Dinitrogen Fixation Across Physico‐Chemical Gradients of the Eastern Tropical North Pacific Oxygen Deficient Zone

The Eastern Tropical North Pacific Ocean hosts one of the world's largest oceanic oxygen deficient zones (ODZs). Hot spots for reactive nitrogen (Nr) removal processes, ODZs generate conditions proposed to promote Nr inputs via dinitrogen (N2) fixation. In this study, we quantified N2 fixation rates by 15N tracer bioassay across oxygen, nutrient, and light gradients within and adjacent to the ODZ. Within subeuphotic oxygen‐deplete waters, N2 fixation was largely undetectable; however, addition of dissolved organic carbon stimulated N2 fixation in suboxic (<20 μmol/kg O2) waters, suggesting that diazotroph communities are likely energy limited or carbon limited and able to fix N2 despite high ambient concentrations of dissolved inorganic nitrogen. Elevated rates (>9 nmol N·L−1·day−1) were also observed in suboxic waters near volcanic islands where N2 fixation was quantifiable to 3,000 m. Within the overlying euphotic waters, N2 fixation rates were highest near the continent, exceeding 500 μmol N·m−2·day−1 at one third of inshore stations. These findings support the expansion of the known range of diazotrophs to deep, cold, and dissolved inorganic nitrogen‐replete waters. Additionally, this work bolsters calls for the reconsideration of ocean margins as important sources of Nr. Despite high rates at some inshore stations, regional N2 fixation appears insufficient to compensate for Nr loss locally as observed previously in the Eastern Tropical South Pacific ODZ.


Introduction
One of the major sources of reactive N (N r ) to the global ocean is dinitrogen (N 2 ) fixation (Gruber & Galloway, 2008), the assimilation of N 2 gas into biomass. Despite the abundance of N 2 in marine systems, only selected prokaryotic "N 2 fixers" (diazotrophs) have the genetic capacity to mediate its intracellular reduction to ammonia, which can then be assimilated via common metabolic pathways (Berges & Mulholland, 2008). Where present, these organisms can increase the N r pool and consequently stimulate primary production in N-limited ocean regions, thereby enhancing atmospheric drawdown of carbon dioxide (CO 2 ) and, potentially, export of this carbon (C) through the biological pump (e.g., Karl et al., 2012). Understanding the factors regulating N r inputs and losses is essential to predicting how oceanic N r inventories vary under changing climatic conditions and affect the ocean's capacity to take up CO 2 .
Pelagic ODZs account for roughly one third of N r loss from the ocean (DeVries et al., 2012). Significant N r deficits are observed in these regions relative to the concentrations of other dissolved constituents including phosphate (soluble reactive phosphate, SRP) and iron (Fe), an essential cofactor in the nitrogenase enzyme (Dixon & Kahn, 2004). As such, these waters are hypothesized to favor diazotrophic activity once advected into the euphotic zone, where the remaining DIN is rapidly depleted, by limiting the growth of competitors who cannot fix N 2 (Deutsch et al., 2007;Monteiro et al., 2011). In so doing, these geochemical signals (e.g., low DIN:SRP) are believed to play a critical role in the feedback mechanism between N 2 fixation and N r losses regulating the ocean's N r inventory (Weber & Deutsch, 2014); however, Fe limitation of N 2 fixation 10.1029/2019GB006242 Global Biogeochemical Cycles may spatially decouple N r inputs from losses (Weber & Deutsch, 2014), as hypothesized for the South Pacific basin Dekaezemacker et al., 2013;Knapp et al., 2016Knapp et al., , 2018Weber & Deutsch, 2014). The North Pacific receives higher aeolian Fe inputs than the South Pacific Ocean (Jickells et al., 2005), but N 2 fixation measurements in the ETNP ODZ region remain sparse despite local observations of diazotrophs (Jayakumar et al., 2017;White et al., 2013).
This study leveraged naturally occurring light, nutrient, and O 2 gradients in the ETNP to characterize N 2 fixation by mixed diazotroph communities with respect to these variables. The influence of dissolved organic C availability on N 2 fixation below the euphotic zone was investigated by amending whole water incubations with either glucose or a mixed amino acid solution. By furthering understanding of the physico-chemical factors regulating diazotroph activity, this study contributes to our evolving view of N 2 fixation in the marine environment and the feedback mechanisms maintaining the ocean's N r inventory.

