Nitrogen cycling in the deep sedimentary biosphere : nitrate isotopes in porewaters underlying the oligotrophic North Atlantic

Nitrogen (N) is a key component of fundamental biomolecules. Hence, its cycling and availability are central factors governing the extent of ecosystems across the Earth. In the organic-lean sediment porewaters underlying the oligotrophic ocean, where low levels of microbial activity persist despite limited organic matter delivery from overlying water, the extent and modes of nitrogen transformations have not been widely investigated. Here we use the N and oxygen (O) isotopic composition of porewater nitrate (NO−3 ) from a site in the oligotrophic North Atlantic (Integrated Ocean Drilling Program – IODP) to determine the extent and magnitude of microbial nitrate production (via nitrification) and consumption (via denitrification). We find that NO−3 accumulates far above bottom seawater concentrations (∼ 21 μM) throughout the sediment column (up to ∼ 50 μM) down to the oceanic basement as deep as 90 m b.s.f. (below sea floor), reflecting the predominance of aerobic nitrification/remineralization within the deep marine sediments. Large changes in the δN and δO of nitrate, however, reveal variable influence of nitrate respiration across the three sites. We use an inverse porewater diffusion–reaction model, constrained by the N and O isotope systematics of nitrification and denitrification and the porewater NO−3 isotopic composition, to estimate rates of nitrification and denitrification throughout the sediment column. Results indicate variability of reaction rates across and within the three boreholes that are generally consistent with the differential distribution of dissolved oxygen at this site, though not necessarily with the canonical view of how redox thresholds separate nitrate regeneration from dissimilative consumption spatially. That is, we provide stable isotopic evidence for expanded zones of co-occurring nitrification and denitrification. The isotope biogeochemical modeling also yielded estimates for the δN and δO of newly produced nitrate (δNNTR (NTR, referring to nitrification) and δONTR), as well as the isotope effect for denitrification (εDNF) (DNF, referring to denitrification), parameters with high relevance to global ocean models of N cycling. Estimated values of δNNTR were generally lower than previously reported δN values for sinking particulate organic nitrogen in this region. We suggest that these values may be, in part, related to sedimentary N2 fixation and remineralization of the newly fixed organic N. Values of δONTR generally ranged between −2.8 and 0.0 ‰, consistent with recent estimates based on lab cultures of nitrifying bacteria. Notably, some δONTR values were elevated, suggesting incorporation of O-enriched dissolved oxygen during nitrification, and possibly indicating a tight coupling of NH+4 and NO − 2 oxidation in this metabolically sluggish environment. Our findings indicate that the production of organic matter by in situ autotrophy (e.g., nitrification, nitrogen fixation) supplies a large fraction of the biomass and organic substrate for heterotrophy in these sediments, supplementing the small organic-matter pool derived from the overlying euphotic zone. This work sheds new light on an active nitrogen cycle operating, despite exceedingly low carbon inputs, in the deep sedimentary biosphere. Published by Copernicus Publications on behalf of the European Geosciences Union. 7484 S. D. Wankel et al.: Nitrogen cycling in deep-sea porewaters

the canonical view of how redox thresholds separate nitrate regeneration from dissimilative consumption spatially.That is, we provide stable isotopic evidence for expanded zones of co-occurring nitrification and denitrification.The isotope biogeochemical modeling also yielded estimates for the δ 15 N and δ 18 O of newly produced nitrate (δ 15 N NTR (NTR, referring to nitrification) and δ 18 O NTR ), as well as the isotope effect for denitrification ( 15 ε DNF ) (DNF, referring to denitrification), parameters with high relevance to global ocean models of N cycling.Estimated values of δ 15 N NTR were generally lower than previously reported δ 15 N values for sinking particulate organic nitrogen in this region.We suggest that these values may be, in part, related to sedimentary N 2 fixation and remineralization of the newly fixed organic N. Values of δ 18 O NTR generally ranged between −2.8 and 0.0 ‰, consistent with recent estimates based on lab cultures of nitrifying bacteria.Notably, some δ 18 O NTR values were elevated, suggesting incorporation of 18 O-enriched dissolved oxygen during nitrification, and possibly indicating a tight coupling of NH + 4 and NO − 2 oxidation in this metabolically sluggish environment.Our findings indicate that the production of organic matter by in situ autotrophy (e.g., nitrification, nitrogen fixation) supplies a large fraction of the biomass and organic substrate for heterotrophy in these sediments, supplementing the small organic-matter pool derived from the overlying euphotic zone.This work sheds new light on an active nitrogen cycle operating, despite exceedingly low carbon inputs, in the deep sedimentary biosphere.

Introduction
Below the surface ocean, the dark ocean, including environments above and below the seafloor, hosts the largest habitable environment on the planet and is home to a wide range of globally relevant biogeochemical processes (Orcutt et al., 2011).While significant progress has been made in recent years toward characterizing the geological, chemical, and ecological composition of a variety of subsurface environments (Orcutt et al., 2011;Edwards et al., 2011Edwards et al., , 2012b)), the potential for impact of these systems on global biogeochemical cycles remains poorly understood.Most of our knowledge about subseafloor microbial activity stems from research focusing on productive continental margins, where relatively high fluxes of organic matter from surface primary productivity support a large heterotrophic and mostly anaerobic microbial community (e.g., Blair and Aller, 2012).By comparison, vast areas of the seafloor, in particular those underlying ocean gyres, characterized by low primary productivity and low organic-matter flux to the sea floor, have received far less attention (D'Hondt et al., 2009;Mason et al., 2010;Fischer et al., 2009).In contrast to well-studied ocean margin sediments, oxygen (O 2 ) and nitrate (NO − 3 ), two powerful oxidants of organic carbon, penetrate deeply into the sediment underlying oligotrophic ocean waters (D'Hondt et al., 2009(D'Hondt et al., , 2015;;Murray and Grundmanis, 1980;Rutgers van der Loeff et al., 1990;Sachs et al., 2009;Røy et al., 2012;Fischer et al., 2009).Furthermore, in the sediments overlying relatively young and permeably ocean crust, O 2 and NO − 3 are also supplied via upward diffusion from oxic and nitrate-replete fluids flowing through basaltic basement as has been shown for the North Pond site, which is located on the western flank of the Mid-Atlantic Ridge (Orcutt et al., 2013;Ziebis et al., 2012).At North Pond, where the sediment cover is thin (< ∼ 25 m), O 2 penetrates the entire sediment column; where sediment thickness is elevated, conditions become anoxic at mid-depths.Aerobic heterotrophic respiration likely dominates organic carbon oxidation in the upper sediment column.However, as organic carbon becomes limiting at greater depths, autotrophic processes (e.g., nitrification) are likely to gain relative importance.Further, there is evidence that the upward supply of oxidants from the basaltic basement supports increased microbial activity (Picard and Ferdelman, 2011).However, a fundamental understanding of the relative importance of specific metabolic activities that drive and sustain subsurface communities is lacking.Because central ocean gyres cover roughly half of the global seafloor, understanding the nature of the biosphere hosted within these sediments may provide important insights into its role in global marine nitrogen and carbon cycling.Here we focus specifically on elucidating subsurface nitrogen cycling and its role in supporting heterotrophic and autotrophic processes in oligotrophic deep-ocean sediments underlying the North Atlantic Gyre, at North Pond (22 • 45 N, 46 • 05 W).
Integrated Ocean Drilling Program (IODP) Expedition 336 (Mid-Atlantic Ridge Microbiology, 16 September-16 November 2011) aimed to directly address the nature of microbial communities in both ocean crust and sediments at North Pond, a characteristic sediment-filled 70 km 2 depression surrounded by high relief topography common to the western flank of the Mid-Atlantic Ridge (Becker et al., 2001;Langseth et al., 1992).While a majority of seafloor subsurface biosphere research has focused on aspects of sedimentary carbon, sulfur, and iron cycles, the potential role of N in supporting subsurface microbial activity has been largely unexplored.Despite exceedingly oligotrophic conditions, life persists and evidence for active heterotrophic and autotrophic microbial communities in North Pond sediments is mounting (Ziebis et al., 2012;Picard and Ferdelman, 2011;Orcutt et al., 2013).
Nitrogen plays a central role as a limiting nutrient in many regions of the sunlit surface ocean (Rabalais, 2002), as nearly 90 % of the biologically available fixed N in the ocean resides below the euphotic zone in the deep ocean NO − 3 reservoir (Gruber, 2008).Globally, sediments (especially continental shelves) are considered a net sink of fixed nitrogen through reductive anaerobic processes including denitrification and anaerobic ammonium oxidation (Christensen et al., 1987;Devol, 1991;Prokopenko et al., 2013).For example, coupled nitrification (the chemolithotrophic oxidation of NH + 4 to NO − 3 ) and denitrification (the typically heterotrophic reduction of NO − 3 to N 2 ) have been shown to be important in N budgets in sediments of continental shelves, ocean margins, and estuaries (Risgaard-Petersen, 2003;Granger et al., 2011;Lehmann et al., 2004Lehmann et al., , 2005;;Wankel et al., 2009).There is abundant evidence demonstrating the importance of both oxidative and reductive N cycling processes (and their tight coupling) operating in sediment environments.In contrast to sediments on continental shelves, however, data from sediments underlying large swaths of the oligotrophic ocean suggest an entirely different framework.For example, NO −  Fawcett et al., 2015).Nitrate-removal processes (whether assimilatory or dissimilatory) have been shown to impart linearly coupled increases in both N and O isotope ratios of the remaining NO − 3 pool (Karsh et al., 2012;Kritee et al., 2012;Granger et al., 2004Granger et al., , 2008)).In contrast, however, nitrification, the two-step oxidation of NH + 4 to NO − 2 followed by NO − 2 oxidation to NO − 3 , represents a decoupling of the N and O isotope systems in the resulting nitrate (Casciotti and McIlvin, 2007;Buchwald and Casciotti, 2013;Wankel et al., 2007).Whereas the N atoms derive from the NH + 4 (which can be assumed to originate from the sedimentary organic nitrogen pool), the oxygen atoms derive, to varying degrees, from both water and dissolved O 2 (Buchwald and Casciotti, 2010;Buchwald et al., 2012;Casciotti et al., 2010).Thus, by combining isotope mass balances of both N and O in the NO − 3 system, along with our understanding of organismlevel constraints on the isotope systematics of these transformations, we can deduce the relative roles of multiple N cycling processes (e.g., Wankel et al., 2009;Lehmann et al. 2004;Bourbonnais et al., 2009).Here we use the dual isotopic composition of nitrate (N and O isotopes) as a record of microbial processes occurring in the low-carbon sediments of North Pond underlying the oligotrophic North Atlantic gyre.By combining the N and O isotope mass balance with an inverse reaction-diffusion model approach, we use these data to estimate rates of nitrification and denitrification, and to provide new constraints on some isotope parameters for these processes.

