Variation in the emission factor of N₂O derived from chemical nitrogen fertilizer and organic matter: A case study of onion fields in Mikasa, Hokkaido, Japan

Abstract

Variability in the emission factors of nitrous oxide (N₂O) associated with the application of chemical fertilizer (EF_F) and organic matter (EF_O) were analyzed in two onion fields (GL, Gray Lowland soil [Gleysol; Food and Agriculture Organization/UNESCO]; BL, Brown Lowland soil [Fluvisol; Food and Agriculture Organization/UNESCO]) in Mikasa, Hokkaido, Japan. Nitrous oxide flux was measured using a closed chamber technique in four treatments (FOP, chemical nitrogen fertilization and organic matter application, with plants; F, chemical nitrogen fertilization only, without plants; OP, organic matter application only, with plants; C, control, no fertilization or organic matter application, without plants) for 4 years in GL (2000, 2003–2005) and for 1 year in BL (2005). The application rate of chemical fertilizer nitrogen ranged from 237 to 242 kg N ha⁻¹ year⁻¹ in GL and was 284 kg N ha⁻¹ year⁻¹ in BL; organic matter nitrogen ranged from 81 to 117 kg N ha⁻¹ year⁻¹ in GL and was 181 kg N ha⁻¹ year⁻¹ in BL. The emission factors (EF) were calculated using the equations: EF_F (%) = (N₂O emission in FOP–N₂O emission in OP)/(applied chemical nitrogen fertilizer) × 100 and EF_O (%) = (N₂O emission in FOP–N₂O emission in F)/(applied organic matter nitrogen) × 100. The annual N₂O emissions for treatments FOP, F, OP and C were 7.2–17, 5.7–17, 3.2–9.9 and 2.0–12 kg N ha⁻¹ year⁻¹, respectively, in GL and 5.6, 2.8, 1.9 and 1.8 kg N ha⁻¹ year⁻¹, respectively, in BL. The EF_F ranged from 1.3% to 5.5% in GL and was 1.3% in BL. The EF_F was positively correlated with the mean annual air temperature (P < 0.01), suggesting that N₂O emission derived from chemical nitrogen fertilizer increases as air temperature rises. The EF_O, however, differed greatly between GL (ranging from −5.2% to 9.1%) and BL (1.5%). The EF_O was positively correlated with the mean annual relative humidity, although the correlation was not significant (P = 0.23). This finding suggests that much wetter climatic conditions may increase N₂O emissions derived from organic matter nitrogen. The estimated N₂O emissions based on these EF values and the rate of nitrogen application coincided well with the measured N₂O emissions in the FOP treatment in both soils.

Key words: :

• chemical fertilizer
• crop residue
• emission factor
• nitrous oxide
• organic fertilizer

INTRODUCTION

Nitrous oxide (N₂O) is a major greenhouse gas. It has a 296-fold higher greenhouse effect than that of carbon dioxide (CO₂), and its concentration in the air is increasing at a rate of 0.8 p.p.b. per year (Intergovernmental Panel on Climate Change 2001). The global emission of N₂O was estimated to be 16.2 Tg N year⁻¹, and the emission from agricultural soils accounted for 24% of all emissions (Intergovernmental Panel on Climate Change 2001; Mosier et al. 1998).

The simplest way to estimate N₂O emission from agricultural soil is to use the emission factor (EF) proposed by the Intergovernmental Panel on Climate Change (2006) and fertilizer nitrogen application rate described to statistical data (e.g. published by the Japanese government). The EF is the ratio of the N₂O emission divided by the amount of nitrogen fertilizer applied. The Intergovernmental Panel on Climate Change has proposed a default EF value of 1.0%, although if an EF value based on actual measurements is available, that value is preferred to the default value. In Japan, the EF was estimated to be 0.62% for all fertilized upland fields (Akiyama et al. 2006). The N₂O emission from agricultural soils usually increases with increases in the application of chemical fertilizer, manure and crop residue (Bouwman 1996; Cochran et al. 1997; Mogge et al. 1999; Mori et al. 2005; Tokuda and Hayatsu 2004; Zou et al. 2005), and EF values differ among the various types of fertilizers (e.g. 1.65% for chemical fertilizer and 0.15–0.59% for residues in winter wheat fields; Zou et al. 2005). In addition, because several types of amendments are usually applied to an agricultural field, N₂O emissions from the soil would be affected by their interactions. Thus, to more accurately assess N₂O emissions from agricultural fields receiving chemical fertilizer, manure and crop residue, each of these EF values needs to be calculated.

