Water quality, vegetation, and management of stormwater ponds draining three distinct urban land uses in central Florida

Stormwater ponds are prevalent engineered water features that are designed to mitigate the effects of urban land use on downstream water bodies. These systems contribute significantly to the total area of aquatic ecosystems in some urban watersheds, making them important integrative features for examining human impacts on water ecosystem services. We investigated the distribution of stormwater ponds in relation to different land uses and examined dissolved nutrients in pond water, pond vegetation, and pond management practices in stormwater ponds receiving runoff from distinct urban lands uses in a rapidly developing suburban watershed of Orlando, FL, USA. Stormwater ponds represented 40.2% of the total area of non-forested freshwater systems in the watershed and were dominated by ponds situated in residential (43.7%), followed by roadways (14.7%), and institutional (2.3%) land uses. We randomly selected 8 ponds receiving runoff from each of these three lands uses, using schools to represent the institutional land use, and expressways to represent roadways, and sampled them monthly from May-Sept. 2015 for nutrient and physical characteristics and assessed their general cover of emergent and submerged vegetation. Total nitrogen (TN) concentrations differed significantly between expressway and school ponds, with expressway ponds having TN concentrations 51.7% higher than schools. Both total phosphorus (TP) and TN varied differently through time in the different lands uses and concentrations of most dissolved nutrients separated out from pond vegetative cover in a principal component analysis. Ponds with higher TN and chlorophyll a (chla) concentrations had lower water clarity. Management intensity for removal of aquatic vegetation and algae was lower in school ponds than in expressway and residential ponds, and school ponds contained the highest abundance and diversity of vegetation. Different urban land uses had varying impacts on water quality, and more intense chemical use to control vegetation and algae was related to greater nutrient and chla concentrations and lower water clarity. Our results indicate that there are important differences in nutrient concentration in stormwater ponds in the different land uses that may be related both to differences in surface runoff, and in management practices use to control overgrowth of vegetation and algae.


Introduction
Urbanization is one of the strongest anthropogenic threats to environmental quality (Groom et al. 2006;Paul and Meyer 2001;Yan and Edwards 2013), and urban development significantly alters the hydrology, flow, and quality of rivers, lakes, and streams (Adler and Tanner 2013). These alterations strongly influence the biological integrity of urban aquatic systems, impair their aesthetic and recreational uses (Booth and Jackson 1997;Chocat et al. 2001), and create serious implications for the quantity of fresh, clean water for both humans and wildlife (Arnold and Gibbons 1996). Many engineered urban stormwater systems are implemented into planning and policy at the national, regional, state, and local levels and are designed to mitigate the negative influences of urban development on the environmental and ecological quality of downstream ecosystems. (Chocat et al. 2001;Hogan and Walbridge 2007;Olding 2000; Thompson and Fryxell 2007).
One prevalent stormwater best management practice (BMP) is construction of stormwater basins that control sedimentation and flooding from runoff, and can enhance water quality, before releasing the runoff into adjoining water bodies (Anderson et al. 2002;Behera and Teegavarupa 2015;Bolund and Hunhammer 1999;Gallagher et al. 2011;Hogan and Walbridge 2007;Grimm et al. 2008). In regions with high water tables, these basins can be permanent ponds that can create extensive novel habitat for aquatic and terrestrial wildlife (Beckingham et al. 2019;Gallagher et al. 2011;Marsalek et al. 2005;McKinney et al. 2011). Additionally, stormwater basins can provide recreational uses, such as boating or fishing, and enhance aesthetics and increase property value in residential areas (Drescher et al. 2011).
The potential for stormwater ponds to detain nutrients and pollutants in runoff, such as phosphorous, nitrogen, pesticides, contaminated sediments, and metals, may cause unfavorable conditions for resident flora and fauna and degrade surface waters (Arnold and Gibbons 1996;Bishop et al. 2000;Chocat et al. 2001;Gallagher et al. 2011;Paul and Meyer 2001;Walsh 2000;Walsh et al. 2005). Although stormwater basins function as habitat for a variety of species, they are not considered natural freshwater ecosystems, and are not always subject to water quality standards. For example, in Florida, governments at the federal, state, and local level have adopted nutrient and pollutant criteria for most streams, spring vents, lakes, and selected estuaries (Florida Department of Environmental Protection 2013), but stormwater ponds are excluded from these criteria, even though they are the de facto primary receiving water bodies in many engineered stormwater systems.
Stormwater basins are often designed with dual function of reducing runoff during storm events and providing nutrient retention. In Florida, they are designed to provide a presumption of reducing at least 80% of the mean annual load of pollutants that violate the State Water Quality Standards, and the rate is 95% reduction if a stormwater system discharges into an Outstanding Florida Water (OFW) (Harper and Baker 2007). Despite this presumption of compliance, reported levels of removal for stormwater ponds in Florida range from 4-63% for N and 39-93% for P. There is little evidence that stormwater ponds in Florida or other areas of the southeastern US are coming anywhere close to meeting presumed nutrient removal performance, or are significantly reducing nutrient impact on downstream waterways (Harper and Baker 2007;Beckingham et al. 2019;Hess et al. 2022).
Considering that stormwater basins are a dominant aquatic feature in many urban landscapes but are not well studied, there is a need for more information on their characteristics and their relationship to different urban land use and management regimes. Additionally, because stormwater basins are the initial receiving system for urban runoff and have a close association with specific urban land use classes (Levia and Page 2000), they can provide important information about the influence of different urban land-uses on stormwater nutrients and pollutants (Kaushal and Belt 2012). Most studies investigating the influence of urbanization and development on stormwater basins are intensive and small scale and focus on one or a few basins (Janke et al. 2014;Marsalek et al. 2002), and there is a need for more information on variability among ponds in different land uses at the landscape scale.
In this study we investigated how ecological conditions varied among stormwater ponds receiving drainage from three different types of urban land use, and to investigate how management of the ponds and general characteristics of the surrounding landscape might contribute to those conditions. The specific objectives were to: 1) quantify the distribution of stormwater basins in a rapidly developing urban/suburban watershed, 2) assess the physical, chemical, and ecological characteristics of the stormwater pond basins based on surrounding urban land-use and land cover, and 3) examine differences in stormwater management practices that might contribute to different ecological outcomes in ponds in the different land uses.

