Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
 Water column depth profiles along the North Pacific margin from Point Conception to the tip of Baja California indicate elevation of nitrate (NO3−) 15N/14N and 18O/16O associated with denitrification in the oxygen-deficient thermocline waters of the eastern tropical North Pacific. The increase in δ18O is up to 3‰ greater than in δ15N, whereas our experiments with denitrifier cultures in seawater medium indicate a 1:1 increase in NO3− δ18O and δ15N during NO3− consumption. Moreover, the maximum in NO3− δ18O is somewhat shallower than the maximum in NO3− δ15N. These two observations can be summarized as an “anomaly” from the 1:1 δ18O-to-δ15N relationship expected from culture results. Comparison among stations and with other data indicates that this anomaly is generated locally. The anomaly has two plausible interpretations: (1) the addition of low-δ15N NO3− to the shallow thermocline by the remineralization of newly fixed nitrogen, or (2) active cycling between NO3− and NO2− (coupled NO3− reduction and NO2− oxidation) in the suboxic zone.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 The oceanic budget of biologically available (or “fixed”) nitrogen is poorly understood. Estimates of the global rate of nitrogen (N) loss by denitrification would leave the ocean N budget far out of balance unless N2 fixation rates are much higher than previously estimated [Brandes and Devol, 2002; Codispoti et al., 2001; Middelburg et al., 1996]. While such imbalances cannot be ruled out, the stability of atmospheric CO2 and of the N isotopic composition of deep sea sediments over the last ∼5 kyr argues against such extreme imbalances [Deutsch et al., 2004; Kienast, 2000].
 Direct measurements of N fluxes in the ocean (e.g., N2 fixation, denitrification, NO3− assimilation, and nitrification) cannot, by themselves, provide a reliable picture of the ocean N cycle. Temporal and spatial complexity, combined with the limitations of shipboard sampling of the ocean, lead to uncertainty in the extrapolation of these measurements to regional and global fluxes. Moreover, assays for N transformations can perturb the samples they are attempting to measure. For these reasons, biogeochemical parameters in ocean water have become important as more integrative measures of the rates of N fluxes.
 Deviations in the [NO3−]-to-[PO43−] relationship from the “Redfield” relationship driven by algal assimilation and remineralization are used to study the rates and distributions of both N2 fixation and denitrification. “N*”, defined as [NO3−] − 16 × [PO43−] + 2.9 (in μmol/kg) [Deutsch et al., 2001], quantifies excesses and deficits in NO3− relative to the globally derived [NO3−]-to-[PO43−] relationship, indicating regions of N2 fixation and denitrification, respectively. When combined with some measure of ocean circulation, rates of these processes can be derived [Deutsch et al., 2001; Gruber and Sarmiento, 1997]. While this use of nutrient data is extremely powerful, it has limitations. First, deviations from the Redfield [NO3−]-to-[PO43−] relationship may not always be due to N inputs or outputs, arising instead from variations in the stoichiometry of nutrient uptake and remineralization. Second and most relevant here, NO3− inputs and losses partially erase one another if they occur in the same water or if their host waters are mixed in a way that cannot be reconstructed.
 However, the N isotopes do not escape the second weakness described above for the N-to-P approach: Because N2 fixation and denitrification have counteracting effects on both N* and NO3− δ15N, gross fluxes from either process cannot be determined with sufficient accuracy. For instance, N2 fixation in the tropical and subtropical Pacific surface adds new NO3− to the Pacific thermocline, increasing N* and decreasing NO3− δ15N, while denitrification removes low-δ15N NO3− from suboxic regions of the eastern Pacific, lowering N* and raising the δ15N of the residual NO3−. If the two processes occur in the same region or if waters from these regions mix vigorously, the tracer signals of both processes are reduced [Deutsch et al., 2001].
 Previous coupled studies of NO3− N and O isotopes have observed a strong correlation between these two isotope systems, with both NO3− δ15N and δ18O increasing as NO3− is consumed by denitrification. Freshwater studies observe an O:N ratio for isotope effects (18ɛ:15ɛ) of ∼0.5–0.6, for their respective isotope systems [Bottcher et al., 1990; Lehmann et al., 2003; Mengis et al., 1999]. However, our culture experiments with denitrifiers in seawater yield an 18ɛ:15ɛ of ∼1 [Granger et al., 2004a], as does a field study of an enclosed marine basin [Sigman et al., 2003b]. This fits with our previous observations for algal NO3− assimilation, for which we also observe an 18ɛ:15ɛ of ∼1 over a broad range of amplitudes for the isotope effect [Casciotti et al., 2002; Granger et al., 2004b]. That we observe the same 18ɛ:15ɛ for both denitrification and NO3− assimilation is consistent with evidence that NO3− reduction is the dominant cause of fractionation in both processes [Needoba et al., 2004; Shearer et al., 1991]. Thus, while we have much to learn about the N:O fractionation ratios, NO3− assimilation and denitrification, the processes of NO3− consumption with the greatest effects on oceanic NO3− distributions, apparently cause similar isotope fractionation of N and O in NO3−.
