The fate of fixed nitrogen in Santa Barbara Basin sediments during seasonal anoxia

Despite long-standing interest in the biogeochemistry of the Santa Barbara Basin (SBB), there are no direct rate measurements of different nitrogen transformation processes. We investigated benthic nitrogen cycling using in situ incubations with ¹⁵NO${}_{\mathrm{3}}^{-}$ addition and quantified the rates of total nitrate (NO${}_{\mathrm{3}}^{-}$) uptake, denitrification, anaerobic ammonia oxidation (anammox), N₂O production, and dissimilatory nitrate reduction to ammonia (DNRA). Denitrification was the dominant NO${}_{\mathrm{3}}^{-}$ reduction process, while anammox contributed 0 %–27 % to total NO${}_{\mathrm{3}}^{-}$ reduction. DNRA accounted for less than half of NO${}_{\mathrm{3}}^{-}$ reduction except at the deepest station at the center of the SBB where NO${}_{\mathrm{3}}^{-}$ concentration was lowest. NO${}_{\mathrm{3}}^{-}$ availability and sediment total organic carbon content appeared to be two key controls on the relative importance of DNRA. The increasing importance of fixed N retention via DNRA relative to fixed N loss as NO${}_{\mathrm{3}}^{-}$ deficit intensifies suggests a negative feedback loop that potentially contributes to stabilizing the fixed N budget in the SBB. Nitrous oxide (N₂O) production as a fraction of total NO${}_{\mathrm{3}}^{-}$ reduction ranged from 0.2 % to 1.5 %, which was higher than previous reports from nearby borderland basins. A large fraction of NO${}_{\mathrm{3}}^{-}$ uptake was unaccounted for by NO${}_{\mathrm{3}}^{-}$ reduction processes, suggesting that intracellular storage may play an important role. Our results indicate that the SBB acts as a strong sink for fixed nitrogen and potentially a net source of N₂O to the water column.

1 Introduction

Oxygen minimum zones (OMZs) in the world's ocean, whether they are formed naturally or induced by human activities, have been expanding in the past century (Horak et al., 2016; Oschlies et al., 2017; Stramma et al., 2008). As oxygen (O₂) concentration is one of the key controls on biogeochemical processes, including nitrogen (N) cycling, N biogeochemistry in OMZs has been extensively studied (Paulmier and Ruiz-Pino, 2009; Zehr, 2009). Denitrification, the reduction of nitrate (NO${}_{\mathrm{3}}^{-}$) to dinitrogen gas (N₂), and anaerobic ammonia oxidation (anammox), where nitrite (NO${}_{\mathrm{2}}^{-}$) and ammonium (NH${}_{\mathrm{4}}^{+}$) are converted into N₂ by comproportionation, are two major sinks of the oceanic fixed N budget (Gruber, 2008). These two processes are inhibited by the presence of O₂ and sulfide, and their rates are sensitive to O₂ at nanomolar concentrations (Dalsgaard et al., 2014; Joye and Hollibaugh, 1995; Caffrey et al., 2019). Because the last step of the sequential reduction of NO${}_{\mathrm{3}}^{-}$ during denitrification, N₂O reduction, is the most sensitive to O₂ (Zumft, 1997), the production of nitrous oxide (N₂O) as a byproduct of denitrification is usually elevated under hypoxic conditions, i.e., in the presence of O₂ (Firestone et al., 1980; Ji et al., 2015). Additionally, nitrification, i.e., the oxidation of NH${}_{\mathrm{4}}^{+}$ and subsequently NO${}_{\mathrm{2}}^{-}$, is another major source of N₂O in the ocean (Elkins et al., 1978), and the relative yield of N₂O from nitrification is high under low-O₂ conditions (<4 µM) (Ji et al., 2018). Under O₂ limitation, dissimilatory nitrate reduction to ammonia (DNRA) coupled to organic matter degradation is another important process that results in fixed N retention instead of removal (Burgin and Hamilton, 2007). When viewed as competing processes, DNRA is favored over denitrification under NO${}_{\mathrm{3}}^{-}$-limited conditions where electron donors are in excess (Tiedje et al., 1983). Additionally, under sulfidic conditions, autotrophic DNRA coupled to sulfide oxidation can become a dominant pathway for NO${}_{\mathrm{3}}^{-}$ reduction (Shao et al., 2011).

The Santa Barbara Basin (SBB) is one of the borderland basins off the southern part of the coast of California and characterized by high export production (Thunell, 1998). Because the bottom water (maximum depth 586 m) in the SBB is separated from the area outside the basin by relatively shallow sills on the eastern end (∼200 m deep) and the western end (∼475 m deep), O₂ concentrations at the basin's bottom are generally low and usually fluctuate between 1 and 30 µM (Bograd et al., 2002; Goericke et al., 2015; Reimers et al., 1990; Sholkovitz and Gieskes, 1971; Myhre et al., 2018). During upwelling seasons (winter and spring), water is advected from outside the basin and replenishes bottom water O₂ in the SBB. However, high export production fuels O₂ demand that maintains low O₂ levels within the basin at depths below the deeper sill (Thunell, 1998). As a consequence, anoxia develops at the bottom of the SBB until the next upwelling event (Goericke et al., 2015), and large coverage of bacterial mats on the sea floor has been reported in the SBB (Valentine et al., 2016).

