This investigation was conducted in the STWNPO, between 15 and 28°N and 135°E (Figure 1A) onboard the R/V Isabu, from May 27^th to June 4^th, 2020. Seawater was collected at eight stations (Figure 1A), and at each station, 24 vertical samples were collected from the surface down to a depth of 5774 m using Niskin bottles on a rosette sampler. At all stations, temperature, salinity, and dissolved oxygen (DO) profiles were determined using an SBE 911 Plus (CTD; Sea-Bird, USA) sensor system attached to a rosette sampler. The accuracies of temperature, conductivity, and oxygen sensors were ± 0.001°C, ± 0.0003 S m^-1, and ± 2% of saturation, respectively. The wind speed above the sea surface (29 m) was also measured using an ultrasonic wind sensor (Ventus, Lufft, Germany) and later converted to a 10 m height following the log wind profile method (Holmes, 2007). The accuracy of the wind sensor is ± 0.2 m s^-1.

To determine nutrient concentrations, seawater samples were collected in 15 mL acid-rinsed bottles and stored at −20°C until further analysis. Nitrate ( NO 3 − ) and phosphate (PO₄^3-) were analyzed in the laboratory using a QuAAtro autoanalyzer (Seal Analytical, Germany). The analytical precision of the nutrients was greater than 1%. To determine chlorophyll-a (Chl-a), ~4 L of seawater was filtered through 47 mm Whatman GF/F filters and frozen at −80°C. In the laboratory, pigments were extracted using 90% acetone (Ritchie, 2006), and the Chl-a concentrations were measured using a Trilogy Fluorometer (Model # 7200-002, Turner designs, USA) with an analytical precision of ± 0.05 µg L^-1. Seawater samples for pH estimation were collected according to the protocol of Dickson et al. (2007) and were later spectrophotometrically assessed using an unpurified meta-cresol purple indicator (Douglas and Byrne, 2017a; Douglas and Byrne, 2017b) with an accuracy of ± 0.004 pH units.

To estimate the historical values of atmospheric N₂O concentrations, a dataset of atmospheric N₂O concentrations was aggregated, spanning AD 800 to AD 2020. Data between 800 and 1977, when atmospheric N₂O had not been monitored, were obtained from studies of Antarctic ice cores (Battle et al., 1996; Schilt et al., 2010). Data from 1979 to 1986 were obtained from the Air Monitoring Program of the Commonwealth Scientific and Industrial Research Organisation (CSIRO) in Cape Grim, Australia. Although there is a limit to the available information representing the spatial variability of atmospheric N₂O concentrations for the entire planet, obtaining this information is simultaneously beneficial and the sole data source.

The longitudinal difference in atmospheric N₂O concentrations in the North Pacific Ocean is insignificant because of the zonally well-mixed atmosphere and lack of strong sources or sinks (Ishijima et al., 2009). However, as N₂O emissions (e.g., from agriculture, industry, and natural soil) are more significant in the Northern Hemisphere than in the Southern Hemisphere, atmospheric N₂O is also higher in the Northern Hemisphere than in the Southern Hemisphere (Supplementary Figure S1A). From 1987 to 2020, the mean difference in atmospheric N₂O concentrations between hemispheres was 1.13 ± 0.59 ppb. Hence, to reflect atmospheric N₂O over the North Pacific Ocean as accurately as possible, the data between 1987 and 2020 were taken from the air monitoring program run by the NOAA Earth System Research Laboratories (ESRL) in Mauna Loa, Hawaii.

Subsequently, spline interpolation was applied according to time to the total aggregated N₂O data because it comprised discrete values (Supplementary Figure S1B). However, our approach does not accurately reflect the spatial variability of the dataset before 1987, when atmospheric N₂O data for the North Pacific Ocean was limited, and may underestimate the equilibrium N₂O of seawater, expressed as N₂O_eq = N₂O_air ∙ β ∙ P, where N₂O_air is the atmospheric N₂O level, β is the Bunsen solubility (nmol L⁻¹ atm⁻¹) determined from the relationship between the potential temperature and salinity of seawater (Weiss and Price, 1980), and P is the atmospheric pressure (atm). However, for the pre-1980s period, estimating the exact impact of utilizing data from the Southern Hemisphere (i.e., Antarctica and Cape Grim, Australia) is still limited by data availability.

The seawater samples used to quantify the dissolved N₂O concentrations were carefully transferred from the Niskin sampler to 120 mL glass bottles, and 100 μL of saturated HgCl₂ was added to inhibit biological activity. The sample bottles were tightly sealed with rubber stoppers and aluminum caps to avoid interaction with ambient air (Wilson et al., 2018). We used the headspace method with a cavity ring-down spectrometer (CRDS; Model G2308, USA) to estimate the dissolved N₂O gas in the water samples. This laser-based method uses gas optical absorbance characteristics with a precision of < 7 ppb (Heo et al., 2021). Subsamples were transferred from 120 mL glass bottles into a 100 mL glass gas-tight syringe, and 40 mL of high-purity N₂O-free air was then added to the syringe. The sample and N₂O-free air in the syringe were gas equilibrated using an action shaker (ASA-026-12, ASIA TESTING MACHINE, South Korea). This equilibrated gas (i.e., gas in the headspace of the syringe) was then injected into the CRDS to measure its N₂O concentrations. The following equation was used to calculate and convert the N₂O concentrations in the equilibrated gas to dissolved N₂O concentrations in the seawater samples (Eq. 1): (1) N 2 O c o n c . = ( β · x · P · V w + x · P R · T · V h s ) / V w where N₂O_conc. is the dissolved N₂O concentration in the seawater sample (nmol L^-1), β is the Bunsen solubility (nmol L⁻¹ atm⁻¹) determined from the relationship between the potential temperature and salinity of seawater (Weiss and Price, 1980), х is the dry N₂O mole fraction (ppb) measured in the headspace, P is the atmospheric pressure (atm), V_w is the volume of the seawater 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) (Wilson et al., 2018).

