Update app.py

app.py

@@ -4,6 +4,85 @@ import streamlit as st
 summarizer = pipeline("summarization", model="facebook/bart-large-cnn")
 
 ARTICLE = """ 
+1
+Vol.:(0123456789)
+Scientifc Reports | (2022) 12:10472 | https://doi.org/10.1038/s41598-022-14569-0
+www.nature.com/scientificreports
+A functional defnition
+to distinguish ponds from lakes
+and wetlands
+David C. Richardson 1,19*, MeredithA. Holgerson 2,19, Matthew J. Farragher 3
+,
+Kathryn K. Hofman 4
+, Katelyn B. S. King 5
+, María B.Alfonso 6
+, Mikkel R.Andersen 7
+, Kendra Spence Cheruveil 8
+, KristenA. Coleman9
+, Mary Jade Farruggia 10,
+Rocio Luz Fernandez 11, Kelly L. Hondula 12, GregorioA. López Moreira Mazacotte 13, Katherine Paul1
+, Benjamin L. Peierls 14, Joseph S. Rabaey 15, Steven Sadro 10,
+María Laura Sánchez 16, Robyn L. Smyth 17 & Jon N. Sweetman 18
+Ponds are often identifed by their small size and shallow depths, but the lack of a universal evidencebased defnition hampers science and weakens legal protection. Here, we compile existing pond
+defnitions, compare ecosystem metrics (e.g., metabolism, nutrient concentrations, and gas fuxes)
+among ponds, wetlands, and lakes, and propose an evidence-based pond defnition. Compiled
+defnitions often mentioned surface area and depth, but were largely qualitative and variable.
+Government legislation rarely defned ponds, despite commonly using the term. Ponds, as defned in
+published studies, varied in origin and hydroperiod and were often distinct from lakes and wetlands
+in water chemistry. We also compared how ecosystem metrics related to three variables often
+seen in waterbody defnitions: waterbody size, maximum depth, and emergent vegetation cover.
+Most ecosystem metrics (e.g., water chemistry, gas fuxes, and metabolism) exhibited nonlinear
+relationships with these variables, with average threshold changes at 3.7± 1.8 ha (median: 1.5 ha)
+in surface area, 5.8 ± 2.5 m (median: 5.2 m) in depth, and 13.4 ± 6.3% (median: 8.2%) emergent
+vegetation cover. We use this evidence and prior defnitions to defne ponds as waterbodies that are
+small (< 5 ha), shallow (< 5 m), with< 30% emergent vegetation and we highlight areas for further
+study near these boundaries. This defnition will inform the science, policy, and management of
+globally abundant and ecologically signifcant pond ecosystems.
+Lentic (still) waterbodies have long been placed into categories to improve our understanding of aquatic ecosystems, aid science communication, and facilitate management decisions1,2
+. For instance, lentic ecosystems
+have been sorted into discrete categories by size or depth3,4
+, trophic status5
+, and mixing regime6,7
+. Ofen, lentic
+waterbodies are categorized by diferent ecosystem types, such as lakes, ponds, and wetlands (Fig. 1). Categorizing
+OPEN
+1
+Biology Department, State University of New York at New Paltz, New Paltz, NY, USA. 2
+Department of Ecology
+and Evolutionary Biology, Cornell University, Ithaca, NY, USA. 3
+School of Biology and Ecology, Climate Change
+Institute, University of Maine, Orono, ME, USA. 4
+Departments of Biology and Environmental Studies, St. Olaf
+College, Northfeld, MN, USA. 5
+Department of Fisheries and Wildlife, Michigan State University, East Lansing,
+MI, USA. 6
+Instituto Argentino de Oceanografía (IADO), Universidad Nacional del Sur (UNS)-CONICET, Florida
+8000, Complejo CCT CONICET Bahía Blanca, Edifcio E1, B8000BFW Bahía Blanca, Argentina. 7
+Centre for
+Freshwater and Environmental Studies, Dundalk Institute of Technology, Dundalk, Ireland. 8
+Department
+of Fisheries and Wildlife and the Lyman Briggs College, Michigan State University, East Lansing, MI,
+USA. 9
+Department of Geography, York University, Toronto, ON, Canada. 10Department of Environmental Science
+and Policy, University of California, Davis, Davis, CA, USA. 11National Scientifc and Technical Research Council
+(CONICET), Cordoba, Argentina. 12Battelle, National Ecological Observatory Network (NEON), Boulder, CO,
+USA. 13Department of Ecohydrology and Biogeochemistry, Leibniz-Institute of Freshwater Ecology and Inland
+Fisheries (IGB), Müggelseedamm 310, 12587 Berlin, Germany. 14Lakes Environmental Association, Bridgton,
+ME, USA. 15Department of Ecology, Evolution, and Behavior, University of Minnesota-Twin Cities, St. Paul, MN,
+USA. 16CONICET-Universidad de Buenos Aires, Buenos Aires, Argentina. 17Environmental and Urban Studies, Bard
+College, Annandale‑on‑Hudson, NY, USA. 18Department of Ecosystem Science and Management, Penn State
+University, University College, PA, USA. 19These authors contributed equally: David C. Richardson and Meredith
+A. Holgerson. *email: richardsond@newpaltz.edu
+2
+Vol:.(1234567890)
+Scientifc Reports | (2022) 12:10472 | https://doi.org/10.1038/s41598-022-14569-0
+www.nature.com/scientificreports/
+waterbodies using physical and biological characteristics facilitates generalizations and decision making, but
+categories may not always align with ecological inferences2
+.
+Categorizing small waterbodies is particularly challenging. Te majority of the world’s lentic waterbodies are
+small: over 95% are less than 10 ha (0.1 km2
+)3,8
 . Across the history of limnology, small and shallow waterbodies
 are widely referred to as ponds8–11, yet pond defnitions difer across the globe and are not based on scientifc
 evidence. Te lack of a universal, scientifcally-based pond defnition that diferentiates ponds from other lentic
@@ -27,5 +106,515 @@ from scientifc literature and evaluated legislative defnitions of ponds, wetland
 a large dataset of pond characteristics and ecosystem function from a global literature survey and compared
 ecosystem structural and functional metrics among ponds, wetlands, and lakes. Finally, we propose an evidencebased pond defnition.
 Results and discussion
+Current scientifc defnitions of ponds. We compiled existing scientifc defnitions of ponds by conducting a backwards and forwards search of papers referenced in or subsequently referencing three seminal
+pond papers8,17,18 (see “Methods”). We ultimately compiled 54 pond defnitions from scientifc literature (data
+available19). Te variables most ofen included in defnitions were surface area (91% of defnitions), depth (48%),
+permanence (48%), origin (i.e., natural or human-made; 33%), and standing water (33%; Fig. 2a). When surface
+area or depth were included in defnitions, they were ofen mentioned qualitatively (e.g., “small” and “shallow”).
