Source or sink? Meta-analysis reveals diverging controls of phosphorus retention and release in restored and constructed wetlands

Wetland restoration is a popular nutrient management strategy for improving water quality in agricultural catchments. However, a wetland’s ability to retain phosphorus is highly variable and wetlands can sometimes be a source of phosphorus to downstream ecosystems. Here, we used a meta-analysis approach to explore the source and sink capacity of 139 wetlands for both total phosphorus (TP) and the more bioavailable form, phosphate (PO4 3−), at seasonal and annual timescales. Median retention efficiency across all studies is 32% for TP and 28% for PO4 3−, however the range is extremely broad. We found that wetlands are often sinks for TP (84% of site-years) and PO4 3− (75% of site years). The median TP retention within wetlands that are sinks (2.0 g·m−2·yr−1) is greater than release by wetlands that are sources (−0.5 g·m−2·yr−1). In contrast, for PO4 3−, median retention within wetlands that are phosphorus sinks (0.8 g·m−2·yr−1) is of similar magnitude to that released by wetlands that are phosphorus sources (−0.7 g·m−2·yr−1). We found that phosphorus release from wetlands coincides with higher hydraulic loading rates, lower influent phosphorus concentration, and legacy soil/sediment phosphorus. Phosphate releases were especially common in wetlands used for treating municipal wastewater, as well as restored and constructed wetlands with flashy, precipitation-driven flow. We found that experimental design may inherently bias our understanding of wetland performance for phosphorus retention as studies conducted in mesocosms outperform other wetland types. Analysis of monthly data demonstrated significant temporal variability in wetland phosphorus dynamics, often switching from retention to release many times within a year, but with no generalizable seasonal trends. Our results highlight the value of restoring wetlands for phosphorus retention and point to ways of furthering their utility towards improving water quality by simultaneously targeting retention enhancing measures and release avoidance.


Introduction
Phosphorus, in excess, is an environmental pollutant that enters surface waters from domestic wastewater, agricultural fertilizer use, and livestock operations (Smil 2000, Bennett et al 2001, Mekonnen and Hoekstra 2018. In phosphorus-limited systems, like most inland water bodies including the North American Great Lakes, excess phosphorus leads to eutrophication and associated water quality problems (Schindler 1977, Correll 1998. Even in coastal waters that are more nitrogen-limited, reducing phosphorus loads will likely play a key role in ameliorating nutrient driven water quality issues (Conley et al 2009, Paerl et al 2014. Restored and constructed wetlands are widely used to improve water quality through removal of nutrients, sediments, and other pollutants. While wetlands are generally effective at removing nitrogen (49.4 ± 25.4% removal efficiency) from surface water (Cheng and Basu 2017), their efficacy at retaining phosphorus is more variable (Land et al 2016) and wetlands can even be a source of phosphorus to downstream water (Audet et al 2020). This variability in phosphorus retention by wetlands challenges the use of wetlands as a nutrient reduction strategy. By further elucidating the drivers that underlie wetland phosphorus source-sink dynamics, we can take a more informed approach to wetland restoration.
Retention and release of phosphorus in wetlands is driven by a combination of physical and biologic controls, as well as design and management factors. Phosphorus enters a wetland in either dissolved or particulate form, with relative contributions of each form driven by the land use characteristics within the drainage basin (Meyer andLikens 1979, Manning et al 2020). Physical processes govern the transport of water and phosphorus into wetlands, as well as the physical processes of settling and resuspension of particulate phosphorus within wetlands (Ardón et al 2010, Jiang andMitsch 2020). Wetland inflow and outflow rates also determine the residence time of water within wetlands, which is an important control of both biotic and abiotic processes (Koskiaho et al 2003, Reinhardt et al 2005, Skinner 2022). Chemical equilibrium drives movement between sediment or soil bound phosphorus and phosphorus dissolved in pore water, while diffusion and bioturbation processes drive movement of dissolved phosphorus between porewater and the overlying water column (Dunne and Reddy 2005, Jarvie et al 2005, Kinsman-Costello et al 2014, Simpson et al 2021. Finally, biologic activity drives assimilation of phosphorus into plants and microorganisms (Reddy et al 1999) as well as the decomposition of organic material and subsequent release of organically bound phosphorus (Carstensen et al 2019). Incomplete decomposition of organic matter is a key process by which phosphorus is retained within wetlands, as partially decomposed material accumulates over time. Temperate wetlands accumulate organic matter at an average rate of 155 g C m −2 yr −1 (Loder and Finkelstein 2020) simultaneously accumulating phosphorus bound up in that organic material. These processes interact and exert influence on the rates of one another, but the relative importance of these processes is highly dependent on the magnitude and forms of phosphorus inputs. Other factors including redox conditions and plant and microbial community composition also influence a wetland's ability to retain phosphorus, and contribute to the variability of phosphorus retention rates observed (Walton et al 2020).
The restoration of wetlands in agricultural landscapes is widely used with the aim of improving water quality from excess nutrients and agrochemicals (Zedler 2003). However, a history of nutrient use in intensively farmed areas has led to an accumulation of legacy soil phosphorus (Sharpley et al 2013, Nair et al 2015, Wiegman et al 2022. Moreover, wetland drainage also facilitates the liberation of phosphorus from accumulated organic matter as it degrades (Zak et al 2018). These sources of legacy and labile phosphorus pose additional challenges for wetland restoration in some locations due to the risk of phosphorus release during the process of restoration and in the years following land conversion. To address this problem, soil assays and soil column flow through experiments have been used to examine the phosphorus release potential of wetland soils. This work demonstrates the role of equilibrium dynamics in governing the bi-directional exchange of phosphorus between wetland soils and the overlying water column (Aldous et al 2005, Dunne and Reddy 2005, Kinsman-Costello et al 2016, Wiegman et al 2022. Other soil characteristics such as pH and iron content dictate the binding potential and overall storage capacity for phosphorus in wetland substrates (Jensen et al 1992, Kinsman-Costello et al 2016, Hu et al 2022. These experiments suggest that under many conditions (i.e., high phosphorus content, low influent phosphorus concentrations, low phosphorus binding capacity) soils in restored wetlands will release phosphorus, leading to wetlands that act as a net source of phosphorus to the landscape.
However, soil characteristics and internal loading of phosphorus within wetlands are not the only controls of wetland phosphorus release. Even phosphorus that is released from soils can be taken up again by wetland plants and retained within the system by the accumulation of organic matter (Pant 2020). However, unlike nitrogen that can ultimately be removed from wetlands via conversion to and release as a gas (e.g. N 2 or N 2 O), the only 'removal' pathway for phosphorus is through sedimentation and burial (Reddy et al 1999, Dunne andReddy 2005). As such, this is not a permanent removal of phosphorus from the system, but rather storage that is potentially reversible. Changes in equilibrium status can stimulate the release of sorbed phosphorus and major hydrologic disturbances can remobilize phosphorus, even from buried wetland soils (Reddy et al 1999).
The combination of internal (e.g., phosphorus legacies) and external (e.g., storms) controls, necessitates whole ecosystem studies to assess overall wetland phosphorus retention potential. Whole wetland phosphorus budget studies are particularly valuable in the context of restored and constructed wetlands, as they help to inform design criteria for improving water quality outcomes. Most wholewetland studies, however, focus on a single wetland or a small group of wetlands in a common geographical context. Despite limited scope, these studies still demonstrate the considerable variability that exists in phosphorus retention both within a single wetland over time (Choate et al 1990) and between similar wetlands from the same study region (Audet et al 2020). This variability includes wetland phosphorus release, either intermittently or from systems that are persistent net sources of phosphorus to the downstream environment.
Despite instances of wetland phosphorus release, a meta-analysis of 67 studies showed that at annual scales, the majority of restored and constructed wetlands are net phosphorus sinks, with median total phosphorus (TP) retention of 46% (95% CI 37%-55%) (Land et al 2016). While Land et al (2016) examined a large number of studies, their analysis was limited to TP, which leaves an important gap in our understanding of how wetlands process and remove phosphorus in its various forms, notably, the more bioreactive form of phosphate (Schindler 1977, Correll 1998. A similar review by Cheng and Basu (2017) found that wetlands are more efficient at retaining TP than phosphate, however, their analysis was limited to only 20 studies with phosphate data and only examined wetlands that retained phosphorus, excluding those that are phosphorus sources. Finally, these previous efforts to synthesize wetland phosphorus retention across multiple studies only take into account net performance at the annual scale, without considering seasonal dynamics that may, in fact, be most relevant to downstream water quality and management decision making. Sub-annual data is necessary for assessing the synchronicityor lack thereof-of phosphorus retention in wetlands with spikes in phosphorus use and runoff, and the timing of seasonal algal blooms (Michalak et al 2013). Although a wetland can be a net phosphorus sink at the annual scale, releases of dissolved phosphate under certain seasonal conditions may disproportionately contribute to downstream water quality problems by triggering algal blooms. Therefore, two key pieces are missing from previous meta-analyses: sub-annual phosphorus retention-release dynamics and the relative retention of phosphate versus TP in wetlands. Here we address these two major gaps in meta-analysis of wetland phosphorus retention. First, we explore wetland phosphorus retention, speciation, and dynamics, and the differential entrapment of phosphate and TP. We also go beyond annual retention summaries to look at seasonal patterns of wetland phosphorus retention. Inference from seasonal data has the potential to be much more relevant to management recommendations as compared to annual data. Finally, we explore how retention dynamics of both forms of phosphorus vary as a function of wetland type, loading, and other characteristics at both annual and sub-annual timesteps. This meta-analysis of 50 studies yields a cross-section of wetland types and characteristics to shed new light on the controls of phosphorus retention in and release from restored and constructed wetlands that will further efforts to maximize water quality benefits through wetland restoration and creation.