Site selection
This study was done in the 72,520-ha Econlockhatchee River watershed, a mixed suburban watershed located east of Orlando, Florida, USA (Fig. 1). Geographic coverages of land use/land cover in the watershed were acquired from the St. John's River Water Management District geospatial open database (St. John's River Water Management District 2014). Stormwater basins have outfalls into tributaries, but they are not required to adhere to freshwater pollution guidelines, even though they are the dominant surface water features in this landscape (Table 1). We visually inspected aerial photographs of 1,587 features designated as reservoirs or ponds in the GIS coverage and eliminated features not considered stormwater ponds (e.g. farm ponds, borrow pits), resulting in an estimated 1,368 stormwater detention ponds. To determine the dominant land use surrounding individual stormwater ponds, we made a near table with a 100-m radius to select ponds that would not overlap when clipped with a 50-m buffer (ArcGIS 10). The procedure produced a group of 663 ponds, which were then categorized according to the surrounding land-uses within a 50-m buffer around the ponds. We chose a 50-m buffer because stormwater ponds in the study area are typically built to fulfill regulatory permitting requirements associated with construction activities occurring within particular land covers/uses. The land cover or land use in close proximity to the pond is thus a good indicator for the specific land use that drains into the pond. For example, stormwater ponds constructed in residential neighborhoods are nearly always surrounded immediately by residential land cover, which is the sole land use that contributes to piped drainage into the pond, as is the case for commercial or institutional parcels. In the case of roadways, there is greater chance that ponds near to roads may receive drainage from different land uses, but most ponds designed to receive drainage from roadways are situated alongside the road they drain such that road land uses make up a good portion of the near buffer. The buffer areas of the subsampled ponds contained eleven different land-use groups dominated by urban land use covers (63.4%). Residential land use was by far the most dominant (43.7%) followed by roadways (14.7%) commercial and services (2.7%) and institutional (2.3%).
We randomly selected study sites from three of the most dominant land uses in the watershed (residential, roadway,  and institutional). Within those land uses we selected specific types of entities that would represent distinct but consistent differences in land use type and pond management. For the institutional land use, we selected ponds surrounded by K-12 schools, all managed by the local Orange County Public school system; for the roadway land use, we selected ponds managed by the Orlando Orange County Expressway Authority (OOCEA); and for the residential land use, we chose ponds managed within homeowners' associations (HOAs) which are a dominant form of residential land use in the region. We refer to the land use for these selected ponds as "school," "expressway," and "residential" in our analysis, as these are the specific categories chosen to represent the general land use type. We did not include the commercial and services category because it was similar in coverage to institutional, and because it was more difficult to gain permission to access the commercial sites within the time frame of this study. We randomly selected eight ponds of similar surface area (0.4 to 2.0 hectares) from each of the three landuse classes (N = 24) using the subset features (Geostatistical Analysis) tool in ArcGIS, avoiding extremes of the size distribution ( Fig. 1). Residential ponds were blocked into four, geographic zones (North, South, East and West) with two ponds per zone. We randomly selected eight ponds from the 63 public schools containing stormwater ponds, using a similar blocking method. Due to distribution limitations of ponds associated with the expressway, we created two blocks (North and South) and randomly selected four ponds in each block (N = 8). Where available we obtained pond construction specifications from the St. John's River Water Management District permitting site (St. Johns River Water Management District 2016) to determine age and surface area of each pond, if available. There was no significant difference among land-uses in size of ponds included in the study (one-way ANOVA.F 2,21 = 0.52, p > 0.05). However, mean pond age varied significantly among land-uses (one-way ANOVA.F 2,18 = 5.08, p < 0.05), with school ponds being younger than expressway ponds. In each case, site visits were performed prior to final selections to verify that all the piped drainage into the selected ponds came exclusively from drains within the selected land use, to ensure that the piped storm inflows were not from mixed sources.

