N2O dynamics in the western Arctic Ocean during the summer of 2017

The western Arctic Ocean (WAO) has experienced increased heat transport into the region, sea-ice reduction, and changes to the WAO nitrous oxide (N2O) cycles from greenhouse gases. We investigated WAO N2O dynamics through an intensive and precise N2O survey during the open-water season of summer 2017. The effects of physical processes (i.e., solubility and advection) were dominant in both the surface (0–50 m) and deep layers (200–2200 m) of the northern Chukchi Sea with an under-saturation of N2O. By contrast, both the surface layer (0–50 m) of the southern Chukchi Sea and the intermediate (50–200 m) layer of the northern Chukchi Sea were significantly influenced by biogeochemically derived N2O production (i.e., through nitrification), with N2O over-saturation. During summer 2017, the southern region acted as a source of atmospheric N2O (mean: + 2.3 ± 2.7 μmol N2O m−2 day−1), whereas the northern region acted as a sink (mean − 1.3 ± 1.5 μmol N2O m−2 day−1). If Arctic environmental changes continue to accelerate and consequently drive the productivity of the Arctic Ocean, the WAO may become a N2O “hot spot”, and therefore, a key region requiring continued observations to both understand N2O dynamics and possibly predict their future changes.

prior to the study presented here, no intensive investigation synthesizing information on the dynamics of WAO N 2 O (i.e., distributions of the concentration and flux and their controlling environmental factors) in the water column (from the surface to the bottom) had been conducted.
The concentration of N 2 O in the atmosphere has been steadily increasing since pre-industrial times, and because the points in time at which the N 2 O concentration of water parcels were in equilibrium with the atmosphere prior to being ventilated are all different, it is necessary to calculate equilibrium N 2 O concentration to accurately estimate the amount of N 2 O production. Although there have been a few studies attempting to accurately calculate the N 2 O concentration of the water column 20,22,25 , the N 2 O concentration of the water column in other studies has been calculated from the contemporary atmospheric N 2 O concentration 13,19,21 . Hence, in this study, the equilibrium N 2 O concentration was calculated using a tracer gas.
Based on an intensive and precise N 2 O survey of the WAO water column during the open-water season of summer 2017, we (1) present spatial distributions of N 2 O concentrations and fluxes, (2) identify physical and/ or biogeochemical factors controlling the distributions, (3) determine whether the WAO is a source or sink for atmospheric N 2 O, and (4) speculate regarding future changes in the WAO N 2 O flux in response to rapid Arctic climate changes.

Materials and methods
Sampling and measurement of basic physical/biogeochemical parameters. In August 2017, the ice breaker R/V Araon collected physical and biogeochemical samples from 30 WAO stations from the Bering Strait to the Chukchi Borderland (Fig. 1a). At each sampling location, the vertical profiles of the potential temperature (θ), salinity (S), and dissolved oxygen (DO) were measured using a conductivity-temperaturedepth instrument (CTD; SBE 911 plus, Sea-Bird Electronics, Inc., USA). The CTD temperature and conductivity accuracies were ± 0.001 °C and ± 0.0003 S m −1 , respectively (sensor specifications can be found at https:// seabi rd. com/). The CTD salinity measurements were calibrated with discrete bottle samples analyzed using a laboratory salinometer (Model 8400B, Guideline Instruments, Canada) 26 . Seawater samples were collected using 10  ]) were analyzed in the onboard laboratory using a continuous flow autoanalyzer (QuAAtro, Seal Analytical, Germany) 27 . The analytical precision of the nutrient measurements was better than 1%.
In this study, the dissolved inorganic nitrogen (DIN) was defined as the sum of [NH 4 + + NO 2 − + NO 3 − ]. Seawater samples for a chlorophyll-a analysis were filtered onto 25 mm Whatman GF/F filters, extracted in 90% acetone at 4 °C for 24 h, and quantified using a Turner Designs fluorometer (Trilogy Fluorometer, Turner Designs, USA) with an analytical precision of ± 0.05 μg L −128 . Samples for N 2 O analysis were transferred to 120-mL glass bottles. To inhibit the biological activity, 100 μL of a saturated mercury chloride (HgCl 2 ) solution was added to each sample and then sealed with rubber stoppers and aluminum caps 29 . The samples were then stored in the laboratory at ambient 'laboratory temperature' (~ 24 °C) until analysis. Wind speeds were observed using a windmill anemometer (05106, RM Young, USA) on the R/V Araon at a height of 30 m (U 30 ) above the sea surface, and were then converted from U 30 to a height of 10 m (U 10 ) using a log wind profile method (refer to Supplementary Text S1 and Table S1).