Materials and Methods
We measured NFR, nutrient concentrations, and hydrographic characteristics within and adjacent to the ETNP ODZ aboard the National Oceanic and Atmospheric Administration vessel Ronald H. Brown in April 2016, during an El Niño-Southern Oscillation (ENSO) event (Climate Prediction Center, 2019). The cruise track extended southeast along the Mexican coastline from the Rosa Seamount (25°N, 115°W) off the Baja peninsula to 15°N and 99°W, proximal to the Guerrero-Oaxaca border, then offshore in a northwesterly direction to 18°N and 113°W (Figure 1a). "Inshore" was defined here as being within 200 km of the coastline; however, all stations were beyond the shelf break and were at least 2,000 m deep. Our sampling strategy was optimized for high-resolution data collection along vertical gradients of light, dissolved nutrients, and O 2 .

Hydrographic and Nutrient Measurements
Vertical profiles of temperature, salinity, O 2 , and chlorophyll a fluorescence were obtained at 16 stations using a Sea-Bird SBE 11plus CTD, equipped with a model 43 dissolved oxygen sensor (detection limit~3 μmol/kg O 2 ), a LI-COR Biospherical Photosynthetically Available Radiation (PAR) Sensor, and a Seapoint Chlorophyll Fluorometer. These instruments were mounted to a sampling rosette holding twenty-four 12-L Niskin bottles from which water samples were collected at selected depths to measure chlorophyll a and dissolved nutrient concentrations. Water for N 2 fixation incubations in the euphotic zone and below the ODZ was also collected from Niskin bottles (see below). Chlorophyll a concentrations were determined via the nonacidification method (Welschmeyer, 1994). NO 3 − plus nitrite (NO 2 − ) and SRP concentrations were measured onboard using an Astoria-Pacific nutrient autoanalyzer following standard colorimetric methods (Parsons et al., 1984) and according to the manufacturer's specifications. A Biosciences Ultrospec 2100 pro spectrophotometer was used for NO 2 − analysis (Parsons et al., 1984), and NO 3 − concentrations were calculated by difference. NH 4 + concentrations were determined fluorometrically using the

10.1029/2019GB006242
Global Biogeochemical Cycles orthophthaldialdehyde method (Holmes et al., 1999). DIN was calculated as the sum of NO 3 − +NO 2 − and NH 4 + concentration. Further details of these analyses are available in the supporting information Text S1.
In addition to discrete measurements, a Pump Profiling System, developed and built by the Monterey Bay Aquarium and Research Institute (Sakamoto et al., 1990), was used to generate near-continuous nutrient profiles in real time to approximately 350 m and to collect low-oxygen water. This system was composed of a cable, hose, and a small rosette to which a submersible water pump, a Sea-Bird SBE 19 SeaCAT CTD, WETStar Fluorometer, Sea Tech Beam Transmissometer, and a LI-COR Biospherical PAR Sensor were mounted. The Pump Profiling System was deployed to a maximum depth of 400 m. Water was pumped directly from depth to the laboratory to generate in situ nutrient profiles and to an on-deck station where samples from the O 2 minimum depth horizon were collected for incubation experiments (see below). In the laboratory, this flow ran first through an MBARI-modified Durafet pH sensor before being shunted to an Alpkem Astoria-Pacific Rapid Flow Analysis System which determined concentrations of NO 3 − plus NO 2 − , NO 2 − , and NH 4 + at a rate of one measurement per second. NO 3 − and NO 2 − measurements were made following manufacturer's instructions (Sakamoto et al., 1990). The orthophthaldialdehyde method (Holmes et al., 1999) was adapted for the Rapid Flow Analysis System and used to determine in situ NH 4 + concentrations. A Fast Repetition Rate Fluorometer collected chlorophyll a fluorescence profiles at a rate of one sample per 30 s.

