Triple oxygen isotope evidence for the pathway of nitrous oxide production in a forested soil with increased emission on rainy days

Continuous increases in atmospheric nitrous oxide (N₂O) concentrations are a global concern. Both nitrification and denitrification are the major pathways of N₂O production in soil, one of the most important sources of tropospheric N₂O. The ¹⁷O excess (Δ¹⁷O) of N₂O can be a promising signature for identifying the main pathway of N₂O production in soil. However, reports on Δ¹⁷O are limited. Thus, we determined temporal variations in the Δ¹⁷O of N₂O emitted from forested soil for more than one year and that of soil nitrite (NO${}_{\mathrm{2}}^{-}$), which is a possible source of O atoms in N₂O. We found that N₂O emitted from the soil exhibited significantly higher Δ¹⁷O values on rainy days (+0.12 ± 0.13 ‰) than on fine days (−0.30 ± 0.09 ‰), and the emission flux of N₂O was significantly higher on rainy days (38.8 ± 28.0 µg N m⁻² h⁻¹) than on fine days (3.8 ± 3.1 µg N m⁻² h⁻¹). Because the Δ¹⁷O values of N₂O emitted on rainy and fine days were close to those of soil NO${}_{\mathrm{2}}^{-}$ (+0.23 ± 0.12 ‰) and O₂ (−0.44 ‰), we concluded that although nitrification was the main pathway of N₂O production in the soil on fine days, denitrification became active on rainy days, resulting in a significant increase in the emission flux of N₂O. This study reveals that the main pathway of N₂O production can be identified by precisely determining the Δ¹⁷O values of N₂O emission from soil and by comparing the Δ¹⁷O values with those of NO${}_{\mathrm{2}}^{-}$, O₂, and H₂O in the soil.

1 Introduction

Nitrous oxide (N₂O) is a strong greenhouse gas and an essential substance in stratospheric ozone depletion (Dickinson and Cicerone, 1986). Since pre-industrial times, the atmospheric N₂O level has increased by 24 % to 335.8 ppb, with an average growth rate of 1.05 ppb yr⁻¹ in the last decade (WMO, 2023). Terrestrial soils account for approximately 60 % of total N₂O emissions (Tian et al., 2020). Therefore, better knowledge of the pathways of N₂O production in soils is required to establish mitigation measures.

Both nitrification and denitrification are representative microbial pathways of N₂O production in soils (Wrage et al., 2001). Nitrification is the oxidation of ammonium (NH${}_{\mathrm{4}}^{+}$) to nitrate (NO${}_{\mathrm{3}}^{-}$) via aerobic microbial activity, during which N₂O is produced as a byproduct of hydroxylamine (NH₂OH) oxidation to nitrite (NO${}_{\mathrm{2}}^{-}$), while denitrification is the reduction of NO${}_{\mathrm{3}}^{-}$ to NO${}_{\mathrm{2}}^{-}$ and then to N₂O which is further reduced to nitrogen (N₂) via facultative anaerobes (Fig. 1). Soil conditions such as moisture content, O₂ availability (Bateman and Baggs, 2005; Zhu et al., 2013), temperature (Luo et al., 2007), and fertilizer types (Zhu et al., 2013) have been proposed as parameters to determine the pathways of N₂O production in soils.

Figure 1 Schematic showing the pathways of N₂O production in soil (Kool et al., 2007, 2011; Wankel et al., 2017; Wrage et al., 2005) and the Δ¹⁷O values of O₂ (Sharp et al., 2016), NO${}_{\mathrm{2}}^{-}$, and H₂O (Uechi and Uemura, 2019). The orange lines, green lines, and blue dash lines indicate the processes of nitrification, denitrification, and the possible contributions of O atoms derived from soil H₂O through nitrification and denitrification, respectively.

Techniques such as acetylene blockage (Balderston et al., 1976; Lin et al., 2019), artificial isotope tracers (¹⁵N and ¹⁸O) (Mulvaney and Kurtz, 1982; Wrage et al., 2004), and natural stable isotopes (Toyoda et al., 2013; Yu et al., 2020) are conventionally used to identify the pathways of N₂O production via nitrification and denitrification. Both acetylene blockage and artificial isotope tracers are mostly performed in laboratory (in vitro) incubations because they are costly, complicated, and time-consuming in field research. Natural stable isotopes such as δ¹⁵N, δ¹⁸O, and SP (¹⁵N site preference) can be used to identify the pathways of N₂O production in soils (Decock and Six, 2013; Toyoda et al., 2017; Verhoeven et al., 2019). However, further reduction of N₂O to N₂ after the production of N₂O until emission from soil to air results in significant changes in the δ¹⁵N, δ¹⁸O, and SP values of N₂O due to the fractionation of isotopes, which makes the identification process difficult (Ostrom et al., 2007).

Recent studies on the Δ¹⁷O value of NO${}_{\mathrm{3}}^{-}$ (the definition detailed in Sect. 2.4) have reported that Δ¹⁷O is a useful natural signature for clarifying the complicated biogeochemical processes in terrestrial ecosystems (Ding et al., 2022, 2023, 2024; Michalski et al., 2004; Tsunogai et al., 2010). Although the values of δ¹⁵N, δ¹⁸O, and SP can vary during various fractionation processes of isotopes within terrestrial ecosystems, the Δ¹⁷O value remains almost stable because possible variations in δ¹⁷O and δ¹⁸O values during the processes of biogeochemical isotope fractionation follow the relation of δ¹⁷O ≈ 0.5 δ¹⁸O, which cancels out the variations in the Δ¹⁷O value (Young et al., 2002). Consequently, the mixing of the same oxygen compounds with different Δ¹⁷O values is the primary cause of variations in Δ¹⁷O values throughout the biogeochemical processes in terrestrial ecosystems.

Because N₂O produced through nitrification is a byproduct of the oxidation reaction between NH${}_{\mathrm{4}}^{+}$ (to NH₂OH) and O₂, the Δ¹⁷O value of N₂O produced through nitrification is expected to be close to that of tropospheric O₂ (Fig. 1) (Kool et al., 2007, 2011; Wrage et al., 2005), with previous studies reporting a Δ¹⁷O value of −0.44 ‰ (Sharp and Wostbrock, 2021). Conversely, the Δ¹⁷O value of N₂O produced through denitrification is expected to be close to that of NO${}_{\mathrm{2}}^{-}$ (Fig. 1) (Kool et al., 2007, 2011; Wankel et al., 2017; Wrage et al., 2005). Because O atoms in NO${}_{\mathrm{2}}^{-}$ are derived from either soil NO${}_{\mathrm{3}}^{-}$ (Δ¹⁷O = from 0 ‰ to +20 ‰) or H₂O (Δ¹⁷O = +0.03 ± 0.01 ‰) (Hattori et al., 2019; Nakagawa et al., 2018; Uechi and Uemura, 2019), significant differences in Δ¹⁷O values between N₂O produced through nitrification and that produced through denitrification are expected if the additional contributions of O atoms derived from soil H₂O are insignificant in N₂O during the processes of N₂O production in soils through nitrification and denitrification (Fig. 1) (Kool et al., 2007).