Using water column NO${}_{\mathrm{3}}^{-}$ concentration data collected in the SBB by the California Cooperative Oceanic Fisheries Investigations (CalCOFI) along longitudinal transects (Koslow et al., 2010), Valentine et al. (2016) estimated the benthic NO${}_{\mathrm{3}}^{-}$ uptake rate to be as high as 11.7 mmol m⁻² d⁻¹, which was one of the highest rates ever reported. However, the fate of the NO${}_{\mathrm{3}}^{-}$ in the sediments remains unclear as there are no direct rate measurements of N cycling processes in the SBB. Indirect estimates using analysis of stable isotopes of water column NO${}_{\mathrm{3}}^{-}$ suggest that benthic denitrification accounts for >75 % of NO${}_{\mathrm{3}}^{-}$ loss in the SBB, and the rates of benthic denitrification were estimated to be the highest among borderland basins in the eastern tropical North Pacific (Sigman et al., 2003). Benthic anammox is expected to occur in the SBB (Prokopenko et al., 2006), but the relative contribution of denitrification and anammox to N₂ production has not been assessed. In addition to different dissimilatory processes that reduce NO${}_{\mathrm{3}}^{-}$, the apparent NO${}_{\mathrm{3}}^{-}$ drawdown could also be attributed to intracellular storage by both prokaryotes and microbial eukaryotes (Kamp et al., 2015; Bernhard et al., 2012; Schulz et al., 1999). With respect to N₂O, these other borderland basins are considered to be a weak sink (Townsend-Small et al., 2014). As the SBB stands out in terms of denitrification, it may be expected that SBB benthic cycling of N₂O is also unique.

To decipher the fate of NO${}_{\mathrm{3}}^{-}$ taken up by SBB sediments, we performed in situ incubations using benthic flux chambers with added ¹⁵NO${}_{\mathrm{3}}^{-}$ along the bottom slope traversing north–south across the deeper portion of the SBB. By calculating the rates of N₂ production by denitrification and anammox, total N₂O production, and DNRA, we assess the overall rates of NO${}_{\mathrm{3}}^{-}$ uptake and reduction rates. Accompanying geochemical data are used to explore the controls on the relative importance of NO${}_{\mathrm{3}}^{-}$ retention via DNRA.

2 Materials and methods

2.1 In situ incubations with benthic flux chambers

Remotely operated vehicle (ROV) Jason deployed automated benthic flux chambers (BFCs) and conducted sediment push coring at seven stations (Fig. 1) in the SBB along a southern and a northern depth and O₂ gradient originating from the depocenter in the deepest point of the basin (Table 1). Station depth, latitude, and longitude were automatically generated by the Jason data processor using navigation data derived from the Doppler velocity log system and the ultrashort baseline positioning system. Bottom water O₂ concentration was determined using a type 4831 O₂ optode sensor (Aanderaa Data Instruments AS, Bergen, NO) on the ROV and calibrated against Winkler titration measurements of seawater collected from Niskin bottles (Qin et al., 2022). Bottom water was collected using Niskin bottles and stored frozen at −30 °C until lab analysis for nitrate (NO${}_{\mathrm{3}}^{-}$) concentration following the spectrophotometric method described by García-Robledo et al. (2014).

Figure 1 Sampling stations in the Santa Barbara Basin. The color contours show bathymetry data from the General Bathymetric Chart of the Oceans at 30 arcsec resolution (Becker et al., 2009) visualized in Ocean Data View v5.6.2 (Schlitzer, 2002).

Table 1 Sampling date, latitude, longitude, depth, bottom water concentrations of oxygen and nitrate, chamber volume, total organic carbon (TOC) and nitrogen (TON), and C : N molar ratio of organic matter in the top 2 cm of the sediment (by dry weight %) at the seven sampling stations in the Santa Barbara Basin. Oxygen concentrations below detection limit of the type 4831 (Aanderaa Data Instruments AS, Bergen, NO) oxygen optode sensor (3 µM) and the Winkler titration method (1 µM) are denoted by “bdl”. Note that oxygen concentrations in the bottom water at NDRO and SDRO were confirmed to be zero through additional analytical methods (see Yousavich et al., 2024).

Sediment samples for total organic carbon (TOC) and total organic nitrogen (TON) analyses were subsampled from push cores (polycarbonate, 30.5 cm length, 6.35 cm inner diameter) retrieved by ROV Jason that were sectioned in 1 cm increments up to 10 cm followed by 2 cm increments below 10 cm (Yousavich et al., 2024). Wet sediments were dried for up to 48 h at 50 °C and treated with 6N HCl to dissolve carbonate minerals (Harris et al., 2001). Samples were then washed with ultrapure water and dried again at 50 °C. An aliquot (∼10–15 mg) was then packed into individual 8×5 mm pressed tin capsules and analyzed at the University of California Davis stable isotope facility using a PDZ Europa 20-20 isotope ratio mass spectrometer (Sercon Ltd., Cheshire, UK). TOC and TON were calculated based on the sample peak area corrected against a reference material (alfalfa flour). Molar concentrations, obtained from measured TOC and TON (in wt %), were used to calculate carbon-to-nitrogen (C : N) ratios.