Limnological measures
We sampled limnological parameters five times from May to September, 2015, during the end of central Florida's dry season to near the end of the wet season. Physical parameters included temperature, pH, and conductivity (YSI Model 63 Handheld pH, Conductivity, Salinity and Temperature System, YSI Inc., Yellow Springs Ohio); dissolved oxygen (YSI Environmental ProODO); and Secchi depth. We measured all physical parameters at the center of the pond. Field procedures followed standard protocols used by the FDEP (Florida Department of Environmental Protection 2017).
We collected a 125-ml filtered water sample and a 1-L unfiltered, water sample at the center and near the outlet and inlet of each pond (United States Environmental Protection Agency 2012). Filtered samples were for dissolved reactive phosphorus (DRP) which was analyzed within 48 h of sampling. Unfiltered samples were acidified with sulphuric acid (H 2 SO 4 ) to below pH 2 and stored it in a refrigerator until analysis. During June through September sampling sessions, we took an additional 1-L unfiltered sample at the center of the pond to measure chlorophyll a. This second unfiltered sample was vacuum-filtered (250-mL) on a glass fiber filter, which was extracted with 90% ethanol to determine chlorophyll a on the day of sampling (United States Environmental Protection Agency 2012). Samples were analyzed for DRP (D'Angelo et al. 2001), nitrate (NO 3 ) (Schnetger and Lehners 2014), and ammonium (NH 4 ) (Sims et al. 1995) and after persulfate digestion (Johnes and Heathwaite 1992), were analyzed for total nitrogen (TN) and total phosphorous (TP). All analyses used standard colorimetric methods on an Epoch microplate spectrophotometer (Biotek Instruments, Inc, Winooski, Vermont).

Pond vegetation
We measured ecological characteristics of pond vegetation at the end of the study (October 2015) at randomly selected 10-m shore lengths centered on points generated in ArcGIS along each pond perimeter. The number of sites per pond was based on sampling approximately 15% of the pond's perimeter, with minimum distances set between points to ensure sampling locations were distributed around the pond. We measured littoral zone width at three randomly selected three points along the 10-m shore length by measuring the perpendicular distance from the shore to the furthest edge of both emergent and submergent littoral vegetation. We determined the percentage coverage and species composition (submergent and emergent) of dominant plants along a 1 m-wide swath centered on the measuring tape at each location. In addition, we recorded upland vegetation 3 m shoreward of the water's edge and recorded algal mat cover in the pond visually from the shore in four subjectively determined partitions for each pond. Vegetation was identified to species when possible, using Cho et al. (2014) and Tobe et al. (1998).