Previous studies have identified the elevated Δ¹⁷O values in atmospheric N₂O (Δ¹⁷O ≈ +0.9 ‰), observed in both stratospheric and tropospheric air (Cliff et al., 1999; Kaiser et al., 2003; Thiemens and Trogler, 1991). Komatsu et al. (2008) subsequently conducted the first Δ¹⁷O measurements of N₂O emitted from a soil to assess whether soil N₂O could be the source of elevated Δ¹⁷O values of atmospheric N₂O. However, the temporal variations of the Δ¹⁷O values for N₂O emitted from soil remain unknown. Whether Δ¹⁷O values of N₂O can be used to identify the pathways of N₂O production in soils has not been discussed. In addition, the advantages of Δ¹⁷O signature, relative to other natural stable isotopes, for identifying the pathways of N₂O production remain unclear. To address these, in this study, we measured precise Δ¹⁷O values for N₂O emitted from forested soil and those for NO${}_{\mathrm{2}}^{-}$ in the soil. Further, we conducted similar observations in the same soil artificially fertilized with Chile saltpeter or urea to investigate the possible contributions of O atoms derived from soil H₂O in N₂O during N₂O production.

2 Methods

2.1 Study site

The study site was located in a secondary warm-temperate forest within an urban area (35°10^′ N, 136°58^′ E, Fig. 2), approximately 50 m from the common building of the Graduate School of Environmental Studies at Nagoya University. The lowest, highest, and mean monthly temperatures recorded at the nearest meteorological station (Nagoya station) were 5.2 °C (in January), 28.9 °C (in July), and 18.5 °C, respectively, from April 2022 to July 2023. The annual mean precipitation was approximately 1800 mm. The soil stratum in the forested field possessed an approximate depth of 20 cm, characterized by a bulk density of 1.12 g cm⁻³. Details of the forest have been described in the previous study (Hiyama et al., 2005).

Figure 2 Map showing the location of Nagoya, Japan, where the studied site is located (a). Map showing the monitoring site of N₂O emitted from forested soil in a secondary warm-temperate forest (yellow square) and the plots fertilized with Chile saltpeter (CS, blue square), urea (U, purple square), and no fertilizer (NF, gray square) (b). Photo showing the plots and flow chambers set on the plots (c).

2.2 Sampling of N₂O

Samples of N₂O emitted from the forested soil under natural conditions were collected 18 times (n=18) from April 2022 to July 2023 in a field with an area of 5 m² (Fig. 2b). Among the samples, 12 were collected on fine days, whereas 6 were collected on rainy days. A fine day is defined as a day without precipitation for 48 h prior to the end of each sampling. The total precipitation within 12 h at the end of each sampling of the rainy days exceeded 12 mm.

The sampling of N₂O emitted from the artificially fertilized soil was performed during a period of fine weather in three plots (1 m² for each located more than 5 m away from each other) within the same forested field, located approximately 3 m away from the plot where we conducted the sampling under natural conditions (Fig. 2b and c). Either urea (CO(NH₂)₂, 46 % TN) or Chile saltpeter (KNO₃, 14 % TN) was applied to two of the plots (U and CS plots) on 16 July 2023 at the same N amount of 250 kg N ha⁻¹. Urea is a synthetic N fertilizer (Sun and Hope Ltd., Japan), and Chile saltpeter (SQM Ltd., USA) contains NO${}_{\mathrm{3}}^{-}$ with a high Δ¹⁷O value of +19 ‰ (determined through the internationally distributed isotope reference materials USGS-34 and USGS-35). The third plot was blank, meaning no fertilizer was added (NF plot). Sampling of N₂O from each plot was performed twice on days 2 and 6 after the addition of each fertilizer.

To precisely determine Δ¹⁷O of N₂O, more than 60 nmol of N₂O is required (Komatsu et al., 2008), which corresponds to more than 4 L of air containing N₂O at atmospheric concentrations. Accordingly, in this study, a flow chamber made of polypropylene with dimensions of 0.8 m × 0.3 m × 0.18 m was deployed onto the sampling site throughout each day of sampling (Fig. S1). This chamber has an inlet and outlet port with an inner diameter of 1 cm. The outlet port was connected to an air pump using Tygon tubing, and the inlet port was open to ambient air. Using the air pump, the air in the chamber was taken into a 5 L aluminum bag, along with the gases emitted by the soil, as illustrated in Fig. S1. The flow rate of the air pump was set at 100 mL min⁻¹ throughout the deployment of the chamber; thus, each sampling lasted 45 min until 4.5 L of gas was collected into the aluminum bag. Each gas sampling was started 2 h after deployment of the flow chamber; thus, it took more than 8 h to collect four samples. In addition to the gas samples emitted from the soil, ambient air in the forest was sampled into two 3 L vacuum stainless steel canisters (SilcoCan, Restek).

2.3 Sampling and analysis of forested soil

After collecting the gas samples to determine N₂O, a soil sample (approximately 150 g) was randomly collected from more than four places beneath the chamber. Approximately 20 g of the soil sample was heated at 80 °C for 48 h to estimate the water content from the weight loss and water-filled pore space (WFPS; the calculation was detailed in Text S1 in the Supplement). Using the remaining soil sample (120 g), NH${}_{\mathrm{4}}^{+}$, NO${}_{\mathrm{3}}^{-}$, and NO${}_{\mathrm{2}}^{-}$ in each soil sample were extracted into 120 mL of a 2 M KCl solution, and their concentrations were determined using a high performance microflow analyzer (QuAAtro 39 Autoanalyzer, BLTEC, Osaka, Japan).

2.4 Concentration and isotopic compositions of N₂O

The gas samples collected in aluminum bags or stainless canisters were subsampled into a 100 mL pre-evacuated glass bottle to determine the concentration ([N₂O]), δ¹⁵N, and δ¹⁸O of N₂O simultaneously. The remaining samples were further subsampled to either 1 or 2 L pre-evacuated glass bottles to determine the Δ¹⁷O of N₂O. The concentration and isotopic compositions (δ¹⁵N, δ¹⁸O, and Δ¹⁷O) of N₂O were determined using a continuous flow isotope ratio mass spectrometry (CF-IRMS; Finnigan MAT252, Thermo Fisher Scientific, Waltham, MA, USA) system that consists of an original pre-concentrator system, chemical traps, and gas chromatograph at Nagoya University (Komatsu et al., 2008). The analytical procedures using the CF-IRMS system were the same as those detailed in previous studies (Hirota et al., 2010; Komatsu et al., 2008).

The isotopic ratios of ¹⁵N $/$ ¹⁴N,¹⁷O $/$ ¹⁶O, and ¹⁸O $/$ ¹⁶O are expressed in the δ notations:

$$\begin{array}{}\text{(1)}& {\mathit{\delta }}^{\mathrm{15}}\mathrm{N},{\mathit{\delta }}^{\mathrm{17}}\mathrm{O},\text{or}{\mathit{\delta }}^{\mathrm{18}}\mathrm{O}=R_{\mathrm{sample}}/R_{\mathrm{standard}}-\mathrm{1},\end{array}$$

where R denotes ¹⁵N $/$ ¹⁴N, ¹⁷O $/$ ¹⁶O, or ¹⁸O $/$ ¹⁶O ratios of the sample and each standard reference material.