 Unlike the consumption of NO3−, NO3− production appears to have very different effects on the N and O isotopes of NO3−. In the open ocean subsurface, at least in oxic waters, almost all of the ammonium generated from organic N is eventually oxidized to NO3−, so that the N isotope effects associated with ammonium production and nitrification do not impact the δ15N of NO3− produced. In this case, the δ15N of newly produced NO3− is primarily controlled by the δ15N of the organic matter being remineralized. The δ18O of newly produced NO3− obviously does not depend on the isotopic composition of the organic matter being remineralized.
 Biochemical studies have derived mechanisms for ammonium oxidation to nitrite (NO2−) in which one O atom is donated from O2 and the other from water [Andersson et al., 1982]. NO2− oxidation to NO3− involves the donation of O only from water [Dispirito and Hooper, 1986; Kumar et al., 1983]. On this basis, the traditional interpretation has been that two thirds of the O atoms in NO3− should originate from water and one third should originate from O2 [Böhlke et al., 1997; Durka et al., 1994; Kendall, 1998; Wassenaar, 1995]. However, the same biochemical studies also demonstrated a strong nitrifier-catalyzed nitrite-water exchange of O atoms [Andersson et al., 1982]. On the basis of these observations, it is likely that much less than one out of two O atoms in NO2− comes from O2 [Casciotti et al., 2002]. A culture experiment in which Nitrosomonas europaea produces NO2− in the presence of 18O-labeled water indicates that at least 50% of the O atoms in NO2− have undergone exchange with water [Casciotti, 2002], such that at least 5 out of the 6 O atoms in NO3− originate from water (i.e., 1 or less out of the 6 comes from O2). It is also possible that catalysis of exchange with water occurs during the oxidation of NO2− to NO3− [Dispirito and Hooper, 1986], reducing further the effective contribution of O atoms from O2 and increasing the contribution from H2O.
 Measurements to date from the ocean indicate that away from regions of known denitrification, subsurface NO3− δ18O varies relatively little and is close to the ambient water (0 ± 1‰ or 3 ± 1‰ different from it; see auxiliary material, endnote i in Auxmat1.txt) the ambient water (δ18Osample = ((18O/16O)sample/(18O/16O)reference − 1) × 1000‰, where the reference is Vienna Standard Mean Ocean Water (VSMOW); see section 2). Deep NO3− δ18O is within ±1‰ among regions with very different deep O2 concentration (the Bering Sea, the North Pacific, the Southern Ocean, and the North Atlantic), arguing against a strong influence from O2 δ18O [Casciotti et al., 2002; Lehmann et al., 2005] (A. Knapp, unpublished data, 2005). This is consistent with ambient water being the dominant source of the O atoms in NO3−, although much work remains to be done on this question.
 In general, the additional insight that NO3− δ18O brings to measurements of NO3− δ15N and N* involves the processes that are not captured by the O isotopes (Figure 1). As seen from the N atom in NO3−, NO3− assimilation and nitrification are part of an internal cycle within the ocean that should cause no net change in the mean δ15N of ocean NO3− over time. N2 fixation and denitrification (plus additional smaller terms) comprise the input/output budget of fixed N and control the mean δ15N of ocean NO3− [Brandes and Devol, 2002; Deutsch et al., 2004]. In contrast, for the O atoms in NO3−, nitrification is an absolute input, while both NO3− assimilation and denitrification are absolute sinks. The δ18O of newly produced NO3− does not depend on the origin of NH4+ being nitrified, be it from newly fixed N, from the biomass of phytoplankton growing in a NO3−-rich environment, or from biomass of phytoplankton in a NO3−-poor environment that assimilate all of the NO3− supplied to them.
 This fundamental difference between the N and O isotopes of NO3− allows their coupled measurement to separate processes that overprint one another when they are monitored using NO3− δ15N alone. For instance, with the added constraint of NO3− δ18O, it should be possible to separate and quantify the impacts of N2 fixation and denitrification. The N and O isotopes of NO3− are fractionated to the same extent by denitrification. Thus NO3− becomes enriched in both 15N and 18O as denitrification proceeds. The difference between the two isotope systems arises with their different sensitivities to N2 fixation. While the nitrification of newly fixed N will work to lower the δ15N of subsurface NO3−, the δ18O of NO3− produced by nitrification is insensitive to the origin of the N being remineralized in the subsurface. Thus O isotopes may indicate when the impact of N2 fixation has caused NO3− δ15N (and N*) to underestimate the NO3− lost to denitrification.
 At the same time, the O isotopes may record other gross fluxes of NO3− that do not impact the N isotopes. For instance, if NO3− is reduced to some other form (organic N, NH4+, or NO2−) and then oxidized back to NO3− without any N loss, the δ15N of NO3− is constrained by mass balance to remain unchanged, whereas the δ18O of NO3− may change (Figure 1). The direction in which δ18O changes will depend on whether the δ18O of the NO3− removed is higher or lower than the δ18O of the NO3− added back. If the NO3− added back is higher in δ18O than that removed, then the NO3− δ18O will drift upward. Because isotope discrimination during NO3− reduction often causes the δ18O of the consumed NO3− to be less than δ18O of newly produced NO3−, it will generally be the case that NO3− δ18O will increase relative to δ15N with the rate of an internal cycle of NO3− consumption and production.
