Combined emission of CH₄ and N₂O from a paddy field was reduced by preceding upland crop cultivation

Abstract

Since crop rotation between paddy rice and upland crops is widely conducted in Japan and other Asian countries, the effect of crop rotation on greenhouse gas emission should be clarified. In this study, methane (CH₄) and nitrous oxide (N₂O) fluxes were simultaneously measured for two years from 2004 to 2005 in a paddy rice field with three different cultivation histories, i.e. consecutive paddy rice cultivation (PR), single cropping of upland rice (UR), and double cropping of soybean and wheat (SW) in the preceding two years from 2002 to 2003. In 2004, the cumulative CH₄ emissions in the UR and SW plots were 511 and 2817 g CH₄ m⁻² y⁻¹, which were 8 and 46%, respectively, of that in the PR plots (6092 g CH₄ m⁻² y⁻¹). In 2005, the cumulative CH₄ emissions in the UR and SW plots were 5123 and 1331 g CH₄ m⁻² y⁻¹, which were 87 and 23%, respectively, of that in the PR plots (5893 g CH₄ m⁻² y⁻¹), although the differences were not statistically significant. The soil reduction/oxidation potential (Eh) in the UR plots was higher than that in the PR plots in 2004. However, no distinctive differences in soil Eh among the three cropping systems were found in 2005. In the spring of 2004, the soil iron (Fe) content determined by extraction with dithionite-ethylenediaminetetraacetic acid (EDTA) solution was higher in the UR plots than in the PR and SW plots. However, no significant differences in Fe content among the three cropping systems were found in the spring of 2002 and 2005. The application of a relatively small amount of residue from the upland rice (c. 30% of that from the paddy rice) and the removal of all aboveground crop residues of soybean and wheat before paddy rice cultivation in 2004 could have contributed significantly to the low CH₄ emissions in the UR and SW plots. In addition, change in the form of soil Fe during the preceding periods with upland crop cultivation may also have been related to the decreases in CH₄ emission. The cumulative N₂O emissions ranged from 39 to 99 mg N m⁻² y⁻¹, and showed no significant difference among the three cropping systems in 2004 and 2005. These results indicate that the combined CH₄ and N₂O emission from paddy soil is reduced by the introduction of the preceding upland crop cultivation when crop residue from the previous upland crop is small or removed before paddy rice cultivation, although this effect was expected only for one year just after the land use change from upland crop cultivation to paddy rice cultivation.

Keywords:

• crop rotation
• methane
• nitrous oxide
• paddy
• upland.

Introduction

Methane (CH₄) and nitrous oxide (N₂O) are major greenhouse gases, which has 25 and 298 times, respectively, higher global warming potential than carbon dioxide (CO₂) in a time horizon of 100 years (Forster et al. 2007). Rice fields have been identified as a major source of atmospheric CH₄ (Smith et al. 2007) and therefore have been the subject of many previous studies. According to the results of these previous studies, water management and organic matter application are the two major factors that regulate CH₄ emission (e.g. Yagi and Minami 1990; Yagi et al. 1996; Yagi 2002). In general, CH₄ emission can be reduced by introducing midseason drainage. On the other hand, it tends to increase with an increase in organic matter application rate.

As for N₂O, its annual emission from paddy fields is generally low compared with that from upland crop fields. Field measurements of N₂O emission in the previous studies revealed that water management and nitrogen fertilization level are the major factors regulating N₂O emission during paddy rice cultivation (Nishimura et al. 2004). In general, N₂O emission tends to increase with the introduction of midseason drainage and with an increase in nitrogen fertilization rate. On the other hand, Akiyama et al. (2005) comprehensively summarized the data in previous peer-reviewed papers with field measurements of N₂O emission from paddy rice fields (113 measurements from 17 sites). According to their report, the background N₂O emission (N₂O emission from bare soil without fertilization) from paddy rice fields was estimated to be 1820 g N ha⁻¹ y⁻¹ on average, and additional fertilizer-induced N₂O emissions were estimated to be 341 ± 474 g N ha⁻¹ season⁻¹ for fields with continuous flood irrigation, and 993 ± 1075 g N ha⁻¹ season for fields with midseason drainage.

In Japan and other Asian countries, consecutive paddy rice cultivation has been conducted for a long time. Nowadays, the crop rotation of paddy rice and upland crops (paddy-upland crop rotation) is widely conducted, and various upland crops are cultivated in drained paddy fields (The Ministry of Agriculture, Forestry and Fisheries of Japan 2003). Drainage of paddy soil over the years in accordance with the introduction of upland crop cultivation may cause significant changes in various soil properties. The possible changes in soil properties include change in soil tillability (Takahashi et al. 1999), enhancement of aerobic decomposition of soil organic carbon and mineralization of soil organic nitrogen (e.g. Takahashi et al. 2003), oxidation of minerals in the soil such as iron (Fe) (Kyuma 2004), and change in the composition of soil microbial community (Chu et al. 2009). According to these possible changes in soil properties, the dynamics of CH₄ and N₂O may also change significantly even during the subsequent periods with paddy rice cultivation.

