Salt Marsh Denitrification Provides a Significant Nitrogen Sink in Barnegat Bay, New Jersey Free

Nutrient enrichment is the main cause of eutrophication and hypoxia in coastal waters (Nixon, 1995). Nutrients enter estuaries and coastal waters from river runoff, direct discharge, atmospheric deposition, and ocean exchange. Nitrogen is exported or removed through burial in the sediments, ocean exchange, and denitrification. Nitrogen tends to limit photosynthetic primary production, and therefore is the major contributor to eutrophication in coastal waters. High N loads modify the N cycle and N transformations, resulting in enhanced magnitude of N flux in estuaries (Elsey-Quirk et al., 2013; Gardner et al., 2009; Groffmann et al., 2004). In the United States, inland and coastal wetlands remove approximately 21% of the transported N load, providing an important ecological service (Jordan, Stoffer, and Nestlerode, 2011), with denitrification being an important mechanism of removal.

The amount of N removed or recycled by coastal wetlands varies with different types of processes and seasons. N can be remineralized into ammonium (⁠-N) and recycled to the water column and sediments, removed as N₂ gas via denitrification, immobilized by microbes, or buried as soil organic N (Mitsch and Gosselink, 1993). Sediment profiles of N concentration reflect physical processes such as advection and diffusion of dissolved nutrients, sediment accretion, and biogeochemical processes including biological uptake and transformation (i.e. plant uptake and denitrification). Although N burial is an important N sink in Barnegat Bay (Seitzinger, 1992; Velinsky et al., 2017), denitrification permanently removes N from the water column (Seitzinger, 1988, 1992; Seitzinger et al., 2006), the rates of which may vary on the basis of temperature and nutrient inputs, as well as between habitat types.

Water quality in Barnegat Bay is affected by persistent high nutrient loads from agricultural runoff, waste discharge, and storm-water discharge (BBNEP, 2005; Kennish et al., 2007). N loading from the watershed occurs at an estimated rate of 7 × 10⁵ kg/y, which amounts to 3.9 kg/ha per year (Kennish et al., 2007). Approximately 15% of the N load to surface waters is derived from the application of fertilizer in the watershed (Ayars and Gao, 2007; N. Borgatti, unpublished data; Castro and Driscoll, 2002). Higher N loading occurs in the northern part of Barnegat Bay (north of Barnegat Inlet), with a large seasonal change in N concentrations (high in wet seasons, i.e. April–October; Baker et al., 2014). As a consequence, the Bay experiences recurring phytoplankton blooms with chlorophyll a concentrations up to 40 to 50 μg/L, harmful algal blooms, and brown tides (Aureococcus anaphagefferans), which have occurred sporadically since 1995 (Gastrich et al., 2004; Olsen and Mahoney, 2001). Poor flushing and relative long residence times contribute to the eutrophic status of the Bay. However, Barnegat Bay has approximately a 1.1 × 10⁸ m² area of tidal marsh (Lathrop and Haag, 2007), which can significantly reduce inorganic N inputs. Many salt marshes along the Bay are subject to open marsh water management (OMWM) for mosquito control. OMWM consists of excavating shallow permanent pools of water in the marsh interior. The excavated sediment is then sprayed around the ponds, effectively becoming a thin-layer deposit, creating elevations approximately 5 cm higher than non-OMWM marshes (Elsey-Quirk and Adamowicz, 2016). Because OMWM ponds are too low in elevation to support emergent vegetation, nutrient cycling may differ from that of vegetated marsh sediments. The lack of vegetation, higher water level, and lower elevation of OMWM ponds compared with salt marshes likely affect sediment redox conditions, affecting sediment N dynamics. Approximately 36 km² of salt marsh in Barnegat Bay has been altered by OMWM activities, and therefore differences in microbial nutrient processing associated with OMWM ponds may have a landscape-level effect on the microbial attenuation of nutrient loads by salt marshes.