N 2 Fixation Incubation Experiments
Whole water was collected from the Niskin bottles or pumped from depth anoxically using the Pump Profiling System for N 2 fixation incubation experiments carried out above, below or within the OMZ, respectively (Table 1). We define the OMZ as the region of the water column in which the lowest O 2 concentrations were observed at a site. This is distinct from the ODZ, operationally defined by O 2 concentrations below the limit of detection (LOD) of the Seabird O 2 sensor (3 μmol/kg) and NO 2 − concentrations exceeding 0.5 μM which has been deemed indicative of functionally anoxic conditions (Thamdrup et al., 2012).
NFR measurements were made using a modified version of the traditional 15 N 2 bubble method (Montoya et al., 1996). In the traditional 15 N 2 bubble method, 15 N 2 gas is injected into a filled sample bottle and 15 N enrichment of the particulate N (PN) pool is measured following an incubation period. The assimilation rate of the 15 N tracer into biomass (i.e., NFR) can then be calculated using a mixing model as shown below (equation (1); Montoya et al., 1996). N 2 gas is, however, slow to equilibrate, causing source pool enrichment (i.e., dissolved N 2 enrichment) to change over the course of the incubation which may result in an underestimation of NFR (Böttjer et al., 2017;Mohr et al., 2010). To address this issue, the dissolution of highly enriched 15 N 2 gas (~99%, Cambridge Isotopes, Tewksbury, MA) was hastened following injection by slowly inverting sample bottles on a large seesaw for 15 min. The seesaw consisted of a flat panel affixed to a central axis with baskets on either side in which incubation bottles were secured laterally. In this configuration, the 15 N 2 gas bubble traveled the full length of each bottle as the panel was gently rocked. Cambridge Isotope 15 N 2 gas was selected because, while significant 15 N-NH 4 + and -NO 3 − concentrations have been observed in other brand stocks, these contaminants have only been reported from Cambridge Isotope stocks at tracer-level concentrations (Dabundo et al., 2014). The remaining gas bubble was then Note. Note that not all "OMZ" samples exhibit suboxia (<20 μmol/kg). See Text S1 for a detailed description of sample handling. OMZ = O 2 minimum zone.

10.1029/2019GB006242
Global Biogeochemical Cycles removed using a syringe prior to the incubation period, and 15 N enrichment of the N 2 pool was measured directly (see below). 15 N 2 uptake experiments were carried out in triplicate. A detailed description of sample collection and incubation procedures is available in the supporting information Text S1.
Euphotic samples were incubated on-deck in tanks equipped with neutral density screens to approximate light levels at the depth of sample collection, which was determined using the PAR sensor mounted to the CTD rosette. Continuously flowing surface seawater maintained near-ambient surface temperatures in the deck incubators. Subeuphotic waters (below the 0.1% light level; Table 2) were incubated in the dark, in either a~12°C cold van or a 4°C walk-in refrigerator, depending on the temperature at the depth of sample collection. Due to a cold van malfunction, incubations from within the OMZ at Stations 8, 9, and 10 were carried out in a darkened room retrofitted with air conditioners, capable of maintaining temperatures at 16°C. While warmer temperatures may increase metabolic activity (Price & Sowers, 2004), the effect of this slight increase in temperature on N 2 fixation was likely minimal as rates from the affected bottles were largely undetectable (Table S2). For samples above and within the OMZ or ODZ, uptake experiments were 24 hr. Water samples collected below the OMZ or ODZ, where biomass is lower and cooler temperatures reduce metabolic rates (Price & Sowers, 2004), were incubated for 48 hr. Incubations were terminated by filtration of the sample onto precombusted (450°C for 2 hr) Whatman GF-75 filters (nominal pore size of 0.3 μm). Immediately prior to filtration, an aliquot of each sample to measure 15 N 2 enrichment in each bottle was transferred to a helium-flushed exetainer using a gas-tight syringe and preserved by adding 50 μl of a helium-flushed zinc chloride or mercuric chloride solution (50% w/v ZnCl 2 and HgCl 2 , Thermo Fisher Scientific, Waltham, MA).
To establish the initial isotopic composition and concentration of PN in water samples, water from each depth sampled was collected separately. Triplicate samples were filtered at the time of collection in a designated laboratory space isolated from where experiments using 15 N tracer were being conducted to avoid isotope contamination. Filters for both initial and final PN analysis were placed in sterile microcentrifuge tubes and frozen until analysis at Old Dominion University. Filters were dried for 48 hr at 50°C then pelletized in tin discs. PN and C concentrations and isotopic enrichment were measured using a Europa 20-20 isotope ratio mass spectrometer (IRMS) equipped with an automated N and C analyzer. Isotopic enrichment of dissolved N 2 was measured using a Europa 20-22 continuous flow IRMS, as described in Jayakumar et al. (2017).

