Effect of biogas slurry application on CH₄ and N₂O emissions, Cu and Zn uptakes by whole crop rice in a paddy field in Japan

Abstract

Effects of 4-year consecutive application of biogas slurry (BS) at rates of 0 (NF), 100 (BS100) and 300 (BS300) kg nitrogen (N) ha⁻¹ on yield of whole crop rice (Oryza sativa L. var. Leaf Star) and environmental impacts were investigated in a field experiment in comparison with a conventional rate of chemical fertilizer CF100 (100 kg N ha⁻¹). Average biomass production was comparable between BS100 (1.9 ± 0.1 kg dry matter m⁻²) and CF100 (1.8 ± 0.1 kg m⁻²) and significantly (P < 0.01) highest in BS300 (2.1 ± 0.1 kg m⁻²). Four years’ average methane (CH₄) emissions during the growing periods were significantly (P < 0.05) highest in BS300 (43.7 ± 18.4 g m⁻² season⁻¹), followed by BS100 (32.0 ± 3.0 g m⁻² season⁻¹) and then NF (23.5 ± 8.2 g m⁻² season⁻¹) and CF100 (20.3 ± 3.3 g m⁻² season⁻¹), indicating that BS application may cause a potential risk of CH₄ emission. There were no significant differences in copper (Cu) and zinc (Zn) uptakes by the rice plant between BS100 and CF100, but significantly higher Zn content was observed in the grain of BS300 in 2011, indicating a potential risk of higher heavy metal uptake in BS300. Compared with CF100, no significant higher accumulations of extractable and total forms of Cu and Zn in soil were observed from four years of consecutive BS application. This study revealed that the application of BS, generated from pig manure, to rice fields at the conventional rate (100 kg N ha⁻¹) may be considered to substitute chemical fertilizer utilization without additional environmental impacts in greenhouse gas emission and heavy metal uptake.

Key words:

• Biogas slurry
• Cu
• methane
• paddy field
• Zn

INTRODUCTION

Rice (Oryza sativa L.) is the most important staple food crop for the demand of the increasing population of the world (FAO 2002) and about 90% of world rice production comes from Asia (Welch et al. 2010). In Japan, however, annual rice consumption (on a calorie supply basis) per capita has been decreasing, from 118 kg in 1962 to 59 kg in 2008. In contrast, consumption of livestock products has increased from 9.2 to 28.5 kg for meat and from 37.5 to 91.8 kg for milk and dairy products during 1965 to 2008 (MAFF 2009). Cultivation of whole crop rice, which produces high biomass and is used as feed for livestock, is increasing in Japan to raise feed self-sufficiency ratio to 50% (on a calorie supply basis) in 2020 targeted by Ministry of Agriculture, Forestry and Fisheries (MAFF), and production of rice for feed increased from 8000 tons in 2008 to 183,000 tons in 2011 (MAFF 2012).

Great attention has been paid to anaerobic digestion of animal wastes, because it produces renewable energy in an environmentally friendly manner and the construction of biogas plants is increasing around the world (Alburquerque et al. 2012). The increase in such biofuel production has simultaneously increased the amount of by-product (Alotaibi and Schoenau 2011, 2013), i.e., digested slurry (biogas slurry, BS) that inevitably remains after biogas production and needs further purification to avoid environmental pollution. BS contains various nutrients and thus can be an effective organic fertilizer in crop production (Debnath et al. 1996; Matsunaka et al. 2006). According to Alburquerque et al. (2012), the sustainability of biogas production depends on an appropriate end use of the BS, avoiding any negative environmental impacts. Rice fields will be one of the best target sites to apply it as an effective organic fertilizer. Thus, we have evaluated BS as a fertilizer for whole crop rice production (Hou et al. 2007; Sunaga et al. 2009; Win et al. 2009, 2010). The previous studies revealed that BS is an alternative fertilizer enabling 25–30 t of dry matter production under an application rate of 300 kg nitrogen (N) ha⁻¹ (Sunaga et al. 2009). By contrast, enhanced methane (CH₄) emission was pointed out as a potential risk in a paddy fertilized with BS (Win et al. 2010). We reported that CH₄ emission was mitigated by an alternative use of pig manure as materials for anaerobic digestion rather than cow manure and concluded that pig manure-derived BS should be used as a fertilizer (Sasada et al. 2011). However, all these studies were conducted in lysimeters of 1 m × 1m × 0.5 m in size. It is empirically known that lysimeter or pot experiments cannot replace long-term field experiments for their reliability (Weatherley et al. 1988) and thus field experiments are essential to confirm the above findings.

