Nitrogen and oxygen availabilities control water column nitrous oxide production during seasonal anoxia in the Chesapeake Bay

Nitrous oxide (N₂O) is a greenhouse gas and an ozone depletion agent. Estuaries that are subject to seasonal anoxia are generally regarded as N₂O sources. However, insufficient understanding of the environmental controls on N₂O production results in large uncertainty about the estuarine contribution to the global N₂O budget. Incubation experiments with nitrogen stable isotope tracer were used to investigate the geochemical factors controlling N₂O production from denitrification in the Chesapeake Bay, the largest estuary in North America. The highest potential rates of water column N₂O production via denitrification (7.5±1.2 nmol-N L⁻¹ h⁻¹) were detected during summer anoxia, during which oxidized nitrogen species (nitrate and nitrite) were absent from the water column. At the top of the anoxic layer, N₂O production from denitrification was stimulated by addition of nitrate and nitrite. The relative contribution of nitrate and nitrite to N₂O production was positively correlated with the ratio of nitrate to nitrite concentrations. Increased oxygen availability, up to 7 µmol L⁻¹ oxygen, inhibited both N₂O production and the reduction of nitrate to nitrite. In spring, high oxygen and low abundance of denitrifying microbes resulted in undetectable N₂O production from denitrification. Thus, decreasing the nitrogen input into the Chesapeake Bay has two potential impacts on the N₂O production: a lower availability of nitrogen substrates may mitigate short-term N₂O emissions during summer anoxia; and, in the long-run (timescale of years), eutrophication will be alleviated and subsequent reoxygenation of the bay will further inhibit N₂O production.

1 Introduction

Nitrous oxide (N₂O) is a strong greenhouse gas with 298-fold higher global warming potential per mole than that of carbon dioxide. N₂O is also a catalyst of ozone depletion in the stratosphere. Since the Industrial Revolution, the N₂O atmospheric concentration has been increasing at an unprecedented rate, and the current concentration is the highest in the last 800 000 years of Earth's history (Schilt et al., 2010). The contribution of N₂O emissions to global warming and ozone depletion will increase because N₂O is not as strictly regulated as are CO₂ and halocarbon compounds. With the successful mitigation of halocarbon compounds accomplished by the Montreal Protocol, N₂O is likely to be the single most important anthropogenically emitted ozone-depleting agent in the 21st century (Ravishankara et al., 2009).

Microbial processes are responsible for the majority of N₂O production, both in natural and anthropogenically impacted environments. These pathways include oxidative and reductive processes occurring at the full range of environmental oxygen concentrations. In the presence of oxygen, N₂O can be produced as a byproduct during autotrophic aerobic ammonium (${\mathrm{NH}}_{\mathrm{4}}^{+}$) oxidation to nitrite (${\mathrm{NO}}_{\mathrm{2}}^{-}$) by bacteria (Arp and Stein, 2003) and archaea (Santoro et al., 2011). The production of N₂O can also occur via ${\mathrm{NO}}_{\mathrm{2}}^{-}$ reduction by nitrifying organisms, termed nitrifier denitrification. This process was demonstrated in cultures (Poth and Focht, 1985; Frame and Casciotti, 2010) and in the water column of the subtropical North Pacific Ocean (Wilson et al., 2014). Under low-oxygen and anoxic conditions, denitrifying bacteria produce N₂O via enzyme-mediated heterotrophic denitrification, which consists of the stepwise reduction of nitrate (${\mathrm{NO}}_{\mathrm{3}}^{-}$), ${\mathrm{NO}}_{\mathrm{2}}^{-}$ and nitric oxide (NO), with organic matter as the electron donor. The nirS gene that encodes the genetic material for nitrite reductase (the enzyme mediating ${\mathrm{NO}}_{\mathrm{2}}^{-}$ reduction to NO) is often used as a proxy for abundance and diversity of denitrifying bacteria and is the gene in the denitrification sequence that is most reliably associated with a complete denitrification pathway (Graf et al., 2014). N₂O is not produced via anaerobic ammonium oxidation (anammox), another important nitrogen removal process in the natural environment (Kartal et al., 2011).

The increase in atmospheric N₂O is attributed to intensification of human activities (e.g., fossil fuel combustion, fertilizer application, human and animal waste disposal), which alter the microbial nitrogen cycle in the biosphere. Increased nitrogen supply from fertilizer and atmospheric deposition causes increased N₂O emission not only from agricultural land, but also in rivers, streams and coastal waters (Ciais et al., 2013; Thompson et al., 2014). Among these aquatic environments, intense N₂O efflux originates from estuaries and associated river networks, which occupy 0.3 % of global waters (Dürr et al., 2011) but could contribute up to 10 % of anthropogenic fluxes (Seitzinger and Kroeze, 1998; Ciais et al., 2013). Being the largest estuary in North America, the Chesapeake Bay and its tributaries have experienced eutrophication and expansion of summertime anoxia due to increased population, expansion of industrialization and land use changes since the 18th century (Cooper and Brush, 1993; Boesch et al., 2001). The Chesapeake tributary is a source of N₂O (indicated by surface N₂O oversaturation) in the summertime between June and September (Elkins et al., 1978; Kaplan et al., 1978; McElroy et al., 1978). The summertime water column is characterized by strong oxygen gradients (equilibrium with atmosphere at the surface and complete anoxia below ∼10 m), depletion of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ and ${\mathrm{NO}}_{\mathrm{2}}^{-}$, and accumulation of ${\mathrm{NH}}_{\mathrm{4}}^{+}$ in the deep water (Lee et al., 2015b). Increased microbial activities driving carbon assimilation and respiration have been demonstrated in the vicinity of the oxic–anoxic interface in the water column (Lee et al., 2015a). However, the N₂O production pathway and the associated environmental controlling factors have not been investigated in the Chesapeake Bay.

Here we report a pilot study using nitrogen stable isotope (¹⁵N) incubation experiments to quantify N₂O production rates and their dependence on the availabilities of oxygen, ${\mathrm{NO}}_{\mathrm{3}}^{-}$ and ${\mathrm{NO}}_{\mathrm{2}}^{-}$ in the Chesapeake Bay. Because seasonal anoxia occurs at the study site in the central region of the Chesapeake Bay, reductive pathways of N₂O production (i.e., reduction of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ and ${\mathrm{NO}}_{\mathrm{2}}^{-}$) are the main focus. Further understanding of the environmental controls on N₂O production in estuaries will facilitate the design of effective environmental engineering projects to mitigate N₂O emission.

