Isotopomer ratios of N2O (bulk nitrogen and oxygen isotope ratios, δ15Nbulk and δ18O, and intramolecular 15N site preference, SP) are useful parameters that characterize sources of this greenhouse gas and also provide insight into production and consumption mechanisms. We measured isotopomer ratios of N2O emitted from typical Japanese agricultural soils (Fluvisols and Andisols) planted with rice, wheat, soybean, and vegetables, and treated with synthetic (urea or ammonium) and organic (poultry manure) fertilizers. The results were analyzed using a previously reported isotopomeric N2O signature produced by nitrifying/denitrifying bacteria and a characteristic relationship between δ15Nbulk and SP during N2O reduction by denitrifying bacteria. Relative contributions from nitrification (hydroxylamine oxidation) and denitrification (nitrite reduction) to gross N2O production deduced from the analysis depended on soil type and fertilizer. The contribution from nitrification was relatively high (40%–70%) in Andisols amended with synthetic ammonium fertilizer, while denitrification was dominant (50%–90%) in the same soils amended with poultry manure during the period when N2O production occurred in the surface layer. This information on production processes is in accordance with that obtained from flux/concentration analysis of N2O and soil inorganic nitrogen. However, isotopomer analysis further revealed that partial reduction of N2O was pronounced in high-bulk density, alluvial soil (Fluvisol) compared to low-bulk density, volcanic ash soil (Andisol), and that the observed difference in N2O flux between normal and pelleted manure could have resulted from a similar mechanism with different rates of gross production and gross consumption. The isotopomeric analysis is based on data from pure culture bacteria and would be improved by further studies on in situ biological processes in soils including those by fungi. When flux/concentration-weighted average isotopomer ratios of N2O from various fertilized soils were examined, linear correlations were found between δ15Nbulk and δ18O, and between SP and δ15Nbulk. These relationships would be useful to parameterize isotopomer ratios of soil-emitted N2O for the modeling of the global N2O isotopomer budget. The results obtained in this study and those from previous firn/ice core studies confirm that the principal source of anthropogenic N2O is fertilized soils.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Nitrous oxide (N2O) is one of the increasing atmospheric trace gases. Although its global average tropospheric concentration is lower than carbon dioxide (CO2) by 3 orders of magnitude (321 ppb in 2009, Carbon Dioxide Information and Analysis Center, http://cdiac.esd.ornl.gov/pns/current_ghg.html), it has about 300 times larger global warming potential than CO2 over a 100 year time scale [Intergovernmental Panel on Climate Change (IPCC), 2007]. In the stratosphere, N2O is decomposed by ultraviolet light and reactive oxygen to form nitric oxide, which destroys stratospheric ozone [Crutzen, 1970]. Recent estimates show that N2O is the single most important ozone-depleting gas since the reduction in CFC emissions [Ravishankara et al., 2009].
 Agriculture accounts for about 16% of global N2O sources and agricultural soils contribute 40%–70% of agricultural N2O emissions [Davidson, 2009; IPCC, 2007], which is the largest among anthropogenic sources. However, there is still large uncertainty in the estimated annual flux of N2O from agriculture because N2O is produced and consumed by several bacterial processes which are sensitive to soil physical/chemical conditions. Therefore its flux is spatially and temporally variable. The major N2O production processes are nitrification (equation (1)) and denitrification (equation (2)). The latter is also a N2O consumption process:
In addition, it is known that some nitrifying bacteria produce N2O by nitrite reduction similar to denitrifying bacteria, which is called as nitrifier-denitrification [Wrage et al., 2001]:
Nitrification occurs under oxic conditions and is generally an autotrophic process not dependent on availability of organic substances, although there are also heterotrophic nitrifiers [Kuenen and Robertson, 1994]. In contrast, denitrification is active under reducing environments and heterotrophic processes, requiring organics. Therefore, it is important to know the relative contributions of these processes to N2O production to mitigate its emission and to estimate N2O flux from soils using regional or global nitrogen cycling models [Sorai et al., 2007]. Although the flux ratio of NO and N2O has been found to correlate with soil moisture and used to assess the dominance of nitrification or denitrification [Davidson, 1993], a more robust indicator is needed for quantitative estimation of the share of nitrification and denitrification.
 Nitrogen and oxygen stable isotope ratios are useful tools to resolve N2O formation/decomposition processes in its global budget because they are determined by isotope ratios of precursor materials. The isotope enrichment factor, ɛ, which is unique to each reaction pathway [Kim and Craig, 1993; Rahn and Wahlen, 2000; Yoshida and Matsuo, 1983] is defined as follows:
where k(iX) and k(jX) are reaction rate constants for heavy molecules containing isotope iX, and the lightest molecules containing isotope jX, respectively.
 In addition to these “bulk” isotope ratios, the use of N2O isotopomer ratios (relative abundance of 14N15N16O and 15N14N16O to that of 14N14N16O) has been proposed to obtain more detailed information on N2O production and consumption processes and sources [Yoshida and Toyoda, 2000]. In fact, studies on N2O formation in pure bacterial cultures showed that the 15N site preference (the difference in 15N/14N ratio between central and peripheral N atoms in the NNO molecule, hereinafter SP) can be used to differentiate N2O produced by hydroxylamine oxidation (equation (1)) and nitrite reduction (equations (2) and (3)) [Sutka et al., 2003, 2004; Sutka et al., 2006]. Isotopomer ratios of N2O in the ocean [e.g., Yamagishi et al., 2007], industrial sources [e.g., Toyoda et al., 2008], rivers [Toyoda et al., 2009], and soils (see below) have been characterized, and a secular trend of atmospheric N2O isotopomer ratios has been reported [Röckmann and Levin, 2005].
