Early mid-season drainage can mitigate greenhouse gas emission from organic rice farming with green manure application

ABSTRACT

Organic rice farming is acknowledged as a safe and environmentally friendly rice production method. However, the application of fresh organic matter as fertilizer can increase methane (CH₄) emissions during the rice-growing period because it is a carbon (C) source for CH₄ under anaerobic conditions. In this study, we evaluated the CH₄ emissions and net greenhouse gas emissions (NGHGE) from rice paddy fields managed by long-term organic farming. We also assessed the effect of early incorporation of green manure and mid-season drainage (to reduce CH₄ emission and NGHGE) on the rice grain yield. In the first year, we measured CH₄ and nitrous oxide (N₂O) emissions, C budget, and yield in conventionally managed (CF) and organically managed (GM) rice paddy fields, in which white clover was grown in the fallow period. In the second year, we set up four treatments with dried hairy vetch (Vicia villosa) as green manure, incorporated in the conventional season (G), one week (GE1), and three weeks (GE3) earlier, and early mid-season drainage (GED), in which green manure was incorporated in the same way as in G. In the second year, we measured the same factors as in the first year. In the first year, C was accumulated in GM due to the application of C from green manure, despite C being lost in CF. On the other hand, the large amount of CH₄ emission induced in GM contributed to global warming due to high NGHGE compared to CF. Early incorporation of green manure in rice paddy fields did not increase CO₂ or decrease CH₄ emissions before rice transplanting in the second year. A significant reduction in grain yield in GE3 suggested that three weeks earlier incorporation of green manure is not appropriate in terms of stable agricultural management. Meanwhile, early mid-season drainage reduced CH₄ emissions, which contributed to both soil C sequestration and NGHGE reduction. These results indicate that mid-season drainage a week earlier than conventional practice could be a workable way of maintaining the brown rice yield and soil C sequestration and mitigating global warming in rice paddy fields receiving green manure.

KEY WORDS:

• Greenhouse gas emission
• carbon sequestration
• green manure
• early incorporation
• mid-season drainage

1. Introduction

Rice, one of the world’s major cereal crops, is a staple food in Japan. Thus, rice paddy fields are dominant agricultural land-use types in Japan. However, the emission of methane (CH₄) from rice paddy fields is a significant source of greenhouse gas from agriculture in the country. Because of the reductive decomposition of organic matter in the soil during the rice-growing season, CH₄ emission increases when organic matter is applied to rice paddy fields even it is composted (Yagi and Minami 1990; Takakai et al. 2020a, 2020b).

Nowadays, organic rice, regarded as safe food, is much in demand, and organic rice farming is increasing to supply the market. There are several types of organic rice farming, one of which utilizes green manure as fertilizer. In Japan, Chinese milk vetch (Astragalus sinicus) is a typical green manure crop traditionally used in rice paddy fields. The seed of Chinese milk vetch is sown in autumn just before or after the rice harvest. Growing Chinese milk vetch, not yet mature, is incorporated into the soil in spring several weeks before rice cultivation. However, the application of fresh organic matter before rice planting can increase CH₄ emissions during the rice-growing season (Suzue et al 2011; Toma et al. 2013). Suzue et al. (2011) observed that the application of Chinese milk vetch as an N source for rice increased CH₄ emissions by approximately 10 times relative to paddy fields managed by conventional fertilization.

Several methods to reduce CH₄ emissions from rice paddy fields have been proposed. These are, for example, rice straw composting (Yagi, Tsuruta, and Minami 1997), mid-season drainage, and early starting of mid-season drainage (Itoh et al. 2011; Nishimura et al. 2020), and application of iron materials (Ali, Oh, and Kim 2008). In a review paper summarizing field studies conducted in Southeast Asian countries (Yagi 2020), the application of alternate wetting and drying (AWD) technique, biochar application, composting of rice straw and livestock manure, application of sulfate-containing fertilizer, and drainage in the fallow season are effective management practices for reducing CH₄ emissions in rice paddy fields. The basic principles underlying these technologies to reduce CH₄ emissions are the removal of labile organic C, which can be a source of CH₄-C and the maintenance of oxidative conditions in the soil. Following this approach, one strategy is to reduce the amount of C in green manure as much as possible to limit the C input from green manure in paddy fields. In addition, returning the labile organic C in green manure to the atmosphere as CO₂ rather than CH₄ could reduce the C in green manure applied before rice cultivation. This approach would be advantageous because, as a greenhouse gas, CH₄ is 34 times stronger than CO₂ (IPCC 2013). Generally, green manure starts to decompose when incorporated into the soil, and the decomposed C is released as CO₂ under aerobic soil conditions before irrigation for rice cultivation. Therefore, earlier application of green manure in paddy fields may reduce the C source of CH₄ during the rice-growing period. However, early incorporation of milk vetch might reduce the green manure biomass because the vetch tends to grow rapidly with increasing temperatures during the spring season. Reducing green manure biomass inputs also means less N supply from the green manure. Therefore, it is important to identify an appropriate time to apply green manure that is effective in reducing CH₄ emissions while maintaining yield.

It is not possible to prevent soil reduction by applying iron oxides because artificially produced materials cannot be used in organic rice farming. Therefore, oxidation of the soil environment by soil drainage may be effective in reducing CH₄ production in rice paddy fields. Mid-season drainage is commonly practiced in Japan. In general, CH₄ flux in paddy fields gradually increases after rice transplanting until mid-season drainage and decreases rapidly after the drainage (e.g., Itoh et al. 2011; Toma et al. 2019). Therefore, early implementation of mid-season drainage can effectively reduce CH₄ emissions by lowering the peak CH₄ flux and thus reducing CH₄ production during the growing period. On the other hand, early implementation of mid-season drainage may suppress the number of rice tillers and reduce grain yield.

