Introduction

Coastal systems in the temperate zone around the globe are being polluted by nitrogen (N) entering from their watersheds as a result of anthropogenic activity (Howarth et al. 2002). Anthropogenic N enters coastal ecosystems primarily as dissolved inorganic N (DIN) and stimulates primary productivity, which can lead to phytoplankton blooms and a cascade of negative ecological effects. In response to nationwide water quality concerns, the federal Clean Water Act (CWA) (Sect. 303 (d)) established regulatory measures to restore and maintain the overall health of the nation’s waters. A total maximum daily load (TMDL) was set for each pollutant within a system not meeting CWA standards. Municipalities are currently grappling with the need to reduce N loading in estuaries to meet these regulatory TMDLs. Traditional wastewater collection and treatment is expensive; therefore, researchers have been intensively working to develop non-infrastructure approaches for removing N before it reaches estuarine waters or to enhance the removal of N within the estuary itself. Similarly, interest in nutrient trading programs has increased in recent years. As a result, environmental managers must quantify N removal for a variety of alternate approaches if they are to be implemented as a N management tools (Ray et al. 2021; Rose et al. 2021).

Oyster aquaculture is being investigated as a sink for N that has reached the estuary from the surrounding airshed and watershed (Ayvazian et al. 2021). Bivalves affect biogeochemical cycling by filtering out phytoplankton and other particulate matter, thus increasing water clarity. Particulate matter captured on the gills can be rejected as pseudofeces, or ingested, digested, and released as feces (Newell and Jordan 1983), both of which increase deposition to bottom sediments. The majority of the digested organic N is assimilated into oyster tissue and shell for maintenance and growth, and some amount is excreted, primarily as ammonium (NH4+; Bayne and Hawkins 1992). Assimilation and subsequent oyster harvest is a potential permanent N removal pathway as the removed oysters store N and make it unavailable for uptake by phytoplankton and microbes.

Pseudofeces and feces, collectively called biodeposits, are denser than ambient suspended particulates. Once released to the water column, biodeposits quickly settle to the sediment surface, where the organic N in the biodeposits may be permanently buried in the anoxic sediment layer or regenerated as NH4+. Ammonium in the oxic sediments can diffuse to the water column or be available for sediment nitrifying bacteria. Nitrifying bacteria derive energy through the metabolic process of nitrification. Nitrification is a two-step process that requires the presence of oxygen to transform NH4+ into nitrite and then nitrate (NO3−); it occurs in the upper oxic sediment layer (Fenchel et al. 1998). The NO3− from nitrification can either be released into the overlying water column or can diffuse to the anoxic sediment layer where it is available to denitrifying bacteria.

Denitrification is limited in sediments with free oxygen, but occurs in anoxic zones if NO3− is present. Denitrification generates energy as denitrifying microbes consume organic matter and convert NO3− to dinitrogen gas, which escapes the aquatic system into the atmosphere. Nitrification provides a NO3− source for denitrification. Therefore, nitrification and denitrification typically occur as coupled processes (termed coupled nitrification–denitrification; CND) in shallow-water estuaries (Hamersley and Howes 2005). Many factors contribute to the potential for CND, including the level of overlap between oxic and anoxic zones in the sediment (Fenchel et al. 1998). In contrast, direct denitrification typically occurs where NO3− is available in waters overlying redox stratified sediments, and NO3− diffuses into the sediments to the anoxic layer to support denitrification (Piehler and Smyth 2011). Direct denitrification depends mainly on the availability of water column NO3−.

Two other N transformations that have received less attention are dissimilatory reduction of NO3− to ammonium (DNRA) and anaerobic ammonium oxidation (anammox). DNRA is the direct conversion of NO3− to ammonium, and may compete with denitrification for NO3−; it is a microbial process that conserves N within a system and therefore is not part of N removal. DNRA has been reported in shallow-water estuaries in Texas, North Carolina, and Virginia (An and Gardner 2002; Smyth et al. 2013; Lunstrum et al. 2018), and can be associated with oyster reefs and off-bottom oyster culture. Anammox is a microbial-mediated process in which ammonium and nitrite are converted directly to dinitrogen gas in the absence of O2. Anammox is a potential N loss pathway similar to denitrification, but it is not believed to play a large role in shallow-water polyhaline estuaries (Rich et al. 2008); studies conducted in MA and NY suggested that anammox contributed < 10% of N2 production (Engström et al. 2005; Koop-Jakobsen and Giblin 2009). To the extent that anammox occurs at oyster sites, the N2 generated would be captured with the N2 from denitrification in assays that quantify N2 production as the measure of microbial N removal.

Oyster aquaculture may increase net denitrification by increasing NH4+ production when bottom water dissolved oxygen (DO) levels are sufficient to maintain an oxygenated surficial sediment layer. Denitrification and denitrification efficiency (generally described as N2 production relative to total N remineralization) appears to be site-specific, and published data are quite variable (Kellogg et al. 2014). The observed variability in denitrification rates and efficiencies is likely attributable to differences in site-specific environmental conditions, aquaculture management practices, and techniques used to measure denitrification. A more quantitative understanding requires collecting data across different settings to identify patterns and determine what features are related to oyster aquaculture and enhanced denitrification. At present, given the significant N management effort, data specific to embayments characteristic of southeastern Massachusetts (i.e., eutrophic shallow-water depositional systems) is needed.

Several previous studies have measured nutrient and N2 flux rates associated with oyster aquaculture, but the systems studied to date have significant differences compared to southeastern MA embayments including depth, hydrodynamics, and water column particulate composition, as well as aquaculture type (e.g., species and gear; Kellogg et al. 2014; Humphries et al. 2016; Lunstrum et al. 2018). In the present study, we quantified N cycling relative to floating oyster aquaculture, specifically, the potential for enhanced denitrification, and the relationship between N deposition and environmental conditions affecting denitrification. We hypothesized that (1) biodeposition resulting from floating oyster aquaculture would enhance sediment denitrification rates, (2) labile biodeposits would continue to enhance denitrification rates after oyster removal from the system and into the following spring, and (3) the degree of enhancement would be a function of both bottom water DO and the amount of biodeposits settling to the sediment surface.