Several studies have reported annual variation in EF values, which may be the result of climatic differences. For example, N₂O emissions ranged from 3.5 to 15.6 kg N ha⁻¹ year⁻¹ and the EF ranged from 1.1 to 6.4% in an onion field in Hokkaido, Japan (Kusa et al. 2002). Dobbie et al. (1999) reported N₂O emissions of 1.0–18.4 kg N ha⁻¹ year⁻¹ and EF values of 0.3–5.8% in grasslands in Scotland. Takakai et al. (2006) reported N₂O emissions of 7.1–256 kg N ha⁻¹ year⁻¹ and the EF ranged from 1.8 to 36% in agricultural peat soil in Indonesia. Smith et al. (1998) noted that the EF calculated in Scottish agricultural soils was lower than 1.25% because of lower temperatures than those recorded by Bouwman (1996).

Nitrous oxide emissions from agricultural soils originate not only from applied nitrogen but also from the nitrogen derived from soil organic matter. The purpose of this study was to calculate EF values for chemical fertilizer and added organic matter and to investigate the annual variation in EF values for agricultural soils in Mikasa, Hokkaido, Japan.

MATERIALS AND METHODS

Site description

Two onion fields with Gray Lowland (GL) soil (Gleysol; Food and Agriculture Organization/UNESCO) and Brown Lowland (BL) soil (Fluvisol; Food and Agriculture Organization/UNESCO) in Mikasa, Hokkaido, Japan, were investigated (GL, 43°14.4′Ν, 141°50′Ε; BL, 43°13.6′Ν, 141°49.1′Ε). Previous crops were onion and onion has been cultivated for more than 10 years at both sites. The distance between the GL and BL sites was 2.7 km. The study was conducted in 2000 and 2003–2005 in GL and in 2005 in BL. The mean annual temperature and precipitation were 7.4°C and 1,155 mm, respectively, over the past 30 years (observed at Iwamizawa weather station, 43°12.6′Ν, 141°47.3′E). Soils at both sites were covered with snow, but did not freeze, in winter. Table 1 lists the soil chemical properties at both sites.

Management of both fields (e.g. application rate and method of fertilizer) treated as farmer's practices around the fields. At the beginning of May at site GL, 242 kg N ha⁻¹ year⁻¹ (NH⁺ ₄, 187; NH⁻ ₄, 55) of chemical fertilizer was applied in 2000, 2003 and 2004 and 237 kg N ha⁻¹ year⁻¹ (NH⁺ ₄, 183; NH⁻ ₄, 54) was applied in 2005. At the end of April 2005 at site BL, 204 kg N ha⁻¹ (NH⁺ ₄, 45; NH⁻ ₄, 159) was applied and 80 (Nitrolime) kg N ha⁻¹ was applied in mid October (total 284 kg N ha⁻¹ year⁻¹). Chemical fertilizers were applied to the surface of the soil afterwhich the soil was plowed to 0–5 or 10 cm depth. At site BL, organic fertilizer was applied in the form of manure made by sludge at a rate of 45 kg N ha⁻¹ year⁻¹ at the end of March and as rice brain at a rate of 38 kg N ha⁻¹ year⁻¹ in mid October. After the harvest, onion residue was left on the field at both sites until plowing. At site GL, the amounts of

Table 1 Chemical properties of the soil layers in Gray Lowland soil (GL) and Brown Lowland soil (BL)

Site	Horizon	Depth	pH(H2O)^†	CEC (cmolc kg^−1)	Sand (%)	Silt (%)	Clay (%)	C (%)	N (%)	C/N
GL	Ap1	0–10	5.8	25.5	12.1	51.2	36.7	3.21	0.28	11.5
	Ap2	10–33	5.2	24.6	11.3	50.9	37.8	3.51	0.28	12.7
	Bg	33–64	4.4	27.1	2.90	47.9	49.2	4.91	0.31	16.0
	C	64–100	4.7	33.1	6.10	56.3	37.6	7.66	0.44	17.4
BL	Ap1	0–15	5.2	24.6	6.00	57.4	36.6	2.17	0.19	11.3
	Ap2	15–35	5.2	24.1	5.50	58.7	35.8	2.06	0.18	11.5
	Bg	35–100+	4.2	20.8	12.6	61.6	25.8	1.36	0.14	9.82
^†Soil : distilled water ratio was 1:2.5 (GL) and 1:5 (BL).

onion residue in 2000, 2003, 2004 and 2005 were 86.9, 81.0, 92.3 and 117 kg N ha⁻¹ year⁻¹, and the C:N ratios of the residues were 20.0, 23.1, 20.5 and 20.1, respectively. At site BL, the amount of onion residue was 98.7 kg N ha⁻¹ year⁻¹ and the C:N ratio of the residue was 20.8. As is common in this area, onions were planted at the beginning of May, root cutting was carried out in mid August, and onions were harvested at the beginning of September at both sites. The fields were plowed in mid October, and crop residue was incorporated at both sites.