A number of studies have measured CH₄ and N₂O emissions also in fields with paddy-upland crop rotation. Field measurements of CH₄ and N₂O emissions from drained paddy soils for upland crop cultivation were conducted in some previous studies (e.g. Abao et al. 2000; Nishimura et al. 2005a, 2008; Yao et al. 2010). In addition, measurements of CH₄ and N₂O emissions during paddy rice cultivation period in fields with preceding upland crop cultivation were also conducted in some previous studies. For example, Abao et al. (2000) conducted a simultaneous measurement of CH₄ and N₂O fluxes in fields with alternate cropping of paddy rice and upland cowpea/wheat. They showed relatively low CH₄ emissions during the paddy rice cultivation period, ranging 707–3440 mg CH₄ m⁻² period⁻¹. Cai et al. (2000) measured CH₄ emission in paddy fields with triple cropping in the preceding year. They found that cumulative CH₄ emission decreased with an increase in the number of upland vegetable cultivations in the preceding year. Yan et al. (2005) summarized data on the CH₄ emission in paddy fields in various locations in China and proposed an equation for estimating CH₄ emission with various environmental factors as parameters. In this equation, the effect of ‘preseason water management' was taken into account, indicating the possible reduction in CH₄ emission with prolonged drainage before paddy rice cultivation. Yao et al. (2010) measured N₂O emission in fields with alternate cropping of paddy rice and upland wheat. They showed that relatively high N₂O emissions (1.6–7.8 kg N ha⁻¹ period⁻¹) during the paddy rice cultivation period were decreased to 1.1–5.9 kg N ha⁻¹ period⁻¹ by the incorporation of composted rice straw or fresh wheat straw prior to the flooding of the field.

However, relatively few studies have directly compared CH₄ and/or N₂O emissions during paddy rice cultivation period in fields with paddy-upland crop rotation to those with continuous paddy rice cultivation. Kumagai and Konno (1998) conducted CH₄ flux measurements for three years in a paddy field with carrot cultivation in the preceding year, and found a significant decrease in CH₄ emission in the first year compared with that in the fields with consecutive paddy rice cultivation. However, they also found that no such significant decrease in CH₄ emission occurred in the second or third year of the experiment. As for N₂O, to the best of our knowledge, there are as yet no previous studies in which its emission in fields with paddy-upland crop rotation was directly compared to that with continuous paddy rice cultivation.

In this study, we focused on the effect of preceding drainage for upland crop cultivation on the CH₄ and N₂O emissions from the fields with paddy rice cultivation. CH₄ and N₂O fluxes were continuously monitored and compared in fields with consecutive paddy rice cultivation and paddy-upland crop rotation. On the basis of the differences in the CH₄ and N₂O emissions, the effect of paddy-upland crop rotation on the reduction in greenhouse gas emission was discussed.

Materials and methods

Outline of the experimental field and crop cultivation

The experiment was conducted for two years from 2004 to 2005 in six paddy soil lysimeter plots at the National Institute for Agro-Environmental Sciences (NIAES) (36°01′N, 140°07′E), Japan. Each lysimeter plot had a 9 m² (3 m × 3 m) cross-sectional area and a 1 m depth. The soil type for the plot was Gray lowland soil (Fluvisols). The topsoil had the texture of clay loam with a clay content of 36%. The bulk density of the topsoil at 0–5 cm depth was 0.84 g cm⁻³, and the soil pH in H₂O was 5.7. According to results of the soil core sampling conducted in the spring of 2004, the carbon and nitrogen contents of the topsoil (0–5 cm depth) were 18.8 mg C g soil⁻¹ and 1.5 mg N g soil⁻¹, respectively, which showed no significant difference among the plots.

