Quantifying Urban Bioswale Nitrogen Cycling in the Soil, Gas, and Plant Phases

water

Article Quantifying Urban Bioswale Nitrogen Cycling in the Soil, Gas, and Plant Phases

Nandan Shetty 1, Ranran Hu 2, Jessica Hoch 3, Brian Mailloux 4, Matthew Palmer 3 , Duncan N. L. Menge 3 , Krista McGuire 5 , Wade McGillis 6 and Patricia Culligan 1,* 1 Department of Civil Engineering and Engineering Mechanics, Columbia University, 500 West 120th Street, 610 Mudd, New York, NY 10027, USA; [email protected] 2 Department of Earth and Environmental Engineering, Columbia University, 500 West 120th Street, 918 Mudd, New York, NY 10027, USA; [email protected] 3 Department of Ecology, Evolution, and Environmental Biology, Columbia University, 1200 Amsterdam Avenue, 10th Floor Schermerhorn Ext, New York, NY 10027, USA; [email protected] (J.H.); [email protected] (M.P.); [email protected] (D.N.L.M.) 4 Department of Environmental Science, Barnard College, 3009 Broadway, 404 Altschul Hall, New York, NY 10027, USA; [email protected] 5 Department of Biology, University of Oregon, 1210 University of Oregon, 77 Klamath Hall, Eugene, OR 97403, USA; [email protected] 6 Lamont-Doherty Earth Observatory, Columbia University, 61 Route 9W, Palisades, NY 10964, USA; [email protected] * Correspondence: [email protected]; Tel.: +1-212-854-3154

Received: 22 September 2018; Accepted: 6 November 2018; Published: 12 November 2018

Abstract: Bioswales are a common feature of urban green infrastructure plans for stormwater management. Despite this fact, the nitrogen (N) cycle in bioswales remains poorly quantified, especially during dry weather in the soil, gas, and plant phases. To quantify the nitrogen cycle across seven bioswale sites located in the Bronx, New York City, we measured rates of ammonium and nitrate production in bioswale soils. We also measured soil nitrous oxide gas emissions and plant foliar nitrogen. We found that all mineralized nitrogen underwent nitrification, indicating that the soils were nitrogen-rich, particularly during summer months when nitrogen cycling rates increase, as indicated by higher levels of ammonium in the soil. In comparison to mineralization (0 to 110 g N m−2 y−1), the amounts of nitrogen uptake by the plants (0 to 5 g N m−2 y−1) and of nitrogen in gas emissions from the soils (1 to 10 g N m−2 y−1) were low, although nitrous oxide gas emissions increased in the summer. The bioswales’ greatest influx of nitrogen was via stormwater (84 to 591 g N m−2 y−1). These findings indicate that bioswale plants receive overabundant nitrogen from stormwater runoff. However, soils currently used for bioswales contain organic matter contributing to the urban nitrogen load. Thus, bioswale designs should use less nitrogen rich soils and minimize fertilization for lower nitrogen runoff.

Keywords: nitrogen cycle; mineralization; nitriﬁcation; nitrous oxide; plant uptake; bioswale

1. Introduction Nitrogen (N) is considered the most problematic pollutant for U.S. coastal waters, having degraded the marine ecosystems, ﬁsh production, and recreation of two-thirds of the U.S. coastline [1]. Due to fossil fuel combustion from power plants and vehicles, 47% more nitrogen accumulates on impervious surfaces in urban compared to nonurban areas [2]. Bioswales, which are a common feature of many cities’ green infrastructure (GI) plans, are vegetated structures designed to reduce water pollution by absorbing stormwater runoff from impervious surfaces [3]. However, recent studies have

Water 2018, 10, 1627; doi:10.3390/w10111627 www.mdpi.com/journal/water Water 2018, 10, 1627 2 of 19

Water 2018, 10, x FOR PEER REVIEW 2 of 18 demonstrated that these GI types can contribute to nitrogen pollution [4–7]. During wet weather studies have demonstrated that these GI types can contribute to nitrogen pollution [4–7]. During wet conditions, high concentrations of nitrogen enter bioswales via stormwater influent, and potentially weather conditions, high concentrations of nitrogen enter bioswales via stormwater influent, and exit via surface overflow and/or subsurface infiltration [8–11]. In dry weather conditions, the nitrogen potentially exit via surface overflow and/or subsurface infiltration [8–11]. In dry weather conditions, source comes from soil mineralization, a commonly used index of soil nitrogen availability [12–14], the nitrogen source comes from soil mineralization, a commonly used index of soil nitrogen which describes the rate that soil microbes decompose long-term stores of organic nitrogen (org–N) availability [12–14], which+ describes the rate that soil microbes decompose long-term stores of organic into ammonium (NH4 ) (Figure1). Mineralization processes occur in all soils. In well-drained soil nitrogen (org–N) into ammonium (NH4+) (Figure 1). Mineralization processes occur in all soils. In environments, nitrifying microbes then further transform this ammonium to nitrate (NO −), the most well-drained soil environments, nitrifying microbes then further transform this ammonium3 to nitrate common drinking water pollutant [15] and a highly mobile anion that readily leaches out of soils [10,16]. (NO3−), the most common drinking water pollutant [15] and a highly mobile anion that readily leaches Bioswalesout of soils are [10,16 most].frequently Bioswales madeare most with frequently fast-draining made soils with that fast are-draining primarily soils aerobic, that are creating primarily a dry environmentaerobic, creating hospitable a dry environment to nitrifying hospitable microbes, andto nitrifying thus have microbes, high levels and of thus nitrification have high [5 levels,10,17 ,of18 ]. Brownnitrification et al. [ 19[5,10,17,18], Lucas and]. Brown Greenway et al. [19 [20],], Lucas and McPhillips and Greenway et al. [ [2021]], suggestand McPhillip that excesss et al mineralization. [21] suggest andthatnitrification excess mineralization account for and why nitrification bioswales have account been for found why tobioswale leach nitrate,s have beenwhich found can be to quickly leach flushednitrate, out which following can be rain quickly events flushed [4,6,10 ]. out Surprisingly, following neither rain events mineralization [4,6,10]. Surprisingly, nor nitrification neither rates havemineralization been well quantifiednor nitrification in bioswales. rates have been well quantified in bioswales.

Plant Uptake: Gas emissions:

• Org - N • N2 + • NH4 • N2O - • NO3 • NO

Stormwater Influent: Bioswale Soil: Stormwater Overflow

• Org - N + - • Org - N + Org - N NH4 NO3 + • NH4 • NH4 - - • NO3 Mineralization Nitrification • NO3

Stormwater Infiltrate • Org - N + • NH4 - • NO3

FigureFigure 1. 1. SimplifiedSimplified nitrogen cycle within bioswales:bioswales: Solid Solid purple purple arrows arrows denote denote nitrogen nitrogen fluxes fluxes quantifiedquantified inin thisthis study,study, whilewhile blueblue dotteddotted arrows denote well well-documented-documented liquid liquid-phase-phase fluxes fluxes that that occuroccur duringduring wetwet weatherweather only.only.

TheThe measurementmeasurement ofof mineralizationmineralization and nitrification nitrification rates rates in in bioswale bioswale soils soils can can help help quantify quantify thethe rate rate atat which which soilssoils releaserelease nitrogennitrogen pollution via the the conversion conversion of of nitrogen nitrogen from from organic organic to to inorganicinorganicforms, forms, whichwhich areare moremore readilyreadily washed out from bioswale soil soils.s. In In addition, addition, the the proportion proportion ofof mineralized mineralized nitrogennitrogen thatthat undergoesundergoes nitrificationnitrification can also also be be used used to to indicate indicate the the concentration concentration of of nitrogennitrogen speciesspecies inin soils.soils. InIn nitrogen-richnitrogen-rich environments, nitrogen nitrogen accumulates accumulates as as nitrate nitrate rather rather than than ammoniumammonium becausebecause nitrifyingnitrifying microbes thrive in in such environments environments.. Conversely, Conversely, when when nitrogen nitrogen is is scarce,scarce, nitrifying nitrifying microbesmicrobes competecompete poorly for ammonium ammonium against against plants plants and and heterotrophic heterotrophic microbe microbes,s, limitinglimiting the the proportionproportion ofof nitrogennitrogen thatthat accumulatesaccumulates as nitrate [ 13,2213,22]].. SimilarSimilar to to mineralization mineralization and and nitrification nitrification rates, rates, the emission the emission rate of rate microbially-mediated of microbially-mediated nitrogen gasnitrogen (including gas (including dinitrogen dinitrogen gas (N2), gas nitrous (N2), oxide nitrous (N oxide2O), and (N2 nitricO), and oxide nitric (NO)) oxide (Figure (NO)) 1(Figure), can also 1), reflectcan also the reflect nitrogen the levelnitrogen in the level soil, in asthe emissions soil, as emissions increase increase when the when microbes the microbes have a greater have a nitrogengreater supplynitrogen [23 supply,24]. McPhillips [23,24]. McPhillips et al. [25] found et al. roadside[25] found ditches roadside to be ditches hotspots to for be nitrogen hotspots emissions for nitrogen due toemissions elevated due nitrogen to elevated influxes nitrogen from stormwater influxes from runoff. stormwater Increased runoff. concentrations Increased ofconcentrations nitrogen species of fromnitrogen fertilization species from [25– 27fertilization] and nitrogen [25–27 deposition] and nitrogen [23] aredeposition also positively [23] are also associated positively with associated increased nitrogenwith increased gas emissions. nitrogen gas In fact,emissions. nitrous In oxidefact, nitrous emissions oxide from emissions irrigated from urban irrigated lawns urban inCalifornia lawns in increasedCalifornia from increased 7.2 µg fromN m− 7.22 h −μ1g[ 28 N ] m to−2 720 h−1 µ[g28 N] to m − 7202 h −μ1g[ 29 N] m after−2 h−1 the [29 application] after the ofapplication fertilizers. of fertilizers.

Water 2018, 10, 1627 3 of 19

Another indication that bioswale soils might be too nitrogen-rich is when sources of nitrogen (stormwater influent and soil mineralization) exceed plant requirements, after accounting for losses from gas emissions. While ornamental landscapes and trees in urban areas require about 5 to 20 g N m−2 y−1 in fertilizer [30], bioswales might not require fertilization because nitrogen inputs through stormwater inflows already contain significant nitrogen loads [21]. The creation of a comprehensive nitrogen mass balance for urban bioswales would allow comparison of nitrogen levels in plant foliage to nitrogen inputs from stormwater and mineralization, and thus help determine whether bioswale soils are at proper nitrogen levels. Better quantification of nitrogen cycling in bioswales is the first step toward improving the design and management of these GI installations in order to minimize their nitrogen export [5,31]. In particular, research that examines the distinctions between mineralization, gas emissions, and plant uptake is greatly needed [5,12,32,33]. This study aimed to meet this need by measuring the levels of nitrogen in soil, gas, and plant forms within seven bioswales located in New York City (NYC) over a period spanning from June 2015 to October 2016. The specific objectives of the study were to: (1) quantify spatiotemporal variability of soil mineralization and nitrification across the study sites, (2) measure nitrogen gas emission rates, (3) estimate maximum nitrogen uptake rates by the vegetation, and (4) use the results to construct an overall nitrogen mass balance for a bioswale. The results suggest that bioswale soils should contain greater C:N ratios from the outset and be maintained without additional fertilizers.