NFR Calculations and Error Analysis
Volumetric NFR were calculated using equation (1): and [PN] represent the average initial and final PN isotopic composition, isotopic composition of the N 2 pool, and PN concentration of three replicate samples. The standard deviation of these replicates was propagated to calculate NFR error following Montoya et al. (1996) and Gradoville et al. (2017). The incubation time is denoted as Δt, and the uncertainty of this value was estimated based on the average time required to filter the given volume of seawater relative to the length of the incubation. A sensitivity analysis of the relative contributions of each measurement to total error is presented in Table S1. See supporting information Text S1 for further details on rate and error calculations.
A change in the atom percent enrichment of the PN pool was considered detectable if it exceeded 3 times the standard deviation of eight 12.5-μg N replicate standards which were run daily (Ripp, 1996). A LOD was calculated for each volumetric rate by substituting this value for A PN f −A PN0 . A limit of quantification was similarly calculated by taking 10 times the standard deviation of the replicate standards (Ripp, 1996). The average LOD for PN mass, calculated based on blanks (3σ, n = 4), among all IRMS runs for this study was 0.83 μg N. This value, however, should not be confused with the minimum mass deemed acceptable for determination of isotope ratios. The accuracy of enrichment measurements decreases with decreasing mass (Sharp, 2017), and so a lower limit of acceptable mass was established per instrument run based on the linearity of the atom percent measured during each standard run; on average this was 3.0 μg N for this study. The mass of standards ranged from 1.17 to 100 μg N.
Areal (depth-integrated) rates were calculated for the euphotic zone (defined as the region extending from the surface to the 0.1% light level; Table 2) by averaging volumetric rates measured for the surface mixed layer, from the bottom of the surface mixed layer to the 1% light level, and from the 1% to 0.1% light level, then depth integrating over each layer and summing the respective contributions. If the volumetric rate at a given depth was below the LOD, then it was assigned a value of 0 nmol N·L −1 ·day −1 for the purpose of calculating the areal rate. If the calculated NFR was above the LOD but below the limit of quantification (i.e., it was detectable but not quantifiable), then the LOD was used in the areal rate calculation. The error associated with both nondetectable and nonquantifiable rates was propagated along with that of all quantifiable rates. NFR were not depth integrated in subeuphotic waters because quantifiable rates were sparse and accurate estimation of subeuphotic areal NFR consequently untenable. Areal NFR from the euphotic zone were compared between inshore (1 to 9) and offshore (10 to 16) stations using a two-tailed Mann-Whitney U test.

Carbon Addition Bioassays
Carbon addition bioassays were carried out at Stations 6, 9, 12, 15, and 16 at depths within and below the core ODZ to determine whether the supply of organic C limited NFR. For these experiments, samples were collected as described above. Once sealed, but prior to 15 N 2 additions, one set of triplicate bottles was amended with glucose (Cambridge Isotopes, Tewksbury, MA) and another with a mixture of 20 amino acids (Cambridge Isotopes, Tewksbury, MA), resulting in a final addition of 40-μmol organic C/L to both sets of incubations. This addition approximately doubled the availability of dissolved organic C (Hansell & Carlson, 1998;Loh & Bauer, 2000), although much of the ambient dissolved organic C pool is thought to be very old and likely refractory (Druffel et al., 1992), thereby augmenting the significance of these C amendments to microbial communities present. Unamended triplicate incubations served as controls. Experimental treatments were compared to controls using a Wilcoxon signed-rank test. Where N 2 fixation was not detected or quantified, the lower limit of these values, zero or the LOD respectively, was used in statistical calculations. This approach increases our ability to detect significant differences (lower Type II error risk) but increases the probability of a false positive (higher Type I error risk).

Regional Hydrography
Surface waters of the ETNP are characterized by high productivity, a strong pycnocline that prevents local ventilation of deep waters, and a thermal front where the cool California Current meets the eastern Pacific warm pool (Fiedler & Talley, 2006). Our study area crossed this frontal region. Alongshore surface waters were cool to the north (<25°C at Stations 1 to 4) and warmer south of the Gulf of California (Table 2 and Figure 2a). Surface waters at Stations 9 to 13 exceeded 27.5°C, characteristic of the eastern Pacific warm pool (Fiedler & Talley, 2006), and salinity was low, below 34 (Table 2 and  These waters typically exhibit little seasonal and ENSO variability in sea surface temperature and salinity; however, ENSO events are associated with a local deepening of the thermocline (Fiedler & Talley, 2006). In the present study, the thermocline shoaled and strengthened alongshore to the southeast (Figures 3i  and 3j), following regional trends observed previously (Fiedler & Talley, 2006). The nitracline and primary chlorophyll a fluorescence maximum also shoaled to the southeast, accompanied by an increase in chlorophyll a fluorescence (Figures 3e-3h). A secondary chlorophyll a fluorescence maximum was evident at Stations 5 to 13, where the core ODZ was thickest and NO 2 − concentrations highest.
O 2 -deficient waters, defined as those in which O 2 concentrations were below detection by the Seabird O 2 sensor (<3 μmol/kg) and NO 2 − concentrations were above 0.5 μM (Thamdrup et al., 2012), were observed at all offshore stations (10 to 16) and inshore Stations 5 to 9 along the west coast of Mexico (Figures 1b  and 1c). The thickness of the ODZ increased toward the southeastern most station, expanding from 200 to 400 m at northerly Stations 15, 16, and 5, to a thickness of 100 to 600 m at Station 9. Stations 1 to 4 were north of the ODZ but exhibited suboxic (<20 μmol/kg O 2 ) conditions below the euphotic zone (Figures 3a-3d). The OMZ, the low-O 2 depth horizon present throughout the region, was characterized by slightly higher salinity (~34.7-34.8) and warmer temperatures (10-16°C) relative to underlying waters (S ≈ 24.6-34.5, T ≤ 10°C) and higher salinity (Table S2) than the California Current water intruding from the north (Fiedler & Talley, 2006). DIN:SRP ratios remained within a narrow range (8.5 to 12) from below the thermocline to approximately 400 m then gradually increased, stabilizing near 14 after 1,000 m. In the North Pacific Ocean, DIN:SRP ratios are typically~14.1, averaging 14.6 below 1,000 m (Tyrrell & Law, 1997).