+Of the 61% of defnitions that included a maximum pond surface area, the range was 0.1 to 100 ha, the median
+was 2 ha, and all but two defnitions were≤10 ha (Fig. 2b). For depth, only 17% of studies provided a maximum
+depth cutof, which ranged 2 to 8 m (Fig. 2c). Of the 26 defnitions mentioning permanence, 22 stated that ponds
+could be temporary or permanent and only three indicated that ponds are exclusively permanent waterbodies.
+Of the 18 defnitions mentioning origin, 17 mentioned that ponds could be natural or human-made with the
+remaining study indicating ponds can have diverse origins.
+Other important factors included in defnitions related to morphometry. For example, 30% of defnitions
+mentioned the potential for plants to colonize the entire basin, which relates to high light penetration (mentioned
+in 11% of defnitions) and/or shallow depths. For example, Wetzel11 defnes ponds as having enough light penetration that macrophyte photosynthesis can occur over the entire waterbody. As such, these conditions may be
+Figure 1. We call lentic waterbodies by a variety of names in the English language including ponds, lakes,
+wetlands, reservoirs, oxbows, prairie potholes, vernal pools, lagoons, dams, puddles, and shallow lakes. Tese
+names may or may not correspond to ecological and systematic diferences. Generally, laypeople and experts, as
+individuals, will quickly diferentiate among broad categories of ponds, lakes, and wetlands; however, individuals
+may respond in diferent ways depending on their background and experiences. We present three diferent
+images of waterbodies that could each be categorized as lake, pond, or wetland using objective (e.g., morphology
+or vegetative cover) or more subjective criteria keeping cognizant of the complexity within and potential overlap
+among waterbody types.
+3
+Vol.:(0123456789)
+Scientifc Reports | (2022) 12:10472 | https://doi.org/10.1038/s41598-022-14569-0
+www.nature.com/scientificreports/
+comparable to the littoral region of lakes (11% of defnitions). Lastly, 7% of pond defnitions mentioned mixing
+versus stratifcation, whereby ponds mix more than lakes20 yet less than shallow lakes due to a smaller fetch16.
+To assess if there was agreement in pond defnitions among papers, we examined the number of times each
+defnition was cited. Across the 54 defnitions, there were 89 citations of 48 unique papers. Ultimately, most
+papers (75%) were only cited only once, indicating no consensus in pond defnition. Te most cited paper was
+Biggs et al.21, which accounted for 15% of citations. Te next two most cited papers were Oertli et al.17 and Sondergaard et al.18, which were seminal papers included in our backwards-forwards search, and each comprised
+8% of citations.
+International defnitions. At an international level, there is no consensus on how to discriminate among
+ponds, lakes, and wetlands. In North America, wetlands are generally considered to be shallow:<2 m in Canada22
+and<2.5 m in the US23, which diferentiates them from lakes. Some nations, such as Australia, South Korea, and
+Uganda, explicitly include ponds and lakes in federal wetland defnitions24 (see also22). Te inclusion of ponds
+and some lakes within wetland defnitions ofen stems from the Ramsar Convention, an international body
+interested in global wetland conservation that has been signed by 172 countries representing 6 continents as
+of 202125. Te Ramsar Convention defned wetlands as “areas of marsh, fen, peatland, or water” across marine,
+brackish, and freshwater with varying degrees of permanence and natural or artifcial states with a maximum
+depth of 6 m26, which overlaps depths found in many defnitions of ponds and shallow lakes. In other countries,
+ponds are included in lake defnitions under federal conservation laws. For example, in the Danish “nature
+protection” law §3, lakes are defned as waterbodies with a surface area of>100 m2
+. As 98% of Danish ‘lakes’ are
+smaller than 1 ha27, this law protects many small waterbodies that may be considered ponds elsewhere. Still other
+agencies have only qualitative pond defnitions: the European Commission simply defnes ponds as “relatively
+shallow” and may also be called “pool, tarn, mere, or small lake,” a defnition also used by the International
+Union for Conservation of Nature28,29. Tese examples underscore that waterbody defnitions vary globally, are
+generally qualitative, and are rarely based on scientifc evidence relating to ecosystem structure or function. Te
+defnitions possibly derive from diferent management, protection, and monitoring strategies; for instance, the
+European Union’s Water Framework Directive excludes waterbodies<50 ha (0.5 km2
+) in size from monitoring30.
+Figure 2. Summary of “pond” defnitions from scientifc literature including (a) presence of various
+morphological, biological, and physical characteristics in the defnition as blue bars (n=54 defnitions total).
+Bold black lines indicate the number of defnitions with surface area and depth values. Histograms of the upper
+limits from “pond” defnitions for (b) surface area and (c) maximum depth.
+4
+Vol:.(1234567890)
+Scientifc Reports | (2022) 12:10472 | https://doi.org/10.1038/s41598-022-14569-0
+www.nature.com/scientificreports/
+Current U.S. Federal and State defnitions. In the US, waterbody defnitions vary among federal agencies, with implications for both legal protection and monitoring. Te US Environmental Protection Agency
+(EPA) and the US Army Corps of Engineers (ACE) defne wetlands based on saturated soils and hydrophytic
+vegetation, which has the potential to include ponds within the category of wetlands. Conversely, the US Fish
+and Wildlife Service (USFWS) distinguishes among wetlands and lakes based on surface area, depth, and emergent vegetation31. Lakes are≥8 ha or if smaller, they must be≥2.5 m in maximum depth. In contrast, wetlands
+are are dominated by>30% emergent plant cover; if there is less, the site may still be a wetland if<2.5 m deep
+and<8 ha in size. Terefore, ponds are ofen considered by USFWS to be wetlands, but this is not always the case:
+ponds have been used as an example of a waterbody that can be classifed as lake, wetland, or both32. Te lack of
+an explicit, unifed, and scientifcally based pond defnitions across three federal agencies (EPA, ACE, USFWS)
+is confusing and contributes to ponds being underrepresented in US aquatic waterbody monitoring relative to
+their numerical dominance on the landscape3,8
+. For example, US EPA monitoring programs include ponds in
+both the National Wetland Condition Assessment and the National Lake Assessment; however, “ponds” represent a small number of waterbodies in each of these surveys (<12% classifed qualitatively as “pond” in 2011
+wetland survey; 13% of waterbodies were<5 ha in 2012 lake survey).