The design of the BFCs has been described previously (Vonnahme et al., 2020). In brief, a stirred cylindrical polycarbonate chamber (inner diameter =19 cm) equipped with conductivity and oxygen sensors in the lid (type 5860 and 4330, respectively, Aanderaa Data Instruments AS, Bergen, NO) was inserted into the sediment to enclose a sediment patch of 284 cm² together with 2.5 to 4.5 L of overlying water. The chambers were outfitted with a syringe sampler hosting one injection syringe and six sampling syringes to inject into and take samples from the overlying water at approximately 60 min intervals. The injection syringe contained 200 µmol of ¹⁵N-labeled potassium nitrate (Cambridge Isotopes) dissolved in 50 mL of deionized water. To minimize the introduction of O₂, the ¹⁵N-labeled potassium nitrate solution was purged by ultra-high-purity helium at 5 mL min⁻¹ for 60 min prior to being loaded into the injection syringe. The post-injection decrease in salinity in the chamber (as detected by the conductivity sensor) was used to calculate the volume of the benthic flux chamber (Kononets et al., 2021). Depending on the chamber volume, the total concentration of NO${}_{\mathrm{3}}^{-}$ ranged between 50 and 100 µM at the beginning of in situ incubations. This level of NO${}_{\mathrm{3}}^{-}$ amendment was intended to prevent its depletion before the end of incubations given the potentially high rates of NO${}_{\mathrm{3}}^{-}$ uptake estimated by a previous study (Valentine et al., 2016).

After recovery, water samples from the BFC were transferred to evacuated 12 mL vials (Exetainer^®, Labco, Lampeter, UK) pre-filled with 0.1 mL of 7 M zinc chloride for preservation. Prior to analysis of the isotopic compositions of N₂ and N₂O, a 5 mL sample was replaced with ultra-high-purity helium to create a headspace. The concentration and δ¹⁵N of dissolved N₂ and N₂O were determined using a Sercon CryoPrep gas concentration system interfaced to a Sercon 20-20 isotope ratio mass spectrometer (IRMS) at the University of California Davis Stable Isotope Facility. The measurement precision was ±0.2 ‰ for δ¹⁵N.

Water samples from the benthic flux chambers for analysis of ¹⁵NH${}_{\mathrm{4}}^{+}$ were filtered through sterile 47 mm syringe filters (0.2 µm pore size) and frozen immediately. The production of ¹⁵NH${}_{\mathrm{4}}^{+}$ in seawater samples was measured using a method adapted from Zhang et al. (2007) and described previously (Peng et al., 2016). In brief, NH${}_{\mathrm{4}}^{+}$ was first oxidized to NO${}_{\mathrm{2}}^{-}$ using hypobromite (BrO⁻) and then reduced to N₂O using an acetic acid azide working solution (McIlvin and Altabet, 2005; Zhang et al., 2007). The δ¹⁵N of the produced N₂O was determined using an Elementar Americas PrecisION continuous flow, multicollector, isotope ratio mass spectrometer (CF-MC-IRMS) coupled to a custom-built automated gas extraction and preparation system similar to the system described in McIlvin and Casciotti (2011). Calibration and correction were performed as described in Zhang et al. (2007). The measurement precision was ±0.2 ‰ for δ¹⁵N. NH${}_{\mathrm{4}}^{+}$ solutions (10 µM) from a mixture of 99 % ¹⁵NH₄Cl (Cambridge Isotopes) and IAEA standard N1 (δ¹⁵N =1.2 ‰) with a final δ¹⁵N of 135 ‰, 676 ‰, 1351 ‰, 5404 ‰, and 10 806 ‰ were prepared and used as in-house reference standards. The IRMS measurements of these in-house reference standards scaled linearly (R²=0.9996) with their δ¹⁵N values.

2.2 Rate calculations and statistics

Production rates of ²⁹N₂, ³⁰N₂, ¹⁵NH${}_{\mathrm{4}}^{+}$, and total N₂O were calculated from the slope of the concentrations of the respective species at the syringe sampling time points by fitting a linear regression multiplied by the overlying water column volume and divided by the chamber area. The linear regressions excluded the last one or two sampling time points if they clearly deviated from a linear trend compared to the first four or five sampling time points. The rates of N₂ production from denitrification and anammox were calculated following a previously described method (Thamdrup and Dalsgaard, 2002) with modifications to account for coupled DNRA–anammox (Peng et al., 2021). The calculation was set up with denitrification rate (R_DN) and anammox rate (R_AMX) as unknowns:

$$\begin{array}{}\text{(1)}& {\displaystyle }{R}_{\mathrm{DN}}\cdot f_{\mathrm{N}}^{\mathrm{2}}+R_{\mathrm{AMX}}\cdot f_{\mathrm{A}}\cdot f_{\mathrm{N}}=P^{\mathrm{30}},\text{(2)}& {\displaystyle }\begin{array}{rl}{R}_{\mathrm{DN}}& \cdot \mathrm{2}\cdot f_{\mathrm{N}}\cdot \left(\mathrm{1}-f_{\mathrm{N}}\right)+R_{\mathrm{AMX}}\cdot [f_{\mathrm{A}}\cdot \left(\mathrm{1}-f_{\mathrm{N}}\right)\\ & +(\mathrm{1}-f_{\mathrm{A}})\cdot f_{\mathrm{N}}]=P^{\mathrm{29}},\end{array}\end{array}$$

where P²⁹ and P³⁰ are the respective production rates of ²⁹N₂ and ³⁰N₂ that were calculated from measured concentrations stated above, f_N is the fraction of ¹⁵N in the NO${}_{\mathrm{3}}^{-}$ pool, and f_A is the fraction of ¹⁵N in the NH${}_{\mathrm{4}}^{+}$ pool. The solution for R_DN and R_AMX is

$$\begin{array}{}\text{(3)}& {\displaystyle }{R}_{\mathrm{DN}}={\displaystyle \frac{\left(f_{\mathrm{A}}+f_{\mathrm{N}}-\mathrm{2}\cdot f_{\mathrm{A}}\cdot f_{\mathrm{N}}\right)\cdot P^{\mathrm{30}}-f_{\mathrm{A}}\cdot f_{\mathrm{N}}\cdot P^{\mathrm{29}}}{f_{\mathrm{N}}^{\mathrm{2}}\cdot (f_{\mathrm{N}}-f_{\mathrm{A}})}},\text{(4)}& {\displaystyle }{R}_{\mathrm{AMX}}={\displaystyle \frac{f_{\mathrm{N}}\cdot P^{\mathrm{29}}-\mathrm{2}\cdot (\mathrm{1}-f_{\mathrm{N}})\cdot P^{\mathrm{30}}}{f_{\mathrm{N}}\cdot (f_{\mathrm{N}}-f_{\mathrm{A}})}}.\end{array}$$