Literature search
We conducted a systematic search of peer-reviewed publications to find studies of restored or constructed wetlands that report at least one year of wetland phosphorus retention for both TP and phosphate (PO 4 3− ). In this paper, wetland 'restoration' is broadly defined, including rewetting as well as more extensive efforts involving excavation and planting, though we acknowledge that true restoration to a predrainage state is not likely achieved due to ecological hysteresis (Hemes et al 2018). For inclusion in this meta-analysis, studies needed to report (1) solute load and flow at both the inlet and outlet of the wetland when flow is present and (2) sufficient information about the hydrology of the system to ensure a nutrient budget of reasonably high quality. In some cases, reported solute concentrations were used to calculate flow from load, or concentration was back-calculated from load and flow as needed to fill in missing data.
We relied on several previous meta-analyses to acquire appropriate studies including the North American Treatment Wetland Database (Knight et al 1994) (n = 8), (Cheng and Basu 2017) (n = 5) and (Land et al 2016) (n = 18). To acquire more recent studies, we followed the approach of Land et al (2016) for keywords (excluding nitrogen keywords) and quality assessment for inclusion/exclusion of studies. We queried web of science with the following terms: (wetland * OR pond OR mire * OR marsh OR fen OR 'wet meadow' OR riparian OR 'floodplain' OR reed) AND (construct * OR creat * OR restor * OR man * made OR flooding OR inundation) AND (phosph * ) AND (retention OR trap * OR uptake OR sedimentation OR remov * OR settling OR accretion OR precipitat * OR * sorption) (Land et al 2016).
Our search yielded 1632 unique records for the period of January 2014-January 2022 (we assumed all relevant papers prior to 2014 appeared in Land et al (2016)). An initial screening selected for studies that measured both TP and PO 4 3− for at least one year and narrowed the results to 173 studies. During full text screening, a further 154 studies were excluded for failing to meet our inclusion criteria or quality standard for water and nutrient budgets. In total, 19 additional studies were kept for a total of 50 studies. The most frequent reasons for exclusion during screenings were if a study reported a phosphorus budget based on concentration only, flow measured only at the inlet or outlet, or only reporting on one of the phosphorus species of interest.
The inclusion criteria of measured flow and solute load at both the wetland inlet and outlet necessitates the exclusion of wetland configurations for which inflow and outflow are not easily defined (e.g. riparian or coastal wetlands). For example, our study does not include riparian or floodplain wetlands, despite the significant role they play in phosphorus cycling (Walton et al 2020). Additionally, terminal wetlands that do not typically spill (i.e. in the prairie pothole region of the US and Canada) are likely also underrepresented in the literature. As such, our meta-analysis focuses on wetlands with a prototypical flow-through configuration, as this is the most common study design and most readily facilitates crosscomparison.
A note about nomenclature: literature studies report phosphate several ways including soluble reactive phosphorus, dissolved reactive phosphorus, and other terms (Skinner 2022). While acknowledging the modes of measuring each phosphorus form alters their specific reactivity, these forms are generally accepted as more bioavailable, therefore we group all of these terms together under the umbrella of phosphate (PO 4 3− ) for the sake of a more holistic interpretation of phosphorus in wetlands (Reynolds and Davies 2007).
Annual water and phosphorus budgets were extracted from each study, digitizing tables and plots where necessary (WebPlotDigitizer Version 4.5 (Rohatgi 2021)). For six studies, monthly data was also available and extracted amounting to monthly phosphorus budgets from 26 individual wetlands and 36 site-years. We also recorded meta-data including date, location, soil information, size, depth, residence time, vegetation, and wetland age, for each site as available.