Sediment and porewater collection
Sediment cores were collected at three sites in the North Pond Basin (Fig. 1) as part of the IODP Leg 336 expedition and have been described extensively elsewhere (Expedition-336-Scientists, 2012).Four boreholes were drilled (U1382B, U1383D, U1383E, and U1384A, referred to hereafter as 2B, 3D, 3E, and 4A).Sites 3D and 3E were next to each other and as drilling logs indicated that the core from 3E showed excessive signs of disturbance upon retrieval and potential contamination by seawater, it was excluded from our study.Sediment cores were retrieved using the advanced piston corer, which penetrated the seafloor sediments until contact with basement, followed by extended core barrel coring of the upper section of basement rock.Site 2B (∼ 90 m sediment thickness, depth to basement) is located in the deeper part of the pond, approximately 25m away from DSDP (Deep Sea Drilling Project) legacy hole 395A, which was instrumented as a CORK (Circulation Obviation Retrofit Kit) observatory (Davis et al., 1992).Site 3D (∼ 42 m sediment thickness, depth to basement) lies in the northeastern region towards the edge of North Pond (∼ 5.9 km away from U1382A), whereas site 4A (∼ 95 m sediment thickness, depth to base- ment) is located on the northwest side in a deeper part of the basin, approximately 3.9 and 6.2 km in distance from U1383 and U1382, respectively (Fig. 1).A total of 2 decades of temperature and flow records revealed vigorous subsurface flow (Becker et al., 1984;Gable et al., 1992), with geothermal surveys indicating that recharge is from the southeastern side of the basin (near 2B and 3D) flowing to the northwest (towards 4A) (Fig. 1).All sediments were comprised of light-brown to brown nannofossil ooze with intercalations of foraminiferal sand.In the lowest few meters close to the sediment/basement contact, sediments exhibited a darker brown color and sometimes rust-colored clay-rich zones (Edwards et al., 2012a;Expedition-336-Scientists, 2012).
Porewater samples for concentration and stable isotope analyses were collected either directly from cores on the shipboard catwalk immediately after core retrieval (and stored at −80 • C until analysis), from whole core rounds (∼ 10 cm core sections) that had been preserved at −80 • C for ∼ 1 year until thawing and porewater extraction, or from subsampled sediments collected shipboard and stored at −20 • C for ∼ 9 months.Porewaters from whole core rounds (∼ 40 mL) and subsampled sediments (∼ 3 mL), were extracted using Rhizon samplers (0.2 µm) (Seeberg-Elverfeldt et al., 2005) and stored frozen (−20

Concentration measurements
Concentrations of NO x (i.e., NO − 3 plus NO − 2 ) from shipboard-extracted porewaters were measured via ion chromatography ∼ 3 months after collection, while concentrations from home-laboratory-extracted porewaters were measured by chemiluminescence after reduction in a hot acidic vanadyl sulfate solution on a NO x analyzer (Braman and Hendrix, 1989) (detection limit < 0.5 µM).Concentrations of NO − 2 were quantified by using the Griess-Ilosvay method followed by measuring absorption at 540 nm (Grasshoff et al., 1999) or by chemiluminescence in a sodium iodide solution on a NO x analyzer (Garside, 1982;Cox, 1980).NO − 3 was quantified by the difference between NO x and NO − 2 (Grasshoff et al., 1999).Ammonium concentrations were measured using the orthophthaldialdehyde fluorescence method with an empirically determined sensitivity limit of ∼ 0.5 µM for porewaters (Holmes et al., 1999).

Nitrate stable isotope composition
Nitrate N and O isotopic composition were measured using the denitrifier method (Casciotti et al., 2002;Sigman et al., 2001), in which sample NO − 3 is quantitatively converted to N 2 O using a lab-grown denitrifying bacterium before being extracted and purified on a purge-and-trap system similar to that previously described (McIlvin and Casciotti, 2010).Where detected (O 2 -depleted zones of 2B), NO − 2 was removed by sulfamic acid addition (Granger and Sigman, 2009) prior to isotopic analysis of the NO − 3 .Isotopic analysis of shipboard-extracted samples (16) were conducted at the University of Basel using a Delta V Advantage (Thermo, Inc.), while all other samples (29) were measured on an Iso-Prime100 (Elementar, Inc.).Corrections for drift, size and fractionation of O isotopes during bacterial conversion were carried out as previously described using NO − 3 standards USGS 32, USGS 34, and USGS 35 (Casciotti et al., 2002;McIlvin and Casciotti, 2011), with a typical reproducibility of 0.2 and 0.4 ‰ for δ 15 N and δ 18 O, respectively.

Previous measurements of dissolved oxygen and organic matter content at North Pond
Oxygen penetration depths, which have been discussed previously (Orcutt et al., 2013), vary distinctly among the three sites at North Pond indicating much greater respiratory consumption in 2B than in the profiles of other two sites, 3D and 4A (Fig. 2).In 2B, dissolved oxygen levels are drawn down to near detection by a depth of about 10 m b.s.f.(below sea floor) (although low levels of dissolved O 2 seem to persist as deep as 30 m b.s.f.).In contrast, at site 3D, dissolved O 2 levels are drawn down close to the detection limit (∼ 5 µM) for an interval of only ∼ 3m between a depth of ∼ 30 and ∼ 33 m b.s.f. and in 4A, zero-level O 2 concentrations were observed over the interval between 32 and 54 m b.s.f.At North Pond, O 2 (and NO − 3 ) is also supplied via diffusion from the underlying basaltic crustal aquifer (Fig. 2) (Orcutt et al., 2013;Ziebis et al., 2012).Although not measured during this work, sediment organic carbon and nitrogen content was measured on several of the piston cores collected during the site survey, averaging 0.15 and 0.02 %, respectively (Ziebis et al., 2012).Although quantification of such low organic-N and organic-C levels is challenging and is afflicted with relatively large uncertainties, no discernible organic matter elemental differences among the sites were noted (Ziebis et al., 2012).Similarly, no differences in organic phosphorus content were noted among the three sites (Defforey and Paytan, 2015).