We determined the measurement accuracy of the N₂O concentration in seawater by measuring the concentration of N₂O standard gas once every 15 seawater samples. The standard was certified as 334.1 ppb (± 1%) by the Korean Research Institute of Standards and Science. The ensemble standard deviation of all gas standard measurements was 2.8% (mean: 328.8 ± 2.3 ppb) (Supplementary Figure S2). As the gas standard measurements overlapped with the precision of the certified standard values and the instrument, the seawater samples were measured without any adjustments. In addition, we measured duplicate seawater samples, with a measurement discrepancy of approximately 4%.

Excess N₂O (ΔN₂O), which is the amount of biogeochemically produced N₂O, was estimated as the difference between the equilibrium N₂O (N₂O_eq, refer to the section 2.2) and the measured N₂O concentration (N₂O_measured) (Yoshinari, 1976; Walter et al., 2006; Kock et al., 2016; Heo et al., 2021) and is expressed as follows:

To accurately estimate ΔN₂O, we first assessed the time when a water parcel was last in contact with the atmosphere (i.e., ventilation age) using the transit time distribution (TTD) method. Historical N₂O_air values varied over time (Supplementary Figure S1B), and given that the Pacific Ocean is the oldest in the world, the age of each water parcel should be considered when selecting a value of N₂O_air for calculations of N₂O_eq away from the surface. The TTD method was initially developed by Waugh et al. (2003) and has been widely applied to various marine environments (e.g., Tanhua et al. (2008); Kim et al. (2013); Sonnerup et al. (2019); Ko and Quay (2020)). Below, we briefly describe how this method was applied in the present study using CFC measurements. In summary, we calculated water mass ages approximately ranging from 0 to 1200 yr. for 0−5774 m (Supplementary Figure S3).

The transit time distribution (TTD) method assumes that inert tracers present at the surface are transported into the ocean interior, and their concentrations are affected by physical processes (e.g., advection and diffusion) (Waugh et al., 2003; Tanhua et al., 2008; Kim et al., 2013). Using this method, ventilation times can be obtained (Waugh et al., 2003). The following equations describe this method: (3) C ( x , t ) = ∫ 0 ∞ C 0 ( t − t ′ ) · G ( x , t ′ ) d t ′ where C(x,t) is the concentration of the inert tracer (i.e., CFCs; see measurement details below) at the interior location (x) and contemporary time (t), C₀ (t–t’) is the surface concentration considering the time taken to enter the ocean interior from the surface (t–t’), and t’ is the transit time from the surface to the interior location. G(x,t’) is the distribution of transit times (i.e., TTD) for a water parcel from its surface of origin to its inner location, and G(x,t’)dt’ represents the mass fraction of the water parcel that last made contact with the surface of origin from time t’ to time (t’ + dt’). For a finite domain with nonzero diffusion, the TTD is always positive and ∫ 0 ∞ G d t ′ equals 1, assuming that all water parcels had surface contact at some time in the past. To characterize the TTD, it can be written in the form of a simple function, the free parameters of which can be estimated from tracer observations using the inverse Gaussian function (Hall et al., 2002; Waugh et al., 2006) as follows: (4) G ( t , Γ , Δ ) = Γ 3 4 π Δ 2 t 3 exp ( − Γ ( t − Γ ) 2 4 Δ 2 t ) where t is t’/Γ as a dimensionless time, Γ is the mean transit time, and Δ is the width (spread) of the transit time. Γ and Δ can be associated with a one-dimensional model using Peclet number corresponding to Pe = (Γ/Δ)² = vL/K, where v is the advective flow, L is the transport length, and K is the diffusivity (mixing) (Waugh and Hall, 2002; Sonnerup et al., 2015). As the inverse Gaussian function can be fully described by two parameters, Γ and Δ, the Δ/Γ ratio plays an important role in determining TTD age. A higher Δ/Γ ratio indicates that physical mixing is vigorous and that the water parcels are older. In contrast, a lower Δ/Γ ratio (i.e., closer to zero) indicates that advection is predominant and that the water parcels are younger. To minimize errors when using the TTD method, selecting a suitable Δ/Γ ratio for the study area is essential.

We used CFC11 and CFC12, collected during the July–August 2016 CLIVAR P09 cruise (Supplementary Figure S3), as inert tracer gases using the TTD method. To identify a suitable Δ/Γ ratio for STWNPO, we compiled the Δ/Γ ratios that have been applied to various regions (Supplementary Table S1). He et al. (2018) used a Δ/Γ ratio of 1.2 in the North Pacific Ocean, and Sonnerup et al. (2013) suggested that possible Δ/Γ ratios for directly ventilated waters in the North Pacific do not exceed 1.0 but are greater than 1.0, for denser waters that are ventilated by physical mixing. Waugh et al. (2006) assumed a Δ/Γ ratio of 1.0, which could accurately describe global tracer distributions, while Wang et al. (2021) used a mean Δ/Γ ratio of 0.8 for the western North Pacific and the northern South China Sea. Based on the compilation of these values, we set the Δ/Γ ratio range to 0.8–1.2, in increments of 0.2.

Ideally, the TTD ages from different tracers should be equal, so we compared the correlations between TTD ages from CFC11 (TTD_CFC11 age) and CFC12 (TTD_CFC12 age) along with Δ/Γ ratios of 0.8, 1.0, and 1.2 (Supplementary Figure S4). For all Δ/Γ ratios, the correlation between TTD_CFC11 age and TTD_CFC12 age was high. We also compared the excess N₂O (ΔN₂O) in the water column, as estimated from the TTD_CFC12 ages for each Δ/Γ ratio (i.e., ΔN₂O_0.8, ΔN₂O_1.0, and ΔN₂O_1.2) (Supplementary Table S2). The mean of the differences between ΔN₂O_0.8 and ΔN₂O_1.0 and between ΔN₂O_1.0 and ΔN₂O_1.2 was 0.07 ± 0.07 nmol L⁻¹ and 0.09 ± 0.11 nmol L⁻¹, respectively. The ΔN₂O values of the available Δ/Γ ratios were not significantly different. Therefore, we used the mean Δ/Γ ratio of 1.0. N₂O saturation (= N₂O_measured/N₂O_eq × 100) was calculated based on N₂O_eq using the water mass age corresponding to Δ/Γ ratio of 1.0.