Pond management
To determine management practices in the ponds, we developed a standard questionnaire about aquatic vegetation and algal control practices (e.g., herbicide, mechanical), chemical usage and rates, and frequency of application. After receiving appropriate institutional approval, the questionnaire to pond managers as determined through homeowners' associations (HOAs) for residential areas, and land and facilities managers to determine who was responsible for pond management. Based on the responses, we developed a subjective pond management intensity index scaling from 1 to 5 that considered whether chemical controls were used, how much chemical was applied, and how frequently controls were applied. Ponds managed frequently (i.e., every 2-6 weeks) with chemicals received a rating between 4 and 5, moderately managed ponds with chemicals a rating between 2 and 3, and ponds infrequently managed with chemicals a rating between 0 and 1.

Statistical methods
We tested the null hypothesis that water quality parameters and vegetation did not differ among land uses. The first analysis explored the correlation (α = 0.05) and variance among the twelve limnological variables using Principal Component Analysis (PCA). The variables included the overall mean within each pond for pH, temperature, conductivity, dissolved oxygen, Secchi depth, DRP, total nitrogen, nitrate, ammonium, chlorophyll a, total phosphorus and total vegetative cover. We used one-way ANOVA on the Principal Component scores to determine whether ponds differed significantly among land uses (α = 0.05). We then used Repeated Measures Full Factorial Design (Mixed-Model) to analyze the limnological parameters most descriptive of axis 1 and used a Tukey-Kramer HSD post-hoc test to test pairwise comparisons, adjusting the alpha value to α = 0.01. We used a one-way ANOVA (α = 0.05) to analyze the littoral zone vegetation data. The independent variable was land use, and the dependent variables were mean distance of emergent vegetation from shore, mean distance of submergent vegetation from the shore, mean percent cover of emergent vegetation, mean percent cover of submergent vegetation, and total cover (m 2 m −1 of shoreline). Pairwise comparisons were analyzed using Tukey-Kramer HSD post-hoc test. All analyses were conducted using JMP Pro 11 software (SAS Institute, Inc., Cary, NC, USA).

Principal components analysis (PCA)
Several limnological variables used in the PCA were significantly correlated at alpha = 0.05 (Table 2). Dissolved oxygen was positively correlated with pH and chlorophyll a. Total nitrogen was positively correlated with pH. Secchi depth was negatively correlated with total nitrogen and chlorophyll a. Chorophyll a was positively correlated with total phosphorus and total nitrogen. Total phosphorus was positively correlated with DRP. Total vegetative cover was positively correlated with Secchi depth. The three remaining variables were independent of the others (Table 2). Average values of limnological variables across all dates are shown in Table 3.
The Principal Components Analysis showed clear relationships among some suites of limnological parameters (Fig. 2). The first four principal components all had eigen vectors greater than one and represented 74.45% of the variance in the  data, so we retained these axes for further analysis. Principal Component axis 1 represents a nutrient gradient with a positive loading for total nitrogen, dissolved oxygen, pH, DRP, total phosphorus, ammonium, and chlorophyll a and a negatively loading for Secchi depth and total vegetative cover loaded. Nutrients loaded on the other axes as well with PC 2 having a negative loading for TP, DRP, ammonium, and nitrate, and positive loading for temperature, pH, total vegetative cover, PC 3 loading positively for nitrate, and PC4 loading positively for DRP and total vegetative cover.
A one-way ANOVA comparing the Principal Component scores from all four axes for individual ponds revealed significant separation between expressway ponds and school ponds (F 2,21 = 9.67, p < 0.05) only with Principal Component 1 (Fig. 2). Based on the PCA results and significant differences between ponds from different land uses along axis 1, repeated measures analysis was used to analyze the parameters most strongly related to this axis: total nitrogen, total phosphorus, dissolved oxygen, and Secchi depth. TN was chosen over chlorophyll a because they are highly correlated, and TN includes the nitrogen within chlorophyll a. The α value was adjusted to 0.01 to account for multiple tests.