 Isotope/isotopomer ratios of N2O emitted from fertilized soils have been investigated by several researchers. Kim and Craig  first reported that bulk N and O isotope ratios of N2O emitted from tropical fertilized soils are lower than those of tropospheric N2O. Pérez et al.  studied temporal variation in SP as well as N2O bulk isotope ratios in subtropical soils, and Yamulki et al.  characterized these in temperate grassland soils. Successive studies have been carried out on temperate arable soils in Germany [Well et al., 2005], the Netherlands [Van Groenigen et al., 2005] and Canada [Rock et al., 2007], with the latter two studies dealing with bulk isotope ratios only. Recent isotopomer analytic studies estimated that N2O emitted from agricultural fields in the northern United States [Opdyke et al., 2009] and in Venezuela [Park et al., 2011] is produced mainly by denitrification, although the authors of each study assumed different end-member values for nitrification and denitrification.
 In addition to field observations, laboratory experiments using soil cores have been conducted. Bol et al.  incubated nitrate-amended grassland soils under an O2/He atmosphere, controlled O2 concentration and measured N2O isotopomer ratios and the flux of N2O and N2. They observed an increase in isotopomer ratios accompanied by a decrease in the N2O/N2 flux ratio, which suggested the occurrence of denitrification. Pérez et al.  incubated nonfertilized tropical forest soils with and without acetylene and estimated ɛ(15N) and SP for nitrification and denitrification. Their reported SP values for nitrification (−16.8‰ ± 8.4‰) and denitrification (9.4‰ ± 8.1‰) are significantly lower than those for pure culture of nitrifiers, and the cause of the difference has been unclear (herein, reported values by Pérez et al.  are converted on to the Toyoda and Yoshida  scale, of which accuracy has been confirmed by other researchers [Griffith et al., 2009; Westley et al., 2007]). Well et al. [2006, 2008] and Well and Flessa  estimated ɛ(15N) and SP during nitrification and denitrification in fertilized arable soils by combining isotopomer analysis at natural abundance level and the tracer approach with 15N-labeled substrates. Their estimated ɛ(15N) and SP were generally within the range reported from pure bacterial cultures, although their definition for “nitrification,” as well as that of Pérez et al. , seems to include both hydroxylamine oxidation (equation (1)) and nitrifier-denitrification (equation (3)). Ostrom et al.  determined ɛ(15N) and ɛ(18O) during N2O reduction in cultivated soils and found that the ratio ɛ(15N)/ɛ(18O) was constant, although the magnitude of each ɛ was variable compared to those in pure cultures of denitrifying bacteria.
 These studies suggest that isotopomer ratios of N2O emitted from agricultural soils are suitable parameters to diagnose soil microbial production and consumption processes. At the same time, however, temporal and spatial variability in N2O isotopomer ratios make it difficult to deduce their representative values for agricultural N2O for a budget analysis based on box models [Kim and Craig, 1993; Yoshida and Toyoda, 2000]. To overcome this problem, two approaches are possible: (1) to obtain flux-weighted average values from various observations [Pérez et al., 2001], (2) to obtain a relationship between isotopomer ratios and other available parameters. The latter approach would be useful to determine appropriate input values for models with higher resolution than simple box models. The purpose of this study is to investigate temporal variability of N2O isotopomer ratios emitted from Japanese agricultural soils to obtain flux-weighted average values, and compare our results with previous studies (approach 1), and to examine factors such as soil type, fertilizer, and crops that control N2O production and consumption processes and hence isotopomer ratios (approach 2).
2. Materials and Methods
2.1. Study Site
 Monitoring of N2O flux and gas sampling for isotopomeric analysis were conducted at an experimental field consisting of twelve lysimeter plots (3 × 3 m, 1–1.2 m depth) in the National Institute for Agro-Environmental Sciences (NIAES) (36°N, 140°E), Japan.
 Plots F1 through F6 were filled with Fluvisols (Gray lowland soils) collected from a neighboring paddy field in 1993, and in which paddy rice (Oryza sativa L., cv. Nipponbare) had been cultivated annually. After the harvest of rice in 2001, plots F2, F3, F5, and F6 were kept drained and two kinds of upland crop cultivation have been conducted since 2002. Plots F3 and F6 were planted with upland rice (cv. Toyohatamochi) (here after referred to UR plots) and F2 and F5 were used for double cropping of soybean (cv. Enrei) and wheat (cv. Norin-61) [Nishimura et al., 2004, 2005a] (SW plots). Plots F1 and F4 were planted with the paddy rice as in previous years (PR plots). Urea was applied as nitrogen fertilizer.
 Plots A1 through A6 were filled with Andisols, in which several kinds of vegetables have been grown since 1996 [Akiyama and Tsuruta, 2002; Hayakawa et al., 2009]. During the period of this experiment, Komatsuna (B. rapa var. peruviridis, cv. Kuromisugi) were grown with three different treatments. Plots A1 and A4 were fertilized with poultry manure (hereafter referred to PM plots); A2 and A4, pelleted poultry manure (PP plots); A3 and A6, chemical fertilizer (ammonium sulfate) (CF plots).
Table 1 summarizes properties of the soils, fertilizers, and agricultural operations.
28 Sep 2006, fertilization and sowing; 13 Nov, harvest; 11 Jun 2007, fertilization and sowing; 17 Jul, harvest
2.2. Flux Measurement
 N2O flux was monitored every 4 h with an automated trace-gas flux monitoring system. This consisted of chambers, valves and tubing, sampling pumps, two gas chromatographs equipped with an electron capture detector (for N2O measurement) and a flame ionization detector (for CH4 measurement), an infrared gas analyzer (for CO2 measurement), and a computer. Details of the system are described elsewhere [Akiyama et al., 2000; Nishimura et al., 2005b]. For the Andisol experiments, NO flux was also monitored [Hayakawa et al., 2009].