2 Methods

2.1 Sample acquisition and processing

Sampling and incubation experiments were carried out on 19 July 2016, 17 November 2016 and 3 May 2017, corresponding to typical conditions of summer, autumn and spring, respectively. Samples were collected at 38.55^∘ N, 76.43^∘ W (bottom depth 26.5 m) close to the mouth of the Choptank River in the central region of the Chesapeake Bay. Conductivity–temperature–depth and dissolved oxygen ([O₂]) were measured with a YSI sonde package (Model 600XLM with a 650 MDS display logger) equipped with a diaphragm pump which was deployed for water sampling. The oxygen sensor had a detection limit of ∼5 µmol L⁻¹. Samples for ${\mathrm{NO}}_{\mathrm{2}}^{-}$ and ${\mathrm{NO}}_{\mathrm{3}}^{-}$ concentration measurements were filtered (0.22 µm pore size, Sterivex-GP, EMD Millipore) and frozen at −80 ^∘C until analysis. Discrete samples for N₂O concentration were collected directly from the pump outlet into the bottom of acid-washed, 60 mL glass serum bottles (catalog no. 223745, Wheaton, Millville, NJ, USA). Bottles were sealed with butyl rubber stoppers (catalog no. W224100-202, Wheaton, Millville, NJ, USA) and aluminium rings while submerged under water pumped from depth to avoid atmospheric N₂O and oxygen contamination. Samples for characterizing the N₂O concentration profile were preserved immediately after filling by injecting 0.1 mL saturated HgCl₂. Samples for N₂O incubation experiments (Sect. 2.2) were acquired from 12, 17 and 19.5 m during July 2016, November 2016 and May 2017, respectively; sealed the same way as described above for discrete N₂O concentration samples; and stored in the dark at 4 ^∘C without adding HgCl₂. Samples for denitrifying nirS gene abundance were collected at 14, 17 and 19.5 m by filtering 600–2000 mL of water through a 0.22 µm filter (Sterivex-GP, EMD Millipore) and frozen at −80 ^∘C until DNA extraction and analysis.

Table 1 Parameters for control, nitrogen manipulation and oxygen manipulation incubation experiments in July 2016, November 2016 and May 2017 sampling. In May 2017, only the control experiment was conducted. The unit “µmol L⁻¹” is represented by “µM”. Bold columns highlight the concentrations for ¹⁵N tracers. In situ nitrate and nitrite concentrations in July 2016 were <0.02 µmol L⁻¹; in November 2016 the concentrations were 5.0 and 0.4 µmol L⁻¹, respectively; in May 2017 the concentrations were 6.3 and 0.4 µmol L⁻¹, respectively.

Samples for total dissolved inorganic carbon ($\mathrm{DIC}=\left[{\mathrm{H}}_{\mathrm{2}}{\mathrm{CO}}_{\mathrm{3}}\right]+\left[{\mathrm{HCO}}_{\mathrm{3}}^{-}\right]+\left[{\mathrm{CO}}_{\mathrm{3}}^{\mathrm{2}-}\right]$) and community respiration rates were collected only in July 2016. The DIC samples were preserved with mercuric chloride (HgCl₂) for initial conditions, while biochemical oxygen demand (BOD) bottles were incubated in a temperature-controlled environmental chamber (±1 ^∘C of in situ water temperatures). After 24 h, samples were siphoned from the vials, preserved with HgCl₂, and respiration rates were determined as the difference in DIC between initial and final samples divided by 24 h (Lee et al., 2015b).

2.2 ¹⁵N incubation experiments for N₂O production

Within 3 h of sampling, incubation experiments were initiated at the Horn Point Laboratory, Cambridge, Maryland. Samples were divided into three sets for control, nitrogen manipulation and oxygen manipulation experiments.

Control experiment. The control experiment was conducted in July 2016, November 2016 and May 2017. A small (3 mL) headspace was created in the serum bottles, which were subsequently flushed with helium for 10 min to minimize oxygen contamination from sampling and transportation. Two suites of ¹⁵N tracer solutions (${}^{\mathrm{15}}{\mathrm{NO}}_{\mathrm{2}}^{-}$ plus ${}^{\mathrm{14}}{\mathrm{NO}}_{\mathrm{3}}^{-}$, ${}^{\mathrm{15}}{\mathrm{NO}}_{\mathrm{3}}^{-}$ plus ${}^{\mathrm{14}}{\mathrm{NO}}_{\mathrm{2}}^{-}$, 0.1 mL) were injected to achieve final concentrations of 5 µmol L⁻¹ ${\mathrm{NO}}_{\mathrm{2}}^{-}$ and ${\mathrm{NO}}_{\mathrm{3}}^{-}$ (see conditions for experiment 1-A and 1-B, 4-A and 4-B, and 6-A and 6-B in Table 1). Tracer solutions were made from deionized water and were flushed with helium prior to addition to incubation experiments. In order to have enough mass to detect N₂O production, ∼1.2 nmol of natural abundance N₂O was injected to each bottle, reaching a concentration of ∼20 nmol L⁻¹ in the water phase (calculated equilibrium concentration according to Weiss and Price, 1980, with 3 mL headspace and 57 mL water). Initial conditions (one bottle for each time course) were sampled within 30 min of tracer addition by injecting 0.1 mL saturated HgCl₂. Incubations lasted ∼2 h at in situ temperature (±0.5 ^∘C), during which duplicate bottles were preserved with a saturated HgCl₂ solution every 40 to 60 min, totalling seven bottles over four time points, including the initial for a time course analysis.

Dissolved inorganic nitrogen (DIN) manipulation. The DIN manipulation experiment was conducted only in July 2016 because ${\mathrm{NO}}_{\mathrm{2}}^{-}$ and ${\mathrm{NO}}_{\mathrm{3}}^{-}$ were absent from the water column (see Sect. 3.1). A 3 mL headspace was created before flushing with helium for 10 min to establish anoxic condition. Then, ∼1.2 nmol N₂O was injected to reach a concentration of ∼20 nmol L⁻¹ in the water phase. Two suites of ¹⁵N tracer solutions (${}^{\mathrm{15}}{\mathrm{NO}}_{\mathrm{2}}^{-}$ plus ${}^{\mathrm{14}}{\mathrm{NO}}_{\mathrm{3}}^{-}$, ${}^{\mathrm{15}}{\mathrm{NO}}_{\mathrm{3}}^{-}$ plus ${}^{\mathrm{14}}{\mathrm{NO}}_{\mathrm{2}}^{-}$, 0.1 mL of total volume of tracer addition) were injected to designated bottles to achieve ratios of ${\mathrm{NO}}_{\mathrm{2}}^{-}:{\mathrm{NO}}_{\mathrm{3}}^{-}\approx \mathrm{1}:\mathrm{10}$, 1:3, 3:1 and 10:1, with ¹⁵N fraction labeled between 0.016 and 0.16 (Table 1, experiment 2-A to 2-H). This allowed simultaneous detection of N₂O production from ${\mathrm{NO}}_{\mathrm{2}}^{-}$ and ${\mathrm{NO}}_{\mathrm{3}}^{-}$ at different ratios of ${\mathrm{NO}}_{\mathrm{2}}^{-}$ to ${\mathrm{NO}}_{\mathrm{3}}^{-}$ concentration. Incubations lasted ∼2 h with the same sampling strategy as the control experiment.