Dissolved N 2 O measurements using a cavity ring-down spectrometer. For ease and convenience of gas extraction, we used the headspace method to extract dissolved N 2 O gas from the samples (see Supplementary Text S2). Subsamples were obtained by transferring 40 mL water samples from 120-mL glass bottles into a 100-mL glass gas-tight syringe, followed by the addition of 40 mL of high-purity N 2 O-free air. The gas-tight syringe was shaken using an action shaker for 10 min to achieve equilibrium of gases between the sample and headspace phases ( Supplementary Fig. S1). This equilibrium gas was injected into a Cavity Ring-Down Spectrometer (CRDS), which is a laser-based technique that uses the optical absorbance characteristics of the gas. CRDS has recently been widely and frequently used to measure greenhouse gases in various marine environments 31−33 . Herein, we used a commercially available CRDS (Model G2308, Picarro Inc., USA) for N 2 O measurements ( Supplementary Fig. S1). As N 2 O concentration obtained by the CRDS is the concentration in the headspace, a calculation is required to determine the concentration of dissolved N 2 O in the seawater sample (Eq. 1): Figure 1. Study area map and physicochemical properties of WAO. (a) Map of sampling stations using the Ice Breaking R/V Araon during August 2017 with bathymetry information (a horizontal white-blue gradient color bar). The sampling locations were filled with chlorophyll-a concentrations (white to green colors). In this study, Sts. 1-9 are located in the SC (i.e., Bering Strait to Chukchi Shelf), and Sts. 10-30 are placed in the NC (i.e., Chukchi Borderland and Canada Basin). The FZ is between St. 9 and 10 (black dotted line). Schematic arrows represent major surface currents (blue) and gyres (yellow) identified in the study area during the summer: Siberian Coastal Current, Alaskan Coastal Current, Bering Sea Water, and Beaufort Gyre. Vertical distributions of (b) potential temperature (°C), and (c) salinity (psu) along a latitudinal transect from the Bering Strait to the Chukchi Borderland (black solid line shown in a). (d) Potential temperature-salinity diagram with N * information (blue to red gradient color bar); vertical distributions of (e) dissolved oxygen (μmol L −1 ) and (f) dissolved inorganic nitrogen (DIN; μmol L −1 ). The inset in (f) shows the DIN inventory (g N m −2 ) between the surface and 10 m at each station (red to blue gradient color bar). Note that this figure was generated using MATLAB program (ver. R2019b and www. mathw orks. com). www.nature.com/scientificreports/ where β is the Bunsen solubility (nmol L −1 atm −1 ) determined from the relationship between seawater θ and S 34 ; x is the dry gas mole fraction (ppb) measured in the headspace; P is the atmospheric pressure (atm); V w is the volume of the water sample (mL); V hs is the volume of the headspace phase (mL); R is the gas constant (0.082057 L atm K −1 mol −1 ); and T is the equilibration temperature in Kelvin (K) 35 .
To validate the CRDS-based N 2 O measurements, the measurement accuracy was examined by repeatedly measuring an N 2 O standard gas, which was certified as 334.1 ppb by the Korea Research Institute of Standards and Science, before and after the sample measurement with an interval of 20 samples. The measurements of standard gas were well reproduced within a deviation of approximately 3% ( Supplementary Fig. S2). In addition, we repeatedly measured the reference water (RW) of known concentration (N 2 O RW = 7.74 nmol L −1 ) obtained by equilibrating the ambient air (N 2 O air = 337.3 ppb) with seawater (T = 20.5 °C and S = 33.93 psu) for 24 h in the laboratory 36 . The N 2 O RW was estimated from the T and S of the equilibrated water 34 . The analytical precision was approximately 4% (Supplementary Table S2).