2.3. Analysis of N2O Isotopomer Ratios
 Each gas sample for isotopomer analysis was collected into one or two plastic bags at the exit of the sampling pumps in phase with GC-ECD analysis. This was transferred immediately into an evacuated 1 L glass bottle equipped with two stopcocks. The sampling was conducted weekly in 2003, and 1–6 times per day during the period when high N2O flux was expected in 2006 and 2007. Ambient air was collected into another glass bottle at 2 m above ground near the experimental field, around noon of the sampling day.
 The N2O isotopomer ratios were measured at Tokyo Institute of Technology (Tokyo Tech) using a gas chromatograph–isotope ratio mass spectrometer (GC-IRMS) (MAT 252; Thermo Fisher Scientific K.K., Yokohama, Japan) system described elsewhere [Toyoda et al., 2005]. Site-specific nitrogen isotope analysis in N2O was conducted using ion detectors that had been modified for mass analysis of fragment N2O ions (NO+) containing N atoms in the central positions of N2O molecules, whereas bulk (average) nitrogen and oxygen isotope ratios were determined from molecular ions [Toyoda and Yoshida, 1999]. Pure N2O (purity > 99.999%; Syowa Denko K.K., Japan) was calibrated with international standards and was used as a working standard for isotopomer ratios. The notation of isotopomer ratios is shown below. The measurement precision was typically better than 0.1‰ for δ15Nbulk and δ18O, and better than 0.5‰ for δ15Nα and δ15Nβ.
Herein, 15Rα and 15Rβ represent 15N/14N ratios at the center and end sites of nitrogen atoms, respectively; 15Rbulk and 18R are average isotope ratios for 15N/14N and 18O/16O, respectively. Subscripts “sample” and “std” indicate isotope ratios for the sample and the standard atmospheric N2 for N and Vienna Standard Mean Ocean Water (V-SMOW) for O. We also define the 15N site preference (SP) as an illustrative parameter of the intramolecular distribution of 15N:
The N2O concentration was measured simultaneously with isotopomer ratios by comparing the peak area of the major ion (mass 44 and 30 in molecular ion analysis and fragment ion analysis, respectively) obtained with sample gas and that with our primary standard (349 ppb N2O in Air; Japan Fine Products Co., Ltd.) [Toyoda et al., 2005].
2.4. Soil Analysis
 Soil volumetric water content at 5 and 10 cm depths was monitored using time domain reflectometry moisture sensors (CS615, Campbell Scientific Instruments, Logan, Utah). The water-filled pore space (WFPS) was calculated from the volumetric water content and porosity.
 Soil samples were collected from 0 to 5 cm depth at 3–7 day intervals. For soil mineral nitrogen analysis, 15 g samples of fresh soil were extracted with 100 mL KCl solution and concentrations of nitrate (NO3−) and ammonium (NH4+) were measured by colorimetry using a continuous flow analyzer (TRRACS, Bran + Luebbe, Nordersterdt, Germany) [Akiyama and Tsuruta, 2003].
 A subset of the KCl solution samples was stored at −35°C for several months and further analyzed for N and O isotope ratios in the nitrate and ammonium.
 The δ15N of NH4+ was measured by the diffusion method [Holmes et al., 1998] where ammonium absorbed onto a glass fiber filter containing H2SO4 was converted to N2, and analyzed using a sensitivity-improved elemental analyzer-isotope ratio mass spectrometer (EA-IRMS) system at the Japan Agency for Marine-Earth Science and Technology (JAMSTEC) [Ogawa et al., 2010]. The δ15N and δ18O of NO3− were measured using the denitrifier method [Casciotti et al., 2002; Sigman et al., 2001] where N2O produced by Paracoccus aureofaciens (NBRC 3521) from the nitrate was analyzed as above.
 The δ15N of fertilizer was measured in a similar manner to soil NH4+, although urea was introduced directly into the EA-IRMS system.
2.5. Data Processing
 Isotopomer ratios for soil-emitted N2O (δsoil) were calculated from those for chamber gas samples (δchamber) and ambient air samples (δair) assuming two-component mixing. In the 2003 experiment, sampling was often conducted three consecutive times during the closure of the chamber, therefore the δsoil was obtained from the slope of the regression line for δchamber versus inverse Cchamber (Keeling plot). In other cases, δsoil was calculated using the following mass balance equation:
where C is N2O concentration and Csoil = Cchamber − Cair. The δsoil values obtained from small Csoil values (<10 ppbv, corresponding to flux of < 10 μgNm−2h−1) were not used for further data analysis because in such cases δchamber was equal to δair within the precision of the analysis and the error propagated in the calculation of δsoil was large.
 Statistical analyses of nitrous oxide isotopomer ratio data were made using Origin(R) (version 8) software (OriginLab Corp., Northampton, Massachusetts, United States). In order to evaluate the effects of crops and fertilizers, we applied analysis of variance (ANOVA) with Tukey's multiple range comparison tests. A value of p < 0.05 was considered statistically significant.
3.1. Fluvisol Experiment
Figure 1 shows the time course of N2O flux, soil NH4+ and NO3− concentrations, WFPS at 5 cm depth, and isotope/isotopomer ratios of NH4+ and N2O.
3.1.1. N2O Flux, Soil NH4+ and NO3− Concentrations, and WFPS
 In PR plots, peaks of N2O flux were observed immediately after the irrigation and basal fertilization (12 May), after the final drainage (16–23 September), and after the harvest (1–10 October) (Figure 1a). Except for these peak periods, N2O flux was < 10 μgNm−2h−1. Ammonium concentration increased after fertilization in May, July, and August. It decreased slightly from ∼2 to 1 mg N kg−1 dry soil around 6 October when a small N2O flux was observed. Nitrate concentration decreased after irrigation and remained 0–1.5 mg N kg−1 dry soil, except for a small increase (2.2 mg N kg−1 dry soil) around 6 October.