 Here we use the coupled N and O isotopes of water column nitrate as complementary constraints on the N transformations at work in and nearby the eastern tropical North Pacific denitrification zone. Our central new observation is that the δ18O of NO3− is up to 3‰ more elevated than is its δ15N relative to “background” (e.g., deep open ocean) NO3−, with the greatest deviation between the two isotope systems at ∼100 m shallower than the previously described δ15N maximum. Given that our culture experiments indicate a 1:1 δ18O:δ15N elevation by denitrification, we attempt to identify and quantify the process responsible for the deviation of the O and N isotopes from denitrification-only behavior.
2. Materials and Methods
2.1. Sample Collection
 Water samples were collected through the water column by hydrocast off the California coast from Point Conception to the southern tip of Baja California during coring cruise OXMZ01MV aboard the RV Melville in November of 1999 (Figure 2) [van Geen, 2001]. Samples were collected in acid- and distilled water-rinsed polyethylene bottles after two rinses with sample water and were preserved by acidification to a pH of 2–3 with 50% reagent-grade hydrochloric acid. Upon arrival at the laboratory 4 months after collection, an aliquot of each sample was frozen, and these aliquots were used for NO3− N and O isotope analysis.
2.2. Dissolved Oxygen and Nutrient Concentration Measurements
 During OXMZ01MV, the concentrations of phosphate (PO43−), nitrate (NO3−), and nitrite (NO2−) were measured at sea by automated colorimetric methods, and the concentration of dissolved O2 was measured by Winkler titration. In the hydrocast profiles from OXMZ01MV, [NO2−] was less than 0.1 μM in all but one 50-m sample and was typically less than 0.05 μM. This is much lower than measured at lower latitudes along the eastern tropical Pacific margin [Codispoti et al., 1986; Lipschultz et al., 1990] but fits with previously reported distributions [Cline and Richards, 1972] (see endnote ii in Auxmat1.txt [Deutsch et al., 2001; Gruber and Sarmiento, 1997]).
2.3. NO3− Isotopic Analysis
 The 15N/14N and 18O/16O of NO3− were determined using the denitrifier method [Casciotti et al., 2002; Sigman et al., 2001]. Briefly, NO3− and NO2− are converted quantitatively to N2O by a strain of bacterial denitrifier that lacks nitrous oxide reductase activity, and the product N2O is extracted, purified, and analyzed by continuous flow isotope ratio mass spectrometry. Individual analyses are referenced to injections of N2O from a pure gas cylinder and then standardized using international NO3− isotopic reference material IAEA-N3. The O isotope data are corrected for exchange with oxygen atoms from water during reduction of NO3− to N2O [Casciotti et al., 2002], which is quantified by analysis of IAEA-N3 in 18O-enriched water and was 5% or less for the analyses reported here. Reproducibility of replicates (which were analyzed for ∼75% of the water samples) was generally consistent with previously reported analysis standard deviations of 0.2‰ for δ15N and 0.5‰ for δ18O (see endnote iii in Auxmat1.txt [Anbar and Gutmann, 1961; Böhlke et al., 2003; Bunton et al., 1952]).
 As stated above, referencing of 15N/14N to atmospheric N2 and of 18O/16O to VSMOW was through comparison to the potassium nitrate (KNO3) reference material IAEA-N3, with an assigned δ15N of +4.7‰ [Gonfiantini et al., 1995] and reported δ18O of +22.7 to +25.6‰ [Böhlke et al., 2003; Lehmann et al., 2003; Revesz et al., 1997; Silva et al., 2000]. We adopt here a δ18O of 22.7‰ [Lehmann et al., 2003; Revesz et al., 1997; Silva et al., 2000], as we have used in previous publications. If we were to assume the most recent and highest estimate for the δ18O of IAEA-N3 (25.6‰, [Böhlke et al., 2003]), then the NO3− δ18O of all of our samples would increase by ∼2.9‰. Indeed, we expect that the new, higher δ18O of IAEA-N3 will prove to be correct, but we wish to guard against using multiple different referencing schemes through time and thus will wait for corroboration of the results of Böhlke et al. . The O isotopic difference between NO3− reference IAEA-N3 (and indeed all NO3− references) and Vienna SMOW is not addressable with the denitrifier method, which can only measure isotopic differences among NO3− samples. The uncertainty in the isotopic difference between IAEA-N3 and VSMOW is an unfortunate source of uncertainty in our reported values. However, our focus here is on the variation of NO3−18O/16O in the ocean, not its relationship to the isotope ratios found in seawater or other O-bearing materials. Our interpretation is not affected by a uniform shift in the δ18O of all of our data sets relative to VSMOW, because all of the O isotope rules used in the calculations below are based on our own NO3− isotope data.
 While N* is generally negative throughout the eastern North Pacific (ENP), there is a thermocline-depth N* minimum that indicates in situ denitrification or rapid exchange with a region of denitrification (Figure 3c). The N* minimum is associated with the [O2] minimum (Figure 3e), with both O2 depletion and the N* minimum becoming more pronounced toward the south among our station locations. This is consistent with a requirement of very low [O2] (<4 μM or so) for denitrification to proceed rapidly in the water column [Lipschultz et al., 1990, and references therein]. The expectation based on the [O2] data is that water column denitrification is only active in the stations south of 25°N (blue symbols in Figure 3e). The N* minimum and NO3− δ15N and δ18O maxima of the more northern stations result from the coastal undercurrent carrying northward these signals of denitrification [Altabet et al., 1999; Liu and Kaplan, 1989; Sigman et al., 2003b; Wooster and Jones, 1970].
