Nitrogen and Warming Effects on Nitrous Oxide Production Associated with Coastal Macroinvertibrates

Bivalve shellfish potentially reduce excess nitrogen in the water column, however they can also be involved in the emission of nitrous oxide (N2O), a potent greenhouse gas. Environmental controls on N2O production from bivalves have not been well quantified. We tested responses of N2O production by three bivalves (Mytilus edulis, Mercenaria mercenaria and Crassostrea virginica) to nitrogen (N) loading and/or warming after immediate (1 day) and short-term (14-28 days) exposure. This twofactor laboratory study had four treatments: (1) ammonium nitrate (N) addition (targeting 100μM-N), (2) warming (22C), (3) N addition + warming and (4) no N addition or warming (control, 19C). Potential N2O production rates were higher in response to N additions for all bivalves, particularly with short-term exposures. Warming had a small but significant impact on N2O production from M. mercenaria, confounded by a significant interaction of exposure X warming and exposure X nitrogen X warming. Similarly, C. virginica also showed a significant interaction of exposure X warming, indicating that longer exposure to warming may influence N2O production from this species. M. edulis showed the highest N2O production rates, reaching 252 nmol N2O ind hr, more than an order of magnitude higher than the previously largest reported rates. However, mass-specific rates (7.5 nmol N2O g hr) were the same order of magnitude as previous studies. Notably, N2O production associated with M. edulis were obtained while the organisms had poor health, which likely induced high respiration rates and was probably caused by hypoxic water conditions. We also examined the influence of macro-epifauna on the N2O production associated with M. edulis via removal of macro-epifauna. There was no significant difference in N2O produced by M. edulis with and without epifauna, which suggests that N2O production may be largely due to gut microbial activity and microbial biofilms on the shells from M. edulis. In summary, our study indicates a strong influence of N on the potential N2O emissions rates of prominent bivalves, which should be considered when incorporating them into coastal N mitigation strategies.


INTRODUCTION
Bivalves are frequently used in water quality restoration projects, as they can mitigate the negative consequences of eutrophication by significantly impacting fixed nitrogen (N) cycling in aquatic systems (Stief 2013;Kellogg et al. 2014). Through filter-feeding, they can influence benthic-pelagic coupling by removing fixed N and phytoplankton biomass from the water column that together contribute to eutrophic systems (Carmichael et al. 2012;Kellogg et al. 2014). Along with fixed N, bivalves ingest denitrifiers that can remain metabolically active in their anoxic gut microenvironment, which further impacts aquatic N cycling (Stief et al. 2009). Also, reef building bivalves can enhance the surface area for N-cycling microbes to grow on their shells, thereby removing or producing N through nitrification, denitrification or coupled processes (Svenningsen et al. 2012;Heisterkamp et al. 2013). Bivalves can also be sources of ammonia through their excretions and biodeposits at high rates such that this may counter the amount of fixed N they can remove from the water column (Stief 2013). Importantly, recent investigations have found that by enhancing Ncycling processes, bivalves can also stimulate the production of nitrous oxide (N 2 O) (Stief et al. 2009;Heisterkamp et al. 2010;Svenningsen et al. 2012), a potent greenhouse gas that is a major contributor to climate change and ozone depletion (Forster et al. 2007).
Denitrification in the anoxic guts of bivalves as well nitrification and/or denitrification on their shell biofilms produce significant amounts of N 2 O to the water column (Svenningsen et al. 2012;Heisterkamp et al. 2013). N 2 O is produced through microbial nitrification (the oxidation of ammonium to nitrate) during the first oxidation step (Goreau et al. 1980) and as a byproduct in denitrification and nitrifier denitrification (Zumft 1997;Wrage et al. 2001). When complete, denitrification is actually a sink for N 2 O (converting it to N 2 ), but in the case of invertebrate guts, the pathway may be terminated early with N 2 O due to inhibition of N 2 O reductase (nosZ) by excess nitrate similar to marine sediments (Miller et al. 1986). The sources of N 2 O production can vary among species as Svennignsen et al. (2012) found that approximately 25% of the N 2 O production associated with the zebra mussel, Dreissena polymorpha was ascribed to the shell biofilm, while Heisterkamp et al. (2013) found that the N 2 O production from the blue mussel, Mytilus edulis, was almost entirely due to shell biofilm microbiota. N 2 O production from invertebrates is also known to be positively related to biomass (Stief and Eller 2006;Stief et al. 2009;Stief and Schramm 2010), whereby large bivalves can have large N 2 O production potentials compared to smaller individuals. This is due to the increase in gut size with increase in biomass and can therefore house more microbial consortia. Bivalve associated N 2 O production is of particular concern due to their large abundances in benthic coastal habitats and in aquaculture (Stief et al. 2009;Svenningsen et al. 2012;Heisterkamp et al. 2013); thus, these areas can be potential "hot spots" of N 2 O emissions to the atmosphere.
Excess fixed N and ocean warming are key environmental stressors that may increase N 2 O emissions, though the effects on marine shellfish have been largely understudied. The increased availability of fixed N fertilizer and atmospheric deposition is affecting coastal habitats worldwide (Gruber and Galloway 2008). N 2 O is known to be emitted in these areas of high concentrations of fixed N in sediments and water column (Seitzinger et al. 1983;Kroeze and Seitzinger 1998;Corredor et al. 1999;Muñoz-Hincapié et al. 2002). Thus, highly eutrophic habitats may significantly increase N 2 O production associated with marine bivalves by providing copious amounts of substrate for nitrification, denitrification and coupled processes to occur. Also, N 2 O production rates can increase as a function of temperature due to higher microbial activities as has been shown in marine sediments, grasslands and in microbial biofilms in rivers (Kroeze and Seitzinger 1998;Boulêtreau et al. 2012).
Consequently, ocean warming may stimulate N 2 O production by increasing the metabolism of nitrifying and denitrifying bacteria on the shells and within the guts of bivalves. Ocean warming, driven by enhanced carbon dioxide in the water, can significantly decrease growth of larvae and juvenile bivalves, increase mortality and increase respiration (Dove and Sammut 2007; Dickinson et al. 2012;Matoo et al. 2013;Mackenzie et al. 2014). Impaired health of the organism due to warming may stimulate more oxygen demand and thus increase filter-feeding, ingestion of denitrifiers and subsequent N 2 O emissions. Further, filtration rates and temperature have a positive relationship (Kittner 2005), which would potentially increase fixed N into bivalve guts, stimulating denitrification. Thus, synergistic interactions between N rich environments and warming can potentially intensify N 2 O production associated with marine bivalves.
The goal of this research was to: (1) examine the effect of N addition and/or warming on N 2 O production from three coastal bivalves (Mytilus edulis, Mercenaria mercenaria and Crassostrea virginica), (2) determine if biomass is a control for N 2 O production and (3) examine the contribution of macro-epifauna on one invertebrate (M. edulis) to overall N 2 O production rates via a macro-epifauna-removal experiment in the presence and absence of N loading. In order to assess how elevated N and/or warming affects rates of N 2 O production from these taxa, a two-factor aquarium study with 4 treatments was used: (1) N addition, (2) warming, (3) N addition + warming and (4) controls (with no N addition or warming). N 2 O production rates were examined after an immediate (1 day) exposure and short-term (14-28 day) exposures.
We hypothesized that both N additions and warming treatments would significantly increase the N 2 O emissions rates from each species and that N 2 O would increase between exposure periods. We also expected M. mercenaria to have the highest N 2 O production per gram of biomass and the smallest species, C. virginica to emit the lowest amount per gram. Lastly, we anticipated macro-epifauna organisms to contribute significantly to the N 2 O production of M. edulis, particularly in the presence of N additions. To identify further potential controls on N 2 O emissions by these invertebrates, we tested relationships of N 2 O production rates to water quality parameters (pH, DO, dissolved inorganic nitrogen) and physiological status of the three bivalve species. Our study contributes to the few studies that have examined bivalve N 2 O production, especially those in response to global change drivers.