 In SW plots, significant N2O flux was observed (1) around the flowering and ripening stages of the wheat in May, (2) after the harvest of wheat (around 17 June), (3) after the fertilization of soybean (around 25 June), (4) 26 July to 9 August, (5) middle September, and (6) middle October (Figure 1a). Ammonium and nitrate showed maximum concentrations at the time of fertilization (20–24 June; see Figure 1b). During the period of the flux peaks (4) – (6), WFPS was decreasing gradually from maximum values of 80%–100% (plot F5) and about 50% (plot F2) (Figure 1c). The very high WFPS in plot F5 was caused by a faulty drainage system.
 In upland rice (UR) plots, high N2O flux was observed (1) 26 July to 9 August, (2) 27 August to 9 September, and (3) after the harvest (24 September to 13 October). Ammonium and nitrate showed maximum concentrations 1 week and 3–4 weeks after the fertilization, respectively. The WFPS showed almost the same temporal change as that of SW plot (F2), and N2O flux peaks 1 and 2 corresponded to the gradual decreasing phase of WFPS.
3.1.2. Isotope/Isotopomer Ratios in N2O and Soil NH4+
 The observed range in δ15Nbulk, δ18O, and SP of N2O was −30‰ to 35‰, 25‰ to 85‰, and −5‰ to 90‰, respectively (Figures 1d–1f). Although there was no statistically significant difference in δ15Nbulk and δ18O between PR, SW, and UR plots, SP was significantly higher in PR and UR plots than in SW plots. Correlations between N2O isotopomer ratios and other parameters (soil temperature, WFPS, N2O flux, ammonium and nitrate concentrations) were examined using flux-weighted average value for each characteristic date or period (Table 2). The δ18O had a positive correlation with nitrate concentration (r2 = 0.329, p = 0.019).
Table 2. Integrated N2O Flux, Average Values of N2O Isotopomer Ratios, and Average Concentrations and Isotope Ratios of NH4+ and NO3− During Selected Periods
 The δ15N of NH4+ in all the plots ranged from −12.5‰ to 18.7‰ (Figure 1d) while that of fertilizer (urea) was −2.4‰ ± 0.1‰. When weighted averages for selected periods are compared, δ15N of N2O is lower than that of NH4+ by on average 10‰ (Table 2).
3.2. Andisol Experiment in 2006
Figure 2 shows the time course of N2O flux, soil NH4+ and NO3− concentrations, WFPS, and isotope/isotopomer ratios of NH4+, NO3−, and N2O. All data except isotopic values are from Hayakawa et al. .
3.2.1. N2O Flux, Soil NH4+ and NO3− Concentrations, and WFPS
 The N2O flux increased immediately after fertilization, reached its maximum within 3–8 days, and decreased to initial levels 2 weeks after fertilization (Figure 2a). Although the nitrogen application rate was the same, the maximum values differed between fertilizers. They were 350, 1700, and 100 μg N m−2 h−1 for PM, PP, and CF, respectively.
 Soil NH4+ concentration showed a similar pattern to N2O flux, but the peak timing and relative magnitude between treatments were different. The concentration peak appeared 4 days after fertilization in PM and CF plots, and a week after fertilization in PP plots (Figure 2b). The peak height decreased in the order of CF, PP, and PM. After the NH4+ peak a smaller but wider NO3− concentration peak was observed. The concentration in different plots had a pattern similar to that of NH4+.
 During the period of large N2O flux, WFPS varied between 44% and 62% following rainfall events (1–2 and 5–7 October; see Figure 2c).
3.2.2. Isotope/Isotopomer Ratios in N2O and Soil NH4+ and NO3−
 The observed range in δ15Nbulk, δ18O, and SP of N2O was −10‰ to 5‰, 30‰ to 70‰, and −15‰ to 10‰, respectively (Figures 2d–2f). There was no significant difference in the isotopomer ratios between PM, PP, and CF plots. There was also no significant correlation between isotopomer ratios and other parameters.
 The δ15N of NH4+ and NO3− and δ18O of NO3− ranged from 5‰ to 20‰, −25‰ to −5‰, and −6‰ to 0‰, respectively (Figures 2d and 2e). The δ15N of NH4+ in fertilizer was 18.5‰, 18.6‰, and −1.6‰, for PM, PP, and CF, respectively. When weighted averages for selected periods were compared, δ15N of N2O was lower than that of NH4+ by 20‰, 16‰, and 4‰–16‰ for PM, PP, and CF plots, respectively, while it was lower than that of NO3− by 4‰, 12‰, and 5‰–27‰, respectively (Table 2). The δ18O of N2O for PM, PP, and CF plots was higher than that of NO3− by 60‰, 30‰–60‰, and 50‰–60‰, respectively (Table 2).
3.3. Andisol Experiment in 2007
Figure 3 shows the time course of N2O flux, soil NH4+ and NO3− concentrations, WFPS, and isotope/isotopomer ratios of NH4+ and N2O. All data except isotopic values are from Hayakawa et al. .
3.3.1. N2O Flux, Soil NH4+ and NO3− Concentrations, and WFPS
 The pattern and maximum value of N2O flux were similar to those of the 2006 experiments. However, the peak shape was slightly different from the previous year in PM and PP plots. The flux increased gradually during the initial 4 days and was followed by a sharp peak on 15 June (Figure 3a). The peak coincided with the peak in WFPS, which increased from 49% to 66% because of rainfall (Figure 3c).
 In PM and CF plots, soil NH4+ seems to have reached to its maximum within a day after fertilization (Figure 3b), after which it decreased. The NH4+ peak in PP plots was observed 4 days after fertilization. The timing of NH4+ peaks and the relationship of peak magnitude between the three treatments were similar to those of the 2006 experiments. Soil NO3− concentration increased 4 days after fertilization in all the plots (Figure 3b). The maximum concentration was 110, 100, and 230 mg N kg−1 dry soil for PP, PM, and CF, respectively. These values were about three times higher compared to those in 2006 experiments.