The goals of this study were to (1) determine differences between vegetated marsh areas and nonvegetated OMWM ponds in denitrification and N fluxes, (2) examine spatial and seasonal variation in denitrification rate in Barnegat Bay salt marsh ecosystems, and (3) quantify the amount of N via denitrification provided by a salt marsh ecosystem relative to N inputs in the Bay. It was also predicted that denitrification rates would be higher in vegetated marsh habitats compared with ponds because of aerobic/anaerobic interfaces and labile C sources present in marshes. It was hypothesized that there would be a strong seasonal increase in denitrification in the summer (Kaplan, Teal, and Valiela, 1977) and spatial differences in denitrification related to areas of higher and lower watershed N loading.

Barnegat Bay–Little Egg Harbor estuary, located along the central New Jersey coastline in the Atlantic Coastal Plain province, is a back-barrier lagoon-type estuary that extends from Point Pleasant south to Little Egg Inlet (Figure 1). Habitats in this system include barrier beach and dune, submerged aquatic vegetation beds, intertidal sand and mudflats, salt marsh islands, fringing tidal salt marshes, freshwater tidal marsh, and palustrine swamps. The Bay is approximately 70 km in length, 2 to 6 km wide, and up to 7 m deep. Tidal waters cover approximately 280 km², with a ratio of watershed area to water area of 6:1. The Bay watershed covers an area of approximately 1700 km² and has been extensively developed over the past 70 years (Lathrop and Haag, 2007). From 1986 to 2012 the urban land-cover area in the Barnegat Bay watershed increased from 23% to 32%, whereas forested land cover has decreased (BBP, 2016). N inputs ranged from 4.6 to 8.6 ×10⁵ kg y⁻¹ (1989–2011; mean value calculated as 6.6 ± 1.1; Baker et al., 2014). Wetland area in the Bay covers 110 km², with an estimated 36 km² altered by OMWM ponds (www.crssa.rutgers.edu/ projects/lc). Three marshes representing the northern (Reedy Creek; RC), midbarrier island (Island Beach State Park; IBSP), and southern (Channel Creek; CC) parts of Barnegat Bay were the focus of the study (Figure 1).

Six sediment cores (∼6.5-cm diam; ∼20-cm sediment depth) and overlying water from each of the three marshes were collected during three seasons: spring (May), summer (July), and fall (October) of 2012. In July, the season with the highest predicted denitrification rates, three habitat types were sampled: (1) vegetated marsh; (2) vegetated marsh that had experienced thin-layer deposition from the spraying of material excavated for OMWM ponds; and (3) the interior of OMWM ponds (n = 6).

Creek water (>20 L) was collected near each marsh during each sampling for core incubations, nutrients, and chlorophyll a analysis. Temperature, dissolved oxygen, pH, conductivity, and salinity at each site were collected using handheld YSI model 556.

Overlying and site water from all sampling dates were retained for filtration (GF/F 0.7 μm) and nutrient analysis. The filtrate was placed into small precleaned high-density polyethylene bottles. Water samples for nutrient and related parameters were filtered and immediately frozen.

Within 12 hours of sampling, water and sediment cores were transported on ice with site water overlying the headspace to the University of North Carolina Institute of Marine Sciences in Morehead City, NC (IMS). At IMS, cores were submerged in an aerated water bath in an environmental chamber (Bally Inc.) overnight at in situ temperatures. The fluxes of nutrients and dissolved gases were measured with continuous-flow incubations of intact cores (Lavrentyev, Gardner, and Yang, 2000; McCarthy et al., 2007). The following morning, each core was capped with an air-tight Plexiglas top equipped with an inflow and outflow sampling port. Aerated and unfiltered water was passed over cores at a flow rate of 1 mL min⁻¹, which created a well-mixed water column within the chamber (Lavrentyev, Gardner, and Yang, 2000). Incubations were conducted in the dark at in situ temperature. Some of the cores from vegetated marsh and vegetated marsh on thin sediment-layer sites had bubbles and thus were removed while calculating denitrification rates.