Oxygen manipulation. The oxygen manipulation experiment was conducted in July 2016 and November 2016. Headspace (3–8 mL) was created before flushing with helium for 10 min. Oxygen-saturated site water was made by air equilibration at in situ temperature. To achieve different oxygen levels, 0.2, 0.5, 1.0, 2.0 or 5.0 mL of oxygen-saturated site water was injected. With a final volume of ∼3 mL of headspace during the course of the incubation, the oxygen concentrations in the water phase were 0.3 to 6.4 µmol L⁻¹ in July 2016 (Table 1, experiment 3-A–3-J) and were 0.2 to 7.3 µmol L⁻¹ in November 2016 (Table 1, experiment 5-A–5-J) after the calculated equilibration between headspace and water (Garcia and Gordon, 1992). In addition, an optical sensor was used to measure oxygen concentrations directly in a parallel experimental setup and the agreement between calculated target concentration and measured concentration was excellent (data not shown). After oxygen adjustment, ∼1.2 nmol N₂O was injected into each bottle, and two suites of ¹⁵N tracer solutions (${}^{\mathrm{15}}{\mathrm{NO}}_{\mathrm{2}}^{-}$ plus ${}^{\mathrm{14}}{\mathrm{NO}}_{\mathrm{3}}^{-}$, ${}^{\mathrm{15}}{\mathrm{NO}}_{\mathrm{3}}^{-}$ plus ${}^{\mathrm{14}}{\mathrm{NO}}_{\mathrm{2}}^{-}$, 0.1 mL) were injected to achieve a final concentration of 5 µmol L⁻¹ ${\mathrm{NO}}_{\mathrm{2}}^{-}$ and ${\mathrm{NO}}_{\mathrm{3}}^{-}$. The ¹⁵N fraction for ${\mathrm{NO}}_{\mathrm{2}}^{-}$ or ${\mathrm{NO}}_{\mathrm{3}}^{-}$ during the incubation experiments is shown in Table 1. Incubations lasted ∼2 h with the same sampling strategy as the control experiment.

2.3 Analytical procedures

For water column nutrients, dissolved ${\mathrm{NO}}_{\mathrm{2}}^{-}$ was measured using a colorimetric method (Hansen and Koroleff, 2007) and ${\mathrm{NO}}_{\mathrm{3}}^{-}+{\mathrm{NO}}_{\mathrm{2}}^{-}$ was measured using a hot (90 ^∘C) acidified vanadium(III) reduction column coupled to a chemiluminescence NO∕NO_x analyzer (Teledyne API, San Diego, CA, USA) (Garside, 1982; Braman and Hendrix, 1989). DIC was measured with an automated infrared analyzer (Apollo SciTech, Newark, DE, USA) as previously reported (Lee et al., 2015b). Preserved N₂O samples were stored in the dark at room temperature (∼22 ^∘C) for less than 3 weeks before analysis. Dissolved N₂O was extracted by flushing with helium for 40 min at a rate of 37 mL min⁻¹ (extraction efficiency 99±2 %) and subsequently cryo-trapped by liquid nitrogen and isolated from interfering compounds (H₂O, CO₂) by gas chromatography (Weigand et al., 2016). Pulses of purified N₂O were injected into an isotope ratio mass spectrometer (Delta V^Plus, Thermo Fisher Scientific, Waltham, MA) for mass ($m/z=\mathrm{44}$, 45, 46) and isotope ratio ($m_{\mathrm{1}}/m_{\mathrm{2}}=\mathrm{45}/\mathrm{44}$, 46∕44) measurements. The amount of N₂O was calibrated with standard N₂O vials, which were made by injecting 1, 2 or 5 nmol N₂O-N into 20 mL glass vials (catalog no. C4020-25, Thermo Fisher Scientific, Waltham, MA).

After N₂O analysis, samples incubated with ${}^{\mathrm{15}}{\mathrm{NO}}_{\mathrm{3}}^{-}$ were also assayed for ${}^{\mathrm{15}}{\mathrm{NO}}_{\mathrm{2}}^{-}$ to determine rates of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ reduction. Two milliliters of each sample were transferred from the 60 mL serum bottle to a 20 mL glass vial and then flushed with helium for 10 min. Dissolved ${}^{\mathrm{15}}{\mathrm{NO}}_{\mathrm{2}}^{-}$ was converted to N₂O using the acetic-acid-treated sodium azide solution for quantitative conversion (McIlvin and Altabet, 2005). Resulting N₂O was measured for nitrogen isotope ratio (¹⁵N∕¹⁴N) so as to determine the ¹⁵N enrichment of ${\mathrm{NO}}_{\mathrm{2}}^{-}$.

For the analysis of nirS gene abundance, DNA extraction and qPCR for the nirS gene using SYBR Green were performed as previously described (Jayakumar et al., 2009, 2013). Extracted DNA was quantified using PicoGreen fluorescence (Molecular Probes, Eugene, OR) prior to the qPCR assay. Samples for qPCR were run in triplicates including a no-template control, a no-primer control and five different dilutions of a nirS standard. Threshold cycle (Ct) values were obtained using automatic analysis settings of the quantitative PCR and further used to calculate the gene copy numbers as described in Jayakumar et al. (2013).

Figure 1 Depth profiles on three sampling dates – 19 July 2016 (filled square), 17 November 2016 (cross), 3 May 2017 (grey circle) – of (a) salinity, (b) temperature, (c) oxygen, (d) nitrous oxide, (e) nitrate and (f) nitrite. Analysis of nirS gene abundance (g) was only conducted at one depth, at which incubations were also performed, during each trip.

2.4 Data analysis

N₂O concentration was calculated from the amount of N₂O detected by mass spectrometry divided by the volume of water in the serum bottles. N₂O production (R) was calculated from the progressive increase in ⁴⁵N₂O and ⁴⁶N₂O concentrations in each serum bottle over the time course experiments.

$$\begin{array}{}\text{(1)}& {\displaystyle }R={\displaystyle \frac{\mathrm{1}}{F}}\times \left({\displaystyle \frac{{\mathrm{d}}^{\mathrm{45}}{\mathrm{N}}_{\mathrm{2}}\mathrm{O}}{\mathrm{d}t}}+\mathrm{2}\times {\displaystyle \frac{{\mathrm{d}}^{\mathrm{46}}{\mathrm{N}}_{\mathrm{2}}\mathrm{O}}{\mathrm{d}t}}\right),\end{array}$$

where d⁴⁵N₂O∕dt and d⁴⁶N₂O∕dt represent the production rates (nmol-N L⁻¹ h⁻¹) of mass 45 and 46 N₂O during incubation. F represents the ¹⁵N fraction in the initial substrate (${\mathrm{NO}}_{\mathrm{2}}^{-}$ or ${\mathrm{NO}}_{\mathrm{3}}^{-}$). Rates were considered significant based on the linear regression of the time course data having p<0.05 (n=7, Student's t test). The detection limit for N₂O production was 0.002 nmol-N L⁻¹ h⁻¹. The ¹⁵N incubation experiments can identify the pathway but cannot distinguish the relative contributions of two or more functioning microbial groups to a single N₂O production pathway (i.e., N₂O production via ${\mathrm{NO}}_{\mathrm{2}}^{-}$ reduction by nitrifier denitrification and/or heterotrophic denitrification).