Data analysis
Wetlands from the selected studies were sorted by type, catchment setting, and hydrologic regime. Wetland types include 'Natural' , 'Restored' , 'Constructed' , and 'Mesocosm' . Because the aim of this study was to look at restored/constructed wetlands, sites designated as 'Natural' are systems in which a naturally occurring wetland treats water that is artificially pumped in for the purpose of water quality improvements. In this way, the hydrology of the system has been artificially reconfigured, but the wetland itself is essentially in its natural state. The difference between 'Restored' and 'Constructed' , as we define it, depends on the previous land use. Restored wetlands are built on land that was historically wetland prior to development or agriculture use. Constructed wetlands are built on land with no history of wetland occurrence, as often is the case on the premises of wastewater treatment facilities for tertiary treatment. Not all studies explicitly stated a land use history prior to wetland construction, in which case a designation of constructed was used unless it had been specifically referred to as restored. Finally, for the purpose of this meta-analysis, a designation of 'Mesocosm' was given to studies conducted in tubs, tanks, or other semi-isolated structures. Mesocosms are useful for ecological investigations, capturing some, but not all variability that exists in the field, depending on their size and the complexity of their design. The inclusion of mesocosm studies was limited to only those constructed outdoors and with an area of 1 m 2 or larger, therefore more closely resembling true wetlands than mesocosms conducted in lab or greenhouse settings.
Wetland catchments are designated based on the predominant land use, as either agricultural, urban, or wetlands associated with waste-water treatment (WWT). The latter refers to wetlands receiving effluent from water treatment facilities (typically municipal wastewater, sometimes combined with stormwater) after primary (sometimes secondary and tertiary) treatment processing. Hydrologic regime is a description of the water flow at the inlet to the wetland (Land et al 2016). The inflow can either be 'Continuous' or 'Intermittent' . The latter refers to systems where the inflow dries up for periods (generally a week or longer, as this is the typical sampling interval). Intermittent wetlands include those experiencing summer dry periods, winter periods where the inlet is completely frozen, (but not those where winter sampling simply ceased due to wintry conditions) and those with managed interruptions in flow (i.e. for maintenance). Additionally, the inflow can either be 'Regulated' or 'Unregulated' . The majority of wetlands have unregulated inflows; however, regulation refers to systems (usually in the context of water treatment facilities) with managed inflows maintained at a consistent inflow rate. 'Unregulated' refers to any system where the flow is influenced by precipitation and thus does not have a consistent inflow rate. This yields a total of four possible hydrologic regimes: 'Continuous Regulated' , 'Continuous Unregulated' , 'Intermittent Regulated,' and 'Intermittent Unregulated.' There are some cases where studies lacked sufficient information to assign a hydrologic regime with confidence and were therefore given a designation of 'Not Specified. ' For ease of analysis, the nine wetland-catchment configurations represented in the study were regrouped into three major wetland types: wetlands associated with WWT, wetland mesocosm, and all other wetlands. The constructed, restored, and natural wetlands were merged given the difficulty of parsing site history in cases where it is not reported. Additionally, urban and agricultural catchment types were merged given the mix of land uses within some wetland catchments.
To assess the drivers of wetland phosphorus retention, a suite of wetland attributes was obtained from the literature including wetland size, age since restoration or construction, catchment area, influent phosphorus concentration and flow rate. Several additional parameters were derived from these attributes using the following approaches. The catchment area to wetland area ratio was calculated as: for sites where catchment area is reported (n = 99 of 273). Hydraulic loading rate (HLR), or the rate with which water enters a wetland, was either extracted directly from the text of each study or calculated as annual inflow divided by wetland surface area. The relative proportion of PO 4 3− to TP loads at the outlet versus the inlet allows us to determine if a wetland is magnifying PO 4 3− (relative to TP) or suppressing it via preferential PO 4 3− uptake. We refer to this metric as the PO 4 3− magnification ratio and it is calculated as: where a ratio greater than one indicates more PO 4 3− leaving the wetland than coming in relative to the amount of TP. We refer to this outcome as PO 4 3− magnification and the opposite outcome (PO 4 3− :TP ratio < 1) as PO 4 3− suppression. All data analyses were conducted in R-3.5.3 with 'ggplot2' , 'tidyverse' , 'cowplot' , ggpmisc' , and 'rstatix' packages (Wickham 2016, Wickham et

Results and discussion
Systematic literature review yielded 50 studies for meta-analysis comprising 139 individual wetlands monitored for at least one year and a total of 273 site-years of data (see supplemental table 1 for complete reference list). Global coverage is biased toward North America and Europe with very poor coverage in the Southern Hemisphere (figure 1(a)). Study duration is typically only one or two years and duration and number of sites within a study tend to tradeoff, meaning studies with more sites are typically shorter in duration and longer duration studies typically examine fewer sites (figure 1(b)). Wetlands in this meta-analysis are mainly embedded in agricultural catchments (76%) with the remaining wetlands used to treat secondary or tertiary wastewater effluent (18%) or located in urban catchments (6%) (figure 1(c)). Constructed wetlands are the most prevalent type in the meta-analysis (n = 64), followed by mesocosms (n = 39), restored wetlands (n = 33) and natural wetlands (n = 3). The dataset is well distributed over a range of wetland sizes from 1 m 2 to 18 000 000 m 2 (18 km 2 ) (figure 1(d)). The wetlands within the data set are on average very young, meaning recently restored or created. The majority of the data comes from sites that are between 0 and 3 years old (178 of 273, 65%) and only 17% of the data comes from study sites more than 5 years post restoration/construction.