+Refecting political and geographic variability at the national scale, most US states have their own waterbody
+protections33. We surveyed US state agencies to examine state defnitions of ponds, lakes, and wetlands (see
+“Methods”). Our survey responses included 42 of 50 (84%) states (Fig. 3). Only one state (Michigan) explicitly
+defned ponds, 11 states defned lakes (26%), and 30 states defned wetlands (71%). While only one state defned
+ponds, half of the surveyed states used the term “pond” in their legislation. Specifcally, ponds were referenced
+as state waters (e.g., Vermont) or were included in state defnitions for lakes (e.g., Kansas) or wetlands (e.g.,
+Rhode Island). It is unclear how these defnitions impact monitoring and protection or why the distinctions
+were originally made. For instance, many states monitor lakes based on minimum size thresholds, which vary
+widely from < 1 ha in Arizona and Alaska, 2–4 ha in many northeastern states, and up to 8 ha in Washington
+and Nebraska. Te variety of defnitions and monitoring size cutofs do not appear to be scientifcally based,
+but may stem from arbitrary decisions, historic references, mapping capabilities from decades ago, and resource
+limitations for monitoring; the same rationale for defnitions likely apply to local, regional, and international
+organizations around the globe.
+Comparing lake, pond, and wetlands characteristics from literature. We compared biological,
+physical, and chemical characteristics of waterbodies that scientists called lakes, ponds, or wetlands in published
+studies. To obtain data for the pond characteristics, we used the same literature search summarized above for
+pond defnitions (also, see “Methods”). From the 519 papers that we examined, we extracted data on sites the
+authors called “ponds” and other variants (e.g., ‘small ponds’, ‘fsh ponds’, but NOT ‘lakes’). We fltered waterbodies that were≤20  ha surface area and≤9  m depth (global distribution; n=1327) to include waterbodies
+slightly greater than the maximum depth and maximum surface area used to defne ponds in prior studies34,35.
+To compare ponds to lakes and wetlands, we used existing lake (US and Europe; n=55,173) and wetland (US;
+n=400) databases; waterbodies were classifed as lake or wetland by the scientists or managers who published
+the database. Wetlands were classifed as<1 m in depth with no defned surface area and lakes were all>0.02 ha
+with no defned depth (see “Methods’’ for details).
+From the waterbodies that scientists called “ponds,” hydroperiod and origin varied over a large range of
+characteristics. Of the 608 ponds with hydroperiod data, permanent ponds accounted for 74% (n =450) and
+temporary ponds for 26% (n=158). Out of 648 ponds with known origins, 65% (n=418) were constructed or
+manipulated and 35% (n=230) were natural. Terefore, scientists consider ponds to be inclusive of both permanent and temporary hydroperiod and have natural or human-made origins.
+Figure 3. US state responses to surveys indicating if the state has a defnition of wetland, lake, or pond and if
+the state used the term “pond” in their legislation. NR=no response.
+5
+Vol.:(0123456789)
+Scientifc Reports | (2022) 12:10472 | https://doi.org/10.1038/s41598-022-14569-0
+www.nature.com/scientificreports/
+When examining water chemistry, nutrients, and biotic data across diferent waterbody types, as defned by
+publishing scientists and managers, we found that ponds were distinct from lakes and wetlands in two metrics
+(TN, pH), similar to wetlands in one metric (TP), and similar to lakes in one metric (chl a; Fig. 4; Tables S1, S2).
+For example, ponds had distinctly high TN concentrations, which were greater than either lakes or wetlands
+(Fig. 4b; Table S2). Ponds and wetlands had similarly high TP concentrations, which were signifcantly greater
+than lakes; ponds were also most variable in TP (Fig. 4a; Table S2). Lastly, ponds chlorophyll (chl) a concentrations were similar to lakes, with wetlands being most variable but lower, on average (Fig. 4d; Table S2).
+Does ecosystem structure and function distinguish ponds from lakes and wetlands? We evaluated the relationship between key metrics of ecosystem structure or function with three quantitative variables
+that ofen showed up in pond, lake, or wetland defnitions: surface area, maximum depth (hereafer depth), and
+emergent vegetation cover. Our metrics of ecosystem structure or function include nutrients (total phosphorus
+(TP), total nitrogen (TN)), water chemistry (pH), primary producer biomass (chl a), metabolism (gross primary
+production—GPP, respiration—R, net ecosystem production—NEP), and heat and gas distributions and movement (diel temperature ranges—DTR, methane fuxes, gas transfer velocities). Te data was collated from global
+surveys of literature and federal or international databases (see “Methods”) with ultimately ten comparisons for
+surface area, six comparisons for depth, and four comparisons for emergent vegetation cover with a range of
+sample sizes for each comparison (n=67 to 7931, see Tables S3, S5, S7). We assessed each relationship for four
+Figure 4. Comparison of various chemical and biological parameters across wetlands, ponds, and lakes, with
+waterbody category based on the term used by publishing scientists and managers (Table S2). Violin plots
+indicate distributions of waterbody characteristics, the white box indicates 25th to 75th percentile with median
+in the middle, whiskers indicate 1.5×interquartile range, and outliers are black closed circles. Letters inside the
+plot indicate signifcant diferences in means (LSD, alpha=0.05). Note all x-axes have logarithmic scales.
+6
+Vol:.(1234567890)
+Scientifc Reports | (2022) 12:10472 | https://doi.org/10.1038/s41598-022-14569-0
+www.nature.com/scientificreports/
+diferent patterns in increasing order of complexity: null, linear, segmented (nonlinear), and logistic (nonlinear)
+patterns and selected the best ft and most parsimonious relationship.
+Ecosystem structure and function were mostly nonlinearly related to surface area (n=9/10 variables), depth
+(n=5/6 variables), and emergent vegetation cover (n=3/4 variables) with both segmented and logistic relationships occurring (Figs. 5, 6, 7; Tables S3–S8). For surface area, six variables had logistic relationships: TP (Fig. 5b),
+methane fuxes (Fig. 5d), chl a (Fig. 5f), diel temperature range (Fig. 5h), gas exchange rates (k600; Fig. 5i), and
+pH (not pictured). Te infection occurred at 0.8 ha for TP, 1.1 ha for methane fuxes, 1.5 ha for chl a, 1.7 ha for
+pH, 4.6 ha for diel temperature range, and 17.5 ha for gas exchange rates (Table S3). NEP (Fig. 5c), R (Fig. 5e),
+and TN (Fig. 5g) all had segmented linear relationships where smaller systems had steeper slopes than larger
+systems (Table S4). Te breakpoint in surface area was 1.0 ha for NEP, 1.2 ha for R, and 3.8 ha for TN (Table S3).
+For depth, two variables had logistic relationships: diel temperature range (Fig. 6e) and chlorophyll a (Fig. 6f),
+with the infection occurring at 5.9 m and 14.9 m, respectively (Table S5). pH (Fig. 6b), TP (Fig. 6c), and TN
+(Fig. 6d) all had segmented linear relationships where smaller systems had steeper slopes than larger systems
+(Table S6) with breakpoints occurring at 1.0, 2.1, and 5.2 m, respectively (Table S5). For emergent vegetation
+cover, TN (Fig. 7b), TP (Fig. 7c), and pH (Fig. 7d) all had segmented linear relationships where systems with
+more emergent vegetation had steeper slopes than more open systems (Table S8). Te breakpoint in emergent
+vegetation cover was 6.0% for TN, 8.2% for TP, and 26.0% for pH (Table S7).