Field site and collection
For this study, multiple individuals of M. edulis, M. mercenaria and C. virginica were collected from subtidal Narragansett Bay, Rhode Island, USA through scuba diving in Narragansett, Conimicut Point and North Kingstown, respectively. C. virginica was collected during the early summer season. M. mercenaria and M. edulis were collected during the late summer. Narragansett Bay, Rhode Island has a well-documented bay-wide gradient in anthropogenic N loads, declining from the north to south (Deacutis 2008;Oczkowski et al. 2008). The nutrient maximum within the bay is in the Providence River and upper bay, with an annual average total nitrogen concentrations (TN) of l70 µM and total phosphorus (TP) of 5 µM, including dissolved particulate, organic and inorganic forms (Oviatt et al. 2002). The lowest nutrient levels are in Rhode Island Sound with an annual average concentrations for TN of 12 µM and TP of 1 µM (Oviatt et al. 2002). There is minimal freshwater input into the bay, thus surface and bottom salinities are high, ranging from 20-32 ppt (Pilson 1985). Water temperatures of the Narragansett Bay range from 1 ºC to 23 ºC in the middle and lower bay with minimal differences between surface and bottom waters (Kremer and Nixon 1979). Warmer waters have been observed over recent decades, with an average 2.2ºC increase during the winter since 1960 (Nixon et al. 2009), along with occasional hypoxic events (Deacutis et al. 2006;Codiga et al. 2009).
We collected seawater (40 mL) at the surface and at the depth of each bivalve collection site (depths ranging from 3-8 meters) and analyzed them on board the research vessel to determine salinity using a handheld refractometer, temperature, pH and dissolved oxygen (DO, Thermo Scientific Orion Star A326 pH/Dissolved Oxygen Portable Multiparameter Meter). To determine dissolved inorganic nitrogen (DIN) concentrations, a subsample of the water was filtered using disposable syringe filters (Advantec; 0.45 µm; sterile) and frozen (-17ºC) at the time of collection and later analyzed for ammonia and nitrate using a micro-segmented continuous flow nutrient analyzer (Astoria Analyzer, model 303a).
Immediately after collection, organisms were transported to environmental temperature chambers (Holman Engineering) at the Marine Sciences Research Facility (MSRF) at the University of Rhode Island Bay Campus. They were acclimated in clear glass "batch style" aquaria ( Figure 1) in aerated, unfiltered seawater at either ambient water temperature (19ºC) or warming treatment (22ºC) for at least 12 hours before start of the incubations. Gas tight tops for the aquaria were equipped with inflow and outflow sampling ports ( Figure 1). A magnetic stir bar inside each tank was used and gently driven at 40 rotations per minute by a magnetic carousel in order to homogenize the water (Figueiredo-Barros et al. 2009). Unfiltered seawater pumped directly from Narragansett Bay was used in order to simulate natural microbial populations and to maintain a food source similar to that in situ.

