Effects of environmental factors on temporal variation in annual carbon dioxide and nitrous oxide emissions from an unfertilized bare field on Gray Lowland soil in Mikasa, Hokkaido, Japan

Abstract

Soil is an important source of atmospheric carbon dioxide (CO₂) and nitrous oxide (N₂O). Studies of CO₂ and N₂O emissions from bare soil may explain annual changes in carbon (C) in soil organic matter (SOM) and help analyze N₂O production from SOM. Therefore, CO₂ and N₂O emissions associated with the decomposition of SOM from bare soil are important factors for assessing the C budget and N₂O emission in agricultural fields. We conducted a study over 7 years to assess the controlling factors of CO₂ and N₂O emissions from unplanted and unfertilized soil in Mikasa, Hokkaido, Japan. The CO₂ flux increased in summer and there were significant positive correlations between the CO₂ flux and soil temperature in the first 4 years. However, apparent relationships between CO₂ flux and water-filled pore space, soil NH₄ and NO₃ concentrations were not observed. The slope of monthly CO₂ emission against mean monthly temperature was positively correlated with monthly precipitation. These results suggest that the response of CO₂ production to increases in soil temperature became more sensitive in wet soils. The average CO₂ emission during the study period was 2.53 Mg C ha⁻¹ year⁻¹, and uncertainty in the annual CO₂ emission was 24%. Annual precipitation explained the yearly variation (CO₂ emission [Mg C ha⁻¹ year⁻¹] = 0.0021 × annual precipitation [mm year⁻¹] −0.0499, R = 0.976, P < 0.001). Nitrous oxide flux increased from July to October and was positively correlated with CO₂ flux. Based on the ratio of N₂O-N : NO-N of fluxes, N₂O appeared to be the main product of denitrification. The average N₂O emission over the study period was 4.88 kg N ha⁻¹ year⁻¹, and uncertainty in the annual N₂O emission was 58.5%. Strong relationships between the monthly emissions of CO₂ and N₂O suggest that N₂O production by denitrification is strongly affected by SOM decomposition. Unlike CO₂ emission, a relationship between N₂O emission and precipitation was not observed because of the multiple pathways of nitrification and denitrification for N₂O production induced by SOM decomposition.

Key words:

• carbon dioxide
• denitrification
• nitrous oxide
• soil organic matter decomposition
• temporal variation

Introduction

The contributions of atmospheric carbon dioxide (CO₂) and nitrous oxide (N₂O) to global warming are reported to be 60 and 6%, respectively. Since the Industrial Revolution, concentrations of these greenhouse gases in the atmosphere have increased at a rate of 1.4 p.p.m. year⁻¹ and 0.8 p.p.b. year⁻¹, respectively (Intergovernmental Panel on Climate Change 2007). Carbon (C) emissions that have originated as a result of changes in land use from 1980 to 1989 contribute 24% of the global annual CO₂ emission. Similarly, the emission of N₂O originating from agricultural fields in 1989 contributed 24% of the total global N₂O emission (Mosier et al. 1998). Recently, annual N₂O emissions from fertilized cropland and grassland at a global level have been estimated to be 3.3 and 0.8 Tg N year⁻¹, respectively (Stehfest and Bouwman 2006).

A number of studies have reported that there is a reduction in C from agricultural soils because microorganisms decompose soil organic matter (SOM) and emit CO₂ (Hu et al. 2004; Koga et al. 2006; Koizumi et al. 1993; Mu et al. 2006, 2008; Shimizu et al. 2009). A reduction in C from soil has been estimated by the difference in soil C over 20 years in a 0–30 cm soil surface (Intergovernmental Panel on Climate Change 2006; Paustian et al. 1997). In recent years, the global warming potential (GWP) has been used to evaluate the effect of agricultural activities on global warming (Chu et al. 2007; Jones et al. 2006; Koga et al. 2006; Mosier et al. 2005; Mu et al. 2006; Robertson and Grace 2004; Six et al. 2004). Therefore, it is necessary to estimate the annual reduction in soil C to evaluate the annual effect of agricultural activities on global warming. However, annual reductions in soil C are difficult to detect from investigations of annual change in the amount of soil C, because soil C is usually very large compared with C emission from the soil. In agricultural fields, decomposition of SOM can be measured as the CO₂ emission from bare fields in which root respiration is excluded (Hanson et al. 2000; Hu et al. 2004; Mu et al. 2006, 2008; Shimizu et al. 2009; Subke et al. 2006). As the decomposition of SOM results from microbial activities, it is influenced by chemical and physical conditions, such as soil temperature, water conditions and pH. As a result of spatial variability in these factors (Yanai et al. 2003), it is difficult to investigate the factors affecting the decomposition of SOM at a field level. Understanding the factors affecting the decomposition of SOM, quantifying the annual CO₂ emission from bare fields and clarifying the factors affecting temporal variations in annual CO₂ emissions from an unfertilized bare soil are needed to estimate C loss from soil. An understanding of these factors will provide beneficial information for the study of the effect of agricultural activities on C cycle or global warming over the long term for eco-balance or life cycle analyses (Kimura et al. 2007; Koga et al. 2006).