The cultivation histories of the plots are shown in Table 1. Before 2001, paddy rice was cultivated in all six plots. From the spring of 2002 to the spring of 2004, single cropping of upland rice (UR) and double cropping of soybean/wheat (SW) were each conducted in two of the lysimeter plots, while single cropping of paddy rice (PR) was consecutively conducted in the remaining two plots. Then, from the spring of 2004, single cropping of paddy rice cultivation was again conducted in all six plots. The cultivated rice cultivar was “Nipponbare”. The planting density of the rice was 22.2 hills m⁻². Urea (50 kg N ha⁻¹ as basal application, and 40 kg N ha⁻¹ as supplemental application), fused magnesium phosphate (80 kg P₂O₅ ha⁻¹ as basal application) and potassium chloride (80 kg K₂O ha⁻¹ as basal application, and 30 kg K₂O ha⁻¹ as supplemental application) were applied as nitrogen, phosphate and potassium fertilizers, respectively. In 2004, the transplanting of rice seedlings to the SW plots was conducted 31 days later than that to the PR and UR plots owing to the preceding winter wheat cultivation. Therefore, rice growth and water management in the SW plots were also delayed compared with those in the PR and SW plots. In 2005, the transplanting of rice seedlings was conducted on the same date (May 16, 2005) in all three cropping systems.

Table 1. Outline of the crop cultivation histories of the single cropping of paddy rice (PR), single cropping of upland rice (UR), and double cropping of soybean and wheat (SW) plots

	1993–2001	2002		2003		2004	2005
PR	Paddy rice	Paddy rice		Paddy rice		Paddy rice	Paddy rice
UR	Paddy rice	Upland rice		Upland rice		Paddy rice	Paddy rice
SW	Paddy rice	Soybean	Wheat	Soybean	Wheat	Paddy rice	Paddy rice

The dates and the amounts of aboveground residue incorporation into the field from 2002 to 2005 are shown in Table 2. In the autumn of 2002 and 2003, residues of the harvested paddy rice and upland rice were air-dried and then respectively incorporated into the soil in the PR and UR plots. In the spring of 2003, residue of the harvested wheat was incorporated in the SW plots, whereas residues of the soybean harvested in the autumn of 2002 and 2003 and those of wheat harvested in the spring of 2004 were removed from the field (not incorporated). In the autumn of 2004 and 2005, residues of the harvested paddy rice were incorporated in all the plots.

Table 2. Dates and amounts of aboveground crop residue incorporated into the field from 2002 to 2005

Season	Cropping system^†	Harvested crop	Incorporated biomass^‡(g dry weight m^−2)	Date of residue incorporation
Autumn 2002–Spring 2003	PR	Paddy rice	708 ± 5	2002/10/17
	UR	Upland rice	339 ± 26	2002/10/17
	SW	Soybean/Wheat	0/436 ± 16	—/2003/06/20
Autumn 2003–Spring 2004	PR	Paddy rice	584 ± 12	2003/10/31
	UR	Upland rice	173 ± 91	2003/10/31
	SW	Soybean/Wheat	0/0	–/–
Autumn 2004	PR	Paddy rice	600 ± 16	2004/11/09
	UR	Paddy rice	650 ± 55	2004/11/09
	SW	Paddy rice	578 ± 0	2004/11/09
Autumn 2005	PR	Paddy rice	811 ± 0	2005/10/28
	UR	Paddy rice	822 ± 31	2005/10/28
	SW	Paddy rice	917 ± 8	2005/10/28
^†The cropping system is described in the text and in Table 1.
^‡The values are shown as mean ± standard deviation.

The outline of the water management during the rice cultivation period was based on conventional Japanese practices, i.e. continuous flood irrigation before summer, drainage and the subsequent intermittent flood irrigation in summer, and final drainage about 20 days before the rice harvest in autumn. In 2005, the fields were once drained briefly around July 3. However, the fields were flooded again by the subsequent precipitation and then maintained flooded until around July 15. Water percolation rate was regulated to be about 1 cm day⁻¹ using a tubing pump system (Model No. 7553-80, Cole-Parmer, USA) installed at the bottom of the soil layer of the lysimeters. During the fallow periods, the underground water table was kept at 90 cm depth, as described by Minamikawa et al. (2010).

Measurement of CH₄ and N₂O fluxes

A chamber made of transparent polycarbonate and acrylic plates was placed at the center of each lysimeter plot. The cross-sectional area of the chamber was 0.81 m² (0.9 m × 0.9 m). The height of the chamber was 0.6 m during the fallow period or when the plants were shorter than 0.6 m. It was changed to 1.2 m by connecting additional sidewalls when the plants grew taller. About every 40 minutes, the lids of one of the chambers were closed with pneumatic cylinders, kept closed for about 30 minutes, and then opened again. All the chambers were closed for separate flux measurements; therefore, each chamber was closed every 4 h (six times per day). During the closed period, the air inside of the chamber was circulated with a pump at flow rates of 5 to 7 L min⁻¹. Part of the circulated air was injected to the two gas chromatograph (GC) systems four times at intervals of 8.5 minutes. A GC (GC-14B, Shimadzu, Japan) equipped with a flame ionization detector and some switching valves was used for CH₄ measurement. Another GC (GC-14B, Shimadzu, Japan) equipped with an electron capture detector and some switching valves was used for N₂O measurement. More details of the flux measurement system are shown in our previous report (Nishimura et al. 2005b).