Annual phosphorus retention dynamics
Annual wetland phosphorus loading and retention span wide ranges across the studies within this metaanalysis, including both positive and negative retention rates (table 1). Across all studies, the annual phosphorus retention rate in wetlands is slightly higher for TP than for PO 4 3− . The median TP retention rate across all site-year combinations (n = 273) is 1.4 g·m −2 ·yr −1 , which corresponds to a retention efficiency of 32%. The median PO 4 3− retention rate is 0.4 g·m −2 ·yr −1 and the retention efficiency is 28% (n = 273). The median annual TP retention rate from this meta-analysis is similar to the median of 1.2 g·m −2 ·yr −1 (n = 146) reported by Land et al (2016), however, our median TP retention efficiency is significantly lower than theirs of 46%. A possible source of bias in our study may be due to our focus on studies that measure multiple form of phosphorus (TP and PO 4 3− ), which is perhaps more common in the context of sites with known or suspected phosphorus retention problems.
Out of 273 wetland site-years, 16% have negative TP retention (i.e., TP sources) and 25% have negative PO 4 3− retention (figure 2(a) and table 1). This finding is similar to Land et al (2016) who reported 17 of 146 (12%) wetlands were TP sources. Despite the frequency with which restored and constructed wetlands release more phosphorus than they retain, it is important to examine the magnitudes of both phosphorus retention and phosphorus release. By grouping wetland site-years as either sources or sinks of phosphorus (table 1) we can assess their relative impacts on downstream phosphorus loads. The  The standard error (SE) is calculated as the standard deviation divided by the square root of n, where n is the number of wetland site-years. median magnitude of TP retention in sink wetlands is 2.0 g·m −2 ·yr −1 compared to −0.5 g·m −2 ·yr −1 in source wetlands. In the case of PO 4 3− , median retention in sink wetlands (0.8 g·m −2 ·yr −1 ) is comparable in magnitude to the PO 4 3− released by source wetlands (0.7 g·m −2 ·yr −1 ). Further examination of wetlands that release phosphate is warranted, to help identify management strategies that promote uptake of this form of phosphorus and better protect downstream water quality.
Further insight is gained from re-grouping wetland site-years based on their retention and release of TP and PO 4 3 −, independently. In 70% of site-years, wetlands are a sink for both forms of phosphorus, the remaining act as a source of one or both phosphorus species ( figure 2(b)). The least common pattern is for a system to be a sink for PO 4 3− and a source of TP (4.4%, lower right-hand quadrant in figure 2(C)). In 13.6% of site years, wetlands are a source of PO 4 3− and a sink for TP while 11.4% are a source of both. More prevalent release of PO 4 3− is concerning for downstream water quality given that this form of phosphate is more readily available for processes that can lead to eutrophication and harmful algal blooms (Correll 1998). For wetlands that function as a sink for both TP and PO 4 3− , retention efficiencies are fairly evenly distributed between 0%-100% (upper right-hand quadrant in figure 2(C)) and generally fall around the 1:1 line, with a slight bias toward better PO 4 3− retention efficiency. Preferential retention of PO 4 3− in wetlands may be explained by plant uptake during the growing season (Reddy et al 1999). In contrast, wetlands that are phosphorus sources tend to fall above the one-to-one line, meaning less efficient retention of phosphate compared to TP (red points in figure 2(C)).
Preferential retention or release of either form of phosphorus, leads to an altered PO 4 3− :TP ratio in the outflow, compared to the water being received by a wetland. There is particular concern about the magnification of PO 4 3− relative to TP, as this phenomenon could exacerbate the consequences of phosphorus pollution (Correll 1998). A PO 4 3− magnification ratio (equation (2)) greater than one indicates more PO 4 3− leaving a wetland than entering, relative to the amount of TP (i.e., PO 4 3− magnification). Across the entire dataset, the median PO 4 3− magnification ratio does not significantly differ from 1 (Wilcoxon test, p = 0.56), however, wetlands that are phosphorus sources (red symbols in figure 2(C)) exhibit a higher PO 4 3− magnification ratio (median = 1.3, interquartile range = 1.5) than those that are phosphorus sinks (median = 0.9, interquartile range = 0.3) and the difference between these groups is statistically significant (Mann-Whitney test, P < 0.0001) (figure 2(d)). That the overall median magnification ratio is close to one suggests that wetlands are not acting as significant hotspots for the magnification of phosphate.

Seasonal phosphorus retention dynamics
While the majority of studies (64%) in our metaanalysis report only annual phosphorus retention, seasonal patterns of phosphorus retention are perhaps more important for understanding the role that wetlands play in phosphorus driven water quality problems. Accessible monthly data is much sparser than annual nutrient budgets, with only six of the 50 studies reporting data (complete with solute concentration/load and flow rate at inlet and outlet) at monthly (or smaller) timesteps. Three of these studies were conducted in Denmark and the remaining studies are from Slovenia, the US and Canada. The Canadian study provides the most extensive dataset, covering eight wetlands for two years of monitoring. Despite the predominance of wetlands as net phosphorus sinks on an annual basis, 75% of wetlands with monthly data available are a TP source for at least one month during the period of study (64% for PO 4 3− ) and 39% are a source of TP for three consecutive months or more (25% for PO 4 3− ). There is a high level of variability in phosphorus retention from one month to the next, both within individual wetlands, and across studies (figure 3). Several wetlands switch between being sources and sinks of phosphorus multiple times throughout the year. On a mass retention basis, similar to annual phosphorus budgets, there is more TP retained on average in sink wetlands compared to that released by source wetlands (table 2, supplemental figure 1). However, PO 4 3− releases from source wetlands are on a similar order of magnitude as the mass retained by sink wetlands (table 2, supplemental figure 1).
No patterns or seasonal trends of wetland phosphorus retention emerge across all the monthly datasets (figure 3). Indeed, we see that wetlands may be a source of phosphorus to downstream environments in all seasons of the year (U-V in winter, C in spring, J-K in summer, Y-Z in fall). There appears to be some clustering of seasonal patterns of wetlands from within the same study region, however, diverging patterns between sites from the same study also present themselves (e.g., site M and N in figure 3). Wetlands tend to behave similarly from one year to the next, but this observation is limited by only having multi-year data for ten sites.
Monthly phosphate magnification ratios range from 0 to 8.9 and are consistent with annual observations. Another pattern that emerges at the monthly scale that is consistent with the annual snapshot is the tendency for PO 4 3− magnification to occur when wetlands are acting as a source of phosphorus (figure 3).