Carbon (C) sequestration in agricultural land is one of the important strategies in mitigating the global warming problem (Mandal et al. 2007; Pan et al. 2009). Organic matter application is expected to increase C contents in soil (Janzen et al. 1998; Su et al. 2006; Yan et al. 2007; Banger et al. 2009; Rui and Zhang 2010). Rice fields sequestrate more C than upland fields because the decomposition rate of added and native organic carbons is lower under anaerobic conditions in rice soils than aerobic conditions in upland soils (Witt et al. 2000; Liping and Erda 2001; Yan et al. 2013). But there is little information about C sequestration in soil planted with whole crop rice and fertilized with BS.

BS contains high concentrations of heavy metal, especially in pig slurry (Nicholson et al. 1999). Therefore, heavy metal contents in the rice plant and their accumulation in the soil should also be measured to evaluate the sustainability of rice production using BS. Lu et al. (2012) evaluated heavy metal concentrations in soil and rice grain after a single application of BS to a paddy field. There was no report on the potential risks for heavy metal accumulation in paddy field applied with BS continuously, and thus, it is necessary to address this issue before paddy fields will be treated with an increasing amount of BS. In this study, therefore, whole crop rice cultivation fertilized with BS was performed in a field during 4 consecutive years to evaluate effects of BS application on the environmental impacts. The objectives of this study were to investigate the effects of different rates of BS application to a field on (1) biomass production of whole crop rice, (2) greenhouse gas [CH₄ and nitrous oxide (N₂O)] emission, (3) C content in soil and (4) uptake of copper (Cu) and zinc (Zn) by plants and their accumulation in soil, in a production system of whole crop rice fertilized with BS in relation to a conventional system using chemical fertilizer.

MATERIALS AND METHODS

Experimental design and treatments

A 4-year field experiment was carried out at Field Museum Hommachi, Field Science Center, Tokyo University of Agriculture and Technology (35°39’57’’N, 139°28’18’’E) from 2009 to 2012. The mean annual temperature in Fuchu was 15.6, 15.8, 15.3 and 14.9°C and the mean annual precipitation in Fuchu was 1566, 1714, 1428 and 1696 mm in 2009, 2010, 2011 and 2012, respectively. The site was used as a paddy field, had not been grown with any crops during 1994 and 2008, and then was used for this experiment. The initial soil properties of this site were as follows: soil type, gray lowland soil; soil texture, clay (sand 25%, silt 21% and clay 54%); bulk density, 1.04 g cm⁻³; porosity, 59.9%; total C, 28.3 g kg⁻¹; total N, 2.6 g kg⁻¹; available phosphorus (P), 0.12 g kg⁻¹; exchangeable potassium (K), 0.23 g kg⁻¹; exchangeable calcium (Ca), 3.08 g kg⁻¹; exchangeable magnesium (Mg), 0.25 g kg⁻¹. Four treatments were tested for 4 consecutive years. The treatments were NF (no fertilizer), CF100 [100 kg ammonium nitrogen (NH₄⁺-N) ha⁻¹, a compound chemical fertilizer with 14:14:14 (N (nitrogen):P₂O₅ (phosphorous):K₂O (potassium oxide)), CO-OP Chemical Company, Tokyo, Japan], BS100 [100 kg NH₄⁺-N ha⁻¹ with biogas slurry (BS, Table 1)] and BS300 (300 kg NH₄⁺-N ha⁻¹ with BS). In this study, 100 kg N was considered conventional, based on Nishimura et al. (2004) and Hayashi et al. (2006) who reported 90 kg N ha⁻¹ as a representation of the Japanese conventional style. Each plot was 2.4 m × 3.6 m in size and prepared in triplicate per treatment in a randomized complete block design. Each plot was separated using plastic boards to prevent the diffusion of applied fertilizer into the other plots. Whole crop rice (Oryza sativa L. var. Leaf Star: Ookawa et al. (2010) was grown in 30 cm × 15 cm spacing. Fertilizer application was split into three times (50% at basal, 30% at tillering stage and 20% at flowering stage). Soil was flooded ca. 1 week before basal fertilization and ploughed immediately after basal application. At topdressing, BS or chemical fertilizer was broadcast into the floodwater and soil was not ploughed. Irrigation was stopped ca. 2 weeks before harvest. Management schedules are shown in Table 2. At harvest each year, all the aboveground biomass of rice was removed by cutting from about 2 cm above the soil surface and the remaining stubble was incorporated into the soil by ploughing in the next year’s growing season.

Table 1 Nutrient contents in the applied biogas slurry

Year	2009	2010	2011	2012
Dry matter (g L^−1)	8.3	12.3	17.5	15.6
pH	8.05	8.07	7.73	7.76
EC (mS cm^−1)	18.5	33.2	15.1	31.2
Total C (g L^−1)	9.1	6.9	23.7	20.9
Total N (g L^−1)	3.7	5.0	5.3	5.7
NH4^+-N (g L^−1)	3.0	3.5	3.6	3.2
Total P (g L^−1)	0.34	0.34	0.58	0.39
Total K (g L^−1)	2.6	3.5	2.4	3.5
Total Cu (mg L^−1)	2.6	3.4	7.3	4.1
Total Zn (mg L^−1)	9.1	10.1	21.9	8.0

Note: EC, electrical conductivity; C, carbon; N, nitrogen; NH₄⁺-N, ammonium nitrogen; P, phosphorus; K, potassium; Cu, copper; Zn, zinc.