Nitrous oxide is produced by nitrification and denitrification processes in the soil. Therefore, N₂O emission is affected by the chemical and physical conditions of the soil (Bremner 1997; Colbourn and Dowdell 1984; Mosier et al. 1998; Stehfest and Bouwman 2006). Because ammonium (NH₄ ⁺) is used in the nitrification process and nitrate (NO₃ ⁻) is used in the denitrification process, N₂O emission from agricultural fields is usually in proportion to the N application rate (Bouwman 1996). Based on this, methods of estimating N₂O emissions from agricultural fields have been proposed (Intergovernmental Panel on Climate Change 2006). In Tier 1 and 2 of the Intergovernmental Panel on Climate Change (IPCC) Guidelines for National Greenhouse Gas Inventories (Intergovernmental Panel on Climate Change 2006), the emission factor (EF) is the ratio of N-induced N₂O emission, which is the difference between N₂O emissions from N-fertilized and unfertilized fields, divided by the amount of applied N fertilizer. However, to calculate original EF values, the N₂O emission originating from soil organic N should be considered because N₂O emission from the soil in agricultural fields is usually composed not only of N₂O emissions originating from the applied N, but also of N₂O emissions induced by decomposition of SOM. The N₂O emission from bare fields is regarded as the emission originating from the soil organic N. Several studies have reported annual N₂O emissions from unfertilized bare fields (Akiyama et al. 2006; Clayton et al. 1997; van Groenigen et al. 2004; Kamp et al. 1998; Koga et al. 2004; Zou et al. 2005). In addition, temporal variation in N₂O emission from agricultural fields has been reported in several studies (Drury et al. 2006; Kusa et al. 2002; Takakai et al. 2006; Zou et al. 2005). These studies, however, have not focused on the factors that affect N₂O flux or annual N₂O emission from SOM. Very few studies have focused on analyzing the mechanisms of N₂O production and emission induced by SOM decomposition in agricultural bare fields over the long term. Knowledge about the controlling factors of N₂O flux and annual emissions will be useful for long-term studies examining the effect of agricultural activities on the C cycle or global warming.

The objective of the present study was to clarify the controlling factors of CO₂ and N₂O emissions from agricultural bare soil based on 7 years of monitoring in central Hokkaido, Japan.

Materials and methods

Site description

The present study was conducted in Mikasa, central Hokkaido, Japan (43°14.4′N, 141°50′E). The soil type is Gray Lowland soil (Gleysol; FAO/UNESCO). The mean annual temperature at the study site is 7.5°C and the mean annual precipitation is 1,164 mm, of which 32% is in the form of snow. The soil does not freeze during winter. The temperature usually increases in August and precipitation increases from July to October (Fig. 1). At a depth of 0–10 cm, the soil pH (H₂O) was 5.8 and the cation exchange capacity was 25.5 cmolc kg⁻¹. The soil was comprised of sand (12.1%), silt (51.2%) and clay (36.7%). The concentrations of soil C and N were 32.1 and 2.8 g kg⁻¹, respectively (Toma and Hatano 2007) and the C:N ratio was 11.5 at a depth of 0–10 cm in 1996. Soil C density was recorded as 108 Mg C ha⁻¹ at a depth of 0–30 cm.

Figure 1 Seasonal variations in (a,b) monthly precipitation and (c,d) mean monthly air temperature.

Experimental design

We set up an unfertilized bare plot (5 m × 8 m) inside a 2-ha onion field owned by a farmer in Mikasa in 1999 and conducted monitoring in 2000, from 2002 to 2005, and in 2007. In 2006, the unfertilized bare plot was shifted to another location in the same field by the farmer owing to unavoidable circumstances. Conventional management practices for onion cultivation were carried out on the studied plot every year, although chemical and organic fertilizers or residues were not applied. Most weeds were removed from the monitoring plot regularly. Onion is usually cultivated from May to September in Mikasa. The soil is usually tilled before fertilization and plowed to a depth of 15 cm at the end of April. Onion roots are cut for harvest from mid to late August. The onion harvest is followed by plowing, which is carried out to incorporate the residues into the soil from the end of September to early October.

Measurement of CO₂, N₂O and NO fluxes

Fluxes of CO₂, N₂O and NO were measured using a closed-chamber method (three or four replicates) using two types of cylindrical stainless steel chambers from 10.00 to 14.00 hours. In 2000, chambers measuring 30 cm in diameter and 35 cm in height were used (Kusa et al. 2002, 2006); chambers measuring 20 cm in diameter and 25 cm in height were used from 2002 to 2007. The cover of the chamber was made of acryl and was equipped with a sample collector, a pressure regulating bag and a Tedlar bag (0.5 L). In 2000 and from 2002 to 2004, we inserted the chambers directly into the soil to a depth of 2 cm and began taking measurements after 15 min. However, in 2005, 2006 and 2007, we used a chamber-base made of stainless steel with a diameter of 20 cm. The upper part of the chamber-base had a slight depression, which was filled with water to seal it during the measurement (Toma and Hatano 2007; Toma et al. 2007). The base was kept on the ground, except during plowing and was reset 1 day before the next sampling.

