Treatment Wetland Plant Harvests as a Tool for Soil Phosphorus Reduction in North Central US Agricultural Watersheds

Abstract

Agricultural watersheds in the North Central United States have been intensively farmed for decades with widespread application of fertilizer and extensive tilling practices. Soil phosphorus built up in sediments over time as a result of these practices may be released under anaerobic conditions, such as flood events. These floods are increasing in frequency and intensity due to climate change, leading to downstream water-quality concerns. Edge-of-field best management practices, including constructed treatment wetlands, provide a natural buffer for excess phosphorus runoff, but may only be a temporary solution if soil becomes oversaturated with phosphorus over extended periods of time. Preventing wetlands from becoming sources of phosphorus to water bodies may be essential for management in future years when considering impacts from climate change. This research assesses how wetland plant harvesting can reduce soil phosphorus accumulation (measured as Olsen phosphorus) in edge-of-field treatment wetlands, thereby preventing these systems from becoming phosphorus sources and ensuring the longevity of water-quality benefits from these systems. Using several 380 L controlled wetland mesocosm experiments in 2018–2019, we assessed above-ground plant material (S. tabernaemontani and B. fluviatilis) and soil Olsen P through the growing season and after harvest. We observed a reduction in soil phosphorus from wetland plant harvesting between 1–50% over one year, with a mean reduction of 7.9 mg/kg. B. fluviatilis initially contained higher P concentration early in the season (0.82% P content) compared to S. tabernaemontani (0.76% P), but S. tabernaemontani retained higher P later in the season (0.3% P content) compared to B. fluviatilis (0.25%). Time of season may significantly impact plant P removal potential, including accessibility of treatment wetland sites. While controlled mesocosm experiments may not always be applicable to real landscape-level management, this study highlights the potential for reductions in soil phosphorus and corresponding downstream phosphorus fluxes in edge-of-field treatment wetlands through plant harvest during the growing season. Plant harvesting can be used by land managers in edge-of-field treatment wetlands as an adaptation mechanism for shifting environmental conditions, such as increased heavy rainfall occurrences and flood events, that are exacerbated by climate change in this region.

1. Introduction

There have been many studies to identify and further understand the numerous ecosystem services that North American wetlands may provide [1,2,3]. Nutrient cycling within some wetlands, including the retention of phosphorus (P) and removal of nitrogen (N), is an essential ecosystem service to some areas [4,5,6,7,8]. In the north central United States, intense agricultural land management and a significant decrease in overall wetland area [9,10] have contributed to a surplus of nutrients (N and P) within many watersheds, leading to water-quality concerns and management challenges [11,12]. Wetlands, including naturally occurring, restored, or constructed wetlands, may serve as an effective barrier to pollutant loading from agricultural fields and water bodies [13]. Treatment wetlands are an engineered edge-of-field management practice that are strategically placed and specifically address water-quality management from adjacent agricultural practices [13,14,15]. Although wetlands have been found to be effective in their ability to remove N from a system through the natural process of denitrification [4,16], P cycling, management, and storage remain critical considerations in some agricultural watersheds in the north central United States [12,17,18].

While wetlands are generally considered nutrient sinks within a landscape [5], under certain circumstances they may be a source of P [19,20], thereby posing problems for water quality and land management. Under climate change conditions such as increased drought and heavy rainfall occurrences, some wetland biogeochemical cycles may be altered [21,22], potentially impacting P sorption and release [23]. Phosphorus loading to a wetland may result from surface water overflow, agricultural tile drainage, erosion and sediment deposition, decomposition of organic matter, or the re-suspension and internal loading of P stored in sediments [24,25,26,27]. Solid forms of P that are bound to sediment may be more easily settled out than soluble forms; however, they tend to remain in the wetland without physical removal.

Storage of bound P within the soils of an edge-of-field treatment wetland is an intended outcome in management decision making but may be a temporary solution if the biogeochemical cycling of P is not considered under fluctuating environmental conditions. Accumulated soil or sediment associated with P can be released through biogeochemical reactions in the soil or during or from heavy rainfall, runoff events, or other disturbances [23,27,28,29,30]. Under anaerobic conditions, stored P that was once bound to sediment may be released, ultimately negatively impacting P cycling and water quality from eutrophication [11,31].

