Role of nitrification and denitrification on the nitrous oxide cycle in the eastern tropical North Pacific and Gulf of California

Authors

  • Hiroaki Yamagishi,

    1. Department of Environmental Science and Technology, Interdisciplinary Graduate School of Science and Engineering, Tokyo Institute of Technology, Yokohama, Japan
    2. SORST Project, Japan Science and Technology Agency (JST), Kawaguchi, Japan
    3. Research Center for the Evolving Earth and Planets, Tokyo Institute of Technology, Okayama, Japan
    4. Now at Atmospheric Environmental Division, National Institute for Environmental Studies, Tsukuba, Japan.
    Search for more papers by this author
  • Marian B. Westley,

    1. Department of Oceanography, School of Ocean and Earth Science and Technology, University of Hawaii, Honolulu, Hawaii, USA
    2. Now at Geophysical Fluid Dynamics Laboratory, National Oceanic and Atmospheric Administration, Princeton, New Jersey, USA.
    Search for more papers by this author
  • Brian N. Popp,

    1. Department of Geology and Geophysics, School of Ocean and Earth Science and Technology, University of Hawaii, Honolulu, Hawaii, USA
    Search for more papers by this author
  • Sakae Toyoda,

    1. SORST Project, Japan Science and Technology Agency (JST), Kawaguchi, Japan
    2. Department of Environmental Chemistry and Engineering, Interdisciplinary Graduate School of Science and Engineering, Tokyo Institute of Technology, Yokohama, Japan
    Search for more papers by this author
  • Naohiro Yoshida,

    1. Department of Environmental Science and Technology, Interdisciplinary Graduate School of Science and Engineering, Tokyo Institute of Technology, Yokohama, Japan
    2. SORST Project, Japan Science and Technology Agency (JST), Kawaguchi, Japan
    3. Research Center for the Evolving Earth and Planets, Tokyo Institute of Technology, Okayama, Japan
    4. Department of Environmental Chemistry and Engineering, Interdisciplinary Graduate School of Science and Engineering, Tokyo Institute of Technology, Yokohama, Japan
    5. Frontier Collaborative Research Center, Tokyo Institute of Technology, Yokohama, Japan
    Search for more papers by this author
  • Shuichi Watanabe,

    1. Department of Environmental Science and Technology, Interdisciplinary Graduate School of Science and Engineering, Tokyo Institute of Technology, Yokohama, Japan
    2. SORST Project, Japan Science and Technology Agency (JST), Kawaguchi, Japan
    3. Mutsu Institute for Oceanography, Japan Agency for Marine-Earth Science and Technology, Mutsu, Japan
    Search for more papers by this author
  • Keisuke Koba,

    1. Department of Environmental Science and Technology, Interdisciplinary Graduate School of Science and Engineering, Tokyo Institute of Technology, Yokohama, Japan
    2. SORST Project, Japan Science and Technology Agency (JST), Kawaguchi, Japan
    3. Now at Institute of Symbiotic Science and Technology, Tokyo University of Agriculture and Technology, Tokyo, Japan.
    Search for more papers by this author
  • Yasuhiro Yamanaka

    1. Faculty of Environmental Earth Science, Hokkaido University, Sapporo, Japan
    2. Frontier Research Center for Global Change, Japan Agency for Marine-Earth Science and Technology, Yokohama, Japan
    Search for more papers by this author

Abstract

[1] Nitrous oxide (N2O) is an important atmospheric greenhouse gas and is involved in stratospheric ozone depletion. Analysis of the isotopomer ratios of N2O (i.e., the intramolecular distribution of 15N within the linear NNO molecule and the conventional N and O isotope ratios) can elucidate the mechanisms of N2O production and destruction. We analyzed the isotopomer ratios of dissolved N2O at a site in the eastern tropical North Pacific (ETNP) and a site in the Gulf of California (GOC). At these sites, the flux of N2O to the atmosphere is extremely high but denitrification activity in the oxygen minimum zone (OMZ) also reduces N2O to N2. We estimated the isotopomeric enrichment factors for N2O reduction by denitrification. The factor was −11.6 ± 1.0‰ for the bulk (average) N, −19.8 ± 2.3‰ for the center N (α-site nitrogen), −3.4 ± 0.3‰ for the end N (β-site nitrogen), and −30.5 ± 3.2‰ for the 18O of N2O. Isotopomer analysis of N2O suggests that nitrifiers should contribute to N2O production more than denitrifiers at the oxycline above the OMZs in the ETNP (50–80 m) and in the GOC (80–300 m). In contrast, denitrifiers should largely contribute to the N2O production and consumption in the OMZs both in the ETNP (120–130 m) and in the GOC (600–800 m). The N2O isotopomer analysis will be a useful tool for resolving the distribution of water masses that carry a signal of N loss by denitrification.

1. Introduction

[2] Nitrous oxide (N2O) is an important greenhouse gas [Ramanathan et al., 1985], and also plays a role in stratospheric ozone chemistry [Crutzen, 1970]. The ocean is an important source of N2O to the atmosphere; however estimates of its annual emission vary widely, for example, 3.0 ± 2 TgN [Intergovernmental Panel on Climate Change, 2001] (17% of the total emission) and 5.8 ± 2 TgN [Nevison et al., 2003] (33% of the total emission), because observations are limited and models of the oceanic N2O cycle are still evolving. Nitrification and denitrification are the main processes affecting the N2O cycle in the oceans [Suntharalingam and Sarmiento, 2000]. Under aerobic conditions, N2O is thought to be produced by autotrophic nitrifiers by chemical decomposition of hydroxylamine or of an intermediate between hydroxylamine and nitrite. Under low–oxygen conditions, N2O production by autotrophic nitrifiers is enhanced, because ammonia oxidation to nitrite is accompanied by reduction of nitrite to N2O [Poth and Focht, 1985; Ritchie and Nicholas, 1972], a process called nitrifier denitrification [Poth and Focht, 1985; Wrage et al., 2001].

[3] The eastern tropical North Pacific (ETNP) is one of the three major low-oxygen regions of the world's ocean, which account for most of the world's water column denitrification [Gruber and Sarmiento, 1997]. Annual N2O emission is high from such low-oxygen regions [Nevison et al., 1995], and has been estimated to be 25–50% of the total oceanic emission [Suntharalingam et al., 2000]. Short-term fluctuations between a nondenitrification mode and a denitrification mode in the ETNP and the Arabian Sea could impact global climate by changes in N2O emission [Suthhof et al., 2001]. Therefore, in order to predict future changes in atmospheric N2O concentration and to understand the impact of N2O on climate, it is essential to understand the N2O cycle in these low-oxygen regions in the oceans.

[4] The production processes of N2O have been the subject of study for many years. Cohen and Gordon [1978] measured N2O concentration in the ETNP and suggested that N2O is produced by nitrification and consumed by denitrification on the basis of the correlation between N2O concentration and apparent oxygen utilization (AOU) and the inverse correlation between N2O concentration and nitrate deficit. In contrast, Pierotti and Rasmussen [1980] suggested that denitrification acts as a net source of N2O under oxygen concentration of 0.1–0.3 mL L−1, which is equivalent to 4.6–13.7 μmol kg−1 O2 when sigmaθ is 26. Yoshida et al. [1984] suggested that the nitrogen isotope ratio of N2O in the ETNP indicates that subsurface water is a source of N2O and that the extremely oxygen-depleted water is a sink. Yoshinari et al. [1997] analyzed nitrogen and oxygen isotope ratios of N2O in the oxic/suboxic boundaries in the ETNP and the Arabian Sea, and suggested that dissolved N2O is enriched in 15N and 18O owing to N2O reduction (N2O→N2) by denitrification and could be an important source of atmospheric N2O enriched in 15N relative to tropospheric N2O. On the basis of the concentrations of nutrients and the bulk nitrogen and oxygen isotope ratios of N2O in the ETNP, Westley et al. [2001] suggested that N2O at the concentration maximum is produced by nitrification within the subsurface oxycline whereas denitrification consumes N2O in the core of the oxygen minimum zone (OMZ). However, the roles of microbial processes responsible for N2O production have not been resolved sufficiently, in part owing to a lack of adequate analytical tools.