In addition, ΔN₂O_1.0 and unmodified ΔN₂O (i.e., estimated ΔN₂O using a unitary contemporary atmospheric N₂O concentration) were compared to obtain the error when estimating ΔN₂O without considering the water mass age. The mean difference was 1.28 ± 1.12 nmol L^-1, which was 11.3 ± 10.9% compared to the ΔN₂O_1.0 (Supplementary Table S2). This suggests that significant distortion may occur when estimating excess N₂O if the same contemporary atmospheric N₂O concentration is used for all water depths without considering the water mass age.

The vertical profiles of the hydrographic parameters in the STWNPO are shown in Figures 1B–F, and the distributions of the potential temperature (θ) and salinity (S) of the water masses are shown in Figure 2. The distributions of θ and S are labeled as N * ( = [ NO 3 − ]   measured   −   R N:P × [ PO 4 3 − ] measured ,  where  R N:P  is the Redfield ratio ) . N^* has been widely used as an indicator of nitrogen surplus (e.g., nitrogen fixation) or sinks (e.g., denitrification: NO 3 − → NO 2 − → N₂O/N₂), relative to phosphorus (Gruber and Sarmiento, 1997). This study utilized an N:P ratio of 16 ± 1, confirmed by Takahashi et al. (1985) and Anderson and Sarmiento (1994). Given the deviation of this stoichiometric value, the error of N^* estimations was 1.65 ± 1.19 μmol L^-1.

These results demonstrated that the distributions of θ, S, NO 3 − , and DO between 0 and 200 m depths were the warmest, highly saline, and N-depleted (i.e., oligotrophic) over the study period. In particular, at shallow depths (0–200 m), St. 1 to St. 4, located in the southern region, had a higher mean θ (26.34°C), maximum S (35.05 psu), and lower mean DO (195.11 µmol L⁻¹), compared to St. 5 to St. 8, which were located in the northern region and had a mean θ of 21.97 °C, maximum S of 34.96 psu, and mean DO of 205.71 µmol L⁻¹. These hydrographic variations are likely the reason for the different water masses being part of the surface waters of the STWNPO, that is, the North Pacific Subtropical Underwater (NPSTUW) (Suga et al., 2000; O’Connor et al., 2002; Behrens et al., 2018) in the southern region and the North Pacific Subtropical Mode Water (NPSTMW) (Hanawa and Talley, 2001; Oka, 2009; Rainville et al., 2014) in the northern region. The NPSTMW had higher average NO 3 − (1.56 µmol L⁻¹) and Chl-a (0.13 µg L⁻¹) concentrations in the northern region, relative to the southern region (1.10 µmol L⁻¹ of NO 3 − and 0.11 µg L⁻¹ of Chl-a). However, in terms of chlorophyll as an indicator of primary productivity, both regions exhibited overall low productivity in the STWNPO during this study (Figures 1E, F).

In contrast to the upper layer, the intermediate profiles (200–1500 m) in this study showed sharp fluctuations in physicochemical parameters. A pycnocline appeared just below the surface, which is a typical characteristic of North Pacific Central Water (NPCW) (Emery, 2001; Behrens et al., 2018), with an extensive range of θ (10–22°C) and S (34.2–35.2 psu) (Figures 1B, C, 2). In addition, just below the NPCW, the North Pacific Intermediate Water (NPIW) was observed with the lowest S in this study (~34.15 psu) (Figure 1C) (Talley, 1993; Behrens et al., 2018). The NPIW is ventilated by the Okhotsk Sea or an adjacent subpolar gyre with a low S and high DO. However, NPIW which had lower DO (107.56 ± 32.96 µmol L⁻¹) than the overlying NPCW (186.12 ± 8.11 µmol L⁻¹) (Figure 1D) was due to its long residence time (Talley, 2011). In this study, under the NPIW, the Pacific Deep Water (PDW) was observed to have minimum DO concentrations (Figure 1D) over an extensive range of depths (800–3000 m), and these core conditions were considered hypoxic (<63 µmol L⁻¹) (Amakawa et al., 2009; Talley, 2011; Behrens et al., 2018). PDW was also identified as having the maximum nutrient concentration, which was inversely proportional to the DO concentration (Figures 1D, E) (Talley, 2011), along with its minimum N^* signature (Figure 2). As the PDW is formed by upwelling, diffusion, and continued mixing with adjacent waters, its attributes become less characteristic toward the southern region (Figures 1D, E) (Talley, 2011).

The water layer beneath the PDW, the Lower Circumpolar Deep Water (LCDW) (Emery, 2001; Kawano et al., 2006; Talley, 2011), was observed to have minimal θ (~1.2 °C) and high S (~34.68 psu). The LCDW was composed of the densest and most homogeneous waters and was also identified by higher DO and N^* and lower nutrient content than the PDW (Figures 1D, E, 2).

Considered together, our results demonstrate that during the investigation (spring, 2020), the STWNPO comprised six different water masses (NPSTUW, NPSTMW, NPCW, NPIW, PDW, and LCDW) and exhibited strong hydrographic variations (see Supplementary Table S3). To better understand N₂O dynamics in this study, we divided the water column into three layers according to the vertical distribution of the water masses: surface (0–200 m), intermediate (200–1500 m), and deep (>1500 m) layers.

During this investigation, N₂O concentrations within the surface layer showed an increasing trend with depth (Figure 3A). The variations in N₂O in the southern region where the NPSTUW exerted influence and in the northern region where the NPSTMW exerted influence ranged from 6.4 nmol L⁻¹ to 9.6 nmol L⁻¹ and from 6.7 nmol L⁻¹ to 9.5 nmol L⁻¹, respectively. Although the N₂O concentration was slightly lower in the southern region (mean N₂O_{0–200m southern}: 7.4 ± 0.9 nmol L⁻¹) than in the northern region (mean N₂O_{0–200m northern}: 7.8 ± 0.7 nmol L⁻¹), the average N₂O saturation was higher in the southern region (120.4 ± 6.4%) than in the northern region (113.9 ± 4.4%) (Supplementary Figure S5). The dependence of solubility on temperature can explain these variations in N₂O concentrations.