3.3.2. Isotope/Isotopomer Ratios in N2O and Soil NH4+ and NO3−
 The observed range in δ15Nbulk, δ18O, and SP of N2O was −45‰ to −5‰, 5‰–50‰, and 5‰–35‰, respectively (Figures 3d–3f). There was no significant difference in the isotopomer ratios between PM, PP, and CF plots, similar to the 2006 experiments. For the flux-weighted average value for each characteristic date or period, δ15Nbulk had a negative correlation with ammonium concentration (r2 = 0.676, p = 0.012).
 The observed range in δ15N of NH4+ and NO3− and δ18O of NO3− was similar to that in the 2006 experiment (Figures 3d and 3e). When weighted averages for selected periods are compared, δ15N of N2O was lower than that of NH4+ by 25‰–40‰, 31‰, and 35‰–60‰ for PM, PP, and CF plots, respectively, while it was lower than that of NO3− by 13‰–20‰, 8‰–16‰, and 17‰, respectively. An exception is that the difference in δ15N between N2O and NO3− was positive on 21 June in CF plots (Table 2). The δ18O of N2O was higher than that of NO3− by 30‰, 20‰–40‰, and 20‰–30‰, respectively (Table 2).
 In this section, we first discuss critical factors that affect N2O production and consumption processes using flux-weighted average isotopomer ratios for the whole observation period. Next, production and consumption processes at specific periods are discussed by comparing flux/concentration data and isotopomeric data. Finally, isotopomeric implications for global N2O balance are discussed.
4.1. Factors Controlling N2O Production and Consumption Processes
 As shown in section 3, statistical analysis of raw isotopomer data showed almost no significant difference between different soil types, fertilizers, and crops. Here we calculate flux-weighted averages for the whole observation period to estimate the representative values of isotopomer ratios for N2O emitted from different soils or soils amended with different fertilizers. The flux-weighted average N2O isotopomer ratios for each defined period (Table 2) were calculated using the observed δsoil and flux at the time of sampling. Then the averaged values were further averaged, whereby integrated N2O flux during the period was taken as the weight. For the Fluvisol experiment and Andisol experiment in 2006, this averaging was done for all the data because there was no significant difference between plots. For the Andisol experiment in 2007, data from PM and PP were combined for the same reason. Obtained isotopomer ratios are plotted in Figure 4, together with the flux-weighted or concentration-weighted average values reported by other researchers who studied N2O surface flux or soil depth profiles and their isotopomer ratios for fertilized agricultural soils. Also shown in Figure 4 is the range of other anthropogenic N2O sources.
4.1.1. Soil Type
 Applied fertilizer and planted crops differed between the two soils tested in this work (Table 1). However, the difference in fertilizer between urea and ammonium sulfate would cause a marginal effect because the former generates ammonium by hydrolysis and δ15N of both fertilizers agreed within 1‰. In addition, there was no significant difference in the flux-weighted average between crops in Fluvisol experiment. Therefore, we compare the flux-weighted average for Fluvisol with that for CF plots of Andisol (Table 2). Andisol data obtained in 2007 were used for this comparison because factors other than fertilizer seem to have controlled the N2O production processes in 2006 (see below).
 The three N2O isotopomer ratios for Fluvisol were larger than those for Andisol07(CF) (Figure 4) indicating a difference in production and/or consumption processes between the two soil types. It is known that N2O isotopomer ratios increase during N2O reduction both in pure culture of denitrifying bacteria and in soil mesocosms [Jinuntuya-Nortman et al., 2008; Ostrom et al., 2007]. Therefore, the difference could be explained by the difference in the progress of bacterial N2O reduction. Fluvisols are alluvial soil and have heavier texture than Andisols that are characterized by low-bulk density and good drainage [Deckers et al., 1998]. The clay contents of Fluvisol and Andisol used in this study were 36% and 18%, respectively. Because of these soil physical properties, Fluvisol tends to be more anaerobic and more likely to develop reductive micro sites than Andisol. Therefore, N2O reduction activity might have been higher in Fluvisols than Andisols.
 As shown in Figure 4, the averaged isotopomer ratios of N2O emitted from different soils including those reported by other researchers have a linear relationship: δ15Nbulk and δ18O are positively correlated (r2 = 0.618 and p = 2.4 × 10−5), whereas SP and δ15Nbulk are negatively correlated (r2 = 0.278 and p = 0.0025), if we exclude our Fluvisol data and those of Cardenas et al.  who incubated grassland soils with sheep manure. The slopes of the regression lines (1.58 ± 0.28 and −0.26 ± 0.10 for δ15Nbulk/δ18O and SP/δ15Nbulk, respectively) are different from those obtained in bacterial N2O reduction (ɛ(15Nbulk)/ɛ(δ18O) = 0.41 ± 0.05 and ɛ(SP)/ɛ(15Nbulk) = 1.12 ± 0.13, [Ostrom et al., 2007]). In addition, we found no significant relationship between isotopomer ratios and bulk density or clay content of the soils. Therefore, the relationship should be caused not only by N2O reduction but also by N2O production processes that depend on redox conditions, substrate availability, and bacterial consortia adapted to the soil. It is also suggested that our Fluvisol data might be significantly affected by N2O reduction, which is tested using a quantitative process model in section 4.2.2.
 The flux-weighted average isotopomer ratios for manure (PM/PMP) plots and synthetic fertilizer (CF) plots in the Andisol experiment in 2007 showed distinct values: the former exhibited higher δ15Nbulk and δ18O and lower SP compared to the latter (Table 2 and Figure 4). The difference in δ15Nbulk (18.3‰) is similar to the difference in δ15N of fertilizer (20.2‰), suggesting that the nitrogen source of N2O was fertilizer-N and that it is difficult to measure the difference in production and consumption processes of N2O between different fertilizers using δ15Nbulk alone, as implied by previous studies [Park et al., 2011; Well et al., 2008]. We therefore use both δ15Nbulk and SP in the process model analysis.