2. Materials and Methods

2.1. Bioswale Sites The seven target bioswales were constructed between April 2013 and April 2014 in the Bronx, NYC, at 40◦490 N, 73◦520 W (Table1). The total area of the bioswales varied from 7 m 2 to 115 m2. Each bioswale contained a 61-cm layer of engineered sandy loam soil. The soil specification used to construct the bioswales is included in the supplemental information. Below the soil lay another 61-cm layer made up of 5-cm-diameter crushed stone. This stone drainage layer was wrapped on all sides with a geotextile filter fabric to separate the stone from the soil without restricting water movement. The soil surface was topped with 7 cm of bark chip mulch, and the bioswales were planted with both native and ornamental plant species including trees, shrubs, grasses, and herbaceous perennials, as detailed in Section 2.4. Stormwater influent entered into the bioswales through curb cut inlets. The bioswales were designed for infiltration and did not contain underdrains. Stormwater that did not infiltrate overflowed through curb cut outlets. Four of the smaller bioswales were termed “right-of-way bioswales” (ROWBs) because they were located in the sidewalk, while the remaining three bioswales, termed “stormwater greenstreets” (SGSs), were larger and extended into the street. Shetty contains a full description of the target bioswale sites [34]. Soil characterization and soil mineralization and nitrification measurements were conducted at all seven sites, with four of the seven sites undergoing more intensive sampling to evaluate spatial variability in soil mineralization and nitrification (Table1). Soil gas emissions were sampled at one ROWB and one SGS to detect if the size of the sampling site affected results. Plant foliar nitrogen was sampled at the four ROWB sites only, as the SGS sites contained too great a variety of plant species for comparison. Water 2018, 10, 1627x FOR PEER REVIEW 44 of of 18 19

Table 1. Bioswale sampling summary. Sites that received more intensive sampling to evaluate within- Table 1. Bioswale sampling summary. Sites that received more intensive sampling to evaluate site spatial variability in soil mineralization and nitrification are designated by “xx”. ROWB: right-of- within-site spatial variability in soil mineralization and nitriﬁcation are designated by “xx”. ROWB: way bioswale; SGS: stormwater greenstreet. right-of-way bioswale; SGS: stormwater greenstreet. Soil Soil Mineralization Gas Plant Bioswale Area (m2) Soil Soil Mineralization Bioswale Area (m2) Gas Emissions Plant Uptake CharacterizationCharacterization andand Nitriﬁcation Nitrification Emissions Uptake ROWB 26B 9.3 x xx x ROWB 26B 9.3 x xx x ROWBROWB 2323 7.0 7.0x x xx xx ROWBROWB 9A9A 9.3 9.3x x xx xx ROWBROWB 9B9B 9.3 9.3x x xxxx xx xx SGS 21 76.2 x xx x SGS 21 76.2 x xx x SGS 11 94.8 x xx SGSSGS 11 2 94.8 115.2 x x xxx SGS 2 115.2 x x

As described in Sections 2.2 and 2.3, soil mineralization and nitriﬁcation and gas emissions As described in Sections 2.2 and 2.3, soil mineralization and nitrification and gas emissions measurements were conducted within one meter of each site’s inlet, whereas soil characterization measurements were conducted within one meter of each site’s inlet, whereas soil characterization and spatial variability measurements were conducted at regular intervals from the inlet to the outlet and spatial variability measurements were conducted at regular intervals from the inlet to the outlet (Figure2). All plants were sampled throughout each site. (Figure 2). All plants were sampled throughout each site.

Inlet Outlet

Gas Emissions Soil Characterization Soil Mineralization and Nitrification All sites

Four sites tested for spatial variability

FFigureigure 2. Photo of representative site with example sampling strategy strategy.. Plant sampling locations are omitted from the figure ﬁgure because all plant species were sampled throughout each site.site.

2.22.2.. Soil

2.2.1. Soil Characterization 2.2.1. Soil Characterization On 30 June 2016, two composite soil samples were taken from each of the seven sites using a On 30 June 2016, two composite soil samples were taken from each of the seven sites using a circular 7.6-cm-diameter barrel auger [35]. The samples were submitted to the Soil Testing Laboratory circular 7.6-cm-diameter barrel auger [35]. The samples were submitted to the Soil Testing Laboratory at Rutgers University and analyzed for soil texture, organic matter, total carbon, and total nitrogen. at Rutgers University and analyzed for soil texture, organic matter, total carbon, and total nitrogen. The two composite samples consisted of a shallow sample taken from a depth from 7.6 cm to 22.9 cm, The two composite samples consisted of a shallow sample taken from a depth from 7.6 cm to 22.9 cm, and a deeper sample from 30.5 cm to 45.7 cm depth. Each composite was comprised of three

Water 2018, 10, 1627 5 of 19 and a deeper sample from 30.5 cm to 45.7 cm depth. Each composite was comprised of three subsamples spatially distributed across the site, one located at the inlet to the bioswale, one in the middle of the bioswale, and one near the outlet of the bioswale (Figure2). The three subsamples were added to the same polyethylene Ziploc bag and homogenized through kneading the bag by hand to create each composite sample.

2.2.2. Soil Mineralization and Nitrification Soil mineralization and nitrification were analyzed with an intact soil core, a standard method for measuring soil mineralization [36,37]. Cores were housed in two 25.4-mm-inner diameter polyvinyl chloride pipes, each 50.8 cm long, that were sharpened at the bottom to minimize soil disturbance during insertion. These pipes were cut in half lengthwise and duct-taped back together so that they could be opened following retrieval. They were pushed into the soil leaving 5 cm above ground. Each core housing had a pre-cut 1-cm hole at the top for a rebar handle that could be used to pull the core out (Figure2). Soil cores were sampled in pairs at each bioswale. The cores were taken within one meter of the bioswale inlet, and 15 cm apart from each other. To test initial ammonium and nitrate levels, the first core was removed immediately, while the other core was left in the ground at the site for a seven-day field incubation period. Neither the initial nor incubated core was consistently in front of the other or closer to the inlet. A photograph was taken to ensure that the same locations would not be sampled on subsequent visits. The hole left after removing each core was backfilled with bioswale soil to restore the site. Following removal, the duct tape on each core was cut, the pipe was opened and the soil core length measured. The samples from the top and bottom half of the core were then homogenized in separate polyethylene Ziploc bags through kneading the bags by hand. In the field, 5 g of soil sample was added to 50-mL vials that were pre-filled with 30 mL of a 2 M KCl solution; the salts in this solution displace ammonium and nitrate from the soil, allowing quantification of nitrogen levels in the soil via measurement of solution concentration levels. Vials were brought back to the Heck Laboratory located at Barnard College, NYC and shaken at 150 rpm for one hour to allow ammonium and nitrate to dissolve into the solution. Following this, vials were then centrifuged at 4000 rpm for four minutes to filter out the soil using Millex-SV Durapore polyvinylidene fluoride (PVDF) filters (0.22 µm pore size); the resultant solution was then frozen within 6 hours of sampling. To calculate the mass of dry soil used in the 5 g sample, soil moisture was also measured using a gravimetric method [38]. One pair of soil cores was sampled and analyzed at the seven bioswales monthly from June 2015 to August 2016, with the exception of January and February 2016 which were not sampled due to frozen soils. This resulted in the measurement of soil nitrogen content on 26 dates, and the calculation of mineralization and nitrification rates for 13 incubation periods. Within-site spatial variability of soil mineralization and nitrification was measured at four of the seven sites in October and November 2015 (Table1). The spatial variability test included five pairs of soil cores rather than just one, with each pair evenly spread down the length of the bioswale from the stormwater inlet to the outlet (Figure2). Ammonium was analyzed by a fluorometric method [ 39] and nitrate analyzed by ion chromatography [40]. Net mineralization and net nitrification were calculated with the following equation:

C + − C Net Mineralization and Net Nitriﬁcation (µg N d−1 g−1 dry soil) = i t i (1) t

− + where C represents the concentration of NO3 + NH4 when calculating net mineralization, − C represents the concentration of NO3 only when calculating net nitriﬁcation, and t = 7 days. Ci refers to concentrations from the immediately sampled soil cores and Ci+t refers to concentrations from the soil cores in place for seven days at a site. Water 2018, 10, 1627 6 of 19

2.3. Soil Gas Emissions Soil gas emissions were sampled at two of the seven sites with a static chamber method, which is the most common method for analysis of N2O fluxes [26,41]. A polyvinyl chloride (PVC) cap soil gas chamber was used, 7 cm in height and 20 cm in diameter, with three Swagelok 3 mm tube compression fittings drilled into the top of the chamber (Figure2). Two fittings were attached to septa to allow for syringe withdrawal of gas from the chamber, while the third was fixed with a 1.2-m-long, 3-mm-diameter coiled stainless steel tube in order to equalize pressure but prevent significant diffusion of gas out of the chamber. Polyethylene plastic sheeting with a thickness of 4 mm tightly wrapped the chamber and a heavy chain was placed on top of the sheeting to create a seal with the uneven soil. The gas concentration in the chamber was sampled over time, and translated into a flux rate. Gas samples with 15 mL of headspace were taken at 2, 5, 10, and 15 min intervals using a 30-mL syringe and stored in over-pressurized pre-evacuated 10 mL vials. Gas samples were analyzed for nitrous oxide using a SRI 8610C gas chromatograph (SRI Instruments, Torrance, CA, USA). For each test day and site, soil gas emission was calculated by determining a linear slope of concentration of the four time points [25,42] while considering the footprint and volume of the chamber and the density of air. Gas fluxes were measured 16 times at two bioswales on eight dates between 15 June 2015 and 21 September 2016. All measurements were conducted during the summer and fall seasons, with the exception of 15 June 2015, which was included with summer samples for statistical analysis. During the sampling dates, the temperature ranged from 8.4 ◦C to 26.9 ◦C and the total rain depth over the previous three days before sampling ranged from 0 to 56.1 mm. All measurements were conducted within one meter of the inlets of each site, an area that may have increased soil gas emissions compared to the rest of the site due to greater nitrogen inputs [24].