Table 2 Management schedules of this study

Year	Basal fertilization	Transplanting	First topdressing	Second topdressing	Harvest
2009	May 26	Jun 1	Aug 12	Sep 5	Nov 10
2010	Jun 1	Jun 3	Aug 13	Sep 5	Nov 11
2011	May 24	Jun 2	Jul 19	Aug 19	Nov 8
2012	May 29	Jun 5	Aug 4	Sep 1	Nov 13

Origin and chemical properties of biogas slurry

In this study, BS made from pig manure was used, since pig manure-derived BS did not enhance CH₄ emission in a paddy soil while cow manure-derived BS did (Sasada et al. 2011). BS used in this experiment was collected from a biogas plant in Aichi prefecture for 2009, 2011 and 2012 and in Hiroshima prefecture for 2010. NH₄⁺-N content was measured with the indophenol blue method after more than 1000 times dilution. For total C and N, BS was separated into the liquid and solid fractions by centrifugation (8000 × g, 5 min); total C and N are expressed as the sum of soluble organic C in the liquid fraction, measured with a TOC/TN meter (TOC-V CSH, Shimadzu Co., Kyoto, Japan), and total C and N in the solid fraction, measured with a CN corder (MT-700, Yanaco New Science Inc., Kyoto, Japan). For the measurement of total Cu, Zn and K, 1 g dry matter of BS was placed in a 200-mL conical flask and digested on a hot plate at 200°C by adding 0.2 mL concentrated sulphuric acid, 1 mL of nitric acid and 4 mL of 70% perchloric acid for about 3 h until white color appeared. Digestate was cooled and diluted into 25-mL flask by washing with 0.1 M hydrochloric acid (HCl) and then filtered with a filter paper (No. 5C, Tokyo Roshi Kaisha Ltd., Tokyo, Japan). Then, total Cu, Zn and K concentrations of each sample solution were measured by an atomic absorption spectrophotometer (Polarized Zeeman Atomic Absorption Spectrophotometer, Z-5010, Hitachi High-Technologies Co., Ltd., Tokyo, Japan). The same digestate was used to measure total P content by the ammonium molybdenum blue method.

Biomass production

At harvest each year, aboveground rice biomass was harvested from 1 m² (including about 20 hills) in each plot (8.64 m²) at a height of about 2 cm from the soil surface and their grain and leafy portions were weighed separately after oven drying at 70°C for about 4 d.

Gas sampling

Methane and N₂O flux was measured at 2–3-week intervals during the growing period by the closed chamber method (Win et al. 2010). For gas sampling, two rice hills were covered with a chamber, 30 cm length × 30 cm width × 100 cm height, connected with a battery-operated air-circulation inner fan and Tedlar® bag for mixing the air inside the chamber and regulating pressure. When the plant height became greater than the chamber height, the chamber was connected with another chamber with 30 cm length, 30 cm width and 50 cm height. Gas samples (30 mL) were taken with a 50-mL syringe at 0-, 10- and 20-min intervals after chamber placement and transferred into 10 mL pre-evacuated vials. The temperature inside the chamber was recorded using a micro-temperature thermometer (PC-9125, AS ONE Co., Tokyo, Japan) inserted into a small hole of the chamber through a rubber septum.

Both CH₄ and N₂O concentrations of the gas samples were quantified by a gas chromatograph (GC) with a flame ionization detector (FID, GC-14B, Shimadzu, Kyoto, Japan) for CH₄ or with an electron capture detector (ECD GC-14A, Shimadzu) for N₂O, respectively. For CH₄ gas analysis, the GC was equipped with a Porapak N (80/100 mesh, 3 mm diameter, 2 m length) column using helium (He) as carrier gas. The temperature of column, injector and detector were maintained at 80, 180 and 180°C respectively. For N₂O gas analysis, the GC was equipped with a Porapak Q (80/100 mesh, 3 mm diameter, 2 m length) column and argon containing 5% CH₄ was used as a carrier gas at a flow rate of 23 mL min⁻¹. The column and detector temperatures were maintained at 90 and 330°C, respectively (Yanai et al. 2007).

