardwood forests. Soil Sci Soc Am J 68:132–8. Bates D, Ma¨chler M, Bolker B, Walker S. 2015. Fitting linear mixed-effects models using lme4. J Stat Softw 67:1–48. Dixon RK, Brown S, Houghton RA, Solomon AM, Trexler MC, Wisniewsik J. 1994. Carbon pools and ﬂux of global forest ecosystems. Science 263:185–90. Berg B, Matzner E. 1997. Effect of N deposition on decomposi- tion of plant litter and soil organic matter in forest systems. Environmental Reviews 5:1–25. Ferna´ndez-Martı´nez M, Vicca S, Janssens IA, Sardans J, Luys- saert S, Campioli M, Chapin FSIII, Ciais P, Malhi Y, Ober- steiner M, Papale D, Piao SL, Reichstein M, Roda` F, Pen˜ uelas Blackwood CB, Waldrop MP, Zak DR, Sinsabaugh RL. 2007. Molecular analysis of fungal communities and laccase genes in decomposing litter reveals differences among forest types but S. J. Forstner and others 398 J. 2014. Nutrient availability as the key regulator of global forest carbon balance. Nature Climate Change 4:471–6. REFERENCES Gundersen P. 1998. Effects of enhanced nitrogen deposition in a spruce forest at Klosterhede, Denmark, examined by moder- ate NH4NO3 addition. For Ecol Manage 101:251–68. Gundersen P, Emmett BA, Kjønaas OJ, Koopmans CJ, Tietema A. 1998. Impact of nitrogen deposition on nitrogen cycling in forests: a synthesis of NITREX data. For Ecol Manage 101:37– 55. Knorr M, Frey SD, Curtis PS. 2005. Nitrogen additions and litter decomposition: a meta-analysis. Ecology 86:3252–7. Krause K, Cherubini P, Bugmann H, Schleppi P. 2012a. Growth enhancement of Picea abies trees under long-term, low-dose N addition is due to morphological more than to physiological changes. Tree Physiol 32:1471–81. Gundersen P, Rasmussen L. 1995. Nitrogen mobility in a nitro- gen limited forest at Klosterhede, Denmark, examined by NH4NO3 addition. For Ecol Manage 71:75–88. Krause K, Niklaus PA, Schleppi P. 2013. Soil-atmosphere ﬂuxes of the greenhouse gases CO2, CH4 and N2O in a mountain spruce forest subjected to long-term N addition and to tree girdling. Agric For Meteorol 181:61–8. Hagedorn F, Bucher JB, Schleppi P. 2001a. Contrasting dynamics of dissolved inorganic and organic nitrogen in soil and surface waters of forested catchments with Gleysols. Geoderma 100:173–92. Long-Term N Addition and Soil Organic C 399 vard Forest LTER, Massachusetts, USA. For Ecol Manage 196:7–28. Krause K, Providoli I, Currie WS, Bugmann H, Schleppi P. 2012b. Long-term tracing of whole catchment 15N additions in a mountain spruce forest: measurements and simulations with the TRACE model. Trees 26:1683–702. Ma¨kipa¨a¨ R. 1995. Effect of nitrogen input on carbon accumu- lation of boreal forest soils and ground vegetation. For Ecol Manage 79:217–26. Kuznetsova A, Brockhoff PB, Christensen RHB. 2017. lmerTest package: tests in linear mixed effects models. J Stat Softw 82(13):1–26. McNulty SG, Aber JD. 1993. Effects of chronic nitrogen additions on nitrogen cycling in a high-elevation spruce–ﬁr stand. Can J For Res 23:1252–63. Lal R. 2005. Forest soils and carbon sequestration. For Ecol Manage 220:242–58. McNulty SG, Aber JD, Newman SD. 1996. Nitrogen saturation in a high elevation New England spruce-ﬁr stand. For Ecol Manage 84:109–21. LeBauer DS, Treseder KK. 2008. Nitrogen limitation of net pri- mary productivity in terrestrial ecosystems is globally dis- tributed. Ecology 89:371–9. McNulty SG, Boggs J, Aber JD, Rustad L, Magill A. 2005. Red spruce ecosystem level changes following 14 years of chronic N fertilization. For Ecol Manage 219:279–91. Lenth RV. 2016. Least-squares means: the R package lsmeans. J Stat Softw 69:1–33. Leupold W. 1942. REFERENCES Neue Beobachtungen zur Gliederung der Flyschbildungen der Alpen zwischen Reuss und Rhein. Eclo- gae Geol Helvetiae 35:247–91. Minocha R, Long S, Magill AH, Aber J, McDowell WH. 2000. Foliar free polyamine and inorganic ion content in relation to soil and soil solution chemistry in two fertilized forest stands at the Harvard Forest, Massachusetts. Plant Soil 222:119–37. Leys C, Ley C, Klein O, Bernard P, Licata L. 2013. Detecting outliers: do not use standard deviation around the mean, use absolute deviation around the median. J Exp Soc Psychol 49:764–6. Mohn J, Schu¨ rmann A, Hagedorn F, Schleppi P, Bachofen R. 2000. Increased rates of denitriﬁcation in nitrogen-treated forest soils. For Ecol Manage 137:113–19. Li W, Jin C, Guan D, Wang Q, Wang A, Yuan F, Wu J. 2015. The effects of simulated nitrogen deposition on plant root traits: a meta-analysis. Soil Biol Biochem 82:112–18. Mu¨ ller CW, Ko¨gel-Knabner I. 2009. Soil organic carbon stocks, distribution, and composition affected by historic land use changes on adjacent sites. Biol Fertil Soils 45:347–59. Nadelhoffer KJ. 2000. The potential effects of nitrogen deposi- tion on ﬁne-root production in forest ecosystems. New Phytol 147:131–9. Litton CM, Giardina CP. 2008. Below-ground carbon ﬂux and partitioning: global patterns and response to temperature. Funct Ecol 22:941–54. Litton CM, Raich JW, Ryan MG. 2007. Carbon allocation in forest ecosystems. Glob Change Biol 13:2089–109. Nadelhoffer KJ, Aber JD, Melillo JM. 1985. Fine roots, net pri- mary production, and soil nitrogen availability: a new hypothesis. Ecology 66:1377–90. Liu L, Greaver TL. 2010. A global perspective on belowground carbon dynamics under nitrogen enrichment: belowground C dynamics under N enrichment. Ecol Lett 13:819–28. Nave LE, Vance ED, Swanston CW, Curtis PS. 2009. Impacts of elevated N inputs on north temperate forest soil C storage, C/ N, and net N-mineralization. Geoderma 153:231–40. Lovett GM, Arthur MA, Weathers KC, Fitzhugh RD, Templer PH. 2013. Nitrogen addition increases carbon storage in soils, but not in trees, in an Eastern U.S. Deciduous Forest. Ecosystems 16:980–1001. Niu S, Classen AT, Dukes JS, Kardol P, Liu L, Luo Y, Rustad L, Sun J, Tang J, Templer PH, Thomas RQ, Tian D, Vicca S, Wang Y-P, Xia J, Zaehle S. 2016. Global patterns and substrate-based mechanisms of the terrestrial nitrogen cycle. Ecol Lett 19:697– 709. Lovett GM, Goodale CL. 2011. A new conceptual model of nitrogen saturation based on experimental nitrogen addition to an oak forest. Ecosystems 14:615–31. REFERENCES No¨mmik H, Vahtras K. 1982. Retention and ﬁxation of ammo- nium and ammonia in soils. Nitrogen Agric Soils 22:123–71. Lu M, Yang Y, Luo Y, Fang C, Zhou X, Chen J, Yang X, Li B. 2011a. Responses of ecosystem nitrogen cycle to nitrogen addition: a meta-analysis. New Phytol 189:1040–50. Olsson P, Linder S, Giesler R, Ho¨gberg P. 2005. Fertilization of boreal forest reduces both autotrophic and heterotrophic soil respiration. Glob Change Biol 11:1745–53. Lu M, Zhou X, Luo Y, Yang Y, Fang C, Chen J, Li B. 2011b. Minor stimulation of soil carbon storage by nitrogen addition: a meta-analysis. Agr Ecosyst Environ 140:234–44. Oren R, Ellsworth DS, Johnsen KH, Phillips N, Ewers BE, Maier C, Scha¨fer KVR, McCarthy H, Hendrey G, McNulty SG, Katul GG. 2001. Soil fertility limits carbon sequestration by forest ecosystems in a CO2-enriched atmosphere. Nature 411:469–72. Lu X, Mao Q, Mo J, Gilliam FS, Zhou G, Luo Y, Zhang W, Huang J. 2015. Divergent responses of soil buffering capacity to long- term N deposition in three typical tropical forests with dif- ferent land-use history. Environ Sci Technol 49:4072–80. Pan YD, Birdsey RA, Fang JY, Houghton R, Kauppi PE, Kurz WA, Phillips OL, Shvidenko A, Lewis SL, Canadell JG, Ciais P, Jackson R, Pacala S, McGuire A, Piao S, Rautiainen A, Sitch S, Hayes D. 2011. A large and persistent carbon sink in the world’s forests. Science 333:988–93. Lucas RW, Klaminder J, Futter MN, Bishop KH, Egnell G, Lau- don H, Ho¨gberg P. 2011. A meta-analysis of the effects of nitrogen additions on base cations: implications for plants, soils, and streams. For Ecol Manage 262:95–104. Peng Y, Guo D, Yang Y. 2017. Global patterns of root dynamics under nitrogen enrichment. Glob Ecol Biogeogr 26:102–14. Lundstro¨m US, van Breemen N, Bain D. 2000. The podzolization process. A review. Geoderma 94:91–107. Poeplau C, Vos C, Don A. 2017. Soil organic carbon stocks are systematically overestimated by misuse of the parameters bulk density and rock fragment content. SOIL 3:61–6. MaarouﬁNI, Nordin A, Hasselquist NJ, Bach LH, Palmqvist K, Gundale MJ. 2015. Anthropogenic nitrogen deposition en- hances carbon sequestration in boreal soils. Glob Change Biol 21:3169–80. Pregitzer KS, Burton AJ, Zak DR, Talhelm AF. 2008. Simulated chronic nitrogen deposition increases carbon storage in Northern Temperate forests. Glob Change Biol 14:142–53. Magill AH, Aber JD, Currie WS, Nadelhoffer KJ, Martin ME, McDowell WH, Melillo JM, Steudler P. 2004. REFERENCES Ecosystem re- sponse to 15 years of chronic nitrogen additions at the Har- R Core Team. 2016. R: a language and environment for statistical computing. Vienna, Austria. https://www.R-project.org. S. J. Forstner and others 400 forest—evidence from C, N and 14C content. For Ecol Manage 263:131–7. Raich JW, Nadelhoffer KJ. 1989. Belowground carbon allocation in forest ecosystems: global trends. Ecology 70:1346–54. forest—evidence from C, N and 14C content. For Ecol Manage 263:131–7. Rasse DP, Rumpel C, Dignac M-F. 2005. Is soil carbon mostly root carbon? Mechanisms for a speciﬁc stabilisation. Plant Soil 269:341–56. Thomas RQ, Canham CD, Weathers KC, Goodale CL. 2010. In- creased tree carbon storage in response to nitrogen deposition in the US. Nat Geosci 3:13–17. Reay DS, Dentener F, Smith P, Grace J, Feely RA. 2008. Global nitrogen deposition and carbon sinks. Nat Geosci 1:430–7. Thorn KA, Mikita MA. 1992. Ammonia ﬁxation by humic sub- stances: a nitrogen-15 and carbon-13 NMR study. Sci Total Environ 113:67–87. Rousseeuw PJ, Croux C. 1993. Alternatives to the median absolute deviation. Journal of the American Statistical Asso- ciation 88:1273–83. Tian D, Niu S. 2015. A global analysis of soil acidiﬁcation caused by nitrogen addition. Environmental Research Letters 10:024019. Schaberg PG, Perkins TD, McNulty SG. 1997. Effects of chronic low-level N additions on foliar elemental concentrations, morphology, and gas exchange of mature montane red spruce. Can J For Res 27:1622–9. Townsend AR, Braswell BH, Holland EA, Penner JE. 1996. Spatial and temporal patterns in terrestrial carbon storage due to deposition of fossil fuel nitrogen. Ecol Appl 6:806–14. van Diepen LTA, Frey SD, Sthultz CM, Morrison EW, Minocha R, Pringle A. 2015. Changes in litter quality caused by simu- lated nitrogen deposition reinforce the N-induced suppression of litter decay. Ecosphere 6:1–16. Schleppi P, Bucher-Wallin L, Siegwolf R, Saurer M, Muller N, Bucher JB. 1999a. Simulation of increased nitrogen deposi- tion to a montane forest ecosystem: partitioning of the added 15N. Water Air Soil Pollut 116:129–34. Schleppi P, Curtaz F, Krause K. 2017. Nitrate leaching from a sub-alpine coniferous forest subjected to experimentally in- creased N deposition for 20 years, and effects of tree girdling and felling. Biogeochemistry 134:319–35. van Diepen LTA, Frey SD, Landis EA, Morrison EW, Pringle A. 2016. Fungi exposed to chronic nitrogen enrichment are less able to decay leaf litter. Ecology 98:5–11. REFERENCES Vet R, Artz RS, Carou S, Shaw M, Ro C-U, Aas W, Baker A, Bowersox VC, Dentener F, Galy-Lacaux C, Hou A, Pienaar JJ, Gillett R, Forti MC, Gromov S, Hara H, Khodzher T, Maho- wald NM, Nickovic S, Rao PSP, Reid NW. 2014. A global assessment of precipitation chemistry and deposition of sulfur, nitrogen, sea salt, base cations, organic acids, acidity and pH, and phosphorus. Atmos Environ 93:3–100. Schleppi P, Muller N, Edwards PJ, Bucher JB. 1999b. Three years of increased nitrogen deposition do not affect the vegetation of a montane forest ecosystem. Phyton 39:197–204. Schleppi P, Muller N, Feyen H, Papritz A, Bucher JB, Flu¨ hler H. 1998. Nitrogen budgets of two small experimental forested catchments at Alptal, Switzerland. For Ecol Manage 101:177– 85. Vogt KA, Grier CC, Vogt D. 1986. Production, turnover, and nutrient dynamics of above-and belowground detritus of world forests. Adv Ecol Res 15:303–77. Schro¨der WH, Bauch J, Endeward R. 1988. Microbeam analysis of Ca exchange and uptake in the ﬁne roots of spruce: inﬂu- ence of pH and aluminum. Trees 2:96–103. Vogt KA, Vogt DJ, Palmiotto PA, Boon P, O’Hara J, Asbjornsen H. 1995. Review of root dynamics in forest ecosystems grouped by climate, climatic forest type and species. Plant Soil 187:159–219. Schulze E. 1989. Air pollution and forest decline in a spruce (Picea abies) forest. Science 244:776. Simpson D, Andersson C, Christensen JH, Engardt M, Geels C, Nyiri A, Posch M, Soares J, Soﬁev M, Wind P, Langner J. 2014. Impacts of climate and emission changes on nitrogen deposition in Europe: a multi-model study. Atmos Chem Phys 14:6995–7017. Wang C, Han S, Zhou Y, Yan C, Cheng X, Zheng X, Li M-H. 2012. Responses of ﬁne roots and soil N availability to short- term nitrogen fertilization in a broad-leaved Korean pine mixed forest in Northeastern China. PLoS ONE 7:e31042. Xia J, Wan S. 2008. Global response patterns of terrestrial plant species to nitrogen addition. New Phytol 179:428–39. Singmann H, Bolker B, Westfall J, Højsgaard S, Fox J, Lawrence M, others. 2015. afex: analysis of factorial experiments. R package version 013-145. Xia M, Talhelm AF, Pregitzer KS. 2018. Long-term simulated atmospheric nitrogen deposition alters leaf and ﬁne root decomposition. Ecosystems 21:1–14. Sinsabaugh RL. 2010. Phenol oxidase, peroxidase and organic matter dynamics of soil. Soil Biol Biochem 42:391–404. Xu G-L, Schleppi P, Li M-H, Fu S-L. 2009. REFERENCES Negative responses of Collembola in a forest soil (Alptal, Switzerland) under experimentally increased N deposition. Environ Pollut 157:2030–6. Sollins P, Homann P, Caldwell BA. 1996. Stabilization and destabilization of soil organic matter: mechanisms and con- trols. Geoderma 74:65–105. Sun T, Dong L, Wang Z, Lu¨ X, Mao Z. 2016. Effects of long-term nitrogen deposition on ﬁne root decomposition and its extracellular enzyme activities in temperate forests. Soil Biol Biochem 93:50–9. Yuan ZY, Chen HYH. 2010. Fine root biomass, production, turnover rates, and nutrient contents in boreal forest ecosys- tems in relation to species, climate, fertility, and stand age: literature review and meta-analyses. Crit Rev Plant Sci 29:204–21. Talbot JM, Allison SD, Treseder KK. 2008. Decomposers in dis- guise: mycorrhizal fungi as regulators of soil C dynamics in ecosystems under global change. Funct Ecol 22:955–63. Yue K, Peng Y, Peng C, Yang W, Peng X, Wu F. 2016. Stimu- lation of terrestrial ecosystem carbon storage by nitrogen addition: a meta-analysis. Scientiﬁc Reports 6:19895. Tamm CO. 1991. Nitrogen-limited and nitrogen-depleted ter- restrial ecosystems: ecological characteristics. Nitrogen in ter- restrial ecosystems. Ecological studies. Berlin: Springer. p 34– 49. Zak DR, Holmes WE, Burton AJ, Pregitzer KS, Talhelm AF. 2008. Simulated atmospheric NO3-deposition increases soil organic matter by slowing decomposition. Ecol Appl 18:2016–27. Tefs C, Gleixner G. 2012. Importance of root derived carbon for soil organic matter storage in a temperate old-growth beech
https://openalex.org/W4200436314	https://amt.copernicus.org/articles/15/1333/2022/amt-15-1333-2022.pdf	English	null	Reply on RC2	null	2,021	cc-by	16,381	Correspondence: Alexander Myagkov (alexander.myagkov@radiometer-physics.de) Received: 24 July 2021 – Discussion started: 18 August 2021 Revised: 8 January 2022 – Accepted: 28 January 2022 – Published: 14 March 2022 Received: 24 July 2021 – Discussion started: 18 August 2021 Revised: 8 January 2022 – Accepted: 28 January 2022 – Published: 14 March 2022 Abstract. This study presents the ﬁrst-ever complete char- acterization of random errors in dual-polarimetric spectral observations of meteorological targets by cloud radars. The characterization is given by means of mathematical equa- tions for joint probability density functions (PDFs) and error covariance matrices. The derived equations are checked for consistency using real radar measurements. One of the main conclusions of the study is that the convenient representation of spectral polarimetric measurements including differential reﬂectivity ZDR, correlation coefﬁcient ρHV , and differential phase 8DP is not suited for the proper characterization of the error covariance matrix. This is because the aforementioned quantities are complex, non-linear functions of the radar raw data, and thus their error covariance matrix is commonly de- rived using simpliﬁed linear relations and by neglecting the correlation of errors. This study formulates the spectral po- larimetric measurements in terms of a different set of quan- tities that allows for a proper analytic treatment of their er- ror covariance matrix. The results given in this study allow for utilization of spectral polarimetric measurements for ad- vanced meteorological applications, among which are varia- tional retrieval techniques, data assimilation, and sensitivity analysis. can independently characterize hydrometeors coexisting in the same volume but moving with different speeds relative to the radar (Kollias et al., 2007). Second, the high sensi- tivity and vast dynamic range make cloud radars capable of measuring return signals from a wide range of particle sizes, which is a challenging task for other instruments like lidars (Bühl et al., 2013). Third, due to relatively low atten- uation of microwave signals by liquid water, cloud radars proﬁle clouds up to the top even in the presence of light to moderate rain. These capabilities promote cloud radars for investigation of different formation and development pro- cesses throughout the life cycle of clouds. Correspondence: Alexander Myagkov (alexander.myagkov@radiometer-physics.de) For instance, cloud radars help to characterize initial ice formation and devel- opment in mixed-phase clouds (Bühl et al., 2019a, b), im- prove characterization of pure liquid clouds (Rusli et al., 2017; Acquistapace et al., 2017), estimate rates of aggre- gation (Kneifel et al., 2015, 2016) and riming (Kalesse et al., 2016; Moisseev et al., 2017; Kneifel and Moisseev, 2020), and quantitatively analyze solid and liquid precipita- tion (Matrosov, 2005; Matrosov et al., 2006, 2008; Tridon and Battaglia, 2015; Tridon et al., 2017, 2019). Many cloud radars have dual-polarization capabilities. An interest in polarimetry-based methods in the cloud radar community has been growing, which is indicated by a num- ber of studies during the last decade (Matrosov et al., 2012; Oue et al., 2015; Lu et al., 2015; Myagkov et al., 2016a, b; Matrosov et al., 2017; Oue et al., 2018; Myagkov et al., 2020). Vertically pointed cloud radars often operate in the LDR (linear depolarization ratio) mode; i.e., they transmit a linearly polarized wave (either horizontally or vertically) and receive co- and cross-polarized components of the backscat- tered signal (e.g., Görsdorf et al., 2015). The LDR mode 1 Introduction Cloud radars are a major component of state-of-the-art, ground-based observation platforms (Illingworth et al., 2007; Kollias et al., 2020). Their unique capabilities make these in- struments extremely valuable for cloud and precipitation re- search. First, these radars have Doppler capabilities; i.e., they Atmos. Meas. Tech., 15, 1333–1354, 2022 https://doi.org/10.5194/amt-15-1333-2022 © Author(s) 2022. This work is distributed under the Creative Commons Attribution 4.0 License. Atmos. Meas. Tech., 15, 1333–1354, 2022 https://doi.org/10.5194/amt-15-1333-2022 © Author(s) 2022. This work is distributed under the Creative Commons Attribution 4.0 License. A. Myagkov and D. Ori: Analytic characterization of random errors This allows for using a similar approach in analytic characterization of errors in spectral polarimetric observations. Operational precipitation radars are used by weather ser- vices to continuously scan the atmosphere, providing polari- metric variables integrated for a scattering volume. In addi- tion to the integrated quantities, cloud radars with the hy- brid mode enable spectrally resolved polarimetric observa- tions and, therefore, can provide the same set of polarimetric variables for different types of cloud particles coexisting in the same resolution volume (Oue et al., 2015; Myagkov et al., 2016b, 2020). Spectral observations are in general possible with precipitation radars (Spek et al., 2008; Dufournet and Russchenberg, 2011; Pﬁtzenmaier et al., 2018). Such mea- surements, however, are not performed by operational radars due to fast azimuth scanning. p p A number of studies (e.g., Hogan, 2007; Cao et al., 2013; Yoshikawa et al., 2014; Chang et al., 2016; Huang et al., 2020) characterize the joint PDF of polarimet- ric radar measurements by the error covariance matrix. There are, however, problems with existing approximations of the error covariance matrix for polarimetric observa- tions. First, the elements in the main diagonal of the er- ror covariance matrix – variances of random errors – are found using the ﬁrst-order Taylor approximation following Bringi and Chandrasekar (2001). Conventional polarimetric variables such as differential reﬂectivity, correlation coefﬁ- cient, and differential phase are, however, highly non-linear functions. Therefore, the approximation may lead to biases in the error variance estimates, especially when signal-to- noise ratios (SNRs) and/or the number of averaged samples is low. This problem becomes important for cloud radars col- lecting polarimetric variables with a high spatial, temporal, and spectral resolution. Second, non-diagonal components of the error covariance matrix are typically set to zero as- suming no correlation between errors in measured quantities, but validity and effects of this assumption are not discussed. The information content of measurements is, however, higher when errors are correlated (chap. 3.2.6 in Rodgers, 2000), and therefore non-negligible off-diagonal elements of the co- variance matrix should not be ignored. Spectral polarimetry can be used for a development of ad- vanced retrieval methods. For example variational retrievals developed for dual-frequency spectra (Tridon and Battaglia, 2015; Tridon et al., 2017) could be applied also to spectral polarimetry. Moisseev and Chandrasekar (2007) presented ﬁrst attempts to retrieve proﬁles of raindrop size distributions using polarimetric spectra from a precipitation radar. A. Myagkov and D. Ori: Analytic characterization of random errors 1334 is efﬁcient for clutter removal and detection of the melt- ing layer and columnar-shaped ice particles. As shown by Matrosov et al. (2001), however, the applicability of the LDR mode at low elevation angles might be limited due to its high sensitivity to the orientation of cloud particles. Therefore, scanning polarimetric cloud radars often have polarimetric modes which are less sensitive to the orientation. One such mode is the hybrid mode (also denoted as the STSR (simul- taneous transmission and simultaneous reception) or STAR (simultaneous transmission and reception) mode in the lit- erature). Radars with the hybrid mode emit the horizontal and vertical components of the transmitted wave simulta- neously (Myagkov et al., 2015; Bringi and Chandrasekar, 2001, Sect. 4.7). Cloud radars with the hybrid mode allow for adoption of polarimetry-based methods developed during the last several decades for centimeter-wavelength meteoro- logical radars (further denoted as precipitation radars). systematic and random measurement errors. The former type of errors is solved by a calibration. Calibration aspects of po- larimetric quantities have been intensively studied for both precipitation and cloud radars (Chandrasekar et al., 2015) and are out of the scope of this study. In the case of radar observations of meteorological targets, random errors can be characterized from measurements if raw (unaveraged) data are available. Cloud radars, however, rarely store raw data because of the high data rate. Therefore, commonly used ap- proaches to characterize random errors are based on statis- tical models of the received radar signals. Random errors in radar signals can be represented by a joint probability density function (PDF) of amplitudes and phases in the two orthog- onal polarimetric channels. The joint PDF for polarimetric observations obtained for a single pulse can be found in Mid- dleton (1996, chap. 9.2). Single-pulse measurements, how- ever, are rarely used in the radar meteorology because of the low sensitivity and higher requirement for storage space. The observed radar spectra almost always result from the averag- ing of a number of return pulses. Lee et al. (1994) showed a derivation of a joint probability density function of polari- metric variables for the case of averaging. The authors used a number of assumptions applicable for Earth’s surface ob- servations using synthetic-aperture radars. It turns out that the same assumptions are applicable to spectral polarimetric observations of meteorological targets. https://doi.org/10.5194/amt-15-1333-2022 Published by Copernicus Publications on behalf of the European Geosciences Union. Published by Copernicus Publications on behalf of the European Geosciences Union. A. Myagkov and D. Ori: Analytic characterization of random errors plitudes of the analyzed spectral component in the horizontal and vertical channels, respectively (the dot hereafter denotes a complex quantity). valid for spectral measurements we extend the error model available in the literature and thus derive mathematical ex- pressions characterizing random errors in spectral polarimet- ric observations of meteorological targets. The study is or- ganized as follows. We review the measurement method of spectral polarimetry with radars operating in the hybrid mode in Sect. 2. In Sect. 3 the likelihood functions of the common polarimetric radar variables are rigorously derived. The error covariance matrix of polarimetric measurements is derived in Sect. 4 by taking into account the correlations among the various measurement random errors. In Sect. 5 the validity of expressions derived for the likelihood functions and error covariance matrix is checked using real raw measurements from a cloud radar. 2.2 Coherency of complex amplitudes in range and velocity domain Different range bins as well as different spectral components are often considered to be statistically independent because the corresponding complex amplitudes result from non- coherent scattering of numerous independently moving parti- cles. Some correlation, however, can be expected due to sam- pling effects and the FFT spectral leakages (e.g., Sect. 5.3 in Marple, 2019). For instance, the power scattered from parti- cles located close to the end of a range bin is distributed be- tween this and the following range bins. These effects depend on ﬁlter properties and used FFT windows. It is challeng- ing to give a general analytical solution taking these effects into account. Therefore, these effects are out of the scope of this study. For the sake of simplicity the following analysis is shown only for a single range bin and a single spectral com- ponent. Since movements of particles in neighboring range and spectral bins are not related, statistical properties of an individual bin considered in the following are not affected by sampling effects and spectral leakages. The neglection of the dependence of the neighboring bins (due to sampling effects and spectral leakages) leads to an underestimation of the in- formation entropy when a complete spectrum and/or spectral proﬁle is analyzed. This worst case assumption, however, al- lows for a relatively easy and universal characterization of measurement errors. Future studies may improve the error characterization by considering the sampling and leakage ef- fects. 2 Basics of radar spectral polarimetry in the hybrid mode This section introduces known relations between a raw cloud radar signal, complex amplitudes, and spectral polarimetric variables for observations of meteorological targets. These relations are based on the same set of assumptions in- troduced in classical works of Doviak et al. (1979) and Bringi and Chandrasekar (2001) for precipitation radars. Note that since pulsed radars are currently more common in the meteorological community, we use the term “pulse” to refer to a type of the transmitted radar signal in Sects. 2– 4. For radars with frequency-modulated continuous wave (FMCW) signals, however, the term “chirp” should be used. Later, in Sect. 5 we use measurements from a FMCW radar, and therefore the term “chirp” is used there. Atmos. Meas. Tech., 15, 1333–1354, 2022 A. Myagkov and D. Ori: Analytic characterization of random errors This ap- proach, however, has not yet been explored in polarimetric cloud radars. Recent review studies (Zhang et al., 2019; Morrison et al., 2020; Ryzhkov et al., 2020) demonstrate that polarimetric observations from precipitation radar networks are highly beneﬁcial for the evaluation and development of numerical weather prediction and cloud resolving models. The high value of polarimetric observations is given by their sensi- tivity to microphysical properties of cloud and precipitation particles such as size, shape, number concentration, state of matter, density, and orientation (Kumjian, 2013). Polarimet- ric cloud radars are not yet widely used for model improve- ment. This, however, does not indicate that cloud radar po- larimetry is not informative relative to precipitation radars. Conversely, the cloud radar spectral polarimetry can essen- tially complement available measurements. This study is based on well-known statistical properties of polarimetric radar signals. Using certain simpliﬁcations The development of both quantitative retrievals and data assimilation algorithms requires the characterization of the Atmos. Meas. Tech., 15, 1333–1354, 2022 https://doi.org/10.5194/amt-15-1333-2022 1335 A. Myagkov and D. Ori: Analytic characterization of random errors Introduce a measurement column vector In this study ZDR, ρHV, and 8DP are deﬁned for each spectral line using elements of corresponding B: 6m =   σ 2 h 0 qσhσv sσhσv 0 σ 2 h −sσhσv qσhσv qσhσv −sσhσv σ 2 v 0 sσhσv qσhσv 0 σ 2 v ,  , (3) (3) ZDR = Bhh Bvv , (10) ρHV = s R2 hv + J 2 hv BhhBvv , (11) 8DP = atan −Jhv Rhv . (12) (10) where σh is the standard deviation of ˆRh and ˆJh, σv is the standard deviation of ˆRv and ˆJv, q is the correlation between ˆRh and ˆRv, and s is the correlation between ˆRh and ˆJv. (11) (12) 2.4 Statistical properties of complex amplitudes the overline indicates the expected value; Bhh and Bvv repre- sent total powers of the horizontal and vertical components of the received signal, respectively; ˙Bhv is the covariance be- tween the horizontal and vertical components of the received signal; and ∗is the complex conjugation sign. Note that in general Bhh, Bvv, and real and imaginary parts of ˙Bhv can be calibrated in any quantity that is proportional to the power (watts) received by the radar, e.g., classical radar reﬂectivity (mm6 m−3) or even arbitrary units (Myagkov et al., 2016a). Recall that in this study the covariance matrix B corresponds to a single spectral component. Such spectral representation of vector signals was introduced by Wiener (1930). Introduce a measurement column vector ˆm = [ ˆRh, ˆJh, ˆRv, ˆJv]T (1) ˆm = [ ˆRh, ˆJh, ˆRv, ˆJv]T (1) with ˆR and ˆJ being real and imaginary parts of a complex amplitude ˙S, indices h and v denoting the polarization state, and T being the transposition sign; the circumﬂex is used hereafter to emphasize measured quantities. The probability density function (PDF) of ˆm, given the true covariance ma- trix 6m of ˆm, can be written as follows: fm ˆm\|6m = (2π)−2det(6m)−1 2 e−1 2 ˆmT6m ˆm. (2) (2) The elements of B are related to the statistics of the com- plex amplitudes ˙Sh and ˙Sv as follows: Note that throughout the study a PDF is a function of measured quantities (e.g., ˆm in Eq. 2) with ﬁxed parame- ters (e.g., 6m in Eq. 2). The same PDF is called a likelihood function if the measured quantities are ﬁxed, and the PDF is viewed as a function of parameters. Bhh = var( ˆRh) + var( ˆJh) = 2σ 2 h , (6) Bvv = var( ˆRv) + var( ˆJv) = 2σ 2 v , (7) ˙Bhv = Rhv + iJhv = (q + js)σhσv, (8) (6) (7) (8) (8) Doviak et al. (1979) showed that for meteorological tar- gets I and Q components are jointly normal with zero mean, zero correlation, and equal standard deviation. The authors explain that these properties are due to scattering from a large number of particles moving in an unpredictable way in a scat- tering volume. Since Nfft is much smaller than the number of particles in a resolution volume, the properties are also valid for relations between ˆRh and ˆJh and between ˆRv and ˆJv. where Rhv and Jhv are real and imaginary parts of ˙Bhv In the precipitation radar community, dual-polarized mea- surements are rarely represented by B. Instead a set of po- larimetric variables are used. Therefore, the same polarimet- ric variables (but spectrally resolved) are introduced in this study. Introduce a vector c = (Bhh,ZDR,ρHV,8DP)T, (9) (9) The measured complex amplitudes ˙Sh and ˙Sv, however, can be correlated. Taking these properties into account, the true covariance matrix 6m is deﬁned in the following way (Eq. 5.178 in Bringi and Chandrasekar, 2001): where ZDR is the differential reﬂectivity, ρHV is the corre- lation coefﬁcient, and 8DP is the differential phase. 2.3 Coherency of complex amplitudes in time domain Radar polarimetric measurements are made on an orthogo- nal measurement basis deﬁned by feeders of the antenna sys- tem. In the hybrid mode the measurement basis is typically Cartesian and formed by the horizontal (h) and vertical (v) components. Further this basis is denoted as the h–v basis. Dual-polarimetric cloud radars have two receivers dedicated to the orthogonal polarimetric components of the received signal. For each transmitted pulse the receivers provide range proﬁles of in-phase Ih,v and quadrature Qh,v components, where indices h and v denote the polarization state. Note that this study does not cover the radar signal processing to get the Ih,v and Qh,v proﬁles. This information can be found in a radar handbook (e.g., Skolnik, 2008, chap. 6). Using Nfft proﬁles of Ih + iQh and Iv + iQv, where i is the imaginary unit, the radar calculates complex Doppler spectra in the hor- izontal and vertical channel, respectively, applying the fast Fourier transformation (FFT) along the time dimension. The complex Doppler spectra are represented by complex ampli- tudes ˙S for each spectral component and each range bin. In the following, ˙Sh and ˙Sv denote the measured complex am- Unlike precipitation radars which perform rapid azimuth scans, cloud radars are typically pointed to a certain direc- tion or make slow scans to get non-broadened Doppler spec- tra. Doviak et al. (1979) showed (Eq. 5.2 therein) that the co- herency between the adjacent samples depends on the wave- length and the sample repetition period. Cloud radars typ- ically have the pulse repetition frequency on the order of 10 kHz and Nftt in the range of 128 to 1024. This results in getting a single spectrum every 0.01–0.1 s. For such sam- pling properties of cloud radars any signiﬁcant coherency be- tween adjacent samples of a spectral line requires the spec- tral broadening not exceeding at most a few centimeters per second. The turbulent spectral broadening, however, ex- ceeds a few centimeters per second even in stratiform non- precipitating clouds (Borque et al., 2016). Therefore, con- secutive samples of complex amplitudes for a spectral line can be considered to be independent. Atmos. Meas. Tech., 15, 1333–1354, 2022 https://doi.org/10.5194/amt-15-1333-2022 1336 https://doi.org/10.5194/amt-15-1333-2022 3 Likelihood of elements of the covariance matrix B Any measurement is affected by inherent uncertainty. As many other measurement devices, radars also attempt to re- duce uncertainty in measurements by means of an average over multiple independent samples. The result of the aver- age maximizes the likelihood of the measurements, while the characterization of the distribution of the observations yields an estimate of the uncertainty in the measurements. In Sect. 3.1 we change the basis from h–v to the one on which elements of the vector ˆb become independent. On the new basis, the joint multivariate likelihood function can be represented by the product of the likelihood functions of each independent element. The likelihood of a single indepen- dent element is relatively simple to describe analytically. In Sect. 3.2 a formal derivation of the likelihood function on this new basis is provided. The solution for fb(ˆb\|b,Ns) is given in Sect. 3.3 converting back to the original space and applying the rule of change in variables. As mentioned above, the radar observations are often represented by the vector c. Therefore, Sect. 3.3 also provides the likelihood fc(ˆc\|b,Ns) for the con- ventional representation of polarimetric measurements. Assume the following problem. The state of the atmo- sphere is represented by the state vector x. A forward model F maps x into a vector F(x) = b = (Bhh,Rhv,Jhv,Bvv)T (13) (13) in the space of observations. The actual measurement vector is in the space of observations. The actual measurement vector is Note that in this section we keep only equations required to understand the principle of the derivation. The extensive calculus required to prove the formulas used in the section is provided in the Appendix. ˆb = ( ˆBhh, ˆRhv, ˆJhv, ˆBvv)T = b + ϵ, (14) (14) where A. Myagkov and D. Ori: Analytic characterization of random errors 1337 of fb(ˆb\|b,Ns) provided in this section is developed in four steps. 3.1 Step 1: change in basis and diagonalization of the covariance matrix B ˆBhh = ⟨˙Sh ˙S∗ h⟩, (15) ˆRhv = Re ⟨˙Sh ˙S∗ v⟩ , (16) ˆJhv = Im ⟨˙Sh ˙S∗ v⟩ ,and (17) ˆBvv = ⟨˙Sv ˙S∗ v⟩ (18) ˆBhh = ⟨˙Sh ˙S∗ h⟩, (15) ˆRhv = Re ⟨˙Sh ˙S∗ v⟩ , (16) ˆJhv = Im ⟨˙Sh ˙S∗ v⟩ ,and (17) ˆBvv = ⟨˙Sv ˙S∗ v⟩ (18) As previously mentioned, ˙Sh and ˙Sv are, in general, corre- lated. There is, however, always a basis on which the pro- jections of ˙Sh and ˙Sv become completely uncorrelated. This basis is further denoted as the c–x (co-polar and cross-polar) basis. The conversion of the vector e on the h–v basis to the vector eD on c–x basis is made using the unitary operator Q: (18) are constituents of the measured covariance matrix ˆB, and ϵ represents the vector of measurement random er- rors in each component of ˆb. In Eqs. (15)–(18) Re and Im are the real and imaginary parts of a complex num- ber; <> denotes averaging over Ns independent complex spectra calculated from non-overlapping time sequences. The estimators Eqs. (15)–(18) are the same as given in Bringi and Chandrasekar (2001, chap. 6.4.5). The only dif- ference is that within this work the variables are calculated using complex amplitudes for a spectral line instead of us- ing in-phase and quadrature components (I/Q hereafter) as is done by precipitation radars. What is the likelihood of ˆb given the state vector x? In the case that the forward model provides a unique and accurate relation between x and b, the problem is equivalent to ﬁnding fb(ˆb\|b,Ns) – the likelihood of ˆb – given the true vector of measurements b and the num- ber of averaged spectra Ns. are constituents of the measured covariance matrix ˆB, and ϵ represents the vector of measurement random er- rors in each component of ˆb. In Eqs. (15)–(18) Re and Im are the real and imaginary parts of a complex num- ber; <> denotes averaging over Ns independent complex spectra calculated from non-overlapping time sequences. The estimators Eqs. (15)–(18) are the same as given in Bringi and Chandrasekar (2001, chap. 6.4.5). The only dif- ference is that within this work the variables are calculated using complex amplitudes for a spectral line instead of us- ing in-phase and quadrature components (I/Q hereafter) as is done by precipitation radars. 2.5 Polarimetric variables Note that elements of the matrix B are in general af- fected by noise. The noise in both polarimetric chan- nels is not known exactly. Typically, it is estimated from spectra using, for example, the algorithm from Hildebrand and Sekhon (1974). A subtraction of noise lev- els from corresponding diagonal terms of the covariance ma- trix B to get an estimate of signal-only powers leads to oc- casions when the covariance matrix is no longer positively semi-deﬁnite. In this case, ρHV calculated from the noise- corrected covariance matrix can exceed 1, which is beyond the range of valid values. In order to avoid this problem, we characterize radar measurements without noise subtraction. A further advantage of this approach is that spectral lines containing noise only can also be correctly characterized. Since for meteorological targets ˆRh is not correlated with ˆJh, and ˆRv is not correlated with ˆJv, the absolute phases of ˙Sh and ˙Sv are uniformly distributed from 0 to 2π and thus un- informative. Therefore, the polarimetric observations in the hybrid mode can be represented by a 2 × 2 covariance ma- trix B (Eq. 4.130 in Bringi and Chandrasekar, 2001) instead of the true covariance matrix 6m: B = eeT = Bhh ˙Bhv ˙B∗ hv Bvv , (4) (4) where e = ( ˙Sh, ˙Sv)T; (5) Atmos. Meas. Tech., 15, 1333–1354, 2022 Atmos. Meas. Tech., 15, 1333–1354, 2022 https://doi.org/10.5194/amt-15-1333-2022 Atmos. Meas. Tech., 15, 1333–1354, 2022 Atmos. Meas. Tech., 15, 1333–1354, 2022 Myagkov and D. Ori: Analytic characterization of random errors 3.3 Step 3: likelihood function on the h–v basis By deﬁnition, the off-diagonal elements of the covariance matrix D are zeros (see Eq. 20). This implies no correla- tion between ˙Sc and ˙Sx. In this case, the likelihood function fd(ˆd\|b,Ns), where In the previous step, the likelihood function of measurements represented on the c–x basis was derived. In this subsection we perform a transformation back from the c–x basis to the original h–v basis that allows for the comparison of radar measurements in a common orthogonal reference frame. ˆd = ˆDcc, ˆRcx, ˆJcx, ˆDxx T , (28) (28) Applying the rule of changing variables in a multivariate PDF (e.g., Walpole et al., 2012, Theorem 7.4), fb(ˆb\|b,Ns) can be found from Eq. (29) as follows: can be written as a multiplication of likelihood functions of individual components: fb(ˆb\|b,Ns) = \|Jbd\|fd(ˆd\|b,Ns). (36) (36) As shown in Appendix E, the determinant of the Jacobian Jbd of the transformation from ˆb to ˆd is equal to 1. As shown in Appendix E, the determinant of the Jacobian Jbd of the transformation from ˆb to ˆd is equal to 1. fd(ˆd\|b,Ns) =f ˆDcc\|b,Ns f ˆRcx\|b,Ns ×f ˆJcx\|b,Ns f ˆDxx\|b,Ns . (29) (29) ˆD = Q† ˆBQ. (27) where χ2 k is the chi-squared distribution with k degrees of freedom, Note that the operator Q here is the same as in Eq. (20) and not recalculated using ˆB. a = (2Ns + 1)/2, (34) b = (1 −2Ns)/2, (35) (34) (34) (35) (34) (35) 3.2 Step 2: likelihood function of the measurements on the c–x basis (35) 0 is the gamma function, and Kµ is the Bessel function of the second kind of order µ. Recall that σc and σx in Eqs. (30)– (33) are derived from the elements of b using Eqs. (21)–(22) and Eqs. (24) and (25). Derivation and Monte Carlo evalua- tion of Eqs. (30)–(33) are given in Appendix B. Appendix B3 shows how to handle Eqs. (32) and (33) when ˆRcx and ˆJcx are close to 0. In the previous step, measurements were represented on a new – c–x – basis on which the orthogonal components of the measurement vector are independent. On this basis the joint multivariate likelihood function can be represented as a product of likelihood functions with a single element as an argument. This allows for a relatively easy mathematical description of the likelihood function on the c–x basis. 3.1 Step 1: change in basis and diagonalization of the covariance matrix B What is the likelihood of ˆb given the state vector x? In the case that the forward model provides a unique and accurate relation between x and b, the problem is equivalent to ﬁnding fb(ˆb\|b,Ns) – the likelihood of ˆb – given the true vector of measurements b and the num- ber of averaged spectra Ns. ˆ eD = ˙Sc ˙Sx = Qe. (19) (19) The calculation of the matrix Q is given in Appendix A. Real and imaginary parts of ˙Sc are jointly distributed normally with the zero mean, zero correlation, and standard deviation σc. Real and imaginary parts of ˙Sx are also jointly distributed normally with zero mean and zero correlation but have, in general, a different standard deviation σx. A transformation from the basis h–v to the basis c–x also changes the covariance matrix of the measurements. The co- variance matrix D of measurements on the c–x basis is diag- onal and can be found as follows: D = Dcc 0 0 Dxx = Q†BQ. (20) (20) In a general case, elements of the vector ˆb can be corre- lated. In this case the derivation of the likelihood function fb(ˆb\|b,Ns) is challenging. In order to simplify the deriva- tion, we follow an approach identical to the one demonstrated in Rodgers (2000, Sect. 2.3.1 therein). The author considers a multivariate PDF with correlated errors. He transforms the coordinate system in such a way that orthogonal compo- nents of the error vector are independent (uncorrelated). On this basis, the joint PDF can be represented by the product of independent, univariate PDFs for each individual com- ponent. Thus, following a similar approach, the derivation In Eq. (20) † is the Hermitian conjugate. Zero off-diagonal terms in D indicate that there is no correlation between the orthogonal components, i.e., ˙Sc and ˙Sx. Expanding Eq. (20), the elements of the matrix D can be found as follows: Dcc = q2 11Bhh + \|˙q12\|2Bvv −2q11 (R12Rhv + J12Jhv) (21) Dxx = \|˙q12\|2Bhh + q2 11Bvv + 2q11 (R12Rhv + J12Jhv), (22) where ˙qnm represents elements of Q, with n and m being indices of row and column, respectively; ˙q12 = R12 + iJ12. (23) (23) ˙q12 = R12 + iJ12. https://doi.org/10.5194/amt-15-1333-2022 Atmos. Meas. Tech., 15, 1333–1354, 2022 Atmos. Meas. Tech., 15, 1333–1354, 2022 A. Myagkov and D. Ori: Analytic characterization of random errors 1338 dividual components can be computed as follows: Similar to relations between the powers and the standard deviations given in Eqs. (6) and (7), σ1 and σ2 are related to Dcc and Dxx, respectively: d to 24) 25) f ˆDcc\|b,Ns = Ns σ 2c χ2 2Ns Ns σ 2c ˆDcc , (30) f ˆDxx\|b,Ns = Ns σ 2x χ2 2Ns Ns σ 2x ˆDxx , (31) deviations given in Eqs. (6) and (7), σ1 and σ2 are related to Dcc and Dxx, respectively: Dcc = var(Rc) + var(Jc) = 2σ 2 c (24) Dxx = var(Rx) + var(Jx) = 2σ 2 x . (25) f ˆDcc\|b,Ns = Ns σ 2c χ2 2Ns Ns σ 2c ˆDcc , (30) f ˆDxx\|b,Ns = Ns σ 2x χ2 2Ns Ns σ 2x ˆDxx , (31) (30) Dcc = var(Rc) + var(Jc) = 2σ 2 c (24) Dxx = var(Rx) + var(Jx) = 2σ 2 x . (25) σc σc f ˆDxx\|b,Ns = Ns σ 2x χ2 2Ns Ns σ 2x ˆDxx , (31) Dcc = var(Rc) + var(Jc) = 2σ 2 c (24) Dxx = var(Rx) + var(Jx) = 2σ 2 x . (25) Dcc = var(Rc) + var(Jc) = 2σ 2 c (24) Dxx = var(Rx) + var(Jx) = 2σ 2 x . (25) (31) (25) f ˆRcx\|b,Ns = (2Ns)a\| ˆRcx\|−b √π22Ns(σcσx)a0(Ns)Kb × \|2Ns ˆRcx\| σcσx ! , (32) f ˆJcx\|b,Ns = (2Ns)a\| ˆJcx\|−b √π22Ns(σcσx)a0(Ns)Kb × \|2Ns ˆJcx\| σcσx ! , (33) f ˆRcx\|b,Ns = (2Ns)a\| ˆRcx\|−b √π22Ns(σcσx)a0(Ns)Kb The measured values ˆDcc, (32) ˆDcx = ˆRcx + i ˆJcx, (26) ˆDcx = ˆRcx + i ˆJcx, (26) and ˆDxx represent elements of the matrix ˆD: and ˆDxx represent elements of the matrix ˆD: × \|2Ns ˆJcx\| σcσx ! , (33) ˆD = Q† ˆBQ. (27) (33) 4.1 Error covariance matrix of b The main result of this subsection – Eq. (46) – was imple- mented as a ready-to-use MATLAB function that is available in the Supplement. In this section we start from the representation of measure- ments on the c–x basis because the measurement errors are independent in this case. Recall that Eq. (27) relates the co- variance matrix ˆD and ˆB. This equation thus can be used to ﬁnd relations between elements of the vector ˆb on the origi- nal h–v basis and elements of the vector ˆd on the c–x basis. In the next subsection we also consider the error covari- ance matrix of the vector ˆc – the conventional representa- tion of polarimetric measurements. Note however that, as is shown in Sect. 5, the approximation of the error covariance matrix of ˆc has issues which may limit its applicability. The covariance matrix ˆB estimated from measurements is related to the matrix ˆD as follows: 4 Error covariance matrices = Mˆd. (45) In the previous section we derived mathematical expres- sions for the likelihood for polarimetric radar observations. A number of scientiﬁc studies, however, require the numer- ical computation of the covariance matrix of the measure- ment errors. For instance, optimal estimation, data assimila- tion, and sensitivity analysis are often performed using error covariance matrices. Unfortunately, an analytical integration of Eqs. (29), (36), and (39) required for the statistical mo- ment calculation is challenging. In this section we therefore follow a different and more viable way to calculate elements of the error covariance matrix. This is done by going back to the representation of the measurements on the convenient c–x basis and applying well-known rules for the calculation of the covariance matrix after a linear transformation. In this case, as shown in Wilks (chap. 10.4.3), the error covariance matrix 6b of ˆb can be calculated from the error covariance matrix 6d of ˆd: 6b = M6dMT, (46) 6b = M6dMT, (46) (46) where where 6d =   4σ 4 c /Ns 0 0 0 0 σ 2 c σ 2 x /Ns 0 0 0 0 σ 2 c σ 2 x /Ns 0 0 0 0 4σ 4 x /Ns  . (47) (47) Recall that the off-diagonal terms of 6d are set to 0 taking into account that the elements of ˆd are not correlated. The derivation of diagonal terms – variances of elements of ˆd – is given in Appendix C. A. Myagkov and D. Ori: Analytic characterization of random errors elements of the vector ˆd: B. It is therefore interesting to derive the likelihood function of those quantities. ˆBhh = q2 11 ˆDcc + \|˙q12\|2 ˆDxx + 2q11 R12 ˆRcx ˆBhh = q2 11 ˆDcc + \|˙q12\|2 ˆDxx + 2q11 R12 ˆRcx +J12 ˆJcx , (41) ˆRhv = q11R12 ˆDxx −ˆDcc + q2 11 −R2 12 + J 2 12 ˆRcx −2R12J12 ˆJcx (42) Likelihood fc(ˆc\|b,Ns) of a vector Bh Likelihood fc(ˆc\|b,Ns) of a vector (41) ˆc = ( ˆBhh, ˆZDR, ˆρHV , ˆ8DP) (37) ˆc = ( ˆBhh, ˆZDR, ˆρHV , ˆ8DP) (37) can be found by multiplying fb(ˆb\|b,Ns) by \|Jcb\|, with (37) can be found by multiplying fb(ˆb\|b,Ns) by \|Jcb\|, with (42) Jcb = −B3 hhZ−3 DRρHV (38) (38) ˆJhv = q11J12 ˆDxx −ˆDcc + q2 11 + R2 12 −J 2 12 ˆJcx −2R12J12 ˆRcx (43) ˆDvv = \|˙q12\|2 ˆDcc + q2 11 ˆDxx −2q11 R12 ˆRcx +J12 ˆJcx . (44) being the Jacobian of the transformation from ˆc to ˆb (see Appendix F): (43) fc(ˆc\|b,Ns) = B3 hhZ−3 DRρHVfb(ˆb\|b,Ns). (39) (39) (44) (44) +J12 ˆJcx . (44) The ﬁnal results of this section – Eqs. (36) and (39) – can be used for the maximum likelihood optimization and Bayesian inference methods. Ready-to-use MATLAB imple- mentations of these equations are provided in the Supple- ment. Or they can be found in matrix form: Or they can be found in matrix form: ˆb=   q2 11 2q11R12 2q11J12 \|˙q12\|2 −q11R12 q2 11 −R2 12 + J 2 12 −2R12J12 q11R12 −q11J12 −2R12J12 q2 11 + R2 12 −J 2 12 q11J12 \|˙q12\|2 −2q11R12 −2q11J12 q2 11  ˆd = Mˆd. ˆd 3.4 Step 4: likelihood for the conventional representation of polarimetric measurements As already mentioned, the polarimetric measurements are commonly described by means of a set of quantities (ZDR, ρHV, 8DP) that are non-linear functions of the elements of The derivation of the formulas for the calculation of the likelihood functions is tedious and provided in full in Ap- pendix B for the interested reader. The likelihoods of the in- Atmos. Meas. Tech., 15, 1333–1354, 2022 https://doi.org/10.5194/amt-15-1333-2022 https://doi.org/10.5194/amt-15-1333-2022 1339 (40) (40) ˆB = Q ˆDQ†. 4.2 Error covariance matrix of the conventional measurement vector c ˆB = Q ˆDQ†. (40) https://doi.org/10.5194/amt-15-1333-2022 ˆB = Q ˆDQ†. As is shown in Sect. 4.1, the error covariance matrix 6b can be used to characterize uncertainties in spectral radar obser- vations. By analogy to what is done in Sect. 3.4, one might After expanding Eq. (40) it can be seen that the elements of the vector ˆb can be found as linear combinations of the Table 1. Operational setting of the used W-band radar. Table 1. Operational setting of the used W-band radar. Parameter Chirp type 1 Chirp type 2 Chirp type 3 Covered distance [km] 0.1–1.2 1.2–4.9 4.9–15 Range resolution [m] 29.8 29.8 55 Number of chirps in a sequence 7168 7168 9216 Chirp repetition frequency [kHz] 9.2 7.5 5 continuous signals. Küchler et al. (2017) explain the opera- tion principle and show that the radar proﬁles the atmosphere using several chirp types. Each chirp type is dedicated to a certain distance range. During measurements chirp types are switched consequently. For each chirp type a number of chirps (chirp sequence hereafter) are processed continuously. Operational settings used during I/Q measurements are listed in Table 1. think about applying again the rules of linear transformation to obtain the error covariance of the vector ˆc. In this section we do that by means of a linearization of the formulas that deﬁne the components of c (Eqs. 10, 11, and 12). It is further demonstrated in Sect. 5 that such representation of measure- ment uncertainties for ˆc is deﬁcient. Recall that the calculation of ˆc includes highly non-linear functions (see Sect. 2.5). Therefore, the error covariance ma- trix 6c of the vector ˆc is estimated using the ﬁrst-order Taylor approximation: Measurements were made during a rain event on 21 June 2021 at 7:44 UTC. I/Q measurements provide a high data rate of about 900 MB min−1. Therefore, about 3 min of I/Q measurements were collected for the analysis. The radar was pointed to 45◦elevation. Since different chirp types have different properties, in the following only I/Q data collected with the ﬁrst chirp type are used. Since the ﬁrst chirp se- quence covers the lowest part of the atmosphere, the ana- lyzed data correspond to rain. As explained in Sect. 2, no noise subtraction is required to describe the statistics of the measurements. We therefore use all available spectral lines, including those containing noise only. A total of 90 % of spectral noise power was from 0.2–1.3×10−3 a.u (arbitrary units). Signal-to-noise ratio (deﬁned here as a ratio of signal power in a spectral line divided by the mean spectral noise power in the same range bin) speciﬁed in linear units was from 0 (no signal) to 106. We would like to emphasize that no ﬁltering based on signal-to-noise ratio was applied. 5.1 Processing All I/Q measurements within a chirp sequence in every po- larimetric channel are split into 224 continuous blocks. Each block contains 32 I/Q pairs. The FFT with the Blackman weighting window is applied to each block to get complex Doppler spectra. Then the 224 blocks are split into 28 sub- blocks with 8 spectra in each sub-block. Within each sub- block elements of the vector ˆb are calculated according to Eqs. (15)–(18) with Ns = 8 for every spectral line. For each ˆb the vector ˆc is obtained. Note that for this, Eqs. (10)–(12) were applied to elements of ˆb instead of b. Using vectors ˆb and ˆc within a sequence, the error covariance matrices ˆ6b and ˆ6c are calculated numerically. The circumﬂex here in- dicates that the error covariance matrices are estimated from measurements. Table 1. Operational setting of the used W-band radar. Tak- ing into account that the ﬁrst chirp type has 37 range bins, in total 2.2×103 chirp sequences (15.9×106 chirps) are avail- able in each polarimetric channel. 6c = S6bST, (48) 6c = S6bST, (48) where S is the sensitivity matrix: where S is the sensitivity matrix: where S is the sensitivity matrix: S =   ∂Bhh ∂Bhh ∂Bhh ∂Rhv ∂Bhh ∂Jhv ∂Bhh ∂Bvv ∂ZDR ∂Bhh ∂ZDR ∂Rhv ∂ZDR ∂Jhv ∂ZDR ∂Bvv ∂ρHV ∂Bhh ∂ρHV ∂Rhv ∂ρHV ∂Jhv ∂ρHV ∂Bvv ∂8DP ∂Bhh ∂8DP ∂Rhv ∂8DP ∂Jhv ∂8DP ∂Bvv   . (49) (49) Note that the utilization of the ﬁrst-order Taylor approxi- mation for variances of polarimetric variables was proposed in the classical book of Bringi and Chandrasekar (2001). Equation (D1) in Appendix D shows the complete matrix S in terms of Bhh, Bvv, ˙Bhv, Rhv, and Jhv. A ready-to-use MATLAB implementation of Eq. (48) is provided in the Supplement. As is shown in the next section, the error covariance matrix 6c does not always reﬂects the true statistical properties of polarimetric observations. There- fore, this approximation is provided only for demonstration purposes, and it is not recommended. A. Myagkov and D. Ori: Analytic characterization of random errors A. Myagkov and D. Ori: Analytic characterization of random errors A. Myagkov and D. Ori: Analytic characterization of random errors 1340 https://doi.org/10.5194/amt-15-1333-2022 https://doi.org/10.5194/amt-15-1333-2022 https://doi.org/10.5194/amt-15-1333-2022 Atmos. Meas. Tech., 15, 1333–1354, 2022 1340 A. Myagkov and D. Ori: Analytic characterization of random errors Table 1. Operational setting of the used W-band radar. Parameter Chirp type 1 Chirp type 2 Chirp type 3 Covered distance [km] 0.1–1.2 1.2–4.9 4.9–15 Range resolution [m] 29.8 29.8 55 Number of chirps in a sequence 7168 7168 9216 Chirp repetition frequency [kHz] 9.2 7.5 5 5.2 Filtering The random error analysis provided in this study is only ap- plicable to volume-distributed scattering and noise. As dis- cussed in Sect. 2, in this case ˆRh is not correlated with ˆJh, and ˆRv is not correlated with ˆJv. However, radar observa- tions in general contain scattering from atmospheric plank- ton, ground clutter, and coherent receiver noise, which do not fulﬁll the assumption. In order to ﬁlter out spectral lines with correlated real and imaginary parts, a simple ﬁltering rule was applied. It is known that for a signal with uncorre- lated in-phase and quadrature components, its mean power and power standard deviation are related to each other (Eq. 5.193 in Bringi and Chandrasekar, 2001). Figure 2 shows distributions of the mean power over the power standard deviation calculated in the horizontal and vertical polariza- tion channels shown by blue and yellow lines, respectively. It can be seen that the mode of the distributions is close to the theoretical value of √Ns = 2.8. The distributions, how- ever, have a considerable tail on the left side. These small values of the ratio are expected for correlated in-phase and quadrature components. Thus, a threshold in the ratio of the mean power over the standard deviation of power can be used to ﬁlter out unwanted spectral lines. In order to specify the threshold, the Monte Carlo approach was used. A total of 15.9×106 random complex values with normal distribution, zero mean, and a standard deviation of 1 were generated. The same processing as for measured I/Q data was applied to the generated complex values. The distribution of the ratio of the mean power over the power standard deviation for the gener- ated data (denoted as expected distribution) is shown in Fig. 2 by the red line. The expected distribution has a much smaller tail on the left side relative to the ones of the measured dis- tributions. The threshold of 2.3 used for ﬁltering is chosen as the 5th percentile of the expected distribution. Vectors ˆb and ˆc are excluded from the analysis if for the corresponding spectral component within a chirp sequence the ratio of the mean power over the power standard deviation is below the threshold in at least one of the polarimetric channels. Around 18 % of the data are excluded. Figure 1. Schematic illustration of the error covariance matrix cal- culation. Figure 2. 5 Consistency checks on radar observations In order to check consistency of Eqs. (36), (39), (46), and (48) with radar measurements, I/Q data collected with a W-band cloud radar with the hybrid polarimetric mode were used (Myagkov and Unal, 2021). The radar is a part of a dual-frequency system owned and operated by the Technical University of Delft in Cabauw, the Nether- lands. Technical speciﬁcations of the radar can be found in Myagkov et al. (2020). The radar uses frequency-modulated https://doi.org/10.5194/amt-15-1333-2022 Atmos. Meas. Tech., 15, 1333–1354, 2022 A. Myagkov and D. Ori: Analytic characterization of random errors 1341 A. Myagkov and D. Ori: Analytic characterization of rand Figure 1. Schematic illustration of the error covariance matrix cal- culation. 5.2 Filtering Distributions of the ratio of mean power over the power standard deviation for the horizontal (blue line) and vertical (yel- low line) channels. The expected distribution is shown with the red line. The vertical black line indicates the threshold corresponding to the 5th percentile of the distribution for the randomly generated complex numbers. Figure 2. Distributions of the ratio of mean power over the power standard deviation for the horizontal (blue line) and vertical (yel- low line) channels. The expected distribution is shown with the red line. The vertical black line indicates the threshold corresponding to the 5th percentile of the distribution for the randomly generated complex numbers. 5.3 Evaluation of fb(ˆb\|b,Ns) and fb(ˆc\|b,Ns) 5.3 Evaluation of fb(ˆb\|b,Ns) and fb(ˆc\|b,Ns) Recall that b is estimated from measurements by averaging all available sub-blocks within a chirp sequence; b, how- ever, can also be estimated by maximization of the likeli- hood functions given in Eqs. (36) and (39). In this case, an optimization algorithm needs to be employed to ﬁnd a set of elements of b corresponding to the global maximum in ei- ther Eq. (36) or Eq. (39). This study uses a derivative-free optimization method available by default in MATLAB (La- garias et al., 1998). Since the optimization method minimizes a function, the likelihood functions were not used directly. Instead, the following cost functions were used for the mini- mization: The calculation of the likelihood functions using Eqs. (36) and (39) requires b. The approximation of covariance matri- ces using Eqs. (46) and (48) requires the matrix B. In order to estimate b and B, elements of the vector ˆb are averaged over 28 sub-blocks available within a single chirp sequence. These averaged values are assumed to be elements of the vector b from which the matrix B is obtained. Using B and Ns = 8, 6b and 6c are calculated for each chirp sequence as shown in Fig. 1. https://doi.org/10.5194/amt-15-1333-2022 https://doi.org/10.5194/amt-15-1333-2022 Atmos. Meas. Tech., 15, 1333–1354, 2022 A. Myagkov and D. Ori: Analytic characterization of random errors A. Myagkov and D. Ori: Analytic characterization of random errors 1342 Cb = − 28 X l=1 log10 fb ˆb\|b,Ns , (50) Cc = − 28 X l=1 log10 fc ˆc\|b,Ns . (51) Here the index l runs over 28 sub-blocks within a chirp VARbhh = B2 hh Ns , (52) VARzdr = 2Z2 DR(1 −ρ2 HV) Ns , (53) VARρ = (1 −ρ2 HV)2 2Nsρ2 HV ,and (54) VAR8 = (1 −ρ2 HV) 2Nsρ2 HV , (55) Cb = − 28 X l=1 log10 fb ˆb\|b,Ns , (50) Cc = − 28 X l=1 log10 fc ˆc\|b,Ns . (51) Here the index l runs over 28 sub-blocks within a chirp E ti (50) d (51) t k i t t th t th VARbhh = B2 hh Ns , (52) VARzdr = 2Z2 DR(1 −ρ2 HV) Ns , (53) VARρ = (1 −ρ2 HV)2 2Nsρ2 HV ,and (54) VAR8 = (1 −ρ2 HV) 2Nsρ2 HV , (55) VARbhh = B2 hh Ns , (52) VARzdr = 2Z2 DR(1 −ρ2 HV) Ns , (53) VARρ = (1 −ρ2 HV)2 2Nsρ2 HV ,and (54) VAR8 = (1 −ρ2 HV) 2Nsρ2 HV , (55) VARbhh = B2 hh Ns , (52) VARzdr = 2Z2 DR(1 −ρ2 HV) Ns , (53) VARρ = (1 −ρ2 HV)2 2Nsρ2 HV ,and (54) VAR8 = (1 −ρ2 HV) 2Nsρ2 HV , (55) Cb = − 28 X l=1 log10 fb ˆb\|b,Ns , (50) Cc = − 28 X l=1 log10 fc ˆc\|b,Ns . (51) Cb = − 28 X l=1 log10 fb ˆb\|b,Ns , (50) Cc = − 28 X l=1 log10 fc ˆc\|b,Ns . (51) (52) (53) (51) (55) Here the index l runs over 28 sub-blocks within a chirp sequence. Equations (50) and (51) take into account that the consecutive ˆb vectors are not correlated. In this case the to- tal likelihood of 28 vector ˆb’s is a product of likelihood of each individual ˆb. In order to avoid an overﬂow of double numbers, the logarithm was used. In this case the logarithm of the product is replaced by the sum of logarithms. respectively. Figure 4 shows that VARbhh, VARzdr, and VAR8 match exactly 6c(1,1), 6c(2,2), and 6c(4,4), respectively. VARρ, however, agrees with 6c(3,3) only at values of ρHV > 0.95. Below this value VARρ overestimates the variance of ˆρHV . At values of ρHV close to 0, VARρ has unrealistically high values, which result from ρHV in the denominator of Eq. (54). Figure 4d also shows unrealistic values with both approx- imations of the ˆ8DP variance. Taking into account that ˆ8DP can take values within the range of 0 to 2π rad, the vari- ance of ˆ8DP exceeding 103 rad2 is deﬁnitely erroneous. The high variance of ˆ8DP corresponds to values of ρHV < 0.3. This effect results from the ﬁrst-order Taylor approximation of Eq. (12), which is a highly non-linear function. ˆ A comparison of the error covariance matrices ˆ6c with the calculated one 6c is shown in Fig. 5. Figure 5f, k, and p indicate considerable differences caused by the ﬁrst-order Taylor approximation in variances of ˆZDR, ˆρHV , and ˆ8DP, respectively. The results also reveal that the ﬁrst-order Taylor approximation cannot adequately represent most of the non- diagonal components of the error covariance matrix. 5.5 Evaluation of 6b Figure 6 shows a comparison of elements of error covari- ance matrices ˆ6b estimated from the radar measurements with those calculated using Eq. (46). Estimated and calcu- lated elements are in a good agreement. Linear regressions shown in the panels by red lines have slopes close to 1. Pear- son correlations between estimated and calculated elements exceed 0.96. These results indicate an agreement of the the- oretical calculation with measurements and thus conﬁrm the correctness of Eq. (46). As expected, Figs. 6a and 5a show equivalent results. This is because the co-polar signal Bhh is effectively the same in both measurement representations and highlights the relevance of the present study only for dual-polarimetric quantities. https://doi.org/10.5194/amt-15-1333-2022 The log- arithm is a monotonically increasing function, and therefore it does not change the position of the maximum of the like- lihood function. Finally, the minus sign was introduced to have a smaller value of a cost function corresponding to a higher value of the likelihood. For the evaluation, 1000 chirp sequences were chosen randomly for the maximum likeli- hood estimation using fb(ˆb\|b,Ns). In each chirp sequence a single spectral line was randomly chosen for the analy- sis. Thus, there are 28 vector ˆb’s available in each of the 1000 chirp sequences. For each sequence, the optimization algorithm requires an initial guess of b. In order to avoid lo- cal minima, ﬁve different initial guesses were used, which are a coefﬁcient P multiplied by the ﬁrst ˆb in the analyzed chirp sequence. The values of P were 0.5, 0.75, 1, 1.25, and 1.5. The solution giving the lowest cost function out of the ﬁve outcomes was chosen as the result. Similarly the maxi- mum likelihood estimation using fc(ˆc\|b,Ns) was done using independently chosen 1000 chirp sequences. Figure 3 shows a comparison of elements of b estimated by averaging over 28 sub-blocks and those estimated by the maximum likeli- hood approach. All panels show a good agreement indicated by the close-to-unity slope of the linear regression. Both fb(ˆb\|b,Ns) (results in the ﬁrst row of Fig. 3) and fc(ˆc\|b,Ns) (results in the second row of Fig. 3) show the same level of agreement and, therefore, can be used with no difference. respectively. https://doi.org/10.5194/amt-15-1333-2022 A. Myagkov and D. Ori: Analytic characterization of random errors Myagkov and D. Ori: Analytic characterization of random errors 1343 A. Myagkov and D. Ori: Analytic characterization of random errors 1343 Figure 3. Comparison of elements of b estimated by averaging over 28 sub-blocks (x axis) with those estimated by the maximum likelihood approach (y axis); fb(ˆb\|b,Ns) was used for (a)–(d). fc(ˆc\|b,Ns) was used for (e)–(h). Each panel contains 1000 points described in the text. Linear regressions are shown by solid red lines. Each panel has a text box with the slope of the corresponding linear regression. Uncertainties in the slopes were estimated using bootstrapping. Note that units are not critical for the evaluation of the correctness of the derived likelihood functions. Therefore, arbitrary units (a.u.) are used. Figure 3. Comparison of elements of b estimated by averaging over 28 sub-blocks (x axis) with those estimated by the maximum likelihood approach (y axis); fb(ˆb\|b,Ns) was used for (a)–(d). fc(ˆc\|b,Ns) was used for (e)–(h). Each panel contains 1000 points described in the text. Linear regressions are shown by solid red lines. Each panel has a text box with the slope of the corresponding linear regression. Uncertainties in the slopes were estimated using bootstrapping. Note that units are not critical for the evaluation of the correctness of the derived likelihood functions. Therefore, arbitrary units (a.u.) are used. Figure 4. Comparison of variances of (a) ˆBhh, (b) ˆZDR, (c) ˆρHV , and (d) ˆ 8DP. Approximations developed in this study are on the x axis. Approximations from Bringi and Chandrasekar (2001) are on the y axis; ρHV is color-coded in (c) and (d) to illustrate the values of ρHV at which approximations lead to erroneous values (see details in text). Note that units are not critical for the evaluation of the derived equations. Therefore, arbitrary units (a.u.) are used in (a). Figure 4. Comparison of variances of (a) ˆBhh, (b) ˆZDR, (c) ˆρHV , and (d) ˆ 8DP. Approximations developed in this study are on the x axis. Approximations from Bringi and Chandrasekar (2001) are on the y axis; ρHV is color-coded in (c) and (d) to illustrate the values of ρHV at which approximations lead to erroneous values (see details in text). Note that units are not critical for the evaluation of the derived equations. Therefore, arbitrary units (a.u.) are used in (a). 5.4 Evaluation of 6c Diagonal elements of 6c – variances of ˆBhh, ˆZDR, ˆρHV , and ˆ8DP – were checked against those calcu- lated using Eqs. (6.139a), (6.141), (6.144), and (6.143) in Bringi and Chandrasekar (2001), respectively. Taking into account that samples for a spectral line are not correlated, approximations for variances of ˆBhh, ˆZDR, ˆρHV , and ˆ8DP based on the equations in Bringi and Chandrasekar (2001) are It is thus concluded that any application of spectral po- larimetric measurements which requires the estimate of the error covariance matrix (e.g., variational retrievals, data as- similation, and sensitivity analysis) should be performed in the space of observations ˆb rather than ˆc. Atmos. Meas. Tech., 15, 1333–1354, 2022 https://doi.org/10.5194/amt-15-1333-2022 A. Myagkov and D. Ori: Analytic characterization of random errors https://doi.org/10.5194/amt-15-1333-2022 6 Summary and outlook (1) a lack of joint PDFs for averaged spectral polarimetric measurements, (2) neglection of non-diagonal components of the error covariance matrix, and (3) inaccuracy of the ﬁrst- order approximation in variances of polarimetric variables. This study thus aims to provide solutions for these three problems. Spectral and polarimetric cloud radar observations have a great potential in the cloud science (Kollias et al., 2020). Decades of such measurements have been already collected by, for example, the ARM (Atmospheric Radiation Mea- surement) and CLOUDNET communities. An advanced ap- plication of these vast datasets requires an accurate char- acterization of measurement uncertainties. Systematic er- rors in moment radar data and polarimetric variables have been discussed in many studies. Random measurement er- rors, in contrast, are rarely considered in the literature. There are three main problems in existing random-error- characterization methods in meteorological studies, namely Equations provided in Sect. 3 give an exact mathemati- cal solution for the joint PDFs of spectral polarimetric obser- vations. The PDFs are given for two equivalent representa- tions of the measurements: (1) b = (Bhh,Rhv,Jhv,Bvv)T and (2) c = (Bhh,ZDR,ρHV,8DP)T. The obtained equations take into account non-coherent averaging of spectra, which is ap- plied by a majority of cloud radars to improve the sensitiv- https://doi.org/10.5194/amt-15-1333-2022 A. Myagkov and D. Ori: Analytic characterization of random errors A. Myagkov and D. Ori: Analytic characterization of random errors 1344 Figure 5. Comparison of ˆ6c estimated from the radar measurements with 6c obtained from Eq. (48). Elements of 6c are given on the x axes. Elements of ˆ6c are given on the y axes. The ﬁrst and the second numbers in brackets indicate the row and the column of the corresponding matrix, respectively. Linear regressions are shown by red lines. Slopes of the linear regressions and Pearson correlations are given in boxes in each panel. Uncertainties in the slope and the correlation are represented by ±1 standard deviation of the corresponding parameter. The standard deviations are obtained using bootstrapping. Panels without linear regressions show elements for which Eq. (48) gives only near- zero values. Note that units are not critical for the evaluation of the derived equations. Therefore, arbitrary units (a.u.) are used. Also note that only values on the x and y axes in an individual panel should be compared. Values in different panels should not be compared. it M i lik lih d ti t f b b d E (36) S ti 4 i f d th i t i Figure 5. Comparison of ˆ6c estimated from the radar measurements with 6c obtained from Eq. (48). Elements of 6c are given on the x axes. Elements of ˆ6c are given on the y axes. The ﬁrst and the second numbers in brackets indicate the row and the column of the corresponding matrix, respectively. Linear regressions are shown by red lines. Slopes of the linear regressions and Pearson correlations are given in boxes in each panel. Uncertainties in the slope and the correlation are represented by ±1 standard deviation of the corresponding parameter. The standard deviations are obtained using bootstrapping. Panels without linear regressions show elements for which Eq. (48) gives only near- zero values. Note that units are not critical for the evaluation of the derived equations. Therefore, arbitrary units (a.u.) are used. Also note that only values on the x and y axes in an individual panel should be compared. Values in different panels should not be compared. Section 4 is focused on the error covariance matrix re- quired for a number of applications such as data assimila- tion, sensitivity analysis, and variational retrievals. https://doi.org/10.5194/amt-15-1333-2022 Atmos. Meas. Tech., 15, 1333–1354, 2022 A. Myagkov and D. Ori: Analytic characterization of random errors A. Myagkov and D. Ori: Analytic characterization of random errors 1345 Figure 6. Comparison of ˆ6b estimated from the radar measurements with 6b obtained from Eq. (46). Elements of 6b are given on the x axes. Elements of ˆ6b are given on the y axes. The ﬁrst and the second numbers in brackets indicate the row and the column of the corresponding matrix, respectively. Linear regressions are shown by red lines. Slopes of the linear regressions and Pearson correlations are given in boxes in each panel. Uncertainties in the slope and the correlation are represented by ±1 standard deviation of the corresponding parameter. The standard deviations are obtained using bootstrapping. Note that units are not critical for the evaluation of the derived equations. Therefore, Figure 6. Comparison of ˆ6b estimated from the radar measurements with 6b obtained from Eq. (46). Elements of 6b are given on the x axes. Elements of ˆ6b are given on the y axes. The ﬁrst and the second numbers in brackets indicate the row and the column of the corresponding matrix, respectively. Linear regressions are shown by red lines. Slopes of the linear regressions and Pearson correlations are given in boxes in each panel. Uncertainties in the slope and the correlation are represented by ±1 standard deviation of the corresponding parameter. The standard deviations are obtained using bootstrapping. Note that units are not critical for the evaluation of the derived equations. Therefore, arbitrary units (a.u.) are used. trated that elements of 6c have considerable differences from those estimated from the measurements. First, we found dif- ferences in variances of ZDR, ρHV, and 8DP of up to a fac- tor of 10, 5, and 100, respectively. Second, the calculated variance of 8DP shows unrealistically high values by far ex- ceeding the range of possible values. Third, most of the off- highly non-linear functions, 6c was derived using the ﬁrst- order Taylor approximation. The same approach was used by Bringi and Chandrasekar (2001) to get equations for vari- ances of polarimetric observations. The error covariance matrices were evaluated using I/Q observations from a polarimetric W-band radar. It is illus- https://doi.org/10.5194/amt-15-1333-2022 A. Myagkov and D. Ori: Analytic characterization of random errors The er- ror covariance matrices 6b and 6c for b and c, respectively, are obtained using the characteristic functions of the PDFs described in Sect. 3. Since the calculation of the c includes ity. Maximum likelihood estimators of b based on Eqs. (36) and (39) were compared with the estimator based on longer averaging. The comparison was based on dual-polarimetric cloud radar observations. The comparison showed a good agreement. Both PDFs can be equivalently used for methods based on the maximum likelihood and Bayesian inference. Atmos. Meas. Tech., 15, 1333–1354, 2022 https://doi.org/10.5194/amt-15-1333-2022 Myagkov and D. Ori: Analytic characterization of random errors B2 Likelihood functions for Dcc and Dxx It is known that the PDF of zs being a sum of squares of independent standard normal samples (i.e., distributed nor- mally with a mean of 0 and standard deviation of 1) is the chi-squared distribution χ2 k (zs), where the degree of freedom k shows how many samples have been summed. Taking into account that In order to demonstrate a practical application of the developed characterization of the measurements er- rors, a few retrieval techniques are currently being de- veloped. The ﬁrst one is an improvement of the ice- shape retrieval described in Myagkov et al. (2016a). An- other one is an adoption of the drop size distribu- tion retrieval from Tridon and Battaglia (2015) for dual- polarimetric cloud radar observations. ˆDcc = N−1 s σ 2 c ( σ −2 c Ns X l=1 Re ˙Sc 2 l + σ −2 c Ns X l=1 Im ˙Sc 2 l ) , (B3) (B3) where the ﬁrst and the second summed terms in the curly brackets are sums of squares of independent standard nor- mal samples, the likelihood function f ˆDcc\|σc,Ns can be found by changing the variable zs to Nsσ −2 c ˆDc: where the ﬁrst and the second summed terms in the curly brackets are sums of squares of independent standard nor- mal samples, the likelihood function f ˆDcc\|σc,Ns can be found by changing the variable zs to Nsσ −2 c ˆDc: B1 Change in variables in a PDF Consider a vector a with n random variables a1...n. Assume the joint PDF fa(a) of the variables is known. The joint PDF y y = G(a) (B1) (B1) y = G(a) can be found by changing the variables in fa(a): Thus, based on the results found within this study, it is rec- ommended to use the vector b to represent polarimetric cloud radar observations for applications requiring the error covari- ance matrix. This representation has a better characterization of random errors in comparison with widely used representa- tion c. When the signal-to-noise ratio is high (> 35 dB), how- ever, the variances are quite low, and the Taylor approxima- tion may give reasonable results. We would like to emphasize that there is no additional processing required to get the vec- tor b. Elements of the vector b are an intermediate process- ing step on the way from I/Q data to conventional spectral polarimetric variables and thus have been already calculated by Doppler cloud radars with the hybrid mode. fy(y) = \|J\|fa h G−1(y) i , (B2) (B2) where G−1 is the reverse transformation from y to a, and J is the determinant of the Jacobian of the transformation a = G−1(y). https://doi.org/10.5194/amt-15-1333-2022 https://doi.org/10.5194/amt-15-1333-2022 Atmos. Meas. Tech., 15, 1333–1354, 2022 1346 A. Myagkov and D. Ori: Analytic characterization of random errors Appendix B: Derivation of likelihood functions diagonal terms of 6c are not correlated with corresponding values estimated from observations. We relate the differences to the ﬁrst-order Taylor approximation. The Taylor approxi- mation assumes linear relations between elements of the vec- tor b and the elements of the vector c, while the relations in- clude highly non-linear functions. In contrast, 6b agrees well with the observations. The correlation between calculated el- ements of 6b with those estimated from the observations ex- ceeds 0.965. A. Myagkov and D. Ori: Analytic characterization of random errors 1. Bhh is a sum of mean powers of signal Psh and noise Pnh. 1. Bhh is a sum of mean powers of signal Psh and noise Pnh. where n is the number of averaged multiplications, 0 is the gamma function, and Kµ is the Bessel function of the second kind of order µ. where n is the number of averaged multiplications, 0 is the gamma function, and Kµ is the Bessel function of the second kind of order µ. ˆ 2. Bvv is a sum of mean powers of signal Psv and noise Pnv. 2. Bvv is a sum of mean powers of signal Psv and noise Pnv. µ ˆRcx is calculated as follows: ˆRcx is calculated as follows: ˆRcx = 2σcσx ( 1 2Nsσcσx " Ns X l=1 Re ˙Sc Re ˙Sx + Ns X l=1 Im ˙Sc Im ˙Sx #) , (B7) 3. Pnh = Pnv = 1. 3. Pnh = Pnv = 1. 4. Psh and Psv were randomly and independently gener- ated using the uniform distribution from 1 to 5. 4. Psh and Psv were randomly and independently gener- ated using the uniform distribution from 1 to 5. (B7) 5. ˙Bhv was calculated as ρHVei8DP√PshPsv. where the term in the curly brackets is an average over 2Ns multiplications of independent standard normal samples. In this case, the likelihood function f ˆRcx\|σc,σx,Ns can be found by changing zm by (2σcσx)−1 ˆRcx: where the term in the curly brackets is an average over 2Ns multiplications of independent standard normal samples. In this case, the likelihood function f ˆRcx\|σc,σx,Ns can be found by changing zm by (2σcσx)−1 ˆRcx: 6. ρHV was chosen randomly using the uniform distribu- tion from 0 to 1. 6. ρHV was chosen randomly using the uniform distribu- tion from 0 to 1. 7. 8DP was chosen randomly using the uniform distribu- tion from 0 to 2π. f ˆRcx\|σc,σx,Ns = (2Ns)a\| ˆRcx\|−b √π22Ns(σcσx)a0(Ns)Kb f ˆRcx\|σc,σx,Ns = (2Ns)a\| ˆRcx\|−b √π22Ns(σcσx)a0(Ns)Kb × Ns\| ˆRcx\| σcσx ! , (B8) 8. Ns was chosen as a random integer number in the range of 2 to 80. × Ns\| ˆRcx\| σcσx ! , (B8) From the covariance matrix B the true covariance matrix 6m was obtained. A total of 105 × Ns vector m’s were gen- erated according to the PDF given in Eq. Appendix A: Diagonalization matrix Q The operator Q, which is used to diagonalize the covariance matrix B in Eq. (20), is calculated as follows (Kanareykin et al., 1968, chap. 2.5): f ˆDcc\|σc,Ns = Ns σ 2c χ2 2Ns Ns σ 2c ˆDcc . (B4) (B4) Q = q11 ˙q12 −˙q∗ 12 q11 , (A1) (A1) The factor of 2 in the degree of freedom is because there are 2Ns summed components in the curly brackets in Eq. (B3). The equation for ˆDxx is derived in a similar man- ner as for ˆDcc, resulting in where where q11 = 1 + ˙d 2−0.5 , (A2) ˙q12 = −˙d∗q11, and (A3) d = ˙B∗ hv 0.5 h TrB + p TrB2 −4det(B) i −Bvv . (A4) q11 = 1 + ˙d 2−0.5 , (A2) ˙q12 = −˙d∗q11, and (A3) d = ˙B∗ hv 0.5 h TrB + p TrB2 −4det(B) i −Bvv . (A4) f ˆDxx\|σx,Ns = Ns σ 2x χ2 2Ns Ns σ 2x ˆDxx . (B5) (B5) (A3) B3 Likelihood functions for Rcx and Jcx Nadarajah and Pogány (2016) provide a solution for the PDF of an averaged multiplication zm of two standard normal vari- ables. For two uncorrelated variables the PDF is deﬁned as follows: In Eq. (A4) Tr is the matrix trace. fz(zm) =n(n+1)/22(1−n)/2\|zm\|(n−1)/2 √π0(n/2) ×K(1−n)/2 (n\|zm\|), (B6) √ ×K(1−n)/2 (n\|zm\|), (B6) (B6) Atmos. Meas. Tech., 15, 1333–1354, 2022 Atmos. Meas. Tech., 15, 1333–1354, 2022 https://doi.org/10.5194/amt-15-1333-2022 https://doi.org/10.5194/amt-15-1333-2022 1347 A. Myagkov and D. Ori: Analytic characterization of random errors (B11) (B11) B4 Monte Carlo evaluation of Eqs. (B4), (B5), (B8), and (B10) B4 Monte Carlo evaluation of Eqs. (B4), (B5), (B8), and (B10) For the equation evaluation a simulated dataset was gener- ated. In total 1000 sets of distributions were simulated using the Monte Carlo approach. A single set included distributions of ˆBcc, ˆBxx, ˆRcx, and ˆJcx. For a single set 105 vector ˆb’s were generated. A single vector ˆb resulted from Ns randomly gen- erated vector m’s. For a single set of distributions a single covariance matrix B was taken. The elements of the covari- ance matrix B and Ns were randomly generated according to the following rules (values have linear arbitrary units): A. Myagkov and D. Ori: Analytic characterization of random errors (2). Then, 105 ele- ments of the ˆb were calculated according to Eqs. (15)–(18). Elements of the vector ˆd were derived from the vector ˆb’s using Eq. (27). (B8) where a = (2Ns + 1)/2, b = (1 −2Ns)/2, 0 is the gamma function, and Kµ is the Bessel function of the second kind of order µ. When ˆRcx →0, the modiﬁed Bessel func- tion Kb Ns\| ˆRcx\|(σcσx)−1 →∞. Therefore, for ˆRcx close to 0, the following approximation based on Eqs. (9.6.6) and (9.6.8) from Abramowitz and Stegun (1972) should be used: Using the 105 vector ˆd’s individual histograms for each of the variables ˆBcc, ˆBxx, ˆRcx, and ˆJcx are derived. A his- togram has 10 bins covering the range from the minimum to maximum values of the corresponding variable. Widths of bins were adjusted to have 10 000 samples in each bin. For the same bins the expected number of samples is calculated using the corresponding PDF. Since integration of Eqs. (B4), (B5), (B8), and (B10) is challenging, the integration is done numerically. Then the Pearson’s chi-squared test is applied. The same procedure is repeated for all 1000 sets of distri- butions. Thus, for each PDF (Eqs. B4, B5, B8, and B10) 1000 test-statistic values were obtained. f ˆRcx\|σc,σx,Ns ≈ Ns0(−b) 2√πσcσx0(Ns). (B9) (B9) Formulas for ˆJcx are deﬁned in a similar manner: Formulas for ˆJcx are deﬁned in a similar manner: f ˆJcx\|σc,σx,Ns = (2Ns)a\| ˆJcx\|−b √π22Ns(σcσx)a0(Ns)Kb × Ns\| ˆJcx\| σcσx ! . (B10) (B10) (B10) The Pearson’s chi-squared test implies a comparison of the test-statistic values with critical values for a given level of signiﬁcance. A test-statistic value exceeding the critical value would indicate that there is a chance (equal to the signiﬁ- cance level) that the data signiﬁcantly differ from the PDF. There is, however, a small chance that the conclusion that the data differ from the PDF is erroneous. Table B1 shows the percentage of the test-statistic values exceeding critical values. It can be seen that the number of test-statistic values exceeding corresponding critical values is very close to the theoretical values, i.e., 5, 2.5, and 1 % at 0.95, 0.975, and 0.99 signiﬁcance levels, respectively. This conﬁrms the va- lidity of the obtained PDFs. The approximation for ˆJcx is close to 0: The approximation for ˆJcx is close to 0: f ˆJcx\|σc,σx,Ns ≈ Ns0(−b) 2√πσcσx0(Ns). Atmos. Meas. Tech., 15, 1333–1354, 2022 A. Myagkov and D. Ori: Analytic characterization of random errors 1348 Table B1. Percentage of test-statistic values exceeding critical values for different signiﬁcance levels. Percentages are given in percent. The names of the four columns on the right side of the table indicate the distribution for which a percentage is given. Signiﬁcance level Critical value f ˆDcc\|σc,Ns f ˆRcx\|σc,Ns f ˆJcx\|σc,Ns f ˆDxx\|σc,Ns 0.95 16.919 6.9 5.2 5.8 5.9 0.975 19.023 3.8 3.4 2.7 2.9 0.99 21.666 1.0 1.3 1.0 1.2 acteristic functions. A γ th raw statistical moment Mγ of a random variable with a characteristic function φ(t) can be found as follows: As expected for a multiplication of two uncorrelated vari- ables, the mean values of ˆRcx and ˆJcx are as follows: ˆRcx = ˆJcx = 1 i dφcx(t) dt t=0 = 0. (C10) (C10) Mγ = i−γ dγ φ(t) dtγ t=0. (C1) (C1) The variance of ˆRcx and ˆJcx can be found as follows: The variance of ˆRcx and ˆJcx can be found as follows: The calculation of derivatives of the characteristic func- tions is in general easier to obtain than integration of the cor- responding PDFs. var ˆRcx = var ˆJcx = −d2φcx(t) dt2 t=0 = σ 2 c σ 2 x 2Ns . (C11) (C11) The characteristic function of the chi-squared distribution χ2 k (zs) is φs (t) = (1 −2it)−k/2. (C2) (C2) Therefore, the characteristic function for ˆDcc for a given σc and Ns can be written in the following way: Therefore, the characteristic function for ˆDcc for a given σc and Ns can be written in the following way: φcc(t) = 1 −2iσ 2 c t Ns −Ns . (C3) (C3) The mean value and variance of ˆDcc are calculated as fol- lows: ˆDcc = 1 i dφcc(t) dt t=0 = 2σ 2 c , (C4) var( ˆDcc) = −d2φcc(t) dt2 t=0 −ˆDcc 2 = 4σ 4 c Ns . (C5) (C4) (C5) Similarly, ˆDxx = 1 i dφxx(t) dt t=0 = 2σ 2 x , and (C6) var( ˆDxx) = 4σ 4 x Ns . (C7) (C6) (C7) Based on Nadarajah and Pogány (2016) the characteristic function corresponding to fz(zm) is φz(t) = 1 + t2 n2 −n/2 . https://doi.org/10.5194/amt-15-1333-2022 Appendix C: Variances of elements of the vector ˆd To derive solutions for the mean and variances of elements of ˆd, the distribution of the elements is represented by char- https://doi.org/10.5194/amt-15-1333-2022 Atmos. Meas. Tech., 15, 1333–1354, 2022 https://doi.org/10.5194/amt-15-1333-2022 A. Myagkov and D. Ori: Analytic characterization of random errors 1349 Appendix D: Sensitivity S S=   ∂Bhh ∂Bhh ∂Bhh ∂Rhv ∂Bhh ∂Jhv ∂Bhh ∂Bvv ∂ZDR ∂Bhh ∂ZDR ∂Rhv ∂ZDR ∂Jhv ∂ZDR ∂Bvv ∂ρHV ∂Bhh ∂ρHV ∂Rhv ∂ρHV ∂Jhv ∂ρHV ∂Bvv ∂8DP ∂Bhh ∂8DP ∂Rhv ∂8DP ∂Jhv ∂8DP ∂Bvv   =   1 0 0 0 B−1 vv 0 0 −BhhB−2 vv −0.5\| ˙Bhv\|B−0.5 vv B−1.5 hh Rhv\| ˙Bhv\|−1(BhhBvv)−0.5 Jhv\| ˙Bhv\|−1(BhhBvv)−0.5 −0.5\| ˙Bhv\|B−0.5 hh B−1.5 vv 0 −Jhv\| ˙Bhv\|−2 Rhv\| ˙Bhv\|−2 0   (D1) Appendix D: Sensitivity S S=   ∂Bhh ∂Bhh ∂Bhh ∂Rhv ∂Bhh ∂Jhv ∂Bhh ∂Bvv ∂ZDR ∂Bhh ∂ZDR ∂Rhv ∂ZDR ∂Jhv ∂ZDR ∂Bvv ∂ρHV ∂Bhh ∂ρHV ∂Rhv ∂ρHV ∂Jhv ∂ρHV ∂Bvv ∂8DP ∂Bhh ∂8DP ∂Rhv ∂8DP ∂Jhv ∂8DP ∂Bvv   =   1 0 0 0 B−1 vv 0 0 −BhhB−2 vv −0.5\| ˙Bhv\|B−0.5 vv B−1.5 hh Rhv\| ˙Bhv\|−1(BhhBvv)−0.5 Jhv\| ˙Bhv\|−1(BhhBvv)−0.5 −0.5\| ˙Bhv\|B−0.5 hh B−1.5 vv 0 −Jhv\| ˙Bhv\|−2 Rhv\| ˙Bhv\|−2 0   (D1) (D1) Appendix E: Jacobian Jbd of the transformation from ˆb to ˆd Appendix E: Jacobian Jbd of the transformation from ˆb to ˆd Appendix E: Jacobian Jbd of the transformation from ˆb to ˆd Using Eqs. (21)–(22) Jbd can be written as follows: Jbd= ∂Dcc ∂Bhh ∂Dcc ∂Rhv ∂Dcc ∂Jhv ∂Dcc ∂Bvv ∂Rcx ∂Bhh ∂Rcx ∂Rhv ∂Rcx ∂Jhv ∂Rcx ∂Bvv ∂Jcx ∂Bhh ∂Jcx ∂Rhv ∂Jcx ∂Jhv ∂Jcx ∂Bvv ∂Dxx ∂Bhh ∂Dxx ∂Rhv ∂Dxx ∂Jhv ∂Dxx ∂Bvv = q2 11 \|˙q12\|2 −2q11R12 −2q11J12 \|˙q12\|2 q2 11 2q11R12 2q11J12 q11R12 −q11R12 q2 11 −R2 12 + J 2 12 −2R12J12 q11J12 −q11J12 −2R12J12 q2 11 + R2 12 −J 2 12 = (q2 11 + \|q12\|2)4 (E1) (E1) Taking into account Eqs. (A2) and (A3), Jbd = 1. Taking into account Eqs. (A2) and (A3), Jbd = 1. Appendix F: Jacobian Jcb of the transformation from ˆc to ˆb A. Myagkov and D. Ori: Analytic characterization of random errors (C8) (C8) Therefore, the characteristic function for ˆRcx and ˆJcx for given σc, σx, and Ns is as follows: φcx(t) = 1 + σ 2 c σ 2 x t2 N2s −Ns/2 . (C9) (C9) Atmos. Meas. Tech., 15, 1333–1354, 2022 https://doi.org/10.5194/amt-15-1333-2022 Appendix G: Table of symbols Appendix G: Table of symbols Table G1. Main symbols used throughout the study. The overdot indicates a complex number. Indices h and v indicate the polarization of the receiver channel. The circumﬂex indicates a measured quantity. Appendix F: Jacobian Jcb of the transformation from ˆc to ˆb Using Eqs. (10)–(12) Jcb can be written as follows: Using Eqs. (10)–(12) Jcb can be written as follows: Using Eqs. (10)–(12) Jcb can be written as follows: Jcb= ∂Bhh ∂Bhh ∂Bhh ∂ZDR ∂Bhh ∂ρHV ∂Bhh ∂8DP ∂Rhv ∂Bhh ∂Rhv ∂ZDR ∂Rhv ∂ρHV ∂Rhv ∂8DP ∂Jhv ∂Bhh ∂Jhv ∂ZDR ∂Jhv ∂ρHV ∂Jhv ∂8DP ∂Bvv ∂Bhh ∂Bvv ∂Rhv ∂Bvv ∂Jhv ∂Bvv ∂Bvv = 1 0 0 0 ρHV cos(8DP)Z−0.5 DR −0.5BhhρHV cos(8DP)Z−1.5 DR Bhh cos(8DP)Z−0.5 DR −BhhρHV sin(8DP)Z−0.5 DR ρHV sin(8DP)Z−0.5 DR −0.5BhhρHV sin(8DP)Z−1.5 DR Bhh sin(8DP)Z−0.5 DR −BhhρHV cos(8DP)Z−0.5 DR Z−1 DR −BhhZ−2 DR 0 0 = −B3 hhZ−3 DRρHV (F1) (F1) https://doi.org/10.5194/amt-15-1333-2022 Atmos. Meas. Tech., 15, 1333–1354, 2022 https://doi.org/10.5194/amt-15-1333-2022 Atmos. Meas. Tech., 15, 1333–1354, 2022 1350 A. Myagkov and D. Ori: Analytic characterization of random errors https://doi.org/10.5194/amt-15-1333-2022 Atmos. Meas. Tech., 15, 1333–1354, 2022 Appendix G: Table of symbols Symbol Description 0 Gamma function ρHV and ˆρHV Correlation coefﬁcient for a spectral line 8DP and ˆ8DP Differential phase for a spectral line σh Standard deviation of ˆRh and ˆJh σv Standard deviation of ˆRv and ˆJv σc Standard deviation of ˆRc and ˆJc σx Standard deviation of ˆRx and ˆJx 6m Error covariance matrix of ˆm 6d Error covariance matrix of ˆd 6b and ˆ6b Error covariance matrix of ˆb 6c and ˆ6c Error covariance matrix of ˆc φ(t) Characteristic function φs(t) Characteristic function for zs φcc(t) Characteristic function for ˆDcc φcx(t) Characteristic function for ˆRcx and ˆJcx φz(t) Characteristic function for zm χ2 k Chi-squared distribution with k degrees of freedom ∗ Complex conjugation sign † Hermitian conjugate sign b Column vector with elements Bhh, Rhv, Jhv, and Bvv ˆb Column vector with elements ˆBhh, ˆRhv, ˆJhv, and ˆBvv B and ˆB 2 × 2 covariance matrix describing polarimetric measurements in a single spectral line on the h–v basis Bhh, ˙Bhv, and Bvv Elements of the covariance matrix B ˆBhh and ˆBvv Diagonal elements of the covariance matrix ˆB c Column vector with elements Bhh, ZDR, ρHV, and 8DP ˆc Column vector with elements ˆBhh, ˆZDR, ˆρHV , and ˆ8DP ˆd Column vector with elements ˆDcc, ˆRcx, ˆJcx, and ˆDxx Dcc and Dxx Diagonal elements of the covariance matrix D ˆDcc, ˆDvv, and ˆDcx Elements of the covariance matrix ˆD D and ˆD 2 × 2 covariance matrix describing polarimetric measurements in a single spectral line on the c–x basis e Measurement column vector on the h–v basis eD Measurement column vector on the c–x basis f ( ˆDcc\|b,Ns) PDF of ˆDcc for a given b and Ns f ( ˆRcx\|b,Ns) PDF of ˆRcx for a given b and Ns f ( ˆJcx\|b,Ns) PDF of ˆJcx for a given b and Ns f ( ˆDxx\|b,Ns) PDF of ˆDxx for a given b and Ns fm( ˆm\|6m) Joint PDF of ˆm for a given 6m fd(ˆd\|b,Ns) Joint PDF of ˆd for a given b and Ns fb(ˆb\|b,Ns) Joint PDF of ˆb for a given b and Ns fc(ˆc\|b,Ns) Joint PDF of ˆc for a given b and Ns i Imaginary unit Ih,v Measured in-phase component measured by the radar receiver in a range bin J12 Imaginary part of ˙q12 ˆJcx Imaginary part of ˆDcx ˆJh and ˆJv Imaginary parts of ˙Sh and ˙Sv, respectively Jhv Imaginary part of ˙Bhv ˆJhv Imaginary part of the covariance between ˙Sh and ˙Sv M T h 15 1333 1354 2022 h //d i /10 5194/ Atmos. Appendix G: Table of symbols Meas. Tech., 15, 1333–1354, 2022 https://doi.org/10.5194/amt-15-13 Atmos. Meas. Tech., 15, 1333–1354, 2022 Atmos. Meas. Tech., 15, 1333–1354, 2022 https://doi.org/10.5194/amt-15-1333-2022 https://doi.org/10.5194/amt-15-1333-2022 A. Myagkov and D. Ori: Analytic characterization of random errors 1351 A. Myagkov and D. Ori: Analytic characterization of random errors Symbol Description Jbd Jacobian of the transformation from ˆb to ˆd Jcb Jacobian of the transformation from ˆc to ˆb Kµ Bessel function of the second kind of order µ ˆm Measurement vector, the elements of which are real and imaginary parts of ˙Sh and ˙Sv Mγ γ th raw statistical moment of a random variable Ns Number of spectra used for averaging Nfft Number of pulses or chirps used to calculate the Doppler spectra Qh.v Measured quadrature component measured by the radar receiver in a range bin Q Matrix used to diagonalize the matrix B q11, ˙q12, and q22 Elements of the matrix Q q Correlation between ˆRh and ˆRv s Correlation between ˆRh and ˆJv R12 Real part of ˙q12 ˆRh and ˆRv Real parts of ˙Sh and ˙Sv, respectively Rhv Real part of ˙Bhv ˆRhv Real part of the covariance between ˙Sh and ˙Sv ˆRcx Real part of ˆDcx ˙Sh,v Measured complex amplitude for a spectral line S The 4 × 4 sensitivity matrix T The transposition sign t Argument of a characteristic function VARbhh Variance of ˆBhh approximated from Bringi and Chandrasekar (2001) VARzdr Variance of ˆZDR approximated from Bringi and Chandrasekar (2001) VARρ Variance of ˆρHV approximated from Bringi and Chandrasekar (2001) VAR8 Variance of ˆ8DP approximated from Bringi and Chandrasekar (2001) ZDR and ˆZDR Differential reﬂectivity for a spectral line zs A sum of squares of independent standard normal samples zm Averaged multiplication of two standard normal variables Disclaimer. Publisher’s note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional afﬁliations. Disclaimer. Publisher’s note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional afﬁliations. Code and data availability. I/Q data used in this study are available on Zenodo (Myagkov and Unal, 2021). MATLAB code used to pro- cess I/Q data is provided in the Supplement to this paper. Ready-to- use MATLAB implementations for Eqs. (36), (39), (46), and (48) are given in the Supplement. Special issue statement. References Hildebrand, P. H. and Sekhon, R. S.: Objective determi- nation of the noise level in Doppler spectra, J. Appl. Meteorol., 13, 808–811, https://doi.org/10.1175/1520- 0450(1974)013<0808:ODOTNL>2.0.CO;2, 1974. Abramowitz, M. and Stegun, I. A.: Handbook of Mathemati- cal Functions With Formulas, Graphs, and Mathematical Ta- bles, in: Applied Mathematics Series 55, 10th edn., edited by: Abramowitz, M. and Stegun, I. A., United States National Bu- reau of Standards, 1972. Hogan, R. J.: A Variational Scheme for Retrieving Rainfall Rate and Hail Reﬂectivity Fraction from Polarization Radar, J. Appl. Meteorol. Clim., 46, 1544, https://doi.org/10.1175/JAM2550.1, 2007. Acquistapace, C., Kneifel, S., Löhnert, U., Kollias, P., Maahn, M., and Bauer-Pfundstein, M.: Optimizing observations of driz- zle onset with millimeter-wavelength radars, Atmos. Meas. Tech., 10, 1783–1802, https://doi.org/10.5194/amt-10-1783- 2017, 2017. Huang, H., Zhao, K., Zhang, G., Hu, D., and Yang, Z.: Opti- mized raindrop size distribution retrieval and quantitative rain- fall estimation from polarimetric radar, J. Hydrol., 580, 124248, https://doi.org/10.1016/j.jhydrol.2019.124248, 2020. Borque, P., Luke, E., and Kollias, P.: On the uniﬁed estima- tion of turbulence eddy dissipation rate using Doppler cloud radars and lidars, J. Geophys. Res.-Atmos., 121, 5972–5989, https://doi.org/10.1002/2015JD024543, 2016. Illingworth, A. J., Hogan, R. J., O’Connor, E. J., Bouniol, D., De- lanoé, J., Pelon, J., Protat, A., Brooks, M. E., Gaussiat, N., Wil- son, D. R., Donovan, D. P., Baltink, H. K., van Zadelhoff, G.-J., Eastment, J. D., Goddard, J. W. F., Wrench, C. L., Haeffelin, M., Krasnov, O. A., Russchenberg, H. W. J., Piriou, J.-M., Vinit, F., Seifert, A., Tompkins, A. M., and Willén, U.: Cloudnet, B. Am. Meteorol. Soc., 88, 883–898, https://doi.org/10.1175/BAMS-88- 6-883, 2007. p g Bringi, V. N. and Chandrasekar, V.: Polarimetric Doppler Weather Radar, Cambridge University Press, ISBN 0521623847, 2001. Bühl, J., Ansmann, A., Seifert, P., Baars, H., and Engelmann, R.: Toward a quantitative characterization of heterogeneous ice formation with lidar/radar: Comparison of CALIPSO/CloudSat with ground-based observations, Geophys. Res. Lett., 40, 4404– 4408, https://doi.org/10.1002/grl.50792, 2013. Kalesse, H., Szyrmer, W., Kneifel, S., Kollias, P., and Luke, E.: Fin- gerprints of a riming event on cloud radar Doppler spectra: ob- servations and modeling, Atmos. Chem. Phys., 16, 2997–3012, https://doi.org/10.5194/acp-16-2997-2016, 2016. Bühl, J., Seifert, P., Engelmann, R., and Ansmann, A.: Im- pact of vertical air motions on ice formation rate in mixed-phase cloud layers, NPJ Clim. Atmos. Sci., 2, 36, https://doi.org/10.1038/s41612-019-0092-6, 2019a. Kanareykin, D. B., Potechin, V. A., and Shishkin, I. F.: Marine Ra- dio Polarimetry, Sudostroenie, 1968. Kneifel, S. Review statement. This paper was edited by Ulrich Löhnert and re- viewed by Dmitri Moisseev and one anonymous referee. Review statement. This paper was edited by Ulrich Löhnert and re- viewed by Dmitri Moisseev and one anonymous referee. Appendix G: Table of symbols This article is part of the special issue “Fu- sion of radar polarimetry and numerical atmospheric modelling to- wards an improved understanding of cloud and precipitation pro- cesses (ACP/AMT/GMD inter-journal SI)”. It is not associated with a conference. Supplement. The supplement related to this article is available on- line at: https://doi.org/10.5194/amt-15-1333-2022-supplement. Author contributions. AM derived equations for PDFs and error covariance matrices, made evaluation using the radar observations, and prepared the ﬁrst draft of the manuscript. DO reviewed the draft and essentially improved the manuscript. Acknowledgements. This work was carried out as a collaboration within the IMPRINT (Understanding Ice Microphysical Processes by combining multi-frequency and spectral Radar polarImetry aNd super-parTicle modelling) project (project no. 408011764), which is a part of the German Research Foundation (DFG) Priority Pro- gram SPP2115 PROM (Fusion of Radar Polarimetry and Numer- ical Atmospheric Modelling Towards an Improved Understanding of Cloud and Precipitation Processes). The authors acknowledge Ruisdael Observatory (the Netherlands) and Christine Unal from TU Delft for granting access to the W-band radar in Cabauw to col- lect I/Q data used in this study. The work of Davide Ori is funded by the German Research Foundation (DFG) under the grant SCHE Competing interests. Alexander Myagkov is an employee of Ra- diometer Physics GmbH, and Davide Ori has no competing inter- ests. https://doi.org/10.5194/amt-15-1333-2022 Atmos. Meas. Tech., 15, 1333–1354, 2022 https://doi.org/10.5194/amt-15-1333-2022 A. Myagkov and D. Ori: Analytic characterization of random errors A. Myagkov and D. Ori: Analytic characterization of random errors 1352 A. Myagkov and D. Ori: Analytic characterization of random errors 2074/1-1 (SPP HALO). The authors thank the editor and the two reviewers for comments and suggestions, which helped to improve the paper. 2074/1-1 (SPP HALO). The authors thank the editor and the two reviewers for comments and suggestions, which helped to improve the paper. Doviak, R. J., Zrnic, D. S., and Sirmans, D. S.: Doppler weather radar, Proc. IEEE, 67, 1522–1553, https://doi.org/10.1109/PROC.1979.11511, 1979. Dufournet, Y. and Russchenberg, H. W. J.: Towards the im- provement of cloud microphysical retrievals using simultane- ous Doppler and polarimetric radar measurements, Atmos. Meas. Tech., 4, 2163–2178, https://doi.org/10.5194/amt-4-2163-2011, 2011. Financial support. Radiometer Physics GmbH covered the publi- cation fees. Görsdorf, U., Lehmann, V., Bauer-Pfundstein, M., Peters, G., Vavriv, D., Vinogradov, V., and Volkov, V.: A 35-GHz polari- metric Doppler radar for long term observations of cloud pa- rameters – Description of system and data processing, J. Atmos. Ocean. Tech., 32, 675–690, https://doi.org/10.1175/JTECH-D- 14-00066.1, 2015. A. Myagkov and D. Ori: Analytic characterization of random errors 1353 Middleton, D.: An Introduction to Statistical Communica- tion Theory: An IEEE Press Classic Reissue, Wiley, ISBN 9780780311787, 1996. Kollias, P., Bharadwaj, N., Clothiaux, E. E., Lamer, K., Oue, M., Hardin, J., Isom, B., Lindenmaier, I., Matthews, A., Luke, E. P., Giangrande, S. E., Johnson, K., Collis, S., Comstock, J., and Mather, J. H.: The ARM Radar Network: At the Leading Edge of Cloud and Precipitation Observations, B. Am. Mete- orol. Soc., 101, E588–E607, https://doi.org/10.1175/BAMS-D- 18-0288.1, 2020. Moisseev, D. N. and Chandrasekar, V.: Nonparametric Estima- tion of Raindrop Size Distributions from Dual-Polarization Radar Spectral Observations, J. Atmos. Ocean. Tech., 24, 1008, https://doi.org/10.1175/JTECH2024.1, 2007. Küchler, N., Kneifel, S., Löhnert, U., Kollias, P., Czekala, H., and Rose, T.: A W-Band Radar-Radiometer System for Accurate and Continuous Monitoring of Clouds and Precipitation, J. Atmos. Ocean. Tech., 34, 2375–2392, https://doi.org/10.1175/JTECH- D-17-0019.1, 2017. Moisseev, D., von Lerber, A., and Tiira, J.: Quantifying the effect of riming on snowfall using ground-based ob- servations, J. Geophys. Res.-Atmos., 122, 4019–4037, https://doi.org/10.1002/2016JD026272, 2017. p g Morrison, H., van Lier-Walqui, M., Fridlind, A. M., Grabowski, W. W., Harrington, J. Y., Hoose, C., Korolev, A., Kumjian, M. R., Milbrandt, J. A., Pawlowska, H., Posselt, D. J., Prat, O. P., Reimel, K. J., Shima, S.-I., van Diedenhoven, B., and Xue, L.: Confronting the Challenge of Modeling Cloud and Pre- cipitation Microphysics, J. Adv. Model. Earth Sy., 12, e01689, https://doi.org/10.1029/2019MS001689, 2020. Kumjian, M.: Principles and Applications of Dual-Polarization Weather Radar. Part I: Description of the Polarimet- ric Radar Variables, J. Operational Meteor., 1, 226–242, https://doi.org/10.15191/nwajom.2013.0119, 2013. Lagarias, J. C., Reeds, J. A., Wright, M. H., and Wright, P. E.: Convergence Properties of the Nelder–Mead Simplex Method in Low Dimensions, SIAM J. Optim., 9, 112–147, https://doi.org/10.1137/S1052623496303470, 1998. Myagkov, A. and Unal, C.: W-band dataset with I/Q mea- surement for an AMT manuscript (1.0), Zenodo [data set], https://doi.org/10.5281/zenodo.5126813, 2021. Lee, J.-S., Hoppel, K. W., Mango, S. A., and Miller, A. R.: Inten- sity and phase statistics of multilook polarimetric and interfero- metric SAR imagery, IEEE T. Geosci. Remote, 32, 1017–1028, https://doi.org/10.1109/36.312890, 1994. Myagkov, A., Seifert, P., Wandinger, U., Bauer-Pfundstein, M., and Matrosov, S. Y.: Effects of antenna patterns on cloud radar polari- metric measurements, J. Atmos. Ocean. Tech., 32, 1813–1828, https://doi.org/10.1175/JTECH-D-15-0045.1, 2015. Lu, Y., Aydin, K., Clothiaux, E. E., and Verlinde, J.: Retrieving Cloud Ice Water Content Using Millimeter- and Centimeter- Wavelength Radar Polarimetric Observables, J. Appl. Meteo- rol. Clim., 54, 596–604, https://doi.org/10.1175/JAMC-D-14- 0169.1, 2015. A. Myagkov and D. Ori: Analytic characterization of random errors Myagkov, A., Seifert, P., Bauer-Pfundstein, M., and Wandinger, U.: Cloud radar with hybrid mode towards estimation of shape and orientation of ice crystals, Atmos. Meas. Tech., 9, 469–489, https://doi.org/10.5194/amt-9-469-2016, 2016a. Marple, S.: Digital Spectral Analysis, 2nd edn., Dover Books on Electrical Engineering, Dover Publications, ISBN 9780486780528, 2019. Myagkov, A., Seifert, P., Wandinger, U., Bühl, J., and Engelmann, R.: Relationship between temperature and apparent shape of pris- tine ice crystals derived from polarimetric cloud radar obser- vations during the ACCEPT campaign, Atmos. Meas. Tech., 9, 3739–3754, https://doi.org/10.5194/amt-9-3739-2016, 2016b. Matrosov, S. Y.: Attenuation-Based Estimates of Rainfall Rates Aloft with Vertically Pointing Ka-Band Radars, J. Atmos. Ocean. Tech., 22, 43, https://doi.org/10.1175/JTECH-1677.1, 2005. Myagkov, A., Kneifel, S., and Rose, T.: Evaluation of the reﬂectivity calibration of W-band radars based on observations in rain, At- mos. Meas. Tech., 13, 5799–5825, https://doi.org/10.5194/amt- 13-5799-2020, 2020. Matrosov, S. Y., Reinking, R. F., Kropﬂi, R. A., Martner, B. E., and Bartram, B. W.: On the Use of Radar Depolarization Ratios for Estimating Shapes of Ice Hydrometeors in Winter Clouds, J. Appl. Meteorol., 40, 479–490, https://doi.org/10.1175/1520- 0450(2001)040<0479:OTUORD>2.0.CO;2, 2001. Nadarajah, S. and Pogány, T. K.: On the distribution of the product of correlated normal random variables, C. R. Math., 354, 201– 204, https://doi.org/10.1016/j.crma.2015.10.019, 2016. Matrosov, S. Y., May, P. T., and Shupe, M. D.: Rainfall Proﬁling Using Atmospheric Radiation Measurement Program Vertically Pointing 8-mm Wavelength Radars, J. Atmos. Ocean. Tech., 23, 1478, https://doi.org/10.1175/JTECH1957.1, 2006. Oue, M., Kumjian, M. R., Lu, Y., Verlinde, J., Aydin, K., and Cloth- iaux, E. E.: Linear depolarization ratios of columnar ice crystals in a deep precipitating system over the Arctic observed by zenith- pointing Ka-band Doppler radar, J. Appl. Meteorol. Clim., 54, 1060–1068, https://doi.org/10.1175/JAMC-D-15-0012.1, 2015. Matrosov, S. Y., Shupe, M. D., and Djalalova, I. V.: Snowfall Retrievals Using Millimeter-Wavelength Cloud Radars, J. Appl. Meteorol. Clim., 47, 769, https://doi.org/10.1175/2007JAMC1768.1, 2008. Oue, M., Kollias, P., Ryzhkov, A., and Luke, E. P.: Toward Ex- ploring the Synergy Between Cloud Radar Polarimetry and Doppler Spectral Analysis in Deep Cold Precipitating Sys- tems in the Arctic, J. Geophys. Res.-Atmos., 123, 2797–2815, https://doi.org/10.1002/2017JD027717, 2018. Matrosov, S. Y., Mace, G. G., Marchand, R., Shupe, M. D., Hallar, A. G., and McCubbin, I. B.: Observations of ice crystal habits with a scanning polarimetric W-band radar at slant linear de- polarization ratio mode, J. Atmos. Ocean. Tech., 29, 989–1008, https://doi.org/10.1175/JTECH-D-11-00131.1, 2012. Pﬁtzenmaier, L., Unal, C. M. H., Dufournet, Y., and Russchen- berg, H. W. References and Moisseev, D.: Long-Term Statistics of Riming in Nonconvective Clouds Derived from Ground-Based Doppler Cloud Radar Observations, J. Atmos. Sci., 77, 3495–3508, https://doi.org/10.1175/JAS-D-20-0007.1, 2020. Bühl, J., Seifert, P., Radenz, M., Baars, H., and Ansmann, A.: Ice crystal number concentration from lidar, cloud radar and radar wind proﬁler measurements, Atmos. Meas. Tech., 12, 6601– 6617, https://doi.org/10.5194/amt-12-6601-2019, 2019b. Kneifel, S., von Lerber, A., Tiira, J., Moisseev, D., Kol- lias, P., and Leinonen, J.: Observed relations between snowfall microphysics and triple-frequency radar measure- ments, J. Geophys. Res.-Atmos., 6034–6055, 2015JD023156, https://doi.org/10.1002/2015JD023156, 2015. Cao, Q., Zhang, G., and Xue, M.: A Variational Approach for Retrieving Raindrop Size Distribution from Polarimetric Radar Measurements in the Presence of Attenuation, J. Appl. Mete- orol. Clim., 52, 169–185, https://doi.org/10.1175/JAMC-D-12- 0101.1, 2013. Kneifel, S., Kollias, P., Battaglia, A., Leinonen, J., Maahn, M., Kalesse, H., and Tridon, F.: First observations of triple- frequency radar Doppler spectra in snowfall: Interpretation and applications, J. Geophys. Res.-Atmos., 43, 2225–2233, https://doi.org/10.1002/2015GL067618, 2016. Chandrasekar, V., Baldini, L., Bharadwaj, N., and Smith, P. L.: Cali- bration procedures for global precipitation-measurement ground- validation radars, URSI Radio Sci. Bull., 2015, 45–73, https: //ieeexplore.ieee.org/document/7909473 (last access: 6 March 2022), 2015. Kollias, P., Clothiaux, E. E., Miller, M. A., Albrecht, B. A., Stephens, G. L., and Ackerman, T. P.: Millimeter-Wavelength Radars: New Frontier in Atmospheric Cloud and Precip- itation Research, B. Am. Meteorol. Soc., 88, 1608–1624, https://doi.org/10.1175/BAMS-88-10-1608, 2007. Chang, W.-Y., Vivekanandan, J., Ikeda, K., and Lin, P.-L.: Quanti- tative Precipitation Estimation of the Epic 2013 Colorado Flood Event: Polarization Radar-Based Variational Scheme, J. Appl. Meteorol. Clim., 55, 1477–1495, https://doi.org/10.1175/JAMC- D-15-0222.1, 2016. https://doi.org/10.5194/amt-15-1333-2022 Atmos. Meas. Tech., 15, 1333–1354, 2022 A. Myagkov and D. Ori: Analytic characterization of random errors 1354 981022740X, World Scientiﬁc, https://doi.org/10.1142/3171, 2000. Walpole, R. E., Myers, R. H., Myers, S. L., and Ye, K.: Probability and Statistics for Engineers and Scientists, 9th edn., edited by: Lynch, D., Prentice Hall, ISBN 9780321629111, 2012. Rusli, S. P., Donovan, D. P., and Russchenberg, H. W. J.: Simultane- ous and synergistic proﬁling of cloud and drizzle properties using ground-based observations, Atmos. Meas. Tech., 10, 4777–4803, https://doi.org/10.5194/amt-10-4777-2017, 2017. Wiener, N.: Generalized harmonic analysis, Acta Math., 55, 117– 258, 1930. Wilks, D. S.: Statistical methods in the atmospheric sciences, 3rd edn., in: International geophysics series, 100, ISBN 9780123850225, 2011. Ryzhkov, A. V., Snyder, J., Carlin, J. T., Khain, A., and Pinsky, M.: What Polarimetric Weather Radars Offer to Cloud Modelers: Forward Radar Operators and Micro- physical/Thermodynamic Retrievals, Atmosphere, 11, 362, https://doi.org/10.3390/atmos11040362, 2020. Yoshikawa, E., Chandrasekar, V., and Ushio, T.: Raindrop Size Distribution (DSD) Retrieval for X-Band Dual- Polarization Radar, J. Atmos. Ocean. Tech., 31, 387–403, https://doi.org/10.1175/JTECH-D-12-00248.1, 2014. Skolnik, M.: Radar Handbook, 3rd edn., McGraw-Hill Education, ISBN 9780071485470, 2008. Zhang, G., Mahale, V. N., Putnam, B. J., Qi, Y., Cao, Q., Byrd, A. D., Bukovcic, P., Zrnic, D. S., Gao, J., Xue, M., Jung, Y., Reeves, H. D., Heinselman, P. L., Ryzhkov, A., Palmer, R. D., Zhang, P., Weber, M., Mcfarquhar, G. M., Moore, B., Zhang, Y., Zhang, J., Vivekanandan, J., Al-Rashid, Y., Ice, R. L., Berkowitz, D. S., Tong, C.-c., Fulton, C., and Doviak, R. J.: Current Status and Future Challenges of Weather Radar Polarimetry: Bridging the Gap between Radar Meteorology/Hydrology/Engineering and Numerical Weather Prediction, Adv. Atmos. Sci., 36, 571– 588, https://doi.org/10.1007/s00376-019-8172-4, 2019. Spek, A. L. J., Unal, C. M. H., Moisseev, D. N., Russchen- berg, H. W. J., Chandrasekar, V., and Dufournet, Y.: A new technique to categorize and retrieve the microphysical proper- ties of ice particles above the melting layer using radar dual- polarization spectral analysis, J. Atmos. Ocean. Tech., 25, 482– 497, https://doi.org/10.1175/2007JTECHA944.1, 2008. Tridon, F. and Battaglia, A.: Dual-frequency radar Doppler spec- tral retrieval of rain drop size distributions and entangled dy- namics variables, J. Geophys. Res.-Atmos., 120, 5585–5601, https://doi.org/10.1002/2014JD023023, 2015. Tridon, F., Battaglia, A., Luke, E., and Kollias, P.: Rain retrieval from dual-frequency radar Doppler spectra: validation and poten- tial for a midlatitude precipitating case-study, Q. J. Roy. Meteor. Soc., 143, 1364–1380, https://doi.org/10.1002/qj.3010, 2017. Tridon, F., Battaglia, A., Chase, R. J., Turk, F. A. Myagkov and D. Ori: Analytic characterization of random errors J.: Observing ice particle growth along fall streaks in mixed-phase clouds using spectral polarimetric radar data, At- mos. Chem. Phys., 18, 7843–7862, https://doi.org/10.5194/acp- 18-7843-2018, 2018. Matrosov, S. Y., Schmitt, C. G., Maahn, M., and de Boer, G.: Atmospheric Ice Particle Shape Estimates from Polarimet- ric Radar Measurements and In Situ Observations, J. Atmos. Ocean. Tech., 34, 2569–2587, https://doi.org/10.1175/JTECH- D-17-0111.1, 2017. Rodgers, C. D.: Inverse Methods for Atmospheric Sounding, Se- ries on Atmospheric, Oceanic and Planetary Physics, 2, ISBN https://doi.org/10.5194/amt-15-1333-2022 Atmos. Meas. Tech., 15, 1333–1354, 2022 https://doi.org/10.5194/amt-15-1333-2022 A. Myagkov and D. Ori: Analytic characterization of random errors J., Leinonen, J., Kneifel, S., Mroz, K., Finlon, J., Bansemer, A., Tanelli, S., Heymsﬁeld, A. J., and Nesbitt, S. W.: The Micro- physics of Stratiform Precipitation During OLYMPEX: Com- patibility Between Triple-Frequency Radar and Airborne In Situ Observations, J. Geophys. Res.-Atmos., 124, 8764–8792, https://doi.org/10.1029/2018JD029858, 2019. Atmos. Meas. Tech., 15, 1333–1354, 2022 https://doi.org/10.5194/amt-15-1333-2022
W4327511997.txt	https://ddd.uab.cat/pub/languesparole/languesparole_a2016n2/languesparole_a2016n2p75.pdf	fr		grammaire des usages du français parlé pour non spécialistes: difficultés et perspectives	Langues & Parole	2,016	cc-by	6,465	Une grammaire des usages du français parlé pour non spécialistes : difficultés et perspectives Sandrine CADDÉO Université Aix Marseille, CNRS, LPL, France Résumé Des grammaires rendant compte des usages oraux du français contemporain devraient pouvoir côtoyer les usuels de la langue pour permettre au public de nonspécialistes d’élargir ses connaissances sur les faits de langues, très partiellement décrits par les grammaires de référence souvent tournées vers le système du français écrit. La question des usages conduit cependant à remettre en question, critiquer ou même invalider les perspectives actuelles sur certains domaines grammaticaux « fossilisés » par plusieurs décennies de tradition grammaticale qui ont façonné la pensée collective (en ce qui concerne entre autres la phrase, le système du genre et du nombre, la notion de non-standard, les genres discursifs). Cet article revient sur quelques-uns des acquis descriptifs qui résultent de l’observation du français parlé et interroge leur recevabilité « didactique ». Mots-clés : Français parlé, grammaire, variété des usages, enseignement. Abstract Grammars of spoken French should be available with the public of nonspecialists to widen his knowledge on the language events, very partially described by the current reference grammars often turned to the system of written French. The writing of such tools could however come up against difficulties: how to make agree to re-examine, to criticize or even to invalidate grammatical domains "fossilized" by several decades of grammatical tradition which shaped the collective thought. This article suggests confronting some cases (among others the sentence, number and gender, the notion of non-standard, register category) and questioning the didactic admissibility. Key words: French spoken, grammar, varieties of use, didactic A PRES le développement, dans les années 2000, de la linguistique de corpus et devant les connaissances accumulées, depuis les années 90, sur le fonctionnement du français parlé, émerge aujourd’hui le souhait de renouveler ou d’enrichir les outils didactiques, comme celui 75 Sandrine Caddéo Une grammaire des usages du français parlé pour non spécialistes : difficultés et perspectives d’élaborer une grammaire de la langue qui rendrait compte du français contemporain tant dans son fonctionnement que dans ses pratiques, à l’oral (Bilger & Cappeau, 2013), ou à l’oral et à l’écrit (Deulofeu & Debaisieux, 2012), sur le modèle de l’anglais (Biber et. al., 1999). Ce type d’ouvrages à destination de non spécialistes fait cruellement défaut pour le français, et les quelques supports recensables qui tiennent compte de la dimension orale et des usages (par exemple Rullier-Theuret, 2010 ou Weber, 2013) sont discrets à côté des grammaires ancrées dans une tradition1 privilégiant la langue standard écrite et une organisation basée sur les catégories grammaticales et la phrase. Aujourd’hui pourtant, il est difficile d’ignorer les apports des travaux sur le français parlé, qui s’appuient sur des données attestées et quantifiées et qui discutent l’existence même d’une seule ou de deux grammaires (l’une de l’écrit, l’autre de l’oral). La tendance est en effet de montrer que les locuteurs exploitent leur grammaire de manière différenciée (Benzitoun et al., à par.). Comme l’écrit Blanche-Benveniste (2010), « La description devra faire état de plusieurs sortes de grammaires d’extensions différentes, qui ne sont pas sollicitées de la même façon par tous les locuteurs ni dans toutes les situations » (préface). Mais ces savoirs peuvent-ils être utilisés à des fins d’exploitation didactique quand on sait qu’ils remettent en question des notions très bien installées dans notre panorama grammatical, comme l’unité « phrase » ou les relations de subordination, qu’ils sont susceptibles de s’appuyer sur des faits non standards pour rendre compte d’organisations inédites ou qu’ils font émerger une grande diversité de micro-systèmes difficilement généralisables ? Nous interrogerons ces différents points en essayant de cerner les difficultés qu’ils posent en lien avec les savoirs véhiculés par les grammaires à vocation pédagogique qui forgent certaines représentations. Nous nous intéresserons principalement à des aspects morphologiques et syntaxiques en portant attention à certains faits de langue parmi les plus problématiques. 1 Dans cet article, nous développons notre réflexion en relation avec une certaine tradition grammaticale sans les précautions qui font l’objet d’une discussion dans Neveu (2007). Les diverses désignations que nous utilisons pourront ainsi correspondre alternativement au discours grammatical ou à l’objet de diffusion des savoirs – les manuels. 76 Sandrine Caddéo Une grammaire des usages du français parlé pour non spécialistes : difficultés et perspectives 1. Quand l’observation des usages oraux réinterroge certaines notions grammaticales établies Force est de constater que dans les grammaires traditionnelles de nombreuses notions grammaticales, et certains critères définitoires qui leur sont associés, n’évoluent guère. Comme l’écrit Béguelin (2000) dans un chapitre traitant du classement en catégories des unités grammaticales : « Qui analyse d’anciens manuels de grammaire s’aperçoit qu’en cent ans, la vision de la langue a évolué sur certains points, mais qu’elle est restée étonnamment stable sur d’autres » (p. 167). 1.1. Une organisation structurelle et informationnelle sans la phrase La notion de phrase – et son corrélat la proposition2 – reste un des points les plus épineux dans les études sur la langue. Du côté des spécialistes du français parlé, elle est largement remise en question et très critiquée car non opératoire pour saisir toutes les organisations exploitées à l’oral (cf. entre autres Béguelin, 2000 ; Blanche-Benveniste, 2002 ; Benzitoun & Sabio, 2010 ; Benzitoun, 2010 ; Sabio, 2011). Mais la notion est au cœur des apprentissages scolaires3, les programmes de français du niveau collège se concentrant même sur « la grammaire de la phrase » (Bulletin Officiel spécial, n°6 du 28 août 2008). Dans les nouvelles directives en lien avec la réforme des programmes du collège intervenue à la rentrée 20164, la définition de la phrase est ramenée au canon « GN sujet – GV prédicat » et c’est une définition tirée de la Grammaire méthodique du Français ([1994] 2009) qui est utilisée pour faire intervenir « la structure fondamentalement bipartite de la phrase canonique » (p. 240). La large diffusion de cette perspective comme un des fondamentaux de la langue française ne permet pas d’être optimiste sur la manière dont pourraient être intégrées les nombreuses autres formes d’organisation que tout locuteur exploite dans ses discours et qu’il est difficile de faire « dériver » de la phrase canonique. Au mieux, seront2 Pour une synthèse de l’histoire et de l’évolution des deux notions, Combettes, 2011. Notons que même chez les spécialistes du langage, la notion est fréquemment utilisée même si elle est critiquée, c’est ce que l’on peut constater dans l’ouvrage de Detey et al. (2010) consacré aux variétés du français parlé et dont le titre d’une des parties est « Syntaxe, ou autour de la phrase » (p.86). 4 Cf. le document intitulé « Etude de la langue au collège : quelles simplifications et pourquoi ? » consultable ici : http://www.lettres.acversailles.fr/IMG/pdf/etude_de_la_langue_ simplifications_ terminologiques.pdf [lien vérifié le 20/10/2016] 3 77 Sandrine Caddéo Une grammaire des usages du français parlé pour non spécialistes : difficultés et perspectives elles classées – comme c’est déjà le cas pour certaines structures – dans les « phrases atypiques » taxées de « rebelles à l’analyse »5 : À côté des phrases canoniques, il existe des phrases atypiques : structures fréquemment employées, notamment à l’oral, mais difficiles, voire impossibles à dériver d’une phrase canonique. Il s’agit des phrases à présentatif (« Il y a longtemps que je t’aime. » ; « Voilà cinq ans qu’il a déménagé. »), des phrases non verbales (« Drôle d’endroit pour une rencontre ! » ; « Génial, ce film ! » ; « À cœur vaillant rien d’impossible. »), des phrases comportant des incises (« Comprends-moi bien, dit Jacques, je ne refuse pas de t’aider. ») ou des incidentes (« Il viendra, j’espère. » ; « Les mythes, c’est bien connu, ont la vie dure. »), etc. Si ces phrases sont souvent rebelles à l’analyse, on doit toutefois pouvoir les exploiter autrement en classe, en faisant travailler les élèves, quand ils les rencontrent, sur leur transformation en phrases canoniques et sur la comparaison entre les deux (ou davantage) manières de dire. (p. 2) C’est ignorer les travaux existants sur ces faits de langue. Par exemple, lorsque Sabio (2006) s’interroge sur l’ordre donné comme contraint du complément d’objet direct par rapport au verbe, il aboutit à deux conclusions. À partir de données attestées – écrites et orales –, il montre que la propriété classificatoire des compléments (position libre pour les circonstanciels et fixe pour les essentiels) est discutable car le complément d’objet direct accepte d’être antéposé. Il cite des exemples du type Un bon disque, on va s’écouter, d’accord ? (écrit) ou Deux cigarettes j’ai fumé (oral). Une analyse distributionnelle qui s’appuie sur des critères prosodiques, syntaxiques et lexicaux lui permet également d’identifier deux organisations distinctes. Grosso modo, au niveau informationnel, dans La bourgeoisie, j’ai pas connu, le focus est mis sur la partie verbale et dans A peine huit ans il a, c’est le complément antéposé qui est saillant (Sabio 2006, 75). Le nœud du problème réside sans doute dans le choix à faire sur le cadre d’analyse, celui du cadre phrastique étant « très contraignant » (Béguelin, 2000, 132), à moins que l’enjeu ne soit le fait de ne pas privilégier un cadre d’analyse, mais faut-il – et peut-on – s’en passer complètement ou, à l’exemple de La grammaire méthodique du français doiton adopter « plus d’un cadre théorique » (Riegel et al., 1994, Avantpropos, p. XVI), mais alors lesquels ? 5 Cf. le document cité note précédente. 78 Sandrine Caddéo 1.2. Une grammaire des usages du français parlé pour non spécialistes : difficultés et perspectives Une organisation morphologique différente à l’écrit et à l’oral L’importance accordée à la maîtrise de l’orthographe dans l’enseignement soumet à un seul point de vue la manière dont est saisie l’organisation de certaines catégories morphologiques comme le genre, le nombre et le verbe en français. Les questions de morphologie issues de l’observation de l’oral redessinent pourtant profondément le système du français, surtout dans le domaine de la flexion : (…) la grammaire morphologique du français écrit et celle du français parlé appartiennent à deux typologies différentes (Blanche-Benveniste, 2008, 200). Sur la question du nombre, par exemple, la description des données orales confirme la faible fréquence de formes audibles de pluriel et surtout l’existence de nombreux autres procédés complémentaires qui ne relèvent pas de la morphologie : La morphologie du nombre est peu marquée en français parlé sur le nom lui-même, et elle a un caractère nettement moins flexionnel que dans le français écrit, où tous les noms sont susceptibles d’avoir une marque. Le nombre y est indiqué bien davantage au niveau du syntagme nominal, par les déterminants et les quantifieurs, par des phénomènes de sélection lexicale, ou simplement par la bonne plausibilité sémantique. Pour le décrire dans toutes ces dimensions, il faut nécessairement adopter une démarche « holistique », qui dépasse le strict cadre morphologique (Blanche-Benveniste, 2004, 149). Et bien que pour chaque catégorie qui accepte la flexion en nombre, des précautions d’usage sont régulièrement apportées par les grammaires sur le fait que tous les membres ne sont pas concernés, le trait morphologique reste un des critères premiers pour répartir les unités selon les deux entrées « variables » et « invariables » ; ce qui amoindrit considérablement d’une part la place accordée aux autres voies que la langue exploite pour signifier le pluriel ; d’autre part l’idée – à accepter – que des noms, des pronoms ou des adjectifs peuvent être « invariables ». Dans une perspective plus générale, Blanche-Benveniste (2007, 131) rappelle un principe de morphologie soustractive qui concerne de nombreuses catégories et dont les grammaires ne parlent pas : 79 Sandrine Caddéo Une grammaire des usages du français parlé pour non spécialistes : difficultés et perspectives On doit mentionner par exemple l’importance prise en français parlé par un type de morphologie soustractive : différence entre les personnes du singulier et du pluriel de certains verbes au présent de l’indicatif, le singulier étant plus court que le pluriel ; différence entre féminin et masculin de certains adjectifs, le masculin étant plus court que le féminin. L’écart important qui existe dans la manière dont s’organisent les phénomènes morphologiques entre l’oral et l’écrit ne peut continuer à être ignoré par les supports pédagogiques. Peut-être bien plus clairement qu’en syntaxe, la dimension morphologique met en évidence le bien fondé d’intégrer aux analyses de la langue la dimension orale. Ses questions relèvent cependant d’un niveau de technicité (dimensions phonétique et phonologique, saisie de la notion de morphèmes, allomorphes, supplétismes, etc.) moins immédiatement accessible que ne sont les règles d’orthographe. Peut-on envisager un compromis ? 2. Quand le non standard participe de la description La prise en compte des fautes dans la perspective dite traditionnelle a souvent comme fonction principale de présenter les contre-exemples (« ce qu’il ne faut pas dire »). Sous la plume des plus puristes, la mention même de ces réalisations non standards de la langue les fait paradoxalement exister. Prisonniers d’un idéal associé au standard dont les modèles sont essentiellement pris dans l’écrit littéraire, les grammairiens qui ont ce discours s’appuient sur ces cas pour mieux les écarter. Dans les approches descriptives, la fréquence de certaines fautes au contraire interroge et fait progresser la réflexion. 2.1. Le non standard sous le prisme de la variation Un manuel de langue française prenant en compte l’oral et l’écrit et qui présente un chapitre intitulé « L’oral littéraire et les fautes typantes » (Rullier-Theuret, 2010) met sur le même plan les fautes sur la sélection du relatif, l’absence du « ne » de négation et certains points ayant trait aux emplois des pronoms6. Or ces phénomènes ne sont pas strictement comparables ; ce qui a conduit Blanche-Benveniste (2000, 37) à proposer de ne plus classer comme fautes celles « fréquemment attestées chez tout 6 L’auteur explique cependant que son objectif est de relever dans la littérature des exemples que les auteurs auraient sélectionnés pour donner un effet d’oralité (p.74). 80 Sandrine Caddéo Une grammaire des usages du français parlé pour non spécialistes : difficultés et perspectives le monde » (p. 37). La présence ou l’absence du « ne » de négation, l’alternance de « on » et « nous » ou de « ça » et « cela » peuvent être traités en termes de variation diaphasique en opposant registre soutenu et registre familier, même si l’observation de leur distribution ne permet pas de réduire l’alternance à une simple question de registres7. Pour un certain nombre de faits linguistiques qu’on taxerait de fautes banales, les locuteurs eux-mêmes conviennent d’un certain degré d’acceptabilité. À l’opposé, les erreurs de sélection des pronoms relatifs sont jugées plus sévèrement et l’emploi systématique d’un « que » à la place de certains autres relatifs résiste au jugement diaphasique. Ces fautes sont dites « typantes » par Blanche-Benveniste (2000, 41) parce qu’elles « agissent comme des marquages sociaux (…) ». Ces relatives « non standards » sont assez fréquemment étudiées (entres autres Gadet, 1989 ; Béguelin, 2000, 311 ; Detey et al., 2010, 94 ; Blanche-Benveniste, 2010, 94), mais le degré d’importance qu’on leur accorde est finalement disproportionné car elles ne sont pas très fréquentes à l’oral (BlancheBenveniste, 2010, 98). Les phénomènes non standards qui sont taxés de fautes typantes correspondent généralement à celles qui sont corrigées par l’école ou l’environnement (Blanche-Benveniste, 2000, 41) : l’indicatif dans des contextes où est attendu le subjonctif (« vous souhaitez que je viens directement chez vous »), un relatif renforcé par « que » (du type « où c’est que »), des accords non conformes (« c’est moi qui a ») ou non réalisés (« la veste que j’ai pris »), la sélection d’une forme de conditionnel derrière « si » (« s’il voudrait trouver un emploi … »), l’emploi de « malgré que », etc. 2.2. Le non-standard intégré dans l’organisation du système Au-delà des considérations variationnistes, les études qui tiennent compte des énoncés non standards éclairent certaines zones de fonctionnement du français contemporain. Comme illustration, nous reviendrons en premier lieu sur le domaine du nombre. Le français parlé est considéré comme « sousmarqué » par rapport à l’écrit. Les différences audibles qui permettraient 7 Pour l’étude du pronom « on », Blanche-Benveniste, 2003. 81 Sandrine Caddéo Une grammaire des usages du français parlé pour non spécialistes : difficultés et perspectives d’identifier un pluriel ou un féminin (les formes dites marquées, ou longues) sont statistiquement peu nombreuses si on les compare aux cas où la différence est neutralisée ou organisée différemment par rapport à l’écrit (Blanche-Benveniste, 2000, 140). La grande fréquence de fautes sur la liaison par le développement d’un [z] non requis est possiblement une des voies compensatoires développées par les locuteurs : Au total, le marquage en nombre, qui dépend des types de noms et de déterminants ainsi que du caractère vocalique ou consonantique des initiales, est assez aléatoire. C’est sans doute pour compenser les lacunes de ce marquage que se sont développés les [z] de « fausses liaisons » qui indiquent le pluriel à différents endroits des syntagmes, bien au-delà de leur domaine normatif d’application (BlancheBenveniste, 2010, 56). De nombreuses autres réalisations sont classées comme non normatives alors qu’elles présentent des distributions remarquables qui ne justifient pas leur rejet. Blanche-Benveniste (2010) en cite plusieurs cas : certains noms à emploi conjonctionnel comme « sous condition de » et « sous condition que » ou à valeur de préposition comme « limite » dans j’avais pas trop peur – limite c’était excitant (p. 108-109) ; des emplois à complément zéro de certains verbes lorsqu’ils sont impliqués dans des relations d’anaphore, comme « aimer » dans « l’hélicoptère c’est bien – j’ai bien aimé » (p. 138) ou bien les emplois de « pas de » et « pas des », souvent traités sous l’angle du raccourcissement de la forme « des » dans des contextes spécifiques (par exemple en contexte négatif). L’auteur explique que dans l’usage non normatif, la répartition est sensible à la catégorie sur laquelle porte la modalité négative ; ce qui est une forme de régularisation : L’usage normatif répartit les deux formes pas de et pas des selon le type de verbe, à valeur identifiante ou non. L’usage non normatif a une autre répartition : pas des pour tous les verbes, quels qu’ils soient ; pas de seulement pour les constructions non verbales (p. 147). Certains phénomènes ont longtemps été écartés des analyses descriptives sur les seuls critères qu’ils étaient mis au service de l’expression et/ou qu’ils relevaient plutôt de l’oral. C’est le cas de la dislocation, qui est associée aux procédés d’emphase. Les nombreux 82 Sandrine Caddéo Une grammaire des usages du français parlé pour non spécialistes : difficultés et perspectives travaux sur la question8, grâce le plus souvent aux données orales, ont permis simultanément d’identifier une grande diversité de formes et de dégager des critères de description permettant de les inscrire dans l’organisation grammaticale du français. À ce titre, il n’est plus possible de les marginaliser. La notion de faute interroge. Un grand nombre d’énoncés taxés de fautifs ne sont pas détectés comme tels par les locuteurs et finalement les plus stigmatisés concernent quasiment toujours les mêmes domaines. Le biais de la variation pourrait permettre de développer chez le nonspécialiste plus de distanciation dans ses jugements. Pour autant, la prise en compte des énoncés non standards dans une optique pédagogique sertelle objectivement l’idée de donner à voir la langue telle qu’elle se parle ? 3. Quand se multiplient les micro-systèmes à l’intérieur du système Les études syntaxiques sur le français parlé s’accordent sur l’idée qu’il existe une collection de micro-systèmes, sensibles à des facteurs divers, qui empêchent d’avoir une vision globale et homogène du fonctionnement de la langue. Plusieurs facteurs peuvent en partie expliquer le morcellement d’un phénomène grammatical ; nous en présenterons deux : ce qui relève du savoir inné et du savoir acquis, réinterprété par Blanche-Benveniste (1990) sous l’opposition « grammaire première/grammaire seconde » et la question des genres, qui reste une problématique très sensible dans la réflexion sur la typologie des discours. 3.1. Les faits de langues et les « savoirs » grammaticaux Dans un même domaine, les faits de langue sont généralement décrits avec un degré d’importance similaire par les grammaires usuelles alors qu’ils sont exploités très inégalement par les locuteurs. Dans une grammaire qui tiendrait compte des usages, une certaine hiérarchisation s’imposerait même si, pour cela, des emplois taxés de familiers viendraient en premier plan. C’est ce que montrent Bilger & 8 Blasco-Dulbecco (1999) y a consacré un ouvrage, et Berrendonner (2015) propose une mise au point plus récente. 83 Sandrine Caddéo Une grammaire des usages du français parlé pour non spécialistes : difficultés et perspectives Cappeau (2013) à partir de l’étude de plusieurs phénomènes différents sur des corpus de français parlé et écrit échantillonnés. À partir de la distribution de même selon sa catégorie d’appartenance et ses valeurs sémantiques, ils concluent que Cette analyse souligne parfois l’existence d’une discordance entre l’usage d’une forme et les développements qu’elle a suscités dans les grammaires ou les manuels, et elle a permis de mettre en lumière des associations parfois négligées (p. 196). Entre autres observations, nous notons, par exemple, que les locuteurs privilégient à l’oral « même pas » plutôt que la forme recommandée « pas même » (p. 194), que la locution « même que », dite typique de l’oral, est absente de leur corpus d’études (p. 195), ainsi que « quand bien même » à valeur concessive, alors que la tournure occupe une place importante dans les grammaires (p. 193). D’autres points d’observations complémentaires sont réunis dans le tableau suivant issu de leur étude (p. 196) : Même forme invariable Quand même Même modalisateur Même + conjonction Utilisation très importante Devant une préposition : Même si : absent de l’écrit à l’oral, rare à l’écrit s’accompagne à l’oral littéraire, bien présent à d’une antéposition l’oral et dans la presse fréquente du Syntagme Préposition (même à un Placé plutôt à l’intérieur Bordelais je le dis) de la construction Même quand : assez rare Pas même : dans l’écrit Quand bien même : souvent littéraire Même que : absent : doit décrit, peu attesté souvent être réanalysé Même pas : à l’oral Tableau 1 : Les principaux emplois de même invariable (Bilger et Cappeau, 2013) Ces fortes disproportions d’emplois que l’on constate régulièrement pour des faits linguistiques selon qu’ils sont étudiés selon leurs usages ou leur traitement dans les grammaires pourraient s’expliquer partiellement à partir des notions de « grammaire première » 84 Sandrine Caddéo Une grammaire des usages du français parlé pour non spécialistes : difficultés et perspectives et de grammaire seconde »9 proposées par Blanche-Benveniste (1990). La « grammaire première », concerne les faits de langue que les locuteurs adoptent naturellement dans les usages, même ceux d’emplois limités, mais qui sont sollicités (presque de manière prédictible) selon divers facteurs. Par exemple, Blanche-Benveniste (2010) affirme, au sujet des passés simples qu’ils « […] subsistent, dans de faibles proportions, mais [qu’]ils subsistent. Ils sont réservés à certaines situations particulières et à certains locuteurs. (p. 206). L’auteur mentionne les enfants, qui y sont attachés, les adultes qui par leur profession (avocats, guides touristiques, professeurs) sont habitués à parler à partir de supports écrits. L’autre observation concerne le type discursif : le récit, qui semble favoriser naturellement le passé simple (p. 106). Le savoir de la « grammaire seconde » relèverait de l’apprentissage scolaire imposant « une tournure qui [se serait] partiellement installée dans les usages, sans jamais avoir été vraiment ‘productive’. » (BlancheBenveniste, 1990, 71). Le cas de « quand bien même » de l’étude précédente pourrait en être une illustration. Deulofeu (2000) propose une extension de la notion qui pourrait rappeler celle de sociolecte lorsqu’il s’interroge sur la manière « de définir un genre au sens grammatical du terme » (p. 272). Le niveau de grammaire seconde tiendrait « à la volonté et à la capacité du locuteur à utiliser des ressources linguistiques codifiées, réservées à des usages qui ne sont pas familiers et spontanés » (p. 273). 3.2. La répartition des faits de langue dans les « genres » Le développement important des études sur corpus, qui est allé de pair avec une réflexion sur les genres à l’oral, a permis de montrer que les faits de langue ne se répartissent pas avec la même fréquence selon les types de discours. Cappeau (2001) rappelle par exemple la forte proportion de sujets lexicaux présents dans les explications techniques, la concentration de séquences en une fois que dans les recettes de cuisine ou encore la fréquence de « certains » non anaphorique dans les débats politiques. Dans les interviews politiques, Blasco-Dulbecco & Cappeau (2012) identifient, entre autres, l’exploitation importante de certaines relatives du type qui est celui, un 9 Une intéressante discussion sur ces notions a été publiée par Elalouf (2012). 85 Sandrine Caddéo Une grammaire des usages du français parlé pour non spécialistes : difficultés et perspectives N1 qui est un N1 et qui est le mien. Benzitoun (2013) montre des régularités de placement de l’adjectif dans l’oral non planifié. BlancheBenveniste (2000) remarque successivement que les passés simples apparaissent souvent « dans des situations que les locuteurs ressentent comme solennelles » (p. 54) ; que « les passifs canoniques (…) sont presque totalement absents des conversations et des récits familiers » (p. 57) ; que le « langage professionnel » recèle de nombreuses nominalisations (p. 63) et « que la langue de conversation utilise régulièrement il y a et qui pour encadrer [des sujets indéfinis] » (p. 93). Les cas à citer pourraient être encore nombreux, mais nous choisissons d’arrêter notre inventaire parce qu’il présente deux défauts. Premièrement, le fait en lui-même n’est parlant qu’en relation avec le faisceau de critères dans lequel il s’inscrit, sans quoi les observations perdent de leur légitimité. Ainsi pour considérer comme remarquable la fréquence importante de sujets lexicaux dans les explications techniques, il faut savoir que l’oral est plutôt caractérisé par des sujets sous forme pronominale : Cette propriété qui peut paraître banale lorsque l’on est habitué à travailler sur l’écrit, devient significative à l’oral (qui comporte une grande quantité de sujets pronominaux) (Cappeau, 2001, 71). Lorsque Blanche-Benveniste (2000, 57) remarque la présence de passifs dans les récits de faits divers ou dans les reportages sportifs, elle les lie à un fonctionnement spécifique à savoir « quand il s’agit d’exprimer un déroulement progressif ou d’enchaîner une série d’événements ». Deuxièmement, comme le montrent Bilger & Cappeau (2004, 28) sur la question des registres de langue : Il n’y a sans doute pas plus de “genre” et de “style” différents à l’oral par rapport à l’écrit, mais il n’y en pas moins non plus. Pour s’en convaincre, il suffit de travailler sur des productions variées et suffisamment longues. En revanche, il semblerait que dans les productions orales, les variations soient à la fois plus rapides et plus nombreuses qu’à l’écrit dans la mesure où elles dépendent d’un faisceau de facteurs hétérogènes comme : - la situation de parole ou le genre qui jouent un rôle déterminant. Le locuteur respecte un certain nombre de contraintes que lui impose le “cadre” dans lequel il parle. Il n’utilisera pas exactement les mêmes outils syntaxiques lorsqu’il raconte un voyage et lorsqu’il explique le fonctionnement d’un appareil ; 86 Sandrine Caddéo Une grammaire des usages du français parlé pour non spécialistes : difficultés et perspectives - la posture prise par le locuteur. Ainsi, le locuteur peut devenir le porte–parole d’un groupe, ce qui modifiera profondément son style. De même, il peut, lorsqu’il parle de son activité professionnelle, recourir à un “jargon” (sous la forme de tournures syntaxiques incluant par exemple des nominalisations) qu’il abandonne dès qu’il s’écarte de ce statut. La saisie des différences de registres est donc plus fine en français parlé puisqu’elle peut apparaître à l’intérieur même d’un discours dont les traits généraux ont pu été pré-caractérisées : Nous avons noté dans les corpus longs des phénomènes qui permettent d’observer à quel moment les locuteurs sont amenés, parfois, à s’exprimer dans un registre différent, en dépit d’une même situation de prise de parole, en dépit aussi du genre dans lequel ils évoluent (Bilger & Cappeau, 2004, 21). Les remarques qui font un lien entre phénomènes linguistiques et registres de langue ne peuvent pas être séparées du contexte d’emploi dans lequel les formes ont été relevées, ni du domaine auquel elles appartiennent afin de pouvoir faire des comparaisons avec d’autres réalisations de même type. Dans cette perspective, les connaissances à transmettre relève du savoir du spécialiste si on ne veut pas voir appauvrir la réflexion sous le couvert de rendre la matière enseignable. N’est-ce pas trop ambitieux ? Enfin, pour dépasser les organisations en registres classiques exploitées par les grammaires et qui relèvent d’oppositions du type langue familière, courante et surveillée, que proposer ? L’absence de typologie encore claire concernant les discours oraux, comme le laissent penser les nombreuses précautions que les auteurs prennent lorsqu’ils présentent les critères d’échantillonnage de leur corpus, n’offre pas de solutions immédiates. 4. Conclusion Nous nous sommes intéressée à la problématique d’une élaboration d’une grammaire des usages pour non spécialistes à la lecture de certains ouvrages didactiques qui ont certes le mérite de tenir compte des données orales, mais qui nous semblaient aussi véhiculer encore les partis-pris les plus classiques. L’ouvrage de FLE de Weber (2013) dont le sous-titre est « Enseigner le français tel qu’il est parlé » développe un chapitre sur la syntaxe à partir des points de différence existant exclusivement entre l’oral et l’écrit (p. 157) ; il exploite l’idée que les organisations spécifiques s’expliquent en partie par des principes de 87 Sandrine Caddéo Une grammaire des usages du français parlé pour non spécialistes : difficultés et perspectives facilitation (§5.4, p. 164), de rapidité (p. 166) et de moindre effort (p. 169) ; enfin il conclue sur l’idée que la spécificité de la syntaxe de l’oral est sa plus grande liberté (p. 173). Ces perspectives – déjà critiquées en 1987 par Blanche-Benveniste et Jeanjean – ne devraient plus se rencontrer, surtout s’ils concernent les faits syntaxiques de la langue. Nous adhérons également aux observations qui relèvent les nombreux défauts des grammaires traditionnelles (Suso, 2004) et qui rendent nécessaires la prise en compte de la langue dans ses usages. Cependant, tout au long de cet article, tout en rendant compte des avancées considérables sur l’analyse de l’oral que les études en linguistique française ont apportées, nous avons émis des réserves sur la manière dont un usuel pourrait « absorber » la dimension des usages. Deulofeu & Debaisieux (2012, 27) exprimaient le vœu que « toute grammaire de référence d’une langue [se présente] à la fois comme une somme organisée des connaissances ou des savoirs que les locuteurs ou les spécialistes ont sur cette langue, et comme un recensement des usages effectifs que font les locuteurs de cette langue dans les divers registres selon lesquels elle est utilisée ». Mais la combinaison des deux dimensions reste peut-être ambitieuse si l’objectif visé est de la rendre accessible à des non spécialistes. Le risque pourrait être de « remplacer » une approche hétérogène (la règle et ses exceptions) par une autre (la multi-dimension des organisations). L’acceptation de la remise en question de certaines idées qui font poids dans la tradition va-t-elle de soi ? Par exemple, certaines catégories ont une forme étroitement associée à une fonction, comme les conjonctions de subordination. De nombreux travaux (Debaisieux, 2007 ; 2013) montrent pourtant que la présence d’une conjonction de subordination ne garantit pas une dépendance hiérarchisée – et donc une subordination. Debaisieux (2002) identifie par exemple deux emplois de « parce que »10 : l’un comme subordonnant dans Il est parti parce qu’il était fatigué, qui exprime très clairement la cause, et l’autre comme connecteur discursif dans il est malade, parce qu’il a de la fièvre, emploi très fréquent dans les usages oraux et qui marque deux actes énonciatifs. 10 Les exemples sont de l’auteur. 88 Sandrine Caddéo Une grammaire des usages du français parlé pour non spécialistes : difficultés et perspectives Sur la question du non standard, les grammaires ont finalement toujours présenté des réalisations langagières non totalement conformes aux règles énoncées. Grevisse et « Le bon usage » (qui en est à sa 16e édition en 2016 et se décline en version électronique) était précurseur dans ce domaine. Mais quelles que soient les sources (elles étaient littéraires dans « Le bon usage »), dans l’esprit du lecteur, les faits restent originaux, fautifs ou stylistiques. L’intégration du non standard dans une grammaire – même si elle est légitimée par la notion de fréquence – nous paraît devoir s’accompagner d’un certain nombre de précautions. La dimension quantitative pourrait être un argument de présentation des phénomènes dans des dimensions comparatives (par exemple dans les registres les plus facilement identifiables malgré les problématiques encore très présentes) afin que tout apprenant (L1 comme L2) – ou enseignant – puisse appréhender les usages d’emblée dans leur diversité tout en étant aidé pour repérer les grandes tendances. Bien comprendre qu’il s’agit de « tendances » nous paraît évidemment primordial parce qu’à souhaiter rendre compte de la langue « telle qu’elle se parle » pour l’enseignement, les risques résident dans la manière dont le message est diffusé du côté des enseignants et reçu du côté des apprenants : il ne s’agit pas de réduire la répartition des faits aux fréquences et à leurs variations au risque de faire apparaître l’idée de nouvelles règles ! Références bibliographiques BEGUELIN, M.-J., De la phrase aux énoncés : grammaire scolaire et descriptions linguistiques, Bruxelles, De Boeck/Duculot, 2000. BENZITOUN, C., Quelle(s) unité(s) syntaxique(s) maximale(s) en français parlé ? Discussions autour de quelques problèmes rencontrés, Travaux de linguistique 1/2010, 60, 2010, 129-126 [URL : www.cairn.info/revue-travaux-de-linguistique2010-1-page-109.htm] BENZITOUN, C., Adjectifs épithètes alternants en français parlé : premiers résultats, TIPA. Travaux interdisciplinaires sur la parole et le langage, 2013 [En ligne], 29 \| 2013, mis en ligne le 19 décembre 2013, consulté le 20 décembre 2013. URL : http://tipa.revues.org/585 89 Sandrine Caddéo Une grammaire des usages du français parlé pour non spécialistes : difficultés et perspectives BENZITOUN, C., SABIO, F., Où finit la phrase ? Où commence le texte ?, Discours, 2010 [En ligne], 7, mis en ligne le 20 décembre 2010. URL http://discours.revues.org/7966 BENZITOUN, C., CAPPEAU, P., CORMINBOEUF, G., (sous presse), Réflexions sur les exploitations différenciées de la grammaire, Revue de Sémantique et de Pragmatique. BERRENDONNER, A., Constructions disloquées, Encyclopédie Grammaticale du Français, 2015 [en ligne : http://encyclogram.fr] BIBER, D., Variation across Speech and Writing, Cambridge, Cambridge University Press, 1988. BILGER, M., CAPPEAU, P., L’oral ou la multiplication des styles, Langage et société, 2004/3, 109, 2004, 13-30. BILGER, M., CAPPEAU, P., Comment les données de corpus pourraient renouveler les manuels de grammaire ?, Linx, 2013 [En ligne], 68-69 \| 2013 [mis en ligne le 19 novembre 2015, consulté le 07/11/2016. URL : http://linx.revues.org/1526 BLANCHE-BENVENISTE, C., JEANJEAN, C., Le français parlé. Transcription et édition, Paris, Didier érudition, 1987. BLANCHE-BENVENISTE, C., Grammaire première et grammaire seconde : l’exemple de en, Recherches sur le français parlé, 10, 1990, 51-73. BLANCHE-BENVENISTE, C., Approches de la langue parlée en français, Paris, Ophrys, 2000. BLANCHE-BENVENISTE, C., Phrase et construction verbale, in CHAROLLES, M. (éd.), Actes du colloque Y a-t-il une syntaxe au-delà de la phrase ?, Verbum, 24/1-2, 2002, 722. BLANCHE-BENVENISTE, C., Le double jeu du pronom ON, La syntaxe raisonnée, Mélanges de linguistique générale et française offerts à Annie Boone à l’occasion de son 60° anniversaire, HADERMANN, P., VAN SLIJCKE, A., BERRE, M. (éds), De Boeck/Duculot, 2003, 43-56. BLANCHE-BENVENISTE, C., Le singulier et le pluriel en français parlé contemporain, Bulletin de la Société de linguistique de Paris, T. XCIX, fasc. 1, 2004, 129-154. BLANCHE-BENVENISTE, C., Corpus de langue parlée et description grammaticale de la langue, Langage et société, 2007/3, 121-122, 2007, 129-141. BLANCHE-BENVENISTE, C., Les unités de la langue écrite et de la langue parlée, Cahiers de l’Université de Perpignan 37, in BILGER, M. (coord.), Données orales. Les enjeux de la transcription, 2008, 192-216. BLANCHE-BENVENISTE, C., Le français. Usages de la langue parlée, Leuven/Paris, Peeters, 2010. BLASCO, M., Les dislocations en français contemporain : étude syntaxique, Paris, Champion, 1999. 90 Sandrine Caddéo Une grammaire des usages du français parlé pour non spécialistes : difficultés et perspectives BLASCO-DULBECCO, M., CAPPEAU, P., L’interface syntaxe/lexique : illustrations à travers les corpus oraux, Cahiers du LRL, 1, Clermont-Ferrand, Presses Universitaires de Blaise Pascal, 2007, 105-121. BLASCO-DULBECCO, M., CAPPEAU, P., Identifier et caractériser un genre : l’exemple des interviews politiques, Langages, 2012/3, 187, 2012, 27-40. CAPPEAU, P., Faits de syntaxe et genres à l’oral, Le français dans le monde, N° spécial, Oral : variabilité et apprentissages, janvier 2001, 2001, 69-77. COMBETTES, B., Phrase et proposition. Histoire et évolution de deux notions grammaticales, Le français aujourd’hui, 2011/2, 173, 2011, 11-20. DEBAISIEUX, J.-M., Le fonctionnement de parce que en français parlé : étude quantitative sur corpus, PUSCH, C.D., RAIBLE, W. (dir.), Romance corpus linguistics: corpora and spoken language, Gunter Narr Verlag Tübingen, 2002, 349-362. DEBAISIEUX, J.-M., La distinction entre dépendance grammaticale et dépendance macrosyntaxique comme moyen de résoudre les paradoxes de la subordination, Faits de langues, Peter Lang, 2007, 119-132. DEBAISIEUX, J.-M., (dir), Analyses linguistiques sur corpus. Subordination et insubordination en français, Traité IC2, série Cognition et traitement de l’information, Lavoisier, 2013. DEULOFEU, J., Les commentaires sportifs constituent-ils un genre au sens linguistique du terme ?, Corpus. Méthodologie et applications linguistiques, dirigé par M. Bilger, Paris, Champion, 2000, 271-295. DEULOFEU, J., DEBAISIEUX, J.-M., Une tâche à accomplir pour la linguistique française du XXIe siècle : élaborer une grammaire des usages du français, Langue française 4, 176, 2012, 2012, 27-46 [URL : www.cairn.info/revue-langue-francaise 2012-4-page-27.htm. ELALOUF, A., La notion de “grammaire seconde” : tentative de reconstruction épistémologique », in Actes du CMLF 2012 – IIIe Congrès mondial de linguistique française. 2012, EDP sciences. [En ligne : http://dx.doi.org/10.1051/shsconf/ 20120100328] GADET, F., Les relatives non standards saisies par les grammaires, LINX, 20, 1989, 7-49. NEVEU, F., LAUWERS, P., La notion de « tradition grammaticale » et son usage en linguistique française, Langages 3/2007, 167, 2007, 7-26 [URL : www.cairn.info/revue-langages-2007-3-page-7.htm] SABIO, F., Phrases et constructions verbales : quelques remarques sur les unités syntaxiques dans le français parlé, Constructions verbales et production de sens, Jan 2006, Besançon, France. Presses Universitaires de Franche-Comté, 2011, 127-139. SUSO LOPEZ, J., La grammaire et les descriptions de la langue : la réflexion sur le fonctionnement de la langue favorise-t-elle l’apprentissage du FLE ?, Phonétique, Lexique, grammaire et enseignement-apprentissage du FLE, Granada, MÉTODO ediciones, 2004, 218-258. 91 Sandrine Caddéo Une grammaire des usages du français parlé pour non spécialistes : difficultés et perspectives RIEGEL, M., PELLAT, J.-C., RIOUL, R., Grammaire méthodique du français, Paris, Presses universitaires de France, 1994. RULLIER-THEURET, F., L’oral et l’écrit, Louvain-la-Neuve, De Boeck-Duculot, coll. Grevisse langue française, 2010. TOMASSONE, R., Pour enseigner la grammaire, Paris, Delagrave, 2002. WEBER C., Pour une didactique de l’oralité. Enseigner le français tel qu’il est parlé, Paris, Didier, 2013. 92
https://openalex.org/W2795829575	https://www.intechopen.com/citation-pdf-url/58590	English	null	Opuntia ficus-indica (Nopal Extract) as Green Inhibitor for Corrosion Protection in Industrial Steels	InTech eBooks	2,018	cc-by	6,871	Selection of our books indexed in the Book Citation Index in Web of Science™ Core Collection (BKCI) Interested in publishing with us? Contact book.department@intechopen.com Numbers displayed above are based on latest data collected. For more information visit www.intechopen.com Open access books available Countries delivered to Contributors from top 500 universities International authors and editors Our authors are among the most cited scientists Downloads We are IntechOpen, the world’s leading publisher of Open Access books Built by scientists, for scientists 14% 191,000 210M TOP 1% 154 7,200 Chapter 7 Abstract Soluble extract from Opuntia ficus-indica (Nopal extract) has been proposed in this chap- ter as a green inhibitor due to its component called mucilage, which has the ability to retain water; for this reason, it has been used as metal corrosion protection in machinery pieces, tools and other metallic components that need to be stored for short periods. In this way, three industrial carbon steels (AISI 1018, 1045 and 4140) have been exposed in sulfuric acid (H2SO4) to evaluate the corrosion behavior with or without Nopal extract (NE). Some electrochemical techniques have been implemented to evaluate the corrosion inhibition efficiency (IE) such as DC linear polarization resistance (LPR) and AC elec- trochemical impedance spectroscopy (EIS). Results indicated a considerable superficial modification of steel in terms of dielectric constant and ion charge capacity. When the NE was added, the corrosion mechanism changed from localized to general attack, decreas- ing the corrosion rate in all cases. More susceptibility to fail by corrosion was observed in the 1045 carbon steel in comparison with the other two studied steels; these results were confirmed by the percentage of inhibitor’s efficiency of about 95%. Keywords: green inhibitor, corrosion protection, Nopal extract, industrial carbon steel, corrosion rate, electrochemical evaluation, electrochemical impedance spectroscopy Additional information is available at the end of the chapter Additional information is available at the end of the chapter Additional information is available at the end of the chapter http://dx.doi.org/10.5772/intechopen.72944 http://dx.doi.org/10.5772/intechopen.72944 Opuntia ficus-indica (Nopal Extract) as Green Inhibitor for Corrosion Protection in Industrial Steels Araceli Mandujano Ruiz, Luis-Enrique Corona Almazán, Héctor Herrera Hernández and Jorge Morales Hernández Luis-Enrique Corona Almazán, Héctor Herrera Hernández and Jorge Morales Hernández Additional information is available at the en 1. Introduction Study of corrosion inhibitors has taken a new role in recent years, since new regulations restrict the use of several components in conventional corrosion inhibitors like nitrites, benzoates, © 2016 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2018 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. ( ) p Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2018 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, and reproduction in any medium, provided the original work is properly cited. Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Corrosion Inhibitors, Principles and Recent Applications 146 chrome, lead, arsenic, among others, because they are toxic and dangerous to health and environ- ment. Formulations of alternative compounds with more biocompatibility and low costs have led to the development of the green inhibitors also called eco-friendly inhibitors [1, 2]. Main sources from which these inhibitors are extracted come from plants, fruits, seeds, leaves and flowers whose components such as flavonoids, alkaloids and natural oils (pennyroyal oil, jojoba oil, etc.) are considered the inhibitor’s active agent. Those extracts have been studied in low concentration and exposed to different aggressive media to protect metals (mainly carbon steels), obtaining good results at the superficial protection against corrosion during the first hours of exposure. Figure 1 shows a general classification of the corrosion inhibitors according to its origin. Organic inhibitors are characterized by their high molecular weight structures and are polar molecules. Most organic inhibitors are adsorbed on the metal surface by displac- ing water molecules and forming a compact barrier. Inorganic inhibitors are salts of some metals, which have a passivation effect and reaction with the metal. Some synthetic com- pounds can reduce corrosion damage in carbon steels [3–5] because their oxygen, nitrogen, and sulfur heteroatoms react on the metal surface, blocking active sites where corrosion occurs. Green corrosion inhibitors are biodegradable and do not contain heavy metals or toxic com- pounds. Mechanisms of the green inhibitors indicate that ions/molecules are adsorbed onto metal surface, interfering with the anodic and/or cathodic reactions and decreasing the diffu- sion rate for reactants to the metal surface. Usually, the electrical resistance of the metal surface is decreasing. Since their innovation, some researchers have proposed many plants as pros- pects to be green corrosion inhibitors. Table 1 shows a review of several papers about green inhibitors collected from 2004 to 2017, where the carbon steels in acid media (HCl and H2SO4) Figure 1. General classification of corrosion inhibitors. Figure 1. General classification of corrosion inhibitors. Opuntia ficus-indica (Nopal Extract) as Green Inhibitor for Corrosion Protection in Industrial Steels http://dx.doi.org/10.5772/intechopen.72944 1 Opuntia ficus-indica (Nopal Extract) as Green Inhibitor for Corrosion Protection in Industrial Steels http://dx.doi.org/10.5772/intechopen.72944 147 Inhibitor Metal Medium IE (%) Year References Zanthoxylum armatum Mild steel H3PO4 > 90 2004 [8] Simmondsia chinensis Carbon steel HCl 98 2004 [9] Artemisia vulgaris Carbon steel HCl 85 2004 [10] Datura metel Carbon steel HCl and H2SO4 84~97 2005 [11] Mentha pulegium Carbon steel HCl 80 2006 [12] Ammi visnaga SX316 HCl 99.3 2006 [13] Carica papaya Mild steel H2SO4 94 2007 [14] Gongronema latifolium Aluminum HCl, NaOH > 90 2007 [15] Opuntia ficus-indica (Cactus) Aluminum HCl 94 2007, 2012 [16, 17] Phyllanthus amarus Mild steel HCl and H2SO4 > 90 2008 [18] Black pepper Carbon steel HCl and H2SO4 98 2008 [19] Zallouh root Carbon steel H2SO4 90 2008 [20] Justicia gendarussa Mild steel HCl 93 2009 [21] Clove oil Nickel, inconel 600, 690 HCl 88~92 2009 [22] Lupinus albus Carbon steel HCl and H2SO4 86 2009 [23] Gossypium hirsutum Aluminum NaOH 97 2009 [24] Ananas sativum Aluminum HCl 96 2010 [25] Murraya koenigii Mild steel HCl ~ 94 2011 [26] Marine Alga Caulerpa racemosa Mild steel H2SO4 ~85 2012 [27] Spirulina Platensis Mild steel HCl and H2SO4 ~80 2012 [28] Aloe vera Zinc, Galvanized iron HCl, H2SO4 88.5 2014 [29] Tagetes erecta Mild steel H2SO4 98.07 2014 [30] Watermelon rind Mild steel HCl, H2SO4 ~86 2014 [31, 32] Mangifera indica (Mango) and Orange Mild steel, Carbon steel H2SO4, HCl 97 2014 [33] Anise Carbon steel HCl 94 2014 [34] Coconut coir Aluminum, Mild Steel HCl and H2SO4 80 2012,2014 [35–37] Musa paradisiaca (banana) Mild steel, SS304 H2SO4, NaCl 82.7 2011,2014 [38, 39] Corrosion Inhibitors, Principles and Recent Applications 148 Inhibitor Metal Medium IE (%) Year References Ficus hispana Mild steel HCl 90 2015 [40] Nicotiana tabacum Mild steel H2SO4 94 2015 [41] Litchi chinensis Mild steel H2SO4 97.8 2015 [42] Capsicum annuum, Citrus aurantium, Moringa oleifera Copper HNO3 60~80 2015 [43] Valoniopsis pachynema Brass H3PO4 96.06 2015 [44] Azadirachta indica Al, Mild steel, Tin, SS, Carbon steel HCl, H2SO4, HNO3 > 85 2015 [45] Petroselinum crispum (Parsley), Eruca sativa (Arugula) Anethum graveolens (Dill) Carbon steel HCl 92 2015 [46] Morinda citrifolia AISI 1045, AISI 8620 HCl > 86 2015,2016 [47, 48] Pomegranate Carbon steel, Mild steel, α-Brass HCl > 90 2015,2013,2017 [49–51] Gentiana olivieri Mild steel HCl 92 2016 [52] Roasted coffee Carbon steel HCl 94 2016 [53] Curcumin Aluminum NaCl 80 2016 [54] Calotropis procera leaves Mild steel NaCl > 90 2016 [55] Ocimum basilicum Carbon steel, Mild steel NaCl + Na2S, H2SO4 88.5 2016,2012 [56, 57] Tilia Mild steel HCl 79 2017 [58] Cedrus atlantica Lead Na2CO3 ~ 70 2017 [59] Table 1. Review of several papers related with green inhibitors from 2004 to 2017. Table 1. Review of several papers related with green inhibitors from 2004 to 2017. were the most studied, possibly because the total annual cost of metallic corrosion in a country is associated mainly with the corrosion of carbon steels of different industrial sectors. Most of these works were focused on demonstrating the efficiency of these extracts as well as their behavior with respect to temperature, identifying the absorption mechanism from the extract components [6, 7], however, other factors like cost and amount are not mentioned in detail. Evaluation of corrosion inhibition efficiency was based on the implementation of electro- chemical and gravimetric methods; the electrochemical techniques consist of linear polar- ization resistance (LPR) and electrochemical impedance spectroscopy (EIS). With LPR was obtained information about the inhibition reaction with the compound, classifying the inhibi- tor action as anodic, cathodic or mixed. EIS allowed to detect the resistance changes on the surface and how the active sites are reordered through changes at the interfacial capacitance of metal-electrolyte. Opuntia ficus-indica (Nopal Extract) as Green Inhibitor for Corrosion Protection in Industrial Steels http://dx.doi.org/10.5772/intechopen.72944 1 Opuntia ficus-indica (Nopal Extract) as Green Inhibitor for Corrosion Protection in Industrial Steels http://dx.doi.org/10.5772/intechopen.72944 149 149 Corrosion inhibition efficiency can be calculated through the coverage degree “θ” on the metal surface by the following equations: Corrosion inhibition efficiency can be calculated through the coverage degree “θ” on the metal surface by the following equations: θ = 1 ___ R ct (blank) − 1 ___ R ct (inhibitor) __________________ 1 ___ R ct (blank) (1) θ = i corr (blank) − i corr (inhibitor) ______________ i corr (blank) (2) θ = V corr (blank) − V corr (inhibitor) _______________ V corr (blank) (3) 𝜃 = Weight loss in blank − Weight loss in inhibitor ___________________________________ Weight loss in blank (4) θ = 1 ___ R ct (blank) − 1 ___ R ct (inhibitor) __________________ 1 ___ R ct (blank) (1) θ = i corr (blank) − i corr (inhibitor) ______________ i corr (blank) (2) θ = V corr (blank) − V corr (inhibitor) _______________ V corr (blank) (3) 𝜃 = Weight loss in blank − Weight loss in inhibitor ___________________________________ Weight loss in blank (4) (1) (2) (3) 𝜃 = Weight loss in blank − Weight loss in inhibitor ___________________________________ Weight loss in blank (4) (4) where Rct is the charge transfer resistance, icorr is the corrosion current and Vcorr is the corrosion rate. Expressions (1), (2) and (3) are obtained from electrochemical techniques like LPR and EIS, and Eq. (4) is calculated using the gravimetric method. With any of those equations, the efficiency is obtained with: where Rct is the charge transfer resistance, icorr is the corrosion current and Vcorr is the corrosion rate. Expressions (1), (2) and (3) are obtained from electrochemical techniques like LPR and EIS, and Eq. (4) is calculated using the gravimetric method. With any of those equations, the efficiency is obtained with: EI (%) = θ × 100 (5) (5) 1.1. Opuntia ficus plant One plant that has caused interest about their potential as a green inhibitor is Opuntia ficus- indica (Nopal), which is a traditional vegetable in Mexico with a high soluble fiber content. Nopal is classified as a wild plant that survives in desert regions and does not require much water for cultivation. It is said that it has an important ecological performance, as it stops the degradation of the deforested soil, making the arid land productive because of its capacity to retain water. There are about 1600 species in 122 genera of the cactus family, from which comes the Nopal. It has fruits that are edible and are known with the name of Tunas. The soluble fiber of Nopal known as mucilage, which has the capacity to retain up to 30 times its weight in water. For this particular compound, it has been promoted as a good candidate for the production of an ecological inhibitor for the protection of the steel against corrosion [64, 65]. Chemical analysis of mucilage indicates that it contains a large amount of pectin as well as minerals such as calcium, potassium and sodium whose quantities depend on the age of the cladodes (leaves). The presence of these minerals suggests that this plant may remain stable over long periods in alkaline media. Mucilage is similar in consistency with Aloe vera (slimy) and has been used for the conserva- tion of building materials. Chandra [66] and Torres Acosta [17] studied the effects of dehy- drated Nopal and Aloe vera powder in the electrochemical performance of reinforcing steel in chloride contaminated concrete. Preliminary results suggest that adding these powders in small concentrations might be suitable for enhancing corrosion resistance of steel in concrete. Therefore, this investigation deals with the performance of Nopal mucilage as a corrosion inhibitor for industrial steels exposed in H2SO4. In general, a good inhibitor must have an efficiency greater than 90%. In general, a good inhibitor must have an efficiency greater than 90%. Efficiency of an organic compound as a corrosion inhibitor depends mainly on the anchor capacity by chemical or physical interactions on the metal surface, forming multilayers of adsorbed organic molecules that block the active sites of corrosion and retarding the anodic and/or cathodic reactions. Thermodynamic studies with adsorption isotherms (Temkin, Frumkin, Langmuir, Freundlich, among others) have allowed to estimate the stability of the layer adsorbed on the surface. These depend on the physicochemical properties of the active molecules related to the inhibitor functional groups like aromatic rings and amino acids with –NH2 groups, which are readily protonated in acid media and adsorbed on the metal surface through acid anions (Cl−, SO 4 2− ) negatively charged. Electron density from the donor atoms, and the interaction between the orbital π from inhibitory and the orbital d of the iron, play an important role in the adsorption type [60–62]. Investment in the use of inhibitors for industrial scale to prevent corrosion represents a small percentage (1%) worldwide compared with the costs in the use of paints, surface treatments and materials selection, which represent 80% of the investment. The main sec- tors investing in inhibitors such as petrochemicals (production, synthesis, and refining), water treatment systems, and chemical and food processes, have reported an efficiency of up to 90% in the mitigation of corrosion, saving billions of dollars. However, despite its effectiveness, many of the commercial synthetic inhibitors that are currently used gener- ate problems of toxicity and contamination to the environment. Environmental Protection Agency (EPA) regulations have determined that inorganic inhibitor compounds such as Corrosion Inhibitors, Principles and Recent Applications 150 salts of chromates, phosphates and molybdates, as well as organic inhibitors like phos- phates, amides or thiols, are dangerous. Due to this problem, green inhibitors open an opportunity for the replacement of these compounds (Table 1) where the synthesis process of natural extracts is of low cost and can be applied in closed systems as pipelines and packaging of pieces for transport. Nowadays, the limitation in the industrial staggering of these green inhibitors is in their preservation, since it has been shown that after 24 h, the natural extract begins to decompose which would imply the use of large volumes of extract to serve the industry [63]. 2.1. Specimens preparation Three types of carbon steels were evaluated in this research: AISI 1045, 1018 and 4140. The composition of each metal is shown in Table 2. Rods of each steel were cut mechanically to obtain specimens of 25 mm of diameter and 5 mm of thickness for the electrochemical evalu- ations. The surface preparation was grinding with SiC paper through numbers 80, 240, 320, 400, 600 and 1500 grits until a mirror finish. Finally, the specimens were washed with distilled water, degreasing with acetone and drying with hot air. Opuntia ficus-indica (Nopal Extract) as Green Inhibitor for Corrosion Protection in Industrial Steels http://dx.doi.org/10.5772/intechopen.72944 151 Opuntia ficus-indica (Nopal Extract) as Green Inhibitor for Corrosion Protection in Industrial Steels http://dx.doi.org/10.5772/intechopen.72944 151 151 Element Wt. % 1045 1018 4140 C 0.45 0.2 0.4 Mn 0.7 0.9 0.85 P 0.04 0.04 0.04 S 0.05 0.05 0.05 Fe 98.76 98.81 97.31 Mo **** ** 0.2 Cr ** ** 0.9 Si ** **** 0.25 Table 2. Metal composition with AISI classification. Table 2. Metal composition with AISI classification. 2.2. Preparation of Nopal extract Nopal extract was obtained through the selection of tender cladodes, a cleaning process and removal of thorns, cut and disinfected with 2 ml of chlorine +1 ml of Cu2SO4, assuring the elimination of microorganism and fungi that degrade the extract. Subsequently, it was crushed in an extractor where the pulp was removed from the juice; this last one was heated for 1 h at 78°C with a volume ratio of 2:1 in distilled water. Finally, the concentrated liquid was filtered through a fine sieve to ensure separation of solid particles larger than 180 μm, obtaining the final product as shown in Figure 2. Figure 2. Nopal extract preparation process. Figure 2. Nopal extract preparation process. Corrosion Inhibitors, Principles and Recent Applications 152 2.3. Experimental sequence Electrochemical characterization was carried out in an acrylic cell for flat probes [65], assem- bled with an arrangement of three electrodes. Steels 1018, 1045 and 4140 were placed as work- ing electrode (WE), a grid of platinum was placed as a counter electrode (CE) and a calomelane electrode was used as reference electrode (RE). Solution of H2SO4 (0.6 mol·l−1) was used as the test electrolyte, prepared with analytical grade reagents and deionized water. All experiments were performed under standard conditions of temperature and pressure. Figure 3 shows the experimental sequence where the steels in H2SO4 with and without Nopal extract (NE) were evaluated. The relation of Nopal extract and H2SO4 ratio was 50:50 volume %. Open circuit potential (OCP) was monitored to ensure the system stability during 1 h, followed by the electrochemical impedance spectroscopy (EIS) technique, applying a sweep of frequencies from 100 kHz to 10 mHz with 10 mV amplitude. OCP measurement was run again for 1 h to continue with linear polarization resistance (LPR) with an overpotential range of ±25 mV and a sweep speed of 1 mV·s−1. Cycle OCP-EIS-OCP-LPR was repeated for 24 h until record the final behavior with a polarization curve (CP) applying an overpotential of ±1 V at a scan rate of 1 mV·s−1. Corrosion rates, capacitances and resistances for all steels were calculated. Figure 3. Experimental sequence. Figure 3. Experimental sequence. Figure 3. Experimental sequence. Figure 3. Experimental sequence. 2.3. Experimental sequence Electrochemical characterization was carried out in an acrylic cell for flat probes [65], assem- bled with an arrangement of three electrodes. Steels 1018, 1045 and 4140 were placed as work- ing electrode (WE), a grid of platinum was placed as a counter electrode (CE) and a calomelane electrode was used as reference electrode (RE). Solution of H2SO4 (0.6 mol·l−1) was used as the test electrolyte, prepared with analytical grade reagents and deionized water. All experiments were performed under standard conditions of temperature and pressure. Figure 3 shows the experimental sequence where the steels in H2SO4 with and without Nopal extract (NE) were evaluated. The relation of Nopal extract and H2SO4 ratio was 50:50 volume %. Open circuit potential (OCP) was monitored to ensure the system stability during 1 h, followed by the electrochemical impedance spectroscopy (EIS) technique, applying a sweep of frequencies from 100 kHz to 10 mHz with 10 mV amplitude. OCP measurement was run again for 1 h to continue with linear polarization resistance (LPR) with an overpotential range of ±25 mV and a sweep speed of 1 mV·s−1. Cycle OCP-EIS-OCP-LPR was repeated for 24 h until record the final behavior with a polarization curve (CP) applying an overpotential of ±1 V at a scan rate of 1 mV·s−1. Corrosion rates, capacitances and resistances for all steels were calculated. 3. Results and discussion Corrosion potential results are shown in Figure 4, where only the steels exposed in acid medium reported positive potentials with a tendency to move toward more negative (active) potentials when the inhibitor is present, indicating a change in the surface activity of the metal. Figure 5 shows the behavior of the resistance between the metal in the acid medium and the medium with inhibitor (H2SO4 + NE). A considerable growth of the Nyquist semicircle was observed with the addition of NE for the three steels, being the widest semicircle for the 4140 steel, which reaches a value of charge transfer resistance (Rct) of 1400 ohms at the beginning Opuntia ficus-indica (Nopal Extract) as Green Inhibitor for Corrosion Protection in Industrial Steels http://dx.doi.org/10.5772/intechopen.72944 153 Opuntia ficus-indica (Nopal Extract) as Green Inhibitor for Corrosion Protection in Industrial Steels http://dx.doi.org/10.5772/intechopen.72944 153 Figure 4. Potential corrosion behavior of metals for 24 h. Figure 5. Nyquist plots corresponding to the three steels in H2SO4 and H2SO4 + Nopal extract. http://dx.doi.org/10.5772/intechopen.72944 Figure 4. Potential corrosion behavior of metals for 24 h. Figure 5. Nyquist plots corresponding to the three steels in H2SO4 and H2SO4 + Nopal extract. Corrosion Inhibitors, Principles and Recent Applications 154 Figure 6. EEC and fit of Nyquist plots corresponding to metals in H2SO4 and H2SO4 + Nopal extract. gure 6. EEC and fit of Nyquist plots corresponding to metals in H2SO4 and H2SO4 + Nopal extract. Figure 6. EEC and fit of Nyquist plots corresponding to metals in H2SO4 and H2SO4 + Nopal extrac of the evaluation. As the exposure time increases, the semicircle shows a tendency to decrease slightly as a result of the alteration of the protection film formed by the inhibitor. At the end of the monitoring, the Rct values from the inhibitor films do not approach the Rct values of the blank, conserving some protection capacity with respect to the metal without inhibitor. Semicircle adjustment was performed for each impedance cycles by taking an equivalent elec- trical circuit (EEC) as shown in Figure 6. Due to the shape of the Nyquist semicircle, the basic configuration called EEC Randles model was selected, which is formed by a resistance in series (solution resistance, Rs) coupled to a constant phase element (CPE) that is in parallel with another resistance attributed to the metal/electrolyte interface (Rct). 3. Results and discussion To obtain the capacitance values (Cdl) from the constant phase element (CPE), it was necessary to obtain the maximum frequency of the Nyquist semicircle (ωθ max) as well as the α exponent that was calculated during the CPE adjustment, since CPE = F٠s(a−1) where F and s indicate Faradios and seconds, respectively. Results in Table 3 are shown in μF cm−2. Equation (6) shows the calculation to Cdl: C dl = CPE ∗ ω θmax (a−1) (6) (6) C dl = CPE ∗ ω θmax (a−1) Opuntia ficus-indica (Nopal Extract) as Green Inhibitor for Corrosion Protection in Industrial Steels http://dx.doi.org/10.5772/intechopen.72944 15 155 Metal Medium LPR EIS Jo (A/cm2) Ecorr vs. SCE (V) Rp (Ohm ·cm2) Rct (Ohm ·cm2) Cdl (μF·cm2) 1018 H2SO4 2.92E-05 −0.511 268.182 187.538 179.36 H2SO4 + NE 7.85E-06 −0.545 1015.013 717.162 155.96 4140 H2SO4 4.74E-05 −0.506 179.379 121.76 1202.42 H2SO4 + NE 7.71E-06 −0.543 1085.28 815.98 185.14 1045 H2SO4 3.34E-04 −0.439 64.865 50.78 492.49 H2SO4 + NE 1.15E-05 −0.505 729.44 514.16 120.41 Table 3. Electrochemical results. Table 3. Electrochemical results. Table 3. Electrochemical results. Polarization curves of carbon steel samples exposed in H2SO4 with inhibitor are shown in Figure 7(A), where a decrease in current corrosion of the three steels studied is observed. Anodic slope shows a change near the 400 mV, associated with a change of their activation by the presence of a coating from the inhibitor’s organic molecules, limiting the ionic interchange between the metal and electrolyte. Figure 7(A) shows the behavior of the polarization in which there is a decrease in the flow of current on the three steels exposed to the inhibitor. Figure 7(B) shows the corrosion behavior for all the steels; these values were obtained by extrapolation of Tafel, where the steels show a lower corrosion rate to 1018, 4140 and 1045, respectively; these values remain stable through- out the 24 h of exposure. Table 3 shows the values obtained with the electrochemical techniques LPR and EIS; a cor- respondence was observed between the values of resistance for both techniques; however, Figure 7. (A) Polarization curve for steels in H2SO4 and H2SO4 + NE medium and (B) corrosion rate of metals for 24 h. Figure 7. (A) Polarization curve for steels in H2SO4 and H2SO4 + NE medium and (B) corrosion rate of metals for 24 h. 3. Results and discussion Corrosion Inhibitors, Principles and Recent Applications 156 the resistance values by the impedance technique are lower because they are depurated due to the contribution of the resistance to the solution that has been separated with the adjust- ment of the EEC. The current indicates that the 1018 and 4140 steels have low corrosion rate. Capacitance values indicate a decrease with the presence of the inhibitor and are lower for 1045 and 1018 steels; these values tend to decrease due to a difficulty of ionic exchange. Figure 8 shows the metallography obtained for all steels on a macroscopic scale, where for (a)–(c), the area exposed to H2SO4 shows that the main mechanism for the three metals is a combination between pitting and generalized corrosion, being more notable the pitting attack for the 1018 and 1045 steels before the NE addition. Figure 8. Metallographic obtained of metals after exposition in H2SO4 and H2SO4 + Nopal extract. Figure 8. Metallographic obtained of metals after exposition in H2SO4 and H2SO4 + Nopal extract. Opuntia ficus-indica (Nopal Extract) as Green Inhibitor for Corrosion Protection in Industrial Steels http://dx.doi.org/10.5772/intechopen.72944 157 Opuntia ficus-indica (Nopal Extract) as Green Inhibitor for Corrosion Protection in Industrial Steels http://dx.doi.org/10.5772/intechopen.72944 157 Figure 9. Inhibition efficiency for the three metals in H2SO4 + Nopal extract after 24 h. Opuntia ficus-indica (Nopal Extract) as Green Inhibitor for Corrosion Protection in Industrial Steels http://dx.doi.org/10.5772/intechopen.72944 Figure 9. Inhibition efficiency for the three metals in H2SO4 + Nopal extract after 24 h. Figure 8(d)–(f) shows the behavior of the steels after exposed to the acid mixture plus the Nopal extract. Change of the corrosion mechanism was observed with the addition of the NE from a localized attack (pitting) to a generalized attack due to the redistribution of the active sites by the presence of the organic molecules from the inhibitor, having better behavior in the steel 1045 where the corrosion was considerably diminished. Figure 9 shows the efficiency calculated for the three steels, observing the performance of the Nopal extract during the 24 h of exposure. It was observed that steel 1045 obtains the highest percentage of efficiency, which is associated with the inhibitor coverage, and remains con- stant from the 10th hour of exposure, while the other two steels (1018 and 4140) do not achieve sufficient efficiency values to guarantee their protection. 4. Conclusions Opuntia ficus-indica (Nopal extract) was used as a green inhibitor to evaluate the corrosion behavior in the protection of three industrial steels (AISI 1018, 1045 and 4140), exposed in sul- furic acid (H2SO4) plus the addition of NE in the same proportion (50/50 volume %). Corrosion potential showed a tendency to be more negative when the NE was added, forming a compact barrier on the metal surface that increased the surface resistance, according to the growth of the Nyquist semicircles and with respect to the steel substrates. Charge transfer resistance (Rct) was reduced as the exposition time increased due to the dete- rioration of the protective film. A 4140 steel reported less corrosion rate with respect to the 1018 and 1045 steels, respectively. Polarization resistance (Rp) showed the same behavior with the best behavior for the steel 4140 after 24 h. Corrosion Inhibitors, Principles and Recent Applications 158 Change of the corrosion mechanism was observed with the addition of the NE from a local- ized attack (pitting) to a generalized attack due to the redistribution of the active sites by the presence of the organic molecules from the inhibitor, having better behavior in the steel 1045 where the corrosion was considerably diminished. Change of the corrosion mechanism was observed with the addition of the NE from a local- ized attack (pitting) to a generalized attack due to the redistribution of the active sites by the presence of the organic molecules from the inhibitor, having better behavior in the steel 1045 where the corrosion was considerably diminished. The NE addition changed the corrosion mechanism in the steels from a localized attack (pit- ting) to a generalized attack due to the adsorbed inhibitor molecules on the metal surface, reducing the anodic and/or cathodic reactions and as a consequence the corrosion rate. Alloyed steel (4140) showed a decrease in the inhibitor efficiency with the exposition time, showing better tendency (above 90%) for the 1045 steel. This is possible because the chemi- cal composition of the steel is interacting with the adsorbed organic molecules, increasing or reducing the actives sites for corrosion. Difference in corrosion rate for the 4140 steel with the others was the combination of the green inhibitor and their alloyed elements (Mo and Cr). Acknowledgements The authors wish to thank our institution (CIDETEQ) and (UAEM) for the facilities and the National Council of Science and Technology (CONACYT) for financial support. We appreci- ate the participation of the student Luis E. Almazán Corona, who actively participated in the experimental development through the program UAEM-SIyEA (Secretary of Investigation and Advance Studies) 3817/2014/CID. Author details Araceli Mandujano Ruiz1, Luis-Enrique Corona Almazán2, Héctor Herrera Hernández2 and Jorge Morales Hernández1 Address all correspondence to: amandujano@cideteq.mx 1 Centro de Investigación y Desarrollo Tecnológico en Electroquímica (CIDETEQ), Querétaro, México 2 Universidad Autónoma del Estado de México, Atizapán de Zaragoza, Estado de México, México 2 Universidad Autónoma del Estado de México, Atizapán de Zaragoza, Estado de México, México [1] Lim HL. Deeper understanding of green inhibitors for corrosion of reinforcing steel in concrete. In: Handbook of Research on Research Development in Materials Science and Corrosion Engineering Education, E.S. Reference: EUA. 2015 References [1] Lim HL. Deeper understanding of green inhibitors for corrosion of reinforcing steel in concrete. In: Handbook of Research on Research Development in Materials Science and Corrosion Engineering Education, E.S. Reference: EUA. 2015 Opuntia ficus-indica (Nopal Extract) as Green Inhibitor for Corrosion Protection in Industrial Steels http://dx.doi.org/10.5772/intechopen.72944 159 159 [2] Bockris JO, Singh WP. Toxicity issues of organic corrosion inhibitors; applications of QSAR Model. In: Proceedings of Corrosion. Houston: NACE, International; 1996 [3] Espinoza VA, Negrón SGE, González OR, Angeles BD, Herrera HH, Romero RM. Mild steel corrosion inhibithion in HCl by di-alkyl and di-1,2,3, triazole derivatives of uracil and thyamine. Materials Chemistry and Physics. 2014:407-417 [4] Espinoza VA, Negrón SGE, Angeles BD, Palomar PME, Romero RMA, Herrera HH. Pantoprazol as inhibitors of API 5L X52 steel corrosion in HCl 1M. ECS Transactions. 2011:207-2016 [5] Palomar PM, Romero RM, Herrera HH, Abfeu QMA, Likhanova N, Uruchurtu J, et. al. Influence of the alkyl chain length of 2 amino 5 alkyl 1,3,4 thidiazole compounds on the corrosion inhibithion of steel immersed in sulfuric acid solutions. Corrosion Science. 2012:231-243 [6] Majdouline L et al. An investigation of mild carbon steel corrosion inhibition in hydro- chloric acid medium by environment friendly green inhibitors. Journal of Coatings Technology and Research. 2011;9:15-26 [7] Ibrahim TH, Zour MA. Corrosion inhibition of mild steel using fig leaves extract in hydro- chloric acid solution. International Journal of Electrochemical Science. 2011;6:6442-6455 [8] Gunasekaran G, Chauchan LR. Eco friendly inhibitor for corrosion inhibition of mild steel in phosphoric acid medium. Electrochimica Acta. 2004;49:4387-4395 [9] Chetouani A, Hammouti B, Benkaddour M. Corrosion inhibition of iron in hydrochloric acid solution by jojoba oil. Pigment & Resin Technology. 2004;33:26-31 [10] Bouyanzer A, Hammouti B, Chetouani A. A study of anti-corrosive effects of Artemisia oil on steel. Pigment & Resin Technology. 2004;33:287-292 [11] Sethuraman MG, Bothi RP. Corrosion inhibition of mild steel by Datura metel. Pigment & Resin Technology. 2005;34:327-331 [12] Bouyanzer A, Hammouti B, Majidi L. Pennyroayl oil from Mentha pulegium as corrosion inhibitor for steel in 1 M HCl. Materials Letters. 2006;60:2840-2843 [13] El-Etre AY. Khilah extract as inhibitor for acid corrosion of SX 316 steel. Applied Surface Science. 2006;252:8521-8525 [14] Okafor PC, Ebenso EE. Inhibitive action of Carica papaya extracts on the corrosion of mild steel in acidic media and their adsorption characteristics. Pigment & Resin Tech- nology. 2007;36:134-140 [15] Oguzie EE, Onuoha GN, Ejike EE. References Effect of Gongronema latifolium extract on alumin- ium corrosion in acidic and alkaline media. Pigment & Resin Technology. 2007;36:44-49 [16] Torres Acosta AA. Opuntia-ficus-indica (Nopal) mucilage as a steel corrosion inhibitor in alkaline media. Journal Application of Electrochemical. 2007;38:835-841 Corrosion Inhibitors, Principles and Recent Applications 160 [17] Torres-Acosta AA, Molina MW, Guzmán AEM. State of the art on cactus additions in alkaline media as corrosion inhibitors. International Journal of Corrosion. 2012:1-9 [18] Okafor PC et al. Inhibitory action of Phyllanthus amarus extracts on the corrosion of mild steel in acidic media. Corrosion Science. 2008;50:2310-2317 [19] Raja PB, Sethuraman MG. Inhibitive effect of black pepper extract on the sulphuric acid corrosion of mild steel. Materials Letters. 2008;62:2977-2979 [20] El- Etre AY. Inhibition of C-steel corrosion in acidic solution using the aqueous extract of zallouh root. Materials Chemistry and Physics. 2008;108:278-282 [21] Satapathy AK et al. Corrosion inhibition by Justicia gendarussa plant extract in hydro- chloric acid solution. Corrosion Science. 2009;51:2848-2856 [22] Abdullah M, Al-Karanee SO, Abdel FAA. Inhibition of the corrosion of nickel and its alloys by Natural Clove Oil. Chemical Engineering Communications. 2009;196:1406-1416 [23] Abdel- Gaber AM, Adb- El NBA. The role of acid anion on the inhibition of the acidic corrosion of steel by lupine extract. Corrosion Science. 2009;51:1038-1042 [24] Abiola OK, Otaigbe JOE, Kio OJ. Gossipium hirsutum L. extracts as green corrosion inhib- itor for aluminum in NaOH solution. Corrosion Science. 2009;51:1879-1881 [25] Atinga EI et al. Leaves extract of Ananas sativum as green corrosion inhibitor for alumin- ium in hydrochloric acid solutions. Green Chemistry Letters and Reviews. 2010;3:61-68 [26] Beenakumari KS. Inhibitory effects of Murraya koenigii (curry leaf) leaf extract on the corrosion of mild steel in 1 M HCl. Green Chemistry Letters and Reviews. 2011;4:117-120 [27] Chennappan K, Mathur GS. Caulerpin—A bis-indole alkaloid as a green inhibitor for the corrosion of mild steel in 1 M HCl solution from the marine alga Caulerpa racemosa. Industrial and Engineering Chemistry Research. 2012:10399-10407. DOI: 10.1021/ie3010379 [28] Kamal C, Sethuraman MG. Spirulina platensis—A novel green inhibitor for acid corro- sion of mild steel. Arabian Journal of Chemistry. 2012;5:155-161 [29] Shah P, Agarwal S. Aloe-Vera: A green corrosion inhibitor. International Journal for Research in Applied Science and Engineering Technology. 2014;2:14-17 [30] Mourya P, Banerjee S, Singh MM. Corrosion inhibition of mild steel in acidic solution by Tagetes erecta (Marigold flower) extract as a green inhibitor. Corrosion Science. References 2014 [31] Odewunmi NA, Umoren SA, Gasem ZM. Utilization of watermelon rind extract as a green corrosion inhibitor for mild steel in acidic media. Journal of Industrial and Engi- neering Chemistry. 2014:1-9 [32] Odewunmi NA, Umoren SA, Gasem ZM. Watermelon waste products as green cor- rosion inhibitors for mild steel 2 in HCl solution. Journal of Environmental Chemical Engineering. 2014:1-11 Opuntia ficus-indica (Nopal Extract) as Green Inhibitor for Corrosion Protection in Industrial Steels http://dx.doi.org/10.5772/intechopen.72944 161 161 [33] Da Rocha CJ, Da Cunha PG, D’Eliab E. Aqueous extracts of mango and orange peel as green inhibitors for carbon steel in hydrochloric acid solution. Materials Research. 2014;17:1581-1587 [34] Fouda AS et al. Anise extract as green corrosion inhibitor for carbon steel in hydrochloric acid solutions. International Journal of Innovative Research in Science Engineering and Technology. 2014;3:11210-11228 [35] Umoren SA et al. Coconut coir dust extract: A novel eco-friendly corrosion inhibitor for Al in HCl solutions. Green Chemistry Letters and Reviews. 2012;5:303-313 [36] Umoren SA, Solomon MM, Eduok UM, Obot IB, Israel AU. Inhibition of mild steel cor- rosion in H2SO4 solution by coconut coir dust extract obtained from different solvent systems and synergistic effect of iodide ions: Ethanol and acetone extracts. Journal of Environmental Chemical Engineering. 2014. DOI: 10.1016/j.jece.2014.03.024 [37] Umoren SA. et al. Inhibition of mild steel corrosion in H2SO4 solution by coconut coir dust extract obtained from different solvent systems and synergistic effect of iodide ions: Ethanol and acetone extracts. Journal of Environmental Chemical Engineering. 2014:1-55 [38] Mayanglambam RS, Sharma V, Singh G. Musa Paradisiaca extract as a green inhibitor for corrosion of mild steel in 0.5 M sulphuric acid solution. Portugaliae Electrochimica Acta. 2011;29:405-417 [39] Ismail A, Tajuddin MAM. Banana peel as green corrosion inhibitor for stainless steel 304. International Conference on X-Rays & Related Techniques in Research & Industry. 2014:109-110 [40] Muthukrishnan P et al. Stigmasterol extracted from Ficus hispida leaves as a green inhib- itor for the mild steel corrosion in 1 M HCl solution. Arabian Journal of Chemistry. 2015:1-12 [41] Bhawsar J, Jain PK, Jain P. Experimental and computational studies of Nicotiana tabacum leaves extract as green corrosion inhibitor for mild steel in acidic medium. Alexandria Engineering Journal. 2015;54:769-775 [42] Singh MR, Gupta P, Gupta K. The litchi (Litchi chinensis) peels extract as a potential green inhibitor in prevention of corrosion of mild steel in 0.5 M H2SO4 solution. Arabian Journal of Chemistry. References 2015:1-7 [43] Savita et al. Fruit extract as a green inhibitor for copper corrosion in nitric acid solution. International Journal of Innovative Research in Science Engineering and Technology. 2015;4:4545-4553 [44] Kumar RS, Chandrasekaran V. Valoniopsis pachynema extract as a green inhibitor for corrosion of brass in 0.1 N phosphoric acid solution. Metallurgical and Materials Transactions. 2015:1-8 Corrosion Inhibitors, Principles and Recent Applications 162 [45] Sharma SK, Peter A, Obot IB. Potential of Azadirachta indica as a green corrosion inhibi- tor against mild steel, aluminum, and tin: A review. Journal of Analytical Science and Technology. 2015;6:1-16 [46] Al-Senani GM. The use of green leafy vegetables extracts as corrosion inhibitors for car- bon steel in acidic medium. World Applied Sciences Journal. 2015;33(10):1659-1666 [47] Herrera HH, et al. Morinda Citrifolia como inhibidor natural de la corrosión de aceros en medio ácido. In: Congreso Internacional de Metalurgia y Materiales 2015. Concepción- Chile. 2015 [48] Herrera HH, et al. Morinda Citrifolia como Inhibidor Natural de la Corrosión de Aceros en Medio Ácido. In: Memorias del Congreso Internacional de Investigación Academica Journals Tabasco 2016; Tabasco; 2016 [49] Abboud Y et al. Corrosion inhibition of carbon steel in hydrochloric acid solution using pomegranate leave extracts. Corrosion Engineering, Science and Technology. 2015:1-9 [50] Abboud Y et al. Punica granatum leave extract as green corrosion inhibitor for mild steel in Hydrochloric acid. MATEC Web of Conferences. 2013 [51] Cheyad MS, Salman TA. Pomegranate peel extract as green corrosion inhibitor for α-brass in 2M HCl solution. Journal of Materials Science & Surface Engineering. 2017;5:597-601 [52] Baran E, Cakir A, Yazici B. Inhibitory effect of Gentiana olivieri extracts on the corrosion of mild steel in 0.5 M HCl: Electrochemical and phytochemical evaluation. Arabian Journal of Chemistry. 2016:1-17 [53] Souzaa ECDCA et al. Roasted coffee extracts as corrosion inhibitors for mild steel in HCl solution. Materials Research. 2016;19:1276-1285 [54] Juhaiman LAA. Curcumin extract as a green inhibitor of leaching from aluminum cook- ware at quasi-cooking conditions. Green and Sustainable Chemistry. 2016;6:57-70 [55] Taleb I, Elron G, Ime BO, Mustafa K, Mohamed AZ. Corrosion inhibition of mild steel by Calotropis procera leaves extract in a CO2 saturated sodium chloride solution. Journal of Adhesion Science and Technology. 2016;30(23):2523-2543 [56] SFA E-A et al. The role of aqueous Ocimum basilicum extract for monitoring the corro- sion inhibition of carbon steel used in sanitation plants and its effect on Escherichia coli. Zastita Materijala. References 2016;57:583-596 [57] Mayanglambam RS, Sharma V, Singh G. Ocimum basilicum extract as a potential green inhibitor for corrosion of mild steel in 0.5M H2SO4 solution. Asian Journal of Chemistry. 2012;24:1773-1778 [58] SFA E-A, SAA E-M, Mostafa MH. Tilia leafs as eco-friendly corrosion inhibitor for mild steel in aqueous solutions. Zastita Materijala. 2017;58:217-282 [59] Elmiziani I et al. Lead corrosion inhibition by Cedrus atlantica as a green inhibitor in 0.1M Na2CO3 solution. International Journal of Advanced Chemistry. 2017;5:1-7 Opuntia ficus-indica (Nopal Extract) as Green Inhibitor for Corrosion Protection in Industrial Steels http://dx.doi.org/10.5772/intechopen.72944 163 163 [60] Khaled KF. Experimental, density function theory calculations and molecular dynamics simulations to investigate the adsorption of some thiourea derivatives on iron surface in nitric acid solutions. Applied Surface Science. 2010;256:6753-6763 [61] Sanyal B. Organic compounds as corrosion inhibitors in different environments—A review. Progress in Organic Coatings. 1981;9:165-236 [62] Ibrahim TH, Chehade Y, Zour MA. Corrosion inhibition of mild steel using potato peel extract in 2M HCL solution. International Journal of Electrochemical Science. 2011;6: 6542-6555 [63] Sastri VS. Challenges in corrosion, costs, causes, consequences and control. Wiley series in corrosion. 2015 [64] De los Ríos JP, Sánchez CM, Nava DC, Chacón NJG, González RJG, Huape PE, Neri FA, Martínez VA. Opuntia ficus-indica extract as green corrosion inhibithor for carbon steel in 1M HCl solution. Journal of Spectroscopy. 2015:1-9 [65] Herrera HH, Franco TMI, Miranda HJG, Herrera SE, Espinoza VA, Fajardo G. Gel de Aloe-Vera como Potencial inhibidor de la corrosión del acero de Refuerzo Estructural. Avances en Ciencias e Ingeniería. 2015;6:9-23 [66] Chandra S, Eklud L, Villareal RR. Use of cactus in mortars and concrete. Cement and Concrete Research. 1998;28:41-51
https://openalex.org/W2010956923	https://europepmc.org/articles/pmc4094383?pdf=render	English	null	Hypoxia and Exercise Increase the Transpulmonary Passage of 99mTc-Labeled Albumin Particles in Humans	PloS one	2,014	cc-by	8,417	Abstract Intrapulmonary arteriovenous anastomoses (IPAVs) are large diameter connections that allow blood to bypass the lung capillaries and may provide a route for right-to-left embolus transmission. These anastomoses are recruited by exercise and catecholamines and hypoxia. Yet, whether IPAVs are recruited via direct, oxygen sensitive regulatory mechanisms or indirect effects secondary to redistribution pulmonary blood flow is unknown. Here, we hypothesized that the addition of exercise to hypoxic gas breathing, which increases cardiac output, would augment IPAVs recruitment in healthy humans. To test this hypothesis, we measured the transpulmonary passage of 99mTc-macroaggregated albumin particles (99mTc-MAA) in seven healthy volunteers, at rest and with exercise at 85% of volitional max, with normoxic (FIO2 = 0.21) and hypoxic (FIO2 = 0.10) gas breathing. We found increased 99mTc-MAA passage in both exercise conditions and resting hypoxia. However, contrary to our hypothesis, we found the greatest 99mTc-MAA passage with resting hypoxia. As an additional, secondary endpoint, we also noted that the transpulmonary passage of 99mTc-MAA was well-correlated with the alveolar-arterial oxygen difference (A-aDO2) during exercise. While increased cardiac output has been proposed as an important modulator of IPAVs recruitment, we provide evidence that the modulation of blood flow through these pathways is more complex and that increasing cardiac output does not necessarily increase IPAVs recruitment. As we discuss, our data suggest that the resistance downstream of IPAVs is an important determinant of their perfusion. Citation: Bates ML, Farrell ET, Drezdon A, Jacobson JE, Perlman SB, et al. (2014) Hypoxia and Exercise Increase the Transpulmonary Passage of 99mTc-Labeled Albumin Particles in Humans. PLoS ONE 9(7): e101146. doi:10.1371/journal.pone.0101146 Editor: Tim Lahm, Indiana University, United States of America Received February 19, 2014; Accepted June 3, 2014; Published July 11, 2014 Received February 19, 2014; Accepted June 3, 2014; Published July 11, 2014 pyright: 2014 Bates et al. This is an open-access article distributed under the terms of the Creative Commons Attribution L restricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: This work was supported by funding from the National Institutes of Health (5R01HL086897 and 5T32HL007654) and the American Heart Association (Postdoctoral Fellowship, Bates). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Melissa L. Bates1, Emily T. Farrell1, Alyssa Drezdon1, Joseph E. Jacobson1,4, Scott B. Perlman2, Marlowe W. Eldridge1,3 1 Department of Pediatrics, Critical Care Division and the John Rankin Laboratory of Pulmonary Medicine, University of Wisconsin School of Medicine and Public Health, Madison, Wisconsin, United States of America, 2 Department of Radiology, University of Wisconsin School of Medicine and Public Health, Madison, Wisconsin, United States of America, 3 Departments of Biomedical Engineering and Kinesiology, University of Wisconsin School of Medicine and Public Health, Madison, Wisconsin, United States of America, 4 Michigan State University College of Human Medicine, East Lansing, Michigan, United States of America July 2014 \| Volume 9 \| Issue 7 \| e101146 Introduction of healthy adults infused with epinephrine. The intensity of right- to-left bubble passage was correlated with cardiac output and these authors concluded that IPAVs recruitment was the result of a catecholamine-induced increase in cardiac output [13]. Addition- ally, Bryan, et al. demonstrated increases in cardiac output, intrapulmonary shunt (QS/QT), and transpulmonary bubble passage in healthy humans with dobutamine infusion [17]. Similarly, they found increased right-to-left bubble passage and also concluded that IPAVs recruitment was related to cardiac output. Inducible intrapulmonary arteriovenous anastamoses (IPAVs) are large diameter conduits that bypass the pulmonary capillary network. The existence of these pathways has been previously demonstrated with saline contrast echocardiography [1–3] and 99mTc-labeled albumin macroaggregates (99mTc-MAA) [4] in humans and solid latex microspheres in animals [5,6] and ex- vivo lung preparations [5–10]. While not typically perfused at rest, IPAVs open in ,90% of humans performing moderate to heavy intensity exercise [1,4]. These pathways are also recruited by hypoxia (FIO2 = 0.10–0.08) [6,11] and their recruitment during exercise can be prevented or reversed by hyperoxia (FIO2 = 1.0) [12,13]. Whether these observations are due to direct oxygen sensitive regulatory mechanisms or indirect effects secondary to redistribution pulmonary blood flow is unknown. It is possible, however, that the recruitment of IPAVs occurs only coincidentally with the elevation in cardiac output. The strongest evidence against a causative role for cardiac output is the fact that hyperoxia nearly universally reverses catecholamine- induced or exercise-induced IPAVs recruitment [12,13]. Further- more IPAVs recruitment can occur at rest with alveolar hypoxia with a minimal increase in cardiac output and pulmonary vascular pressure [11]. Indeed, these data provide compelling evidence that IPAVs regulation is in part mediated by an oxygen sensitive mechanism and not primarily by mechanical factors such as increases pulmonary blood flow and driving pressure. The mechanism by which IPAVs are recruited by exercise and hypoxia is not known. Several investigators have suggested that the transpulmonary passage of particles or bubbles through IPAVs is the result of elevated cardiac output and may occur in order to moderate pulmonary vascular pressure and decrease right ventricular stress [14–16]. Laurie, et al. Study design – Visits 1, 2, & 5 We recruited twelve individuals (18–35 years) from the general population of the University of Wisconsin-Madison. Data are available from the authors upon request. A 22-gauge catheter was placed in an antecubital vein for the injection of 99mTc-MAA. Participants then laid supine, donned nose clips, and breathed FIO2 = 0.21 or 0.10 (balance N2) through a mouthpiece fitted to a 2-way non-rebreathing valve. After three minutes, arterial blood was drawn via radial arterial puncture. Participants were then seated upright and continued breathing for an additional seven minutes. At the end of the breathing trial, 99mTc-MAA was injected into the intravenous catheter. Whole body images were then acquired to quantify the fraction of 99mTc- MAA bypassing the pulmonary circulation. Because a right-to-left intracardiac shunt could allow 99mTc- MAA to bypass the lung independent of IPAVs, we screened participants for septal shunts, including a patent foramen ovale (PFO). To observe these shunts, a saline contrast echocardiogram with a Valsalva maneuver was performed prior to participation. The Valsalva maneuver was standardized such that each participant donned nose clips and exhaled against a fixed resistance to generate +40 cmH2O mouth pressure for 10– 15 seconds, coinciding with the injection of saline contrast [18]. This pressure was then released and the transeptal passage of bubbles was assessed. This was repeated without Valsalva in order to exclude individuals with pathological pulmonary arteriovenous malformations. A PFO was observed in five individuals (42%) and they were excluded from participation. During visits 1&2, participants also performed a graded exercise test, on a magnetically-braked cycle ergometer (Velotron, RacerMate, Seattle, WA) while breathing either FIO2 = 0.21 or 0.10. Participants wore noseclips and a 2-way non-rebreathing mouthpiece for the measurement of ventilation and metabolic function. Beginning at 60 W, the workload was increased 25– 50 W every five minutes. The exercise test was ended when the participant could no longer maintain a pedaling cadence $55 rpm for three minutes. Each test lasted 25–30 minutes. Stages longer than those of a typical max test were used to mirror the exercise performed in visits 3&4, where additional time was required to accommodate the 99mTc-MAA injection and to allow sufficient time for complete circulation of the 99mTc-MAA particles. A single 99mTc-MAA injection requires about two minutes, including agitation of the material, injection, and thorough flushing of the infusion line and syringe. Study design – Visits 1, 2, & 5 Thus, a workload thought to be maintainable for the full duration of the 5-minute stage (85% of max wattage) was chosen. An exercise test was not performed at visit 5. The purpose of visit 5 was to evaluate the reproducibility of the 99mTc-MAA injection. Because 99mTc-MAA is contraindicated in pulmonary hyper- tension [19], the peak velocity of the tricuspid regurgitation jet [20,21] was evaluated in the remaining participants and found to be normal. Forced spirometry and single breath carbon monoxide diffusion capacity were measured using commercially-available equipment (Optima, Medgraphics, St. Paul, MN) using standard- ized techniques [22,23]. Percent predicted values were calculated using previously published regression equations [24–26]. Pulmo- nary function equipment was calibrated prior to each use with a 3L syringe and manufacturer-supplied carbon monoxide and neon control gases. Ethics statement Participants gave written, informed consent. The consent procedure and materials and the experimental protocol were approved by the Institutional Review Board at the University of Wisconsin School of Medicine and Public Health. Preparation of the 99mTc-MAA injection 99 Preparation of the 99mTc-MAA injection Briefly, 99mTc-MAA (Pulmolite, Pharmalucence, Bedford, MA) was prepared by a nuclear pharmacist such that each injection contained ,300,000 particles with 3–4 miC of activity. The fraction of unbound 99mTc was determined for each injection using thin-layer chromatography and was 4.862.8%. The same lot of MAA was used throughout the study. Microscopic analysis (ImageJ, National Institutes of Health, Bethesda, MD) of 16106 particles, sampled from three different vials revealed a mean particle diameter of 29 mm. with 97% of particles greater than 10 mm on the shortest axis, but none was greater than 120 mm. We investigated the feasibility of removing particles ,20 mm by vacuum filtration prior to injection (Steriflip, Millipore, Billerica, MA). Vacuum filtration was highly effective in removing particles ,20 mm, but resulted in the formation of several particles . 500 mm. Because the injection of particles .150 mm is not recommended in humans, we did not filter 99mTc-MAA prior to injection. p j Briefly, 99mTc-MAA (Pulmolite, Pharmalucence, Bedford, MA) was prepared by a nuclear pharmacist such that each injection contained ,300,000 particles with 3–4 miC of activity. The fraction of unbound 99mTc was determined for each injection using thin-layer chromatography and was 4.862.8%. The same lot of MAA was used throughout the study. Microscopic analysis (ImageJ, National Institutes of Health, Bethesda, MD) of 16106 particles, sampled from three different vials revealed a mean particle diameter of 29 mm. with 97% of particles greater than 10 mm on the shortest axis, but none was greater than 120 mm. We investigated the feasibility of removing particles ,20 mm by vacuum filtration prior to injection (Steriflip, Millipore, Billerica, MA). Vacuum filtration was highly effective in removing particles ,20 mm, but resulted in the formation of several particles . 500 mm. Because the injection of particles .150 mm is not recommended in humans, we did not filter 99mTc-MAA prior to injection. Overall study design Participants visited the laboratory on five occasions with at least one week between visits. This interval was selected to ensure complete clearance of any previously-injected 99mTc-MAA (99mTc half-life = 6.0 hours). During visits 1&2, 99mTc-MAA was injected intravenously after ten minutes of FIO2 = 0.21 or 0.10 breathing. During visits 3&4, 99mTc-MAA was injected while the participant breathed either FIO2 = 0.21 or 0.10 while exercising at 85% of their maximal exercise capacity. On the 5th visit, 99mTc-MAA was again injected after ten minutes of FIO2 = 0.21 breathing. Because Subject population and screening Study design – Visits 1, 2, & 5 Introduction observed right-to-left transpulmonary saline bubble passage through IPAVs in a group 1 July 2014 \| Volume 9 \| Issue 7 \| e101146 July 2014 \| Volume 9 \| Issue 7 \| e101146 PLOS ONE \| www.plosone.org IPAVs Recruitment in Hypoxia Given that IPAVs recruitment has never been quantified in the context of hypoxia or hypoxic exercise, it is also unknown if IPAVs recruitment is similar to that observed under other conditions. If IPAVs recruitment and cardiac output are indeed linked, then we would predict that their recruitment by hypoxia would be further enhanced by the addition of exercise. To test this hypothesis, we measured the transpulmonary passage of 99mTc-MAA particles at rest and at 85% of maximal exercise with normoxic and hypoxic gas breathing, according to previously described methods [4]. As a secondary endpoint, arterial blood gases were measured in each condition. By assessing IPAVs recruitment under these conditions, we are able to further contribute to our understanding of how these pathways are regulated. Surprisingly, and contrary to our hypothesis, we found the transpulmonary passage of 99mTc-MAA to be greatest with resting hypoxic gas breathing. the resting FIO2 = 0.21 injection is used as a baseline by which changes in 99mTc-MAA passage in the other conditions are compared [4], the 5th visit served as a control in order to evaluate the reproducibility of transpulmonary 99mTc-MAA passage. Preparation of the 99mTc-MAA injection Imaging Methods g g We have previously discussed, in detail, the use of 99mTc-MAA to detect IPAVs recruitment, including the limitations of this technique [4], and would direct the reader to this publication for further information. Participants were imaged supine with anterior and posterior whole body images acquired simultaneously by gamma camera (VPS-45, General Electric Medical Systems, Waukesha, WI) using a 20-minute ‘‘step and shoot’’ protocol. The anterior and posterior images were reconstructed into a single whole body image where each voxel represents the geometric mean of the anterior and posterior image (GE eNTEGRA software, GE Healthcare). The fraction of 99mTc-MAA traversing the pulmonary circulation was calculated by quantifying one minus the lung-to-whole body ratio from this unified image and correcting it for the amount of unbound 99mTc99. As described previously [4], the lung region of interest (ROI) was defined using the normoxic exercise scan and included all voxels with at least 3– 5% of the peak voxel intensity. The ROI was then applied to the other images and visually inspected to ensure that the ROI contained the entirety of the lung. Because saline contrast and solid microsphere studies have repeatedly demonstrated that IPAVs are not recruited at rest in normoxia, the lung-to-whole body ratio in the normoxic resting condition was set as a baseline. Changes with hypoxia and exercise are described subsequently as relative to this baseline. Again, to verify the reproducibility of the resting normoxic injection, this procedure was repeated on a second occasion. Discussion Our interest in intrapulmonary arteriovenous anastomoses (IPAVs) stems from the finding that these large diameter pathways can be recruited in ,90% of healthy individuals and may provide a conduit for right-to-left embolus transmission. Unlike our previous work investigating IPAVs recruitment with normoxic exercise [4], we have not directly quantified IPAVs recruitment with hypoxia or hypoxic exercise. To that end, our primary goal was to quantify the transpulmonary passage of 99mTc-MAA particles in the face of hypoxia. We aimed to further understand factors that influence their recruitment by specifically asking if exercise universally increases IPAVs recruitment. Here, we provide evidence suggesting that exercise, which increases cardiac Alveolar PO2 was calculated according to the following equation: PAO2 = PIO2–(PaCO2/R)+[FIO26PaCO26(1-R)/R], where PIO2 is the dry inspired partial pressure of oxygen and R is the respiratory exchange ratio [27]. The A-aDO2 was then calculated as the difference between the alveolar and arterial PO2. Blood gas and metabolic measurements For each sampling, arterial blood was drawn into three airtight, heparinized syringes. Each syringe was analyzed once (pHOx Basic, Nova Biomedical, Waltham, MA) immediately after collection and corrected for body temperature. No more than five minutes passed between the collection and analysis. A three- point calibration was done prior to the study using manufacturer- supplied control solutions. Blood gas parameters for normoxic and hypoxic rest and exercise are given in Table 2. Exercise widened the A-aDO2, although there was no difference between normoxic and hypoxic exercise (p = 0.43). The A-aDO2 during the last exercise stage was significantly related to the fraction of transpulmonary 99mTc-MAA passage (p = 0.01) and 63% of the variance in the A-aDO2 was accounted for by the percent passage of 99mTc-MAA (Figure 3). Equipment used to measure metabolic and ventilatory param- eters (Optima, Medgraphics, St. Paul, MN) was calibrated prior to use with a 3L syringe and manufacturer-supplied carbon dioxide and oxygen control gases. Study design – Visits 3&4 Prior to exercise, participants swallowed a pill-shaped core temperature probe (CorTemp, HQ Inc., Palmetto, FL). A 22- gauge catheter was placed in an antecubital vein for 99mTc-MAA injection and a 3F angiocath was placed in a brachial artery for blood sampling. Participants were seated on the ergometer and fitted with noseclips. Ventilation and metabolic function were measured as in PLOS ONE \| www.plosone.org July 2014 \| Volume 9 \| Issue 7 \| e101146 2 IPAVs Recruitment in Hypoxia visits 1&2. After being seated at rest for 10 minutes, participants began exercising at 30% of the maximal wattage achieved during visit 1&2. They completed five minute stages at 30%, 45%, 60%, 75%, and 85% of maximal capacity. During the third minute of each stage, arterial blood gases were drawn. One participant during normoxic exercise and two participants during hypoxic exercise became exhausted prior to reaching 85% of max and measurements were made in the 75% stage. During the third minute of the final stage, 99mTc-MAA was injected into the intravenous catheter such that the 99mTc-MAA injection coincided with the drawing of arterial blood. evaluate differences in 99mTc-MAA passage with hypoxia and exercise. Significance was determined when p,0.05 with a Bonferroni correction applied for multiple comparisons. Predictors of the A-aDO2 during exercise were evaluated by repeated measures general linear model (Minitab, State College, PA) in which the participant was treated as a random variable, the FIO2 as a fixed variable, and the 99mTc-MAA fraction as a covariate. Reproducibility of the baseline 99mTc-MAA injection was assessed by calculating the intra-class correlation coefficient between the repeated visits and the repeatability coefficient [28]. If 99mTc- MAA passage was beyond the repeatability coefficient, then it was concluded that the difference was outside of the measurement’s error range. Baseline transpulmonary 9 Baseline transpulmonary 99mTc-MAA passage Transpulmonary passage of 99mTc-MAA was 11.769.1% and 11.469.4% during repeated normoxic rest (visits 1 and 5, respectively) and were not statistically different (p = 0.41). This is consistent with the percentage of particles known to be #15 mm. The repeatability coefficient was 0.92%. As we have done previously [4], each individual’s resting value was used as a correction factor for all subsequent scans. At rest, hypoxia increased the passage of 99mTc-MAA particles in 6/7 participants (D5.362.9%) (Figure 1). Normoxic exercise increased 99mTc-MAA passage in 5/7 participants (71%) (D3.562.3%) and hypoxic exercise increased 99mTc-MAA passage in 4/7 participants (57%) (D3.562.5). The overall population means in normoxic and hypoxic exercise were not different (D2.562.6% vs D1.763.0%, p = 0.60), nor were the percent changes in transpulmonary 99mTc- MAA passage correlated between the two conditions (R2,0.01, p = 0.996), meaning that the same individuals did not necessarily recruit IPAVs similarly in both conditions (Figure 2A). While the percent changes in transpulmonary passage of 99mTc-MAA for each individual were well-correlated between hypoxic rest and hypoxic exercise (R2 = 0.82, p = 0.005), the overall percentage of 99mTc-MAA passage was higher with hypoxic rest than hypoxic exercise (D4.663.1% vs D2.562.6%, p = 0.001) (Figure 2B). Results Anthropometric characteristics, pulmonary function, and exer- cise capacity data are given in Table 1. July 2014 \| Volume 9 \| Issue 7 \| e101146 July 2014 \| Volume 9 \| Issue 7 \| e101146 Methodological considerations We have previously validated the existence of IPAVs in intact rats and dogs and isolated rat, baboon, dog, and human lungs using solid latex particles of known diameter [5,7,29]. Although the lung is not permeable to solid latex particles 15–70 mm under resting conditions, these large particles traverse the lung with the addition of hypoxia or exercise [5,29]. A limitation of the study of IPAVs in humans is that, in contrast to our animal studies, we are unable to directly measure the size of the 99mTc-MAA particles that traverse the lung. However, we previously have demonstrated that the fraction of 99mTc-MAA particles that traverse the lung in exercising humans is similar to the fraction of radiolabeled latex particles that traverse the lungs of dogs performing similar exercise (1.3 and 1.4%, respectively) [4]. Figure 2. Relationship between transpulmonary 99mTc-MAA passage with exercise in normoxia vs. hypoxia (A) and between rest and exercise in hypoxia (B). The transpulmonary passage of 99mTc-MAA with exercise in hypoxia was well-correlated with that measured at rest with hypoxic gas breathing. Dashed line indicates the line of identity. doi:10.1371/journal.pone.0101146.g002 It has been suggested that small particles traverse the lung by ‘‘squeezing’’ through distended pulmonary capillaries when pulmonary vascular pressure is elevated [15,16]. Assuming that the pulmonary vascular distends 2% per mmHg increase in pulmonary vascular pressure [30], at physiological perfusion pressures we would expect a 6 mm pulmonary capillary to distend to no more than 8 mm. Manohar and Goetz demonstrated that the lung of the thoroughbred race horse, which apparently lacks IPAVs but is perfused at extraordinarily high pressure during exercise, is not permeable to 15mm microspheres, suggesting that the maximal capillary diameter is ,15 mm [31]. Furthermore, the binding of 99mTc to albumin is proportional to the particle mass and our previous calculations demonstrate that the contribution of the additional 99mTc-MAA particles #15 mm that might pass through pulmonary capillaries during exercise is ,0.4% of the total signal. [4]. Thus, we are confident that the additional 99mTc signal with exercise and hypoxia is the result of large diameter particles traversing IPAVs. Furthermore, if 99mTc-MAA is traversing the pulmonary capillaries, we would expect that the passage of 99mTc-MAA would be greatest during hypoxic exercise when pulmonary vascular pressures are highest and to increase in all participants. Statistical Analysis Change in the transpulmonary passage (%) of 99mTc- MAA compared to resting, normoxic gas breathing. Dashed line indicates the repeatability coefficient (0.92%) Transpulmonary 99mTc- MAA passage was noted in 6/7 participants performing exercise in normoxia and 4/7 participants performing exercise in hypoxia. Breathing hypoxic gas at rest increased 99mTc-MAA passage in all participants relative to hypoxic exercise. * indicates a difference compared to hypoxic rest (p = 0.001). doi:10.1371/journal.pone.0101146.g001 Figure 1. Change in the transpulmonary passage (%) of 99mTc- MAA compared to resting, normoxic gas breathing. Dashed line indicates the repeatability coefficient (0.92%) Transpulmonary 99mTc- MAA passage was noted in 6/7 participants performing exercise in normoxia and 4/7 participants performing exercise in hypoxia. Breathing hypoxic gas at rest increased 99mTc-MAA passage in all participants relative to hypoxic exercise. ** indicates a difference compared to hypoxic rest (p = 0.001). doi:10.1371/journal.pone.0101146.g001 output, does not necessarily increase IPAVs recruitment. Indeed, if cardiac output were a major modulator of IPAVS recruitment in hypoxia, we would expect 99mTc-MAA passage to be higher during exercise than at rest. Instead, we found it to be greatest with resting hypoxia and similar between both exercise conditions. Statistical Analysis All data are displayed as 6 standard deviation unless otherwise noted. Blood gas and metabolic parameters were compared using repeated measures ANOVA with post hoc analysis (Student’s t- test) to compare differences between exercise and rest, as well as between normoxia and hypoxia. Paired t-tests were used to PLOS ONE \| www.plosone.org 3 July 2014 \| Volume 9 \| Issue 7 \| e101146 IPAVs Recruitment in Hypoxia Table 1. Anthropometric characteristics of the seven participants completing the study. VO2max VO2max Age Height Weight FEV1 DLCO FIO2 = 0.21 FIO2 = 0.10 Subject (yrs) Sex (cm) (kg) (L) FEV1/FVC (mL/min/Torr) (mL/kg/min) (mL/kg/min) 01 23 F 165 56 3.54 0.83 30.74 30.9 20.6 (113%) (97%) (92.3) 02 22 M 183 80 5.10 0.85 45.06 56.5 32.4 (101%) (119%) (117.1) 03 21 F 171 61 3.98 0.88 32.12 43.3 30.2 (116%) (95%) (98.4) 04 31 M 172 88 4.02 0.88 37.14 33.6 18.6 (106%) (109%) (76.4) 05 33 F 171 62 3.90 0.83 32.57 57.1 32.3 (121%) (96%) (132.4) 06 21 M 178 76 4.42 0.86 34.59 45.2 23.8 (95%) (96%) (93.3) 07 20 M 184 72 4.36 0.75 42.92 58.0 30.5 (103%) (101%) (118.6) Mean 6 2465 17567 71612 4.1960.50 0.8460.04 36.4565.57 46.4611.3 26.965.8* SD (10869) (10269) (104619) FVC, forced vital capacity; FEV1, forced expired volume in 1 second; DLCO, diffusion capacity for carbon monoxide; VO2max; relative maximal oxygen uptake. indicates p,0 05 Values in parentheses are percent predicted (23 25) PLOS ONE \| www.plosone.org 4 IPAVs Recruitment in Hypoxia does not necessarily increase IPAVs recruitment. Indeed, if c output were a major modulator of IPAVS recruitment in a, we would expect 99mTc-MAA passage to be higher exercise than at rest. Instead, we found it to be greatest with hypoxia and similar between both exercise conditions. odological considerations have previously validated the existence of IPAVs in intact nd dogs and isolated rat, baboon, dog, and human lungs olid latex particles of known diameter [5,7,29]. Although g is not permeable to solid latex particles 15–70 mm under conditions, these large particles traverse the lung with the n of hypoxia or exercise [5,29]. A limitation of the study of in humans is that, in contrast to our animal studies, we are to directly measure the size of the 99mTc-MAA particles averse the lung. Statistical Analysis However, we previously have demonstrated e fraction of 99mTc-MAA particles that traverse the lung in ing humans is similar to the fraction of radiolabeled latex es that traverse the lungs of dogs performing similar exercise d 1.4%, respectively) [4]. as been suggested that small particles traverse the lung by zing’’ through distended pulmonary capillaries when nary vascular pressure is elevated [15,16]. Assuming that ulmonary vascular distends 2% per mmHg increase in nary vascular pressure [30], at physiological perfusion ld 6 l ill di d 1. Change in the transpulmonary passage (%) of 99mTc- ompared to resting, normoxic gas breathing. Dashed line es the repeatability coefficient (0.92%) Transpulmonary 99mTc- assage was noted in 6/7 participants performing exercise in xia and 4/7 participants performing exercise in hypoxia. ng hypoxic gas at rest increased 99mTc-MAA passage in all ants relative to hypoxic exercise. * indicates a difference ed to hypoxic rest (p = 0.001). 371/journal.pone.0101146.g001 Figure 2. Relationship between transpulmonary 99mTc-MAA passage with exercise in normoxia vs. hypoxia (A) and between rest and exercise in hypoxia (B). The transpulmonary passage of 99mTc-MAA with exercise in hypoxia was well-correlated with that measured at rest with hypoxic gas breathing. Dashed line indicates the line of identity. doi:10.1371/journal.pone.0101146.g002 Figure 2. Relationship between transpulmonary 99mTc-MAA passage with exercise in normoxia vs. hypoxia (A) and between rest and exercise in hypoxia (B). The transpulmonary passage of 99mTc-MAA with exercise in hypoxia was well-correlated with that measured at rest with hypoxic gas breathing. Dashed line indicates the line of identity. doi:10.1371/journal.pone.0101146.g002 Figure 1. Change in the transpulmonary passage (%) of 99mTc- MAA compared to resting, normoxic gas breathing. Dashed line indicates the repeatability coefficient (0.92%) Transpulmonary 99mTc- MAA passage was noted in 6/7 participants performing exercise in normoxia and 4/7 participants performing exercise in hypoxia. Breathing hypoxic gas at rest increased 99mTc-MAA passage in all participants relative to hypoxic exercise. indicates a difference compared to hypoxic rest (p = 0.001). doi:10.1371/journal.pone.0101146.g001 Figure 1. Change in the transpulmonary passage (%) of 99mTc- MAA compared to resting, normoxic gas breathing. Dashed line indicates the repeatability coefficient (0.92%) Transpulmonary 99mTc- MAA passage was noted in 6/7 participants performing exercise in normoxia and 4/7 participants performing exercise in hypoxia. Breathing hypoxic gas at rest increased 99mTc-MAA passage in all participants relative to hypoxic exercise. indicates a difference compared to hypoxic rest (p = 0.001). doi:10 1371/journal pone 0101146 g001 Figure 1. What influences the recruitment of large diameter IPAVs? In humans and experimental animal models with lesions that result in unventilated lung regions, increasing cardiac output typically increases physiological intrapulmonary shunt (QS/QT) [23,33–38]. It might, therefore, seem surprising and counterintu- itive that we found a decrease in IPAVs perfusion with hypoxic exercise, where the cardiac output is higher, compared to rest. Our finding are, in fact, congruent with previous work in this area and understanding the interaction between FIO2, cardiac output, and QS/QT provides insight into what factors may regulate IPAVs perfusion. Many previous studies have relied on models of lung injury where the FIO2 is normoxic or hyperoxic and injured regions of the lung are focally hypoxic because of the poor ventilation (ie, respiratory distress syndrome, pulmonary edema and atelectasis, arteriovenous malformation and methacholine challenge) [33–38]. However, Wagner, et al. measured the increase in QS/QT in a methacholine-challenged dog model with normoxic, hyperoxic, and hypoxic ventilation [39], thereby fundamentally improving our understanding of this phenomenon. In all conditions, methacholine challenge resulted in an initial 28% shunt. With normoxic and hyperoxic ventilation, this shunt increased 7% and 11% per L/min increase in cardiac output, but only 2% with hypoxic ventilation. The authors explain these different changes in QS/QT by asking the reader to consider a lung with two vascular compartments with parallel resistances – one that is normal and well-ventilated and one that is injured and poorly ventilated. Under normoxic and hyperoxic conditions, the poorly ventilated compartment is hypoxic, resulting in a high degree of vasocon- striction and high vascular resistance. The resistance in the well- ventilated compartment is lower than the injured compartment. As cardiac output increases, there is a greater fall in resistance in the poorly ventilated compartment relative to the well-ventilated compartment, thereby redistributing the cardiac output to the injured compartment. When both compartments are hypoxic, the resistances are similar and the cardiac output is distributed similarly between the compartments. To understand our finding that IPAVs perfusion decreases with exercise, we would similarly ask the reader to consider a lung with two vascular compartments. Unlike Wagner et al.’s model, these resistances are morphometrically distinct. One compartment contains large diameter of IPAVs, which lie upstream of a second compartment of pulmonary arteries, capillaries, and veins. Methodological considerations Instead, we found passage to be the greatest during resting hypoxia, suggesting that distended capillaries are not the primary source of transpulmonary capillary passage. The excellent repeatability of the measurement in resting, normoxic individuals gives us confidence that the differences observed with hypoxia and exercise represent true increases in transpulmonary particle passage and are not caused by free pertechnate or inter-injection variability. Indeed, we took care to perform all scans within 10 minutes of the conclusion of exercise and found very little free label (,0.1% of the total extrapulmonary signal), in the bladder, gut, and thyroid, where free pertechnate accumulates. If our findings were caused by free pertechnate, we would expect to find substantial labeling of these organs [32]. It is also attractive to consider filtering very small particles from the injectate. However, in our hands this resulted in the significant aggregation of particles, resulting in the formation of large particles that are unsafe for injection in humans. July 2014 \| Volume 9 \| Issue 7 \| e101146 PLOS ONE \| www.plosone.org July 2014 \| Volume 9 \| Issue 7 \| e101146 5 IPAVs Recruitment in Hypoxia Figure 3. Relationship between transpulmonary 99mTc-MAA passage and the alveolar-arterial oxygen difference (A-aDO2) in normoxic and hypoxic exercise. Line represents the result of a general linear model analysis in which the transpulmonary 99mTc-MAA passage was linearly correlated with the A-aDO2 (R2 = 0.63), but this relationship was not dependent on the FIO2 (p.0.05). doi:10.1371/journal.pone.0101146.g003 Table 2. Arterial blood gases, respiratory quotient (R), and the alveolar-arterial PO2 difference (A-aDO2), measured at rest and at 85% of the maximal attainable wattage during th resting and exercise visits. Condition FIO2 R pH PCO2 [mmHg] PO2 [mmHg] A-aDO2 [mmHg] Exercise Visit Rest 0.21 0.8660.11 7.4660.02 38.864.9 102.168.3 21.265.0 Rest 0.10 1.3460.15 7.4860.02 34.363.9 43.863.6 21.164.7 85% Max Wattage 0.21 1.1560.11* 7.3860.07* 33.564.0* 101.365.7 14.165.2* 85% Max Wattage 0.10 1.2760.16* 7.3960.07* 27.163.0{ 37.860.5{ 11.165.9* Resting Visit Rest 0.21 0.8760.09 7.4860.05 38.764.9 103.869.8 2.667.1 Rest 0.10 1.1960.11 7.4860.01 37.963.3 42.165.0 21.864.7 *i di t ,0 05 d t l d t t th d PLOS ONE \| www.plosone.org Figure 3. Relationship between transpulmonary 99mTc-MAA passage and the alveolar-arterial oxygen difference (A-aDO2) in normoxic and hypoxic exercise. Methodological considerations Line represents the result of a general linear model analysis in which the transpulmonary 99mTc-MAA passage was linearly correlated with the A-aDO2 (R2 = 0.63), but this relationship was not dependent on the FIO2 (p.0.05). doi:10.1371/journal.pone.0101146.g003 July 2014 \| Volume 9 \| Issue 7 \| e101146 What influences the recruitment of large diameter IPAVs? However, we acknowledge that an important limitation of our work, and indeed the field at this point, is the inability to prevent IPAVs recruitment during exercise in an O2-independent manner, thereby allowing the contribution of IPAVs to the A-aDO2 to be evaluated during exercise without changes in FIO2 that might alter IPAVs patency. We also recognize that it is possible that the increase in 99mTc- MAA occurs only coincidentally with the increased in A-aDO2. The strong correlation between the transpulmonary passage of 99mTc-MAA during hypoxic rest and hypoxic exercise supports this interpretation and suggests that when IPAVS are open, the fraction of the cardiac output that perfuses them is determined by two factors – 1) their relative abundance, which contributes to intersubject variability and 2) the downstream resistance, which redistributes blood flow through IPAVs and contributes to intercondition variability. Moreover, Tobin used a mechanism very similar to that we propose to identify IPAVs in isolated lungs [41]. By strategically elevating the downstream vascular resistance, he was able to preferentially fill these pathways with casting material. If IPAVs are located upstream of the vessels that are the largest contributors to the vascular resistance [43–45], then the tone of the resistance vessels could determine flow through IPAVs. We note a similar increase in IPAVs recruitment proportional to the increase in A-aDO2 in both normoxia and hypoxia, but the relative contributions of ventilation-perfusion inequality, diffusion limitation and shunt to the increased A-aDO2 during exercise is certainly dependent on the FIO2. Vogiatzis, et al. found a seven- fold increase in Qs/QT with hypoxic exercise, yet we found no additional increase in IPAVs recruitment with our particle method. If we speculate that IPAVs do contribute to Qs/QT, then additional contributors, including diffusion limitation, would need to account for the remaining A-aDO2. While we note a positive relationship between A-aDO2 and the fractional IPAVs recruitment, future studies will be needed to separate the simultaneous contributions of IPAVs, ventilation-perfusion in- equality, and diffusion limitation in normoxia and hypoxia, posing an interesting challenge to investigators motivated to understand the mechanism of the widened A-aDO2 in exercising humans. Eldridge, et al. [1] previously speculated that IPAVs exist as a ‘‘pressure relief valve’’ to normalize pulmonary pressure in the face of high cardiac output. Although this explanation seems physio- logically plausible, it is not supported by these data and challenges our own previous ideas about IPAVs recruitment. What influences the recruitment of large diameter IPAVs? It is possible that the recruitment of IPAVs is not caused by increased cardiac output and pressure, but by biochemical signals that directly impact their tone. Hyperoxia consistently prevents or reverses the recruitment of IPAVs, even in the face of high intensity exercise [12] and epinephrine and dobutamine infusion [13], suggesting that IPAVs regulation by hyperoxia occurs as a result of a distinct and secondary O2-mediated mechanism. Further supporting a role for oxygen sensitivity, we found previously that PvO2,22 mmHg is required for the recruitment of IPAVs in rats and we speculated that IPAVs are recruited via an oxygen-sensitive mechanism, in response to pulmonary arterial hypoxemia [6]. Epinephrine, dobutamine, and dopamine are vasoactive and may affect the tone of IPAVs independently of their effect on total pulmonary vascular flow. Isolated lung models may be valuable in future investigations, allowing the evaluation of biochemically-mediated effects while controlling pulmonary vascular pressure and flow. g It may seem counterintuitive that the A-aDO2 would be unchanged with resting hypoxia, despite the fact that the IPAVS recruitment was greatest in this condition. This is likely the result of the narrow arteriovenous oxygen difference in this condition. Relying on historical data obtained from resting humans breathing FIO2 = 0.10 [53], we can estimate that the mixed venous saturation of hemoglobin (SO2) is 50% in our participants. Using our participants own mean value of alveolar PO2 and the Kelman equation to convert between PO2 and SO2 [54], we can estimate venous and alveolar oxygen content [27]. Using the standard shunt equation [27], we can then calculate the contribution of a 5% anatomic shunt on the A-aDO2. Indeed, this calculation reveals that the impact of a 5% anatomic shunt on the A-aDO2 is trivial under these conditions (1.2 mmHg), thereby demonstrating that IPAVs recruitment need not always result in a widened A- aDO2. What influences the recruitment of large diameter IPAVs? We have previously shown in rats that IPAVs are $70 mm [6], placing them early in the vascular tree and upstream of the primary resistance arterioles [40] that constrict in response to hypoxia. This location is further supported by anatomic studies that identify July 2014 \| Volume 9 \| Issue 7 \| e101146 6 IPAVs Recruitment in Hypoxia arteriovenous connections early in the vascular tree [41]. With resting hypoxia, large diameter IPAVs are open and there is also considerable vasoconstriction of the resistance arterioles, directing blood flow through IPAVs. With exercise, the post-IPAVs resistance falls tremendously [42]. Thus, following similar logic to that used by Wagner, et al. [39], we would expect the perfusion to IPAVs to fall as a function of the fall in resistance of the non- IPAVs compartment. arteriovenous connections early in the vascular tree [41]. With resting hypoxia, large diameter IPAVs are open and there is also considerable vasoconstriction of the resistance arterioles, directing blood flow through IPAVs. With exercise, the post-IPAVs resistance falls tremendously [42]. Thus, following similar logic to that used by Wagner, et al. [39], we would expect the perfusion to IPAVs to fall as a function of the fall in resistance of the non- IPAVs compartment. that Gledhill et al. calculated [52] would be required to explain the remaining A-aDO2 after accounting for ventilation-perfusion inequality. This suggests that the shunting of venous admixture through intrapulmonary pathways may, in fact, be an important contributor to Qs/QT in normoxic exercise. However, we acknowledge that an important limitation of our work, and indeed the field at this point, is the inability to prevent IPAVs recruitment during exercise in an O2-independent manner, thereby allowing the contribution of IPAVs to the A-aDO2 to be evaluated during exercise without changes in FIO2 that might alter IPAVs patency. We also recognize that it is possible that the increase in 99mTc- MAA occurs only coincidentally with the increased in A-aDO2. that Gledhill et al. calculated [52] would be required to explain the remaining A-aDO2 after accounting for ventilation-perfusion inequality. This suggests that the shunting of venous admixture through intrapulmonary pathways may, in fact, be an important contributor to Qs/QT in normoxic exercise. Conclusion Here, we demonstrate the perfusion of IPAVs is greatest in hypoxic individuals and that elevating the cardiac output with exercise does not increase IPAVs recruitment. The A-aDO2 is related to the magnitude of transpulmonary 99mTc-MAA passage We appreciate that, because of the number of visits required for the completion of this study, blood gases were obtained via arterial puncture preceding the resting injections rather than by placing an in-dwelling catheter. Our participants found this to be more comfortable in the supine position, which may impact the A-aDO2 and limit our ability to interpret these data. There was, however, no significant difference between the A-aDO2 measured during the resting visit in the supine position and the A-aDO2 measured at rest in the upright position during the exercising visits. p y p g Whether IPAVs contribute to exercise-induced arterial hypox- emia [46] has been a controversial question [47,48]. We have been motivated to use particles to quantify the recruitment of large diameter pathways that could contribute to anatomic shunt by the recognition of substantial pre-capillary gas exchange [49,50] that might confound measurements made with SF6 or 100% O2, thereby underestimating shunt under normoxic or hypoxic conditions. Although 99mTc-MAA passage has been related to SpO2, it has not been related to the A-aDO2. Thus, as a secondary endpoint of this study, we measured arterial blood gases in our participants and found the magnitude of IPAVs recruitment to be well-correlated with the A-aDO2. Indeed, we were intrigued by the fact that the transpulmonary passage of 99mTc-MAA was quite similar to actual values of physiological shunt (QS/QT) reported elsewhere. Vogiatzis, et al. reported Qs/QT in humans performing normoxic exercise at power outputs similar to ours and found Qs/ QT to be 3.5% [51]. This is similar to the 2.5% passage of 99mTc- MAA reported here. This is also similar to the amount of shunt References Wagner PD, Schaffartzik W, Prediletto R, Knight DR (1991) Relationship among cardiac-output, shunt, and inspired O2 concentration J Appl Physiol 71: 2191–2197. 15. La Gerche A, MacIsaac AI, Burns AT, Mooney DJ, Inder WJ, et al. (2010) Pulmonary transit of agitated contrast is associated with enhanced pulmonary vascular reserve and right ventricular function during exercise. J Appl Physiol 109: 1307–1317. 40. Jiang ZL, Kassab GS, Fung YC (1994) Diameter-defined Strahler system and connectivity matrix of the pulmonary arterial tree. J Appl Physiol 76: 882–892. 41. Tobin CE (1966) Arteriovenous shunts in the peropheral pulmonary circulation in the human lung. Thorax 21: 197–204. 16. Lalande S, Yerly P, Faoro V, Naeije R (2012) Pulmonary vascular distensibility predicts aerobic capacity in healthy individuals. J Physiol 590: 4279–4288. 42. Eldridge MW, Podolsky A, Richardson RS, Johnson DH, Knight DR, et al. (1996) Pulmonary hemodynamic response to exercise in subjects with prior high- altitude pulmonary edema. J Appl Physiol 81: 911–921. 17. Bryan TL, van Diepen S, Bhutani M, Shanks M, Welsh RC, et al. (2012) The effects of dobutamine and dopamine on intrapulmonary shunt and gas exchange in healthy humans. J Appl Physiol 113: 541–548. 43. Dingemans KP, Wagenvoort CA (1978) Pulmonary arteries and veins in experimental hypoxia. An ultrastructural study. Am J Pathol 93: 353–368. 18. Marriott K, Manins V, Forshaw A, Wright J, Pascoe R (2013) Detection of Right-to-Left Atrial Communication Using Agitated Saline Contrast Imaging: Experience with 1162 Patients and Recommendations for Echocardiography. Journal of the American Society of Echocardiography 26: 96–102. perimental hypoxia. An ultrastructural study. Am J Pathol 93: 353–3 44. Kato M, Staub NC (1966) Response of small pulmonary arteries to unilobar hypoxia and hypercapnia. Circ Res 19: 426–440. p Journal of the American Society of Echocardiography 26: 96–102. 45. Staub NC (1985) Site of hypoxic pulmonary vasoconstriction. Chest 88: 240S– 245S. 19. Pharmalucence (2010) Pulmolite, kit for the preparation of technetium Tc99m albumin aggregated for injection U.S. National Library of Medicine. 46. Dempsey JA, Wagner PD (1999) Exercise-induced arterial hypoxemia. J Appl Physiol 87: 1997–2006. albumin aggregated for injection U.S. National Library of Medicin 20. McLaughlin VV, Archer SL, Badesch DB, Barst RJ, Farber HW, et al. References 25. Cotes JE, Chinn DJ, Quanjer PH, Roca J, Yernault JC (1993) Standardization of the measurement of transfer-factor (diffusing capacity)-report working party standardization of lung-function tests European-community for steel and coal- official statement of the european respiratory society. European Respiratory Journal 6: 41–52. 1. Eldridge MW, Dempsey JA, Haverkamp HC, Lovering AT, Hokanson JS (2004) Exercise-induced intrapulmonary arteriovenous shunting in healthy humans. J Appl Physiol 97: 797–805. J pp y 2. Elliott JE, Choi Y, Laurie SS, Yang X, Gladstone IM, et al. (2011) Effect of initial gas bubble composition on detection of inducible intrapulmonary arteriovenous shunt during exercise in normoxia, hypoxia, or hyperoxia. J Appl Physiol 110: 35–45. 26. Jones NL, Makrides L, Hitchcock C, Chypchar T, McCartney N (1985) Normal standards for an incremental progressive cycle ergometer test. Am Rev Respir Dis 131: 700–708. J pp y 3. Lovering AT, Romer LM, Haverkamp HC, Pegelow DF, Hokanson JS, et al. (2008) Intrapulmonary shunting and pulmonary gas exchange during normoxic and hypoxic exercise in healthy humans. J Appl Physiol 104: 1418–1425. 27. West JB (2008) Respiratory Physiology: The Essentials 8th Edition. Philadelphia: Wolters Kluwer. 208 p. 28. Cleophas TJ, Zwinderman AH, Cleophas TF (2006) Statistics Applied to Clinical Trials: Springer. 4. Lovering AT, Haverkamp HC, Romer LM, Hokanson JS, Eldridge MW (2009) Transpulmonary passage of 99mTc macroaggregated albumin in healthy humans at rest and during maximal exercise. J Appl Physiol 106: 1986–1992. p g 29. Bates ML, Fulmer BR, Farrell ET, Drezdon A, Pegelow DF, et al. (2012) Hypoxia recruits intrapulmonary arteriovenous pathways in intact rats but not isolated rat lungs. J Appl Physiol 112: 1915–1920. 5. Stickland MK, Lovering AT, Eldridge MW (2007) Exercise-induced arteriove- nous intrapulmonary shunting in dogs. Am J Respir Crit Care Med 176: 300– 305. g pp y 30. Reeves JT, Linehan JH, Stenmark KR (2005) Distensibility of the normal human lung circulation during exercise. Am J Physiol Lung Cell Mol Physiol 288: L419–425. 6. Bates M, Fulmer B, Farrell E, Drezdon A, Pegelow D, et al. (2012) Hypoxia recruits intrapulmonary arteriovenous pathways in intact rats but not isolated rat lungs. J Appl Physiol. 31. Manohar M, Goetz TE (2005) Intrapulmonary arteriovenous shunts of .15 mu m in diameter probably do not contribute to arterial hypoxemia in maximally exercising Thoroughbred horses. J Appl Physiol 99: 224–229. g J pp y 7. References (2009) ACCF/AHA 2009 expert consensus document on pulmonary hypertension: a report of the American College of Cardiology Foundation Task Force on Expert Consensus Documents and the American Heart Association: developed in collaboration with the American College of Chest Physicians, American Thoracic Society, Inc., and the Pulmonary Hypertension Association. Circulation 119: 2250–2294. y 47. Lovering AT, Eldridge MW, Stickland MK (2009) Counterpoint: Exercise- induced intrapulmonary shunting is real. J Appl Physiol 107: 994–997. 48. Hopkins SR, Olfert IM, Wagner PD (2009) Point: Exercise-induced intrapul- monary shunting is imaginary. J Appl Physiol 107: 993–994. 48. Hopkins SR, Olfert IM, Wagner PD (2009) Point: Exercise- monary shunting is imaginary. J Appl Physiol 107: 993–994 49. Tabuchi A, Styp-Rekowska B, Slutsky AS, Wagner PD, Pries AR, et al. (2013) Precapillary Oxygenation Contributes Relevantly to Gas Exchange in the Intact Lung. Am J Respir Crit Care Med 188: 474–481. 21. Galie N, Hoeper MM, Humbert M, Torbicki A, Vachiery JL, et al. (2009) Guidelines for the diagnosis and treatment of pulmonary hypertension: the Task Force for the Diagnosis and Treatment of Pulmonary Hypertension of the European Society of Cardiology (ESC) and the European Respiratory Society (ERS), endorsed by the International Society of Heart and Lung Transplanta- tion (ISHLT). Eur Heart J 30: 2493–2537. g J p 50. Conhaim RL, Staub NC (1980) Reflection spectrophotometric measurement of O2 uptake in pulmonary arterioles of cats. J Appl Physiol 48: 848–856. O2 uptake in pulmonary arterioles of cats. J Appl Physiol 48: 848–8 51. Vogiatzis I, Zakynthinos S, Boushel R, Athanasopoulos D, Guenette JA, et al. (2008) The contribution of intrapulmonary shunts to the alveolar-to-arterial oxygen difference during exercise is very small. J Physiol 586: 2381–2391. yg g y y 52. Gledhill N, Froese A, Dempsey J (1977) Ventilation to perfusion distribution during exercise in health. In: Dempsey J, Reed C, editors. Muscular Exercise and the Lung. Madison, WI: University of Wisconsin Press. 325–344. 22. Miller MR, Hankinson J, Brusasco V, Burgos F, Casaburi R, et al. (2005) Standardisation of spirometry. European Respiratory Journal 26: 319–338. and the Lung. Madison, WI: University of Wisconsin Press. 325–34 23. American Thoracic Society (1995) Single-breath carbon monoxide diffusing capacity (transfer factor). Recommendations for a standard technique–1995 update. Am J Respir Crit Care Med 152: 2185–2198. 53. References Lovering AT, Stickland MK, Kelso AJ, Eldridge MW (2007) Direct demonstration of 25- and 50-microm arteriovenous pathways in healthy human and baboon lungs. Am J Physiol Heart Circ Physiol 292: H1777–1781. 32. Dogan AS, Rezai K, Kirchner PT, Stuhlmuller JE (1993) A scintographic sign for the detection of right-to-left shunts. Journal of Nuclear Medicine 34: 1607– 1611. 8. Niden AH, Aviado DM Jr (1956) Effects of pulmonary embolism on the pulmonary circulation with special reference to arteriovenous shunts in the lung. Circ Res 4: 67–73. 33. Bishop MJ, Cheney FW (1983) Effects of pulmonary blood-flow and mixed venous O2 tension on gas-exchange in dogs. Anesthesiology 58: 130–135. 9. Tobin CE (1966) Arteriovenous Shunts in Peripheral Pulmonary Circulation in Human Lung. Thorax 21: 197-&. 34. Breen PH, Schumacker PT, Hedenstierna G, Ali J, Wagner PD, et al. (1982) How does increased cardiac-output increase shunt in pulmonary edema. J Appl Physiol 53: 1273–1280. 10. Prinzmetal M, Ornitz EM Jr, Simkin B, Bergman HC (1948) Arterio-venous anastomoses in liver, spleen, and lungs. Am J Physiol 152: 48–52. 35. Breen PH, Schumacker PT, Sandoval J, Mayers I, Oppenheimer L, et al. (1985) Increased cardiac-output increases shunt-role of pulmonary-edema and perfusion J Appl Physiol 59: 1313–1321. 11. Laurie SS, Yang X, Elliott JE, Beasley KM, Lovering AT (2011) Hypoxia- induced intrapulmonary arteriovenous shunting at rest in healthy humans (vol 109, pg 1072, 2010). J Appl Physiol 110: 1502–1502. p J pp y 36. Colley PS, Cheney FW (1977) Sodium nitroprusside increases QS-QT in dogs with regional atelectasis. Anesthesiology 47: 338–341. pg ) J pp y 12. Lovering AT, Stickland MK, Amann M, Murphy JC, O’Brien MJ, et al. (2008) Hyperoxia prevents exercise-induced intrapulmonary arteriovenous shunt in healthy humans. J Physiol 586: 4559–4565. 37. Colley PS, Cheney FW, Hlastala MP (1979) Ventilation-perfusion and gas- exchange effects of sodium nitroprusside in dogs with normal and edematous lungs. Anesthesiology 50: 489–495. y J y 13. Laurie SS, Elliott JE, Goodman RD, Lovering AT (2012) Catecholamine- induced opening of intrapulmonary arteriovenous anastomoses in healthy humans at rest. J Appl Physiol. g gy 38. Dantzker DR, Lynch JP, Weg JG (1980) Depression of cardiac-output is a mechanism of shunt reduction in acute respiratory failure Chest 77: 636–642. J pp y 14. Lovering AT, Elliott JE, Beasley KM, Laurie SS (2010) Pulmonary pathways and mechanisms regulating transpulmonary shunting into the general circula- tion: an update. Injury 41 Suppl 2: S16–23. 39. Author Contributions Conceived and designed the experiments: MB SP ME EF JJ. Performed the experiments: MB ME EF JJ AD. Analyzed the data: MB SP ME EF JJ AD. Contributed reagents/materials/analysis tools: SP. Wrote the paper: MB SP ME EF JJ AD. July 2014 \| Volume 9 \| Issue 7 \| e101146 PLOS ONE \| www.plosone.org 7 IPAVs Recruitment in Hypoxia July 2014 \| Volume 9 \| Issue 7 \| e101146 IPAVs Recruitment in Hypoxia References Motley HL, Cournand A, Werko L, Himmelstein A, Dresdale D (1947) The influence of short periods of induced acute anoxia upon pulmonary artery pressures in man. Am J Physiol 150: 315–320. p J p 24. Knudson RJ, Lebowitz MD, Burrows B (1980) The maximal expiratory flow- volume curve-Normal standards, variability and effects of age. American Review of Respiratory Disease 122: 990–991. 54. Kelman GR (1966) Digital computer subroutine for the conversion of oxygen tension into saturation. J Appl Physiol 21: 1375–1376. PLOS ONE \| www.plosone.org July 2014 \| Volume 9 \| Issue 7 \| e101146 July 2014 \| Volume 9 \| Issue 7 \| e101146 8