Wetland characteristics and drivers of phosphorus retention
We examined the relationships between annual phosphorus retention efficiency (%) and eight wetland characteristics including wetland type, hydrologic regime, influent concentrations of TP and PO 4 3− , wetland size, wetland age, HLR, and wetland to catchment area ratio (figure 4). To examine these characteristics across TP and PO 4 3− retention simultaneously, wetlands were first binned by retention of both species (figure 4(a)). Binning the data helps elucidate patterns of phosphorus retention, particularly the difference between wetlands that are phosphorus sources (left most bars in figure 4) and those that are phosphorus sinks (three bars on the right, partitioned by retention efficiency into bins of 0%-33%, 33%-67% and 67%-100%). Bins have been normalized to 100% for ease of visualization but note that 67%-100% retention is the smallest bin (n = 32, see the unscaled bar plots in supplemental figure 2).
Both wetland type and hydrologic regime influence phosphorus behavior in wetlands. We find phosphorus releases to be most common in restored/constructed wetlands ( figure 4(b)) and wetlands with unregulated (i.e. precipitation driven) inflow ( figure 4(c)). Conversely, phosphorus retention tends to be highest in mesocosm studies and wetlands with regulated flows, though the difference is only significant with respect to phosphate retention (figure 5). The range of observed phosphorus retention from unregulated flow systems is very large, particularly with respect to intermittent, unregulated flows (figures 5(b)-(d)). Mesocosm studies exhibit a much narrower retention range and very rarely exhibit phosphorus release. Across all hydrologic regimes, median TP retention in mesocosms (46.7%) is higher than that of the other wetland types (31.4% for restored/constructed and 31.5% for wastewater treatment wetlands). The enhanced performance of mesocosms is even more apparent for PO 4 3− where median retention is 59.5% compared to 28.6% for restored/constructed and 1.6% for wastewater treatment wetlands across all flow types. The spread of the data is much larger across restored/constructed wetlands and wastewater treatment wetlands and the latter are particularly poor at retaining phosphate (figures 5(c) and (d)).
Both wetland age and wetland size exhibit a 'U-shaped' relationship with phosphorus retention (figures 4(d) and (e)) where older and larger wetlands show up more in the outermost bars, whereas younger and smaller wetlands have lower retention, . Letters indicate unique wetlands and numbers indicate different data-years for the same wetland. Blank cells indicate no data reported for that month (e.g., flow drying up in the summer) and gray cells (NA) indicate phosphorus measured as zero or below detection at the inflow and outflow. (c) Monthly phosphate magnification ratio, where values less than one (shades of green) indicate greater PO4 3− uptake relative to TP and values greater than one (shades of purple) indicate PO4 3− magnification.
Here NA (gray) indicates a month when outflow TP or inflow TP or PO4 3− are zero and the ratio is therefore incalculable due to division by zero. The standard error (SE) is calculated as the standard deviation divided by the square root of n, where n is the number of wetland site-months. but are also less frequently phosphorus sources. This suggests that wetlands that are sources of phosphorus diverge from expected relationships between retention and the attributes that affect retention. A compelling illustration of this idea is shown with catchment to wetland area ratio (figure 4(f)). Here we see the highest phosphorus retention bin is dominated by wetlands with small catchments relative to their size, and retention decreases as catchment size increasesexcept for wetlands that release phosphorus. Wetlands that are a source of phosphorus are not characterized exclusively by large catchments, instead, it is apparent that a wetland may be a source of phosphorus with any given catchment to wetland area ratio.
To further illustrate this point, we can examine the relationship between phosphorus retention and catchment to wetland area ratio directly (figures 6(a) and (b)). By separating 'sink wetlands' from 'source wetlands,' we see a weak, but significant negative relationship between catchment to wetland area ratio and phosphorus retention for both TP (R 2 = 0.06, p = 0.032) and PO 4 3− (R 2 = 0.27, p < 0.001), but only for 'sink' wetlands. For wetlands that release either TP or PO 4 3− (shown in red in figures 6(a) and (b)), there is no relationship with catchment to wetland area ratio. Despite the lack of a relationship with respect to source wetlands, catchment size is a critical aspect of wetland design for maximizing retention efficiency, perhaps more so than wetland size alone (see supplemental figure 3).
Further evidence of the divergence between 'sink wetlands' and 'source wetlands' is apparent when examining influent phosphorus concentration. Lower influent phosphorus is associated with both greater phosphorus retention, and phosphorus release (figures 4(g), (h) and 6(c), (d)). That source wetlands have significantly lower influent phosphorus concentrations than sink wetlands (pairwise t-test, Bonferroni adjusted p TP = 0.004, p PO4 = 0.002) suggests the role of equilibrium dynamics contributing to wetland phosphorus release (Kinsman-Costello et al 2014, Badiou et al 2018. Furthermore, the lowest Figure 6. Drivers of total phosphorus (TP) and phosphate (PO4 3− ) retention in wetlands. Explanatory variables include (a), (b) catchment to watershed area ratio, (c) inflow TP concentration, (d) influent PO4 3concentration, and (e), (f) hydraulic loading rate (HLR). Percent retention of TP is the dependent variable in panels (a), (c), and (e), and percent retention of PO4 3− is the dependent variable in panels (b), (d), and (f). Blue points, lines, and equations represent site-years that are a sink for the respective phosphorus species, and red represents phosphorus sources (percent retention less than zero). Gray lines and equations model the relationship of all points, both source and sink. Only relationships with p < 0.05 are shown. retention efficiencies in source wetlands are observed when influent phosphorus concentrations are also low, but the trend is not maintained in sink wetlands (figures 6(c) and (d)).
Finally, we see that wetlands with smaller HLRs tend to have higher relative phosphorus retention ( figure 4(i)). There are weak, but significant inverse relationships between HLR (m·yr −1 ) and percent retention of both TP (R 2 = 0.07, p < 0.001) and PO 4 3− (R 2 = 0.11, p < 0.001) in sink wetlands, wherein lower HLR contributes to higher phosphorus retention (figures 6(c) and (d)). This finding is consistent with the relationship with catchment to wetland area ratio, as wetlands with proportionally larger catchments tend to experience greater hydraulic loading. These trends point to the governing role of hydrologic fluxes in phosphorus retention in wetlands.
Repeatedly, we see that 'source wetlands' diverge from the patterns exhibited by wetlands that retain phosphorus with respect to various wetland characteristics. We therefore assert that the processes driving phosphorus release act independently from those more generally associated with improving phosphorus retention. Not only does this framing help to explain the large amount of variability observed within this meta-analysis, but it is also useful for developing strategies for better phosphorus management in wetlands going forward.
We also examined drivers of phosphorus retention within the subset of studies for which monthly data was available. The finer temporal resolution of monthly data further highlights the role of hydrologic dynamics as a driver of phosphorus retention. There is a significant inverse relationship between monthly HLR and retention of TP (R 2 = 0.38, p < 0.001) and PO 4 3− (R 2 = 0.28, p < 0.001) (supplemental figure 4). Monthly data is better able to capture the influence of large rain events, thus these correlations are much stronger than those between annual HLR and retention of TP (R 2 = 0.05, p < 0.001) or PO 4 3− (R 2 = 0.03, p = 0.003) (figures 6(c) and (f)). In addition to better elucidating the role that weather and extreme events play in wetland phosphorus retention, monthly wetland phosphorus budgets are much more valuable than annually aggregated data for identifying seasonal patterns. For example, observations by Hoffmann et al (2012) in Denmark (Y and Z in figure 3), where seasonal wetland drying in the summer is followed by phosphorus release upon rewetting in the fall. Seasonal observations of phosphorus retention and release dynamics can help in identifying the mechanisms controlling these processes and potential management strategies, which we address in the next section.