 Comparison of profiles shows qualitatively that NO3−15N/14N (Figure 3b) is strongly anti-correlated with N* (Figure 3c), as would be expected from N isotope discrimination during denitrification. For a range of models of NO3− supply and consumption, a N isotope effect (15ɛ) for denitrification of ∼25‰ has been estimated [Sigman et al., 2003b], consistent with other studies referenced above. A much lower net isotope effect applies in the Santa Barbara Basin (station 3) because of denitrification in the sediments of that basin [Sigman et al., 2003b].
 The overall δ18O:δ15N trend of 1.25 in the ENP data actually hides systematic depth-variations in the relationship between δ18O and δ15N. At ∼350 m, as N* begins its upward increase and NO3− δ15N begins to decrease, NO3− δ18O holds steady or continues to increase an additional 100 m toward the surface before decreasing again, resulting in a NO3− δ18O maximum that is ∼100 m shallower than the δ15N maximum and the N* minimum. In our plots of NO3− δ18O versus NO3− δ15N (Figure 4), this leads to a “loop” (counterclockwise up) pattern: shoaling from the deepest samples, the isotopic composition of NO3− progresses upward and to the right along a slope of ∼1.25 in δ18O/δ15N space, then shifts toward a more vertical path as δ18O continues to increase but δ15N remains unchanged or decreases, then returns downward and to the left, typically reaching a δ18O-to-δ15N relationship at ∼100 m that is similar to that of deep waters.
4.1. Quantifying the Deviation Between NO3− O and N Isotopes in the Thermocline
 We focus first on the ENP profiles from the southern tip of Baja (stations 7–16), where conditions are appropriate for water column denitrification. As described above, the relationship between the NO3− N and O isotopes within the suboxic zone (200–800 m) cannot be explained solely by denitrification with an 18ɛ:15ɛ ∼ 1, especially in its shallow portion (e.g., at ∼200 m). Graphically, the discrepancy from a 1:1 fractionation relationship expected for denitrification can be visualized as the horizontal distance in δ18O-versus-δ15N space between the data and a line with a slope (18ɛ/15ɛ) of 1 appropriate for denitrification running through the mean δ15N and δ18O of ENP deep water (Figure 4a). We formalize this as “Δ(15,18)”,
where δ15Nm and δ18Om are the mean δ15N and δ18O of eastern North Pacific deep water, which is taken to approximate the source of NO3− to the upper water column of the eastern North Pacific, and 18ɛ:15ɛ is the N-to-O isotope effect ratio for denitrification, which our culture studies indicate to be 1 [Granger et al., 2004a]. We use here 5‰ and −0.5‰ for δ15Nm and δ18Om (based on samples taken from 3500 m and below at HOT station ALOHA (D. M. Sigman and D. Karl, unpublished data, 2005)), such that the 800–1450 m data from stations 7–16 yield a Δ(15,18) close to 0‰ (+0.2‰, Table 1) (see endnote v in Auxmat1.txt).
Table 1. Water Column Parameters for Model Targets
Stations 7–16, 200–800 m
Stations 7–16, 800–1450 m
δ15N, ‰ versus air
δ18O, ‰ versus VSMOW
 For stations 7–16, Δ(15,18) varies coherently with depth (Figure 5c), being close to zero below 800 m (by definition) and decreasing upward to a minimum of −2.5‰ at 200 m, with a sharp increase to 100 m and above. The minimum in Δ(15,18) is ≥100 m shallower than the δ15N maximum (Figure 5b) and the N* minimum (Figure 5e). Given that the deviation is not proportional to δ15N or N*, it is not well explained by a uniform deviation in 18ɛ:15ɛ from the culture-derived estimate of 1. Moreover, this sense of deviation would require an 18ɛ:15ɛ > 1, for which there is no support from previous work in seawater or freshwater. Finally, an 18ɛ:15ɛ of 1 yields an excellent fit to the data from the Santa Barbara Basin (indicated red circles in Figure 4c), in which denitrification is progressively drawing down NO3− after a springtime flushing event [Sigman et al., 2003b].
4.2. Regional Extent of the Δ(15,18) Minimum
 The ∼200-m-centered minimum in Δ(15,18) weakens as one moves north along the California margin and is not evident near Point Conception (Figure 6). The shallowest samples in the more northern profiles tend to reach positive values for Δ(15,18), which can be explained as a result of the algal uptake/remineralization cycle (see below). The lack of a strong Δ(15,18) minimum in the more northern profiles rules out the possibility that the minimum near the tip of Baja originates from advection from the north, for instance, because of a negative Δ(15,18) in preformed NO3− from regions of ventilation to the north. Comparison with Hawaii Ocean Time series station ALOHA shows clearly that the Δ(15,18) minimum in the ENP is also not transported into the eastern North Pacific margin from the west (D. M. Sigman and D. Karl, unpublished data, 2005). While it is still possible that the suboxic zone to the South represents a source for the Δ(15,18) minimum in the ENP near the tip of Baja, the data in hand indicate no role for transport and suggest that the Δ(15,18) minimum is generated locally.