[5] Recently, techniques have been developed to measure the isotopomer ratios of N2O (intramolecular distribution of 15N within the linear NNO molecule) [Toyoda and Yoshida, 1999; Brenninkmeijer and Röckmann, 1999]. Toyoda and Yoshida [1999] defined the center and end positions of the N2O molecule as α and β nitrogen, respectively, and the difference in isotope ratios between α and β-site nitrogen as the “15N-site preference” (SP = δ15Nα − δ15Nβ). Isotopomer analysis of N2O has been an important tool for resolving the N2O cycle in the oceans. For example, in the subtropical North Pacific gyre, the minima o f δ15N and δ18O indicated subsurface production of N2O and accounted for more than 40% of its emission from the surface to the atmosphere [Dore et al., 1998]. Subsequent isotopomer analysis of N2O constrained the ratio to 40–75% and moreover suggested that nitrification (via nitrifier-denitrification) accounted for the in situ N2O production [Popp et al., 2002].

[6] For relating isotopomer measurements to the production and consumption processes of N2O, isotopomeric enrichment factors are essential. Recently, isotopomeric enrichment factors for N2O production by nitrifiers [Sutka et al., 2003, 2004a] and denitrifiers [Sutka et al., 2006; Toyoda et al., 2005] have been studied. However, there are few reports on the isotopomeric enrichment factor for N2O reduction by denitrifiers, which is necessary in order to use N2O isotopomer ratios quantitatively. Thus the objectives of this research are, first, to estimate the isotopomeric enrichment factor for N2O reduction and, second, to resolve quantitatively production and consumption processes of nitrous oxide in the ETNP and Gulf of California (GOC) using N2O isotopomer analysis.

[7] Here we estimated the isotopomeric enrichment factor for N2O reduction in the core of the oxygen minimum zone in the ETNP. Applying this enrichment factor in a steady state model for the N2O release from denitrifiers, we evaluated the contribution of each production process to the N2O cycle in the ETNP and GOC. The isotopomer analysis suggests that nitrifiers play a more important role for N2O production than denitrifiers in the subsurface layers, whereas denitrifiers produce and consume N2O at the oxygen levels below about 1 μmol kg−1 in the upper and lower portions of the OMZs. We found that the SP of N2O is a useful tool for resolving the N2O cycle under the suboxic conditions in the oceans (Here we defined suboxic condition as 1 ≤ [O2] < 50 μmol kg−1).

2. Methods

2.1. Sampling

[8] Samples were collected in the ETNP (16°N, 107°W) at a site located in the core of the low-oxygen region during May–June 2000 on the R/V Revelle. The site is the same as Station 5 of the Eastern Pacific Redox Experiment (EPREX) research cruise [Sansone et al., 2001; Sutka et al., 2004b; Westley et al., 2001]. Samples were also collected in the central GOC (26.30°N, 110.13°W) during September 2001 on the R/V New Horizon. The oxygen concentration was determined by Winkler titration [Grasshoff et al., 1983]. Nitrite and ammonium were measured using colorimetry [Grasshoff et al., 1983] and the fluorescence method of Jones [1991], respectively. Samples for N2O concentration and isotopomer ratios were collected in 250-mL glass serum vials, preserved with HgCl2, and sealed with butyl rubber stoppers (see Popp et al. [2002] for details).

2.2. N2O Isotopomer Analysis

[9] For isotopomeric analysis, seawater was transferred from a serum vial to a sparging column and sparged with ultrapure helium (purity > 99.99995%). Water vapor and CO2 were removed by magnesium perchlorate and NaOH (Ascarite II, GL Sciences Inc., Tokyo, Japan), respectively, and then extracted gases were collected on a stainless steel trap immersed in liquid nitrogen. These gases were carried by helium through Ascarite II and magnesium perchlorate again, and then cryofocused. N2O was purified using a PoraPLOT-Q capillary column (25 m, 0.32 mm i.d., df = 10 μm, Varian, Inc., Palo Alto, USA) at 27°C. The N2O isotopomer ratios were determined by duplicate measurements on a Finnigan MAT 252 with modified collectors (Thermo electron corporation, Bremen, Germany). First the ratios of δ(45/44) and δ(46/44) and the N2O concentration were measured by monitoring mass (m/z) 44, 45, and 46 ions simultaneously. Then the ratio of δ(31/30) was measured by monitoring mass (m/z) 30 and 31 fragment ions. For the analysis of δ(31/30), samples were further purified using a precolumn inserted after the preconcentration trap in order to remove an interfering substance detected in mass (m/z) 31. The substance is most likely a fluorocarbon (m/z 31 from CF+) [Kaiser et al., 2003]. The precolumn is a 1/4 inch o.d., 150-cm-long s.s. tube packed with silica gel (80/100 mesh) [Toyoda et al., 2005]. The final isotopomer ratios were calculated following Toyoda and Yoshida [1999]. We adopted the definition of Toyoda and Yoshida [1999] for the isotopomer/isotope ratios of N2O, namely δ15Nα, δ15Nβ, δ15Nbulk (conventional δ15N), and δ18O. Isotopomer ratios of N2O were calibrated with a standard gas produced by thermal decomposition of NH4NO3 [Toyoda and Yoshida, 1999]. The precision was estimated on the basis of the measurement of a laboratory reference gas dissolved in pure water to be less than ±0.2‰ for δ15Nbulk, ±0.4‰ for δ15Nα, ±0.5‰ for δ15Nβ, ±0.4‰ for δ18O, and ±0.9‰ for SP (1σ) of dissolved N2O in duplicate 100 mL samples containing more than 20 nmol kg−1 N2O. Equilibrium concentration of N2O is calculated from potential temperature and salinity using the equation of Weiss and Price [1980].

2.3. Advection-Diffusion-Reaction Model

[10] We estimated the isotopomeric enrichment factors for N2O reduction from the concentration and isotopomer ratios of N2O at the ETNP site using one-dimensional, vertical advection-diffusion-reaction models containing one reaction. Assuming steady state and that the vertical diffusion coefficient (K) is independent of depth, we obtain the following differential equation:

equation image

where K and V are the diffusion coefficient and advection velocity, respectively, in the z direction. The concentration and reduction rate of nitrous oxide are denoted by N and Fz, respectively. The values of each parameter are described in the Appendix. Assuming that N2O reduction by denitrifiers can be a pseudo-first-order reaction of N2O with the N2O reduction rate (Fz) proportional only to the concentration of N2O, the rate Fz is described as follows:

equation image

where, m is a reaction rate constant for the concentration model (validation of this assumption is discussed in section 4.2 and section I of auxiliary material Text S1). The differential operators in equation (1) were replaced by the Euler method of forward-difference approximations. The detailed procedure for estimating enrichment factors follows the model of Cline and Kaplan [1975]. In the models, the step interval was set to 5 m. Fitting the model to the measured N2O concentration gives N2O reduction rates at each depth.

[11] In our model, we designate N2O isotope or isotopomer ratios with δi:

equation image

where δi is δ15Nbulk, δ18O, δ15Nα, or δ15Nβ, and std is a standard. Ri is 15Rbulk, 18R, 15Rα, or 15Rβ, respectively, which are {[14N15N16O] + [15N14N16O]}/2 [14N14N16O], [14N14N18O]/[14N14N16O], [14N15N16O]/[14N14N16O], or [15N14N16O]/[14N14N16O], respectively. The fractionation factors (αi) are defined following Mariotti et al. [1981],

equation image

where m is a reaction rate constant of the most abundant isotopomer, namely, 14N14N16O, for N2O reduction, i describes bulk-N, 18O, α, or β, and mi are reaction rate constants for either 15N- or 18O-isotopes, or either 14N15N16O- (15Nα-) or 15N14N16O- (15Nβ-) isotopomers, respectively. Enrichment factors (ɛi) are defined following Mariotti et al. [1981],

equation image

[12] This definition gives a negative value for an isotopic/isotopomeric enrichment factor, which is used in some reports [Mariotti et al., 1981; Ostrom et al., 2002]. The benefit of the definition is that it gives consistency between the value of the site preference of produced N2O and the isotopomeric enrichment factors for decomposition of N2O (see section II of auxiliary material Text S1). Substituting the concentrations of 15N- and 18O-isotopes, and 15Nα- and 15Nβ-isotopomers into equation (1), and using the definitions of δi and αi from equations (3) and (4), we obtain equation (6) for the isotopomer model,

equation image

where the primes refer to successive differentiation with respect to depth. The differential operators in equation 6 were replaced by the Euler method of forward-difference approximations [Cline and Kaplan, 1975]. The isotopomeric enrichment factors were estimated by fitting the models to the observed isotopomer ratios.