Methods

Study Site and Oyster Aquaculture Characteristics

Lonnie’s Pond (Orleans, MA; 41.7693° N, − 69.9768° W), a eutrophic shallow-water depositional system, was used to examine the effect of oyster aquaculture on N cycling previously described in Labrie et al. (2022). Lonnie’s Pond (0.06 km2) is a drown kettle pond with an average depth of 3.0 m; it receives surface water flow via a freshwater stream along the southwest shore (stream enters 50 m from nearest edge of the floating bags) and tidal input through an inlet on the southeast section of the pond (Fig. 1a). It is a tributary to the Pleasant Bay Estuary (lagoon) located on the eastern-most shore of Cape Cod on the Atlantic Ocean. Lonnie’s Pond and other tributaries to the Pleasant Bay Estuary are impaired by excessive N inputs from their watershed. Findings of the MA Estuaries Project (Howes et al. 2006) indicated that the N load to Lonnie’s Pond would need to be lowered by 300 kg N year−1 (33% of the N loading to its watershed) to mitigate impairment, with other tributaries to the Pleasant Bay Estuary also requiring similar reductions. In 2016, floating oyster aquaculture was established in Lonnie’s Pond to test its use for N removal through N assimilation with oyster harvest and enhanced sediment denitrification. Two oyster cohorts, first (0 +) and second year (1 +), were deployed in spring or early summer 2016–2018 and harvested or overwintered in late fall of the same year. Aquaculture deployments included 200,000 oysters (800 bags), 600,000 oysters (1012 bags), and 2.1 million oysters (1515 bags) in 2016, 2017, and 2018, respectively. First-year oyster seed (2–3 mm) was deployed in 0.75-mm mesh modified spat bags with attached floats, and then size-sorted and redeployed in 6-mm mesh bags once shell lengths reached approximately 12 mm. The 6-mm diamond mesh bags consisted of high-density polyethylene (Intermas brand; 1 m × 0.5 m) and were made positively buoyant with two attached foam floats. Floating bags were routinely cleaned, and oysters were size graded to maximize food availability to promote biodeposition and oyster growth. Bags were flipped in place periodically (weekly to bi-weekly) to reduce biofouling and to redistribute oysters within each bag. Aquaculture plots and the degree of overlap across the 3 years of the study are shown in Fig. 1b.

Fig. 1
figure 1

Satellite image of Lonnie’s Pond, Orleans (a). Composite diagram of 2016–2018 deployment areas showing relative size and position of aquaculture plots. The 2016 deployment area held 8 longlines, installed ~ 3 m apart with 2 rows of 50 bags on each longline. 2017 and 2018 deployment areas were divided into square plots (378 m2) each holding 30 rows of 17 bags; this higher density layout resulted in a 50% increase in the number of bags per unit area compared to 2016. The 2018 SW and SE plots were the same as the 2017 W and E plots, respectively. The 2018 NW plot was 2.8 m water depth: 1 m deeper than the other areas (b)

Full size image

Whole bag oyster weights were determined for each bag upon deployment and removal from the pond at the end of the growing season. Total biomass increase over the deployment period was determined by weight increase. In addition, 25–30 oysters from each aquaculture plot and cohort were collected upon deployment in early spring and removal in late fall for determination of whole oyster wet weight, dry tissue weight (DTW), dry shell weight, and N content. Oyster tissue and shell were dried for a minimum of 72 h at 60 °C. Nitrogen content of tissue and shell and the mass of N removed at the end of the growing season were determined following procedures described by Kellogg et al. (2013) and Reitsma et al. (2014) using elemental analysis (measured at Boston University’s Stable Isotope Laboratory) following standard procedures. Biomass increase was multiplied by percent N content of whole oysters to determine net N assimilation at time of removal from Lonnie’s Pond.

Field Sampling and Water Chemistry

Time series measurements of DO, temperature, and salinity were made using moored YSI-6600 sondes during the 2016–2018 growing seasons. Sensors were 30 cm above the sediment surface within the oyster deployment area and approximately 4 m to its east and west (total depth ~ 1.8 m) to measure oxygen conditions near the sediment surface. A fourth mooring was placed approximately 4 m north of the deployment area in surface waters ~ 3 m above the sediments (total depth ~ 3.5 m) to monitor surface/mixed-layer water quality outside the influence of oyster biodeposit degradation in the sediments. Measurements were recorded every 15 min and sensors calibrated bi-weekly with triplicate samples for DO by Winkler (Oudot et al. 1988). Water quality sampling was conducted prior to aquaculture deployment and throughout the oyster growing season, and also included collecting grab samples and flow measurements of the stream entering the pond. Bi-weekly water quality sampling began in June 2016, corresponding with oyster deployment, and in April in 2017 and 2018.

Water depth, Secchi depth, temperature, and DO were recorded during water column sampling. Whole water was held in acid-leached 1-L brown HDPE bottles for laboratory collection of particulate carbon (C) and N, and 60 mL subsamples were field filtered (0.2 µm) into acid-leached polyethylene bottles for analysis of dissolved inorganic and organic N and P constituents. Water samples were stored dark on ice for transport to the Coastal Systems Analytical Facility at UMassD/SMAST. Filtered samples were assayed for ammonium (NH4+) following the indophenol/hypochlorite method (Scheiner 1976) and ortho-phosphate (Murphy and Reilly 1958) within 24 h of collection. The remaining sample was frozen (− 20 °C) until assayed for combined nitrate + nitrite (NOx−  = NO2−  + NO3−) by cadmium reduction on a QuikChem 8000 Lachat auto analyzer (Wood et al. 1967) and dissolved organic N by persulfate digestion (D’Elia et al. 1977). Particulate organic C and PON were determined by filtering whole water through pre-combusted 25-mm Whatman glass fiber filters within 24 h of collection, drying (65 °C), and assayed using a Perkin Elmer Series II CHN elemental analyzer (Kristen 1983).

Sediment Core Collection and Incubation

The spatial extent of biodeposition to sediments (biodeposit area) from overlying floating oyster bags was determined using in situ biodeposition rates (standardized to a 1-g dry tissue weight oyster) measured over a range of water quality conditions, and a site-specific particle advection model (Labrie et al. 2022). To measure biodeposition, biodeposit traps (based on a design developed by Higgins et al. 2013) were secured directly below individual unmodified floating bags in Lonnie’s Pond for 24–72 h to capture biodeposits settling from the bags. A control biodeposit trap was deployed during each event to determine background particle settling rates, which was then subtracted from measured total deposition rates to yield biodeposition rates. Biodeposited C and N masses were based on field biodeposition rates and C and N content of collected biodeposits as determined by elemental analysis. Per the particle advection model described in Labrie et al. (2022), for 2017 and 2018, oyster biodeposits were advected by horizontal currents upon release from floating oyster bags. Measured horizontal velocities (Acoustic Doppler Current Profiler (ADCP); Aquadopp Profiler HR 2 MHz ADCP, Nortek USA Inc., Boston, MA, USA), biodeposit settling rates, and water column depth were used to predict horizontal displacement of biodeposits settling to the sediment surface relative to the biodeposit source (i.e., oysters within the floating bags). The upward-facing ADCP was deployed in Lonnie’s Pond in October 2017 and July and October 2018 over multiple tidal cycles to capture seasonal variability. During each event, the ADCP was mounted on a 12-mm thick rectangular base and was deployed directly below or next to (1.5 m north) the floating bags. Individual horizontal displacements were summed across a 2-D grid to estimate biodeposition intensity relative to the location of the floating bags and sediment core collections. The summed displacements were used to determine the degree to which sediments received labile nutrient-rich biodeposits. In particular, N deposition (biodeposition N + background N) relevant to each core location was predicted using the biodeposit model (N biodeposition standardized to a 1-g dry tissue weight oyster) plus inputs due to background settling of N (based on measured particulate organic N, total depth, and a turnover of 15.5% per day; Lohrenz et al. 1987), all referenced to a 2-D grid of oyster dry tissue weights (representing the floating oyster bags). In the present study, for 2016 only, biodeposit area was determined using the 2017 predicted ratio between modeled biodeposit area and deployment area. A maximum distance of biodeposit displacement was then determined based on barotropic mean velocity, maximum water column depth, and biodeposit settling velocity.