+To summarize across all three metrics (surface area, depth, and emergent vegetation cover), we evaluated
+where the boundaries of nonlinear relationships generally occurred, which informs boundaries between ponds,
+lakes, and wetlands (Table 1). For surface area, the boundary was 3.7±1.8 ha (mean±standard error) and the
+median was 1.5 ha, consistent with the median of 2 ha from scientifc defnitions (Fig. 2b). Te depth boundary
+was 5.8 ± 2.5 m (mean ± standard error) and the median was 5.2 m, within the range of scientifc defnitions
+(Fig. 2c). Te emergent vegetation cover boundary was 13.4±6.3% (mean±standard error) and the median was
+8.2%, both of which were lower than the previously identifed wetland lower bound of 30%31.
+Figure 5. Relationships between lentic waterbody size (excluding wetlands) and ecosystem structure and
+function metrics: (a) gross primary production (GPP), (b) total phosphorus concentrations (TP), (c) net
+ecosystem production (NEP), (d) methane fuxes (CH4 fux), (e) respiration (R), (f) chlorophyll a concentrations
+(Chl a), (g) total nitrogen concentrations (TN), (h) diel temperature ranges (DTR), and (i) gas transfer piston
+velocity (k600). Optimal model fts from null, linear, segmented, and logistic curves in bold foreground lines. For
+nonlinear segmented and logistic models (b–i), plots are ordered by boundaries between ponds and lakes, as
+defned by model breakpoints or infection points (vertical background lines).
+7
+Vol.:(0123456789)
+Scientifc Reports | (2022) 12:10472 | https://doi.org/10.1038/s41598-022-14569-0
+www.nature.com/scientificreports/
+Pond morphology (e.g., size and depth) creates fundamentally distinct conditions that govern ecosystem
+structure and function. Specifcally, ponds experience less wind-driven turbulence than larger waterbodies due
+to small fetch and sheltering from the landscape36. We found that gas exchange rates (k600) decreased at~18 ha,
+presumably due to reduced wind shear (Fig. 5i; also supported by37) and altered thermal dynamics. For instance,
+ponds and shallow lakes can warm dramatically during the day, inducing stratifcation, and cool of and mix completely overnight38. We found higher diel temperature ranges were more common in waterbodies<5 ha (Fig. 5h)
+and<6 m (Fig. 6e; see also39). Such diferences in temperature and mixing can promote internal nutrient loading40
+and ecosystem respiration41, which may explain the higher TN (Figs. 4b, 5g), TP (Figs. 4a, 5b) and ecosystem
+respiration (Fig. 5e) found in ponds. Lastly, diferences in water column mixing, increased nutrients, and higher
+respiration can all contribute to the higher greenhouse gas emissions found in ponds relative to lakes (Fig. 5d)4,42.
+Metrics of phytoplankton biomass (chl a) and total ecosystem production in the water (GPP) exhibited weak
+or inconsistent relationships with surface area and depth, likely due to diferences in the location and types of
+primary production across waterbody types. While total primary production in deep lakes is ofen dominated
+by phytoplankton43, shallow waterbodies can shif toward non-planktonic primary production like benthic algae
+or foating, emergent, or submerged macrophytes44. Ponds have pelagic phytoplankton, benthic algae (i.e., periphyton), and sediment rooted-submerged or foating macrophytes. In contrast, wetland productivity ofen predominantly occurs above the air–water interface45. Where emergent vegetation dominates, they may limit light
+and reduce water column nutrients, both of which are needed by phytoplankton and periphyton. Macrophytes
+can also modify water column and sediment geochemistry by providing autotrophic organic carbon and oxygen
+to rooting systems in the sediments46. Consequently, these opposing drivers can explain the high variability in
+primary production we observed (Fig. 5f, Table S2). Distinguishing ponds from wetlands will ultimately be aided
+by additional ecosystem measurements of metabolism, greenhouse gas production, and additional metrics (e.g.,
+carbon burial) across shallow waterbodies with a range of emergent vegetation cover.
+Figure 6. Relationships between lentic waterbody maximum depth (Max depth) and various ecosystem
+structure and function metrics: (a) methane fuxes (CH4 fux), (b) pH, (c) total phosphorus concentrations (TP),
+(d) total nitrogen concentrations (TN), (e) diel temperature ranges (DTR), and (f) chlorophyll a concentrations
+(Chl a) from literature data extraction with optimal model fts from null, linear or null, segmented linear, and
+logistic curves in bold foreground lines. For nonlinear segmented and logistic models (b–f), plots are ordered by
+model breakpoints or infection points (vertical background lines), indicative of boundaries between ponds and
+lakes.
+8
+Vol:.(1234567890)
+Scientifc Reports | (2022) 12:10472 | https://doi.org/10.1038/s41598-022-14569-0
+www.nature.com/scientificreports/
+A functional pond defnition. Our review of existing pond defnitions highlights that surface area and
+depth are the most common variables used to defne ponds; yet how small and how shallow a waterbody must
+be to classify as a pond is unclear, with defnitions ranging by orders of magnitude. Emergent vegetation is a
+third variable useful in distinguishing wetlands from ponds, but the threshold value,>30% emergent vegetation
+coverage for wetlands established at the US federal level, is not based on documented changes in ecosystem function. Comparing characteristics among waterbodies that scientists self-categorized into lakes, ponds, or wetlands, ponds were sometimes distinct from lakes and wetlands (pH, TN), sometimes similar to wetlands (TP),
+and sometimes similar to lakes (chl a), suggesting ponds are an ecologically distinct type of ecosystem. Lastly, we
+Figure 7. Relationships between lentic waterbody emergent vegetation cover (Emergent veg.) and various
+ecosystem structure and function metrics: (a) chlorophyll a concentrations (Chl a), (b) total nitrogen
+concentrations (TN), (c) total phosphorus concentrations (TP), (d) pH from literature data extraction with
+optimal model fts from null, linear or null, segmented linear, and logistic curves in bold foreground lines. For
+nonlinear segmented and logistic models (b–d), plots are ordered by model breakpoints or infection points
+(vertical background lines), indicative of boundaries between ponds and wetlands.
+Table 1. Nonlinear boundary values, parameter estimate±standard error (SE), from comparisons between
+surface area, maximum (max.) depth, and emergent vegetation (veg.) cover and ecosystem structure/function
+metrics including gross primary production (GPP), total phosphorus concentrations (TP), methane fuxes
+(CH4 fux), respiration (R), net ecosystem production (NEP), chlorophyll a concentrations (Chl a), pH, total
+nitrogen concentrations (TN), diel temperature ranges (DTR), and gas transfer piston velocity (k600). Boundary
+estimates are included if the nonlinear models (segmented regression or logistic relationships) were selected as
+optimal fts with standard error as determined when ftting the parameter. NA indicates a null or linear ft, –
+indicates not enough data was available to perform the analysis.