Wetland vegetation also plays a significant role in P cycling, with studies showing wetland species removing up to 5% of Total P (TP) [32]. Some wetland species that have shown promise for P removal potential include cattails (Typha spp.) [33] and mixtures of wet prairie species [34]. Hybrid cattail (Typha × glauca), however, is an invasive species in North America and so is not planted intentionally by resource managers.

Climate change is expected to increase the variability in rainfall events, drought, and other conditions that directly impact the cycling of P [21] and may lead to increased loading and mobility of P in agricultural watersheds, making wetlands and wetland plants a key component of climate adaptation. Edge-of-field constructed treatment wetlands located in agricultural watersheds that contain high soil P content run the risk of becoming sources instead of sinks for different forms of P [11]. It is critical that land managers consider soil P content and cycling, wetland vegetation that is present, and the impacts from climate change on P cycling to prevent further water-quality pollution. This research aims to address these management challenges and decisions by assessing the potential for soil P retention and removal from treatment wetland plant species, thereby reducing the risk of further water-quality degradation of valuable water sources. The edge-of-field research in particular addresses the challenge of limited land availability for wetland restoration in agricultural watersheds. Plant harvesting maximizes the nutrient-removal potential of small wetlands, which are needed in intensively farmed landscapes.

Plant uptake of nutrients within the root systems and shoots [14,35,36] can be one of the main pools of bioavailable forms of P within a wetland or aquatic ecosystem [14]. In a study of 35 wetland species, P concentrations ranged from 0.08% to 0.63% by mass [37,38]. Indeed, Fraser et al. [39] conducted a mesocosm experiment in which S. tabernaemontani harvesting reduced overall soil TP content. Results showed that fall harvests may be less effective at soil P reduction from biomass harvest and removal. Considering the large pool of P that can be stored within wetland soils [30], often measured using the Olsen P soil test, and the water-quality risks associated with soil P release [29], land managers should consider harvesting treatment wetland vegetation annually for their desired nutrient removal or management goals.

Building on research from Fraser et. al. and Jiang et. al., the research questions ad-dressed in this project were: How effective is the management practice of wetland plant harvesting for reducing bioavailable soil P, measured as Olsen P? Does the plant species being removed significantly impact the removal potential of P from the soil profile of an edge-of-field treatment wetland? Can edge-of-field treatment wetland vegetation harvesting be used as an effective management tool for water-quality protection? Wetland plant species that are effective at capturing bioavailable forms of P [37,38] were tested for P removal capacity from the soil over the two growing seasons with repeated vegetation harvests as a management practice. The impacts of climate change, including in some areas an increased risk of flood events and increased risk of drought conditions [21] on the biogeochemical cycling of P throughout edge-of-field treatment wetlands in the north central United States, remain to be fully understood as adaptation measures continue to be explored. To address these questions, we drew comparisons between a controlled mesocosm experiment and an edge-of-field treatment wetland located in the southwestern region of Minnesota. Although biomass and P content are large factors to consider for the proposed research questions, we also recognize that the practicality of some wetland plants and harvesting of them may not be accessible in some locations and for some land managers, depending on seasonality. Ultimately, we hypothesize that stored P content within the soils and established plant biomass within the wetland will largely determine P removal capacity from plant harvests.

2. Materials and Methods

2.1. University of Minnesota Mesocosm Experiment

The St. Paul, Minnesota mesocosm experiment began in 2017 and was modeled after similar wetland mesocosm experiments by Mitsch [18,40] and Jiang et. al. [41]. The mesocosm experiment design consisted of thirty (30) 378-liter (L) tanks that were placed in an excavated portion of land on the University of Minnesota agricultural experimental station (44.990535, −93.181045). Each tank was set up with approximately 25–30 cm of soil collected from a nearby wetland area and combined with farmland soil to create a loamy soil mix. In 2017, vegetation within the tanks included 3 native Minnesota wetland species monocultures that were planted via seedlings: prairie cordgrass (Spartina pectinata), Canada bluejoint grass (Calamagrostis canadensis), and tussock sedge (Carex stricta). These plant species were selected due to being wetland species native to the region and similar species to those found in the Granada treatment wetland. Jiang et. al. [41] used softstem bulrush in similar mesocosm experiments and helped inform our selection of species, including those of varying flood tolerance and different taxonomic groups. Soil samples were also collected from all the mesocosm tanks in the fall of 2017 and tested for Olsen P, which may be a strong indicator for soil available P [42]. Experiment site design and methodology can be explored further in Alsadi, N. [43].