Mechanisms governing phosphorus dynamics in wetlands and management recommendations
It is critical to develop a conceptual understanding of what drives phosphorus retention/release dynamics in wetlands, such that future restoration and construction of wetlands will achieve maximum benefits in terms of water quality improvements. Formal driver analyses, like our section 3.3, focus on easily measurable wetland attributes like wetland size, age, and loading rates. But not all controls of phosphorus cycling are so easily measured. Qualitative assessment of the papers within this meta-analysis reveals common underlying causes for phosphorus release that, in some cases, act independently of quantifiable wetland attributes (see supplemental table 2 for narrative descriptions from each case study). In this section, we enumerate the specific mechanisms underlying phosphorus dynamics in restored and constructed wetlands with particular attention to the distinction between mechanisms of phosphorus retention versus mechanisms of phosphorus release (figure 7). This framing helps to emphasize the idea that the relationships between drivers of phosphorus retention and release are not always reciprocal (e.g., small versus large wetlands) and is ultimately used to provide appropriate design and management recommendations for future wetland restoration efforts. This discussion of phosphorus retention/release mechanisms and management recommendations is organized with respect to transport (hydrology), pools (legacies and loading), transformation (biogeochemistry), and intrinsic wetland properties (age and design).

Wetland hydrology and morphometry
Flow regulation and wetland morphometry (size and catchment area) emerged as key controls of phosphorus retention. We found that wetlands with regulated flow regimes are stronger phosphorus sinks, while intermittent and unregulated flow in wetlands can contribute to larger phosphorus releases. Wetlands with regulated inflows are designed to achieve sufficient residence times for adequate phosphorus removal to occur via uptake and settling. Conversely, unregulated flow systems (driven mainly by precipitation), can have very variable residence times. In the context of wastewater treatment, we find evidence that regulating flow can improve phosphorus retention and avoid releases. In unregulated flow systems, large rainfall events can reduce the potential for settling and uptake and lead to poor phosphorus retention.
Extreme rain events are deemed one of the main causes of poor phosphorus retention, sometimes accounting for significant portions of annual wetland phosphorus export (Ardón et al 2010, Jiang andMitsch 2020). However the link between extreme rain events and phosphorus export is not apparent in all studies. Dunne et al (2015) explicitly rules out this driver based on evidence that removal efficiency for total suspended solids does not change with storm intensity. Both Jordan et al (2003) and Tanner and Sukias (2011) found that some, but not all, high flow events drive phosphorus release, indicating that there may yet be some unaccounted-for interaction effects. An example of such an interaction effect is antecedent conditions, where alternation between dry and wet cycles has been shown to promote metabolism of organic matter and the release of organic phosphorus. In these cases, continuous water flow and stable water levels favor phosphorus retention (Kieckbusch and Schrautzer 2007).
Wetland morphometry, including wetland size and catchment to wetland area ratio, influences inundation dynamics in wetlands, and thus impacts phosphorus retention. We found evidence of lower retention and phosphorus release at higher HLRs and catchment to wetland area ratios. This is likely because a wetland's water-holding capacity is often exceeded at higher HLRs, contributing to limited retention capacity. Controlling inflow rate and flow variability is important for optimizing phosphorus retention. Given that water management and flow regulation in wetlands is costly, natural flow regimes are much more common in restored and constructed wetlands, particularly those in agricultural landscapes. In natural systems, flow variability is not as easily controlled, but it is possible to design wetlands with water-holding capacity in mind (considering catchment size, slope, infiltration rates and tile drainage) to improve phosphorus retention during major runoff events.
One particular strategy to help achieve phosphorus load reductions at the landscape scale is promoting the restoration of geographically isolated wetlands, with small catchment to wetland area ratios and low HLRs, such that they never or rarely spill (Richardson andQian 1999, Cheng et al 2023). Another strategy for achieving maximum phosphorus retention is the use of varied flow configurations, particularly the combined use of surface and subsurface flowing wetlands (Beutel et al 2014, David et al 2022. Designs that incorporate a sedimentation basin can further help to dissipate flow energy from large storm events (Beutel et al 2014). Lastly, regional planning of wetland networks will achieve greater phosphorus retention at the landscape scale than uncoordinated efforts of single restoration projects.