A closed chamber equation (Rolston 1986) was used to estimate both CH₄ and N₂O flux in each treatment as follows:

(1)

where Q is the CH₄ or N₂O fluxes (mg m⁻² min⁻¹), V is the volume of chamber (m³), A is the base area of chamber (m²), (∆c/∆t) is the increase or decrease rate of gas concentration (mg m⁻³) per unit time T (min), M is the molar weight of the gas and K is the Kelvin temperature of the air inside the chamber. Total emissions were calculated by multiplying the daily gas flux at each measurement for the time interval and summing up the values for the growing period (from June to October for 2009 and 2010, from June to September for 2011 and from June to November for 2012).

Soil sampling

Soil samples were collected from a ploughed layer (0 to 15 cm depth) at the beginning of this experiment (May 2009) and after harvesting each year (Dec 2009, Nov 2010, Dec 2011 and Dec 2012) with a root auger. Soil samples were collected from five points in each plot and a composite soil was made per plot. Samples were air dried and sieved through a 0.5-mm mesh sieve and stored at room temperature. Total C and N contents of the soil samples were measured with the CN corder. For available Cu and Zn, five grams of each soil sample was mixed with 50 mL of 1 M ammonium acetate (pH 7.0) and 0.1 N HCl, respectively, shaken for 30 min at 200 rpm and filtered with filter paper. For total Cu and Zn, 1 g of each soil sample was digested using the same digestion method mentioned above. Then Cu and Zn contents were measured with the atomic absorption spectrophotometer.

Heavy metal contents of rice plants

For the measurement of Cu and Zn contents of plants, two plants per plot were separately used. Their leafy and grain parts were separately pulverized with a motor operated grinder (Wonder Blender, WB-1, Osaka Chemical Co., Ltd., Japan) after oven drying at 70°C for about 48 h at harvesting time of each year. After digesting 1 g of each sample with the same digestion method mentioned above, the Cu and Zn concentrations were measured by the atomic absorption spectrophotometer.

Statistical analysis

Treatments and annual effects on biomass production, CH₄, N₂O, soil C and heavy metal contents in plant and soil were analyzed by one-way and two-way analysis of variance (ANOVA) and by least significant difference (LSD_0.05, Fisher) for mean comparison. The statistical analyses were carried out using the software Excel Statistics version 12 (SPSS Inc., Tokyo, Japan).

RESULTS

Biomass production

Biomass production of whole crop rice showed no significant difference between the same amounts of N treatments (CF100 and BS100) in each year (Fig. 1). Moreover, higher biomass production was obtained from BS300 in each year. It was significantly lowest (P < 0.01) in NF each year.

Figure 1 Plant biomass production (dry matter, DM) of a whole crop rice variety (Oryza sativa L. var. Leaf Star) under different fertilizer treatments; NF: no fertilization, CF100: chemical fertilizer [100 kg ammonium nitrogen (NH₄⁺-N) ha⁻¹], BS100: biogas slurry (100 kg NH₄⁺-N ha⁻¹), BS300: biogas slurry (300 kg NH₄⁺-N ha⁻¹). Different small letters show significant differences among the treatments in each year (P < 0.05) and different capital letters show significant differences among the years in each treatment (P < 0.05). Bars represent standard deviation of the mean (n = 3).

Methane and nitrous oxide emissions

During the growing period of each year, average CH₄ flux was very low in the first year (from June 30 to October 24). It increased by 10 times in the second year (from June 17 to October 16), 3.3 times from the second to the third year (from June 21 to September 23) and 1.3 times from the third to the fourth year (from June 23 to November 5) (Fig. 2). Mean total CH₄ emission during the growing seasons of this 4-year study period was significantly (P < 0.05) highest in BS300 (43.7 ± 18.4 g m⁻² season ⁻¹), followed by BS100 (32.0 ± 3.0 g m⁻² season ⁻¹), and then NF (23.5 ± 8.5 g m⁻² season⁻¹) and CF100 (20.3 ± 3.3 g m⁻² season ⁻¹). The difference between BS100 and CF100 was not significant (P < 0.05), but BS100 showed 1.6 times higher CH₄ emissions than CF100. Total CH₄ emissions increased from 2009 to 2011 in all the treatments and there was no increase in NF and CF100 from 2011 to 2012 while BS treatments showed further increase until 2012 (Fig. 3). The highest emissions in each treatment were found in the 2012 growing season (109 ± 61 g m⁻² in BS300 and 65.2 ± 6.1g m⁻² in BS100).

Figure 2 Seasonal variation of CH₄ flux during the growing season of (a) 2009, (b) 2010, (c) 2011 and (d) 2012 among fertilizer treatments; NF: no fertilization, CF100: chemical fertilizer [100 kg ammonium nitrogen (NH₄⁺-N) ha⁻¹], BS100: biogas slurry (100 kg NH₄⁺-N ha⁻¹), BS300: biogas slurry (300 kg NH₄⁺-N) ha⁻¹). Bars represent standard deviation of the mean (n = 3). Solid arrows indicate 1^st top fertilizer dressing and dotted arrows indicate 2^nd top fertilizer dressing.