In 2018, the wetland plants within the mesocosms were replaced by monoculture stands of Minnesota native softstem bulrush (Schoenoplectus tabernaemontani) and river bulrush (Bolboschoenus fluviatilis) throughout the entire tank. These were selected due to being native to Minnesota while also being able to withstand varying water depths throughout a season. The bulrush species were placed in the tanks (15 tanks of each species) via plugs in the spring of 2018. Approximately 12 weeks of fertilizer-enhanced water treatments were added to each tank, simulating agricultural runoff, which totaled approximately 10.22 mg of P applied manually to each mesocosm each season. Agricultural runoff was simulated in the mesocosm tanks by applying fertilizer-water treatments to each tank weekly. This included filling a large tank on-site with water and fertilizer and distributing 5-gallon buckets of the fertilizer-water treatment to the soil surface of each mesocosm tank. It was necessary to include nutrient inputs to the mesocosms to better align with edge-of-field treatment wetlands and their direct collection and filtering of runoff from an adjacent farmland, since the mesocosms were not located at the edge of a farm field. We sampled bulrush species within each mesocosm (approximately 3 samples per selected tank) on 5 distinctive dates from May–September throughout each season to assess P content (ppm) and biomass (mg) above the soil level. Above-ground plant samples were harvested at 5 distinct times throughout the entire growing season (May–September) through 2018 and 2019. Of the 30 mesocosms, 3 tanks of each type of species (total 6 tanks) were sampled at each distinct sampling date. Sampling dates were selected to be representative of the growing season with a few weeks between each. Plant samples were collected at just above the soil line, with 3 of the 6 quadrants of each tank selected via an online randomized number generator (limited to 1–6 to represent quadrants within each mesocosm tank). Within the 2 seasons of P content and biomass assessment, the selected 5 wetland species were assessed for plant biomass and P content upon harvest of the mesocosm tanks each fall.

In the 2018–2019 seasons, soil samples were taken at the beginning (June) and end (October) of each growing season within all tanks and assessed for Olsen P. Samples were gathered within the upper 15 cm of soil within the tanks using a small soil auger. Soil samples were collected from near the center of each mesocosm tank. End-of-season soil samples corresponded with the annual end-of-season wetland plant harvest that included all vegetation in all tanks to assess soil P removal capacity from vegetation harvest.

Soils were refrigerated for storage, then dried and assessed at the Research Analytical Lab at the University of Minnesota (ral.cfans.umn.edu) for Olsen P. Reduction in P was observed throughout the monitoring seasons. The mesocosm experiment, located in St. Paul, MN, is shown in Scheme 1.

2.2. Edge-of-Field Treatment Wetland in South-Central Minnesota

In the Minnesota River Basin, agricultural practices and repeated applications of fertilizer to the landscape have created water-quality concerns from N, P, and sediment loading [12]. The Granada, Minnesota (43.756562 N, 94.343852 W) edge-of-field treatment wetland is a constructed subsurface tile drainage wetland that was initially installed in an effort to target and reduce the high nutrient loading into the nearby Minnesota River tributary, Elm Creek. With high demand for agricultural land in this region of the US, this effort in installing the constructed wetland looked to minimize the amount of converted land while maximizing treatment services. Phosphorus and N loads into the creek were primarily from agricultural fertilizer applications by the landowner to the adjacent cropped field [34]. The entire treatment wetland is approximately 0.1 ha in size and collects approximately 10.1 ha of agricultural tile drainage. Each cell within the treatment wetland is approximately 13.7 m × 26.7 m, with wet prairie plant species present. Groundwater wells were located throughout the wetland as well as at the inlet and outlet for monitoring [34].