This study aimed to evaluate the organic rice farming practices utilizing green manure in terms of CH₄ emission and net greenhouse gas emission (NGHGE) compared to conventional rice farming practices during rice growing period in the first year. Furthermore, we assessed the impact of early incorporation of green manure and mid-season drainage on the CH₄ emission and NGHGE from spring till rice harvest in the second year. We also evaluated the effects of the above practices on rice yield.

2. Materials and methods

2.1. Site description and treatments

This study was conducted at the University Farm, Ehime University, Japan (33°57ʹ48” N, 132°47ʹ20” E, 11 m asl.). The mean annual air temperature was 16.5°C, and annual precipitation was 1315 mm (mean values over 30 years from 1981 to 2010). In the first year, 2017, the study was conducted in one conventionally managed (use of chemical fertilizer) (CF, 323 m²) and one organically managed (use of green manure) (GM, 3500 m²) rice paddy fields at the University Farm. Since 2007, only organic fertilizers have been applied in the GM. Especially in 2007, 2008, and from 2012 to 2016, white clover (Trifolium repens) has been applied as green manure. The soil characteristics in CF and GM are shown in Table 1. Available phosphorus content is relatively higher in CF, although exchangeable K and Mg contents and free Fe oxides content are higher in GM.

Table 1. Soil characteristics in chemical fertilizer applied (CF) and green manure (GM) applied paddy fields in 2017

Treatment	Depth	Available phosphorus content^‡	Ex. K content	Ex. Mg content	Ex. Ca content	CEC^§	Free Fe oxides content	Carbon content	Nitrogen content	C:N ratio	Sand	Silt	Clay
	(cm)	(g P2O5 kg^−1)	(mg kg^−1)	(mg kg^−1)	(g kg^−1)	(cmolc kg^−1)	(g Fe kg^−1)	(%)	(%)		(%)	(%)	(%)
CF	0–5	2.10	94.7	129	1.40	10.9	5.46	1.74	0.18	9.46	73.1	14.2	12.7
	5–10	2.15	86.7	124	1.35	10.3	5.21	1.52	0.16	9.28	71.7	13.1	15.2
	10–20	2.20	54.9	123	1.41	9.73	10.9	1.12	0.13	8.48	71.7	15.0	13.3
	20–30	1.05	51.8	136	1.11	8.48	12.8	0.37	0.05	7.89	69.7	13.2	17.2
GM	0–5	0.96	109	196	1.32	11.6	8.09	1.52	0.17	9.04	74.5	9.10	16.4
	5–10	1.07	123	180	1.41	10.1	7.85	1.44	0.16	8.99	70.5	13.8	15.7
	10–20	0.92	62.7	193	1.33	9.76	9.25	1.04	0.12	8.45	72.0	12.8	15.2
	20–30	0.29	31.2	201	1.60	8.16	10.3	0.24	0.05	4.54	71.9	12.0	16.2

‡ Bray II method, § Cation exchange capacity.

In the second year (2018), the study was conducted in sixteen small-scale experimental plots at the University Farm. The area of each plot was 20.8 m² (2.5 m × 8.3 m). In the surface soil layer (approximately 0–21-cm depth), cation exchange capacity, free Fe oxides contents were 8.77 cmol_c kg⁻¹ and 3.46 g Fe kg⁻¹, respectively. Soil carbon and nitrogen contents and C:N ratios were 1.04%, 0.10%, and 10.4, respectively. The sand, silt, and clay contents were 62.6%, 10.9%, and 26.5%, respectively. We set up four treatments. In the G, GE1, and GE3 treatments, dried hairy vetch (Vicia villosa), sold as feed for livestock, was incorporated as green manure in the conventional season, one week earlier (GE1), and three weeks earlier, respectively. In the GED treatment, the hairy vetch was incorporated in the conventional timing same as G, and early mid-season drainage was conducted. Each treatment was monitored for four iterations.

2.2. Land management

In the first year, 2017, seeds of white clover (10 kg ha⁻¹) were broadcast on October 27^th, 2016 in GM. Land was plowed on May 25^th and 30^th, 2017 in GM. In CF, chemical fertilizer (N: P₂O₅: K₂O = 84: 60: 60 kg ha⁻¹) was applied and incorporated on June 1^st. In both treatments, irrigation and puddling were conducted on June 6^th. Rice seedlings (15.2 hills m⁻², c.v. Nikomaru) were transplanted on June 9^th. Herbicides and pesticides were applied in CF, and any chemicals were not applied in GM. In GM, weeding was conducted on June 20^th (11 days after transplanting [DAT]), July 4^th (25 DAT), and 14^th (35 DAT). The soil was dried for a week from July 20^th to effect mid-season drainage in both treatments. Intermittent irrigation was practiced after the mid-season drainage. Both fields were naturally dried about one week before the harvest. Rice grains were harvested on October 15^th.

In the second year, 2018, hairy vetch (Vicia villosa), which contained 42.2% of C and 3.21% of N, was applied on May 4^th in GE3 (2.50 Mg ha⁻¹), 22^th in GE1 (4.00 Mg ha⁻¹), and 29^th in G and GED (5.00 Mg ha⁻¹). The application rates were decided based on the data of green manure growth in GM measured for several years collected before this experiment (data not shown). On each day, hairy vetch was incorporated into the soil at a depth of 0–10–cm just after the application. In all treatments, irrigation and puddling were conducted on June 18^th. Rice seedlings were transplanted on June 20^th at the same density as the first year. The soil was dried for a week to effect mid-season drainage from July 28^th to August 4^th in GED and from August 2^nd to 8^th in G, GE1, and GE3. After mid-season drainage, intermittent irrigation was conducted for all treatments. All experimental plots were naturally dried about one week before the harvest. Rice grains were harvested on October 11^th.