Treatments

We set up four treatments in both GL and BL: FOP, chemical nitrogen fertilization and organic matter application, with plants; F, chemical nitrogen fertilization only, without plants; OP, organic matter application only, with plants; and C, control, no fertilization or organic matter application, without plants. Organic matter application included onion residue for treatments FOP and OP at both sites, and also included organic fertilizer at the BL site. The area of each treatment plot was 40 m² (5 m × 8 m) at GL and 8 m² (2 m × 4 m) at BL. Each treatment was set up at the beginning of April (one replicate at each site) and management (e.g. ploughing) was the same for each treatment. The treatment plot at site GL was created at the same place throughout the investigation period.

N₂O and NO flux measurements

N₂O and NO fluxes were measured using a closed chamber method (four replications) with two types of cylindrical stainless steel chambers. In 2000, a chamber that was 30 cm in diameter and 35 cm in height was used (Kusa et al. 2002). In 2003, 2004 and 2005, the diameter of the chamber was 20 cm and the height was 25 cm (Toma and Hatano 2007). The cover of the chamber was made of acryl and was equipped with a sample collector, pressure regulating bag and a Tedlar bag (0.5 L). In 2000, 2003 and 2004, we put the chambers 2 or 3 cm into the soil and started measurements after 15 min. In 2005, however, the chamber had a base made of stainless steel with a diameter of 20 cm; the upper part of the base was equipped with a slight depression, which was filled with water to seal it during the measurement (Toma and Hatano 2007). The base was kept on the ground, except during harvesting and cultivation, and the base was reset 1 day before the next sampling. Onion plants and their living parts were removed from the chamber base, but the litter covering the base was left inside during the measurement.

We took the gas samples before and 15 min after closing the chamber. A 250-mL gas sample was taken using a syringe (25 mL) and was injected into a Tedlar bag (0.5 L). The bags were then brought to the laboratory and 20 mL of each gas sample was immediately transferred into a glass vial (10 mL). From these bag samples, NO was analyzed using a chemoluminescence nitrogen oxide analyzer (model 265P, Kimoto Electric, Osaka, Japan) within 12 h, and from the samples of vials N₂O was analyzed with an ECD (Electron capture detector) gas chromatograph (model GC-14B, Shimadzu, Kyoto, Japan) within 1 month.

Gas fluxes were calculated using the following equation:

where F is the flux (mg N m⁻² h⁻¹), ρ is the gas density (ρN₂O-N = 1.26 × 10⁶ and ρNO-N = 0.63 × 10⁶ mg N m⁻³), V is the volume of the chamber (m³), A is the area of the chamber (m²), Δc/Δt is the ratio of change in the gas concentration inside the chamber (10⁻⁶ m³ m⁻³ h⁻¹), T is the air temperature inside the chamber (°C) and P is the air pressure (mm Hg; see Kusa et al. 2002). Due to considerations of machinery precision, N₂O flux values ranging from −6 to 6 µg N m⁻² h⁻¹ and NO flux values ranging from −0.2 to 0.2 µg N m⁻² h⁻¹ were regarded as 0 µg N m⁻² h⁻¹. Monthly and annual N₂O emissions were calculated assuming linear changes between two sampling occasions.

Calculations of EF_F, EF_O and estimated N₂O emission from the FOP treatment

If we assume that N₂O emission is composed of N₂O derived from applied chemical nitrogen fertilizer, applied organic matter and soil organic matter decomposition, N₂O emission can be estimated as:

where EF_F is the emission factor of N₂O associated with the application of chemical nitrogen fertilizer, EF_O is the emission factor of N₂O associated with the application of organic nitrogen and [C] is N₂O emission in the C treatment. EF_F and EF_O were calculated using the following equations:

where [FOP] is the N₂O emission measured in the FOP treatment (kg N ha⁻¹ year⁻¹), [OP] is the N₂O emission measured in the OP treatment (kg N ha⁻¹ year⁻¹) and [F] is the N₂O emission measured in the F treatment (kg N ha⁻¹ year⁻¹). Onion residue might be decomposed during the snow-covered season because N₂O fluxes were low after the snowmelt (Kusa et al. 2002). In addition, Toma and Hatano (2007) reported that approximately 45% of the carbon in onion residue was decomposed within 2 months. Therefore, we assumed that the onion residue was almost decomposed within the year.