Water volume, loading rates, and nutrient concentrations varied in most of the prior years of analysis of the Granada treatment wetland. In 2013, 2014, 2016 and 2018, flooding occurred throughout the wetland due to overflows of Elm Creek. In these years of monitoring, floodwaters stood within the wetland for approximately 3–4 weeks each season [34,44]. Phosphorus loading to the wetland was approximately 0.144 kg TP in 2017 and 0.464 kg TP in 2018 [34].

Vegetation in the wetland was sampled at various locations within the wetland cells (3) using a standardized (1 square-meter) sampling size within each cell at just above the soil level. These samples were then scaled up to be representative of the treatment wetland in our estimates. The vegetation located throughout the edge-of-farm treatment wetland included a wet prairie species mix planted by seed during the construction of the wetland in 2013. Some of the more dominant vegetation included native species prairie cordgrass (Spartina pectinata) as well as some non-native species, including hybrid cattail (Typha × glauca). The wetland vegetation was harvested at the near-soil surface level by hand using large hedge-trimmers over 1-square-meter plots, air-dried at the University of Minnesota, and assessed for P content in a University of Minnesota lab (ral.cfans.umn.edu). The soil samples were gathered from each of the individual wetland cells and intermittently collected over the 2017–2018 seasons at varying depths (shallow: upper 15 cm; deep: below 15 cm). Soil samples were taken at various depths within the wetland and assessed for Olsen P and TP [34] using methods in which soil is dried and mixed with sodium bicarbonate and later measured with a Brinkman PC 900 probe colorimeter [45]. The Granada treatment wetland is shown in Scheme 2.

3. Results

3.1. Mesocosm Experiment

3.1.1. Species Biomass and P Uptake

In the 2017 experiment, C. canadensis displayed the lowest biomass overall (84.43 g) but had higher P concentration (2575.5 ppm P) compared to the other species, with almost twice the P concentration as S. pectinata (1299.04 ppm P). However, results showed that S. pectinata ultimately removed 0.337 P(g)/DW(dry weight)(g) at the end-of-season harvest, which was slightly higher than the other species (C. canadensis 0.217 P(g)/DW(g); C. stricta 0.314 P(g)/DW(g)). In a comparison of all the mesocosm vegetation tanks in the 2017–2018 seasons, the two bulrush species removed more P(g)/DW(g) (Figure 1) than the other species assessed; the 2018 S. tabernaemontani and B. fluviatilis removed 1.11 P(g)/DW(g) and 0.595 P(g)/DW(g). Results did not indicate statistically significant differences in P removal between species. However, the results indicate the potential for all assessed treatment wetland species to retain and remove P from soils saturated with fertilizer upon vegetation harvest. Selection of treatment wetland vegetation may be dependent upon the location and availability to the land manager. Nutrient management strategies should consider the P content and biomass of vegetation when considering selecting and planting vegetation within a treatment wetland. While individual wetlands are effective at nutrient removal, scaling up to meet larger watershed goals for P and N total load reduction would require adaptation at a much greater rate and scale.

3.1.2. Seasonal Wetland Vegetation Harvest

In 2018, softstem bulrush (S. tabernaemontani) had higher P content (3205.03 ppm) but a lower overall biomass. This resulted in similar amounts of P removed between S. tabernaemontani and the river bulrush (B. fluviatilis) species; approximately 0.0016 kg (1.6 g) of P were removed per tank on average from the harvested biomass (Figure 2). B. fluviatilis initially contained higher P content in early season samples with an average 0.82% P content compared to 0.76% P in S. tabernaemontani, but S. tabernaemontani retained higher P later in the season with 0.3% phosphorus content compared to 0.25% in B. fluviatilis (Figure 3 and Figure 4). Contributing factors to the varying P content between species may include growth rates of each species and the demand for P uptake at the start of the growing season upon establishment of the species compared to later in the season. The comparable P removal rates indicate similar potential from both bulrush species to retain and remove P upon harvest. Seasonal variability in plant biomass and P content indicates P removal may be optimized depending on species and timing of harvest.