2.4. Interpreting Origins of N2O From Isotopomer Analysis

[13] When interpreting N2O isotopomer ratios of (δ15Nα, δ15Nβ, δ18O) quantitatively, it is reasonable to use isotopomeric enrichment factors of (ɛα, ɛβ, ɛ18O). However, quantitative analysis of δ18O-N2O is difficult because oxygen in a substrate, nitrite, can exchange with that in water during denitrification [Casciotti et al., 2002; Shearer and Kohl, 1988], and moreover there are no data on the isotope composition of O2 in the ETNP and GOC sites and of δ18O of nitrate in the GOC. On the other hand, the site preference of N2O is considered to be independent of the isotope ratios of the substrate during N2O production by nitrifiers and denitrifiers because the SP of N2O produced by nitrifiers [Sutka et al., 2003] and denitrifiers [Toyoda et al., 2005] is nearly constant even if the δ15Nbulk and δ18O of N2O are changed owing to the increase of the isotope ratios of residual substrates. Considering that SP of N2O should not be influenced by the isotope ratio of its substrates and that δ15Nbulk of N2O can be interpreted using enrichment factors of bulk nitrogen, analyzing N2O on a plot of SP versus δ15Nbulk should have more benefit than on a plot of δ15Nα versus δ15Nβ.

[14] Thus we demonstrate the validity of the analysis of isotopomer fractionation of N2O reduction on the plot of SP versus δ15Nbulk (see section II of auxiliary material Text S1 for details) and then we analyze the production processes of N2O in the ETNP and GOC on a plot of SP versus δ15Nbulk. In the case of a one step reaction (substrate → product), we define apparent isotopomeric fractionation factors of α-site or β-site nitrogens in N2O following Mariotti et al. [1981],

equation image
equation image

where subscripts of p and s are product and substrate, respectively. When equations (3) and (5) are inserted into equations (7a) and (7b), and then the logarithm is taken, the following equations are obtained:

equation image
equation image

We define the isotopomeric enrichment factor for SP in N2O as ɛSP ≡ ɛα − ɛβ. When the following conditions (equation (10)) can be applied, equations (8) and (9) can be combined to give equation (11),

equation image
equation image

Therefore a plot of SP versus δ15Nbulk can be substituted for a plot of δ15Nα versus δ15Nβ during the analysis of isotopomer effect for N2O production and consumption.

[15] Since N2O is an intermediate of denitrification, the isotopomer ratios of N2O inside the cell of denitrifiers and N2O dissolved in the water column should be different. Here we designate N2O inside denitrifiers and in the water column as N2Ovivo and N2Ositu, respectively. Assuming that the flux of N2O released from the cells of denitrifiers during denitrification (the flux from N2Ovivo to N2Ositu) and the consumption flux of dissolved N2O (the flux from N2Ositu to N2Ovivo) are negligible compared with the rate of denitrification (the fluxes from NO3 to N2Ovivo and N2Ovivo to N2), isotopomer ratios of N2O released from the cell of denitrifiers can be calculated [Barford et al., 1999]. The N2O isotopomer ratios should be controlled by production and reduction processes as follows:

equation image

where the bold italic symbols of equation image and δ indicate the matrices of (δ15Nbulk, SP). The SP of δ-NO3 is recognized as zero for convenience. The symbols of ɛNaR and ɛN2OR indicate the isotopomeric enrichment factors for nitrate reduction to N2O and for N2O reduction to N2, respectively. The instantaneous products of N2O and N2 are expressed as δ-N2OD-pro and δ-N2D-pro, respectively. Barford et al. [1999] suggest the following correlations under steady state:

equation image
equation image
equation image

Isotopomer ratios of N2O produced by denitrifiers (δ-N2O D-pro) and N2O released from denitrifiers (δ-N2O D-rel) are calculated from equations (13), (14), and (15) as follows:

equation image
equation image

N2O is released by nitrifiers and denitrifiers, and simultaneously dissolved N2O is consumed by denitrifiers, therefore the isotopomer ratios of dissolved N2O (δ-N2Ositu) can be written as follows:

equation image

where t (0 ≤ t ≤ 1) is the ratio of the N2O released from nitrifiers to the total N2O released into the water column (abbreviated as N-ratio), δ-N2ON-pro represents the isotopomer ratios of N2O produced by nitrifiers, and ν is an arbitrary positive quantity. Because the parameters of δ-N2Ositu, δ-N2ON-pro, δ-N2OD-rel, and equation imageN2OR are the matrixes of δ15Nbulk and SP (δ15Nbulk, SP), ratios of N2O released from nitrifiers and denitrifiers can be estimated on the basis of plots of SP versus δ15Nbulk (Figure 1). The ratio t is estimated following equation (19) (see section III of auxiliary material Text S1 for the derivation of equation (19)).

equation image

where each plot is defined as follows: δ-N2OD-rel (x1, y1), δ-N2ON-pro (x2, y2), and δ-N2Ositu (x3, y3). The constant, a, is equivalent to the slope of the regression line for N2O reduction on the SP-δ15Nbluk plot. The error of the ratio t is estimated following propagation of errors in equation (19) using a = 1.43 ± 0.14, which was estimated from the regression line over the depth range 400–700 m at the ETNP site.

Figure 1.

Plots of the site preference (SP) versus δ15Nbulk of N2O for estimating the contribution of nitrification and denitrification as the production processes of N2O. The t is the ratio of N2O released from nitrifiers to the total N2O released from both nitrifiers and denitrifiers. Each notation is defined in text (section 2.4).

3. Results

[16] The N2O concentration maximum (85 nmol kg−1) in the ETNP was found at the depth of 65–85 m in the oxycline (Figure 2a). Below 85 m, N2O concentration decreased rapidly to 7.8 nmol kg−1 (130 m), which is below the equilibrium concentration. The O2 concentration was <1.5 μmol kg−1 O2 below 100 m (Table 1). A secondary N2O concentration maximum (56.5 nmol kg−1) was observed at 800 m under suboxic conditions (2.7 μmol kg−1 O2). These features are consistent with previous observations in the ETNP [Cohen and Gordon, 1978; Nevison et al., 2003; Pierotti and Rasmussen, 1980; Yoshinari et al., 1997].

Figure 2.

Concentration and isotopomer ratios of N2O and concentrations of O2 and nitrogen compounds plotted versus depth in (a, b, c) the ETNP and (d, e, f) the Gulf of California. Figures 2a and 2d show N2O concentration (open diamond), O2 concentration (solid circle), equilibrium N2O concentration (dashed line), salinity (solid line), and potential temperature (dotted line). Figures 2b and 2e show δ15Nbulk (solid diamond), δ15Nα (solid square), δ15Nβ (solid triangle), δ18O (solid circle), and 15N-site preference (solid star) of dissolved N2O; δ15Nbulk (open diamond), δ15Nα (open square), δ15Nβ (open triangle), δ18O (open circle), and 15N-site preference (open star) of tropospheric N2O [Yoshida and Toyoda, 2000]. Figures 2c and 2f show nitrate (solid circle), nitrite (solid diamond), and ammonia (solid triangle) concentrations, and N* (solid star). In Figure 2f, nitrate concentration and N* were calculated from the annual analyzed mean nutrient, salinity, and temperature at 26.5°N, 110.5°W [Boyer et al., 2002; Conkright et al., 2002; Stephens et al., 2002]. All isotopic values are expressed as permil (‰) deviations from atmospheric N2 (for nitrogen) and VSMOW (for oxygen). The equilibrium concentration of N2O was calculated following Weiss and Price [1980].