2.4. Plant Uptake We measured the level of nitrogen in samples of tissue from ten leaves per plant species, collected among the different plant species present at four different sites (Table1). The vegetation in the four sampled bioswales was planted such that there were discrete patches covered by each species. We also measured the spatial extent of each plant species by measuring the area covered by each distinct patch for a given species. We selected leaf area indices from published literature, and summed data from all plant patches to estimate total nitrogen content in aboveground foliage. We sampled 10 leaves collected across the different patches of each plant species at each bioswale on 5 October 2016. On this date, ROWB 9A and ROWB 9B both contained Hemerocallis ‘Lady Florence’, Liriope muscari, Panicum virgatum, Pennisetum alopecuroides, Quercus palustris, Rosa ‘Radrazz’ Knockout, and Symphyotrichum novae-angliae, ROWB 26B contained Amelanchier canadensis, Aronia melanocarpa, Nepeta x faassennii ‘Walker’s Low’, Panicum virgatum, and Spiraea tomentosa; and ROWB 23 contained Aronia melanocarpa, Echinacea purpurea, Eupatorium sp., Nepeta x faassennii ‘Walker’s Low’, and Panicum virgatum. Each collected leaf was photographed in the field immediately after collection with a 7.62 cm × 12.7 cm card for determining scale. Using ImageJ, the perimeter of each leaf was traced to calculate the area. After air drying in the laboratory, all leaves were massed, crushed, ground, and analyzed for carbon and nitrogen using a FlashEA® 1112 Carbon and Nitrogen analyzer (Thermo Scientific, Waltham, MA, USA). We quantified carbon and nitrogen ratios (C:N), nitrogen per unit mass, nitrogen per unit area, and nitrogen per leaf. For statistical analysis, plant species were grouped into shrubs, grasses, and forbs. To calculate the area of each patch, top-down photos were taken on October 12th, 2016, capturing the extent of each patch. A 7.62 cm × 12.7 cm notecard was included in each photo for scale. Using the program ImageJ, the perimeter of each plant patch was traced to measure area. To estimate total nitrogen content in aboveground foliage, we reviewed published research conducted in the United States or Canada that used a LI-3100 or LAI-2000 Plant Canopy Analyzer Water 2018, 10, 1627 7 of 19

(Li-COR Inc., St. Lincoln, NE, USA) to calculate leaf area index. The average leaf area indexes for shrubs, grasses, and forbs were 2.1 [43–45], 3.7 [46–50], and 3.9 [51,52], respectively. We did not capture top-down photos for the bioswale trees. Instead, we did an approximate count of the total number of leaves for the Quercus palustris trees at ROWB 9A and ROWB 9B.

2.5. Statistical Analysis We conducted statistical analyses in R v. 3.1.3 (The R Project for Statistical Computing, 2015). We exclusively used non-parametric statistics, including Wilcoxon rank-sum tests to distinguish if different sites, sampling dates, and soil depths were statistically different from one another, and Spearman rank correlation coefﬁcients to identify spatial trends from the inlet to outlet of bioswales and to determine correlations between our measurements, air temperature and rainfall [53].

3. Results

3.1. Soil

3.1.1. Soil Characterization The shallow (7.6–22.9 cm) and deep (30.5–45.7 cm) soil samples were not statistically different from one another in terms of their soil texture, organic matter, total carbon, and total nitrogen (p > 0.05) (Table2).

Table 2. Soil macronutrient and texture means ± standard error, and Wilcoxon rank-sum test p-values demonstrating lack of statistical difference (p < 0.05) between shallow and deep samples.

Depth Total N (%) Total C (%) C:N ratio Organic (%) Sand (%) Silt (%) Clay (%) 7.6–22.9 cm (n = 7) 0.15 ± 0.02 2.9 ± 0.5 19.4 ± 0.4 5.0 ± 0.8 80.1 ± 1.2 13.4 ± 1.3 6.5 ± 0.4 30.5–45.7 cm (n = 7) 0.13 ± 0.01 2.4 ± 0.3 18.7 ± 1.2 4.1 ± 0.5 78.6 ± 1.9 13.5 ± 1.5 7.9 ± 0.6 p-value 0.71 0.80 0.22 0.75 0.48 0.95 0.14

3.1.2. Soil Mineralization and Nitrification Although the core housings were 50.8 cm in length, the average soil core length during the 15-month monitoring period was 36.5 cm, because 5 cm of the core housing was above ground and the soil core itself would often compact during sampling. Consequently, the shallow and deep halves of soil cores averaged 18.25 cm each. Soil inorganic nitrogen concentrations and transformation rates were more impacted by the depth than placement within the bioswale. The five soil cores evenly spaced down the length of the four sites (ROWB 26B, ROWB 9B, SGS 11, SGS 21), which were used to test within-site spatial variability from bioswale inlet to outlet, did not demonstrate an increasing or decreasing trend for soil nitrate (p = 0.127, n = 75), ammonium (p = 0.533, n = 75), mineralization (p = 0.939, n = 36), or nitrification (p = 0.639, n = 36) (Figure3). On the other hand, the shallow halves of soil cores for all sites overall had greater nitrate (p = 0.008, n = 423), mineralization (p = 0.044, n = 209) and nitrification (p = 0.043, n = 209) compared to the deep halves, but had no overall difference for ammonium (p = 0.131, n = 423). The seven different sites produced similar results to one another (Figures4 and5), although site ROWB 23 had the most soil inorganic nitrogen in soil extracts, with statistically greater concentrations of extractable ammonium than SGS 21 (p = 0.019, n = 119) and SGS 11 (p = 0.047, n = 119). ROWB 23 also had greater extractable nitrate than SGS 21 (p = 0.006, n = 119), SGS 2 (p = 0.011, n = 104), and ROWB 9B (p = 0.014, n = 117). Inorganic nitrogen concentrations, mineralization, and nitrification rates did not show statistically significant differences (p < 0.05) among the other sites. Within each site, the shallow halves had statistically greater nitrification and mineralization for site ROWB 9A only. Water 2018, 10, x FOR PEER REVIEW 7 of 18

2.5. Statistical Analysis We conducted statistical analyses in R v. 3.1.3 (The R Project for Statistical Computing, 2015). We exclusively used non-parametric statistics, including Wilcoxon rank-sum tests to distinguish if different sites, sampling dates, and soil depths were statistically different from one another, and Spearman rank correlation coefficients to identify spatial trends from the inlet to outlet of bioswales and to determine correlations between our measurements, air temperature and rainfall [53].

3. Results

3.1. Soil

Table 2. Soil macronutrient and texture means ± standard error, and Wilcoxon rank-sum test p-values demonstrating lack of statistical difference (p < 0.05) between shallow and deep samples.

3.1.2. Soil Mineralization and Nitrification Although the core housings were 50.8 cm in length, the average soil core length during the 15- month monitoring period was 36.5 cm, because 5 cm of the core housing was above ground and the soil core itself would often compact during sampling. Consequently, the shallow and deep halves of soil cores averaged 18.25 cm each. Soil inorganic nitrogen concentrations and transformation rates were more impacted by the depth than placement within the bioswale. The five soil cores evenly spaced down the length of the four sites (ROWB 26B, ROWB 9B, SGS 11, SGS 21), which were used to test within-site spatial variability from bioswale inlet to outlet, did not demonstrate an increasing or decreasing trend for Water 2018soil, 10 nitrate, 1627 (p = 0.127, n = 75), ammonium (p = 0.533, n = 75), mineralization (p = 0.939, n = 36), or 8 of 19

nitrification (p = 0.639, n = 36) (Figure 3).

)

) -

4 1 1

2 g 0 (

depth m depth

(

o n i 0 to 18 cm

t 0 to 18 cm

i a

18 to 36 cm t 18 to 36 cm

c i

Water 2018, 10a , x FOR PEER REVIEW 8 of 18

Water 20180 , 10, x FOR PEER REVIEW f -1 8 of 18

e t

On theOn other the other hand, hand, the shallow the shallow halves halves of soil of coressoil cores forn all for sites all sitesoverall overall had greaterhad greater nitrate nitrate (p = (p = e 0.008, n = 423n -2 ), mineralization (p = 0.044, n = 209) and nitrification-2 (p = 0.043, n = 209) compared to the 0.008, n = 423a ), mineralizationb c d (pe = 0.044, n = 209) and nitrificationa b (p = 0.043c ,d n = 20e 9) compared to the deep halves, but hadwi tnohin-si overallte locati odifferencen for ammonium (p = 0.131,w nith =in 423-site )lo. ca tion deep halves, but had no overall difference for ammonium (p = 0.131, n = 423). The sevenThe seven different different sites (sitesproduceda) produced similar similar results results to one to an oneother another (Figures (Figures (4b and) 4 5) and, although 5), although site site ROWBROWB 23 had 23 the had most the most soil inorganic soil inorganic nitrogen nitrogen in soil in extracts, soil extracts, with withstatistically statistically greater greater concentrations concentrations Figure 3. Within-site spatial variability from bioswale inlet to outlet for (a) net mineralization and (b) of extractableFigureof extractable 3. Within-site ammonium ammonium spatial than variabilitythanSGS 21SGS (p 21 from= 0.019,(p = bioswale 0.019, n = 119) n inlet= and119) to SGSand outlet 11SGS for(p 11 (=a 0.047,)( netp = mineralization0.047, n = 119). n = 119).ROWB andROWB 23 (b ) net23 net nitrification. Nitrogen transformation rates (μg N− d−1 −g−1 dry soil) as calculated with Equation (1) alsonitriﬁcation. hadalso greaterhad greater Nitrogenextractable extractable transformation nitrate nitrate than thanSGS rates 21SGS ((µp g21 = N 0.006,(p d = 10.006, gn =1 119),dry n = soil)119),SGS as2SGS ( calculatedp =2 0.011,(p = 0.011, n with = 104), nEquation = 104),and and (1) quantify differences between initial and one-week incubations for shallow and deep halves of soil ROWquantifyROWB 9B B( differencesp9B = 0.014,(p = 0.014, n between = 117). n = 117).Inorganic initial Inorganic and nitrogen one-week nitrogen concentrations, incubations concentrations, for mineralization, shallow mineralization, and deep and halves nitrificationand nitrification of soil cores cores at four bioswales (ROWB 26B, ROWB 9B, SGS 11, SGS 21) (n = 36). “a” denotes the soil cores rates ratesdid not did show not show statistically statistically significant significant differences differences (p < 0.05)(p < 0.05)among among the other the other sites. sites. Within Within each each at fourclosest bioswales to the (ROWBinlet, and 26B, “e” denotes ROWB the 9B, soil SGS cores 11, closest SGS 21) to the (n =outlet. 36). “a”ROWB: denotes right-of the-way soil bioswale; cores closest site, the shallow halves had statistically greater nitrification and mineralization for site ROWB 9A tosite, the inlet,SGS:the shallow stormwater and “e” halves denotes greenstreet. had the statistically soil cores greater closest nitrif to theication outlet. and ROWB: mineralization right-of-way for site bioswale; ROWB SGS: 9A only.only. stormwater greenstreet.

* *

)

d d

( (

n depth

i t

0 t 0 0 to 180 cmto 18 cm a

a 18 to 3168 cmto 36 cm

n i

-2 i -2

n n

ROWBR O23WRB O2W3 BR O26WBBR 2O6BWBR O9AWBR 9OAWBR O9BWB SG9BS 1SG1 S 11SGS 2SGS 2SGS 2SG1 S 21 site site

−1 −1 FigureFigureFigure 4. 4.Net Net 4 . mineralization mineralizationNet mineralization rates rates rates(μg (µ nitrogen g (μ nitrogeng nitrogen (N) (N)d (−1N g) d −1d −1dry g−1 soil)drydry as soil) soil) calculated as as calculated calculated in Equation in Equation in (1)Equation (1) (1) quantifyingquantifyingquantifying differencesdifferences differences between between between initial initial initial and and one and one-week-week one- weekincubations incubationsincubations for all for soil for all cores all soil soil coresat coresseven at seven bioswales at seven bioswales bioswales sampledsampledsampled at at two two at depths depths two depths (n = 209) 209).(n =. *209) *signifies signiﬁes. * signifies bioswales bioswales bioswales where where where shallow shallow shallow (0 to (0 18(0 to tocm) 18 18 andcm) cm) deep andand deep(18 to (18 (3618 to to 36 36 cm) samplescm) samplescm) were samples were statistically werestatistically statistically different. different. different. ROWB: ROWB: ROWB: right-of-wayright- rightof-way-of -bioswale;way bioswale; bioswale; SGS: SGS: stormwaterSGS: stormwater stormwater greenstreet. greenstreet.greenstreet.