Other measurements and sampling frequency

The soil temperature at 5 cm depth was measured five times in all treatments during the gas measurements. Disturbed soil samples were taken from depths of 0–5 cm, as a mixing soil sample from five places in each plot. Samples were extracted using distilled water, and NH⁺ ₄ and NH⁻ ₄ concentrations were analyzed by colorimetry with indophenol-blue and ion chromatography (QIC analyzer, Dionex Japan, Osaka, Japan), respectively. Three replicate undisturbed soil samples were collected using a steel corer (100 mL) and the water-filled pore space (WFPS) was measured.

After fertilizer application, samples were collected three times per week from May to June, three times per month from July to August, and once or twice per month during the snow-covered season. From after snow melt until before fertilization, and from mid October until snow cover, sampling was conducted two or three times per month.

Statistical analysis

The relationships between the meteorological condition and N₂O emission or EFs were determined using regression analyses. We used Excel Tahenryoukaiseki Ver.4.0 for Windows (Social Survey Research Information Company, Tokyo, Japan) for the regression analysis.

RESULTS

N₂O flux

Figure 1 shows the seasonal variation in N₂O flux at each site. Two peaks were observed each year at GL. The first peak was observed from May to June in the FOP and F treatments after chemical fertilizer application. The second peak occurred from late August to October in all treatments. At BL, only a negligible increase in N₂O flux occurred in May in the FOP treatment, and the flux further increased during July and August. This was also found in the other treatments.

From May to June, at GL, the maximum N₂O flux values in the FOP and F treatments in 2000 and 2004 were similar, and those in 2003 and 2005 resembled each other. From May to June in 2000 and 2004, the maximum N₂O fluxes in FOP were 1.86 and 2.16 mg N m⁻² h⁻¹ and those in F were 1.19 and 1.04 mg N m⁻² h⁻¹, respectively. And in 2003 and 2005, the maximum N₂O fluxes in FOP were 0.26 and 0.33 mg N m⁻² h⁻¹ and those in F were 0.13 and 0.16 mg N m⁻² h⁻¹, respectively. In BL, the maximum N₂O flux was 0.12 mg N m⁻² h⁻¹ from May to June in 2005 in the FOP treatment.

From the end of August to the end of October each year, N₂O flux in the FOP, F and OP treatments at GL ranged from 0.02 to 2.07, 0.02 to 2.10 and 0.01 to 1.65 mg N m⁻² h⁻¹, respectively, which was slightly higher than the ranges in the C treatment (0.01–1.33 mg N m⁻² h⁻¹). At BL, these values ranged from 0.001 to 0.44, –0.002 to 0.37 and 0.004 to 0.12 mg N m⁻² h⁻¹, respectively, and the values of C ranged from 0.01 to 0.17 mg N m⁻² h⁻¹.

NO flux and N₂O-N/NO-N

The maximum NO fluxes in the FOP and F treatments at GL were 3.30 and 0.78 mg N m⁻² h⁻¹, respectively, and these fluxes were observed from May to June in all years. At BL, the maximum NO flux in the FOP and F treatments was 0.12 mg N m⁻² h⁻¹ at the end of June and 0.48 mg N m⁻² h⁻¹ in early May. The NO fluxes in the OP and C treatments at both sites were low and stable throughout all seasons (data not shown). Based on these values, the ratio between N₂O-N and NO-N (N₂O-N/NO-N) decreased to less than 1.0 from May to June and increased to approximately 100 from September to October in the FOP and F treatments at GL (Fig. 2). At BL, N₂O-N/NO-N decreased to less than 1.0 in June and increased to approximately 100 in August in the FOP and F treatments. In the OP and C treatments at both sites, the N₂O-N/NO-N remained above 1.0 from May to June and the seasonal change in the ratio was small compared with that in FOP and F.