N 2 Fixation in the Euphotic Zone
NFR were relatively high in the warm, sunlit, oxic waters (Figures 3k and 3l), where NO 3 − and SRP concentrations were low (Figures 3e, 3f, S1a, and S1b) and DIN:SRP was consistently below 3 (Figures S1c and S1d), suggestive of N r limitation. N 2 fixation was detected in euphotic waters at 10 of the 16 stations, in 17 of 62 samples collected (Table S2). Most of these samples came from Stations 1 to 8 where surface salinity and concentrations of chlorophyll a and SRP were comparatively high (Figures 2b, 2c, S1a, and S1b). NFR were highest and extended deeper into the water column where the thermocline was weakest, constricting toward the southeast as the underlying ODZ shoaled (Figures 3a, 3b, 3k, and 3l and Table S2). Depth-integrated euphotic NFR among inshore stations (1 to 9) were significantly greater than at offshore stations (10 to 16; Mann Whitney U, n 1 = 7, n 2 = 9, U = 93, p = 0.003), ranging from below detection to 6,230 ± 5,500 μmol N·m −2 ·day −1 (Table 2) Table 2 for values). White dots depict locations where N 2 fixation was undetectable throughout the euphotic zone.

10.1029/2019GB006242
Global Biogeochemical Cycles  Loescher et al., 2014;Sohm et al., 2011;White et al., 2013). Comparison of these systems to N rdeplete ocean gyres, where the bulk of global N 2 fixation has historically been ascribed, is, however, complicated by methodological biases-chiefly, the potential underestimation of the classic 15 N 2 bubble method due to the slow dissolution of 15 N 2 (Böttjer et al., 2017;Mohr et al., 2010), as well as underestimation or overestimation of N 2 fixation when measured indirectly via the acetylene reduction assay (Mulholland et al., 2004). Luo et al. (2012) calculated an arithmetic mean areal NFR for North Pacific (0°N to 55°N) euphotic zone of 120 ± 22 μmol N·m −2 ·day −1 based on studies applying these methods primarily in the basin's interior. Böttjer et al. (2017) accounted for the approximately twofold underestimation of the 15 N 2 bubble method in time series data from station ALOHA (North Pacific subtropical gyre) and calculated an average rate of 230 ± 136 μmol N·m −2 ·day −1 . These values are in relatively good agreement if a correction factor of 2 (Böttjer et al., 2017; is also applied to Luo and colleague's basin-wide estimate. Of the 10 stations where N 2 fixation was detected in surface waters for this study, areal rates exceeded the station ALOHA mean (Böttjer et al., 2017) at five-all located near the coastline (Table 2). If, however, we account for the error inherent to the areal rates presented in this study and variation from the mean at station ALOHA, only rates at Stations 1, 2, and 5 clearly exceed mean N 2 fixation at station ALOHA, and only at Stations 1 and 2 does N 2 fixation exceed the maximum rate at ALOHA where NFR range from 21 to 676 μmol N·m −2 ·day −1 (Böttjer et al., 2017). Nevertheless, the observed distribution of euphotic zone N 2 fixation in this study and the elevation of these rates above the regional average suggest that conditions in the cooler, saltier, more productive water mass near the continent favored the growth and activity of local diazotroph communities relative to those in the central North Pacific basin. This finding supports the observation of Jayakumar et al. (2017) that euphotic, inshore NFR in the ETNP ODZ region exceeded those offshore.
Euphotic NFR reported here (with a median value of 176 μmol N·m −2 ·day −1 ) were also high relative to waters overlying the ETSP ODZ, where Bonnet et al. (2013) and Fernandez et al. (2011) estimate average areal rates of 43 ± 6 and 48 ± 68 μmol N·m −2 ·day −1 , respectively, using the classic 15 N 2 bubble method. These values are in relatively good agreement with other incubation-based estimates from the ETSP euphotic zone Knapp et al., 2016) and are comparable to those observed in the central South Pacific gyre (94 ± 61 μmol N·m −2 ·day −1 ; Halm et al., 2012), indicating that surface waters of the ETSP ODZ do not support elevated NFR as predicted by Deutsch et al. (2007). Despite these low regional means, Loescher et al. (2014) observed N 2 fixation to exceed 800 μmol N·m −2 ·day −1 at one coastal sulfidic station in the ETSP, suggesting that sulfidic events may sporadically enhance rates. Nevertheless, N 2 fixation is thought to be only a minor contributor of N r to export production in the ETSP (Chang et al., 2019;Knapp et al., 2016). Note. Rate measurement propagated error, calculated as described in section 2, is given in parentheses.