2.3. Greenhouse gas flux and cumulative emission measurements

Fluxes of CH₄, N₂O, and CO₂ were measured using the closed chamber technique. Acrylic chambers (30 cm × 60 cm) divided into upper (55 cm height) and lower (65 cm height) compartments were used for CH₄ and N₂O flux measurements in 2017 (Toma et al. 2019). Gas fluxes were measured at three different points in both CF and GM. During the early growing season, only the upper compartment was used, but both were used in the late growing season. Methane and N₂O gas samples were collected at 4, 14, and 24 min after the chambers were deployed. Stainless-steel bases were installed between rows for the CO₂ flux measurement. Polyvinyl chloride collars (20 cm high) were placed under stainless-steel bases to prevent the invasion of roots under the base area, consequently preventing CO₂ contamination from roots (Raich and Tefekciogul 2000; Toma et al. 2019). Therefore, the CO₂ flux from the soil surface in the bases was regarded as heterotrophic respiration (R_h). The stainless-steel chamber described by Toma et al. (2016) was used for measuring CO₂ flux from the soil surface in 2017. Carbon dioxide gas samples were collected at 0, 10, and 20 min after the chambers were deployed. The fluxes of all greenhouse gas (GHG) were measured approximately once every 10 days.

In the fallow period before rice cultivation in 2018 (from May 6^th till June 17^th), two of the stainless-steel bases, one for CH₄ and N₂O flux measurements and the other for CO₂ flux measurement were installed in the soil in each plot. For CO₂ flux measurement, herbicide was applied around the bases to prevent weed growth and CO₂ contamination from the weed root. A stainless-steel chamber was used for all GHG fluxes during the fallow season before rice cultivation. The GHG gas samples were collected as well as the method for CO₂ flux measurement in the first year. During the rice-growing period in 2018, all GHG fluxes were measured by the methods adopted in the first year, although the height of the lower acrylic chamber was 30 cm. The GHG flux measurement frequency was approximately once every 10 days, though GHG flux were measured more frequently just after the incorporation of hairy vetch and during the mid-season drainage.

Concentrations of CH₄ and N₂O were measured by gas chromatography (GC-14A, Shimadzu, Kyoto, Japan) equipped with a flame-ionization detector and an electron-capture detector, respectively. The concentration of CO₂ was analyzed with a CO₂ analyzer (ZFP-9, Fuji Electric, Tokyo, Japan). Fluxes of CH₄, N₂O, and CO₂ were calculated using linear regression. Cumulative CH₄, N₂O, and CO₂ (R_h) emissions were determined by the trapezoidal method, according to Toma et al. (2016).

2.4. Carbon budget and net greenhouse gas emission

In the paddy fields, the aboveground parts of the rice plants, grain, and straw were removed from the field. Thus, the removed C in the grain and straw from the field can be regarded as C-neutral. We also hypothesized that belowground parts of rice plants do not vary every year and decompose similarly after the plowing in every autumn. We also ignored C input from weed in CF in 2017, since weed was removed in CF during fallow season. Therefore, C input by green manure and C output by R_h and CH₄ emissions are C inflow and C outflow, respectively, as shown by the following equation for estimating C budget (Mg C ha⁻¹).

(1) $\mathrm{C}\mathrm{b}\mathrm{u}\mathrm{d}\mathrm{g}\mathrm{e}\mathrm{t}=\mathrm{C}\mathrm{i}\mathrm{n}\mathrm{g}\mathrm{r}\mathrm{e}\mathrm{e}\mathrm{n}\mathrm{m}\mathrm{a}\mathrm{n}\mathrm{u}\mathrm{r}\mathrm{e}-\mathrm{C}\mathrm{i}\mathrm{n}{\mathrm{R}}_{\mathrm{h}}\mathrm{a}\mathrm{n}\mathrm{d}\mathrm{C}{\mathrm{H}}_4\mathrm{e}\mathrm{m}\mathrm{i}\mathrm{s}\mathrm{s}\mathrm{i}\mathrm{o}\mathrm{n}$(1)

The global warming potentials (GWP), including climate-carbon feedbacks, of CH₄ and N₂O, were 34 and 298 times higher, respectively, than the GWP of CO₂ over a 100-year time horizon (IPCC 2013). The net greenhouse gas emission (NGHGE, Mg CO₂eq ha⁻¹) during the study period was calculated as the sum of the GWPs of CH₄ (GWP_CH4), N₂O (GWP_N2O), R_h (GWP_Rh), and green manure C (GWP_GM), which is applied-C from white clover, other plants, and roots in 2017 and hairy vetch in 2018:

(2) $\mathrm{N}\mathrm{G}\mathrm{H}\mathrm{G}\mathrm{E}=\mathrm{G}\mathrm{W}{\mathrm{P}}_{\mathrm{C}\mathrm{H}4}+\mathrm{G}\mathrm{W}{\mathrm{P}}_{\mathrm{N}2\mathrm{O}}+\mathrm{G}\mathrm{W}{\mathrm{P}}_{\mathrm{R}\mathrm{h}}-\mathrm{G}\mathrm{W}{\mathrm{P}}_{\mathrm{G}\mathrm{M}}$(2)

Details for estimating each GWP value were explained by Toma et al. (2019). Yield scale NGHGE (NGHGE_Y) was calculated by dividing NGHGE by brown rice yield.