The rate of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ reduction to ${\mathrm{NO}}_{\mathrm{2}}^{-}$ was calculated as

$$\begin{array}{}\text{(2)}& {\displaystyle }{\mathrm{NO}}_{\mathrm{2}}^{-}\mathrm{production}=({\mathrm{d}}^{\mathrm{15}}{\mathrm{NO}}_{\mathrm{2}}^{-}/\mathrm{d}t)/F,\end{array}$$

where ${\mathrm{d}}^{\mathrm{15}}{\mathrm{NO}}_{\mathrm{2}}^{-}/\mathrm{d}t$ represents the production rate of ¹⁵${\mathrm{NO}}_{\mathrm{2}}^{-}$ (nmol-N L⁻¹ h⁻¹). F represents initial ¹⁵N enrichment of substrate ${\mathrm{NO}}_{\mathrm{3}}^{-}$. Rates were considered significant based on linear regression of the time course data having p<0.05 (Student's t test). The detection limit for ${\mathrm{NO}}_{\mathrm{2}}^{-}$ production was 0.05 nmol-N L⁻¹ h⁻¹.

3 Results and discussion

3.1 Water column features

The physical and chemical properties of the water column in the central Chesapeake Bay showed seasonal variation (Fig. 1). Temperature and salinity differed among the three seasons but were essentially constant in the top 7 m of the water column on the three sampling dates. In July, the water column was stratified because of lower salinity (∼16 PSU) and higher temperature (∼28.5 ^∘C) in the top ∼10 m, resulting in a pronounced halocline and thermocline (Fig. 1a and b). Less pronounced stratification in May and November was due to a weaker temperature difference between the top 10 m and below. The July oxygen profile showed a significant concentration decrease between 3 and 10 m (Fig. 1c), with a sharp oxycline (∼30 µmol L⁻¹ m⁻¹). Below 10 m, the oxygen concentration was below detection of the sensor (∼5 µmol L⁻¹) and was likely anoxic. The water samples were free of any hydrogen sulfide odor, so we conclude that sulfide was either absent or was present at a very low level (<1 µmol L⁻¹). No anoxic layer was observed in May and November (Fig. 1c), and previous studies showed that the water column of the Chesapeake Bay was reoxygenated following summertime anoxia during winter and spring (Lee et al., 2015a).

The surface N₂O saturation concentrations in July, November and May were 6.6, 10.4 and 12.0 nmol L⁻¹, respectively. In July, N₂O concentration was close to air-saturation level (6.6 nmol L⁻¹) at the surface layer. In the low oxygen layer (below 12 m), N₂O was apparently undersaturated (2.0–3.7 nmol L⁻¹, 20 %–50 % air saturation, Fig. 1d). In November, the surface N₂O concentration was slightly oversaturated (11.3 nmol L⁻¹, 108 % air saturation). N₂O concentrations at depth were oversaturated; the concentrations varied between 11.0 and 11.5 nmol L⁻¹, corresponding to 109 %–115 % air saturation. In May, both the surface and water column N₂O concentrations were air undersaturated; the surface concentration was 9.1 nmol L⁻¹, 76 % air saturation; concentrations between 8 and 17 m ranged from 9.4 to 11.0 nmol L⁻¹, corresponding to 82 %–97 % air saturation. As the surface and water column N₂O saturation levels vary greatly between seasons, the assessment of the N₂O dynamics of the Chesapeake Bay requires expanding the temporal and spatial coverage of the field sampling. In the following, we focus on N₂O production and its environmental controlling factors.

The concentrations of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ and ${\mathrm{NO}}_{\mathrm{2}}^{-}$ (Fig. 1e and f) in July were below 0.02 µmol L⁻¹ within the sampling depth interval (top 17 m of water column). Measurable levels of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ and ${\mathrm{NO}}_{\mathrm{2}}^{-}$ species were found in May and November. The surface concentrations of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ and ${\mathrm{NO}}_{\mathrm{2}}^{-}$ in May were 20 and 0.5 µmol L⁻¹, respectively; and the concentrations decreased with depth. In November, ${\mathrm{NO}}_{\mathrm{3}}^{-}$ and ${\mathrm{NO}}_{\mathrm{2}}^{-}$ were depleted at the surface (∼3 m) and their concentrations increased with depth; at 17 m the concentrations of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ and ${\mathrm{NO}}_{\mathrm{2}}^{-}$ were 5.0 and 0.4 µmol L⁻¹, respectively. The increase in water column ${\mathrm{NO}}_{\mathrm{3}}^{-}$ and ${\mathrm{NO}}_{\mathrm{2}}^{-}$ concentrations in May and November can be attributed to increased runoff from the anthropogenically influenced watershed. Water column depletion of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ and ${\mathrm{NO}}_{\mathrm{2}}^{-}$ in the summer is the result of denitrification (Baird et al., 1995; Boynton et al., 1995), which indicates potential water column N₂O production via denitrification (discussed in Sect. 3.2).

As a proxy for the size of the denitrifying community, the abundance of the nirS gene was (5.91±0.1) × 10⁴ copy mL⁻¹ at 14 m in July, which was the highest among the three sampling trips (Fig. 1g). The lowest nirS gene abundance (9.1±1.3) × 10³ copy mL⁻¹ was observed in May at 19.5 m. The abundance of nirS was measured only at the depths at which incubations were performed, and the nirS abundance increased with increasing rates of N₂O production (see Sect. 3.2). In July 2016, water column DIC concentrations ranged from 1377 to 1831 µmol L⁻¹, with the highest concentrations below 10 m. Average community respiration rates at 3 and 14 m depth were 2.01 and 0.63 µmol L⁻¹ h⁻¹, respectively.