Repeated-measures analysis of limnological parameters
Two water quality measures differed significantly among land uses: TN and Secchi depth (Table 4). Total nitrogen concentrations were 51.7% higher in expressway ponds than in school ponds (Fig. 3). Secchi depth differed significantly between expressway and school ponds, with expressway ponds being 52.4% lower than school ponds (Fig. 4). For both TN and Secchi depth, residential ponds had values intermediate between school and expressway ponds. The other parameters did not differ significantly among land uses. The TN concentrations differed significantly over time (Table 4). For both TN and TP there was a significant interaction between land use and time (Table 3). Total N declined slightly over time in expressway ponds but increased in July and September in the residential and school ponds,  (Fig. 5). Total phosphorus in school ponds increased over time then decreased in September. In contrast, TP in expressway ponds decreased in May and July, increased in August, then decreased in September. In residential ponds, TP increased in June, decreased in July and August then increased in September (Fig. 5).

Differences among lands use in intensity of pond management
Based on self-reported vegetation management concerns and treatments, expressway and residential ponds were managed at similar high intensity and school ponds were managed at lower intensity (Table 5). A common management concern for all land uses was vegetation overgrowth. Both chemical and mechanical treatments were applied to control the overgrowth. The Central Florida Expressway Authority hired private contractors to manage expressway ponds using chemical means. Emergent vegetation was chemically treated eight times per year with Rodeo (glyphosate), Hydrothol 191 (Mono (N, N-dimethylalkylamine) salt of endothall), and Diquat (diquat dibromide), and submerged vegetation was chemically treated three times per year with Fluridone (1-methyl-3-phenyl-5-3-(trifluoromethyl) phenyl|-41H|pyridinone). Managers at the Expressway Authority stated that they were preparing to introduce grass carp (Ctenopharyngodon idella) for vegetation control. Six of the eight residential ponds were managed by the Stormwater Management Division, Orange County Public Works Department through chemical means. Rodeo and Aquatol K (Dipotassium salt of endothall) were used every 4-6 weeks for vegetation, and copper sulfate was used as needed for algae. Two residential ponds were managed by the neighborhood homeowners' association, which hired private and public companies to manage the ponds chemically every 2-4 weeks. The Orange County Public School's Grounds Department maintained school ponds down to the water's edge mechanically and with Round Up (glyphosate). The schools stated that they may hire outside companies for aquatic herbicide treatment on an as needed basis, but much less frequently than for residential and expressway ponds.

Prevalance of stormwater basins with the watershed
The aerial extent of stormwater basins in the Econlockhatchee watershed exceeded that of natural lakes, indicating they contribute substantially to open water habitat in this region ( Table 1). The dominance of residential land use surrounding the ponds was expected because stormwater ponds are built primarily to mitigate the impacts of impervious areas associated with urbanization (Hogan and Walbridge 2007), and residential development is a major land use in the watershed. Stormwater ponds were rare in the more rural fringes of the watershed, where they were associated with major roadways.

Differences in water quality measures among land-uses
The higher nutrient concentrations and lower water clarity of expressway ponds, followed closely by residential ponds may be related to the fact that expressway ponds have more impervious surfaces and traffic flow than residential and school areas. This higher impervious area can lead to higher nutrient loads, and more runoff of road surface materials, sediments, and atmospheric deposition, and runoff from motor vehicle operations (Harper and Baker 2007). Areas drained by residential ponds generally only have 20-40% impervious surfaces and receive nutrient inputs from fertilizer and pet excrement, as well as atmospheric deposition residential roads and other pavement, and organic matter in the form of leaves, grass clippings and other detritus (Carpenter et al. 2014;Lusk et al. 2020;Krimsky et al. 2021;Toor 2016, 2017). These inputs are not unique to residential area and may also occur in the other land uses investigated in this study. Schools receive most of their runoff from parking lots with limited traffic flow and occasional landscape drains (Brattebo and Booth 2003), and fertilization at schools is minimal except on some athletic fields. It is possible that the more limited fertilizer use at schools contributed to the low TN and chlorophyll a and high clarity in their associated ponds compared to expressway and residential ponds, which overall represent more intensive lands uses with more intensive pond management than the school ponds examined in this study.
It is unclear why temporal patterns in certain water quality parameters differed markedly among land uses. Seasonal trends in rainfall likely had some influence on stormwater constituents due to dilution or flushing effects. For example, TN levels in expressway ponds were elevated even in the dry season, when nutrient levels overall in the other land uses were at their lowest, but TN concentrations in these ponds dropped in the rainy season, which may have been a dilution effect due to increased volume and flow of water into the ponds. By contrast, TN levels in residential and school ponds were low in the dry season and increased later in the wet season, possibly due to more nutrients flushing into the ponds with increased runoff in the rainy season. The decrease of conductivity through time among all land-uses is indicative of a dilution effect with increased rainfall throughout the rainy season (Kantrowitz and Woodham 1995). Secchi depth may also be affected by rainfall patterns, as water clarity can depend upon weather (Nellis et al. 1997). School ponds had consistently high Secchi depths, whereas expressway ponds remained consistently low. Secchi depths in residential ponds decreased during the July sampling event and remained at that lower level through September. Nutrient levels in residential ponds had the opposite pattern, and the increase in nutrients during the wet season in residential ponds was positively related to the decline in water clarity in those ponds, possibly due to proliferation of algae and increase in other suspended solids associated with runoff in that land use.