Soil temperature, WFPS, and soil NH⁻ ₄ and NO⁻ ₃ concentrations

Soil temperature and WFPS are shown in Fig. 3. Seasonal variation in soil temperature was similar at both sites and resembled that of air temperature, with maximum values observed from July to August. At GL, a positive significant relationship between N₂O flux and soil temperature was observed throughout all seasons in the OP and C treatments (OP, n = 104, r ² = 0.13, P < 0.01; C, n = 104, r ² = 0.07, P < 0.01). Considering only those periods when high N₂O fluxes were observed (from May to June and from September to October), however, a positive significant relationship between N₂O flux and soil temperature was observed only in OP (May–June, n = 32, r ² = 0.12, P < 0.05; September–October, n = 30, r ² = 0.26, P < 0.01). At BL, a positive significant relationship between N₂O flux and soil temperature was observed throughout all seasons in the FOP, F and C treatments (FOP, n = 22, r ² = 0.21, P < 0.05; OF, n = 20,

Figure 1  Seasonal variation in N2O flux at the Gray Lowland soil (GL) and Brown Lowland soil (BL) sites. The arrows indicate the time of chemical fertilizer application (CF), organic fertilizer application (OF), root cutting (RC) and harvest and residue application (H). The four treatments were: FOP, chemical nitrogen fertilization and organic matter application, with plants; F, chemical nitrogen fertilization only, without plants; OP, organic matter application, with plants; C, no fertilization or organic matter, and no plants. Error bars indicate standard deviation.

r ² = 0.30, P < 0.05; C, n = 22, r ² = 0.40, P < 0.01). From September to October, when high N₂O emission was recorded, a significant positive relationship was observed only in C (n = 8, r ² = 0.54, P < 0.05).

The WFPS in all treatments at GL and BL were high after snow melt, decreased to 25–50% in summer and increased again in autumn (Fig. 3). Beginning in August, the WFPS at BL increased rapidly from 25% to approximately 75%, whereas that at GL increased more slowly. Across each year there was no relationship between N₂O flux and WFPS in any treatment at either site. However, N₂O flux was positively correlated with WFPS in the F treatment at GL from May to June (n = 30, r ² = 0.23, P < 0.01) and in the OP treatment at BL from September to October (n = 8, r ² = 0.54, P < 0.05).

Maximum soil NH⁺ ₄ concentrations in the FOP and F treatments at both sites were observed in May; after this time, however, NH⁺ ₄ was not detected (Fig. 4). Throughout each year, there was no significant relationship between N₂O flux and soil NH⁺ ₄ concentration. However, a positive relationship was observed in the OP treatment at site GL from September to October (n = 20, r ² = 0.31, P < 0.05).

Soil NH⁻ ₄ concentration increased after fertilizer application in the FOP and F treatments at both sites, and maximum concentration was observed from June to August (Fig. 4). Soil NH⁻ ₄ concentrations in the OP and C treatments also increased from June to August, but the concentrations were lower than those in FOP and F. In all treatments, soil NH⁻ ₄ concentration decreased in mid August. Throughout all seasons, a positive significant relationship between N₂O flux and soil NH⁻ ₄ concentration was found in the C treatment at GL (n = 76, r ² = 0.16, P < 0.01).

N₂O emission, EF_F and EF_O

Annual N₂O emissions differed greatly among treatments and years (Table 2). At GL, the annual N₂O emissions ranged

Figure 2  Seasonal variation in N2O-N/NO-N at the Gray Lowland soil (GL) and Brown Lowland soil (BL) sites. See Fig. 1 for an explanation of the arrows and treatments.

from 7.19 to 17.3 mg N m⁻² h⁻¹ year⁻¹ in FOP, 5.72 to 16.7 mg N m⁻² h⁻¹ year⁻¹ in F, 3.24 to 9.90 mg N m⁻² h⁻¹ year⁻¹ in OP and 2.01 to 12.1 mg N m⁻² h⁻¹ year⁻¹ in C. At BL, the annual N₂O emission in the FOP, F, OP and C treatments were 5.56, 2.83, 1.88 and 1.81 mg N m⁻² h⁻¹ year⁻¹, respectively. The 2-month N₂O emission from May to June was positively correlated with the mean temperature during these months (n = 8, r ² = 0.53, P < 0.05; Fig. 5a) in the FOP and F treatments at site GL. The 2-month N₂O emissions from September to October in the FOP and OP treatments at GL, however, were positively correlated with the precipitation during these months (n = 8, r ² = 0.39, P < 0.1; Fig. 5b).