NO −
3 and NO − 2 concentration profiles Bottom seawater NO − 3 concentration at North Pond is approximately 21.6 µM (Ziebis et al., 2012).At all depths in all three profiles, porewater NO − 3 concentrations exceeded bottom water NO − 3 concentrations, reflecting the production of NO − 3 by nitrification and the net flux of NO − 3 to the overlying water from this site of ∼ 4.6 µmoles m −2 d −1 (Ziebis et al., 2012), consistent with other studies of NO − 3 fluxes in pelagic deep-sea sediments (Berelson et al., 1990;Goloway and Bender, 1982;Hammond et al., 1996;Jahnke et al., 1982;Grundmanis and Murray, 1982).More precisely, below the sediment-water interface, NO − 3 concentrations increased significantly with depth (Fig. 2), before decreasing again with proximity to the basement/sediment contact.Midprofile NO − 3 concentration maxima reached 38.2, 42.2, and 49.1 µM at depths of 19.1, 23.0, and 56.3 m b.s.f. in the cores from sites 2B, 3D, and 4A, respectively -depths that generally coincided with O 2 concentrations below 10 µM.Nitrite was below detection at sties 3D and 4A and was only detected at anoxic depths in site 2B (Fig. 2), where concentrations of up to 6.0 and 6.6 µM were observed at depths of 36 and 59 m, respectively.

NO − 3 N and O isotopic composition
Down-core changes in δ 15 N and δ 18 O varied markedly among the three cores (Fig. 2).The most prominent changes in both δ 15 N and δ 18 O were observed at site 2B (which had the most extensive O 2 -depleted zone), in which δ 15 N increased with depth from a value of +5.4 ‰ (bottom seawater) to a maximum of +23.3 ‰ at a depth of 59.2 m b.s.f. and δ 18 O increased from a bottom seawater value of +1.8 ‰ to a maximum of +23.8 ‰ at a much shallower depth of 32.1 m b.s.f.Isotopic maxima generally coincided with depths of lowest O 2 concentrations, except in 3D, where the maximum was observed at slightly greater depth than the O 2 minimum (Fig. 2).Substantial N and O isotopic shifts were also observed at site 3D, in which increases above bot- ).The red circle at the top of the profiles denotes the bottom seawater NO − 3 concentration of 21.6 µM (Ziebis et al., 2012).Horizontal black lines indicate depth of contact with ocean crust.Gray boxes indicate "transitional" zones in which a reaction-diffusion model is used to calculate co-occurring nitrification and denitrification (see text for details).Strong increases in δ 15 N and δ 18 O in U1382B coincide with depths having the lowest O 2 concentration and are indicative of the influence of denitrification.While the NO − 3 concentrations profiles of U1383D and U1384A appear similar, stark differences in the nitrate dual isotopic composition reflect the generally low level of microbial activity in U1384A.tom water NO − 3 values to δ 15 N of +11.8 ‰ and a δ 18 O of +21.7 ‰ were observed, with both maxima occurring at a depth of 37 m b.s.f.In contrast, site 4A exhibited only very modest changes relative to bottom seawater, with a maximum δ 15 N value of +7.0 ‰ at a depth of 38.8 m b.s.f. and a maximum δ 18 O value of +6.3 ‰ at a depth of 44.1 m b.s.f.Strong correlations were also observed between δ 15 N NO 3 and δ 18 O NO 3 (Fig. 3), with δ 18 O NO 3 values always increasing more than the corresponding δ 15 N − NO 3 .The relationship between δ 18 O NO 3 and δ 15 N NO 3 exhibited a slope of 1.8 for the upper portion of the 2B profile, 3.0 for the 3D profile and 2.4 for the 4A profile -consistently exceeding the 1 : 1 relationship expected from NO − 3 consumption alone.In 2B, sampling points near the most O 2 -depleted depths and the lower portion of the profile fell closer to the expected 1 : 1 line for NO − 3 consumption by denitrification (Fig. 3).

Discussion
The distribution of porewater nitrate in deep-sea sediments is controlled by the combined influence of diffusion as well as several biologically catalyzed diagenetic processes including nitrification (ammonia and nitrite oxidation to nitrate) and denitrification (loss of N via nitrate reduction to gaseous products, NO, N 2 O or N 2 ).Here we use the concentration and dual N and O stable isotope composition of porewater NO − 3 to gain insight into the magnitude and distribution of N transformation processes.In comparison to models that predict the rates of these processes based solely on concentration profiles of NO − 3 and O 2 , for example, our approach estimates rates using the added constraints provided by recent studies of N and O isotope systematics of nitrification and denitrification (Granger et al., 2008;Buchwald and Casciotti, 2010;Casciotti et al., 2010;Buchwald et al., 2012).In particular, while there are strong and related N and O isotope effects during denitrification (Granger et al., 2008), the isotopic transformations of N and O are decoupled due to differently sourced N and O atoms during nitrate production (Buchwald and Casciotti, 2010;Casciotti et al., 2010;Buchwald et al., 2012).Thus, changes in N and O isotopic composition between intervals of any one depth are the combined result of both diffusion of NO − 3 to/from overlying (and underlying) seawater, together with microbially mediated production and/or consumption of NO − 3 within porewaters.Under low oxygen, NO − 3 respiration by denitrification leads to a well-characterized increase in both δ 15 N and δ 18 O in conjunction with decreasing NO − 3 replete) porewaters also suggest a high potential for denitrification, which can be verified using nitrate dual isotope measurements.In this way, our modeling approach provides an assessment of the distribution of these N transformations, as well as some additional insights on the nature of N and O source atoms to NO − 3 in these energy-lean systems.Specifically, we show below that the profiles of δ 15 N NO 3 and δ 18 O NO 3 can be explained by variations in the magnitude of nitrification and denitrification occurring throughout the sediment column, including substantial zones of overlap of these aerobic/anaerobic processes.Finally, we use our model to predict the δ 18 O and δ 15 N stemming from nitrate production by nitrification, offering insights into both the nature of processes setting the O isotopic composition of oceanic NO − 3 , as well as the sources of N and/or the isotopic partitioning of available N sources in global ocean sediments.