3.2 Active water column N₂O production

The anoxic control experiment (anoxic condition with 5 µmol L⁻¹ ${}^{\mathrm{15}}{\mathrm{NO}}_{\mathrm{2}}^{-}$ or ${}^{\mathrm{15}}{\mathrm{NO}}_{\mathrm{3}}^{-}$) was used to demonstrate active N₂O production: in July 2016, at the top of the anoxic layer (∼12.3 m), rates of N₂O production from ${\mathrm{NO}}_{\mathrm{2}}^{-}$ and ${\mathrm{NO}}_{\mathrm{3}}^{-}$ reduction were 5.42±0.35 and 2.04±0.86 nmol-N L⁻¹ h⁻¹, respectively (Fig. 2). In November 2016, at 17 m within the oxygenated water column ([O₂] >180 µmol L⁻¹), rates of N₂O production were 0.33±0.01 and 0.95±0.35 nmol-N L⁻¹ h⁻¹, respectively. In May 2017, no N₂O production was detected at 19.5 m.

Figure 2 Abundances of nirS gene and rates of N₂O production from nitrate plus nitrite reduction at three sampling times. The nirS gene abundances were analyzed at 14.1, 17.0 and 19.5 m during July 2016, November 2016 and May 2017, respectively. The N₂O production rates were measured in the control experiment (helium-flushed anoxic incubation) at 12.3, 17.0 and 19.5 m during July 2016, November 2016 and May 2017, respectively.

The total N₂O production rate of 7.5±1.2 nmol-N L⁻¹ h⁻¹ in July 2016 is lower than the measurements (18–77 nmol-N L⁻¹ h⁻¹) made 40 years ago in the Potomac River (McElroy et al., 1978), a tributary to the Chesapeake Bay. This difference could be due to much higher water column nutrients in the Potomac River (${\mathrm{NO}}_{\mathrm{2}}^{-}$ plus ${\mathrm{NO}}_{\mathrm{3}}^{-}$ concentration >30 µmol L⁻¹) at that time and presumably denser microbial populations because of sediment resuspension (4–10 m water depth). With added substrates (${\mathrm{NO}}_{\mathrm{2}}^{-}$ and ${\mathrm{NO}}_{\mathrm{3}}^{-}$) being more than an order of magnitude higher than in situ levels in July 2016, and the anoxic conditions being used in the November 2016 experiments (in situ [O₂]>180 µmol L⁻¹), N₂O production rates reported here are potential rates, which nevertheless highlight the potential for N₂O production in anoxic waters responding rapidly (within hours) to pulses of ${\mathrm{NO}}_{\mathrm{2}}^{-}$ or ${\mathrm{NO}}_{\mathrm{3}}^{-}$.

Based on the nirS gene abundance, the denitrifying population was more abundant in July than in November and was the smallest in May in the lower water column (14–19.5 m) of the Chesapeake Bay (Fig. 1g). In July the highest N₂O production rates from denitrification co-occurred with the highest nirS abundances (Fig. 2). While the water column oxygen in November was >180 µmol L⁻¹, the nirS gene abundance supported potential denitrification at a N₂O production rate of 1.28±0.35 nmol-N L⁻¹ h⁻¹ in anoxic incubation experiments. In May when hypoxic conditions had not yet developed, reduction of ${\mathrm{NO}}_{\mathrm{2}}^{-}$ or ${\mathrm{NO}}_{\mathrm{3}}^{-}$ to N₂O did not occur, and the nirS abundance (9.1×10³ copies mL⁻¹) was the lowest among three seasons. It is likely that the denitrifying community did not recover from oxygen inhibition during the 2 h anoxic incubation. A metatranscriptome analysis showed that the transcript ratios for denitrification were the lowest in June before the onset of hypoxia, and the highest ratios were in August when anoxia was most pronounced (Eggleston et al., 2015).

3.3 N₂O production pathways regulated by availability of nitrogen substrate

The ratio of the rates of N₂O production from ${\mathrm{NO}}_{\mathrm{2}}^{-}$ reduction vs. N₂O production from ${\mathrm{NO}}_{\mathrm{3}}^{-}$ reduction positively correlates with the ratio of ${\mathrm{NO}}_{\mathrm{2}}^{-}:{\mathrm{NO}}_{\mathrm{3}}^{-}$ concentrations (Fig. 3). This suggests increasing ${\mathrm{NO}}_{\mathrm{2}}^{-}$ or ${\mathrm{NO}}_{\mathrm{3}}^{-}$ availability favors N₂O production from the reduction of the respective substrate. At concentration ratios of ${\mathrm{NO}}_{\mathrm{2}}^{-}:{\mathrm{NO}}_{\mathrm{3}}^{-}<\mathrm{0.5}$, the ratios of rates were similar to the concentration ratio, 0.3±0.2. At a concentration ratio of ${\mathrm{NO}}_{\mathrm{2}}^{-}:{\mathrm{NO}}_{\mathrm{3}}^{-}=\mathrm{1}:\mathrm{1}$, the ratio of rates of N₂O production from respective substrates measured from replicate experiments varied from 0.6 to 2.6. At ${\mathrm{NO}}_{\mathrm{2}}^{-}:{\mathrm{NO}}_{\mathrm{3}}^{-}=\mathrm{10}$, the ratio of rates was greater than 10. Therefore, the primary nitrogen source of N₂O production via denitrification depends in part on the relative availability of the substrate (${\mathrm{NO}}_{\mathrm{2}}^{-}$ or ${\mathrm{NO}}_{\mathrm{3}}^{-}$).

Figure 3 Ratio of rates of N₂O production from ${\mathrm{NO}}_{\mathrm{2}}^{-}$ reduction and ${\mathrm{NO}}_{\mathrm{3}}^{-}$ reduction plotted with the respective ratio of ${\mathrm{NO}}_{\mathrm{2}}^{-}$ to ${\mathrm{NO}}_{\mathrm{3}}^{-}$ concentration in the DIN manipulation experiment from July 2016 sampling. Log scale on both axes is for clarity at the low values.

As denitrification is a stepwise enzymatic reduction from ${\mathrm{NO}}_{\mathrm{3}}^{-}$, ${\mathrm{NO}}_{\mathrm{2}}^{-}$, NO, N₂O to N₂, the pathway can be somewhat modular (Graf et al., 2014); i.e., many organisms possess only one or a few steps, rather than the complete pathway. In complete denitrifiers (organisms capable of reducing ${\mathrm{NO}}_{\mathrm{3}}^{-}$ to N₂), the degree to which intermediates (i.e., ${\mathrm{NO}}_{\mathrm{2}}^{-}$) exchange across cellular membranes with the ambient environment is unknown (Moir and Wood, 2001). We use data from the DIN manipulation experiment (conducted in July 2016) to show that full exchange between intracellular and ambient ${\mathrm{NO}}_{\mathrm{2}}^{-}$ during ${\mathrm{NO}}_{\mathrm{3}}^{-}$ reduction to N₂O is unlikely, as explained below.