Prior to 2010, the N₂O surface concentrations observed in the vicinity of the Western North Pacific tended to be lower (~5.6 nmol L⁻¹), but remained slightly supersaturated compared to equilibrium values (Butler et al., 1989; Yoshida et al., 1989; Toyoda et al., 2002). However, recent observations have indicated that surface N₂O concentrations increase with increasing latitude—for example, N₂O concentrations in tropical regions were ~6.0 nmol L⁻¹ and those at 30°N were under 8.0 nmol L⁻¹ (Zhan et al., 2017). In addition, other reports from a nearby station (30°N, 145°E) reported that values increase from the surface (~6.6 nmol kg⁻¹) toward the euphotic zone limit (~10 nmol kg⁻¹) (Breider et al., 2015; Yoshikawa et al., 2016).

In the intermediate layer, the N₂O concentrations rapidly increased up to the oxygen minimum layer (Figure 3A), and the NPCW exhibited the lowest N₂O concentrations (13.4 ± 3.1 nmol L⁻¹) and saturation values (149.6 ± 28.1%) (Supplementary Figure S5). In contrast, beneath the NPCW, the NPIW (27.4 ± 5.8 nmol L⁻¹) exhibited higher N₂O concentrations. Within the intermediate layer of the PDW, the highest N₂O concentration and saturation were observed (31.1 ± 2.7 nmol L⁻¹ and 280.4 ± 29.9%, respectively), and the maximum N₂O (35.3 nmol L⁻¹ and 324.5%) was observed in the core of the oxygen minimum layer (Figures 1D, 3A). Below the core depths, N₂O concentration tended to decrease as DO concentration increased with depth.

Our observations are consistent with the findings of previous studies. An earlier study reported higher N₂O concentrations (~50 nmol L⁻¹) at similar latitudes in the eastern region (160°E) of the PDW (Butler et al., 1989), with decreasing DO concentrations toward the southeast. Toyoda et al. (2002) and Breider et al. (2015) also reported similar N₂O profiles in the intermediate layer of nearby stations. However, the maximum N₂O was higher in the latter by approximately 42 nmol kg⁻¹ (approximately 43.2 nmol L^-1 at a density of 27.4 kg m^-3) and the DO was lower (approximately 40 μmol L⁻¹).

In contrast, the N₂O concentrations of the PDW in the deep layer showed a gradually decreasing trend (~23.9 ± 2.1 nmol L⁻¹) with increasing depth and DO concentrations (Figure 3A). Additionally, with a stable physicochemical environment (Figure 1B–E), the underlying LCDW comprised homogeneous N₂O distributions (20.8 ± 1.1 nmol L⁻¹ and 171.8 ± 9.5%) (Supplementary Figure S5). These results are consistent with previous studies, where N₂O concentrations in the deep layer decreased with increasing depth and converged to approximately 20 nmol kg⁻¹ (approximately 20.6 nmol L^-1 at an average depth of 27.74 kg m^-3) (Toyoda et al., 2002). Moreover, in the deep layers (depths below 2000 m), the N₂O concentrations of the North Atlantic Ocean (i.e., 13.6 ± 2.3 nmol L⁻¹) were lower than those in the North Pacific Ocean, indicating an accumulation of N₂O in deep waters over time (Bange and Andreae, 1999).

The relationship between ΔN₂O and other biogeochemical tracers, represented by the apparent oxygen utilization (AOU = [DO]_{eq (θ, S)} − [DO]_measured) and N^*, has been widely used to estimate the biogeochemical production and/or consumption of N₂O in various marine environments (Nevison et al., 2003; Forster et al., 2009; Zhang et al., 2015; Chen et al., 2021). The AOU is typically interpreted as the amount of DO consumed during remineralization (Redfield et al., 1963; Sarmiento and Gruber, 2006), and a positive correlation between ΔN₂O and AOU indicates that nitrification ( NH 4 + → NO 2 − → NO 3 − ) is the main pathway for ΔN₂O production (Yoshinari, 1976; Nevison et al., 2003). In addition, a negative correlation between ΔN₂O and N^* indicates N₂O production via denitrification (Yamagishi et al., 2005; Kock et al., 2016).

The distribution of surface N₂O increased with depth, with slightly supersaturated values relative to equilibrium values. Higher N₂O concentrations than the N₂O capacity of seawater within the surface layer indicated that excess N₂O (i.e., ΔN₂O) was produced within the water column. In addition, fluctuations in physical parameters (i.e., temperature and salinity) could be responsible for the supersaturated N₂O observed within the surface layer. Because the rate of N₂O exchange is slower than that of heat exchange, a discrepancy between gas and thermal equilibrium may occur in the mixed layer, resulting in the N₂O supersaturation of surface water (Zhan et al., 2017).

To investigate the biological effects on N₂O dynamics, we evaluated the linear relationships between ΔN₂O and AOU (Figure 3B) and the linear relationship between ΔN₂O and NO 3 − (Supplementary Figure S6). However, in the mixed layer, the AOU may have been distorted because of biological activity (photosynthesis) and interactions with the atmosphere. The correlation between ΔN₂O and AOU in the 0–200 m layer was weaker (R² = 0.39, Supplementary Figure S7) than that (R² = 0.48) for the 100–200 m layer. Therefore, we set a depth of 100 m as the boundary of the mixed layer (Yoon et al., 2022), and discussed the results below 100 m. In the surface layer, ΔN₂O had relatively strong correlations with AOU and NO 3 − (R² = 0.48 and R² = 0.34, respectively) (Figure 3B and Supplementary Figure S6). Generally, nitrification in the surface layer is limited by the photoinhibition of nitrifying microorganisms and competition with phytoplankton for ammonia (Lomas and Lipschultz, 2006). Several studies have suggested the possibility of nitrification within the euphotic zone (Wankel et al., 2007; Rafter and Sigman, 2016; Stephens et al., 2020). Therefore, ΔN₂O–AOU and NO 3 − correlations suggested that the slightly supersaturated N₂O of the surface layer during this investigation may be due to the products derived from nitrification within the euphotic zone, and gradually increasing nitrification near the boundary of the euphotic zone. Recent observations of N₂O production reinforce this suggestion via nitrification at a nearby location (30°N, 145°E) (Breider et al., 2015; Yoshikawa et al., 2016). In addition, the ΔN₂O of the surface layer included both in-situ produced N₂O and transported N₂O. Approximately 30–50% of the surface ΔN₂O may originate from production within the subsurface water at 100–300 m in the subtropical North Pacific Ocean (Popp et al., 2002; Breider et al., 2015).