Table 1. Concentration and Isotopomer Ratios of N2O in the Eastern Tropical North Pacific and Gulf of California
Depth, mSigmaθO2, μmol kg−1N2O, nmol kg−1Saturation,a %δ15Nbulk, ‰δ15Nα, ‰δ15Nβ, ‰δ18O, ‰SP, ‰
  • a

    Saturation (%) = [N2O]measured/[N2O]equilibrium × 100. Equilibrated concentration was calculated following Weiss and Price [1980].

  • b

    Oxygen concentration was measured from a different cast with the cast of N2O.

Eastern Tropical North Pacific (16°N, 107°W)
1022.34205.4110.071726.1513.31−1.0246.4514.33
2522.42205.4910.371776.2312.56−0.1046.3712.66
4522.55185.6422.453794.5110.40−1.3846.8711.78
5523.48105.9452.378103.758.92−1.4247.1110.35
6525.0230.7784.7910993.609.93−2.7447.8112.68
8525.675.5484.5710094.6211.44−2.2055.2113.64
10526.020.6526.893068.2623.43−6.9178.1930.34
12026.121.4110.581199.5829.43−10.2789.5239.70
13026.160.407.84889.7028.71−9.3092.2638.01
16526.380.47 b6.106614.3731.96−3.2293.3435.17
35026.730.193.803821.8235.088.57100.7526.50
40026.830.396.276022.7446.42−0.95105.3047.37
50026.960.769.829118.5040.56−3.5695.1544.12
60027.090.2822.1919811.7327.58−4.1278.1931.70
70027.190.4547.484117.8521.47−5.7766.6427.24
80027.272.7056.514777.7118.73−3.3059.9722.03
125027.4927.5041.773309.5221.93−2.8956.5124.82
151027.5747.4935.922779.7621.73−2.2056.8823.93
199527.6680.1328.852169.9221.15−1.3257.0622.47
 
Gulf of California (26.30°N, 110.13°W)
523.38208.8112.191946.9014.47−0.6751.2915.14
3023.43216.7712.612014.1414.20−5.9250.4220.12
7524.7967.8834.404595.7012.15−0.7554.0812.89
15025.9034.2133.583858.6017.62−0.4155.5218.03
30026.383.0540.5441110.9719.262.6862.2116.58
40526.591.2248.204609.9020.78−0.9963.1221.76
50526.760.6153.874906.8919.50−5.7260.3425.23
55526.830.6158.045176.5518.78−5.6961.0924.46
60526.880.6166.595844.9519.40−9.4958.8128.88
70526.940.6182.177061.9416.55−12.6753.3929.22
80527.031.2277.846521.7615.68−12.1751.8827.85
97027.122.4475.406143.6216.30−9.0657.6025.35

[17] N2O production is enhanced under suboxic conditions; thus N2O should be produced at the oxycline around 75 m, where the N2O concentration maximum formed a sharp peak in the N2O concentration profile at the ETNP site. Potential temperature and salinity at the depth of 50–75 m plot linearly on a T-S diagram, which indicates the mixing of two water masses. Thus N2O produced at the oxycline should diffuse to the mixed layer by vertical mixing and then be released to the troposphere by air-sea gas exchange. N2O isotopomer ratios at the surface were similar to those of the troposphere (Figure 2b), which indicates that N2O in the troposphere should be dissolved at least in the mixed layer. Assuming that the N2O profile from the N2O concentration maximum to the surface is formed by mixing of the background tropospheric N2O and the N2O produced at the oxycline, we can use a Keeling Plot, a plot of isotope ratios versus the reciprocal N2O concentration (Figure 3), to estimate the isotopomer ratios of N2O produced at the oxycline. Intercepts of regression lines estimate isotopomer ratios of a produced substance [Pataki et al., 2003].

Figure 3.

Isotopomer ratios of N2O versus reciprocal N2O concentration at the ETNP site (10–65 m). The value of intercept represents isotopomer ratio of a source (i.e., an end-member). The end-member equals the isotopomer ratios of N2O produced at the concentration maximum. Each plot represents δ15Nbulk (open diamond), δ15Nα (open square), δ15Nβ (open triangle), and δ18O (open circle).

[18] The isotopomer ratios of N2O produced at the subsurface N2O maximum in the ETNP were 3.2 ± 0.1, 8.7 ± 0.6, −2.3 ± 0.6, 47.6 ± 0.3, and 11.0 ± 1.2‰, for δ15Nbulk, δ15Nα, δ15Nβ, δ18O, and the site preference, respectively (1σ, n = 5). The δ15Nbulk and δ18O estimates are similar to the measured subsurface values in the ETNP (δ15Nbulk = 5 ± 2‰ and δ18OVSMOW = 50 ± 2‰ [Yoshinari et al., 1997]) and the Arabian Sea (δ15Nbulk = 7 ± 2‰ and δ18OVSMOW = 49 ± 2‰ [Yoshinari et al., 1997]; δ15Nbulk = 2 ± 2‰ and δ18OVSMOW = 47 ± 6‰ [Naqvi et al., 1998]). The δ15Nbulk estimate is also similar to the δ15Nbulk estimated from Keeling Plots in the subtropical gyre (δ15Nbulk = 3.7‰ [Popp et al., 2002]). The remarkable consistency of the isotope ratios of N2O strongly suggests that the production process is similar in the subsurface layer despite the large disparity in geographic regions. The consistent δ15N values of the y-intercept of the Keeling Plot for such a wide range of oceanic environments is startling and further suggests uniformity in the processes that form N2O in oceanic surface waters.

[19] At the GOC site, the suboxic interface was located at about 300 m (Figure 2d) and the N2O concentration was highest at 700 m, well within the oxygen minimum zone (O2< 1 μmol kg−1). At 700 m, site preference was high, whereas δ15Nbulk and δ18O were low, relative to those of N2O at the suboxic interface (300 m; see Figure 2e). Incubation of sediments indicates that the dominant source of N2O is denitrification at oxygen levels <1 μmol kg−1 [Jørgensen et al. [1984]. Considering the low O2 concentration (<1 μmol kg−1) and the negative N* (Figure 2f), which is an indicator for nitrate deficit (N* = [NO3] − 16 [PO43−] + 2.90 μmol kg−1 [Deutsch et al., 2001]), denitrification should be the primary source of N2O at 700 m.

4. Discussion

4.1. Nitrous Oxide Cycle in the Oxygen Minimum Zone of the ETNP

[20] At oxygen levels below 1 μmol kg−1, the consumption of N2O by denitrification becomes the dominant process and drives N2O concentrations below equilibrium concentrations [Codispoti et al., 1992; Prosperie et al., 1996; Suntharalingam et al., 2000]. Actually, N2O concentration was lower than its equilibrium concentration at the depth of 130–500 m at the ETNP site (Table 1 and Figure 2a), which indicates that N2O reduction by denitrifiers is active. Therefore isotopomer ratios of N2O should be affected by the isotope effect associated with N2O reduction by denitrifiers.

[21] The δ15Nα, δ15Nβ and δ15Nbulk of N2O are all linearly correlated with the δ18O of N2O from 400 to 700 m (Figure 4a). Strong linear correlation between these three independent parameters (δ15Nbulk, δ18O, and SP) is strong evidence for either isotopomer fractionation associated with a one-step reaction or mixing of two different pools of N2O. Since N2O consumption is surely active at oxygen levels below 1 μmol kg−1 and mixing of waters by cross-isopycnal diffusion across 300 m is unlikely to account for the linear correlation, we suggest that the highly linear correlations between 400 and 700 m (Figure 4a) result from N2O consumption by denitrification rather than by mixing between isotopically distinct pools of N2O. The δ15Nbulk, δ15Nα, δ18O, and SP of N2O increased as N2O concentration decreased, and the δ15Nβ slightly increased from 700 m to 400 m (Figures 2a and 2b). This correlation is consistent with kinetic isotope effects associated with the cleavage of the N-O bonds during N2O reduction by denitrification [Popp et al., 2002; Toyoda et al., 2002; Westley et al., 2006]. Therefore the concentration and isotopomer ratios of N2O from 700 to 400 m suggest that the N2O is produced in the suboxic zone (the area below 750 m, where [O2] ≥ 1 μmol kg−1) and then diffuses and advects into the OMZ up to 400 m, which is the strongest anoxic zone at the ETNP site on the basis of ammonium accumulation (Figure 2c).