The Δ¹⁷O of N₂O, including NO${}_{\mathrm{2}}^{-}$, NO${}_{\mathrm{3}}^{-}$, H₂O, and O₂, is defined by Eq. (2) (Kaiser et al., 2007; Miller, 2002):

$$\begin{array}{}\text{(2)}& {\mathrm{\Delta }}^{\mathrm{17}}\mathrm{O}={\displaystyle \frac{\mathrm{1}+{\mathit{\delta }}^{\mathrm{17}}\mathrm{O}}{{\left(\mathrm{1}+{\mathit{\delta }}^{\mathrm{18}}\mathrm{O}\right)}^{\mathit{\beta }}}}-\mathrm{1},\end{array}$$

where β denotes the slope of the reference line in the δ¹⁷O−δ¹⁸O space. Previous studies have proposed values ranging from 0.525 to 0.5305 for β during the various processes of isotope fractionation through experimental measurements and/or theoretical calculations (Cao and Liu, 2011; Matsuhisa et al., 1978; Pack and Herwartz, 2014; Sharp and Wostbrock, 2021). In this study, we adopted a value of 0.528 for β to define Δ¹⁷O. The details of the ranges of the possible Δ¹⁷O variations due to the ranges of β are presented in Sect. 4.1.

To calibrate the δ¹⁵N and δ¹⁸O of N₂O to the international scale, N₂O in a tropospheric air sample collected at Hateruma Island in 2010 (Japan) was used as the standard with a δ¹⁵N value of +6.5 ‰ and a δ¹⁸O value of +44.3 ‰ (Toyoda et al., 2013). To calibrate the Δ¹⁷O of N₂O on the international Vienna Standard Mean Ocean Water (VSMOW) scale, we prepared two kinds of N₂O standards with different Δ¹⁷O values calibrated using a conventional method (Thiemens and Trogler, 1991). The procedures for this calibration are presented in Sect. 2.6, with the details of the N₂O standards. Through repeated measurements of N₂O in a tropospheric air sample collected at Nagoya University, the analytical precisions (1σ) of the measurements were estimated to be ± 10.0 ppb, ± 0.5 ‰, ± 0.6 ‰, and ± 0.11 ‰ for concentration, δ¹⁵N, δ¹⁸O, and Δ¹⁷O, respectively (Fig. S2). To achieve higher precision, analyses of Δ¹⁷O were performed at least three times for each sample, resulting in a standard error (SE) of ± 0.06 ‰.

2.5 Emission flux

Based on the change in the concentration of N₂O from the inlet to the outlet, the emission flux of N₂O from the soil was calculated using Eq. (3):

$$\begin{array}{}\text{(3)}& \mathrm{Flux}={\displaystyle \frac{P\times V\times (C_{\mathrm{final}}-C_{\mathrm{air}})\times M}{R\times T\times t\times A}},\end{array}$$

where Flux denotes the emission flux of N₂O (µg N m⁻² h⁻¹), P denotes the pressure (Pa), V represents the volume of the gas sample in the aluminum bag (0.0045 m³), C_final denotes the concentration of N₂O in the gas sample taken at the end of each deployment of the chamber (µmol mol⁻¹), C_air denotes the concentration of N₂O in the ambient air (µmol mol⁻¹), M represents the molecular weight of N in N₂O (28 µg N µmol⁻¹), R represents the universal gas constant (8.314 m³ Pa K⁻¹ mol⁻¹), T represents the air temperature in the forest (K), t represents the duration of each gas sampling (45 min), and A represents the surface area of soil covered by the chamber (0.24 m²).

2.6 Calibration of the Δ¹⁷O values of N₂O

To determine the Δ¹⁷O values of N₂O in the samples on the VSMOW scale, we prepared two standards (STD1 and STD2) containing N₂O. The Δ¹⁷O values of N₂O in the standards were calibrated to the VSMOW scale using the conventional method reported in Thiemens and Trogler (1991), where N₂O was quantitatively converted to O₂ using BrF₅ and a Ni catalytic container. The details are presented below.

A calibrated quantity of N₂O (50–170 µmol) was subsampled and transferred into a nickel tube (approximately 60 cm³) under liquid N₂ temperature. The coexisting components of N₂O, such as helium in the case of STD2, were evacuated from the nickel tube after N₂O was trapped in the nickel tube under liquid N₂ temperature. The nickel tube was then heated at 725 °C for 2.5 h to convert N₂O to NiO and N₂. After evacuating N₂ from the nickel tube, a 10-fold quantity of BrF₅ was introduced into the nickel tube and heated at 725 °C for 12 h to convert NiO to O₂ and NiF₂. After the purification of O₂, both δ¹⁸O and Δ¹⁷O of O₂ were determined on the VSMOW scale using IRMS, with the quantity of O₂ evolved from N₂O. Details on the procedures of O₂ purification and the measurement of O₂ using IRMS on the VSMOW scale have been described in previous studies (Sambuichi et al., 2021, 2023). STD1 is pure N₂O gas prepared from N₂O in a gas cylinder (more than 99.9 %; Koike Medical Ltd., Japan). The yield ratios of O₂ and Δ¹⁷O of STD1 were 103 ± 7 % and −0.22 ± 0.07 ‰, respectively (Fig. S3). The N₂O in STD2 is a mixture of helium and N₂O (N₂O $/$ He ≈ 1.5) produced from NO${}_{\mathrm{2}}^{-}$ that had been under oxygen isotope exchange equilibrium with H₂O with a Δ¹⁷O value of +1.2 ‰ originally, under a pH of 1.2. NO${}_{\mathrm{2}}^{-}$ was then converted to N₂O through a reaction with hydrazoic acid (N₃H), as described by Tsunogai et al. (2008). The reaction product (N₂O) was purged from the vial using pure helium (more than 99.9 %). After the removal of H₂O by passing a trap under the temperature of dry ice + ethanol, N₂O was captured in a trap at the temperature of liquid O₂ and then transported into a 1 L stainless steel canister together with helium. The yield of O₂ and Δ¹⁷O of STD2 were 97 ± 5 % and +1.13 ± 0.02 ‰, respectively (Fig. S3). To calibrate the Δ¹⁷O values of the samples measured using CF-IRMS, approximately 1 mL of each STD was subsampled into a 200-mL pre-evacuated glass bottle and diluted using pure helium to 1 atm. The Δ¹⁷O values of N₂O in the diluted standards were then determined using CF-IRMS like the procedure used on the samples before the sample measurements by introducing 30–60 nmol of N₂O. This allowed us to calibrate the Δ¹⁷O values of the samples to the VSMOW scale (Fig. S4).