 Although most of the previously reported N2O isotopomer ratios were obtained using synthetic fertilizers, they show a wide range (Figure 4; “Silty clay loam” [Yamulki et al., 2001], one of the three data from “Hapludalf” [Opdyke et al., 2009], and “Rheidol” [Cardenas et al., 2007] were derived from soils amended with manure only). This means that factors other than fertilizer also play an important role in N2O production.
 Although large flux of N2O often coincided with high WFPS, we could not find significant relationship between N2O isotopomer ratios and WFPS. Davidson  hypothesized that high WFPS enhances denitrification from the observed inverse relationship between NO/N2O flux ratio and WFPS. If this is the case, SP of N2O should correlate with WFPS because N2O produced by nitrification (hydroxylamine oxidation) and denitrification (nitrite reduction) is known to have different SP values (Table 3). Therefore, the apparent insignificant relationship between SP and WFPS might suggest that N2O reduction perturbed SP and that relative magnitude of reduction to production was not dependent on WFPS alone. This effect can be considered using the process model in the following section.
Table 3. Reported Enrichment Factors for Bacterial N2O Production and Reduction Processes
These values were excluded for the determination of “best estimate” because production rate and yield of N2O in the experiment were quite low and SP might have been affected by processes other than bacterial denitrification (e.g., abiological formation).
4.2. Production and Consumption Processes of N2O at Specific Periods
4.2.1. Implications From Flux and Concentration Data
 In PR plots of the Fluvisol experiments, the first N2O flux peak observed immediately after irrigation was accompanied by increase in ammonium concentration and decrease in nitrate concentration (Figures 1a and 1b). While the ammonium increase should have resulted from fertilization, the decrease in nitrate suggests nitrate consumption (denitrification) under a reducing environment induced by the irrigation. Therefore, N2O was probably produced by denitrification. The smaller but wider subsequent two flux peaks coincided with a decrease in ammonium and an increase in nitrate. This indicates that N2O was produced from ammonium supplied by the fertilizer (September peak) or crop residue (October peak) under relatively oxidized conditions induced by drainage. However, no N2O emission was observed after the supplemental fertilization in July and August. This can be explained by the paddy soil becoming reducing because of flooding, which prevented N2O production or stimulated N2O reduction to N2.
 In SW and UR plots, N2O emissions accompanied by soil inorganic nitrogen concentration peaks (Figures 1a and 1b) would have originated from fertilizer-derived ammonium. However, there were several other N2O peaks which were not in phase with ammonium or nitrate concentrations. This can be explained by (1) N2O being produced at depth in the soil where inorganic nitrogen accumulated [Minamikawa et al., 2010] or (2) drastic change in redox conditions stimulating microbial N2O production. These processes likely occurred when large variations in WFPS were observed, e.g., between the end of July and the beginning of August.
 In the Andisol experiments, time lags between N2O emissions, soil inorganic nitrogen concentrations and fertilization were smaller than in the Fluvisol experiments (Figures 2a, 2b, 3a, and 3b). This indicates that N2O was produced in the surface soil layer from fertilizer-derived ammonium. However, observed differences in N2O emissions from PM, PP, and CF plots suggests different mechanisms of N2O production. In CF plots, ammonium might have been assimilated effectively by plants or microbes or leached from the surface layer, which could result in lower N2O emission. In the case of PM/PP plots, easily degradable organics from manure would have enhanced the activities of heterotrophic denitrifying bacteria, which could increase N2O emissions [Hayakawa et al., 2009]. On the basis of the NO/N2O emission ratio, Hayakawa et al.  deduced that nitrification and denitrification were dominant in CF and PM/PP plots, respectively, and that denitrification was enhanced in PP because of reducing micro sites inside the pelleted manure.
4.2.2. Implications From Isotope/Isotopomer Ratios
 Isotopomer ratios of N2O reflect source materials or production and consumption processes. The δ15Nbulk of newly produced N2O is a function of δ15N of substrates and the isotope/isotopomer enrichment factor, ɛ(15N)pro, of the production reaction [Mariotti et al., 1981]:
In the case of δ18O of N2O, however, it cannot be expressed simply by an equation similar to (9). For example, in nitrifier-denitrification, molecular oxygen (O2) and water are the oxygen source of nitrite [Andersson and Hooper, 1983; Hollocher et al., 1981], and three of the four oxygen atoms in two nitrite molecules are lost during N2O formation. We therefore have to consider δ18O of O2 and H2O, the ɛ(18O) of each reaction step that incorporates or releases oxygen atoms. Furthermore, nitrite is known to exchange its oxygen atom(s) with water by the reactions catalyzed by acid [Van Etten and Risley, 1981] or bacteria [Andersson and Hooper, 1983; Casciotti et al., 2002]. Because we obtained only δ18O of N2O and nitrate, and because the literature on ɛ(18O) is limited to denitrification (nitrate reduction to N2O; see Table 3), we only applied equation (9) to δ15N in the following discussion.
 In contrast to δ15Nbulk, SP of N2O depends only on the production pathway and not on δ15N of the substrate, provided that two chemically equivalent molecules (e.g., NH2OH and NO) are combined to form N2O [Sutka et al., 2003; Toyoda et al., 2005]. The reported SP values for N2O produced by hydroxylamine oxidation (nitrification, SPnit) and nitrite reduction (denitrification, SPdenit, and nitrifier-denitrification, SPnit-denit) are 33‰ ± 4‰ and −1.0‰ ± 5.5‰, respectively (Table 3; SPdenit and SPnit-denit are assumed to be equal and estimated by averaging the three “best estimate” values). Therefore, if bacterial nitrification and denitrification are responsible for N2O production, and if N2O reduction is not occurring, the relative contribution of nitrification, x, can be estimated from the following equation:
Recently, it has been reported that N2O production by fungal denitrification is significant in grassland soils [Crenshaw et al., 2008; Laughlin and Stevens, 2002], and Sutka et al.  found that SP values (SPfungi) of N2O produced by two species of fungi (Fusarium oxysporum and Cylindrocarpon tonkinense) are similar to those of bacterial nitrification. If fungal denitrification were also significant in this study, the x obtained from equation (10) would represent the relative contribution from the combined source consisting of bacterial nitrification and fungal denitrification [Opdyke et al., 2009].