Gas samples were taken at time 0 min from inside the chamber and 6 min (for CO₂) or 15 min (for N₂O and NO) after closing the chamber (Nakano et al. 2004; Toma and Hatano 2007). Using a 25 mL syringe, 10 gas samples were taken (total volume of 250 mL) and each gas sample was injected into a Tedlar bag (0.5 L). The bags were then taken to the laboratory and 20 mL of each gas sample was immediately transferred into glass vials (10 mL). From the samples in the bags, CO₂ and NO were analyzed using an Infrared CO₂ Analyzer (Model ZFP5YA3I; Fuji Electric, Tokyo, Japan) and a Chemoluminescence Nitrogen Oxide Analyzer (Model 265P; Kimoto Electric, Osaka, Japan), respectively. From the samples in the vials, N₂O was analyzed with an ECD Gas Chromatograph (Model GC-14B; Shimadzu, Kyoto, Japan).

The gas fluxes were calculated using the following equation:

where, F is the flux (mg N or C m⁻² h⁻¹), ρ is the gas density (ρN₂O-N = 1.259 × 10⁶, ρCO₂-C = 0.536 × 10⁶ and ρNO-N = 0.625 × 10⁶ mg 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; Toma and Hatano 2007).

A positive value of F indicates gas emission from the soil to the atmosphere and a negative value indicates gas uptake by the soil from the atmosphere. Considering the precision of the machinery, N₂O flux values ranging from −6 to 6 μg N m⁻² h⁻¹, CO₂ flux values from −3.7 to 3.7 mg C m⁻¹ h⁻¹ and NO flux values from −0.2 to 0.2 μg N m⁻² h⁻¹ were regarded as 0 μg N or C m⁻² h⁻¹. Annual CO₂ and N₂O emissions were calculated assuming linear changes between two sampling occasions.

The value of Q₁₀, which is a relative increase in CO₂ flux for a 10°C change in soil temperature, was calculated based on the Arrhenius equation and the following equation (Hu et al. 2004):

where, k is the CO₂ flux, A is a constant, E is the activation energy, R is the universal gas constant (8.3144 J mol⁻¹ K) and T is the absolute temperature (K). The logarithm of the CO₂ flux and the reciprocal of the absolute temperature of the soil were plotted, and E was determined from the slope. The value of Q₁₀ was calculated using T = 283 (K) because mean monthly air temperatures at the study site ranged from 10 to 20°C (Fig. 1c,d).

Measurement of other variables and the frequency of sampling

Soil temperature at a depth of 5 cm was measured five times at the same time as the CO₂, N₂O and NO fluxes were measured. Five replicate disturbed soil samples (0–5 cm depth) were taken and combined to form one composite sample. Soil solutions were extracted using distilled water throughout the study period and NH₄ ⁺-N, NO₃ ⁻-N and water-soluble organic carbon (WSOC) concentrations were analyzed by colorimetry with indophenol-blue, ion chromatography (QIC Analyzer, Dionex, Osaka, Japan) and a total organic carbon analyzer (Model TOC-5000A; Shimadzu), respectively. In 2007 the soil NH₄ ⁺-N concentration extracted by distilled water was not measured, and in 2006 and 2007 the soil NH₄ ⁺-N concentration extracted with 2 mol L^–1 KCl was analyzed. The WSOC was not measured in 2000. Three replicate undisturbed soil samples were collected using a steel cylinder (100 mL) and the water-filled pore space (WFPS) was measured. Daily or annual meteorological data were obtained from the local Iwamizawa Weather Station (43°12.6′N, 141°47.3′E).

After plowing, all 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-cover season. During the period from snowmelt to onion cultivation and from mid October to snow cover, sampling was conducted two or three times per month.

Statistical analyses

Relationships between CO₂ and N₂O fluxes and soil physical and chemical properties, or annual CO₂ and N₂O emissions and mean annual air temperature or annual precipitation were analyzed using linear or quadratic curve regressions. Variations in annual CO₂ or N₂O emissions during the study period were evaluated using Tukey–Kramer tests. A least significant difference test was used to determine significant differences (P < 0.05).

Uncertainties were calculated using the following equation:

The two-sided 95% confidence intervals of the CO₂ and N₂O emissions were calculated using the following equation:

where, d.f. is the degrees of freedom and t(d.f., 0.05) is the t value at P < 0.05 with a two-sided alternative.

Results

Seasonal variation in CO₂ and N₂O fluxes

The CO₂ flux peaked from August to October (Figs 2a, 3a), whereas the N₂O flux increased from the end of July to October in all years (Figs 2b,3b). Soil temperature increased in August and this trend was similar to that of air temperature (Fig. 4a,f). The WFPS was high in early April and November, and low in August (Fig. 4b,g). From August to October, the WFPS was as low as approximately 50%. Soil NH₄ ⁺-N concentrations extracted by distilled water ranged from 0 to 7.6 mg N kg⁻¹ each year and did not exhibit any consistent seasonal pattern (Fig. 4c,h). Although the NH₄ ⁺-N concentration extracted with 2 mol L^–1 KCl in 2006 and 2007 did not show a distinct seasonal pattern (Fig. 4i), the range of concentrations (1.67–11.3 mg N kg⁻¹ in 2006, 0.84–29.9 mg N kg⁻¹) was higher than the values obtained through extraction with water. Soil NO₃ ⁻-N concentrations increased from August to September each year (Fig. 4d,j). Fluctuations in soil WSOC concentrations were small throughout the year (Fig. 4e,k).