+Ecosystem metric
+Surface area
+Boundary est.±SE (ha)
+Max. depth
+Boundary est.±SE (m)
+Emergent veg. cover
+Boundary est.±SE (%)
+GPP NA – –
+TP 0.8±1.2 2.1±1.2 8.2±1.2
+NEP 1.0±1.4 – –
+CH4 fux 1.1±1.7 NA –
+R 1.2±1.5 – –
+Chl a 1.5±1.7 14.9±1.2 NA
+pH 1.7±1.5 1.0±1.4 26.0±1.3
+TN 3.8±1.4 5.2±1.4 6.0±1.3
+DTR 4.6±1.3 5.9±1.3 –
+k600 17.5±1.5 – –
+Mean 3.7±1.8 5.8±2.5 13.4±6.3
+Median 1.5 5.2 8.2
+9
+Vol.:(0123456789)
+Scientifc Reports | (2022) 12:10472 | https://doi.org/10.1038/s41598-022-14569-0
+www.nature.com/scientificreports/
+found clear nonlinear relationships when we examined relationships between ecosystem structure or function
+and surface area, depth, and emergent vegetation cover; these boundaries help to quantitatively defne ponds.
+Specifcally, we found that across available ecosystem metrics, ecosystems shif in structure and function at
+average (±SE) values of 3.7 (±1.8) ha in size, 5.8 (±2.5) m in depth, and 13.4 (±6.3) % emergent vegetation cover
+(Table 1). For surface area, all but one ecosystem metric (k600) was below 5 ha in surface area, which fts well
+within the range of most existing defnitions (≤10 ha; Fig. 2), and we suggest may be used to distinguish ponds
+from lakes. For maximum depth, all but one ecosystem metric (chl. a) was below a 6 m depth threshold, which
+also fts well within the range of depths reported in pond defnitions (Fig. 2), and matches the published threshold of 5 m maximum depth for shallow lakes44. Our depth analysis was less robust than surface area because we
+had less depth data, a common challenge in lentic studies47; we therefore advise further studies in waterbodies
+to explicitly evaluate this threshold. Until further work is done, we recommend using 5 m as a maximum depth
+threshold for ponds as it is close to both threshold shifs in ecosystem function and matches with the shallow
+lake literature44,48. We had the fewest ecosystem metric comparisons for emergent vegetative cover, and observed
+three nonlinear boundaries ranging from 6 to 26% cover. Te mean (13.4%), though smaller, is not statistically
+diferent than the 30% emergent vegetation cover (one sample t-test, t2=− 2.6, p=0.12) proposed by Cowardin
+et al.31 to separate wetlands from lakes. We recommend separating ponds and wetlands using the 30% coverage
+in emergent vegetation threshold for now, but recognize that the Cowardin et al.31 metric is not data driven and
+our analysis was limited by existing data. Future studies must examine how ecosystem structure and function
+shifs across a gradient of emergent vegetation cover to better functionally distinguish wetlands from ponds and
+could ultimately lower that boundary.
+Our review of data from the literature showed scientists and managers view ponds as permanent or temporary
+and natural or human made in origin. Terefore, we felt it necessary to provide the inclusivity of these concepts
+in a pond defnition. Other defnitions also link depth to light availability, where light penetrates to the sediments
+across the pond (e.g.,11). However, light availability is not only mediated by depth; even in the shallowest systems
+light can be limiting due to turbidity, dissolved organic matter, and submerged or foating plants (e.g.,49,50). For
+example, foating duckweed can cover most of a pond’s surface area and reduce light penetration to<1% relative
+to the light above the water’s surface49, and dramatically change the ecology of shallow systems51.
+As our analyses indicate that ponds are functionally distinct from lakes and wetlands, we propose the following scientifcally informed pond defnition (Fig. 8):
+Ponds are small and shallow waterbodies with a maximum surface area of 5 ha, a maximum depth of 5 m,
+and < 30% coverage of emergent vegetation. Ponds will have light penetration to the sediments if water
+clarity permits and can be permanent or temporary and natural or human-made.
+Our proposed defnition is based on the current state of the science; we anticipate that future research will
+further resolve diferences among these fve categories. For example, we call for future research to examine how
+ecosystem structure and function shif across our proposed boundaries, particularly for depth and emergent
+Figure 8. Conceptual model to defne lentic waterbodies based on three diferent criteria (depth, surface area,
+and emergent vegetation). Boundaries for all three axes come from our analysis and are informed by existing
+pond, lake, and wetland defnitions. Figure by Visualizing Science.
+10
+Vol:.(1234567890)
+Scientifc Reports | (2022) 12:10472 | https://doi.org/10.1038/s41598-022-14569-0
+www.nature.com/scientificreports/
+vegetation, which had smaller sample sizes and fewer ecosystem metrics than surface area. Additional variables
+such as basin geometry (e.g., area:volume ratios), sheltering from wind, water residence time, water clarity, and
+geographic location, may also afect a waterbody’s ecosystem structure and function, creating some overlap
+between classifcations especially along the upper and lower bounds of our pond defnition. For instance, a
+landscape with little wind sheltering increases water column mixing that could cause a waterbody the size of a
+pond to function more like a shallow lake. We therefore advocate for additional sampling of lentic waterbodies,
+especially in locations where lentic waterbodies are understudied or being rapidly constructed like tropical and
+subtropical regions52, to help resolve boundaries among waterbody types and further refne the pond defnition.
+Conclusion
+Scientists, policy makers, water resource managers and the public all use the word “pond” to describe small and
+shallow waterbodies, which are globally abundant8
+ and hotspots for biogeochemistry4,8
+ and biodiversity53. Yet,
+the lack of a universal pond defnition means that ponds can fall between lake and wetland jurisdictions and
+categorizations22, thus potentially limiting their legal protections. Globally, the situation is similar to US policy
+as some nations defne ponds as wetlands (e.g., the Ramsar Convention), some as lakes (e.g., Denmark), and
+others specifcally defne ponds (e.g., United Kingdom). Te pond defnition presented here will favor more
+frequent and consistent use of the term and ultimately improve the protection, monitoring, and scientifc study
+of ponds, which are globally abundant and structurally and functionally distinct from other lentic waterbodies.