Because we collected single seawater samples to measure the dissolved N 2 O concentrations during the 2017 summer survey, we conducted measurements of duplicate samples collected at different times and in different environments ( Supplementary Fig. S3). The measurement discrepancy between duplicate samples was no greater than 3% (Supplementary Tables S3 and S4).  29,37,38 and is expressed as follows:

Estimations of excess N 2 O and biogeochemical tracers (AOU and N *
To accurately estimate ΔN 2 O, we need to estimate the time at which a water parcel has its last contact with the atmosphere (i.e., ventilation age) because N 2 O air varies temporally 39 . Although this concept has not been applied in many studies, a few studies have estimated ΔN 2 O by analyzing isotopes 20,25 or estimating the convection rate 22 . However, in these previous studies, there was also a limitation in that the uncertainty was large for the isotopic composition value or the convection rate in the field was not always constant. Hence, for our calculation, we used sulfur hexafluoride (SF 6 ), which displays linear growth over time in the atmosphere ( Supplementary Fig. S4). The SF 6 tracer data were collected during the summer 2015 CLIVAR ARC01 cruise 40 (Supplementary Fig. S5). The SF 6 -derived ventilation age of a water parcel collected at time t was calculated by first converting the measured SF 6 concentration (femtomoles kg −1 ) into its partial pressure (pSF 6 ), based on the θ and S of the sample. The pSF 6 value was then matched to the atmospheric growth record for SF 6 to determine the calendar year (τ) in which the SF 6 concentration in the water sample would have been in equilibrium with the atmosphere (i.e., the "ventilation date") 41 . The pSF 6 age or "ventilation age" of a water parcel is given as t − τ (Fine, 2011). The N 2 O air history affecting the WAO water column was based on the SF 6 -derived calendar age (Supplementary Table S5 and Fig. S6).
The relationship between ΔN 2 O and other biogeochemical tracers, such as the apparent oxygen utiliza- 4 3− ] measured , where R N:P is the Redfield ratio) has been widely used to estimate the biogeochemical production and/or consumption of N 2 O in various marine environments 19,37,38,42−46 . AOU is typically interpreted as the amount of DO consumed during remineralization 47,48 , a positive linear relationship between ΔN 2 O and AOU indicates that nitrification (NH 4 + → NO 2 − → NO 3 − ) is the main pathway of ΔN 2 O production 29,44 . In addition, N * has been widely used as an indicator of excess nitrogen (e.g., a nitrogen fixation) or deficit (e.g., denitrification: NO 3 − → NO 2 − → N 2 O/ N 2 ) relative to phosphorus 49 . In this study, N * was calculated as [DIN] − 16 × [PO 4 3− ], where 16 is the Redfield ratio of N to P. A negative linear relationship between ΔN 2 O and N * indicates that ΔN 2 O is mainly produced through denitrification 13,46,50 .

Results and discussion
Summer oceanographic conditions. During the summer open-water season, it is apparent that the physical and biogeochemical properties of the WAO surface waters show a latitudinal gradient from the Bering Strait to the Chukchi Borderland 12 . The southern Chukchi (SC) (i.e., Bering Strait to Chukchi Shelf), which is mainly influenced by relatively warm and nutrient-enriched Pacific waters, is a highly biologically productive marine environment displaying a generally high chlorophyll-a biomass 12,51 (Fig. 1a). By contrast, the northern Chukchi (NC) (i.e., the Chukchi Borderland and Canada Basin) is primarily driven by freshwater inputs from melting sea ice and rivers, and is characterized as cold, fresh, and oligotrophic, displaying a generally low chlorophyll-a biomass (Fig. 1a) 52−54 .
To investigate the hydrographic conditions during the summer of 2017 in the WAO, we analyzed the vertical distributions of θ, S, DO, and DIN along a latitudinal transect from the Bering Strait to the Chukchi Borderland ( Fig. 1a-c,e,f) and used a θ-S diagram to examine the composition of water masses (Fig. 1d). The distributions of θ, S, and DIN in the surface waters (< 50 m depth) suggest generally warmer, more saline, and DIN-richer waters in the SC (mean θ 0-50 m of 5.14 °C, S 0-50 m of 32.33 psu, and DIN 0-50 m of 5.31 μmol L −1 ) compared to the NC (mean θ 0-50 m of − 1.00 °C, S 0-50 m of 29.65 psu, and DIN 0-50 m of 1.20 μmol L −1 ) (Fig. 1b,c,f). In addition, the DIN 0-10 m inventory is higher in the SC (mean of 0.61 g m −2 ) than in the NC (mean of 0.01 mg m −2 ) (Fig. 1f), indicating a greater potential for driving a higher primary production in the SC, as found by Grebmeier et al. 55 . www.nature.com/scientificreports/ Meanwhile, the distribution of DO exhibited the opposite behavior (mean DO SC(0-50 m) of 333.8 μmol L −1 and DO NC(0-50 m) of 385.4 μmol L −1 ) (Fig. 1e). Based on the geographical distribution, as shown in the θ-S diagram (Fig. 1d), two different water masses are likely to be involved in the mixing process within the surface waters of the study area: warm, saline, and nutrient-enriched SC waters (i.e., Pacific Summer Water (PSW), which is also called the Bering Summer Water) 56−59 and cold, fresh, and nutrient-depleted NC waters (herein referred to as freshwater (FW)) 56,57,60,61 . Owing to the distinct physicochemical contrast between mixing PSW and FW, a frontal zone (FZ) arises between them (located between St. 9 and St. 10 (~ 73° N), Fig. 1) 62 .