2.5. Measurements of green manure biomass, rice growth, and yield components

In GM, plant biomass was measured on May 22^nd, 2017. All the aboveground plants were clipped from six quadrats at 2500 cm² (50 cm × 50 cm) followed by root collection from small quadrats at 625 cm² (25 cm × 25 cm, 15-cm-depth) within each quadrat. Collected aboveground plant materials were separated into white clover and other plants. All plant samples were dried at 70°C for at least 48 h and weighed. Carbon and N contents were measured using a CN analyzer (Sumigraph NC-80 auto, Sumika Chemical Analysis Service, Tokyo, Japan) after the dried plant samples had been finely powdered. The C and N contents in hairy vetch applied in 2018 were also measured using the same method.

The growth of the rice plants, described by plant height, tiller number, and chlorophyll content (SPAD-502, MINOLTA, Osaka, Japan), was measured at five replications during the rice-growing season. Nine rice plants per plot were clipped at harvest time and air-dried for about 10 days. Panicles were counted, and grains were separated from the straw. Rice straw was dried at 70°C for at least 48 h and weighed. Rice husks were removed, and 1000 brown rice grains and percentage of whole grains in brown rice were weighed using a grain inspector (RGQI10A, Satake, Hiroshima, Japan). The brown rice yield per unit area was calculated from plant density and brown rice yield per plant. The protein content of 15% moisture was measured using a composition analyzer (AN-700, Kett, Tokyo, Japan).

2.6. Soil analysis and environmental factor measurements

In both years, intact soil core samples for the measurements of soil Fe²⁺ content, completely enclosed to avoid air contamination, were collected from the top 0–10 cm and extracted by 1 M sodium acetate at pH 3.0 immediately after opening in the laboratory. The Fe²⁺ content in the extraction was analyzed by 0.2% o-phenanthroline colorimetry on the day of extraction. Ammonium-nitrogen (NH₄⁺) and nitrate-nitrogen (NO₃⁻) contents in the soil samples were also extracted from the soil samples by 2 M KCl (1:10). The NH₄⁺ and NO₃⁻ contents in the extraction were analyzed by indophenol blue and vanadium (III) chloride–nitrogen-ethylenediamine dihydrochloride colorimetry, respectively.

In the fallow period in 2018, soil water content was measured in the top 0–10 cm soil. Soil redox potential (Eh) was measured by a portable soil Eh meter (PRN-41, Fujiwara, Tokyo, Japan) and a platinum electrode (EP-201 type, Fujiwara, Tokyo, Japan) at a depth of 5 cm in the rice-growing period in both years. Mean daily air temperature and daily precipitation were recorded at the meteorological station in the University Farm.

2.7. Statistical analysis

Welch’s t-test was used to evaluate the differences in GHG emissions (CH₄, R_h, N₂O), C budget, and NGHGE among C and G treatments in 2017. Welch’s analysis of variance, which did not consider the homogeneity of variance, and Games-Howell’s multiple comparison test were adapted for the evaluation of the difference in dry matter application, N content, C:N ratio, and N application rate between the white clover, other plants, and roots in 2017, and for GHG emissions, C budget, and NGHGE between the treatments in 2018. Statistically significant differences were evaluated at the 5% probability level.

3. Results

3.1. Weather, greenhouse gas fluxes, carbon budget, and net greenhouse gas emission

In 2017, the mean daily air temperature increased from June to August and decreased toward the end of October (Figure 1(a)). The average daily air temperature (23.7°C) from June to October was 0.4°C lower than usual in this region. The highest and lowest temperatures during the rice-growing period were 30.3°C on August 25^th and 17.2°C on October 15^th, respectively. Precipitation from June to August (1064 m) was 332 mm higher than usual. However, precipitation in July (69 mm) was only 36% of the usual.

Figure 1. Seasonal variation in mean daily air temperature and precipitation (a), CH₄ (b), CO₂ (c), and N₂O (d) fluxes in chemical fertilizer (CF) and green manure (GM) applied treatments in 2017. F, T, P, TP, W, and H represents chemical fertilizer application, tillage, puddling, transplanting, weeding, and harvest, respectively

Total N and C application rates from plants in 2017 were 178 ± 24.9 (Average±SD) kg N ha⁻¹ and 4488 ± 435 kg C ha⁻¹, respectively, in GM. Nitrogen application from other plants and roots was higher than that of white clover due to higher dry matter, although the N content of white clover was significantly higher than that of other plants and roots (Table 2). The C:N ratio of white clover was significantly lower than that of other plants and roots. Carbon inputs from other plants and roots were 5.56 and 8.58 times higher than that of white clover (Table 3).

Table 2. Dry matter and nitrogen input from plant in GM in 2017

		Aboveground		Root	Total
		White clover	Other plant
Dry matter	(Mg ha^−1)	0.72 ± 0.78 c	3.36 ± 1.42 b	6.66 ± 0.98 a	10.7 ± 0.60
N content	(%)	2.62 ± 0.42 a	1.65 ± 0.62 b	1.47 ± 0.13 b	ND
C:N ratio		16.1 ± 2.48 b	30.5 ± 7.04 a	26.2 ± 3.12 a	ND
N application rate	(kg N ha^−1)	19.7 ± 21.5 b	61.0 ± 44.2 ab	97.4 ± 11.4 a	178 ± 24.9

Values are represented by Average ± Standard deviation.