 Here we use SP and ɛ(15N)pro for bacterial N2O production reported by pure culture incubation studies. Although these parameters are also available from soil incubation studies and they might be suitable for analysis of soil processes, it is problematic whether SPnit and SPdenit obtained by soil incubation are identical to those defined in this study. For example, Pérez et al.  assumed that N2O which was produced under atmosphere containing 10 kPa of acetylene (C2H2) derived from denitrification and obtained SPnit by comparison with the result obtained without C2H2 that inhibit NH3 oxidation and N2O reduction. Therefore, their reported SPnit might have been decreased by the contribution from nitrifier-denitrification whereas SPdenit might have been increased by the contribution from fungal denitrification. Well et al. [2006, 2008] used 15N-labeled nitrate to partition the N2O production into nitrification and denitrification, which means that nitrifier-denitrification is treated as nitrification. Soil incubation using reagents that inhibit nitrite or NO reduction and bacterial or fungal activity would give invaluable isotopomeric information that can be used in soil process modeling.
Equation (10) is not always applicable to the soil environment because N2O isotopomer ratios would also change by N2O reduction as follows (the “Rayleigh equation”):
where C is the N2O concentration, ɛred is the enrichment factor for reduction, and subscript 0 means the initial value. Therefore, we analyzed N2O production and consumption processes on the basis of equations (9)–(11), similar to previous studies on N2O dissolved in seawater [Yamagishi et al., 2007] or groundwater [Koba et al., 2009]. Briefly, ranges of δ15Nbulk and SP for N2O produced by each microbial process were estimated from observed δ15N of substrate and literature values for ɛ(15N)pro and SP. Then, observed δ15Nbulk and SP of soil-emitted N2O were compared with each end-member on SP − δ15Nbulk space, and the relative fraction of N2O derived from nitrification, x (0 ≤ x ≤ 1), and the approximate measure of progress of N2O reduction, Fr (0 ≤ Fr ≤ 1), were calculated using Monte Carlo method (Figure 5; see Text S1). We made several calculations with respect to the choice of two end-members in equation (10): (1) nitrification and denitrification, (2) nitrification and nitrifier-denitrification, (3) fungal denitrification and denitrification, and (4) fungal denitrification versus nitrifier-denitrification. The results in cases 3 and 4 were similar to those in cases 1 and 2, respectively. This is because the estimated ranges of δ15Nbulk of N2O produced by nitrification and fungal denitrification often overlaps in addition to the similarity between SPnit and SPfungi as already mentioned. Therefore, we show the results obtained by considering only bacterial processes. Although fungal biomass could be reduced by tillage [Frey et al., 1999], the significance of fungal denitrification in cultivated soils should be examined using other approaches as well as N2O isotopomer ratios.
 Nitrous oxide reduction occurred in almost all cases (Figure 6a). The Fr was generally higher in Fluvisols than in Andisols. In contrast, x varied both with time, soils, and treatments. In PR plots, nitrification and denitrification/nitrifier-denitrification contributed equally in October (Figure 6b). This is in accordance with the temporal minimum and maximum of ammonium and nitrate concentrations, respectively, observed for the period. In SW plots, denitrification or nitrifier-denitrification was dominant for all five dates/periods (Figure 6b). On June 17, flux/concentration analysis indicated that ammonium released from the fertilizer was the N2O substrate; hence nitrifier-denitrification is the process responsible. In contrast, N2O was likely produced from denitrification in the deep soil layer in July–September [Minamikawa et al., 2010], considering the low surface inorganic nitrogen concentrations and high WFPS in this period. In UR plots, equal contributions from nitrification and denitrification/nitrifier-denitrification were deduced for the periods 29 July to 5 August and 24 September to 15 October, but a higher (70%–80%) share of nitrification was estimated on 2 September. The dominance of nitrification is in line with the decrease in WFPS and the small nitrate peak observed around 2 September (Figures 1b and 1c).
 For the Andisol experiment in 2006, our analysis based on the SP − δ15Nbulk relationship did not give meaningful Fr and x for four out of the seven dates/periods. This was because observed N2O had relatively high δ15Nbulk values compared to those of bacterial N2O calculated from δ15N of ammonium or nitrate, and therefore SP became extraordinary low value if we reconstruct isotopomer ratios before N2O was reduced. Considering the three successive calculation results that indicated very high (∼90%) contribution from denitrification/nitrifier-denitrification (Figure 6b), there are two possible causes for the high δ15Nbulk of N2O: (1) δ15N of nitrite was significantly higher than expected or (2) N2O production occurred in a deeper soil layer where δ15N of nitrate might have been higher than in the surface layer. The latter possibility is likely when we take into account there was almost no difference in δ15Nbulk of N2O between different fertilizer treatments (PM/PP versus CF plots), and that most of the isotopic data were obtained in the initial or final phase of N2O emission. The nitrate concentrations were lower in 2006 than in 2007, especially for the period of our sample collection for N2O isotopes (Figure 2). There were heavy rainfall events immediately before the fertilization (26–27 September) and in the later phase of N2O emission (5–7 October) which increased WFPS (Figure 2c) and might have suppressed nitrification, and therefore N2O production, in the surface soil layer. In the deeper layer however, denitrification could be enhanced by the increased soil moisture. The substrate for the denitrification could be residual nitrate originating from fertilizer of the previous year (chemical fertilizer was applied to all the plots in 2005) or nitrate mineralized from soil organic nitrogen. In either case, nitrate would have been isotopically “heavier” than surface N species because of partial reduction by denitrifiers, which could be the source of heavy N2O. This presumed process is supported by the observation that annual emission of N2O dissolved in the subsurface drainage from Fluvisol plots in 2006–2007 was as high as 50%–67% of the aboveground gaseous emission [Minamikawa et al., 2010], although the groundwater table was set to a depth of 0.9 m from the soil surface for the Fluvisol plots whereas the drainage from Andisol plots was allowed to drain freely.