https://openalex.org/W2787613942	http://eprints.whiterose.ac.uk/128132/1/oe-26-3-3188%20%281%29.pdf	English	null	Quantitative thermal imaging using single-pixel Si APD and MEMS mirror	Optics express	2,018	cc-by	6,213	Article: https://doi.org/10.1364/OE.26.003188 https://doi.org/10.1364/OE.26.003188 Reuse This article is distributed under the terms of the Creative Commons Attribution (CC BY) licence. This licence allows you to distribute, remix, tweak, and build upon the work, even commercially, as long as you credit the authors for the original work. More information and the full terms of the licence here: https://creativecommons.org/licenses/ This is a repository copy of Quantitative thermal imaging using single-pixel Si APD and MEMS mirror. White Rose Research Online URL for this paper: http://eprints.whiterose.ac.uk/128132/ Version: Published Version Version: Published Version Article: Hobbs, M.J., Grainger, M.P., Zhu, C. et al. (2 more authors) (2018) Quantitative thermal imaging using single-pixel Si APD and MEMS mirror. Optics Express, 26 (3). pp. 3188-3198. ISSN 1094-4087 MATTHEW J. HOBBS, MATTHEW P. GRAINGER, CHENGXI ZHU, CHEE HING TAN, AND JONATHAN R. WILLMOTT* Department of Electronic and Electrical Engineering, The University of Sheffield, Sheffield, S1 4DE, UK Abstract: Accurate quantitative temperature measurements are difficult to achieve using focal-plane array sensors. This is due to reflections inside the instrument and the difficulty of calibrating a matrix of pixels as identical radiation thermometers. Size-of-source effect (SSE), which is the dependence of an infrared temperature measurement on the area surrounding the target area, is a major contributor to this problem and cannot be reduced using glare stops. Measurements are affected by power received from outside the field-of-view (FOV), leading to increased measurement uncertainty. In this work, we present a micromechanical systems (MEMS) mirror based scanning thermal imaging camera with reduced measurement uncertainty compared to focal-plane array based systems. We demonstrate our flexible imaging approach using a Si avalanche photodiode (APD), which utilises high internal gain to enable the measurement of lower target temperatures with an effective wavelength of 1 µm and compare results with a Si photodiode. We compare measurements from our APD thermal imaging instrument against a commercial bolometer based focal-plane array camera. Our scanning approach results in a reduction in SSE related temperature error by 66 °C for the measurement of a spatially uniform 800 °C target when the target aperture diameter is increased from 10 to 20 mm. We also find that our APD instrument is capable of measuring target temperatures below 700 °C, over these near infrared wavelengths, with D* related measurement uncertainty of ± 0.5 °C. Published by The Optical Society under the terms of the Creative Commons Attribution 4.0 License. Further distribution of this work must maintain attribution to the author(s) and the published article’s title, journal citation, and DOI. OCIS codes: (040.1345) Avalanche photodiodes (APDs); (110.6820) Thermal imaging; (120.6780) Temperature. Takedown If you consider content in White Rose Research Online to be in breach of UK law, please notify us by emailing eprints@whiterose.ac.uk including the URL of the record and the reason for the withdrawal request. eprints@whiterose.ac.uk https://eprints.whiterose.ac.uk/ eprints@whiterose.ac.uk https://eprints.whiterose.ac.uk/ eprints@whiterose.ac.uk https://eprints.whiterose.ac.uk/ Vol. 26, No. 3 \| 5 Feb 2018 \| OPTICS EXPRESS 3188 References and links 1. J. Dixon, “Radiation thermometry,” J. Phys. E Sci. Instrum. 21(5), 425–436 (1988). 2. P. Saunders, Radiation Thermometry: Fundamentals and Applications in the Petrochemical Industry (SPIE Ppress, 2007). 3. H. W. Yoon and G. P. Eppeldauer, “Measurement of thermal radiation using regular glass optics and short-wave infrared detectors,” Opt. Express 16(2), 937–949 (2008). p p ( ) ( ) 4. J. Spannar, P. Wide, and B. Sohlberg, “A method for measuring strip temperature in the steel industry,” IEEE Trans. Instrum. Meas. 51(6), 1240–1245 (2002). ( ), ( ) 5. F. J. Rodriguez, F. J. Meca, J. A. Jimenez, and E. J. Bueno, “Monitoring and quality improvement of pharmaceutical glass container’s manufacturing process,” IEEE Trans. Instrum. Meas. 57(3), 584–590 (2008). 6 A Bendada K Cole M Lamontagne and Y Simard “Infrared radiometry using a dielectric-silver-coated ( ) ( ) 5. F. J. Rodriguez, F. J. Meca, J. A. Jimenez, and E. J. Bueno, “Monitoring and quality improvement of pharmaceutical glass container’s manufacturing process,” IEEE Trans. Instrum. Meas. 57(3), 584–590 (2008). 6. A. Bendada, K. Cole, M. Lamontagne, and Y. Simard, “Infrared radiometry using a dielectric-silver-coated hollow glass waveguide for polymer processing,” Infrared Phys. Technol. 45(1), 59–68 (2004). g g p y p g y 7. Q. He, Z. Su, Z. Xie, Z. Zhong, and Q. Yao, “A Novel Principle for Molten Steel Level Measurement in Tundish by Using Temperature Gradient,” IEEE Trans. Instrum. Meas. 66(7), 1809–1819 (2017). y g p 8. R. Usamentiaga, J. Molleda, D. F. Garcia, J. C. Granda, and J. L. Rendueles, “Temperature measurement of molten pig iron with slag characterization and detection using infrared computer vision,” IEEE Trans. Instrum. Meas. 61(5), 1149–1159 (2012). 8. R. Usamentiaga, J. Molleda, D. F. Garcia, J. C. Granda, and J. L. Rendueles, “Temperature measurement of molten pig iron with slag characterization and detection using infrared computer vision,” IEEE Trans. Instrum. Meas. 61(5), 1149–1159 (2012). 9. J. Envall, S. Mekhontsev, Y. Zong, and L. Hanssen, “Spatial scatter effects in the calibration of IR pyrometers and imagers,” Int. J. Thermophys. 30(1), 167–178 (2009). 9. J. Envall, S. Mekhontsev, Y. Zong, and L. Hanssen, “Spatial scatter effects in the calibration of IR pyrometers and imagers,” Int. J. Thermophys. 30(1), 167–178 (2009). g p y ( ) ( ) 10. G. Grgić and I. Pušnik, “Analysis of thermal imagers,” Int. J. Thermophys. 32(1–2), 237–247 (2011). Hamamatsu Photonics, “Si photodiodes - S1087/S1133 series,” http://www.hamamatsu.com/resources/pdf/ssd/s1087_etc_kspd1039e.pdf. Last Accessed: August, 2017. 22. Hamamatsu Photonics, “Si photodiodes - S1087/S1133 series,” http://www.hamamatsu.com/resources/pdf/ssd/s1087_etc_kspd1039e.pdf. Last Accessed: August, 2017. 22. Hamamatsu Photonics, Si photodiodes S1087/S1133 series, http://www.hamamatsu.com/resources/pdf/ssd/s1087_etc_kspd1039e.pdf. Last Accessed: August, 2017. 23. Hamamatsu Photonics, “Si APD - S9251/S12092 series,” http://www.hamamatsu.com/resources/pdf/ssd/s9251_etc_k 23. Hamamatsu Photonics, “Si APD - S9251/S12092 series,” http://www.hamamatsu.com/resources/pdf/ssd/s9251_etc_kapd1013e.pdf. Last Accessed: August, 2017. 23. Hamamatsu Photonics, “Si APD - S9251/S12092 series,” http://www.hamamatsu.com/resources/pdf/ssd/s9251_etc_kapd1013e.pdf. Last Accessed: August, 2017. 24 h d C h “ l i i f h lib i f i f d ” l i 31(1) 2 http://www.hamamatsu.com/resources/pdf/ssd/s9251_etc_kapd1013e.pdf. Last Accessed: August, 2017. p p p p g 24. J. Hahn and C. Rhee, “Interpolation equation for the calibration of infrared pyrometers,” Metrologia 31(1), 27– 32 (1994). 24. J. Hahn and C. Rhee, “Interpolation equation for the calibration of infrared pyrometers,” Metrologia 31(1), 27– 32 (1994). ( ) 25. J. Bezemer, “Spectral sensitivity corrections for optical standard pyrometers,” Metrologia 10(2), 47–52 (19 25. J. Bezemer, “Spectral sensitivity corrections for optical standard pyrometers,” Metrologia 10(2), 47–52 ( Spectral sensitivity corrections for optical standard pyrometers 26. P. Saunders, “Uncertainty Arising from the Use of the Mean Effective Wavelength in Realizing ITS-90,” in AIP Conference Proceedings (AIP, 2003), pp. 639–644. 27. A. Whittam, R. Simpson, and H. McEvoy, “Performance tests of thermal imaging systems to assess their suitability for quantitative temperature measurements,” in 12th International Conference on Quantitative InfraRed Thermography (QIRT, 2014). 27. A. Whittam, R. Simpson, and H. McEvoy, “Performance tests of thermal imaging systems to assess their suitability for quantitative temperature measurements,” in 12th International Conference on Quantitative InfraRed Thermography (QIRT, 2014). References and links 11 H B d i d G G l h “C lib i f l d h l i f d ” J S S S 4(1) 18 g p y 10. G. Grgić and I. Pušnik, “Analysis of thermal imagers,” Int. J. Thermophys. 32(1–2), 237–247 (2011). 11 H B d i d G G l h “C lib i f l d h l i f d ” J S S S 4(1 11. H. Budzier and G. Gerlach, “Calibration of uncooled thermal infrared cameras,” J. Sens. Sens. Syst. 4(1), 187– 197 (2015). 11. H. Budzier and G. Gerlach, “Calibration of uncooled thermal infrared cameras,” J. Sens. Sens. Syst. 4(1), 187– 197 (2015). #305917 https://doi.org/10.1364/OE.26.003188 Received 14 Sep 2017; revised 8 Dec 2017; accepted 20 Dec 2017; published 30 Jan 2018 Journal © 2018 Vol. 26, No. 3 \| 5 Feb 2018 \| OPTICS EXPRESS 3189 12. M. F. Duarte, M. A. Davenport, D. Takhar, J. N. Laska, T. Sun, K. F. Kelly, and R. G. Baraniuk, “Single-pixel imaging via compressive sampling,” IEEE Signal Process. Mag. 25(2), 83–91 (2008). g g p p g, g g ( ), ( ) 13. M.-J. Sun, M. P. Edgar, D. B. Phillips, G. M. Gibson, and M. J. Padgett, “Improving the signal-to-noise ratio of single-pixel imaging using digital microscanning,” Opt. Express 24(10), 10476–10485 (2016). 14. R. Usamentiaga, D. F. Garcia, and J. Molleda, “Uncertainty analysis in spatial thermal measurements using infrared line scanners,” IEEE Trans. Instrum. Meas. 57(9), 2074–2082 (2008). 15. L. Li, V. Stankovic, L. Stankovic, L. Li, S. Cheng, and D. Uttamchandani, “Single pixel optical imaging using a scanning MEMS mirror,” J. Micromech. Microeng. 21(2), 025022 (2011). 16. W. Aenchbacher and T. Kurzweg, “Single-pixel, MEMS scanning mirror camera,” in IEEE Photonic Society 24th Annual Meeting (IEEE, 2011), pp. 849–850. 17. R. Moss, P. Yuan, X. Bai, E. Quesada, R. Sudharsanan, B. L. Stann, J. F. Dammann, M. M. Giza, and W. B. Lawler, “Low-cost compact MEMS scanning LADAR system for robotic applications,” Proc. SPIE 8379, 837903 (2012). 18. V. Milanović, A. Kasturi, J. Yang, and F. Hu, “Closed-Loop Control of Gimbal-less MEMS Mirrors for Increased Bandwidth in LiDAR Applications,” Proc. SPIE 10191, 101910N (2017). 18. V. Milanović, A. Kasturi, J. Yang, and F. Hu, “Closed-Loop Control of Gimbal-less M Increased Bandwidth in LiDAR Applications,” Proc. SPIE 10191, 101910N (2017). g p Increased Bandwidth in LiDAR Applications,” Proc. SPIE 10191, 101910N (2017). 19. M. Hobbs, C. Tan, and J. Willmott, “Evaluation of phase sensitive detection method photodiode for radiation thermometry,” J. Instrum. 8(3), P03016 (2013). 20. G. R. Peacock, “A review of non-contact process temperature measurements in steel manufacturing,” Proc. SPIE 3700, 171–189 (1999). 21. L. Chen, H. Gao, Y. Zhao, Y. Yu, and D. Wang, “Temperature error in casting radiation thermometry caused by emissivity error and measurement wavelength,” in ICIEA 2008: 3rd IEEE Conference on Industrial Electronics and Applications (IEEE, 2008), pp. 1047–1049. 21. L. Chen, H. Gao, Y. Zhao, Y. Yu, and D. Wang, “Temperature error in casting radiation thermometry caused by emissivity error and measurement wavelength,” in ICIEA 2008: 3rd IEEE Conference on Industrial Electronics and Applications (IEEE, 2008), pp. 1047–1049. pp ( , ), pp 22. Hamamatsu Photonics, “Si photodiodes - S1087/S1133 series,” http://www.hamamatsu.com/resources/pdf/ssd/s1087 etc kspd1 pp ( , ), pp 22. 1. Introduction Thermal imaging cameras require more steps in the calibration process, with each individual pixel requiring calibration and unresponsive pixels being accounted for [11]. Other concerns include inter-pixel crosstalk and non-uniformities across the array, both in terms of within the semiconductor material, electronic noise or the ambient temperature. The price of focal-plane array based thermal imaging cameras increases dramatically as the operating wavelength, Ȝop, increases (especially if cooling is required). Cheaper uncooled thermal detector based arrays can be used at longer wavelengths, at the expense of lower sensitivity and lower speed operation. Single-pixel imaging has been explored for different applications using digital micromirror devices (DMDs) utilising compressive sampling algorithms [12,13]. Unfortunately, availability of window materials limits their operation to Ȝop < 2.5 µm. An approach commonly used in industry is that of rotating mirror scanners which rapidly move the measurement FOV across the target in 1-dimension, and are particularly suited to moving targets, such as on a conveyor belt or for metal strips being rolled [14]. They offer advantages over conventional focal-plane array based imaging cameras: reduced SSE, simplified calibration and reduced cost for longer wavelength operation. A new approach to mirror scanning in recent years is the use of microelectromechanical systems (MEMS) mirror based imaging, which has been previously demonstrated using visible light [15,16] and LiDAR [17,18]. Such mirrors offer high speed 2-dimensional beam steering by devices which can be fabricated using low cost Si wafers. Limitations in technology constrains MEMS mirrors to be comparatively small; typically only a few millimetres in diameter. The power reaching the detector within a MEMS based scanning system is, in turn, limited by the small size of currently available mirrors. This vignettes the optical system or leads to a slow optical system with a large f-number, therefore, limiting the lowest temperature which can be measured by the instrument. High resolution imaging (for both scanned imagers and focal-plane array imagers), requires small active area detectors or pixels. However, use of smaller area detectors leads to comparatively reduced irradiance at the detector surface, also limiting the lowest temperature which can be measured by the instrument with acceptable signal-to-noise-ratio (SNR). We have previously demonstrated a Si avalanche photodiode (APD) as an alternative detector to a Si photodiode for radiation thermometry operating at Ȝop = 1 ȝm, successfully demonstrating that the high internal gain of the APD allows lower target temperatures to be measured [19]. 1. Introduction Radiation thermometers are commonly used to measure the temperature of target objects by measuring their emitted infrared radiation [1–3]. They can be used for industrial processing applications, for instance for various metals, glass and plastics during manufacture [4–6], where accurate temperature measurement is required for quality control and energy efficiency. The non-contact measurement nature of the radiation thermometer allows it to be easily moved to measure the temperature of different points on the target by re-sighting the instrument, with both fixed and portable instruments available. Compared to contact temperature measurements, it offers the added benefits of higher speed operation whilst eliminating the risk of contamination of the target object. Taking the non-contact measurement approach further, the use of infrared detector arrays within thermal imaging cameras allow for measurement over a wider area, enabling a temperature map to be produced without having to move the sighting position [7,8]. The use of thermal imaging cameras is particularly helpful for qualitative thermal imaging applications, although they are relatively expensive. For example, they can be used to detect spatial temperature variations across the target, which could indicate problems within the manufacturing process. However, thermal imaging cameras have disadvantages compared to quantitative radiation thermometry, when measurement of target temperature with low uncertainty is required. Quantitative thermal imaging allows SI unit traceable temperature measurements with defined or quantifiable uncertainty made on a grid across the camera scene. This is in contrast to thermal imaging cameras that provide qualitative temperature maps that can distinguish between different gradients of temperature. Under the precise methodology in which they were calibrated they will provide accurate temperature measurement; however away from those experimental conditions they provide only indicative Vol. 26, No. 3 \| 5 Feb 2018 \| OPTICS EXPRESS 3190 temperature measurements. A significant problem within thermal imaging cameras is the size- of-source effect (SSE), which arises due to imperfections within the optical system leading to reflections and scattering. Each pixel receives power over a wider area of the target than the camera was designed for [9,10]. Glare stops may reduce the problem, though this approach is far less effective than for single-pixel systems where all but the single-pixel surface can be made highly absorbing. With an array of pixels, the measured temperature becomes dependent upon the size of the target or objects next to the target. uniform 800 °C target when the target aperture diameter is increased from 10 to 20 mm. Our instrument is capable of measuring a target temperature below 700 °C with a noise performance of better than ± 0.5 °C. To the best of our knowledge, this is the first use of a Si APD (or any APD) for thermal imaging for quantitative temperature measurement. 1. Introduction Therefore a Si APD would appear to be an ideal candidate for single-pixel thermal imaging with Ȝop = 1 µm. Radiation thermometry with Ȝop = 1 ȝm is commonly used for general purpose high temperature (> 600 °C) applications such as in the steel industry [20], where its spectral response is within the range of the corresponding blackbody spectra at such temperatures. Operating at this short wavelength also has the advantage that it will result in reduced error in the measured temperature caused by unknown target emissivity compared to longer wavelength operation [21]. In this work, we present a 1 µm MEMS mirror based single-pixel thermal imaging camera for quantitative thermal imaging, comparing both a Si APD and a Si photodiode as the chosen sensor. The comparison enables a demonstration of the benefits provided by a Si APD for single-pixel thermal imaging over a Si photodiode. To further demonstrate the benefits of this imaging approach for quantitative thermal imaging, comparison is also made with a commercial bolometer based focal-plane array imager. Our scanning approach results in a reduction in SSE related temperature error by 66 °C for the measurement of a spatially uniform 800 °C target when the target aperture diameter is increased from 10 to 20 mm. Our instrument is capable of measuring a target temperature below 700 °C with a noise performance of better than ± 0.5 °C. To the best of our knowledge, this is the first use of a Si APD (or any APD) for thermal imaging for quantitative temperature measurement. Vol. 26, No. 3 \| 5 Feb 2018 \| OPTICS EXPRESS 3191 Vol. 26, No. 3 \| 5 Feb 2018 \| OPTICS EXPRESS 3191 2. Experimental design and procedures The experimental setup for the MEMS single-pixel thermal imaging camera is shown in Fig. 1. Fig. 1. MEMS single-pixel thermal imaging setup diagram. Fig. 1. MEMS single-pixel thermal imaging setup diagram. The instrument generates a bidirectional raster scanning pattern of the single-pixel’s FOV. The FOV is scanned across 320 horizontal positions and 78 vertical positions with the signal measured by the detector at each scan position or ‘pixel’. The signal is amplified by a transimpedance amplifier (TIA) and logged by a MATLAB controlled digital multimeter (DMM). In this work, we compared a Si photodiode and a Si APD, using Hamamatsu S1133- 01 and Hamamatsu S12092-2, respectively [22,23]. The TIA circuit was of conventional configuration, and comprised a Texas Instruments OPA657 op-amp and a resistor-capacitor (RC) feedback network (2 Mȍ and 1.8 pF), giving a circuit rise time of ~8 µs. The TIA circuit was followed by a first order RC filter (1 kȍ and 1 nF) for additional noise rejection. In order to supply a constant high voltage to the APD, a Laser Components ABC550-06 biasing module was used. For data capture, the voltage measurement for each pixel takes approximately 4.4 ms (of which 4 ms is the integration time of the DMM); this is significantly slower than the TIA circuit response time. The full image can be captured within approximately 110 seconds; with the MEMS mirror positions pre-loaded and stored locally within the flash memory of the MEMS controller to avoid adding additional time to the image capture. This slow acquisition time could be improved by implementing a faster method of analogue-to-digital conversion. A 0.2 mm field stop was located in front of the Si photodiode to enable resolution and power throughput measurements to be compared with the 0.2 mm diameter active area of the APD. Under these conditions, the FOV of the single-pixel was found to be approximately 5 mm in diameter for both the Si photodiode and Si APD imagers. An RG850 daylight filter was placed in front of each detector, to reduce the effect of background light on the measurement and to define the short wavelength spectral response of the detectors. Vol. 26, No. 3 \| 5 Feb 2018 \| OPTICS EXPRESS 3192 Planck’s law can be used to mathematically model the radiant exitance from a blackbody as a function of wavelength and temperature. 3. Results and discussion The assessment of our scanning thermal imaging camera was limited to those parameters most salient to quantitative radiation thermometry: SSE, noise and measurement uncertainty. This is sufficient because we are free from the usual problems of focal-plane array based thermal imaging. In order to use an APD within any optical system, the optimum biasing voltage should first be evaluated for the circuit and application. Increased APD gain will provide benefit to the system up to a point where the excess noise generated by the APD becomes larger than the noise of the TIA. For our APD-TIA circuit, a biasing voltage of approximately −220 to −225 V was found to achieve peak SNR. We have not attempted to accurately quantify the level of internal gain for the APD; this requires knowledge of the APD’s field profile to establish a unity gain reference [19]. However, with reference to the datasheet, the gain at this bias voltage should be greater than 2 orders of magnitude [23]. Note, this bias value for the APD is approximate, it is adjusted by the temperature compensation of the biasing module in order to keep the gain constant to overcome ambient temperature drift. For the Si photodiode, without internal gain, an operating bias of 0 V was used to minimise dark current. The mean output voltage is shown in Fig. 3(a) as a function of furnace temperature for both the Si photodiode and Si APD. The output voltage of the Si APD camera is significantly higher across the measurement range (by over 2 orders of magnitude) compared to the Si photodiode based instrument. This is due to the high internal gain of the APD, which provides electronic gain to the output voltage measured, by increasing the photocurrent, without concomitant increase in noise seen in other methods of providing amplification. The APD can, therefore, be used to measure lower target temperature before the noise floor is reached. For instance, if a minimum output voltage of 1 mV is specified, the APD circuit produced this at a target temperature of ~584 °C, which is ~328 °C lower than the photodiode’s minimum temperature of ~912 °C. p From Fig. 3(a) the APD output was found to obey Eq. (2) and hence is linear for measured temperatures down to 700 °C. At lower temperatures, non-linearity was observed. This calibration equation was used to calculate the output voltage expected for different target temperatures, or conversely the temperature each measured voltage corresponds to, (c). This calibration can therefore be applied to the thermal image to produce a temperature map showing how the measured temperature varies across the image, (d). This calibration equation was used to calculate the output voltage expected for different target temperatures, or conversely the temperature each measured voltage corresponds to, (c). This calibration can therefore be applied to the thermal image to produce a temperature map showing how the measured temperature varies across the image, (d). The D* related temperature noise was estimated by extracting the mean and standard deviation of the temperature measurement data with the MEMS mirror stationary at its origin position whilst imaging a 10 mm target aperture. The root-mean-squared (RMS) noise of the instrument was calculated by taking 2 × the standard deviation value, which therefore represents the fluctuation in the measured temperature within ± 1 standard deviation from its mean. 2. Experimental design and procedures At wavelengths that are short compared to the peak in emission, an approximation that can be algebraically manipulated is given by the Wien law, as shown in (1). 1 2 5 ( ) exp c c L T λ λ λ −   =     (1) (1) L is the spectral radiance of the emitted radiation of an ideal blackbody, c1 is Planck’s first radiation constant, c2 is Planck’s second radiation constant, Ȝ is wavelength and T is the blackbody temperature in kelvin. From (1), the relationship between the measured output voltage, V, and the target temperature can be approximated by plotting ln(V) against 1/T [24]; V is proportional to L. The gradient of this plot enables us to calculate a useful approximation of the mean effective wavelength for the instrument, Ȝm, which is an average wavelength of the detector spectral response over the measured temperature range [25,26]. Ideally, the ln(V) against 1/T relationship would be linear in radiance and would allow a simple calibration at a single point in the instrument temperature measurement range. A non-linear response is also usable, when used with multiple calibration temperatures and a more involved calibration procedure. Si photodiodes are highly linear, which makes them ideal for calibration, whereas Si APDs as a function of radiance may be non-linear and will be a subject of our analysis. Fig. 2. Single-pixel imager calibration steps. Fig. 2. Single-pixel imager calibration steps. The basic calibration procedure for the single-pixel thermal camera is shown in Fig. 2. The output voltage as a function of furnace temperature was measured, with the dark offset voltage first measured and subtracted from each point, (a). A 10 mm target aperture was used for calibration in this work, which was approximately twice the diameter of the FOV of the single-pixel. The factor of two accounts for the design FOV and, in addition, any radiation from outside the FOV due to the SSE. The ln(V) against 1/T relationship was plotted, (b), and the gradient, m, and intercept, c, were found and used in the calibration equation, Eq. (2). 1 exp V m c T   = +     (2) (2) Vol. 26, No. 3 \| 5 Feb 2018 \| OPTICS EXPRESS 3193 3. Results and discussion The non- linearity is not due to the total optical power, as the power is weak at all measured target temperatures. However, when the temperature reduces, the profile of photons from different wavelength changes. Since avalanche gain is also dependent on carrier injection profile, the gain therefore changes with the photon injection profile leading to a non-linear response. Our results are valid over the measured target temperature range due to the calibration process. Fig. 3. (a) Mean output voltage and (b) noise for Si photodiode and Si APD imagers with furnace temperature. Vol. 26, No. 3 \| 5 Feb 2018 \| OPTICS EXPRESS 3194 Fig. 3. (a) Mean output voltage and (b) noise for Si photodiode and Si APD imagers with furnace temperature. Figure 3(b) shows the noise as a function of target temperature for both the Si photodiode and Si APD imagers. D* related noise dictates the minimum temperature that can be resolved for a temperature fluctuation and can be expressed as a temperature measurement uncertainty in degrees Celsius (e.g., within ± 0.5 °C of the measured temperature, which is a typical specification for a high quality radiation thermometer). Therefore, if the target signal measured is too weak, a reliable measure of the target temperature will not be possible. The noise of the Si APD is significantly less than that of the Si photodiode. If a ± 0.5 °C noise specification is required for the instrument, this would not be achieved by the Si photodiode over the measured temperature range. However, the Si APD would achieve the ± 0.5 °C noise specification for target temperatures of approximately 700 °C and above. The improved noise performance of the Si APD instrument is due to two reasons; the high internal gain of the APD, and its low junction capacitance compared to the Si photodiode. Circuit noise analysis calculations produced a total integrated circuit noise of 48 µV for the Si APD, and 880 µV for the Si photodiode; an improvement factor of over 18. The high capacitance of the large area Si photodiode led to increased gain at high frequencies, resulting in its increased noise. However, the factor of improvement in the noise measurement across all temperatures is greater than 18, and significantly better as the target temperature decreases. 3. Results and discussion Mid-point cross sections of the 10 mm target aperture thermal images for (a) Si photodiode and (b) Si APD imagers at a furnace temperature of 800 °C. Vol. 26, No. 3 \| 5 Feb 2018 \| OPTICS EXPRESS 3196 Vol. 26, No. 3 \| 5 Feb 2018 \| OPTICS EXPRESS 3196 Vol. 26, No. 3 \| 5 Feb 2018 \| OPTICS EXPRESS 3196 Fig. 5. Mid-point cross sections of the 10 mm target aperture thermal images for (a) Si photodiode and (b) Si APD imagers at a furnace temperature of 800 °C. Moving from the centre to the periphery of each aperture for each of the two systems, Fig. 4 shows there is a reduction in the measured temperature and some blurring in the image at its very edges. This is believed to be due to the translation of the target spot over the aperture edge, where the FOV is elongated due to the finite response time of the detection electronics. As the target aperture size increases for both imagers, so does the measured temperature. This is believed to be due to residual SSE within the instrument, and is particularly clear as we increase the size of the target aperture. For the 20 mm target aperture, we measured approximately 10 °C increase in measured temperature compared with the 10 mm target aperture, which corresponds to 0.1% of the imager’s dynamic range. p p g y g The primary motivation behind this work is to improve the accuracy of quantitative thermal imaging compared to what is currently possible with conventional focal-plane array thermal imaging. Issues with thermal imaging cameras for quantitative thermal imaging, primarily due to SSE issues, are well known but not well publicised. Unfortunately, this leads to thermal imaging cameras being used for quantitative temperature measurement applications without an understanding of their limitation in providing accuracy. For instance, a thermal detector based thermal imager could be used as a ‘general purpose’ method for measuring temperature without consideration of SSE, therefore resulting in a temperature error which is often ignored by the user. In order for us to make a direct comparison with a commercial thermal imaging camera, we repeated the imaging measurements with a bolometer based thermal imaging camera. The camera was focused onto the same target apertures as before. 3. Results and discussion Therefore, even if the capacitance of the Si photodiode were to be the same as that of the APD, significant improvement would still be expected due to the high gain of the APD, especially at lower target temperatures. The improved noise and mean output voltage performance highlight both the advantage of, and requirement for, the Si APD in such an aperture limited system; the Si photodiode performance without high internal gain would not be sufficient. Fig. 4. SSE analysis for (a) Si photodiode and (b) Si APD imagers with increasing target aperture size at a furnace temperature of 800 °C. Vol. 26, No. 3 \| 5 Feb 2018 \| OPTICS EXPRESS 3195 Fig. 4. SSE analysis for (a) Si photodiode and (b) Si APD imagers with increasing target aperture size at a furnace temperature of 800 °C. Fully calibrated thermal imaging measurements are shown in Fig. 4 at a furnace temperature of 800 °C for both the Si photodiode and Si APD imagers. The imaged target aperture was increased in size to perform an SSE analysis; the raw data was calibrated at the 10 mm target aperture to provide a quantitative temperature measurement assessment. The imaged area is approximately 16 mm by 5 mm. By closely examining the images in Fig. 4, the imaged apertures do indeed correspond to the size of the target apertures used, hence demonstrating that the proof-of-concept works. Regards the temperature measurement within the image, there is a clear distinction between the Si photodiode and Si APD results. The Si APD’s image is relatively uniform in temperature, whilst the Si photodiode’s image shows a larger temperature variation. This is due to the much higher noise of the Si photodiode producing such a large fluctuation in temperature. To demonstrate the effect of this noise more clearly, cross sections were taken at the mid-points of the thermal images for the 10 mm target aperture, as shown in Fig. 5. The fluctuation in the measured temperature is clearly much less for the Si APD imager, which corresponds to its lower noise. These results highlight the importance of noise within a radiation thermometry / thermal imaging system and the benefit the Si APD provides due to its high internal gain leading to improved SNR performance. Fig. 5. 3. Results and discussion The approximate blackbody calibration furnace was set to 800 °C, and the images were cropped to show the same target area as our single-pixel camera. Results are shown in Fig. 6, with increasing target aperture size. The Si APD single-pixel camera was compared and temperature calibration was again relative to the 10 mm target aperture; cross sections of all the image mid-points are shown in Fig. 7. Although the imaging data for the bolometer camera provides faithful spatial information of the size of the target apertures, there is a large difference in the measured temperature when moving from the centre to the periphery, as is highlighted by examining the cross sections. Although this is the case, to some extent, for the Si APD single-pixel camera, it is a marked improvement over the bolometer camera. This SSE problem with the focal-plane array is particularly striking as the target aperture is increased in size. For the 20 mm aperture, the measured temperature is 876 °C, therefore an error of 76 °C compared to the 10 mm target aperture, corresponding to 3.5% of its dynamic range. This is considerably more than that of the Si APD single-pixel change of 10 °C. These results highlight the problem of SSE within staring array imagers for quantitative temperature measurement, and the benefits of our single-pixel imaging approach using a Si APD. When the target aperture diameter is increased from 10 to 20 mm, our single pixel thermal camera offers 66 °C improvement in SSE related temperature error for the measurement of the 800 °C target. Our findings reflect the general understanding of this issue that has also be highlighted in investigations made by others [27]. Vol. 26, No. 3 \| 5 Feb 2018 \| OPTICS EXPRESS 3197 measurement of the 800 °C target. Our findings reflect the general understanding of this issue that has also be highlighted in investigations made by others [27]. Vol. 26, No. 3 \| 5 Feb 2018 \| OPTICS EXPRESS 3197 Vol. 26, No. 3 \| 5 Feb 2018 \| OPTICS EXPRESS 3197 measurement of the 800 °C target. Our findings reflect the general understanding of this issue that has also be highlighted in investigations made by others [27]. measurement of the 800 °C target. Our findings reflect the general understanding of this issue that has also be highlighted in investigations made by others [27]. Fig. 6. be identified and greatly reduced. Suitable methods would be to fully enclose the setup, incorporate appropriate stops and baffles and to apply anti-reflection coating to the lens and MEMS mirror window. 4. Conclusion In this work we have demonstrated a single-pixel MEMS based thermal imaging camera for quantitative temperature mapping. Both a Si photodiode and a Si APD were assessed and compared for the application. The high internal gain of the Si APD was shown to have a clear advantage over the Si photodiode, allowing for a lower temperature target to be imaged with a much lower noise quantitative temperature measurement. The Si APD single-pixel thermal imaging camera measures a target temperature of 700 °C with a noise performance better than ± 0.5 °C. In comparison with a commercial focal-plane array bolometer based camera, the greatly reduced SSE of our Si APD imager led to significant reduction in temperature error within the quantitative temperature measurement. For a target temperature of 800 °C, our single-pixel imager was shown to offer a 66 °C improvement in the SSE related temperature error when increasing the target aperture diameter from 10 to 20 mm. With further investigations and optimisation of the SSE within the Si APD imager, we would expect to reduce the measurement uncertainties further. 3. Results and discussion SSE analysis for (a) Si APD and (b) bolometer camera imagers with increasing target aperture size at a furnace temperature of 800 °C. Fig. 7. Mid-point cross sections of the thermal images for (a) Si APD and (b) bolometer camera imagers at a furnace temperature of 800 °C. With further investigation into causes of SSE within our Si APD thermal imager, we would expect to improve the performance even further. Primarily, causes of reflections should Fig. 6. SSE analysis for (a) Si APD and (b) bolometer camera imagers with increasing target aperture size at a furnace temperature of 800 °C. Fig. 7. Mid-point cross sections of the thermal images for (a) Si APD and (b) bolometer camera imagers at a furnace temperature of 800 °C. With further investigation into causes of SSE within our Si APD thermal imager, we Fig. 7. Mid-point cross sections of the thermal images for (a) Si APD and (b) bolometer camera imagers at a furnace temperature of 800 °C. With further investigation into causes of SSE within our Si APD thermal imager, we would expect to improve the performance even further. Primarily, causes of reflections should Fig. 7. Mid-point cross sections of the thermal images for (a) Si APD and (b) bolometer camera imagers at a furnace temperature of 800 °C. With further investigation into causes of SSE within our Si APD thermal imager, we would expect to improve the performance even further. Primarily, causes of reflections should With further investigation into causes of SSE within our Si APD thermal imager, we would expect to improve the performance even further. Primarily, causes of reflections should Vol. 26, No. 3 \| 5 Feb 2018 \| OPTICS EXPRESS 3198 Engineering and Physical Sciences Research Council (EPSRC) fellowship EP/M009106/1. Funding Engineering and Physical Sciences Research Council (EPSRC) fellowship EP/M009106/1.
https://openalex.org/W2805105373	https://www.oatext.com/pdf/SCRR-2-112.pdf	English	null	The body's operating system	Surgical case reports and reviews	2,018	cc-by	2,912	h Andrew Hague* h Andrew Hague* Professor of Advanced Medicine, President, Cellsonic, Manufacturers of Medical Equipment, UK By borrowing a term from the computer industry, we find a way to understand the body. Our bodies are managed by an Operating System just as the software of a computer works with the hardware. Break the rules of that System and the body breaks down. Earth. Fifty years ago, we did not think of Operating Systems. Now the interconnectedness of things is understood, of humans in their environment and inside the humans the relationship between mind and body. The inability of humans to control themselves because they are driven by a force hidden inside their brain may seem fanciful until you go down to the Accident and Emergency Department on a Saturday night to see the victims of senseless violence. Why do they get drunk? Why is it always violence? Are they really human? This insight has come from observing cures wrought by the CellSonic VIPP machine which does not use drugs and is non- invasive. How this cure is unclear, so we presume it is triggering the immune system. That then leads us to ask how the immune system is so powerful. This approach gives us a better understanding that should guide doctors and regulators. Step back from the chaos and blood. Central to the Operating System is belonging. We are born to parents and belong to a family. Criminologists show that most criminals grew up without love [2]. The family is the essential support for all humans. Families live in communities which in turn associate with other groups and this is where the relationships, to my mind, can go wrong. In the world today, we see countries and parts of countries wanting to be independent, to break off relationships. It is xenophobia on a national scale. Why the aggression and hatred? In sport, competition should be friendly but is often not. Aggression is built into our brains. The Operating System triggers violence. Is this a survival mechanism dating back to when we were animals competing with tigers and hyenas for food? Why did it not become redundant like our monkey tails or the ability to breath underwater. We worry. Do other animals? For sure they display anxiety and fear but do not go to the extent of humans to manage problems. If we did not worry, there would be no doctors. h Andrew Hague* Thanks to the brain mutations that occurred about 60,000 years ago, homo sapiens became more manipulative than other animals. Even so, it is only recently, within the last few centuries, that the specialist function of medical doctors became established. Prior to that, humans lived and died as do most other animals. The speed of understanding of the body is rapid and getting close to the stage where death could be prevented. Many people believe this is a desirable goal and they have surely not worked out the consequences [1]. It highlights the first rule of the operating system for all creatures, that the species takes priority over the individual. Can this violence be bred out of us? Ask your ethics committees if you can run a trial on selective breeding of humans to find a cure for violence. Humans devote great resources to healing individuals and they are applied right up to death. In other animals we see abandonment. It raises the question of whether we need doctors, but this is not an acceptable question to any individual who is suffering. They want help and interestingly their family and friends also want them to get help. This is another unique characteristic of humans. It is built into our Operating System, hard wired as some would say. Ensuring it is applied is our guilt and fear; when our time comes we too want to be helped. Fortunately, not everyone takes fear of strangers and anger to extremes. What I see on my travels are parents each from different continents with bright children. I listen to them in the queues at the airport. In a restaurant I may be eavesdropping on the family, parents of different colours and face shapes, pretty children with a blend of both and conversations that put the Saturday night brawlers to shame. I have hope. Humans can be delightful. They can all love each other. Marry someone from a different continent. The further apart the gene pool, the better the offspring. These systems over which we have no control determine the role of medicine and its practitioners. They come as part of the package developed during the mutation of the brain. We must also understand that the brain is not static. It adapts to circumstances and as change occurs, usually caused by humans, the human brain evolves, always selfishly putting the individual before the species. Surgical Case Reports and Reviews Surgical Case Reports and Reviews ISSN: 2516-1806 Commentary ISSN: 2516-1806 Correspondence to: Andrew Hague, Professor of Advanced Medicine, President, Cellsonic, Manufacturers of Medical Equipment, UK, Tel: +1 315 210 6307; E-mail: cellsonic.beauty@gmail.com Received: February 05, 2018; Accepted: February 23, 2018; Published: February 26, 2018 Surg Case Rep Rev, 2018 The body’s operating system h Andrew Hague* h Andrew Hague* What I want to see are statistics correlating sleep with Alzheimer’s disease. There is a hypothesis that sleep cleans the detritus from the brain that would otherwise clog up the cells. The pressure to work and so-called play reduces the time in darkness needed to re-charge the brain. All this sleep is fundamental to the Operating System, so damage caused by breaking the rules cannot be repaired with drugs; the rules have to be obeyed [6]. Teenagers who sleep 12 hours are letting their growing bodies develop. Chastising parents accusing them of being lazy are causing problems. If those same teenagers, then indulge in alcohol and narcotics then they are certainly asking for trouble. The Operating System never wants the brain to be detached. Contentment and pleasure is sufficient. Acting the fool and raving is the behaviour of the deranged who before long will have a body that does not perform without pain. Using Appleton’s figures, Otto Schuman in Germany calculated that there is an electro-magnetic field of 7.83 hertz on the Earth. It is called the Schuman Resonance [4]. It means that all living things are affected by electrical and magnetic influences. Living cells have evolved in the presence of this field. Bradford University has an anechoic chamber from which all electrical fields are blocked. Spend a few minutes inside it and you feel weird, almost ill. All corrective measures (medicine) have to take the electrical properties into account and any oncologist who does not understand this can only kill and killing is not curing [5]. Is there a link between aggression and alcohol detachment? Is this the flaw in our constitution? Is further evolution required? Cancer is the replication of mutated cells. Mutations are happening all the time in all of us. To cure cancer, it is necessary to stop the cell mutating which means changing the behaviour of the cell, not necessarily killing the cell. No drugs can do it, they are only able to kill. The body has a built-in system to remove mutant cells. It is needed because most organs in the body replace their cells regularly, the process of cell division. Old species such as ants and sharks have stable organs, but humans are still settling down. The immune system is constantly searching for aberrant cells and will eliminate them when it can. A weak immune system cannot perform this function. h Andrew Hague* The Operating System that makes us fight in the false belief that it protects us has to be overcome. As doctors, you cannot do much about it except recognise it for what it is, a built-in fault that is inappropriate today. In 1972 I was cycling through Holland on my way to Rome and caught up with an entomology student. I asked him how mosquitos served humans. Remember that in those days we had a view of the world in which everything was for the benefit of humans. We had a right to interfere with nature to make it suit our purpose. Rivers were diverted and dammed. Forests cut down and wild animals slaughtered. The concept of a chosen race still haunted Europe. I was therefore surprised by his answer, that the purpose of mosquitos could be to control humans. We certainly are the most damaging species ever known on Volume 2(1): 1-3 doi: 10.15761/SCRR.1000112 Surg Case Rep Rev, 2018 Hague A (2018) The body’s operating system Hague A (2018) The body’s operating system many were rulers due to their physical prowess. Today’s (democratic) rulers are assessed by their intellect, not muscles. Long before humans evolved, there was a more profound condition that affects all living things; the formation of the Earth with an atmosphere and around it the ionosphere. What came into being millions or billions of years ago affects us today. Understanding it has come about in my lifetime. In the 1950s, at school in Bradford in England, our maths teacher referred to the Appleton Layer, something in the ionosphere of which he was very proud because a Bradfordian called Edward Appleton [3] had discovered it by mathematical calculation and showed that it affected us. It was not until last year when in conversation with Professor Simon Shepherd at Bradford University that I remembered the Appleton Layer and suddenly that piece of jig saw added to the picture and I saw a fundamental cause of the body’s Operating System. Our bodies struggle when they are only exercised for a few minutes a day. Without exercise, it is the brain that consumes most calories, so concentration leads to hunger with inadequate muscles allowing sugars to damage the pancreas. The other problem is sleep. Apparently civilised people, usually meaning urbanised, get less sleep than primitives where there are no power lines or concrete. h Andrew Hague* Vitamins, lots of sleep and exercise all help the immune system. It diminishes with age, we suspect, so a patient cured of cancer may still have an immune system that fails to manage the cell replication process leaving mutant cells to multiply. This need not be a problem because further CellSonic treatments taking only a few minutes will stop the replication of the mutant cells. It would not be a case of the cancer having returned; it is just that the original conditions apply, and the immune system has to be supported. All this is part of the body’s Operating System. My understanding of the Operating System comes from observing and deducing what is happening. I have refrained from trying to guess. Acceptance does not diminish the intellect, it is pragmatic. An explanation can follow. Priority is to know what is going on. Modern doctors struggle when faced with unexplained facts. Did they refuse to buy a bicycle when the shopkeeper could not explain how it would balance for mile after mile on two wheels? I am sure they never asked because they had seen bicycles travelling along the street without falling down. Why then do many countries refuse to allow a cure known to work until it has been explained in molecular detail. By law, patients are forced to suffer amputation to protect the smug ignorance of bureaucrats and politicians. The CellSonic VIPP technology (very intense pressure pulses) non-invasively achieves many cures without side effects and no drugs are used. It may be many years before researchers can explain the phenomenon but that should not prevent patients being cured. On a recent tour, I met a patient cured of liver cancer that had spread throughout her body and she had declared herself cured three days after the first treatment. Blood tests and scans later confirmed her prognosis. The level at which the immune system kicks in, meaning when it is started, is determined by the Operating System and was set during evolution and probably earlier before the brain mutation defined the species as homo sapiens. Life back then was arduous by today’s standards. Survival depended on hunting for and gathering food. It was all muscle action. With the new brain, ingenuity and tools were possible and the desire was and is always for less or easier muscle action. Hague A (2018) The body’s operating system h Andrew Hague* Only in the last few years, still within my lifetime, has it become possible to exist with minimum movement. In civilised society, we have to make an effort to exercise. Writing this paper has taken three hours typing plus three hours cycling. Thinking occurs whilst pedalling, not sitting at the keyboard. There is a correlation between lack of exercise and increases of disease and cancer. The body’s Operating System is in danger of being contravened. Another lady at the same CellSonic Clinic was down to one kidney dialysis a week from three after regular treatments and she felt very much better. She was in her mid-sixties and did not need an explanation. All she wanted was the visit to the clinic and that made her happy. That feeling of happiness, by the way, is another observation which is always reported, and I have asked for it to be recorded in patient’s logs because it is a feature of the treatment. Strange but true. Have we discovered a cure for depression? The third patient to whom I was introduced was equally remarkable. Her spinal cord had been severed four years before. After a series of CellSonic treatments to her spine and both legs down to the ankles she is regaining control of her legs, bowel and bladder. With delight she showed me a video of her swimming and she was moving her To know at what level of exertion the immune system begins to operate, look only at life in medieval times. You don’t need to go back to the Palaeolithic. Even monarchs and rulers had to exert themselves and Volume 2(1): 2-3 Volume 2(1): 2-3 doi: 10.15761/SCRR.1000112 Surg Case Rep Rev, 2018 Hague A (2018) The body’s operating system legs. New nerves are forming most likely as a network by-passing the severed main nerves. Whether she will completely recover back to normal remains to be seen but this is good medicine by any measure. She is not being drugged so she has no side effects and there is no surgery. Copyright: ©2018 Hague A. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. References 1. The consequences of curing death are dire. Our species survival depends on a steady evolution to produce individuals with immunities in the world of bacteria in and around us. Break that interaction and the immunity is broken resulting in mass extinction. Cellsonic. 1. The consequences of curing death are dire. Our species survival depends on a steady evolution to produce individuals with immunities in the world of bacteria in and around us. Break that interaction and the immunity is broken resulting in mass extinction. Cellsonic. 1. The consequences of curing death are dire. Our species survival depends on a steady evolution to produce individuals with immunities in the world of bacteria in and around us. Break that interaction and the immunity is broken resulting in mass extinction. Cellsonic. 2. Child care and the Growth of Love by John Bowlby My edition is in Pelican published 1965 His obituary was published in The Independent 5th September 1990 2. Child care and the Growth of Love by John Bowlby My edition is in Pelican published 1965 His obituary was published in The Independent 5th September 1990 Despite my own protestations, I still ask “how” after conceding the “what”. The sharp pulses of CellSonic are breaching the threshold of the immune system and causing a self-induced repair. I bounce this suggestion off many doctors and none have rejected the idea. My message is, trust your observations and make the patient’s health your priority. Admit that in many cases drugs and surgery do not work. The poisons versus pulsations argument is gathering force and will keep researchers active for years to come. 3. https://en.wikipedia.org/wiki/Edward_Victor_Appleton 3. https://en.wikipedia.org/wiki/Edward_Victor_Appleton 4. (1952) This global electromagnetic resonance phenomenon is named after physicist Winfried Otto Schumann who predicted it mathematically in 1952. 4. (1952) This global electromagnetic resonance phenomenon is named after physicist Winfried Otto Schumann who predicted it mathematically in 1952. 5. Review of reports of curing cancer with CellSonic VIPP machines https:// symbiosisonlinepublishing.com/palliative-medicine-care/Volume4-Issue4.php 5. Review of reports of curing cancer with CellSonic VIPP machines https:// symbiosisonlinepublishing.com/palliative-medicine-care/Volume4-Issue4.php 6. Travelling frequently, I know the problem of jet lag and the only correction is to accept that it takes one day for every hour of change. There are no drugs that will help without causing damaging side effects. 6. Surg Case Rep Rev, 2018 References Travelling frequently, I know the problem of jet lag and the only correction is to accept that it takes one day for every hour of change. There are no drugs that will help without causing damaging side effects. Copyright: ©2018 Hague A. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Volume 2(1): 3-3 Volume 2(1): 3-3 doi: 10.15761/SCRR.1000112 Surg Case Rep Rev, 2018
https://openalex.org/W4205950389	https://formative.jmir.org/2021/8/e21817/PDF	English	null	Online Search Trends Influencing Anticoagulation in Patients With COVID-19: Observational Study (Preprint)	null	2,020	cc-by	3,635	Worrall et al Worrall et al JMIR FORMATIVE RESEARCH Original Paper Online Search Trends Influencing Anticoagulation in Patients With COVID-19: Observational Study Amy P Worrall1, BA, BSc, MB BAO MCh, MSc; Claire Kelly2, MB BAO BCh, MSc; Aine O'Neill1, BSc, MB BAO BCh; Murray O'Doherty1, BA, MB BAO BCh; Eoin Kelleher3, MB BAO BCh; Anne Marie Cushen4, BSc, MSc; Cora McNally1, MB BAO BCh, MRCPI; Samuel McConkey1,5, MB BAO, MD, DMTH, MRCPI; Siobhan Glavey2, BSc, MB BAO BCh, PhD, MRCPI; Michelle Lavin2,6, MB BAO BCh, PhD, MRCPI; Eoghan de Barra1,5, BSc, MB BAO BCh, MD, MRCPI 1Department of Infectious Diseases, Beaumont Hospital, Dublin, Ireland 2Department of Haematology, Beaumont Hospital, Dublin, Ireland 3Department of Anaesthesiology, Beaumont Hospital, Dublin, Ireland 4Department of Pharmacy, Beaumont Hospital, Dublin, Ireland 5Department of International Health and Tropical Medicine, Royal College of Surgeons Ireland, Dublin, Ireland 6Irish Centre for Vascular Biology, School of Pharmacy & Biomedical Sciences, Royal College of Surgeons Ireland, Dublin, Ireland these authors contributed equally 1Department of Infectious Diseases, Beaumont Hospital, Dublin, Ireland 2Department of Haematology, Beaumont Hospital, Dublin, Ireland 3Department of Anaesthesiology, Beaumont Hospital, Dublin, Ireland 4Department of Pharmacy, Beaumont Hospital, Dublin, Ireland 5Department of International Health and Tropical Medicine, Royal College of Surgeons Ireland, Dublin, Ireland 6Irish Centre for Vascular Biology, School of Pharmacy & Biomedical Sciences, Royal College of Surgeons Ireland, Dublin, Ireland these authors contributed equally Corresponding Author: Amy P Worrall, BA, BSc, MB BAO MCh, MSc Department of Infectious Diseases Beaumont Hospital P.O. Box 1297 Beaumont Road Dublin, 9 Ireland Phone: 353 1 8093000 Email: worralap@tcd.ie Corresponding Author: Amy P Worrall, BA, BSc, MB BAO MCh, MSc Department of Infectious Diseases Beaumont Hospital P.O. Box 1297 Beaumont Road Dublin, 9 Ireland Phone: 353 1 8093000 Email: worralap@tcd.ie https://formative.jmir.org/2021/8/e21817 Methods We conducted a retrospective audit of patients who tested positive for COVID-19 by real-time reverse transcription polymerase chain reaction at Beaumont Hospital in Dublin, Ireland, a tertiary 820-bed hospital, from March 1 to May 31, 2020. The audit received approval from the audit department and the research ethics committee of Beaumont Hospital. All patients were followed until the June 20, 2020. Data regarding venous thromboembolism (VTE) risk was assessed using the Padua Prediction Score system. The appropriate prescribing of thromboprophylaxis or anticoagulation therapy within 24 hours of admission or 24 hours of a positive COVID-19 result were collected. Three local interventions in the hospital were noted as key events that influenced anticoagulation and thromboprophylaxis guidelines for patients with COVID-19. These three interventions included the following: a COVID-19 teaching session by the coagulation specialist, a consultant hematologist (attending); a hospital-wide email with VTE prophylaxis guidelines; and hospital-wide circulation of an infographic about VTE prophylaxis in patients with COVID-19 (Figure 1). The COVID-19 pandemic has rapidly evolved, spreading throughout the globe from its origin in China. The growing need to quickly report clinical findings, guidance, and recommendations became paramount as the seriousness of the situation unfolded. Chinese reports and studies were soon followed by those of our Italian colleagues. Italy became the first country to provide a European angle on treatment and management of COVID-19. Dramatic changes occurred worldwide, forcing the public into the safety of their homes. Social distancing measures, including quarantine, business closures, and travel bans, meant the ability for academic, clinical, and public discourse was limited. In order to gain information on COVID-19, the most recent updates, and suggested management, clinicians began engaging with online and social media platforms, ones that have already been described in the literature [5]. Data from Google Trends were retrieved online in comma-separated values (CSV) format and used to compile a representative trend of worldwide searches for “COVID-19” and synonyms (eg, “coronavirus” and “covid”), together with the following terms: “anticoagulation,” “VTE prophylaxis,” “thrombosis,” “clots,” and “clotting.” The search results were then summated. Data collection included results from February 29 until May 31, 2020. The data were worldwide data, rather than European or Irish Google Trends data, as Google search results at the time were very scarce in individual countries. The Google Trends data were being used to infer online activity on numerous social media platforms during the early period of the COVID-19 pandemic. Abstract Background: Early evidence of COVID-19–associated coagulopathy disseminated rapidly online during the first months of 2020, followed by clinical debate about how best to manage thrombotic risks in these patients. The rapid online spread of case reports was followed by online interim guidelines, discussions, and worldwide online searches for further information. The impact of global online search trends and online discussion on local approaches to coagulopathy in patients with COVID-19 has not been studied. Objective: The goal of this study was to investigate the relationship between online search trends using Google Trends and the rate of appropriate venous thromboembolism (VTE) prophylaxis and anticoagulation therapy in a cohort of patients with COVID-19 admitted to a tertiary hospital in Ireland. Methods: A retrospective audit of anticoagulation therapy and VTE prophylaxis among patients with COVID-19 who were admitted to a tertiary hospital was conducted between February 29 and May 31, 2020. Worldwide Google search trends of the term “COVID-19” and anticoagulation synonyms during this time period were determined and correlated against one another using a Spearman correlation. A P value of <.05 was considered significant, and analysis was completed using Prism, version 8 (GraphPad). Results: A statistically significant Spearman correlation (P<.001, r=0.71) was found between the two data sets, showing an increase in VTE prophylaxis in patients with COVID-19 with increasing online searches worldwide. This represents a proxy for online searches and discussion, dissemination of information, and Google search trends relating to COVID-19 and clotting risk, in particular, which correlated with an increasing trend of providing thromboprophylaxis and anticoagulation therapy to patients with COVID-19 in our tertiary center. Conclusions: We described a correlation of local change in clinical practice with worldwide online dialogue and digital search trends that influenced individual clinicians, prior to the publication of formal guidelines or a local quality-improvement intervention. JMIR Form Res 2021 \| vol. 5 \| iss. 8 \| e21817 \| p. 1 (page number not for citation purposes) XSL•FO RenderX XSL•FO RenderX Worrall et al JMIR FORMATIVE RESEARCH outbreak [6]. Similarly, infodemiology metrics for search trends on Google across European countries showed a strong correlation between COVID-19 cases locally and worldwide, identifying new avenues for online disease surveillance and response efforts [7]. Introduction Since late 2019, the knowledge of the clinical sequelae of COVID-19 has increased largely through the rapid dissemination of information through various information platforms. In addition to asymptomatic infection, SARS-CoV-2 can cause a broad range of symptoms, from mild coryzal symptoms to neurological and gastrointestinal presentations and, most worryingly, severe acute respiratory failure. In the early months of 2020, reports of a high incidence of COVID-19–related coagulopathy circulated online, with anecdotal evidence and case reports emerging initially. This was soon followed by marked increases in clinical thrombosis, including deep vein thromboses, pulmonary emboli [1,2], and microthrombi in pulmonary vasculature in postmortem pathological studies [3]. Clinical debate ensued surrounding the best strategy to prevent and treat COVID-19–associated coagulopathy. By the middle of April 2020, an international position paper recommended the provision of adequate thromboprophylaxis in hospitalized patients with COVID-19 [4]. Debate began on the use of therapeutic anticoagulation in the absence of confirmed thrombosis, or an intermediary dose of low-molecular-weight heparin as a thromboprophylaxis in this cohort. The risk-benefit ratio of bleeding versus clotting had to be taken into consideration, especially as some patients displayed signs of disseminated intravascular coagulopathy. However, the role of online search trends and online health information tools in the spread of this information has yet to be assessed. Here, we investigate the power of search trends as representative of discussion and dissemination of clinical information and recommendations online as well as how trending themes and discussion have influenced clinical practice in our local center. JMIR Form Res 2021 \| vol. 5 \| iss. 8 \| e21817 \| p. 2 (page number not for citation purposes) Methods Data were analyzed using Prism, version 8 (GraphPad), according to nonparametric Spearman correlations. A P value of <.05 was considered statistically significant. Infodemiology studies have already shown that search trends are useful parameters in the measurement of this pandemic; for example, search trends for symptoms correlating with disease https://formative.jmir.org/2021/8/e21817 https://formative.jmir.org/2021/8/e21817 XSL•FO RenderX XSL•FO RenderX Worrall et al JMIR FORMATIVE RESEARCH Figure 1. Local infographic used to encourage venous thromboembolism (VTE) prophylaxis in patients with COVID-19, designed by Dr Eoin Kelleher (@eoinkr) and circulated in print and on digital platforms throughout our tertiary center. Figure 1. Local infographic used to encourage venous thromboembolism (VTE) prophylaxis in patients with COVID-19, designed by Dr Eoin Kelleher (@eoinkr) and circulated in print and on digital platforms throughout our tertiary center. Figure 1. Local infographic used to encourage venous thromboembolism (VTE) prophylaxis in patients with COVID-19, designed by Dr Eoin Kelleher (@eoinkr) and circulated in print and on digital platforms throughout our tertiary center. A total of 6361 individual searches worldwide were collated from Google Trends over the study period relating to the COVID-19 pandemic and thrombosis. The percentage of patients with COVID-19 who were on appropriate anticoagulation therapy was graphed as a percentage of the total number of patients during the admission period (Figure 2 [4,8]). A statistically significant Spearman correlation (P<.001, r=0.71) was found between the two data sets. Our study demonstrated that online searches and discussion, dissemination of information, and Google search trends relating to COVID-19 and clotting risk, in particular, correlated with an increasing trend of providing thromboprophylaxis and anticoagulation therapy to patients with COVID-19 in our tertiary center. Following the publication of two major guideline papers [4,8] (Figure 2) and three local VTE interventions, there was significant improvement in VTE prophylaxis among our cohort of patients with COVID-19 infection (Figure 2). https://formative.jmir.org/2021/8/e21817 Results A total of 399 patients consecutively diagnosed with COVID-19 were reviewed during the study period. The median age of the patients was 70 years (IQR 27, range 21-99), the majority were male (247/399, 61.9%) and Caucasian (360/399, 90.2%), and 81.5% (325/399) had underlying comorbidities. On admission, patients were assessed for thrombotic and bleeding risks. The median Padua Prediction Score was 4 (IQR 4, range 1-12). A total of 14.3% of patients (57/399) were on anticoagulation therapy for thrombotic or cardiac disorders on admission. A total of 55.1% of patients (220/399) had commenced standard thromboprophylaxis doses of heparin within the first 24 hours of admission or within 24 hours of a positive COVID-19 result. A total of 89.1% of anticoagulation prescriptions (196/220) were correctly adjusted for BMI or renal function within the same admission time frame. JMIR Form Res 2021 \| vol. 5 \| iss. 8 \| e21817 \| p. 3 (page number not for citation purposes) JMIR Form Res 2021 \| vol. 5 \| iss. 8 \| e21817 \| p. 3 (page number not for citation purposes) XSL•FO RenderX JMIR FORMATIVE RESEARCH Worrall et al Figure 2. Run chart of daily online searches relating to COVID-19 and anticoagulation (blue, left axis) and the percentage of correct anticoagulation medications administered to the cohort of patients with COVID-19 (red, right axis). Black dashes indicate the publication of two major international, interim, guidance papers for patients with COVID-19 [4,8], and green dashes indicate three local venous thromboembolism (VTE) interventions in our tertiary center. tertiary center. th i fl f li h t d li i l di tertiary center. Discussion the influence from online search trends, online social media platform discussions, and dissemination of information is not the influence from online search trends, online social media platform discussions, and dissemination of information is not always reliable and can also be harmful [10]. Limitations Online learning platforms, information and news sources, as well as social media do impact and influence a subgroup of physicians that use those resources [5], and it is not outlandish to suggest that the number of clinicians engaging in online health information discussions has increased during the COVID-19 pandemic. The consequences of this can be positive; for example, the use of our local VTE prophylaxis increased significantly within our cohort prior to the three local VTE anticoagulation interventions, most likely secondary to two major interim guidance documents being published and significant international online discussion surrounding COVID-19 coagulopathy. However, as seen with the rapid and somewhat premature use of hydroxychloroquine in patients with COVID-19 or the hasty alarm surrounding ibuprofen use in COVID-19 cohorts, in an evolving pandemic, knee-jerk medical management and interventions can occur without stringent scientific or medical evidence to back up these actions. These actions are often a result of the rapid dissemination of data and research that were in the preprint stage and had not undergone full peer review [10,11]. We must acknowledge that This study is limited by the single local center analyzed in Ireland. The search terms and trends utilized were from one social media platform and can only provide a tentative inference of impact on practice. The Google Trends data were only collected on a worldwide sample, as the number of local and regional searches were too small for appropriate analysis. We also note that values from Google Trends are not absolute values of search results, but rather normalized values. However, the use of Google Trends in this study was used to represent a general trend of online discussion that occurred online during the early months of the COVID-19 pandemic, whereby online platforms, such as Google, Twitter, and Facebook, provided platforms for swift data dissemination for health practitioners [13]. Principal Findings While social media and online health information sources have almost always universally been appreciated for their potential, they have often not been successfully implemented as key components of public health strategy [12]. Only a small minority of medical and public health researchers would have actively engaged in this space in a professional capacity, as social media has been seen as a means of disseminating information rather than obtaining it [12]. The COVID-19 pandemic has reshaped the use of online platforms from online tools, search engines, and social media sites as potential tools of public health strategy, but also for disease surveillance and monitoring and, indeed, for more immediate transmission of acute changes in disease management [7]. A balance must be struck between the lethargic delay in translational medicine and the impulsive dissemination of as yet–unproven research and information dissemination, in particular during a global health crisis. Mavragani and colleagues demonstrated that Google Trends search queries correlated strongly with COVID-19 cases and COVID-19 deaths, but also that their correlations were most accurate during the initial months of a region’s outbreak [7]. Our correlation similarly reflected search trends that correlated with a change in clinical practice locally. However, in 2015, Narayanaswami et al published a longitudinal study of the effects of traditional versus novel information dissemination (online resources, social media platforms, etc) on implementing and disseminating clinical guidelines [9]. They found no additional benefit from online resources in increasing awareness and implementation of clinical guidelines, though we hypothesize that this question may need revisiting if considering the developments in online resources since 2015 and the unique global impact and increased reliance on online assets during the COVID-19 pandemic. https://formative.jmir.org/2021/8/e21817 Authors' Contributions APW, CK, AO’N, MO’D, and ML collected the data. APW and CK analyzed the data. All authors were involved in writing and reviewing the paper. EK designed and kindly provided the graphic design for the local “Think VTE Prophylaxis in Patients With COVID-19” intervention in Beaumont Hospital. Conclusions In this paper, we described a phenomenon of local change in clinical practice following worldwide online conversation and digital search trends that influenced individual clinicians before https://formative.jmir.org/2021/8/e21817 JMIR Form Res 2021 \| vol. 5 \| iss. 8 \| e21817 \| p. 4 (page number not for citation purposes) XSL•FO RenderX Worrall et al JMIR FORMATIVE RESEARCH the formation of formal clinical guidelines, with tentative natural improvement and significant improvement following a quality-improvement intervention. Conflicts of Interest None declared. JMIR Form Res 2021 \| vol. 5 \| iss. 8 \| e21817 \| p. 5 (page number not for citation purposes) References Tracking COVID-19 in Europe: Infodemiology approach. JMIR Pu 20;6(2):e18941 [FREE Full text] [doi: 10.2196/18941] [Medline: 32250957] 8. Barrett CD, Moore HB, Yaffe MB, Moore EE. ISTH interim guidance on recognition and management of coagulopathy in COVID-19: A comment. J Thromb Haemost 2020 Aug;18(8):2060-2063. [doi: 10.1111/jth.14860] [Medline: 32302462 8. Barrett CD, Moore HB, Yaffe MB, Moore EE. ISTH interim guidance on recognition and management of coagulopathy in COVID-19: A comment. J Thromb Haemost 2020 Aug;18(8):2060-2063. [doi: 10.1111/jth.14860] [Medline: 32302462] 9. Narayanaswami P, Gronseth G, Dubinsky R, Penfold-Murray R, Cox J, Bever C, et al. The impact of social media on dissemination and implementation of clinical practice guidelines: A longitudinal observational study. J Med Internet Res 2015 Aug 13;17(8):e193 [FREE Full text] [doi: 10.2196/jmir.4414] [Medline: 26272267] g; ( ) [ j ] [ ] 9. Narayanaswami P, Gronseth G, Dubinsky R, Penfold-Murray R, Cox J, Bever C, et al. The impact of social media on dissemination and implementation of clinical practice guidelines: A longitudinal observational study. J Med Internet Res 2015 Aug 13;17(8):e193 [FREE Full text] [doi: 10.2196/jmir.4414] [Medline: 26272267] 11. Carius BM, Schauer SG. Ibuprofen during the COVID-19 pandemic: Social media precautions and implications. West J Emerg Med 2020 Apr 24;21(3):497-498 [FREE Full text] [doi: 10.5811/westjem.2020.4.47686] [Medline: 32421496] 11. Carius BM, Schauer SG. Ibuprofen during the COVID-19 pandemic: Social media precautions and implications. West J Emerg Med 2020 Apr 24;21(3):497-498 [FREE Full text] [doi: 10.5811/westjem.2020.4.47686] [Medline: 32421496] 12. Keller B, Labrique A, Jain KM, Pekosz A, Levine O. Mind the gap: Social media engagement by public health researchers. J Med Internet Res 2014 Jan 14;16(1):e8 [FREE Full text] [doi: 10.2196/jmir.2982] [Medline: 24425670] 13. Chan AKM, Nickson CP, Rudolph JW, Lee A, Joynt GM. Social media for rapid knowledge dissemination: Early experience from the COVID-19 pandemic. Anaesthesia 2020 Dec;75(12):1579-1582 [FREE Full text] [doi: 10.1111/anae.15057] [Medline: 32227594] Abbreviations CSV: comma-separated values VTE: venous thromboembolism References 1. Middeldorp S, Coppens M, van Haaps TF, Foppen M, Vlaar AP, Müller MCA, et al. Incidence of venous thromboembolism in hospitalized patients with COVID-19. J Thromb Haemost 2020 Aug;18(8):1995-2002 [FREE Full text] [doi: 10.1111/jth.14888] [Medline: 32369666] 1. Middeldorp S, Coppens M, van Haaps TF, Foppen M, Vlaar AP, Müller MCA, et al. Incidence of venous thromboembolism in hospitalized patients with COVID-19. J Thromb Haemost 2020 Aug;18(8):1995-2002 [FREE Full text] [doi: 10.1111/jth.14888] [Medline: 32369666] j 2. Helms J, Tacquard C, Severac F, Leonard-Lorant I, Ohana M, Delabranche X, CRICS TRIGGERSEP Group (Clinical Research in Intensive Care and Sepsis Trial Group for Global Evaluation and Research in Sepsis). High risk of thrombosis in patients with severe SARS-CoV-2 infection: A multicenter prospective cohort study. Intensive Care Med 2020 Jun;46(6):1089-1098 [FREE Full text] [doi: 10.1007/s00134-020-06062-x] [Medline: 32367170] ( ) 3. Fox SE, Akmatbekov A, Harbert JL, Li G, Quincy Brown J, Vander Heide RS. Pulmonary and cardiac pathology in African American patients with COVID-19: An autopsy series from New Orleans. Lancet Respir Med 2020 Jul;8(7):681-686 [FREE Full text] [doi: 10.1016/S2213-2600(20)30243-5] [Medline: 32473124] 4. Spyropoulos AC, Levy JH, Ageno W, Connors JM, Hunt BJ, Iba T, Subcommittee on Perioperative‚ Critical Care Thrombosis‚ Haemostasis of the Scientific‚ Standardization Committee of the International Society on Thrombosis and Haemostasis. Scientific and Standardization Committee communication: Clinical guidance on the diagnosis, prevention, and treatment of venous thromboembolism in hospitalized patients with COVID-19. J Thromb Haemost 2020 Aug;18(8):1859-1865 [FREE Full text] [doi: 10.1111/jth.14929] [Medline: 32459046] j 5. Chan T, Trueger NS, Roland D, Thoma B. Evidence-based medicine in the era of social media: Scholarly engagement through participation and online interaction. CJEM 2018 Jan;20(1):3-8. [doi: 10.1017/cem.2016.407] [Medline: 28077195] 5. Chan T, Trueger NS, Roland D, Thoma B. Evidence-based medicine in the era of social media: Scholarly engagement through participation and online interaction. CJEM 2018 Jan;20(1):3-8. [doi: 10.1017/cem.2016.407] [Medline: 28077195] 6. Higgins TS, Wu AW, Sharma D, Illing EA, Rubel K, Ting JY, Snot Force Alliance. Correlations of online search engine trends with coronavirus disease (COVID-19) incidence: Infodemiology study. JMIR Public Health Surveill 2020 May 21;6(2):e19702 [FREE Full text] [doi: 10.2196/19702] [Medline: 32401211] 6. Higgins TS, Wu AW, Sharma D, Illing EA, Rubel K, Ting JY, Snot Force Alliance. Correlations of online search engine trends with coronavirus disease (COVID-19) incidence: Infodemiology study. JMIR Public Health Surveill 2020 May 21;6(2):e19702 [FREE Full text] [doi: 10.2196/19702] [Medline: 32401211] 7. Mavragani A. JMIR Form Res 2021 \| vol. 5 \| iss. 8 \| e21817 \| p. 6 (page number not for citation purposes) Abbreviations bbreviations CSV: comma-separated values VTE: venous thromboembolism https://formative.jmir.org/2021/8/e21817 XSL•FO RenderX Worrall et al JMIR FORMATIVE RESEARCH Edited by G Eysenbach; submitted 13.08.20; peer-reviewed by T Chan, A Mavragani; comments to author 19.09.2 received 11.04.21; accepted 11.04.21; published 31.08.21 ©Amy P Worrall, Claire Kelly, Aine O'Neill, Murray O'Doherty, Eoin Kelleher, Anne Marie Cushen, Cora McNally, Samuel McConkey, Siobhan Glavey, Michelle Lavin, Eoghan de Barra. Originally published in JMIR Formative Research (https://formative.jmir.org), 31.08.2021. This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Formative Research, is properly cited. The complete bibliographic information, a link to the original publication on https://formative.jmir.org, as well as this copyright and license information must be included. ©Amy P Worrall, Claire Kelly, Aine O'Neill, Murray O'Doherty, Eoin Kelleher, Anne Marie Cushen, Cora McNally, Samuel McConkey, Siobhan Glavey, Michelle Lavin, Eoghan de Barra. Originally published in JMIR Formative Research (https://formative.jmir.org), 31.08.2021. This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Formative Research, is properly cited. The complete bibliographic information, a link to the original publication on https://formative.jmir.org, as well as this copyright and license information must be included. https://formative.jmir.org/2021/8/e21817 https://formative.jmir.org/2021/8/e21817 XSL•FO RenderX
https://openalex.org/W2921785649	https://europepmc.org/articles/pmc6488531?pdf=render	English	null	Racial/ethnic disparities in prevalence, treatment, and control of hypertension among US adults following application of the 2017 American College of Cardiology/American Heart Association guideline	Preventive medicine reports	2,019	cc-by	11,526	A R T I C L E I N F O The 2017 American College of Cardiology/American Heart Association (ACC/AHA) Guideline for the Prevention, Detection, Evaluation, and Management of High Blood Pressure in Adults recommends reduced systolic/diastolic blood pressure (SBP/DBP) cutoffs to define hypertension (i.e., by changing these from ≥140/ 90 to ≥130/80 mmHg), including new recommendations about indications and goals of antihypertensive treatment. This study reported the differences in age-adjusted prevalence and treatment status of hypertension according to race among US adults per the 2017 ACC/AHA guideline. Keywords: Hypertension Blood pressure Blood pressure control Hypertension treatment 2017 ACC/AHA Race Racial/ethnic disparities NHANES US The National Health and Nutrition Examination Survey 2011–16 data was analyzed. The main outcomes were age-adjusted prevalence and treatment status of hypertension among adults aged ≥20 years. After prevalence estimation, other proportions were obtained. The analysis included 16,103 adults (mean age: 47.6 years, 51.8% women). The age-adjusted proportions of adults with hypertension (59.0%, 95% confidence interval [CI]: 57.4%–60.6%), treatment-eligible for hy- pertension (49.3%, 95% CI: 47.7%–50.8%), and unmet treatment goals (63.8%, 95% CI: 60.0%–67.5%) among the treated were highest among non-Hispanic blacks. A large proportion of Mexican-Americans (46.5%, 95% CI: 42.0%–51.0%) and people of other races/ethnicities (49.3%, 95% CI: 45.5%–53.0%) were not receiving treat- ment despite having indication. Non-Hispanic blacks also had the highest prevalence of stage 2 hypertension. Among all races, prevalence, treatment-eligibility, and unmet treatment goals were higher among people with older age, male gender, diabetes, higher body weight, and higher cardiovascular disease risk while the majority of younger, lower/normal body weight, or non-diabetic people were untreated despite being eligible for treat- ment. The prevalence, treatment-eligibility, and unmet goals were substantially higher among non-Hispanic blacks. Moreover, disparities exist in treatment where Mexican-Americans and people of ‘other races/ethnicities’ were largely untreated despite having indication. Contents lists available at ScienceDirect Contents lists available at ScienceDirect Racial/ethnic disparities in prevalence, treatment, and control of hypertension among US adults following application of the 2017 American College of Cardiology/American Heart Association guideline Gulam Muhammed Al Kibria⁎ University of Maryland School of Medicine, Baltimore, MD 21201, USA Gulam Muhammed Al Kibria⁎ University of Maryland School of Medicine, Baltimore, MD 21201, USA Gulam Muhammed Al Kibria⁎ https://doi.org/10.1016/j.pmedr.2019.100850 Received 11 February 2019; Received in revised form 2 March 2019; Accepted 14 March 2019 Available online 16 M arch 2019 2211-3355/ © 2019 The Author. Published by Elsevier Inc. This is an open access article under the CC BY license (http://creativecom m ons.org/licenses/BY/4.0/). Preventive M edicine Reports 14 (2019) 100850 Preventive M edicine Reports 14 (2019) 100850 Abbreviations: ACC/AHA, American College of Cardiology/American Heart Association; BP, Blood pressure; CI, Confidence interval; CKD, Chronic kidney disease; CVD, Cardiovascular disease; DBP, Diastolic blood pressure; JNC, Joint National Committee; NHANES, National Health and Nutrition Examination Survey; SBP, Systolic blood pressure; SE, Standard error Abbreviations: ACC/AHA, American College of Cardiology/American Heart Association; BP, Blood pressure; CI, Confidence interval; CKD, Chronic kidney disease CVD, Cardiovascular disease; DBP, Diastolic blood pressure; JNC, Joint National Committee; NHANES, National Health and Nutrition Examination Survey; SBP Systolic blood pressure; SE, Standard error ⁎ Corresponding author at: University of Maryland School of Medicine, Howard Hall, 133-B, 660 W. Redwood Street, Baltimore, MD 21201, USA. E-mail address: gkibria1@outlook.com. 1. Introduction Other studies also support this finding (Dorans et al., 2018; Khera et al., 2018). Although this guideline has classified a substantial proportion of adults as hypertensive who would have been known as prehypertensive per the previous guidelines, it provides a greater opportunity to take prevention and treatment measures in the early stages of hypertension (Muntner et al., 2017; Whelton et al., 2017). Additionally, adoption of the new guideline could substantially reduce the cardiovascular events resulting from hypertension (Bundy et al., 2018; Whelton et al., 2017; Yano et al., 2018). Participants reported their race/ethnicity, and were grouped as non- Hispanic whites, non-Hispanic blacks, Mexican-Americans, and other races/ethnicities. Due to a lower proportion of respondents from races/ ethnicities other than the first 3 races, all other races (i.e., non-Hispanic Asians, other Hispanics, and other races including multi-races) were grouped as other races/ethnicities. Participants also reported their age (in years) and gender. Age was stratified into 20–44, 45–54, 55–64, and ≥65 years (Muntner et al., 2017). Borderline elevated and high cho- lesterol levels were defined as 200–239 and ≥240 mg/dl cholesterol levels, respectively. If the high-density lipoprotein (HDL) was < 40 mg/ dl for men and < 50 mg/dl for women, it was categorized as low (Ostchega et al., 2018). Self-reports for diagnosis of prediabetes and diabetes were obtained. The chronic kidney disease Epidemiology (CKD-EPI) equation was used to estimate the glomerular filtration rate (GFR). A person was categorized as CKD if the albumin-creatinine ratio was ≥30 mg/g or the GFR was < 60 ml/min per 1.73 m2 (Levey et al., 2009). ‘Weight in kilograms’ was divided by ‘height in meters squared’ to obtain body mass index (BMI). The BMI cutoffs to define under-/ normal weight, overweight, and obesity were < 25, 25–29.9, and ≥30 kg/m2, respectively. NHANES also reports family income to pov- erty ratio. This is the ratio of family's income and poverty threshold based on the number of family members; a higher ratio indicates a higher income (Ostchega et al., 2018; U.S. Department of Health and Human Services, 2019). This ratio was stratified as ‘ < 2’ and ‘≥2’. Participants were also asked about the number of health care visits they have done over the past year; which was categorized as 0, 1–3, and ≥4 (Ostchega et al., 2018). CVD event was defined if a person had a history of myocardial infarction, coronary heart disease, stroke, or heart failure. 1. Introduction Among people without CVD event, 10-year predicted CVD risk was obtained with pooled cohort risk equations (Goff et al., 2014). Previous CVD event or ≥10% 10-year CVD risk was considered as high CVD risk (Muntner et al., 2017; Whelton et al., 2017). Studies from the US have consistently reported that the prevalence and likelihoods of hypertension differ by race. According to Dorans et al., the age-adjusted prevalence of hypertension was about 8.5% higher among non-Hispanic blacks compared to non-Hispanic whites in 2015–16. In addition to non-Hispanic blacks, the age-adjusted pre- valence of hypertension was higher among Mexican-Americans or other races (Dorans et al., 2018). The proportion of people who would require pharmacotherapy for hypertension or who have unmet treatment goals could also be substantially higher among non-Hispanic blacks com- pared to other races (Muntner et al., 2017). Moreover, studies reported that Mexican-Americans could remain untreated despite having in- dications for treatment (Gu et al., 2017; Park et al., 2018). Differences in awareness, lifestyle, and socioeconomic characteristics could be re- sponsible for these disparities (Gu et al., 2017; Whelton et al., 2017). Considering the clinical and public health importance of hypertension, estimates of prevalence, control, or treatment of hypertension ac- cording to characteristics such as age, gender, kidney disease, body weight, or cholesterol levels among racial groups in the US are also essential to investigate. While earlier studies found that the new guideline could change the estimates of overall hypertension burden per race in the US, how these estimates of hypertension could differ according to characteristics among different races have not been esti- mated yet. Furthermore, there have been limited recent studies that investigate racial disparities related to prevalence, control, and treat- ment of hypertension together. To understand the racial disparities related to hypertension burden, new studies are needed that quantify the prevalence, control, and treatment status of hypertensive in- dividuals according to race with recent data per the latest guideline. To address these gaps in the literature, this study obtained the aforemen- tioned estimates by race among US adults. Persons with stage 2 hypertension, stage 1 hypertension with dia- betes, CKD, or high CVD risk, and SBP ≥130 mmHg with ≥65 years of age were considered as treatment-eligible for hypertension (Supplemental Table 1). 1. Introduction hypertension (Whelton et al., 2017). Previous guidelines such as the Seventh Joint National Committee Guideline (JNC 7) recommends the SBP/DBP thresholds as 140/90 mmHg; however, the new guideline recommends the cutoffs as 130/80 mmHg, 10 mmHg lower than the SBP/DBP thresholds of the previous guidelines (Chalmers et al., 1999; Chobanian et al., 2003; Whelton et al., 2017). Furthermore, the term ‘prehypertension’ has been replaced with a new term ‘elevated BP’ which recommends the SBP cutoff as 120–129 mmHg while the person has a normal DBP (i.e., < 80 mmHg) (Whelton et al., 2017). The 2017 hypertension (Whelton et al., 2017). Previous guidelines such as the Seventh Joint National Committee Guideline (JNC 7) recommends the SBP/DBP thresholds as 140/90 mmHg; however, the new guideline recommends the cutoffs as 130/80 mmHg, 10 mmHg lower than the SBP/DBP thresholds of the previous guidelines (Chalmers et al., 1999; Chobanian et al., 2003; Whelton et al., 2017). Furthermore, the term ‘prehypertension’ has been replaced with a new term ‘elevated BP’ which recommends the SBP cutoff as 120–129 mmHg while the person has a normal DBP (i.e., < 80 mmHg) (Whelton et al., 2017). The 2017 Uncontrolled hypertension is the most common risk factor for car- diovascular disease (CVD) and is responsible for a large number of deaths and disabilities within the US (Forouzanfar et al., 2017; Heron, 2018). The 2017 American College of Cardiology/American Heart As- sociation (ACC/AHA) Guideline for the Prevention, Detection, Evalua- tion, and Management of High Blood Pressure in Adults has changed the systolic/diastolic blood pressure (SBP/DBP) thresholds to define Preventive M edicine Reports 14 (2019) 100850 G.M. Al Kibria calculate BP levels. If an individual had an SBP of 120–129 mmHg with a normal DBP (i.e., < 80 mmHg), then the BP level was classified as elevated BP. To classify an individual as stage 1 and stage 2 hy- pertensive, the following SBP/DBP thresholds were used, respectively (in mmHg), 130–139/80–89 and ≥140/90. Hypertension includes both stage 1 and stage 2 hypertension. Individuals who reported that they were currently taking BP lowering drugs were categorized as hy- pertensive regardless of BP levels (Supplemental Table 1) (Muntner et al., 2017; Whelton et al., 2017). ACC/AHA guideline has also changed the recommendations about treatment initiations and treatment goals. More than 45% of US adults (i.e., aged ≥20 years) could be hypertensive as per the new guideline (Muntner et al., 2017). 2.1. Data source This study used the National Health and Nutrition Examination Survey (NHANES) data. The continuous NHANES is a cross-sectional survey. The primary aim of this biennial survey is to obtain nationally representative samples of the non-institutionalized US population. Details of the survey are available elsewhere (Zipf et al., 2013). This analysis was limited to 2011–16 survey years to report the most recent estimates. These publicly available datasets were downloaded and merged according to unique identification numbers. The Ethics Review Board of the National Center for Health Statistics approved the survey protocols (National Center for Health Statistics, 2017). 2.3. Statistical analysis The background characteristics of the hypertensive and overall study participants were reported according to their race. Mean and standard error (SE) were used to report continuous variables while categorical variables were reported with weighted percentages and unweighted numbers. During comparison, continuous variables were tested with t-tests and categorical variables were tested with chi-square tests. Then, the age-adjusted proportion (with 95% confidence interval [CI]) of individuals with hypertension, people with treatment-eligibility for hypertension, people who were not taking antihypertensive drugs despite being treatment-eligible, and people with unmet treatment goals among treated were reported according to race as well as the overall population. The age distributions from the 2015 population census were used to obtain age-adjusted estimates (US Census Bureau, 1. Introduction Among individuals who reported that they were currently taking antihypertensive drugs, the following individuals were considered as not meeting treatment goals: if the SBP/DBP was ≥130/80 mmHg among age groups < 65 years and if the SBP was ≥130 mmHg among age groups ≥65 years (Supplemental Table 1) (Muntner et al., 2017; Whelton et al., 2017). 2.2. Study variables Using factory-calibrated ‘Baumanometer® mercury true gravity wall model sphygmomanometers’, trained physicians measured the BP in mobile examination centers. The BP was recorded for four times in the seated position after resting for 5 min. Appropriate cuff sizes were used (Centers for Disease Control and Prevention, 2009; Zipf et al., 2013). In this study, the mean of the first 3 BP measurements was used to 2 Preventive M edicine Reports 14 (2019) 100850 G.M. Al Kibria G.M. Al Kibria Table 1 Background characteristics of the hypertensive study participants by race, NHANES 2011–16a,b. 2.2. Study variables Table 1 Background characteristics of the hypertensive study participants by race, NHANES 2011–16a,b. y p g c Other races/ethnicities include non-Hispanic Asians, other Hispanics, and other races including multi-races. d l b d b h f l bl bl c Other races/ethnicities include non-Hispanic Asians, other Hispanics, and other races including multi-race d p-Values obtained by chi-square tests (for categorical variables) or t-tests (continuous variables). d p-Values obtained by chi-square tests (for categorical variables) or t-tests (continuous variables) (Table 1). The mean age of the participants was 47.6 years (SE: 0.4). About 51.8% of the respondents were females. Most of the character- istics differed according to race (p < 0.05). The proportion of hy- pertensive people with a high cholesterol level was higher among non- Hispanic whites while Mexican-Americans had a higher proportion of hypertensive people with young age, diabetes, obesity, and low family income to poverty ratio. Overall, 24.8% people had high CVD risk. 2015). The analysis also accounted for multistage cluster sampling design of the NHANES to obtain all the estimates (Johnson et al., 2014); using the mobile examination center's weights, all the weighted esti- mates were reported. Overall, due to a lower proportion of missing data (< 10%), variables with missing data were neither dropped nor im- puted (i.e., available case analysis) (Kang, 2013). Stata 14.0 was used to analyze data. The ‘svy’ command in Stata allows to adjust for the multistage cluster-sampling design (Stata Corporation, College Station, Texas USA, 2015). Table 2 shows the age-adjusted prevalence of hypertension ac- cording to race and the prevalence among the overall population. The prevalence of hypertension was 45.7% (95% CI: 44.1%–47.3%) among non-Hispanic whites, 59.0% (95% CI: 57.4%–60.6%) among non-His- panic blacks, 46.1% (95% CI: 43.8%–48.3%) among Mexican-Amer- icans, 45.2% (95% CI: 43.3%–47.2%) among other races/ethnicities, 2.2. Study variables Characteristics Overall study population (N = 16,103) Hypertensive participants by race Non-Hispanic White (n = 3132/6066) Non-Hispanic Blacks, (n = 2253/3665) Mexican-Americans, (n = 981/2164) Other races/ethnicities c, (n = 1885/4208) All races, (n = 8251/16,103) p-Valued SBP, mean (SE), mmHg 122.4 (0.3) 133.0 (0.5) 136.8 (0.6) 134.8 (0.8) 134.8 (0.6) 133.9 (0.4) < 0.01 DBP, mean (SE), mmHg 70.7 (0.2) 74.4 (0.4) 76.1 (0.6) 76.1 (0.7) 76.2 (0.6) 74.9 (0.3) < 0.01 Age (in years) Mean (SE) 47.6 (0.4) 57.8 (0.4) 52.8 (0.6) 49.9 (1.1) 53.6 (0.7) 56.1 (0.3) < 0.001 20–44 44.7 (6797) 19.8 (587) 29.1 (493) 39.1 (229) 29.5 (414) 23.6 (1723) < 0.001 45–54 18.7 (2720) 18.9 (487) 23.8 (445) 23.3 (178) 20.7 (354) 20.1 (1464) 55–64 17.3 (2798) 25.5 (593) 24.9 (634) 19.2 (271) 21.7 (480) 24.5 (1978) ≥65 19.2 (3788) 35.8 (1465) 22.2 (681) 18.4 (303) 28.0 (637) 31.8 (3086) Gender Male 48.2 (7811) 51.0 (1616) 46.2 (1133) 56.5 (516) 52.6 (963) 50.9 (4228) < 0.001 Female 51.8 (8292) 49.0 (1516) 53.8 (1120) 43.5 (465) 47.4 (922) 49.1 (4023) Cholesterol level (in mg/dl) Normal (< 200) 45.9 (7039) 29.7 (905) 43.5 (832) 37.8 (312) 34.8 (568) 32.6 (2617) < 0.001 Borderline (200–239) 24.0 (3504) 23.0 (631) 19.7 (398) 26.8 (229) 23.8 (423) 22.9 (1681) High (≥240) 30.0 (4734) 47.4 (1470) 36.8 (848) 35.5 (391) 41.4 (780) 44.5 (3489 High-density lipoprotein cholesterol (in mg/dl) Normal 70.5 (10357) 67.9 (1924) 72.1 (1467) 60.9 (565) 62.8 (1096) 67.4 (5052) < 0.001 Low (< 40 for men & < 50 for women) 29.5 (4671) 32.1 (1003) 27.9 (526) 39.1 (356) 37.2 (624) 32.6 (2509) Chronic kidney disease No 85.1 (12981) 76.4 (2118) 75.5 (1583) 76.9 (706) 78.2 (1408) 76.5 (5815) 0.47 Yes 14.9 (2814) 23.6 (928) 24.5 (586) 23.1 (254) 21.8 (416) 23.5 (2184) Diabetes mellitus status No 83.4 (12515) 75.7 (2229) 71.1 (1466) 68.7 (596) 71.2 (1270) 74.0 (5561) < 0.001 Prediabetes 6.1 (950) 8.2 (238) 6.8 (144) 7.1 (70) 9.4 (160) 8.1 (612) Diabetes 10.5 (2221) 16.1 (550) 22.1 (548) 24.2 (284) 19.4 (387) 17.9 (1769) Body mass index (in kg/m2) Normal/underweight (< 25) 29.5 (4695) 19.5 (642) 18.4 (421) 10.4 (108) 28.5 (552) 19.8 (1723) < 0.001 Overweight (25–29.9) 32.8 (5038) 34.1 (1000) 25.9 (590) 31.5 (311) 31.2 (615) 32.5 (2516) Obese (≥30) 37.8 (6013) 46.4 (1368) 55.7 (1164) 58.1 (533) 40.2 (649) 47.7 (3714) Number of health care visits within the past year 0 15.3 (2600) 9.3 (276) 12.1 (246) 21.6 (168) 16.7 (282) 11.4 (972) < 0.001 1–3 64.8 (10264) 63.7 (1904) 67.5 (1528) 61.4 (608) 62.7 (1194) 63.9 (5234) ≥4 19.9 (3223) 27.0 (948) 20.4 (478) 17.0 (203) 20.6 (407) 24.7 (2036) Family income-to-poverty ratio < 2 36.7 (7305) 29.1 (1331) 53.7 (1083) 64.6 (566) 46.2 (821) 36.6 (3801) 0.07 ≥2 63.3 (7383) 70.9 (1609) 46.3 (925) 35.4 (291) 53.8 (836) 63.4 (3661) 10-yr CVD risk categories Low 75.2 (9652) 53.7 (1159) 55.8 (865) 70.1 (460) 60.4 (845) 55.9 (3329) < 0.001 High 24.8 (4322) 46.3 (1520) 44.2 (988) 29.9 (374) 39.3 (694) 44.1 (3576) Abbreviations: CVD: Cardiovascular disease, DBP: Diastolic blood pressure, NHANES: National Health and Nutrition Examination Survey, SBP: Systolic blood pressure, SE: Standard error.i Table 1 Background characteristics of the hypertensive study participants by race, NHANES 2011–16a,b. p , a Numbers are presented as weighted percentages and unweighted numbers unless otherwise specified. b Numbers may not add up to total because of missing values. 3. Results The analysis included 16,103 respondents aged ≥20 years 3 Preventive M edicine Reports 14 (2019) 100850 G.M. Al Kibria Table 2 Age-adjusted prevalence (with 95% confidence interval) of hypertension according to race, NHANES 2011–16a. 3. Results Characteristics Non-Hispanic Whites Non-Hispanic Blacks Mexican-Americans Other races/ethnicitiesb In all races Age (in years) 20–44 24.6 (22.3–27.0) 32.3 (29.5–35.2) 22.3 (19.5–25.4) 21.2 (18.6–24.0) 24.7 (23.0–26.4) 45–54 47.8 (43.3–52.4) 68.3 (63.9–72.4) 49.0 (43.2–54.7) 46.4 (42.6–50.3) 50.1 (47.3–53.0) 55–64 63.8 (60.4–67.1) 83.8 (80.5–86.7) 66.1 (60.1–71.5) 65.6 (61.0–70.0) 66.3 (63.8–68.8) ≥65 75.8 (73.1–78.4) 89.7 (87.3–91.8) 79.8 (73.1–85.2) 80.9 (77.3–84.1) 77.7 (75.3–79.8) Sex Male 49.3 (46.9–51.7) 60.9 (58.9–62.9) 48.2 (45.3–51.2) 49.1 (46.6–51.7) 50.2 (48.4–51.9) Female 42.0 (40.1–43.9) 57.4 (55.2–59.7) 43.6 (41.3–46.0) 41.6 (39.4–43.8) 43.7 (42.3–45.1) Cholesterol level (in mg/dl) Normal (< 200) 40.3 (37.9–42.8) 56.0 (53.9–58.2) 39.4 (37.0–42.0) 41.3 (38.2–44.5) 42.2 (40.4–44.1) Borderline (200–239) 44.4 (41.4–47.4) 55.6 (52.1–59.1) 45.7 (40.0–51.6) 44.9 (41.2–48.6) 45.2 (42.9–47.5) High (≥240) 57.9 (54.3–61.5) 73.5 (68.3–78.1) 59.1 (53.4–64.7) 56.2 (52.1–60.3) 59.2 (56.3–61.9) High-density lipoprotein cholesterol (in mg/dl) Normal 42.0 (40.1–43.9) 56.6 (54.6–58.5) 44.8 (42.2–47.4) 43.0 (41.0–45.1) 43.8 (42.4–45.2) Low (< 40 for men & < 50 for women) 53.8 (51.5–56.1) 64.9 (62.0–67.8) 47.1 (43.2–51.0) 49.4 (46.7–52.1) 53.2 (51.5–54.9) Chronic kidney disease No 43.7 (42.0–45.4) 56.4 (54.6–58.3) 42.8 (40.9–44.8) 43.0 (40.9–45.1) 44.6 (43.3–45.9) Yes 58.7 (53.8–63.4) 74.1 (69.2–78.5) 65.5 (60.4–70.3) 61.5 (56.4–66.3) 62.2 (58.9–65.4) Diabetes mellitus status No 42.8 (40.9–44.6) 56.3 (54.1–58.4) 42.5 (40.0–44.9) 42.6 (40.4–44.8) 43.9 (42.6–45.3) Prediabetes 54.1 (46.8–61.1) 64.8 (53.2–74.9) 52.0 (41.7–62.1) 57.2 (51.8–62.4) 55.8 (50.8–60.6) Diabetes 73.9 (67.8–79.2) 76.9 (68.8–83.4) 65.8 (58.6–72.4) 57.5 (49.8–64.8) 70.2 (66.3–73.8) Body mass index (in kg/m2) Normal/underweight (< 25) 32.4 (30.0–34.8) 49.7 (46.8–52.6) 32.8 (28.0–38.1) 36.8 (33.7–40.0) 34.6 (32.8–36.5) Overweight (25–29.9) 44.7 (42.4–46.9) 53.9 (51.7–56.1) 42.3 (39.0–45.6) 44.0 (41.8–46.1) 44.9 (43.3–46.5) Obese (≥30) 57.9 (55.6–60.0) 67.3 (64.9–69.6) 53.1 (50.0–56.1) 57.1 (53.4–60.7) 58.2 (56.6–59.7) Family income to poverty ratio < 2 47.8 (45.3–50.4) 60.4 (58.2–62.7) 47.2 (44.7–49.7) 46.1 (43.9–48.3) 49.5 (48.1–50.9) ≥2 45.2 (43.2–47.2) 57.7 (55.3–60.0) 44.5 (41.1–47.9) 43.9 (41.0–47.0) 45.8 (44.3–47.3) Number of health care visits within the past year 0 42.6 (37.4–47.9) 58.5 (53.8–63.1) .40.8 (36.4–45.3) 40.4 (35.3–45.8) 43.2 (39.8–46.7) 1–3 44.4 (42.7–46.2) 58.1 (56.5–59.7) 46.7 (44.0–49.5) 44.5 (42.0–47.0) 46.0 (44.6–47.4) ≥4 50.4 (46.9–53.9) 64.1 (60.1–67.9) 49.8 (44.2–55.4) 48.9 (44.5–53.3) 51.4 (48.7–54.1) 10-yr CVD risk categories Low 36.1 (33.9–38.3) 51.6 (48.3–54.8) 39.7 (36.9–42.6) 36.0 (33.5–38.7) 37.3 (35.6–39.0) High 63.0 (56.3–69.2) 85.7 (78.9–90.6) 61.5 (49.3–72.5) 68.6 (59.2–76.7) 67.4 (62.7–71.8) Overall 45.7 (44.1–47.3) 59.0 (57.4–60.6) 46.1 (43.8–48.3) 45.2 (43.3–47.2) 46.9 (45.8–48.1) Abbreviations: CVD: Cardiovascular disease, NHANES: National Health and Nutrition Examination Survey. a If the systolic/diastolic blood pressure cutoff was ≥130/80 mmHg or a person reported taking antihypertensive medication. hypertension. b Other races/ethnicities include non-Hispanic Asians, other Hispanics, and other races including multi-races. 3. Results This includes both stages 1 and 2 hypertension. b Other races/ethnicities include non-Hispanic Asians, other Hispanics, and other races including multi-races. Abbreviations: CVD: Cardiovascular disease, NHANES: National Health and Nutrition Examination Survey. a If the systolic/diastolic blood pressure cutoff was ≥130/80 mmHg or a person reported taking antihypertensive medication. This includes both stages 1 and 2 hypertension. b Other races/ethnicities include non-Hispanic Asians, other Hispanics, and other races including multi-races. . ces/ethnicities include non-Hispanic Asians, other Hispanics, and other races including multi-races. like high CVD risk, diabetes mellitus, CKD, and high cholesterol level, about three-fourths of non-Hispanic blacks had hypertension. Figure 1 shows the age-adjusted prevalence of blood pressure stages by race. The overall prevalence of stage 1 hypertension was similar across races; however, the prevalence of stage 2 hypertension and the proportion of people on antihypertensive medication was highest among non-His- panic blacks. and 46.9% (95% CI: 45.8%–48.1%) among overall population. In all race categories, males had a higher prevalence of hypertension com- pared to females. The characteristics (e.g., overweight/obesity, high cholesterol, CKD, or the number of health care visits) that had higher prevalence in one racial category also had higher prevalence in other racial categories. Regardless of characteristics, non-Hispanic blacks had the highest prevalence among all races. For some of the characteristics Fig. 1. Age-adjusted prevalence of blood pressure stages according to race, National Health and Nutrition Examination Survey 2011–16. sted prevalence of blood pressure stages according to race, National Health and Nutrition Examination Survey 2011–16. Fig. 1. Age-adjusted prevalence of blood pressure stages according to race, National Health and Nutrition Exam 4 Preventive M edicine Reports 14 (2019) 100850 G.M. Al Kibria Table 3 Age-adjusted proportion (with 95% confidence interval) of individuals require antihypertensive medication with characteristics stratified by race, NHANES 2011–16a. Table 3 Age-adjusted proportion (with 95% confidence interval) of individuals require antihypertensive medication with characteristics stratified by race, NHANES 2011–16a. Characteristics Non-Hispanic Whites Non-Hispanic Blacks Mexican-Americans Other races/ethnicitiesb In all races Table 3 Age-adjusted proportion (with 95% confidence interval) of individuals require antihypertensive medication with characteristics stratified by race, NHANES 2011–16a. 3. Results Characteristics Non-Hispanic Whites Non-Hispanic Blacks Mexican-Americans Other races/ethnicitiesb In all races Age (in years) 20-44 12.2 (10.5-14.0) 19.3 (17.1-21.6) 11.2 (9.3-13.5) 10.1 (8.6-11.8) 12.6 (11.4-13.9) 45-54 32.4 (28.6-36.4) 53.4 (48.9-57.9) 32.4 (26.3-39.3) 33.0 (29.5-36.8) 34.9 (32.3-37.7) 55-64 55.0 (51.0-59.0) 77.0 (73.0-80.5) 57.1 (50.6-63.3) 55.4 (50.7-60.0) 57.6 (54.6-60.5) ≥65 75.1 (72.4-77.6) 89.2 (86.5-91.5) 79.8 (73.1-85.2) 80.6 (76.8-83.8) 77.0 (74.7-79.2) Sex Male 37.7 (35.7-39.8) 48.9 (46.9-50.8) 36.0 (32.6-39.5) 37.3 (35.1-39.5) 38.4 (36.9-40.1) Female 33.6 (32.0-35.3) 49.5 (47.4-51.6) 37.1 (34.6-39.7) 34.9 (32.8-37.1) 35.8 (34.5-37.2) Cholesterol level (in mg/dl) Normal (< 200) 31.1 (28.9-33.5) 45.7 (43.0-48.3) 30.7 (28.0-33.6) 33.2 (30.5-36.0) 33.2 (31.5-35.0) Borderline (200-239) 32.7 (29.8-35.7) 43.9 (41.3-46.5) 35.0 (28.9-41.5) 33.5 (30.5-36.7) 33.6 (31.4-35.8) High (≥240) 48.5 (44.8-52.2) 64.3 (59.3-69.0) 47.3 (41.7-52.9) 45.9 (41.7-50.2) 49.3 (46.2-52.3) High-density lipoprotein cholesterol (in mg/dl) Normal 32.6 (30.7-34.6) 46.5 (44.5-48.6) 35.3 (32.6-38.1) 33.4 (31.6-35.3) 34.3 (32.8-35.8) Low (< 40 for men & < 50 for women) 43.0 (40.4-45.6) 55.8 (52.6-59.0) 38.8 (35.6-42.1) 41.4 (38.4-44.5) 43.5 (41.6-45.4) Chronic kidney disease No 32.7 (31.3-34.2) 45.2 (43.3-47.0) 32.2 (30.0-34.5) 32.8 (31.0-34.8) 33.8 (32.6-35.0) Yes 58.7 (53.8-63.4) 74.1 (69.2-78.5) 65.5 (60.4-70.3) 61.5 (56.4-66.3) 62.2 (58.9-65.4) Diabetes mellitus status No 31.7 (30.1-33.3) 45.0 (43.1-47.0) 31.6 (28.9-34.3) 32.4 (30.3-34.6) 33.0 (31.7-34.2) Prediabetes 54.1 (46.8-61.1) 64.8 (53.2-74.9) 52.0 (41.7-62.1) 57.2 (51.8-62.4) 55.8 (50.8-60.6) Diabetes 68.0 (60.7-74.5) 72.5 (64.7-79.0) 55.8 (49.2-62.3) 51.5 (43.7-59.2) 63.9 (59.5-68.0) Body mass index (in kg/m2) Normal/underweight (< 25) 24.5 (22.5-26.7) 40.3 (37.7-43.0) 25.4 (20.9-30.6) 28.7 (26.1-31.5) 26.6 (25.0-28.2) Overweight (25-29.9) 33.4 (31.9-35.0) 43.4 (41.0-45.8) 32.8 (29.2-36.6) 35.0 (32.7-37.4) 34.3 (33.1-35.6) Obese (≥30) 46.2 (44.2-48.1) 57.5 (55.2-59.7) 42.9 (39.7-46.1) 45.9 (42.7-49.2) 47.1 (45.6-48.7) Family income to poverty ratio < 2 39.6 (37.5-41.8) 51.1 (49.1-53.0) 37.0 (34.0-40.0) 37.8 (35.4-40.2) 40.8 (39.4-42.2) ≥2 34.4 (32.6-36.2) 47.8 (45.6-50.0) 35.7 (33.0-38.6) 34.2 (31.6-36.8) 35.3 (33.9-36.8) Number of health care visits within the past year 0 29.1 (25.2-33.4) 42.3 (39.1-45.8) 30.6 (25.7-35.9) 27.7 (22.5-33.6) 30.1 (27.1-33.2) 1-3 34.6 (32.9-36.3) 49.1 (47.4-50.8) 36.6 (33.9-39.3) 35.9 (33.7-38.2) .36.5 (35.1-37.9) ≥4 42.5 (39.4-45.7) 58.2 (54.0-62.3) 45.6 (39.9-51.5) 42.8 (38.9-46.7) 44.2 (41.7-46.7) 10-yr CVD risk categories Low 23.7 (21.7-25.8) 37.6 (33.5-41.9) 28.5 (25.1-32.2) 24.8 (22.1-27.8) 25.0 (23.3-26.7) High 63.0 (56.3-69.2) 85.7 (78.9-90.6) 61.5 (49.3-72.5) 68.6 (59.2-76.7) 67.4 (62.7-71.8) Overall 35.7 (34.3-37.2) 49.3 (47.7-50.8) 36.6 (34.1-39.2) 36.0 (34.3-37.8) 37.2 (36.0-38.3) Abbreviations: CVD: Cardiovascular disease, NHANES: National Health and Nutrition Examination Survey. 3. Results a Individuals with stage 2 hypertension, stage 1 hypertension with diabetes, chronic kidney disease or high CVD risk, and systolic blood pressure ≥130 mmHg with ≥65 years of age were considered as treatment-eligible for hypertension. b Other races/ethnicities include non-Hispanic Asians, other Hispanics, and other races including multi-races. a Individuals with stage 2 hypertension, stage 1 hypertension with diabetes, chronic kidney disease or high C with ≥65 years of age were considered as treatment-eligible for hypertension. a Individuals with stage 2 hypertension, stage 1 hypertension with diabetes, chronic kidney disease or high CVD risk, and systolic blood pressure ≥130 mmHg with ≥65 years of age were considered as treatment-eligible for hypertension. b Other races/ethnicities include non-Hispanic Asians, other Hispanics, and other races including multi-races. y g g y b Other races/ethnicities include non-Hispanic Asians, other Hispanics, and other races including multi-races. characteristics, despite having indications for treatment, Mexican- Americans and people of other races/ethnicities had a greater propor- tion of untreated people compared to the other 2 races. The proportion of age-adjusted treatment-eligible individuals was 35.7% (95% CI: 34.3%–37.2%), 49.3% (95% CI: 47.7%–50.8%), 36.6% (95% CI: 34.1%–39.2%), 36.0% (95% CI: 34.3%–37.8%), and 37.2% (95% CI: 36.0%–38.3%) among non-Hispanic whites, non-Hispanic blacks, Mexican-Americans, other races/ethnicities, and the overall population, respectively (Table 3). Similar to estimates from Table 2, characteristics that had a higher prevalence of hypertension also had increased indication for treatment, and non-Hispanic blacks were more likely to be eligible compared to other races as per most characteristics. As shown in Table 5, according to race, ordered from the highest age-adjusted proportion, 63.8% (95% CI: 60.0%–67.5%) of non-His- panic blacks, 60.5% (95% CI: 54.1%–66.5%) of other races/ethnicities, 58.2% (95% CI: 51.7%–64.5%) of Mexican-Americans, and 49.7% (95% CI: 45.0%–54.4%) of non-Hispanic whites were not meeting the treat- ment goals. Per most of the characteristics, non-Hispanic blacks had higher proportions of people with unmet treatment goals compared to other 3 races. Figure 2 summarizes all four studied prevalence ac- cording to race as well as the overall population. Table 4 shows the age-adjusted proportion of individuals who were not receiving treatment despite having indication; contrary to the findings of Tables 2 & 3, a lower proportion of non-Hispanic blacks were untreated regardless of characteristics. 3. Results Overall, about half of the people with other races/ethnicities were not receiving treatment, fol- lowed by Mexican-Americans, non-Hispanic whites, and non-Hispanic blacks, 49.3% (95% CI: 45.5%–53.0%), 46.5% (95% CI: 42.0%–51.0%), 38.9% (95% CI: 35.3%–42.6%), and 35.1% (95% CI: 32.4%–37.9%), respectively. In all races, younger people, males, people with normal cholesterol, normal HDL, non-diabetic, normal weight, low family in- come to poverty ratio, and low CVD risk were more likely to remain untreated despite having treatment-indication. For most of the 4. Discussion This study revealed the racial disparities in age-adjusted prevalence and treatment status of hypertension within the US. Compared to other races, non-Hispanic blacks had higher prevalence and treatment in- dication for hypertension. Although they were more likely to be treated, about two-thirds were not meeting treatment goals; this proportion was also the highest among all races. The proportion of individuals who 5 Preventive M edicine Reports 14 (2019) 100850 G.M. Al Kibria Table 4 Age-adjusted proportion (with 95% confidence interval) of untreated hypertensive adults among who have antihypertensive treatment indication per characteristics stratified by race, NHANES 2011–16a. Table 4 Age-adjusted proportion (with 95% confidence interval) of untreated hypertensive adults among who have antihypertensive treatment indication per characteristics stratified by race, NHANES 2011–16a. 4. Discussion Characteristics Non-Hispanic Whites Non-Hispanic Blacks Mexican-Americans Other races/ethnicitiesb In all races Age (in years) 20-44 48.6 (42.1-55.1) 47.9 (41.9-53.9) 58.1 (48.7-67.0) 65.7 (58.1-72.6) 51.9 (47.6-56.2) 45-54 37.2 (31.2-43.7) 28.3 (22.0-35.6) 50.0 (39.8-60.2) 44.2 (36.5-52.2) 37.5 (33.4-41.7) 55-64 29.1 (24.9-33.6) 24.2 (19.7-29.3) 30.2 (22.8-38.8) 34.1 (28.2-40.5) 29.0 (25.9-32.2) ≥65 27.0 (23.6-30.7) 22.1 (19.2-25.2) 30.8 (26.5-35.6) 29.8 (26.0-33.8) 27.0 (24.3-29.8) Sex Male 45.9 (41.5-50.2) 42.9 (38.0-48.0) 56.4 (49.1-63.5) 52.9 (47.9-57.7) 47.3 (44.4-50.3) Female 29.5 (25.0-34.4) 28.7 (24.9-32.8) 34.8 (27.2-43.2) 45.4 (39.4-51.5) 32.3 (29.2-35.5) Cholesterol level (in mg/dl) Normal (< 200) 41.4 (36.5-46.5) 37.8 (33.7-42.0) 47.4 (39.1-55.8) 56.8 (50.3-63.0) 43.2 (39.9-46.6) Borderline (200-239) 49.4 (41.2-57.7) 49.2 (42.0-56.4) 55.5 (43.9-66.6) 56.9 (49.9-63.6) 50.8 (45.3-56.3) High (≥240) 28.2 (24.2-32.7) 17.8 (13.0-24.0) 39.1 (28.8-50.5) 40.1 (30.4-50.5) 29.4 (26.0-33.0) High-density lipoprotein cholesterol (in mg/dl) Normal 41.3 (36.7-46.0) 37.2 (33.6-40.9) 51.9 (44.0-59.8) 52.7 (47.8-57.7) 43.0 (39.9-46.1) Low (< 40 for men & < 50 for women) 34.2 (28.7-40.2) 28.3 (23.6-33.5) 37.8 (30.8-45.4) 44.4 (37.9-51.1) 35.4 (31.6-39.4) Chronic kidney disease No 36.3 (32.3-40.6) 33.8 (30.6-37.2) 43.6 (37.8-49.7) 48.4 (44.2-52.7) 38.3 (35.4-41.2) Yes 53.7 (46.9-60.3) 40.3 (35.9-44.8) 51.5 (42.7-60.1) 52.9 (43.9-61.6) 50.2 (46.5-54.0) Diabetes mellitus status No 41.9 (38.2-45.7) 40.7 (38.0-43.5) 53.7 (46.7-60.6) 57.2 (52.9-61.4) 44.6 (42.3-47.0) Prediabetes 40.3 (30.9-50.5) 38.8 (28.6-50.1) 53.3 (38.2-67.8) 42.4 (30.8-54.9) 41.1 (34.7-47.9) Diabetes 25.1 (16.0-37.2) 17.5 (11.4-25.9) 28.5 (19.5-39.8) 27.2 (14.5-45.2) 24.2 (18.6-30.8) Body mass index (in kg/m2) Normal/underweight (< 25) 47.4 (36.7-58.3) 55.9 (47.1-64.4) 61.4 (40.5-78.8) 58.8 (52.2-65.1) 51.2 (43.8-58.5) Overweight (25-29.9) 42.0 (34.6-49.7) 31.4 (24.3-39.5) 59.0 (46.5-70.4) 54.4 (47.6-61.0) 43.9 (39.0-49.0) Obese (≥30) 34.9 (31.0-39.1) 31.6 (28.2-35.2) 39.4 (33.0-46.2) 42.7 (36.4-49.2) 35.7 (33.1-38.3) Family income to poverty ratio < 2 41.3 (36.6-46.1) 35.9 (30.6-41.6) 48.9 (40.3-57.6) 51.6 (47.1-56.1) 42.3 (39.4-45.3) ≥2 38.0 (33.3-42.9) 34.7 (30.5-39.2) 43.4 (33.8-53.5) 48.5 (42.0-54.9) 39.2 (35.7-42.8) Number of health care visits within the past year 0 82.1 (72.8-88.7) 85.3 (75.4-91.6) 92.4 (84.5-96.4) 87.2 (77.4-93.1) 87.2 (77.4-93.1) 1-3 38.1 (33.3-43.1) 32.6 (29.3-36.0) 39.7 (32.0-47.9) 47.0 (42.3-51.9) 47.0 (42.3-51.9) ≥4 25.2 (20.2-30.8) 21.3 (16.1-27.7) 25.0 (14.6-39.2) 33.9 (26.3-42.5) 33.9 (26.3- 42.5) 10-yr CVD risk categories Low 41.1 (35.4-47.0) 34.8 (30.6-39.2) 48.3 (41.1-55.5) 52.1 (46.4-57.8) 42.6 (38.9-46.4) High 36.7 (26.2-48.7) 25.2 (17.3-35.1) 43.8 (28.1-60.8) 52.2 (43.0-61.3) 37.6 (31.6-43.9) Overall 38.9 (35.3-42.6) 35.1 (32.4-37.9) 46.5 (42.0-51.0) 49.3 (45.5-53.0) 40.4 (38.0-42.8) Abbreviations: CVD: Cardiovascular disease, NHANES: National Health and Nutrition Examination Survey. 4. Discussion a Individuals with stage 2 hypertension, stage 1 hypertension with diabetes, chronic kidney disease or high CVD risk, and systolic blood pressure ≥130 mmHg with ≥65 years of age were considered as treatment-eligible for hypertension. b Other races/ethnicities include non-Hispanic Asians, other Hispanics, and other races including multi-races. Table 4 Age-adjusted proportion (with 95% confidence interval) of untreated hypertensive adults among who have antihypertensive treatment indication per characteristic stratified by race, NHANES 2011–16a. y a Individuals with stage 2 hypertension, stage 1 hypertension with diabetes, chronic kidney disease or high CVD risk, and systolic blood pressure ≥130 mmHg with ≥65 years of age were considered as treatment-eligible for hypertension. b Other races/ethnicities include non-Hispanic Asians, other Hispanics, and other races including multi-races. measures that could reduce the higher prevalence of hypertension among African-Americans. could be hypertensive or treatment-eligible for hypertension among races other than the non-Hispanic blacks were similar; however, the proportion of untreated individuals despite being treatment-eligible was substantially higher among Mexican-Americans and other races/ ethnicities. To the best of this author's knowledge, this is the first epi- demiological study to report the racial/ethnic disparities in prevalence, treatment, and control of hypertension among US adults per the 2017 ACC/AHA guideline. The prevalence of overall and uncontrolled hypertension among Mexican-Americans was higher than the prevalence estimates by Muntner and colleagues that also analyzed the NHANES datasets from 2011 to 14 survey years (Muntner et al., 2017); this could be due to age- adjustment. Without age-adjustment, a large proportion of younger people among Mexican-Americans could show a lower prevalence of hypertension among them (Table 1). Dorans et al. reported the age- adjusted prevalence of hypertension according to race following ap- plication of the 2017 guideline. The estimates reported by the authors were similar to the estimates of this study (Dorans et al., 2018). The overall estimates of prevalence and treatment status observed by this study are similar to findings of other studies (Dorans et al., 2018; Khera et al., 2018; Muntner et al., 2017). Over the past few decades, non-Hispanic blacks had the highest prevalence and unmet treatment goals among all races in the US (Dorans et al., 2018; Whelton et al., 2017). However, the prevalence of hypertension among Africans in Africa are considerably lower than this prevalence. 4. Discussion Although genetic similarities exist between Africans in Africa and African-Americans in the US, differences in lifestyles and socioeconomic factors between the African-Americans and Africans in Africa could cause this higher pre- valence of hypertension (Dorans et al., 2018; Forouzanfar et al., 2017; Ostchega et al., 2018; Whelton et al., 2017). These findings highlight the importance of adopting adequate lifestyle and other preventive Even though the non-Hispanic blacks were receiving treatment, the treatment was not adequate to reduce the pressure level (i.e., to meet the treatment goals). Regular monitoring of treated individuals is thus essential. This proportion was also substantially higher among people from other races/ethnicities as well as Mexican-Americans. The un- controlled hypertension could put them at a greater risk of complica- tions as also seen by previous studies (Muntner et al., 2017; Ostchega et al., 2018). Even a large proportion of people with high CVD risks had uncontrolled hypertension. Furthermore, the overall age-adjusted 6 Preventive M edicine Reports 14 (2019) 100850 G.M. Al Kibria Table 5 Age-stratified and overall proportion (with 95% confidence interval) of treated hypertensive adults who have unmet treatment goals per characteristics stratified by race, NHANES 2011–16a. 4. Discussion Characteristics Non-Hispanic Whites Non-Hispanic Blacks Mexican-Americans Other races/ethnicitiesb In all races Age (in years) 20-44 45.9 (36.2-55.9) 61.6 (54.8-67.9) 60.2 (42.5-75.6) 62.2 (49.1-73.7) 52.2 (45.4-58.9) 45-54 46.8 (37.6-56.2) 64.9 (58.7-70.6) 47.9 (30.4-66.0) 56.4 (47.7-64.8) 51.7 (45.6-57.8) 55-64 48.6 (42.2-55.1) 62.0 (57.4-66.5) 57.1 (49.5-64.3) 53.8 (44.2-63.2) 51.6 (46.9-56.4) ≥65 61.7 (57.3-66.0) 69.5 (65.5-73.2) 64.2 (54.1-73.3) 66.0 (59.8-71.6) 63.1 (59.4-66.5) Sex Male 52.8 (45.6-59.8) 68.7 (62.3-74.4) 66.5 (53.2-77.6) 61.8 (52.9-70.0) 57.1 (52.0-62.1) Female 46.7 (41.5-51.9) 61.0 (55.2-66.5) 53.3 (44.1-62.3) 59.0 (49.8-67.7) 51.8 (48.0-55.7) Cholesterol level (in mg/dl) Normal (< 200) 50.3 (42.0-58.6) 63.0 (55.8-69.7) 62.1 (50.4-72.5) 64.4 (54.1-73.4) 55.3 (49.5-61.0) Borderline (200-239) 50.2 (40.2-60.1) 75.1 (61.1-85.3) 51.6 (38.0-65.0) 64.2 (50.3-76.0) 55.6 (48.6-62.4) High (≥240) 48.0 (40.8-55.1) 61.0 (52.7-68.8) 57.0 (32.1-78.8) 56.4 (42.0-69.9) 51.4 (45.7-57.1) High-density lipoprotein cholesterol (in mg/dl) Normal 47.9 (42.9-53.0) 64.8 (58.7-70.5) 59.0 (46.8-70.2) 60.6 (49.6-70.7) 55.4 (50.8-59.8) Low (< 40 for men & < 50 for women) 63.6 (47.9-76.9) 61.1 (52.8-68.9) 55.8 (46.6-64.6) 58.5 (49.1-67.3) 51.4 (46.6-56.3) Chronic kidney disease No 48.7 (43.6-53.8) 61.8 (56.4-66.9) 57.6 (50.6-64.3) 58.2 (51.3-64.8) 52.7 (48.8-56.5) Yes 56.9 (42.7-70.1) 67.0 (57.0-75.6) 59.2 (36.6-78.6) 68.0 (53.3-79.8) 60.3 (53.0-67.1) Diabetes mellitus status No 49.1 (43.6-54.6) 62.3 (56.7-67.6) 61.3 (51.5-70.2) 65.9 (58.3-72.7) 53.8 (49.6-58.0) Prediabetes 49.1 (34.5-63.8) 64.0 (43.9-80.1) 71.6 (44.6-88.8) 55.3 (40.1-69.6) 54.3 (45.1-63.1) Diabetes 52.3 (42.0-62.3) 68.0 (60.4-74.7) 50.0 (31.9-68.2) 51.8 (36.3-66.9) 55.1 (48.1-62.0) Body mass index (in kg/m2) Normal/underweight (< 25) 47.1 (34.6-60.0) 64.6 (46.8-79.1) 72.2 (63.0-79.8) 57.1 (41.1-71.7) 52.3 (42.1-62.4) Overweight (25-29.9) 49.3 (41.0-57.7) 69.6 (60.7-77.3) 50.1 (33.9-66.3) 66.7 (53.3-77.9) 54.6 (47.8-61.3) Obese (≥30) 50.2 (44.7-55.7) 62.1 (57.5-66.4) 57.2 (49.0-65.0) 57.0 (50.0-63.7) 54.1 (50.4-57.6) Family income to poverty ratio < 2 46.1 (39.2-53.1) 64.3 (57.9-70.3) 56.9 (46.5-66.8) 58.2 (49.2-66.6) 54.4 (50.0-58.8) ≥2 50.3 (44.3-56.3) 64.8 (59.8-69.5) 56.9 (42.8-69.9) 63.1 (53.3-72.0) 53.7 (49.1-58.1) Number of health care visits within the past year 0 57.4 (30.8-80.4) 84.5 (69.5-92.8) 30.4 (20.5-42.4) 87.5 (57.8-97.3) 60.2 (42.0-76.0) 1-3 49.5 (43.5-55.6) 64.5 (59.7-69.1) 61.6 (51.6-70.7) 58.0 (51.0-64.7) 54.8 (50.4-59.1) ≥4 49.2 (41.6-56.8) 60.0 (53.5-66.1) 54.7 (38.7-69.8) 64.2 (53.0-74.1) 52.4 (46.9-57.9) 10-yr CVD risk categories Low 40.4 (33.4-47.7) 45.5 (40.2-51.0) 45.0 (29.7-61.3) 51.6 (43.2-59.9) 43.1 (37.8-48.5) High 51.5 (41.6-61.3) 76.3 (65.6-84.5) 65.0 (39.7-84.0) 68.2 (50.7-81.8) 60.7 (53.9-67.1) Overall 49.7 (45.0-54.4) 63.8 (60.0-67.5) 58.2 (51.7-64.5) 60.5 (54.1-66.5) 54.2 (50.7-57.6) Abbreviations: CVD: Cardiovascular disease, NHANES: National Health and Nutrition Examination Survey. 4. Discussion a If the systolic/diastolic blood pressure was ≥130/80 mmHg among individuals taking any BP lowering drugs (age groups < 65 years) or the systolic blood pressure was ≥130 mmHg (age groups ≥65 years), then they were considered as persons with unmet treatment goals. b Other races/ethnicities include non-Hispanic Asians, other Hispanics, and other races including multi-races. Table 5 Age-stratified and overall proportion (with 95% confidence interval) of treated hypertensive adults who have unmet treatment goals per characteristics stratified by race, NHANES 2011–16a. Table 5 Age-stratified and overall proportion (with 95% confidence interval) of treated hypertensive adults who have unmet treatment goals per characteristics stratified by race, NHANES 2011–16a. Ch t i ti N Hi i Whit N Hi i Bl k M i A i Oth / th i iti b I ll a If the systolic/diastolic blood pressure was ≥130/80 mmHg among individuals taking any BP lowering drugs (age groups < 65 years) or the systolic blood pressure was ≥130 mmHg (age groups ≥65 years), then they were considered as persons with unmet treatment goals. b Other races/ethnicities include non-Hispanic Asians, other Hispanics, and other races including multi-races. participants among these races could be responsible for this under- utilization of treatment (Gu et al., 2017). Multiple individual, social, and healthcare-related factors could also cause this underutilization (Egan et al., 2014; Gu et al., 2017; Wang and Vasan, 2005). Ensuring adequate treatment for all these untreated and uncontrolled individuals proportion of untreated treatment-eligible people was more than 40%, this proportion was higher among other races/ethnicities and Mexican- Americans; the age-adjusted number of health care visits over the past year was also lower among them (Supplemental Table 2). A previous study by Gu et al. suggests that a large proportion of uninsured Fig. 2. Age-adjusted proportion of people with hypertension, treatment-eligible for hypertension, untreated among treatment-eligible, and unmet treatment goals among treated according to race, National Health and Nutrition Examination Survey 2011–16. Fig. 2. Age-adjusted proportion of people with hypertension, treatment-eligible for hypertension, untreated among treatment-eligible, and unmet treatment goals among treated according to race, National Health and Nutrition Examination Survey 2011–16. ople with hypertension, treatment-eligible for hypertension, untreated among treatment-eligible, and unmet treatment goals ional Health and Nutrition Examination Survey 2011–16. 7 7 Preventive M edicine Reports 14 (2019) 100850 G.M. 5. Conclusion This study reported the racial/ethnic disparities associated with prevalence, treatment, and control of hypertension among US adults per the 2017 ACC/AHA guideline. Although the age-adjusted pre- valence of hypertension was substantially higher among all races, the non-Hispanic blacks had the highest prevalence regardless of char- acteristics. Treatment and control disparities also exist where the ma- jority of non-Hispanic blacks were not meeting the treatment goals, and a substantial proportion of Mexican-Americans or people of ‘other races/ethnicities’ were not receiving treatment despite being eligible for antihypertensive treatment. Interventions are needed to increase prevention, treatment, and control measures by all races to reduce ra- cial disparities associated with hypertension burden, including the complications resulting from it. If any of the participants of the present study (i.e., NHANES 2011–16) ever received an evaluation, were evaluated or treated under previous guidelines such as the JNC 7 guideline (Chalmers et al., 1999; Chobanian et al., 2003). As those guidelines recommend different cut- offs to define hypertension, treatment indication, and treatment goals, reevaluation of previously treated people is thus important. Despite a substantial increase in prevalence of hypertension following application of the 2017 ACC/AHA guideline, the overall proportion of treatment- eligible people would be similar to previous guidelines (Muntner et al., 2017). Furthermore, people who would be classified as stage 1 hy- pertensive per the 2017 ACC/AHA guideline, were classified as pre- hypertensive per the previous guidelines (Whelton et al., 2017). Pre- vious guidelines also recommend preventive measures for prehypertensive individuals (Chalmers et al., 1999; Chobanian et al., 2003). However, adoption of the new guideline could substantially reduce the complications resulting from hypertension and increase the awareness among all people from early stages of hypertension (Bundy et al., 2018; Whelton et al., 2017; Yano et al., 2018). Thus, the new guideline has potentials to reduce the racial disparities associated with hypertension and its complications if it is implemented, and adequate preventive and treatment measures are taken (Muntner et al., 2017; Whelton et al., 2017). Supplementary data to this article can be found online at https:// doi.org/10.1016/j.pmedr.2019.100850. Funding Not received for this study. 4. Discussion Al Kibria by overcoming these barriers would be helpful to reduce the compli- cations of hypertension including the disparities associated with treat- ment. (i.e., NHANES) used standardized validated methods to obtain BP le- vels. Using adequate statistical procedures to adjust for sample weights and age along with a large sample size also enabled reporting of pre- valence on a wide range of background characteristics (Johnson et al., 2014; Muntner et al., 2017). The limitations of the present study also merit discussion. The NHANES data was cross-sectional, and the BP was measured on a single day; this could cause some overestimation as the 2017 ACC/AHA recommends using measures of multiple days to con- firm the diagnosis of hypertension (Muntner et al., 2017; Whelton et al., 2017). Moreover, as earlier studies found a higher proportion of un- treated people among Hispanics other than Mexican-Americans, grouping them with non-Hispanic Asians and other races as a separate category (i.e., other races/ethnicities) could lead to some over- estimation in that group. Despite the dissimilarities, all four age-adjusted proportions were substantially higher among all studied races. To prevent and treat hy- pertension or its complications, all hypertensive individuals need to adopt lifestyle and other preventive measures such as increasing phy- sical activity or restricting salt intake (Whelton et al., 2017). Ad- ditionally, controlling other associated conditions such as diabetes, dyslipidemia, increased body weight, or CKD that increases likelihoods of hypertension would be essential. Some of the conditions such as CVD risk and CKD were disproportionately higher among non-Hispanic blacks compared to other 3 races (Supplemental Table 2), that could cause the difference. Preventing and treating these conditions would be necessary to reduce the racial disparities associated with prevention, treatment, and control of hypertension and its complications including the overall burden. These conditions also share common risk factors; the prevention and control strategies of these conditions are similar to hypertension (Berrington de Gonzalez et al., 2010; Burke et al., 2008; Forouzanfar et al., 2015; Hypertension Study Group, 2001; Scholes et al., 2012). Author's contributions GMAK had full access to all the data in the study and takes re- sponsibility for the integrity of the data and the accuracy of the data analysis.i Concept, first draft, statistical analysis, writing, review, and editing: GMAK Acknowledgements The author thanks Dr. Mohammad Rashidul Hashan, MBBS and Mr. Reese Crispen for proof-reading the document. Conflicts of interest No conflicts of interest to disclose. Racial disparities in prevalence, treatment, and control of hy- pertension and other chronic diseases have been studied widely over the past two decades (Giles et al., 2007; Gu et al., 2017; Karlamangla et al., 2010; Redmond et al., 2011). Studies have also shown how the awareness or outcome of diseases differ (Batson et al., 2010; Giles et al., 2007; Gu et al., 2017). Interventions are needed to uptake preventive and treatment measures as well as awareness. Prevention and control programs should incorporate the findings of the present study to reduce these disparities. Strategies such as prescribing (e.g., once-daily re- gimen if possible), educating (e.g., clearly written instructions), and tracking/encouraging (e.g., encouraging patients to use reminders) have been shown effective to increase medication adherence (Chang et al., 2018). These strategies require interactions of patients, physi- cians, and health care delivery systems (Bradbury et al., 2018; Chang et al., 2018). Successful implementation of these strategies is essential to reduce the racial disparities associated with prevention and treat- ment measures including the overall burden. In addition to races that have poor treatment or control of hypertension, people with the char- acteristics that were more likely to remain untreated (e.g., younger age and male gender) or uncontrolled (e.g., CKD and high CVD risk) are essential to prioritize. Batson, B., Belletti, D., Wogen, J., 2010. Effect of African American race on hypertension management: a real-world observational study among 28 US physician practices. Ethn Dis 20, 409–415. Batson, B., Belletti, D., Wogen, J., 2010. Effect of African American race on hypertension management: a real-world observational study among 28 US physician practices. Ethn Dis 20, 409–415. Berrington de Gonzalez, A., Hartge, P., Cerhan, J.R., Flint, A.J., Hannan, L., MacInnis, R.J., Moore, S.C., Tobias, G.S., Anton-Culver, H., Freeman, L.B., Beeson, W.L., Clipp, S.L., English, D.R., Folsom, A.R., Freedman, D.M., Giles, G., Hakansson, N., Henderson, K.D., Hoffman-Bolton, J., Hoppin, J.A., Koenig, K.L., Lee, I.-M., Linet, M.S., Park, Y., Pocobelli, G., Schatzkin, A., Sesso, H.D., Weiderpass, E., Willcox, B.J., Wolk, A., Zeleniuch-Jacquotte, A., Willett, W.C., Thun, M.J., 2010. Body-mass index and mortality among 1.46 million white adults. N. Engl. J. Med. 363, 2211–2219. https://doi.org/10.1056/NEJMoa1000367. Bradbury, K., Morton, K., Band, R., van Woezik, A., Grist, R., McManus, R.J., Little, P., Berrington de Gonzalez, A., Hartge, P., Cerhan, J.R., Flint, A.J., Hannan, L., MacInnis, R.J., Moore, S.C., Tobias, G.S., Anton-Culver, H., Freeman, L.B., Beeson, W.L., Clipp, S.L., English, D.R., Folsom, A.R., Freedman, D.M., Giles, G., Hakansson, N., Henderson, K.D., Hoffman-Bolton, J., Hoppin, J.A., Koenig, K.L., Lee, I.-M., Linet, M.S., Park, Y., Pocobelli, G., Schatzkin, A., Sesso, H.D., Weiderpass, E., Willcox, B.J., Wolk, A., Zeleniuch-Jacquotte, A., Willett, W.C., Thun, M.J., 2010. Body-mass index and mortality among 1.46 million white adults. N. Engl. J. Med. 363, 2211–2219. https://doi.org/10.1056/NEJMoa1000367. References Marks, G.B., Martin, R.V., Marzan, M.B., Mashal, M.T., Masiye, F., Mason-Jones, A.J., Matsushita, K., Matzopoulos, R., Mayosi, B.M., Mazorodze, T.T., McKay, A.C., McKee, Mekonnen, W., Melaku, Y.A., Meltzer, M., Memish, Z.A., Mendoza, W., Mensah, G.A., Chobanian, A.V., Bakris, G.L., Black, H.R., Cushman, W.C., Green, L.A., Izzo, J.L., Jones, D.W., Materson, B.J., Oparil, S., Wright, J.T., Roccella, E.J., National Heart, Lung, and Blood Institute Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure, National High Blood Pressure Education Program Coordinating Committee, 2003. The seventh report of the joint national committee on prevention, detection, evaluation, and treatment of high blood pres- sure: the JNC 7 report. JAMA 289, 2560–2572. https://doi.org/10.1001/jama.289. 19.2560. f urthy, K.S., Naghavi, M., Nahas, Z., Naheed, A., Naidoo, K.S., Naldi, L., Nand, D., Dorans, K.S., Mills, K.T., Liu, Y., He, J., 2018. Trends in prevalence and control of hy- pertension according to the 2017 American College of Cardiology/American Heart Association (ACC/AHA) guideline. J. Am. Heart Assoc. 7, e008888. https://doi.org/ 10.1161/JAHA.118.008888. ving, B., Nyakarahuka, L., Oh, I.-H., Ohkubo, T., Olusanya, B.O., Om Opio, J.N., Orozco, R., Pagcatipunan, R.S., Pain, A.W., Pandian, J.D., Panelo, C.I.A., Pavlin, B.I., Pearce, N., Pedraza, L.S., Pedroza, A., Stokic, L.P., Pekericli, A., Pereira, Pavlin, B.I., Pearce, N., Pedraza, L.S., Pedroza, A., Stokic, L.P., Pekericli, A., Pereira, D.M., Perez-Padilla, R., Perez-Ruiz, F., Perico, N., Perry, S.A.L., Pervaiz, A., Pesudovs, Egan, B.M., Li, J., Small, J., Nietert, P.J., Sinopoli, A., 2014. The growing gap in hy- pertension control between insured and uninsured adults: National Health and nu- trition examination survey 1988 to 2010. Hypertension 64, 997–1004. https://doi. org/10.1161/HYPERTENSIONAHA.114.04276. D.M., Perez-Padilla, R., Perez-Ruiz, F., Perico, N., Perry, S.A.L., Pervaiz, A., Pesudovs, eterson, C.B., Petzold, M., Phillips, M.R., Phua, H.P., Plass, D., Poenaru Polanczyk, G.V., Polinder, S., Pond, C.D., Pope, C.A., Pope, D., Popova, S., malek, F., Powles, J., Prabhakaran, D., Prasad, N.M., Qato, D.M., Queza Pourmalek, F., Powles, J., Prabhakaran, D., Prasad, N.M., Qato, D.M., Quezada, A.D., Quistberg, D.A.A., Racapé, L., Rafay, A., Rahimi, K., Rahimi-Movaghar, V., Rahman, S j k S i dd S f A Quistberg, D.A.A., Racapé, L., Rafay, A., Rahimi, K., Rahimi-Movaghar, V., Rahman, S U Raju M Rakovac I Rana S M Rao M Razavi H Reddy K S Refaat A H Forouzanfar, M.H., Alexander, L., Anderson, H.R., Bachman, V.F., Biryukov, S., Brauer, M., Burnett, R., Casey, D., Coates, M.M., Cohen, A., Delwiche, K., Estep, K., Frostad, J. References J., Kc, A., Kyu, H.H., Moradi-Lakeh, M., Ng, M., Slepak, E.L., Thomas, B.A., Wagner, J., Aasvang, G.M., Abbafati, C., Ozgoren, A.A., Abd-Allah, F., Abera, S.F., Aboyans, V., Abraham, B., Abraham, J.P., Abubakar, I., Abu-Rmeileh, N.M.E., Aburto, T.C., Achoki, T., Adelekan, A., Adofo, K., Adou, A.K., Adsuar, J.C., Afshin, A., Agardh, E.E., Al Khabouri, M.J., Al Lami, F.H., Alam, S.S., Alasfoor, D., Albittar, M.I., Alegretti, M. A., Aleman, A.V., Alemu, Z.A., Alfonso-Cristancho, R., Alhabib, S., Ali, R., Ali, M.K., Alla, F., Allebeck, P., Allen, P.J., Alsharif, U., Alvarez, E., Alvis-Guzman, N., Amankwaa, A.A., Amare, A.T., Ameh, E.A., Ameli, O., Amini, H., Ammar, W., Anderson, B.O., Antonio, C.A.T., Anwari, P., Cunningham, S.A., Arnlöv, J., Arsenijevic, V.S.A., Artaman, A., Asghar, R.J., Assadi, R., Atkins, L.S., Atkinson, C., Avila, M.A., Awuah, B., Badawi, A., Bahit, M.C., Bakfalouni, T., Balakrishnan, K., Balalla, S., Balu, R.K., Banerjee, A., Barber, R.M., Barker-Collo, S.L., Barquera, S., Barregard, L., Barrero, L.H., Barrientos-Gutierrez, T., Basto-Abreu, A.C., Basu, A., Basu, S., Basulaiman, M.O., Ruvalcaba, C.B., Beardsley, J., Bedi, N., Bekele, T., Bell, M.L., Benjet, C., Bennett, D.A., Benzian, H., Bernabé, E., Beyene, T.J., Bhala, N., Bhalla, A., Bhutta, Z.A., Bikbov, B., Abdulhak, A.A.B., Blore, J.D., Blyth, F.M., Bohensky, M.A., Başara, B.B., Borges, G., Bornstein, N.M., Bose, D., Boufous, S., Bourne, R.R., Brainin, M., Brazinova, A., Breitborde, N.J., Brenner, H., Briggs, A.D. M., Broday, D.M., Brooks, P.M., Bruce, N.G., Brugha, T.S., Brunekreef, B., Buchbinder, R., Bui, L.N., Bukhman, G., Bulloch, A.G., Burch, M., Burney, P.G.J., Campos-Nonato, I.R., Campuzano, J.C., Cantoral, A.J., Caravanos, J., Cárdenas, R., Cardis, E., Carpenter, D.O., Caso, V., Castañeda-Orjuela, C.A., Castro, R.E., Catalá- López, F., Cavalleri, F., Çavlin, A., Chadha, V.K., Chang, J., Charlson, F.J., Chen, H., Chen, W., Chen, Z., Chiang, P.P., Chimed-Ochir, O., Chowdhury, R., Christophi, C.A., Chuang, T.-W., Chugh, S.S., Cirillo, M., Claßen, T.K., Colistro, V., Colomar, M., Colquhoun, S.M., Contreras, A.G., Cooper, C., Cooperrider, K., Cooper, L.T., Coresh, J., Courville, K.J., Criqui, M.H., Cuevas-Nasu, L., Damsere-Derry, J., Danawi, H., Dandona, L., Dandona, R., Dargan, P.I., Davis, A., Davitoiu, D.V., Dayama, A., de Castro, E.F., De la Cruz-Góngora, V., De Leo, D., de Lima, G., Degenhardt, L., del Pozo-Cruz, B., Dellavalle, R.P., Deribe, K., Derrett, S., Jarlais, D.C.D., Dessalegn, M., deVeber, G.A., Devries, K.M., Dharmaratne, S.D., Dherani, M.K., Dicker, D., Ding, E. References Velasquez-Melendez, G., Venketasubramanian, N., Vijayakumar, L., Villalpando, S., Violante, F.S., Vlassov, V.V., Vollset, S.E., Wagner, G.R., Waller, S.G., Wallin, M.T., López, F., Cavalleri, F., Çavlin, A., Chadha, V.K., Chang, J., Charlson, F.J., Chen, H., Chen, W., Chen, Z., Chiang, P.P., Chimed-Ochir, O., Chowdhury, R., Christophi, C.A., p , , , , Ç , , , , g, , , , , , Chen, W., Chen, Z., Chiang, P.P., Chimed-Ochir, O., Chowdhury, R., Christophi, C.A., Chuang, T.-W., Chugh, S.S., Cirillo, M., Claßen, T.K., Colistro, V., Colomar, M., Wan, X., Wang, H., Wang, J., Wang, L., Wang, W., Wang, Y., Warouw, T.S., Watts, C. H., Weichenthal, S., Weiderpass, E., Weintraub, R.G., Werdecker, A., Wessells, K.R., Colquhoun, S.M., Contreras, A.G., Cooper, C., Cooperrider, K., Cooper, L.T., Coresh, J., Courville, K.J., Criqui, M.H., Cuevas-Nasu, L., Damsere-Derry, J., Danawi, H., Colquhoun, S.M., Contreras, A.G., Cooper, C., Cooperrider, K., Cooper, L.T., Coresh, J., Courville, K.J., Criqui, M.H., Cuevas-Nasu, L., Damsere-Derry, J., Danawi, H., Westerman, R., Whiteford, H.A., Wilkinson, J.D., Williams, H.C., Williams, T.N., Dandona, L., Dandona, R., Dargan, P.I., Davis, A., Davitoiu, D.V., Dayama, A., de C t E F D l C Gó V D L D d Li G D h dt L d l Dandona, L., Dandona, R., Dargan, P.I., Davis, A., Davitoiu, D.V., Dayama, A., de Castro, E.F., De la Cruz-Góngora, V., De Leo, D., de Lima, G., Degenhardt, L., del Dandona, L., Dandona, R., Dargan, P.I., Davis, A., Davitoiu, D.V., Dayama, A., de Castro, E.F., De la Cruz-Góngora, V., De Leo, D., de Lima, G., Degenhardt, L., del Pozo-Cruz, B., Dellavalle, R.P., Deribe, K., Derrett, S., Jarlais, D.C.D., Dessalegn, M., Pozo Cruz, B., Dellavalle, R.P., Deribe, K., Derrett, S., Jarlais, D.C.D., Dessalegn, M., deVeber, G.A., Devries, K.M., Dharmaratne, S.D., Dherani, M.K., Dicker, D., Ding, E. L., Dokova, K., Dorsey, E.R., Driscoll, T.R., Duan, L., Durrani, A.M., Ebel, B.E., Ellenbogen, R.G., Elshrek, Y.M., Endres, M., Ermakov, S.P., Erskine, H.E., Eshrati, B., Esteghamati, A., Fahimi, S., Faraon, E.J.A., Farzadfar, F., Fay, D.F.J., Feigin, V.L., g , , , , , , , , , , , , Esteghamati, A., Fahimi, S., Faraon, E.J.A., Farzadfar, F., Fay, D.F.J., Feigin, V.L., Feigl, A.B., Fereshtehnejad, S.-M., Ferrari, A.J., Ferri, C.P., Flaxman, A.D., Fleming, T.fi Feigl, A.B., Fereshtehnejad, S.-M., Ferrari, A.J., Ferri, C.P., Flaxman, A.D., Fleming, T. References L., Dokova, K., Dorsey, E.R., Driscoll, T.R., Duan, L., Durrani, A.M., Ebel, B.E., Ellenbogen, R.G., Elshrek, Y.M., Endres, M., Ermakov, S.P., Erskine, H.E., Eshrati, B., Esteghamati, A., Fahimi, S., Faraon, E.J.A., Farzadfar, F., Fay, D.F.J., Feigin, V.L., Feigl, A.B., Fereshtehnejad, S.-M., Ferrari, A.J., Ferri, C.P., Flaxman, A.D., Fleming, T. D., Foigt, N., Foreman, K.J., Paleo, U.F., Franklin, R.C., Gabbe, B., Gaffikin, L., Gakidou, E., Gamkrelidze, A., Gankpé, F.G., Gansevoort, R.T., García-Guerra, F.A., Gasana, E., Geleijnse, J.M., Gessner, B.D., Gething, P., Gibney, K.B., Gillum, R.F., Ginawi, I.A.M., Giroud, M., Giussani, G., Goenka, S., Goginashvili, K., Dantes, H.G., Gona, P., de Cosio, T.G., González-Castell, D., Gotay, C.C., Goto, A., Gouda, H.N., Guerrant, R.L., Gugnani, H.C., Guillemin, F., Gunnell, D., Gupta, Rahul, Gupta, Rajeev, Gutiérrez, R.A., Hafezi-Nejad, N., Hagan, H., Hagstromer, M., Halasa, Y.A., Hamadeh, R.R., Hammami, M., Hankey, G.J., Hao, Y., Harb, H.L., Haregu, T.N., Haro, Forouzanfar, M.H., Alexander, L., Anderson, H.R., Bachman, V.F., Biryukov, S., Brauer, M., Burnett, R., Casey, D., Coates, M.M., Cohen, A., Delwiche, K., Estep, K., Frostad, J. J., Kc, A., Kyu, H.H., Moradi-Lakeh, M., Ng, M., Slepak, E.L., Thomas, B.A., Wagner, J., Aasvang, G.M., Abbafati, C., Ozgoren, A.A., Abd-Allah, F., Abera, S.F., Aboyans, V., Abraham, B., Abraham, J.P., Abubakar, I., Abu-Rmeileh, N.M.E., Aburto, T.C., Achoki, T., Adelekan, A., Adofo, K., Adou, A.K., Adsuar, J.C., Afshin, A., Agardh, E.E., Al Khabouri, M.J., Al Lami, F.H., Alam, S.S., Alasfoor, D., Albittar, M.I., Alegretti, M. A., Aleman, A.V., Alemu, Z.A., Alfonso-Cristancho, R., Alhabib, S., Ali, R., Ali, M.K., Alla, F., Allebeck, P., Allen, P.J., Alsharif, U., Alvarez, E., Alvis-Guzman, N., Amankwaa, A.A., Amare, A.T., Ameh, E.A., Ameli, O., Amini, H., Ammar, W., Anderson, B.O., Antonio, C.A.T., Anwari, P., Cunningham, S.A., Arnlöv, J., Arsenijevic, V.S.A., Artaman, A., Asghar, R.J., Assadi, R., Atkins, L.S., Atkinson, C., Avila, M.A., Awuah, B., Badawi, A., Bahit, M.C., Bakfalouni, T., Balakrishnan, K., Balalla, S., Balu, R.K., Banerjee, A., Barber, R.M., Barker-Collo, S.L., Barquera, S., Barregard, L., Barrero, L.H., Barrientos-Gutierrez, T., Basto-Abreu, A.C., Basu, A., Basu, S., Basulaiman, M.O., Ruvalcaba, C.B., Beardsley, J., Bedi, N., Bekele, T., Bell, M.L., Benjet, C., Bennett, D.A., Benzian, H., Bernabé, E., Beyene, T.J., Bhala, N., Bhalla, A., Bhutta, Z.A., Bikbov, B., Abdulhak, A.A.B., Blore, J.D., Blyth, F.M., Bohensky, M.A., Başara, B.B., Borges, G., Bornstein, N.M., Bose, D., Boufous, S., Bourne, R.R., Brainin, M., Brazinova, A., Breitborde, N.J., Brenner, H., Briggs, A.D. References M., Broday, D.M., Brooks, P.M., Bruce, N.G., Brugha, T.S., Brunekreef, B., Buchbinder, R., Bui, L.N., Bukhman, G., Bulloch, A.G., Burch, M., Burney, P.G.J., Campos-Nonato, I.R., Campuzano, J.C., Cantoral, A.J., Caravanos, J., Cárdenas, R., Cardis, E., Carpenter, D.O., Caso, V., Castañeda-Orjuela, C.A., Castro, R.E., Catalá- López, F., Cavalleri, F., Çavlin, A., Chadha, V.K., Chang, J., Charlson, F.J., Chen, H., Quistberg, D.A.A., Racapé, L., Rafay, A., Rahimi, K., Rahimi-Movaghar, V., Rahman, S.U., Raju, M., Rakovac, I., Rana, S.M., Rao, M., Razavi, H., Reddy, K.S., Refaat, A.H., S.U., Raju, M., Rakovac, I., Rana, S.M., Rao, M., Razavi, H., Reddy, K.S., Refaat, A.H., , j , , , , , , , , , , y, , , , Rehm, J., Remuzzi, G., Ribeiro, A.L., Riccio, P.M., Richardson, L., Riederer, A., Robinson, M., Roca, A., Rodriguez, A., Rojas-Rueda, D., Romieu, I., Ronfani, L., om, R., Roy, N., Ruhago, G.M., Rushton, L., Sabin, N., Sacco, R.L., Sa Sahathevan, R., Sahraian, M.A., Salomon, J.A., Salvo, D., Sampson, U.K., Sanabria, J. Sahathevan, R., Sahraian, M.A., Salomon, J.A., Salvo, D., Sampson, U.K., Sanabria, J. R Sanchez L M Sánchez-Pimienta T G Sanchez-Riera L Sandar L Santos I S Sahathevan, R., Sahraian, M.A., Salomon, J.A., Salvo, D., Sampson, U.K., Sanabria, J. References uchbinder, R., Bui, L.N., Bukhman, G., Bulloch, A.G., Burch, M., Burney, P.G.J., , , , , , , , , , , y, , Campos-Nonato, I.R., Campuzano, J.C., Cantoral, A.J., Caravanos, J., Cárdenas, R., C di E C D O C V C ñ d O j l C A C R E C lá N., Uzun, S.B., van de Vijver, S., Van Dingenen, R., van Gool, C.H., van Os, J., Varakin, Y.Y., Vasankari, T.J., Vasconcelos, A.M.N., Vavilala, M.S., Veerman, L.J., N., Uzun, S.B., van de Vijver, S., Van Dingenen, R., van Gool, C.H., van Os, J., ampos-Nonato, I.R., Campuzano, J.C., Cantoral, A.J., Caravanos, J., Cá Campos Nonato, I.R., Campuzano, J.C., Cantoral, A.J., Caravanos, J., Cárdenas, R., Cardis, E., Carpenter, D.O., Caso, V., Castañeda-Orjuela, C.A., Castro, R.E., Catalá- Cardis, E., Carpenter, D.O., Caso, V., Castañeda-Orjuela, C.A., Castro, R.E., Catalá- arakin, Y.Y., Vasankari, T.J., Vasconcelos, A.M.N., Vavilala, M.S., Veerman, L.J., Varakin, Y.Y., Vasankari, T.J., Vasconcelos, A.M.N., Vavilala, M.S., Veerman, L.J., Velasquez-Melendez, G., Venketasubramanian, N., Vijayakumar, L., Villalpando, S., Violante, F.S., Vlassov, V.V., Vollset, S.E., Wagner, G.R., Waller, S.G., Wallin, M.T., Wan, X., Wang, H., Wang, J., Wang, L., Wang, W., Wang, Y., Warouw, T.S., Watts, C. H., Weichenthal, S., Weiderpass, E., Weintraub, R.G., Werdecker, A., Wessells, K.R., Westerman, R., Whiteford, H.A., Wilkinson, J.D., Williams, H.C., Williams, T.N., Woldeyohannes, S.M., Wolfe, C.D.A., Wong, J.Q., Woolf, A.D., Wright, J.L., Wurtz, B., Xu, G., Yan, L.L., Yang, G., Yano, Y., Ye, P., Yenesew, M., Yentür, G.K., Yip, P., Yonemoto, N., Yoon, S.-J., Younis, M.Z., Younoussi, Z., Yu, C., Zaki, M.E., Zhao, Y., Zheng, Y., Zhou, M., Zhu, J., Zhu, S., Zou, X., Zunt, J.R., Lopez, A.D., Vos, T., Murray, C.J., 2015. Global, regional, and national comparative risk assessment of 79 beha- vioural, environmental and occupational, and metabolic risks or clusters of risks in 188 countries, 1990–2013: a systematic analysis for the Global Burden of Disease Study 2013. Lancet 386, 2287–2323. doi:https://doi.org/10.1016/S0140-6736(15) 00128-2. References R., Sanchez, L.M., Sánchez-Pimienta, T.G., Sanchez-Riera, L., Sandar, L., Santos, I.S., Sapkota, A., Satpathy, M., Saunders, J.E., Sawhney, M., Saylan, M.I., Scarborough, P., Schmidt, J.C., Schneider, I.J.C., Schöttker, B., Schwebel, D.C., Scott, J.G., Seedat, S., Sepanlou, S.G., Serdar, B., Servan-Mori, E.E., Shaddick, G., Shahraz, S., Levy, T.S., Shangguan, S., She, J., Sheikhbahaei, S., Shibuya, K., Shin, H.H., Shinohara, Y., Shiri, R., Shishani, K., Shiue, I., Sigfusdottir, I.D., Silberberg, D.H., Simard, E.P., Sindi, S., Anderson, B.O., Antonio, C.A.T., Anwari, P., Cunningham, S.A., Arn Shangguan, S., She, J., Sheikhbahaei, S., Shibuya, K., Shin, H.H., Shinohara, Y., Shiri, R., Shishani, K., Shiue, I., Sigfusdottir, I.D., Silberberg, D.H., Simard, E.P., Sindi, S., Arsenijevic, V.S.A., Artaman, A., Asghar, R.J., Assadi, R., Atkins, L.S., Atkinson, C., Avila M A Awuah B Badawi A Bahit M C Bakfalouni T Balakrishnan K , , , , , g , , g, , , , , , Singh, A., Singh, G.M., Singh, J.A., Skirbekk, V., Sliwa, K., Soljak, M., Soneji, S., Balalla, S., Balu, R.K., Banerjee, A., Barber, R.M., Barker-Collo, S.L., Barquera, S., øreide, K., Soshnikov, S., Sposato, L.A., Sreeramareddy, C.T., Stapelberg, N.J.C., Straif, K., Stroumpoulis, K., Sturua, L., Sunguya, B.F., Swaminathan, S., Swaroop, M., Basu, S., Basulaiman, M.O., Ruvalcaba, C.B., Beardsley, J., Bedi, N., Bekele, T., Bell, L., Benjet, C., Bennett, D.A., Benzian, H., Bernabé, E., Beyene, T.J., Bhala, N., ykes, B.L., Tabb, K.M., Takahashi, K., Talongwa, R.T., Tandon, N., Tanne, D., A., Bhutta, Z.A., Bikbov, B., Abdulhak, A.A.B., Blore, J.D., Blyth, F.M S., Texcalac-Sangrador, J.L., Thackway, S.V., Thomson, B., Thorne-Lyman, A.L., Bohensky, M.A., Başara, B.B., Borges, G., Bornstein, N.M., Bose, D., Boufous, S., Thrift, A.G., Thurston, G.D., Tillmann, T., Tobollik, M., Tonelli, M., Topouzis, F., Towbin J A Toyoshima H Traebert J Tran B X Trasande L Trillini M Bourne, R.R., Brainin, M., Brazinova, A., Breitborde, N.J., Brenner, H., Briggs, A.D. M., Broday, D.M., Brooks, P.M., Bruce, N.G., Brugha, T.S., Brunekreef, B., Thrift, A.G., Thurston, G.D., Tillmann, T., Tobollik, M., Tonelli, M., Topouzis, F., Towbin, J.A., Toyoshima, H., Traebert, J., Tran, B.X., Trasande, L., Trillini, M., Trujillo, U., Dimbuene, Z.T., Tsilimbaris, M., Tuzcu, E.M., Uchendu, U.S., Ukwaja, K. References This study has several notable strengths. First, the estimates were derived from three large nationally representative samples. This survey 8 Preventive M edicine Reports 14 (2019) 100850 G.M. Al Kibria Preventive M edicine Reports 14 (2019) 100850 Yardley, L., 2018. Using the person-based approach to optimise a digital intervention for the management of hypertension. PLoS One 13, e0196868. https://doi.org/10. 1371/journal.pone.0196868. Bundy, J.D., Mills, K.T., Chen, J., Li, C., Greenland, P., He, J., 2018. Estimating the as- sociation of the 2017 and 2014 hypertension guidelines with cardiovascular events and deaths in US adults: an analysis of national data. JAMA Cardiol. 3, 572. https:// doi.org/10.1001/jamacardio.2018.1240. Burke, G.L., Bertoni, A.G., Shea, S., Tracy, R., Watson, K.E., Blumenthal, R.S., Chung, H., Carnethon, M.R., 2008. The impact of obesity on cardiovascular disease risk factors and subclinical vascular disease: the multi-ethnic study of atherosclerosis. Arch. Intern. Med. 168, 928. https://doi.org/10.1001/archinte.168.9.928. shnaswami, S., Kromhout, H., Ku, T., Defo, B.K., Bicer, B.K., Kuipers, Kulkarni, C., Kulkarni, V.S., Kumar, G.A., Kwan, G.F., Lai, T., Balaji, A.L., Lalloo, R., Lallukka, T., Lam, H., Lan, Q., Lansingh, V.C., Larson, H.J., Larsson, A., Laryea, D.O., , , , , , , , , , , j , , , , Lallukka, T., Lam, H., Lan, Q., Lansingh, V.C., Larson, H.J., Larsson, A., Laryea, D.O., Centers for Disease Control and Prevention, 2009. National Health and Nutrition Examination Survey - Health Tech/Blood Pressure Procedures Manual. Chalmers, J., MacMahon, S., Mancia, G., Whitworth, J., Beilin, L., Hansson, L., Neal, B., Rodgers, A., Ni Mhurchu, C., Clark, T., 1999. 1999 World Health Organization- International Society of Hypertension Guidelines for the management of hyperten- sion. Guidelines sub-committee of the World Health Organization. Clin. Exp. Hypertens. 21, 1009–1060. https://doi.org/10.3109/10641969909061028. M., Li, Yichong, Li, Yongmei, Liang, J., Liang, X., Lim, S.S., Lindsay, M.P., Lipshultz, S.E., Liu, S., Liu, Y., Lloyd, B.K., Logroscino, G., London, S.J., Lopez, N., Lortet- y g p Tieulent, J., Lotufo, P.A., Lozano, R., Lunevicius, R., Ma, J., Ma, S., Machado, V.M.P., acIntyre, M.F., Magis-Rodriguez, C., Mahdi, A.A., Majdan, M., Malekzadeh, R., Chang, T.E., Ritchey, M.D., Ayala, C., Durthaler, J.M., Loustalot, F., 2018. Use of stra- tegies to improve antihypertensive medication adherence within United States out- patient health care practices, DocStyles 2015–2016. The Journal of Clinical Hypertension 20, 225–232. https://doi.org/10.1111/jch.13188. References Modeled state-level estimates of hypertension prevalence and un- diagnosed hypertension among US adults during 2013–2015. The Journal of Clinical Hypertension 20, 1395–1410. https://doi.org/10.1111/jch.13388. Goff, D.C., Lloyd-Jones, D.M., Bennett, G., Coady, S., D’Agostino, R.B., Gibbons, R., Greenland, P., Lackland, D.T., Levy, D., O’Donnell, C.J., Robinson, J.G., Schwartz, J.S., Shero, S.T., Smith, S.C., Sorlie, P., Stone, N.J., Wilson, P.W.F., 2014. American College of Cardiology/American Heart Association Task Force on Practice Guidelines. In: 2013 ACC/AHA guideline on the assessment of cardiovascular risk: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. J. Am. Coll. Cardiol. 63. pp. 2935–2959. https://doi.org/10.1016/j.jacc. 2013.11.005.f Redmond, N., Baer, H.J., Hicks, L.S., 2011. Health behaviors and racial disparity in blood pressure control in the National Health and nutrition examination survey. Hypertension 57, 383–389. https://doi.org/10.1161/HYPERTENSIONAHA.110. 161950. Scholes, S., Bajekal, M., Love, H., Hawkins, N., Raine, R., O'Flaherty, M., Capewell, S., 2012. Persistent socioeconomic inequalities in cardiovascular risk factors in England over 1994–2008: a time-trend analysis of repeated cross-sectional data. BMC Public Health 12. https://doi.org/10.1186/1471-2458-12-129. Gu, A., Yue, Y., Desai, R.P., Argulian, E., 2017. Racial and ethnic differences in anti- hypertensive medication use and blood pressure control among US adults with hy- pertension: the National Health and Nutrition Examination Survey, 2003 to 2012. Circulation: Cardiovascular Quality and Outcomes 10. https://doi.org/10.1161/ CIRCOUTCOMES.116.003166. Health 12. https://doi.org/10.1186/1471-2458-12-129. Stata Corporation, College Station, Texas USA, 2015. Stata 14.0 [WWW Document]. URL (accessed 5.8.17). U.S. Department of Health & Human Services, 2019. Poverty guidelines [WWW docu- ment]. URL. https://aspe.hhs.gov/poverty-guidelines, Accessed date: 16 February 2019. Heron, M., 2018. National Vital Statistics Reports National Center for Health Statistics. 2018. (Annual No. 67 (6)). Hyattsville, MD. US Census Bureau, 2015. Age and sex composition in the United States: 2015. [WWW Document]. URL. https://www.census.gov/data/tables/2015/demo/age-and-sex/ 2015-age-sex-composition.html, Accessed date: 16 December 2018. Hypertension Study Group, 2001. Prevalence, awareness, treatment and control of hy- pertension among the elderly in Bangladesh and India: a multicentre study. Bull. World Health Organ. 79, 490–500. Johnson, C., Burt, Vi.L, Mohadjer, L.K., 2014. National Health and Nutrition Examination Survey: Sample Design, 2011–2014. Wang, T.J., Vasan, R.S., 2005. Epidemiology of uncontrolled hypertension in the United States. Circulation 112, 1651–1662. https://doi.org/10.1161/CIRCULATIONAHA. 104.490599. Kang, H., 2013. The prevention and handling of the missing data. Korean Journal of Anesthesiology 64, 402. https://doi.org/10.4097/kjae.2013.64.5.402. References D., Foigt, N., Foreman, K.J., Paleo, U.F., Franklin, R.C., Gabbe, B., Gaffikin, L., fi Gakidou, E., Gamkrelidze, A., Gankpé, F.G., Gansevoort, R.T., García-Guerra, F.A., Gasana, E., Geleijnse, J.M., Gessner, B.D., Gething, P., Gibney, K.B., Gillum, R.F., Forouzanfar, M.H., Liu, P., Roth, G.A., Ng, M., Biryukov, S., Marczak, L., Alexander, L., Estep, K., Hassen Abate, K., Akinyemiju, T.F., Ali, R., Alvis-Guzman, N., Azzopardi, P., Banerjee, A., Bärnighausen, T., Basu, A., Bekele, T., Bennett, D.A., Biadgilign, S., Catalá-López, F., Feigin, V.L., Fernandes, J.C., Fischer, F., Gebru, A.A., Gona, P., Gupta, R., Hankey, G.J., Jonas, J.B., Judd, S.E., Khang, Y.-H., Khosravi, A., Kim, Y.J., Kimokoti, R.W., Kokubo, Y., Kolte, D., Lopez, A., Lotufo, P.A., Malekzadeh, R., Melaku, Y.A., Mensah, G.A., Misganaw, A., Mokdad, A.H., Moran, A.E., Nawaz, H., Neal, B., Ngalesoni, F.N., Ohkubo, T., Pourmalek, F., Rafay, A., Rai, R.K., Rojas- Rueda, D., Sampson, U.K., Santos, I.S., Sawhney, M., Schutte, A.E., Sepanlou, S.G., Shifa, G.T., Shiue, I., Tedla, B.A., Thrift, A.G., Tonelli, M., Truelsen, T., Tsilimparis, Ginawi, I.A.M., Giroud, M., Giussani, G., Goenka, S., Goginashvili, K., Dantes, H.G., Gona, P., de Cosio, T.G., González-Castell, D., Gotay, C.C., Goto, A., Gouda, H.N., R.L., Gugnani, H.C., Guillemin, F., Gunnell, D., Gupta, Rahul, Gupta, Rajeev, Gutiérrez, R.A., Hafezi-Nejad, N., Hagan, H., Hagstromer, M., Halasa, Y.A., Heuton, K.R., Heydarpour, P., Hijar, M., Hoek, H.W., Hoffman, H.J., Hornberger, J. C., Hosgood, H.D., Hoy, D.G., Hsairi, M., Hu, G., Hu, H., Huang, C., Huang, J.J., 9 Preventive M edicine Reports 14 (2019) 100850 G.M. Al Kibria N., Ukwaja, K.N., Uthman, O.A., Vasankari, T., Venketasubramanian, N., Vlassov, V.V., Vos, T., Westerman, R., Yan, L.L., Yano, Y., Yonemoto, N., Zaki, M.E.S., Murray, C.J.L., 2017. Global burden of hypertension and systolic blood pressure of at least 110 to 115 mm Hg, 1990–2015. JAMA 317, 165–182. https://doi.org/10.1001/jama. 2016.19043. approval [WWW document]. URL. https://www.cdc.gov/nchs/nhanes/irba98.htm, Accessed date: 11 September 2018. approval [WWW document]. URL. https://www.cdc.gov/nchs/nhanes/irba98.htm, Accessed date: 11 September 2018. Ostchega, Y., Zhang, G., Hughes, J.P., Nwankwo, T., 2018. Factors associated with hy- pertension control in US adults using 2017 ACC/AHA guidelines: national health and nutrition examination survey 1999–2016. Am. J. Hypertens. 31, 886–894. https:// doi.org/10.1093/ajh/hpy047. Giles, T., Aranda, J.M., Suh, D.-C., Choi, I.-S., Preblick, R., Rocha, R., Frech-Tamas, F., 2007. Ethnic/racial variations in blood pressure awareness, treatment, and control. J Clin Hypertens (Greenwich) 9, 345–354.f Park, S., Gillespie, C., Baumgardner, J., Yang, Q., Valderrama, A.L., Fang, J., Loustalot, F., Hong, Y., 2018. References Whelton, P.K., Carey, R.M., Aronow, W.S., Casey, D.E., Collins, K.J., Dennison Himmelfarb, C., DePalma, S.M., Gidding, S., Jamerson, K.A., Jones, D.W., MacLaughlin, E.J., Muntner, P., Ovbiagele, B., Smith, S.C., Spencer, C.C., Stafford, R.S., Taler, S.J., Thomas, R.J., Williams, K.A., Williamson, J.D., Wright, J.T., 2017. 2017 ACC/AHA/AAPA/ABC/ACPM/AGS/APhA/ASH/ASPC/NMA/PCNA guideline for the prevention, detection, evaluation, and management of high blood pressure in adults: a report of the American College of Cardiology/American Heart Association task force on clinical practice guidelines. Hypertension. https://doi.org/10.1161/ HYP.0000000000000065. HYP.0000000000000065. Karlamangla, A.S., Merkin, S.S., Crimmins, E.M., Seeman, T.E., 2010. Socioeconomic and ethnic disparities in cardiovascular risk in the United States, 2001-2006. Ann. Epidemiol. 20, 617–628. https://doi.org/10.1016/j.annepidem.2010.05.003. Khera, R., Lu, Y., Lu, J., Saxena, A., Nasir, K., Jiang, L., Krumholz, H.M., 2018. Impact of 2017 ACC/AHA guidelines on prevalence of hypertension and eligibility for anti- hypertensive treatment in United States and China: nationally representative cross sectional study. BMJ. https://doi.org/10.1136/bmj.k2357. k2357. Levey, A.S., Stevens, L.A., Schmid, C.H., Zhang, Y.L., Castro, A.F., Feldman, H.I., Kusek, J.W., Eggers, P., Van Lente, F., Greene, T., Coresh, J., CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration), 2009. A new equation to estimate glomerular filtration rate. Ann. Intern. Med. 150, 604–612. Yano, Y., Reis, J.P., Colangelo, L.A., Shimbo, D., Viera, A.J., Allen, N.B., Gidding, S.S., Bress, A.P., Greenland, P., Muntner, P., Lloyd-Jones, D.M., 2018. Association of blood pressure classification in young adults using the 2017 American College of Cardiology/American Heart Association blood pressure guideline with cardiovascular events later in life. JAMA 320, 1774. https://doi.org/10.1001/jama.2018.13551. i Muntner, P., Carey, R.M., Gidding, S., Jones, D.W., Taler, S.J., Wright, J.T., Whelton, P.K., 2017. Potential U.S. population impact of the 2017 American College of Cardiology/ American Heart Association high blood pressure guideline. Circulation. https://doi. org/10.1161/CIRCULATIONAHA.117.032582. Zipf, G., Chiappa, M., Porter, K.S., Ostchega, Y., Lewis, B.G., Dostal, J., 2013. National health and nutrition examination survey: plan and operations, 1999–2010. Vital Health Stat. 1, 1–37. g/ / National Center for Health Statistics, 2017. NCHS research ethics review board (ERB) 10 10
https://openalex.org/W4223901416	https://pubs.rsc.org/en/content/articlepdf/2022/ra/d2ra00697a	English	null	Experimentally designed chemometric models for the assay of toxic adulterants in turmeric powder	RSC advances	2,022	cc-by	6,929	aAnalytical Chemistry Department, Faculty of Pharmacy, October 6 University, PO Box 12858, 6 October City, Giza, Egypt bPharmacognosy Department, Faculty of Pharmacy, October 6 University, PO Box 12858, 6 October City, Giza, Egypt cAnalytical Chemistry Department, Faculty of Pharmacy, Cairo University, El-Kasr El- Aini Street, Cairo, 11562, Egypt. E-mail: ahmed.bayoumy@pharma.cu.edu.eg; ahmedss_pharm@yahoo.com dMedicinal Chemistry Department, PharmD Program, Egypt-Japan University of Science and Technology (E-JUST), New Borg El-Arab City, Alexandria, 21934, Egypt † Electronic supplementary information (ESI) available. See DOI: 10.1039/d2ra00697a Open Access Article. Published on 23 March 2022. Downloaded on 10/24/2024 6: This article is licensed under a Creative Commons Attribution 3.0 U Turmeric is an indispensable culinary spice in diﬀerent cultures and a principal component in traditional remedies. Toxic metanil yellow (MY), acid orange 7 (AO) and lead chromate (LCM) are deliberately added to adulterate turmeric powder. This work compares the ability of multivariate chemometric models with those of artiﬁcial intelligent networks to enhance the selectivity of spectral data for the rapid assay of these three adulterants in turmeric powder. Using a custom experimental design, we provide a data- driven optimization for the sensitive parameters of the partial least squares model (PLS), artiﬁcial neural network (ANN) and genetic algorithm (GA). The optimized models are validated using sets of genuine turmeric samples from ﬁve diﬀerent geographical regions spiked with standard adulterant concentrations. The optimized GA-PLS and GA-ANN models reduce the root mean square error of prediction by 18.4%, 31.1% and 55.3% and 25.0%, 69.9% and 88.4% for MY, AO and LCM, respectively. Received 1st February 2022 Accepted 7th March 2022 DOI: 10.1039/d2ra00697a rsc.li/rsc-advances eye conditions,5 kidney and heart diseases,3 cancer,6 rheuma- toid arthritis,7 and several psychiatric disorders.8 1. Introduction Food provides the human body with the energy necessary to function and exist, provided that safety and quality are guar- anteed. The latter is sometimes perturbed deliberately through the inclusion of inferior admixtures or the exclusion of valuable ingredients to increase prot margins at the expense of customer health and social consequences. Laws, traditions and religions incriminate any form of the above practices. The relatively high global consumption and demand for turmeric in diﬀerent applications (such as nutraceuticals, food avorings and cosmetics) make it more vulnerable to adultera- tion with low-quality ingredients such as starch, chalk, yellow soapstone, lead chromate and synthetic dyes.9 The mentioned adulterants can result in cardiovascular, neurological, hepato- toxic and nephrotoxic health hazards for consumers. For instance, lead chromate (LCM) and synthetic dyes such as met- anil yellow (MY) and acid orange 7 (AO) are deliberately added to mimic the color appearance of turmeric despite their hazardous health eﬀects.9 The electron-withdrawing character of the azo group in the synthetic dyes sometimes develops an electron deciency and is reduced to carcinogenic amino compounds.10 Turmeric rhizome, known as Indian saﬀron, is one of the largest selling natural food products in the world, and receives a great deal of attention from both medical and culinary specialists.1 Turmeric comprises more than 100 compounds. Curcumin is the main active compound and is credited with most of turmeric's health benets due to its antioxidant and anti-inammatory properties.2 Turmeric can be used to avor or color many food substances, such as curry powder, mustard, butter and cheese. Medically, it treats several inammatory conditions,3 metabolic syndromes,4 inammatory degenerative Long-term consumption of metanil yellow (MY) (sodium 3- [4-anilinophenylazo]benzene sulfonate) causes severe damage to the heart and nervous tissues,11,12 degenerative changes in the lining of the stomach, kidneys and liver,13 as well as adversely aﬀects the ovaries and testes.14,15 Acid orange 7 (AO) (sodium 4-[(2E)-2-(2-oxo-naphthalene-1- ylidene)hydrazinyl]benzene sulfonate) and lead chromate are deliberately added to adulterate turmeric powder.16,17 Acid orange 7 irritates the eyes, skin, mucous membrane and upper respiratory tract in addition to causing severe headaches, nausea, water-borne diseases, such as dermatitis, and to loss of bone marrow, leading to anemia.16 Lead chromate (LCM) can lead to lung and renal cancer, neuropathy, osteopathy, and respiratory tract toxicity.18 Analytical chemistry guarantees food safety and quality through the development of reliable sensitive and selective © 2022 The Author(s). RSC Advances Cite this: RSC Adv., 2022, 12, 9087 Received 1st February 2022 Accepted 7th March 2022 DOI: 10.1039/d2ra00697a rsc.li/rsc-advances PAPER der a Creative Commons Attribution 3.0 Unported Licence. Open Access Article. Published on 23 March 2022. Downloaded on 10/24/2024 6:29:14 AM. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. Experimentally designed chemometric models for the assay of toxic adulterants in turmeric powder† Shymaa S. Soliman,a Alaadin E. El-Haddad,b Ghada A. Sedik,c Mohamed R. Elghobashy,ca Hala E. Zaazaac and Ahmed S. Saad cd © 2022 The Author(s). Published by the Royal Society of Chemistry Experimentally designed chemometric models for the assay of toxic adulterants in turmeric powder† Shymaa S. Soliman,a Alaadin E. El-Haddad,b Ghada A. Sedik,c Mohamed R. Elghobashy,ca Hala E. Zaazaac and Ahmed S. Saad cd Cite this: RSC Adv., 2022, 12, 9087 Shymaa S. Soliman,a Alaadin E. El-Haddad,b Ghada A. Sedik,c Mohamed R. Elghobashy,ca Hala E. Zaazaac and Ahmed S. Saad *cd 1. Introduction Published by the Royal Society of Chemistry RSC Adv., 2022, 12, 9087–9094 \| 9087 Paper View Article Online Paper View Article Online View Article Online P View Article Online RSC Advances analytical methods to detect and quantify food adulterants and food ingredients.19–22 Analytical reports provide authorities, food suppliers and consumers with the evidence necessary to build condence in food authenticity, safety and quality.23 2.3 Standard solutions Stock standard solutions (500 mg mL1) of MY, AO and LCM were prepared using NaOH (0.2 M) as a solvent. 2.2 Instrumentation The spectrophotometric measurements were carried out using a Shimadzu UV-visible spectrophotometer dual beam, model UV-1800 with a 1 cm quartz cell supplied with UV-Probe 2.32 soware (Shimadzu Scientic Instruments inc., Kyoto, Japan). All chemometric methods were implemented in MATLAB® 8.1.0.604 (R2013a) and the PLS version 2.1 toolbox. Design- Expert® 13.0.1.0 soware was used to analyze the results. Open Access Article. Published on 23 March 2022. Downloaded on 10/24/2024 6:29:14 AM. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. Methanol, ethanol, acetonitrile, acetone, acetic acid, nitric acid and sodium hydroxide (NaOH) pellets were purchased from El-NASR Pharmaceutical Chemicals Co., (Egypt). Sodium hydroxide solution (0.2 M) was prepared in distilled water and used as a solvent. The DoE technique has been used extensively in several elds, such as chemistry,24–29 agriculture,30,31 engineering,32,33 industry34,35 (in particular food industry36–38), and even in non- scientic aspects, such as predicting sports results,39 espe- cially soccer and basketball. DoE can be used in other disci- plines, such as energy40–44 and sensors optimization.45–49 Previously, many empirical studies in science materials have been made through one factor at a time experimentation (OFAT), which provides uncorrelated material-based systems. This may be attributed to the complexity, safety-critical nature of the energy systems, the high number of components, and the processing conditions. However, DoE evaluates the contribu- tion of diﬀerent factors simultaneously and denes the needed redundancies for meaningful statistical assessment of the outcomes, allowing for the consistent establishment of energy behavioral-based strategies, a good prediction of energy yields, and integral optimization of electrochemical sensors for higher sensitivity and selectivity. 2.4 Procedure 2.4.1 Spectral characteristics. Zero-order (D0) absorption spectra of separate solutions of 10 mg mL1 MY, AO and LCM, and 200 mg mL1 clear supernatant of each turmeric extract were recorded against NaOH solution (0.2 M) as a blank over the wavelength range of 200–800 nm. A literature survey revealed several methods to detect and/or quantify the adulteration of turmeric, such as HPLC,50,51 HPTLC,52,53 voltammetry,54 multispectral imaging,55 spectro- photometry56 and FTIR.57–60 However, the reported literature failed to reveal a single analytical method for the simultaneous assay of MY, AO and LCM in turmeric powder. 2.4.2 Preparation of the calibration and validation sets. Sets of 25 and 5 mixtures were used for the construction of the calibration set and the validation set, respectively. Each mixture contains diﬀerent concentrations of the three adulterants spiked to pure turmeric powder. Accurate 20 mg weights of pure turmeric powder (from ve diﬀerent sources: RG, ABU, HZ, MD and DA) were transferred separately into falcon tubes (50 mL), and spiked with standard MY (in the range 200, 400, 600, 800 and 1000 mg), standard AO (in the range 800, 1000, 1200, 1400 and 1600 mg) and standard LCM (in the range 2000, 2500, 3000, 3500 and 4000 mg) according to the multilevel multifactor design. The spiked turmeric powder was sonicated with NaOH (0.2 M, 50 mL, for 15 min), ltered, and the clear ltrate was quantitatively transferred into a volumetric ask (100 mL) and completed using the same solvent. In the current work, an investigation and comparison were conducted on the ability of (1) traditional chemometric models, (2) articially intelligent neural networks, (3) genetic algorithm variable selection tool, and nally (4) experimentally designed optimization for discrimination and simultaneous quantitation of toxic adulterants in the complex natural matrix of turmeric powder. Once fed with a single UV-spectrum of the turmeric powder extract, the models decode the UV-absorbances to discriminate and quantify the adulterants within the complex turmeric powder. The ICH validation parameters were computed to ensure the validity of the method. The model represents a simple, direct and fast analytical tool for quality control laboratories to investigate turmeric samples, and support food suppliers and authorities with scientic evidence regarding food safety and quality. 2.4.3 Wavelength range selection. Diﬀerent wavelength ranges were sought to select the optimum range that achieves 2.4.3 Wavelength range selection. 9088 \| RSC Adv., 2022, 12, 9087–9094 © 2022 The Author(s). Published by the Royal Society of Chemistry 2.1 Samples and reagents Standard metanil yellow (purity 98.00%) and acid orange 7 (purity 98.00%) were purchased from Techno Pharmchem (India). Lead chromate standard (purity 98.00%) was purchased from OTTO (India). Genuine turmeric rhizomes (Curcuma longa L.) were purchased from the local markets in Egypt: Ragab (RG), Abu-Auf (ABU), Harraz (HZ) and Medrar (MD). In addition, rhizomes were obtained from Danube (DA) in a Saudi Arabian market. They were kindly identied by the Pharmacognosy Department, Faculty of Pharmacy, October 6 University, Cairo, Egypt. The rhizomes were ground separately to obtain the pure turmeric powder. Analyzing pharmaceutical and food products is a highly multivariate process, as many factors (parameters) may be involved during the analysis. Powerful data collection and statistical tools are required to identify the factors that may aﬀect the results of the experiment. Design of experiment technique (DoE) is a robust statistical tool that is successfully deployed in diﬀerent types of systems, product design, process development and optimization. It plays a prevalent role in decision-making processes through providing useful experi- mental information, and planning the type and the number of experiments to reach the optimal parameters settings. Conse- quently, this allows for overcoming the inability defects of generating a large amount of useful experimental data with proper human interpretation. Open Access Article. Published on 23 March 2022. Downloaded on 10/24/2024 6:29:14 AM. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. 2.4 Procedure Diﬀerent wavelength ranges were sought to select the optimum range that achieves 9088 \| RSC Adv., 2022, 12, 9087–9094 © 2022 The Author(s). Published by the Royal Society of Chemistry RSC Advances View Article Online RSC Advances View Article Online View Article Online Paper higher sensitivity and selectivity of the proposed models to avoid the noisy regions and poorly informative wavelength range. adulterants. Three concentrations of MY (2, 4 and 10 mg mL1), AO (10, 12 and 16 mg mL1) and LCM (20, 25 and 40 mg mL1) were analyzed three times intra-daily and on three successive days. The % RSD of the three adulterants were calculated. adulterants. Three concentrations of MY (2, 4 and 10 mg mL1), AO (10, 12 and 16 mg mL1) and LCM (20, 25 and 40 mg mL1) were analyzed three times intra-daily and on three successive days. The % RSD of the three adulterants were calculated. adulterants. Three concentrations of MY (2, 4 and 10 mg mL1), AO (10, 12 and 16 mg mL1) and LCM (20, 25 and 40 mg mL1) were analyzed three times intra-daily and on three successive days. The % RSD of the three adulterants were calculated. 2.4.4 Construction of the PLS models. Zero-order absorp- tion spectra of the three adulterants (MY, AO, LCM) were recorded in the wavelength range of 200–800 nm using NaOH (0.2 M) as solvent. The wavelengths in the range of 230–570 nm were selected during the analysis, as the three analytes exhibit adequate absorbance within the working concentration ranges. Cross-validation was carried out using the leave-one-out method, and the root mean square error of cross-validation (RMSECV) was computed and used to obtain the optimum number of latent variables. 2.4.4 Construction of the PLS models. Zero-order absorp- tion spectra of the three adulterants (MY, AO, LCM) were recorded in the wavelength range of 200–800 nm using NaOH (0.2 M) as solvent. The wavelengths in the range of 230–570 nm were selected during the analysis, as the three analytes exhibit adequate absorbance within the working concentration ranges. Cross-validation was carried out using the leave-one-out method, and the root mean square error of cross-validation (RMSECV) was computed and used to obtain the optimum number of latent variables. 2.4.10 Application. The literature failed to reveal a quanti- tative method to assay the three adulterants simultaneously in pure turmeric powder. 3.2 Multivariate analysis of diﬀerent turmeric rhizomes Five diﬀerent sources of turmeric rhizomes were purchased from diﬀerent markets and ground to obtain the pure turmeric powder. Then, the spiked turmeric powders were successfully analyzed using four diﬀerent multivariate chemometric models, such as partial least square (PLS), genetic algorithm—partial least square (GA-PLS), optimized genetic algorithm—partial least square (GA(DoE)-PLS), and articial neural network using opti- mized genetic algorithm dataset (GA(DoE)-ANN), in which each source was given a specic level coded as (2, 1, 0, 1 and 2). 2.4.8 Analysis of the calibration and validation sets. The PLS, GA-PLS, GA(DoE)-PLS and GA(DoE)-ANN models were constructed to determine the concentration of each analyte in the calibration set, the recovery percent, standard deviation, relative standard deviation (% RSD), and root mean square error of calibration (RMSEC). Aerward, the developed models determined the concentration of each adulterant in the vali- dation set mixtures. The recovery percent, standard deviation, % RSD and RMSEP were calculated. Open Access Article. Published on 23 March 2022. Downloaded on 10/24/2024 6:29:1 This article is licensed under a Creative Commons Attribution 3.0 Unpor 2.4.5 Variable selection using the genetic algorithm tool. The genetic algorithm (GA) parameters were congured (Table S1†). The GA procedure was repeated several times to select the relevant wavelengths out of the 341 wavelengths in the range (230 nm–570 nm). The selected wavelengths were used to build the GA-PLS model. The model was used to determine the concentration of MY, AO and LCM. 3.1 Spectral characteristics 2.4.7 Articial neural network. Articial neural network (ANN) is a computing system that mimics how the human brain analyzes and processes data. The optimized GA data were used to construct the articial neural network GA(DoE)-ANN. The absorbance matrix was reduced from 341 wavelengths to 123 wavelengths for the three adulterants before presenting them into the network to save the modeling time. The absorbances of the selected wavelengths (123 wavelengths) were used as inputs, while the concentration matrix of the three adulterants was used as output for the GA(DoE)-ANN model. The ANN parame- ters were adjusted using the Plackett–Burman design (Table S3†). The absorption spectra of the diﬀerent turmeric extracts absorb light with a similar pattern, but at diﬀerent extents. While the absorption spectra of the three adulterants (MY, AO, LCM) show diﬀerent absorption patterns and extents from each other and turmeric powder (Fig. 1), the application of chemometric was applied for the resolution of the three adulterants in turmeric powder from their spectral data. 3. Results and discussion Detection of food adulteration has become an increasing concern for governments and institutions to guarantee food safety and quality. Therefore, a single, fast and cheap analytical method was developed for the detection of possible turmeric adulterants. The method should be able to assay the hazardous adulterants in diﬀerent sources of turmeric powder with minimal sample preparation, only extract and measure. Thus, a simple and cheap UV-visible spectrophotometry method was a good choice to apply. The spectral data were analyzed using multivariate chemometric models. Chemometric combined with spectrophotometric techniques to recognize and assay compounds from their combined spectral data. 2.4.6 Design of experiment for the optimization of the GA parameters. A two-level (1, +1) factorial design was followed using center points to optimize the genetic algorithm parame- ters. Three numeric factors were manipulated, and two of them were GA parameters. The optimized parameters were the maximum number of latent variables (ml) required to build the GA model, the included tness percentage (t%), and the number of the latent variables used to build the PLS model (LV) (Table S2†). The design included three levels for ml (3, 8 and 13), t% (50%, 70% and 90%) and LV (4, 5 and 6). The opti- mized parameters of the GA were used to construct the PLS model, and the results were analyzed using Design-Expert® soware. RMSEP was calculated to assess the predictive ability of the model. 2.4 Procedure According to ICH recommendations,61 the validation was performed through the determination of the adulterant standards spiked to pure turmeric powder samples from diﬀerent sources. Accurate 20 mg weights of turmeric powder were spiked with standard MY (in the range of 200 and 400 mg), AO (in the range of 800 and 1000 mg) and LCM (in the range of 2000 and 2500 mg). The mixtures were then sonicated with 50 mL NaOH (0.2 M) in a 50 mL falcon tube, and ltered. The ltrate was then quantitatively transferred into a volumetric ask (100 mL), and completed to the volume using NaOH (0.2 M). 3.3 Solvent selection Many trials were carried out trying to nd the suitable solvent to freely dissolve the three adulterants. Diﬀerent solvents were 2.4.9 Reproducibility of the models. The reproducibility of the models was tested using diﬀerent concentrations of the © 2022 The Author(s). Published by the Royal Society of Chemistry RSC Adv., 2022, 12, 9087–9094 \| 9089 RSC Advances Fig. 1 Zero-order absorption spectra of (A) 200 mg mL1 turmeric extracts from diﬀerent sources, (B) 10 mg mL1 metanil yellow (—), 10 mg mL1 acid orange 7 (- - -) and 10 mg mL1 lead chromate (..). Fig. 1 Zero-order absorption spectra of (A) 200 mg mL1 turmeric extracts from diﬀerent sources, (B) 10 mg mL1 metanil yellow (—), 10 mg mL1 acid orange 7 (- - -) and 10 mg mL1 lead chromate (..). Open Access Article. Published on 23 March 2022. Downloaded on 10/24/2024 6:29:14 AM. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. Fig. 1 Zero-order absorption spectra of (A) 200 mg mL1 turmeric extracts from diﬀerent sources, (B) 10 mg mL1 metanil yellow (—), 10 mg mL1 acid orange 7 (- - -) and 10 mg mL1 lead chromate (..). Fig. 1 Zero-order absorption spectra of (A) 200 mg mL1 turmeric extracts from diﬀerent sources, (B) 10 mg mL acid orange 7 (- - -) and 10 mg mL1 lead chromate (..). 3.5 Construction of the models A multilevel multifactor design was used to prepare a calibra- tion set of 25 laboratory prepared mixtures and a validation set of 5 mixtures containing diﬀerent concentration levels of the three adulterants ranging from 2–10 mg mL1 for MY, 8–16 mg mL1 for AO and 20–40 mg mL1 for LCM. Each model was constructed using the optimum number of latent variables to avoid the unnecessary noise and loss of meaningful data required to build the models. 3.4 Wavelength selection The wavelengths used were in the range of 230–570 nm, which achieved good linearity for the three adulterants. Meanwhile, the other wavelengths were discarded due to the noise appear- ing within the range of 200–229 nm and the poor absorbance within the range of 571–800 nm. The small RMSEP values indicate good predictability and high-resolution power of the model. A good correlation coeﬃ- cient (r) for each adulterant was achieved, which indicates a good t between the predicted and the actual concentrations. Genetic algorithm tool improved the predictive power of the PLS model by selecting the most informative wavelengths and excluding the less informative ones. It reduced the number of wavelengths to about 68.9% of the original ones (106 wave- lengths for the three adulterants). We constructed the GA-PLS model using the GA-selected wavelength of the calibration set. RMSECV calculations show that six latent variables are adequate for the construction of the GA-PLS model (Fig. S1†). Unfortunately, GA did not reduce the number of latent variables compared to the previously mentioned PLS model. The RMSEP values were relatively small compared to that of the PLS model (Fig. 2), which indicates an increase in the predictive power of the GA-PLS method compared to the classical PLS method. 3.7 Predictive powers of PLS and GA-PLS models tried, such as methanol, ethanol, acetonitrile, acetone, acetic acid, diluted nitric acid and aqueous NaOH. The three adul- terants were freely soluble in aqueous NaOH (0.2 M). A PLS model was constructed using six latent variables (Fig. S1†). The model successfully determined MY, AO and LCM in spiked turmeric samples within the calibration and valida- tion sets. The mean recovery, % RSD, RMSE and other statistical parameters were calculated for each adulterant (Table 1). A PLS model was constructed using six latent variables (Fig. S1†). The model successfully determined MY, AO and LCM in spiked turmeric samples within the calibration and valida- tion sets. The mean recovery, % RSD, RMSE and other statistical parameters were calculated for each adulterant (Table 1). © 2022 The Author(s). Published by the Royal Society of Chemistry 3.8 Optimization of GA-PLS model The GA reduced the RMSEP, and the standard deviation of the results was obtained using the PLS model. The quality-by- design principles was applied to maximize the accuracy and precision using an experimental design to optimize the parameters of the genetic algorithm-PLS model (ml, t% and LV). The latter reduced the error and increased the precision of the PLS model for the prediction of MY, AO, and LCM in the ve commercially available sources of turmeric. Finally, the reso- lution of the classical PLS chemometric model was compared to an articially intelligent neural network. The latter exhibited better predictability for the three adulterants in the ve commercial sources of turmeric powder. Even with the preference of the GA-PLS model over the classical one, there was a necessity to optimize its predictive ability using the design of experiment technique (DoE). The DoE evaluates the impact of diﬀerent factors on the desired response, and identies important interactions that are missed when exper- imenting with one factor at a time. The results were analyzed using one-way ANOVA (Table S4†). The small p-values (less than 0.05) and large F-value prove the tness of the model, and its ability to determine the concentration of the three adulterants. These results indicate that there is only a 0.01% chance that an F-value this large could occur due to noise. However, the lack of © 2022 The Author(s). 3.8 Optimization of GA-PLS model Published by the Royal Society of Chemistry 9090 \| RSC Adv., 2022, 12, 9087–9094 Paper Table 1 Assay validation sheet of the chemometric models for the calibration and validation sets Table 1 Assay validation sheet of the chemometric models for the calibration and validation sets Parameters Calibration set Validation set Metanil yellow Acid orange 7 Lead chromate Metanil yellow Acid orange 7 Lead chromate PLS modeld Concentration range (mg mL1) 2–10 8–16 20–40 2–10 8–16 20–40 Mean 100.09 100.51 100.05 99.00 100.03 99.36 % RSD 1.892 1.984 1.794 2.120 2.097 1.756 Repeatability precision (RSD) 1.903 1.924 1.536 — — — Intermediate precision (RSD) 1.917 2.084 1.969 — — — RMSE 0.101a 0.224a 0.592a 0.057b 0.166b 0.377b Slopec 0.9996 0.9943 0.9920 1.0025 0.8327 0.9884 Intercept 0.0019 0.1163 0.2327 0.0378 1.5232 0.1192 r 0.9993 0.9970 0.9965 0.9987 0.9998 0.9900 Latent variables 6 GA-PLS modele Concentration range (mg mL1) 2–10 8–16 20–40 2–10 8–16 20–40 Mean 100.14 100.54 99.97 98.95 100.21 98.82 % RSD 1.814 1.629 1.604 1.894 1.714 0.813 Repeatability precision RSD 1.799 2.066 1.122 — — — Intermediate precision RSD 1.887 2.082 1.642 — — — RMSE 0.090a 0.171a 0.547a 0.049b 0.135b 0.309b Slopec 0.9990 0.9865 0.9976 1.0123 0.8790 1.0020 Intercept 0.0061 0.2099 0.0531 0.0663 1.1197 0.3112 r 0.9995 0.9984 0.9969 0.9992 0.9972 0.9979 Latent variables 6 GA(DoE)-PLS modelf Concentration range (mg mL1) 2–10 8–16 20–40 2–10 8–16 20–40 Mean 100.22 100.24 100.05 99.40 99.52 99.58 % RSD 1.542 1.143 1.388 0.918 0.919 0.576 Repeatability precision (% RSD) 1.540 1.785 1.398 — — — Intermediate precision (% RSD) 1.687 1.859 1.589 — — — RMSE 0.071a 0.127a 0.428a 0.040b 0.093b 0.138b Slopec 1.0000 0.9968 0.9878 0.9856 0.9596 1.0166 Intercept 0.0034 0.0624 0.3587 0.0256 0.3245 0.4702 r 0.9997 0.9989 0.9982 0.9994 0.9972 0.9993 Latent variables 5 GA(DoE)-ANN modelg Concentration range (mg mL1) 2–10 8–16 20–40 2–10 8–16 20–40 Mean 100.02 100.05 100.17 100.15 99.86 100.02 % RSD 0.415 0.631 0.613 0.782 0.313 0.054 Repeatability precision RSD 0.564 0.932 0.584 — — — Intermediate precision RSD 0.722 1.167 1.052 — — — RMSE 0.030a 0.062a 0.191a 0.030b 0.028b 0.016b Slopec 1.0015 0.9974 1.0039 1.0113 0.9879 1.0019 Intercept 0.0011 0.0400 0.0615 0.0226 0.1016 0.0343 r 0.9999 0.9997 0.9997 0.9997 0.9997 1.0000 Latent variables 5 a Root mean square error of calibration. b Root mean square error of prediction. c Data of the straight line plotted between predicted versus actual concentrations. 3.8 Optimization of GA-PLS model d Partial least square. e Genetic algorithm-partial least square. f Optimized genetic algorithm-partial least square. g Articial neural network using optimized genetic-algorithm. Open Access Article. Published on 23 March 2022. Downloaded on 10/24/2024 6:29:14 AM. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. ot mean square error of prediction. c Data of the straight line plotted between predicted versus actual ic algorithm-partial least square. f Optimized genetic algorithm-partial least square. g Articial neural a Root mean square error of calibration. b Root mean square error of prediction. c Data of the straight line plotted between predicted versus actual concentrations. d Partial least square. e Genetic algorithm-partial least square. f Optimized genetic algorithm-partial least square. g Articial neural network using optimized genetic-algorithm. factors was observed. The desirability function selected the best level of each factor, which was then used to optimize the predictive ability of the GA-PLS model. The function suggested using ml, t% and LV of 10, 82% and 5, respectively (Fig. S3†), with 97.60% desirability. t (p-values more than 0.05) implies that the curvature of the model is not signicant relative to the pure error, which could occur due to noise. Non-signicant lack of t is desirable as we want the model to t, which means that the curvature is not signicant. The relatively small diﬀerence between the adjusted and predicted R2 indicates good predictability of the model. Adequate precision measures the signal-to-noise ratio. A ratio greater than 4 is achieved, which indicates an adequate signal and proves that the model can be used to navigate the design space. Residuals versus predicted data plots (Fig. S2†) indicated the good tness of the model as the residuals (errors) are within the specied limits. No signicant interaction between the The suggested ml was applied to GA parameters, while the suggested t% was used to select the optimum set of wave- lengths. The PLS model was constructed using the calibration set and the optimum number of LV (Fig. S1†). The smaller RMSEP of the GA(DoE)-PLS model relative to that of the GA-PLS model suggests better predictive ability of the former model (Fig. 2). © 2022 The Author(s). Published by the Royal Society of Chemistry RSC Adv., 2022, 12, 9087–9094 \| 9091 Fig. 2 Root mean square error of calibration and validation sets for the three adulterants using the four chemometric models. RSC Advances nloaded on 10/24/2024 6:29:14 AM. 3.8 Optimization of GA-PLS model mmons Attribution 3.0 Unported Licence. View A Fig. 2 Root mean square error of calibration and validation sets for the three adulterants using the four chemometric models. RSC Advances Pa nloaded on 10/24/2024 6:29:14 AM. mmons Attribution 3.0 Unported Licence. View Article O RSC Advances Fig. 2 Root mean square error of calibration and validation sets for the three adulterants using the four chemometric models. Fig. 3 Plot of the mean square error (MSE) and the correlation coeﬃcient (r) against the number of neurons in the architecture of GA(DoE)-AN p Open Access Article. Published on 23 March 2022. Downloaded on 10/24/2024 6:29:14 AM. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. Fig. 2 Root mean square error of calibration and validation sets for the three adulterants using the four chemometric models. Fig. 3 Plot of the mean square error (MSE) and the correlation coeﬃcient (r) against the number of neurons in the architecture of GA(DoE)-ANN. Open Access Article. Published on 23 March 2022. Dow This article is licensed under a Creative Co Fig. 3 Plot of the mean square error (MSE) and the correlation coeﬃcient (r) against the number of neurons in t ror (MSE) and the correlation coeﬃcient (r) against the number of neurons in the architecture of GA(DoE)-ANN. (Fig. 3). The ANN architecture indicated the diﬀerent layers used for the prediction of the concentration of the three adul- terants (Fig. S4†). Purelin–Purelin transfer function was suitable owing to the linear absorbances–concentrations relationship. The TRAINLM-Levenberg–Marquardt backpropagation was preferred as a training function. The GA(DoE)-ANN model was performed on the validation set, and RMSEP was calculated. 3.9 Optimization of ANN model The ANN model was constructed using the GA (DoE) dataset. Diﬀerent numbers of hidden neurons were tested to select the optimum number of neurons that improved the ANN predictive ability. The optimum number of hidden neurons was seven, which gave a small MSE and correlation coeﬃcient r near unity Table 2 Determination of metanil yellow, acid orange 7 and lead chromate in spiked turmeric samples Models Metanil yellow (recovery% RSD%)a Acid orange 7 (recovery% RSD%)a Lead chromate (recovery% RSD%)a Partial least square 98.67 1.876 100.07 1.840 100.28 1.650 Genetic algorithm-partial least square 99.00 1.746 100.33 1.685 99.65 0.923 Optimized genetic algorithm-partial least square 99.42 1.338 99.28 0.976 100.05 0.477 Articial neural network using optimized genetic algorithm 99.75 0.938 99.93 0.577 99.97 0.200 a Average of three determinations. 9092 \| RSC Adv., 2022, 12, 9087–9094 © 2022 The Author(s). Published by the Royal Society of Chemistry Table 2 Determination of metanil yellow, acid orange 7 and lead chromate in spiked turmeric samples Models Metanil yellow (recovery% RSD%)a Acid orange 7 (recovery% RSD%)a Lead chromate (recovery% RSD%)a Partial least square 98.67 1.876 100.07 1.840 100.28 1.650 Genetic algorithm-partial least square 99.00 1.746 100.33 1.685 99.65 0.923 Optimized genetic algorithm-partial least square 99.42 1.338 99.28 0.976 100.05 0.477 Articial neural network using optimized genetic algorithm 99.75 0.938 99.93 0.577 99.97 0.200 Table 2 Determination of metanil yellow, acid orange 7 and lead chromate in spiked turmeric samples © 2022 The Author(s). Published by the Royal Society of Chemistry RSC Advances View Article Online RSC Advances View Article Online View Article Online View Article Online Paper RSC Advances Access Article. Published on 23 March 2022. Downloaded on 10/24/2024 6:29:14 AM. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. 4 Y. Panahi, M. S. Hosseini, N. Khalili, E. Naimi, L. E. Simental- Mend´ıa, M. Majeed and A. Sahebkar, Biomed. Pharmacother., 2016, 82, 578–582. 5 S. Togni and F. Mazzolani, Clin. Ophthalmol., 2013, 7, 939. 3.11 Validation of the models 6 W. Li, L. Sun, J. Lei, Z. Wu, Q. Ma and Z. Wang, Oncol. Rep., 2020, 44, 382–392. Few methods were reported for the assay of MY in turmeric powder. However, no method was reported for the simulta- neous assay of the three adulterants in turmeric powder. Accordingly, we validated the developed chemometric models as per the ICH guidelines.61 This was achieved through spiking of the pure turmeric powder with known concentrations of the three adulterants, and the ability of the proposed models to recover the concentration of the spiked adulterants was then assessed. The recovery percentages and the relative standard deviation percentages were calculated (Table 2). 7 B. Chandran and A. Goel, Phytother. Res., 2012, 26, 1719– 1725. 8 L. L. Hurley, L. Akinresoye, E. Nwulia, A. Kamiya, A. A. Kulkarni and Y. Tizabi, Behav. Brain Res., 2013, 239, 27–30. 9 B. Sasikumar, J. Spices Aromat. Crop., 1970, 28, 96–105. 10 R. A. Cartwright, Environ. Health Perspect., 1983, 49, 13–19. 11 R. N. Dome, S. Hazra, D. Ghosh and S. Ghosh, Int. J. Pharm. Pharm. Sci., 2017, 9, 203. 12 T. N. Nagaraja and T. Desiraju, Food Chem. Toxicol., 1993, 31, 41–44. Author contributions 21 N. Kusnin, N. A. Yusof, J. Abdullah, S. Sabri, F. Mohammad, S. Mustafa, N. A. Ab Mutalib, S. Sato, S. Takenaka, N. A. Parmin and H. A. Al-Lohedan, RSC Adv., 2020, 10, 27336–27345. 21 N. Kusnin, N. A. Yusof, J. Abdullah, S. Sabri, F. Mohammad, All authors contributed to the conceptualization of the work, analysis, investigation and visualization of the data. 22 S. Karimi, J. Feizy, F. Mehrjo and M. Farrokhnia, RSC Adv., 2016, 6, 23085–23093. 4. Conclusion 13 P. P. Nath, K. Sarkar, M. Mondal and G. Paul, Environ. Toxicol., 2010, 29, 453–460. Four multivariate chemometric models (PLS, GA-PLS, GA (DoE)- PLS and GA(DoE)-ANN) successfully quantied three toxic turmeric adulterants (MY, AO and LCM) in turmeric powder samples to determine the type and amount of the adulterant. The work demonstrated the augmented inuence of the GA and the experimental design on the predictive ability of the PLS and ANN models. The four chemometric methods were validated as per the ICH guidelines. The developed methods proved to be fast, cheap and involved minimal sample preparation. The methods can be used for fast surveillance of turmeric powder adulteration in quality control laboratories before including the powder in food recipes to guarantee authenticity, safety and quality. 14 P. P. Nath, K. Sarkar, M. Mondal and G. Paul, Environ. Toxicol., 2016, 31, 2057–2067. 15 R. Sarkar and A. R. Ghosh, Int. J. Basic Appl. Med. Sci., 2012, 2, 40–42. 16 V. K. Gupta, A. Mittal, V. Gajbe and J. Mittal, Ind. Eng. Chem. Res., 2006, 45, 1446–1453. 17 W. Cowell, T. Ireland, D. Vorhees and W. Heiger-Bernays, Public Health Rep., 2017, 132, 289–293. 18 J. Singh, D. E. Pritchard, D. L. Carlisle, J. A. Mclean, A. Montaser, J. M. Orenstein and S. R. Patierno, Toxicol. Appl. Pharmacol., 1999, 161, 240–248. 19 D. I. Ellis, H. Muhamadali, S. A. Haughey, C. T. Elliott and R. Goodacre, Anal. Methods, 2015, 7, 9401–9414. 20 I. Khalil, W. A. Yehye, N. Muhd Julkapli, A. A. I. Sina, S. Rahmati, W. J. Basirun and A. Seyfoddin, Analyst, 2020, 145, 1414–1426. Funding 23 Y. Huang, X. Mu, J. Wang, Y. Wang, J. Xie, R. Ying and E. Su, J. Mater. Chem. B, 2022, 10, 1359. This research did not receive any specic grant from funding agencies in the public, commercial or not-for-prot sectors. 24 P. M. Murray, F. Bellany, L. Benhamou, D.-K. Buˇcar, A. B. Tabor and T. D. Sheppard, Org. Biomol. Chem., 2016, 14, 2373–2384. There are no conicts to declare. There are no conicts to declare. RSC Adv., 2022, 12, 9087–9094 \| 9093 © 2022 The Author(s). Published by the Royal Society of Chemistry © 2022 The Author(s). Published by the Royal Society of Chemistry References The four models successfully predicted the concentration of the three analytes in the calibration and validation sets (Table 1). Reproducible and precise results (% RSD) were obtained for repeatability and intermediate precision, indicating the success of the four models in analyzing the three adulterants (Table 1). The GA(DoE)-PLS and GA(DoE)-ANN models achieved the best accuracy (low RMSEP) and precision (low % RSD) compared to the other models (Fig. S5 and S6†), owing to the useful GA variable selection tool, the eﬀective DoE optimization approach, and the powerful ANN articial intelligence (Fig. 2). 1 S. M. A. El-Halim, M. A. Mamdouh, A. E. El-Haddad and S. M. Soliman, Sci. Pharm., 2020, 88, 1–16. 2 F. Forouzanfar, M. Majeed, T. Jamialahmadi and A. Sahebkar, in Advances in Experimental Medicine and Biology, Springer, 2021, vol. 1291, pp. 363–373. 3 B. B. Aggarwal and K. B. Harikumar, Int. J. Biochem. Cell Biol., 2009, 41, 40–59. Open Access Article. Published on 23 March 2022. Downloaded on 10/24/2024 6:2 This article is licensed under a Creative Commons Attribution 3.0 Unp 31 A. Łacka, Agronomy, 2021, 11, 2406. 51 L. Sabatino, M. Scordino, M. Gargano, A. Belligno, P. Traulo and G. Gagliano, Nat. Prod. Commun., 2011, 6, 1873–1876. 32 M. Picard, A. K. Mohanty and M. Misra, RSC Adv., 2020, 10, 36058–36089. 52 A. Girme, G. Saste, A. K. Balasubramaniam, S. Pawar, C. Ghule and L. Hingorani, J. Pharm. Biomed. Anal., 2020, 191, 113603. 33 R. W. Baker, L. Forfar, X. Liang and P. J. Cameron, React. Chem. Eng., 2021, 6, 709–719. 34 N. S. Eyke, W. H. Green and K. F. Jensen, React. Chem. Eng., 2020, 5, 1963–1972. 53 A. Rohman, T. Wijayanti, A. Windarsih and S. Riyanto, Molecules, 2020, 25, 1–13. 35 M. A. Abou-Shamat, J. L. Stair, S. B. Kirton, J. Calvo-Castro and M. T. Cook, Mol. Syst. Des. Eng., 2020, 5, 1538–1546. 54 R. M. Shereema, T. P. Rao, V. B. Sameer Kumar, T. V. Sruthi, R. Vishnu, G. R. D. Prabhu and S. Sharath Shankar, Mater. Sci. Eng., C, 2018, 93, 21–27. 36 Y. Ayaz Atalan and A. Atalan, ˙Istatistik ve Uygulamalı Bilim. Derg., 2021, 2, 1–7. 55 W. G. Chaminda Bandara, G. W. Kasun Prabhath, D. W. Sahan Chinthana Bandara Dissanayake, V. R. Herath, G. M. Roshan Indika Godaliyadda, M. P. Bandara Ekanayake, D. Demini and T. Madhujith, J. Food Eng., 2020, 266, 109700. 37 D. Galvan, L. Eing, H. Cremasco and C. A. Conte-Junior, Foods, 2021, 10, 1941. 38 M. Fidaleo, Food Bioprocess Technol., 2020, 13, 1035–1047. 39 F. Liu, Y. Shi and L. Najjar, Procedia Comput. Sci., 2017, 122, 720–726. 56 S. Varliklioz Er, H. Eksi-Kocak, H. Yetim and I. H. Boyaci, Food Anal. Methods, 2017, 10, 1547–1555. 40 D. Pugliese, F. Bella, V. Cauda, A. Lamberti, A. Sacco, E. Tresso and S. Bianco, ACS Appl. Mater. Interfaces, 2013, 5, 11288–11295. 57 S. Kar, B. Tudu, A. K. Bag and R. Bandyopadhyay, Food Anal. Methods, 2018, 11, 1291–1302. 41 S. Galliano, F. Bella, M. Bonomo, F. Giordano, M. Gr¨atzel, G. Viscardi, A. Hagfeldt, C. Gerbaldi and C. Barolo, Sol. RRL, 2021, 5, 2000823. 58 S. Dhakal, K. Chao, J. Qin, M. Kim, W. Schmidt and D. Chan, Sens. Agric. Food Qual. Saf. VIII, 2016, vol. 9864, p. 98640A. 59 I. R. Rukundo, M. G. C. Danao, C. L. Weller, R. L. Wehling and K. M. Eskridge, J. Near Infrared Spectrosc., 2020, 28, 81–92. 42 S. Galliano, F. Bella, M. Bonomo, F. Giordano, M. © 2022 The Author(s). Published by the Royal Society of Chemistry View Article Online RSC Advances © 2022 The Author(s). Published by the Royal Society of Chemistry Paper 45 I. Reichert, P. Olney and T. Lahmer, J. Civ. Struct. Health Monit., 2021, 11, 223–234. 25 S. Choudhary, K. Sharma, M. S. Bhatti, V. Sharma and V. Kumar, RSC Adv., 2022, 12, 4780–4794. 26 C. Waldron, A. Pankajakshan, M. Quaglio, E. Cao, F. Galvanin and A. Gavriilidis, React. Chem. Eng., 2020, 5, 112–123. 46 P. Garc´ıa-Manrique, M. B. Gonz´alez-Garc´ıa and M. C. Blanco-L´opez, in Laboratory Methods in Dynamic Electroanalysis, Elsevier, 2020, pp. 361–371. 27 A. Elik, A. Demirbas¸ and N. Altunay, Food Chem., 2022, 371, 131138. 47 Y. Xu, X. Gao, C. Yang, B. Man and J. Leng, J. Mater. Chem. C, 2021, 9, 2763–2774. Open Access Article. Published on 23 March 2022. Downloaded on 10/24/2024 6:29:14 AM. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. 48 D. Suriano, G. Cassano and M. Penza, J. Sens., 2020, 2020, 1– 48 D. Suriano, G. Cassano and M. Penza, J. Sens., 2020, 2020, 1– 20. 28 N. Taouk, W. Boumya, R. Elmoubarki, A. Elhalil, M. Achak, M. Abdennouri and N. Barka, Mater. Today Chem., 2022, 23, 100732. 49 M. E. Wahba, D. El Wasseef, A. S. Saad and M. E. Draz, RSC Adv., 2021, 11, 12833–12844. 29 F. H. Edrees, A. S. Saad, M. T. Alsaadi, N. H. Amin and N. S. Abdelwahab, RSC Adv., 2021, 11, 1450–1460. 50 P. K. Sahu, J. Panda, Y. V. V. Jogendra Kumar and S. K. Ranjitha, J. Liq. Chromatogr. Relat. Technol., 2020, 43, 247–254. 30 S. B. Kim, D. S. Kim and C. Magana-Ramirez, PLoS One, 2021, 16, e0257472. Open Access Article. Published on 23 March 2022. Downloaded on 10/24/2024 6:2 This article is licensed under a Creative Commons Attribution 3.0 Unp Gr¨atzel, G. Viscardi, A. Hagfeldt, C. Gerbaldi and C. Barolo, Sol. RRL, 2021, 5, 2000823. 60 E. A. Petrakis and M. G. Polissiou, Talanta, 2017, 162, 558– 566. 43 F. Bella, M. Imperiyka and A. Ahmad, J. Photochem. Photobiol., A, 2014, 289, 73–80. 61 N. Baber, Br. J. Clin. Pharmacol., 1994, 37, 401–404. 44 S. Galliano, F. Bella, G. Piana, G. Giacona, G. Viscardi, C. Gerbaldi, M. Gr¨atzel and C. Barolo, Sol. Energy, 2018, 163, 251–255. 9094 \| RSC Adv., 2022, 12, 9087–9094 © 2022 The Author(s). Published by the Royal Society of Chemistry