The highest N₂O concentration and saturation levels found in the intermediate layer in this investigation suggest that N₂O production was vigorous. In addition, the correlation between ΔN₂O and AOU and between ΔN₂O and NO 3 − was significant (R² = 0.92 and R² = 0.91, respectively) (Figure 3B and Supplementary Figure S6). In addition, a negative correlation between ΔN₂O and N^* was evaluated in this layer, where denitrification signals were most likely to be detected, and the correlation between ΔN₂O and N^* was relatively strong (R² = 0.60) (Figure 3C). These results suggest that nitrification and denitrification may contribute to the ΔN₂O production in the intermediate layer, as discussed in the next section. Yoshida et al. (1989) were the first to suggest that both nitrification and denitrification are potential sources of N₂O through isotopic composition analysis of the western Pacific Ocean (WPO). Similarly, other studies have observed nitrification as a significant source of the highest N₂O levels (Toyoda et al., 2002; Breider et al., 2015).

Supersaturated N₂O in the deep layer was not as high as in the intermediate layer, but a significant amount of N₂O was observed in this study. We determined that the relationships between ΔN₂O and AOU and between ΔN₂O and NO 3 − were strongly linear (positively correlated at R² = 0.92 and R² = 0.81, respectively) (Figure 3B and Supplementary Figure S6). The relatively homogeneous N₂O concentrations and hydrographic conditions at the bottom, and the tendency of N₂O and AOU (and NO 3 − ) to increase with shallower depths suggest that nitrification is the primary source of N₂O accumulation in the deep layer.

The significant correlation between N₂O and N^* in the intermediate layer is likely represented by the N₂O production signal via denitrification (hereafter referred to as ‘potential denitrification’). However, despite hypoxic conditions, the oxygen concentration in the oxygen minimum layer was higher than that in the oxygen minimum zone (DO < 20 µmol L^-1) (Paulmier and Ruiz-Pino, 2009), where denitrification usually occurs (Bianchi et al., 2012). A possible explanation for potential denitrification in the open ocean is the formation of ‘micro-reducing environments’ provided by marine snow (Kim et al., 2013).

As aggregates descend through the water column, they entangle fine suspended particles and accumulate more in the intermediate layer (Alldredge et al., 1990). In the oxygen minimum layer, especially under hypoxic conditions, oxygen influx to the aggregates may be lower. These conditions may form a more favorable reducing microenvironment (Shanks and Reeder, 1993) and allow potential denitrification in the microzones despite aerobic conditions (Li and Peng, 2002). Indeed, aggregates consisting of not only nitrifiers but also denitrifiers on the suspended particles were observed, suggesting that the suspended particles may provide a niche in which coupled nitrification and denitrification processes can occur (Zhu et al., 2018). Yamagishi et al. (2005) also suggested the possibility of denitrification within microzones of aggregates despite the aerobic conditions in the North Pacific Ocean (44°N, 155°W).

As Frey et al. (2020) found that the addition of particulate organic matter increased N₂O production by denitrification, high availability of NO 3 − increases N₂O production through denitrification (Weier et al., 1993; Ji et al., 2015). Furthermore, the high NO 3 − concentration in the oxygen minimum layer may provide easy access to NO 3 − . Marchant et al. (2017) also suggested that denitrification aids in respiration under aerobic conditions when the electron acceptor and donor concentrations fluctuate over short temporal and spatial scales, as in permeable environments. This background and our results support potential denitrification. However, it is necessary to exclude the possibility of alternative explanations for the potential denitrification signals, although there is no direct evidence to support the occurrence of potential denitrification in the intermediate layer in the study area.

First, the low N^* values and high N₂O concentrations in the intermediate layer are likely due to the influence of the NPIW, which may transport denitrification signals from the eastern North Pacific Ocean, where oxygen is completely depleted in oxygen-deficient zones. Although the NPIW contained relatively low N^* values and high N₂O concentrations at depths adjacent to the oxygen minimum layer, the core of the NPIW exhibited N^* of -1.09 μmol L^-1 (Figures 2, 3A). We determined that the N^* minimum (-7.28 μmol L^-1) and N₂O maximum (35.3 nmol L⁻¹) values were identified in the oxygen minimum layer occupied by the PDW (Figures 1D, 3). Moreover, while the NPIW core was found at ~600 m (salinity minimum in Figure 1C), the lowest N^* signals appeared at ~1250 m, dominating the PDW. These results indicate that NPIW is not a significant contributor to potential denitrification signals.

Second, the N^* technique can not distinguish between denitrification signals from the water column and sedimentary interactions, assuming that all nitrate sinks result from potential denitrification. Hence, we focused on the intermediate layer, which has no direct contact with the bottom layer. Nevertheless, as the PDW originates from the upwelling of bottom water and a mixture of deep waters (Talley, 2011), the water mass history that may involve substantial exposure to the sediment needs to be considered. If the interaction with the sediment affects the signal, the N^* value decreases as it approaches the bottom. However, the core of the LCDW contained a relatively high N^* value (-2.68 μmol L^-1), and low N₂O concentrations (20.0 nmol L⁻¹) (Figures 2, 3A) compared to the PDW. These results suggest that the influence of NPIW and sedimentary interactions were not significant contributors to the potential denitrification signals in the oxygen minimum layer.

Based on these N₂O mechanisms (i.e., nitrification and potential denitrification), we attempted to estimate the fractions of N₂O production derived from physical and biogeochemical processes in the intermediate layer. N₂O is derived from two physical processes. First, the dissolved N₂O concentration of a water parcel can interact with the atmosphere in the surface layer, thereby contributing to the internal N₂O concentration. Second, mixing with water parcels originating from other surface layers can contribute to the internal N₂O concentration as the water parcels move from the surface layer to an internal location.