* *

)

1 -

2 - 2

N 1

g m

m depthdepth

(

( n 0 n 0 to 18 cm

o 0 0 to 18 cm

i i

t 18 to 36 cm

i r

-1 r

t -1

e n -2 n -2

ROWBR O23WRB O2W3 BR O26WBBR 2O6BWBR O9AWBR 9OAWBR O9BWB SG9BS 1SG1 S 11SGS 2SGS 2SGS 2SG1 S 21 site site

−1 −1 FigureFigureFigure 5. 5. NetN et5 . nitriﬁcation nNitrificationet nitrification rates rates r (atesμ (gµ nitrogeng (μ nitrogeng nitrogen (N) (N) d(−1N )g d −1d −1dry g−1 soildrydry) as soil soil) calculated) as as calculated calculated in Equation in Equation in Equation(1) (1) (1) quantifyingquantifyingquantifying differencesdifferences differences between between between initial initial initial and and one and one-week-week one- weekincubations incubationsincubations for all for soil for all cores all soil soil coresat coresseven at seven bioswales at seven bioswales bioswales sampledsampledsampled at at two two at depths depths two depths ((n = 20 209).(n9 )=. *20 *signifies9 signiﬁes). * signifies bioswales bioswales bioswales where where where shallow shallow shallow (0 to(0 18(0 to tocm) 18 18 and cm)cm) deep andand deep(18 to (18 (3618 to to 36 36 cm) samplescm) samplescm) were samples were statistically werestatistically stati different.stically different. different. ROWB: ROWB: ROWB: right-of-wayright- rightof-way-of -bioswale;way bioswale; bioswale; SGS: SGS: stormwaterSGS: stormwater stormwater greenstreet. greenstreet.greenstreet.

Water 2018, 10, 1627 9 of 19 WaterWater 20182018,, 1010,, xx FORFOR PEERPEER REVIEWREVIEW 99 ofof 1818

SoilSSoiloil extractable extractableextractable ammonium ammoniumammonium displayed displayed displayed a seasonal a a seasonal seasonal trend, as the trend, trend, greatest asas ammonium the the greategreatestst concentrations ammoniumammonium occurredconcentrationsconcentrations in the summer occurredoccurred (Figure inin thethe6 summer).summer Speciﬁcally, (Figure(Figure June 66)). 2015. Specifically,Specifically, demonstrated JuneJune statistically20152015 demonstrateddemonstrated greater ammonium statisticallystatistically concentrationsgreatergreater ammoniumammonium than concentrationsconcentrations every other period, thanthan everyevery with other theother exception periodperiod,, withwith of June thethe exceptionexception 2016. The ofof other JuneJune dates 2016.2016. ThewereThe otherother not statisticallydatesdates were were different not not statistically statistically from one different different another. from Junefrom 2015 one one was another another also.. theJuneJune only 2015 2015 period was was with also also statistically the the only only periodperiod signiﬁcant withwith differencesstatisticallystatistically between significsignificantant initial differencesdifferences and incubated betweenbetween concentrations. initialinitial andand incubatedincubated concentrations.concentrations.

0 i

( 20

(

40 m

60 m

60 )

)

6/156/15A*A* 7/157/15BB 8/158/15BB 9/159/15BB 10/1510/15BB 11/1511/15BB 12/1512/15BB 1/161/16 2/162/16 3/163/16BB 4/164/16BB 5/165/16BB 6/166/16ABAB 7/167/16BB 8/168/16BB )

) 7.5 month N month

7.5

p (

p type

( 5.0 type

5.0

m initial m

u initial

u i

n incubated

n incubated o

o 2.52.5

a a 0.00.0 6/156/15 7/157/15 8/158/15 9/159/15 10/1510/15 11/1511/15 12/1512/15 1/161/16 2/162/16 3/163/16 4/164/16 5/165/16 6/166/16 7/167/16 8/168/16 momonntthh

−1 −1 FigureFigureFigure 6.66.. SoilSoil extractableextractable ammoniumammonium ammonium ((parts (partsparts perper per millionmillion million nitrogennitrogen nitrogen ((ppmppm (ppm NN) N)) gg−1 g drydrydry soil)soil) soil) inin initialinitial in initial andand n andoneone-- one-weekweekweek incubationsincubations incubations atat sevenseven at seven bioswalebioswale bioswalesss sampledsampled sampled atat twotwo at depthsdepths two depths monthlymonthly monthly ((nn == 423)423) ( =ffroro 423)mm JuneJune from 20152015 June toto 2015AugustAugust to August 20162016,, omitting 2016,omitting omitting two two twowinter winter winter months months months... TheThe The lettersletters letters aboveabove above representrepresent represent statistically statistically significant significant significant differences among periods (two periods that share a letter are not statistically different), and * signifies differencesdifferences amongamong periodsperiods (two(two periodsperiods thatthat shareshare a a letterletter areare not not statisticallystatistically different),different), andand * * signifiessignifies periods where initial and incubated samples were statistically different. The bars at the top denote rain periodsperiods wherewhere initialinitial andand incubatedincubated samsamplesples werewere statisticallystatistically different.different. TheThe barsbars atat thethe toptop denotedenote amounts during the one-week incubation periods. rainrain amountsamounts duringduring thethe oneone--weekweek incubationincubation periods.periods. While soil extractable nitrate concentrations also displayed a possible seasonal trend, as the lowest WWhilehile s soiloil extractable extractable nitrate nitrate concentrations concentrations also also displayeddisplayed a a possible possible seasonal seasonal trend, trend, as as the the nitrate concentrations were found in the cooler months of November and March 2016 (Figure7), nitrate lowestlowest nitratenitrate concentrationsconcentrations werewere foundfound inin ththee coolercooler monthsmonths ofof NovemberNovember andand MarchMarch 20162016 (Figure(Figure 77)),, levels were more affected by precipitation. The only statistically significant increase between initial nnitrateitrate levelslevels werewere moremore affectedaffected byby precipitation.precipitation. TheThe onlyonly statisticallystatistically significantsignificant increaseincrease betweenbetween and incubated soil levels for nitrate occurred in September 2015 and December 2015, when rain did not initialinitial andand incubatedincubated soilsoil levelslevels forfor nitratenitrate occurredoccurred inin SeptemberSeptember 20152015 andand DecemberDecember 2015,2015, whenwhen rainrain occur during the incubation period. In contrast, in June 2016 there was a clear decrease between initial diddid not not occur occur during during the the incubation incubation period period.. InIn contrast, contrast, inin June June 20162016 ttherehere was was a a clear clear decreasedecrease and incubated soil levels for nitrate, as this period had the greatest rain depth (Figure7). Ammonium betweenbetween initialinitial andand incubatedincubated soilsoil llevelsevels forfor nitratenitrate,, asas thisthis periodperiod hadhad thethe greatestgreatest rainrain depthdepth (Figure(Figure 7)7).. levels however showed mostly seasonal variability (Figure6). AmmoniumAmmonium levelslevels howeverhowever showedshowed mostlymostly seasonalseasonal variabilityvariability (Figure(Figure 66))..

0 i

( 20

(

40 m

60 m

60 )

) 6/15AB 7/15AB 8/15AB 9/15A* 10/15AB* 11/15B 12/15AB* 1/16 2/16 3/16B 4/16AB 5/16AB 6/16AB* 7/16AB 8/16AB 10.0 6/15AB 7/15AB 8/15AB 9/15A* 10/15AB* 11/15B 12/15AB* 1/16 2/16 3/16B 4/16AB 5/16AB 6/16AB* 7/16AB 8/16AB

10.0 momonntthh

)

N 7.5

7.5 m

m type

p type

( p

initial ( 5.0

5.0 initial

e t

e incubated t

a incubated

i t

i 2.5

n 2.5 n

0.00.0 6/156/15 7/157/15 8/158/15 9/159/15 10/1510/15 11/1511/15 12/1512/15 1/161/16 2/162/16 3/163/16 4/164/16 5/165/16 6/166/16 7/167/16 8/168/16 momonntthh

−1 Figure 7. Soil extractable nitrate (parts per million nitrogen (ppm N)−1 g dry soil) in initial and FigureFigure 77.. SoilSoil extractableextractable nitratenitrate ((partsparts perper millionmillion nitrogennitrogen ((ppmppm NN)) gg−1 drydry soil)soil) inin initialinitial andand oneone-- one-week incubations at seven bioswales sampled at two depths monthly (n = 423) from June, 2015 to weekweek incubationsincubations atat sevenseven bioswalebioswaless sampledsampled atat twotwo depthsdepths monthlymonthly ((nn == 423)423) fromfrom June,June, 20152015 toto August, omitting two winter months. The letters above represent statistically signiﬁcant differences AugustAugust,, omittingomitting twotwo winterwinter months.months. TheThe lettersletters aboveabove representrepresent statisticallystatistically significantsignificant differencesdifferences among periods and * signiﬁes periods where initial and incubated samples were statistically different. amongamong periodsperiods andand ** signifiessignifies periodsperiods wherewhere initialinitial and and incubatedincubated samplessamples werewere statistically statistically different.different. The bars at the top denote rain amounts during the one-week incubation periods. TheThe barsbars atat thethe toptop denotedenote rainrain amountsamounts duringduring thethe oneone--wweekeek incubationincubation periods.periods.

Water 2018, 10, 1627 10 of 19 Water 2018, 10, x FOR PEER REVIEW 10 of 18

One ofof thethegreatest greatest net net nitriﬁcation nitrification rates rate wass was found found during during the the rain-free rain-free period period of September, of September, 2015 (Table2015 (Table3), and 3), the and lowest the lowest rate was rate found was duringfound Juneduring 2016, June when 2016, there when was there the largestwas the rain largest amount. rain However,amount. However, there were there no consistentwere no consisten patternst inpatterns mineralization in mineralization rates among rate samplings among sampling periods (Table periods3). There(Tablewas 3). There no increasing was no increasing or decreasing or decreasing trend over trend time fromover time 2015 from to 2016 2015 for to mineralization 2016 for mineralization (p = 0.502, n(p= = 209)0.502, or n nitriﬁcation = 209) or nitrification (p = 0.470, (np == 209).0.470, n = 209).

−1 −1 Table 3.3. MeansMeans ofof netnet mineralizationmineralization and and net net nitriﬁcation nitrification (µ (gμg nitrogen nitrogen (N) (N) d d−1g g−1 dry soil) averaged for shallowshallow andand deepdeep halveshalves ofof soilsoil corescores atat sevenseven bioswalesbioswales ((nn == 209).209). LettersLetters indicateindicate statisticallystatistically signiﬁcantsignificant differences amongamong periods.periods.