Different letters represent significant difference between white clover, other plant, and root. ND: no data.

Table 3. Carbon budget (kg C ha⁻¹) from June 11 to October 15 in 2017

			Treatment
			CF	GM	P value
Input	Aboveground	White clover	0	296 ± 217	NA
		Other plant	0	1649 ± 734	NA
	Root		0	2543 ± 357	NA
	Total		0	4488 ± 211	NA
Output	CH4 emission		194 ± 71.7	1535 ± 420	< 0.05
	Rh		2094 ± 429	2610 ± 204	0.39
	Total		2288 ± 364	4145 ± 421	< 0.01
C budget			−2288 ± 364	343 ± 632	< 0.01

NA: not available.

Values are represented by Average ± Standard deviation.

Positive and negative values represent net C sink and source of atmospheric C, respectively.

Methane and CO₂ fluxes before mid-season drainage in 2017 were higher in GM than in CF, respectively. Differences in fluxes between the treatments were not apparent after mid-season drainage (Figure 1(b, c)). The CH₄ flux was higher before mid-season drainage than after. The N₂O fluxes were relatively lower before mid-season drainage and higher after mid-season drainage (Figure 1(d)). There were no differences in N₂O flux among the treatments before mid-season drainage, although N₂O flux in GM tended to be higher in GM.

Cumulative CH₄ emissions were significantly higher in GM than in CF (Table 3). The R_h was relatively higher in GM than in CF, although a significant difference in R_h was not observed between the treatments (Table 3). Cumulative N₂O emissions in GM (3.12 kg N ha⁻¹) were 2.88 times higher than in CF (1.08 kg N ha⁻¹), although there was no significant difference in cumulative N₂O emissions between the treatments.

The positive C budget in GM (0.34 ± 0.63 Mg C ha⁻¹) showed net C sequestration in 2017 (Table 3). On the other hand, a negative C budget in CF (−2.29 ± 0.36 Mg C ha⁻¹) showed net C loss and a significantly lower C budget compared to that in GM. The NGHGE in both treatments showed positive values (Table 4), and NGHGE in GM was 3.5 times higher than in CF (P = 0.05). A higher contribution (85%) of GWP_CH4 to total GWP_OUT was observed in GM.

Table 4. Global warming potentials (GWP) and net greenhouse gas emission (NGHGE) (Mg CO₂eq ha⁻¹) from June 11 to October 15 in 2017

		Treatment		P value
		CF	GM
GWPIN	GWPGM of white clover	0	−1.09 ± 0.80	NA
	GWPGM of other plant	0	−6.04 ± 2.69	NA
	GWPGM of root	0	−9.32 ± 1.31	NA
	Total	0	−16.5 ± 0.77	NA
GWPOUT	GWPCH4	8.82 ± 3.25	69.6 ± 19.1	< 0.05
	GWPRh	7.68 ± 1.57	9.57 ± 0.75	0.16
	GWPN2O	0.51 ± 1.15	1.46 ± 1.26	0.39
	Total	17.0 ± 2.02	80.6 ± 18.2	< 0.05
NGHGE		17.0 ± 2.02	64.2 ± 18.9	0.05

NA: not available.

Values are represented by Average ± Standard deviation.

Positive and negative values represent acceleration or mitigation of global warming, respectively.

In 2018, the mean daily air temperature increased from May to August and decreased toward the end of October (Figure 2(a)). The average daily air temperature (24.0°C) from June to October was almost averages in this region. The highest and lowest temperatures during the rice-growing period were 31.6°C on August 24^th and 17.2°C on October 11^th, respectively. Precipitation from June to August (1156 mm) was 426 mm higher than usual. On the other hand, precipitation in July (526 mm) was 2.75 times higher than usual.

Figure 2. Seasonal variation in mean daily air temperature (a), CH₄ (b), CO₂ (c), and N₂O (d) fluxes in green manure (G) fertilization, green manure fertilization at 1 (GE1) and 3 (GE3) weeks earlier treatments and combination of green manure fertilization and earlier mid-season drainage (GED) in 2018. HV, T, P, and H represents hairy vetch application, tillage, puddling, and harvest, respectively

The CH₄ flux in the fallow period in 2018 was lower than that during the rice-growing period in all treatments (Figure 2(b)). The CH₄ flux increased after the transplanting of rice seedlings in all treatments. However, in GED, CH₄ flux decreased earlier than in other treatments after the mid-season drainage. In G, GE1, and GE3, CH₄ flux also decreased after mid-season drainage. Methane fluxes increased in mid-September, although they were not higher than those before the mid-season drainage in all treatments. Carbon dioxide fluxes in the fallow period tended to be higher than those in the rice-growing period (Figure 2(c)). No apparent differences in CO₂ flux among the treatments were observed during the rice-growing season. The N₂O flux in the fallow period increased just after the application of hairy vetch in all treatments in 2018 (Figure 2(d)). In the rice-growing season, N₂O flux increased during mid-season drainage and decreased after in all treatments.