https://openalex.org/W4367601216	https://www.zora.uzh.ch/id/eprint/214203/8/CIIT_of_Personality_and_Psychopathology-2021-12-15T15_58_18.841Z.pdf	English	null	Contemporary integrative interpersonal theory: Integrating structure, dynamics, temporal scale, and levels of analysis.	Journal of psychopathology and clinical science	2,023	cc-by	20,199	Running head: INTERPERSONAL THEORY Running head: INTERPERSONAL THEORY Running head: INTERPERSONAL THEORY 1 1 Contemporary Integrative Interpersonal Theory: Integrating Structure, Dynamics, Temporal Scale, and Levels of Analysis Aidan G.C. Wright, Department of Psychology, University of Pittsburgh Aaron L. Pincus, Department of Psychology, The Pennsylvania State University Christopher J. Hopwood, Department of Psychology, University of California, Davis Note: This manuscript has been accepted for publication in the Journal of Abnormal Psychology. Supplemental material follows main text in this document. Correspondence concerning this article should be addressed to Aidan G.C. Wright, Department of Psychology, University of Pittsburgh, 4119 Sennott Square, 210 S. Bouquet St., Pittsburgh, PA, 15260. E-mail: aidan@pitt.edu INTERPERSONAL THEORY 2 2 Abstract Theoretical accounts of psychopathology often emphasize social context as etiologically central to psychological dysfunction, and interpersonal impairments are widely implicated for many legacy diagnostic categories that span domains of psychopathology (e.g., affective, personality, thought disorders). Contemporary Integrative Interpersonal Theory (CIIT) seeks to explain the emergence, expression, and maintenance of socio-affective functioning and dysfunction across levels and timescales of analysis. We emphasize the importance of cohesively addressing the often-segregated challenges of establishing empirically supported structure, functional accounts of dynamic processes, and how together these facilitate theoretical and methodological Theoretical accounts of psychopathology often emphasize social context as etiologically central to psychological dysfunction, and interpersonal impairments are widely implicated for many legacy diagnostic categories that span domains of psychopathology (e.g., affective, personality, thought disorders). Contemporary Integrative Interpersonal Theory (CIIT) seeks to explain the emergence, expression, and maintenance of socio-affective functioning and dysfunction across levels and timescales of analysis. We emphasize the importance of cohesively addressing the often-segregated challenges of establishing empirically supported structure, functional accounts of dynamic processes, and how together these facilitate theoretical and methodological consistency across levels of analysis ranging from biology to behavior. We illustrate CIIT’s potential to serve as an integrative theory for generating falsifiable hypotheses that support strong inference investigations into the nature of psychological dysfunction across a range of traditional diagnostic constructs and superordinate spectra of psychopathology. consistency across levels of analysis ranging from biology to behavior. We illustrate CIIT’s potential to serve as an integrative theory for generating falsifiable hypotheses that support strong inference investigations into the nature of psychological dysfunction across a range of traditional diagnostic constructs and superordinate spectra of psychopathology. Keywords: Interpersonal Theory; Social Processes; Transdiagnostic; Personality; Psychopathology Psychopathology General Scientific Summary: The field of psychopathology is transitioning away from discrete categories of mental disorders, and shifting towards understanding psychopathology using dimensions that cut across traditional dimensions. However, several challenges are associated with these changes, including linking a scientifically supported structure with basic dynamic processes, and how levels of analysis can be integrated into coherent systems of functioning. This paper describes how Contemporary Integrative Interpersonal Theory has anticipated and offers potential solutions to these challenges. 1 The National Institute of Mental Health’s Research Domain Criteria uses the term “unit” in much the same way, to mean different levels of functioning or modalities of analysis. INTERPERSONAL THEORY 3 1 The National Institute of Mental Health’s Research Domain Criteria uses the term “unit” in mu mean different levels of functioning or modalities of analysis 1 The National Institute of Mental Health’s Research Domain Criteria uses the term “unit” in mu mean different levels of functioning or modalities of analysis. INTERPERSONAL THEORY 4 and effects of functioning as well as dysfunction with broad patterns of interpersonal motives, perceptions, cognitions, behavior, and affect, it is a theory of personality and psychopathology. Contemporary Integrative Interpersonal Theory and effects of functioning as well as dysfunction with broad patterns of interpersonal motives, perceptions, cognitions, behavior, and affect, it is a theory of personality and psychopathology. Contemporary Integrative Interpersonal Theory and effects of functioning as well as dysfunction with broad patterns of interpersonal motives, perceptions, cognitions, behavior, and affect, it is a theory of personality and psychopathology. Integrating Structure, Dynamics, Temporal Scale, and Levels of Analysis The interpersonal dance begins at least as early as birth, and ends only with death. Virtually all of the most important events in life are interpersonal in nature and most of what we call personality is interpersonal in expression. – Allen Francis, M.D. (1996) An understanding of psychological functioning is necessary to understand its dysfunction (i.e., psychopathology), and this understanding must cut across levels1 of analysis and span temporal scales. Comprehensive theoretical models of psychopathology need to address each of these components, and major ongoing efforts to develop revised psychiatric taxonomies – notably the Hierarchical Taxonomy of Psychopathology (HiTOP; Kotov et al., 2017) and the Research Domain Criteria (RDoC; Cuthbert, 2014) – have set their sights on some of these challenges. These efforts, though valuable, lack an organizing theory to guide their focus, coherence, and integration with a broader understanding of human behavior (Dalgleish et al., 2020). Here we outline, in broad strokes, how contemporary integrative interpersonal theory (CIIT; Dawood et al., 2018; Kiesler, 1996; Pincus, 2005; Pincus & Ansell, 2013) provides an established theoretical framework that can address the key challenges of modern psychopathology science. CIIT seeks to integrate an empirically based structural model with both short-term (momentary) and long-term (developmental) dynamics, across levels of analysis, and thus inform cause and effect models of psychopathology. Consistent with Frances’ (1996) assertion above (see also Pincus & Wright, 2011), because CIIT seeks to account for the causes Contemporary Integrative Interpersonal Theory CIIT (Pincus, 2005) emerged from the relational movement in response to psychoanalytic drive theory in the first half of the 1900’s (Sullivan, 1953a, 1953b, 1954), but has kept pace with major advances in psychological science over the past 70 years (Dawood et al., 2018; Laforge, 2004; Wiggins & Trapnell, 1996). From Sullivan’s work to present day CIIT, the priority focus is on interpersonal functioning, which is organized around the “interpersonal situation” as a fundamental element of personality and psychopathology. Personality and psychopathology are explicitly linked in this perspective, because who we are, and the problems we have, bi- directionally influence each other (Pincus & Wright, 2011; Sullivan, 1948). Yet, this emphasis on the interpersonal should not be understood as narrow or disconnected from other important areas of psychological experience (Benjamin, 2002; Horowitz et al., 2006; Kiesler et al., 1997; Lukowitsky & Pincus, 2011). Indeed, virtually all domains of personality and psychopathology implicate interpersonal functioning and vice versa. CIIT elevates the interpersonal not only because humans are a social species but because functioning cannot be defined in a vacuum; it must serve some purpose, it must have an object (Leary, 1957). We first outline the major tenets of CIIT before discussing how they address the major challenges facing modern theories of psychopathology. These assertions elaborate previous arguments (Kiesler, 1996; Leary, 1957; Pincus & Ansell, 2013; Sullivan, 1953a), and are intended to be broad. Prior descriptions of CIIT have addressed pieces of the full model (e.g., structure, development, dynamics, levels of analysis), though this is first to articulate how all of these components fit together in a cohesive model that integrates structure, processes, temporal INTERPERSONAL THEORY 5 scales, and levels of analysis. Though we limit the summary of supporting research here, interested readers are directed to an expanded treatment in an on-line supplement. scales, and levels of analysis. Though we limit the summary of supporting research here, interested readers are directed to an expanded treatment in an on-line supplement. 1. Fundamentally important functional expressions of personality and psychopathology typically occur in interpersonal situations. This articulates the purpose of personality (as Allport [1937] argued, personality is and personality does) and is supported by the near ubiquitous inclusion of interpersonal elements in major structural and process- based models of psychological and behavioral functioning (e.g., systems theory, attachment theory, dimensional models of normal and pathological traits and temperament). In CIIT, the interpersonal situation is considered a fundamental unit of personality and psychopathology, and it encompasses direct in-person interactions as well as mental representations of interactions (recollected and imagined). 2. Interpersonal functioning can be organized using the dimensions of agency and communion. The empirically derived structural model used to organize interpersonal functioning is the interpersonal circumplex (top left of Figure 1). At the broadest and most abstract level, the vertical axis can be construed as agency and the horizontal axis as communion. Agency is defined by individuation and self-differentiation, encompassing needs and strivings for achievement, mastery, dominance, and control. Communion is defined by a focus on and connection with others, encompassing needs and strivings for belonging, intimacy, and nurturance. These constructs link CIIT to vast literatures that span the behavioral and social sciences (Wiggins, 1991). Labels for these dimensions have varied by author, depending in part on their focus or level of construal (e.g., communal motives vs. warm behavior). This structure not only serves the purpose of identifying and operationalizing the primary domains of INTERPERSONAL THEORY 6 focus, but also how they relate to each other in mathematically precise ways that form a basis for falsifiable inferences (Gurtman & Pincus, 2003; Zimmermann & Wright, 2017). Moreover, these same domains are used to organize functioning and dysfunction; there are not separate models for normal functioning and psychopathology (Leary, 1957). focus, but also how they relate to each other in mathematically precise ways that form a basis for falsifiable inferences (Gurtman & Pincus, 2003; Zimmermann & Wright, 2017). Moreover, these same domains are used to organize functioning and dysfunction; there are not separate models for normal functioning and psychopathology (Leary, 1957). 3. The same structure organizes functioning across levels of experience. From the outset, interpersonal theory recognized the need for a “multi-level” understanding of psychological functioning. Humans are complex systems that span modalities of experience and expression. These include, but are not limited to, overt interpersonal behaviors, perceptual foci, construal processes, motivations, goals, and neurobehavioral structures. As alluded to in Tenet 2, the structure of the interpersonal circle can be used to organize functioning across levels of inquiry. For instance, research has identified neurobehavioral agentic and communal reward systems as well as validated methods for assessing discrete overt dominant and affiliative behavior. 4. Satisfying agentic and communal motives drives interpersonal behavior. Sullivan’s initial socio-affective formulation of personality’s function involved seeking out self- esteem and security in social situations (1953a). As such, it can be understood as a control theory model (viz. Carver & Scheier, 1998), in which interpersonal behavior serves to satisfy agentic and communal motives, sometimes referred to colloquially as succeeding at work and love, or getting ahead and getting along. The theory posits that satisfaction of motives leads to decreased negative affect and vulnerability, increased positive affect, felt security, and self-esteem, whereas frustration of motives INTERPERSONAL THEORY 7 7 leads to the opposite. These contingencies serve to reinforce/maintain or punish/decrease specific behaviors in the short and long-term. This tenet also provides an explicit link between the interpersonal and affective domains, highlighting that they are inexorably intertwined. Individuals differ from each other in their average levels of these motives based on innate biology that has interacted with life experience. Additionally, individuals differ dynamically from themselves moment to moment and across situations as they encounter differences in affordances and opportunities for motive satisfaction. leads to the opposite. These contingencies serve to reinforce/maintain or punish/decrease specific behaviors in the short and long-term. This tenet also provides an explicit link between the interpersonal and affective domains, provides an explicit link between the interpersonal and affective domains, highlighting that they are inexorably intertwined. Individuals differ from each other in their average levels of these motives based on innate biology that has interacted with life experience. Additionally, individuals differ dynamically from themselves moment to moment and across situations as they encounter differences in affordances and opportunities for motive satisfaction. 5. The “interpersonal transaction cycle” provides probabilistic expectations for patterns of behavior and falsifiable predictions about behavioral sequences. The dyadic processes that drive an interpersonal situation can be elaborated to include key components (e.g., motivation, perception, behavior, affect) of a transaction cycle that involves the interaction among modalities or levels of experience referenced in Tenet 3 (see Top right of Figure 1). CIIT makes specific predictions about the patterns of an interaction between individuals, such as complementarity (Sadler et al., 2011). 6. Sustained deviations from these expected patterns may reflect psychopathology. Typical patterns of functioning, either normatively or for the individual, provide a preliminary basis for comparison to identify maladaptive functioning. This respects individual differences and within-person fluctuations, but nevertheless can be used to describe and explain dysfunctional processes both nomothetically and idiographically. Taken together, these principles can be used to generically define adaptive psychological functioning and dysfunction (i.e., psychopathology) from the perspective of CIIT. Adaptive INTERPERSONAL THEORY 9 9 2013; Pincus et al., 2020), or interpersonal dysfunction could result from psychopathology for which the primary impairments are commonly considered non-interpersonal, as is the case for many internalizing and thought disorders (e.g., the majority of delusions have social themes [Bell et al., 2020]; both depression [Hames et al., 2013] and eating disorders [Monteleone et al., 2018] broadly and uniquely impact social perception and behavior). Importantly, the content and processes outlined in CIIT may provide valuable insights into an individual’s unique manifestation of any form of psychopathology. In each case, CIIT can serve as a useful lens for understanding the processes operant in the key domain of social functioning, and be leveraged to intervene and effect some desirable change for the individual (Anchin & Pincus, 2010; Benjamin, 2018). We next elaborate how these basic principles give rise to an integrated theory of structure and process, spanning timescales and levels of analysis, and inform notions of cause and effect in psychopathology. Like most theories, some aspects of CIIT have achieved strong support, whereas other aspects remain speculative and in some cases are under active inquiry (see supplement). INTERPERSONAL THEORY 8 8 functioning is defined as a sustained ability to engage interpersonally in ways that coordinate and satisfy the agentic and communal needs of self and other, relatively consistent with one’s developmental stage and socio-cultural context, through the flexible, stable, and effective regulation of self, affect, and interpersonal behavior. As stated in Tenet 4, CIIT recognizes and respects the rich heterogeneity in individual differences in dispositional levels of agentic and communal motivation, which are the result of accumulated life experience interacting with one’s unique biological makeup. This means that adaptive functioning is not defined by dispositional levels per se: rather, it is defined by the ability to stably yet flexibly coordinate and satisfy self and others’ motives within the contexts of developmental, socio-cultural, and situational demands. Accordingly, dysfunction would be defined as the sustained breakdown in any of the processes that support and maintain the flexible, stable, and effective regulation of self, affect, and/or interpersonal behavior. The emphasis is on sustained difficulties so as to accommodate the expected transient deviations and regulation of processes in reaction to life events and stressors. Examples of contributing processes might be impairments, biases, or instability of perceptions of self and other; overvalued or extreme motivations or beliefs; interpersonal repertoire/skills deficits; idiosyncratic, deviant, inconsistent, or rigidly applied behavioral rules, such as overvalued or disavowed goals, amplified or exaggerated links among perceptions, Accordingly, dysfunction would be defined as the sustained breakdown in any of the processes that support and maintain the flexible, stable, and effective regulation of self, affect, and/or interpersonal behavior. The emphasis is on sustained difficulties so as to accommodate the expected transient deviations and regulation of processes in reaction to life events and stressors. Examples of contributing processes might be impairments, biases, or instability of perceptions of self and other; overvalued or extreme motivations or beliefs; interpersonal repertoire/skills deficits; idiosyncratic, deviant, inconsistent, or rigidly applied behavioral rules, such as overvalued or disavowed goals, amplified or exaggerated links among perceptions, affect, and behavior (e.g., rejection sensitivity), among others. The key is that any one of these could disrupt the successful and mutual navigation of interpersonal transaction cycles, thereby leading to self, affect, and behavioral dysregulation, and ultimately unstable, dissatisfying, and even destructive interpersonal relations. As such, an individual’s psychopathology could reflect primary interpersonal dysfunction, as is often the case for personality disorders (Hopwood et al., Integration: From Structure to Function Psychopathology is transitioning away from constructs (i.e., circumscribed syndromes and diagnostic categories) that were established in a top-down manner but have not held up to validation scrutiny, towards evidence-based “transdiagnostic” dimensions (Dalgleish et al., Psychopathology is transitioning away from constructs (i.e., circumscribed syndromes and diagnostic categories) that were established in a top-down manner but have not held up to validation scrutiny, towards evidence-based “transdiagnostic” dimensions (Dalgleish et al., 2020). The dual challenges facing the field are establishing the relevant domains of (dys)function and defining pathology within this space. Establishing the domains of functioning largely corresponds to structural questions, akin to anatomy. Defining pathology, however, requires elaborating the dynamic processes that correspond to the function of those domains, validation scrutiny, towards evidence-based transdiagnostic dimensions (Dalgleish et al., 2020). The dual challenges facing the field are establishing the relevant domains of (dys)function and defining pathology within this space. Establishing the domains of functioning largely corresponds to structural questions, akin to anatomy. Defining pathology, however, requires elaborating the dynamic processes that correspond to the function of those domains, 10 INTERPERSONAL THEORY akin to physiology. These are not independent goals, because mapping the structure of psychological domains facilitates accurate description of functional processes (Hopwood et al., 2015; Wright & Kaurin, 2020). In trying to address these demands, emerging systems of psychopathology have prioritized different aspects of the challenges (Dalgleish et al., 2020). For example, the HiTOP has targeted structure, taking a decidedly empirical approach by using quantitative models to establish points of convergence and divergence across traditional diagnoses and symptoms (Kotov et al., 2021). However, thus far the HiTOP Consortium has not targeted the functional processes associated with the identified domains. It therefore remains a descriptive structural model to date. In contrast, other approaches, often emerging directly from clinical intervention, have focused on transdiagnostic processes but have not directly sought to challenge the organizing structure of psychiatric disorders (e.g., Harvey et al., 2004; Hayes & Hofmann, 2018). Arguably, the RDoC framework falls somewhere in between, because it proposes novel structural elements organized around basic neurobehavioral processes with psychiatric relevance (Cuthbert, 2014). These models each have important strengths, but are incomplete because they lack organizing functional theories. For instance, the HiTOP Consortium follows the guiding principle that quantitative empirical structural modeling will lead to a valid model for psychopathology. Its strength is its structural validation, but it is limited by its lack of articulation of functioning within the domains it identifies. Because it is primarily descriptive rather than process-focused or explanatory, it cannot distinguish between function and dysfunction. In contrast, theoretical models of specific psychiatric disorders have focused on the mechanistic functional processes that offer prime targets for intervention in treatment, though INTERPERSONAL THEORY 12 Wilson et al., 2017 for a meta-analysis). Relatedly, agency and communion provide two of the pillars of the DSM-5 Alternative Model for Personality Disorders (Williams & Simms, 2016; Wright, Pincus, et al. 2012; Wright, Thomas, et al., 2012) and the HiTOP (Girard et al., 2017). The interpersonal content and implications of personality and psychopathology are not limited to these two domains, and all major domains of adaptive and maladaptive personality models (e.g., negative affectivity, disinhibition) are reflected in interpersonal profiles (Ansell & Pincus, 2004; Du et al., 2021; Wright, Pincus, et al. 2012). For many other forms of psychopathology, CIIT assumes a pathoplastic, or moderating relationship with interpersonal dispositions. Pathoplasticity is characterized by a mutually influencing non-etiological The interpersonal content and implications of personality and psychopathology are not limited to these two domains, and all major domains of adaptive and maladaptive personality models (e.g., negative affectivity, disinhibition) are reflected in interpersonal profiles (Ansell & Pincus, 2004; Du et al., 2021; Wright, Pincus, et al. 2012). For many other forms of psychopathology, CIIT assumes a pathoplastic, or moderating relationship with interpersonal dispositions. Pathoplasticity is characterized by a mutually influencing non-etiological relationship between psychopathology and another psychological system (Widiger & Smith, 2008). Pathoplastic relationships between interpersonal dispositions and psychopathology have been found in social phobia (Cain et al., 2010; Kachin et al., 2001), generalized anxiety disorder (Przeworski et al., 2011; Salzer et al., 2008, 2011), panic disorder (Zilcha-Mano et al., 2015), depression (Dawood et al., 2013; Cain et al., 2012; Simon et al., 2015), post-traumatic stress disorder (Thomas et al., 2014), eating pathology (Ambwani & Hopwood, 2009; Hopwood et al., 2007), maladaptive perfectionism (Slaney et al., 2006), and fear of failure (Wright et al., 2009). 2008). Pathoplastic relationships between interpersonal dispositions and psychopathology have been found in social phobia (Cain et al., 2010; Kachin et al., 2001), generalized anxiety disorder (Przeworski et al., 2011; Salzer et al., 2008, 2011), panic disorder (Zilcha-Mano et al., 2015), depression (Dawood et al., 2013; Cain et al., 2012; Simon et al., 2015), post-traumatic stress disorder (Thomas et al., 2014), eating pathology (Ambwani & Hopwood, 2009; Hopwood et al., 2007), maladaptive perfectionism (Slaney et al., 2006), and fear of failure (Wright et al., 2009). However, although the interpersonal circumplex structural model is centrally important to CIIT, the theory asserts that personality is dynamic and transactional (Dawood et al., 2018; Kiesler, 1996; Pincus & Hopwood, 2012). INTERPERSONAL THEORY 11 these disorder-specific theories (and subsequent treatments) suffer because they are not organized within valid structure. For instance, myriad publications propose pathogenesis and maintenance mechanisms of major depression, generalized anxiety, borderline personality disorder, and so forth. As such, they remain overly specific and compartmentalized from a broader integrative perspective represented in the structural literature (cf. unified protocol; Barlow et al., 2017). Taken together, the strengths and limitations of each of these efforts highlight that integrative models of psychopathology should be built upon a structurally valid model, but must also be focused and context specific, incorporating dynamic processes in those contexts (Hopwood, 2018; Hopwood et al., 2015; Leary, 1957; Wright & Kaurin, 2020). This is because domains of functioning are organized around specific purposes in relevant contexts. CIIT focuses on delimited domains of functioning, with identified functional purposes, providing a tractable level of breadth. Within these domains it has independently developed a well- established structural model, the interpersonal circumplex (top left of Figure 1), which provides the coordinates for mapping processes within and across levels of functioning (i.e., the interpersonal situation; top right of Figure 1). Structurally, the dimensions of the interpersonal circumplex can be conceptualized as domains that can be found within extant models of personality and psychopathology. For instance, the first two dimensions of the Big-Five, extraversion and agreeableness (Goldberg, 1990), can be understood as rotational variants of the domains of agency and communion (DeYoung et al., 2013; McCrae & Costa, 1989; Pincus, 2002; Wiggins & Trapnell, 1996). Traditional personality disorder diagnoses reflect key points within this space, such that distinct dispositional interpersonal styles are associated with most personality disorder diagnoses (see INTERPERSONAL THEORY Functioning fundamentally concerns dynamic processes that unfold over time at different temporal scales, from the micro (the milliseconds it takes to perceive and react to stimuli) to the macro (change in traits across developmental epochs). These processes interface with context to variably determine (mal)adaptivity. For instance, well-functioning perceptual, construal, and behavioral selection processes allow INTERPERSONAL THEORY 13 flexible and adaptive moment-to-moment interaction with others during social interactions. And, what is developmentally appropriate at one stage of life (e.g., the demanding behavior of a 3- year-old) is often considered maladaptive if they persist beyond that stage. Interpersonal situations involve more than one person (we use only dyadic examples for simplicity), only one of which must be present or real. That is, the interpersonal situation encompasses direct in-person interactions as well as mental representations of interactions (recollected and imagined), and everything in between. Consistent patterns of functioning in these situations are a major part of what define personality and psychopathology (Cain & Pincus, 2016; Pincus & Wright, 2011; Sullivan, 1953a). By definition, interpersonal situations are dynamic and transactional, involving an ebb and flow of behavior as well as a give and take between participants. Kiesler (1996) unpacked the key components of the interpersonal transaction cycle, which involve both overt behavior and each person’s mental construal of the other. These are detailed in the top right of Figure 1. As a cyclical process, where it begins is arbitrary, but we might begin with Person A’s overt behavior (Box 1). This then is perceived by Person B (Path 1), which can then be said to impact Person B (i.e., their internal reaction; Box 2), in turn Person B responds with an overt response to Person A (the process of selecting this is reflected in Path 2), which is perceived and has its own impact. This sequence continuously cycles throughout the interaction. The boxes and paths in this diagram can be isolated to consider the constituents and components. For example, Box 1 includes overt behavior, Path 1 involves attention and perception, Box 2 includes construal and affective reactions, and Path 2 might involve motivation. Although this sort of transactional process and diagram is widely represented generically in social-cognitive formulations of behavior, CIIT brings the contextual focus and the structural model of the interpersonal circumplex to populate these processes with INTERPERSONAL THEORY 14 specific content. By doing so, it tethers the structural to the dynamic and mechanistic, which provides a critical bridge to function. The dimensions of agency and communion define the purpose and meaning of behavior, its perceptions, impacts, and the motivations that drive it. Within interpersonal situations, affect serves as the primary reinforcer and punisher, signaling whether an individual’s motives and goals are being satisfied. Motivations to engage in interpersonal situations and for which behaviors are expressed are organized around agency and communion (Horowitz et al., 2006) and are represented by strivings for self-esteem (an agentic motivation) and security (a communal motivation) in interpersonal situations, and when satisfied lead to decreases in negative affect and vulnerability, and increases in positive affect. Individual differences in motivations, imbued through basic biological temperament (Depue & Collins, 1999; Depue & Marrone-Strupinsky, 2005) and shaped through development (e.g., Critchfield & Benjamin, 2008; Wright, Pincus, & Lenzenweger, 2012) and experience, influence the construal of situations, the types of situations entered and avoided, and the behaviors exhibited within situations in a probabilistic fashion based on available affordances. Individual differences in interpersonal functioning are shaped and can lead to longstanding change via accumulated outcomes of those situations. In this way, the trait and process approaches interpenetrate and form a coherent whole across brief and extended timescales (Bottom of Figure 1). Interpersonal theory provides a network of falsifiable hypotheses about the patterns of behaviors that contribute to increased positive affect and stability, or negative affect and dissolution, of interactions. The most prominent is complementarity, which, in simple terms, states that in a probabilistic fashion dominant behavior pulls for or invites submission from the other, and warmth invites warmth. Transactions that adhere to these patterns should lead to INTERPERSONAL THEORY 16 child interaction patterns were associated with less affiliative behavior as a function of ADHD diagnosis (Nilsen et al., 2015), maternal control is influenced by evocative gene-environment correlations (Klahr et al., 2013), and that wives’ depressive symptoms influenced dynamics of dominance (wife increased while husband decreased over the interaction) but husbands’ depressive symptoms influenced dynamics of affiliation (affiliation in husband and wife decreased over the interaction; Lizdek et al., 2016). Looking across situations in daily life using ecological momentary assessment, detailed articulations of contextualized dynamic processes associated with psychopathology have been empirically modeled, such as the perceptions of withdrawal→negative affect→hostility characteristic of borderline pathology (Sadikaj et al., 2013), or perceptions of dominance→negative affect→hostility characteristic of narcissistic pathology (Wright et al., 2017). Studies have also shown that decreases in binge eating were linked to increased interpersonal complementarity principles (Ambwani et al., 2015), marijuana use was associated with greater hostility in perceptions and behavior (Ansell et al., 2015), and stronger links have been found between social perception, behavior, and embarrassment among those diagnosed with social phobia relative to controls (Sadikaj et al., 2015). child interaction patterns were associated with less affiliative behavior as a function of ADHD diagnosis (Nilsen et al., 2015), maternal control is influenced by evocative gene-environment correlations (Klahr et al., 2013), and that wives’ depressive symptoms influenced dynamics of dominance (wife increased while husband decreased over the interaction) but husbands’ depressive symptoms influenced dynamics of affiliation (affiliation in husband and wife decreased over the interaction; Lizdek et al., 2016). Looking across situations in daily life using ecological momentary assessment, detailed articulations of contextualized dynamic processes associated with psychopathology have been empirically modeled, such as the perceptions of withdrawal→negative affect→hostility characteristic of borderline pathology (Sadikaj et al., 2013), or perceptions of dominance→negative affect→hostility characteristic of narcissistic pathology (Wright et al., 2017). Studies have also shown that decreases in binge eating were linked to increased interpersonal complementarity principles (Ambwani et al., 2015), marijuana use was associated with greater hostility in perceptions and behavior (Ansell et al., 2015), and stronger links have been found between social perception, behavior, and embarrassment among those diagnosed with social phobia relative to controls (Sadikaj et al., 2015). In sum, functional interpersonal behavior serves to satisfy individuals’ motivations and goals within the bounds of developmentally and culturally appropriate expectations (Tenet 4). Rich individual differences in basic motivations and goals allow for wide variability in functional behavior. INTERPERSONAL THEORY 15 greater positive affect and stability of the transaction or relationship, and those that do not lead to renegotiation and possibly negative affect and dissolution. A comprehensive treatment of the interpersonal complementarity literature is beyond the scope of this article (see Sadler et al., greater positive affect and stability of the transaction or relationship, and those that do not lead to renegotiation and possibly negative affect and dissolution. A comprehensive treatment of the interpersonal complementarity literature is beyond the scope of this article (see Sadler et al., 2011 for a review). However, it is informative because it illustrates a well-articulated hypothesis for normative transactional processes, linking the dynamics of behavior and affect within and across interpersonal situations. It therefore offers one potential framework for defining specific adaptive and maladaptive processes. Indeed, this generative concept has fostered a large supportive literature, yet ongoing research addresses remaining questions. For instance, what is the exact timescale on which complementarity occurs, is it consistent across the full range of behavior, and how is it influenced by contextual constraints (e.g., coworker vs. intimate relationships)? Emphasizing the dynamics of interpersonal functioning within and across situations has led to the development of a sophisticated assessment and modeling apparatus (Pincus et al., 2014), including measuring behaviors unfolding moment-to-moment in laboratory-based interpersonal transactions (e.g., Ross et al., 2017; Sadler et al., 2009), behaviors expressed in interpersonal situations encountered in daily life (e.g., Moskowitz, 2009), and traditional dispositional summaries (Locke, 2011). To expand on the brief review of dispositional interpersonal findings above, results using these “dynamic” approaches to examine interpersonal functioning find that individuals vary considerably in their interpersonal behavior within and across situations in predictable ways. For instance, the fine-grained second-by-second coding method of Continuous Assessment of Interpersonal Dynamics (Lizdek et al., 2012; Girard & Wright, 2018), has revealed that conflict tasks are characterized by more cold behavior than non- conflict tasks and more negative affect in response to coldness (Hopwood et al., 2020), mother- INTERPERSONAL THEORY 17 about whether observed patterns of dysfunction are explained by biased perception, differences in contingencies between affect and perception, or selection of maladaptive responses in response to others’ behavior to explain hostile dynamics (viz. Hopwood, 2018; Pincus & Hopwood, 2012), to name some possible examples. We now shift to discussing the importance of placing questions about mechanisms within a framework that spans different levels of analysis. about whether observed patterns of dysfunction are explained by biased perception, differences in contingencies between affect and perception, or selection of maladaptive responses in response to others’ behavior to explain hostile dynamics (viz. Hopwood, 2018; Pincus & Hopwood, 2012), to name some possible examples. We now shift to discussing the importance of placing questions about mechanisms within a framework that spans different levels of analysis. INTERPERSONAL THEORY Dysfunction is therefore defined as sustained departures from these expected patterns, but may manifest in different ways and due to multiple potential process breakdowns. The key point is that the interpersonal situation allows for strong inference (Platt, 1964) tests of the components of the process giving rise to or impacted by psychopathology, thereby suggesting optimal interventions. In other words, one can pit competing hypotheses Levels of Analysis: From Striatum to the Street Though theories of psychopathology often prioritize one level of analysis (e.g., biological, motivational, cognitive, affective, behavioral), multi-level formulation is necessary for a comprehensive understanding of dysfunction because humans are complexes of multiple interacting biobehavioral systems. Moving from static structural models to functional models of complex behavioral dynamics necessitates integration across levels of analysis to achieve a mechanistic or explanatory account of psychopathology. This is the logic underlying the RDoC approach, which seeks to span “units” of analysis using domains that are presumed to reflect key circuits of functioning. In much the same way, CIIT’s two principal domains of agency and communion serve as the organizing framework to span levels of functioning in a coherent and cohesive manner. Using the same structural model for different levels of functioning was among the major early innovations of interpersonal theory (Leary, 1957; see Tenet 3 in text and supplement). The levels that have received explicit attention in CIIT have varied over time and with evolving emphases in psychological science (Pincus et al., 2014). Early mainstays included overt behavior, values, and conscious description (i.e., self-report), as well as others less common in psychopathology research today, such as private symbolism and unexpressed INTERPERSONAL THEORY 18 unconscious. In more recent articulations, although there remains a sustained interest in overt behavior, values, and self-concept (e.g., global self-report), there has also been expansion of interpersonal emphases to include perceptual, cognitive (e.g., efficacy), and neurobiological levels of functioning. Our goal here is not to define the exact set of levels that are necessary or valuable for understanding functioning and psychopathology. Instead, we aim to make the more general points that a multilevel understanding is necessary, and the interpersonal circumplex fulfils the need for an orienting structure to chart interpersonal functioning across levels of analysis in personality and psychopathology. However, acknowledging that human functioning and psychopathology are realized across multiple levels does not resolve the difficulty of understanding how these levels associate and interact as an integrated whole (e.g., Bilder et al., 2013; Eronen, 2021). Structure and function are unlikely to be mirrored across levels and contexts with simple one-to-one correspondence. That is, agency and communion serve as broad abstract domains that will be instantiated more specifically at each level in relevant manifestations (see supplemental material Tenet 3) of genetic profiles, neural circuitries (e.g., corticolimbic-striatal networks), motivational strivings (e.g., affiliative incentive reward), mental representation (e.g., memories of warm interactions with others), and specific behaviors (e.g., smiling, affectionate touch). From a descriptive standpoint, there may be reasons to expect some degree of alignment or cohesion across levels (e.g., individual differences in average domain-relevant behavior correspond to neural structural or network differences). But from a functional standpoint, we would also expect interplay and reciprocal influence across units. The interpersonal situation provides the necessary contextual crucible for elaborating how these levels interface, and links trait structure and dynamic processes over diverse INTERPERSONAL THEORY 20 specific situations, how they respond, and whether those situations are reinforcing or punishing, which in turn (slowly and over time) influence trait levels. These sorts of models have become popular in a generic fashion in contemporary personality theory (e.g., Baumert et al., 2017; specific situations, how they respond, and whether those situations are reinforcing or punishing, which in turn (slowly and over time) influence trait levels. These sorts of models have become popular in a generic fashion in contemporary personality theory (e.g., Baumert et al., 2017; Wrzus & Roberts, 2017). However, CIIT is unique in providing the necessary content (agency and communion) and context (social relating) to generate specific hypotheses about how these processes occur. These include normative trait differences and development, as well as departures from expected developmental trajectories. In this section we have described the importance and potential of using the same structure to conceptualize variation in personality and psychopathology across interpenetrating levels of analysis. This shared structure across levels is what allows us to formulate connections from the striatum to the street and from development to daily life within the same theoretical model (e.g., biological bases establish set-points/reaction ranges that are shaped through development, setting up motivations that guide goal-directed pursuits in adulthood that are instantiated in day-to-day behavior). An elaborated multi-level dynamic model facilitates the articulation of specific dysfunctions emblematic of psychopathology, and supports strong inference tests of competing hypotheses of the location of the dysfunction (e.g., instability or excess of behavior vs. poorly applied behavior, unstable or inaccurate perceptions or construal of situational cues vs. mis- calibrated response). In the final section, we illustrate how CIIT can provide a model to define primarily interpersonal psychopathology, as well as how other-domain primary psychopathology results in interpersonal dysfunction. INTERPERSONAL THEORY 19 timescales. First, over a very brief timescale, as illustrated in the top right of Figure 1, different modalities of functioning weave together to form what is experienced as a seamless process (e.g., motivation→perception→construal→physiology→affect→behavior). Thus, the dynamics of the interpersonal situation serve as the context to study how the levels interact with each other and interpenetrate to form a coherent system. Although aspects of these sequences may be experienced as essentially automatic, distinguishing these levels provides entry points to articulating functional and dysfunctional aspects of the process. For instance, does an individual’s overt behavior match their intended interpersonal communication? Or, does their perception and construal of another’s behavior match a consensus perception and construal of the same person? Second, over much larger timescales, by emphasizing a multilevel understanding of functioning, CIIT can account for individual differences in trait development and change. Individuals start life with genetically driven differences in relevant neural structures that manifest in the degree to which they are motivated for and experience reward from social interaction. Innate temperamental differences are upregulated or downregulated through positive and negative feedback loops in early life interpersonal situations, which lead to lasting individual differences in personality and its development through continued interactions. These processes are detailed by Depue and colleagues (Depue, 2006; Depue & Collins, 1999; Depue & Lenzenweger, 2016; Depue & Marrone-Strupinsky, 2005). More recently, Waller and Wagner (2019) have outlined how these long-term developmental processes may give rise to interpersonal psychopathology in a novel neurobehavioral model entirely consistent with CIIT. The bottom of Figure 1 also illustrates how individual differences and the interpersonal situation reciprocally influence each other, such that an individual’s motives influence the selection of Beyond Description: From Cause to Effect The interpersonal situation distinguishes multiple kinds of problems, offering a clear, clinically useful, and theory-guided taxonomy for dysfunctional processes. These process-based INTERPERSONAL THEORY 21 formulations serve not only to define psychopathology that is distinctly interpersonal (cause), but also can be used to understand how other forms of psychopathology can significantly impair basic interpersonal functioning (effect). Examples of traditional psychiatric diagnoses that are primarily interpersonal include the personality disorders (e.g., borderline, narcissistic, antisocial). Yet, affective, anxiety, externalizing, thought, eating, and other disorders often have significant interpersonal impacts as well. For instance, affect can serve as a moderator of situational construal and behavior through its well-known effects on perceptual and motivational processes. CIIT provides a framework or lens for understanding dysfunctional processes that are caused by interpersonal impairments or that are the effect of other impairments. We end with several exemplars intended to bring these articulations to life. formulations serve not only to define psychopathology that is distinctly interpersonal (cause), but also can be used to understand how other forms of psychopathology can significantly impair basic interpersonal functioning (effect). Examples of traditional psychiatric diagnoses that are primarily interpersonal include the personality disorders (e.g., borderline, narcissistic, antisocial). Yet, affective, anxiety, externalizing, thought, eating, and other disorders often have significant interpersonal impacts as well. For instance, affect can serve as a moderator of situational construal and behavior through its well-known effects on perceptual and motivational processes. The personality disorders have often been considered the “interpersonal disorders,” and therefore detailed interpersonal formulations of these DSM constructs exist (Benjamin, 1996; Hopwood et al., 2013; Pincus & Hopwood, 2012). Some prime examples include characteristic interpersonal situations defined by distorted perceptions of or hypersensitive attunement to others’ interpersonal behavior, followed by affective and/or self-dysregulation, giving rise to self-protective motives, and maladaptive behaviors. In the case of borderline pathology, these revolve around affiliation/withdrawal, whereas in the case of narcissistic pathology The personality disorders have often been considered the “interpersonal disorders,” and therefore detailed interpersonal formulations of these DSM constructs exist (Benjamin, 1996; Hopwood et al., 2013; Pincus & Hopwood, 2012). Some prime examples include characteristic interpersonal situations defined by distorted perceptions of or hypersensitive attunement to others’ interpersonal behavior, followed by affective and/or self-dysregulation, giving rise to self-protective motives, and maladaptive behaviors. In the case of borderline pathology, these revolve around affiliation/withdrawal, whereas in the case of narcissistic pathology dominance/status is the main organizing theme. INTERPERSONAL THEORY 22 though the core dysfunction is often presumed to be non-interpersonal, they nevertheless result in interpersonal impairments. Social anxiety is an apt example that often involves self-fulfilling interpersonal processes. For instance, strong threat-sensitivity manifests as hypersensitivity to rejection/evaluation, which leads to a variety of self-protective interpersonal behaviors that often ensure the individual is rejected, thereby confirming their fears and negatively rewarding their continued withdrawal. Other classical syndromes, such as major depression and generalized anxiety, are similarly associated with interpersonal impairments that can be enhanced with a CIIT formulation (Cain et al., 2012; Przeworski et al., 2011). Underscoring the integrative power of CIIT, both the psychoanalytic and cognitive-behavioral traditions have identified interpersonal dispositions as risk factors for depression, though using alternative names for agentic (introjective/autonomy) and communal (anaclitic/sociotropic) motives (Blatt, 2008). In either case, the deeply held motives, when thwarted, leave the individual vulnerable to depression, which then results in, among other things, interpersonal passivity and withdrawal. These are prime targets for intervention in well-supported treatments for disrupting depressive dynamics, such as behavioral activation. though the core dysfunction is often presumed to be non-interpersonal, they nevertheless result in interpersonal impairments. Social anxiety is an apt example that often involves self-fulfilling interpersonal processes. For instance, strong threat-sensitivity manifests as hypersensitivity to rejection/evaluation, which leads to a variety of self-protective interpersonal behaviors that often ensure the individual is rejected, thereby confirming their fears and negatively rewarding their continued withdrawal. Other classical syndromes, such as major depression and generalized anxiety, are similarly associated with interpersonal impairments that can be enhanced with a CIIT formulation (Cain et al., 2012; Przeworski et al., 2011). Underscoring the integrative power of CIIT, both the psychoanalytic and cognitive-behavioral traditions have identified interpersonal dispositions as risk factors for depression, though using alternative names for agentic (introjective/autonomy) and communal (anaclitic/sociotropic) motives (Blatt, 2008). In either case, the deeply held motives, when thwarted, leave the individual vulnerable to depression, which then results in, among other things, interpersonal passivity and withdrawal. These are prime targets for intervention in well-supported treatments for disrupting depressive dynamics, such as behavioral activation. The thought-disorder spectrum, which encompasses both positive and negative symptoms, is also associated with significant interpersonal impairments. It is readily apparent how perceptions distorted by positive symptoms, such as paranoia and persecutory delusions, might interfere with the natural harmony of an interpersonal situation (e.g., Lewis & Ridenour, 2020). INTERPERSONAL THEORY Alternatively, some manifestations of pathology reflect overvalued/extreme motivations that color most interpersonal situations, leading to inflexible or misapplied behavior (e.g., the tendency for narcissistic individuals to view all situations as opportunities to self-promote without consideration of others’ goals, or the tendency for paranoid individuals to perceive nearly all ambiguous behavior as threatening). dominance/status is the main organizing theme. Alternatively, some manifestations of pathology reflect overvalued/extreme motivations that color most interpersonal situations, leading to inflexible or misapplied behavior (e.g., the tendency for narcissistic individuals to view all situations as opportunities to self-promote without consideration of others’ goals, or the tendency for paranoid individuals to perceive nearly all ambiguous behavior as threatening). When considering other domains of psychopathology, such as the internalizing spectrum, INTERPERSONAL THEORY Some have even argued that delusions may best be construed as impairments in basic social-cognitive processes (Bell et al., 2020). So, too, do negative symptoms, such as avolition, apathy, and flattened affect, dampen transactional processes at various points in the interpersonal cycle. One aspect of the broad, complex neural circuitry of negative symptoms (e.g., ventral 23 INTERPERSONAL THEORY tegmental area to nucleus accumbens; see supplemental Tenet 3) likely reflects, in part, disruptions in the same dopaminergic neural circuits as agentic incentive sensitivity, and manifests in fewer interpersonal pursuits and lower responsivity when opportunities or requests from others arise (Brown & Pluck, 2000). Consistent with CIIT, the cognitive-interpersonal model of eating disorders suggests that difficulties with interpersonal relatedness predispose to the development of anorexia nervosa and are also secondary consequences of the disorder, contributing to the maintenance of the illness (Treasure et al., 2020). Adversities in social development and impairments in social cognition can impact interpersonal behavior and disrupt identity development, potentially stimulating perfectionistic strivings and greater internalization of body image ideals to define the self. People with anorexia nervosa are more likely to report pervasive negative schemas regarding self and others and difficulties in emotion regulation, with a greater use of dysfunctional interpersonal emotion-regulation strategies (Oldershaw et al., 2015). In the acute phase, starvation has numerous interpersonal effects via its impact on social cognition, including deficits in nonverbal emotional expression, an inability to provide a spontaneous social narrative and to identify the social salience of stimuli, threat sensitivity, maladaptive social comparisons, misalignments in social reciprocity, and the avoidance of conflict (Ambwani et al., 2016; Oldershaw et al., 2018). Impairments in nonverbal communication and a lack of reciprocation of warmth can elicit aversive responses in others via interpersonal complementarity, leading to social exclusion and isolation. This can further the overvaluation of thinness and self-control for self-image. Consistent with CIIT, the cognitive-interpersonal model of eating disorders suggests that difficulties with interpersonal relatedness predispose to the development of anorexia nervosa and are also secondary consequences of the disorder, contributing to the maintenance of the illness (Treasure et al., 2020). Adversities in social development and impairments in social cognition can impact interpersonal behavior and disrupt identity development, potentially stimulating perfectionistic strivings and greater internalization of body image ideals to define the self. People with anorexia nervosa are more likely to report pervasive negative schemas regarding self and others and difficulties in emotion regulation, with a greater use of dysfunctional interpersonal emotion-regulation strategies (Oldershaw et al., 2015). INTERPERSONAL THEORY 24 INTERPERSONAL THEORY In the acute phase, starvation has numerous interpersonal effects via its impact on social cognition, including deficits in nonverbal emotional expression, an inability to provide a spontaneous social narrative and to identify the social salience of stimuli, threat sensitivity, maladaptive social comparisons, misalignments in social reciprocity, and the avoidance of conflict (Ambwani et al., 2016; Oldershaw et al., 2018). Impairments in nonverbal communication and a lack of reciprocation of warmth can elicit aversive responses in others via interpersonal complementarity, leading to social exclusion and isolation. This can further the overvaluation of thinness and self-control for self-image. Interpersonal dysfunctions represent both antecedents of problems in living and also the signs of psychopathologies with origins in other (e.g., affective) systems. Thus, CIIT can serve as a uniting and organizing framework for both causes and effects of psychiatric problems. Conclusion In summary, psychopathology is facing an exciting time of change and revision. There is considerable momentum to make paradigmatic shifts from longstanding, albeit fundamentally flawed, conceptualizations of psychological dysfunction. At the same time, as Dalgleish and colleagues (2020) argue, these efforts suffer from a lack of organizing theories that limit their ability to provide a comprehensive and cohesive account of psychopathology. Here we have presented CIIT as an exemplar paradigm for defining psychological functioning and dysfunction, bridging structural and mechanistic models, relevant across levels of analysis, and spanning temporal scales. We have highlighted how many of the interpersonal tradition’s early innovations such as empirical structure, dimensional constructs, and defining function across levels and modalities, have proven prescient, and are now cornerstones of contemporary psychopathology science. Furthermore, CIIT’s integrative nature has allowed it to remain current, incorporating the advances of other areas reflected in longstanding tenets of interpersonal theory (e.g., control theory, neurobehavioral circuitry). Though many of CIIT’s tenets are rooted in substantial empirical support, it remains an active driver of research programs that seek to elaborate these tenets and expand the theory’s reach. It can serve double duty as a way to explain psychopathology that is primarily interpersonal in nature, as well as a critical lens to understand interpersonal dysfunction secondary to other forms of psychopathology. INTERPERSONAL THEORY 25 References Allen, T.A., & DeYoung, C.G. (2017). Personality neuroscience and the five factor model. Oxford handbook of the five factor model, 319-352. Allport, G. (1937). Personality: A psychological interpretation. Holt. Ambwani, S., & Hopwood, C.J. (2009). The utility of considering interpersonal problems in the assessment of bulimic features. Eating Behaviors, 10, 247–253. Ambwani, S., & Hopwood, C.J. (2009). The utility of considering interpersonal problems in the assessment of bulimic features. Eating Behaviors, 10, 247–253. Ambwani, S., Roche, M.J., Minnick, A.M., & Pincus, A.L. (2015). Negative affect, interpersonal perception, and binge eating behavior: An experience sampling study. International Journal of Eating Disorders, 48, 715-726. American Psychiatric Association. (2013). Diagnostic and Statistical Manual of Mental Disorders, fifth edition (DSM-5). Authors. Anchin, J.C., & Pincus, A.L. (2010). Evidence-based interpersonal psychotherapy with personality disorders: Theory, components, and strategies. In Evidence-based treatment of personality dysfunction: Principles, methods, and processes. (pp. 113-166). American Psychological Association. Anchin, J.C., & Pincus, A.L. (2010). Evidence-based interpersonal psychotherapy with personality disorders: Theory, components, and strategies. In Evidence-based treatment of personality dysfunction: Principles, methods, and processes. (pp. 113-166). American Psychological Association. Ansell, E.B., Kurtz, J.E., & Markey, P.M. (2008). Gender differences in interpersonal complementarity within roommate dyads. Personality and Social Psychology Bulletin, 34(4), 502-512. Ansell, E.B., Laws, H.B., Roche, M.J., & Sinha, R. (2015). Effects of marijuana use on impulsivity and hostility in daily life. Drug and Alcohol Dependence, 148, 136-142. Ansell, E.B., & Pincus, A.L. (2004). Interpersonal perceptions of the five-factor model of personality: An examination using the structural summary method for circumplex data. Multivariate Behavioral Research, 39, 167-201. Ansell, E.B., Laws, H.B., Roche, M.J., & Sinha, R. (2015). Effects of marijuana use on impulsivity and hostility in daily life. Drug and Alcohol Dependence, 148, 136-142. Ansell, E.B., & Pincus, A.L. (2004). Interpersonal perceptions of the five-factor model of personality: An examination using the structural summary method for circumplex data. Multivariate Behavioral Research, 39, 167-201. INTERPERSONAL THEORY 26 Bakan, D. (1966). The duality of human existence: Isolation and communion in Western man. Boston, MA: Beacon Press. Barlow, D.H., Farchione, T.J., Sauer-Zavala, S., Latin, H.M., Ellard, K.K., Bullis, J.R., ... & Cassiello-Robbins, C. (2017). Unified Protocol for Transdiagnostic Treatment of Emotional Disorders. Oxford: Oxford University Press. Bartholomew, K., & Horowitz, L.M. (1991). Attachment styles among young adults: A test of a four-category model. Journal of Personality and Social Psychology, 61, 226–244. Baumeister, R.F., & Leary, M.R. (1995). The need to belong: desire for interpersonal attachments as a fundamental human motivation. Psychological Bulletin, 117, 497-529. Baumert, A., Schmitt, M., Perugini, M., Johnson, W., Blum, G., Borkenau, P., ... & Wrzus, C. (2017). Integrating personality structure, personality process, and personality development. European Journal of Personality, 31(5), 503-528. Bell, V., Raihani, N., & Wilkinson, S. (2020). Derationalizing delusions. Clinical Psychological Science. Advanced online publication https://doi.org/10.1177/2167702620951553. Benjamin, L.S. (1993). Every psychopathology is a gift of love. Psychotherapy Research, 3, 1- 24. Benjamin, L.S. (2002). Interpersonal Diagnosis and Treatment of Personality Disorders. Guilford Press. Benjamin, L.S. (2018). Interpersonal reconstructive therapy for anger, anxiety, and depression. Washington, DC: American Psychological Association. Bilder, R. M., Howe, A. G., & Sabb, F. W. (2013). Multilevel models from biology to psychology: Mission impossible? Journal of Abnormal Psychology, 122(3), 917-927. Bliton, C.F., & Pincus, A.L. (2020). Initial construction and validation the Inventory of Bakan, D. (1966). The duality of human existence: Isolation and communion in Western man. Boston, MA: Beacon Press. Bakan, D. (1966). The duality of human existence: Isolation and communion in Western man. Boston, MA: Beacon Press. Barlow, D.H., Farchione, T.J., Sauer-Zavala, S., Latin, H.M., Ellard, K.K., Bullis, J.R., ... & Cassiello-Robbins, C. (2017). Unified Protocol for Transdiagnostic Treatment of Baumeister, R.F., & Leary, M.R. (1995). The need to belong: desire for interpersonal Baumeister, R.F., & Leary, M.R. (1995). The need to belong: desire for interpersonal attachments as a fundamental human motivation. Psychological Bulletin, 117, 497-529. Baumert, A., Schmitt, M., Perugini, M., Johnson, W., Blum, G., Borkenau, P., ... & Wrzus, C. (2017). Integrating personality structure, personality process, and personality development. European Journal of Personality, 31(5), 503-528. Bell, V., Raihani, N., & Wilkinson, S. (2020). Derationalizing delusions. Clinical Psychological Science. Advanced online publication https://doi.org/10.1177/2167702620951553. Benjamin, L.S. (1993). Every psychopathology is a gift of love. Psychotherapy Research, 3, 1- 24. Benjamin, L.S. (2002). Interpersonal Diagnosis and Treatment of Personality Disorders. Guilford Press. Benjamin, L.S. (2018). INTERPERSONAL THEORY Interpersonal reconstructive therapy for anger, anxiety, and depression. Washington, DC: American Psychological Association. Bilder, R. M., Howe, A. G., & Sabb, F. W. (2013). Multilevel models from biology to attachments as a fundamental human motivation. Psychological Bulletin, 117, 497-529. ., Schmitt, M., Perugini, M., Johnson, W., Blum, G., Borkenau, P., ... & Wrzus, C. (2017). Integrating personality structure, personality process, and personality development. European Journal of Personality, 31(5), 503-528. Bell, V., Raihani, N., & Wilkinson, S. (2020). Derationalizing delusions. Clinical Psychological Science. Advanced online publication https://doi.org/10.1177/2167702620951553. Benjamin, L.S. (1993). Every psychopathology is a gift of love. Psychotherapy Research, 3, 1- 24. 24. Benjamin, L.S. (2002). Interpersonal Diagnosis and Treatment of Personality Disorders. Guilford Press. Benjamin, L.S. (2018). Interpersonal reconstructive therapy for anger, anxiety, and depression. Washington, DC: American Psychological Association. Bilder, R. M., Howe, A. G., & Sabb, F. W. (2013). Multilevel models from biology to psychology: Mission impossible? Journal of Abnormal Psychology, 122(3), 917-927. Bliton, C.F., & Pincus, A.L. (2020). Initial construction and validation the Inventory of INTERPERSONAL THEORY 27 rpersonal Influence Tactics Circumplex (IIT-C) scales. Assessment, 27, 688-705. differences. Journal of Personality and Social Psychology, 58(3), 464–471. Brown, K.W., & Moskowitz, D.S. (1998). Dynamic stability of behavior: The rhythms of our interpersonal lives. Journal of personality, 66(1), 105-134. interpersonal lives. Journal of personality, 66(1), 105-134. Brown, R.G., & Pluck, G. (2000). Negative symptoms: the ‘pathology’ of motivation and goal- directed behaviour. Trends in neurosciences, 23(9), 412-417. directed behaviour. Trends in neurosciences, 23(9), 412-417. Cain, N.M., Ansell, E.B., Wright, A.G.C., Hopwood, C.J., Thomas, K.M., Pinto, A.,… Grilo, C.M. (2012) Interpersonal pathoplasticity in the course of major depression. Journal of Consulting and Clinical Psychology, 80, 78-86. Cain, N.M., & Pincus, A.L. (2016). Treating maladaptive interpersonal signatures. In W.J. Livesley, G. Dimaggio, & J.F. Clarkin (Eds.), Integrated treatment for personality disorder: A modular approach (pp. 305–324). Guilford Press. Cain, N.M., Pincus, A.L., & Grosse Holtforth, M. (2010). Interpersonal subtypes in social phobia: Diagnostic and treatment implications. Journal of Personality Assessment, 92, 514–527. Caligor, E., Kernberg, O.F., Clarkin, J.F., & Yeomans, F.E. (2018). Psychodynamic therapy for personality pathology: Treating self and interpersonal functioning. Washington, DC: American Psychiatric Association Publishing. Carson, R. C. (1969). Interaction concepts of personality. Routledge. Carver, C., & Scheier, M. (1998). On the self-regulation of behavior. New York: Cambridge University Press. Clegg, K. A., Moskowitz, D. S., Miners, C. T., Andrevski, G., Sadikaj, G., & Zuroff, D. C. 28 INTERPERSONAL THEORY (2021). Interpersonal perception and interpersonal spin. Journal of Personality, 89(3), 483-499. Corr, P. J., DeYoung, C. G., & McNaughton, N. (2013). Motivation and personality: A neuropsychological perspective. Social and Personality Psychology Compass, 7(3), 158- 175. Côté, S., Moskowitz, D.S., & Zuroff, D.C. (2012). Social relationships and intraindividual variability in interpersonal behavior: Correlates of interpersonal spin. Journal of Personality and Social Psychology, 102(3), 646–659. Critchfield, K.L., & Benjamin, L.S. (2008). Internalized representations of early interpersonal experience and adult relationships: A test of copy process theory in clinical and non- clinical settings. Psychiatry; Interpersonal and Biological Processes, 71(1), 71-92. Critchfield, K.L., & Benjamin, L.S. (2010). Assessment of repeated relational patterns for individual cases using the SASB-based Intrex questionnaire. Journal of Personality Assessment, 92(6), 480-489. Critchfield, K.L., Benjamin, L.S., & Levenick, K. (2015). Reliability, sensitivity, and specificity of case formulations for comorbid profiles in interpersonal reconstructive therapy: Addressing mechanisms of psychopathology. Journal of Personality Disorders, 29(4), 547-57. Cuthbert, B.N. (2014). The RDoC framework: facilitating transition from ICD/DSM to dimensional approaches that integrate neuroscience and psychopathology. World Psychiatry 13(1) 28-35 21). Interpersonal perception and interpersonal spin. Journal of Personality, 89(3), Corr, P. J., DeYoung, C. G., & McNaughton, N. (2013). Motivation and personality: A neuropsychological perspective. Social and Personality Psychology Compass, 7(3), 158- 175. Côté, S., Moskowitz, D.S., & Zuroff, D.C. (2012). Social relationships and intraindividual variability in interpersonal behavior: Correlates of interpersonal spin. Journal of Personality and Social Psychology, 102(3), 646–659. Critchfield, K.L., & Benjamin, L.S. (2008). Internalized representations of early interpersonal experience and adult relationships: A test of copy process theory in clinical and non- clinical settings. Psychiatry; Interpersonal and Biological Processes, 71(1), 71-92. Critchfield, K.L., & Benjamin, L.S. (2010). Assessment of repeated relational patterns for Critchfield, K.L., & Benjamin, L.S. (2008). Internalized representations of early interpersonal experience and adult relationships: A test of copy process theory in clinical and non- clinical settings. Psychiatry; Interpersonal and Biological Processes, 71(1), 71-92. Critchfield, K.L., & Benjamin, L.S. (2010). Assessment of repeated relational patterns for individual cases using the SASB-based Intrex questionnaire. Journal of Personality Assessment 92(6) 480 489 Critchfield, K.L., Benjamin, L.S., & Levenick, K. (2015). Reliability, sensitivity, and specificity of case formulations for comorbid profiles in interpersonal reconstructive therapy: Addressing mechanisms of psychopathology. Journal of Personality Disorders, 29(4), 547-57. Cuthbert, B.N. (2014). The RDoC framework: facilitating transition from ICD/DSM to dimensional approaches that integrate neuroscience and psychopathology. World Psychiatry, 13(1), 28-35. INTERPERSONAL THEORY Dalgleish, T., Black, M., Johnston, D., & Bevan, A. (2020). Transdiagnostic approaches to mental health problems: Current status and future directions. Journal of Consulting and Critchfield, K.L., Benjamin, L.S., & Levenick, K. (2015). Reliability, sensitivity, and specificity of case formulations for comorbid profiles in interpersonal reconstructive therapy: Addressing mechanisms of psychopathology. Journal of Personality Disorders, 29(4), 547-57. Cuthbert, B.N. (2014). The RDoC framework: facilitating transition from ICD/DSM to dimensional approaches that integrate neuroscience and psychopathology. World Psychiatry, 13(1), 28-35. Dalgleish, T., Black, M., Johnston, D., & Bevan, A. (2020). Transdiagnostic approaches to mental health problems: Current status and future directions. Journal of Consulting and INTERPERSONAL THEORY 29 Dawood, S., Dowgwillo, E.A., Wu, L.Z., & Pincus, A.L. (2018). Contemporary integrative interpersonal theory of personality. In V. Zeigler-Hill & T.K. Shackelford (Eds.), The SAGE handbook of personality and individual differences: The science of personality and individual differences (p. 171–202). Los Angeles: Sage. SAGE handbook of personality and individual differences: The science of personality and individual differences (p. 171–202). Los Angeles: Sage. Dawood, S., & Pincus, A.L. (2016). Multisurface interpersonal assessment in a cognitive- behavioral therapy context. Journal of personality assessment, 98(5), 449-460. Dawood, S., Thomas, K.M., Wright, A.G.C., & Hopwood, C.J. (2013). Heterogeneity of interpersonal problems among depressed young adults: Associations with substance abuse and pathological personality traits. Journal of Personality Assessment, 95, 513-522. D R A (2006) I l b h i d h f li N b h i l Depue, R.A. (2006). Interpersonal behavior and the structure of personality: Neurobehavioral foundation of agentic extraversion and affiliation. In T. Canli (Ed.), Biology of personality and individual differences (p. 60–92). Guilford Press. Depue, R.A., & Collins, P.F. (1999). Neurobiology of the structure of personality: Dopamine, facilitation of incentive motivation, and extraversion. Behavioral and brain sciences, 22(3), 491-517. Depue, R.A., & Lenzenweger, M.F. (2005). A Neurobehavioral Dimensional Model of Personality Disturbance. In M.F. Lenzenweger & J.F. Clarkin (Eds.), Major theories of personality disorder (p. 391–453). Guilford Press. Depue, R.A., & Morrone-Strupinsky, J.V. (2005). A neurobehavioral model of affiliative bonding: Implications for conceptualizing a human trait of affiliation. Behavioral and Brain Sciences, 28(3), 313-349. Depue, R. A., & Spoont, M. R. (1986). A behavioral dimension of constraint. Annals of the New Depue, R. A., & Spoont, M. R. (1986). A behavioral dimension of constraint. Annals of the New INTERPERSONAL THEORY 30 York Academy of Sciences, 487(1), 47-62. DeYoung, C.G., Weisberg, Y.J., Quilty, L.C., & Peterson, J.B. (2013). Unifying the aspects of the Big Five, the interpersonal circumplex, and trait affiliation. Journal of personality, 81(5), 465-475. Dowgwillo, E.A., Pincus, A.L., Newman, M.G., Wilson, S.J., Molenaar, P.C.M., & Levy, K.N. (2019). Two methods for operationalizing the interpersonal situation to investigate personality pathology and interpersonal perception in daily life. In L.I. Truslow & J.M. Rahmaan (Eds.), Personality disorders: What we know and future directions for research (pp. 31-106). Hauppauge, NY: Nova Science Publishers. Dryer, D.C., & Horowitz, L.M. (1997). When do opposites attract? Interpersonal complementarity versus similarity. Journal of Personality and Social Psychology, 72(3), 592–603. Du, T. V., Yardley, A. E., & Thomas, K. M. (2021). Mapping big five personality traits within and across domains of interpersonal functioning. Assessment, 28(5), 1358-1375. Du, T. V., Yardley, A. E., & Thomas, K. M. (2021). Mapping big five personality traits within and across domains of interpersonal functioning. Assessment, 28(5), 1358-1375. Erickson, T.M., Newman, M.G., & Pincus, A.L. (2009). Predicting unpredictability: Do measures of interpersonal rigidity/flexibility and distress predict intraindividual variability in social perceptions and behavior? Journal of Personality and Social Psychology, 97(5), 893–912. Erickson, T.M., Newman, M.G., & Pincus, A.L. (2009). Predicting unpredictability: Do measures of interpersonal rigidity/flexibility and distress predict intraindividual variability in social perceptions and behavior? Journal of Personality and Social Psychology, 97(5), 893–912. Eronen, M.I. (2021). The levels problem in psychopathology. Psychological Medicine, 51(6), 927-933. Fabrigar, L.R., Visser, P.S., & Browne, M.W. (1997). Conceptual and methodological issues in testing the circumplex structure of data in personality and social psychology. Personality and Social Psychology Review, 1(3), 184-203. 31 INTERPERSONAL THEORY Feldman, R. (2014). Synchrony and the neurobiological basis of social affiliation. In M. Mikulincer & P. R. Shaver (Eds.), The Herzliya series on personality and social psychology. Mechanisms of social connection: From brain to group (p. 145–166). Feldman, R. (2014). Synchrony and the neurobiological basis of social affiliation. In M. Mikulincer & P. R. Shaver (Eds.), The Herzliya series on personality and social psychology. Mechanisms of social connection: From brain to group (p. 145–166). American Psychological Association Florsheim, P., & McArthur, L. (2009). An interpersonal approach to attachment and change. In J. H. Obegi & E. Berent (Eds.), Attachment theory and research in clinical work with adults (pp. 379-409). New York: Guilford. American Psychological Association Florsheim, P., & McArthur, L. (2009). An interpersonal approach to attachment and change. In J. H. Obegi & E. Berent (Eds.), Attachment theory and research in clinical work with adults (pp. 379-409). New York: Guilford. Fournier, M.A., Moskowitz, D.S., & Zuroff, D.C. (2009). The interpersonal signature. Journal of Research in Personality, 43(2), 155-162. Fournier, M.A., Moskowitz, D.S., & Zuroff, D.C. (2009). The interpersonal signature. Journal of Research in Personality, 43(2), 155-162. Fournier, M.A., Moskowitz, D.S., & Zuroff, D.C. (2009). The interpersonal signature. Journal of Research in Personality, 43(2), 155-162. Fournier, M.A., Moskowitz, D.S., & Zuroff, D.C. (2011). Origins and applications of the interpersonal circumplex. In L.M. Horowitz & S. Strack (Eds.), Handbook of interpersonal psychology (pp. 57-74). Wiley. Fournier, M.A., Zuroff, D.C., & Moskowitz, D.S. (2007). The social competition theory of depression: Gaining from an evolutionary approach to losing. Journal of Social and Clinical Psychology, 26, 786-790. Freedman, M.B. (1985). Symposium: Interpersonal circumplex models (1948-1983). Journal of Personality assessment, 49(6), 622-625. Freedman, M.B., Leary, T.F., Ossorio, A.G., & Coffey, H.S. (1951). The Interpersonal Freedman, M.B., Leary, T.F., Ossorio, A.G., & Coffey, H.S. (1951). The Interpersonal Dimension of Personality 1. Journal of personality, 20(2), 143-161. Dimension of Personality 1. Journal of personality, 20(2), 143-161. Gifford, R. (1991). Mapping Nonverbal Behavior on the Interpersonal Circle. Journal of Personality and Social Psychology, 61(2), 279-288. Girard, J.M., & Wright, A.G.C. (2018). DARMA: Software for dual axis rating and media annotation. Behavior Research Methods, 50, 902-909. INTERPERSONAL THEORY 32 Girard, J.M., Wright, A.G.C., Beeney, J.E., Lazarus, S., Scott, L.N., Stepp, S.D., & Pilkonis, P.A. (2017). Interpersonal problems across levels of the psychopathology hierarchy. Comprehensive Psychiatry, 79, 53-69. Girard, J.M., Wright, A.G.C., Beeney, J.E., Lazarus, S., Scott, L.N., Stepp, S.D., & Pilkonis, P.A. (2017). Interpersonal problems across levels of the psychopathology hierarchy. Comprehensive Psychiatry, 79, 53-69. Gray, J.A. and McNaughton, N. 2000. The neuropsychology of anxiety: an enquiry into the functions of the septo-hippocampal system, 2nd edn. New York: Oxford University Press Grosse-Holtforth, M., Thomas, A., & Caspar, F. (2011). Interpersonal motivation. In L.M. Horowitz & S. Strack (Eds.), Handbook of interpersonal psychology: Theory, research, assessment, and therapeutic interventions (pp. 107–122). Hoboken, NJ: John Wiley & Sons. Guttman, L. (1954). A new approach to factor analysis: The radex. In P.F. Lazarsfeld (Ed.), Mathematical thinking in the social sciences (pp.258-348). New York: ColumbiaUniversity Press. Gurtman, M. B. (1991). Evaluating the interpersonalness of personality scales. Personality and Social Psychology Bulletin, 17(6), 670-677. Gurtman, M.B., & Pincus, A.L. (2003). The circumplex model: Methods and research applications. In J. Schinka & W. Velicer (Eds.), Handbook of psychology: Research methods in psychology (2nd ed., pp. 407–428). Wiley. Hames, J.L., Hagan, C.R., & Joiner, T.E. (2013). Interpersonal processes in depression. Annual Review of Clinical Psychology, 9, 355-377. Harvey, A.G., Watkins, E., Mansell, W., & Shafran R. (2004). Cognitive behavioral processes across psychological disorders: A transdiagnostic approach to research and treatment. Oxford University Press. Hayes, S.C., & Hofmann, S.G. (2018). Process-based CBT: The science and core clinical INTERPERSONAL THEORY 33 competencies of Cognitive Behavioral Therapy. New Harbinger Publications. Hogan, R.J., & Roberts, B.W. (2000). A socioanalytic perspective on person- environment. Person-environment psychology: New directions and perspectives. 2nd ed. M h h N J L E lb A i t 1 24 Mahwah, New Jersey: Lawrence Erlbaum Associates, 1-24. Hopwood, C.J. (2018). Interpersonal dynamics in personality and personality disorders. European Journal of Personality, 32(5), 499-524. Hopwood, C.J., Ansell, E.B., Pincus, A.L., Wright, A.G.C., Lukowitsky, M.R., & Roche, M.J. (2011). The circumplex structure of interpersonal sensitivities. Journal of Personality, 79, 707-740. Hopwood, C.J., Clarke, A.N., & Perez, M. (2007). Pathoplasticity of bulimic features and interpersonal problems. International Journal of Eating Disorders, 40, 652–658. Hopwood, C.J., & Good, E.W. (2019). Structure and correlates of interpersonal problems and sensitivities. Journal of personality, 87(4), 843-855. Hopwood, C.J., Harrison, A.L., Amole, M., Girard, J.M., Wright, A.G., Thomas, K.M., … & Kashy, D.A. (2020). Properties of the continuous assessment of interpersonal dynamics across sex, level of familiarity, and interpersonal conflict. Assessment, 27(1), 40-56. Hopwood, C.J., Pincus, A.L., & Wright, A.G.C. (2019). The interpersonal situation: Integrating clinical assessment, formulation, and intervention. In D. Samuel & D. Lynam (Eds.), Hopwood, C.J., Pincus, A.L., & Wright, A.G.C. (2019). The interpersonal situation: Integrating clinical assessment, formulation, and intervention. In D. Samuel & D. Lynam (Eds.), Using basic personality research to inform personality pathology (pp. 94-121). Oxford University Press. clinical assessment, formulation, and intervention. In D. Samuel & D. Lynam (Eds.), Using basic personality research to inform personality pathology (pp. 94-121). Oxford University Press. Hopwood, C.J., Wright, A.G.C., Ansell, E.B., & Pincus, A.L. (2013). The interpersonal core of personality pathology. Journal of Personality Disorders, 27(3), 271-295. Hopwood, C.J., Zimmerman, J., Pincus, A.L, & Krueger, R.F. (2015). Connecting personality INTERPERSONAL THEORY 34 structure and dynamics: Towards a more evidence-based and clinically useful diagnostic scheme. Journal of Personality Disorders, 29(4), 431-448. structure and dynamics: Towards a more evidence-based and clinically useful diagnostic scheme. Journal of Personality Disorders, 29(4), 431-448. Horowitz, L.M. (2004). Interpersonal foundations of psychopathology. Washington, DC: American Psychological Association Horowitz, L.M., & Strack, S. (2011). Handbook of interpersonal psychology: Theory, research, assessment, and therapeutic interventions. Hoboken, NJ: Wiley. Horowitz, L.M., & Wilson, K.R. (2005). Interpersonal motives and personality disorders. In S. Strack (Ed.), Handbook of Personology and Psychopathology (pp. 495–510). Hoboken, NJ: Wiley. Insel, T.R., Cuthbert, B.N., Garvey, M.A., Heinssen, R.K., & Pine, D.S. (2010). Research Domain Criteria (RDoC): Toward a New Classification Framework for Research on Mental Disorders. Am J Psychiatry, 167(7), 748-751. Johnson, S.L., Leedom, L.J., & Muhtadie, L. (2012). The dominance behavioral system and psychopathology: Evidence from self-report, observational, and biological studies. Psychological Bulletin, 138, 692-743. Kachin, K.E., Newman, M.G., & Pincus, A.L. (2001). An interpersonal problem approach to the INTERPERSONAL THEORY 35 division of social phobia subtypes. Behavior Therapy, 32, 479–501. Kaurin, A., Sequeira, S. L., Ladouceur, C., McKone, K. M., Rosen, D., Jones, N. P., ... & Silk, J. (in press). Modeling sensitivity to social threat in adolescent girls: A psychoneurometric approach. Journal of Abnormal Psychology. Kehl, M., Edershile, E.A., Hopwood, C.J., & Wright, A.G.C. (in press). A response surface analysis investigation of the effects of (mis)alignment between interpersonal values and efficacies on interpersonal problems. Journal of Personality. https://psyarxiv.com/ncpy3/ Kiesler, D.J. (1979). An interpersonal communication analysis of relationship in psychotherapy. Psychiatry, 42, 299-311. Kiesler, D. J. (1983). The 1982 interpersonal circle: A taxonomy for complementarity in human transactions. Psychological review, 90(3), 185. Kiesler, D.J. (1996). Contemporary interpersonal theory and research: Personality, psychopathology, and psychotherapy. John Wiley & Sons. Kiesler, D.J., Schmidt, J.A., & Wagner, C.C. (1997). A circumplex inventory of impact messages: An operational bridge between emotion and interpersonal behavior. In R. messages: An operational bridge between emotion and interpersonal behavior. In R. Plutchik & H.R. Conte (Eds.), Circumplex models of personality and emotions (p. 221– 244). American Psychological Association. Plutchik & H.R. Conte (Eds.), Circumplex models of personality and emotions (p. 221 244). American Psychological Association. 244). American Psychological Association. 244). American Psychological Association. Klahr, A.M., Thomas, K.M., Hopwood, C.J., Klump, K.L., & Burt, S.A. (2013). Evocative gene– environment correlation in the mother–child relationship: A twin study of interpersonal processes. Development and Psychopathology, 25, 105-118. Klemm W. (2015) Neurobiological Perspectives on Agency: Ten Axioms and Ten Propositions. In C. Gruber, M. Clark, S. Klempe, & J. Valsiner (Eds.), Constraints of agency: Annals of Theoretical Psychology—Volume 12 (pp. 51-88). Springer, Cham. f Theoretical Psychology—Volume 12 (pp. 51-88). Springer, Cham. INTERPERSONAL THEORY 36 Kotov, R., Krueger, R.F., Watson, D., Achenbach, T.M., Althoff, R.R., Bagby, R.M., … Zimmerman, M. (2017). The Hierarchical Taxonomy of Psychopathology (HiTOP): A dimensional alternative to traditional nosologies. Journal of Abnormal Psychology, 126(4), 454–477. v, R., Krueger, R.F., Watson, D., Achenbach, T.M., Althoff, R.R., Bagby, R.M., … Kotov, R., Krueger, R.F., Watson, D., Cicero, D.C., Conway, C.C., DeYoung, C.G., ... & Wright, A.G.C. (2021). The Hierarchical Taxonomy of Psychopathology (HiTOP): A quantitative nosology based on consensus of evidence. Annual Review of Clinical Psychology, 17. Advanced online publication. LaForge, R. (2004). The early development of the Interpersonal System of Personality (ISP). Multivariate Behavioral Research, 39(2), 359-378. LaForge, R., & Suczek, R. F. (1955). The Interpersonal Dimension of Personality: III. An Interpersonal Check List 1. Journal of Personality, 24(1), 94-112. Lakey, B. & Orehek, E. (2011). Relational regulation theory: A new approach to explain perceived social support’s link to mental health. Psychological Review, 118,482-495. Leary, T. (1957). Interpersonal diagnosis of personality: A functional theory and methodology for personality evaluation. Ronald Press. Leary, T., & Coffey, H.S. (1955). Interpersonal diagnosis: Some problems of methodology and validation. The Journal of Abnormal and Social Psychology, 50(1), 110-124. Leary, T., & Coffey, H.S. (1955). Interpersonal diagnosis: Some problems of methodology and lid i Th J l f Ab l d S l P h l 50(1) 110 124 y, , y, ( ) p g p gy validation. The Journal of Abnormal and Social Psychology, 50(1), 110-124. Leary, M. R., Tambor, E. S., Terdal, S. K., & Downs, D. L. (1995). Self-esteem as an interpersonal monitor: The sociometer hypothesis. Journal of personality and social psychology, 68(3), 518. Lenzenweger, M.F., & Depue, R.A. (2016). Toward a developmental psychopathology of personality disturbance: A neurobehavioral dimensional model incorporating genetic, Lenzenweger, M.F., & Depue, R.A. (2016). Toward a developmental psychopathology of personality disturbance: A neurobehavioral dimensional model incorporating genetic, INTERPERSONAL THEORY 37 ronmental, and epigenetic factors. Developmental psychopathology, 1-31. environmental, and epigenetic factors. Developmental psychopathology, 1-31. Lenzenweger, M.F., & Depue, R.A. (2020). Personality disturbances as emergent phenomena reflective of underlying neurobehavioral systems: Beyond dimensional measurement, phenotypic trait descriptors, and factor analysis. Psychopathology, 53, 213-220. Lewis, K.C., & Ridenour, J.M. (2020). The integration of EMA and single-occasion multimethod assessment data for a complex psychiatric patient. Assessment, 27, 1532-1546. I., Sadler, P., Woody, E., Ethier, N., & Malet, G. (2012). Capturing the stream of Lizdek, I., Sadler, P., Woody, E., Ethier, N., & Malet, G. (2012). Capturing the stream of behavior: A computer-joystick method for coding interpersonal behavior continuously over time. Social Science Computer Review, 30, 513-521. Lizdek, I., Woody, E., Sadler, P., & Rehman, U. S. (2016). How do depressive symptoms in husbands and wives relate to the interpersonal dynamics of marital interactions?. Journal of Counseling Psychology, 63, 721-735. Liu, H., Xie, Q.W., & Lou, V.W. (2019). Everyday social interactions and intra-individual variability in affect: A systematic review and meta-analysis of ecological momentary assessment studies Motivation and Emotion 43(2) 339 353 assessment studies. Motivation and Emotion, 43(2), 339-353. Locke, K.D. (2011). Circumplex Measures of Interpersonal Constructs. In Horowitz, L.M., & Strack, S. Handbook of interpersonal psychology (pp. 313-324). Wiley. Locke, K.D. (2015). Agentic and communal social motives. Social and Personality Psychology Compass, 9, 525-538. Locke, K.D., & Sadler, P. (2007). Self-efficacy, values, and complementarity in dyadic interactions: Integrating interpersonal and social-cognitive theory. Personality and Social Psychology Bulletin, 33, 94–109. Lorr, M., & McNair, D. M. (1963). An interpersonal behavior circle. The Journal of Abnormal Lorr, M., & McNair, D. M. (1963). An interpersonal behavior circle. The Journal of Abnormal INTERPERSONAL THEORY 38 Lukowitsky, M.R., & Pincus, A.L. (2011). The pantheoretical nature of mental representations and their ability to predict interpersonal adjustment in a nonclinical and their ability to predict interpersonal adjustment in a nonclinical and their ability to predict interpersonal adjustment in a nonclinical sample. Psychoanalytic Psychology, 28(1), 48–74. Luo, X., Nuttall, A., Locke, K., & Hopwood, C. (2018). Dynamic Longitudinal Relations Between Binge Eating Symptoms and Severity and Style of Interpersonal Problems. Journal of Abnormal Psychology, 127, 30-42. Luyten, P., & Blatt, S.J. (2011). Integrating theory-driven and empirically-derived models of Luyten, P., & Blatt, S.J. (2011). Integrating theory-driven and empirically-derived models of personality development and psychopathology: A proposal for DSM V. Clinical Psychology Review, 31, 52-68. Mahadevan, N., Gregg, A.P., & Sedikides, C. (2019). Is self-regard a sociometer or a hierometer?: Self-esteem tracks status and inclusion, narcissism tracks status. Journal of Personality and Social Psychology, 116(3), 444-466. hierometer?: Self-esteem tracks status and inclusion, narcissism tracks status. Journal of Personality and Social Psychology, 116(3), 444-466. Personality and Social Psychology, 116(3), 444-466. Mahadevan, N., Gregg, A.P., & Sedikides, C. (2020). The ups and downs of social life: Within‐ person variations in daily status and inclusion differentially predict self‐regard and i t l b h i J l f P li 88(6) 1111 1128 interpersonal behavior. Journal of Personality, 88(6), 1111-1128. Markey, P.M., Funder, D.C., & Ozer, D.J. (2003). Complementarity of interpersonal behaviors in dyadic interactions. Personality and Social Psychology Bulletin, 29(9), 1082-1090. dyadic interactions. Personality and Social Psychology Bulletin, 29(9), 1082-1090. McCabe, K. O., & Fleeson, W. (2012). What is extraversion for? Integrating trait and motivational perspectives and identifying the purpose of extraversion. Psychological science, 23(12), 1498-1505. McCrae, R., & Costa, P. (1989). The Structure of Interpersonal Traits: Wiggins's Circumplex and the Five-Factor Model. Journal of Personality and Social Psychology, 56(4), 586-595. INTERPERSONAL THEORY Nilsen, E.S., Lizdek, I., & Ethier, N. (2015). Mother-child interpersonal dynamics: The influence INTERPERSONAL THEORY 39 McNair, D.M. (1963). An interpersonal behavior circle. The Journal of Abnormal and Social Psychology, 67(1), 68-75. Meisel, S.N., Colder, C.R., & Hopwood, C.J. (under review). Using the Continuous Assessment of Interpersonal Dynamics to characterize adolescent substance use discussions. Meisel, S.N., Colder, C.R., & Hopwood, C.J. (under review). Using the Continuous Assessment of Interpersonal Dynamics to characterize adolescent substance use discussions. Monteleone, A.M., Treasure, J., Kan, C., & Cardi, V. (2018). Reactivity to interpersonal stress in patients with eating disorders: A systematic review and meta-analysis of studies using an experimental paradigm. Neuroscience & Biobehavioral Reviews, 87, 133-150. Moskowitz, D.S. (1994). Cross-situational generality and the interpersonal circumplex. Journal of Personality and Social Psychology, 66, 921–933. Moskowitz, D.S. (2005). Unfolding interpersonal behavior. Journal of Personality, 73, 1607– 1632. Moskowitz, D.S. (2009). Coming full circle: Conceptualizing the study of interpersonal behaviour. Canadian Psychology, 50(1), 33–41. Moskowitz, D.S., & Zuroff, D.C. (2004). Flux, pulse, and spin: Dynamic additions to the personality lexicon. Journal of Personality and Social Psychology, 86(6), 880–893. Moskowitz, D. S., & Zuroff, D. C. (2005). Assessing interpersonal perceptions using the interpersonal grid. Psychological assessment, 17(2), 218. interpersonal grid. Psychological assessment, 17(2), 218. Moskowitz, D. S., Zuroff, D. C., aan het Rot, M., & Young, S. N. (2011). Tryptophan and interpersonal spin. Journal of Research in Personality, 45(6), 692-696. Moskowitz, D. S., Zuroff, D. C., aan het Rot, M., & Young, S. N. (2011). Tryptophan and interpersonal spin. Journal of Research in Personality, 45(6), 692-696. Moskowitz, D. S., Ho, M. H. R., & Turcotte-Tremblay, A. M. (2007). Contextual influences on interpersonal complementarity. Personality and Social Psychology Bulletin, 33(8), 1051- 1063. Nilsen, E.S., Lizdek, I., & Ethier, N. (2015). Mother-child interpersonal dynamics: The influence 40 INTERPERSONAL THEORY of maternal and child ADHD symptoms. Journal of Experimental Psychopathology, 6, 313-329. Palumbo, I., Perkins, E., Yancey, J., Brislin, S., Patrick, C., & Latzman, R. (2020). Toward a multimodal measurement model for the neurobehavioral trait of affiliative capacity. Palumbo, I., Perkins, E., Yancey, J., Brislin, S., Patrick, C., & Latzman, R. (2020). Toward a multimodal measurement model for the neurobehavioral trait of affiliative capacity. Personality Neuroscience, 3, E11. doi:10.1017/pen.2020.9 Pickering, A.D. and Gray, J.A. 1999. The neuroscience of personality, in L.A. Pervin and O.P. John (eds.), Handbook of personality: theory and research, 2nd edn, pp. 277–99. New York: Guilford Press Pincus, A.L. (2002). Constellations of dependency within the five-factor model of personality. In P.T. Costa, Jr. & T.A. Widiger (Eds.), Personality disorders and the five-factor model of personality (p. 203–214). American Psychological Association. Pincus, A.L. (2002). Constellations of dependency within the five-factor model of personality. In P.T. Costa, Jr. & T.A. Widiger (Eds.), Personality disorders and the five-factor model of personality (p. 203–214). American Psychological Association. Pincus, A.L. (2005). A contemporary integrative interpersonal theory of personality disorders. In M.F. Lenzenweger & J.F. Clarkin (Eds.), Major theories of personality disorder (p. 282– 331). Guilford Press. Pincus, A.L., & Ansell, E.B. (2013). Interpersonal theory of personality. In J. Suls & H. Tennen (Eds.), Handbook of psychology Vol. 5: Personality and social psychology (2nd ed., pp. 141–159). Wiley. Pincus, A.L., Cain, N.M., & Halberstadt, A.L. (2020b). Importance of self and other in defining personality pathology. Psychopathology, 53, 133-140. Pincus, A.L., & Hopwood, C.F. (2012). A contemporary interpersonal model of personality pathology and personality disorder. In T.A. Widiger (Ed.), The Oxford handbook of personality disorders (p. 372–398). Oxford University Press. Pincus, A.L., Hopwood, C.J., & Wright, A.G.C. (2020a). The interpersonal situation: An Pincus, A.L., Hopwood, C.J., & Wright, A.G.C. (2020a). The interpersonal situation: An INTERPERSONAL THEORY 41 integrative framework for the study of personality, psychopathology, and psychotherapy. In D. Funder, J.F. Rauthmann, & R. Sherman (Eds), Oxford handbook of psychological situations (pp. 124-142). Oxford. Pincus, A.L., Lukowitsky, M.R., Wright, A.G.C., & Eichler, W.C. (2009). The interpersonal nexus of persons, situations, and psychopathology. Journal of Research in Personality, 43, 264-265. Pincus, A.L., Sadler, P., Woody, E., Roche, M. J., Thomas, K.M., & Wright, A.G.C. (2014). Multimethod assessment of interpersonal dynamics. In C.J. Hopwood & R.F. Bornstein (Eds.), Multimethod clinical assessment (pp. 52-91). Guilford Press. Pincus, A.L., & Wiggins, J.S. (1990). Interpersonal problems and conceptions of personality disorders. Journal of personality disorders, 4(4), 342-352. Pincus, A.L. & Wright, A.G.C. (2011). Interpersonal diagnosis of psychopathology. In L. M. Horowitz and S. Strack (Eds.) Handbook of interpersonal psychology (pp. 359-381). Wiley. Przeworski, A., Newman, M. G., Pincus, A.L., Kasoff, M.B., Yamasaki, A.S., Castonguay, L. G., & Berlin, K.S. (2011). Interpersonal pathoplasticity in individuals with generalized anxiety disorder. Journal of Abnormal Psychology, 120, 286–298. Rappaport, L.M., Moskowitz, D.S., & D’Antono, B. (2014). Naturalistic interpersonal behavior patterns differentiate depression and anxiety symptoms in the community. Journal of Counseling Psychology, 61, 253-263. Reis, H., Collins, W., & Berscheid, E. (2000). The Relationship Context of Human Behavior and Development. Psychological Bulletin, 126, 844-872. Ringwald, W.R., Dombrovski, A.Y., Hallquist, M.N., & Wright, A.G.C. (2020a). INTERPERSONAL THEORY 42 Transdiagnostic predictors of interpersonal and affective variability in borderline personality pathology. Manuscript submitted for publication. https://psyarxiv.com/sh7cz/ Ringwald, W.R., Hopwood, C.J., Pilkonis, P.A., & Wright, A.G.C. (2020b). Dynamic features of affect and interpersonal behavior in relation to general and specific personality pathology. Personality Disorder: Theory, Research, and Treatment. https://osf.io/4qevt/ Roberts, B.W., & DelVecchio, W.F. (2000). The rank-order consistency of personality traits from childhood to old age: A quantitative review of longitudinal studies. Psychological Bulletin, 126(1), 3–25. Transdiagnostic predictors of interpersonal and affective variability in borderline personality pathology. Manuscript submitted for publication. https://psyarxiv.com/sh7cz/ Ringwald, W.R., Hopwood, C.J., Pilkonis, P.A., & Wright, A.G.C. (2020b). Dynamic features of affect and interpersonal behavior in relation to general and specific personality pathology. Personality Disorder: Theory, Research, and Treatment. https://osf.io/4qevt/ Roberts, B.W., & DelVecchio, W.F. (2000). The rank-order consistency of personality traits from childhood to old age: A quantitative review of longitudinal studies. Psychological Bulletin, 126(1), 3–25. Roberts, B.W., Walton, K.E., & Viechtbauer, W. (2006). Patterns of mean-level change in personality traits across the life course: A meta-analysis of longitudinal studies. Psychological Bulletin, 132(1), 1–25. Roche, M.J., Pincus, A.L., Conroy, D.E., Hyde, A.L., & Ram, N. (2013). Pathological narcissism and interpersonal behavior in daily life. Personality Disorders: Theory, Research, and Treatment, 4(4), 315–323. Roche, M.J., Pincus, A.L., Hyde, A.L., Conroy, D.E., & Ram, N. (2013). Within-person covariation of agentic and communal perceptions: Implications for interpersonal theory and assessment. Journal of Research in Personality, 47(4), 445-452. Ross, J.M., Girard, J.M., Wright, A.G.C., Beeney, J.E., Scott, L.N., Hallquist, M.N., … & Pilkonis, P.A. (2017). Momentary patterns of covariation between specific affects and interpersonal behavior: Linking relationship science and personality assessment. Psychological Assessment, 29(2), 123-134. Russell, J.J., Moskowitz, D.S., Zuroff, D.C., Sookman, D., & Paris, J. (2007). Stability and variability of affective experience and interpersonal behavior in borderline personality INTERPERSONAL THEORY 43 Sadikaj, G., Moskowitz, D.S., Russell, J.J., & Zuroff, D.C. (2015). Submissiveness in social anxiety disorder: The role of interpersonal perception and embarrassment. Journal of Social and Clinical Psychology, 34, 1-27. Sadikaj, G., Moskowitz, D.S., Russell, J.J., Zuroff, D.C., & Paris, J. (2013). Quarrelsome behavior in borderline personality disorder: Influence of behavioral and affective reactivity to perceptions of others. Journal of Abnormal Psychology, 122, 195-207. behavior in borderline personality disorder: Influence of behavioral and affective reactivity to perceptions of others. Journal of Abnormal Psychology, 122, 195-207. Sadikaj, G., Moskowitz, D.S., & Zuroff, D.C. (2011). Attachment-related affective dynamics: Differential reactivity to others’ interpersonal behavior. Journal of Personality and Social Psychology, 100, 905–917. Sadler, P., Ethier, N., Gunn, G.R., Duong, D., & Woody, E. (2009). Are we on the same wavelength? Interpersonal complementarity as shared cyclical patterns during interactions. Journal of Personality and Social Psychology, 97, 1005-1020. S. Strack (Eds.), Handbook of interpersonal psychology (pp. 123-142). Wiley. Sadler, P., Howard, A.P., Lizdek, I., & Woody, E. (2020). The Interpersonal Transaction Scales– 8 (ITS-8): A circumplex-based, behaviorally anchored instrument based on the CLOIT-R. Assessment, 27, 675-687. Sadler, P., Woody, E., McDonald, K., Lizdek, I., & Little, J. (2015). A lot can happen in a few minutes: Examining dynamic patterns within an interaction to illuminate the interpersonal nature of personality disorders. Journal of Personality Disorders, 29, 526-546. Safran, J.D. (1990a). Towards a refinement in cognitive therapy in light of interpersonal theory: I. Theory. Clinical Psychology Review, 10, 87-105. INTERPERSONAL THEORY 44 Safran. J.D. (1990b). Towards a refinement of cognitive therapy in light of interpersonal theory: II. Practice. Clinical Psychology Review, 10, 107-121. Salzer, S., Pincus, A.L., Hoyer, J., Kreische, R., Leichsenring, F., & Leibling, E. (2008).Interpersonal subtypes within generalized anxiety disorder. Journal of Personality Assessment, 90, 292–299. Salzer, S., Pincus, A.L., Winkelbach, C., Leichsenring, F., & Leibing, E. (2011). Interpersonal subtypes and change of interpersonal problems in the treatment of patients with generalized anxiety disorder: A pilot study. Psychotherapy: Theory, Research, Practice, & Training, 48, 304-310. Schmidt, J., Wagner, C., & Kiesler, D. (1999). Psychometric and Circumplex Properties of the Octant Scale Impact Message Inventory (IMI-C): A Structural Evaluation. Journal of Counseling Psychology, 46(3), 325-334. Shoda, Y., Mischel, W., & Wright, J.C. (1994). Intraindividual stability in the organization and patterning of behavior: Incorporating psychological situations into the idiographic analysis of personality. Journal of Personality and Social Psychology, 67, 674–687. Simon, S., Cain, N.M., Samstag, L.W., Meehan, K.B., & Muran, J.C. (2015). Assessing interpersonal subtypes in depression. Journal of Personality Assessment, 97, 364-373. Simpson, J.A., Griskevicius, V., & Kim J.S. (2011). Evolution, life history theory, and personality. In L.M. Horowitz & S. Strack (Eds.), Handbook of interpersonal psychology: Theory, research, assessment, and therapeutic interventions (pp. 75-89). Hoboken, NJ: Wiley. Shoda, Y., Mischel, W., & Wright, J.C. (1994). Intraindividual stability in the organization and patterning of behavior: Incorporating psychological situations into the idiographic analysis of personality. Journal of Personality and Social Psychology, 67, 674–687. Simon, S., Cain, N.M., Samstag, L.W., Meehan, K.B., & Muran, J.C. (2015). Assessing interpersonal subtypes in depression. Journal of Personality Assessment, 97, 364-373. Simpson, J.A., Griskevicius, V., & Kim J.S. (2011). Evolution, life history theory, and personality. In L.M. Horowitz & S. Strack (Eds.), Handbook of interpersonal psychology: Theory, research, assessment, and therapeutic interventions (pp. 75-89). Hoboken, NJ: Wiley. Slaney, R.B., Pincus, A.L., Uliaszek, A.A., & Wang, K.T. (2006). Conceptions of perfectionism and interpersonal problems: Evaluating groups using the structural summary method for Slaney, R.B., Pincus, A.L., Uliaszek, A.A., & Wang, K.T. (2006). Conceptions of perfectionism and interpersonal problems: Evaluating groups using the structural summary method for 45 INTERPERSONAL THEORY circumplex data. Assessment, 13, 138–153. p , ( ) y p g implications for human psychopathology. Psychological bulletin, 112(2), 330. Sullivan, H. S. (1948). The meaning of anxiety in psychiatry and in life. Psychiatry, 11(1), 1-13. Sullivan, H.S. (1953a). Conceptions of modern psychiatry. Norton. Sullivan, H.S. (1953b). The interpersonal theory of psychiatry. Norton. Sullivan, H.S. (1954). The psychiatric interview. Norton. Tchalova, K., Sadikaj, G., Moskowitz, D.S., Zuroff, D.C., & Bartz, J.A. (2019). et al. Variation in the μ-opioid receptor gene (OPRM1) and experiences of felt security in response to a romantic partner’s quarrelsome behavior. Molecular Psychiatry. Advanced online publication. https://doi.org/10.1038/s41380-019-0600-4 Tchalova, K., Sadikaj, G., Moskowitz, D.S., Zuroff, D.C., & Bartz, J.A. (2019). et al. Variation in the μ-opioid receptor gene (OPRM1) and experiences of felt security in response to a romantic partner’s quarrelsome behavior. Molecular Psychiatry. Advanced online publication. https://doi.org/10.1038/s41380-019-0600-4 Thomas, K.M., Hopwood, C.J., Donnellan, M.B., Wright, A.G.C., Sanislow, C.A., McDevitt- Murphy, M.E.,… & Morey, L.C. (2014). Personality heterogeneity in PTSD: Distinct temperament and interpersonal typologies. Psychological Assessment, 26, 23–34 Tracey, T.J. (1993). An interpersonal stage model of the therapeutic process. Journal of Counseling Psychology, 40(4), 396–409. Tracey, T.J. (1994). An examination of the complementarity of interpersonal behavior. Journal of Personality and Social Psychology, 67(5), 864–878. Tracey, T.J. (2004). Levels of interpersonal complementarity: A simplex representation. Personality and social psychology bulletin, 30(9), 1211-1225. Tracey, T.J.G., Bludworth, J., & Glidden-Tracey, C.E. (2012). Are there parallel processes in Tracey, T.J.G., Bludworth, J., & Glidden-Tracey, C.E. (2012). Are there parallel processes in psychotherapy supervision? An empirical examination. Psychotherapy, 49(3), 330–343. T ll P D & P lh D L (2012) A ti d l l Th i d Trapnell, P.D., & Paulhus, D. L. (2012). Agentic and communal values: Their scope and INTERPERSONAL THEORY 46 measurement. Journal of Personality Assessment, 94, 39-52. Treasure, J., Willmott, D., Ambwani, S., Cardi, V., Clark Bryan, D., Rowlands, K., & Schmidt, perpetuating factors that contribute to the development of the severe and enduring illness. Journal of Clinical Medicine, 9, 630. Trucco, E.M., Wright, A.G.C., & Colder, C.R. (2014). Stability and change of social goals in adolescence. Journal of Personality, 82(5), 379-389. Vazire, S. (2010). Who knows what about a person? The self-other knowledge asymmetry (SOKA) model. Journal of Personality and Social Psychology, 98(2), 281-300. Venables, C.N, & Patrick, C.J. (2020). In Wright, A.G.C., & Hallquist, M.N. (Eds.). Building models of psychopathology spanning multiple modalities of measurement. Cambridge handbook of research methods in clinical psychology (pp. 18-32). New York, NY: Cambridge University Press. Wagner, C.C., Kiesler, D.J., & Schmidt, J. A. (1995). Assessing the interpersonal transaction cycle: Convergence of action and reaction interpersonal circumplex measures. Journal of Personality and Social Psychology, 69(5), 938–949. Waller, R., & Wagner, N. (2019). The Sensitivity to Threat and Affiliative Reward (STAR) Waller, R., & Wagner, N. (2019). The Sensitivity to Threat and Affiliative Reward (STAR) model and the development of callous-unemotional traits. Neuroscience & Biobehavioral Reviews, 107, 656-671. Wang, S., Roche, M.J., Pincus, A.L., Conroy, D.E., Rebar, A.L., & Ram, N. (2014). Interpersonal dependency and emotion in everyday life. Journal of Research in Personality, 53, 5-12. Widiger, T.A., & Smith, G.T. (2008). Personality and psychopathology. In O.P. John, R. Robins, INTERPERSONAL THEORY 47 & L.A. Pervin (Eds.), Handbook of personality: Theory and research (3rd ed., pp. 743– 769). New York, NY: Guilford Press. & L.A. Pervin (Eds.), Handbook of personality: Theory and research (3rd ed., pp. 743– 769). New York, NY: Guilford Press. Wiggins, J.S. (1979). A psychological taxonomy of trait descriptive terms: The interpersonal domain. Journal of Personality and Social Psychology, 37, 395–412. Wiggins, J.S. (1991). Agency and communion as conceptual coordinates for the understanding and measurement of interpersonal behavior. In D. Cicchetti & W.M. Grove (Eds.), Thinking clearly about psychology: Essays in honor of Paul E. Meehl, Vol. 2: Personality and psychopathology (pp. 89-113). University of Minnesota Press. Wiggins, J.S. (1997a). Circumnavigating Dodge Morgan’s interpersonal style. Journal of Personality, 65, 1069–1086. Wiggins, J.S. (1997b). In defense of traits. In R. Hogan, J. A. Johnson, & S.R. Briggs (Eds.), Handbook of personality psychology (pp. 95–141). San Diego, CA: Academic Press. Wiggins, J.S. (2003). Paradigms of personality assessment. New York, NY: Guilford Press. Wiggins, J.S., & Broughton, R. (1991). A geometric taxonomy of personality scales. European Journal of Personality, 5, 343-365. Wiggins, J.S. & Trapnell, P.D. (1996). A dyadic-interactional perspective on the five-factor model. In J.S. Wiggins (Ed.), The five-factor model of personality: Theoretical perspectives (pp. 88-162). Guilford. Williams, T., & Simms, L. (2016). Personality disorder models and their coverage of interpersonal problems. Personality Disorders: Theory, Research, and Treatment, 7(1), 15-27. Wilson, S., Stroud, C.B., & Durbin, C.E. (2017). Interpersonal dysfunction in personality disorders: A meta-analytic review. Psychological Bulletin, 143(7), 677-734. INTERPERSONAL THEORY 48 Woods, W.C., & Wright, A.G.C. (2019). Approaches to modeling interpersonal complementarity in intensive longitudinal data. https://doi.org/10.31234/osf.io/5xj7k ht, A.G.C., Creswell, K.G., Flory, J.D., Muldoon, M.F., & Manuck, S.B. (2019). Wright, A.G.C., & Kaurin, A. (2020). Integrating structure and function in conceptualizing and assessing pathological traits. Psychopathology. Wright, A.G.C., Pincus, A. L., Hopwood, C. J., Thomas, K. M., Markon, K. E., & Krueger, R. F. (2012). An interpersonal analysis of pathological personality traits in DSM- (2012). An interpersonal analysis of pathological personality traits in DSM- 5. Assessment, 19(3), 263-275. (2012). An interpersonal analysis of pathological personality traits in DSM 5. Assessment, 19(3), 263-275. Wright, A. G. C., Pincus, A. L., Conroy, D. E., & Elliot, A. J. (2009a). The pathoplastic relationship between interpersonal problems and fear of failure. Journal of Personality, 77, 997–1024. Wright, A.G.C., Pincus, A.L., & Lenzenweger, M.F. (2012). Interpersonal development, Wright, A.G.C., Pincus, A.L., & Lenzenweger, M.F. (2012). Interpersonal development, stability, and change in young adulthood. Journal of Personality, 80(5), 1339-1372. Wright, A.G.C., Stepp, S.D., Scott, L.N., Hallquist, M.N., Beeney, J.E., Lazarus, S.A., & Pilkonis, P.A. (2017). The effect of pathological narcissism on interpersonal and affective processes in social interactions. Journal of Abnormal Psychology, 126(7), 898–910. Wright, A.G.C., Thomas, K.M., Hopwood, C.J., Markon, K.E., Pincus, A.L. & Krueger, R.F. (2012). The hierarchical structure of DSM-5 pathological personality traits. Journal of 1. The focus: Fundamentally important functional expressions of personality and psychopathology typically occur in interpersonal situations. 1. The focus: Fundamentally important functional expressions of personality and psychopathology typically occur in interpersonal situations. Basic needs for belongingness (Baumeister & Leary, 1995), the ability to negotiate social dominance hierarchies (Johnson et al., 2012), and the relational context of personality development itself (Reis et al., 2000) all speak to the importance of interpersonal functioning for personality, psychopathology, and well-being. Though virtually all theories of personality and psychopathology touch upon interpersonal functioning, in Contemporary Integrative Interpersonal Theory (CIIT; Pincus, 2005) it is a central proposition—integrating personality structures and dynamic processes in relation to interpersonal functioning is of fundamental importance for understanding and investigating personality, psychopathology, and their substrates. The integrative nature of CIIT can accommodate findings from several theoretical and research traditions that bear on personality and relational functioning (Horowitz, 2004; Horowitz & Strack, 2011; Pincus & Ansell, 2013). As Horowitz and colleagues stated, “Because the interpersonal approach harmonizes so well with all of these theoretical approaches, it is integrative: It draws from the wisdom of all major approaches to systematize our understanding of interpersonal phenomena. Although it is integrative, however, it is also unique, posing characteristic questions of its own” (Horowitz et al., 2006, p. 82). Interpersonal models have been integrated conceptually and empirically with attachment (e.g., Benjamin, 1993; Bartholomew & Horowitz, 1991; Florsheim & McArthur, 2009) and evolutionary (Fournier et al., 2007; Hoyenga et al., 1998; Simpson et al., 2011; Zuroff et al., 1999) theories of personality and development, recognizing interpersonal relations as necessary for individual and species survival. CIIT also integrates social-cognitive (Locke & Sadler, 2007; Safran, 1990a, 1990b; Shoda et al., 1994) and psychodynamic (Caligor et al., 2018; Lukowitsky & Pincus, 2011; Luyten & Blatt, 2011) personality process models, by proposing that mediating internal psychological features (e.g., self–other schemas, motives and needs embedded in these schemas, and emerging emotional experiences) influence interpersonal patterns (Pincus, 2005). Finally, interpersonal functioning is also rooted in basic neurobiological systems underlying dominance/agency and communion/affiliation (Depue, 2006; Feldman, 2014; Klemm, 2015; Lenzenweger & Depue, 2020; Palumbo et al., 2020) and interpersonal functioning is impacted by pharmacological changes in brain chemistry (e.g., Moskowitz et al., 2011; Wright et al., 2019) and genetic variation (e.g., Tchalova et al., 2019). The breadth and scope of CIIT promote the “interpersonal situation” (Hopwood et al., 2019; Pincus et al, 2020a; Sullivan, 1953a, 1953b) as a uniquely valuable unit of analysis for studying psychological phenomena at multiple levels. INTERPERSONAL THEORY 49 Abnormal Psychology, 121(4), 951-957. Wright, A. G. C., & Zimmermann, J. (2019). Applied ambulatory assessment: Integrating idiographic and nomothetic principles of measurement. Psychological Assessment, 31(12), 1467–1480. Zilcha-Mano, S., McCarthy, K. S., Dinger, U., Chambless, D.L., Milrod, B.L., Kunik, L., & Barber, J. P. (2015). Are there subtypes of panic disorder? An interpersonal perspective. Journal of Consulting and Clinical Psychology 83 938 950 Journal of Consulting and Clinical Psychology, 83, 938-950. Zimmermann, J. & Wright, A.G.C. (2017). Beyond description in interpersonal construct validation: Methodological advances in the circumplex structural summary approach. Assessment, 24(1), 3-23. Zuroff, D. C., Moskowitz, D. S., & Côté, S. (1999). Dependency, self-criticism, interpersonal behavior, and affect: Evolutionary perspectives. British Journal of Clinical Psychology, 38, 231-250. INTERPERSONAL THEORY 50 Figure 1. Diagram of Contemporary Integrative Interpersonal Theory Features Figure 1. Diagram of Contemporary Integrative Interpersonal Theory Features Figure 1. Diagram of Contemporary Integrative Interpersonal Theory Features Figure 1. Diagram of Contemporary Integrative Interpersonal Theory Features 51 INTERPERSONAL THEORY This is a supplement to elaborate the fundamental tenets of Contemporary Integrative Interpersonal Theory, as part of the manuscript, “Contemporary Integrative Interpersonal Theory: Integrating Structure, Dynamics, Temporal Scale, and Levels of Analysis” authored by Aidan G.C. Wright, Aaron L. Pincus, and Christopher J. Hopwood. 2. The structure: Interpersonal functioning can be organized using the dimensions of agency and communion. Although the Kaiser-Berkeley group were not directly influenced by his work, their early efforts coincided with Guttman’s (1954) articulation of a circumplex model, which was quickly enlisted to validate measures of the interpersonal circle (e.g., Lorr & McNair, 1963). Although the Kaiser-Berkeley group were not directly influenced by his work, their early efforts coincided with Guttman’s (1954) articulation of a circumplex model, which was quickly enlisted to validate measures of the interpersonal circle (e.g., Lorr & McNair, 1963). Conceptually, a circumplex is a formal quantitative model of a circular array of variables whose associations are non-hierarchical, or not simply a matter of degree but rather of kind (Fabrigar et al., 1997; Guttman, 1954). Formally, a circumplex is represented in a circulant correlation matrix, which is defined by a pattern of decreasing associations moving away from the main diagonal. These associations can also be decomposed into two dimensions to summarize the pattern of associations. Unlike typical simple structure factor models, in a circumplex structure, it is expected that the entire two-dimensional space is populated by relevant content. That is, theoretically all combinations of the two dimensions manifest in relevant patters of functioning. The original interpersonal circle, then formalized mathematically as the interpersonal circumplex (IPC), has provided CIIT with a firm structural basis and nomological net for tests of theory, measure development, and external construct evaluation (Gurtman, 1991; Gurtman & Pincus, 2003; Wiggins & Broughton, 1991; Zimmermann & Wright, 2017). ; gg g , ; g , ) The IPC has been established in dispositional models and measures of an evolving set of interpersonal constructs including general traits, problems, impacts, strengths, sensitivities, efficacies, influence tactics, and values (Bliton & Pincus, 2020; Fournier et al., 2011; Hopwood et al., 2011; Locke, 2011; Wiggins, 1979). Agentic and communal dispositions imply enduring patterns of perceiving, thinking, feeling, and behaving that are probabilistic in nature, and describe an individual’s interpersonal tendencies aggregated across time, place, and relationships. An individual’s interpersonal dispositions, when understood in relation to their motives and goals, are assumed to give rise to variations in adaptive and maladaptive behavior that are contextualized yet consistent across relevant interpersonal situations (Horowitz & Wilson, 2005; Wiggins, 1997b). Thus, at this level IPC models can describe a person’s typical ways of relating to others and refer to aspects of their interpersonal style. 1. The focus: Fundamentally important functional expressions of personality and psychopathology typically occur in interpersonal situations. The interpersonal situation is the dynamically unfolding context in which social-learning takes place across the lifespan, promoting personality organization, development, and adjustment through the continuous patterning and re-patterning of interpersonal experience. Although interpersonal dysfunction is central to conceptualization of personality pathology (e.g., Pincus et al., 2020b), the structural (i.e., interpersonal dispositions) and process (i.e., patterns of interpersonal perception, affect, and behavior) aspects of CIIT have been empirically investigated across the spectra of psychopathology. INTERPERSONAL THEORY 2. The structure: Interpersonal functioning can be organized using the dimensions of agency and communion. 2. The structure: Interpersonal functioning can be organized using the dimensions of agency and communion Early efforts to outline a comprehensive interpersonal model of personality and psychopathology amenable to systematic scientific inquiry recognized the need for a valid organizational scheme, or structure (Freedman et al., 1951; Leary & Coffey, 1955). Deeply influenced by the relational psychoanalytic theories of Sullivan (1953a), among others, the Kaiser-Berkley group set about establishing a structural model that could account for the patterns of behaviors observed in psychotherapy groups, eventually arriving at a circular array Early efforts to outline a comprehensive interpersonal model of personality and psychopathology amenable to systematic scientific inquiry recognized the need for a valid organizational scheme, or structure (Freedman et al., 1951; Leary & Coffey, 1955). Deeply influenced by the relational psychoanalytic theories of Sullivan (1953a), among others, the Kaiser-Berkley group set about establishing a structural model that could account for the patterns of behaviors observed in psychotherapy groups, eventually arriving at a circular array (Freedman, 1985). The two primary axes (i.e., a vertical dimension and a horizontal dimension) of this circular array were initially used as a practical shorthand for summarizing the full system, and they were defined in terms of Dominance (vertical) and Love (horizontal). Since then, these two dimensions have been expanded and taken on various names depending on the author and specific application, but these include Control, Power, and Mastery for Dominance, and Affiliation, Warmth, and Connectedness for Love. More broadly, Wiggins (1991; 1997a; 2003) has construed them as the superordinate dimensions of Agency and Communion, which have propaedeutic explanatory power across scientific disciplines (Bakan, 1966). of behaviors observed in psychotherapy groups, eventually arriving at a circular array (Freedman, 1985). The two primary axes (i.e., a vertical dimension and a horizontal dimension) of this circular array were initially used as a practical shorthand for summarizing the full system, and they were defined in terms of Dominance (vertical) and Love (horizontal). Since then, these two dimensions have been expanded and taken on various names depending on the author and specific application, but these include Control, Power, and Mastery for Dominance, and Affiliation, Warmth, and Connectedness for Love. More broadly, Wiggins (1991; 1997a; 2003) has construed them as the superordinate dimensions of Agency and Communion, which have propaedeutic explanatory power across scientific disciplines (Bakan, 1966). INTERPERSONAL THEORY 52 2. The structure: Interpersonal functioning can be organized using the dimensions of agency and communion. 3. The levels: The same structure organizes functioning across levels of experience (i.e., from brain to behavior) ) Early articulations of interpersonal theory recognized the need for organizing experience and functioning across multiple “levels” (Leary & Coffey, 1955). The basic issue is that human personality functioning spans different levels that interpenetrate and interact to emerge as a coherent whole. The IPC would then serve as the common conceptual map for ensuring consistency when traveling within and across these levels. Influenced by psychodynamic thinking, early description of levels included public communication, conscious description, private symbolization, unexpressed unconscious, and values (Leary, 1957). Although the taxonomy of levels is open to continuous updating in CIIT and other areas of psychology, there are clear parallels between the levels proposed by Leary and concepts in contemporary personality science (e.g., self-reports, overt behavioral ratings, self-other agreement; Vazire, 2010). The broader point is that the notion that human psychology cannot be summarized with a single level of analysis has perhaps never been more apparent than now (Venables & Patrick, 2020). For instance, this is a foundational principle of the National Institute of Mental Health’s Research Domain Criteria approach (Insel et al., 2010). Thus, it is important to retain this fundamental tenet of interpersonal theory in CIIT, even as we may seek to update or augment the specific set of levels that are under consideration. CIIT is therefore not a theory about traits, overt behavior, or even neural circuits, it is, at its core, an integrative theory about the interpersonal system as a whole. Interpersonal traits are, like all personality traits, best understood to be abstractions or probabilistic descriptions of an individual’s tendency to think, feel, or behave in a particular manner. They are emergent phenomena, arising from the interplay among the various components of the interpersonal system. In the preceding section (Tenet 2), we mentioned several of the key domains that have been a focus in contemporary interpersonal research, including the basics of motivations (i.e., values), cognitions (e.g., efficacies), perception, behavior, and mental representations (Lukowitsky & Pincus, 2011), among others. Each of these has been fruitfully organized and operationalized using the agency and communion framework. However, these are not “levels” in the strictest sense, in that these are not all hierarchically organized. Rather, they might be better construed as different components or modalities of functioning that work in concert. Some of their key interaction processes are discussed below in Tenet 6, as we elaborate the dynamics of the interpersonal transaction cycle. INTERPERSONAL THEORY Schmidt et al., 1999). At the level of specific behaviors, interpersonal description permits moment-to-moment and transactional analyses of unfolding interpersonal processes. Thus, the agency and communion meta-structure and the IPC provide a “key conceptual map” (Kiesler, 1996, p. 172) for CIIT’s interpersonal description of personality structures and personality processes. Notably, interpersonal research on psychopathology is conducted at many levels of specificity (see Tenet 3; e.g., Bell et al., 2020; Luo et al., 2018; Sadler et al., 2015). 2. The structure: Interpersonal functioning can be organized using the dimensions of agency and communion. gg g g ) The IPC has been established in dispositional models and measures of an evolving set of interpersonal constructs including general traits, problems, impacts, strengths, sensitivities, efficacies, influence tactics, and values (Bliton & Pincus, 2020; Fournier et al., 2011; Hopwood et al., 2011; Locke, 2011; Wiggins, 1979). Agentic and communal dispositions imply enduring patterns of perceiving, thinking, feeling, and behaving that are probabilistic in nature, and describe an individual’s interpersonal tendencies aggregated across time, place, and relationships An individual’s interpersonal dispositions when understood in relation to their relationships. An individual’s interpersonal dispositions, when understood in relation to their motives and goals, are assumed to give rise to variations in adaptive and maladaptive behavior that are contextualized yet consistent across relevant interpersonal situations (Horowitz & Wilson, 2005; Wiggins, 1997b). Thus, at this level IPC models can describe a person’s typical ways of relating to others and refer to aspects of their interpersonal style. y g p p y At a more fine-grained level, the IPC structure can be used to classify the nature and intensity of specific interpersonal behaviors (Gifford, 1991; Moskowitz, 1994, 2005, 2009; Sadler et al., 2020), perceptions (Moskowitz & Zuroff, 2005), and impacts (Kiesler et al., 1997; 53 INTERPERSONAL THEORY we would expect that there are specific neurobiological features (e.g., structures, circuits, transmitters) that support or facilitate agentic and communal reward systems. By integrating the work from pioneering personality neuroscientists, the “levels” of CIIT now include key neurobiological systems and components. This draws on work directly referencing agentic and communal traits (Lenzenweger & Depue, 2020; Depue, 2006; Depue & Collins, 1999; Depue & Lenzenweger, 2005; Depue & Marrone-Strapinsky, 1999; Feldman, 2014; Klemm, 2015; Lenzenweger & Depue, 2016; Palumbo et al., 2020) as well as the Big Five/Five Factor Model (See Allen & DeYoung, 2017 for a review). Because dominance and affiliation, the two primary axes of the trait IPC, can be construed as rotational variants of extraversion and agreeableness (DeYoung et al., 2013; McCrae & Costa, 1989; Pincus, 2002), the Big Five personality neuroscience literature directly informs the biological levels of CIIT. Recall that the phenotypic topography mapped by the IPC is fully continuous and populated in its entirety by relevant markers (see Tenet 2), and the broad constructs of extraversion and agreeableness reflect blends of agency and communion. As such, there is no need to make assumptions of isomorphism between biological systems and specific articulations of phenotypic interpersonal traits as instantiated in particular questionnaires. we would expect that there are specific neurobiological features (e.g., structures, circuits, transmitters) that support or facilitate agentic and communal reward systems. By integrating the work from pioneering personality neuroscientists, the “levels” of CIIT now include key neurobiological systems and components. This draws on work directly referencing agentic and communal traits (Lenzenweger & Depue, 2020; Depue, 2006; Depue & Collins, 1999; Depue & Lenzenweger, 2005; Depue & Marrone-Strapinsky, 1999; Feldman, 2014; Klemm, 2015; Though a full review is beyond the scope of the current treatment, there now appears to be strong support for neurobehavioral structures and systems associated with sensitivity to reward aligned with agency and communion (see Allen & DeYoung, 2017; Depue & Collins, 1999; Depue, 2006; Depue & Marrone-Strapinsky, 1999; Waller & Wagner, 2019 for detailed reviews). These models generally assume that individual differences in the sensitivity and reactivity to relevant stimuli classes are reflective of individual differences in biological functioning. The agentic reward system involves, in large part, dopaminergic activity. Dopamine acts as a broad neuromodulator supporting appetitive strivings, incentive reward pursuit, and feelings such as desire, potency, and self-efficacy. 3. The levels: The same structure organizes functioning across levels of experience (i.e., from brain to behavior) As fundamental organizing constructs, agency and communion are part of the evolutionary architecture of human functioning. They have been selected to help humans resolve the major life tasks of getting ahead and getting along (Hogan & Roberts, 2000). As such, they too should be reflected in the functional architecture of human biology, including devoted or supporting neural circuits, neurotransmitters, and hormones. Here there is no need to assume a literal representation of the IPC “in the brain,” and instead we can leverage Gray’s (Gray & McNaughton, 2000; Pickering & Gray, 1999) notion of a “conceptual nervous system.” That is, 54 INTERPERSONAL THEORY 4. What gets us moving: Satisfying agentic and communal motives drive interpersonal behavior In what is arguably CIIT’s broadest and most interdisciplinary level, agency and communion classify the interpersonal motives, goals, and values of human relations (Horowitz, 2004). In interpersonal situations, motivation can reflect the agentic and communal nature of the individual’s long-term strivings, or more specific agentic and communal goals and values (e.g., to be in control; to be close) that specific behaviors are enacted to achieve (Grosse Holtforth et al., 2011; Horowitz et al., 2006; Locke, 2015; Trapnell & Paulhus, 2012). Sullivan (1953a, 1953b) suggested individuals are motivated to increase security and self-esteem (positively reinforcing) and avoid anxiety (negatively reinforcing). Thus, although there are individual differences in the strengths and patterning of these motivations, they are assumed to broadly be human universals. Satisfaction of these motives leads to decreased negative affect and vulnerability, increased positive affect, felt security, and self-esteem, whereas frustration of motives leads to the opposite (Horowitz et al., 2006; Leary et al., 1995; Lakey & Orehek, 2011; Mahadevan et al., 2019; Pincus & Hopwood, 2012). , ; p , ) In this respect, CIIT shares much in common with a control-theory approach to personality functioning (e.g., Carver & Scheier, 1998). Individuals seek to satisfy these two broad, biologically-based approach motives (Tenet 3) through a variety of characteristic adaptations, scaffolding strategies, and specific goals and behavior designed or selected to fulfill the motivation in particular interpersonal situations (Corr et al., 2013; Grosse Holtforth et al., 2011; Horowitz et al., 2006). Individuals differ from each other in their level of these motivations, and they differ within themselves over time, both in the long and the short term. Individuals differ from each other because of innate biology, life experiences and social learning, and the ongoing patterning and repatterning of interpersonal experiences in situations and their reinforcement (Benjamin, 1993; Pincus & Ansell, 2013). Within-person variation over the long term is due to development in typical maturation patterns, life events, and social role transitions (Roberts & DelVecchio, 2000; Roberts et al., 2006; Trucco et al., 2014; Wright et al., 2012), whereas short term variation reflects the varying demands of specific interpersonal contexts (e.g., McCabe & Fleeson, 2012; Moskowitz & Zuroff, 2004). Successful navigation of each of these dynamic time-scales involves the flexible adaptation and reconciliation of one’s motives with the demands of one’s developmental stage, social roles, and specific interpersonal contexts. p g p p A growing body of research supports these propositions. INTERPERSONAL THEORY articulation of human functioning has proved remarkably prescient, as this remains, or arguably has become, one of the major challenges facing the field. As we briefly summarize above, the agentic and communal meta-structure has proven useful for organizing the accumulating neurobehavioral findings, even as we integrate basic concepts from the broader field of personality neuroscience to extend interpersonal theory’s original levels. Future work is needed to continue to flesh out this longstanding, yet still nascent, notion of a multi-level model of personality functioning and dysfunction, but the necessary perspective and structure are features of CIIT. INTERPERSONAL THEORY As such it is not specifically implicated in agentic reward, but in incentive reward more broadly. Neurological structures associated with the agentic reward system include the ventral tegmental area to nucleus accumbens (i.e., in the striatum) pathway. In contrast, the communal reward system involves endogenous opiate activity, which supports feelings of satisfaction and affection, though oxytocin and vasopressin are also implicated. Opioid receptors in the ventral tegmental area- nucleus accumbens pathways are similarly involved in affiliative reward, as well as networks of connections among corticolimbic regions. In addition, agreeableness appears to be related to serotonergic functioning, which, is a broad neurotransmitter that is implicated in regulatory mechanisms writ large (Depue & Spoont, 1986; Spoont, 1992; Wright et al., 2019). However, the majority of work has examined the agentic and affiliative rewards systems as operationalized in extraversion scales and facets, and less work has investigated the neurobehavioral underpinnings of agreeableness per se. In sum, since its beginnings, interpersonal theory has explicitly incorporated a multi- level understanding of personality and psychopathology. In contrast to other structural models of personality and psychopathology, from the outset the aim was to find a structure that could serve as the common organizational scheme across levels and modalities of experience and functioning. Early theory emphasized how the dynamics of interactions among various levels gave rise to important personality processes (e.g., defense mechanisms). With some notable exceptions (e.g., Dawood & Pincus, 2016; Hopwood & Good, 2019; Kehl et al., 2020; Pincus et al., 2014), the particular manner in which levels interact with each other has received little empirical attention. That notwithstanding, the fundamental insight of needing a multi-level 55 INTERPERSONAL THEORY 5. How we move: The “interpersonal transaction cycle” provides probabilistic expectations for patterns of behavior and falsifiable predictions about behavioral sequences. An implication of Sullivan’s (1953a) definition of personality as the “enduring pattern of interpersonal situations that characterize a life” is that individual differences in personality and psychopathology are best characterized in terms of relatively stable patterns of within-person dynamics, contextualized within interpersonal situations. These patterns reflect interpersonal transactions cycles rooted in probabilistic functional connections between the thoughts, perceptions, feelings, and behaviors of individuals interacting with one another, and the situations contextualizing the interaction. Thus, CIIT is rooted in testable predictions about how interactions tend to unfold, and interpersonal researchers have pioneered methods for testing hypotheses about stability and variability in transactional social processes (e.g., Moskowitz & Zuroff, 2004; Sadler et al., 2009; Wright & Zimmermann, 2019). CIIT is a rich theory that makes a range of specific hypotheses about normative transaction. Here we briefly review connections between: a) two peoples’ interpersonal behavior (i.e., complementarity), b) one person’s behavior and another’s internal experience, c) past experiences and current behavior (i.e., copy processes), and d) situations and behavior. The prediction for which there is the most robust evidence within the interpersonal research tradition is complementarity (Carson, 1969; Kiesler, 1983; Sadler, Ethier, & Woody, 2011; Tracey, 1994), or the principle that during an interaction, the behavioral dominance of two people will tend to be reciprocal, whereas behavioral warmth will tend to be similar. There is strong evidence for complementarity, particularly when it is measured as a dynamic process that occurs as a situation unfolds (Bluhm et al., 1992; Markey et al., 2003; Sadler et al., 2009; Hopwood et al., 2020; Tracey, 1994, 2004). A second predicted pattern is between the behaviors of one person and the covert, internal reactions and experiences of another. Early work showed that different interpersonal behaviors provoked relatively predictable patterns of reaction from others in experimental studies (Wagner et al., 1995). Subsequent work in experience sampling studies has likewise shown systematic and normative connections between one person’s interpersonal behavior and another’s internal experience. For instance, there are reliable connections between one person’s coldness or dominance and another’s negative affect (Kaurin et al., 2020; Sadikaj et al., 2011; Sadikaj et al., 2013; Wang et al., 2014; Wright et al., 2017). A third predicted pattern is between the transaction cycles characteristic of early development and the interactions characteristic of present functioning. INTERPERSONAL THEORY the moment, goals to behave in an extraverted fashion explain observed extraverted behavior, which partially explains the link between goals and positive affect (McCabe & Fleeson, 2012). In each of these studies, though the average effect is consistent with theoretical predictions, individual differences in the strength of association between agentic and communal strivings and affective reactions were observed. the moment, goals to behave in an extraverted fashion explain observed extraverted behavior, which partially explains the link between goals and positive affect (McCabe & Fleeson, 2012). In each of these studies, though the average effect is consistent with theoretical predictions, individual differences in the strength of association between agentic and communal strivings and affective reactions were observed. 4. What gets us moving: Satisfying agentic and communal motives drive interpersonal behavior For instance, within the laboratory, experimental manipulations of inclusion/exclusion and status are associated with changes in self-esteem (Leary et al., 1995; Mahadevan et al., 2019). The same is true of such experiences in daily life (Mahadevan et al., 2020). In naturally occurring interpersonal situations, perceiving others as behaving more coldly or dominantly is associated with experiencing less positive and more negative affect (Sadikaj et al., 2013; Wright et al., 2017). In 56 INTERPERSONAL THEORY 57 However, longitudinal studies are needed to rule out alternative explanations to the interpersonal proposition that developmental processes are “copied” in adult interactions. A final prediction is about the link between aspects of situations and interpersonal behavior. There is strong evidence that different situations (including settings, interaction partners, topics, etc.) influence levels of interpersonal behavior (e.g., Brown & Moskowitz, 1998; Moskowitz et al., 2007). There is also evidence that situational factors impact levels of complementarity (Hopwood et al., 2020) and may moderate the link between interpersonal behaviors and psychopathology symptoms (Dowgwillo et al., 2019; Roche, Pincus, Conroy et al., 2013; Sadikaj et al., 2010). j ) Overall, these data strongly support the assumption there are reliable and culturally normative dynamic connections between certain patterns of interpersonal transaction. While many of probabilistic predictions about CIIT have been demonstrated empirically, most of these tests have occurred in isolation. More work is needed to combine the various hypotheses into coherent sequences that distinguish people from one another in terms of their characteristic personality style and psychopathology symptoms. Ultimately, the promise of the interpersonal approach is to piece these findings together in order to establish normative cycles, which would enable tests of the temporal sequences characteristic of patients’ difficulties (Pincus et al., 2009; Pincus & Wright, 2011). 5. How we move: The “interpersonal transaction cycle” provides probabilistic expectations for patterns of behavior and falsifiable predictions about behavioral sequences. Critchfield and Benjamin (2008, 2010; Critchfield et al., 2015) showed that there is specificity in the recalled scripts of developmental interactions with caregivers and patterns of contemporary interpersonal behavior among both patients and non-patients. These studies advanced research in this area by going beyond the general proposition that maladaptive behavior in childhood begets maladaptive behavior in adulthood, to empirically test and confirm specific links between the manner of positive and negative behavior recalled in childhood and the manner of behavior in adulthood. INTERPERSONAL THEORY 6. How we know we’re moving in the right direction (or not): Sustained deviations from these expected patterns (likely) reflect psychopathology. Several authors have elaborated interpersonal transaction cycles for different psychological conditions (e.g., Benjamin, 2002; Hopwood, 2018) and particular patients (Cain & Pincus, 2016; Pincus & Hopwood, 2012). These cycles are characterized by systematic deviance from normative functional interpersonal patterns. As with the probabilistic patterns discussed above, most research to date has focused on specific segments of broader dynamic processes, and researchers have yet to test these processes as complete units. However, many specific findings speak to the general proposition that sustained deviations from normative interpersonal processes are associated with psychopathology and dysfunction. There is robust evidence that maladaptive manifestations of interpersonal dispositions are characteristic of psychopathology (Pincus & Wiggins, 1990; Wilson et al., 2017; Wright et al., 2012) as well as levels and variability in negative affect over time (Liu et al., 2019; Wright et al., 2015). Interpersonal behaviors assessed at the momentary level are also related to levels of psychopathology (Dowgwillo et al., 2019; Roche, Pincus, Conroy et al., 2013; Sadikaj et al., 2010), and variability in momentary interpersonal behavior predicts interpersonal negative affect and interpersonal distress (Erickson et al., 2009). Interpersonal research from a variety of paradigms indicates that non-normative levels of variability in interpersonal behaviors, relative to the norm, is generally maladaptive (Côté et al., 2012; Erickson et al. 2009; Moskowitz et al., 2011; Rappaport et al., 2014; Ringwald et al., 2020a, 2020b; Russell et al., 2007). Likewise, a number of studies using different methods suggests that psychopathology is associated with interpersonal perceptions in naturalistic settings (Ambwani et al., 2015; Dowgwillo et al., 2019; Roche, Pincus, Hyde, et al., 2013). An interesting recent study combined these effects, showing that people who tend to exhibit greater levels of behavioral variability in interpersonal interactions also tend to have systematic biases in perceiving those interactions (Clegg et al., in press). INTERPERSONAL THEORY 58 Whereas an individuals’ warmth is normatively unrelated to that individuals’ dominance, some research suggests that interpersonal warmth and dominance are more likely to correlate within person in individuals with psychopathology, and that the nature of this correlation depends on the specific type of disorder (Fournier et al., 2009; Roche, Pincus, Hyde et al., 2013). Finally, a number of studies have suggested that deviations from complementarity are associated with psychopathology and dysfunction (Ansell et al., 2008; Dryer & Horowitz, 1997; Meisel et al., under review; Woods & Wright, 2019). There is also evidence that manipulating patterns of complementarity can be useful in psychotherapeutic treatment (Tracey, 1993; Tracey, Bludworth, & Glidden-Tracey, 2012).
https://openalex.org/W2012654782	https://europepmc.org/articles/pmc5424107?pdf=render	English	null	Orally active microtubule-targeting agent, MPT0B271, for the treatment of human non-small cell lung cancer, alone and in combination with erlotinib	Cell death and disease	2,014	cc-by	9,106	Orally active microtubule-targeting agent, MPT0B271, for the treatment of human non-small cell lung cancer, alone and in combination with erlotinib Tsai1,6, C-Y Wang2,6, J-P Liou1, H-C Pai2, C-J Hsiao2, J-Y Chang3,4, J-C Wang5, C-M Teng2 and S-L Pan,5 Microtubule-binding agents, such as taxanes and vinca alkaloids, are used in the treatment of cancer. The limitations of these treatments, such as resistance to therapy and the need for intravenous administration, have encouraged the development of new agents. MPT0B271 (N-[1-(4-Methoxy-benzenesulfonyl)-2,3-dihydro-1H-indol-7-yl]-1-oxy-isonicotinamide), an orally active microtubule-targeting agent, is a completely synthetic compound that possesses potent anticancer effects in vitro and in vivo. Tubulin polymerization assay and immunoﬂuorescence experiment showed that MPT0B271 caused depolymerization of tubulin at both molecular and cellular levels. MPT0B271 reduced cell growth and viability at nanomolar concentrations in numerous cancer cell lines, including a multidrug-resistant cancer cell line NCI/ADR-RES. Further studies indicated that MPT0B271 is not a substrate of P-glycoprotein (P-gp), as determined by ﬂow cytometric analysis of rhodamine-123 (Rh-123) dye efﬂux and the calcein acetoxymethyl ester (calcein AM) assay. MPT0B271 also caused G2/M cell-cycle arrest, accompanied by the up-regulation of cyclin B1, p-Thr161 Cdc2/p34, serine/threonine kinases polo-like kinase 1, aurora kinase A and B and the downregulation of Cdc25C and p-Tyr15 Cdc2/p34 protein levels. The appearance of MPM2 and the nuclear translocation of cyclin B1 denoted M phase arrest in MPT0B271-treated cells. Moreover, MPT0B271 induced cell apoptosis in a concentration-dependent manner; it also reduced the expression of Bcl-2, Bcl-xL, and Mcl-1 and increased the cleavage of caspase-3 and -7 and poly (ADP-ribose) polymerase (PARP). Finally, this study demonstrated that MPT0B271 in combination with erlotinib signiﬁcantly inhibits the growth of the human non-small cell lung cancer A549 cells as compared with erlotinib treatment alone, both in vitro and in vivo. These ﬁndings identify MPT0B271 as a promising new tubulin-binding compound for the treatment of various cancers. Cell Death and Disease (2014) 5, e1162; doi:10.1038/cddis.2014.128; published online 10 April 2014 Subject Category: Cancer structurally different compounds. The most common form of drug resistance to taxanes and other microtubule-targeting agents is overexpression of the efﬂux pump P-glycoprotein (P-gp)/multidrug resistance (MDR) protein.7 Clinically used microtubule-targeting agents, such as paclitaxel and vinblas- tine, are substrates for P-gp. Thus, a microtubule-targeting agent is required, which circumvents these mechanisms of drug resistance. 1School of Pharmacy, College of Pharmacy, Taipei Medical University, No. 250, Wu-hsing Street, Taipei 11031, Taiwan; 2Pharmacological Institute, College of Medicine, National Taiwan University, Taipei, Taiwan; 3National Institute of Cancer Research, National Health Research Institutes, Tainan, Taiwan; 4Division of Hematology and Oncology, Department of Internal Medicine, National Cheng Kung University Hospital, Tainan, Taiwan and 5The Ph.D. Program for Cancer Biology and Drug Discovery, College of Medical Science and Technology, Taipei Medical University, No. 250, Wu-hsing Street, Taipei 11031, Taiwan Corresponding author: S-L Pan, The Ph.D. Program for Cancer Biology and Drug Discovery, College of Medical Science and Technology, Taipei Medical University, No. 250, Wu-hsing Street, Taipei 11031, Taiwan. Tel: +886 2 27361661 ext 7671; Fax: +886 2 23221742; E-mail: slpan@tmu.edu.tw or psl0826@ms13.hinet.net 6These authors contributed equally to this work. Received 21.10.13; revised 15.2.14; accepted 28.2.14; Edited by G Ciliberto Keywords: non-small cell lung cancer; microtubule-binding agents; apoptosis; erlotinib Abbreviations: P-gp, P-glycoprotein; Rh-123, rhodamine-123; calcein AM, calcein acetoxymethyl ester; PARP, poly (ADP-ribose) polymerase; MDR, multidrug resistance; MRP, multidrug resistance-associated protein; STAT3, signal transducer and activator of transcription 3; EGFR, epidermal growth factor receptor; NSCLC, non-small cell lung cancer; PLK1, serine/threonine kinases polo-like kinase 1; DAPI, 4,6-diamidino-2-phenylindole; PI, propidium iodide; siRNA, small interfering RNA Citation: Cell Death and Disease (2014) 5, e1162; doi:10.1038/cddis.2014.128 & 2014 Macmillan Publishers Limited All rights reserved 2041-4889/14 Citation: Cell Death and Disease (2014) 5, e1162; doi:10.1038/cddis.2014.128 & 2014 Macmillan Publishers Limited All rights reserved 2041-4889/14 OPEN OPEN Orally active microtubule-targeting agent, MPT0B271, for the treatment of human non-small cell lung cancer, alone and in combination with erlotinib y p p y Epidermal growth factor receptor (EGFR), a trans- membrane glycoprotein, is frequently overexpressed in human tumors such as breast, ovarian, prostate, pancreatic, and non-small cell lung cancer (NSCLC). This overexpression is correlated with poor prognosis and worse clinical outcome. Thus, targeted therapies directed at blocking EGFR function are attractive potential therapeutics for cancer.19 Erlotinib, a reversible EGFR tyrosine kinase inhibitor, was initially approved for treatment of patients with advanced NSCLC who had failed one chemotherapy regimen. Although clinical results have demonstrated that erlotinib monotherapy showed a survival beneﬁt in comparison with gemcitabine for patients with NSCLC or pancreatic cancer, resistance to erlotinib reduces its efﬁcacy.20,21 The T790M point mutation in EGFR causes a conformational change at the ATP binding pocket, thus decreasing the afﬁnity between erlotinib and EGFR and leading to acquired/secondary resistance.22,23 MPT0B271 inhibits cancer cell proliferation and induces cytotoxicity in vitro. The ability of MPT0B271 to inhibit cancer cell growth was evaluated using the Sulforhodamine B (SRB) assay. As shown in Figure 2a, MPT0B271 inhibited the proliferation of A549, AsPC-1, HCT116, Hep3B, MDA- MB-231, PC3, SKOV3, and NCI/ADR-RES cells with GI50 values of 27.9, 23.3, 21.0, 35.5, 19.0, 20.4, 18.5, and 18 nM, respectively. MPT0B271 reduced the viability of several human cancer cell lines with IC50 values in the low nanomolar range of 21–73 nM, as measured by the 3-(4,5- dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay (Figure 2b). In the present study, we demonstrate that MPT0B271 (N-[1-(4-methoxy-benzenesulfonyl)-2,3-dihydro-1H-indol-7-yl]- 1-oxy-isonicotinamide), an orally active compound, represses microtubule polymerization. We evaluated the antiproliferative activity of MPT0B271 against several cancer cell lines as well as P-gp-overexpressing NCI/ADR-RES cells. We investigated the action mechanism of MPT0B271 on cell-cycle progression and apoptosis. Furthermore, we explored the anticancer activity of MPT0B271 alone and in combination with erlotinib in vitro and in vivo. Our results suggest that MPT0B271 is a promising therapeutic candidate for the treatment of human cancers. Effects of paclitaxel and vincristine on the growth of NCI/ADR-RES cells. MPT0B271 was equally potent against the multidrug-resistant cancer cell line NCI/ADR-RES, as measured by both SRB and MTT assays. The growth inhibition of NCI/ADR-RES cells treated with paclitaxel or vincristine was then evaluated. As shown in Figure 2c, the mean GI50 values of paclitaxel and vincristine were 7.670±1.228 mM and 8.156±0.854 mM, respectively, indicat- ing that the effect of MPT0B271 was more potent than that of paclitaxel or vincristine in NCI/ADR-RES cells. Orally active microtubule-targeting agent, MPT0B271, for the treatment of human non-small cell lung cancer, alone and in combination with erlotinib Such a treatment could be advantageous for patients with drug-resistant tumors.8–10 Furthermore, another limitation of current treatments is their requirement for intravenous delivery, which leads to hypersensitivity reactions.11,12 Microtubules are the main component of the cytoskeleton and have an essential role in cell division, intracellular transport, and motility.1 Several clinically important microtubule-target- ing agents, such as taxanes and vinca alkaloids, bind to microtubules, thereby altering the normal dynamic equilibrium and resulting in either the stabilization or destabilization of microtubules.2 The consequence of disrupting the microtu- bule organization with these drugs is the G2/M-phase arrest of the cell cycle and ultimately apoptotic or non-apoptotic cell death.3,4 Even though microtubule-targeting agents are used clini- cally to treat patients with neoplastic disease, they have substantial drawbacks such as the development of resistance over time.5 Drug resistance can be intrinsic, in which case the ﬁrst chemotherapy fails, or acquired, in which case there is a response to the ﬁrst round of chemotherapy but failure of the second.6 In either case, tumors become refractory to various Signal transducer and activator of transcription 3 (STAT3) is inappropriately activated in various tumor types, such as lymphoma, breast, ovarian, pancreatic, and lung cancer, and is particularly activated in aggressive and invasive tumors.13,14 STAT3 is a transcription factor that translocates Treatment of MPT0B271 alone or in combination with erlotinib in human NSCLC A-C Tsai et al 2 2 into the nucleus, binding to its responsive element after activation. STAT3 regulates the expression of various genes including Bcl-xL, Mcl-1, cyclin D1, and p53, which regulate cell-cycle progression, cell survival, and proliferation.15,16 In addition to its transcriptional function, STAT3 associates with the cytoskeleton and may thus regulate microtubule function. Microtubule-interfering agents, such as paclitaxel and vinor- elbine, can inhibit STAT3 tyrosine phosphorylation.17,18 the absolute oral bioavailability being 26%. At an oral dose of 20 mg/kg, MPT0B271 showed rapid absorption in mice with a short maximal plasma concentration time (Tmax) of 15 min (the ﬁrst sampling time point). The mean Cmax and AUC(0–27) of MPT0B271 was 402 mg/ml and 558 ng h/ml, respectively. To evaluate the effect of MPT0B271 in vivo, nude mice that received tumor xenografts of the human NSCLC cell line A549 were used. As illustrated in Figure 1e (upper panel), oral administration of MPT0B271 resulted in a dose- dependent inhibition of tumor growth, and animals did not lose 410% of their body weight during treatment (Figure 1e, lower panel). Orally active microtubule-targeting agent, MPT0B271, for the treatment of human non-small cell lung cancer, alone and in combination with erlotinib Next, we performed ﬂow cytometric analysis to measure the accumulation of rhodamine-123 (Rh-123), a P-gp-transported ﬂuorescent dye, in NCI/ADR-RES cells following the treatment with MPT0B271. After incubation with classic P-gp inhibitors, verapamil and cyclosporine A, we observed a large increase in Rh-123 accumulation in comparison with the vehicle-treated cells. However, Rh-123 accumulation did not increase in NCI/ADR-RES cells after incubation with MPT0B271 (Figure 2d). Calcein acetoxymethyl ester (calcein AM) is a ﬂuorescent substrate for P-gp and can therefore be used to measure the transport activity of P-gp. Thus, P-gp inhibition was examined by measuring calcein AM retention using the Multidrug Resistant Assay.24 As illustrated in Figure 2e, the increase in intracellular ﬂuorescence was indicative of P-gp inhibition by verapamil or cyclosporine A. In contrast, the efﬂux of calcein AM was not affected by MPT0B271. These results suggest that MPT0B271 circumvents P-gp-mediated drug resistance. Results MPT0B271 inhibits tubulin polymerization. MPT0B271 (Figure 1a) inhibited microtubule formation in an in vitro tubulin polymerization assay, and the effect was similar to the inhibition induced by the microtubule-destabilizing agents colchicine and vincristine. In contrast, the widely used microtubule-stabilizing agent paclitaxel induced a marked increase in tubulin polymerization (Figure 1b). The effect of MPT0B271 on the arrangement and distribution of the microtubule network in cells was studied in situ by immuno- ﬂuorescence. As shown in Figure 1c, treatment of A549 cells with MPT0B271 disrupted the microtubule cytoskeleton. These data indicate that MPT0B271 depolymerizes micro- tubules in tumor cells. MPT0B271 is an orally bioavailable synthetic inhibitor of tumor growth in vivo. The MPT0B271 plasma concentra- tion versus time proﬁle following intravenous and oral administration to CD-1 (Crl.) mice is shown in Figure 1d. The half-life (t1/2) for oral administration (2.0 h) was slightly longer than the half-life for intravenous administration with MPT0B271 arrests the cell cycle in M phase. Most tubulin-targeting agents induce cell-cycle arrests;4 therefore, the effect of MPT0B271 on cell-cycle progression was assessed by FACScan ﬂow cytometry using propidium iodide (PI) staining. The data showed that treatment of Cell Death and Disease Treatment of MPT0B271 alone or in combination with erlotinib in human NSCLC A-C Tsai et al Figure 1 (a) Chemical structure of MPT0B271. (b) Effect of MPT0B271 on tubulin polymerization. Puriﬁed tubulin in reaction buffer was incubated at 371C in the absence or presence of increasing concentrations of MPT0B271, 10 mM paclitaxel, 10 mM colchicine, or 10 mM vincristine. Assembly of microtubules was then measured over 30 min at 1 min intervals at an absorbance of 340 nm using a spectrophotometer. (c) Immunoﬂuorescence staining of microtubules in A549 cells. Cells were treated with vehicle (DMSO), 0.3 mM MPT0B271, paclitaxel, or vincristine for 24 h. Cells were labeled with a b-tubulin antibody and an FITC-conjugated anti-mouse IgG antibody, were counterstained with 4,6-diamidino-2-phenylindole (DAPI) and observed by confocal microscopy. Left, DAPI; middle, microtubule network; right, merged microtubule network and DAPI. (d) PK properties, plasma concentration versus time proﬁles of MPT0B271 after i.v. (2 mg/kg) and p.o. (20 mg/kg) dosing of fasted male CD-1 (Crl.) mice. (e) Efﬁcacy of MPT0B271, dosed orally, on tumor xenografts. Upper panel, tumor growth of A549 xenografts in nude mice that were orally treated with or without MPT0B271 (5, 10, and 20 mg/kg). Tumor growth is presented as the mean tumor volume (mm3)±S.E. Results Tumor volume was determined using caliper measurements and was calculated as the product of 1/2 length width2. Lower panel, body weight (g) of the mice. Po0.05 and Po0.01 as compared with the control group ymerization. Puriﬁed tubulin in reaction buffer was incubated at 371C in the absence e, or 10 mM vincristine. Assembly of microtubules was then measured over 30 min at orescence staining of microtubules in A549 cells. Cells were treated with vehicle h a b-tubulin antibody and an FITC-conjugated anti-mouse IgG antibody, were scopy. Left, DAPI; middle, microtubule network; right, merged microtubule network after i.v. (2 mg/kg) and p.o. (20 mg/kg) dosing of fasted male CD-1 (Crl.) mice. of A549 xenografts in nude mice that were orally treated with or without MPT0B271 . Tumor volume was determined using caliper measurements and was calculated as 05 and Po0.01 as compared with the control group Figure 1 (a) Chemical structure of MPT0B271. (b) Effect of MPT0B271 on tubulin polymerization. Puriﬁed tubulin in reaction buffer was incubated at 371C in the absence or presence of increasing concentrations of MPT0B271, 10 mM paclitaxel, 10 mM colchicine, or 10 mM vincristine. Assembly of microtubules was then measured over 30 min at 1 min intervals at an absorbance of 340 nm using a spectrophotometer. (c) Immunoﬂuorescence staining of microtubules in A549 cells. Cells were treated with vehicle (DMSO), 0.3 mM MPT0B271, paclitaxel, or vincristine for 24 h. Cells were labeled with a b-tubulin antibody and an FITC-conjugated anti-mouse IgG antibody, were counterstained with 4,6-diamidino-2-phenylindole (DAPI) and observed by confocal microscopy. Left, DAPI; middle, microtubule network; right, merged microtubule network and DAPI. (d) PK properties, plasma concentration versus time proﬁles of MPT0B271 after i.v. (2 mg/kg) and p.o. (20 mg/kg) dosing of fasted male CD-1 (Crl.) mice. (e) Efﬁcacy of MPT0B271, dosed orally, on tumor xenografts. Upper panel, tumor growth of A549 xenografts in nude mice that were orally treated with or without MPT0B271 (5, 10, and 20 mg/kg). Tumor growth is presented as the mean tumor volume (mm3)±S.E. Tumor volume was determined using caliper measurements and was calculated as the product of 1/2 length width2. Lower panel, body weight (g) of the mice. Results Po0.05 and Po0.01 as compared with the control group Cell Death and Disease Cell Death and Disease t of MPT0B271 alone or in combination with erlotinib in human NSCLC A-C Tsai et al Treatment of MPT0B271 alone or in combination with erlotinib in human NSCLC A-C Tsai et al A C Tsai et al 4 Figure 2 (a) Various types of human cancer cells were treated with the indicated concentrations of MPT0B271 for 48h. Then, cell growth inhibition was determined using the SRB assay, and the GI50 of each cell line is expressed as the mean±S.E. of four independent determinations. (b) The cytotoxic effects of various human cancer cell lines were determined using an MTT assay. The IC50 of each cell line is expressed as the mean±S.E. of four independent determinations. (c) NCI/ADR-RES cells were treated with the indicated concentration of paclitaxel or vincristine for 48h, and cell growth was determined by the SRB assay. Data are expressed as the mean±S.E. of at least three independent experiments. Po0.01 as compared with the control group. (d) Effect of MPT0B271 on P-gp activity. NCI/ADR-RES cells were pretreated with or without MPT0B271 (0.025, 0.05, and 0.1mM), verapamil (50mM), cyclosporine A (10 mM), or vincristine (30mM) for 1h and then co-treated with 10mM rhodamine 123 (Rh-123). After 1h incubation at 371C, cells were washed with PBS, collected by trypsinization and detected byﬂow cytometry. (e) NCI/ADR-RES cells were incubated in the absence or presence of the indicated agents for 30min and then stained with calcein AM ﬂuorescent dye. Fluorescence was measured at an excitation wavelength of 485nm and an emission wavelength of 535nm. Po0.05 as compared with the control group Figure 2 (a) Various types of human cancer cells were treated with the indicated concentrations of MPT0B271 for 48h. Then, cell growth inhibition was determined using the SRB assay, and the GI50 of each cell line is expressed as the mean±S.E. of four independent determinations. (b) The cytotoxic effects of various human cancer cell lines were determined using an MTT assay. The IC50 of each cell line is expressed as the mean±S.E. of four independent determinations. (c) NCI/ADR-RES cells were treated with the indicated concentration of paclitaxel or vincristine for 48h, and cell growth was determined by the SRB assay. Data are expressed as the mean±S.E. of at least three independent experiments. **Po0.01 as compared with the control group. Results The effect of MPT0B271 on p53 was evaluated by treating A549 cells for 6–48 h with MPT0B271 and then by immunoblotting with p53 and phospho-p53 (Ser15) anti- bodies. Treatment with MPT0B271 upregulated the expres- sion of p53 protein and the phosphorylation of p53 at Ser15 (Figure 5a). However, knockdown of p53 by p53 small interfering RNA (siRNA) did not attenuate the induction of PARP and prevent the cytotoxic effects (Figure 5b). In addition, we further evaluated the in vitro cytotoxic activities of MPT0B271 in other human NSCLC cell lines, H1299 (null p53) and H226 (mutant p53), using the MTT assay. Treatment of H1299 and H226 cells with MPT0B271 reduced cell viability in a concentration-dependent manner, with mean IC50 values of 0.110±0.014 mM and 0.046±0.003 mM, respectively (Figure 5c). Figure 3 (a) Effect of MPT0B271 on cell-cycle progression. A549 cells were exposed to 0.3 mM MPT0B271 for the indicated times and then stained with PI to determine the proportion of DNA. Data acquisition and analysis were performed on a FACScan ﬂow cytometer. The data are expressed as the mean±S.E. of at least three independent experiments. (b) The effect of MPT0B271 on G2/M cell-cycle regulatory proteins. A549 cells were treated with 0.3 mM MPT0B271 for the indicated times. Whole-cell extracts were subjected to SDS-PAGE and immunoblotted with the indicated antibodies. (c) Treatment with 0.3 mM MPT0B271 for the indicated times. Nuclear lysates were subjected to western blot analysis using an antibody speciﬁc for cyclin B1 Figure 3 (a) Effect of MPT0B271 on cell-cycle progression. A549 cells were exposed to 0.3 mM MPT0B271 for the indicated times and then stained with PI to determine the proportion of DNA. Data acquisition and analysis were performed on a FACScan ﬂow cytometer. The data are expressed as the mean±S.E. of at least three independent experiments. (b) The effect of MPT0B271 on G2/M cell-cycle regulatory proteins. A549 cells were treated with 0.3 mM MPT0B271 for the indicated times. Whole-cell extracts were subjected to SDS-PAGE and immunoblotted with the indicated antibodies. (c) Treatment with 0.3 mM MPT0B271 for the indicated times. Nuclear lysates were subjected to western blot analysis using an antibody speciﬁc for cyclin B1 STAT3 tyrosine phosphorylation declined dramatically in MPT0B271-treated A549 cells, as determined by the PathScan Phospho-STAT3 (Tyr705) Sandwich ELISA kit (Figure 5e). However, transfection with constitutively active STAT3 only slightly rescued cell viability in A549 cells (Figure 5f). Results (d) Effect of MPT0B271 on P-gp activity. NCI/ADR-RES cells were pretreated with or without MPT0B271 (0.025, 0.05, and 0.1mM), verapamil (50mM), cyclosporine A (10 mM), or vincristine (30mM) for 1h and then co-treated with 10mM rhodamine 123 (Rh-123). After 1h incubation at 371C, cells were washed with PBS, collected by trypsinization and detected byﬂow cytometry. (e) NCI/ADR-RES cells were incubated in the absence or presence of the indicated agents for 30min and then stained with calcein AM ﬂuorescent dye. Fluorescence was measured at an excitation wavelength of 485nm and an emission wavelength of 535nm. Po0.05 as compared with the control group Cell Death and Disease Treatment of MPT0B271 alone or in combination with erlotinib in human NSCLC A-C Tsai et al A549 cells with MPT0B271 for 6–72 h resulted in a time- dependent accumulation of cell populations at the G2/M phase followed by a subsequent increase in the hypodiploid sub-G1 cell population, which indicated apoptosis (Figure 3a). To elucidate the mechanisms through which MPT0B271 induced the G2/M-phase arrests, the levels of the G2/M phase regulatory proteins were assessed. Western blot analysis revealed that MPT0B271 induced upregulation of cyclin B1, p-Thr161 Cdc2/p34, serine/threonine kinases polo-like kinase 1 (PLK1), and aurora kinase A and B; downregulation of Cdc25C and p-Tyr15 Cdc2/p34 protein levels; and a marked increase in mitosis-speciﬁc MPM2 phosphoprotein expression (Figure 3b). In addition, a remarkable nuclear accumulation of cyclin B1 protein was demonstrated in MPT0B271-treated A549 cells as compared with untreated cells (Figure 3c). MPT0B271 triggers apoptosis in A549 cells. The ability of MPT0B271 to induce apoptosis was examined by measuring cytoplasmic histone-associated DNA fragments using the Cell Death Detection ELISAPlus kit. The data show that increasing concentrations of MPT0B271 increased the level of DNA fragments in A549 cells after 48 h of treatment (Figure 4a). Moreover, MPT0B271 caused a concentration- dependent increase in cleaved executioner caspase-3 and its downstream substrate poly (ADP-Ribose) polymerase (PARP) (Figure 4b). Based on these results, we investigated the effect of MPT0B271 on the expression of anti-apoptotic proteins. The protein levels of Bcl-2, Bcl-xL, and Mcl-1 in A549 cells decreased in a time-dependent manner after MPT0B271 treatment (Figure 4c). Furthermore, exogenous Mcl-1 overexpression partially abolished the activation of PARP cleavage and rescued cell viability in A549 cells, suggesting that MPT0B271 triggers apoptotic cell death through the suppression of Mcl-1 expression (Figure 4d). MPT0B271 inhibits constitutive p53 activation in A549 cells. Results MPT0B271 inhibits constitutive STAT3 activation in A549 cells. Next, we investigated the effect of MPT0B271 on the modulation of constitutive STAT-3 phosphorylation in A549 cells by western blot analysis using an antibody that recognizes phosphorylation at Tyr705. As shown in Figure 5d, MPT0B271 signiﬁcantly inhibited the constitutive phosphoryla- tion of STAT3 in a time-dependent manner. The level of MPT0B271 in combination with erlotinib increased tumor cell growth inhibition in vitro and in vivo. Erlotinib has proven efﬁcacy in advanced NSCLC, but resistance to Cell Death and Disease Cell Death and Disease Treatment of MPT0B271 alone or in combination with erlotinib in human NSCLC A-C Tsai et al Figure 4 (a) Measurement of apoptosis. A549 cells were treated with the indicated concentration of MPT0B271, and oligonucleosomal DNA fragmentation was quantitatively assessed with the Cell Death ELISAPLUS kit. Apoptosis was enhanced in relation to control cells. Data are expressed as the mean±S.E. of at least three independent experiments. *Po0.001, as compared with the control group. (b) A549 cells were exposed to serial concentrations of MPT0B271 for 48 h, and whole-cell lysates were collected and immunoblotted with antibodies against caspase-3, -7, -8, and -9 and PARP. (c) After treatment with vehicle or MPT0B271 (0.3 mM) for the indicated times, A549 cells were harvested and lysed. Equal amounts of lysate protein were run on an SDS-PAGE gel, transferred onto nitrocellulose membrane and incubated with the indicated antibodies. (d) Effect of ectopic Mcl-1 on MPT0B271-induced cell apoptosis. A549 cells were transfected with vector or Mcl-1 plasmid for 24 h and then incubated with or without MPT0B271 for 48 h. Whole-cell lysates were subjected to western blot analysis, and cell viability was measured by the SRB assay Treatment of MPT0B271 alone or in combination with erlotinib in human NSCLC A-C Tsai et al 6 6 Figure 4 (a) Measurement of apoptosis. A549 cells were treated with the indicated concentration of MPT0B271, and oligonucleosomal DNA fragmentation was quantitatively assessed with the Cell Death ELISAPLUS kit. Apoptosis was enhanced in relation to control cells. Data are expressed as the mean±S.E. of at least three independent experiments. Po0.001, as compared with the control group. (b) A549 cells were exposed to serial concentrations of MPT0B271 for 48 h, and whole-cell lysates were collected and immunoblotted with antibodies against caspase-3, -7, -8, and -9 and PARP. Results (c) After treatment with vehicle or MPT0B271 (0.3 mM) for the indicated times, A549 cells were harvested and lysed. Equal amounts of lysate protein were run on an SDS-PAGE gel, transferred onto nitrocellulose membrane and incubated with the indicated antibodies. (d) Effect of ectopic Mcl-1 on MPT0B271-induced cell apoptosis. A549 cells were transfected with vector or Mcl-1 plasmid for 24 h and then incubated with or without MPT0B271 for 48 h. Whole-cell lysates were subjected to western blot analysis, and cell viability was measured by the SRB assay erlotinib in combination with MPT0B271 is effective against erlotinib-resistant NSCLC cell lines. erlotinib in combination with MPT0B271 is effective against erlotinib-resistant NSCLC cell lines. erlotinib also occurs.23 Thus, we investigated whether the combination of erlotinib and MPT0B271 is effective against erlotinib-resistant human NSCLC A549 cells. We chose these concentrations from dose-response curves (data not shown). As shown in Figure 6a and Supplementary Figure 2, erlotinib (5 mM) in combination with MPT0B271 (0.0125 or 0.025 mM) resulted in signiﬁcantly higher cell death than the monother- apeutic treatments, as measured using an enzyme immu- noassay for histone-associated DNA fragments. Finally, treatment of A549 xenograft-bearing nude mice with a combination of erlotinib and MPT0B271 resulted in signiﬁ- cantly decreased tumor progression without loss of their body weight as compared with either erlotinib or MPT0B271 monotherapy (Figures 6b and c). These data indicate that Discussion In the present study, MPT0B271 was identiﬁed as a microtubule-depolymerizing agent and exhibited signiﬁcant antitumor activity against various cancer cell lines, as well as drug-resistant sublines. It has been illustrated that p53 serves as a key player in cellular response to various extracellular and intracellular stresses such as DNA damage, oncogenic action, and microtubule disruption.25 Our results showed that MPT0B271 induces the expression of p53 protein in a time- dependent manner, but the knockdown of p53 cannot prevent Cell Death and Disease Treatment of MPT0B271 alone or in combination with erlotinib in human NSCLC A-C Tsai et al 7 Figure 5 (a) The effect of MPT0B271 on p53 expression and phosphorylation in A549 cells. Cells were treated with MPT0B271 (0.3 mM) for the indicated times, and whole-cell extracts were prepared and underwent western blot analysis using the indicated antibodies. (b) A549 cells were pretreated with or without p53 siRNA for 24 h and then incubated with or without MPT0B271 (0.1 mM) for 48 h. Upper panel, total cellular lysates were subjected to western blot analysis of p53 and PARP. Lower panel, cell viability was measured by the SRB assay and expressed as a percentage of the untreated control. (c) Concentration-dependent effect of MPT0B271 on cell viability. H1299 and H226 cells were treated with or without the indicated concentration of MPT0B271 for 48 h, and the cytotoxic effect was determined with the MTT assay. Data are expressed as the mean±S.E. of at least three independent experiments. Po0.05; Po0.01, and Po0.001 as compared with the control group. (d) The effect of MPT0B271 on STAT3 phosphorylation in A549 cells. Cells were treated with MPT0B271 (0.3 mM) for the indicated times, and whole-cell extracts were prepared and analyzed for STAT-3 phosphorylation (at Tyr705). (e) A549 cells were treated with various concentrations (0.025–0.1 mM) of MPT0B271 for 24 h, after which the level of STAT3 tyrosine phosphorylation in cells was measured using the PathScan Phospho-Stat3 (Try705) Sandwich ELISA kit and spectrophotometry at 450 nm. Data represent the mean±S.E. of at least three independent experiments. Po0.05; **Po0.001 as compared with the control group. (f) The effect of ectopic STAT3 on MPT0B271-induced cell apoptosis. A549 cells were transfected with vector or STAT3 plasmid for 24 h and then incubated with or without MPT0B271 for 48 h. Discussion We also tested the effect of verapamil in combination with MPT0B271 using SRB assay (Supplementary Figure 1). The GI50 values of the combination of verapamil with MPT0B271 in NCI/ADR cells were 16 nM, which indicated that the antiproliferative effect of combination was similar with that of treatment with MPT0B271 alone. These results demonstrated that MPT0B271 may not be a substrate or modulator of the P-gp efﬂux pump. Similar to other microtubule-targeting agents, MPT0B271 induced a concentration-dependent G2/M blockade, as indi- cated by ﬂow cytometric analysis and expression of the MPM-2 epitope, a mitosis-speciﬁc marker. Another marker of the G2/M phase is cyclin B1, which is localized in the cytoplasm in G2 phase but moves rapidly into the nucleus at the beginning of mitosis.31 Western blot analysis showed that the protein level of cyclin B1 in the nuclear fraction of MPT0B271-treated cells was signiﬁcantly enhanced compared with that of untreated cells, implying that cells treated with MPT0B271 were arrested in the mitotic phase. Cdc2/p34 is a cell-cycle kinase responsible for the regulation of G2 progression and G2/M transition in all eukaryotic cells. It has been reported that the activity of Cdc2/ p34 kinase depends not only on its association with cyclin B1 but also on its phosphorylation state.32 We found that MPT0B271 markedly reduced p-Tyr15 Cdc2/p34 levels and increased p-Thr161 Cdc2/p34 levels and was associated with cyclin B1 upregulation, but had no effect on Cdc2/p34 expression. The PLK-1 and aurora kinase A and B are active during mitosis.33 PLK-1 is an early trigger of G2/M transition in mammalian cells and has been implicated in the regulation of different processes including mitotic entry, spindle formation, and cytokinesis. Aurora kinase A and B are required for centrosome separation, mitotic spindle assembly, chromo- some biorientation, and cytokinesis during mitosis.34,35 Our results showed that MPT0B271 enhances the expression of PLK-1 and aurora kinase A and B proteins in a time-dependent manner, supporting the notion that MPT0B271 induces M-phase arrest in A549 cells. Our results showed that in addition to inducing cell-cycle arrests in mitosis, treatment with MPT0B271 is followed by a subsequent increase in the hypodiploid (apoptotic sub-G1 peak) population of the cell cycle. The presence of cytoplasmic histone-associated DNA fragments induced by MPT0B271 was also observed, providing additional evidence that MPT0B271 induces apoptosis in A549 cells. Cysteine pro- teases (caspases) have a critical role in apoptosis through the proteolysis of speciﬁc targets. Discussion Whole-cell lysates were subjected to western blot analysis, and cell viability was measured by the SRB assay Drug resistance is a serious problem that restricts the use of microtubule-targeting agents for clinical therapy. The MDR P-gp, a plasma membrane protein overexpressed in multi- drug-resistant tumor cells, is a major obstacle to the success of chemotherapy. Efﬂux of drugs caused by MDR proteins augments the elimination of drugs from target cells and leads to drug resistance.28 Many microtubule-targeting agents are substrates of P-gp, and higher doses of these drugs are required to achieve adequate intracellular concentrations in multidrug-resistant cancer cells.8 We employed Rh-123 and calcein AM, both substrates of P-gp and multidrug resistance- associated protein (MRP), as probes to detect chemical apoptosis in cells treated with MPT0B271. In addition, A549 (wild-type p53), H1299 (p53 null), and H226 (p53 mutant) exhibited similar IC50 values, indicating that the antitumor effect of MPT0B271 may not correlate with the p53 status. The traditional microtubule-targeting agents in clinical use, such as paclitaxel, require intravenous administration with a long-term remedial course, causing physical and mental suffering and reducing the patients’ quality of life. In addition, they often lead to hypersensitivity reactions and neuro- pathy.26,27 Here, it is noteworthy that MPT0B271 possesses oral availability, improves solubility, and is efﬁcacious in vivo against NSCLC tumor xenografts. Cell Death and Disease Cell Death and Disease Treatment of MPT0B271 alone or in combination with erlotinib in human NSCLC A-C Tsai et al 8 Figure 6 In vitro and in vivo antitumor activity of MPT0B271 in combination with erlotinib. (a) A549 cells were treated with erlotinib (5 mM) in combination with MPT0B271 (0.0125 or 0.025 mM) for 48 h, and cell apoptosis was measured using the Cell Death ELISAPLUS kit. Data are expressed as the mean±S.E. of at least three independent determinations. (b) A549 xenograft model. A549-tumor-bearing nude mice were treated with vehicle, MPT0B271 (20 mg/kg/day by oral gavage q2d), erlotinib (25 mg/kg/day by oral gavage once a day), or MPT0B271 in combination with erlotinib. Tumor was excised when the tumor size reached 1200 mm3. (c) The body weight of the mice measured daily during the ﬁrst week and then at the days of administration. Po0.05; Po0.01, and *Po0.001 as compared with the control group that the accumulation of Rh-123 or the intracellular ﬂuores- cence of Calcein is not increased in MPT0B271-treated NCI/ ADR-RES cells. Discussion Our data showed that erlotinib monotherapy did not result in signiﬁcant cell death in A549 cells as compared with the control group, which was assessed using a cell death detection ELISA assay kit. However, combination therapy of MPT0B271 with erlotinib showed a synergistic effect in A549 cells. A similar result was obtained in the A549 subcutaneous xenograft mouse model. Erlotinib had no effect on tumor progression, but in comparison with MPT0B271 could produce a stronger antitumor effect. These results may indicate that MPT0B271 in combination with erlotinib would be a useful treatment in cases of NSCLC, which have developed progressive disease. In vitro tubulin polymerization assay. The effect of identiﬁed compounds on tubulin polymerization was determined kinetically using the CytoDYNAMIX Screen kit (BK006P, Cytoskeleton Inc., Denver, CO, USA). Cold porcine tubulin protein (499% purity) was added to G-PEM buffer (80 mM PIPES, 2 mM MgCl2, 0.5 mM EGTA, 1 mM GTP, pH 6.9) containing 15% glycerol with or without the identiﬁed compounds. The sample mixture was dotted onto a prewarmed 96-well plate, which was immediately transferred to a 371C plate reader (SpectraMax Plus, Molecular Devices Inc., Sunnyvale, CA, USA). The absorbance was read every minute for 30 min at 340 nm. Immunoﬂuorescence confocal microscopy. A549 cells were seeded sparsely in eight-well chamber slides and treated with or without identiﬁed compounds for 24 h. Following treatment, cells were ﬁxed with cold methanol at 201C for 15 min, washed three times with phosphate-buffered saline (PBS) and blocked with 1% PBS plus 0.1% Triton X-100 for 30 min at 371C. Microtubules were detected by incubation with a monoclonal anti-b-tubulin at 371C for 1 h. Then, the cells were washed with PBS and incubated with an FITC-conjugated anti-mouse IgG antibody. Nuclei were stained with DAPI, and microtubule distribution images were acquired with a Leica TCS SP2 Confocal Spectral Microscope (Leica, Wetzlar, Germany). Taken together, our results indicated that MPT0B271 could be an effective orally administered microtubule-destabilizing agent with poor susceptibility to P-gp that possesses potent cytotoxic activity. We demonstrated that MPT0B271 inhibits tubulin polymerization, leading to mitotic arrest of the cell cycle and subsequently triggering apoptotic signaling path- ways in human NSCLC cells. In addition, the antitumor growth effect of MPT0B271 in combination with the EGFR inhibition effect of erlotinib is more potent than the same drug used alone in NSCLC cells in vitro and in vivo. Discussion Both the death receptor- mediated pathway and the mitochondrial-mediated pathway result in the activation of caspase-3, and subsequently lead to the cleavage of PARP.36 Our results showed that treatment with MPT0B271 results in a signiﬁcant induction of caspase-3 and -7 activation, but fails to promote the cleavage of caspase- 8 and -9 in A549 cells. Caspases-8 and -9 probably have a minor role in apoptosis in A549 cells. Mcl-1 is required to prevent apoptosis induced by intrinsic and extrinsic path- ways.37 Treatment with MPT0B271 reduced the expression of Mcl-1 in A549 cells. In addition, cell apoptosis was rescued by exogenous Mcl-1 overexpression, which signiﬁes the role of Mcl-1 degradation in MPT0B271-induced cell apoptosis. compounds interacting with MDR proteins. The calcein generated from calcein AM by esterase in a viable cell emits strong green ﬂuorescence.24,29,30 Our results demonstrated compounds interacting with MDR proteins. The calcein generated from calcein AM by esterase in a viable cell emits strong green ﬂuorescence.24,29,30 Our results demonstrated Cell Death and Disease Treatment of MPT0B271 alone or in combination with erlotinib in human NSCLC A-C Tsai et al 9 9 Sigma (St. Louis, MO, USA). 4’,6-diamidino-2-phenylindole (DAPI) was purchased from Roche Molecular Biochemicals (Cat. 10236276001, Mannheim, Germany). Antibodies against cdc2 (Thr161) (Cat. 9114), cdc2 (Tyr15) (Cat. 4539), Aurora A (Cat. 4718), Aurora B (Cat. 3094), PLK1 (Cat. 4513), p53 (Sser15) (Cat. 2527), Bid (Cat. 2002), caspase-8 (Cat. 9746), and caspase-9 (Cat. 9502) were purchased from Cell Signaling Technology (Beverly, MA, USA). Antibodies against cyclin B1 (Cat. sc-594), cdc25C (Cat. sc-13138), cdc2 (Cat. sc-54), Bcl-2 (Cat. sc- 7382), Bcl-xL (Cat. sc-8392), Bax (Cat. sc-7480), Mcl-1 (Cat. sc-819), Bak (Cat. sc-832), and PARP (Cat. sc-7150) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). The antibody against caspase-3 (Cat. IMG- 144A) was purchased from Imgenex (San Diego, CA, USA), and the antibody to MPM2 (Ser161/Thr97) (Cat. 05-368) was obtained from Upstate Biotechnology Inc. (Temecula, CA, USA). Antibodies to STAT3 (Cat. 610189), p53 (Cat. 554166), and caspase-7 (Cat. 556541) were purchased from BD Biosciences (San Jose, CA, USA). The antibody to STAT3 (Tyr705) (Cat. 2236-1) was obtained from Epitomics (Burlingame, CA, USA), and the antibody to actin (Cat. MAB1501) was purchased from Millipore (Temecula, CA, USA). The p53 si RNA (Cat. 106141) was obtained from Invitrogen (Carlsbad, CA, USA). STAT3 is constitutively activated in most tumor cells and persistent STAT3 activation has been associated with both chemoresistance and radioresistance. Discussion The acetyl STAT3 translocates into the nucleus where it promotes cell prolifera- tion and survival through transactivation of related genes. In addition, STAT3 can also associate with microtubules and mitochondria and regulate cell behavior. It was found that microtubule-targeting agents decrease the tyrosine phos- phorylation-induced activation of STAT3 (Tyr705) in tumor cells, inhibit the expression of STAT3 target genes, and correlate with its cytotoxic effect.38–40 In our study, we found that MPT0B271 speciﬁcally reduced STAT3 phosphorylation at Tyr705. In A549 cells transfected with the STAT3-C plasmid, a minor reversal of cell cytotoxicity was observed, suggesting that STAT3 may have a minor role in MPT0B271- induced apoptosis in A549 cells and that other mechanisms may be involved in this phenomenon. Cell culture. The human NSCLC cell lines A549, NCI-H1299, and NCI-H226; the human pancreatic adenocarcinoma cell line AsPC-1; the human colorectal cancer cell line HCT-116; the human hepatoma cell line Hep3B; the human breast carcinoma cell line MDA-MB-231; the human prostate cancer cell line PC-3; the human ovarian carcinoma cell line SKOV3; and the human promyelocytic leukemia cell line HL60 were obtained from American Type Culture Collection (ATCC) (Manassas, VA, USA). The NCI/ADR-RES cell line was obtained from the DTP Human Tumor Cell Line Screen (Developmental Therapeutics Program, NCI). Cells were maintained in RPMI-1640 medium or DMEM with 10% fetal bovine serum (FBS) and penicillin (100 U/ml)/ streptomycin (100 mg/ml)/amphotericin (0.25 mg/ml) at 371C in a humidiﬁed incubator with 5% CO2. y p The overexpression of EGFR has previously been reported in a wide range of human malignancies including NSCLC and is a factor that is indicative of poor prognosis. The oral EGFR tyrosine kinase inhibitor Erlotinib (Terceva, Roche, Mannheim, Germany) reversibly binds to the intracellular domain of EGFR and blocks autophosphorylation of EGFR with subsequent suppression of the downstream signaling pathways that causes uncontrolled tumor cell growth and proliferation. Although erlotinib has proven efﬁcacy in metastatic NSCLC and has been reported to confer a survival beneﬁt for advanced NSCLC patients harboring EGFR mutations, resistance to erlotinib also occurs and reduces its efﬁcacy.41–43 To overcome the problem of resistance, we combined the microtubule-binding agent MPT0B271 with erlotinib to increase the antitumor effects in the erlotinib-resistant human NSCLC cell line A549. Conﬂict of Interest The authors declare no conﬂict of interest. Acknowledgements. This work was supported by research grants from the National Science Council of Taiwan NSC 99-2320-B-038-020-MY3 and Taipei Medical University TMU101-AE1-B48. Acknowledgements. This work was supported by research grants from the National Science Council of Taiwan NSC 99-2320-B-038-020-MY3 and Taipei Medical University TMU101-AE1-B48. P-gp activity assay. Rh-123 and a Multi-Drug Resistance Assay kit (Calcein AM, Cayman Chemical, Ann Arbor, MI, USA), both of which include ﬂuorescent substrates of P-gp, were used to measure the transport activity of P-gp. For the Rh-123 assay, NCI/ADR-RES cells were pretreated with or without the indicated compounds in culture medium in the dark at 371C for 1 h and then co-treated with Rh-123 for an additional hour. After Rh-123 accumulation, the cells were washed with ice-cold PBS and collected by trypsinization. The intracellular ﬂuorescence of Rh-123 was measured using a FACScan ﬂow cytometer (BD Biosciences). The calcein AM efﬂux assay was performed according to the manufacturer’s protocol. In brief, NCI/ADR-RES cells were seeded in 96-well culture plates at a density of 5 104 cells/well and then treated with or without the experimental compounds. Calcein AM was added to each well, and the ﬂuorescence was measured at an excitation wavelength of 485 nm and an emission wavelength of 535 nm. 1. Hawkins T, Mirigian M, Selcuk Yasar M, Ross JL. Mechanics of microtubules. J Biomech 2010; 43: 23–30. 1. Hawkins T, Mirigian M, Selcuk Yasar M, Ross JL. Mechanics of microtubules. J Biomech 2010; 43: 23–30. 2. Stanton RA, Gernert KM, Nettles JH, Aneja R. Drugs that target dynamic microtubules: a new molecular perspective. Med Res Rev 2011; 31: 443–481. 2. Stanton RA, Gernert KM, Nettles JH, Aneja R. Drugs that target dynamic microtubules: a new molecular perspective. Med Res Rev 2011; 31: 443–481. 3. Schmidt M, Bastians H. Mitotic drug targets and the development of novel anti-mitotic anticancer drugs. Drug Resist Updat 2007; 10: 162–181. 3. Schmidt M, Bastians H. Mitotic drug targets and the development of novel anti-mitotic anticancer drugs. Drug Resist Updat 2007; 10: 162–181. 4. Mollinedo F, Gajate C. Microtubules, microtubule-interfering agents and apoptosis. Apoptosis 2003; 8: 413–450. 4. Mollinedo F, Gajate C. Microtubules, microtubule-interfering agents and apoptosis. Apoptosis 2003; 8: 413–450. 5. Perez EA. Microtubule inhibitors: Differentiating tubulin-inhibiting agents based on mechanisms of action, clinical activity, and resistance. Mol Cancer Ther 2009; 8: 2086–2095. Cell-cycle analysis. Conﬂict of Interest Following treatment, the cells were harvested by trypsinization and centrifuged. The pellets were ﬁxed with 70% (v/v) ethanol at 201C overnight. The cells were again centrifuged at 2000 r.p.m. for 3 min, and the supernatant was discarded. Pellets were washed once with ice-cold PBS and resuspended in 0.1ml phosphate/citric acid buffer (0.2 M Na2HPO4, 0.1 M citric acid, pH 7.8) for 30 min at room temperature. After incubation, the cells were stained with a PI working solution (0.1% Triton X-100, 100 mg/ml RNase and 80 mg/ml PI in PBS) in the dark. The ﬂuorescence of the samples was measured with a FACScan ﬂow cytometer with the CellQuest software (Becton Dickinson, San Jose, CA, USA). 6. Gottesman MM. Mechanisms of cancer drug resistance. Annu Rev Med 2002; 53: 615–627. 7. Dumontet C, Sikic BI. Mechanisms of action of and resistance to antitubulin agents: microtubule dynamics, drug transport, and cell death. J Clin Oncol 1999; 17: 1061–1070. 8. Fojo T, Menefee M. Mechanisms of multidrug resistance: the potential role of microtubule- stabilizing agents. Ann Oncol 2007; 18(Suppl 5): v3–v8. 9. Kavallaris M. Microtubules and resistance to tubulin-binding agents. Nat Rev Cancer 2010; 10: 194–204. 10. Murray S, Briasoulis E, Linardou H, Bafaloukos D, Papadimitriou C. Taxane resistance in breast cancer: mechanisms, predictive biomarkers and circumvention strategies. Cancer Treat Rev 2012; 38: 890–903. Nuclear extracts and western blot analysis. Cytoplasmic and nuclear protein extraction was performed as described previously.44 For western blot analysis, cell lysates were prepared, and proteins were separated by 7.5–15% SDS-PAGE, transferred onto PVDF membrane, and then immunoblotted with speciﬁc antibodies. Proteins were visualized with an ECL detection system (GE Healthcare Bio-Sciences, Pittsburgh, PA, USA). 11. Gelderblom H, Verweij J, Nooter K, Sparreboom A. Cremophor EL: the drawbacks and advantages of vehicle selection for drug formulation. Eur J Cancer 2001; 37: 1590–1598. 12. Koolen SL, Beijnen JH, Schellens JH. Intravenous-to-oral switch in anticancer chemotherapy: a focus on docetaxel and paclitaxel. Clin Pharmacol Ther 2010; 87: 126–129. 13. Haftchenary S, Avadisian M, Gunning PT. Inhibiting aberrant Stat3 function with molecular therapeutics: a progress report. Anticancer Drugs 2011; 22: 115–127. 14. Yin ZJ, Zhang Y, Li Y, Lv T, Liu J, Wang X. Prognostic signiﬁcance of STAT3 expression and its correlation with chemoresistance of non-small cell lung cancer cells. Acta Histochem 2012; 114: 151–158. Evaluation of apoptosis. Discussion The inhibition rate on cell proliferation (GI50) as a function of test drug concentration was calculated for each well as 100 [(A515treated cells A515time)]. Once the tumor size was B100 mm3, mice were allocated at random to treatment groups: vehicle (0.5% CMC/0.1% Tween-80 in ddH2O); 5 or 10 mg/kg of MPT0B271 once a day (qd) or 20 mg/kg once every 2 days (q2d) by oral gavage; 25 mg/kg of erlotinib qd; or 25 mg/kg of erlotinib qd in combination with 20 mg/kg of MPT0B271 q2d. Caliper measurements were used to calculate tumor volume (V, mm3) using the formula V ¼ lw2/2, with l being the length and the w being the width of the tumor. All animal studies were conducted in accordance with the guidelines of the Animal Care and Use Committee at National Taiwan University. Statistical analysis. Results are expressed as the mean±S.E. for the indicated number of separate experiments. Means were assessed for signiﬁcant differences using t-test and P-valueso0.05 were considered as signiﬁcant. MTT assay. Cells were incubated in the absence or presence of MPT0B271 for 48 h and then incubated with 0.5 mg/ml MTT solution for 1 h at 371C. At the end of the incubation, the purple formazan crystals were dissolved in 100 ml DMSO, and the absorbance was measured at 550 nm using an ELISA reader. The percentage of cell survival was plotted as a percentage to determine the IC50 values (50% growth inhibition). Discussion These ﬁndings suggest that MPT0B271 may foster novel therapeutic strategies for NSCLC. Pharmacokinetic analysis. Male CD-1 (Crl.) mice (n ¼ 4 per group) were obtained from Lasco (Taipei, Taiwan) and used to examine the pharmacokinetics (PK) of the MPT0B271 compound. MPT0B271 was dissolved in polyethylene glycol 400/ethanol/water (60/5/35, v/v) and administered by a single intravenous tail vein injection at 2.0 mg/kg. For oral dosing, MPT0B271 was suspended in 0.5% methylcellulose (MC)/water and given via oral gavage with a dosing volume of 0.26 ml per animal (10 mg/kg). The animals were fasted for 16 h before dosing and allowed to consume standard chow 4 h post dosing. For mice treated via tail vein injection, plasma samples were collected before dosing and then 2, 5, 15, and 30 min and 1, 1.5, 2, 4, 6, 9, 24, and 27 h after dosing. For mice treated via oral administration, blood samples were collected before dosing and then 15 and 30 min and 1, 1.5, 2, 4, 6, 9, 24, and 27 h after dosing. The plasma concentrations of MPT0B271 were measured by LC-MS/MS (API4000; PE Sciex, Concord, ON, Canada) with a reverse-phase ODS column. The PK parameters were calculated from the mean plasma concentrations with the WinNonlin Professional program (version 5.2, Pharsight Corp., Mountain View, CA, USA). Materials and Methods Reagents. SRB (230162; Sigma-Aldrich, St. Louis, MO, USA), MTT (M2128; Sigma-Aldrich), PI (P4170; Sigma-Aldrich), FITC-conjugated anti-mouse IgG (F9295; Sigma-Aldrich), and all other chemical reagents were obtained from Cell Death and Disease Treatment of MPT0B271 alone or in combination with erlotinib in human NSCLC A-C Tsai et al Treatment of MPT0B271 alone or in combination with erlotinib in human NSCLC A-C Tsai et al 10 SRB assay. All cells were seeded in 96-well culture plates at a density of 3–5 103 cells/well. After attachment, cells were ﬁxed with 10% trichloroacetic acid (TCA) to provide a measurement of the cell population at the time of drug addition. The following day, cells were treated with vehicle (0.1% DMSO) or an increasing gradient concentration of the indicated compounds for 48 h, after which cells were ﬁxed with 10% TCA and stained with 0.4% (w/v) SRB dissolved in 1% acetic acid. The protein-bound dye was subsequently extracted with 10 mM trizma base to determine the absorbance at a wavelength of 515 nm. Conﬂict of Interest Cell Death and Disease Treatment of MPT0B271 alone or in combination with erlotinib in human NSCLC A-C Tsai et al 11 22. Ghosh G, Lian X, Kron SJ, Palecek SP. Properties of resistant cells generated from lung cancer cell lines treated with EGFR inhibitors. BMC Cancer 2012; 12: p 95. 37. Thomas LW, Lam C, Edwards SW. Mcl-1; the molecular regulation of protein function. FEBS Lett 2010; 584: 2981–2989. 38. Aggarwal BB, Sethi G, Ahn KS, Sandur SK, Pandey MK, Kunnumakkara AB et al. Targeting signal-transducer-and-activator-of-transcription-3 for prevention and therapy of cancer: modern target but ancient solution. Ann NY Acad Sci 2006; 1091: 151–169. 23. Wang M, Zhao J, Zhang LM, Li H, Yu JP, Ren XB et al. Combined Erlotinib and Cetuximab overcome the acquired resistance to epidermal growth factor receptors tyrosine kinase inhibitor in non-small-cell lung cancer. J Cancer Res Clin Oncol 2012; 138: 2069–2077. cer: modern target but ancient solution. Ann NY Acad Sci 2006; 1091: 151–16 39. Walker SR, Chaudhury M, Nelson EA, Frank DA. Microtubule-targeted chemotherapeutic agents inhibit signal transducer and activator of transcription 3 (STAT3) signaling. Mol Pharmacol 2010; 78: 903–908. 24. Szakacs G, Jakab K, Antal F, Sarkadi B. Diagnostics of multidrug resistance in cancer. Pathol Oncol Res 1998; 4: 251–257. 25. Stewart ZA, Tang LJ, Pietenpol JA. Increased p53 phosphorylation after microtubule disruption is mediated in a microtubule inhibitor- and cell-speciﬁc manner. Oncogene 2001; 20: 113–124. 40. Liu H, Tekle C, Chen YW, Kristian A, Zhao Y, Zhou M et al. B7-H3 silencing increases paclitaxel sensitivity by abrogating Jak2/Stat3 phosphorylation. Mol Cancer Ther 2011; 10: 960–971. 26. Dumontet C, Jordan MA. Microtubule-binding agents: a dynamic ﬁeld of cancer therapeutics. Nat Rev Drug Discov 2010; 9: 790–803. 41. Sangodkar J, Katz S, Melville H, Narla G. Lung adenocarcinoma: lessons in translation from bench to bedside. Mt Sinai J Med 2010; 77: 597–605. 27. Gelderblom H, Verweij J, Nooter K, Sparreboom A. Cremophor EL: the drawbacks and advantages of vehicle selection for drug formulation. Eur J Cancer 2001; 37: 1590–1598. 42. Francis H, Solomon B. The current status of targeted therapy for non-small cell lung cancer. Intern Med J 2010; 40: 611–618. 28. Nieto Montesinos R, Be´duneau A, Pellequer Y, Lamprecht A. Delivery of P-glycoprotein substrates using chemosensitizers and nanotechnology for selective and efﬁcient therapeutic outcomes. J Control Release 2012; 161: 50–61. 43. Uribe MC, Grier HJ, Parenti LR. Conﬂict of Interest Quantitative oligonucleosomal DNA fragmenta- tion was used to determine the ability of the identiﬁed compound to induce apoptosis. The Cell Death Detection ELISAPLUS kit (Roche) was used according to the manufacturer’s protocol. 15. Germain D, Frank DA. Targeting the cytoplasmic and nuclear functions of signal transducers and activators of transcription 3 for cancer therapy. Clin Cancer Res 2007; 13: 5665–5669. STAT3 kit. The level of STAT3 phosphorylation in response to MPT0B271 treatment was determined using the PathScan Phospho-Stat3 (Try705) Sandwich ELISA kit (Cell Signaling Technology) according to the manufacturer’s instructions. The absorbance was read within 30min at 405 nm using a microplate ELISA reader. 16. Aggarwal BB, Kunnumakkara AB, Harikumar KB, Gupta SR, Tharakan ST, Koca C et al. Signal transducer and activator of transcription-3, inﬂammation, and cancer: how intimate is the relationship? Ann NY Acad Sci 2009; 1171: 59–76. 17. Walker SR, Chaudhury M, Nelson EA, Frank DA. Microtubule-targeted chemotherapeutic agents inhibit signal transducer and activator of transcription 3 (STAT3) signaling. Mol Pharmacol 2010; 78: 903–908. Transient transfection. The siRNA against p53 and the negative control were purchased from Invitrogen. The Mcl-1 (25375) and STAT3-C plasmids were purchased from Addgene (Cambridge, MA, USA). A549 cells were grown to 70% conﬂuence in a culture dish, and transfection was performed using Lipofectamine reagent (Invitrogen) according to the manufacturer’s instructions. Following transfection, cells were allowed to recover for 24 h, seeded in 96-well or 6-well plates and then treated for another 48 h. 18. Walker SR, Chaudhury M, Frank DA. STAT3 inhibition by microtubule-targeted drugs: dual molecular effects of chemotherapeutic agents. Mol Cell Pharmacol 2011; 3: 13–19. 19. Seshacharyulu P, Ponnusamy MP, Haridas D, Jain M, Ganti AK, Batra SK. Targeting the EGFR signaling pathway in cancer therapy. Expert Opin Ther Targets 2012; 16: 15–31. 20. Shepherd FA, Rodrigues Pereira J, Ciuleanu T, Tan EH, Hirsh V, Thongprasert S et al. Erlotinib in previously treated non-small-cell lung cancer. N Engl J Med 2005; 353: 123–132. 21. Moore MJ, Goldstein D, Hamm J, Figer A, Hecht JR, Gallinger S et al. Erlotinib plus gemcitabine compared with gemcitabine alone in patients with advanced pancreatic cancer: a phase III trial of the National Cancer Institute of Canada Clinical Trials Group. J Clin Oncol 2007; 25: 1960–1966. A549 xenograft models. A549 cells (2 106 cells/mice) in 0.2 ml culture medium were inoculated subcutaneously (sc) in nu/nu mice (male, 5–6 weeks). Conﬂict of Interest Ovarian structure and oogenesis of the oviparous goodeids Crenichthys baileyi (Gilbert, 1893) and Empetrichthys latos Miller, 1948 (teleostei, Cyprinodontiformes). J Morphol 2012; 273: 371–387. 43. Uribe MC, Grier HJ, Parenti LR. Ovarian structure and oogenesis of the oviparous goodeids Crenichthys baileyi (Gilbert, 1893) and Empetrichthys latos Miller, 1948 (teleostei, Cyprinodontiformes). J Morphol 2012; 273: 371–387. p 29. Canitrot Y, Lautier D. [Use of rhodamine 123 for the detection of multidrug resistance]. Bull Cancer 1995; 82: 687–697. 44. Tsai AC, Pan SL, Sun HL, Wang CY, Peng CY, Wang SW et al. CHM-1, a new vascular targeting agent, induces apoptosis of human umbilical vein endothelial cells via p53-mediated death receptor 5 up-regulation. J Biol Chem 2010; 285: 5497–5506. 30. Shapiro AB, Ling V. The mechanism of ATP-dependent multidrug transport by P-glycoprotein. Acta Physiol Scand Suppl 1998; 643: 227–234. 31. Gavet O, Pines J. Activation of cyclin B1-Cdk1 synchronizes events in the nucleus and the cytoplasm at mitosis. J Cell Biol 2010; 189: 247–259. Cell Death and Disease is an open-access journal published by Nature Publishing Group. This work is licensed under a Creative Commons Attribution 3.0 Unported License. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/3.0/ 32. Coulonval K, Kooken H, Roger PP. Coupling of T161 and T14 phosphorylations protects cyclin B-CDK1 from premature activation. Mol Biol Cell 2011; 22: 3971–3985. 33. Malumbres M. Physiological relevance of cell cycle kinases. Physiol Rev 2011; 91: 973–1007. 34. Andrews PD, Knatko E, Moore WJ, Swedlow JR. Mitotic mechanics: the auroras come into view. Curr Opin Cell Biol 2003; 15: 672–683. 35. Lens SM, Voest EE, Medema RH. Shared and separate functions of polo-like kinases and aurora kinases in cancer. Nat Rev Cancer 2010; 10: 825–841. 36. Fan TJ, Han LH, Cong RS, Liang J. Caspase family proteases and apoptosis. Acta Biochim Biophys Sin (Shanghai) 2005; 37: 719–727. formation accompanies this paper on Cell Death and Disease website (http://www.nature.com/cddis) Cell Death and Disease Cell Death and Disease
https://openalex.org/W3186836918	https://www.research.unipd.it/bitstream/11577/3396325/1/sustainability-13-08276.pdf	English	null	Collaborations in Environmental Initiatives for an Effective “Adaptive Governance” of Social–Ecological Systems: What Existing Literature Suggests	Sustainability	2,021	cc-by	18,333	Elena Andriollo 1,, Alberto Caimo 2, Laura Secco 1 and Elena Pisani 1 1 Dipartimento Territorio e Sistemi Agro-Forestali (TESAF), Università degli Studi di Padova, Via dell’Università, 16, 35020 Legnaro, PD, Italy; laura.secco@unipd.it (L.S.); elena.pisani@unipd.it (E.P.) 2 School of Mathematical Sciences, Technological University Dublin, D07 ADY7 Dublin, Ireland; Alberto.Caimo@tudublin.ie g y p p 2 School of Mathematical Sciences, Technological University Dublin, D07 ADY7 Dublin, Ir Alberto.Caimo@tudublin.ie Correspondence: elena.andriollo.1@phd.unipd.it Abstract: Moving from the scientific literature on the evaluation of environmental projects and pro- grams, this study identifies how and under which conditions collaborations in environmentally sus- tainable projects are considered effective for the adaptive governance of SES. The method adopted is a systematic literature review based on the quantitative and qualitative analysis of 56 articles se- lected through specific queries on the SCOPUS database and published from 2004 to 2020. Results of the quantitative analysis identify conditions able to evaluate collaborations, highlighting the need to adopt a transdisciplinary approach analysing both social and ecological challenges and assessing both social and ecological results. Moreover, they suggest preferring using primary data involving multi-sector and multi-scale actors and enlarging the geographical context to the most vulnerable countries. The results of the qualitative analysis provide specific recommendations for collabora- tions being effective when related to communication, equity, foresight, and respect, which need to be further strengthened by all actors. Multiplicity in visions and approaches should be seen as a resource able to stimulate creativity in social arrangements and environmental practices, making collaborations in environmental projects instrumental for the effectiveness of adaptive governance of SES. Citation: Andriollo, E.; Caimo, A.; Secco, L.; Pisani, E. Collaborations in Environmental Initiatives for an Effective “Adaptive Governance” of Social–Ecological Systems: What Existing Literature Suggests. Sustainability 2021, 13, 8276. https://doi.org/10.3390/su13158276 Keywords: collaboration; adaptive governance; sustainability transformations; social–ecological systems; evaluation; systematic literature review Received: 28 June 2021 Accepted: 20 July 2021 Published: 24 July 2021 Collaborations in Environmental Initiatives for an Effective “Adaptive Governance” of Social–Ecological Systems: What Existing Literature Suggests Elena Andriollo 1,*, Alberto Caimo 2, Laura Secco 1 and Elena Pisani 1 Citation: Andriollo, E.; Caimo, A.; Secco, L.; Pisani, E. Collaborations in Environmental Initiatives for an Effective “Adaptive Governance” of Social–Ecological Systems: What Existing Literature Suggests. Sustainability 2021, 13, 8276. https://doi.org/10.3390/su13158276 1. Introduction Human activities are exerting an increasing impact on the environment at all scales, from local to global, endangering the conditions of ecosystems [1–11]. Emergencies that global society is fighting nowadays are evidence of this close connection. Specifically, the COVID-19 pandemic has reinforced this awareness within the scientific community [12] and has probably diffused it to a broader public [13], calling for real sustainable develop- ment action [14,15]. Nevertheless, at present, sustainable development is still far from be- ing achieved: “The world today is not sustainable, not resilient and not fair for the major- ity of mankind” [14] (p. 1). Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. Copyright: © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses /by/4.0/). Complexities and uncertainties characterizing both environmental and social chal- lenges limit the implementation of activities able to effectively catalyze sustainable trans- formations [16]. In particular, the governance of the environment is challenging because many natural resources are shared among multiple competing actors, provoking conflicts. That is why collaboration is proposed as a promising approach able to address such issues [17]. However, collaboration improves the governance of natural resources if it is effective [18]. Otherwise, collaboration could be seen as a sort of panacea solution that can have no www.mdpi.com/journal/sustainability Sustainability 2021, 13, 8276. https://doi.org/10.3390/su13158276 Sustainability 2021, 13, 8276 2 of 29 value or even make counterproductive effects [19,20]. To overcome this limitation, at pre- sent, the existing literature reveals an urgent need to provide additional knowledge con- cerning the effectiveness of collaborations both in terms of evaluation approaches [18,21– 23] and conditions able to foster them [24]. Hence, this study identifies how and under which conditions collaborations are considered instrumental for an effective “adaptive governance” in terms of sustainable transformations in Social–Ecological Systems (SES). The study analyses collaborative relations among different actors involved in environ- mental programs and projects as analyzed in the scientific literature on environmental evaluation. Specifically, this study aims to provide additional knowledge for (i) improv- ing evaluations of collaborations in future and (ii) providing guidelines for actors to foster effective collaborations. Identifying and classifying findings emerging from evaluations of real experiences allows an understanding of why some collaborations are effective while others fail or collapse [18,25]. 1. Introduction The analysis of project and program evaluations, which have been scrutinized through peer-review scientific articles, is at the core of this study. This is motivated by the awareness that evaluations are instrumental to increase the effectiveness of environmental actions, adjusting them to new needs emerging over time through the stimulation of the learning-by-doing process, which identifies previous failures and successes and high- lights current needs [26,27]. Specifically, evaluation, and more specifically, self-evalua- tion, could enhance the performance of future initiatives—through an individual and col- lective practice of reflection on the process undertaken during the action implementa- tion—if its results pave the way to changing community routines and individual and col- lective practices and behaviors [28]. Its findings, indeed, can help policymakers to reform or re-design policy instruments but also be helpful for practitioners and, generally, all stakeholders to identify the most relevant and critical aspects for promoting and making valuable and successful their entrepreneurial and social initiatives in the environmental realm [29]. In this perspective, the role of the evaluation is further strengthened if the ini- tiatives can have a clear transformative impact and become utilized and available to the entire society by proposing evidence-based examples on transition practices for sustaina- ble transformations [16]. Effective sustainable transformations are fostered by the capacity of all actors com- posing society to respond to change through adaptation [30]. This can be achieved through an ongoing individual and collective adjustment aiming at revising environmen- tal activities [31,32]. Accordingly, adaptive governance has been defined as the set of in- teractions between actors, networks, organizations, and institutions that aims to facilitate transformations to achieve the desired state for SES [32,33]. The SES concept [34] high- lights that nature and society coevolve through a reciprocal adaptation process based on interdependencies [8,35–37]. In particular, it clarifies that society—intended as people, communities, economies, and cultures [38]—is part of the biosphere and it is entirely de- pendent on nature [8]. In this paper, we focus on the societal component, and therefore on collaborations amongst humans and their organizations, while exploring in detail the eco- logical component and networks remains out of the scope. We focus on collaborations among different types of social actors because sustaina- bility transformations are usually multi-actor and multi-level processes [39] that are char- acterized by differences in interests, perspectives, needs, knowledge, and resources among stakeholders, leading to possible conflicts, e.g., [40–42]. 1. Introduction The scientific literature agrees in considering collaborative relationships the most suitable means to support sus- tainability transformations [21,43–47]. The literature reports examples of projects which are characterized by good performances in term of effectiveness due to collaboration be- tween different types of stakeholders, such as in biodiversity conservation projects [42], land use planning [48], and protected areas management projects [49]. Collaboration can be seen as “a set of organizational and interpersonal relationships shaped by the nature of the problems being addressed, the predispositions and capabilities of key actors, and the characteristics of the places in which the problems occur” [43] (p. 85). Collaborative Sustainability 2021, 13, 8276 3 of 29 relationships are characterized by strong interactions between all types of actors involved in the process and by trust and honesty [50]. When they are characterized by accountabil- ity and transparency, they contribute to building knowledge, solving conflicts, develop- ing trust or trustworthiness among actors, and connecting different types of actors and sectors that previously worked in isolation to identify common solutions [20,41,51,52]. Collaboration concretely happens through the creation of partnerships. Partnerships arise when different actors share their resources in order to achieve a common goal. Accord- ingly, creating collaborative partnerships composed of multiple actors is considered an essential tool to face uncertainties and complexities characterizing environmental chal- lenges [53]. g The needs, ideas, and actions that emerge from collaborative relationships trigger the coevolving process between society and nature by establishing new social arrangements [33], intended as new roles and interactions of actors [54]. Hence, evaluations of new col- laborative interactions emerged from adaptive governance initiatives, and when scruti- nized through scientific articles, could identify aspects able to improve their effectiveness and encourage sustainability transformations with a consequent improvement in the quality of SES. q y The paper is organized into five sections. After this introduction, the theoretical framework is presented in Section 2, then the materials and methods are specified in Sec- tion 3. The quantitative and qualitative results are described in Section 4 and further dis- cussed in Section 5 with concluding remarks in Section 6. 2.1. On Key Basic Concepts 2.1. On Key Basic Concepts This article bases its foundations on the theoretical concepts of sustainable transfor- mations and adaptive governance. Sustainable transformations refer to changes in social and environmental interactions and feedback in all dimensions of SES by considering resilience and adaptation [30,55]. Transformations are recognized as deliberative actions activated intentionally by actors to realize a significant change (i.e., radical and non-linear social changes able to cross thresholds into new development trajectories—[55,56]) to achieve adaptation in SES [57]. Transformations can be different in focus and can be distinguished between ecological (e.g., changes of landscape, ecosystem services, and assemblages of species) and social (e.g., new values, norms, institutions, changes in governance arrangements and everyday practices), with a continuous interplay between these two sets of transformations, which depends on each other [58]. Focusing mainly on social transformations, the assumption at the basis of this study is that changes in social values, rules, and knowledge may impact decisions of individuals and organizations, fostering transformative adaptations based on shared solutions and learning by improving SES quality [31]. Adaptive governance integrates the concepts of transformations, SES, and govern- ance [33]. The governance concept refers as the set of rules, structures, processes, and tra- ditions that determine how people make decisions, share power, exercise responsibilities, and ensure accountability [32,59]. Adaptive governance of SES is, indeed, characterized by participation, experimentation, and collective learning of the different stakeholders in- volved in diverse phases of collaborative activities, such as the identification, formulation, implementation, and evaluation of environmental policies, programs, or initiatives [35]. Adaptive governance reaches its effectiveness if it is fit-for-purpose, that is, when “(i) its structure enables multiple actors to purposely guide, control, manage or steer societies through network structures that fit with their social and ecological context, (ii) its pro- cesses fit with both the network structures in which they take place and the purposes for which they are being used” [19] (p. 76). Consequently, adaptive governance should “(i) provide information (science and local knowledge); (ii) deal with conflict; (iii) induce rule Sustainability 2021, 13, 8276 4 of 29 compliance; (iv) provide infrastructure for capacity building; and (v) be prepared for change” [33] (p. 4). compliance; (iv) provide infrastructure for capacity building; and (v) be prepared for change” [33] (p. 4). 2.2. On Collaborations 5 of 29 Sustainability 2021, 13, 8276 5 of 29 An SES is constituted by interdependent social and ecological systems whose pecu- liarities are due to their specific context (grey down arrows). If the ecological system can be conceptualized as an interdependent system of organisms or biological units [66] (the green nodes connected through ties in Figure 1), the actors that constitute the social sys- tem (the red nodes connected through social relations) could be defined as individuals or organizations intended to generate changes through activities that have environmental impacts [67]. The two systems, i.e., social and ecological, are connected to each other through social and ecological interactions occurring at multiple levels of adaptive govern- ance [68] where individuals and organizations exert a pressure on the ecological compo- nents of the SES and, vice versa, the induced changes on the environment influence actions (the orange arrows). These interactions influence both the flows among resources com- posing the ecological system (the green lines) and the relations within the social system (the red lines). Effective collaborations in the adaptive governance of SES require that actors guide, control, manage, and steer environmental resources by considering both components, so- cial and ecological. The literature recognizes that by increasing the social connectivity in SES, collaborative activities can improve effective management of the ecological compo- nent through the creation of flexible connections among stakeholders formalized in part- nerships [20,69]. Connections require sharing of material and non-material resources, fa- cilitating trust-building relations needed to resolve conflicts [17]. Connections can sustain adaptation and trigger sustainability transformations [8,30]. By identifying, formulating and implementing environmental project activities (P), actors can concretize environmen- tal collaborations based on adaptive governance that are able to synergically consider both the social and the ecological systems (blue arrow), fostering governance activities that could be more fit-for-purpose [19] in producing outputs, outcomes, and impacts ([70] de- fines outputs as the tangible results made by activities that are relevant for the achieve- ment of outcomes. Outcomes are defined as likely or achieved short-term or medium-term effects. Impacts are defined as positive or negative long-term effects produced by activi- ties) (Figure 1) [71,72]. The results of projects could negatively or positively affect the con- text where they act and both the social and the ecological systems (grey arrows). 2.2. On Collaborations Analyzing how and in which conditions collaborations contribute to the achievement of effectiveness in adaptive governance processes requires focusing on the behaviours, decisions, and activities at the individual and collective levels which determine the effects on the biosphere [8]. Following the adaptive governance concept, the literature on sus- tainability transformations recognizes the critical role played by individuals and their in- teractions in social transformations, meaning “a set of recognizable activities and attitudes used by an actor to address the recurring situation” [60] (p. 49). The role appears because of interactions between different social groups and implies expected behaviors, rights, and duties [61]. Accordingly, actors are not passive rule-followers, but they can be active agents in systemic changes, i.e., changes in the institutional structure such as thinking, everyday habits, management practices, and resource flows [30]. Actors can exert power and influence the magnitude and effectiveness of transformations through their agency [30,60]. Specifically, [62] identifies four actor categories involved in sustainability trans- formations: the State, market actors, community, and the third sector (e.g., labor unions, NGOs, and science). Different features typify them in the following axes: (i) informal– formal, (ii) profit–non-profit, and (iii) public–private. The State is formal, public, and not- for-profit; the market is formal, private, and for-profit; the community is informal, private, and not-for-profit; and the Third Sector is conceptualized as an intermediary form be- tween the three axes [37], allowing the inclusion of different organizational forms such as social entrepreneurs, social enterprises, and cooperative organizations. Collaborative interactions between these different typologies of actors create new hy- brid forms of governance and evidence the change of the conventional role attributed to a specific actor needed to compensate for limitations of other social agents [60,63], encour- aging creativity and, consequently, the development of experimentations through the identification of new ideas, innovative organizational models, new social and environ- mental practices, novel arrangements, and agreements that potentially could contribute to the achievement of sustainability [30,64]. Moreover, interacting actors define and guide governance processes necessarily impacting (positively or negatively) on nature because they are related transversally with natural components of SES through their decisions and activities [8,22,65]. Such interactions between society and nature constitute SES [8] and are shown in Figure 1. Figure 1. Result chain of adaptive governance activities in SES. Figure 1. Result chain of adaptive governance activities in SES. 2.2. On Collaborations Moreo- ver, the ongoing learning-by-doing process fostered by evaluations allows identifying im- provements in governance activities through an adaptive cycle as shown by dashed grey arrows (Figure 1) [30]. 3. Materials and Methods To understand how and under which conditions collaborations could contribute to effective adaptive governance of SES, we perform a systematic literature review through both a quantitative and qualitative analysis based on reliable and high-quality evaluations reported in the scientific literature [73]. The systematic review is performed to collect and synthesize pieces of evidence emerging from scientific articles focused on the results of the evaluations of environmental activities and extrapolate knowledge on collaborations in adaptive governance of SES [74]. We opted for a systematic review because it allows summarize existing and fragmented knowledge discussed in multiple scientific articles in order to handle the research questions in a sounder way [75]. Specifically, we want to reorganize the scientific knowledge that emerged from experiences already analyzed and evaluated by the scientific community, focusing on evaluating and fostering collaboration in adaptive governance. Steps constituting the literature review process are reported in Figure 2. 6 of 29 Sustainability 2021, 13, 8276 Figure 2. Schematization of the sequence of steps constituting the literature review. Figure 2. Schematization of the sequence of steps constituting the literature review. 3.2. Quantitative Analysis of Relevant Data The extraction of relevant data for the quantitative analysis, i.e., the third step of sys- tematic literature review, was done through a Sankey diagram. The Sankey diagram is a visual tool able to define a flow from one set of values to another, highlighting their rela- tionships. Flows and quantities are visualized from the size of lines connecting a value to another one, evidencing the magnitude of relationships. Accordingly, the wider the lines are, the larger the quantity of the flow is [86]. Here, the flow visualized by the Sankey diagram represents the coexistence in the same article of multiple attributes used for its classification, which define its peculiarities. Every article is classified through the identi- fication of levels pertaining to six different scales. Thus, we transform qualitative infor- mation to quantitative data (i.e., number of articles in a certain level, and number of rela- tionships between levels of two consecutive scales) in order to better identify what levels are most addressed by evaluations and what are the most recurrent relationships among levels of different scales. We first identify the year of publication of articles, in which journals the articles are published, and in which scientific areas articles are included according to the subject areas specified by the journals. After that, we classify articles based on the following scales de- tailed into different levels: (i) type of evaluations, i.e., Assessment based on indicators or indices, Pure qualitative evaluations, and Integrated evaluations [87]; (ii) scale of intervention of projects or programs evaluated, i.e., Local, Sub-national, Na- tional, International, Global [68]; (iii) geographical localization, i.e., Africa, America, Asia, Europe, Oceania (https://un- stats.un.org/unsd/methodology/m49/, accessed on 1 June 2021); (iv) human pressures on environmental resources, i.e., Agriculture, Forestry, Fishing and hunting, Tourism, Industry, Transport, Urban areas, Waste, Energy, and Climate change [88]; g (v) environmental issues, i.e., Biodiversity, Freshwater, Land and soil, Ocean and coasts, and Air [88]; (vi) sustainability transformations addressed, i.e., Sustainable food, land, water and oceans, Health, well-being and demography, Sustainable cities and communities, En- ergy decarbonization and sustainable industry, Digital revolution for sustainable de- velopment, Education, gender, and inequality [36]. 3.1. Identification and Extraction of Scientific Papers identify evidence related to collaborations through codes reported in MS Excel files and then elaborated through quantitative and qualitative analyses (e.g., [85]). 3.1. Identification and Extraction of Scientific Papers 3.1. Identification and Extraction of Scientific Papers For the extraction of scientific articles, we choose the SCOPUS database, whose pe- culiarities guarantee high quality and reliability. SCOPUS provides the most extensive availability of high-quality journals [76] and articles from around the world, especially on environmental science, and the possibility to have easy access to abstracts for most papers compared to other academic research databases such as Web of Science [77–79]. SCOPUS assures the extraction of reliable data through the analysis of scientific articles subjected to peer review process, compared to other databases characterized by a more extensive coverage such as Google Scholar, whose citations derive from multiple sources, which also includes preprints. [80,81]. Accordingly, SCOPUS can be considered the largest cu- rated abstract and citations database, characterized by a selection process on its contents that contribute to preserving the integrity of science. The reliability of such a database is already demonstrated by using SCOPUS for multiple evaluations, such as national assess- ments, government science policy evaluations, and university rankings [82]. Additionally, SCOPUS better support the implementation of systematic reviews based on key words search than other databases, especially new databases such as Dimensions or Microsoft Academic [83]. To identify articles, we used the following key words combination (string search): “environmental evaluation” AND “governance OR institution” AND “social AND ecological”, in order to gather a collection of scientific articles treating environmental evaluations of programs or projects aiming at fostering sustainability transformations in both the social and the ecological dimensions of SES, with a focus on governance arrangements. Then, we identified the papers that fit the purpose of the research by reading abstracts using a specific set of selection criteria of the abstracts, as proposed by [84]. The selected abstracts have to: (i) deal with social and ecological variables (ii) provide an evaluation of completed environmental programs or projects (iii) describe activities aiming to foster sustainability transformations (iv) be oriented to a governance approach. The reduced numerosity of papers allows us to analyze them deeply by reading the whole text and valorizing every statement. By reading their texts, we classify articles and Sustainability 2021, 13, 8276 7 of 29 identify evidence related to collaborations through codes reported in MS Excel files and then elaborated through quantitative and qualitative analyses (e.g., [85]). 3.3. Qualitative Analysis of Relevant Data The fourth step is the extrapolation of pieces of evidence (statements as reported in the text of the article) on interactions between different types of actors as categorized by [62], e.g., in the case of the article proposed by [41], State actors are the federal and pro- vincial fisheries departments, market actors are local fishers and aquaculture operators, community actors are local and aboriginal communities, and third sector actors are re- search institutions and multiple NGOs. All statements related to pieces of evidence on interactions are collected in an Excel spreadsheet file, clarifying: (i) what are the categories of the actors involved in the relationship (i.e., the State, mar- ket, community, and the third sector); (ii) if and how the relationship has been effective or not in dealing with the environmen- tal challenge in the analyzed SES (e.g., resolution of conflicts around multiple uses of marine space through the development of a new institution [41]). Finally, a qualitative content analysis reviews and summarizes the heterogeneous knowledge by grouping the statements (narrative text) with an equal or similar meaning Finally, a qualitative content analysis reviews and summarizes the heterogeneous knowledge by grouping the statements (narrative text) with an equal or similar meaning Sustainability 2021, 13, 8276 8 of 29 into homogeneous categories aggregated around broad concepts emerging from our in- terpretation of contents reported in articles. 4.1. Selected Papers The selection of articles on the SCOPUS database identifies 194 articles, which are consequently filtered, considering only articles and reviews written in the English lan- guage (147). After analysing abstracts, 56 papers (listed in Appendix A) fit with all the four criteria identified in Section 3 to address the research purposes and are used. 4.2. Quantitative Results The analysis reveals that selected articles are relatively recent, and the oldest is pub- lished in 2004. Figure 3 shows that evaluations of environmental governance activities fitting with the research criteria are mostly published in the last decade, i.e., after 2010, with a maximum value in 2016 (nine articles published). Then, the number of articles reaches stability with five to six papers published every year. Figure 3. Numerosity of articles selected by the systematic literature review per year. Figure 3. Numerosity of articles selected by the systematic literature review per year. As reported in Appendix B, the selected articles are published in several journals and subject areas, which mainly belong to the Environmental Sciences (54 articles), followed by Social Sciences (22), and Agricultural and Biological Sciences (16). In addition, the clas- sification identifies other subject areas such as Medicine (7), Economics, econometrics and finance (9), and Energy (4), evidencing the transdisciplinary nature of the topic we are exploring. p g The Sankey diagram (Figure 4) shows relationships between all the scales and levels used for classification purposes. Each paper can be part of multiple classification scales and levels at the same time. Thus, the total numbers specified for each scale and for each level do not align with the total number of 56 articles. 9 of 29 Sustainability 2021, 13, 8276 Figure 4. Classification of articles by Sankey diagram. Figure 4. Classification of articles by Sankey diagram. Starting from the scale “Type of evaluations” as reported in the articles, we observe that evaluations using indices or indicators create 27 relationships, evaluations using pure qualitative methods create 19 relationships, and evaluations using a combination of par- ticipative approaches and multicriteria assessments create 15 relationships. Moving to the “Scale of intervention”, it is possible to observe that indicator assess- ments and pure qualitative methods are used transversally for all the levels from local to global, while integrated assessments are mostly used in evaluations at a minor scale, mostly sub-national and local. The 77% of relationships constituting the Sankey diagram focuses on program or project activities implemented at the sub-national and local levels. A minor number of relationships focuses on a national (14%) or international scale (3%), and only one article refers to a global scale (it creates five relationships because it relates with all continents). 4.2. Quantitative Results Focusing on the “Geographical localization”, it is possible to observe that studies are mostly localized in developed countries. In fact, the geographical area with the highest number of activities analyzed is Europe with 68 relationships (34%). The review selects articles that analyze initiatives placed in all continents: Africa (8%), America (28%), Asia (17%), and Oceania (13%). However, it reveals that the poorest areas remain understudied (e.g., Sub-Saharan Africa or the Middle East). Observing the “Human pressures” scale, it emerges that Agriculture is the most re- current pressure in terms of relationships (17%), followed by Fishing and Hunting (13%), Industry (11%), Urban Areas (11%), and Climate Change (11%). Forestry (8%), Tourism (8%), Waste Production (8%), and Transport (9%) are less investigated, and Energy re- ceives a little attention (3%). Moving to “Environmental issues”, the analysis reveals that Land and Soil counts 124 relationships (30%), while Freshwater and Biodiversity total 97 and 96 relationships, re- spectively (23% for both). Then, Ocean and Coasts attest 70 relations (17%), followed by Air with 26 relations (6%). 10 of 29 Sustainability 2021, 13, 8276 10 of 29 More specifically, if Agriculture, Climate Change, and Fishing and Hunting seem to be transversal pressures impacting all of the most addressed environmental issues, from the Sankey diagram it emerges that Forestry and Tourism mainly impact on Biodiversity and Land and Soil, while Industry and Urban Areas mainly impact on Freshwater and Land and Soil. Transport, Waste, and Energy production are mainly related to Land and Soil and Freshwater, but it is also possible to appreciate a relevant number of relationships targeting the environmental issue of Air. g g Finally, focusing on the framework proposed by [36] on sustainability transfor- mation, the classification highlights that most of the initiatives relate to the achievement of sustainability in food production, land use, water use, and oceans (115), followed by initiatives aimed to improve community health and well-being (45) and by initiatives which aim at achieving sustainability in cities and communities (30). A minor number of relationships are related to energy decarbonization and sustainable industry and educa- tion, gender, and inequality (16 both), then followed by digital revolution for sustainable development (8). 4.3. Qualitative Results The selected articles offer several examples of evaluations of interventions dealing with the improvement of adaptive governance of SES through the identification of novel solutions. Examples of evaluations undertaken are: (i) ex-ante evaluations of the impact caused by specific types of land use in protected areas [89]; (ii) participative evaluations aimed at creating awareness on environmental issues [90]; and (iii) the identification of best practices for resilient environmental management [24]. The following paragraphs summarize the recommendations on how and under which conditions collaborations con- tribute to the effective adaptive governance of SES as highlighted and suggested by eval- uation results. In order to facilitate the comprehension, the qualitative results are grouped into four categories having a common conceptual significance: (i) Communication, (ii) Eq- uity, (iii) Foresight, and (iv) Respect. These categories and their main components emerged from the analysis of the articles are summarized in Figure 5. 4.3.1. Communication Most of the selected articles highlights the importance of clear communication among multiple stakeholders, where individuals, groups, and organizations can express their values and perceptions. Developing a common language, specifically if it is informal and not technical, helps to avoid misunderstandings among actors [41,89,91,92]. Instrumental for effective collaborations is the use of visual tools—more user-friendly and for all types of people (also for illiterate people)—in communicating environmental issues or in par- ticipative evaluation processes [24,49]. Therefore, evaluations recommend clearness and transparency in communicating the contents of regulations, recommendations, directives, and so on from public bodies to all the other types of stakeholders, especially on the con- tent of policy objectives both general and specific [48,93–97]. Scientific communication is fundamental for community education. Third sector actors as proposed by [62]—espe- cially researchers, but also NGOs and generally all public actors—to play a fundamental role in the transmission of scientific knowledge to all other actors [96,98–100]. To be effec- tive, the content of scientific communication has to be clear and make use of tools able to be applied by non-experts [42,101,102], especially by policymakers who normally steer, guide, control, and manage natural resources. Moreover, public actors are invited to in- crease the number of communicative initiatives and tools aimed to make the community aware of environmental challenges and to propose everyday practices able to foster sus- tainable behaviors through the awareness that sustainable actions are more convenient for their well-being [92,103,104]. 