Figure 4.

Relationship between nitrogen isotope/isotopomer ratios and oxygen isotope ratio of N2O (a) between 400 and 700 m and (b) between 65 and 130 m at the ETNP site. Each plot represents δ15Nbulk (open diamond), δ15Nα (open square), δ15Nβ (open triangle), and 15N-site preference (cross). Regression lines of 400–700 m are as follows, where a cross denotes δ18O and the value in parentheses is R2 (n = 4): δ15Nbulk = −18.21 + 0.387 X (0.998), δ15Nα = −23.36 + 0.665 X (0.995), δ15Nβ = −13.05 + 0.109 X (0.890), and SP = −10.31 + 0.555 X (0.981). Regression lines of 65 − 130 m are as follows (n = 5): δ15Nbulk = −3.14 + 0.142 X (0.995), δ15Nα = −12.94 + 0.462 X (0.989), δ15Nβ = 6.66 − 0.178 X (0.944), and SP = −19.60 + 0.640 X (0.980).

[22] Linear correlations between δ18O-N2O and δ15Nα-, δ15Nβ-, or δ15Nbulk-N2O were also observed over the depth range 65–130 m (Figure 4b). At 65 m, low δ15Nbulk, δ18O, and SP of N2O relative to that in the troposphere can be attributed to the N2O production under suboxic conditions. From 65 m to 130 m, oxygen concentration decreases with depth, which results in the shift of the N2O production process from nitrification to denitrification, and the increase of the ratio of N2O reduction to its production. This suggestion is consistent with the profile of N* (Figure 2c). The value of N* decreased as depth increased to 100 m, which suggests denitrification is active at least below the depth of 100 m. Over the depth range 65–130 m, δ15Nβ-N2O decreased with depth, which could be caused by production of N2O with low δ15Nbulk and subsequent consumption of dissolved N2O. A decrease of δ15Nβ accompanied by increases of δ18O and SP was also observed in the Black Sea [Westley et al., 2006]. Therefore the correlation between δ18O-N2O and δ15Nα-, δ15Nβ-, or δ15Nbulk-N2O at 65–130 m should be the result of a shift in the production and consumption processes of N2O. Potential temperature and salinity at 65–130 m follow a linear trend on a T-S diagram (data not shown), which indicates the mixing of two water masses. Therefore the good correlation between N2O isotopomer ratios over the depth range 65 to 130 m could be attributed to mixing of water masses. (Correlation coefficient (R2) is listed in the caption of Figure 4).

4.2. Isotopomeric Enrichment Factors for N2O Reduction

[23] If in situ production or isopycnal diffusion affects the distribution of N2O between 400 and 700 m at the ETNP site in addition to the in situ N2O reduction, then N2O isotopomer ratios should plot as curves in δ15N versus δ18O space. The observed linear correlation between N2O isotopomer ratios between 400 and 700 m indicates that only one of these processes affects the N2O pool. The OMZs in the ETNP are considered as stagnant layers ventilated only by mixing with surface and deep oxygenated waters [Wyrtki, 1962]. Actually, distribution of N* at Sigmaθ = 26.9 and 27.1, which are equivalent to ∼ 450 and 600 m, respectively [Castro et al., 2001], indicates that the water mass at the ETNP site is stagnant and appears to be isolated from other water masses. Therefore it is reasonable to assume that only in situ N2O reduction affects the pool of N2O and to analyze the N2O isotopomer ratios between 400 to 700 m using vertical advection-diffusion-reaction models.

[24] There are several factors that could control N2O reduction rate, such as N2O, O2, or nitrate concentrations, or potential denitrification activity. Oxygen and nitrate can compete with N2O as electron accepters. However, the oxygen concentration was under 1 μmol/kg from 400–700 m and showed no concentration gradients. Therefore it is not necessary to include O2 concentration among the factors controlling N2O reduction rate as a function of depth. Negative N* from 400–700 m indicates that nitrate is a competitor of N2O over this depth interval (Figure 2c). The concentration of nitrate was approximately a thousand times higher than the N2O concentration. Therefore N2O concentration rather than nitrate concentration seems to constrain the N2O reduction rate. During nitrate reduction, organic matter is thought to regulate the nitrate reduction rate because denitrification activity can be limited by the supply of electron donors (organic compounds) rather than electron accepters such as nitrate [Cline and Kaplan, 1975; Sutka et al., 2004b]. However, considering that the nitrate reduction rate is likely at least 2 orders of magnitude higher than the N2O reduction rate, it is probable that electron donors should be sufficient for N2O reduction and that N2O reduction can be a pseudo-first-order reaction of N2O. Thus we developed the advection-diffusion-reaction model with the assumption that the N2O reduction rate is only proportional to the concentration of N2O (see section 2.3 for the description of the model). We also tested the advection-diffusion-reaction model with the assumption that the N2O reduction rate is proportional to the concentration of both N2O and electron donors (see section I of auxiliary material Text S1). Although the second model contains large uncertainties in POC concentration, the estimated isotopomeric enrichment factors agree with those of the model based on N2O concentration within errors.

[25] We estimated isotopomeric enrichment factors for N2O reduction (see Table 2 and Figure 5 for the results) by applying a vertical advection-diffusion-reaction model to the N2O concentration and isotopomer ratios over the depth range 400–700 m at the ETNP site. Current physical oceanographic theory suggests that the predominant form of ventilation in the oceans below the mixed layer is mixing along isopycnal surfaces [Ledwell et al., 1993; Luyten et al., 1983]. Enrichment factors for nitrate reduction by denitrification estimated by lateral reaction-diffusion models (ɛ = −30 ± 3‰ [Brandes et al., 1998]; ɛ = −30 ± 7.5‰ [Voss et al., 2001]) are identical within the range of the errors to those estimated using vertical advection-diffusion-reaction models (ɛ = −30 ± 5‰: [Sutka et al., 2004b] [see also Cline and Kaplan, 1975]). The agreement between these results on nitrate suggests that our estimate of the enrichment factors for N2O should be reliable in the ETNP.

Figure 5.

(a) Concentration and (b) isotopomer ratios of N2O at the ETNP site versus depth. Solid lines are best fit, advection-diffusion-reaction model results. Figure 5a shows observed concentration (open circle), and Figure 5 b shows δ15Nbulk (open diamond), δ15Nα (open square), δ15Nβ (open triangle), and δ18O (open circle) of observed N2O. (c) N2O reduction rates calculated by the model with different values of diffusion coefficient (K) and advection velocity (V). The values of V/K and V for each curve named in legend are listed in Table A1. The solid curve named as “Optimal-high” is the “best fit” result using the values of V/K = −1.4 × 10−3 m−1 and V = −4.0 m yr−1.

Table 2. Isotopomeric Enrichment Factors for N2O Reductiona
Isotopomeric enrichment factor, ‰ConditionsReference
ɛbulk-Nɛαɛβɛ180ɛ180bulk-Nb
  • a

    N2O → N2.

  • b

    The ratio of ɛ18Obulk-N is equivalent to the slope of the regression line on δ18O versus δ15N plot (n = 4, R2 = 0.998). The error of ɛ18Obulk-N is equivalent to the error of the slope of the regression line.

  • c

    Average value of replicate experiments.

  • d

    A N2-fixing bacterium.