 One aspect of the Δ(15,18) minimum that seems to have a simple cause is the upward increase in Δ(15,18) from the 200 m minimum toward the surface; this is well explained by the NO3− assimilation/remineralization cycle. The lack of significant surface NO3− in the region indicates that upwelled NO3− is consumed to completion by algal uptake. Thus the organic matter produced and exported into the subsurface will have the same δ15N as the upwelled NO3−. However, the nitrification of this organic N produces NO3− with a δ18O of ∼0‰ (i.e., close to that of water), essentially “washing” the 18O enrichment from the NO3− pool. This should tend to increase the Δ(15,18) as one approaches the top of the thermocline. Indeed, Δ(15,18) reaches positive values in many cases (Figures 5 and 6), most likely because of this effect. These samples are evident in δ18O-vs-δ15N space as the points that reach below the 1:1 line in the lower left sector of the plot (Figures 4a and 4c).
 In the subtropical thermocline of the North Atlantic and North Pacific, there is evidence for the production of a sizable NO3− excess relative to expectations based on PO43− concentration and Redfield ratios; this finding has been interpreted to indicate that newly fixed N is accumulating as NO3− in the thermocline waters of these regions [Deutsch et al., 2001; Gruber and Sarmiento, 1997; Hansell et al., 2004; Michaels et al., 1996]. NO3− in the subtropical thermocline of both the Pacific and the North Atlantic has been observed to have a low δ15N, as low as 2‰ [Karl et al., 2002; Knapp et al., 2005; Liu et al., 1996]. Given the low δ15N of newly fixed N, the low δ15N of subtropical thermocline NO3− is consistent with the N*-based interpretation of the accumulation of newly fixed N as thermocline NO3− [Gruber and Sarmiento, 1997]. More work is needed to validate this interpretation, but it would seem difficult for it to be strictly incorrect.
 On the basis of similar logic, Brandes et al.  explain the upward decrease in δ15N above denitrification zones in the Arabian Sea and eastern tropical North Pacific as the result of oxidation of low-δ15N, newly fixed N to NO3−. This explanation fits with the upward change in the δ18O/δ15N relationship reported here. That is, the shallower δ18O maximum suggests that the nitrification of newly fixed N is “eroding” the tops of the NO3− δ15N maximum and the N* minimum. It is not clear whether nitrification would be limited within the suboxic zone of our study region [Lipschultz et al., 1990]. In any case, the suboxia does not extend far offshore at the latitudes of our stations [Conkright et al., 2002], so NO3− could be produced from nitrification in the oxic waters just to the west and imported along isopycnals.
 As described above, the upward increase in Δ(15,18) above its minimum at 200 m is well explained by complete assimilation of upwelled NO3− and subsequent remineralization of most of the exported organic N in the shallow subsurface. That the minimum in Δ(15,18) is strongest at 200 m and not deeper could be explained by (1) the lower [NO3−] at shallower depths, which requires a smaller amount of newly fixed N to cause the same decrease in Δ(15,18), (2) the tendency for nitrification at the upper margin of the suboxic zone [Lipschultz et al., 1990], and/or (3) the rapid decrease in the sinking N flux with depth in the water column.
4.3.2. Nitrate/Nitrite Redox Cycling
 Since the work of Anderson [Anderson, 1982; Anderson et al., 1982], it has been hypothesized that there is significant redox cycling between nitrate and nitrite in ocean suboxic zones, with NO3− reduction to NO2− in the core of the suboxic zones, mixing of the NO2− to the margins of the suboxic zone, and reoxidation of the NO2− once it reaches higher [O2] waters. Anderson suggested that roughly half of the nitrate reduction in open ocean suboxic zones can be coupled to nitrite oxidation, the other half proceeding to denitrification. This exact process is not plausibly significant in our study region, as the measured [NO2−] rarely climbed above 0.05 μM (typically ∼0.01 μM) in the subsurface samples. However, there might be exchange along isopycnals with waters to the south where that process could occur. Moreover, there might well be simultaneous NO3− reduction and NO2− oxidation in the same water parcel within our study region [Lipschultz et al., 1990].
 Such a cycle might explain the deviation of NO3− δ18O and δ15N from 1:1 covariation. NO3− reduction will consume NO3− with the N and O isotope effects of denitrification. If the ambient NO3− δ15N and δ18O are 14‰ and 10‰, respectively, an isotope effect of 20‰ (for both N and O) will make the consumed NO3− approximately −6‰ and −10‰, respectively. When the NO2− produced is reoxidized to NO3−, it will return NO3− with roughly the same δ15N as the loss, so that the ambient NO3− δ15N is, in net, unchanged. The δ18O of the reoxidized NO3−, however, would most likely be higher than the δ18O of the NO3− consumed. The preferential extraction of 16O from the chain of N species (i.e., a “branching fractionation”) yields NO2− with a δ18O higher than that of the NO3− consumed [Casciotti et al., 2002], such that its recycling back into the NO3− pool may cause a net increase in NO3− δ18O. In addition, the reduction to NO2− and reoxidation to NO3− will work to incorporate O atoms from H2O, such that the reoxidized NO3− would likely be shifted toward a δ18O of 0‰. This shift might be complete if there is rapid O atom exchange with water in the enzyme active site of NO2− oxidase [Dispirito and Hooper, 1986], as has been observed to occur in the presence of enzymes catalyzing ammonium oxidation to NO2− [Andersson et al., 1982]. Alternatively, the only O atoms added from H2O may be the single O required to convert NO2− to NO3−, so that the δ18O of NO3− from NO2− reoxidation has some memory of the δ18O of NO2− produced (as well as of NO2− reduction, which would increase the δ15N and δ18O of NO2−; see below and endnote vii in Auxmat1.txt [Bryan et al., 1983; Casciotti, 2002]). Details aside, the coupling of NO3− reduction and NO2− reoxidation should work to raise the δ18O of ambient NO3− relative to its δ15N, thereby generating a negative Δ(15,18).