2.7 Isotopic composition of NO${}_{\mathrm{2}}^{-}$

To determine the δ¹⁸O and Δ¹⁷O values of soil NO${}_{\mathrm{2}}^{-}$ that had been extracted in the KCl solution, the NO${}_{\mathrm{2}}^{-}$ in the KCl solution was chemically converted to N₂O using the method originally developed to determine the δ¹⁸O of NO${}_{\mathrm{2}}^{-}$ (McIlvin and Altabet, 2005), with several modifications for Δ¹⁷O (Xu et al., 2021), as explained below. Approximately 40 mL of each solution was pipetted into a glass vial (66.7 mL) and sealed with a butyl rubber septum cap. After purging the solution using high-purity helium for 45 min, 1.8 mL of an azide-acetic acid buffer (0.1 mol L⁻¹ NaN₃ in 1 vol % acetic acid), which had been purged using pure helium as well, was added to the solution to convert NO${}_{\mathrm{2}}^{-}$ to N₂O:

$$\begin{array}{}\text{(R1)}& {\mathrm{HNO}}_{\mathrm{2}}+{\mathrm{HN}}_{\mathrm{3}}\to {\mathrm{N}}_{\mathrm{2}}\mathrm{O}+{\mathrm{H}}_{\mathrm{2}}\mathrm{O}+{\mathrm{N}}_{\mathrm{2}}.\end{array}$$

After the vials were shaken for 1 h at a rate of 2 cycles s⁻¹, 0.9 mL of 6 M NaOH was added to each vial and shaken for 15 min.

The δ¹⁸O and Δ¹⁷O of N₂O converted from NO${}_{\mathrm{2}}^{-}$ in each vial were determined using the CF-IRMS system. We repeated the analyses for each solution sample at least three times to obtain better precision for Δ¹⁷O.

The δ¹⁸O values of NO${}_{\mathrm{2}}^{-}$ were calibrated to the VSMOW scale using three in-house nitrite standards (STD10, STD11, and STD12), the δ¹⁸O values of which had been determined using a thermal conversion elemental analyzer IRMS system, where oxygen atoms in each nitrite $/$ nitrate had been converted into CO using a glassy carbon tube at 1400 °C (Xu et al., 2021) and calibrated to the VSMOW scale using the international nitrate standards USGS34 (δ¹⁸O = −27.9 ‰) and IAEA-NO-3 (δ¹⁸O = +25.6 ‰) as the primary standards. Isotope fractionations during chemical conversion into N₂O were corrected by measuring the nitrite standards in the same way as samples were measured using the CF-IRMS system. In addition, the extent of oxygen isotope exchange between NO${}_{\mathrm{2}}^{-}$ and H₂O during the conversion was quantified using the relation between δ¹⁸O of the nitrite standards and that of N₂O (Xu et al., 2021). The Δ¹⁷O values of NO${}_{\mathrm{2}}^{-}$ were calibrated to the VSMOW scale by comparing N₂O derived from NO${}_{\mathrm{2}}^{-}$ with N₂O standards (STD1 and STD2) while assuming that the changes in Δ¹⁷O were negligible during the conversion from NO${}_{\mathrm{2}}^{-}$ into N₂O, except for the oxygen isotope exchange reaction between NO${}_{\mathrm{2}}^{-}$ and H₂O during the conversion to N₂O. The progress of oxygen isotope exchange between NO${}_{\mathrm{2}}^{-}$ and H₂O was calibrated from the Δ¹⁷O values of NO${}_{\mathrm{2}}^{-}$ using the exchange rate estimated by calculating δ¹⁸O values while assuming that the Δ¹⁷O value of H₂O was 0 ‰.

While the KCl solutions were widely used for the extraction of soil NO${}_{\mathrm{2}}^{-}$ (e.g., Lewicka-Szczebak et al., 2021; Shen et al., 2003), Homyak et al. (2015) raised the concerns that the recovery of soil NO${}_{\mathrm{2}}^{-}$ could be low when using KCl solutions compared to deionized water. Therefore, we conducted a comparative experiment to evaluate this potential issue and concluded that the use of KCl solution introduced negligible bias in terms of soil NO${}_{\mathrm{2}}^{-}$ recovery or Δ¹⁷O measurements compared to deionized water extraction for the soil type and experimental conditions in this study. The details are described in the Supplement (Text S2).

3 Results

3.1 Flux and isotopic compositions of N₂O emitted from forested soil

Almost all of the concentrations of N₂O ([N₂O]) in the samples collected in aluminum bags were higher than that of N₂O in ambient air (Figs. 3a and S5), implying that N₂O in the aluminum bags was a mixture of N₂O in ambient air and N₂O emitted from the forested soil. To determine the isotopic compositions (δ¹⁵N, δ¹⁸O, and Δ¹⁷O) of N₂O emitted from the soil, N₂O derived from ambient air was excluded using the linear correlation between $\mathrm{1}/$ [N₂O] and the isotopic compositions (δ¹⁵N, δ¹⁸O, and Δ¹⁷O) during mixing (Figs. 3b, c, d, and S5), also was known as Keeling plot approach (Keeling, 1958; Tsunogai et al., 1998, 2003). This method assumes that the concentrations of N₂O (N₂O $/$ (N₂O + N₂)) in the gases emitted from the soil were more than 3 %, allowing $\mathrm{1}/$ [N₂O] to be approximated to be 0 (Text S3). The uncertainties associated with the isotopic compositions of N₂O emitted from soil (i.e., the intercept) were estimated by applying the York method (Tsunogai et al., 2011; York et al., 2004) to the obtained relationship between $\mathrm{1}/$ [N₂O] as the independent variable and the isotopic compositions as the dependent variable in which uncertainties of both independent and dependent variables for individual data are considered.

Figure 3 An example of changes in the concentration of N₂O ([N₂O]) in gas samples during the observation on 8 September 2022, plotted as a function of the elapsed time since the deployment of the flow chamber on the forested soil (a), and the δ¹⁵N (b), δ¹⁸O (c), and Δ¹⁷O (d) values of N₂O plotted as a function of the reciprocal of [N₂O] ($\mathrm{1}/$ [N₂O]) during the observation. Each solid line is the least squares fitting of the samples, while each dotted line is the 2σ confidence interval of the fitting line. Error bars smaller than the sizes of the symbols are not shown.

The flux of N₂O emitted from the forested soil determined on fine days varied from −0.2 to 9.8 µg N m⁻² h⁻¹, with an average of 3.8 ± 3.1 µg N m⁻² h⁻¹ (1 SD; n=12). In addition, the emission flux during the warm seasons (from April to October; 5.1 ± 2.8 µg N m⁻² h⁻¹) was significantly higher than that during the cold seasons (from November to March; 1.0 ± 1.1 µg N m⁻² h⁻¹) (Fig. 4a; Table S1), implying that the emission flux of N₂O on fine days exhibited clear seasonal variation. Furthermore, the average emission flux of N₂O determined on rainy days (38.8 ± 28.0 µg N m⁻² h⁻¹; n=6) was significantly higher than that determined on fine days (3.8 ± 3.1 µg N m⁻² h⁻¹) (Fig. 4a and b). These patterns of N₂O emissions were in accordance with those of agricultural and forested soils reported in previous studies (Anthony et al., 2023; Chen et al., 2012; Choudhary et al., 2002; Yan et al., 2008).