Diffusion-reaction model
The diffusion-reaction inverse modeling approach used here is conceptually similar to other early diagenetic models that simulate porewater profiles of dissolved species through a sediment column harboring both oxic and anoxic organic matter remineralization (Christensen and Rowe, 1984;Goloway and Bender, 1982;Jahnke et al., 1982).It is an inverse modeling approach adapted to distinguish between heavy and light nitrate isotopologues (e.g., Lehmann et al., 2007).Specifically, we use the model to estimate rates of nitrification and denitrification required to fit the concentration profiles of each isotopologue, 14 N 16 O − 3 , 15 N 16 O − 3 , and 14 N 18 O 16 O − 2 (and, thus, δ 15 N NO 3 and δ 18 O NO 3 values) under the assumption of steady-state diffusion and microbial production (by nitrification) and/or consumption (by denitrification).Rates of nitrification and denitrification in each porewater sampling interval (e.g., defined as the distance between the lower and upper midpoints between sampling depths) were estimated numerically by least-squares fitting of the system of equations describing the distribution of each isotopologue (using a genetic algorithm included in the Solver package of Microsoft Excel 2011).This approach involves determination of a non-unique solution using numerical iteration and optimization, and is repeatedly iterated to evaluate robustness of model fits.Certain parameters are allowed to be optimizable by the model, including both the magnitude of, and connection between, the N and O isotope effects for denitrification ( 15 ε DNF and 18 ε DNF : 15 ε DNF , respectively), as well as the N and O isotopic composition of new NO − 3 produced by nitrification (δ 15 N NTR and δ 18 O NTR , respectively).Uncertainty in model estimates is expressed as the standard error of 10 model-run estimates (Table 1).Conditions at the uppermost part of the sediment column were constrained by measured concentrations and isotope ratios in bottom seawater.Measured concentrations of the NO − 3 isotopologues within each interval, together with the diffusive fluxes defined by the concentration gradients between the over/underlying intervals, were used for model fitting by least-squares optimization of microbial rates of nitrification and/or denitrification.
As measurable NH + 4 was not observed at any depths, it is not included in the model.NO − 3 is the only dissolved N species included in the model and we assume that all      NH + 4 generated by remineralization is completely oxidized to NO − 3 (see below).To minimize complexity, other diagenetic reactions that may be important in many sedimentary environments, including anaerobic NH + 4 oxidation, removal of N species through interactions with Fe or Mn and the adsorption and retention of NH + 4 by clay minerals are not specifically addressed.We also neglect effects of compaction as well as potential changes in organic matter reactivity with depth.No difference in the diffusivity among NO − 3 isotopologues was included, since these differences are considered to be very small (Clark and Fritz, 1997).
Resolving the vertical dimension only, the mass balance differential equations are as follows: such that for denitrification (DNF): where D refers to the molecular diffusion coefficient for NO − 3 adjusted for porosity, DNF and NTR refer to the reaction rate of denitrification or nitrification, respectively (in mass volume −1 time −1 ), C refers to the concentration of each isotopologue (in mass volume −1 ) and k refers to the firstorder rate constant (time −1 ).The fractionation factor, α, is defined as the ratio of rate constants for the light isotope over the heavy isotope (e.g., 15 α= 14 k/ 15 k) for a given process, alternatively expressed in terms of epsilon where ε = (α − 1) × 1000 in units of permil (‰).The term 18 ε : 15 ε DNF refers to the degree of coupling between the N and O isotope fractionation during denitrification, with a value of 1 indicating that the two isotope effects are identical.
For the δ 15 N and δ 18 O of nitrification (NTR) where f refers to the fractional abundance of a particular isotopologue and and and 15 R NTR and 18 R NTR used to calculate the δ 15 N NTR and δ 18 O NTR , respectively.
For parameterizing diffusion, we use a porewater diffusion coefficient (D s ) based on the molecular diffusion coefficient (D m at 5 • C) for NO − 3 of 1.05 × 10 −5 cm 2 s −1 (Li and Gregory, 1974) adjusted for an average porosity (ϕ) of North Pond sediments of 64 % (Expedition-336-Scientists, 2012), where D s = ϕ k D m and k is an empirically derived factor (we use 2.6) accounting for tortuosity of pore space (Hammond et al., 1996;McManus et al., 1995).
Compared with contemporaneous profiles of O 2 and Sr (Orcutt et al., 2013) and other dissolved ions, the NO − 3 concentration profiles suffer from some apparent analytical noise.The nature of the heterogeneity for NO − 3 concentration measurements was unclear.However, it is unlikely that this heterogeneity is environmental and we attribute it to small amounts of evaporation during freezer storage of the sediments, which is supported by the apparent smoothness of the isotopic measurements (evaporation would change the apparent concentrations without influencing the isotopic composition of solutes such as NO − 3 ).As such, for the purpose of the diffusion-reaction model, we applied a 5-point-weighted triangular smoothing to the concentration data to eliminate outliers and unrealistically sharp gradients (Fig. 2).Given the relatively smooth and contiguous vertical profiles of δ 15 N NO 3 and δ 18 O NO 3 , only very minor smoothing to these data was required using a similar approach (Fig. 2).
Within this model architecture, we explore the influence of four key parameters that could affect the estimation of nitrification and denitrification rates by this isotope mass balance approach, in particular, 15 ε DNF , 18 ε : 15 ε DNF , δ 15 N NTR , and δ 18 O NTR .Specifically, the expression of the full enzymaticlevel isotope effect ( 15 ε DNF ) for denitrification (27 ‰) can be influenced by electron donor, carbon substrate quality, denitrification rate and metabolic activity (Kritee et al., 2012).Moreover, although the relationship between the kinetic isotope effects for 18 O and 15 N during respiratory consumption of NO − 3 by denitrification (e.g., 18 ε : 15 ε DNF ) has been shown to remain consistent at 1:1, the potential influence of nitrate reduction by periplasmic nitrate reductase (Nap, periplasmic nitrate reductase.),which imparts a lower 18 ε : 15 ε DNF value of 0.6 (Granger et al., 2008;Frey et al., 2014), could play a role in the dual isotope trajectory of NO − 3 consumption (Wenk et al., 2014).Further, in the absence of NH + 4 accumulation in these sediments, the δ 15 N NTR is equal to the source of NH + 4 being oxidized to NO − 3 , which is related to the δ 15 N of the organic matter being remineralized.The δ 18 O NTR stems from a combination of factors including the δ 18 O of the water and dissolved O 2 as well as the expression of kinetic isotope effects associated with the incorporation of O atoms from these pools (Buchwald and Casciotti, 2010;Casciotti et al., 2010;Andersson and Hooper, 1983).Below, we use the model to optimize and predict these values and to explore the sensitivity of rate estimates to 15 ε DNF .
The model contains more parameters than can be explicitly estimated from the small number of data points measured.To minimize the number of variables as much as possible (and maximize the utility of the approach for constraining other variables), we adapt the model implementation for three different O 2 regimes: (1) oxic intervals where O 2 is poised as the more energy-yielding oxidant with respect to NO − 3 (here generally O 2 > ∼ 40 µM) and in which only nitrification is allowed to occur, (2) transitional intervals in which both denitrification and nitrification may occur (O 2 between ∼ 40 and 2 µM) and (3) anoxic intervals where O 2 is < 2 µM and in which only denitrification is allowed to occur.We choose 40 µM as a boundary for the onset of denitrification based on estimates of thermodynamic energy yield under these conditions for aerobic vs. denitrification-based respiration (Brewer et al., 2014).In the oxic intervals, the model is used for parameter estimation of both δ 15 N NTR and δ 18 O NTR (in addition to nitrification rate), while in the anoxic intervals the model is used to estimate 15 ε DNF and 18 ε : 15 ε DNF (in addition to denitrification rate).In transitional intervals, 15 ε DNF and 18 ε : 15 ε DNF are held constant at 25 ‰ and 1, respectively, and the parameter δ 15 N NTR and δ 18 O NTR are estimated through model fitting, together with rates of both nitrification and denitrification.In accordance with previous experimental work (Buchwald and Casciotti, 2010;Buchwald et al., 2012;Casciotti et al., 2010), the allowed range of values for δ 15 N NTR and δ 18 O NTR were set to −5 to +10 and −5 to +20 ‰, respectively.When 15 ε DNF and 18 ε : 15 ε DNF were determined by model fitting, parameter estimates were allowed to range between 0 and 30 ‰ and between 0.6 and 1.2, respectively.