Phosphorus legacies and loading
High flows can reduce phosphorus retention in wetlands, but only under certain conditions do they spur phosphorus releases. There must be a pool of phosphorus, either in decaying organic matter, or built up in wetland soils, in order for mechanism such as scouring and resuspension to drive out more phosphorus than is entering at any given time. While addressing hydrologic factors may help improve phosphorus retention, addressing wetlands that are either occasionally, or persistently sources of phosphorus to downstream environments requires attention to phosphorus pools. Large pools of sorbed phosphorus in wetland soils can arise as it accumulates over a wetland's lifespan, or from legacy phosphorus from previous land use (Nair et al 2015, Wiegman et al 2022. Sediment testing confirmed unusually high legacy phosphorus to be the root of phosphorus release from one of the eight small wetlands in the study by Page et al (2022), causing this wetland to be a source of phosphorus to downstream ecosystems nearly all year long (Site O, figure 3). Unfortunately, phosphorus levels in wetland soils are neither widely, nor consistently reported in the studies within this meta-analysis, making it difficult to develop generalizable relationships across scales.
Management strategies to address legacy phosphorus include removing local substrates during wetland construction (Zak et al 2018). Phosphorus retention can also be improved through the use of substrate amendments that increase phosphorus binding capacity and avoiding those that could potentially be a source of phosphorus themselves through leaching (Emsens et al 2017, Carstensen et al 2019. While some amendments such as limestone, alum, or phosphorus poor subsoils, improve substrate phosphorus binding capacity, others may not provide enough binding sites, resulting in a time lag between restoration and sediment accumulation sufficient for phosphorus sorption (Ballantine and Tanner 2010, Tanner and Sukias 2011, Díaz et al 2012. Finally, achieving an optimal ratio between iron and phosphorus within wetland substrates could improve phosphorus retention as suggested in several studies (Zak et al 2004, Mendes et al 2018; attaining an iron to phosphorus ratio of three or larger in anaerobic pore water is necessary to promote precipitation of phosphorus (Kieckbusch and Schrautzer 2007), though other suggest a ratio even larger (Hoffmann et al 2012).
In addition to legacy phosphorus stored within wetlands, we also find that a wetland's ability to retain phosphorus is dependent on the concentration of phosphorus in wetland inflows. There is evidence that low influent phosphorus concentrations can lead to phosphorus releases, while in wetlands that are phosphorus sinks, high concentrations lead to poorer retention. Phosphorus releases arise when equilibrium dynamics favor the movement of phosphorus from soils into overlying water (Herskowitz 1986, Kinsman-Costello et al 2016, Badiou et al 2018, a phenomenon that has also been observed in riverine and floodplain sediments (Jarvie et al 2005, Simpson et al 2021. However, as this mechanism operates only when influent concentrations are relatively low, the magnitudes of these phosphorus releases can also be quite small. This pathway may help explain why wetlands tend to release more PO 4 3− relative to TP (figure 2(C)), and why, in some cases, the magnitudes of these releases are relatively low. There are limited recommendations for managing phosphorus influent to wetlands beyond farm nutrient management strategies, and even this has limited efficacy due to fertilizer legacies (Hamilton 2012, Melland et al 2018.

Wetland biogeochemistry
Hydrology is an important driver of wetland phosphorus retention from a physical perspective (flow) but also a biogeochemical perspective in terms of controlling the redox status of wetland soils. Prolonged inundation drives oxygen depletion in wetlands and activates anaerobic metabolism of organic material (see the introduction of Carstensen et al (2019) for a full description of these pathways). Mineralization and release of organic phosphorus is widely attributed as the rationale for source wetlands (Díaz et al 2012, Carstensen et al 2019, Jiang and Mitsch 2020. A primary consequence of anaerobic metabolism on wetland phosphorus cycling is the release of iron-bound phosphorus, when iron is present and reduced as the preferential terminal electron acceptor (Herskowitz 1986, Mendes et al 2018. Other biogeochemical factors, such as soil type, and peat and iron content, are important controls of both anaerobic release potential and phosphorus sorption capacity. However, like wetland soil phosphorus content, these other soil characteristics and redox status are not widely measured or reported, leaving a gap in our understanding of the relative importance of these processes governing phosphorus source/sink dynamics.
As with redox status, water depth in wetlands plays an important role in controlling phosphorus cycling by moderating temperature and plant establishment (Zak and McInnes 2022). Uptake of dissolved phosphorus, either externally or internally loaded, prevents it from leaving the wetland and impacting downstream environments. Temperature plays a role in governing this internal phosphorus uptake by mediating plant growth and microbial activity rates (Jordan et al 2003). Summer is typically a period of relatively high treatment efficiency as a result of this uptake (Herskowitz 1986). However, some studies observe poorer retention (Beutel et al 2014, Dunne et al 2015 or net release (Toet et al 2005) during the growing season, which is partially attributed to warmer temperatures speeding the decay of organic matter. It is suggested that with wellestablished vegetation, phosphorus uptake by plants and microbes can be sufficient to overcome decomposition releases of organic phosphorus (Herskowitz 1986, Jiang andMitsch 2020). Not only does a well vegetated wetland promote direct phosphorus uptake, but it also provides a physical structure for slowing flow and trapping sediment and particulate organic matter. However, some organisms promote phosphorus release; macroinvertebrates, muskrats, invasive carp and other organisms can drive phosphorus release through bioturbation and disturbance of wetland substrates (Barten 1987, Beutel et al 2014. Temperature also governs the seasonal cycles of wetland vegetation and other land use patterns. Two studies report minimum removal rates in the winter and early spring, suggesting a mismatch between winter loading (fall fertilizer application) and reduced plant uptake during winter dormancy (Kovacic et al 2000, Koskiaho et al 2003. Temperature also interacts with hydrologic processes, such as the case when early spring rains flush out decomposing organic matter before new spring growth is established to slow flows and trap sediment (Kovacic et al 2000). Our metaanalysis does not elucidate any overarching seasonal patterns (figure 3), though it is important to note that we have no sub-annual data from studies of longer than two years, and it is very likely that seasonal patterns only become apparent after a restored wetland has reached a steady state with respect to substrate equilibrium dynamics and plant community establishment (Dunne and Reddy 2005).

Wetland age and design
Trends emerge as wetlands age, but the direction of the relationship between wetland age and phosphorus retention can go either way. A wetlands' phosphorus retention capacity might increase over time due to plant establishment or changing wetland soil dynamics (Carstensen et al 2019). For example, wetlands established with phosphorus laden soils or substrates (such as wood chips) may leach phosphorus in early years until equilibrium is established (Dunne et al 2015). Or the initial substrate of a constructed wetland may lack phosphorus binding affinity and subsequently sediment with greater affinity for P binding might accrue over the first couple years of a wetland's life (Carstensen et al 2019). Conversely, a wetland may lose phosphorus retention efficiency over time if soil phosphorus binding capacity becomes saturated (Jordan et al 2003, Díaz et al 2012. It is therefore not surprising that we do not see an overall trend of phosphorus retention as wetlands age given the fact that soil binding capacity might trend in either direction over time. Audet and others (2020) observe this lack of a pattern in their study of eight wetlands that span an age range of 3-13 years and suggest that restoration design and soil properties confound potential patterns driven by wetland aging. There is a clear need for more long-term data to help assess phosphorus retention as wetlands age (see section 3.5).
Finally, wetland design and configuration are largely beyond the scope of this meta-analysis but are worth exploring as means to improve phosphorus retention. Sinuous flow paths and designs that prevent channelization helps evenly distribute water, slowing flow and promoting phosphorus uptake (Jordan et al 2003, Dunne et al 2015. Additionally, the use of sedimentation basins or multiple wetlands in sequence, also known as a treatment train, may be more effective than wetlands with a single basin (Beutel et al 2014). Achieving maximum phosphorus retention in wetlands might require employing multiple strategies (i.e., design aspects and substrate removal) within a single wetland restoration project.