Nitrogen and warming manipulations
Nitrogen and warming manipulations (NxW) were applied to invertebrates with 3-4 replicate aquaria (10.1 L, Figure 1) which had been acclimating overnight in the same aquarium. For N addition treatments, each aquarium received a pulse of 10mM ammonium nitrate prepared in unfiltered seawater at a volume sufficient to increase the overall DIN concentration to 100 µM-N. This N level was selected in order to target maximal total N concentrations in the bay that were previously reported (Oviatt et al. 2002). N additions were made to the same water that the invertebrates had been acclimating in and immediately prior to sealing the aquarium for each of the five-hour incubations. All aquaria were maintained in the dark (except during sampling and tank maintenance) to limit DIN uptake by any active phytoplankton.
Based on approximate field abundances in New England waters (Schulte et al. 2009;Tam and Scrosati 2011), three individuals were added per aquarium for M. edulis and C. virginica. M. mercenaria aquaria received 1 individual per tank (Thelen and Thiet 2009). Aquaria were randomly assigned to treatments. Seawater was replaced (with appropriate N levels and water temperatures) every other day in order to maintain adequate food supply for the organisms and in attempt to control ammonia accumulation from animal excretions.
Temperatures were manipulated by placing the aquaria into one of two environmentally controlled chambers. Control treatments were kept at 19 o C based on average summer temperatures in the bay over the past decade (Nixon et al. 2009).
Since the temperature of the Narragansett Bay generally does not vary significantly between the surface and bottom during the summer months, aquaria receiving the warming treatment were maintained at 22ºC based on the average projected increase of ocean surface temperatures in New England by 2100 (Mora et al. 2013).
N 2 O production rates were determined for bivalve in each aquaria over immediate (1-day) five hour long incubations in each of the four treatments. To compare responses over longer terms, the N 2 O production rates were determined again after 14-days (for M. edulis) and 28-days (for M. mercenaria and C. virginica). These time periods will hereafter be referred to as "immediate" and "short-term", respectively. The "short-term" time period was shorter for M. edulis than for the other taxa due to mortality of the organisms during the course of the experiment. For M. edulis, two weeks was the longest time period over which at least three live organisms could be maintained per aquarium, and three replicate aquaria were maintained rather than four for the other species. Each bivalve species was tested in a separate, sequential experiment using the same aquarium, which were cleaned with freshwater and bleach between experiments.

Sea water only experiment (no invertebrates)
To test potential N 2 O production from the water column alone (without invertebrates) in the presence or absence of N additions we employed a two-factor design with the following 4 treatments: (1) filtered seawater (0.2µm), (2) filtered seawater with N addition (100 µM-N as in the bivalve experiments), (3) unfiltered seawater and (4) unfiltered seawater with N addition. Water temperatures were kept at 19ºC and followed the same protocols for the bivalve experiments as described above with a five-hour total incubation period. edulis individuals were obtained from a seawater out-take pipe at the MSRF and acclimated overnight in aquaria. Individuals were randomly assigned to each tank in a two-factor design with the following treatments: (1) no macro-epifauna (experimentally removed), (2) no macro-epifauna + N addition (at the same levels as the N addition experiment above, 100 µM-N), (3) intact macro-epifauna (not removed), (4) intact macro-epifauna + N addition. In the "no macro-epifauna" treatments, macro-epifauna were gently removed from M. edulis individuals using a knife. Four M. edulis individuals were used per aquarium. Individuals assigned to the "macro-epifauna" treatment were handled similarly but not scraped with a knife, in attempt to maintain consistency. All aquaria for this experiment were maintained in the environmental control chambers at 19 o C. The duration of this experiment was the same as "immediate exposures" in the N addition and warming experiment. They were not maintained longer than one day and N 2 O production was determined within five hours of the experimental manipulations.
Water samples (35mL) from the aquaria were taken with 60 mL nylon syringes equipped with stopcock valves (Cole Parmer) immediately after sealing the aquaria with a gas tight cover by first establishing a gentle siphon in the outflow sampling port, flushing out the first 35mL in the syringe with water from the aquaria, and then collecting the next 35mL of water in the syringes (T 0 ). The same procedure was performed for three and five hours. For aquaria with N treatments, water samples were taken after immediate N additions immediately after sealing of the lid (T 0 ). Samples for the M. mercenaria and M. edulis short-term N addition and warming experiments were preserved with 1mL of 50% w/v zinc chloride in gas tight septum bottles and stored at room temperature until further analysis. At time of analysis, water samples were equilibrated with ultra-high purity helium (Moseman-Valtierra et al. 2015) approximately three weeks after collection. All other samples were equilibrated within fifteen minutes of collection and headspace gas concentrations were measured on a gas chromatograph (Shimadzu GC-2014) equipped with an electron capture detector (325ºC).

N 2 O production rates
For gas chromatography, helium was used as a carrier gas and a 5% methane mixture (balance argon) was used as a makeup gas with a flow rate of 2.5 mL min -1 .
The SPL-2014 capillary column with copper packing had a flow rate was 25 mL min -1 based on specifications made by manufacturer (Shimadzu). Two specialty gas standards (Airgas, Billerica, MA) were used to construct standard curves with concentrations of 0.5 ppm to 2.1 ppm for N 2 O. The detection limit of the GC is where C w is the dissolved concentration of N 2 O (nmol L -1 ), K 0 is the solubility coefficient for N 2 O (mol L -1 atm -1 ), x' is the dry gas mole fraction of N 2 O in the sample headspace (ppb), P is the atmospheric pressure (1 atm), V wp is the volume of water phase (mL), V hs is the volume of the headspace (mL), R is the gas constant (L atm K -1 mol -1 ), and T is temperature upon equilibration (K) (Weiss and Price 1980;Walter et al. 2005).

Water quality parameters
Before and after each incubation, at 0 and 5 hours, pH and DO were measured as described above. 40 mL of water samples were also taken at the same time points, filtered (0.45 µm) and frozen for subsequent N analysis as described above.
For both experiments, changes in water properties (pH, DO) within each aquarium were tested before and after the five-hour periods of each N 2 O production assay. DO was not measured for either of the M. mercenaria experiments or for the M. edulis short-term exposure due to the meter being unavailable for use. As with N 2 O production rates, these were performed separately for each bivalve species.