3.1.3. Mesocosm Soil P

Soil samples were compared from beginning to end of season in coordination with the seasonal vegetation harvest. Following the 2017 and 2018 tank harvests, soil P content (mg/kg) was visibly reduced in every tank, ranging from 1 mg/kg to 20 mg/kg per tank (mean reduction of 7.9 mg/kg of Olsen P with a standard deviation of approximately 5.7 mg/kg), which is approximately a 1–50% P reduction range (Figure 5). Loading was approximately 10.2 mg of P total in 2018, while removal for the season was on average 33.3 mg/kg Olsen P per tank.

Olsen P reductions occurred in the 2017–2018 season within all the mesocosm tanks through plant uptake and harvest, with softstem bulrush removing more Olsen P from the soil on average than river bulrush. However, further statistical analysis including a T-test indicated that there was no significant statistical difference in Olsen P reductions at the 5% level. Total P was analyzed in 2018 soil samples as well. In 2018, average TP content in the softstem bulrush tanks was 736.5 mg/kg and 847.9 mg/kg in the river bulrush tanks.

The large range of soil P removal (1–50%) was a surprising finding from this study and indicates the need for further long-term analysis including several rounds of annual harvesting to fully assess the potential for long-term soil P removal. A longer-term mesocosm study would help determine what other contributing factors may cause higher or lower removal potential from harvesting. The observed trends in the soil P reductions may indicate a reduced removal potential over time as soil P becomes less plant-available following several rounds of annual harvesting. Long-term effectiveness of soil P removal potential may decline over time as legacy P may be reduced.

3.2. Edge-of-Field Treatment Wetland Case Study

3.2.1. Species Biomass and P Uptake

The early seasons of monitoring at the Granada, MN treatment wetland began around 2013 and assessed nutrient reduction within the site [34]. Inputs to the wetland included an average loading rate of TP into the wetland of approximately 0.144 kg in 2017 and 0.464 kg in 2018 [34]. In 2017–2018, results of harvested biomass from each individual cell were compared to understand which cell may have higher P stored in the vegetation. In 2018, Cell 1 and Cell 3 removed similar amounts of P per sample (approximately 0.4 P(g)/DW(g)). There was no significant difference in P removal within each individual cell within the treatment wetland between both seasons of monitoring. The 2018 season wet prairie mix harvest removed more P on average than the 2017 season. Estimated TP removed from the wetland area in 2017 by vegetative harvest was approximately 2.3 kg/ha; in 2018, estimated P removal within the wetland was approximately 3.2 kg/ha. In all of the seasons of sampling and vegetation harvesting, 2014 had the highest estimated removal rate by harvested biomass with approximately 4.5 kg/ha of P and approximately 0.44 kg of P potentially removed by vegetation throughout the wetland area (0.1 ha) (Figure 6). These results indicate that higher P removal may be witnessed in the earlier seasons of harvesting due to higher plant-available P that has been built up over time within the soil, with higher P removal seen in the 2014 and 2015 seasons. Other external factors that may have contributed to the variations in removal rates include the precipitations levels to the site and management decisions and practices in the adjacent lands. Land managers should consider both short- and long-term management practices for the treatment wetland in mitigating nutrient pollution to the adjacent water body.

3.2.2. Granada Wetland Soil P

Soil P content varied among sampling seasons [13,34]. In 2012, soil samples within the wetland contained approximately 10.6 mg/kg P. In 2018, soil P was reduced to approximately 1.6 mg/kg. Similarly, Olsen P measurements were collected to assess the more bioavailable forms of P. In 2014, average Olsen P was 3.9 mg/kg. In 2019, soil Olsen P was reduced to an average of 2.8 mg/kg. Shallow soil samples (upper 15 cm) within the wetland displayed higher Olsen P values than deeper soil samples (below 15 cm). In 2014, Cell 3 had higher phosphorus content (6.25 mg/kg) compared to Cell 1 (2 mg/kg). In 2019, Cell 3 continued to display higher phosphorus content (4 mg/kg) than Cell 1 (2 mg/kg) (Figure 7).