Cores were acclimated in the continuous-flow system for a period of no less than 18 hours before sampling to allow the system to reach equilibrium (Eyre and Ferguson, 2002; Eyre et al., 2002). Water samples (5 mL) were collected from the outflow of each core at 18-, 24-, 36- and 48-hour increments to ensure that steady-state conditions were present for analysis of dissolved gases. Inflow concentrations were measured from a bypass line that flowed directly into the sample vials. Gas samples were analyzed for N₂, O₂, and Ar using membrane inlet mass spectrometry (Kana et al., 1994, 1998).

In addition to gas samples, at the 24-hour sampling, 50-mL water samples were collected for nutrient analysis from the inflow and outflow of each core. Water was filtered through Whatman GF/F filters (25-mm diam, 0.7-μm nominal pore size) and the filtrate was analyzed for (⁠+ ⁠) and using Alpkem segmented flow analyzer following standard methods (Paudel et al., 2017).

Flux calculations were based on the assumption of steady-state gradients that match in situ gradients and a homogenous water column. Benthic fluxes were calculated using the equation (C_out − C_in) × F/A, where C represents the concentration of analyte, C_in and C_out are the inflow and outflow concentrations (μM), respectively, F is the peristaltic pump flow rate (l h⁻¹), and A is the surface area of the core (m²) (Miller-Way and Twilley, 1996). Net N₂ fluxes were calculated from the N₂/Ar ratio, where the positive flux of N₂ out of the sediment was denitrification minus N fixation (An, Gardner, and Kana, 2001; Kana et al., 1994). A positive net N₂ flux indicates that sediments were net denitrifying and were considered rates of denitrification. Oxygen fluxes and sediment oxygen demand (SOD) were calculated from O₂/Ar as the flux of O₂ into the sediment (Kana et al., 1994; Smith et al., 2006). For and ⁠, a positive flux indicated production from the sediment to the water column and a negative flux indicated uptake from the overlying water. Individual measurements from each core over the incubation were averaged to yield core-specific values. Denitrification data were extrapolated on the basis of a 12-hour day to reflect our assumption of very low rates during the day due to both competition with benthic microalgae for N and increased oxygen concentrations (Hochard et al., 2010; Piehler and Smyth, 2011; Tobias, 2007).

Water samples, both from the adjacent creek and core incubations, were analyzed for ⁠, ⁠, and soluble reactive phosphorus (SRP). Nitrate–nitrite and -N concentrations were determined using an Alpkem 300 segmented flow autoanalyzer with a detection limit of 0.006 and 0.005 mg/L for and NH₄, respectively, and 0.002 mg P/L for SRP.

After the water/gas exchange experiments, individual cores were sectioned into top (0–2 cm), middle (5–7 cm), and bottom (9–11 cm) for organic C, total N, and total P composition. Sediment total organic carbon (SOC) and total nitrogen (TN) were measured using a CE Instruments Flash EA 1112 series following the guidelines in U.S. Environmental Protection Agency 440.0, manufacturer instructions, and Academy of Natural Sciences of Philadelphia personal computer standard operating procedures. Samples were ground to a powder, pretreated with fuming HCl to remove inorganic C, redried, and ground. Samples were weighed into tin boats using a microbalance (in duplicate) and analyzed using the FLASH 1112 elemental analyzer.

The effects of marsh and season and their interaction on denitrification and dissolved inorganic N (DIN) flux rates were tested using an analysis of variance (ANOVA). A one-way ANOVA tested the influence of habitat type on denitrification and N fluxes. Differences in water-column chemistry (e.g., dissolved oxygen, salinity, nutrient concentrations) among marshes and differences in soil properties among marshes and depths were tested using ANOVA. A post hoc Tukey test was used for multiple comparisons if significant effects were found. Correlation analyses were conducted to examine relationships among denitrification, N fluxes, and water-column nutrients and soil C and nutrients. All data analyses were conducted using JMP version 12.1, SAS Institute Inc., Cary, NC, 1989–2017.