Figure 3 Total methane (CH₄) emission (for the growing period of each year) from 2009 to 2012 among fertilizer treatments; NF: no fertilization, CF100: chemical fertilizer [100 kg ammonium nitrogen (NH₄⁺-N) ha⁻¹], BS100: biogas slurry (100 kg NH₄⁺-N ha⁻¹), BS300: biogas slurry (300 kg NH₄⁺-N ha⁻¹). Different lower case letters show significant differences among the treatments in each year (P < 0.05) and different capital letters show significant differences among the years in each treatment (P < 0.05). Bars represent standard deviation of the mean (n = 3).

Both the maximum positive N₂O flux (5.5 ± 7.8 mg N₂O m⁻² day⁻¹) and the maximum negative flux (–3.7 ± 4.4 mg N₂O m⁻² day⁻¹) were observed in the BS300 treatment in October 2010 and June 2011, respectively (Fig. 4). The averages of 4 years of N₂O fluxes were 0.12 ± 1.20, 0.22 ± 1.49, – 0.16 ± 1.23 and 0.26 ± 1.33 mg N₂O m⁻² day⁻¹ in NF, CF100, BS100 and BS300, respectively.

Figure 4 Seasonal variation of N₂O flux during the growing season of (a) 2009, (b) 2010, (c) 2011 and (d) 2012 among fertilizer treatments; NF: no fertilization, CF100: chemical fertilizer [100 kg ammonium nitrogen (NH₄⁺-N) ha⁻¹], BS100: biogas slurry (100 kg NH₄⁺-N ha⁻¹), BS300: biogas slurry (300 kg NH₄⁺-N) ha⁻¹). Bars represent standard deviation of the mean (n = 3). Solid arrows indicate 1^st top fertilizer dressing and dotted arrows indicate 2^nd top fertilizer dressing.

Soil C contents

In all treatments, soil C contents increased from the first to the fourth year. Increased C contents from the first to the fourth year were 3.7 ± 0.5, 5.5 ± 1.4, 7.3 ± 1.2 and 6.2 ± 2.9 g kg⁻¹ in NF, CF100, BS100 and BS300, respectively. BS treatments showed higher C contents than NF and CF100, except BS300 in 2012 which showed lower C content than CF100. The NF treatment showed the lowest in all the years, but there was no significant difference among the treatments in each year (Fig. 5).

Figure 5 Changes of soil total carbon (C) contents from 2009 to 2012 among fertilizer treatments; NF: no fertilization, CF100: chemical fertilizer [100 kg ammonium nitrogen (NH₄⁺-N) ha⁻¹], BS100: biogas slurry (100 kg NH₄⁺-N ha⁻¹), BS300: biogas slurry (300 kg NH₄⁺-N) ha⁻¹). Soil sampling were done on December 2009, November 2010, December 2011 and December 2012. Different lower case letters show significant differences among the treatments in each year (P < 0.05) and different capital letters show significant differences among the years in each treatment (P < 0.05). Bars represent standard deviation of the mean (n = 3).

Heavy metal contents in rice plants and soil

There was no significant difference in the Cu contents in either the grain or stem/leaves portions of the harvested rice plants among the treatments in each year (Fig. 6a). No significant differences in the Zn contents in either the grain or stem/leaves portions were found among the treatments in each year, except in 2011 in which significantly higher Zn content in the grain portion was found in BS300 than in NF and CF100 (Fig. 6b). Although the differences were not significant, the Cu and Zn contents in both the grain and stem/leaves portions showed slightly higher values in BS300 than in the other treatments in most years. No significantly higher Cu and Zn contents were found in either the grain or stem/leaves in the later years, when compared to the start time.

Figure 6 (a) Copper (Cu) and (b) zinc (Zn) contents of rice (Oryza sativa L. var. Leaf Star) grain and stem/leaves portions during four-year study period. Different lower case letters show significant differences among the treatments in each year (P < 0.05) and different capital letters show significant differences among the years in each treatment (P < 0.05). Bars represent standard deviation of the mean (n = 3). NF: no fertilization, CF100: chemical fertilizer [100 kg ammonium nitrogen (NH₄⁺-N) ha⁻¹], BS100: biogas slurry (100 kg NH₄⁺-N ha⁻¹), BS300: biogas slurry (300 kg NH₄⁺-N) ha⁻¹).

Soil exchangeable Cu (1M ammonium acetate extraction) contents increased from the first to second year, but there was no further increase in the later years and also there were no significant differences among the treatments in each year (Fig. 7a). Although soil exchangeable Zn contents showed no significant differences among the treatments in each year, the 4-year average concentration was 11 and 19% higher in BS100 and BS300, respectively, than in CF100. BS100 showed significantly higher soil exchangeable Zn contents than NF only in 2011. BS300 showed significantly higher exchangeable Zn contents than NF in 2011 and 2012 (Fig. 7b).

