Identifying the origin of nitrous oxide dissolved in deep ocean by concentration and isotopocule analyses

Nitrous oxide (N2O) contributes to global warming and stratospheric ozone depletion. Although its major sources are regarded as bacterial or archaeal nitrification and denitrification in soil and water, the origins of ubiquitous marine N2O maximum at depths of 100–800 m and N2O dissolved in deeper seawater have not been identified. We examined N2O production processes in the middle and deep sea by analyzing vertical profiles of N2O concentration and isotopocule ratios, abundance ratios of molecules substituted with rare stable isotopes 15N or 18O to common molecules 14N14N16O, in the Atlantic, Pacific, Indian, and Southern oceans. Isotopocule ratios suggest that the N2O concentration maxima is generated by in situ microbial processes rather than lateral advection or diffusion from biologically active sea areas such as the eastern tropical North Pacific. Major production process is nitrification by ammonia-oxidizing archaea (AOA) in the North Pacific although other processes such as bacterial nitrification/denitrification and nitrifier-denitrification also significantly contribute in the equatorial Pacific, eastern South Pacific, Southern Ocean/southeastern Indian Ocean, and tropical South Atlantic. Concentrations of N2O below 2000 m show significant correlation with the water mass age, which supports an earlier report suggesting production of N2O during deep water circulation. Furthermore, the isotopocule ratios suggest that AOA produce N2O in deep waters. These facts indicate that AOA have a more important role in marine N2O production than bacteria and that change in global deep water circulation could affect concentration and isotopocule ratios of atmospheric N2O in a millennium time scale.

Isotopocule ratios (δ 15 N, δ 18 O, and SP, 15 N-site preference in NNO molecule) of N 2 O are useful parameters to identify the origin and production-consumption processes of N 2 O because they depend on the isotopic ratios in precursors and isotope effects of chemical or biochemical reactions 6 . Reportedly, the magnitudes of isotope effects on nitrogen during N 2 O production as byproduct of nitrification by ammonia oxidizing bacteria (AOB) 7,8 and ammonia oxidizing archaea (AOA) 9,10 differ. Moreover, SP values of N 2 O produced by AOB via hydroxylamine (NH 2 OH) 9,11 and by AOA 9,10,12 are distinct from the values of N 2 O produced by AOB via nitrite (NO 2 − ) 8,11,13,14 or by denitrification 11,15 (Supplementary Fig. S1, Table S2). When N 2 O is partially reduced to N 2 in denitrification, all isotopocule ratios increase [16][17][18][19][20] . Although the isotope effect during N 2 O reduction varies among bacterial species or pure culture and community incubations, consistent relations have been found between isotope effects on N, O, and SP 16 .
Several earlier studies have analyzed vertical profiles of concentration and isotopocule ratios of N 2 O in the ocean and have investigated production-consumption processes specific to the depths or sites 17,[21][22][23][24][25] . Nevertheless, it remains unclear whether N 2 O is produced in situ or transported from other regions because existing data do not cover the wide range of oceanic setting such as circulation age of deep seawater. This report describes new isotopocule analyses of N 2 O in the northern North Pacific (NNP), equatorial Pacific (EQP), and Southern Ocean and southeastern Indian Ocean (SO/SIO) where younger, medium, and older water respectively exist in deep layers (Fig. 1). The respective results are compared with existing observations to ascertain the origin of N 2 O from the perspective of global deep water circulation.

Vertical Profiles of Dissolved N 2 o
In the surface layer, concentrations of N 2 O measured in the eight regions except the eastern tropical North Pacific (ETNP) are as low as those expected under the dissolution equilibrium between atmosphere and seawater (6-8 nmol kg −1 ). However, they increase with depth and reach a maximum (25-65 nmol kg −1 ) (Fig. 2a). The depth of the maximum is 100-400 m in the tropical South Atlantic (TSA), SO, and EQP, and 400-1500 m in other regions. Those values correspond to almost identical seawater density (potential density anomaly, σ θ , of about 27) (Supplementary Table S1). Below the maximum, the N 2 O concentration decreases with depth. It reaches 15-40 nmol kg −1 in the deep layer (>2000 m depth or σ θ = 27.7-27.8). In the ETNP, two concentration maxima exist at 60 m (σ θ = 25.0) and 800 m (σ θ = 27.3) 17 .
Each of the three independent isotopocule ratios shows a unique profile that is different from that of concentration ( Fig. 2b-d). The bulk (or average) nitrogen isotope ratio (δ 15 N bulk ) in the eight regions except ETNP shows no vertical gradient from the surface to the depth corresponding to σ θ ≅ 25. Then it decreases slightly with depth, showing a minimum value at σ θ ≅ 26. Below the small minimum, it increases monotonically with depth and reaches its maximum (9-10‰) at the bottom layer. In the ETNP, it shows a minimum and a maximum respectively corresponding to the shallow and deep concentration maxima. The oxygen isotope ratio (δ 18 O) and 15 N-site preference (SP) also show almost constant values from the surface to the depth above the concentration maximum (σ θ ≅ 26) in the eight regions. In deeper water, however, their vertical profiles are markedly different among oceanic regions. Although δ 18 O and SP respectively show a parallel increase and decrease with concentration in the North Pacific stations, they exhibit a monotonic increase in EQP, SO/SIO, and ESP. In TSA, they show minima at the concentration maximum.  www.nature.com/scientificreports www.nature.com/scientificreports/