To calculate total denitrification N removal rate, two main factors were applied to the denitrification rates. First, the incubations for this study were done under dark conditions to directly measure N₂ production from denitrification. Importantly, however, algal–plant uptake of nitrate and oxygen production during the day in the surface sediments would limit denitrification. Therefore to scale our numbers up for the Bay, denitrification rates were divided by two assuming dark conditions for half the time. Second, since the marsh sites (especially nonpond sites) are only flooded for specific time periods, in each tide, periods of inundations (Ensign, Piehler, and Doyle, 2008; Smyth et al., 2013) were given specific importance when calculating denitrification rates in the marsh. A water-level data logger at IBSP and CC estimated 3–4 hours of tidal inundation for Barnegat Bay (U.S. Geological Survey, unpublished data). In addition, once water inundates the marsh there may be a lag time before onset of denitrification. For this study, it was assumed that there was no lag time and the average inundation time between the mid- and lower Bay (∼12 h) was used. We assumed no denitrification in winter, on the basis of low rates of denitrification in October.

Water chemistry, sediment C and N, and denitrification rates in the three different habitats were studied and the results are summarized below.

Water-column temperature across marsh sites averaged 19°C ± 1°C in May, 25°C ± 1°C in July, and 15°C ± 2°C in October. Salinity was lowest in RC (19.2 ± 0.1 practical salinity units [psu]) and highest at the midbay site (31.4 ± 0.2 psu) (F_2,42 = 437.51, p < 0.0001; Table 1). Dissolved oxygen (DO) concentrations differed among marshes depending on season (season × marsh: F_4,36 = 82.60, p < 0.0001; Table 1). DO was generally lower in July than in October at all marshes. Spatially, DO was lower in RC than down-bay marshes in both May and July. pH ranged from 6.8 to 8.1, with slightly lower values at the upstream location (i.e. RC site).

The seasonality of concentration in the water column differed among marshes (season × marsh interaction: F_4,35 = 27.37, p < 0.0001). Generally, nutrient concentrations in the northern part of the Bay were the highest in May, whereas the southern part of the Bay had the highest nutrient concentrations in October (Figure 2). Nitrate concentration at all sites was very low in July, with most of the water samples from RC and IBSP below the detection limit. -N concentration also varied seasonally depending on location (season × marsh interaction: F_4,36 = 77.08, p < 0.0001). -N concentration was the highest in CC in October, followed by RC in May (Figure 2). SRP concentration was the highest in the southern part of the Bay at CC in October (season × marsh interaction: F_4,36 = 11.67, p < 0.0001; Figure 2).

Organic matter ranged from 23% to 64% and was the highest at IBSP and the lowest at CC across depth (p < 0.0001; Table 2). Organic carbon was higher at RC and IBSP than at CC (p < 0.0001). TN was highest at RC, followed by IBSP and then by CC. C-to-N molar ratio was relatively higher in CC than in the other two sites. There was no statistical difference between vegetated and OMWM ponds in SOM, SOC, or TN (data not shown).

Denitrification rates (net positive N₂ flux) were similar among marsh locations, but differed among seasons. Denitrification rates were higher in July, averaging 121 ± 20 μmol m⁻² h⁻¹, than in October (49 ± 19 μmol m⁻² h⁻¹; F_2,49 = 4.02, p = 0.0242; Table 3). Denitrification rate in May (83 ± 14 μmol N m⁻² h⁻¹) was not different from those in the summer and fall. SOD ranged from 240 to 3400 μmol O₂ m⁻² h⁻¹ across sites, with the highest demand in the July when temperatures were highest (F_2,45 = 7.12, p = 0.0019). The relationship between SOD and N₂ production varied seasonally (Figure 3). Rates of SOD were significantly related to N₂ production in May and July and less so in October (Figure 3); with the strongest relationship in July (r² = 0.90, p < 0.0001) when temperatures were the highest. Rates of denitrification were not correlated with water-column nutrient concentrations.

fluxes ranged from −34 to +28 μmol N m⁻² h⁻¹ across all sites and seasons. Nitrate fluxes varied among marshes depending on season (season × marsh interaction: F_4,42 = 5.7, p < 0.0001). In May, RC marsh was taking up N, whereas IBSP and CC were a source of oxidized N (Figure 4). In July, N effluxes in IBSP were greater than that in CC, which averaged close to zero. N fluxes in October were variable, particularly in CC, and did not differ among marshes. Nitrate fluxes were directed out of the sediments and at a higher magnitude when and where water-column concentration of nitrate was low, and conversely, were directed into the sediments when water-column nitrate concentration was high, which occurred during different months at different sites (Figure 5).