−11.6 ± 1.0−19.8 ± 2.3−3.4 ± 0.3−30.5 ± 3.22.58 ± 0.08eastern tropical North Pacific(advection-diffusion-reaction models)this study (400–700 m in the ETNP)
−6.9– −9.8  −12.6– −24.92.51incubation of soil collected under larch in SiberiaMenyailo and Hungate [2006]
−6.3– −8.3  −16– −212.55– 2.56incubation of soil collected under birch in SiberiaMenyailo and Hungate [2006]
−12.9cchemostat culture of Paracoccus denitrificansBarford et al. [1999]
−39pure culture of Azotobacter vinelandii at 25°CdYamazaki et al. [1987]
−42pure culture of Pseudomonas aeruginosa at 10°CWahlen and Yoshinari [1985]
−37pure culture of Pseudomonas aeruginosa at 26°CWahlen and Yoshinari [1985]
−2.4−4.92.0incubation of landfill soil in ambient air at 24°CMandernack et al. [2000]
∼2correlation between δ15Nbulk and δ18O in N2O dissolved in the eastern tropical North PacificYoshinari et al. [1997]
∼3correlation between δ15Nbulk and δ18O in N2O dissolved in the Arabian SeaYoshinari et al. [1997]

[26] The enrichment factor was estimated at −11.6 ± 1.0‰ for the bulk (average) N, −19.8 ± 2.3‰ for the center N (α-site nitrogen), −3.4 ± 0.3‰ for the end N (β-site nitrogen), −30.5 ± 3.2‰ for the 18O of N2O (Table 2). The range of the enrichment factors is based on a sensitivity test (see Appendix A). Our estimated enrichment factors of bulk nitrogen and oxygen for N2O reduction are within the range of reported values (Table 2). Our enrichment factor of bulk nitrogen is similar to that of Barford et al. [1999] (−12.9‰), which was estimated using pure cultures of a denitrifier, Paracoccus denitrificans, grown under steady state in a chemostat. Ostrom et al. [2007] also estimated the enrichment factor of bulk nitrogen, using a denitrifier, Paracoccus denitrificans, as −10.9‰. The oxygen isotopic enrichment factor of our estimation is similar to that of Wahlen and Yoshinari [1985] (−37‰), which was estimated by pure cultures of a denitrifier, Paracoccus aeruginosa, grown in batch culture. The ratio of oxygen to nitrogen isotope fractionation at 400–700 m at the ETNP site (ɛ18Obulk-N = 2.58 ± 0.08; see Table 2 for details) is similar to the ratios of N2O reduction by a denitrifier, Paracoccus denitrificans18Obulk-N = 2.28 [Ostrom et al., 2007]), and soils (ɛ18Obulk-N = 2.5 [Ostrom et al., 2007], 2.51–2.56 [Menyailo and Hungate, 2006], and 2.0 [Mandernack et al., 2000]), although soil incubation experiments give smaller enrichment factors for bulk nitrogen and oxygen. Moreover, the ratio of oxygen to nitrogen enrichment factors (ɛ18Obulk-N) at the ETNP site is also similar to the isotope fractionation observed in the ETNP (Δδ18O/Δδ15Nbulk ≈ 2) and in the Arabian Sea (Δδ18O/Δδ15Nbulk ≈ 3) [Yoshinari et al., 1997].

[27] A constant 1:1 ratio of 15N/14N and 18O/16O isotope fractionation in nitrate has been observed during denitrification [Durka et al., 1994; Bottcher et al., 1990; Sigman et al., 2005] and nitrate assimilation by marine phytoplankton [Granger et al., 2004]. Granger et al. [2004] suggested that total isotope fractionation in phytoplankton consists of two steps, namely reversible flow across the membrane of a cell (first step) and enzyme reaction (second step), and that isotope fractionation for diffusion of substrates into and out of a cell is negligible. The two-step model has also been suggested to explain nitrogen isotope fractionation for nitrate reduction by denitrifiers [Mariotti et al., 1982; Ostrom et al., 2002, Sigman et al., 2005,2006]. This two-step model, which is similar to the two-step models describing carbon isotope fractionation [e.g., Laws et al., 1995], could also apply to isotopomer fractionation for N2O reduction by denitrifiers as follows:

equation image

Assuming steady state, there are steady flows, Fab, Fba, and Fbc from A to B, from B to A, and from B to C, respectively. Isotopomeric enrichment factors are described as equation imageab, equation imageba, and equation imagebc, respectively, for each step. In case of N2O reduction, A is N2O outside a cell, B is N2O inside a cell, and C is the N2 inside a cell. Then, isotopomeric enrichment factor for N2O reduction can be calculated by the equation of Rees [1973],

equation image

where Xb = Fba/Fab.Isotope fractionations for diffusion of nitrate [Bryan et al., 1983; Granger et al., 2004; Mariotti et al., 1982], N2 [Handley and Raven, 1992], and CO2 [Farquhar et al., 1982; O'Leary, 1984] through the cell membrane (equation imageab and equation imageba) are thought to be negligible. Assuming that isotopomer fractionations for diffusion (equation imageab and equation imageba) of N2O are also negligible, the ratios between isotopomeric enrichment factors for N2O reduction are constant, even if the ratio Xb changes. The steady state denitrifier model of Barford et al. [1999] (see section 2.4 for details) estimates the isotopic enrichment factors for enzyme reaction (equation imagebc). Similarity of nitrogen isotopic enrichment factors estimated by Barford et al. [1999] (equation imagebc) and this study (equation imageac) suggests that the ratio Xb at 400–700 m at the ETNP site is close to 1 (equation (21) estimates the ratio Xb.).

4.3. Estimating the Contribution of Nitrifiers and Denitrifiers to N2O Production

[28] We analyzed the contribution of nitrifiers and denitrifiers to N2O production at the ETNP and GOC sites using SP-δ15Nbulk plot (see section 2.4 for description and see Figure 6 for the results). For the SP-δ15Nbulk analysis, we calculated the isotopomer ratios of N2O released by denitrifiers (δ-N2OD-rel) and adopted the reference value estimated by Popp et al. [2002] for the isotopomer ratios of N2O released by nitrifiers (δ-N2ON-pro) (see Tables 3a and 3b for detail calculations). We estimated the range of nitrogen isotope ratio in nitrate in the ETNP and GOC on the basis of previous studies at the depth of our observations (see references in Table 3a). The value of δ-N2OD-rel was calculated using equations (16) and (17), and isotopomeric enrichment factors of equation imageNaR (NO3 → N2O) and equation imageN2OR (N2O → N2) listed in Table 3b. Variation of δ-N2OD-rel was determined using a propagation of error calculation.

Figure 6.

Plots of (a) 15N-site preference versus δ15Nbulk and (b) δ15Nbulk versus δ18O. Symbols representing the observations and previous reports are defined in legend. Figure 6a shows isotopomer ratios of N2O produced by nitrifiers (red pentagon) and denitrifiers (open triangle), and isotopomer ratios of N2O released from denitrifiers (solid triangle). Red dash-dotted line and blue dotted line are in parallel with the isotopomer fractionation for N2O reduction. In Figures 6a and 6b, orange and blue numbers are the depths for each sample at the ETNP and GC sites, respectively. Black solid line and broken line are the regression lines between 400 and 700 m and between 65 and 130 m, respectively, at the ETNP site. All isotopic values are expressed as permil (‰) deviations from atmospheric N2 (for nitrogen) and VSMOW (for oxygen).

Table 3a. Estimates of Isotopomer Ratios of N2O Released From Nitrifiers and Denitrifiers
Isotopomer Ratios, ‰Symbolδ15NSPNotes
  • a

    The data at the depth of 200–600 m is excluded for the eastern tropical North Pacific because nitrate at the depth should not be the substrate of the N2O that we analyze.

  • b

    Range of the value is calculated from propagation of errors: σ(A) = [{σ(B)}2 + {σ(C)}2](1/2) when A = B + C.