Errors calculated from the linear regression of ²⁹N₂ and ³⁰N₂ production rates were propagated to R_DN and R_AMX following established statistical methods (Deming, 1943). Detection limits of the calculated rates were estimated as double the standard deviation from linear regressions. Depending on the in situ NO${}_{\mathrm{3}}^{-}$ concentration, the detection limit for total N₂ production from denitrification and anammox ranged between 0.04 and 0.17 mmol m⁻² d⁻¹ and 0.04 and 0.24 mmol m⁻² d⁻¹ (Table S1 in the Supplement), respectively. The detection limit for N₂O production ranged between 1.1 and 5.6 µmol m⁻² d⁻¹. DNRA rates were calculated as the rates of increase in ¹⁵NH${}_{\mathrm{4}}^{+}$ divided by f¹⁵, where f¹⁵ is the fraction of ¹⁵N in the NO${}_{\mathrm{3}}^{-}$ pool. Because part of the produced ¹⁵NH${}_{\mathrm{4}}^{+}$ would be adsorbed to sediment minerals, the rates of ¹⁵NH${}_{\mathrm{4}}^{+}$ production were further multiplied by a factor of 2 (De Brabandere et al., 2015; Laima, 1994). Depending on the in situ NH${}_{\mathrm{4}}^{+}$ concentration, the detection limit for total NH${}_{\mathrm{4}}^{+}$ production rates ranged between 0.01 and 0.07 mmol m⁻² d⁻¹ (Table S1).

3 Results and discussion

3.1 Interpretation of rate measurements from benthic flux chamber incubations

The use of benthic flux chambers to perform ¹⁵NO${}_{\mathrm{3}}^{-}$ incubation experiments in situ offers multiple advantages over other techniques such as slurry or whole-core incubations, including minimal disturbance of the sediment, maintenance of in situ pressure and temperature, and relatively large surface area which can account for spatial heterogeneity (Aller et al., 1998; Hall et al., 2007; Nielsen and Glud, 1996; Robertson et al., 2019). On the other hand, several limitations of using tracer incubations with benthic flux chambers can lead to either underestimated or overestimated rates. First, the diffusion of added ¹⁵NO${}_{\mathrm{3}}^{-}$ into sediments and the labeled ¹⁵NO${}_{\mathrm{3}}^{-}$ reduction products out of sediments in this study was unlikely at steady state. ¹⁵NO${}_{\mathrm{3}}^{-}$ added to the overlying water of the chambers diffuses into sediment porewater where O₂ is depleted within the first few millimeters, sustaining benthic NO${}_{\mathrm{3}}^{-}$ reduction. However, a share of the labeled N compounds that are produced will diffuse to pore waters in deeper sediment layers and, hence, cannot be detected in samples taken from the overlying waters. This can lead to an underestimation of NO${}_{\mathrm{3}}^{-}$ reduction rates.

Second, the addition of NO${}_{\mathrm{3}}^{-}$ at concentrations that were 1.6–6.2 (median =2.3) times as high as ambient concentrations could lead to overestimation of rates. The NO${}_{\mathrm{3}}^{-}$ uptake rates calculated as the decrease in total NO${}_{\mathrm{3}}^{-}$ concentration over time was 1.9–6.4 (median =3.8) times higher than those measured in parallel chambers deployed at the same time without any added ¹⁵NO${}_{\mathrm{3}}^{-}$ (Table S2; Yousavich et al., 2024). While the diffusive loss of NO${}_{\mathrm{3}}^{-}$ to the sediment porewater is expected to account for the stimulated NO${}_{\mathrm{3}}^{-}$ uptake partially, NO${}_{\mathrm{3}}^{-}$ addition also likely stimulated the rates of NO${}_{\mathrm{3}}^{-}$ reduction and intracellular storage. However, it remains unclear whether the accelerated NO${}_{\mathrm{3}}^{-}$ uptake is partitioned between intracellular storage and reduction in the same proportion as under unamended conditions, which would partially depend on the carrying capacity of NO${}_{\mathrm{3}}^{-}$ storage vs. reduction.