Invertebrate condition index
After incubations, organisms were frozen (-17 o C) and later analyzed to determine their condition index (CI) in order to assess their physiological status. They were shucked using a blade and wet weights of soft tissues and shells were determined on an analytical balance. Organisms were then placed in a drying oven at 70°C for 48 h (when a constant weight was reached) and re-weighed. The following formula by Crosby and Gale (1990) was used to calculate the CI:

STATISTICAL ANALYSIS
A mixed model ANOVA was used to determine the statistical significance of differences in N 2 O production rates (normalized per individual) among experimental treatments for N+W experiment (factor 1= nitrogen, factor 2= warming, factor 3=exposure) for each bivalve species. Due to gas loss from two syringes during transportation, one outlier from the immediate exposure and one from the short-term exposure were removed from the M. mercenaria data set. N 2 O production by M.
mercenaria was log transformed to achieve equal variance among treatments. A twofactor ANOVA was used to compare biomass normalized N 2 O production, mortality rates for M. edulis and condition indices among treatments for the three bivalve species.
Linear regressions were performed to test relationships between N 2 O concentrations and water quality parameters (ammonium, nitrate, pH, DO), wet mass of the organisms and invertebrate CI. Regressions were made using all treatments and exposures per species.
A mixed model was applied to test for differences among treatments in the N addition and warming experiment for ammonium, nitrate, pH and DO.
For aquaria without invertebrates (seawater controls), differences in N 2 O production rates from the water column were also tested with a two-factor ANOVA (filtration, nitrogen).
A two-factor ANOVA was used to test for significant differences in N 2 O production rates by M. edulis in the epifauna removal experiment (epifauna presence, nitrogen).
For all ANOVAs, assumptions of equal variance and normality were tested using the Bartlett test and Shapiro-Wilk test, respectively. All statistical analyses were performed with JMP 10.0 software and significance levels of a=0.05 were used.

Field site characterization
Surface water ammonia (NH 4 + ) concentrations ranged between 25.0-31.5 µM-N between collection sites, while bottom water concentrations ranged from 22.6-24.4 µM-N (Table 1). Surface and bottom water nitrate (NO 3 -) concentrations ranged from 0.6-1.8 µM-N and 1.3-2.3 µM-N, respectively (Table 1). Conimicut Point had highest NH 4 + and NO 3 concentrations compared to Narragansett and North Kingstown collection sites, as well as lowest pH and DO concentrations (Table 1). Temperatures ranged between 12-23ºC and were generally similar between depths (Table 1). Salinity ranged 28-32 ppt and was similar among sites and depths ( Table 1).

Mytilus edulis
M. edulis incubations under short-term exposures had the highest N 2 O production rates observed in all of the experiments, averaging 251.9 ± 11.6 (+N) and 206.0 ± 47.8 (+N+W) nmol ind -1 hr -1 ( Figure 2). N 2 O emissions in N addition treatments were approximately five times greater in both immediate and short-term exposures than in -N-W controls, but were not impacted by warming, with no significant nitrogen X warming interaction ( Figure 2, Table 2). N 2 O production was approximately five times higher in the short-term exposure compared to the immediate exposure, with a significant exposure x nutrient interaction ( Figure 2, Table 2).
N 2 O emission rates normalized per wet mass (g) for M. edulis under short-term exposure (Table 3) were significantly affected by N and NW treatments, though no significant effect was seen by warming alone (Table 4).

Mercenaria mercenaria
N 2 O production rates of M. mercenaria were significantly affected by nitrogen, warming and exposure ( Figure 3, Table 2). There were also significant interactions among exposure X nitrogen, exposure X warming, and exposure X nitrogen X warming ( Figure 3, Table 2). The largest change in N 2 O production between immediate and short term exposures was the five-fold increase between in the N addition only treatment ( Figure 3). The presence of N alone significantly affected N 2 O production rates when quantified per wet mass (g) of organism for the short-term exposure treatments (Table 3,4).

Crassostrea virginica
In contrast to the other bivalves, N 2 O consumption was observed in several of C. virginica treatments. Highest average rates of N 2 O consumption were -5.5 ± 1.3 nmol N 2 O ind -1 h -1 (control treatments; Figure 4). There was a significant effect of N addition on the N 2 O production by C. virginica, and largest N 2 O emissions were found in the short-term N addition only treatments ( Figure 4, Table 2). There was also a significant antagonistic effect of exposure X warming, with a lowered N 2 O production in the short-term W and NW treatments ( Figure 4, Table 2).
When short-term N 2 O production rates were normalized for C. virginica per wet mass (g) for the short-term exposure period (Table 3), N 2 O emissions were significantly affected by nitrogen, warming and their interaction (Table 4).

Dissolved inorganic nitrogen
In the N+W experiment, NH 4 + and NO 3 concentrations were generally higher in the N addition and NW treatments for the three bivalve species studied (Figure 5-6), but initial (T 0 ) concentrations of NH 4 + and NO 3 varied between bivalve species despite similar N manipulations in both immediate and short-term exposures.
NH 4 + concentrations were high in all M. edulis treatments and were approximately four times higher compared to the other species studied ( Figure 5A). In M. edulis, there were unexpectedly high NH 4 + values in both immediate and shortterm exposure even in control treatments, and a decrease of NH 4 + concentrations in the N treatments during both exposure times ( Figure 5A). There was a significant positive effect of nitrogen X warming and exposure on final NH 4 + concentrations for M. edulis (Table 5). The change of NH 4 + between start and end of incubations was also significantly affected by nitrogen, warming and their interaction (Table 5). In the M.
edulis aquaria, T 0 and T 5 NH 4 + concentrations declined as N 2 O concentrations increased, in the N addition treatment only (R 2 =0.54, p=0.04). NO 3 concentrations were generally higher in the short-term exposure than the immediate and were higher than expected at the start of the incubation for all treatments ( Figure 6A). There was a significant effect of nitrogen, nitrogen X warming and exposure X warming on the final NO 3 concentrations of M. edulis (Table 5). There was no significant treatment or exposure effect on the change of NO 3 concentration between the start and end of the incubation (Table 5). No relationship was found between N 2 O production rates and change in NO 3 concentrations in the M. edulis experiments.
In M. mercenaria incubations, initial NH 4 + ( Figure 5B) and NO 3 concentrations ( Figure 6B) were two to three times higher in treatments with N additions, for both immediate and short-term periods ( C. virginica incubations showed higher NH 4 + and NO 3 concentrations at T 0 in the N addition treatments for the immediate and short-term exposures ( Figure 5C, 6C).
There was also a decrease of NH 4 + and NO 3 between T 0 and T 5 for C. virginica ( Figure 6C). Final concentrations of NH 4 + + and NO 3 were significantly affected by all treatments and interactions (  Figure 6C).