The EF_F values, calculated using the annual N₂O emission in the FOP and OP treatments according to Eq. 3, ranged from 1.32 to 5.50% at GL and the EF_F value at BL was 1.30%. The EF_O values, calculated using the annual N₂O emission in the FOP and F treatments according to Eq. 4, ranged from −5.22% to 9.06% at GL and the EF_O value at BL was 1.51% (Table 2). The EF_F was positively correlated with mean annual temperature at GL (n = 4, r ² = 0.96, P < 0.01; Fig. 6a), and the EF_O increased as mean annual relative humidity increased at GL, although the relationship was not statistically significant (n = 4, r ² = 0.59, P = 0.23; Fig. 6b). The EF_F values at GL and BL in 2005

Figure 3  Seasonal variation in air and soil temperatures and water-filled pore space (WFPS) at the Gray Lowland soil (GL) and Brown Lowland soil (BL) sites. See Fig. 1 for an explanation of the treatments.

were very similar (GL, 1.32%; BL, 1.30%), whereas the EF_O values differed greatly (GL, −5.22%; BL, 1.51%).

By assuming that N₂O emission is composed of N₂O derived from applied chemical nitrogen fertilizer, applied organic matter and soil organic matter decomposition, N₂O emission was estimated according to Eq. 2. Chemical fertilizer induced N₂O emissions were calculated by multiplying the EF_F of the actual measurement by the chemical nitrogen application rate, and organic fertilizer induced N₂O emissions were calculated by multiplying the EF_O of the actual measurement by organic nitrogen application rate. N₂O emissions in the C treatment were assumed to have originated from soil organic matter. According to our field measurements, the estimated N₂O emission in the FOP treatment was approximately 20% larger than the measured emission (Fig. 7).

DISCUSSION

Emission factors of N₂O associated with the application of chemical fertilizer (EF_F)

The EF _F estimated in this study showed high variability, ranging from 1.30 to 5.50% (Table 2); these values were larger than the 1.0% proposed by the Intergovernmental Panel on Climate Change (2006) and the values reported in other studies (Akiyama and Tsuruta 2002; Smith et al. 1998; Zou et al. 2005). In this study, the EF _F value was calculated by dividing the difference in N ₂ O emissions between the FOP and OP treatments by the chemical nitrogen fertilizer application rate (Eq. 3). Because the amounts of applied chemical nitrogen fertilizer each year were nearly the same at GL, the annual variability in EF _F was likely to be affected by the difference in the N ₂ O emissions in the FOP and OP treatments

Figure 4  Seasonal variation in soil concentration and soil concentration at the Gray Lowland soil (GL) and Brown Lowland soil (BL) sites. See Fig. 1 for an explanation of the arrows and treatments.

Table 2 Nitrogen application, N2O emission, EFF and EFO in Gray Lowland soil (GL) and Brown Lowland soil (BL)

Year	Site	Chemical N (kg N ha^−1 year^−1)	Residue N (kg N ha^−1 year^−1)	Organic N (kg N ha^−1 year^−1)	N2O emission				EFF (%)	EFO (%)
					FOP	F	OP	C
2000	GL	242	86.9	0	16.7	8.78	9.90	4.44	2.79	9.06
2003		242	81.0	0	7.71	5.72	3.24	2.01	1.84	2.45
2004		242	92.3	0	17.3	16.7	4.00	7.34	5.50	0.64
2005		237	117	0	7.19	13.3	4.06	12.1	1.32	–5.22
2005	BL	284	98.7	82.5	5.56	2.83	1.88	1.81	1.30	1.51
FOP, chemical nitrogen fertilization and organic matter application, with plants; F, chemical nitrogen fertilization only, without plants; OP, organic matter application, with plants; C, no fertilization or organic matter, no plants; EFF, emission factor of N2O associated with the application of chemical nitrogen fertilizer; EFO, emission factor of N2O associated with the application of organic nitrogen.

after fertilizer application, especially from May to June. During these months, N ₂ O flux increased in the FOP and F treatments and N ₂ O-N/NO-N decreased to below 1.0 at GL and BL (Figs 1,2). Bouwman (1990) summarized the reports written by Anderson and Levine (1986) and Lipschultz et al. (1981) and reported that a value of N ₂ O-N/NO-N below 1.0 indicates that N ₂ O was produced in the soil mainly by the nitrification process, whereas a value above 100 indicates that denitrification is the dominant process for N ₂ O

Figure 5  Relationship between N2O emission and mean temperature and precipitation at the Gray Lowland soil (GL) site. (a) Relationship between the 2-month N2O emission and mean annual air temperature in the chemical nitrogen fertilization and organic matter application, with plants (FOP) and the chemical nitrogen fertilization only, without plants (F) treatments from May to June, (b) relationship between the 2-month N2O emission and precipitation in the FOP and the organic matter application, with plants (OP) treatments from September to October.