 In the Andisol 2007 experiment, the contribution of nitrification was higher in CF plots (40%–70%; see Figure 6b) than in PM/PP plots (10%–50%), which is in accordance with results from the NO/N2O flux ratio [Hayakawa et al., 2009]. Synthetic fertilizer such as ammonium salt easily releases ammonium which can be oxidized subsequently to hydroxylamine and nitrite by nitrifying bacteria. However, ammonium often accumulated immediately after the application of synthetic fertilizer (Figures 2 and 3). This suggests that uptake by plant or oxidation by nitrifiers was rate-limiting. This situation would increase the possibility of N2O formation from hydroxylamine. In contrast, organic molecules supplied by the organic fertilizer could enhance N2O production from nitrite because they are essential as electron donors for the heterotrophic denitrifiers.
 The isotopomeric analysis also showed that similar N2O production and consumption processes were responsible for the N2O emission peak observed in both PM and PP plots in 2007. This suggests that higher N2O emission from PP than that from PM is caused by rapid rate of net production ( = production − consumption). Substrates for denitrification/nitrifier-denitrification (i.e., ammonium, nitrite, nitrate, and organic substances) are concentrated in the pelleted manure, whereas they would diffuse into surrounding soils in the case of normally applied manure. Therefore, if the reaction rates were limited by substrate concentrations, PP could release more N2O than PM.
 The positive correlation between δ18O of N2O and nitrate concentration observed in Fluvisol plots and the negative correlation between δ15Nbulk and ammonium concentration observed in Andisol 2007 plots are interpreted as follows on the basis of above discussion. In Fluvisols, N2O produced by nitrification/nitrifier-denitrification/denitrification was reduced intensively by denitrification. In the initial stage, δ18O value of N2O would be low because production is likely to dominate over reduction, whereas in the later stage it would be high because of N2O reduction (e.g., Fr value for SW plots increased from June to October; see Figure 6a). On the other hand, nitrate accumulates as nitrification proceeds. This would result in an apparent positive relationship between δ18ON2O and nitrate concentration. The absence of such a relationship between δ15Nbulk or SP and nitrate concentration might be because absolute values of ɛ(15N) or ɛ(SP) values for N2O reduction are lower than ɛ(18O) (Table 3). In the Andisol 2007 experiment, the contribution of nitrification to N2O production was relatively larger in the initial stage than in the later stage (Figure 6b). Therefore, in the initial stage when ammonium concentration was high, δ15Nbulk would be low because of the large fractionation during ammonium oxidation by nitrifiers (Table 3). As ammonium is consumed in the later stage, δ15N of the substrate pool would become higher. In addition, the magnitude of 15N fractionation in nitrite reduction by nitrifiers or denitrifiers is not as large as nitrification. Therefore, a negative correlation between δ15Nbulk and ammonium concentration might have been observed.
4.3. Implications for Global N2O Budget
 Although the validity and cause of the relationship between isotopomer ratios of N2O emitted from fertilized soils (Figure 4) should be examined by further studies on other type of soils and fertilizers, it can be used to parameterize the isotopomeric signature of N2O from fertilized soils, which can then be used for modeling of the global N2O budget.
 Recent studies on archived air samples in firn or ice cores show that N2O concentrations and isotopomer ratios show a secular trend [Ishijima et al., 2007; Röckmann et al., 2003; Sowers et al., 2002]. Simple box model calculations predicted isotopomer ratios for anthropogenic N2O added to the atmosphere since the Industrial Revolution. The values predicted by Röckmann et al.  and Ishijima et al.  are within the range of δ15Nbulk, δ18O, and SP of soil N2O (Figure 4). Since other anthropogenic N2O sources so far studied show different isotopomeric compositions (Figure 4), this work provides further evidence of the major role that agricultural soils have in the increase of N2O in the atmosphere as it has previously been shown from bulk 15N data [Pérez et al., 2001; Rahn and Wahlen, 2000].
 Isotopomeric analysis of N2O emitted from temperate agricultural soils indicated that relative contributions from nitrification (hydroxylamine oxidation) and denitrification (nitrite reduction) to gross N2O production depend on soil type and fertilizer type, and in some cases, on soil moisture. In the two soils studied in this work, partial reduction of N2O was pronounced in high-bulk density, alluvial soil (Fluvisol) compared to low-bulk density, volcanic ash soil (Andisol). When N2O production occurred in the surface layer, the contribution from nitrification was relatively high in soils amended with synthetic ammonium fertilizer, while denitrification was dominant in the soils amended with poultry manure. However, the process information obtained here relies on isotopomer characteristics found in pure culture of bacteria. Further studies are needed on isotopomer enrichment factors for each N2O production and consumption process by bacteria and fungi in intact soils. From combined data derived from the literature and this study, linear relationships were found between flux-weighted average isotopomer ratios of N2O emitted from fertilized soils. The relationships would be useful to parameterize isotopomer ratios of soil-emitted N2O for modeling the global N2O isotopomer budget. The results obtained in this study confirm that the major source of anthropogenic N2O is fertilized soils.
 We are grateful to S. Oshita and H. Fujimura for sampling support and to Y. Ueno, A. Fujii, N. Suzuki, H. Yamagishi, and H. Nara for helpful discussions. N. E. Ostrom and T. Perez are acknowledged for constructive review. This work was supported by the Global Environment Research Fund (A-0904) of the Ministry of the Environment, Japan, and Grants-in-Aid for Creative Scientific Research (2005/17GS0203) of the Ministry of Education, Culture, Sports, Science, and Technology, Japan.