Per Labrie et al. (2022), the particle advection model was verified by comparing predicted C deposition with measured sediment oxygen demand (SOD) rates assuming that the molar ratio of CO2:O2 during respiration was 1:1. Findings indicated that predicted C deposition was positively correlated with measured SOD. The implication was that the spatial extent of biodeposition to the sediments (i.e., organic C deposition) was largely accounted for by on-site C mineralization, and therefore that particle transport beyond the biodeposit area was minimal. In the present study, predicted N deposition was compared with measured N remineralization.

Determination of the biodeposit area was used to select coring locations affected by biodeposit loading, vs. those receiving only background (non-oyster) deposition. Sediment cores collected outside of the biodeposit area represent background benthic flux rates and are referred to as control sediments or control cores, while sediments collected within the biodeposit area are referred to as treated sediments or treated cores.

Twelve to sixteen cores were collected on each sediment collection date depending on the area and heterogeneity of sediments being examined. In general, core locations were distributed to capture potential variation in nutrient flux rates throughout the biodeposit area, as well as the variability of control sediments. Sampling within the biodeposit area aimed to capture differences attributable to different oyster sizes, ages, and stocking densities, as well as the amount of time oysters were deployed. In 2016 only, sediment cores were only collected directly below the floating bags or beyond the maximum biodeposit deployment distance. Sediment collections began after at least 30 days of oyster deployment and continued to capture seasonal variations focusing on periods of high oyster and sediment metabolic activity in late summer, early fall, and early spring (2017 and 2019). July, August, and September incubations captured warmer summer water conditions and peak temperatures and biodeposition rates, whereas October incubations captured conditions of cooler more oxygenated bottom water and maximum oyster biomass. Finally, early spring incubations were conducted prior to oyster deployment to capture any residual effects of the prior year’s accumulated biodeposits that overwintered as temperatures began to warm and sediment metabolism increased.

Intact sediment cores were collected by SCUBA diver in core tubes (15 cm i.d., 30 cm height, 15 cm of sediment) and were immediately transported at in situ temperature by boat to a nearby field laboratory for incubation in pre-equilibrated insulated water baths. Headspace water was replaced with filtered (0.22 µm) water (e.g., “fill water”) from the core site, aerated, and mixed at speeds that replicated in situ conditions without inducing resuspension. Incubation baths and core headspace water were held within 1–2 °C of in situ temperatures. Following incubations, sediment cores were sub-sampled (upper 2 cm sediment, 66.37 cm3) for N content by drying (60 °C drying oven, 72 h) and homogenizing with mortar and pestle, and analyzed using a Perkin Elmer 2400 elemental analyzer.

Time series measurements of DO in headspace water were made through a port in the gas tight top using a YSI ProDO optical probe and DO was kept > 3 mg L−1. Oxygen uptake rate was determined by linear regression of the time course of headspace DO concentrations (N ≥ 5). Parallel time series measurements of NH4+, ortho-phosphate, NOx−, and dissolved organic N were conducted over a 16- to 18-h period under dark conditions, mimicking the low bottom water light levels at the coring site. An experiment to confirm the use of dark incubation was conducted under in situ light (measured as 43–88 µE m−2 s−1) and dark conditions in late September and no difference was found between DO uptake under low in situ light and dark incubation (t = 1.151, df = 15, p = 0.268). September cores were collected during a period of reduced phytoplankton biomass compared to spring and summer highs; however, the lack of difference between light and dark incubations indicates that light availability at the sediment surface was not sufficient to support benthic photosynthesis. Similar to DO, nutrient uptake/release rates were based on linear regression (N ≥ 5) of constituent concentration. The effects of the two treatments (treated vs. control) were examined by season and project year, and their interactions on SOD and nutrient fluxes were determined using a 3-way ANOVA. Data were transformed to meet assumptions of normality and equality of variance when necessary. A Tukey–Kramer correction was applied to adjust the p-values for multiple comparisons of treatments with unequal sample sizes (cores). Outlier analysis was performed prior to treatment comparison; cores were removed if surficial sediments were sulfidic or if a large burrowing organism, such as a quahog, was found upon breakdown of the experiment.

After completing DO and nutrient time series measurements, cores were again capped gas tight with a water headspace for time series measurements of N2 production. Measurements did not differentiate between coupled nitrification–denitrification, direct denitrification, or anammox; therefore, N2 production was used as a measure of total denitrification. Headspace samples were collected in 40-mL glass serum vials and preserved with 0.25 mL of 25% hydrochloric acid and capped with butyl rubber stoppers (Charoenpong et al. 2014). DO concentrations within the headspace were not allowed to drop below 3 mg L−1.

Denitrification was determined from linear regression of time series measurements of N2 excess as assayed using a Multicollector Isotopic Ratio Mass Spectrometry (MC-IRMS with GV IsoPrime analyzer; Charoenpong et al. 2014). N2 produced by denitrification is precisely detected from its ratio with the naturally occurring inert gas argon. Samples were pumped through a gas permeable membrane to extract gas into the mass spectrometer inlet. The inlet was fitted with cryogenic traps to remove water vapor and CO2. Sample gas was analyzed for masses 28 (N2) and 40 (Ar) for determining the N2 to Ar ratio. Calibration used a certified reference gas mixture of argon, N2, and oxygen. A one-tailed t test was used to assess the difference between denitrification rates in sediments from biodeposit affected versus background (control) areas, and oyster-mediated enhanced denitrification was determined as the difference between rates in treated sediments and control sediment areas.

Results

Oyster Growth and N Assimilation

Deployment and removal dates for each year, and dry tissue weights relative to bag count and deployment surface area, are shown in Table 1. Oysters were deployed for 175, 230, and 241 days (first date of deployment to last day of recovery) in 2016, 2017, and 2018, respectively, yielding a live whole oyster biomass increase through growth of 8,458, 8,769, and 10,079 kg, respectively. Nitrogen content of whole oysters was 0.31% by mass (live whole oyster biomass) for 0 + and 1 + oysters for all years except 2018; N content of 1 + oysters in 2018 was 0.40%.