 The plausibility of a role for the NO3−/NO2− redox cycle in explaining the Δ(15,18) minimum is unclear. The minimum in Δ(15,18) at the top of the suboxic zone agrees with the expectation that the NO3−/NO2− redox cycle would be most intense where the vertical [O2] gradient is steepest [Lipschultz et al., 1990]. It is troubling that NO2− is so scarce in this region of the ENP, although this does not absolutely preclude a tight balance between the reduction, release, and reoxidation of NO2−. An additional argument against the NO3−/NO2− redox cycle explanation for the Δ(15,18) minimum is the lack of any anomaly in Δ(15,18) associated with denitrification in the Santa Barbara Basin (Figure 4c), where we presume such a NO3−/NO2− redox cycle should be equally active.
4.4. Steady State Model of the Candidate Processes
 We describe the results from the simplest possible quantitative model that we could conceive to estimate the fluxes of the two alternative processes that we have proposed to explain the Δ(15,18) minimum (Figure 7). This model represents the effects of five N cycle processes acting simultaneously on the suboxic thermocline zone of the ENP: (1) mixing with the deeper ENP (M), (2) denitrification (D), (3) mixing with a biologically active and NO3− deplete surface ocean (S), (4) addition of NO3− from the nitrification of N from new N2 fixation (F), and (5) redox cycling between NO3− and NO2− (C; C1 is NO3− reduction, C2 is NO2− oxidation). The following rules apply to the fluxes.
 1. Mixing with deeper eastern North Pacific water (M in Figure 7) introduces NO3− with a concentration, δ15N, and δ18O measured in the water below the suboxic zone by our study (Table 1), while it removes NO3− with whatever concentration and isotope composition occurs in the thermocline box.
 2. Denitrification (D in Figure 7) consumes NO3− with a kinetic isotope effect that is equivalent for 15N/14N and 18O/16O. The amplitude of the isotope effect is adjusted to fit the data and is reported below.
 3. Mixing with the surface ocean (S in Figure 7) has no effect on [NO3−] or NO3− δ15N because all NO3− mixed upward into the surface is consumed in the surface and exported as organic N back into the reservoir, where it is completely remineralized to NO3−. However, the nitrification of this organic N export produces NO3− with a δ18O of 0‰, essentially ‘washing’ the 18O enrichment from the NO3−.
 4. The NO3− added from newly fixed N (F in Figure 7) has a δ15N of −1‰ and a δ18O of 0‰.
 5. In the NO3−/NO2− redox cycle (C in Figure 7), NO3− reduction (C1) occurs with the same 15ɛ and 18ɛ as denitrification (D). For a given 15ɛ for denitrification, the δ15N of the NO3− reoxidized from NO2− depends on the relative amplitudes of 15ɛ for NO2− reduction and NO2− oxidation; we assume that these isotope effects are equal in the calculations but consider other cases in the text. The δ18O of the NO3− reoxidized from NO2− depends on the same factors as does its δ15N; however, the δ18O is also affected by two additional factors. First, 16O is preferentially lost from the nitrogen species in the denitrification pathway [Casciotti et al., 2002]. This “branching fractionation” during NO3− reduction (assumed here to be equivalent to the 18ɛ of nitrate consumption by denitrification) yields NO2− with a δ18O ∼ 18ɛ‰ higher than that of the NO3− consumed, such that its recycling back into the NO3− pool may cause a net NO3− δ18O increase. Second, incorporation of O from H2O during (1) NO2−/H2O exchange and (2) NO2− oxidation drives the δ18O of the reoxidized NO3− toward 0‰. While Figure 7 shows only the case for complete O exchange between NO2− and H2O, the cases of complete O exchange and no exchange are both considered in the calculations below. For lack of better information, we assume that the 18ɛ/15ɛ ratio is the same for NO2− reduction as for NO2− oxidation, regardless of what that ratio might be.
 Here we consider only the model steady state. Varying D, F, and C, we fit [NO3−], NO3− δ15N, and Δ(15,18) for the means for the 200–800 m depth zone from stations 7–16, using the 800–1450 m data from the same stations to estimate the values for background ENP conditions (Table 1). The nitrate isotopes and N* of the 800–1450 m water indicate that it is impacted by denitrification, by exchange with the eastern tropical Pacific suboxic zones and by sedimentary denitrification (P. DiFiore, unpublished results, 2005), and is thus far from reflecting the mean conditions of the global ocean or even the whole North Pacific. We address here only the fluxes that drive the isotopic and concentration differences between the suboxic thermocline box and the 800–1450 m water below it.