Dissolved oxygen and pH
Between T 0 and T 5 , pH and dissolved oxygen (DO) generally significantly decreased for all organisms and all treatments ( In M. mercenaria experiments, pH was significantly affected by time (Table   7), such that pH was lower by the end of the incubation compared to the start (Table   6). pH was also significantly affected by warming and the interaction of warming X time (Table 7).
Time points also affected C. virginica pH (Table 7) driven by lower pH at the end of the incubations (Table 6,7). Nitrogen, warming and warming X time also significantly affected pH (Table 7). DO levels were significantly affected by time and warming (Table 6,7). Though levels did significantly drop between start and end of incubations, DO levels did not fall below 3 mg L -1 for C. virginica in either immediate or short-term incubations (Table 6).

Macro epifaunal contributions associated with M. edulis N 2 O production
The macro-epifauna removal and N-addition experiment with M. edulis showed lower average N 2 O production rates (Figure 7) than M. edulis N+W experiment ( Figure 2). There was a significant effect of N treatment on N 2 O production (F 3,10 = 6.7, p=0.03). However, there was no significant impact of the presence of macro-epifauna (F 3,10 =2.1, p=0.18) nor the combination of nitrogen X epifauna (F 3,10 =0.03, p=0.87). Although N 2 O production rates were qualitatively higher for M. edulis with macro-epifauna than those without them, only increased N levels had a significant effect on the rates of N 2 O production.
pH (t 13 =-7.3, p<0.01) and DO levels (t 13 = -13.6, p<0.01) were significantly lower between the start and end of the experiment for the M. edulis macro-epifauna removal experiment (Table 6), but DO concentrations did not reach hypoxic levels.
N 2 O production rates from all treatments showed a weak negative relationship towards pH (R 2 =0.37, p=0.02) as well as a weak relationship towards dissolved oxygen (R 2 =0.33, p=0.03) of the aquarium water.
Physical condition of bivalves M. edulis individuals had a 31% mortality rate over the short-term 14-day incubation period. During this period, the control treatments had a total of 3 mortalities and the experimental treatments had 4 mortalities per treatment (out of 12 individuals for each treatment). There was no significant difference in mortality rates per treatment (F 3,12 =0.04, p=0.99). The other species did not experience mortality during this study, nor was there any mortality in the M. edulis macro-epifauna experiment.
By the end of the experiment, M. mercenaria had the highest condition index (CI) among the three species while C. virginica had the lowest (Figure 8). There was a significant interaction between nitrogen and warming on the CI of M. edulis (F 3,44 =2.12, p=0.02). C. virginica CI was significantly affected by warming (Figure 8, F 3,44 =0.67, p<0.01). No significant relationships were found between CI and N 2 O production for all taxa between different experimental treatments, but there was a positive relationship for C. virginica (R 2 = 0.17, p<0.01).
There was a weak but significant positive relationship between wet weights of the individuals and N 2 O production rates, though was largely driven by large variability in weights and production rates of M. edulis.

DISCUSSION
This study revealed that prominent benthic bivalves can enhance water column N 2 O, particularly in response to prolonged exposure to fixed N. The increase of N 2 O production in the N addition treatments for all three bivalves is reasonable, as more N is available for various metabolic pathways of microbes associated with the bivalves.
We virginica between exposure times (Figures 2-4). The N 2 O production per individual from the -N-W control treatments are an order of magnitude greater than in previous studies examining M. edulis without N or warming manipulations (Steif et al. 2009, Heisterkamp et al. 2013). However, the organisms used in our study were ~60 times heavier (per wet mass) than those used in previous studies, and when compared to mass specific N 2 O, our results for the N+W experiment are within the same range as previous studies (Stief et al. 2009;Heisterkamp et al. 2013).
The high N 2 O emissions from M. edulis may be also due to their high N   We expected M. mercenaria to be the highest contributor to N 2 O production; however, this species showed the second highest bivalve associated N 2 O production, whereby N addition treatments significantly increased N 2 O production particularly over prolonged exposure to N (Figure 3). Studies have found that filtration rates are significantly lower by M. mercenaria when they are assayed in aquaria without sediments compared to those with sediments where they are allowed to burrow (Caughlan and Ansell 1964;Riisgård 1988). Therefore, this species may have not been filter-feeding at high enough rates to store enough denitrifying bacteria in their large guts. This species also did not show evidence of high excretion rates as shown by M. edulis (Figure 4), which could have prevented the induction of high rates of coupled nitrification-denitrification processes.
C. virginica showed the lowest overall N 2 O production rates, although the N 2 O production associated with C. virginica in the -N -W control treatments (0.5-2.8 nmol N 2 O ind -1 h -1 ) is the same order of magnitude as other aquatic bivalves (Stief et al. 2009: 200;Heisterkamp et al. 2010;Svenningsen et al. 2012). We hypothesize that consumption of N 2 O, which was observed in the immediate control and N addition only treatments, was likely due to complete denitrification. N cycling related to C.
virginica found that denitrification rates increase in sediments with oyster reefs relative to bare sediments (Piehler and Smyth 2011;Smyth et al. 2013;Kellogg et al. 2014). It is likely that N 2 O production will vary when C. virginica is assayed with sediments and may in fact increase N 2 O when sediment bacteria are allowed to interact with the fixed N.