Correlation coefficients between the CO₂ or N₂O fluxes and soil physical and chemical properties are given in Table 1. The CO₂ flux was significantly and positively correlated with soil temperature from 2000 to 2004 (Table 1), and reached its maximum at 48.8% of WFPS (Fig. 5a). Although the N₂O flux was significantly and positively correlated with soil temperature in 2000 and 2004, N₂O flux was significantly and positively correlated with CO₂ flux from 2000 to 2006. Carbon dioxide and N₂O fluxes were significantly correlated with soil NO₃ ⁻-N concentration, and N₂O flux was significantly correlated with soil temperature in 2000 and 2004. However, no significant correlation was observed between CO₂ or N₂O fluxes and soil chemical or physical properties over the study period. There were no significant correlations between N₂O flux and WFPS over the study period except for 2003, but N₂O showed high fluxes when WFPS ranged from 30 to 60% and reached its maximum at 48.4% (Fig. 5b).

Figure 2 Seasonal variations in (a) CO2, (b) N2O and (c) NO fluxes in 2000, 2002 and 2003. Arrows indicate the timing of plowing. sd, standard deviation.

Figure 3 Seasonal variations in (a) CO2, (b) N2O and (c) NO fluxes from 2004 to 2007. Arrows indicate the timing of plowing. sd, standard deviation.

The relationship between N₂O flux and the N₂O-N:NO-N ratio is given in Fig. 6. The N₂O flux increased with an increase in the N₂O-N:NO-N ratio, with 18 out of 49 samples of N₂O flux over 0.079 mg N m⁻² h⁻¹, which resulted in a ratio of N₂O-N:NO-N over 100. The average N₂O flux was 0.079 mg N m⁻² h⁻¹. Therefore, approximately 37% of the N₂O fluxes were higher than the average value.

Cumulative CO₂ and N₂O emissions

Table 2 shows the annual CO₂ and N₂O emissions and Q₁₀ values. Annual CO₂ emissions ranged from 2.04 to 3.32 Mg C ha⁻¹ year⁻¹ and were significantly correlated with annual precipitation (Fig. 7; y = 0.0021x −0.0499, R = 0.98, P < 0.01). The average CO₂ emission during the study period was 2.53 Mg C ha⁻¹ year⁻¹ and uncertainty in the annual CO₂ emission was 24%. The Q₁₀ values ranged from 1.11 to 2.38. There was no significant correlation between the Q₁₀ value and annual precipitation (y = 0.0012x + 0.1918, R = 0.56, P = 0.11). However, the slope of the monthly CO₂ emission against mean monthly temperature from April to October was significantly correlated with precipitation from April to October (Fig. 8; y = 0.0688x − 27.67, R = 0.90, P < 0.01). Carbon dioxide flux in the present study was measured only during the daytime, and annual CO₂ emission was calculated assuming linear changes between two sampling occasions. That is, the decrease in CO₂ flux during the night, when the temperature decreased, was not considered when calculating the annual CO₂ emission from 2000 to 2004. The difference in the maximum and minimum air temperatures from July to September, which was the period of high air temperature, was less than 10°C (8.8°C). In addition, the average Q₁₀ value from 2000 to 2004 was 1.99 (Table 2). If we assume that the CO₂ flux during the maximum daily air temperature is 1.99-fold higher than the CO₂ flux during the minimum daily air temperature, the measured annual CO₂ emission would be 1.33-fold higher than the annual CO₂ emission that was calculated by considering daily change in the CO₂ flux.

Figure 4 Seasonal variations in (a,f) soil temperature, (b,g) water-filled pore space (WFPS), (c,h,i) soil NH4+-N, (d,j) NO3−-N and (e,k) water-soluble organic carbon (WSOC) concentrations.

Annual N₂O emission ranged from 1.62 to 12.1 kg N ha⁻¹ year⁻¹. The average N₂O emission during the study period was 4.88 kg N ha⁻¹ year⁻¹ and uncertainty in the annual N₂O emission was 58.5%. The uncertainties in yearly variation in CO₂ and N₂O emissions were 17.9 and 76.1%, respectively. Although correlations between N₂O emission and annual air temperature or precipitation were not significant, N₂O emission increased with an increase in CO₂ emission (Fig. 9). Significant differences in annual CO₂ and N₂O emissions among the study years indicate yearly variation in CO₂ and N₂O emissions at the study site (Table 2).