 We opted here to use mixing with the deeper water from the same stations, as opposed to lateral exchange, as the mechanism for refreshing the 200–800 m suboxic thermocline box. This allowed the current study to be self-contained with respect to measurements. Efforts to use other mixing end-members (e.g., the thermocline from the open subtropical Pacific as measured at station ALOHA (D. M. Sigman and D. Karl, unpublished data, 2005) or the thermocline from our more northern stations (Figures 3 and 6)), yielded similar results that nevertheless require the consideration of additional factors (calculations not shown).
 The lack of a time-keeping constraint in our model means that we can only explore ratios of fluxes (i.e., the ratio of F or C to D), not the absolute magnitude of each flux. For flux magnitudes to be at least physically reasonable, we assume a value for M that yields a water residence time in the suboxic thermocline box of 10 years, intended to be roughly consistent with previous studies [Deutsch et al., 2001].
 Since F and C are alternative plausible explanations for the Δ(15,18) minimum, we explore these two terms separately in the sections below. However, they may both be at work.
4.4.1. Quantifying the Needed N2 Fixation
 The results from the model are largely intuitive. First, more of an assumption than an observation, the difference in [NO3−] between the suboxic thermocline box and the deeper water ([NO3−]B − [NO3−]M) is logically equivalent to the N* difference from the deeper water, to which we refer below as the “NO3− deficit” of the box. Second, the steady state NO3− deficit is affected solely by (and is proportional to) the ratio (D − F)/M. Since M is held constant in our calculations, Figure 8 indicates that the NO3− deficit is a function of D − F (Figure 8, green contours). Third, NO3− δ15N (and δ18O) increases with D and decreases with F (Figure 8, red contours). Fourth, Δ(15,18) decreases as N2 fixation increases, almost regardless of D (Figure 8, blue contours). Further visualization of model results are in the Auxiliary Materials (see endnote viii in Auxmat1.txt).
 In order to fit the 200–800 m data from stations 7–16 (Table 1), we find that the needed N2 fixation/denitrification ratio (F/D) is roughly 0.65 (black circle in Figure 8). This suggests that 65% of the denitrification occurring in the 200–800 m suboxic zone is countered by the nitrification of newly fixed N. N* in the suboxic zone (200–800 m) is 6.2 μM lower than in the deeper waters between 800 and 1450 m (−12.6 μM and −6.4 μM, respectively; Table 1). Thus our results would require that N2 fixation is erasing a NO3− deficit of (0.65/(1 − 0.65)) × 6.2 μM, or −11.6 μM. Added to the observed N* of −12.6 μM, this would yield a N* in the suboxic zone of −24.2 μM, were it not for N2 fixation, that is, a total N* minimum of roughly twice the observed amplitude.
 The isotope effect for denitrification that is required to simultaneously fit the N*, NO3− δ15N, and NO3− δ18O (or Δ(15,18)) data is 18.9‰, ∼5–10‰ lower than derived previously from regression of NO3− δ15N against N* in field data [Altabet et al., 1999; Brandes et al., 1998; Sigman et al., 2003b]. The need for a lower isotope effect than previous field studies at least partially arises from our isotope-derived inference that N2 fixation causes the N*-derived NO3− deficit to be less than the actual amount of NO3− consumed by denitrification. While the true biological isotope effect amplitude for denitrifiers in the ENP is not known, the value required by the model may be lower than that value, in which case it may indicate that a fraction of the NO3− consumption occurring within the suboxic zone is driven by sedimentary denitrification along the margin [Sigman et al., 2003b]. However, the isotope effect amplitude required by the model would also increase modestly if spatial heterogeneity were included in the model [Deutsch et al., 2004].
 The assimilation/remineralization cycle (S in Figure 7), in the case of complete NO3− consumption in the surface, decreases NO3− δ18O toward its nitrification production value (∼0‰) while not affecting NO3− δ15N; in net, the effect is to increase Δ(15,18), that is, erode the Δ(15,18) minimum (see above). We neglect this term in the calculation shown in Figure 8, setting S to 0. Including this term would yield an even higher N2 fixation/denitrification ratio (see endnote ix in Auxmat1.txt), but estimating the amplitude of S is difficult.
4.4.2. Quantifying the Needed Nitrate/Nitrite Redox Cycling
 Because there is essentially no NO2− in this region of the ENP, if the signal is generated locally, a putative NO3−/NO2− redox cycle must occur within a given water sample (i.e., without NO2− transport). Thus we can meaningfully compare the model results to the peak amplitude of the Δ(15,18) minimum (−2.51‰) at 200 m as well as to the mean Δ(15,18) of the 200–800 m interval (−1.15‰, Figure 9, dotted and solid gray bars, respectively). For the cases considered here, NO2− oxidation must be ∼0.7–0.95 and ∼0.35–0.45 times the rate of NO2− reduction to fit the observations at 200 m and over the 200–800 m interval, respectively (Figure 9). The preference for 16O-NO3− during NO3− reduction and the preferential extraction of 16O from the NO2− produced (the “branching fractionation”), which we assume here to have the same 18ɛ, offset one another to yield NO2− with a δ18O close to that of the NO3− in the water and thus <10‰ greater than the δ18O of the water. Therefore, for a given amount of NO3−/NO2− cycling, the case of no NO2−/H2O exchange yields only slightly greater Δ(15,18) than the case with complete exchange (solid versus dashed line in Figure 9).