Another possible reason for N 2 O consumption is competition for DIN between
N cycling bacteria and phytoplankton in the aquaria (Sundbäck et al. 2000). The phytoplankton density in the unfiltered aquarium water column may have been large enough to outcompete the ammonia oxidizing bacteria and denitrifiers for fixed N (Vieillard and Fulweiler 2014), though this should have been minimized as we performed this study in the dark to minimize the role of phytoplankton. Further, the organisms may have not been excreting at such high rates since there was no NH 4 + production by the end of the five-hour incubation ( Figure 5C). There was minimal consumption of NH 4 + and NO 3 -( Figure 5C), which also suggests low rates of nitrification and denitrification, respectively.

Warming impact on bivalve N 2 O production
Our findings indicate that warming had a slight impact on N 2 O emissions associated with these bivalves. M. edulis was not significantly affect by warming, however M. mercenaria production rates were affected by warming alone and C.
virginica only showed significantly warming impact when combined with exposure ( Figures 2-4, Table 2). Warming impacts on M. mercenaria production rates were likely pulled by the interaction of exposure X warming and exposure X warming X nitrogen ( Figure 3, Table 2 There was a transient increase in N 2 O production for C. virginica under immediate exposures to warming (Figure 4). In the immediate exposure, the production of N 2 O in the NW treatment was possibly due to increased rates of coupled nitrification-denitrification activity since there was higher NH 4 + and NO 3 decrease compared to the non-warming treatments ( Figure 5C, 6C). There was less NH 4 + in the short-term NW treatment compared to the immediate exposure NW treatment ( Figure   5C), which may not have given the ammonia oxidizing bacteria enough substrate to metabolize even under warming conditions. our results and past research that has found shell biota to be significant contributors (Heisterkamp et al. 2010;Svenningsen et al. 2012;Heisterkamp et al. 2013).
M. edulis individuals from the epifaunal removal experiment (Figure 7) also had lower N 2 O production when compared to the N+W experiment (Figure 2), and the former were more consistent with previous production rates of M. edulis Heisterkamp et al. 2010Heisterkamp et al. , 2013. We hypothesize that the difference between these experiments is due to the filtered vs. unfiltered water used, respectively. N 2 O production in the epifauna (with no N addition, Figure 7) treatment was ~3 times lower than immediate exposure control results for M. edulis in the NW experiment ( Figure 2). The N 2 O production from the epifauna + nitrogen addition treatment was also ~3 times lower than M. edulis N addition treatment in the N+W experiment.
Filtration of the water column likely removed N 2 O producing bacteria that M. edulis would normally ingest from the water column, therefore lowering the N 2 O production rate associated with M. edulis. Stief (2013) suggests microbes ingested by bivalves may not be digested and survive in the gut, remaining metabolically active to facilitate production of N 2 O. M. edulis is an efficient filter feeder that ingests large amounts of bacteria (McHenery and Birkbeck 1985); therefore, the rates described associated with M. edulis in the epifauna removal experiment (Figure 7) are likely produced from N 2 O producing microbes already present within the gut of the organism before the start of this experiment as well as potential microbial biofilms on their shells.

CONCLUSION
Our study illustrates that N loading increases N 2 O emissions from bivalve shellfish. We show little evidence that warming or macro-epifauna significantly contribute to N 2 O production in this short term lab study, however future longer duration experiments are needed to better simulate the natural environment. Potential N 2 O emissions from bivalves differed between species and were not biomass dependent as has been shown for other invertebrate taxa. M. edulis produced the highest rates of N 2 O rather than Mercenaria mercenaria, which we believe to be a combination of high NH 4 + production and the induction of hypoxic conditions due to potentially poor health of one source population. Notably, N 2 O emissions increased via exposure to N loading and will possibly further increase in natural systems where they can interact with marine sediments and increase benthic N cycling. Since marine bivalves are found in high abundances in coastal systems where the combination of eutrophic, warming and low DO often coincide, their potential contribution to N 2 O emissions from benthic systems may be higher than previously understood and warrants further investigation.

Bay, Puerto Rico
Carbon dioxide (CO 2 ), nitrous oxide (N 2 O) and methane (CH 4 ) are potent greenhouse gases (GHGs) that are readily increasing in the atmosphere due to human activities and are significant drivers of climate change (Forster et al. 2007). Mangrove wetlands are important ecosystems for potentially mitigating climate change as they have significant carbon stocks (Donato et al. 2011). However, with large inputs of human-derived pollutants such as nitrogen (N), mangroves may shift from being sinks of GHGs to sources (Corredor et al. 1999;Muñoz-Hincapié et al. 2002;Chen et al. 2012). The purpose of this experiment was to compare GHG emissions from Rhizophora dominant mangrove sediments in Jobos Bay, Puerto Rico at a pristine site and an anthropogenically influenced site. We hypothesized higher GHGs emissions from mangrove sediments in the anthropogenically influenced site compared to the pristine. We also wanted to determine if pulses of nitrogen (N) would stimulate metabolism by N-cycling bacteria and produce significant amounts of N 2 O.