11 of 29 Sustainability 2021, 13, 8276 11 of 29 4.3.2. Equity The integration of different typologies of actors, especially underprivileged stake- holders, and the respect of equity also within participative initiatives are essential for es- tablishing relationships based on trust and respect [24,26,41,42,48,89,92,95,98,99,105–108]. In participative processes, there is always the risk that interests of the elites prevail or that some groups of relevant actors are excluded in the decision-making processes [24,48,91,96,109]. Therefore, moderators or facilitators have the fundamental role in assur- ing equity through an objective and impartial management of trade-offs on interests and needs among actors [26,99,106]. The need for equity explains why collective initiatives are often sustained by external groups of experts, mostly NGOs and universities [100,110] who involve local stakeholders through, e.g., citizen-science tools, trips, workshops, and practical exercises of participative multicriteria assessments [24,26,90,93,94,100,111,112]. In order to guarantee equity in participative decision-making processes, facilitators and moderators have to assure the respect of privacy and allocate time allowing all actors to equally express their opinions and values [42,89]. In addition, public and third sector ac- tors are requested to coordinate and stimulate people to think and act for the good of all communities and to recognize valuable allies in local stakeholders [23,92,97,99,108,109,113,114]. Additionally, public authorities are required to devolve some power and autonomy to bottom-up initiatives that emerge from adaptive govern- ance processes [41,90,99,100,115]. Accordingly, all actors are invited to share material and non-material resources by considering a self-help perspective [96,98] in order to overcome limits that could preclude sustainability transformations, (e.g., the creation of ecotourism infrastructures in Amazon villages as suggested by [24]). Specifically, private actors are invited to avoid influencing scientific activities and research themes through the alloca- tion of private funds on specific research themes that do not positively impact on the so- ciety [93]. 4.3.3. Foresight Sustainability transformations require interventions producing effects in the long term, which contrast with individual needs focused on short-term outputs. Following this view, [116] underlines that several environmental projects are funded on a short-term pe- riod. To address this weakness, public bodies are requested to maintain the attention and the support on environmental initiatives in the long term by developing solid and coher- ent planning instruments. Institutional stability seems able to reduce the “stakeholder apathy” [42] and to assure continuity in environmental adaptive governance initiatives [41,42,49,91,96,117–119]. Considering public actors, [99] highlights the need to also sup- port collaboration between partners after the end of the project through the creation of a stable network of actors sharing common objectives and working together for a more ex- tended period. This could be fostered by programs having a long- or medium-term vision that can promote the resilience of ecosystems [99,102,120–124]. Experiences highlight the strategic nature of proposing tools to motivate private actors to be involved in sustaina- bility transformations. Accordingly, private actors are more likely to act when it is easy and convenient to do the right thing [26]. Sustainability transformations need to be pro- posed as means able to increase their well-being through, for example, the introduction of incentives [26,97,106,125]. The incentive has not to be only monetary (e.g., payments, sub- sidiaries) but also of a different nature (e.g., new job opportunities) [48,91,117]. To sustain innovations that foster sustainability transformations, donors are invited to sustain trans- disciplinary research [94,96]. On the other side, third sector and State actors are invited to create common spaces for boosting innovations [42,126]. Equally, private actors, and in particular market actors, must be encouraged to sustain scientific research, especially for the development of innovative eco-friendly technologies [127]. Moreover, they are in- vited to trust in science and accept changes in their everyday lives, even if it is difficult to see the short-term advantages [128]. 12 of 29 12 of 29 Sustainability 2021, 13, 8276 4.3.4. Respect Relevant and suitable sustainability transformations need dynamic and flexible reg- ulations and policies that take into consideration social and ecological characteristics and the scale where interventions take place in order to address specific emerging needs that continuously evolve among time and space [26,95,97,118,129–131]. This is why the State and third sector actors are requested to comprehend real problems dealt by local stakeholders (both community and market) [92,100,110]. In addition, public interventions need to be culturally contextualized, and they have to respect traditions (e.g., everyday practices and taboos) of communities where they are placed, especially in non-Western countries, in order to build trust and legitimation [24,41,91,98,99,108,111,128,132]. To do so, the literature invites the promotion of participative evaluations processes [108,133]. Accordingly, evaluations need to provide specific information on both the environmental and social contexts and to include indicators related to the quality of life of locals, espe- cially of indigenous communities, which very often appear as the most marginalized groups [93,94,96,105,122,129,134–136]. Figure 5. Categories and components fostering effective collaborations as highlighted by articles selected by the systematic review. Figure 5. Categories and components fostering effective collaborations as highlighted by articles selected by the systematic review. Figure 5. Categories and components fostering effective collaborations as highlighted by articles selected by the systematic review. 1. All typologies of evaluations should adopt a transdisciplinary approach when deal- ing with the assessment of collaborations for the adaptive governance of SES. 1. All typologies of evaluations should adopt a transdisciplinary approach when deal- ing with the assessment of collaborations for the adaptive governance of SES. The analysis of evaluations reported in the articles demonstrate that articles use ba- sically a transdisciplinary approach. This is corroborated by results showing that the ma- jority of articles selected by the review (42 out 56) are published in journals belonging to multiple subject areas. In addition, the analysis identifies two main approaches used to assess adaptive governance initiatives: (i) articles dealing only with social variables (18) and (ii) transdisciplinary articles dealing with both social and ecological variables (38). The heterogeneous variables used in the 56 analyzed articles attest that transdisciplinary research is instrumental to provide a transversal knowledge fitting all dimensions of sus- tainability [137], as reported by, e.g., [101,111]. Nevertheless, transdisciplinary approaches involve difficulties in their operationalization, specifically related to the diversity of inter- ests, values, and perceptions of actors involved in adaptative governance initiatives [138]. 2. Evaluations normally centered on secondary data should also use participative tech- niques for primary data collection. This will allow to reach a better understanding of real situations of evaluated contexts, which is a necessary pre-condition for effective collaborations. The scientific literature recognizes the central importance of using participative ap- proaches in all the phases of the project cycle, e.g., [48,49,114], in order to determine a real impact in the target context. Nevertheless, moving to the classification of evaluations re- ported in articles, we observe that articles using social and environmental indicators or indices limit the use of participative approaches in evaluations, e.g., [115,127]. Pure qual- itative articles can be subdivided into two categories: on one side, some articles use par- ticipative approaches in projects, e.g., [90,100], and on the other side, some articles focus on analyses of policy, e.g., [93,119]. Conversely, articles based on integrated assessments reveal that the involvement of the community by using participatory approaches plays a determinant role in (i) the identification of needs or environmental challenges tacking lo- cal communities, e.g., [24,89,109]; (ii) the implementation of project activities, e.g., [49,105,108]; (iii) the evaluation of undertaken actions, the successful reaching of their ob- jectives, and consequently, the impacts of the initiatives, e.g., [42]. 1. All typologies of evaluations should adopt a transdisciplinary approach when deal- ing with the assessment of collaborations for the adaptive governance of SES. Experiences demon- strate that knowledge sharing among local actors helps identify the specific needs of local communities and the interlinks among environmental and social problems, which are not immediately visible to the external managers, who typically adopt a sectorial problem- solving approach. In addition, ex-ante participative evaluations allow discussing local problems permitting people to take consciousness of the importance of the environmental challenge and identify context-based solutions that local community supports [52,89]. Re- sults support the need to enlarge the use of participative approaches in all types of evalu- ations, specifically in evaluations based on indicators and indices that normally rely on secondary data to better represent real situations of evaluated contexts. 3. Evaluations of adaptive governance initiatives of SES should involve actors from multiple spatial scales. 3. Evaluations of adaptive governance initiatives of SES should involve actors from multiple spatial scales. 5. Discussion The evaluations presented in the 56 articles highlight: (i) how effective governance of SES is difficult to achieve due to complexities and uncertainties which characterize envi- ronmental and social challenges presented in the different contexts analyzed and (ii) a specific and context-based environmental issue is typically characterized by multiple so- cial and institutional stakeholders interconnected through different ties with a set of in- terrelated environmental resources, as already pointed out by e.g., [30,32]. Through the results of the quantitative and qualitative analyses, we want to provide and help to (i) improve evaluations in future and (ii) provide guidelines to actors to foster effective collaborations. The following discussion firstly presents specific indications on how to assess the effectiveness of collaborations, and secondly, examines how actors could foster them. Sustainability 2021, 13, 8276 13 of 29 13 of 29 5.1. How to Assess the Effectiveness of Collaborations 5.1. How to Assess the Effectiveness of Collaborations 1. All typologies of evaluations should adopt a transdisciplinary approach when deal- ing with the assessment of collaborations for the adaptive governance of SES. 5. Evaluations of adaptive governance should focus on both environmental and social challenges to identify collaborations able to foster synergies in SES. 5. Evaluations of adaptive governance should focus on both environmental and social challenges to identify collaborations able to foster synergies in SES. 5. Evaluations of adaptive governance should focus on both environmental and social challenges to identify collaborations able to foster synergies in SES. The focus on human pressures evaluated by this study highlights that environmental actions reported in the articles have repercussions in addressing crucial social challenges that humanity, at present, has to deal with. Sustainability practices in food supply chain and in actions undertaken in urban areas are the most recurrent challenges in the selected articles. Agriculture and Fishing and Hunting, followed by Urban Areas and Industry, are human pressures with the highest number of relationships within the Sankey diagram. This could be explained because one of the most critical challenges that humanity must deal with in the future years will be the exponential increase of the global population and the consequent increasing demand for food to assure food security for all people [88,146] and the migration of people from rural to urban areas [147]. Accordingly, most of the adaptive activities reported in articles selected by the review focus on (i) food production in rural and urban areas, e.g., [106,112]; (ii) fishing activities, e.g., [41,105]; and (iii) the evaluation of sustainable practices in urban areas, e.g., [103,121]. Synergies between social and ecological challenges are supported in multiple international agreements and policies, e.g., Agenda 2030 and Sustainable Development Goals (SDGs) or the Farm to Fork EU Strategy [4,148]. Accordingly, it is not surprising that the number of articles selected by this study is highest in 2015 and 2016, when SDGs emerge in the international policies. Hence, evaluations need to focus on both environmental and social outputs that emerge from collaborations, as already exemplified by, e.g., [24,94]. 6. In order to assess the effectiveness of collaborations within environmental projects and programs, evaluations should focus on synergies and trade-offs among multiple environmental challenges determined by human actions at the same time. Therefore, they should be multi-sectorial. 6. In order to assess the effectiveness of collaborations within environmental projects and programs, evaluations should focus on synergies and trade-offs among multiple environmental challenges determined by human actions at the same time. Therefore, they should be multi-sectorial. 3. Evaluations of adaptive governance initiatives of SES should involve actors from multiple spatial scales. In terms of scale of intervention, our systematic literature review shows that evalua- tions are mainly focused on sub-national or local levels. Conversely, it evidences a low number of evaluations implemented at national, international, and global scales. Evalua- tions of sub-national and local interventions are characterized by a high frequency of ac- tivities based on participative approaches also implemented through multicriteria assess- ments, e.g., [48,99]. The local scale of intervention probably fosters the generation of effec- tive collaborations in SES [139,140]. Consequently, evaluations based on sub-national and local scales would be more prone to assess these collaborations. p Nevertheless, the literature highlights the need to avoid inward-looking approaches because the majority of SES does not limit to the narrow boundaries of the SES analyzed Sustainability 2021, 13, 8276 14 of 29 14 of 29 but is open and susceptible to external changes [48,141]. Consequently, [142] propose to involve actors from multiple scales in participative approaches, combining together dif- ferent interests to compensate for this weakness. Equally, multiple evaluations selected by the review adopt the same approach, involving stakeholders from different spatial scales, e.g., [99,107]. A quantitative analysis of relationships constituting networks through the Social Network Analysis (SNA) could be helpful in the assessment of connec- tivity between actors of adaptive governance initiatives, e.g., [69,107,143]. 4. Evaluations on collaborations for adaptive governance of SES should enlarge the context of analysis to countries that, at present, are most vulnerable to climate change and natural resources depletion. 4. Evaluations on collaborations for adaptive governance of SES should enlarge the context of analysis to countries that, at present, are most vulnerable to climate change and natural resources depletion. 4. Evaluations on collaborations for adaptive governance of SES should enlarge the context of analysis to countries that, at present, are most vulnerable to climate change and natural resources depletion. The geographical analysis of articles reveals a high concentration of evaluations in Europe, America, and Australia. Sub-Saharan Africa and the Middle East, which are more vulnerable to climate change effects and natural resources depletion, do not attest to a scientific discussion on the research topic. Studies placed in poor areas mainly deal with activities related to Western countries activities such as wildlife tourism, e.g., [91,107,109]. Thus, the geographical analysis of articles denotes a Western-centric vision in scientific research related to the evaluation of sustainability issues and environmental challenges. 3. Evaluations of adaptive governance initiatives of SES should involve actors from multiple spatial scales. This evidence is confirmed by multiple other studies related to sustainability analysis, e.g., higher education for sustainable development in [144] and resilience thinking in [145]. 5. Evaluations of adaptive governance should focus on both environmental and social challenges to identify collaborations able to foster synergies in SES. Environmental issues dealt by articles are mostly related to the use of (i) Land and Soil and (ii) Freshwater, and the conservation of (i) Biodiversity and (ii) Oceans and Coasts. Little attention is devoted to the Air. Evaluations analyzed by our literature review demonstrate the necessity to consider multiple environmental issues simultaneously, such as the interdependencies between land use and biodiversity as pointed out by, e.g., [52,107]. Accordingly, the scientific literature highlights the need to consider synergies and trade-offs among multiple environmental issues generated by implementing human activities [36]. For example, in the case of agricultural activities negatively impacting on Sustainability 2021, 13, 8276 15 of 29 15 of 29 the environment, the evaluation should consider the interactions among food supply, wa- ter use, and biodiversity loss [149,150]. The generation of effective collaborations, able to cope with multiple negative effects determined by human actions, can be stimulated by the inclusion of actors of multiple sectors, as evidenced in the scientific literature, e.g., by [151,152]. Our review provides examples of various cross-sector collaborations, such as the participative evaluations that involve fishers and tourist operators, e.g., [42,99,109]. Nevertheless, despite the recognition that multi-sectoriality is fundamental for an effec- tive environmental governance, the experiences highlight difficulties in its concretization due to different needs, visions of the world, problems to be addressed, terminology, etc., in multiple sectors [40–42]. From this background emerges the relevance and the need of trade-offs in identifying common and shared strategies to be implemented by collabora- tions of multiple and different actors, which, at present, are scarcely examined by the sci- entific literature [15]. 7. Evaluations of the governance of SES should consider the role of effective collabora- tions to promote transformations towards improved community well-being. 7. Evaluations of the governance of SES should consider the role of effective collabora- tions to promote transformations towards improved community well-being. 7. Evaluations of the governance of SES should consider the role of effective collabora- tions to promote transformations towards improved community well-being. Articles selected by the literature review are mainly focused on transformations re- lated to the sustainable use of natural resources such as land and oceans, followed by transformations aimed at fostering human well-being and the sustainability of urban ar- eas. Evaluations should focus on interventions not only in terms of assessment of the qual- ity of ecosystems, but also as opportunities to foster community well-being through the catalyzation of multiple facts such as inclusiveness, equality, trust, education of the com- munity, and the respect of rights and cultures, which can lead to the achievement of a thriving global society [153,154]. Accordingly, selected articles provide multiple examples of environmental evaluations which consider environmental interventions as means able to foster community well-being. For example, [24,49] demonstrate how effective manage- ment requires the involvement of indigenous communities and the respect of their cul- tures and lifestyles. Furthermore, [106] shows that the environmental projects placed in post-industrial cities not only impact the environmental quality, but they also accelerate environmental justice and social equity. However, at present, top-down and centralized approaches neglecting the fundamental role of local community and of peculiarities of contexts and cultures are still the most used in the governance of environmental resources [93]. Evaluations on collaborative efforts addressing specific environmental challenges through a bottom-up approach could be useful in the identification of new solutions able to improve both natural ecosystems and human well-being [17]. 5.2. How to Foster Effective Collaborations. 4. Respect of social and ecological contexts leads to the design and implementation of relevant activities, building trust and legitimation, and, consequently, fostering ef- fective collaborations. The initiatives described in the analyzed articles show that the respect of both eco- logical and social contexts is a prerequisite for implementing effective initiatives and col- laborations. Context-based approaches lead to the design and implementation of relevant initiatives that consider both (i) the ecological conditions evolving in time and space and (ii) local cultures and lifestyles. From the articles selected by the literature review emerges the fundamental role of policies able to adapt to every specific area and social need, which, consequently, can support new governance arrangements generated by adaptive govern- ance initiatives [90,93,96,100]. Relevant projects can build trust and legitimacy, helping with the generation of effective collaborations between the local community and external actors proposing initiatives, e.g., [24,49,99]. 5.2. How to Foster Effective Collaborations. 5.2. How to Foster Effective Collaborations. 1. A clear communication fosters community support to environmental activities, and consequently, it increases the possibility to foster effective collaborations through community awareness on environmental challenges. The qualitative analysis of the final considerations reported in the analyzed articles shows that a clear communication can empower locals, help in resolving conflicts, and help communities define good practices for contributing to sustainability transformations. In addition, a transparent information on activities and outputs can favor the reliability of actions undertaken by the promoters of adaptive governance initiatives, facilitating com- munity trust [91,136]. Clear communication fosters community awareness on environ- mental challenges and its support on environmental activities, e.g., [90,103], especially in contexts of poverty and marginalization, where people have little chances to be empow- ered through traditional channels, e.g., schooling [24,121]. For example, [106] observes that a clear communication in relation to urban community gardening has the possibility to include the most marginal groups in community activities to empower them and foster their pro-environmental behavior, and, consequently, their support of the objectives of the initiative. Sustainability 2021, 13, 8276 16 of 29 16 of 29 2. Equity fosters the emergence of a conscious and shared environmental responsibility through the identification of common strategies by multiple stakeholders that sup- port effective collaborations. 2. Equity fosters the emergence of a conscious and shared environmental responsibility through the identification of common strategies by multiple stakeholders that sup- port effective collaborations. Evaluations analyzed by the qualitative analysis highlight that equity in participative processes stimulates the emergence of a conscious and shared environmental responsibil- ity among all stakeholders who have different rights and duties related to the environ- mental issue to be tackled [41]. Adaptive governance initiatives characterized by equity are more prone to generate meaningful dialogue between different actors, and, conse- quently, the identification of strategies in agreement with all parts involved, which con- siderate needs and opportunities for all actors, including the less powerful, e.g., [24,89,103,105,106,108,109,114,128]. 3. Foresight in the governance of SES fosters a constant process of adaptation, support- ing effective collaborations in the long run. 3. Foresight in the governance of SES fosters a constant process of adaptation, support- ing effective collaborations in the long run. 3. Foresight in the governance of SES fosters a constant process of adaptation, support- ing effective collaborations in the long run. The qualitative analysis reveals that foresight is necessary for sustaining the trans- formative process that essentially constitutes adaptive governance as described by the adaptive cycle [30]. Accordingly, foresight is crucial in fostering changes in natural re- source management through the introduction or development of new tools or novel ap- proaches that could lead to the implementation of innovations [49,52,90,110,123,127,132]. Forward-looking initiatives can assure continuity in the transformative process also after the end of projects through the creation of networks of actors who continue to collaborate in order to stimulate additional improvements of the governance of SES (e.g., through the creation of new governance arrangements such as alliances and spin-offs, as reported by [90]). The continuity of collaborations in the long-term period through, e.g., regular peri- odical meetings [42], is, in turn, instrumental in avoiding the stakeholder apathy charac- terized by the declining of exchange of knowledge and the engagement of stakeholders and leadership [42,96]. 4. Respect of social and ecological contexts leads to the design and implementation of relevant activities, building trust and legitimation, and, consequently, fostering ef- fective collaborations. 4. Respect of social and ecological contexts leads to the design and implementation of relevant activities, building trust and legitimation, and, consequently, fostering ef- fective collaborations. 5.3. Managerial Implications Clear communication, equity, foresight, and respect also need to be considered from a managerial point of view. In particular, if interventions concern SES, they are requested to focus on both the social and the ecological peculiarities from their starting phases. This is highlighted in multiple policy documents (e.g., Agenda 2030 [155]; Paris Agreements [156]; The European Green Deal [157]) and program regulations (e.g., LIFE Programme [158]; Interreg Europe [159]). In particular, the most general indication that emerged from this study is to valorize the fundamental role of community involvement from the very beginning of every project. Local actors, whom project managers often consider as passive beneficiaries of project results [160], need to be involved and converted into active stake- holders through the devolution of responsibilities and autonomy in actions implementa- tion [161]. Accordingly, building a shared environmental responsibility among jurisdic- Sustainability 2021, 13, 8276 17 of 29 17 of 29 tional levels is functional for increasing the effectiveness of activities [162]. In the follow- ing paragraph, we provide indications on how to increase people engagement and, thus, sustain collaborations in the different phases constituting the project cycle. tional levels is functional for increasing the effectiveness of activities [162]. In the follow- ing paragraph, we provide indications on how to increase people engagement and, thus, sustain collaborations in the different phases constituting the project cycle. (i) Identification, formulation: project designers are requested to clearly identify stake- holders and their potential role in the phases of identification and formulation through the stakeholder analysis (e.g., influence and matrix) and through the SWOT analysis [48,163]. In addition, to guarantee equity, the identification of possible coa- litions constitutes a necessary step in order to prevent that elites prevail in decision- making processes [89]. Project designers should involve local communities from the very beginning by respecting the values and culture that could be better understood through the use of both informal conversations and well-designed questionnaires and surveys [100] or multicriteria assessments to be performed with the active par- ticipation of representatives of the local communities [48,52,89]. Instrumental, since the identification of the project, is the schedule of different meetings among stake- holders aimed to identify problems and resolve possible conflicts and identify possi- ble trade-offs in the decision-making process [41]. 5.3. Managerial Implications (ii) (ii) Implementation, monitoring: effective collaborations among multiple stakeholders could be supported and stimulated through the involvement of stakeholders in reg- ular meetings in order to avoid stakeholder apathy [42] and the promotion of the creation of new bottom-up experiences such as spin-offs and alliances [90]. Trips, fes- tivals, and special events are fundamental for communicating project objectives and results, stimulating a pro-environmental behavioral change of community that, con- sequently, is more prone to support project interventions [90]. Citizen science is in- strumental for the involvement of people but also for the monitoring of activities [93]. (iii) Evaluation: evaluations of projects need to consider the environmental results derived by projects implementation and social outcomes derived from them. As highlighted by [94] is fundamental to assessing all dimensions of sustainability (i.e., environmen- tal, social, and economic) through developing suitable indicators. Additionally, eval- uations need to be participative and include all types of stakeholders, especially local actors, as reported by [49], using tools that better fit with people cultures and peculi- arities. Conversely, evaluations and results diffusion need to be clearly communi- cated to everybody through, e.g., public events designed not only for technicians but also for non-experts [24]. Instrumental in communicating project results is the iden- tification and spreading of best practices [106]. 6. Conclusions Our review of evaluations makes evident that transdisciplinary, multi-scale, and multi-sector approaches should be applied to assess the effectiveness of collaborations in adaptive governance of SES. Moreover, it shows that participative approaches are instru- mental in understanding the context where initiatives are placed and demonstrate that environmental actions implemented through effective collaborations should promote so- cial well-being. Four broad concepts can resume the conditions able to catalyze effective collaborations in the governance of SES. They include clear communication, equity, fore- sight, and respect. They are seen as characteristics able to incentive the inclusion of stake- holders, their trust, and consequently, their support in the definition and implementation of relevant initiatives, and to assure the continuing of the transformative process that con- stitutes the adaptive governance of SES. From our analysis, it emerges that the effectiveness of adaptive governance initiatives is essentially based on processes established through the involvement of multiple actors and the consequent emergence of social networks. Future studies and evaluations of en- vironmental projects and programs could better analyze the connectivity between actors, for example, increasing the application and use of Social Network Analysis. Sustainability 2021, 13, 8276 18 of 29 18 of 29 Despite the abundance of recommendations that emerge from the analysis of articles related to interactions among different actors, evaluations mainly focus on the role of pub- lic actors (i.e., State and third sector), with little attention on the contribution of private actors (market and community). To address this gap, future studies could focus on the side of private actors and develop user-friendly tools to foster sustainability in everyday behaviors. Moreover, this analysis highlights the need to highlight and valorize the most mar- ginal voices embedded in adaptive governance. Evaluations about adaptive governance of SES located in developing countries could be opportunities for the creation of new knowledge through the sharing of both scientific and traditional/indigenous knowledge, which could propose new effective solutions and approaches useful for sustainability transformations to be also implemented in different contexts. p Even if this study is limited to the selection of articles written only in English and retrieved from the SCOPUS database, we believe it provides a useful initial overview of the current knowledge and possible improvements in evaluation of collaborations within adaptive governance of social–ecological systems and their global to local challenges. Author Contributions: Conceptualization, E.A., E.P. and L.S.; methodology, E.A. Appendix A List of articles selected for the systematic review. List of articles selected for the systematic review. 1. Ancuta, C.; Olaru, M.; Popa, N.; Isfanescu, R.; Jigoria-Oprea, L. Evaluation of the sus-tainable development of rural settlements. Case study: Rural settlementd from Romanian Banat. Car- path. J. Earth Envrion. 2015, 10, 67–80. 1. Ancuta, C.; Olaru, M.; Popa, N.; Isfanescu, R.; Jigoria-Oprea, L. Evaluation of the sus-tainable development of rural settlements. Case study: Rural settlementd from Romanian Banat. Car- path. J. Earth Envrion. 2015, 10, 67–80. p 2. Armitage, D.; Marschke, M.; Plummer, R. Adaptive co-management and the paradox of learn- ing. Glob. Environ. Chang. 2008, 18, 86–98, doi:10.1016/j.gloenvcha.2007.07.002. p 2. Armitage, D.; Marschke, M.; Plummer, R. Adaptive co-management and the paradox of learn- ing. Glob. Environ. Chang. 2008, 18, 86–98, doi:10.1016/j.gloenvcha.2007.07.002. 3. Benitez-Capistros, F.; Hugé, J.; Koedam, N. Environmental impacts on the Galapagos Islands: Identification of interactions, perceptions and steps ahead. Ecol. Indic. 2014, 38, 113–123, doi:10.1016/j.ecolind.2013.10.019. 3. Benitez-Capistros, F.; Hugé, J.; Koedam, N. Environmental impacts on the Galapagos Islands: Identification of interactions, perceptions and steps ahead. Ecol. Indic. 2014, 38, 113–123, doi:10.1016/j.ecolind.2013.10.019. 4. Bergquist, D.A.; Cavalett, O.; Rydberg, T. Participatory emergy synthesis of integrated food and biofuel production: A case study from Brazil. Environ. Dev. Sustain. 2011, 14, 167–182, doi:10.1007/s10668-011-9314-8. 4. Bergquist, D.A.; Cavalett, O.; Rydberg, T. Participatory emergy synthesis of integrated food and biofuel production: A case study from Brazil. Environ. Dev. Sustain. 2011, 14, 167–182, doi:10.1007/s10668-011-9314-8. 5. Brown, P.R.; Jacobs, B.; Leith, P. Participatory monitoring and evaluation to aid investment in natural resource manager capacity at a range of scales. Environ. Monit. Assess. 2012, 184, 7207– 7220, doi:10.1007/s10661-011-2491-y. 5. Brown, P.R.; Jacobs, B.; Leith, P. Participatory monitoring and evaluation to aid investment in natural resource manager capacity at a range of scales. Environ. Monit. Assess. 2012, 184, 7207– 7220, doi:10.1007/s10661-011-2491-y. y 6. Bundy, A.; Chuenpagdee, R.; Boldt, J.L.; Borges, M.D.F.; Camara, M.L.; Coll, M.; Diallo, I.; Fox, C.; Fulton, E.A.; Gazihan, A.; et al. Strong fisheries management and governance positively impact ecosystem status. Fish Fish. 2016, 18, 412–439, doi:10.1111/faf.12184. 6. Bundy, A.; Chuenpagdee, R.; Boldt, J.L.; Borges, M.D.F.; Camara, M.L.; Coll, M.; Diallo, I.; Fox, C.; Fulton, E.A.; Gazihan, A.; et al. Strong fisheries management and governance positively impact ecosystem status. Fish Fish. 2016, 18, 412–439, doi:10.1111/faf.12184. p y 7. 6. Conclusions and E.P.; data cu- ration, E.A.; writing—original draft preparation, E.A.; writing—review and editing, E.A., E.P. and L.S.; supervision, E.P, L.S. and A.C. All authors have read and agreed to the published version of the manuscript. Funding: This research received no external funding Funding: This research received no external funding Institutional Review Board Statement: Not applicable. Informed Consent Statement: Not applicable. Conflicts of Interest: The authors declare no conflict of interest. Conflicts of Interest: The authors declare no conflict of interest. Appendix A Butler, J.; Young, J.; McMyn, I.; Leyshon, B.; Graham, I.; Walker, I.; Baxter, J.; Dodd, J.; War- burton, C. Evaluating adaptive co-management as conservation conflict resolution: Learning from seals and salmon. J. Environ. Manag. 2015, 160, 212–225, doi:10.1016/j.jen- vman.2015.06.019. 8. Chu, J.; Garlock, T.; Sayon, P.; Asche, F.; Anderson, J.L. Impact evaluation of a fisheries devel- opment project. Mar. Policy 2017, 85, 141–149, doi:10.1016/j.marpol.2017.08.024. p p j y j p 9. Clark, T.W.; Padwe, J. The Ecuadorian Condor Bioreserve Initiative. J. Sustain. For. 2004, 18, 297–324, doi:10.1300/j091v18n02_14. j 10. De Alencar, N.P.; Le Tissier, M.; Paterson, S.; Newton, A. Circles of Coastal Sustainability: A Framework for Coastal Management. Sustainability 2020, 12, 4886, doi:10.3390/su12124886. 19 of 29 Sustainability 2021, 13, 8276 19 of 29 11. Dressel, S.; Ericsson, G.; Sandström, C. Mapping social-ecological systems to understand the challenges underlying wildlife management. Environ. Sci. Policy 2018, 84, 105–112, doi:10.1016/j.envsci.2018.03.007. j 12. Etxano, I.; Garmendia, E.; Pascual, U.; Hoyos, D.; Díez, M.; Cadiñanos, J.A.; Lozano, P.J. A participatory integrated assessment approach for Natura 2000 network sites. Environ. Plan. C: Gov. Policy 2015, 33, 1207–1232, doi:10.1177/0263774x15612318. 13. Foley, P.; Okyere, D.A.; Mather, C. Alternative environmentalities: Recasting the assessment of Canada’s first Marine Stewardship Council-certified fishery in social terms. Ecol. Soc. 2018, 23, doi:10.5751/es-10382-230337. 14. Forster, J.; Turner, R.; Fitzsimmons, C.; Peterson, A.M.; Mahon, R.; Stead, S.M. Evidence of a common understanding of proximate and distal drivers of reef health. Mar. Policy 2017, 84, 263–272, doi:10.1016/j.marpol.2017.07.017. 15. Gerhardinger, L.C.; Godoy, E.A.S.; Jones, P.; Sales, G.; Ferreira, B.P. Marine Protected Dramas: The Flaws of the Brazilian National System of Marine Protected Areas. Environ. Manag. 2010, 47, 630–643, doi:10.1007/s00267-010-9554-7. 16. Gilioli, G.; Tikubet, G.; Herren, H.R.; Baumgartner, J. Assessment of social–ecological transi- tions in a peri-urban Ethiopian farming community. Int. J. Agric. Sustain. 2014, 13, 204–221, doi:10.1080/14735903.2014.954452. 17. Gillon, S.; Booth, E.G.; Rissman, A.R. Shifting drivers and static baselines in environmental governance: Challenges for improving and proving water quality outcomes. Reg. Environ. Chang. 2015, 16, 759–775, doi:10.1007/s10113-015-0787-0. 18. Guerrero, A.M.; Bodin, Ö.; McAllister, R.R.J.; Wilson, K. Achieving social-ecological fit through bottom-up collaborative governance: An empirical investigation. Ecol. Soc. 2015, 20, doi:10.5751/es-08035-200441. 19. He, R.; Tang, Z.; Dong, Z.; Wang, S. Performance Evaluation of Regional Water Environment Integrated Governance: Case Study from Henan Province, China. Int. J. Environ. Res. Public Heal. 2020, 17, 2501, doi:10.3390/ijerph17072501. j p 20. Appendix A Community owned solutions: Identifying local best practices for social-ecological sustainability. Ecol. Soc. 2016, 21, doi:10.5751/es-08496-210242. 33. Nilsson, A.K.; Bohman, B. Legal prerequisites for ecosystem-based management in the Baltic Sea area: The example of eutrophication. Ambio 2015, 44, 370–380, doi:10.1007/s13280-015-0656- 6. 34. Nuno, A.; Bunnefeld, N.; Milner-Gulland, E. Managing social–ecological systems under uncer- tainty: Implementation in the real world. Ecol. Soc. 2014, 19, doi:10.5751/es-06490-190252. 35. Oviedo, A.F.P.; Bursztyn, M. The Fortune of the Commons: Participatory Evaluation of Small- Scale Fisheries in the Brazilian Amazon. Environ. Manag. 2016, 57, 1009–1023, doi:10.1007/s00267-016-0660-z. 36. Parlee, C.E.; Wiber, M.G. Using conflict over risk management in the marine environment to strengthen measures of governance. Ecol. Soc. 2018, 23, doi:10.5751/es-10334-230405. 37. Pearson, L.J.; Collins, K. Does social-ecological context influence state-based water manage- ment decisions? Case study from Queensland, Australia (1980–2006). Hydrol. Res. 2009, 12, 186– 202, doi:10.2166/wp.2009.055. 38. Petursdottir, T.; Arnalds, O.; Baker, S.; Montanarella, L.; Aradottir, A.L. A Social–Ecological System Approach to Analyze Stakeholders’ Interactions within a Large-Scale Rangeland Res- toration Program. Ecol. Soc. 2013, 18, doi:10.5751/es-05399-180229. 39. Robinson, C.; Bark, R.H.; Garrick, D.; Pollino, C.A. Sustaining local values through river basin governance: Community-based initiatives in Australia’s Murray–Darling basin. J. Environ. Plan. Manag. 2014, 58, 2212–2227, doi:10.1080/09640568.2014.976699. 40. Schouten, M.A.; van der Heide, C.M.; Heijman, W.J.; Opdam, P.F. A resilience-based policy evaluation framework: Application to European rural development policies. Ecol. Econ. 2012, 81, 165–175, doi:10.1016/j.ecolecon.2012.07.004. j 41. Schultz, L.; Lundholm, C. Learning for resilience? Exploring learning opportunities in bio- sphere reserves. Environ. Educ. Res. 2010, 16, 645–663, doi:10.1080/13504622.2010.505442. 42. Seyfang, G. Sustainable consumption, the new economics and community currencies: Devel- oping new institutions for environmental governance. Reg. Stud. 2006, 40, 781–791, doi:10.1080/00343400600959173. 43. Sheng, R.; Lin, T. Evolutionary Assessment of the Ecological Governance under the Metropol- itan Background: Evidence from Chongming Eco-Island, Shanghai, China. Sustainability 2019, 11, 5327, doi:10.3390/su11195327. 44. Shkaruba, A.; Kireyeu, V. Recognising ecological and institutional landscapes in adaptive gov- ernance of natural resources. For. Policy Econ. 2013, 36, 87–97, doi:10.1016/j.forpol.2012.10.004. 45. Smedstad, J.A.; Gosnell, H. Do Adaptive Comanagement Processes Lead to Adaptive Coman- agement Outcomes? A Multicase Study of Long-term Outcomes Associated with the National Riparian Service Team’s Place-based Riparian Assistance. Ecol. Soc. 2013, 18, doi:10.5751/es- 05793-180408. 46. Söderberg, C. Complex governance structures and incoherent policies: Implementing the EU water framework directive in Sweden. J. Environ. Manag. 2016, 183, 90–97, doi:10.1016/j.jen- vman.2016.08.040. 47. Sparrevik, M.; Breedveld, G.D. Appendix A Jennings, S.; Pascoe, S.; Hall-Aspland, S.; Le Bouhellec, B.; Norman-Lopez, A.; Sullivan, A.; Pecl, G. Setting objectives for evaluating management adaptation actions to address climate change impacts in south-eastern Australian fisheries. Fish. Oceanogr. 2016, 25, 29–44, doi:10.1111/fog.12137. g 21. Johnson, F.A.; Eaton, M.J.; Mikels-Carrasco, J.; Case, D. Building adaptive capacity in a coastal region experiencing global change. Ecol. Soc. 2020, 25, doi:10.5751/es-11700-250309. 22. Jones, O.P.; Stephenson, R.L. Practical use of full-spectrum sustainability in the Bay of Fundy. Ecol. Soc. 2019, 24, doi:10.5751/es-11010-240325. 23. Kimario, F.F.; Botha, N.; Kisingo, A.; Job, H. Theory and practice of conservancies: Evidence from wildlife management areas in Tanzania. Erdkd. 2020, 117–143, doi:10.3112/erd- kunde.2020.02.03. 24. Koenigstein, S.; Ruth, M.; Gößling-Reisemann, S. Stakeholder-Informed Ecosystem Modeling of Ocean Warming and Acidification Impacts in the Barents Sea Region. Front. Mar. Sci. 2016, 3, doi:10.3389/fmars.2016.00093. 25. Langemeyer, J.; Gómez-Baggethun, E.; Haase, D.; Scheuer, S.; Elmqvist, T. Bridging the gap between ecosystem service assessments and land-use planning through Multi-Criteria Deci- sion Analysis (MCDA). Environ. Sci. Policy 2016, 62, 45–56, doi:10.1016/j.envsci.2016.02.013. y y j 26. Li, J.; Pan, S.-Y.; Kim, H.; Linn, J.H.; Chiang, P.-C. Building green supply chains in eco-indus- trial parks towards a green economy: Barriers and strategies. J. Environ. Manag. 2015, 162, 158– 170, doi:10.1016/j.jenvman.2015.07.030. j j 27. Lin, G.; Wu, B.; Lin, X.; Fan, A.; Tian, S. Ecological Study on the Index System and Methodol- ogy of Performance Quantization for Sustainable Forest Management. Ekoloji 2019, 28, 1365– 1372. 28. Liu, B.; Wang, J.; Jing, Z.; Tang, Q. Measurement of sustainable transformation capability of resource-based cities based on fuzzy membership function: A case study of Shanxi Province, China. Resour. Policy 2020, 68, 101739, doi:10.1016/j.resourpol.2020.101739. 29. Lopes, R.; Videira, N. Bringing stakeholders together to articulate multiple value dimensions of ecosystem services. Ocean Coast. Manag. 2018, 165, 215–224, doi:10.1016/j.oce- coaman.2018.08.026. 30. Luisetti, T.; Turner, R.; Jickells, T.; Andrews, J.; Elliott, M.; Schaafsma, M.; Beaumont, N.; Mal- colm, S.; Burdon, D.; Adams, C.; et al. Coastal Zone Ecosystem Services: From science to values and decision making; a case study. Sci. Total. Environ. 2014, 493, 682–693, doi:10.1016/j.sci- totenv.2014.05.099. Sustainability 2021, 13, 8276 20 of 29 20 of 29 31. Marshall, G.R. Transaction costs, collective action and adaptation in managing complex social– ecological systems. Ecol. Econ. 2013, 88, 185–194, doi:10.1016/j.ecolecon.2012.12.030. 32. Mistry, J.; Berardi, A.; Tschirhart, C.; Bignante, E.; Haynes, L.; Benjamin, R.; Albert, G.; Xavier, R.; Robertson, B.; Davis, O.; et al. Appendix A Integrated Sustainability Assessment of Public Rental Housing Community Based on a Hybrid Method of AHP-Entropy Weight and Cloud Model. Sustainability 2017, 9, 603, doi:10.3390/su9040603. Appendix A From Ecological Risk Assessments to Risk Governance. Evalu- ation of the Norwegian Management System for Contaminated Sediments. Integr. Environ. As- sess. Manag. 2007, 6, 240–248, doi:10.1897/ieam_2009-049.1. g 48. Stacey, N.; Izurieta, A.; Garnett, S.T. Collaborative Measurement of Performance of Jointly Managed Protected Areas in Northern Australia. Ecol. Soc. 2013, 18, doi:10.5751/es-05273- 180119. 49. Stephenson, R.L.; Paul, S.; Wiber, M.; Angel, E.; Benson, A.J.; Charles, A.; Chouinard, O.; Clem- ens, M.; Edwards, D.; Foley, P.; et al. Evaluating and implementing social-ecological systems: A comprehensive approach to sustainable fisheries. Fish Fish. 2018, 19, 853–873, doi:10.1111/faf.12296. 50. Thiel, A.; Schleyer, C.; Hinkel, J.; Schlüter, M.; Hagedorn, K.; Bisaro, S.; Bobojonov, I.; Hamidov, A. Transferring Williamson’s discriminating alignment to the analysis of environ- mental governance of social-ecological interdependence. Ecol. Econ. 2016, 128, 159–168, doi:10.1016/j.ecolecon.2016.04.018. Sustainability 2021, 13, 8276 21 of 29 21 of 29 51. Thompson, B.S.; Friess, D.A. Stakeholder preferences for payments for ecosystem services (PES) versus other environmental management approaches for mangrove forests. J. Environ. Manag. 2019, 233, 636–648, doi:10.1016/j.jenvman.2018.12.032. g j j 52. Treemore-Spears, L.J.; Grove, J.M.; Harris, C.K.; Lemke, L.D.; Miller, C.J.; Pothukuchi, K.; Zhang, Y.; Zhang, Y.L. A workshop on transitioning cities at the food-energy-water nexus. J. Environ. Stud. Sci. 2016, 6, 90–103, doi:10.1007/s13412-016-0381-x. 53. Uchiyama, Y.; Kohsaka, R. Application of the City Biodiversity Index to populated cities in Japan: Influence of the social and ecological characteristics on indicator-based management. Ecol. Indic. 2019, 106, 105420, doi:10.1016/j.ecolind.2019.05.051. 54. Waylen, K.A.; Blackstock, K.L. Monitoring for Adaptive Management or Modernity: Lessons from recent initiatives for holistic environmental management. Environ. Policy Gov. 2017, 27, 311–324, doi:10.1002/eet.1758. 55. Waylen, K.A.; Blackstock, K.L.; Van Hulst, F.; Damian, C.; Horváth, F.; Johnson, R.K.; Kanka, R.; Külvik, M.; Macleod, C.J.; Meissner, K.; et al. Policy-driven monitoring and evaluation: Does it support adaptive management of socio-ecological systems?. Sci. Total. Environ. 2019, 662, 373–384, doi:10.1016/j.scitotenv.2018.12.462. 55. Waylen, K.A.; Blackstock, K.L.; Van Hulst, F.; Damian, C.; Horváth, F.; Johnson, R.K.; Kanka, R.; Külvik, M.; Macleod, C.J.; Meissner, K.; et al. Policy-driven monitoring and evaluation: Does it support adaptive management of socio-ecological systems?. Sci. Total. Environ. 2019, 662, 373–384, doi:10.1016/j.scitotenv.2018.12.462. j 56. Duan, K.; Zuo, J.; Zhao, X.; Tang, D. Integrated Sustainability Assessment of Public Rental Housing Community Based on a Hybrid Method of AHP-Entropy Weight and Cloud Model. Sustainability 2017, 9, 603, doi:10.3390/su9040603. 56. Duan, K.; Zuo, J.; Zhao, X.; Tang, D. 1. Chakraborty, I.; Maity, P. COVID-19 outbreak: Migration, effects on society, global environment and prevention. Sci. Total Environ. 2020, 728, 138882, doi:10.1016/j.scitotenv.2020.138882. 2. European Commission. EU Biodiversity Strategy for 2030. Bringing Nature Back into Our Lives. 2020. Available online: https://eur-lex.europa.eu/resource.html?uri=cellar:a3c806a6-9ab3-11ea-9d2d-01aa75ed71a1.0001.02/DOC_1&format=PDF (accessed on 1 June 2021). ( ) 3. Allen, M.R.; Dube, O.P.; Solecki, W.; Aragón-Durand, F.; Cramer, W.; Humphreys, S.; Kainuma, M.; Kala, J.; Mahowald, N.; Mulugetta, Y.; et al. Framing and Context. In Global Warming of 1.5 °C. An IPCC Special Report on the Impacts of Global Warming of 1.5 °C above Pre-Industrial Levels and Related Global Greenhouse Gas Emission Pathways, in the Context of Strengthening the Global Response to the Threat of Climate Change, Sustainable Development, and Efforts to Eradicate Poverty. Masson-Delmotte, V.P., Appendix B Table A1. Journals where selected articles are published and their subject areas. Journals No. of Articles Subject Areas Ambio 1 Environmental Science Medicine Social Sciences Carpathian Journal of Earth and Environmental Sciences 1 Earth and Planetary Sciences Environmental Science Ecological Economics 3 Economics, Econometrics and Finance Environmental Science Ecological Indicators 2 Agricultural and Biological Sciences Decision Sciences Environmental Science Ecology and Society 10 Environmental Science Ekoloji 1 Agricultural and Biological Sciences Environmental Science Environment and Planning C: Government and Policy 1 Environmental Science Social Sciences Environment, Development and Sustainability 1 Economics, Econometrics and Finance Environmental Science Social Sciences Environmental Education Research 1 Social Sciences Environmental Management 2 Environmental Science Medicine Environmental Monitoring and Assessment 1 Environmental Science Medicine Environmental Policy and Governance 1 Environmental Science Social Sciences Environmental Science and Policy 2 Environmental Science Social Sciences Erdkunde 1 Earth and Planetary Sciences Environmental Science Social Sciences Table A1. Journals where selected articles are published and their subject areas. Sustainability 2021, 13, 8276 22 of 29 Fish and Fisheries 2 Agricultural and Biological Sciences Earth and Planetary Sciences Environmental Science Fisheries Oceanography 1 Agricultural and Biological Sciences Earth and Planetary Sciences Forest Policy and Economics 1 Agricultural and Biological Sciences Economic, Econometrics and Finance Environmental Science Social Sciences Frontiers in Marine Science 1 Agricultural and Biological Sciences Earth and Planetary Sciences Engineering Environmental Science Global Environmental Change 1 Environmental Science Social Sciences Integrated Environmental As- sessment and Management 1 Environmental Science Social Sciences Medicine International Journal of Agricul- tural Sustainability 1 Agricultural and Biological Sciences Economics, Econometrics and Finance International Journal of Envi- ronmental Research and Public Health 1 Environmental Science Medicine Journal of Environmental Management 4 Environmental Science Medicine Journal of Environmental Plan- ning and Management 1 Chemical Engineering Environmental Science Social Sciences Journal of Environmental Stud- ies and Sciences 1 Environmental Science Social Sciences Journal of Sustainable Forestry 1 Agricultural and Biological Sciences Energy Environmental Science Social Sciences Marine Policy 2 Agricultural and Biological Sciences Economic, Econometrics and Finance Environmental Science Social Sciences Ocean and Coastal Management 1 Agricultural and Biological Sciences Earth and Planetary Sciences Environmental Science Regional Environmental Change 1 Environmental Science Regional Studies 1 Environmental Science Social Sciences Resources Policy 1 Economics, Econometrics and Finance Environmental Science Social Sciences Science of the Total Environ- ment 2 Environmental Science Sustainability (Switzerland) 3 Energy Environmental Science Social Sciences Water Policy 1 Environmental Science Social Sciences References Folke, C.; Biggs, R.; Norström, A.V.; Reyers, B.; Rockström, J. Social-ecological resilience and biosphere-based sustainability science. Ecol. Soc. 2016, 21, 41, doi:10.5751/ES-08748-210341. 9. World Health Organization. Our Planet, Our Health, Our Future Human Health and the Rio Conventions: Biological Diversity, Climate Change and Desertification; World Health Organization: Geneva, Switzerland, 2020. 10. Butchart, S.H.M.; Walpole, M.; Collen, B.; van Strien, A.; Scharlemann, J.P.W.; Almond, R.E.A.; Baillie, J.E.M.; Bomhard, B.; Brown, C.; Bruno, J.; et al. Global Biodiversity: Indicators of Recent Declines. Science 2010, 328, 5982, doi:10.1126/sci- ence.1187512. 11. Crutzen, P.J. The “Anthropocene”. In Earth System Science in the Anthropocene.; Ehlers, E., Krafft, T., Eds.; Springer: Berlin/Heidelberg, Germany, 2006; doi:10.1007/3-540-26590-2_3. g y 12. Gautam, S.; Hens, L. Covid-19: Impact by and on the environment, health and economy. Environ. Dev. Sustain. 2020, 22, 4953–4954, doi:10.1007/s10668-020-00818-7. 13. Severo, E.A.; Ferro De Guimarães, J.C.; Dellarmelin, M.L. Impact of the COVID-19 pandemic on environmental awareness, sustainable consumption and social responsibility: Evidence from generations in Brazil and Portugal. J. Clean. Prod. 2021, 286, 124947, doi:10.1016/j.jclepro.2020.124947. j j p 14. Hummels, H.; Argyrou, A. Planetary demands: Redefining sustainable development and sustainable entrepreneurship. J. Clean. Prod. 2021, 278, 123804, doi:10.1016/j.jclepro.2020.123804. 15. Bowen, K.J.; Cradock-Henry, N.A.; Koch, F.; Patterson, J.; Häyhä, T.; Vogt, J.; Barbi, F. Implementing the “Sustainable De- velopment Goals”: Towards addressing three key governance challenges—collective action, trade-offs, and accountability. Curr. Opin. Environ. Sustain. 2017, 26–27, 90–96, doi:10.1016/j.cosust.2017.05.002. p j 16. Loorbach, D.; Wittmayer, J.; Avelino, F.; von Wirth, T.; Frantzeskaki, N. Transformative innovation and translocal diffusion. Environ. Innov. Soc. Transit. 2020, 35, 251–260, doi:10.1016/j.eist.2020.01.009. 17. Bodin, Ö.; Crona, B. The role of social networks in natural resource governance: What relational patterns make a difference? Glob. Environ. Chang. 2009, 19, 366–374, doi:10.1016/j.gloenvcha.2009.05.002. 18. Bodin, Ö.; Robins, G.; McAllister, R.J.; Guerrero, A.; Crona, B.; Tengö, M.; Lubell, M. Theorizing benefits and constraints in collaborative environmental governance: A transdisciplinary social-ecological network approach for empirical investiga- tions. Ecol. Soc. 2016, 21, 40, doi:10.5751/ES-08368-210140. 19. Rijke, J.; Brown, R.; Zevenbergen, C.; Ashley, R.; Farrelly, M.; Morison, P.; van Herk, S. Fit-for-purpose governance: A frame- work to make adaptive governance operational. Environ. Sci. Policy 2012, 22, 73–84, doi:10.1016/j.envsci.2012.06.010. 19. Rijke, J.; Brown, R.; Zevenbergen, C.; Ashley, R.; Farrelly, M.; Morison, P.; van Herk, S. Fit-for-purpose governance: A frame- work to make adaptive governance operational. Environ. Sci. Policy 2012, 22, 73–84, doi:10.1016/j.envsci.2012.06.010. 20. Bodin, Ö. Collaborative environmental governance: Achieving collective action in social-ecological systems. References 1. Chakraborty, I.; Maity, P. COVID-19 outbreak: Migration, effects on society, global environment and prevention. Sci. Total Environ. 2020, 728, 138882, doi:10.1016/j.scitotenv.2020.138882. 2. European Commission. EU Biodiversity Strategy for 2030. Bringing Nature Back into Our Lives. 2020. Available online: https://eur-lex.europa.eu/resource.html?uri=cellar:a3c806a6-9ab3-11ea-9d2d-01aa75ed71a1.0001.02/DOC_1&format=PDF (accessed on 1 June 2021). ( ) 3. Allen, M.R.; Dube, O.P.; Solecki, W.; Aragón-Durand, F.; Cramer, W.; Humphreys, S.; Kainuma, M.; Kala, J.; Mahowald, N.; Mulugetta, Y.; et al. Framing and Context. In Global Warming of 1.5 °C. An IPCC Special Report on the Impacts of Global Warming of 1.5 °C above Pre-Industrial Levels and Related Global Greenhouse Gas Emission Pathways, in the Context of Strengthening the Global Response to the Threat of Climate Change, Sustainable Development, and Efforts to Eradicate Poverty. Masson-Delmotte, V.P., Sustainability 2021, 13, 8276 23 of 29 23 of 29 Zhai, H.-O., Pörtner, D., Roberts, J., Skea, P.R., Shukla, A., Pirani, W., Moufouma-Okia, C., Péan, R., Pidcock, S., et al., Eds.; Intergovernmental Panel on Climate Change (IPCC): Geneva, Swizerland, 2018. g g 4. Campagnolo, L.; Marinella, D. Can the Paris deal boost SDGs achievement? An assessment of climate mitigation co-benefits or side-effects on poverty and inequality. World Dev. 2019, 122, 96–109, doi:10.1016/j.worlddev.2019.05.015. 5. Gills, B.; Morgan, J. Global Climate Emergency: After COP24, climate science, urgency, and the threat to humanity. Globali- zations 2020, 17, 885–902, doi:10.1080/14747731.2019.1669915. 6. Smith, P.; Nkem, J.; Calvin, K.; Campbell, D.; Cherubini, F.; Grassi, G.; Korotkov, V.; Hoang, A.L.; Lwasa, S.; McElwee, P.; et al. Interlinkages between Desertification, Land Degradation, Food Security and Greenhouse Gas Fluxes: Synergies, Trade- offs and Integrated Response Options. In Climate Change and Land: An IPCC Special Report on Climate Change, Desertification, Land Degradation, Sustainable Land Management, Food Security, and Greenhouse Gas Fluxes in Terrestrial Ecosystems; Shukla, P.R., Skea, J., Calvo Buendia, E., Masson-Delmotte, V., Portner, H.-O., Roberts, D.C., Zhai, P., Slade, R., Connors, S., van Diemen, R., et al., Eds.; Intergovernmental Panel on Climate Change (IPCC): Geneva, Swizerland 2019. g g y y Skea, J., Calvo Buendia, E., Masson-Delmotte, V., Portner, H.-O., Roberts, D.C., Zhai, P., Slade, R., Connors, S., van Diemen, R., et al., Eds.; Intergovernmental Panel on Climate Change (IPCC): Geneva, Swizerland 2019. 7. de Araujo Barbosa, C.C.; Atkinson, P.M.; Dearing, J.A. Extravagance in the commons: Resource exploitation and the fron- tiers of ecosystem service depletion in the Amazon estuary. Sci. Total Environ. 2016, 550, 6–16, doi:10.1016/j.sci- totenv.2016.01.072. 8. 26. Armitage, D.; Marschke, M.; Plummer, R. Adaptive co-management and the paradox of learning. Glob. Environ. Chang. 2008, 18, 86–98, doi:10.1016/j.gloenvcha.2007.07.002. References Folke, C.; Hahn, T.; Olsson, P.; Norberg, J. Adaptive Governance of Social-Ecological Systems. Annu. Rev. Environ. Resour. 2005, 30, 441–473, doi:10.1146/annurev.energy.30.050504.144511. 33. Chaffin, B.C.; Gosnell, H.; Cosens, B.A. A decade of adaptive governance scholarship: Synthesis and future directions. Ecol. Soc. 2014, 19, 56, doi:10.5751/ES-06824-190356. 34. Folke, C.; Berkes, F. Understanding Dynamics of Ecosystem-Institution Linkages for Building Resilience; Beijer Discussion Paper No. 112; The Beijer Institute of Ecological Economics, Royal Academy of Sciences: Stokholm, Sweden, 1998. j g y y 35. Colloff, M.J.; Wise, R.M.; Palomo, I.; Lavorel, S.; Pascual, U. Nature’s contribution to adaptation: Insig the transformation of social-ecological systems. Ecosyst. People 2020, 16, 137–150, doi:10.1080/26395916 g y y p 36. Sachs, J.D.; Schmidt-Traub, G.; Mazzucato, M.; Messner, D.; Nakicenovic, N.; Rockström, J. Six Transformations to achieve the Sustainable Development Goals. Nat. Sustain. 2019, 2, 805–814, doi:10.1038/s41893-019-0352-9. p 37. Folke, C.; Carpenter, S.; Walker, B.; Scheffer, M.; Elmqvist, T.; Gunderson, L.; Holling, C.S. Regime Shifts, Resilience, and Biodiversity in Ecosystem Management. Annu. Rev. Ecol. Evol. Syst. 2004, 35, 557–581, doi:10.1146/annurev.ecol- sys.35.021103.105711. y 38. Gallopín, G.C. Branch Points: Global Scenarios and Human Choice, 1997 Available online: https://g chives/other/Branch%20Points.pdf (accessed on 1 June 2021). C. Branch Points: Global Scenarios and Human Choice, 1997 Available online: https://greattransition.org Branch%20Points.pdf (accessed on 1 June 2021). 39. Köhler, J.; Geels, F.W.; Kern, F.; Markard, J.; Onsongo, E.; Wieczorek, A.; Alkemade, F.; Avelino, F.; Bergek, A.; Boons, F.; et al. An agenda for sustainability transitions research: State of the art and future directions. Environ. Innov. Soc. Transit. 2019, 31, 1–32, doi:10.1016/j.eist.2019.01.004. j 40. Staniscia, B.; Komatsu, G.; Staniscia, A. Nature Park establishment and environmental conflicts in coa the Costa Teatina National Park in central Italy. Ocean Coast. Manag. 2019, 182, 104947, doi:10.1016/j.oce 40. Staniscia, B.; Komatsu, G.; Staniscia, A. Nature Park establishment and environmental conflicts in coastal areas: The case of the Costa Teatina National Park in central Italy. Ocean Coast. Manag. 2019, 182, 104947, doi:10.1016/j.ocecoaman.2019.104947. 41. Parlee, C.E.; Wiber, M.G. Using conflict over risk management in the marine environment to strengthen measures of gov- ernance. Ecol. Soc. 2018, 23, 5, doi:10.5751/ES-10334-230405. y g j 41. Parlee, C.E.; Wiber, M.G. Using conflict over risk management in the marine environment to strengthen measures of gov- ernance. Ecol. Soc. 2018, 23, 5, doi:10.5751/ES-10334-230405. 42. Butler, J.R.A.; Young, J.C.; McMyn, I.A.G.; Leyshon, B.; Graham, I.M.; Walker, I.; Baxter, J.M.; Dodd, J.; Warburton, C. References Science 2017, 357, eaan1114, doi:10.1126/science.aan1114. p g p y j 20. Bodin, Ö. Collaborative environmental governance: Achieving collective action in social-ecological systems. Science 2017, 357, eaan1114, doi:10.1126/science.aan1114. 21. Cockburn, J.; Schoon, M.; Cundill, G.; Robinson, C.; Aburto, J.A.; Alexander, S.M.; Baggio, J.A.; Barnaud, C.; Chapman, M.; Garcia Llorente, M.; et al. Understanding the context of multifaceted collaborations for social-ecological sustainability: A methodology for cross-case analysis. Ecol. Soc. 2020, 25, 7, doi:10.5751/ES-11527-250307. gy y 22. Barnes, M.L.; Bodin, Ö.; Guerrero, A.M.; McAllister, R.J.; Alexander, S.M.; Robins, G. The social structural foundations of adaptation and transformation in social–ecological systems. Ecol. Soc. 2017, 22, 16, doi:10.5751/ES-09769-220416. 23. Guerrero, A.M.; Bodin, Ö.; McAllister, R.R.; Wilson, K.A. Achieving social-ecological fit through bottom-up collaborative governance: An empirical investigation. Ecol. Soc. 2015, 20, 41, doi:10.5751/ES-08035-200441. 24. Mistry, J.; Berardi, A.; Tschirhart, C.; Bignante, E.; Haynes, L.; Benjamin, R.; Albert, G.; Xavier, R.; Robertson, B.; Davis, O.; et al. Community owned solutions: Identifying local best practices for social-ecological sustainability. Ecol. Soc. 2016, 21, 42, doi:10.5751/ES-08496-210242. 25. Ostrom, E. A General Framework for Analyzing Sustainability of Social-Ecological Systems. Science 2009, 24, 419–422, doi:10.1126/science.1172133. 26. Armitage, D.; Marschke, M.; Plummer, R. Adaptive co-management and the paradox of learning. Glob. Environ. Chang. 2008, 18, 86–98, doi:10.1016/j.gloenvcha.2007.07.002. Sustainability 2021, 13, 8276 24 of 29 24 of 29 27. Folke, C.; Carpenter, S.; Elmqvist, T.; Gunderson, L.; Holling, C.S.; Walker, B. Resilience and sustainable development: Building adaptive capacity in a world of transformations. Ambio 2002, 31, 437–40, doi:10.1579/0044-7447-31.5.437. g p p y 28. OECD. Principles for Evaluation of Development Assistance; 1991. Available online: https://www.oecd.org/dac/evalua- tion/2755284.pdf (accessed on 1 June 2021). p ( ) 29. Secco, L.; Pisani, E.; Da Re, R.; Rogelja, T.; Burlando, C.; Vicentini, K.; Pettenella, D.; Masiero, M.; Miller, D.; Nijkj, M. To- wards a method of evaluating social innovation in forest-dependent rural communities: First suggestions from a science- stakeholder collaboration. For. Policy Econ. 2019, 104, 9–22, doi:10.1016/j.forpol.2019.03.011. y j p 30. Westley, F.R.; Tjornbo, O.; Schultz, L.; Olsson, P.; Folke, C.; Crona, B.; Bodin, Ö. A theory of transformative agency in linked social-ecological systems. Ecol. Soc. 2013, 18, 27, doi:10.5751/ES-05072-180327. 31. Colloff, M.J.; Martín-López, B.; Lavorel, S.; Locatelli, B.; Gorddard, R.; Longarettig, P.Y.; Walters,G.; van Kerkhoff, L.; Wy- born, C.; Coreau, A.; Wise, R.M.; et al. An integrative research framework for enabling transformative adaptation. Environ. Sci. Policy 2017, 68, 87–96, doi:10.1016/j.envsci.2016.11.007. y j 32. References Resilience thinking: Integrating resilience, adaptability and transformability. Ecol. Soc. 2010, 15, 20. p y y 56. Hölscher, K.; Wittmayer, M.; Loorbach, D. Transition versus transformation: What’s the difference? Environ. Innov. Soc. Transit. 2018, 27, 1–3, doi:10.1016/j.eist.2017.10.007. 57. Few, R.; Morchain, D.; Spear, D.; Mensah, A.; Bendapudi, R. Transformation, adaptation and development: Relating con- cepts to practice. Palgrave Commun. 2017, 3, 17092, doi:10.1057/palcomms.2017.92. 58. Andrachuk, M.; Armitage, D. Understanding social-ecological change and transformation through co of system identity. Ecol. Soc. 2015, 20, 26, doi:10.5751/ES-07759-200426. Armitage, D. Understanding social-ecological change and transformation through community perceptions ty. Ecol. Soc. 2015, 20, 26, doi:10.5751/ES-07759-200426. y y 59. Lebel, L.; Anderies, M.; Campbell, B.; Folke, C.; Hatfield-Dodds, S.; Hughes, T.P.; Wilson, J. Governance and the capacity to manage resilience in regional social-ecological systems. Ecol. Soc. 2006, 11, 19. 60. Wittmayer, J.M.; Avelino, F.; van Steenbergen, F.; Loorbach, D. Actor roles in transition: Insights from sociological perspec- tives. Environ. Innov. Soc. Transit. 2017, 24, 45–56, doi:10.1016/j.eist.2016.10.003. j 61. Turner, R.H. Role Change. Annu. Rev. Sociol. 1990, 16, 87–110, doi:10.1146/annurev.so.16.080190.00051 62. Avelino, F.; Wittmayer, J.M. Shifting Power Relations in Sustainability Transitions: A Multi-actor Perspective. J. Environ. Policy Plan. 2016, 18, 628–649, doi:10.1080/1523908X.2015.1112259. y 63. Lemos, M.C.; Agrawal, A. Environmental Governance. Rev. Environ. Resour. 2006, 31, 297–325, doi:10.1146/annurev.en- ergy.31.042605.135621. gy 64. Davies, A.L.; White, R.M. Collaboration in natural resource governance: Reconciling stakeholder expectations in deer man- agement in Scotland. J. Environ. Manag. 2012, 15, 160–169, doi:10.1016/j.jenvman.2012.07.032. 65. Schoon, M.; Van der Leeuw, S. The shift toward social-ecological systems perspectives: Insights into the human-nature relationship. Nat. Sci. Soc. 2015, 23, 166–174, doi:10.1051/nss/2015034. 66. Anderies, J.M.; Folke, C.; Walker, B.; Ostrom, E. Aligning key concepts for global change policy: Robustness, resilience, and sustainability. Ecol. Soc. 2013, 18, 8, doi:10.5751/ES-05178-180208. ewig, J. Importance of Actors and Agency in Sustainability Transitions: A Systematic Exploration of the inability 2016, 8, 476, doi:10.3390/su8050476. 67. Fischer, L.B.; Newig, J. Importance of Actors and Agency in Sustainability Transitions: A Systemat Literature. Sustainability 2016, 8, 476, doi:10.3390/su8050476. y 68. Cash, D.W.; Adger, W.; Berkes, F.; Garden, P.; Lebel, L.; Olsson, P.; Pritchard, L.; Young, O. Scale and cross-scale dynamics: Governance and information in a multilevel world. Ecol. Soc. 2006, 11, 8. 69. Schoon, M.; York, A.; Sullivan, A.; Baggio, J. The emergence of an environmental governance network: The case of the Arizona Borderlands. Reg. Environ. Chang. 2017, 17, 677–689, doi:10.1007/s10113-016-1060-x. 70. OECD. References Eval- uating adaptive co-management as conservation conflict resolution: Learning from seals and salmon. J. Environ. Manag. 2015, 160, 212–225, doi:10.1016/j.jenvman.2015.06.019. j j 43. Butcher, J.R.; Gilchrist, D.J.; Phillimore, J.; Wanna, J. Attributes of effective collaboration: Insights from five case studies in Australia and New Zealand. Policy Des. Pract. 2019, 2, 75–89, doi:10.1080/25741292.2018.1561815. y 44. Nohrstedt, D.; Bodin, Ö. Collective Action Problem Characteristics and Partner Uncertainty as Drivers of Social Tie For- mation in Collaborative Networks. Policy Stud. J. 2019, 48, 1082–1108, doi:10.1111/psj.12309. y J p j 45. Alexander, S.M.; Armitage, D.; Charles, A. Social networks and transitions to co-management in Jamaican marine reserves and small-scale fisheries. Glob. Environ. Chang. 2015, 35, 213–225, doi:10.1016/j.gloenvcha.2015.09.001. Alexander, S.M.; Armitage, D.; Charles, A. Social networks and transitions to co-management in Jamaic and small-scale fisheries. Glob. Environ. Chang. 2015, 35, 213–225, doi:10.1016/j.gloenvcha.2015.09.001. , ; g , ; , g J nd small-scale fisheries. Glob. Environ. Chang. 2015, 35, 213–225, doi:10.1016/j.gloenvcha.2015.09.001. 46. Ingold, K.; Fischer, M. Drivers of collaboration to mitigate climate change: An illustration of Swiss climate policy over 15 years. Glob. Environ. Chang. 2014, 24, 88–98, doi:10.1016/j.gloenvcha.2013.11.021. 47. Nyaga, G.N.; Whipple, J.M. Relationship Quality and Performance Outcomes: Achieving a Sustainable Competitive Ad- vantage. J. Bus. Logist. 2011, 32, 345–360, doi:10.1111/j.0000-0000.2011.01030.x. 48. Langemeyer, J.; Gómez-Baggethun, H.D.; Scheuer, S.D, Elmqvist, T. Bridging the gap between ecosystem service assess- ments and land-use planning through Multi-Criteria Decision Analysis (MCDA). Environ. Sci. Policy 2015. 62, 45–56, doi:10.1016/j.envsci.2016.02.013. j 49. Stacey, N.; Izurieta, A.; Garnett, S.T. Collaborative measurement of performance of jointly managed protected areas in northern Australia. Ecol. Soc. 2013 18, 19, doi:10.5751/ES-05273-180119. 50. Herremans, I.M.; Nazari Mahmoudian, F. Stakeholder Relationships, Engagement, and Sustainability ics 2016, 138, 417–435, doi:10.1007/s10551-015-2634-0. 51. Bodin, Ö.; Mancilla García, M.; Robins, G. Reconciling Conflict and Cooperation in Environmental Governance: A Social Network Perspective. Annu. Rev. Environ. Resour. 2020, 45, 471–495, doi:10.1146/annurev-environ-011020-064352. p , , , 52. Lopes, R.; Videira, N. Bringing stakeholders together to articulate multiple value dimensions of ecosystem services. Ocean. Coast. Manag. 2018, 165, 215–224, doi:10.1016/j.ocecoaman.2018.08.026. g j 53. Dannenberg, A.; Barrett, S. Cooperating to avoid catastrophe. Nat. Hum. Behav. 2018, 2, 435–437, doi:10.1038/s41562-018- 0374-8. Sustainability 2021, 13, 8276 25 of 29 54. Gjorgievski, V.Z.; Cundeva, S.; Georghiou, G.E. Social arrangements, technical designs and impacts of energy communities: A review. Renew. Energ. 2021, 169, 1138–1156, doi:10.1016/j.renene.2021.01.078. g j 55. Folke, C.; Carpenter, S.R.; Walker, B.; Scheffer, M.; Chapin, T.; Rockström, J. References Measuring and Managing Results in Development Co-Operation: A Review of Challenges and Practices among DAC Members and Observers; OECD Publishing: Paris, France, 2014. Available online: https://www.oecd.org/dac/peer-reviews/Measuring- and-managing-results.pdf (accessed on 1 June 2021). g g p 71. Allen, W.; Cruz, J.; Warburton, B. How Decision Support Systems Can Benefit from a Theory of Change Approach. Environ. Manag. 2017, 59, 956–965, doi:10.1007/s00267-017-0839-y. 72. Margoluis, R.; Stem, C.; Swaminathan, V.; Brown, M.; Johnson, A.; Placci, G.; Salafsky, N.; Tilders, I. Results chains: A tool for conservation action design, management, and evaluation. Ecol. Soc. 2013, 18, 22, doi:10.5751/ES-05610-180322. 73. Snyder, H. Literature review as a research methodology: An overview and guidelines. J. Bus. Res. 2019, 104, 333–339, doi:10.1016/j.jbusres.2019.07.039. 74. Berrang-Ford, L.; Pearce, T.; Ford, J.D. Systematic review approaches for climate change adaptation research. Reg. Environ. Chang. 2015, 15, 755–769, doi:10.1007/s10113-014-0708-7. ght, T.M. Effectiveness in conservation practice: Pointers from medicine and public health. Conserv. Biol. doi:10.1046/j.1523-1739.2001.99499.x. 75. Pullin, A.S.; Knight, T.M. Effectiveness in conservation practice: Pointers from medicine and public 2001, 15, 50–54, doi:10.1046/j.1523-1739.2001.99499.x. 76. Bosman, J.; van Mourik, I.; Rasch, M.; Sieverts, E.; Verhoeff, H. Scopus Reviewed and Compared. The Coverage and Functionality of the Citation Database Scopus, Including Comparisons with Web of Science and Google Scholar; Universiteitsbibliotheek Utrecht/Utrecht University Library: Utrecht, The Netherlands, 2006. y y itsouni, E.I., Malietzis, G.A., Pappas, G. Comparison of PubMed, Scopus, Web of Science, and Google hs and weaknesses. FASEB J. 2008, 22, 338-42. doi: 10.1096/fj.07-9492LSF y y 77. Falagas, M.E., Pitsouni, E.I., Malietzis, G.A., Pappas, G. Comparison of PubMed, Scopus, Web of Scholar: strengths and weaknesses. FASEB J. 2008, 22, 338-42. doi: 10.1096/fj.07-9492LSF 78. Chadegani, A.A.; Salehi, H.; Yunus, M.M.; Farhadi, H.; Fooladi, M.; Farhadi, M.; Ebrahim, N.A. A Comparison between Two Main Academic Literature Collections: Web of Science and Scopus Databases. Asian Soc. Sci. 2013, 9, 18–26, doi:10.5539/ass.v9n5p18. p 79. Li, J.; Burnham, J.F.; Lemley, T.; Britton, R.M. Citation Analysis: Comparison of Web of Science®, Scop Google Scholar. J. Electron. Resour. Med Libr. 2013, 7, 196–217, doi:10.1080/15424065.2010.505518. 80. Martín-Martín, A.; Orduna-Malea, E.; Thelwall, M.; López-Cózar, E.D. Google Scholar, Web of Science, and Scopus: A sys- tematic comparison of citations in 252 subject categories. J. Informetr. 2018, 12, 1160–1177, doi:10.1016/j.joi.2018.09.002. p j g f j j 81. Singh, V.K.; Singh, P.; Karmakar, M.; Leta, J.; Mayr, P. The journal coverage of Web of Science, Scopus and Dimensions: A comparative analysis. Scientometrics 2021, 126, 5113–5142, doi:10.1007/s11192-021-03948-5. References p y 82. Baas, J.; Schotten, M.; Plume, A.; Côté, G.; Karimi, R. Scopus as a curated, high-quality bibliometric data source for academic research in quantitative science studies. Quant. Sci. Stud. 2020, 1, 377–386, doi:10.1162/qss_a_00019. Sustainability 2021, 13, 8276 26 of 29 26 of 29 83. Martín-Martín, A.; Thelwall, M.; Orduna-Malea, E.; López-Cózar, E.D. Google Scholar, Microsoft Academic, Scopus, Di- mensions, Web of Science, and OpenCitations’ COCI: A multidisciplinary comparison of coverage via citations. Scientomet- rics 2021, 126, 871–906, doi:10.1007/s11192-020-03690-4. 84. Linnenluecke, M.K.; Marrone, M.; Singh, A.K. Conducting systematic literature reviews and bibliometric analyses. Aust. J. Manag. 2020, 45, 175–194, doi:10.1177/0312896219877678. g 85. Ose, S.O. Using Excel and Word to Structure Qualitative Data. J. Appl. Soc. Sci. 2016, 10, 147–162, doi:10.1177/1936724416664948. 86. Schmidt, M. The Sankey Diagram in Energy and Material Flow Management. Part II: Methodology and Current Applica- tions. J. Ind. Ecol. 2008, 173–185, doi:10.1111/j.1530-9290.2008.00015.x. 87. Ness, B.; Urbel-Piirsalu, E.; Anderberg, S.; Olsson, L. Categorising tools for sustainability assessment. Ecol. Econ. 2007, 60, 498–508, doi:10.1016/j.ecolecon.2006.07.023. 88. UNEP. Global Environment Outlook. GEO-6. Healthy Planet, Healthy People; Cambridge University Press: Cambridge, UK, 2019; doi:10.1017/9781108627146. 89. Etxano, I.; Garmendia, E.; Pascual, U.; Hoyos, D.; Díez, M.A.; Cadiñanos, J.; Lozano, P.J. A participatory integrated assess- ment approach for Natura 2000 network sites. Environ. Plan. C Gov. Policy 2015, 33, 1207–1232, doi:10.1177/0263774X15612318. 90. Smedstad, J.A.; Gosnell, H. Do adaptive comanagement processes lead to adaptive comanagement outcomes? A multicase study of long-term outcomes associated with the National Riparian Service Team’s place-based riparian assistance. Ecol. Soc. 2013, 18, 8, doi:10.5751/ES-05793-180408. , , , / 91. Kimario, F.F.; Botha, N.; Kisingo, A.; Job, H. Theory and practice and practice of conservancies: Evidence from wildlife management areas in Tanzania. Erdkunde 2020, 74, 117–141, doi:10.3112/erdkunde.2020.02.03. 92. Bergquist, D.A.; Cavalett, O.; Rydberg, T. Participatory emergy synthesis of integrated food and biofuel production: A case study from Brazil. Environ. Dev. Sustain. 2012, 14, 167–182, doi:10.1007/s10668-011-9314-8. 93. Waylen, K.A.; Blackstock, K.L.; van Hulst, F.J.; Damian, C.; Horváth, F.; Johnson, R.K.; Kanka, R.; Külvik, M.; Macleo, C.J.A.; Meissner, K.; et al. Policy-driven monitoring and evaluation: Does it support adaptive management of socio-ecological sys- tems? Sci. Total Environ. 2019, 662, 373–384, doi:10.1016/j.scitotenv.2018.12.462. 94. Stephenson, R.L.; Wiber, S.P.M.; Angel, E.; Benson, A.J.; Charles, A.; Chouinard, O.; Dan Edwards, M.D.; Foley, P.; Jennings, L.; Jones, O.; et al. Evaluating and implementing social–ecological systems: A comprehensive approach to sustainable fish- eries. Fish Fish. 2018, 19, 853–873, doi:10.1111/faf.12296. References 95. Söderberg, C. Complex governance structures and incoherent policies: Implementing the EU water framework directive in Sweden. J. Environ. Manag. 2016, 183, 90–97, doi:10.1016/j.jenvman.2016.08.040. 96. Gerhardinger, L.C.; Godoy, E.A.S.; Jones, P.J.S.; Sales, G.; Ferreira, B.P. Marine Protected Dramas: The Flaws of the Brazilian National System of Marine Protected Areas. Environ. Manag. 2014, 47, 630–643, doi:10.1007/s00267-010-9554-7. y g 97. Clark, T.W.; Padwe, J. The Ecuadorian Condor Bioreserve Initiative. J. Sustain. For. 2004, 18, 297–324, doi:10.1300/J091v18n02_14. 98. Johnson, F.A.; M J Eaton, J.M.; Case, D. Building adaptive capacity in a coastal region experiencing glo 2020, 25, 9, doi:10.5751/ES-11700-250309. 99. Thompson, S.T.; Friess, D.A. Stakeholder preferences for payments for ecosystem services (PES) versus other environmental management approaches for mangrove forests. J. Environ. Manag. 2019, 233, 636–648, doi:10.1016/j.jenvman.2018.12.032. g pp g g j j 100. Petursdottir, T.; Arnalds, O.; Baker, S.; Montanarella, L.; Aradóttir, Á. A social–ecological system approach to analyze stake- holders’ interactions within a large-scale rangeland restoration program. Ecol. Soc. 2013, 18, 29, doi:10.5751/ES-05399-180229. 100. Petursdottir, T.; Arnalds, O.; Baker, S.; Montanarella, L.; Aradóttir, Á. A social–ecological system approach to analyze stake- holders’ interactions within a large-scale rangeland restoration program. Ecol. Soc. 2013, 18, 29, doi:10.5751/ES-05399-180229. 101. de Alencar, N.M.P.; Le Tissier, M.; Paterson, S.K.; Newton, A. Circles of Coastal Sustainability: A Framework for Coastal Management. Sustainability 2020, 12, 4886, doi:10.3390/su12124886. g g p g , , , 101. de Alencar, N.M.P.; Le Tissier, M.; Paterson, S.K.; Newton, A. Circles of Coastal Sustainability: A Framework for Coastal Management. Sustainability 2020, 12, 4886, doi:10.3390/su12124886. g y 102. Schouten, M.A.H.; van der Heide, M.; Heijman, W.J.M.; Opdam, P.F.M. A resilience-based policy evaluation framework: Application to European rural development policies. Ecol. Econ. 2012, 81, 165–175, doi:10.1016/j.ecolecon.2012.07.004. 103. Li, J.; Pan, S.-Y.; Kim, H.; Linn, J.H.; Chiang, P.-C. Building green supply chains in eco-industrial parks towards a green economy: Barriers and strategies. J. Environ. Manag. 2015, 162, 158–170, doi:10.1016/j.jenvman.2015.07.030. 105. Jennings, S.; Pascoe, S.; Hall-Aspland, S.; Bouhellec, B.; Norman-Lopez, A.; Sullivan, A.; Pecl, G. Setting objectives for eval- uating management adaptation actions to address climate change impacts in south-eastern Australian fisheries. Fish. Ocean- ogr. 2016, 25, 29–44, doi:10.1111/fog.12137. g g 106. Treemore-Spears, L.J.; Grove, J.M.; Harris, C.K.; Lemke, L.D.; Miller, C.J.; Pothukuchi, K.; Zhang, Y.; Zhang, Y.L. A work- shop on transitioning cities at the food-energy-water nexus. J. Environ. Stud. Sci. 2016, 6, 90–103, doi:10.1007/s13412-016- 0381-x. 107. Nuno, A.; Bunnefeld, N.; Milner-Gulland, E. References Managing social–ecological systems under uncertainty: Implementation in the real world. Ecol. Soc. 2014, 19, 52, doi:10.5751/ES-06490-190252. 108. Robinson, C.J.; Bark, R.H.; Garrick, D.; Pollino, C.A. Sustaining local values through river basin governance: Community- based initiatives in Australia's Murray–Darling basin. J. Environ. Plan. Manag. 2014, 58, 2212–2227, doi:10.1080/09640568.2014.976699. 27 of 29 27 of 29 Sustainability 2021, 13, 8276 109. Benitez-Capistros, F.; Hugé, J.; Koedama, N. Environmental impacts on the Galapagos Islands: Identificationof interactions, perceptions and steps ahead. Ecol. Indic. 2014, 38, 113–123, doi:10.1016/j.ecolind.2013.10.019. p p p j 110. Schultz, L.; Lundholm, C. Learning for resilience? Exploring learning opportunities in biosphere reserves. Environ. Educ. Res. 2010, 16, 645–663, doi:10.1080/13504622.2010.505442. 111. Jones, O.P.; Stephenson, R.L. Practical use of full-spectrum sustainability in the Bay of Fundy. Ecol. Soc. 2019, 24, 25, doi:10.5751/ES-11010-240325. 112. Gilioli, G.; Tikubet, G.; Herren, H.R.; Baumgärtner, J. Assessment of social–ecological transitions in a peri-urban Ethiopian farming community. Int. J. Agric. Sustain. 2014, doi:10.1080/14735903.2014.954452. 113. Gillon, S.; Booth, E.G.; Rissman, A.R. Shifting drivers and static baselines in environmental govern improving and proving water quality outcomes. Reg. Environ. Chang. 2015, doi:10.1007/s10113-015-078 114. Brown, P.R.; Jacobs, B.; Leith, P. Participatory monitoring and evaluation to aid investment in natural resource manager capacity at a range of scales. Environ. Monit. Assess. 2012, 184, 7207–7220, doi:10.1007/s10661-011-2491-y. p y g y 115. Chu, J.; Garlock, T.M.; Sayon, P.; Asche, F.; Anderson, J.L. Impact evaluation of a fisheries development project. Mar. Policy 2017, 85, 141–149, doi:10.1016/j.marpol.2017.08.024. 116. Waylen, K.A.; Blackstock, K.L. Monitoring for Adaptive Management or Modernity: Lessons from recent initiatives for ho- listic environmental management. Environ. Policy Gov. 2017, 27, 311–324, doi:10.1002/eet.1758. g y 117. Lin, G.; Wu, B.; Lin, X.; Fan, A.; Tian, S. Ecological Study on the Index System and Methodology of Performance Quantiza- tion for Sustainable Forest Management. Ekoloji 2019, 28, 1365–1372. 118. Nilsson, A.K.; Bohman, B. Legal prerequisites for ecosystem-based management in the Baltic Sea are trophication. Ambio 2015, 44, S370–S380, doi:10.1007/s13280-015-0656-6. p 119. Pearson, J.; Collins, K. Does social-ecological context influence state-based water management decisions? Case study from Queensland, Australia (1980–2006). Water Policy 2010, 12, 186–202, doi:10.2166/wp.2009.055. Q , ( ) y , , , p 120. He, R.; Tang, Z.; Dong, Z.; Wang, S. Performance Evaluation of Regional Water Environment Integrated Governance: Case Study from Henan Province, China. Int. J. Environ. Res. Public Health 2020, 17, 2501, doi:10.3390/ijerph17072501. 121. Liu, B.; Wang, J.; Jing, Z.; Tang, Q. References Measurement of sustainable transformation capability of resource-based cities based on fuzzy membership function: A case study of Shanxi Province, China. Resour. Policy 2020, 68, 101739, doi:10.1016/j.resour- pol.2020.101739. p 122. Bundy, A.; Chuenpagdee, R.; Boldt, J.L.; de Fatima Borges, M.; Camara, M.L.; Coll, M.; Diallo, I.; Clive Fox, C.; Fulton, E.A.; Gazihan, A.; et al. Strong fisheries management and governance positively impact ecosystem status. Fish Fish. 2017, 18, 412– 439, doi:10.1111/faf.12184. 123. Oviedo, A.F.P.; Bursztyn, M. The Fortune of the Commons: Participatory Evaluation of Small-Scale Fis Amazon. Environ. Manag. 2016, 5, 1009–1023, doi:10.1007/s00267-016-0660-z. 124. Marshall, G.R. Transaction costs, collective action and adaptation in managing complex social–ecological systems. Ecol. Econ. 2013, 88, 185–194, doi:10.1016/j.ecolecon.2012.12.030. j 125. Thiel, A.; Schleyer, C.; Hinkel, J.; Schlüter, M.; Hagedorn, K.; Bisaro, S.; Bobojonov, I.; Hamidov, A. Transferring William- son's discriminating alignment to the analysis of environmental governance of social-ecological interdependence. Ecol. Econ. 2016, 128, 159–168, doi:10.1016/j.ecolecon.2016.04.018. j 126. Ancuta, C.; Olaru, M.; Popa, N.; Isfanescu, R.; Jigoria-Oprea, L. Evaluation of the sustainable development of rural settle- ments. Case study: Rural settlement from romanian Banat. Carpathian J. Earth Environ. Sci. 2015, 10, 67–80. y p 127. Sheng, R.; Lin, T. Evolutionary Assessment of the Ecological Governance under the Metropolitan Background: Evidence from Chongming Eco-Island, Shanghai, China. Sustainability 2019, 11, 5327, doi:10.3390/su11195327. from Chongming Eco-Island, Shanghai, China. Sustainability 2019, 11, 5327, doi:10.3390/su11195327. 128. Koenigstein, S.; Ruth, M.; Gößling-Reisemann, S. Stakeholder-Informed Ecosystem Modeling of Ocean Warming and Acid- ification Impacts in the Barents Sea Region. Front. Mar. Sci. 2016, 3, 93, doi:10.3389/fmars.2016.00093. g g , g , y , , , 128. Koenigstein, S.; Ruth, M.; Gößling-Reisemann, S. Stakeholder-Informed Ecosystem Modeling of Ocean Warming and Acid- ification Impacts in the Barents Sea Region. Front. Mar. Sci. 2016, 3, 93, doi:10.3389/fmars.2016.00093. 129. Dressel, S.; Ericsson, G.; Sandström, C. Mapping social-ecological systems to understand the challenges underlying wildlife management. Environ. Sci. Policy 2018, 84, 105–112, doi:10.1016/j.envsci.2018.03.007. g y j 130. Shkaruba, A.; Kireyeu, V. Recognizing ecological and institutional landscapes in adaptive governance of natural resources. For. Policy Econ. 2013, 36, 87–97, doi:10.1016/j.forpol.2012.10.004. 131. Uchiyama, Y.; Kohsaka, R. Application of the City Biodiversity Index to populated cities in Japan: Influence of the social a d e olo i al ha a te i ti o i di ato ba ed a a e e t E ol I di 2019 106 105420 130. Shkaruba, A.; Kireyeu, V. p y 135. Luisetti, T.; Turner, R.K.; Jickells, T.; Andrews, J.; Elliott, M.; Schaafsma, M.; Beaumont, N.; Malcolm, S.; Burdon, D.; Adams, C.; et al. Coastal Zone Ecosystem Services: From science to values and decision making; a case study. Sci. Total Environ. 2014, 493, 682–693, doi:10.1016/j.scitotenv.2014.05.099. References Research pathways to foster transformation: Linking sustaina- bility science and social-ecological systems research. Ecol. Soc. 2020, 25, 13, doi:10.5751/ES-11332-250113. 138. Holzer, J.M.; Adamescu, C.M.; Cazacu, C.; Díaz-Delgado, R.; Dick, J.; Méndez, P.F.; Santamaría, L.; Orenstein, D.E. Evalu- ating transdisciplinary science to open research-implementation spaces in European social-ecological systems. Biol. Conserv. 2019, 238, 108228, doi:10.1016/j.biocon.2019.108228. j 139. Wyborn, C.; Bixler, R.P. Collaboration and nested environmental governance: Scale dependency, scale framing, and cross- scale interactions in collaborative conservation. J. Environ. Manag. 2013, 15, 58–67, doi:10.1016/j.jenvman.2013.03.014. 140. Ostrom, E. Governing the Commons: The Evolution of Institutions for Collective Action; Cambridge University Press: Cambridge, Uk, 1990. 141. Linstädter, A.; Kuhn, A.; Naumann, C.; Rasch, S.; Sandhage-Hofmann, A.; Amelung, W.; Jordaan, J.; Du Preez, C.C.; Bollig, M. Assessing the resilience of a real-world social-ecological system: Lessons from a multidisciplinary evaluation of a South African pastoral system. Ecol. Soc. 2016, 21, 35, doi:10.5751/ES-08737-210335. p y 142. Sanon, S.; Hein, T.; Douven, W.; Winkler, P. Quantifying ES trade-offs: The case of an urban floodplain in Vienna, Austria. J. Environ. Manag. 2012, 111, 159–172, doi:10.1016/j.jenvman.2012.06.008. 143. Pisani, E.; Andriollo, E.; Masiero, M.; Secco, L. Intermediary Organisations in Collaborative Environmental Governance: Evidence of the EU-funded LIFE Sub-Programme for the Environment (LIFE-ENV). Heliyon 2020, 4 e04251, doi:10.1016/j.heliyon.2020.e04251. j y 144. Hallinger, P.; Chatpinyakoop, C. A Bibliometric Review of Research on Higher Education for Sustainable Development, 1998–2018. Sustainability 2019, 11, 2401, doi:10.3390/su11082401. 145. Xu, L.; Marinova, D. Resilience thinking: A bibliometric analysis of socio-ecological research. Scientometrics 2013, 96, 911– 927, doi:10.1007/s11192-013-0957-0. 146. FAO. The State of Food Security and Nutrition in the World 2020. Transforming Food Systems for Affordable Healthy Diets; 2020. Available online: http://www.fao.org/3/ca9692en/CA9692EN.pdf (accessed on 1 June 2021). 147. UN DESA. World Population Prospects. The 2015 Revision. Key Findings and Advance Tables; 2015. Availab ulation.un.org/wpp/publications/files/key_findings_wpp_2015.pdf (accessed on 1 June 2021). 148. Schebesta, H.; Candel, J.J.L. Game-changing potential of the EU’s Farm to Fork Strategy. Nat. Food 2020, 1, 586–588, doi:10.1038/s43016-020-00166-9. 149. Ramankutty, N.; Mehrabi, Z.; Waha, K.; Jarvis, L.; Kremen, C.; Herrero, M.; Rieseberg, L.H. Trends in Global Agricultural Land Use: Implications for Environmental Health and Food Security. Annu. Rev. Plant Biol. 2018, 69, 789–815, doi:10.1146/an- nurev-arplant-042817-040256. p 150. FAO. Building a Common Vision for Sustainable Food and Agriculture. Principles and Approaches; 2014. Available online: http://www.fao.org/3/a-i3940e.pdf (accessed on 1 June 2021). p g p 151. Hossu, C.A.; Ioja, I.; Nita, M.R.; Hartel, T.; Badiu, D.L.; Hersperger, A.M. References Recognizing ecological and institutional landscapes in adaptive governance of natural resources. For. Policy Econ. 2013, 36, 87–97, doi:10.1016/j.forpol.2012.10.004. 131. Uchiyama, Y.; Kohsaka, R. Application of the City Biodiversity Index to populated cities in Japan: Influence of the social and ecological characteristics on indicator-based management. Ecol. Indic. 2019, 106, 105420, doi:10.1016/j.ecolind.2019.05.051. j 132. Forster, J.; Turner, R.A.; Fitzsimmons, C.; Angeli, M.; Peterson, A.M.; Mahon, R.; Steada, S.M. Evidence of a common un- derstanding of proximate and distal drivers of reef health. Mar. Policy 2017, 84, 263–272, doi:10.1016/j.marpol.2017.07.017. g p y j 133. Wu., G.; Duan, K.; Zuo, J.; Zhao, X.; Tang, D. Integrated Sustainability Assessment of Public Rental Based on a Hybrid Method of AHP-Entropy Weight and Cloud Model. Sustainability 2017, 9, 603, doi: 133. Wu., G.; Duan, K.; Zuo, J.; Zhao, X.; Tang, D. Integrated Sustainability Assessment of Public Rental Housing Community Based on a Hybrid Method of AHP-Entropy Weight and Cloud Model. Sustainability 2017, 9, 603, doi:10.3390/su9040603. 134. Foley, P.; Okyere, D.A.; Mather, C. Alternative environmentalities: Recasting the assessment of Canada’s first Marine Stew- ardship Council-certified fishery in social terms. Ecol. Soc. 2018, 23, 37, doi:10.5751/ES-10382-230337. Based on a Hybrid Method of AHP Entropy Weight and Cloud Model. Sustainability 2017, 9, 603, doi:10.3390/su9040603. 134. Foley, P.; Okyere, D.A.; Mather, C. Alternative environmentalities: Recasting the assessment of Canada’s first Marine Stew- ardship Council-certified fishery in social terms. Ecol. Soc. 2018, 23, 37, doi:10.5751/ES-10382-230337. 135. Luisetti, T.; Turner, R.K.; Jickells, T.; Andrews, J.; Elliott, M.; Schaafsma, M.; Beaumont, N.; Malcolm, S.; Burdon, D.; Adams, C.; et al. Coastal Zone Ecosystem Services: From science to values and decision making; a case study. Sci. Total Environ. 2014, 493, 682–693, doi:10.1016/j.scitotenv.2014.05.099. Sustainability 2021, 13, 8276 28 of 29 28 of 29 136. Sparrevik, M.; Breedveldy, G.D. From Ecological Risk Assessments to Risk Governance: Evaluation of the Norwegian Man- agement System for Contaminated Sediments. Integr. Environ. Assess. Manag. 2009, 6, 240–248, doi:10.1897/IEAM_2009-049.1. 137. Horcea-Milcu, A.I.; Martín-López, B.; Lam, D.P.; Lang, D.J. Research pathways to foster transformation: Linking sustaina- bility science and social-ecological systems research. Ecol. Soc. 2020, 25, 13, doi:10.5751/ES-11332-250113. 136. Sparrevik, M.; Breedveldy, G.D. From Ecological Risk Assessments to Risk Governance: Evaluation of the Norwegian Man- agement System for Contaminated Sediments. Integr. Environ. Assess. Manag. 2009, 6, 240–248, doi:10.1897/IEAM_2009-049.1. g y g g 137. Horcea-Milcu, A.I.; Martín-López, B.; Lam, D.P.; Lang, D.J. References Need for a cross-sector approach in protected area management. Land Use Policy 2017, 69, 586–597, doi:10.1016/j.landusepol.2017.10.012. 152. Roux, D.; Ashton, P.; Nel, J.; MacKay, H. Improving Cross-Sector Policy Integration and Cooperation in Support of Fresh- water Conservation. Conserv. Biol. 2008, 22, 1382–1387, doi:10.1111/j.1523-1739.2008.01080.x. 153. Talmage, C.; Knopf, R.C. Rethinking Diversity, Inclusion, and Inclusiveness: The Quest to Better Understand Indicators of Community Enrichment and Well-Being. In New Dimensions in Community Well-Being. Community Quality-of-Life and Well- Being; Kraeger, P., Cloutier, S., Talmage, C., Eds.; Springer: Cham, Switzerland, 2017; doi:10.1007/978-3-319-55408-2_2. y g y g y Q y f f Being; Kraeger, P., Cloutier, S., Talmage, C., Eds.; Springer: Cham, Switzerland, 2017; doi:10.1007/978-3-319-55408-2_2. 154. Griggs, D.; Stafford-Smith, M.; Gaffney, O.; Rockström, J.; Öhman, M.C.; Shyamsundar, P.; Steffen, W.; Glaser, G.; Kanie, N ; Noble I Sustainable development goals for people and planet Nature 2013 495 305 307 doi:10 1038/495305a Being; Kraeger, P., Cloutier, S., Talmage, C., Eds.; Springer: Cham, Switzerland, 2017; doi:10.1007/978-3-319-55408-2_2. 154. Griggs, D.; Stafford-Smith, M.; Gaffney, O.; Rockström, J.; Öhman, M.C.; Shyamsundar, P.; Steffen, W.; Glaser, G.; Kanie, N.; Noble, I. Sustainable development goals for people and planet. Nature 2013, 495, 305–307, doi:10.1038/495305a. 154. Griggs, D.; Stafford-Smith, M.; Gaffney, O.; Rockström, J.; Öhman, M.C.; Shyamsundar, P.; Steffen, W.; Glaser, G.; Kanie, N.; Noble, I. Sustainable development goals for people and planet. Nature 2013, 495, 305–307, doi:10.1038/495305a. p g p p p 155. UN. Resolution adopted by the General Assembly on 25 September 2015. A/RES/70/1; 2015. Available online: https://www.un.org/en/development/desa/population/migration/generalassembly/docs/globalcom- pact/A_RES_70_1_E.pdf (accessed on 1 June 2021). 156. UN. Report of the Conference of the Parties on Its Twenty-First Session, Held in Paris from 30 November to 13 December 2015. FCCC/CP/2015/10. 2015. Available online: https://unfccc.int/resource/docs/2015/cop21/eng/10.pdf (accessed on 1 June 2021). 157. EC. Communication from the Commission to the European Parliament, the European Council, the Council, the European Economic and Social Committee and the Committee of the Regions. The European Green Deal. 2019. Available online: https://eur-lex.europa.eu/resource.html?uri=cellar:b828d165-1c22-11ea-8c1f-01aa75ed71a1.0002.02/DOC_1&format=PDF (accessed on 1 June 2021). 158. L. 172/53. Regulation (EU) 20217783 of the European Parliament and of the Council of 29 April 2021 Establishing a Pro- gramme for the Environment and Climate Action (LIFE), and Repealing Regulation (EU) No 1293/2013. Available online: https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:32021R0783&from=EN (accessed on 1 June 2021). https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:32021R0783&from=EN (accessed on 1 June 2021). 159. EC. Interreg Europe 2014-2020 CCI 2014 TC 16 RFIR 001 Cooperation Programme Document. Available online: https://www.in- terregeurope.eu/fileadmin/user_upload/documents/Interreg_Europe_-_CP_final.pdf (accessed on 1 June 2021). 160. g g y 162. Gerlak, A.K.; Heikkila, T.; Newig, J. Learning in environmental governance: Opportunities for translat J. Environ. Policy Plan. 2020, 22, 653–666, doi:10.1080/1523908X.2020.1776100. 161. Stringer, L.C.; Dougill, A.J.; Fraser, E.; Hubacek, K.; Prell, C.; Reed, M.S. Unpacking “participation” in the adaptive man- agement of social–ecological systems: A critical review. Ecol. Soc. 2006, 11, 39. J y , , , 163. de Bisthoven, L.J.; Vanhove, M.; Rochette, A.-J.; Hugé, J.; Luc Brendonck, L. Stakeholder Analysis on Ecosystem Services of Lake Manyara Sub-basin (Tanzania): How to Overcome Confounding Factors. Environ. Manag. 2021, doi:10.1007/s00267- 021-01466-x. g y , , A.K.; Heikkila, T.; Newig, J. Learning in environmental governance: Opportunities for translating theory to p k, A.K.; Heikkila, T.; Newig, J. Learning in environmental governance: Opportunities for translating theory iron. Policy Plan. 2020, 22, 653–666, doi:10.1080/1523908X.2020.1776100. 162. Gerlak, A.K.; Heikkila, T.; Newig, J. Learning in environmental governance: Opportunities for translating theory to practice. J. Environ. Policy Plan. 2020, 22, 653–666, doi:10.1080/1523908X.2020.1776100. ikkila, T.; Newig, J. Learning in environmental governance: Opportunities for translating theory to practice Plan. 2020, 22, 653–666, doi:10.1080/1523908X.2020.1776100. References Coy, D.; Malekpour, S.; Saer, A.K.; Dargaville, R. Rethinking community empowerment in the energy transformation: A p p g ( ) 159. EC. Interreg Europe 2014-2020 CCI 2014 TC 16 RFIR 001 Cooperation Programme Document. Available online: https://www.in- terregeurope.eu/fileadmin/user_upload/documents/Interreg_Europe_-_CP_final.pdf (accessed on 1 June 2021). g p p g p p 160. Coy, D.; Malekpour, S.; Saer, A.K.; Dargaville, R. Rethinking community empowerment in the energy transformation: A critical review of the definitions, drivers and outcomes. Energy Res. Soc. Sci. 2021, 72, 101871, doi:10.1016/j.erss.2020.101871. Sustainability 2021, 13, 8276 29 of 29 29 of 29
https://openalex.org/W2729623630	https://europepmc.org/articles/pmc5642502?pdf=render	English	null	Identification of two novel biomarkers of rectal carcinoma progression and prognosis via co-expression network analysis	Oncotarget	2,017	cc-by	10,374	Min Sun1,, Taojiao Sun2,, Zhongshi He3 and Bin Xiong1 Min Sun1,, Taojiao Sun2,, Zhongshi He3 and Bin Xiong1 Department of Oncology, Zhongnan Hospital of Wuhan University, Hubei Key Laboratory of Tumo Hubei Cancer Clinical Study Center, Wuhan 430071, P.R. China Department of Stomatology, Zhongnan Hospital of Wuhan University, Wuhan 430071, P.R. China 2Department of Stomatology, Zhongnan Hospital of Wuhan University, Wuhan 430071, P.R. China 2Department of Stomatology, Zhongnan Hospital of Wuhan University, Wuhan 430071, P.R. China Department of Radiation and Medical Oncology, Zhongnan Hospital of Wuhan University, Wuhan 43 3Department of Radiation and Medical Oncology, Zhongnan Hospital of Wuhan University, Wuhan 430071, P.R. China These authors have contributed equally to this work 3Department of Radiation and Medical Oncology, Zhongnan Hospital of Wuhan University, Wuhan 430071, P.R. China These authors have contributed equally to this work Correspondence to: Bin Xiong, email: binxiong1961@whu.edu.cn Correspondence to: Bin Xiong, email: binxiong1961@whu.edu.cn Keywords: rectal cancer, The Cancer Genome Atlas, weighted gene co-expression network analysis, prognosis, disease progression Received: March 06 2017 Accepted: May 22 2017 Published: June 27 2017 Correspondence to: Bin Xiong, email: binxiong1961@whu.edu.cn Keywords: rectal cancer, The Cancer Genome Atlas, weighted gene co-expression network analysis, prognosis, disease progression Received: March 06, 2017 Accepted: May 22, 2017 Published: June 27, 2017 Keywords: rectal cancer, The Cancer Genome Atlas, weighted gene co-expression network analysis, prog Received: March 06, 2017 Accepted: May 22, 2017 Published: June 27, 2017 Copyright: Sun et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License 3.0 (CC BY 3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. (CC BY 3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. ABSTRACT mRNA expression profiles provide important insights on a diversity of biological processes involved in rectal carcinoma (RC). Our aim was to comprehensively map complex interactions between the mRNA expression patterns and the clinical traits of RC. We employed the integrated analysis of five microarray datasets and The Cancer Genome Atlas rectal adenocarcinoma database to identify 2118 consensual differentially expressed genes (DEGs) in RC and adjacent normal tissue samples, and then applied weighted gene co-expression network analysis to parse DEGs and eight clinical traits in 66 eligible RC samples. A total of 16 co-expressed gene modules were identified. The green-yellow and salmon modules were most appropriate to the pathological stage (R = 0.36) and the overall survival (HR =13.534, P = 0.014), respectively. A diagnostic model of the five pathological stage hub genes (SCG3, SYP, CDK5R2, AP3B2, and RUNDC3A) provided a powerful classification accuracy between localized RC and non-localized RC. We also found increased Secretogranin III (SCG3) expression with higher pathological stage and poorer prognosis in the test and validation set. The increased Homer scaffolding protein 2 (HOMER2) expression with the favorable survival prediction efficiency significantly correlated with the markedly reduced overall survival of RC patients and the higher pathological stage during the test and validation set. Our findings indicate that the SCG3 and HOMER2 mRNA levels should be further evaluated as predictors of pathological stage and survival in patients with RC. www.impactjournals.com/oncotarget www.impactjournals.com/oncotarget/ www.impactjournals.com/oncotarget/ INTRODUCTION for RC, especially for modules having a strong correlation between genes with similar expression patterns, to predict the pathological stage and survival outcome, which could help to understand disease pathogenesis and provide personalized treatment, have been rarely reported. However, previous research based on examining genetic mutations and different expression patterns associating with colorectal carcinogenesis has largely ignored the relationship between the genes and clinical characteristics [4, 5]. In addition, RC progression involves several critical stages, although most studies have only evaluated the Accurate staging of rectal cancer (RC) is essential for carrying out precise therapies that increase the survival rate of patients [1]. Previous studies have found no differences in specific clinicopathological risk factors between patients with a high or low risk of RC progression [2]. Recent technological breakthroughs in genome-wide sequencing have shed new insights on deregulated mRNAs that have been identified and characterized in the past few years [3]. Genetic biomarkers www.impactjournals.com/oncotarget Oncotarget 69594 Table 1: Quality control results of RC in the different datasets Table 1: Quality control results of RC in the different datasets No. Study IQC EQC CQCg CQCp AQCg AQCp Rank 1 GSE75548 3.92 2 1.8* 2.23 1.9* 1.13* 1.83 2 GSE12225 4.74 2 0.68* 1.66* 1.08* 2.38 2.00 3 GSE35982 5.62 2 0.02* 10.03 0* 0.07* 2.67 4 GSE34472 1.3* 0.4* 0.24* 0.36* 0.56* 0.11* 3.50 RC, rectal cancer; GSE, GEO dataset; IQC, internal quality control indexes; EQC, external quality control indexes; CQCg and CQCp, consistency of differential expression quality control indexes for genes and pathways; AQCg and AQCp, accuracy quality control indexes for genes and pathways. * P-value not significant after Bonferroni correction. RC, rectal cancer; GSE, GEO dataset; IQC, internal quality control indexes; EQC, external quality control indexes; CQCg and CQCp, consistency of differential expression quality control indexes for genes and pathways; AQCg and AQCp, accuracy quality control indexes for genes and pathways. * P-value not significant after Bonferroni correction. colon cancer, there was very limited mRNA expression data relating to rectal cancer. We investigated and manually curated four public datasets (GSE12225, GSE34472, GSE35982, and GSE75548) as the training set (Supplementary Table 1) [12–15]. differences between RC and adjacent normal tissue (ANT), regardless of the intermediate stages [6]. Therefore, there is an urgency to include stage and prognostic predictive modules to the current staging system, which could be achieved by combining information on the clinical characteristics and validated gene-specific biomarkers. INTRODUCTION For these reasons, we aimed to establish comprehensive mRNA expression patterns of the module genes and clinical traits, especially those relating to multi-stage disease progression which directly affects the prognosis through weighted gene co-expression network analysis (WGCNA) [7]. The training set calculated the six quantitative quality control (QC) measures by standardized mean ranks and principal component analysis (PCA) biplots within MetaQC (Table 1 and Figure 1A) [16]. GSE34472 was detected in low quality RC samples and was omitted from the training set [17]. The training set finally consisted of 65 RC samples and 42 ANTs, and it had 8153 gene symbols in common. A total of 6603 gene symbol sets passed the filtering criteria. We identified 4091 mRNAs that were consistently DEGs using the moderated t test, Fisher’s method by summarizing -log(p-value) across studies and running 300 permutations for the meta-analysis to infer the P-values (Figure 1B) [18]. As expected, hierarchical clustering of the three datasets using the remaining 4091 DEGs distinguished RC from ANT samples (Figure 2A). WGCNA results in the construction of free-scale gene co-expression networks that explore the relationships between the gene sets and clinical features [8]. WGCNA groups prerogative genes into modules based on their co- expression similarities and analogous functions across a population of samples [9–11]. In the present work, we carried out the pooled analysis of RC mRNA raw microarrays and the network level analysis of The Cancer Genome Atlas (TCGA) rectal adenocarcinoma (READ) data to decipher the relationships between the module genes and clinical traits. To the best of our knowledge, we are the first group to use meta-analysis and WGCNA to identify modules displaying a nominal evidence association with RC clinical traits. By characterizing module content and topology, we identified clinical traits, modules, and network concepts that play important roles in the regulation of RC at the level of the differentially expressed genes (DEGs). Moreover, to define the prognostic value of the cancer-specific module, which was related to tumor progression, further analysis was performed to validate candidate markers by combining survival analysis with an independent validation cohort. The raw level-3 RNAseq data from 20,501 mRNAs in 95 RC samples and 10 ANT samples were downloaded from TCGA [1, 8, 10]. A total of 2177 DEGs were identified by linear models within the microarray analysis (LIMMA), among which 1092 were up-regulated and 1085 were down-regulated in RC versus ANT (Supplementary Table 2). INTRODUCTION This mRNA signature allowed for the separation of RC samples from ANT samples in the 2-way hierarchical cluster (Figure 2B) and in the PCA plot (Figure 2C). A total of 2118 overlapping DEGs were chosen by both training and test sets (Figure 2D). RESULTS We included 2118 DEGs from 70 RC patients with complete clinical traits and prognostic information of the TCGA-READ test set to construct co-expression networks via WGCNA. After four outlier samples were discarded, the connections between the genes in the gene network were in line with a scale-free network distribution when the soft threshold power β was set at 3 (Supplementary www.impactjournals.com/oncotarget Identification of clinicopathological modules It is important to identify modules that have the most significant associations with clinical features. We sought to explore whether any of the groups of genes from each of the identified modules were correlated with the clinical variables of RC. The list of genes in each module is presented in Supplementary Table 3. The magenta (94 genes) module yielded significant Pearson’s correlation coefficient (PCC) with the pathological stage (R = 0.32, P = 0.003) and the pathology N stage (R = 0.32, P = 0.008). The salmon (41 genes) module Identification of consensual DEGs in training and test sets in rectal cancer patients Identification of consensual DEGs in training and test sets in rectal cancer patients The flow chart outlining the methods used in this study is shown in Supplementary Figure 1. In contrast to www.impactjournals.com/oncotarget Oncotarget 69595 69595 yielded significant PCC with the pathological stage (R = 0.25, P = 0.04) and pathology T stage (R = 0.33, P = 0.007). The red (126 genes) module yielded significant PCC with the pathological stage (R = 0.37, P = 0.002), the pathology T stage (R = 0.29, P = 0.02) and pathology N stage (R = 0.31, P = 0.01). The tan (44 genes) module yielded significant PCC with the pathological stage (R = 0.31, P = 0.01), the pathology T stage (R = 0.26, P = 0.03), pathology N stage (R = 0.38, P = 0.002), pathology M stage (R = 0.32, P = 0.009) and lymphatic invasion (R = 0.27, P = 0.03). The green-yellow (54 genes) module yielded significant PCC with the pathological stage (R = 0.38, P = 0.002), the pathology T stage (R = 0.34, P = 0.006), pathology N stage (R = 0.36, P = 0.006) and lymphatic invasion (R = 0.26, P = 0.04) (Figure 3C), which was used as the pathological stage module in subsequent analyses. These results suggested that the highly co-expressed genes in the same module have potential biological significance [6]. Each module might represent specific clinical features of RC patients [3, 22]. Figure 2A-2D). The dynamic tree cut method identified modules with similar expression profiles (Figure 3A, 3B). After highly similar modules were merged (Figure 3B), a total of 16 co-expressed modules were identified, ranging in size from 11 to 623 genes, whereas the “grey” module was reserved for genes that were not co-expressed (Figure 3C). After comparing the TCGA-READ test set with the validation set (GSE29621) [19], the summary preservation statistics used in determining whether a reference network is found in another test network were visualized [20]. The green-yellow and salmon modules were well preserved, with low median Rank statistics and Z-summary statistics larger than 10 (Supplementary Figure 3) [7, 21]. Identification of overall survival modules We also explored the significant associations of these modules to overall survival given their biological importance. We applied the Cox regression model in 66 RC patients with complete survival data to calculate the Figure 1: Meta-analysis of differentially expressed genes involved in rectal cancer by combining P-values. (A) Principal component analysis (PCA) biplot of quality control measures in four RC studies. (B) The number of differentially expressed genes plotted as a function of false discovery rate (FDR) in the analysis of four different datasets and four different meta-analysis algorithms (maxP, Fisher roP and adaptively weighted statistic). Figure 1: Meta-analysis of differentially expressed genes involved in rectal cancer by combining P-values. (A) Principal component analysis (PCA) biplot of quality control measures in four RC studies. (B) The number of differentially expressed genes plotted as a function of false discovery rate (FDR) in the analysis of four different datasets and four different meta-analysis algorithms (maxP, Fisher roP and adaptively weighted statistic). www.impactjournals.com/oncotarget Oncotarget 69596 outcomes of overall survival (HR = 13.534, P = 0.014). From the Kaplan-Meier curves, we also found that the decreased expression of genes within the salmon module indicated better outcomes of RC patients in the salmon module (P = 0.019) (Figure 4A). HRs and corresponding P-values for each dichotomized module (Table 2). The salmon module, which was defined as the overall survival module within the sixteen merged modules, was shown to have significant associations with the prognosis of RC patients. The upregulated expression of the genes within the salmon module indicated poor HRs and corresponding P-values for each dichotomized module (Table 2). The salmon module, which was defined as the overall survival module within the sixteen merged modules, was shown to have significant associations with the prognosis of RC patients. The upregulated expression of the genes within the salmon module indicated poor Oncotarget 69597 w.impactjournals.com/oncotarget he prognosis of RC patients. The upregulated expression f the genes within the salmon module indicated poor module (P 0.019) (Figure 4A). igure 2: Identification of consensus DEGs in the training and test sets of rectal cancer patients. (A) Heat map and two- ay hierarchical clustering based on 4091 DEGs that were differentially expressed between RC and ANT samples of the training set. ANT green label) and RC (red label) samples fell into separate clusters. (B) The 2177 DEGs RC (red label) vs. ANT (green label) of the TCGA- EAD test set. Identification of overall survival modules Each column represents a sample, whereas each row represents the mRNA expression level. The color scale represents the aw Z-score ranging from blue (low expression) to red (high expression). Dendrograms beside each heat map correspond to the hierarchical ustering of the 2177 DEGs by the expression level. (C) PCA plot showing complete, unsupervised separation of the 105 array samples nto 95 RC (red) and 10 ANT (green) samples. (D) A Venn diagram showing the overlap of DEGs detected by the training and test sets. Figure 2: Identification of consensus DEGs in the training and test sets of rectal cancer patients. (A) Heat map and two- way hierarchical clustering based on 4091 DEGs that were differentially expressed between RC and ANT samples of the training set. ANT (green label) and RC (red label) samples fell into separate clusters. (B) The 2177 DEGs RC (red label) vs. ANT (green label) of the TCGA- READ test set. Each column represents a sample, whereas each row represents the mRNA expression level. The color scale represents the raw Z-score ranging from blue (low expression) to red (high expression). Dendrograms beside each heat map correspond to the hierarchical clustering of the 2177 DEGs by the expression level. (C) PCA plot showing complete, unsupervised separation of the 105 array samples into 95 RC (red) and 10 ANT (green) samples. (D) A Venn diagram showing the overlap of DEGs detected by the training and test sets. www.impactjournals.com/oncotarget Oncotarget 69597 Figure 3: Network construction of the weighted co-expressed genes and their associations with clinical traits. (A) Hi hi l l t i t f th TCGA READ l b d th DEG D d ti l b l d ith th TCGA READ i Figure 3: Network construction of the weighted co-expressed genes and their associations with clinical traits. (A) Hierarchical clustering tree of the TCGA-READ samples based on the DEGs. Dendrogram tips are labeled with the TCGA-READ unique name and experiment identifier. In the hierarchical dendrogram, lower branches correspond to higher co-expression (height = Euclidean distance). Identical colors in the eight bands below the dendrogram depict the TCGA-READ clinical traits. (B) Heat map view of topological overlap of co-expressed genes in different modules. The heat map was generated from the topological overlap values between genes. The genes were grouped into modules labelled by a color code, which are given under the gene dendrogram on both sides. Identification of overall survival modules The topological overlap was high among genes of same module. (C) Module-trait relationships for age at diagnosis, gender, histological type, lymphatic invasion and pathologic stage. Numbers shown represent Pearson correlations between the modules and traits. P-values are in parentheses. Numbers on the color bar refer to the strength of the correlation in the table (red = 1, green = -1). (T: extent of the tumor; N: extent of spread to the lymph nodes; M: presence of metastasi). Figure 3: Network construction of the weighted co-expressed genes and their associations with clinical traits. (A Hierarchical clustering tree of the TCGA-READ samples based on the DEGs. Dendrogram tips are labeled with the TCGA-READ uniqu name and experiment identifier. In the hierarchical dendrogram, lower branches correspond to higher co-expression (height = Euclidea distance). Identical colors in the eight bands below the dendrogram depict the TCGA-READ clinical traits. (B) Heat map view of topologica overlap of co-expressed genes in different modules. The heat map was generated from the topological overlap values between genes. Th genes were grouped into modules labelled by a color code, which are given under the gene dendrogram on both sides. The topologica overlap was high among genes of same module. (C) Module-trait relationships for age at diagnosis, gender, histological type, lymphati invasion and pathologic stage. Numbers shown represent Pearson correlations between the modules and traits. P-values are in parentheses Numbers on the color bar refer to the strength of the correlation in the table (red = 1, green = -1). (T: extent of the tumor; N: extent of sprea to the lymph nodes; M: presence of metastasi). Figure 3: Network construction of the weighted co-expressed genes and their associations with clinical traits. (A) Hierarchical clustering tree of the TCGA-READ samples based on the DEGs. Dendrogram tips are labeled with the TCGA-READ unique name and experiment identifier. In the hierarchical dendrogram, lower branches correspond to higher co-expression (height = Euclidean distance). Identical colors in the eight bands below the dendrogram depict the TCGA-READ clinical traits. (B) Heat map view of topological overlap of co-expressed genes in different modules. The heat map was generated from the topological overlap values between genes. The genes were grouped into modules labelled by a color code, which are given under the gene dendrogram on both sides. The topological overlap was high among genes of same module. Identification of overall survival modules (C) Module-trait relationships for age at diagnosis, gender, histological type, lymphatic invasion and pathologic stage. Numbers shown represent Pearson correlations between the modules and traits. P-values are in parentheses. Numbers on the color bar refer to the strength of the correlation in the table (red = 1, green = -1). (T: extent of the tumor; N: extent of spread to the lymph nodes; M: presence of metastasi). Figure 3: Network construction of the weighted co-expressed genes and their associations with clinical traits. (A) Hierarchical clustering tree of the TCGA-READ samples based on the DEGs. Dendrogram tips are labeled with the TCGA-READ unique name and experiment identifier. In the hierarchical dendrogram, lower branches correspond to higher co-expression (height = Euclidean distance). Identical colors in the eight bands below the dendrogram depict the TCGA-READ clinical traits. (B) Heat map view of topological overlap of co-expressed genes in different modules. The heat map was generated from the topological overlap values between genes. The genes were grouped into modules labelled by a color code, which are given under the gene dendrogram on both sides. The topological overlap was high among genes of same module. (C) Module-trait relationships for age at diagnosis, gender, histological type, lymphatic invasion and pathologic stage. Numbers shown represent Pearson correlations between the modules and traits. P-values are in parentheses. Numbers on the color bar refer to the strength of the correlation in the table (red = 1, green = -1). (T: extent of the tumor; N: extent of spread to the lymph nodes; M: presence of metastasi). Enrichment analysis and sub-network genes of green-yellow and salmon modules We focused on the green-yellow module and carried out enrichment analysis because this module correlated strongly with the clinical features of RC patients, especially with the pathological stage. However, other modules showed weaker correlations with the phenotypic characteristics of RC. Interestingly, the green-yellow module was significantly enriched for cell adhesion for Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) (Figure 4B). We used the molecular complex detection (MCODE) algorithm (http:// apps.cytoscape.org/apps/mcode) to analyze a subset of the co-expression network (threshold = 0.4). When the node density cut-off was set at 2, the node score cut-off at 0.2, the k-core at 2, and the maximum depth at 100, the rank 1 cluster was identified. We found that there were 15 stage sub-network genes in this cluster (CHGB, SCG3, SYP, www.impactjournals.com/oncotarget www.impactjournals.com/oncotarget Oncotarget 69598 Table 2: Correlation between gene co-expression modules and overall survival Table 2: Correlation between gene co-expression modules and overall survival Overall survival HR CI (95% Cl) P-value ME black 0.258 0.054-1.220 0.087 ME blue 2.075 0.596-7.224 0.252 ME brown 0.448 0.115-1.747 0.248 ME cyan 1.245 0.375-4.146 0.721 ME green 0.736 0.206-2.628 0.637 ME green-yellow 2.307 0.605-8.791 0.221 ME grey 1.232 0.376-4.041 0.731 ME magenta 1.080 0.324-3.600 0.900 ME midnight-blue 1.868 0.545-6.396 0.320 ME pink 0.913 0.274-3.046 0.883 ME purple 0.234 0.049-1.106 0.067 ME red 2.802 0.742-10.579 0.129 ME salmon 13.534 1.712-106.958 0.014 ME tan 4.242 0.912-19.725 0.065 ME turquoise 2.487 0.658-9.400 0.180 ME yellow 1.440 0.433-4.790 0.552 SNAP25, SCN3B, C19orf30, LRRC55, FMO2, CDK5R2, AP3B2, FAM123C, CTXN2, RUNDC3A, IL1F5, and MATN3) (Figure 4C). SNAP25, SCN3B, C19orf30, LRRC55, FMO2, CDK5R2, AP3B2, FAM123C, CTXN2, RUNDC3A, IL1F5, and MATN3) (Figure 4C). The results of the survival analysis of the salmon module were consistent with their biological significance found by PCC. The salmon module was previously shown to correlate with the pathological stage (PCC = 0.25, P = 0.04) and the pathological T stage (PCC = 0.33, P = 0.007). Thus, the high expression of genes within the salmon module might represent a higher pathological stage and the T stage of RC with poor prognosis. The GO and KEGG pathway enrichment analysis of the genes within the salmon module showed that cancer (P = 0.046) and calcium-mediated signaling (P = 0.039) pathways were significantly affected when the condition of P < 0.05 was applied (Figure 4D). We utilized the MCODE algorithm to obtain seven sub-network genes which were defined as overall survival hub genes in the rank 1 cluster (GLT25D2, LRRTM1, ARMC4, CSMD3, SUSD4, HOMER2, and PTPRZ1) (Figure 4E). Further screening and identification of pathological stage candidate biomarkers by ANOVA, survival and ROC curve analysis ANOVA was carried out to determine the 15 pathological stage sub-network genes that were expressed separately and differentially. Meanwhile, the difference between localized RC (pathological stages I and II) and non-localized RC (pathological stages III and IV) by an independent t test for each of the 15 sub-network hub genes was determined. As a result, seven pathological stage hub genes (SCG3, SYP, SNAP25, CDK5R2, AP3B2, FAM123C, and RUNDC3A) that had a P-value less than 0.05 were designated to be significantly expressed (Supplementary Figure 4). www.impactjournals.com/oncotarget Oncotarget 69599 To further investigate whether the seven pathological stage hub genes correlated with the survival of RC patients, Kaplan-Meier analysis revealed that the five pathological stage hub genes (SCG3, SYP, CDK5R2, AP3B2, and RUNDC3A) with the highest levels significantly correlated with the markedly reduced overall survival of RC patients (Figure 5A), suggesting the important roles of five pathological stage hub genes in progress and prognosis of RC patients. Five pathological stage hub genes provided a high classification accuracy between localized RC and non-localized RC, which was estimated using receiver operating characteristic (ROC) curve analysis. AUC values for five pathological stage hub genes were greater than 0.64 in the TCGA-READ test set (Figure 5B). We could attain a best performance on accuracy by a univariate linear regression model built on Figure 4: Enrichment analysis and sub-network genes of the green-yellow and salmon modules. (A) Survival analysis based on the gene expression pattern in the salmon module. Impact of the expressed genes in the salmon module on the overall survival. (B) GO enrichment and KEGG analysis for the 54 module genes included in the green-yellow module. The original significance was transformed to “ – log10(P)” to plot the curve. (C) Visualization of the network connections among the most connected genes in the green- yellow module. Lines connecting two genes represent an association between the genes, and the color of the connecting line represents the weighted value of the two genes in the network. The colors of the nodes corresponded to the weighted value of the modular assignments. Elliptic and sexangular nodes indicate upregulated or downregulated sub-network genes in the modules. The nodes with a bold circle represent network hub genes identified by further analysis. (D) GO enrichment and KEGG analysis for the 41 module genes included in the salmon module. (E) Visualization of the network connections among the most connected genes in the salmon module. Further screening and identification of pathological stage candidate biomarkers by ANOVA, survival and ROC curve analysis Figure 4: Enrichment analysis and sub-network genes of the green-yellow and salmon modules. (A) Survival analysis based on the gene expression pattern in the salmon module. Impact of the expressed genes in the salmon module on the overall survival. (B) GO enrichment and KEGG analysis for the 54 module genes included in the green-yellow module. The original significance was transformed to “ – log10(P)” to plot the curve. (C) Visualization of the network connections among the most connected genes in the green- yellow module. Lines connecting two genes represent an association between the genes, and the color of the connecting line represents the weighted value of the two genes in the network. The colors of the nodes corresponded to the weighted value of the modular assignments. Elliptic and sexangular nodes indicate upregulated or downregulated sub-network genes in the modules. The nodes with a bold circle represent network hub genes identified by further analysis. (D) GO enrichment and KEGG analysis for the 41 module genes included in the salmon module. (E) Visualization of the network connections among the most connected genes in the salmon module. www.impactjournals.com/oncotarget www.impactjournals.com/oncotarget Oncotarget 69600 a panel of the combined 5 pathological stage hub genes (AUC = 0.744): risk score = 0.369 × SCG3 + 0.296 × SYP + 0.128 × RUNDC3A -0.073 × CDK5R2 + 0.115 × AP3B2 -3.55 (Figure 5B).i conducted to validate the five pathological stage hub genes in the validation set (GSE29621). Only SCG3 with a P-value less than 0.05 were declared to be significantly expressed in four pathological stages of RC (ANOVA test, P = 0.043) (Figure 5C). In addition, we carried out a single-gene survival analysis in order to reveal which genes were most significantly associated with overall Thus, five pathological stage hub genes were selected as the candidate genes for further validation. ANOVA, survival and ROC curve analysis were Oncotarge 69601 impactjournals com/oncotarget gure 5: Further screening and validation of pathologic stage candidate biomarkers by survival and ROC curv alyses. (A) Survival analysis of the five hub genes in the TCGA-READ test set. For the survival curves of the patients in differe ups, solid red lines represent the high expression of hub genes and solid black lines represent the low expression of hub genes. (B) RO lysis of the five hub genes in the TCGA- READ test set. Receiver operating characteristic (ROC) curve and area under the curve (AUC istics to evaluate the diagnostic efficiency of the hub genes in the TCGA-READ test set, which distinguished between localized an n-localized RC patients. (C) Boxplots of SCG3 mRNA expression across different pathologic stages in the validation set. (D) Surviv lysis of SCG3 mRNA expression in the validation set. (E) ROC analysis of SCG3 mRNA expression in the independent validation se Figure 5: Further screening and validation of pathologic stage candidate biomarkers by survival and ROC curve analyses. (A) Survival analysis of the five hub genes in the TCGA-READ test set. For the survival curves of the patients in different groups, solid red lines represent the high expression of hub genes and solid black lines represent the low expression of hub genes. (B) ROC analysis of the five hub genes in the TCGA- READ test set. Receiver operating characteristic (ROC) curve and area under the curve (AUC) statistics to evaluate the diagnostic efficiency of the hub genes in the TCGA-READ test set, which distinguished between localized and non-localized RC patients. (C) Boxplots of SCG3 mRNA expression across different pathologic stages in the validation set. www.impactjournals.com/oncotarget (D) Survival analysis of SCG3 mRNA expression in the validation set. (E) ROC analysis of SCG3 mRNA expression in the independent validation set. www.impactjournals.com/oncotarget Oncotarget 69601 survival hub genes in the salmon module was conducted for 93 RC patients, and we found that only HOMER2 significantly associated with patient overall survival in the TCGA-READ set (Supplementary Figure 5). Kaplan-Meier analysis revealed that the high HOMER2 level significantly correlated with the markedly reduced overall survival of RC patients (HR 7 [95% CI 1.03 − 52.82], P = 0.028) (Figure 6A), suggesting the important roles of HOMER2 in the prognosis of RC patients. The sensitivity and specificity of the HOMER2 expression level on the survival outcome was assessed by ROC curve analysis, and the area under the ROC curve was used to evaluate survival prediction efficiency of HOMER2. The AUC value of HOMER2 was 0.673, which was close to 0.7 (Figure 6B). Thus, HOMER2 was chosen as the overall survival candidate biomarker for further validation to determine whether there was a significant correlation between HOMER2 and the prognosis of RC patients. ANOVA, survival and ROC curve analysis were carried out to confirm HOMER2 in the validation survival. Unfortunately, we only found an increased expression of the SCG3 (HR 2.381 [95 % CI 1.01 – 5.88], P = 0.043) with poor prognosis (Figure 5D). SCG3, which was chosen as the pathological stage candidate biomarker, provided a high classification accuracy between localized RC and non-localized RC, which was estimated using ROC curve analysis (Figure 5E). Further screen and validation of novel overall survival candidate biomarkers To further narrow the overall survival candidate biomarkers that harbor great significance among the declared seven overall survival hub genes, we chose to use the survival and ROC curve analysis to summarize the expression patterns of the hub genes. Survival analysis was performed for each overall survival hub gene. The patients were dichotomized into two equal groups by the expression level of the overall survival hub genes. Survival analysis for the seven overall Figure 6: Further screening and validation of novel overall survival candidate biomarkers by survival and ROC curve analyses. (A) Survival analysis of HOMER2 mRNA expression in the TCGA-READ test set. For survival curves of patients in different groups, solid red lines represent high expression and solid black lines represent low expression. (B) ROC analysis of HOMER2 mRNA expression in the TCGA-READ test set. Receiver operating characteristic (ROC) curve and area under the curve (AUC) statistics to evaluate the diagnostic efficiency of HOMER2 mRNA expression in the TCGA-READ test set, which distinguished between high risk and low risk RC patients. (C) Boxplots of HOMER2 mRNA expression across different pathologic stages in the validation set. (D) Survival analysis of HOMER2 mRNA expression in the validation set. (E) ROC analysis of HOMER2 mRNA expression in the independent validation set to distinguish between high risk and low risk RC patients. Figure 6: Further screening and validation of novel overall survival candidate biomarkers by survival and ROC curve analyses. (A) Survival analysis of HOMER2 mRNA expression in the TCGA-READ test set. For survival curves of patients in different groups, solid red lines represent high expression and solid black lines represent low expression. (B) ROC analysis of HOMER2 mRNA expression in the TCGA-READ test set. Receiver operating characteristic (ROC) curve and area under the curve (AUC) statistics to evaluate the diagnostic efficiency of HOMER2 mRNA expression in the TCGA-READ test set, which distinguished between high risk and low risk RC patients. (C) Boxplots of HOMER2 mRNA expression across different pathologic stages in the validation set. (D) Survival analysis of HOMER2 mRNA expression in the validation set. (E) ROC analysis of HOMER2 mRNA expression in the independent validation set to distinguish between high risk and low risk RC patients. www.impactjournals.com/oncotarget Oncotarget 69602 set (GSE29621). Only HOMER2 had a P-value less than 0.05 and was determined to be significantly differentially expressed in the four pathological stages of RC (ANOVA test, P = 0.008) (Figure 6C). Further screen and validation of novel overall survival candidate biomarkers In addition, we carried out a single-gene survival analysis to reveal which genes were most significantly associated with overall survival. Unfortunately, we only found the increased expression of HOMER2 (HR 2.632 [95 % CI 1.163 – 6.840], P = 0.015) to associate with poor prognosis (Figure 6D).i the red, tan, and green-yellow modules and pathological stages T and N, suggesting that the highly co-expressed genes within the same module were of similar biological significance.i Given that the biological significance of these modules might be related to the potential clinical manifestations, we performed Cox regression analysis for each module to determine their survival outcomes [1, 11, 39, 40]. Poor outcomes were found for the high expression group of the salmon module, which was consistent with previous results indicating that the high expression of the salmon module might represent a higher pathological stage and the T stage [2]. The sensitivity and specificity of the expression level of HOMER2 on the survival outcome were assessed by ROC curve analysis, and AUC was used to evaluate survival prediction efficiency of HOMER2. The AUC value of HOMER2 was 0.685 (Figure 6E). The green-yellow and salmon modules were found to correlate with the pathological stage and overall survival, respectively. Thus, we selected the green- yellow module for the subsequent analysis because this module is likely to represent tumor staging characteristics more accurately [41]. Cell adhesion pathways suggested by GO and KEGG were over-represented in the green- yellow module. To some extent, these results also partially explained the increasing stages of RC patients. By means of one-way ANOVA and an independent sample t test, seven pathological stage hub genes (SCG3, SYP, SNAP25, CDK5R2, AP3B2, FAM123C, and RUNDC3A) could effectively distinguish non-localized RC from localized RC. Furthermore, the expression levels of five pathological stage hub genes (SCG3, SYP, CDK5R2, AP3B2, RUNDC3A) significantly associated with both the pathological stage and overall survival, but only SCG3 mRNA expression was replicated in the validation set. The diagnosis model of the combined five pathological stage hub genes, which provided a high classification accuracy, might be good biomarkers for distinguishing between DISCUSSION Gene signatures identified from genome-based assays are known to contribute to RC stratification [23]. Numerous studies have defined in part the gene signatures predicting the survival outcome or recurrence of RC [24–34]. In this study, six mRNA datasets were subjected to meta-analysis and WGCNA to identify hub genes associating with clinical characteristics as well as RC progression and prognosis [4, 5, 35].i We identified sixteen distinct modules from 2118 consensus genes that passed the meta-analysis filtering criteria and TCGA-READ validation for WGCNA [36]. Among the identified modules, the magenta, red, tan, and green-yellow modules associated with the pathological stage, especially the green-yellow module which had the strongest correlation with the pathological stage [12, 37, 38]. In addition, the brown module correlated with the age of diagnosis. Correlations were also found between Figure 7: SCG3 and HOMER2 are overexpressed in rectal cancer. (A) Forest plot of SCG3 expression across all training and test pooled analysis datasets. (B) Forest plot of HOMER2 expression across all training and test pooled analysis datasets. The x-axis is the standardized mean difference between rectal cancer (RC) and adjacent normal tissue (ANT) on a log2 scale. Thus, a value of 1 signifies a 2-fold difference in gene expression between cancer and normal. Figure 7: SCG3 and HOMER2 are overexpressed in rectal cancer. (A) Forest plot of SCG3 expression across all training and test pooled analysis datasets. (B) Forest plot of HOMER2 expression across all training and test pooled analysis datasets. The x-axis is the standardized mean difference between rectal cancer (RC) and adjacent normal tissue (ANT) on a log2 scale. Thus, a value of 1 signifies a 2-fold difference in gene expression between cancer and normal. www.impactjournals.com/oncotarget www.impactjournals.com/oncotarget Oncotarget 69603 WGCNA [47]. We associated a number of these network modules to RC clinicopathological variables, as well as to overall survival, and uncovered the gene expression signature associated with RC pathology stage and overall survival. We identified several significant pathways, as well as five potential pathological stage hub genes and seven overall survival hub genes using MCODE. Utilizing the TCGA-READ dataset, we constructed a diagnosis model of five mRNA signatures, and the validation cohort confirmed that the panel may be a useful biomarker for prognosis in RC. Of greatest interest, we found that the increased SCG3 and HOMER2 expression associated with the increased pathological stage and poor survival in the test and validation set among RC patients, providing two useful markers of RC and suggesting that it may help to identify those with more aggressive disease. Nevertheless, multicenter randomized controlled studies and in vivo and in vitro experiments are still required to evaluate the possible application of the molecular signatures for survival prediction and to characterize the hub genes functionally for the application of this approach in specific clinical settings. WGCNA [47]. We associated a number of these network modules to RC clinicopathological variables, as well as to overall survival, and uncovered the gene expression signature associated with RC pathology stage and overall survival. We identified several significant pathways, as well as five potential pathological stage hub genes and seven overall survival hub genes using MCODE. Utilizing the TCGA-READ dataset, we constructed a diagnosis model of five mRNA signatures, and the validation cohort confirmed that the panel may be a useful biomarker for prognosis in RC. Of greatest interest, we found that the increased SCG3 and HOMER2 expression associated with the increased pathological stage and poor survival in the test and validation set among RC patients, providing two useful markers of RC and suggesting that it may help to identify those with more aggressive disease. Nevertheless, multicenter randomized controlled studies and in vivo and in vitro experiments are still required to evaluate the possible application of the molecular signatures for survival prediction and to characterize the hub genes functionally for the application of this approach in specific clinical settings. localized and non-localized RC (Figure 5B). SCG3, a member of the multifunctional granin family, played a key role in secretory granule biogenesis, which involves the cellular uptake of endogenous and exogenous toxins [42]. Data collection, preprocessing, and normalization A public microarray repository was curated to search through PubMed, Gene Expression Omnibus (GEO) (accession numbers GSE75548, GSE34472, GSE35982, GSE12225, and GSE29621), ArrayExpress (accession number E-GEOD-34472), and TCGA-READ datasets up to September 2016. Only initial experimental studies that screened different mRNAs from RC and adjacent normal tissue samples in humans were included. After writing off the duplicated datasets, the combined datasets (GSE75548, GSE34472, GSE35982, GSE12225, and TCGA-READ) containing 168 RC and 60 normal rectal tissue samples were generated (Supplementary Table 1). The raw datasets were preprocessed individually using the LIMMA software package with log2 transformation and annotated by converting different probe IDs to the respective gene symbols. Duplicate gene expression values were averaged. In addition, we tentatively estimated between- studies heterogeneity in effect sizes of twelve hub genes (Supplementary Figure 6), especially SCG3 and HOMER2 (Figure 7A, 7B). Only one gene (SUSD4) had significant between-studies heterogeneity (I2 = 67.32%) and was ejected because we did not have sufficient information available to explore heterogeneity (Supplementary Table 1).i www.impactjournals.com/oncotarget SCG3 was identified as the most sensitive and specific marker for circulating tumor cells in small cell lung cancer and was indicative of a worse survival outcome. As is well known, response to standard chemotherapy is important in determining survival. SCG3 was also evident in patients with poor response to standard chemotherapy [24]. There is potential mechanistic association between SCG3 expression in tumors and response to platinum- based therapy or topoisomerase II inhibitors [24]. As RE-1 silencing transcription factor is a transcriptional repressor in cancer, higher expression of SCG3 mRNA may increase the aggressive potential of the tumor or reduce the drug sensitivity of RE-1 silencing transcription factor depleted tumors. For the first time, we show SCG3 as a biomarker of RC pathological stage and prognosis.i Further analysis identified the overall survival candidate biomarkers from the salmon module, thus demonstrating the significant association with survival in the test and validation sets. Among the seven hub genes related to overall survival, increased HOMER2 expression associated with the increased pathological stage and poor survival of RC patients. Moreover, such expression may be useful in evaluating the survival risk because the AUC, which was used to evaluate the survival prediction efficiency of HOMER2, was 0.673 and 0.685 in the test and validation set, respectively. HOMER2 gene encodes a member of the homer family of dendritic proteins and regulates group 1 metabotrophic glutamate receptor function. HOMER2 is also a promising biomarker for cancer prognosis. HOMER2 is known to associate with overall survival and disease-free survival in early stage non-small cell lung cancer [43]. HOMER2 which was identified as a binding partner of MYO18B, interacted with the C-terminal region of MYO18B, a candidate tumor suppressor gene involved in the pathogenesis of human cancers including colorectal cancer [44]. Additionally, methylation of HOMER2 was reported to be a valuable biomarker which significantly discriminated CRC patients from controls [45]. Integrated-signature gene analysis of TCGA- READ as the test set and consensus screening of the DEGs To further limit analyses to genes common to all datasets, the results of the DEG training set were validated in the TCGA-READ dataset, which was considered as the test set. The TCGA-READ mRNA and clinical data (level 3) of the corresponding patients (RC and ANT) were downloaded from the TCGA data portal (up to May 20, 2016). The TCGA-READ DEGs were analyzed using an empirical Bayes approach within the LIMMA software package. The DEGs of the test set with a \|log2 fold change (FC)\| ≥ 0.5 and an adjusted P-value less than 0.05 were selected for subsequent analysis. A total of 2177 gene symbols of the test set passed the filtering criteria. We then created an overlapping gene set by selecting common official gene symbols in both training and test sets, resulting in a total of 2118 intersecting genes. The consensus DEGs, which ensured that RC and ANT samples were well characterized, were chosen for WGCNA [10, 37, 52]. Preprocessing of clinical information Clinical information obtained from the TCGA- READ dataset (American Joint Committee on Cancer pathological TNM stage, gender, age at initial pathological diagnosis and histological type (mucinous adenocarcinoma or adenocarcinoma), especially vital status and time to last follow-up) for 70 patients was used after eliminating incomplete clinical traits and gene expression. Integrated analysis of the gene expression datasets of the training set To make the five microarray datasets derived from the five different platforms compatible for data analysis, we used the MetaOmics software package (http://www. pitt.edu/~tsengweb/MetaOmicsHome.htm) to integrate and analyze the GEO datasets [48]. Firstly, the MetaQC software package, which provided a quantitative and objective quality control tool for determining the inclusion/ exclusion criteria for the meta-analysis, eliminated To our knowledge, this is the first study investigating the relationship between the co-expression network of clinical traits and genes in patients with RC [46]. In summary, we identified sixteen gene co-expression modules from five RC datasets using meta-analysis and www.impactjournals.com/oncotarget Oncotarget 69604 the GSE34472 dataset [48, 49]. Thus, the training set included three GEO datasets (GSE75548, GSE35982, and GSE12225) after Quality Control. Secondly, to distinguish the DEGs between RC and ANT, the MetaDE software package limited the mean and standard deviation (SD) filter thresholds, which were set at 10% to filter minor changes in gene expression levels [50]. Considering the different stringencies of the various methods, Fisher’s method was favored for the meta-analysis (Figure 1B). For Fisher’s method, the modified t test and the permutation method (nPermutations = 300) were used to extrapolate the P-values [51]. P-values less than 0.05 were considered statistically significant for the DEGs. The heatmaps illustrating the DEG patterns were also generated [50]. estimated and generated as described for the standard scale-free networks. In this case, the power of β, which was set at 3 (scale-free R2 = 0.89), was auto-selected. Moreover, the weighted adjacency matrix was constructed using the power function ADJmn = \|CORmn\|β (CORmn = Pearson’s correlation between gene m and gene n; ADJmn = adjacency between gene m and gene n). β was the soft thresholding parameter, which was used to transform adjacencies and correlations into a Topological Overlap Matrix (TOM), and then the corresponding dissimilarity (1-TOM) was calculated. Finally, module identification was carried out with the dynamic tree cut method by hierarchically clustering the genes using 1-TOM as the distance measure with a deep split value of 2 and a minimum size cut-off of 30 for the resulting dendrogram. Highly similar modules were marked by clustering and merged with a height cut-off of 0.25. Module preservation and quality statistics were computed using the module Preservation function (nPermutations = 200) within the WGCNA software package between the TCGA-READ test set and the GSE29621 validation set (Supplementary Table 1) [20]. Identification of clinical feature modules, survival analysis and efficacy evaluation of pathological stage hub genes Module eigengenes (MEs), which are the first principal components in the PCA for each gene module, summarized the expression patterns of all genes into a single characteristic expression profile within a given module. The dynamic decision-making tree, node splitting method and cluster analysis of the squared Euclidean distance were used to identify MEs related to these clinical features, especially those involved in the progression and prognosis of RC. Spearman’s correlation analysis was carried out to confirm the object module, which was the most relevant module between the MEs and clinical traits. Depending on these, the module that had the highest Spearman’s correlation coefficient for the pathological stage and MEs in the object module was defined as the pathological stage module. Hub genes that had been selected in the pathological stage module were obtained using the MCODE algorithm plugin within the Cytoscape software package (version 3.4.0) [53]. Weighted gene co-expression network construction and module preservation analysis For the single module survival analysis, the TCGA-READ test data was dichotomized around the median expression of each gene module. The overall survival was used as the survival endpoint, which was determined via the “survival” R software package. Cox regression analysis was performed to evaluate the hazard ratio (HR). Survival curves of the significant objected The TCGA-READ dataset with intact clinical features and prognostic information was selected for constructing the scale-free gene co-expression networks within the WGCNA software package [8]. Firstly, the appropriate soft threshold power was automatically www.impactjournals.com/oncotarget Abbreviations The abbreviations used are: RC, rectal cancer; ANT, adjacent normal tissue; WGCNA, weighted gene co-expression network analysis; TCGA, The Cancer Genome Atlas; READ, rectal adenocarcinoma; DEGs, differentially expressed genes; HR, hazard ratio; AUC, area under the curve; GEO, Gene Expression Omnibus; GSE, gene expression data series; LIMMA, linear models for microarray analysis; QC, quality control; SD, standard deviation; TOM, Topological Overlap Matrix; ME, Module eigengene; MCODE, molecular complex detection; GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes; ANOVA, analysis of variance; AJCC, American Joint Committee on Cancer; PCC, Pearson’s correlation coefficient; PCA, principal component analysis; ROC, receiver operating characteristic; DAVID, the Database for Annotation, Visualization, and Integrated Discovery. www.impactjournals.com/oncotarget Oncotarget 69605 heterogeneity was examined by the Cochran’s Q-test and I2 statistic, with P values for heterogeneity by the I2 value > 50% indicating substantial heterogeneity [56]. We used 300 Monte Carlo permutation tests after combining P values (Supplementary Figure 1) [57], and a DerSimonian and Laird random effects model for meta-analysis by combing effect size to select hub genes in RC. module, which was defined as the survival module, were constructed by the Kaplan-Meier method and compared by the log-rank test. Survival hub genes were obtained by MCODE. Survival curves and ROC analysis were designed to identify candidate markers. The results of the survival curves and ROC analysis were next verified on an independent validation set (GSE29621) using the survival hub genes as the candidate marker input to predict the classes of prognosis in RC. GRANT SUPPORT This research was supported by National Natural Science Foundation of China (81572874 and 81472799), and Project of Hubei Medical Talents Training Program. Author contributions Significance differences between stage groups were determined by analysis of variance (ANOVA) or the Student’s t test. P-values less than 0.05 were considered statistically significant. The preliminary relationships between the hub genes of stage modules were demonstrated by boxplot graphs. Survival curves were constructed by the Kaplan-Meier method and compared by the log-rank test for the expression level with the stage and survival hub genes for calculating the overall survival. All Cox regression models were tested based on Schoenfeld residuals to evaluate the hazard ratio (HR) for each module. Statistical analysis was performed using the R software package (version 3.2.3). ROC analysis was used to evaluate the diagnostic value on the outcome for the expression level of each stage and survival hub genes in RC via the code of Mihaly (Supplementary File 2) [55]. Using univariate linear logistic regression of the hub genes, a classification model of the combined hub genes was built to evaluate the discriminatory capacity of localized RC and non-localized RC. Finally, the results of the survival curves and ROC analysis were validated on an independent validation set (GSE29621) using the pathological stage and survival hub genes as candidate biomarker input to predict the classes of pathological stage as well as survival of neoplastic progression and prognosis in RC patients. Because of the heterogeneity of hub genes within and among samples and datasets, the between-study Participated in research design: Min Sun, Bin Xiong. Performed data analysis: Min Sun, Zhongshi He and Taojiao Sun. Wrote or contributed to the writing of the manuscript: Min Sun and Taojiao Sun. Functional annotation and network visualization within pathological stage and survival modules The pathological stage and survival modules were functionally annotated based on the analytical results of their gene compositions. GO functional-related to specific Biological Processes and KEGG pathway enrichment analyses were performed for the object module by using the Database for Annotation, Visualization, and Integrated Discovery (DAVID) [54]. P-values less than 0.05 were used as the cut-off value. Network visualization of the stage and survival modules were carried out within the Cytoscape software package. Moreover, there were strong correlations between the co-expressed module genes and their functions, and MCODE was utilized to provide deeper insights for hub genes. ACKNOWLEDGMENTS We thank the Cancer Genome Atlas and the Gene Expression Omnibus for providing the data for this study. CONFLICTS OF INTEREST The authors declare no conflicts of interest. REFERENCES 1. Heiland DH, Haaker G, Delev D, Mercas B, Masalha W, Heynckes S, Gabelein A, Pfeifer D, Carro MS, Weyerbrock A, Prinz M, Schnell O. Comprehensive analysis of PD-L1 www.impactjournals.com/oncotarget Oncotarget 69606 69606 expression in glioblastoma multiforme. Oncotarget. 2017; 8:42214-42225. doi: 10.18632/oncotarget.15031. 14. Modena BD, Bleecker ER, Busse WW, Erzurum SC, Gaston BM, Jarjour NN, Meyers DA, Milosevic J, Tedrow JR, Wu W, Kaminski N, Wenzel SE. Gene expression correlated to severe asthma characteristics reveals heterogeneous mechanisms of severe disease. Am J Respir Crit Care Med. 2017; 195:1449-1463. 2. Feng Y, Wang X. Systematic analysis of microarray datasets to identify Parkinson's diseaseassociated pathways and genes. Mol Med Rep. 2017; 15:1252-1262. 3. Broadbent JA, Sampson D, Sabapathy S, Haseler LJ, Wagner KH, Bulmer AC, Peake JM, Neubauer O. Gene networks in skeletal muscle following endurance exercise are co-expressed in blood neutrophils and linked with blood inflammation markers. J Appl Physiol (1985). 2017; 122:752-766. 15. Miao X, Luo Q, Zhao H, Qin X. Co-expression analysis and identification of fecundity-related long non-coding RNAs in sheep ovaries. Sci Rep. 2016; 6:39398. 16. Wang Q, Huang H, Zeng X, Ma Y, Zhao X, Huang M. Single-agent maintenance therapy for advanced non- small cell lung cancer (NSCLC): a systematic review and Bayesian network meta-analysis of 26 randomized controlled trials. PeerJ. 2016; 4:e2550. 4. Gawlick U, Lu KC, Douthit MA, Diggs BS, Schuff KG, Herzig DO, Tsikitis VL. Stage III & IV colon and rectal cancers share a similar genetic profile: a review of the Oregon Colorectal Cancer Registry. Am J Surg. 2013; 205:608-612; discussion 612. 17. Luykx JJ, Olde Loohuis LM, Neeleman M, Strengman E, Bakker SC, Lentjes E, Borgdorff P, van Dongen EP, Bruins P, Kahn RS, Horvath S, de Jong S, Ophoff RA. Peripheral blood gene expression profiles linked to monoamine metabolite levels in cerebrospinal fluid. Transl Psychiatry. 2016; 6:e983. 5. Gantt GA, Chen Y, Dejulius K, Mace AG, Barnholtz-Sloan J, Kalady MF. Gene expression profile is associated with chemoradiation resistance in rectal cancer. Colorectal Dis. 2014; 16:57-66. 18. Lu S, Li J, Song C, Shen K, Tseng GC. Biomarker detection in the integration of multiple multi-class genomic studies. Bioinformatics. 2010; 26:333-340. 6. Pei G, Chen L, Zhang W. WGCNA application to proteomic and metabolomic data analysis. Methods Enzymol. 2017; 585:135-158. 19. Bogenpohl JW, Mignogna KM, Smith ML, Miles MF. REFERENCES Integrative analysis of genetic, genomic, and phenotypic data for ethanol behaviors: a network-based pipeline for identifying mechanisms and potential drug targets. Methods Mol Biol. 2017; 1488:531-549. 7. Langfelder P, Horvath S. WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics. 2008; 9:559. 8. Zhang J, Baddoo M, Han C, Strong MJ, Cvitanovic J, Moroz K, Dash S, Flemington EK, Wu T. Gene network analysis reveals a novel 22-gene signature of carbon metabolism in hepatocellular carcinoma. Oncotarget. 2016; 7:49232-49245. doi: 10.18632/oncotarget.10249. 20. Langfelder P, Luo R, Oldham MC, Horvath S. Is my network module preserved and reproducible? PLoS Comput Biol. 2011; 7:e1001057. 21. Langfelder P, Horvath S. Eigengene networks for studying the relationships between co-expression modules. BMC Syst Biol. 2007; 1:54. 9. Esposti DD, Hernandez-Vargas H, Voegele C, Fernandez- Jimenez N, Forey N, Bancel B, Le Calvez-Kelm F, McKay J, Merle P, Herceg Z. Identification of novel long non- coding RNAs deregulated in hepatocellular carcinoma using RNA-sequencing. Oncotarget. 2016; 7:31862-31877. doi: 10.18632/oncotarget.7364. 22. Wilbe M, Kozyrev SV, Farias FH, Bremer HD, Hedlund A, Pielberg GR, Seppala EH, Gustafson U, Lohi H, Carlborg O, Andersson G, Hansson-Hamlin H, Lindblad-Toh K. Multiple changes of gene expression and function reveal genomic and phenotypic complexity in SLE-like disease. PLoS Genet. 2015; 11:e1005248. 10. Heiland DH, Demerath T, Kellner E, Kiselev VG, Pfeifer D, Schnell O, Staszewski O, Urbach H, Weyerbrock A, Mader I. Molecular differences between cerebral blood volume and vessel size in glioblastoma multiforme. Oncotarget. 2017; 8:11083-11093. doi: 10.18632/oncotarget.11522. 23. Dearmon B. Implementing measures to improve cancer program practice profile report (CP3R) compliance rates for breast, colon, and rectal cancers. J Registry Manag. 2010; 37:65-66. 11. Yu X, Feng L, Liu D, Zhang L, Wu B, Jiang W, Han Z, Cheng S. Quantitative proteomics reveals the novel co-expression signatures in early brain development for prognosis of glioblastoma multiforme. Oncotarget. 2016; 7:14161-14171. doi: 10.18632/oncotarget.7416. 24. de Maat MF, van de Velde CJ, Benard A, Putter H, Morreau H, van Krieken JH, Meershoek Klein-Kranenbarg E, de Graaf EJ, Tollenaar RA, Hoon DS. Identification of a quantitative MINT locus methylation profile predicting local regional recurrence of rectal cancer. Clin Cancer Res. 2010; 16:2811-2818. 12. Pacholewska A, Marti E, Leeb T, Jagannathan V, Gerber V. LPS-induced modules of co-expressed genes in equine peripheral blood mononuclear cells. BMC Genomics. 2017; 18:34. 25. Herrera L, Brown MT. Prognostic profile in rectal cancer. Dis Colon Rectum. 1994; 37:S1-5. 13. REFERENCES Chen J, Yu L, Zhang S, Chen X. Network analysis-based approach for exploring the potential diagnostic biomarkers of acute myocardial infarction. Front Physiol. 2016; 7:615. 26. Morandin C, Tin MM, Abril S, Gomez C, Pontieri L, Schiott M, Sundstrom L, Tsuji K, Pedersen JS, Helantera H, Mikheyev AS. Comparative transcriptomics reveals the www.impactjournals.com/oncotarget Oncotarget 69607 40. Presson AP, Yoon NK, Bagryanova L, Mah V, Alavi M, Maresh EL, Rajasekaran AK, Goodglick L, Chia D, Horvath S. Protein expression based multimarker analysis of breast cancer samples. BMC Cancer. 2011; 11:230. conserved building blocks involved in parallel evolution of diverse phenotypic traits in ants. Genome Biol. 2016; 17:43. 27. Liu J, Jing L, Tu X. Weighted gene co-expression network analysis identifies specific modules and hub genes related to coronary artery disease. BMC Cardiovasc Disord. 2016; 16:54. 41. Xu X, Zhou Y, Miao R, Chen W, Qu K, Pang Q, Liu C. Transcriptional modules related to hepatocellular carcinoma survival: coexpression network analysis. Front Med. 2016; 10:183-190. 28. Zhou R, Man Y. Integrated analysis of DNA methylation profiles and gene expression profiles to identify genes associated with pilocytic astrocytomas. Mol Med Rep. 2016; 13:3491-3497. 42. Cecchin E, Agostini M, Pucciarelli S, De Paoli A, Canzonieri V, Sigon R, De Mattia E, Friso ML, Biason P, Visentin M, Nitti D, Toffoli G. Tumor response is predicted by patient genetic profile in rectal cancer patients treated with neo-adjuvant chemo-radiotherapy. Pharmacogenomics J. 2011; 11:214-226. 29. Wang QL, Chen X, Zhang MH, Shen QH, Qin ZM. Identification of hub genes and pathways associated with retinoblastoma based on co-expression network analysis. Genet Mol Res. 2015; 14:16151-16161. 43. Naeger NL, Robinson GE. Transcriptomic analysis of instinctive and learned reward-related behaviors in honey bees. J Exp Biol. 2016; 219:3554-3561. 30. Riordan AM, Thomas MK, Ronnekleiv-Kelly S, Warner T, Geiger PG, Kennedy GD. Utility of micro-ribonucleic acid profile for predicting recurrence of rectal cancer. J Surg Res. 2012; 177:87-92. 44. Ajima R, Kajiya K, Inoue T, Tani M, Shiraishi-Yamaguchi Y, Maeda M, Segawa T, Furuichi T, Sutoh K, Yokota J. HOMER2 binds MYO18B and enhances its activity to suppress anchorage independent growth. Biochem Biophys Res Commun. 2007; 356:851-856. 31. Guo Y, Xing Y. Weighted gene co-expression network analysis of pneumocytes under exposure to a carcinogenic dose of chloroprene. Life Sci. 2016; 151:339-347. 32. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2016. CA: a cancer journal for clinicians. 2016; 66:7-30. 45. REFERENCES Brohawn DG, O'Brien LC, Bennett JP Jr. RNAseq analyses identify tumor necrosis factor-mediated inflammation as a major abnormality in ALS spinal cord. PLoS One. 2016; 11:e0160520. 33. Cantor RM, Cordell HJ. Gene expression in large pedigrees: analytic approaches. BMC Genet. 2016; 17 Suppl 2:3. 34. Deo S, Kumar S, Shukla NK, Kar M, Mohanti BK, Sharma A, Raina V, Rath GK. Patient profile and treatment outcome of rectal cancer patients treated with multimodality therapy at a regional cancer center. Indian J Cancer. 2004; 41:120-124. 46. Gao B, Shao Q, Choudhry H, Marcus V, Dong K, Ragoussis J, Gao ZH. Weighted gene co-expression network analysis of colorectal cancer liver metastasis genome sequencing data and screening of anti-metastasis drugs. Int J Oncol. 2016; 49:1108-1118. 35. Gonzalez-Gonzalez M, Garcia J, Alcazar JA, Gutierrez ML, Gonzalez LM, Bengoechea O, Abad MM, Santos- Briz A, Blanco O, Martin M, Rodriguez A, Fuentes M, Munoz-Bellvis L, et al. Association between the cytogenetic profile of tumor cells and response to preoperative radiochemotherapy in locally advanced rectal cancer. Medicine (Baltimore). 2014; 93:e153. 47. Holtman IR, Raj DD, Miller JA, Schaafsma W, Yin Z, Brouwer N, Wes PD, Moller T, Orre M, Kamphuis W, Hol EM, Boddeke EW, Eggen BJ. Induction of a common microglia gene expression signature by aging and neurodegenerative conditions: a co-expression meta- analysis. Acta Neuropathol Commun. 2015; 3:31. 48. Wang X, Kang DD, Shen K, Song C, Lu S, Chang LC, Liao SG, Huo Z, Tang S, Ding Y, Kaminski N, Sibille E, Lin Y, et al. An R package suite for microarray meta-analysis in quality control, differentially expressed gene analysis and pathway enrichment detection. Bioinformatics. 2012; 28:2534-2536. 36. Langfelder P, Mischel PS, Horvath S. When is hub gene selection better than standard meta-analysis? PLoS One. 2013; 8:e61505. 37. Tian F, Zhao J, Fan X, Kang Z. Weighted gene co-expression network analysis in identification of metastasis-related genes of lung squamous cell carcinoma based on the Cancer Genome Atlas database. J Thorac Dis. 2017; 9:42-53. 49. Kang DD, Sibille E, Kaminski N, Tseng GC. MetaQC: objective quality control and inclusion/exclusion criteria for genomic meta-analysis. Nucleic Acids Res. 2012; 40:e15. 38. Mulligan MK, Mozhui K, Pandey AK, Smith ML, Gong S, Ingels J, Miles MF, Lopez MF, Lu L, Williams RW. Genetic divergence in the transcriptional engram of chronic alcohol abuse: a laser-capture RNA-seq study of the mouse mesocorticolimbic system. Alcohol. 2017; 58:61-72. 50. www.impactjournals.com/oncotarget REFERENCES Qi C, Hong L, Cheng Z, Yin Q. Identification of metastasis- associated genes in colorectal cancer using metaDE and survival analysis. Oncol Lett. 2016; 11:568-574. 51. Tseng GC, Ghosh D, Feingold E. Comprehensive literature review and statistical considerations for microarray meta- analysis. Nucleic Acids Res. 2012; 40:3785-3799. 39. Huang Y, Ma SF, Vij R, Oldham JM, Herazo-Maya J, Broderick SM, Strek ME, White SR, Hogarth DK, Sandbo NK, Lussier YA, Gibson KF, Kaminski N, et al. A functional genomic model for predicting prognosis in idiopathic pulmonary fibrosis. BMC Pulm Med. 2015; 15:147. 52. Oros Klein K, Oualkacha K, Lafond MH, Bhatnagar S, Tonin PN, Greenwood CM. Gene coexpression analyses www.impactjournals.com/oncotarget Oncotarget 69608 differentiate networks associated with diverse cancers harboring TP53 Missense or null mutations. Front Genet. 2016; 7:137. expression-based biomarkers predicting outcome after tamoxifen treatment in breast cancer. Breast Cancer Res Treat. 2013; 140:219-232. 56. Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med. 2002; 21:1539-1558. 53. Bader GD, Hogue CW. An automated method for finding molecular complexes in large protein interaction networks. BMC Bioinformatics. 2003; 4:2. 57. Zintzaras E, Ioannidis JP. Meta-analysis for ranked discovery datasets: theoretical framework and empirical demonstration for microarrays. Comput Biol Chem. 2008; 32:39-47. 54. Huang da W, Sherman BT, Lempicki RA. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protoc. 2009; 4:44-57. 55. Mihaly Z, Kormos M, Lanczky A, Dank M, Budczies J, Szasz MA, Gyorffy B. A meta-analysis of gene www.impactjournals.com/oncotarget Oncotarget 69609
https://openalex.org/W3173499163	https://ojs.unpkediri.ac.id/index.php/pgsd/article/download/15742/2168	Indonesian	null	Analisis Manfaat Penggunaan Aplikasi Zoom Dalam Pembelajaran Daring Bagi Guru Dan Siswa Di Sekolah Dasar	Jurnal pendidikan dasar nusantara	2,021	cc-by-sa	4,417	at Penggunaan Aplikasi... Volume 7│Nomor 1│Juli 2021│ DOI:https://doi.org/10.29407/jpdn.v7i1.15742 at Penggunaan Aplikasi... Volume 7│Nomor 1│Juli 2021│ DOI:https://doi.org/10.29407/jpdn.v7i1.15742 at Penggunaan Aplikasi Volume 7│Nomor 1│Juli 2021│ DOI:https://doi.org/10.29407/jpdn.v7i1.15742 ANALISIS MANFAAT PENGGUNAAN APLIKASI ZOOM DALAM PEMBELAJARAN DARING BAGI GURU DAN SISWA DI SEKOLAH DASAR Eko Kuntarto1, Muhammad Sofwan2, Nurlaili Mulyani3. ekokuntarto28@unja.ac.id1, muhammad.sofwan@unja.ac.id2, nurlailimulyani@gmail.com3 PGSD, FKIP, Universitas Jammbi1,2,3 Abstrak: Penelitian ini bertujuan untuk mendeskripsikan manfaat dari penggunaan aplikasi zoom dalam pembelajaran daring bagi guru dan siswa di Sekolah Dasar.Jenis penelitian dalam penelitian ini yaitu penelitian kualitatif dengan jenis fenomenologi. Sampel dalam penelitian ini sebanyak 8 informan, terdiri dari dua guru kelas V dan enam siswa kelas Vc. Hasil dari penelitian ini adalah penggunaan aplikasi zoom memberikan banyak manfaat bagi guru dan siswa selama pembelajaran daring di masa pandemi. Manfaat bagi guru diantaranya memudahkan mencapai tujuan pembelajaran, memudahkan berinteraksi bersama siswa, memberikan variasi dalam pembelajaran daring, menciptakan pembelajaran daring yang aktif , sedangkan manfaat penggunaan zoom bagi siswa diantaranya memudahkan memahami materi, meningkatkan semangat belajar daring, menghilangkan rasa bosan belajar daring. Manfaat yang dirasakan guru dan siswa tersebut memberikan dampak yang positif dalam pembelajaran daring sehingga menjadi lebih efektif. Kata kunci: manfaat zoom, pembelajaran daring, guru dan siswa, sekolah dasar. Keywords: benefits of zoom, online learning, teachers and students, elementary school. Keywords: benefits of zoom, online learning, teachers and students, elementary school. ANALYSIS OF THE BENEFITS OF USING THE ZOOM APP IN ONLINE LEARNING FOR TEACHERS AND STUDENTS IN ELEMENTARY SCHOOLS Abstract: This study aims to describe the benefits of using the zoom app in online learning for teachers and students in elementary schools. The type of research in this study is qualitative research with phenomenological type. The sample in this study was 8 informants, consisting of two grade v teachers and six vc grade students. Benefits for teachers include making it easier to achieve learning objectives, facilitating interaction with students, providing variations in online learning, creating active online learning, http://ojs.unpkediri.ac.id/pgsd │Volume 7│Nomor 1│Juli 2021 49 │Volume 7│Nomor 1│Juli 2021 Eko, Muhammad, Nurlaili. Analisis Manfaat Penggunaan Aplikasi... while the benefits of using zoom for students include making it easier to understand materials, improving the spirit of online learning, eliminating the boredom of online learning. The benefits felt by teachers and students have a positive impact in online learning so that it becomes more effective. Keywords: benefits of zoom, online learning, teachers and students, elementary school. Eko, Muhammad, Nurlaili. Analisis Manfaat Penggunaan Aplikasi... while the benefits of using zoom for students include making it easier to understand materials, improving the spirit of online learning, eliminating the boredom of online learning. The benefits felt by teachers and students have a positive impact in online learning so that it becomes more effective. uhammad, Nurlaili. Analisis Manfaat Penggunaan Aplikasi... Eko, Muhammad, Nurlaili. Analisis Manfaat Penggunaan Aplikasi... while the benefits of using zoom for students include making it easier to understand materials, improving the spirit of online learning, eliminating the boredom of online learning. The benefits felt by teachers and students have a positive impact in online learning so that it becomes more effective. while the benefits of using zoom for students include making it easier to understan materials, improving the spirit of online learning, eliminating the boredom of onlin learning. The benefits felt by teachers and students have a positive impact in onlin learning so that it becomes more effective. PENDAHULUAN Dengan adanya covid-19 yang melanda Indonesia di awal Maret 2020, hingga saat ini masih memberikan dampak di setiap aspek kehidupan termasuk di bidang pendidikan. Covid-19 merupakan penyakit menular, yang berarti dapat menyebar, baik secara langsung maupun tidak langsung, dari satu orang ke orang lain (Mustakim, 2020). Penyakit ini menyerang sistem pernapasan pada manusia, yang mana penyebaran Covid-19 masih berlangsung dengan cepat. Salah satu cara memutuskan rantai penyebaran Covid-19 yaitu dengan melakukan pembatasan interaksi masyarakat. Kemendikbud untuk memutuskan rantai penyebaran covid-19 melalui Surat Edaran No 4 Tahun 2020 yang dikeluarkan Kementrian Pendidikan dan Kebudayaan Nasional tentang Pelaksanaan Kebijakan Pendidikan Dalam Masa Darurat Penyebaran Covid-19, dimana pembelajaran yang ada wajib dilaksanakan secara daring. Meskipun pendidikan dilaksanakan secara daring tetapi pendidikan memegang peran penting dalam perbaikan kualitas generasi sebuah negara, tujuan dari pendidikan harus tetap diwujudkan. Pendidikan memiliki tujuan untuk mengembangkan potensi siswa supaya melahirkan manusia yang bertakwa kepada Tuhan Yang Maha Esa, menjadikan manusia dan warga negara yang seutuhnya (UU No 20 Tahun 2003). Salah satu dampak dari wabah pandemi Covid-19 pada bidang pendidikan yaitu pembelajaran yang awalnya dilaksanakan secara luring kini telah dialihkan secara daring. Pembelajaran daring merupakan sistem pembelajaran yang dilakukan dengan menggunakan platform yang dapat menunjang proses pembelajaran meskipun jarak jauh (Handarini & Wulandari, 2020). Pembelajaran daring adalah kegiatan belajar yang dilaksanakan dalam jarak jauh melalui media berupa internet dan alat seperti komputer dan smartphone (Putria dkk, 2020). Pembelajaran daring lebih menekankan pada ketelitian dan kejelian peserta didik dalam menerima dan mengolah informasi yang disajikan secara online (Riyana, 2019). Maka dapat disimpulkan bahwa pembelajaran daring adalah pembelajaran yang dilaksanakan dalam kondisi berjauhan dengan http://ojs.unpkediri.ac.id/pgsd │Volume 7│Nomor 1│Juli 2021 50 http://ojs.unpkediri.ac.id/pgsd │Volume 7│Nomor 1│Juli 2021 50 50 50 http://ojs.unpkediri.ac.id/pgsd │Volume 7│Nomor 1│Juli 2021 http://ojs.unpkediri.ac.id/pgsd http://ojs.unpkediri.ac.id/pgsd Eko, Muhammad, Nurlaili. Analisis Manfaat Penggunaan Aplikasi... memanfaatkan media elektronik dan jaringan internet, sehingga memudahkan siswa untuk belajar. Pembelajaran masa pandemi yang menggunakan pembelajaran daring dipilih untuk merencanakan proses pembelajaran yang sesuai dengan kondisi saat ini dimana pembelajaran tatap muka tidak dapat dilaksanakan. Pelaksanaan pembelajaran daring mampu untuk menunjang dan memudahkan proses pembelajaran di sekolah (Sadikin & Hamidah, 2020). Proses pembelajaran daring bertujuan agar kegiatan belajar mengajar menjadi lebih mudah dan lancar serta dapat meningkatkan minat dan hasil proses belajar mengajar. Pembelajaran daring bertujuan memberikan fasilitas pembelajaran yang bermutu dalam jaringan (daring) yang bersifat utuh dan terbuka untuk menjangkau peminat yang lebih banyak dan luas (Sofyana & Roza, 2019) Untuk menunjang pembelajaran daring yang dilakukan demi mencapai tujuan pembelajaran, guru harus memanfaatkan perkembangan Teknologi Informasi dan Komunikasi (TIK). Dimanfaatkannya teknologi sebagai media pembelajaran dalam proses belajar mengajar, dapat mempermudah cara mengajar dalam berkomunikasi dan berinteraksi dengan para siswa baik di dalam kelas maupun di luar kelas (Anshori, 2018). Penggunaan TIK tidak hanya dapat mendukung perkembangan kognitif siswa tetapi juga meningkatkan motivasi belajar dan interaksi siswa dalam pembelajaran (Suryani, 2010). Untuk memenuhi standart pendidikan dengan pemanfaatan Teknologi Informasi dengan menggunakan perangkat komputer atau gadget yang saling terhubung antara siswa dan guru sehingga melalui pemanfaatan teknologi tersebut proses belajar mengajar bisa tetap dilaksanakan dengan baik (Pakpahan & Fitriani, 2020). Perkembangan TIK sebagai media pembelajaran harus benar-benar dipertimbangakan karena jka tidak tepat guna dapat memberikan dampak buruk pada manfaat belajaran. Salah satu hasil dari perkembangan TIK yang dapat digunakan guru sebagai media guna menunjang pembelajaran daring di masa pandemi yaitu aplikasi zoom. Zoom dapat dikategorikan sebagai media pembelajaran online yang dapat diartikan sebagai suatu jenis belajar mengajar yang memungkinkan tersampaikannya bahan ajar ke siswa dengan menggunakan media Internet (Monica & Fitriawati, 2020). Aplikasi zoom merupakan sebuah sebuah media pengajaran daring yang berupa aplikasi yang mempunyai fitur konferensi jarak jauh dengan menyatukan konferensi video, obrolan online, pertemuan online dan juga bisa dikolaborasikan dengan seluler (Rosyid http://ojs.unpkediri.ac.id/pgsd 51 51 │Volume 7│Nomor 1│Juli 2021 │Volume 7│Nomor 1│Juli 2021 http://ojs.unpkediri.ac.id/pgsd http://ojs.unpkediri.ac.id/pgsd Eko, Muhammad, Nurlaili. Analisis Manfaat Penggunaan Aplikasi... , , gg p dkk., 2020). Selain itu penggunaan zoom dapat memberikan dampak yang positif bagi sistem pembelajaran daring di sekolah-sekolah. Melalui zoom dijadikan pembelajaran online jarak jauh menjadikan pembelajaran lebih efektif, karena zoom menyediakan video konferensi yang dapat dijangkau oleh seluruh partisipan atau siswa dan guru (Brahma, 2020). , , gg p dkk., 2020). Selain itu penggunaan zoom dapat memberikan dampak yang positif bagi sistem pembelajaran daring di sekolah-sekolah. Melalui zoom dijadikan pembelajaran online jarak jauh menjadikan pembelajaran lebih efektif, karena zoom menyediakan video konferensi yang dapat dijangkau oleh seluruh partisipan atau siswa dan guru (Brahma, 2020). Berdasarkan hasil studi pendahuluan yang dilasanakan di Sekolah Islam Terpadu Ahmad Dahlan Kota Jambi kelas VC didapati bahwa proses pembelajaran yang dilaksanakan di sekolah ini dilakukan secara daring sejak adanya pandemi Covid-19. Terdapat beberapa media pembelajaran yang digunakan dalam menunjang kegiatan pembelajaran daring diantaranya aplikasi zoom, google classroom, google form, WhatsApp Group dan video pembelajaran. Menurut wali kelas dalam menunjang pembelajaran daring untuk siswa, maka beliau menggunakan salah satu perkembangan TIK yaitu aplikasi Zoom. Selain itu pembelajaran daring dikombinasikan juga dengan grup whatsapp dan video pembelajaran yang dibuat sendiri. Guru mengemukakan bahwa aplikasi zoom yang digunakan dalam pembelajaran daring saat ini sangat banyak memberikan manfaat baik bagi guru maupun siswa. Berdasarkan latar belakang yang telah diapaparkan tujuan dalam penelitian ini yaitu untuk menganalisis manfaat dari penggunaan aplikasi zoom dalam pembelajaran daring bagi guru dan siswa di sekolah dasar. Hal ini dikarenakan dalam pembelajaran daring ataupun pembelajaran jarak jauh yang dilaksanakan khususya pada jenjang sekolah dasar, guru dituntut untuk mampu memilih media yang efektif sebagai penunjang pembelajaran yang dilaksanakan. METODE Penelitian ini menggunakan pendekatan kualitatif. Metode penelitian kualitatif digunakan untuk menjawab masalah penelitian terkait data berupa narasi yang berasal dari wawancara, observasi, dan ekstraksi dokumen (Wahidmurni, 2017). Dengan jenis penelitian fenomenologi. Studi fenomenologi bertujuan untuk menggali informasi secara mendalam para subjek mengenai pengalamannya dalam satu peristiwa (Habiansyah, 2008). Penelitian ini dilaksanakan di kelas VC Sekolah Dasar Islam Terpadu Ahmad Dahlan, pada semester genap tahun ajaran 2020/2021. Dengan tahap persiapan, tahap pelaksanaan, dan tahap penyelesaian. http://ojs.unpkediri.ac.id/pgsd │Volume 7│Nomor 1│Juli 2021 52 http://ojs.unpkediri.ac.id/pgsd │Volume 7│Nomor 1│Juli 2021 52 │Volume 7│Nomor 1│Juli 2021 http://ojs.unpkediri.ac.id/pgsd HASIL Berdasarkan hasil observasi dan wawancara bersama guru dan siswa, didapati beberapa manfaat penggunaan aplikasi zoom dalam pembelajaran daring bagi guru dan siswa. 1. Manfaat Penggunaan Zoom Bagi Guru a. Memudahkan Mencapai Tujuan Pembelajaran http://ojs.unpkediri.ac.id/pgsd http://ojs.unpkediri.ac.id/pgsd Eko, Muhammad, Nurlaili. Analisis Manfaat Penggunaan Aplikasi... Eko, Muhammad, Nurlaili. Analisis Manfaat Penggunaan Aplikasi... Data dalam penelitian ini dibedakan menjadi dua yaitu data primer dan data sekunder. Sumber data primer dalam penelitian ini terdiri dari 2 orang guru kelas V dan 6 orang siswa kelas VC. Sedangkan sumber data sekunder dalam penelitian ini yaitu diperoleh dari buku-buku dan jurnal yang relevan dengan penelitian ini. Teknik pengumpulan data yang digunakan dalam penelitian ini yaitu observasi dan wawancara mendalam. Untuk menyajikan data agar mudah dipahami, maka peneliti menggunakan teknik analisis model Miles dan Huberman dengan tahapan 1) Pengumpulan data, pada analisis model pertama dilakukan pengumpulan data dari berbagai kegiatan antara lain wawancara, observasi, dan berbagai dokumentasi berdasarkan topik masalah penelitian yang kemudian dikembangkan penajaman data melalui pencarian data selanjutnya; 2) Reduksi data, reduksi data adalah kegiatan mengolah data yang diperoleh dan dianalisis dengan sedemikian rupa sehingga dapat menarik simpulan dan diverifikasi (Miles dan Huberman, 2007); 3) Penyajian data, penyajian data dilakukan guna menemukan pola- pola yang bermakna. memberikan adanya penarikan simpulan dan memberikan tindakan (Miles dan Huberman, 2007); 4) Penarikan Kesimpulan, tahap penarikan kesimpulan adalah bagian dari suatu kegiatan konfigurasi yang utuh (Miles dan Huberman, 2007). b. Memudahkan Berinteraksi Bersama Siswa b. Memudahkan Berinteraksi Bersama Siswa b. Memudahkan Berinteraksi Bersama Siswa Berdasarkan hasil wawancara dan observasi didapati bahwa aplikasi zoom ini mampu menampung seluruh siswa dengan satu kali pertemuan. Guru mengungkapkan bahwa dalam suasana pembelajaran daring ini aplikasi zoom membantu guru untuk mempermudah berkomunikasi dengan siswa dibanding aplikasi whatsapp. Disamping itu banyak fitur yang menunjang pembelajaran daring lebih mudah tersampaikan dengan baik kepada siswa di rumah. Guru dapat menampilkan materi sekaligus menjelaskannya langsung kepada siswa melalui aplikasi zoom. Siswa dapat bertanya langsung kepada guru layaknya di dalam kelas. Pembelajaran tidak monoton hanya mengerjakan tugas saja, tetapi guru juga memberikan kuis kepada siswa. kuis tersebut berupa pertanyaan mengenai materi pembelajaran yang telah dilaksanakan. Siswa berlomba-lomba menjawab pertanyaan tersebut agar bisa mendapatkan point berupa bintang. Kemudian bintang tersebut dikumpulkan agar dapat menjadi bintang kelas. Bintang kelas bukan dia yang mendapatkan rangking 1 tetapi dia yang sering menjawab ketika pembelajaran berlangsung. Interaksi yang tercipta antara guru dengan siswa dalam pembelajaran melalui zoom mampu membuat pembelajaran terlaksana dengan baik. a. Memudahkan Mencapai Tujuan Pembelajaran Berdasarkan hasil wawancara dan observasi, diketahui bahwa dalam pembelajaran daring guru menyampaikan tujuan diawal pembelajaran. Penyampaian ini bertujuan untuk memberitahukan ke siswa mengenai apa saja yang harus mereka capai saat pembelajaran. Guru mengungkapkan dengan menggunakan aplikasi zoom mampu untuk mencapai tujuan pembelajaran yang telah dirancang hingga 80%. Demi mencapai tujuan tersebut guru kelas membuat suasana pembelajaran di zoom sama seperti pembelajaran tatap muka di kelas. Guru tp://ojs.unpkediri.ac.id/pgsd │Volume 7│Nomor 1│Juli 2021 53 http://ojs.unpkediri.ac.id/pgsd │Volume 7│Nomor 1│Juli 2021 53 http://ojs.unpkediri.ac.id/pgsd Eko, Muhammad, Nurlaili. Analisis Manfaat Penggunaan Aplikasi... memulai pembelajaran dengan memberikan semangat kepada siswa, mengkondisikan siswa agar siap untuk mengikuti pembelajaran. Setelah itu guru menyampaikan tujuan, menjelaskan materi pembelajaran, memberikan kesempatan siswa untuk bertanya, memberikan penugasan pada siswa, dan mengevaluasi hasil yang didapat oleh siswa. Eko, Muhammad, Nurlaili. Analisis Manfaat Penggunaan Aplikasi... memulai pembelajaran dengan memberikan semangat kepada siswa, mengkondisikan siswa agar siap untuk mengikuti pembelajaran. Setelah itu guru menyampaikan tujuan, menjelaskan materi pembelajaran, memberikan kesempatan siswa untuk bertanya, memberikan penugasan pada siswa, dan mengevaluasi hasil yang didapat oleh siswa. c. Memberikan Variasi Dalam Pembelajaran Daring c. Memberikan Variasi Dalam Pembelajaran Daring Berdasarkan hasil wawancara dan observasi, diketahui bahwa penggunaan aplikasi zoom ini menjadi salah satu variasi yang digunakan guru untuk menunjang proses pembelajaran daring. Pada awalnya guru hanya menggunakan aplikasi whatsapp untuk menunjang proses pembelajaran daring. Kemudian guru mencoba untuk menggunakan aplikasi yang lain dan aplikasi tersebut adalah aplikasi zoom. Guru mengungkapkan bahwa sebagai seorang guru harus mampu menguasai perkembangan teknologi informasi dan http://ojs.unpkediri.ac.id/pgsd │Volume 7│Nomor 1│Juli 2021 54 http://ojs.unpkediri.ac.id/pgsd │Volume 7│Nomor 1│Juli 2021 54 │Volume 7│Nomor 1│Juli 2021 http://ojs.unpkediri.ac.id/pgsd http://ojs.unpkediri.ac.id/pgsd Eko, Muhammad, Nurlaili. Analisis Manfaat Penggunaan Aplikasi... komunikasi agar dapat menciptakan suasana baru bagi siswa supaya tidak bosan dalam belajar terutama belajar daring. Guru juga mengungkapkan bahwa setelah menggunakan aplikasi zoom untuk menunjang pembelajaran daring ternyata banyak kelebihan yang diperolehnya. d. Menciptakan Pembelajaran Daring Yang Aktif Berdasarkan hasil wawancara dan observasi didapati bahwa proses pembelajaran menggunakan aplikasi zoom ini cukup aktif. Dimana proses pembelajaran tidak monoton mengerjakan tugas saja, tetapi ada interaksi secara langsung antara guru dengan siswa dan siswa dengan siswa. sebelum pembelajaran dimulai guru memberikan kesempatan kepada siswa untuk saling menyapa satu sama lain sambil menunggu tema-teman yang lain bergabung ke dalam zoom. Siswa selalu diberi kesempatan oleh guru untuk bertanya apabila ada materi yang belum atau kurang dipahami. Kemudian di akhir pembelajaran guru memberikan kuis berupa pertanyaan seputar materi yang telah dipelajari. Siswa yang dapat menjawab pertanyaan diberikan reward berupa poin. c. Menghilangkan Rasa Bosan Belajar Daring Berdasarkan hasil wawancara dan observasi, didapati bahwa belajar melalui aplikasi zoom dapat menghilangkan rasa bosan siswa mengikuti pembelajaran daring. Siswa mengungkapkan bahwa karena mereka dapat bertemu serta belajar bersama guru dan teman-temannya serta materi yang diberikan guru juga menarik. Siswa terlihat sangat antusias belajar melalui zoom karena kegiatan pembelajaran tidak monoton, tetapi siswa dapat mengikuti kegiatan pembelajaran layaknya di dalam kelas walaupun melalui aplikasi zoom. Beda halnya dengan belajar melalui whatsapp, siswa hanya diberikan tugas terus menerus, hal ini yang membuat siswa bosan. Berdasarkan hasil penelitian yang telah didapatkan, aplikasi zoom memberikan banyak manfaat bagi guru dan siswa pada pembelajaran daring di masa pandemi ini. Manfaat tersebut akan membuat pembelajaran yang dilakukan oleh guru dapat mencapai tujuan yang telah dirancang sebelumnya. Adapun manfaat bagi guru diantaranya memudahkan mencapai tujuan, memudahkan berinteraksi bersama siswa, memberikan variasi dalam pembelajaran daring untuk siswa, dan menciptakan pembelajaran daring yang aktif. a. Memudahkan Memahami Materi Berdasarkan hasil wawancara dan observasi didapati bahwa guru menyampaikan tujuan pembelajaran melalui fitur yang terdapat pada zoom yaitu share screen. Kemudian guru menjelaskan materi dengan menampilkan materi pembelajaran. penjelasan oleh guru dengan tampilan materi pembelajaran secara bersamaan membuat siswa lebih mudah memahami materi. Siswa juga mengungkapkan bahwa belajar dengan menggunakan zoom ini lebih mudah untuk memahami materi dibandingkan dengan aplikasi whatsapp. Materi yang belum dipahami siswa dapat ditanyakan secara langsung kepada uru dan guru dapat menjawab pertanyaan tersebut secara langsung. ada yang menarik dari kegiatan pembelajaran melalui aplikasi zoom yaitu setelah proses pembelajaran berakhir uru menanyakan kepada siswa siapa saja yang belum paham terhadap pembelajaran yang telah dilaksanakan. Kemudian guru memberikan kesempatan kepada siswa yang belum paham untuk tetap berada http://ojs.unpkediri.ac.id/pgsd │Volume 7│Nomor 1│Juli 2021 55 http://ojs.unpkediri.ac.id/pgsd │Volume 7│Nomor 1│Juli 2021 55 │Volume 7│Nomor 1│Juli 2021 http://ojs.unpkediri.ac.id/pgsd http://ojs.unpkediri.ac.id/pgsd http://ojs.unpkediri.ac.id/pgsd Eko, Muhammad, Nurlaili. Analisis Manfaat Penggunaan Aplikasi... di dalam zoom untuk mengikuti kembali pembelajaran dari awal yang akan diulang oleh guru dan yang sudah paham dapat meninggalkan zoom (leave). b. Meningkatkan Semangat Belajar Daring Berdasarkan hasil wawancara dan observasi, diketahui bahwa pada kegiatan pembelajaran siswa bersemangat karena dapat belajar bersama teman- teman dan dapat melihat langsung guru menjelaskan pembelajaran. Pembelajaran daring yang tidak menggunakan zoom rentan membuat siswa tidak bersemangat dalam mengikuti pembelajaran karena mereka tidak dapat belajar bersama. Ketika menggunakan zoom mereka mengatakan bahwa lebih bersemangat belajarnya daripada belajar menggunakan media yang lain. PEMBAHASAN http://ojs.unpkediri.ac.id/pgsd │Volume 7│Nomor 1│Juli 2021 56 │Volume 7│Nomor 1│Juli 2021 http://ojs.unpkediri.ac.id/pgsd http://ojs.unpkediri.ac.id/pgsd Eko, Muhammad, Nurlaili. Analisis Manfaat Penggunaan Aplikasi... Dalam pembelajaran daring saat ini pencapaian tujuan pembelajaran tetap harus diperhatikan oleh guru, hal ini guna untuk memenuhi kebutuhan siswa akan pemahaman terhadap materi yang diberikan. Menurut Kurniasari dkk., (2020) mengungkapkan bahwa baik pembelajaran daring maupun luring guru harus memperhatikan tercapainya tujuan pembelajaran. Guna mencapai tujuan pembelajaran daring diperlukan penggunaan media yang merupakan hasil dari perkembangan teknologi yang memungkinkan pembelajaran lebih efektif dan berorientasi tujuan (Kuntarto & Asyhar, 2016). Salah satu media yang tepat untuk mencapai tujuan pembelajaran di masa pandemi saat ini adalah aplikasi zoom. Hal ini sejalan dengan Liu & Ilyas, (2020) yang mengungkapkan bahwa pembelajaran daring yang menggunakan zoom memiliki pengaruh terhadap hasil belajar siswa, yang mana tercapainya minimal nilai 70. Selain itu demi mencapai tujuan pembelajaran tersebut diperlukan interaksi yang aktif antara guru dan siswa, terlebih di masa pandemi covid-19. Buana (2020) mengemukakan salah satu prinsip terjadinya pembelajaran daring yaitu tersedianya ruang bagi guru dan siswa untuk berinteraksi dua arah dan bersifat interaktif. Dengan meggunakan aplikasi zoom yang memiliki berbagai fitur didalamnya mampu untuk memfasilitasi interaksi yang aktif antara guru dan siswa. Pembelajaran daring menggunakan zoom yang memiliki fitur video conference dapat mengoptimalkan interaksi siswa dan siswa maupun siswa dan guru (Ismawati & Prasetyo, 2020). Pembelajaran daring saat ini rentan membuat siswa merasakan kebosanan saat belajar, seperti bosan ketika belajar dengan metode/media yang monoton. Oleh karena ini guru harus membuat sebuah variasi dalam pembelajaran daring agar siswa memiliki minat belajar yang baik. Salah satu upaya yang dapat dilakukan guru yaitu dengan cara menggunakan aplikasi zoom sebagai media pembelajaran, yang mana dapat memfasilitasi siswa untuk merasakan suasana beajar yang baru. Seperti dapat bertatap muka bersama teman dan guru, dan belajar bersama. Menurut Siahaan (2020) menyatakan bahwa pembelajaran yang bervariasi dapat memberikan manfaat untuk mengembangkan pemikiran siswa melalui analisis mereka sendiri. Selain memberikan variasi dalam pembelajaran, penggunaan zoom juga mampu menciptakan pembelajaran daring yang lebih aktif, daripada media pembelaan yang lain. Dengan menggunakan zoom, guru bisa memberikan pembelajaran dan siswa bisa berkomunikasi langsung untuk menanyakan berbagai permasalahan dalam materi http://ojs.unpkediri.ac.id/pgsd │Volume 7│No 57 │Volume 7│Nomor 1│Juli 2021 57 http://ojs.unpkediri.ac.id/pgsd http://ojs.unpkediri.ac.id/pgsd │Volume 7│Nomor 1│Juli 2021 Eko, Muhammad, Nurlaili. Analisis Manfaat Penggunaan Aplikasi... Eko, Muhammad, Nurlaili. Analisis Manfaat Penggunaan Aplikasi... Eko, Muhammad, Nurlaili. Analisis Manfaat Penggunaan Aplikasi... pembelajaran secara interaktif dengan guru (Solihin, 2020). Dengan menggunakan zoom guru dapat membuat pembelajaran lebih menarik, karena aplikasi zoom ini memiliki fitur-fitur yang dapat menampilkan bahan ajar yang menarik kepada peserta didik. Pratama & Mulyati (2020) mengungkapkan bahwa dalam pembelajaran daring siswa cenderung lebih menyukai pembelajaran yang menarik dan akan membuat siswa penasaran serta siswa menjadi aktif dalam proses belajar. Tidak hanya bermanfaat bagi guru, penggunaan aplikasi zoom juga memberikan manfaat yang baik bagi siswa untuk melaksanakan pembelajaran daring di masa pandemi. Manfaat tersebut diantaranya siswa lebih mudah memahami materi pembelajaran, meningkatkan semangat belajar siswa, dan menghilangkan rasa bosan dalam belajar daring. Meskipun pembelajaran dilaksanakan secara daring, pemahaman siswa terhadap materi pelajaran yang diberikan oleh guru juga perlu diperhatikan. Pemahaman akan materi pelajaran oleh siswa tidak hanya memberikan pengaruh terhadap ketercapaian pembelajaran akan tetapi berpengaruh juga terhadap perkembangan kognitif siswa tersebut. Oleh karena itu di dalam pembelajaran daring penggunaan media online merupakan salah satu solusi agar siswa dapat memahami materi pelajaran dengan baik (Mustakim, 2020). Aplikasi zoom merupakan media berbasis multimedia yang memiliki fitur-fitur pendukung dalam memudahkan siswa untuk memahami materi. Sejalan dengan Monica & Fitriawati (2020) yang mengungkapkan sebagian besar siswa dapat memahami materi melalui pembelajaran daring yang menggunakan aplikasi zoom. Selain itu penggunaan zoom mampu meningkatkan semangat siswa untuk belajar dan menghilangkan rasa bosan untuk belajar daring. Kondisi pandemi saat ini semangat belajar siswa menjadi salah satu aspek yang sangat penting dalam pembelajaran. Salah satu strategi yang dapat dilakukan guru agar siswa bersemangat dan tidak bosan belajar daring yaitu dengan memfasilitasi siswa agar dapat belajar bersama teman-temannya. Karena belajar bersama dengan teman sebaya akan membuat siswa lebih semangat untuk belajar dan ketika siswa merasa bingung, bimbang, dan bosan maka teman sebaya yang bisa mengurangi perasaan tersebut (Ningsih & Djumali, 2020). Aplikasi zoom dapat menjadi media untuk melakukan pembelajaran bersama-sama, karena memiliki video conference yang mampu menampung banyak peserta. Pembelajaran menggunakan zoom, membuat guru dapat menampilkan materi pembelajaran dan http://ojs.unpkediri.ac.id/pgsd │Volume 7│Nomor 1│Juli 2021 58 58 http://ojs.unpkediri.ac.id/pgsd │Volume 7│Nomor 1│Juli 2021 58 58 │Volume 7│Nomor 1│Juli 2021 │Volume 7│Nomor 1│Juli 2021 http://ojs.unpkediri.ac.id/pgsd http://ojs.unpkediri.ac.id/pgsd Eko, Muhammad, Nurlaili. Analisis Manfaat Penggunaan Aplikasi... Eko, Muhammad, Nurlaili. Analisis Manfaat Penggunaan Aplikasi... menjelaskannya secara bersamaan kepada siswa. Hal tersebut dikarenakan zoom memiliki fitur share screen yaitu fitur yang dapat menampilkan materi ajar baik berupa power point, video, dokumen, dan pdf yang membuat pembelajaran menyenangkan (Kurniawansyah, 2020). Berdasarkan hasil penelitian dan pembahasan yang telah dipaparkan, maka dapat ditarik kesimpulan bahwa pembelajaran daring di masa pandemi saat ini harus dapat berjalan dengan efektif. Salah satu media yang dapat digunakan guru dalam pembelajaran daring adalah aplikasi zoom. Penggunaan aplikasi ini mampu memberikan manfaat yang baik bagi guru dan siswa, sehingga pembelajaran daring dapat berjalan dengan efektif. SIMPULAN Akibat dari adanya pandemic Covid-19 menyebabkan berubahnya sistem pendidikan yang ada, yang semulanya pendidikan dilakukan tatap muka berubah menjadi daring. Pembelajaran secara daring menuntut setiap pendidik untuk memanfaatkan perkembangan teknologi yang ada. Aplikasi zoom merupakan salah satu produk dari perkembangan teknologi saat ini, selain itu zoom dapat digunakan sebagai media pembelajaran daring. Zoom yang digunakan sebagai media pembelajaran daring dimasa pandemi memberikan manfaat yang baik bagi guru dan siswa. Manfaat bagi guru diantaranya memudahkan mencapai tujuan pembelajaran, memudahkan berinteraksi bersama siswa, memberikan variasi dalam pembelajaran daring, menciptakan pembelajaran daring yang aktif , sedangkan manfaat penggunaan zoom bagi siswa diantaranya memudahkan memahami materi, meningkatkan semangat belajar daring, menghilangkan rasa bosan belajar daringManfaat yang dirasakan guru dan siswa tersebut memberikan dampak yang positif dalam pembelajaran daring sehingga menjadi lebih efektif. DAFTAR RUJUKAN Anshori, S. (2018). Pemanfaatan Teknologi Informasi Dan Komunikasi Sebagai Media Pembelajaran. Civic-Culture: Jurnal Ilmu Pendidikan PKn dan Sosial Budaya, 2(1). Anshori, S. (2018). Pemanfaatan Teknologi Informasi Dan Komunikasi Sebagai Media Pembelajaran. Civic-Culture: Jurnal Ilmu Pendidikan PKn dan Sosial Budaya, 2(1). http://ojs.unpkediri.ac.id/pgsd │Volume 7│Nomor 1│Juli 2021 59 http://ojs.unpkediri.ac.id/pgsd │Volume 7│Nomor 1│Juli 2021 59 │Volume 7│Nomor 1│Juli 2021 http://ojs.unpkediri.ac.id/pgsd Eko, Muhammad, Nurlaili. Analisis Manfaat Penggunaan Aplikasi... gg p Brahma, I. A. (2020). Penggunaan zoom sebagai pembelajaran berbasis online dalam mata kuliah sosiologi dan antropologi pada mahasiswa PPKN di STKIP Kusumanegara Jakarta. Aksara: Jurnal Ilmu Pendidikan Nonformal, 6(2), 97- 102. Buana, C. A. (2020). Penerapan desain rencana pembelajaran sebagai upaya membangun interaksi dalam pembelajaran daring (Doctoral dissertation, Universitas Pelita Harapan). Handarini, O. I., & Wulandari, S. S. (2020). Pembelajaran Daring Sebagai Upaya Study From Home (SFH) Selama Pandemi Covid 19. Jurnal Pendidikan Administrasi Perkantoran (JPAP), 8(3), 496-503. Hasbiansyah, O. (2008). Pendekatan fenomenologi: Pengantar praktik penelitian dalam Ilmu Sosial dan Komunikasi. Mediator: Jurnal Komunikasi, 9(1), 163-180. Ismawati, Dwi, and Iis Prasetyo. 2020. “Efektivitas Pembelajaran Menggunakan Video Zoom Cloud Meeting Pada Anak Usia Dini Era Pandemi Covid-19.” Jurnal Obsesi : Jurnal Pendidikan Anak Usia Dini 5(1): 665. Kuntarto, E., & Asyhar, R. (2016). Pengembangan Model Pembelajaran Blended Learning Pada Aspek Learning Design Dengan Platform Media Sosial Online Sebagai Pendukung Perkuliahan Mahasiswa. Repository Unja. Kurniasari, A., Pribowo, F. S. P., & Putra, D. A. (2020).Analisis Efektivitas Pelaksanaan Belajar Dari Rumah (Bdr) Selama Pandemi Covid-19. Jurnal Review Pendidikan Dasar: Jurnal Kajian Pendidikan Dan Hasil Penelitian, 6(3), 246-253. Kurniawansyah, A. S. (2020). Kolaborasi Whatsapp Group, Zoom Cloud Meeting, Dan Google Drive Sebagai Formula Dalam Pelaksanaan Kegiatan Perkuliahan Online Di Masa Pandemi Covid-19. JURNAL MEDIA INFOTAMA, 16(2). Liu, A. N. A. M. M., & Ilyas, I. (2020). Pengaruh Pembelajaran Online Berbasis Zoom Cloud Meeting Terhadap Hasil Belajar Mahasiswa Fisika Universitas Flores. Jurnal Pendidikan Fisika Dan Keilmuan (JPFK), 6(1), 34-38. http://ojs.unpkediri.ac.id/pgsd │Volume 7│Nomor 1│Juli 2021 60 60 60 http://ojs.unpkediri.ac.id/pgsd │Volume 7│Nomor 1│Juli 2021 Eko, Muhammad, Nurlaili. Analisis Manfaat Penggunaan Aplikasi... Eko, Muhammad, Nurlaili. Analisis Manfaat Penggunaan Aplikasi... Miles, M. B. & Huberman, A. M. (2007). Analisis Data Kualitatif. Diterjemahkan oleh Rohidi. Jakarta: UI-Press. Monica, J., & Fitriawati, D. (2020). Efektivitas Penggunaan Aplikasi Zoom Sebagai Media Pembelajaran Online Pada Mahasiswa Saat Pandemi Covid-19. Jurnal Communio: Jurnal Jurusan Ilmu Komunikasi, 9(2), 1630-1640. Monica, J., & Fitriawati, D. (2020). Efektivitas Penggunaan Aplikasi Zoom Sebagai Media Pembelajaran Online Pada Mahasiswa Saat Pandemi Covid-19. DAFTAR RUJUKAN Jurnal Communio: Jurnal Jurusan Ilmu Komunikasi, 9(2), 1630-1640. Mustakim, M. (2020). Efektivitas pembelajaran daring menggunakan media online selama pandemi covid-19 pada mata pelajaran matematika. Al asma: Journal of Islamic Education, 2(1), 1-12. Mustakim, M. (2020). Efektivitas pembelajaran daring menggunakan media online selama pandemi covid-19 pada mata pelajaran matematika. Al asma: Journal of Islamic Education, 2(1), 1-12. Ningsih, L. K., & Djumali, M. P. (2020). Kejenuhan Belajar Masa Pandemi Covid-19 Siswa SMTA Di Kedungwungu Indramayu (Doctoral dissertation, Universitas Muhammadiyah Surakarta). Pakpahan, R., & Fitriani, Y. (2020). Analisa pemanfaatan teknologi informasi dalam pembelajaran jarak jauh di tengah pandemi virus corona covid-19. Journal of Information System, Applied, Management, Accounting and Research, 4(2), 30- 36. Putria, H., Maula, L. H., & Uswatun, D. A. (2020). Analisis proses pembelajaran dalam jaringan (daring) masa pandemi covid-19 pada guru sekolah dasar. Jurnal Basicedu, 4(4), 861-870. Riyana, C. (2019). Produksi Bahan Pembelajaran Berbasis Online. Universitas Terbuka. Rosyid, N. M., Thohari, I., & Lismanda, Y. F. (2020). Penggunaan Aplikasi Zoom Cloud Meetings Dalam Kuliah Statistik Pendidikan Di Fakultas Agama Islam Universitas Islam Malang. Vicratina: Jurnal Pendidikan Islam, 5(11), 46-52. 61 │Volume 7│Nomor 1│Juli 2021 http://ojs.unpkediri.ac.id/pgsd Eko, Muhammad, Nurlaili. Analisis Manfaat Penggunaan Aplikasi... Sadikin, A., & Hamidah, A. (2020). Pembelajaran Daring di Tengah Wabah Covid- 19:(Online Learning in the Middle of the Covid-19 Pandemic). Biodik, 6(2), 214-224. Siahaan, M. (2020). Dampak Pandemi Covid-19 Terhadap Dunia Pendidikan. Dampak Pandemi Covid-19 Terhadap Dunia Pendidikan, 20(2). Sofyana, L., & Rozaq, A. (2019). Pembelajaran Daring Kombinasi Berbasis Whatsapp Pada Kelas Karyawan Prodi Teknik Informatika Universitas Pgri Madiun. Jurnal Nasional Pendidikan Teknik Informatika: JANAPATI, 8(1), 81- 86. Solihin, A. (2020). Pembelajaran online dengan Aplikasi Zoom Meeting di Kelas 5 SDN 1 Selaawi di Masa Pandemi Covid-19. Gunahumas, 3(2). Suryani, A. (2010). ICT in education: Its benefits, difficulties, and organizational development issues. Jurnal Sosial Humaniora (JSH), 3(1), 13-33. Wahidmurni, W. (2017). Pemaparan metode penelitian kualitatif. 1-17. http://ojs.unpkediri.ac.id/pgsd │Volume 7│Nomor 1│Juli 2021 62 62
https://openalex.org/W4246902141	https://acp.copernicus.org/articles/21/14293/2021/acp-21-14293-2021.pdf	English	null	Comment on acp-2021-23	null	2,021	cc-by	13,059	Correspondence: F. Zhang (fang.zhang@bnu.edu.cn) Received: 12 January 2021 – Discussion started: 1 March 2021 Revised: 25 June 2021 – Accepted: 19 July 2021 – Published: 27 September 2021 Received: 12 January 2021 – Discussion started: 1 March 2021 Revised: 25 June 2021 Accepted: 19 July 2021 Published: 27 September 2021 Received: 12 January 2021 – Discussion started: 1 March 2021 Revised: 25 June 2021 – Accepted: 19 July 2021 – Published: 27 September 2021 Abstract. The effect of new particle formation (NPF) on cloud condensation nuclei (CCN) varies widely in diverse environments. CCN or cloud droplets from NPF sources re- main highly uncertain in the urban atmosphere; they are greatly affected by the high background aerosols and fre- quent local emissions. In this study, we quantiﬁed the ef- fect of NPF on cloud droplet number concentration (CDNC, or Nd) at typical updraft velocities (V ) in clouds based on ﬁeld observations on 25 May–18 June 2017 in urban Bei- jing. We show that NPF increases the Nd by 32 %–40 % at V = 0.3–3 m s−1 during the studied period. The Nd is re- duced by 11.8 ± 5.0 % at V = 3 m s−1 and 19.0 ± 4.5 % at V = 0.3 m s−1 compared to that calculated from constant supersaturations due to the water vapor competition effect, which suppresses the cloud droplet formation by decreasing the environmental maximum supersaturation (Smax). The ef- fect of water vapor competition becomes smaller at larger V that can provide more sufﬁcient water vapor. However, under extremely high aerosol particle number concentrations, the effect of water vapor competition becomes more pronounced. As a result, although a larger increase of CCN-sized parti- cles by NPF events is derived on clean NPF days when the number concentration of preexisting background aerosol par- ticles is very low, no large discrepancy is presented in the en- hancement of Nd by NPF between clean and polluted NPF days. We ﬁnally reveal a considerable impact of the primary sources on the evaluation of the contribution of NPF to CCN number concentration (NCCN) and Nd based on a case study. Our study highlights the importance of full consideration of both the environmental meteorological conditions and multi- ple sources (i.e., secondary and primary) to evaluate the ef- fect of NPF on clouds and the associated climate effects in polluted regions. 1 Introduction In the global climate system, aerosols, cloud condensation nuclei (CCN) and cloud droplets are very important com- ponents. Clouds, serving as a bridge connecting aerosols and climate, are the most uncertain factor of climate change (IPCC, 2013; Seinfeld et al., 2016; Cai et al., 2020). The mi- crophysical link between aerosols and clouds as the most im- portant part has received extensive attention. Cloud droplet activation is a key process from aerosol to clouds, and researchers have tried to simulate the microphysical pro- cesses using numerical activation models (e.g., Boucher and Lohmann, 1995; Abdul-Razzak et al., 1998; Ghan et al., 1993; Khvorostyanov and Curry, 1999; Abdul-Razzak and Ghan, 2000; Nenes et al., 2001, 2003; Petters and Kreiden- weis, 2007; Ren et al., 2018; Genz et al., 2020). New particle formation (NPF) events have been observed and occur frequently in different atmospheric environments in the world (Spracklen et al., 2010; Yue et al., 2011; Peng Atmos. Chem. Phys., 21, 14293–14308, 2021 https://doi.org/10.5194/acp-21-14293-2021 © Author(s) 2021. This work is distributed under the Creative Commons Attribution 4.0 License. Evaluation of the contribution of new particle formation to cloud droplet number concentration in the urban atmosphere Sihui Jiang1, Fang Zhang2, Jingye Ren1, Lu Chen1, Xing Yan1, Jieyao Liu1, Yele Sun3, and Zhanqing 1College of Global Change and Earth System Science, Beijing Normal University, Beijing 100875, China 2Environmental Science and Engineering Research Center, School of Civil and Environmental Engineering, Harbin Institute of Technology (Shenzhen), 518055 Shenzhen, China 3State Key Laboratory of Atmospheric Boundary Layer Physics and Atmospheric Chemistry, Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, China 4Earth System Science Interdisciplinary Center and Department of Atmospheric and Oceanic Science, University of Maryland, College Park, Maryland, USA Correspondence: F. Zhang (fang.zhang@bnu.edu.cn) S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration 14294 et al., 2017; Kerminen et al., 2018; Bousiotis et al., 2019; Zimmerman et al., 2020). NPF events are one of the most signiﬁcant sources of ﬁne particles in the atmosphere (Shi et al., 1999; Stanier et al., 2004; Kulmala and Kerminen, 2008). For example, it has been found that NPF contributed about 76 % of the total ﬁne particle number concentration in urban Beijing (Wu et al., 2011). These nucleated particles subse- quently grow through coagulation or condensation processes to CCN-relevant sizes or act as CCN in convective clouds (Fan et al., 2013; Li et al., 2010). In reality, ﬁeld studies have shown that these ﬁne particles produced from NPF can sub- sequently result in an enhancement in NCCN at cloud-relevant supersaturation (Kalkavouras et al., 2017; Peng et al., 2014; Wu et al., 2015; Ma et al., 2016; Z. Li et al., 2017; Zhang et al., 2019). It was estimated that up to 80 % of CCN number concentration (NCCN) is from the nucleation process in urban Beijing (Wiedensohler et al., 2008). larger than those of a relatively clean atmosphere. Wieden- sohler et al. (2012) found that, under the high concentration levels of gaseous pollutants and strong oxidation in polluted areas, the high concentration of nanoparticles generated by NPF events can rapidly grow to tens or even hundreds of nanometers in a few hours. Zhang et al. (2019) observed that the subsequent growth of newly formed particles can last 2– 3 d in urban Beijing, producing more CCN-sized particles. Previous studies in polluted regions demonstrated the com- plex and nonlinear relationship between aerosol particles and CCN due to multiple emission sources (Zhang et al., 2014, 2016, 2017, 2019; Ren et al., 2018; Fan et al., 2020), high- lighting the importance of understanding the connections be- tween aerosols and CCN or cloud droplets close to the source regions. Particularly, owing to the extremely high CN num- ber concentrations (with order of magnitude as high as 104 or even 105 cm−3) during NPF events in the urban area, the effect of competition for water vapor and reduction in cloud supersaturation is expected to be exacerbated. However, the NCCN only reﬂects the cloud-forming po- tential of aerosol particles at a given supersaturation. The measurement of CCN is usually carried out at constant su- persaturations. S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration Different from the prescribed supersaturation used in the evaluation of NCCN, when calculating the cloud droplet number concentration (CDNC, or Nd), researchers considered the dynamic situations in clouds. In clouds, the supersaturation exhibits variable levels that instantaneously adjust to the intensity of cloud updrafts and the particle num- ber size distribution (PNSD) (Nenes et al., 2003; Hudson et al., 2015). So the CDNC (or Nd) depends on the size distri- bution, chemical properties of aerosol and the cloud updraft velocity, all of which regulate the maximum supersaturation (Smax) that can be formed in a cloud parcel (Nenes and Se- infeld, 2003). Studies have shown that the CDNC in clouds exhibits a sublinear relationship to aerosol number concen- tration (NCN) (Twomey, 1977; Leaitch et al., 1986; Ghan et al., 1993; Boucher and Lohmann, 1995; Nenes et al., 2001; Ramanathan et al., 2001; Sullivan et al., 2016); this is differ- ent from CCN due to the limitation of the water vapor in the actual environment. Using the prescribed supersaturation to calculate CDNC may therefore provide a bias in the evalua- tion of the aerosol indirect effect. For example, Kalkavouras et al. (2017, 2019) reported an average 12 % enhancement of CDNC during two consecutive NPF episodes in the eastern Mediterranean, which was signiﬁcantly smaller than the en- hancement of NCCN (∼87 %) during the NPF events. Hence, it is critical to fully consider the background meteorological conditions (e.g., using dynamic water vapor under different updraft velocities) to simulate the Smax when evaluating the effect of NPF on clouds and the associated climate effects. The current study quantiﬁes the contribution of NPF to NCCN and CDNC in the polluted urban atmosphere of Bei- jing using ﬁeld measurements of aerosol number size distri- butions and chemical composition. The effect of water vapor competition on evaluating Nd during NPF events is exam- ined. The impact of the background preexisting particles on the enhancement of CCN and CDNC is also discussed by contrasting the results on typical clean NPF days and pol- luted NPF days. Given the strong local primary sources like trafﬁc emissions in the urban area, a case study is conducted to investigate the impact of primary emissions on the evalua- tion of the effect of NPF on Nd. Published by Copernicus Publications on behalf of the European Geosciences Union. ublished by Copernicus Publications on behalf of the European Geosciences Union. https://doi.org/10.5194/acp-21-14293-2021 S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration 14295 The number size distribution of particles in the size range from 10 to 550 nm (scanned range) was measured with a time resolution of 5 min by a scanning mobility particle sizer (SMPS; Wang and Flagan, 1990; Collins et al., 2002), which consists of a differential mobility analyzer (DMA, model 3081L, TSI Inc.) to classify particles with different sizes of particles, and a condensation particle counter (CPC, model 3772, TSI Inc.) to detect the size-classiﬁed particles. The sampled particles were dried to a relative humidity < 30 % before entering the DMA. The non-refractory chemical com- position of PM1 is measured by an aerosol chemical specia- tion monitor (ACSM), which consists of an aerodynamic lens to efﬁciently sample and focus submicron particles into the ACSM (Ng et al., 2011). Before sampling into the ACSM, aerosol particles are dried by silica gel desiccant. The ACSM was operated at a time resolution of 15 min. And the non- refractory chemical components that can be measured mainly include organics, sulfate salts (SO2− 4 ), nitrate salts (NO− 3 ), ammonium (NH+ 4 ) and chloride (Cl−) (Ng et al., 2011). The refractory components mainly include black carbon (BC), and the BC mass concentration was measured using a seven- wavelength aethalometer (AE33, Magee Scientiﬁc Corp) . where κi and εi are the hygroscopic parameter and volume fraction for each individual (dry) component in the mixture, respectively. The κ value and density (ρ) of each species used in the calculation are given in Table 1, which are from Petters and Kreidenweis (2007) and Topping (2005). In Eq. (1), the corresponding Dc can be obtained from a given Sc, and all particles with diameters larger than Dc can be activated. So the NCCN can be calculated by integrating the PNSD from Dc to the largest particle size measured: CCN(Dc) = Z Dc 550n(logDp)d logDp, (3) (3) where n(logDp) is the particle number that corresponds to each particle size bin d logDp in the aerosol number size dis- tribution. 2.3 Calculation of Nd The Nd depends on the Smax that can be formed in adiabatic ascending clouds. And this “cloud-relevant” supersaturation varies at different updraft velocity. A global scheme of cloud droplet parameterization has been established and developed for the calculation of the Nd and Smax (Nenes and Seinfeld, 2001, 2003; Fountoukis and Nenes, 2005). In this study, the Smax was calculated from an equation that expresses the wa- ter vapor balance in adiabatic ascending cloud (Nenes and Seinfeld, 2003): 2.1 Site and experiment A ﬁeld campaign was conducted from 25 May to 18 June 2017 at the Institute of Atmospheric Physics (IAP), Chinese Academy of Sciences (39.98◦N, 116.39◦E), for measurements of aerosol physical and chemical properties. The IAP is located between the North Third Ring Road and Fourth Ring Road in northern Beijing. It is a typical urban background site, mainly affected by trafﬁc and cooking emis- sions. Beijing is hot in summer, with high ambient relative humidity, which is conductive to the generation of atmo- spheric convection and reduces the high background aerosol condensation sink. The radiation in summer is stronger than other seasons, which promotes the generation of nucleated particles. In addition, local sources from trafﬁc and cooking emissions, which may contribute many CCN size-relevant particles, can be important at the site (Sun et al., 2015). The instruments during the campaign were deployed in a con- tainer at ground level (∼8 m on a meteorological tower). Relevant studies have been carried out in clean regions, but fewer were conducted in polluted urban areas. Field studies have shown that NPF events can occur frequently in polluted urban sites, although the high concentration of background particles is not conducive to the generation of new particles (Wu et al., 2011; Peng et al., 2014; Zimmerman et al., 2020), and the formation and growth rate of new particles may be https://doi.org/10.5194/acp-21-14293-2021 Atmos. Chem. Phys., 21, 14293–14308, 2021 S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration 2.2 Calculation of NCCN According to the hygroscopic growth process of particles de- scribed by Köhler theory (Köhler, 1936), the particles with a dry particle diameter (Dp) larger than the critical dry par- ticle diameter (Dc) can be activated to form a cloud droplet. In this study, the κ-Köhler theory (Petters and Kreidenweis, 2007), which simply describes the approximate relationship between the Dc with the critical supersaturation (Sc), is ap- plied as follows, when κ > 0.1: ds dt = αV −γ dw dt , (4) (4) where α and γ are two coefﬁcients that can be calculated by meteorological constants, the product of α and V expresses the increase of supersaturation due to the adiabatic cooling of the parcel, and dw dt denotes the water condensation rate dur- ing the aerosol activation and subsequent growth processes, shown in detail in Eq. (4). And ds dt expresses the growth rate of supersaturation; when it is equal to 0, the supersaturation reaches the maximum value. K = 4A3 27D3cln2Sc , A = 4σwMw RT ρw , (1) (1) where Mw is the molecular weight of water (Mw = 0.018015 kg mol−1), ρw is the density of water (ρw = 997.1 kg m−3), T is the parcel temperature (T = 298.15 K), where σw is the droplet surface tension at the point of acti- vation (σw = 0.072 J m−2), and R is the universal gas con- stant (R = 8.315 J K−1 mol−1). κ is a hygroscopic parameter which depends on the chemical composition of the particle. In this study, based on the assumption that particles are in- ternally mixed and their chemical composition will not be impacted by changes in particle size, we derived the κ with a simple mixing rule on the basis of chemical volume frac- tions (Petters and Kreidenweis, 2007; Gunthe et al., 2009). We used ACSM data, combined with the positive matrix fac- torization (PMF) analysis data to calculate the organic and inorganic volume fraction, according to the following equa- tion: where Mw is the molecular weight of water (Mw = 0.018015 kg mol−1), ρw is the density of water (ρw = 997.1 kg m−3), T is the parcel temperature (T = 298.15 K), where σw is the droplet surface tension at the point of acti- vation (σw = 0.072 J m−2), and R is the universal gas con- stant (R = 8.315 J K−1 mol−1). S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration This case on 11 June was not a unique case, and similar patterns are also shown on other NPF days during the campaign (Figs. S3–S8). S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration Owing to the fact that the direct measurement of cloud-scale updraft ve- locity in the atmosphere is almost impossible, the prescribed updraft velocity used in this study is from previous studies. Generally, the updraft velocities are reported to be very small (Martin and Johnson, 1994) and range from 0.1 to 1.0 m s−1 in stratocumulus and cumulus clouds in the remote or marine boundary layer (Meskhidze et al., 2005; Morales and Nenes, 2010). The vertical updraft velocities were derived to vary from 0.3 to 3 m s−1 (Zheng et al., 2015), which is typical of cumulus and convective clouds in summer in north China; thus they were selected and applied in this study. RS = CCNS CCNS,tstart , (8) RS = CCNS CCNS,tstart , (8) where S represents the supersaturation. Before the new parti- cles reach a large enough size to impact NCCN, the variations of RS should remain constant for different supersaturations if the concentrations of the background or preexisting aerosols change insigniﬁcantly. And at tstart,CCN when NPF begins to impact the NCCN, an apparent increase in RS is observed, taking the observation on 11 June as an example (Fig. 1a). Also, due to the heterogenous composition and distinct CCN activity of the newly formed particles (Duan et al., 2018; Ren et al., 2018; Zhang et al., 2019; Tao et al., 2021), a pa- rameter, Rd, which was calculated with the relative standard deviation of the RS of different supersaturations at a given time, is applied to ﬁx the tstart,CCN and tend,CCN. Then the tstart,CCN and tend,CCN correspond to the moments when the Rd starts to increase and goes back to nearly zero (Fig. 1b) between the tstart and tend. The same method is used to de- termine the time that NPF begins and ends its impact on the Nd, denoted as tstart,Nd and tend,Nd respectively (Fig. 1d, e). More details about the method can be found in Kalkavouras et al. (2019). As shown in Fig. 1, it is clear that both the NCCN and Nd exhibit a large increase in the NPF-impacted time zone between tstart,CCN and tend,CCN (Fig. 1c) and be- tween tstart,Nd and tend,Nd (Fig. 1f). The average time lag between tstart and tstart,Nd was about 3–5 h, which is short- ened by 50 % compared to that reported by Kalkavouras et al. (2019). S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration 14296 Table 1. Densities of different chemical species and their κ measured by the laboratory. Species NH4NO3 (NH4)2SO4 NH4HSO4 H2SO4 POA∗ SOA∗ BC ρ (kg m−3) 1720 1769 1780 1830 1000 1400 1700 κ 0.58 0.48 0.56 1.19 0 0.09 0 POA∗refers to primary organic aerosol, and SOA∗refers to secondary organic aerosol. Species NH4NO3 (NH4)2SO4 NH4HSO4 H2SO4 POA∗ SOA∗ ρ (kg m−3) 1720 1769 1780 1830 1000 1400 κ 0.58 0.48 0.56 1.19 0 0.09 POA∗refers to primary organic aerosol, and SOA∗refers to secondary organic aerosol. POA∗refers to primary organic aerosol, and SOA∗refers to secondary organic aerosol. F S(Sx) can be calculated by the integration of nS(S′) from the lower limit 0 to upper limit Sx. If the Smax is known, the activated Nd can be calculated from Eq. (7), as F S(Sx) can be calculated by the integration of nS(S′) from the lower limit 0 to upper limit Sx. If the Smax is known, the activated Nd can be calculated from Eq. (7), as generated. To determine the time that NPF begins and ends its impact on the NCCN, denoted as tstart,CCN and tend,CCN re- spectively, the time series of NCCN was ﬁrstly divided by the NCCN at tstart at each prescribed supersaturation, to derive the normalized time series of NCCN, denoted as RS. The equation is written as follows: Nd = F S(Smax). Nd = F S(Smax). (7) (7) In this study, we used the PNSD, chemical components, and empirical values of cloud updraft velocity to determine the Smax and Nd during NPF days in urban Beijing. Owing to the fact that the direct measurement of cloud-scale updraft ve- locity in the atmosphere is almost impossible, the prescribed updraft velocity used in this study is from previous studies. Generally, the updraft velocities are reported to be very small (Martin and Johnson, 1994) and range from 0.1 to 1.0 m s−1 in stratocumulus and cumulus clouds in the remote or marine boundary layer (Meskhidze et al., 2005; Morales and Nenes, 2010). The vertical updraft velocities were derived to vary from 0.3 to 3 m s−1 (Zheng et al., 2015), which is typical of cumulus and convective clouds in summer in north China; thus they were selected and applied in this study. In this study, we used the PNSD, chemical components, and empirical values of cloud updraft velocity to determine the Smax and Nd during NPF days in urban Beijing. 2.2 Calculation of NCCN κ is a hygroscopic parameter which depends on the chemical composition of the particle. In this study, based on the assumption that particles are in- ternally mixed and their chemical composition will not be impacted by changes in particle size, we derived the κ with a simple mixing rule on the basis of chemical volume frac- tions (Petters and Kreidenweis, 2007; Gunthe et al., 2009). We used ACSM data, combined with the positive matrix fac- torization (PMF) analysis data to calculate the organic and inorganic volume fraction, according to the following equa- tion: dw dt = π 2 ρw S Z 0 D2 p dDp dt nS(S′)ds′, (5) (5) where ρw is the density of water. ns (S′)ds′ is the number concentration of particles activated between S′ and S′ + ds. Nenes et al. (2001) used a sectional representation of the CCN spectrum (i.e., particle number supersaturation distri- bution nS(S′)) and total number of particles with Sc smaller than S, F S(S), which is given by F S (Sx) = Sx Z 0 nS(S′)ds′, (6) (6) (6) where Sx is the supersaturation in the environment and nS(S′) represents the number concentration of particles ac- tivated between S′ and S′ + ds′ in the CCN spectrum. The where Sx is the supersaturation in the environment and nS(S′) represents the number concentration of particles ac- tivated between S′ and S′ + ds′ in the CCN spectrum. The κchem = X i εiκi, (2) κchem = X i εiκi, (2) https://doi.org/10.5194/acp-21-14293-2021 Atmos. Chem. Phys., 21, 14293–14308, 2021 Atmos. Chem. Phys., 21, 14293–14308, 2021 Atmos. Chem. Phys., 21, 14293–14308, 2021 14296 S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration Table 1. Densities of different chemical species and their κ measured by the laboratory. Species NH4NO3 (NH4)2SO4 NH4HSO4 H2SO4 POA∗ SOA∗ BC ρ (kg m−3) 1720 1769 1780 1830 1000 1400 1700 κ 0.58 0.48 0.56 1.19 0 0.09 0 POA∗refers to primary organic aerosol, and SOA∗refers to secondary organic aerosol. S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration ng et al.: The contribution of new particle formation to cloud droplet number concentration 2.4 Method for calculating the contribution of NPF to NCCN and Nd The increment of NCCN or Nd by the NPF (1NCCN or 1Nd) is usually quantiﬁed by comparing the NCCN or Nd prior to and after the NPF event (Peng et al., 2014; Wu et al., 2015; Ma et al., 2016; Ren et al., 2018; Zhang et al., 2019; Fan et al., 2020). In this study, the NCCN or Nd prior to the NPF event was determined as a 2 h average of NCCN or Nd before the burst of newly formed nucleated particles. And the NCCN and Nd after the NPF event were calculated as the average of NCCN or Nd from the beginning to the end of the NPF impact on the NCCN or Nd. So it is critical to determine when a NPF event starts and ends or when NPF begins and ends its impact on the NCCN or Nd. Generally, the burst in the nucleation-mode particles sym- bolizes the beginning of an NPF event. Here, the moment when a half-hour concentration of the nucleation-mode par- ticles suddenly increases by orders of magnitude as high as ∼104 cm−3 during NPF cases was deﬁned as tstart. The end time of an NPF event, tend, is deﬁned by the moment when the half-hour concentration of nucleated particle is lower than that at tstart. Note that this method operates with an assumption of the unchanged background preexisting aerosols during the NPF events, without consideration of the impacts from local emis- sion sources and diurnal changes in the planetary boundary layer (PBL). As shown in Fig. 2b, the time series of NCN presents a baseline which indicates that the concentrations of the background aerosols on each of the 7 typical NPF days do not vary much; the impact from the variation of background aerosol particles thus should be insigniﬁcant. The impact of Since some time is needed for the newly formed nucle- ated particles to grow to a sufﬁcient size to act as CCN, the NCCN would not be enhanced as soon as new particles are Atmos. Chem. Phys., 21, 14293–14308, 2021 https://doi.org/10.5194/acp-21-14293-2021 S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration 14297 S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration 14297 Figure 1. 2.4 Method for calculating the contribution of NPF to NCCN and Nd The diurnal evolution of (a) the RS of NCCN at different supersaturations, (b) the relative dispersion of RS and Rd, for NCCN at different supersaturations, (c) the calculated NCCN under different supersaturations, (d) the RS of Nd under updraft velocities from 0.3 to 3.0 m s−1, (e) the Rd for Nd and (f) the calculated Nd at updraft velocities of 0.3 and 2.1 m s−1 on 11 June 2017. Figure 1. The diurnal evolution of (a) the RS of NCCN at different supersaturations, (b) the relative dispersion of RS and Rd, for NCCN at different supersaturations, (c) the calculated NCCN under different supersaturations, (d) the RS of Nd under updraft velocities from 0.3 to 3.0 m s−1, (e) the Rd for Nd and (f) the calculated Nd at updraft velocities of 0.3 and 2.1 m s−1 on 11 June 2017. ure 2b, c and d present the time series of NCN, NCCN and Nd. The ﬁgure shows that the NPF event drives the variation of NCCN and Nd, conﬁrming that the occurrence of NPF events is an important source of CCN. The variation trend of NCN is more correlated with that of NCCN than Nd (see also Fig. 4, Table S5). This is because the NCCN was calculated based on a constant S rather than referring to the availability of wa- ter vapor, while the calculation of Nd is based on the Smax that can be reached in the real atmosphere at a given updraft velocity. In the cloud, the change in the quantity of cloud particles can be directly reﬂected by the change in Smax. As shown in Fig. 2e, the average Smax for the two vertical up- draft velocity of V = 0.3 and V = 2.1 m s−1 was calculated to be under 0.2 % and 0.4 %, varying largely with the vari- ation of NCN due to the effect of water vapor competition, which will be discussed in Sect. 3.3. PBL is expected to be small when the growth of the newly formed particles spans only a few hours. However, when the growth continues over a longer period to evening or at night, which may coincide with the period during which the PBL height changes from high to low (Kerminen et al., 2012; Alt- städter, et al., 2015; Z. 3 Results and discussion 3.1 Time series of observed NPF events and calculated NCCN and Nd 2.4 Method for calculating the contribution of NPF to NCCN and Nd Li et al., 2017), it will result in a larger NCCN and Nd, leading to an overestimation of the contribu- tion of NPF to NCCN and Nd. A quantitative evaluation of such an impact is difﬁcult due to the contemporary PBL data not being available. Therefore, here we only investigate the impact of local emissions on the evaluation of the effect of NPF on Nd based on a case study. 3.2 Quantitative evaluation of the NPF impact on NCCN and Nd Time series of (a) particle number size distribution (PNSD) (the selected seven typical NPF events are marked by pink shading), (b) the total particle number concentration (Ntotal), (c) CCN number concentration (NCCN), (d) cloud droplet number concentration (Nd) and (e) the maximum supersaturation (Smax) from 25 May to 15 June 2017. Figure 2. Time series of (a) particle number size distribution (PNSD) (the selected seven typical NPF events are marked by pink shading), (b) the total particle number concentration (Ntotal), (c) CCN number concentration (NCCN), (d) cloud droplet number concentration (Nd) and (e) the maximum supersaturation (Smax) from 25 May to 15 June 2017. the relative contribution of NPF to NCCN is more signiﬁcant in remote clean regions. relatively independent of the variation of V . This is primarily due to the water vapor competition effect under very high CN number concentrations when calculating the Nd. Under high NCN, the water vapor competition effect will lead to lower Smax, which is smaller than the constant S for calculating NCCN. Roughly, the Nd at V of 0.3–3 m s−1 corresponds to the NCCN at S of 0.1 %–0.5 %, within which the percentages of 1NCCN and the contributions of the NPF to NCCN do not change much either. The effect of water vapor competition will be further examined in the following section. The estimated results of Nd for selected vertical updraft velocities are shown in Fig. 3d–f. Generally, the average Smax was calculated to be under 0.4 % and 0.2 % for V = 2.1 and V = 0.3 m s−1 respectively (Fig. 2e), corresponding to criti- cal particle sizes (Dc) of ∼70 and ∼110 nm. This means that most activated drops are from accumulation-mode particles and larger particles in Aitken mode. The large contribution of the Aitken-mode particles leads to a large number of cloud droplets in urban Beijing, especially for high updraft veloc- ity. Basically, the average 1Nd values (increased by NPF) are 433, 854, 1117, 1281 and 1523 cm−3 at updraft velocities of 0.3, 0.9, 1.5, 2.1 and 3 m s−1 respectively (Fig. 3d), which is a much larger magnitude compared with that in the clean areas (Morales Betancourt et al., 2014; Sullivan et al., 2016; Kalka- vouras et al., 2019). 3.2 Quantitative evaluation of the NPF impact on NCCN and Nd This is equal to the Nd being enhanced by 32 %, 37 %, 38 %, 38 % and 40 % at updraft velocities of 0.3, 0.9, 1.5, 2.1 and 3 m s−1 respectively (Fig. 3e), suggest- ing that the higher cloud updraft velocity not only generates more cloud droplets, but also induces larger enhancements in Nd. We also show that the NPF contributes about 30 % to the total Nd during the studied period in urban Beijing (Fig. 3f). And the rest (about 70 %) of cloud droplets are from the other sources or preexisting particles. With the increase of the S, the percentages of NPF-initiated NCCN and the contributions of the NPF to NCCN increased more signiﬁcantly than for Nd with the increase of V . In other words, the percentages of NPF-initiated Nd and the contributions of the NPF to Nd are 3.2 Quantitative evaluation of the NPF impact on NCCN and Nd During the observed periods from 25 May to 15 June 2017, the NPF events occurred on most days (∼13 d) (Fig. 2a). Ac- cording to Dal et al. (2005) and Wu et al. (2015), a typical NPF event includes the sudden appearance and continuous growth of particles smaller than 25 nm, and a “banana” shape can be seen on the particle number size spectrum. While non-NPF events may also have sudden increases of ﬁne par- ticles at a short timescale (e.g., local sources from vehicle or cooking emissions), they do not show a banana shape. Therefore, those cases with a typical banana shape (seven NPF events in total), which presents a complete NPF evolu- tion process from nucleation to subsequent growth (not in- terrupted by meteorological conditions either), are selected for further study (marked by pink shading in Fig. 2). Fig- Based on the method in Sect. 2.4, the contribution of the NPF to NCCN and Nd is calculated and is shown in Fig. 3. The re- sults show that the NCCN is increased on average by 32.0 %, 43.0 %, 53.0 % and 65.0 % at S of 0.2 %, 0.4 %, 0.6 % and 0.8 % respectively during NPF events (Fig. 3b, c, Table S3), amounting to about 24 %–37 % of environment CCN at the cloud-relevant supersaturation being directly originated from NPF during the studied period in urban Beijing. And the rest (about 63 %–76 %) of CCN are from other sources or preex- isting particles, which is a much larger proportion than that derived in remote Finokalia, Crete, Greece, by Kalkavouras et al (2019). In other words, due to the higher background concentration of aerosol particles in the polluted urban area, https://doi.org/10.5194/acp-21-14293-2021 Atmos. Chem. Phys., 21, 14293–14308, 2021 S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentra 14298 14298 S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration Figure 2. Time series of (a) particle number size distribution (PNSD) (the selected seven typical NPF events are marked by pink shading), (b) the total particle number concentration (Ntotal), (c) CCN number concentration (NCCN), (d) cloud droplet number concentration (Nd) and (e) the maximum supersaturation (Smax) from 25 May to 15 June 2017. J g p p Figure 2. https://doi.org/10.5194/acp-21-14293-2021 3.4 The variations of CCN and cloud droplet on typical clean and polluted NPF days: a case study Generally, lower PM2.5 means a low background condensa- tion sink (CS), which is conducive to the condensation and coagulation of nucleation particles (Wu et al., 2011; Yue et al., 2011; Wiedensohler et al., 2012). Different from the re- mote clean area, some of the NPF events in urban Beijing during the campaign occurred with background pollution (with daily mass concentrations of PM2.5 of ∼40 µg m−3) or are impacted by local primary emissions. This kind of NPF event has different characteristics from that in clean con- ditions, as the sudden increase of nucleation particles less than 25 nm is often accompanied by an increase of large particles at the beginning of NPF. Here, they are named as polluted and clean NPF events respectively. Two days, on 27 May and 11 June, representing the typical polluted and clean NPF events respectively, are selected for contrasting the effect of the two kinds of NPF on CCN and CDNC. As shown in Fig. 7, there is a higher preexisting background of accumulation-mode particles throughout the day on the pol- luted NPF day of 27 May than on the clean NPF day of 11 June. On the clean NPF days, many more nucleation- and Aitken-mode particles, with NCN enhancement 2-fold higher than that on polluted days (Fig. 7a), were generated, and NPF g To evaluate the effect of water vapor competition on Nd, by taking the case on 11 June as an example, we compare the Nd calculated from the varied Smax at different updraft velocities with the Nd at constant S (Fig. 5). The results from other NPF cases are also summarized in Tables S6 and S7. Obviously, after the tstart, the Smax starts to decrease and was negatively correlated with Nd for both the updraft veloc- ities, reﬂecting the enhanced effect of competition for wa- ter vapor from the growing number of droplets (Fig. 5a and b). It is shown that Smax was decreased by 14.5 ± 3.5 %, 13.3 ± 4.0 %, 13.4 ± 4.2 %, 12.0 ± 4.1 % and 11.7 ± 3.9 % for V = 0.3, 0.9, 1.5, 2.1 and 3 m s−1 respectively (Fig. 5c, d). S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration 14299 Figure 3. Box diagram of the increment of (a) CCN number concentration NCCN (1NCCN), (b) enhanced percentage of NCCN and (c) the contribution to total NCCN by NPF for different supersaturations (0.2 %–0.8 %) and (d) cloud droplet number concentration Nd (1Nd), (e) enhanced percentage of Nd and (f) the contribution to total Nd by NPF under different updraft velocities (0.3–3.0 m s−1). Figure 3. Box diagram of the increment of (a) CCN number concentration NCCN (1NCCN), (b) enhanced percentage of NCCN and (c) the contribution to total NCCN by NPF for different supersaturations (0.2 %–0.8 %) and (d) cloud droplet number concentration Nd (1Nd), (e) enhanced percentage of Nd and (f) the contribution to total Nd by NPF under different updraft velocities (0.3–3.0 m s−1). to a certain number. This is strongly suggested by the dif- ference between the calculated NCCN using the constant S and the Nd using the variable Smax in the air parcels because in the actual environment, it is often not possible to achieve sufﬁcient supersaturation compared to the prescribed levels that are preset by the instrument. For example, the average Smax is lower than 0.5 % at the maximum cloud updraft ve- locity of 3 m s−1 according to the calculation results in this study. Therefore, although NPF events may strongly increase NCCN, the formed Nd is eventually limited by water vapor competition, which determines the Smax that varies in the cloud. The Smax is related to the cloud formation dynamics and the aerosol levels in the region. to a certain number. This is strongly suggested by the dif- ference between the calculated NCCN using the constant S and the Nd using the variable Smax in the air parcels because in the actual environment, it is often not possible to achieve sufﬁcient supersaturation compared to the prescribed levels that are preset by the instrument. For example, the average Smax is lower than 0.5 % at the maximum cloud updraft ve- locity of 3 m s−1 according to the calculation results in this study. Therefore, although NPF events may strongly increase NCCN, the formed Nd is eventually limited by water vapor competition, which determines the Smax that varies in the cloud. The Smax is related to the cloud formation dynamics and the aerosol levels in the region. S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration ilar to those reported by Kalkavouras et al. (2017), who showed that these competition effects suppress Nd by 20 % for V = 0.3 and 12.3 % for V = 0.6 m s−1. In addition, the declined percentages with an increase in updraft velocity suggest that the effect becomes smaller at larger V , achiev- ing greater Smax in the environment. Essentially, water vapor competition led to the reduction in Nd by decreasing the re- quired Smax for the CN activation. 