+Methods
+Literature survey. To compile biological, physical, and chemical characteristics of ponds, we conducted a
+literature search based on three seminal papers establishing the ecological importance of ponds: Oertli et al.35,
+Søndergaard et al.18, and Downing8
+, each of which has>100 citations and is more than ten years old. We conducted a backwards and forwards search in April 2019 to compile all papers cited by these three papers, and
+all papers that cited them, yielding 519 unique papers. We extracted physical, chemical, and biological data for
+papers that reported data for individual waterbodies defned as ponds by the publishing scientists. To ensure
+consideration of all potential ponds, we checked that waterbodies selected were small (≤20 ha in surface area)
+and shallow (≤9 m in maximum or mean depth), boundaries that are slightly greater than the maximum depth
+(8 m)35 and maximum surface area (10 ha)34 used to defne ponds in a few prior studies. We used the resulting
+1327 waterbodies in our analysis, which had a global distribution (Fig. S1)19.
+Scientifc defnitions. To investigate how scientifc researchers defned ponds, we reviewed all 519 papers
+for pond defnitions. We included defnitions where the authors explicitly referred to their study waterbodies
+as ponds (e.g., we excluded “shallow lakes” and “small lakes”), yielding 40 pond defnitions. Te defnitions
+included 89 citations of 48 unique papers; we evaluated all cited papers that were not already in our compilation
+for additional defnitions and citations. Tis process added 14 defnitions, plus an additional fve cited papers not
+assessed due to our inability to access or translate them (data available19).
+Federal and state defnitions. We examined policy defnitions using the United States (US) federal and
+state legislation as an example because we posited diferences at this scale would refect the challenges faced by
+governments from other countries in formulating a unifed pond defnition. At the federal level, we examined
+three agencies with monitoring or regulatory responsibilities: US Environmental Protection Agency (EPA), US
+Army Corps of Engineers (ACE), and US Fish and Wildlife Service (USFWS). Due to the difculty of fnding all
+state policies, we sent electronic surveys to individuals working in state environmental agencies in all states. We
+asked whether their state defned lakes, ponds, and wetlands, and requested the legislative sources. We received
+responses from 42/50 states and evaluated all defnitions provided and their associated legislation.
+Pond, lake, and wetland data. We compared chemical and biological characteristics among various
+lentic waterbodies as defned by scientists as ponds, wetlands, or lakes. For ponds, we used data from the literature data extraction as described above (n=1327). Wetland data came from the US EPA’s 2011 National
+Wetland Condition Assessment, which surveyed wetlands with standing water<1 m in depth and variable surface area54,55. We selected wetland sites that were freshwater and had water chemistry data (n=400). Lake data
+was extracted from LAGOS-NE (lakes≥4 ha; n=51,101)56, EPA’s 2012 National Lake Assessment (lakes≥1 ha;
+n=1130)57,58, and the European Environmental Agency’s Waterbase database (lakes>0.02 ha; n=2942)59. From
+these sources, we compared nutrients (total phosphorus (TP), total nitrogen (TN)), water chemistry (pH), and
+primary producer biomass (chl a) among waterbody types (ponds, wetlands, and lakes).
+We also examined diferences in six additional metrics of ecosystem function across waterbodies using data
+from a variety of sources and ranging in sample size from 67 to 198 global sites (gross primary production—GPP,
+respiration—R, net ecosystem production—NEP, diel temperature ranges—DTR, methane fuxes, gas transfer
+velocities). We extracted metabolism metrics (GPP and R) from an existing literature review60 and two published
+studies of various sized lentic ecosystems41,61. DTR, calculated as the diel diference between the maximum and
+minimum surface temperature for each waterbody, were extracted from multiple studies38,39. Areal methane
+fuxes42 and gas transfer velocities (k600)37 were extracted from existing literature reviews.
+Comparing lake, pond, and wetlands characteristics from literature. We evaluated whether there
+were diferences among waterbody types as defned by scientists and managers. We determined signifcant differences in waterbody characteristics across waterbody types using ANOVA and post-hoc Least Signifcant
+11
+Vol.:(0123456789)
+Scientifc Reports | (2022) 12:10472 | https://doi.org/10.1038/s41598-022-14569-0
+www.nature.com/scientificreports/
+Diference (LSD) analysis. We determined the variation within each freshwater type using the coefcient of
+variation (cv) and tested for signifcant diferences using Levene’s test. We acknowledge that the confounding
+defnitions of waterbodies resulted in overlapping size distributions; therefore any statistical diferences across
+waterbody types will be conservative.
+Does pond ecosystem structure and function distinguish ponds difer from lakes and wet- lands? We evaluated where the cutofs might exist along pond to lake and pond to wetland gradients. We
+used the relationship between surface area, depth, or emergent vegetation cover represented by x below and
+each ecosystem variable (n=10 variables for surface area; n=6 variables for depth, n=4 for emergent vegetation cover) represented by y below for four diferent patterns in increasing order of complexity: null, linear,
+segmented (nonlinear), and logistic (nonlinear) patterns. We ft null models by taking the arithmetic mean
+(Eq. 1), linear models using ordinary least-squares linear regression (Eq. 2), segmented using regressions with
+one breakpoint (Eq. 3) via the “segmented” package62, and logistic using sigmoidal curves (Eq. 4) via the nls
+function in R (Fig. S2). Parameters a – h and bp (breakpoint) were ft using the methods above.
+We log-transformed surface area and depth to account for the several order of magnitude scale and nonnormality. Similarly, we transformed some of the ecosystem variables depending on normality and distributions.
+To select among the four models for each relationship, we examined the AICc fts and selected the minimum
+AICc as the optimal ft with consideration of other model fts within 11 units of the minimum AICc using root
+mean squared error and parsimony63. If one of the nonlinear models was selected, we then objectively quantifed
+the boundary among ecosystem types (i.e., pond vs. lake or pond vs. wetland) using either the breakpoint or the
+infection point parameter from the segmented regression or sigmoid curve, respectively (Fig. S2c,d).
+Data availability
+All data used for this manuscript is available through an Environmental Data Initiative data publication (Richardson et al. 2022: https://doi.org/10.6073/pasta/ec507ac70846b17d0633d95aa3c680c6).
+Received: 7 January 2022; Accepted: 8 June 2022
+References
+1. Alexander, L. C. Science at the boundaries: Scientifc support for the Clean Water Rule. Freshw. Sci. 34, 1588–1594 (2015).
+2. Kraemer, B. M. Rethinking discretization to advance limnology amid the ongoing information explosion. Water Res. 178, 115801
+(2020).
+3. Verpoorter, C., Kutser, T., Seekell, D. A. & Tranvik, L. J. A global inventory of lakes based on high-resolution satellite imagery.
+Geophys. Res. Lett. 41, 6396–6402 (2014).
+4. Holgerson, M. A. & Raymond, P. A. Large contribution to inland water CO2 and CH4 emissions from very small ponds. Nat. Geosci.
+9, 222–226 (2016).
+5. Dodds, W. K. & Cole, J. J. Expanding the concept of trophic state in aquatic ecosystems: It’s not just the autotrophs. Aquat. Sci. 69,
+427–439 (2007).