The trends in soil Olsen P indicate that reductions in more bioavailable forms of soil P can be achieved over time through annual vegetation harvesting. Variability in soil P over the sampling seasons indicates nutrient fluctuations within the wetland ecosystem over time, which may be related to the decisions of the land manager including fertilizer inputs and overall productivity. Other factors impacting the nutrient dynamics within the wetland ecosystem may include climatic conditions including precipitation. The observed soil P content reductions over time align with the broader goals of water-quality improvement and protection within the Minnesota River Basin, as this management practice may remove soil P that is bioavailable (Olsen P) as well as soil P that has been built up over time (legacy P), thereby reducing the risk for future P runoff into adjacent water sources.

4. Discussion

4.1. Plant Harvesting as Management Practice for Reducing Soil P

While it is widely known that wetlands may provide nutrient retention ecosystem services, the added management practice of harvesting vegetation within these systems may provide additional water-quality benefits to an agricultural watershed. Accumulation of P in agricultural wetland soils can be a risk to water quality, and the remediation of this legacy P from fertilizer will be important to consider to prevent P fluxes under varying hydrologic conditions. Wetland vegetation requires essential nutrients, including P, to grow every season, utilizing what reserves are available within the soil. When harvesting the above-ground biomass and removing it from the site, the nutrients that have been utilized by the plant and are stored within the above-ground biomass are also removed. Over time, this may reduce the amount of stored legacy P within the soil of a treatment wetland that has become oversaturated with P. The management decision to annually harvest treatment wetland vegetation will aid in reducing the risk and amount of P loading to nearby water sources. The high removal rate of P by plant harvest at the edge-of-field Granada treatment wetland site (2.3 kg/ha in 2017; 3.2 kg/ha in 2018) compared to the P loading rate (0.144 kg TP in 2017; 0.464 kg in 2018) indicates that harvesting native wet prairie species mix may reduce P that has been accumulated within soils over time with annual vegetation harvesting as an effective management practice.

In the 2017–2018 mesocosm experiments, soil Olsen P (mg/kg) was reduced in every mesocosm tank (30 total), ranging from 1 mg/kg to 20 mg/kg per tank reduction (mean reduction of 7.9 mg/kg of Olsen P with a standard deviation of approximately 5.7 mg/kg). This displayed a 1–50% P reduction range throughout all tanks, indicating high potential and variability for P removal. Loading of P was approximately 10.2 mg in 2018, while removal from harvest was on average 33.3 mg/kg P per tank. This aligns with similar studies in this region, including Canadian marsh vegetation harvesting [1], that resulted in removal of up to 4.7% of total P loading. Strong plant establishment and varying water levels within the mesocosms may impact the removal efficiency from plant harvesting. The mesocosms favored plant growth due to the regular watering and nutrient loading and likely higher temperatures than in the Granada wetland. Given the variability in soil P removal within the mesocosm tanks, effectiveness of plant harvesting for P remediation at larger scales will need to account for drainage-area-to-wetland-harvesting-area ratios to determine how much harvesting would be needed to achieve nutrient-reduction goals. Plant-harvest effectiveness may also be impacted by the hydrology of a given site, with saturation levels impacting P uptake potential, as well as feasibility of access of the site at the optimal time of year for harvest. Water levels are an important consideration for nutrient cycling and potential for P removal [41], and the Granada wetland experienced varying water levels and precipitation over the years of monitoring compared to the controlled hydrologic regime within the mesocosms. The storage of P within the wetland vegetation is another benefit of these systems [32] that may aid in preventing water-quality degradation and has the potential to be enhanced via harvesting and removal. Given potential trade-offs between productivity on agricultural land and treatment wetland services, land managers can consider the amount of wetland land to harvest to meet P reduction goals in partnership with consideration of valuable wetland ecosystem services including habitat for flora and fauna. With a high demand for productive agricultural land, it is especially important to maximize the amount of nutrient retention and removal services in these wetlands in the spaces provided. Long-term plant harvesting may impact the wetland ecosystem, including the existing plant communities and habitat. Harvesting of plant material should be considered to reduce P runoff and buildup within treatment wetland soils and may not be necessary once P levels have been reduced. Annual harvesting of edge-of-field treatment wetland vegetation may reduce stored P content and reduce risks of water-quality degradation in agricultural watersheds while maximizing the potential for nutrient-removal services.