Discussion

Key factors determining the CO₂ flux in bare soil

Several studies have reported that CO₂ flux from soil shows an exponential increase with an increase in soil temperature (Boone et al. 1998; Hu et al. 2001; Jones et al. 2006; Schindlbacher et al. 2009). In the present study, there were significant positive correlations between CO₂ flux and soil temperature (in 2000 and in 2002–2004) (Table 1). In contrast, there was no significant correlation between CO₂ flux and WFPS, except in 2003 (Table 1). Gulledge and Schimel (1998) reported that microbial respiration was directly proportional to the water-holding capacity (10–60%). Moreover, Linn and Doran (1984) and Gulledge and Schimel (1998) reported that the maximum CO₂ production by microorganisms was at 60% WFPS. In our study, CO₂ flux peaked at 49.1% of WFPS. Therefore, soil moisture might not influence the production or emission of CO₂ linearly, and there might be an appropriate condition of soil moisture for CO₂ production in relation to soil ventilation. In contrast, significant positive correlations between annual CO₂ emission and annual precipitation and between the slope of monthly CO₂ emission against mean monthly temperature from April to October and precipitation from April to October suggested that the production of CO₂ was greater in wet conditions even if temperatures were similar (Figs 7, 8). Therefore, CO₂ production increased with an increase in soil temperature, but CO₂ production at the same temperature was enhanced in wet conditions. High moisture content prevents soil gas from diffusing from the soil to the atmosphere. This means that even if CO₂ was produced in the soil under appropriate conditions for CO₂ production, CO₂ might not necessarily be diffused at that time. There could be possibilities of high CO₂ flux to be detected during the season of high precipitation. Differences in the timing between CO₂ production and diffusion from the soil to the atmosphere could possibly be explained by the lack of correlation between CO₂ flux and WFPS and by a significant positive correlation between annual CO₂ emission and precipitation (Table 1; Fig. 7).

Table 1 Correlation coefficients for the relationships between CO2 or N2O fluxes and soil physical and chemical properties

Year	Soil temperature† (°C)		WFPS (%)		NH4 ^+-N (H2O) (mg N kg^−1)		NH4 ^+-N (KCl) (mg N kg^−1)		NO3 ^−-N (mg N kg^−1)		WSOC mg N kg^−1		CO2 flux mg C m^−2  h^−1
	n	R	n	R	n	R	n	R	n	R	n	R	n	R
CO2 flux (mg C m^−2 h^−1)
2000	19	0.54*	19	−0.17	19	−0.25	ND	ND	19	0.56*	ND	ND	–	–
2002	51	0.37*	43	0.09	33	−0.47**	ND	ND	33	0.08	33	0.06	–	–
2003	34	0.68**	30	−0.51**	13	0.56*	ND	ND	13	0.66*	13	−0.27	–	–
2004	26	0.68**	24	−0.34	18	−0.35	ND	ND	18	0.71**	18	−0.08	–	–
2005	24	0.35	24	0.09	24	−0.37	ND	ND	24	0.33	22	0.37	–	–
2006	27	0.31	29	0.22	29	−0.05	29	0.15	28	0.10	29	0.13	–	–
2007	25	0.13	21	−0.11	ND	ND	21	−0.22	22	0.12	22	−0.60*	–	–
N2O flux (mg N m^−2 h^−1)
2000	22	0.54**	19	−0.19	22	−0.23	ND	ND	22	0.68**	ND	ND	23	0.59**
2002	51	0.26	43	−0.11	33	−0.30	ND	ND	33	0.04	33	0.08	60	0.74**
2003	34	0.25	30	−0.42*	13	0.52	ND	ND	13	0.36	13	−0.34	39	0.51**
2004	26	0.50**	24	−0.32	18	−0.36	ND	ND	18	0.78**	18	0.03	29	0.71**
2005	24	0.19	24	0.28	24	−0.28	ND	ND	24	0.14	22	0.49*	28	0.79**
2006	28	0.21	29	0.09	29	−0.01	29	0.10	28	0.23	29	−0.10	32	0.55**
2007	24	−0.11	21	0.23	ND	ND	21	−0.20	22	−0.37	22	−0.37	25	0.27
*P < 0.05, **P < 0.01; ND, no significant difference; †Exponential curve regression analysis between CO2 flux and soil temperature. WFPS, water-filled pore space.

Figure 5 Relationships between (a) CO2 or (b) N2O fluxes and water-filled pore space (WFPS).

Figure 6 Relationship between N2O flux and the N2O-N:NO-N ratio.

Table 2 Annual CO2 and N2O emissions from unfertilized bare soil in an onion field in Mikasa

Year	Annual precipitation (mm year^−1)	Mean annual air temperature (°C)	Period day	CO2 emission (Mg C ha^−1 year^−1)	Uncer-tainty (%)	Q10	N2O emission (kg N ha^−1 year^−1	Unvcer- tainty (%)
2000	1576	7.7	302	3.32^a	14.2	2.38	4.60^bc	18.8
2002	1187	7.7	251	2.42^abc	16.9	2.03	3.94^bc	23.6
2003	986	7.4	318	2.04^c	20.7	1.80	2.01^c	74.0
2004	1294	8.4	280	2.48^bc	13.2	1.73	7.34^b	15.0
2005	1398	7.5	297	2.95^ab	21.2	1.53	12.1^a	19.0
2006	1124	7.8	271	2.25^bc	22.8	1.38	1.62^c	106
2007	1015	8.2	256	2.20^bc	58.9	1.11	2.56^c	153
Average	1226	7.81	282	2.53	24.0	1.71	4.88	58.5
Values within the same column with different lowercase letters differ significantly (P < 0.05). Q10, relative increase in CO2 flux for a 10°C change in soil temperature.