Table 1 Dates of oyster deployment and removal, and dry tissue weight (DTW) densities relative to the number of bags deployed (g DTW bag−1) and pond surface area over which floating bags were deployed (g DTW m−2)
Full size table

Bottom Water DO: Continuous Autonomous Monitoring

The greatest DO depletion and daily O2 excursion occurred in late summer when temperature-dependent oxygen solubility was low and oxygen demand by marine communities was high; DO availability increased towards fall as waters cooled and fall storms increased water column mixing. Temperature profiles (not shown) indicated weak water column stratification during July and August. We observed consistent overlap between time series records of surface and bottom water DO concentrations for the duration of sonde deployment (Fig. 2). Here, the 2018 sonde data is shown because it provides the longest uninterrupted DO time series and is consistent with 2016 and 2017 results. During August, bottom water (0.3 m from bottom; ~ 1.8 m total depth) time series typically showed greater DO variability compared to surface waters (0.5 m from surface; ~ 3.5 m total depth), with daily O2 excursions up to ~ 6 mg L−1 (daylight O2 above atmospheric equilibration levels and nighttime O2 depletion). DO concentrations above Commonwealth MA water quality standards (> 6 mg L−1) accounted for 68% and 63% of the 2018 surface and bottom records, respectively. However, DO concentrations less than 6 mg L−1 occurred during the warmest period from early August to mid-September when SOD was greatest and bottom water re-aeration was periodically limited by low water column mixing. Despite direct contact with the atmosphere, O2 consumption in surface waters exceeded rates of re-aeration, resulting in DO declines to less than 4 mg L−1 during late summer.

Fig. 2
figure 2

August to November 2018 time series DO concentrations from surface and bottom water column moored sondes. The bottom sonde was deployed directly below the floating oyster bags (most biodeposits settle directly below source bags) to capture impacts of increased deposition on DO. Periodic low DO events (< 4 mg L−1) occurred when temperatures were above 20 °C; however, these events were typically brief (< 24 h) and DO was typically above 2 mg L−1

Full size image

Water Column N

Stream inputs of DIN to Lonnie’s Pond were highest in May and decreased through the summer. Interannual differences in rainfall and stream-related inputs resulted in a two-fold increase from 0.1 to 0.2 mg L−1 in average monthly DIN levels due to the loading variation from 2017 to 2018; in addition, the major fraction of NH4+ and NOx− in DIN shifted from NH4+ in 2017 to NOx− in 2018. In contrast to stream loads, water column NOx− was consistently the smallest fraction of DIN and TN pools; NOx− concentrations were typically ~ 3.3 µM lower compared to NH4+. Monthly NOx− and NH4+ concentrations showed a clear temporal trend with maximum levels in October (Fig. 3). October NOx− concentrations were ~ 16 times higher compared to June; therefore, the diffusional supply of water column NOx− (potentially available to denitrifying bacteria) was highest in October and lowest in June. Bottom water NH4+ concentrations in July, August, and September (~ 5–6 µM) represented summer highs likely because of higher rates of NH4+ production during the warmest months. Elevated levels of water column NOx−  + NH4+ in October, which could not be explained by stream input (DIN loading decreased ~ 80% from May to Oct/Sep), were likely the result of internal processes, such as regeneration of deposited organic matter and reduced primary productivity in response to cooling waters and decreasing light availability. PON concentration, like DIN loading, was highest in May and decreased towards fall (Fig. 3), suggesting that stream inputs of DIN to the pond were rapidly taken up by phytoplankton and converted from dissolved N to particulate N in phytoplankton biomass.

Fig. 3
figure 3

Mean monthly water column combined nitrate (NOx−; a), ammonium (NH4+; b), particulate organic N (PON; c), and bioactive N (NOx−  + NH4+  + PON; d) concentrations from April–November samplings of water quality sites distributed throughout the pond (2016–2018). Vertical error bars represent 1 standard deviation. Water column NOx.− can diffuse from the water column into the sediment for use by denitrifying bacteria via direct denitrification. Note the different y-axis limits

Full size image

Sediment Oxygen and Nutrient Fluxes

The highest SOD flux rates of each core sampling event were measured in treated cores, with the highest overall being in summer 2016 and 2018 (Table 2). The lower summer 2017 SOD rates corresponded with observations of minimal redox potential discontinuity (RPD; i.e., sediment color change from brown to grey/black) depth and sulfidic surficial sediments and the absence of benthic animals. The effect of the RPD depth can also be seen in the lower NH4+ and NOx− flux rates in summer 2017 compared to summer 2016 and 2018; N fluxes are described below. In addition, there was a significant interaction effect between treatment and season for SOD rates (n = 117) where summer treated cores had higher SOD compared to control cores and fall treated and control cores F(1, 105) = 5.33, p = 0.023 (Table 3). This interactive effect is the result of greater rates of bacterial activity during the warm summer months when there is higher organic matter availability in the biodeposit area.

Table 2 Comparison of mean ± SD sediment oxygen demand (SOD) and flux of dissolved inorganic nutrients (mmol O2, N, or P m−2 day−1) from treated (within biodeposit area) and control areas (outside biodeposit area), collected in Lonnie’s Pond during summer, fall, and spring* of 2016, 2017, and 2018. A positive value for nutrients indicates an average outward flux from the sediments, and a negative value an average uptake by the sediment from the water column
Full size table
Table 3 F-values of 3-way ANOVA for the main and interactive effects of treatment (biodeposit affected and background), season (summer and fall), and year (2016, 2017, and 2018) on sediment oxygen demand (SOD), and ortho-phosphate (PO43−), ammonium (NH4+), and combined nitrate (NOx−) sediment flux rates
Full size table

On average, there was net inorganic phosphorus release from the sediments to the water column in summer. However, we found a significant interaction between treatment, season, and year (3-way ANOVA, F(2, 89) = 3.98, p = 0.0221; Table 3). The effect of treatment and season differed by year because of the high mean summer treated PO43− flux measured in 2016 (Table 2). PO43− efflux from treated cores collected during summer months was significantly greater compared to both fall treatments and summer control cores. Also, PO43− efflux measured in August 2016 was significantly greater compared to October 2016 and summer and fall of 2017 and 2018. PO43− efflux from treated and control sediments occurred during periods of low bottom water DO concentrations (~ 1 mg L−1 in 2016) when RPD was very shallow/non-existent.

On average, NH4+ fluxes to bottom water in the biodeposit area was significantly greater than from control areas in all years and seasons; however, there was a significant interaction between treatment, season, and year (3-way ANOVA, F(2, 103) = 3.41, p = 0.0367; Table 3). The interaction stems from the greater release in summer versus fall in all years and both treatments where summer months had higher temperatures and biodeposition rates are greatest. However, NH4+ efflux from treated cores collected during summer months in 2016 and 2018 for all cores was significantly greater than 2017 summer rates.

Uptake and efflux of NOx− across the sediment surface varied seasonally with NOx− uptake during fall and spring when SOD is also lowest and net efflux during the warmer summer months (Table 2). Only August 2017 showed summer NOx− uptake. We found a significant interaction between season and year (3-way ANOVA, F(2, 97) = 5.41, p = 0.0059; Table 3). The average NOx− efflux from sediments receiving biodeposits in summer 2018 was about 6 times greater compared to 2016 and 11 to 33 times greater compared to 2017 (Table 2). This suggests that NH4+ production and a thicker oxidized surface layer (see below) favored nitrification. However, the treatment effect for NOx− efflux was not significant during any year. NOx− fluxes were generally similar in summer and fall both within and between years 2016 and 2017.