-N fluxes were similar across seasons, but differed significantly among marshes (F_2,45 = 5.72, p = 0.0064). -N flux averaged 74 ± 12 μmol m⁻² h⁻¹ across seasons in RC, which was over two times greater than in IBSP and CC. Nitrate fluxes were positivity related to -N fluxes for IBSP (r² = 0.39, p = 0.0129). -N fluxes were negatively related to water-column nutrients, which, overall, explained 90% of the variation in -N fluxes (Figure 6).

Denitrification in vegetated areas had similar variability compared with the OMWM ponds and rates were within 2 standard deviations of each other (Table 4). Rates in vegetated marsh on thin sediment deposition areas were approximately twice as much and with greater variability. Similarly, SOD followed the same pattern except the OMWM ponds were substantially less variable. Ammonium efflux from the pond habitat was 10 times greater than fluxes from the vegetated sites. The vegetated areas on thin sediment-deposited sites exhibited net uptake of ammonium.

Denitrification rates follow the general water-column temperature trend; i.e. highest during July (average temperature 26°C) and lowest during October (average temperature 15.5°C). Midsummer high-temperature preference of denitrifying bacteria is associated with a greater abundance of denitrifying bacteria as well as a specific increase in N₂ production at temperatures above 5°C potentially starting around 10°C (Kaplan, Teal, and Valiela, 1977; Nowicki, 1994). Greater nitrate production by nitrification with higher temperatures is likely enhancing N₂ production rate during the summer, whereas the relative importance of nitrate supplied by the water column declines (Poulin, Pelletier, and Saint-Louis, 2007). Denitrification rates ranged between 12 and 290 μmol m⁻² h⁻¹ in marshes of the NE United States (Valiela et al., 2000), whereas gross denitrification rates in vegetated marshes (variable vegetation) ranged between 36 and 4129 μmol m⁻² d⁻¹, with a median value of 1000 μmol m⁻² d⁻¹ (n = 16) (Hopkinson and Giblin, 2008). Denitrification rates in the present study resembled rates reported by Valiela et al. (2000). In general, higher denitrification rates are associated with higher oxygen demand, which is associated with greater nitrate production, and thus increased coupled nitrification–denitrification (Piehler and Smyth 2011; Smyth et al 2013; Tobias and Neubauer, 2009).

Nitrate fluxes from marsh sediments were positive in May and July, indicating that production exceeded denitrification demand or autotrophic uptake. Nitrate fluxes were not related to SOD, whereas -N fluxes were positively related to SOD in May and July, only suggesting limitation in nitrification during oxygen consumption and indicating possible recycling release of N in the sediments. Rates of SOD indicate nitrification: the oxidation of -N to nitrite and eventually nitrate (Ward, 1996). Lower oxygen availability would limit nitrification and the coupling to denitrification. Measured -N fluxes from pore-water profiles and flux chambers in Great Marsh, DE over a year ranged from −6 to 6 μmol N m⁻² h⁻¹and 5.1 to 206 μmol N m⁻² h⁻¹ respectively (Scudlark and Church, 1989). Similarly, measured -N fluxes ranged between 3 and 435 μmol N m⁻² h⁻¹ in the tidal marsh of Virginia, with highest flux rates identified in midsummer (August) (Chambers, Harvey, and Odum, 1992). These data indicate that diagenesis of organic matter is producing dissolved -N in excess of that used in coupled nitrification–denitrification, with remaining -N escaping the marsh surface during tidal inundation. The low water-column concentration of dissolved nitrate suggests that external sources of nitrate transport to the Barnegat Bay system limited denitrification rates and coupled reactions.