3.3 The effect of water vapor competition on evaluating Nd Figure 4 shows the scatter plots of correlations between NCN and NCCN at supersaturations of 0.6 % and 0.8 % and the correlations between NCN and Nd under updraft vertical ve- locities of 2.1 and 3.0 m s−1. The NCCN and NCN were ob- viously linearly related, but the correlation between Nd and NCN was nonlinear. When shown as average values with er- ror bars, the Nd increases linearly as NCN increases when the NCN is below 15 000, then the Nd begins to decrease with the further increase of NCN. This has been presented in pre- vious studies (Nenes et al., 2001; Ramanathan et al., 2001; Sullivan et al., 2016) and is believed to be caused by the water vapor competition of the aerosol particles. Although the larger updraft velocities can achieve greater supersatu- ration in adiabatic ascending clouds, and more particles can be activated into cloud droplets, the water vapor competition still occurred when background aerosol particles increased https://doi.org/10.5194/acp-21-14293-2021 Atmos. Chem. Phys., 21, 14293–14308, 2021 S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration 14300 Figure 4. Scatter plots of correlation between total number concentration (NCN) and CCN number concentration (NCCN) at supersaturation of (a) 0.6 % and (c) 0.8 % respectively. Scatter plot of correlation between NCN and cloud droplet number concentration (Nd) at updraft vertical velocities of (b) 2.1 m s−1 and (d) 3.0 m s−1 respectively. Figure 4. Scatter plots of correlation between total number concentration (NCN) and CCN number concentration (NCCN) at supersaturation of (a) 0.6 % and (c) 0.8 % respectively. Scatter plot of correlation between NCN and cloud droplet number concentration (Nd) at updraft vertical velocities of (b) 2.1 m s−1 and (d) 3.0 m s−1 respectively. Figure 5. (a) The diurnal changes of the calculated maximum supersaturation (Smax) (referring to the constant supersaturation S) at V = 0.3 and 2.1 m s−1 on 11 June 2017. (b) Comparison of the cloud droplet number concentration (Nd) under the constant S and Smax on 11 June 2017. The suppression percentage of (c) Smax and (d) Nd due to the competition of water vapor. Figure 5. (a) The diurnal changes of the calculated maximum supersaturation (Smax) (referring to the constant supersaturation S) at V = 0.3 and 2.1 m s−1 on 11 June 2017. (b) Comparison of the cloud droplet number concentration (Nd) under the constant S and Smax on 11 June 2017. The suppression percentage of (c) Smax and (d) Nd due to the competition of water vapor. events developed more strongly in the initial stage. The be- ginning of NPF events (tstart) in the polluted case (11:00) was about 2 h later than that in the clean case (∼09:00). clean and polluted days respectively (Fig. 7b). For Nd, the percentage increase on clean days was 22 % and 37 % and on polluted days 34 % and 26 % under updraft velocities of 0.3 and 2.1 m s−1. The percentage increases in Nd between clean and polluted days are comparable. This result further illustrates the effect of water vapor competition on Nd un- der high NCN in the polluted atmosphere. This suggests that it is critical to fully consider the background meteorological conditions (e.g., using dynamic water vapor under different updraft velocities) to simulate the Nd when evaluating the effect of NPF on clouds and the associated climate effects. For both cases, the NCCN is increased with the evolution of the NPF events (Fig. 6a, b, e, f). 3.4 The variations of CCN and cloud droplet on typical clean and polluted NPF days: a case study Therefore, in comparison to the Nd calculated from the constant S, the Nd calculated from the variable Smax is greatly reduced at both the updraft velocities of 0.3 m s−1 and 2.1 m s−1, suggesting a signiﬁcant suppression of cloud droplet formation. Quantitatively, the Nd is reduced by 19.0 ± 4.5 %, 15.7 ± 4.7 %, 14.8 ± 5.6 %, 12.3 ± 4.9 % and 11.8 ± 5.0 % at updraft velocity of 0.3, 0.9, 1.5, 2.1 and 3 m s−1 respectively on the NPF days. Our results are sim- https://doi.org/10.5194/acp-21-14293-2021 Atmos. Chem. Phys., 21, 14293–14308, 2021 S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration But the magnitude of the enhancements in the two cases is quite different. The NCCN during NPF events on polluted days was generally twice than that of clean days (Fig. 6e and f) because there were a large number of preexisting CCN-sized aerosol par- ticles on polluted NPF days. As a result, a larger increment of NCCN is derived on clean NPF days, showing 37 %–80 % and 25 %–41 % percentage increases of NCCN from NPF on https://doi.org/10.5194/acp-21-14293-2021 Atmos. Chem. Phys., 21, 14293–14308, 2021 S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration 14301 Figure 6. Comparison of (a, b) the particle number size distribution PNSD, (c, d) aerosol particle number concentration NCN, (e, f) CCN number concentration NCCN and (g, h) cloud droplet number concentration Nd between a clean and a polluted NPF event. The clean NPF day is with a clean background (PM2.5 = 14 (µg/m3), and the polluted NPF day is with PM2.5 of 73 µg/m3. Figure 6. Comparison of (a, b) the particle number size distribution PNSD, (c, d) aerosol particle number concentration NCN, (e, f) CCN number concentration NCCN and (g, h) cloud droplet number concentration Nd between a clean and a polluted NPF event. The clean NPF day is with a clean background (PM2.5 = 14 (µg/m3), and the polluted NPF day is with PM2.5 of 73 µg/m3. Figure 7. Comparison of the increments of (a) total particle number concentration (NCN) and (b) CCN number concentration (NCCN) and cloud droplet number concentration (Nd) between the two different typical NPF events. Figure 7. Comparison of the increments of (a) total particle number concentration (NCN) and (b) CCN number concentration (NCCN) and cloud droplet number concentration (Nd) between the two different typical NPF events. Atmos. Chem. Phys., 21, 14293–14308, 2021 https://doi.org/10.5194/acp-21-14293-2021 Atmos. Chem. Phys., 21, 14293–14308, 2021 3.5 The impact of primary emissions during evening rush hour on the calculation of the contribution of NPF to NCCN and Nd: a case study also can be found in the supporting information of Liu et al. (2021). To evaluate the impact of the primary emissions, it is crit- ical to separate the particle modes representing the primary aerosols from the observed PNSD. According to the observed characteristics of PNSD, the newly formed particles con- tinue to grow and are dominated by Aitken mode for sev- eral hours after the NPF occurred (Fig. 8a). The size mode of the newly formed particles during the rush hour is esti- mated by applying a growth rate of 3.2 ± 0.5 nm h−1, which is calculated by the variation of median particle size dur- ing 12:00–18:00. The calculation results show that the NPF- tracked particles can grow to ∼50–60 nm during the rush hour period. The primary particles from vehicles or cooking generally have a smaller size (∼30 nm) than the NPF-tracked mode and accumulation mode (∼100–120 nm) (Brines et al., 2015; Dall’Osto, et al., 2011; Harrison, et al., 2011), so we applied three modes to ﬁt the PNSD from the beginning of the evening rush hour to the end, assuming a normal distribu- tion. Note that the size mode for background aerosols almost coincides with the accumulation mode of primary emitted particles during the period. Since the mode and concentra- tion of background aerosols do not change much before and after the occurrence of new particles (Fig. 8a, b), the impact of background aerosol is thus deducted from the ﬁtting ac- cumulation mode. The ﬁtted result shows a major peak in the Aitken mode at ∼50 nm that is related to the NPF event and two minor peaks in Aitken (∼30 nm) and accumulation (∼100–120 nm) mode (Fig. 8e, f) that are associated with the primary vehicle or cooking emissions. Figure 8g and h show the separated PNSD of the NPF-related and primary aerosols respectively. Then the increments of NCCN and Nd from NPF are obtained from the PNSD of NPF mode, and the increments of NCCN and Nd from primary emissions are During the campaign, very high number concentrations of ﬁne particles were observed during evening rush hour (as shown in Fig. 2a) when primary emissions related to auto- mobile exhaust or cooking activities near the site may im- pact the PNSD. Those particles from primary emissions can serve as CCN and thereby impact the evaluation of the con- tribution of NPF to Nd. https://doi.org/10.5194/acp-21-14293-2021 https://doi.org/10.5194/acp-21-14293-2021 14302 S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration Table 2. Quantitative evaluation of the contribution of primary emissions to Nd and NCCN. V or S Dc Nd_NPF or 1NCCN_NPF Na d_PE or 1Na CCN_PF Nd_total or NCCN_total (m s−1 or %) (nm) (cm−3) (%) (cm−3) (%) (cm−3) Evaluation of the contribution of primary emissions to Nd 0.3 140 200 84.4 % 37 15.6 % 237 0.9 107 543 86.6 % 84 13.4 % 627 1.5 93 676 87.5 % 97 12.5 % 773 2.1 84 750 83.1 % 153 16.9 % 903 3.0 75 942 77.1 % 279 22.9 % 1221 Evaluation of the contribution of primary emissions to NCCN 0.2 % 109 654 92.0 % 57 8.0 % 711 0.4 % 69 1356 87.2 % 199 12.8 % 1555 0.6 % 52 1680 87.1 % 249 12.9 % 1929 0.8 % 43 1801 85.0 % 318 15.0 % 2119 a PE, primary emission. ng et al.: The contribution of new particle formation to cloud droplet number concentration 14302 Table 2. Quantitative evaluation of the contribution of primary emissions to Nd and NCCN. 3.5 The impact of primary emissions during evening rush hour on the calculation of the contribution of NPF to NCCN and Nd: a case study Therefore, taking the day of 11 June as an example, such an effect from primary emissions dur- ing evening rush hour is investigated (Fig. 8). On the day, 1 h after the burst of newly formed particles at ∼12:00, the Nd began to rise rapidly, and the increase of Nd continued until 21:30 (Fig. 1f). At ∼18:00, the primary emissions also begin to impact the NCCN and Nd. Note that a sudden de- crease and dilution in the PNSD are due to a precipitation event at ∼21:30. From 18:00–21:30, the CCN and cloud droplets were from both the NPF source and the primary emissions. The impact of primary emissions is also indicated by the variations of particles composition during 18:00– 22:00, when both the primary organic aerosols (POAs) and BC show a rapid increase in the mass concentration and fraction (Fig. 8c and d). Here, a positive matrix factoriza- tion (PMF) analysis was performed to separate the primary and secondary organic aerosol factors quantitatively for the purpose of source apportionment based on ﬁeld measure- ment by an Aerodyne high-resolution time-of-ﬂight aerosol mass spectrometer (HR-ToF-AMS) (Xu et al., 2017; Zhang et al., 2011). The PMF algorithm in the robust mode (Paatero and Tapper, 1994) was applied to the high-resolution mass spectra to resolve distinct organic aerosol factors represent- ing primary and secondary sources and processes. More de- tails about operation of the HR-ToF-AMS and PMF analysis https://doi.org/10.5194/acp-21-14293-2021 Atmos. Chem. Phys., 21, 14293–14308, 2021 S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration The contribution of new particle formation to cloud droplet number concentration 14303 g p p re 8. Diurnal variations of the (a) aerosol size distribution, (b) particle number concentrations for different size modes, (c) mass con- rations of aerosol chemical composition and (d) mass fraction of aerosol chemical components, (e, f) ﬁtted three modes of the particle ber concentration PNSD at 18:00 and 21:30, and (g) diurnal variations of the separated NPF-related PNSD and (h) the PNSD of primary sols. Figure 8. Atmos. Chem. Phys., 21, 14293–14308, 2021 4 Conclusions In this study, we quantiﬁed the contribution of NPF to Nd at typical updraft velocities in clouds using ﬁeld measurements of aerosol number size distributions and chemical composi- tion in urban Beijing. We show that NPF drives the variations of NCCN and Nd. About 32 %–65 % NCCN are increased by NPF events for supersaturation of 0.2 %–0.8 % in the pol- luted atmosphere. And the Nd is increased about 32 %–40 % by NPF at V = 0.3–3 m s−1 accordingly. A signiﬁcant re- duction in Nd is observed due to water vapor competition with consideration of actual environmental updraft veloc- ity, with decrease rates of 11.8 % ± 5.0 % at V = 3 m s−1 and 19.0 % ± 4.5 % at V = 0.3 m s−1 in comparison with a constant supersaturation. The effect of water vapor compe- tition becomes smaller at larger V , at which a greater Smax can be achieved. Essentially, water vapor competition led to the reduction in Nd by decreasing the environmental Smax for the activation of aerosol particles. It is shown that Smax was decreased by 14.5 ± 3.5 % to 11.7 ± 3.9 % for V = 0.3– 3 m s−1. Our results suggest signiﬁcant suppression of cloud droplet formation due to the water vapor competition, partic- ularly at extremely high aerosol particle number concentra- tions. As a result, although a larger enhancement of CCN- sized particles by NPF event is derived on clean NPF days when there are few preexisting background aerosol particles, there is no large discrepancy in the enhancement of Nd by NPF between the clean and polluted NPF days. Finally, we show a considerable impact of the primary sources when evaluating the contribution of NPF to cloud droplet number using a case study. Our study highlights the importance of full consideration of both the environmental meteorological conditions and multiple sources (i.e., secondary and primary) to evaluate the effect of NPF on clouds and the associated climate effects. For example, Merikanto et al. (2010) used model to simulate the variation of CDNC from the year 1850 to 2000 and showed that NPF made a nearly equal contribu- tion (16 %–13.5 %) to global CDNC in all those years, lead- ing to about a 50 % enhancement in the year from 1850 to 2000 in cloud albedo. S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration parameterize the impact of NPF events on cloud, precipita- tion, and radiative forcing in models. cated by the sharply raised particle number concentrations during the rush hour (Fig. 8b). Data availability. All data used in the study are available at https: //data.mendeley.com/datasets/hkkzbn4zv3/1 (last access: 4 January 2021, Zhang, 2021) or from the corresponding author upon request (fang.zhang@bnu.edu.cn). 4 Conclusions There are still large uncertainties in how to accurate quantitatively assess the response of these climate effects to NPF. This study is carried out in a polluted urban area and is beneﬁcial for the research on aerosol–cloud microphysical processes. It provides a new perspective for follow-up research in the urban atmosphere. Note that there are still limitations of our studies, as we only investigated several NPF cases within a short period due to the limited measurement data. The small sample size might cause bias in the results. Further studies based on more measurement data, i.e., with longer time periods and more observational sites, are warranted to verify and reﬁne our results, so as to Supplement. The supplement related to this article is available on- line at: https://doi.org/10.5194/acp-21-14293-2021-supplement. Author contributions. FZ and SJ conceived the conceptual devel- opment of the paper. LC, YS and JR contributed measurements. SJ directed and performed the experiments with LC, JL, JR, XY, ZL and FZ. SJ conducted the data analysis and wrote the draft of the paper. All authors commented on the paper. Competing interests. The authors declare that they have no conﬂict of interest. Disclaimer. Publisher’s note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional afﬁliations. Acknowledgements. This work was supported by the National Ba- sic Research Program of China (grant no. 2017YFC1501702) and the National Natural Science Foundation of China (grant nos. 41675141 and 41975174). Financial support. This research has been supported by the Na- tional Key Research and Development Program of China (grant no. 2017YFC1501702) and the National Natural Science Foundation of China (grant nos. 41675141 and 41975174). Review statement. This paper was edited by Veli-Matti Kerminen and reviewed by two anonymous referees. 3.5 The impact of primary emissions during evening rush hour on the calculation of the contribution of NPF to NCCN and Nd: a case study Diurnal variations of the (a) aerosol size distribution, (b) particle number concentrations for different size modes, (c) mass con- centrations of aerosol chemical composition and (d) mass fraction of aerosol chemical components, (e, f) ﬁtted three modes of the particle number concentration PNSD at 18:00 and 21:30, and (g) diurnal variations of the separated NPF-related PNSD and (h) the PNSD of primary aerosols. and 0.8 % respectively. Compared with Nd, the contribution percentage of primary emission to NCCN is smaller due to the total NCCN being much more than the total Nd. Our re- sults show a considerable impact of these primary sources when evaluating the contribution of NPF to cloud droplet number, highlighting the importance of considering the in- ﬂuence from multiple (i.e., secondary and primary) sources on clouds in the polluted atmosphere. Finally, it is worth noting that the dynamic changes of PBL would also impact the NCCN and Nd during the period, and the decrease in the height of PBL from the daytime to evening will result in an increase of NCCN or Nd. However, for this case, the impact from primary emissions is much more prominent, as indi- obtained by subtracting the increment of NCCN and Nd by NPF from the total increment of Nd. The calculated results are summarized in Table 2. For Nd, the average contribution of primary emission to Nd is 15.6 %, 13.4 %, 12.5 %, 16.9 % and 22.9 % cm−3 for updraft veloci- ties of 0.3, 0.9, 1.5, 2.1 and 3 m s−1 respectively. The propor- tion of contribution from NPF and primary emission to the Nd increment changes with the variation of V . The higher proportion of contribution from primary emission is obtained at higher V , which may be determined by the different char- acteristics between atmospheric particles emitted from the evening trafﬁc sources and generated from NPF events. For NCCN, the average contribution from primary emissions is 8.0 %, 12.8 %, 12.9 % and 15.0 % at S of 0.2 %, 0.4 %, 0.6 % Atmos. Chem. Phys., 21, 14293–14308, 2021 https://doi.org/10.5194/acp-21-14293-2021 14304 S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration Cai, M., Liang, B., Sun, Q., Zhou, S., Chen, X., Yuan, B., Shao, M., Tan, H., and Zhao, J.: Effects of continental emissions on cloud condensation nuclei (CCN) activity in the northern South China Sea during summertime 2018, Atmos. Chem. Phys., 20, 9153–9167, https://doi.org/10.5194/acp-20-9153-2020, 2020. Kalkavouras, P., Bossioli, E., Bezantakos, S., Bougiatioti, A., Kalivitis, N., Stavroulas, I., Kouvarakis, G., Protonotariou, A. P., Dandou, A., Biskos, G., Mihalopoulos, N., Nenes, A., and Tombrou, M.: New particle formation in the southern Aegean Sea during the Etesians: importance for CCN production and cloud droplet number, Atmos. Chem. Phys., 17, 175–192, https://doi.org/10.5194/acp-17-175-2017, 2017. Collins, D. R., Flagan, R. C., and Seinfeld, J. H.: Improved inversion of scanning DMA data, Aerosol Sci. Tech., 36, 1–9, 2002. Dal, M., Kulmala, M., Riipinen, I., Wagner, R., Hussein, T., Aalto, P. P., and Lehtinen, K. E. J.: Formation and growth of fresh at- mospheric aerosols: eight years of aerosol size distribution data from SMEAR II, Hyytiälä, Finland, Boreal Environ. Res., 10, 323–336, 2005. Kalkavouras, P., Bougiatioti, A., Kalivitis, N., Stavroulas, I., Tombrou, M., Nenes, A., and Mihalopoulos, N.: Regional new particle formation as modulators of cloud condensation nuclei and cloud droplet number in the eastern Mediterranean, At- mos. Chem. Phys., 19, 6185–6203, https://doi.org/10.5194/acp- 19-6185-2019, 2019. Dall’Osto, M., Thorpe, A., Beddows, D. C. S., Harrison, R. M., Barlow, J. F., Dunbar, T., Williams, P. I., and Coe, H.: Remark- able dynamics of nanoparticles in the urban atmosphere, At- mos. Chem. Phys., 11, 6623–6637, https://doi.org/10.5194/acp- 11-6623-2011, 2011. Kazil, J., Stier, P., Zhang, K., Quaas, J., Kinne, S., O’Donnell, D., Rast, S., Esch, M., Ferrachat, S., Lohmann, U., and Feichter, J.: Aerosol nucleation and its role for clouds and Earth’s radiative forcing in the aerosol-climate model ECHAM5-HAM, Atmos. Chem. Phys., 10, 10733–10752, https://doi.org/10.5194/acp-10- 10733-2010, 2010. Duan, J., Wang, Y., Xie, X., Li, M., Tao, J., Wu, Y., Cheng, T.: Inﬂuence of pollutants on activity of aerosol cloud condensa- tion nuclei (CCN) during pollution and post-rain periods in Guangzhou, southern China, Sci. Tot. Environ., 642, 1008–1019, https://doi.org/10.1016/j.scitotenv.2018.06.053, 2018. Kerminen, V.-M., Paramonov, M., Anttila, T., Riipinen, I., Foun- toukis, C., Korhonen, H., Asmi, E., Laakso, L., Lihavainen, H., Swietlicki, E., Svenningsson, B., Asmi, A., Pandis, S. N., Kul- mala, M., and Petäjä, T.: Cloud condensation nuclei production associated with atmospheric nucleation: a synthesis based on ex- isting literature and new results, Atmos. Chem. Phys., 12, 12037– 12059, https://doi.org/10.5194/acp-12-12037-2012, 2012. S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration 14305 Abdul-Razzak, H., Ghan, S., and Rivera-Carpio, C.: A parame- treization of aerosol activation: 1. Single aerosol types, J. Geo- phys. Res., 103, 6123–6131, 1998. Fountoukis, C. and Nenes, A.: Continued development of a cloud droplet formation parameterization for global climate models, J. Geophys. Res., 110, D11212, https://doi.org/10.1029/2004JD005591, 2005. Abdul-Razzak, H. and Ghan, S.: A parametreization of aerosol ac- tivation: 2. Multiple aerosol types, J. Geophys. Res., 105, 6837– 6844, 2000. Genz, C., Schrödner, R., Heinold, B., Henning, S., Baars, H., Spindler, G., and Tegen, I.: Estimation of cloud condensa- tion nuclei number concentrations and comparison to in situ and lidar observations during the HOPE experiments, Atmos. Chem. Phys., 20, 8787–8806, https://doi.org/10.5194/acp-20- 8787-2020, 2020. Asmi, E., Kivekäs, N., Kerminen, V.-M., Komppula, M., Hyvärinen, A.-P., Hatakka, J., Viisanen, Y., and Lihavainen, H.: Secondary new particle formation in Northern Finland Pallas site between the years 2000 and 2010, Atmos. Chem. Phys., 11, 12959–12972, https://doi.org/10.5194/acp-11-12959-2011, 2011. Ghan, S., Chung, C., and Penner, J.: A parameterization of cloud droplet nucleation part I: single aerosol type, Atmos. Res., 30, 198–221, https://doi.org/10.1016/0169-8095(93)90024-I, 1993. Boucher, O. and Lohmann, U.: The sulfate-CCN-cloud albedo ef- fect, Tellus, 47, 281–300, 1995. Bousiotis, D., Dall’Osto, M., Beddows, D. C. S., Pope, F. D., and Harrison, R. M.: Analysis of new particle formation (NPF) events at nearby rural, urban background and urban roadside sites, At- mos. Chem. Phys., 19, 5679–5694, https://doi.org/10.5194/acp- 19-5679-2019, 2019. Gunthe, S. S., King, S. M., Rose, D., Chen, Q., Roldin, P., Farmer, D. K., Jimenez, J. L., Artaxo, P., Andreae, M. O., Martin, S. T., and Pöschl, U.: Cloud condensation nuclei in pristine tropi- cal rainforest air of Amazonia: size-resolved measurements and modeling of atmospheric aerosol composition and CCN activity, Atmos. Chem. Phys., 9, 7551–7575, https://doi.org/10.5194/acp- 9-7551-2009, 2009. Brines, M., Dall’Osto, M., Beddows, D. C. S., Harrison, R. M., Gómez-Moreno, F., Núñez, L., Artíñano, B., Costabile, F., Gobbi, G. P., Salimi, F., Morawska, L., Sioutas, C., and Querol, X.: Trafﬁc and nucleation events as main sources of ul- traﬁne particles in high-insolation developed world cities, At- mos. Chem. Phys., 15, 5929–5945, https://doi.org/10.5194/acp- 15-5929-2015, 2015. Harrison, R. M., Beddows, D. and Dall’Osto, M.: PMF Analysis of Wide-Range Particle Size Spectra Collected on a Major Highway, Environ. Sci. Tech., 45, 5522, https://doi.org/10.1021/es201998m, 2011. Hudson, J. G. , Noble, S., and Tabor, S.: Cloud supersaturations from CCN spectra hoppel minima, J. Geophys. Res.-Atmos., 120, 3436–3452, https://doi.org/10.1002/2014JD022669, 2015. References Albrecht, B. A.: Aerosols, cloud microphysics, and fractional cloudiness, Science, 245, 1227–1230, 1989. Altstädter, B., Platis, A., Wehner, B., Scholtz, A., Wildmann, N., Hermann, M., Käthner, R., Baars, H., Bange, J., and Lampert, A.: ALADINA – an unmanned research aircraft for observing vertical and horizontal distributions of ultraﬁne particles within the atmospheric boundary layer, Atmos. Meas. Tech., 8, 1627– 1639, https://doi.org/10.5194/amt-8-1627-2015, 2015. Atmos. Chem. Phys., 21, 14293–14308, 2021 https://doi.org/10.5194/acp-21-14293-2021 https://doi.org/10.5194/acp-21-14293-2021 S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration 14306 growth: 1. Equilibrium size spectra and cloud condensation nuclei activity spectra, J. Geophys. Res., 104, 2175–2184, https://doi.org/10.1029/98JD02687, 1999. Meskhidze, N., Nenes, A., Conant, W., and Seinfeld, J.: Evaluation of a new cloud droplet activation parameterization with in situ data from CRYSTAL-FACE and CSTRIPE, J. Geophys. Res., 110, D16202, https://doi.org/10.1029/2004JD005703, 2005. Köhler, H.: The nucleus in and the growth of hygroscopic droplets, Transactions of the Faraday Society, 32, 1152–1161, https://doi.org/10.1039/tf9363201152, 1936. Morales, R. and Nenes, A.: Characteristic updrafts for com- puting distribution-averaged cloud droplet number and stra- tocumulus cloud properties, J. Geophys. Res., 115, D18220, https://doi.org/10.1029/2009JD013233, 2010. Kulmala, M. and Kerminen, V. M.: On the formation and growth of atmospheric nanoparticles, Atmos. Res., 90, 132–150, https://doi.org/10.1016/j.atmosres.2008.01.005, 2008. Morales Betancourt, R. and Nenes, A.: Understanding the contri- butions of aerosol properties and parameterization discrepan- cies to droplet number variability in a global climate model, At- mos. Chem. Phys., 14, 4809–4826, https://doi.org/10.5194/acp- 14-4809-2014, 2014. Kulmala, M., Petäjä, T., Nieminen, T., Sipilä, M., Manninen, H. E., Lehtipalo, K., Dal, Maso, M., Aalto, P., Junninen, H., Paasonen, P., Riipinen, I., Lehtinen, K. E. J., Laaksonen, A., and Kerminen, V.-M.: Measurement of the nucleation of atmospheric aerosol particles, Nat. Protocol., 7, 1651–1667, https://doi.org/10.1038/nprot.2012.091, 2012. Nenes, A. and Seinfeld, J. H.: Parameterization of cloud droplet formation in global climate models, J. Geophys. Res, 108, 4415, https://doi.org/10.1029/2002JD002911, 2003. Leaitch, W. R., Strapp, J. W., and Isaac, G. A.: Cloud droplet nucle- ation and cloud scavenging of aerosol sulphate in polluted atmo- spheres, Tellus B, 38, 328–344, https://doi.org/10.1111/j.1600- 0889.1986.tb00258.x ,1986 Nenes, A., Chan, S., Abdul-Razzak, H., Chuang, P., and Se- infeld, J. H.: Kinetic limitations on cloud droplet forma- tion and impact on cloud albedo, Tellus B, 53, 133–149, https://doi.org/10.3402/tellusb.v53i2.16569, 2001. Leino, K., Nieminen, T., Manninen, H. E., Petäjä, T., Kerminen, V. M., and Kulmala, M.: Intermediate ions as a strong indicator of new particleformation bursts in a boreal forest, Boreal Environ. Res., 21, 274–286, 2016. Ng, N. L., Herndon, S. C., Trimborn, A., Canagaratna, M. R., Croteau, P. L., and Onasch, T. B.: An aerosol chemical speciation monitor (ACSM) for routine monitoring of the composition and mass concentrations of ambient aerosol, Aerosol Sci. Tech., 45, 780-794, https://doi.org/10.1080/02786826.2011.560211, 2011. S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration Li, Y., Zhang, F., Li, Z., Sun, L., Wang, Z., Li, P., Sun, Y., Ren, J., Wang, Y., and Cribb, M.: Inﬂuences of aerosol physiochemical properties and new particle formation on CCN activity from ob- servation at a suburban site of China, Atmos. Res., 188, 80–89, https://doi.org/10.1016/j.atmosres.2017.01.009, 2017. Paatero, P. and Tapper, U.: Positive matrix factorization: A nonnegative factor model with optimal utilization of error- estimates of data values, J. Citation Rep., 5, 112–126, https://doi.org/10.1002/env.3170050203, 1994. Li, Z., Guo, J., Ding, A., Liao, H., Liu, J., Sun, Y., Wang, T., Xue, H., Zhang, H., and Zhu, B.: Aerosols and boundary-layer inter- actions and impact on air quality, Natl. Sci. Rev., 4, 810–833, https://doi.org/10.1093/nsr/nwx117, 2017. Peng, J. F., Hu, M., Wang, Z. B., Huang, X. F., Kumar, P., Wu, Z. J., Guo, S., Yue, D. L., Shang, D. J., Zheng, Z., and He, L. Y.: Submicron aerosols at thirteen diversiﬁed sites in China: size distribution, new particle formation and corresponding contri- bution to cloud condensation nuclei production, Atmos. Chem. Phys., 14, 10249–10265, https://doi.org/10.5194/acp-14-10249- 2014, 2014. Li, Z., Feng, N., Fan, J., Liu, Y., and Ding, Y.: Long-term impacts of aerosols on the vertical development of clouds and precipitation, Nat. Geosci., 4, 888–894, https://doi.org/10.1038/ngeo1313, 2011. Liu, C., Wang, T., Rosenfeld, D., Zhu, Y., Yue, Z., and Yu, X.: Anthropogenic effects on cloud condensation nuclei distribu- tion and rain initiation in East Asia, Geophys. Res. Lett., 47, e2019GL086184, https://doi.org/10.1029/2019GL086184, 2019. Peng,Y., Dong, Y., and Li, X. M.: Different Character- istics of New Particle Formation Events at Two Sub- urban Sites in Northern China, Atmosphere, 8, 258, https://doi.org/10.3390/atmos8120258, 2017. Liu, J., Zhang, F., Xu, W., Sun, Y., Chen, L., and Li, S.: Hygroscopicity of organic aerosols linked to forma- tion mechanisms. Geophys. Res. Lett., 48, e2020GL091683, https://doi.org/10.1029/2020GL091683, 2021. Petters, M. D. and Kreidenweis, S. M.: A single parameter representation of hygroscopic growth and cloud condensa- tion nucleus activity, Atmos. Chem. Phys., 7, 1961–1971, https://doi.org/10.5194/acp-7-1961-2007, 2007. Ramanathan, V., Crutzen, P. J., Kiehl, J. T., and Rosenfeld, D.: Aerosols, climate, and the hydrological cycle, Science, 294, 2119–2124, https://doi.org/10.1126/science.1064034, 2001. Ma, N., Zhao, C., Tao, J., Wu, Z., Kecorius, S., Wang, Z., Größ, J., Liu, H., Bian, Y., Kuang, Y., Teich, M., Spindler, G., Müller, K., van Pinxteren, D., Herrmann, H., Hu, M., and Wiedensohler, A.: Variation of CCN activity during new particle formation events in the North China Plain, Atmos. Chem. Phys., 16, 8593–8607, https://doi.org/10.5194/acp-16-8593-2016, 2016. S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration Fan, J., Leung, R., Rosenfeld, D., Chen, Q., Li, Z., Zhang, J., and Yan, H.: Microphysical effects determine macro- physical response for aerosol impacts on deep convec- tive clouds, P. Natl. Aacad. Sci. USA, 110, 4581–4590, https://doi.org/10.1073/pnas.1316830110, 2013. Fan, X., Liu, J., Zhang, F., Chen, L., Collins, D., Xu, W., Jin, X., Ren, J., Wang, Y., Wu, H., Li, S., Sun, Y., and Li, Z.: Contrasting size-resolved hygroscopicity of ﬁne particles de- rived by HTDMA and HR-ToF-AMS measurements between summer and winter in Beijing: the impacts of aerosol ag- ing and local emissions, Atmos. Chem. Phys., 20, 915–929, https://doi.org/10.5194/acp-20-915-2020, 2020. Kerminen, V. M., Chen, X., Vakkari, V., Petäjä, T., Kulmala, M., and Bianchi, F.: Atmospheric new particle formation and growth: review of ﬁeld observations, Environ. Res. Lett., 13, 103003, https://doi.org/10.1088/1748-9326/aadf3c, 2018. Khvorostyanov, V. I., and Curry, J. A.: A simple analytical model of aerosol properties with account for hygroscopic https://doi.org/10.5194/acp-21-14293-2021 Atmos. Chem. Phys., 21, 14293–14308, 2021 S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration S., Rose, D., Su, H., and Pöschl, U.: Rapid aerosol particle growth and increase of cloud condensation nucleus activity by secondary aerosol for- mation and condensation: A case study for regional air pol- lution in northeastern China, J. Geophys. Res., 114, D00G08, https://doi.org/10.1029/2008JD010884, 2008. B., Krejci, R., Laaksonen, A., Hamed, A., Minikin, A., Harri- son, R. M., Talbot, R., and Sun, J.: Explaining global surface aerosol number concentrations in terms of primary emissions and particle formation, Atmos. Chem. Phys., 10, 4775–4793, https://doi.org/10.5194/acp-10-4775-2010, 2010. Wu, Z., Hu, M., Liu, S., Wehner, B., and Wiedensohler, A.: Particle number size distribution in the urban atmo- sphere of Beijing, China, Atmos. Environ., 42, 7967–7980, https://doi.org/10.1016/j.atmosenv.2008.06.022, 2008. Stanier, C., Khlystov, A., and Pandis, S.: Ambient aerosol size dis- tributions and number concentrations measured during the Pitts- burgh Air Quality Study (PAQS), Atmos. Environ., 38, 3275– 3284, https://doi.org/10.1016/j.atmosenv.2004.03.020, 2004. Sullivan, S. C., Lee, D., Oreopoulos, L., and Nenes, A.: The role of updraft velocity in temporal variability of cloud hy- drometeor number, P. Natl. Acad. Sci. USA, 113, 5781–5790, https://doi.org/10.1073/pnas.1514043113, 2016. Wu, Z. J., Hu, M., and Yue, D. L.: Evolution of particle number size distribution in an urban atmosphere during episodes of heavy pollution and new particle formation, Sci. China Earth Sci., 54, 1772–1778, https://doi.org/10.1007/s11430-011-4227-9, 2011. Wu, Z. J., Poulain, L., Birmili, W., Größ, J., Niedermeier, N., Wang, Z. B., Herrmann, H., and Wiedensohler, A.: Some insights into the condensing vapors driving new particle growth to CCN sizes on the basis of hygroscopicity measurements, Atmos. Chem. Phys., 15, 13071–13083, https://doi.org/10.5194/acp-15-13071- 2015, 2015. Sun, Y. L., Wang, Z. F., Du, W., Zhang, Q., Wang, Q. Q., Fu, P. Q., Pan, X. L., Li, J., Jayne, J., and Worsnop, D. R.: Long- term real-time measurements of aerosol particle composition in Beijing, China: seasonal variations, meteorological effects, and source analysis, Atmos. Chem. Phys., 15, 10149–10165, https://doi.org/10.5194/acp-15-10149-2015, 2015. Tao, J., Kuang, Y., Ma, N., Hong, J., Sun, Y., Xu, W., Zhang, Y., He, Y., Luo, Q., Xie, L., Su, H., and Cheng, Y.: Secondary aerosol formation alters CCN activity in the North China Plain, At- mos. Chem. Phys., 21, 7409–7427, https://doi.org/10.5194/acp- 21-7409-2021, 2021. Xu, W. Q., Han, T. T., Du, W., Wang, Q. Q., Chen, C., Zhao, J.: Effects of aqueous-phase and photochemical pro- cessing on secondary organic aerosol formation and evolu- tion in Beijing, China, Environ. Sci. Tech., 51, 762–770, https://doi.org/10.1021/acs.est.6b04498, 2017. Topping, D. O., McFiggans, G. S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration 14307 Wiedensohler, A., Birmili, W., Nowak, A., Sonntag, A., Weinhold, K., Merkel, M., Wehner, B., Tuch, T., Pfeifer, S., Fiebig, M., Fjäraa, A. M., Asmi, E., Sellegri, K., Depuy, R., Venzac, H., Vil- lani, P., Laj, P., Aalto, P., Ogren, J. A., Swietlicki, E., Williams, P., Roldin, P., Quincey, P., Hüglin, C., Fierz-Schmidhauser, R., Gysel, M., Weingartner, E., Riccobono, F., Santos, S., Grün- ing, C., Faloon, K., Beddows, D., Harrison, R., Monahan, C., Jennings, S. G., O’Dowd, C. D., Marinoni, A., Horn, H.-G., Keck, L., Jiang, J., Scheckman, J., McMurry, P. H., Deng, Z., Zhao, C. S., Moerman, M., Henzing, B., de Leeuw, G., Löschau, G., and Bastian, S.: Mobility particle size spectrometers: har- monization of technical standards and data structure to facili- tate high quality long-term observations of atmospheric parti- cle number size distributions, Atmos. Meas. Tech., 5, 657–685, https://doi.org/10.5194/amt-5-657-2012, 2012. Seinfeld, J. H., Bretherton, C. S., Carslaw, K. S., Coe, H., De- Mott, P. J., Dunlea, E. J., Feingold, G., Ghan, S. J., Guenther, A.B., Kahn, R. A., Kracunas, I. P., Kreidenweis, S. M., Molina, M. J., Nenes, A., Penner, J. E., Prather, K. A., Ramanathan, V., Ramaswamy, V., Rasch, P. J., Ravishankara, A. R., Rosen- feld, D., Stephens, G., and Wood R.: Improving Our Fundamen- tal Understanding of the Role of Aerosol-Cloud Interactions in the Climate System, P. Nat. Acad. Sci. USA, 113, 5781–5790, https://doi.org/10.1073/pnas.1514043113, 2016. Shi, J., Khan, A., Harrison, R.: Measurements of ultraﬁne parti- cle concentration and size distribution in the urban atmosphere, Sci. Tot. Environ., 235, 51–64, https://doi.org/10.1016/S0048- 9697(99)00189-8, 1999. Spracklen, D. V., Carslaw, K. S., Merikanto, J., Mann, G. W., Red- dington, C. L., Pickering, S., Ogren, J. A., Andrews, E., Bal- tensperger, U., Weingartner, E., Boy, M., Kulmala, M., Laakso, L., Lihavainen, H., Kivekäs, N., Komppula, M., Mihalopoulos, N., Kouvarakis, G., Jennings, S. G., O’Dowd, C., Birmili, W., Wiedensohler, A., Weller, R., Gras, J., Laj, P., Sellegri, K., Bonn, B., Krejci, R., Laaksonen, A., Hamed, A., Minikin, A., Harri- son, R. M., Talbot, R., and Sun, J.: Explaining global surface aerosol number concentrations in terms of primary emissions and particle formation, Atmos. Chem. Phys., 10, 4775–4793, https://doi.org/10.5194/acp-10-4775-2010, 2010. Wiedensohler, A., Chen, Y. F., Nowak, A., Wehner, B., Achtert, P., Berghof, M., Birmili, W., Wu, Z. J., Hu, M., Zhu, T., Takegawa, N., Kita, K., Kondo, Y., Lou, S. R., Hofzumahaus, A., Holland, F., Wahner, A., Gunthe, S. S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration Ren, J., Zhang, F., Wang, Y., Collins, D., Fan, X., Jin, X., Xu, W., Sun, Y., Cribb, M., and Li, Z.: Using different assumptions of aerosol mixing state and chemical composition to predict CCN concentrations based on ﬁeld measurements in urban Beijing, At- mos. Chem. Phys., 18, 6907–6921, https://doi.org/10.5194/acp- 18-6907-2018, 2018. Martin, G. and Johnson, D.: The measurement and parameteriza- tion of effective radius of droplets in warm stratocumulus clouds, J. Atmos. Sci., 51, 1823–1842, https://doi.org/10.1175/1520- 0469(1994)051<1823:TMAPOE>2.0.CO;2, 1994. Rose, C., Sellegri, K., Moreno, I., Velarde, F., Ramonet, M., Wein- hold, K., Krejci, R., Andrade, M., Wiedensohler, A., Ginot, P., and Laj, P.: CCN production by new particle formation in the free troposphere, Atmos. Chem. Phys., 17, 1529–1541, https://doi.org/10.5194/acp-17-1529-2017, 2017. Merikanto, J., Spracklen, D. V., Pringle, K. J., and Carslaw, K. S.: Effects of boundary layer particle formation on cloud droplet number and changes in cloud albedo from 1850 to 2000, Atmos. Chem. Phys., 10, 695–705, https://doi.org/10.5194/acp-10-695- 2010, 2010. https://doi.org/10.5194/acp-21-14293-2021 Atmos. Chem. Phys., 21, 14293–14308, 2021 https://doi.org/10.5194/acp-21-14293-2021 S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration B., and Coe, H.: A curved multi- component aerosol hygroscopicity model framework: Part 1 – Inorganic compounds, Atmos. Chem. Phys., 5, 1205–1222, https://doi.org/10.5194/acp-5-1205-2005, 2005. Yue, D. L., Hu, M., Zhang, R. Y., Wu, Z. J., Su, H., Wang, Z. B., Peng, J. F., He, L. Y., Huang, X. F., Gong, Y. G., and Wiedensohler, A.: Potential contribution of new particle forma- tion to cloud condensation nuclei in Beijing, Atmos. Environ., 45, 6070–6077, https://doi.org/10.1016/j.atmosenv.2011.07.037, 2011. Twomey, S.: The inﬂuence of pollution on the short- wave albedo of clouds, J. Atmos. Sci., 34, 1149–1152, https://doi.org/10.1175/1520-0469(1977)0342.0.CO;2, 1977. Zimmerman, A. , Petters, M. D., and Meskhidze, N.: Ob- servations of new particle formation, modal growth rates, and direct emissions of sub-10 nm particles in an urban environment, Atmos. Environ., 242, 117835, https://doi.org/10.1016/j.atmosenv.2020.117835, 2020. Wang, S. C. and Flagan, R. C.: Scanning Electrical Mo- bility Spectrometer, Aerosol Sci. Tech., 13, 230–240, https://doi.org/10.1080/02786829008959441, 1990. Wang, Y., Zhang, F., Li, Z., Tan, H., Xu, H., Ren, J., Zhao, J., Du, W., and Sun, Y.: Enhanced hydrophobicity and volatil- ity of submicron aerosols under severe emission control con- ditions in Beijing, Atmos. Chem. Phys., 17, 5239–5251, https://doi.org/10.5194/acp-17-5239-2017, 2017. Zhang, F., Li, Y., Li, Z., Sun, L., Li, R., Zhao, C., Wang, P., Sun, Y., Liu, X., Li, J., Li, P., Ren, G., and Fan, T.: Aerosol hygroscopicity and cloud condensation nuclei activity during the AC3Exp campaign: implications for cloud condensation nu- https://doi.org/10.5194/acp-21-14293-2021 Atmos. Chem. Phys., 21, 14293–14308, 2021 https://doi.org/10.5194/acp-21-14293-2021 Atmos. Chem. Phys., 21, 14293–14308, 2021 S. Jiang et al.: The contribution of new particle formation to cloud droplet number concentration 14308 clei parameterization, Atmos. Chem. Phys., 14, 13423–13437, https://doi.org/10.5194/acp-14-13423-2014, 2014. clei parameterization, Atmos. Chem. Phys., 14, 13423–13437, https://doi.org/10.5194/acp-14-13423-2014, 2014. Zhang, Q., Jimenez, J. L., Canagaratna, M. R., Ulbrich, I. M., Ng, N. L., Worsnop, D. R., and Sun, Y. L.: Understanding at- mospheric organic aerosols via factor analysis of aerosol mass spectrometry: A review, Anal. Bioanal. Chem., 401, 3045–3067, https://doi.org/10.1007/s00216-011-5355-y, 2011. Zhang, F., Li, Z., Li, Y., Sun, Y., Wang, Z., Li, P., Sun, L., Wang, P., Cribb, M., Zhao, C., Fan, T., Yang, X., and Wang, Q.: Im- pacts of organic aerosols and its oxidation level on CCN ac- tivity from measurement at a suburban site in China, Atmos. Chem. Phys., 16, 5413–5425, https://doi.org/10.5194/acp-16- 5413-2016, 2016. Zhang, F.: Evaluation of the contribution of new particle forma- tion to cloud droplet in urban atmosphere [data set], available at: https://data.mendeley.com/datasets/hkkzbn4zv3/1, last access: 4 January 2021. Zhang, F., Wang, Y., Peng, J., Ren, J., Collins, D., Zhang, R., and Li, Z.: Uncertainty in predicting CCN activity of aged and primary aerosols, J. Geophys. Res.-Atmos., 122, 11723–11736, https://doi.org/10.1002/2017JD027058, 2017. Zheng, Y. T., Rosenfeld, D., and Li, Z. Q.: Satellite Inference of Thermals and Cloud-Base Updraft Speeds Based on Retrieved Surface and Cloud-Base Temperatures, J. Atmos. Sci., 72, 2411– 2428, https://doi.org/10.1175/JAS-D-14-0283.1, 2015. Zhang, F., Ren, J., Fan, T., Chen, L., Xu, W., and Sun, Y.: Sig- niﬁcantly enhanced aerosol CCN activity and number con- centrations by nucleation haze events: A case study in ur- ban Beijing, J. Geophys. Res.-Atmos., 124, 14102–14113, https://doi.org/10.1029/ 2019JD031457, 2019. Atmos. Chem. Phys., 21, 14293–14308, 2021 https://doi.org/10.5194/acp-21-14293-2021
https://openalex.org/W2044098535	https://link.springer.com/content/pdf/10.1007/JHEP07(2015)154.pdf	English	null	The two ∇6R4 type invariants and their higher order generalisation	The Journal of high energy physics/The journal of high energy physics	2,015	cc-by	25,306	Published for SISSA by Springer Published for SISSA by Springer Published for SISSA by Springer Received: April 1, 2015 Accepted: June 12, 2015 Published: July 29, 2015 Received: April 1, 2015 Accepted: June 12, 2015 Published: July 29, 2015 Open Access, c⃝The Authors. Article funded by SCOAP3. Keywords: Extended Supersymmetry, Nonperturbative Eﬀects, Supergravity Models, Supersymmetric Eﬀective Theories 3 Supergravity in higher dimensions 3.1 N = 4 supergravity in ﬁve dimensions 3.2 N = (2, 2) supergravity in six dimensions 3.3 N = 2 supergravity in seven dimensions 3.4 N = 2 supergravity in eight dimensions 3.4 N = 2 supergravity in eight dimensions A Ed(d) Eisenstein series, and tensorial diﬀerential equations B Some additional computations on Eisenstein series 41 B.1 E6(6) Eisenstein series in the fundamental representation 41 B.2 SO(6, 6) Eisenstein series 43 B.3 SO(n, n) Eisenstein series in the adjoint 46 p B.1 E6(6) Eisenstein series in the fundamental representation B.2 SO(6, 6) Eisenstein series B.3 SO(n, n) Eisenstein series in the adjoint The two ∇6R4 type invariants and their higher order generalisation doi:10.1007/JHEP07(2015)154 doi:10.1007/JHEP07(2015)154 Contents 1 Introduction 1 2 N = 8 supergravity in four dimensions 3 2.1 The standard ∇6R4 type invariant 5 2.2 F 2∇4R4 type invariant and its relation to ∇6R4 8 2.3 Dimensional reduction to three dimensions 11 2.4 E7(7) Eisenstein series 13 2.5 F 2k∇4R4 type invariants 18 2.6 Wavefront set and Poisson equation source term 20 2.7 String theory perturbation theory 24 3 Supergravity in higher dimensions 27 3.1 N = 4 supergravity in ﬁve dimensions 27 3.2 N = (2, 2) supergravity in six dimensions 30 3.3 N = 2 supergravity in seven dimensions 34 3.4 N = 2 supergravity in eight dimensions 35 A Ed(d) Eisenstein series, and tensorial diﬀerential equations 38 B Some additional computations on Eisenstein series 41 B.1 E6(6) Eisenstein series in the fundamental representation 41 B.2 SO(6, 6) Eisenstein series 43 B.3 SO(n, n) Eisenstein series in the adjoint 46 Contents 1 Introduction 1 2 N = 8 supergravity in four dimensions 3 2.1 The standard ∇6R4 type invariant 5 2.2 F 2∇4R4 type invariant and its relation to ∇6R4 8 2.3 Dimensional reduction to three dimensions 11 2.4 E7(7) Eisenstein series 13 2.5 F 2k∇4R4 type invariants 18 2.6 Wavefront set and Poisson equation source term 20 2.7 String theory perturbation theory 24 3 Supergravity in higher dimensions 27 3.1 N = 4 supergravity in ﬁve dimensions 27 3.2 N = (2, 2) supergravity in six dimensions 30 3.3 N = 2 supergravity in seven dimensions 34 3.4 N = 2 supergravity in eight dimensions 35 A Ed(d) Eisenstein series, and tensorial diﬀerential equations 38 B Some additional computations on Eisenstein series 41 B.1 E6(6) Eisenstein series in the fundamental representation 41 B.2 SO(6, 6) Eisenstein series 43 B.3 SO(n, n) Eisenstein series in the adjoint 46 1 Introduction 1 2 N = 8 supergravity in four dimensions 3 2.1 The standard ∇6R4 type invariant 5 2.2 F 2∇4R4 type invariant and its relation to ∇6R4 8 2.3 Dimensional reduction to three dimensions 11 2.4 E7(7) Eisenstein series 13 2.5 F 2k∇4R4 type invariants 18 2.6 Wavefront set and Poisson equation source term 20 2.7 String theory perturbation theory 24 3 Supergravity in higher dimensions 27 3.1 N = 4 supergravity in ﬁve dimensions 27 3.2 N = (2, 2) supergravity in six dimensions 30 3.3 N = 2 supergravity in seven dimensions 34 3.4 N = 2 supergravity in eight dimensions 35 A Ed(d) Eisenstein series, and tensorial diﬀerential equations 38 B Some additional computations on Eisenstein series 41 B.1 E6(6) Eisenstein series in the fundamental representation 41 B.2 SO(6, 6) Eisenstein series 43 B.3 SO(n, n) Eisenstein series in the adjoint 46 1 Introduction 2 N = 8 supergravity in four dimensions 2.1 The standard ∇6R4 type invariant 2.2 F 2∇4R4 type invariant and its relation to ∇6R4 2.3 Dimensional reduction to three dimensions 2.4 E7(7) Eisenstein series 2.5 F 2k∇4R4 type invariants 2.6 Wavefront set and Poisson equation source term 2.7 String theory perturbation theory JHEP07(2015)154 The two ∇6R4 type invariants and their higher order generalisation JHEP07(2015)154 Guillaume Bossard and Valentin Verschinin Centre de Physique Th´eorique, Ecole Polytechnique, CNRS, 91128 Palaiseau cedex, France E-mail: bossard@cpht.polytechnique.fr, valentin.verschinin@cpht.polytechnique.fr Guillaume Bossard and Valentin Verschinin Centre de Physique Th´eorique, Ecole Polytechnique, CNRS, 91128 Palaiseau cedex, France E-mail: bossard@cpht.polytechnique.fr, valentin.verschinin@cpht.polytechnique.fr Abstract: We show that there are two distinct classes of ∇6R4 type supersymmetry in- variants in maximal supergravity. The second class includes a coupling in F 2∇4R4 that generalises to 1/8 BPS protected F 2k∇4R4 couplings. We work out the supersymmetry constraints on the corresponding threshold functions, and argue that the functions in the second class satisfy to homogeneous diﬀerential equations for arbitrary k ≥1, such that the corresponding exact threshold functions in type II string theory should be proportional to Eisenstein series, which we identify. This analysis explains in particular that the ex- act ∇6R4 threshold function is the sum of an Eisenstein function and a solution to an inhomogeneous Poisson equation in string theory. Keywords: Extended Supersymmetry, Nonperturbative Eﬀects, Supergravity Models, Supersymmetric Eﬀective Theories Keywords: Extended Supersymmetry, Nonperturbative Eﬀects, Supergravity Models, Supersymmetric Eﬀective Theories ArXiv ePrint: 1503.04230 ArXiv ePrint: 1503.04230 Open Access, c⃝The Authors. Article funded by SCOAP3. 1 Introduction The determination of the exact string theory low energy eﬀective action is a very diﬃcult problem in general. In the case of type II string theory on R1,10−d ×T d−1, the lowest order non-perturbative corrections could nonetheless have been computed [1–3]. Although there is no non-perturbative formulation of the theory, the constraints following from supersym- metry and U-duality have permitted to determine the non-perturbative low energy eﬀective action from perturbative computations in string theory [4–8] and in eleven-dimensional su- pergravity [2, 9–12]. The four-graviton amplitude allows in particular to determine the ∇2kR4 type correction in the eﬀective action, L ∼1 κ2 R + X p,q κ2 d−3+4p+6q 9−d E(p,q)∇4p+6qR4 + . . . (1.1) (1.1) p,q – 1 – where the dots stand for other terms including the supersymmetric completion, (p, q) labels the diﬀerent invariant combinations of derivatives compatible with supersymmetry accord- ing to the notations used in [13], and E(p,q) are automorphic functions of the scalar ﬁelds deﬁned on Ed(d)(Z)\Ed(d)/Kd. For (p, q) = (0, 0), (1, 0) and (0, 1), the complete eﬀective action at this order is determined by these functions E(p,q), which have been extensively studied [14–33]. E(0,0) is an Eisenstein series associated to the minimal unitary representation [25, 27], E(1,0) is an (or a sum of two) Eisenstein series associated to the next to minimal unitary representation(s) [27], and both are therefore relatively well understood. They are nonetheless very complicated functions, and the explicit expansion of E(1,0) in Fourier modes is not yet determined [28, 29, 32]. E(0,1) is not even an Eisenstein series, and was shown in [10] to satisfy to an inhomogeneous Poisson equation in type IIB. A proposal for this function in eight dimensions [26], suggested a split of the function into the sum of an Eisenstein series and an inhomogeneous solution, which was subsequently generalised in seven and six dimensions [13, 27], and recently clariﬁed in [33]. JHEP07(2015)154 In this paper we extend the analysis carried out in [30, 31] to the study of E(0,1). We show that this function indeed splits into the sum of two functions that are associated to two distinct supersymmetry invariants, and therefore satisfy to inequivalent tensorial diﬀerential equations. In particular, the second satisﬁes to a homogeneous equation, which is solved by the Eisenstein function appearing in [13, 26, 33]. 1 Introduction One can distinguish the two functions by looking at speciﬁc higher point couplings that we identify. The new class of invariants generalises to an inﬁnite class admitting a coupling in F 2k∇4R4, and we identify a unique Eisenstein function solving the corresponding tensorial diﬀerential equations in all dimensions greater than four. This function turns out to be compatible with perturbative string theory, and only admits three perturbative contributions in four dimensions, at 1-loop, (k + 2)-loop, and 2k-loop. However, the only amplitude that seems to unambiguously distinguish it from others is the (k + 2)-loop four-graviton amplitude in a non-trivial Ramond-Ramond background, which makes an explicit check extremely challenging. We start with the analysis of the supersymmetry invariants in four dimensions. The two ∇6R4 type invariants in the linear approximation are associated to two distinct classes of chiral primary operators of SU(2, 2\|8) discussed in [34]. We identify the corresponding representations of E7(7) associated to nilpotent coadjoint orbits [35] that are summarised in ﬁgure 1. In the linearised approximation, the F 2∇4R4 type invariant does not carry a ∇6R4 coupling, but we explain that the structure of the linearised invariant allows for this mixing at the non-linear level, and that the latter must occur because the two classes of invariants merge in one single E8(8) representation in three dimensions. We conclude that the exact threshold function in four dimensions takes the form E(0,1) = ˆE(8,1,1) + 32 189π ˆEh 0 000005 i , (1.2) (1.2) where ˆE(8,1,1) is the solution to the inhomogeneous diﬀerential equation (2.143) that is consis- tent with perturbative string theory. The explicit relation between the tensorial diﬀerential where ˆE(8,1,1) is the solution to the inhomogeneous diﬀerential equation (2.143) that is consis- tent with perturbative string theory. The explicit relation between the tensorial diﬀerential – 2 – • • • • • • • • R R4 ∇4R4 F 2k∇4R4 ∇6R4 0 - - - - - - - - 34 52 54 64 66 70 76 dim Figure 1. Closure diagram of nilpotent orbits of E7(7) of dimension smaller than 76. JHEP07(2015)154 Figure 1. Closure diagram of nilpotent orbits of E7(7) of dimension smaller than 76. equations and the associated nilpotent orbits permits us to determine the wavefront set of the associated functions, extending the results of [28, 29] to the ∇6R4 threshold function. 1 Introduction It appears, as can be seen in ﬁgure 1, that the two functions admit distinct wavefront sets. In particular we show that although ˆE(8,1,1) is not an Eisenstein series, it admits the same wavefront set as ˆE h 0 00000 i. We then consider the uplift of our results in higher dimensions, and exhibit that this general structure extends to all dimensions lower than eight, and is in perfect agreement with the exact threshold functions proposed in [13, 26, 33]. In each dimension, the su- persymmetry invariants transform in irreducible representations of Ed(d), deﬁned by the representation of Ed(d) on the associated function on Ed(d)/Kd satisfying to the relevant dif- ferential equations implied by supersymmetry. The inequivalent invariants are summarised in ﬁgure 2. The tensorial diﬀerential equations satisﬁed by Eisenstein functions relevant to our analysis are reviewed in the appendices. 2 N = 8 supergravity in four dimensions Maximal supergravity includes 70 scalar ﬁelds parametrising the symmetric space E7(7)/SUc(8) [36], and can be deﬁned in superspace by promoting these ﬁelds to super- ﬁelds φµ [37, 38]. One deﬁnes the Maurer-Cartan form dV V−1 = 2δ[k [i ωl]j] Pijkl P ijkl −2δ[i [kωj]l] ! , (2.1) (2.1) with P ijkl = 1 24εijklpqrsPpqrs . (2.2) with with with P ijkl = 1 24εijklpqrsPpqrs . (2.2) th P ijkl = 1 24εijklpqrsPpqrs . (2.2) – 3 – IIA IIB IIA IIB IIA IIB - - • • • ◦ - • • ◦ - • • • - • • - • • - • • 10 8 7 6 5 4 3 dim • • • ◦ • ◦ • ◦ ◦ ◦ ◦ ◦ R4 ∇4R4 ∇6R4 Figure 2. Each node corresponds to an inequivalent supersymmetry invariant, white if it cannot be written in harmonic superspace in the linearised approximation, and red if the corresponding harmonic superspace is chiral. For ∇6R4, the links to 10 dimensions are valid for the homogeneous solution, while all the eight-dimensional invariants uplift to type IIA for the inhomogeneous solution. IIA IIB IIA IIB IIA IIB - - • • • ◦ - • • ◦ - • • • - • • - • • - • • 10 8 7 6 5 4 3 dim • • • ◦ • ◦ • ◦ ◦ ◦ ◦ ◦ R4 ∇4R4 ∇6R4 Figure 2. Each node corresponds to an inequivalent supersymmetry invariant, white if it ca IIA IIB JHEP07(2015)154 JHEP07(2015)154 Figure 2. Each node corresponds to an inequivalent supersymmetry invariant, white if it cannot be written in harmonic superspace in the linearised approximation, and red if the corresponding harmonic superspace is chiral. For ∇6R4, the links to 10 dimensions are valid for the homogeneous solution, while all the eight-dimensional invariants uplift to type IIA for the inhomogeneous solution. Figure 2. Each node corresponds to an inequivalent supersymmetry invariant, white if it cannot be written in harmonic superspace in the linearised approximation, and red if the corresponding harmonic superspace is chiral. For ∇6R4, the links to 10 dimensions are valid for the homogeneous solution, while all the eight-dimensional invariants uplift to type IIA for the inhomogeneous solution. 2 N = 8 supergravity in four dimensions In this approximation the superﬁeld W ijkl transforms in the minimal unitary represen- tation of the superconformal group SU(2, 2\|8) [39]. This property permits a complete classiﬁcation of supersymmetry invariants in the linearised approximation in terms of irre- ducible representations of SU(2, 2\|8) of Lorentz invariant top component [34, 40]. In our analysis, we rely on the assumption of absence of supersymmetry anomaly, such that there is no algebraic obstruction to the extension of a linearised invariant to a full non-linear invariant. This implies a bijective correspondence between the set of linearised invariants and the non-linear invariants, such that one can deduce the explicit gradient expansion of the functions (or tensor functions) of the scalar ﬁelds on E7(7)/SUc(8) that determine the invariants. JHEP07(2015)154 2 N = 8 supergravity in four dimensions The metric on E7(7)/SUc(8) is deﬁned as The metric on E7(7)/SUc(8) is deﬁned as Gµν(φ)dφµdφν = 1 3PijklP ijkl , (2.3) (2.3) and the derivative in tangent frame is deﬁned such that for any function and the derivative in tangent frame is deﬁned such that for any function dE = 3P ijklDijklE . (2.4) (2.4) The superﬁelds satisfy to Di αE = 1 4εijklpqrsχαjkl DpqrsE , ¯D ˙αiE = 6¯χjkl ˙α DijklE , (2.5) (2.5) where χαijk is the Dirac superﬁeld in Weyl components, and ¯χijk ˙α its complex conjugate. The expansion of the scalar ﬁelds include the 28 Maxwell ﬁeld strengths Fαβij, the 8 Rarita-Schwinger ﬁeld strengths ραβγi and the Weyl tensor Cαβγδ, satisfying to N = 8 supergravity classical (two derivatives) ﬁeld equations. The supervielbeins are the solutions to the Bianchi identities deﬁned such that the Riemann tensor is valued in sl(2, C) ⊕su(8) and the su(8) component is identiﬁed with the scalar ﬁeld curvature [37, 38], Rij = 1 3Pjklp ∧P iklp . (2.6) (2.6) The covariant derivative on E7(7)/SUc(8) in tangent frame satisﬁes to The covariant derivative on E7(7)/SUc(8) in tangent frame satisﬁes to Dijkl, Dpqrs Dtuvw = −24δijkl qrs][tDuvw][p + 3δijkl pqrsDtuvw , (2.7) (2.7) – 4 – and the Laplace operator is ∆= 1 3DijklDijkl . (2.8) ∆= 1 3DijklDijkl . (2. (2.8) In the linearised approximation, the scalar superﬁeld Wijkl satisﬁes to the reality con- straint (2.2) and to In the linearised approximation, the scalar superﬁeld Wijkl satisﬁes to the reality con- straint (2.2) and to Dp αWijkl = 2δp [iχαjkl] , ¯D ˙αpWijkl = 1 12εijklpqrs ¯χqrs ˙α . (2.9) (2.9) In this approximation the superﬁeld W ijkl transforms in the minimal unitary represen- tation of the superconformal group SU(2, 2\|8) [39]. This property permits a complete classiﬁcation of supersymmetry invariants in the linearised approximation in terms of irre- ducible representations of SU(2, 2\|8) of Lorentz invariant top component [34, 40]. In our analysis, we rely on the assumption of absence of supersymmetry anomaly, such that there is no algebraic obstruction to the extension of a linearised invariant to a full non-linear invariant. This implies a bijective correspondence between the set of linearised invariants and the non-linear invariants, such that one can deduce the explicit gradient expansion of the functions (or tensor functions) of the scalar ﬁelds on E7(7)/SUc(8) that determine the invariants. 2.1 The standard ∇6R4 type invariant JHEP07(2015)154 Z duu8iurjuskutlWrst = Wijkl , (2.15) Z duu8iusju1kurlW rtuWstu = WijpqW klpq −1 28δkl ij WpqrsW pqrs , (2.15) Z duu1qu8pu8iurjuskutlWu[rsWt]vwW uvw= Wpo[ijWkl]mnW qomn−\|W\|2 108 δq pWijkl−δp [iWjkl]p , Z duu1ku1lu8iu8jWursWtvwW uvwW rst = Wnpq(iWj)mp′q′W np′q′(kW l)pqm −δ(k (i δl) j)(. . . ) , which are respectively in the [0, 0, 0, 1, 0, 0, 0], the [0, 1, 0, 0, 0, 1, 0], the [1, 0, 0, 1, 0, 0, 1] and the [2, 0, 0, 0, 0, 0, 2] irreducible representations of SU(8), whereas which are respectively in the [0, 0, 0, 1, 0, 0, 0], the [0, 1, 0, 0, 0, 1, 0], the [1, 0, 0, 1, 0, 0, 1] and the [2, 0, 0, 0, 0, 0, 2] irreducible representations of SU(8), whereas Z duu1mu1nu8ku8luriusjutpuuqWvwrW vwtWsxyW uxy = Wi′j′[ijWk]lk′l′W i′j′[pqW m]nk′l′+. . (2 16) (2.16) (2.16) gives rise to the fourth order monomial in the [1, 0, 1, 0, 1, 0, 1] irreducible representation. One obtains in this way that the harmonic superspace integral of a general monomial of order n1 + 2n2 + 3n3 + 4n4 + 4n′ 4 + 4 in the [n2, n4, n1 + n3, n4, n2] of SU(6) gives rise to a term in ∇6R4 with a monomial of order n1 + 2n2 + 3n3 + 4n4 + 4n′ 4 in the [n3 + n4 + 2n′ 4, n2, n4, n1 + n3, n4, n2, n3 + n4 + 2n′ 4] of SU(8), i.e. Z duD14 ¯D14F(u)[n2,n4,n1+n3,n4,n2] [n3+n4+2n′ 4,n2,n4,n1+n3,n4,n2,n3+n4+2n′ 4]W n1+2n2+3n3+4n4+4n′ 4+4\|[n2,n4,n1+n3,n4,n2] ∼∇6R4 W n1+2n2+3n3+4n4+4n′ 4\|[n3+n4+2n′ 4,n2,n4,n1+n3,n4,n2,n3+n4+2n′ 4] + . . . (2.17) ∇6R4 W n1+2n2+3n3+4n4+4n′ 4\|[n3+n4+2n′ 4,n2,n4,n1+n3,n4,n2,n3+n4+2n′ 4] + . . . (2.17) (2.17) where the function F(u) is the function of the harmonic variable deﬁned as a product of the generating functions deﬁned in (2.15), (2.16). One needs at least one quartic singlet in the G-analytic superﬁeld to get a non-vanishing integral [34]. Referring to the one to one correspondence between linearised and non-linear invari- ants [34], one deduces that the non-linear invariant must admit the same gradient expan- sion, i.e. 2.1 The standard ∇6R4 type invariant One can deﬁne a ∇6R4 type invariant in harmonic superspace, using the harmonic variables u1i, uri, u8i parametrising SU(8)/S(U(1)×U(6)×U(1)), such that r = 2 to 7 of SU(6) [34, 40, 43]. In this case the harmonic superspace integral can be deﬁned at the non-linear level [44], but we will only consider its linearised approximation. The superﬁeld in the 20 of SU(6) Wrst = ui8ujruksultWijkl , (2.10) (2.10) satisﬁes to the G-analyticity constraints satisﬁes to the G-analyticity constraints u1iDi αWrst = 0 , ui8 ¯D ˙αiWrst = 0 . (2.11) (2.11) One can therefore integrate any function of Wrst on the associated analytic superspace. To understand the most general integrand, we must decompose monomials of Wrst in irre- ducible representations of SU(6). At quadratic order we have the representation [0, 0, 2, 0, 0] and the combination 1 One can therefore integrate any function of Wrst on the associated analytic superspace. To understand the most general integrand, we must decompose monomials of Wrst in irre- ducible representations of SU(6). At quadratic order we have the representation [0, 0, 2, 0, 0] and the combination W rtuWstu = 1 6εrtuvwxWstuWvwx (2.12) (2.12) in the [1, 0, 0, 0, 1]. Because one obtains the [0, 0, 2, 0, 0] by simply adding the Dynkin labels of Wrst, we will say that this representation is freely generated, whereas we shall consider the [1, 0, 0, 0, 1] as a new generator at order two. At cubic order, we have the two elements freely generated by the ones already discussed, i.e. [0, 0, 3, 0, 0] and [1, 0, 1, 0, 1], and the additional combination Wu[rsWt]vwW uvw , (2.13) (2.13) – 5 – – 5 – in the [0, 0, 1, 0, 0]. At quartic order we have the four elements freely generated by the ones already discussed, and the two additional elements Wvw[rW vw[tWs]xyW u]xy , WursWtvwW uvwW rst , (2.14) (2.14) that decompose into the [0, 1, 0, 1, 0] and the singlet representation. One checks that these elements freely generate the general polynomials in Wrst, such that the latter are labeled by ﬁve integers. To integrate such a function in analytic superspace, one needs to consider these gen- erating monomials with additional harmonic variables in order to compensate for the S(U(1) × U(6) × U(1)) representation, i.e. 2.1 The standard ∇6R4 type invariant L(8,1,1)[E(8,1,1)] = X n1,n2,n3,n4,n′ 4 Dn1+2n2+3n3+4n4+4n′ 4 [n3+n4+2n′ 4,n2,n4,n1+n3,n4,n2,n3+n4+2n′ 4] E(8,1,1) L [n3+n4+2n′ 4,n2,n4,n1+n3,n4,n2,n3+n4+2n′ 4] (8,1,1) L(8,1,1)[E(8,1,1)] = X Dn1+2n2+3n3+4n4+4n′ 4 [n3+n4+2n′ 4 n2 n4 n1+n3 n L(8,1,1)[E(8,1,1)] = X n1,n2,n3,n4,n′ 4 Dn1+2n2+3n3+4n4+4n′ 4 [n3+n4+2n′ 4,n2,n4,n1+n3,n4,n2,n3+n4+2n′ 4] E(8,1,1) L [n3+n4+2n′ 4,n2,n4,n1+n3,n4,n2,n3+n4+2n′ 4] (8,1,1) L(8,1,1)[E(8,1,1)] = X Dn1+2n2+3n3+4n4+4n′ 4 [n3+n4+2n′ n2 n4 n1+n3 n n1,n2,n3,n4,n′ 4 (2.18) – 6 – where each L [n3+n4+2n′ 4,n2,n4,n1+n3,n4,n2,n3+n4+2n′ 4] (8,1,1) is an E7(7) invariant superform in the cor- responding representation of SU(8). Note that although the irreducible representation remains unchanged under the substitution where each L [n3+n4+2n′ 4,n2,n4,n1+n3,n4,n2,n3+n4+2n′ 4] (8,1,1) is an E7(7) invariant superform in the cor- responding representation of SU(8). Note that although the irreducible representation remains unchanged under the substitution n1, n3, n′ 4 → n1 + 2, n3 −2, n′ 4 + 1 (2.19) (2.19) the corresponding superforms and the tensor structure of the derivative are diﬀerent, and are really labelled by the ﬁve integers n1, n2, n3, n4, n′ 4 without any further identiﬁcation. Of course the mass dimension implies that these integers are bounded from above, and the maximal weight terms in χ14 ¯χ14 can only be in representations like [2, 6, 0, 8, 0, 6, 2], [2, 6, 1, 6, 1, 6, 2], . . . [2, 10, 0, 0, 0, 10, 2], . . . [11, 1, 0, 0, 0, 1, 11]. JHEP07(2015)154 This gradient expansion implies in particular that the third order derivative of E(8,1,1) in the [0, 2, 0, 0, 0, 0, 0] and its complex conjugate must vanish, i.e. (4DijpqDpqmnDmnkl −Dijkl(∆+ 24)) E(8,1,1) = 0 , 4DijpqDpqmnDmnkl −Dijkl(∆+ 24) E(8,1,1) = 0 . (2.20) (2.20) These equations imply all the higher order constraints on the function such that its gradient expansion is in agreement with (2.18). Deﬁning the covariant derivative in tangent frame as a Lie algebra generator in the fundamental representation of E7(7), this equation reads equivalently D 3 56E(8,1,1) = D56 6 + 1 4∆ E(8,1,1) . (2.21) (2.21) This implies in particular that all the Casimir operators are determined by the quadratic one such that This implies in particular that all the Casimir operators are determined by the quadratic one such that tr D 2+2n 56 E(8,1,1) = 6∆ 6 + 1 4∆ n E(8,1,1) , (2.22) (2.22) but the quadratic Casimir is not a priori determined by equation (2.20) alone. F 2∇4R4 type invariant and its relation to ∇6R4 2.2 Although the ∇6R4 type invariant provides the unique supesymmetric invariant preserving SU(8) one can write at this order, there is another class of invariants that can be deﬁned form the chiral harmonic superspace deﬁned in terms of the harmonic variables uˆri, uri parametrising SU(8)/S(U(2) × U(6)) [34, 43], with ˆr, ˆs equal to 1, 2 of SU(2), and r, s running from 3 to 8 of SU(6). One deﬁnes the superﬁeld W rs = u1iu2jurkuslW ijkl (2.24) (2.24) that satisﬁes to the G-analiticity constraint JHEP07(2015)154 JHEP07(2015)154 uˆri ¯Di αW rs = 0 . (2.25) (2.25) Similarly as in the preceding section, the most general function of W rs is freely generated by the three monomials Similarly as in the preceding section, the most general function of W rs is freely generated by the three monomials W rs , 1 2εrstuvwW tuW vw , 1 2εrstuvwW rsW tuW vw . (2.26) (2.26) One must supplement them with harmonic variables to preserve S(U(2) × U(6)) invari- ance, using One must supplement them with harmonic variables to preserve S(U(2) × U(6)) invari- ance, using Z du ui1uj2ukruls W rs = W ijkl , (2.27) Z du ui1uj2urkusl 1 2εrstuvwW tuW vw = W ijpqWklpq −1 28δij klW pqrsWpqrs , Z du ui1uj2uk1ul2 1 2εrstuvwW rsW tuW vw = W ijpqWpqrsW rskl −1 12W ijklWpqrsW pqrs . Z du ui1uj2ukruls W rs = W ijkl , (2.27) 1 1 (2.27) Z du ui1uj2urkusl 1 2εrstuvwW tuW vw = W ijpqWklpq −1 28δij klW pqrsWpqrs , Z du ui1uj2uk1ul2 1 2εrstuvwW rsW tuW vw = W ijpqWpqrsW rskl −1 12W ijklWpqrsW pqrs . One only gets a non-trivial integral if the cubic SU(6) singlet in W rs appears at least quadratically, which can be understood from the property that the associated chiral primary operator of SU(2, 2\|8) is otherwise in a short representation [34]. Because the U(1) weight of the measure is compensated by a single factor of this cubic SU(6) singlet, it appears that there is no SU(8) invariant that exists in this class. One only gets a non-trivial integral if the cubic SU(6) singlet in W rs appears at least quadratically, which can be understood from the property that the associated chiral primary operator of SU(2, 2\|8) is otherwise in a short representation [34]. 2.1 The standard ∇6R4 type invariant We will need to consider the other invariants to ﬁnally conclude that supersymmetry moreover implies [27] ∆E(8,1,1) = −60E(8,1,1) −(E(8,4,4))2 . (2.23) (2.23) Equation (2.21) deﬁnes a qantization of the algebraic condition Q 3 56 = 0 associated to the complex nilpotent orbit of E7 of Dynkin label h 0 200000 i, while the condition that the fourth order derivative does not vanish generically in the [2,0,0,0,0,0,2] distinguishes its real form of SU(8) Dynkin label [2000002] [35], which deﬁnes the graded decomposition of SU(8) associated to the (8, 1, 1) harmonic superspace we consider in this section. The property that the linearised structure does not permit to determine the eigenvalue of the Laplace operator in this case, implies that the quantization of the associated nilpotent orbit is not unique, and depends on one free parameter. This property follows from the fact that a nilpotent element of this kind can be obtained as the appropriate limit of a semi-simple element satisfying to the characteristic equation Q 3 56 = 1 24tr(Q 2 56)Q56. – 7 – F 2∇4R4 type invariant and its relation to ∇6R4 Because the U(1) weight of the measure is compensated by a single factor of this cubic SU(6) singlet, it appears that there is no SU(8) invariant that exists in this class. For a general monomial, one gets an invariant of the form Z du ¯D16D12 F(u)[0,n1,0,n2,0] [0,n2+2n3+2,0,n1,0,n2,0] W n1+2n2+3n3+6\|[0,n1,0,n2,0] (2.28) ∼W n1+2n2+3n3 [0,n2+2n3,0,n1,0,n2,0] ¯F 2 [0,2,0,0,0,0,0]∇4R4 + . . . +W n1+2n2+n3−22 [0,n2+2n3−8,0,n1−8,0,n2−4,0] ¯χ16 [0,8,0,4,0,0,0]χ12 [0,2,0,4,0,4,0] , (2.28) where all terms are projected to the [0,n2+2n3+2,0,n1,0,n2,0] irreducible representation, and the term in ¯F 2 is ˙ ˙ ¯F ij ˙α ˙β ¯F ˙α ˙βkl −¯F [ij ˙α ˙β ¯F kl] ˙α ˙β . (2.29) (2.29) For a generic function F[W] of Wrs, one obtains For a generic function F[W] of Wrs, one obtains D16 ¯D12F[W] = X n1,n2,n3 ∂n1+2n2+3n3+6F[W] ∂W n1+2n2+3n3+6 [0,n2,0,n1,0]L[0,n1,0,n2,0](n1+2n2+3n3+3) (8,2,0) lin , (2.30) (2.30) – 8 – where the densities L[0,n1,0,n2,0](n1+2n2+3n3+3) (8,2,0) lin are of order n1 + 2n2 + 3n3 + 6 in the ﬁelds here the densities L[0,n1,0,n2,0](n1+2n2+3n3+3) (8,2,0) lin are of order n1 + 2n2 + 3n3 + 6 in the ﬁelds and only depend on the scalar ﬁelds through their space-time derivative. The number n1 + 2n2 + 3n3 + 3 is the U(1) weight of the density. These densities determine by construction covariant superforms in the linearised approximation [45–47], such that d(0)Lij,kl (8,2,0) lin = 0 , d(0)Lij,kl,pqrs (8,2,0) lin + 3P pqrs ∧Lij,kl (8,2,0) lin [0,2,0,1,0,0,0] = 0 , d(0)Lij,kl,pqrs,mntu (8,2,0) lin + 3P pqrs ∧Lij,kl,mntu (8,2,0) lin [0,2,0,2,0,0,0] = 0 , d(0)Lij,kl,pq (8,2,0) linrs + 18Prsmn ∧Lij,kl,pqmn (8,2,0) lin [0,3,0,0,0,1,0] = 0 , (2.31) JHEP07(2015)154 (2.31) where d(0) is the superspace exterior derivative in the linear approximation. At the next order, because d = ∞ X n=0 d(n) (2.32) (2.32) satisﬁes to d2 = 0, one has {d(0), d(1)} = 0 , (2.33) (2.33) and therefore and therefore d(0) d(1)Lij,kl (8,2,0) lin = 0 . (2.34) (2.34) We assume in this paper that the structure of superconformal multiplets implies the absence of supersymmetry anomaly, or equivalently that the ﬁfth cohomology of d(0) is empty. Nev- ertheless, even if d(1)Lij,kl (8,2,0) lin only depends on the covariant superﬁelds, nothing prevents its d(0) antecedent to depend explicitly on the scalar ﬁelds. F 2∇4R4 type invariant and its relation to ∇6R4 Following (2.17), we have d(0)L(8,1,1) lin = 0 , d(0)Lijkl (8,1,1) lin = −3P ijkl ∧L(8,1,1) lin , d(0)Lijkl,pqrs (8,1,1) lin = −3 P ijkl ∧Lpqrs (8,1,1) lin [0,0,0,1,0,0,0] , d(0)L ij (8,1,1) lin kl = −18 Pklpq ∧Lijpq (8,1,1) lin [0,1,0,0,0,1,0] , (2.37) (2.37) and therefore and therefore d(0) W ijpqWpqrsW rskl −1 12W ijklWpqrsW pqrs L(8,1,1) lin +W ijpqWpqrsLrskl (8,1,1) lin + W ijpqW klrsL(8,1,1) linpqrs + W klpqWpqrsLrsij (8,1,1) lin +6W pqijL kl (8,1,1) lin pq + 6W pqklL ij (8,1,1) lin pq −12W i]pq[kL l][j (8,1,1) lin pq = 18 P pqij ∧L kl (8,1,1) lin pq + P pqkl ∧L ij (8,1,1) lin pq −2P i]pq[k ∧L l][j (8,1,1) lin pq , (2.38) d(0) W ijpqWpqrsW rskl −1 12W ijklWpqrsW pqrs L(8,1,1) lin JHEP07(2015)154 +W ijpqWpqrsLrskl (8,1,1) lin + W ijpqW klrsL(8,1,1) linpqrs + W klpqWpqrsLrsij (8,1,1) lin +6W pqijL kl (8,1,1) lin pq + 6W pqklL ij (8,1,1) lin pq −12W i]pq[kL l][j (8,1,1) lin pq = 18 P pqij ∧L kl (8,1,1) lin pq + P pqkl ∧L ij (8,1,1) lin pq −2P i]pq[k ∧L l][j (8,1,1) lin pq , (2.38) (2.38) such that L ij (8,1,1) lin kl is indeed a consistent candidate. Moreover, the structure of the linearised (8, 1, 1) invariant does not permit to have the tensor function W ijpqWpqrsW rskl, such that (2.38) is not the exterior derivative of a superform that does not depend on the naked scalar ﬁelds (uncovered by a space-time derivative). It follows that such a correction, if it appeared in (2.35), could not be reabsorbed in a redeﬁnition of Lij,kl (8,2,0) (1). ( ) ( ) If this mixing between the (8, 2, 0) and the (8, 1, 1) superforms was not appearing at the non-linear level, then the action of the exterior derivative in the function of the scalar ﬁelds should not introduce lower derivative terms such that it should satisfy then to Dijpq(4DpqrsDrsmnDmnkl −Dpqkl(∆+ 24)) E(8,2,0) = 0 . (2.39) (2.39) If the mixing did appear, then the unicity of the linearised invariants (2.37) would imply that the corresponding non-linear superform should be the same as in (2.18), such that once again the exterior derivative acting on D3 [0,2,0,0,0,0,0]E(8,2,0) should not generate lower derivative terms and one would conclude again that (2.39) must be satisﬁed. F 2∇4R4 type invariant and its relation to ∇6R4 This implies in this case that d(1)Lij,kl (8,2,0) lin = −d(0)Lij,kl (8,2,0) (1) + Ppqrs ∧Mij,kl,pqrs + P pqij ∧Mpqkl + P pqkl ∧Mpqij −2P i]pq[k ∧Mpql][j , (2.35) (2.35) where Lij,kl (8,2,0) (1) is the covariant correction to the superform, whereas Mij,kl,pqrs and Mijkl are superforms of order six in the ﬁelds in the [0, 2, 0, 1, 0, 0, 0] and the [0, 1, 0, 0, 0, 1, 0], respectively, that must satisfy to where Lij,kl (8,2,0) (1) is the covariant correction to the superform, whereas Mij,kl,pqrs and Mijkl are superforms of order six in the ﬁelds in the [0, 2, 0, 1, 0, 0, 0] and the [0, 1, 0, 0, 0, 1, 0], respectively, that must satisfy to d(0)Mij,kl,pqrs = P pqrs ∧N ij,kl [0,2,0,1,0,0,0] , d(0)Mijkl = P ijpq ∧Nklpq −1 28δij klP pqrs ∧Npqrs . (2.36) (2.36) In order to have such corrections that could not be reabsorbed in a covariant correction as Lij,kl (8,2,0) (1), one must have a corresponding short multiplet associated to a linearised invariant of the same dimension. The only candidate for a superform Mij,kl,pqrs is Lij,kl,pqrs (8,2,0) lin , but it is of order seven in the ﬁelds, and therefore Mij,kl,pqrs = 0 at this order. However, there is a candidate for Mijkl which is L ij (8,1,1) lin kl, the superform that appears in the ∇6R4 type – 9 – invariant discussed in the last section. Following (2.17), we have invariant discussed in the last section. F 2∇4R4 type invariant and its relation to ∇6R4 Therefore this equation must be satisﬁed in either cases. Using moreover the property that the gradient expansion of the linearised invariant is inconsistent with the presence of the third order derivative in the [1, 0, 0, 1, 0, 0, 1] of SU(8), one requires 36Djr[klDirmnDpq]mn −δi jDklpq(∆+ 42) + δi [kDlpq]j(∆−120) E(8,2,0) = 0 . (2.40) (2.40) Using this equation one computes independently of (2.39) that Dijpq(4DpqrsDrsmnDmnkl−Dpqkl(∆+ 24)) E(8,2,0) = 1 12 28DijpqDklpq−3δij kl∆ (∆+60) E(8,2,0) (2.41) d l d th t (2 39) d (2 40) i l t th Dijpq(4DpqrsDrsmnDmnkl−Dpqkl(∆+ 24)) E(8,2,0) = 1 12 28DijpqDklpq−3δij kl∆ (∆+60) E(8,2, (2 4 (2.41) and we conclude that (2.39) and (2.40) imply together and we conclude that (2.39) and (2.40) imply together ∆E(8,2,0) = −60E(8,2,0) . (2.42) ∆E(8,2,0) = −60E(8,2,0) . (2.42) – 10 – This eigenvalue is such that the structure of the invariant is consistent with the mixing between the (8, 2, 0) and the (8, 1, 1) superforms. Only in this case can they reduce to the same invariant for a function E(8,2,2) satisfying to both (2.20) and (2.40), as for the ∇4R4 type invariant. We are going to argue now that this chiral invariant must indeed include a ∇6R4 coupling, because the two classes of invariants reduce to one single class in three dimensions. But before to do this, let us mention that (2.40) can be rewritten as D 3 133E(8,2,0) = 1 3D133∆E(8,2,0) , (2.43) (2.43) JHEP07(2015)154 which deﬁnes a qantization of the algebraic eqation Q 3 133 = 0 associated to the complex nilpotent orbit of E7 of Dynkin label h 0 000002 i with the real form deﬁned with the SU(8) Dynkin label [0200000] [35], which deﬁnes the graded decomposition of SU(8) associated to the (8, 2, 0) harmonic superspace we consider in this section. In this case the choice of real form moreover implies that the complex charge in the 70 deﬁning the nilpotent orbit through the Kostant-Sekiguchi correspondence satisﬁes to QijpqQpqmnQmnkl = 0 , (2.44) (2.44) such that it admits a unique quantization, with the eigenvalue of the Laplace operator −60. However, we will see in the following that the constraint (2.39) can be relaxed while keeping the property that the associated representation of E7(7) is a highest weight representation. such that it admits a unique quantization, with the eigenvalue of the Laplace operator −60. F 2∇4R4 type invariant and its relation to ∇6R4 However, we will see in the following that the constraint (2.39) can be relaxed while keeping the property that the associated representation of E7(7) is a highest weight representation. 1This property follows from the fact that the classiﬁcation of duality orbits of the black hole charges are the same in the N = 2 supergravity theories of duality group SO∗(12) and SU(3, 3) [50]. 2.3 Dimensional reduction to three dimensions In three dimensions, the duality group is E8(8), of maximal compact subgroup Spin(16)/Z2. We denote i, j the SO(16) vector indices and A, B the positive chirality Weyl spinor in- dices. The covariant derivative in tangent frame is a chiral Weyl spinor, i.e. in the 0 0000001 representation. In the linearised approximation, the covariant ﬁelds all descend from the Weyl spinor scalar ﬁeld, satisfying to [48] Di αW A = ΓiA ˙Aχα ˙A . (2.45) (2.45) Both four-dimensional (8, 1, 1) and (8, 2, 0) harmonic superspaces descend to the same (16, 2) harmonic superspace in three dimensions, deﬁned through the introduction of har- monic variables parametrising SO(16)/(U(2) × SO(12)) [49]. The Weyl spinor representa- tion decomposes with respect to U(2) × Spin(12) as Both four-dimensional (8, 1, 1) and (8, 2, 0) harmonic superspaces descend to the same (16, 2) harmonic superspace in three dimensions, deﬁned through the introduction of har- monic variables parametrising SO(16)/(U(2) × SO(12)) [49]. The Weyl spinor representa- tion decomposes with respect to U(2) × Spin(12) as 128 ∼= 32 (−1) + ⊕(2 ⊗32−) (0) ⊕32 (1) + , (2.46) (2.46) such that the grad 1 Weyl spinor W of Spin(12) satisﬁes to a G-analyticity constraint with respect to the positive grad covariant derivative in the 2 of U(2). The general polynomial in the Spin(12) Weyl spinor is parametrised by four integers, just as for the rank three antisymmetric tensor of SU(8) in section (2.1).1 One computes in a similar way the general such that the grad 1 Weyl spinor W of Spin(12) satisﬁes to a G-analyticity constraint with respect to the positive grad covariant derivative in the 2 of U(2). The general polynomial in the Spin(12) Weyl spinor is parametrised by four integers, just as for the rank three antisymmetric tensor of SU(8) in section (2.1).1 One computes in a similar way the general – 11 – integral Z duF(u) 0 0n2 0n4n1+n3 0 0n3+n4+2n′ 4 0n2 0n4n1+n3 D28W n1+2n2+3n3+4n4+4n′ 4+4\| 0 0n2 0n4n1+n3 ∼∇10P 4 W n1+2n2+3n3+4n4+4n′ 4\| 0 0n3+n4+2n′ 4 0n2 0n4n1+n3 + . . . (2.47) (2.47) where ∇10P 4 is a Spin(16) invariant quartic term in the scalar ﬁeld momentum, that replaces the ∇6R4 type term that vanishes modulo the equations of motion in three di- mensions. 2.3 Dimensional reduction to three dimensions In three dimensions it is not established if there is a one to one correspondence between non-linear and linear invariants deﬁned as harmonic superspace integrals. Nev- ertheless, the class of invariants we discuss descends from four dimensions, and we can therefore assume they admit the same structure, i.e. JHEP07(2015)154 L(16,2)[E(16,2)] = X n1,n2,n3,n4,n′ 4 Dn1+2n2+3n3+4n4+4n′ 4 0 0n3+n4+2n′ 4 0n2 0n4n1+n3 E(16,2) L 0 0n3+n4+2n′ 4 0n2 0n4n1+n3 . (2.48) (2.48) This expansion implies that the fourth order derivative of the function E(16,2) restricted to the 0 1000100 must vanish, i.e. DΓi[jk rD DΓlpq]rD E(16,2) = −δi[j DΓklpq]D (∆+ 48)E(16,2) , (2.49) (2.49) where the Laplace operator ∆is deﬁned as where the Laplace operator ∆is deﬁned as where the Laplace operator ∆is deﬁned as ∆= DADA . (2.50) (2.50) By dimensional reduction of the four-dimensional equation (2.42), one computes that ∆E(16,2) = −198E(16,2) . (2.51) (2.51) One can understand that the two kinds of 1/8 BPS invariants discussed in the preceding section dimensionally reduce to this single class. If one consider the decomposition of (2.46) with respect to U(6) ⊂Spin(12), one obtains for one embedding One can understand that the two kinds of 1/8 BPS invariants discussed in the preceding section dimensionally reduce to this single class. If one consider the decomposition of (2.46) with respect to U(6) ⊂Spin(12), one obtains for one embedding 32+ ∼= 6(−2) ⊕20(0) ⊕6 (2) , (2.52) (2.52) such that the G-analytic superﬁeld in the 32+ includes the four-dimensional (8, 1, 1) G- analytic scalar W rst as well as some components of the vector ﬁelds. A generic spinor of non-zero quartic invariant can be represented by W rst. For the other embedding U(6) ⊂ Spin(12), one gets 32+ ∼= 1 (−3) ⊕15(−1) ⊕15 (1) ⊕1(3) , (2.53) (2.53) such that the G-analytic superﬁeld in the 32+ includes the four-dimensional (8, 2, 0) G- analytic scalar W rs as well as some components of the vector ﬁelds, and a Ehlers complex scalar parametrising the four-dimensional metric. The scalar ﬁeld alone only parametrises a null spinor of Spin(12) of vanishing quartic invariant, and only together with the Ehlers – 12 – scalar ﬁeld it can provide a representative of a generic spinor. 2.4 E7(7) Eisenstein series In this section we shall discuss some properties of Einstein series that solve the diﬀerential equations we have derived for the ∇6R4 type invariants. 2.3 Dimensional reduction to three dimensions One could have naively con- cluded that the function E(8,2,0) should give rise to a function on E8(8)/Spinc(16) satisfying moreover to scalar ﬁeld it can provide a representative of a generic spinor. One could have naively con- cluded that the function E(8,2,0) should give rise to a function on E8(8)/Spinc(16) satisfying moreover to 5(DΓijpqD) DΓklpqD E = −20 DΓijklD (∆+ 48) E + 28δkl ij ∆(∆+ 120) E , (2.54) (2.54) ut this equation only admits solutions for functions satisfying to the Laplace equation ∆E = −210 E , (2.55) (2.55) excepted for the functions satisfying to the quadratic and cubic constraints that deﬁne the R4 and ∇4R4 type invariants. We see therefore that this equation is incompatible with supersymmetry. JHEP07(2015)154 It follows that both (8, 1, 1) and (8, 2, 0) type invariants dimensionally reduce to three- dimensional invariants depending of functions on E8(8)/Spinc(16) that belong to the same representation of E8(8). Being in the same representation, they both carry a quartic com- ponent in the linearised approximation and they must both include a ∇6R4 type term in their uplift to four dimensions. This proves that the mixing between the two diﬀerent linearised structures must occur such that the non-linear ¯F 2∇4R4 type invariant cannot exist without including a ∇6R4 type term as well. Before to end this section on the three-dimensional theory, let us discuss the modiﬁ- cation of the supersymmetry constraint due to the completion of the R4 type invariant at the next order. As it is argued in [10], the appearance of a R4 correction with threshold function E(16,8), will modify the Laplace equation with a non-zero right-hand-side, i.e. ∆E(16,2) = −198E(16,2) −E 2 (16,8) . (2.56) (2.56) Because the function E(16,8) satisﬁes to [30] Because the function E(16,8) satisﬁes to [30] Because the function E(16,8) satisﬁes to [30] (DΓijklD) E(16,8) = 0 , (2.57) (2.57) the second derivative of its square must necessarily vanish in the 0 1000100 , and we get accordingly a modiﬁcation of (2.49) to DΓi[jk rD DΓlpq]rD E(16,2) = 150δi[j DΓklpq]D E(16,2) + δi[j DΓklpq]D E 2 (16,8) . (2.58) (2.58) 2.4.1 Fundamental representation As discussed in [25, 31], one can deﬁne the Eisenstein series As discussed in [25, 31], one can deﬁne the Eisenstein series Eh 0 00000s i = X Γ∈Z56 I′′ 4 (Γ)\|133=0 \|Z(Γ)ijZ(Γ)ij\|−s , (2.59) (2.59) – 13 – as a sum over the rank one integral charge vectors Γ in the 56 of E7(7) satisfying to the constraint that the quadratic tensor Γ ⊗Γ vanishes in the adjoint representation. This formula is rather useful to identify the diﬀerential equations satisﬁed by the Eisenstein function, because one can simply consider the case of one charge Γ, with Z(Γ)ij = VijIΓI, such that the quadratic constraint becomes Z[ijZkl] = 1 24εijklpqrsZpqZrs , ZikZjk = 1 8δj i ZklZkl , (2.60) (2.60) and the diﬀerential operator acts on Zij as an element of e7(7) and the diﬀerential operator acts on Zij as an element of e7(7) JHEP07(2015)154 DijklZpq = 3δpq [ijZkl] , DijklZpq = 1 8εijklpqrsZrs . (2.61) (2.61) Using the deﬁnition \|Z\|2 = ZijZij, one computes that the function \|Z\|−2s satisﬁes to DijpqDklpq\|Z\|−2s = 2s(s −2)ZijZkl\|Z\|−2s−2 + s(s −11) 4 δkl ij \|Z\|−2s , DijpqDpqrsDrskl\|Z\|−2s = −3s(s −2)(s −4)ZijZkl\|Z\|−2s−2 + s2 −15s + 8 4 Dijkl\|Z\|−2s , Djr[klDirmnDpq]mn\|Z\|−2s = (s −2)(s −7) 12 δi jDklpq\|Z\|−2s −s2 −9s −40 12 δi [kDpql]j\|Z\|−2s , (2.62) (2.62) (2.62) and to the Laplace equation ∆\|Z\|−2s = 3s(s −9)\|Z\|−2s . (2.63) (2.63) For s ̸= 2, 4, the function admits a generic gradient expansion in the irreducible represen- tations [0, n2 + 2n3, 0, n1, 0, n2, 0] and their complex conjugate. To exhibit this property, it is convenient to consider a restricted set of indices as follows For s ̸= 2, 4, the function admits a generic gradient expansion in the irreducible represen- tations [0, n2 + 2n3, 0, n1, 0, n2, 0] and their complex conjugate. To exhibit this property, it is convenient to consider a restricted set of indices as follows D12ijDijklDkl12 n3D12pqD78pqn2(D1234)n1 \|Z\|−2s (2.64) = (s + n1 + n2 + n3 −1)!(s + n2 + n3 −3)!(s + n3 −5)! (s −1)!(s −3)!(s −5)! × × -3Z 2 12 n32Z12Z78n2-6Z[12Z34] n1 \|Z\|−2(s+n1+n2+n3) . (2.64) One computes moreover that for m ≤n One computes moreover that for m ≤n D78ijDijklDkl78m (D12pqDpqrsDrs12)n \|Z\|−2s (2.65) (2.65) = (s + n−1)!(s + n−3)!(s + n−5)!(s + n + m−1)!(s + n + m −3)!(s −n + m −5)! 2.4.1 Fundamental representation (s −1)!(s −3)!(s −5)!(s + n −1)!(s + n −3)!(s −n −5)! × × -3Z78 2m-3Z 2 12 n \|Z\|−2(s+n+m) = -3 2 n+m (s + n −5)!(s + n + m −1)!(s + n + m −3)!(s −n + m −5)! (s + n −m −5)!(s + 2n −1)!(s + 2n −3)!(s −n −5)! × × D12ijDijklDkl12 n−mD12pqD78pqn+m \|Z\|−2s = (s + n−1)!(s + n−3)!(s + n−5)!(s + n + m−1)!(s + n + m −3)!(s −n + m −5)! (s −1)!(s −3)!(s −5)!(s + n −1)!(s + n −3)!(s −n −5)! × × -3Z78 2m-3Z 2 12 n \|Z\|−2(s+n+m) – 14 – such that acting with a derivative operator in the conjugate representation [0, 0, 0, 0, 0, 2m, 0] does not produce an independent tensor. One has in particular for s an integer greater than 5 such that acting with a derivative operator in the conjugate representation [0, 0, 0, 0, 0, 2m, 0] does not produce an independent tensor. One has in particular for s an integer greater than 5 D78ijDijklDkl78 (D12pqDpqrsDrs12)s−4 \|Z\|−2s = 0 . (2.66) (2.66) This equation is the equivalent on E7(7)/SUc(8) of the equation on SL(2)/SO(2) This equation is the equivalent on E7(7)/SUc(8) of the equation on SL(2)/SO(2) ¯DDs−1E 0 ¯DDs−1E[s] = 0 , (2.67) (2.67) for integral s, and we would like to see that the function E h 0 00000s i also decomposes somehow into a “holomorphic” part Fs and a “anti-holomorphic” part ¯Fs, satisfying respectively to for integral s, and we would like to see that the function E h 0 00000s i also decomposes somehow into a “holomorphic” part Fs and a “anti-holomorphic” part ¯Fs, satisfying respectively to JHEP07(2015)154 JHEP07(2015)154 (D12pqDpqrsDrs12)s−4 ¯Fs = 0 , D78ijDijklDkl78s−4 Fs = 0 , (2.68) (2.68) such that such that (D12pqDpqrsDrs12)s−4 Eh 0 00000s i =(D12pqDpqrsDrs12)s−4 Fs , (2.69) (2.69) and respectively for the complex conjugate. By consistency, this requires for instance that acting with further derivatives on this tensor does not permit to get back lower order tensors with n3 < s −4 in (2.64). Through representation theory, one obtains that Through representation theory, one obtains that D[0,0,0,1,0,0,0]Dn1+2n2+3n3 [0,n2+2n3,0,n1,0,n2,0]\|Z\|−2s (2.70) ∼ Dn1+1+2n2+3n3 [0,n2+2n3,0,n1+1,0,n2,0] + Dn1−1+2(n2+1)+3n3 [0,n2+2n3+1,0,n1−1,0,n2+1,0] + Dn1+2(n2−1)+3(n3+1) [0,n2+2n3+1,0,n1,0,n2−1,0] +Dn1−1+2n2+3n3 [0,n2+2n3,0,n1−1,0,n2,0] + Dn1+1+2(n2−1)+3n3 [0,n2+2n3−1,0,n1+1,0,n2−1,0] + Dn1+2(n2+1)+3(n3−1) [0,n2+2n3−1,0,n1,0,n2+1,0] \|Z\|−2s (2.70) for some coeﬃcients that are not speciﬁed. 2.4.1 Fundamental representation So the only way to reduce n3, is to increase n2 by 1 unit. We will check this equation in the case n1 = n2 = 0. The restriction of the derivative D3n\|Z\|−2s to the [0, 2n, 0, 0, 0, 0, 0] with two free indices reads D3n [0,2n,0,0,0,0,0] ij12n−122n−1 \|Z\|−2s = (s+n−1)!(s+n−3)!(s+n−5)! (s −1)!(s −3)!(s −5)! (−3)n 2n ZijZ 2n−1 12 −(2n−1)Z1[iZj]2Z 2n−2 12 \|Z\|−2(s+n) , = (s+n−1)!(s+n−3)!(s+n−5)! (s −1)!(s −3)!(s −5)! (−3)n 2n ZijZ 2n−1 12 −(2n−1)Z1[iZj]2Z 2n−2 12 \|Z\|−2(s+n) , (2.71) (2.71) (2.71) and one computes that and one computes that D78ij 1 2n ZijZ 2n−1 12 −(2n −1)Z1[iZj]2Z 2n−2 12 \|Z\|−2(s+n) = (2n + 5)(n −s + 4) 8n Z78Z 2n−1 12 \|Z\|−2s , (2.72) D78ij 1 2n ZijZ 2n−1 12 −(2n −1)Z1[iZj]2Z 2n−2 12 \|Z\|−2(s+n) 2n j 12 ( ) [ j] 12 \| \| = (2n + 5)(n −s + 4) 8n Z78Z 2n−1 12 \|Z\|−2s , (2.72) (2.72) such that such that such that D78ijD3n [0,2n,0,0,0,0,0] ij12n−122n−1 \|Z\|−2s = 3(s + n −5)(2n + 5)(s −n −4) 16n D12ijDijklDkl12 n−1D12pqD78pq \|Z\|−2s . (2.73) 16n – 15 – In particular we have that In particular we have that D78ijD3(s−4) [0,2(s−4),0,0,0,0,0] ij12s−522s−5 \|Z\|−2s = 0 , (2.74) (2.74) consistently with the assumption that no lower order tensor is produced out of the tensor function (2.69). We conclude therefore that the tensor FE[0,2(s−4),0,0,0,0,0] = D3(s−4) [0,2(s−4),0,0,0,0,0]Eh 0 00000s i , (2.75) (2.75) is an E7(7)(Z) modular form that is in some sense holomorphic, such that its gradient expansion is restricted to derivative of this tensor in the symmetric representations [0, n2 + 2(n3 + s −4), 0, n1, 0, n2, 0], i.e. JHEP07(2015)154 Dn1+2n2+3n3 [0,n2+2n3,0,n1,0,n2,0]FE[0,2(s−4),0,0,0,0,0] ∈[0, n2 + 2(n3 + s −4), 0, n1, 0, n2, 0] . 2.4.1 Fundamental representation (2.76) (2.76) Using Langlands functional identity [27], one computes that the only integer values of s ≥5 for which the function diverges are Using Langlands functional identity [27], one computes that the only integer values of s ≥5 for which the function diverges are Eh 0 000005+ϵ i = 63 16π ϵEh 0 000004 i + ˆEh 0 000005 i + O(ϵ) , Eh 0 000007+ϵ i = 1 964 655ζ(5) 2048π5 ϵ Eh 0 000002 i + ˆEh 0 000007 i + O(ϵ) , Eh 0 000009+ϵ i = 12 642 554 925ζ(5)ζ(9) 2 097 152π9 ϵ + ˆEh 0 000009 i + O(ϵ) . (2.77) Eh 0 000007+ϵ i = 2048π5 ϵ Eh 0 000002 i + Eh 0 000007 i + O(ϵ) , Eh 0 000009+ϵ i = 12 642 554 925ζ(5)ζ(9) 2 097 152π9 ϵ + ˆEh 0 000009 i + O(ϵ) . (2.77) (2.77) However, according to (2.62), the function Eh 0 00000s i satisﬁes to D3 [0,2,0,0,0,0,0]Eh 0 00000s i = 0 , for s = 0, 2, 4 , (2.78) (2.78) and it follows that the tensor FE[0,2(s−4),0,0,0,0,0] (2.75) is ﬁnite for all s. However, we have argued in the preceding section that the ¯F 2∇4R4 type invariant must include a ∇6R4 type term, which will be multiplied by the function itself. In this case the relevant Eisenstein function diverges, and one must regularise it such that the diﬀerential equation (2.42) will be modiﬁed to 189 ∆ˆEh 0 000005 i = −60 ˆEh 0 000005 i + 189 16πEh 0 000004 i . (2.79) (2.79) Such a correction is reminiscent of a 1-loop logarithm divergence of the ∇4R4 type invariant form factor. Such a correction is reminiscent of a 1-loop logarithm divergence of the ∇4R4 type invariant form factor. 2.4.2 Adjoint representation One can also consider the Eisenstein series2 Eh 0 s00000 i = X Q∈Z133\|Q2=0 X(Q)ijklX(Q)ijkl−s , (2.80) (2.80) 2We assume here that all the elements of Z133 are in the E7(7)(Z) orbit of a relative integer times a normalised representative of the continuous orbit. This property does not aﬀect our conclusions in any case, which only requires the generating character to satisfy to the diﬀerential equations we discuss. 2We assume here that all the elements of Z133 are in the E7(7)(Z) orbit of a relative integer times a normalised representative of the continuous orbit. This property does not aﬀect our conclusions in any case, which only requires the generating character to satisfy to the diﬀerential equations we discuss. – 16 – as the sum over integral charges Q ∈e7(7) satisfying to the constraint Q2 = 0, and such that the adjoint action of the coset representative V on Q decomposes into the anti-Hermitian traceless matrix Λij and the complex-selfdual antisymmetric tensor Xijkl satisfying to the constraints ΛikΛkj = −1 48δi jXklpqXklpq , Λ[i [kΛj] l] = −1 2XijpqXklpq + 1 48δij klXpqrsXpqrs , Λ[ipXj]pkl = Λ[kpXl]pij . (2.81) (2.81) JHEP07(2015)154 The action of the derivative on these tensors is deﬁned as the e7(7) action DijklXpqrs = 12δ[pqr [ijk Λs] l] , DijklΛpq = 2δp [iXjkl]q + 1 4δp qXijkl . (2.82) DijklXpqrs = 12δ[pqr [ijk Λs] l] , DijklΛpq = 2δp [iXjkl]q + 1 4δp qXijkl . (2.82) omputes for \|X\|2 = X Xijkl that (2.82) One computes for \|X\|2 = XijklXijkl that Dijkl\|X\|2 = −24Xp[ijkΛp l] , DijpqXklpq = 10δ[k [i Λj] l] , (2.8 Dijkl\|X\|2 = −24Xp[ijkΛp l] , DijpqXklpq = 10δ[k [i Λj] l] , (2.83) DijpqDklpq\|X\|2 = 30XijpqXklpq + 3δkl ij \|X\|2 , Dijpq\|X\|2Dklpq\|X\|2 = 12XijpqXklpq\|X\|2 , (2.83) (2.83) [ DijpqDklpq\|X\|2 = 30XijpqXklpq + 3δkl ij \|X\|2 , Dijpq\|X\|2Dklpq\|X\|2 = 12XijpqXklpq\|X\|2 , which permits to derive that DijpqDklpq\|X\|−2s = 6s(2s −3)XijpqXklpq\|X\|−2s−2 −3sδkl ij \|X\|−2s . (2.84) (2.84) One gets therefore a solution to the equation One gets therefore a solution to the equation One gets therefore a solution to the equation DijpqDklpqE(8,4,4) = −9 2δkl ij E(8,4,4) , (2.85) (2.85) for s = 3 2. One computes in general that DijpqDpqrsDrskl\|X\|−2s = s2 −17 2 s + 6 Dijkl\|X\|−2s , (2.86) (2.86) and the function satisﬁes to (2.20) and its complex conjugate for all s. 3We are grateful to Axel Kleinschmidt who provided the explicit coeﬃcients 2.4.2 Adjoint representation The restriction of the third order derivative to the [1, 0, 0, 1, 0, 0, 1] gives D1k[12D34]ijD8ijk\|X\|−2s = −3 4s(2s −3)(2s −5)Λ81X1234\|X\|−2s−2 , (2.87) (2.87) showing that the function solves the cubic equation (2.40) for s = 5 2. These functions satisfy the same equations as their analog Eisenstein functions deﬁned in the fundamental representation, consistently with the property that3 Eh 0 3 2 00000 i = 2 πEh 0 000002 i , Eh 0 5 2 00000 i = 8 15πEh 0 000004 i . (2.88) (2.88) ne can also consider the restriction of the fourth order derivative to the [2, 0, 0, 0, 0, 0, 0, 2] vanish, which deﬁnes a further restriction on solutions to (2.20). In this case one obtains D8kijD1lijD1kpqD8lpq\|X\|−2s = −9 2s(2s −3)(2s −5)(s −4)Λ81Λ81\|X\|−2s−2 , (2.89) D8kijD1lijD1kpqD8lpq\|X\|−2s = −9 2s(2s −3)(2s −5)(s −4)Λ81Λ81\|X\|−2s−2 , (2.89) (2.89) 3We are grateful to Axel Kleinschmidt who provided the explicit coeﬃcients. 3We are grateful to Axel Kleinschmidt who provided the explicit coeﬃcients. – 17 – and this further restriction distinguishes the value s = 4. We have in general ∆\|X\|−2s = 2s(2s −17)\|X\|−2s , (2.90) (2.90) and the function E h 0 400000 i does not solve the 1/8 BPS equation. Using the same normalisa- tion as [27] with a factor of 2ζ(2s) in the deﬁnition of the Eisenstein function, one computes using Langlands formula and the function E h 0 400000 i does not solve the 1/8 BPS equation. Using the same normalisa- tion as [27] with a factor of 2ζ(2s) in the deﬁnition of the Eisenstein function, one computes using Langlands formula Eh 0 s00000 i = π 33 2 Γ(s−8)Γ(s−13 2 )Γ(s−11 2 )Γ(2s−17 2 )ζ(2s−16)ζ(2s−13)ζ(2s−11)ζ(4s−17) Γ(s −5 2)Γ(s −3 2)Γ(s)Γ(2s −8)ζ(17 −2s)ζ(2s −5)ζ(2s −3)ζ(4s −16) ×Eh 0 17 2 -s00000 i . (2.91) (2.91) (2.91) JHEP07(2015)154 The function is singular for various values of s, i.e. 9 2, 11 2 , 6, 13 2 , 7, and 17 2 , and in particular for s = 6, which is the relevant value to solve equation (2.20) with (2.23). One should therefore consider the regularised series The function is singular for various values of s, i.e. 9 2, 11 2 , 6, 13 2 , 7, and 17 2 , and in particular for s = 6, which is the relevant value to solve equation (2.20) with (2.23). 2.4.2 Adjoint representation One should therefore consider the regularised series Eh 0 6+ϵ00000 i = π5 8ζ(9) ϵEh 0 5 2 00000 i + ˆEh 0 600000 i + O(ϵ) . (2.92) (2.92) However, we will see in the following that this function does not appear in string theory, similarly as the ∇6R4 threshold function is not described by an Eisenstein series in type IIB supergravity. Nonetheless, some components of this function should appear, as we will argue in the following. However, we will see in the following that this function does not appear in string theory, similarly as the ∇6R4 threshold function is not described by an Eisenstein series in type IIB supergravity. Nonetheless, some components of this function should appear, as we will argue in the following. 2.5 F 2k∇4R4 type invariants The F 2∇4R4 type invariants we have discussed in section 2.2 have a natural generalisation to higher order invariants. Considering the same chiral harmonic superspace deﬁned in terms of the harmonic variables uˆri, uri parametrising SU(8)/S(U(2) × U(6)) [34], one can deﬁne the G-analytic superﬁelds ¯F 12 ˙α ˙β = u1iu2jF ij ˙α ˙β , ¯χ12r ˙α = u1iu2jurk ¯χijk ˙α . (2.93) (2.93) They do not permit to deﬁne directly chiral primary operators of SU(2, 2\|8), because to deﬁne directly chiral primary operators of SU(2, 2\|8), because ¯D ˙αtW rs = δ[r t ¯χ12s] ˙α , ¯D ˙αrχ12s ˙β = δs r ¯F 12 ˙α ˙β . (2.94) (2.94) Chiral primary operators are annihilated by the special supersymmetry generators at the origin, i.e. Sr ˙γ ¯F 12 ˙α ˙β = ε ˙γ( ˙α ¯χ12r ˙β) , Sr ˙α ¯χ12s ˙β = ε ˙α ˙βW rs , St ˙αW rs = 0 . (2.95) (2.95) One can enforce this property by deﬁning a chiral primary as One can enforce this property by deﬁning a chiral primary as O (k) F = (S)12 ¯F 12 ˙α ˙β ¯F ˙α ˙β122+k F[W] ∝ εrstuvwW rsW tuW vw2 ¯F 12 ˙α ˙β ¯F ˙α ˙β12k−1 F[W] + . . . (2.96) ( ) for an arbitrary function F of the G-analytic superﬁeld W rs. By construction such a chiral primary operator is never short, and deﬁnes a non-trivial integrand for the (8, 2, 0) measure. for an arbitrary function F of the G-analytic superﬁeld W rs. By construction such a chiral primary operator is never short, and deﬁnes a non-trivial integrand for the (8, 2, 0) measure. – 18 – Because one can consider an arbitrary representative up to a total fermionic derivative, one can as well consider the ﬁrst term in (2.96) as the integrand. Because one can consider an arbitrary representative up to a total fermionic derivative, one can as well consider the ﬁrst term in (2.96) as the integrand. So similarly as in section 2.2, we get the general class of linearised invariants for an arbitrary positive integer k, Z du ¯D16D12 F(u)[0,n1,0,n2,0] [0,n2+2n3+2k,0,n1,0,n2,0] ¯F 2k−2 W n1+2n2+3n3+6\|[0,n1,0,n2,0] (2.97) ∼W n1+2n2+3n3 [0,n2+2n3,0,n1,0,n2,0] ¯F 2k [0,2k,0,0,0,0,0]∇4R4 + . . . +W n1+2n2+n3−22 [0,n2+2n3−8,0,n1−8,0,n2−4,0] ¯F 2k−2 [0,2k−2,0,0,0,0,0] ¯χ16 [0,8,0,4,0,0,0]χ12 [0,2,0,4,0,4,0] . 2.6 Wavefront set and Poisson equation source term We have seen that there are two classes of ∇6R4 type invariants in four dimensions, that preserve tree-level supersymmetry modulo the classical ﬁeld equations. However, consid- ering that the eﬀective action already includes an R4 type correction, we must take into account the action of the accordingly modiﬁed supersymmetry transformation on the R4 type invariant itself. This is a very diﬃcult task to carry out in practice, but one can nonetheless show general properties on these corrections. We recall that the R4 type in- variant admits the following gradient expansion in derivatives of the function E(8,4,4) JHEP07(2015)154 L(8,4,4)[E(8,4,4)] = 12 X n=0 Dn [0,0,0,n,0,0,0]E(8,4,4)L[0,0,0,n,0,0,0] (8,4,4) , (2.103) (2.103) with E(8,4,4) satisfying to (2.85). The ﬁrst order modiﬁcation of the supersymmetry trans- formations will therefore necessarily admit the same gradient expansion in the function E(8,4,4), such that δ = δ(0) + 12 X n=0 Dn [0,0,0,n,0,0,0]E(8,4,4)δ(1) [0,0,0,n,0,0,0] + . . . , (2.104) (2.104) where the dots stand for higher order corrections. It follows that the correction at second order will admit the expansion δ Z L(8,4,4)[E(8,4,4)] = Z 12 X n=0 Dn [0,0,0,n,0,0,0]E(8,4,4)δ(1) [0,0,0,n,0,0,0] ! 12 X m=0 Dm [0,0,0,m,0,0,0]E(8,4,4)L[0,0,0,m,0,0,0] (8,4,4) ! = Z X m n X R Dn [0,0,0,n,0,0,0]E(8,4,4)Dm [0,0,0,m,0,0,0]E(8,4,4) R ΨR m,n (2.105) (2.105) where the sum over R runs over all irreducible representations of SU(8) in the tensor product [0, 0, 0, n, 0, 0, 0]⊗[0, 0, 0, m, 0, 0, 0], and ΨR m,n are understood to be E7(7) invariant densities function of the ﬁelds and their covariant derivatives in the irreducible repre- sentation R. One checks that all the appearing irreducible representations R are self- conjugate, i.e. of the type [n4, n3, n2, n1, n2, n3, n4], by property of the tensor product [0, 0, 0, n, 0, 0, 0] ⊗[0, 0, 0, m, 0, 0, 0]. The ¯F 2∇4R4 type invariant admits a gradient ex- pansion with non-self conjugate irreducible representations, and all its components in self- conjugate representations do in fact coincide with ones appearing in the (8, 1, 1) ∇6R4 type invariant. It follows that the analysis of the supersymmetry constraints on the ¯F 2∇4R4 type invariant is not modiﬁed by the presence of the R4 correction, and equations (2.40), (2.42) are the exact equations to be solved by the corresponding function E(8,2,0) in the Wilso- nian action. 2.5 F 2k∇4R4 type invariants (2.97) JHEP07(2015)154 We conclude that the corresponding supersymmetry invariants admit the same gradient expansion in We conclude that the corresponding supersymmetry invariants admit the same gradient expansion in L (k) (8,2,0)[E (k) (8,2,0)] = X n1≥0,n2≥0,n3≥k Dn1+2n2+3n3 [0,n2,0,n1,0,n2+2n3,0]E (k) (8,2,0) Lk [0,n2+2n3,0,n1,0,n2,0] (8,2,0) , (2.98) (2.98) for a function E (k) (8,2,0) satisfying to (2.40), and k ≥2. The coupling at the lowest number of points is then of the kind D3k [0,0,0,0,0,2k,0]E (k) (8,2,0) Lk [0,2k,0,0,0,0,0] (8,2,0) = D3k [0,0,0,0,0,2k,0]E (k) (8,2,0) ¯F 2k [0,2k,0,0,0,0,0] ∇4R4 + . . . (2.9 (2.99) In principle one could expect to have a non-trivial mixing with another class of linearised invariant at the non-linear level, just as the one of the ¯F 2∇4R4 type invariant with the ∇6R4 type invariant described in section 2.2. However, there is no higher order chiral primary operator that can deﬁne a non-trivial (8, 1, 1) harmonic superspace integral, and we did not ﬁnd any linearised invariant with the right structure to deﬁne a possible cohomology class as does (2.38). Therefore we expect these invariants to have the same structure as the associated linearised invariants, i.e. to only contribute to (4 + 2k)-point amplitudes and higher. Independently of this assumption, the structure of these invariants requires that the action of the derivative Dijkl on D3kE (k) (8,2,0) does not generate lower order derivatives of the function. This condition is precisely (2.74), and we conclude therefore that the eigenvalue of the Laplace operator is determined in the same way as ∆E (k) (8,2,0) = 3(k + 4)(k −5)E (k) (8,2,0) , (2.100) (2.100) such that the function satisﬁes to such that the function satisﬁes to 12Djr[klDirmnDpq]mnE (k) (8,2,0) = (k + 2)(k −3)δi jDklpqE (k) (8,2,0) −(k(k −1) −60)δi [kDlpq]jE (k) (8,2,0) . (2.101) ( It is therefore tempting to conjecture that It is therefore tempting to conjecture that E (k) (8,2,0) ∝Eh 0 000004+k i , (2.102) (2.102) in the string theory eﬀective action, and we will indeed show in section 2.7 that this function admits a consistent perturbative string theory limit. Moreover, we will see in section 3 that it also admits an appropriate decompactiﬁcation limit. – 19 – – 19 – 2.6 Wavefront set and Poisson equation source term 2.6 Wavefront set and Poisson equation source term However, all the irreducible representations that appear in the gradient expan- sion (2.18) are included in the tensor product [0, 0, 0, n, 0, 0, 0] ⊗[0, 0, 0, m, 0, 0, 0] for m – 20 – and n running from 1 to twelve, and the diﬀerential equations (2.20), (2.23) must be modi- ﬁed in the presence of the R4 type correction. Following the analysis carried out in [10, 27], we conclude that and n running from 1 to twelve, and the diﬀerential equations (2.20), (2.23) must be modi- ﬁed in the presence of the R4 type correction. Following the analysis carried out in [10, 27], we conclude that ∆E(8,1,1) = −60E(8,1,1) −(E(8,4,4))2 . (2.106) (2.106) As explained in [30], this requires then to modify (2.20) to As explained in [30], this requires then to modify (2.20) to As explained in [30], this requires then to modify (2.20) to DijpqDpqmnDmnklE(8,1,1) = −9DijklE(8,1,1) −1 2E(8,4,4)DijklE(8,4,4) , DijpqDpqmnDmnklE(8,1,1) = −9DijklE(8,1,1) −1 2E(8,4,4)DijklE(8,4,4) . (2.107) (2.107) JHEP07(2015)154 Using in particular the tensor product Using in particular the tensor product [0,0,0,2,0,0,0] ⊗[0,0,0,1,0,0,0] ∼= [0,0,0,3,0,0,0] ⊕[0,1,0,1,0,1,0] ⊕[1,0,0,1,0,0,1] ⊕[0,0,1,1,1,0,0] ⊕[0,0,0,1,0,0,0] one shows that [0,0,0,2,0,0,0] ⊗[0,0,0,1,0,0,0] ∼= [0,0,0,3,0,0,0] ⊕[0,1,0,1,0,1,0] ⊕[1,0,0,1,0,0,1] ⊕[0,0,1,1,1,0,0] ⊕[0,0,0,1,0,0,0] one shows that (4DijpqDpqmnDmnkl −Dijkl(∆+ 24)) (E(8,4,4))2 = 0 , (2.108) (2.108) whereas 36Djr[klDirmnDpq]mn −δi jDklpq(∆+ 42) + δi [kDlpq]j(∆−120) (E(8,4,4))2 ̸= 0 . (2.109) We therefore conclude that no higher derivative correction in (E(8,4,4))2 can consistently modify (2.107) without contradicting (2.106). These properties of the source term in the Poisson equation (2.106) can also be under- stood through the structure of the Fourier modes of these functions. In the decompactiﬁ- cation limit, the Fourier modes of a function are the coeﬃcients, functions on E6(6)/Spc(4), of e2πi(q,a), with the axion ﬁeld a in the 27 of the E6(6) subgroup. We have shown in [30] that (2.85) implies then that the associated momenta are rank one vectors, i.e. using the cubic Jordan norm 3 det (q) = tr q(q × q), q × q = 0 , (2.110) (2.110) consistently with the properties of the R4 threshold function. If we consider the square of E(8,4,4), it admits by construction Fourier modes of momenta q1 +q2 where q1 and q2 satisfy to (2.110), such that consistently with the properties of the R4 threshold function. 2.6 Wavefront set and Poisson equation source term If we consider the square of E(8,4,4), it admits by construction Fourier modes of momenta q1 +q2 where q1 and q2 satisfy to (2.110), such that det (q1 + q2) = det (q1) + tr q1(q2 × q2) + tr q2(q1 × q1) + det (q2) = 0 , (2.111) (2.111) As one can see in [30], equation (2.106) implies that the Fourier modes of the function E(8,1,1) must indeed carry momenta satisfying to the rank 2 constraint det (q) = 0, whereas the Fourier modes of the function E(8,2,0) are generic by construction in the parabolic de- composition. The nilpotent orbit associated to E(8,2,0) is indeed deﬁned from the graded decomposition As one can see in [30], equation (2.106) implies that the Fourier modes of the function E(8,1,1) must indeed carry momenta satisfying to the rank 2 constraint det (q) = 0, whereas the Fourier modes of the function E(8,2,0) are generic by construction in the parabolic de- composition. The nilpotent orbit associated to E(8,2,0) is indeed deﬁned from the graded decomposition e7(7) ∼= 27 (−2) ⊕ gl1 ⊕e6(6) (0) ⊕27(2) , (2.112) (2.112) – 21 – such that a representative of the nilpotent orbit is a generic element of the grad two component in the 27. Considering instead the string theory limit, the non-abelian Fourier modes are deﬁned over a Heisenberg algebra with 32 momenta in the positive chirality Weyl spinor represen- tation of Spin(6, 6) associated to Ramond-Ramond D-brane charge Q, and an additional momentum associated to the Neveu-Schwarz 5-brane charge N5. The nilpotent orbit asso- ciated to E(8,1,1) is deﬁned from the associated graded decomposition e7(7) ∼= 1(−4) ⊕32(−2) ⊕(gl1 ⊕so(6, 6)) (0) ⊕32(2) ⊕1(4) . (2.113) (2.113) JHEP07(2015)154 A representative of the nilpotent orbit is deﬁned as a generic Weyl spinor in the grad 2 component [51], Q ∈Spin(6, 6)/SU(2, 4) , or Q ∈Spin(6, 6)/SL(6) , (2.114) (2.114) to which one can add an arbitrary element of the grad 4 component N5. This implies in particular that equation (2.107) does not imply any constraint on the Fourier modes. Equation (2.85) implies instead that Q must be a rank 1 spinor, [51] to which one can add an arbitrary element of the grad 4 component N5. This implies in particular that equation (2.107) does not imply any constraint on the Fourier modes. 2.6 Wavefront set and Poisson equation source term Equation (2.85) implies instead that Q must be a rank 1 spinor, [51] (Q2)\|66 ˆ= (QΓMNQ) = 0 , Q ∈Spin(6, 6)/ SL(6) ⋉R15 , (2.115) (2.115) as for example the grad 3 singlet in the decomposition so(6, 6) ∼= 15 (−2) ⊕(gl1 ⊕sl6) (0) ⊕15(2) , 32 ∼= 1(−3) ⊕15(−1) ⊕15 (1) ⊕1(3) . (2.116) (2.116) A generic rank 1 charge vector can always be rotated to the grad 3 component. Considering the sum of two rank one charges, respectively in the grad -3 and the grad 3 components, one obtains a generic rank 4 spinor of stabilizer SL(6) ⊂Spin(6, 6). All the rank four charges deﬁned as the sum of two rank 1 charges with a non-trivial symplectic product can be written in this form. Therefore the right-hand-side in (2.107) indeed sources generic Fourier modes of E(8,1,1). More precisely, all the Fourier modes with a negative quartic invariant I4(Q) ≤0 (belonging to the second orbit in (2.114)) are sourced by the function E 2 (8,1,1), whereas the Fourier modes with a strictly positive quartic invariant I4(Q) (belonging to the ﬁrst orbit in (2.114)) satisfy to a homogeneous equation. On the contrary, a representative of the nilpotent orbit associated to E(8,2,0) satisﬁes that its third power in the adjoint representation vanishes, which according to (2.113) implies that [51] Q3)\|32 ˆ= QΓMNQ ΓMNQ = 0 , Q ∈Spin(6, 6)/ SL(2) × Spin(3, 4) ⋉R2×8+1 . (2.117 (2.117) The relation with the Fourier modes is not completely straightforward in the presence of a non-trivial NS5-brane charge, because in that case the nilpotent subgroup is a non-abelian Heisemberg group, such that the corresponding Killing vector κα = ∂ ∂aα −1 2Cαβaβ ∂ ∂b , k5 = ∂ ∂b , (2.118) (2.118) – 22 – satisfy to [κα, κβ] = Cαβk5 , (2.119) (2.119) [κα, κβ] = Cαβk5 , where Cαβ is the antisymmetric charge conjugation matrix of Spin(6, 6). For a Fourier mode of vanishing NS5-brane charge, k5EQ,0 = 0, and one can deﬁne the spinor charge Q such that καEQ,0 = iQαEQ,0, and Q must satisfy to the same algebraic equations as the representatives of the nilpotent orbits associated to the diﬀerential equations. For a non-zero NS5-brane charge the relevant equations are more complicated, but still involve the Killing vector κα to the third order in the same combination. 2.6 Wavefront set and Poisson equation source term Let us now consider the M-theory limit, for which one considers the decomposition JHEP07(2015)154 e7(7) ∼= 7(−4) ⊕35 (−2) ⊕(gl1 ⊕sl7) (0) ⊕35(2) ⊕7 (4) , (2.120) (2.120) In this case the nilpotent subgroup also generate a non-abelian Heisenberg type algebra ilpotent subgroup also generate a non-abelian Heisenberg type algebra κmnp = ∂ ∂amnp −1 12εmnpqrstaqrs ∂ ∂bt , km = ∂ ∂bm , (2.121) (2.121) such that such that [κmnp, κqrs] = 1 6εmnpqrstkt . (2.122) (2.122) For a Fourier mode of vanishing M5-brane charge, kmEq,0 = 0, and one can deﬁne the M2- brane charge κmnpEq,0 = iqmnpEq,0. For a non-zero M5-brane charge kmEq,p = ipmEq,p the relevant equations are more complicated, but still involve the Killing vector in a way similar as does the corresponding nilpotent orbit characteristic equation involves the algebraic charges. For a 1/2 BPS charge satisfying to the quadratic constraint, one obtains [52] For a Fourier mode of vanishing M5-brane charge, kmEq,0 = 0, and one can deﬁne the M2- brane charge κmnpEq,0 = iqmnpEq,0. For a non-zero M5-brane charge kmEq,p = ipmEq,p the relevant equations are more complicated, but still involve the Killing vector in a way similar as does the corresponding nilpotent orbit characteristic equation involves the algebraic charges. For a 1/2 BPS charge satisfying to the quadratic constraint, one obtains [52] εmnqrstuqpqrqstu = 0 , qmnppp = 0 , (2.123) (2.123) giving 17 = 13 + 4 linearly independent solutions, with typical representative giving 17 = 13 + 4 linearly independent solutions, with typical representative 1 6qmnpdym ∧dyn ∧dyp = q1dy1 ∧dy2 ∧dy3 . (2.124) (2.124) The cubic constraint in the adjoint representation implies The cubic constraint in the adjoint representation implies εnrstuvwqrstquv[pqqm]w = 0 , εmnqrstuqpqrqstupp = 0 , (2.125) (2.125) that gives 27 = 21 + 6 linearly independent solutions, with typical representative 1 6qmnpdym ∧dyn ∧dyp = dy1 ∧ q1dy2 ∧dy3 + q2dy4 ∧dy5 + q3dy6 ∧dy7 . 2.6 Wavefront set and Poisson equation source term As we have already explained, the product of two functions including non-perturbative corrections admits generic Fourier modes in the string theory limit, because the sum of two pure spinors can be a generic spinor. We see therefore that a source term modifying (2.101) would necessarily involve the third order diﬀerential operator such as to source these Fourier modes. Such a modiﬁcation would destroy completely the structure of the equations, which would reduce then to some kind of Poisson equation. However, the solutions to the diﬀerential equation (2.101) admit restricted Fourier modes in the string theory limit, satisfying to (2.117) for a vanishing NS5-brane charge. As we have already explained, the product of two functions including non-perturbative corrections admits generic Fourier modes in the string theory limit, because the sum of two pure spinors can be a generic spinor. We see therefore that a source term modifying (2.101) would necessarily involve the third order diﬀerential operator such as to source these Fourier modes. Such a modiﬁcation would destroy completely the structure of the equations, which would reduce then to some kind of Poisson equation. 2.6 Wavefront set and Poisson equation source term (2.126) The cubic constraint in the fundamental implies instead εmnrstuvqmnrqst(pqq)uv = 0 , (2.127) (2.127) that gives 33 = 26+7 linearly independent solutions, such that the SL(7) M2-brane charge orbits are either that gives 33 = 26+7 linearly independent solutions, such that the SL(7) M2-brane charge orbits are either q ∈SL(7, R)/(SL(3, C) ⋉C3) , or q ∈SL(7, R)/(SL(3, R) ⋉R3)×2 , (2.128) q ∈SL(7, R)/(SL(3, C) ⋉C3) , or q ∈SL(7, R)/(SL(3, R) ⋉R3)×2 , (2.128) (2.128) – 23 – – 23 – with typical representative with typical representative 1 6qmnpdym∧dyn∧dyp = q1dy1∧dy2∧dy3+q2dy1∧dy4∧dy5+q3dy2∧dy6∧dy4+q4dy3∧dy5∧dy6, (2.129) (2.129) and and and 1 2 × 123 εpmnrstuεqm′n′r′s′t′u′qmnrqstm′qn′r′uqs′t′u′ = q1q2q3q4δp 7δq 7 , (2.130) 1 2 × 123 εpmnrstuεqm′n′r′s′t′u′qmnrqstm′qn′r′uqs′t′u′ = q1q2q3q4δp 7δq 7 , (2.130) (2.130) such that the orbit (2.128) is determined by the sign of the eigenvalue of this rank one symmetric tensor, I4 = q1q2q3q4. The generic sum of two rank one charges takes the form JHEP07(2015)154 1 6qmnpdym ∧dyn ∧dyp = q1dy1 ∧dy2 ∧dy3 + q2dy4 ∧dy5 ∧dy6 , (2.131) (2.131) and is a generic solution to (2.127) associated to the second orbit (i.e. I4 < 0), and violates equation (2.126). Therefore we conﬁrm that a quadratic source in E(8,4,4) is in contradiction with the cubic equation satisﬁed by E(8,2,0), whereas it is consistent with the one satisﬁed by E(8,1,1). Let us now argue that all the invariants of the inﬁnite series of ¯F 2k∇4R4 do not get modiﬁed at the same order by lower order modiﬁcations to the supersymmetry transfor- mations. By power counting, the next order correction to the R4 type invariant and a ¯F 2k∇4R4 type invariant can in principle contribute to a right-hand-side for the classical supersymmetry variation of a ¯F 2k+6∇4R4 type invariant. So in principle one could expect that the function E (k) (8,2,0) satisﬁes to a Poisson equation of the kind = 3(k + 4)(k −5)E (k) (8,2,0) −a (k) 3 E(8,4,4) E (k−3) (8,2,0) −a (k) 5 E(8,2,2) E (k−5) (8,2,0) −a (k) 6 (E(8,4,4))2 E (k−6) (8,2,0) − k−8 X p=0 b(k) p E (k−8−p) (8,2,0) E (p) (8,2,0) + . . . (2.132) (2.132) However, the solutions to the diﬀerential equation (2.101) admit restricted Fourier modes in the string theory limit, satisfying to (2.117) for a vanishing NS5-brane charge. 2.7 String theory perturbation theory In order to deduce constraints on the contributions that can possibly appear in perturbative string theory, it is important to solve the diﬀerential equations satisﬁed by the threshold functions in the parabolic gauge with manifest T-duality symmetry (2.113). In this section we will solve these equations on an ansatz function depending only on the string theory dilaton e2φ and the scalar ﬁelds parametrising SO(6, 6)/(SO(6) × SO(6)). We have not computed explicitly the decomposition of the diﬀerential equations, but using the manifest – 24 – covariance, and the known solutions for the R4 and the ∇4R4 threshold functions [27], we can determine unambiguously all the unknown coeﬃcients. covariance, and the known solutions for the R4 and the ∇4R4 threshold functions [27], we can determine unambiguously all the unknown coeﬃcients. We deﬁne the covariant derivative Daˆb on SO(6, 6)/(SO(6) × SO(6)) in tangent frame, such that a = 1 to 6 of one SO(6) and ˆb = 1 to 6 of the other. It is convenient to deﬁne the covariant derivative as an SU(4) × SU(4) tensor Dijˆkˆl = 1 4γaijγ ˆbˆkˆlDaˆb , (2.133) (2.133) with i = 1, 4 of one SU(4) and ˆı = 1, 4 of the other. with i = 1, 4 of one SU(4) and ˆı = 1, 4 of the other. Calibrating the equations on the known solutions, one obtains that with i = 1, 4 of one SU(4) and ˆı = 1, 4 of the other. 2.7 String theory perturbation theory Calibrating the equations on the known solutions, one obtains that JHEP07(2015)154 D 3 56E = s2 −17 2 s + 6 D56E (2.134) (2.134) decomposes on R∗ + × SO(6, 6)/(SO(6) × SO(6)) as 1 64∂3 φ + 17 32∂2 φ + 3 2∂φ−DaˆbDaˆb δb a+ 3 4∂φ+6 DaˆcDbˆc E = s2−17 2 s+6 δb a 1 4∂φE , DaˆcDdˆcDdˆb + 3 16∂2 φ + 31 8 ∂φ + 9 Daˆb E = s2 −17 2 s + 6 DaˆbE , 8Dip ˆkˆqDpr ˆqˆsDjr ˆlˆs + 5 4∂φ + 2 2Dij ˆkˆl E = 2 s2 −17 2 s + 6 Dij ˆkˆlE , (2.135) whereas D 3 133E = s(s −9)D133E (2.136) D 3 133E = s(s −9)D133E gives the components in the 32 of Spin(6, 6) 1 64∂3 φ + 5 8∂2 φ −5 16∂φ −DpqˆrˆsDpqˆrˆs δk i δ ˆl ˆ+3(∂φ+6)DipˆˆqDkpˆlˆq E = s(s−9)δk i δ ˆl ˆ 1 4∂φE , 8DipˆˆqDprˆqˆsDkrˆlˆs + 3 16∂3 φ + 31 8 ∂φ 2Dikˆˆl E = 2s(s−9)DikˆˆlE. The ∇4R4 threshold function solves (2.134) for s = 3 2 and (2.137) for s = 4. One reads di- rectly from these equations, that a solution of type eaφED6 on R∗ +×SO(6, 6)/(SO(6)×SO(6)) must be such that ED6 satisﬁes to the quadratic equations in all fundamental representa- tions (i.e. the vector and Weyl spinor of positive and negative chirality), unless a = −6 or a = −8. The only other solutions are therefore such that ED6 is either a constant, or solves (B.24). One ﬁnds the unique solution e−10φ. For the values a = −8, the function ED6 satisﬁes to a quadratic equation in the spinor representation, and solve (B.30) for s = 4 (or 1 which is equivalent). For a = −6, ED6 satisﬁes to a quadratic equation in the vector representation, and cubic equations in the two spinor representations, and must therefore satisfy to (B.25). We ﬁnd therefore that supersymmetry and T-duality alone already deter- mine the ∇4R4 type corrections in perturbative string theory, up to three free coeﬃcients, – 25 – that are given in [27] 1 2Eh 0 5 2 00000 i = ζ(5)e−10φ + 4 15πe−8φE 0 40000 + 2 3e−6φE 2 00000 + O e−e−φ . 2.7 String theory perturbation theory (2.138) (2.138) This conﬁrms that supersymmetry alone already prevents any perturbative correction to the ∇4R4 threshold function beyond 2-loop in perturbative string theory. The functions deﬁning the ¯F 2k∇4R4 solve equation (2.137) for s = k + 4. Similarly one obtains the general SO(6, 6, Z) invariant solution (k) (8,2,0) = c (k) 1 e−2(k+4)φE 0 k+40000 + c (k) k+2e−6φE k+2 0000 0 + c (k) 2k e2(k−5)φE 0 k0000 + O e−e−φ . (2.139) JHEP07(2015)154 (2.139) It is quite remarkable that the only solutions we get all correspond to a strictly positive number of loops in perturbative string theory. After implementing the Weyl rescaling, one obtains indeed that c (k) ℓ is a coeﬃcient for a ℓ-loop correction in string theory for the ¯F 2k∇4R4 threshold function. For k = 1 and k = 2, equation (2.137) is exact for the Wilsonian eﬀective action (not taking into account linear corrections associated to logarithms in the complete eﬀective action). U-duality therefore implies that E(8,2,0) must be an Eisenstein function as in (2.102). Assuming that our argumentation in the preceding section is correct, and that equation (2.137) is satisﬁed for all k, we arrive at the conjecture that the ¯F 2k∇4R4 threshold function is deﬁned by the Eisenstein series E h 0 000004+k i for all k. It is rather remarkable that this coupling would only get three corrections in perturbation theory, at 1-loop, k + 2-loop and 2k-loop. It is quite remarkable that the only solutions we get all correspond to a strictly positive number of loops in perturbative string theory. After implementing the Weyl rescaling, one obtains indeed that c (k) ℓ is a coeﬃcient for a ℓ-loop correction in string theory for the ¯F 2k∇4R4 threshold function. For k = 1 and k = 2, equation (2.137) is exact for the Wilsonian eﬀective action (not taking into account linear corrections associated to logarithms in the complete eﬀective action). U-duality therefore implies that E(8,2,0) must be an Eisenstein function as in (2.102). Assuming that our argumentation in the preceding section is correct, and that equation (2.137) is satisﬁed for all k, we arrive at the conjecture that the ¯F 2k∇4R4 threshold function is deﬁned by the Eisenstein series E h 0 000004+k i for all k. 2.7 String theory perturbation theory It is rather remarkable that this coupling would only get three corrections in perturbation theory, at 1-loop, k + 2-loop and 2k-loop. This Eisenstein function diverges precisely for k = 1, corresponding to the ¯F 2∇4R4 threshold related to the ∇6R4 threshold function by supersymmetry. One must therefore consider the regularised Eisenstein series 16 63 ˆEh 0 000005 i = 16 63e−10φ ˆE 0 50000 −15ζ(5) 4 φ e−10φ+ 1 πe−8φ 2φE 0 40000 −∂sE 0 s0000 s=4 + π 9 e−6φ ˆE 3 00000 + O e−e−φ . (2.140) Here we have ﬁxed all the coeﬃcients by consistency with (2.77) and (2.138). The logarithm of the dilaton indicates a divergence of the ∇4R4 form factor into ∇6R4 in supergravity. Note nonetheless that the 3-loop contribution in the last line violate T-duality parity in O(6, 6, Z), and the string theory eﬀective action must include the same function with oppo- site chirality. Because it is a three-loop contribution, it cannot come from the completion of the R4 type invariant and it must appear as a solution to equation (2.134) for s = 6. Considering the general SO(6, 6, Z) invariant solution of (2.134), one ﬁnds indeed Considering the general SO(6, 6, Z) invariant solution of (2.134), one ﬁnds E(8,1,1) = c-6e−4sφ + c- 1 2 e−(2s+1)φE 0 0000s-1 2 + c1e2(2s−17)φ + c2e−8φE 0 0s-2000 + c3e2(s−9)φE 0 0000s-3 (2.141) (2.141) where the coeﬃcients cℓare constants that would correspond to ℓ-loop contributions for the ∇6R4 threshold function in string theory. Note that the ﬁrst two terms do not make – 26 – sense in perturbative string theory. The corresponding Eisenstein function E h 0 s00000 i includes generically all these terms, and therefore cannot deﬁne the string theory threshold function, consistently with the property that (2.134) is corrected by a source term (2.107). However, the three-loop contribution is not aﬀected by the source term, and one can take seriously the last contribution, which is precisely the one required to restore O(6, 6, Z) invariance for s = 6. 2.7 String theory perturbation theory This is indeed conﬁrmed by the expression obtained in [33] for the ∇6R4 threshold function, and using these results we conclude therefore that the exact threshold function for the ∇6R4 coupling is deﬁned as JHEP07(2015)154 E(0,1) = ˆE(8,1,1) + 32 189π ˆEh 0 000005 i , (2.142) (2.142) where the function E(8,1,1) solve the diﬀerential equation where the function E(8,1,1) solve the diﬀerential equation ∆ˆE(8,1,1) = −60 ˆE(8,1,1) − Eh 0  00000 i 2 + 35 π 1 2Eh 0  00000 i (2.143) DijpqDpqmnDmnkl ˆE(8,1,1) = −9Dijkl ˆE(8,1,1) −1 2Eh 0  00000 iDijklEh 0  00000 i + 35 4πDijkl 1 2Eh 0  00000 i , DijpqDpqmnDmnkl ˆE(8,1,1) = −9Dijkl ˆE(8,1,1) −1 2Eh 0  00000 iDijklEh 0  00000 i + 35 4πDijkl 1 2Eh 0  00000 i . (2.143) Here the anomalous right-hand-side is determined such as to coincide with the one obtained in [33] for the complete function E(0,1). These coeﬃcients can also be directly computed from the properties of the Eisenstein functions and the structure of the diﬀerential equa- tions [53]. 3 Supergravity in higher dimensions In this section we will consider the extension of the results of the preceding section in ﬁve, six, seven and eight dimensions. We will see that the two ∇6R4 type invariants both lift to higher dimensions, even if they cannot be deﬁned as harmonic superspace integrals in the linearised approximation in general. 3.1.1 Linearised ∇6R4 type invariants 3.1.1 Linearised ∇6R4 type invariants In ﬁve dimensions there is only one kind of 1/8 BPS harmonic superspace integral that one can deﬁne [41]. For this purpose, one considers Sp(4)/(U(1) × Sp(3)) harmonic variables (u1i, uri, u8i) with r = 2, ..., 7 in the fundamental of Sp(3), and the decomposition sp(4) ∼= 6(−1) ⊕(u(1) ⊕sp(3)) (0) ⊕6(1) 42 ∼= 14 (−1) 3 ⊕14 (0) 2 ⊕14 (1) 3 . (3.1) (3.1) One deﬁnes the G-analytic superﬁeld W rst in the [0, 0, 1] of Sp(3) W rst = u1iurjusjutkLijkl , (3.2) One deﬁnes the G-analytic superﬁeld W rst in the [0, 0, 1] of Sp(3) W rst = u1iurjusjutkLijkl , (3.2) One deﬁnes the G-analytic superﬁeld W rst in the [0, 0, 1] of Sp(3) W rst = u1iurjusjutkLijkl , (3.2) JHEP07(2015)154 which satisﬁes the constraint which satisﬁes the constraint which satisﬁes the constraint u1iDi αW rst = 0 . (3.3) (3.3) Following the same reasoning as in section 2.1, we consider a general monomial of W rst in an irreducible representation of Sp(3). In this case we obtain equivalently that the monomials are freely generated by W rst in the [0, 0, 1], the elements W rtpW sqrΩtqΩpr , (3.4) (3.4) in the [2, 0, 0], in the [2, 0, 0], W rpqW supW tqu , (3.5) (3.5) in the [0, 0, 1], and W r]tuW [stuW p]vwW [qvw −“symp trace” , W rstW pqrW uspW tqu , (3.6) W r]tuW [stuW p]vwW [qvw −“symp trace” , W rstW pqrW uspW tqu , (3.6) respectively in the [0 2 0] and the singlet representation The general linearised invariant W r]tuW [stuW p]vwW [qvw −“symp trace” , W rstW pqrW uspW tqu , (3.6) respectively in the [0, 2, 0] and the singlet representation. The general linearised invariant t k th f th f (3.6) respectively in the [0, 2, 0] and the singlet representation. The general linearised invariant takes therefore the form respectively in the [0, 2, 0] and the singlet representation. The general linearised invariant takes therefore the form Z duD28F(u)[2n2,2n4,n1+n3] [2n3+2n4+4n′ 4,2n2,2n4,n1+n3]W 4+n1+2n2+3n3+4n4+4n′ 4 [2n2,2n4,n1+n3] = Ln1+2n2+3n3+4n4+4n′ 4 [2n3+2n4+4n′ 4,2n2,2n4,n1+n3] ∇6R4 + . . . + . . . 3.1 N = 4 supergravity in ﬁve dimensions Let us recall in a ﬁrst place some properties of maximal supergravity in ﬁve dimensions. The scalar ﬁelds parametrise the symmetric space E6(6)/Sp(4)c. We use i, j = 1, ..., 8 as indices in the fundamental representation of Sp(4), and Ωij deﬁnes the symplectic form with the normalisation ΩikΩjk = δi j. The covariant derivative in tangent frame Dijkl is a symplectic traceless rank four antisymmetric tensor in the representation [0, 0, 0, 1] of Sp(4). – 27 – 3.1.1 Linearised ∇6R4 type invariants (3.7) (3.7) The structure of these linearised invariants suggests that the complete non-linear invariant admits the following gradient expansion The structure of these linearised invariants suggests that the complete non-linear invariant admits the following gradient expansion L(4,1)[E(4,1)] = X n1,n2,n3,n4,n′ 4 Dn1+2n2+3n3+4n4+4n′ 4 [2n3+2n4+4n′ 4,2n2,2n4,n1+n3]E(4,1)L[2n3+2n4+4n′ 4,2n2,2n4,n1+n3] (4,1) . (3 (3.8) The consistency of this ansatz requires that the function E(4,1) must be an eigenfunction of the Laplace operator, and that its third order derivative restricted to the [0, 2, 0, 0] is proportional to its second derivative in the same representation. This linearised analysis is consistent with the one of the (8, 1, 1) type invariant in four dimensions, and we are going to see that the relevant equation is The consistency of this ansatz requires that the function E(4,1) must be an eigenfunction of the Laplace operator, and that its third order derivative restricted to the [0, 2, 0, 0] is proportional to its second derivative in the same representation. This linearised analysis is consistent with the one of the (8, 1, 1) type invariant in four dimensions, and we are going to see that the relevant equation is DijpqDpqrsDrsklE(4,1) = 1 4Dijkl(34 + ∆) E(4,1) . (3.9) (3.9) However, the (8, 2, 0) type invariants cannot be deﬁned in the linearised approximation through a harmonic superspace integral, and we shall instead consider the uplift of the general invariant to ﬁve dimensions. However, the (8, 2, 0) type invariants cannot be deﬁned in the linearised approximation through a harmonic superspace integral, and we shall instead consider the uplift of the general invariant to ﬁve dimensions. – 28 – 3.1.2 Decompactiﬁcation limit from four to ﬁve dimensions We are therefore going to solve the diﬀerential equations (2.40) and (2.143) for a function depending only of the Levi subgroup R+ ∗× E6(6) of the parabolic subgroup associated to the decompactiﬁcation limit, such that e7(7) ∼= 27 (−2) ⊕ gl1 ⊕e6(6) (0) ⊕27(2) , 56 ∼= 1(−3) ⊕27(−1) ⊕27 (1) ⊕1(3) . (3.10) (3.10) r this purpose we use the same conventions as in [30], such that the coset representative E7(7)/SU(8)c is deﬁned as r this purpose we use the same conventions as in [30], such that the coset representative E7(7)/SU(8)c is deﬁned as V =      e3φ 0 0 0 0 eφVijI 0 0 0 0 e−φV -1Iij 0 0 0 0 e−3φ           1 aJ 1 2tJKLaKaL 1 3tKLP aKaLaP 0 δJ I tIJKaK 1 2tIKLaKaL 0 0 δI J aI 0 0 0 1     , (3.11) JHEP07(2015)154 (3.11) where VijI is the coset representative in E6(6) with the Sp(4) pair ij being antisymmetric symplectic traceless and the index I in the fundamental of E6(6). tIJK is the E6(6) invariant symmetric tensor normalised as in [30]. We have already computed the decomposition of the cubic equation (2.21) in [30], which is 1 64∂3 φ + 21 32∂2 φ + 9 2∂φ −3 4∆ E(8,1,1) = −1 4∂φ 9E(8,1,1) + 1 4E 2 (8,4,4) (3.12) DijpqDpqrsDrskl + Dijkl 1 48∂2 φ + 27 24∂φ + 7 2 +DijpqDklpq 1 4∂φ + 3 +δkl ij 1 123 ∂3 φ + 5 96∂2 φ + 1 6∂φ−1 4∆ ! E(8,1,1) = − 1 12δkl ij ∂φ+Dijkl 9E(8,1,1)+ 1 4E 2 (8,4,4) , (3.12) where δkl ij = δ[k [i δl] j] −1 8ΩijΩkl , ∆= 1 3DijklDijkl , (3.13) (3.13) and indices are raised and lowered with the symplectic matrix Ωij. Because of the Weyl rescaling required to stay in Einstein frame, the relevant radius power in the decompactiﬁ- cation limit for a ∇2nR4 threshold function is such that EE7 = e−(6+2n)φEE6, and because we are interested in the constraint on the ∇6R4 threshold function, we use the ansatz E(8,1,1) = e−12φE(8,1) , E(8,4,4) = e−6φE(8,4) , (3.14) (3.14) where E(8,1) and E(8,4) are functions on E6(6)/Sp(4)c. Using this ansatz, one derives where E(8,1) and E(8,4) are functions on E6(6)/Sp(4)c. 2 N = (2, 2) supergravity in six dimensions We shall now discuss these invariants in N = (2, 2) supergravity in six dimensions. We recall that the scalar ﬁelds parametrise in this case the symmetric space SO(5, 5)/(SO(5)× SO(5)). 3.1.2 Decompactiﬁcation limit from four to ﬁve dimensions We therefore obtain that the threshold function is the regularised Eisenstein series ˆE (1) 1 8 = 5 108 ˆEh 0 0000 9 2 i , (3.19) (3.19) such that the exact ∇6R4 threshold function E(0,1) is E(0,1) = ˆE(8,1) + 5 108 ˆEh 0 0000 9 2 i . (3.20) (3.20) The series E h 0 0000s i admits a pole at s = 9 2 proportional to the series E h 0 3 2 0000 i deﬁning the R4 threshold, exhibiting that the R4 invariant form factor diverges at two loop into the ∇6R4 form factor associated to the same function. This is in agreement with [33], where the explicit coeﬃcient is computed. 3.1.2 Decompactiﬁcation limit from four to ﬁve dimensions Using this ansatz, one derives where E(8,1) and E(8,4) are functions on E6(6)/Sp(4)c. Using this ansatz, one derives ∆E(8,1) = −18E(8,1) −E 2 (8,4) , (3.15) (3.15) and DijpqDpqrsDrskl + 2Dijkl E(8,1) = −1 4DijklE 2 (8,4) . (3.16) (3.16) These equations are satisﬁed by the 1/8 BPS threshold functions in the Wilsonian eﬀective action, but the U-duality invariant function appearing in the 1PI eﬀective action satisﬁed to anomalous equations with additional terms linear in E in the right hand side [33] These equations are satisﬁed by the 1/8 BPS threshold functions in the Wilsonian eﬀective action, but the U-duality invariant function appearing in the 1PI eﬀective action satisﬁed to anomalous equations with additional terms linear in E(8,4) in the right-hand-side [33]. alous equations with additional terms linear in E(8,4) in the right-hand-side [33]. – 29 – We shall now consider the uplift of the F 2k∇4R4 type invariants, but for this purpose it will be more convenient to consider directly the decompactiﬁcation limit of the Eisenstein function E h 0 00000k+4 i. We shall only consider the term with the correct power of the com- pactiﬁcation radius r to lift to a diﬀeomorphism invariant in ﬁve dimensions, as computed in appendix B.1, Z d4x √−g Eh 0 00000k+4 i ∇4+2kR4 →π 1 2 Γ(k + 7 2) Γ(k + 4) Z d5x √−g Eh 0 0000k+ 7 2 i∇4+2kR4 . (3.17) (3.17) We conclude in this way that the threshold function E (k) 1 8 deﬁning the F 2k∇4R4 type in- variants satisﬁes to JHEP07(2015)154 JHEP07(2015)154 DijpqDpqrsDrskl−2k(k+1)−10 3 Dijkl E (k) 1 8 = k+2 2 DijpqDklpq−2(2k−5)(2k+7) 27 δkl ij E (k) 1 8 DijpqDpqrsDrskl−2k(k+1)−10 3 Dijkl E (k) 1 8 = k+2 2 DijpqDklpq−2(2k−5)(2k+7) 27 δkl ij E (k) 1 8 ∆E (k) 1 8 = 2 3(2k −5)(2k + 7)E (k) 1 8 , D3 [2,0,0,1]E (k) 1 8 = 0 . (3.18) ∆E (k) 1 8 = 2 3(2k −5)(2k + 7)E (k) 1 8 , D3 [2,0,0,1]E (k) 1 8 = 0 . (3.18) (3.18) It follows from representation theory that such equations are indeed implied by (2.136), and this explicit example permits to determine them uniquely. 3.2.1 Linearised invariant In the linearised approximation, the theory is deﬁned from the scalar superﬁeld Lijˆıˆin the [0, 1] × [0, 1] of Sp(2) × Sp(2), where i, j and ˆı, ˆrun from 1 to 4 in the fundamental of the two respective Sp(2). One can deﬁne a ∇6R4 type invariant by considering harmonic variables u1i, uri, u4i parametrising Sp(2)/(U(1) × Sp(1)) associated to one Sp(2) factor, with r = 2, 3 of Sp(1), such that sp(2) ∼= 2(−1) ⊕(u(1) ⊕sp(1)) (0) ⊕2(1) , 4 ∼= 1(−1) ⊕2(0) ⊕1(1) . (3.21) (3.21) – 30 – One can in this way introduce the G-analytic superﬁeld [41] One can in this way introduce the G-analytic superﬁeld [41] W rˆıˆ= u1iurjLijˆıˆ, (3.22) (3.22) that transforms in the fundamental of Sp(1) and as a vector of SO(5) ∼= Sp(2)/Z2, and satisﬁes to the 1/8 BPS G-analyticity constraint that transforms in the fundamental of Sp(1) and as a vector of SO(5) ∼= Sp(2)/Z2, and satisﬁes to the 1/8 BPS G-analyticity constraint u1iDi αW r,ˆıˆ= 0 . (3.23) (3.23) A general polynomial in W r,ˆıˆdecomposes into irreducible representations of Sp(1)×Sp(2). Similarly as in lower dimensions, one shows that the latter are freely generated by W r,ˆıˆ itself in the [1] × [0, 1] of SU(2) × Sp(2), the two quadratic monomials JHEP07(2015)154 W rˆıˆW sˆıˆ, W rˆıˆkWrˆˆk , (3.24) W rˆıˆW sˆıˆ, W rˆıˆkWrˆˆk , (3.24) in the [2] and the [2, 0], respectively, the cubic monomial W sˆıˆW rˆkˆlWsˆkˆl , (3.25) (3.25) in the [1] × [0, 1], and the two quartic monomials W sˆıˆpWsˆˆpW pˆk ˆqWpˆlˆq −1 6δˆıˆ ˆkˆlW sˆıˆpWsˆˆpW pˆıˆqWpˆˆq , W sˆıˆpWsˆˆpW pˆıˆqWpˆˆq , (3.26) (3.26) in the [0, 2] and the singlet representation, respectively. One concludes that the most general monomial is labeled by 6 integers, such that in the [0, 2] and the singlet representation, respectively. One concludes that the most general monomial is labeled by 6 integers, such that Z duD12 ¯D16F(u)[n1+2n2+n3][2n′ 2,n1+n3+2n4] [2n′ 2+2n3+4n4+4n′ 4,n1+2n2+n3][2n′ 2,n1+n3+2n4]W 4+n1+2n2+2n′ 2+3n3+4n4+4n′ 4 [n1+2n2+n3][2n′ 2,n1+n3+2n4] Z duD12 ¯D16F(u)[n1+2n2+n3][2n′ 2,n1+n3+2n4] [2n′ 2+2n3+4n4+4n′ 4,n1+2n2+n3][2n′ 2,n1+n3+2n4]W 4+n1+2n2+2n′ 2+3n3+4n4+4n′ 4 [n1+2n2+n3][2n′ 2,n1+n3+2n4] = Ln1+2n2+2n′ 2+3n3+4n4+4n′ 4 [2n′ 2+2n3+4n4+4n′ 4,n1+2n2+n3][2n′ 2,n1+n3+2n4]∇6R4 + . . . (3.27) Z = Ln1+2n2+2n′ 2+3n3+4n4+4n′ 4 [2n′ 2+2n3+4n4+4n′ 4,n1+2n2+n3][2n′ 2,n1+n3+2n4]∇6R4 + . . . 3.2.1 Linearised invariant (3.2 (3.27) The linear analysis therefore suggests the form of the nonlinear invariant L(4,1,0)[E(4,1,0)] (3.28) = X n1,n2,n′ 2 n3,n4,n′ 4 Dn1+2n2+2n′ 2+3n3+4n4+4n′ 4 [2n′ 2+2n3+4n4+4n′ 4,n1+2n2+n3][2n′ 2,n1+n3+2n4]E(4,1,0)L [2n′ 2+2n3+4n4+4n′ 4,n1+2n2+n3][2n′ 2,n1+n3+2n4] (4,1,0) X +2 +2 ′ +3 +4 +4 ′ [ ′ ][ ′ ′ ] (3.28) L(4,1,0)[E(4,1,0)] L(4,1,0)[E(4,1,0)] (3.28) = X n1,n2,n′ 2 n3,n4,n′ 4 Dn1+2n2+2n′ 2+3n3+4n4+4n′ 4 [2n′ 2+2n3+4n4+4n′ 4,n1+2n2+n3][2n′ 2,n1+n3+2n4]E(4,1,0)L [2n′ 2+2n3+4n4+4n′ 4,n1+2n2+n3][2n′ 2,n1+n3+2n4] (4,1,0) + X n1,n2,n′ 2 n3,n4,n′ 4 Dn1+2n2+2n′ 2+3n3+4n4+4n′ 4 [2n′ 2,n1+n3+2n4][2n′ 2+2n3+4n4+4n′ 4,n1+2n2+n3]E(4,1,0)L [2n′ 2,n1+n3+2n4][2n′ 2+2n3+4n4+4n′ 4,n1+2n2+n3] (4,1,0) L(4,1,0)[E(4,1,0)] (3.2 = X n1 n2 n′ 2 Dn1+2n2+2n′ 2+3n3+4n4+4n′ 4 [2n′ 2+2n3+4n4+4n′ 4,n1+2n2+n3][2n′ 2,n1+n3+2n4]E(4,1,0)L [2n′ 2+2n3+4n4+4n′ 4,n1+2n2+n3][2n′ 2,n1+n3+2n4] (4,1,0) = X Dn1+2n2+2n′ 2+3n3+4n4+4n′ 4 ′ ′ 3, 4, 4 + X n1,n2,n′ 2 n3,n4,n′ 4 Dn1+2n2+2n′ 2+3n3+4n4+4n′ 4 [2n′ 2,n1+n3+2n4][2n′ 2+2n3+4n4+4n′ 4,n1+2n2+n3]E(4,1,0)L [2n′ 2,n1+n3+2n4][2n′ 2+2n3+4n4+4n′ 4,n1+2n2+n3] (4,1,0) + X n1,n2,n′ 2 n3,n4,n′ 4 Dn1+2n2+2n′ 2+3n3+4n4+4n′ 4 [2n′ 2,n1+n3+2n4][2n′ 2+2n3+4n4+4n′ 4,n1+2n2+n3]E(4,1,0)L [2n′ 2,n1+n3+2n4][2n′ 2+2n3+4n4+4n′ 4,n1+2n2+n3] (4,1,0) where we consider the possibility of a mixing between the invariant L(4,1,0) with its conju- gate obtained by exchanging the two Sp(2) factors, according to the observation in [31] for the ∇4R4 type invariant in eight dimensions. where we consider the possibility of a mixing between the invariant L(4,1,0) with its conju- gate obtained by exchanging the two Sp(2) factors, according to the observation in [31] for the ∇4R4 type invariant in eight dimensions. From this structure one deduces that supersymmetry requires the function E(4,1,0) to be an eigenfunction of the Laplace operator, and to satisfy equations of the form 1,0) , D3 [0,1],[0,1]E(4,1,0) ∝D[0,1],[0,1]E(4,1,0) , (3.29) D3 [2,0],[2,0]E(4,1,0) ∝D 2 [2,0],[2,0]E(4,1,0) , D3 [0,1],[0,1]E(4,1,0) ∝D[0,1],[0,1]E(4,1,0) , (3.29) (3.29) as well has a highest weight constraint as well has a highest weight constraint D2k [0,0],[0,2k]E(4,1,0) = 0 , (3.30) (3.30) for some integer k. We will see in the next section that the standard ∇6R4 type invariant threshold function indeed satisﬁes to these equations for k = 2. for some integer k. We will see in the next section that the standard ∇6R4 type invariant threshold function indeed satisﬁes to these equations for k = 2. – 31 – – 31 – 3.2.2 Decompactiﬁcation limit from ﬁve to six dimensions We are now going to solve the diﬀerential equations (3.16) and (3.18) for a function de- pending only of the Levi subgroup R+ ∗× SO(5, 5) of the parabolic subgroup associated to the decompactiﬁcation limit, such that e6(6) ∼= 16(−3) ⊕(gl1 ⊕so(5, 5)) (0) ⊕16 (3) , 27 ∼= 10(−2) ⊕16(1) ⊕1(4) . (3.31) (3.31) The covariant derivative on E6(6)/Spc(4) acting on such a function takes the block diago- nal form 1 1 1 D27 = diag 1 6∂φ, D16 + 1 24116∂φ, D10 −1 12110∂φ . (3.32) (3.32) JHEP07(2015)154 To check the diﬀerential equations (3.16) and (3.18) we need to compute the block diagonal decomposition of the higher order diﬀerential operators. In order to do this computation we consider a general ansatz and determine all the free coeﬃcients by consistency with the various diﬀerential equations displayed in appendix A. We obtain in this way D 2 27 = diag 1 62 ∂2 φ + 1 2∂φ , D 2 16 + 1 2D16 1 6∂φ + 1 + 1 16116 1 62 ∂2 φ + 3∂φ , D 2 10 −D10 1 6∂φ + 1 + 1 4110 1 62 ∂2 φ + ∂φ (3.33 (3.33) and D 3 27 = diag 1 63 ∂3 φ + 3 16∂2 φ + 5 4∂φ −1 2∆, D 3 16 + 3 4D 2 16 1 6∂φ+2 +D16 3 16 1 62 ∂2 φ + 2∂φ +1 + 1 64116 1 63 ∂3 φ + 1 2∂2 φ −8∆ , D 3 10 −3 2D 2 10 1 6∂φ+2 +D10 3 4 1 62 ∂2 φ + ∂φ + 5 2 −1 8110 1 63 ∂3 φ + 1 4∂2 φ +∂φ−2∆ , (3.34) where ∆≡Tr D 2 10. In order to determine the constraints on the threshold function is six dimensions, we consider an ansatz with the appropriate power of the radius modulus e−3φ such as to compensate for the Weyl rescaling to Einstein frame, i.e. E(8,4) = e−6φE(4,2,2) , E(8,1) = e−12φE(4,1,0) . (3.35) (3.35) The singlet component of (3.16) gives directly the Poisson equation The singlet component of (3.16) gives directly the Poisson equation ∆E(4,1,0) = −E 2 (4,2,2) , (3.36) (3.36) which is indeed consistent with (3.14). 3.2.2 Decompactiﬁcation limit from ﬁve to six dimensions Working out spinor and the vector equations, using the Poisson equation (3.36), one obtains similarly which is indeed consistent with (3.14). Working out spinor and the vector equations, using the Poisson equation (3.36), one obtains similarly D 3 16 −3 4D16 E(4,1,0) = −1 4D16E 2 (4,2,2) , D 3 10 −3 2D10 E(4,1,0) = −1 4D10E 2 (4,2,2). (3.37) (3.37) – 32 – – 32 – The only Eisenstein function that solves this homogenous equation (for E(4,2,2) = 0) is the Eisenstein series ˆE 0 0 7 2 00 , but its expansion in the string theory limit is inconsistent with perturbation theory. As is computed in appendix B.3, one has moreover Ddˆa D(a ˆbDb\|ˆbDcˆcDd)ˆc \|[0,4] ˆE 0 0 7 2 00 = 0 , (3.38) (3.38) which deﬁnes an integrability condition for the function to decompose into the sum of two functions satisfying to (3.30) and its conjugate obtained by exchange of the two Sp(2) for k = 2. Let us now consider the diﬀerential equation (3.18) for the function E(k) 1 8 E6 deﬁning the F 2k∇4R4 threshold function in ﬁve dimensions. Diﬀeomorphism invariance in six dimensions requires an ansatz of the form JHEP07(2015)154 E(k) 1 8 E6 = e−2(k+5)φE(k) 1 8 D5 , (3.39) (3.39) and using this ansatz, one obtains from the singlet component of (3.18) the Laplace equa- tion 5 ∆E(k) 1 8 = 5 2(k + 3)(k −1)E(k) 1 8 , (3.40) (3.40) here we removed the D5 label for simplicity. The spinor and the vector equations give then D 3 16 −13(k + 3)(k −1) + 24 16 D16 E(k) 1 8 = −3(k + 1) 4 D 2 16 −5(k + 3)(k −1) 16 116 E(k) 1 8 , D 2 10 E(k) 1 8 = (k + 3)(k −1) 4 110 E(k) 1 8 , (3.41) (3.41) where we used that the even and odd powers of D10 lie in diﬀerent irreducible represen- tations of Sp(2) × Sp(2), and must therefore vanish separately. The unique Eisenstein function satisfying to this equation is E(k) 1 8 ∝E 0 000k+3 , (3.42) (3.42) consistently with the decompactiﬁcation limit of the ﬁve-dimensional function Eh 0 0000k+ 7 2 i = 2ζ(2k + 7)e−4(2k+7)φ + π 1 2 Γ(k + 3) Γ(k + 7 2) e−(2k+10)φE 0 000k+3 + . 3.2.2 Decompactiﬁcation limit from ﬁve to six dimensions . . (3.43) (3.43) We conclude therefore that the exact ∇6R4 function is deﬁned as E(0,1) = ˆE(4,1,0) + ˆE(4,0,1) + 8 189 ˆE 0 0004 , (3.44) (3.44) where ˆE(4,1,0) satisﬁes to (3.37), and to an anomalous Poisson equation with an additional constant source term. This function is consistent with [33], where the second Eisenstein function appears with this normalisation, and the 2-loop ﬁve dimensional threshold function must indeed solve (3.37), because the equation is parity invariant with respect to O(5, 5). – 33 – 3.3 N = 2 supergravity in seven dimensions 3.3 N = 2 supergravity in seven dimensions None of the ∇6R4 type invariants can be deﬁned in the linearised approximation as har- monic superspace integrals in seven dimensions. We will therefore consider the uplift of the four-dimensional invariants in the decompactiﬁcation limit. In seven dimensions the scalar ﬁelds parametrize the symmetric space SL(5)/SO(5), and the covariant derivative Dij transforms as a symmetric traceless tensor of SO(5), with i, j = 1, . . . , 5 of SO(5). We con- sider therefore the parabolic subgroup of E7(7) of semi-simple Levi subgroup SL(5) ×SL(3) associated to the decomposition None of the ∇6R4 type invariants can be deﬁned in the linearised approximation as har- monic superspace integrals in seven dimensions. We will therefore consider the uplift of the four-dimensional invariants in the decompactiﬁcation limit. In seven dimensions the scalar ﬁelds parametrize the symmetric space SL(5)/SO(5), and the covariant derivative Dij transforms as a symmetric traceless tensor of SO(5), with i, j = 1, . . . , 5 of SO(5). We con- sider therefore the parabolic subgroup of E7(7) of semi-simple Levi subgroup SL(5) ×SL(3) associated to the decomposition e7(7) ∼= ¯5(−6)⊕(3 ⊗5)(−4)⊕(¯3 ⊗10)(−2)⊕(gl1⊕sl3⊕sl5) (0)⊕(3 ⊗10)(2) ⊕(¯3 ⊗¯5)(4) ⊕5(6) 56 ∼= 3 (−5) ⊕10(−3) ⊕(3 ⊗5)(−1) ⊕(3 ⊗5)(1) ⊕10 (3) ⊕3(5) . (3.45) JHEP07(2015)154 (3.45) We will use the same conventions as in [31], where the decompactiﬁcation limit of equa- tions (2.40) and (2.143) is already discussed in details. We consider the ansatz E(8,1,1) = e−36φE 1 8 , E(8,4,4) = e−18φE(4,2) , (3.46) (3.46) with E 1 8 and E(4,2) deﬁned on SL(5)/SO(5), and the appropriate power of the volume mod- ulus e−3φ required by diﬀeomorphism invariance in seven dimensions. The 3(3) component of the equation reduces to the Poisson equation ∆E 1 8 ≡2DijDijE 1 8 = 42 5 E 1 8 −E 2 (4,2) . (3.47) (3.47) Using this equation in the (3 ⊗5)(1) component of (2.143), one obtains DikDklDlj + 1 5DikDkj −1053 400 Dij + 177 500δj i E 1 8 = 1 4 1 10δj i −Dij E 2 (4,2) . (3.48) (3.48) Using moreover these equations to simplify the 10 (3) components, one obtains 6D[i [kDj]pDp l] + 3 10D[i [kDj] l] E 1 8 = 3 5δ[i [k Dj]pDp l] −71 20Dj] l] + 9 25δj] l] E 1 8 . 3.3 N = 2 supergravity in seven dimensions (3.49) (3.49) he solution to the homogenous equation (with E(4,2) = 0) can be written as the Eisenstein The solution to the homogenous equation (with E(4,2) = 0) can be written as the Eisenstein function E[3,0,0, 5 2 ], using the formulae of appendix A.5. [ , , , 2 ] We consider now the F 2k∇4R4 threshold function, with the ansatz E (k) (8,2,0) = e−6(k+5)φE (k) 1 8 . (3.50) (3.50) Appropriate linear combinations of the grad 6 and 4 components of (2.40) in (3.45) give the two equations DikDklDlj+ 1 2DikDkj E (k) 1 8 = 16k(7k+20)+75 400 Dij+ 3k(k+5)(2k+5) 125 δj i −1 40δj i ∆ E (k) 1 8 k DikDkjE (k) 1 8 = k4k + 5 20 Dij + 3k(k + 5)(2k + 5) 25 δj i −1 2δj i ∆ E (k) 1 8 . (3.51) (3.51) – 34 – – 34 – For k ≥1 we conclude that the function must satisfy to the quadratic equation DipDpjE (k) 1 8 = 4k + 5 20 DijE (k) 1 8 + 3k(2k + 5) 25 δj i E (k) 1 8 , (3.52) (3.52) such that such that ∆E (k) 1 8 = 6k(2k + 5) 5 E (k) 1 8 . (3.53) (3.53) One can then check that all the other equations implied by (2.40) are indeed satisﬁed provided that (3.52) is. This equation is satisﬁed by the Eisenstein function E[0,0,k+ 5 2 ,0], which appears in the decompactiﬁcation limit of the corresponding Eisenstein function on E7(7)/SUc(8), i.e. JHEP07(2015)154 Eh 0 00000k+4 i = e−10(k+4)φE[k+4,0] + π 3 2 Γ k + 5 2 Γ(k + 4) e−6(k+5)φE[0,0,k+ 5 2 ,0] + . . . (3.54) (3.54) where the Eisenstein series is normalised with an extra 2ζ(2s) factor with respect to the Langlands normalisation. We conclude therefore that the exact ∇6R4 threshold function is E(0,1) = E 1 8 + 5π 378E[0,0, 7 2 ,0] , (3.55) (3.55) where E 1 8 is a solution to (3.47), (3.48), (3.49) in agreement with [13].4 4Note that the normalisation of the Eisenstein function E[0,0,s,0] does not include the additional factor of ζ(2s −1) as in [13]. 3.4 N = 2 supergravity in eight dimensions (3.58) (3.58) One computes that the higher order derivative operators obtained as products of D5 in the 5 of SL(5) decompose similarly as D 2 5 = diag Dαβ 1 10∂φ + 3 4 + 1 400∂2 φ + 1 16∂φ + 1 2DγδDγδ δβ α, DacDcb + Dab 1 15∂φ + 1 2 + 1 900∂2 φ + 1 24∂φ δb a , (3.59) JHEP07(2015)154 (3.59) and and D 3 5 = diag −1 4Dαβ 3 102 ∂2 φ + 7 10∂φ + 9 4 + 2DγδDγδ −1 8 1 103 ∂3 φ + 1 30∂2 φ + 1 8∂φ + 6DγδDγδ 1 10∂φ + 1 −2DabDab δβ α , DacDcdDdb + DacDcb 1 10∂φ + 1 + 1 3Dab 1 102 ∂2 φ + 11 40∂φ + 3 4 + 1 8 1 153 ∂3 φ + 1 180∂2 φ −1 12∂φ −2DαβDαβ δb a . (3.60 DacDcdDdb + DacDcb 1 10∂φ + 1 + 1 3Dab 1 102 ∂2 φ + 11 40∂φ + 3 4 + 1 8 1 153 ∂3 φ + 1 180∂2 φ −1 12∂φ −2DαβDαβ δb a . (3.60) + 1 8 1 153 ∂3 φ + 1 180∂2 φ −1 12∂φ −2DαβDαβ δb a . (3.60) (3.60) We can now solve equation (3.48) in the docompactiﬁcation limit, with E 1 8 (φ, τ, t) and [1 E(4,2) = e−6φ 2 ˆE[1](τ) + ˆE[ 3 2 ,0](t) −20π(φ −φ0) , (3.61) (3.61) φ0 is a constant that depends on the renormalisation scheme. 3.4 N = 2 supergravity in eight dimensions We shall now consider the oxidation of the seven-dimensional ∇6R4 and F 2n∇4R4 type invariants to eight dimensions. Because there is a 1-loop divergence in eight dimensions, the exact R4 threshold function diﬀers from the Wilsonian eﬀective action function. In the dimensional reduction, the divergence appears to be absorbed into the inﬁnite sum of Kaluza-Klein states over the circle such that the function is ﬁnite in seven dimensions, but involves a logarithm of the radius modulus in the decompactiﬁcation limit [42]. In order to consider the non-analytic terms in eight dimensions, we will take these logarithms into account in the decompactiﬁcation limit. We shall use the same conventions as in [31], i.e. the complex scalar ﬁeld τ parametrises the coset representative vαj ∈SL(2)/SO(2), with α, β = 1, 2 of SO(2) and i, j = 1, 2 of SL(2), whereas the ﬁve real scalar ﬁelds t parametrise the coset representative V aJ ∈ SL(3)/SO(3), with a, b = 1, 2, 3 of SO(3) and I, J = 1, 2, 3 of SL(3). The corresponding covariant derivative in tangent frame are then traceless symmetric tensors Dαβ and Dab, respectively. In the decompactiﬁcation limit, one writes the SL(5)/SO(5) coset element in the parabolic gauge V = e−3φv-1 j α 0 e2φV aKaK j e2φV aJ ! , (3.56) (3.56) associated to the graded decomposition associated to the graded decomposition sl5 ∼= (2 ⊗3)(−5) ⊕(gl1 ⊕sl2 ⊕sl3) (0) ⊕(2 ⊗3)(5) . (3.57) (3.57) 4Note that the normalisation of the Eisenstein function E[0,0,s,0] does not include the additional factor of ζ(2s −1) as in [13]. – 35 – In this way one computes that the covariant derivative over SL(5)/SO(5) in tangent frame acts on a function of φ, τ, t as [31] In this way one computes that the covariant derivative over SL(5)/SO(5) in tangent frame acts on a function of φ, τ, t as [31] D5 = diag −1 20∂φδβ α −Dαβ, 1 30∂φδb a + Dab . 3.4 N = 2 supergravity in eight dimensions Using the property that DacDbc ˆE[ 3 2 ,0](t) = −1 4Dab ˆE[ 3 2 ,0](t) + 2π 3 δb a , 2DαβDαβ ˆE[1](τ) = π , (3.62) DacDbc ˆE[ 3 2 ,0](t) = −1 4Dab ˆE[ 3 2 ,0](t) + 2π 3 δb a , 2DαβDαβ ˆE[1](τ) = π , (3.62) (3.62) one shows that the general solution to (3.48), as a function of φ, τ and t takes the gen- eral form E 1 8 (φ, τ, t) = e−12φ ˆF[4](τ) + ˆF[4,−2](t) + 1 3 ˆE[1](τ) ˆE[ 3 2 ,0](t) + π 18 ˆE[1](τ) + 19π 12 −10π 3 (φ −φ0) 2 ˆE[1](τ) + ˆE[ 3 2 ,0](t) + π 3 + 100π2 3 (φ −φ0)2 , (3.63) (3.63) where F[4](τ) and F[4,−2](t) are solutions to where F[4](τ) and F[4,−2](t) are solutions to ∆ˆF[4](τ) = 12 ˆF[4](τ) −( ˆE[1](τ))2 , ∆ˆF[4,−2](t) = 12 ˆF[4,−2](t) −( ˆE[ 3 2 ,0](t))2 , DacDcdDdb ˆF[4,−2](t) = 49 16Dab ˆF[4,−2](t) −3 2δab ˆF[4,−2](t) −1 2 ˆE[ 3 2 ,0](t)Dab ˆE[ 3 2 ,0](t) +1 2δab 1 4( ˆE[ 3 2 ,0](t))2 + π 9 ˆE[ 3 2 ,0](t) + π2 27 . (3.64) (3.64) – 36 – Here the notation is used to emphasize that these solutions are deﬁned modulo the homo- geneous solutions E[4](τ) and E[4,−2](t), respectively, as one can see using the formulae of appendix A.6. Note nonetheless that these homogeneous solutions are inconsistent with the string theory perturbation expansion, and the exact threshold function is uniquely determined by these equations and consistency with string theory [26]. Here the notation is used to emphasize that these solutions are deﬁned modulo the homo- geneous solutions E[4](τ) and E[4,−2](t), respectively, as one can see using the formulae of appendix A.6. Note nonetheless that these homogeneous solutions are inconsistent with the string theory perturbation expansion, and the exact threshold function is uniquely determined by these equations and consistency with string theory [26]. The structure of the threshold function exhibits that there is a 1-loop divergence of the R4 type invariant form factor proportional to the ∇6R4 type invariant. This implies the presence of an addition renormalisation scheme ambiguity in the deﬁnition of the analytic part of the eﬀective action. 3.4 N = 2 supergravity in eight dimensions It appears that the renormalisation scheme used in [13, 26], cannot be obtained by simply neglecting the terms in φ−φ0, but one ﬁnds nonetheless that the threshold function only diﬀers by terms proportional to the linear and the quadratic term in φ −φ0, i.e. JHEP07(2015)154 ˆE 1 8 (τ, t) = ˆF[4](τ) + ˆF[4,−2](t) + 1 3 ˆE[1](τ) ˆE[ 3 2 ,0](t) + π 18 ˆE[1](τ) + 19π 12 + π 36 2 ˆE[1](τ) + ˆE[ 3 2 ,0](t) −5π 2 . (3.65) (3.65) Let us now consider the oxidation of the F 2k∇4R4 type invariants, i.e. solve the dif- ferential equation (3.52) for a function of the form e−(10+2k)φE (k) 1 8 (τ, t), as required by dif- feomorphism invariance in eight dimensions. One obtains straightforwardly 2DαβDαβE (k) 1 8 (τ, t) = (1 + k)(2 + k)E (k) 1 8 (τ, t) , DacDcbE (k) 1 8 (τ, t) = (5 + 4k) 12 DabE (k) 1 8 (τ, t) + (2 + k)(1 + 2k) 9 δb aE (k) 1 8 (τ, t) . (3.66) (3.66) Using the results of appendix A.6 one obtain that the solution can be written in terms of Eisenstein functions as E (k) 1 8 (τ, t) ∝E[k+2](τ)E[0,k+2](t) , (3.67) (3.67) consistently with the decompactiﬁcation limit of the SL(5)/SO(5) Eisenstein function [31], E[0,0,k+ 5 2 ,0] = 2ζ(2k + 5)ζ(2k + 4)e−6(2k+5)φ + 2π2ζ(2k + 2) (2k + 3)(k + 1)e8kφE[k+ 3 2 ,0](t) + √πΓ(k + 2) 2Γ(k + 5 2) e−2(k+5)φE[k+2](τ)E[0,k+2](t) + O e−e−5φ . (3.68) (3.68) The sum of the two functions reproduces correctly the threshold function obtained in [13, 26]. The sum of the two functions reproduces correctly the threshold function obtained in [13, 26]. A Ed(d) Eisenstein series, and tensorial diﬀerential equations In this appendix we collect the diﬀerential equations satisﬁed by the Eisenstein functions that are relevant in the analysis of BPS threshold functions in string theory, and their related coadjoint nilpotent orbits. We write them in terms of the covariant derivative in tangent frame valued in the Lie algebra in some particular representations, which are speciﬁed by their dimension. Acknowledgments We would like to thank Axel Kleinschmidt and Boris Pioline for useful discussions, and are particularly grateful to Axel Kleinschmidt for providing some explicit relations between Eisenstein series. Eisenstein series. – 37 – A.1 E8(8) The Eisenstein function in the adjoint representation is associated to the nilpotent orbit of Dynkin label h 0 0000002 i, with D8 Dynkin label h 0 ·0000020 i, and satisﬁes in general to the diﬀerential equation JHEP07(2015)154 DΓi[jk rD DΓlpq]rD Eh 0 000000s i = −δi[j DΓklpq]D (2s(2s −29) + 48)Eh 0 000000s i . (A.1) (A.1) For the following two special values of s, the function is associated to lower dimensional nilpotent orbits, and satisﬁes moreover to For the following two special values of s, the function is associated to lower dimensional nilpotent orbits, and satisﬁes moreover to ΓklD(DΓijklD) Eh 0 000000 9 2 i = −168 ΓijDEh 0 000000 9 2 i , (DΓijklD) Eh 0 000000 5 2 i = 0 . (A.2) (A.2) A.2 E7(7) The Eisenstein function in the adjoint representation is associated to the nilpotent orbit of Dynkin label h 0 200000 i, with A7 Dynkin label [2000002], and satisﬁes in general to the diﬀerential equation D 3 56Eh 0 s00000 i = s(2s−17) 2 +6 D56Eh 0 s00000 i , ∆Eh 0 s00000 i = 2s(2s−17)Eh 0 s00000 i. (A.3) D 3 56Eh 0 s00000 i = s(2s−17) 2 +6 D56Eh 0 s00000 i , ∆Eh 0 s00000 i = 2s(2s−17)Eh 0 s00000 i. (A.3) (A.3) (A.3) For the following two special values of s, the function is associated to lower dimensional nilpotent orbits, and satisﬁes moreover to For the following two special values of s, the function is associated to lower dimensional nilpotent orbits, and satisﬁes moreover to D 3 133Eh 0 5 2 00000 i = −20D133Eh 0 5 2 00000 i , D 2 56Eh 0 3 2 00000 i = −9 2156Eh 0 3 2 00000 i . (A.4) (A.4) The Eisenstein series E h 0 00000 i is generated by a character satisfying to a stronger quartic constraint (2.89) also associated to a lower dimensional nilpotent orbit, but does not itself satisfy this equation [53]. The Eisenstein function in the fundamental representation is associated to the nilpotent orbit of Dynkin label h 0 000002 i, with A7 Dynkin label [0200000] and its conjugate, and satisﬁes in general to the diﬀerential equation D 3 133Eh 0 00000s i = s(s −9)D133Eh 0 00000s i , ∆Eh 0 00000s i = 3s(s −9)Eh 0 00000s i . (A.5) (A.5) The function moreover satisﬁes to highest weight representations constraints for integral s, and is associated to lower dimensional nilpotent orbits for s = 2 and 4. The relation of these Eisenstein functions with nilpotent orbits can be summarised in the following truncated closure diagram 3 [35]. The function moreover satisﬁes to highest weight representations constraints for integral s, and is associated to lower dimensional nilpotent orbits for s = 2 and 4. The relation of these Eisenstein functions with nilpotent orbits can be summarised in the following truncated closure diagram 3 [35]. A.2 E7(7) – 38 – • • • • • • • • Eh 0 000002 i Eh 0 3 2 00000 i Eh 0 000004 i Eh 0 5 2 00000 i Eh 0 00000s i Eh 0 s00000 i 0 - - - - - - - - 34 52 54 64 66 70 76 dim Figure 3. Nilpotent orbits associated to Eisenstein series in the E7(7) closure diagram. JHEP07(2015)154 Figure 3. Nilpotent orbits associated to Eisenstein series in the E7(7) closure diagram. A.3 E6(6) The Eisenstein function in the adjoint representation is associated to the nilpotent orbit of Dynkin label h 2 00000 i, with C4 Dynkin label [2, 0, 0, 0] [54], and satisﬁes in general to the diﬀerential equations D 3 27Eh s 00000 i = 1 2(s −5)(2s −1)D27Eh s 00000 i , ∆Eh s 00000 i = 2s(2s −11)Eh s 00000 i . (A. (A.6) The function is associated to the next to minimal nilpotent orbit for s = 3 2 and to the minimal nilpotent orbit for s = 1. However, there is a 1-parameter family of Eisenstein functions associated to the next to minimal nilpotent orbit. It is the Eisenstein function in the fundamental representation, that satisﬁes to D 3 27 − 2s(s −6) 3 + 5 2 D27 Eh 0 s0000 i = (3 −s) D 2 27 −8 27s(s −6)127 Eh 0 s0000 i , ( (A.7) and its third order derivative restricted to the [2, 0, 0, 1] of Sp(4) vanishes. It is functionally related to the Eisenstein function in the anti-fundamental representation at 6 −s, and reduces to the unique Eisenstein function associated to the minimal nilpotent orbit at s = 3 2. A.4 SO(5, 5) A.4 SO(5, 5) The Eisenstein function in the adjoint representation is associated to the nilpotent orbit of Dynkin label 0 0200 , with C2 × C2 Dynkin label [2, 0] × [0, 0] and [0, 0] × [2, 0], and satisﬁes in general to the diﬀerential equations D 3 16E 0 0s00 = 2s(2s −7) + 3 4 D16E 0 0s00 , D 3 10E 0 0s00 = s(2s −7) + 3 2 D10E 0 0s00 , ∆E 0 0s00 = 2s(2s −7)E 0 0s00 . (A.8) (A.8) – 39 – The function is associated to lower dimensional nilpotent orbits for s = 1, 1 2. The Eisenstein function in the Weyl spinor representation is associated to the largest next to minimal nilpotent orbit, and satisﬁes in general to the diﬀerential equations D 3 16 −13s(s −4) + 24 16 D16 E 0 000s = −3(s −2) 4 D 2 16 −5s(s −4) 16 116 E 0 000s , D 2 10E 0 000s = s(s −4) 4 110E 0 000s . (A.9) (A.9) It is functionally related to the Eisenstein function in the conjugate representation at 4 −s. The Eisenstein function in the vector representation is associated to the smallest next to minimal nilpotent orbit, and satisﬁes in general to the diﬀerential equations It is functionally related to the Eisenstein function in the conjugate representation at 4 −s. The Eisenstein function in the vector representation is associated to the smallest next to minimal nilpotent orbit, and satisﬁes in general to the diﬀerential equations JHEP07(2015)154 JHEP07(2015)154 D 2 16E 0 s000 = s(s −4) 4 116E 0 s000 , D 3 10E 0 s000 = (s −1)(s −3)D10E 0 s000 . (A.10) (A.10) A.5 SL(5) The Eisenstein function in the adjoint representation is associated to the nilpotent or- bit of weighted Dynkin diagram [2, 0, 0, 2], and depends on two parameters. Weyl group symmetry implies functional relations between the functions E[s,t,0,0] ∝E[1−s,s+t−1 2 ,0,0] ∝E[t,0,0, 5 2 −s−t] ∝E[s+t−1 2 ,0,0,2−t] , (A.11) (A.11) and the former satisﬁes to the diﬀerential equations 2D[i [kDj]pDp l]E[s,t,0,0] + 2s + 4t −5 10 D[i [kDj] l]E[s,t,0,0] = δ[i [k 2s + 4t −5 5 Dj]pDp l] + 2s + 4t −5 5 2 −3 4 Dj] l] 2 2D[i [kDj]pDp l]E[s,t,0,0] + 2s + 4t −5 10 D[i [kDj] l]E[s,t,0,0] = δ[i [k 2s + 4t −5 5 Dj]pDp l] + 2s + 4t −5 5 2 −3 4 Dj] l] −32s + 4t −5 40 2s + 4t −5 5 2 −1 δj] l] E[s,t,0,0] , DikDklDlj + 2s + 4t −5 5 DikDkj − 3(2s + 4t −5)2 400 + 2s2 −2s −3 8 Dij E[s,t,0,0] = 2s + 4t −5 160 9(2s + 4t −5)2 25 −4s2 + 4s −9 δi j E[s,t,0,0] , (A.12) (A.12) (A.12) as well as to the Laplace equation as well as to the Laplace equation as well as to the Laplace equation ∆E[s,t,0,0] = s(s −1) + 3 20(2s + 4t −5)2 −15 4 E[s,t,0,0] . (A.13) (A.13) The antisymmetric tensor Eisenstein function is associated to the next to minimal nilpotent orbit, and satisﬁes to the diﬀerential equation DikDkj E[0,0,s,0] = 4s −5 20 Dij E[0,0,s,0] + 3s(2s −5) 25 δj i E[0,0,s,0] , (A.14) (A.14) – 40 – – 40 – whereas the vector Eisenstein function is associated to the minimal nilpotent orbit, and satisﬁes to both whereas the vector Eisenstein function is associated to the minimal nilpotent orbit, and satisﬁes to both DikDkj E[s,0,0,0] = −3(4s −5) 20 Dij E[s,0,0,0] + 2s(2s −5) 25 δj i E[s,0,0,0] , D[i [kDj] l] E[s,0,0,0] = 4s −5 10 δ[k [i Dj] l] E[s,0,0,0] −s(2s −5) 50 δkl ij E[s,0,0,0] . (A.15) (A.15) JHEP07(2015)154 A.6 SL(3) A.6 SL(3) There are only two nilpotent orbits of SL(3), the general Eisenstein function satisﬁes to There are only two nilpotent orbits of SL(3), the general Eisenstein function satisﬁes to D 3 3 E[s,t] = 2s2 + (s + t)(2t −3) + 3 8 6 D3E[s,t] + (s −t)(4s + 2t −3)(2s + 4t −3) 108 13E[s,t] ∆E[s,t] = 2 2s2 + (s + t)(2t −3) 3 E[s,t] , (A.16) (A.16) and the Eisenstein function associated to the minimal nilpotent orbit satisﬁes and the Eisenstein function associated to the minimal nilpotent orbit satisﬁes D 2 3 E[s,0] = −4s −3 12 D3E[s,0] + s(2s −3) 9 13E[s,0] . (A.17) (A.17) B Some additional computations on Eisenstein series B.1 E6(6) Eisenstein series in the fundamental representation In the decompactiﬁcation limit, the series deﬁnition (2.59) of the Eisenstein series E h 0 00000s i as a sum over the rank 1 charges in the 56 or E7(7), decomposes into the four compo- nents (3.10) p0, pI, qI, q0 of grad −3, −1, 1, 3 respectively, with the rank one constraint 1 2tIJKqJqK = q0pI , tIKP tJLP qKpL −pIqJ = δI Jq0p0 , 1 2tIJKpJpK = p0qI . (B.1) The E invariant norm then reads 1 2tIJKqJqK = q0pI , tIKP tJLP qKpL −pIqJ = δI Jq0p0 , 1 2tIJKpJpK = p0qI . (B (B.1) The E7(7) invariant norm then reads The E7(7) invariant norm then reads Z(Γ)2 = e6φ q0 + aIqI + 1 2tIJKaIaJpK + 1 6tIJLaIaJaLp0 2 + e2φ Z qI + tIJKaJpK + 1 2tIJKaJaKp0 2 + e−2φZ(pI + aIp0) 2 + e−6φ(p0)2 , (B.2) (B.2) – 41 – where e−2φ is the radius moduli, whereas \|Z(q)\|2 now represents the E6(6) invariant norm. At large e−2φ we will only consider the sum over the maximal weight charges q0, qI where e−2φ is the radius moduli, whereas \|Z(q)\|2 now represents the E6(6) invariant norm. At large e−2φ we will only consider the sum over the maximal weight charges q0, qI X Γ∈Z56 ∗ Γ×Γ=0 \|Z(Γ)\|−2s X Γ∈Z56 ∗ Γ×Γ=0 \|Z(Γ)\|−2s = 2ζ(2s)e−6sφ + X q∈Z27 ∗ q×q=0 X q0∈Z πs Γ(s) Z ∞ 0 dt t1+s e−π t e6φ(q0+aIqI)2+e2φ\|Z(q)\|2 + . . . = 2ζ(2s)e−6sφ + X q∈Z27 ∗ q×q=0 X ˜q0∈Z πs Γ(s) Z ∞ 0 dt t 1 2 +s e−3φe−π t e2φ\|Z(q)\|2−πte−6φ˜q0+2πi˜q0qIaI + . . . = 2ζ(2s)e−6sφ + π 1 2 Γ(s −1 2) Γ(s) e−2(s+1)φ X q∈Z27 ∗ q×q=0 \|Z(q)\|−2(s−1 2 ) + . . . (B.3) = 2ζ(2s)e−6sφ + X q∈Z27 ∗ q×q 0 X q0∈Z πs Γ(s) Z ∞ 0 dt t1+s e−π t e6φ(q0+aIqI)2+e2φ\|Z(q)\|2 + . . . q q = 2ζ(2s)e−6sφ + X q∈Z27 ∗ q×q=0 X ˜q0∈Z πs Γ(s) Z ∞ 0 dt t 1 2 +s e−3φe−π t e2φ\|Z(q)\|2−πte−6φ˜q0+2πi˜q0qIaI + . . . = 2ζ(2s)e−6sφ + π 1 2 Γ(s −1 2) Γ(s) e−2(s+1)φ X q∈Z27 ∗ q×q=0 \|Z(q)\|−2(s−1 2 ) + . . . (B.3) = 2ζ(2s)e−6sφ + X q∈Z27 ∗ X ˜q0∈Z πs Γ(s) Z ∞ 0 dt t 1 2 +s e−3φe−π t e2φ\|Z(q)\|2−πte−6φ˜q0+2πi˜q0qIaI + . . . B.1 E6(6) Eisenstein series in the fundamental representation JHEP07(2015)154 (B.3) The other terms are more complicate to obtain explicitly, but they follow the same pattern such that the perturbative terms reduce to sum over the charges of grad −1 and −3 after Poisson resumation. The complete perturbative expansion in e−2φ is then determined by compatibility with the Langlands functional identity to be Eh 0 00000s i = 2ζ(2s)e−6sφ + π 1 2 Γ s −1 2 Γ(s) e−2(s+1)φEh 0 0000s- 1 2 i + π5Γ s −9 2 Γ s −5 2 ζ(2s −9) Γ(s −2)Γ(s)ζ(2s −4) e2(s−10)φEh 0 s- 5 2 0000 i + 2π 27 2 Γ s −17 2 Γ s −13 2 Γ s −9 2 ζ(2s −17)ζ(2s −13)ζ(2s −9) Γ(s −4)Γ(s −2)Γ(s)ζ(2s −8)ζ(2s −4) e6(s−9)φ (B.4) (B.4) The generating character of the function Eh 0 0000s i is deﬁned in terms of the central charge of qI Zij(q) = VijIqI , \|Z(q)\|2 = Zij(q)Zij(q) , (B.5) (B.5) ij(q) ij qI , \| (q)\| ij(q) (q) , ( ) and the quadratic constraint tIJKqJqK = 0 is equivalent to Zik(q)Zjk(q) = 1 8δj i \|Z(q)\|2 . (B.6) Zik(q)Zjk(q) = 1 8δj i \|Z(q)\|2 . (B.6) (B.6) The covariant derivative in tangent frame acts on the central charge as DijklZpq(q) = 3 δ[ij pq Zkl](q) −Ω[ijδk [pZq] l](q) −1 4ΩpqΩ[ijZkl](q) −1 12Ω[ijΩkl]Zpq(q) , (B.7) uch that jklZpq(q) = 3 δ[ij pq Zkl](q) −Ω[ijδk [pZq] l](q) −1 4ΩpqΩ[ijZkl](q) −1 12Ω[ijΩkl]Zpq(q) , such that such that Dijkl\|Z(q)\|2 = 1 4Z[ijZkl] + 1 96Ω[ijΩkl]\|Z(q)\|2 . (B.8) (B.8) One computes that Dijkl\|Z(q)\|−2s = s 1 4Z(q)[ijZ(q)kl] + 1 96Ω[ijΩkl]\|Z(q)\|2 \|Z(q)\|−2(s+1) , (B.9) Dijkl\|Z(q)\|−2s = s 1 4Z(q)[ijZ(q)kl] + 1 96Ω[ijΩkl]\|Z(q)\|2 \|Z(q)\|−2(s+1) , (B.9) (B.9) – 42 – – 42 – and and DijpqDklpq\|Z(q)\|−2s = −2 3s(3 −2s) Z(q)ijZ(q)kl −Z(q)i[kZ(q)l]j + 1 36s(63 −10s)Ωi[kΩl]j\|Z(q)\|2 + 1 36s(9 −2s)ΩijΩkl\|Z(q)\|2 \|Z(q)\|−2(s+1) . B.1 E6(6) Eisenstein series in the fundamental representation (B.10) (B.10) The right-hand-side decomposes into an Sp(4) singlet and a tensor in the irreducible rep- resention [0, 2, 0, 0] JHEP07(2015)154 JHEP07(2015)154 Z(q)ijZ(q)kl [0,2,0,0] = 1 2 Z(q)ijZ(q)kl−Z(q)i[kZ(q)l]j−1 72ΩijΩkl\|Z(q)\|2+ 1 72Ωi[kΩl]j\|Z(q) (B 1 One deduces the Laplace equation One deduces the Laplace equation ∆\|Z(q)\|−2s = 1 3DijpqDijpq\|Z(q)\|−2s = 8 3s(s −6)\|Z(q)\|−2s , (B.12) and one gets at third order DijrsDrspqDklpq\|Z(q)\|−2s (B.13) = −1 3s 8s2 −42s + 33 Z(q)ijZ(q)kl − 1 108s 14s2 −81s + 54 Ωi[kΩl]j\|Z(q)\|2 −1 216 8s3−54s2+27s ΩijΩkl\|Z(q)\|2−2 3s 2s2−15s+6 Z(q)i[kZ(q)l]j \|Z(q)\|−2(s+1) . (B.13) The right hand side can be expressed in terms of lower oder derivatives of \|Z(q)\|−2s using the relations (B.9) and (B.10), such that DijrsDrspqDklpq\|Z(q)\|−2s = 2 3s(s −6) + 5 2 Dijkl\|Z(q)\|−2s +(3 −s) DijpqDklpq + 1 27 Ωi[kΩl]j + 1 8ΩijΩkl DpqrsDpqrs \|Z(q)\|−2s . (B.14) Moreover, one straightforwardly works out that the third order derivative projected to the [2, 0, 0, 1] irreducible representation of Sp(4) vanishes Moreover, one straightforwardly works out that the third order derivative projected to the [2, 0, 0, 1] irreducible representation of Sp(4) vanishes D3 [2,0,0,1]\|Z(q)\|−2s = 0 . (B.15) (B.15) B.2 SO(6, 6) Eisenstein series B.2 SO(6, 6) Eisenstein series We deﬁne ﬁrst the series associated to anti-chiral spinors. The associated ‘central charge’ is Ziˆı and its complex conjugate Ziˆı, where i = 1, 4 of one SU(4) factor and ˆı = 1, 4 of the other. The rank one constraint on the spinor is the pure spinor constraint, that reads ZiˆkZj ˆk = 1 4δi jZkˆlZk ˆl , ZkˆıZkˆ= 1 4δˆ ˆıZkˆlZk ˆl , 1 2εijpqZpˆkZqˆl = 1 2εˆkˆlˆpˆqZi ˆpZj ˆq . (B.16) (B.16) – 43 – – 43 – One computes then that the covariant derivative over SO(6, 6)/(SO(6) × SO(6)) acts on the central charge as One computes then that the covariant derivative over SO(6, 6)/(SO(6) × SO(6)) acts on the central charge as DijˆkˆlZpˆq = 1 2εˆkˆlˆqˆrδp [iZj] ˆr , DijˆkˆlZpˆq = 1 2εijprδˆq [ˆkZrˆl] . (B.17) (B.17) Considering a homogeneous function of \|Z\|2 = Zi ˆZiˆ, one has Considering a homogeneous function of \|Z\|2 = Zi ˆZiˆ, one has Considering a homogeneous function of \|Z\|2 = Zi ˆZiˆ, one has Dijˆkˆl\|Z\|−2s = −s εˆkˆlˆpˆqZi ˆpZj ˆq\|Z\|−2s−2 , (B.18) Dijˆkˆl\|Z\|−2s = −s εˆkˆlˆpˆqZi ˆpZj ˆq\|Z\|−2s−2 , (B.18) and more generally in the vector representation (note that Dijˆkˆl = 1 2εijpqDpqˆkˆl and etc...) JHEP07(2015)154 DijˆpˆqDklˆpˆq\|Z\|−2s = s(s −5) 4 δij kl \|Z\|−2s , (B.19) (B.19) in the chiral spinor representation DipˆkˆqDjpˆlˆq\|Z\|−2s = −s(s −1) 2 ZiˆlZj ˆk\|Z\|−2s−2 + s(s −4) 8 δi jδ ˆk ˆl \|Z\|−2s , DipˆkˆqDprˆqˆsDjrˆlˆs\|Z\|−2s = 2s2 −10s + 5 8 Dijˆkˆl\|Z\|−2s , (B.20) (B.20) (B.20) and in the anti-chiral representation Dip ˆkˆqDjpˆlˆq\|Z\|−2s = s(s −1) 2 Zi ˆkZjˆl\|Z\|−2s−2 + s(s −7) 16 δj i δ ˆk ˆl \|Z\|−2s , Dip ˆkˆqDpr ˆqˆsDjr ˆlˆs\|Z\|−2s = −3s(s −1)(s −2) 8 Zi ˆkZj ˆl\|Z\|−2s−2 + s2 −11s + 4 16 Dij ˆkˆl\|Z\|−2s . (B.21) (B.21) As in the preceding section, one can deﬁne the series E s 00000 = X Λ∈Z32 ΛΓMNΛ=0 Z(Λ)iˆZ(Λ)iˆ−s . (B.22) The series only converges for s > 5, and satisﬁes to The series only converges for s > 5, and satisﬁes to The series only converges for s > 5, and satisﬁes to The series only converges for s > 5, and satisﬁes to E s 00000 = π 15 2 Γ s −9 2 Γ s −7 2 Γ s −5 2 ζ(2s −9)ζ(2s −7)ζ(2s −5) Γ(s −2)Γ(s −1)Γ(s)ζ(2s −4)ζ(2s −2)ζ(10 −2s) E 5-s 0000 0 . B.2 SO(6, 6) Eisenstein series (B 23) E s 00000 = π 15 2 Γ s −9 2 Γ s −7 2 Γ s −5 2 ζ(2s −9)ζ(2s −7)ζ(2s −5) Γ(s −2)Γ(s −1)Γ(s)ζ(2s −4)ζ(2s −2)ζ(10 −2s) E 5-s 0000 0 . (B (B.23) (B.23) ( ) The ﬁrst critical function is E 1 00000 , which solves a quadratic equation in all three funda- mental representation, ( ) The ﬁrst critical function is E 1 00000 , which solves a quadratic equation in all three funda- mental representation, DijˆpˆqDklˆpˆq\|Z\|−2 = −δij kl \|Z\|−2 , DipˆkˆqDjpˆlˆq\|Z\|−2 = −3 8δi jδ ˆk ˆl \|Z\|−2 , Dip ˆkˆqDjpˆlˆq\|Z\|−2 = −3 8δj i δ ˆk ˆl \|Z\|−2 . (B.24) (B.24) – 44 – and is in fact proportional to E 0 00001 and E 0 20000 . This function is associated to the min- imal unitary representation of SO(6, 6). The next one is E 2 00000 , which solves a quadratic vector equation and two cubic spinor equations vector equation and two cubic spinor equations DijˆpˆqDklˆpˆq\|Z\|−4 = −3 2δij kl \|Z\|−4 , DipˆkˆqDprˆqˆsDjrˆlˆs\|Z\|−4 = −7 2Dijˆkˆl\|Z\|−4 , Dip ˆkˆqDpr ˆqˆsDjr ˆlˆs\|Z\|−4 = −7 2Dij ˆkˆl\|Z\|−4 . (B.25) (B.25) JHEP07(2015)154 JHEP07(2015)154 It is equal to E 0 0000 . The divergent Eisenstein series are It is equal to E 0 0000 . The divergent Eisenstein series are E 0 0000+ϵ = 45 2π ϵE 0 0000 + ˆE 0 0000 + O(ϵ) , E 0 0000+ϵ = 14 175 ζ(3) 8π3 ϵ E 0 0000 + ˆE 0 0000 + O(ϵ) , E 0 0000+ϵ = 1 488 375 ζ(3)ζ(5) 256π5 ϵ + ˆE 0 0000 + O(ϵ) , (B.26) (B.26) We will now consider a charge Q in the vector representation, satisfying ⟨Q, Q⟩= 0. In this case it is convenient to use vector indices, with the deﬁnition We will now consider a charge Q in the vector representation, satisfying ⟨Q, Q⟩= 0. In this case it is convenient to use vector indices, with the deﬁnition Dijˆkˆl = 1 4γaijγ ˆbˆkˆlDaˆb . (B.27) (B.27) The real ‘central charges’ Za and Zˆa then satisfy to the constraint The real ‘central charges’ Za and Zˆa then satisfy to the constraint ZaZa = ZˆaZˆa , (B.28) (B.28) and DaˆbZc = 1 2δacZˆb , DaˆbZˆc = 1 2δˆbˆcZa . B.2 SO(6, 6) Eisenstein series (B.29) (B.29) One computes then that a homogeneous function of ZaZa satisﬁes to Daˆb(ZcZc)−s = −sZaZˆb(ZcZc)−s , D[a [ˆcDb] ˆd](ZeZe)−s = 0 , DaˆcDbˆc(ZdZd)−s = s(s −2)ZaZb(ZcZc)−s−1 −s 2δab(ZcZc)−s , DaˆcDdˆcDb ˆd(ZeZe)−s = (s2 −5s + 5)Daˆb (ZcZc)−s . (B.30) (B.30) The second equation implies that this function always satisﬁes to a quadratic equation in the two spinor representations, whereas it only satisﬁes to a quadratic constraint in the vector representation for the critical value s = 2. As in the preceding section, one can deﬁne the series As in the preceding section, one can deﬁne the series E 0 s0000 = X Q∈Z12 ⟨Q,Q⟩=0 (Z(Q)aZ(Q)a)−s . (B.31) (B.31) – 45 – – 45 – The series only converges for s > 5, and satisﬁes to E 0 s0000 = π5 Γ s −9 2 Γ s −5 2 ζ(2s −9)ζ(2s −5) Γ(s −2)Γ(s)ζ(2s −4)ζ(10 −2s) E 0 -s0000 . (B.32) (B.32) The divergent series are The divergent series are E 0 +ϵ0000 = 3 2 ϵE 0 0000 + ˆE 0 0000 + O(ϵ) , E 0 +ϵ0000 = 945 ζ(5) 128 ϵ + ˆE 0 0000 + O(ϵ) . (B.33) (B.33) JHEP07(2015)154 However the function is ﬁnite at s = 4 and However the function is ﬁnite at s = 4 and However the function is ﬁnite at s = 4 and E 0 0000 = 15 ζ(3) 2 E 0 0000 . (B.34) (B.34) satisfying moreover to satisfying moreover to DaˆcDbˆc Eα2, n 2 = 3 −n 2 δabEα2, n 2 . (B.46) (B.46) Similarly as for the E7(7) Eisenstein series in the fundamental representation, we expect this property to generalise to s = n 2 +k for any integer k, such that D2+2kEα2, n 2 +k restricted to the symmetric rank 2 + 2k representation of SO(n) vanishes. For k = 1 on computes indeed that Ddˆa X(a ˆbXb\|ˆbXcˆcXd)ˆc\|X\|−2s−4− 4 n+4δ(abXc ˆbXd)ˆb\|X\|−2s−2+ 2 (n+2)(n+4)δ(abδcd)\|X\|−2s = n(n + 2 −2s) 2(n + 4) X(a ˆbXb\|ˆbXc) ˆcΛˆaˆc\|X\|−2s−4 − 3 n + 2δ(abXc) ˆbΛˆaˆb\|X\|−2s−2 . (B.47) Open Access. This article is distributed under the terms of the Creative Commons Attribution License (CC-BY 4.0), which permits any use, distribution and reproduction in any medium, provided the original author(s) and source are credited. Open Access. This article is distributed under the terms of the Creative Commons Attribution License (CC-BY 4.0), which permits any use, distribution and reproduction in any medium, provided the original author(s) and source are credited. B.3 SO(n, n) Eisenstein series in the adjoint (B.41) (B.41) This implies that in particular an equation of the type This implies that in particular an equation of the type This implies that in particular an equation of the type D 3 2n−1\|X\|−2s = asD2n−1\|X\|−2s , (B.42) (B.42) in the spinor representation, for a coeﬃcient that can straightforwardly be ﬁxed. Moreover JHEP07(2015)154 Moreover Dcˆb nXa ˆdXc ˆd\|X\|−2s−2 −δc a\|X\|−2s = −(n −2)(2s −n) 2 ΛacXcˆb\|X\|−2s−2 , (B.43) (B.43) suggesting that the function Eα2, n 2 = X Q∈so(n,n) Q2=0 \|X(Q)\|−2s (B.44) at s = n 2 decomposes into the sum Eα2,s = Eα2,s + ¯Eα2,s , (B.45) (B.45) B.3 SO(n, n) Eisenstein series in the adjoint For SO(n, n) the adjoint representation decomposes with respect to SO(n) × SO(n) with a running from 1 to n of the ﬁrst SO(n) and ˆa running from 1 to n of the second. We decompose therefore the adjoint into the coset component Xaˆb and the two antisymmetric tensors Λab and Λˆaˆb. The minimal representation is such that a charge Q ∈so(n, n) is nilpotent in all three fundamental representations, which reads explicitly ΛacΛbc = XaˆcXbˆc , Λ[abΛcd] = 0 , ΛacXcˆb = −XaˆcΛˆcˆb , Λ[abXc] ˆd = 0 , ΛˆaˆcΛˆbˆc = XcˆaXcˆb , Xa[ˆbΛˆc ˆd] = 0 , ΛabΛˆaˆb = −2X[a [ˆcXb] ˆd] , Λ[ˆaˆbΛˆc ˆd] = 0 . (B.35) (B.35) They satisfy to They satisfy to DaˆbXc ˆd = 1 2δacΛˆb ˆd + 1 2δˆb ˆdΛac , DaˆbΛcd = δa[cXd]ˆb , DaˆbΛˆc ˆd = δˆb[ˆcXa\| ˆd] . (B.36) Using this one computes DaˆbDcˆb\|X\|−2s = s(2s −n + 3)XaˆbXcˆb\|X\|−2s−2 −s δc a \|X\|−2s , (B.37) DaˆbDcˆb\|X\|−2s = s(2s −n + 3)XaˆbXcˆb\|X\|−2s−2 −s δc a \|X\|−2s , (B.37) (B.37) such that such that ∆\|X\|−2s ≡2DaˆbDaˆb\|X\|−2s = 2s(2s −2n + 3)\|X\|−2s . (B.38) ∆\|X\|−2s ≡2DaˆbDaˆb\|X\|−2s = 2s(2s −2n + 3)\|X\|−2s . (B.38) (B.38) Note that the case s = n−3 2 is special, and reduces then to a spinor representation Eisenstein function. In general one has still 2 function. In general one has still DaˆcDdˆcDdˆb\|X\|−2s = s(2s −2n + 3) 2 + (n −2)(n −3) 4 Daˆb\|X\|−2s. (B.39) (B.39) One computes moreover that One computes moreover that D[a [ˆcDb] ˆd]\|X\|−2s = s(2s −1)X[a [ˆcXb] ˆd]\|X\|−2s−2 , (B.40) (B.40) – 46 – – 46 – such that the representation s = 1 2 is special, and then reduce to a vector representation Eisenstein function. Using representation theory, one straightforwardly check that there is no possible rank 3 antisymmetric tensor that one can write, such that such that the representation s = 1 2 is special, and then reduce to a vector representation Eisenstein function. Using representation theory, one straightforwardly check that there is no possible rank 3 antisymmetric tensor that one can write, such that D[a [ ˆdDb ˆdDc] ˆf]\|X\|−2s = 0 . [1] M.B. Green and M. Gutperle, Eﬀects of D instantons, Nucl. Phys. B 498 (1997) 195 [hep-th/9701093] [INSPIRE]. [2] M.B. Green, M. Gutperle and P. Vanhove, One loop in eleven-dimensions, Phys. Lett. B 409 (1997) 177 [hep-th/9706175] [INSPIRE]. References [1] M.B. Green and M. Gutperle, Eﬀects of D instantons, Nucl. Phys. B 498 (1997) 195 [hep-th/9701093] [INSPIRE]. [2] M.B. Green, M. Gutperle and P. Vanhove, One loop in eleven-dimensions, Phys. Lett. B 409 (1997) 177 [hep-th/9706175] [INSPIRE]. – 47 – [3] E. Kiritsis and B. Pioline, On R4 threshold corrections in IIB string theory and (p, q) string instantons, Nucl. Phys. B 508 (1997) 509 [hep-th/9707018] [INSPIRE]. [4] E. D’Hoker and D.H. Phong, Two-loop superstrings VI: Non-renormalization theorems and the 4-point function, Nucl. Phys. B 715 (2005) 3 [hep-th/0501197] [INSPIRE]. [5] M.B. Green, J.G. Russo and P. Vanhove, Low energy expansion of the four-particle genus-one amplitude in type-II superstring theory, JHEP 02 (2008) 020 [arXiv:0801.0322] [INSPIRE]. [6] H. Gomez and C.R. Mafra, The closed-string 3-loop amplitude and S-duality, JHEP 10 (2013) 217 [arXiv:1308.6567] [INSPIRE]. 7] M.B. Green, S.D. Miller and P. Vanhove, SL(2, Z)-invariance and D-instanton contributions to the D6R4 interaction, arXiv:1404.2192 [INSPIRE]. JHEP07(2015)154 JHEP07(2015)154 [8] E. D’Hoker, M.B. Green, B. Pioline and R. Russo, Matching the D6R4 interaction at two-loops, JHEP 01 (2015) 031 [arXiv:1405.6226] [INSPIRE]. [9] M.B. Green, H.-h. Kwon and P. Vanhove, Two loops in eleven-dimensions, Phys. Rev. D 61 (2000) 104010 [hep-th/9910055] [INSPIRE]. [10] M.B. Green and P. Vanhove, Duality and higher derivative terms in M-theory, JHEP 01 (2006) 093 [hep-th/0510027] [INSPIRE]. [11] M.B. Green, J.G. Russo and P. Vanhove, Modular properties of two-loop maximal supergravity and connections with string theory, JHEP 07 (2008) 126 [arXiv:0807.0389] [INSPIRE]. [12] A. Basu, Constraining non–BPS interactions from counterterms in three loop maximal supergravity, Class. Quant. Grav. 32 (2015) 045012 [arXiv:1408.0094] [INSPIRE]. [13] M.B. Green, J.G. Russo and P. Vanhove, Automorphic properties of low energy string amplitudes in various dimensions, Phys. Rev. D 81 (2010) 086008 [arXiv:1001.2535] [INSPIRE]. [14] M.B. Green and S. Sethi, Supersymmetry constraints on type IIB supergravity, Phys. Rev. D 59 (1999) 046006 [hep-th/9808061] [INSPIRE]. [15] A. Sinha, The ˆG4λ16 term in IIB supergravity, JHEP 08 (2002) 017 [hep-th/0207070] [INSPIRE]. [16] A. Basu, Supersymmetry constraints on the R4 multiplet in type IIB on T 2, Class. Quant. Grav. 28 (2011) 225018 [arXiv:1107.3353] [INSPIRE]. [17] A. Basu, The D4R4 term in type IIB string theory on T 2 and U-duality, Phys. Rev. D 77 (2008) 106003 [arXiv:0708.2950] [INSPIRE]. [18] B. Pioline, H. Nicolai, J. Plefka and A. Waldron, R4 couplings, the fundamental membrane and exceptional theta correspondences, JHEP 03 (2001) 036 [hep-th/0102123] [INSPIRE]. [18] B. Pioline, H. References Nicolai, J. Plefka and A. Waldron, R4 couplings, the fundamental membrane and exceptional theta correspondences, JHEP 03 (2001) 036 [hep-th/0102123] [INSPIRE]. [19] D. Kazhdan, B. Pioline and A. Waldron, Minimal representations, spherical vectors and exceptional theta series, Commun. Math. Phys. 226 (2002) 1 [hep-th/0107222] [INSPIRE]. [19] D. Kazhdan, B. Pioline and A. Waldron, Minimal representations, spherical vectors and exceptional theta series, Commun. Math. Phys. 226 (2002) 1 [hep-th/0107222] [INSPIRE]. [20] B. Pioline and A. Waldron, The Automorphic membrane, JHEP 06 (2004) 009 [hep-th/0404018] [INSPIRE]. [20] B. Pioline and A. Waldron, The Automorphic membrane, JHEP 06 (2004) 009 [hep-th/0404018] [INSPIRE]. [21] B. Pioline, R4 couplings and automorphic unipotent representations, JHEP 03 (2010) 116 [arXiv:1001.3647] [INSPIRE]. [21] B. Pioline, R4 couplings and automorphic unipotent representations, JHEP 03 (2010) 116 [arXiv:1001.3647] [INSPIRE]. [22] E. D’Hoker, M.B. Green, B. Pioline and R. Russo, Matching the D6R4 interaction at two-loops, JHEP 01 (2015) 031 [arXiv:1405.6226] [INSPIRE]. – 48 – [23] A. Basu, The D6R4 term from three loop maximal supergravity, Class. Quant. Grav. 31 (2014) 245002 [arXiv:1407.0535] [INSPIRE]. [24] E. D’Hoker, M.B. Green and P. Vanhove, On the modular structure of the genus-one Type II superstring low energy expansion, arXiv:1502.06698 [INSPIRE]. [25] N.A. Obers and B. Pioline, Eisenstein series and string thresholds, Commun. Math. Phys. 209 (2000) 275 [hep-th/9903113] [INSPIRE]. [26] A. Basu, The D6R4 term in type IIB string theory on T 2 and U-duality, Phys. Rev. D 77 (2008) 106004 [arXiv:0712.1252] [INSPIRE]. [27] M.B. Green, S.D. Miller, J.G. Russo and P. Vanhove, Eisenstein series for higher-rank groups and string theory amplitudes, Commun. Num. Theor. Phys. 4 (2010) 551 [arXiv:1004.0163] [INSPIRE]. JHEP07(2015)154 JHEP07(2015)154 [28] M.B. Green, S.D. Miller and P. Vanhove, Small representations, string instantons and Fourier modes of Eisenstein series (with an appendix by D. Ciubotaru and P. Trapa), arXiv:1111.2983 [INSPIRE]. [29] P. Fleig, A. Kleinschmidt and D. Persson, Fourier expansions of Kac-Moody Eisenstein series and degenerate Whittaker vectors, Commun. Num. Theor. Phys. 08 (2014) 41 [arXiv:1312.3643] [INSPIRE]. [30] G. Bossard and V. Verschinin, Minimal unitary representations from supersymmetry, JHEP 10 (2014) 008 [arXiv:1406.5527] [INSPIRE]. [31] G. Bossard and V. Verschinin, E∇4R4 type invariants and their gradient expansion, JHEP 03 (2015) 089 [arXiv:1411.3373] [INSPIRE]. [32] H.P.A. Gustafsson, A. Kleinschmidt and D. Persson, Small automorphic representations and degenerate Whittaker vectors, arXiv:1412.5625 [INSPIRE]. [33] B. Pioline, D6R4 amplitudes in various dimensions, JHEP 04 (2015) 057 [arXiv:1502.03377] [INSPIRE]. [34] J.M. Drummond, P.J. Heslop, P.S. References Howe and S.F. Kerstan, Integral invariants in N = 4 SYM and the eﬀective action for coincident D-branes, JHEP 08 (2003) 016 [hep-th/0305202] [INSPIRE]. [35] D.ˇZ. D– okovi´c, The closure diagrams for nilpotent orbits of real forms EVI and EVII of E7, Representation Theor. 5 (2001) 17. [36] E. Cremmer and B. Julia, The SO(8) Supergravity, Nucl. Phys. B 159 (1979) 141 [INSPIRE]. [37] L. Brink and P.S. Howe, The N = 8 Supergravity in Superspace, Phys. Lett. B 88 (1979) 268 [INSPIRE]. [38] P.S. Howe, Supergravity in Superspace, Nucl. Phys. B 199 (1982) 309 [INSPIRE]. [39] M. G¨unaydin and N. Marcus, The Unitary Supermultiplet of N = 8 Conformal Superalgebra Involving Fields of Spin ≤2, Class. Quant. Grav. 2 (1985) L19 [INSPIRE]. [40] J.M. Drummond, P.J. Heslop and P.S. Howe, A Note on N = 8 counterterms, arXiv:1008.4939 [INSPIRE]. [41] G. Bossard, P.S. Howe and K.S. Stelle, The Ultra-violet question in maximally supersymmetric ﬁeld theories, Gen. Rel. Grav. 41 (2009) 919 [arXiv:0901.4661] [INSPIRE]. – 49 – [42] M.B. Green, J.G. Russo and P. Vanhove, String theory dualities and supergravity divergences, JHEP 06 (2010) 075 [arXiv:1002.3805] [INSPIRE]. [43] G.G. Hartwell and P.S. Howe, (N, p, q) harmonic superspace, Int. J. Mod. Phys. A 10 (1995) 3901 [hep-th/9412147] [INSPIRE]. [44] G. Bossard, P.S. Howe, K.S. Stelle and P. Vanhove, The vanishing volume of D = 4 superspace, Class. Quant. Grav. 28 (2011) 215005 [arXiv:1105.6087] [INSPIRE]. [45] T. Voronov, Geometric integration theory on supermanifolds, volume 9, part 1 of Soviet scientiﬁc reviews, Section C: Mathematical physics reviews, CRC Press (1991). [46] S.J. Gates Jr., Ectoplasm has no topology: The Prelude, hep-th/9709104 [INSPIRE]. JHEP07(2015)154 [47] S.J. Gates Jr., M.T. Grisaru, M.E. Knutt-Wehlau and W. Siegel, Component actions from curved superspace: Normal coordinates and ectoplasm, Phys. Lett. B 421 (1998) 203 [hep-th/9711151] [INSPIRE]. [48] J. Greitz and P.S. Howe, Maximal supergravity in three dimensions: supergeometry and diﬀerential forms, JHEP 07 (2011) 071 [arXiv:1103.2730] [INSPIRE]. [49] P.S. Howe and M.I. Leeming, Harmonic superspaces in low dimensions, Class. Quant. Grav. 11 (1994) 2843 [hep-th/9408062] [INSPIRE]. [50] S. Ferrara and M. G¨unaydin, Orbits of exceptional groups, duality and BPS states in string theory, Int. J. Mod. Phys. A 13 (1998) 2075 [hep-th/9708025] [INSPIRE]. [51] J. Igusa, A classiﬁcation of spinors up to dimension twelve, Amer. J. Math. 92 (1970) 997. [52] H. L¨u, C.N. Pope and K.S. Stelle, Multiplet structures of BPS solitons, Class. Quant. Grav. References 15 (1998) 537 [hep-th/9708109] [INSPIRE]. [53] G. Bossard and A. Kleinschmidt, Supergravity divergences, supersymmetry and automorphic forms, to appear. [54] D.ˇZ. D– okovi´c, The closure diagrams for nilpotent orbits of real form of E6, J. Lie Theor. 11 (2001) 381. [54] D.ˇZ. D– okovi´c, The closure diagrams for nilpotent orbits of real form of E6, J. Lie Theor. 11 (2001) 381. – 50 –
https://openalex.org/W4292188484	https://pubs.aip.org/asa/jasa/article-pdf/152/2/1003/16527094/1003_1_online.pdf	English	null	Benchmark problems for transcranial ultrasound simulation: Intercomparison of compressional wave models	The Journal of the Acoustical Society of America/The journal of the Acoustical Society of America	2,022	cc-by	13,735	Benchmark problems for transcranial ultrasound simulation: Intercomparison of compressional wave models a) Jean-Francois Aubry; Oscar Bates; Christian Boehm; Kim Butts Pauly; Douglas Christensen; Carlos Cueto; Pierre Gélat; Lluis Guasch; Jiri Jaros ; Yun Jing; Rebecca Jones; Ningrui Li; Patrick Marty; Hazael Montanaro; Esra Neufeld; Samuel Pichardo; Gianmarco Pinton; Aki Pulkkinen; Antonio Stanziola; Axel Thielscher; Bradley Treeby ; Elwin van 't Wout Benchmark problems for transcranial ultrasound simulation: Intercomparison of compressional wave models a) Benchmark problems for transcranial ultrasound simulation: Intercomparison of compressional wave models a) Jean-Francois Aubry; Oscar Bates; Christian Boehm; Kim Butts Pauly; Douglas Christensen; Carlos Cueto; Pierre Gélat; Lluis Guasch; Jiri Jaros ; Yun Jing; Rebecca Jones; Ningrui Li; Patrick Marty; Hazael Montanaro; Esra Neufeld; Samuel Pichardo; Gianmarco Pinton; Aki Pulkkinen; Antonio Stanziola; Axel Thielscher; J. Acoust. Soc. Am. 152, 1003–1019 (2022) https://doi.org/10.1121/10.0013426 )Invited. b)Also at: Laboratory for Acoustics/Noise Control, Empa, Swiss Federal Laboratories for Materials Science and Technology, Dubendorf, Switzerland. c)Also at: Department of Information Technology and Electrical Engineering, Swiss Federal Institute of Technology (ETH), Zurich, Switzerland. d)Also at: Danish Research Center for Magnetic Resonance, Copenhagen University Hospital Hvidovre, Hvidovre, Denmark. e)Electronic mail: b.treeby@ucl.ac.uk ARTICLE ................................... Benchmark problems for transcranial ultrasound simulation: Intercomparison of compressional wave modelsa) Jean-Francois Aubry,1 Oscar Bates,2 Christian Boehm,3 Kim Butts Pauly,4 Douglas Christensen,5 Carlos Cueto,2 Pierre Gelat,6 Lluis Guasch,7 Jiri Jaros,8 Yun Jing,9 Rebecca Jones,10 Ningrui Li,11 Patrick Marty,3 Hazael Montanaro,12,b) Esra Neufeld,12,c) Samuel Pichardo,13 Gianmarco Pinton,10 Aki Pulkkinen,14 Antonio Stanziola,15 Axel Thielscher,16,d) Bradley Treeby,15,e) and Elwin van ’t Wout17 1Physics for Medicine Paris, National Institute of Health and Medical Research (INSERM) U1273, ESPCI Paris, Paris Sciences and Lettres University, French National Centre for Scientiﬁc Research (CNRS) UMR 8063, Paris, France 2Department of Bioengineering, Imperial College London, Exhibition Road, London SW7 2AZ, United Kingdom 3Institute of Geophysics, Swiss Federal Institute of Technology (ETH) Z€urich, Sonneggstrasse 5, 8092 Z€urich, Sw 4Department of Radiology, Stanford University, Stanford, California 94305, USA 5Department of Biomedical Engineering and Department of Electrical and Computer Engineering, University of Utah, Salt Lake City, Utah 84112, USA 6Department of Surgical Biotechnology, Division of Surgery and Interventional Science, University College London, London NW3 2PF, United Kingdom 7Earth Science and Engineering Department, Imperial College London, London, United Kingdom 8Centre of Excellence IT4Innovations, Faculty of Information Technology, Brno University of Technology, Bozetechova 2, Brno 612 00, Czech Republic 9Graduate Program in Acoustics, The Pennsylvania State University, University Park, Pennsylvania 16802, USA 10 9Graduate Program in Acoustics, The Pennsylvania State University, University Park, Pennsylvania 16802, USA 0Joint Department of Biomedical Engineering, University of North Carolina at Chapel Hill, Chapel Hill, North North Carolina State University, Raleigh, North Carolina 27695, USA 11Department of Electrical Engineering, Stanford University, Stanford, California 94305, USA 11Department of Electrical Engineering, Stanford University, Stanford, California 94305, USA 12 24 October 2024 05:17:09 24 October 2024 05:17:09 12Foundation for Research on Information Technologies in Society (IT’IS), Zurich, Switzerland 13Radiology and Clinical Neurosciences Departments, Cumming School of Medicine, University of Calgary, Calgary, Alberta, Canada 14Department of Applied Physics, University of Eastern Finland, 70211 Kuopio, Finland 15Department of Medical Physics and Biomedical Engineering, University College London, Gower Street, London WC1E 6BT, United Kingdom 16Technical University of Denmark, Kongens Lyngby, Denmark 17Institute for Mathematical and Computational Engineering, School of Engineering and Faculty of Mathematics, Pontiﬁcia Universidad Catolica de Chile, Santiago, Chile 17Institute for Mathematical and Computational Engineering, School of Engineering and Faculty of Mathematics, Pontiﬁcia Universidad Catolica de Chile, Santiago, Chile 17Institute for Mathematical and Computational Engineering, School of Engineering and Faculty of Mathematics, Pontiﬁcia Universidad Catolica de Chile, Santiago, Chile ARTICLE ................................... Benchmark problems for transcranial ultrasound simulation: Intercomparison of compressional wave modelsa) Jean-Francois Aubry,1 Oscar Bates,2 Christian Boehm,3 Kim Butts Pauly,4 Douglas Christensen,5 Carlos Cueto,2 Pierre Gelat,6 Lluis Guasch,7 Jiri Jaros,8 Yun Jing,9 Rebecca Jones,10 Ningrui Li,11 Patrick Marty,3 Hazael Montanaro,12,b) Esra Neufeld,12,c) Samuel Pichardo,13 Gianmarco Pinton,10 Aki Pulkkinen,14 Antonio Stanziola,15 Axel Thielscher,16,d) Bradley Treeby,15,e) and Elwin van ’t Wout17 1Physics for Medicine Paris, National Institute of Health and Medical Research (INSERM) U1273, ESPCI Paris, Paris Sciences and Lettres University, French National Centre for Scientiﬁc Research (CNRS) UMR 8063, Paris, France 2Department of Bioengineering, Imperial College London, Exhibition Road, London SW7 2AZ, United Kingdom 3Institute of Geophysics, Swiss Federal Institute of Technology (ETH) Z€urich, Sonneggstrasse 5, 8092 Z€urich, Switzerland 4Department of Radiology, Stanford University, Stanford, California 94305, USA 5Department of Biomedical Engineering and Department of Electrical and Computer Engineering, University of Utah, Salt Lake City, Utah 84112, USA 6Department of Surgical Biotechnology, Division of Surgery and Interventional Science, University College London, London NW3 2PF, United Kingdom 7Earth Science and Engineering Department, Imperial College London, London, United Kingdom 8C f ll 4 l f f h l f h l h 2 612 00 ARTICLE ................................... ARTICLE ................................... Articles You May Be Interested In Articles You May Be Interested In Viscoelastic transcranial wave propagation modeling for neuronavigated focused ultrasound procedures in humans J Acoust Soc Am (October 2022) Transcranial ultrasound simulation with uncertainty estimation JASA Express Lett. (May 2023) A simulation study on the sensitivity of transcranial ray-tracing ultrasound modeling to skull properties J. Acoust. Soc. Am. (August 2023) Viscoelastic transcranial wave propagation modeling for neuronavigated focused ultrasound procedures in humans J Acoust Soc Am (October 2022) 24 October 2024 05:17:09 Benchmark problems for transcranial ultrasound simulation: Intercomparison of compressional wave modelsa) Jean-Francois Aubry,1 Oscar Bates,2 Christian Boehm,3 Kim Butts Pauly,4 Douglas Christensen,5 Carlos Cueto,2 Pierre Gelat,6 Lluis Guasch,7 Jiri Jaros,8 Yun Jing,9 Rebecca Jones,10 Ningrui Li,11 Patrick Marty,3 Hazael Montanaro,12,b) Esra Neufeld,12,c) Samuel Pichardo,13 Gianmarco Pinton,10 Aki Pulkkinen,14 Antonio Stanziola,15 Axel Thielscher,16,d) Bradley Treeby,15,e) and Elwin van ’t Wout17 1Physics for Medicine Paris, National Institute of Health and Medical Research (INSERM) U1273, ESPCI Paris, Paris Sciences and Lettres University, French National Centre for Scientiﬁc Research (CNRS) UMR 8063, Paris, France 2Department of Bioengineering, Imperial College London, Exhibition Road, London SW7 2AZ, United Kingdom 3Institute of Geophysics, Swiss Federal Institute of Technology (ETH) Z€urich, Sonneggstrasse 5, 8092 Z€urich, Switzerland 4Department of Radiology, Stanford University, Stanford, California 94305, USA 5Department of Biomedical Engineering and Department of Electrical and Computer Engineering, University of Utah, Salt Lake City, Utah 84112, USA 6Department of Surgical Biotechnology, Division of Surgery and Interventional Science, University College London, London NW3 2PF, United Kingdom 7Earth Science and Engineering Department, Imperial College London, London, United Kingdom 8Centre of Excellence IT4Innovations, Faculty of Information Technology, Brno University of Technology, Bozetechova 2, Brno 612 00, Czech Republic 9Graduate Program in Acoustics, The Pennsylvania State University, University Park, Pennsylvania 16802, USA 10Joint Department of Biomedical Engineering, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina 27599, USA and North Carolina State University, Raleigh, North Carolina 27695, USA 11Department of Electrical Engineering, Stanford University, Stanford, California 94305, USA 12Foundation for Research on Information Technologies in Society (IT’IS), Zurich, Switzerland 13Radiology and Clinical Neurosciences Departments, Cumming School of Medicine, University of Calgary, Calgary, Alberta, Canada 14Department of Applied Physics, University of Eastern Finland, 70211 Kuopio, Finland 15 f d l h d d l C ll d G d C1 6 ARTICLE ................................... I. INTRODUCTION The working group met regularly throughout 2021. The list of benchmarks (discussed in Sec. II) was iteratively reﬁned, including the source deﬁnitions, the medium geome- try, the material properties, and the output domain size. File submission formats, mechanisms for data sharing, and com- parison metrics (along with codes to compute them)8 were also deﬁned. Benchmark submissions were non-blinded with multiple resubmissions allowed. The goal was not a competi- tion to establish which model was the “best” by some deﬁni- tion. Rather, it was to establish consensus on different approaches to transcranial ultrasound modeling and how to implement these correctly using a range of modeling tools available to the community. In this spirit, work-in-progress results and comparison metrics were discussed at regular inter- vals. These discussions, along with the sharing of code, approaches, and processing steps, etc., ultimately allowed the benchmarks to be computed with a wide set of simulation tools with excellent agreement (see Sec. IV). Ultrasound is increasingly used for therapeutic applica- tions in the brain, including for tissue ablation,1 opening of the blood-brain barrier,2 and modulation of brain activity.3 One challenge is the non-invasive delivery of ultrasound through the skull bone, which can signiﬁcantly distort and attenuate the transmitted waves.4 To account for this, com- puter simulations are now frequently used to make predic- tions of the intracranial pressure ﬁeld5 and to correct for phase aberrations due to the skull.6 This is particularly important for transcranial ultrasound stimulation (TUS), as the low ultrasound intensities make it highly challenging to measure the delivered energy in vivo, e.g., using magnetic resonance (MR)-guided thermometry.7 At a high level, there are four main steps in the setup of an acoustic model for transcranial ultrasound: (1) deﬁning the medium parameters, including the skull and soft-tissue geom- etry and the acoustic properties (using a medical image, for example); (2) deﬁning the transducer characteristics, including the geometry, driving parameters, and relative position; (3) deﬁning the numerical parameters for the model, including the grid resolution and boundary conditions; and (4) processing and interpreting the simulated results. One challenge for the community is that there is a large variation in these steps in the published literature, and there is currently little consensus on the best approach or the uncertainties associated with numerical modeling more generally. https://doi.org/10.1121/10.0013426 comparison, the median values for each benchmark for the difference in focal pressure and position are less than 10% and 1 mm, respectively. The benchmark deﬁnitions, model results, and intercomparison codes are freely available to facilitate further comparisons. comparison, the median values for each benchmark for the difference in focal pressure and position are less than 10% and 1 mm, respectively. The benchmark deﬁnitions, model results, and intercomparison codes are freely available to facilitate further comparisons. V C 2022 Author(s). All article content, except where otherwise noted, is licensed under a Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/). https://doi.org/10.1121/10.0013426 p V C 2022 Author(s). All article content, except where otherwise noted, is licensed under a Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/). https://doi.org/10.1121/10.0013426 (Received 9 February 2022; revised 18 July 2022; accepted 22 July 2022; published online 16 August 2022) [Editor: James F. Lynch] Pages: 1004– Pages: 1004–1019 I. INTRODUCTION 24 October 2024 05:17:09 The primary goal of this phase of the intercomparison exercise was to establish a series of benchmarks relevant to transcranial ultrasound, along with consensus on the correct numerical solutions for these benchmarks. Consequently, sim- ulations were typically performed with very high sampling to maximize accuracy. Because of this emphasis and the differ- ent computational resources available to each group, a com- parison of the computational performance of the individual models was considered out-of-scope from the outset. However, it is still important to note that some models used in the intercomparison, in particular, those based on the angular spectrum method, have an efﬁciency/accuracy trade-off inher- ent in their formulation (see, e.g., Ref. 9). This should be con- sidered when interpreting the intercomparison metrics presented in Sec. IV and the supplementary information.10 As part of the International Transcranial Ultrasonic Stimulation Safety and Standards (ITRUSST) consortium, a working group focused on simulation and planning was con- vened. The primary goal was to perform a modeling inter- comparison to systematically evaluate the steps involved in transcranial simulation, with a view to establishing best practice. A number of researchers active in the development of tools for transcranial ultrasound simulation were invited to take part. The ﬁrst phase, which is reported here, was a model intercomparison using a series of numerical bench- marks relevant to transcranial ultrasound where the medium parameters and transducer characteristics were well deﬁned. The primary research question was do different modeling techniques and computer codes give the same answer when the inputs to the model are well speciﬁed? This was taken as the ﬁrst step to ensure that any differences in more compli- cated scenarios (e.g., where the skull properties are mapped from a medical image or the transducer properties are mapped from a hydrophone measurement) could be evalu- ated as systematically and independently as possible. The ﬁnal output from the intercomparison exercise is a set of nine well-deﬁned numerical benchmarks relevant to transcranial ultrasound (with a total of 18 permutations of these benchmarks), along with publicly available simulation results for these benchmarks computed using 11 different modeling codes.8,11 ABSTRACT: Computational models of acoustic wave propagation are frequently used in transcranial ultrasound therapy, for example, to calculate the intracranial pressure ﬁeld or to calculate phase delays to correct for skull distortions. To allow intercomparison between the different modeling tools and techniques used by the community, an international working group was convened to formulate a set of numerical benchmarks. Here, these benchmarks are presented, along with intercomparison results. Nine different benchmarks of increasing geometric complexity are deﬁned. These include a single-layer planar bone immersed in water, a multi-layer bone, and a whole skull. Two transducer conﬁgurations are considered (a focused bowl and a plane piston operating at 500 kHz), giving a total of 18 permutations of the benchmarks. Eleven different modeling tools are used to compute the benchmark results. The models span a wide range of numerical techniques, including the ﬁnite-difference time-domain method, angular spectrum method, pseudospectral method, boundary-element method, and spectral-element method. Good agreement is found between the models, particularly for the position, size, and magnitude of the acoustic focus within the skull. When comparing results for each model with every other model in a cross- 1003 0001-4966/2022/152(2)/1003/17 J. Acoust. Soc. Am. 152 (2), August 2022 V C Author(s) 2022. V C Author(s) 2022. V C Author(s) 2022. . . https://doi.org/10.1121/10.0013426 500 kHz with a constant surface velocity of 0.04 m/s.17 Assuming an acoustic impedance of 1.5 megarayls, this is equivalent to modeling the sources as a distribution of free- ﬁeld monopole radiators with a source pressure of 60 kPa. linear wave propagation (previous studies have shown that nonlinear effects are negligible for typical TUS parame- ters)12 Only compressional waves were considered for this stage of the intercomparison. In some circumstances, elastic wave effects will also play a role, particularly if the ultra- sound waves are not close to normal incidence with respect to the skull bone,13,14 but these effects were not investigated here. All simulations were conducted in three dimensions. B. Transducer characteristics Two transducer deﬁnitions were used (see Fig. 1). The ﬁrst was a spherically curved transducer with a 64 mm radius of curvature and a 64 mm aperture diameter. This is representative of the single-element transducers frequently used for TUS.15 The second was a plane piston transducer with a diameter of 20 mm. Piston transducers are often used in multi-element arrays. While the typical diameter of an element in a multi-element array is smaller than 20 mm, this diameter was used to provide identiﬁable beam characteris- tics within the simulation domain. For some numerical tech- niques, piston transducers are easier to model, particularly when aligned with the computational grid, which avoids staircasing artifacts.16 Both transducers were driven at C. Material properties The material properties used for the benchmarks are given in Table I (all materials are modeled as acoustic media, i.e., ﬂuids). These are intended to be representative (rather than deﬁnitive) values and were taken from the range presented in the literature.18–25 For the simulations including absorption, the loss is deﬁned to be non-dispersive, i.e., either frequency independent or, for power law models, dependent on frequency squared. A. Overview The benchmarks were deﬁned considering typical TUS scenarios, although they are also relevant to other therapeu- tic applications of transcranial ultrasound. Simulations were single frequency (time-harmonic) and performed assuming 1004 J. Acoust. Soc. Am. 152 (2), August 2022 Aubry et al. F. Benchmarks A total of nine benchmarks relevant to transcranial ultrasound were devised. These are summarized in Table II. The benchmarks gradually increase in complexity, both add- ing additional tissue layers and increasing the geometric complexity of the skull. Benchmarks 1–7 are illustrated in Fig. 1, while benchmarks 8 and 9 are illustrated in Fig. 2. Regardless of the sampling or mesh used for the simula- tions, the outputs stored in the comparison ﬁles were resampled onto a uniform Cartesian grid with 0.5 mm grid sampling. This corresponds to six points per wavelength (PPW) in water. For benchmarks 1–6, the comparison domain size was a 120 70 mm (axial lateral) slice through the central z-coordinate, corresponding to a grid size of 241 141 grid points. For benchmark 7, the comparison domain was 120 70 70 mm (241 141 141 grid points). For bench- mark 8, the comparison domain was 225 170 190 mm (451 341 381 grid points). For benchmark 9, the compari- son domain was 212 224 184 mm (425 449 369 grid points). Benchmark 1 considers the bowl and piston transducers in water (free-ﬁeld) using the properties given in Table I. Benchmark 2 adds uniform artiﬁcial absorption of 1 dB/cm at 500 kHz. During the intercomparison exercise, these benchmarks served as a helpful reference to ensure the transducer properties, absorption units, and comparison domain were correctly speciﬁed. For these simulations, ref- erence simulations were also computed using the fast near- ﬁeld method as implemented in the FOCUS toolbox.26–28 Calculations using FOCUS were performed using 5000 inte- gration points to give a high level of precision. Several mod- els used the ﬁelds computed using FOCUS across a transverse y-z plane as the source deﬁnition (see Sec. III). 24 October 2024 05:17:09 For all benchmarks, the transducer was oriented such that the beam axis pointed in the x dimension, with the transducer positioned in the center of the y/z dimensions. Using 1-based indexing, the center of the source (rear of the bowl or center of the piston) relative to the output grid was positioned at [1, 71] for benchmarks 1–6, [1, 71, 71] for benchmark 7, [1, 171, 191] for benchmark 8, and [1, 225, 185] for benchmark 9. Note the comparisons for benchmarks 1–6 were made in two dimensions due to the axisymmetry of the geometry. https://doi.org/10.1121/10.0013426 TABLE I. Compressional sound speed (c), mass density (q), and absorption coefﬁcient (a) used in the benchmark simulations. E. Naming convention TABLE I. Compressional sound speed (c), mass density (q), and absorption coefﬁcient (a) used in the benchmark simulations. c (m/s) q (kg/m3) a (dB/cm at 500 kHz) Water 1500 1000 0 Skin 1610 1090 0.2 Brain 1560 1040 0.3 Cortical bone 2800 1850 4 Trabecular bone 2300 1700 8 The benchmarks were given unique identiﬁers in the following format: PH<NUM>-BM<NUM>-SC<NUM>. PH (phase) identiﬁes the intercomparison phase (in this case 1). BM (benchmark) identiﬁes the benchmark number within the phase. SC (source) identiﬁes the source condition, where 1 is the bowl source and 2 is the plane piston source. A sum- mary of the different benchmarks is given in Table II. File names for the intercomparison results follow the same con- vention with the model name appended (see Table IV): PH<NUM>-BM<NUM>-SC<NUM>_<MODELNAME>. The simulation outputs for each model for each benchmark are publicly available.11 number of periods, and then extracting the amplitude and phase at the driving frequency using a Fourier transform. Note that the phase is optional and was not used for the compari- sons presented in Sec. IV but was included for completeness. The results were saved either as MATLAB .mat ﬁles using the “-v7.3” ﬂag where possible (this format can be easily opened as a HDF5 ﬁle outside MATLAB) or as HDF5 .h5 ﬁles with the variables saved as datasets in the root group. D. Simulation outputs The simulation results were stored as two variables named p_amp and p_phase. These represent the ampli- tude and phase of the complex pressure ﬁeld at 500 kHz over the speciﬁed comparison domain. For time domain models, these parameters can be extracted precisely by set- ting the time step to an integer number of points per period (PPP), recording the steady-state pressure ﬁeld for an integer 24 October 2024 05:17:09 24 October 2024 05:17:09 FIG. 1. (Color online) Transducer deﬁnitions and simulation layouts for benchmarks 1–7. Benchmarks 1–6 use a two-dimensional (2D) comparison domain of 120 mm (axial) by 70 mm (lateral) through the central z plane. Benchmark 7 uses a 3D comparison domain of 120 70 by 70 mm. The material properties used are given in Table I. J Acoust Soc Am 152 (2) August 2022 Aubry et al 1005 FIG. 1. (Color online) Transducer deﬁnitions and simulation layouts for benchmarks 1–7. Benchmarks 1–6 use a two-dimensional (2D) comparison domain of 120 mm (axial) by 70 mm (lateral) through the central z plane. Benchmark 7 uses a 3D comparison domain of 120 70 by 70 mm. The material properties used are given in Table I. J. Acoust. Soc. Am. 152 (2), August 2022 Aubry et al. 1005 J. Acoust. Soc. Am. 152 (2), August 2022 J. Acoust. Soc. Am. 152 (2), August 2022 J. Acoust. Soc. Am. 152 (2), August 2022 Aubry et al. 1005 1005 F. Benchmarks All simulations were con- ducted in three dimensions. Benchmark 3 introduces a single ﬂat 6.5 mm layer of cortical bone immersed in water, positioned 30 mm from the transducer as shown in Fig. 1. Benchmark 4 extends this to include a 4 mm skin layer, a three-layered skull (consisting of 1.5 mm cortical bone for the outer table, 4 mm trabecular bone, and 1 mm cortical bone for the inner table, giving the same overall skull thickness and position as benchmark 3) TABLE II. Summary of benchmarks in phase 1 of the intercomparison. SC1 corresponds to the focused bowl transducer and SC2 to the plane piston trans- ducer. Outputs are resampled to a regular Cartesian mesh with a grid spacing of 0.5 mm. Simulation layouts are shown in Figs. 1 and 2. gp ¼ grid points. TABLE II. Summary of benchmarks in phase 1 of the intercomparison. SC1 corresponds to the focused bowl transducer and SC2 to the plane piston trans- ducer. Outputs are resampled to a regular Cartesian mesh with a grid spacing of 0.5 mm. Simulation layouts are shown in Figs. 1 and 2. gp ¼ grid points. ducer. Outputs are resampled to a regular Cartesian mesh with a grid spacing of 0.5 mm. Simulation layouts are shown in Figs. 1 and 2. gp ¼ grid points. Label Description Output grid size PH1-BM1-SC1/2 Water (lossless) 120 70 mm (241 141 gp) PH1-BM2-SC1/2 Water (artiﬁcial absorption of 1 dB/cm at 500 kHz) 120 70 mm (241 141 gp) PH1-BM3-SC1/2 Flat, single-layer skull (cortical bone) in water 120 70 mm (241 141 gp) PH1-BM4-SC1/2 Flat, skin, three-layered skull, and brain 120 70 mm (241 141 gp) PH1-BM5-SC1/2 Curved, single-layer skull (cortical bone) in water 120 70 mm (241 141 gp) PH1-BM6-SC1/2 Curved, skin, three-layered skull, and brain 120 70 mm (241 141 gp) PH1-BM7-SC1/2 Truncated skull mesh in water, target in visual cortex 120 70 70 mm (241 141 141 gp) PH1-BM8-SC1/2 Whole skull mesh, target in visual cortex 225 170 190 mm (451 341 381 gp) PH1-BM9-SC1/2 Whole skull mesh, target in motor cortex 212 224 184 mm (425 449 369 gp) 1006 J. Acoust. Soc. Am. 152 (2), August 2022 Aubry et al. FIG. 2. https://doi.org/10.1121/10.0013426 FIG. 2. (Color online) Simulation lay- outs for benchmarks 8 (top row) and 9 (bottom row) showing the central x-y and x-z slices. The position of the bowl transducer is shown for reference. Benchmark 7 (shown in Fig. 1) uses a subset of the skull mask and the same relative transducer position as bench- mark 8, with a reduced comparison domain size as shown with the dashed line. The material properties used are given in Table I. 24 October 2024 05:17:09 The skull mesh was stored as two .stl ﬁles represent- ing the inner and outer surfaces of the skull bone. Position transforms were stored as three-dimensional (3D) afﬁne transformations that position the transducer relative to the coordinates in the .stl ﬁles. Grid-based discretizations containing a binary skull mask were also generated using the iso2mesh MATLAB toolbox.34,35 These were generated on a regular Cartesian mesh at a range of resolutions after applying the appropriate inverse position transforms (to move the skull mesh relative to the transducer) and were truncated to the appropriate comparison domain (see Sec. II D). The skull ﬁles and position transforms are available alongside the simulation results.11 with water on the exterior and brain on the interior as shown in Fig. 1. The thickness values are based on average values for parietal bone29 and scalp.30 Benchmark 5 increases the geometric complexity of benchmark 3 by using a curved 6.5 mm layer of cortical bone immersed in water, with inner and outer radii of 68.5 and 75 mm, respectively. Note that the bone layer is spheri- cally (not cylindrically) curved, meaning the curvature in the out-of-plane dimension is the same as that shown in Fig. 1. Benchmark 6 is a curved extension of benchmark 4, where the thickness values correspond to differences in the curvature radii. Benchmarks 7–9 increase the geometric complexity fur- ther by using a homogeneous skull mesh generated from the MNI152_T1_1 mm magnetic resonance imaging template brain.31,32 The template image was run through an adapted version of SimNIBS headreco.33 Additional smoothing of the tissue surfaces while simultaneously preventing intersec- tions between neighboring surfaces was performed using SimNIBS functions. Benchmarks 7 and 8 use a transducer position targeted at the foveal representation of the primary visual cortex, while benchmark 9 uses a transducer position targeted at the hand area of the primary motor cortex. J. Acoust. Soc. Am. 152 (2), August 2022 F. Benchmarks (Color online) Simulation lay- outs for benchmarks 8 (top row) and 9 (bottom row) showing the central x-y and x-z slices. The position of the bowl transducer is shown for reference. Benchmark 7 (shown in Fig. 1) uses a subset of the skull mask and the same relative transducer position as bench- mark 8, with a reduced comparison domain size as shown with the dashed line. The material properties used are given in Table I. https://doi.org/10.1121/10.0013426 https://doi.org/10.1121/10.0013426 TABLE IV. Summary of models used to calculate the benchmark results. Additional details are given in the supplementary material (Ref. 10). Authors correspond to the authors of the current manuscript directly con- tributing to the intercomparison exercise, not necessarily the authors of the model. TABLE III. Difference metrics used for the intercomparison. Here, p1 and p2 are the amplitude of the pressure ﬁeld over the 2D or 3D comparison domains for the reference ﬁeld and comparison ﬁeld, respectively (these are assumed to be positive). Sums and maximum values are assumed to be over all values in the comparison domain starting from the exit plane of the transducer. Focal values are taken from inside the brain (or post-skull) region only. pos max is used to denote the position of the maximum value in the comparison domain. Label Authors Domain Method BABELVISCOFDTD S.P. Time FDTDa FULLWAVE R.J., G.P. Time FDTD GMFDTD A.P. Time FDTD HAS N.L., K.B.P. Frequency HASb JWAVE A.S. Frequency Fourier spectral method with iterative solver KWAVE B.T., J.J. Time Pseudospectral time domain MSOUND Y.J. Frequency Modiﬁed angular spectrum OPTIMUS P.G., E.v.W. Frequency BEMc SALVUS P.M., C.B. Time Spectral-element SIM4LIFE H.M., E.N. Time FDTD STRIDE C.C., O.B., L.G. Time FDTD aFinite-difference time-domain (FDTD). bHybrid angular spectrum (HAS). cBoundary-element method (BEM). Metric Definition Relative L2 ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ P ðp1 p2Þ2 P p2 1 s Relative L1 maxjp1 p2j maxðp1Þ Focal (peak) pressure jmaxðp1Þ maxðp2Þj maxðp1Þ Focal position jjpos maxðp1Þ pos maxðp2Þjj2 However, for more complex geometries, these become dom- inated by differences in the rapidly varying near-ﬁeld region between the source and the skull. For this reason, differ- ences in the focal characteristics were also computed. This included the magnitude and position of the peak pressure within the brain and differences in the full-width at half- maximum (FWHM) and 6 dB focal volume. The FWHM values were taken in each Cartesian direction present in the comparison domain (i.e., in the x and y dimensions for benchmarks 1–6 and in the x, y, and z dimensions for bench- marks 7–9). For benchmarks 3–9 for the piston transducer, the acoustic ﬁeld gradually decays within the brain; thus, there is no natural focus in the axial direction. In this case, the axial focal position and lateral proﬁles and FWHM val- ues were taken at x ¼ 60 mm (corresponding to a grid index of 121). A. Overview A total of 11 modeling tools were used for the inter- comparison, in addition to the free-ﬁeld reference values calculated using FOCUS discussed in Sec. II F. These are sum- marized in Table IV. A short description of each model is given in Secs. III B–III L, with additional details given in the supplementary material.10 C. FULLWAVE FULLWAVE2 3D solves the wave equation with quadratic nonlinearity and multiple relaxations using a staggered-grid FDTD approach with fourth-order accuracy in time and var- iable accuracy in space.43,44 This model uses a staggered- grid Cartesian mesh with a convolutional PML at the bound- aries, utilizing high-order adaptive stencils that minimize dispersion and dissipation errors. The source and output can take the shape of any arbitrary geometry that can be deﬁned on a Cartesian grid, with sources modeled either as free- https://doi.org/10.1121/10.0013426 For benchmarks 7–9, differences in the 6 dB focal volume were also computed. The focal volume was calcu- lated by thresholding the pressure ﬁeld inside the brain to 50% of the maximum value and then counting the voxels in the largest connected component. Code to compute the intercomparison metrics is available on GitHub.8 Calculations are solved using a 4th-order in space and 2nd- order in time FDTD scheme in Cartesian coordinates.37,38 Calculations are solved using a 4th-order in space and 2nd- order in time FDTD scheme in Cartesian coordinates.37,38 Stress tensors and displacement vectors are solved a half time step separated from each other. Attenuation losses are modeled using a quality factor for narrowband condi- tions.38,39 Liquid-bone interfaces and heterogeneity of tissue material are modeled using averaging operators.40 Optional reduction of staircasing artifacts can be enabled using a superposition operator.41 A perfectly matched layer (PML) condition for viscoelastic propagation is used to absorb waves at the boundaries.42 24 October 2024 05:17:09 All benchmarks were computed using a resolution of 12 grid PPW. Sources were modeled as stress nodes using the same staircase-free formulation and dispersion correction as in the k-Wave model (see Sec. III G). The time PPP for benchmarks 1 and 2 was 25, and for benchmarks 3–9, it was 48. Benchmarks 1–7 used a total grid size of 305 305 521 grid points, including the PML. For benchmarks 8 and 9, the grid size was, respectively, 785 705 941 and 761 921 889. Simulation outputs were resampled to the comparison grid using a spline interpolation of order 3. G. Intercomparison metrics A number of metrics were chosen to compare the simu- lated ﬁelds. Mathematical deﬁnitions for some metrics are given in Table III. Metrics based on the entire ﬁeld were taken from the exit plane of the source, excluding the ﬁrst grid point in the x-direction for the piston transducer and the ﬁrst 19 grid points in the x-direction for the bowl transducer. The relative L2 and L1 errors provide a useful (and strict) measure of the overall differences between simulations. Aubry et al. 1007 J. Acoust. Soc. Am. 152 (2), August 2022 1007 F. JWAVE JWAVE simulates the solution of time-harmonic wave propagation problems by solving the heterogeneous Helmholtz equation in the complex domain, using a regular Fourier spectral collocation method and linear iterative solv- ers such as restarted generalized minimal residual method (GMRES).49,50 Absorbing boundary conditions are enforced using a PML,51 while the deﬁnition of the sources is done by projecting them on the discrete collocation grid by approximately convolving them with the band limited inter- polant.52 The source ﬁeld is modeled as a mass source. JWAVE is a PYTHON software written using JaxDF,53 which in turn is based on JAX.54 The code is just-in-time compiled for the hardware at hand [e.g., graphical processing units (GPUs) or tensor processing units (TPUs)] and allows for automatic differentiation to be applied with respect to any continuous parameter. 24 October 2024 05:17:09 All simulations were computed using 12 PPW and 75 PPP. Grid sizes for the simulations were 576 376 376 grid points for benchmarks 1–7, 996 776 856 for bench- mark 8, and 944 992 832 for benchmark 9. The grid sizes include a 48 grid point absorbing layer surrounding the domain, which had attenuation linearly increasing from zero to 50 Np/m, corresponding to about 94% amplitude attenua- tion for a normally incident reﬂected wave. Simulations were computed for 3900 time steps for benchmarks 1–7, 15 075 time steps for benchmark 8, and 15 075 time steps for benchmark 9. The simulations produced a complex val- ued steady-state pressure ﬁeld, which was resampled to the comparison grid using spline interpolation before computing the pressure amplitude and the phase angle. Benchmarks 1 and 2 were computed using 6 PPW, while benchmarks 3–7 were computed using 12 PPW. The PML size was ﬁxed to 30 voxels. To reduce the computation time of the FFTs, the domain dimensions were padded to the nearest integers with prime factors smaller than 7. When required, the results were resampled to the intercomparison grid using Fourier interpolation. Benchmarks 8 and 9 were too large for the available computational resources, so results for these benchmarks were not computed. B. BABELVISCOFDTD BABELVISCOFDTD solves the viscoelastic wave equation expressed in stress tensors and displacement vectors, where the bone material is modeled as a viscoelastic isotropic medium.23 The term “Babel” refers to the multiple comput- ing backends (CUDA, OpenCL, Apple Metal, and X86–64) that are supported for calculations. Nodes of stress and dis- placement are placed in a staggered-grid arrangement.36 Aubry et al. 1008 J. Acoust. Soc. Am. 152 (2), August 2022 https://doi.org/10.1121/10.0013426 ﬁeld particle displacement, velocity, or a monopole pressure source. using a spatial step between each spatial-frequency step. Reﬂected pressures are saved, backpropagated, and summed with the incident pressure ﬁeld, and this process is repeated until convergence to produce the ﬁnal steady-state pressure ﬁeld. For the benchmark comparison, the bowl and piston geometries were modeled as monopole pressure sources on a Cartesian grid, emitting a continuous sinusoidal wave. All benchmarks were computed with 12 PPW and 60 PPP, giv- ing a Courant–Fredrichs–Lewy condition (CFL)45 of 0.2. This created a simulation grid 2 times the size of the com- parison grid. To account for this, the simulations were run with a spatial step size of 2 voxels in each direction, down- sampling the output grid to the comparison grid size. The output over one steady-state cycle was then scaled based on the CFL and driving signal to account for the use of additive sources. Initial pressure ﬁelds were computed using the fast near-ﬁeld method as implemented in the FOCUS toolbox (see Sec. II F). Benchmarks 1–6 were computed using a grid size of 1001 1001 1001 with 6 PPW in the transverse direc- tions and 24 PPW in the axial direction. Benchmarks 7 and 8 were computed using a grid size of 1401 1401 501 with an isotropic resolution of 12 PPW. Benchmark 9 was computed using a grid size of 1201 1201 501 with an isotropic resolution of 12 PPW. Calculated pressure ﬁelds were resampled to the comparison grid using bilateral interpolation. D. GMFDTD The GMFDTD model simulates acoustic wave propagation based on coupling of the second-order acoustic and visco- elastic wave equations using a combined grid method and FDTD method. The model operates using a regular Cartesian mesh. For ﬂuid simulations (as described in this work), GMFDTD solves the acoustic wave equation using a FDTD approach with fourth-order spatial and second-order time stencils. First-order absorbing boundary conditions are used on exterior boundaries of the simulation domain. A ﬁnite thickness absorbing layer was placed on the exterior boundaries to further reduce acoustic reﬂections. A hetero- geneous Neumann boundary condition is used to model the sound sources making the source-medium interface work as an acoustically hard reﬂector for incoming sound waves. For a more thorough description of the model, see Ref. 46. G. KWAVE KWAVE solves three coupled equations equivalent to a generalized Westervelt equation, where spatial gradients are calculated using a Fourier collocation spectral method, and time integration is performed using a dispersion-corrected ﬁnite-difference scheme.55,56 Calculations are performed on a regular Cartesian mesh with a space and time staggered grid. A split-ﬁeld PML is used to absorb the waves at the domain boundaries. Sources are modeled as free-ﬁeld monopoles (injection of mass) using a staircase-free formu- lation to represent the bowl and piston geometries52 and a dispersion-corrected time-stepping scheme.57 KWAVE solves three coupled equations equivalent to a generalized Westervelt equation, where spatial gradients are calculated using a Fourier collocation spectral method, and time integration is performed using a dispersion-corrected ﬁnite-difference scheme.55,56 Calculations are performed on https://doi.org/10.1121/10.0013426 Benchmarks 1–6 were computed using the axisymmetric version of k-Wave to provide a high-resolution reference sim- ulation.58 Benchmarks 1, 2, 3, and 5 used 60 PPW and 2400 PPP, while benchmarks 4 and 6 used 60 PPW and 6000 PPP. In both cases, the total grid size was 2700 864 grid points, including the PML, and the simulation time was 120 ls, giv- ing 144 000 and 360 000 time steps, respectively. Benchmarks 7–9 were computed using the 3D version of k-Wave opti- mized for high performance computing clusters.59 Benchmark 7 used 30 PPW and 1200 PPP, with a grid size of 1296 768 768 grid points and 72 000 time steps (120 ls simula- tion time). Benchmarks 8 and 9 used 18 PPW and 360 PPP with 72 000 time steps (400 ls simulation time). The grid sizes were 1458 1080 1200 and 1350 1440 1152, respectively. The simulation times were sufﬁcient to reach steady state and were chosen via a convergence test. All simu- lations used a grid spacing that was an integer division of the comparison resolution (0.5 mm); thus, simulation outputs were resampled to the comparison grid using decimation. hierarchical matrix compression.65 The convergence of the iterative GMRES linear solver was improved with OSRC pre- conditioning.66 All models were implemented in PYTHON, using version 3 of the open-source BEMPP library.67 The triangular surface meshes were created with Gmsh68 for benchmarks 3–6 and using Meshmixer69 for benchmarks 7–9. The size of the mesh elements was speciﬁed as 4.3 PPW (0.7 mm) in benchmark 3, 6 PPW (0.5 mm) in benchmarks 4 and 5, 4 PPW (0.75 mm) for benchmark 6, and 10 PPW (0.3 mm) for benchmark 7. A compromise in terms of mem- ory requirements and accuracy of results had to be sought on benchmarks 8 and 9, and a value of 4 PPW (0.75 mm) was used on the skull mesh in the vicinity of the transducer with a value of 2.4 PPW (1.25 mm) elsewhere. The bowl and piston transducers were implemented using a Rayleigh integral for- mulation, consisting of a summation of evenly spaced mono- pole radiators positioned on their surface. The transducer surfaces were discretized using 23 and 6 monopole sources per wavelength in water for benchmarks 1–7 and benchmarks 8 and 9, respectively. https://doi.org/10.1121/10.0013426 In cases where the position of mono- pole sources coincided with a ﬁeld evaluation point, NaN was assigned to the acoustic pressure. The acoustic ﬁeld was eval- uated from the surface potentials by interpolation for points on, or very close to, the material interface and with Green’s functions for points in the material volume. I. OPTIMUS OPTIMUS is a full wave solver based on the BEM.61,88 The BEM employs the Green’s function of the Helmholtz equation to reformulate the volumetric wave problem into a boundary integral equation at the interfaces of piecewise homogeneous domains embedded in free space.62 Benchmarks 3, 5, and 7 were modeled with the Poggio–Miller–Chew–Harrington– Wu–Tsai (PMCHWT) formulation,63 benchmarks 4 and 6 were solved with a multi-trace formulation,64 and a nested ver- sion of the PMCHWT formulation solves benchmarks 8 and 9. The numerical discretization leads to a dense system of linear equations, whose computational footprint is reduced through H. MSOUND MSOUND solves the Helmholtz equation with the absorp- tion term for linear acoustics cases.60 For layered media, the conventional angular spectrum approach coupled with the analytical plane wave transmission and reﬂection coefﬁ- cients is used. For arbitrarily heterogeneous media, a split- step Fourier method with interpolation is used. Calculations are performed on a regular Cartesian mesh in space. A non- reﬂecting layer can be used to reduce the spatial aliasing error. Sources are modeled by assigning the complex pres- sure distribution on the initial plane. In these simulations, the initial plane pressure ﬁelds were obtained by FOCUS, as MSOUND currently only considers the pressure-release bound- ary condition (p ¼ 0) for the region outside the source. All benchmarks were computed using the function Forward3D_fund. Benchmarks 1 and 2 were computed using 6 PPW in all directions. Benchmarks 7–9 were com- puted using 12 PPW in all directions. Benchmarks 3 and 4 were computed using 6 PPW in the lateral directions and 48 PPW in the axial (propagation) direction. Benchmarks 5 and 6 were computed using 6 PPW in the lateral directions and 24 PPW in the axial direction. For benchmarks 3–9, simula- tion outputs were down-sampled to the comparison grid. 24 October 2024 05:17:09 J. SALVUS SALVUS solves the second-order linear wave equation in the time domain and can handle acoustic and elastic media.70 It utilizes a matrix-free implementation of the continuous- Galerkin spectral-element method71 and an explicit second- order Newmark time-stepping scheme. The computational domain is discretized using unstructured conforming hexahe- dral meshes,72 which enable the exact representation of inter- faces and discontinuities in the tissue parameters. Absorbing boundaries are imposed using the ﬁrst-order Sommerfeld radiation condition in addition to sponge layers.73 The trans- ducers are modeled as a collection of monopole point sources distributed evenly over the surface of the transducer. Spectral elements of order 4 were utilized for all simu- lations; this corresponds to 125 nodes per element. Due to the interfaces being represented precisely using hexahedral meshes generated within Coreform Cubit 2021.5,74 utilizing 2–3 elements per wavelength for all benchmarks proved to be sufﬁcient. The maximum pressure distributions were computed by propagating the waveﬁeld in the time domain and then applying the on-the-ﬂy temporal Fourier trans- form.75 All simulation results were output on the same hexa- hedral discretizations used as inputs and were subsequently resampled onto the comparison grid using fourth-order Lagrange polynomials in the spectral-element basis. E. HAS The HAS method is a generalization of the angular spec- trum method, enabling propagation of pressure ﬁelds in het- erogeneous media.47,48 An initial pressure distribution is ﬁrst deﬁned on a plane perpendicular to the direction of propagation. To produce the full 3D steady-state pressure ﬁeld, pressures on subsequent planes are calculated in the spatial-frequency domain by solving the Helmholtz equation using the angular spectrum method. Errors due to local var- iations in attenuation and acoustic velocity are corrected for a regular Cartesian mesh with a space and time staggered grid. A split-ﬁeld PML is used to absorb the waves at the domain boundaries. Sources are modeled as free-ﬁeld monopoles (injection of mass) using a staircase-free formu- lation to represent the bowl and piston geometries52 and a dispersion-corrected time-stepping scheme.57 Aubry et al. 1009 J. Acoust. Soc. Am. 152 (2), August 2022 https://doi.org/10.1121/10.0013426 amplitudes over the comparison domains given in Table II. The beam shapes for benchmarks 1 and 2 are characteristic of focused bowl and unfocused piston transducers. The introduction of a ﬂat skull bone with a single layer (bench- mark 3) or multiple layers (benchmark 4) causes a drop in the focal pressure. Hot-spots (localized regions of increased pressure) are introduced on the skull surface, and the reﬂected waves generate a complex interference pattern between the transducer and the skull. For the focused bowl transducer (PH1-BM3-SC1 and PH1-BM4-SC1), the reﬂected waves also generate a secondary focus near the rear surface of the transducer. When a curved skull is used (benchmarks 5 and 6), the hot-spots and secondary focus are reduced. For all benchmarks with the piston transducer, a distinct last-axial maximum is no longer present after the introduction of the skull. Instead, the spatial peak pressure is typically either inside or immediately adjacent to the skull bone, and the acoustic beam gradually diverges after the skull surface. The introduction of a more complex skull geometry in benchmarks 7–9 generates additional features in the pressure ﬁelds. For benchmarks 7 and 8, the internal occipital protuberance of the skull bone causes a noticeable deﬂection of the acoustic beam. The use of the whole skull for benchmarks 8 and 9 also introduces small amplitude reﬂections from the opposite skull surface (e.g., see PH1- BM8-SC2 in Fig. 4). method on adaptive, rectilinear meshes (to adapt grid-steps to the local wavelength and reﬁne relevant geometric fea- tures) with cell-centered pressure degrees-of-freedom. Flux conserving virtual auxiliary points are used to improve accu- racy at interfaces and boundaries, and PMLs—according to the stretched coordinate formulation76—are used to avoid reﬂections at domain boundaries (for more details on the numerical methods, see Ref. 77). Results can be recorded as phasors (at the base frequency and, if relevant, higher har- monics) or transient 3 þ 1D ﬁelds, and the solver has been veriﬁed and validated,78 also for transcranial focused ultra- sound modeling.79,80 The original hard sources (imposed pressure; sinusoidal with rise time or user-deﬁned transient proﬁles) were extended for the purpose of this work by soft sources (cosine function to avoid slowly decaying low fre- quency components). L. STRIDE STRIDE solves the second-order, isotropic, linear acoustic wave equation using an FDTD approximation over a rectan- gular Cartesian grid,81 which is generated using the domain- speciﬁc language Devito.82 Spatial derivatives are calcu- lated using a 10th-order ﬁnite-difference approximation, while time integration is performed using a 4th-order time- stepping scheme optimized for increased stability.83 Acoustic waves at the boundaries are absorbed using either a sponge absorbing boundary84 or a complex frequency- shifted PML.85 Sources are introduced as free-ﬁeld monop- oles, which can be deﬁned at locations both on and off the grid.86 Benchmarks 1–7 were computed using 24 PPW and 120 PPP, resulting in a grid size of 1061 661 661, including absorbing boundaries. Benchmarks 8 and 9 were computed using 18 PPW and 90 PPP, with grid sizes of 1451 1121 1241 and 1373 1445 1205, respectively. A complex frequency-shifted PML was used as the absorbing boundary for all benchmarks. Computed results were resampled onto the comparison mesh using linear interpolation. Figure 5 gives a summary of the L1 and L2 intercom- parison metrics computed across the comparison domains outlined in Table II. Results are presented for each bench- mark (summarizing the cross-comparison results across all codes) and for each code (summarizing the cross- comparison results across all benchmarks). For benchmarks 1 and 2 (water and water with artiﬁcial absorption), the level of agreement is very high. For the bowl transducer, seven https://doi.org/10.1121/10.0013426 The present benchmarks were simulated using isotropic voxel meshes (24 voxels per wavelength, 0.125 mm resolu- tion) over the prescribed simulation domain padded with 96 layers of inhomogeneous PML, with a time step chosen to satisfy the CFL stability criterion (0.026 ls or 76.9 PPP with bone, 0.048 ls or 41.7 PPP without). Fifty periods were sim- ulated for benchmarks 1–7 (561 561 961 voxels), while 200 periods were simulated for benchhmark 8 (1521 1361 1801) and benchmark 9 (1473 1793 1697). To facili- tate comparison, voxeling was offset by half a cell compared to the deﬁned transducer surface, such that transducer grid points correspond to voxel cell centers and material interfa- ces to voxel faces. 