In 2018, cumulative CH₄ emissions were significantly higher in GE1 than those in GE3 and GED (Table 5). Methane emissions in GED were 32.8% and 30.1% of those in G and GE1, respectively. The highest emissions of R_h and N₂O were observed in GED, although there was no significant difference among the treatments. R_h in fallow season in G, GE1, GE3, and GED were 545 ± 59.7 (25.8% of applied C), 596 ± 141 (35.3%), 363 ± 159 (34.4%), and 468 ± 179 (22.2%) kg C ha⁻¹, respectively. There was no significant difference in R_h among the treatments

Table 5. Carbon budget (kg C ha⁻¹) from May 5 to October 11 in 2018

		Treatment
		G	GE1	GE3	GED
Input	Hairy vetch^§	2112	1690	1056	2112
Output	CH4 emission	276 ± 185 abc	300 ± 56.9 a	171 ± 34.1 b	90.4 ± 30.7 c
	Rh	1675 ± 185 a	1751 ± 223 a	1449 ± 203 a	1778 ± 489 a
	Total	1951 ± 211 a	2051 ± 260 a	1620 ± 210 a	1688 ± 468 a
Carbon budget		161 ± 211 a	−361 ± 260 ab	−664 ± 210 b	244 ± 468 ab

Values are represented by Average ± Standard deviation.

Different letters represent significant difference between the treatments.

Positive and negative values represent net C sink and source of atmospheric C, respectively.§ differences in carbon input by hairy vetch incorporation was not statistically analyzed due to artificial application.

In 2018, the C budget in G and GED showed net C sequestration (Table 5). There was a significant difference in the C budget between G and GE3. The C budget in GED, which was the highest among the treatments, was not significantly different from that in other treatments due to its higher variability. In all treatments, Rh, which contributed to more than 82.6% (69.8–89.4%) in the C output, was not significantly different among the treatments. The NGHGE in all treatments showed positive values (Table 6); NGHGE in GE1 was highest and significantly higher than that in GED. The GWP_OUT in GE1 was significantly higher than in GE3 and GED. A higher contribution of GWP_CH4 to GWP_OUT (66.0%) was also observed in GE1.

Table 6. Global warming potentials (GWP) and net greenhouse gas emission (NGHGE) (Mg CO₂eq ha⁻¹) from May 5 to October 11 in 2018

		Treatment
		G	GE1	GE3	GED
GWPIN	GWPGM^§	−7.74	−6.20	−3.87	−7.74
GWPOUT	GWPCH4	12.5 ± 8.39 abc	13.6 ± 2.58 a	7.75 ± 1.55 b	4.10 ± 1.39 c
	GWPRh	6.14 ± 0.68 a	6.42 ± 0.82 a	5.31 ± 0.74 a	6.52 ± 1.79 a
	GWPN2O	0.57 ± 0.40 a	0.56 ± 0.50 a	0.37 ± 0.17 a	0.66 ± 0.28 a
	Total	19.2 ± 8.30 ab	20.6 ± 2.78 a	13.4 ± 1.93 b	11.3 ± 1.53 b
NGHGE		11.5 ± 8.30 ab	14.4 ± 2.78 a	9.55 ± 1.93 a	3.53 ± 1.53 b

Values are represented by Average ± Standard deviation.

Different letters represent significant difference between the treatments.

Positive and negative values represent acceleration or mitigation of global warming, respectively.

§ differences in carbon input by green manure, which was hairy vetch, incorporation was not statistically analyzed due to artificial application.

3.2. Soil redox condition and soil inorganic nitrogen concentrations

In 2017, soil Eh and Fe²⁺ content were lower and higher in GM than CF in the period from rice transplanting to mid-season drainage, respectively (Figure 3(a, b)). After mid-season drainage, there was no apparent difference in soil Eh and soil Fe²⁺ content between GM and CF. Soil NH₄⁺ and NO₃⁻ contents in GM were relatively higher and lower in GM than in CF, respectively (Figure 3(c, d)). However, after the mid-season drainage, differences between the treatments were unclear.

Figure 3. Seasonal variation in soil Eh at 5-cm depth (a), soil Fe₂⁺ (b), NH₄⁺, and NO₃⁻ contents in chemical fertilizer (CF) and green manure (G) applied treatments in 2017. F, T, P, TP, W, and H represents chemical fertilizer application, tillage, puddling, transplanting, weeding, and harvest, respectively

In 2018, soil water content increased in all treatments due to rainfall after hairy vetch application in GE3 (Figure 4(a)). However, soil water content did not vary among the treatments in the fallow period. Soil Eh in the GED increased earlier than in the other treatments because of earlier mid-season drainage. After mid-season drainage, there was no significant tendency of soil Eh among the treatments (Figure 4(b)). Due to the earlier mid-season drainage, soil Fe²⁺ content in the GED decreased earlier than in the other treatments. In late September, both soil Eh and soil Fe²⁺ content in GED showed the highest and lowest values compared to other treatments. Soil NH₄⁺ contents in the fallow period were lower than those in the rice-growing period (Figure 4(c)). On the contrary, difference in soil NO₃⁻ contents were not clear among the treatment. In GE3, to which the lowest amount of hairy vetch was applied, lower soil NH₄⁺ and NO₃⁻ content were apparent after and before rice transplanting, respectively.

Figure 4. Seasonal variation in soil Eh at 5-cm depth (a), soil Fe₂⁺ (b), NH₄⁺, and NO₃⁻ contents in green manure (G) fertilization, green manure fertilization at 1 (GE1) and 3 (GE3) weeks earlier treatments and combination of green manure fertilization and earlier mid-season drainage (GED) in 2018. HV, T, P, and H represents hairy vetch application, tillage, puddling, and harvest, respectively

3.3. Plant growth, yield components, and yield scale net greenhouse gas emission

In 2017, the tiller number and leaf chlorophyll content in CF were relatively higher than those in GM, although differences in plant height were not apparent (Figure S1). Brown rice yields and panicle numbers in CF were significantly higher than those in GM, and the 1000 grain weight in CF was significantly lower than that in GM (Table 7).