Model-estimated rates of nitrification and denitrification
Profiles of sedimentary porewater solutes reflect the combined influence of many processes including diagenetic reactions, which are intimately related to the availability, abundance and quality of organic carbon.In particular, the distribution of dissolved substrates that are available as electron acceptors for microbial respiration of organic carbon, generally reflect stepwise consumption by the most thermodynamically (and kinetically) favorable metabolic processes (e.g., O 2 consumption precedes NO − 3 consumption, which precedes sulfate reduction).While in organic-rich estuarine and continental shelf sediments, dissolved O 2 and NO − 3 are typically consumed within a few centimeters or millimeters below the sediment-seawater interface, sediments underlying large areas of the oligotrophic ocean are characterized by very deep penetration of O 2 , in some cases even penetrating to the underlying ocean crust (D'Hondt et al., 2009(D'Hondt et al., , 2015;;Orcutt et al., 2013;Ziebis et al., 2012).In connection with this deep penetration of O 2 , deep-sea sediment porewaters also often exhibit extensive accumulation of NO − 3 above ambient seawater concentrations, associated with the oxidation of NH + 4 released by aerobic remineralization of sediment organic matter (and linked to the consumption of O 2 through Redfield stoichiometry) (Berelson et al., 1990;Christensen and Rowe, 1984;D'Hondt et al., 2009;Goloway and Bender, 1982;Seitzinger et al., 1984).In organic-rich sediments, NO − 3 concentration profiles may exhibit maxima only a few millimeters or centimeters below the sedimentwater interface.In contrast, in the deep-sea sediments underlying the oligotrophic regions of the ocean, the sedimentary zone where NO − 3 accumulates to 10 to 30 µM above bottom seawater concentrations can extend over a much larger vertical extent and nitrate maxima can be found tens of meters below the seafloor.In effect, the redox zonation of O 2 respiration, NH + 4 oxidation, NO − 2 oxidation, and NO − 3 reduction extends over larger depth ranges, and, depending on sediment thickness and organic carbon content -the redox state of these sediments may never reach the potential for NO − 3 reduction to play a role as a thermodynamically viable metabolic pathway.
While it is not necessarily apparent whether any NO − 3 respiration is occurring based on North Pond concentration profiles alone, dramatic increases in the δ 15 N and δ 18 O of NO − 3 with depth into the anoxic sediment intervals were observed in both 2B and 3D (Fig. 2 2).The marked distinction between 2B/3D and 4A notwithstanding, the dual isotopic composition and concentrations of NO − 3 in these porewaters reveal active nitrogen cycling within all three sites.
Using these changes in dual NO − 3 isotopic composition and concentration, we calculated the rates of nitrification and denitrification necessary to produce the observed patterns within each interval (in the transitional intervals, here we prescribe a value of 25 ‰ for 15 ε DNF , but explore the model sensitivity to this value later).Rates of nitrification and denitrification varied with depth, as well as across the three sites (Fig. 4; Table 1).Estimated rates of nitrification in the oxic and transitional intervals were up to 871 nmol cm −3 yr −1 , while rates of denitrification in the anoxic and transitional intervals reached up to 579 nmol cm −3 yr −1 (Fig. 4; Table 1).By comparison, maximum rates of nitrification estimated from O 2 and NO − 3 profiles in sediments below the South Pacific gyre, perhaps the most organic-matter-depleted seafloor sediments in the world, were predicted to be only as high as 18-74 nmol cm −3 yr −1 (D' Hondt et al., 2009).The estimated nitrification rates at North Pond are still orders of magnitude lower than those typically measured in coastal and estuarine sediments (Rysgaard et al., 1993;Usui et al., 2001;Wankel et al., 2011), and are, thus, consistent with lower total cell abundances and lower abundance of functional genes involved in nitrification recently measured in North Pond sediments (Zhao and Jørgensen, 2015) as compared to typical estuarine sediments (Mosier and Francis, 2008;Wankel et al., 2011).
In general, nitrification rates in the oxic intervals near the seafloor were comparable to rates in the oxic layers near the underlying crust.The highest denitrification rates typically coincided with depths where the lowest O 2 was observed, with the exception of rather low rates near the central anoxic zone of core 2B.Nitrification, which requires O 2 , is observed as deep as 28 m in 2B and all the way to the underlying crust at the other two sites.Interestingly, modeled rates of nitrification were typically the highest at comparatively low levels of dissolved O 2 (∼ 15 µM in 2B, ∼ 10 µM in 3D and ∼ 35 µM in 4A) -suggesting an important role for micro-aerophillic nitrification (e.g., organisms adapted to respiration under low O 2 conditions).The depths of maximum denitrification rates generally coincided with the onset of O 2 levels below ∼ 2 µM.In 2B, denitrification rates were the highest at depths of ∼ 28 and ∼ 72 m (Fig. 4).In 3D, maximum denitrification rates were observed at 21 m -with similar rates between 20 and 25 m.Interestingly, the model indicated no denitrification within the 3 m anoxic zone of this core -possibly indicating limitation by organic substrate availability.In contrast, rates of denitrification were only estimated to occur within the anoxic zone of site 4A (e.g., not within the transitional intervals, although this was somewhat sensitive to prescribed 15 ε DNF , see below), with the highest rate of 24 nmol cm −3 yr −1 at a depth of 44 m.Overall, rates of both nitrification and denitrification were highest at site 2B and lowest in 4A, consistent with a greater amount of microbial activity revealed by the steeper O 2 gradients in 2B and recent functional gene quantification in NP sediments (Zhao and Jørgensen, 2015).More precisely, the abundance of 16S rRNA, and several functional genes, including those for both archaeal and bacterial ammonia oxidation, nitrite oxidation, and denitrification, were all significantly lower in 4A than in the other cores (Zhao and Jørgensen, 2015).In all three cores, maximum rates of nitrification exceeded those of denitrification, consistent with the net accumulation of NO − 3 throughout the sediment column.Even in 2B, where O 2 is below 2 µM over an interval of ∼ 40 m, the NO − 3 concentration profile exhibits no obvious influence by NO − 3 reduction (Fig. 2).Finally, the model suggests the co-occurrence of nitrification and denitrification (Fig. 4) in the transition zones of our model (depths at which O 2 is between 2 and 40 µM).Although denitrification rates generally did not exceed those of nitrification where the two processes are co-occurring (i.e., no net nitrate consumption), the increasing δ 15 N and δ 18 O of NO − 3 clearly reflects the influence of NO − 3 loss via denitrification.Indeed, similar inferences have also been made about such overlap of nitrification and denitrification albeit in the much deeper and organic-rich hadopelagic sediments of the Ogasawara Trench (Nunoura et al., 2013).This observation illustrates the exceptionally extended vertical redox zonation of these sediments -and highlights the potential interaction between nitrogen transformations that are classically considered spatially explicit.