The biological processes involved in N₂O dynamics in the ocean include nitrification and denitrification. Nitrification is a remineralization process, which consumes oxygen (AOU) and produces N₂O as a by-product. Freing et al. (2009) used ΔN₂O/AOU as a quantitative indicator of N₂O production through nitrification. In contrast, denitrification uses nitrate as an electron acceptor instead of oxygen in oxygen-deficient environments, thereby losing nitrogen. Accordingly, N^* represents the nitrogen lost compared with phosphate through denitrification. As denitrification also produces N₂O as a by-product, the relationship between N^* and ΔN₂O was used as an indicator of N₂O production through denitrification (Kock et al., 2016; Fenwick et al., 2017).

Before calculating these fractions, we made the following assumptions: (1) N₂O derived from physical processes is maintained in equilibrium with the atmosphere before ventilation of the water parcels and (2) N₂O derived from biogeochemical processes is produced only through nitrification and potential denitrification. Although diverse processes interact in the ocean environment, we focused on evaluating these assumptions. The fractions of N₂O produced were calculated as follows: (5) N 2 O m e a s u r e d   ≈ N 2 O P h y s i c a l +   N 2 O N i t r i f i c a t i o n +   N 2 O d e n i t r i f i c a t i o n P o t e n t i a l (6) N 2 O N i t r i f i c a t i o n  =  r Δ N 2 O : A O U   × A O U Z i (7) N 2 O d e n i t r i f i c a t i o n P o t e n t i a l   =   r Δ N 2 O : N * × N Z i * where N₂O_measured is the measured N₂O concentration, N₂O_Physical is the N₂O concentration derived from the physical process (i.e., N₂O_eq), N₂O_{Nitrification} and   N 2 O d e n i t r i f i c a t i o n P o t e n t i a l are N₂O produced by nitrification and potential denitrification, r_{ΔN 20:AOU} and r_{ΔN 20:N *} are the linear regression slopes between ΔN₂O and AOU and between ΔN₂O and N^*, respectively (Figure 3B), and Zi is the depth (Z) at each station (i). The relationships between ΔN₂O and AOU and between ΔN₂O and N^* were based on N₂O production according to the oxygen consumption rate through nitrification and N₂O production according to the nitrate sink through denitrification, respectively. The errors in r_{ΔN 20:AOU} and r_{ΔN 20:N *} were ±7.1% and ±19.3%, respectively, at a confidence level of 95%. Nitrification is a remineralization process, which consumes oxygen, in turn provides nitrate ( NO 3 − ) to the water column. Because NO 3 − is also a by-product of nitrification, the linear regression slope between ΔN₂O and NO 3 − (i.e., r Δ N 2 O : N O 3 − ) represents N₂O production through nitrification. The error in r Δ N 2 O : N O 3 − was ±7.5% at a confidence level of 95%.

To ensure the reliability of the slopes in estimating N₂O production derived from biogeochemical processes, we compiled a list of slopes obtained for diverse regions (Supplementary Table S4). The slopes of the STWNPO were lower than those in other regions known to have higher productivity (e.g., the eastern tropical Pacific Ocean and eastern tropical Atlantic Ocean). The N₂O distributions and biogeochemical tracer slopes for the tropical and subtropical North Atlantic Ocean were similar to those in our observations for STWNPO. Although their extremes differ, the two regions show similar hydrographic trends because the upper circulations in both regions are mainly driven by the wind (Talley, 2011). These environments influence N₂O dynamics and may explain the similar results observed in both regions. Therefore, we assumed that the slope values observed in this study were acceptable for use in our assessment.

The mean fractions of N₂O production in the intermediate layer were estimated using the AOU and NO 3 − observations (Supplementary Figure S8). Although there was a slight variation in the fraction values between the AOU and NO 3 − , the trend was so similar that we employed the estimations obtained using the AOU. Given the error of r_{ΔN 20:AOU} the error for estimating N₂O_{Nitrification} was ±1.8%. Given the errors of N^* and r_{ΔN 20:N *}, the errors for estimating   N 2 O denitrification Potential were ±6.9% and ±3.1%, respectively. Accordingly, the combined uncertainty in   N 2 O denitrification Potential was estimated to be ±12.4%. Among the total N₂O concentrations within the intermediate layer, the fractions of biogeochemically derived N₂O were greater than those of physical processes. Because of the influence of the oxygen minimum layer in the intermediate water column, N₂O production from potential denitrification in this study was higher at high latitudes, where lower DO was observed (Supplementary Figure S8). The fractions of N₂O produced by potential denitrification in biogeochemical processes range from 17 to 25%, suggesting that N₂O may be produced by potential denitrification in hypoxic waters.

In addition, the N₂O production rates (i.e., r_{ΔN 20:AOU} and r_{ΔN 20:N *}) of water masses adjacent to the oxygen minimum layer were compared to investigate the potential N₂O contribution during the mixing of the water masses (Supplementary Table S5). The NPIW overlying the upper boundary of the oxygen minimum layer exhibited a relatively high r_{ΔN 20:AOU} (0.11) and r_{ΔN 20:N *} (-1.35). Compared with the production rates in the intermediate layer, the NPIW may contribute approximately 55% of N₂O_{Nitrification} and 37% of   N 2 O denitrification Potential in proportion to the mixing ratio. The PDW in the deep layer, which is underlying the lower boundary of the oxygen minimum layer, also exhibited higher r_{ΔN 20:AOU} (0.16) and r_{ΔN 20:N *} (-0.84) and may contribute 64% of N₂O_{Nitrification} and 27% of   N 2 O denitrification Potential in proportion to mixing ratio.

For global biological N₂O estimation (Freing et al., 2012), the impact of denitrification on N₂O production was considered to occur only in the Arabian Sea and eastern tropical Pacific Ocean. In the oxygen minimum layer of the Arabian Sea and eastern tropical North and South Pacific, the fractions of N₂O produced through denitrification ranged up to 89%, 35%, and 14%, respectively. They produced higher estimates of the fractions in the Arabian Sea and eastern tropical North Pacific Ocean and lower fractions in the eastern tropical South Pacific Ocean than the present estimates for the STWNPO, where we observed higher DO concentrations in the oxygen minimum layer.