4.2. Implications for Treatment Wetland Vegetation Harvesting as a Management Practice in Minnesota and the North Central United States

Agricultural watersheds throughout the upper Midwest transport high loads of N and P due to routine agricultural practices [17,44,46,47,48]. Phosphorus can be especially challenging to manage or remove from a highly concentrated site due to mainly being present in a liquid or solid form [49]. To enhance the abilities of treatment wetlands to capture and remove P, natural solutions such as selecting native plant establishment and annual harvesting may be a necessary management solution for land managers [1]. These watersheds are ideal for wetland vegetation harvesting due to the high nutrient loading and mobility, leading to water-quality issues including eutrophication. This study highlighted five wetland species for comparison. These species ranged in their tolerance to inundation and drought. Land managers considering treatment wetland plant species for harvest should consider plants that may be highly durable under varying hydrologic conditions. Similarly, some species may be more or less tolerant to repeated harvesting; therefore, it is important that land managers consider the length of time in which they intend to annually harvest and if it aligns with the plant species selected. The implications of directly harvesting above-ground vegetation include the temporary removal of wetland habitat in some cases which may have impacts on the area’s wildlife. Land managers considering harvesting as a practice for soil P remediation should also consider the species of plants selected and their resiliency to repeated harvests to determine how feasible this practice may be for a given treatment wetland. This research supports a holistic approach to edge-of-field best management practices and regenerative agricultural in a changing climate. This research is a proactive approach to optimizing the nutrient-removal efficiency of a treatment wetland and removing the buildup of soil P in intensively managed agricultural landscapes.

Harvesting of wetland vegetation can be an effective nutrient-removal strategy, as shown in this study, but may be largely dependent upon the accessibility of the plants throughout the season as well as equipment that is available to the land manager [50]. Harvesting of treatment wetland vegetation may not be a practical option if the wetland has high water levels in late fall, which is when harvests are typically conducted in the upper Midwest. Drainage of the wetland using water control structures would be necessary in these cases. Biomass and P content may both influence the amount of P that may be removed per harvest. In this study, in 2018, P content within the shoots of both species of plants (B. fluviatilis and S. tabernaemontani) was highest in the early season (June–July) and slowly decreased over time towards the fall (September–October) (Figure 4). The varying levels of P within the vegetation biomass throughout the growing season can be indicative of growth rates and may provide insight into management protocols for desired outcomes with differing species. Timing of harvesting may align with highest P uptake rates, such as harvesting of plant material in June versus end-of-season harvesting. This may be a future research question and is also related to regional or temporal differences in precipitation.

Increased precipitation in recent years throughout parts of Minnesota has resulted in higher water levels into the fall season. Accessibility of wetland plants for harvesting could be a challenge due to the unpredictability of a changing climate. Higher water levels or flooding for longer periods of time throughout the upper Midwest is predicted with climate change [51] and may cause further issues for treatment wetland nutrient-removal efficiency [52]. It is suggested that wetland vegetation harvesting for soil P remediation be considered for seasonal to shallow wetlands that may allow for harvesting access on a seasonal basis. This may include wet meadows or shallow marshes and is less likely a practical management option for deep marshes, forested swamps, and similar systems. Harvesting of plant biomass should correspond with time of season that may optimize plant P removal while ensuring adequate access with available equipment. Future research should build off these findings to determine which wetland plant species may be better suited per region, given a changing climate, to allow for harvesting access and optimized P removal per time of season, including a comparison of efficient methods and equipment for this. Treatment wetland vegetation harvesting can be an effective management practice and may be part of a larger nutrient management plan. Landowners should consider other regenerative agriculture practices, such as cover cropping or other controlled drainage systems along with edge-of-field treatment wetlands to maximize the amount of captured P as well as N and further prevent water-quality degradation through a variety of climate-resilient management practices. Questions that can be hypothesized based on these experiments may be centered around water level influence on P uptake and removal from wetland soil. While many wetlands may be more shallow seasonally, it will be necessary to determine which species may be best suited for optimized P removal.