METHODS
Greenhouse gas emissions were measured in the red mangrove zone in Jobos Bay, Puerto Rico at a pristine site (Cayes) and an anthropogenically influenced site (Mar Negro). Jobos Bay is on the south coast of the island and dominated by agriculture. Mar Negro is directly influenced by human pollutants as it is adjacent to a residential area that commonly disposes of their untreated human waste directly on the mangrove watershed (Bowen and Valiela 2008). The pristine site is one of fifteen small mangrove fringe islands that have no direct connection to the mainland and therefore little anthropogenic influence. These measurements were collected during the wet season, though there was a drought occurring in this region of Puerto Rico during this time. Each site had 6 replicates that were ~1m apart from each other.
Fluxes were measured using a cavity ring-down spectrophotometer (Picarro G2508) during low tide using 100 foot tubing that was used to connect the analyzer to a static flux chamber. The chamber (volume approximately 10 liters) was equipped with an inflow port, outflow port, fan to homogenize the air and pigtail to equalize pressure.
The chamber was placed on the mangrove sediment on preinstalled (2 days) aluminum collars for 8 minutes. GHG fluxes were calculated using the Ideal Gas Law (PV = nRT) using field-measured air temperatures (HOBO, Inc) and atmospheric pressure. If there was no change in gas concentration over time during the measurement period, fluxes were reported as negligible.
Salinity, pH, soil moisture, and redox potential were measured to determine correlations between edaphic parameters and gas fluxes. Soil salinity was measured from the top 3cm by pressing the soil against paper filters within a plastic syringe (15mL). The extracted water was then measured for salinity on a handheld portable refractometer. To measure pH, a soil slurry was made by collecting approximately 10 mL of surface soil using a cut off syringe and adding it to a 50mL falcon tube with 15mL of deionized water. Soil pH was then measured using a pH meter (Thermo Scientific Orion Star A326 pH/Dissolved Oxygen Portable Multiparameter Meter) after the slurry was shaken for approximately 10 seconds. Soil moisture content was measured using a volumetric water content sensor (Decagon Devices, Pullman, WA) inserted 5 cm into the soil and soil oxidation reduction (redox) potential was measured using a pH/ORP meter (Metler Toledo, Greifensee, Switzerland). 10 cm deep porewater (5 mL) was obtained using rhizons (Rhizosphere), filtered with a 0.45 µm filter and frozen in order to measure dissolved inorganic nitrate (DIN).
In order to examine if nutrient pulses stimulate N 2 O fluxes, we also performed a N enrichment experiment in Mar Negro. N treatments received 500 mL of 300 µM potassium nitrate made with unfiltered site water which had salinities of 31ppt prior to flux measurement. Control plots received 500 mL of site water only. Each treatment had three replicates. N manipulations and seawater controls were applied by slowing pouring the amount over the designated plot area. Fluxes were measured within one minute of experimental manipulations, 1.5 hours later, and 72 hours (3 days) later using methods as described above. Due to variability in raw data, N 2 O fluxes were calculated using twenty second blocked averages. Edaphic parameters and porewater were measured as described above immediately after gas flux measurements.

STATISTICAL ANALYSIS
A paired t-test was used to determine statistical significant differences for each greenhouse gas and edaphic parameters between sites. A two-factor ANOVA was used to determine if there was a statistical significant influence on N 2 O fluxes by N pulse and time (1 factor=treatment, 2 factor=time).
Unequal variance and normality were tested using the Bartlett test and Shapiro-Wilk test, respectively, in order to ensure assumptions were met for each statistic.
Correlations were used to test relationships between edaphic parameters and greenhouse gas fluxes.
All statistical analyses were performed with JMP 10.0 software and significant levels of a=0.05 were used.

RESULTS AND DISCUSSION
Our study found that CH 4 and CO 2 fluxes from mangrove sediments were positive at both sites, while N 2 O was negligible. CO 2 fluxes were significantly higher (t 1,10 =5.07, p=0.05) in the anthropogenically influenced site compared to the pristine ( Figure 1). A similar pattern was found for CH 4 (t 1,10 =18.42, p<0.01) such that CH 4 fluxes from the anthropogenic site was ~11x higher than the pristine.
Edaphic results are listed in Table 1. There were no significant relationships between edaphic parameters and GHG fluxes, except for a weak positive relationship between soil moisture and CO 2 fluxes in the pristine site (R 2 =0.67, p=0.05). Salinity was significantly higher in the pristine site compared to the anthropogenic site (t 1,10 =2.18, p=0.05), likely due to its closer proximity to the open ocean. All other edaphic parameters were similar between sites. Porewater ammonium concentrations were higher in the anthropogenically influenced site and were positively related to CH 4 fluxes (R 2 =0.50, p=0.03), suggesting anthropogenic nitrogen inputs may potentially enhance microbial metabolism and enhance GHG emissions from these mangroves.
Our study found no effects of N additions on CH 4 or CO 2 at 0, 1.5 hours or 72 hours. N pulses did significantly increase N 2 O emissions compared to controls at 1.5 and 72 hours (F 3,7 =4.6, p=0.01), though the time or the combination of time and treatment did not have a significant affect (Figure 2). N 2 O flux was negligible at 0 hours (not shown). Increase in N 2 O suggests that microbial metabolism was enhanced with the N pulse. Since redox levels were negative (Table 1), denitrification was the likely source of N 2 O emissions as denitrification occurs in anoxic conditions (Zumft 1997).
These fluxes are toward the lower end of the GHG flux range compared to other anthropogenically impacted mangroves ecosystems (Chauhan et al. 2008;Chen et al. 2012;Call et al. 2015). This may be due to differences in hydrology of the sites (diurnal tide cycle) or C:N ratio of the sediments. Altering oxic to anoxic conditions (ie: semi-diurnal tides) can result in higher denitrification rates from sediments than those under relatively more continuous anoxic conditions (diurnal tides). Higher sediment C:N typically yield higher respiration rates (Rivera-Monroy and Twilley 1996), thus further analysis is needed to determine the potential influence of C:N on GHG fluxes. Further analysis is also needed during different seasonal and temporal scales to quantify the extent of their potential affect on carbon sequestration rates.