These differences may be attributed to several factors that were not considered in the present study. Non-constant remineralization ratios, including N:P, can vary with depth. The N:P ratio in the Pacific Ocean has been reported to be approximately 12–14 and 15–16 between 1000 and 3000 m and 3000–4000 m, respectively, which influences N^* (Anderson and Sarmiento, 1994). However, the present estimation was performed using a constant Redfield ratio of 16, and the preformed nutrients were not considered. Furthermore, as N^* is indirect evidence for denitrification, our estimations using N^* involve uncertainty. Our estimates could not distinguish N₂O production via nitrifier denitrification, although nitrifier denitrification is known to contribute to N₂O production in STWNPO (Breider et al., 2015; Yoshikawa et al., 2016). In addition, we simplified the physical processes involved in the N₂O dynamics. Significant uncertainties likely accompany these factors. However, as providing the possibility of potential denitrification in the STWNPO is still crucial in the global N cycle, future studies involving tracers such as nitrogen and oxygen isotopes and information on aggregate microzones are needed.

To determine whether the STWNPO was a source or sink for atmospheric N₂O during this investigation, we used the air–sea gas exchange equation presented below: (8) N 2 O f l u x   =   k w · ( [ N 2 O ] w a t e r s u r f a c e − [ N 2 O ] e q ) where [ N 2 O ] water surface is the N₂O concentration in the surface water adjacent to the atmosphere and k_w is the gas transfer velocity (cm h⁻¹). k_w was determined by physical factors (i.e., temperature and wind speed at 10 m). It is significantly influenced by the wind speed because it is proportional to the exponent of the wind speed. During this survey, the maximum wind speed observed at St. 8 was nearly double the overall average wind speed (Supplementary Table S6). Weighted mean wind data (e.g., wind speed data for the 60 days prior to sampling) are sometimes used to avoid overestimating short-term weather fluctuations, such as gusts (Fenwick et al., 2017; Zhan et al., 2021). Although this approach reduces the impact of meteorological changes, it does not accurately reflect hydrographical changes in the ocean. We used average wind speed during the survey to obtain better observation-based results. In addition, to mitigate the differences in k_w owing to dynamic wind speed intensities, we used three k_w models with different exponents of wind speed, as follows: (9) k w W M 1999   =   0.0283 × U 10 3 × ( S c 660 ) − 1 2 (10) k w N 2000   = ( 0.333 × U 10 + 0.222 × U 10 ) 2 × ( S c 600 ) − 1 2 (11) k w W 2014   =   0.251 × U 10 2 × ( S c 660 ) − 1 2 where WM1999, N2000, and W2014 represent the model references (i.e., Wanninkhof and McGillis (1999), WM1999; Nightingale et al. (2000), N2000; Wanninkhof (2014), W2014), U₁₀ is the wind speed 10 m above the sea surface. Sc is the Schmidt number, defined as the kinematic viscosity/molecular diffusion (Bergman et al., 2011). The empirical equation proposed by Wanninkhof (2014) was used to calculate the Schmidt number. The three k_w models are widely used for ocean environments and have been verified to yield reasonable results [e.g., in the Bering Sea and Southern Ocean (Chen et al., 2014); in the Gulf of Mexico (Kim, 2018); in the STWNPO (Zhan et al., 2017)]. In addition to N₂O flux due to gusts, we directly compared N₂O fluxes using three k_w models to assess how wind influences N₂O flux estimations in terms of N₂O dynamics. The N₂O fluxes estimated by the models during this investigation of the STWNPO are provided in Supplementary Table S7, where the average values of the N₂O flux from the three models are used to facilitate the presentation of the current results. In addition, both unmodified N₂O_eq and N₂O_eq adjusted by TTD ages were calculated using contemporary atmospheric N₂O values at depths adjacent to the atmosphere. The standard deviation of contemporary atmospheric N₂O used in calculating N₂O fluxes was ±0.86 ppb. Given the standard deviation, the uncertainty in calculating the N₂O flux was remarkably low at ±0.05 µmol m⁻² d⁻¹. When calculating N₂O fluxes, the uncertainty of the measured N₂O should also be considered. Given the standard deviation of gas standard measurements, the error of N₂O fluxes was 0.6 ± 0.7 µmol m⁻² d⁻¹.

During this study, the average N₂O flux of the STWNPO was 2.0 ± 0.3 µmol m⁻² d⁻¹ (Supplementary Table S7). Excluding St. 8 (where it was windy during the observations), the average N₂O flux of the stations was 1.5 ± 0.2 µmol m⁻² d⁻¹, and in contrast, the St. 8 flux was comparatively high (5.6 ± 1.3 µmol m⁻² d⁻¹). Although the N₂O saturation at St. 8 was comparatively lower than at lower latitudes (see Section 3.2), the high N₂O flux would be due to intense wind speeds. This suggests that in STWNPO, surface waters are already supersaturated with N₂O and could release more N₂O into the atmosphere during physical stress.

The N₂O fluxes of the STWNPO varied with time and the estimation methods used. Butler et al. (1989) estimated that N₂O fluxes ranged from 0.04 to 0.13 µmol m⁻² d⁻¹ in the subtropical gyre of the West Pacific Ocean. In that study, the average surface anomaly was lower (~2.5%). In general, open oceans at low latitudes, such as the tropical and subtropical Pacific Ocean, have low variability, and their effects on N₂O dynamics are small. For example, simulated seasonal variations in parameters related to N₂O dynamics showed little variation in STWNPO (30° N, 145° E) (Yoshikawa et al., 2016). The N₂O flux fluctuated slightly higher in winter, but this was because of the wind speed. The surface N₂O of the STWNPO originates from N₂O production through nitrification, and the nitrate concentration also shows little seasonal variation. For there to be some effect on existing N₂O dynamics, a relatively large event must occur. Butler et al. (1989) highlighted El Niño as an event present in their study. El Niño, which is a natural climate variability in the Pacific Ocean, can reduce the upward diffusion of rich subsurface N₂O by suppressing upwelling, which consequently reduces N₂O efflux (Cline et al., 1987; Ji et al., 2019). The present study area is distant from this interannual event centered on the equator, which may have caused the differences in the results.