The conditions and results from experiment 2-H (Table 1) were used because this experiment had the highest ambient ${\mathrm{NO}}_{\mathrm{2}}^{-}$ pool; an exchange between the pools could be easily detected. During ${\mathrm{NO}}_{\mathrm{3}}^{-}$ reduction to N₂O, if denitrifiers reduce ${}^{\mathrm{15}}{\mathrm{NO}}_{\mathrm{3}}^{-}$ (total 1.2 µmol L⁻¹, ¹⁵N fraction labeled 0.16) to ${}^{\mathrm{15}}{\mathrm{NO}}_{\mathrm{2}}^{-}$ at the maximal rate (0.2 µmol-N L⁻¹ h⁻¹; see Sect. 3.4) and the product fully exchanges with the ambient ${}^{\mathrm{14}}{\mathrm{NO}}_{\mathrm{2}}^{-}$ (10 µmol L⁻¹, ¹⁵N fraction labeled 0.0037), after 2 h, the ¹⁵N addition to the total ${\mathrm{NO}}_{\mathrm{2}}^{-}$ pool will be 0.064 µmol L⁻¹:

$$\begin{array}{ll}{\displaystyle }& {\displaystyle }(\text{Rate of}{\mathrm{NO}}_{\mathrm{2}}^{-}\text{production from}{\mathrm{NO}}_{\mathrm{3}}^{-}\times \text{incubation time}\\ {\displaystyle }& {\displaystyle }\times \text{initial}{}^{\mathrm{15}}\mathrm{N}\text{fraction of}{\mathrm{NO}}_{\mathrm{3}}^{-})\\ {\displaystyle }& {\displaystyle }=(\mathrm{0.2}\mathrm{\mu }\mathrm{mol}\text{-N}{\mathrm{L}}^{-\mathrm{1}}{\mathrm{h}}^{-\mathrm{1}}\times \mathrm{2}\mathrm{h}\times \mathrm{0.16})=\mathrm{0.064}\mathrm{\mu }\mathrm{mol}{\mathrm{L}}^{-\mathrm{1}};\end{array}$$

and the resulting ¹⁵N fraction (unitless) of ${\mathrm{NO}}_{\mathrm{2}}^{-}$ will be 0.01:

$$\begin{array}{ll}{\displaystyle }& {\displaystyle }({}^{\mathrm{15}}\mathrm{N}\text{addition to}{\mathrm{NO}}_{\mathrm{2}}^{-}+\text{initial}{}^{\mathrm{15}}\mathrm{N}\text{fraction of}{\mathrm{NO}}_{\mathrm{2}}^{-}\\ {\displaystyle }& {\displaystyle }\times \text{initial concentration of}{\mathrm{NO}}_{\mathrm{2}}^{-})/\\ {\displaystyle }& {\displaystyle }\left(\text{total concentration of}{\mathrm{NO}}_{\mathrm{2}}^{-}\right)\\ {\displaystyle }& {\displaystyle }=(\mathrm{0.064}\mathrm{\mu }\mathrm{mol}{\mathrm{L}}^{-\mathrm{1}}+\mathrm{0.0037}\times \mathrm{10}\mathrm{\mu }\mathrm{mol}{\mathrm{L}}^{-\mathrm{1}})/\\ {\displaystyle }& {\displaystyle }(\mathrm{10}+\mathrm{0.064})\mathrm{\mu }\mathrm{mol}{\mathrm{L}}^{-\mathrm{1}}\approx \mathrm{0.01}.\end{array}$$

Assuming 6 nmol-N L⁻¹ h⁻¹ as the rate of N₂O production from ${\mathrm{NO}}_{\mathrm{2}}^{-}$ reduction (the ${\mathrm{NO}}_{\mathrm{2}}^{-}\to {\mathrm{N}}_{\mathrm{2}}\mathrm{O}$ rate shown in Fig. 3; ¹⁵N fraction of ${\mathrm{NO}}_{\mathrm{2}}^{-}=\mathrm{0.01}$), and the initial N₂O concentration as 20 nmol L⁻¹ (described in Sect. 2.2; ¹⁵N fraction of N₂O=0.0037), after 2 h, the resulting ¹⁵N fraction of N₂O will be 0.0052:

$$\begin{array}{ll}{\displaystyle }& {\displaystyle }\left(\right({}^{\mathrm{15}}\mathrm{N}\text{fraction of}{\mathrm{NO}}_{\mathrm{2}}^{-}\times \text{rate of}{\mathrm{N}}_{\mathrm{2}}\mathrm{O}\text{production}\\ {\displaystyle }& {\displaystyle }\text{from}{\mathrm{NO}}_{\mathrm{2}}^{-}\times \text{incubation time})+(\text{initial}{}^{\mathrm{15}}\mathrm{N}\text{fraction}\\ {\displaystyle }& {\displaystyle }\text{of}{\mathrm{N}}_{\mathrm{2}}\mathrm{O}\times \text{initial concentration of}{\mathrm{N}}_{\mathrm{2}}\mathrm{O}\times \text{molar nitrogen}\\ {\displaystyle }& {\displaystyle }\text{in molar}{\mathrm{N}}_{\mathrm{2}}\mathrm{O}\left)\right)/\left(\right(\text{rate of}{\mathrm{N}}_{\mathrm{2}}\mathrm{O}\text{production from}{\mathrm{NO}}_{\mathrm{2}}^{-}\\ {\displaystyle }& {\displaystyle }\times \text{incubation time})+(\text{initial concentration of}{\mathrm{N}}_{\mathrm{2}}\mathrm{O}\\ {\displaystyle }& {\displaystyle }\times \text{molar nitrogen in molar}{\mathrm{N}}_{\mathrm{2}}\mathrm{O}\left)\right)\\ {\displaystyle }& {\displaystyle }=\left(\right(\mathrm{0.01}\times \mathrm{6}\mathrm{nmol}\text{-N}{\mathrm{L}}^{-\mathrm{1}}{\mathrm{h}}^{-\mathrm{1}}\times \mathrm{2}\mathrm{h})\\ {\displaystyle }& {\displaystyle }+(\mathrm{0.0037}\times \mathrm{20}\mathrm{nmol}\text{-}{\mathrm{N}}_{\mathrm{2}}\mathrm{O}{\mathrm{L}}^{-\mathrm{1}}\times \mathrm{2}\mathrm{N}/{\mathrm{N}}_{\mathrm{2}}\mathrm{O}))/\\ {\displaystyle }& {\displaystyle }(\mathrm{6}\times \mathrm{2}+\mathrm{20}\times \mathrm{2})\mathrm{nmol}\text{-N}{\mathrm{L}}^{-\mathrm{1}}=\mathrm{0.0052}.\end{array}$$