Figure 4 Temporal variations in the flux (a), Δ¹⁷O (c), δ¹⁸O (e), and δ¹⁵N (g) values of N₂O emitted from the forested soil, and the δ¹⁸O and Δ¹⁷O values of soil NO${}_{\mathrm{2}}^{-}$ (green triangles), O₂ (orange lines), and soil H₂O (blue area or line). Sampling performed on fine and rainy days is indicated by the open (white) and solid (yellow) circles, respectively, with the box plots of the emission flux (b), Δ¹⁷O (d), δ¹⁸O (f), and δ¹⁵N (h) of N₂O on fine and rainy days. The black lines of the box plots indicate the median values. The lower and upper boundaries of the box plots indicate the lower (25 %) and upper (75 %) quartiles of data for each component, respectively. The whiskers of the box plots denote the entire range of values for each component. Error bars smaller than the sizes of the symbols are not shown.

Because of the small emission flux of N₂O during the cold seasons, the linear relationships between the isotopic compositions and $\mathrm{1}/$ [N₂O] became insignificant in some of the observations performed during the cold seasons (Fig. S5, from November 2022 to January 2023). Thus, the uncertainties associated with the isotopic compositions estimated for N₂O emitted from the soil became enormous. Consequently, the isotopic compositions of N₂O emitted from the soil are not shown under the following conditions: (1) the [N₂O] in the gas sample collected at the end of each deployment of the chamber did not exceed 130 % of that of ambient air, and (2) the linear correlation between $\mathrm{1}/$ [N₂O] and the isotopic compositions was statistically insignificant (P>0.05). Similar criteria have been adopted in previous studies (Kaushal et al., 2022; Opdyke et al., 2009).

The N₂O emitted from the forested soil on fine days exhibited δ¹⁵N, δ¹⁸O, and Δ¹⁷O values ranging from −27.5 ‰ to −17.9 ‰, from +26.1 ‰ to +37.6 ‰, and from −0.40 ‰ to −0.11 ‰, respectively, with average values and standard deviations (1 SD) of −22.5 ± 2.8 ‰, +30.9 ± 4.3 ‰, and −0.30 ± 0.09 ‰, respectively (Fig. 4g, e, and c). On the other hand, N₂O emitted from the forested soil on rainy days exhibited δ¹⁵N, δ¹⁸O, and Δ¹⁷O values ranging from −26.6 ‰ to −13.8 ‰, from +18.4 ‰ to +36.2 ‰, and from −0.06 ‰ to +0.26 ‰, respectively, with average values and standard deviations (1SD) of −20.4 ± 5.0 ‰, +27.9 ± 6.4 ‰, and +0.12 ± 0.13 ‰, respectively (Fig. 4g, e, and c).

The NO${}_{\mathrm{2}}^{-}$ exhibited δ¹⁸O and Δ¹⁷O values ranging from +2.4 ‰ to +12.0 ‰ and from +0.04 ‰ to +0.50 ‰, respectively, with average values of +6.0 ± 2.0 ‰ and +0.23 ± 0.12 ‰, respectively (n=18, Fig. 4e and c). These δ¹⁸O values of NO${}_{\mathrm{2}}^{-}$ coincided well with those determined in a previous study (Lewicka-Szczebak et al., 2021).

3.2 Flux and isotopic compositions of N₂O emitted from artificially fertilized soils

The fluxes of N₂O emitted from the NF (no fertilizer), U (fertilized with urea, CO(NH₂)₂), and CS (fertilized with Chile saltpeter, KNO₃) plots were 5.2, 70.6, and 112.3 µg N m⁻² h⁻¹, respectively, 2 d after fertilization and 4.2, 56.7, and 39.4 µg N m⁻² h⁻¹, respectively, 6 d after fertilization (Table S1). The fluxes of N₂O emitted from the U and CS plots were significantly higher than that from the NF plot, indicating that the flux of N₂O emitted from the soil increased significantly because of fertilization, supporting the results reported in previous studies (Kaushal et al., 2022; McKenney et al., 1978; Toyoda et al., 2011, 2017).

The δ¹⁵N, δ¹⁸O, and Δ¹⁷O values of N₂O emitted from the NF plot 2 d after fertilization were −17.1 ± 6.4 ‰, +36.1 ± 6.7 ‰, and −0.37 ± 0.20 ‰, respectively, whereas those emitted from the NF plot 6 d after fertilization were −12.2 ± 3.2 ‰, +40.0 ± 13.3 ‰, and −0.32 ± 0.23 ‰, respectively. The δ¹⁵N, δ¹⁸O, and Δ¹⁷O values of N₂O emitted from the U plot 2 d after fertilization were −39.3 ± 0.7 ‰, +34.4 ± 0.4 ‰, and −0.14 ± 0.06 ‰, respectively, whereas those emitted from the U plot 6 d after fertilization were −33.3 ± 0.5 ‰, +25.7 ± 0.6 ‰, and −0.16 ± 0.05 ‰, respectively. The δ¹⁵N, δ¹⁸O, and Δ¹⁷O values of N₂O emitted from the CS plot 2 d after fertilization were −19.3 ± 0.6 ‰, +54.1 ± 0.8 ‰, and +8.22 ± 0.03 ‰, respectively, whereas those emitted from the CS plot 6 d after fertilization were −11.3 ± 0.7 ‰, +58.7 ± 1.2 ‰, and +7.36 ± 0.17 ‰, respectively (Fig. 5). These flux, δ¹⁵N, and δ¹⁸O of N₂O emitted from the NF, U, and CS plots correspond well with the results of many previous studies on forested and artificial soils (or agricultural soils) (Kaushal et al., 2022; Kim and Craig, 1993; Snider et al., 2009; Toyoda et al., 2017; Wrage et al., 2004).

Figure 5 Changes in [N₂O] of gas samples collected from the plots of NF (gray), U (purple), and CS (blue) 2 d after fertilization (a) and 6 d after fertilization (e) and plotted as a function of the elapsed time since the deployment of the flow chamber; the δ¹⁵N (b, f), δ¹⁸O (c, g), and Δ¹⁷O (d, h) values of N₂O plotted as a function of the reciprocal of [N₂O] ($\mathrm{1}/$ [N₂O]). Error bars smaller than the sizes of the symbols are not shown.

The δ¹⁸O and Δ¹⁷O values of NO${}_{\mathrm{2}}^{-}$ in the NF plot 2 d after fertilization were +2.7 ‰ and +0.42 ‰, respectively, whereas those in the NF plot 6 d after fertilization were +1.3 ‰ and +0.35 ‰, respectively. The δ¹⁸O and Δ¹⁷O values of NO${}_{\mathrm{2}}^{-}$ in the U plot 2 d after fertilization were +7.6 ‰ and +0.31 ‰, respectively, whereas those in the U plot 6 d after fertilization were +5.4 ‰ and +0.17 ‰, respectively. The δ¹⁸O and Δ¹⁷O values of NO${}_{\mathrm{2}}^{-}$ in the CS plot 2 d after fertilization were +29.0 ‰ and +8.26 ‰, respectively, whereas those in the CS plot 6 d after fertilization were +45.2 ‰ and +12.32 ‰, respectively (Fig. 6).