MonthMonth of of Incubation Incubation Period Period NetNet Mineralization Mineralization (µ g(µg N dN− d1−1g −g−11) N Netet Nitrification Nitriﬁcation ((µgµg NN dd−−11 gg−1−) 1) JuneJune 2015 2015 −0.81−0.81 bb 0.01 0.01 ab ab JulyJuly 2015 2015 0.680.68 aa 0.49 0.49 ab ab AugustAugust 2015 2015 −0.11−0.11 abab −−0.050.05 ab ab SeptemberSeptember 2015 2015 0.020.02 abab 0.30 0.30 a a OctoberOctober 2015 2015 −0.11−0.11 abab −−0.310.31 b b NovemberNovember 2015 2015 0.100.10 abab 0.03 0.03 ab ab DecemberDecember 2015 2015 0.040.04 abab 0.11 0.11 ab ab March 2016 0.23 ab 0.07 ab March 2016 0.23 ab 0.07 ab April 2016 0.28 ab 0.12 ab April 2016 0.28 ab 0.12 ab May 2016 −0.05 ab 0.02 ab JuneMay 2016 2016 −0.72−0.05 bab −0.020.50 ab b JulyJune 2016 2016 0.19−0.72 abb −0.50 0.12 b a AugustJuly 2016 2016 0.200.19 abab 0.12 0.29 a ab August 2016 0.20 ab 0.29 ab

Soil mineralization and nitrification followed a “one to one” relationship, especially for shallow Soil mineralization and nitrification followed a “one to one” relationship, especially for shallow soils (Figure8). This ratio implies that nearly all organic nitrogen was eventually transformed into soils (Figure 8). This ratio implies that nearly all organic nitrogen was eventually transformed into nitrate, because ammonium resulting from mineralization was quickly nitrified and stored as nitrate. nitrate, because ammonium resulting from mineralization was quickly nitrified and stored as nitrate. Shallow samples generally had greater nitrification than deeper samples (Figure8a). The full soil data Shallow samples generally had greater nitrification than deeper samples (Figure 8a). The full soil data are found in the Supplemental Material.

are found in the Supplemental Material.

)

1 t

- 3

- i

2 t

a 2

1 i

( 0 /

0 to 18 cm n

n 18 to 36 cm

o o

-1 i

a z

i 0 l

c -2

r e

t -3

m -1

t e

-6 -4 -2 0 2 4 e

n n 0 to 18 cm 18 to 36 cm net mineralization (mg N d-1 g-1 ) depth (a) (b)

Figure 8. 8.RelationshipRelationship between net between nitrogen net mineralization nitrogen mineralizationand net nitrification and (μg net nitrogen nitriﬁcation (N) d−1 (gµ−1g dry nitrogen soil) (N)for shallow d−1 g−1 anddry deep soil) halves for shallow of soil and cores deep at seven halves bioswale of soils cores (n = 20 at9 seven) (a) Dotted bioswales line (denotesn = 209) 1:1 (a) ratio Dotted; (b line) Net denotes mineralization 1:1 ratio;/n (itrificationb) Net mineralization/nitriﬁcation ratio. ratio.

3.2.3.2. Soil Gas Emissions µ −2 −1 µ Soil N 2OO gas gas emissions emissions overall overall averaged averaged 35 35 μg gN N2O2-O-NN m−2 m h−1, hwith, witha maximum a maximum of 149 of μ 149g N2Og- −2 −1 −2 −1 −2 −1 µ −2 −1 N2 mO-N h m, andh one, and negative one negative flux (N2 fluxO consumption) (N2O consumption) of −78 μg of N−2O78-N gm N 2hO-N. As m shownh .in As Figure shown 9, inthe Figure two sites9, the where two gases sites where were sampled gases were were sampled not statistically were not different statistically from different each other from (p = each 0.959) other but (thep = six 0.959 measured) but the summer six measured fluxes were summer statistically fluxes were greater statistically than the greaterten fall fluxes than the (p = ten 0.042). fall fluxesThere were neither significant correlations with 3-day antecedent rain (p = 0.515) nor with temperature (p = 0.081). The full gas data are found in the Supplemental Material.

Water 2018, 10, 1627 11 of 19

(p = 0.042). There were neither signiﬁcant correlations with 3-day antecedent rain (p = 0.515) nor with Water 2018, 10, x FOR PEER REVIEW 11 of 18 temperatureWater 2018, 10, (xp FOR= 0.081). PEER REVIEW The full gas data are found in the Supplemental Material. 11 of 18

) 150

1 )

- 150

2 -

m 100

(

m 50 Season

( x Season

u 50 x

l Summer

F l

FallSummer

F e

Fall

d e

i 0

d x

i 0

u s

o -50

r t

o -50

N N ROWB 9B SGS 21 ROWB 9B Site SGS 21 Site

Figure 9. Soil nitrous oxide gas emissions (n = 16). FigureFigure 9. 9.Soil Soil nitrousnitrous oxide oxide gas gas emissions emissions ( n(n= =16). 16). 3.3. Plant Uptake 3.3. Plant Uptake The shrubs shrubs,, including including RRosaosa ‘Radrazz’‘Radrazz’ Knockout Knockout, Spiraea, Spiraea tomentosa, tomentosa, andand AroniaAronia melanocarpa melanocarpa, had, The shrubs, including Rosa ‘Radrazz’ Knockout, Spiraea tomentosa, and Aronia melanocarpa, had hadlower lower foliar foliar nitrogen nitrogen per perbioswale bioswale area area (Fig (Figureure 10 10a) a)than than the the grasses grasses (p (p == 0.003 0.003),), which included included lower foliar nitrogen per bioswale area (Figure 10a) than the grasses (p = 0.003), which included PanicumPanicum virgatum virgatum andand PennisetumPennisetum alopecuroides alopecuroides, and, andthe forbs the ( forbsp = 0.0001 (p =), which 0.0001), include whichd Echinacaea included Panicum virgatum and Pennisetum alopecuroides, and the forbs (p = 0.0001), which included Echinacaea Echinacaeapurpurea, Hemerocallis purpurea, Hemerocallis ‘Lady Florence’‘Lady, Symphyotrichum Florence’, Symphyotrichum novae-angliae novae-angliae, Liriope muscari, Liriope, and muscari Nepeta, purpurea, Hemerocallis ‘Lady Florence’, Symphyotrichum novae-angliae, Liriope muscari, and Nepeta andfaassenniiNepeta. The faassennii grasses. and The forbs grasses did and not forbsshow didstatistically not show significant statistically differences signiﬁcant for differencesfoliar nitrogen for faassennii. The grasses and forbs did not show statistically significant differences for foliar nitrogen foliarper bioswale nitrogen area per ( bioswalep = 0.333 area). Hemerocallis (p = 0.333). hadHemerocallis the most,had averaging the most, 8.0averaging g N m−2, with 8.0 g a N maximum m−2, with of a per bioswale area (p = 0.333). Hemerocallis had the most, averaging 8.0 g N m−2, with a maximum of maximum11.6 g N m of−2. 11.6 g N m−2. 11.6 g N m−2.

A B B A B B 2

1 A B B

A B B 7

)

(

. 0

Shrub Grass Forb 0 Shrub Grass Forb Shrub Grass Forb Shrub Grass Forb (a) (b) (a) (b) Figure 10.10. ((aa)) Foliar Foliar nitrogen nitrogen per per bioswale bioswale area area (n ( n= 42=); 42); (b) ( bFoliar) Foliar nitrogen nitrogen per perleaf leaf area area (n = (179)n =. 179). The Figure 10. (a) Foliar nitrogen per bioswale area (n = 42); (b) Foliar nitrogen per leaf area (n = 179). The Theletters letters at the at top the toprepresent represent statistically statistically significant signiﬁcant differences differences among among groups groups (two (two groups groups that that share share a letters at the top represent statistically significant differences among groups (two groups that share a aletter letter are are not not statistically statistically different) different).. letter are not statistically different). The shrubs also had about one-thirdone-third lower nitrogennitrogen onon aa per leaf area basis (Figure 1010b)b) thanthan the The shrubs also had about one-third lower nitrogen on a per leaf area basis (Figure 10b) than the grasses ((pp< < 0.0001)0.0001) and and the the forbs forbs (p (

Water 2018, 10, 1627 12 of 19

16.7 (p < 0.0001). The only index where shrubs indicated greater uptake was patch area, for which the shrubs averaged 0.5 m2, while the shrubs and grasses each averaged 0.4 m2, although patch area for shrubs was neither statistically greater than patch area for the grasses (p = 0.539) or the forbs (p = 0.408). Total foliar nitrogen for shrubs, grasses, and forbs at the four sites averaged 30.2 g N and the trees contained an additional 14.1 g N on average. Total nitrogen content in aboveground foliage at the 9.3 m2 sites averaged 4.8 g N m−2. The full plant data are found in the Supplemental Material.

4. Discussion

4.1. Soil

4.1.1. Soil Characterization Our data quantify nitrogen cycling in the soil, gas, and plant phase. These are poorly documented ﬂuxes of the bioswale nitrogen cycle that occur during dry weather. In particular, the data indicate that bioswale soils contain overabundant nitrogen in excess of plant and microbial uptake. As such, the C:N ratios of the soils were below 25:1 (Table2), which is considered the critical C:N ratio. Above this ratio, microbes may be nitrogen limited and below it there may be net nitrogen release from decomposing organic matter [16,54]. Soils with greater C:N ratios tend to be more capable of removing nitrogen from stormwater by promoting microbial uptake (immobilization) over mineralization [12,21,55,56].