Relationships among limnological measures
The relationship between TN, TP, and chlorophyll a is consistent with known influence of these two nutrients on algal growth, although the relationship between chlorophyll a and these factors depends upon many factors including  Table 5 Management rating of pond maintenance concerns for each land-use and an overall subjective intensity index based on selfreported levels of treatment. N = number of managers maintaining ponds within each land use. Ratings are on a scale from 1 to 5 (1 being least concern or lowest intensity and 5 being of most concern or highest intensity) *Residential pond ratings are based on the mean of three different management companies; expressway and school ponds are based on a single management entity the trophic state of the system (Chin 2015). Increased chlorophyll a can reduce light penetration and negatively influence water clarity, and this strong inverse relationship between chlorophyll a and Secchi depth is well documented in the literature (Borkma and Smayda 2016;Hoyer et al. 2002). The positive correlation between dissolved oxygen and pH was expected because as dissolved oxygen increases with photosynthesis, carbon dioxide decreases, which can increase pH. The TN, TP, and DRP concentrations observed in this study are lower than the mean values as reported and reviewed by and Harper and Baker (2007) for stormwater ponds in Florida and indicate that the ponds included in this study are not eutrophic. Although some data are available for TN and TP in different land uses, there is limited data on nitrate, ammonium, and ortho-phosphate levels (Harper and Baker 2007).

Pond vegetation
The much greater abundance of submergent and emergent vegetation in school ponds relative to the other two land uses is consistent with lower management intensity in school ponds. Increased vegetation may have contributed to the higher water clarity in school ponds, because vegetation provides protection from non-point source pollutants, and improves water quality directly through chemical uptake and indirectly supplying chemically active organic matter (i.e., leaf detritus), and by modifying water movement and increasing soil stabilization (Dosskey et al. 2010). Abundant submergent and emergent vegetation can absorb nutrients within the pond to suppress phytoplankton growth and make the water clearer. Appropriate management of pond vegetation can effectively improve water quality, but this depends on land and water resource managers understanding vegetation composition and the previously discussed processes. Eutrophication can occur if these factors are not taken into consideration (Dosskey et al. 2010;Stoler and Relyea 2015).
Our results suggest that the lowest intensity of herbicide application to ponds is associated with most plant growth and diversity among these three different land uses.

Differences in management practices among land-uses
The primary concern motivating pond management practices across all land uses was removal of aquatic vegetation for aesthetics and function, but vegetation distribution and control may have direct impact on the physical and chemical characteristics of the ponds. Chemical controls were the primary management tool, but frequency of application differed among land uses. School ponds were the least intensely managed, contained the highest abundance and diversity of vegetation, and had the best water quality. Infrequent management of these ponds is likely due to lack of available funds relative to the other land uses and less concern of school district administrators over maintaining consistently cleared vegetation in the ponds. Less intensive management can increase vegetation, which aids in nutrient removal. However, as noted above public K-12 schools also likely have low nutrient inputs in runoff due to the lower intensity of management or use in the other land uses, especially chemical use in residential areas, and traffic on urban expressways.