24 October 2024 05:17:09 B. Difference metrics Aggregated difference metrics are given in Figs. 5–7. These were calculated by comparing each model with every other model in a cross-comparison and then computing the metrics described in Sec. II G. The box plots (generated using boxchart in MATLAB) illustrate the minimum, maxi- mum, median, and ﬁrst and third quartiles, along with any outliers. The same metrics were also computed for each model and benchmark using KWAVE as a reference. This ref- erence was used due to the very high spatial and temporal sampling possible for the KWAVE simulations, particularly for benchmarks 1–6, which allowed an axisymmetric formu- lation to be used. Field plots, axial and lateral proﬁles, dif- ference plots, and summary tables against FOCUS (for benchmarks 1 and 2) and KWAVE (for benchmarks 1–9) for each model are given in the supplementary material.10 These outputs are grouped both by benchmark and by model for ease of reference. Note that the simulation results and the comparison codes are freely available;8,11 thus, it is straightforward to generate other comparisons as required or add new modeling results to the intercomparison. 2 K. SIM4LIFE SIM4LIFE solves acoustic pressure wave equations (linear, or Westervelt–Lighthill, which considers dispersion and fre- quency mixing), using a multi-GPU-accelerated FDTD Aubry et al. 1010 J. Acoust. Soc. Am. 152 (2), August 2022 A. Field characteristics Representative simulation results for all benchmarks are given in Figs. 3 and 4. These illustrate the pressure Aubry et al. 1011 Aubry et al. 1011 J. Acoust. Soc. Am. 152 (2), August 2022 https://doi.org/10.1121/10.0013426 https://doi.org/10.1121/10.0013426 FIG. 3. (Color online) Pressure amplitudes computed using KWAVE for benchmarks 1–6 showing x-y slices through the central z plane for a comparison domain of 120 mm (axial) by 70 mm (lateral). 24 October 2024 05:17:09 24 October 2024 05:17:09 24 October 2024 05:17:09 FIG. 3. (Color online) Pressure amplitudes computed using KWAVE for benchmarks 1–6 showing x-y slices through the central z plane for a comparison domain of 120 mm (axial) by 70 mm (lateral) FIG. 3. (Color online) Pressure amplitudes computed using KWAVE for benchmarks 1–6 showing x-y slices through the central z plane for a comparison domain of 120 mm (axial) by 70 mm (lateral). models have L1 values of less than 1% when compared to FOCUS, and all values are less than 10% (see supplementary material). For the piston transducer, the simulations are slightly less accurate. Four models have L1 values of less than 1% when compared to FOCUS, and the maximum L1 value against FOCUS is 15%. Examining the difference plots (see supplementary material),10 the largest differences are in the complex near-ﬁeld pattern close to the transducer sur- face, where the pressure varies rapidly. models have L1 values of less than 1% when compared to FOCUS, and all values are less than 10% (see supplementary material). For the piston transducer, the simulations are slightly less accurate. Four models have L1 values of less than 1% when compared to FOCUS, and the maximum L1 value against FOCUS is 15%. Examining the difference plots (see supplementary material),10 the largest differences are in the complex near-ﬁeld pattern close to the transducer sur- face, where the pressure varies rapidly. models have median L1 values less than 10% across all benchmarks when compared to KWAVE [see Fig. 5(b)]. However, in general, the differences are larger than those found for benchmarks 1 and 2. Examining the difference plots (see supplementary material),10 the largest variations are in the region between the transducer and skull bone. A. Field characteristics These arise due to a combination of errors in modeling the near-ﬁeld of the transducer, even in free-ﬁeld (described above), along with errors in modeling the reﬂection from the bone and soft- tissue surfaces (e.g., due to errors in the positions of the inter- faces and amplitude and phase errors in the reﬂected waves). 2 For benchmarks 3–9, the L1 and L2 metrics both increase noticeably, with median values for the cross- comparison between 10% and 100% [Fig. 5(a)]. There is still close agreement between some models, for example, three Overall, the L1 and L2 intercomparison metrics demon- strate that, on a pixel-by-pixel basis, there are often large Aubry et al. 1012 J. Acoust. Soc. Am. 152 (2), August 2022 FIG. 4. (Color online) Pressure amplitudes computed using KWAVE for benchmarks 7–9 showing x-y (left) and x-z (right) slices through the location of the peak pressure. The approximate location of the skull is shown with the white overlay. The size of the comparison domain for each benchmark is given in Table II. 24 October 2024 05:17:09 FIG. 4. (Color online) Pressure amplitudes computed using KWAVE for benchmarks 7–9 showing x-y (left) and x-z (right) slices through the location of the peak pressure. The approximate location of the skull is shown with the white overlay. The size of the comparison domain for each benchmark is given in Table II. variations between the model outputs. This is true despite there being no uncertainty in the material parameters and transducer characteristics. This highlights the inherent uncertainties when using computational models for transcra- nial ultrasound simulation, which must be considered when interpreting model results. models have median differences across all benchmarks on the order of 1% or less [see Fig. 6(b)]. Considering the focal position, all values including outliers are within 2.5 mm [see Fig. 6(a)], with median values for all benchmarks of 1 mm or less. Compared to KWAVE, the median values for all mod- els are less than 0.5 mm, with seven models having a median value of 0 mm [see Fig. 6(b)]. Figures 6 and 7 give a summary of the intercomparison metrics for the focal position, size, and pressure. Despite the variations in the full-ﬁeld error norms discussed above, there is very close agreement in the focal metrics. When com- pared by benchmark, the median values for the difference in focal pressure are all less than 10% [see Fig. 6(a)]. J. Acoust. Soc. Am. 152 (2), August 2022 A. Field characteristics Similarly, when compared by code, 10 of 11 models have median differences less than 10%. Differences of this level are on par with experimental repeatability and reproducibil- ity measurements conducted using similar ultrasound trans- ducers and a range of hydrophones.87 Compared to KWAVE, seven models have maximum differences in the focal pres- sure across all benchmarks of less than 10%, and ﬁve Figure 7 gives a summary of the intercomparison met- rics for focal size. Note, as mentioned in Sec. II G, the axial focal size for the piston transducer (SC2) for benchmarks 3–9 is not calculated as there is no focus after propagation through the skull. For reference, in water (BM1), the axial and lateral focal size for the focused bowl transducer is 26.2 and 4.1 mm, respectively, and the lateral focal size for the piston transducer at x ¼ 60 mm is 13.2 mm. For all bench- marks, the median differences in the axial focal size for the focused bowl transducer are less than 0.6 mm [Fig. 7(a)], although there are a small number of outliers with differ- ences up to 2.3 mm. The median differences in the lateral Aubry et al. 1 1013 1013 J. Acoust. Soc. Am. 152 (2), August 2022 https://doi.org/10.1121/10.0013426 G. 5. (Color online) Summary of relative L1 and L2 difference metrics computed across the entire ﬁeld taken from the exit plane of the transducer. oss-comparison (all codes compared with all codes). (b) Comparison with KWAVE. 24 October 2024 05:17:09 IG. 5. (Color online) Summary of relative L1 and L2 difference metrics computed across the entire ﬁeld taken from the exit ross-comparison (all codes compared with all codes). (b) Comparison with KWAVE. ry of relative L1 and L2 difference metrics computed across the entire ﬁeld taken from the exit plane of the transducer. (a) ompared with all codes). (b) Comparison with KWAVE. FIG. 5. (Color online) Summary of relative L1 and L2 difference metrics computed across the entire ﬁeld taken from the exit plane of the transducer. (a) Cross-comparison (all codes compared with all codes). (b) Comparison with KWAVE. focal size for the focused bowl transducer for all bench- marks are 0.2 mm or less. Variations in the lateral focal size for the piston transducer are generally larger, noting the lateral focal size is also larger for this transducer. Similar results are evident for the comparison against KWAVE [Fig. 7(b)]. A. Field characteristics relevant metrics to compare, along with acceptable limits on the differences between models, depend strongly on the intended application of the computational results. For exam- ple, calculating phase delays, calculating the approximate position and size of the acoustic focus in the brain, and cal- culating the pressure in the skin and skull to subsequently estimate skull heating may each have different constraints and accuracy requirements. An analysis of these factors is beyond the scope of the current work. However, it is hoped that the benchmarks and computational results presented here may help to facilitate such investigations in the future. Overall, there is very close agreement for all bench- marks in the characteristics of the focal pressure ﬁeld after propagation through the skull bone. Larger differences are evident in the full-ﬁeld metrics, dominated by differences in the ﬁeld between the transducer and the skull. The most Aubry et al. 1014 J. Acoust. Soc. Am. 152 (2), August 2022 https://doi.org/10.1121/10.0013426 6. (Color online) Summary of focal (peak) pressure and focal position metrics. (a) Cross-comparison (all codes compared with all codes) mparison with KWAVE. 24 October 2024 05:17:09 FIG. 6. (Color online) Summary of focal (peak) pressure and focal position metrics. (a) Cross-comparison (all codes compared with all codes). (b) Comparison with KWAVE. V. SUMMARY than 10% and 1 mm for all benchmarks. The differences in focal pressure are comparable to variations in experimental measurements,87 and the median differences in the axial and lateral focal position (0.6 and 0.2 mm) for the focused trans- ducer are small compared to the corresponding size of the 6 dB focal volume (26.2 and 4.1 mm). These results build conﬁdence in the ability of the described computational models to produce consistent results when simulating wave propagation through skull layers at 500 kHz. The benchmark deﬁnitions and associated data ﬁles, simulation results, and A series of numerical benchmarks relevant to transcra- nial ultrasound simulation are presented, along with inter- comparison results for 11 modeling tools used in the community. The intercomparison results show close agree- ment between the models, particularly for the position, size, and magnitude of the acoustic focus after propagating through the skull. When comparing each model with every other model in a cross-comparison, the median values for the difference in focal pressure and focal position are less 1015 Aubry et al. Aubry et al. J. Acoust. Soc. Am. 152 (2), August 2022 https://doi.org/10.1121/10.0013426 7. (Color online) Summary of axial and lateral focal size metrics. Note that axial focal size was not computed for benchmarks 3–9 when using the p n source (SC2), as the ﬁeld in this case did not have an axial maximum in the post-skull region. (a) Cross-comparison (all codes compared with s). (b) Comparison with KWAVE. 24 October 2024 05:17:09 FIG. 7. (Color online) Summary of axial and lateral focal size metrics. Note that axial focal size was not computed for benchmarks 3–9 when using the plane piston source (SC2), as the ﬁeld in this case did not have an axial maximum in the post-skull region. (a) Cross-comparison (all codes compared with all codes). (b) Comparison with KWAVE. models and model comparisons when using material param- eters derived from CT images. codes to compute the intercomparison metrics are all freely available.8,11 This allows the results to be replicated or fur- ther analysis to be conducted. Additional model results can also be easily added to the intercomparison, for example, to validate newly developed solvers. More generally, the inter- comparison exercise provides a framework for creating benchmarks and performing model cross-comparisons. Further phases of the intercomparison exercise are currently under discussion, including benchmarks for elastic wave https://doi.org/10.1121/10.0013426 y ( ) 11J.-F. Aubry, O. Bates, C. Boehm, K. Butts Pauly, D. Christensen, C. Cueto, P. Gelat, L. Guasch, J. Jaros, Y. Jing, R. Jones, N. Li, P. Marty, H. Montanaro, E. Neufeld, S. Pichardo, G. Pinton, A. Pulkkinen, A. Stanziola, A. Thielscher, B. Treeby, and E. van ’t Wout, “Benchmark problems for transcranial ultrasound simulation: Datasets for intercompar- ison of compressional wave models (version 1.0) [dataset],” Zenodo, https://doi.org/10.5281/zenodo.6020543 (2022). 12J. K. Mueller, L. Ai, P. Bansal, and W. Legon, “Numerical evaluation of the skull for human neuromodulation with transcranial focused ultra- sound,” J. Neural Eng. 14(6), 066012 (2017). g ( ) ( ) 13G. T. Clement, P. J. White, and K. Hynynen, “Enhanced ultrasound trans- mission through the human skull using shear mode conversion,” J. Acoust. Soc. Am. 115(3), 1356–1364 (2004). 14X.-D. Wang, W.-J. Lin, C. Su, and X.-M. Wang, “Inﬂuence of mode con- versions in the skull on transcranial focused ultrasound and temperature ﬁelds utilizing the wave ﬁeld separation method: A numerical study,” Chin. Phys. B 27(2), 024302 (2018). y 15Y. Younan, T. Defﬁeux, B. Larrat, M. Fink, M. Tanter, and J.-F. Aubry, “Inﬂuence of the pressure ﬁeld distribution in transcranial ultrasonic neu- rostimulation,” Med. Phys. 40(8), 082902 (2013). 16J. L. Robertson, B. T. Cox, J. Jaros, and B. E. Treeby, “Accurate simula- tion of transcranial ultrasound propagation for ultrasonic neuromodulation and stimulation,” J. Acoust. Soc. Am. 141(3), 1726–1738 (2017). 17 1W. J. Elias, N. Lipsman, W. G. Ondo, P. Ghanouni, Y. G. Kim, W. Lee, M. Schwartz, K. Hynynen, A. M. Lozano, B. B. Shah, D. Huss, R. F. Dallapiazza, R. Gwinn, J. Witt, S. Ro, H. M. Eisenberg, P. S. Fishman, D. Gandhi, C. H. Halpern, R. Chuang, K. Butts Pauly, T. S. Tierney, M. T. Hayes, G. Rees Cosgrove, T. Yamaguchi, K. Abe, T. Taira, and J. W. Chang, “A randomized trial of focused ultrasound thalamotomy for essen- tial tremor,” N. Engl. J. Med. 375(8), 730–739 (2016). 17H. O’Neil, “Theory of focusing radiators,” J. Acoust. Soc. Am. 21(5), 516–526 (1949). 18F. J. Fry and J. E. Barger, “Acoustical properties of the human skull,” J. Acoust. Soc. Am. 63(5), 1576–1590 (1978). 19 19T. D. Mast, “Empirical relationships between acoustic parameters in human soft tissues,” Acoust. Res. Lett. Online 1(2), 37–42 (2000). 20 2A. Abrahao, Y. Meng, M. Llinas, Y. Huang, C. Hamani, T. Mainprize, I. Aubert, C. Heyn, S. E. Black, K. Hynynen, N. https://doi.org/10.1121/10.0013426 10See supplementary material at https://www.scitation.org/doi/suppl/ 10.1121/10.0013426 for a table form of the model summaries given in Sec. III and summaries of the complete set of intercomparison results. SuppPub1.xlsx gives an alternate table form of the model summaries given in Sec. III. SuppPub2.zip provides summaries of the comparison results (including metrics, ﬁeld plots, axial proﬁles, and difference plots) for each model compared against FOCUS (for benchmarks 1 and 2) and KWAVE (for benchmarks 1–9). The .zip ﬁle contains separate pdf ﬁles for each model and for each benchmark, as a well as a summary of the cross-comparison metrics. The raw data ﬁles and MATLAB codes to process the results are also freely available (Refs. 8 and 11). helpful discussions regarding the use of FOCUS as a reference simulation. O.B. was supported by the Engineering and Physical Sciences Research Council (EPSRC) Centre for Doctoral Training in Neurotechnology, Grant No. EP/L016737/ 1. K.B.P. and N.L. acknowledge the support of National Institutes of Health (NIH) Grant Nos. R01 CA227687, NIH T32 EB009653, and NSF DGE 1656518. D.C. acknowledges support from the Focused Ultrasound Foundation and NIH Grant Nos. R01 EB013433, R01 CA172787, R01 EB028316, and R37 CA224141. C.C. acknowledges the support of EPSRC Grant No. EP/T51780X/1. P.G. and E.v.W. acknowledge the support of EPSRC Grant No. EP/P012434/1 and use of the UCL Myriad High Performance Computing Facility (Myriad@UCL) and associated support services. J.J. was supported by Brno University of Technology under Project No. FIT-S-20–6309. Y.J. acknowledges the support of NIH Grant No. R01EB025205. P.M. and C.B. acknowledge support from the Swiss National Supercomputing Centre (CSCS) under Project Nos. s1040 and sm59. S.P. acknowledges the support of the Natural Sciences and Engineering Research Council of Canada and the Canada Foundation for Innovation. A.P. would like to acknowledge Academy of Finland Project Nos. 320166, 336119, and 336799. A.S. and B.T. were supported by EPSRC Grant No. EP/S026371/1. A.T. was supported by Lundbeck Foundation Grant No. R313-2019-622. the results are also freely available (Refs. 8 and 11). 11J.-F. Aubry, O. Bates, C. Boehm, K. Butts Pauly, D. Christensen, C. Cueto, P. Gelat, L. Guasch, J. Jaros, Y. Jing, R. Jones, N. Li, P. Marty, H. Montanaro, E. Neufeld, S. Pichardo, G. Pinton, A. Pulkkinen, A. Stanziola, A. Thielscher, B. Treeby, and E. van ’t Wout, “Benchmark problems for transcranial ultrasound simulation: Datasets for intercompar- ison of compressional wave models (version 1.0) [dataset],” Zenodo, https://doi.org/10.5281/zenodo.6020543 (2022). https://doi.org/10.1121/10.0013426 Lipsman, and L. Zinman, “First-in-human trial of blood–brain barrier opening in amyotrophic lat- eral sclerosis using MR-guided focused ultrasound,” Nat. Commun. 10(1), 4373 (2019). 20G. Clement and K. Hynynen, “Correlation of ultrasound phase with physi- cal skull properties,” Ultrasound Med. Biol. 28(5), 617–624 (2002). y y p p y cal skull properties,” Ultrasound Med. Biol. 28(5), 617–624 (2002). cal skull properties,” Ultrasound Med. Biol. 28(5), 617–624 ( 21 p p 21S. Pichardo, V. W. Sin, and K. Hynynen, “Multi-frequency characteriza- tion of the speed of sound and attenuation coefﬁcient for longitudinal transmission of freshly excised human skulls,” Phys. Med. Biol. 56(1), 219–250 (2011). 3W. Legon, T. F. Sato, A. Opitz, J. Mueller, A. Barbour, A. Williams, and W. J. Tyler, “Transcranial focused ultrasound modulates the activity of primary somatosensory cortex in humans,” Nat. Neurosci. 17(2), 322–329 (2014). 4 22G. Pinton, J.-F. Aubry, E. Bossy, M. Muller, M. Pernot, and M. Tanter, “Attenuation, scattering, and absorption of ultrasound in the skull bone,” Med. Phys. 39(1), 299–307 (2011). 23S. Pichardo, C. Moreno-Hernandez, R. A. Drainville, V. Sin, L. Curiel, and K. Hynynen, “A viscoelastic model for the prediction of transcranial ultrasound propagation: Application for the estimation of shear acoustic properties in the human skull,” Phys. Med. Biol. 62(17), 6938–6962 (2017). 4K. Hynynen and F. A. Jolesz, “Demonstration of potential noninvasive ultrasound brain therapy through an intact skull,” Ultrasound Med. Biol. 24(2), 275–283 (1998). 5G. Bouchoux, K. B. Bader, J. J. Korfhagen, J. L. Raymond, R. Shivashankar, T. A. Abruzzo, and C. K. Holland, “Experimental valida- tion of a ﬁnite-difference model for the prediction of transcranial ultra- sound ﬁelds based on CT images,” Phys. Med. Biol. 57(23), 8005–8022 (2012). 24P. J. White, G. T. Clement, and K. Hynynen, “Longitudinal and shear mode ultrasound propagation in human skull bone,” Ultrasound Med. Biol. 32(7), 1085–1096 (2006). 25 6F. Marquet, M. Pernot, J.-F. Aubry, G. Montaldo, L. Marsac, M. Tanter, and M. Fink, “Non-invasive transcranial ultrasound therapy based on a 3D CT scan: Protocol validation and in vitro results,” Phys. Med. Biol. 54(9), 2597–2613 (2009). 25T. D. Webb, S. A. Leung, P. Ghanouni, J. J. Dahl, N. J. Pelc, and K. B. Pauly, “Acoustic attenuation: Multifrequency measurement and relation- ship to CT and MR imaging,” IEEE Trans. Ultrason. Ferroelectr. Freq. Control 68(5), 1532–1545 (2021). 26 Pauly, Acoustic attenuation: Multifrequency measurement and relation- ship to CT and MR imaging,” IEEE Trans. Ultrason. Ferroelectr. Freq. ACKNOWLEDGMENTS The authors would like to thank the International Transcranial Ultrasonic Stimulation Safety and Standards (ITRUSST) consortium for providing the motivation and framework to conduct this work and Robert McGough for J. Acoust. Soc. Am. 152 (2), August 2022 Aubry et al. Aubry et al. 1W. J. Elias, N. Lipsman, W. G. Ondo, P. Ghanouni, Y. G. Kim, W. Lee, M. Schwartz, K. Hynynen, A. M. Lozano, B. B. Shah, D. Huss, R. F. Dallapiazza, R. Gwinn, J. Witt, S. Ro, H. M. Eisenberg, P. S. Fishman, D. Gandhi, C. H. Halpern, R. Chuang, K. Butts Pauly, T. S. Tierney, M. T. Hayes, G. Rees Cosgrove, T. Yamaguchi, K. Abe, T. Taira, and J. W. Chang, “A randomized trial of focused ultrasound thalamotomy for essen- tial tremor,” N. Engl. J. Med. 375(8), 730–739 (2016). https://doi.org/10.1121/10.0013426 J. Numer. Methods Eng. 122(20), 5873–5897 (2021). 65 41R. A. Drainville, L. Curiel, and S. Pichardo, “Superposition method for modelling boundaries between media in viscoelastic ﬁnite difference time domain simulations,” J. Acoust. Soc. Am. 146(6), 4382–4401 (2019). 42 65T. Betcke, E. van ’t Wout, and P. Gelat, “Computationally efﬁcient boundary element methods for high-frequency Helmholtz problems in unbounded domains,” in Modern Solvers for Helmholtz Problems, edited by D. Lahaye, J. Tang, and K. Vuik (Birkh€auser, Cham, Switzerland, 2017), pp. 215–243. 42F. Collino and C. Tsogka, “Application of the perfectly matched absorb- ing layer model to the linear elastodynamic problem in anisotropic hetero- geneous media,” Geophysics 66(1), 294–307 (2001). 65T. Betcke, E. van ’t Wout, and P. Gelat, “Computationally efﬁcient boundary element methods for high-frequency Helmholtz problems in unbounded domains,” in Modern Solvers for Helmholtz Problems, edited 43G. F. Pinton, J. Dahl, S. Rosenzweig, and G. E. Trahey, “A heterogeneous nonlinear attenuating full-wave model of ultrasound,” IEEE Trans. Ultrason. Ferroelectr. Freq. Control 56(3), 474–488 (2009). by D. Lahaye, J. Tang, and K. Vuik (Birkh€auser, Cham, Switzerland, 2017), pp. 215–243. 66E. van ’t Wout, S. R. Haqshenas, P. Gelat, T. Betcke, and N. Saffari, “Frequency-robust preconditioning of boundary integral equations for ti t i i ” J C t Ph 462 111229 (2022) E. van t Wout, S. R. Haqshenas, P. Gelat, T. Betcke, and N. Saffari, “Frequency-robust preconditioning of boundary integral equations for acoustic transmission,” J. Comput. Phys. 462, 111229 (2022). 67 44G. Pinton, “A fullwave model of the nonlinear wave equation with multi- ple relaxations and relaxing perfectly matched layers for high-order numerical ﬁnite-difference solutions,” arXiv:2106.11476 (2021). 45 “Frequency-robust preconditioning of boundary integral equations for acoustic transmission,” J. Comput. Phys. 462, 111229 (2022). 67 q y p g y g q acoustic transmission,” J. Comput. Phys. 462, 111229 (2022). 67 67W. Smigaj, T. Betcke, S. Arridge, J. Phillips, and M. Schweiger, “Solving boundary integral problems with BEM,” ACM Trans. Math. Softw. 41, 1–40 (2015). 45R. Courant, K. Friedrichs, and H. Lewy, “€Uber die partiellen differenzen- gleichungen der mathematischen physik” (“On the partial differential equations of mathematical physics”), Math. Ann. 100(1), 32–74 (1928). 68C. Geuzaine and J.-F. Remacle, “Gmsh: A 3-D ﬁnite element mesh gener- ator with built-in pre- and post-processing facilities,” Int. J. Numer. Methods Eng. 79(11), 1309–1331 (2009). 69 q p y 46A. Pulkkinen, B. Werner, E. Martin, and K. https://doi.org/10.1121/10.0013426 59 58B. E. Treeby, E. S. Wise, F. Kuklis, J. Jaros, and B. Cox, “Nonlinear ultra- sound simulation in an axisymmetric coordinate system using a k-space pseudospectral method,” J. Acoust. Soc. Am. 148(4), 2288–2300 (2020). 59J. Jaros, A. P. Rendell, and B. E. Treeby, “Full-wave nonlinear ultrasound simulation on distributed clusters with applications in high-intensity focused ultrasound,” Int. J. High Perform. Comput. Appl. 30(2), 137–155 (2016). 35Q. Fang and D. A. Boas, “Tetrahedral mesh generation from volumetric binary and grayscale images,” in Proceedings of the 2009 IEEE International Symposium on Biomedical Imaging: From Nano to Macro, Boston, MA (June 28–July 1, 2009), pp. 1142–1145. p p 59J. Jaros, A. P. Rendell, and B. E. Treeby, “Full-wave nonlinear ultrasound simulation on distributed clusters with applications in high-intensity focused ultrasound,” Int. J. High Perform. Comput. Appl. 30(2), 137–155 (2016). 60 36J. Virieux, “P-SV wave propagation in heterogeneous media: Velocity- stress ﬁnite-difference method,” Geophysics 51(4), 889–901 (1986). 60J. Gu and Y. Jing, “MSOUND: An open source toolbox for modeling acous- tic wave propagation in heterogeneous media,” IEEE Trans. Ultrason. Ferroelectr. Freq. Control 68(5), 1476–1486 (2021). 37A. R. Levander, “Fourth-order ﬁnite-difference P-SV seismograms,” Geophysics 53(11), 1425–1436 (1988). 38T. Bohlen, “Parallel 3-D viscoelastic ﬁnite difference seismic modelling,” Comput. Geosci. 28(8), 887–899 (2002). 61E. van ’t Wout, P. Gelat, T. Betcke, and S. Arridge, “A fast boundary ele- ment method for the scattering analysis of high-intensity focused ultra- sound,” J. Acoust. Soc. Am. 138(5), 2726–2737 (2015). 39J. O. Blanch, J. O. Robertsson, and W. W. Symes, “Modeling of a con- stant Q: Methodology and algorithm for an efﬁcient and optimally inex- pensive viscoelastic technique,” Geophysics 60(1), 176–184 (1995). 40 62E. van ’t Wout, S. R. Haqshenas, P. Gelat, T. Betcke, and N. Saffari, “Boundary integral formulations for acoustic modelling of high-contrast media,” Comput. Math. Appl. 105, 136–149 (2022). 40P. Moczo, J. Kristek, V. Vavrycuk, R. J. Archuleta, and L. Halada, “3D heterogeneous staggered-grid ﬁnite-difference modeling of seismic motion with volume harmonic and arithmetic averaging of elastic moduli and densities,” Bull. Seismol. Soc. Am. 92(8), 3042–3066 (2002). 41 media, Comput. Math. Appl. 105, 136 149 (2022). 63S. R. Haqshenas, P. Gelat, E. van ’t Wout, T. Betcke, and N. Saffari, “A fast full-wave solver for calculating ultrasound propagation in the body,” Ultrasonics 110, 106240 (2021). 64E. van ’t Wout, S. R. Haqshenas, P. Gelat, T. Betcke, and N. Saffari, “Benchmarking preconditioned boundary integral formulations for acoustics,” Int. https://doi.org/10.1121/10.0013426 Hynynen, “Numerical simula- tions of clinical focused ultrasound functional neurosurgery,” Phys. Med. Biol. 59(7), 1679–1700 (2014). 69R. Schmidt and K. Singh, “Meshmixer: An interface for rapid mesh composition,” in Proceedings of ACM SIGGRAPH 2010 Talks, Los Angeles, CA (July 26–30, 2010). 47U. Vyas and D. Christensen, “Ultrasound beam simulations in inhomoge- neous tissue geometries using the hybrid angular spectrum method,” IEEE Trans. Ultrason. Ferroelectr. Freq. Control 59(6), 1093–1100 (2012). 48 70M. Afanasiev, C. Boehm, M. van Driel, L. Krischer, M. Rietmann, D. A. May, M. G. Knepley, and A. Fichtner, “Modular and ﬂexible spectral- element waveform modelling in two and three dimensions,” Geophys. J. Int. 216(3), 1675–1692 (2019). 70M. Afanasiev, C. Boehm, M. van Driel, L. Krischer, M. Rietmann, D. A. May, M. G. Knepley, and A. Fichtner, “Modular and ﬂexible spectral- 48S. Almquist, D. L. Parker, and D. A. Christensen, “Rapid full-wave phase aberration correction method for transcranial high-intensity focused ultra- sound therapies,” J. Ther. Ultrasound 4(1), 30 (2016). element waveform modelling in two and three dimensions,” Geophys. J. Int. 216(3), 1675–1692 (2019). 49A. Stanziola, S. R. Arridge, B. T. Cox, and B. E. Treeby, “j-Wave: An open-source differentiable wave simulator,” arXiv:2207.01499 (2022). 50 71A. Ferroni, P. F. Antonietti, I. Mazzieri, and A. Quarteroni, “Dispersion- dissipation analysis of 3-D continuous and discontinuous spectral element methods for the elastodynamics equation,” Geophys. J. Int. 211(3), 1554–1574 (2017). 50Y. Saad and M. H. Schultz, “GMRES: A generalized minimal residual algorithm for solving nonsymmetric linear systems,” SIAM J. Sci. Stat. Comput. 7(3), 856–869 (1986). ( ) 72V. Hapla, M. G. Knepley, M. Afanasiev, C. Boehm, M. van Driel, L. Krischer, and A. Fichtner, “Fully parallel mesh I/O using PETSc DMPlex with an application to waveform modeling,” SIAM J. Sci. Comput. 43(2), C127–C153 (2021). p 51A. Bermudez, L. Hervella-Nieto, A. Prieto, and R. Rodrıguez, “An opti- mal perfectly matched layer with unbounded absorbing function for time- harmonic acoustic scattering problems,” J. Comput. Phys. 223(2), 469–488 (2007). 73R. Kosloff and D. Kosloff, “Absorbing boundaries for wave propagation problems,” J. Comput. Phys. 63(2), 363–376 (1986). 4 52E. S. Wise, B. Cox, J. Jaros, and B. E. Treeby, “Representing arbitrary acoustic source and sensor distributions in Fourier collocation methods,” J. Acoust. Soc. Am. 146(1), 278–288 (2019). 74Coreform LLC, “Coreform Cubit (version 2021.5) [computer software],” http://coreform.com (Last viewed August 11, 2022). 75P. A. Witte, M. Louboutin, F. Luporini, G. J. Gorman, and F. J. https://doi.org/10.1121/10.0013426 54J. Bradbury, R. Frostig, P. Hawkins, M. J. Johnson, C. Leary, D. Maclaurin, G. Necula, A. Paszke, J. VanderPlas, S. Wanderman-Milne, and Q. Zhang, “JAX: Composable transformations of PythonþNumPy programs ” http://github com/google/jax (Last viewed August 11 2022) 30H. Hori, G. Moretti, A. Rebora, and F. Crovato, “The thickness of human scalp: Normal and bald,” J. Invest. Dermatol. 58(6), 396–399 (1972). 54J. Bradbury, R. Frostig, P. Hawkins, M. J. Johnson, C. Leary, D. Maclaurin, G. Necula, A. Paszke, J. VanderPlas, S. Wanderman-Milne, and Q. Zhang, “JAX: Composable transformations of PythonþNumPy programs,” http://github.com/google/jax (Last viewed August 11, 2022). 55B. E. Treeby and B. T. Cox, “k-Wave: MATLAB toolbox for the simula- tion and reconstruction of photoacoustic wave ﬁelds,” J. Biomed. Opt. 15(2), 021314 (2010). 31V. S. Fonov, A. C. Evans, R. C. McKinstry, C. Almli, and D. Collins, “Unbiased nonlinear average age-appropriate brain templates from birth to adulthood,” NeuroImage 47(Suppl. 1), S39–S41 (2009). p g , p g g g j ( g , ) 55B. E. Treeby and B. T. Cox, “k-Wave: MATLAB toolbox for the simula- tion and reconstruction of photoacoustic wave ﬁelds,” J. Biomed. Opt. 15(2), 021314 (2010). 32V. Fonov, A. C. Evans, K. Botteron, C. R. Almli, R. C. McKinstry, D. L. Collins, and Brain Development Cooperative Group, “Unbiased average age-appropriate atlases for pediatric studies,” NeuroImage 54(1), 313–327 (2011). ( ) ( ) 56B. E. Treeby, J. Jaros, A. P. Rendell, and B. Cox, “Modeling nonlinear ultrasound propagation in heterogeneous media with power law absorp- tion using a k-space pseudospectral method,” J. Acoust. Soc. Am. 131(6), 4324–4336 (2012). 33J. D. Nielsen, K. H. Madsen, O. Puonti, H. R. Siebner, C. Bauer, C. G. Madsen, G. B. Saturnino, and A. Thielscher, “Automatic skull segmenta- tion from MR images for realistic volume conductor models of the head: Assessment of the state-of-the-art,” NeuroImage 174, 587–598 (2018). 34 57B. Cox and B. Treeby, “Accurate time-varying sources in k-space pseudo- spectral time domain acoustic simulations,” in Proceedings of the 2018 IEEE International Ultrasonics Symposium (IUS), Kobe, Japan (October 22–25, 2018), pp. 1–4. Assessment of the state-of-the-art,” NeuroImage 174, 587–598 ( 34Q. Fang, “iso2mesh: a 3D surface and volumetric mesh generator for MATLAB/Octave,” http://iso2mesh.sourceforge.net (Last viewed August 11, 2022). 58B. E. Treeby, E. S. Wise, F. Kuklis, J. Jaros, and B. Cox, “Nonlinear ultra- sound simulation in an axisymmetric coordinate system using a k-space pseudospectral method,” J. Acoust. Soc. Am. 148(4), 2288–2300 (2020). 1018 J. Acoust. Soc. Am. 152 (2), August 2022 https://doi.org/10.1121/10.0013426 Control 68(5), 1532–1545 (2021). 6 26R. J. McGough, T. V. Samulski, and J. F. Kelly, “An efﬁcient grid sector- ing method for calculations of the near-ﬁeld pressure generated by a cir- cular piston,” J. Acoust. Soc. Am. 115(5), 1942–1954 (2004). 7R. F. Dallapiazza, K. F. Timbie, S. Holmberg, J. Gatesman, M. B. Lopes, R. J. Price, G. W. Miller, and W. J. Elias, “Noninvasive neuromodulation and thalamic mapping with low-intensity focused ultrasound,” J. Neurosurg. 128(3), 875–884 (2018). cular piston,” J. Acoust. Soc. Am. 115(5), 1942–1954 (2004). 27 27D. Chen and R. J. McGough, “A 2D fast near-ﬁeld method for calculating near-ﬁeld pressures generated by apodized rectangular pistons,” J. Acoust. Soc. Am. 124(3), 1526–1537 (2008). 28 g 8B. Treeby, “Benchmark problems for transcranial ultrasound simulation: Intercomparison library (version 1.0) [computer code],” https://github. com/ucl-bug/transcranial-ultrasound-benchmarks (Last viewed August 11, 2022). 28J. F. Kelly and R. J. McGough, “Transient ﬁelds generated by spherical shells in viscous media,” AIP Conf. Proc. 1113, 210–214 (2009). 29 29S. L. Alexander, K. Rafaels, C. A. Gunnarsson, and T. Weerasooriya, “Structural analysis of the frontal and parietal bones of the human skull,” 29S. L. Alexander, K. Rafaels, C. A. Gunnarsson, and T. Weerasooriya, “Structural analysis of the frontal and parietal bones of the human skull,” J. Mech. Behav. Biomed. Mater. 90, 689–701 (2019). 9J. Gu and Y. Jing, “A modiﬁed mixed domain method for modeling acoustic wave propagation in strongly heterogeneous media,” J. Acoust. Soc. Am. 147(6), 4055–4068 (2020). 9J. Gu and Y. Jing, “A modiﬁed mixed domain method for modeling acoustic wave propagation in strongly heterogeneous media,” J. Acoust. Soc. Am. 147(6), 4055–4068 (2020). Aubry et al. 1017 Aubry et al. J. Acoust. Soc. Am. 152 (2), August 2022 J. Acoust. Soc. Am. 152 (2), August 2022 https://doi.org/10.1121/10.0013426 82M. Louboutin, M. Lange, F. Luporini, N. Kukreja, P. A. Witte, F. J. Herrmann, P. Velesko, and G. J. Gorman, “Devito (v3. 1.0): An embedded domain-speciﬁc language for ﬁnite differences and geophysical exploration,” Geosci. Model Dev. 12(3), 1165–1187 (2019). 76W. C. Chew and W. H. Weedon, “A 3D perfectly matched medium from modiﬁed Maxwell’s equations with stretched coordinates,” Microw. Opt. Technol. Lett. 7(13), 599–604 (1994). 76W. C. Chew and W. H. Weedon, “A 3D perfectly matched medium from modiﬁed Maxwell’s equations with stretched coordinates,” Microw. Opt. Technol. Lett. 7(13), 599–604 (1994). 77A. Kyriakou, E. Neufeld, B. Werner, G. Szekely, and N. Kuster, “Full- wave acoustic and thermal modeling of transcranial ultrasound propaga- tion and investigation of skull-induced aberration correction techniques: A feasibility study,” J. Ther. Ultrasound 3(1), 11 (2015). 83L. Amundsen and Ø. Pedersen, “Time step n-tupling for wave equations,” Geophysics 82(6), T249–T254 (2017). 84 84G. Yao, N. V. Da Silva, and D. Wu, “An effective absorbing layer for the boundary condition in acoustic seismic wave simulation,” J. Geophys. Eng. 15(2), 495–511 (2018). 78E. Neufeld, A. Kyriacou, W. Kainz, and N. Kuster, “Approach to validate simulation-based distribution predictions combining the gamma-method and uncertainty assessment: Application to focused ultrasound,” J. Verif. Valid. Uncertain Quantif. 1(3), 031006 (2016). 85Y. Gao, J. Zhang, and Z. Yao, “Unsplit complex frequency shifted perfectly matched layer for second-order wave equation using auxiliary differential equations,” J. Acoust. Soc. Am. 138(6), EL551–EL557 (2015). 79C. Pasquinelli, H. Montanaro, H. J. Lee, L. G. Hanson, H. Kim, N. Kuster, H. R. Siebner, E. Neufeld, and A. Thielscher, “Transducer model- ing for accurate acoustic simulations of transcranial focused ultrasound stimulation,” J. Neural Eng. 17(4), 046010 (2020). 86G. J. Hicks, “Arbitrary source and receiver positioning in ﬁnite-difference schemes using Kaiser windowed sinc functions,” Geophysics 67(1), 156–165 (2002). 80H. Montanaro, C. Pasquinelli, H. J. Lee, H. Kim, H. R. Siebner, N. Kuster, A. Thielscher, and E. Neufeld, “The impact of CT image parame- ters and skull heterogeneity modeling on the accuracy of transcranial focused ultrasound simulations,” J. Neural Eng. 18(4), 046041 (2021). 81 87E. Martin and B. Treeby, “Investigation of the repeatability and reproduc- ibility of hydrophone measurements of medical ultrasound ﬁelds,” J. Acoust. Soc. Am. 145(3), 1270–1282 (2019). 81C. Cueto, O. Bates, G. Strong, J. Cudeiro, F. Luporini, O. C. Agudo, G. Gorman, L. Guasch, and M.-X. Tang, “Stride: A ﬂexible software plat- form for high-performance ultrasound computed tomography,” Comput. J. Acoust. Soc. Am. 152 (2), August 2022 https://doi.org/10.1121/10.0013426 Herrmann, “Compressive least-squares migration with on-the-ﬂy Fourier transforms,” Geophysics 84(5), R655–R672 (2019). 53A. Stanziola, S. R. Arridge, B. T. Cox, and B. E. Treeby, “A research framework for writing differentiable PDE discretizations in JAX,” arXiv:2111.05218 (2021). Aubry et al. 1018 J. Acoust. Soc. Am. 152 (2), August 2022 1018 J. Acoust. Soc. Am. 152 (2), August 2022 https://doi.org/10.1121/10.0013426 Methods Programs Biomed. 221, 106855 (2022). 88P. Gelat, S. R. Haqshenas, and E. van 0t Wout, “OptimUS: A Python library for solving 3D acoustic wave propagation,” https://github.com/ optimuslib (Last viewed August 11, 2022). 24 October 2024 05:17:09 Aubry et al. 1019
https://openalex.org/W2167844299	https://europepmc.org/articles/pmc4610456?pdf=render	English	null	Design and Realization of a Three Degrees of Freedom Displacement Measurement System Composed of Hall Sensors Based on Magnetic Field Fitting by an Elliptic Function	Sensors	2,015	cc-by	7,074	Design and Realization of a Three Degrees of Freedom Displacement Measurement System Composed of Hall Sensors Based on Magnetic Field Fitting by an Elliptic Function Bo Zhao, Lei Wang * and Jiu-Bin Tan Sensors 2015, 15, 22530-22546; doi:10.3390/s150922530 Sensors 2015, 15, 22530-22546; doi:10.3390/s150922530 sensors ISSN 1424-8220 www.mdpi.com/journal/sensors OPEN ACCESS sensors ISSN 1424-8220 www.mdpi.com/journal/sensors OPEN ACCESS Bo Zhao, Lei Wang * and Jiu-Bin Tan Harbin Institute of Technology, D-403 Science Park, No. 2 Yikuang Street, Harbin 150080, China; E-Mails: hitzhaobo@hit.edu.cn (B.Z.); Jbtan@hit.edu.cn (J.-B.T.) * Author to whom correspondence should be addressed; E-Mail: hit_wanglei@hit.edu.cn; Tel.: +86-451-8641-2041 (ext. 815); Fax: +86-451-8640-2258. Academic Editor: Vittorio M. N. Passaro Received: 15 July 2015 / Accepted: 31 August 2015 / Published: 8 September 2015 Received: 15 July 2015 / Accepted: 31 August 2015 / Published: 8 September 2015 Received: 15 July 2015 / Accepted: 31 August 2015 / Published: 8 September 2015 Abstract: This paper presents the design and realization of a three degrees of freedom (DOFs) displacement measurement system composed of Hall sensors, which is built for the XYθz displacement measurement of the short stroke stage of the reticle stage of lithography. The measurement system consists of three pairs of permanent magnets mounted on the same plane on the short stroke stage along the Y, Y, X directions, and three single axis Hall sensors correspondingly mounted on the frame of the reticle stage. The emphasis is placed on the decoupling and magnetic field fitting of the three DOFs measurement system. The model of the measurement system is illustrated, and the XY positions and θZ rotation of the short stroke stage can be obtained by decoupling the sensor outputs. A magnetic field fitting by an elliptic function-based compensation method is proposed. The practical field intensity of a permanent magnet at a certain plane height can be substituted for the output voltage of a Hall sensors, which can be expressed by the elliptic function through experimental data as the crucial issue to calculate the three DOFs displacement. Experimental results of the Hall sensor displacement measurement system are presented to validate the proposed three DOFs measurement system. Keywords: Hall sensor; displacement measurement; magnetic field fitting; elliptic function Sensors 2015, 15 Sensors 2015, 15 22531 1. Introduction Precision metrology plays an important role in on-machine measurement, positioning and manufacturing, such as machine tools, coordinate measuring machines and semiconductor manufacturing [1]. Particularly, the accuracy of these machines is significantly determined by the translational and rotational stages, so multi-DOF precision measurement of the stages is crucial. In recent years, abundant research on multi-DOF measurement systems for precision stages has been carried out. Aktakka developed a microactuation and sensing platform which can be used to provide precise physical reference for calibration of multi-degrees-of-freedom inertial sensors [2]. A six-DOF displacement measurement system using a diffraction grating as a cooperative target was proposed by Kim, combined with optical triangulation [3]. The Stewart parallel structure was applied in multi-DOF measurement as a MEMS sensor by Mura [4], and a sensitivity analysis of the configuration was carried on concerning geometrical characteristic and displacement amplitude [5]. Allred developed another Gough-Stewart platform-based measurement solution which integrates linear displacement sensors into a high capacity laminate bearing [6]. Hall magnetic field sensors feature contactless measurement, robustness, tolerance to harsh environments [7], simplicity and versatility [8], so they are widely applied in industrial control systems, precise instruments [9] and consumer electronic products, not only for direct measurement of magnetic fields [10], but also for non-direct measurements, such as speed or position [11–13], sometimes shape detection [14] or current measurement [15,16]. This paper presents a three DOFs displacement measurement system composed by three Hall sensors for precision positioning, which plays a crucial role while manufacturing or manipulating on the micro/nano level, where the key role of a precision positioning stage is to load, position and keep an object stable [17]. In this paper two translational DOFs and one rotational DOF of the stage are measured by three linear Hall sensors located in the same plane. A decoupling model of the measurement system is constructed, the stage positions can be obtained by the output of the sensors and the decoupling matrix. Then a magnetic field fitting method by an elliptic function is proposed to analytically express the magnet field line at a certain height, by which the practical field intensity of a permanent magnet can be substituted for the output voltage of the Hall sensors to solve the position of the stage. The decoupling and magnetic field fitting method is finally validated by experiments. 2. Description of the Measurement System The XYθZ displacements of the SS (short stroke) stage of the RS (reticle stage) of the lithography are measured by the cooperation of three Hall sensors in this paper. The schematic illustration of the SS stage is shown in Figure 1, in which the basic components of the SS stage are given. The SS stage is suspended by three gravity compensators (GCs) of cylindrical shape based on magnetic levitation. The GC is actuated by a voice coil motor along the Z axis, the GC rotors are connected with the SS stage, and the GC stators are fixed on the SS frame. Each GC can provide enough electromagnetic force along the Z axis for positioning. The radial gap between the magnet of the compensator and the coil is 2 mm, so besides Z displacement, the SS stage features θX and θY displacements which can be also activated by the cooperation of the compensators. Meanwhile, the SS 22532 Sensors 2015, 15 Sensors 2015, 15 state can move virtually without friction in the XYθZ DOFs. The ZθXθY displacements can be measured by three Linear Variable Differential Transformers (LVDTs) mounted under the SS stage, by which the SS stage can be controlled to maintain an expected height. Figure 1. Schematic illustration of the proposed short stroke stage with 3D Hall sensors measurement system, in which the LS (long stroke) motors and the aerostatic guides are not included. Figure 1. Schematic illustration of the proposed short stroke stage with 3D Hall sensors measurement system, in which the LS (long stroke) motors and the aerostatic guides are not included. The XYθZ DOFs of the SS stage can be achieved by the linear motors on the premise of suspended condition. In the XOY plane, the SS stage is actuated by three linear voice coil motors (LVCM) which are placed along the Y, Y, X directions, respectively, with the magnet of the motors fixed on the SS frame and the coil fixed on the SS stage. The center of gravity of the SS stage is not on the action line of the driving force of the motors, and each motor can generate both one force along its own axis and one torque along the z axis, so the motion in each of the three DOFs should be realized by the cooperation of the three motors. 2. Description of the Measurement System The movement along the y axis can be achieved by synchronous control of the two Y motors, with non-output of the X motor. The movement of the X motor can be achieved by the driving force along the x axis from the X motor, and a torque from two Y motors balanced with the torque from the X motor. Rotation along the z axis can be achieved by the reverse movement of the two Y motors which can simultaneously generate two opposite forces along the y axis and one torque along the z axis. The gap between the magnet and the coil of the LVCMs guarantees the movement of XYθZ DOFs. The stroke of linear movement of the SS stage is 4 mm, that of rotation movement is 0.6°. The absolute position of the SS stage to the zero point of the lithography in the XOY plane is measured by three laser interferometers. Three pairs of permanent magnets used for XYθZ displacements measurement are fixed on the SS stage along the X and Y directions, and the three corresponding Hall sensors are mounted on the SS frame above the permanent magnets along the direction of the equipotential line, so the movement of the SS stage relative to the SS frame perpendicular to the equipotential line in the XOY plane can be measured on the premise that the SS stage is held to the expected height by the three GCs, which is crucial for the relative motion control between the LS motion and SS motion. 3.1. Installation of the Hall Sensor The installation of each Hall sensor is shown in Figure 2. The output voltage of the Hall sensor is related to the magnetic flux density of the magnetic field lines passing through the sensor, which is the 22533 Sensors 2015, 15 Sensors 2015, 15 zero offset voltage in the middle section. Considering the motion range of the SS stage and the mounting space, the two pieces of permanent magnet are placed at a certain distance to construct an arc-shaped magnetic field above the upper surface of the magnet. The Hall sensor is fixed on the SS frame, and the mounting plate of the two pieces of permanent magnet is fixed on the SS stage. The permanent magnets are of the same size, with 12.7 mm length, and 6.4 mm width and height, and the installation dimensions of Hall sensor, including the terminal, are 43 mm × 25.5 mm × 6.5 mm. There is no mechanical interference during measuring with this installation method. The distribution of magnetic field above the upper surface of the magnet at different heights is simulated as shown in Figure 3, in which the curves of the relationship between magnetic flux density at a certain height and variation of position along y direction are given. The linear interval of the magnetic flux density can be used in displacement measuring. Although the nonlinear variation of magnetic flux density in response to the y position has disadvantages for measuring, a linear interval on the curves near the middle section exists, which is sufficient for measuring y displacements. Figure 2. Installation of one Hall sensor and the permanent magnet. Figure 3. Simulation of the magnetic field, where h represents the height between the Hall sensor and the upper surface of the magnets. Figure 2. Installation of one Hall sensor and the permanent magnet. Figure 2. Installation of one Hall sensor and the permanent magnet. Figure 3. Simulation of the magnetic field, where h represents the height between the Hall sensor and the upper surface of the magnets. Figure 3. Simulation of the magnetic field, where h represents the height between the Hall sensor and the upper surface of the magnets. 3.2. 3-DOF Measuring Principle The measurement model of the Hall sensors with three DOFs is shown in Figure 4. The triangle ABC represents the arrangement of the three Hall sensors, which is invariable during measurements. In the Figure, the motion of the three pairs of magnets is replaced by the movement of the triangle ABC to A'B´C´ for concise illustration. The coordinate system of the SS stage is denoted by XOY, and the sub-coordinate 22534 Sensors 2015, 15 Sensors 2015, 15 systems of the three Hall sensors are denoted by X1O1Y1, X2O2Y2 and X3O3Y3. In the sub-coordinate systems, the directions of measurement of the Hall sensors are along their own X axes which orient the Y, Y, X directions under the XOY coordinate system, respectively. The three pairs of magnets move with the SS stage, and the Hall sensors remain relatively stationary. Figure 4. Measurement model of Hall sensors with three DOFs. Figure 4. Measurement model of Hall sensors with three DOFs. The points A, B and C represent the original positions of the three Hall sensors relative to the magnet, and the points A´, B´ and C´ represent the current positions after a translational movement and rotation of the SS stage. Let (x1, y1), (x2, y2) and (x3, y3) be the coordinates of points A, B and C in the sub-coordinate systems, and (X1, Y1), (X2, Y2) and (X3, Y3) be the coordinates in the XOY coordinate system. Taking point A for example, which equipotential line the point A is standing on can be solved by the output of the Hall sensor, but the exact coordinates of A cannot be worked out. However, the shape of triangle ABC is constant, as it is related to the installation of the magnet. Once the equipotential line of each point is known, the exact coordinates of points A, B and C can be solved. Let ݉ሬሬറ = (xm, ym) be the translation vector which represents the movement of the SS stage along the X and Y axes, and θ be the rotation angle about point O. Sensors 2015, 15 Sensors 2015, 15 3 3 3 2 2 2 2 1 2 2 2 1 1 1 2 1 2 1 2 1 2 2 1 1 2 2 2 1 2 1 2 2 2 2 2 1 2 1 2 1 2 1 2 1 2 1 2 1 2 cos(2arctan ) sin(2arctan ) ( ) ( ) ( )( ) 2( )( ) tan ( ) ( ) 2arctan 2[ ( )( )] tan ( m m x x x y y x y y x y x x y y y y x y x y y y y x x y y x x y y x x y y y y y y x x α α α θ α α ′ = − + Δ + + Δ + − + Δ + Δ + − − = − + − = − + Δ Δ − −Δ −Δ − −Δ −Δ + + = Δ 1 2 1 2 ) 2( ) y y y y         −Δ + −  (3) 2 2 2 2 1 2 2 2 1 1 1 2 1 2 1 2 1 2 2 1 1 2 2 2 1 2 1 2 ( ) ( ) ( )( ) 2( )( ) tan ( ) ( ) m y x y y x y x x y y y y x y x y y y y x x y y α Δ + + Δ + − + Δ + Δ + − − = − + −    (3) (Δ 1 2 1 2 ) 2( ) y y y y −Δ + − where Δyi = yi´− yi (I = 1, 2). In practical applications, the value of θ is less than 2º, and the line connecting the center of two Hall sensors of Y direction is not parallel to the X axis, so the denominator in Equation (3) is not zero, and the equation has a solution. Sensors 2015, 15 It can be found in Figure 4 that the equipotential line of the magnet is a series of nonlinear curves, that is, the displacement of magnet in the measuring direction of Hall sensor cannot be directly calculated by the output of Hall sensors unless the distribution of the magnetic field line is known, so the key point of measuring the three DOFs is working out the relationship between the output of the Hall sensors and the relative position of the sensors in the magnetic field. 3.2. 3-DOF Measuring Principle The relationship between ABC and A´B´C´ can be written as: cos sin ( 1,2,3) sin cos i i m i i m x x x i y y y θ θ θ θ ′ −         = + =         ′         (1) (1) Let (XO1, YO1), (XO2, YO2) and (XO3, YO3) be the coordinates of the symmetric centers of the pairs of magnets in the XOY coordinate system, so the points A, B and C in the sub-coordinate systems can be transformed into the XOY coordinate system by: Let (XO1, YO1), (XO2, YO2) and (XO3, YO3) be the coordinates of the symmetric centers of the pairs of magnets in the XOY coordinate system, so the points A, B and C in the sub-coordinate systems can be transformed into the XOY coordinate system by: 1 1 1 1 2 2 2 2 3 3 3 3 cos( / 2) sin( / 2) 0 0 0 0 sin( / 2) cos( / 2) 0 0 0 0 0 0 cos( / 2) sin( / 2) 0 0 0 0 sin( / 2) cos( / 2) 0 0 0 0 0 0 cos0 sin 0 0 0 0 0 sin 0 cos0 T T O O O O O O x X X y Y x X y Y x X y Y π π π π π π π π             −            + =       −                   −           1 1 2 2 3 3 T Y X Y X Y                     (2) (2) So if the value of points (xi, yi) and (xi´, yi´) can be obtained in response to the output of Hall sensors, the movement of the three DOFs of the SS stage in XOY plane can be given by: So if the value of points (xi, yi) and (xi´, yi´) can be obtained in response to the output of Hall sensors, the movement of the three DOFs of the SS stage in XOY plane can be given by: So if the value of points (xi, yi) and (xi´, yi´) can be obtained in response to the output of Hall sensors, the movement of the three DOFs of the SS stage in XOY plane can be given by: 22535 4. Magnetic Field Fitting by an Elliptic Function The Hall sensors are transducers which change the output voltage in response to a magnetic field while measuring. Therefore, Equation (4) representing the relationship between the output of the Hall sensors and the relative positions of the sensors in the magnetic field is crucial for displacement measuring, that is, once the output voltage of the Hall sensor is known, the equipotential line on which the Hall sensor is located can be found: ( , ) U F x y = (4) ( , ) U F x y = (4) In this paper, a magnetic field fitting method is proposed to get the position coordinates from the output voltage of the Hall sensors. The equipotential lines of the magnet are approximated to conics, so the magnetic field could be fitted by conic curves. The possible types of conic curves are given in Table 1, which is used for solving the coordinates of the points ABC through the outputs voltage of the Hall sensors. Table 1. Possible types of conic curves. Type Expression Elliptic 2 2 1 3 2 4 ( ) ( ) ( ) ( ) 1, 1,2,3; ( ) ( ) n n n n n n x f U y f U n f U f U − − + = = Parabolic 2 1 2 3 ( )[ ( )] ( ), 1,2,3; n n n n x f U y f U f U n = − + = Hyperbolic 2 2 1 3 2 4 ( ) ( ) ( ) ( ) 1, 1,2,3; ( ) ( ) n n n n n n x f U y f U n f U f U − − − = = Table 1. Possible types of conic curves. Sensors 2015, 15 Sensors 2015, 15 22536 4.1. Data Collecting System Sensors 2015, 15 22537 Sensors 2015, 15 Sensors 2015, 15 and the magnets is regulated by the lifting platform, and the displacement along the measurement direction is generated by the linear motion platform. The stroke of the platform in each plane is from 2.5 mm to 2.5 mm, with a 100 μm step size. It can be found that the sensitivity of Hall sensor is not monotonic versus h, and the optimal sensitivity occurred when h = 5 mm, where the output of the Hall sensor varies 1.53 V while the motion platform moves 1 mm. When h = 8 mm, the sensitivity of the Hall sensor is 1.15 V/mm, which is lower, but still sufficient for the displacement measurement. However, the installation space is limited by the frocks of Hall sensor and other restrictions from the motors and cables, so the h is determined to be 8 mm. Considering the installation, voltage sampling range of the A/D chip and measurement range of the Hall sensor, the height between the Hall sensor and the magnets is selected to be 8 mm. The resolution of the Hall sensor in this case is 1 μm, as shown in Figure 7c. (a) (b) (c) Figure 7. Performance of single Hall sensor: (a) Sensitivity on different height; (b) Output voltage along measurement direction on the plane of h = 0.8 mm; (c) Resolution of the Hall sensor along the measurement direction. (a) (b) (c) (a) (b) (c) Figure 7. Performance of single Hall sensor: (a) Sensitivity on different height; (b) Output voltage along measurement direction on the plane of h = 0.8 mm; (c) Resolution of the Hall sensor along the measurement direction. 4.1. Data Collecting System A four DOFs magnetic field intensity collecting system is built for magnetic field fitting, which is composed of an air-bearing turntable, a lifting platform and a linear motion platform with two DOFs as shown in Figure 5. The linear motion platform is mounted on the upper surface of the lifting platform which is supported by the air-bearing turntable. The θZ displacement can be achieved by the air-bearing turntable which features high rotation positioning accuracy of 1″. The resolution of X and Y displacements is 0.1 μm. The three pairs of magnets are fixed on the upper surface of the linear motion platform, and the corresponding Hall sensors are fixed on a gantry stack. During the data collection procedure, the Hall sensors remain stationary while the magnets move with the platforms, which is consistent with the real application. Figure 5. Data collection system with four DOFs. Figure 5. Data collection system with four DOFs. The data acquisition circuits are developed to improve the resolution and stability of the Hall sensors. According to the output signal and the magnetic field of the magnets, each Hall sensor features its own data acquisition circuit, of which the structure is shown in Figure 6. SS495A is the Hall sensor. The voltage adjust module is built by INA128U as a preamplifier which can satisfy the need of resolution and Signal to Noise Ratio (SNR). The Sallen-Key second order low pass filter is built with an AD8639 to acquire the rapidly decreasing signal at the cut off frequency. The output voltage of the low pass filter is converted to a differential signal by an ADA4932-1. The Field-Programmable Gate Array (FPGA) is the main controller which controls data acquisition and transmits digital signals to computer through RS485. Figure 6. Structure of data acquisition circuit. Figure 6. Structure of data acquisition circuit. The performance of a single Hall sensor is tested by the data acquisition system. The sensitivity of the Hall sensor at different heights is examined as shown in Figure 7a; the height between the Hall sensor 4.2. Magnetic Field Fitting by Elliptic Function The magnetic field intensity can be substituted for the output voltage of the Hall sensor. Therefore, the data of output voltage on the plane of h = 0.8 mm in the area of {(x, y)\|\|x\| ≤ 3.0 mm, \|y\| ≤ 4.0 mm} is collected, in which the measurement direction of the Hall sensor is along the X axis. Taking the installation errors and machining errors of the system into account, the data collection range is larger than the measurement range of the sensors. The interval of the data is 200 μm along the X axis, and 400 μm along the Y axis, so the voltage result data can make up to a 31 × 21 matrix. The equipotential line of the magnetic field is drawn by the data of output voltage as shown in Figure 8. The intensity of the magnetic field is indicated by the absolute value of the voltage, and the different colors represent the output voltage of the Hall sensor, which is related to the measurement direction. The voltage interval is 0.2 V. Then the equipotential line is fitted by the three types of function as listed in Table 1, respectively, to find out the most appropriate one to describe the real distribution. 22538 Sensors 2015, 15 Sensors 2015, 15 Figure 8. Equipotential line of the magnetic field represented by different colors. Figure 8. Equipotential line of the magnetic field represented by different colors. As shown in Figure 8, every equipotential line is considered to be composed of a series of discrete points of the voltage matrix, so taking the voltage of U = 0.4 V for example, the curve can be fitted by a parabolic equation (Equation (5)) with a certain value of f11(U), f12(U) and f13(U). The parameter f11(U) is one element of fn1(U) in the parabolic type as listed in Table 1, as well as the parameter f12(U) and f13(U). 2 11 12 13 ( )[ ( )] ( ) x f U y f U f U = − + (5) (5) It can be found in Figure 9 that not all points lie on the curve, so the degree of fitting is evaluated by the variance of the points. After all equipotential lines are fitted by a parabolic function, the equipotential lines in Figure 8 are also fitted by an elliptic function and hyperbolic function. 4.2. Magnetic Field Fitting by Elliptic Function The variance of every equipotential line is listed in Table 2. It can be concluded that the equipotential lines fitted by the elliptic function can achieve the optimal result. Figure 9. Equipotential line of U = 0.4 V fitted by Equation (5). Figure 9. Equipotential line of U = 0.4 V fitted by Equation (5). Sensors 2015, 15 Sensors 2015, 15 22539 Table 2. Variance of fitting data by different models. Table 2. Variance of fitting data by different models. Therefore it is decided to fit the magnetic field distribution with an elliptic equation (Equation (6)), and the focus is placed on the solution of parameters fn1(U), fn2(U), fn3(U) and fn4(U), which is also solved by a curve fitting method: 4.2. Magnetic Field Fitting by Elliptic Function Voltage of Equipotential Line Variance of Fitting Data Voltage of Equipotential Line Variance of Fitting Data Parabola Ellipse Hyperbola Parabola Ellipse Hyperbola −2.0 1.36 × 10−5 1.97 × 10−6 1.37 × 10−5 0.0 1.66 × 10−7 1.65 × 10−7 1.72 × 10−6 −1.8 1.29 × 10−4 9.33 × 10−6 1.30 × 10−4 0.2 4.97 × 10−6 3.21 × 10−6 4.98 × 10−6 −1.6 1.47 × 10−4 1.03 × 10−6 1.48 × 10−4 0.4 4.54 × 10−6 1.71 × 10−7 4.58 × 10−6 −1.4 9.28 × 10−5 1.06 × 10−6 9.33 × 10−5 0.6 1.02 × 10−5 1.45 × 10−7 1.03 × 10−5 −1.2 5.90 × 10−5 8.40 × 10−6 5.93 × 10−5 0.8 2.43 × 10−5 5.40 × 10−7 2.45 × 10−5 −1.0 3.71 × 10−5 3.75 × 10−7 3.73 × 10−5 1.0 3.94 × 10−5 5.03 × 10−7 3.96 × 10−5 −0.8 1.90 × 10−5 2.58 × 10−7 1.91 × 10−5 1.2 6.06 × 10−5 4.74 × 10−7 6.10 × 10−5 −0.6 1.01 × 10−5 1.96 × 10−7 1.01 × 10−5 1.4 2.00 × 10−5 1.91 × 10−7 2.02 × 10−5 −0.4 4.82 × 10−6 1.21 × 10−7 4.84 × 10−6 1.6 1.02 × 10−7 1.29 × 10−9 1.02 × 10−7 −0.2 1.20 × 10−6 2.05 × 10−7 1.20 × 10−6 -- -- -- -- Variance of mean 3.57 × 10−5 6.52 × 10−7 3.60 × 10−5 -- -- -- -- Therefore it is decided to fit the magnetic field distribution with an elliptic equation (Equation (6)), and the focus is placed on the solution of parameters fn1(U), fn2(U), fn3(U) and fn4(U), which is also solved by a curve fitting method: 2 2 1 3 2 4 ( ) ( ) ( ) ( ) 1 ( ) ( ) n n n n n n x f U y f U f U f U − − + = (6) (6) (6) To decrease the fitting error of the parameters, the voltage interval of the equipotential line is reduced to 0.1 V. The parameters of every equipotential line are solved as shown in Figure 10, each parameter is composed of a series of voltage, which can also be fitted by the proper function. The characteristics of each parameter are as follows: ve represented by fn1(U) is approximately linear if the outliers on both ends are ignored, 1. 4.2. Magnetic Field Fitting by Elliptic Function The curve represented by fn1(U) is approximately linear if the outliers on both ends are ignored, and can be fitted by a linear function given as fn1(U) = a·U + b; 1. The curve represented by fn1(U) is approximately linear if the outliers on both ends are ignored, and can be fitted by a linear function given as fn1(U) = a·U + b; 2. The curve represented by fn2(U) is symmetric to the vertical axis U = 0 V, and is also approximately linear on one side, so it can be fitted by the absolute form of a linear function as fn2(U) = −\|c·U + d\|. Taking the quadratic form of parameter fn2(U) in Equation (6) into account, the term fn2(U) could also be expressed by fn2(U) = c·U + d; 3. The curve represented by fn3(U) is similar to an inversely proportional function, which can be written as fn3(U) = p/(U + q) + m; 4. The curve represented by fn4(U) is composed of irregular points. However, the effect of fn4(U) on the elliptic function mainly concentrates on the degree of convergence of the curvature radius of the equipotential lines, which influences more the points far from the origin but less near the origin, so that the fn4(U) is considered to be constant in the elliptic function. nsors 2015, 15 225 Figure 10. Parameters of the elliptic function. Each curve represents one parameter in Equation (6), which should be given by analytic expression. 22540 Sensors 2015, 15 ors 2015, 15 Figure 10. Parameters of the elliptic function. Each curve represents one parameter in Equation (6), which should be given by analytic expression. Figure 10. Parameters of the elliptic function. Each curve represents one parameter in Equation (6), which should be given by analytic expression. Therefore, the four parameters in the elliptic function can be given by: Therefore, the four parameters in the elliptic function can be given by: 1 2 3 4 ( ) 1.479 0.027 ( ) 0.565 0.025 ( ) 0.042 / ( 0.031) 0.2661 ( ) 5.20 f U U f U U f U U f U = +   = +  = − + −   =  (7) (7) The fitting performance of the parameters fn1(U), fn2(U) and fn3(U) is shown in Figure 11. The parameters of the elliptic function could be well fitted by Equation (7). 4.2. Magnetic Field Fitting by Elliptic Function So the elliptic function is given by: 2 2 1.479 0.027 0.042 / ( 0.031) 0.2661 1 0.565 0.025 5.20 x U y U U − − + + +     + =     +     (8) (8) 1 0.565 0.025 5.20 U + =     +     (8) Figure 11. Fitting performance of the parameters fn1(U), fn2(U) and fn3(U). Figure 11. Fitting performance of the parameters fn1(U), fn2(U) and fn3(U). Figure 12 shows the equipotential lines of the magnetic field fitted by Equation (8). The blue points represent the output voltage collected by the data collection system, and the curves with different colors represent the equipotential lines drawn by Equation (8). In the range of \|x\| ≤ 2.0 mm, corresponding to a voltage from −1.8 V to 1.8 V, the discrete points from the experiment can be substituted by curves. Considering measurement range requirement of the Hall sensor is 4 mm, it can be theoretically concluded that the magnetic field fitting method is applicable. 22541 Sensors 2015, 15 Figure 12. Fitting performance of the parameters fn1(U), fn2(U) and fn3(U). Figure 12. Fitting performance of the parameters fn1(U), fn2(U) and fn3(U). 5. Performance The performance of the three DOFs displacement measurement system is tested in a real application as shown in Figure 13. The three Hall sensors are installed along the Y, Y, X directions, which are the same as in the data collection system. The relative height of the Hall sensor and magnets is maintained the same as in the data collection system by the three GCs. The SS stage in the experiment corresponds to a closed-loop system which is driven by the three motors and measured by laser interferometers in the XOY plane, with the LS stage in a stationary state, so the data from the laser interferometers can be used as the reference of the Hall sensors. Before moving the SS stage, the initial Hall voltage Un of each sensor is recorded. Then the SS stage is actuated to move along the X and Y directions, respectively, to obtain the displacement sensitivity kn of the X Hall sensor and Y Hall sensor. Taking the X Hall sensor for example, as shown in Figure 14, the equipotential lines of the magnetic field are replaced by a voltage with a step of 0.1 V. The Hall sensor is located on the red elliptic curve if Un = −0.8 V. Then the SS stage moves along the X direction, and the displacement sensitivity kn can be calculated by the X displacement measured by laser interferometers and the variation of the Hall voltage, which is the red line in Figure 14. The magnetic field is fitted in the data collection system, that is, the corresponding Hall voltage of the magnetic field is known. Un is the initial voltage of the X Hall sensor, kn is the displacement sensitivity along the measuring direction. The intersection of the two red curves is the initial position of the X Hall sensor, so the initial position of X Hall sensor (xn, yn) is located on the intersection of the two red curves. 22542 Sensors 2015, 15 Sensors 2015, 15 Figure 13. Three DOFs displacement measurement system installed on reticle stage of the lithography. Figure 13. Three DOFs displacement measurement system installed on reticle stage of the lithography. Figure 14. Calibration of the initial position of X Hall sensor. Figure 14. Calibration of the initial position of X Hall sensor. The resolutions of the three DOFs displacement measurement system are shown in Figures 15–17, the resolutions of X and Y displacements is 1 μm, and the resolution of the θZ displacement is 1.3″. Figure 15. The resolution of the Hall sensors in X direction. Figure 15. The resolution of the Hall sensors in X direction. 22543 Sensors 2015, 15 Figure 16. The resolution of the Hall sensors in Y direction. Figure 17. The resolution of the Hall sensors in θZ direction. The measurement error of the XY displacements is shown in Figure 18, the SS stage is controlled to ve from −2 mm to 2 mm along the X direction and Y direction, and the errors at different positions obtained by comparison of the data from the laser interferometers and the Hall sensors. It can be nd that the measurement error of the X displacement is less than 4.6 μm, and that of the Y displacement Figure 16. The resolution of the Hall sensors in Y direction. Figure 17. The resolution of the Hall sensors in θZ direction. Figure 17. The resolution of the Hall sensors in θZ direction. The measurement error of the XY displacements is shown in Figure 18, the SS stage is controlled to move from −2 mm to 2 mm along the X direction and Y direction, and the errors at different positions are obtained by comparison of the data from the laser interferometers and the Hall sensors. It can be found that the measurement error of the X displacement is less than 4.6 μm, and that of the Y displacement is less than 4.8 μm in the whole stroke. 22544 Sensors 2015, 15 Sensors 2015, 15 225 Figure 18. Measurement error of XY displacements. , Figure 18. Measurement error of XY displacements. Figure 18. Measurement error of XY displacements. The measurement error of the θZ displacement is shown in Figure 19, the SS stage is controlled to rotate from −0.3° to 0.3° about the Z direction. Sensors 2015, 15 It can be found that the measurement error of the θZ displacement is less than 7.6″ in the whole stroke. Figure 19. Measurement error of θZ displacements. Figure 19. Measurement error of θZ displacements. Sensors 2015, 15 Sensors 2015, 15 Sensors 2015, 15 Sensors 2015, 15 22545 Acknowledgments This work was supported by the National Natural Science Foundation grants 51205092, China Postdoctoral Science Foundation grants 2012M540280 and Special Postdoctoral Science Foundation grants 2014T70323. Author Contributions Bo Zhao contributed to developing the ideas of this research. Bo Zhao and Lei Wang were involved in the measurement system modeling and experimental setup, as well as the drafting of the paper. Bo Zhao and Jiu-Bin Tan carried out the experiments, including data analysis. Jiu-Bin Tan critically reviewed the paper. Conflicts of Interest The authors declare no conflict of interest. 6. Conclusions In this paper, a three degrees of freedom displacement measurement system composed of three Hall sensors is presented. The measuring principle is analyzed, and the relationship between the variation of the position of the Hall sensors and the XYθz displacements of the SS stage is given. The output of one Hall sensor may correspond with many relative positions of the magnet and the sensor, but the combination of the output of three sensors can work out the only XYθz displacements of the SS stage. Therefore a magnetic field fitting method is proposed to build the relationship between the output of three sensors and the XYθz displacements, that is, the magnetic field of the magnets is a known quantity which is given by an elliptic function at a certain measurement height during displacement measurements. The parameters of the elliptic function are also fitted by analytic expression, so that the relative movement of the SS stage can be measured by the Hall sensors. The three degrees of freedom displacement measurement system is finally validated experimentally. References 1. Lee, C.B.; Lee, S.K. Multi-degree-of-freedom motion error measurement in an ultraprecision machine using laser encoder—Review. J. Mech. Sci. Technol. 2013, 27, 141–152. 1. Lee, C.B.; Lee, S.K. Multi-degree-of-freedom motion error measurement in an ultraprecision machine using laser encoder—Review. J. Mech. Sci. Technol. 2013, 27, 141–152. 2. Aktakka, E.E.; Woo, J.K.; Egert, D.; Gordenker, R.J.M.; Najafi, K.A. Microactuation and sensing platform with active lockdown for in situ calibration of scale factor drifts in dual-axis gyroscopes. IEEE/ASME T. Mech. 2015, 20, 934–943. 3. Kim, J.A.; Bae, E.W.; Kim, S.H.; Kwak, Y.K. Design methods for six-degree-of-freedom displacement measurement systems using cooperative targets. Precis. Eng. 2002, 26, 99–104. 3. Kim, J.A.; Bae, E.W.; Kim, S.H.; Kwak, Y.K. Design methods for six-degree-of-freedom displacement measurement systems using cooperative targets. Precis. Eng. 2002, 26, 99–104. 4. Mura, A. Multi-dofs MEMS displacement sensors based on the Stewart platform theory. Microsyst. Technol. 2012, 18, 575–579. 4. Mura, A. Multi-dofs MEMS displacement sensors based on the Stewart platform theory. Microsyst. Technol. 2012, 18, 575–579. 5. Mura, A. Sensitivity analysis of a six degrees of freedom displacement measuring device. J. Mech. Eng. Sci. 2014, 228, 158–168. 5. Mura, A. Sensitivity analysis of a six degrees of freedom displacement measuring device. J. Mech. Eng. Sci. 2014, 228, 158–168. 6. Allred, C.J.; Mark, R.J.; Gregory, D.B. Real-time estimation of helicopter blade kinematics using integrated linear displacement sensors. Aerosp. Sci. Technol. 2015, 42, 274–286. 7. Rhyu, S.H.; Jung, I.S.; Kwon, B.I. 2-D modeling and characteristic analysis of a magnetic position sensor. IEEE Trans. Magn. 2005, 41, 1828–1831. 8. Han, X.T.; Cao, Q.L.; Wang, M. A linear Hall Effect displacement sensor using a stationary two-pair coil system. In Proceedings of IEEE International Instrumentation and Measurement Technology Conference, Hangzhou, China, 10–12 May 2011; pp. 1342–1345. 9. Mohammed, H.A.; Bending, S.J. Fabrication of nanoscale Bi Hall sensors by lift-off technique for applications in scanning probe microscope. Semicond. Sci. Tech. 2014, 29, doi:10.1088/0268-1242/29/8/085007. 10. Manzin, A.; Nabaei, V. Modelling of micro-Hall sensors for magnetization imaging. J. Appl. Phys. 2014, 115, doi:10.1063/1.4862090. 11. Xu, Y.; Pan, H.B.; He S.Z; Li, L. A highly sensitive CMOS digital Hall Sensor for low magnetic field applications. Sensors 2012, 12, 2162–2174. Sensors 2015, 15 Sensors 2015, 15 22546 12. Hyeonh, J.A.; Kyoung, R.K. 2D Hall sensor array for measuring the position of a magnet matrix. Int. J. Precis. Eng. Man. Green Technol. 2014, 1, 125–129. 13. Kim, S.Y.; Choi, C.; Lee, K.; Lee, W. An improved rotor position estimation with vector—Tracking observer in PMSM drives with low-resolution Hall-effect sensors. IEEE Trans. Ind. Electron. 2011, 58, 4078–4086. 14. Norhisam, M.; Ng, W.S.; Suhaidi, S.; Mohd, H.M.; Nashiren, F.M. A mobile ferromagnetic shape detection sensor using a Hall sensor array and magnetic imaging. Sensors 2011, 11, 10474–10489. 15. Kim, K.W.; Torati, S.R.; Reddy, V.; Yoon, S.S. Planar Hall resistance sensor for monitoring current. J. Magn. 2014, 19, 151–154. 16. Jiang, J.; Makinwa, K.A.A.; Kindt, W.J. A continuous-time ripple reduction technique for spinning- current Hall sensors. IEEE J. Solid-St. Circ. 2014, 49, 1525–1534. 17. Kim, W.J.; Shobhit, V.; Huzefa, S. Design and precision construction of novel magnetic-levitation- based multi-axis nanoscale positioning systems. Precis. Eng. 2007, 31, 337–350. © 2015 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/4.0/).
https://openalex.org/W2802497061	https://europepmc.org/articles/pmc5832855?pdf=render	English	null	Large-scale patterning of single cells and cell clusters in hydrogels	Scientific reports	2,018	cc-by	12,076	Large-scale patterning of single cells and cell clusters in hydrogels Xiangyu Gong1,2 & Kristen L. Mills1,2 Biophysical properties of the extracellular matrix (ECM) are known to play a significant role in cell behavior. To gain a better understanding of the effects of the biophysical microenvironment on cell behavior, the practical challenge is longitudinally monitoring behavioral variations within a population to make statistically powerful assessments. Population-level measurements mask heterogeneity in cell responses, and large-scale individual cell measurements are often performed in a one-time, snapshot manner after removing cells from their matrix. Here we present an easy and low-cost method for large- scale, longitudinal studies of heterogeneous cell behavior in 3D hydrogel matrices. Using a platform we term “the drop-patterning chip”, thousands of cells were simultaneously transferred from microwell arrays and fully embedded, only using the force of gravity, in precise patterns in 3D collagen I or Matrigel. This method allows for throughputs approaching 2D patterning methods that lack phenotypic information on cell-matrix interactions, and does not rely on special equipment and cell treatments that may result in a proximal stiff surface. With a large and yet well-organized group of cells captured in 3D matrices, we demonstrated the capability of locating selected individual cells and monitoring cell division, migration, and proliferation for multiple days. Received: 29 June 2017 Accepted: 14 February 2018 Published: xx xx xxxx Received: 29 June 2017 Accepted: 14 February 2018 Published: xx xx xxxx Received: 29 June 2017 Accepted: 14 February 2018 Published: xx xx xxxx Cell behavior is markedly variable not only between populations of cells of different types or from different tis- sues, but also within a population of cells1–4. To understand the extent of variability between or within popula- tions of cells, it is desirous to characterize a large sample of them. Typically, physical measurements on a large number of cells means removing them from physiologically relevant matrices and only capturing data at one time point (i.e., snapshot measurements)5. However, it is becoming increasingly apparent that important aspects of cell behavior are elicited by their interactions with the extracellular matrix (ECM)6–9. An example of this is the drastic difference in exhibited morphology dependent upon whether cells are plated on a 2D substrate or within a 3D matrix (Figure S1). Therefore, it would benefit a wide variety of studies to have a simple method to pattern cells within 3D matrices for observation of their behavior over extended periods of time (longitudinal). www.nature.com/scientificreports www.nature.com/scientificreports www.nature.com/scientificreports Received: 29 June 2017 Accepted: 14 February 2018 Published: xx xx xxxx Results Our platform capitalizes on the single-cell arraying power of microwells, which are fabricated by stand- ard soft lithography techniques. Once the cells are trapped in the microwells, the platform is inverted and the cells are allowed to drop into a hydrogel that is in the process of gelation. Therefore, we term the platform “the drop-patterning chip”. In this article, the detailed operation of the drop-patterning chip is first presented for large-scale arrays based on collagen I. Performance of the method is evaluated by its patterning efficiency, single-cell occupancy rate, and spatial distribution of cells in 3D collagen gel. The gentle transfer method fur- thermore leads to excellent cell viability. Finally, we present longitudinal observations of cell motility and in situ development of multicellular tumor spheroids (MCTS) from these individual cells in two representative, widely used matrices: collagen I and Matrigel. The drop-patterning chip. The drop-patterning chip consists of three layers (Fig. 1a), which create a closed chamber that is filled with hydrogel (Fig. 1b). The three layers, from bottom to top, are a poly dimethylsiloxane (PDMS) microwell substrate, a PDMS square frame spacer (thickness: approximately 600 µm), and a standard glass coverslip (thickness: 170 µm). Due to the natural adhesion of PDMS to itself and between PDMS and glass, the three layers spontaneously bond together with slight pressure applied by hand. This bond is reversible so the three components can be cleaned and reused. The microwell dimensions—30 µm in diameter and 27 µm in depth—are on the order of the size of a single cell. All data presented here are based on the experiments carried out on microwell arrays with center-to-center spacing (along both columns and rows) of 100 µm, 150 µm, or 200 µm. On each drop-patterning chip, we designed four arrays with 100-µm spacing, fours arrays of 150-µm spacing, and three arrays of 200-µm spacing. Each array is comprised of 400 microwells (20 × 20). Hundreds to thousands of cells may be captured in the wells on a chip for subsequent simultaneous patterning in a hydrogel (Fig. 1c). Step-by-step drop-patterning in collagen. The first set of steps of the drop-patterning technique includes priming the microwells with cells and assembling the chip (Fig. 2a, steps 1–5). A cell suspension (approx- imately 250 μL, 1 × 106 cells/mL) was deposited on to the PDMS substrate patterned with microwell arrays. www.nature.com/scientificreports/ supports in a 3D space. Whereas these methods allow for cell anchorage and ease of locating and image collecting, the stiff and/or 2D nature of the substrates (e.g., glass or plastic surfaces, 2–4 GPa) do not provide an accurate analog to the soft, 3D nature of the in vivo environment (e.g., breast tissue, hundreds of Pascals; human intestinal tissue, thousands of Pascals)25,26. , ) In between 2D and 3D patterning methods are overlay methods, where cells are patterned on a substrate and then covered with a layer of hydrogel or other 3D matrix. Some innovative methods to manipulate cells into pat- terns include anchoring DNA-labeled cells on a DNA-patterned substrate27 and using dielectrophoretic (DEP) forces to attract cells to patterned nodes28–30. After the cells are positioned, a layer of hydrogel may be formed on top. Researchers have also used an array of magnetic nodes to trap magnetically labeled cells in between two layers of collagen31. Position control over cell placement is indeed accurate, however these methods require special tools (e.g., molecular printing, gold coated nodes, specially treated cells) not easily accessible in every lab. Another drawback of some of these methods for mechanobiological experiments is the presence of stiff substrates and/or interfaces necessitated by the patterning methods, which may prevent the full encapsulation of the cells.it y p g y p p Fully embedding a large population of single cells at specified locations in a 3D, uniform, and soft environ- ment presents a particular challenge. No method has yet been devised in which single cells or small cell clusters may be accurately patterned, without using any externally applied forces or chemical treatments, in various soft and continuous 3D matrices, allowing for dynamic studies of individual cells’ responses to biomechanical stimuli. The goal of this work was to develop a simple and low-cost platform for patterning a large population of single cells or cell clusters completely within a 3D environment (i.e., without any undesirable discontinuous or stiff boundaries). The resulting platform should allow for longitudinal observation of the single cells or cell clusters and control over the cell-cell distance within various cell array patterns. In this article, we describe the design, fabrication, method, and performance of the platform that we developed. Large-scale patterning of single cells and cell clusters in hydrogels Embedding cells in a 3D matrix is most simply achieved by mixing cells with a liquid precursor to a syn- thetic or biological hydrogel and allowing the gelation process to encapsulate the cells. Long-term monitoring of selected single cells or cell clusters in a mid- to high-throughput fashion then becomes a significant challenge, if not impossible, as the cells are positioned randomly. Researchers have resorted to embedding small numbers of cells into a matrix for long-term studies of single-cell behavior, which eases the experimentalist’s efforts to locate cells7, but often does not provide a large enough sample set for significant statistical analyses. One way of achieving better statistics on observable cell behavior in 3D culture has been to employ a modified hanging drop protocol. Using a hydrogel precursor mixed with cells to form the hanging drops is a simple way to encapsulate cells in controllable positions for high-throughput analyses10,11. However, this method only creates macro-scale arrays and is not suitable for single-cell analysis because the number of cells in each drop will vary.f y g y p y Patterning methods and scaffolds have been devised in order to controllably position single cells or cell clusters for gathering large, longitudinal sets of data. These methods often take advantage of material surface properties, morphologies, or micropatterns to capture cells in fixed positions to promote cell attachment and elicit a mechanobiological response12–15. Microwells, for example, can be used to rather simply achieve cell place- ment16–19. Furthermore, they have not only been used as a niche where cells may proliferate, but they have also been used as a tool for transferring cells into other 2D environments20,21. Surface acoustic waves have been used to move single cells to desired positions on a 2D substrate22. Engineered scaffolds, such as polymer structures fabri- cated via direct laser writing (DLW)23 and crack-based patterning24, provide single cells with adhesive, topological 1Department of Mechanical, Aerospace, and Nuclear Engineering, Rensselaer Polytechnic Institute, 110 8th St, Troy, NY, 12180, USA. 2Center for Biotechnology and Interdisciplinary Studies, Rensselaer Polytechnic Institute, 110 8th St, Troy, NY, 12180, USA. Correspondence and requests for materials should be addressed to K.L.M. (email: millsk2@rpi.edu) Scientific REPOrTS \| (2018) 8:3849 \| DOI:10.1038/s41598-018-21989-4 1 www.nature.com/scientificreports/ Results Treating the PDMS substrate with air plasma and then incubating in bovine serum albumin (BSA) prior to seed- ing promoted surface wetting and prevented cell attachment, respectively. After allowing the cells to settle into the microwells for 5 minutes, excess medium and cells were removed from the surface by gently flushing with phosphate-buffered saline (PBS). The PDMS spacer (middle component, Fig. 1a) was then assembled atop the substrate, creating a chamber. This chamber was filled with about 300 µL collagen I solution (1.0 mg/mL) and then sealed with the coverslip (top component, Fig. 1a).h p p p g The second set of steps of the drop-patterning technique transfers the array of cells into the collagen gel (Fig. 2a, steps 6–9). The enclosed chip was inverted and the trapped cells were allowed to fall, due to gravity, out of the microwells. Adhesion of the cells to the PDMS microwell walls varied, therefore not all cells fell out of the microwells simultaneously. This means pattern fidelity may be lost and some cells may not be fully transferred into the gel. To improve the pattern fidelity, we first let the cells fall to the coverslip at room temperature, which took about 10 minutes. The chip, with most cells on the coverslip, was then prewarmed at 37 °C in an incubator for 4 minutes and 30 seconds, allowing the collagen to start gelling. Then we inverted the chip once again upon which time all cells fell from the coverslip simultaneously; adhesive interactions between the cells and the BSA-treated coverslip were weaker and more uniform than those between the cells and microwells. The synergy of the rate of collagen gelation and velocity of the falling cells resulted in the cell array becoming fully encapsulated approxi- mately on the same focal plane. After the cells were patterned in collagen gel, the whole chip was submerged in medium for on-chip 3D cell culture. PDMS chambers have been proved to be suitable for 3D cell culture due to Scientific REPOrTS \| (2018) 8:3849 \| DOI:10.1038/s41598-018-21989-4 2 www.nature.com/scientificreports/ Figure 1. Drop-patterning chip design. (a) Schematic of the three-layer configuration of the drop-patterning chip. From bottom to top: PDMS microwell substrate (1.5 mm thick), PDMS square frame spacer (inner side lengths 15 mm, thickness: approximately 600 µm), standard glass coverslip (22-mm square) (b) Photograph of an assembled chip with collagen I. Results The chip was designed to fit in a 35-mm Petri dish lid. The inset shows a section of a microwell array with inter-well spacing of 150 µm. (Scale bar: 25 mm) (c) The hierarchical organization of cells patterned in a hydrogel enables repeated tracking and longitudinal observation (time: t0 to t1) of a large sample of cells. Figure 1. Drop-patterning chip design. (a) Schematic of the three-layer configuration of the drop-patterning chip. From bottom to top: PDMS microwell substrate (1.5 mm thick), PDMS square frame spacer (inner side lengths 15 mm, thickness: approximately 600 µm), standard glass coverslip (22-mm square) (b) Photograph of an assembled chip with collagen I. The chip was designed to fit in a 35-mm Petri dish lid. The inset shows a section of a microwell array with inter-well spacing of 150 µm. (Scale bar: 25 mm) (c) The hierarchical organization of cells patterned in a hydrogel enables repeated tracking and longitudinal observation (time: t0 to t1) of a large sample of cells. the oxygen permeability of PDMS32. The coverslip was also carefully slid aside for nutrient and oxygen exchange (Figure S2). To visualize the drop-patterning method, the whole process was monitored under a bright field microscope the oxygen permeability of PDMS32. The coverslip was also carefully slid aside for nutrient and oxygen exchange (Figure S2).i To visualize the drop-patterning method, the whole process was monitored under a bright field microscope (Zeiss, Axio Vert.A1) (Fig. 2b and Movie S1). Focus was maintained on a single cell (indicated with red triangles) in the array. Immediately after the first chip inversion (labeled Time 0), the cells were still trapped in the microw- ells. After 2 minutes, the microwells were out of focus, which means cells left the microwells and were traveling through the collagen solution. It took approximately 8 minutes for the single cells and 11 minutes for most the cells and cell clusters (i.e., a doublet of cells, indicated with a blue arrow) to settle on the coverslip. Another dou- blet remained stuck in a microwell (indicated with a red circle). Then the chip was transferred into an incubator for prewarming. After an incubation time of 4 minutes and 30 seconds at 37 °C, the chip was inverted a second Scientific REPOrTS \| (2018) 8:3849 \| DOI:10.1038/s41598-018-21989-4 3 www.nature.com/scientificreports/ Figure 2. Schematics and images to illustrate the drop-patterning method. (a) 1. Results A cell suspension is seeded on a substrate containing arrays of microwells. The cells are allowed to settle for 5 minutes. 2. Excess cells are gently flushed away with PBS. 3. A PDMS spacer is assembled onto the microwell substrate to create a chamber. 4–5. Once filled with collagen solution, the entire chamber is sealed with a coverslip. 6. The chip is inverted and the trapped cells begin falling out. 7–9. When most cells settle on the glass coverslip, the chip is prewarmed at 37 °C for 4.5 minutes and then inverted again. The cells start falling back into the collagen precursor and the collagen is allowed to gel at 37 °C. After the cell pattern is encapsulated in the collagen gel, the coverslip is carefully slid aside for nutrient exchange (see also Figure S2). (b) On chip, real-time monitoring (bright-field imaging) of cell-array embedding in collagen by drop-patterning. (Movie S1) The imaging started immediately after the first inversion, and always focused on a single cell indicated with a red triangle. The cell doublet in the red circle was stuck in the well and did not fall into the collagen precursor. The blue arrow may be used to keep track of a cell/position before and after the second inversion. Bright-field images of the cells in collagen before and after gelation are presented in Figure S4 to show the structural difference between collagen precursor and collagen gel. (Scale bar: 50 µm). Figure 2. Schematics and images to illustrate the drop-patterning method. (a) 1. A cell suspension is seeded on a substrate containing arrays of microwells. The cells are allowed to settle for 5 minutes. 2. Excess cells are gently flushed away with PBS. 3. A PDMS spacer is assembled onto the microwell substrate to create a chamber. 4–5. Once filled with collagen solution, the entire chamber is sealed with a coverslip. 6. The chip is inverted and the trapped cells begin falling out. 7–9. When most cells settle on the glass coverslip, the chip is prewarmed at 37 °C for 4.5 minutes and then inverted again. The cells start falling back into the collagen precursor and the collagen is allowed to gel at 37 °C. After the cell pattern is encapsulated in the collagen gel, the coverslip is carefully slid aside for nutrient exchange (see also Figure S2). Results (b) On chip, real-time monitoring (bright-field imaging) of cell-array embedding in collagen by drop-patterning. (Movie S1) The imaging started immediately after the first inversion, and always focused on a single cell indicated with a red triangle. The cell doublet in the red circle was stuck in the well and did not fall into the collagen precursor. The blue arrow may be used to keep track of a cell/position before and after the second inversion. Bright-field images of the cells in collagen before and after gelation are presented in Figure S4 to show the structural difference between collagen precursor and collagen gel. (Scale bar: 50 µm). time and incubated at 37 °C for 30 minutes longer. The resulting, fully embedded, 3 × 3 cell array is shown in the final image of Fig. 2b. We verified the cell arrays were fully embedded in the gel-filled chamber without contacting either top or bottom surface by imaging the chip on different focal planes after the collagen gelled (Figure S3). Cell patterning efficiency. Cell trapping efficiency is considered an important measure of the performance of microwell arrays in high-throughput research on anchorage dependent cells17,33. Similarly, the percentage of cell-occupied positions in the embedded 3D pattern reflects the overall efficiency of the drop-patterning method, which we defined as cell-patterning efficiency. As shown previously, not all trapped cells fell out of the wells, which means there was a nominal discrepancy between the cell-trapping efficiency in the microwells and the cell-patterning efficiency in collagen. Both the cell-trapping efficiency and cell-patterning efficiency were calcu- lated according to equation (1), by dividing the total number of cell-occupied positions either in the microwells or in the collagen by the total number of microwells. Efficiency (%) Number of Occupied Positions Total Number of Microwells 100 (1) = × (a) Trapping efficiency (T) and patterning efficiency (P) of arrays with inter-well spacing of 100 μm, 150 μm, and 200 μm on four chips. Kruskal-Wallis test with Dunn’s post hoc for multiple comparisons was performed between efficiency data sets within individual chips. Boxes represent 25th to 75th percentile and whiskers represent minimum- maximum. No statistical difference (N.S.) occurred in the efficiencies between the different inter-well spacings except in one case. (P < 0.05). (b) Mean efficiencies of all arrays on each chip were depicted on a bar plot (mean ± s.d.) overlaid with data (P < 0.05, **P < 0.001 based on Kruskal-Wallis test with Dunn’s post hoc for multiple comparisons). (c) The number of arrays used to measure efficiencies on each chip. (d) The probability distribution of number of cells per occupied position determined from observations of all 20 × 20 arrays on the four chips. (e) Schematics illustrating the observed cell occupancy scenarios at the drop-patterned positions. increasing the trapping efficiency would lead to a higher patterning efficiency. The average cell-trapping efficiency and cell-patterning efficiency of all arrays on the four chips were 65.7 ± 11.5% and 51.2 ± 11.7%, respectively. In addition to single cells, small cell clusters, such as doublets and triplets, were trapped in the microwells and patterned in collagen (Fig. 3e). This occurred due to a combination of factors including cell size variation and the susceptibility of cells to dissociation with enzymatic digestion. For the cell line and treatment we used (see Materials and Methods), the probability of finding a given number of cells in an occupied position corresponding to the 20 × 20 positions of an array is presented in Fig. 3d. Each data point present a probability measurement based on an array. The single-cell probability was 40–60% on average and depended on how well the cells were disassociated. Even though cell-patterning efficiency may vary, the probability of the number of occupant cells per position was relatively consistent across chips. Large scale array patterning and characterization of spatial distribution in collagen. Compared to the common in vitro 3D cell culture protocol of randomly embedding hundreds or thousands of cells in a matrix, the drop-patterning chip is able to embed a similar sample size of cells in a matrix, approximately on the same focal plane in a well-organized way (Fig. 4a). The cells may be cultured for time periods up to several days (Fig. Efficiency (%) Number of Occupied Positions Total Number of Microwells 100 (1) = × (1) Cell-trapping (T) and cell-patterning (P) efficiencies for this method were measured on a total of forty 20 × 20 microwell arrays distributed over four chips (Fig. 3a). All three different inter-well spacings (100 µm, 150 µm, and 200 µm) were used. Figure 3c summarizes the number of arrays that were measured on each chip. No statistical difference was measured in the patterning and trapping efficiencies between different inter-well spacings within a chip (Fig. 3a), which suggests that inter-well distance does not markedly affect efficiencies. The discrepancies between patterning and trapping efficiencies on different chips showed consistency regardless of the fluctuation in trapping efficiency (details in Figure S5), which implies that this discrepancy is mainly due to the inherent adhesive interactions between cells and the microwell substrate material. In Fig. 3b, the overall trapping and pat- terning efficiencies on the four individual chips were calculated as the mean of the efficiencies of all arrays on each chip from Fig. 3a. The difference in trapping efficiency was statistically significant between chips. We attribute this fluctuation in trapping efficiency to the imprecise nature of the manual trapping operations. In other words, Scientific REPOrTS \| (2018) 8:3849 \| DOI:10.1038/s41598-018-21989-4 4 www.nature.com/scientificreports/ Figure 3. Cell trapping and patterning efficiencies of the drop-patterning method. (a) Trapping efficiency (T) and patterning efficiency (P) of arrays with inter-well spacing of 100 μm, 150 μm, and 200 μm on four chips. Kruskal-Wallis test with Dunn’s post hoc for multiple comparisons was performed between efficiency data sets within individual chips. Boxes represent 25th to 75th percentile and whiskers represent minimum- maximum. No statistical difference (N.S.) occurred in the efficiencies between the different inter-well spacings except in one case. (P < 0.05). (b) Mean efficiencies of all arrays on each chip were depicted on a bar plot (mean ± s.d.) overlaid with data (P < 0.05, **P < 0.001 based on Kruskal-Wallis test with Dunn’s post hoc for multiple comparisons). (c) The number of arrays used to measure efficiencies on each chip. (d) The probability distribution of number of cells per occupied position determined from observations of all 20 × 20 arrays on the four chips. (e) Schematics illustrating the observed cell occupancy scenarios at the drop-patterned positions. Figure 3. Cell trapping and patterning efficiencies of the drop-patterning method. Efficiency (%) Number of Occupied Positions Total Number of Microwells 100 (1) = × 4b). Cells can always be located in an array of hundreds of cells during multi-day incubation times. To illustrate this, we used two ROIs, which are drawn on images at the time of drop-patterning (Fig. 4a) and after three days (Fig. 4b and Figure S6). On day 3, the cells (phase-contrast images shown in Fig. 4c) were fixed and their nuclei and actin fluorescently labeled in order to visualize cell morphology as the result of the 3-day incubation within the drop-patterning chip (Fig. 4d). The majority of cells formed small MCTS with diameters of 30–60 μm. However, there were a few positions where cells exhibited a much more migratory phenotype (e.g., white arrowheads in Fig. 4d). The drop-patterning chip also demonstrated the ease of protein expression assay for the cells in selected positions via immunostaining. For example, the intermediate filament protein vimentin, Scientific REPOrTS \| (2018) 8:3849 \| DOI:10.1038/s41598-018-21989-4 5 www.nature.com/scientificreports/ Figure 4. Large-scale cell array patterning in 3D collagen. (a) An array (patterned by 20×20 microwell array, inter-well spacing: 150 μm, patterning efficiency: 61%) of single cells and small cell clusters developed into (b) an array of MCTS after three days. The air bubble trapped between the PDMS and Petri dish lid in (a) did not interfere with drop-patterning. Region of interest 1 (ROI1) and region of interest 2 (ROI2) are enclosed by blue and red lines, respectively. (Scale bars: 200 μm) (c) Phase-contrast images of the MCTS array developed from the cells in ROI1 and ROI2 after three days. (Scale bars: 50 μm) (d) Immunofluorescence imaging (blue: Hoechst, red: F-actin, maximum projection) of the MCTS array developed from the cells in ROI1 and ROI2 (enclosed in white dotted lines in c) after three days, showing the difference in morphology between MCTS. Cell migration into the collagen matrix is observed (arrow heads). (Scale bars: 50 μm). (e) Viability assay (19 hours after embedding) of drop-patterned cells with calcein AM (live cells, green) and EthD-1 (dead cells, red). (f) Cell array patterned (patterning efficiency: 65.8%) in collagen by 20×20 microwell array with inter-well spacing of 150 μm in the columns and rows showing how the cell-cell distance characterization was performed (Scale bar: 200 μm). Kernel distributions of cell-cell distances (g) in the rows (X) and (h) in the columns (Y) on four arrays are overlaid on the corresponding histograms. Efficiency (%) Number of Occupied Positions Total Number of Microwells 100 (1) = × (i) Distribution of all distance measurements overlaid with mean ± standard deviation. Figure 4. Large-scale cell array patterning in 3D collagen. (a) An array (patterned by 20×20 microwell array, inter-well spacing: 150 μm, patterning efficiency: 61%) of single cells and small cell clusters developed into (b) an array of MCTS after three days. The air bubble trapped between the PDMS and Petri dish lid in (a) did not interfere with drop-patterning. Region of interest 1 (ROI1) and region of interest 2 (ROI2) are enclosed by blue and red lines, respectively. (Scale bars: 200 μm) (c) Phase-contrast images of the MCTS array developed from the cells in ROI1 and ROI2 after three days. (Scale bars: 50 μm) (d) Immunofluorescence imaging (blue: Hoechst, red: F-actin, maximum projection) of the MCTS array developed from the cells in ROI1 and ROI2 (enclosed in white dotted lines in c) after three days, showing the difference in morphology between MCTS. Cell migration into the collagen matrix is observed (arrow heads). (Scale bars: 50 μm). (e) Viability assay (19 hours after embedding) of drop-patterned cells with calcein AM (live cells, green) and EthD-1 (dead cells, red). (f) Cell array patterned (patterning efficiency: 65.8%) in collagen by 20×20 microwell array with inter-well spacing of 150 μm in the columns and rows showing how the cell-cell distance characterization was performed (Scale bar: 200 μm). Kernel distributions of cell-cell distances (g) in the rows (X) and (h) in the columns (Y) on four arrays are overlaid on the corresponding histograms. (i) Distribution of all distance measurements overlaid with mean ± standard deviation. which is typically upregulated in cells undergoing epithelial-to-mesenchymal transition (EMT)34,35, was labeled on the array of MCTS, as shown in Figure S6.i In order to confirm that drop patterning does not damage living cells, cell viability assays were conducted via calcein acetoxymethyl ester (AM) and ethidium homodimer-1 (EthD-1) staining 19 hours after the cells had been patterned in the collagen gel. Figure 4e is a representative image of the resulting fluorescent signals on a section of an array. Two replicates (two chips prepared at different times) following the same drop pattering protocol were tested. Random regions on the chips were imaged. Efficiency (%) Number of Occupied Positions Total Number of Microwells 100 (1) = × We achieved an average viability of 98%: 458 of 470 cells in the first chip and 358 of 361 cells in the second chip survived.t i p p While cells are being drop-patterned in the collagen solution, cell drifting in the plane of the array may occur. Drifting influences the patterning fidelity (Fig. 4f). To quantify the fidelity of the drop-patterning method, we measured the horizontal (X) and vertical (Y) distances (arrows in Fig. 4f) between two cells or clusters that were patterned by neighboring microwells (150 µm spacing) immediately after collagen gelation. Kernel distri- butions of the cell-cell distance (Fig. 4g,h) were based on a total of 558 measurements in the X direction and 515 measurements in the Y direction on four arrays from four individual chips. On each array, at least 100 cell-cell spacing distances in each direction were measured. The number of measurements depended on the cell pattern- ing efficiency. The cell-cell distances in both directions in each array are also presented as the mean ± standard deviation overlaid with all measurements in Fig. 4i. Arrays from difference chips showed similar distributions in both directions. The averages and standard deviations of all measurements are 143 ± 26 µm in the X- and 144 ± 24 µm in the Y-direction. Scientific REPOrTS \| (2018) 8:3849 \| DOI:10.1038/s41598-018-21989-4 6 www.nature.com/scientificreports/ Figure 5. Longitudinal observations of 3D on-chip single cell motility and MCTS development in collagen. (a) Phase-contrast images of MCTS progression from a 3 × 3 cell array in collagen (1.0 mg/mL). The field of view is divided into nine zones (I-IX) with dotted lines. Cell protrusions over time are indicated with arrowheads (red: zone I, blue: zone II, green: zone V, white: IX) (Scale bars: 50 μm). (b) Immunofluorescence imaging (red: F-actin, blue: Hoechst) of the array if four MCTS that developed from the cells in zone V, VI, VIII, and IX (Scale bar: 50 μm). (c) Confocal reflectance imaging shows the collagen I (1.0 mg/mL) microstructure (thickness: 7 μm, maximum projection), and verifies the MCTS were fully embedding when developing (Scale bar: 50 μm). (d) 3D rendering of the MCTS array in (b) showing single cell migration (arrows) and collective migration (arrowheads) in 3D (Scale bar: 40 μm). The MCTS are color-coded by their volumes. Figure 5. Longitudinal observations of 3D on-chip single cell motility and MCTS development in collagen. www.nature.com/scientificreports/ matrix starting from day 2, with the protrusions changing directions during the first 7 hours of day 2. Also in the first half of day 2, cells from zones I, III and VII underwent phenotypic change and migrated away from their respective initial positions. After 3.5 days in the collagen gel, cells in zones II, III, IV and VII displayed the highest motility. They continued to proliferate and migrate into neighboring zones. yh y p g g g In contrast, the cells in zones V, VI, VIII and IX were not, at least initially, as motile, however, they did pro- liferate. The three single cells and one cell doublet began forming MCTS within the first two days. Only after three days did cells from these MCTS begin to form protrusive actions and migrate away from the MCTS in both single-cell (green arrowhead) and collective (white arrowheads) manners.ti g (g ) ( ) After 3.5 days, the cells in the 3 × 3 array were fixed and their nuclei (Hoechst, blue) and F-actin (phalloidin, red) were stained. A confocal image stack (height: 112 µm) was acquired from a focal plane above a tumor array to a focal plane below the tumor array (Fig. 5b). Collagen fibers were auto-fluorescent in the blue channel on both the first and the last several images with no F-actin (red) observed, which verified that all the tumors were fully encapsulated in collagen. Reflectance confocal microscopy was used to visualize detailed microstructure of collagen (thickness: 7 μm) surrounding the migrating cells (Fig. 5c). 3D-reconstruction (Fig. 5d) of the array of four solid tumors in Zones V, VI, VIII, and IX clearly shows the spatial relationship between tumors and the cells (marked with arrows) that were escaping the tumors, and reveals that cell migration in collagen did not only happen within the plane of the cell array. Longitudinal observations of cell morphology, division, and tumor development in Matrigel. Using collagen I for 3D culturing of cells provides them with a fibrillar network similar to the native ECM in which a mesenchymal phenotype and protrusive sensing are promoted. Matrigel, on the other hand, originating from the epithelial basement membrane, provides a different set of physical and biochemical cues for cells in 3D culture36,37, which promote organoid formation and tumor growth38,39. www.nature.com/scientificreports/ As Matrigel is a popular 3D culture matrix, we demonstrated that our drop-patterning method would be amenable to its usage. A Matrigel concentration of 5 mg/mL was used. According to the manufacturer, Matrigel gels rapidly at 22 °C to 37 °C. Indeed, when dropping the cells into Matrigel by inverting the chip at room temperature, we found that before cells reached the coverslip (spacer height: 600 μm), the rapid gelation of Matrigel had already captured the cells in the chamber. This process took only about 10 minutes. After the cells were no longer sinking, the chip was transferred to a 37 °C incubator for 10 minutes to fully gel the matrix. Verification of full cell embedding is shown in Figure S7. i Large-scale cell arrays were patterned in Matrigel with a nominal cell-cell spacing of 100 μm (Fig. 6a) and incubated for 3.5 days (Fig. 6b). Within the larger array, two regions of interest are outlined (ROI1: blue and ROI2: red) from which higher magnification insets are presented. The multiple-day monitoring of cell prolif- eration in ROI1 is shown in Fig. 6c. Again, ROI1 was divided into eight zones. Zone I and III were vacant. The single cells in Zones V and VII divided into doublets after one day. Instead of protruding and migrating through the matrix as in collagen I, the HCT-116 cells tended to maintain their positions and develop into MCTS in the Matrigel. After fixing the tumor arrays in Matrigel at Day 3.5, we acquired fluorescence (red: F-actin, blue: nuclei) confocal images of the solid tumors in ROI1 (Fig. 6d) and ROI2 (Fig. 6f), to further compare the morphologies of these tumors. Figure 6e shows the 3D reconstruction of ROI1, color-coded by tumor volume. From Fig. 6d,f, it is clear that no cells formed protrusions into the Matrigel. Other single-cell resolution patterns in 3D. Array patterns may be an efficient tool for ease of locating individual cells when one studies single cell behaviors in 3D matrices. However, the drop-patterning method may also be used to produce other, more complex patterns that may be designed, for example, to induce or prohibit cell-cell interactions using spatial variations in the cell density. We demonstrated this capability with a couple of different patterns that we dropped into collagen I (Fig. 7a). We illustrate the flexibility in pattern design with the letters R, P, and I (abbreviation for Rensselaer Polytechnic Institute) (Fig. www.nature.com/scientificreports/ 7b). To demonstrate how variable spac- ing may be used to investigate cell-cell interaction distances as a function of matrix properties, a concentric-circle pattern consisting of 16 radial lines was drop-patterned and the cells were incubated (Fig. 7c,d) for six days. g g y In the concentric-circle pattern design, the cell-cell spacing along the radii is 100 μm (red double-headed arrow, Fig. 7c). The cell-cell spacing on a circle is proportional to its radius (i.e., larger when the radius is larger). Thus, the area density of cells varies in radial direction and the cell-cell distance (black double-headed arrows) can be varied within one pattern. We observed that varying cell spacing might be used to probe the sensitivity of cells to signaling from neighboring cells as a function of spacing. Comparing Fig. 7c,d, we found MCTS merged when they were patterned close to each other (blue dotted ellipses). Overlaid on the image in Fig. 7d (at day 6) are the radial patterns (dotted yellow lines) and short solid lines (red and black) parallel to cell protrusions from the tumors. The lengths of the solid lines indicate the protrusion lengths. The black lines represent the cells pro- truding towards adjacent MCTS in the radial direction, while the red lines indicate protrusions between MCTS in the circumferential direction. The green circle (radius R = 500 μm) with a nominal cell-cell distance (black double-headed arrow) of approximately 200 μm encloses most cell protrusions occurring between MCTS in the circumferential direction (i.e., red solid lines). Cell patterns like the concentric circles may be potentially used as a tool to study cell-cell interactions and characterize the distance range for mechanical or chemical communica- tions between cells with respect to matrix properties. Efficiency (%) Number of Occupied Positions Total Number of Microwells 100 (1) = × (a) Phase-contrast images of MCTS progression from a 3 × 3 cell array in collagen (1.0 mg/mL). The field of view is divided into nine zones (I-IX) with dotted lines. Cell protrusions over time are indicated with arrowheads (red: zone I, blue: zone II, green: zone V, white: IX) (Scale bars: 50 μm). (b) Immunofluorescence imaging (red: F-actin, blue: Hoechst) of the array if four MCTS that developed from the cells in zone V, VI, VIII, and IX (Scale bar: 50 μm). (c) Confocal reflectance imaging shows the collagen I (1.0 mg/mL) microstructure (thickness: 7 μm, maximum projection), and verifies the MCTS were fully embedding when developing (Scale bar: 50 μm). (d) 3D rendering of the MCTS array in (b) showing single cell migration (arrows) and collective migration (arrowheads) in 3D (Scale bar: 40 μm). The MCTS are color-coded by their volumes. Longitudinal observations of single cell motility, division, and tumor development in colla- gen. As discussed previously, the drop-patterned cell arrays enable the study of single-cell behavior in 3D gels over extended periods of time (Fig. 4 and Figure S6). In Fig. 5a, an array of cells was drop-patterned into collagen I and the cells were tracked in nine positions over 3.5 days. Here, we demonstrated the ease of following the pro- gression of multiple cells over periods of time, even when not kept directly on an incubator-equipped microscope stage. Pattern fidelity allowed for easy, repeated location of the cells. The position shift of the collagen gel with respect to the PDMS microwell substrate—the gel detached after cell culture medium was added—did not affect the fidelity of the cell array. For illustration, the region with the 3 × 3 cell array has been visually divided and labeled as zones I through IX. The patterned array consisted of eight single cells (I-VIII) and one doublet (IX). All cells survived, and they divided within two days. y y Evidence of protrusive (Fig. 5a, red and blue arrowheads) and motile behaviors was readily observed over the 3.5-day period. Notably, the cells in zones I and II exhibited extensive protrusive behaviors into the collagen Scientific REPOrTS \| (2018) 8:3849 \| DOI:10.1038/s41598-018-21989-4 7 www.nature.com/scientificreports/ Discussion and Outlook Here we presented the development and performance evaluation of a method for embedding arrays and planar patterns of cells completely within hydrogels. The cells do not sit within discontinuous interfaces or at bounda- ries between stiffness-mismatched materials. Our desire for developing such a tool arose from facing the chal- lenges of obtaining significant numbers of data points on the physical behavior of single cells and/or aggregates of cells embedded in 3D matrices over time. The drop-patterning method is easy and inexpensive to employ and 8 Scientific REPOrTS \| (2018) 8:3849 \| DOI:10.1038/s41598-018-21989-4 www.nature.com/scientificreports/ Figure 6. 3D on-chip MCTS development in the basement membrane matrix Matrigel. (a array in Matrigel, which developed into (b) an array of MCTS after 3.5 days. Region of inte region of interest 2 (ROI2) are enclosed by red and blue lines, respectively. (Scale bars: 200 observation of tumor development in ROI1 over 3 days. (Scale bars: 50 μm) (d) Fluorescen (red: F-actin, blue: Hoechst) of the MCTS array developed from the cells in ROI1 after 3.5 50μm) (e) 3D rendering of the MCTS array in (d) color coded by volume (f) Fluorescenc if Figure 6. 3D on-chip MCTS development in the basement membrane matrix Matrigel. (a) Large-scale cell array in Matrigel, which developed into (b) an array of MCTS after 3.5 days. Region of interest 1 (ROI1) and region of interest 2 (ROI2) are enclosed by red and blue lines, respectively. (Scale bars: 200 μm) (c) Longitudinal observation of tumor development in ROI1 over 3 days. (Scale bars: 50 μm) (d) Fluorescence confocal imaging (red: F-actin, blue: Hoechst) of the MCTS array developed from the cells in ROI1 after 3.5 days. (Scale bar: 50 μm) (e) 3D rendering of the MCTS array in (d) color-coded by volume. (f) Fluorescence imaging (red: F-actin, blue: Hoechst) of the MCTS array developed from the cells in ROI2 after 3.5 days. (Scale bar: 50 μm). Figure 6. 3D on-chip MCTS development in the basement membrane matrix Matrigel. (a) Large-scale cell array in Matrigel, which developed into (b) an array of MCTS after 3.5 days. Region of interest 1 (ROI1) and region of interest 2 (ROI2) are enclosed by red and blue lines, respectively. (Scale bars: 200 μm) (c) Longitudinal observation of tumor development in ROI1 over 3 days. Discussion and Outlook Our experience and other reports, however, indicate that the single-cell trapping efficiency of passive methods such as microwells is highly dependent on cell type and the ease with which the cells significantly eases the task of collecting large numbers of data points for many different types of 3D cell studies including: cell-matrix interactions, cell-cell interactions, cell division mechanics, and organoid or MCTS growth. significantly eases the task of collecting large numbers of data points for many different types of 3D cell studies including: cell-matrix interactions, cell-cell interactions, cell division mechanics, and organoid or MCTS growth. To evaluate the performance of our method, we drop-patterned cells into collagen I gels using evenly spaced arrays of single-cell-sized microwells. We reported the trapping, patterning, and single-cell efficiencies from hun- dreds of cells patterned in multiple trials. Reported for similar microwell trapping technologies are trapping effi- ciencies as high as 85% to 92%16. Our experience and other reports, however, indicate that the single-cell trapping efficiency of passive methods such as microwells is highly dependent on cell type and the ease with which the cells may be dissociated. This results in single-cell efficiencies on the order of 26%40 to 34%33, and probabilities of cell numbers per well or per chamber fitting the Poisson distribution41. Our single-cell efficiency is currently on par with these other studies. We believe this may significantly be improved by designing the trapping steps around a more efficient microfluidic method42,43. including: cell-matrix interactions, cell-cell interactions, cell division mechanics, and organoid or MCTS growth. To evaluate the performance of our method, we drop-patterned cells into collagen I gels using evenly spaced arrays of single-cell-sized microwells. We reported the trapping, patterning, and single-cell efficiencies from hun- dreds of cells patterned in multiple trials. Reported for similar microwell trapping technologies are trapping effi- ciencies as high as 85% to 92%16. Our experience and other reports, however, indicate that the single-cell trapping efficiency of passive methods such as microwells is highly dependent on cell type and the ease with which the cells may be dissociated. This results in single-cell efficiencies on the order of 26%40 to 34%33, and probabilities of cell numbers per well or per chamber fitting the Poisson distribution41. Our single-cell efficiency is currently on par with these other studies. We believe this may significantly be improved by designing the trapping steps around a more efficient microfluidic method42,43. Discussion and Outlook (Scale bars: 50 μm) (d) Fluorescence confocal imaging (red: F-actin, blue: Hoechst) of the MCTS array developed from the cells in ROI1 after 3.5 days. (Scale bar: 50 μm) (e) 3D rendering of the MCTS array in (d) color-coded by volume. (f) Fluorescence imaging (red: F-actin, blue: Hoechst) of the MCTS array developed from the cells in ROI2 after 3.5 days. (Scale bar: 50 μm). Figure 6. 3D on-chip MCTS development in the basement membrane matrix Matrigel. (a) Large-sca array in Matrigel, which developed into (b) an array of MCTS after 3.5 days. Region of interest 1 (ROI region of interest 2 (ROI2) are enclosed by red and blue lines, respectively. (Scale bars: 200 μm) (c) Lon observation of tumor development in ROI1 over 3 days. (Scale bars: 50μm) (d) Fluorescence confocal Figure 6. 3D on-chip MCTS development in the basement membrane matrix Matrigel. (a) Large-scale cell array in Matrigel, which developed into (b) an array of MCTS after 3.5 days. Region of interest 1 (ROI1) and region of interest 2 (ROI2) are enclosed by red and blue lines, respectively. (Scale bars: 200 μm) (c) Longitudinal observation of tumor development in ROI1 over 3 days. (Scale bars: 50 μm) (d) Fluorescence confocal imaging (red: F-actin, blue: Hoechst) of the MCTS array developed from the cells in ROI1 after 3.5 days. (Scale bar: 50 μm) (e) 3D rendering of the MCTS array in (d) color-coded by volume. (f) Fluorescence imaging (red: F-actin, blue: Hoechst) of the MCTS array developed from the cells in ROI2 after 3.5 days. (Scale bar: 50 μm). significantly eases the task of collecting large numbers of data points for many different types of 3D cell studies including: cell-matrix interactions, cell-cell interactions, cell division mechanics, and organoid or MCTS growth. To evaluate the performance of our method, we drop-patterned cells into collagen I gels using evenly spaced arrays of single-cell-sized microwells. We reported the trapping, patterning, and single-cell efficiencies from hun- dreds of cells patterned in multiple trials. Reported for similar microwell trapping technologies are trapping effi- ciencies as high as 85% to 92%16. Scientific REPOrTS \| (2018) 8:3849 \| DOI:10.1038/s41598-018-21989-4 Discussion and Outlook fil Matrix materials and their biophysical properties are important considerations when designing and conduct- ing studies of cell behavior in 3D. Depending on the type of study, a model of the mechanical, morphological, Scientific REPOrTS \| (2018) 8:3849 \| DOI:10.1038/s41598-018-21989-4 Scientific REPOrTS \| (2018) 8:3849 \| DOI:10.1038/s41598-018-21989-4 9 www.nature.com/scientificreports/ Figure 7. Various patterns at the single-cell level in collagen produced with the drop-patterning method. (a) Photograph of the drop-patterned collagen gel in a 35-mm glass-bottomed Petri dish with millimeter-scale tumor patterns growing for six days from the single-cell level. (Scale bar: 15 mm) (b) Fluorescence images (Hoechst, maximum projection) of cell-assembled capital letters R, P, and I (abbreviation for Rensselaer Polytechnic Institute) after six days of growth in the collagen. (c) Single cells and small cell clusters were drop- patterned in collagen in concentric circles demonstrating the capability of producing varying intercellular spacing within one pattern (double-headed arrows). Ninety-nine of a total of 128 positions were occupied by cells (78% patterning efficiency). (d) Six-day culture of the cell pattern in (c) showing MCTS growth, cell migration, and cell protrusions in between MCTS (red and black solid lines). The radius R of the green circle is 500 μm. Blue dotted ellipses indicate two separate cells in (c) proliferating and merging into one MCTS in (d) after six days. (Scale bars: 200 μm). Figure 7. Various patterns at the single-cell level in collagen produced with the drop-patterning method. (a) Photograph of the drop-patterned collagen gel in a 35-mm glass-bottomed Petri dish with millimeter-scale tumor patterns growing for six days from the single-cell level. (Scale bar: 15 mm) (b) Fluorescence images (Hoechst, maximum projection) of cell-assembled capital letters R, P, and I (abbreviation for Rensselaer Polytechnic Institute) after six days of growth in the collagen. (c) Single cells and small cell clusters were drop- patterned in collagen in concentric circles demonstrating the capability of producing varying intercellular spacing within one pattern (double-headed arrows). Ninety-nine of a total of 128 positions were occupied by cells (78% patterning efficiency). (d) Six-day culture of the cell pattern in (c) showing MCTS growth, cell migration, and cell protrusions in between MCTS (red and black solid lines). The radius R of the green circle is 500 μm. Blue dotted ellipses indicate two separate cells in (c) proliferating and merging into one MCTS in (d) after six days. (Scale bars: 200 μm). Discussion and Outlook and/or biological activity of the ECM may be desired. We demonstrated that the drop-patterning method is applicable to three different types of matrix material: collagen I, Matrigel, and agarose (Figure S9). The results presented here focused on collagen I and Matrigel, but this method may be furthermore expandable to any hydro- gel formulation for which the gelation kinetics may be controlled on the time scale of minutes. This provides the basis for which cell-matrix interactions may be studied with respect to many different biochemical compositions, morphologies, and stiffness of ECM at the single-cell level in a more controllable way. p g ,f g y Compared to random mixing of cells in a gel, this well-organized method of 3D cell culture made it possible to keep track of individual cell positions in a large sample size on one chip. Individual MCTS could be easily traced back to the corresponding single cells or cell clusters that they developed from. In a demonstration of the applica- bility of this method for tracking the manifestation of cell behavioral heterogeneity within a matrix, we point out the distinctly different behaviors of individual cell development in one array in both the collagen I and Matrige. In collagen I, we observed some cells to display more spread and motile behaviors whereas others displayed a more proliferative phenotype (Fig. 5). In Matrigel, although no motile behavior was observed, the morphologies of MCTS (Fig. 6d–f) were distinctly varied. Whereas some MCTS adopted round and smooth boundaries, some MCTS (outlined with dotted lines) were less packed and adopted a “grape-like” morphology. Morphological classifications such as this have been previously described for a panel of human breast cancer cell lines44, where the loosely organized grape-like morphology has been attributed to weaker cell-cell adhesion. We believe the drop-patterning chip will become a useful tool to isolate and study the biological basis of different phenotypic behaviors displayed within the same cell line with respect to interactions with matrix materials.l p y p Technologies, such as cell microdroplets45 and micro-chambers along a microfluidic channel33, are able to provide cells a 3D environment while possibly allowing for cell proliferation to be tracked in individual droplets or chambers. However, these methods lack the ability to expand the complexity of the patterning and accurately control cell-cell interactions. The drop-patterning technique allows for a large degree of flexibility in pattern geometry (Fig. Scientific REPOrTS \| (2018) 8:3849 \| DOI:10.1038/s41598-018-21989-4 Materials and Methods Design and fabrication of the drop-patterning chip. We designed the microwell patterns on com- puter aid design (CAD) software SolidWorks (Dassault Systèmes), and all the patterns were printed on a chrome mask by a high-resolution printing service (Front Range Photo Mask, CO, USA). Micro-posts on the silicon wafer were fabricated in a negative photoresist (SU-8 3025, MicroChem, MA, USA) through the techniques of photolithography. In brief, a layer of SU-8 (27.5 μm thick) was spin-coated on a 3-inch silicon wafer for 30 sec at 2,800 rpm. After soft baking at 95 °C for 10 min, the SU-8 coating was crosslinked under the exposure of UV light through a chrome mask. After post-exposure baking at 65 °C for 1 min and 95 °C for 8 min, non-crosslinked SU-8 photoresist was washed off through the developing process, and all the crosslinked micro-posts remained on the silicon wafer. The silicon wafer was then hard baked at 180 °C for 10 min. The heights of these SU-8 features were inspected by a stylus profilometer (Veeco, DekTak 8).t i Using soft lithography techniques, we molded microwells onto a PDMS (Sylgard 184, Dow Corning) sheet via the silicon master with micro-posts. First, we treated the surface with Tridecafluoro-1,1,2,2-tetrahydrooctyl- 1-trichlorosilane (TFOCS, Gelest, PA, USA) to prevent cured PDMS sticking to the master. Then, we poured a thoroughly mixed, degassed PDMS precursor (ratio of base and curing agent = 10:1 by weight) onto the silicon master in a plastic Petri dish and allowed the PDMS to cure at 70 °C overnight. After the PDMS was fully cured, we peeled it off and cut it into the desired size (25 mm × 25 mm). A PDMS spacer ring (thickness: approximately 600 μm) was directly cut out of a plain PDMS sheet as a square, matching the size of the patterned PDMS sub- strate. A square coverslip with standard dimensions (22 mm × 22 mm, thickness: 120–160 μm) was used to seal the whole device. Before assembling, we treated the PDMS substrate and spacer ring in a plasma cleaner (Harrick Plasma) and allowed them to partially recover its natural hydrophobicity in a sterile ambient environment over- night. All three components were sterilized all the three components with 70% ethanol and then with UV light for 15 minutes. Preparation of collagen gel and Matrigel. Materials and Methods Eight parts of type I bovine collagen monomer solution (3.1 mg/mL, pH 2, PureCol, Advanced Matrix, USA) was diluted with one part of 10× PBS, and then neutral- ized to a pH of 7.2–7.6 with 0.1 M sodium hydroxide (NaOH) solution. To avoid local pH variance, the solution was pipetted up and down each time a fraction of NaOH was added. The final volume was adjusted to ten parts with ultrapure water. Using this method, the concentration of the neutralized collagen solution was 2.48 mg/mL. Based on the final concentration desired, we were able to adjust the collagen solution to any concentration lower than 2.48 mg/mL, by diluting with cell culture medium. Matrigel (growth factor-reduced, Corning Life Sciences, Lowell, MA) was diluted with cell culture medium to a final concentration of 5 mg/mL. To prevent local gelation, all the solutions and tubes were chilled and all mixing operations were conducted on ice, for both gels. Since air could be introduced via mixing process, the final collagen solution was degassed on ice in a vacuum desiccator to eliminate bubbles during gelation. Cell culture. In the study of cell proliferation and tumor growth in 3D, we chose a human colon cancer cell line HCT-116 (ATCC) as a cell model. Before loading the cells in the chip, we cultured them on tissue culture flasks in McCoy’s 5 A modified medium (Corning) with 10% (vol/vol) FBS (Gibco) and 1% penicillin/strepto- mycin (Gibco) at 37 °C and 5% CO2 in a humidified incubator. The cell culture medium was changed every other day and passaged when cells reached over 80% confluency. When the cells were embedded in 3D collagen, we continued culturing them by submerging the chip in fresh cell culture medium. Passage numbers of the cells used in this research did not exceed ten. Device preparation and assembly. Before drop-patterning, the surface of the inherently hydrophobic PDMS microwell substrate (bottom component, Fig. 1a) was treated with air plasma (Harrick Plasma) for 30 seconds, a process that renders its surface hydrophilic. The substrate was then placed in a sterile ambient envi- ronment overnight to partially recover its natural hydrophobicity. This step allowed for optimal wetting behavior while preventing cell attachment to the microwell walls. Discussion and Outlook 7) to elicit or minimize various cell interactions via mechanical or chemical cues. We furthermore envision that the drop-patterning method may be used for 3D co-culture studies with, for example, cancer asso- ciated fibroblasts, endothelial cells, or immune cells on one chip. Although single-cell microwells and arrays were used here to illustrate the use of this new method, we would like to point out that it is possible to tune the size of the microwells in order to drop-pattern aggregates of cells or pre-produced MCTS. Scientific REPOrTS \| (2018) 8:3849 \| DOI:10.1038/s41598-018-21989-4 10 www.nature.com/scientificreports/ The flexibility with which one may choose the cell patterns as well as the types of hydrogel makes this a gener- ally useful tool for fields as varied as tissue engineering, stem cell research, ex vivo cancer assays, in vitro studies on intercellular communication (e.g., neurite outgrowth of primary neurons), and tumor-microenvironment interaction (e.g., tumor angiogenesis). The drop-patterning method may also be used as part of a multiplexed design, integrated with other platforms—such as a chemotaxis device46 or an engineered blood vessel47—to create a complex engineered niche for large scale studies cell-matrix interactions. Materials and Methods The surface of the plasma-treated microwell substrate and the glass coverslip were then incubated at room temperature for one hour with a 10% bovine serum albumin (BSA) solution to further prevent cell attachment.t p After the BSA treatment, 250 μL cell suspension (1 × 106 cells/mL) was seeded on the microwell substrate and the cells were allowed to settle into the microwells for about 5 min. Then the supernatant was removed and excess, untrapped cells were gently flushed away with phosphate-buffered saline (PBS). A Kimwipe was used to carefully dry the unpatterned area near the edges of the substrate, while keeping the central, patterned area wet. A dry PDMS spacer (middle component, Fig. 1a) was then rapidly placed onto the substrate to create a chamber. This chamber was filled with a collagen solution (1.0 mg/mL) and then sealed with the coverslip (top component, Fig. 1a). Immunohistochemistry. HCT-116 tumor arrays were grown in 3D collagen I or Matrigel for multiple days. Collagen gels were then transferred into a glass bottom dish (World Precision Instruments, FL, USA) for fixation and imaging. Because Matrigel tended to break when we tried to transfer it from a chip to a new dish, we directly stained the MCTS in the Matrigel on chip and conducted confocal imaging through the coverslip (Figure S8). Scientific REPOrTS \| (2018) 8:3849 \| DOI:10.1038/s41598-018-21989-4 11 www.nature.com/scientificreports/ Specifically, the samples were washed in PBS, fixed with 3.7% paraformaldehyde at 37 °C for 30 min, and perme- abilized with 0.5% Triton X-100 at 37 °C for 30 min. After washing with PBS three times for 30 min, the samples were blocked for 10 hours in 5% BSA in PBS at room temperature. Samples were then incubated at 4 °C with a primary vimentin (Figure S6) antibody (1:50, mouse, Santa Cruz) diluted in incubation solution (PBS with 0.1% BSA) overnight. An F-actin probe rhodamine phalloidin (1:50, R415, Thermo Fisher) and a second antibody (1:50, mouse, Alexa Fluor 488, Santa Cruz) were then applied in dark at 4 °C overnight. The nuclei were then stained with Hoechst (0.2 μg/mL, Hoechst 33342, Thermo Fisher) at room temperature in dark for 4 hours. In the cell viability assays based on collagen, HCT-116 single cells or small cell clusters were stained with calcein AM and EthD-1 (LIVE/DEAD Viability Kit, Invitrogen) 19 hours after drop-patterning. Image acquisition and statistical analysis. Materials and Methods Bright field images of microwells and cells were obtained with an inverted microscope (Zeiss, Axio Vert.A1). Fluorescence images of cells/MCTS and reflectance imaged of collagen microstructure were acquired with a laser scanning confocal microscope (Zeiss, LSM 510 META). 3D reconstructions of z-stacks of MCTS arrays were performed on software ZEN (Zeiss) and 3D rendering based on confocal microscopic images was made by software Imaris 8 (Bitplane). Cell-cell distances were measured with ImageJ48. All data was presented as mean ± standard deviation (s.d.), except for Fig. 3a where the boxes represent the 25th to 75th percentile and the whiskers represent the minimum-maximum range. Statistical difference was determined by Kruskal–Wallis test with Dunn’s post hoc testing for multiple comparisons. Differences were con- sidered significant at P < 0.05. References 1. Bhatia, S., Frangioni, J. V., Hoffman, R. M., Iafrate, A. J. & Polyak, K. The challenges posed by cancer heterogeneity. Nat. Biotechnol. 30, 604–610 (2012). 2. Wang, Y., Jeong, Y., Jhiang, S. M., Yu, L. & Menq, C.-H. Quantitative Characterization of Cell Behaviors through Cell Cycle Progression via Automated Cell Tracking. PLOS ONE 9, e98762 (2014). g ( ) S., Rancourt, A., Sato, Y. & Satoh, M. S. Single-cell lineage tracking analysis reveals that an established cell line comprises ll d h i h S i R 6 (2016) g g 3. Sato, S., Rancourt, A., Sato, Y. & Satoh, M. S. Single-cell lineage tracking analysis reveals that an established cell line comprises putative cancer stem cells and their heterogeneous progeny Sci Rep 6 (2016) 3. Sato, S., Rancourt, A., Sato, Y. & Satoh, M. S. Single-cell lineage tracking analy putative cancer stem cells and their heterogeneous progeny. Sci. Rep. 6 (2016).h . Soeda, A. et al. The Evidence of Glioblastoma Heterogeneity. Sci. h g y p 5. Sung, M.-H. & McNally, J. G. Live cell imaging and systems biology. Wiley Interdiscip. Rev. Syst. Biol. Med. 3, 167–182 (2011).f h g y p 5. Sung, M.-H. & McNally, J. G. Live cell imaging and systems biology. Wiley Interdiscip. R 6. Rijal, G. & Li, W. 3D scaffolds in breast cancer research. Biomaterials 81, 135–156 (2016).i f 7. He, L. et al. Local 3D matrix confinement determines division a f 7. He, L. et al. Local 3D matrix confinement determines division axis through cell shape. Oncotarget 7, 6994–7011 (2016). 8 P l N t l A th di i l i tt d t d l f b t ll i ti t di Bi t i l 8. Peela, N. et al. A three dimensional micropatterned tumor model for breast cancer cell migration studies. Biomaterials 81, (2016).h 9. Cukierman, E., Pankov, R., Stevens, D. R. & Yamada, K. M. Taking Cell-Matrix Adhesions to the Third Dimension. Science 294 1708–1712 (2001). 0. Tung, Y.-C. et al. High-throughput 3D spheroid culture and drug testing using a 384 hanging drop array. Analyst 136, 473–478 (2011). ( ) 1. Kuo, C.-T. et al. Three-dimensional spheroid culture targeting versatile tissue bioassays using a PDMS-based hanging drop array. Sci Rep. 7, 4363 (2017). p 2. Tan, J. L., Liu, W., Nelson, C. M., Raghavan, S. & Chen, C. S. Scientific REPOrTS \| (2018) 8:3849 \| DOI:10.1038/s41598-018-21989-4 References Simple approach to micropattern cells on common culture substrates by tuning substrate wettability. Tissue Eng. 10, 865–872 (2004). y g y g 3. Leong, K., Boardman, A. K., Ma, H. & Jen, A. K.-Y. Single-Cell Patterning and Adhesion on Chemically Engineered P l (di h l il ) S f L i 25 4615 4620 (2009) y g y g 13. Leong, K., Boardman, A. K., Ma, H. & Jen, A. K.-Y. Single-Cell Patterning and Adhesion on Chemically Engineered Poly(dimethylsiloxane) Surface. Langmuir 25, 4615–4620 (2009). Poly(dimethylsiloxane) Surface. Langmuir 25, 4615–4620 (2009) y y g 14. Singhvi, R. et al. Engineering cell shape and function. Science 264, 696–698 (1994).h 14. Singhvi, R. et al. Engineering cell shape and function. Science 2 15. Théry, M., Jiménez-Dalmaroni, A., Racine, V., Bornens, M. & Jülicher, F. Experimental and theoretical study of mitotic spindle orientation. Nature 447, 493–496 (2007). 6. Lee, W. C., Rigante, S., Pisano, A. P. & Kuypers, F. A. Large-scale arrays of picolitre chambers for single-cell analysis of large cel populations. Lab. Chip 10, 2952–2958 (2010). p p p 7. Rettig, J. R. & Folch, A. Large-Scale Single-Cell Trapping And Imaging Using Microwell Arrays. Anal. Chem. 77, 5628–5634 (2005) lf l k l b f l h ll f 17. Rettig, J. R. & Folch, A. Large-Scale Single-Cell Trapping And Imaging Using Microwell Arrays. Anal. Chem. 77, 5628–5634 (2005). 18. Lutolf, M. P., Doyonnas, R., Havenstrite, K., Koleckar, K. & Blau, H. M. Perturbation of single hematopoietic stem cell fates in artificial niches. Integr. Biol. 1, 59–69 (2009).h g J g g pp g g g g y ( ) 18. Lutolf, M. P., Doyonnas, R., Havenstrite, K., Koleckar, K. & Blau, H. M. Perturbation of single hematopoietic stem cell fates in artificial niches. Integr. Biol. 1, 59–69 (2009).h i g 9. Mercey, E. et al. The application of 3D micropatterning of agarose substrate for cell culture and in situ comet assays. Biomaterials 31 3156–3165 (2010). 20. Rosenthal, A., Macdonald, A. & Voldman, J. Cell patterning chip for controlling the stem cell microenvironment. Biomaterials 28, 3208–3216 (2007).l 21. Lin, C.-H. et al. A microfluidic dual-well device for high-throughput single-cell capture and culture. Lab. Chip 15, 2928–2938 (2015). 22. Guo, F. et al. Controlling cell–cell interactions using surface acoustic waves. Proc. Natl. Acad. Sci. 112, 43–48 (2015).fh 21. Lin, C.-H. et al. A microfluidic dual-well device for high-throughput single-cell capture and culture. Lab. References Chip 15, 2928 21. Lin, C.-H. et al. A microfluidic dual-well device for high-throughput single-cell capture and culture. Lab. Chip 15, 2928–2938 (2 22 Guo F et al Controlling cell cell interactions using surface acoustic waves Proc Natl Acad Sci 112 43 48 (2015) 21. Lin, C. H. et al. A microfluidic dual well device for high throughput single cell capture and culture. Lab. Chip 15, 2928 2938 (2015). 22. Guo, F. et al. Controlling cell–cell interactions using surface acoustic waves. Proc. Natl. Acad. Sci. 112, 43–48 (2015).fh l al. Controlling cell–cell interactions using surface acoustic waves. l 22. Guo, F. et al. Controlling cell–cell interactions using surface aco 23. Klein, F. et al. Two-Component Polymer Scaffolds for Controlled Three-Dimensional Cell Culture. Adv. Mater. 23, 1341– (2011).ii 24. Moraes, C. et al. Defined topologically-complex protein matrices to manipulate cell shape via three-dimensional fiber-like patterns. Lab Chip 14, 2191–2201 (2014). p 25. Butcher, D. T., Alliston, T. & Weaver, V. M. A tense situation: forcing tumour progression. Nat. Rev. Cancer 9, 108–122 (2009). 26. Johnson, L. A. et al. Matrix Stiffness Corresponding to Strictured Bowel Induces a Fibrogenic Response in Human Co Fibroblasts. Inflamm. Bowel Dis. 19, 891–903 (2013). fl ( ) 27. Todhunter, M. E. et al. Programmed synthesis of three-dimensional tissues. Nat. Methods 12, 975–981 (2015). 28. Gray, D. S., Tan, J. L., Voldman, J. & Chen, C. S. Dielectrophoretic registration of living cells to a microelectrode array. Bio Bioelectron. 19, 1765–1774 (2004). 29. Albrecht, D. R., Underhill, G. H., Wassermann, T. B., Sah, R. L. & Bhatia, S. N. Probing the role of multicellular organization in three- dimensional microenvironments. Nat. Methods 3, 369–375 (2006). 0. Albrecht, D. R., Tsang, V. L., Sah, R. L. & Bhatia, S. N. Photo- and electropatterning of hydrogel-encapsulated living cell arrays. Lab Chip 5, 111–118 (2005).h p , ( ) 1. Okochi, M. et al. Three-dimensional cell culture array using magnetic force-based cell patterning for analysis of invasive capacity o BALB/3T3/ r L b Chip 9 3378 3384 (2009) p 31. Okochi, M. et al. Three-dimensional cell culture array using magnetic force-based cell patterning for analysis of invasive capacity of BALB/3T3/v-src. Lab. Chip 9, 3378–3384 (2009). h BALB/3T3/v-src. Lab. Chip 9, 3378–3384 (2009). p 2. Gheibi, P. et al. Microchamber Cultures of Bladder Cancer: A Platform for Characterizing Drug Responsiveness and Resistance in PDX and Primary Cancer Cells. Sci. Rep. www.nature.com/scientificreports/ www.nature.com/scientificreports/ 34. Ivaska, J. Vimentin. Small GTPases 2, 51–53 (2011). 35. Mendez, M. G., Kojima, S.-I. & Goldman, R. D. Vimentin induces changes in cell shape, motility, and adhesion during the epithelial to mesenchymal transition. FASEB J. 24, 1838–1851 (2010).h Mendez, M. G., Kojima, S.-I. & Goldman, R. D. Vimentin induces changes in cell shape, motility, and adhesion during the epithelial mesenchymal transition FASEB J 24 1838 1851 (2010) , , K j , , R g p , y, g p to mesenchymal transition. FASEB J. 24, 1838–1851 (2010).h y . Frantz, C., Stewart, K. M. & Weaver, V. M. The extracellular matri h 37. Carey, S. P., Martin, K. E. & Reinhart-King, C. A. Three-dimensional collagen matrix induces a mechanosensitive invasive epithelial phenotype. Sci. Rep. 7, srep42088 (2017).f p yp p p 38. Yamamura, K., Kibbey, M. C., Jun, S. H. & Kleinman, H. K. Effect of Matrigel and laminin peptide YIGSR on tumor growth metastasis. Semin. Cancer Biol. 4, 259–265 (1993).i 39. Fridman, R., Benton, G., Aranoutova, I., Kleinman, H. K. & Bonfil, R. D. Increased initiation and growth of tumor cell lines, cancer stem cells and biopsy material in mice using basement membrane matrix protein (Cultrex or Matrigel) co-injection. Nat. Protoc. 7, 1138 (2012). ( ) 40. Guan, Z., Jia, S., Zhu, Z., Zhang, M. & Yang, C. J. Facile and Rapid Generation of Large-Scale Microcollagen Gel Array for Long- Term Single-Cell 3D Culture and Cell Proliferation Heterogeneity Analysis. Anal. Chem. 86, 2789–2797 (2014). ll ld d ll Th d b d b d h d f fl d d l g g y y , ( ) 41. Collins, D. J., Neild, A., deMello, A., Liu, A.-Q. & Ai, Y. The Poisson distribution and beyond: methods for microfluidic droplet production and single cell encapsulation. Lab. Chip 15, 3439–3459 (2015). p g p p ( ) 42. Carlo, D. D., Wu, L. Y. & Lee, L. P. Dynamic single cell culture array. Lab. Chip 6, 1445–1449 (2006). y g y p ( ) 43. Zhang, K., Chou, C.-K., Xia, X., Hung, M.-C. & Qin, L. Block-Cell-Printing for live single-cell printing. Proc. Natl. Acad. Sci. 111, 2948–2953 (2014).hi 44. Kenny, P. A. et al. The morphologies of breast cancer cell lines in three-dimensional assays correlate with their profiles of expression. Mol. Oncol. 1, 84–96 (2007). p 45. Pan, J. et al. Acknowledgements g We thank Dr. Aram Chung, Dr. Leo Wan, and Y. Deng for technical assistance. Photolithography for this work was performed in the Micro and Nano Fabrication Clean Room (MNCR) at RPI. Confocal microscopy was done in the Microscopy Core with Dr. Sergey Pryshchep’s assistance. g We thank Dr. Aram Chung, Dr. Leo Wan, and Y. Deng for technical assistance. Photolithography for this work was performed in the Micro and Nano Fabrication Clean Room (MNCR) at RPI. Confocal microscopy was done in the Microscopy Core with Dr. Sergey Pryshchep’s assistance. Author Contributions ut o Co t but o s X.G. and K.L.M. designed the experiments and wrote the main manuscript. X.G. conducted all the experiments analyzed the data, and prepared the figures. X.G. and K.L.M. designed the experiments and wrote the main manuscript. X.G. conducted all the experiments, analyzed the data, and prepared the figures. www.nature.com/scientificreports/ Quantitative tracking of the growth of individual algal cells in microdroplet compartments. Integr. Biol. 3, 1043–1051 (2011).lf 46. Haessler, U., Kalinin, Y., Swartz, M. A. & Wu, M. An agarose-based microfluidic platform with a gradient buffer for 3D chemotaxis studies. Biomed. Microdevices 11, 827–835 (2009). 7. Zheng, Y. et al. In vitro microvessels for the study of angiogenesis and thrombosis. Proc. Natl. Acad. Sci. 109, 9342–9347 (2012). 8 S h id C A R b d W S & Eli i i K W NIH I t I J 25 f i l i N t M th d 9 671 675 (2012) 7. Zheng, Y. et al. In vitro microvessels for the study of angiogenesis and thrombosis. Proc. Natl. Acad. Sci. 109, 9342–9347 (2012). 8. Schneider, C. A., Rasband, W. S. & Eliceiri, K. W. NIH Image to ImageJ: 25 years of image analysis. Nat. Methods 9, 671–675 (2012) References 7, 12277 (2017).l 32. Gheibi, P. et al. Microchamber Cultures of Bladder Cancer: A Platform for Characterizing Drug Responsiveness and Resistance in PDX and Primary Cancer Cells. Sci. Rep. 7, 12277 (2017). 33 Zh Z Ch Y C Ch Y H L Y & Y E Mi fl idi 3D l i l d hip f i l ll i l ti d li y p ( ) 3. Zhang, Z., Chen, Y.-C., Cheng, Y.-H., Luan, Y. & Yoon, E. Microfluidics 3D gel-island chip for single cell isolation and lineage dependent drug responses study. Lab. Chip 16, 2504–2512 (2016). Scientific REPOrTS \| (2018) 8:3849 \| DOI:10.1038/s41598-018-21989-4 12 Scientific REPOrTS \| (2018) 8:3849 \| DOI:10.1038/s41598-018-21989-4 © The Author(s) 2018 Additional Information Supplementary information accompanies this paper at https://doi.org/10.1038/s41598-018-21989-4. Supplementary information accompanies this paper at https://doi.org/10.1038/s41598-018-21989-4. Competing Interests: The authors declare no competing interests. Competing Interests: The authors declare no competing interests. Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ublisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and nstitutional affiliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Cre- ative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not per- mitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. © The Author(s) 2018 Scientific REPOrTS \| (2018) 8:3849 \| DOI:10.1038/s41598-018-21989-4 13
https://openalex.org/W2751588619	http://sedici.unlp.edu.ar/bitstream/handle/10915/68065/Documento_completo.pdf?sequence=1	English	null	Structural Characterization of Natural and Processed Zircons with X-Rays and Nuclear Techniques	Advances in condensed matter physics	2,017	cc-by	7,747	Hindawi Advances in Condensed Matter Physics Volume 2017, Article ID 9707604, 9 pages https://doi.org/10.1155/2017/9707604 Hindawi Laura C. Damonte,1 Patricia C. Rivas,2 Alberto F. Pasquevich,3 Fernanda Andreola,4 Federica Bondioli,5 Anna M. Ferrari,6 Laura Tositti,7 and Giorgia Cinelli7 1Departamento de F´ısica, IFLP-CCT-CONICET, Facultad de Ciencias Exactas, Universidad Nacional de La Plata, 2Facultad de Ciencias Agrarias y Forestales, Universidad Nacional de La Plata, IFLP-CCT-CONICET, 60 y 119 s/n, 1900 La Plata, Argentina 3Departamento de F´ısica, IFLP-CCT-CONICET, Facultad de Ciencias Exactas, Universidad Nacional de La Plata CICPBA, CC 67, 1900 La Plata, Argentina g 4Department of Engineering “Enzo Ferrari”, University of Modena and Reggio Emilia, Via P. Vivarelli 10, 41125 Modena, Italy 5Department of Engineering and Architecture, University of Parma, Parco Area delle Scienze 181/A, 43124 Parma, Italy 6Department of Science and Methods of Engineering, University of Modena and Reggio Emilia, Via Amendola 2, 42100 Reggio Emilia, Italy gg y 7Department of Chemistry “G. Ciamician”, University of Bologna, Via Selmi 2, 40126 Bologna, Italy Correspondence should be addressed to Laura C. Damonte; lcdamonte@gmail.com Correspondence should be addressed to Laura C. Damonte; lcdamonte@gmail.com Received 20 March 2017; Accepted 18 June 2017; Published 10 September 2017 Received 20 March 2017; Accepted 18 June 2017; Published 10 September 2017 Academic Editor: Sergei Sergeenkov Academic Editor: Sergei Sergeenkov Copyright © 2017 Laura C. Damonte et al. Tis is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. In ceramic industry, zircon sand is widely used in diﬀerent applications because zirconia plays a role as common opacifying constituent. In particular, it is used as a basic component of glazes applied to ceramic tiles and sanitary ware as well as an opaciﬁer in unglazed bulk porcelain stoneware. Natural zircon sands are the major source of zirconium minerals for industrial applications. In this paper, long, medium, and short range studies were conducted on zirconium minerals originated from Australia, South Africa, and United States of America using conventional and less conventional techniques (i.e., X-Ray Diﬀraction (XRD), Positron Annihilation Lifetime Spectroscopy (PALS), and Perturbed Angular Correlations (PAC)) in order to reveal the type and the extension of the regions that constitute the metamict state of zircon sands and the modiﬁcations therein produced as a consequence of the industrial milling process and the thermal treatment in the production line. Additionally, HPGe gamma-ray spectroscopy conﬁrms the occurrence of signiﬁcant levels of natural radioactivity responsible for metamictization in the investigated zircon samples. Results from XRD, PALS, and PAC analysis conﬁrm that the metamict state of zircon is a dispersion of submicron disordered domains in a crystalline matrix of zircon. 1. Introduction coeﬃcient of thermal expansion, low electrical and thermal conductivity, and good corrosion resistance [1]. Zirconium is the 18th most abundant element on Earth. In nature it mainly occurs as the free oxide ZrO2 (baddeley- ite), but more commonly as zircon (ZrSiO4). Mineral sands, the common name for zircon in connection with its most typical ore, are mainly exploited in the refractory and ceramic industries. Te former application originates from zircon excellent thermophysical properties such as the low On the other hand, the widespread use of zircon sand in the ceramic industry (49% of the total consumption) depends on its use as a glazing and opacifying constituent. Natural zircon sands whose average grain size, for example, in granite rocks, is about 100–200 휇m are the major source of zirconium minerals for industrial applications; in this grain size range zircon can be used either as a feedstock to fused zirconia or in Advances in Condensed Matter Physics 2 zirconium chemical plants as foundry sand or directly in the manufacture of some refractory materials. Other industrial applications require lower particle size usually restricted to two basic types: zircon ﬂour (40–50 휇m) and micronized zircon (퐷< 5 휇m) used as an opaciﬁer [2, 3]. Te use of micronized zircon and zircon ﬂour in ceramic products has been forecast to reach the 56% in 2012 of the total world- wide industrial consumption. Australia (the main producer), together with South Africa, United States, and India, accounts for over 80% of the global zircon production [3]. included in Si-rich domains; (c) oxygen as bridges among domains, O bridges being bound to two Zr, and one Si as in the crystalline structure of zircon. Furthermore, Trachenko et al. [17] have studied the eﬀects of radiation damage on structural changes in zircon by computer simulations, proposing damaged regions with nonuniform density, increasing it from depleted core to a densiﬁed boundary. Further information on the surrounding of Zr and Hf both in natural zircons from Canada and in synthetic zircons and hafnon was collected by the determination of the quadrupole hyperﬁne interactions [18, 19]. Te remark- able sensitivity of the techniques employed in these studies allowed obtaining valuable in depth information, but the vari- ability of situations and states presented by natural zircons due to their highly diﬀerentiated geological histories requires the contribution of extensive and demanding experimental research to fulﬁll statistical representativeness of zircon com- plexity. 1. Introduction Beside about 2% Hafnium (Hf), zircon sands typically hosts REEs (Rare Earth Elements) and relevant amount of naturally occurring radionuclides, namely, tetravalent uranium (U) and thorium (T), replacing zirconium in the crystal lattice. Uranium and thorium concentration in zircon sands range, respectively, between 5–4000 ppm and 2–2000 ppm, depending on age and location of the deposits [4]. Te levels of U and T in zircon sands are therefore frequently high enough to include these materials within the so-called NORMs, Naturally Occurring Radioactive Materials requiring handling and management according to appropriate radioprotection standards [3]. As parent nuclides of 238U, 235U, and 232T radioactive families they are accompanied by their respective radioactive descendants which altogether contribute to signiﬁcant radioactivity levels in zircons compared to average rocks. Cumulative radioactivity of zircon sands ranges between 5000 and 12000 Bq/kg depending on ore age, weathering, and deposit location [5, 6]. Due to radioactive decays of U and T families, in particular in the form of alpha radiation with energy ≥4,6 MeV whose emission is accompanied by relevant recoil interactions, the structure of zircon can be heavily damaged over geological times, resulting in a partially aperiodic state, the so-called metamict state [7–9]. Te metamict state can be described as a mixture of crystalline and amorphous domains whose extent depends on the dose of radiation absorbed by zircon [8, 9]. For this reason exposure time, or, in other words, the age of zircons together with the enrichment in uranium and thorium, plays a relevant role in the degree of metamictization of each deposit. Te amorphous domains due to radiation can be restored by recrystallization during thermal processes on geological times [10]. Te crystalline and amorphous domains can be eﬃciently characterized by several instrumental techniques such as the widely employed electron microscopy (SEM [11], TEM [12]), X-Ray Diﬀraction (XRD) [8, 13], Raman spectroscopy [7], and, less frequently, Neutron Diﬀraction (ND) [14], while the structure of the domains damaged by irradiation is not well known due to the aperiodicity of the system [10, 15]. Te aperiodic or amorphous domains remain poorly studied since short range techniques of nanoscopic resolution are needed. Te EXAFS experiments by Farges and Calas in metamict zircons from Madagascar and Japan show the coexistence of Zr(VII) and Zr(VIII) coordination, Zr(VIII) corresponding to the fundamental structure of the crystalline zircon and Zr(VII) produced by metamictization [15, 16]. 1. Introduction Tis observation allowed Farges to suggest that the metamict zircon structure could be supported by the existence of three diﬀerent types of oxygen: (a) oxygen included in Zr-rich domains; (b) oxygen In this work structural analysis of several zircon samples has been carried out using short range nuclear techniques, that is, Perturbed Angular Correlation (PAC) spectroscopy and Positron Annihilation Lifetime Spectroscopy (PALS) whose results were compared and integrated with those from the long-range X-Ray Diﬀraction. In particular XRD and PAC were used to both determine the amorphous fraction derived from cumulative alpha-decay events and analyze the atomic nanoconﬁgurations and defects of zircon sands [19]. Te study covers the comparative analysis of sands of diﬀerent origins (United States, Australia, and South Africa) with the aim of characterizing the type and the extension of the regions constituting the metamict state in untreated materials from geographically distinct areas as well as the modiﬁcations observed as a consequence of the industrial milling process and the thermal treatment (annealing) in the production line. 2. Experimental Procedure Qualitative and quantitative analysis of spectra were carried out relatively to a mixed radionuclide standard solution using the Ametek-Gamma Vision sofware; analytical results were checked by certiﬁed reference materials, namely, UTS-3 and DH1-a both by CANMET. Analysis of NORM 훾-emitters was performed using a library of 훾-emissions chosen to this scope based upon the highest photonic yields % among those available for the members of all the three natural radioactive families [20]. 238U concentration was determined by the 226Ra emission at 186 keV corrected for the 235U interference, while the typical emissions of the 214Bi and 214Pb daughters were taken independently, since they are likely inﬂuenced by loss of gaseous radon intermediate, which would lead to underes- timation of the 226Ra and 238U [20]. 232T was determined through the 228Ac emission at 338 keV. Concentrations in ppm for elemental uranium and thorium were determined by the following equations in the assumption of natural isotopic composition for the former element [21]: weight of corundum originally added to the mixtures as an internal standard and therefore internally normalized [22]. Te background was successfully ﬁtted by a Chebyshev function with a variable number of coeﬃcients depending on its complexity. Te peak proﬁles were modeled using a pseudo-Voigt function with one Gaussian (GW) and two Lorentzian (퐿 , 퐿 ) coeﬃcients. Lattice constants, phase frac- tion, and coeﬃcients corresponding to sample displacement and asymmetry were also reﬁned. Te cut-oﬀvalue for the calculation of the peak proﬁles in all reﬁnements was 0.05%. 2.3. PALS Analysis. Positron Annihilation Lifetime Spec- troscopy (PALS) is based on the capability of positron to sense the electron density in the material; in consequence it is highly sensitive to open volumes in solids. In matter, a positron and an electron mainly annihilate by the emission of two gamma rays of 511 keV. Under given circumstances, both particle and antiparticle can form a bound state, called positronium (Ps) [23]. Te ortho-positronium (o-Ps, the spins of the two particles are parallel oriented) is commonly annihilated with an electron of the medium to give two gamma radiations. Te measured positron and o-Ps lifetimes are related to the open volume defect size allowing char- acterizing the defect structure of the material under study. PALS measurements were done in a conventional fast-fast coincidence system with two scintillator detectors (one BaF2 and one plastic BURLE) provided a time resolution (FWHM) of 260 ps. 2. Experimental Procedure Te samples studied in the present work are natural zircons from the United States, Australia, and South Africa (provided by Endeka Ceramics) in two diﬀerent stages of their industrial elaboration: sands (raw material) were named according to their origin as USA, AUS, and SUD, respectively (see Table 1 for chemical composition and granulometry), and sands micronized at a size of 퐷50 = 4 휇m were named as USA4, AUS4, and SUD4. Te milling process was conducted with an air-jet milling system (INCO) that is able to produce, without the help of moving and classifying parts, micronized parti- cles typically 100% below 5 휇m (laser particle size analyzer, Mastersizer 2000, Malvern) without causing damage to the products. Te micronized sands were subsequently annealed at 1000∘C for 2 h in air atmosphere using an electric oven and the processed samples obtained will be indicated as aUSA4, aAUS4, and aSUD4, respectively. 2.1. Radioactivity Measurements. Te radioactive content of the raw sand samples investigated and of the corresponding micronized derivatives was determined by high resolution 훾-ray spectrometry using a HPGe p-type detector (Ortec, now Ametek). Aliquots of about 50 g of each sample were Advances in Condensed Matter Physics 3 Table 1: Chemical composition (provided by Endeka Ceramics) for natural zircon sands. U and T content and granulometry determined in natural and micronized zircon sands. USA USA4 AUS AUS4 SUD SUD4 ZrO2 + HfO2 (%) 66.3 62.2 66.5 SiO2 32.7 33.3 32.7 TiO2 0.09 0.15 0.11 Al2O3 0.58 0.33 0.18 Fe2O3 0.058 0.04 0.06 Cr2O3 — — <0.01 CaO — — 0.01 U (ppm) 17010 1449 36022 31119 31019 33620 T (ppm) 674 644 21913 17511 1439 1569 Total activity (Bq/kg) 130101041 11390911 278402227 239501916 236001880 265202122 퐷50 (휇m) 125 4.1 108 4.6 84 4.0 Table 1: Chemical composition (provided by Endeka Ceramics) for natural zircon sands. U and T content and granulometry determined in natural and micronized zircon sands. Table 1: Chemical composition (provided by Endeka Ceramics) for natural zircon sands. U and T content and granulometry determined in natural and micronized zircon sands. accurately weighted and made up to a standard geometry in a polystyrene jar. All the samples were counted with the same procedure. Te gamma spectra were acquired for 24 hours each in order to optimize radioactivity measurements. 2. Experimental Procedure A 22NaCl (10 휇Ci) radioactive source deposited onto a kapton foil (1.42 g/cm3) and sandwiched between two identical samples was used. Te analyzed sands were uniaxial pressed into 8 mm diameter pellets of around 1 mm thick. Te source contribution and the response function were evaluated from an Hf metal reference sample using the RESOLUTION code [24]. Positron lifetime spectra of 3 × 106 counts each were recorded at room temperature and analyzed with the POSITRONFIT program [24]. Afer background and source contribution correction, the lifetime spectra 1 ppm U = 12.35 Bq/kg 238U. 1 ppm T = 4.072 Bq/kg 232T. 1 ppm U = 12.35 Bq/kg 238U. 2.2. XRD Analysis. Te X-Ray Diﬀraction spectra were recorded at room temperature using Cu-K훼radiation in a Philips PW1700 diﬀractometer. Quantitative analysis of the samples was performed by the combined Rietveld RIR (Reference Intensity Ratio) method. 10 wt% of corundum (NIST SRM 674a annealed at 1500∘C for 1 day to increase the crystallinity to 100 wt%) was added to all samples as an internal standard. Te mixtures, ground in an agate mortar, were side-loaded in an aluminum ﬂat holder in order to mini- mize the preferred orientation problems. Data were recorded in the 5∘–140∘2휃range (step size 0.02∘and 6 s counting time for each step). Te phase fractions extracted by the Rietveld RIR reﬁnements, using GSAS sofware and EXPGUI as graphical interface, were rescaled on the basis of the absolute ( ) = ∑ 푖 푖푒−푡/휏푖 (1) (1) Advances in Condensed Matter Physics 4 Counts × 105 1.0 0.5 0.0 100 200 211 312 301 20.0 40.0 60.0 80.0 100.0 120.0 2- eta (deg) Counts × 105 1.0 0.5 0.0 20.0 40.0 60.0 80.0 100.0 120.0 2- eta (deg) Counts × 105 1.0 0.5 0.0 20.0 40.0 60.0 80.0 100.0 120.0 2- eta (deg) USA4 AUS4 SUD4 Figure 1: Rietveld reﬁnements results for micronized zircon sands. Points are the experimental data while full line is the results from the ﬁts. Column bars below represent peak position for zircon and corundum phases. Te full line below represents the diﬀerence from data to reﬁnement. Counts × 105 1.0 0.5 0.0 100 200 211 312 301 20.0 40.0 60.0 80.0 100.0 120.0 2- eta (deg) USA4 were decomposed into diﬀerent exponential decays, each positron state being characterized by a positron lifetime, 휏푖, with certain intensity, 푖(normalized). 2.4. PAC Analysis. 2. Experimental Procedure Te Perturbed Angular Correlation (PAC) method is based on the hyperﬁne interactions of nuclear moments with extra nuclear ﬁelds. Tis technique provides a nanoscopic (i.e., short range) description of the Zr-ions nearest neighborhoods. Te nuclei of 181Hf, obtained by irra- diation with thermal neutrons (RA3-reactor of the Comisi´on Nacional de Energ´ıa At´omica (CNEA)), of the 180Hf natural impurities located at Zr sites constitute the radioactive probe nuclei in the zircon powders. Te 훾–훾cascade at (133–482) keV, populated by the 훽-decay of 181Hf, was used to measure the quadrupole interaction of the 482 keV (+5/2) state of 181Ta. Te time diﬀerential anisotropy was calculated from the coincidence spectra 푁(휃, ), where 휃is the angle between detectors and is the time delay between the two gamma events. Afer subtraction of chance coincidence background, time spectra corresponding to angles of 90∘and 180∘between detectors were combined to obtain the ratio Counts × 105 1.0 0.5 0.0 20.0 40.0 60.0 80.0 100.0 120.0 2- eta (deg) AUS4 푅( ) = 2 푁(180∘, ) −푁(90∘, ) 푁(180∘, ) + 2푁(90∘, ) = 퐴22퐺22 ( ) . (2) Te 퐺22( ), the perturbation function, for a static quadrupole interaction has the following expression: 퐺22 ( ) = 3 ∑ 푛=0 푆2푛푒− 푛푡cos (휔푛 ) . (3) (3) Te frequencies, 휔푛, are related to the quadrupole frequency 휔 = 휋푒푄푉 /20ℎ, by 휔푛= 푔푛(휂)휔 . Te 푔푛(휂) coeﬃcients are known functions of the asymmetry parameter, 휂, deﬁned by 휂= (푉 −푉 )/푉푧푧, where 푉푖푖are the principal com- ponents of the EFG tensor. Te exponential function takes into account a Lorentzian frequency distribution of width 훿 around the 휔 values due to disorder and/or impurities or defects present in the lattice. On account of the −3 dependence of the quadrupole hyperﬁne interaction, the technique is extremely localized and diﬀerent surroundings can be determined. Tus, a relative fraction or intensity (!푖) and the corresponding EFG parameters (휔 푖, 휂푖, 훿푖) can be determined for each neighborhood. Finally, the experimental ratio 푅( ) is ﬁtted by 푅( ) = 퐴22 ∑!푖퐺푖 22( ), where !푖are the relative fraction of nuclei experiencing a given perturbation 퐺푖 22( ) [25, 26]. Figure 1: Rietveld reﬁnements results for micronized zircon sands. Points are the experimental data while full line is the results from the ﬁts. Column bars below represent peak position for zircon and corundum phases. 2. Experimental Procedure Te full line below represents the diﬀerence from data to reﬁnement. with important ceramic companies of the Modena Ceramic district. Consequently the NORM levels detected suggest that the zircons analyzed in this work are partially metamicted. p g g p 22 PAC measurements were performed at room temperature using a standard setup with four conical BaF2 scintillation detectors with a time resolution of 0.6 ns (FWHM). Details on the experimental setup adopted in this work can be found elsewhere [27]. Te XRD patterns obtained in natural zircon sands only reveal the principal lines corresponding to ZrSiO4. Rietveld RIR reﬁnements were performed on natural and micronized samples. Te obtained cell parameters and quantitative anal- ysis are shown in Table 2 while Figure 1 shows the reﬁnement results for the micronized samples. A small decrease in lattice parameters is observed with respect to the standard ZrSiO4, probably consistent with the presence of interstitial defects typical of metamictic state. In agreement with U and T 3. Results and Discussion Te U and T contents determined in the investigated samples (see Table 1) are comparable to those published by Bruzzi et al. [5] and to those regularly checked by the authors from Bologna University in the long term collaboration Advances in Condensed Matter Physics 5 Table 2: Results of Rietveld quantitative analysis (wt%) RIR. Standard parameters for ZrSiO4 are from literature (ICSD 9582). Sample ZrSiO4 Amorphous # = $ % &2 푅wp 푅p USA 49.91 50.12 6.60591 5.98781 1.884 0.089 0.069 USA4 66.82 33.24 6.60581 5.98641 3.347 0.122 0.096 AUS 48.51 51.52 6.60751 5.99271 2.182 0.095 0.074 AUS4 59.11 40.83 6.60771 5.99241 3.675 0.124 0.091 SOU 39.71 60.31 6.60661 5.99182 2.054 0.093 0.071 SOU4 57.61 42.44 6.60721 5.99251 3.587 0.123 0.094 ZrSiO4 6.6122 5.9942 of Rietveld quantitative analysis (wt%) RIR. Standard parameters for ZrSiO4 are from literature (ICSD 9582). Table 2: Results of Rietveld quantitative analysis (wt%) RIR. Standard parameters for ZrSiO4 are from lite perﬁne quadrupole parameters ﬁtted to the experimental PAC results. Errors are indicated as subindexes. Table 3: Hyperﬁne quadrupole parameters ﬁtted to the experimental PAC results. Errors are indicated as subindexes. Sample Relative fraction ! (%) Quadrupole frequency 휔 (Mrad/s) Asymmetry parameter 휂 Frequency distribution width 훿(%) Interaction label USA 491 1011 0.061 31 1 211 1091 0.871 71 2 221 761 0.622 121 3 81 1911 0.732 31 4 AUS 401 1011 0.091 41 351 1041 0.782 152 161 661 0.774 114 91 1902 0.652 41 SUD 471 1011 0.101 51 341 1051 0.731 111 111 671 0.773 51 81 1912 0.622 41 As a consequence of the micronization process the cav- ities disappear and only crystalline materials with regions displaying intrinsic defects coexist. Further annealing heals the defects, leading to a crystalline state (Figure 2(c)). content (Table 1) USA sand is the most crystalline material as compared with the other two samples investigated. In addi- tion, the milling process reduced grain size and amorphous content (Tables 1 and 2), without changing cell parameters. g y g Te results of the quantitative Rietveld RIR analysis are in good agreement with PALS observations. Table 2 shows that the amorphous content decreases around 30% from zircon natural sands to the micronized ones favoring the crystalline ZrSiO4 phase. Te lifetime components resolved in the PALS spectra diﬀer in the natural and processed zircon (Figure 2). 3. Results and Discussion aUSA4 aAUS4 aSUD4 0 25 50 75 100 Intensity USA4 AUS4 SUD4 0 25 50 75 100 Intensity USA AUS SUD 0 25 50 75 100 Intensity Intensity Intensity aUSA4 aAUS4 aSUD4 500 1000 1500 2000 (ps) (c) USA4 AUS4 SUD4 500 1000 1500 2000 (ps) (b) USA AUS SUD 500 1000 1500 2000 (ps) (a) (b) Figure 2: Lifetime components and their intensities determined by PALS for the natural (a), the micronized (b), and the annealed micronized sands (c). Te comparison of the ! values in Table 3 and Figure 4 shows that the most relevant hyperﬁne interaction ( 1, 휔 1 = 101 Mrad/s and 40 < !1% < 49) corresponds to Zr lattice site in crystalline zircons in agreement with quadrupole fre- quency values reported previously [19, 27]. With signiﬁcant relative abundance (20 < !2% < 35) very asymmetric and slightly distributed interaction 2 of quadrupole frequency (휔 2 = 104–109 Mrad/s) similar to that of the crystalline zircon is also observed in the analyzed samples. Tis inter- action has been observed in fully metamict zircons [19, 29]. Te above-mentioned interactions are accompanied by the interactions of similar or lower abundance 3 (11 < !3% < 22) and 4 (8 < !4% < 9). PAC studies on synthetic zircon and mullite-zirconia system [27] determined the presence of a similar 3 interaction associated with aperiodic regions. Tis interaction which exists in synthetic zircons cannot be attributed to the metamict state. Interaction 4 has not been reported in natural zircons so far. Tus, PAC results show that the metamict state of zircon may consist in an atomic arrangement with two diﬀerent Zr surroundings described by 2 and 4 interactions. Furthermore, PAC allows validating the model of the metamict state of three regions (a Si-rich region, a Zr-rich region, and a bridge region) suggested by Farges [15] since the model admits two nonequivalent sites for the Zr(VII). Te fact that the 2 interaction has a quadrupole frequency similar to that of crystalline zircon ( 1) but more asymmetric and slightly distributed gives an experimental support to assigned it to Zr probes involved in the bridge zone since they resemble the zirconium environment in the crystalline zircon. 3. Results and Discussion Te evolution of the spectra appears to be consistent with the structure of natural zircon where the longest lifetimes (휏= 1800 ps) originate from positronium annihilating in cavities. Te lowest lifetimes present both in natural and in processed zircon seem rather to correspond to positron annihilating in zircon crystalline bulk (휏= 300 ps) or in vacancy-type defects of smaller size (휏= 500 ps) than the cavities where positronium annihilates. Te PAC spectra (Figure 3(a)) and their associated Fourier transform (Figure 3(b)) for natural sands are typical for polycrystalline samples. All of them were analyzed using four quadrupole components according to the perturbation factor (3) (see Table 3). Fitting the PAC spin rotation curves both for micronized sands and for those thermally treated revealed diﬀerent combinations of the four interactions determined in the natural sands. Te whole evolution of the relative fractions obtained for all the samples is shown in Figure 4. Relative to the micronization of the sands, PAC shows no changes in the amounts of the interactions. Te thermal annealing of the micronized sands instead indicates a remarkable increase of the fraction ! for interaction 1; in addition, a systematic decrease of 2 fraction is observed. Te decrease is higher for the SUD than for the USA and AUS zircon sands. If the formation of o-positronium in the cavities of the natural sands is assumed, using the semiempirical relation between the o-positronium half-life 휏(ns) and the average cavity dimension 푅(˚A) where it is annihilated [28] 휏= 1 2 [1 − 푅 푅+ 1.66 + 1 2휋sin ( 2휋푅 푅+ 1.66)] −1 (4) (4) it is possible to estimate that the natural sands cavities are approximately of 80 ˚A 3 (휏≅1800 ps). Advances in Condensed Matter Physics 6 6 USA AUS SUD USA AUS SUD 0 25 50 75 100 Intensity 500 1000 1500 2000 (ps) (a) USA4 AUS4 SUD4 USA4 AUS4 SUD4 0 25 50 75 100 Intensity 500 1000 1500 2000 (ps) (b) aUSA4 aAUS4 aSUD4 aUSA4 aAUS4 aSUD4 0 25 50 75 100 Intensity 500 1000 1500 2000 (ps) (c) Figure 2: Lifetime components and their intensities determined by PALS for the natural (a), the micronized (b), and the annealed micronized sands (c). 3. Results and Discussion On the other hand, quadrupole parameters of the 4 interaction are close to those reported for the Zr(VII) in the tetragonal metastable form of ZrO2 [30]; therefore this interaction can be directly associated with the Zr-rich regions proposed in the model. Simple calculations of the quadrupole interaction based on the zircon structure, using the point charge model [31] for the ﬁrst Zr coordination, provide addi- tional support since they suggest the existence of interactions of high asymmetry parameter 휂for conﬁgurations of one oxygen vacancy in the ZrO8 group. Summing up, in terms of the model proposed by Farges [15] for the metamict state of the zircon, it is possible to assign the following interpretation to the quadrupole interactions observed in the described experiment: Tis interaction which exists in synthetic zircons cannot be attributed to the metamict state. Interaction 4 has not been reported in natural zircons so far. Tus, PAC results show that the metamict state of zircon may consist in an atomic arrangement with two diﬀerent Zr surroundings described by 2 and 4 interactions. Furthermore, PAC allows validating the model of the metamict state of three regions (a Si-rich region, a Zr-rich region, and a bridge region) suggested by Farges [15] since the model admits two nonequivalent sites for the Zr(VII). Te fact that the 2 interaction has a quadrupole frequency similar to that of crystalline zircon ( 1) but more (i) 1 and 3 interactions are the corresponding ones to probes immersed in crystalline zircon ( 1) that coexist with aperiodic zircon ( 3). (ii) 2 and 4 correspond to two nonequivalent zirconium sites of the metamict zircon in Zr-rich regions ( 4) and in the bridge region that contains Zr and Si ( 2). In this circumstance, the addition of the relative fractions of 2 and 4 represents the extent or degree of metamictization Advances in Condensed Matter Physics 7 USA AUS SUD 0.00 0.05 −A2G2 (t) 0 10 30 40 20 Time (ns) 0 10 30 40 20 Time (ns) 0 10 30 40 20 Time (ns) (a) USA AUS SUD 1000 2000 3000 0 4000 Frequency (krad/s) 1000 2000 3000 0 4000 Frequency (krad/s) 1000 2000 3000 0 4000 Frequency (krad/s) (b) Figure 3: PAC spectra for zircon sands. Te experimental spin rotation curves 퐴2퐺2( ) are shown on (a); the Fourier transform amplitude spectra are plotted on (b). 3. Results and Discussion Solid lines represent the least-squares-ﬁtted PAC spectra and their respective Fourier transforms obtained with the values given in Table 3 for the ﬁtting parameters. SU 0 10 30 40 20 Time (ns) (a) SUD 1000 2000 3000 0 4000 Frequency (krad/s) (b) (a) (b) Figure 3: PAC spectra for zircon sands. Te experimental spin rotation curves 퐴2퐺2( ) are shown on (a); the Fourier transform amplitude spectra are plotted on (b). Solid lines represent the least-squares-ﬁtted PAC spectra and their respective Fourier transforms obtained with the values given in Table 3 for the ﬁtting parameters. of the zircons: 29, 44, and 42% for USA, AUS, and SUD, respectively (see Table 3). Te values show a reasonable agree- ment with those expected from the content of radioactive ele- ments determined in the sands (see Table 1) and consequently with radiation dose [9]. According to Farges and modeling the Zr-rich region as the core of a sphere surrounded by the bridge region as shown in Figure 5, it is possible to estimate an order of a magnitude for the size of the metamict regions based on the relative abundance of the 2 and 4 interactions. In fact, the ratio of the relative fractions !2/(!2 + !4) is proportional to the ratio between the volume containing PAC probes experiencing 2 interaction and the volume containing all PAC probes of the metamict zone. Tis last volume ratio region has a radius of about 40 ˚A, similar to other author’s proposals (Weber et al. 1999 and [17]), our experimental results of the ratio !2/(!2 + !4), shown in Table 4, allow estimating the width bridge zone in about 15 ˚A (dashed sectors of Figure 6). g Te three applied techniques, XRD, PALS, and PAC, report similar degree of metamictization for all natural sands in agreement with the determined U and T content. Te proposed model based on PAC results to describe the metamictic region is consistent with the presence of the small cavities determined by PALS within the Zr-rich region which is the less dense one. Tis feature is also in agreement with the computer simulations reported by Trachenko et al. [17]. A gradual diminution of the metamictic state as process- ing advances (micronized and annealed samples) is observed by all the techniques. 3. Results and Discussion Te zircon sands contain depleted cavities of about 80 ˚A 3 while the metamict region has an extension of the order of 40 ˚A consisting in defective regions. Tese regions present Zr-rich and Si-rich zones sep- d b h f b d f l Table 4: Relative abundances representing the amount of PAC probes in the bridge region (!2) relative to all the PAC probes in the metamict region (!2 + !4). Geological origin of the samples Natural sand Micronized sand Annealed micronized sand USA 74 71 59 AUS 80 80 87 SUD 80 76 73 Sands Micronized sands sands Annealed micronized Sands Micronized sands sands Annealed micronized Sands Micronized sands sands Annealed micronized USA AUS SUD 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 f (%) f (%) f (%) Figure 4: PAC relative fractions evolution determined for the natural, the micronized, and the annealed micronized zircon sands. Symbols represent the relative fractions of the four hyperﬁne interactions reported in Table 3: 1 (full square), 2 (open square), 3 (full circle), and 4 (open circle). Zr-rich region Si-rich region R Bridge region thickness r Figure 5: Schematic representation of metamict state as proposed by the authors. R = 40 Å R = 80 Å R = 150 Å R = 300 Å R = 1150 Å 0 10 20 30 40 50 60 70 80 90 100 Relative abundance f2/(f2 + f4) (%) 0 20 50 10 30 40 Bridge region thickness (Å) Figure 6: Relative abundances of the various surroundings of the PAC probe nuclei !2/(!2 + !4) in the metamict state as a function of the bridge region thickness. degree decreases from AUS, SUD, and USA natural sands. Besides, USA sands display better zircon crystallinity than the h l d Sands Micronized sands sands Annealed micronized USA 0 20 40 60 80 100 f (%) Zr-rich region Si-rich region R Bridge region thickness r Figure 5: Schematic representation of metamict state as proposed by the authors. 3. Results and Discussion R = 40 Å R = 80 Å R = 150 Å R = 300 Å R = 1150 Å 0 10 20 30 40 50 60 70 80 90 100 Relative abundance f2/(f2 + f4) (%) 0 20 50 10 30 40 Bridge region thickness (Å) Figure 6: Relative abundances of the various surroundings of the PAC probe nuclei !2/(!2 + !4) in the metamict state as a function of the bridge region thickness. R = 40 Å R = 80 Å R = 150 Å R = 300 Å R = 1150 Å 0 10 20 30 40 50 60 70 80 90 100 Relative abundance f2/(f2 + f4) (%) 0 20 50 10 30 40 Bridge region thickness (Å) Bridge region thickness (Å) Figure 6: Relative abundances of the various surroundings of the PAC probe nuclei !2/(!2 + !4) in the metamict state as a function of the bridge region thickness. Figure 4: PAC relative fractions evolution determined for the natural, the micronized, and the annealed micronized zircon sands. Symbols represent the relative fractions of the four hyperﬁne interactions reported in Table 3: 1 (full square), 2 (open square), 3 (full circle), and 4 (open circle). degree decreases from AUS, SUD, and USA natural sands. Besides, USA sands display better zircon crystallinity than the other two natural sands. Te results have conﬁrmed that the metamict state of zircon natural sands is a dispersion of submicron domains in a crystalline matrix of zircon. Te zircon sands contain depleted cavities of about 80 ˚A 3 while the metamict region has an extension of the order of 40 ˚A consisting in defective regions. Tese regions present Zr-rich and Si-rich zones sep- arated by a thin interface or bridge region of approximately 15 ˚A thick (2-3 zircon cell constants). degree decreases from AUS, SUD, and USA natural sands. Besides, USA sands display better zircon crystallinity than the other two natural sands. Te results have conﬁrmed that the metamict state of zircon natural sands is a dispersion of submicron domains in a crystalline matrix of zircon. Te zircon sands contain depleted cavities of about 80 ˚A 3 while the metamict region has an extension of the order of 40 ˚A consisting in defective regions. Tese regions present Zr-rich and Si-rich zones sep- arated by a thin interface or bridge region of approximately 15 ˚A thick (2-3 zircon cell constants). 3. Results and Discussion However, while XRD and PALS show an apparent complete recovery of the crystalline state, PAC senses residual Zr environments inside the metamictic state. 100 [1 −(푅− 푅 ) 3 ] (5) (5) is plotted in Figure 6 versus the bridge region thickness , for diﬀerent spheres sizes 푅. If we assume that the metamictic Advances in Condensed Matter Physics 8 Table 4: Relative abundances representing the amount of PAC probes in the bridge region (!2) relative to all the PAC probes in the metamict region (!2 + !4). Geological origin of the samples Natural sand Micronized sand Annealed micronized sand USA 74 71 59 AUS 80 80 87 SUD 80 76 73 Sands Micronized sands sands Annealed micronized Sands Micronized sands sands Annealed micronized Sands Micronized sands sands Annealed micronized USA AUS SUD 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 f (%) f (%) f (%) Figure 4: PAC relative fractions evolution determined for the natural, the micronized, and the annealed micronized zircon sands. Symbols represent the relative fractions of the four hyperﬁne interactions reported in Table 3: 1 (full square), 2 (open square), 3 (full circle), and 4 (open circle). 4. Conclusions In this paper long, medium, and short range studies (X-Ray Diﬀraction (XRD), Positron Annihilation Lifetime Spectro- scopy (PALS), and Perturbed Angular Correlations, (PAC)) were successfully applied to characterize the zirconium min- l d f l h f d d Zr-rich region Si-rich region R Bridge region thickness r Figure 5: Schematic representation of metamict state as proposed by the authors. R = 40 Å R = 80 Å R = 150 Å R = 300 Å R = 1150 Å 0 10 20 30 40 50 60 70 80 90 100 Relative abundance f2/(f2 + f4) (%) 0 20 50 10 30 40 Bridge region thickness (Å) Figure 6: Relative abundances of the various surroundings of the PAC probe nuclei !2/(!2 + !4) in the metamict state as a function of the bridge region thickness. degree decreases from AUS, SUD, and USA natural sands. Besides, USA sands display better zircon crystallinity than the other two natural sands. Te results have conﬁrmed that the metamict state of zircon natural sands is a dispersion of submicron domains in a crystalline matrix of zircon. References [21] D. C. Stromswold, “Calibration facilites for borehole and surface environmental radiation measurements,” Journal of Radioana- lytical and Nuclear Chemistry Articles, vol. 194, no. 2, pp. 393– 401, 1995. [1] S. Motoi, “Review of ceramic materials in zircon and zirconia system,” Bulletin of the Ceramic Society of Japan, vol. 7, no. 3, pp. 162–168, 1972. [22] B. H. Toby, “EXPGUI, a graphical user interface for GSAS,” Jour- nal of Applied Crystallography, vol. 34, no. 2, pp. 210–213, 2001. [2] M. Lynch, “Zircon opaciﬁer markets and trends,” Asian Ceram- ics, pp. 36–41, 2007. [23] W. Brandt and A. Dupasquier, Positron Solid State Physics, North-Holland, Amsterdam, Netherlands, 1983. [3] IAEA, “Radiation protection and NORM residue management in the zircon and zirconia industries,” Tech. Rep. 51, Interna- tional Atomic Energy Agency ed., Vienna, Austria, 2007. [24] P. Kirkegaard and M. Eldrup, “Positronﬁt extended: a new version of a program for analysing position lifetime spectra,” Computer Physics Communications, vol. 7, no. 7, pp. 401–409, 1974. [4] S. Ellsworth, A. Navrotsky, and R. C. Ewing, “Energetics of radiation damage in natural zircon (ZrSiO4),” Physics and Chemistry of Minerals, vol. 21, no. 3, pp. 140–149, 1994. [25] H. Frauenfelder and R. M. Steﬀen, “Alpha, Beta- and Gamma- ray spectroscopy,” in Angular Correlation, K. Siegbahn, Ed., p. 997, North-Holland, Amsterdam, Netherlands, 1965. [5] L. Bruzzi, M. Baroni, G. Mazzotti, R. Mele, and S. Righi, “Radio- activity in raw materials and end products in the Italian ceramics industry,” Journal of Environmental Radioactivity, vol. 47, no. 2, pp. 171–181, 2000. [26] R. Dogra, A. P. Byrne, and M. C. Ridgway, “Te potential of the perturbed angular correlation technique in characterizing semiconductors,” Journal of Electronic Materials, vol. 38, no. 5, pp. 623–634, 2009. [6] J. M. Hanchar, Zircon, vol. 53 of Reviews in Mineralogy and Geo- chemistry, Mineralogical Society of America, USA, 1 edition, 2003. [27] N. M. Rendtorﬀ, M. S. Conconi, E. F. Aglietti et al., “Phase quan- tiﬁcation of mullite-zirconia and zircon commercial powders using PAC and XRD techniques,” Hyperfine Interactions, vol. 198, no. 1, pp. 211–218, 2010. [7] M. Zhang, E. K. H. Salje, I. Farnan et al., “Metamictization of zir- con: Raman spectroscopic study,” Journal of Physics Condensed Matter, vol. 12, no. 8, pp. 1915–1925, 2000. [8] S. R´ıos, T. Malcherek, E. K. Salje, and C. Domeneghetti, “Local- ized defects in radiation-damaged zircon,” Acta Crystallograph- ica Section B: Structural Science, vol. 56, no. 6, pp. Conflicts of Interest [17] K. Trachenko, M. T. Dove, and E. K. Salje, “Structural changes in zircon under @-decay irradiation,” Physical Review B, vol. 65, no. 18, Article ID 189902, 2002. Te authors declare that there are no conﬂicts of interest regarding the publication of this paper. [18] H. Jaeger and S. P. McBride, “Perturbed angular correlation measurement of the electric ﬁeld gradient at 181Ta in ZrSiO4 and HfSiO4,” Hyperfine Interactions, vol. 177, no. 1-3, pp. 51–56, 2007. Acknowledgments [19] H. Jaeger, M. P. Rambo, and R. E. Klueg, “Hyperﬁne interaction of metamict zircon studied by PAC spectroscopy,” Hyperfine Interactions, vol. 136-137, no. 3-8, pp. 515–521, 2001. Te authors are grateful to Endeka Ceramics (Modena, Italy) for providing the zircon sand samples used in the present study. [20] G. R. Gilmour, Gamma-Ray Spectrometry, John Wiley & Sons, 2008. Disclosure [15] F. Farges, “Te structure of metamict zircon: a temperature- dependent EXAFS study,” Physics and Chemistry of Minerals, vol. 20, no. 7, pp. 504–514, 1994. Present aﬃliation for Giorgia Cinelli is European Commis- sion, Joint Research Centre, Directorate G, Nuclear Safety & Security, Unit G.10, Knowledge for Nuclear Safety, Safeguards and Security, Via Enrico Fermi 2749, 21027 Ispra, Italy. [16] F. Farges and G. Calas, “Structural analysis of radiation damage in zircon and thorite: an X- ray absorption spectroscopic study,” American Mineralogist, vol. 76, no. 1-2, pp. 60–73, 1991. Advances in Condensed Matter Physics Te complete recovery of the zircon lattice afer an annealing treatment was conﬁrmed by XRD and PALS results. Instead, the very local technique (PAC) points out the existence of metamict residues in all the samples. [13] M. Colombo, J. Chrosch, and E. K. H. Salje, “Annealing meta- mict zircon: a powder X-ray diﬀraction study of a highly defect- ive phase,” Journal of the American Ceramic Society, vol. 82, no. 10, pp. 2711–2716, 1999. [14] Z. Mursic, T. Vogt, and F. Frey, “High-temperature neutron powder diﬀraction study of ZrSiO4 up to 1900 K,” Acta Crys- tallographica Section B, vol. 48, no. 5, pp. 584–590, 1992. 4. Conclusions In this paper long, medium, and short range studies (X-Ray Diﬀraction (XRD), Positron Annihilation Lifetime Spectro- scopy (PALS), and Perturbed Angular Correlations, (PAC)) were successfully applied to characterize the zirconium min- erals originated from Australia, South Africa, and United States of America. It was established that the metamictization 9 References 947–952, 2000. [28] P. Montes-Navajas, L. C. Damonte, and H. Garc´ıa, “Positron annihilation lifetimes in cucurbiturils: evidence of internal inclusion of gold in CB[7],” ChemPhysChem, vol. 10, no. 5, pp. 812–816, 2009. [9] S. R´ıos, E. K. H. Salje, M. Zhang, and R. C. Ewing, “Amorphiza- tion in zircon: evidence for direct impact damage,” Journal of Physics Condensed Matter, vol. 12, no. 11, pp. 2401–2412, 2000. [29] P. Lacentre and M. C. Caracoche, “Te eﬀects of radiation damage and annealing on the microstructure of a natural zircon observed by TDPAC,” Radiation Effects and Defects in Solids, vol. 129, no. 3-4, pp. 181–191, 1994. [10] F. Poitrasson, J. M. Hanchar, and U. Schaltegger, “Te current state and future of accessory mineral research,” Chemical Geology, vol. 191, no. 1-3, pp. 3–24, 2002. [30] M. C. Caracoche, J. A. Mart´ınez, A. F. Pasquevich, P. C. Rivas, E. Djurado, and F. Boulc’h, “PAC characterization of Gd and Y doped nanostructured zirconia solid solutions,” Physica B: Condensed Matter, vol. 389, no. 1, pp. 98–102, 2007. [11] A. Kaiser, M. Lobert, and R. Telle, “Termal stability of zircon (ZrSiO4),” Journal of the European Ceramic Society, vol. 28, no. 11, pp. 2199–2211, 2008. [31] D. Wiarda, M. Uhrmacher, A. Bartos, and K. P. Lieb, “Electric ﬁeld gradients at 111Cd in binary oxides,” Journal of Physics: Condensed Matter, vol. 5, no. 25, article 002, pp. 4111–4124, 1993. [12] W. J. Weber, R. C. Ewing, and L.-M. Wang, “Te radiation- induced crystalline-to-amorphous transition in zircon,” Journal of Materials Research, vol. 9, no. 3, pp. 688–698, 1994. References Submit your manuscripts at https://www.hindawi.com Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 High Energy Physics Advances in The Scientific World Journal Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Fluids Journal of Atomic and Molecular Physics Journal of Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Advances in Condensed Matter Physics Optics International Journal of Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Astronomy Advances in International Journal of Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Superconductivity Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Statistical Mechanics International Journal of Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Gravity Journal of Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Astrophysics Journal of Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Physics Research International Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Solid State Physics Journal of Computational Methods in Physics Journal of Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Soft Matter Journal of Hindawi Publishing Corporation http://www.hindawi.com Aerodynamics Journal of Volume 2014 Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Photonics Journal of Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Journal of Biophysics Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Thermodynamics Journal of Submit your manuscripts at https://www.hindawi.com The Scientific World Journal Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Fluids Journal of Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Advances in Condensed Matter Physics Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Gravity Journal of Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Soft Matter Journal of Hindawi Publishing Corporation http://www.hindawi.com Aerodynamics Journal of Volume 2014 Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Photonics Journal of The Scientific World Journal Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Hind http A C Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Gravity Journal of Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Soft Matter Journal of Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Photonics Journal of The Scientific World Journal Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Advances in Condensed Matter Physics Volume 2014 Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Soft Matter Journal of Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Photonics Journal of The Scientific World Journal Hindawi Publishing Corporation Hindawi Publishing Corporation http://www.hindawi.com Aerodynamics Journal of Volume 2014 Optics International Journal of Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Optics International Journal of Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Submit your manuscripts at https://www.hindawi.com Submit your manuscripts at https://www.hindawi.com Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 High Energy Physics Advances in Atomic and Molecular Physics Journal of Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Astronomy Advances in International Journal of Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Superconductivity Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Astrophysics Journal of Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Physics Research International Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Solid State Physics Journal of Computational Methods in Physics Journal of Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Journal of Biophysics Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Thermodynamics Journal of Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Astronomy Advances in Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 Solid State Physics Journal of
https://openalex.org/W2776235052	https://interactive-plus.ru/e-articles/358/Action358-465304.pdf	Russian	null	The analysis of international and Russian rating systems for assessing the performance of universities	Interaktivnaâ nauka	2,017	cc-by	2,638	Scientific Cooperation Center "Interactive plus" Scientific Cooperation Center "Interactive plus" УДК 69 DOI 10.21661/r-465304 I.A. Shagabaev, I.P. Vasiliuk THE ANALYSIS OF INTERNATIONAL AND RUSSIAN RATING SYSTEMS FOR ASSESSING THE PERFORMANCE OF UNIVERSITIES И.А. Шагабаев, И.П. Василюк АНАЛИЗ МЕЖДУНАРОДНЫХ И РОССИЙСКИХ РЕЙТИНГОВЫХ СИСТЕМ ОЦЕНКИ ДЕЯТЕЛЬНОСТИ УНИВЕРСИТЕТОВ Аннотация: работа посвящена анализу международных и российских рей- тинговых систем, используемых для оценки деятельности университетов раз- личных стран мира. Приведены примеры рейтинговых оценок ряда ведущих ву- зов России. Выведены показатели для российского университета дружбы наро- дов. Ключевые слова: международные рейтинговые системы, ведущие россий- ские университеты, рейтинговые оценки. 2 https://interactive-plus.ru THE ANALYSIS OF INTERNATIONAL AND RUSSIAN RATING SYSTEMS FOR ASSESSING THE PERFORMANCE OF UNIVERSITIES Abstract: the study is dedicated to the analysis of International and Russian rat- ing systems for assessing the performance of Universities of different World countries. The examples of rating scores of some of the Russian Leading Universities are given here. The indicators for Peoples' Friendship University of Russia (RUDN/PFUR) have also been deducted. Keywords: international rating systems, leading Russian Universities, rating scores. В последнее десятилетие образовательные организации высшего образова- ния стали значительно большее внимание уделять рейтинговым оценкам своей деятельности. Лидирующие позиции в мировых и отечественных рейтингах поз- воляют вузам привлекать одаренных студентов, высокопрофессиональных пре- подавателей и опытных ученых. Высокие рейтинговые оценки вузов в автори- тетных международных рейтинговых системах являются в настоящее время Content is licensed under the Creative Commons Attribution 4.0 license (CC-BY 4.0) Центр научного сотрудничества «Интерактив плюс» главными показателями конкурентоспособности и качества системы высшего образования страны, что помогает привлекать дополнительные источники инве- стирования. главными показателями конкурентоспособности и качества системы высшего образования страны, что помогает привлекать дополнительные источники инве- стирования. Приведем примеры наиболее важных современных международных рейтин- говых систем и критериев оценки деятельности университетов [8; 9]. 1) Академический рейтинг университетов Мира (Academic Ranking of World Universities – ARWU). Критерии оценки: 1. Число выпускников-лауреатов Нобелевской или Филдсовской премии (Alumni) – 10% 2. Число сотрудников-лауреатов Нобелевской или Филдсовской премии (Award) – 20% 3. Число наиболее часто цитируемых исследователей в различных предмет- ных областях (HiCi) – 20% 4. Число статей, опубликованных в журналах Nature и Science (N&S) – 20% 4. Число статей, опубликованных в журналах Nature и Science (N&S) – 20% 5. Число статей, проиндексированных в ScienceCitationIndex – Expanded и 5. Число статей, проиндексированных в ScienceCitationIndex – Expanded и SocialSciencesCitationIndex (PUB) – 20% 6. Академическая производительность на одного представителя научно- преподавательского состава вуза – результат деления суммы баллов по предыду- щим пяти показателям на число эквивалентов полной ставки академического персонала (PCP) – 10% 7. Отраслевые и предметные оценки 7. Отраслевые и предметные оценки 2) рейтинги мировых университетов компании QS (QS World University Rankings) Rankings) Критерии оценки: 1. Академическая репутация (40%) – производится на основе глобального опроса академического сообщества. 1. Академическая репутация (40%) – производится на основе глобального опроса академического сообщества. 2. Репутация среди работодателей, преподавательского состава и студентов (20%). 3. Доля иностранцев в научно-преподавательском составе (5%). ttps://interactive-plus.ru Содержимое доступно по лицензии Creative Commons Attribution 4.0 license (CC-BY 4.0) p p Содержимое доступно по лицензии Creative Commons Attribution 4.0 license (CC-BY 4.0) Scientific Cooperation Center "Interactive plus" 4. THE ANALYSIS OF INTERNATIONAL AND RUSSIAN RATING SYSTEMS FOR ASSESSING THE PERFORMANCE OF UNIVERSITIES Доля иностранцев в общем числе студентов (5%). 4. Доля иностранцев в общем числе студентов (5%). 5. Число цитирований на одного сотрудника (20%). 5. Число цитирований на одного сотрудника (20%). 6. По отраслям. 6. По отраслям. 7. По специальностям компании(мнение академического сообщества, мне- ние работодателей, индекс цитирования и индекс Хирша). 7. По специальностям компании(мнение академического сообщества, мне- ние работодателей, индекс цитирования и индекс Хирша). 8. Рейтинг университетов стран БРИКС(Академическая репутация (30%); Репутация среди работодателей (20%); Соотношение числа студентов и числа НПС (20%); Доля НПС с ученой степенью (10%); Количество опубликованных статей, приходящихся на одного НПС (10%); Количество цитат, приходящихся на одну опубликованную статью (5%); Доля НПС-иностранцев в общем числе НПС(2,5%); Доля студентов-иностранцев в общем числе студентов (2,5%)). 9. Рейтинг университетов развивающихся стран Европы и Центральной Азии(Академическая репутация (30%); Репутация среди работодателей (20%); Соотношение численности НПР и студентов (15%); Web-влияние (10%) (резуль- таты Webometrics); Число статей на одного НПС (10%); Доля НПР, обладающих ученой степенью в общей численности НПР (5%); Среднее число цитирований на одну статью (5%); Доля иностранных НПР в общем числе НПС(2.5%); Доля студентов-иностранцев в общем числе студентов (2.5%)). 10. Рейтинг молодых университетов(по критериям рейтинга мировых уни- верситетов). 11. Рейтинг силы систем высшего образования(Сила системы, доступность, ведущий университет, экономические условия). Определим позиции российских вузов в международных рейтинговых си- стемах, описанных выше: Определим позиции российских вузов в международных рейтинговых си- стемах, описанных выше: 3 Content is licensed under the Creative Commons Attribution 4.0 license (CC-BY 4.0) 1. Академический рейтинг университетов Мира (Academic Ranking of World Universities – ARWU) Таблица 1 Академический рейтинг университетов Мира Наименование учреждения Позиция в рейтинге 1. МГУ им. М.В. Ломоносова 86 1. Академический рейтинг университетов Мира (Academic Ranking of World Universities – ARWU) Центр научного сотрудничества «Интерактив плюс» Центр научного сотрудничества «Интерактив плюс» 2. СПбГУ группа 301–400 3. МФТИ группа 501–600 4. НГУ группа 501–600 5. СПбГТИ группа 901–1000 6. УФУ группа 901–1000 7. КФУ группа 901–1000 2) рейтинги мировых университетов компании QS (QS World University Rankings). Таблица 2 Рейтинги компании QS Наименование учреждения Рейтинг Позиция в рейтинге МГУ им. М.В. Ломоносова Топ-200 лучших университетов Мира 108 МГУ им. М.В. Ломоносова Естественные науки 60 МГТУ им. Баумана Страны БРИКС 35 МГИМО Страны БРИКС 39 МФТИ Страны БРИКС 45 МИФИ Страны БРИКС 51 СПбПУ Страны БРИКС 60 ВШЭ Страны БРИКС 63 ТПУ Страны БРИКС 64 К(П)ФУ Страны БРИКС 72 УрФУ Страны БРИКС 77 ЮФУ Страны БРИКС 81 РУДН Страны БРИКС 84 ННГУ им. THE ANALYSIS OF INTERNATIONAL AND RUSSIAN RATING SYSTEMS FOR ASSESSING THE PERFORMANCE OF UNIVERSITIES Лобачевского Страны БРИКС 86 МИСиС Страны БРИКС 89 РЭУ им. Плеханова Страны БРИКС 91 ДВФУ Страны БРИКС 98 СПбГУИТМО Страны БРИКС 99 МГУ Развивающиеся страны Европы 1 НГУ Развивающиеся страны Европы 2 2. СПбГУ группа 301–400 3. МФТИ группа 501–600 4. НГУ группа 501–600 5. СПбГТИ группа 901–1000 6. УФУ группа 901–1000 7. КФУ группа 901–1000 Таблица 2 Рейтинги компании QS Наименование учреждения Рейтинг Позиция в рейтинге МГУ им. М.В. Ломоносова Топ-200 лучших университетов Мира 108 МГУ им. М.В. Ломоносова Естественные науки 60 МГТУ им. Баумана Страны БРИКС 35 МГИМО Страны БРИКС 39 МФТИ Страны БРИКС 45 МИФИ Страны БРИКС 51 СПбПУ Страны БРИКС 60 ВШЭ Страны БРИКС 63 ТПУ Страны БРИКС 64 К(П)ФУ Страны БРИКС 72 УрФУ Страны БРИКС 77 ЮФУ Страны БРИКС 81 РУДН Страны БРИКС 84 ННГУ им. Лобачевского Страны БРИКС 86 МИСиС Страны БРИКС 89 РЭУ им. Плеханова Страны БРИКС 91 ДВФУ Страны БРИКС 98 СПбГУИТМО Страны БРИКС 99 МГУ Развивающиеся страны Европы 1 НГУ Развивающиеся страны Европы 2 ВШЭ Молодые университеты группа 81–90 Оценим современное положение в рейтингах одного из ведущих отече- ственных вузов – Российского университета дружбы народов (РУДН). Оценим современное положение в рейтингах одного из ведущих отече- ственных вузов – Российского университета дружбы народов (РУДН). Scientific Cooperation Center "Interactive plus" Content is licensed under the Creative Commons Attribution 4.0 license (CC-BY 4.0) РУДН в международных рейтингах тояние на данный момент: Состояние на данный момент: 1. Первый российский вуз, удостоенный оценки «5 звезд» по 5 категориям международного рейтинга QS Stars. 1. Первый российский вуз, удостоенный оценки «5 звезд» по 5 категориям международного рейтинга QS Stars. 2. РУДН в мировом рейтинге THE. 3. RUDN University в группе 501–550 в рейтинге QS World University Rank- ings 2018. 4. РУДН – в ТОП-100 мирового рейтинга THE. 4. РУДН – в ТОП-100 мирового рейтинга THE. 5. РУДН в рейтинге лучших вузов Европы Times Higher Education. 6. РУДН в ТОП-20 лучших вузов по версии первого национального рей- тинга вузов «Три миссии университетов». 7. Российский университет дружбы народов в рейтинге UI GreenMetric World University Ranking – «зеленый» рейтинг вузов мира, публикуемый универ- ситетом Индонезии. 8. РУДН занимает 5 место в рейтинге «Интерфакс». 8. РУДН занимает 5 место в рейтинге «Интерфакс». 9. РУДН – в ТОП-100 из более 7000 университетов стран БРИКС (Бразилия, Россия, Индия, Китай, Южная Африка) в мировом рейтинге QS. 10. В глобальном рейтинге Webometrics РУДН занимает 18 место среди уни- верситетов России. 11. В ежегодном Национальном рейтинге российских вузов РУДН занимает 5–6 место. 12. РУДН в топ-10 Национального рейтинга университетов 2016. 12. РУДН в топ-10 Национального рейтинга университетов 2016. 13. РУДН занимает 15 место в Национальном рейтинге университетов по параметру «Исследования». 13. РУДН занимает 15 место в Национальном рейтинге университетов по параметру «Исследования». 14. РУДН в топ-200 лучших вузов мира и третий среди российских универ- ситетов по качеству преподавания в рейтинге Round University Ranking (RUR). 15. 456 место в Международном рейтинге университетов Round University Ranking (RUR). Заметим, что актуальной является задача повышения международной рей- тинговой оценки РУДН, а стратегической целью – выполнение проекта 5–100, в 5 5 Content is licensed under the Creative Commons Attribution 4.0 license (CC-BY 4.0) Центр научного сотрудничества «Интерактив плюс» соответствии с Постановлением Правительства РФ 16 марта 2013 г. №211 «О ме- рах государственной поддержки ведущих университетов Российской Федерации в целях повышения их конкурентоспособности среди ведущих мировых научно- образовательных центров», которым утвержден план мероприятий по повыше- нию глобальной конкурентоспособности ведущих российских университетов среди ведущих мировых научно-образовательных центров (проект 5–100) [1]. Основным инструментом и основой повышения конкурентоспособности РУДН является модернизация информационной инфраструктуры университета. Целевое состояние ИТ-инфраструктуры к 2020 году включает следующие со- ставляющие: 1. ttps://interactive-plus.ru Содержимое доступно по лицензии Creative Commons Attribution 4.0 license (CC-BY 4.0) 6 ttps:// te act ve p us. u Содержимое доступно по лицензии Creative Commons Attribution 4.0 license (CC-BY 4.0) РУДН в международных рейтингах Высокопроизводительная коммуникационная среда, обеспечивающая ре- шение задач управления, организацию как трансграничного, так и внутрикорпо- ративного электронного обучения на базе многофункциональной системы элек- тронного обучения и интернет-портала университета, коммуникационную среду потребителей услуг университета, а также повышение узнаваемости и признания университета на международном уровне [4]. 1. Высокопроизводительная коммуникационная среда, обеспечивающая ре- шение задач управления, организацию как трансграничного, так и внутрикорпо- ративного электронного обучения на базе многофункциональной системы элек- тронного обучения и интернет-портала университета, коммуникационную среду потребителей услуг университета, а также повышение узнаваемости и признания университета на международном уровне [4]. 2. Интеграция информационных систем университета на базе собственной единой информационной платформы с целью повышения их производительно- сти, обеспечения достоверности данных и формирования на их основе системы непрерывного систематического наблюдения (мониторинга) за основными про- цессами и показателями деятельности университета [5; 6]. 2. Интеграция информационных систем университета на базе собственной единой информационной платформы с целью повышения их производительно- сти, обеспечения достоверности данных и формирования на их основе системы непрерывного систематического наблюдения (мониторинга) за основными про- цессами и показателями деятельности университета [5; 6]. 3. Многофункциональная система электронного обучения (МСЭО), обеспе- чивающая реализацию как основных профессиональных образовательных про- грамм, так и программ дополнительного образования всех видов [7]. 4. Расширение пропускной способности внутрикорпоративной локально- вычислительной сети университета до скорости не менее 10 Гбит/с. Создание бесшовной высокоскоростной беспроводной сети с увеличением зоны покрытия во всех учебных и административных корпусах Университета, а также на терри- тории общежитий. Подключение к международному роуминговому сервису WiFi eduroam (education roaming) [2; 3]. Scientific Cooperation Center "Interactive plus" 5. Запуск и ввод в эксплуатацию центра обработки данных (ЦОД). На базе ЦОДа создается высокопроизводительный кластер (суперкомпьютер), который позволит производить параллельные вычисления и расчеты, а также обрабаты- вать и хранить большие данные. 6. Создание собственной облачной инфраструктуры (private cloud) с высо- кой степенью отказоустойчивости, безопасности и масштабируемости, обеспе- чивающей централизованный и высокоскоростной доступ к различным элек- тронным ресурсам университета студентам, преподавателям и сотрудникам. 6. Создание собственной облачной инфраструктуры (private cloud) с высо- кой степенью отказоустойчивости, безопасности и масштабируемости, обеспе- чивающей централизованный и высокоскоростной доступ к различным элек- тронным ресурсам университета студентам, преподавателям и сотрудникам. 7. Сотрудничество университета с различными компаниями – производите- лями вычислительного оборудования и программного обеспечения. 7. Сотрудничество университета с различными компаниями – производите- лями вычислительного оборудования и программного обеспечения. 8. Разработка приложений под современные мобильные платформы iOS, Android и Windows Phone. 8. Разработка приложений под современные мобильные платформы iOS, Android и Windows Phone. РУДН в международных рейтингах В результате реализации проектов по модернизации и развитию информа- ционной инфраструктуры в указанных областях университет должен получить к 2020 году следующие преимущества:  переход на новую систему организации учебного процесса – From Teaching to Learning;  обеспечение мобильности преподавателей и студентов за счет удаленного доступа к электронным ресурсам и образовательным программам;  широкое привлечение иностранных обучающихся к дистанционным обра- зовательным программам, в том числе к программам повышения квалификации и профессиональной переподготовки;  снижение непроизводственных затрат времени сотрудников. Content is licensed under the Creative Commons Attribution 4.0 license (CC-BY 4.0) Центр научного сотрудничества «Интерактив плюс» 2. Горелов Г.В., Ромашкова О.Н. Оценка качества обслуживания в сетях с пакетной передачей речи и данных // Вестник Российского университета дружбы народов. Серия: Прикладная и компьютерная математика. – 2003. – Т. 2. – №1. – С. 23–31. 2. Горелов Г.В., Ромашкова О.Н. Оценка качества обслуживания в сетях с пакетной передачей речи и данных // Вестник Российского университета дружбы народов. Серия: Прикладная и компьютерная математика. – 2003. – Т. 2. – №1. – С. 23–31. 3. Лукова О.Н. Анализ качества стохастической цифровой передачи рече- вой информации (методика и ее использование при разработке информационных систем) // Диссертация на соискание ученой степени кандидата технических наук. – М.: МИИТ, 1994. – 149 с. 3. Лукова О.Н. Анализ качества стохастической цифровой передачи рече- вой информации (методика и ее использование при разработке информационных систем) // Диссертация на соискание ученой степени кандидата технических наук. – М.: МИИТ, 1994. – 149 с. 3. Лукова О.Н. Анализ качества стохастической цифровой передачи рече- вой информации (методика и ее использование при разработке информационных систем) // Диссертация на соискание ученой степени кандидата технических наук. – М.: МИИТ, 1994. – 149 с. 4. Prokhorov E.I., Ponomareva L.A., Permyakov E.A., Kumskov M.I. Fuzzy Classification and Fast Rules for Refusal in the QSAR Problem // Pattern Recognition and Image Analysis (Advances in Mathematical Theory and Applications). – 2011. – Т. 21. – №3. – С. 542–544. 5. Ермакова Т.Н. Повышение эффективности управления информацион- ными потоками в образовательном комплексе / Т.Н. Ермакова, О.Н. Ромашкова // Вестник РГРТУ. – 2016. – №57. – С. 82–87. 5. Ермакова Т.Н. Повышение эффективности управления информацион- ными потоками в образовательном комплексе / Т.Н. Ермакова, О.Н. Ромашкова // Вестник РГРТУ. – 2016. – №57. – С. 82–87. 6. Ромашкова О.Н. Алгоритм работы смодулем «УЧЕБНАЯ ДЕЯТЕЛЬ- НОСТЬ» управленческой информационной системы для образовательного ком- плекса / О.Н. Ромашкова, Т.Н. Ермакова // Исследование различных направле- ний современной науки. VIII Международная научно-практическая конферен- ция. – 2016. – С. 917–924. 7. Кунтикова Е.С. Проблемы автоматизации учета инновационной деятель- ности в образовательном учреждении / Е.С. Кунтикова, С.В. Чискидов, Е.Н. Пав- личева // Информационные ресурсы России. – 2014. – №3 (139). – С. 25–29. 7. Кунтикова Е.С. Проблемы автоматизации учета инновационной деятель- ности в образовательном учреждении / Е.С. Кунтикова, С.В. Чискидов, Е.Н. Пав- личева // Информационные ресурсы России. – 2014. – №3 (139). – С. 25–29. 8. Балацкий Е.В. Рейтинги национальных систем высшего образования / Е.В. Балацкий, А.Б. Гусев // Мир измерений. – 2008. – №4 (86). 9. Список литературы 1. Утверждение плана мероприятий по развитию ведущих университетов, предусматривающих повышение их конкурентоспособности среди ведущих ми- ровых научно-образовательных центров от 26 октября 2012 г. Распоряжение Правительства РФ N2006-р [Текст] / Собрание законодательства РФ. – 2012. – 05 ноября. – СТ. 6288. 7 Content is licensed under the Creative Commons Attribution 4.0 license (CC-BY 4.0) Центр научного сотрудничества «Интерактив плюс» 8 https://interactive-plus.ru 8 https://interactive-plus.ru Содержимое доступно по лицензии Creative Commons Attribution 4.0 license (CC-BY 4.0) ttps://interactive-plus.ru Содержимое доступно по лицензии Creative Commons Attribution 4.0 license (CC-BY 4.0) Content is licensed under the Creative Commons Attribution 4.0 license (CC-BY 4.0) Центр научного сотрудничества «Интерактив плюс» Bates T. (2010). World University Rankings: a Reality Based on a Fraud // E- learning&Distance Education. 17.09.2010 [Электронный ресурс]. – Режим доступа: http://www.tonybates.ca/2010/09/17/world-university-rankings-a-realitybased-on-a- fraud/, свободный. Scientific Cooperation Center "Interactive plus" Шагабаев Илья Арсенович – магистрант ФГБОУ ВО «Московский педа- гогический государственный университет», Россия, Москва. Shagabaev Ilia Arsenovich – graduate student at Moscow State University of Education, Russia, Moscow. Василюк Игорь Петрович – соискатель ФГБОУ ВО «Московский педаго- гический государственный университет», Россия, Москва. Vasiliuk Igor Petrovich – applicant at Moscow State University of Education, Russia, Moscow. 9 Content is licensed under the Creative Commons Attribution 4.0 license (CC-BY 4.0)
https://openalex.org/W1989145185	https://zenodo.org/records/2423227/files/article.pdf	English	null	MR. DOOLEY, 2D, ON THE DISCUSSION METHOD	School science and mathematics	1,918	public-domain	2,098	MR. DOOLEY, 2D, ON THE DISCUSSION METHOD. (With Apologies to the Original Mr. Dooley.) BY JEAN BROADHUBST, Teachers College, Columbia University. MR. DOOLEY, 2D, ON THE DISCUSSION METHOD. (With Apologies to the Original Mr. Dooley.) BY JEAN BROADHUBST, Teachers College, Columbia University. BY JEAN BROADHUBST, Teachers College, Columbia University. Ye’ see, Michael, me b’y, since me fayther settled so manny questions av internashunal dispute, it cum to me natshral loike to be inthrusted in the prisint day and its problims. And me posishun av windy washin’ at collige is furtherin’ me in the way av it, for windy washin’ is a quiet job and slow, as is well known. So I hear manny a thing that me puir fayther missed altogither. And the quarest av thim all is the way thim big perfessors is a tryin’ to git shut av their jobs, entoirely. Do ye iver hear me, Michael, a sayin’ that the windys don’t need washin’ or cud wash thimsilves? Be gorrah, a man ’ud be two fules in wan, he wud, to do thatfor he’d sure lose his job altogither and the windys wudn’t be washed at all, at all. But this discussin method is a quare one. As I kin make it out, it’s a way the perfessors have av lettin’ the scholars run on by thimsilves loike a horse widout anny driver at all. No, it’s not loike Murphy it is, fur his old white horse stops at ivry house and waits till Murphy gives the baby his bit av pasteurized milk for the day, and comes out wid the empties. That’s not loike the discussin way at all, for old Murphy knows where his horse is ivry minute and where he’s goin’, and what’s more, the old horse fetches up at the right places ivry time. But in the discussin way, a horse ud av to av the wings av a bird and legs av a kangrue, for he’d av to jump from one street to anuther and skip whole blocks intoirely; and to kape up wid him Murphy wud av need av an aeroplane wid all the accelirators and patent brakes tha<t av yit to be invinted. Besides, Murphy’s horse niver goes backward, me b’y. He’s a wise old horse, and he’d know it av he did. Now, as I kin make it out, in the discussin way the per- fessor starts ’em on a little way wid his highbrow stuff, and thin he drops the lines intoirely. MR. DOOLEY ON THE DISCUSSION METHOD MR. DOOLEY ON THE DISCUSSION METHOD 681 MR. DOOLEY, 2D, ON THE DISCUSSION METHOD. (With Apologies to the Original Mr. Dooley.) BY JEAN BROADHUBST, Teachers College, Columbia University. And ivry horse pulls a dif- ferent way, zigzaggin’ around as if the royal road to learnin’ was the Abbey Inn hill for the steepness av it. And thin, whin thim as don’t discuss, the passingers, so to speak, have been yanked this way and that till they don’t know which way they’re goin’, sumthin’ breaks, and the hull thing stops, wid each av thim discussers in a diffrunt place; and the passingers begin wanderin’ about aimless-like, some av thim SCHOOL SCIENCE AND MATHEMATICS 682 evin goin5 back where they started at the fut av the hill and not knowin’ it at all, and thinkin’ they’re goin’ ahead, bedad. It’s all true. They’re befuddled as no whisky cud do it in so short a time evin in this prohibishun town. But the per- fessor, ye say. Well, it depinds on the perfessor. There’s nothin’ harder to ketch than a horse broke loose, but some av thim perfessors kin do it. Some av thim let the discussers tire thim- selves out, and some av thim ketches and houlds thim by feeding a scatterin^ the grain around. A handful av wheat goes a long way, if you mix it wid chaff, me b’y. But some aV thim per- fessors’ll be caught by the pure food laws yit. There’s only wan thing that’s saved thim so farthat their stuff ain’t done in packidges! There’s no label to say whales chaff, but the adul- tration’s somethin grand. IVd make anyone av thim rich, if it was gasoline or evin sugar he was by the way av dealin’ in. ; No, it’s a quare way to me way av thinking The perfessor starts ’em off a discussin’ by telling ’em a little av what he knows, the straight stuff, and thin the discussers break in to tell about what they don’t know about. They actually git up\and say, ^Pve niver had any experience in this field, but some ye^rs ago, I"would ye believe it! And often and often agin they say, "I am unable to offer anything in the direct line av this discussin," and thin talk for twinty minutes! ’ But I’m thinkin’, some av the perfessors av caught onto the way to stop it. MR. DOOLEY, 2D, ON THE DISCUSSION METHOD. (With Apologies to the Original Mr. Dooley.) BY JEAN BROADHUBST, Teachers College, Columbia University. For some av thim begin, "This mornin’ we will discuss the educashun av the child from the standpoint av-" ye can’t expect me to remimber the rist av itand thin talk thimsilves till the bell rings! They’re the wise guys, they be, for they git sumthin’ done. Thin iverbody’s more satisfied, fur the classes thinks they’ve been discussing which flatters the edicated mind, and they’ve learned sumthin’, too, which ain’t so common in the rale discussin way. And the worst av it is it’s a spilin’ the schools, too. Me little Jimmie is a bein’ learned the same way. The taycher starts out to learn thim that Connecticut raises tobaccy and the kids wid opporchunities or imaginashuns get up and say, ^When I was to me grannie’s in Maryland I seen fields and fields of to- baccy"; and "Me uncle owns two tobaccy farms in Kentucky"; and "When we go on our vacashun in our motor car, maybe we’ll go to Connecticut and see thim tobaccy fields." Why can’t they "believe the painter" whin the taycber tells thim it grows in Connecticut widout all this introducshun? And MR. DOOLEY ON THE DISCUSSION METHOD 683 they discuss, too, just like the grown ups, wid the taycher lead- in7 thim on, wid ^Why is Connecticut a tobacco state?" And they guess all the raysons they can think ofthe ones they had for corn and watches and firearms and shoes till somebody strikes the one the taycher knows about. And thin they start over agin a keepin7 the smart Alecky ones a talkin7 all the time; and the kids don7! hear no good grammar day in an7 out wid the glib ones a gettin7 glibber and the slow ones overworked sortin7 out which is talk and which is learnin7. And all this discussin7 and spreadin7 thimsilves out thin spoils thim fur rale work, Pm thinkin7. There^ some classes in collige the discussers niver takeor if they do, they drop it afore the day av reckonin7, the examinashun, so to speak. Why, there was wan wuman discusserthere^ comparative few faymale discussers, but some av thim/s as bad as the men who got into a regular class ^tead av a discussin one. She^ been fed up on discussin where Ws ^all a matter av opinion any- how,77 as one perfesser always says giib-like when the discussers get too hot and nobudy7!! MR. DOOLEY, 2D, ON THE DISCUSSION METHOD. (With Apologies to the Original Mr. Dooley.) BY JEAN BROADHUBST, Teachers College, Columbia University. give an inchand so she^ cum to where she cudn7! settle down to learn annythin7 for hersilf or by hersilf, as Abraham Lincoln said. And wan day, whin she^ been partic^ar stoopid, she jest pushed her glasses and spiders back, and said, lofty-like, <c! don7! approve av this way av studyin7 at all. I don7! want to see that a spidery got six legs. In Perfessor Blacky class we’d discuss it.77 But the bug per- fessor just said, ^If the Lord^ kindly left you the wan sinse of eyesight, that^ wan av the manny things you can’t discuss wid his eight legs stickin7 right out in plain sight.77 Manny^ the time a discusser has interrupted a flow av rale knowludge, that^d make ye thrill, evin wid your feet the other side av the windy in the ice and snow, and Pll hear the pas- singers, as we called thim, heave a sigh as the same old dis- cussers git up and say nuthin7 at all. It’s always the same dis- cussers, about five or six to a class, maybe. And if they av to git what they don7! know out av their system, you^d think they^ put thim discussers all in a class by thimsilves, where they cud waste their own time and let the rist av the class go on widout all this zigzaggin7 round. But the perfessor is so busy holdin7 fast to his lead-horse lines that he don7! hear thim sighin7 on the back seats. I expect the perfessors cum from the city and don7! know abou!t horses as we do, me Vy. For i! ain7! the prancin7 leaders that really pulls the tally-ho, is it? And for all their SCHOOL SCIENCE AND MATHEMATICS 684 cavorting the coach keeps to the same old road just the same. So their leading evin, ain^t much to brag av. And did ye iver see a tally-ho team doin’ anny rale workand neither will you, me b^y. All ye kin say is they^re havin7 a good toime av it, but iVs the hefty Percheruns ye see t7 the coal wagons. Some day the remarks av thim back seats^ll git louder. They^re a gittin7 perty well bumped around, thim passingers, and they don^t loike it. MR. DOOLEY, 2D, ON THE DISCUSSION METHOD. (With Apologies to the Original Mr. Dooley.) BY JEAN BROADHUBST, Teachers College, Columbia University. And the day av reckonin7 is comin^ whin they^ll rise up and demand their rights, and we’ll live to see it, be gor- rah, fur it’s cummin7 soon. Who knows but the Collige’ll have guards to kape out thim discussers. There’ll be me opporchunity, for I kin spot a dis- cusser ivery toime. Fur what I’ve suffered from some av thim discussers, evin on the windy sills, which is remote, so to speak, wud make me the best bouncer on the force, and widout anny eye to the Carnegie medals, either. And it’ll cum, for this discussion way ain’t what some av these perfessors think it is. For it’s only a few that do the dis- cussin’ and whin they start, the majority begins a cussin\ And it’s the majority that rules in this free country av America, praise be, except in prisidintial elections, av course. RAINFALL AND GUNFIRE. It is, of course, impossible to ascertain definitely whether bombard- ment can produce rain or not, because one cannot determine in any spe- cific case what the rainfall would have been in the absence of bombard- ment. But there is a very widespread notion that gunfire may exercise an influence on the weather and there is always some one to seize upon any opportunity to show a possible connection between them. Since the experiments of C. T. R. Wilson showed that ions may act as nuclei for the condensation of water vapor, a new explanation has been put forward by several persons who claim that condensation may occur as a result of ionization of the air caused by bombardment. A As a matter of fact, Wilson has shown that in order for the ions to be effective in producing condensation a "fourfold" degree of supersatura- tion is required for the negative ions, and a considerably higher degree for the positive. Others have shown that supersaturation is impossible so long as there are any particles of dust in the air. Dust is always found in the air, even in samples taken from the tops of high mountains far from towns. Consequently, we cannot suppose that ions are the immediate cause of any condensation produced by bombardment. No rational account of any complete process by which gunfire could affect the weather has ever been put forward. Nor does the statistical evidence that is available support the proposition. It shows, rather, that the condition of the weather is a factor in determining the time of battles. An examination of the rainfall in France since the commencement of the war shows no outstanding features which could not be accounted for by the pressure distribution existing at the times of the individual falls.