origin of N 2 O at Concentration Maximum
Isotopic signatures of N 2 O at density level of σ θ = 27.3 which corresponds to concentration maximum in ETNP show systematic difference between oceanic regions (Fig. 3). For example, in spite of similarity in water mass property (temperature and salinity, data not shown) between NNP and WNP and between EQP and ETNP, they are distinguished each other. Variations of isotopocule ratios are caused by (1) mixing of N 2 O with different isotopic signatures such as N 2 O produced in situ, advected from other regions, and dissolved atmospheric N 2 O or (2) decomposition of N 2 O during which the remaining N 2 O is isotopically fractionated 6 . Based on a global  Table S1 for data sources.
www.nature.com/scientificreports www.nature.com/scientificreports/ distribution of N 2 O concentration, it has been proposed that N 2 O produced in the eastern tropical Pacific is exported to other regions in the Pacific by lateral or isopycnal advection 4 . If this is the case, and if we consider mixing of two endmembers, namely, N 2 O produced in the ETNP and background N 2 O from the atmosphere, the isotopic data points for the regions outside the ETNP is expected to fall on the mixing line in isotope-reciprocal concentration plot. However, the results show that mixing between N 2 O in the ETNP and N 2 O in the water equilibrated with the atmosphere cannot explain the observed isotopocule ratios in the eight regions (Fig. 3). In  Table S1).
In Fig. 3, N 2 O maxima in NNP could be regarded as an alternative endmember because of its high concentration. However, lateral advection of N 2 O from NNP is also improbable for the following reasons. First, the distribution of data obtained in other regions still cannot be explained consistently by the isopycnal mixing in each panel of Fig. 3. Second, vertical profiles of temperature and salinity at the stations in this work do not show discontinuous feature which is expected if different water mass is advected laterally.
Contribution from partial reduction of N 2 O (the second factor noted above) is also unlikely because all the isotopic signatures should be increased with decreasing concentration and because nitrate reduction, which is the first step of denitrification and prerequisite for N 2 O reduction, is not prominent in the observed oxic water columns. Therefore, we infer that the N 2 O maximum is a result of in situ production rather than advection from other regions. Possible production mechanisms are discussed below together with those for deep water N 2 O.

origin of N 2 O in Deep Layer
The concentration of N 2 O averaged for water below 2000 m depth shows a positive correlation with the circulation age of seawater estimated from the 14 C content 26 (Fig. 5a). Bange and Andreae found a similar relation by compiling 56 observations from the North/South Atlantic, North/South Indian, and North Pacific oceans 5 . Based on insignificant fluxes estimated for hydrothermal or sediment N 2 O, they concluded that N 2 O is produced in the deep ocean mainly by nitrification. The slope of the regression line in Fig. 5a ((1.0 ± 0.2) × 10 −2 nmol kg −1 yr −1 ) is comparable to their reported value of (5.7 ± 1.0) × 10 −3 nmol L −1 yr −1 and the y-intercept (14.3 ± 2.2 nmol kg −1 ) agrees with the earlier report (13.1 nmol kg −1 ) 5 . Here we show that isotopocule ratios also increase with the circulation age. When plotted against inverse N 2 O concentration, they are distributed on a line that passes ranges of surface water that is saturated with the atmosphere (Fig. 5b). This confirms that N 2 O is added during the circulation of deep seawater after losing contact with the atmosphere, and the following isotopocule ratios of the produced N 2 O are obtained as y-intercepts of the regression lines: 10.2 ± 0.4‰, 62.2 ± 1.6‰, and 28.1 ± 1.6‰, respectively, for δ 15 N bulk , δ 18 O, and SP.
A closer look at Fig. 5a and Supplementary Table S1 reveals systematic difference in N 2 O concentration between NNP, STNP, WNP, and ESNP. Since it has been suggested that deep water circulates from the South Pacific to these north Pacific regions via different routes 27,28 , N 2 O production rate during the circulation might vary with the pathways. For instance, dissolved oxygen concentration in the deep layer is lowest in NNP 29 , which could explain the higher N 2 O concentration in NNP than other North Pacific regions because N 2 O production is enhanced under low oxygen condition 12,30 . Additional factor that could partly explain the positive deviations of NNP data from the fitted line is underestimation of age for the NNP waters (Matsumoto, personal communication).  Fig. 5b respectively show the values for the surface water in the subduction region (off Greenland and off Antarctica) which is equilibrated with modern and preindustrial atmosphere 31 . The negative deviation of the SO/SIO data might indicate that relatively younger water mass in this region is affected by the mixing of modern surface water during its advection.