Third, the slight increase in O₂ concentration in benthic chambers could have affected the rates of dissimilatory NO${}_{\mathrm{3}}^{-}$ reduction and led to underestimates. O₂ in bottom water (and, therefore, also in pore waters) was depleted (below detection of the Winkler titration method, 1 µM) at the deepest stations SDRO and NDRO (southern and northern depocenter radial origin, respectively; Table 1). O₂ concentrations in the overlying water in most incubations were slightly increasing over the period of the incubation with an average rate of 0.11±0.44 µmol h⁻¹. The increase is attributed to a release of O₂ from the polycarbonate walls and lids of the chambers that were exposed to air until shortly before deployment. The net increase in O₂ in the overlying water indicates that rates of O₂ provision from the plastics were in most cases higher than the rates of O₂ uptake by the enclosed sediment. A release of O₂ from plastics has been reported by a previous study which showed rates of O₂ provided from polycarbonate to O₂-poor waters were among the highest of all plastics tested (Stevens, 1992). The extent to which the artificial elevation of O₂ levels in the water overlaying the sediment in the chambers may have affected N transformation pathways and rates will depend on the O₂ sensitivity of the respective processes and the penetration depth of O₂ into the sediment. This effect was likely insignificant in our incubations in the SBB because the rate of O₂ change was minimal compared to ambient O₂ concentrations except for station NDRO, where O₂ concentrations in the chamber water rose from below detection to 10 µM (Table 1). While there are limitations that can lead to both underestimates and overestimates, there is the possibility that they level each other out and our observations are close to in situ production rates of N₂, N₂O, and NH${}_{\mathrm{4}}^{+}$. Despite this concern, the relative contribution of different NO${}_{\mathrm{3}}^{-}$ reduction processes and the general trend of NO${}_{\mathrm{3}}^{-}$ reduction rates across the surveyed transect in the SBB are likely representative of in situ conditions.

The NO${}_{\mathrm{3}}^{-}$ reduction rates measured in our experiments represent only the benthic contribution because the water samples in the six sampling syringes were subsampled simultaneously after recovery and no preservative was added inside the sampling syringe to terminate reactions. Therefore, we assume that NO${}_{\mathrm{3}}^{-}$ reduction in the overlying water (and in the syringes after respective samples have been taken) contributed equally among all six samples to the production of N₂, N₂O, and NH${}_{\mathrm{4}}^{+}$ and does not interfere with our rate calculations. Separate water incubations would be needed to determine the rates of NO${}_{\mathrm{3}}^{-}$ reduction in the water column. To account for NH${}_{\mathrm{4}}^{+}$ adsorption which could lead to an underestimate of DNRA, we made the assumption that an amount of ¹⁵NH${}_{\mathrm{4}}^{+}$ that equals the measured increase in the benthic flux chambers is adsorbed to sediment minerals (Hall et al., 2017; Laima, 1994). The rates determined in this study were determined during seasonal anoxia when bottom water O₂ was below detection at the depocenter of the basin. Additional expeditions are required to capture seasonal variations in these N cycling processes.

3.2 Denitrification was the dominant NO${}_{\mathrm{3}}^{-}$ reduction pathway

On average, N₂ production by denitrification and anammox was dominant over DNRA in this study, accounting for 70.4±16.4 % of total NO${}_{\mathrm{3}}^{-}$ reduction (Fig. 2 and Table 2). Total N₂ production rates ranged from 0.89 to 3.60 mmol N m⁻² d⁻¹, which were lower compared to a previous estimate (∼4.5 mmol N m⁻² d⁻¹) based on NO${}_{\mathrm{3}}^{-}$ stable isotope mass balance calculations for the SBB (Sigman et al., 2003). Nevertheless, the previous estimate includes large uncertainties, and the rates calculated from stable isotope mass balance represent signals integrated over multiple seasons (Sigman et al., 2003), whereas our measurements represent snapshots obtained in one season of one year when the bottom water NO${}_{\mathrm{3}}^{-}$ was not depleted. N₂ production rates at seasons more depleted in NO${}_{\mathrm{3}}^{-}$ concentrations in the bottom water compared to our study might more closely resemble rates estimated by Sigman et al. (2003). Season-resolving studies are needed in the future to understand the natural variability in the system and assess potential effects of stressors such as deoxygenation and rising temperature.

Figure 2 Inorganic N-species production rates determined from ¹⁵N–NO${}_{\mathrm{3}}^{-}$ labeling studies with in situ benthic flux chambers: N₂ production rate from denitrification, N₂ production rate from anammox, NH${}_{\mathrm{4}}^{+}$ production rate, and N₂O production rate. Note the lower range (right y axis) for N₂O production. Error bars represent standard errors of the calculated slope from linear regressions of N₂ and N₂O production over time.

N₂ production rates in this study were higher than most of those reported in other studies using in situ incubations with benthic flux chambers (Bonaglia et al., 2017; De Brabandere et al., 2015; Hall et al., 2017; van Helmond et al., 2020; Hylén et al., 2022). Elevated rates in the SBB are likely a result of the high organic matter content of sediment (4.1 %–6.8 % total organic carbon; Table 1), supporting high microbial respiration rates, and little (max 20 mm) to zero O₂ penetration into the sediment (Yousavich et al., 2024). Compared to the SBB, organic matter content in sediment of previous studies, including the anoxic Eastern Gotland Basin (Hall et al., 2017), the largely pristine and oxygenated Gulf of Bothnia (Bonaglia et al., 2017), and an anoxic fjord basin in the By Fjord on the Swedish west coast (De Brabandere et al., 2015), was lower and the N₂ production rates were typically <1 mmol N m⁻² d⁻¹. In comparison, N₂ production rates reached 1.72±0.77 mmol N m⁻² d⁻¹ in the sediment underlying eutrophic waters of the Stockholm archipelago, where organic matter content was similar to SBB sediment (6.3 % $w/w$) and O₂ penetration depth was <4 mm (van Helmond et al., 2020). Additionally, benthic denitrification rates in the SBB (1.37±0.64 mmol N m⁻² d⁻¹) were similar to those reported from the Peruvian OMZ (1.31±0.60 mmol N m⁻² d⁻¹) where bottom water O₂ was lower than 10 µM and the organic matter content was similar (up to 7.5 % TOC and 0.9 % TON) to that in SBB sediments (Bohlen et al., 2011; Henrichs and Farrington, 1984; Sommer et al., 2016).