https://openalex.org/W4244638840	https://www.ccsenet.org/journal/index.php/mas/article/download/68078/36923	English	null	Reviewer Acknowledgements for Modern Applied Science, Vol. 10, No. 10	Modern applied science	2,016	cc-by	371	Modern Applied Science; Vol. 10, No. 10; 2016 ISSN 1913-1844 E-ISSN 1913-1852 Published by Canadian Center of Science and Education Modern Applied Science; Vol. 10, No. 10; 2016 ISSN 1913-1844 E-ISSN 1913-1852 Published by Canadian Center of Science and Education Modern Applied Science; Vol. 10, No. 10; 2016 ISSN 1913-1844 E-ISSN 1913-1852 Published by Canadian Center of Science and Education Reviewer Acknowledgements Modern Applied Science wishes to acknowledge the following individuals for their assistance with peer review of manuscripts for this issue. Their help and contributions in maintaining the quality of the journal are greatly appreciated. Modern Applied Science is recruiting reviewers for the journal. If you are interested in becoming a reviewer, we welcome you to join us. Please find the application form and details at http://recruitment.ccsenet.org and e-mail the completed application form to mas@ccsenet.org. Reviewers for Volume 10, Number 10 , Antonio Camarena-Ibarrola, Electrical Engineering School of UMSNH, Mexico Assad Omar, Al-Batra University, United Kingdom Chanchai Chitlaoarporn, Rangsit University, Thailand Christoph Maria Ravesloot, Rotterdam University of Applied Science, The Netherlands Ebrahim Moradi, University of Sistan & Baluchestan, Iran Ganesan P., Sathyabama University, India Hassan Boland, University of Guilan, Iran Huda Samee, Zarqa Private University, Jordan Jamal Kheiri, University of Tehran, Iran Janar Jarasovna, L.N.Gumilyov Eurasian National University, Kazakhstan Jill Smith, University of York, United Kingdom JIN Lahua, Jinan University, China Kamal Gholipour, Tabriz University of Medical Sciences, Iran Liang Yu, Washington State University, China Lizziane Winkelstroter, University of São Paulo, Brazil Mansoor Maitah, The Czech University of Life Sciences Prague, Czech Republic Marek Brabec, Academy of Sciences of the Czech republic, Czech Republic Miloš Hitka, Technical University in Zvolen, Slovakia Miroslaw Majkut, Silesian University of Technology, Poland Mohammad Biria, Islamic Azad of Rasht, Iran Mohammad Mehdi Rashidi, Bu-Ali Sina University, Iran Mojtaba Abbasian, University of Chabahar, Iran Peng Zhang, State University of New York at Stony Brook, United States Praveen Kumar, Texas Tech University, United States Pri Priyono, Economics Faculty Universities PGRI Adi Buana, Indonesia Rostyslav Sklyar, Independent professional, Ukraine Sankaranarayanan Murugan, Sathyabama University, India Antonio Camarena-Ibarrola, Electrical Engineering School of UMSNH, Mexico Antonio Camarena-Ibarrola, Electrical Engineering School of UMSNH, Mexico Assad Omar, Al-Batra University, United Kingdom Assad Omar, Al-Batra University, United Kingdom Ganesan P., Sathyabama University, India 297
https://openalex.org/W2150041642	https://ijponline.biomedcentral.com/counter/pdf/10.1186/s13052-015-0120-z	English	null	Assessment of epicardial adipose tissue thickness and the mean platelet volume in children with familial Mediterranean fever	The Italian Journal of Pediatrics/Italian journal of pediatrics	2,015	cc-by	4,574	Uluca et al. Italian Journal of Pediatrics (2015) 41:15 DOI 10.1186/s13052-015-0120-z Uluca et al. Italian Journal of Pediatrics (2015) 41:15 DOI 10.1186/s13052-015-0120-z ITALIAN JOURNAL OF PEDIATRICS ITALIAN JOURNAL OF PEDIATRICS RESEARCH Open Access Abstract Background: Familial Mediterranean fever (FMF) is an inflammatory disease, which is suggested to be associated with increased risk of atherosclerosis. Epicardial adipose tissue (EAT) thickness and the mean platelet volume (MPV) are parameters used in prediction of atherosclerotic risk in various conditions. These parameters were evaluated in children with FMF and compared with healthy controls. Methods: Forty-five patients with FMF and 54 age- and gender-matched healthy controls were assessed. Duration of symptoms, age at diagnosis, duration of delay in diagnosis, frequency and duration of FMF attacks, disease severity scores, response to colchicine therapy, MEditerraneanFeVer (MEFV) gene mutations, and MPV values were recorded. EAT thicknesses were measured by echocardiography. Results: Epicardial adipose tissue thicknesses of the children with FMF were found to be significantly greater than that of controls (5.1 ± 1.4 vs. 4.5 ± 0.9 mm, p = 0.036). FMF patients had significantly higher MPV values compared with the controls (7.8 ± 1.1 vs. 7.3 ± 1.4 fl, p = 0.044). Age at diagnosis, duration of delay in diagnosis, and MPV values were found to be correlated with EAT thickness in the patient group (r = 0.49, p = 0.001 for the former parameters and r = 0.32, p = 0.04 for MPV). Conclusion: Epicardial adipose tissue thickness and MPV values seem to be increased in children with FMF. These findings may indicate an increased risk of atherosclerosis in FMF patients. Keywords: Familial Mediterranean fever, Epicardial adipose tissue thickness, Mean platelet volume, Atherosclerosis, Inflammation allowing evaluation of carotid intima media thickness and endothelial dysfunction found a predisposition of FMF patients to atherosclerosis [4,5]. Mean platelet vol- ume (MPV) and epicardial adipose tissue (EAT) thick- ness are novel parameters used in prediction of the risk of atherosclerosis [3,6-8]. Albeit contradictory results have been reported, there are some studies investigating the usefulness of MPV in assessment of atherosclerotic risk in cases with FMF [6,9,10]. However, no study evaluating EAT thicknesses of FMF patients was re- ported so far. Assessment of epicardial adipose tissue thickness and the mean platelet volume in children with familial Mediterranean fever Ünal Uluca1, Fikri Demir2, Aydın Ece3, Velat Şen1, Ali Güneş1, Fesih Aktar1, İlhan Tan1, Duran Karabel1, Ümitcan Yazgan4 and Muhammed Nurullah Sabaz1 © 2015 Uluca et al.; licensee BioMed Central. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Correspondence: ulucaunal@hotmail.com 1Department of Pediatrics, Dicle University Medical School, Diyarbakir, Turkey Full list of author information is available at the end of the article Measurement of epicardial adipose tissue thickness All h i h di hi i i Measurement of epicardial adipose tissue thickness All transthoracic echocardiographic examinations were performed by an experienced pediatric cardiologist (F.D.). A Vivid 5 Ultrasound System with appropriate trans- ducers (GE Medical Systems, Horten, Norway) was used for echocardiographic evaluations. The epicardial adipose tissue was seen as an echo-free space between the outer surface of myocardium and visceral pericar- dium on two-dimensional echocardiography, and its thickness was measured on the free wall of right ven- tricle at end-diastole in parasternal long-axis view [13]. Discussion Systemic inflammation is an important risk factor for ath- erosclerosis. Carotid intima media thickness, endothelial Table 1 The main characteristics of the patients and the control group FMF group Control group p (n = 45) (n = 54) Gender (M/F) 19 / 26 24 / 30 NS Age (years) 8.1 ± 4.1 7.9 ± 4.6 NS BMI (kg/m2) 16.2 ± 2.3 16.3 ± 3.4 NS EAT thickness (mm) 5.1 ± 1.4 4.5 ± 0.9 0.036 MPV (fl) 7.8 ± 1.1 7.3 ± 1.4 0.044 Onset of symptoms (years) 4.1 ± 3.0 Age of diagnosis (years) 5.8 ± 3.0 Delay in diagnosis (years) 1.7 ± 1.9 BMI: Body mass index, EAT: Epicardial adipose tissue, FMF: Familial Mediterranean fever, F: Female, M: Male, MPV: Mean platelet volume, NS: Not significant. Table 1 The main characteristics of the patients and the control group Introduction Familial Mediterranean fever (FMF) is the most common cause of hereditary recurrent fever. It is an auto- inflammatory disease, characterized by attacks of fever, abdominal pain, chest pain, arthritis, and skin rashes. Familial Mediterranean fever is common in people from Eastern Mediterranean region, especially in Sephardic Jewish, Armenian, Turkish, and Arabic population [1,2]. Systemic inflammation plays an important role in de- velopment and progression of atherosclerosis. The possi- bility of atherosclerosis was reported to be increased in cases with rheumatic diseases [1,3]. Case-control studies In this study, we investigated whether EAT thickness and MPV values differ in cases with FMF, when com- pared with healthy controls and assessed the correlation between EAT and MPV. * Correspondence: ulucaunal@hotmail.com 1Department of Pediatrics, Dicle University Medical School, Diyarbakir, Turkey Full list of author information is available at the end of the article Uluca et al. Italian Journal of Pediatrics (2015) 41:15 Page 2 of 4 Uluca et al. Italian Journal of Pediatrics (2015) 41:15 Results The study included 45 children with FMF (19 male, 26 fe- male) and 54 healthy matched controls (24 male, 30 fe- male). The main characteristics of the patient and the control groups are summarized in Table 1. The mean epi- cardial adipose tissue thicknesses of the patient and the control groups were 5.1 ± 1.4 and 4.5 ± 0.9 mm, respect- ively and the mean EAT thickness of the children with FMF were found significantly higher than that of controls (p = 0.036). FMF patients had significantly higher MPV values compared with the controls as well (7.8 ± 1.1 vs. 7.3 ± 1.4 fl, p = 0.044). There were no significant differ- ences between the groups in terms of gender, age, or BMI (p > 0.05, for each comparisons). Age at diagnosis, dur- ation of delay in diagnosis, and MPV values were found to be correlated with EAT thickness in the patient group (r = 0.49, p = 0.001 for the former parameters and r = 0.32, p = 0.04 for MPV). However, no correlation was found between disease severity score and EAT thickness. Forty-two of 45 FMF patients (93.3%) had a MEFV gene mutation. The patients were divided into three subgroups according to MEFV mutations (homozygote, heterozygote, and compound heterozygote) and the comparison of those subgroups in terms of EAT thickness and MPV revealed no significant difference (p > 0.05). Nine (21.4%) patients had M694V mutation and 33 (78.6%) patients with muta- tions other than M694V mutation such as V726A, M680I or R761H etc. Although, M694V mutation group had in- creased EAT (5.84 ± 1.94 vs. 4.79 ± 1.31 mm) and MPV (7.90 ± 0.96 vs. 7.73 ± 1.10 fl) values, while compared with the other MEFV mutations’ group, these differences did not reach to statistically significant levels (p = 0.107 and p = 0.497, respectively). Study population Forty-five FMF patients diagnosed according to Tel- Hashomer criteria [11] and age- and gender-matched 54 healthy normal weighted controls were enrolled into the study. Patients who experienced an FMF attack within two weeks prior to admission, overweighed children, and cases with dyslipidemia or any coexistent inflammatory disease were excluded from the study. The study proto- col was approved by the Dicle University Hospital Ethics Committee. Data regarding age, gender, weight, height, body mass index (BMI) which was calculated as weight/(height)2 (kg/m2), and MPV values of the patients and controls were detected and recorded. Duration of symptoms, age at diagnosis, duration of delay in diagnosis, frequency and duration of FMF attacks, response to colchicine, and determined MEditerraneanFeVer (MEFV) gene muta- tions in FMF patients were recorded. The severity of dis- ease was determined according to the criteria adapted by Ozen et al. [12] for affected children. Scores ranging from 3 to 5 points indicated mild disease, between 6 and 9 points showed moderate disease, and greater than 9 points was considered as severe disease. Statistical analysis h d The data were processed with SPSS (Statistical Pack- age for Social Sciences) 16.0 program for Windows. Kolmogorov-Smirnov test was used to determine the distribution pattern of data. Normally distributed numer- ical variables were shown as mean ± standard deviation. Because of normal distribution, numeric variables of the patients and controls were compared with Student T-test. Kruskal-Wallis test was used for the comparison of three mutation-based patient subgroups. Chi-square test was used to compare categorical variables between the study groups. Correlations between numerical variables were evaluated by Pearson’s or Spearman’s correlation analysis. A p value less than 0.05 was accepted as statistically significant. Uluca et al. Italian Journal of Pediatrics (2015) 41:15 Page 3 of 4 dysfunction, MPV and EAT thickness are some parame- ters that suggested for evaluation of increased atheroscler- otic risk [3-5,7]. Familial Mediterranean fever is an inflammatory disease suggested to predispose to athero- sclerosis [4-6]. The current study has found that FMF cases had significantly higher EAT thicknesses and MPV values when compared to healthy controls and EAT thick- ness has correlated with age of diagnosis, duration of delay in diagnosis, and MPV values. higher in patients with coronary artery disease [15]. The cause of increased EAT thickness in FMF cases remains unexplained. However, similarity in cytokine profiles of EAT and FMF disease makes us think that, these cyto- kines may have some roles in thickening of EAT and FMF. FMF may be associated with increased atheroscler- otic risk through the cytokines released by thickened EAT and/or by inflammatory cells functioning in the pathogenesis of FMF. Platelet activation indicated by in- creased MPV values in our patients may also contribute to this atherosclerotic risk. g Epicardial adipose tissue is a structure surrounding myocardium. It is composed mainly of adipocytes, but inflammatory, stromal, vascular, and nerve cells may also exist [14]. Epicardial adipose tissue expresses and se- cretes several bioactive molecules including interleukin (IL)-1, IL-1β, IL-6, IL-8, and IL-10, the levels of these in- terleukins were found as significantly higher in subjects with coronary artery disease when compared to healthy subjects [15]. Additionally, it was found that epicardial adipose tissue produces higher levels of inflammatory cytokines and reactive oxygen species than subcutaneous fat tissue in cases with coronary artery disease [16]. Re- cently, a meta-analysis of 16 studies revealed a correl- ation between EAT thickness and coronary artery disease [17]. Statistical analysis h d Increased EAT volume was reported to be a stronger coronary risk factor than increased fat volume in any other part of the body. This may be related to proximity of EAT to coronary arteries. Possibly because of above-mentioned chronic low-grade inflammation and endothelial injury associated with reactive oxygen species, increased EAT seems to play important roles in pathogenesis of atherosclerosis. Mean platelet volume and EAT thickness are sug- gested to be used in prediction of the risk of atheroscler- osis in various conditions including psoriatic arthritis, Behçet’s disease, ankylosing spondylitis, etc. [7,8,22]. Studies evaluating carotid intima media thickness and endothelial dysfunction in FMF patients have found an increased risk of atherosclerosis [4,5]. Usefulness of MPV in assessment of atherosclerotic risk in cases with FMF was also investigated. In accordance with our re- sults, most of the studies found significantly higher MPV values in FMF patients and stated that FMF is as- sociated with increased risk of atherosclerosis [6,9,10]. MPV values were found higher even in attack-free FMF cases. However, some other investigators did not deter- mine any increase in platelet volumes [23,24]. Only one research reported lower MPV values in cases having FMF [25]. No study evaluated the atherosclerotic risk of FMF patients through measurement of EAT thickness so far. Atherosclerotic risk resulted from FMF-related chronic inflammation may be further increased by EAT thickness and MPV. We think the current investigation is unique because of simultaneous assessment of MPV values and EAT thicknesses in children with FMF. The finding that both of the parameters indicating athero- sclerotic risk were increased in FMF patients may be as- sociated with higher probability of atherogenesis. In previous studies, M694V mutation has been found to- gether with severe disease and more frequent renal amyloidosis development in patients with FMF [2,12]. In our study, despite increased mean values of both EAT and MPV in M694V mutation group in comparison with other MEFV mutations’ group, these differences did not reach to significant levels, probably because of small number (n = 9) of patients in M694V subgroup. Further- more, the mean age of our study group was 8.1 years; therefore these non-significant differences can reach to significant levels with increasing age especially for EAT. Further studies with more patients are needed to clarify this point. Statistical analysis h d Platelet activation has been suggested as a triggering factor in development of atherosclerosis and the mean platelet volume is accepted as a good indicator of plate- let activation. CXCL4 is a chemokine of platelets having atherogenic activities by altering differentiation of T cells and macrophages, inhibiting apoptosis of neutrophils and monocytes, and stimulating CCL5. CCL5 is depos- ited on vascular endothelium and triggers atherogenic activities of monocytes. Platelet surface molecules such as GPIIb/IIIa, GP1bα, and P-selectin interact with endo- thelial cells, leukocytes and matrix components, which resultantly affect atherogenesis. Besides these, platelets also contribute to formation of lipid laden macrophages by modifying and endocytosing LDL particles [18,19]. Familial Mediterranean fever is an IL-1β dependent auto-inflammatory disease and is related to dysregula- tion of nod-like receptor family pyrin domain-containing 3 inflammasome (NLRP3) [20]. IL-6 and TNF-α levels were found higher in patients with FMF during both at- tack and attack-free period [21]. As mentioned above, inflammation is an important risk factor for atheroscler- osis and EAT secretes various cytokines including IL-1β, IL-6, and TNF-α, the levels of which were significantly Determination of increased risk of atherosclerosis in FMF patients even in childhood is important. MPV is a cheap and easily available parameter. Echocardiography provides simple and noninvasive assessment of EAT Uluca et al. Italian Journal of Pediatrics (2015) 41:15 Page 4 of 4 Page 4 of 4 thickness. Given that children are expected to live long years; such an atherosclerotic risk may be associated with significant morbidity or mortality. Increased aware- ness about this risk may provide to take some measures for prevention of potential morbidities. 8. Resorlu H, Akbal A, Resorlu M, Gokmen F, Ates C, Uysal F, et al. Epicardial adipose tissue thickness in patients with ankylosing spondylitis. Clin Rheumatol. 2014;34:295–9. 9. Coban E, Adanir H. Platelet activation in patients with Familial Mediterranean Fever. Platelets. 2008;19:405–8. 9. Coban E, Adanir H. Platelet activation in patients with Familial Mediterranean Fever. Platelets. 2008;19:405–8. 10. Arica S, Ozer C, Arica V, Karakus A, Celik T, Gunesacar R. Evaluation of the mean platelet volume in children with familial Mediterranean fever. Rheumatol Int. 2012;32:3559–63. 10. Arica S, Ozer C, Arica V, Karakus A, Celik T, Gunesacar R. Evaluation of the mean platelet volume in children with familial Mediterranean fever. Rheumatol Int. 2012;32:3559–63. There are some limitations of the present study to be mentioned. Competing interests Competing interests The authors declare that they have no competing interests. y 18. von Hundelshausen P, Schmitt MM. Platelets and their chemokines in atherosclerosis-clinical applications. Front Physiol. 2014;5:294. 18. von Hundelshausen P, Schmitt MM. Platelets and their chemokines in atherosclerosis-clinical applications. Front Physiol. 2014;5:294. Statistical analysis h d How to interpret epicardial adipose tissue as a cause of coronary artery disease: a meta-analysis. Coron Artery Dis. 2012;23:227–33. 18. von Hundelshausen P, Schmitt MM. Platelets and their chemokines in atherosclerosis-clinical applications. Front Physiol. 2014;5:294. 17. Xu Y, Cheng X, Hong K, Huang C, Wan L. How to interpret epicardial adipose tissue as a cause of coronary artery disease: a meta-analysis. Coron Artery Dis. 2012;23:227–33. 17. Xu Y, Cheng X, Hong K, Huang C, Wan L. How to interpret epicardial adipose tissue as a cause of coronary artery disease: a meta-analysis. Coron Artery Dis. 2012;23:227–33. Authors’ contributions Ü 19. Murat SN, Duran M, Kalay N, Gunebakmaz O, Akpek M, Doger C, et al. Relation between mean platelet volume and severity of atherosclerosis in patients with acute coronary syndromes. Angiology. 2013;64:131–6. ÜU, FA and AE have made substantial contributions to conception and design, ÜY, MNS and AG contribute to acquisition of data, ÜU, VŞ and DK contribute to analysis and interpretation of data; ÜU, FD and İT have been involved in drafting the manuscript, AE performed statistical analysis and FD carried out echocardiographic exam. All of the authors have given final approval of the version to be published. All authors read and approved the final manuscript. 20. Mitroulis I, Skendros P, Ritis K. Targeting IL-1beta in disease; the expanding role of NLRP3 inflammasome. Eur J Intern Med. 2010;21:157–63. 21. Ben-Zvi I, Livneh A. Chronic inflammation in FMF: markers, risk factors, outcomes and therapy. Nat Rev Rheumatol. 2011;7:105–12. 22. Canpolat F, Akpinar H, Eskioglu F. Mean platelet volume in psoriasis and psoriatic arthritis. Clin Rheumatol. 2010;29:325–8. Author details 1 23. Uluca U, Ece A, Sen V, Karabel D, Yel S, Gunes A, et al. Usefulness of Mean Platelet Volume and Neutrophil-to-Lymphocyte Ratio for Evaluation of Children with Familial Mediterranean Fever. Med Sci Monit. 2014;20:1578–82. 1Department of Pediatrics, Dicle University Medical School, Diyarbakir, Turkey. 2Department of Pediatric Cardiology, Dicle University Medical School, 1Department of Pediatrics, Dicle University Medical School, Diyarbakir, Turkey. 2Department of Pediatric Cardiology, Dicle University Medical School, Diyarbakir, Turkey. 3Department of Pediatric Rheumatology, Dicle University Medical School, Diyarbakir, Turkey. 4Department of Physiology, Dicle University Medical School, Diyarbakir, Turkey. Diyarbakir, Turkey. 3Department of Pediatric Rheumatology, Dicle University Medical School, Diyarbakir, Turkey. 4Department of Physiology, Dicle University Medical School, Diyarbakir, Turkey. Diyarbakir, Turkey. 3Department of Pediatric Rheumatology, Dicle University Medical School, Diyarbakir, Turkey. 4Department of Physiology, Dicle University Medical School, Diyarbakir, Turkey. 24. Makay B, Turkyilmaz Z, Unsal E. Mean platelet volume in children with familial Mediterranean fever. Clin Rheumatol. 2009;28:975–8. 24. Makay B, Turkyilmaz Z, Unsal E. Mean platelet volume in children with familial Mediterranean fever. Clin Rheumatol. 2009;28:975–8. 25. Sahin S, Senel S, Ataseven H, Yalcin I. Does mean platelet volume influence the attack or attack-free period in the patients with Familial Mediterranean fever? Platelets. 2013;24:320–3. 25. Sahin S, Senel S, Ataseven H, Yalcin I. Does mean platelet volume influence the attack or attack-free period in the patients with Familial Mediterranean fever? Platelets. 2013;24:320–3. Received: 23 December 2014 Accepted: 11 February 2015 Received: 23 December 2014 Accepted: 11 February 2015 References Karakurt Ariturk O, Ureten K, Sari M, Yazihan N, Ermis E, Erguder I. Relationship of paraoxonase-1, malondialdehyde and mean platelet volume with markers of atherosclerosis in familial Mediterranean fever: an observational study. Anadolu Kardiyol Derg. 2013;13:357–62. 7. Tasolar H, Tasolar S, Kurtulus D, Altun B, Bayramoglu A, Otlu YO, et al. Increased epicardial adipose tissue thickness on transthoracic echocardiography in patients with Behcet disease. J Ultrasound Med. 2014;33:1393–400. 2. Ece A, Cakmak E, Uluca U, Kelekci S, Yolbas I, Gunes A, et al. The MEFV mutations and their clinical correlations in children with familial Mediterranean fever in southeast Turkey. Rheumatol Int. 2014;34:207–12. 3. Beinsberger J, Heemskerk JW, Cosemans JM. Chronic arthritis and cardiovascular disease: Altered blood parameters give rise to a prothrombotic propensity. Semin Arthritis Rheum. 2014;44:345–52. 4. Bilginer Y, Ozaltin F, Basaran C, Duzova A, Besbas N, Topaloglu R, et al. Evaluation of intima media thickness of the common and internal carotid arteries with inflammatory markers in familial Mediterranean fever as possible predictors for atherosclerosis. Rheumatol Int. 2008;28:1211–6. Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color ﬁgure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color ﬁgure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color ﬁgure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit Submit your next manuscript to BioMed Central and take full advantage of: Submit your next manuscript to BioMed Central and take full advantage of: p p 5. Akdogan A, Calguneri M, Yavuz B, Arslan EB, Kalyoncu U, Sahiner L, et al. Are familial Mediterranean fever (FMF) patients at increased risk for atherosclerosis? Impaired endothelial function and increased intima media thickness are found in FMF. Rheumatol. 2008;22:811–29. 2. Ece A, Cakmak E, Uluca U, Kelekci S, Yolbas I, Gunes A, et al. The MEFV mutations and their clinical correlations in children with familial Mediterranean fever in southeast Turkey. Rheumatol Int. 2014;34:207–12. 3. Beinsberger J, Heemskerk JW, Cosemans JM. Chronic arthritis and cardiovascular disease: Altered blood parameters give rise to a prothrombotic propensity. Semin Arthritis Rheum. 2014;44:345–52. 4. Bilginer Y, Ozaltin F, Basaran C, Duzova A, Besbas N, Topaloglu R, et al. Evaluation of intima media thickness of the common and internal carotid arteries with inflammatory markers in familial Mediterranean fever as possible predictors for atherosclerosis. Rheumatol Int. 2008;28:1211–6. 5. Akdogan A, Calguneri M, Yavuz B, Arslan EB, Kalyoncu U, Sahiner L, et al. Are familial Mediterranean fever (FMF) patients at increased risk for atherosclerosis? Impaired endothelial function and increased intima media thickness are found in FMF. J Am Coll Cardiol. 2006;48:2351–3. 6. Karakurt Ariturk O, Ureten K, Sari M, Yazihan N, Ermis E, Erguder I. Relationship of paraoxonase-1, malondialdehyde and mean platelet volume with markers of atherosclerosis in familial Mediterranean fever: an observational study. Anadolu Kardiyol Derg. 2013;13:357–62. 7. Tasolar H, Tasolar S, Kurtulus D, Altun B, Bayramoglu A, Otlu YO, et al. Increased epicardial adipose tissue thickness on transthoracic echocardiography in patients with Behcet disease. J Ultrasound Med. 2014;33:1393–400. References 1. Touitou I, Kone-Paut I. Autoinflammatory diseases. Best Pract Res Clin Rheumatol. 2008;22:811–29. 1. Touitou I, Kone-Paut I. Autoinflammatory diseases. Best Pract Res Clin Rheumatol. 2008;22:811–29. Rheumatol. 2008;22:811–29. 2. Ece A, Cakmak E, Uluca U, Kelekci S, Yolbas I, Gunes A, et al. The MEFV mutations and their clinical correlations in children with familial Mediterranean fever in southeast Turkey. Rheumatol Int. 2014;34:207–12. 3. Beinsberger J, Heemskerk JW, Cosemans JM. Chronic arthritis and cardiovascular disease: Altered blood parameters give rise to a prothrombotic propensity. Semin Arthritis Rheum. 2014;44:345–52. 4. Bilginer Y, Ozaltin F, Basaran C, Duzova A, Besbas N, Topaloglu R, et al. Evaluation of intima media thickness of the common and internal carotid arteries with inflammatory markers in familial Mediterranean fever as possible predictors for atherosclerosis. Rheumatol Int. 2008;28:1211–6. 5. Akdogan A, Calguneri M, Yavuz B, Arslan EB, Kalyoncu U, Sahiner L, et al. Are familial Mediterranean fever (FMF) patients at increased risk for atherosclerosis? Impaired endothelial function and increased intima media thickness are found in FMF. J Am Coll Cardiol. 2006;48:2351–3. 6. Karakurt Ariturk O, Ureten K, Sari M, Yazihan N, Ermis E, Erguder I. Relationship of paraoxonase-1, malondialdehyde and mean platelet volume with markers of atherosclerosis in familial Mediterranean fever: an observational study. Anadolu Kardiyol Derg. 2013;13:357–62. 7. Tasolar H, Tasolar S, Kurtulus D, Altun B, Bayramoglu A, Otlu YO, et al. Increased epicardial adipose tissue thickness on transthoracic echocardiography in patients with Behcet disease. J Ultrasound Med. 2014;33:1393–400. Rheumatol. 2008;22:811 29. 2. Ece A, Cakmak E, Uluca U, Kelekci S, Yolbas I, Gunes A, et al. The MEFV mutations and their clinical correlations in children with familial Mediterranean fever in southeast Turkey. Rheumatol Int. 2014;34:207–12. 3. Beinsberger J, Heemskerk JW, Cosemans JM. Chronic arthritis and cardiovascular disease: Altered blood parameters give rise to a prothrombotic propensity. Semin Arthritis Rheum. 2014;44:345–52. 4. Bilginer Y, Ozaltin F, Basaran C, Duzova A, Besbas N, Topaloglu R, et al. Evaluation of intima media thickness of the common and internal carotid arteries with inflammatory markers in familial Mediterranean fever as possible predictors for atherosclerosis. Rheumatol Int. 2008;28:1211–6. 5. Akdogan A, Calguneri M, Yavuz B, Arslan EB, Kalyoncu U, Sahiner L, et al. Are familial Mediterranean fever (FMF) patients at increased risk for atherosclerosis? Impaired endothelial function and increased intima media thickness are found in FMF. J Am Coll Cardiol. 2006;48:2351–3. 6. Statistical analysis h d The main limitation of the study is that pa- rameters suggested to be related with atherosclerosis were evaluated cross-sectionally and follow-up measure- ments were not obtained. Relatively small sample size is the other limitation. 11. Livneh A, Langevitz P, Zemer D, Zaks N, Kees S, Lidar T, et al. Criteria for the diagnosis of familial Mediterranean fever. Arthritis Rheum. 1997;40:1879–85. 11. Livneh A, Langevitz P, Zemer D, Zaks N, Kees S, Lidar T, et al. Criteria for the diagnosis of familial Mediterranean fever. Arthritis Rheum. 1997;40:1879–85. 12. Ozen S, Aktay N, Lainka E, Duzova A, Bakkaloglu A, Kallinich T. Disease severity in children and adolescents with familial Mediterranean fever: a comparative study to explore environmental effects on a monogenic disease. Ann Rheum Dis. 2009;68:246–8. 12. Ozen S, Aktay N, Lainka E, Duzova A, Bakkaloglu A, Kallinich T. Disease severity in children and adolescents with familial Mediterranean fever: a comparative study to explore environmental effects on a monogenic disease. Ann Rheum Dis. 2009;68:246–8. 13. Bertaso AG, Bertol D, Duncan BB, Foppa M. Epicardial fat: definition, measurements and systematic review of main outcomes. Arq Bras Cardiol. 2013;101:e18–28. 13. Bertaso AG, Bertol D, Duncan BB, Foppa M. Epicardial fat: definition, measurements and systematic review of main outcomes. Arq Bras Cardiol. 2013;101:e18–28. In conclusion, epicardial adipose tissue thickness and MPV values seem to be increased in children with FMF. We think EAT thickness and MPV are useful variables indicating the risk of atherosclerosis and may be evalu- ated in FMF patients of all ages. These findings may ne- cessitate close follow-up and taking proper preventive measures. 14. Verhagen SN, Visseren FL. Perivascular adipose tissue as a cause of atherosclerosis. Atherosclerosis. 2011;214:3–10. 14. Verhagen SN, Visseren FL. Perivascular adipose tissue as a cause of atherosclerosis. Atherosclerosis. 2011;214:3–10. 15. Iacobellis G, Malavazos AE, Corsi MM. Epicardial fat: from the biomolecular aspects to the clinical practice. Int J Biochem Cell Biol. 2011;43:1651–4. 15. Iacobellis G, Malavazos AE, Corsi MM. Epicardial fat: from the biomolecular aspects to the clinical practice. Int J Biochem Cell Biol. 2011;43:1651–4. 16. Katsiki N, Mikhailidis DP, Wierzbicki AS. Epicardial fat and vascular risk: a narrative review. Curr Opin Cardiol. 2013;28:458–63. 16. Katsiki N, Mikhailidis DP, Wierzbicki AS. Epicardial fat and vascular risk: a narrative review. Curr Opin Cardiol. 2013;28:458–63. 17. Xu Y, Cheng X, Hong K, Huang C, Wan L. References J Am Coll Cardiol. 2006;48:2351–3. • Convenient online submission • Thorough peer review 6. Karakurt Ariturk O, Ureten K, Sari M, Yazihan N, Ermis E, Erguder I. Relationship of paraoxonase-1, malondialdehyde and mean platelet volume with markers of atherosclerosis in familial Mediterranean fever: an observational study. Anadolu Kardiyol Derg. 2013;13:357–62. 7. Tasolar H, Tasolar S, Kurtulus D, Altun B, Bayramoglu A, Otlu YO, et al. Increased epicardial adipose tissue thickness on transthoracic echocardiography in patients with Behcet disease. J Ultrasound Med. 2014;33:1393–400.
https://openalex.org/W2769351021	https://publications.rwth-aachen.de/record/717008/files/717008.pdf	English	null	A new multiple trauma model of the mouse	BMC musculoskeletal disorders	2,017	cc-by	9,699	© The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. * Correspondence: Stefanie_Oestern@hotmail.com 1Department of Trauma Surgery, University Medical Center of Schleswig-Holstein, Arnold-Heller Straße 7, 24105, Campus Kiel, Kiel, Germany Full list of author information is available at the end of the article A new multiple trauma model of the mouse Stefanie Fitschen-Oestern1*, Sebastian Lippross1, Tim Klueter1, Matthias Weuster1, Deike Varoga1, Mersedeh Tohidnezhad2, Thomas Pufe2, Stefan Rose-John3, Hagen Andruszkow4, Frank Hildebrand4, Nadine Steubesand1, Andreas Seekamp1 and Claudia Neunaber4 Abstract Background: Blunt trauma is the most frequent mechanism of injury in multiple trauma, commonly resulting from road traffic collisions or falls. Two of the most frequent injuries in patients with multiple trauma are chest trauma and extremity fracture. Several trauma mouse models combine chest trauma and head injury, but no trauma mouse model to date includes the combination of long bone fractures and chest trauma. Outcome is essentially determined by the combination of these injuries. In this study, we attempted to establish a reproducible novel multiple trauma model in mice that combines blunt trauma, major injuries and simple practicability. Methods: Ninety-six male C57BL/6 N mice (n = 8/group) were subjected to trauma for isolated femur fracture and a combination of femur fracture and chest injury. Serum samples of mice were obtained by heart puncture at defined time points of 0 h (hour), 6 h, 12 h, 24 h, 3 d (days), and 7 d. Results: A tendency toward reduced weight and temperature was observed at 24 h after chest trauma and femur fracture. Blood analyses revealed a decrease in hemoglobin during the first 24 h after trauma. Some animals were killed by heart puncture immediately after chest contusion; these animals showed the most severe lung contusion and hemorrhage. The extent of structural lung injury varied in different mice but was evident in all animals. Representative H&E-stained (Haematoxylin and Eosin-stained) paraffin lung sections of mice with multiple trauma revealed hemorrhage and an inflammatory immune response. Plasma samples of mice with chest trauma and femur fracture showed an up-regulation of IL-1β (Interleukin-1β), IL-6, IL-10, IL-12p70 and TNF-α (Tumor necrosis factor- α) compared with the control group. Mice with femur fracture and chest trauma showed a significant up-regulation of IL-6 compared to group with isolated femur fracture. Conclusions: The multiple trauma mouse model comprising chest trauma and femur fracture enables many analogies to clinical cases of multiple trauma in humans and demonstrates associated characteristic clinical and pathophysiological changes. This model is easy to perform, is economical and can be used for further research examining specific immunological questions. Fitschen-Oestern et al. BMC Musculoskeletal Disorders (2017) 18:468 DOI 10.1186/s12891-017-1813-9 Fitschen-Oestern et al. BMC Musculoskeletal Disorders (2017) 18:468 DOI 10.1186/s12891-017-1813-9 Background l l visceral injuries can lead to life-threatening complica- tions if not anticipated and treated [3]. Extremity fractures such as a femur fracture must be stabilized. Multiple trauma accounts for a significant proportion of deaths worldwide [1]. The most frequent injuries in trauma patients are chest trauma, extremity fractures and head injuries [2]. The time point of operative treatment is still contro- versially discussed, although most of the literature recommends early surgical stabilization of these fractures. Respiratory deterioration can be exacerbated by the presence of unstable long bone fractures. Blunt chest trauma can result in significant morbidity in injured patients, and both chest wall and intrathoracic Several trauma mouse models focus on blunt chest trauma and head injury, but to our knowledge the Page 2 of 12 Page 2 of 12 Page 2 of 12 Fitschen-Oestern et al. BMC Musculoskeletal Disorders (2017) 18:468 combination of chest trauma and long bone fractures has not been previously emphasized. by the local institutional animal care and research advis- ory committee and permitted by the local government of Lower Saxony, Germany (AZ 10AO29). The study was performed at the experimental trauma surgery labora- tory of Hannover Medical School (MHH). Trauma causes tissue damage, blood loss and activation of the immune system. The extent of the inflammatory im- mune response correlates with the degree of tissue damage [4], whereas extremity fractures in particular are generally accompanied by extensive soft tissue damage in multiple trauma patients [5]. In addition, extremity fractures are as- sociated with an increased risk of complications [6], which might explain why damage control surgery is currently a point of interest [7]. Cytokines are important components of the immune response, and their release correlates with the degree of trauma depending on the extent of the associ- ated injury [8]. Cytokines such as IL-6 serve as markers for the severity of trauma and early identification of high-risk patients for the development of posttraumatic MODS (multi organ dysfunction syndrome) [9]. y Experiments were conducted in an operating room at the animal research facility. One hundred twelve male C57BL/6 N mice (Charles River, Germany) weighing 22 ± 3 g (gram) were used for the study. Twenty male mice were used in preliminary experiments to determine the weight needed for induction of chest trauma. All mice were handled at room temperature for 14 days before treatment, and all mice were age-matched (12 weeks old). Background l l We used only male C57BL/6 N mice for this pri- mary study because gender of mice affects hormones and cell-mediated immune response [20, 21]. Cytokine expression also differs between male and female mice [22]. Further studies with female mice will be necessary. Animals were maintained under standardized conditions in a controlled environment at 21 ± 2 °C (Celsius), with a relative humidity of 50% and artificial light (14 h light, 10 h dark). They received a commercial pellet diet (altromin 1320, Altromin, Lage, Germany) and water ad libitum. Analgesic treatment was administered to all animals in the form of metamizol-sodium (200 mg/kg (milligram/kg) body weight; Novalgin® Hoechst, Unterschleiβheim, Germany) throughout the study. Mice were injected subcutaneously under deep anesthesia prior to induction of the thorax trauma and after induction of the femur fracture. For postop- erative analgesia, 0.8 mg/mL (milligram/ml) Novamin- sulfon Lichtenstein 500 mg (Zentiva Pharma GmbH, Frankfurt am Main, Germany) was added to the drinking water for the first 3 days after trauma. Chest trauma is associated with a serious risk of posttrau- matic complications, including hypoxia caused by lung con- tusion, blood loss, heart contusion, pericardial tamponade or sepsis due to esophageal or tracheal perforations. Pulmonary contusion is the most frequently diagnosed in- trathoracic injury related to blunt chest trauma, affecting 17–25% of adult blunt chest trauma patients [10]. It is also an independent risk factor for the development of pneumo- nia, severe clinical acute lung injury (ALI) and acute re- spiratory distress syndrome (ARDS) [11]. Lung contusion affects approximately 17–25% of adult patients with blunt trauma and is the leading cause of death from blunt thor- acic injury [11]. Patients with concurrent blunt chest trauma and long bone fractures have an especially higher incidence of pulmonary damage [12]. The molecular mechanisms of the immune response after multiple trauma are highly complex and not yet completely understood. Only a few murine chest trauma models have been established to date, but there is no multiple trauma model that combine a chest trauma with an isolated femur fracture. All surgical procedures were performed under deep anesthesia with isoflurane ((Minrad, Bethlehem, PA)) and local application of xylazine (16 mg/kg) (Rompun®, Bayer, Leverkusen, Germany). The mice were warmed to 36 °C using infrared heat lamps after the surgical proce- dures were complete. Wound closure was performed before recovery from the anesthesia. Group distribution and experimental procedures In addition to clinical parameters, we focused on post- traumatic cytokine release based on knowledge of the tight correlation between immunological changes and the degree of tissue damage as well as the severity of ischemia [4]. p p p Animals were randomly assigned to one of three groups. In the first group, mice received an isolated femur shaft fracture after stabilization with a pin (Fx). In the second group, mice received a combined intramedullary femoral fracture stabilization and blunt thoracic trauma (group TTFx). In the third group, mice underwent a control op- eration with intramedullary pin implantation in an intact femur without fracture (control). Background l l We aimed to develop a standardized, reproducible, and clinically relevant multiple trauma mouse model of chest trauma [13] and femur fracture [14] to investigate the pathophysiologic changes, especially cytokine ex- pression, after multiple trauma. Both methods have been evaluated in isolation in several studies and described as reproducible [15–19]. For quantification of activity as a measure of clinical status after trauma, a scoring system was used [23]. Animal care Experiments were carried out in accordance with the German Animal Welfare Legislation and were approved Page 3 of 12 Page 3 of 12 Fitschen-Oestern et al. BMC Musculoskeletal Disorders (2017) 18:468 Group Fx and TTFx were divided into six subgroups (n = 8) depending on the time point of sacrifice: 0 h, 6 h, 12 h, 24 h, 3 d and 7 d. The control group was sacrificed at 0 h (n = 8). calculations assumed that all the potential energy of the weight was transferred to the animal, neglecting fric- tional dissipation. The platform was suspended on Teflon guides to minimize friction and facilitate energy transfer to the anesthetized animal. The shield was reproducibly placed entirely over the chest without in- trusion onto the abdomen. One hundred twelve multiple trauma, femur fracture and control mice were tested (48 multiple trauma and 48 isolated femur fracture). The control group consisted of 16 mice (16,6% (percent)) that underwent an oper- ation (femur stabilization) in the absence of fracture or chest trauma (Fig. 1c). Experiments were undertaken by three different surgeons. There were no significant dif- ferences with regard to moribund animals (surgeon 1: 6 moribund mice, surgeon 2: 4 moribund mice, surgeon 3: 6 moribund mice). The experiments were performed by three different autonomous surgeons. All data were examined by a stat- istician. The cause of death was determined during organ removal immediately after death. Femur fracture The experimental design of the multiple trauma model is based on a two-hit model. The first hit consists of a closed femur fracture on the right side as described by Bonnarens [14]. In brief, under deep anesthesia with iso- flurane, a 20 gauge needle was first inserted into the canal of the mouse femur as an intramedullary pin (Fig. 1 c). After primary wound closure, a standardized femur fracture was induced in both groups using a blunt guillo- tine device weighing 500 g (0.784 J) after primary stabilization. This procedure resulted in an A-type femoral fracture combined with a moderate soft-tissue injury. The type of fracture (A fracture, AO classifica- tion) was controlled after sacrifice. Blunt thoracic trauma After induction of the femur fracture, a blunt thoracic trauma was induced by a modified version of a previ- ously described model for rats of bilateral lung contu- sion in the TTFx group [13, 24]. The method has been previously described as reproducible. Blunt thoracic trauma was induced in anesthetized mice by dropping a hollow aluminum cylindrical weight (300 g) from a height of 55 cm (centimeter) through a vertical stainless steel tube onto a Lexon platform rest- ing on the chest (Fig. 1 a, b). The impact energy E (1.617 J (Joule)) of the falling weight was calculated using the eq. E = m x g x h, where m = mass of aluminum weight (in kilograms), g = gravitational accel- eration (9.8 ms−2 (milliseconds−2)) and h = height of weight above the Lexon platform (in meters). The Body weight, activity and body temperature y g , y y p The body weight, body temperature and activity of mice were measured in all groups before trauma and after trauma before sacrifice. Fig. 1 a, b Apparatus used for the multiple trauma mouse model. A cylindrical weight of 500 g was dropped through a tube onto a plunger in contact with the chest of an anesthetized mouse. Mice were placed on a cross on a platform of acrylic glass immediately below the plunger. The plunger allowed induction of a bilateral pulmonary contusion. c Induction of femur fracture. Before a closed femur fracture was induced, the femur was stabilized using a cannula (Sterican 0.55 × 25 LILA lI) from the knee joint to the femoral neck. The method used to induce femur fracture was first described by Bonnarens. After primary wound closure, a standardized femur fracture was induced using a blunt guillotine device with a weight of 500 g (0.784 J) Fig. 1 a, b Apparatus used for the multiple trauma mouse model. A cylindrical weight of 500 g was dropped through a tube onto a plunger in contact with the chest of an anesthetized mouse. Mice were placed on a cross on a platform of acrylic glass immediately below the plunger. The plunger allowed induction of a bilateral pulmonary contusion. c Induction of femur fracture. Before a closed femur fracture was induced, the femur was stabilized using a cannula (Sterican 0.55 × 25 LILA lI) from the knee joint to the femoral neck. The method used to induce femur fracture was first described by Bonnarens. After primary wound closure, a standardized femur fracture was induced using a blunt guillotine device with a weight of 500 g (0.784 J) Page 4 of 12 Page 4 of 12 Fitschen-Oestern et al. BMC Musculoskeletal Disorders (2017) 18:468 Fitschen-Oestern et al. BMC Musculoskeletal Disorders (2017) 18:468 Weight and temperature g p All mice that underwent chest trauma and femur frac- ture showed a tendency of reduced weight after 6 h. Representative the weight loss was shown for 5 mice with an initial weight of 25,62 g, 25,58 g, 26,54 g, 24,40 g and 24,83 g. After 6 h the weight was reduced to 24,06 g, 24,95 g, 25,39 g, 24,16 g and 23,17 g (Fig. 2). After 3 days the weight had returned completely to baseline values. Temperature declined until 24 h after trauma (Fig. 3). Representative temperature was shown for 5 mice before (38,7 °C, 37,6 °C, 37,8 °C, 37,2 °C, 38,7 °C) and 24 h after the trauma (37,2 °C, 36,8 °C, 37,6 °C, 37 °C, 37,5 °C). After 24 h weight and temperature returned to baseline values. There was no significant decrease in either temperature or weight after trauma. The total scan time was approximately 14 min. Scan- ning of mice lungs has been described previously [27]. The lungs of mice, which were killed immediately after trauma, were scanned using a Novotec MicroScope (Novotec Medical GmbH, Pforzheim) at an isotropic nominal spatial resolution (voxel size) of 15–20 μm. Samples were transported on ice before the scanned lungs were positioned on a special platform to prevent artifacts. Image analysis was performed using ImageJ software. Harvesting procedure Animals were sacrificed under deep anesthesia with isoflur- ane at 0 h, 6 h, 12 h, 24 h and 3 d after trauma induction. Heparinized blood was obtained via cardiac puncture. Histology Tissue samples were embedded in paraffin. Sections (5 μm (micrometer)) were obtained from the central portion of the lung with a sliding microtome (HM 430; Microm International), placed on Superfrost Plus microscope slides (Thermo Scientific) and incubated overnight at 60 °C. The sections were routinely stained with hematoxylin and eosin (H&E). Safranin O staining was carried out for 6 min using a 0.1% aqueous solution at pH 3.0. Specimens Animals were sacrificed immediately after trauma, after 6 h, 12 h, 24 h, 3 d and 7 d to obtain samples for histo- logic examination. Tissue samples from lung were collected and stored at −20 °C until processed. Protein analysis of cytokines To analyze concentrations of different cytokines, blood samples obtained by heart puncture of the mice were centrifuged for five minutes. The supernatant was removed and stored at 20 °C until processing. The con- centrations of IL-1β, IL-6, IL-10, IL-12p70 and TNFα in plasma samples were analyzed using a Luminex assay ac- cording to standard protocols with LiquiChip200 (Qiagen). A Milliplex cytokine multiplex immunoassay kit (MPXHCYTO-60 K-01; Millipore) was used for protein detection. Micro computed tomography Chest trauma of multiple trauma mice was assessed by micro computed tomography (μCT). The CT scan was performed at the Molecular Imaging North Competence Center (Am Botanischen Garten 14, 24,118 Kiel). Micro computed tomography in Kiel has been applied previously to mice in several studies [25, 26]. Two radiologists planed every scan. Statistics Statistical analysis was performed using a standard soft- ware application (SPSS Inc., Chicago, IL, USA). Differ- ences between the sham group and the other groups were evaluated using the Wilcoxon signed-rank test. For compromise of mice with an isolated fracture and mice with chest trauma and a femur fracture we used further the Mann-Whitney U test. Probability values less than 0.05 were considered statistically significant. The Data are shown as box-and-whisker plot with median and interquartile range. Results Survival Regarding the reproducibility of the model in the group of multiple trauma mice, 16 mice (33%) died after chest trauma because of hemorrhage and 32 mice (66%) survived. Assessment of blood parameters Blood was centrifuged at 2500×g for 5 min (minutes) at room temperature (Eppendorf 3200, Hamburg, Germany). After centrifugation, the plasma was transferred into a fresh tube, snap-frozen and stored at −80 °C. Samples for the control group were collected using the retrobulbar technique during the preliminary test. Post- traumatic control was performed by heart puncture. By using 2-mL (milliliter) syringes (Pico50, Radiometer Medical, Brønshøj, Denmark) containing 80 IU electrolyte-balanced heparin, blood samples (0.7 ml) for blood gas analysis and assessment of marker enzyme ac- tivities were collected from heart. The animals were sacrificed by heart puncture under deep anesthesia. The hemoglobin concentration, hematocrit and metabolic parameters (lactate, glucose), and osmolality were assessed using a blood gas analyzer (ABL 715, Radiom- eter, Copenhagen, Denmark). Ct Macroscopic and microscopic analyses showed that im- mediate death was caused by intrathoracic bleeding or heart contusion. CT scans were performed on mice that were killed immediately after chest trauma (Fig. 5). Hemothorax and lung contusion could be observed on the thoracic CT. The injured mice had no rib fractures. All mice showed comparable results on CT scan. Fig. 2 Weight of the mice after multiple trauma. The mouse numbers are equal to the order of the data. The weight of mice was determined before multiple trauma and before each mouse was sacrificed. A decrease in weight was observed during the first 24 h following trauma (shown for 6 h after trauma) Histology Histological sections were examined to assess the severity of pulmonary tissue injury in mice (Fig. 6). HE staining was performed for lung samples of mice with chest trauma and an unoperated control group. At 24 h post- Fig. 4 Hemoglobin measurements in mice after multiple trauma. The hemoglobin level in mice without trauma was compared to hemoglobin level in mice after multiple trauma. Five mice without trauma underwent retrobulbar puncture to measure hemoglobin as a control. Mice that underwent trauma were sacrificed by heart puncture at 6 h, 12 h, 24 h, 3 d, 7 d, 14 d, and 28 d after trauma. Data are shown for 5 mice each at 6, 12, and 24 h after trauma. Blood samples were analyzed immediately after puncture using a Radiometer ABL 700. Hemoglobin levels decreased up to 24 h Fig. 3 Body temperature of the mice after multiple trauma. The mouse mice are equal to the order of the data. The body temperature (°C) of mice was measured before multiple trauma was induced and before each mouse was sacrificed. A drop in temperature was observed during the first 24 h after trauma (shown for 6 h after trauma) Fig. 4 Hemoglobin measurements in mice after multiple trauma. Th h l bi l l i i i h d Fig. 4 Hemoglobin measurements in mice after multiple trauma. The hemoglobin level in mice without trauma was compared to hemoglobin level in mice after multiple trauma. Five mice without trauma underwent retrobulbar puncture to measure hemoglobin as a control. Mice that underwent trauma were sacrificed by heart puncture at 6 h, 12 h, 24 h, 3 d, 7 d, 14 d, and 28 d after trauma. Data are shown for 5 mice each at 6, 12, and 24 h after trauma. Blood samples were analyzed immediately after puncture using a Radiometer ABL 700. Hemoglobin levels decreased up to 24 h Fig. 3 Body temperature of the mice after multiple trauma. The mouse mice are equal to the order of the data. The body temperature (°C) of mice was measured before multiple trauma was induced and before each mouse was sacrificed. A drop in temperature was observed during the first 24 h after trauma (shown for 6 h after trauma) Fig. 3 Body temperature of the mice after multiple trauma. The mouse mice are equal to the order of the data. Hemoglobin In a previous examination 5 mice of the same age and weight were punctured retrobulbar before trauma induction Fitschen-Oestern et al. BMC Musculoskeletal Disorders (2017) 18:468 Page 5 of 12 Page 5 of 12 Fig. 2 Weight of the mice after multiple trauma. The mouse numbers are equal to the order of the data. The weight of mice was determined before multiple trauma and before each mouse was sacrificed. A decrease in weight was observed during the first 24 h following trauma (shown for 6 h after trauma) as a control to measure the baseline hemoglobin values. Compared to the control group (14,8 g/dl (gram/deciliter), 15,3 g/dl, 14,4 g/dl, 14 g/dl, 15,3 g/dl, 14,9 g/dl), hemoglobin declined after 6 h (13,7 g/dl, 13 g/dl, 13,5 g/dl, 11,7 g/dl, 13,7 g/dl, 12,2 g/dl) and 12 h (13,7 g/dl, 11,7 g/dl, 13,2 g/dl, 14,7 g/dl, 12,5 g/dl, 14,1 g/dl) until 24 h after induction of chest trauma (14,4 g/dl, 13,4 g/dl, 13,2 g/dl, 12 g/dl, 12,3 g/dl, 13,7 g/dl) (Fig. 4). Hemoglobin values had returned to baseline values 3 days after trauma. All tested mice showed compar- able results. Ct Histology The body temperature (°C) of mice was measured before multiple trauma was induced and before each mouse was sacrificed. A drop in temperature was observed during the first 24 h after trauma (shown for 6 h after trauma) Page 6 of 12 Fitschen-Oestern et al. BMC Musculoskeletal Disorders (2017) 18:468 Fitschen-Oestern et al. BMC Musculoskeletal Disorders (2017) 18:468 Fig. 5 CT scan of chest trauma. Computerized tomography was performed for 5 mice that died immediately after chest trauma. The CT scans revealed intrathoracic bleeding and hemopneumothorax. Scans were performed in the Molecular Imaging North Competence Center, CAU Kiel Fig. 5 CT scan of chest trauma. Computerized tomography was performed for 5 mice that died immediately after chest trauma. The CT scans revealed intrathoracic bleeding and hemopneumothorax. Scans were performed in the Molecular Imaging North Competence Center, CAU Kiel contusion, HE staining of lung samples showed thickening of the alveolar lining with ongoing leukocytic infiltration. All of the stained lungs showed comparable results. After an isolated femur fracture, an up-regulation of IL-12p70 could be detected from 0 h (679,24 ± 578,52) to 6 h (1523,32 ± 480,26). Similar to mice with an isolated fracture, IL-12p70 was up-regulated and could be detected from 0 h (321,64 ± 294,74 pg/ml) to 6 h after multiple trauma (1671,78 ± 350,74 pg/ml). Inflammatory markers Different cytokines were analyzed by multiplex immunoassay in plasma samples of mice with an isolated femur fracture and in mice with a chest trauma and femur fracture (Fig. 7 a, b, c, d, e). Mice with an isolated femur fracture showed a down-regulation of IL1β from 0 h (808,84 ± 190,93 pg/ml (picogram/ml)) to 12 h (492,81 ± 190,93 pg/ml) and an up- regulation of IL-6 from 0 h (39,32 ± 18,58) to 6 h (76,67 ± 16,40). IL1β and IL-6 were up-regulated from 0 h (IL1β 209,12 ± 166,51 pg/ml; IL-6 22,98 ± 9,59 pg/ml) to 6 h after multiple trauma (IL1β 829,40 ± 163,87 pg/ml; IL-6 99,88 ± 65,18 pg/ml). IL-10 expression was down-regulated from 0 h (2464,06 ± 894,49) to 12 h (1773,15 ± 742,4) in mice with an isolated fracture. In contrast, the maximum of IL-10 expres- sion was reached at 12 h after multiple trauma (2519,12 ± 1782,87 pg/ml), whereas IL-10 expression was reduced dir- ectly after trauma (0 h 1085,22 ± 702,85 pg/ml) compared with the control group (1644,08 ± 1001,46 pg/ml). g A slight regulation of TNFα expression was observed after an isolated fracture, whereas TNFα expression declined immediately after multiple trauma (0 h 578,93 ± 232,48 pg/ml) compared with the control group (1241,5 ± 266,22 pg/ml). The baseline level of TNFα expression was recovered at 6 h after trauma (1288,76 ± 693,74 pg/ml). After multiple trauma all cytokines (IL1β, IL-6, IL-12p70, IL-10 and TNFα) were up-regulated (Fig. 7 a, b, c, d, e). Mice with multiple trauma showed a significant up-regulation for IL-6 compared to mice with an isolated fracture. Discussion Blunt chest trauma represents one of the most common injuries in multiple trauma patients [28], while lung con- tusion is one of the most important factors contributing Fig. 6 Microscopic evaluation of histological changes in the lung of mice after pulmonary contusion. Mice subjected to chest injury showed extensive intra-alveolar and intrabronchial hemorrhaging with consecutive atelectasis, while sham mice did not show such pulmonary changes Fig. 6 Microscopic evaluation of histological changes in the lung of mice after pulmonary contusion. Mice subjected to chest injury showed extensive intra-alveolar and intrabronchial hemorrhaging with consecutive atelectasis, while sham mice did not show such pulmonary changes Fig. 6 Microscopic evaluation of histological changes in the lung of mice after pulmonary contusion. Mice subjected to extensive intra-alveolar and intrabronchial hemorrhaging with consecutive atelectasis, while sham mice did not show su Page 7 of 12 Fitschen-Oestern et al. BMC Musculoskeletal Disorders (2017) 18:468 a b c d e a, b, c, d, e Cytokine analysis assessed by the Luminex assay. Serum samples of mice with an isolated femur fracture (Fx) and mice with rauma and femur fracture (TTFx) were analysed by multiplex immunoassay. All Data are shown as box-and-whisker plot with median and uartile range. Concentration of different cytokines (IL-1β, IL-6, IL-10, IL-12p70 and TNFα) was measured in different groups of mice 0 h, 6 h, d and 7 d (n = 8) after trauma. Figures show regulation of IL-1α, IL-6, IL-10, IL-12p70 and TNFα after an isolated femur fracture from 0 h to er trauma and cytokine regulation after induction of chest trauma and femur fracture from 0 h to 7 d after trauma. Cytokine expression mpared respectively to Sham group a a b c c c d d e e Fig. 7 a, b, c, d, e Cytokine analysis assessed by the Luminex assay. Serum samples of mice with an isolated femur fracture (Fx) and mice with chest trauma and femur fracture (TTFx) were analysed by multiplex immunoassay. All Data are shown as box-and-whisker plot with median and interquartile range. Concentration of different cytokines (IL-1β, IL-6, IL-10, IL-12p70 and TNFα) was measured in different groups of mice 0 h, 6 h, 12 h, 3 d and 7 d (n = 8) after trauma. Discussion Figures show regulation of IL-1α, IL-6, IL-10, IL-12p70 and TNFα after an isolated femur fracture from 0 h to 7 d after trauma and cytokine regulation after induction of chest trauma and femur fracture from 0 h to 7 d after trauma. Cytokine expression was compared respectively to Sham group Fitschen-Oestern et al. BMC Musculoskeletal Disorders (2017) 18:468 Page 8 of 12 to the increased morbidity and mortality of multiple trauma patients [29]. Femoral fractures represent one of the most prevalent associated injuries in multiple trauma patients with blunt thoracic trauma [30]. The presence of long bone fractures causing respiratory deterioration and respiratory dysfunction may preclude orthopedic surgical intervention for several days. however, they show suppression of compensatory mechanisms. Uncontrolled hemorrhagic shock models represent the clinical situation but are less standardized [40]. The manipulation of a single variable such as vol- ume, blood pressure and time may cause unpredictable, irreproducible results so that hemorrhagic shock models are difficult to compare [40]. In our study, we investigated a reproducible new mul- tiple trauma mouse model using the combination of these two major injuries. The main questions of the present study may be summarized as follows: Chest trauma in small animals can be induced by, for instance, a blast wave generator [41] or weight-induced bilateral lung contusion [13], which we used in our model. Blast injury is an important cause of trauma in military conflicts or terrorism, whereas weight-induced trauma imitates the trauma that occurs in traffic acci- dents [42]. Why were mice chosen as our experimental animal? Why did we choose the combination of chest trauma and femur fracture? Why did we choose the combination of chest trauma and femur fracture? What are the influences of chest trauma and femur fractures? A blast generator created laser induced stress waves and the intensity of the shock wave is flexible by varying the laser energy [42]. The trauma model of bilateral lung contusion induced by a focused external blunt chest trauma (Fig. 1a, b) has the advantages of being specific in terms of lung contusion (Figs 5 and 6) and uses a method of impact induction, which is reproducible and highly relevant to the chest trauma that occurs in motor vehicle accidents [13]. How is the immune response altered in terms of cyto- kine expression? Discussion Mice are currently the experimental tool of choice for the majority of immunologists, and the study of their immune responses has offered tremendous insight into the functions of the human immune system [31]. Humans and mice share approximately 80% of their genes [32], and unlike large animal models, mice are technically easier to implement, have lower acquisition and housing costs and superior ethical acceptance and are available as knockout animals. Despite the high incidence of chest trauma and femur fracture, there are no trauma mouse models combining these two injuries to date, and the immunologic alterations following pulmonary contusion remain insuf- ficiently elucidated. Traumatic brain injury, thoracic trauma, hemorrhagic shock and long bone fracture are the focus of most mural trauma models. All these models have advantages and disadvantages. Apart from thoracic injuries, long bone fractures are particularly critical and represent one of the most preva- lent associated injuries in multiple trauma patients with blunt thoracic injuries [30, 43]. Tibia fracture models have the advantage of easier intramedullary access com- pared to the femur [44]. The diameter of the femur is relatively consistent and large compared with that of the tibia, which facilitates the use of larger implants and the bone is more thickly covered by muscle [44]. Some trauma models focus on an isolated organ or tis- sue injury, and some models focus on the combination of several severe injuries. To concentrate on a particular injury might be an advantage in some ways, but it does not replicate multiple trauma in humans [33]. Several trauma studies focus on traumatic brain injur- ies [34–36]. The knowledge about outcome rates after concomitant traumatic brain injuries may help prioritize the research in this regard. However traumatic brain in- jury models have the limitation of not reflecting exactly the clinical setting and posttraumatic intensive monitor- ing in humans [37]. In an open femur fracture model the bone is exposed and fractured via osteotomy or by weakening the bone with several drill holes [45]. Open fracture models are stabilized by extramedullary fixation techniques like a locking plate or an external fixator [39]. Induction of an open fracture and extramedullary stabilization generates considerable soft tissue injury. External fixation has the disadvantage of high implant weight and the large vari- ation in implant stiffness [46]. An external plate fixation may damage the periosteum and perfusion and nutrition [46]. Discussion Interestingly, no rib fractures were found in our study or mentioned in previous evaluations [56, 57]. The murine chest exhibits high flexibility because of the cos- tal dorso-ventral joints, which are not present in the hu- man thorax [58, 59]. In humans, the metabolic response to severe injury results in hypothermia and weight loss. Similar to clinical conditions, the mice displayed a decrease in body temperature, weight loss and blood loss after multiple trauma during the first 24 h after trauma (Fig. 2, Fig. 3, Fig. 4). Accidental hypothermia is a serious problem in multiple trauma patients because of the negative pathophysiological effects [60]. Early rewarming ap- pears to be essential for the treatment of hypothermic trauma patients. In our study, the mice were placed under an incubator lamp for the first 12 h after in- duction of trauma, but body temperature did not re- cover until 24 h post-trauma potentially because only external warming was applied without the donation of warm infusions or a blood supply, which is normally administered to trauma patients. Although early fracture stabilization can minimize sev- eral pulmonary complications such as fat embolism [50], damage control surgery and the timing of optimal treat- ment in multiple trauma patients are still points of interest [51]. Damage control during femur fracture stabilization has been shown to be beneficial for preventing the second hit inflammatory reaction and is associated with decreased blood loss and decreased mortality and morbidity in trauma patients [52]. There is evidence that early fracture fixation reduces the incidence of fracture-related compli- cations and improves fracture outcome. The posttraumatic inflammatory reaction in humans and mice is essentially regulated by cytokine expression [61, 62]. The magnitude of cytokine expression is regu- lated by the trauma severity in humans [63]. TNFα, IL-6, IL-1β and IL-10 correlate with the systemic inflamma- tory response and injury severity [64, 65], and therefore we focused on these mediators in mice. Multiple trauma patients with severe damage or limited lung function ex- hibit significantly higher cytokine patterns in the early post-injury phase, with elevations of TNFα, IL-6, IL-10 and IL-1β compared with other trauma patients [66]. We detected an increase in TNFα, IL-6 and IL-1β in the plasma samples from multiple trauma mice. The mortality rates of multiple trauma patients range from 7% to 45%, depending on the injury severity [53]. Discussion In contrast to most of the intramedullary stabilization techniques the external stabilization pro- vides rotation stability after fracture [47]. Hemorrhagic shock ist the leading cause of morbidity and mortality in trauma patients [38]. In mouse models hemorrhagic shock can be induced volume controlled, pressure controlled or uncontrolled [39]. While volume- controlled hemorrhagic shock shows compensatory physiological mechanisms and is easy and less invasive to perform, it provides the disadvantage of an uncertain severity of hemorrhage [40]. Pressure-controlled hemorrhagic shock models are standardized and repro- ducible models that allow the analysis of severe shock states and the monitoring of physiological parameters; In a closed femur fracture model the fracture usually followed by placement of intramedullary screws, pins or locking nails [14, 46, 47]. Closed fracture models have the advantage of reduced risk of wound infection Fitschen-Oestern et al. BMC Musculoskeletal Disorders (2017) 18:468 Page 9 of 12 Page 9 of 12 of 22 ± 3 g. The impact energy we choose might be high in relation to the small mural chest. compared to an open osteotomy [14]. Several intrame- dullary stabilization systems are not stable against longi- tudinal and rotational deformations [44]. A locking nail offers higher stability compared to pin fixation but is not a rigid fixation technique and the operation time is lon- ger which might be a relevant disadvantage in a multiple trauma model [44]. Rotation stability can be achieved by flattening the tip and the distal end of a needle or a pin [46]. In our trauma mouse model, femur fracture was stabilized by minimal invasive surgery before the fracture was induced. Prior to fracturing the femur a 20 gauge needle was inserted in the medullary cavity of the femur to maintain axial alignment during the fracture and avoid large displacements. We chose this method de- scribed by Bonnarens [14] because it offers accurate re- duction, a reduced operation time, less costs and less blood loss than external fixation [48]. While stabilization is performed immediately before the induction of frac- ture, the second hit inflammatory reaction can also be prohibited similar to damage control. Nevertheless, a higher incidence of complications and higher mortality after fracture stabilization is always observed in the pres- ence of severe thoracic injuries [49]. The animals that did not survive died immediately after the chest trauma due to intrathoracic hemorrhage, which was confirmed by CT scan and removal of the or- gans. Discussion In multiple trauma patients, 20–25% of deaths are at- tributed to chest injury [54]. In our study, 16 mice (33%) died within the first 30 min after chest trauma, and 64 mice (66%) survived. The mortality rate in our trauma model was high compared to other studies [41, 42, 55] but most of the thoracic trauma models focus on an iso- lated thoracic injury [41, 42, 55]. Examining isolated organ injury may be of benefit; however, this does not accurately replicate human trauma, which often involves multiple organ systems [33]. We focused on IL-12, which is produced at high levels by monocytes and macrophages. Blunt chest trauma induces mediator-dependent monocyte migra- tion to the lung [67], and high expression of IL-12 can be detected in the monocytes of trauma patients [68]. An increase of IL-12p70 could also be detected in multiple trauma mice. Additionally, in some models only one side or a special part of the chest is affected [42, 55] whereas our model is a bilateral contusion model [13]. In preliminary test we determined a high impact energy to generate a severe chest trauma. The thoracic trauma was first described for rats (body weight 250–300 g) with an impact energy of 2,45 J [13]. The falling weight in our study had an im- pact energy of 1617 J, and the mice had a body weight In comparison to humans, a correlation between cytokine expression and the severity of trauma could also be detected in our trauma model. Mice with a single fracture generally showed reduced cytokine expression compared with multiple trauma mice. TNFα, IL-6 and IL-1β are rapidly acting cytokines in humans, and peak levels can be detected within 24 h Page 10 of 12 Fitschen-Oestern et al. BMC Musculoskeletal Disorders (2017) 18:468 Page 10 of 12 Page 10 of 12 after trauma [69]. Similar results were obtained for mice after trauma, with an increase observed at 24 h. Our re- sults using the murine trauma model were also consist- ent with the results of an isolated blast wave trauma model or burn injury model [41]. In summary, mice seems to show comparable cytokine expression patterns to human trauma patients. samples from all animals. HA helped to control all animals in the first 24 h. Elisa experiments were performed by SFO with the help of MT, TP, NS and SRJ. References %: percent; °C: celsius; ALI: acute lung injury; ARDS: acute respiratory distress syndrome; Cm: centimeter; D: day; dl: deciliter; E: impact energy; Fx group: fracture group; g: gram; g: gravitational acceleration; h: height of weight above the Lexon platform (in meters); h: hour; H&E- stained: Haematoxylin and Eosin-stained; IL: Interleukin; J: Joule; Kg: kilogram; m: mass of aluminum weight; mg: milligram; min: minutes; ml: milliliter; MODS: multi organ dysfunction syndrome; ms: millisecond; pg: picogram; TNF-α: Tumor necrosis factor- α; TTFx group: thorax trauma fracture; μCT: micro computed tomography; μm: micrometer 1. Lenz A, Franklin GA, Cheadle WG. Systemic inflammation after trauma. Injury, Int. J. Care Injured. 2007;38:1336–45. 1. Lenz A, Franklin GA, Cheadle WG. Systemic inflammation after trauma. Injury, Int. J. Care Injured. 2007;38:1336–45. 2. Bardenheuer M, Obertacke U, Waydhas C, Nast-Kolb D. Epidemiology of the severely injured patient. A prospective assessment of preclinical and clinical management. AG Polytrauma of DGU. Unfallchirurg. 2000;103:355–63. 3. Wanek S, Mayberry JC. Blunt thoracic trauma: flail chest, pulmonary contusion, and blast injury. Crit Care Clin. 2004;20:71–81. 4. Hildebrand F, Pape HC, Krettek C. The importance of cytokines in the posttraumatic inflammatory reaction. Unfallchirurg. 2005;108(10):793–4. 796– 803 Competing interests h h d l h The authors declare that they have no competing interests. This trauma model will be extremely helpful to answer outstanding questions concerning whether cytokine blockade, which is available in the clinic, is helpful for the treatment of trauma patients. Such studies can be further complemented by the evaluation of genetically modified mice that lack particular cytokines, in terms of the course of multiple traumatic situations. These stud- ies will eventually lead to better therapeutic approaches to this life-threatening condition. Specifically, it is even more important to develop new animal models with the most frequently encountered injuries for further medical improvement, necessitating further studies. Availability of data and materials The datasets used and analysed during the current study are available from the corresponding author on reasonable request. All data generated or analysed during this study are included in this published article. The manuscript, including related data, figures and tables have not been previously published and are not under consideration elsewhere. 9. Guisasola MC, Ortiz A, Chana F, Alonso B, Vaquero J. Early inflammatory response in polytraumatized patients: cytokines and heat shock proteins. A pilot study. Orthop Traumatol Surg Res. 2015;101(5):607–11. pilot study. Orthop Traumatol Surg Res. 2015;101(5):607–11. 10. Crohn SM. Pulmonary contusion: review of the clinical entity. J Trauma. 1997;42:973–9. Consent for publication The manuscript does not contain any individual person’s data. Not applicable The manuscript does not contain any individual person’s data. Not applicable Discussion Data analysis was supported by AS, SL, FH, MW, TK and DV. All authors were involved in reading and approving the final manuscript. Funding All i All experiments were financed by research funds of the department. The funds are not bound to any data collection, data analysis or result. Further an AO funds (Arbeitsgemeinschaft Osteosynthese, F379014) was used to finance part of elisa experiments. Before starting with the experiments an application with the experimental set-up ws sent to the AO. AO funding had no influence on data analysis, process, results or writing the manuscript. 7. Nahm NJ, Vallier HA. Timing of definitive treatment of femoral shaft fractures in patients with multiple injuries: a systematic review of randomized and nonrandomized trials. J Trauma Acute Care Surg. 2012; 73(5):1046–63. 8. Seekamp A, Jochum M, Ziegler M, van Griensven M, Martin M, Regel G. Cytokines and adhesion molecules in elective and accidental trauma-related ischemia/reperfusion. J Trauma. 1998;44(5):874–82. 8. Seekamp A, Jochum M, Ziegler M, van Griensven M, Martin M, Regel G. Cytokines and adhesion molecules in elective and accidental trauma-related ischemia/reperfusion. J Trauma. 1998;44(5):874–82. Conclusion We have established a new multiple trauma mouse model that better recapitulates the immunological response of severely injured patients. Despite clear dif- ferences between humans and animals, animal studies are necessary to gain further insight into the physio- logical mechanisms underlying multiple trauma. Ethics approval and consent to participate Experiments were carried out in accordance with the German Animal Welfare Legislation, and were approved by the local institutional animal care and research advisory committee (Kiel and Hannover, Germany) and permitted by the local government of Lower Saxony, Germany (reference number: AZ 10AO29). Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Acknowledgements Not applicable. 5. Kauvar D, Sarfati MR, Kraiss LW. National trauma databank analysis of mortality and limb loss in isolated lower extremity vascular trauma. J Vasc Surg. 2011;53(6) 6. Kobbe P, Lichte P, Pape HC. Complex extremity fractures following high energy injuries: the limited value of existing classifications and a proposal for a treatment-guide. Injury. 2009;40(4):69–74. Author details 1 1Department of Trauma Surgery, University Medical Center of Schleswig-Holstein, Arnold-Heller Straße 7, 24105, Campus Kiel, Kiel, Germany. 2Department of Anatomy and Cell Biology, RWTH Aachen University, Wendlingweg 2, D-52074 Aachen, Germany. 3Department of Biochemistry, Medical Faculty, Olshausenstr. 40, 24098 Kiel, Germany. 4Department of Trauma Surgery, RWTH Aachen University, Pauwelsstraße 30, 52074 Aachen, Germany. 1Department of Trauma Surgery, University Medical Center of Schleswig-Holstein, Arnold-Heller Straße 7, 24105, Campus Kiel, Kiel, Germany. 2Department of Anatomy and Cell Biology, RWTH Aachen University, Wendlingweg 2, D-52074 Aachen, Germany. 3Department of Biochemistry, Medical Faculty, Olshausenstr. 40, 24098 Kiel, Germany. 4 4Department of Trauma Surgery, RWTH Aachen University, Pauwelsstraße 30, 52074 Aachen, Germany. 4Department of Trauma Surgery, RWTH Aachen University, Pauwelsstraße 30, 52074 Aachen, Germany. Received: 11 June 2017 Accepted: 6 November 2017 Received: 11 June 2017 Accepted: 6 November 2017 Received: 11 June 2017 Accepted: 6 November 2017 Authors’ contributions Models of hemorrhagic shock: differences in the physiological and inflammatory response. Cytokine. 2013;61(2):585–90. 17. Dolgachev VA, Yu B, Reinke JM, Raghavendran K, Hemmila MR. Host susceptibility to gram-negative pneumonia after lung contusion. J Trauma Acute Care Surg. 2012;72(3):614–22. 41. Knöferl MW, Liener UC, Seitz DH, Perl M, Brückner UB, Kinzl L, Gebhard F. Cardiopulmonary, histological, and inflammatory alterations after lung contusion in a novel mouse model of blunt chest trauma. Shock. 2003;19:519–25. 18. Drissi H, Paglia DN. Surgical procedures and experimental outcomes of closed fractures in rodent models. Femur fracture. Methods Mol Biol. 2015;1226:193–211. 42. Satoh Y, Sato S, Saitoh D, Tokuno S, Hatano B, Shimokawaji T, Kobayashi H, Takishima K. Pulmonary blast injury in mice: a novel model for studying blast injury in the laboratory using laser-induced stress waves. Lasers Surg Med. 2010;42(4):313–8. 19. Aurégan JC, Coyle RM, Danoff JR, Burky RE, Akelina Y, Rosenwasser MP. The rat model of femur fracture for bone and mineral research: An improved description of expected comminution, quantity of soft callus and incidence of complications. Bone Joint Res. 2013;2(8):149–54. 43. Rixen D, Sauerland S, Oestern H-J, Bouillon B. Management strategies in the first operative phase after long-bone injury of the lower extremity in multiple-injured patients. A systematic literature review. Unfallchirurg. 2005;108(10):829–42. 20. Nadkarni S, McArthur S. Oestrogen and immunomodulation: new mechanisms that impact on peripheral and central immunity. Curr Opin Pharmacol. 2013;13:576–81. 44. Qi B, Yu J, Zhao Y, Zhu D, Yu T. Review article mouse fracture models: a primer. Int J Clin Exp Med. 2016;9(7):12418–29. 21. Hernández-Bello R, Nava-Castro K, Muñiz-Hernández S, Nava-Luna P, Trejo- Sánchez I, Tiempos-Guzmán N, Mendoza-Rodríguez Y, Morales-Montor J. Beyond the reproductive effect of sex steroids: their role during immunity to helminth parasite infections. Mini Rev Med Chem. 2012;12:1071–80. 45. Klein M, Stieger A, Stenger D, Scheuer C, Holstein JH, Pohlemann T, Menger MD, Histing T. Comparison of healing process in open osteotomy model and open fracture model: Delayed healing of osteotomies after intramedullary screw fixation. J Orthop Res. 2015;33(7):971-8. 46. Holstein JH, Menger MD, Culemann U, Meier C, Pohlemann T. Development of a locking femur nail for mice. J Biomech. 2007;40(1):215–9. 22. Crockett ET, Spielman W, Dowlatshahi S, He J. Sex differences in inflammatory cytokine production in hepatic ischemia-reperfusion. J Inflamm (Lond). 2006;3:16. 47. Manigrasso MB, O’Connor JP. Characterization of a closed femur fracture model in mice. J Orthop Trauma. 2004;18(10):687–95. 23. Authors’ contributions 11. Miller PR, Croce MA, Bee TK, Qaisi WG, Smith CP, Collins GL, Fabian TC. ARDS after pulmonary contusion: accurate measurement of contusion volume identifies high-risk patients. J Trauma. 2001;51:223–8. We declare, that all authors have made substantial contributions to the conception and design of the study, acquisition of data, analysis and interpretation of data. All authors drafted the article critically for important intellectual content and made substantial contributions to final approval of the version to be submitted. SFO and CN operated all animals and retrieved 12. Pelias ME, Townsend MC, Flancbaum L. Long bone fractures predispose to pulmonary dysfunction in blunt chest trauma despite early operative fixation. Surgery. 1992;111(5):576–9. Page 11 of 12 Page 11 of 12 Page 11 of 12 Fitschen-Oestern et al. BMC Musculoskeletal Disorders (2017) 18:468 13. Raghavendran K, Davidson BA, Helinski JD, Marschke CJ, Manderscheid P, Woytash JA, Notter RH, Knight PR. A rat model for isolated bilateral lung contusion from blunt chest trauma. Anesth Analg. 2005;101(5):1482–9. 36. Flierl MA, Stahel PF, Beauchamp KM, Morgan SJ, Smith WR, Shohami E. Mouse closed head injury model induced by a weight-drop device. Nat Protoc. 2009;4(9):1328–37. 13. Raghavendran K, Davidson BA, Helinski JD, Marschke CJ, Manderscheid P, Woytash JA, Notter RH, Knight PR. A rat model for isolated bilateral lung contusion from blunt chest trauma. Anesth Analg. 2005;101(5):1482–9. 14. Bonnarens F, Einhorn TA. Production of a standard closed fracture in laboratory animal bone. J Orthop Res. 1984;2:97–101. 37. Marklund N. Rodent models of traumatic brain injury: methods and challenges. Methods Mol Biol. 2016;1462:29–46. 14. Bonnarens F, Einhorn TA. Production of a standard closed fractur laboratory animal bone. J Orthop Res. 1984;2:97–101. 15. Suresh MV, Ramakrishnan SK, Thomas B, Machado-Aranda D, Bi Y, Talarico N, Anderson E, Yatrik SM, Raghavendran K. Activation of hypoxia-inducible factor-1α in type 2 alveolar epithelial cell is a major driver of acute inflammation following lung contusion. Crit Care Med. 2014;42(10):642–53. 38. Nunez TC, Cotton BA. Transfusion therapy in hemorrhagic shock. Curr Opin Crit Care. 2009;15(6):536–41. 39. Frink M, Andruszkow H, Zeckey C, Krettek C, Hildebrand F. Experimental Trauma Models: An Update. J Biomed Biotechnol. 2011;2011:797383. 16. Dolgachev VA, Yu B, Sun L, Shanley TP, Raghavendran K, Hemmila MR. Interleukin 10 overexpression alters survival in the setting of gram-negative pneumonia following lung contusion. Shock. 2014;41(4):301–10. 40. Pfeifer R, Lichte P, Schreiber H, Sellei RM, Dienstknecht T, Sadeghi C, Pape HC, Kobbe P. Authors’ contributions Neunaber C, Oestern S, Andruszkow H, Zeckey C, Mommsen P, Kutter D, Stöfen M, Krettek C, Hildebrand F. Cytokine productive capacity of alveolar macrophages and Kupffer cells after femoral fracture and blunt chest trauma in a murine trauma model. Immunol Lett. 2013;152(2):159–66. 48. Cheung KM, Kaluarachi K, Andrew G, Lu W, Chan D, Cheah KS. An externally fixed femoral fracture model for mice. J Orthop Res. 2003;21(4):685–90. 49. Bone LB, Giannoudis P. Femoral shaft fracture fixation and chest injury after Polytrauma. J Bone Joint Surg Am. 2011;93:311–7. 24. Stockmann U, Roscher R. Pulmonary contusion after blunt chest trauma: an experimental model. Thoraxchir Vask Chir. 1977;25:211–3. 50. Riska EB, von Bonsdorff H, Hakkinen S, Jaroma H, Kiviluoto O, Paavilainen T. Prevention of fat embolism by early internal fixation of fractures in patients with multiple injuries. Injury 1976, 8: 110–116. 25. Tower RJ, Campbell GM, Müller M, Will O, Glüer CC, Tiwari S. Binding kinetics of a fluorescently labeled bisphosphonate as a tool for dynamic monitoring of bone mineral deposition in vivo. J Bone Miner Res. 2014;29(9):1993–2003. 51. Schreiber VM, Tarkin IS, Hildebrand F, Darwiche S, Pfeifer R, Chelly J, Giannoudis P, Pape HC. The timing of definitive fixation for major fractures in polytrauma-a matched-pair comparison between a US and European level I centres: analysis of current fracture management practice in polytrauma. Injury. 2011;42(7):650–4. 26. Tiwari S, Schem C, Lorenzen AC, Kayser O, Wiese C, Graeff C, Peña J, Marshall RP, Heller M, Kalthoff H, Jonat W, Glüer CC. Application of ex vivo micro-computed tomography for assessment of in vivo fluorescence and plain radiographic imaging for monitoring bone metastases and osteolytic lesions. J Bone Miner Metab. 2012;30(3):373–80. 52. Roberts CS, Pape HC, Jones AL, Malkani AL, Rodriguez JL, Giannoudis PV. Damage control orthopaedics: evolving concepts in the treatment of patients who have sustained orthopaedic trauma. Instr Course Lect. 2005;54:447–62. 27. Jackson IL, Vujaskovic Z, Down JD. Revisiting strain-related differences in radiation sensitivity of the mouse lung: recognizing and avoiding the confounding effects of pleural effusions. Radiat Res. 2010;173(1):10–20. 53. Chawda MN, Hildebrand F, Pape HC, Giannoudis PV. Predicting outcome after multiple trauma: which scoring system? Injury. 2004;35(4):347–58. 53. Chawda MN, Hildebrand F, Pape HC, Giannoudis PV. Predicting outcome after multiple trauma: which scoring system? Injury. 2004;35(4):347–58. 54. Clark GC, Schecter WP, Trunkey DD. Variables affecting outcome in blunt chest trauma: flail chest vs. pulmonary contusion. J Trauma. 1988;28(3):298–304. 28. Authors’ contributions Hildebrand F, Giannoudis PV, Mv G, Zelle B, Ulmer B, Krettek C, Bellamy MC, Pape HC. Management of polytraumatized patients with associated blunt chest trauma: a comparison of two European countries. Injury. 2005;36(2):293–302. 54. Clark GC, Schecter WP, Trunkey DD. Variables affecting outcome in blunt chest trauma: flail chest vs. pulmonary contusion. J Trauma. 1988;28(3):298–304. 29. Ganie FA, Lone H, Lone GN, Wani ML, Singh S, Dar AM, Wani N, Wani SN, Nazeer N. Lung contusion: a Clinico-pathological entity with unpredictable clinical course. Bull Emerg Trauma. 2013;1(1):7–16. 55. Hoth JJ, Hudson WP, Brownlee NA, Yoza BK, Hiltbold EM, Meredith JW, CE MC. Toll-like receptor 2 participates in the response to lung injury in a murine model of pulmonary contusion. Shock. 2007;28(4):447–52. 30. Hildebrand F, Giannoudis P, van Griensven M, Chawda M, Probst C, Harms O. Secondary effects of femoral instrumentation on pulmonary physiology in a standardised sheep model: what is the effect of lung contusion and reaming. Injury 2005; 36: 544–555. 56. Raghavendran K, Davidson B, Woytash JA, Helinski JD, Marschke CJ, Manderscheid P, Notter R, Knight P. The evolution of isolated bilateral lung contusion from blunt chest trauma in rats: cellular and cytokine responses. Shock. 2005;24(2):132–8. 31. Mestas J, Hughes C. Of mice and not men: differences between mouse and human immunology. J Immunol. 2004;172:2731–8. 57. Perl M, Gebhard F, Braumüller S, Tauchmann B, Brückner U, Kinzl L, Knöferl M. The pulmonary and hepatic immune microenvironment and its contribution to the early systemic inflammation following blunt chest trauma. Crit Care Med. 2006;34(4) 32. Tsukamoto T, Pape HC. Animal models for trauma research: what are the options? Shock. 2009;31(1):3110. 33. Stortz JA, Raymond SL, Mira JC, Moldawer LL, Mohr AM, Efron PA. Murine models of sepsis and trauma: can we bridge the gap? ILAR J. 2017;58(1):90–105. 34. Carbonell WS, Maris DO, McCall T, Grady MS. Adaptation of the fluid i i j d l h J N 1998 15(3) 217 29 58. Genton L, Pichard C. Protein catabolism and requirements in severe illness. Int J Vitam Nutr Res. 2011;81(2–3):143–52. 33. Stortz JA, Raymond SL, Mira JC, Moldawer LL, Mohr AM, Efron PA. Murine models of sepsis and trauma: can we bridge the gap? ILAR J. 2017;58(1):90–105. 59. Omerbegović M, Durić A, Muratović N, Mulalić L, Hamzanija E. Metabolic response to trauma and stress. Med Arh. 2003, 57;4(1):57–60. 34. Carbonell WS, Maris DO, McCall T, Grady MS. Fitschen-Oestern et al. BMC Musculoskeletal Disorders (2017) 18:468 Authors’ contributions Adaptation of the fluid percussion injury model to the mouse. J Neurotrauma. 1998;15(3):217–29. 35. Smith DH, Soares HD, Pierce JS, Perlman KG, Saatman KE, Meaney DF, Dixon CE, McIntosh TKA. Model of parasagittal controlled cortical impact in the mouse: cognitive and histopathologic effects. J Neurotrauma. 1995;12(2):169–78. 60. Mommsen P, Zeckey C, Frink M, Krettek C, Hildebrand F. Accidental hypothermia in multiple trauma patients. Zentralbl Chir. 2012;137(3):264-9. 61. Gentile LF, Nacionales DC, Cuenca AG, Armbruster M, Ungaro RF, Abouhamze AS, Lopez C, Baker HV, Moore FA, Ang DN, Efron PA. Page 12 of 12 Fitschen-Oestern et al. BMC Musculoskeletal Disorders (2017) 18:468 Identification and description of a novel murine model for polytrauma and shock. Crit Care Med. 2013;41(4):1075–85. Identification and description of a novel murine model for polytrauma and shock. Crit Care Med. 2013;41(4):1075–85. Identification and description of a novel murine model for polytrauma and shock. Crit Care Med. 2013;41(4):1075–85. 62. Bone RC. Towards a theory regarding the pathogenesis of the systemic inflammatory response syndrome: what we do and do not know about cytokine regulation. Crit Care Med. 1996;24(1):163–72. cytokine regulation. Crit Care Med. 1996;24(1):163–72. 63. Pfeifer R, Darwiche S, Kohut L, Billiar T, Pape HC. Cumulative effects of bone and soft tissue injury on systemic inflammation: a pilot study. Clin Orthop Relat Res. 2013;471(9):2815–21. 64. Akkose S, Ozgurer A, Bulut M, Koksal O, Ozdemir F, Ozguc H. Relationships between markers of inflammation, severity of injury, and clinical outcomes in hemorrhagic shock. Adv Ther. 2007;24:955–62. 65. Jaffer U, Wade RG, Gourlay T. Cytokines in the systemic inflammatory response syndrome: a review. HSR Proc Intensive Care Cardiovasc Anesth. 2010;2(3):161–75. 66. Roumen RM, Hendriks T, van der Ven-Jongekrijg J, Nieuwenhuijzen GA, Sauerwein RW, van der Meer JW, Goris RJ. Cytokine patterns in patients after major vascular surgery, hemorrhagic shock, and severe blunt trauma. Relation with subsequent adult respiratory distress syndrome and multiple organ failure. Ann Surg. 1993;218(6):769–76. 67. Seitz DH, Niesler U, Palmer A, Sulger M, Braumüller ST, Perl M, Gebhard F, Knöferl MW. Blunt chest trauma induces mediator-dependent monocyte migration to the lung. Crit Care Med. 2010;38(9):1852–9. 68. Köller M, Dávid A, Hahn MP, Muhr G. Liberation of interleukin 12 (IL12) after trauma and polytrauma. Langenbecks Arch Chir Suppl Kongressbd. 1998;115(1):453–6. 69. Li J, Li NP, YF G, Yang X, XB L, Cong JN, Ling Y, Tang JA, Yuan XY, Wang H. Fitschen-Oestern et al. BMC Musculoskeletal Disorders (2017) 18:468 Authors’ contributions Dynamic activity of NF-kappaB in multiple trauma patients and protective effects of ulinastain. Chin J Traumatol. 2011;14(6):354–8. • We accept pre-submission inquiries • Our selector tool helps you to ﬁnd the most relevant journal • We provide round the clock customer support • Convenient online submission • Thorough peer review • Inclusion in PubMed and all major indexing services • Maximum visibility for your research Submit your manuscript at www.biomedcentral.com/submit Submit your next manuscript to BioMed Central and we will help you at every step: • We accept pre-submission inquiries • Our selector tool helps you to ﬁnd the most relevant journal • We provide round the clock customer support • Convenient online submission • Thorough peer review • Inclusion in PubMed and all major indexing services • Maximum visibility for your research Submit your manuscript at www.biomedcentral.com/submit Submit your next manuscript to BioMed Central and we will help you at every step: • We accept pre-submission inquiries • Our selector tool helps you to ﬁnd the most relevant journal • We provide round the clock customer support • Convenient online submission • Thorough peer review • Inclusion in PubMed and all major indexing services • Maximum visibility for your research Submit your manuscript at www.biomedcentral.com/submit Submit your next manuscript to BioMed Central and we will help you at every step: Submit your next manuscript to BioMed Central and we will help you at every step:
https://openalex.org/W3001321088	https://europepmc.org/articles/pmc6996127?pdf=render	English	null	The comparison of plasma fibronectin in term and preterm delivery: A cross-sectional, descriptive-analytical study	International journal of reproductive biomedicine	2,020	cc-by	5,371	The comparison of plasma fibronectin in term and preterm delivery: A cross-sectional, descriptive-analytical study Zahra Moradi1 M.Sc., Parvin Moradi2 M.D., Mohamad Hassan Meshkibaf3 Ph.D., Mehrnoosh Aleosfoor1 B.Sc., Mehdi Sharafi4 M.Sc., Saeedeh Jafarzadeh1 M.Sc. 1School of Nursing, Fasa University of Medical Sciences, Fasa, Iran. 1School of Nursing, Fasa University of Medical Sciences, Fasa, Iran. 2Department of Obstetrics and Gynecology, Medical School, Fasa University of Medical Sciences, Fasa, Iran. g, y , , 2Department of Obstetrics and Gynecology, Medical School, Fasa University of Medical Sciences, Fasa, Iran. 2Department of Obstetrics and Gynecology, Medical School, Fasa University of Medical Sciences, F I 3Department of Clinical Biochemistry, Fasa University of Medical Sciences, Fasa, Iran. 4N C i bl Di R h C t F U i it f M di l S i F I 3Department of Clinical Biochemistry, Fasa University of Medical Sciences, Fasa, Iran. 4Non-Communicable Diseases Research Center, Fasa University of Medical Sciences, Fasa, Iran. Corresponding Author: Saeedeh Jafarzadeh; School of Nursing, Fasa University of Medical Sciences, Fasa, Iran. Postal Code: 0987461686688 Tel: (+98) 9171875089 Email: saeedeh.jafarzadeh@yahoo.com Received 13 December 2018 Revised 6 July 2019 Accepted 14 August 2019 Production and Hosting by Knowledge E Moradi et al. This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited. Editor-in-Chief: Aflatoonian Abbas M.D. Abstract Background: Preterm delivery is one of the main causes of infant death. Therefore, prediction of preterm delivery may eliminate a large number of prenatal complications. Objective: The present study aimed to understand if preterm delivery can be predicted by assessing maternal plasma fibronectin concentration. Background: Preterm delivery is one of the main causes of infant death. Therefore, prediction of preterm delivery may eliminate a large number of prenatal complications. Objective: The present study aimed to understand if preterm delivery can be predicted by assessing maternal plasma fibronectin concentration. Materials and Methods: Serum samples from 105 pregnant women participating in this study were collected. The plasma fibronectin were measured at 24-28 wk of gestation and again at 32-36 wk of gestation. Unfortunately, only 65 of the 105 pregnant women, returned for the second sampling. The plasma fibronectin was analyzed using ELISA method and its concentration in term and preterm deliveries was compared. The delivery dates of all the women were also recorded. Results: Out of 105 pregnant women, 28 delivered preterm (26.7%). The Plasma fibronectin concentrations in women with preterm delivery were higher than in those who delivered at term (p = 0.001). Accordingly, Plasma fibronectin concentrations were significantly higher in the second serum samples (p = 0.01). Plasma fibronectin concentrations was also higher in obese women and in those suffering from preeclampsia (p = 0.12) and gestational diabetes (p = 0.81). Conclusion: Plasma fibronectin concentrations test could be used as an optional screening test for preterm delivery at 28 to 34 wk of gestation in pregnant women who prefer to avoid vaginal sampling. Materials and Methods: Serum samples from 105 pregnant women participating in this study were collected. The plasma fibronectin were measured at 24-28 wk of gestation and again at 32-36 wk of gestation. Unfortunately, only 65 of the 105 pregnant women, returned for the second sampling. The plasma fibronectin was analyzed using ELISA method and its concentration in term and preterm deliveries was compared. The delivery dates of all the women were also recorded. Results: Out of 105 pregnant women, 28 delivered preterm (26.7%). The Plasma fibronectin concentrations in women with preterm delivery were higher than in those who delivered at term (p = 0.001). Accordingly, Plasma fibronectin concentrations were significantly higher in the second serum samples (p = 0.01). Key words: Premature birth, Fibronectin, Maternal serum screening tests. How to cite this article: Moradi Z, Moradi P, Meshkibaf MH, Aleosfoor M, Sharafi M, Jafarzadeh S. “The comparison of plasma fibronectin in term and preterm delivery: A cross-sectional, descriptive-analytical study,” Int J Reprod BioMed 2020; 18: 11–20. https://doi.org/10.18502/ijrm.v18i1.6191 Page 11 International Journal of Reproductive BioMedicine Volume 18, Issue no. 1, https://doi.org/10.18502/ijrm.v18i1.6191 Production and Hosting by Knowledge E Research Article The comparison of plasma fibronectin in term and preterm delivery: A cross-sectional, descriptive-analytical study Zahra Moradi1 M.Sc., Parvin Moradi2 M.D., Mohamad Hassan Meshkibaf3 Ph.D., Mehrnoosh Aleosfoor1 B.Sc., Mehdi Sharafi4 M.Sc., Saeedeh Jafarzadeh1 M.Sc. 1School of Nursing, Fasa University of Medical Sciences, Fasa, Iran. 2Department of Obstetrics and Gynecology, Medical School, Fasa University of Medical Sciences, Fasa, Iran. 3Department of Clinical Biochemistry, Fasa University of Medical Sciences, Fasa, Iran. 4N C i bl Di R h C F U i i f M di l S i F I Research Article The comparison of plasma fibronectin in term and preterm delivery: A cross-sectional, descriptive-analytical study Abstract Plasma fibronectin concentrations was also higher in obese women and in those suffering from preeclampsia (p = 0.12) and gestational diabetes (p = 0.81).i Production and Hosting by Knowledge E Conclusion: Plasma fibronectin concentrations test could be used as an optional screening test for preterm delivery at 28 to 34 wk of gestation in pregnant women who prefer to avoid vaginal sampling. Editor-in-Chief: Aflatoonian Abbas M.D. How to cite this article: Moradi Z, Moradi P, Meshkibaf MH, Aleosfoor M, Sharafi M, Jafarzadeh S. “The comparison of plasma fibronectin in term and preterm delivery: A cross-sectional, descriptive-analytical study,” Int J Reprod BioMed 2020; 18: 11–20. https://doi.org/10.18502/ijrm.v18i1.6191 Page 11 International Journal of Reproductive BioMedicine Moradi et al. 1. Introduction could be effective in the prognosis of preterm delivery. Plasma fibronectin concentration follows an ascending trend in high-risk pregnancies (13). Assessment of Plasma fibronectin concentration is a simple, non-invasive, and accurate method for investigation of endothelial function (11). In Iran, pregnant women are reluctant to undergo vaginal examination during pregnancy, believing it is hazardous to infant health and may itself cause preterm delivery. The availability of another sampling option may help the patient as well as the physician to evaluate the chance of preterm delivery. Because blood sampling is simpler and less invasive than vaginal sampling, the purpose of the present study is to evaluate and compare Plasma fibronectin concentrations in women that delivered preterm and at term?” Preterm delivery is defined as delivery before 37 wk of gestation or 259 days after the last menstrual period (LMP) (1, 2). It is one of the main reasons for infant mortality in the world, accounting for 35% of infant deaths annually (3, 4). Preterm delivery is still responsible for 70% of neurological complications, disabilities, and deaths, which impose a substantial economic healthcare burden (5, 6). Several methods are available for diagnosing, including investigation of cervical dilation, assessment of cervical activity with a dynamometer, performing sonography to determine gestational age, evaluation of cervical length using abdominal or vaginal ultrasound, and evaluation of vaginal fibronectin levels and salivary estriol (2). However, identification by a reliable screening marker can predict preterm delivery before the incidence of clinical symptoms. Vaginal fibronectin concentration is a reliable marker in the prognosis of preterm delivery. However, its effectiveness may be limited because of anxiety and cultural misgivings among pregnant women about the method of sampling, fibronectin secretion increases due to endothelial stress (7-10). Thus, in the case of preterm delivery, preeclampsia, and intrauterine growth restriction (IUGR), fibronectin concentration increases weeks and even months before delivery. It is rarely found in discharges after the 21st wk of gestation. However, it increases again before delivery. Thus, the early presence of fibronectin in cervical and vaginal discharges can predict preterm delivery (11, 12). 2.2. Statistical analysis All data were analyzed using SPSS statistical software (IBM Corp. Released 2013. IBM SPSS Statistics for Windows, Version 22.0. Armonk, NY: IBM Corp.”). Descriptive statistics such as mean, standard deviation (SD), and percentage were used. The chi-square independent and paired t-tests and a logistic regression analysis were also employed. In addition, the receiver operating characteristic (ROC) curve was used to assess the cutoff points. The significance level was set at 𝛼< 0.05. 2. Materials and Methods This cross-sectional, descriptive-analytical study was conducted in 105 pregnant women after considering the inclusion and exclusion criteria. The pregnant women had been referred to the gynecology clinic in Vali-e-Asr Hospital, Fasa, Iran for routine prenatal care between March 2014 and April 2016. The inclusion criteria included age (18-35 years old), single pregnancy, gestational age of 24-28 and 32-36 wk; the exclusion criteria included a history of receiving tocolytic agents (terbutaline, ritodrine, magnesium sulfate, salbutamol, and isoxsuprine) and suffering from chronic hypertension, diabetes, renal problems, and inflammatory disorders such as lupus. The gestational age was determined according to LMP. In case of doubt about the date of LMP, the first-trimester sonography was taken into consideration. Demographic data (parents’ ages, occupations, and education levels) and obstetric information were collected for each participant. For fetal fibronectin test, most samples are taken through the vaginal route when the process of preterm delivery begins. Because fibronectin concentration increases earlier in maternal plasma compared with vaginal discharges, Zygmunt proposed that high Plasma fibronectin levels https://doi.org/10.18502/ijrm.v18i1.6191 Page 12 Plasma fibronectin in term and preterm delivery International Journal of Reproductive BioMedicine deliveries with respect to the mother’s age, father’s age, and the interval between marriage and first pregnancy (p = 0.51). There were also no significant differences between the women with term and preterm deliveries with respect to occupation, education level, consanguineous (family) marriage, parity, history of miscarriage, husband’s cigarette smoking status, family history of preterm delivery, contraception method, and utilization of assisted reproductive techniques (ART) (p = 0.49). However, the body mass index (BMI) before pregnancy was higher among women in the preterm delivery group (p = 0.01), which continued throughout pregnancy. Higher fibronectin concentration was accompanied by lower infant weight (p = 0.001) (Table I). A venous blood sample was collected, serum separated, and kept at 𝜇70°C. All the samples were assessed for fibronectin concentrations using 96- well enzyme-linked immunosorbent assay (ELISA) kits. The first sampling was undertaken at 24-28 wk of gestation and second sampling at 32-36 wk of gestation. However, samples could only be taken from 65 women during the second sampling period. All participants were followed until delivery and information about infants’ weight, type of delivery, and gestational age at the time of delivery was collected and recorded. Plasma fibronectin concentrations were assessed using ELISA kits (BE59341 IBL international GMBH). 2.1. Ethical considerations The results revealed that 28 participants had preterm delivery (26.7%). Among the women with preterm delivery, 13 had natural vaginal delivery (NVD) (46.4%) and 15 had a cesarean section(C/S) (53.6%). These measures (NVD or C/S) were 57.1% and 42.9%, respectively, in the term delivery group (p = 0.33). According to the results, two women in the preterm delivery group (7.1%) and three women in the term delivery group (3.9%) had unplanned pregnancies (p = 0.009). Besides, 53.6% of the women in the preterm delivery group and 31.2% of those in the term delivery group had a family history of diabetes and hypertension (p = 0.03). Additionally, 17.9% of the women in the preterm delivery group and 10.4% of those in the term delivery group reported a history of preterm delivery in their mothers or sisters. However, the difference was not statistically significant (p = 0.30) (Table I). Out of 65 women who took part in the second sampling, 11 had preterm delivery and 54 had term delivery. At the first sampling (24−28 wk), the results of fibronectin concentrations revealed no significant difference (p = 0.66). Plasma fibronectin concentrations in the term delivery group decreased in the second sampling, but the difference was not statistically significant (p Women were informed about the study and those who were interested to join, provided written informed consents. This article was extracted from a research proposal (93104) approved by Fasa University of Medical Sciences. 3. Results The results showed no significant differences between the women with term and preterm https://doi.org/10.18502/ijrm.v18i1.6191 Page 13 International Journal of Reproductive BioMedicine Moradi et al. and preeclampsia (p = 0.009) (Tables IV and V). = 0.16). However, in contrast, the mean plasma fibronectin concentration significantly increased in the preterm delivery group (p = 0.01). In other words, the increase in gestational age was accompanied by a significant increase in the fibronectin concentration in the preterm delivery group (p = 0.001) (Table II and III). With respect to pregnancy complications, the incidence of diabetes and hypertension was higher in women in the preterm delivery group. Moreover, plasma fibronectin concentrations were significantly higher in the women suffering from gestational diabetes According to the ROC curve, the best cutoff point in the first sampling was 1.750 ng/ml with a sensitivity of 80.26%, specificity of 17.85%, positive predictive value of 73/08%, and negative predictive value of 26/92%. In the second sampling, the best cutoff point was 4,226 ng/ml with a sensitivity of 100%, specificity of 74%, positive predictive value of 76/67%, and negative predictive value of 23/33%. According to these measures, this test should be used after the 28th wk of gestation (Figures 1, 2). Table I. 3. Results Frequency distribution and comparison of demographic, labor, and individual factors in two groups Personal and delivery details, and maternal factors Preterm N = 28 (26.7%) Term N = 77 (73.3%) Total N = 150 (100%) P-value Mother’s age (yr) * 29.47 ± 5.56 28.57 ± 6.50 28.81 ± 6.257 0.51∗ Husband’s age (yr)* 35.39 ± 7.28 33.45 ± 6.03 33.97 ± 6.417 0.17 The average age of marriage for women (yr) * 22.21 ± 5.39 21.70 ± 4.66 21.84 ± 4.84 0.63 Time span between the age of marriage and the first pregnancy (yr) * 2.25 ± 1.79 2.57 ± 3.20 2.49 ± 2.84 0.63 Gestational age in the first sampling (wk) * 26.45 ± 2.68 26.16 ± 2.33 26.1 ± 2.98 0.87 Gestational age in the second sampling (wk) * 31.18 ± 3.10 32.33 ± 2.74 30.90 ± 2.35 0.42 Gestational age at birth (wk) * 35.12 ± 1.10 39.25 ± 2.13 38.24 ± 2.03 0.012 Mother’s weight at first sampling (kg) * 65.3 ± 10.56 60.42 ± 10.74 61.72 ± 10.867 0.42 Maternal weight at delivery (Kg) * 74.25 ± 12.33 67.54 ± 12.24 69.33 ± 12.566 0.17 Maternal BMI before pregnancy * 26.24 ± 346 24.18 ± 3.88 24.73 ± 3.87 0.01 Delivery type NVD 13 (46.4) 44 (57.1) C/S 15 (53.6) 37 (42.9) 0.33∗∗ Mother’s job Housewife 22 (78.6) 71 (92.2) Employed 6 (21.4) 6 (7.8) 0.15 Husband’s job The manual worker 2 (7.1) 16 (20.8) Employee 8 (28.6) 16 (20.8) Freelancer 18 (64.3) 45 (58.4) 0.34 Mother’s education Illiterate 2 (7.1) 1 (1.3) Under the diploma 7 (15) 27 (35.1) Academic 19 (67.8) 49 (63.7) 0.37 n and comparison of demographic, labor, and individual factors in two groups Table I. Frequency distribution and comparison of demographic, labor, and individual factors in two groups Page 14 https://doi.org/10.18502/ijrm.v18i1.6191 https://doi.org/10.18502/ijrm.v18i1.6191 https://doi.org/10.18502/ijrm.v18i1.6191 Plasma fibronectin in term and preterm delivery International Journal of Reproductive BioMedicine International Journal of Reproductive BioMedicine Table I. Continued. Personal and delivery details, and maternal factors Preterm N = 28 (26.7%) Term N = 77 (73.3%) Total N = 150 (100%) P-value Husband’s education Illiterate 1 (3.6) 5 (6.5) Under the diploma 15 (53.6) 53 (68.8) Academic 12 (42.8) 19 (24.7) 0.08 Gravid 1 13 (6.4) 27 (35.1) 2 8 (28.6) 27 (35.1) 3≥ 7 (25) 23 (29.8) 0.70 Abortion history Yes 4 (14.3) 19 (24.7) No 24 (85.7) 58 (75.3) 0.25 Smoking by husband Yes 6 (21.4) 14 (18.2) No 22 (78.6) 63 (81.8) 0.70 Preterm labor history Yes 23 (82.1) 69 (89.6) No 5 (17.9) 8 (10.4) 0.30 Use of fertility assisted methods Yes 2 (7.1) 3 (3.9) No 26 (92.9) 74 (96.1) 0.49 Disease in the current pregnancy ** GDM 4 (14.28) 1 (1.31) PEC 4 (14.28) 2 (2.63) Infection 7 (25) 19 (25) Vaginal bleeding 1 (3.57) 10 (13.15) 0.009 * Mean ± SD, *n (%), P-value∗, Qualitative variables were analyzed using an independent t-test, P-value∗∗, Chi-square test NVD: Normal vaginal delivery, C/S: Cesarean section, GDM: Gestational diabetic mellitus, PEC: Preeclampsia Mean ± SD, *n (%), P-value∗, Qualitative variables were analyzed using an independent t-test, P-value∗∗, Chi-square test NVD: Normal vaginal delivery, C/S: Cesarean section, GDM: Gestational diabetic mellitus, PEC: Preeclampsia Mean ± SD, *n (%), P-value∗, Qualitative variables were analyzed using an independent t-test, P-value∗∗, Chi-square test NVD: Normal vaginal delivery, C/S: Cesarean section, GDM: Gestational diabetic mellitus, PEC: Preeclampsia Table II. Comparison of mean plasma concentrations of fibronectin in two groups in double sampling Variable Fibronectin plasma1 Fibronectin plasma2 P-value∗ Term 7727.50 ± 11506.734 4059.57 ± 3175.744 0.16 Preterm 6726.43 ± 7174.916 10507.14 ± 5477.106 0.01 P-value∗∗ 0.668 0.001 Mean ± SD, Qualitative variables were analyzed using an independent and dependent t-test and quantitative variables were analyzed using logistic regression test, P-value∗, paired sample t-test; P-value∗∗, independent t-test Table II. International Journal of Reproductive BioMedicine Comparison of mean plasma concentrations of fibronectin in two groups in double sampling * Mean ± SD, Qualitative variables were analyzed using an independent and dependent t-test and quantitative variables were analyzed using logistic regression test, P-value∗, paired sample t-test; P-value∗∗, independent t-test * Mean ± SD, Qualitative variables were analyzed using an independent and dependent t-test and quantitative variables were analyzed using logistic regression test, P-value∗, paired sample t-test; P-value∗∗, independent t-test https://doi.org/10.18502/ijrm.v18i1.6191 Page 15 International Journal of Reproductive BioMedicine Moradi et al. Table III. Comparison of mean of fibronectin concentrations in the first and second sampling groups Variable N Mean ± SD p-value Fibronectin plasma1 30 4366.66 ± 5209.09 Fibronectin plasma2 30 5564 ± 4642.26 0.009 * Mean ± SD, P-value, paired sample t-test Table III. Comparison of mean of fibronectin concentrations in the first and second sampling groups Table III. Comparison of mean of fibronectin concentrations in the first and second sampling groups * Mean ± SD, P-value, paired sample t-test Table IV. Comparison of fibronectin concentration in women with gestational diabetes and non-diabetes individuals omparison of fibronectin concentration in women with gestational diabetes and non-diabetes individuals Table IV. Comparison of fibronectin concentration in women with gestational diabetes and non-diabetes indiv Table IV. Comparison of fibronectin concentration in women with gestational diabetes and non-diabetes individuals Fibronectin GDM Fibronectin plasma 1 Fibronectin plasma 2 Mean ± SD N Mean ± SD N Yes 6508.75 ± 5175.02 8 6166.67 ± 3663.78 3 No 7537.08 ± 10831.50 96 5497.04 ± 4791.46 27 P-value 0.791 0.817 * Mean ± SD, P-value: Independent t-test, GDM: Gestational diabetic mellitus Fibronectin plasma 2 Mean ± SD N 6166.67 ± 3663.78 3 5497.04 ± 4791.46 27 0.817 * Mean ± SD, P-value: Independent t-test, GDM: Gestational diabetic mellitus Table V. Comparison of fibronectin concentration in women with preeclampsia and non-affected patients Fibronectin Preeclampsia Fibronectin plasma 1 Fibronectin plasma 2 Mean ± SD N Mean ± SD N Yes 9581.67 ± 7356.95 6 12600 1 No 7327.96 ± 10669.99 96 5321.38 ± 4526.73 29 P-value 0.612 0.125 * Mean ± SD, P-value: Independent t-test Figure 1. Receiver operating characteristic curve for fibronectin plasma 1. Table V. International Journal of Reproductive BioMedicine Comparison of fibronectin concentration in women with preeclampsia and non-affected patients Fibronectin Preeclampsia Fibronectin plasma 1 Fibronectin plasma 2 Mean ± SD N Mean ± SD N Yes 9581.67 ± 7356.95 6 12600 1 No 7327.96 ± 10669.99 96 5321.38 ± 4526.73 29 P-value 0.612 0.125 * Mean ± SD, P-value: Independent t-test Table V. Comparison of fibronectin concentration in women with preeclampsia and non-affected patients Fibronectin Preeclampsia Fibronectin plasma 1 Fibronectin plasma 2 Mean ± SD N Mean ± SD N Yes 9581.67 ± 7356.95 6 12600 1 No 7327.96 ± 10669.99 96 5321.38 ± 4526.73 29 P-value 0.612 0.125 * Mean ± SD, P-value: Independent t-test Table V. Comparison of fibronectin concentration in women with preeclampsia and non-affected patients Figure 1. Receiver operating characteristic curve for fibronectin plasma 1. https://doi.org/10.18502/ijrm.v18i1.6191 Page 16 International Journal of Reproductive BioMedicine Plasma fibronectin in term and preterm delivery Figure 2. Receiver operating characteristic curve for fibronectin plasma 2. Figure 2. Receiver operating characteristic curve for fibronectin plasma 2. 4. Discussion The results of the present study revealed that an increase in BMI before pregnancy and an increase in the mother’s weight at the time of delivery led to an increase in plasma fibronectin concentration. In contrast, an early increase in plasma fibronectin concentration leads to preterm delivery. Moreover, five women developed gestational diabetes and six developed preeclampsia. Of these, eight in total underwent preterm delivery. Fibronectin concentration was also higher in women with gestational diabetes and hypertension. These results were similar to those obtained by Ekaidem and Rasanen, which showed that plasma fibronectin concentration and the risk of preeclampsia were higher among obese women (14, 16). The relationship between increased plasma fibronectin concentration and risk of preeclampsia in obese women may be attributed to obesity complications. Hypertension in obese individuals can reduce placental perfusion and increase vascular problems, which eventually increase the production and secretion of fibronectin as well as the production of pro- inflammatory products (14, 16). Parker and Kim also demonstrated that maternal obesity was accompanied by increased risk of preterm delivery due to medical complications such as gestational diabetes and blood pressure disorders. Thus, interventional strategies for reducing the risk of gestational diabetes and blood pressure disorders could decrease the risk of preterm delivery as well (17, 18). Thus, evaluation of mothers’ weight and BMI before pregnancy and at the first visit after that, nutritional consultation, and provision of necessary care services could play an important and lifestyle modifications, as well as weight adjustment, are essential during pregnancy in order to have a healthy delivery and avoid complications and negative outcomes (19-22). The results of the present study showed that plasma fibronectin concentration decreased in the second sampling compared with the first sampling in the term delivery group. Considering the fact that the increase in plasma fibronectin concentration is indicative of preterm delivery, this finding followed the normal trend of changes in fibronectin concentration during pregnancy (10). However, this measure was significantly higher in the second sampling compared with the first sampling in the preterm delivery group. This implies that factors leading to preterm delivery follow an ascending trend and that plasma fibronectin concentration increases over time. In the studies by Zigmunt and Forouhari (2, 13), plasma fibronectin concentration was assessed in women who showed symptoms of preterm delivery. However, the present study aimed to determine whether plasma fibronectin concentration could be used as a screening test. 4. Discussion the pre-pregnancy value at wk 16th (11). The results of our study are consistent with the findings of the above studies. Thus, it can be concluded that plasma fibronectin concentration follows an ascending trend in high-risk pregnancies. It usually increases weeks and even months before the incidence of complications, such as preterm delivery, preeclampsia, and IUGR (13). Increased plasma fibronectin concentration may result from its secretion from damaged endothelial cells occurring in placentas of women experiencing preterm delivery (13). In this regard, vascular problems and inflammatory markers affect plasma fibronectin concentration. These changes are similar to those occurring in IUGR and preeclampsia (14). In the present study, only 65 out of the 105 participants took part in the second sampling. Among these women, 58 and 7 belonged to the term and preterm delivery groups, respectively. Comparing the results of fibronectin concentration in the first and second sampling for those who attended for the second time showed that an increase in gestational age was accompanied by an increase in The results indicated that plasma fibronectin concentration was higher in the women who had given birth to their children before 37 wk of gestation, which is consistent with the study by Zigmunt (13). Forouhari and colleagues compared maternal plasma fibronectin concentration in three study groups, including women with symptoms and risk factors of preterm delivery, women with symptoms but without risk factors of preterm delivery, and healthy pregnant women. The results showed that plasma fibronectin concentration was significantly higher in the preterm delivery group compared with the term delivery group (2). Severens-Rijvers also evaluated fibronectin concentration before pregnancy and at 12 and 16 wk of gestation. Based on the results, fibronectin concentration was higher in women with maternal placental syndrome and had increased since before pregnancy up to the 16th wk of gestation. However, in women without the maternal placental syndrome, this value increased up to the 12th wk of gestation but decreased below https://doi.org/10.18502/ijrm.v18i1.6191 Page 17 International Journal of Reproductive BioMedicine Moradi et al. fibronectin concentration, which is consistent with studies by Ekaidem and Dane (14, 15). Production and secretion of fibronectin into body fluids is accompanied by the natural growth of placental and trophoblastic villi, which increase with an increase in gestational age (14). 4. Discussion Therefore, the participants were selected from women who had no symptoms of preterm delivery. In addition, blood was taken at two stages with a 4- 6 wk interval in order to determine the best time for performing the test. Zigmunt et al. also reported the highest plasma fibronectin concentration (600 ± 160) in the women who had given birth to their children before 32 wk of gestation (21). In Forouhari’s study, plasma fibronectin concentration was higher in the preterm delivery group compared with the term delivery group, with the best cutoff point being 700 ng/ml. The difference between the results may be because of the kits utilized. Moreover, Forouhari reported sensitivity, specificity, positive predictive value, and negative predictive value to be 100%, 61.1%, 54.3%, and 100%, respectively. The present study benefitted from higher specificity (2). However, https://doi.org/10.18502/ijrm.v18i1.6191 Page 18 Plasma fibronectin in term and preterm delivery International Journal of Reproductive BioMedicine Zigmunt et al. measured fibronectin concentration using the nephelometry method and reported the cutoff point to be 450 ng/ml. Indeed, sensitivity, specificity, positive predictive value, and negative predictive value were reported to be 50%, 95%, 50%, and 90%, respectively (13). As mentioned above, Zigmunt assessed fibronectin concentration using the nephelometry method, while ELISA is a more accurate technique. In the present study, the prevalence of preterm delivery was 26.7%. Various prevalence rates have been reported in different studies. For instance, Forouhari et al. reported the prevalence of preterm delivery to be 31.6% in Shiraz (2). Additionally, Shojae and Shariati showed that the prevalence of the condition was 7.2% in Bojnourd and 15.4% in Tehran and around the world (6, 23). These differences may be attributed to various causes of preterm delivery in different cultural and geographical regions. The findings of the present study demonstrated that an increase in plasma fibronectin concentration resulted in a decrease in birth weight. This decrease may have resulted because of preterm delivery. Similarly, Forouhari and Zigmunt found that an increase in plasma fibronectin concentration caused a decrease in birth weight (2, 13). Given that only two studies have been conducted on the assessment of plasma fibronectin concentration for predicting preterm delivery, limited studies were available for comparison of the results. Moreover, sampling was not undertaken in several stages in the previous studies to evaluate changes in fibronectin concentration in preterm delivery. Thus, further studies may help to achieve more reliable results regarding this issue. Acknowledgments This article was extracted from a research proposal (93104) approved by Fasa University of Medical Sciences. Hereby, the authors would like to thank the Fasa University of Medical Sciences for their financial assistance for this paper, the Clinical Trial Research Center of Vali-e-Asr Hospital, the Dean of the hospital, and all hospital personnel for their cooperation. We are also thankful to the women for their kind participation in the study. References [1] Sultana Z, Maiti K, Aitken J, Morris J, Dedman L, Smith R. Oxidative stress, placental ageing-related pathologies and adverse pregnancy outcomes. Am J Reprod Immunol 2017; 77: doi: 10.1111/aji.12653. [2] Forouhari S, Ghaemi SZ, Azadian M, Parsanezhad ME, Sarvestani E, Jokar A, et al. Predicting preterm delivery by measuring plasma fibronectin concentration. Int J Res Stud Biosci 2014; 2: 1–7. [3] Howson CP, Kinney MV, McDougall L, Lawn JE. Born too soon: preterm birth matters. Reprod Health 2013; 10: S1. [4] James DK, Steer PJ, Weiner CP, Gonik B, Robson SC. High- risk pregnancy: management options. UK, Cambridge University Press; 2017. [5] Halimi Asl AA, Safari S, Parvareshi Hamrah M. Epidemiology and related risk factors of preterm labor as an obstetrics emergency. Emerg 2017; 5: e3. [6] Khoshnood Shariati M, Karimi Z, Rezaienejad M, Basiri A, Torkestani F, Saleh Gargari S. Perinatal complications associated with preterm deliveries at 24 to 33 weeks and 6 days gestation (2011-2012): A hospital-based retrospective study. Iran J Reprod Med 2015; 13: 697–702.i 4. Discussion We suggest the best time for analysis to be after 30 wk of pregnancy. We have undertaken another study to evaluate and compare fetal and plasma fibronectin concentrations at the same time in the same patients. 5. Conclusion Plasma fibronectin concentration can predict the process of preterm delivery earlier than the fetal fibronectin test. Therefore, it can be used as a screening test in the prognosis of preterm delivery. [7] Purisch SE, Gyamfi-Bannerman C. Epidemiology of preterm birth. Seminars in perinatology 2017; 41: 387–391. [8] Martin RJ, Fanaroff AA, Walsh MC. Fanaroff and Martin’s Neonatal-Perinatal Medicine E-Book: Diseases of the Fetus and Infant: Elsevier Health Sciences; 2010. https://doi.org/10.18502/ijrm.v18i1.6191 Page 19 International Journal of Reproductive BioMedicine Moradi et al. of preeclampsia. Am J Obstet Gynecol 2015; 212: 82. e1– e9. [9] van der Krogt L, Ridout AE, Seed PT, Shennan AH. Placental inflammation and its relationship to cervicovaginal fetal fibronectin in preterm birth. Eur J Obstet Gynecol Reprod Biol 2017; 214: 173–177. [17] Parker MG, Ouyang F, Pearson C, Gillman MW, Belfort MB, Hong X, et al. Prepregnancy body mass index and risk of preterm birth: association heterogeneity by preterm subgroups. BMC Pregnancy Childbirth 2014; 14: 153. [10] McLaren JS, Hezelgrave NL, Ayubi H, Seed PT, Shennan AH. Prediction of spontaneous preterm birth using quantitative fetal fibronectin after recent sexual intercourse. Am J Obstet Gynecol 2015; 212: 89. e1– e5. [18] Kim SS, Mendola P, Zhu Y, Hwang BS, Grantz KL. Spontaneous and indicated preterm delivery risk is increased among overweight and obese women without prepregnancy chronic disease. BJOG 2017; 124: 1708– 1716. [11] Severens-Rijvers CAH, Al-Nasiry S, Ghossein-Doha C, Marzano S, Ten Cate H, Winkens B, et al. Circulating fibronectin and plasminogen activator inhibitor-2 levels as possible predictors of recurrent placental syndrome: An exploratory study. Gynecol Obstet Invest 2017; 82: 355– 360. [19] Bahrami Taghanaki H, Hashemian M, Lotfalizadeh M, Noras M. [The relationship between Body Mass Index (BMI) and birth weight and some pregnancy outcomes.] Iran J Obstet Gynecol Infertil 2016; 19: 1–8. (in Persian) [12] Macones GA. Fetal fibronectin testing in threatened preterm labor: time to stop. Am J Obstet Gynecol 2016; 215: 405. [20] Ahmadi Taheri S, Ramazani Ahmadi AH, Javadi M, Barikani A. [Comparison of dietary patterns during pregnancy in the mothers of the infants with low birth weight and normal weight]. Iran J Obstet Gynecol Infertil 2018; 21: 80–89. (in Persian) [13] Zygmunt M, Lang U, Katz N, Künzel W. Maternal plasma fibronectin: a predictor of preterm delivery. Eur J Obstet Gynecol Reprod Biol 1997; 72: 121–126. [21] Salem Z, Ebrahimi F, Aminzadeh F, Asadolahi Z. 5. Conclusion The prevalence of malnutrition and its association with pregnancy outcome among pregnant women in Rafsanjan, Iran, in 2016. J Occup Health Epidemiol 2017; 6: 106– 113. [14] Ekaidem IS, Bolarin DM, Udoh AE, Etuk SJ, Udiong CE. Plasma fibronectin concentration in obese/overweight pregnant women: a possible risk factor for preeclampsia. Indian J Clin Biochem 2011; 26: 187–192. [15] Dane C, Buyukasik H, Dane B, Yayla M. Maternal plasma fibronectin and advanced oxidative protein products for the prediction of preeclampsia in high risk pregnancies: a prospective cohort study. Fetal Diagn Ther 2009; 26: 189–194. [22] Chen X, Zhao D, Mao X, Xia Y, Baker PN, Zhang H. Maternal dietary patterns and pregnancy outcome. Nutrients 2016; 8: 351. [23] Shoja M, Shoja E, Shoja E, Gharaei M. Prevalence and affecting factors on preterm birth in pregnant women Referred to Bentolhoda hospital-Bojnurd. J North Khorasan Univ Med Sci 2016; 7: 855–863. [16] Rasanen J, Quinn MJ, Laurie A, Bean E, Roberts CT Jr, Nagalla SR, et al. Maternal serum glycosylated fibronectin as a point-of-care biomarker for assessment Page 20 Page 20 https://doi.org/10.18502/ijrm.v18i1.6191
https://openalex.org/W4291605540	https://www.cambridge.org/core/services/aop-cambridge-core/content/view/A187A2137606276CB3B3E66906FC47B7/S002237782200068Xa.pdf/div-class-title-theory-of-gradient-drift-instabilities-in-low-temperature-partially-magnetised-plasmas-div.pdf	English	null	Theory of gradient drift instabilities in low-temperature, partially magnetised plasmas	Journal of plasma physics	2,022	cc-by	10,827	Theory of gradient drift instabilities in low-temperature, partially magnetised plasmas Kentaro Hara 1,†, Adnan R. Mansour 1 and Sedina Tsikata 2 1Department of Aeronautics and Astronautics, Stanford University, Stanford, CA 94305, USA 2Centre National de la Recherche Scientiﬁque (CNRS), ICARE UPR 3021, 45071 Orléans, France (Received 25 March 2022; revised 29 June 2022; accepted 30 June 2022) A ﬂuid dispersion theory in partially magnetised plasmas is analysed to examine the conditions under which large-wavelength modes develop in Penning-type conﬁgurations, that is, where an electric ﬁeld is imposed perpendicular to a homogeneous magnetic ﬁeld. The ﬂuid dispersion relation assuming a slab geometry shows that two types of low-frequency, gradient drift instabilities occur in the direction of the E × B and diamagnetic drifts. One type of instability, observed when the equilibrium electric ﬁeld and plasma density gradient are in the same direction, is similar to the classic modiﬁed Simon–Hoh instability. A second instability is found for conditions in which (i) the diamagnetic drift is in the direction opposite to the E × B drift and (ii) the magnitude of the diamagnetic drift is sufﬁciently larger than the electron thermal speed. The present ﬂuid dispersion theory suggests that the rotating spokes driven by such ﬂuid instabilities propagate in the same direction as the diamagnetic drift, which can be in the same direction as or opposite to the E × B drift, depending on the plasma conditions. This ﬁnding may account for the observation, in some plasma devices, of the rotation of large-scale structures in both the E × B and −E × B directions. Key words: plasma devices, plasma dynamics, plasma instabilities J. Plasma Phys. (2022), vol. 88, 905880408 © The Author(s), 2022. Published by Cambridge University Press This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution and reproduction, provided the original article is properly cited. doi:10.1017/S002237782200068X † Email address for correspondence: kenhara@stanford.edu https://doi.org/10.1017/S002237782200068X Published online by Cambridge University Press 1. Introduction Low-temperature magnetised plasmas can be found in a variety of applications and natural phenomena, including magnetron discharges (Keidar & Beilis 2006; Anders 2012; Ito, Young & Cappelli 2015; Hecimovic & von Keudell 2018), Penning discharges (Quraishi, Robertson & Walch 2002; Abolmasov 2012), Hall effect thrusters (HETs) and accelerators (Ellison, Raitses & Fisch 2012; Sekerak et al. 2015; Romadanov et al. 2016; Mazouffre et al. 2019), high-power microwaves (Lau 2001; Benford, Swegle & Schamiloglu 2015), dusty (complex) plasmas (Bal & Bose 2010) and interplanetary and interstellar environments (Breneman et al. 2013). An applied magnetic ﬁeld traps the charged particles, thus increasing ionisation efﬁciency and reducing the diffusivity, serving as a critical path to control ﬂows and chemistry in low-temperature plasmas. https://doi.org/10.1017/S002237782200068X Published online by Cambridge University Press 2 K. Hara, A.R. Mansour and S. Tsikata Depending on the magnetic ﬁeld strength and collision frequency, plasmas can be either partially magnetised (i.e. electrons are magnetised but ions are non-magnetised) or fully magnetised (i.e. both ions and electrons are magnetised). In low-temperature plasmas, the dynamics of the neutral gas also play an important role in the transport properties and time-dependent plasma behaviour. The multiscale nature of the plasma ﬂow in such low-temperature magnetised plasmas leads to various oscillation modes from high frequency (of the order of gigahertz) to low frequency (of the order of kilohertz) (Hara 2019). Recent studies suggest that the anomalous electron transport across magnetic ﬁelds can be due to plasma–wall interaction (Kaganovich et al. 2007; Sydorenko et al. 2008), the plasma waves initiated by kinetic instabilities (Adam, Héron & Laval 2004; Héron & Adam 2013; Boeuf 2017; Janhunen et al. 2018) or a combination of these effects (Héron & Adam 2013; Villafana et al. 2021). A few examples of the kinetic instabilities in the high-frequency range include the electron cyclotron drift instability (Forslund, Morse & Nielson 1971), modiﬁed two-stream instability (McBride et al. 1972), and ion–ion two-stream instability (Gary 1991; Tsikata et al. 2014). On the other hand, low-frequency plasma oscillations include breathing mode oscillations (Boeuf & Garrigues 1998; Barral & Ahedo 2009; Hara et al. 2014a, b; Dale & Jorns 2019) and azimuthally rotating spokes (Ellison et al. 2012; Sekerak et al. 2015; Kawashima, Hara & Komurasaki 2018). The coupling between different instabilities in various spatial and temporal scales plays an important role in determining the transport coefﬁcients and turbulent phenomena in such low-temperature magnetised plasmas. https://doi.org/10.1017/S002237782200068X Published online by Cambridge University Press 2.1. Governing equations The ﬂuid equations are used for both ions and electrons, i.e. when kinetic effects, such as the Bernstein modes (Bernstein 1958), can be neglected. Conservation of mass can be constructed by taking the zeroth moment of the kinetic equation. Here, ionisation and recombination are neglected. Thus, the continuity equation can be written as ∂ns ∂t + ∇· (nsus) = 0, (2.1) ∂ns ∂t + ∇· (nsus) = 0, (2.1) ∂ns ∂t + ∇· (nsus) = 0, (2.1) (2.1) where ns is the number density and us is the bulk velocity for species s. The equation for the ﬂuid momentum can be formulated by taking the ﬁrst moment of the kinetic equation, which can be written using conservative or primitive variables. Assuming that the plasma is collisionless and the distribution function is close to an isotropic Maxwellian distribution function, the conservation of momentum can be written as ∂(msnsus) ∂t + ∇· (msnsusus) = −∇ps + qsns(E + us × B), (2.2) (2.2) where ms is the mass, ps is the pressure, qs is the charge, E is the electric ﬁeld and B is the magnetic ﬁeld. Using the source-less continuity equation, as shown in (2.1), the momentum equation can also be given, using the primitive variables, by where ms is the mass, ps is the pressure, qs is the charge, E is the electric ﬁeld and B is the magnetic ﬁeld. Using the source-less continuity equation, as shown in (2.1), the momentum equation can also be given, using the primitive variables, by ∂us ∂t + (us · ∇)us = −∇ps msns + qs ms (E + us × B). (2.3) (2.3) Note that the pressure is a scalar term, which is valid when the velocity distribution function (VDF) is close to an isotropic Maxwellian distribution function, that is, the temperatures in three directions are equal. Under this condition, the pressure can be written using the ideal gas law: ps = nskBTs, where kB is the Boltzmann constant and Ts is the temperature for species s. Note that the pressure is a scalar term, which is valid when the velocity distribution function (VDF) is close to an isotropic Maxwellian distribution function, that is, the temperatures in three directions are equal. Under this condition, the pressure can be written using the ideal gas law: ps = nskBTs, where kB is the Boltzmann constant and Ts is the temperature for species s. 2. Linearised plasma equations for low-temperature magnetised plasmas Although the rotating spokes are observed in cylindrical (axisymmetric) systems, here we simplify the dispersion relation assuming a slab (Cartesian) geometry. A static, uniform magnetic ﬁeld in z direction B = B0ˆz and an equilibrium electric ﬁeld (applied electric ﬁeld) are considered: E0 = E0ˆx. It is assumed that the equilibrium plasma is quasineutral and a plasma density gradient exists locally in the x direction: E0 ̸= 0 and dn0/dx ̸= 0, generating E × B and diamagnetic drifts in the ±y direction, for the equilibrium condition. In addition, the following assumptions are made: (i) although the temperature gradient may affect the instabilities, we assume that the temperature is uniform and constant, for simplicity; (ii) only ions and electrons are assumed for the species; (iii) the plasma is assumed to be partially magnetised (i.e. magnetised electrons and unmagnetised ions); (iv) collisions are assumed negligible; and (v) the plasma is electrostatic, that is, effects of the induced magnetic ﬁeld are negligible compared with the applied magnetic ﬁeld. https://doi.org/10.1017/S002237782200068X Published online by Cambridge University Press 1. Introduction p g p In this paper, we derive the dispersion relation of low-frequency, large-wavelength gradient drift instabilities, which may lead to the self-organising patterns in the low-temperature magnetised plasmas. We consider a Penning-type conﬁguration, where a homogeneous axial magnetic ﬁeld is applied and an electric ﬁeld and density gradient exist in the radial direction. Although the spokes are often observed in numerical simulations (Boeuf 2014; Powis et al. 2018; Boeuf & Takahashi 2020) and in experiments (Raitses, Kaganovich & Smolyakov 2015; Marcovati, Ito & Cappelli 2020), the mechanism of the spoke formation is still not well understood. One of the most accepted theories attributes the formation of rotating spokes in a Penning-type discharge to the class of Simon–Hoh instabilities (SHIs) (Simon 1963; Hoh 1963). This terminology was ﬁrst used by Sakawa et al. (1993), who proposed the modiﬁed Simon–Hoh instability (MSHI), which comes from a dispersion relation for partially magnetised plasma, whereas the original work by Simon and Hoh individually focused on fully magnetised plasmas. In work by Simon (1963), the criterion for rotating spoke formation has been proposed to be E0 · ∇n0 > 0, where E0 is the applied electric ﬁeld and ∇n0 is the equilibrium plasma density gradient. Recent experimental studies in low-temperature magnetised plasma sources have shown, however, that the rotation direction and speed can depend on various plasma parameters, such as the current (Anders & Yang 2017), indicating that the instability criterion needs to be revisited. The theory of gradient-drift instabilities in low-temperature magnetised plasmas is introduced in this paper. Section 2 discusses the linearised ﬂuid equations for magnetised plasmas under an applied magnetic ﬁeld, assuming that the gyroviscosity effects are negligible. Section 3 shows a low-frequency plasma dispersion relation of the partially magnetised plasma considering a slab geometry. The criteria for the large-wavelength, low-frequency, gradient drift instability under a homogeneous magnetic ﬁeld are discussed. Section 4 illustrates the results of the dispersion relation for partially magnetised plasmas, showing results consistent with the instability criteria derived in § 3. The gradient drift instability theory is applied to various cross-ﬁeld plasma devices in § 5. https://doi.org/10.1017/S002237782200068X Published online by Cambridge University Press Fluid theory of gradient drift instabilities 3 2. Linearised plasma equations for low-temperature magnetised plasmas 2.2. Linear perturbation analysis for partially magnetised plasmas We consider the growth of the instabilities in y direction, that is, the direction in which the electrons drift. Under the linear perturbation analysis, a plasma property Q can be described as a sum of the steady-state quantity and a linear perturbation, such that Q = Q0 + Q1 exp(−iωt + ikyy), (2.4) (2.4) where Q0 and Q1 are the equilibrium (steady-state) and ﬁrst-order perturbation terms of a plasma property Q, respectively, ω is the frequency, t is time and ky is the wave number in the y direction. Here, ω = ωr + iγ , where ωr is the real frequency and γ is the imaginary part which corresponds to the growth rate. 2.1. Governing equations In the present model, the ideal gas law is assumed for the electron ﬂuid model. It is to be noted that standard drift models account for gyroviscosity effects, which arise due to the non-Maxwellian distribution and lead to cancellation of the diamagnetic drift in the inertia term in the momentum equation (Ramos 2005; Schnack et al. 2006). Although the gyroviscosity effects may play an important role in low-temperature, partially magnetised plasmas (Smolyakov et al. 2016), the validity of the drift models and the necessity of including gyroviscosity effects in low-temperature magnetised plasmas needs to be K. Hara, A.R. Mansour and S. Tsikata 4 investigated. In the state-of-the-art computational models for low-temperature plasmas, simpliﬁed ﬂuid models such as the drift-diffusion model are known to represent the physical processes (Kushner 2009; Hara 2019). In this paper, a ﬁve-moment model that neglects the gyroviscosity effects is used, based on recent numerical simulations that show low-frequency rotating spokes (Mansour & Hara 2022). The inclusion of the gyroviscosity effects is reserved for future work. https://doi.org/10.1017/S002237782200068X Published online by Cambridge University Press 2.3. Zeroth-order (equilibrium) equations for magnetised electrons Note that ue0y is given in (2.7). where ˜ω = ω −kyue0y. Note that ue0y is given in (2.7). y y y g The ﬁrst-order momentum equation can be derived from (2.2). Note that pressure term leads to y y y The ﬁrst-order momentum equation can be derived from (2.2). Note that pressure term leads to kBTe∇(n0 + ne1) me(n0 + ne1) = kBTe me ∇n0 n0 1 −ne1 n0 + ∇ne1 n0 , (2.11) (2.11) which results in two contributions to the linearised momentum equation. Thus, using (2.11), the ﬁrst-order momentum equation for magnetised electrons can be given by −i ˜ωue1 = v2 thkn ne1 n0 ˆx −v2 th ∇ne1 n0 −e me (E1 + ue1 × B0), (2.12) (2.12) where vth = (kBTe/me)1/2 is the electron thermal speed and kn = n′ 0/n0 is the inverse of the density gradient length scale, which is also deﬁned similarly by Sakawa et al. (1993) and Smolyakov et al. (2016). Taking the perturbation terms in y direction, e.g. φ1 exp(−iωt + ikyy) for the electric ﬁeld using the electrostatic assumption: E = −∇φ, (2.12) can be written for x and y directions as where vth = (kBTe/me)1/2 is the electron thermal speed and kn = n′ 0/n0 is the inverse of the density gradient length scale, which is also deﬁned similarly by Sakawa et al. (1993) and Smolyakov et al. (2016). Taking the perturbation terms in y direction, e.g. φ1 exp(−iωt + ikyy) for the electric ﬁeld using the electrostatic assumption: E = −∇φ, (2.12) can be written for x and y directions as −i ˜ω ωce −ωce −i ˜ω ue1x ue1y = v2 thkn ne1 n0 −iv2 thky ne1 n0 + i e me kyφ1 , (2.13) (2.13) where ωce = eB0/me is the electron gyrofrequency. Solving for ue1x and ue1y in (2.13) gives ue1x = i ˜ω2 −ω2 ce ˜ωknv2 th ne1 n0 −ωceky v2 th ne1 n0 −e me φ1 , (2.14) ue1y = 1 ˜ω2 −ω2 ce −ωceknv2 th n1 n0 + ˜ωky v2 th ne1 n0 −e me φ1 . (2.15) (2.15) Equations (2.14) and (2.15) are similarly derived in the Rayleigh–Taylor instability analysis (Chen 1984). If cold electrons are assumed (i.e. vth = 0) and one considers the low-frequency approximation (i.e. 2.3. Zeroth-order (equilibrium) equations for magnetised electrons Inserting (2.4) into the governing equations and considering the zeroth-order terms leads to the equilibrium equations. For magnetised electrons, using (2.3) and considering the equilibrium bulk velocity ue0 = (ue0x, ue0y, ue0z)⊺, where subscripts x, y and z denote the direction, the steady-state momentum equation in x and y directions can be written as 0 = −kBTe men0 ∂n0 ∂x −e me (E0 + ue0yB0), (2.5) 0 = e me ue0xB0, (2.6) (2.5) (2.6) where e is the elementary charge, Te is the electron temperature, me is the electron mass and n0 is the equilibrium density, assuming quasineutrality for the equilibrium condition. Recall that the equilibrium density gradient and electric ﬁeld are considered to exist locally only in x direction. Although (2.6) results in ue0x = 0, (2.5) yields the well-known drifts: ue0y = −E0 B0 −kBTe en0 n′ 0 B0 , (2.7) (2.7) where n′ 0 = dn0/dx is the plasma density gradient. The ﬁrst term in (2.7) is the E × B drift and the second term is the diamagnetic drift, which can be written as uE and u∗, respectively. The diamagnetic drift does not come from the single-particle trajectory analysis but appears as an equilibrium drift from the ﬂuid theory, whereas the E × B drift can be derived from single particle trajectories. Nonetheless, the diamagnetic drift is a steady-state bulk velocity that can propagate in the same or opposite direction of the E × B drift. As the density gradient is only considered in x direction and using ue0x = 0 obtained from (2.6), the steady-state conservation of mass can be written as ∇· ue0 = 0. (2.8) (2.8) This relation shows that ue0y is constant in y direction, i.e. homogeneous, which is consistent with the conﬁguration considered. https://doi.org/10.1017/S002237782200068X Published online by Cambridge University Press Fluid theory of gradient drift instabilities 5 5 2.4. First-order (linear perturbation) equations for magnetised electrons Let us consider the perturbation terms of the electron bulk velocity to be ue1 = (ue1x, ue1y, 0)⊺for magnetised electrons. Using (2.8), the ﬁrst-order conservation of mass can be derived from (2.1) as ∂ne1 ∂t + n0∇· ue1 + ue0 · ∇ne1 + ue1 · ∇n0 = 0. (2.9) (2.9) Using the linear perturbation shown in (2.4), the perturbed electron density can be written as ne1 = n0kyue1y −iue1xn′ 0 ˜ω , (2.10) (2.10) where ˜ω = ω −kyue0y. https://doi.org/10.1017/S002237782200068X Published online by Cambridge University Press 2.3. Zeroth-order (equilibrium) equations for magnetised electrons ˜ω2 ≪ω2 ce), (2.14) and (2.15) reduce to ue1x = E1y/B0 and ue1y = i ˜ωue1x/ωce, which are equivalent to the perturbed E × B and polarisation drifts, respectively. However, it can be seen that the determinant of the matrix in (2.13) becomes negative if ˜ω2 ≫ω2 ce. In this case, ue1x moves in the opposite direction of the perturbed E × B drift, which seems to be non-physical (see Appendix A for some discussions). https://doi.org/10.1017/S002237782200068X Published online by Cambridge University Press K. Hara, A.R. Mansour and S. Tsikata K. Hara, A.R. Mansour and S. Tsikata 6 Therefore, assuming ˜ω2 ≪ω2 ce and using (2.14) and (2.15), (2.10) can be written as Therefore, assuming ˜ω2 ≪ω2 ce and using (2.14) and (2.15), (2.10) can be written as ne1 n0 = eφ1 me k2 y ˜ω −knkyωce ω2 ce ˜ω + k2 y + k2 n v2 th ˜ω −2knkyωcev2 th . (2.16) (2.16) Note that this equation is similar to (10) in Sakawa et al. (1993), except for the coefﬁcient of the last term in the denominator. 2.5. Dispersion function for unmagnetised ions Let us derive the zeroth-order equations for unmagnetised ions. If restricting the argument to the cross-ﬁeld direction (x and y), that is, neglecting the plasma dynamics in z direction (along the magnetic ﬁeld), the steady-state continuity equation gives, n0ui0x = const., where ui0x is the equilibrium ion bulk velocity in x direction. Assuming that the ion bulk velocity is negligible in y direction (i.e. ui0y = 0), the linear perturbation of the ion number density, ni1, can be derived from the linearised conservation of mass as ni1 = n0kyui1y −iui1xn′ 0 ω −iknui0x , (2.17) (2.17) where ui1 = (ui1x, ui1y, 0)⊺is the linear perturbation of the ion bulk velocities. where ui1 = (ui1x, ui1y, 0)⊺is the linear perturbation of the ion bulk velocities. i1 ( i1x, i1y, ) p Assuming cold ions (Ti = 0, where Ti is the ion temperature) and only considering perturbation in y direction, as shown in (2.4), the linearised momentum equation for unmagnetised ions can be written as u1x = 0, (2.18) u1y = ky ω e mi φ1, (2.19) (2.18) (2.19) where mi is the ion mass. If one further assumes that the effects of knui0x to be small, (2.17) can be reduced to where mi is the ion mass. If one further assumes that the effects of knui0x to be small, (2.17) can be reduced to ni1 n0 = ek2 y miω2 φ1. (2.20) (2.20) This is consistent with the dispersion function used for unmagnetised ions in Sakawa et al. (1993). This is consistent with the dispersion function used for unmagnetised ions in Sakawa et al. (1993). https://doi.org/10.1017/S002237782200068X Published online by Cambridge University Press 3.2. Instability criteria in the large-wavelength limit 3.2. Instability criteria in the large-wavelength limit In the limit of a large-wavelength mode, a simpliﬁed form of (3.1) can be obtained by neglecting the ky ˜ω term in the numerator of (3.1) compared with knωce, that is, ky ˜ω ≪ knωce. In addition, u∗ωce = −knv2 th gives k2 nv2 th = ˜u2 ∗ω2 ce, where ˜u∗= u∗/vth. Therefore, in this limit, (3.1) can be written as 0 = −knmi meωce ω2 −ky(1 + ˜u2 ∗)ω + ˜u2 ∗k2 y(uE + u∗) + k2 y(uE −u∗). (3.2) (3.2) For the solution to have an unstable mode, that is, a root with a positive growth rate, the discriminant of (3.2) must be negative. The instability condition can therefore be given by For the solution to have an unstable mode, that is, a root with a positive growth rate, the discriminant of (3.2) must be negative. The instability condition can therefore be given by me mi (1 + ˜u2 ∗)2 −4˜u∗ (1 + ˜u2 ∗)˜uE −(1 −˜u2 ∗)˜u∗ < 0, (3.3) (3.3) where ˜uE = uE/vth. By rewriting (3.3) considering me ≪mi, the condition for the partially magnetised plasma to have an unstable mode in the limit of large wavelength can be written as where ˜uE = uE/vth. By rewriting (3.3) considering me ≪mi, the condition for the partially magnetised plasma to have an unstable mode in the limit of large wavelength can be written as ˜u∗F(˜u∗, ˜uE) > 0, (3.4) (3.4) where where F(˜u∗, ˜uE) = ˜u3 ∗+ ˜uE˜u2 ∗−˜u∗+ ˜uE. (3.5) (3.5) As can be seen from (3.4), the two instability criteria can be obtained as (i) ˜u∗> 0 and F(˜u∗, ˜uE) > 0 and (ii) ˜u∗< 0 and F(˜u∗, ˜uE) < 0. It can be seen that (3.3), in the limit of \|˜u∗\| →0 and considering me ≪mi, reduces to uEu∗> 0, which is the instability criterion for the MSHI (Sakawa et al. 1993). See Appendix B for further discussion about the comparison of the present dispersion relation and MSHI. Figure 1 shows the shape of (3.5) for two values of ˜uE, that is, ˜uE = −0.1 and −0.6, to illustrate the range of unstable and stable roots in the limit of large wavelength and low frequency. Here, negative uE values are chosen, as the E × B drift occurs in −y direction when considering E0 in x direction and B0 in z direction. https://doi.org/10.1017/S002237782200068X Published online by Cambridge University Press 3.1. Fluid dispersion relation Assuming quasineutrality and perturbations occur in the azimuthal direction, which are similar assumptions employed for the Rayleigh–Taylor instability theory (Chen 1984), and using (2.16) and (2.20), the dispersion relation for partially magnetised plasmas can be derived as meky miω2 = ky ˜ω −knωce [ω2 ce + k2 y + k2 n v2 th] ˜ω −2knkyωcev2 th , (3.1) (3.1) where ˜ω = ω −kyue0y is used throughout the derivation. Here, this drift-shifted frequency can be written as ˜ω = ω −ωE −ω∗, where ωE = kyuE, ω∗= kyu∗, uE = −E0/B0 and u∗= −knkBTe/(eB0), as can be seen from (2.7). It can be seen that (3.1) yields a third-order equation for ω, from which the damping and linear instability growth can be evaluated. Fluid theory of gradient drift instabilities 7 Fluid theory of gradient drift instabilities 7 Fluid theory of gradient drift instabilities 7 7 3.2. Instability criteria in the large-wavelength limit Unlike the MSHI that predicts one region for unstable modes, that is, ˜u∗˜uE > 0, two unstable regions can be seen in ﬁgure 1(a), whereas three unstable regions can be seen in ﬁgure 1(b). Figure 2 shows the unstable and stable regions of the gradient drift instability in the large-wavelength limit, as a function of the diamagnetic drift and E × B drift. The unstable roots are obtained according to the instability condition shown in (3.4), and the results are consistent with ﬁgure 1, showing the two and three unstable regions depending on u∗and uE. A few observations about the instability criteria can be made as follows. (i) When \|˜uE\| > 0.3, there are two regions that satisfy the instability condition shown in (3.4). One is uEu∗> 0, leading to a condition that can be written as E0 · ∇n0 > 0, which is similar to the MSHI. In this regime, both E × B and diamagnetic drifts are in the same direction, leading to instability. The other region that results in instability can be found at u∗> vth while uE < 0, as shown in ﬁgures 1(a) and 2. It is interesting to note that E0 · ∇n0 < 0 in this second region, which is a different instability criterion compared to the MSHI. This indicates that a strong diamagnetic drift (compared with the magnitude of the E × B drift and the electron thermal speed) can drive an instability at large wavelength whereas the diamagnetic drift is in the opposite direction to the E × B drift. K. Hara, A.R. Mansour and S. Tsikata 8 8 K. Hara, A.R. Mansour and S. Tsikata (b) (a) FIGURE 1. Cubic function, F(˜u∗, ˜uE), shown in (3.5) for (a) uE = −0.6vth and (b) uE = −0.1vth, resulting in two and three instability regions, respectively. Instabilities occur according to (3.4) in the regions where the product of the u∗and the function F(˜u∗, ˜uE) is positive. The red dashed line denotes F = 0 and the unstable regions are indicated in blue. (b) (a) FIGURE 1. Cubic function, F(˜u∗, ˜uE), shown in (3.5) for (a) uE = −0.6vth and (b) uE = −0.1vth, resulting in two and three instability regions, respectively. Instabilities occur according to (3.4) in the regions where the product of the u∗and the function F(˜u∗, ˜uE) is positive. 3.2. Instability criteria in the large-wavelength limit The red dashed line denotes F = 0 and the unstable regions are indicated in blue. (ii) When 0 < \|˜uE\| ⪅0.3, there are three regions that result in linear instability. As can be seen from ﬁgure 1(b), F becomes positive in a certain range within u∗< 0, leading to a stable root because u∗F < 0. The appearance of a stable mode in the E0 · ∇n0 > 0 region can also be seen from ﬁgure 2. It can be shown from (3.5) that F ≈(˜u∗−˜uE)(˜u∗+ ˜uE −1)(˜u∗+ ˜uE + 1) when assuming \|˜uE\| ≪1. Hence, the three instability regions can be observed approximately at (i) ˜u∗< −1 −˜uE, (ii) ˜uE < ˜u∗< 0 and (iii) ˜u∗> 1 −˜uE for ˜uE < 0. (ii) When 0 < \|˜uE\| ⪅0.3, there are three regions that result in linear instability. As can be seen from ﬁgure 1(b), F becomes positive in a certain range within u∗< 0, leading to a stable root because u∗F < 0. The appearance of a stable mode in the E0 · ∇n0 > 0 region can also be seen from ﬁgure 2. It can be shown from (3.5) that F ≈(˜u∗−˜uE)(˜u∗+ ˜uE −1)(˜u∗+ ˜uE + 1) when assuming \|˜uE\| ≪1. Hence, the three instability regions can be observed approximately at (i) ˜u∗< −1 −˜uE, (ii) ˜uE < ˜u∗< 0 and (iii) ˜u∗> 1 −˜uE for ˜uE < 0. (ii) When 0 < \|˜uE\| ⪅0.3, there are three regions that result in linear instability. As can be seen from ﬁgure 1(b), F becomes positive in a certain range within u∗< 0, leading to a stable root because u∗F < 0. The appearance of a stable mode in the E0 · ∇n0 > 0 region can also be seen from ﬁgure 2. It can be shown from (3.5) that F ≈(˜u∗−˜uE)(˜u∗+ ˜uE −1)(˜u∗+ ˜uE + 1) when assuming \|˜uE\| ≪1. Hence, the three instability regions can be observed approximately at (i) ˜u∗< −1 −˜uE, (ii) ˜uE < ˜u∗< 0 and (iii) ˜u∗> 1 −˜uE for ˜uE < 0. (iii) When ˜uE = 0, there are two regions where the unstable roots exist. Equation (3.5) reduces to F = ˜u∗(˜u∗−1)(˜u∗+ 1) when ˜uE = 0. Hence, the instability regions can be observed at (i) ˜u∗< −1, and (ii) ˜u∗> 1, which can be seen from ﬁgure 2. https://doi.org/10.1017/S002237782200068X Published online by Cambridge University Press 3.2. Instability criteria in the large-wavelength limit The solutions indicate that a low-frequency, large-wavelength instability can occur with a strong diamagnetic drift in the absence of any electric ﬁeld. (iii) When ˜uE = 0, there are two regions where the unstable roots exist. Equation (3.5) reduces to F = ˜u∗(˜u∗−1)(˜u∗+ 1) when ˜uE = 0. Hence, the instability regions can be observed at (i) ˜u∗< −1, and (ii) ˜u∗> 1, which can be seen from ﬁgure 2. The solutions indicate that a low-frequency, large-wavelength instability can occur with a strong diamagnetic drift in the absence of any electric ﬁeld. Figure 3 shows the schematic of the instability criteria for the low-frequency rotating spokes. The ﬁrst condition uEu∗> 0 is similar to the MSHI, in which it is predicted that the E × B and diamagnetic drifts must be in the same direction. Another instability condition observed from the derivations shown in this section (i.e. (3.4) and (3.5)) illustrates that the partially magnetised plasma can be unstable when the diamagnetic drift is sufﬁciently https://doi.org/10.1017/S002237782200068X Published online by Cambridge University Press Fluid theory of gradient drift instabilities 9 y f g f FIGURE 2. Unstable and stable regions of the gradient drift instability in the large-wavelength limit, as a function of uE and u∗. The blue region indicates where the unstable modes exist, whereas unstable modes do not exist in the stable region, shown in white. FIGURE 2. Unstable and stable regions of the gradient drift instability in the large-wavelength limit, as a function of uE and u∗. The blue region indicates where the unstable modes exist, whereas unstable modes do not exist in the stable region, shown in white. (b) (a) FIGURE 3. Instability criteria for low-frequency rotating spokes. (a) The E × B drift and diamagnetic drift are in the same direction. (b) The E × B drift and diamagnetic drift are in the opposite direction and the diamagnetic drift must be larger than the electron thermal speed. (a) (b) FIGURE 3. Instability criteria for low-frequency rotating spokes. (a) The E × B drift and diamagnetic drift are in the same direction. (b) The E × B drift and diamagnetic drift are in the opposite direction and the diamagnetic drift must be larger than the electron thermal speed. larger than the electron thermal speed, even if the electric ﬁeld and the density gradient exist in the opposite direction. 237782200068X Published online by Cambridge University Press 3.2. Instability criteria in the large-wavelength limit larger than the electron thermal speed, even if the electric ﬁeld and the density gradient exist in the opposite direction. larger than the electron thermal speed, even if the electric ﬁeld and the density gradient exist in the opposite direction. 3.4. Resonance It can be seen from (2.14) and (2.15) that the linear perturbation of the electron bulk velocities shows resonance when ˜ω2 ≈ω2 ce. In other words, ω ≈±ωce + kyue0y, where ue0y is the sum of the E × B drift and diamagnetic drift, given in (2.7). If the real frequency is of the order of ion plasma frequency, ωr ≪ωce, as shown later in § 4, a resonance condition for the dispersion relation can be considered to be ky = ∓ωce ue0y . (3.7) (3.7) It can be seen that the smaller (larger) the electron drift, the larger (smaller) ky at which resonance occurs. Equation (3.7) indicates that the electron drift is in resonance with the electron gyromotion, which is akin to the electrostatic two-stream instability where the electron drift is in resonance with the electron plasma frequency. Equation (3.7) can also be written as kyλD = ˜u−1 e0yωce/ωpe or kyrL = ˜u−1 e0y, where ωpe = (e2n0/meϵ0)1/2 is the electron plasma frequency, λD = (ϵ0kBTe/e2n0)1/2 is the Debye length, ϵ0 is the vacuum permittivity, ˜ue0y = ue0y/vth and rL = vth/ωce is the electron Larmor radius. Although the kinetic effects, such as the electron Bernstein mode leading to electron cyclotron drift instability (Cavalier et al. 2013; Hara & Tsikata 2020), may play an important role when kyrL = O(1), the results in the present ﬂuid theory are applicable for the smaller ky range, e.g. kyrL < 1. Investigation of the coupling of the present ﬂuid theory and the kinetic dispersion relation (Chang & Callen 1992) is reserved for future work. 4. Results In this section, the results from the linear perturbation theory are discussed. Although the analytic discussions made in § 3.2 are applicable only in the large-wavelength, low-frequency limit, the dispersion relation introduced in (3.1) is valid for a wide range of ky under the assumption of the ﬂuid approach and before the resonance condition as shown in (3.7). It is discussed in this section that the growth rate exhibits a broadband proﬁle and the real frequency is of the order of the ion plasma frequency, which suggests that the wave propagation speed is of the order of the ion acoustic speed. 5 p p g p p The following conditions are considered for two cases: B0 = 200 G, n0 = 1015 m−3, the electron temperature is 4 eV and mi = 40 amu (assuming singly charged argon ions). These plasma parameters are representative of low-temperature cross-ﬁeld plasma sources. Here, the two cases considered are (I) E0 = 10 kV m−1 and (II) E0 = 4 kV m−1. Cases I and II result in uE ≈−0.6vth and uE ≈−0.24vth, respectively, which correspond to the two and three instability region cases as shown in ﬁgure 1. The solution to (3.1) is evaluated for several representative values of u∗. 3.3. Direction of the wave propagation 3.3. Direction of the wave propagation 3.3. Direction of the wave propagation When the instability condition is met in (3.3), the real part of the solution in (3.2) can be obtained as When the instability condition is met in (3.3), the real part of the solution in (3.2) can be obtained as ωr = (1 + ˜u2 ∗)cs 2u∗ kycs, (3.6) (3.6) where cs = (kBTe/mi)1/2 is the ion acoustic speed. The phase velocity of the wave can be obtained from vφ = ωr/ky. It can therefore be seen from (3.6) that the wave propagates in https://doi.org/10.1017/S002237782200068X Published online by Cambridge University Press https://doi.org/10.1017/S002237782200068X Published online by Cambridge University Press 10 K. Hara, A.R. Mansour and S. Tsikata the direction of the diamagnetic drift, which may lead to the rotating spokes to propagate in the direction of E × B drift or −E × B drift, depending on the instability criteria discussed in § 3.2. the direction of the diamagnetic drift, which may lead to the rotating spokes to propagate in the direction of E × B drift or −E × B drift, depending on the instability criteria discussed in § 3.2. https://doi.org/10.1017/S002237782200068X Published online by Cambridge University Press 4.1. Case I: two instability regions, uE ≈−0.6vth 4.1. Case I: two instability regions, uE ≈−0.6vth Figure 4 shows the real and imaginary parts of the solution of (3.1) for different u∗values for uE ≈−0.6vth. This is consistent with the two instability region case, which occurs when \|uE\| > 0.3, as shown in ﬁgure 1(a). In the limit of a large positive diamagnetic drift (e.g. u∗= 1.67vth as shown in ﬁgure 4a), a low-frequency mode is indeed observed in the limit of large wavelength. https://doi.org/10.1017/S002237782200068X Published online by Cambridge University Press Fluid theory of gradient drift instabilities 11 Fluid theory of gradient drift instabilities 11 / pi 0 5 10 15 20 25 / pi -10 -5 0 5 10 ky D / pi -15 -10 -5 0 5 10 15 ky D 0 1 2 3 0 1 2 3 r u* = -0.60 vth u* = 1.67 vth u* = -0.30 vth u* = 1.37 vth u* = 1.01 vth u* = -0.06 vth (e) (b) (a) (c) (d ) ( f ) FIGURE 4. Case I: two instability regions at large wavelength, for different values of u∗with uE ﬁxed at uE = −0.6vth. This value of uE = −0.6vth corresponds to the case illustrated in ﬁgure 1(a). The real frequency, ωr, and growth rate, γ , are shown in blue dashed and black solid lines, respectively, for each value of u∗. FIGURE 4. Case I: two instability regions at large wavelength, for different values of u∗with uE ﬁxed at uE = −0.6vth. This value of uE = −0.6vth corresponds to the case illustrated in ﬁgure 1(a). The real frequency, ωr, and growth rate, γ , are shown in blue dashed and black solid lines, respectively, for each value of u∗. The real frequency, ωr, is positive, indicating that wave propagates in the direction of the diamagnetic drift as opposed to the E × B drift. This observation is consistent with the theoretical observation in § 3.2. The real frequency, ωr, is positive, indicating that wave propagates in the direction of the diamagnetic drift as opposed to the E × B drift. This observation is consistent with the theoretical observation in § 3.2. Figures 4(b) and 4(c) show that the dispersion relation yields roots with a positive growth rate at a ﬁnite ky ̸= 0 but yields stable solutions in the large-wavelength limit, as the magnitude of the diamagnetic drift decreases. https://doi.org/10.1017/S002237782200068X Published online by Cambridge University Press 4.2. Case II: three instability regions, uE ≈−0.24vth 4.2. Case II: three instability regions, uE ≈−0.24vth Figure 5 shows the real and imaginary parts of the solution of (3.1) for several representative u∗values for uE ≈−0.24vth. As discussed in § 3.2 and shown in ﬁgure 1(b), three instability regions can be seen in the large-wavelength limit, when \|uE\| ⪅0.3vth. Figure 5 shows the real and imaginary parts of the solution of (3.1) for several representative u∗values for uE ≈−0.24vth. As discussed in § 3.2 and shown in ﬁgure 1(b), three instability regions can be seen in the large-wavelength limit, when \|uE\| ⪅0.3vth. ⪅ Similar to Case I, there is a low-frequency mode at large wavelength when the diamagnetic drift is opposite to the E × B drift and the magnitude of the diamagnetic drift is sufﬁciently larger than the electron thermal speed, as shown in ﬁgure 5(a). Under this situation, the wave propagation occurs in the direction of the diamagnetic drift. As the magnitude of the diamagnetic drift is decreased, the partially magnetised plasma becomes stable in the large-wavelength limit but a resonance type mode appears, as shown in ﬁgures 5(b) and 5(c), similar to Case I. In addition, when the diamagnetic drift is in the same direction as the E × B drift, the partially magnetised plasma is unstable in the large-wavelength limit, which can be seen in ﬁgures 5(d) and 5( f). However, it is interesting to note that the growth rate becomes zero at ky →0 for the cases with intermediate u∗values, as shown in ﬁgure 5(e). This is consistent with the theoretical prediction in § 3.2 and ﬁgure 1(b): when \|uE\| ⪅0.3vth, the partially magnetised plasma becomes stable at ky →0 for a range of u∗in which u∗F < 0, as shown in ﬁgure 1(b). y In summary, the partially magnetised plasma dispersion relation using the ﬂuid approach shows that the linear growth rate is positive in the large-wavelength limit not only (i) when E0 · ∇n0 > 0 but also (ii) when E0 · ∇n0 < 0 and the diamagnetic drift is sufﬁciently larger than the electron thermal speed. Although the former is similar to the so-called MSHI, the latter suggests that a strong diamagnetic drift may excite large-wavelength, low-frequency plasma oscillations that propagate in the direction of the diamagnetic drift. https://doi.org/10.1017/S002237782200068X Published online by Cambridge University Press 4.1. Case I: two instability regions, uE ≈−0.6vth Note that for these cases, uEu∗< 0, that is, the diamagnetic drift is in the direction opposite to the E × B drift. These results are consistent with the prediction from the analytic theory obtained in § 3.2. It is to be noted that the sign of the real frequency ﬂips, which can be seen from ﬁgures 4(b) and 4(c), indicating that the resonance type phenomenon is driven by the diamagnetic drift in ﬁgure 4(b). However, the wave propagates in the direction of the E × B drift in ﬁgure 4(c) https://doi.org/10.1017/S002237782200068X Published online by Cambridge University Press K. Hara, A.R. Mansour and S. Tsikata 12 despite the fact that the magnitude of diamagnetic drift is larger than the E × B drift. As the diamagnetic drift approaches zero, the resonance condition shifts towards a larger kyλD, as shown in (3.7), and the partially magnetised plasma is stable for a wide range of ky. y When the diamagnetic drift is in the same direction as the E × B drift (i.e. uEu∗> 0), the results in ﬁgure 4(d–f) illustrate that the large-wavelength mode is indeed unstable. This is consistent with § 3.2, in which it is observed that uEu∗> 0 yields a root with a positive growth rate at the large-wavelength, low-frequency limit, if \|uE\| > 0.3vth. Another observation that can be made is that the real frequency is negative, illustrating that the wave propagates in the direction of both E × B drift and diamagnetic drift. The real frequency where the growth rate is maximum is approximately (5-6)ωpi, suggesting that the wave propagation speed is of the order of the ion acoustic speed. The cutoff where γ becomes zero at a ﬁnite ky in ﬁgure 4(d–f) occurs before the resonance condition discussed in § 3.4. 4.2. Case II: three instability regions, uE ≈−0.24vth 5. Application of the gradient drift instability to cross-ﬁeld plasma sources Now that we have established the theory of the gradient drift instability, in this section, the theory is applied to various cross-ﬁeld plasma conﬁgurations. Although the theory discussed in the present paper assumes a slab geometry and the real systems are cylindrical, the slab approximation may provide ﬁrst-order estimates for the local instabilities in cross-ﬁeld plasma sources. Here, four conﬁgurations are considered: (a) Penning discharge, (b) cylindrical magnetron (e.g. for high-power microwave generation), (c) planar magnetron discharge (e.g. for plasma-assisted deposition) and (d) Hall effect thruster (HET), as shown in ﬁgure 6. Fluid theory of gradient drift instabilities 13 Fluid theory of gradient drift instabilities 13 / pi 0 5 10 15 20 / pi -10 -5 0 5 10 ky D / pi 0 1 2 3 -15 -10 -5 0 5 10 15 =0 ky D 0 1 2 3 r u* = -0.83 vth u* = 1.31 v th u* = -0.48 vth u* = 0.95 v th u* = 0.48 v th u* = -0.24 v th (e) (b) (a) (c) (d ) ( f ) FIGURE 5. Case II: three instability regions at large wavelength, for different values of u∗with uE ﬁxed at uE = −0.24vth. The real frequency, ωr, and growth rate, γ , are shown in blue dashed and black solid lines, respectively, for each value of u∗. In ﬁgure 5(e), the growth rate is zero at kyλD ≪1, which corresponds to the region where u∗< 0 and F > 0, shown in ﬁgure 1(b). FIGURE 5. Case II: three instability regions at large wavelength, for different values of u∗with uE ﬁxed at uE = −0.24vth. The real frequency, ωr, and growth rate, γ , are shown in blue dashed and black solid lines, respectively, for each value of u∗. In ﬁgure 5(e), the growth rate is zero at kyλD ≪1, which corresponds to the region where u∗< 0 and F > 0, shown in ﬁgure 1(b). https://doi.org/10.1017/S002237782200068X Published online by Cambridge University Press 5.2. Cylindrical magnetron The cylindrical magnetron, shown in ﬁgure 6(b), is used to study critical ionisation velocity (CIV) phenomena (Brenning et al. 2013). In addition, such a cross-ﬁeld conﬁguration is used in high-power microwave sources. For its use as a microwave generator, ion formation is to be avoided because the working principle is that the electrons from the cathode are trapped (i.e. insulated) by the applied or induced magnetic ﬁeld to generate high-power microwaves (Benford et al. 2015). However, as the devices increase in power and become more compact, plasma generation is unavoidable as the current density in the system increases (Hadas et al. 2008). Although low-frequency plasma oscillations are not well-studied due to the short pulse operation, it can be seen from ﬁgure 6(b) that the partially magnetised plasma is unstable in such conﬁgurations, similar to the Penning discharge. 5.1. Penning discharge The Penning discharge, as shown in ﬁgure 6(a), operates using an outer cylinder as an anode with a cathode placed along the centreline, in addition to an applied axial magnetic ﬁeld (Hoh 1963; Simon 1963). The plasma density is typically largest near the cathode, generating a radially inward plasma density gradient, ∇rn0 < 0. At the same time, the applied electric ﬁeld is also radially inward, Er < 0. Penning discharge typically naturally satisﬁes the condition E0 · ∇n0 > 0, that is, the E × B and diamagnetic drifts are in the same direction. Therefore, as shown in ﬁgure 3(a), the Penning-type discharge satisﬁes (3.4), generating a gradient drift instability at low frequency. K. Hara, A.R. Mansour and S. Tsikata 14 (b) (a) (c) (d ) (b) (a) (a) (c) (d ) (d ) FIGURE 6. Application of the gradient drift instability theory to various cross-ﬁeld plasma devices. Note that the geometries between top and bottom ﬁgures are rotated to keep the magnetic ﬁeld direction consistent, that is, into the page. Blue arrows, that is, u0r in (a,b) or u0z in (c,d), indicate the cross-ﬁeld electron ﬂow, whereas the red arrows indicate the electron drift in the azimuthal direction. https://doi.org/10.1017/S002237782200068X Published online by Cambridge University Press 5.4. HET The HET operates using a radial applied magnetic ﬁeld and axial electric ﬁeld, as shown in ﬁgure 6(d). In the plume of a HET, the E × B drift is much larger than the diamagnetic drift and the diamagnetic drift is in the opposite direction to the E × B drift, which is shown as region 2 in ﬁgure 6(d). Consequently, using the instability criteria discussed in (3.4), the partially magnetised plasma in the HET plume is stable against the Penning-type instability. The simulation results by Kawashima et al. (2018) were compared with a class of gradient drift instabilities in the presence of a magnetic ﬁeld gradient (Esipchuk & Tilinin 1976; Tilinin 1977; Frias et al. 2013) and showed that the gradient drift instability due to the magnetic ﬁeld gradient grows in the plume region, which leads to the excitation of rotating spokes. However, if one considers the plasma inside the channel, shown as region 1 in ﬁgure 6(d), the cross-ﬁeld electron transport is primarily driven by the pressure gradient near the anode. It can therefore be considered that the magnitude of the diamagnetic drift is larger than that of the E × B drift, while the two drifts may be in the opposite direction. Thus, this condition may lead to the gradient drift instability presented in this paper to be unstable. The spoke rotation will be in the direction of the diamagnetic drift driven by the plasma density gradient inside the channel, which turns out to be in the same direction as the E × B drift due to the applied electric ﬁeld (set up by the anode and cathode) in the system. In summary, there are two scenarios in which azimuthally rotating spokes can occur in HETs and in the planar magnetron discharge. One is due to the instabilities that are caused by the magnetic ﬁeld gradient. The other possibility is the gradient drift instability, discussed in the present paper, which is caused by the diamagnetic drift in the absence of magnetic ﬁeld gradients. 5.3. Planar magnetron discharge Figure 6(c) shows a schematic of the planar magnetron, used for plasma-assisted sputtering and deposition (Waits 1978). An axial electric ﬁeld and a radial magnetic ﬁeld are applied close to the cathode surface. The crossed electric and magnetic ﬁelds generate electron drifts in the azimuthal direction. Recent experimental observations in high-power impulse magnetron sputtering (HiPIMS) show that the rotating spoke direction may be a function of the current (Anders & Yang 2017) and can be reversed during a certain operation (Hecimovic et al. 2016). Similar reversal of the rotating spoke propagation is found in a micro-magnetron discharge https://doi.org/10.1017/S002237782200068X Published online by Cambridge University Press Fluid theory of gradient drift instabilities 15 (Ito et al. 2015; Marcovati et al. 2020). The global axial plasma proﬁle, such as the location and amplitude of the density and potential gradients, can affect the characteristics of the gradient drift instability, leading to rotating spokes. (Ito et al. 2015; Marcovati et al. 2020). The global axial plasma proﬁle, such as the location and amplitude of the density and potential gradients, can affect the characteristics of the gradient drift instability, leading to rotating spokes. In addition, magnetron sputtering devices have a large magnetic ﬁeld inhomogeneity in the region where the rotating spokes occur, leading to the possibility of gradient drift instabilities due to magnetic ﬁeld gradients (Esipchuk & Tilinin 1976; Tilinin 1977; Frias et al. 2013) playing an important role in the formation of rotating spokes. 6. Conclusion The present paper has reviewed the ﬂuid dispersion relation of the partially magnetised plasmas. The dispersion relation derived in this paper shows that the partially magnetised plasmas are unstable (i) when E0 · ∇n0 > 0 or (ii) in the presence of a large diamagnetic drift while E0 · ∇n0 < 0. The former is consistent with the so-called MSHI where the E × B drift and the diamagnetic drift are in the same direction. The latter is an instability that occurs when the diamagnetic drift is sufﬁciently larger than the electron thermal speed while occurring in the direction against the E × B drift. One similarity between the two instability conditions is that the plasma wave in the large-wavelength limit propagates in the direction of the diamagnetic drift. This indicates that the low-frequency, large-wavelength, partially magnetised plasma oscillations can occur in either E × B or −E × B direction, depending on the plasma conditions. https://doi.org/10.1017/S002237782200068X Published online by Cambridge University Press Declaration of interests The authors report no conﬂict of interest. Data availability The data that support the ﬁndings of this study are available from the corresponding author upon reasonable request. Acknowledgements This material is based on work supported by the US Department of Energy, Ofﬁce of Science, Ofﬁce of Fusion Energy Sciences, under Award No. DE-SC0020623, by https://doi.org/10.1017/S002237782200068X Published online by Cambridge University Press K. Hara, A.R. Mansour and S. Tsikata 16 the Air Force Ofﬁce of Scientiﬁc Research under Awards No. FA9550-18-1-0090 and No. FA9550-21-1-0433, and by the Ofﬁce of Naval Research under Award No. N00014-21-1-2698. the Air Force Ofﬁce of Scientiﬁc Research under Awards No. FA9550-18-1-0090 and No. FA9550-21-1-0433, and by the Ofﬁce of Naval Research under Award No. N00014-21-1-2698. tor Edward Thomas, Jr. thanks the referees for their advice in evaluating this article Authors contributions K.H. derived the theory and held discussions with A.R.M. and S.T. Results were prepared by K.H. and all authors contributed to analysing data, reaching conclusions and in writing the paper. https://doi.org/10.1017/S002237782200068X Published online by Cambridge University Press .1017/S002237782200068X Published online by Cambridge University Press 237782200068X Published online by Cambridge University Press Appendix A. Eﬀect of ˜ω2 in the electron bulk velocity dispersion In (2.16), it is assumed that ˜ω2 ≪ω2 ce. In the limit of large ky, the ﬂuid dispersion relation, shown in (3.1), can be written as 0 = (ω2 −k2 yc2 s) ˜ω. (A1) (A1) The solution to the dispersion relation yields three roots, that is, ω = ±kycs, kyue0y. Thus, there are no growth rates in the limit of ky →∞in ﬁgures 4 and 5. The solution to the dispersion relation yields three roots, that is, ω = ±kycs, kyue0y. Thus, there are no growth rates in the limit of ky →∞in ﬁgures 4 and 5. y If one retains the ˜ω2 term in (2.14) and (2.15), the electron density perturbation term can be written as, ne1 n0 = eφ1 me k2 y ˜ω −knkyωce −˜ω3 + ω2 ce ˜ω + k2 y + k2 n v2 th ˜ω −2knkyωcev2 th . (A2) (A2) Combining (A2) and (2.20), a third-order equation can be constructed for ω. For this case, the dispersion relation in the limit of ky →∞can be given by Combining (A2) and (2.20), a third-order equation can be constructed for ω. For this case, the dispersion relation in the limit of ky →∞can be given by 0 = ω2 −k2 yc2 s + me mi ˜ω2 ˜ω. (A3) (A3) In (A3), one solution is always real, that is, ω = kyue0y. However, the other two solutions must be evaluated separately. Using me ≪mi, (A3) can be rewritten as In (A3), one solution is always real, that is, ω = kyue0y. However, the other two solutions must be evaluated separately. Using me ≪mi, (A3) can be rewritten as 0 = ω2 −2me mi kyue0yω −k2 y c2 s + me mi u2 e0y . (A4) (A4) The discriminant of (A4) can be obtained as The discriminant of (A4) can be obtained as D = me mi kyue0y 2 −k2 y c2 s + me mi u2 e0y , (A5) (A5) which can be seen to be negative (i.e. D < 0), because me ≪mi. Thus, these two solutions are not real and become imaginary at ky →∞. which can be seen to be negative (i.e. D < 0), because me ≪mi. Thus, these two solutions are not real and become imaginary at ky →∞. Appendix A. Eﬀect of ˜ω2 in the electron bulk velocity dispersion https://doi.org/10.1017/S002237782200068X Published online by Cambridge University Press 17 Fluid theory of gradient drift instabilities Although the inclusion of the ˜ω3 term affects the results in the larger ky region, the large-wavelength solutions (kyλD < 1) shown in ﬁgures 4 and 5 are not affected by the inclusion of the ˜ω3 term. Therefore, in the present paper, the derivations and results neglecting the ˜ω3 term in (A2) are shown. Appendix B. Comparison with MSHI theory The condition for gradient drift instability shown in (3.4) and (3.5) is different from the condition that is originally proposed by Simon (Simon 1963), that is, E0 · ∇n0 > 0. To derive a dispersion relation similar to the MSHI (Sakawa et al. 1993), \|˜u∗\| ≪1 and ky →0 can be assumed so that (3.1) reduces to The condition for gradient drift instability shown in (3.4) and (3.5) is different from the condition that is originally proposed by Simon (Simon 1963), that is, E0 · ∇n0 > 0. To derive a dispersion relation similar to the MSHI (Sakawa et al. 1993), \|˜u∗\| ≪1 and ky →0 can be assumed so that (3.1) reduces to k2 yc2 s ω2 = ω∗ ω −ωE + ω∗ , (B1) (B1) where cs = (kBTe/mi)1/2 is the ion acoustic speed. Assuming me ≪mi, the instability criterion from (B1) can be derived as u∗(uE −u∗) > 0, which leads to \|uE\| > \|u∗\| and uEu∗> 0. This means that, assuming \|˜u∗\| ≪1, a positive growth rate can be obtained only when the magnitude of diamagnetic drift is larger than that of E × B drift and E0 · ∇n0 > 0 are both satisﬁed. It should be noted that the dispersion relation proposed by Sakawa et al. (1993) omitted a factor of two in the last term of the denominator of the right-hand side in (3.1). This results in a dispersion relation that omits the ω∗term in the denominator in (B1), leading to the instability condition to be unconditionally u∗uE > 0 and therefore E0 · ∇n0 > 0. It is discussed in § 3.2 that the MSHI can be recovered in the limit of \|˜u∗\| →0, which can be seen from (B1) as well. REFERENCES 2019 Non-invasive time-resolved measurements of anomalous collision frequency in a Hall thruster. Phys. Plasmas 26 (1), 013516. q y y ( ), ELLISON, C.L., RAITSES, Y. & FISCH, N.J. 2012 Cross-ﬁeld electron transport induced by a rotating spoke in a cylindrical Hall thruster. Phys. Plasmas 19 (1), 013503. p y ESIPCHUK, Y.B. & TILININ, G.N. 1976 Drift instability in a Hall-current plasma accelerator. Sov. Phys. Tech. Phys. 21, 417. FORSLUND, D.W., MORSE, R.L. & NIELSON, C.W. 1971 Nonlinear electron-cyclotron drift instability and turbulence. Phys. Rev. Lett. 27 (21), 1424. FRIAS, W., SMOLYAKOV, A.I., KAGANOVICH, I.D. & RAITSES, Y. 2013 Long wavelength gradient drift instability in Hall plasma devices. II. Applications. Phys. Plasmas 20 (5), 052108. GARY, S.P. 1991 Electromagnetic ion/ion instabilities and their consequences in space plasmas: a review. Space Sci. Rev. 56 (3), 373–415. HADAS, Y., SAYAPIN, A., KRASIK, YA.E., BERNSHTAM, V. & SCHNITZER, I. 2008 Plasma dynamics during relativistic S-band magnetron operation. J. Appl. Phys. 104 (6), 064125. HARA, K 2019 An overview of discharge plasma modeling for Hall effect thrusters. Plasma Sources Sci. Technol. 28, 044001. HARA, K., SEKERAK, M.J., BOYD, I.D. & GALLIMORE, A.D. 2014a Mode transition of a Hall thruster discharge plasma. J. Appl. Phys. 115 (20), 203304. HARA, K., SEKERAK, M.J., BOYD, I.D. & GALLIMORE, A.D. 2014b Perturbation analysis of ionization oscillations in Hall effect thrusters. Phys. Plasmas 21 (12), 122103. HARA, K. & TSIKATA, S. 2020 Cross-ﬁeld electron diffusion due to the coupling of drift-driven microinstabilities. Phys. Rev. E 102, 023202. HECIMOVIC, A., MASZL, C., VON DER GATHEN, V.S., BÖKE, M. & VON KEUDELL, A. 2016 Spoke rotation reversal in magnetron discharges of aluminium, chromium and titanium. Plasma Sources Sci. Technol. 25 (3), 035001. ( ), HECIMOVIC, A. & VON KEUDELL, A. 2018 Spokes in high power impulse magnetron sputtering plasmas. J. Phys. D: Appl. Phys. 51 (45), 453001. HÉRON, A. & ADAM, J.C. 2013 Anomalous conductivity in Hall thrusters: effects of the non-linear coupling of the electron-cyclotron drift instability with secondary electron emission of the walls. Phys. Plasmas 20 (8), 082313. y HOH, F.C. 1963 Instability of Penning-type discharges. Phys. Fluids 6 (8), 1184–1191. ITO, T., YOUNG, C.V. & CAPPELLI, M.A. 2015 Self-organization in planar magnetron microdischarge plasmas. Appl. Phys. Lett. 106 (25), 254104. JANHUNEN, A., SMOLYAKOV, A., SYDORENKO, D., JIMENEZ, M., KAGANOVICH, I. & RAITSES, Y. 2018 Evolution of the electron cyclotron drift instability in two-dimensions. Phys. Plasmas 25 (8), 082308. REFERENCES ABOLMASOV, S.N. 2012 Physics and engineering of crossed-ﬁeld discharge devices. Plasma Sources Sci. Technol. 21 (3), 035006. ADAM, J.C., HÉRON, A. & LAVAL, G. 2004 Study of stationary plasma thrusters using two-dimensional fully kinetic simulations. Phys. Plasmas 11 (1), 295–305. ANDERS, A. 2012 Self-organization and self-limitation in high power impulse magnetron sputtering. Appl. Phys. Lett. 100 (22), 224104. ANDERS, A. & YANG, Y. 2017 Direct observation of spoke evolution in magnetron sputtering. Appl. Phys. Lett. 111 (6), 064103. BAL, S. & BOSE, M. 2010 Modiﬁed Simon–Hoh instability in a magnetized inhomogeneous dusty plasma. J. Phys.: Conf. Ser. 208, 012081. BARRAL, S. & AHEDO, E. 2009 Low-frequency model of breathing oscillations in Hall discharges. Phys. Rev. E 79, 046401. BENFORD, J., SWEGLE, J.A. & SCHAMILOGLU, E. 2015 High Power Microwaves, 3rd edn. CR BERNSTEIN, I.B. 1958 Waves in a plasma in a magnetic ﬁeld. Phys. Rev. 109, 10–21. BOEUF, J.-P. 2014 Rotating structures in low temperature magnetized plasmas—insight from particle simulations. Front. Phys. 2, 74. OEUF, J.-P. 2017 Tutorial: physics and modeling of Hall thrusters. J. Appl. Phys. 121 (1), 011101. BOEUF, J.-P. & GARRIGUES, L. 1998 Low frequency oscillations in a stationary plasma thruster. J. Appl. Phys. 84 (7), 3541–3554. BOEUF, J.-P. & TAKAHASHI, M. 2020 Rotating spokes, ionization instability, and electron vortices in partially magnetized E × B plasmas. Phys. Rev. Lett. 124, 185005. BRENEMAN, A.W., CATTELL, C.A., KERSTEN, K., PARADISE, A., SCHREINER, S., KELLOGG, P.J., GOETZ, K. & WILSON, L.B. III 2013 Stereo and wind observations of intense cyclotron harmonic waves at the Earth’s bow shock and inside the magnetosheath. J. Geophys. Res.: Space 118 (12), 7654–7664. https://doi.org/10.1017/S002237782200068X Published online by Cambridge University Press K. Hara, A.R. Mansour and S. Tsikata 18 BRENNING, N., LUNDIN, D., MINEA, T., COSTIN, C. & VITELARU, C. 2013 Spokes and charged particle transport in HiPIMS magnetrons. J. Phys. D: Appl. Phys. 46 (8), 084005. CAVALIER, J., LEMOINE, N., BONHOMME, G., TSIKATA, S., HONORÉ, C. & GRÉSILLON, D. 2013 Hall thruster plasma ﬂuctuations identiﬁed as the E × B electron drift instability: modeling and ﬁtting on experimental data. Phys. Plasmas 20 (8), 082107. CHANG, Z. & CALLEN, J.D. 1992 Uniﬁed ﬂuid/kinetic description of plasma microinstabilities. Part I: basic equations in a sheared slab geometry. Phys. Fluids B: Plasma 4 (5), 1167–1181. CHEN, F.F. 1984 Introduction to Plasma Physics and Controlled Fusion. Springer. DALE, E.T. & JORNS, B.A. REFERENCES KAGANOVICH, I.D., RAITSES, Y., SYDORENKO, D. & SMOLYAKOV, A. 2007 Kinetic effects in a Hall thruster discharge. Phys. Plasmas 14 (5), 057104. KAWASHIMA, R., HARA, K. & KOMURASAKI, K. 2018 Numerical analysis of azimuthal rotating spokes in a crossed-ﬁeld discharge plasma. Plasma Sources Sci. Technol. 27 (3), 035010. KEIDAR, M. & BEILIS, I.I. 2006 Electron transport phenomena in plasma devices with E × B drift. IEEE Trans. Plasma Sci. 34 (3), 804–814. KUSHNER, M.J. 2009 Hybrid modelling of low temperature plasmas for fundamental investigations and equipment design. J. Phys. D: Appl. Phys. 42 (19), 194013. LAU, Y.Y. 2001 Simple theory for the two-dimensional Child–Langmuir law. Phys. Rev. Lett. 87, 278301. https://doi.org/10.1017/S002237782200068X Published online by Cambridge University Press 19 Fluid theory of gradient drift instabilities MANSOUR, A.R. & HARA, K. 2022 Full ﬂuid moment modeling of rotating spokes in Penning-type conﬁguration. Plasma Sources Sci. Technol. 31, 055012. MARCOVATI, A., ITO, T. & CAPPELLI, M.A. 2020 The dynamics of coherent modes of gra instabilities in a small magnetron discharge plasma. J. Appl. Phys. 127 (22), 223301. ZOUFFRE, S., GRIMAUD, L., TSIKATA, S., MATYASH, K. & SCHNEIDER, R. 2019 Rotating sp instabilities in a wall-less Hall thruster: experiments. Plasma Sources Sci. Technol. 28 (5), 0540 MCBRIDE, J.B., OTT, E., BORIS, J.P. & ORENS, J.H. 1972 Theory and simulation of turbulent heating by the modiﬁed two-stream instability. Phys. Fluids 15 (12), 2367–2383. POWIS, A.T., CARLSSON, J.A., KAGANOVICH, I.D., RAITSES, Y. & SMOLYAKOV, A. 2018 Scaling of spoke rotation frequency within a Penning discharge. Phys. Plasmas 25 (7), 072110. QURAISHI, Q., ROBERTSON, S. & WALCH, R. 2002 Electron diffusion in the annular Penning trap. Phys. Plasmas 9 (8), 3264–3271. RAITSES, Y., KAGANOVICH, I.D. & SMOLYAKOV, A. 2015 Effects of the gas pressure on low frequency oscillations in E × B discharges. In Proceedings of Joint Conference of 30th International Symposium on Space Technology and Science 34th International Electric Propulsion Conference and 6th Nano-satellite Symposium, pp. IEPC–2015–307. pp RAMOS, J.J. 2005 General expression of the gyroviscous force. Phys. Plasmas 12 (11), 112301. ROMADANOV, I., SMOLYAKOV, A., RAITSES, Y., KAGANOVICH, I., TANG, T. & RYZHKOV, S. 2016 Structure of nonlocal gradient-drift instabilities in Hall E × B discharges. Phys. Plasmas 23 (12), 122111. SAKAWA, Y., JOSHI, C., KAW, P.K., CHEN, F.F. & JAIN, V.K. 1993 Excitation of the modiﬁed Simon–Hoh instability in an electron beam produced plasma. Phys. Fluids B: Plasma 5 (6), 1681–1694. WAITS, R.K. 1978 Planar magnetron sputtering. J. Vac. Sci. Technol. 15 (2), 179–187. REFERENCES SCHNACK, D.D., BARNES, D.C., BRENNAN, D.P., HEGNA, C.C., HELD, E., KIM, C.C., KRUGER, S.E., PANKIN, A.Y. & SOVINEC, C.R. 2006 Computational modeling of fully ionized magnetized plasmas using the ﬂuid approximation. Phys. Plasmas 13 (5), 058103. SEKERAK, M.J., LONGMIER, B.W., GALLIMORE, A.D., BROWN, D.L., HOFER, R.R. & POLK, J.E. 2015 Azimuthal spoke propagation in Hall effect thrusters. IEEE Trans. Plasma Sci. 43 (1), 72–85. SIMON, A. 1963 Instability of a partially ionized plasma in crossed electric and magnetic ﬁelds. Phys. SEKERAK, M.J., LONGMIER, B.W., GALLIMORE, A.D., BROWN, D.L., HOFER, R.R. & POLK, J.E. 2015 Azimuthal spoke propagation in Hall effect thrusters. IEEE Trans. Plasma Sci. 43 (1), 72–85. 2015 Azimuthal spoke propagation in Hall effect thrusters. IEEE Trans. Plasma Sci. 43 (1), 72–85. SIMON, A. 1963 Instability of a partially ionized plasma in crossed electric and magnetic ﬁelds. Phys. Fluids 6 (3), 382–388. SIMON, A. 1963 Instability of a partially ionized plasma in crossed electric and magnetic ﬁelds. Phys. Fluids 6 (3), 382–388. SMOLYAKOV, A.I., CHAPURIN, O., FRIAS, W., KOSHKAROV, O., ROMADANOV, I., TANG, T., UMANSKY, M., RAITSES, Y., KAGANOVICH, I.D. & LAKHIN, V.P. 2016 Fluid theory and simulations of instabilities, turbulent transport and coherent structures in partially-magnetized plasmas of E × B discharges. Plasma Phys. Control. Fusion 59 (1), 014041. SYDORENKO, D., SMOLYAKOV, A., KAGANOVICH, I. & RAITSES, Y. 2008 Plasma-sheath instability in Hall thrusters due to periodic modulation of the energy of secondary electrons in cyclotron motion. Phys. Plasmas 15 (5), 053506. TILININ, G.N. 1977 High-frequency plasma waves in a Hall accelerator with an extended acceleration zone. Zh. Tekh. Fiz. 47, 1684–1691. TSIKATA, S., CAVALIER, J., HÉRON, A., HONORÉ, C., LEMOINE, N., GRÉSILLON, D. & COULETTE, D. 2014 An axially propagating two-stream instability in the Hall thruster plasma. Phys. Plasmas 21 (7), 072116. VILLAFANA, W., PETRONIO, F., DENIG, A.C., JIMENEZ, M.J., EREMIN, D., GARRIGUES, L., TACCOGNA, F., ALVAREZ-LAGUNA, A., BOEUF, J.P., BOURDON, A., et al. 2021 2D radial-azimuthal particle-in-cell benchmark for E × B discharges. Plasma Sources Sci. Technol. 30 (7), 075002. WAITS, R.K. 1978 Planar magnetron sputtering. J. Vac. Sci. Technol. 15 (2), 179–187. https://doi.org/10.1017/S002237782200068X Published online by Cambridge University Press