Below the PWW, the maximum θ (~ 1.25 °C) and high salinity (~ 34.89 psu) water observed between 200 and 1000 m in the deep NC (Fig. 1b,c) is typically called Atlantic Water (AW) 56,58,60 . In contrast to PWW, AW is associated with a relative maximum N * (Fig. 1d). The densest waters (θ < 0 °C and S = ~ 34.95 psu), with a relatively uniform θ/S, are distributed from below ~ 1000 m to the bottom and are defined as Arctic Bottom Water (ABW) 61,63 . ABW is associated with a maximum N * signature along the transect (Fig. 1d).
In summary, during the summer of 2017, the study area consisted of five water masses: PSW, FW, PWW, AW, and ABW, recognizable in both the vertical and horizontal directions. Further details on the physicochemical characteristics of these water masses are provided in Supplementary Table S6. To discuss N 2 O dynamics, the water column was divided into three layers based on the vertical distribution of the water masses in the study area: surface (0-50 m), intermediate (50-200 m), and deep (200-2200 m) areas (Fig. 2). The deep layer is mainly composed of AW (generally distributed between 200 and 1000 m with a mean SF 6 -based ventilation age of 24.3 ± 3.9 years) and ABW (generally, below ~ 1000 m with an SF 6 -based ventilation age of 46.6 ± 14.5 years) ( Supplementary Fig. S6). Under these conditions of great age and relative stability, N 2 O concentrations should show little variation (Supplementary Fig. S7 and Supplementary Table S7). The N 2 O concentrations were constant at 13.9 ± 1.0 nmol L −1 in the AW zone, and showed a slightly decreasing trend in the ABW zone (bottom, 12.9 ± 0.8 nmol L −1 ). The saturation values in both zones are mostly less than 100% (i.e., under saturated conditions).
The distribution of SC N 2 O exhibited different patterns from the distribution of NC N 2 O within the surface layer. These results are indicators of the effect of the physical solubility, which is mainly determined by T and S 34 , and is dominant in the NC (cold and fresh) compared to the SC (warm and saline) 23 . In addition to the physical solubility, Randall et al. 64 reported that the N 2 O of sea-ice meltwater was greatly under-saturated, and several studies 13,19,21,24 have suggested that the under-saturated N 2 O in the NC surface water may be related to the dilution of melting sea ice.
The over-saturated SC, which is known to be a high biological productivity region 51 , is presumed to have biogeochemically derived N 2 O (i.e., ΔN 2 O) production that also contributes to the concentration. In addition, the ΔN 2 O production of each water parcel was precisely calculated, resulting from the SF 6 -derived ventilation date. To identify potential ΔN 2 O production sources in the SC, we evaluated both the negative linear relationship between ΔN 2 O and N * and the positive linear relationship between ΔN 2 O and DIN. The relationship between ΔN 2 O and AOU was not considered. This approach was taken because the SC is a shallow shelf region where the entire water column is kept in relatively close contact with the atmosphere. Plots of ΔN 2 O versus DIN show significantly positive correlations (R 2 = 0.50, p < 0.05) (Fig. 2), suggesting that nitrification is likely to serve as the main sources of ΔN 2 O in the SC 65,66 . Interestingly, plots of ΔN 2 O versus N * show weak negative correlations (R 2 = 0.10, p < 0.05) and scattered distributions. It has been suggested by both Hirota et al. 20 13 , the N 2 O concentration in the SC increased with depth, and both the ΔN 2 O-AOU and ΔN 2 O-N * relations were significant, suggesting that vigorous ΔN 2 O production is generated through sedimentary nitrification and denitrification. Wu et al. 21 also observed high N 2 O concentrations corresponding simultaneously to the oxygen minimum and high concentrations of NH 4 + over the Chukchi Sea continental shelf, and suggested that N 2 O production is derived from sedimentary nitrification, denitrification, and nitrification in the water column. Fenwick et al. 19 also suggested that the significant relationship between ΔN 2 O and N * represents the primary source of denitrification and that the significant scatter found in this relationship is due to the influence of other nitrogen cycling processes on ΔN 2 O production, albeit an insignificant relationship between ΔN 2 O and AOU. The results of these studies support our findings.