Model-predicted values of δ 15 N NTR : implications for N sources and processes in oligotrophic sediments
In general, where NH + 4 from remineralization does not accumulate, it is expected that the δ 15 N of NO − 3 produced by NH + 4 oxidation will be equivalent to the δ 15 N of the NH + 4 deriving from remineralization of organic nitrogen.The δ 15 N of organic N of the North Pond sediments was not quantified in this study (N content as measured by Ziebis et al. (2012) within the upper 8 m was < 0.03 %).Yet, the average modelpredicted δ 15 N of newly produced NO − 3 (δ 15 N NTR ) (Fig. 5) ranged from −3.1 to +1.1 ‰ (standard error typically ±0.3 to 0.4 ‰), generally lower than that expected based on the δ 15 N of sinking organic matter from the surface ocean of ∼ +3.7 ‰ (Altabet, 1988(Altabet, , 1989;;Knapp et al., 2005;Ren et al., 2012).Values of δ 15 N NTR >+1 ‰ were only observed just above the O 2 minimum at sites 2B and 3D (Fig. 5).
Confronted with this difference, we turn to other factors that might play a role in setting the δ 15 N NTR .The lower δ 15 N values of newly produced NO − 3 can potentially be explained by a number of possible processes including: (1) isotopic fractionation during remineralization, ( 2 branching between NH + 4 oxidation (whether anaerobic or aerobic) and NH + 4 assimilation or (3) contribution of low δ 15 N through organic matter derived from sedimentary N 2 fixation.
Nitrogen isotope fractionation during organic matter remineralization has been reported (Altabet, 1988;Altabet et al., 1999;Estep and Macko, 1984;Lehmann et al., 2002), whereby the preferential remineralization of low δ 15 N organic matter leads to production of low δ 15 N NH + 4 (which could feed into the production of low δ 15 N NO − 3 ).The influence of this phenomenon is more likely, however, during remineralization of fresh organic matter and where the heterotrophic community has abundant access to highly labile proteinaceous organic matter (Altabet, 1988;Estep and Macko, 1984).At North Pond, given the extremely low levels of organic material present in the sediments, it seems unlikely that preferential utilization of low δ 15 N organic material during diagenesis is responsible for the low δ 15 N NTR .
Competitive branching of NH + 4 supporting simultaneous nutritional supply (as an N source) and energy supply (via autotrophic ammonia oxidation) has been used to explain NO − 3 dual isotopic signatures in N-rich surface waters (Wankel et al., 2007).Because N isotope fractionation during ammonia oxidation is generally thought to be stronger than that of NH + 4 assimilation by phytoplankton under surfacewater conditions, it was argued that this competitive branching can lead to a shunt of low δ 15 N into the NO − 3 pool via ammonia oxidation (Wankel et al., 2007).In contrast to the high-nutrient, sunlit, productive waters of Monterey Bay, however, under the energy-limited, and extremely low-NH + 4production conditions in North Pond porewaters, it is not clear whether the same mechanisms of NH + 4 partitioning are operating.
Although North Pond porewaters contain abundant NO − 3 , assimilation of NO − 3 as a nutritional N source requires an associated metabolic energy for reduction of NO − 3 (via an assimilatory nitrate reductase), a costly process in this energypoor environment.On the other hand, although NH + 4 is more easily assimilated by most microbes, its low abundance in North Pond porewaters reflect its scarcity as a source of N required for cell growth.Based on the estimated values of the N isotope effect for NO − 3 consumption (by denitrification) in the porewaters of 2B that average ∼ 20 ‰ (Table 2), there is little suggestion of NO − 3 assimilation, which would lead to much lower estimated values of 15 ε DNF (Granger et al., 2010).Thus, it is more likely that nutritional N originates from a reduced form such as NH + 4 and/or organic N. If the N isotope effect for ammonia oxidation is much larger than that of ammonia assimilation, then such a competitive branching effect may also be contributing to a low δ 15 N NTR .Although under such low concentrations, it is likely that microbial acquisition of NH + 4 (whether for assimilation or for oxidation) is diffusion-limited, conditions under which the N isotope effect is expected to be near 0 ‰ (Hoch et al., 1992); a strong difference in the isotope effects for assimilation and oxidation of NH + 4 could still contribute to the production of low δ 15 N nitrate.A final possibility for the low values of δ 15 N NTR could reflect the relative importance of benthic N 2 fixation operating in North Pond sediments.Bacterial N 2 fixation is generally thought to result in biomass having a δ 15 N between −2 and 0 ‰ (Delwiche et al., 1979;Meador et al., 2007;Minigawa and Wada, 1986), which could effectively introduce new N and decrease the bulk δ 15 N available for oxidation.Benthic N 2 fixation is not generally considered to be an important enough contributor to the total sediment organic matter to influence the bulk δ 15 N values of sediment organic matter.However, given the low organic-matter flux to these sediments from the overlying oligotrophic surface waters, a proportionately smaller amount of N 2 fixation would be required to significantly impact the sediment organic δ 15 N value.While N 2 fixation is an energetically costly metabolism and might seem an unlikely strategy given the abundant porewater NO − 3 pool, it has been recently acknowledged that N 2 fixation in benthic environments may be widely underestimated, despite high levels of porewater DIN including NO − 3 and/or NH + 4 (Knapp, 2012).In fact, N 2 fixation could be ecologically favored in organic-lean sediments like those at North Pond owing to the formation of H 2 as an end product, which might afford some community level.Although primarily recycled by highly efficient hydrogenases in N-fixing bacteria (Bothe et al., 2010), a small efflux of H 2 could help to fuel other autotrophic metabolisms including both NO − 3 reduction (Nakagawa et al., 2005) or the Knallgas reaction (H 2 + O 2 ), perhaps providing some mutualistic benefit.Hydrogen-based metabolism has been proposed as a significant contributor to subsurface autotrophy underlying the oligotrophic South Pacific gyre (D'Hondt et al., 2009).Although the involvement of so-called alternative nitrogenases (the Fe and V forms), which have been shown to display an even larger kinetic isotope effect (−6 to −7 ‰) than the Mo-bearing form (Zhang et al., 2014), could offer even greater leverage on lowering of the bulk δ 15 N (and source of N for nitrification), their involvement in non-sulfidic marine systems, where Mo is re-plete and soluble Fe and V is scarce, is expected to be minimal (Zhang et al., 2014).A similar argument was also made for the cryptic involvement of N 2 fixation as source of low δ 15 N and explanation for dual NO − 3 isotopic patterns in the large oxygen minimum zone of the eastern tropical North Pacific (Sigman et al., 2005).Thus, we conclude that the low predicted values of δ 15 N NTR provide compelling evidence for an important role of in situ N 2 fixation in these organiclean sediments.
Finally, there is a conspicuous increase in the predicted δ 15 N NTR at sites 2B and 3D (up to +1.8 and +1.1 ‰, respectively) between 15 and 35 m.Although we have no compelling explanation for these observations, it is interesting that these values coincide with the transitional intervals over which δ 18 O NO 3 values increase much more rapidly than δ 15 N NO 3 .While it is possible that our model is insufficient, it may also be that these suboxic depths support differential amounts of in situ nitrogen fixation leading to shifts in the bulk δ 15 N available for oxidation by nitrification.
The contribution of competitive branching during NH + 4 consumption notwithstanding, the predicted δ 15 N NTR values can serve as an index of the degree of organic matter deriving from in situ N 2 fixation versus delivery from the overlying water (δ 15 N ∼ +3.7 ‰).Indeed, using a value of −2 ‰ to represent organic matter derived from biological N 2 fixation (Delwiche et al., 1979;Meador et al., 2007;Minigawa and Wada, 1986), our results suggest that between 25 and 100 % (with an average of 80 ± 20 %) of the organic nitrogen supply in these sediments from biological nitrogen fixation, representing a potentially enormous relative role for in situ autotrophy in sustaining these microbial communities.Under an assumption of steady state, and ignoring a potential contribution of competitive branching in NH + 4 consumption, rates of N 2 fixation were estimated as a fraction of nitrification and ranged up to 345 nmol cm −3 yr −1 (Table 1).The distribution of N 2 fixation rates were generally similar to nitrification -with much lower rates (8-45 nmol cm −3 yr −1 ) at site 4A -again consistent with overall lower cell abundance (Zhao and Jørgensen, 2015).These rates are 2-3 orders of magnitude lower than rates measured in coastal and estuarine sediments (e.g., Bertics et al., 2010;Rao and Charette, 2012;Joye and Paerl, 1993), although still much higher than rates measured in studies of N 2 fixation in oligotrophic water column (e.g., Montoya et al., 2004;Capone et al., 2005).Overall, North Pond sediments appear to harbor a spectrum of microbially mediated N transformations, with rates lower than those found in most sedimentary systems, yet still generally higher than those observed in overlying oligotrophic waters.Thus, while the influence of both sediment hosted N 2 fixation and competitive branching during NH + 4 consumption may not be mutually exclusive, our analysis places important upper limits on the nature of autotrophic lifestyles (including N 2 fixation) and the nitrogen economy in the deep subsurface.