Small-scale sediment heterogeneity was observed by divers during all 3 years of the study, including differences in coloration of surficial sediments ranging from light brown to grey/black, and the presence of faunal burrows and burrow dwelling macrofauna (e.g., amphipods and polychaetes). Redox potential discontinuity (RPD) depth in individual cores varied between flux dates and core location on a single flux date. RPD depths tended to be shallower during periods of high organic matter loading and low (< 6 mg L−1) bottom water DO. Treated cores (i.e., biodeposit affected) collected in summer 2016 and 2017 were characterized by shallow or negligible RPD depths. September 2017 sediments were notable because nearly all treated cores had large surficial patches of black sulfidic mud. Sulfidic sediments persisted into October 2017, but to a lesser extent. Compared to previous summer fluxes, all July 2018 treated cores had a thicker oxidized surface layer. However, the oxidized surface layer gradually thinned from July to October 2018 with only a thin oxidized layer in several treated cores collected in October. April treated and control cores typically had a light brown uniform oxidized surface. In addition, it is important to note that shallow RPD depths were not limited to treated cores. Some August 2017 and July 2018 control cores also had patches of gray/black in the surficial sediments.

Nitrogen Deposition and Cycling in Sediments

Monthly N deposition rates (biodeposition + background deposition) associated with individual core locations were modeled for summer and fall 2017 and 2018 (Labrie et al. 2022); mass balance plots (Fig. 4, x-axis) represent four of these six dates. June 2017 data are not shown since the experiment was designed to capture summer, fall, and following early spring conditions; July 2017 is not shown since no sediment flux data were collected during that period; and September 2017 and October 2018 fluxes (not shown) were excluded because sulfidic surface sediments dominated the treated area. More specifically, September 2017 treated cores had the highest N deposition rates (563.7 mmol N m−2 month−1). Minimum N deposition rates (control) ranged from 40.5 in October 2018 to 111.7 mmol N m−2 month−1 in July 2018. In contrast, 2018 deposition rates in treated cores averaged 350.0 ± 191.7 mmol N m−2 month−1 in July and 220.9 ± 104.7 mmol N m−2 month−1 in October, whereas control areas averaged 191.9 ± 49.5 mmol N m−2 month−1 in July and 52.8 ± 16.1 mmol N m−2 month−1 in October. The decrease in background deposition rates from July to October reflected seasonality of water column PON concentrations, with July having the highest deposition rate followed by June 2017 (148.7 ± 16.1 mmol N m−2 month−1). The low oyster biomass in June 2017 and low growth rates in 2018 produced low deposition rates (primarily biodeposition N). For these reasons, June 2017 and July 2018 deposition rates did not differ between treated and control areas (June 2017: t = 1.95, df = 14, p = 0.072; July 2018: t = 1.96, df = 14, p = 0.071). In contrast, the deposition rates were higher in treated and control cores for all other fluxes dates (August 2017: t = 2.54, df = 14, p = 0.024; September 2017: t = 4.59, df = 14, p = 0.0004; October 2017: t = 4.68, df = 14, p = 0.0004; and October 2018: t = 3.85, df = 14, p = 0.002).

Fig. 4
figure 4

Mass balance of monthly predicted deposited N (N; biodeposition + background) and measured flux of N2-N (left axes), and total N (NH4+, NOx.−, N2-N, plus dissolved organic N; right axes) flux to the water column from the sediments in June 2017 (a), July 2018 (b), August 2017 (c), and October 2017 (d). Data points represent individual cores; triangles (open and filled) are control cores collected outside of the biodeposit area; and circles (open and filled) are treated cores collected within the biodeposit area. Deposited N is based on a biodeposit model (Labrie et al. 2022) specific to each core location. Linear regression of deposited N and N2-N flux (open triangles/circles) and deposited N and total N (filled triangles/circles) is shown. N2-N increased linearly with N deposition in June (95% CI 0.074 to 0.146, n = 9), July (95% CI 0.107 to 0.394, n = 16), and October (95% CI 0.008 to 0.129, n = 15). Total N release increased linearly with N deposition in June (95% CI 0.776 to 1.303, n = 16), July (95% CI 0.692 to 1.768, n = 16), August (95% CI 0.575 to 1.44, n = 14), and October (0.332 to 0.612, n = 15). Gray dashed lines represent 1:1 lines corresponding to total N flux on the right y-axis

Full size image

Monthly denitrification rates increased linearly with N deposition (biodeposition + background deposition) in June 2017, July 2018, and October 2017 (left y-axis; Fig. 4). Of the 3 months, June showed the closest relationship between N deposition and denitrification, explaining 80% of the variation. On average, 8% ± 5% of N deposition ended up denitrified, consistent with the slope of the linear regression (10.8%). June 2017 also coincided with the lowest measured water column NOx− levels. In October and July, denitrification also showed a positive linear increase with N deposition, but the proportion of variance in N2-N was greater than in June 2017 as seen in the lower coefficients of determination (r2 = 0.31 and 0.50 for October 2017 and July 2018, respectively; left y-axis; Fig. 4). The July 2018 results showed three times higher mass of N deposition denitrified (21% ± 12% of N deposition) than in June 2017, suggesting that the amount of N deposited and the environmental conditions in July were slightly more conducive to denitrification compared to June. The August 2017 flux did not yield a significant linear regression due to a large range of denitrification rates under low deposition rates, around 100 mmol N m−2 month−1 (left y-axis; Fig. 4). Denitrification accounted for 34% ± 21% (treated) and 88% ± 47% (control) of total deposition. Only a weak positive relationship was found between N deposition and denitrification in September 2017 (r2 = 0.079) or October 2018 (r2 = 0.026).

Monthly total N (dissolved organic and inorganic N, and N2-N) efflux from sediments was positively correlated with N deposition in June 2017, July 2018, August 2017, October 2017 (Fig. 4, right axis; see figure and caption for regression statistics), September 2017 (r2 = 0.472, slope = 0.278, 95% CI 0.071 to 0.485, n = 12), and October 2018 (r2 = 0.423, slope = 0.244, 95% CI 0.043 to 0.445, n = 12). During the summer months (June, July, and August), the slopes from linear regressions of deposition versus regeneration remained above the 1:1 line (i.e., slope ≥ 1). Similarly in fall (September 2017, October 2018), linear regressions also had slopes less than one. Furthermore, we calculated monthly rates of total N regeneration as a percent of monthly mass of N deposition to the sediment surface. This provides an estimate of the portion of total N deposition that was regenerated in the control vs. treated areas. Percentages did not differ between treated and control areas for June (t =  − 0.641, df = 14, p = 0.532), August (t =  − 1.147, df = 13, p = 0.272), and October 2017 (t =  − 1.499, df = 14, p = 0.156) and July 2018 (t =  − 0.416, df = 14, p = 0.684), indicating that regeneration was directly linked to deposition, and that the deposited organic matter is likely similarly labile between the two areas, which is reasonable since both are phytoplankton based. Treated and control data were thus pooled to examine the overall relationship of deposition to regeneration. The results suggest that an amount of N equivalent to the deposited N is remineralized rapidly, and percentages > 100% indicated some regeneration of residual sediment N from prior deposition. The percent regeneration of deposited organic N during summer months ranged from 107 to 186% and decreased to 55% ± 28% in October 2017. The proportion of total N regenerated to N deposited was lower in treated cores than control cores in September 2017 (treated = 51%, control = 195%) and October 2018 (treated = 59%, control = 130%). The lower regeneration in fall is consistent with the high rates of regeneration the following spring prior to the onset of biodeposition from spring oyster deployment.