Figure 7 Exchangeable forms (extracted with 1M ammonium acetate, pH 7) of (a) copper (Cu) and (b) zinc (Zn), [(extracted with 0.1N hydrochloric acid (HCl)] of (c) Cu and (d) Zn, total forms (digestion with strong acid) of (e) Cu and (f) Zn contents of soil at 0–15 cm depth during four-year study period. Different lower case letters show significant differences among the treatments in each year (P < 0.05) and different capital letters show significant differences among the years in each treatment (P < 0.05). Bars represent standard deviation of the mean (n = 3). NF: no fertilization, CF100: chemical fertilizer [100 kg ammonium nitrogen (NH₄⁺-N) ha⁻¹], BS100: biogas slurry (100 kg NH₄⁺-N ha⁻¹), BS300: biogas slurry (300 kg NH₄⁺-N) ha⁻¹).

Soil 0.1N HCl extractable forms of Cu contents significantly increased from the first to second year in all the treatments, but there was no further increase in the later years and also there were no significant differences among the treatments in each year (Fig. 7c). In 2009, 0.1N HCl extractable forms of Zn contents in soil were significantly higher in BS300 than in CF100 and NF, but there was no significant difference between BS100 and CF100. There were no significant differences in Zn contents among the treatments in the later years (Fig. 7d).

Soil total Cu and Zn contents showed no significant differences among the treatments in each year. Moreover, all treatments showed no yearly increase, except Cu contents in the BS300 treatment, which were significantly higher in 2010 and 2012 than in 2009 (Fig. 7e–f).

DISCUSSION

Biogas slurry gave high biomass production similar to chemical fertilizer, and higher biomass production was obtained from a greater application rate (BS300), suggesting that BS is an alternative organic fertilizer to whole crop rice.

In this study, average CH₄ emission during the 4-year growing period was 1.6 times higher in BS100 than in CF100, under conditions of the same N application rate, and a higher amount of BS application (BS300) caused 2.1 times higher CH₄ emission than CF100, suggesting a potential risk of CH₄ emission in BS application. The input C from BS may cause higher CH₄ emission in the BS treated soils, because it is well known that CH₄ emission is affected by the application rate of organic C (Naser et al. 2007). However, this result contradicts our previous paper reporting on experiments conducted in lysimeters (Sasada et al. 2011). In that study, CH₄ emission was not significantly different under the same application rate (300 kg N ha⁻¹) between in a treatment with BS deriving from pig slurry with low organic matter content and one with chemical fertilizer. The difference in the CH₄ emissions between the lysimeter study and this field study might derive from the difference in the soil C contents because soil C is highly correlated with CH₄ emission (Minamikawa et al. 2005). The original soil C contents ranged from 35 to 38.5 g kg⁻¹ in the lysimeter study while they were 28 g kg⁻¹ in the field. Thus, additional C input through BS to the field soil with low original soil C content might have a higher stimulating effect on CH₄ emissions. Moreover, the different result from the lysimeter study and this study might be caused not only by the original soil carbon concentrations but also by other conditions such as management practices, weather condition, weed growing condition, plant growth performance and root exudation capacity, changes in soil microbial activities and so on. However the results found from this field study would be more important for their reliability than those from the lysimeter study.

The highest CH₄ emission was observed in the fourth year in BS treatments during the growing season (65 and 109 g CH₄ m⁻² in BS100 and BS300), but it was still comparable to average CH₄ emission values (72 g m⁻²) in paddy fields applied with rice straw and higher than those (20 g m⁻²) with compost (Kanno et al. 1997). According to their study, compost application did not increase CH₄ emission compared with chemical fertilizer. Therefore, BS is considered a similar type of organic matter to rice straw, rather than similar to compost. However, caution is required for the cumulative CH₄ emissions because our gas sampling frequency was low in this study.

Methane emissions increased from the first year up to the third year in all treatments and until the fourth year in the BS treatments in this study. Increased CH₄ emission in successive cultivations was also reported in other studies (Yang and Chang 2001; Eusufzai et al. 2010). Kumagai and Konno (1998) reported that 17 g CH₄ m⁻² in the first year increased to 93 g CH₄ m⁻² in the second year after changing an upland field to a flooded paddy field. Cultivation was started in 2009 in the field used in this study and before 2009, the field had not been cultivated with any crop for more than 10 years, and this site had been used as a greenhouse for storing field equipment. The increased CH₄ emission in the later years might be due to the successive cultivation of submerged rice plants (Kanno et al. 1997) in the non-cultivated soil and thereby to the return of stubble and root residues every year, making the soil physically better and favorable to microbial activity (Fraser et al. 1994).