4.1.2. Soil Mineralization and Nitrification Soil nitrogen concentrations did not vary with sampling location, as there were few differences between sites and no clear spatial trends from the inlet to the outlet. However, shallower bioswale soil samples had greater soil nitrate concentrations, mineralization rates, and nitrification rates compared to deeper soil, which might be expected because inputs from stormwater influent and decomposing plant organic matter increase nitrogen levels in shallow soil. In most soils, the quantity and quality of detrital inputs and fertilization are the main factors that control rates and patterns of mineralization and immobilization [16,36]. Shallow soil is also topped with 7 cm of bark chip mulch, which can remove organic nitrogen from stormwater runoff through sorption processes, and also support substantial microbe populations to mineralize and nitrify captured organic nitrogen between rain events [10]. Moreover, shallow soils experience greater temperature fluctuations, supporting faster decomposition than found in deeper soil [57]. Likewise, moisture levels and temperature are more stable in the subsoil [58], and deeper soils retain less carbon and nitrogen [7]. There are also increased oxygen levels closer to the surface, supporting autotrophic nitrification and aerobic metabolism, which are aerobic processes. Other bioswale monitoring studies found most total nitrogen [59] and nitrate [60] to be produced in shallow soils, suggesting that nitrification mainly occurs in the upper soil layers. The overabundance of nitrogen in shallow soils is also evidenced by the 1:1 relationship between net mineralization and net nitrification in Figure8a,b. Our research shows that in urban bioswales, nitrogen accumulates as nitrate, rarely as ammonium. Net nitrification is typically 100% of net mineralization in nitrogen-abundant tropical ecosystems [61] and agricultural systems [13], while only a small fraction of net mineralization in temperate ecosystems [62], because nitrifying microbes thrive in nitrogen-abundant environments, but compete poorly for ammonium against plants and heterotrophic microbes when nitrogen is scarce. In nitrogen-abundant environments, nitrifiers live in close association with mineralizers, making nitrate the dominant nitrogen form moving through the soil, and plants rely on nitrate for nitrogen [13]. However, gross rates of nitrification exceed rates of nitrate uptake by plants and microorganisms, resulting in the accumulation of nitrate; this indicates that excess nitrogen is cycling relative to the ability of plants and microbes to assimilate it [22]. Our data suggest that bioswale soils supply much more nitrogen than the plants need, especially at shallow Water 2018, 10, 1627 13 of 19 depths. The excess nitrogen may not only originate from the initial soil design, but also from inputs from stormwater nitrogen and decomposing plant organic nitrogen that the soil makes available through mineralization and nitrification. We also found that the overabundance of nitrogen increased in warmer months, as indicated by greater concentrations of ammonium in the summer (Figure6). Warmer weather stimulates the mineralization process [63,64], and a similar trend is visible in grassland ecosystems, which show increased ammonium and nitrate concentrations in the summer and decreased concentrations in the fall [65]. Compared to ammonium levels, levels of nitrate fluctuated greatly through high rates of accumulation and leaching (Figures7 and8). The lower left quadrant of Figure8a shows a tight relationship between mineralization and nitrification rates, indicating only nitrate levels to be decreasing, as nitrate is a highly mobile anion that does not bind to soil particles and so is readily washed out [10,16], whereas ammonium is stable.

4.2. Soil Gas Emissions We also found greater nitrous oxide emissions in summer than fall. This increase might be due to the connection between warmer temperatures and high rates of nitrous oxide gas emissions from nitrification and denitrification [27,66]. High levels of ammonium in the soil also leads to increased nitrification and denitrification [15,54], so the summer’s increase in ammonium (Figure6) possibly further increased summer gas emissions of nitrous oxide. Our results might be applicable to bioswales in other cities because the levels of nitrous oxide emitted by our bioswales were similar to rates found in similar studies around the world: our mean −2 −1 reported N2O flux of 35 µg N m h was in the mid-range, as per the mean N2O fluxes of 13.8 and −2 −1 65.6 µg N2O-N m h found in two bioswales in Melbourne, Australia [24], the 23.4 µg of N2O-N −2 −1 −2 −1 m h found in a bioswale in New York, USA [21], and the 10 µg of N2O-N m h found in −2 −1 bioswales in Vermont, USA [67]. Our peak emissions of 148.6 µg N2O-N m h were lower than the −2 −1 −2 −1 peak of 1100 µg N2O-N m h found by Grover et al. [24] and the 330 µg of N2O-N m h found −2 −1 by Shrestha et al. [67], but similar to the 125 µg N2O-N m h found by McPhillips et al. [21].

4.3. Plant Uptake Compared to the forbs and grasses, the leaves of the shrubs had lower nitrogen per site area, lower nitrogen per leaf area, and greater C:N ratios, all showing a lower rate of nitrogen uptake. While Collins et al. [5] suggested that trees and shrubs may remove more nitrogen in the long run due to deeper rooting systems and greater biomass, Turk et al. [68] found that herbaceous perennials outperformed trees and shrubs on nitrogen removal per unit area. Plants with the most nitrogen removal grow rapidly in nitrogen-rich environments and have the ability to store excess nitrogen [59]. Perhaps the grasses and perennial herbs were growing more rapidly, while the shrubs were storing excess nitrogen in woody biomass [32], which was not sampled in this study.

4.4. Nitrogen Mass Balance We estimated minimum and maximum fluxes for each aspect of the nitrogen cycle by extrapolating from our measurements with results from other literature as explained in the Supplemental Material. We found that soil mineralization amounted to 0 to 110 g N m−2 y−1, gas losses were 1 to 10 g N m−2 y−1, and plant uptake was 0 to 5 g N m−2 y−1. We also compared our results to water measurements conducted at the same sites, also as explained in the Supplemental Material. Stormwater influent added 84 to 591 g N m−2 y−1, while 29 to 951 g N m−2 y−1 infiltrated via stormwater into the subsoil, and 19 to 332 g N m−2 y−1 overflowed back into the sewer system (Figure 11). Water 2018, 10, 1627 14 of 19 Water 2018, 10, x FOR PEER REVIEW 14 of 18

Inputs and Outputs for N cycle

1 to 10 g N m-2 Gas emissions

Mineralization 0 to 110 g N m-2 -2 Influent 0 to 5 g N m 84 to 591 g N m-2 Plant uptake 19 to 332 g N m-2 Overflow

Infiltrate 29 to 951 g N m-2

Figure 11.11. Approximate nitrogen mass balance with site ROWB 9A in the foreground and site ROWB 9B in the background.

4.5. LimitationsThus, bioswale nitrogen cycling is dominated by liquid and soil phase fluxes, with relatively much smaller fluxes in the gas and plant phases. In contrast to natural ecosystems, where external Our mineralization data were limited by periodic rains washing away accumulated nitrate, inputs of nitrogen are about 10% of the amount of nitrogen that annually cycles [54], stormwater preventing a clear measurement of inorganic nitrogen changes. However, this condition allowed us inputs of nitrogen in the studied urban bioswales constitute most of the nitrogen that annually cycles. to contrast the fluctuating changes of soil nitrate due to leaching with the more gradual seasonal We found that gas emissions were a small portion of the overall mass balance, which was similar to changes of soil ammonium. Future research could quantify such washouts by sealing off the top and findings by Payne et al. [69] and Morse et al. [70], who found that N O emissions were only 1 to 2% of bottom of incubated soil cores with ion exchange resin bags, which2 prevent loss of ammonium and the incoming nitrogen amount. nitrate [37], and would clarify the negative fluxes we often found for mineralization and nitrification 4.5.due Limitationsto rainfall. The act of taking net mineralization measurements has the side effect of cutting soil roots with the sharpenedOur mineralization edge of the data soil werecorer, limited eliminating by periodic plant uptake rains of washing nitrogen away sources, accumulated which alt nitrate,ers the preventingnitrogen system, a clear enabling measurement higher ofmicrobe inorganic uptake nitrogen and lower changes. net However,mineralization this condition[13]. In addition, allowed the us tosoil contrast core may the sever fluctuating roots that changes nevertheless of soil nitratemay continue due to to leaching take up with nitrogen the more within gradual the core, seasonal again changesresulting of in soil underestimates ammonium. of Future net mineralizatio research couldn [14,36 quantify]. On such the other washouts hand, by the sealing incubated off the core top may and bottomhave slightly of incubated greater soil soil moisture cores with than ion the exchange rest of the resin site bags, [13,37 which], potentially prevent leading loss of ammoniumto overestimates and nitrateof the [mineralization37], and would rate clarify [57 the]. Ultimately, negative fluxes no we method often forfound assessing for mineralization soil nitrogen and provides nitrification an dueunequivocal to rainfall. estimate [37]. Our intact soil core method may reduce soil disturbance compared to other field Themethods act of of taking quantifying net mineralization soil mineralization, measurements which hasare themore side reliable effect than of cutting laboratory soil roots incubations with the sharpened[36]. edge of the soil corer, eliminating plant uptake of nitrogen sources, which alters the nitrogen system,We enablingfound that higher nitrogen microbe content uptake in aboveground and lower net foliage mineralization totaled about [13]. 5 In g addition,m−2. However, the soil plants core may severstore nitrogen roots that in nevertheless areas other maythan continueleaves. Total to take plant up nitrogen nitrogen uptake within theper core,year againcould resultingbe much ingreater, underestimates as a study offound net mineralizationwetland plant nitrogen [14,36]. Onuptake the otherlevels hand,to be 51 the g incubatedN m−2 y−1 [ core10], and may Lucas have slightlyand Greenway greater soil[17] moisturefound that than nitrogen the rest uptake of the siteranged [13, 37from], potentially 51 to 65 g leading N m−2 y to−1 overestimatesfor plants grown of the in mineralizationlab mesocosms. rateFuture [57 ].analyses Ultimately, should no consider method forall plant assessing parts, soil including nitrogen stems, provides roots, an flowers, unequivocal pods, estimateand seeds. [37 ]. Our intact soil core method may reduce soil disturbance compared to other field methods of quantifyingAs discussed soil mineralization, in the Supplemental which areMaterial more, reliable this study than relied laboratory on numerous incubations assumptions [36]. in generalizing monitored measurements into an annual budget. While our simplifying assumptions

Water 2018, 10, 1627 15 of 19

We found that nitrogen content in aboveground foliage totaled about 5 g m−2. However, plants may store nitrogen in areas other than leaves. Total plant nitrogen uptake per year could be much greater, as a study found wetland plant nitrogen uptake levels to be 51 g N m−2 y−1 [10], and Lucas and Greenway [17] found that nitrogen uptake ranged from 51 to 65 g N m−2 y−1 for plants grown in lab mesocosms. Future analyses should consider all plant parts, including stems, roots, ﬂowers, pods, and seeds. As discussed in the Supplemental Material, this study relied on numerous assumptions in generalizing monitored measurements into an annual budget. While our simplifying assumptions limit the precision of our overall estimate, even a conservative estimate of inputs to the studied bioswales in stormwater (84–591 g N m−2 y−1) and mineralization (0–110 g N m−2 y−1) greatly exceed fertilization guidelines (5–20 g N m−2 y−1)[30]. Future research should study how to provide balanced nutrition for vegetation when planted in soil with overabundant nitrogen. Future studies should also test the degree to which plants can absorb the nitrogen in stormwater given that it comes in rapid waves.

4.6. Implications Bioswale soil could better account for the overabundance of nitrogen within stormwater, in order to reduce nitrogen leaching. Detailed design guidance is not available for bioswales [71], and high organic matter contents are frequently specified for bioswale soil, intended to aid plant growth rather than improve water quality [21,72]. Due to this finding we advocate for more stable and more carbon-based organic matter used in soil mixes to increase the C:N ratio for better nitrogen management. The engineered soil specification for NYC bioswales (Supplemental Materials) specifies leaf compost as the only approved amendment to achieve the required organic matter content of 3 to 8%. While leaf compost is already high in carbon, the soil specification could further require that compost stability meet a requirement, such as (1) a Solvita maturity index score of 7 or 8 [30,73], or (2) less than 3 mg CO2-C per g compost organic matter per day [74,75]. Furthermore, the soil specification total nitrogen maximum of 0.25% could be reduced to the 0.15% levels we found in situ (Table2), given that the levels in our study were nitrogen rich. Our data indicate that stormwater and mineralization inputs of nitrogen are more than enough to promote healthy vegetation. A less nitrogen rich soil composition with stable soil carbon enhancements such as coconut coir pith as a replacement for compost [76] might enhance long-term net immobilization in order to reduce nitrogen leaching. We also suggest that the bioswales be maintained without additional fertilizers. Although plant health was not quantified in this study, excess nitrogen is not only linked with increased pollution, but even with declines in forest productivity and increased tree mortality [54]. Surplus fertilizer promotes shoot growth and diverts resources from root growth and production of defensive chemicals, which decreases drought tolerance and pest resistance [77]. When planted in a less nitrogen rich soil, native plants may also better outcompete weeds, potentially reducing maintenance [78]. Our recommendations may therefore support healthier plants with reduced maintenance, in addition to decreasing urban sources of nitrogen pollution.