Figure 7 Relationship between CO2 emission and annual precipitation.

Figure 8 Relationship between the slope of the monthly CO2 emission against mean monthly temperature and precipitation from April to October.

Figure 9 Relationship between monthly CO2 emission and N2O emission from April to October.

Key factors determining N₂O flux from bare soil

Because N₂O is mainly produced by the processes of nitrification and denitrification in soil (Tiedje 1994), N₂O flux is often affected by soil NH₄ ⁺-N or NO₃ ⁻-N concentrations (Livesley et al. 2009; Thornton and Valente 1996; Toma et al. 2007). However, no apparent correlations between N₂O flux and soil NH₄ ⁺-N or NO₃ ⁻-N concentrations were observed in our study. Instead, we observed significant positive correlations between N₂O and CO₂ fluxes in all years, except 2007 (Table 1). Similar positive correlations between N₂O and CO₂ fluxes were observed when crop residues were mixed with surface soil (Huang et al. 2004; Toma and Hatano 2007). In the present study, the substrates for nitrification and denitrification originated from SOM because no fertilizer was applied in the experimental plot, suggesting that N₂O production was strongly related to SOM decomposition. Therefore, one of the factors controlling the production of N₂O could be the supply of C and inorganic N from SOM decomposition. A number of studies have reported that soil temperature is one factor that affects N₂O production (Kamp et al. 1998; Mori et al. 2005; Smith et al. 1998; Tokuda and Hayatsu 2004). As the CO₂ flux increased with an increase in soil temperature in our study (Table 1), soil temperature might affect not only the activity of N₂O production, but also the supply of substrates for N₂O production. Bouwman (1990) summarized the results presented by Anderson and Levine (1986) and Lipschultz et al. (1981) and reported that a N₂O-N:NO-N ratio below 1.0 indicates that N₂O was produced in the soil mainly by nitrification, whereas a value above 100 indicates that denitrification was the dominant process for N₂O production. In our study, approximately 37% of the N₂O fluxes that were higher than the average value occurred when the N₂O-N:NO-N ratio was above 100. In general, an increase in soil moisture contributes to N₂O production by denitrification (Linn and Doran 1984). Studies have shown that a WFPS of 50–60% is suitable for the decomposition of soil organic matter and that these conditions accelerate denitrification as a result of the consumption of oxygen (Maag and Vinther 1999). Moreover, N₂O is mainly produced by denitrification when the WFPS is 50–60% (Davidson et al. 2000). In our study, the WFPS increased from approximately 40 to 60% from the end of July to early October, and the N₂O flux was highest when the WFPS was at 46.5% (Figs 4,5). Sawamoto and Hatano (2000) reported that an increase in N₂O production in autumn might be caused by nitrate transportation and diffusion inside soil aggregates after rainfall and by the development of denitrification areas inside the aggregates of well-structured soil. Therefore, there is a possibility of aerobic and anaerobic conditions occurring simultaneously inside the aggregates and inter-aggregate pores in the well-structured soil examined in the present study. This would suggest that the high N₂O fluxes were induced not only by denitrification, but also by nitrification. Nitrous oxide was the main product of the denitrification process in the present study.

Nitrous oxide fluxes were significantly correlated with the CO₂ flux, and production of CO₂ was affected by soil temperature and precipitation. However, a significant correlation between N₂O flux or annual N₂O emission and WFPS or annual precipitation was not recorded. This might be because N₂O production in soil is a complex process. In the present study, the source of inorganic N might be mainly SOM. Therefore, NH₄ and NO₃, which are substrates for nitrification and denitrification, could have been produced by mineralization and nitrification processes in the soil. In a study of N₂O emission from organic fertilizer, similar correlations between N₂O and CO₂ emission were observed (Hayakawa et al. 2009). When N originates from organic matter, decomposition of the organic matter is probably essential for the production of N₂O. In any case, significant correlations between CO₂ flux and N₂O flux or monthly emissions of CO₂ and N₂O suggest that climate change, which will affect SOM decomposition, might greatly affect N₂O emission induced by SOM decomposition.