Figure 6 The δ¹⁸O (a) and Δ¹⁷O (b) values of N₂O (yellow circles) and NO${}_{\mathrm{2}}^{-}$ (green triangles) in NF, U, and CS plots determined 2 and 6 d after fertilization, and the δ¹⁸O and Δ¹⁷O values of O₂ (orange lines) and soil H₂O (blue area or line). Error bars smaller than the sizes of the symbols are not shown.

4 Discussion

4.1 Identification of N₂O production pathways in forested soil using Δ¹⁷O signature

Because O atoms in N₂O emitted from soil can be derived from those in NO${}_{\mathrm{2}}^{-}$, O₂, or H₂O in soil (Fig. 1), we can constrain the pathways of N₂O production by comparing the δ¹⁸O and Δ¹⁷O values of N₂O with those of NO${}_{\mathrm{2}}^{-}$, O₂, and H₂O in soil. Consequently, we compiled the δ¹⁸O and Δ¹⁷O values of atmospheric O₂ (+23.88 ‰ for δ¹⁸O and −0.44 ‰ for Δ¹⁷O, Sharp and Wostbrock, 2021) and rainwater (ranging from −2 ‰ to −10 ‰ for δ¹⁸O in Japan, Nakagawa et al., 2018; Takahashi, 1998; Uechi and Uemura, 2019; Zou et al., 2015; +0.03 ‰ for Δ¹⁷O in Japan, Uechi and Uemura, 2019), as shown in Figs. 4 and 6, along with those of soil NO${}_{\mathrm{2}}^{-}$ measured in this study.

The Δ¹⁷O of N₂O produced in the soil may differ from that of the source of O atoms (O₂, NO${}_{\mathrm{2}}^{-}$, H₂O) because of oxygen isotope fractionation during nitrification and denitrification, as the value of β in Eq. (2) may vary depending on the reactions. Thus, prior to using Δ¹⁷O values to identify the pathways of N₂O production in soils, we quantified the possible variations in the Δ¹⁷O values of N₂O during each reaction. The details are presented below.

The fractionation of oxygen isotopes during the transformation of the O atoms in O₂ to those in N₂O through nitrification accompanies significant variations in the value of δ¹⁸O from O₂ to N₂O (Figs. 4e and 6a). In addition to δ¹⁸O, the Δ¹⁷O value of N₂O produced through nitrification could be somewhat different from that of O₂, even if all O atoms in N₂O were derived from O₂, due to the possible differences in β from 0.528 during the reaction (Fig. 7). The average variation in δ¹⁸O from O₂ to N₂O due to nitrification (Δδ¹⁸O (N₂O−O₂)) was estimated to be 9 ‰ on average (Figs. 4e and 6a) based on the difference in δ¹⁸O values between N₂O emitted from the soil in this study (+33 ± 10 ‰; n=19) and O₂ in the literature (Sharp and Wostbrock, 2021). Conversely, we can expect values from 0.525 to 0.5305 for β in the various reactions (Cao and Liu, 2011; Matsuhisa et al., 1978; Pack and Herwartz, 2014; Sharp and Wostbrock, 2021), where the β of nitrification may be included. Thus, we quantified the possible range of variations in the Δ¹⁷O value of N₂O from that of O₂ to be less than 0.027 ‰ (Fig. 7), based on the observed Δδ¹⁸O(N₂O−O₂) and the possible variation range of β.

Figure 7 Schematic showing the possible variations in the Δ¹⁷O value of N₂O from that of the source of O atoms (O₂ and NO${}_{\mathrm{2}}^{-}$) during transformations, including nitrification (orange circles), denitrification (green circles), and reduction (yellow circles), due to variations in isotope fractionation and β from 0.525 to 0.5305.

Similarly, the fractionation of oxygen isotopes during the transformation of O atoms in NO${}_{\mathrm{2}}^{-}$ to those in N₂O through denitrification accompanies significant variations in δ¹⁸O from NO${}_{\mathrm{2}}^{-}$ to N₂O as well. The Δ¹⁷O value of N₂O produced through NO${}_{\mathrm{2}}^{-}$ reduction could be somewhat different from that of NO${}_{\mathrm{2}}^{-}$, even if all O atoms in N₂O were derived from NO${}_{\mathrm{2}}^{-}$, due to the possible differences in β from 0.528 during the reaction (Fig. 7). The average variation in δ¹⁸O from NO${}_{\mathrm{2}}^{-}$ to N₂O due to fractionation (Δδ¹⁸O (N₂O−NO${}_{\mathrm{2}}^{-}$)) was estimated to be 25 ‰ on average (Figs. 4e and 6a) based on the difference in δ¹⁸O values between N₂O (+33 ± 10 ‰; n=19) and NO${}_{\mathrm{2}}^{-}$ in this study (+8 ± 9 ‰; n=24). Thus, we quantified the possible range of variations in the Δ¹⁷O value of N₂O from that of NO${}_{\mathrm{2}}^{-}$ to be less than 0.075 ‰ (Fig. 7), based on the observed Δδ¹⁸O (N₂O–NO${}_{\mathrm{2}}^{-}$) and the possible variation range of β, from 0.525 to 0.5305.

Similarly, kinetic fractionation during the reduction of N₂O to N₂ accompanies variation in δ¹⁸O from original N₂O to residual N₂O as well. The Δ¹⁷O value of residual N₂O could somewhat differ from that of the original N₂O. Previous studies have reported the range of variations in δ¹⁸O from original N₂O to residual N₂O due to kinetic fractionation to be less than 10 ‰ on average through incubation experiments (Lewicka-Szczebak et al., 2014, 2015). Thus, we quantified the possible range of variations in the Δ¹⁷O value of residual N₂O from that of original N₂O to be less than 0.03 ‰ (Fig. 7), based on Δδ¹⁸O (less than 10 ‰) and the variation range of β, from 0.525 to 0.5305.

These possible variations in Δ¹⁷O (less than 0.075 ‰) were much less than the difference in Δ¹⁷O values between O₂ and NO${}_{\mathrm{2}}^{-}$ in the forested soil (0.7 ‰ on average; Fig. 4c). In addition, the possible variation ranges in Δ¹⁷O become much smaller if the differences in β from 0.528 were smaller than those used in the calculations (from 0.525 to 0.5305). Thus, we concluded that the possible variations in the Δ¹⁷O value of N₂O from that of the source molecules of O atoms (O₂, H₂O, and NO${}_{\mathrm{2}}^{-}$) during the transformations, including nitrification, denitrification, and reduction, were negligible.

While the Δ¹⁷O values of soil O₂ and H₂O used in this study were referred from atmospheric O₂ and rainwater, respectively, the processes in soil, including diffusion and respiration of O₂ and evaporation and infiltration of rainwater, may cause significant isotopic fractionations of δ¹⁸O, which could consequently alter the Δ¹⁷O values of atmospheric O₂ and rainwater. Thus, prior to using Δ¹⁷O values to identify the pathways of N₂O production in soils, we evaluated the possible variations in the Δ¹⁷O values of O₂ and H₂O in soil compared to those of atmospheric O₂ and rainwater. The details are presented below.