Nitrous oxide emission was very low compared with CH₄ emission in this study. The N application rate was determined based on the NH₄⁺-N contents, and thus total N application rate ranged from 123 to 178 kg ha⁻¹ in BS100. Higher total N application rates in BS100, however, caused no significantly higher N₂O emissions compared with those in CF in this study. In general, large N₂O emissions are observed in mid-season drainage, and a trade-off between CH₄ and N₂O emissions has been well documented in paddy fields (Cai et al. 1997; Zou et al. 2005; Zhang et al. 2010). Since mid-season drainage was not conducted in this study, this may be the reason for the very low N₂O emission.

Soil total C increased in all the treatments from the beginning to the fourth year. It increased by 3.7 and 5.5 g C kg⁻¹ soil within 4 years even in NF and CF100, respectively, in which no organic matter had been applied. It further increased by 1.7 and 0.7 g C kg⁻¹ soil in BS100 and BS300 compared with that in CF100. The amount of C input from BS during 4 years was about 1.2 g kg⁻¹ in BS100 in this study and a majority of the C would be decomposed during the cultivation period. Thus, most of the increased soil C observed in this study might be from the returning crop residues (root and stubble) every year as well as input C through BS. Generally, soil C sequestration is investigated in long-term studies. Some papers report that soil C increases linearly (Kong et al. 2005; Majumder et al. 2008; Li et al. 2010; Ghosh et al. 2012), whereas others observe a logarithmic correlation (Fan et al. 2005; Cai and Qin 2006; Pan et al. 2006; West and Six 2007). These different relations demonstrate the differences in C sequestration efficiency [i.e. ∆ Soil Organic Carbon (SOC)/∆C input] and C saturation limit (Yan et al. 2013). Also, an increase or decrease of SOC may be related to its initial value (Bolinder et al. 2010) and other factors such as soil physical, chemical and biological properties, management practices and climatic conditions (Six et al. 2004; Bronick and Lal 2005). Moreover, no increase or a slight increase of SOC might be due to the decreasing C sequestration efficiency in the case that the SOC approaches the saturation level (Six et al. 2002; Steward et al. 2007), especially in the later years in long-term studies. The amounts of increased soil C in this study seem to be much higher than those in other studies (Gao et al. 2008; Nayak et al. 2009; Zhang et al. 2012). However, such high increases in SOC are also observed in the study of Jiang et al. (2013), who reported a higher and linear increase in SOC (ca. 0.7 and 1.3 g C kg⁻¹ yr⁻¹ in no fertilization and NPK plus pig manure treatments, respectively) in the first four years during their 27-year study period in a rice-rice cropping system. In addition, Dong et al. (2012) found a high increase (ca. 10.2 g kg⁻¹) in SOC within 6 years of pig manure application in a paddy field although the control treatment showed relatively stable C contents. Moreover, even after the cessation of cultivation, Shimoda and Koga (2013) observed a high increase of soil C storage (ca. 5 g kg⁻¹) during 3 years after abandoning a paddy field. In our study, the high increase of soil C within the first 4 years is reasonable because of starting cultivation in non-cultivated field, by which soil physical and microbial conditions are improved (Styla and Sawicka 2009). In addition, the conversion of upland soil to paddy soil increases SOC due to its anaerobic conditions (Liping and Erda 2001; Pan et al. 2004). However, investigation of soil C for more successive years is desirable to evaluate the sustainability of this agricultural system and to determine the effect of cropping and BS application on soil C sequestration.

Some heavy metals are used as essential additives for feed production and, in particular, Cu and Zn are often used in relatively high doses because of their growth-promoting effects (Eckel et al. 2008; Poulsen and Carlson 2008). Among slurries, pig slurry contains higher amounts of heavy metals, especially Cu and Zn (L’Herroux et al. 1997; Nicholson et al. 1999). Therefore, the application of pig slurry might increase a potential risk for heavy metal uptake by plants and also accumulation in soil. The application of a lower amount of BS (BS100) had no risk of Cu and Zn uptake by rice plants until the fourth year of successive application. But a higher amount of BS (BS300) showed slightly higher Cu and Zn contents in both grain and stem/leaves portions of rice in most years, indicating a potential risk for Zn uptake in a high dosage of BS application. However, the highest values (14.7 mg kg⁻¹ for Cu and 40.0 mg kg⁻¹ for Zn) found in the stem/leaves portions of BS300 in this study were still very much lower than the upper limits (Cu = 125 mg kg⁻¹ and Zn = 120 mg kg⁻¹: Mori 2010) for the additives of animal feed in Japan. Besides, though the high dosage of BS showed higher uptake than the other treatments, there was no tendency for Cu and Zn contents in the grain and stem/leaves to increase each year. Rather, they decreased in the stem/leaves in 2012. These results indicate few risks for the enrichment of these elements by the successive application of BS.