Table 7. Yield component and quality of rice in 2017

	Unit	Treatment		P value
		CF	GM
Grain weight	(g m^−2)	650 ± 68.2	538 ± 23.8	0.09
Straw weight	(g m^−2)	1080 ± 95.6	889 ± 19.6	0.07
Grain/Straw		0.60 ± 0.04	0.60 ± 0.03	0.98
Panicle number	(number m^−2)	311 ± 14.8	273 ± 5.13	< 0.05
1000 grain weight^§	(g)	22.8 ± 0.41	23.8 ± 0.18	< 0.05
Brown rice yield^§	(g m^−2)	537 ± 54.8	416 ± 31.3	< 0.05
Protein content^§	(%)	5.50 ± 0.30	5.40 ± 0.26	0.69
Perfect grain ratio	(%)	83.5 ± 0.86	76.8 ± 6.88	0.23

Values are represented by Average ± Standard deviation.

§ 15% of moisture.

In 2018, plant height, tiller number, and leaf chlorophyll content did not vary between the treatments (Figure S2). In 2018, significantly higher brown rice yield and grain weight were observed in G than in GE3 (Table 8). Protein content in GED showed the lowest values among the treatments.

Table 8. Yield component and quality of rice in 2018

	Unit	Treatment
		G	GE1	GE3	GED
Grain weight	(g m^−2)	606 ± 25.6 a	568 ± 56.0 ab	521 ± 30.0 b	598 ± 23.8 a
Straw weight	(g m^−2)	694 ± 9.01 a	719 ± 86.1 a	643 ± 44.0 a	700 ± 23.4 a
Grain/Straw		0.87 ± 0.03 a	0.79 ± 0.04 a	0.81 ± 0.02 a	0.85 ± 0.04 a
Panicle number	(number m^−2)	378 ± 13.8 a	372 ± 25.7 a	352 ± 29.0 a	369 ± 21.6 a
1000grain weight^§	(g)	20.8 ± 0.14 a	21.0 ± 0.24 a	21.1 ± 0.13 a	21.1 ± 0.08 a
Brown rice yield^§	(g m^−2)	369 ± 8.16 a	349 ± 37.9 ab	317 ± 19.8 b	363 ± 25.0 ab
Protein content^§	(%)	7.15 ± 0.26 ab	7.00 ± 0.08 a	6.89 ± 0.16 ab	6.66 ± 0.06 b
Perfect grain ratio	(%)	83.2 ± 1.52 a	84.8 ± 0.54 a	85.9 ± 2.40 a	86.2 ± 1.00 a

Values are represented by Average ± Standard deviation.

Different letters represent significant difference between the treatments.

§ 15% of moisture.

In 2017, the NGHGE_Y in GM (15.7 ± 5.38 Mg CO₂eq Mg⁻¹) was 4.95 times higher than that in CF (3.16 ± 0.11 Mg CO₂eq Mg⁻¹). In 2018, the NGHGE_Y in G, GE1, GE3, and GED were 3.08 ± 2.15, 4.18 ± 1.07, 3.04 ± 0.76, and 0.97 ± 0.40 Mg CO₂eq Mg⁻¹, respectively. Lowest value of NGHGE_Y was observed in GED among the treatments

4. Discussion

4.1. Rice paddy field organically managed by green manure

In the study site, C sink occurred in greater amounts in the green manure-applied fields, indicating that green manure applied to rice paddy fields is a sink of atmospheric C. The contribution of the root C supply by white clover and other plants, in particular, to the C budget in the field was significant in 2017. Roots had a higher C content and C:N ratio than other parts of plant. Furthermore, plant roots are resistant to decomposition and relatively stable in soil compared to leaves and stems due to their high lignin content (Abiven et al. 2005). In this study, significantly higher CH₄ emissions and relatively higher R_h in green manure plots were observed in the green manure-applied field in 2017. Thus, the C output was higher with the application of green manure. Even under conventional rice cultivation practices, paddy fields have been reported to have lower C losses than upland fields (Nishimura et al. 2008; Yan et al. 2013). One of the reasons for this is that paddy rice fields are waterlogged during the rice cultivation period, which results in less soil organic matter decomposition. This suggests that although R_h and CH₄ emissions increased with the application of green manure, there was a net C accumulation in the field due to the mainly root-derived C.