Greenhouse gas fluxes associated with salt marsh dieback
Salt marsh habitats hold many ecosystem services such as storm protection, nursery habitats and carbon sequestration (Chmura et al. 2003). Unfortunately, these vital ecosystems are experiencing significant dieback events worldwide due to direct and indirect anthropogenic disturbances, particularly in regards to climate change (Alber et al. 2008). Dieback in New England salt marshes, for example, have been attributed to increases in herbivory by crabs whose populations have increased due to human-induced decrease in predatory pressure Holdredge et al. 2009;Altieri et al. 2012). Another major concern is rising sea level, which causes prolonged inundation of salt marsh grasses, eventually leading to plant loss and shifts in plant community structure (Craft et al. 2009). Dieback events may significantly alter the ecosystem services of a saltmarsh, including greatly reducing ability to sequester carbon and become sources rather than sinks of greenhouse gases (GHGs).
The purpose of our experiment was to examine the combined effects of crab behavior and sea level rise on the greenhouse gas fluxes of the dominant high marsh plant, Spartina patens in a mesocosm study. We focused on the herbivorous crab Sesarma and also on the burrowing decapod, Uca puglator, which may also have a significant impact on the salt marsh soils though have been largely understudied. We hypothesized that Sesarma would significantly increase GHG emissions (carbon dioxide and methane) due to its herbivory activity while Uca would increase the carbon sequestration rates as it has been positively related to salt marsh plant health.
We also examined the GHG fluxes of a New England salt marsh that is currently undergoing a dieback event compared to its intact area in situ. We hypothesized that the salt marsh dieback zone would have significantly higher GHG emissions compared to the intact zone (control), due to the reduction of photosynthesis by plants and subsequent increase in microbial respiration in the soils.

Our mesocosm experiment was conducted at the Environmental Protection
Agency greenhouse facility in Narragansett, RI using six flow-through mesocosm tanks (900L) under a semi-diurnal tidal regime (Oczkowski et al. 2015 GHG measurements were taken using a cavity ringdown spectrometer (Picarro G2508) in three randomly selected tanks on two dates (2 and 31 July 2014), which were approximately in the beginning and end of the fall season. Nylon tubing (50 feet) was used to connect the analyzer to a static flux chamber for GHG measurements.
GHG measurements and edaphic parameters (pH, redox, soil moisture, salinity) were measured as stated in Appendix I. The only exception is that a larger chamber (~13 L) was used for this study.
GHG measurements were made in a dieback and an intact salt marsh zone in Passeonkquis Cove, Rhode Island (41°44'45.0"N, 71°23'25.5"W) on during early (October) and late (December) fall. The dieback region was characterized by a visual reduction of Spartina alterniflora aboveground biomass and was largely comprised of organic soils. The intact zone was comprised of visually healthy and continuous low marsh Spartina alterniflora. Each zone had four replicates at least 1m apart. GHG measurements were taken as stated above for the mesocosm study, with the exception that the intact zone measures were taken on preinstalled collars since plants and roots would normally obstruct a gas tight seal. Edaphic parameters were measured as stated above with the exception that soil temperature was also measured for this experiment by inserting a handheld soil thermometer 5cm into the soil.

STATISTICAL ANALYSIS
A two-factor ANOVA was used to determine statistical significant differences in greenhouse gases and in edaphic parameters between mesocosm experimental treatments (1 factor= crab, 2 factor=inundation). A two-factor ANOVA was also used to determine statistical significance in in situ salt marsh study (1 factor= treatment, 2 factor=time of data collection). Unequal variance and normality were tested using the Bartlett test and Shapiro-Wilk test, respectively, in order to ensure assumptions were met for the two-factor ANOVA.
Correlations were used to test relationships between edaphic parameters and greenhouse gas fluxes.
All statistical analyses were performed with JMP 10.0 software and significant levels of a=0.05 were used.

RESULTS
Positive carbon dioxide (CO 2 ) fluxes were significantly affected by inundation on 2 July 2014, but were not affected by crab presence or the combination ( Figure 3A, Table 2). The low inundation treatments had higher CO 2 emissions compared to the control ( Figure 3A). Towards the end of the experiment on 31 July 2014, inundation did not individually impact CO 2 fluxes, however crab and the combination did significantly affect CO 2 ( Figure 3B, Table 2). The crab control (no crabs) treatment under low inundation showed similar CO 2 uptake compared to Uca low inundation treatments, while Sesarma treatments showed production of CO 2 . Our hypothesis was partly confirmed such that CO 2 emissions between dates in the Sesarma treatments increased by 31 July 2014. We attribute this increase to the total consumption of S.
patens by Sesarma and subsequent respiration by the remaining plant roots and soils.
Uca did not have a significant impact in the low inundation treatments compared to the controls, however it did significantly reduce the CO 2 uptake in the high inundation treatment, which was unexpected. Methane (CH 4 ) fluxes were negligible on 2 July 2014 (not shown) and were variable and not significantly affected by treatments on 31 July 2014 ( Figure 4, Table 2).
Soil pH was significantly higher in the high elevation treatments on 2 July 2014, but crabs did not have a significant affect (Table 3- (Table 2-3). Average soil salinities ranged between 28-55 ppt but did not significantly differ between treatments.
There were no strong significant relationships between edaphic parameters and GHG fluxes.
There was a surprising significantly lower emission of CO 2 in the dieback zone compared to the control, which does not support our hypothesis ( Figure 6A, Table 5).
There was a significant effect of date such that there was consumption of CO 2 by the late fall compared to the early fall collection dates ( Figure 6A). There was no significant difference in CH 4 fluxes between treatments or date of collection though there was also a trend of lower fluxes for the late fall sample date ( Figure 6B, Table   5). N 2 O was negligible in all treatments during all sample dates.
Redox potential, soil moisture and temperature were significantly different between treatments, date, and the combination of treatment x date (Table 5-6). There was a significant negative relationship between moisture and CO 2 (R 2 =0.45, p=0.01) driven by lower moisture levels in the control plots, likely due to greater uptake by the salt marsh grasses compared to the dieback zone. There was a positive relationship between pH and CH 4 (R 2 =0.50, p<0.01) driven by higher pH levels in the controls plots.
These data were collected during the fall season where microbial activity tend to be lower due to lower temperatures (Table 6), thus the lower CO 2 in the dieback region is likely due to lower microbial activity compared to high respiration by the plants, which is common for the fall months. Future research should focus on the growing season, which is the time salt marshes grasses and associated microbial communities are more active.  Table 3. Average edaphic parameter results for 2 and 31 July 2014 for crab and sea level rise study ± standard error.