The calculated ¹⁵N fraction of N₂O (0.0052) is much lower than the measured ¹⁵N fraction of N₂O (>0.02) in experiment 2H. This means that full exchange of ${\mathrm{NO}}_{\mathrm{2}}^{-}$ during ${\mathrm{NO}}_{\mathrm{3}}^{-}$ reduction to N₂O, at maximum possible rates of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ reduction to ${\mathrm{NO}}_{\mathrm{2}}^{-}$ and N₂O, would yield a rate of N₂O production from ${\mathrm{NO}}_{\mathrm{3}}^{-}$ much lower than observed in the experimental results. Thus, we concluded that the intracellular exchange of ${\mathrm{NO}}_{\mathrm{2}}^{-}$ during ${\mathrm{NO}}_{\mathrm{3}}^{-}$ reduction to N₂O by the denitrifying community in the Chesapeake Bay is limited. Such a tight coupling among nitrate reduction, nitrite reduction and nitric oxide reduction suggests the co-occurrence of the respective functional genes and enzymes in the cell of nitrate reducers. Both dissimilatory nitrate and nitrite reducers are able to produce N₂O independently, so total N₂O production can be quantified accurately by separate measurement of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ and ${\mathrm{NO}}_{\mathrm{2}}^{-}$ reduction.

Figure 4 Rates of N₂O production from ${\mathrm{NO}}_{\mathrm{2}}^{-}$ reduction (orange circles), ${\mathrm{NO}}_{\mathrm{3}}^{-}$ reduction (green squares), and combined ${\mathrm{NO}}_{\mathrm{2}}^{-}$ and ${\mathrm{NO}}_{\mathrm{3}}^{-}$ reduction (black diamonds) under increasing oxygen concentrations in July 2016 (a) and November 2016 (b). The standard deviation of rates in most of the samples were small so that error bars are not visible. Note the scale break at 2 µmol L⁻¹ [O₂] on x axis.

3.4 Oxygen inhibits N₂O production by denitrification

The sensitivities to increasing [O₂] of ${\mathrm{NO}}_{\mathrm{2}}^{-}$ reduction and ${\mathrm{NO}}_{\mathrm{3}}^{-}$ reduction to N₂O were evaluated in samples from July and November 2016 (Fig. 4). The control experiments (anoxic incubation; see Sect. 3.2) in July 2016 and November 2016 showed rates of N₂O production from denitrification of 7.5±1.2 and 1.28±0.35 nmol-N L⁻¹ h⁻¹, respectively. Increasing [O₂] generally decreased N₂O production rates from denitrification. In July 2016, under [O₂] = 0.3 µmol L⁻¹, N₂O production from ${\mathrm{NO}}_{\mathrm{2}}^{-}$ reduction decreased from 5.4 to 2.5 nmol-N L⁻¹ h⁻¹, whereas the rate of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ reduction to N₂O increased from 2.0 to 3.5 nmol-N L⁻¹ h⁻¹. Further increase in [O₂], up to 6.4 µmol L⁻¹, did not fully inhibit N₂O production from ${\mathrm{NO}}_{\mathrm{2}}^{-}$ reduction, the rate of which was 0.08 nmol-N L⁻¹ h⁻¹. However, N₂O production from ${\mathrm{NO}}_{\mathrm{3}}^{-}$ reduction was completely inhibited when [O₂] >0.6 µmol L⁻¹ (Fig. 4a). In November 2016, increasing [O₂] gradually decreased rates of ${\mathrm{NO}}_{\mathrm{2}}^{-}$ reduction to N₂O; no rates were detected when [O₂]>2 µmol L⁻¹. Rates of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ reduction to N₂O were not detected at [O₂]>0 µmol L⁻¹ (Fig. 4b).

Rates of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ reduction to ${\mathrm{NO}}_{\mathrm{2}}^{-}$ under increasing [O₂] were also measured in July 2016 to supplement the sensitivity analysis of denitrification to oxygen. The rate of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ reduction to ${\mathrm{NO}}_{\mathrm{2}}^{-}$ was 100 nmol-N L⁻¹ h⁻¹ under anoxic condition. At [O₂]=0.3 µmol L⁻¹, the rate doubled to 200 nmol-N L⁻¹ h⁻¹ (Fig. 4). Further increase in [O₂] significantly decreased the rate of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ reduction to ${\mathrm{NO}}_{\mathrm{2}}^{-}$. However, at [O₂]=6.4 µmol L⁻¹ ${\mathrm{NO}}_{\mathrm{3}}^{-}$ reduction to ${\mathrm{NO}}_{\mathrm{2}}^{-}$ was still detectable at 0.82±0.06 nmol-N L⁻¹ h⁻¹ (Fig. 5).

Figure 5 Rates of ${\mathrm{NO}}_{\mathrm{2}}^{-}$ production from ${\mathrm{NO}}_{\mathrm{3}}^{-}$ reduction under increasing oxygen concentrations. Error bar indicates the standard deviation of rates from linear regression of three time points (n=7).

These results suggest that oxygenation of the water column in the Chesapeake Bay, even micro-molar level oxygen, would significantly mitigate N₂O production from denitrification. Both July 2016 and November 2016 data showed the difference in the effect of oxygen on N₂O production from ${\mathrm{NO}}_{\mathrm{2}}^{-}$ vs. ${\mathrm{NO}}_{\mathrm{3}}^{-}$ reduction. Samples from July 2016 showed 98 % and complete inhibition on N₂O production from ${\mathrm{NO}}_{\mathrm{2}}^{-}$ and ${\mathrm{NO}}_{\mathrm{3}}^{-}$ reduction at [O₂]=6 µmol L⁻¹, respectively. The November 2016 samples showed 94 % and complete inhibition on N₂O production from ${\mathrm{NO}}_{\mathrm{2}}^{-}$ and ${\mathrm{NO}}_{\mathrm{3}}^{-}$ reduction at [O₂]=0.4 µmol L⁻¹, respectively. Furthermore, N₂O production in the Chesapeake Bay was likely attributed to both heterotrophic denitrification and nitrifier denitrification. Studies have shown that both nitrifiers and denitrifiers are present in the Chesapeake Bay (Bouskill et al., 2012; Hong et al., 2014) and they are capable of ${\mathrm{NO}}_{\mathrm{2}}^{-}$ reduction to N₂O, whereas ${\mathrm{NO}}_{\mathrm{3}}^{-}$ reduction to N₂O is solely mediated by heterotrophic denitrifiers. N₂O production via nitrifier denitrification occurs under the full range of oxygen environments in agricultural soil (Zhu et al., 2013) and the open ocean (Wilson et al., 2014). Partial denitrification (${\mathrm{NO}}_{\mathrm{3}}^{-}$ reduction to N₂O), however, is moderately oxygen sensitive. Thus, increasing oxygen inhibits the activities of denitrifiers, as demonstrated in decreasing rates of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ reduction to N₂O (Fig. 3) and ${\mathrm{NO}}_{\mathrm{3}}^{-}$ reduction to ${\mathrm{NO}}_{\mathrm{2}}^{-}$ (Fig. 5). Increasing oxygen does not completely inhibit N₂O production activity of nitrifiers but probably lowers the N₂O production rates by nitrifier denitrification.