Supplementary Materials: The following are available online at http://www.mdpi.com/2073-4441/10/11/1627/ s1. Author Contributions: Conceptualization, N.S., D.N.L.M., K.M., and P.C.; Formal analysis, N.S., B.M., and P.C.; Funding acquisition, P.C.; Investigation, N.S., R.H., J.H., and B.M.; Methodology, N.S., R.H., B.M., M.P., D.N.L.M., K.M., and W.M.; Project administration, P.C.; Resources, B.M., M.P., D.N.L.M., and P.C.; Supervision, B.M., M.P., and P.C.; Writing—original draft, N.S.; Writing—review and editing, R.H., J.H., B.M., M.P., D.N.L.M., K.M., and P.C. Funding: This work was funded, in part, by the National Science Foundation awards CMMI-1325676 and CBET-1444745, and the Environmental Protection Agency contract EP-15-C-000016. Water 2018, 10, 1627 16 of 19

Acknowledgments: The authors wish to thank the NYC Department of Parks & Recreation and the NYC Department of Environmental Protection for supporting our monitoring work. Any opinions, findings, and conclusions expressed in this paper are those of the authors and not meant to represent the views of any supporting institution. Conflicts of Interest: The authors declare no conflicts of interest.

References

1. Howarth, R.W.; Marino, R. Nitrogen as the Limiting Nutrient for Eutrophication in Coastal Marine Ecosystems: Evolving Views over Three Decades. Limnol. Oceanogr. 2006, 51, 364–376. [CrossRef] 2. Bettez, N.D.; Groffman, P.M. Nitrogen Deposition in and near an Urban Ecosystem. Environ. Sci. Technol. 2013, 47, 6047–6051. [CrossRef][PubMed] 3. Clar, M.; Green, R. Design Manual for Use of Bioretention in Stormwater Management; Department of Environmental Resources: Prince George’s County, MD, USA, 1993. 4. Bratieres, K.; Fletcher, T.D.; Deletic, A.; Zinger, Y. Nutrient and Sediment Removal by Stormwater Biofilters: A Large-Scale Design Optimisation Study. Water Res. 2008, 42, 3930–3940. [CrossRef][PubMed] 5. Collins, K.A.; Lawrence, T.J.; Stander, E.K.; Jontos, R.J.; Kaushal, S.S.; Newcomer, T.A.; Grimm, N.B.; Cole Ekberg, M.L. Opportunities and Challenges for Managing Nitrogen in Urban Stormwater: A Review and Synthesis. Ecol. Eng. 2010, 36, 1507–1519. [CrossRef] 6. Hunt, W.F.; Jarrett, A.R.; Smith, J.T.; Sharkey, L.J. Evaluating Bioretention Hydrology and Nutrient Removal at Three Field Sites in North Carolina. J. Irrig. Drain. Eng. 2006, 132, 600–608. [CrossRef] 7. Li, L.; Davis, A.P. Urban Stormwater Runoff Nitrogen Composition and Fate in Bioretention Systems. Environ. Sci. Technol. 2014, 48, 3403–3410. [CrossRef][PubMed] 8. Passeport, E.; Hunt, W.F. Asphalt Parking Lot Runoff Nutrient Characterization for Eight Sites in North Carolina, USA. J. Hydrol. Eng. 2009, 14, 352–361. [CrossRef] 9. Peterson, I.J.; Igielski, S.; Davis, A.P. Enhanced Denitrification in Bioretention Using Woodchips as an Organic Carbon Source. J. Sustain. Water Built Environ. 2015, 9, 1–9. [CrossRef] 10. Davis, A.P.; Shokouhian, M.; Sharma, H.; Minami, C. Water Quality Improvement through Bioretention Media: Nitrogen and Phosphorus Removal. Water Environ. Res. 2006, 78, 284–293. [CrossRef][PubMed] 11. McNett, J.K.; Hunt, W.F.; Davis, A.P. Influent Pollutant Concentrations as Predictors of Effluent Pollutant Concentrations for Mid-Atlantic Bioretention. J. Environ. Eng. 2011, 137, 790–799. [CrossRef] 12. Payne, E.G.I.; Fletcher, T.D.; Cook, P.L.M.; Deletic, A.; Hatt, B.E. Processes and Drivers of Nitrogen Removal in Stormwater Biofiltration. Crit. Rev. Environ. Sci. Technol. 2014, 44, 796–846. [CrossRef] 13. Schimel, J.P.; Bennett, J. Nitrogen Mineralization: Challenges of a Changing Paradigm. Ecology 2004, 85, 591–602. [CrossRef] 14. Robertson, P.G.; Wedin, D.; Groffman, P.M.; Blair, J.M.; Holland, E.A.; Nadelhoffer, K.J.; Harris, D. Soil Carbon and Nitrogen Availability: Nitrogen Mineralization, Nitrification, and Soil Respiration Potentials. In Standard Soil Methods for Long-Term Ecological Research; Robertson, G.P., Bledsoe, C.S., Coleman, D.C., Sollins, P., Eds.; Oxford University Press: New York, NY, USA, 1999; pp. 258–271. 15. Groffman, P.M.; Boulware, N.J.; Zipperer, W.C.; Pouyat, R.V.; Band, L.E.; Colosimo, M.F. Soil Nitrogen Cycle Processes in Urban Riparian Zones. Environ. Sci. Technol. 2002, 36, 4547–4552. [CrossRef][PubMed] 16. Robertson, G.P.; Groffman, P.M. Chapter 14—Nitrogen Transformations. In Soil Microbiology Ecology and Biochemistry; Elsevier Inc.: Amsterdam, The Netherlands, 2015; pp. 421–446. 17. Lucas, W.C.; Greenway, M. Hydraulic Response and Nitrogen Retention in Bioretention Mesocosms with Regulated Outlets: Part II—Nitrogen Retention. Water Environ. Res. 2011, 83.[CrossRef] 18. Shetty, N.H.; Culligan, P.J.; Mailloux, B.; McGillis, W.R.; Do, H.Y. Bioretention Infrastructure to Manage the Nutrient Runoff from Coastal Cities. In Geo-Chicago 2016 GSP 273; ASCE: Reston, VA, USA, 2016; pp. 402–411. 19. Brown, R.A.; Birgand, F.; Hunt, W.F. Analysis of Consecutive Events for Nutrient and Sediment Treatment in Field-Monitored Bioretention Cells. Water Air Soil Pollut. 2013, 224, 1581. [CrossRef] 20. Lucas, W.C.; Greenway, M. Nutrient Retention in Vegetated and Nonvegetated Bioretention Mesocosms. J. Irrig. Drain. Eng. 2008, 134, 613–623. [CrossRef] 21. McPhillips, L.; Goodale, C.; Walter, M.T. Nutrient Leaching and Greenhouse Gas Emissions in Grassed Detention and Bioretention Stormwater Basins. J. Sustain. Water Built Environ. 2018, 4, 1–10. [CrossRef] Water 2018, 10, 1627 17 of 19

22. Davidson, E.A.; Keller, M.; Erickson, H.E.; Verchot, L.V.; Veldkamp, E. Testing a Conceptual Model of Soil Emissions of Nitrous and Nitric Oxides. Bioscience 2000, 50, 667. [CrossRef] 23. Templer, P.H.; Pinder, R.W.; Goodale, C.L. Effects of Nitrogen Deposition on Greenhouse-Gas Fluxes for Forests and Grasslands of North America. Front. Ecol. Environ. 2012, 10, 547–553. [CrossRef] 24. Grover, S.P.P.; Cohan, A.; Chan, H.S.; Livesley, S.J.; Beringer, J.; Daly, E. Occasional Large Emissions of Nitrous Oxide and Methane Observed in Stormwater Biofiltration Systems. Sci. Total Environ. 2013, 465, 64–71. [CrossRef][PubMed] 25. McPhillips, L.E.; Groffman, P.M.; Schneider, R.L.; Walter, M.T. Nutrient Cycling in Grassed Roadside Ditches and Lawns in a Suburban Watershed. J. Environ. Qual. 2016, 45, 1901. [CrossRef][PubMed] 26. Oertel, C.; Matschullat, J.; Zurba, K.; Zimmermann, F.; Erasmi, S. Greenhouse Gas Emissions from Soils—A Review. Chem. Erde Geochem. 2016, 76, 327–352. [CrossRef] 27. Livesley, S.J.; Dougherty, B.J.; Smith, A.J.; Navaud, D.; Wylie, L.J.; Arndt, S.K. Soil-Atmosphere Exchange of Carbon Dioxide, Methane and Nitrous Oxide in Urban Garden Systems: Impact of Irrigation, Fertiliser and Mulch. Urban Ecosyst. 2010, 13, 273–293. [CrossRef] 28. Townsend-Small, A.; Czimczik, C.I. Carbon Sequestration and Greenhouse Gas Emissions in Urban Turf. Geophys. Res. Lett. 2010, 37, 1–5. [CrossRef] 29. Townsend-Small, A.; Pataki, D.E.; Czimczik, C.I.; Tyler, S.C. Nitrous Oxide Emissions and Isotopic Composition in Urban and Agricultural Systems in Southern California. J. Geophys. Res. 2011, 116, G01013. [CrossRef] 30. Goatley, M.; Hensler, K. Urban Nutrient Management Handbook; Virginia Cooperative Extension: Petersburg, VA, USA, 2011. 31. Roy-Poirier, A.; Champagne, P.; Filion, Y. Review of Bioretention System Research and Design: Past, Present, and Future. J. Environ. Eng. 2010, 136, 878–889. [CrossRef] 32. Payne, E.; Fletcher, T.D.; Russell, D.G.; Grace, M.R.; Cavagnaro, T.R.; Evrard, V.; Deletic, A.; Hatt, B.E.; Cook, P.L.M. Temporary Storage or Permanent Removal? The Division of Nitrogen between Biotic Assimilation and Denitrification in Stormwater Biofiltration Systems. PLoS ONE 2014, 9, e90890. [CrossRef] [PubMed] 33. Gilchrist, S.; Borst, M.; Stander, E.K. Factorial Study of Rain Garden Design for Nitrogen Removal. J. Irrig. Drain. Eng. 2014, 140, 04013016. [CrossRef] 34. Shetty, N.H. New York City’s Green Infrastructure: Impacts on Nutrient Cycling and Improvements in Performance. Ph.D. Thesis, Columbia University, New York, NY, USA, 2018. 35. ASTM. Standard Practice for Soil Exploration and Sampling by Auger Borings D1452/D1452M—16; ASTM: West Conshohocken, PA, USA, 2016; pp. 3–9. 36. Raison, R.J.; Connel, M.J.; Khanna, P.K. Methodology for Studying Fluxes of Soil Mineral-N in Situ. Soil Biol. Biochem. 1987, 19, 521–530. [CrossRef] 37. Hart, S.C.; Stark, J.M.; Davidson, E.A.; Firestone, M.K. Nitrogen Mineralization, Immobilization, and Nitrification. In Methods of Soil Analysis, Part 2. Microbiological and Biochemical Properties; Weaver, R.W., Angle, J.S., Bottomley, P.J., Bezdiek, D.F., Smith, M.S., Tabatabai, M.A., Eds.; Soil Science Society of America: Madison, WI, USA, 1994; pp. 985–1018. 38. Klute, A. Methods of Soil Analysis: Part 1—Physical and Mineralogical Methods, 2nd ed.; Soil Science Society of America: Madison, WI, USA, 1986; Volume 9. 39. Holmes, R.M.; Aminot, A.; Kerouel, R.; Hooker, B.A.; Peterson, B.J. A Simple and Precise Method for Measuring Ammonium in Marine and Freshwater Ecosystems. Can. J. Fish. Aquat. Sci. 1999, 56, 1801–1808. [CrossRef] 40. US EPA. Method 9056A Determination of Inorganic Anions by Ion Chromatography. 2007; No. February; pp. 1–19. 41. Pihlatie, M.K.; Christiansen, J.R.; Aaltonen, H.; Korhonen, J.F.J.; Nordbo, A.; Rasilo, T.; Benanti, G.; Giebels, M.; Helmy, M.; Sheehy, J.; et al. Comparison of Static Chambers to Measure CH4 Emissions from Soils. Agric. For. Meteorol. 2013, 171–172, 124–136. [CrossRef] 42. Norton, R.A.; Harrison, J.A.; Kent Keller, C.; Moffett, K.B. Effects of Storm Size and Frequency on Nitrogen Retention, Denitrification, and N2O Production in Bioretention Swale Mesocosms. Biogeochemistry 2017, 134, 353–370. [CrossRef] Water 2018, 10, 1627 18 of 19