In addition, Zhan et al. (2017) estimated N₂O fluxes in STWNPO using the k w W 2014 method. Their flux estimations were higher (14.8 ± 2.8 µmol m⁻² d⁻¹) than the present result, despite the surface saturation anomaly being similar (~8%). As we calculated the N₂O flux analogously (i.e., following Eqs. 8-11), the difference between our flux values and those of Zhan et al. (2017) are most likely due to differences in wind speed around the time of sampling. Some attempts have been made to estimate N₂O flux in the western North Pacific using the isotopic mass balance method (Breider et al., 2015; Yoshikawa et al., 2016). Their findings showed that the N₂O distributions in the subtropical and subarctic regions were similar (~6–8 nmol kg⁻¹) and higher (~10–11.9 nmol kg⁻¹) than those in our results, and the N₂O fluxes were lower in the subtropical region (0.5–0.9 µmol m⁻² d⁻¹) than in the subarctic region (~6.3 µmol m⁻² d⁻¹), but the overall flux values were lower than those in the present observation. Although the spatial and temporal variability of N₂O production in the STWNPO is very low, our study showed different results related to N₂O flux compared with other studies. These variations could be the reasons for the different approaches to estimating the N₂O fluxes in the ocean environment.

The STWNPO serves as an apparent source of atmospheric N₂O. However, like many other regions worldwide, this region is undergoing various climatic changes that may affect N₂O dynamics. To understand the variation in N₂O flux in the STWNPO, we compared the present results with those from other regions dominated by different N₂O dynamics (Figure 4 and Supplementary Table S8). The N₂O flux in global ocean conditions fluctuates, but the average flux of a normoxic ocean is estimated to be under 1 µmol m⁻² d⁻¹ (Naqvi et al., 2010). N₂O flux values tended to be higher in the coastal region than in the open ocean (Figure 4). However, they were occasionally significantly higher in the several regions characterized by high productivity and depleted oxygen (upwelling region). In an upwelling environment, subsurface water is directed toward the surface, accompanied by accumulated N₂O. This condition also provides a nutrient-rich environment with low DO and triggers high productivity and remineralization. The depletion of oxygen produces N₂O from nitrification and denitrification. As N₂O is also consumed by further denitrification (N₂O → N₂) in oxygen-depleted waters, relatively low N₂O emissions (but still high compared with normal oceans) into the atmosphere were observed in the eastern tropical South Pacific Ocean at 12.7–30.7 µmol m⁻² d⁻¹ (Farías et al., 2009). Nevertheless, significant N₂O effluxes in the upwelling region have been recorded at up to 3243 µmol m⁻² d⁻¹ (Naqvi et al., 2006; Arevalo-Martínez et al., 2015; Fenwick and Tortell, 2018; Ji et al., 2019; Morgan et al., 2019). In addition, N₂O effluxes (2.3 ± 2.7 μmol m^-2 d^-1) were emitted from the southern Chukchi Sea (Heo et al., 2021), which is a highly productive area (Bates and Mathis, 2009). Here, N₂O was derived from nitrification in the water column, and upward transport originated from nitrification or denitrification in the sediment due to the shallow depths, despite oxic environmental conditions.

Based on our results, we speculated on the implications of ocean warming, deoxygenation, acidification, and anthropogenic nitrogen deposition (AND) on the overall N₂O dynamics of STWNPO. Since 1900, the western boundary current of the North Pacific Ocean has experienced two-fold greater warming trends than the global average (Wu et al., 2012). Ocean warming is a significant driver of deoxygenation (Levin, 2018), the Oyashio region (NPIW origin) has undergone deoxygenation and increased stratification. The deoxygenation originating therein has propagated into the interior of the North Pacific Ocean (Sasano et al., 2018). The intermediate water column of the middle-and high-latitude North Pacific may lose more oxygen in the future (Bopp et al., 2017). These environmental changes could expand the oxygen minimum layer of STWPO. In addition, ocean acidification in the STWNPO has accelerated in recent years (Ono et al., 2019), and evidence of acidification was observed in this investigation (Supplementary Figure S9). Furthermore, AND has also been increasing because of human activities that disturb the N cycle in the western North Pacific Ocean (Kim et al., 2014).

The increase in the thermal energy transmitted into the western Pacific Ocean, where N₂O is slightly supersaturated due to its production via nitrification (Figure 5), decreases gas solubility. Subsequently, the increased N₂O efflux into the atmosphere can accelerate the greenhouse effect (positive feedback). The expansion of the oxygen minimum layer could potentially elevate denitrification-based N₂O production (positive feedback) (Figure 5). Decreased pH conditions in the ocean also decrease nitrification and N₂O production (Beman et al., 2011). For instance, recent reports have observed decreases in N₂O production from 44% to 2.4% (Rees et al., 2016), and from 72.38 to 50.77% (Gu et al., 2021), along with decreased pH values. Further, in the western North Pacific Ocean, acidified conditions decreased nitrification but increased N₂O production through the ammonia-oxidizing archaeal process (Breider et al., 2019) and which has also been observed in a variety of other environments, including in freshwater and upwelling regions (Frame et al., 2017). Despite this evidence, apparent inhibition of nitrification by acidification has been observed in the Pacific and Atlantic oceans (Beman et al., 2011), suggesting that accelerating acidification in the western North Pacific may decrease nitrification rates and N₂O production (negative feedback) (Figure 5). AND stimulates productivity by supplying limited nutrients and it consequently elevates nitrification and N₂O production (positive feedback) (Suntharalingam et al., 2012). Atmospheric dust deposition, the vehicle for AND, directly acidifies the seawater. Nitrification elevated by AND-driven productivity also provides a greater H⁺ concentration in the water column, leading to ocean acidification (negative feedback) (Doney et al., 2007; Sarma et al., 2021). Therefore, AND exerts a combination of positive and negative effects on N₂O production in STWNPO. However, our present speculations lack direct observations of related parameters, making it difficult to determine a causal relationship. With regard to these speculations, future studies must focus on the impacts of climate change to better understand N₂O dynamics and the associated air–sea gas exchange in STWNPO.