Nitrate in the eastern tropical North Pacific and Gulf of Californiaδ-NO37.5 ± 6.50the eastern tropical North Pacific: Brandes et al. [1998], Cline and Kaplan [1975], Sigman et al. [2005], Sutka et al. [2004b], and Voss et al. [2001]; the Gulf of California: Altabet et al. [1999]a
N2O produced by denitrifiersδ-N2OD-pro−22.5 ± 10.0−2 ± 5bδ-N2OD-pro = δ-NO3 + equation imageNaR
N2O released from denitrifiersδ-N2OD-rel−10.9 ± 10.014.4 ± 5.5bimage = δ-N2D-proequation imageN2OR
N2O produced by nitrifiersδ-N2ON-pro4 ± 14 ± 4N2O produced in the subsurface layer in the subtropical North Pacific Gyre [Popp et al., 2002]
Table 3b. Isotopomeric Enrichment Factors for the Calculation of N2O Isotopomer Ratios of Sources
Isotopomeric Enrichment Factors, ‰Symbolɛbulk-NɛSPaNotes
  • a

    ɛSP is calculated by the following equation: ɛsp = 2 ɛα − 2 ɛbulk-N. Range of the value is calculated from propagation of errors: σ(SP) = {4 σ(δ15Nα)2 + 4 σ(δ15Nbulk)2}1/2.

Nitrate reduction by denitrifiers: NO3 → N2OδNaR−30 ± 7.5−2 ± 5ɛbulk-N: lateral diffusion-reaction model in the ETNP [Brandes et al., 1998; Voss et al., 2001] and vertical advection-diffusion-reaction model in the ETNP [Cline and Kaplan, 1975; Sutka et al., 2004b], ɛSP: pure culture of denitrifiers: Pseudomonas chlororaphis,Pseudomonas aurelofaciens [Sutka et al., 2006], and Paracoccus denitrificans [Toyoda et al., 2005]
N2O reduction by denitrifiers: N2O → N2δN2OR−11.6 ± 1.0−16.4 ± 2.3Vertical advection-diffusion-reaction model in the ETNP (this study)

[29] There are two choices for the isotopomer ratios of N2O produced by nitrifiers (δ-N2ON-pro): estimation based on isotopomer analysis of N2O produced presumably by nitrifier-denitrification in the shallow aphotic zone in the subtropical North Pacific Gyre [Popp et al., 2002] or calculation using isotopomeric enrichment factors estimated from pure cultures [Sutka et al., 2006, 2003,2004a]. Sutka et al. [2006, 2003, 2004a] found that for pure culture experiments SP of N2O is close to 0‰ when N2O is produced via nitrite by nitrite reductase and that SP of N2O is around 30‰ when N2O is produced by NH2OH oxidation. At station ALOHA (22°45′N, 158°00′W) in the aphotic zone in the subtropical North Pacific Gyre, SP of N2O produced in situ was estimated at ∼4‰ [Popp et al., 2002]. The low SP observed in this subsurface layer at station ALOHA and the ETNP site is consistent with the production of N2O by nitrifier denitrification as observed in pure culture experiments by Sutka et al. [2006, 2003, 2004a]. Comparison of δ18O-N2O with δ18O-O2 also indicates the contribution of nitrifier denitrification for N2O production at the depth of 350–500 m at station ALOHA [Ostrom et al., 2000; Popp et al., 2002]. However, we do not adopt the SP of 0‰ for nitrifier denitrification as the end-member of δ-N2ON-pro, because we could not exclude the possibility of contribution of NH2OH oxidation to N2O production in the aerobic oceans. It is also difficult to use isotopic enrichment factors for the N2O production by nitrifiers, because there are no data for nitrogen isotope ratios of substrates available and, furthermore, calculation of the nitrogen isotope ratio of N2O produced by marine nitrifiers using isotopic enrichment factors is not straightforward. Casciotti et al. [2002] found that the extent of isotope fractionation for ammonia oxidation by marine nitrifiers is about 10–25‰ smaller than that of a terrestrial nitrifier such as Nitrosospira tenuis and a well-studied nitrifier, Nitrosomonas europaea. Therefore, considering the difficulties in estimating the value for δ-N2ON-pro using isotopomer enrichment factors and the consistency in isotopic values observed in ocean environments, we adopt the isotopomer ratios of N2O produced in situ in the aerobic subsurface layer in the subtropical North Pacific Gyre [Popp et al., 2002] as the value of δ-N2ON-pro (Table 3a).

[30] To calculate the isotopomer ratios of N2O released from denitrifiers (δ-N2OD-rel), we assumed that the ratio Xb in the two-step model is one. Then we applied our estimated enrichment factor equation imageN2OR(= equation imageac; N2Ositu → N2) to equation imagebc (N2Ovivo → N2) of the steady state denitrifier model to calculate δ-N2OD-rel (Table 3b). However, even if the ratio Xb is far from one, the slope of SP to δ15Nbulk is constant. Thus the ratio Xb should not affect the analysis of the origin of N2O.

[31] The isotopomer ratios of N2O at 700 m at the GOC site and at 120 m at the ETNP site plot close to the isotopomer fractionation line for N2O reduction from δ-N2O D-rel (blue dotted line at Figure 6a), which suggests that N2O is released from and consumed by denitrifiers in these water masses. On the contrary, the isotopomer ratios of N2O dissolved at the N2O concentration maximum (∼65 m) at the ETNP site and the suboxic zone above the OMZ (∼300 m) at the GOC site plot close to the isotopomer fractionation line of N2O reduction from δ-N2O N-pro (red dash-dotted line at Figure 6a), which suggests that N2O is mainly released from nitrifiers in these water masses.

[32] We estimated the contribution ratio of N2O production by nitrifiers to the total N2O emission (N-ratio) and its error using equation 19 and propagation of error, respectively. Quantitative analysis indicates that 71 ± 17% of the dissolved N2O is released from nitrifiers at the subsurface N2O maximum (at 65 m) at the ETNP site. In the suboxic zone under the core of the OMZ (at 800 m) at the ETNP site, the estimated N-ratio was 60 ± 21%. Nitrifiers primarily contribute to N2O production at 55 m at the ETNP site (N-ratio = 79 ± 15%), and at 75 m at the GOC site (N-ratio = 80 ± 15%). The model results are consistent with the minimum SP values found at these depths, which is a signal of nitrifier-denitrification (Figures 2b and 2e), and with the nitrite concentration peak, a signal of ammonia oxidation by ammonia oxidizers (Figure 2c). N-ratios were also estimated at 92 ± 13% and 11 ± 43% at 300 m and 700 m at the GOC site, respectively.

[33] In order to resolve the contribution of each process more precisely, evaluation of the tropospheric N2O is necessary in subsurface layers and aerobic waters. In contrast, in suboxic zones where N2O is produced and consumed by denitrification, the modern and preindustrial tropospheric N2O should be consumed and thus should not influence the estimates of the N-ratio. Although there are large variations in δ15N-nitrate and isotopic/isotopomeric enrichment factors (see references listed in notes in Tables 3a and 3b), we applied unique values for the end-members, δ-N2ON-pro and δ-N2OD-rel, for estimating N-ratios using SP-δ15Nbulk analysis. The large errors in N-ratios can be attributed to the wide range of the end-member values of δ-N2OD-rel. To reduce errors in δ-N2OD-rel, it is essential to constrain the error in δ15N-nitrate and the nitrogen isotopic enrichment factor for nitrate reduction. Furthermore, even though we estimated errors in N-ratios, there remains additional uncertainty relating to the end-member of δ-N2ON-pro. It is essential to establish a methodology for estimating the end-member of δ-N2ON-pro in the oceans. Given the influence of diffusion and advection on the distribution of N2O and nitrate, it is insufficient to use δ15N-nitrate at each point as the substrate of N2O. In future studies, it is desirable to estimate the N-ratio using an advection-diffusion-reaction model with distribution of isotope ratios of substrates and isotopic/isotopomeric enrichment factors, which are functions of the concentrations of substrates and environmental parameters related to microbial activities such as temperature.