Water column C:N molar ratios (Table 4) averaged 7.2 ± 1.0 (POC:1PON) for the combined flux dates. Molar ratios for the sediment–water column exchange rates (C:N and N:P) were calculated for each core where C:N was determined as the ratio of the rates of SOD and total N efflux, using a 1: 1 ratio of CO2:O2 and the empirical molar N:P ratio determined from the ratio of total N efflux rate to PO43− efflux rate (Table 4). Measured flux rates in treated and control cores did not yield significantly different C:N and N:P molar ratios (p > 0.05); therefore, treated and control ratios were pooled to examine the overall relationship. Averaging sediment flux C:N and N:P ratios over the 3 project years resulted in ratios of 9.9 ± 2.4 C:1 N and 17.7 ± 8.6 N:1P.

Table 4 Mean ± SD C:N molar ratios of water column particulate organic carbon (POC) and N (PON) and sediment flux of C and N. Molar ratios using individual core sediment oxygen demand and total N efflux (NH4+, NOx−, N2-N, and dissolved organic N). We assumed that the molar ratio of CO2:O2 during respiration was 1:1 on the condition that the reduced end-products of anaerobic respiration were oxidized by O2, in the surficial sediments
Full size table

Annual N Removed by Aquaculture: Denitrification and Assimilation/Harvest

Total annual N biodeposition was highest in 2017 (Table 6). 2018 had the largest initial oyster population; however, N biodeposition (9.2 g N m−2) was only ~ 40% of that in 2017 (23.3 g N m−2) and ~ 75% of that in 2016 (12.3 g N m−2). N biodeposition rates peaked in early September of all 3 years. September DTWs were 248 g DTW m−2, 420 g DTW m−2, and 226 g DTW m−2 in 2016, 2017, and 2018, respectively.

Enhancement of microbial denitrification within the biodeposit area resulted in 5.0–12.1 kg N (3.778 to 4.445 g m−2) removal per year over control areas. We observed significant denitrification enhancement on 80% of the flux dates; only August and September 2017 did not have significant denitrification enhancement (Table 5). In August 2017, we found NOx− uptake in six of the eight treated cores and one of the six control cores (pooled average =  − 0.06 ± 0.03 mmol N m−2 day−1); the percentage of total denitrification from direct denitrification (assuming measured NOx− uptake is equivalent to direct denitrification) averaged 2.6 ± 1.2% in August 2017 treated cores. Nitrate uptake was observed during fall of all 3 years. The percentage of total denitrification from direct denitrification in October 2016, 2017, and 2018 could be as high as 13.4% ± 7.6%, 11.9 ± 4.3%, and 32.4 ± 22.6%, respectively. Finally, early spring denitrification rates in sediments receiving biodeposits (pooled average year 1 and year 3 = 2.29 ± 1.51 mmol m−2 day−1) were nearly triple control rates (pooled average year 1 and year 3 = 0.84 ± 0.37 mmol m−2 day−1), but NOx− uptake was similar between treated and control cores (the pooled average for all cores year 1 and year 2 was − 0.06 ± 0.03 mmol m−2 day−1). During spring, direct denitrification contributed more to total denitrification in control areas than treated, and in year 1, the percent direct denitrification of total denitrification (treated and control) was more than double the percent direct denitrification of total denitrification in year 3. Looking at direct denitrification rates as a percent of total denitrification in year 1, the treated percentage was 3.3 ± 1.7% and the control percentage was 10.3 ± 3.8%, while in year 3, the treated was 1.7 ± 1.6% and control was 4.8 ± 3.2%.

Table 5 Incubation temperature (°C), denitrification (mmol N2-N m−2 day−1), and nitrogen (N) efflux (mmol N m−2 day−1) measured in intact sediment cores collected inside vs. outside biodeposit areas (treated vs. control). Difference between the two rates represents denitrification enhancement in biodeposit-affected sediments above background rates. The percentage of the total N denitrified is the portion of measured total N efflux that was N2-N. Nitrogen rates are averages ± SD
Full size table

The mass of N removed through harvest (i.e., oyster removal at end of the growing season) increased from 2016 to 2018, coinciding with increasing level of oyster deployment (Table 6). N assimilation normalized to deployment duration shows that oysters assimilated 148.3, 141.7, and 174.2 g N day−1 overall in 2016, 2017, and 2018, respectively. In 2016, the 1 + cohort contributed 76% of the total assimilated N from both cohorts, while in 2017 and 2018, 0 + and 1 + cohorts each contributed ~ 50%; the relative contribution reflected the proportion of 0 + and 1 + oysters deployed in each year. However, in 2018, 1 + oysters removed 3.3 more kg N compared to 0 + oysters because percent N content of 1 + oysters was higher compared to all other years and cohorts.

Table 6 Annual N removal budget for the 3 study years showing contributions from enhanced denitrification and oyster harvest
Full size table

Annual N Mass Balance

To examine how N dynamics are altered by floating oyster aquaculture, a mass balance of N cycling was constructed for the two water column/sediment systems in which measurements were made: (1) floating oyster aquaculture areas and (2) natural areas without aquaculture (Fig. 5). Sediment–water column exchange (flux) rates were expressed as g N g−1 DTW m−2 and g N m−2 for treated and control areas, respectively. Oyster N assimilation, N biodeposition, and dissolved inorganic N and N2-N sediment fluxes were determined through direct measurements. Settling of background N particles was estimated from average daily water column PON concentrations assessed through linear interpolation between water quality sampling events and a turnover of 15.5% per day (Lohrenz et al. 1987). Oyster filtration rate used published studies on the volume of water cleared of particles by oysters per unit time (Comeau, 2013), and average daily water column PON, to get the mass of N filtered from the water column per day. Comeau (2013) determined a clearance rate of 4.34 l h−1 standardized to 1-g DTW suspended oysters based on laboratory measurements on single oysters. Biodeposition rates standardized to a 1-g DTW oyster measured in situ from both single oysters and whole oyster bags in Lonnie’s Pond suggest that single oyster rates were on average 10 times higher than rates measured from whole oyster bags under comparable environmental conditions. Biodeposit traps deployed in Lonnie’s Pond during summer and fall measuring biodeposition rates (standardized to a 1-g DTW oyster) of individual oysters and whole oyster bags had an average (± SD) biodeposition rate of 158.8 ± 88.3 mg dry wt DTW−1 day−1 and 16.4 ± 6.6 mg dry wt DTW−1 day−1, respectively. The observed rate difference appears to result from the effects of crowding and competition between oysters for food particles in a floating bag versus when individually deployed. To account for this difference, we multiplied the Comeau (2013) clearance rate by a correction factor (0.10) to determine the volume of water cleared of N particles per day by the full oyster deployment. We estimated N excretion using an average rate of 6.0 µmol NH4+ g−1 h−1 (Magni et al. 2000) for 11 species of bivalve mollusc, which agrees with data from Srna and Baggaley (1976) for eastern oysters. Like clearance rate (i.e., filtration), the excretion rate was multiplied by 0.10 assuming a proportional decrease in excreted N with decreased N filtration. Last, burial rates of PON from biodeposition and background settling were estimated by applying a sediment accretion rate typical of Cape Cod embayments and measurements of N content in the upper 2 cm of sediment in treated and control areas. We used an average accretion rate calculated from annual accretion rate measurements in nearby Little Pond, Falmouth, based on Pb210 (3 mm year−1); accretion rates were generally comparable to rates of sea level rise.