Production Mechanisms of N 2 O in Concentration Maximum and Deep Layer
We compare the isotopocule ratio values of excess N 2 O at concentration maximum and accumulated N 2 O during deep water circulation with those values reported or estimated for currently known biological processes of oceanic N 2 O (Fig. 4, Supplementary Table S4). The ranges of δ 15 N bulk and SP values for the N 2 O maximum at most of the stations in the North Pacific and the values for deep N 2 O overlap with that of N 2 O produced by archaeal nitrification conducted by ammonia-oxidizing archaea (AOA) 9,10,12 (Fig. 4a). At several stations in the North Pacific and SO/SIO, SP values of N 2 O at maximum concentration are larger than the range of AOA-derived N 2 O, suggesting additional contribution from bacterial nitrification. In the Atlantic and at several stations in the equatorial/South Pacific and SO/SIO, N 2 O maxima are accompanied by lower SP values, which indicates significant role of nitrifier-denitrification/denitrification. In contrast, δ 18 O values obtained for both N 2 O maximum and deep water N 2 O are higher than that of AOA by 5-20‰ (Fig. 4b). If N 2 O produced by AOA or bacterial processes is further reduced by denitrification, δ 18 O of residual N 2 O increases due to isotope fractionation [16][17][18][19][20] . However, the parallel increases in δ 15 N bulk and SP expected for N 2 O reduction (slope of broken lines in Fig. 4) are not prominent especially for δ 15 N bulk and it is unlikely that N 2 O reduction occurs in oxic seawater because N 2 O reduction has been found only in anoxic environment such as ETNP or Arabian Sea 17,32 . Therefore, the discrepancy of δ 18 O might be explained by variation of oxygen isotope effects during N 2 O production by AOA. This should be tested using further laboratory studies with several species of AOA. Reported oxygen isotope fractionation during the incorporation of O 2 into N 2 O is −2.2 to 13.2‰ in two marine and five soil archaeal species/isolates 9,10 . Those figures might become larger if the reaction proceeds with the equilibrium step because O-isotope exchange equilibrium fractionation between N 2 O and O 2 is calculated theoretically as about 20‰ at 0-10 °C 33 .
Several reports have described the dominance of AOA in nitrification in the ocean. Archaeal amoA gene abundance was positively correlated with potential NH 3 oxidation rate and N 2 O concentration in the upper oxycline of the eastern tropical South Pacific 34 . In the oxygen minimum zone (OMZ) of the eastern tropical North Atlantic, comparable patterns of abundance and expression of archaeal amoA genes and N 2 O co-occurred 12 . Our results suggest that AOA play a major role not only in the OMZ of tropical ocean but also in the OMZ in temperate and subarctic ocean areas and deeper waters. The average rate of N 2 O production during deep water circulation estimated from the relation between N 2 O concentration and circulation age of seawater is 28 f mol L −1 d −1 , Our results suggest that ubiquitous N 2 O maximum in the middle layer in the North Pacific is formed by in situ production (mainly by AOA) rather than advection from the eastern tropical Pacific and that N 2 O is also produced by AOA and accumulated during the global deep water circulation. If one simply assumes that the N 2 O dissolved in the oldest deep water in NNP (34 nmol kg −1 ) is ultimately released to the atmosphere by the circulation driven by seawater subduction in the polar regions (3 × 10 7 m 3 s −1 ) 35 , magnitude of this deep N 2 O source is estimated at up to 1 Tg N yr −1 compared to the estimated surface oceanic source of 3.8 Tg N yr −1 1 which seems to be based on limited surface observations. This implies that change in global deep water circulation could affect concentration and isotopocule ratios of atmospheric N 2 O in a millennium time scale. . At each station, samples were taken at 15-28 depths in the range of 0-6000 m using 12 L Niskin bottles mounted on a conductivity-temperature-depth Rosette sampler. Each sample was collected in a 230 mL glass vial followed by addition by 1 mL saturated HgCl 2 solution for sterilization and by sealing with a butyl rubber stopper. Each was then preserved at 4 °C in the dark.
Isotopocule ratios of excess N 2 O at its maximum and those of N 2 O produced in situ during deep water circulation were estimated by assuming the mixing of two end members. Equations (4) and (5) respectively describe the mass balance of light and heavy N 2 O molecules; C obs , C pro , C atm respectively denote the observed, produced, and atmospheric equilibrium concentrations; δ obs , δ pro , and δ atm are the respective isotopocule ratios. By eliminating C pro from Eqs (4) and (5), we obtain the following: pro o bs obs a tm atm o bs atm The circulation age of deep seawater at each observational station was calculated from the objectively mapped circulation 14 C age below 1500 m in the literature 26 using "2D estimation" tool of Ocean Data View software 38 .