10.1029/2019GB006242
Global Biogeochemical Cycles N 2 fixation may be greater in ETNP surface waters relative to those of the ETSP because the North Pacific receives higher aeolian Fe inputs (Jickells et al., 2005). Indeed, Fe amendments have been observed to increase NFR in ETSP surface waters . It must be noted, however, that the present study occurred during an ENSO event, which has been associated with enhanced NFR in the ETSP . Further work is therefore necessary to elucidate whether this event affected N 2 fixation in the ETNP and consequently whether these regional differences are truly significant.
High rates of N 2 fixation-on par with those observed here at Stations 1, 2, and 5 (Table 2)-have also been observed along other ocean margins (e.g., Carpenter et al., 1999;Mulholland et al., 2012Mulholland et al., , 2014Rees et al., 2009;Shiozaki et al., 2015;Wen et al., 2017) including areas characterized by low concentrations of DIN relative to SRP and other essential elements (e.g., Fe, silicon for diatom-diazotroph symbioses) resulting from riverine inputs (Grosse et al., 2010;Subramaniam et al., 2008) and advection of ODZ-generated N r -deplete waters into marginal seas (White et al., 2013). In the Amazon River plume and Gulf of California, the latter of which receives N r -deplete waters from the ETNP ODZ via the California Undercurrent, areal NFR as high as~8,000 μmol N·m −2 ·day −1 (Subramaniam et al., 2008) and~900 μmol N·m −2 ·day −1 (White et al., 2013) have been reported, respectively. Along the Southern New England shelf, NFR reached 4,106 μmol N·m −2 ·day −1 in Fall . In euphotic waters of the Benguela Upwelling System, another eastern boundary upwelling system predicted by Deutsch et al. (2007) to support elevated N 2 fixation, rates were far lower but increased inshore (Sohm et al., 2011) as observed in this study. These findings, along with the present study, suggest that diazotroph activity is heightened along ocean margins, particularly where N r is drawn down relative to SRP and other nutrients as occurs in the ETNP (Table 2).
Proximity to the continent likely offers advantages to some diazotrophic groups relative to others, depending on the physical and chemical properties of the water column, their metabolic requirements, and genetic capabilities. Enhanced availability of organic C in productive inshore waters may provide energy to heterotrophic diazotrophs and low-oxygen/anoxic microzones for noncyanobacterial diazotrophs fixing N 2 in fully oxic waters (Bombar et al., 2016), such as those observed in the ETNP and ETSP ODZs (Chang et al., 2019;Jayakumar et al., 2017). Indeed, sinking particles have been identified as loci of both cyanobacterial and noncyanobacterial diazotrophs in the North Pacific subtropical gyre (Farnelid et al., 2019). Additionally, Fe can be delivered to the ocean from the continents via aeolian transport and riverine inputs (Hunter & Boyd, 2007), the latter of which may also serve as a source of Fe-binding ligands (e.g., Bundy et al., 2015;Laglera & van den Berg, 2009) that maintain dissolved Fe in solution (Rue & Bruland, 1995;van den Berg, 1995).
Of the four stations (6, 9, 12, and 16) where NFR were measured in samples collected below the ODZ (≥500 m), rates were only detected at Stations 9 and 16 and were lower than measurable rates within the ODZ at these sites ( Figure 4). O 2 concentrations at these stations remained low, below 150 μmol/kg, even in deep waters (Table S2). At Station 9, a NFR of 0.87 ± 0.99 nmol N·m −2 ·day −1 (n = 3) was measured at 1,000 m; N 2 fixation was detectable but not quantifiable at 2,000 m. Rates at station 16 exceeded those measured elsewhere by about an order of magnitude (9.88 ± 3.85 nmol N·m −2 ·day −1 at 184 m and 35.9 ± 12.0 nmol N·m −2 ·day −1 at 200 m, n = 3), peaking at the top of the ODZ but remaining quantifiable to 3,000 m (0.36 ± 0.15 nmol N·m −2 ·day −1 , n = 3; Figure 4).
If low-O 2 conditions within the water column favor N 2 fixation, then one would expect NFR to be commensurately higher within and around the ODZ than in other oceanic environments. Both within and below the OMZ at all sites, however, NFR appeared patchy. Of the 51 subeuphotic samples collected, N 2 fixation was only detected in 10 (Table S2), and 5 of these were from one station, 16, located near the inner Revillagigedo Islands. NFR previously reported in oxygenated, aphotic marine waters are typically on the order of 1 nmol N·L −1 ·day −1 or lower (Moisander et al., 2017, and references therein). This rate is comparable with measurements presented here from within and below the ETNP OMZ including in ODZ waters, except for at station 10.1029/2019GB006242 Global Biogeochemical Cycles 16 (Table S2 and Figure 4), suggesting that suboxic/anoxic conditions alone do not result in elevated subeuphotic N 2 fixation in this region. This appraisal is consistent with work from Jayakumar et al. (2017) who previously observed few low rates (<1 nmol N·L −1 ·day −1 ) of N 2 fixation within the upper ETNP ODZ.
The patchy distribution of subeuphotic N 2 fixation presented here contrasts with some observations from the ETSP ODZ. In the ETSP, Bonnet et al. (2013) reported that low (<1 nmol N·L −1 ·day −1 ) but persistent rates below the euphotic zone account for~90% of total N 2 fixation. Similarly, Fernandez et al. (2011) estimated that the ODZ contributed about 5 times as much newly fixed N as oxic, euphotic waters. More recent work from Chang et al. (2019), which applied the same 15 N 2 bubble removal method as this study, diverges from these accounts. Chang et al. (2019) found no detectable NFR below the euphotic zone. This stark difference may be partially the result of methodological disparities, including the erroneous assumption in earlier ODZ work (e.g., Fernandez et al., 2011) that atom percent 15 N of the initial PN pool is equivalent to atmospheric values, which may inflate reported rate measurements (Chang et al., 2019;Voss et al., 2001). Whether the subeuphotic environment contributes significantly to total regional N r inputs also depends upon the manner in which detection limits are calculated given that even very small rates may be substantive when depth integrated through a deep water column. Regardless, NFR measurements in subeuphotic ETSP waters are largely within the range of those observed elsewhere in the ocean's interior (≤1 nmol N·L −1 ·day −1 ; Moisander et al., 2017, and references therein) and generally appear less patchy than those reported here from the ETNP.
Though NFR were undetectable throughout most of the ETNP OMZ, dissolved organic C additions of either glucose or dissolved free amino acids (DFAA) stimulated N 2 fixation within all OMZ waters surveyed, including within the ODZ at Stations 6 and 12 where NFR were otherwise undetectable (Figure 5a). This finding indicates that the ETNP OMZ hosts populations of diazotrophs capable of fixing N 2 in suboxic and anoxic waters, despite significant concentrations of ambient DIN. Additions largely failed to stimulate N 2 fixation below the OMZ (Figure 5b). As energy-rich molecules, DFAA and glucose may enter both Figure 4. Profiles of N 2 fixation at stations 6 (inshore), 9 (southerly), and 16 (northerly, offshore). Black stars, open circles, and filled circles represent rates that were below detection, detectable but nonquantifiable, and quantifiable. For rate measurements below detection, the limit of detection was plotted; for nonquantifiable rates, the limit of quantification was plotted. Error bars for quantifiable rates represent the propagated error. The shaded region represents where core oxygen deficient zone conditions (O 2 concentrations below detection and >0.5-μM NO 2 − ) occurred. A deep profile of N 2 fixation rates was also produced at station 12 (offshore); however, no N 2 fixation was detected at this site (Table S2). catabolic and anabolic pathways. Consequently, our findings could indicate that diazotrophic activity within the OMZ is limited by either energy or assimilable C. DFAA additions stimulated NFR at five of seven sampling locations within the OMZ, although this lacked statistical significance when compared across all sites (Wilcoxon signed rank, n 1 = n 2 = 7, p = 0.44), potentially due to the small sample size. Conversely, DFAA additions inhibited while glucose additions stimulated rates at certain depths at Station 16 ( Figure 5). Within the ETSP ODZ, Bonnet et al. (2013) found that DFAA additions stimulated N 2 fixation in all ODZ waters surveyed while mixed carbohydrate or glucose additions only stimulated N 2 fixation at one third of the stations surveyed. DFAA additions also enhanced N 2 fixation in the Red Sea (Rahav et al., 2013(Rahav et al., , 2015 and mesopelagic waters of the southwest Pacific Ocean (Benavides Figure 5. Results of carbon addition bioassay experiments within (a) and below (b) the O 2 minimum zone. Nitrogen fixation rates (NFR) marked with an open circle were detectable but not quantifiable. Stars indicate rates that were below the detection limit. For quantifiable rates, error bars express the propagated error. For DNQ rates, error bars express the limit of quantification calculated for that sample. Where no N 2 fixation was detected, the error bar marks the limit of detection. Neither additions of glucose nor dissolved free amino acids (DFAA) increased rates within (n 1 = n 2 = 7) or below (n 1 = n 2 = 8) the O 2 minimum zone significantly when compared across all sites (Wilcoxon signed rank, p > 0.05). ODZ = oxygen deficient zone.