[34] We also plotted the isotopomer ratios of N2O at the concentration maximum (600 m) at Station KNOT in the western North Pacific (data for 991013 from Toyoda et al. [2002]) as the open black star on Figure 6a. Considering that δ15N-NO3 in the northern part of the North Pacific is ∼5‰ [Casciotti et al., 2002] and that the value of δ15N-NO3 is within the range of δ15N-NO3 for the SP-δ15Nbulk analysis (7.5 ± 6.5‰), the contribution of each process to N2O production at Station KNOT can be resolved using a plot of SP versus δ15Nbulk (Figure 6a). The SP-δ15Nbulk plot indicates that denitrification dominates N2O production in the OMZ at Station KNOT in the North Pacific (N-ratio = 31 ± 34%). Thus isotopomer analysis of N2O with steady state model for denitrifiers supports the denitrification hypothesis of Yamagishi et al. [2005] and Yoshida et al. [1989]. Isotopomer analysis of N2O has a potential to resolve the dispute between the nitrification and denitrification hypotheses for the N2O cycle in the oceans.

[35] The production and consumption processes of N2O indicated by the SP-δ15Nbulk analysis are consistent with the processes indicated by N* values and the concentrations of nitrite and oxygen. However, the conventional analysis of δ15Nbulk and δ18O of N2O at the GOC site leads to the opposite conclusion: the smaller δ15Nbulk and δ18O at 700 m than those at 300 m indicates that contribution of nitrification at 700 m is larger than that at 300 m (Figure 6b). This suggestion is not consistent with N* and O2 concentration (Figures 2d and 2f). Therefore the analysis of bulk nitrogen and oxygen isotope ratios may not be an effective tool for resolving production processes of N2O under suboxic conditions in the oceans. In contrast, the plot of the SP versus δ15Nbulk is very useful for resolving production and consumption processes.

5. Conclusions and Implications

[36] In the ETNP, N2O is mainly produced by nitrifiers at the concentration maximum (60–80 m deep) located at the subsurface oxycline. Isotopomer ratios of N2O produced at the maximum are 3.2 ± 0.1, 8.7 ± 0.6, −2.3 ± 0.6, 47.6 ± 0.3, and 11.0 ± 1.2‰ (1σ, n = 5) for δ15Nbulk, δ15Nα, δ15Nβ, δ18O, and 15N-site preference, respectively. At the bottom of the subsurface N2O concentration peak, N2O is produced and consumed by denitrifiers. At the suboxic zone under the OMZ at 800 m, N2O is strongly consumed by denitrifiers after release by nitrifiers and denitrifiers. In the Gulf of California, N2O is dominantly produced by nitrifiers from the subsurface to 300 m. At the N2O concentration maximum at the lower OMZ (700 m), N2O is produced and consumed by denitrifiers.

[37] Isotopomeric fractionation for N2O reduction was analyzed using a one-dimensional, vertical advection-diffusion-reaction model over the depth range 400–700 m in the ETNP OMZ, where N2O diffuses and advects from the suboxic zone under the OMZ at 800 m and where N2O is only consumed by denitrifiers. Isotopomeric enrichment factors for N2O reduction were estimated to be −11.6 ± 1.0, −19.8±2.3, −3.4 ± 0.3, and −30.5 ± 3.2‰ for δ15Nbulk, δ15Nα, δ15Nβ, and δ18O, respectively. Isotopomer analysis of N2O with isotopomeric enrichment factors is a powerful tool for estimating the ratios of N2O released from nitrifiers and denitrifiers in the oceans. Isotopomer analysis of N2O suggests that conventional bulk nitrogen and oxygen isotope ratios cannot effectively differentiate the N2O released from denitrifiers and marine nitrifiers in the low-oxygen regions although the analysis on the plot of δ15N versus δ18O gives information on the extent of its consumption. Thus isotopomer analysis is a powerful tool for resolving the N2O cycle in the oceans.

[38] Isotopomer ratios of N2O will be a sensitive signal of N2O reduction, namely N2 production, by denitrification, which is the main process of N loss from the oceans. Our plot of 15N-site preference versus δ15Nbulk indicates that water masses under suboxic conditions in the ETNP, Gulf of California, and even in the OMZ in the western North Pacific show evidence of N loss by denitrification. The suggestion of N loss from the northern Pacific Ocean is consistent with the suggestion by Li and Peng [2002], who estimated remineralization ratios in the northern Pacific from nutrient and oxygen concentrations using a three-part mixing model. They argued that the low N/P ratio for the northern Pacific basins indicates that organic nitrogen is converted partly into gaseous N2O and N2 through nitrification or denitrification processes in low-oxygen regions, or in the reducing microenvironments of organic matter within an oxygenated water column. N* has been used as a signal of N loss in the oceans. However, N* is affected by several factors such as nitrogen fixation, remineralization ratio of nutrients, and the amount of DON that has not been remineralized. Isotopomer analysis of N2O will be a useful tool for resolving the distributions of water masses that have lost nitrogen by denitrification.

Appendix A:: Model Parameters and Sensitivity Test

A1. Ratio of Advection Velocity to Eddy Diffusion Coefficient (V/K)

[39] We estimated the value of V/K using the one dimensional, advection-diffusion model of Tsunogai [1972] and Munk [1966] with the potential temperature from our cruise (170–810 m at the ETNP site). We selected this depth range because of a highly linear mixing line exists on a T-S diagram. The value of V/K was −1.4 ± 0.4 × 10−3 m−1.

A2. Advection Velocity (V)

[40] We adopt the advection velocity estimated by Chung and Craig [1973] using 226Ra. The velocities in the z direction are −2.7 m yr−1 for minimum and −4.0 m yr−1 for maximum at Station SCAN X-56 (08°07′N, 113°55′W; Bottom = 4035 m), which is close to our ETNP site (16°10′N, 106°59′W).

A3. N2O Reduction Rate (Fz)

[41] The reaction rate constant of m was estimated by fitting the concentration model (equations (1) and (2)) to the measured N2O concentration between 400 and 700 m (O2 < 1 μmol kg−1) at the ETNP site, as shown in Figure 5a. The value of m was estimated to make an asymptotic curve close to 0 nmol kg−1 at 0 m. Several results of the N2O reduction rate are shown in Figure 5c.

A4. Sensitivity

[42] Sensitivity was tested to estimate the range of enrichment factors. The fractionation factor was dependent mainly on the shape of the curve fit to the N2O concentration profile. The values of the parameters (V/K and V) were varied within their ranges to estimate errors of the fractionation factors. We used parameters of “Optimal-high,” as shown in Table A1, for estimating the optimal factors. Shifting the value of V did not change the estimated value of fractionation factors. The range of N2O reduction rates was also estimated. Estimated profiles were shown in Figure 5c, and sets of parameters are shown in Table A1.

Table A1. Parameters for the Advection-Diffusion-Reaction Model
Name103 V/K, m−1V, m yr−1
Optimal-high−1.4−4.0
Optimal-low−1.4−2.7
Maximum−1.0−4.0
Minimum−1.8−2.7

Acknowledgments

[43] We thank Nils Napp for dissolved O2 measurements, and all scientists working on the EPREX for help with sample collection in the ETNP. We are indebted to A. W. Graham for dissolved O2 measurement, sample collection, and cruise information including CTD data from the Gulf of California. We wish to thank the crew of the R/V Revelle for assistance on our cruise in the ETNP, and to thank the crew of the R/V NewHorizon for assistance on the cruise in the Gulf of California. We are indebted to H. Nara for advice on the advection-diffusion-reaction model and for his comments on this paper, and to T. Rust and F. J. Sansone for assistance in this research. We thank R. L. Sutka and N. E. Ostrom for interlaboratory calibration of N2O isotopomer ratios. We also thank R. L. Sutka, J. van-Haren, L. Farıacute;as, N. Handa, J. P. Montoya, and Y. H. Li for valuable discussions. This research was supported by U.S. National Science Foundation grants OCE 9817064 (BNP, Francis J. Sansone and Edward A. Laws) and OCE 0240787 (BNP). Reviews by N. E. Ostrom and two anonymous reviewers were very helpful in improving this paper. This is SOEST contribution 7021.

Ancillary