Table 2 The relative contribution of different processes (total N₂ production, N₂ from denitrification, N₂ from anammox, NH${}_{\mathrm{4}}^{+}$ from DNRA, and N₂O production) to total NO${}_{\mathrm{3}}^{-}$ reduction (upper part) and the relative contribution of total NO${}_{\mathrm{3}}^{-}$ reduction to total NO${}_{\mathrm{3}}^{-}$ uptake (lower part) in the Santa Barbara Basin. Total N₂ production consists of N₂ from denitrification and N₂ from anammox. Total NO${}_{\mathrm{3}}^{-}$ reduction consists of total N₂ production, NH${}_{\mathrm{4}}^{+}$ from DNRA, and N₂O production. Total NO${}_{\mathrm{3}}^{-}$ uptake consists of total NO${}_{\mathrm{3}}^{-}$ reduction and other NO${}_{\mathrm{3}}^{-}$ sinks (e.g., intracellular storage).

Benthic denitrification rates exceeded anammox rates at all sampling sites (Fig. 2 and Table 2). This relationship agrees with the paradigm that denitrification is typically favored over anammox in organic-rich sediments (Dalsgaard et al., 2005; Devol, 2015). Anammox bacteria can reduce NO${}_{\mathrm{3}}^{-}$ to NO${}_{\mathrm{2}}^{-}$, which is then used to oxidize ammonia (NH₃) to N₂ (Kartal et al., 2007). In our in situ incubations, coupled DNRA–anammox in which DNRA produces a substrate (NH₃) required by anammox could result in the production of ³⁰N₂ (Prokopenko et al., 2006), which is accounted for by our rate calculation method (detailed in Sect. 2.2). However, because the porewater NH${}_{\mathrm{4}}^{+}$ concentration was high (>100 µM) (Yousavich et al., 2024), the fraction of ¹⁵N in the NH${}_{\mathrm{4}}^{+}$ pool remained low (up to 2.1 % after ∼1 h of incubation and up to 4.3 % after 6 h of incubation). Therefore, the contribution of anammox to ³⁰N₂ production was below 2.0 % (Table S3). Overall, anammox contributed up to 26.6 % of NO${}_{\mathrm{3}}^{-}$ reduction in the SBB (Table 2), indicating that anammox was a significant process in benthic SBB N cycling. Because the N isotope fractionation during the reduction of nitrite (NO${}_{\mathrm{2}}^{-})$ to N₂ by anammox bacteria ($+\mathrm{16.0}\pm \mathrm{4.5}\mathrm{‰}$) is lower than that of denitrification used for isotope mass balance calculations (∼25 ‰), anammox likely contributed to the lower-than-expected ¹⁵N enrichment in the SBB water column NO${}_{\mathrm{3}}^{-}$ pool previously measured (Brunner et al., 2013; Sigman et al., 2003). When NO${}_{\mathrm{3}}^{-}$ is not limiting, denitrification typically dominates as the denitrifier population has a shorter generation time than DNRA bacteria (Kraft et al., 2014).

3.3 NO${}_{\mathrm{3}}^{-}$ availability and TOC control the relative importance of DNRA

The contribution of DNRA to total NO${}_{\mathrm{3}}^{-}$ reduction was lower than denitrification at all stations except for the deepest station SDRO (Fig. 2), where NH${}_{\mathrm{4}}^{+}$ production by DNRA contributed more than half of the NO${}_{\mathrm{3}}^{-}$ reduction (Table 2). The relative contribution of DNRA to total NO${}_{\mathrm{3}}^{-}$ reduction was positively correlated with TOC in the top 2 cm of the sediment (Fig. 3a) and negatively correlated with the bottom water NO${}_{\mathrm{3}}^{-}$ concentration (Fig. 3b). These trends are consistent with previous findings showing that DNRA tends to be favored in environments with a high availability of electron donors such as organic carbon (Hardison et al., 2015; Kraft et al., 2014; Tiedje et al., 1983) and limited by NO${}_{\mathrm{3}}^{-}$ (van den Berg et al., 2015; Kessler et al., 2018; Peng et al., 2016). Another example where DNRA dominated under limited NO${}_{\mathrm{3}}^{-}$ availability is reported from measurements along a bottom water O₂ and NO${}_{\mathrm{3}}^{-}$ gradient traversing the Peruvian OMZ (Bohlen et al., 2011). One explanation for the increasing importance of DNRA under NO${}_{\mathrm{3}}^{-}$-limited conditions is that the growth yields calculated per mole electron acceptor from DNRA (consumes eight electrons) is higher than from denitrification (consumes five electrons) despite the greater amount of free energy provided by denitrification than DNRA per mole of NO${}_{\mathrm{3}}^{-}$, which was demonstrated by bacterial cultures capable of denitrification and DNRA (Strohm et al., 2007).

Figure 3 The correlation between the contribution of DNRA to NO${}_{\mathrm{3}}^{-}$ reduction (in %) and (a) the C : N ratio of sediment organic matter and (b) the bottom water NO${}_{\mathrm{3}}^{-}$ concentration in the Santa Barbara Basin. Linear regressions were performed excluding one outlier from station SDT3-A. The solid line represents the best fit, and the dashed lines represent the 95 % confidence interval band.