Nitrate fluxes were strongly seasonal, but varied among marshes associated with the local concentration of nitrate in the water column. Although studies have shown a positive relationship with seasonal temperature and nitrate and -N fluxes (Poulin, Pelletier, and Saint-Louis, 2007), we illustrate that seasonal variation can also be influenced by the timing of peak nutrient concentrations, which can vary spatially within an estuary. In the northern part of Barnegat Bay, near the RC marsh, concentration of nitrate was greatest in May, and nitrate uptake by RC marsh was also highest. Nitrate concentrations in the water column at RC were significantly lower in July and October when nitrate was generated by marsh sediments. In the lower part of the estuary, nitrate concentrations were greatest in October when nitrate uptake was greatest at IBSP, and most variable at CC. Similarly, -N consumption rates were positively associated with nutrient availability in the water. Although nitrate consumption can be positively related to -N generation rates (Poulin, Pelletier, and Saint-Louis, 2007), we only found this trend at one marsh, IBSP. The influence of water-column nutrient availability on nutrient fluxes may be due to the low overall nutrient concentrations, with seasonal maximums of approximately 10–30 μg -+ -N/L and 50–100 μg -N/L, and often below detection. The effect of fluxes may be caused by and change in the diffusion gradient rather than microbial processes, as neither SOD nor denitrification were influenced by temporal and spatial changes in the nutrient concentration in the water. At higher nutrient concentrations, such as those supplied in fertilization studies, denitrification rates can be enhanced. However, habitat and elevation within marshes may play a role in fertilization effects on denitrification. Large increases in denitrification rates have been documented in low Spartina alterniflora marshes (Hamersley and Howes, 2005), whereas, despite increases in pore-water nutrient concentration, bacterial abundance, and production, denitrification rate was not affected by fertilization in a high Spartina patens marsh (Caffrey et al. 2007). There were no differences among vegetated S. alterniflora areas and nearby OMWM ponds, whereas vegetated marsh on thin-layer sediment deposition had higher denitrification rates. Despite the lack of difference in the denitrification rates between vegetated marsh and OMWM ponds, the ponds had higher ammonium efflux, indicating that the pond was a net source of remineralized N.

Denitrification in Barnegat Bay salt marshes ranged from 1.1 to 2.6 × 10⁵ kg N y⁻¹ depending on season. Using recent N load estimates (Baker et al., 2014), TN removal via denitrification averaged 27.9% ± 6.9% of the TN load to Barnegat Bay (Table 5). Potential sources of error in scaling up the calculations to the Bay include potential lag time in the onset of this process and the actual amount of wetland area that is inundated during a tidal cycle, a month and a year. However, these data and calculations show that marshes can sequester or biologically remove up to a quarter of the N loads to the Bay from denitrification. The estimates provided above show that the marshes as well as subtidal areas (Seitzinger, 1992) have a potential to trap N before being exported to the Bay and highlight the importance of ecosystem services that marshes provide (i.e., water filtration) and the potential cost of water treatment if marsh areas are reduced by either land development or sea-level rise.

The present study identified Barnegat Bay marshes as an important ecosystem in removing nitrogen via denitrification. The study identified almost similar rates of denitrification between vegetated marsh and OMWM ponds, but had higher denitrification rates from vegetated marsh on a thin layer of sediment. On average, the study estimated that approximately 28% of the TN load to Barnegat Bay was annually removed by salt marshes.

We thank Paul Kiry, Mike Schafer, Will Whalon, Roger Thomas, Michelle Gannon, and Paula Zelanko for field and laboratory assistance as well data interpretation. Tom Belton (New Jersey Department of Environmental Protection [NJ DEP]) provided background, field, and editorial support throughout this project. We also thank Scot Haag for his help in the creating maps for the project. Funds for this project were provided by NJ DEP with additional support from Patrick Center and Academy endowments.