+6. Hutchinson, G. E. & Lofer, H. Te thermal classifcation of lakes. Proc. Natl. Acad. Sci. 42, 84–86 (1956).
+7. Lewis, W. M. Jr. A revised classifcation of lakes based on mixing. Can. J. Fish. Aquat. Sci. 40, 1779–1787 (1983).
+8. Downing, J. Emerging global role of small lakes and ponds: Little things mean a lot. Limnetica 29, 9–24 (2010).
+9. Tienemann, A. Die binnengewässer mitteleuropas: eine limnologische einfurung (E. Schweizerbart, 1925).
+10. Welch, P. S. Limnology (McGraw-Hill, 1952).
+11. Wetzel, R. Limnology (Academic Press, 2001).
+12. Wisconsin DNR. Wisconsin Lakes. (2009).
+13. Minnesota Department of Natural Resources. Lakes, rivers, and wetlands facts. Minnesota Lakes, rivers, and wetlands facts https://
+www.dnr.state.mn.us/faq/mnfacts/water.html (2013).
+14. Litke, E. Who has more lakes: Minnesota or Wisconsin? politifact.com https://www.politifact.com/factchecks/2019/may/23/sarameaney/who-has-more-lakes-minnesota-or-wisconsin/ (2019).
+15. Tornton, B. F., Wik, M. & Crill, P. M. Double-counting challenges the accuracy of high-latitude methane inventories. Geophys.
+Res. Lett. 43, 12–569 (2016).
+16. Fairchild, G. W., Anderson, J. N. & Velinsky, D. J. Te trophic state ‘chain of relationships’ in ponds: Does size matter?. Hydrobiologia
+539, 35–46 (2005).
+17. Oertli, B., Céréghino, R., Hull, A. & Miracle, R. Pond conservation: From science to practice. Hydrobiologia 634, 1–9 (2009).
+18. Søndergaard, M., Jeppesen, E. & Jensen, J. P. Pond or lake: Does it make any diference?. Arch. Für Hydrobiol. 162, 143–165 (2005).
+19. Richardson, D. C. et al. Pond data: Physical, chemical, and biological characteristics with scientifc and United States of America
+state defnitions from literature and legislative surveys. Environ. Data Initiat. https://doi.org/10.6073/pasta/ec507ac70846b17d0633
+d95aa3c680c6 (2022).
+20. Chofel, Q., Touchart, L., Bartout, P. & Al Domany, M. Temporal and spatial variations in heat content of a French pond. Geogr.
+Tech. 12, 9–22 (2017).
+21. Biggs, J., Williams, P., Whitfeld, M., Nicolet, P. & Weatherby, A. 15 years of pond assessment in Britain: Results and lessons learned
+from the work of Pond Conservation. Aquat. Conserv. Mar. Freshw. Ecosyst. 15, 693–714 (2005).
+(1) y = a
+y = bx + c (2)
+y = {d1 ∗ x + e1, x ≤ bp d2 ∗ x + e2, x > bp (3)
+y = f + (4) g − f
+1 + e(bp−x)/h
+12
+Vol:.(1234567890)
+Scientifc Reports | (2022) 12:10472 | https://doi.org/10.1038/s41598-022-14569-0
+www.nature.com/scientificreports/
+22. Tiner, R. W. A Guide to Wetland Formation, Identifcation, Delineation, Classifcation, and Mapping 2nd edn. (CRC Press, 2016).
+23. Federal Geographic Data Committee. Classifcation of wetlands and deepwater habitats of the United States. FGDC-STD-004-2013.
+2nd ed (2013).
+24. Kiai, S. P. M. & Mailu, G. M. Kenya Country Paper (FAO–Food and Agriculture Organization of the United Nations, 1998).
+25. Bridgewater, P. & Kim, R. E. Te Ramsar convention on wetlands at 50. Nat. Ecol. Evol. 5, 268–270 (2021).
+26. Ramsar Information Bureau. What are Wetlands? Ramsar Information Paper No. 1. https://www.ramsar.org/sites/default/fles/
+documents/library/info2007-01-e.pdf (2007).
+27. Sand-Jensen, K. Nature in Denmark: Te Fresh Water. (Gyldendal Trade 150, 2013).
+28. European Commission. Pond. Knowledge for policy glossary https://knowledge4policy.ec.europa.eu/glossary-item/pond_en (2018).
+29. IUCN (International Union for Conservation of Nature). International Union for Conservation of Nature glossary of defnitions.
+https://www.iucn.org/sites/dev/fles/iucn-glossary-of-defnitions_en_2021.05.pdf (2021).
+30. Hill, M. J. et al. Pond ecology and conservation: Research priorities and knowledge gaps. Ecosphere 12, e03853 (2021).
+31. Cowardin, L. M., Carter, V., Golet, F. C. & LaRoe, E. T. Classifcation of Wetlands and Deepwater Habitats of the United States. (Fish
+and Wildlife Service, U.S. Department of the Interior, 1979).
+32. Tiner, R. W. Dichotomous Keys and Mapping Codes for Wetland Landscape Position, Landform, Water Flow Path, and Waterbody
+Type: Version 3.0. 65 (2014).
+33. Sullivan, S. M. P., Rains, M. C. & Rodewald, A. D. Opinion: Te proposed change to the defnition of “waters of the United States”
+fouts sound science. Proc. Natl. Acad. Sci. 116, 11558–11561 (2019).
+34. Kalf, J. Limnology (Prentice-Hall Inc., 2002).
+35. Oertli, B. et al. Conservation and monitoring of pond biodiversity: Introduction. Aquat. Conserv. Mar. Freshw. Ecosyst. 15, 535–540
+(2005).
+36. Markfort, C. D. et al. Wind sheltering of a lake by a tree canopy or bluf topography. Water Resour. Res. 46, W03530 (2010).
+37. Holgerson, M. A., Farr, E. R. & Raymond, P. A. Gas transfer velocities in small forested ponds. J. Geophys. Res. Biogeosci. 122,
+1011–1021 (2017).
+38. Martinsen, K. T., Andersen, M. R. & Sand-Jensen, K. Water temperature dynamics and the prevalence of daytime stratifcation in
+small temperate shallow lakes. Hydrobiologia 826, 247–262 (2019).
+39. Woolway, R. I. et al. Diel surface temperature range scales with lake size. PLoS ONE 11, e0152466 (2016).
+40. Wilhelm, S. & Adrian, R. Impact of summer warming on the thermal characteristics of a polymictic lake and consequences for
+oxygen, nutrients and phytoplankton. Freshw. Biol. 53, 226–237 (2008).
+41. Staehr, P. A., Baastrup-Spohr, L., Sand-Jensen, K. & Stedmon, C. Lake metabolism scales with lake morphometry and catchment
+conditions. Aquat. Sci. 74, 155–169 (2012).