Stormwater ponds as key managed elements in the urban landscape
Environmental degradation due to urbanization is a global concern (Yan and Edwards 2013), and the deterioration of water quality from runoff remains a significant challenge. Stormwater basins in the Econlockhatchee River Watershed were dominated by urban land uses and are a prominent feature of the urbanized areas within the watershed, suggesting that ponds contribute importantly not only to the extent of aquatic environments but also to water quality. Understanding and predicting the influence of land use on water quality is essential to developing policies, standards, and best management practices (BMP) to reduce the impact of urban runoff on downstream ecosystems. There has been progress towards identifying differences between developed land uses with respect to their impact on nonpoint pollution (e.g. Zivkovich and Mays 2018). However, there is a relative lack of information on the assessment and management of stormwater ponds as key elements in the aquatic landscape, and the degree to which they meet their expected performance for water quality and contribute to overall ecological status of urban waters. They are the primary receiving water body and are built based on permitting and engineering guidelines that include assumptions about nutrient performance criteria that may not be met. The ecological characteristics of these ponds, and how they can contribute to overall regional goals for conservation and restoration of aquatic systems are not well developed, and, at least in Florida, are not including in permit conditions. The main driver of management of these systems is to control vegetation and algae overgrowth, but that management is done haphazardly, mostly in reaction to current conditions or according to arbitrary management regimes, and management approaches vary significantly among land uses. Our analysis focused on very coarse-level measures of vegetative and algal communities typical of rapid assessments performed by management agencies. More detailed analysis of the ecological services provided by these systems and how they are affected by factors at the local and landscape scape is warranted.
In regions such Central Florida, where stormwater ponds are the dominant primary receiving water bodies in the built environment, these basins not only provide insight into the impact of different land uses on water quality, but also are a significant management interface between human engineered and natural downstream systems. In our study, the relationship between abundant submerged and emergent vegetation and better water quality in school ponds relative to other land uses, suggest the importance of littoral-zone vegetation in these systems. However the co-occurrence of these factors in our study do not necessary reflect a cause and effect relationship, and other factors related to differences in intensity of land use and pond management may have impacted differences among land sues in water and habitat quality in these ponds. Variation among land uses in seasonal patterns of nutrients indicate that there are differences in underlying mechanisms controlling nutrient dynamics in these systems that need to be further verified and elucidated.
The United States Environmental Protection Agency (USEPA) established state-specific water quality criteria for natural bodies of water in adherence to the Clean Water Act. The standards are representative of surface water conditions minimally influenced by humans and protective of aquatic life and recreational uses. Unfortunately, recommendations have not yet been established for stormwater ponds, even though, as we have shown in this study, such ponds can dominate aquatic habitats and primary receiving water bodies in some watersheds. Although the standards for natural lakes and wetlands are not necessarily applicable to stormwater ponds due to differing hydrology, landscape position, and management status, their critical position in the stormwater continuum suggests that establishing such standards for stormwater basins is warranted (Rooney et al. 2015).
The ponds in this study did not show signs of severe eutrophication or deterioration, but they are relatively young (10-25 years) and their water quality may change or decline over time depending upon how they are maintained or if nutrients accumulate or change over time. Improving he littoral zone by encouraging vegetative growth may contribute to better water quality (i.e., higher water clarity and lower nutrient concentrations), and is likely to provide other ecological benefits, such as habitat and foodweb support. Other options include designating buffer zones (at least 3.7 m away) where no fertilizer or chemicals can be sprayed on lawns, constructing permeable pavement where practical, and developing a comprehensive maintenance plan for the entire drainage basin that includes education and outreach to homeowners and management entities. Heavy dependence on chemical controls for algae and pond vegetation contributes to the chemicalization of the environment and more ecological approaches could reduce pesticide overuse.
Improving the ecological conditions of urban stormwater ponds is critical for preserving preservation and restoring biodiversity and maintaining sustainable habitat in urban areas (Tixier et al. 2011), as well as protecting human health. Our results suggest that different strategies may be required in different urban environments and that there is a need for more research and improved management approaches for these key landscape features.
Author contributions Both authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by Lindsay Skovira. The first draft of the manuscript was written by Lindsay Skovira and Patrick Bohlen provided extensive edits to the final and previous versions of the manuscript. Both authors read and approved the final manuscript.
Funding This work was supported by the University of Central Florida.
Data repository Summary data for pond water quality and vegetation cover data are available at https:// zenodo. org/ record/ 32418 08#. Y724N HbMK70.
Code availability Not applicable.

Declarations
Ethics approval Not applicable.

Consent to participate Not applicable.
Consent for publication Both authors have reviewed the final manuscript and consent to its publication if it is accepted.