4.3. Role of Plant Harvest in Climate Change Adaptation

Global annual average surface temperatures have increased 1.8 degrees F from 1885 to 2016, with greater increases witnessed in some northern latitudes. In the upper Midwest of the United States, climate change models have predicted exacerbated drought conditions along with flood or heavy rainfall events in some agricultural watersheds [21]. The increased frequency of rainfall events and flood occurrences in the upper Midwest has a direct impact on soil nutrient concentrations and P discharges throughout agricultural watersheds [53]. Increased intensity and frequency of these events with climate change may result in increased P transport and mobility from wetlands [54]. However, some studies have shown that with continued flooding, draining, and re-saturation, as we tend to predict with climate change, there may be an increased nutrient-removal efficiency compared to a continuous flow, with some wetland plant species removing up to 20–30% more P under the varied hydrologic conditions [55]. Wetland soils that have been saturated for long periods of time under anoxic conditions, such as with flooding, may release P from bound sediment and flush it downstream or to adjacent water sources. In Minnesota, precipitation is expected to increase, with the potential for increased flooding. It is important to offset the potential for P release from oversaturated soils by integrating management practices such as treatment wetlands on edge-of-field and plant harvesting for soil P removal.

The results from our study indicate the role of wetland plants in removing P from saturated soils and mitigating potential negative impacts resulting from climate change in the north central United States. The P uptake and removal from wetland plants in the mesocosms and edge-of-field wetlands (Figure 1) indicates significant potential for P removal with harvest. Wetlands have the potential to be significant sources or sinks for nutrients including phosphorus, and climate change may exacerbate the risks for eutrophication in some agricultural watersheds without considerable P management, such as optimized wetland vegetation harvesting for soil P remediation. Land managers and policy makers should consider both short- and long-term goals of management decisions to maintain sustainability over time. Edge-of-field best management practices should be incentivized by policy makers to ensure limited to no financial losses to the landowner. Agricultural watersheds should seek to strike a balance between wetland conservation and other climate change adaptation strategies by including a variety of methods, including the consideration of wetland plant harvesting, to maximize benefits and reduce pollution from agricultural practices. With the demand for space in agricultural land of the Midwestern US, it is difficult to get farmers to enroll productive farmland in wetland restoration programs. Therefore, it is important to maximize the nutrient-removal benefits of the wetlands we do restore or create to help reduce nutrient loading downstream and achieve our goals.

5. Conclusions

Edge-of-field treatment wetlands in upper Midwest agricultural landscapes may provide a variety of ecosystem services, including plant uptake and storage of excess P that has been built up within the soil profile from routine practices. Managing these types of constructed wetlands by harvesting vegetation annually may increase the potential for soil P reduction over time, lessening the risk of water-quality contamination during rain or flood events. Here, we find that vegetation harvesting is an effective technique for soil P removal in constructed edge-of-field wetlands with special consideration for water level and plant types. The mesocosm study found that soil P can be reduced by 1–50% following harvest of plant material and that P removal per species was comparable among all species assessed. Although bulrush species had higher P content earlier in the season, lower biomass resulted in similar amounts of P removal between bulrushes. Results indicate time of season for harvest, with less P retained in above-ground biomass later in the growing season, as a large contributing factor in overall P removal. This practice may enhance the efficiency of nutrient-removal efforts and prevent water-quality degradation in agricultural watersheds. When considering harvesting vegetation in edge-of-field treatment wetlands, land managers should consider several key factors, including but not limited to overall nutrient-removal goals of the treatment wetland, wetland plant species type, existing soil P content, timing of harvest, and overall accessibility of the wetland for harvest. This study highlights the additional potential for treatment wetlands to further remediate legacy P buildup in agricultural watersheds through the process of harvesting. It is necessary to consider multi-faceted approaches to sustainability and resiliency, and this research provides evidence towards an approach to soil P remediation in landscapes that may be vulnerable in a rapidly changing climate. Future research should explore the selection of species type per hydrologic regime in a geographic region for optimized P removal or impacts of soil properties on the potential for P uptake by vegetation. Furthermore, land managers may consider vegetation harvesting within their edge-of-field treatment wetlands as a climate adaptation strategy and practice to prevent water-quality degradation within agricultural watersheds.