In accordance with the major maintenance of the entire monitoring system, the CH₄ and N₂O flux measurements were stopped from March 15 to May 12, 2005 (59 days) and from February 21 to March 15, 2005 (23 days), respectively. Cumulative CH₄ and N₂O emissions in 2005 were calculated without data for these periods. There were other 26 to 34 days for CH₄ and 25 to 32 days for N₂O (different among the plots) with flux data deficit (wholly or partly) during the whole experimental period, due to malfunctions or minor system maintenances. For these dates, daily CH₄ and N₂O emissions were estimated by linear interpolation using the daily cumulative emission data of the two adjacent dates (i.e. the dates immediately before and after the period with flux data deficits) without flux data deficits.

Combined annual global warming potential (GWP) was calculated using:

Other data measurements

The temperature of air inside the chambers was measured to calculate gas flux, with a platinum resistance thermometer placed at about 30 cm above the soil surface and recorded on a data logger (HR2400, Yokogawa, Japan). Ambient air temperature and precipitation data were provided hourly by the climate data acquisition station of the NIAES.

Soil core samples at 0–5 cm depth were collected for the analysis of reducible iron (Fe) content in the spring of 2002, 2004 and 2005 before the first flood irrigation. Reducible Fe content in the air-dried soil was analyzed by the dithionite- ethylenediaminetetraacetic acid (EDTA) extraction methods proposed by Asami and Kumada (1959).

Soil reduction/oxidation potential (Eh) was measured with platinum-tipped electrodes. The electrodes were inserted into the soil at depths of 2, 5 and 10 cm (three electrodes per depth of each lysimeter), and left there throughout the rice cultivation period. The potential, measured against a silver chloride reference electrode, was converted to that measured against a hydrogen electrode.

Soil core samples at 0–5 cm depth were collected occasionally during the study period for the analysis of soil inorganic nitrogen content. Soil samples were collected randomly from five locations in each plot and then mixed. The 15-g samples of the collected fresh soil were used for extraction with 100 mL of potassium chloride (KCl) solution (100 g KCl L⁻¹). Nitrate nitrogen (NO₃-N) was analyzed by the copper-cadmium reduction method, and ammonium nitrogen (NH₄-N) was analyzed by the indophenol blue method in a continuous flow analyzer (TRRACS, Bran + Luebbe, Germany).

Differences in the cumulative CH₄ and N₂O emissions and soil reducible Fe content among the cropping systems were analyzed with one-way analysis of variance (ANOVA) and Tukey's multiple comparison test.

Results

Seasonal courses of CH₄ and N₂O fluxes

The seasonal courses of CH₄ flux and soil Eh values are shown in Fig. 1. In 2004 and 2005, CH₄ flux gradually increased during the period with continuous flood irrigation, i.e. from the first spring irrigation to the first summer drainage. In 2005, CH₄ flux decreased once in accordance with the brief drainage of the fields around July 3, although it increased again until the fields were completely drained around July 15. In 2004, the CH₄ flux was significantly higher in the PR plots than in the UR plots throughout the period with continuous flood irrigation. The increase in CH₄ flux in the SW plots was delayed compared with those in the PR and UR plots owing to the delayed flood irrigation. The CH₄ flux in the SW plots during this period was lower than that in the PR plots but higher than that in the UR plots. In 2005, the differences in CH₄ flux between the two replications were large in the PR and UR plots. The CH₄ flux in the SW plots in 2005 was apparently lower than those in the PR and UR plots. Methane flux dropped rapidly to a low level within a few days after the first summer drainage of the field in all the plots in 2004 and 2005. Methane flux then slightly increased again in accordance with the intermittent flood irrigation, but the magnitude of the flux was much lower than that before the drainage. During the periods without rice cultivation including fallow or wheat cultivation periods, CH₄ flux continuously remained low, as shown by the slightly negative fluxes in all three cropping systems (on average, −0.20 to −0.16 mg CH₄ m⁻² day⁻¹) (data not shown).

Seasonal courses of N₂O flux, soil inorganic nitrogen content and precipitation are shown in Fig. 2. A temporal high peak of N₂O flux was observed just after the first flood irrigation in all the cropping systems in 2004 and 2005, which lasted only a few days. Nitrous oxide flux then rapidly decreased and remained low during the period with continuous flood irrigation. During the summer period with drainage and intermittent irrigation, temporal small increases in N₂O flux were observed occasionally. After the final drainage in September and in the following fallow periods, some broad peaks of N₂O flux were observed. The broad peak of N₂O flux observed in the SW plots from January to April 2004 probably had close relationships to the wheat cultivation and supplemental fertilizer application for wheat on March 3. From January to April 2005, the magnitude of N₂O flux continuously remained low at less than 0.2 mg N m⁻² day⁻¹. In the comparison of 2004 and 2005 data, the N₂O flux was generally higher in 2004 than in 2005.