production. Therefore, during this season in the FOP and F treatments at both GL and BL, N ₂ O was likely to be emitted as a result of nitrification of the applied chemical nitrogen. Similar results were reported for GL by Kusa et al. (2002). In May and June, there was no clear relationship between N ₂ O flux and soil temperature (Fig. 3) or soil NH⁺ ₄ concentration (Fig. 4) in the FOP and F treatments at GL and BL. Because NH⁺ ₄ concentration was extracted by distilled water, we might not observe clear relationships between N ₂ O flux and soil NH⁺ ₄ concentration. However, there was a significant positive relationship between the 2-month N ₂ O emission and the 2-month mean air temperature (Fig. 5a). Therefore, N ₂ O production resulting from the nitrification process may be affected by temperature. Figure 6 also shows that EF _F was significantly correlated with mean annual temperature, although further research is required to fully understanding this relationship.

Figure 6  Relationships between emission factors (EF) and climatic conditions. (a) Relationship between the emission factor of N2O associated with the application of chemical nitrogen fertilizer (EFF) and mean annual air temperature, (b) relationship between the emission factor of N2O associated with the application of organic nitrogen (EFO) and mean annual relative humidity. Regression analysis carried out only on the Gray Lowland soil (GL) data.

Emission factors of N₂O associated with applied organic matter (EF_O)

In this study, we believe that the effect of nitrogen uptake by plants to the N₂O emission in FOP and OP treatments may be small. Variation in nitrogen uptake by plants usually increases from June to July (Hayashi and Hatano 1999). In 2000 and 2003–2005, variation in nitrogen uptake by plants showed the same pattern as previous studies (data not shown). Although nitrogen uptake by plants increased after June, the variation in nitrate concentration in soil in the F treatment changed in a similar way to FOP. Moreover in the C treatment, soil nitrate concentration also changed in a similar way to OP (Fig. 4). It is clear that the plants assimilated the nitrogen in FOP and OP. However, we do not understand why the concentrations of nitrate in non-planted treatments did not differ from the planted treatments.

Figure 7  Estimated versus the measured N2O emissions in the chemical nitrogen fertilization and organic matter application, with plants (FOP) treatment. •, Gray Lowland soil (GL); ○: Brown Lowland soil (BL).

At GL and BL, EF_O values ranged from −5.22 to 9.06% (Table 2). At GL, the coefficient of variation (CV; calculated from data in Table 2) of the amount of the N in the applied organic matter was 16% and the CV of the N₂O emission in the FOP treatment was 45%. Therefore, change in N₂O emission may have had a greater effect on the fluctuation of EF_O than did the variation in applied organic matter. In this study, we calculated the EF_O based on the difference in N₂O emissions in the FOP and F treatments (Eq. 4). The EF_O appeared to be strongly affected by the N₂O emission from the end of August to the end of October because N₂O emission originating from chemical fertilizer was released from May to June (Figs 2,4). The N₂O-N/NO-N in the FOP treatment was approximately 100 from early September to the end of October, and during this season N₂O was likely to be produced mainly by the denitrification process (Bouwman 1990). Similar results were reported by Kusa et al. (2002, 2006). From September to October we found no correlations between N₂O flux and soil environmental factors, such as soil temperature, WFPS or soil NH⁺ ₄ and NH⁻ ₄ concentrations, although a positive correlation between N₂O flux and WFPS has been reported in previous studies (e.g. Dobbie and Smith 2003). When the WFPS is approximately 60%, N₂O might be produced in soil (Granli and Bockman 1994). In this study, we found no relationship between N₂O flux and WFPS, and the reason for this is unclear. However, from September to October, the 2-month N₂O emission and 2-month precipitation tended to be positively related, indicating that the time lag between N₂O production and diffusion disappeared (Fig. 5b). This relationship between 2-month N₂O emission and 2-month precipitation from September to October indicated that N₂O production was affected by the water conditions of the soil. In addition, EF_O and annual mean relative humidity tended to have a positive relationship, with EF_O values increasing under wetter climatic conditions (Fig. 6b). CO₂ flux from soil is seen as an indicator of microbial activity and is affected by temperature, substrate and water conditions. Gulledge and Schimel (1998) reported that microbial respiration increased as the ratio of water to the water holding capacity (10–60%) increased. In addition, Linn and Doran (1984) and Gulledge and Schimel (1998) reported that CO₂ production by microorganisms showed maximum production at 60% WFPS. In this study, because WFPS increased to approxiately 60% from August to October, restriction of the water condition might be lifted and microbial activity, including nitrifier and denitrifier activity, might become high over this period. Therefore, wetter conditions increased the EFo because wetter conditions might not only lead to increases in microbial activity and the supply of nitrogen to the nitrifier and denitrifier, but also to the production of more N₂O in the denitrification process.