Annual CO₂ emission and its spatial and temporal variations

The annual CO₂ emission associated with decomposition of SOM ranged from 2.04 to 3.32 Mg C ha⁻¹ year⁻¹ (Table 2). The CO₂ emission values from 2000 to 2004, however, might have been overestimated by approximately 1.33-fold because daily variation in CO₂ flux was not considered. Therefore, if we consider that the CO₂ emission was overestimated then annual CO₂ emission ranged from 1.11 to 2.95 Mg C ha⁻¹ year⁻¹. Mu et al. (2006) reported that CO₂ emitted from SOM in bare treatments under several types of land use (e.g. wheat, maize, onion) ranged from 3.01 to 5.68 Mg C ha⁻¹ year⁻¹ in Hokkaido, Japan. Jacinthe et al. (2002) reported that 4.37 Mg C ha⁻¹ year⁻¹ of CO₂ was emitted from a bare plot in Ohio, USA. Decomposition of SOM in these two studies might have been overestimated similar to our study because they used the trapezoidal rule to calculate annual CO₂ emission. Koizumi et al. (1993) reported that heterotrophic respiration ranged from 7.16 to 10.5 Mg C ha⁻¹ year⁻¹ in three double-cropping agro-ecosystems (rice–barley, peanut–wheat and dentcorn–italian ryegrass). They calculated SOM decomposition as the difference between total soil respiration and plant respiration by calculating soil and root respirations using continuous data of soil temperature and correlations between soil temperature and soil or root respirations. Koga et al. (2006) reported average SOM decomposition over a 20-year time scale from 1981 to be 1.04 to 1.34 Mg C ha⁻¹year⁻¹ from the difference in total soil C within a depth of 20 cm in Hokkaido, Japan. The decomposition rate of SOM reported by Koizumi et al. (1994) and Koga et al. (2006) might not have been overestimated, and their values were larger and smaller than the values in our study, respectively. Differences in SOM decomposition may result from variations in location, cultivation system and soil type at the study site. The mean annual temperature (13.1°C) of the study site reported by Koizumi et al. (1994) was higher than that at our study site. In addition, there were frequent tillage operations because of a double-cropping system. High temperature and frequency of plowing may have increased SOM decomposition in the study reported by Koizumi et al. (1994). In contrast, the study site of Koga et al. (2006) was located at a similar latitude (42°53′N) to our study. Therefore, differences in SOM decomposition between our study and the study conducted by Koga et al. (2006) might result from differences in crop management tillage and/or soil type. Mu et al. (2008) reported that CO₂ emission from bare soil reached its maximum when the silt and clay content of the soil was 60–70%. Moreover, Bellamy et al. (2005) reported that the amount of C over 25 years decreased with an increase in soil C content. In these studies, however, Andisols were not included. Further study examining differences in SOM decomposition in various soil types is required.

Carbon dioxide emission was measured at the same location at our study site from 2000 to 2005. Over this period the average annual CO₂ emission was 2.84 Mg C ha⁻¹ year⁻¹, and 17 Mg C ha⁻¹ was estimated to be released from the soil in 2000–2005. Because the mass of C in the surface (0–30cm) soil was 108 Mg C ha⁻¹, a C loss of 17 Mg C ha⁻¹ corresponds to 15.7% of the surface soil C. We did not measure changes in soil C storage over the study period so the actual reduction in soil C could not be determined in the present study. If changes in soil C were measured in our study then the soil C mass could possibly have been detected. Bellamy et al. (2005), however, reported that loss of soil C was not detectable over 10 years when the soil C content was lower than 50 g kg⁻¹. Detecting changes in soil C at our study site might be difficult. There are several ways for C input in agricultural fields, including rainfall, weeds and algae. However, very little weed or algae existed during the study period. Carbon input from rainfall was not included in the IPCC guidelines for calculating soil C change (Intergovernmental Panel on Climate Change 2006). Thus, there might be no major C inflow that supplemented soil C loss by SOM decomposition. Further study is required to match the reduction in soil C and CO₂ emission from unfertilized bare soil.

Large variation in CO₂ fluxes and annual CO₂ emissions among replicates was observed owing to spatial variation in the CO₂ flux (Figs 2a,3a; Table 2). Average uncertainty in annual CO₂ emission (24%) was higher than of the uncertainty in annual CO₂ emission (17.9%). In general, the decomposition of SOM is affected by soil properties (Koizumi et al. 1993; Mu et al. 2008), method of cultivation (Koga et al. 2006) or land-use practices (Mu et al. 2006). Mu et al. (2008) reported a significant correlation between CO₂ emission and clay and silt content in a bare field in the same district as our study site. In the study field, spatial variations in total C content, the soil C:N ratio and microbial biomass C have been reported (Yanai et al. 2003). These factors possibly cause large uncertainty in the annual CO₂ emission. Using a larger chamber or taking more CO₂ flux measurements might be required to improve the accuracy of annual CO₂ emission values.

Significant differences in annual CO₂ emissions from the bare field during the study period show that there is yearly variation in annual CO₂ emission that is induced by SOM decomposition (Table 2). Large uncertainly (17.9%) in the average annual CO₂ emission during the study period shows that continuous monitoring of CO₂ emission is required to determine a representative value of annual CO₂ emission. This yearly variation in annual CO₂ emission suggests that climate will have a great impact on yearly CO₂ emission from SOM. Raich and Schlesinger (1992) reported that soil respiration, including root respiration, increased with an increase in annual precipitation or mean annual air temperature. Our study also showed that annual precipitation affects CO₂ emission from mineralization of SOM.