Soil reduction/oxidation potential

In general, the soil Eh gradually decreased in periods with continuous flood irrigation, and then rapidly fluctuated in accordance with the subsequent drainage and intermittent flood irrigation in summer. In 2004, the soil Eh during the continuous flood irrigation in the UR plots was consistently higher than that in the PR plots at all the soil depths examined. In particular, the soil Eh at 10 cm depth remained positive even just before the summer drainage (minimum +107 mV) (Fig. 1f). The delayed decrease in the soil Eh in the SW plots in 2004 was due to the delayed rice cultivation. The minimum Eh values in the SW plots just before the summer drainage at 2 and 5 cm depths were similar to those in the PR plots, whereas that at 10 cm depth (−9 mV) was 95 mV higher than that in the PR plots. In 2005, the seasonal courses of soil Eh were similar in all three cropping systems.

Figure 1. Seasonal courses of methane (CH₄) flux (a–c) and soil reduction/oxidation potentials (Eh values) at 2, 5 and 10 cm depths (d–f) in experimental paddy fields with consecutive paddy rice cultivation (PR), those with preceding upland rice cultivation (UR) and those with preceding soybean/wheat cultivations (SW) around the rice cultivation periods in 2004 (left) and 2005 (right). Methane fluxes are averages of two plots, which are daily cumulative values. The vertical bars represent standard deviation of the two plots. The horizontal solid and broken bars represent periods with continuous flood irrigation and those with drainage and intermittent flood irrigation, respectively. Note that wheat sown in the autumn of 2003 was cultivated until June 3, 2004 in the SW plots, and that the rice cultivation period in 2004 differed from that in the PR and UR plots.

Figure 2. Seasonal courses of nitrous oxide (N₂O) flux (a–c), soil ammonium nitrogen (NH₄-N) (d) and nitrate nitrogen (NO₃-N) (e) contents and precipitation (f) in experimental paddy fields with consecutive paddy rice cultivation (PR), those with preceding upland rice cultivation (UR) and those with preceding soybean/wheat cultivations (SW) from 2004 to 2005. Nitrous oxide fluxes are averages of two plots, which are daily cumulative values. The vertical bars represent standard deviation of the two plots. The vertical arrows represent the dates of nitrogen fertilization, with fertilization rates in kg N ha⁻¹. The horizontal solid and broken bars represent periods with continuous flood irrigation and those with drainage and intermittent flood irrigation, respectively. Note that wheat sown in the autumn of 2003 was cultivated until June 3, 2004 in the SW plots and that the rice cultivation period in 2004 differed from that in the PR and UR plots.

Soil reducible iron content

The reducible Fe contents of the air-dried samples of the topsoil in the spring of 2002, 2004 and 2005 are shown in Table 3. In the soil samples collected in the spring of 2004, the reducible Fe content determined by extraction with dithionite-EDTA solution was slightly but significantly higher in the UR plots than in the PR and SW plots. However, no significant differences among the cropping systems were found in 2002 and 2005. We also measured oxalate-extractable Fe content and no significant differences among the cropping systems were found in all the years studied (2002, 2004, and 2005) (data not shown).

Table 3. Reduceble iron (Fe) contents determined by extraction with dithionite- ethylenediaminetetraacetic acid (EDTA) solution using the air-dried topsoil

Year^†	Cropping system^‡	Reducible iron^§ (mg Fe g soil^−1)
2002	PR	14.0 ± 0.5^n.s
	UR	14.0 ± 0.1^n.s
	SW	14.0 ± 0.1^n.s
2004	PR	13.4 ± 0.2^*a
	UR	14.0 ± 0.1^*b
	SW	13.2 ± 0.1^*a
2005	PR	13.8 ± 0.0^n.s
	UR	13.7 ± 0.4^n.s
	SW	13.9 ± 0.0^n.s
Different letters with * show significant differences between the cropping systems within each year (P < 0.05) by Tukey's multiple comparison test, and n.s. shows no significant difference by one-way analysis of variance (ANOVA).
^†Soil samples were collected in the spring of each year before the first flood irrigation.
^‡The cropping system is described in the text and in Table 1.
^§The values are shown as mean ± standard deviation.