10.1029/2019GB006242
Global Biogeochemical Cycles et al., 2015). These observations are perhaps surprising given that DFAA offer a source of N as well as C for heterotrophs and might thus be expected to suppress NFR. If, however, DFAA were used in catabolic rather than anabolic (assimilatory) processes, organic N might be excreted in the form of a waste product. We hypothesize that site-specific differences in microbial response to DFAA and glucose additions are due to population-specific substrate preferences, variability in energy/C limitation, or both. Regarding the response of diazotrophic communities below the OMZ, lower temperatures can slow bacterial metabolic rates (Price & Sowers, 2004) and decrease affinity for organic substrates (Nedwell, 1999). The response of any diazotrophs present in deep waters may consequently have been dampened within the timeframe of incubations, or there are no diazotrophs present and active at this depth.
Our findings suggest that diazotrophs inhabit the ETNP region and fix N 2 at elevated rates in euphotic waters near the coast but are largely inactive in OMZ and ODZ waters, as well as in the water column below, likely due to energy or C limitation at the OMZ depth horizon. In subeuphotic waters, detectable NFR were mostly on par with previous reports from oxygenated aphotic ocean waters (Moisander et al., 2017, and references therein). High NFR were, however, observed within suboxic waters at one station, 16, located near volcanic islands, suggesting that noncyanobacterial diazotrophs may thrive in DIN-replete waters given the right conditions. Here, NFR were highest in suboxic waters and decreased below the ODZ (Figure 4) following increased O 2 concentrations (Table S2) and decreased provision of organic C given that particulate C fluxes decrease with depth (Martin et al., 1987). High-carbon/low-oxygen conditions have previously been associated with increased activity of noncyanobacterial diazotrophs (Severin et al., 2015), and further investigation into how these factors may affect the range and magnitude of noncyanobacterial N 2 fixation is warranted.