43. Sonnentag, O.; Talbot, J.; Chen, J.M.; Roulet, N.T. Using Direct and Indirect Measurements of Leaf Area Index to Characterize the Shrub Canopy in an Ombrotrophic Peatland. Agric. For. Meteorol. 2007, 144, 200–212. [CrossRef] 44. Trammell, T.L.; Ralston, H.A.; Scroggins, S.A.; Carreiro, M.M. Foliar Production and Decomposition Rates in Urban Forests Invaded by the Exotic Invasive Shrub, Lonicera Maackii. Biol. Invasions 2012, 14, 529–545. [CrossRef] 45. Xu, C.Y.; Griffin, K.L.; Schuster, W.S.F. Leaf Phenology and Seasonal Variation of Photosynthesis of Invasive Berberis Thunbergii (Japanese Barberry) and Two Co-Occurring Native Understory Shrubs in a Northeastern United States Deciduous Forest. Oecologia 2007, 154, 11–21. [CrossRef][PubMed] 46. Madakadze, I.C.; Stewart, K.; Peterson, P.R.; Coulman, B.E.; Samson, R.; Smith, D.L. Light Interception, Use-Efficiency and Energy Yield of Switchgrass (Panicum virgatum L.) Grown in a Short Season Area. Biomass Bioenergy 1998, 15, 475–482. [CrossRef] 47. Zeri, M.; Anderson-Teixeira, K.; Hickman, G.; Masters, M.; DeLucia, E.; Bernacchi, C.J. Carbon Exchange by Establishing Biofuel Crops in Central Illinois. Agric. Ecosyst. Environ. 2011, 144, 319–329. [CrossRef] 48. Kiniry, J.R.; Anderson, L.C.; Johnson, M.V.; Behrman, K.D.; Brakie, M.; Burner, D.; Hawkes, C. Perennial Biomass Grasses and the Mason–Dixon Line: Comparative Productivity across Latitudes in the Southern Great Plains. Bioenergy Res. 2013, 6, 276–291. [CrossRef] 49. Mitchell, R.B.; Moser, L.E.; Moore, K.J.; Redfearn, D.D. Tiller Demographics and Leaf Area Index of Four Perennial Pasture Grasses. Agron. J. 1998, 90, 47–53. [CrossRef] 50. Redfearn, D.D.; Moore, K.J.; Vogel, K.P.; Waller, S.S.; Mitchell, R.B. Canopy Architecture and Morphology of Switchgrass Populations Differing in Forage Yield. Agron. J. 1997, 89, 262–269. [CrossRef] 51. Boyd, N.S.; Gordon, R.; Martin, R.C. Relationship between Leaf Area Index and Ground Cover in Potato under Different Management Conditions. Potato Res. 2002, 45, 117–129. [CrossRef] 52. Gordon, R.; Brown, D.M.; Dixon, M.A. Estimating Potato Leaf Area Index for Specific Cultivars. Potato Res. 1997, 40, 251–266. [CrossRef] 53. Helsel, D.R.; Hirsch, R.M. Statistical Methods in Water Resources; Elsevier Inc.: Amsterdam, The Netherlands, 2002; Volume 4. 54. Chapin, F.S.; Matson, P.A.; Mooney, H.A. Principles of Terrestrial Ecosystem Ecology; Springer: New York, NY, USA, 2002. 55. Bernot, M.J.; Dodds, W.K. Nitrogen Retention, Removal, and Saturation in Lotic Ecosystems. Ecosystems 2005, 8, 442–453. [CrossRef] 56. Liu, J.; Sample, D.J.; Owen, J.S.; Li, J.; Evanylo, G. Assessment of Selected Bioretention Blends for Nutrient Retention Using Mesocosm Experiments. J. Environ. Qual. 2014, 43, 1754. [CrossRef][PubMed] 57. Sanaullah, M.; Chabbi, A.; Leifeld, J.; Bardoux, G.; Billou, D.; Rumpel, C. Decomposition and Stabilization of Root Litter in Top- and Subsoil Horizons: What Is the Difference? Plant Soil 2011, 338, 127–141. [CrossRef] 58. Dungait, J.A.J.; Hopkins, D.W.; Gregory, A.S.; Whitmore, A.P. Soil Organic Matter Turnover Is Governed by Accessibility Not Recalcitrance. Glob. Chang. Biol. 2012, 18, 1781–1796. [CrossRef] 59. Zhang, Z.; Rengel, Z.; Liaghati, T.; Antoniette, T.; Meney, K. Influence of Plant Species and Submerged Zone with Carbon Addition on Nutrient Removal in Stormwater Biofilter. Ecol. Eng. 2011, 37, 1833–1841. [CrossRef] 60. Elliott, S.; Meyer, M.H.; Sands, G.R.; Horgan, B. Water Quality Characteristics of Three Rain Gardens Located Within the Twin Cities Metropolitan Area, Minnesota. Cities Environ. 2011, 4, 1–15. [CrossRef] 61. Vitousek, P.M.; Matson, P. Nitrogen Transformations in a Range of Tropical Forest Soils. Soil Biol. Biochem. 1988, 20, 361–367. [CrossRef] 62. McNulty, S.G.; Aber, J.D.; Mclellan, T.M.; Katt, S.M. Nitrogen Cycling in High Elevation Forests of the Northeastern US in Relation to Nitrogen Deposition. Ambio 1990, 19, 38–40. 63. Auyeung, D.S.N.; Suseela, V.; Dukes, J.S. Warming and Drought Reduce Temperature Sensitivity of Nitrogen Transformations. Glob. Chang. Biol. 2013, 19, 662–676. [CrossRef][PubMed] 64. Nadelhoffer, K.J.; Aber, J.D.; Melillo, J.M. Seasonal Patterns of Ammonium and Nitrate Uptake in Nine Temperate Forest Ecosystems. Plant Soil 1984, 80, 321–335. [CrossRef] 65. Kastovska, E.; Edwards, K.; Picek, T.; Santruckova, H. A Larger Investment into Exudation by Competitive versus Conservative Plants Is Connected to More Coupled Plant-Microbe N Cycling. Biogeochemistry 2015, 122, 47–59. [CrossRef] Water 2018, 10, 1627 19 of 19

66. Hatt, B.E.; Fletcher, T.D.; Deletic, A. Hydrologic and Pollutant Removal Performance of Stormwater Bioﬁltration Systems at the Field Scale. J. Hydrol. 2009, 365, 310–321. [CrossRef]

67. Shrestha, P.; Hurley, S.E.; Carol Adair, E. Soil Media CO2 and N2O Fluxes Dynamics from Sand-Based Roadside Bioretention Systems. Water 2018, 10, 185. [CrossRef] 68. Turk, R.P.; Kraus, H.T.; Hunt, W.F.; Carmen, N.B.; Bilderback, T.E. Nutrient Sequestration by Vegetation in Bioretention Cells Receiving High Nutrient Loads. J. Environ. Eng. 2017, 143, 06016009. [CrossRef] 69. Payne, E.G.I.; Pham, T.; Cook, P.L.M.; Deletic, A.; Hatt, B.E.; Fletcher, T.D. Inside Story of Gas Processes within Stormwater Biofilters: Does Greenhouse Gas Production Tarnish the Benefits of Nitrogen Removal? Environ. Sci. Technol. 2017, 51, 3703–3713. [CrossRef][PubMed] 70. Morse, N.R.; McPhillips, L.E.; Shapleigh, J.P.; Walter, M.T. The Role of Denitrification in Stormwater Detention Basin Treatment of Nitrogen. Environ. Sci. Technol. 2017, 51, 7928–7935. [CrossRef][PubMed] 71. Davis, A.P.; Hunt, W.F.; Traver, R.G.; Clar, M. Bioretention Technology: Overview of Current Practice and Future Needs. J. Environ. Eng. 2009, 135, 109–117. [CrossRef] 72. Hunt, W.F.; Davis, A.P.; Traver, R.G. Meeting Hydrologic and Water Quality Goals through Targeted Bioretention Design. J. Environ. Eng. 2012, 138, 698–707. [CrossRef] 73. Woods End Research Laboratory. Guide to Solvita ®Testing for Compost Maturity Index; Woods End Research Laboratory: Mount Vernon, ME, USA, 2002; Volume 3. 74. Thompson, W.H.; Leege, P.B.; Millner, P.D.; Watson, M.E. Test Methods for the Examination of Composting and Compost (TMECC); Holbrook: New York, NY, USA, 2001. 75. Matteson, T.L.; Sullivan, D.M. Stability Evaluation of Mixed Food Waste Composts. Compost Sci. Util. 2006, 14, 170–177. [CrossRef] 76. Analysis of Bioretention Soil Media for Improved Nitrogen, Phosphorus and Copper Retention; Herrera Environmental Consultants: Seattle, WA, USA, 2015. 77. Herms, D.A. Understanding Tree Responses to Abiotic and Biotic Stress Complexes. Arborist News 1998, 7, 21–26. 78. Levin, L.A.; Mehring, A.S. Optimization of Bioretention Systems through Application of Ecological Theory. Wiley Interdiscip. Rev. Water 2015, 2, 259–270. [CrossRef]

© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).