Model-predicted values of the δ 18 O of nitrification (δ 18 O NTR )
We also observed variation in estimated values of the δ 18 O of newly produced NO − 3 (δ 18 O NTR ), ranging from −2.8 to as high as +4.1 ‰ (at the O 2 minimum in 3D), which may offer some insight into the nature of nitrification in these sediments and the deep ocean in general.The oxygen isotope composition of newly produced NO − 3 reflects the combination of several complex factors including (1) the δ 18 O of the ambient water and dissolved O 2 , (2) kinetic isotope effects during the enzymatically catalyzed incorporation of O atoms during oxidation of NH + 4 and NO − 2 , as well as (3) the potential influence of oxygen isotope equilibration between water and NO − 2 (Casciotti et al., 2010).In the upper profile of site 2B and throughout the profile of site 4A, predicted values of δ 18 O NTR clustered between −2.8 and 0.0 % with no clear trends related to downcore concentrations of O 2 or NO − 3 (Fig. 6).Although slightly lower, this range of values agrees remarkably well with values predicted by experiments using a co-culture of NH + 4 and NO − 2 oxidizing bacteria, which ranged from −1.5 to +1.3 ‰ (Buchwald et al., 2012).In systems where NH + 4 and NO − 2 oxidizing bacteria co-exist and are not substratelimited, NO − 2 does not generally accumulate and the importance of oxygen isotope equilibration between NO − 2 and water can be considered minor (∼ 3%) (Buchwald et al., 2012).In this case, the δ 18 O NTR is primarily set by the δ 18 O of water (seawater δ 18 O ∼ 0 ‰) and dissolved O 2 (∼ +26.4 ‰ for the deep N. Atlantic; Kroopnick et al., 1972) and the three kinetic isotope effects during the sequential oxidation of NH + 4 to NO − 3 (Buchwald et al., 2012;Casciotti et al., 2010).The resulting δ 18 O NTR can be described as where 18 ε is the kinetic isotope effect of O atom incorporation from O 2 during NH + 4 oxidation to NH 2 OH ( 18 ε O 2 ), and from water during NH 2 OH oxidation to NO − 2 ( 18 ε H 2 O,1 ) and NO − 2 oxidation to NO − 3 ( 18 ε H 2 O,2 ) (Buchwald et al., 2012).While the value of seawater δ 18 O can be considered to be relatively constant (∼ 0 ‰) in North Pond porewaters, the respiratory consumption of O 2 , as evident in the observed concentration profiles, imparts a relatively strong isotopic fractionation (Bender, 1990;Kroopnick and Craig, 1976;Lehmann et al., 2009) and will cause elevated δ 18 O O 2 values.Using a separate reaction-diffusion model (not shown), we estimated the δ 18 O O 2 to be as high as +70 ‰ where concentrations of O 2 have been drawn down > 95 % of the level found in bottom seawater.Incorporation of this highly 18 Oenriched O 2 by nitrification in these low O 2 intervals may contribute to observed increases in δ 18 O NTR predicted by our model.In particular, in the low O 2 intervals of 2B and 3D, δ 18 O NTR values as high as +4.1 ‰ at the four intervals coinciding with the maximum O 2 drawdown (Fig. 6), and thus may point to incorporation of high-δ 18 O O 2 .For example, assuming a δ 18 O value of 0 ‰ for seawater, using Eq. ( 11) and a combined isotope effect of 18 ‰ for the two steps of NH + 4 oxidation to NO − 2 ( 18 ε O 2 + 18 ε H 2 O,1 ; the two have not yet been resolved from one another; Casciotti et al., 2010) and a value of 15 ‰ for 18 ε H 2 O,2 (O atom incorporation during NO − 2 oxidation to NO − 3 ; Buchwald et al., 2012), δ 18 O NTR values of −2 ‰, +2, or +6 ‰ would imply incorporation of O from an O 2 pool with a value of ∼ +45, +57, or +69 ‰, respectively.
If these higher values are indeed the result of enriched O 2 incorporation, then they also provide indirect information on the degree of oxygen isotope equilibration occurring between NO − 2 and water.Specifically, if some proportion of an elevated δ 18 O O 2 signal is propagated into the NO − 3 pool, then this suggests that the intermediate NO − 2 pool did not completely equilibrate with ambient water (which would effectively erase all signs of precursor molecule δ 18 O).Within these low O 2 transitional intervals in 2B and 3D, it appears that the turnover of the very small NO − 2 intermediate pool may be faster than the time required for complete equilibration between NO − 2 and water (Buchwald and Casciotti, 2013).In contrast, the low O 2 interval from site 4A does not exhibit elevated δ 18 O NTR values near the oxygen minimum, perhaps suggesting that the turnover of NO − 2 here is slower (allowing complete equilibration) or that equilibration is biologically catalyzed (e.g., enhanced by enzymatic activity; Buchwald et al., 2012).Although the concentrations of the NO − 2 pool were generally below detection, making the accurate determination of its turnover time impossible (via δ 18 O), the use of NO − 2 oxygen isotopes as an independent measure of metabolic processes where concentrations persist at measurable levels may be a potentially powerful indicator of biological turnover of NO − 2 (δ 15 N and δ 18 O of NO − 2 in 2B, where NO − 2 was detected at two depths, were not determined as part of this study).Future studies should target this pool as a complementary dimension for constraining subsurface biosphere metabolic rates.

Model sensitivity to prescribed 15 ε DNF in transitional intervals
In the transitional intervals, where both nitrification and denitrification are allowed to co-occur, the model is underdetermined and requires some variables to be prescribed.We chose to prescribe a value for the kinetic isotope effect of denitrification ( 15 ε DNF ) and here examine the sensitivity of estimated rates nitrification and denitrification, as well as predicted values of δ 15 N NTR and δ 18 O NTR .Given the rather tightly confined range of determined values for 15 ε DNF in the anoxic zone of site 2B, averaging 20 ± 1.8 ‰, for illustration, we bracket our prescribed 15 ε DNF in transitional intervals with values of 15 and 25 ‰ (rate estimates where the prescribed 15 ε DNF is as low as 5 ‰ are given in Table 1).Overall, the model-predicted rates of nitrification and denitrification, as well as values of δ 15 N NTR and δ 18 O NTR were largely insensitive to changes in the prescribed strength of the isotope effect for denitrification ( 15 ε DNF ) (Fig. 4).Specifically, when the prescribed value of 15 ε DNF decreased from 25 to 15 ‰ , changes in the predicted values of δ 15 N NTR and δ 18 O NTR were generally small (Figs. 5 and  6), varying by a maximum of 0.9 ‰ (average 0.5 ‰) and 2.3 ‰ (0.6 ‰), respectively.An exception to this is the intervals bracketing the anoxic zone of the profile at site 2B (at depths of 27.9 and 70.8, 72.9 m), which yielded predicted δ 15 N NTR that were either 2.1 ‰ lower (at 27.9 m) or ∼ 5 ‰ higher (at 70.8 and 72.9 m).Predicted δ 18 O NTR values were also quite sensitive to 15 ε DNF in this interval with values that were 3.8 ‰ (at 27.9 m) and 7.2-8.0‰ higher (at 70.8 and 72.9 m).While we cannot rule out the potential influence of changes in physiological expression of isotope effects, the sensitivity of δ 15 N NTR and δ 18 O NTR to 15 ε DNF at these depths may point to an unresolvable artifact of this model approach.Further work being indicated, incorporation of dual nitrite isotopes could certainly aid in resolving this apparent sensitivity.However, this sensitivity was not observed in the other transitional intervals of 2B, 3D, or 4A and conclusions regarding δ 15 N NTR and δ 18 O NTR still appear robust.
Finally, although literature values of 15 ε DNF almost uniformly fall between values of 13 and 30 ‰ , values of 15 ε DNF as low as 2-5 ‰ have been observed occasionally in culture studies (Granger et al., 2008;Wada et al., 1975).While we have no direct evidence that such low values would be relevant in our study, we report the sensitivity of rate estimates and δ 15 N NTR and δ 18 O NTR (Table 1).In short, a prescribed value for 15 ε DNF of 5 ‰ leads to increased estimates of δ 15 N NTR and δ 18 O NTR by an average of 2.3 and 1.4 ‰, respectively (Figs. 5 and 6).These higher estimates of δ 15 N NTR would implicitly require a lower contribution of N 2 fixation derived nitrogen as argued for above, though not eliminate its role completely, especially in 4A where δ 15 N NTR values remain between −1 and +1 ‰ (Fig. 5).While rates of nitrification were less sensitive (somewhat higher in the upper layers of 4A), this very low prescribed value of 15 ε DNF often lead to dramatically increased estimates of denitrification rates -in particular in the upper transitional layers of profiles at 2B and 3D where ∼ 10-20-fold higher maximum denitrification rates are required to reconcile nitrate concentration and isotope data (Table 1).