+42. Deemer, B. R. & Holgerson, M. A. Drivers of methane fux difer between lakes and reservoirs, complicating global upscaling
+eforts. J. Geophys. Res. Biogeosci. 126, e2019JG005600 (2021).
+43. Vadeboncoeur, Y., Peterson, G., Vander Zanden, M. J. & Kalf, J. Benthic algal production across lake size gradients: Interactions
+among morphometry, nutrients, and light. Ecology 89, 2542–2552 (2008).
+44. Schefer, M. Te story of some shallow lakes. In Ecology of Shallow Lakes (ed. Schefer, M.) 1–19 (Springer Netherlands, 2004).
+https://doi.org/10.1007/978-1-4020-3154-0_1.
+45. Hagerthey, S. E., Cole, J. J. & Kilbane, D. Aquatic metabolism in the Everglades: Dominance of water column heterotrophy. Limnol.
+Oceanogr. 55, 653–666 (2010).
+46. Benelli, S. & Bartoli, M. Worms and submersed macrophytes reduce methane release and increase nutrient removal in organic
+sediments. Limnol. Oceanogr. Lett. 6, 329–338 (2021).
+47. Oliver, S. K. et al. Prediction of lake depth across a 17-state region in the United States. Inland Waters 6, 314–324 (2016).
+48. Padisák, J. & Reynolds, C. S. Shallow lakes: Te absolute, the relative, the functional and the pragmatic. Hydrobiologia 506, 1–11
+(2003).
+49. Holgerson, M. A., Lambert, M. R., Freidenburg, L. K. & Skelly, D. K. Suburbanization alters small pond ecosystems: Shifs in
+nitrogen and food web dynamics. Can. J. Fish. Aquat. Sci. 75, 641–652 (2018).
+50. Schefer, M. et al. Floating plant dominance as a stable state. Proc. Natl. Acad. Sci. 100, 4040–4045 (2003).
+51. Schefer, M. & van Nes, E. H. Shallow lakes theory revisited: various alternative regimes driven by climate, nutrients, depth and
+lake size. In Shallow Lakes in a Changing World (eds Gulati, R. D. et al.) 455–466 (Springer Netherlands, 2007). https://doi.org/10.
+1007/978-1-4020-6399-2_41.
+52. Yuan, J. et al. Rapid growth in greenhouse gas emissions from the adoption of industrial-scale aquaculture. Nat. Clim. Change 9,
+318–322 (2019).
+53. Biggs, J., von Fumetti, S. & Kelly-Quinn, M. Te importance of small waterbodies for biodiversity and ecosystem services: implications for policy makers. Hydrobiologia 793, 3–39 (2017).
+54. USEPA. National wetland condition assessment 2011: a collaborative survey of the nation’s wetlands. EPA 843-R-15-005. (2016).
+55. USEPA. United States Environmental Protection Agency National Wetland Condition Assessment 2011 (nwca2011_siteinfo.csv,
+nwca2011_waterchem.csv, nwca2011_chla.csv, and corresponding metadata fles). (2016).
+56. Soranno, P. A. et al. LAGOS-NE: A multi-scaled geospatial and temporal database of lake ecological context and water quality for
+thousands of US lakes. GigaScience 6, gix101 (2017).
+57. USEPA. National Lakes assessment 2012: a collaborative survey of lakes in the United States. EPA 841-R-16-113. (2016).
+58. USEPA. National Lakes Assessment 2012 (nla2012_waterchem_wide.csv, nla2012_wide_siteinfo_08232 016.csv, and corresponding
+metadata fles). (2016).
+59. EEA. Waterbase: European Environment Agency (EEA) water quality data. https://www.eea.europa.eu/data-and-maps/data/water
+base-water-quality-2 (2020).
+60. Hoellein, T. J., Bruesewitz, D. A. & Richardson, D. C. Revisiting Odum (1956): A synthesis of aquatic ecosystem metabolism.
+Limnol. Oceanogr. 58, 2089–2100 (2013).
+61. Hornbach, D. J., Schilling, E. G. & Kundel, H. Ecosystem metabolism in small ponds: Te efects of foating-leaved macrophytes.
+Water 12, 1458 (2020).
+62. Muggeo, V. M. R. segmented: An R package to ft regression models with broken-line relationships. R News 8, 7 (2008).
+63. Burnham, K. P., Anderson, D. R. & Huyvaert, K. P. AIC model selection and multimodel inference in behavioral ecology: Some
+background, observations, and comparisons. Behav. Ecol. Sociobiol. 65, 23–35 (2011).
+Acknowledgements
+Tis work was supported by the Global Lake Ecological Observatory Network (GLEON). We thank participants
+in the GLEON 19 pond ad hoc meeting for discussions that formed the basis for this project. MRA was supported
+as part of the BEYOND 2020 project (grant-aid agreement no. PBA/FS/16/02) by the Marine Institute and funded
+under the Marine Research Programme by the Irish Government. Funding to KSC and KBSK was from the
+US National Science Foundation (EF-1638679 and EF-1638539). KKH was supported by the Adam S. Tomas
+Endowment and by the St. Olaf Collaborative Undergraduate Research and Inquiry Program. Tis material is
+13
+Vol.:(0123456789)
+Scientifc Reports | (2022) 12:10472 | https://doi.org/10.1038/s41598-022-14569-0
+www.nature.com/scientificreports/
+based upon work supported by the National Science Foundation Graduate Research Fellowship Program under
+Grant No. 2036201 to MJF (Farruggia). Any opinions, fndings, and conclusions or recommendations expressed
+in this material are those of the authors and do not necessarily refect the views of the National Science Foundation. Credit for Fig. 8 to Fiona Martin at Visualizing Science (https://www.visualizingscience.com/).
+Author contributions
+D.C.R. and M.A.H. led the project. D.C.R., M.A.H., M.J.F., K.K.H., and K.B.S.K. organized the literature searching, survey, data analysis, fgure preparation, and wrote the manuscript. All authors completed data mining from
+literature survey and were substantively involved in two of the following three categories: (1) project conceptual development, (2) data extraction/analysis/interpretation, and (3) writing/revising manuscript. All authors
+approved the fnal version of this manuscript prior to submission.
+Competing interests
+Te authors declare no competing interests.
+Additional information
+Supplementary Information Te online version contains supplementary material available at https://doi.org/
+10.1038/s41598-022-14569-0.
+Correspondence and requests for materials should be addressed to D.C.R.
+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 afliations.
+Open Access Tis 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. Te 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/.
+© Te Author(s) 2022
 """
-st.write(summarizer(ARTICLE, max_length=130, min_length=30, do_sample=False))
+st.write(summarizer(ARTICLE, max_length=6000, min_length=30, do_sample=False))