For soil O₂, Aggarwal and Dillon (1998) measured the δ¹⁸O values in soil gas at a depth of 3–4 m at a site near Lincoln, Nebraska, USA ranged from +23.3 ‰ to +27.2 ‰, showing the values were comparable with that of atmospheric O₂ (+23.5 ‰ after adjustment in Aggarwal and Dillon, 1998). This confirms that the isotopic fractionations of soil O₂ induced from soil respiration and diffusion processes were not significant. Because the maximum variation in δ¹⁸O from atmospheric O₂ to soil O₂ was less than 3.7 ‰ (27.2 ‰–23.5 ‰), using the method presented in Fig. 7, we quantified the possible variations in the Δ¹⁷O value of soil O₂ from that of atmospheric O₂ to be less than 0.01 ‰. Thus, we ignored the negligible variations in this study.

Similarly, for soil H₂O, Lyu (2021) observed that δ¹⁸O values in soil H₂O at the depths of 0–5, 15–20, and 40–45 cm in a subtropical forest plantation ranged from −4 ‰ to −10 ‰, which fully overlapped with local rainwater (−1 ‰ to −16 ‰), indicating insignificant isotopic fractionations of soil H₂O during hydrological processes such as infiltration and evaporation. In addition, Aron et al. (2021) compiled Δ¹⁷O values of terrestrial H₂O including rainwater, surface, and subsurface water in earth, ranging from +0.06 ‰ to −0.06 ‰ which did not show significant differences with each other, also indicating that the possible variations of Δ¹⁷O values of soil H₂O compared to that of rainwater should be negligible. Finally, we added the variations of Δ¹⁷O values (+0.06 ‰ to −0.06 ‰) of terrestrial H₂O reported in Aron et al. (2021) to Figs. 4 and 6 as the uncertainties of Δ¹⁷O values of soil H₂O.

In the forested soil, N₂O exhibited Δ¹⁷O values (−0.30 ± 0.09 ‰ on average) that were close to that of O₂ (−0.44 ‰) but deviated from those of soil NO${}_{\mathrm{2}}^{-}$ on fine days (+0.24 ± 0.14 ‰; Fig. 4c and d), implying that nitrification was the main pathway to produce N₂O in the soil on fine days. Conversely, N₂O emitted from the soil on rainy days exhibited Δ¹⁷O values (+0.12 ± 0.13 ‰) that were close to those of soil NO${}_{\mathrm{2}}^{-}$ (+0.22 ± 0.09 ‰) and soil H₂O (+0.03 ‰) but deviated from that of O₂ (Fig. 4c and d), implying that (1) the main pathway to produce N₂O changed from nitrification on fine days to denitrification on rainy days and/or (2) the possible contribution of O atoms derived from soil H₂O became more active during the production of N₂O in the soil on rainy days.

4.2 Changes in the Δ¹⁷O of N₂O emitted from artificially fertilized soils

To quantitatively constrain the possible contributions of O atoms derived from soil H₂O during the production of N₂O in the soil, we observed changes in the isotopic compositions of N₂O from the same soil in response to artificial fertilization. In the plot fertilized with CS, the Δ¹⁷O value of N₂O emitted from the soil (+7.79 ± 0.61 ‰ on the average of 2 and 6 d after the fertilization) became significantly closer to that of soil NO${}_{\mathrm{2}}^{-}$ (+10.3 ± 2.9 ‰) compared with that of atmospheric O₂ (−0.44 ‰; Fig. 6b). This suggested that denitrification became the main pathway of N₂O production, probably because of fertilization, which resulted in a significantly higher concentration of NO${}_{\mathrm{3}}^{-}$ (278.4 ± 43.2 mg N kg⁻¹; Table S1) than that of NH${}_{\mathrm{4}}^{+}$ (15.8 ± 4.1 mg N kg⁻¹) in the CS plot. In addition, N₂O emitted from the CS plot exhibited Δ¹⁷O values that were significantly different from those of soil H₂O (+0.03 ‰; Fig. 6b), implying that the contribution of O atoms derived from soil H₂O was minor during the reduction of NO${}_{\mathrm{2}}^{-}$ to produce N₂O. If all the O atoms with low Δ¹⁷O values in N₂O were derived from soil H₂O (+0.03 ‰) in the CS plot, the contribution of O atoms derived from soil H₂O was calculated to be 24 % ((10.30 ‰–7.79 ‰) $/$ (10.30 ‰–0.03 ‰)), based on the isotopic mass balance. If the O₂ also contributed to the N₂O production in the CS plot, the contribution of O atoms derived from soil H₂O should be further reduced. As a result, we determined that the maximum possible contribution of O atoms derived from soil H₂O during the reduction of NO${}_{\mathrm{2}}^{-}$ to N₂O was 24 %.

On the other hand, in the plot fertilized with urea (U plot), the Δ¹⁷O value of N₂O (−0.15 ± 0.01 ‰) was close to that of O₂ (−0.44 ‰) compared with that of soil NO${}_{\mathrm{2}}^{-}$ (+0.24 ± 0.10 ‰). This suggested that nitrification was the main pathway of N₂O production (Fig. 6b), probably due to the enhancement of NH${}_{\mathrm{4}}^{+}$ concentration (423.1 ± 18.2 mg N kg⁻¹; Table S1) compared with that of NO${}_{\mathrm{3}}^{-}$ (13.0 ± 10.7 mg N kg⁻¹) in the U plot. In addition, N₂O emitted from the U plot exhibited Δ¹⁷O values that were significantly different from that of soil H₂O (+0.03 ‰; Fig. 6b), implying that the contribution of O atoms derived from soil H₂O was also minor during the oxidation of NH${}_{\mathrm{4}}^{+}$ to produce N₂O. Consequently, the contribution of O atoms derived from soil H₂O was minor in the soil during N₂O production, irrespective of the pathways of N₂O production being either nitrification or denitrification. In addition, it is difficult to explain the observed increases in the emission flux of N₂O from the soil on rainy days based only on the active contribution of O atoms derived from soil H₂O. Consequently, we concluded that N₂O production through denitrification became active in the soil on rainy days, which resulted in increased N₂O emission and higher Δ¹⁷O values.

4.3 Verification of active N₂O emission by denitrification on rainy days

The forested soil exhibited significantly lower WFPS on fine days (66.1 ± 6.2 %; Table S1) than on rainy days (95.6 ± 19.1 %), implying that the O₂ concentration in the soil was higher on fine days than on rainy days. Using the isotope tracer enriched in ¹⁵N (¹⁵NO${}_{\mathrm{3}}^{-}$ or ¹⁵NH${}_{\mathrm{4}}^{+}$), Mathieu et al. (2006) estimated the relative importance of nitrification and denitrification to produce N₂O in soil. They found that nitrification produced the majority of N₂O under low WFPS conditions (75 %), whereas denitrification accounted for more than 85 % of N₂O produced under high WFPS conditions (150 %). Similarly, using natural stable isotopes (SP), Ibraim et al. (2019) reported the primary pathway for N₂O production in a grassland shifted from nitrification to denitrification as increasing WFPS, when WFPS was below 90 %. Thus, we conclude that the lower WFPS in the soil caused oxic conditions on fine days, resulting in nitrification as the primary pathway for N₂O production in the soil. Conversely, the higher WFPS caused redox conditions in the soil on rainy days, resulting in active N₂O production through denitrification in the soil (Fig. 4a and b).