Conclusions
Here, we present the most extensive data set to date of N 2 fixation above, within, and below the ETNP ODZ. Our observation of high rates (>500 μmol N·m −2 ·day −1 ) at inshore stations supports the growing body of evidence that ocean margins contribute a greater amount of newly fixed N than previously thought Tang et al., 2019). We hypothesize that continental inputs, inshore upwelling, or both alleviate growth limitation by an essential factor and play a key role in shaping the distribution of diazotrophs in N r -deplete waters regionally. Furthermore, we speculate that diazotrophs residing near the coast (this study) and in marginal seas (White et al., 2013) where N r -deplete waters upwell are important to basinscale compensation of N r deficits generated in the ETNP ODZ and merit further investigation.
This study also demonstrates that subeuphotic, suboxic waters in the region harbor diazotrophs capable of fixing N 2 despite high (>20 μmol N·L −1 ·day −1 ) ambient DIN concentrations, but their activity is highly patchy and appears C limited. While it is known that ambient DIN does not necessarily preclude N 2 fixation (Knapp, 2012), the observation of relatively high NFR (>9 nmol N·L −1 ·day −1 ) in DIN-replete deep waters challenges the prevailing hypothesis that N cycle homeostasis is maintained, in part, by the occurrence of N 2 fixation where N r is drawn down (Deutsch et al., 2007;Weber & Deutsch, 2014). Further work is needed to elucidate the response of diverse diazotrophic communities to DIN. Despite observation of patchy but high rates within and around the ETNP ODZ, N 2 fixation throughout the region appears too low to compensate for local N r losses. This finding adds to the mounting evidence that N r inputs and losses are spatially decoupled Knapp et al., 2016Knapp et al., , 2018Weber & Deutsch, 2014). By elucidating the distribution of N 2 fixation across physico-chemical gradients in an undersampled but biogeochemically important region, this study contributes to our evolving understanding of the factors that regulate marine N 2 fixation and the ocean's N r inventory.