4.2.5
Model-predicted values of 18 ε: 15 ε DNF and 15 ε DNF In the anoxic intervals, estimated values of 18 ε : 15 ε DNF ranged from 0.83 to 1.11 with an average value of 0.99 ± 0.1 (Table 2), consistent with a prominent role of respiratory nitrate reductase (Nar), which imparts a 18 ε : 15 ε DNF of ∼ 0.96 ± 0.01 (Granger et al., 2008).Notably, however, the lower values of 0.86 and 0.83 observed near the top and the core of the anoxic zone in site 2B could suggest influence of nitrate reduction by periplasmic nitrate reductase (Nap) (Granger et al., 2008) and chemolithotrophic NO − 3 reduction (Frey et al., 2014;Wenk et al., 2014), which has been shown to impart a lower 18 ε : 15 ε DNF closer to 0.6.In this particular interval, this could suggest that as much as 43 % of nitrate reduction is chemolithotrophic and perhaps metabolically linked to the oxidation of inorganic substrates such as reduced iron or sulfur species.Although outside the scope of this study, interrogation of genetic markers of respiratory and periplasmic nitrate reductase could shed more light on the role nitrate use by subsurface microbial communities.
As discussed above, the model-estimated values of 15 ε DNF (averaging 20.0 ‰ ± 1.8 ‰; Table 2) at site 2B are quite consistent with values from a wide range of studies (Granger et al., 2008), and references therein).Notably however, a different pattern emerges from the two anoxic intervals of site 4A.Although model-estimated values of 15 ε DNF were unresolvable at 38.8 m (likely because the changes in δ 15 N and δ 18 O were too small for reliable model fits), estimated 15 ε DNF values at 44.1 m were ∼ 8.1 ± 0.4 ‰, much lower than observed in 2B.In general, the values observed in 2B are consistent with observations from other environments hosting denitrification (Granger et al., 2008), and suggest that denitrifying organisms may be adapted to low levels of carbon (here < 0.2 % sediment organic carbon) and that their physiological poise may be similar to those found in other anaerobic environments (albeit adapted to grow at exceedingly slow nitrate reduction rates).However, the lower estimated 15 ε DNF values in 4A might also reflect something else.Given the apparent low reactivity of the sediments of site 4A, it is also possible that these particularly low 15 ε DNF values stem from denitrification operating under extreme physiological energy limitation -as discussed below.
While a number of studies have shown that the apparent N isotopic effect for nitrate reduction by denitrification can vary from 5 to 30 ‰ (e.g., Barford et al., 1999;Delwiche and Steyn, 1970;Granger et al., 2008), recent evidence suggests these variations are largely regulated by changes in the combination of cellular uptake and efflux of NO − 3 leading to the expression (or repression) of the enzyme-level isotope effect outside the cell (Granger et al., 2008;Kritee et al., 2012;Needoba et al., 2004).For example, at low extracellular NO − 3 concentrations -low 15 ε DNF values suggest that nitrate transport (having a low 15 ε) becomes the rate-limiting step (Granger et al., 2008;Lehmann et al., 2007;Shearer et al., 1991).In North Pond porewaters, however, at depths where O 2 is low enough for denitrification to occur, NO − In general, greater expression of the intrinsic enzymatic isotope effect (e.g., higher observed 15 ε DNF ) should occur under conditions in which there is a higher efflux of intracellular NO − 3 relative to NO − 3 uptake (Kritee et al., 2012).Interestingly, this efflux/uptake ratio appears to be linked to nitrate reduction rates in denitrifying bacteria, with lower cellspecific nitrate reduction rates leading to lower efflux/uptake ratios and lower observed cellular level 15 ε DNF (Kritee et al., 2012).Indeed, evidence seems to indicate that this efflux/uptake ratio in denitrifying bacteria is highly regulated and that NO − 3 uptake is sensitive to cellular-level energy supply.For example, under conditions in which organisms are required to maintain a careful regulation of energetically costly metabolic processes, it is logical that there would be a lower density of NO − 3 transporters and that intracellular NO − 3 concentrations would be maintained at or near-optimal levels for reduction by nitrate reductase.Similarly, growth under energy-poor carbon substrate supply may also lead to lower observed 15 ε DNF , due to an energy-driven decrease in NO − 3 uptake, lower intracellular NO − 3 concentrations and a lower efflux/uptake ratio.
We suggest that the difference between the lower 15 ε DNF value estimated from the anoxic interval of 4A and the more conventional values from deeper within anoxic intervals of 2B could stem from physiological-level controls on the cellular-level expression of 15 ε DNF .Specifically, as all porewater evidence from site 4A (O 2 , NO − 3 , N and O isotopes) indicates substantially lower levels of microbial activity, denitrification may actually be more energy-limited by carbon (compared to denitrification in the deeper intervals of 2B).This suggests that the operation of denitrification under extremely carbon-poor environments (4A) may lead to conditions where the enzyme-level N isotope fractionation of denitrification is under-expressed on both the cellular, and ecosystem levels, and 15 ε DNF values are much lower than commonly encountered under even just slightly more energyreplete conditions (e.g., 2B).

Summary
In summary, the porewater nitrate isotopic composition reflects the active redox cycling of nitrogen by the subsurface microbial community -including both oxidative and reductive transformations.The variations in reaction rates across and within the three North Pond sites are generally consistent with the distribution of dissolved oxygen, but not necessarily with the canonical view of how redox thresholds act to spatially separate nitrate regeneration from dissimilatory consumption (e.g., denitrification).The incorporation of nitrate dual isotopes into an inverse reaction-diffusion model provides evidence for extensive zones of overlap where O 2 and NO − 3 respiration (nitrification and denitrification) co-occur.The isotope modeling also yielded estimates for the δ 15 N and δ 18 O of newly produced nitrate (δ 15 N NTR and δ 18 O NTR ), as well as the isotope effect for denitrification ( 15 ε DNF ), parameters with high relevance to global ocean models of N cycling (Sigman et al., 2009).Estimated values of δ 15 N NTR were generally lower than previously reported δ 15 N values for sinking particulate organic matter in this region, suggesting the potential influence of sedimentary N 2 fixation and remineralization/oxidation of the newly fixed organic N. Modelestimated values of δ 18 O NTR generally ranged between −2.8 and 0.0 ‰, consistent with lab studies of nitrifying bacteria cultures.Notably, however, some δ 18 O NTR values were elevated, suggesting incorporation of 18 O-enriched dissolved oxygen during the nitrification process, and implying relatively rapid rates of nitrite turnover in environments supporting nitrification.In contrast, the accumulation of NO − 2 under denitrifying conditions likely reflects limitation of NO − 2 reduction by organic matter availability and generally low rates of N-based heterotrophic respiration.Importantly, our findings indicate that the production of organic matter by in situ autotrophy (e.g., nitrification and nitrogen fixation) must supply a substantial fraction of the biomass and organic substrate for heterotrophy in these sediments, supplementing the small organic-matter pool derived from the overlying euphotic zone.Thus, despite exceedingly low exogenous organic matter input, this work sheds new light on an active nitrogen cycle in the deep sedimentary biosphere underlying half of the global ocean.

Figure 1 .
Figure 1.Map of North Pond study site (created using the default Multi-Resolution Topography Synthesis base map in GeoMapApp ver.3.5.1).Color scale reflects water depth in meters with contour intervals of 100 m.

Figure 2 .
Figure 2. Depth profiles from IODP sites U1382B, U1383D, and U1384A at North Pond of porewater concentrations of O 2 (from Orcutt et al., 2013) and NO − 3 as well as the N and O isotopic composition of NO − 3 (δ 15 N NO 3 and δ 18 O − NO 3

Figure 4 .
Figure 4. Model-estimated rates of denitrification and nitrification (nmol cm −3 yr −1 ) based on fitting of nitrate concentration and N and O isotopic composition.Estimates within transitional intervals are calculated using a value for 15 ε DNF of either 15 ‰ (dashed) or 25 ‰ (solid).Error bars are shown for the 15 ε DNF = 25 ‰ only and indicate the standard error of 10 model runs (see text).Note the different scales for rates among the three profiles.

Figure 5 .
Figure 5. Model-estimated values for the N isotopic composition of new nitrate produced by nitrification (δ 15 N NTR ) occurring within porewaters.Values are calculated for depths at which O 2 concentrations were > 2 µM (e.g., oxic and transitional intervals).Sensitivity of δ 15 N NTR to prescribed values of the isotope effect for denitrification ( 15 ε DNF ) in transitional intervals, where both nitrification and denitrification can co-occur, is also indicated.

Figure 6 .
Figure 6.Model-estimated values for the O isotopic composition of new nitrate produced by nitrification (δ 18 O NTR ) occurring within porewaters.Values are calculated for depths at which O 2 concentrations were > 2 µM (e.g., oxic and transitional intervals).Sensitivity of δ 18 O NTR to prescribed values of the isotope effect for denitrification ( 15 ε DNF ) in transitional intervals, where both nitrification and denitrification can co-occur, is also indicated.

Table 1 .
Modeled N transformation rates in porewaters of North Pond sediments.Sensitivity of rates to model-prescribed values of 15 ) -indicating isotope fractionation by microbial denitrification.The highest δ 15 N and δ 18 O values in 2B (+22.2 and +21.8 ‰ , respectively) generally coincide with the lowest dissolved O 2 , while the highest 15 N and δ 18 O values in 3D (+11.8 and + 19.7 ‰ , respectively) fall just below the depth of lowest O 2 .In stark contrast, 4A porewaters exhibit only a minor increase in δ 18 O of ∼ 2.7 ‰ within the anoxic interval, while δ 15 N increased by only 0.8 ‰ (Fig.

Table 2 .
Denitrification isotope effects 15 ε DNF and 18 ε : 15 ε DNF estimated from anoxic porewaters at North Pond.Changes in δ 15 N and δ 18 O were too small over this interval for resolving a reliable estimate of 15 ε or 18 ε: 15 ε *