Both exchangeable (extracted with 1M ammonium acetate, pH 7 and 0.1N HCl) forms of Cu contents in soil increased from the first to second year, but did not increase in the later years in all the treatments, and there were no significant differences among the treatments each year, suggesting no risk of Cu accumulation in soil during the 4-year period. Soil exchangeable Zn (1M ammonium acetate extraction) contents were higher in the BS treatments than in CF100, but there were no significant differences in each year. Only in 2009, the 0.1N HCl extractable Zn content of soil was significantly higher in BS300 than in CF100, but there was no difference between BS100 and CF100 and also no difference among treatments in the later years, indicating no risk of Zn accumulation from BS application compared with chemical fertilizer. Moreover, according to the Agricultural Land Soil Pollution Prevention Law of Japan, the upper limit for Cu for paddy field soil is recorded as 125 mg kg⁻¹ (0.1N HCl extraction) (Makino et al. 2010) and our value was still under this limit.

Total Cu and Zn contents in soil also showed no significant differences among the treatments each year and no increase from the first to the later years in most treatments, except in BS300, in which significantly higher total Cu contents were observed in 2010 and 2012 than in 2009, but not significantly higher than those in CF100 at the respective year, indicating no risk of Cu and Zn accumulation in soil from 4 years’ successive application of BS. Moreover, the concentrations of total Cu and Zn found in this study were within the range of the native abundance in Japanese soil (0.9 to 234.9 mg kg⁻¹ for Cu and 2.5 to 331 mg kg⁻¹ for Zn; Yamasaki et al. (2001). While Cu and Zn are essential elements for plants, accumulation in soil can be poisonous to microorganisms, higher plants and fauna. Although soil Zn contents (170–203 mg kg⁻¹) in this study were still under the upper limit of the maximum permitted concentrations for soil [150–300 mg kg⁻¹ (CEC 1986)], total Cu (104–144 mg kg⁻¹) contents reached the upper limit [50–140 mg kg⁻¹ (CEC 1986)]. However, the high concentration of Cu found in this study was not derived from BS application because no difference was observed between BS treatments and CF100 in this study. Since Cu and Zn contents in the groundwater were not measured, these measurements should be included in further studies because groundwater quality is affected by leachate percolation (Mor et al. 2006; Reddy et al. 2012).

In this study, we did not evaluate other environmental impacts such as ammonia (NH₃) volatilization or N leaching deriving from BS application. Our previous studies have demonstrated that NH₃ volatilization from the application of BS (generated from cow manure) could be mitigated by the simultaneous application of wood vinegar or by keeping a deep flood water level (10 cm) (Win et al. 2009). The field in this study was kept under continuous flooding conditions throughout the growing seasons and thus the NH₃ volatilization risk may be alleviated by keeping deep floodwater conditions at BS applications. This management did not enhance CH₄ emissions (Win et al. 2010), indicating that keeping deep floodwater is an effective way of decreasing NH₃ volatilization without stimulating CH₄ emission. Moreover, very low nitrate concentrations (< 0.5 mg N L⁻¹) were observed in the drainage water in a lysimeter applied with 300 kg N ha⁻¹ of BS (generated from pig and cow manure) (Sasada et al. 2011). Thus, risk of polluting the water environment with BS application may be very low, but further investigation is essential for nitrate (NO₃^–) leaching if BS is applied for the long term.

The present study revealed that the application of BS to paddy fields enhances the CH₄ emission. However, there was no risk of heavy metal uptake by plants and also no accumulation in soil in 4 years’ successive application of BS in comparison with the same dosage of N application with chemical fertilizer. Moreover, higher biomass production was obtained from high dosage of BS application. Thus, the application of BS, generated from pig manure, to rice fields at the conventional rate (100 kg N ha⁻¹) could be considered to substitute for chemical fertilizer utilization without additional environmental impacts in greenhouse gas emission and heavy metal uptake. But for long-term usage, it is necessary to further investigate environmental impacts, especially CH₄ emission.

ACKNOWLEDGMEMTS

This study was supported by Japan Science and Technology (JST), the National Natural Science Foundation of China (NSFC, No. 40821140540), a Grant-in-Aid for Scientific Research (No. 19201018), and the Green Biomass Research for Improvement of Local Energy self-sufficiency from the Ministry of Education, Science, Sports and Culture of Japan. The authors thank Dr. Masanori Okazaki, Tokyo University of Agriculture and Technology (TUAT), for permitting C, N, Cu and Zn analysis, and for invaluable suggestions. We also gratefully thank Mr. Akira Watanabe, Ebara Co., and Taiyo Kosan Ltd., for providing the biogas slurries.