Fig. 5
figure 5

Mass balance comparison of nitrogen (N) cycling in areas receiving biodeposits from floating oyster aquaculture (red arrows) vs. control/natural areas (beige arrows) without floating oyster aquaculture. All arrows represent a 3-year (2016–2018) average of N fluxes (mean ± SD) over the oyster growing season. Treated fluxes were normalized by the season average of dry tissue weight (DTW) biomass and biodeposit area for each project year, expressed as g N m−2 for a 225 kg DTW deployment. Control fluxes were normalized by the biodeposit area for each project year, expressed as g N m−.2. Transformations are as follows: (1) N removed from the water column via oyster filtration, (2) net N assimilated into oyster biomass over the growing season, (3) oyster excretion, (4) biodeposition, (5) background N deposition to the biodeposit area, (6) burial, (7) dissolved inorganic N efflux from mineralization and nitrification, (8) denitrification

Full size image

Oyster physiological processes vary seasonally with seasonal changes in temperature, food availability, and oyster biomass. Oysters directly contributed to permanent N removal through filtration and assimilation by incorporating 40% of ingested PON into oyster biomass (Fig. 5). In addition, oyster filtration and egestion transferred large amounts of PON from the water column to the sediments; this biodeposition accounted for 51% of total N deposition. The main pathways of N removal from active cycling associated with floating oyster aquaculture were assimilation/removal (64%), enhanced denitrification (22%), and burial (14%).

Discussion

Effect of Oyster Aquaculture on N Cycling

Comparison of the N mass balance of a floating oyster aquaculture system with the nearby estuarine area not affected by oyster aquaculture revealed that oysters have a significant effect on N and C cycling in their local environment. Results also show that significant N can be removed through their growth and subsequent removal from the system, and by their influence on sediment denitrification through their biodeposits (Fig. 5). Before floating oyster aquaculture was established in the relatively shallow waters of Lonnie’s Pond, deposition of organic N and C (primarily phytoplankton) to the sediment via zooplankton grazing and direct settling determined sediment respiration rates as modified by temperature. After aquaculture was established, oyster filtration/biodeposition more than doubled the mass of N reaching the sediments. Nitrogen mass balances within the floating oyster aquaculture system indicate that ~ 54% of PON filtered from the water column was incorporated into oyster biomass or excreted as dissolved N. Of the 54% PON that was assimilated, 74% was incorporated by the oysters into biomass. The amount of biodeposits was less well constrained, ranging from 33 to 46% (Fig. 5). A study by Songsangjinda et al. (2000) using floating bag oyster aquaculture found comparable apportioning with only a slightly higher value of 44% biodeposits. In addition, our percent biodeposit apportionment was lower than that reported by Newell and Jordan (1983; 50% filtered N released as biodeposit N) for bottom raised oysters, but this difference was expected given the physiological differences between surface and bottom raised oysters (Comeau, 2013).

A clear response in the cycling of sediment organic matter was found using measurements of oxygen consumption and NH4+ efflux within the biodeposit area (treated) compared to the control area, especially during the summer months (Table 2). The results suggest that biodeposits enhanced remineralization of organic C and N in surficial sediments, and more importantly, that denitrification was enhanced where biodeposition was occurring (Table 5; Fig. 5). The mass balance of N cycling in a floating oyster aquaculture system and an estuarine system not affected by oyster aquaculture indicates that, on an annual basis, the presence of oysters significantly increases respiration and denitrification in sediments. Oyster biodeposit production thus accelerates N cycling within the biodeposit area, which in turn enhances N removal from the estuarine ecosystem.

Seasonality of Biodeposit Remineralization

Tracking of sediment C and N remineralization and deposition indicated that sediments tend to remineralize most fresh deposits and even some stored deposits in summer, and then store organic matter in fall into the following spring. Efflux of remineralized N was well predicted based on biodeposition and an empirical model (Labrie et al. 2022). In addition, average C:N and N:P ratios of nutrient regeneration of all cores (9.9C:1 N and 17.7 N:1P) were comparable to the Redfield ratios for marine phytoplankton (i.e., 6.6C:N and 16.1 N:P). Mass balance regressions of predicted N deposition and sediment N efflux suggested rapid regeneration of organic N deposited during summer months as seen in linear regressions (Fig. 4, slopes above 1.0), with data above the 1:1 lines indicating net organic matter regeneration rather than storage. Rapid cycling of summer biodeposits was also evidenced by 3-way ANOVA results, where SOD and nutrient fluxes were significantly greater during summer months when phytoplankton biomass was high as a result of warm temperatures and light availability. The result was maximum oyster filtration and biodeposition and high background deposition rates. Particulate organic N, an indicator of food availability, peaked between 15 and 20 µmols L−1 during May and June, with concentrations trending downward towards fall (Fig. 3). The slope of the October 2017 linear regression of N deposition and N efflux indicated a ~ 50% lower regeneration rate per unit deposition compared to June 2017 and July 2018, suggesting a larger fraction of deposited N was being stored in the sediments in fall. These results supported our hypothesis that a seasonal lag exists between fall deposition of organic matter and the regeneration of nutrients and denitrification in early spring. We suggest that cooler fall temperatures lead to lower sediment flux rates and greater accumulation of labile organic matter in subsurface sediments. Organic matter respiration is a function of temperature (Jorgensen, 1975); therefore, decreasing temperatures between fall and winter lead to a concomitant decrease in microbial activity, which favors accumulation and storage of deposited N in sediments below the thin surficial oxic layer (Rudnick and Oviatt 1986). Enhanced denitrification in early spring increased annual N removal in oyster areas by 1.8 kg and 3.2 kg above previous summer and fall enhancement, yielding a 42% and 37% increase in the total annual N removed via enhanced denitrification in 2016 and 2018, respectively. Based on these results, future assessments of oyster aquaculture enhanced denitrification should include measurements the following early spring to capture the effects of overwintered biodeposits. Biodeposits accumulate during fall and may persist for months to 1 + years. Thus, the period of enhanced denitrification appears to extend beyond the period of oyster deployment.