Soil inorganic nitrogen content

A slight increase in soil NH₄-N content was found from March to April 2004 in all the plots. The distinctive increase in the SW plots was probably related to the fertilizer application for wheat. In all the plots, soil NH₄-N content rapidly increased after the basal fertilizer application for rice cultivation with the first flood irrigation in the spring of 2004 and 2005. Then, the soil NH₄-N contents remained higher than 30 mg N kg soil⁻¹ for more than one month in all the plots and then gradually decreased in 2004. In 2005, however, the soil NH₄-N contents gradually decreased throughout the period with continuous flood irrigation. A temporal increase on July 13 was probably caused by the brief drainage of the field around July 3. During the periods with intermittent drainage and flood irrigation in summer, the soil NH₄-N contents generally remained low. From the autumn of 2004 to the spring of 2005 and from the autumn to the end of 2005, the soil NH₄-N contents also remained low (mostly lower than 3 mg N kg soil⁻¹) in all the plots.

The soil NO₃-N contents remained lower than 3 mg N kg soil⁻¹ throughout the entire experimental period. In addition, detectable soil NO₃-N contents were observed only before the first flood irrigation in 2004, and just before the first flood irrigation in 2005. During the other periods, including those with continuous or intermittent flood irrigation for rice cultivation and from the final drainage in autumn to winter, the soil NO₃-N contents generally ranged below the detection limit of the analyzer.

Cumulative CH₄ and N₂O emissions

The cumulative CH₄ and N₂O emissions and combined GWPs are summarized in Table 4. The cumulative CH₄ emissions in 2004 were significantly different among the three cropping systems, with the highest in the PR plots. The cumulative CH₄ emissions in the UR and SW plots were 8 and 46% of that in the PR plots, respectively. In 2005, the differences in the cumulative CH₄ emissions among the plots were not statistically significant owing to the large differences among the two replications in the PR and UR plots. However, the cumulative emission in the SW plots was apparently lower than those in the PR and UR plots.

Table 4. Cumulative methane (CH₄) and nitrous oxide (N₂O) emissions and the combined global warming potentials (GWPs) in 2004 and 2005 in the experimental paddy fields

Year	Cropping system^†	CH4 emission^‡ (mg CH4 m^−2 y^−1)	N2O emission^‡ (mg N m^−2 y^−1)	Combined GWP ^‡,§ (g CO2eq m^−2 y^−1)
2004	PR	6092 ± 5^**a	55 ± 4^n.s	178 ± 1^**a
	UR	511 ± 311^**b	71 ± 13^n.s	46 ± 2^**b
	SW	2817 ± 373^**c	99 ± 18^n.s	117 ± 1^**c
2005	PR	5893 ± 1888^n.s	41 ± 1^n.s	167 ± 47^n.s
	UR	5123 ± 3292^n.s	39 ± 12^n.s	146 ± 77^n.s
	SW	1331 ± 522^n.s	57 ± 6^n.s	60 ± 10^n.s
Different letter with ** show significant differences between the cropping systems within each year (P < 0.01) by Tukey's multiple comparison test, and n.s. shows no significant difference by one-way analysis of variance (ANOVA).
^†The cropping system is described in the text and in Table 1.
^‡The values are shown as mean ± standard deviation.
^§The equation for calculating combined GWP is described in the text.

The cumulative N₂O emissions ranged from 39 to 99 g N m⁻² y⁻¹, and were higher in 2004 than in 2005 in all three cropping systems. Differences in the cumulative N₂O emission were not statistically significant among the three cropping systems in 2004 and 2005.

In 2004, the combined GWP was highest in the PR plots and lowest in the UR plots owing to the contributions of the highest and lowest CH₄ emissions, respectively. In 2005, the combined GWP was distinctively lower in the SW plots than in the PR and UR plots, although the differences among the cropping systems were not statistically significant.

Discussion

The outline of the seasonal courses of the CH₄ flux observed in this study followed the typical patterns reported in many previous reports (e.g. Sass et al. 1992; Yagi et al. 1996; Corton et al. 2000; Wang et al. 2000; Wassmann et al. 2000; Nishimura et al. 2004), i.e. a gradual increase during the period with continuous flood irrigation and a rapid decrease to a low level in accordance with the subsequent summer drainage. The cumulative CH₄ emissions obtained in this study ranged from 511 to 6092 mg CH₄ m⁻² y⁻¹ (Table 4), all of which were much lower than the average reported in Japanese paddy fields in lowland soil with straw application, i.e. 19,100 mg CH₄ m⁻² y⁻¹ (Greenhouse Gas Inventory Office of Japan 2009).