 The unknown isotope systematics of NO2− represent a major weakness in this modeling exercise (also see endnote x in Auxmat1.txt [Bryan et al., 1983; Casciotti, 2002]). Nevertheless, the ratios given above for NO2− oxidation to reduction are generally within the range of those originally proposed as part of a transport cycle (0.65–1.50) [Anderson, 1982; Anderson et al., 1982] or measured within individual water samples [Lipschultz et al., 1990]. However, we again note that those rates involved waters with 5–10 μM NO2−, whereas there is essentially no NO2− in our profiles. Thus, if the isotopic signal of this process is important, it may be through exchange with NO2−-bearing waters to the south.
5. Summary and Conclusions
 Here we report coupled N and O isotope measurements of NO3− from a set of hydrocast stations collected along the continental margin from Point Conception to the southern tip of Baja California. The isotope data from the California margin show a distinct anomaly in the δ18O:δ15N relationship from expectations for denitrification alone, with δ15N being lower than expected from δ18O. This isotope anomaly (described as a negative value for “Δ(15,18)”) is present from 200 to 800 m but peaks at 200 m, above the maximum in NO3− δ15N. Comparison of the data from the tip of Baja with the stations from further north and with data from near Hawaii (D. M. Sigman and D. Karl, unpublished data, 2005) indicates that the anomaly originates in or near the region of denitrification.
 One plausible explanation for the Δ(15,18) minimum is the addition of low-δ15N NO3− to the shallow thermocline in the same region where denitrification occurs, which “erodes” the tops of the denitrification-driven maximum in NO3− δ15N and minimum in N*. The most likely origin of this low-δ15N NO3− is N2 fixation in the surface ocean, the rain of this newly fixed N out of the surface ocean, and the subsequent nitrification of its products to NO3− in the thermocline. This is consistent with a previous interpretation of NO3− δ15N data alone from the eastern tropical North Pacific and Arabian Sea that N2 fixation was adding significant amounts of low-δ15N NO3− to the shallow thermocline in these regions [Brandes et al., 1998]. We use the coupled N and O isotope data, in the context of a simple model, to estimate that the rate of this putative N2 fixation is roughly 0.65 of the rate of water column denitrification in the same region.
 Were the N2 fixation input found to be the correct explanation for the Δ(15,18) minimum, it would indicate that a significant fraction of the NO3− loss to denitrification is subsequently compensated by N2 fixation in the surface waters overlying or adjacent to the zone of denitrification. This would explain why PO43−-bearing waters are not observed penetrating far into the eastern ranges of the Pacific subtropical gyres: N2 fixers strip out this P in the waters proximal to the upwelling zones. Moreover, it would bolster the view that oceanic N2 fixation is strongly controlled by N/P variations in the waters supplied to the surface, with diazotrophs succeeding under N-poor, P-bearing conditions [Broecker and Peng, 1982; Redfield, 1958; Tyrrell, 1999], a situation that has been demonstrated in lakes [Schindler, 1977; Smith, 1983].
 An alternative plausible mechanism for the development of the Δ(15,18) minimum is the redox cycling of NO3− and NO2− within suboxic zones. The logic is that NO3− δ18O can be gradually increased if the NO3− reduced to NO2− is lower in δ18O than the NO3− produced from the reoxidation of NO2−. However, the isotope dynamics of NO2− are poorly understood and essentially unknown in the case of the O isotopes. For reasonable assumptions, the mechanism can explain the Δ(15,18) minimum with a ratio of NO2− oxidation to NO2− reduction of as little as 0.7.
 Looking forward, several routes can be imagined that should allow for these two plausible explanations to be tested. First, work on the isotope systematics of NO2− (especially the O isotope systematics) is clearly needed and would provide an immediate test of the premises behind the NO3−/NO2− redox cycling scenario. Second, studies of other ocean regions, including model systems such as well-described isolated basins, would provide critical constraints on the coupled N and O isotopic effects of both N2 fixation and NO3− cycling through other oxidation states. For instance, it is not difficult to identify regions where N2 fixation is occurring without denitrification, and vice versa.
 The isotopic impact of redox cycling of NO3− and NO2− represents something of a liability in the current study because of the uncertainties in its isotope systematics, especially with regard to the O isotopes. However, one can imagine circumstances where the rate of cyclic consumption and production of NO3− could be well constrained by the N and O isotopes. At the base of the euphotic zone, a cycle of NO3− assimilation and remineralization back to NO3− should cause NO3− δ18O to rise above the 1:1 δ18O:δ15N increase expected from NO3− assimilation alone, because the δ18O of the NO3− being consumed by assimilation is lower than the δ18O of NO3− being produced by remineralization/nitrification [Granger et al., 2004b]. In this case, the N and O isotopes should allow for the NO3− recycling to be more accurately quantified.
 We thank M. Bender, K. Casciotti, C. Deutsch, N. Gruber, and B. Ward for discussions. This work was supported by U.S. NSF OCE-0136449 and Biocomplexity grants OCE-9981479 (to D. M. S., through the MANTRA project) and DEB-0083566 (to Simon Levin), and by British Petroleum and Ford Motor Company through the Carbon Mitigation Initiative at Princeton University. M. F. L. acknowledges support from the DFG through grant LE 1326/1-1. Cruise OXMZ01MV was supported by grant NSF OCE 9809026.