The low temperatures (i.e., high gas solubility) characteristic of PWW may be a potential cause of the N 2 O maximum observed within the intermediate layer. If, however, N 2 O concentrations were high solely due to the water parcel's high solute capacity, we should expect dissolved DO concentrations to be similarly elevated. However, the DO concentrations measured within the intermediate layer were low and the AOU was high. SF 6 is likewise affected by solubility as other gases. Based on the assumption that dissolved SF 6 concentrations in any given water parcel will be in equilibrium with the adjacent atmosphere prior to being ventilated, one can determine the ventilation date and the precise equilibrium N 2 O concentrations of the water parcels corresponding to that ventilation date through reference to the SF 6 concentrations. Compared to the equilibrium N 2 O levels at  O and N * were weak (R 2 = 0.13, p < 0.05) potentially as a result of interaction between the bottom water and sediments on the shelf slope.
In addition, the lateral input of shelf waters (i.e., PSW) might contribute to the N 2 O concentrations of the intermediate layer 20 . Zhang et al. 13 , Wu et al. 21 , and Toyoda et al. 25 have all suggested that the subsurface N 2 O maximum may be attributable not only to its local production within the water column (e.g., nitrification), but also to its northward transportation from the SC. Given that PSW input increases with latitude 10 , lateral transport of N 2 O may be a significant factor in determining the characteristics of surface/intermediate layers within the NC.
According to Zhan et al. 22 and Fenwick et al. 19 , in the deep layer of the Arctic Ocean, both the decreasing N 2 O and oxygen with depth and the estimated NO 3 − regeneration rate (2.3 × 10 −5 μmol L −1 a −1 ) indicate that nitrification may be insignificant for N 2 O accumulation. Denitrification may also be insignificant for N 2 O accumulation because of the relatively high oxygen concentration. It has been suggested that the N 2 O levels observed in the deep layer samples may have occurred because the water was last ventilated during pre-industrial times. This hypothesis is based on the estimated convection rate. Offering their own interpretation of the isotopic data, Toyoda et al. 25 suggested that the N 2 O concentrations observed in the deep layer were derived from a mixture of water ventilated under pre-industrial atmospheric conditions and N 2 O produced by nitrification occurring within the water column.
Here, based on the SF 6 -based ventilation age, ventilation dates of the deep water masses (i.e., AW and ABW) were determined to be from circa 1955.9 to 1995.1. The N 2 O concentrations of the deep layer were undersaturated, compared to equilibrium values in atmospheric N 2 O of the ventilation dates. These uniformly undersaturated N 2 O concentrations and the relatively homogeneous hydrographic properties suggest that deep N 2 O concentrations are mainly determined by physical mixing processes such as advection and formation, rather than the involvement of biogeochemical processes (i.e., nitrification and denitrification).  69 . It should be noted that we used the k w of Wanninkhof 70 instead of that of Wanninkhof 67 to more accurately reflect k w in Eq. (3) (refer to Supplementary Text S1). Fenwick et al. 19 and Zhan et al. 24 used the weighted mean wind data (60 days prior to sampling) to avoid an overestimation of the instantaneous wind data in the process of calculating k w . However, we used the mean wind data during sampling to provide results as observation-based snapshots. The mean differences in the estimated N 2 O flux from the three models are 0.3 μmol N 2 O m −2 day −1 in the SC and 0.2 μmol N 2 O m −2 day −1 in the NC (Supplementary Table S8). To facilitate the presentation of our results, we employed the mean value of the N 2 O flux averaged from the three models.