Areas for further research
Although we have come a long way in understanding the use of restored and constructed wetlands for phosphorus retention, there are key gaps that should be addressed to further progress toward improving water quality. This meta-analysis lacks studies from many parts of the world, particularly the Southern Hemisphere and only a small fraction of wetland restoration studies present data from pre-restoration conditions (Audet et al 2020). There is a lack of longterm monitoring studies, and studies on aging wetlands, which leaves a question of wetland longevity: how long do restored and constructed wetlands maintain their functions and how do these functions change over time? And how long do wetlands persist on the landscape, or do they eventually fill in, as is the case with many farm ponds (Swartz and Miller 2021)? Long-term studies are especially important for improving our understanding of interannual climatic variability that can be used for predicting future longterm trends.
Another experimental design choice that may bias our understanding of phosphorus retention in wetlands is the use of mesocosms for studying this problem. While mesocosms are highly effective for studying mechanistic questions, they influence our perception of wetland phosphorus retention by predicting greater retention efficiencies than in most real-world systems. Mesocosm wetlands are rarely observed to be phosphorus sources ( figure 4(b)) and when mesocosm studies are excluded from the analysis, overall median PO 4 3− retention drops from 28.2% to 22.7%. This is likely because mesocosms are more controlled systems and subjected to less extreme conditions than constructed and restored wetlands. Given that 24% of the dataset comes from mesocosm studies, this is a potential source of bias for our collective understanding of the efficacy of wetlands in retaining phosphorus. A major advantage of mesocosms is the ability to track the water balance more accurately; improving hydrologic accounting in field studies should be a priority for future research efforts. Limits to determining the water balance also contribute to the dominance of flow-through wetlands in the literature and results in a gap in understanding the roles of riparian and terminal wetlands as nutrient filters (Walton et al 2020, Cheng et al 2023.
Another major gap in our understanding of wetlands for phosphorus retention is in the availability of data with high temporal resolution. Greater accessibility of sub-annual wetland nutrients budgets would help further our understanding of the temporal dynamics and hydrologic drivers of phosphorus retention, particularly for systems with intermittent flow, like unregulated wastewater treatment wetlands. We encourage the publication of these data sets in a way that facilitates meta-analysis, in order to improve our understanding of seasonal dynamics and uncertainty in this kind of data. Furthermore, although regular sampling helps elucidate broad seasonal patterns and the effects of agricultural practices, weekly sampling, or other fixed-time interval approaches are limited in their ability to capture influential storm events (Vargas and Le 2023). The effect of highflow events on wetland phosphorus retention is most likely underrepresented in most studies (Audet et al 2020). Flow proportional sampling with an emphasis on high flow events is more expensive and time consuming, however, when it comes to phosphorus, storm events exert a disproportionate effect on annual budgets (Jordan et al 2003). This phenomenon has been discussed at length by many (Jordan et al 2003, Mendes et al 2018, Audet et al 2020, but it merits restating as we have demonstrated here the difficulty in capturing the effect of hydrologic drivers of phosphorus retention at the annual scale. Storm sampling and the publishing of high frequency data would make it possible to refine our understanding of the extent that hydrologic dynamics control wetland phosphorus retention. Finally, a major knowledge gap exists in the role of soil and sediment characteristics, both phosphorus content, and other parameters influencing a wetland's phosphorus retention-release dynamics. Though often characterized as either mineral or peat/organic soils, detailed characterization of wetland soils would greatly augment our ability to compare phosphorus retention between different sites. Contextual information about soil type, soil phosphorus content, phosphorus fractionation and other soil characteristics such as pH and iron content are seldom reported. For rewetted peat soils in particular, the age and degree of decomposition affects the form and mobility of potential phosphorus sources (Zak et al 2008). Measurements of multiple forms of phosphorus, particularly phosphate, are also critical to our understanding of wetland processes. Without this information it is difficult to tease apart the mechanisms controlling phosphorus release. Lab experiments (i.e., experiments by Zak et al (2008)) and surveys across multiple wetland sites could help address these knowledge gaps.

Conclusions
Wetlands have been employed to improve water quality in catchments impaired by agricultural runoff and nutrient pollution. This meta-analysis of 139 wetlands demonstrates that a wetland's ability to retain phosphorus is highly variable, albeit wetlands primarily act as phosphorus sinks (84% of site-years for TP, 75% for PO 4 3− ). While the median retention of TP in wetlands that are phosphorus sinks is larger than what is released by wetlands that are phosphorus sources (2.0 g·m −2 ·yr −1 compared to −0.5 g·m −2 ·yr −1 ) releases of phosphate are more common and more significant in their magnitudes. Wetland characteristics and hydrologic regime govern phosphorus retention dynamics in wetlands. Mesocosm studies retain the most phosphorus and impart a potential bias into our understanding of the overall efficiency of wetlands for retaining phosphorus. Wetlands associated with wastewater treatment and those in agricultural landscapes exhibit more variable phosphorus retention efficiency, and occasionally are phosphorus sources, particularly those with unregulated hydrologic regimes. We also found wetland phosphorus releases to coincide with high HLRs, large catchment area to wetland area ratios, low influent phosphorus concentrations, and legacy phosphorus in wetland soils. Monthly data is highly valuable for elucidating the importance of hydrologic flow in controlling phosphorus retention, but limited availability for meta-analysis and synthetic work hinders our understanding of these processes as well as seasonal variability and temporal trends. Finally, we discussed mechanisms of wetland phosphorus retention and release, as well as recommendations for promoting the former and minimizing the latter for helping to achieve maximum water quality improvements. We conclude that the use of wetlands as a means for improving water quality should be continued and expanded, especially considering the many co-benefits wetland provides in terms of wildlife habitat and improvements to water quality through removal of other harmful pollutants.

Data availability statement
The data that support the findings of this study are openly available at the following URL/DOI: https:// doi.org/10.6084/m9.figshare.22471912.v1.