During continuous monitoring of the emission flux of N₂O from an agricultural soil for four years, Anthony et al. (2023) found short-term increases in the emission flux during or immediately after rainfall or irrigation. They referred to this high emission flux as “hot moments” and defined it as exceeding four standard deviations of that of normal periods. They also found significant correlations between the emission flux and WFPS, leading to the conclusion that variations in the concentrations of O₂ in surface soils were responsible for the hot moments of N₂O emissions. Although the hot moments accounted for 1 % of all measurements, they contributed up to 57 % of the annual emissions, indicating their significance as a source of atmospheric emissions. In this study, the emission flux of N₂O on rainy days also exceeded four standard deviations of that on fine days (Fig. 4a and b). The Δ¹⁷O evidence of N₂O found in this study further verified that denitrification was mainly responsible for the enhancement of N₂O production during the hot moments.

4.4 Changes in the pathway of N₂O production due to fertilization with urea

During our observation on the plot fertilized with urea (U plot), N₂O emitted from the plot exhibited Δ¹⁷O values (−0.15 ± 0.01 ‰ on average; Fig. 6b) that were significantly higher than those of the plot without fertilization (NF plot; −0.35 ± 0.04 ‰ on average). Although an increase in the contribution of O atoms derived from soil H₂O could be responsible for the Δ¹⁷O values in addition to an increase in N₂O production through nitrification, we concluded that an increase in N₂O production through NO${}_{\mathrm{2}}^{-}$ reduction was responsible for the Δ¹⁷O values (−0.15 ± 0.01 ‰ on average) of N₂O produced in the plot in response to fertilization of urea $/$ NH${}_{\mathrm{4}}^{+}$ for the following reasons.

Avrahami et al. (2002) reported that fertilization with urea $/$ NH${}_{\mathrm{4}}^{+}$ in soil promoted the oxidation of NH${}_{\mathrm{4}}^{+}$ and thus provided electron acceptors for denitrification. That is, the enrichment of nitrate through nitrification also promotes denitrification. Based on the stable isotopes of N₂O (δ¹⁵N, δ¹⁸O, and SP), along with in vitro acetylene blockage experiments on agricultural soils fertilized with NH${}_{\mathrm{4}}^{+}$, Zhang et al. (2016) reported that while 50 %–70 % of N₂O was produced through nitrification, nitrifier denitrification (NH${}_{\mathrm{4}}^{+}\to$ NO${}_{\mathrm{2}}^{-}\to$ N₂O) and/or heterotrophic denitrification (NH${}_{\mathrm{4}}^{+}\to$ NO${}_{\mathrm{3}}^{-}\to$ NO${}_{\mathrm{2}}^{-}\to$ N₂O) accounted for 30 %–50 % of N₂O production. Similar results have also been reported in previous studies. Although N₂O production through nitrification was simulated by fertilization with urea $/$ NH${}_{\mathrm{4}}^{+}$ in various soils, denitrification also accounted for a significant portion of N₂O production (Kaushal et al., 2022; Khalil et al., 2004; Zhu et al., 2013). In addition to nitrifier and heterotrophic denitrification, N₂O produced through the anammox process (NH${}_{\mathrm{4}}^{+}+$ NO${}_{\mathrm{2}}^{-}\to$ N₂O, Okabe et al., 2011; Tang et al., 2011; Tsushima et al., 2007) can be responsible for the reduction of NO${}_{\mathrm{2}}^{-}$ as well. Zhu et al. (2011) found that the highest rate of anammox was comparable with that of denitrification in soils fertilized with NH${}_{\mathrm{4}}^{+}$ (6.2–178.8 mg N kg⁻¹). These previous experiments support our observation on the U plot that the addition of urea $/$ NH${}_{\mathrm{4}}^{+}$ stimulates N₂O production through nitrifier denitrification and/or heterotrophic denitrification, and/or anammox reaction in addition to nitrification. The increased NO${}_{\mathrm{3}}^{-}$ concentration in the U plot (13.0 ± 10.7 mg N kg⁻¹) compared with those in the NF plot (2.3 ± 0.5 mg N kg⁻¹) probably due to nitrification stimulated by the addition of NH${}_{\mathrm{4}}^{+}$ may be responsible for the active reduction of NO${}_{\mathrm{2}}^{-}$.

4.5 Stable Δ¹⁷O as a natural signature for identifying N₂O production pathways

Although the δ¹⁸O values of N₂O emitted from the soil were significantly higher than those of the sources of O atoms in N₂O (NO${}_{\mathrm{2}}^{-}$, O₂, and H₂O; Figs. 4e and 6a) due to the fractionations of oxygen isotopes during the production and/or reduction of N₂O, the Δ¹⁷O values of N₂O remained within the range of these sources. This indicates that Δ¹⁷O primarily reflects the pathways of N₂O production, providing information distinct from the δ¹⁸O signature because Δ¹⁷O is stable during the processes of biogeochemical isotope fractionation. Moreover, while N₂O emission from the forested soil did not show significant differences in δ¹⁵N and δ¹⁸O values between fine and rainy days due to the fractionations of nitrogen and oxygen isotopes (Fig. 4f and h), the significant difference in the Δ¹⁷O values of N₂O between fine and rainy days (Fig. 4d) highlights Δ¹⁷O to be a promising natural signature for identifying the pathways of N₂O production in soils.

In addition to natural soils, the stable Δ¹⁷O signature is expected to be useful for identifying the pathways of N₂O production in various ecosystems, such as agricultural soils and aquatic environments, where the isotopic fractionations of nitrogen and oxygen isotopes involving biogeochemical processes are significant as well. However, to identify the pathways of N₂O production quantitatively, the uncertainties, including the β values of each reaction during N₂O production and the contributions of O atoms derived from soil H₂O during N₂O production, should be quantified precisely in the future studies.

5 Conclusions

Temporal variations in Δ¹⁷O of N₂O emitted from forested soil were determined to identify the main pathway of N₂O production. Both Δ¹⁷O values and fluxes of N₂O were significantly higher on rainy days compared to fine days. In addition, the Δ¹⁷O values of N₂O emitted on rainy and fine days were close to those of soil NO${}_{\mathrm{2}}^{-}$ and O₂, respectively. Because NO${}_{\mathrm{2}}^{-}$ and O₂ were the source of O-atoms in N₂O production through denitrification and nitrification, respectively, we concluded that while nitrification dominated N₂O production on fine days, denitrification became active on rainy days, resulting in the N₂O flux increasing. In addition, the Δ¹⁷O of N₂O emitted from the same soil fertilized with either Chile saltpeter or urea exhibited values that were significantly different from those of soil H₂O, implying that the contributions of O atoms derived from soil H₂O during N₂O production were minor. Furthermore, while N₂O emitted from the forested soil did not show significant differences in δ¹⁵N and δ¹⁸O values between fine and rainy days, the significant difference in the Δ¹⁷O values of N₂O highlights Δ¹⁷O to be a promising natural signature for identifying the pathways of N₂O production in soils, because Δ¹⁷O is almost stable during isotope fractionation processes such as N₂O production and reduction.