Denitrification Increases with Organic Matter Loading to Oxidized Sediments

Our study showed that denitrification enhancement in biodeposit-affected sediments is a function of bottom water DO and the rate of biodeposition. Denitrification rates were found to be higher in sediments receiving biodeposition than in control areas on all sampling dates. Consistent with previous findings of seasonality (Fulweiler et al. 2013; Piehler and Smyth 2011), denitrification rates were highest in July/August. Interannual differences in summer and fall denitrification rates indicated that sediment redox conditions also influenced CND rates.

Denitrification rates were partially controlled by sediment oxidation status with higher rates observed when bottom water DO was greater than 5 mg L−1 and the sediment surface showed a clearly visible oxidized surface. Time series DO records showed periodic basin-wide water column hypoxia at some point during each of the 3 years of study, with the timing, frequency, and duration of hypoxia being comparable between years. Sediments collected within the area receiving biodeposition in late summer 2017 were characterized by areas of surficial sulfidic sediments and shallow/poorly defined RPD depth due to the “extra” organic matter loading. Denitrification was not significantly enhanced on these occasions. In fact, cores with little or no visible surface oxidization had lower than expected denitrification rates despite greater N deposition compared to cores collected outside of the biodeposit area, which maintained an oxidized surface. Furthermore, cores within the biodeposit area that had a well-defined oxic layer had the highest enhanced denitrification rates during the study (i.e., July 2018; Table 5). Studies have demonstrated that oxygen is required for nitrification and CND are diminished or inhibited in sulfidic sediments (Fenchel et al. 1998; Joye and Hollibaugh 1995). The degree of inhibition is determined by the porewater sulfide concentration, and duration and frequency of exposure of the nitrifying and denitrifying microbial community to sulfides (Caffrey et al. 2019). From a management perspective, it is important to emphasize that predictions of N removal via denitrification will be an overestimate if surficial sediments are sulfidic.

Denitrification efficiencies (denitrification as % of total N mineralization) remained relatively constant across a range of organic matter loading rates but varied temporally. Biodeposits from floating oyster aquaculture appeared to increase denitrification rates and total N regeneration proportionally, as seen in the similar denitrification efficiencies in biodeposit affected and background sediments. However, denitrification efficiencies varied between seasons and years and were significantly affected by sediment oxidation status due to the reduction of denitrification in sulfidic sediments. Efficiencies were highest in spring, and the year 1 spring flux was the highest recorded (Table 5). Spring efficiencies were similar to findings by Owens (2009), who reported spring flux denitrification efficiencies > 50%. Efficiencies were lowest in summer when NH4+ production was highest.

A key finding of this study is that biodeposition increases the rate of sediment denitrification, and that the extent of the increase depends upon the rate of deposition to the sediment surface, as long as the surficial sediments remain oxidized. Significant positive linear relationships between N deposition and denitrification suggest the potential for modeled N deposition to predict CND in estuaries when this condition is met. When this condition is not met, such as when summer stratification of the water column results in hypoxic bottom waters, the remineralization of organic deposits results in a loss of the oxidized surface layer of the sediments, which inhibits CND. To this latter point, we note that in the present study, reduction in sediment denitrification was measured at the Lonnie’s Pond aquaculture site at very high oyster densities (initial oyster densities per square meter of surface area were 224 oysters m−2, 794 oysters m−2, and 1852 oysters m−2 in 2016, 2017, and 2018, respectively) and where low bottom water oxygen conditions occasionally resulted in the loss of the surface oxidized layer, leading to sulfidic surficial sediments. Therefore, caution must be taken in floating aquaculture to not increase the stocking density to the point of causing bottom water hypoxia and reducing N loss via denitrification.

Maximizing N Removal from an Estuary via Floating Oyster Aquaculture

Our N mass balance reveals that floating oyster aquaculture enhances N removal via three pathways. Within the oyster system, oyster N assimilation (and subsequent removal from the system) accounts for the largest N removal, followed by CND enhancement, and then sediment N burial. This study quantified N removal via sediment burial, which enables a full accounting of N removal by oyster aquaculture; previous estimates typically include only N assimilation and denitrification. Although the N burial component is small, it provides additional evidence that oyster activities, namely, biodeposition of labile organic matter, stimulate N removal and accelerate the entire N cycle. Permanent N removal through N assimilation and subsequent harvest (oyster removal from the system at the end of the growing season) was significantly greater than N removal via enhanced dinitrogen gas production. This suggests that oyster deployments can maximize N assimilation by maximizing growth rate, and that denitrification can be maximized by sustaining oxidized surficial sediments. Aquaculture practices such as deployment and removal dates, initial oyster size, stocking density, gear type, maintenance activities, and deployment location and orientation relative to flow patterns can all influence oyster production (Ferreira et al. 2007) and N removal via assimilation (Kellogg et al. 2014; Rose et al. 2015). In the present study, N removed via assimilation only increased 5% from year 1 to year 2 despite a 200% increase in population size and 31% increase in deployment duration (Table 6). Farm-scale ecosystem models suggest that there is a site-specific optimal stocking density and deployment area configuration for all potential aquaculture sites, including Lonnie’s Pond, which maximizes growth rates without maximizing biodeposit or ammonia release (Ferreira et al. 2007). However, the usefulness of models to optimize an ecosystem service is potentially limited by the number and quality of the inputs and site-specific characteristics.

Conclusions

We quantified permanent N removal associated with floating oyster aquaculture in a small tidal salt pond on Cape Cod, southeastern MA. The results of this study suggest that when oysters are added to an estuary, their activities alter N cycling by accelerating the transfer of organic matter from the water column to the associated sediment area (biodeposit area) leading to increased sediment respiration. Without oysters, organic matter in the form of phytoplankton eventually settles to the bottom or is flushed from the estuary via the tidal outlet. We showed that denitrification increases with increasing N deposition in sediments that maintain an oxic surface layer. The N mass balance of the floating oyster aquaculture system indicates that the largest component of N removal is via assimilation into oyster tissue and shell and subsequent oyster removal from the system, followed by denitrification, and then burial. Nitrogen assimilation and denitrification were directly measured in this study. Enhanced denitrification (denitrification over background) within the sediment area receiving oyster biodeposits contributed an additional 27.9% removal of N from the system beyond that assimilated by oysters within their shells and tissue at time of removal from the system. Quantification of oyster-associated denitrification requires measurements both within the growing season and in the following spring, as labile biodeposits overwinter and continue to support denitrification into the following year. Site-specific aquaculture management practices (e.g., oyster deployment timing and oyster stocking density) can optimize oyster production of market-sized oysters, as well as increase biomass, which in turn increases N incorporated into tissue and shell over the growing season. These practices should support organic matter loading to sediments without affecting sediment redox conditions. Overall, the present study has showed that floating oyster aquaculture can significantly enhance permanent N removal primarily through denitrification and oyster harvest.