The regressions we performed between the relative importance of DNRA vs. TOC and bottom water NO${}_{\mathrm{3}}^{-}$ concentration excluded one data point from the station SDT3-A that deviated from the overall trend (Fig. 3). The DNRA rate at SDT3-A (0.14±0.005 mmol N m⁻² d⁻¹) was similarly low compared to NDT3-D (0.14±0.003 mmol N m⁻² d⁻¹), but the N₂ production rates by both denitrification and anammox were the highest among all stations (Fig. 2), resulting in the lowest relative importance of DNRA. Porewater sulfide concentration was high at SDT3-A (Yousavich et al., 2024), so the DNRA bacteria should not be limited by the availability of electron donors. Sediments at SDT3-A were characterized by the highest TOC and TON content among all sites (Table 1), which may have fueled the highest rates of denitrification and anammox (Middelburg et al., 1996; Devol, 2015).

The frequency and magnitude of seasonal anoxia in the SBB have been increasing in the past 4 decades, which is expected to intensify fixed N loss and NO${}_{\mathrm{3}}^{-}$ deficit in the water column (Goericke et al., 2015). Time-series measurements of water column NO${}_{\mathrm{3}}^{-}$ revealed that bottom water NO${}_{\mathrm{3}}^{-}$ depletion has become more frequent since 2003 compared to the time between 1986 and 2003. While seasonal flushing of the SBB not only oxygenates the bottom water but also increases bottom water NO${}_{\mathrm{3}}^{-}$, our results suggest that fixed N retention via DNRA will increase in response to NO${}_{\mathrm{3}}^{-}$ drawdown even before NO${}_{\mathrm{3}}^{-}$ is near depletion, which effectively forms a negative feedback that could potentially prevent the depletion of fixed N in the SBB. On the other hand, when NO${}_{\mathrm{3}}^{-}$ is no longer limiting, perhaps due to slowdown of bottom water deoxygenation, the relative importance of DNRA would decrease, allowing denitrification to dominate NO${}_{\mathrm{3}}^{-}$ reduction pathways.

3.4 N₂O production and saturation

N₂O production rates measured by in situ chamber incubations ranged from 4.8±1.1 to 38.8±5.6 µmol m⁻² d⁻¹ (Fig. 2). These rates were up to an order of magnitude higher than those measured using shipboard whole-core incubations (3.5±1.0 µmol m⁻² d⁻¹) with samples from a similar depth (544 m) in the anoxic part of the Soledad Basin (Townsend-Small et al., 2014). A recent study using in situ chamber incubations with ¹⁵NO${}_{\mathrm{3}}^{-}$ in the Eastern Gotland Basin reported rates (∼15–68 µmol m⁻² d⁻¹) similar to or higher than the rates we measured in the SBB (Hylén et al., 2022). Because the physicochemical context of the Soledad Basin is more similar to the SBB than the Eastern Gotland Basin, we expected the N₂O production rates in the Soledad Basin to be close to those in the SBB. The much lower N₂O production rates reported from the Soledad Basin may be partially attributed to the whole-core incubations that were not performed in situ.

N₂O production as a fraction of total NO${}_{\mathrm{3}}^{-}$ reduction ranged from 0.2 % to 1.5 % (Table 2), which fell in the typical range of N₂O yield from both nitrification and denitrification (Ji et al., 2015, 2018). Although our measurements do not allow the distinction between N₂O production from nitrification and denitrification, it is likely that both processes contributed with the respective share depending on ambient O₂ concentration. At the deepest stations where bottom water O₂ was depleted (Table 1), denitrification was likely the main source of N₂O. At other stations, where bottom water O₂ ranged from 3.1–9.2 µM, nitrification likely also contributed to N₂O production.

Although we observed N₂O production in all in situ ¹⁵NO${}_{\mathrm{3}}^{-}$ incubations, N₂O concentration in the chambers at the start of the incubations was far below saturation level (9 %–12 %) at the two deepest stations SDRO and NDRO (Table S4). In contrast, N₂O was either close to or above saturation at all other stations (Table S4). The low concentration of dissolved N₂O at the two deepest stations is consistent with our finding that N₂ production (i.e., N₂O consumption) rates by denitrification were the highest there (Fig. 2), indicating that the deepest part of the SBB typically acts as a sink for N₂O. The shallower parts of the SBB were characterized by a lower NO${}_{\mathrm{3}}^{-}$ uptake rate (Table S2; Fig. S1), but they had a stronger potential for N₂O production than the deepest stations (Fig. S2). In the case of a eutrophication event, enhanced surface primary productivity could stimulate denitrification as well as N₂O production in the shallower parts of the SBB where bottom water O₂ is not depleted and where benthic N₂O production is more likely to contribute to N₂O efflux from the water column during upwelling events.