The EF_O differed between GL and BL (−5.22% and 1.51%, respectively) in 2005. One reason for this might be a difference in the denitrification activities between sites GL and BL. From August to September, N₂O flux and N₂O-N/NO-N did not increase at BL, whereas these values did increase at GL, indicating that the denitrification activity at BL might be weaker than that at GL (Fig. 2). Because organic matter was needed for the denitrification process, stronger denitrification activity at GL would reduce more N₂O to N₂ in the FOP and OP treatments than F and C treatments. At BL, however, weaker denitrification activity would produce mainly N₂O in the denitrification process in all treatments. Furthermore, the difference in the carbon content of the soil between sites GL and BL might affect the denitrification activities. This explains why the N₂O emission in the FOP and OP treatments at GL was lower than that in the F and C treatments, resulting in a negative EF_O value. Another reason may be the effect of microorganisms in the rhizosphere on N₂O production in the FOP and OP treatments. However, we could not confirm this effect because root cutting did not influence N₂O fluxes in this study. A further reason might be a difference in the application methods and amounts of organic matter between the sites. Only onion residue was applied to the GL field, whereas at BL half of the applied organic matter was organic fertilizer. However, these effects on the EF_O values could not be determined in the present study.

Estimation of N₂O emission

The N₂O emission estimated using Eq. 2 coincided well with the measured values (Fig. 7). Based on data from Dobbie et al. (1999), Flynn et al. (2005) estimated the N₂O emission from agricultural soils in Scotland using chemical nitrogen fertilizer application rate and the EF as a function of average monthly air temperature or monthly precipitation. In addition, Dobbie and Smith (2003) reported that the EF changed with crop type: in grassland EF ranged from 0.4 to 6.5%, while in upland it ranged from 0.5 to 1.5%. Huang et al. (2004) also reported that the EF for N₂O emissions that originated from residues was positively affected by the residue C/N ratio. Therefore, it should be noted that our results may be specific to onion fields. Moreover, in this study there was variation in the N₂O emission from the C treatment (Table 2). The effect of previous cultivation or management of the C plot on N₂O emission was not clear. Therefore, further study of the variation of N₂O emission from bare soil is necessary to more accurately estimate N₂O emissions from agricultural soils based on EF values (Akiyama et al. 2006). In addition, further research into the relationships between EF and meteorological conditions in different land uses would enable estimation of N₂O emission at a regional scale and would lead to a better understanding of temporal and spatial variations in N₂O emissions.

Conclusions

N₂O emissions from soil cultivated with onion showed large temporal variability. In this study, we examined the emission factors for N₂O emissions from plots receiving applied chemical nitrogen fertilizer (EF_F) or applied organic matter (EF_O) and N₂O emissions from soil organic matter (control, bare soil). The EF_F estimated based on 4 years of measurements in two soil types ranged from 1.3 to 5.5%, which is larger than the emission factor recommended by the Intergovernmental Panel on Climate Change (1.0%). In contrast, the EF_O ranged from −5.22 to 9.06%. The two soil types had similar EF_F values, but different EF_O values. N₂O emission derived from applied chemical nitrogen fertilizer appeared to be influenced by nitrification, whereas that derived from applied organic matter appeared to be influenced by denitrification. The EF_F was correlated significantly with mean annual temperature and there was a tendency for higher mean annual relative humidity to increase EF_O. Thus, it is important to consider denitrification activity in soil when estimating N₂O emissions derived from applied organic matter as well as N₂O emission associated with soil organic matter decomposition.

ACKNOWLEDGMENTS

This paper was presented at the International Workshop on Monsoon Asia Agricultural Greenhouse Gas Emissions (MAGE-WS), 7–9 March 2006, Tsukuba, Japan. This work was financially supported by the Global Environmental Research Program of the Ministry of the Environment of Japan (No. S-2). The authors thank Takeshi Morimoto and Masahiro Yamazaki for their permission to use their onion fields for this study.