In the UR plots in 2004, the cumulative CH₄ emission was extremely low at only 8% of that in the PR plots (Table 4). The soil Eh in the UR plots was consistently higher than that in the PR plots at all the soil depths from 2 to 10 cm throughout the period with continuous flood irrigation (>Fig. 1d–f). In addition, the amount of residue incorporation in the UR plots was distinctively lower than that in the PR plots from 2002 to 2003 owing to the small amount of aboveground biomass of the upland rice (Table 2). Therefore, although the contents of total carbon in the topsoil did not differ significantly, the amount of labile organic carbon that could serve as the carbon source of CH₄ may have been lower in the UR plots than in the PR plots in the spring of 2004. Furthermore, the dithionite-EDTA extractable soil Fe content was significantly higher in the UR plots. Takahashi et al. (1999) reported that the reducible Fe content of the air-dried paddy soil determined by the dithionite-citrate extraction method did not significantly change by the preceding upland crop cultivation. However, they also found that Fe²⁺ production rate after submergence of the soil significantly decreased with an increase in the number of upland crop (soybean) cultivations in the preceding year. Based on these results, they discussed that crystallization of the ferric iron may have proceeded gradually during the period of upland crop cultivation and thus the soil may have become more resistant to microbiological reduction. Taking the results of this report into account, the soil of the UR plots in this study may have become also resistant to microbiological reduction during the preceding two years (2002–2003) with upland rice cultivation, although the difference in the soil reducible Fe content was small (Table 3). These results strongly indicate that the preceding upland rice cultivation was effective in reducing the CH₄ emission from paddy fields when the amount of preceding residue application of the upland rice is less than those of the paddy rice.

In the SW plots in 2004, the cumulative CH₄ emission was significantly higher than that in the UR plots but significantly lower than that in the PR plots (Table 4). Although the rice cultivation period in the SW plots differed from that in the PR and UR plots, the minimum soil Eh values in the SW plots just before the first summer drainage were similar to those in the PR plots at 2 and 5 cm depths. However, that at 10 cm depth was higher by 95 mV than that in the PR plots, which indicates less reduced condition in the soil of 10 cm depth in the SW plots (Fig. 1f). As for the soil reducible Fe content, significant differences were not found between the SW and PR plots. From 2002 to 2003, the crop residue incorporation was conducted only once (wheat residue incorporation in the spring of 2003) in the SW plots (Table 2), which may have left a small amount of labile organic carbon for the source of CH₄ in the spring of 2004 compared with that in the PR plots. However, harvest of the wheat in the spring of 2004 was conducted on June 3, which was only eight days before the flood irrigation for the subsequent paddy rice transplanting. Therefore, although the aboveground wheat residue was removed from the field, a significant amount of labile organic carbon may have been supplied through the decomposition of the remaining roots and stubbles of the wheat in the soil, since the application of fresh organic matter just before flood irrigation often enhances CH₄ emission significantly (e.g. Lu et al. 2000; Xu et al. 2000). We did not measure biomass of the remaining dead roots and stubbles of the wheat in the spring of 2004. According to the investigation in the spring of 2003, the biomass of the remaining dead roots and stubbles of the wheat after harvest was 78 g m⁻², and the similar amount of biomass was thought to have also remained in the spring of 2004. Taking these results into account, the effect of reducing CH₄ emission during the subsequent paddy rice cultivation period was also indicated in the SW plots, but the effect was thought to be smaller than that in the UR plots.

In 2005, the cumulative CH₄ emission of the SW plots was lower than those in the PR and UR plots. However, owing to the large differences between the two replications in the PR and UR plots (Fig. 1a and b), the differences among the cropping systems were not statistically significant (Table 4). The seasonal courses of soil Eh were similar in all three cropping systems, and no distinctive differences were found (Fig. 1d–f). The soil reducible Fe contents showed no difference in all three cropping systems either (Table 3). Taking these results into account, we conclude that no significant difference in CH₄ emission was confirmed in 2005, and that no changes in the soil properties that may cause a significant difference in CH₄ production were found. The effect of the preceding upland crop cultivation on CH₄ emission reduction during the subsequent paddy rice cultivation may be generally limited to one year. After the second year, no decrease in CH₄ emission may be expected.

The results on the CH₄ emission obtained in this study were generally similar to those in the previous studies by Cai et al. (2000) and Kumagai and Konno (1998). According to the results of the field experiments by Cai et al. (2000), the CH₄ emission in the paddy field with triple upland vegetable cultivation in the preceding year was 51% of that with double paddy rice and single upland vegetable cultivations in the preceding year. Kumagai and Konno (1998) found that the CH₄ emission in the paddy field with preceding carrot cultivation was reduced at 45% of that with consecutive paddy rice cultivation. However, they also reported that the CH₄ emission rather increased in the second and third years, at 214 and 127%, respectively, compared with that in the field with consecutive paddy rice cultivation. In addition, they revealed by an anaerobic soil incubation experiment in the laboratory that the CH₄ production potential in the soil with preceding carrot cultivation decreased (69%) in the first year but rather increased in the second and third years (515 and 143%, respectively) compared with that with consecutive paddy rice cultivation.