While application of green manure increased the C budget in the paddy field, it was shown to have a net global warming accelerating effect, when CH₄ and N₂O emissions were considered. In particular, the contribution of CH₄ to NGHGE is significant. In general, the application of fresh organic matter, such as rice straw, to rice paddy fields before cultivation is known to increase CH₄ emissions during the rice-growing period (Yagi and Minami 1990). In this study, almost the same amount of green manure as rice straw was applied in GM in 2017, but the amount of CH₄ emission was much higher than values reported from studies in paddy fields in Japan. Yagi and Minami (1990) reported CH₄ emissions of up to 336 kg C ha⁻¹ yr⁻¹ (448 kg CH₄ ha⁻¹ yr⁻¹) in rice paddy fields receiving rice straw. Regardless of the application of rice straw and other organic matter, 903 kg C ha⁻¹ in the rice-growing season in Thailand (1204 kg CH₄ ha⁻¹, Cha-un et al. 2017) and 562 kg C ha⁻¹ in rice-growing season in Vietnam (749 kg CH₄ ha⁻¹, Tariq et al. 2017) and 894 kg C ha⁻¹ in rice-growing season in Vietnam (1192 kg CH₄ ha⁻¹, Trinh et al. 2017) and 499 kg C ha⁻¹ yr⁻¹ in China (665 kg CH₄ ha⁻¹ yr⁻¹, Liu et al. 2019) have been reported. Compared to these studies in Asian countries, a large amount of CH₄ was emitted from the green manure-applied paddy fields in GM in this study. The C:N ratio of rice straw is approximately 70, while the C:N ratio of the plants applied in this study was lower. These findings suggested that CH₄ production accelerated by rapid decomposition of green manure and reduction of soil after plowing may be greater than in other rice paddy fields. Application of a fresh weight of 36 Mg ha⁻¹ of barley and hairy vetch, which were grown as winter cover crops, was reported to emit 1100 kg C ha⁻¹ in the rice-growing season (1466 kg CH₄ ha⁻¹) of CH₄ in the paddy fields (Haque et al. 2013). In a previous study with the small-scale experimental plot used in this study, in which Chinese milk vetch with approximately 3.78 Mg ha⁻¹ of dry matter was incorporated, CH₄ emissions of 1027–1229 kg C ha⁻¹ were reported (Toma et al. 2019). These previous studies reported higher values of CH₄ emissions than others. However, a higher CH₄ emission in GM in 2017 was observed in this study, suggesting that the potential for CH₄ production may be high in long-term green manure-applied paddy fields. Furthermore, continuous rice cultivation utilizing green manure can obtain lower rice yield. Therefore, in rice paddy fields managed by green manure applications, the practice of reducing CH₄ emissions and increasing brown rice yield should be considered to mitigate the acceleration of global warming and for establishing acceptable cultivation practice by organically managed by green manure.

4.2. Earlier incorporation of green manure before rice cultivation

Early incorporation of green manure was ineffective with respect to increasing C budget, reducing NGHGE, and increasing brown rice yield in 2018 in this study. Green manure and other plants during the fallow period had lower biomass C when plowing was carried out three weeks earlier than usual because plants grow in the spring season. In addition, the average of R_h in all treatment before rice cultivation was 29.4% of the applied green manure C, with no significant difference in treatment in 2018. Since organic matter decomposition is generally positively influenced by temperature (e.g., Toma et al. 2019), it is possible that early plowing may not have resulted in as much decomposition due to low temperatures. The fact that CH₄ emissions tended to be lower in GE3 than in G or GE1 during the rice-growing season is assumed to be due to lower green manure C inputs because there was no significant difference in R_h between the treatments in the fallow period in 2018. Regarding NGHGE, the green manure incorporation three weeks before the conventional practice had about half the C applied, whereas the CH₄ emission was only about 40% lower. As a result, the NGHGE was not significantly lower due to earlier incorporation of green manure.

The low brown rice yield in GE3 could be due to the lower application of N in green manure in 2018. In particular, the relatively lower chlorophyll content, panicle number, and protein content observed in GE3 compared to other treatments in this study represent this effect. This finding demonstrated that earlier incorporation of green manure to reduce GHG emissions was not the preferred management method due to low brown rice yields, although NGHGE_Y in GE3 was comparable to that in G.

4.3. Earlier mid-season drainage

Early mid-season drainage in rice paddy fields has been reported to be effective in reducing CH₄ emissions in both chemical fertilizers and green manure applied fields (Itoh et al. 2011; Toma et al. 2019). In this study, one week earlier mid-season drainage was the most favorable for C sequestration, NGHGE, and brown rice yield in 2018. Both the C budget and NGHGE were the highest or lowest with the early mid-season drainage, respectively, because of the significant reduction in CH₄ emissions, even though the C input was the same as in the conventional management (Tables 5 and 6). Because CH₄ fluxes in the study site increased in the period before the mid-season drainage, earlier drainage lowered the maximum CH₄ fluxes, and this was considered to be the main reason for the lower CH₄ emissions in 2018 in this study. The number of panicles tended to be lower with earlier drainage receiving same amount of organic fertilizer in G. However, leaf chlorophyll content values during the ripening period were similar to those of G, except at harvest. These trends may have led to a concentrated supply of photosynthetic products to a limited sink, resulting in higher grain weight. Thus, there was no significant difference in brown rice yield between the treatments with the conventional and earlier operation of mid-season drainage (Table 8). In addition, lower soil NH₄⁺ content (Figure 4(c)) and leaf chlorophyll content at harvest time (Figure S2c), which tended to be lower in the GED than in the other treatments since the mid-drying season, may be responsible for the lower protein content of brown rice. In Japan, brown rice with lower protein content is considered to taste better and be of higher quality (Ishima et al. 1974). Therefore, it was also shown that early mid-season drainage could improve the quality of brown rice.

5. Conclusions

In this study, it was observed that the management of long-term green manure application in paddy fields acted as a C sink when compared to conventional rice cultivation. On the other hand, the large amount of CH₄ induced in the green manure-applied paddy field was found to contribute to global warming due to high NGHGE. Early incorporation of green manure in rice paddy fields did not increase CO₂ or decrease CH₄ emissions before rice transplanting. In addition, the reduction of N input by early incorporation was not considered a practical management method because of concerns about the reduction in brown rice yield. Early mid-drying reduced CH₄ emission by decreasing the peak CH₄ flux before mid-season drainage, which contributed to both soil C sequestration and NGHGE reduction. Furthermore, the technique was shown to have the potential to improve brown rice quality while ensuring the same yield as the conventional practice of mid-season drainage. Therefore, it was suggested that mid-season drainage a week early could be a practical system for maintaining brown rice yield and soil C sequestration and mitigating global warming in rice paddy fields receiving green manure.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This study was supported by JSPS KAKENHI Grant Number JP17H03951.