3.5 Total NO${}_{\mathrm{3}}^{-}$ uptake suggests high potential for intracellular NO${}_{\mathrm{3}}^{-}$ storage

Although the N₂ production rates we measured in the SBB were among the highest reported values for any marine sediments, total NO${}_{\mathrm{3}}^{-}$ reduction, which also includes DNRA and N₂O production, only accounted for 23.9±16.9 % of the total NO${}_{\mathrm{3}}^{-}$ uptake in benthic flux chambers amended with ¹⁵NO${}_{\mathrm{3}}^{-}$ (Table 2). Intracellular NO${}_{\mathrm{3}}^{-}$ storage by bacteria and microbial eukaryotes was likely responsible for the majority of the NO${}_{\mathrm{3}}^{-}$ uptake unaccounted for by the different NO${}_{\mathrm{3}}^{-}$ reduction pathways. Marine Beggiatoa spp. can hyper-accumulate NO${}_{\mathrm{3}}^{-}$ intracellularly at concentrations 3000- to 4000-fold above ambient levels (McHatton et al., 1996). Other microbial lineages including Thioploca, foraminifera, and Gromiida are also known to store NO${}_{\mathrm{3}}^{-}$ intracellularly (Piña-Ochoa et al., 2010; Zopfi et al., 2001). In two of the porewater profiles sampled during the same cruise, NO${}_{\mathrm{3}}^{-}$ concentrations at 1 cm depth reached 80–390 µM, which we interpreted as evidence of NO${}_{\mathrm{3}}^{-}$ leakage from bacterial cells during porewater handling (Yousavich et al., 2024). While it is difficult to directly constrain the contribution of intracellular NO${}_{\mathrm{3}}^{-}$ storage to total NO${}_{\mathrm{3}}^{-}$ uptake, it can be indirectly inferred by calculating the diffusive loss (both upward and downward) of added ¹⁵NO${}_{\mathrm{3}}^{-}$ if porewater concentrations in sediments underlying the benthic flux chamber were available.

The total NO${}_{\mathrm{3}}^{-}$ uptake in the SBB measured from parallel benthic flux chambers without substrate amendment at the same stations (3.26±0.72 mmol N m⁻² d⁻¹) (Yousavich et al., 2024) was higher than that in other nearby borderland basins such as the San Nicolas Basin (0.38±0.03 mmol N m⁻² d⁻¹), the San Pedro Basin (0.78±0.11 mmol N m⁻² d⁻¹) (Berelson et al., 1987), and the Santa Monica Basin (1.10±0.31 mmol N m⁻² d⁻¹) (Jahnke, 1990). As mentioned above (Sect. 3.1), the addition of ¹⁵NO${}_{\mathrm{3}}^{-}$ stimulated NO${}_{\mathrm{3}}^{-}$ uptake rates by multiple folds (compared to BFC incubations without ¹⁵NO${}_{\mathrm{3}}^{-}$ additions) and to a level (11.60±4.15 mmol N m⁻² d⁻¹) similar to a previous estimate (11.7 mmol N m⁻² d⁻¹) based on water column NO${}_{\mathrm{3}}^{-}$ deficit (Valentine et al., 2016). Since bottom water NO${}_{\mathrm{3}}^{-}$ during our sampling time (>12.5 µM in November 2019) was not as depleted as in October 2013 (∼2 µM NO${}_{\mathrm{3}}^{-}$) (Valentine et al., 2016), these results indicate that the microbial community in SBB sediments have the metabolic potential to further consume NO${}_{\mathrm{3}}^{-}$ when SBB bottom water undergoes extended periods (months) of anoxia during autumn and winter. Assuming that NO${}_{\mathrm{3}}^{-}$ in the lowermost 10 m of the water column is under the direct influence of benthic NO${}_{\mathrm{3}}^{-}$ uptake, we estimate it would take between 1 to 4 months to deplete bottom water NO${}_{\mathrm{3}}^{-}$ with a starting concentration of 30 µM, with the shortest depletion time at the depocenter and the longest at the periphery of the SBB. This timescale agrees with time-series measurements of water column NO${}_{\mathrm{3}}^{-}$ concentrations in the SBB (Goericke et al., 2015), and it implies that bottom water NO${}_{\mathrm{3}}^{-}$ is unlikely to become depleted at depths shallower than 500 m. Furthermore, we identified a significant negative correlation between NO${}_{\mathrm{3}}^{-}$ uptake rates without substrate amendments and the fold change after ¹⁵NO${}_{\mathrm{3}}^{-}$ addition (Fig. S3). This negative correlation indicates that benthic NO${}_{\mathrm{3}}^{-}$ uptake rates at the shallow stations were the most responsive to exogenous NO${}_{\mathrm{3}}^{-}$ supply, while on the other hand NO${}_{\mathrm{3}}^{-}$ uptake rates at the deep and anoxic stations were closer to an upper limit that is determined by the microbial community present in the SBB sediments.

4 Summary

We investigated benthic nitrogen cycling processes using in situ incubations with ¹⁵NO${}_{\mathrm{3}}^{-}$ addition and quantified the rates of total NO${}_{\mathrm{3}}^{-}$ uptake, denitrification, anammox, N₂O production, and DNRA. Our results indicate the role of the SBB sediments as a strong sink for fixed N. Denitrification was the dominant NO${}_{\mathrm{3}}^{-}$ reduction process (38 %–76 %), while anammox contributed up to 27 %. DNRA accounted for less than half of NO${}_{\mathrm{3}}^{-}$ reduction except at the deepest station (586 m) at the center of the SBB, where bottom water O₂ concentrations were zero. The elevated relative importance of DNRA under high TOC and low NO${}_{\mathrm{3}}^{-}$ conditions suggests that the fixed N loss in the SBB, especially during seasons of high surface primary productivity and thus high export production, could potentially be balanced by the N retention pathway. The higher N₂O production measured in this study compared to nearby borderland basins may have stemmed from the use of benthic chambers instead of whole-core incubations, which highlights the advantage of in situ incubations in determining benthic N cycling processes. The high potential and relative importance of intracellular NO${}_{\mathrm{3}}^{-}$ storage implied by our data pose a challenge to fully constrain the fixed N budget in the SBB, but it also presents an opportunity for future investigations targeting intracellular storage. Future intensification of water column anoxia may elevate the importance of fixed N retention via DNRA by keeping N in the system as NH${}_{\mathrm{4}}^{+}$, forming negative feedback that could overall reduce fixed N loss in the SBB.