Estimation of N
A map illustrating the spatial distribution of the summer WAO N 2 O fluxes (Fig. 3a) indicates that the SC (Sts. [1][2][3][4][5][6][7][8][9] 19 estimated relatively lower fluxes, suggesting that these results may be due to either of the different calculation approaches (e.g., weighted mean wind data over 60 days prior to sampling), varying oceanographic conditions (e.g., dilution by melting of sea ice with low N 2 O concentration) or both. In addition, the locations are intensive near the coast. Zhan et al. 24 also used the weighted mean wind data for three different models, and the SC was identified as a source of atmospheric N 2 O, but the NC was not identified as either a source or a sink. Despite the similar surface N 2 O concentrations (~ 16.5 nmol L −1 ) with our dataset, these different results may be due to different calculation approaches (i.e., different air-sea exchange models and mean wind data). Toyoda et al. 25 (Fig. 3b). The results showed that the relationships between the N 2 O flux and the SST (R 2 = 0.48, p < 0.05), SSS (R 2 = 0.27, p < 0.05), and mean ΔN 2 O (R 2 = 0.24, p < 0.05) are significant (Fig. 3b), whereas the correlation with U 10 is not. Taken together, these results suggest that during the summer of 2017, the SC acted as a source (mean of + 2.3 ± 2.7 μmol N 2 O m −2 day −1 ) and the NC served as a sink (mean of − 1.3 ± 1.5 μmol N 2 O m −2 day −1 ) for atmospheric N 2 O. The summer WAO N 2 O fluxes were significantly influenced by physical variables associated with the solubility (i.e., SST and SSS) and biogeochemically derived N 2 O production, implying that the distribution of the WAO N 2 O flux is typically strongly susceptible to environmental changes.
A multitude of environmental changes that occur in the WAO may directly and indirectly influence the distribution of WAO N 2 O fluxes (see Fig. 3a). Among the changes observed, the increasing inflow of the Pacific waters 10,72−73 and the rapidly declining sea-ice extent 9,11,74,75 have received substantial attention to date. Based on these two phenomena, we can speculate that the distribution of WAO N 2 O fluxes revealed in this study will change in the future (Fig. 4). Lewis et al. 76 suggested that the increased phytoplankton biomass sustained by an influx of new nutrients, in addition to sea-ice reduction, has driven the Arctic Ocean (e.g., Chukchi Sea) to be increasingly more productive. The increased biomass would lead to intense nitrification and potentially benthic denitrification, resulting in increased N 2 O production within the water column. The increasing inflow of warm and nutrient-enriched Pacific waters into the WAO would likely extend the productive SC region northward, leading to an enlarged WAO role as an N 2 O source (positive, sea → air), whereas a rapid loss of the sea ice extent may ultimately lead to a sea-ice-free NC with a northward shift, resulting in a diminished role as an N 2 O sink (negative, air → sea). Should this potential scenario come to pass, we would expect the WAO to become an oceanic N 2 O "hot spot" source region, and we therefore suggest that this positive feedback scenario should be considered in an effort to improve the future trajectory of WAO changes.

Summary and conclusions
We investigated the distributions of the N 2 O concentration and flux, their controlling factors, and the role of the WAO as a source or sink for atmospheric N 2 O during the summer of 2017. In the surface layer (0-50 m, consisting of PSW + FW), the mean N 2 O concentration of the SC and NC is estimated to be 14.7 ± 2.1 and 15.9 ± 0.8 nmol L −1 , respectively. However, the mean N 2 O saturation was higher in the SC (113% ± 10%, over-saturation) than in the NC (95% ± 5%, under-saturation). This result indicates that the effect of the physical solubility is dominant in the NC (cold and fresh) compared to the SC (warm and saline), and that the over-saturated SC is likely to gain additional biogeochemically derived N 2 O (i.e., ΔN 2 O) production through nitrification. The intermediate layer (50- As our study was based on a single investigation, it is impossible for us to represent the entire 2017 calendar year, or even the entire summer of 2017. We are not, however, alone in suffering from this 'limited data' impediment. The Arctic Ocean is an extreme environment, acquiring year-round data is very difficult and extremely www.nature.com/scientificreports/ costly. Consequently, our results are a mere snapshot of what is undoubtedly a much bigger picture. We intend to propose a direction for future work based on our experience of undertaking this study. If Arctic changes are accelerated and consequently drive the Arctic Ocean in a more productive manner, the WAO may become an oceanic N 2 O "hot spot" source region. Given that these processes are relevant to global climate change, additional observations of the time series and more open-water seasons are required to support this scenario. Therefore, attention should be paid to future N 2 O dynamics in the WAO.

Data availability
Hydrographic data are available in Korea Arctic Ocean-data System (http:// kaos. kopri. re. kr/ cmm/ main/ mainP age. do). Atmospheric N 2 O data are available in ESRL (https:// www. esrl. noaa. gov). The SF 6 data collected from the 2015 CLIVAR ARC01 cruise are available in CCHDO (https:// cchdo. ucsd. edu/). The N 2 O flux and wind data are presented in the Supporting Information.