The outline of the seasonal courses of N₂O flux during the rice cultivation periods in this experiment followed those in our previous study (Nishimura et al. 2004). A remarkable increase in N₂O emission just after the first flood irrigation was observed in 2004 and 2005, which may have been due to the rapid reduction of NO₃-N in the topsoil. Then, N₂O flux remained low, with little NO₃-N in the topsoil (Fig. 2a–c and e). The slight temporal increases in N₂O flux during the drainage and intermittent flood irrigation in summer were probably closely related to the timing of the supplemental fertilizer application and drainage, i.e. drainage just after supplemental fertilizer application often causes a temporal weak N₂O emission (Nishimura et al. 2004).

During the fallow periods, small and broad peaks of N₂O flux were also observed occasionally. The increase in N₂O flux in autumn, including around the periods just after the final drainage and rice harvest, was probably caused by the labile carbon and nitrogen supply by the dead rootmat of aquatic weeds and algae and the remaining stubbles of the harvested rice (Nishimura et al. 2004). However, such labile carbon and nitrogen supply may not be expected from winter to the next spring. During this period, the gradual decomposition and mineralization of soil organic matter were thought to be the only sources of labile carbon and nitrogen supply for N₂O production. The lowest N₂O flux continuously observed from January to April 2005 was probably caused by the combined condition of low labile organic carbon and inorganic nitrogen (NH₄-N and NO₃-N) contents in the soil.

The cumulative N₂O emissions obtained in this study ranged from 39 to 99 mg N m⁻² y⁻¹. Among these values, that in the SW plots in 2004 was the highest (99 mg N m⁻² y⁻¹), which was mainly caused by the high N₂O flux in the spring with wheat cultivation. However, no significant differences in cumulative N₂O emission among the cropping systems were found statistically (Table 4). In addition, the results obtained were similar to those obtained in the field in this study in 2002 (60 mg N m⁻² y⁻¹) (Nishimura et al. 2004). In comparison to other previous studies, the cumulative N₂O emissions obtained in this field were comparatively low. According to the report by Akiyama et al. (2005), which shows a summary of data of the previous field experiments on N₂O emission in paddy rice fields, the mean background N₂O emission (annual N₂O emission from bare soil with no fertilization) was estimated to be 182 mg N m⁻² y⁻¹. The cumulative emissions in the field in this study were even lower than this background N₂O emission. The comparatively low N₂O emissions during the rice cultivation period in the field in this study were probably due to the comparatively low nitrogen fertilization rate, and the drainage and intermittent flood irrigation of short intervals in summer, as discussed in our previous report (Nishimura et al. 2004). On the other hand, further investigation of the factors affecting the dynamics of N₂O during the fallow period is required in future studies.

Studies focused on the effect of paddy-upland crop rotation on greenhouse gas emission have been strictly limited to date. Detailed mechanisms of the reduction in CH₄ emission should be intensively investigated in future studies. In particular, effect of organic matter incorporation and possible change in the form of Fe in the soil during the preceding period with upland crop cultivation on the reduction of submerged soil should be elucidated separately. In addition, a comprehensive evaluation of paddy-upland crop rotation should be conducted in future studies taking greenhouse gas emissions and other various factors such as food production and CO₂ emission (carbon balance as agro-ecosystem) into account together.

Conclusion

The results obtained here and in previous studies indicated a significant reduction in CH₄ emission in the paddy fields with preceding upland crop cultivation when crop residue from the previous upland crop is small or removed before paddy rice cultivation. This effect of CH₄ emission reduction, however, may not be expected in the following year and thereafter. On the other hand, no significant effect of preceding upland crop cultivation on N₂O emission was found. Therefore, the combined GWP by the CH₄ and N₂O emissions was also expected to be reduced by the preceding upland crop cultivation owing to the significant decrease in CH₄ emission, although this effect was expected only for one year just after the land use change from upland crop cultivation to paddy rice cultivation.

Acknowledgments

The authors thank Sachiko Banzawa and Akiko Yoshizawa, and the staff of the experimental field management section in the NIAES for their assistance with the experiment and management of the experimental field. The authors also thank the editor and the anonymous reviewers for providing many valuable suggestions and comments for improving the manuscript. This study was supported by a project entitled “Elucidation of Vulnerability in Agriculture, Forestry and Fishery to Global Warming and Development of Mitigation Techniques” conducted by “Research Initiatives” in the field of agro-environmental studies of the Ministry of Agriculture, Forestry and Fisheries from 2002 to 2006.