https://openalex.org/W3112077729	https://www.e3s-conferences.org/10.1051/e3sconf/202022001081/pdf	English	null	On Methods of Checking Digital Out-Of-Step Protection Devices	E3S web of conferences	2,020	cc-by	4,653	Corresponding author: litvinovii@mail.ru © The Authors, published by EDP Sciences. This is an open access article distributed under the terms of the Creative Commons Attribution License 4.0 (http://creativecommons.org/licenses/by/4.0/). Ilya I. Litvinov1,, Alexandra A. Marchenko1, and Elena I. Gracheva2 1Novosibirsk State Technical University, K.Marksa prospect, 20, 630073 Novosibirsk, Russia 2Kazan State Power Engineering University, Krasnoselskaya st,. 51, 420066 Kazan, Russia Abstract. The article addressed the issue of commissioning task that must take place before putting out-of- step protection into operation. Most of the developers of digital out-of-step protection devices utilize control of impedance vector locus parameter during transient in a power system to detect its out-of-step operation, however checking the setting of such protection in the field conditions using existing means is often a challenging work. To facilitate the commissioning of such devices and provide means to automatize the process so as to rid of human factor the authors suggest means of simulating impedance vector locus with the desired parameters on the complex plane to check distance out-of-step protection. The means for checking current-oscillation based out-of-step protection is also considered. The results of a transient simulation representing instantaneous values of currents and voltages are saved in the international oscillography recording format COMTRADE, which makes it possible to carry out comprehensive verification of emergency control equipment using any kind of testing equipment that is able to generate the corresponding electrical signals stored in such format. achieve the task set, it is necessary to use mathematical simulation. 1 Introduction Out-of-step protection (OOSP) is one of the most important types of emergency control automation designed to prevent the development of an emergency situation caused by a loss of dynamic stability of a power system. Basically, OOSP can be based on different principles of operation: current, distance and angular OOSP [1]. At present, each design organization is developing its own OOSP operation algorithm based on one or another concept of detection of out-of-step operation, which undoubtedly increases the reliability of the power system operation [2-14]. However, OOSP algorithms simultaneously control many parameters of the electric power network, and therefore are very difficult to implement, and checking their settings in the process of commissioning is labor-consuming. Usually verification of such automation is carried out with use of “RL-model of the power system” software which is included in the set of basic programs of the RETOM device (hardware-software testing equipment manufactured by research-and-production company “Dinamika”). However, the lack of available methodological materials on the selection of model parameters for commissioning of a specific device complicates the performance of such work. Thus, the relevant task is to obtain easy-to-use means of checking OOSP. The means should allow to simulate power system in the loss of synchronism conditions and in the case of stable power swings. The creation of such tools would not only reduce labor costs, but also improve the quality of commissioning. The widespread use of such approach is recommended by the standard [15]. To E3S Web of Conferences 220, 01081 (2020) SES-2020 E3S Web of Conferences 220, 01081 (2020) SES-2020 https://doi.org/10.1051/e3sconf/202022001081 2.2 Mathematical model for testing distance OOSP The maximum Imax and minimum Imin values of the current are derived from the given settings of the protection [3] among the values describing behavior of asynchronous running and the behavior of protection in time domain (for example, duration of a cycle, time delay before tripping, and etc.). The parameters Z1 and Z2 are set by the user at will. Then, with the given values of Imax, Imin and Z1, Z2, we obtain the following relation for the model under consideration, which makes it possible to determine the value of E1: For the assumed model of the power system (Fig. 1), the time dependence of the measured impedance vector Zp at the place of installation of the protection is determined as [18]: ( ) ( )    = − + = . (t) E (t) E (t) q ; (t) q Z (t) q Z р Z 1 2 1 1 2 (3) (3) With a constant ratio of the EMFs moduli, the locus of the Zp corresponds to a circle on the complex plane with a radius R0 and a center Z0: ( ) ( ) ( ) ( ) ( )     = − +  = + − = . ; 1 ; 1 2 2 1 max 1 min max min max E q E q Z Z I E I I I I q (2)        = = − + − =     −     + − = . 1 2 ; 1 2 2 1 0 ; 1 2 2 1 2 0 const E E q q Z Z q R q Z Z q Z (4) (2) (4) Here, the sum of resistances to the right (Z2) and to the left (Z1) from the place of installation of the protection P is taken as a constant, and therefore does not affect the formation of instantaneous values of currents. Let’s suppose that the values q, R0, Z0 are determined (the parameter q is determined arbitrarily, and the remaining parameters are uniquely determined by the protection settings), then the values of Z1 and Z2 can be found by expanding the modules in the system of equations (4). 2 Materials and Methods Distance OOSP is the most common type of OOSP in Russian power systems [16]. Current OOSP are also somewhat widespread. For this reason, the work addresses the means of checking current and distance OOSP. The protection should form a control action aimed at tie-tripping only in the case when only two parts of the power system loss the synchronism (note that none of the generators in these parts experience loss of excitation) [17]. Therefore, to compile a mathematical description of the considered conditions, a basic model of an electrical network consisting of two power systems is used (Fig. 1). In Fig. 1, the following designations are assumed: E1, E2 are EMFs of the two power systems; Z1, Z2 are complex impedances from the side of power system 1 and 2 respectively; the protection P is between the power systems and controls the current Ir flowing through a tie-line. Z1 P Z2 E2 E1 Ir Fig. 1.Electrical drawing of the power system for the calculations. Fig. 1.Electrical drawing of the power system for the calculations. https://doi.org/10.1051/e3sconf/202022001081 E3S Web of Conferences 220, 01081 (2020) SES-2020 E3S Web of Conferences 220, 01081 (2020) 2.1 The development of the mathematical model for checking the current OOSP the matrices the instantaneous values of currents and voltages for normal and emergency conditions are derived afterwards. The features of the implementation of OOSP algorithms differ depending on the manufacturer. Therefore, from this point onward, it is assumed that we deal with OOSP algorithms developed by Institute of Power System Automation JSC [3]. To use the method for different algorithms, the described approach can be slightly adjusted without changing its basic essence. The obtained instantaneous values of currents and voltages are stored in the COMTRADE format which allows to reproduce the obtained electrical signals by using a software-and-hardware testing equipment of any manufacturer (the only requirement is that equipment has a function to read COMTRADE files and generate the mentioned signals which is a common function for such devices).With sufficient information about a specific protection algorithm and its settings it is possible to predict the behavior of the protection under test. When checking the correct functioning of the current OOSP, it is necessary to establish the following: reliable inaction of the protection in case when the value of the current flowing through the measuring unit is equal to 90% (or lower)of the set pick-up value; reliable operation of the protection when the RMS current value is equal to 110% (or higher) of the set pick-up value; the protection must not operate if current lowers below the drop-away setting (when it previously raised above the pick-up value) and vice-versa. 2.2 Mathematical model for testing distance OOSP Distance OOSP controls the behavior of the vector of complex impedance, measured at the place of installation of the protection P. When checking such an OOSP, it is necessary to make sure of its reliable non-operation in case of external asynchronous run or in case of stable power swings and a short-circuit fault and reliable operation in case of internal asynchronous run with a given slip period. For the computational model (Fig. 1), we assume that under the conditions of the loss of synchronism, the EMF vector E1 rotates relative to the vector E2, i.e. the angle δ1 of the EMF vector E1 over time remains unchanged, and the angle δ2 of the EMF vector E2 changes over time according to the law: t, (t) s ω 10 δ 1 δ  = (1) Unlike the process of obtaining the method for checking the current OOSP, where the setpoint is set directly for the controlled value, when considering the formation of instantaneous values of currents and voltages for testing a distance OOSP, it is important to control the location of the locus of the impedance vector Zp on the complex plane. (1) where δ10 is the initial value of the phase of the vector E1, rad; ωs is the slip frequency, rad/sec. In the process of loss of synchronism, the current through the power transmission line changes from a certain maximum Imax to a minimum value Imin. The maximum Imax and minimum Imin values of the current are derived from the given settings of the protection [3] among the values describing behavior of asynchronous running and the behavior of protection in time domain (for example, duration of a cycle, time delay before tripping, and etc.). The parameters Z1 and Z2 are set by the user at will. Then, with the given values of Imax, Imin and Z1, Z2, we obtain the following relation for the model under consideration, which makes it possible to determine the value of E1: where δ10 is the initial value of the phase of the vector E1, rad; ωs is the slip frequency, rad/sec. In the process of loss of synchronism, the current through the power transmission line changes from a certain maximum Imax to a minimum value Imin. 1 2 –39.53 17.68 , 47.74 – 17.68 . Z j Ohms Z j Ohms = + = (5) New values of these impedances after transformations (8) are equal to: 1 2 ' –40.45 – 15.45 , ' 46.26 21.26 . Z j Ohms Z j Ohms = = + The obtained values of complex impedances Z'1 and Z'2 are substituted into the power system model (Fig.1) and the loss of synchronism simulation begins according to equation (1). The result of the simulation is the time dependence of the vector values of the voltage supplied to the relay Up (t) and the measured current Ip (t). From this, instantaneous values of current and voltage are formed(as imaginary components of complex values for the considered moment in time t) and COMTRADE files format are made. (6) q System (4), with the impedance values found according to either (5) or (6) makes it possible to obtain a circular locus, however, the region with the highest density of samples (corresponds to the region where the speed of movement of the impedance vector along the locus is the lowest) cannot be determined by a user. In the example under consideration, maximum RMS values of currents and voltages are 1.17 Amps and 55.49 V, respectively (the value of EMF E1 equal to 7 V was taken). These values can be changes while the locus parameters remain the same. The way to do so is to change either parameter q, or the value of EMF E1, or use system of equation (5) instead of (6). The desired distribution of samples on the locus can be achieved in 2 stages, making the following transformations. First, when calculating the values of impedances Z1, Z2 according to (5) or (6), instead of the previously used value of Z0, it is necessary to substitute its corrected value Z'0which is equal to One of the important parameters of the described method is the correction angle φcor, it allows to set the region where the speed of impedance vector is the lowest, as it’s illustrated in Fig. 2, 3. cor j e Z Z  0 0' = (7) (7) -50 0 50 100 150 200 -50 0 50 100 150 200 Z0 R Direction of spin First sample Reactive impedance, Ohms Active impedance, Ohms Fig. 2. Impedance locus with given characteristics, φcor = –45°. 1 2 –39.53 17.68 , 47.74 – 17.68 . Z j Ohms Z j Ohms = + = where φcor is the correction angle with which the desired location of the maximum sample density on the locus is shifted along the locus relative to the initially obtained location. Secondly, in the formula (4) it is necessary to substitute the values of resistances Z1, Z2 not obtained by (5) or (6), but their corrected values Z'1, Z'2, which are:       − = − = . 2 ' 2 ; 1 ' 1 cor j e Z Z cor j e Z Z   (8) (8) Let's consider a numerical example. Suppose it is necessary to obtain the impedance locus centered at the point Z0 = 100 + j75 Ohm and with the radius R = 80 Ohm. Let us also take q = 0.95 and φcor = –45 degrees, and calculate the impedance by expression (6). Then, in accordance with the procedure presented above, we get: Fig. 2. Impedance locus with given characteristics, φcor = –45°. Another test that has to be performed when checking the settings of distance OOSP is its inaction the during stable power swings in the power system. For a comprehensive check, it is necessary to make changes to redefine the value of the angle of EMF E1. Another test that has to be performed when checking the settings of distance OOSP is its inaction the during stable power swings in the power system. For a comprehensive check, it is necessary to make changes to redefine the value of the angle of EMF E1. Assume that at the initial moment of time the phase shift between the vectors E1 and E2 is δ12, while in the process of power swings the angle δ1 deviates from its initial position by δ1m clockwise and by δ2m counterclockwise, as shown in Fig. 4a. . Ohms 68 . 17 74 . 123 0 ' 0 j cor j e Z Z − = =  Assume that at the initial moment of time the phase shift between the vectors E1 and E2 is δ12, while in the process of power swings the angle δ1 deviates from its initial position by δ1m clockwise and by δ2m counterclockwise, as shown in Fig. 4a. 2.2 Mathematical model for testing distance OOSP The sought values of Z1 and Z2 are satisfied by solutions of two systems of equations: In the considered model, all the parameters are now determined, which allows (taking into account the required duration of normal and emergency conditions, and also taking into account the specified parameters of the operation of the protection in time) to obtain the matrices of vector values of EMFs, currents and voltages in the place of the installation of the device P. From this, 2 2 https://doi.org/10.1051/e3sconf/202022001081 E3S Web of Conferences 220, 01081 (2020) E3S Web of Conferences 220, 01081 (2020) SES-2020          − − − =           − + − − − = .1 1 2 0 2 ; 1 2 0 )1 2 ( 0 1 2 1 1 Z q q R Z q q R q Z q Z (5)          − − =           − − − − − = .1 1 2 0 2 ; 1 2 0 )1 2 ( 0 1 2 1 1 Z q q R Z q q R q Z q Z (6) 1 2 –39.53 17.68 , 47.74 – 17.68 . Z j Ohms Z j Ohms = + = 1 2 –39.53 17.68 , 47.74 – 17.68 . Z j Ohms Z j Ohms = + = Assume that at the initial moment of time the phase shift between the vectors E1 and E2 is δ12, while in the process of power swings the angle δ1 deviates from its initial position by δ1m clockwise and by δ2m counterclockwise, as shown in Fig. 4a. The values of Z1, Z2 calculated according to (6) for the given value of Z'0 are equal to: 3 E3S Web of Conferences 220, 01081 (2020) E3S Web of Conferences 220, 01081 (2020) SES-2020 https://doi.org/10.1051/e3sconf/202022001081 -200 -150 -100 -50 0 50 100 150 200 -50 0 50 100 150 200 250 300 Sensitive distance relay characteristic Coarse distance relay characteristic Directional element Imaginary impedance, Ohms Real impedance, Ohms First sample Fig. 5. Impedance locus obtained by simulating power swings in the case of δ1 = 179°, δ2 = 100°, δ12 = 1° and φcor = –81°. -200 -150 -100 -50 0 50 100 150 200 -50 0 50 100 150 200 250 300 Sensitive distance relay characteristic Coarse distance relay characteristic Directional element Imaginary impedance, Ohms Real impedance, Ohms First sample Fig. 5. Impedance locus obtained by simulating power swings in the case of δ1 = 179°, δ2 = 100°, δ12 = 1° and φcor = –81°. -50 0 50 100 150 200 -50 0 50 100 150 200 First sample Z0 R Active impedance, Ohms Reactive impedance, Ohms Direction of spin Fig. 3.Impedance locus with given characteristics, φcor = 135°. -50 0 50 100 150 200 -50 0 50 100 150 200 First sample Z0 R Active impedance, Ohms Reactive impedance, Ohms Direction of spin Fig. 3.Impedance locus with given characteristics, φcor = 135°. Reactive impedance, Ohms First sample Fig. 3.Impedance locus with given characteristics, φcor = 135°. Fig. 5. Impedance locus obtained by simulating power swings in the case of δ1 = 179°, δ2 = 100°, δ12 = 1° and φcor = –81°. Suppose that the change in the angle δ1 occurs according to a sinusoidal law, as shown in Fig. 4b. In this case, the law of variation of the angle δ1 (t) looks as follows (9): 3 Results and Discussion Experience has shown the possibility of using the developed method in practice. Experimental studies of the current-type and distance-type OOSP were carried out on the basis of the multifunctional emergency control system series KPA-M produced by Power Systems Automation Institute JSC using the RETOM software- hardware testing equipment. Analysis of the data recorded by the terminal as a result of the experiments carried out confirms the complete correctness of the considered method, as well as the similarity of the parameters of the simulated locus with the real one observed in the power systems. Some experimental results are shown in Fig. 6-10. Thus, proposed means have more functional abilities and are more easy-to-use that presented in [19] or in [20]. ( ) ( ) ( )       + = − = − + + = ), 2 1 ( 5,0 1 ), 2 1 ( 5,0 0 , 1 / 0 arcsin 2 sin 1 12 0 ) ( 1 m m A m m A A A t sw f A A t        (9) (9) where the parameter fsw characterizes the speed of power swings. In system (9), it is taken into account that the numerical values of δ1m and δ2m should be substituted with a plus sign, then the upper limit of the angle δ1 will be δ12 + δ1m, and the lower one is, respectively, δ12– δ2m. δ12 t δ1 (t) 0 A0 а) b) δ12+δ1m δ12-δ2m A1 E2 E1 δ12 δ2m δ1m Fig. 4. The change in the mutual angle between power systems during power swings: a) phasor diagram; b) change in δ1 over time. δ12 t δ1 (t) 0 A0 b) δ12+δ1m δ12-δ2m A1 а) E2 E1 δ12 δ2m δ1m Fig. 6.Visualization of simulation results in the MatLab software: operation of the current OOSP during the check of pick-up current in specified time delay after the beginning of asynchronous running. а) b) Fig. 4. The change in the mutual angle between power systems during power swings: a) phasor diagram; b) change in δ1 over time. An example of a locus of power swings is shown in Fig. 5, which also shows the settings of distance relay of OOSP in accordance with the algorithm [3]. Fig. 6.Visualization of simulation results in the MatLab Fig. 3 Results and Discussion 6.Visualization of simulation results in the MatLab software: operation of the current OOSP during the check of pick-up current in specified time delay after the beginning of asynchronous running. Fig. 6.Visualization of simulation results in the MatLab software: operation of the current OOSP during the check of pick-up current in specified time delay after the beginning of asynchronous running. 4 4 https://doi.org/10.1051/e3sconf/202022001081 E3S Web of Conferences 220, 01081 (2020) E3S Web of Conferences 220, 01081 (2020) SES-2020 Fig. 7.The record of the test report in the specialized FastView software: the operation of current OOSP after the specified cycles of asynchronous run at the certain pick-up current setting. Fig. 8.The oscillography of current and voltages forming impedance locus shown in Fig.2. The record is processed with the tools available in the specialized FastView software. Currents and voltages are on the left side of the figure; the locus of positive sequence impedance is presented on the right side. Fig. 7.The record of the test report in the specialized FastView software: the operation of current OOSP after the specified cycles of asynchronous run at the certain pick-up current setting. specific type of test equipment, the only requirement for which is the ability to read and reproduce signals saved in a COMTRADE file. The proposed algorithms for the formation of input signals for checking the OOSP are now actively used in the commissioning tasks of emergency control devices series KPA-M. This use significantly reduces the time required to complete commissioning and improves its quality. References 9.The impedance locus recorded by digital terminal KPA- M (the reddots on the locus show the sequential position of the impedance vector Zp at time intervals equal to the terminal sampling period): checking of the settings of the distance OSSP device installed at Ust-Ilimsk HPP (left) and the locus obtain in the process of certification trial of the distance OOSP device by using a RTDS (right). Fig. 9.The impedance locus recorded by digital terminal KPA- M (the reddots on the locus show the sequential position of the impedance vector Zp at time intervals equal to the terminal sampling period): checking of the settings of the distance OSSP device installed at Ust-Ilimsk HPP (left) and the locus obtain in the process of certification trial of the distance OOSP device by using a RTDS (right). [8] Digital device of out-of-step protection, Scientific and technical center of the Unified Energy System [Electronic resource] Available at: https://ntcees.ru/departments/products/alar3.php [9] [9] Operation manual series TIYaK.648229.001.RE: Digital automation for elimination of out-of-step operation ALAR-Ts, Modifications ALAR-Ts-02, ALAR-Ts-03, ALAR-Ts-04 (Saint Petersburg, NIIPT, 2008) References Fig. 7.The record of the test report in the specialized FastView software: the operation of current OOSP after the specified cycles of asynchronous run at the certain pick-up current setting. Fig. 8.The oscillography of current and voltages forming impedance locus shown in Fig.2. The record is processed with the tools available in the specialized FastView software. Currents and voltages are on the left side of the figure; the locus of positive sequence impedance is presented on the right side. Fig. 7.The record of the test report in the specialized FastView software: the operation of current OOSP after the specified cycles of asynchronous run at the certain pick-up current setting. [1] Y.E. Gonik, E.S. Iglitsky, Out-of-step protection (Moscow, Publishing house «Energoatomizdat», 1988) 112 [2] V.G. Narovlyansky, A.B. Vaganov, Application of the ALAR-M device to identify and eliminate the asynchronous mode of the electric power system: an article in scientific, Energetic (Moscow, Publishing house «Energoprogress», 2011) 17-20 [3] [3] Multifunctional emergency control complex KPA-M-03.04.05.05.06.09.11.15.-10001-UHL4 [Electronic resource], User’s manual, Power Systems Automation Institute JSC (2017) Available at: http://iaes.ru/uploads/pages/files/iaes_local_re20 8.pdf [4] [4] Collection of technical descriptions, Complex of devices for emergency automatics, Relematika LLC [Electronic resource] Available at: https://storage.energybase.ru/source/112/NzJaiiW qUK_PCVy98EJDPsldFTprmVs3.pdf Fig. 8.The oscillography of current and voltages forming impedance locus shown in Fig.2. The record is processed with the tools available in the specialized FastView software. Currents and voltages are on the left side of the figure; the locus of positive sequence impedance is presented on the right side. [5] [5] A.P. Maliy, A.A. Shurupov, S.A. Ivanov, et al., Implementation of automatic equipment for opt- ofstep protection in terminals of the ShE2607 series, Relay protection and automation 1, 1, 39- 52 (2010) Fig. 9.The impedance locus recorded by digital terminal KPA- M (the reddots on the locus show the sequential position of the impedance vector Zp at time intervals equal to the terminal sampling period): checking of the settings of the distance OSSP device installed at Ust-Ilimsk HPP (left) and the locus obtain in the process of certification trial of the distance OOSP device by using a RTDS (right). [6] Multifunction Generator, Motor and Transformer Protection Relay SIPROTEC4-7UM62: Siemens AG, User’s manual, Catalog SIP6.2 (Germany, 2001) 556 [7] V.G. Narovlyansky, Modern methods and means of preventing the out-of-step operation of the electric power system (Moscow, Publishing house «Energoatomizdat», 2004) 360 Fig. 9.The impedance locus recorded by digital terminal KPA- Fig. 4 Conclusion The developed algorithms form a full-fledged basis for checking the settings of OOSP of different type in case of the asynchronous run and power swings. [10] Demetrios Tziouvaras, Daqing Hou, Out-of-Step Protection Fundamentals and Advancements, 30th Annual Western Protective Relay It is important that the given approach makes it possible to test OOSP devices without being bound to a 5 E3S Web of Conferences 220, 01081 (2020) E3S Web of Conferences 220, 01081 (2020) SES-2020 https://doi.org/10.1051/e3sconf/202022001081 Conference, Spokane, Washington (21-23 October 2003) [11] G. Benmouyal, D. Hou, D. Tziouvaras, Zero- setting Power-Swing Blocking Protection, 31st Annual Western Protective Relay Conference, Spokane, Washington, 27 (19-21 October 2004) [12] J. Blumschein, Y. Yelgin, M. Kereit, Proper detection and treatment of power swing to reduce the risk of Blackouts, Siemens AG, Energy (2011) [13] Dr. Jürgen Holbach, Siemens AG, Raleigh, NC, USA: New Out Of Step Blocking Algorithm for Detecting Fast Power Swing Frequencies, 30th Annual Western Protective Relay Conference, Washington State University, Spokane, Washington, (21-23 October 2003) [14] Typical instruction on the organization and production of work in relay protection devices and electrical automation of substations [Electronic resource] Available at: https://www.fsk-ees.ru/upload/docs/56947007- 29.240.30.004-2008.pdf [15] [15] N.A. Volokhov, Investigation of the impact on the operation of ALAR devices of disturbances arising after the start of asynchronous mode. Electric power industry through the eyes of youth-2019: materials of the 10th International Scientific and Technical Conference, Irkutsk (Irkutsk, Irkutsk National Research Technical University, 16-20 September 2019) 243-246 [16] [16] Organization standard of System Operator of the Unified Energy System JSC, Relay protection and automation, Automatic emergency control of the power system, Out-of-step protection, Norms and requirements [Electronic resource] Available at: https://so- ups.ru/fileadmin/files/laws/standards/st_rza_alar_ 241215.pdf [17] E.M. Shneerson, Digital relay protection (Moscow, Publishing house «Energoatomizdat», 2007) 549 [18] Q. Verzosa, Realistic testing of power swing blocking and out-of-step tripping functions, 2013 66th Annual Conference for Protective Relay Engineers, College Station, TX, 420-449 (2013) [19] Operation manual: RETOM-61 software and hardware test complex (Cheboksary, Research and Production enterprise «Dinamika») 219 6 6
https://openalex.org/W2907323237	https://europepmc.org/articles/pmc6337303?pdf=render	English	null	Recent Developments in Tandem White Organic Light-Emitting Diodes	Molecules/Molecules online/Molecules annual	2,019	cc-by	21,531	Peng Xiao 1, Junhua Huang 1, Yicong Yu 1,* and Baiquan Liu 2,3,* 1 School of Physics and Optoelectronic Engineering, Foshan University, Foshan 528000, China; xiaopeng@fosu.edu.cn (P.X.); jamha1212@163.com (J.H.) 1 School of Physics and Optoelectronic Engineering, Foshan University, Foshan 528000, China; xiaopeng@fosu.edu.cn (P.X.); jamha1212@163.com (J.H.) 2 LUMINOUS! Centre of Excellent for Semiconductor Lighting and Displays, School of Electrical and Electronic Engineering, Nanyang Technological University, Nanyang Avenue, Singapore 639798, Singapore 3 Institute of Polymer Optoelectronic Materials and Devices, State Key Laboratory of Luminescent Materials and Devices, South China University of Technology, Guangzhou 510640, China * Correspondence: yicong2007@gmail.com (Y.Y.); l.baiquan@mail.scut.edu.cn (B.L.) R i d 7 D b 2018 A t d 25 D b 2018 P bli h d 2 J 2019 1 School of Physics and Optoelectronic Engineering, Foshan University, Foshan 528000, China; xiaopeng@fosu.edu.cn (P.X.); jamha1212@163.com (J.H.) 2 LUMINOUS! Centre of Excellent for Semiconductor Lighting and Displays, School of Electrical and Electronic Engineering, Nanyang Technological University, Nanyang Avenue, Singapore 639798, Singapore 3 Institute of Polymer Optoelectronic Materials and Devices, State Key Laboratory of Luminescent Materials and Devices, South China University of Technology, Guangzhou 510640, China * Correspondence: yicong2007@gmail.com (Y.Y.); l.baiquan@mail.scut.edu.cn (B.L.) Received: 7 December 2018; Accepted: 25 December 2018; Published: 2 January 2019 Received: 7 December 2018; Accepted: 25 December 2018; Published: 2 January 2019 Abstract: Tandem white organic light-emitting diodes (WOLEDs) are promising for the lighting and displays ﬁeld since their current efﬁciency, external quantum efﬁciency and lifetime can be strikingly enhanced compared with single-unit devices. In this invited review, we have ﬁrstly described fundamental concepts of tandem device architectures and their use in WOLEDs. Then, we have summarized the state-of-the-art strategies to achieve high-performance tandem WOLEDs in recent years. Speciﬁcally, we have highlighted the developments in the four types of tandem WOLEDs (i.e., tandem ﬂuorescent WOLEDs, tandem phosphorescent WOLEDs, tandem thermally activated delayed ﬂuorescent WOLEDs, and tandem hybrid WOLEDs). Furthermore, we have introduced doping-free tandem WOLEDs. In the end, we have given an outlook for the future development of tandem WOLEDs. Keywords: tandem; organic light-emitting diode; white; charge generation layer; doping-free molecules molecules Recent Developments in Tandem White Organic Light-Emitting Diodes Peng Xiao 1, Junhua Huang 1, Yicong Yu 1,* and Baiquan Liu 2,3,* molecules Review Recent Developments in Tandem White Organic Light-Emitting Diodes Peng Xiao 1, Junhua Huang 1, Yicong Yu 1,* and Baiquan Liu 2,3,* 1 School of Physics and Optoelectronic Engineering, Foshan University, Foshan 528000, China; xiaopeng@fosu.edu.cn (P.X.); jamha1212@163.com (J.H.) 2 LUMINOUS! Centre of Excellent for Semiconductor Lighting and Displays, School of Electrical and Electronic Engineering, Nanyang Technological University, Nanyang Avenue, Singapore 639798, Singapore 3 Institute of Polymer Optoelectronic Materials and Devices, State Key Laboratory of Luminescent Materials and Devices, South China University of Technology, Guangzhou 510640, China * Correspondence: yicong2007@gmail.com (Y.Y.); l.baiquan@mail.scut.edu.cn (B.L.) Received: 7 December 2018; Accepted: 25 December 2018; Published: 2 January 2019 Abstract: Tandem white organic light-emitting diodes (WOLEDs) are promising for the lighting and displays ﬁeld since their current efﬁciency, external quantum efﬁciency and lifetime can be strikingly enhanced compared with single-unit devices. In this invited review, we have ﬁrstly described fundamental concepts of tandem device architectures and their use in WOLEDs. Then, we have summarized the state-of-the-art strategies to achieve high-performance tandem WOLEDs in recent years. Speciﬁcally, we have highlighted the developments in the four types of tandem WOLEDs (i.e., tandem ﬂuorescent WOLEDs, tandem phosphorescent WOLEDs, tandem thermally activated delayed ﬂuorescent WOLEDs, and tandem hybrid WOLEDs). Furthermore, we have introduced doping-free tandem WOLEDs. In the end, we have given an outlook for the future development of tandem WOLEDs. Keywords: tandem; organic light-emitting diode; white; charge generation layer; doping-free Besides, lighting, which may be able to compete with GaN-based ossess many merits, including high efﬁciency, low power esponse, thin thickness, solution-processed compatibility as ion of emitters and enhancement of device engineering, both delayed ﬂuorescence (TADF) materials-based OLEDs can um efﬁciency (IQE) due to the harvest of totally singlet and cent materials, they harness singlets and triplets due to the ADF emitters, a small energy gap for triplet excited state (T1) which is beneﬁcial to the reverse intersystem process for the ]. ology for high-quality displays and energy-saving lighting tly attracted both industrial and academic interest [27–31]. y Kido and coworkers, showing a maximum power efﬁciency t twenty-four years, the performance of WOLEDs has been rescent emitters and advanced outcoupling technique, the PE W−1 at the illumination-related luminance of 1000 cd m−2 with molecules molecules Review Recent Developments in Tandem White Organic Light-Emitting Diodes Peng Xiao 1, Junhua Huang 1, Yicong Yu 1,* and Baiquan Liu 2,3,* 1 School of Physics and Optoelectronic Engineering, Foshan University, Foshan 528000, China; xiaopeng@fosu.edu.cn (P.X.); jamha1212@163.com (J.H.) 2 LUMINOUS! Centre of Excellent for Semiconductor Lighting and Displays, School of Electrical and Electronic Engineering, Nanyang Technological University, Nanyang Avenue, Singapore 639798, Singapore 3 Institute of Polymer Optoelectronic Materials and Devices, State Key Laboratory of Luminescent Materials and Devices, South China University of Technology, Guangzhou 510640, China * Correspondence: yicong2007@gmail.com (Y.Y.); l.baiquan@mail.scut.edu.cn (B.L.) Received: 7 December 2018; Accepted: 25 December 2018; Published: 2 January 2019 Abstract: Tandem white organic light-emitting diodes (WOLEDs) are promising for the lighting and displays ﬁeld since their current efﬁciency, external quantum efﬁciency and lifetime can be strikingly enhanced compared with single-unit devices. In this invited review, we have ﬁrstly described fundamental concepts of tandem device architectures and their use in WOLEDs. Then, we have summarized the state-of-the-art strategies to achieve high-performance tandem WOLEDs in recent years. Speciﬁcally, we have highlighted the developments in the four types of tandem WOLEDs (i.e., tandem ﬂuorescent WOLEDs, tandem phosphorescent WOLEDs, tandem thermally activated delayed ﬂuorescent WOLEDs, and tandem hybrid WOLEDs). Furthermore, we have introduced doping-free tandem WOLEDs. In the end, we have given an outlook for the future development of tandem WOLEDs. Keywords: tandem; organic light-emitting diode; white; charge generation layer; doping-free molecules Review Recent Developments in Tandem White Organic Light-Emitting Diodes Peng Xiao 1, Junhua Huang 1, Yicong Yu 1,* and Baiquan Liu 2,3,* 1 School of Physics and Optoelectronic Engineering, Foshan University, Foshan 528000, China; xiaopeng@fosu.edu.cn (P.X.); jamha1212@163.com (J.H.) 2 LUMINOUS! Centre of Excellent for Semiconductor Lighting and Displays, School of Electrical and Electronic Engineering, Nanyang Technological University, Nanyang Avenue, Singapore 639798, Singapore 3 Institute of Polymer Optoelectronic Materials and Devices, State Key Laboratory of Luminescent Materials and Devices, South China University of Technology, Guangzhou 510640, China * Correspondence: yicong2007@gmail.com (Y.Y.); l.baiquan@mail.scut.edu.cn (B.L.) Received: 7 December 2018; Accepted: 25 December 2018; Published: 2 January 2019 Abstract: Tandem white organic light-emitting diodes (WOLEDs) are promising for the lighting and displays ﬁeld since their current efﬁciency, external quantum efﬁciency and lifetime can be strikingly enhanced compared with single-unit devices. In this invited review, we have ﬁrstly described fundamental concepts of tandem device architectures and their use in WOLEDs. Then, we have summarized the state-of-the-art strategies to achieve high-performance tandem WOLEDs in recent years. Speciﬁcally, we have highlighted the developments in the four types of tandem WOLEDs (i.e., tandem ﬂuorescent WOLEDs, tandem phosphorescent WOLEDs, tandem thermally activated delayed ﬂuorescent WOLEDs, and tandem hybrid WOLEDs). Furthermore, we have introduced doping-free tandem WOLEDs. In the end, we have given an outlook for the future development of tandem WOLEDs. Keywords: tandem; organic light-emitting diode; white; charge generation layer; doping-free in Tandem White Organic s 1,* and Baiquan Liu 2,3,* gineering, Foshan University, Foshan 528000, China; @163.com (J.H.) miconductor Lighting and Displays, School of Electrical and logical University, Nanyang Avenue, Singapore 639798, Singapore erials and Devices, State Key Laboratory of Luminescent Materials Technology, Guangzhou 510640, China m (Y.Y.); l.baiquan@mail.scut.edu.cn (B.L.) ecember 2018; Published: 2 January 2019 mitting diodes (WOLEDs) are promising for the lighting and cy, external quantum efﬁciency and lifetime can be strikingly devices. In this invited review, we have ﬁrstly described ce architectures and their use in WOLEDs. Then, we have es to achieve high-performance tandem WOLEDs in recent d the developments in the four types of tandem WOLEDs dem phosphorescent WOLEDs, tandem thermally activated dem hybrid WOLEDs). Furthermore, we have introduced nd, we have given an outlook for the future development of ng diode; white; charge generation layer; doping-free ng diodes (OLEDs) have entered the mainstream display performance with the liquid crystal displays [1–5]. 1. Introduction In recent years, organic light-emitting diodes (OLEDs) have entered the mainstream display market, since they can show comparable performance with the liquid crystal displays [1–5]. Besides, OLEDs are promising for the solid-state lighting, which may be able to compete with GaN-based LEDs [6–10]. This is because OLEDs possess many merits, including high efﬁciency, low power consumption, broad viewing angle, fast response, thin thickness, solution-processed compatibility as well as ﬂexibility [11–16]. With the evolution of emitters and enhancement of device engineering, both phosphorescent and thermally activated delayed ﬂuorescence (TADF) materials-based OLEDs can realize a theoretical unity internal quantum efﬁciency (IQE) due to the harvest of totally singlet and triplet excitons [17–20]. For phosphorescent materials, they harness singlets and triplets due to the heavy-atom effect [21–23]. In terms of TADF emitters, a small energy gap for triplet excited state (T1) and singlet excited state (S1) is required, which is beneﬁcial to the reverse intersystem process for the 100% exciton harvesting efﬁciency [24–26]. g y To further promote the OLED technology for high-quality displays and energy-saving lighting ﬁeld, white OLEDs (WOLEDs) have vastly attracted both industrial and academic interest [27–31]. In 1994, the ﬁrst WOLED was developed by Kido and coworkers, showing a maximum power efﬁciency (PE) of 0.83 lm W−1 [32,33]. Over the last twenty-four years, the performance of WOLEDs has been remarkably enhanced. By dint of phosphorescent emitters and advanced outcoupling technique, the PE of WOLEDs could show a PE of 123.4 lm W−1 at the illumination-related luminance of 1000 cd m−2 with www.mdpi.com/journal/molecules Molecules 2019, 24, 151; doi:10.3390/molecules24010151 www.mdpi.com/journal/molecules 2 of 28 Molecules 2019, 24, 151 an external quantum efﬁciency (EQE) of 54.6%, which remains 106.5 lm W−1 at an ultrahigh luminance of 5000 cd m−2 [34]. Even for ﬂexible phosphorescent WOLEDs, their PE can overtake 100 lm W−1 [35]. Currently, the EQE of TADF emitter-based WOLEDs has also been demonstrated to exceed 20% [36–39]. Besides, the luminance >100,000 cd m−2 [40], color rendering index (CRI) >90 [41–46], correlated color temperature (CCT) mimicking sunlight (2500–8000 K) [47–49], and extremely stable color without variation of Commission International de L’Eclairage (CIE) chromaticity coordinates for WOLEDs have also been investigated [50–52]. Hence, these superior properties have demonstrated the great potential of the WOLED technology in the application of displays and lighting. 1. Introduction For the real commercialization, the lifetime is an essential parameter for WOLEDs (i.e., ≥10,000 h at ≥1000 cd m−2), aside from the general target for the PE of 40–70 lm W−1 [53–58]. In the case of ﬂuorescent WOLEDs, their lifetime could overtake 150,000 h at 1000 cd m−2 [59]. However, the EQE, current efﬁciency (CE) and PE are unsatisfactory. To boost the lifetime and efﬁciency simultaneously, the introduction of tandem device architectures is very helpful [60–64]. This is because the lifetime of tandem OLEDs can be N-fold (N, the number of electroluminescent (EL) units) enhanced, since N-fold luminance can be achieved with the current density similar to that of single-unit OLEDs [65–68]. As a consequence, the EQE and CE can be also N-fold improved. With the careful manipulation of charge generation layers (CGLs), the PE is possible to be increased [69–71]. Therefore, tandem OLEDs have aroused both academic and industry interest. In 2005, Ma et al. reported the ﬁrst tandem WOLEDs, where the CE and brightness equal basically to the sum of the two EL units [72]. Since then, a large number of endeavors have been taken to tandem WOLEDs [73–77]. As a matter of fact, the current commercial WOLEDs are almost based on tandem architectures. With the development of tandem WOLEDs, the ﬂuorescent, phosphorescent and TADF emitters have been well connected with CGLs to realize desirable performance. Herein, the fundamental concepts of tandem device architectures and their use in WOLEDs will be ﬁrstly described. Then, the state-of-the-art strategies to achieve high-performance tandem WOLEDs in recent years will be summarized. Speciﬁcally, the research progresses in the four types of tandem WOLEDs (i.e., tandem ﬂuorescent WOLEDs, tandem phosphorescent WOLEDs, tandem TADF WOLEDs, and tandem hybrid WOLEDs) will be highlighted. Furthermore, doping-free tandem WOLEDs will be introduced. Finally, an outlook for the future development of tandem WOLEDs will be presented. 2.1. The Device Architecture of Tandem OLEDs To furnish white emissions, complementary-color emitters, three-color emitters or four-color emitters (e.g., blue/yellow, blue/orange, blue/red, blue/green/red, blue/yellow/red, blue/green/yellow/red) are commonly required [78–85]. For single-unit WOLEDs, the emitting layers (EMLs) can be composed of single or multiple EMLs. In general, single-EML WOLEDs only need one EML comprised of a versatile host doped with different color guests, while multiple-EML WOLEDs require at least two EMLs to produce white emissions [86–91]. Accordingly, the EL unit of tandem WOLEDs could be comprised of non-white or monochromatic EMLs, where various-color EMLs are located in different EL unit to combine for the white emission (type-I tandem WOLED architecture), as shown in Figure 1a. Besides, each unit of tandem of WOLEDs can be made up of multiple-EML white unit (type-II tandem WOLED architecture, Figure 1b) or single-EML white unit (type-III tandem WOLED architecture, Figure 1c). Alternatively, two unit-based tandem WOLEDs can be constructed by a white unit and a non-white unit (type-IV tandem WOLED architecture, Figure 1d). Based on the above facts, highly efﬁcient phosphorescent and TADF emitters are usually adopted in tandem WOLEDs, which is expected to harvest both singlet and triplet excitons for the high performance [92–97]. After the optimization of EMLs, CGLs are selected to interconnect each EL unit. 3 of 28 igh ach Molecules 2019, 24, 151 tandem WOLED f [92 As a result, tandem WOLEDs will be formed. It is deserved to note that most of tandem WOLEDs are focused on two or three EL units, although more than four units can be connected with CGLs. EL unit. As a result, tandem WOLEDs will be formed. It is deserved to note that most of tandem WOLEDs are focused on two or three EL units, although more than four units can be connected with CGLs. As a result, tandem WOLEDs will be formed. It is deserved to note that most of tandem WOLEDs are focused on two or three EL units, although more than four units can be connected with CGLs. EL unit. As a result, tandem WOLEDs will be formed. It is deserved to note that most of tandem WOLEDs are focused on two or three EL units, although more than four units can be connected with CGLs. Figure 1. Typical device architectures for tandem white organic light-emitting diodes (WOLEDs). 2.1. The Device Architecture of Tandem OLEDs (b) Type-II tandem WOLED architecture, where RYGB emitting layers (EMLs) mean that more than two EMLs generating white emission in each EL unit. (c) Type-III tandem WOLED architecture, where each EML of the EL unit is single white EML. (d) Type-IV tandem WOLED architecture, where the non-white EML unit is combined with single white EML or RYGB EMLs. W is white emission, HIL is hole injection layer, HTL is hole transport layer, ETL is electron transport layer, EIL is electron injection layer. Figure 1. Typical device architectures for tandem white organic light-emitting diodes (WOLEDs). (a) Type-I tandem WOLED architecture, in which the white emission is generated via the combination of nonwhite or monochromatic emission from each electroluminescent (EL) unit. (b) Type-II tandem WOLED architecture, where RYGB emitting layers (EMLs) mean that more than two EMLs generating white emission in each EL unit. (c) Type-III tandem WOLED architecture, where each EML of the EL unit is single white EML. (d) Type-IV tandem WOLED architecture, where the non-white EML unit is combined with single white EML or RYGB EMLs. W is white emission, HIL is hole injection layer, HTL is hole transport Figure 1. Typical device architectures for tandem white organic light-emitting diodes (WOLEDs). (a) Type-I tandem WOLED architecture, in which the white emission is generated via the combination of nonwhite or monochromatic emission from each electroluminescent (EL) unit. (b) Type-II tandem WOLED architecture, where RYGB emitting layers (EMLs) mean that more than two EMLs generating white emission in each EL unit. (c) Type-III tandem WOLED architecture, where each EML of the EL unit is single white EML. (d) Type-IV tandem WOLED architecture, where the non-white EML unit is combined with single white EML or RYGB EMLs. W is white emission, HIL is hole injection layer, HTL is hole transport layer, ETL is electron transport layer, EIL is electron injection layer. 2.1. The Device Architecture of Tandem OLEDs (a) Type-I tandem WOLED architecture, in which the white emission is generated via the combination of nonwhite or monochromatic emission from each electroluminescent (EL) unit. (b) Type-II tandem WOLED architecture, where RYGB emitting layers (EMLs) mean that more than two EMLs generating white emission in each EL unit. (c) Type-III tandem WOLED architecture, where each EML of the EL unit is single white EML. (d) Type-IV tandem WOLED architecture, where the non-white EML unit is combined with single white EML or RYGB EMLs. W is white emission, HIL is hole injection layer, HTL is hole transport layer ETL is electron transport layer EIL is electron injection layer Figure 1. Typical device architectures for tandem white organic light-emitting diodes (WOLEDs). (a) Type-I tandem WOLED architecture, in which the white emission is generated via the combination of nonwhite or monochromatic emission from each electroluminescent (EL) unit. (b) Type-II tandem WOLED architecture, where RYGB emitting layers (EMLs) mean that more than two EMLs generating white emission in each EL unit. (c) Type-III tandem WOLED architecture, where each EML of the EL unit is single white EML. (d) Type-IV tandem WOLED architecture, where the non-white EML unit is combined with single white EML or RYGB EMLs. W is white emission, HIL is hole injection layer, HTL is hole transport layer, ETL is electron transport layer, EIL is electron injection layer. Figure 1. Typical device architectures for tandem white organic light-emitting diodes Figure 1. Typical device architectures for tandem white organic light-emitting diodes (WOLEDs). Figure 1. Typical device architectures for tandem white organic light-emitting diodes (WOLEDs). (a) Type-I tandem WOLED architecture, in which the white emission is generated via the combination of nonwhite or monochromatic emission from each electroluminescent (EL) unit. (b) Type-II tandem WOLED architecture, where RYGB emitting layers (EMLs) mean that more than two EMLs generating white emission in each EL unit. (c) Type-III tandem WOLED architecture, where each EML of the EL unit is single white EML. (d) Type-IV tandem WOLED architecture, where the non-white EML unit is combined with single white EML or RYGB EMLs. W is white emission, HIL is hole injection layer, HTL is hole transport Figure 1. Typical device architectures for tandem white organic light-emitting diodes (WOLEDs). (a) Type-I tandem WOLED architecture, in which the white emission is generated via the combination of nonwhite or monochromatic emission from each electroluminescent (EL) unit. layer, ETL is electron transport laye 2.2. The Role of Charge Generation Layer 2.2. The Role of Charge Generation Layer To guarantee the high performance, the CGL plays a vital role in tandem WOLEDs, apart from the efficient each EL unit [98–100]. In 2003, Kido et al. reported the first tandem OLED by using n-doped organic/transparent conductive layer (i.e., Cs:2,9-dimethyl-4,7-diphenyl-1,10-phenanthroline (BCP)/indium tin oxide (ITO)) and n-doped organic/insulating materials (Cs:BCP/V2O5 or Cs:BCP/4F-TCNQ) as the CGLs [98]. Since then, numerous effective CGLs have been demonstrated. Briefly, a CGL acts as internal anode as well as cathode to generate charges and separate the opposite charges injecting to nearby EL unit [101]. Hence, tandem OLEDs can possibly convert one injected electron to multiple photons, obtaining higher luminance and CE at low current density. Such effect is beneficial to To guarantee the high performance, the CGL plays a vital role in tandem WOLEDs, apart from the efﬁcient each EL unit [98–100]. In 2003, Kido et al. reported the ﬁrst tandem OLED by using n-doped organic/transparent conductive layer (i.e., Cs:2,9-dimethyl-4,7-diphenyl-1,10-phenanthroline (BCP)/indium tin oxide (ITO)) and n-doped organic/insulating materials (Cs:BCP/V2O5 or Cs:BCP/4F-TCNQ) as the CGLs [98]. Since then, numerous effective CGLs have been demonstrated. Brieﬂy, a CGL acts as internal anode as well as cathode to generate charges and separate the opposite charges injecting to nearby EL unit [101]. Hence, tandem OLEDs can possibly convert one injected electron to multiple photons, obtaining higher luminance and CE at low current density. Such effect is beneﬁcial to prolong the lifetime due to the avoidance of current leakage and breakdown caused by electric ﬁeld [102–105]. 3 p , g g y For the formation of a CGL, a n-p semiconductor heterojunction is typically needed for the charge generation, which is located at the interface of n-type and p-type layers [106–110]. Besides, some demands are required for an efﬁcient CGL, including excellent charge generation, small barrier for charge separation and injection into the nearby EL unit, outstanding transparency for visible emissions (e.g., ≥75%), good conductivity and long working stability [111–113]. So far, a large number 4 of 28 Molecules 2019, 24, 151 of CGLs have been put forward. Apart from the above mentioned n-doped organic material/metal oxide CGL [98], n-doped organic material/p-doped organic material CGL (e.g., Li:tris(8-quinolinolato) aluminum(III) (Alq3)/FeCl3:4,4′-N,N’-bis[N-(1-naphthyl)-N-phenylamino]biphenyl (NPB) [66], 4,7-diphenyl-1,10-phenanthroline (BPhen):Rb2CO3/NPB:ReO3 [114], Bphen:Rb2CO3/ReO3/NPB:ReO3 [114]) and organic heterojunction CGL (e.g., copper hexadecaﬂuorophthalocyanine (F16CuPc)/copper phthalocyanine (CuPc) [115], fullerene (C60)/pentacene [116], zinc phthalocyanine (ZnPc):C60 [117]) are the most popular types. 3.1. Basic Aspects of Tandem WOLEDs Based on the employed emitters, tandem WOLEDs can be classiﬁed into four types, i.e., tandem ﬂuorescent WOLEDs, tandem phosphorescent WOLEDs, tandem TADF WOLEDs, and tandem hybrid WOLEDs. Thus, the selection of efﬁcient emitters is essential to the high performance [127–132]. Unlike single-unit WOLEDs, the adoption of effective CGLs is another essential for high-performance tandem WOLEDs according to the above mentioned concepts. Therefore, the core features of strategies for tandem WOLEDs are the simultaneous management of efﬁcient emitters and effective CGLs. To ensure the efﬁcient emission can be produced in each EL unit of tandem WOLEDs, the careful manipulation of charges and excitons distribution is necessary [133–138]. Speciﬁcally, the issues of charge injection from the electrodes, charge transport and charge balance in EMLs, exciton generation, exciton harvesting, exciton recombination, exciton diffusion and exciton quenching should be well manipulated [139–142]. On the other hand, the CGL should not only be effective to generate and separate charges but also be able to optimize the optical effect to reduce the light loss. In the following sections, the design strategy, device architecture, emission mechanism as well as the effect of CGL in tandem WOLEDs have been highlighted, including tandem ﬂuorescent, phosphorescent, TADF and hybrid WOLEDs. Additionally, the application of doping-free technique in tandem WOLEDs has also been described. layer, ETL is electron transport laye 2.2. The Role of Charge Generation Layer In the physical processes of CGLs, charge generation and charge separation are the two main features. Hence, upon charges generating at the n-p junction, electrons and holes should be rapidly injected into nearby EL unit. To guarantee such charge separation, the CGL materials with high charge mobility or conductivity are ideal [118–122]. Besides, the energy barrier between CGLs and ETL or HTL in each EL unit should be minimized, which cannot only increase the charge injection into the EL unit but also reduce the charge accumulation [123–126]. To enhance the PE, the CGL is required to possess outstanding charge generation as well as charge separation. Particularly, negligible voltage drop across the CGL in the process of charge generation is required, otherwise the PE cannot be largely enhanced. This is because the voltage of tandem OLEDs usually increases N-fold compared with single-unit devices, leading to the fact that the PE of tandem OLEDs is similar to that of single-unit OLEDs. 3.2. Tandem Fluorescent WOLEDs R, G and B represent red, green and blue emission, respectively. Left: the molecular structures of 4-(dicyanomethylene)-2-t-butyle-6-(1,1,7,7-tetramethyljulolidyl-9-enyl) 4H-pyran (DCJTB) and 9,10-bis-(β-naphthyl)-anthrene (AND) [72]. Figure 2. Right: the device structure of tandem WOLEDs. R, G and B represent red, green and blue emission, respectively. Left: the molecular structures of 4-(dicyanomethylene)-2-t-butyle-6-(1,1,7,7-tetramethyljulolidyl-9-enyl) 4H-pyran (DCJTB) and 9,10-bis-(β-naphthyl)-anthrene (AND) [72]. 5 After Ma’s pioneering work [72], many efforts have been made on tandem fluorescent WOLEDs [143]. For example, Ho et al. used n-doped organic material/p-doped organic material CGL (i.e., Bphen:2% cesium carbonate (Cs2CO3)/NPB:50% v/v tungsten oxide (WO3)) to interconnect blue fluorescent EL unit and red fluorescent EL unit (type-I tandem WOLED architecture) [144]. By virtue of the p-i-n technology, their device showed a CE of 23.9 cd A−1 and PE of 7.8 lm W−1, which are one of the best for tandem fluorescent WOLEDs. However, despite the luminance and efficiency of tandem fluorescent WOLEDs can be scale linearly with the number of EL units, triplet excitons still decay nonradiatively, leading to the fact that After Ma’s pioneering work [72], many efforts have been made on tandem ﬂuorescent WOLEDs [143]. For example, Ho et al. used n-doped organic material/p-doped organic material CGL (i.e., Bphen:2% cesium carbonate (Cs2CO3)/NPB:50% v/v tungsten oxide (WO3)) to interconnect blue ﬂuorescent EL unit and red ﬂuorescent EL unit (type-I tandem WOLED architecture) [144]. By virtue of the p-i-n technology, their device showed a CE of 23.9 cd A−1 and PE of 7.8 lm W−1, which are one of the best for tandem ﬂuorescent WOLEDs. However, despite the luminance and efﬁciency of tandem ﬂuorescent WOLEDs can be scale linearly with the number of EL units, triplet excitons still decay nonradiatively, leading to the fact that the performance of this kind of devices is not satisfactory enough [145–149]. Therefore, more works are focused on phosphorescent or TADF emitters-based tandem WOLEDs. 3.2. Tandem Fluorescent WOLEDs The ﬁrst tandem WOLED was based on ﬂuorescent emitters, in which 4-(dicyanomethylene)-2-t-butyle-6-(1,1,7,7-tetramethyljulolidyl-9-enyl) 4H-pyran (DCJTB), Alq3 and 9,10-bis-(β-naphthyl)-anthrene (ADN) were used as the red (600 nm), green (505 nm) and blue (435 nm) ﬂuorescent emitter, respectively [72]. The device structure was ITO/NPB (50 nm)/ADN (30 nm)/BCP (10 nm)/Alq3 (40 nm)/BCP:Li (10 nm)/V2O5 (30 nm)/NPB (50 nm)/Alq3:DCJTB (40 nm)/Alq3 (40 nm)/LiF (1 nm)/Al (200 nm), which can be classiﬁed into type-I tandem WOLED architecture, as shown in Figure 2. As a result, the tandem WOLED showed a stable white light with CIE coordinates from (0.35, 0.32) at 18 V to (0.36, 0.36) at 50 V. The maximum luminance of 10,200 cd m−2 and CE was 10.7 cd A−1, which were equal basically to the sum of the two EL units. For the origin of strikingly enhanced performance, the CGL of BCP:Li (10 nm)/V2O5 (30 nm) played a key role. In brief, electrons and holes were generated within the CGL and then reached nearby Alq3 and NPB, respectively. On one hand, these electrons could arrive at the blue EML ADN to 5 of 28 ted ese Molecules 2019, 24, 151 BCP:Li (10 nm)/ within the CGL recombine with holes injected from anode, generating blue emission. Since BCP is a hole-blocking material, the holes distribution could be manipulated by adjusting the thickness of BCP layer, assuring blue and green emissions. On the other hand, holes generated from the CGL could reach the red EML Alq3:DCJTB and recombine with electrons injected from cathode, producing the red emission. Therefore, white emission has been achieved. generating blue emission. Since BCP is a hole-blocking material, the holes distribution could be manipulated by adjusting the thickness of BCP layer, assuring blue and green emissions. On the other hand, holes generated from the CGL could reach the red EML Alq3:DCJTB and recombine with electrons injected from cathode, producing the red emission. Therefore, white emission has been achieved. Figure 2. Right: the device structure of tandem WOLEDs. R, G and B represent red, green and blue emission, respectively. Left: the molecular structures of 4-(dicyanomethylene)-2-t-butyle-6-(1,1,7,7-tetramethyljulolidyl-9-enyl) 4H-pyran (DCJTB) and 9,10-bis-(β-naphthyl)-anthrene (AND) [72]. Figure 2. Right: the device structure of tandem WOLEDs. R, G and B represent red, green and blue emission, respectively. Left: the molecular structures of 4-(dicyanomethylene)-2-t-butyle-6-(1,1,7,7-tetramethyljulolidyl-9-enyl) 4H-pyran (DCJTB) and 9,10-bis-(β-naphthyl)-anthrene (AND) [72]. Figure 2. Right: the device structure of tandem WOLEDs. 3.3. Tandem Phosphorescent WOLEDs In the case of phosphorescent emitters, triplet excitons can be harvested via the triplet-triplet energy transfer, while singlet excitons are harnessed via the singlet-triplet intersystem crossing process due to the heavy-atom effect, leading to a maximum IQE of 100% [150]. As an excellent result, single-unit phosphorescent WOLED with ﬂuorescent tube efﬁciency has been demonstrated [151]. With the improvement of material design and device engineering, it is not very surprising to see that the EQE of single-unit phosphorescent WOLED can be >20% currently [152–154]. By extending the application of phosphorescent emitters into tandem WOLEDs, high-performance devices can be also developed with the management of effective CGLs. In 2006, Kanno et al. reported the ﬁrst tandem phosphorescent WOLED [155]. By using a n-doped organic material/metal oxide CGL (i.e., Bphen:Li/MoO3) to interconnect the double-EML 6 of 28 Molecules 2019, 24, 151 white EL unit (i.e., 10 wt. fac-tris(1-(9,9-dimethyl-2-ﬂuorenyl)pyrazolylN,C2′)iridium(III) (Ir(ﬂz)3):N,N′-dicarbazolyl-3,5-benzene (mCP, 20 nm)/10 iridium(III) bis(2-phenylquinolyl-N,C2′)acetylacetonate (PQIr):3 fac-tris(2-phenylpyridinato-N,C2′)iridium(III) (Ir(ppy)3):4,4′-N,N′-dicarbazole-biphenyl (CBP, 5 nm)), tandem WOLEDs with 2 and 3 EL units were developed, which can be classiﬁed into type-II tandem WOLED architecture, as shown in Figure 3. As a result, the tandem WOLED with 3 EL units reached a peak forward-viewing EQE of 34.9% and total EQE of 51.0% at 500 cd m−2. Brieﬂy, there were four factors for the high performance. (i) The Li doping of Bphen in the CGL allowed for efﬁcient electron injection, otherwise inefﬁcient charge generation would occur. (ii) Charges and excitons were conﬁned within the EMLs of each subpixel in the stack, by using materials with wide energy gaps i.e., fac-tris(1-phenypyrazolyl-N,C2′)iridium(III) (Ir(ppz)3) and Bphen as barriers to exciton and charge diffusion across the EML/HTL and EML/ETL interfaces, respectively. (iii) All exciton formation occurred within each EML by direct excitation of the triplet, either by injection from the lowest unoccupied molecular orbital (LUMO) CBP or by electron transfer from the mCP LUMO. (iv) Optical interference and weak microcavity effects were controlled by varying the thickness of HTL 4,4′-bis[N-(1-naphthyl)-N-phenyl-amino]biphenyl (NPD) with high hole mobility, since its thickness variation (<100 nm) would not signiﬁcantly affect the voltage nor charge balance and hence the position of the recombination zone. Therefore, the charge conﬁning structure and the effective CGL ensured a uniform white color balance for each subpixel, along with high efﬁciency for charge injection. Molecules 2018, 23, x FOR PEER REVIEW 7 of 3 Figure 3. Top:Proposed energy-level diagram of tandem WOLEDs. 3.3. Tandem Phosphorescent WOLEDs Numbers indicate the HOMO and LUMO energies relative to vacuum. The arrows indicate the carrier injection from electrodes and the MoO3 CGL. Bottom:The molecular structures of Ir(flz)3 and mCP. Reproduced from reference [155]. In general, the architecture of type-III tandem WOLEDs is simpler than that of type-II tandem WOLEDs due to the reduced number of EMLs. However, it is usually more challenging to achieve high efficiency for single-EML WOLEDs compared with multiple-EML WOLEDs [156–160]. To loosen this bottleneck, Wang et al. have first realized highly efficient single-unit WOLED with the EML of mCP:iridium(III)[bis(4,6-difluorophenyl)-pyridinato-N, C20]picolinate(FIrpic):bis(2-(9,9-diethyl-9H-fluoren-2-yl)-1-phenyl-1H-benzoimidazolN,C3)iri Figure 3. Top: Proposed energy-level diagram of tandem WOLEDs. Numbers indicate the HOMO and LUMO energies relative to vacuum. The arrows indicate the carrier injection from electrodes and the MoO3 CGL. Bottom: The molecular structures of Ir(ﬂz)3 and mCP. Reproduced from reference [155]. In general, the architecture of type-III tandem WOLEDs is simpler than that of type-II tandem WOLEDs due to the reduced number of EMLs. However, it is usually more challenging to achieve gh efﬁciency for single-EML WOLEDs compared with multiple-EML WOLEDs [156–160]. To loosen is bottleneck, Wang et al. have ﬁrst realized highly efﬁcient single-unit WOLED with the EML of CP:iridium(III)[bis(4,6-diﬂuorophenyl)-pyridinato-N,C20]picolinate(FIrpic):bis(2-(9,9-diethyl-9H- uoren-2-yl)-1-phenyl-1H-benzoimidazolN,C3)iridium(acetylacetonate) ((fbi)2Ir(acac)), by harvesting Figure 3. Top:Proposed energy-level diagram of tandem WOLEDs. Numbers indicate the HOMO and LUMO energies relative to vacuum. The arrows indicate the carrier injection from electrodes and the MoO3 CGL. Bottom:The molecular structures of Ir(flz)3 and mCP. Figure 3. Top: Proposed energy-level diagram of tandem WOLEDs. Numbers indicate the HOMO and LUMO energies relative to vacuum. The arrows indicate the carrier injection from electrodes and the MoO3 CGL. Bottom: The molecular structures of Ir(ﬂz)3 and mCP. Reproduced from reference [155]. Reproduced from reference [155]. In general, the architecture of type-III tandem WOLEDs is simpler than that of type-II tandem WOLEDs due to the reduced number of EMLs. However, it is usually more challenging to achieve high efficiency for single-EML WOLEDs compared with multiple-EML WOLEDs [156–160]. To loosen this bottleneck, Wang et al. have first realized highly efficient single-unit WOLED with the EML of mCP:iridium(III)[bis(4,6-difluorophenyl)-pyridinato-N, In general, the architecture of type-III tandem WOLEDs is simpler than that of type-II tandem WOLEDs due to the reduced number of EMLs. However, it is usually more challenging to achieve high efﬁciency for single-EML WOLEDs compared with multiple-EML WOLEDs [156–160]. To loosen this bottleneck, Wang et al. 3.3. Tandem Phosphorescent WOLEDs have ﬁrst realized highly efﬁcient single-unit WOLED with the EML of mCP:iridium(III)[bis(4,6-diﬂuorophenyl)-pyridinato-N,C20]picolinate(FIrpic):bis(2-(9,9-diethyl-9H- ﬂuoren-2-yl)-1-phenyl-1H-benzoimidazolN,C3)iridium(acetylacetonate) ((fbi)2Ir(acac)), by harvesting Reproduced from reference [155]. In general, the architecture of type-III tandem WOLEDs is simpler than that of type-II tandem WOLEDs due to the reduced number of EMLs. However, it is usually more challenging to achieve high efficiency for single-EML WOLEDs compared with multiple-EML WOLEDs [156–160]. To loosen this bottleneck, Wang et al. have first realized highly efficient single-unit WOLED with the EML of mCP:iridium(III)[bis(4,6-difluorophenyl)-pyridinato-N, In general, the architecture of type-III tandem WOLEDs is simpler than that of type-II tandem WOLEDs due to the reduced number of EMLs. However, it is usually more challenging to achieve high efﬁciency for single-EML WOLEDs compared with multiple-EML WOLEDs [156–160]. To loosen this bottleneck, Wang et al. have ﬁrst realized highly efﬁcient single-unit WOLED with the EML of mCP:iridium(III)[bis(4,6-diﬂuorophenyl)-pyridinato-N,C20]picolinate(FIrpic):bis(2-(9,9-diethyl-9H- ﬂuoren-2-yl)-1-phenyl-1H-benzoimidazolN,C3)iridium(acetylacetonate) ((fbi)2Ir(acac)), by harvesting 7 of 28 7 of 28 Molecules 2019, 24, 151 all excitons via two parallel channels:host-guest energy transfer for FIrpic and direct exciton formation following charge trapping for (fbi)2Ir(acac) [161]. Then, they have used such single-EML architecture as the EL unit of a tandem WOLED via the interconnection of a n-doped organic material/metal oxide CGL (i.e., Li-doped BCP/MoO3), which can be classiﬁed into type-III tandem WOLED architecture, as shown in Figure 4 [162]. As a result, the tandem WOLED exhibited a maximum forward viewing CE of 110.9 cd A−1, EQE of 43.3% and PE of 45.2 lm W−1, which were the best for tandem WOLEDs with two EL units at that time. Thus, the high performance was attributed to the combination of effective single units and CGL. Molecules 2018, 23, x FOR PEER REVIEW 8 of 3 Figure 4. Transmittance spectrum of Li:BCP (10 nm)/MoO3 (7 nm) thin film. Inset:schematic device structure of the tandem WOLED. Reproduced from reference [162]. Figure 4. Transmittance spectrum of Li:BCP (10 nm)/MoO3 (7 nm) thin ﬁlm. Inset: schematic device structure of the tandem WOLED. Reproduced from reference [162]. Figure 4. Transmittance spectrum of Li:BCP (10 nm)/MoO3 (7 nm) thin film. Inset:schematic device structure of the tandem WOLED. Reproduced from reference [162]. Figure 4. Transmittance spectrum of Li:BCP (10 nm)/MoO3 (7 nm) thin ﬁlm. Inset: schematic device structure of the tandem WOLED. Reproduced from reference [162]. After the selection of CGLs, the optimization of efficient phosphorescent emitters is key to the performance. Towards this end, Lee et al. 3.3. Tandem Phosphorescent WOLEDs (a) Structure of tandem WOLEDs. (b) Mode analysis of blue light as a function of HATCN thickness with a total HTL thickness of 80 nm. (a) Structure of tandem WOLEDs. (b) Mode analysis of blue light as a function of thickness with a total HTL thickness of 80 nm. indicates the thickness of when an out-coupled mode is maximized. (c) Mode analysis of white light as a of the total HTL thickness at an HATCN thickness of 40 nm. (d) Mode analysis of ht as a function of HATCN thickness with a total HTL thickness of 80 nm. indicates the thickness of HATCN when an out-coupled mode is maximized. (c) Mode analysis of white light as a function of the total HTL thickness at an HATCN thickness of 40 nm. (d) Mode analysis of white light as a function of HATCN thickness with a total HTL thickness of 80 nm. Reproduced from reference [163]. Reproduced from reference [163]. It is well-known that blue phosphorescent emitters usually exhibit poor stability [169–173]. Hence, the lifetime of tandem phosphorescent WOLEDs is limited. Recently, a strategy to address this issue has been proposed by Coburn et al. [174]. The device structure was glass substrate/150 nm ITO/10 nm HATCN/30 nm NPD/1-3 red-green emitter-CGL pairs (D3-D5, respectively)/blue element/CGL/red-green element/50 nm BPyTP2/1.5 nm 8-hydroxyquinolinato lithium (LiQ)/100 nm Al, where the red-green element was 10 nm 4,40-bis(3-methylcarbazol-9-yl)-2,20-biphenyl (mCBP):8 vol % Ir(5′-Phppy)3:10 vol % PQIr/25 nm mCBP:9 vol % Ir(5-Ph-ppy)3/3 nm BAlq/5 nm BAlq:10 vol % PQIr/5 nm BAlq, the blue element was 20 nm mCBP:18 → 14 vol % Ir(dmp)3/10 nm mCBP:11 → 19 vol % Ir(dmp)3:3 vol % mer-tris(N-phenyl, Nmethyl-pyridoimidazol-2-yl)iridium(III) (mer-Ir(pmp)3)/20 nm oduced from reference [163]. well-known that blue phosphorescent emitters usually exhibit poor stability . Hence, the lifetime of tandem phosphorescent WOLEDs is limited. Recently, a to address this issue has been proposed by Coburn et al. [174]. The device structure s substrate/150 nm ITO/10 nm HATCN/30 nm NPD/1-3 red-green emitter-CGL pairs respectively)/blue element/CGL/red-green element/50 nm BPyTP2/1.5 nm yquinolinato lithium (LiQ)/100 nm Al, where the red-green element was 10 nm -methylcarbazol-9-yl)-2,20-biphenyl (mCBP):8 vol % Ir(5′-Phppy)3:10 vol % PQIr/25 P:9 vol % Ir(5-Ph-ppy)3/3 nm BAlq/5 nm BAlq:10 vol % PQIr/5 nm BAlq, the blue was 20 nm mCBP:18 → 14 vol % Ir(dmp)3/10 nm mCBP:11 → 19 vol % Ir(dmp)3:3 vol tris(N-phenyl, Nmethyl-pyridoimidazol-2-yl)iridium(III) (mer-Ir(pmp)3)/20 nm It is well-known that blue phosphorescent emitters usually exhibit poor stability [169–173]. 3.3. Tandem Phosphorescent WOLEDs used an exciplex-forming co-host and red- and green-phosphorescent dyes with horizontally oriented transition dipoles to optimize an orange OLED with a maximum EQE of 32% [163]. Next, by connecting such efficient orange EL unit with a blue phosphorescent EL unit via a CGL of Rb2CO3-doped BPhen/1,4,5,8,9,11-hexaazatriphenylene hexacarbonitrile (HATCN)/4,7-diphenyl-1,10-phenanthroline (TAPC), tandem WOLEDs showed a maximum EQE of 54.3% without outcoupling technology or EQE of 90.6% at 1000 cd m-2 by attaching an index-matched glass half sphere onto the glass substrate. The device structure was ITO (70 nm)/4 % ReO3-doped mCP (x nm)/mCP (15 nm)/mCP:bis-4,6-(3,5-di-3-pyridylphenyl)-2-methylpyrimidine (B3PYMPM):FIrpic (15 nm, 0.7:0.3:0.057 molar ratio)/B3PYMPM (15 nm)/4 wt% Rb2CO3-doped B3PYMPM (25 nm)/23 wt% Rb2CO3-doped BPhen (10 nm)/HATCN (y nm)/TAPC (20 After the selection of CGLs, the optimization of efﬁcient phosphorescent emitters is key to the performance. Towards this end, Lee et al. used an exciplex-forming co-host and red- and green-phosphorescent dyes with horizontally oriented transition dipoles to optimize an orange OLED with a maximum EQE of 32% [163]. Next, by connecting such efﬁcient orange EL unit with a blue phosphorescent EL unit via a CGL of Rb2CO3-doped BPhen/1,4,5,8,9,11-hexaazatriphenylene hexacarbonitrile (HATCN)/4,7-diphenyl-1,10-phenanthroline (TAPC), tandem WOLEDs showed a maximum EQE of 54.3% without outcoupling technology or EQE of 90.6% at 1000 cd m−2 by attaching an index-matched glass half sphere onto the glass substrate. The device structure was ITO (70 nm)/4% ReO3-doped mCP (x nm)/mCP (15 nm)/mCP:bis-4,6-(3,5-di-3-pyridylphenyl)-2-methylpyrimidine (B3PYMPM):FIrpic (15 nm, 0.7:0.3:0.057 molar ratio)/B3PYMPM (15 nm)/4 wt% Rb2CO3-doped B3PYMPM (25 nm)/23 wt% Rb2CO3-doped BPhen (10 nm)/HATCN (y nm)/TAPC (20 nm)/4,4′,4”-tri(N-carbazolyl)triphen-ylamine (TCTA, 10 nm)/TCTA:B3PYMPM:iridium(III) bis(2-phenylquinoline) tetramethylheptadionate (Ir(ppy)2(tmd)):Iridium(III) Rb2CO3 doped BPhen (10 nm)/HATCN (y nm)/TAPC (20 nm)/4,4′,4″-tri(N-carbazolyl)triphen-ylamine (TCTA, 10 nm)/TCTA:B3PYMPM:iridium(III) bis(2-phenylquinoline) tetramethylheptadionate (Ir(ppy)2(tmd)):Iridium(III) bis(4-methyl-2-(3,5-dimethylphenyl)quinolinato-N,C2′) tetramethylheptadionate (Ir(mphmq)2(tmd)) (15 nm, 0.5:0.5:0.1:0.002 molar ratio)/B3PYMPM (60 nm)/LiF (0.7 nm)/Al (100 nm), which can be classified into type-I tandem WOLED architecture, as shown in Figure 5a. A key feature for the high performance was the introduction of the exciplex host p y q y p ppy bis(4-methyl-2-(3,5-dimethylphenyl)quinolinato-N,C2′) tetramethylheptadionate (Ir(mphmq)2(tmd)) (15 nm, 0.5:0.5:0.1:0.002 molar ratio)/B3PYMPM (60 nm)/LiF (0.7 nm)/Al (100 nm), which can be classiﬁed into type-I tandem WOLED architecture, as shown in Figure 5a. A key feature for the high performance was the introduction of the exciplex host TCTA:B3PYMPM for Ir(ppy)2(tmd) and Ir(mphmq)2(tmd) having highly oriented triplet transition dipole moments along the horizontal direction and high photoluminescence quantum yields (PLQY) [164–168]. 3.3. Tandem Phosphorescent WOLEDs Besides, the tandem device 8 of 28 Molecules 2019, 24, 151 architecture was optimized by an optical simulation to maximize the outcoupling of the emitted light. For example, the location for the blue EL unit, orange EL unit and the total thickness of ITO and organic layers have been optimized via the simulation based on the classical dipole model, as shown in Figure 5b–d. Molecules 2018, 23, x FOR PEER REVIEW 9 of 018, 23, x FOR PEER REVIEW 9 of 30 Figure 5. (a) Structure of tandem WOLEDs. (b) Mode analysis of blue light as a function of HATCN thickness with a total HTL thickness of 80 nm. indicates the thickness of HATCN when an out-coupled mode is maximized. (c) Mode analysis of white light as a function of the total HTL thickness at an HATCN thickness of 40 nm. (d) Mode analysis of white light as a function of HATCN thickness with a total HTL thickness of 80 nm. Figure 5. (a) Structure of tandem WOLEDs. (b) Mode analysis of blue light as a function of HATCN thickness with a total HTL thickness of 80 nm. (a) Structure of tandem WOLEDs. (b) Mode analysis of blue light as a function of thickness with a total HTL thickness of 80 nm. indicates the thickness of when an out-coupled mode is maximized. (c) Mode analysis of white light as a of the total HTL thickness at an HATCN thickness of 40 nm. (d) Mode analysis of ht as a function of HATCN thickness with a total HTL thickness of 80 nm. indicates the thickness of HATCN when an out-coupled mode is maximized. (c) Mode analysis of white light as a function of the total HTL thickness at an HATCN thickness of 40 nm. (d) Mode analysis of white light as a function of HATCN thickness with a total HTL thickness of 80 nm. Reproduced from reference [163]. Figure 5. (a) Structure of tandem WOLEDs. (b) Mode analysis of blue light as a function of HATCN thickness with a total HTL thickness of 80 nm. indicates the thickness of HATCN when an out-coupled mode is maximized. (c) Mode analysis of white light as a function of the total HTL thickness at an HATCN thickness of 40 nm. (d) Mode analysis of white light as a function of HATCN thickness with a total HTL thickness of 80 nm. Figure 5. 3.3. Tandem Phosphorescent WOLEDs More specifically, (i) stable bis(8-hydroxy-2-methylquinoline)-(4-phenylphenoxy)aluminum (BAlq) was employed as a hole blocking layer for green element and a thin spacer between the green EML and red doped region, which could also reduce the loss of excitons transferred to its low triplet energy; (ii) the exciton confinement at the EML interface with NPD was obtained by doping red phosphor into a thin green EML adjacent to HTL; (iii) placing red EMLs on both sides of green EML reduced the color shift; iv) dopant grading balanced hole and electron transport in blue EML, broadening the exciton recombination zone and reducing bimolecular annihilation rates that lead to molecular dissociation, which could increase the blue element stability; (v) mer-Ir(pmp)3 was used to improve the reliability of the blue element by reducing the probability that hot excited states degrade host or emitter molecules; (vi) the CGL was stable and possessed high charge mobility; (vii) an outcoupling improvement of 2.2 ± 0.2 times over substrate emission by outcoupling substrate modes using index matching fluid between the device substrate and photodetector during the EQE measurement. Figure 6. Device structure: (left) Tandem WOLEDs D3, D4, and D5 have one, two, and three CGL/red-green element pairs below the blue element, respectively. (Right) Layers of the red-green element, blue element and CGL. Reproduced from reference [174]. Figure 6. Device structure: (left) Tandem WOLEDs D3, D4, and D5 have one, two, and three CGL/red-green element pairs below the blue element, respectively. (Right) Layers of the red-green element, blue element and CGL. Reproduced from reference [174]. Figure 6. Device structure: (left) Tandem WOLEDs D3, D4, and D5 have one, two, and three CGL/red-green element pairs below the blue element, respectively. (Right) Layers of the red-green element, blue element and CGL. Reproduced from reference [174]. Figure 6. Device structure: (left) Tandem WOLEDs D3, D4, and D5 have one, two, and three CGL/red-green element pairs below the blue element, respectively. (Right) Layers of the red-green element, blue element and CGL. Reproduced from reference [174]. 3.3. Tandem Phosphorescent WOLEDs Hence, the lifetime of tandem phosphorescent WOLEDs is limited. Recently, a strategy to address this issue has been proposed by Coburn et al. [174]. The device structure was glass substrate/150 nm ITO/10 nm HATCN/30 nm NPD/1-3 red-green emitter-CGL pairs (D3-D5, respectively)/blue element/CGL/red-green element/50 nm BPyTP2/1.5 nm 8-hydroxyquinolinato lithium (LiQ)/100 nm Al, where the red-green element was 10 nm 4,40-bis(3-methylcarbazol-9-yl)-2,20-biphenyl (mCBP):8 vol % Ir(5′-Phppy)3:10 vol % PQIr/25 nm mCBP:9 vol % Ir(5-Ph-ppy)3/3 nm BAlq/5 nm BAlq:10 vol % PQIr/5 nm BAlq, the blue element was 20 nm mCBP:18 →14 vol % Ir(dmp)3/10 nm mCBP:11 →19 vol % Ir(dmp)3:3 vol % mer-tris(N-phenyl, Nmethyl-pyridoimidazol-2-yl)iridium(III) (mer-Ir(pmp)3)/20 nm mCBP:12 →8 vol % Ir(dmp)3/5 nm mCBP:8 vol % Ir(dmp)3/5 nm mCBP/10 nm BPyTP2, the CGL was 8 nm 2,7-bis(2,20-bipyridine-5-yl)triphenylene (BPyTP2)/12 nm BPyTP2:3 vol % 9 of 28 Molecules 2019, 24, 151 Li/12 nm HATCN/5 nm NPD, D3, D4, and D5 indicated the total number of stacked elements separated by CGLs, as shown in Figure 6. As a result, the optimized tandem WOLED D5 showed a CCT of 2780 K, CRI of 89, a peak PE of 50 lm W−1 and a T70 lifetime (deﬁned as the time corresponding to 30% decrease in luminance from an initial value of 1000 cd m−2) of 80,000 ± 2000 h, with minimal spectral shifts during aging. The key features for the high performance were red emissive blocking layers in the red-green element, graded doping and hot excited state management in the blue element, stable and low voltage CGLs, and effective outcoupling technique. 3.3. Tandem Phosphorescent WOLEDs More speciﬁcally, (i) stable bis(8-hydroxy-2-methylquinoline)-(4-phenylphenoxy)aluminum (BAlq) was employed as a hole blocking layer for green element and a thin spacer between the green EML and red doped region, which could also reduce the loss of excitons transferred to its low triplet energy; (ii) the exciton conﬁnement at the EML interface with NPD was obtained by doping red phosphor into a thin green EML adjacent to HTL; (iii) placing red EMLs on both sides of green EML reduced the color shift; iv) dopant grading balanced hole and electron transport in blue EML, broadening the exciton recombination zone and reducing bimolecular annihilation rates that lead to molecular dissociation, which could increase the blue element stability; (v) mer-Ir(pmp)3 was used to improve the reliability of the blue element by reducing the probability that hot excited states degrade host or emitter molecules; (vi) the CGL was stable and possessed high charge mobility; (vii) an outcoupling improvement of 2.2 ± 0.2 times over substrate emission by outcoupling substrate modes using index matching ﬂuid between the device substrate and photodetector during the EQE measurement. time corresponding to 30% decrease in luminance from an initial value of 1000 cd m−2) of 80,000 ± 2000 h, with minimal spectral shifts during aging. The key features for the high performance were red emissive blocking layers in the red-green element, graded doping and hot excited state management in the blue element, stable and low voltage CGLs, and effective outcoupling technique. 3 4 Tandem TADF WOLE 3.4. Tandem TADF WOLEDs For TADF emitters, triplet excitons could be harnessed as delayed fluorescence through their up-conversion from a lowest triplet state to a lowest singlet state by inducing efficient reverse intersystem crossing [175–180]. Typically, the energy gap between T1 and S1 of ≤ 0.2 eV is favorable to the thermal up-conversion [181–183]. Similar to phosphorescent emitters, 100% IQE can be attained for TADF emitters [184–186]. Thus, the excellent characteristics of TADF emitters render that they are promising for WOLEDs. Since the first single-unit TADF WOLED reported in 2014 [187], the EQE of TADF-based WOLEDs has been demonstrated to be as high as 20%, which is comparable to state-of-the-art phosphorescent WOLEDs [188–190]. For TADF emitters, triplet excitons could be harnessed as delayed ﬂuorescence through their up-conversion from a lowest triplet state to a lowest singlet state by inducing efﬁcient reverse intersystem crossing [175–180]. Typically, the energy gap between T1 and S1 of ≤0.2 eV is favorable to the thermal up-conversion [181–183]. Similar to phosphorescent emitters, 100% IQE can be attained for TADF emitters [184–186]. Thus, the excellent characteristics of TADF emitters render that they are promising for WOLEDs. Since the ﬁrst single-unit TADF WOLED reported in 2014 [187], the EQE of TADF-based WOLEDs has been demonstrated to be as high as 20%, which is comparable to state-of-the-art phosphorescent WOLEDs [188–190]. So far, two kinds of TADF emitters have been Molecules 2019, 24, 151 Molecules 2018, 23, x 10 of 28 11 of 10 of 28 11 of reported, TADF materials and TADF exciplexes [191–195]. In general, these two kinds of emitters can be used to develop high-performance WOLEDs by carefully manipulating the charges and excitons distribution [196]. [191–195]. In general, these two kinds of emitters can be used to develop high-performance WOLEDs by carefully manipulating the charges and excitons distribution [196]. By using effective CGLs to interconnect the TADF emitters (TADF materials or TADF exciplexes), tandem TADF WOLEDs can be achieved. Toward this target, Hung and By using effective CGLs to interconnect the TADF emitters (TADF materials or TADF exciplexes), tandem TADF WOLEDs can be achieved. Toward this target, Hung and coworkers demonstrated a tandem WOLED, in which blue TADF exciplex EL unit and yellow exciplex EL unit were interconnected by a CGL of 9,9-di[4-(di-p-tolyl)aminophenyl]ﬂuorine (DTAF)/MoO3/Al/Liq (holes and electrons were generated from the DTAF/MoO3 interface) [197]. 3 4 Tandem TADF WOLE 3.4. Tandem TADF WOLEDs The structure was ITO/polyethylene dioxythiophene:polystyrene sulfonate (PEDOT:PSS, 30 nm)/TAPC (20 nm)/mCP (15 nm)/mCP:(1,3,5-triazine-2,4,6-triyl)tris(benzene-3,1-diyl))tris(diphenylphosphine oxide) (PO-T2T) (1:1, 20 nm)/PO-T2T (45 nm)/Liq (1 nm)/Al (1 nm)/MoO3 (5 nm)/DTAF (20 nm)/DTAF:PO-T2T (1:1, 20 nm)/PO-T2T (50 nm)/Liq (0.5 nm)/Al (100 nm), as shown in Figure 7. As a result, am EQE of 11.6% was realized. For such high-efﬁciency exciplex based tandem WOLED, the efﬁcient blue TADF exciplex is important. To accomplish this goal, Hung et al. synthesized PO-T2T having a low HOMO, low LUMO, high T1 of 2.99 eV and electron mobility of >10−3 cm2 V−1 s−1 as the acceptor. Combined with the mCP donor, the blue TADF exciplex emission could show an EQE of 8.0%, which ensured the high performance of tandem WOLED [188]. In Hung′s device [197], the blue TADF exciplex has been used. To extend this strategy, Zhao et al. used a CGL of 2,4,6-tris(3-(1H-pyrazol-1-yl)phenyl)-1,3,5-triazine (3P-T2T):(Cs2CO3)/Al/MoO3 to interconnect both blue TADF exciplex EL unit (TCTA:Bphen) and orange TADF exciplex unit (TAPC:2,4,6-tris(3-(1H-pyrazol-1-yl)phenyl)-1,3,5-triazine), achieving a tandem TADF WOLED with an EQE of 9.17% [198]. exciplexes), tandem TADF WOLEDs can be achieved. Toward this target, Hung and coworkers demonstrated a tandem WOLED, in which blue TADF exciplex EL unit and yellow exciplex EL unit were interconnected by a CGL of 9,9-di[4-(di-p-tolyl)aminophenyl]fluorine (DTAF)/MoO3/Al/Liq (holes and electrons were generated from the DTAF/MoO3 interface) [197]. The structure was ITO/polyethylene dioxythiophene:polystyrene sulfonate (PEDOT:PSS, 30 nm)/TAPC (20 nm)/mCP (15 nm)/mCP:(1,3,5-triazine-2,4,6-triyl)tris(benzene-3,1-diyl))tris(diphenylphosphine oxide) (PO-T2T) (1:1, 20 nm)/PO-T2T (45 nm)/Liq (1 nm)/Al (1 nm)/MoO3 (5 nm)/DTAF (20 nm)/DTAF:PO-T2T (1:1, 20 nm)/PO-T2T (50 nm)/Liq (0.5 nm)/Al (100 nm), as shown in Figure 7. As a result, am EQE of 11.6% was realized. For such high-efficiency exciplex based tandem WOLED, the efficient blue TADF exciplex is important. To accomplish this goal, Hung et al. synthesized PO-T2T having a low HOMO, low LUMO, high T1 of 2.99 eV and electron mobility of >10−3 cm2 V−1 s−1 as the acceptor. Combined with the mCP donor, the blue TADF exciplex emission could show an EQE of 8.0%, which ensured the high performance of tandem WOLED [188]. In Hung′s device [197], the blue TADF exciplex has been used. To extend this strategy, Zhao et al. used a CGL of 2,4,6-tris(3-(1H-pyrazol-1-yl)phenyl)-1,3,5-triazine (3P-T2T):(Cs2CO3)/Al/MoO3 to interconnect both blue TADF exciplex EL unit (TCTA:Bphen) and orange TADF exciplex unit (TAPC:2,4,6-tris(3-(1H-pyrazol-1-yl)phenyl)-1,3,5-triazine), achieving a tandem TADF WOLED with an EQE of 9.17% [198]. Figure 7. (a) Energy levels for the tandem WOLED. 3 4 Tandem TADF WOLE 3.4. Tandem TADF WOLEDs (b) EL spectra at various luminance, and two decomposed bands were blue and yellow exciplex emissions. Reproduced from reference [197]. Figure 7. (a) Energy levels for the tandem WOLED. (b) EL spectra at various luminance, and two decomposed bands were blue and yellow exciplex emissions. Reproduced from reference [197]. Figure 7. (a) Energy levels for the tandem WOLED. (b) EL spectra at various luminance, and two decomposed bands were blue and yellow exciplex emissions. Reproduced from reference [197] Figure 7. (a) Energy levels for the tandem WOLED. (b) EL spectra at various luminance, and two decomposed bands were blue and yellow exciplex emissions. Reproduced from reference [197]. [197]. 3.5. Tandem Hybrid WOLEDs Therefore, BCzVBi harvested a majority of singlet excitons, with the remainder of lower energy triplets diffusing through the conductive host CBP to directly excite the green and red phosphors. Such structure allowed high PE via the resonant energy transfer from the conductive host into both the singlet and triplet energy levels. Besides, the NPD thickness in each EL unit was optimized to form the desired white balance in the presence of weak optical interference. showed a maximum total EQE of 57% at a luminance of 1000 cd m , representing a 25% increase relative their previous tandem phosphorescent WOLEDs [155]. For the origin of such high performance, an efficient management of singlet and triplet excitons has been accomplished in the EML of each white EL unit, by locating a CBP spacer between blue fluorescent emitting zone CBP:4,4′-bis(9-ethyl-3-carbazovinylene)-1,1′-biphenyl (BCzVBi) and the green and red phosphorescent regions containing CBP:Ir(ppy)3 and CBP:PQIr, as shown in Figure 8. Therefore, BCzVBi harvested a majority of singlet excitons, with the remainder of lower energy triplets diffusing through the conductive host CBP to directly excite the green and red phosphors. Such structure allowed high PE via the resonant energy transfer from the conductive host into both the singlet and triplet energy levels. Besides, the NPD thickness in each EL unit was optimized to form the desired white balance in the presence of weak optical interference. Figure 8. Structure of organic layers of a two-element WOLED. Reproduced from reference Figure 8. Structure of organic layers of a two-element WOLED. Reproduced from reference [206]. Figure 8. Structure of organic layers of a two-element WOLED. Reproduced from reference Figure 8. Structure of organic layers of a two-element WOLED. Reproduced from reference [206]. 12 [206]. In Kanno’s tandem hybrid WOLEDs, the T1 of BCzVBi is lower than that of phosphorescent emitters. As a result, some triplet excitons are inevitably quenched by the high concentration of blue fluorophor, leading to the fact such device architecture is difficult to achieve 100% EQE [207–209]. To ensure the 100% IQE of tandem hybrid WOLEDs, Leo et al. proposed an effective strategy to develop tandem hybrid WOLEDs by using a CGL to interconnect the triplet-harvesting unit and mixed phosphorescent unit, as shown in Figure 9 [210]. [197]. 3.5. Tandem Hybrid WOLEDs 3.5. Tandem Hybrid WOLEDs With the combination of blue fluorescent/TADF emitters and green-red/yellow/orange phosphorescent emitters, hybrid WOLEDs can be achieved [199–205]. The first hybrid WOLED-based on blue fluorescent emitter was reported in 2003 [199]. Then, Sun et al. demonstrated the first efficient hybrid WOLED with a total EQE of 18.7% and a total PE of 37.6 lm W−1 [200]. In 2014, Zhang et al. realized the first hybrid WOLED based on blue TADF emitter, achieving a maximum forward-viewing EQE of 22.5% and a peak PE of 47.6 lm W−1 [55]. Due to the stable blue fluorescent emitters, most of available products in the WOLED With the combination of blue ﬂuorescent/TADF emitters and green-red/yellow/orange phosphorescent emitters, hybrid WOLEDs can be achieved [199–205]. The ﬁrst hybrid WOLED-based on blue ﬂuorescent emitter was reported in 2003 [199]. Then, Sun et al. demonstrated the ﬁrst efﬁcient hybrid WOLED with a total EQE of 18.7% and a total PE of 37.6 lm W−1 [200]. In 2014, Zhang et al. realized the ﬁrst hybrid WOLED based on blue TADF emitter, achieving a maximum forward-viewing EQE of 22.5% and a peak PE of 47.6 lm W−1 [55]. Due to the stable blue ﬂuorescent emitters, most of available products in the WOLED market are adopted the hybrid WOLED technology. Particularly, tandem hybrid WOLEDs are promising for the practical applications. [ ] p market are adopted the hybrid WOLED technology. Particularly, tandem hybrid WOLEDs are promising for the practical applications. To realize tandem hybrid WOLEDs, Kanno et al. used a CGL of Bphen:Li/MoO3 to interconnect two hybrid white EL units [206]. Although this CGL was the same as that of their tandem phosphorescent WOLEDs [155], the tandem hybrid WOLED with three EL units showed 11 of 28 heir nits 11 of 28 heir nits Molecules 2019, 24, 151 interconnect two t d h h a maximum total EQE of 57% at a luminance of 1000 cd m−2, representing a 25% increase relative their previous tandem phosphorescent WOLEDs [155]. For the origin of such high performance, an efﬁcient management of singlet and triplet excitons has been accomplished in the EML of each white EL unit, by locating a CBP spacer between blue ﬂuorescent emitting zone CBP:4,4′-bis(9-ethyl-3-carbazovinylene)-1,1′-biphenyl (BCzVBi) and the green and red phosphorescent regions containing CBP:Ir(ppy)3 and CBP:PQIr, as shown in Figure 8. [197]. 3.5. Tandem Hybrid WOLEDs have then developed efficient color stable inverted top-emitting tandem hybrid WOLEDs with ultra-thin wetting layer top electrodes [211], ITO-free tandem hybrid WOLEDs with angular color stability [212] and top-emitting tandem hybrid WOLEDs comprising laminated microlens films [213]. Figure 9. (a) Diffusively harvesting triplets. Due to the primarily hole-conducting character of 4P-NPD, excitons were formed close to Bphen. Whereas singlets recombine rapidly after creation, triplets diffuse to Ir(MDQ)2acac. (b) Patterned surface for enhanced light extraction from the substrate Reproduced from reference [210] Figure 9. (a) Diffusively harvesting triplets. Due to the primarily hole-conducting character of 4P-NPD, excitons were formed close to Bphen. Whereas singlets recombine rapidly after creation, triplets diffuse to Ir(MDQ)2acac. (b) Patterned surface for enhanced light extraction from the substrate. Reproduced from reference [210]. Figure 9. (a) Diffusively harvesting triplets. Due to the primarily hole-conducting character of 4P-NPD, excitons were formed close to Bphen. Whereas singlets recombine rapidly after creation, triplets diffuse to Ir(MDQ)2acac. (b) Patterned surface for enhanced light extraction from the substrate Reproduced from reference [210] Figure 9. (a) Diffusively harvesting triplets. Due to the primarily hole-conducting character of 4P-NPD, excitons were formed close to Bphen. Whereas singlets recombine rapidly after creation, triplets diffuse to Ir(MDQ)2acac. (b) Patterned surface for enhanced light extraction from the substrate. Reproduced from reference [210]. 13 Since blue TADF emitters can produce highly efficient blue emission by harvest both singlet and triplet excitons, such kind of emitters have great potential to develop tandem hybrid WOLEDs [214,215]. Recently, Hung et al. reported the first tandem hybrid WOLED by utilizing blue TADF emitter bis[4-(9,9-dimethyl-9,10-dihydroacridine)phenyl]sulfone Since blue TADF emitters can produce highly efﬁcient blue emission by harvest both singlet and triplet excitons, such kind of emitters have great potential to develop tandem hybrid WOLEDs [214,215]. Recently, Hung et al. reported the ﬁrst tandem hybrid WOLED by utilizing blue TADF emitter bis[4-(9,9-dimethyl-9,10-dihydroacridine)phenyl]sulfone (DMAC-DPS) and orange phosphorescent emitter:bis(4-phenylthieno-[3,2-c]pyridine) (acetylacetonate)iridium(III) (PO-01), showing the peak CE of 78.5 cd A−1 and EQE of 28.5% with the CIE coordinates of (0.33, 0.45) at 1000 cd m−2 [216]. The device structure was ITO/HAT-CN (10 nm)/TAPC (40 nm)/TCTA (10 nm)/DMAC-DPS (20 nm)/DPEPO (5 nm)/TPBI (40 nm)/Bphen:Li (1.2 , 15 nm)/HAT-CN (10 nm)/TAPC (40 nm)/TCTA (10 nm)/mCP:PO-01 (1.4 , 20 nm)/1,3,5-tri[(3-pyridyl)-phen-3-yl (TmPyPB, 40 nm)/Liq (2 nm)/Al (120 nm), as shown in Figure 10. Prior to the tandem hybrid WOLED, Hung et al. [197]. 3.5. Tandem Hybrid WOLEDs In the triplet-harvesting unit, highly efficient fluorescent blue bulk emitter 4P-NPD was used as the blue emitter and host of red phosphorescent emitter Ir(MDQ)2acac, since the T1 of 4P-NPD (2.3 eV) is higher than that of Ir(MDQ)2acac. Due to the p-type property of 4P-NPD, the structure of 65 nm HTL/10 nm Spiro-TAD/5 nm 4P-NPD:5 wt% Ir(MDQ)2acac/5 nm 4P-NPD/10 nm BPhen/55 nm BPhen:Cs/100 nm Al rendered that a narrow recombination zone In Kanno’s tandem hybrid WOLEDs, the T1 of BCzVBi is lower than that of phosphorescent emitters. As a result, some triplet excitons are inevitably quenched by the high concentration of blue ﬂuorophor, leading to the fact such device architecture is difﬁcult to achieve 100% EQE [207–209]. To ensure the 100% IQE of tandem hybrid WOLEDs, Leo et al. proposed an effective strategy to develop tandem hybrid WOLEDs by using a CGL to interconnect the triplet-harvesting unit and mixed phosphorescent unit, as shown in Figure 9 [210]. In the triplet-harvesting unit, highly efﬁcient ﬂuorescent blue bulk emitter 4P-NPD was used as the blue emitter and host of red phosphorescent emitter Ir(MDQ)2acac, since the T1 of 4P-NPD (2.3 eV) is higher than that of Ir(MDQ)2acac. Due to the p-type property of 4P-NPD, the structure of 65 nm HTL/10 nm Spiro-TAD/5 nm 4P-NPD:5 wt% Ir(MDQ)2acac/5 nm 4P-NPD/10 nm BPhen/55 nm BPhen:Cs/100 nm Al rendered that a narrow recombination zone close to the hole blocking layer BPhen. By optimizing the thickness of undoped 4P-NPD layer to be larger than the extension of the recombination zone, Ir(MDQ)2acac could only be excited by exciton diffusion owing to the different lifetimes of singlets and triplets (Figure 9). Therefore, the undoped 4P-NPD harvested the singlets, while Ir(MDQ)2acac consumed the triplets, producing 100% IQE white emission. In the mixed phosphorescent unit, the structure of 50 nm HTL/10 nm Spiro-TAD/10 nm TCTA:8 wt% Ir(ppy)3:1 wt% Ir(dhfpy)2acac/10 nm TPBi/50 nm BPhen:Cs/100 nm 12 of 28 nm and 12 of 28 nm and Molecules 2019, 24, 151 of 50 nm HTL/ TPBi/50 nm BPh Al ensured the green and yellow emissions from Ir(ppy)3 and Ir(dhfpy)2acac, respectively. The two phosphorescent emitters mixed in a common matrixTCTA were without loss of efﬁciency, which also reduced the voltage due to the reduction from two to one EML. As a consequence, both of the individual units had the ability to reach a 100% IQE. [197]. 3.5. Tandem Hybrid WOLEDs By stacking both units using a CGL consisting of a p/n-junction with a thin metal layer in between, the resultant architecture was 45 nm HTL (HTL1)/10 nm Spiro-TAD/5 nm 4P-NPD:5 wt% Ir(MDQ)2acac/5 nm 4P-NPD/10 nm BPhen/90 nm BPhen:Cs(ETL1)/0.5 nm Al/85 nm HTL (HTL2)/10 nm Spiro-TAD/10nm 8 wt% Ir(ppy)3:1 wt% Ir(dhfpy)2acac/10 nm TPBi/60 nm BPhen:Cs (ETL2)/100 nm Al. Thus, such device can be classiﬁed into type-IV tandem WOLED architecture. The PE and EQE were 33 lm W−1 and 26%, respectively. Furthermore, by using a quadratic pyramid pattern and a high-refractive-index hemisphere to harvest all the light coupled into the substrate, silver cathode to reduce the absorption, optimized thickness of each transport layer, a PE of 90.5 lm W−1 and EQE of 75.8% at 1000 cd m−2 were obtained with the help of strongly increased light extraction. By extending this design strategy, Leo et al. have then developed efﬁcient color stable inverted top-emitting tandem hybrid WOLEDs with ultra-thin wetting layer top electrodes [211], ITO-free tandem hybrid WOLEDs with angular color stability [212] and top-emitting tandem hybrid WOLEDs comprising laminated microlens ﬁlms [213]. I ( py) a a , e pe i e y e o p o p o e e e i e i e i a o o matrixTCTA were without loss of efficiency, which also reduced the voltage due to the reduction from two to one EML. As a consequence, both of the individual units had the ability to reach a 100% IQE. By stacking both units using a CGL consisting of a p/n-junction with a thin metal layer in between, the resultant architecture was 45 nm HTL (HTL1)/10 nm Spiro-TAD/5 nm 4P-NPD:5 wt% Ir(MDQ)2acac/5 nm 4P-NPD/10 nm BPhen/90 nm BPhen:Cs(ETL1)/0.5 nm Al/85 nm HTL (HTL2)/10 nm Spiro-TAD/10nm 8 wt% Ir(ppy)3:1 wt% Ir(dhfpy)2acac/10 nm TPBi/60 nm BPhen:Cs (ETL2)/100 nm Al. Thus, such device can be classified into type-IV tandem WOLED architecture. The PE and EQE were 33 lm W-1 and 26%, respectively. Furthermore, by using a quadratic pyramid pattern and a high-refractive-index hemisphere to harvest all the light coupled into the substrate, silver cathode to reduce the absorption, optimized thickness of each transport layer, a PE of 90.5 lm W-1 and EQE of 75.8% at 1000 cd m-2 were obtained with the help of strongly increased light extraction. By extending this design strategy, Leo et al. 3 6 Doing-Free Tandem WOLE 3.6. Doing-Free Tandem WOLEDs g Tandem WOLEDs can greatly boost the performance, however, their structures are intrinsically complicated compared with those of single-unit OLEDs. Additionally, the doping technology is required for high-performance tandem WOLEDs (e.g., the p-doping and n-doping charge transport layers, doping different-color EMLs, doping CGLs), which further complicating the structures [217–221]. To simplify the tandem WOLEDs, the doping-free technique may be conducive, since it can simplify the device engineering, shorten the fabrication procedure avert the utilization of host and lower the cost [222 224] Tandem WOLEDs can greatly boost the performance, however, their structures are intrinsically complicated compared with those of single-unit OLEDs. Additionally, the doping technology is required for high-performance tandem WOLEDs (e.g., the p-doping and n-doping charge transport layers, doping different-color EMLs, doping CGLs), which further complicating the structures [217–221]. To simplify the tandem WOLEDs, the doping-free technique may be conducive, since it can simplify the device engineering, shorten the fabrication procedure, avert the utilization of host and lower the cost [222–224]. 14 fabrication procedure, avert the utilization of host and lower the cost [222–224]. In 2007, Liu et al. demonstrated the doping-free tandem WOLEDs by managing an effective doping-free CGL of Bepp2 (25 nm) /KBH4 (1 nm)/Ag (0.5 nm)/HAT-CN (130 nm)/NPB (15 nm) to interconnect doping-free EMLs and doping-free charge transport layers [224]. The device architecture was ITO/HAT-CN (100 nm)/NPB (15 nm)/TAPC (5 In 2007, Liu et al. demonstrated the doping-free tandem WOLEDs by managing an effective doping-free CGL of Bepp2 (25 nm) /KBH4 (1 nm)/Ag (0.5 nm)/HAT-CN (130 nm)/NPB (15 nm) to interconnect doping-free EMLs and doping-free charge transport layers [224]. The device architecture was ITO/HAT-CN (100 nm)/NPB (15 nm)/TAPC (5 nm)/bis(2-phenyl-4,5-dimethylpyridinato)[2-(biphenyl-3-yl)pyridinato] iridium(III) (Ir(dmppy)2(dpp), 0.6 nm)/CGL/TAPC (5 nm)/FIrpic (0.5 nm)/TmPyPB (50 nm)/LiF (1 nm)/Al (200 nm), where Ir(dmppy)2(dpp) and FIrpic were the yellow and blue emitters (device W1), respectively, as shown in Figure 11. Additionally, by using the red emitter Ir(MDQ)2(acac) to replace Ir(dmppy)2(dpp) as well as optimizing the optical effect, another doping-free tandem WOLED was constructed (device W5). As a result, device W1 could accomplish the simpliﬁed structure/short fabrication time/reduced cost/high efﬁciency (81.2 cd A−1)/low efﬁciency roll-off/low voltage/high luminance (44,886 cd m−2) trade-off, while device W5 could possess an acceptable CRI of 67. For the high performance, the doping-free CGL was effective to ensure the charge generation and separation. [197]. 3.5. Tandem Hybrid WOLEDs realized an efﬁcient tandem green TADF OLED with the CGL of Bphen:Li/HAT-CN, achieving the EQE of 32.5%. Such results also demonstrated the CGL was effective, which was attributed to the efﬁcient charge 13 of 28 -CN, was 13 of 28 -CN, h was Molecules 2019, 24, 151 realized an eff achieving the EQ generation, excellent optical transparency and good electron transporting properties of Bphen; Li. On the other hand, since both blue TADF emitter and phosphorescent emitter could harvest singlets and triplets in their individual EL unit, high efﬁciency was attained. g g , p p y g transporting properties of Bphen; Li. On the other hand, since both blue TADF emitter and phosphorescent emitter could harvest singlets and triplets in their individual EL unit, high efficiency was attained. Figure 10. Device configurations of green (G1) and white (W1) OLEDs. Key molecular materials used for green emission (left) and those used for white emission (right) are shown. Reproduced from reference [216]. Figure 10. Device conﬁgurations of green (G1) and white (W1) OLEDs. Key molecular materials used for green emission (left) and those used for white emission (right) are shown. Reproduced from reference [216]. Figure 10. Device configurations of green (G1) and white (W1) OLEDs. Key molecular materials used for green emission (left) and those used for white emission (right) are shown. Reproduced from reference [216]. Figure 10. Device conﬁgurations of green (G1) and white (W1) OLEDs. Key molecular materials used for green emission (left) and those used for white emission (right) are shown. Reproduced from reference [216]. 3 6 Doing-Free Tandem WOLE 3.6. Doing-Free Tandem WOLEDs With the combination of KBH4 and Ag to modify the Bepp2/HAT-CN interface, the electron injection was improved, since metallic K was 14 of 28 e to dif 14 of 28 e to dif Molecules 2019, 24, 151 an acceptable C released by the thermal decomposition of KBH4 and the surface of Bepp2 ﬁlms was not an absolute plane. Hence, a thin K-doping Bepp2 layer at the KBH4/Bepp2 interface was formed. Additionally, Ag functioned as electrodes for both units, also improving electron injection. On the other hand, phosphorescent emitters were adopted in each unit to not only harvest both singlet and triplet excitons but also ensure the white light. the Bepp2/HAT-CN interface, the electron injection was improved, since metallic K was released by the thermal decomposition of KBH4 and the surface of Bepp2 films was not an absolute plane. Hence, a thin K-doping Bepp2 layer at the KBH4/Bepp2 interface was formed. Additionally, Ag functioned as electrodes for both units, also improving electron injection. On the other hand, phosphorescent emitters were adopted in each unit to not only harvest both singlet and triplet excitons but also ensure the white light. Figure 11. (Left): The structures of tandem WOLEDs. (Right): The chemical structure of used emitters. Reproduced from reference [224]. Figure 11. (Left): The structures of tandem WOLEDs. (Right): The chemical structure of used emitters. Reproduced from reference [224]. Figure 11. (Left): The structures of tandem WOLEDs. (Right): The chemical structure of used emitters. Reproduced from reference [224]. Figure 11. (Left): The structures of tandem WOLEDs. (Right): The chemical structure of used emitters. Reproduced from reference [224]. emitters. Reproduced from refe g Reproduced from reference [224]. 4. Summary and Outloo 4. Summary and Outlook Since tandem device architectures can impressively boost the luminance, efficiency and lifetime, the excellent characteristics render that tandem WOLED have been extensively investigated. In this review, we have mainly focused on recent developments in tandem WOLEDs and summarized the advanced strategies to achieve high-performance tandem WOLEDs. Particularly, we have emphasized representative tandem fluorescent WOLEDs, tandem phosphorescent WOLEDs, tandem TADF WOLEDs, and tandem hybrid WOLEDs. Additionally, we have also presented doping-free tandem WOLEDs. The detailed f f d O E h b d b d bl Since tandem device architectures can impressively boost the luminance, efﬁciency and lifetime, the excellent characteristics render that tandem WOLED have been extensively investigated. In this review, we have mainly focused on recent developments in tandem WOLEDs and summarized the advanced strategies to achieve high-performance tandem WOLEDs. Particularly, we have emphasized representative tandem ﬂuorescent WOLEDs, tandem phosphorescent WOLEDs, tandem TADF WOLEDs, and tandem hybrid WOLEDs. Additionally, we have also presented doping-free tandem WOLEDs. The detailed performances for tandem WOLEDs have been described in Table 1. 15 of 28 Molecules 2019, 24, 151 Table 1. Summarized performances for representative tandem WOLEDs. Table 1. Summarized performances for representative tandem WOLEDs. WOLEDs a Von b (v) EQEmax c (%) PEmax d (lm W−1) CEmax e (cd A−1) CIE f CRI g Ref. [72] - - 2.6 10.7 (0.36, 0.34) - Ref. [155] - 34.9 22.7 77.0 (0.35, 0.44) 66 Ref. [162] - 43.3 45.2 110.9 (0.34, 0.41) - Ref. [163] 5.7 54.3 63 - (0.359, 0.498) - Ref. [163] h - 92.4 100 - (0.336, 0.452) - Ref. [174] - 74.3 24 - (0.46, 0.43) 88.6 Ref. [174] h - 171 50 - (0.49, 0.43) 89.4 Ref. [197] 4.0 11.6 15.8 27.7 (0.29, 0.35) 70.6 Ref. [206] - 33 14 - (0.38, 0.44) 82 Ref. [210] - ~26 ~40 - (0.505, 0.422) 77.6 Ref. [210] h - ~78 ~100 - - - Ref. [216] 7.4 28.5 - 78.5 (0.33, 0.45) 82 Ref. [224] 5.1 - 81.2 42.9 (0.35, 0.47) - a Representative tandem WOLEDs. b Turn-on voltage. c Peak EQE. d Peak PE. e Peak CE. f CIE coordinates at ~1000 cd m−2. g Peak CRI. h With the use of outcoupling technique. a Representative tandem WOLEDs. b Turn-on voltage. c Peak EQE. d Peak PE. e Peak CE. f CIE coordinates at ~1000 cd m−2. g Peak CRI. h With the use of outcoupling technique. 4. Summary and Outloo 4. Summary and Outlook Over the past few years, the performance of tandem WOLEDs has incrementally enhanced and nowadays can satisfy the demand of real commercialization for handphones, lamps and televisions. In particular, phosphorescent and TADF emitters are favorable to increase the performance. To date, there are still many challenges hindering the further development of tandem WOLEDs, including the efﬁciency, driving voltage, operational stability as well as viewing angle dependence. First, the PE of tandem WOLEDs still lags behind. For example, the theoretical efﬁciency limit for WOLEDs is about 248 lm W−1, indicating that there is still much room for tandem WOLEDs [225]. Besides, the voltage of tandem WOLEDs is much higher than that of single-unit WOLEDs, which increases the power consumption and easily leads to the low PE [226–228]. Therefore, each EL unit is required to harvest all excitons for efﬁcient emission, while the CGL is needed to be effective for the charge generation and charge separation. In the internal physical processes, the elaborative management of the charge and exciton distribution is conducive. For the external procedures of light propagation, the utilization of outcoupling technologies can vastly enhance the efﬁciency as well as the lifetime [229–231]. This is because the use of light extraction layers can extract the light trapped by the substrate or the inner layers due to the total reﬂection. In addition, the light extraction layer (e.g., laminated microlens ﬁlms) has been utilized to improve the CRI of tandem WOLEDs [214]. Therefore, although only a few outcoupling technologies were exclusively reported to enhance the performance of tandem WOLEDs, it is believed that the light extraction layers are promising for the further development of tandem WOLEDs. Additionally, the adoption of stable emitters and CGLs is also helpful to the lifetime. Speciﬁcally, since the total thickness of tandem WOLEDs is very thick and the number of layers is large, the microcavity effect easily occurs because of the refractive index differences between adjacent layers (e.g., n = 1.7 for organic ﬁlms, n = 1.9 for ITO, and n = 2.2 for MoO3). Thus, tandem WOLEDs should have the Lambertian emission and the color should be stable over all angles (regardless of the center or the corner), otherwise chromaticity angular dependence will become a serious issue since the appearance of items can be rely on the locations [232]. Additionally, the device architecture of tandem WOLEDs is usually complicated. Author Contributions: P.X. and B.L. conceived the idea; P.X. and B.L. wrote the paper; J.H. and Y.Y. advised the paper; B.L. and Y.Y. supervised the project. All authors reviewed the paper. Funding: The authors are grateful to the ﬁnancial support from the National Natural Science Foundation of China (Grant Nos. 61804029, 11504050), Guangdong Natural Science Foundation (Grant No. 2018A030310353), Scientiﬁc References Met. 1998, 94, 245–248. [CrossRef] 7. Ma, Y.; Zhang, H.; Shen, J.; Che, C. Electroluminescence from triplet metal-ligand charge-transfer excited state of transition metal complexes. Synth. Met. 1998, 94, 245–248. [CrossRef] 8. Baldo, M.A.; O’brien, D.F.; You, Y.; Shoustikov, A.; Sibley, S.; Thompson, M.E.; Forrest, S.R. Highly Efﬁcient Phosphorescent Emission From Organic Electroluminescent Devices. Nature 1998, 395, 151–154. [CrossRef] 9. Liu, B.; Gao, D.; Wang, J.; Wang, X.; Wang, L.; Zou, J.; Ning, H.; Peng, J. Progress of White Organic Light-Emitting Diodes. Acta Phys. Chim. Sin. 2015, 31, 1823–1852. 10. Seifert, R.; Moraes, I.R.D.; Scholz, S.; Gather, M.C.; Lüssem, B.; Leo, K. Chemical Degradation Mechanisms of Highly Efﬁcient Blue Phosphorescent Emitters Used for Organic Light Emitting Diodes. Org. Electron. 2013, 14, 115–123. [CrossRef] 11. Liu, B.; Xu, M.; Tao, H.; Ying, L.; Zou, J.; Wu, H.; Peng, J. Highly Efﬁcient Red Phosphorescent Organic Light-Emitting Diodes Based on Solution Processed Emissive Layer. J. Lumin. 2013, 142, 35–39. [CrossRef] 12. Xiang, C.; Koo, W.; So, F.; Sasabe, H.; Kido, J. A Systematic Study on Efﬁciency Enhancements in Phosphorescent Green, Red and Blue Microcavity Organic Light Emitting Devices. Light Sci. Appl. 2013, 2, e74. [CrossRef] 13. Rao, M.V.M.; Su, Y.K.; Huang, T.S.; Chen, Y.C. White Organic Light Emitting Devices Based on Multiple Emissive Nanolayers. Nano-Micro Lett. 2010, 2, 242–246. 14. Eslamian, M. Inorganic and Organic Solution-Processed Thin Film Devices. Nano-Micro Lett. 2017, 9, 3. [CrossRef] [PubMed] 15. Burroughes, J.H.; Bradley, D.D.C.; Brown, A.R.; Marks, R.N.; Mackay, K.; Friend, R.H.; Burns, P.L.; Homes, A.B. Light-Emitting Diodes Based on Conjugated Polymers. Nature 1990, 347, 539–541. [CrossRef] 16. Uoyama, H.; Goushi, K.; Shizu, K.; Nomura, H.; Adachi, C. Highly Efﬁcient Organic Light-Emitting Diodes from Delayed Fluorescence. Nature 2012, 492, 234–238. [CrossRef] [PubMed] 15. Burroughes, J.H.; Bradley, D.D.C.; Brown, A.R.; Marks, R.N.; Mackay, K.; Friend, R.H.; Bur Homes, A.B. Light-Emitting Diodes Based on Conjugated Polymers. Nature 1990, 347, 539–541. [Cr , g g j g y , , [ ] 16. Uoyama, H.; Goushi, K.; Shizu, K.; Nomura, H.; Adachi, C. Highly Efﬁcient Organic Light-Emitting Diodes from Delayed Fluorescence. Nature 2012, 492, 234–238. [CrossRef] [PubMed] 17. Nishimoto, T.; Yasuda, T.; Lee, S.Y.; Kondo, R.; Adachi, C. A Six-carbazole-decorated Cyclophosphazene as a Host with High Triplet Energy to Realize Efﬁcient Delayed-Fluorescence OLEDs. Mater. Horiz. 2014, 1, 264–269. [CrossRef] 18. Zhang, Q.; Tsang, D.; Kuwabara, H.; Hatae, Y.; Li, B.; Takahashi, T.; Lee, S.Y.; Yasuda, T.; Adachi, C. 4. Summary and Outloo 4. Summary and Outlook The introduction of a doping-free technique or solution-processed fabrication method is expected to be useful. After solving the aforementioned problems, the prospect of mass production for tandem WOLEDs will be possible and the proposed solutions are also beneﬁcial to the related optoelectronic ﬁeld (e.g., display, lighting, laser, solar cell, photodetectors and sensors) [233–238]. Author Contributions: P.X. and B.L. conceived the idea; P.X. and B.L. wrote the paper; J.H. and Y.Y. advised the paper; B.L. and Y.Y. supervised the project. All authors reviewed the paper. Funding: The authors are grateful to the ﬁnancial support from the National Natural Science Foundation of China (Grant Nos. 61804029, 11504050), Guangdong Natural Science Foundation (Grant No. 2018A030310353), Scientiﬁc 16 of 28 16 of 28 Molecules 2019, 24, 151 Research Starting Foundation of Foshan University (Grang No. Gg040926), Foshan Science and technology innovation special funds (Grant No. 2017EZ100111). Research Starting Foundation of Foshan University (Grang No. Gg040926), Foshan Science and technology innovation special funds (Grant No. 2017EZ100111). Conﬂicts of Interest: The authors declare no conﬂict of interest. References 1. Yang, X.; Zhou, G.; Wong, W.-Y. Functionalization of Phosphorescent Emitters and Their Host Materials by Main-Group Elements for Phosphorescent Organic Light-Emitting Devices. Chem. Soc. Rev. 2015, 44, 8484–8575. [CrossRef] [PubMed] 1. Yang, X.; Zhou, G.; Wong, W.-Y. Functionalization of Phosphorescent Emitters and Their Host Materials by Main-Group Elements for Phosphorescent Organic Light-Emitting Devices. Chem. Soc. Rev. 2015, 44, 8484–8575. [CrossRef] [PubMed] 2. Jou, J.-H.; Kumar, S.; Agrawal, A.; Li, T.-H.; Sahoo, S. Approaches for Fabricating High Efﬁciency Organic Light Emitting Diodes. J. Mater. Chem. C 2015, 3, 2974–3002. [CrossRef] 2. Jou, J.-H.; Kumar, S.; Agrawal, A.; Li, T.-H.; Sahoo, S. Approaches for Fabricating High Efﬁciency Organic Light Emitting Diodes. J. Mater. Chem. C 2015, 3, 2974–3002. [CrossRef] 3. Liu, B.; Li, X.; Tao, H.; Zou, J.; Xu, M.; Wang, L.; Peng, J.; Cao, Y. Manipulation of Exciton Distribution for High-Performance Fluorescent/Phosphorescent Hybrid White Organic Light-Emitting Diodes. J. Mater. Chem. C 2017, 5, 7668–7683. [CrossRef] 3. Liu, B.; Li, X.; Tao, H.; Zou, J.; Xu, M.; Wang, L.; Peng, J.; Cao, Y. Manipulation of Exciton Distribution for High-Performance Fluorescent/Phosphorescent Hybrid White Organic Light-Emitting Diodes. J. Mater. Chem. C 2017, 5, 7668–7683. [CrossRef] 4. Tyan, Y.S. Organic Light-Emitting-Diode Lighting Overview. J. Photon. Energy 2011, 1, 011009. [CrossRef] 5 T C W V Sl k VA O i El t l i t Di d A l Ph L tt 1987 51 913 [C R f] 4. Tyan, Y.S. Organic Light-Emitting-Diode Lighting Overview. J. Photon. Energy 2011, 1, 011009. [CrossRef] 5. Tang, C.W.; VanSlyke, V.A. Organic Electroluminescent Diodes. Appl. Phys. Lett. 1987, 51, 913. [CrossRef] 4. Tyan, Y.S. Organic Light-Emitting-Diode Lighting Overview. J. Photon. Energy 2011, 1, 011009. [CrossRef] 5. Tang, C.W.; VanSlyke, V.A. Organic Electroluminescent Diodes. Appl. Phys. Lett. 1987, 51, 913. [CrossRef] 6. Fan, C.; Yang, C. Yellow/orange emissive heavy-metal complexes asphosphors in monochromatic and white organic light-emitting devices. Chem. Soc. Rev. 2014, 43, 6439–6469. [CrossRef] 5. Tang, C.W.; VanSlyke, V.A. Organic Electroluminescent Diodes. Appl. Phys. Lett. 1987, 51, 913. [CrossRef] 6. Fan, C.; Yang, C. Yellow/orange emissive heavy-metal complexes asphosphors in monochromatic and white organic light-emitting devices. Chem. Soc. Rev. 2014, 43, 6439–6469. [CrossRef] 6. Fan, C.; Yang, C. Yellow/orange emissive heavy-metal complexes asphosphors in monochromatic and white organic light-emitting devices. Chem. Soc. Rev. 2014, 43, 6439–6469. [CrossRef] 7. Ma, Y.; Zhang, H.; Shen, J.; Che, C. Electroluminescence from triplet metal-ligand charge-transfer excited state of transition metal complexes. Synth. References Nearly 100% Internal Quantum Efﬁciency in Undoped Electroluminescent Devices Employing Pure Organic Emitters. Adv. Mater. 2015, 27, 2096–2100. [CrossRef] [PubMed] 19. Zhang, Q.S.; Li, B.; Huang, S.P.; Nomura, H.; Tanaka, H.; Adachi, C. Efﬁcient blue organic light-emitting diodes employing thermally activated delayed ﬂuorescence. Nat. Photon. 2014, 8, 326–332. [CrossRef] 20. Ying, L.; Ho, C.L.; Wu, H.; Cao, Y.; Wong, W.Y. White Polymer Light-Emitting Devices for Solid-State Lighting: Materials, Devices, and Recent Progress. Adv. Mater. 2014, 26, 2459–2473. [CrossRef] 21. Zou, J.; Wu, H.; Lam, C.S.; Wang, C.; Zhong, C.; Hu, S.; Ho, C.L.; Zhou, G.J.; Wu, H.; Choy, W.C.; et al. Simultaneous Optimization of Charge-Carrier Balance and Luminous Efﬁcacy in Highly Efﬁcient White Polymer Light-Emitting Devices. Adv. Mater. 2011, 23, 2976–2980. [CrossRef] [PubMed] 17 of 28 17 of 28 Molecules 2019, 24, 151 22. Niu, Y.H.; Liu, M.; Ka, J.W.; Bardeker, J.; Zin, M.; Schoﬁeld, Y.; Chi, Y.; Jen, A.K.-Y. Crosslinkable Hole-Transport Layer on Conducting Polymer for High-Efﬁciency White Polymer Light-Emitting Diodes. Adv. Mater. 2010, 9, 300–304. [CrossRef] 23. Wang, H.; Meng, L.Q.; Shen, X.X.; Wei, X.F.; Zheng, X.L.; Lv, X.P.; Yi, Y.P.; Wang, Y.; Wang, P.F. Light-Emitting Diodes: Highly Efﬁcient Orange and Red Phosphorescent Organic Light-Emitting Diodes with Low Roll-Off of Efﬁciency using a Novel Thermally Activated Delayed Fluorescence Material as Host. Adv. Mater. 2015, 27, 4041–4047. [CrossRef] 24. Kim, B.S.; Lee, J.Y. Engineering of Mixed Host for High External Quantum Efﬁciency above 25% in Green Thermally Activated Delayed Fluorescence Device. Adv. Funct. Mater. 2015, 24, 3970–3977. [CrossRef] 25. Luo, D.X.; Yang, Y.B.; Huang, L.; Liu, B.; Zhao, Y. High-performance hybrid white organic light-emitting diodes exploiting blue thermally activated delayed ﬂuorescent dyes. Dyes Pigments 2017, 147, 83–89. [CrossRef] 26. Sasabe, H.; Kido, J. Development of High Performance OLEDs for General Lighting. J. Mater. Chem. C 2013, 1, 1699–1707. [CrossRef] 27. Wang, Q.; Ma, D. Management of Charges and Excitons for High-performance White Organic Light-emitting Diodes. Chem. Soc. Rev. 2010, 39, 2387–2398. [CrossRef] [PubMed] Chem. Soc. Rev. 2010, 39, 2387–2398. [CrossRef] [PubM 8. Chen, J.; Zhao, F.; Ma, D. Hybrid White OLEDs with Fluorophors and Phosphors. Mater. Today 2014 175–183. [CrossRef] 29. Zhang, L.; Li, X.-L.; Luo, D.; Xiao, P.; Xiao, W.; Song, Y.; Ang, Q.; Liu, B. Strategies to Achieve High-Performance White Organic Light-Emitting Diodes. Materials 2017, 10, 1378. [CrossRef] 30. Gather, M.C.; Reineke, S. Recent Advances in Light Outcoupling from White Organic Light-Emitting Diodes. J. Photon. References [CrossRef] 18 of 28 Molecules 2019, 24, 151 18 of 28 41. Sun, N.; Zhao, Y.B.; Zhao, F.C.; Chen, Y.H.; Yang, D.Z.; Chen, J.S.; Ma, D.G. A white organic light-emitting diode with ultra-high color rendering index, high efﬁciency, and extremely low efﬁciency roll-off. Appl. Phys. Lett. 2014, 105, 013303. [CrossRef] 2. Jou, J.-H.; Wu, R.-Z.; Yu, H.-H.; Li, C.-J.; Jou, Y.-C.; Peng, S.-H.; Chen, Y.-L.; Chen, C.-T.; Shen, S.-M.; Joer et al. Artiﬁcial Dusk-Light Based on Organic Light Emitting Diodes. ACS Photon. 2014, 1, 27–31. [Cross g g g g [ ] 43. Jou, J.-H.; Hsieh, C.-Y.; Tseng, J.-R.; Peng, S.-H.; Jou, Y.-C.; Hong, J.H.; Shen, S.-M.; Tang, M.-C.; Chen, P.-C.; Lin, C.-H. Candle Light-Style Organic Light-Emitting Diodes. Adv. Funct. Mater. 2013, 23, 2750–2757. [CrossRef] 44. Shen, S.M.; Lin, C.R.; Wang, Y.S.; Chou, Y.C.; Chen, S.Z.; Jou, J.H. Efﬁcient Very-High Color Rendering index Organic Light-Emitting Diode. Org. Electron. 2011, 12, 865–868. 45. Liu, B.; Xu, M.; Wang, L.; Tao, H.; Su, Y.; Gao, D.; Lan, L.; Zou, J.; Peng, J. Very-High Color Rendering Index Hybrid White Organic Light-Emitting Diodes with Double Emitting Nanolayers. Nano-Micro Lett. 2014, 6, 335–339. [CrossRef] 46. Liu, B.; Luo, D.; Gao, D.; Wang, X.; Xu, M.; Zou, J.; Ning, H.; Wang, L.; Peng, J.; Cao, Y. An ideal host-guest system to accomplish high-performance greenishyellow and hybrid white organic light-emitting diodes. Org. Electron. 2015, 27, 29–34. [CrossRef] 7. Jou, J.H.; Wu, M.H.; Shen, S.M.; Wang, H.C.; Chen, S.Z.; Chen, S.H.; Lin, C.R.; Hsieh, Y.L. Sunlight-S Color-Temperature Tunable Organic Light-Emitting Diode. Appl. Phys. Lett. 2009, 95, 013307. [CrossRe 48. Liu, B.; Nie, H.; Zhou, X.B.; Hu, S.B.; Luo, D.X.; Gao, D.Y.; Zou, J.H.; Xu, M.; Wang, L.; Zhao, Z.J.; et al. Manipulation of Charge and Exciton Distribution Based on Blue Aggregation-Induced Emission Fluorophors: A Novel Concept to Achieve High-Performance Hybrid White Organic Light-Emitting Diodes. Adv. Funct. Mater. 2016, 26, 776–783. [CrossRef] 49. Liu, B.; Nie, H.; Lin, G.; Hu, S.; Gao, D.; Zou, J.; Xu, M.; Wang, L.; Zhao, Z.; Ning, H.; et al. High-Performance Doping-Free Hybrid White OLEDs Based on Blue Aggregation-Induced Emission Luminogens. ACS Appl. Mater. Interfaces 2017, 9, 34162–34171. [CrossRef] 50. Liu, B.; Wang, L.; Gao, D.Y.; Xu, M.; Zhu, X.H.; Zou, J.H.; Lan, L.F.; Ning, H.L.; Peng, J.B.; Cao, Y. Harnessing charge and exciton distribution towards extremely high performance: The critical role of guests in single-emitting-layer white OLEDs. Mater. Horiz. 2015, 2, 536–544. References Energy 2015, 5, 057607. [CrossRef] 31. Huang, H.; Yang, X.; Wang, Y.; Pan, B.; Wang, L.; Chen, J.; Ma, D.; Yang, C. Optimizing the conjugation between N,N′-dicarbazolyl-3,5-benzene and triphenylphosphine oxide as bipolar hybrids for highly efﬁcient blue and single emissive layer white phosphorescent OLEDs. Org. Electron. 2013, 14, 2573–2581. [CrossRef] 32. Kido, J.; Hongawa, K.; Okuyama, K.; Nagai, K. White Light-Emitting Organic Electroluminescent Devices Using The Poly (N-vinylcarbazole) Emitter Layer Doped with Three Fluorescent Dyes. Appl. Phys. Lett. 1994, 64, 815. [CrossRef] 33. Kido, J.; Kimura, M.; Nagai, K. Multilayer White Light-Emitting Organic Electroluminescent Device. Science 1995, 267, 1332–1334. [CrossRef] [PubMed] 34. Ou, Q.D.; Zhou, L.; Li, Y.Q.; Chen, S.; Chen, J.D.; Li, C.; Wang, Q.K.; Lee, S.T.; Tang, J.X. Light-Emitting Diodes: Extremely Efﬁcient White Organic Light-Emitting Diodes for General Lighting. Adv. Funct. Mater. 2014, 24, 7249. [CrossRef] 35. Liu, B.; Wang, L.; Xu, M.; Tao, H.; Gao, D.; Zou, J.; Lan, L.; Ning, H.; Peng, J.; Cao, Y. Extremely Stable-color Flexible White Organic Light-emitting Diodes with Efﬁciency Exceeding 100 lm W−1. J. Mater. Chem. C 2014, 2, 9836. [CrossRef] 36. Liu, X.-K.; Zhan, C.; Jian, Q.; Wen-Jun, Z.; Bo, W.; Hoi Lam, T.; Zhu, F.; Zhang, X.-H.; Lee, C.-S. Organic light-emitting devices: Remanagement of singlet and triplet excitons in single-emissive-layer hybrid white organic light-emitting devices using thermally activated delayed ﬂuorescent blue exciplex. Adv. Mater. 2015, 27, 7079–7085. [CrossRef] 37. Wu, Z.; Yu, L.; Zhao, F.; Qiao, X.; Chen, J.; Ni, F.; Yang, C.; Ahamad, T.; Alshehri, S.M.; Ma, D. Precise Exciton Allocation for Highly Effcient White Organic Light-Emitting Diodes with Low Effciency Roll-Off Based on Blue Thermally Activated Delayed Fluorescent Exciplex Emission. Adv. Opt. Mater. 2017, 5, 1700415. [CrossRef] 38. Xiao, P.; Dong, T.; Xie, J.; Luo, D.; Yuan, J.; Liu, B. Emergence of White Organic Light-Emitting Diodes Based on Thermally Activated Delayed Fluorescence. Appl. Sci. 2018, 8, 299. [CrossRef] 39. Li, X.L.; Xie, G.Z.; Liu, M.; Chen, D.C.; Cai, X.Y.; Peng, J.B.; Cao, Y.; Su, S.J. High-Efﬁciency WOLEDs with High Color-Rendering Index based on a Chromaticity-Adjustable Yellow Thermally Activated Delayed Fluorescence Emitter. Adv. Mater. 2016, 28, 4614–4619. [CrossRef] 40. Liu, B.; Zou, J.H.; Zhou, Z.W.; Wang, L.; Xu, M.; Tao, H.; Gao, D.Y.; Lan, L.Y.; Ning, H.L.; Peng, J.B. Efﬁcient single-emitting layer hybrid white organic light-emitting diodes with low efﬁciency roll-off, stable color and extremely high luminance. J. Ind. Eng. Chem. 2015, 30, 85–91. References [CrossRef] 51. Sun, N.; Wang, Q.; Zhao, Y.B.; Yang, D.Z.; Zhao, F.C.; Chen, J.S.; Ma, D.G. A hybrid white organic light-emitting diode with above 20% external quantum efﬁciency and extremely low efﬁciency roll-off. J. Mater. Chem. C 2014, 2, 7494–7504. [CrossRef] 52. Zhao, F.C.; Zhang, Z.Q.; Liu, Y.P.; Dai, Y.F.; Chen, J.S.; Ma, D. A Hybrid White Organic Light-Emitting Diode with Stable Color and Reduced Efﬁciency Roll-Off by Using a Bipolar Charge Carrier Switch. Org. Electron. 2012, 13, 1049–1055. [CrossRef] 53. Schwartz, G.; Reineke, S.; Rosenow, T.C.; Walzer, K.; Leo, K. Triplet Harvesting in Hybrid White Organic Light-Emitting Diodes. Adv. Funct. Mater. 2009, 19, 1319–1333. [CrossRef] 54. Zhang, D.D.; Duan, L.; Zhang, Y.G.; Cai, M.H.; Zhang, D.Q.; Qiu, Y. Highly efﬁcient hybrid warm white organic light-emitting diodes using a blue thermally activated delayed ﬂuorescence emitter: Exploiting the external heavy-atom effect. Light Sci. Appl. 2015, 4, e232. [CrossRef] 55. Zhang, D.D.; Duan, L.; Li, Y.L.; Zhang, D.Q.; Qiu, Y. Highly efﬁcient and color-stable hybrid warm white organic light-emitting diodes using a blue material with thermally activated delayed ﬂuorescence. J. Mater. Chem. C 2014, 2, 8191–8197. [CrossRef] 56. Zhang, D.D.; Zhang, D.Q.; Duan, L. Exploiting p-Type Delayed Fluorescence in Hybrid White OLEDs: Breaking the Trade-off between High Device Efﬁciency and Long Lifetime. ACS Appl. Mater. Interfaces 2016, 8, 23197–23203. [CrossRef] 57. Liu, B.; Wang, L.; Xu, M.; Tao, H.; Zou, J.H.; Gao, D.Y.; Lan, L.F.; Ning, H.L.; Peng, J.B.; Cao, Y. Efﬁcient Hybrid White Organic Light-emitting Diodes with Extremely Long Lifetime: The Effect of n-type Interlayer. Sci. Rep. 2014, 4, 7198. [CrossRef] [PubMed] 58. Kim, G.H.; Lampande, R.; Im, J.B.; Lee, J.M.; Lee, J.Y.; Kwon, J.H. Controlling the exciton lifetime of blue thermally activated delayed ﬂuorescence emitters using a heteroatom-containing pyridoindole donor moiety. Mater. Horiz. 2017, 4, 619–624. [CrossRef] 19 of 28 19 of 28 Molecules 2019, 24, 151 59. Duan, L.; Zhang, D.Q.; Wu, K.W.; Huang, X.Q.; Wang, L.D.; Qiu, Y. Controlling the Recombination Zone of White Organic Light-Emitting Diodes with Extremely Long Lifetimes. Adv. Funct. Mater. 2011, 21, 3540–3545. [CrossRef] 60. Yook, K.S.; Jeon, S.O.; Min, S.Y.; Lee, J.Y.; Yang, H.J.; Noh, T.; Kang, S.K.; Lee, T.W. Highly Efﬁcient p-i-n and Tandem Organic Light-Emitting Devices Using an Air-Stable and Low-Temperature-Evaporable Metal Azide as an n-Dopant. Adv. Funct. Mater. 2010, 20, 1797–1802. [CrossRef] 61. Sun, H.D.; Guo, Q.X.; Yang, D.Z.; Chen, Y.H.; Chen, J.S.; Ma, D.G. References High Efﬁciency Tandem Organic Light Emitting Diode Using an Organic Heterojunction as the Charge Generation Layer: An Investigation into the Charge Generation Model and Device Performance. ACS Photon. 2015, 2, 271–279. [CrossRef] 62. Ding, L.; Sun, Y.Q.; Chen, H.; Zu, F.S.; Wang, Z.K.; Liao, L.S. A novel intermediate connector with improved charge generation and separation for large-area tandem white organic lighting devices. J. Mater. Chem. C 2014, 2, 10403–10408. [CrossRef] 63. Chang, C.C.; Chen, J.F.; Hwang, S.W.; Chen, C.H. Highly efﬁcient white organic electroluminescent devices based on tandem architecture. Appl. Phys. Lett. 2005, 87, 253501. [CrossRef] 64. Hong, T.; Gao, D.; Liu, B.; Wang, L.; Zou, J.; Xu, M.; Peng, J. Enhancement of Tandem Organic Light-Emitting Diode Performance by Inserting An Ultra-Thin Ag Layer in Charge Generation Layer. Acta Phys. Sin. 2017, 1, 017302. 65. Lei, D.; Tang, X.; Xu, M.F.; Shi, X.B.; Wang, Z.K.; Liao, L.S. Lithium Hydride Doped Intermediate Connector for High-Efﬁciency and Long-Term Stable Tandem Organic Light-Emitting Diodes. ACS Appl. Mater. Interfaces 2014, 6, 18228–18232. 66. Liao, L.S.; Klubek, K.P.; Tang, C.W. High-Efﬁciency Tandem Organic Light-Emitting Diodes. Appl. Phys. Lett. 2004, 84, 167–169. [CrossRef] 67. Liu, J.; Wang, J.; Huang, S.; Shi, X.; Wu, X.; He, G. A highly efﬁcient, transparent and stable charge generation unit based on a p-doped monolayer. Org. Electron. 2013, 14, 1337–1343. [CrossRef] 68. Bao, Q.Y.; Yang, J.P.; Tang, J.X.; Li, Y.; Lee, C.S.; Lee, S.T. Interfacial Electronic Structures of WO3-Based Intermediate Connectors in Tandem Organic Light-Emitting Diodes. Org. Electron. 2010, 11, 1578–1583. [CrossRef] 9. Liao, L.S.; Klubek, K.P. Power Efﬁciency Improvement in a Tandem Organic Light-Emitting Diode. A Phys. Lett. 2008, 92, 223311. [CrossRef] Phys. Lett. 2008, 92, 223311. [CrossRef] 70. Yang, Y.; Peng, T.; Ye, K.Q.; Wu, Y.; Liu, Y.; Wang, Y. High-efﬁciency and high-quality white organic light-emitting diode employing ﬂuorescent emitters. Org. Electron. 2011, 12, 29–33. [CrossRef] 70. Yang, Y.; Peng, T.; Ye, K.Q.; Wu, Y.; Liu, Y.; Wang, Y. High-efﬁciency and high-quality white organic light-emitting diode employing ﬂuorescent emitters. Org. Electron. 2011, 12, 29–33. [CrossRef] 71. Chen, Y.H.; Wang, Q.; Chen, J.S.; Ma, D.; Yan, D.H.; Wang, L.X. Organic Semiconductor Heterojunction as Charge Generation Layer in Tandem Organic Light-Emitting Diodes for High Power Efﬁciency. Org. Electron. 2012, 13, 1121–1128. [CrossRef] 72. Guo, F.; Ma, D. White organic light-emitting diodes based on tandem structures. Appl. Phys. Lett. 2005, 87, 173510. [CrossRef] 73. References High-Efﬁciency and High-Luminance Three-Color White Organic Light-Emitting Diodes with Low Efﬁciency Roll-Off. ECS J. Solid State Sci. Technol. 2018, 7, R99–R103. [CrossRef] 85. Huang, W.Y.; Chen, Z.W.; You, H.W.; Fan, F.C.; Chen, H.F.; Wong, K.T. Efﬁcient carrier- and exciton-conﬁning device structure that enhances blue PhOLED efﬁciency and reduces efﬁciency roll-off. Org. Electron. 2011, 12, 575–581. [CrossRef] 86. Zhang, B.; Tan, G.; Lam, C.S.; Yao, B.; Ho, C.L.; Liu, L.; Xie, Z.; Wong, W.Y.; Ding, J.; Wang, L. High-Efﬁciency Single Emissive Layer White Organic Light-Emitting Diodes Based on Solution-Processed Dendritic Host and New Orange-Emitting Iridium Complex. Adv. Mater. 2012, 24, 1873–1877. [CrossRef] [PubMed] 87. Yin, Y.; Piao, X.; Li, Y.; Wang, Y. High-Efﬁciency and Low-Efﬁciency-Roll-Off Single-Layer White Organic Light-Emitting Devices with a Bipolar Transport Host. Appl. Phys. Lett. 2012, 101, 063306. [CrossRef] 88. Chang, Y.L.; Song, Y.; Wang, Z.; Helander, M.G.; Qiu, J.; Chai, L.; Liu, Z.; Scholes, G.D.; Lu, Z. Highly Efﬁcient Warm White Organic Light-Emitting Diodes by Triplet Exciton Conversion. Adv. Funct. Mater. 2013, 23, 705–712. [CrossRef] 89. Liu, B.; Luo, D.X.; Zou, J.H.; Gao, D.Y.; Ning, H.L.; Wang, L.; Peng, J.B.; Cao, Y. A host-guest system comprising high guest concentration to achieve simpliﬁed and high-performance hybrid white organic light-emitting diodes. J. Mater. Chem. C 2015, 3, 6359–6366. [CrossRef] 90. Ouyang, X.H.; Li, X.L.; Bai, Y.Q.; Mi, D.B.; Ge, Z.Y.; Su, S.J. Highly-efﬁcient hybrid white organic light-emitting diodes based on a high radiative exciton ratio deepblue emitter with improved concentration of phosphorescent dopant. RSC Adv. 2015, 5, 32298–32306. [CrossRef] 91. Fleetham, T.; Huang, L.; Li, J. Tetradentate Platinum Complexes for Efﬁcient and Stable Excimer-Based White OLEDs. Adv. Funct. Mater. 2015, 24, 6066–6073. [CrossRef] 92. Chen, P.; Xie, W.F.; Li, K.; Guan, T.; Duan, Y.; Zhao, Y.; Liu, S.Y.; Ma, C.S.; Zhang, L.Y.; Li, B. White organic light-emitting devices with a bipolar transport layer between blue ﬂuorescent and orange phosphorescent emitting layers. Appl. Phys. Lett. 2007, 91, 023505. [CrossRef] 93. Liu, B.; Xu, M.; Tao, H.; Su, Y.J.; Gao, D.Y.; Zou, J.H.; Lan, L.F.; Peng, J.B. The effect of spacer in hybri organic light emitting diodes. Chin. Sci. Bull. 2014, 59, 3090–3097. [CrossRef] 94. Zou, J.; Liu, J.; Wu, H.; Yang, W.; Peng, J.; Cao, Y. High-Efﬁciency and Good Color Quality White Light-Emitting Devices Based on Polymer Blend. Org. Electron. 2009, 10, 843. [CrossRef] 95. Chen, D.; Xie, G.; Cai, X.; Liu, M.; Cao, Y.; Su, S.-J. References Lee, T.W.; Noh, T.; Choi, B.K.; Kim, M.S.; Shin, D.W. High-Efﬁciency Stacked White Organic Light-Emitting Diodes. Appl. Phys. Lett. 2008, 92, 43301. [CrossRef] 74. Yu, Y.; Chen, X.; Jin, Y.; Wu, Z.; Yu, Y.; Lin, W. Electron-Transporting Layer Doped With Cesium Azide for High-Performance Phosphorescent and Tandem White Organic Light-Emitting Devices. J. Appl. Phys. D 2017, 50, 275104. [CrossRef] 75. Shi, C.; Sun, N.; Wu, Z.; Chen, J.; Ma, D. High Performance Hybrid Tandem White Organic Light-Emitting Diodes by Using A Novel Intermediate Connector. J. Mater. Chem. C 2018, 6, 767–772. [CrossRef] 76. Kim, H.S.; Joo, C.W.; Pyo, B.; Lee, J.M.; Suh, C. Improvement of Viewing Angle Dependence of the White Organic Light Emitting Diodes With Tandem Structure by Introduction of Nanoporous Ppolymer Films. Org. Electron. 2017, 40, 88–96. [CrossRef] 77. Cho, H.; Song, J.; Kwon, B.H.; Choi, S.; Lee, H.; Joo, C.W.; Ahn, S.D.; Kang, S.Y.; Yoo, S.; Moon, J. Stabilizing Color Shift of Tandem White Organic Light-Emitting Diodes. J. Ind. Eng. Chem. 2018, 69, 414–421. [CrossRef] 78. Chen, Y.; Zhao, F.; Zhao, Y.; Chen, J.; Ma, D. Ultra-simple hybrid white organic light-emitting diodes with high efﬁciency and CRI trade-off: Fabrication and emission-mechanism analysis. Org. Electron. 2012, 13, 2807–2815. [CrossRef] 20 of 28 Molecules 2019, 24, 151 79. Zhao, Y.; Zhu, L.; Chen, J.; Ma, D. Improving Color Stability of Blue/Orange Complementary White OLEDs by Using Single-Host Double-Emissive Layer Structure: Comprehensive Experimental Investigation into the Device Working Mechanism. Org. Electron. 2012, 13, 1340–1348. [CrossRef] 80. Luo, D.X.; Yang, Y.F.; Xiao, Y.; Zhao, Y.; Yang, Y.B.; Liu, B.Q. Regulating Charge and Exciton Distribution in High-Performance Hybrid White Organic Light-Emitting Diodes with n-Type Interlayer Switch. Nano-Micro Lett. 2017, 9, 37–44. [CrossRef] 81. Liu, B.; Xu, M.; Wang, L.; Su, Y.J.; Gao, D.Y.; Tao, H.; Lan, L.F.; Zou, J.H.; Peng, J.B. High-Performance Hybrid White Organic Light-Emitting Diodes Comprising Ultrathin Blue and Orange Emissive Layers. Appl. Phys. Express 2013, 6, 122101. [CrossRef] 82. Poloek, A.; Chen, C.-T.; Chen, C.-T. High performance hybrid white and multi-colour electroluminescence from a new host material for a heteroleptic naphthyridinolate platinum complex dopant. J. Mater. Chem. C 2014, 2, 1376–1380. [CrossRef] 83. Schwartz, G.; Fehse, K.; Pfeiffer, M.; Walzer, K.; Leo, K. Highly efﬁcient white organic light emitting diodes comprising an interlayer to separate ﬂuorescent and phosphorescent regions. Appl. Phys. Lett. 2006, 89, 083509. [CrossRef] 84. Zhou, J.; Zou, J.; Dai, C.; Zhang, Y.; Luo, X.; Liu, B. References Fluorescent Organic Planar pn Heterojunction Light-Emitting Diodes with Simpliﬁed Structure, Extremely Low Driving Voltage, and High Efﬁciency. Adv. Mater. 2016, 28, 239–244. [CrossRef] 96. Chen, D.; Wang, Z.; Wang, D.; Wu, Y.-C.; Lo, C.-C.; Lien, A.; Cao, Y.; Su, S.-J. Efﬁcient Exciplex Organic Light-Emitting Diodes with a Bipolar Acceptor. Org. Electron. 2015, 25, 79–84. [CrossRef] 97. Wu, S.F.; Li, S.H.; Wang, Y.K.; Huang, C.C.; Sun, Q.; Liang, J.J.; Liao, L.S.; Fung, M.-K. White Organic LED with a Luminous Efﬁcacy Exceeding 100 lm w−1 without Light Out-Coupling Enhancement Techniques. Adv. Funct. Mater. 2017, 27, 1701314. [CrossRef] 21 of 28 21 of 28 Molecules 2019, 24, 151 98. Kido, J.; Matsumoto, T.; Nakada, T.; Endo, J.; Mori, K.; Kawamura, N.; Yokoi, A. Dig. 27.1: Invited Paper: High Efﬁciency Organic EL Devices having Charge Generation Layers. J. Tech. Pap. Soc. Inf. Disp. Int. Symp. 2003, 34, 964–965. [CrossRef] 99. Pu, Y.J.; Chiba, T.; Ideta, K.; Takahashi, S.; Aizawa, N.; Hikichi, T.; Kido, J. Fabrication of Organic Light-Emitting Devices Comprising Stacked Light-Emitting Units by Solution-Based Processes. Adv. Mater. 2015, 27, 1327–1332. [CrossRef] 100. Yu, J.; Yin, Y.M.; Liu, W.B.; Zhang, W.; Zhang, L.T.; Xie, W.F.; Zhao, H.Y. Effect of the greenish-yellow emission on the color rendering index of white organic light-emitting devices. Org. Electron. 2014, 15, 2817–2821. [CrossRef] 101. Chen, Y.H.; Ma, D. Organic semiconductor heterojunctions as charge generation layers and their application in tandem organic light-emitting diodes for high power efﬁciency. J. Mater. Chem. 2012, 22, 18718–18734. [CrossRef] 102. Kim, M.; Jeon, S.K.; Hwang, S.H.; Lee, J.Y. Stable Blue Thermally Activated Delayed Fluorescent Organic Light-Emitting Diodes with Three Times Longer Lifetime than Phosphorescent Organic Light-Emitting Diodes. Adv. Mater. 2015, 27, 2515–2520. [CrossRef] [PubMed] 103. Shi, Z.; Li, Y.; Li, S.; Li, X.; Wu, D.; Xu, T.; Tian, Y.; Chen, Y.; Zhang, Y.; Zhang, B.; et al. Localized Surface Plasmon Enhanced All-Inorganic Perovskite Quantum Dot Light-Emitting Diodes Based on Coaxial Core/Shell Heterojunction Architecture. Adv. Funct. Mater. 2018, 28, 1707031. [CrossRef] 104. Xiao, P.; Huang, J.; Yu, Y.; Yuan, J.; Luo, D.; Liu, B.; Liang, D. Recent Advances of Exciplex-Based White Organic Light-Emitting Diodes. Appl. Sci. 2018, 8, 1449. [CrossRef] 105. Liu, B.; Xu, M.; Wang, L.; Yan, X.; Tao, H.; Su, Y.; Gao, D.; Lan, L.; Zou, J.; Peng, J. Investigation and Optimization of Each Organic Layer: A Simple But Effective Approach Towards Achieving High-Efﬁciency Hybrid White Organic Light-Emitting Diodes. Org. Electron. References High Power Efﬁciency Tandem Organic Light-Emitting Diodes Based on Bulk Heterojunction Organic Bipolar Charge Generatio Layer. Appl. Phys. Lett. 2011, 98, 243309. [CrossRef] 118. Liu, J.; Huang, S.; Shi, X.; Wu, X.; Wang, J.; He, G. Charge Separation Process in an Ultrathin Electron-Injecting Bilayer-Assisted Charge Generation Unit for Tandem Organic Light-Emitting Diodes. J. Phys. Chem. C. 2013, 117, 13887–13893. [CrossRef] 119. Liu, B.; Xu, M.; Wang, L.; Tao, H.; Su, Y.; Gao, D.; Zou, J.; Lan, L.; Peng, J. Comprehensive Study on the Electron Transport Layer in Blue Flourescent Organic Light-Emitting Diodes. ECS J. Solid State Sci. Technol. 2015, 2, R258–R261. [CrossRef] 120. Liu, B.; Xu, M.; Wang, L.; Tao, H.; Su, Y.J.; Gao, D.Y.; Lan, L.F.; Zou, J.H.; Peng, J.B. Simpliﬁed hybrid white organic light-emitting diodes with efﬁciency/efﬁciency roll-off/color rendering index/color-stability trade-off. Phys. Status Solidi RRL 2014, 8, 719–723. [CrossRef] 121. Lee, J.; Chopra, N.; Bera, D.; Maslov, S.; Eom, S.H.; Zheng, Y.; Holloway, P.; Xue, J.; So, F. Down-Conversion White Organic Light-Emitting Diodes Using Microcavity Structure. Adv. Eng. Mater. 2011, 1, 174–178. [CrossRef] 122. Du, X.; Tao, S.; Huang, Y.; Yang, X.; Ding, X.; Zhang, X. Efﬁcient Fluorescence/Phosphorescence White Organic Light-Emitting Diodes with Ultra High Color Stability and Mild Efﬁciency Roll-Off. Appl. Phys. Lett. 2015, 107, 183304. [CrossRef] 123. Chen, Y.-H.; Ma, D.-G.; Sun, H.-D.; Chen, J.-S.; Guo, Q.-X.; Wang, Q.; Zhao, Y.-B. Organic Semiconductor Heterojunctions: Electrode-Independent Charge Injectors for High-Performance Organic Light-Emitting Diodes. Light Sci. Appl. 2016, 5, e16042. [CrossRef] [PubMed] 124. Sasabe, H.; Takamatsu, J.; Motoyama, T.; Watanabe, S.; Wagenblast, G.; Langer, N.; Molt, O.; Fuchs, E.; Lennartz, C.; Kido, J. High-Efﬁciency Blue and White Organic Light-Emitting Devices Incorporating a Blue Iridium Carbene Complex. Adv. Mater. 2010, 22, 5003–5007. [CrossRef] [PubMed] 125. Liu, B.; Lan, L.; Zou, J.; Peng, J. A novel organic light-emitting diode by utilizing double hole injectio Acta Phys. Sin. 2013, 62, 087302. 126. Jiang, C.; Liu, H.; Liu, B.; Zhong, Z.; Zou, J.; Wang, J.; Wang, L.; Peng, J.; Cao, Y. Improved Performance of Inverted Quantum Dots Light Emitting Devices by Introducing Double Hole Transport Layers. Org. Electron. 2016, 31, 82–89. [CrossRef] 127. Son, Y.H.; Kim, Y.J.; Park, M.J.; Oh, H.Y.; Park, J.S.; Yang, J.H.; Suh, M.C.; Kwon, J.H. Small single–triplet energy gap bipolar host materials for phosphorescent blue and white organic light emitting diodes. J. Mater. Chem. C 2013, 1, 5008–5014. [CrossRef] 128. References 2014, 15, 926–936. [CrossRef] 106. Li, X.L.; Ouyang, X.H.; Liu, M.; Ge, Z.Y.; Peng, J.B.; Cao, Y.; Su, S.J. Highly efﬁcient single-and multi-emission-layer ﬂuorescent/phosphorescent hybrid white organic light-emitting diodes with ~20% external quantum efﬁciency. J. Mater. Chem. C 2015, 3, 9233–9239. [CrossRef] 107. Kim, D.Y.; Park, J.H.; Lee, J.W.; Hwang, S.; Oh, S.J.; Kim, J.; Sone, C.; Schubert, E.F.; Kim, J.K. Overcoming the fundamental light-extraction efﬁciency limitations of deep ultraviolet light-emitting diodes by utilizing transverse-magnetic-dominant emission. Light Sci. Appl. 2015, 4, e263. [CrossRef] 108. Xiao, P.; Huang, J.; Dong, T.; Xie, J.; Yuan, J.; Luo, D.; Liu, B. Room-temperature Fabricated Thin-Film Transistors Based on Compounds with Lanthanum and Main Family Element Boron. Molecules 2018, 23, 1373. [CrossRef] 109. Liu, B.; Xu, Z.P.; Zou, J.H.; Tao, H.; Xu, M.; Gao, D.Y.; Lan, L.F.; Wang, L.; Ning, H.L.; Peng, J.B. High-performance hybrid white organic light-emitting diodes employing p-type interlayers. J. Ind. Eng. Chem. 2015, 27, 240–244. [CrossRef] 110. Liu, B.; Tao, H.; Su, Y.J.; Gao, D.Y.; Lan, L.F.; Zou, J.H.; Peng, J.B. Color-stable, reduced efﬁciency roll-off hybrid white organic light emitting diodes with ultra high brightness. Chin. Phys. B 2013, 22, 077303. [CrossRef] 111. Sun, H.; Chen, Y.H.; Chen, J.S.; Ma, D. Interconnectors in Tandem Organic Light Emitting Diodes and Their Inﬂuence on Device Performance. IEEE J. Sel. Top. Quantum Electron. 2015, 22, 154–163. [CrossRef] 112. Liao, L.S.; Slusarek, W.K.; Hatwar, T.K.; Ricks, M.L.; Comfort, D.L. Tandem Organic Light-Emitting Diode Using Hexaazatriphenylene Hexacarbonitrile in the Intermediate Connector. Adv. Mater. 2008, 20, 324–329. [CrossRef] 113. Tang, J.X.; Fung, M.K.; Lee, C.S.; Lee, S.T. Interface Studies of Intermediate Connectors and Their Roles in Tandem OLEDs. J. Mater. Chem. 2010, 20, 2539–2548. [CrossRef] 114. Leem, D.S.; Lee, J.H.; Kim, J.J.; Kang, J.W. Highly Efﬁcient Tandem p-i-n Organic Light-Emitting Diodes Adopting a Low Temperature Evaporated Rhenium Oxide Interconnecting Layer. Appl. Phys. Lett. 2008, 93, 103304. [CrossRef] 115. Lai, S.L.; Chan, M.Y.; Fung, M.K.; Lee, C.S. Copper Hexadeca-fluorophthalocyanine and Copper Phthalocyanine as a Pure Organic Connecting Unit in Blue Tandem Organic Lght-Emitting Devices. J. Appl. Phys. 2007, 101, 14509. [CrossRef] 22 of 28 Molecules 2019, 24, 151 116. Chen, Y.H.; Chen, J.S.; Ma, D.G.; Yan, D.H.; Wang, L.X.; Chen, Y. Effect of Organic Bulk Heterojunction as Charge Generation Layer on The Performance of Tandem Organic Light-Emitting Diodes. J. Appl. Phys. 2011, 110, 074504. [CrossRef] 117. Chen, Y.H.; Chen, J.S.; Ma, D.G.; Yan, D.H.; Wang, L.X.; Zhu, F.R. References Schwartz, G.; Pfeiffer, M.; Reineke, S.; Walzer, K.; Leo, K. Harvesting Triplet Excitons from Fluorescent Blue Emitters in White Organic Light-Emitting Diodes. Adv. Mater. 2007, 19, 3672–3676. [CrossRef] 139. Ye, J.; Zheng, C.-J.; Ou, X.-M.; Zhang, X.-H.; Fung, M.-K.; Lee, C.-S. Management of Singlet and Triplet Excitons in a Single Emission Layer: A Simple Approach for a High-Efﬁciency Fluorescence/Phosphorescence Hybrid White Organic Light-Emitting Device. Adv. Mater. 2012, 24, 3410–3414. [CrossRef] 140. Liu, B.; Delikanli, S.; Gao, Y.; Gungor, K.; Demir, H.V. Nanocrystal light-emitting diodes based on type II nanoplatelets. Nano Energy 2018, 47, 115–122. [CrossRef] 141. Chapran, M.; Angioni, E.; Findlay, N.J.; Breig, B.; Cherpak, V.; Stakhira, P.; Tuttle, T.; Volyniuk, D.; Grazulevicius, J.V.; Nastishin, Y.A.; et al. An ambipolar BODIPY derivative for a white exciplex OLED and cholesteric liquid crystal laser towards multi-functional devices. ACS Appl. Mater. Interfaces 2017, 9, 4750–4757. [CrossRef] [PubMed] 142. Cekaviciute, M.; Simokaitiene, J.; Volyniuk, D.; Sini, G.; Grazulevicius, J.V. Arylﬂuorenyl-substituted metoxytriphenylamines as deep blue exciplex forming bipolar semiconductors for white and blue organic light emitting diodes. Dyes Pigments 2017, 140, 187–202. [CrossRef] 143. Yokoyama, M.; Su, S.H.; Hou, C.C.; Wu, C.T.; Kung, C.H. Highly Efﬁcient White Organic Light-Emitting Diodes with a p-i-n Tandem Structure. Jpn. J. Appl. Phys. 2011, 50, 04DK06. [CrossRef] 144. Hoa, M.H.; Chen, T.M. Highly Efﬁcient p-i-n White Organic Light Emitting Devices with Tandem Structure. Appl. Phys. Lett. 2007, 91, 233507. [CrossRef] 145. Berggren, M.; Gustafsson, G.; Inganäs, O.; Andersson, M.R.; Hjertberg, T.; Wennerström, O. White Light from an Electroluminescent Diode Made from Poly[3(4-Octylphenyl)-2,2′-Bithiophene] and an Oxadiazole Derivative. J. Appl. Phys. 1994, 76, 7530–7534. [CrossRef] 146. Feng, J.; Li, F.; Gao, W.; Liu, S.; Wang, Y. White Light Emission from Exciplex Using Tris-(8-Hydroxyquinoline)Aluminum as Chromaticity-Tuning Layer. Appl. Phys. Lett. 2001, 78, 3947–3949. [CrossRef] 147. Gebler, D.D.; Wang, Y.Z.; Blatchford, J.W.; Jessen, S.W. Exciplex Emission in Bilayer Polymer Light-Emitting Devices. Appl. Phys. Lett. 1997, 70, 1644–1646. [CrossRef] 148. Chen, S.; Zhao, X.; Wu, Q.; Shi, H.; Mei, Y.; Zhang, R.; Wang, L.; Huang, W. Efﬁcient, Color-Stable Flexible White Top-Emitting Organic Light-Emitting Diodes. Org. Electron. 2013, 14, 3037–3045. [CrossRef] 149. Yook, K.S.; Jeon, S.O.; Joo, C.W.; Lee, J.Y. Color stability and suppressed efﬁciency roll-off in white organic light-emitting diodes through management of interlayer and host properties. J. Ind. Eng. Chem. 2009, 15, 420–422. [CrossRef] 150. Su, S.-J.; Gonmori, E.; Sasabe, H.; Kido, J. References Zhao, B.; Zhang, T.Y.; Li, W.L.; Su, Z.S.; Chu, B.; Yan, X.W.; Jin, F.M.; Gao, Y.; Wu, H. Organic Electronics Highly efﬁcient and color stable single-emitting-layer ﬂuorescent WOLEDs with delayed ﬂuorescent host. Org. Electron. 2015, 23, 208–212. [CrossRef] 129. Liu, B.; Wang, L.; Xu, M.; Tao, H.; Xia, X.; Zou, J.; Su, Y.; Gao, D.; Lan, L.; Peng, J. Simultaneous Achievement of Low Efﬁciency Roll-Off and Stable Color in Highly Efﬁcient Single-Emitting-Layer Phosphorescent White Organic Light-Emitting Diodes. J. Mater. Chem. C 2014, 2, 5870–5877. [CrossRef] 130. Liu, B.; Wang, L.; Zou, J.H.; Tao, H.; Su, Y.J.; Gao, D.Y.; Xu, M.; Lan, L.F.; Peng, J.B. Investigation on spacers and structures: A simple but effective approach toward high-performance hybrid white organic light emitting diodes. Synth. Met. 2013, 184, 5–9. [CrossRef] g g y 131. Yin, Y.M.; Yu, J.; Cao, H.T.; Zhang, L.T.; Sun, H.Z.; Xie, W.F. Efﬁcient non-doped phosphorescent orange, blue and white organic light-emitting devices. Sci. Rep. 2014, 4, 6754. [CrossRef] [PubMed] 132. Wang, L.; Liu, B.; Zhao, X.; Demir, H.V.; Gu, H.; Sun, H. Solvent-Assisted Surface Engineering for High Performance All-Inorganic Perovskite Nanocrystals Light-Emitting Diodes. ACS Appl. Mater. Interfaces 2018, 10, 19828–19835. [CrossRef] [PubMed] 133. Liu, B.; Xu, M.; Wang, L.; Zou, J.H.; Tao, H.; Su, Y.J.; Gao, D.Y.; Ning, H.L.; Lan, L.F.; Peng, J.B. Regulating charges and excitons in simpliﬁed hybrid white organic light-emitting diodes: The key role of concentration in single dopant host-guest systems. Org. Electron. 2014, 15, 2616–2623. [CrossRef] 23 of 28 Molecules 2019, 24, 151 134. Liu, B.; Zou, J.H.; Su, Y.J.; Gao, D.Y.; Lan, L.F.; Tao, H.; Peng, J.B. Hybrid white organic light emitting diodes with low efﬁciency roll-off, stable color and extreme brightness. J. Lumin. 2014, 151, 161–164. [CrossRef] 135. Chen, B.; Liu, B.; Zeng, J.; Nie, H.; Xiong, Y.; Zou, J.; Ning, H.; Wang, Z.; Zhao, Z.; Tang, B.Z. Efﬁcient Bipolar Blue AIEgens for High-Performance Nondoped Blue OLEDs and Hybrid White OLEDs. Adv. Func. Mater. 2018, 28, 1803369. [CrossRef] 136. Zhang, M.; Chen, Z.; Xiao, L.; Qu, B.; Gong, Q. High-Color-Quality White Top-Emitting Organic Electroluminescent Devices Based on Both Exciton and Electroplex Emission. Appl. Phys. Express 2013, 4, 784–788. [CrossRef] 137. Luo, D.X.; Xiao, Y.; Hao, M.M.; Zhao, Y.; Yang, Y.B.; Gao, Y.; Liu, B. Doping-free white organic light-emitting diodes without blue molecular emitter: An unexplored approach to achieve high performance via exciplex emission. Appl. Phys. Lett. 2017, 110, 061105. [CrossRef] 138. References Highly efﬁcient organic blue- and white-light-emitting devices having a carrier- and exciton-conﬁning structure for reduced efﬁciency roll-off. Adv. Mater. 2008, 20, 4189–4194. [CrossRef] 151. Reineke, S.; Lindner, F.; Schwartz, G.; Seidler, N.; Walzer, K.; Lüssem, B.; Leo, K. White organic light-emitting diodes with ﬂuorescent tube efﬁciency. Nature 2009, 459, 234–238. [CrossRef] [PubMed] 152. Zhu, L.; Wu, Z.; Chen, J.; Ma, D. Reduced Efﬁciency Roll-Off in All-Phosphores-cent White Organic Light-Emitting Diodes with an External Quantum Efﬁciency of Over 20%. J. Mater. Chem. C 2015, 3, 3304–3310. [CrossRef] 153. Fleetham, T.; Ecton, J.; Wang, Z.; Bakken, N. Single-Doped White Organic Light-Emitting Device with an External Quantum Efﬁciency Over 20%. Adv. Mater. 2013, 25, 2573–2576. [CrossRef] [PubMed] 24 of 28 Molecules 2019, 24, 151 24 of 28 154. Xu, L.; Tang, C.W.; Rothberg, L.J. High Efﬁciency Phosphorescent White Organic Light-Emitting Diodes with An Ultra-Thin Red and Green Co-Doped Layer and Dual Blue Emitting Layers. Org. Electron. 2016, 32, 54–58. [CrossRef] 155. Kanno, H.; Holmes, R.J.; Sun, Y.; Kena, S.; Forrest, S.R. White Stacked Electrophosphorescent Organic Light-Emitting Devices Employing MoO3 as a Charge-Generation Layer. Adv. Mater. 2006, 18, 339–342. [CrossRef] 156. Wu, H.; Zou, J.; Liu, F.; Wang, L.; Mikhailovsky, A.; Bazan, G.C.; Yang, W.; Cao, Y. Efﬁcient Single Active Layer Electrophosphorescent White Polymer Light-Emitting Diodes. Adv. Mater. 2008, 20, 696. [CrossRef] 156. Wu, H.; Zou, J.; Liu, F.; Wang, L.; Mikhailovsky, A.; Bazan, G.C.; Yang, W.; Cao, Y. Efﬁcient Single Active Layer Electrophosphorescent White Polymer Light-Emitting Diodes. Adv. Mater. 2008, 20, 696. [CrossRef] 157. Wei, P.C.; Zhang, D.D.; Cai, M.H.; Song, X.Z.; Wang, Z.Y.; Duan, L. Simpliﬁed Single-Emitting-Layer Hybrid 157. Wei, P.C.; Zhang, D.D.; Cai, M.H.; Song, X.Z.; Wang, Z.Y.; Duan, L. Simpliﬁed Single-Emitting-Layer Hybrid White Organic Light-Emitting Diodes with High Efﬁciency, Low Efﬁciency Roll-Off, High Color Rendering Index and Superior Color. Org. Electron. 2017, 49, 242–248. [CrossRef] 158. Pereira, D.; Pinto, A.; Califórnia, A.; Gomes, J. Control of A White Organic Light Emitting Diode Emission Parameters Using A Single Doped RGB Active Layer. Mater. Sci. Eng. B 2016, 211, 156–165. [CrossRef] 159. Sun, N.; Wang, Q.; Zhao, Y.B.; Chen, Y.H.; Yang, D.Z.; Zhao, F.C.; Chen, J.S.; Ma, D.G. High-Performance Hybrid White Organic Light-Emitting Devices without Interlayer between Fluorescent and Phosphorescent Emissive Regions. Adv. Mater. 2014, 26, 1617–1621. [CrossRef] 160. Ban, X.; Sun, K.; Sun, Y.; Huang, B.; Jiang, W. References Design of High Triplet Energy Electron Transporting Material for Exciplex-Type Host: Efﬁcient Blue and White Phosphorescent OLEDs based on Solution Processing. Org. Electron. 2016, 33, 9–14. [CrossRef] 161. Wang, Q.; Ding, J.; Ma, D.; Cheng, Y.; Wang, L.; Jing, X.; Wang, F. Harvesting Excitons Via Two Parallel Channels for Efﬁcient White Organic LEDs with Nearly 100% Internal Quantum Efﬁciency: Fabrication and Emission Mechanism Analysis. Adv. Funct. Mater. 2009, 19, 84–95. [CrossRef] 162. Wang, Q.; Ding, J.Q.; Zhang, Z.Q.; Ma, D.G.; Cheng, Y.X.; Wang, L.X.; Wang, F.S. A high-performance tandem white organic light-emitting diode combining highly effective white-units and their interconnection layer. J. Appl. Phys. 2009, 105, 076101. [CrossRef] 163. Lee, S.; Shin, H.; Kim, J.J. High-Efﬁciency Orange and Tandem White Organic Light-Emitting Diodes Using Phosphorescent Dyes with Horizontally Oriented Emitting Dipoles. Adv. Mater. 2014, 26, 5864–5868. [CrossRef] [PubMed] 164. Lee, S.; Kim, K.H.; Limbach, D.; Park, Y.S.; Kim, J.J. Organic Light Emitting Diodes: Low Roll-Off and High Efﬁciency Orange Organic Light Emitting Diodes with Controlled Co-Doping of Green and Red Phosphorescent Dopants in an Exciplex Forming Co-Host. Adv. Funct. Mater. 2013, 23, 4061–4062. [CrossRef] 165. Lee, S.; Limbach, D.; Kim, K.H.; Yoo, S.J.; Park, Y.S.; Kim, J.J. High Efﬁciency and Non-Color-Changing Orange Organic Light Emitting Diodes with Red and Green Emitting Layers. Org. Electron. 2013, 14, 1856–1860. [CrossRef] 166. Park, Y.S.; Lee, S.; Kim, K.H.; Kim, S.Y.; Lee, J.H.; Kim, J.J. Exciplex-Forming Co-host for Organic Light-Emitting Diodes with Ultimate Efﬁciency. Adv. Funct. Mater. 2013, 23, 4914–4920. [CrossRef] 167. Chao, C.I.; Chen, S.A. White Light Emission from Exciplex in a Bilayer Device with Two Blue Light-Emitting Polymers. Appl. Phys. Lett. 1998, 73, 426–428. [CrossRef] 168. Cherpak, V.; Stakhira, P.; Minaev, B.; Baryshnikov, G.; Stromylo, E.; Helzhynskyy, I.; Chapran, M.; Volyniuk, D.; Hotra, Z.; Dabuliene, A.; et al. Mixing of Phosphorescent and Exciplex Emission in Efﬁcient Organic Electroluminescent Devices. ACS Appl. Mater. Interfaces 2015, 7, 1219–1225. [CrossRef] g 169. Zhao, Y.B.; Chen, J.S.; Ma, D.G. Ultrathin Nondoped Emissive Layers for Efﬁcient and Simple Monochrome and White Organic Light-Emitting Diodes. ACS Appl. Mater. Interfaces 2013, 5, 965–971. [CrossRef] 170. Xue, K.W.; Han, G.G.; Duan, Y.; Chen, P.; Yang, Y.Q.; Yang, D.; Duan, Y.H.; Wang, X.; Zhao, Y. Doping-free orange and white phosphorescent organic light-emitting diodes with ultra-simply structure and excellent color stability. Org. Electron. 2015, 18, 84–88. [CrossRef] 171. References Xue, K.W.; Sheng, R.; Duan, Y.; Chen, P.; Chen, B.Y.; Wang, X.; Duan, Y.H.; Zhao, Y. Efﬁcient non-doped monochrome and white phosphorescent organic light-emitting diodes based on ultrathin emissive layers. Org. Electron. 2015, 26, 451–457. [CrossRef] 172. Fleetham, T.; Li, G.; Li, J. Phosphorescent Pt (II) and Pd (II) Complexes for Efﬁcient, High-Color-Quality, and Stable OLEDs. Adv. Mater. 2017, 29, 1601861. [CrossRef] [PubMed] 173. Fleetham, T.; Li, G.; Wen, L.; Li, J. Efﬁcient “Pure” Blue OLEDs Employing Tetradentate Pt Complexes with a Narrow Spectral Bandwidth. Adv. Mater. 2014, 26, 7116–7121. [CrossRef] [PubMed] 25 of 28 Molecules 2019, 24, 151 174. Coburn, C.; Jeong, C.; Forrest, S.R. Reliable, All-Phosphorescent Stacked White Organic Light Emitting Devices with a High Color Rendering Index. ACS Photon. 2018, 5, 630–635. [CrossRef] 175. Rajamalli, P.; Senthilkumar, N.; Gandeepan, P.; Huang, P.-Y.; Huang, M.-J.; Yang, C.-Y.; Chiu, M.-J.; Chu, L.-K.; Lin, H.-W.; Cheng, C.-H. A New Molecular Design Based on Thermally Activated Delayed Fluorescence for Highly Efﬁcient Organic Light Emitting Diodes. J. Am. Chem. Soc. 2016, 138, 628–634. [CrossRef] [PubMed] 176. Yang, Z.Y.; Mao, Z.; Xie, Z.L.; Zhang, Y.; Liu, S.W.; Zhao, J.; Xu, J.R.; Chi, Z.G.; Aldred, M.P. Recent advances in organic thermally activated delayed ﬂuorescence materials. Chem. Soc. Rev. 2017, 46, 915–1016. [CrossRef] [PubMed] 177. Zhang, D.D.; Cai, M.H.; Zhang, Y.G.; Zhang, D.Q.; Duan, L. Sterically Shielded Blue Thermally Activated Delayed Fluorescence Emitters with Improved Efﬁciency and Stability. Mater. Horiz. 2016, 3, 145–151. [CrossRef] 178. Tao, Y.; Yuan, K.; Chen, T.; Xu, P.; Li, H.; Chen, R.; Zheng, C.; Zhang, L.; Huang, W. Thermally Activated Delayed Fluorescence Materials Towards the Breakthrough of Organoelectronics. Adv. Mater. 2014, 26, 7931–7958. [CrossRef] 179. Masui, K.; Nakanotani, H.; Adachi, C. Analysis of exciton annihilation in high efﬁciency sky-blue organic light-emitting diodes with thermally activated delayed ﬂuorescence. Org. Electron. 2013, 14, 2721–2726. [CrossRef] 180. Zhang, D.; Duan, L.; Li, C.; Li, Y.; Li, H.; Zhang, D.; Qiu, Y. High-Efﬁciency Fluorescent Organic Light-Emitting Devices Using Sensitizing Hosts with a Small Singlet-Triplet Exchange Energy. Adv. Mater. 2014, 26, 5050–5055. [CrossRef] 181. Guo, J.J.; Li, X.L.; Nie, H.; Luo, W.W.; Gan, S.F.; Hu, S.M.; Hu, R.R.; Qin, A.J.; Zhao, Z.J.; Su, S.J.; et al. Achieving High-Performance Nondoped OLEDs with Extremely Small Efﬁciency Roll-Off by Combining Aggregation-Induced Emission and Thermally Activated Delayed Fluorescence. Adv. Funct. Mater. 2017, 27, 1606458. [CrossRef] 182. Yong, J.C.; Yook, K.S.; Lee, J.Y. References Cool and Warm Hybrid White Organic Light-Emitting Diode with Blue Delayed Fluorescent Emitter Both as Blue Emitter and Triplet Host. Sci. Rep. 2015, 5, 7859. 183. Wu, Z.; Qi, W.; Ling, Y.; Chen, J.; Qiao, X.; Ahamad, T.; Alshehri, S.M.; Yang, C.; Ma, D. Managing excitons and charges for high-performance ﬂuorescent white organic light-emitting diodes. ACS Appl. Mater. Interfaces 2016, 8, 28780–28788. [CrossRef] [PubMed] 184. Hou, L.; Duan, L.; Qiao, J.; Zhang, D.; Dong, G.; Wang, L.; Qiu, Y. Efﬁcient Solution-Processed Small-Molecule Single Emitting Layer Electrophosphorescent White Light-Emitting Diodes. Org. Electron. 2010, 11, 1344–1350. [CrossRef] 185. Wang, Q.; Oswald, I.W.H.; Perez, M.R.; Jia, H.P.; Shahub, A.A.; Qiao, Q.Q.; Gnade, B.E.; Omary, M.A. Doping-Free Organic Light-Emitting Diodes with Very High Power Efﬁciency, Simple Device Structure, and Superior Spectral Performance. Adv. Funct. Mater. 2014, 24, 4746–4752. [CrossRef] 186. Wang, J.; Chen, J.; Qiao, X.; Alshehri, S.M.; Ahamad, T.; Ma, F. Simple-Structured Phosphorescent Warm White Organic Light-Emitting Diodes with High Power Efﬁciency and Low Efﬁciency Roll-off. ACS Appl. Mater. Interfaces 2016, 8, 10093–10097. [CrossRef] [PubMed] 187. Nishide, J.-I.; Nakanotani, H.; Hiraga, Y.; Adachi, C. High-efﬁciency white organic light-emitting diodes using thermally activated delayed ﬂuorescence. Appl. Phys. Lett. 2014, 104, 233304. [CrossRef] 188. Wu, Z.; Luo, J.J.; Sun, N.; Zhu, L.P.; Sun, H.D.; Yu, L.; Yang, D.Z.; Qiao, X.F.; Chen, J.S.; Yang, C.L.; et al. HighPerformance Hybrid White Organic Light-Emitting Diodes with Superior Efﬁciency/Color Rendering Index/Color Stability and Low Efﬁciency Roll-Off Based on a Blue Thermally Activated Delayed Fluorescent Emitter. Adv. Funct. Mater. 2016, 26, 3306–3313. [CrossRef] 189. Luo, D.; Chen, Q.; Gao, Y.; Zhang, M.; Liu, B. Extremely Simpliﬁed, High-Performance and Doping-Free White Organic Light-Emitting Diodes Based on Single Thermally Activated Delayed Fluorescent Emitter. ACS Energy Lett. 2018, 3, 1531–1538. [CrossRef] 190. Kim, B.S.; Yook, K.S.; Lee, J.Y. Above 20% External Quantum Efﬁciency in Novel Hybrid White Organic Light-Emitting Diodes Having Green Thermally Activated Delayed Fluorescent Emitter. Sci. Rep. 2014, 4, 6019. [CrossRef] 191. Goushi, K.; Yoshida, K.; Sato, K.; Adachi, C. Organic light-emitting diodes employing efﬁcient reverse intersystem crossing for triplet-to-singlet state conversion. Nat. Photon. 2012, 6, 253–258. [CrossRef] Molecules 2019, 24, 151 26 of 28 26 of 28 192. Goushi, K.; Adachi, C. Efﬁcient Organic Light-Emitting Diodes through Up-Conversion from Triplet to Singlet Excited States of Exciplexes. Appl. Phys. Lett. 2012, 101, 3174–3187. [CrossRef] 193. References Chen, D.; Liu, K.; Gan, L.; Liu, M.; Gao, K.; Xie, G.; Ma, Y.; Cao, Y.; Su, S.-J. Modulation of Exciton Generation in Organic Active Planar pn Heterojunction: Toward Low Driving Voltage and High-Efﬁciency OLEDs Employing Conventional and Thermally Activated Delayed Fluorescent Emitters. Adv. Mater. 2016, 28, 6758–6765. [CrossRef] [PubMed] 194. Liu, X.; Chen, Z.; Zheng, C.; Liu, C.; Lee, C.; Li, F.; Ou, X.-M.; Zhang, X.-H. Prediction and Design of Efﬁcient Exciplex Emitters for High-Efﬁciency, Thermally Activated Delayed-Fluorescence Organic Light-Emitting Diodes. Adv. Mater. 2015, 27, 2378–2383. [CrossRef] [PubMed] 195. Liu, X.; Chen, Z.; Zheng, C.-J.; Chen, M.; Liu, W.; Zhang, X.-H.; Lee, C.-S. Nearly 100% Triplet Harvesting in Conventional Fluorescent Dopant-Based Organic Light-Emitting Devices through Energy Transfer from Exciplex. Adv. Mater. 2015, 27, 2025–2030. [CrossRef] [PubMed] 196. Zhang, T.; Chu, B.; Li, W.; Su, Z.; Peng, Q.M.; Zhao, B.; Luo, Y.; Jin, F.; Yan, X.; Gao, Y.; et al. Efﬁcient Triplet Application in Exciplex Delayed-Fluorescence OLEDs Using a Reverse Intersystem Crossing Mechanism Based on A ∆ES-T of Around Zero. ACS Appl. Mater. Interfaces 2014, 6, 11907. [CrossRef] [PubMed] pp f 197. Hung, W.Y.; Fang, G.C.; Lin, S.W.; Cheng, S.H.; Wong, K.T.; Kuo, T.Y.; Chou, P.T. The First Tandem, All-exciplex-based WOLED. Sci. Rep. 2014, 4, 5161. [CrossRef] 198. Zhao, B.; Zhang, T.; Chu, B.; Li, W.; Su, Z.; Luo, Y.; Li, R.; Yan, X.; Jin, F.; Gao, Y.; et al. Highly efﬁcient tandem full exciplex orange and warm white OLEDs based on thermally activated delayed ﬂuorescence mechanism. Org. Electron. 2015, 17, 15–21. [CrossRef] 199. Li, F.; Cheng, G.; Zhao, Y.; Feng, J.; Liu, S.Y.; Zhang, M.; Ma, Y.G.; Shen, J.C. White-electrophosphorescence devices based on rhenium complexes. Appl. Phys. Lett. 2003, 83, 4716. [CrossRef] 200. Sun, Y.R.; Giebink, N.C.; Kanno, H.; Ma, B.W.; Thompson, M.E.; Forrest, S.R. Management of Singlet and Triplet Excitons for Efﬁcient White Organic Light-Emitting Devices. Nature 2006, 440, 908–912. [CrossRef] 201. Chen, Z.; Liu, X.K.; Zheng, C.J.; Ye, J.; Liu, C.L.; Li, F.; Ou, X.M.; Lee, C.S.; Zhang, X.H. High Performance Exciplex-Based Fluorescence-Phosphorescence White Organic Light-Emitting Device with Highly Simpliﬁed Structure. Chem. Mater. 2015, 27, 5206–5211. [CrossRef] 202. Li, X.L.; Ouyang, X.H.; Chen, D.C.; Cai, X.Y.; Liu, M.; Ge, Z.Y.; Cao, Y.; Su, S.J. Highly efﬁcient blue and warm white organic light-emitting diodes with a simpliﬁed structure. Nanotechnology 2016, 27, 124001. [CrossRef] [PubMed] 203. References Liu, B.; Wang, L.; Gao, D.Y.; Zou, J.H.; Ning, H.L.; Peng, J.B.; Cao, Y. Extremely high-efﬁciency and ultrasimpliﬁed hybrid white organic light-emitting diodes exploiting double multifunctional blue emitting layers. Light Sci. Appl. 2016, 5, e16137. [CrossRef] [PubMed] 204. Zheng, C.-J.; Wang, J.; Ye, J.; Lo, M.-F.; Liu, X.-K.; Fung, M.-K.; Zhang, X.-H.; Lee, C.-S. Novel Efﬁcient Blue Fluorophors with Small Singlet-Triplet Splitting: Hosts for Highly Efﬁcient Fluorescence and Phosphorescence Hybrid WOLEDs with Simpliﬁed Structure. Adv. Mater. 2013, 25, 2205–2211. [CrossRef] [PubMed] 205. Schwartz, G.; Reineke, S.; Walzer, K.; Leo, K. Reduced efﬁciency roll-off in high-efﬁciency hybrid white organic light-emitting diodes. Appl. Phys. Lett. 2008, 92, 053311. [CrossRef] 206. Kanno, H.; Giebink, N.C.; Sun, Y.; Forrest, S.R. Stacked white organic light-emitting devices based on a combination of ﬂuorescent and phosphorescent emitters. Appl. Phys. Lett. 2006, 89, 023503. [CrossRef] 207. Schwartz, G.; Ke, T.-H.; Wu, C.-C.; Walzer, K.; Leo, K. Balanced ambipolar charge carrier mobility in mixed layers for application in hybrid white organic light-emitting diodes. Appl. Phys. Lett. 2008, 93, 073304. [CrossRef] 208. Zhang, T.; Zhao, B.; Chu, B.; Li, W.; Su, Z.; Yan, X.; Liu, C.; Wu, H.; Gao, Y.; Jin, F.; et al. Simple structured hybrid WOLEDs based on incomplete energy transfer mechanism: From blue exciplex to orange dopant. Sci. Rep. 2015, 5, 10234. [CrossRef] 209. Luo, D.; Li, X.-L.; Zhao, Y.; Gao, Y.; Liu, B. High-Performance Blue Molecular Emitter-Free and Doping-Free Hybrid White Organic Light-Emitting Diodes: An Alternative Concept to Manipulate Charges and Excitons Based on Exciplex and Electroplex Emission. ACS Photon. 2017, 4, 1566–1575. [CrossRef] 210. Rosenow, T.C.; Furno, M.; Reineke, S.; Olthof, S.; Lussem, B.; Leo, K. Highly efﬁcient white organic light-emitting diodes based on ﬂuorescent blue emitters. J. Appl. Phys. 2010, 108, 113113. [CrossRef] 27 of 28 Molecules 2019, 24, 151 27 of 28 211. Schwab, T.; Schubert, S.; Hofmann, S.; Frobel, M.; Fuchs, C.; Thomschke, M.; Muller-Meskamp, L.; Leo, K.; Gather, M.C. Highly Efﬁcient and Color Stable Inverted White Top-emitting OLEDs with Ultra-thin Wetting Layer Top Electrodes. Adv. Opt. Mater. 2013, 1, 707–713. [CrossRef] 212. Gaynor, W.; Hofmann, S.; Christoforo, M.G.; Sachse, C.; Mehra, S.; Salleo, A.; McGehee, M.D.; Gather, M.C.; Lussem, B.; Muller-Meskamp, L. Color in the Corners: ITO-Free White OLEDs with Angular Color Stability. Adv. Mater. 2013, 25, 4006–4013. [CrossRef] [PubMed] 213. Thomschke, M.; Reineke, S.; Lussem, B.; Leo, K. Highly efﬁcient white top-emitting organic light-emitting diodes comprising laminated microlens ﬁlms. References Nano Lett. 2012, 12, 424–428. [CrossRef] [PubMed] 214. Zhang, M.; Wang, K.; Zheng, C.-J.; Liu, W.; Lin, H.; Tao, S.-L.; Zhang, X.-H. Efﬁcient, color-stable and high color-rendering-index white organic light-emitting diodes employing full thermally activated delayed ﬂuorescence system. Org. Electron. 2017, 50, 466–472. [CrossRef] 215. Duan, L.; Tsuboi, T.; Qiu, Y.; Li, Y.R.; Zhang, G.H. Tandem organic light-emitting diodes with KBH4 doped 9,10-bis(3-(pyridin-3-yl)phenyl) anthracene connected to the charge generation layer. Opt. Express 2012, 20, 14564. [CrossRef] [PubMed] 216. Huang, C.; Xie, Y.; Wu, S.; Li, S.H.; Liang, J.J.; Fung, M.K. Thermally activated delayed ﬂuorescence-based tandem OLEDs with very high external quantum efﬁciency. Phys. Status Solidi A 2017, 214, 1700240. [CrossRef] 217. Chen, S.; Qu, Q.; Kong, M.; Zhao, X.; Yu, Z.; Jia, P.; Huang, W. On the Origin of the Shift in Color in White Organic Light-Emitting Diodes. J. Mater. Chem. C 2013, 1, 3508. [CrossRef] 218. Zhao, B.; Zhang, H.; Miao, Y.; Wang, Z.; Gao, L.; Wang, H.; Hao, Y.; Li, W. High Color Stability and CRI (>80) Fluorescent White Organic Light-Emitting Diode based Pure Emission of Exciplexes by Employing Merely Complementary colors. J. Mater. Chem. C 2018, 6, 304–311. [CrossRef] 219. Fröbel, M.; Schwab, T.; Kliem, M.; Hofmann, S.; Leo, K.; Gather, M.C. Get It White: Color-Tunable Ac/Dc OLEDs. Light Sci. Appl. 2015, 4, e247. [CrossRef] 220. Wang, D.; Li, W.L.; Su, Z.S.; Li, T.L.; Chu, B.; Bi, D.F.; Chen, L.L.; Su, W.M.; He, H. Broad Wavelength Modulating and Design of Organic White Diode Based on Lighting by Using Exciplex Emission from Mixed Acceptors. Appl. Phys. Lett. 2006, 89, 53. [CrossRef] 221. Gong, S.; Chen, Y.; Luo, J.; Yang, C.; Zhong, C.; Qin, J.; Ma, D. Bipolar Tetraarylsilanes as Universal Hosts for Blue, Green, Orange, and White Electrophosphorescence with High Efﬁciency and Low Efﬁciency Roll-Off. Adv. Funct. Mater. 2011, 21, 1168–1178. [CrossRef] 222. Liu, B.; Tao, H.; Wang, L.; Gao, D.Y.; Liu, W.C.; Zou, J.H.; Xu, M.; Ning, H.L.; Peng, J.B.; Cao, Y. High-performance doping-free hybrid white organic light-emitting diodes: The exploitation of ultrathin emitting nanolayers (<1 nm). Nano Energy 2016, 26, 26–36. 223. Luo, D.; Xiao, P.; Liu, B. Doping-Free White Organic Light-Emitting Diodes. Chem. Rec. 2018. [CrossRef] 224. Liu, B.; Wang, L.; Tao, H.; Xu, M.; Zou, J.; Ning, H.; Peng, J.; Cao, Y. Doping-Free Tandem White Organic Light-Emitting Diodes. Sci. Bull. 2017, 62, 1193–1200. [CrossRef] 25. Yoshihiro, O. References Color rendering and luminous efﬁcacy of white LED spectra. Proc. SPIE 2004, 88, 5530. 226. Adachi, C.; Baldo, M.A.; Thompson, M.E.; Forrest, S.R. Nearly 100% Internal Phosphorescence Efﬁciency in an Organic Light-Emitting Device. J. Appl. Phys. 2001, 90, 5048–5051. [CrossRef] g g g pp y 227. Nozoe, S.; Matsuda, M. Enhanced emission by accumulated charges at organic/metal interfaces generated during the reverse bias of organic light emitting diodes. Appl. Sci. 2017, 7, 1045. [CrossRef] 228. Brovelli, S.; Sforazzini, G.; Serri, M.; Winroth, G.; Suzuk, K.; Meinardi, F.; Anderson, H.L.; Cacialli, F. Emission Color Trajectory and White Electroluminescence Through Supramolecular Control of Energy Transfer and Exciplex Formation in Binary Blends of Conjugated Polyrotaxanes. Adv. Funct. Mater. 2012, 22, 4284–4291. [CrossRef] 229. Preinfalk, J.B.; Eiselt, T.; Wehlus, T.; Rohnacher, V.; Hanemann, T.; Gomard, G.; Lemmer, U. Large-Area Screen-Printed Internal Extraction Layers for Organic Light-Emitting Diodes. ACS Photon. 2017, 4, 928–933. [CrossRef] 230. Koh, T.-W.; Spechler, J.A.; Lee, K.M.; Arnold, C.B.; Rand, B.P. Enhanced Outcoupling in Organic Light-Emitting Diodes via a High-Index Contrast Scattering Layer. ACS Photon. 2015, 2, 1366–1372. [CrossRef] 231. Krotkus, S.; Kasemann, D.; Lenk, S.; Leo, K.; Reineke, S. Adjustable White-Light Emission from a Photo-Structured Micro-OLED Array. Light Sci. Appl. 2016, 5, e16121. [CrossRef] [PubMed] 28 of 28 28 of 28 Molecules 2019, 24, 151 232. Ji, W.; Zhang, L.; Gao, R.; Zhang, L.; Xie, W.; Zhang, H.; Li, B. Top-Emitting White Organic Light-Emitting Devices with Down-Conversion Phosphors: Theory and Experiment. Opt. Express 2008, 16, 15489. [CrossRef] [PubMed] 233. Jiang, C.; Zhong, Z.; Liu, B.; He, Z.; Zou, J.; Wang, L.; Wang, J.; Peng, J.B.; Cao, Y. Coffee-Ring-Free Quantum Dot Thin Film Using Inkjet Printing from a Mixed-Solvent System on Modiﬁed ZnO Transport Layer for Light-Emitting Devices. ACS Appl. Mater. Interfaces 2016, 8, 26162–26168. [CrossRef] [PubMed] 234. Liu, B.; Wang, L.; Gu, H.; Sun, H.; Demir, H.V. Highly Efﬁcient Green Light-Emitting Diodes from All-Inorganic Perovskite Nanocrystals Enabled by a New Electron Transport Layer. Adv. Opt. Mater. 2018, 5, 1800220. [CrossRef] 235. Xiao, P.; Huang, J.; Yan, D.; Luo, D.; Yuan, J.; Liu, B.; Liang, D. Emergence of Nanoplatelet Light-Emitting Diodes. Materials 2018, 11, 1376. [CrossRef] [PubMed] 236. Gao, Y.; Li, M.; Delikanli, S.; Zheng, H.; Liu, B.; Dang, C.; Sum, T.C.; Demir, H.V. Low-threshold lasing from colloidal CdSe/CdSeTe core/alloyed-crown type-II heteronanoplatelets. Nanoscale 2018, 10, 9466–9475. [CrossRef] [PubMed] 237. References Shi, Z.; Li, Y.; Zhang, Y.; Chen, Y.; Li, X.; Wu, D.; Xu, T.; Shan, C.; Du, G. High-efﬁciency and air-stable perovskite quantum dots light-emitting diodes with an all-inorganic heterostructure. Nano Lett. 2016, 17, 313–321. [CrossRef] [PubMed] 238. Shi, Z.; Li, S.; Li, Y.; Ji, H.; Li, X.; Wu, D.; Xu, T.; Chen, Y.; Tian, Y.; Zhang, Y.; et al. Strategy of solution-processed all-inorganic heterostructure for humidity/temperature-stable perovskite quantum dot light-emitting diodes. ACS Nano 2018, 12, 1462–1472. [CrossRef] [PubMed] © 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

Subsets and Splits

No community queries yet

The top public SQL queries from the community will appear here once available.