4 Conclusion and outlook

The Chesapeake Bay is a potential N₂O source via denitrification when ${\mathrm{NO}}_{\mathrm{3}}^{-}$ and ${\mathrm{NO}}_{\mathrm{2}}^{-}$ are present under anoxic conditions. Relative rates of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ and ${\mathrm{NO}}_{\mathrm{2}}^{-}$ reduction to N₂O were positively correlated with relative concentrations of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ and ${\mathrm{NO}}_{\mathrm{2}}^{-}$. Increased oxygen, either by natural water column oxygenation or by experimental manipulation, decreased N₂O production rates via denitrification. The size of the denitrifying community increased with increasing rates of N₂O production via denitrification. The potential N₂O production in the summertime suggests that intermittent N₂O efflux to the atmosphere could occur when a shallow oxic–anoxic interface (typically 10–15 m) is present (Taft et al., 1980; Kemp et al., 1992; Lee et al., 2015a), as well as frequent disturbance of water column stratification by storm events, boat traffic and surface cooling. The seasonal variation of surface and water column N₂O saturation levels (air undersaturated in May and air oversaturated in November) and the detection of significant N₂O production in July (summer) when N₂O concentrations were the lowest imply that N₂O consumption was also occurring in the Chesapeake Bay and probably minimizing N₂O efflux to the atmosphere. A long-term, comprehensive survey with wide spatial coverage will help (i) assess if the Chesapeake Bay is a net N₂O source or sink on an annual scale and (ii) to investigate the physical, chemical and biological controls of N₂O emission in the Chesapeake Bay.

Denitrification is critical for complete removal of fixed nitrogen so as to mitigate eutrophication in natural waters. The N₂O production rates could serve as a proxy for estimating nitrogen loss. It is estimated that 1 % of total denitrified nitrogen is converted to N₂O in river networks (Beaulieu et al., 2011) so the ratio of N₂O:N₂ during denitrification is 1:100. Assuming that N₂O production occurs at a rate of 7 nmol-N L⁻¹ h⁻¹ within 0.2 m of the oxic–anoxic interface in summertime (based on the July 2016 control data, N₂O production from ${\mathrm{NO}}_{\mathrm{3}}^{-}$ plus ${\mathrm{NO}}_{\mathrm{2}}^{-}$), denitrification yields a potential water column nitrogen removal rate of 140 µmol-N m⁻² h⁻¹, or 0.24 mg-N m⁻² d⁻¹. In addition, the sediment in the bay is capable of anaerobic ammonia oxidation (Rich et al., 2008) and denitrification (Kemp et al., 1990; Kana et al., 2006). Total sedimentary N₂ production, measured by the acetylene block reduction method (Kemp et al., 1990) and N₂ accumulation method (Kana et al., 2006), recorded areal rates of 50–70 µmol-N m⁻² h⁻¹. Therefore, expansion of anoxia in the Chesapeake Bay could increase the potential of biological nitrogen removal by the sediment–water system that counteracts the increase in nitrogen loading from anthropogenic activities.

The oxidation of ${\mathrm{NH}}_{\mathrm{4}}^{+}$, although not the focus of this study, is a possible pathway for N₂O production under low-oxygen conditions (Anderson, 1964). The yield of N₂O (molar ratio of N₂O production to ${\mathrm{NH}}_{\mathrm{4}}^{+}$ oxidation) increases with decreasing oxygen (Goreau et al., 1980). Culture (Qin et al., 2017) and field studies (Bristow et al., 2016; Peng et al., 2016) have shown high affinity of oxygen (<5 µmol L⁻¹) during ${\mathrm{NH}}_{\mathrm{4}}^{+}$ oxidation. The main sources of ${\mathrm{NH}}_{\mathrm{4}}^{+}$ in the Chesapeake Bay include remineralization of organic matter in the oxygenated water column and sediments (Kemp et al., 1990) and atmospheric deposition (Larsen et al., 2001). Onset of ${\mathrm{NH}}_{\mathrm{4}}^{+}$ oxidation is viable at ${\mathrm{NH}}_{\mathrm{4}}^{+}$ concentration <100 nmol L⁻¹ by the natural ammonia-oxidizing community (Horak et al., 2013). Thus, N₂O production from ${\mathrm{NH}}_{\mathrm{4}}^{+}$ oxidation might be stimulated under low-oxygen condition by influx of ammonium near the oxic–anoxic interface, which deserves future research efforts.

The inhibition of N₂O production by oxygen highlights the positive outcomes of reoxygenation of the Chesapeake Bay. Since the late 20th century, the Chesapeake Bay has received increased anthropogenic nitrogen loading from various sources including fertilizer (Groffman et al., 2009), untreated sewage (Kaplan et al., 1978) and atmospheric deposition (Russell et al., 1998; Loughner et al., 2016). Fueled by increased nitrogen input, elevated primary production in the surface layer stimulates aerobic remineralization at depth, which consumes oxygen rapidly. In summertime, water column stratification restricts influx of oxygen to depth, creating seasonal anoxia/hypoxia in the bay. The documented eutrophication and expansion of anoxia/hypoxia in the Chesapeake Bay in the late 20th century attracted public attention because of increasing mortality of organisms with high commercial and recreational value (Cooper and Brush, 1993). Moreover, expansion of the volume of low-oxygen waters will result in more “hot spots” for N₂O production. The key factor for mitigating anoxia is to control the nitrogen input to the bay (Hagy et al., 2004; Zhou et al., 2014). Effective fertilizer application, sewage treatment, natural nitrogen removal by denitrification and anammox and plant uptake have been successfully enforced to control the nitrogen runoff into the bay from the tributaries (Boesch et al., 2001). The near absence of summertime water column ${\mathrm{NO}}_{\mathrm{2}}^{-}+{\mathrm{NO}}_{\mathrm{3}}^{-}$ concentrations close to the central Chesapeake Bay as shown in this study and others (Lee et al., 2015a) could prevent N₂O production. Reducing the nitrogen input into the Chesapeake Bay will help mitigate N₂O efflux: In the short term (timescale of days to months), nitrogen sources (${\mathrm{NH}}_{\mathrm{4}}^{+}$, ${\mathrm{NO}}_{\mathrm{2}}^{-}$ and ${\mathrm{NO}}_{\mathrm{3}}^{-}$) for N₂O production will be decreased. In the long run (inter-annual timescale), eutrophication will be alleviated, which will reoxygenate the water column, and inhibit N₂O production.