Previous studies have reported that the effect of chemical fertilizer application on microbial respiration is negligible (Ginting et al. 2003; Hu et al. 2004; Jacinthe et al. 2002). Hence, CO₂ emissions obtained in the present study could be an approximation of SOM decomposition in a fertilized field. Several studies have reported that soil C decreases in agricultural fields with low levels of organic matter application (Ginting et al. 2003; Hu et al. 2004; Jacinthe et al. 2002; Mu et al. 2008; Shimizu et al. 2009). The loss of soil C is reported to have a greater influence on global warming than the emissions of N₂O or CH₄ (Jones et al. 2006; Koga et al. 2006; Mosier et al. 2005; Mu et al. 2006). Although it is difficult to calculate an annual loss of C by measuring soil C reduction, it could be done by measuring CO₂ emission from bare soil in an agricultural field. Assessing CO₂ emissions from bare fields, using the correlation between CO₂ emission and available environmental factors such as meteorological data or soil type, will be useful for cultivation management or planning of farm activities at a regional scale to mitigate C loss from soil.

Annual N₂O emission and its spatial and temporal variations

Annual N₂O emission in the present study ranged from 1.62 to 12.1 kg N ha⁻¹ year⁻¹ (average 4.88 kg N ha⁻¹year⁻¹). These values are higher than most reported values for bare fields (1.00 kg N ha⁻¹ year⁻¹ [Intergovernmental Panel on Climate Change 2006], −0.02–0.13 kg N ha⁻¹year⁻¹ [Koga et al. 2004], 0.14–1.52 kg N ha⁻¹ year⁻¹ [van Groenigen et al. 2004] or 0.36–0.14 kg N ha⁻¹year⁻¹ [Akiyama et al. 2006], 4.80 kg N ha⁻¹ year⁻¹ [Kamp et al. 1998] or 4.25 kg N ha⁻¹ year⁻¹ [Zou et al. 2005]). This difference in annual N₂O emission might result from soil physical and chemical properties. The N content in the soil may affect N₂O production because there is no N source for a substrate of N₂O in bare soil except for the N in SOM. Klemedtsson et al. (2005) reported a significant negative relationship between N₂O emission and soil C:N ratio in forested peat soil in Europe, indicating that N₂O emission increased with an increase in relative N to C in peat. Nitrogen contents of the soils in the study sites of Kamp et al. (1998) (1.7 g kg⁻¹) and Zou et al. (2005) (1.1 g kg⁻¹), in which N₂O emissions exceeded 4.0 kg N ha⁻¹, were lower than the value recorded at our study site (2.8 g kg⁻¹). Production of N₂O in poorly drained soil increases as a result of anaerobic conditions in the soil (van Groenigen et al. 2004). The N₂O emission reported by Koga et al. (2004) was lower than the value recorded here even when the N content at their study site (2.4–3.1 g kg⁻¹) was similar to that at our site. This could possibly be because of the drainage quality of the soil. The soil type at the study site of Koga et al. (2004) was a well-drained volcanic ash soil. Therefore, N concentration as a source of substrate for N₂O and drainage conditions, which might control the aerobic and anaerobic conditions of the soil, might be an indicator of the potential N₂O source from unfertilized bare fields.

Large variations in flux and annual emission of both N₂O and CO₂ were observed in our study (Figs 1b,2b; Table 2). There is considerable uncertainty in the amount of N₂O emission from unfertilized bare fields (Akiyama et al. 2006; Zou et al. 2005). Both CO₂ and N₂O emissions are influenced by soil properties (van Groenigen et al. 2004) and cultivation methods (Koga et al. 2004). At our study site, high spatial variability (CV 217%) in the N₂O flux has been reported (Yanai et al. 2003). Yanai et al. (2003) also reported that soil organic matter and soil pH were the main soil-related determining factors for N₂O flux.

Temporal variation in N₂O was confirmed to be similar to that recorded in a number of previous studies (Drury et al. 2006; Kusa et al. 2002; Takakai et al. 2006; Zou et al. 2005). In our study, neither N fertilizer nor crop residue was applied to the experimental plot from 1999 to 2005. Therefore, it is expected that the substrate required for the production of N₂O had been reduced. However, annual N₂O emission did not consistently decrease from 2000 to 2005, resulting in large uncertainty in annual N₂O emission (Table 2). This suggests that the effect of climatic factors on N₂O emission might be greater than the reduction in substrate required for N₂O production. Annual N₂O emission tended to increase with an increase in annual precipitation. The relationship between annual N₂O emission and annual precipitation was not significant, whereas annual CO₂ emission was significantly correlated with annual precipitation (Fig. 7); this may have occurred because there is a difference in the environmental factors that are required for mineralization, nitrification and denitrification. In mineralization and nitrification processes, oxygen is required for oxidation of organic C or NH₄. However, anaerobic conditions are required for denitrification because NO₃ is used for electron capture (Bouwman et al. 2001). As N₂O was produced through these three biological processes in the soil, a correlation between annual N₂O emission and annual precipitation may not be significant.

Acknowledgments

This study 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 for allowing us to conduct the study on his farm.