Effect of the long-term application of anaerobically digested residual slurry on methane emissions in a rice paddy field

ABSTRACT

The anaerobic digestion of livestock manure is an environmentally compatible technology used for the production of renewable energy. Anaerobically digested residual slurry has been used worldwide as a liquid fertilizer in both upland and paddy fields. However, a controversial question remains as to whether the application of slurry to rice paddy fields increases methane emissions; although methane is one of the most prevalent greenhouse gases, little is known about the effects of the long-term application of residual slurry on methane emission. In this study, we repeatedly applied slurry to a paddy field for six years at different application rates (10, 15, and 20 g N m⁻² based on ammonium-nitrogen content). At the fifth and sixth years of application, we evaluated the effect in terms of methane flux and soil total carbon content. The effect of the long-term application of the slurry (10 g N m⁻²) on grain yield was equivalent to that of chemical fertilizer (10 g N m⁻²). The application of the residual slurry was likely to increase the cumulative methane emissions during rice growing season in both 2006 and 2007. On the other hand, we observed that soil total carbon did not accumulate significantly in the soil. Thus, we cannot rule out the potential risk of additional methane emissions caused by the application of the residuary slurry to paddy fields.

KEYWORDS:

• Anaerobic digestion
• methane emission
• paddy field
• residual slurry
• soil carbon dynamics

1. Introduction

Anaerobic digestion is gaining popularity worldwide as a way to produce renewable energy by fermenting livestock manure in biogas plants. However, a serious question has arisen concerning the treatment of residual slurry, which has been used as liquid fertilizer due to its high ammonium-nitrogen (NH₄⁺-N) content. Previous studies in upland fields have demonstrated that the effect of the application of slurry on plant growth and grain yield is equivalent or superior to chemical fertilizer for corn (Loria et al. 2007; Morris and Lathwell 2004) and forage production (Saunders et al. 2012).

In East Asia, recent rapid economic growth has increased the demand for meat and dairy products (Thornton 2010). To treat dairy and municipal organic waste, developed and developing Asian countries have begun to construct large-scale biogas plants (Lu et al. 2012; Nishikawa et al. 2012). Currently, large areas of paddy fields, which serve as the main source for staple foods, are widely distributed across Asia. Furthermore, the food demands for livestock products are expected to nearly double in sub-Saharan Africa and South Asia from 2000 to 2050 (Thornton 2010). In sub-Saharan Africa, strong demand has driven an increase in rice cultivation area (Balasubramanian et al. 2007). Thus, the application of anaerobically digested residual slurry to paddy fields is likely inevitable and has attracted attention as a way to reduce the application of chemical fertilizer (Chen et al. 2013; Zhang et al. 2015).

The mitigation of methane emission from paddy fields is another crucial topic, as it represents an important category of anthropogenic methane emissions (IPCC 1994). Methane is one of the most notorious greenhouse gases, responsible for a 34-fold higher global warming potential than carbon dioxide (IPCC 2013). Many studies have indicated that the incorporation of organic material into paddy soils increases methane emissions because of the accumulation of carbon (C), which is a substrate for methanogens (Chen et al. 2011; Watanabe et al. 1999; Yagi and Minami 1990). Although several studies have indicated that the application of residual slurry might cause a potential risk of methane emissions from paddy fields (Sasada et al. 2011; Singla and Inubushi 2014; Win et al. 2010; Win et al. 2014), a few studies have shown that the application of residual slurry decreased the methane emissions (Singla et al. 2015; Singla and Inubushi 2013). Therefore, the results have been contradictory.

Nishikawa et al. (2012) demonstrated that residual slurry can be continuously used as an alternative to chemical fertilizer under appropriate fertilization conditions, corresponding to an application rate of approximately 15 g N m⁻². However, the application rate of residual slurry is primarily dependent on plant nutritional requirements, especially the crop N requirements, to maximize grain yield; little attention has been paid to the C content. Furthermore, decomposition-resistant organic materials derived from residual slurry may accumulate and degrade slowly from a long-term perspective. Therefore, long-term experiments are needed to fully understand the impact of the application of residual slurry on C dynamics and methane flux.

We conducted long-term application of residual slurry with different application rates in a paddy field for seven years. The methane flux was estimated by the closed chamber method in the field at the sixth year of repeated application. This study aimed to answer the question of whether the continuous application of residual slurry has the cumulative effect on methane emissions from a paddy field.

2. Materials and methods

2.1. Experimental design

To determine the effect of the long-term application of residual slurry on crop productivity and methane emission, a field experiment was conducted from 2002 to 2007 at a paddy field in the Experimental Farm of Kyoto University located in Osaka, Japan (34°51ʹ16″ N, 135°37ʹ46″ E), as described by Nishikawa et al. (2012; 2013; 2014). The soil at the site was characterized as sandy gray lowland soil, classified as a Gleysol according to the Food and Agriculture Organization of the United Nations’ soil taxonomy, with a particle composition of 68.4% sand, 17.5% silt, and 14.1% clay. Just prior to transplanting in 2002, soil total N content ranged from 1.70 to 2.09 g N kg⁻¹; available N content ranged from 80.6 to 100.5 mg N kg⁻¹; and available P ranged from 993 to 1112 mg P₂O₅ kg⁻¹ (Bray-2) (Nishikawa et al. 2012). The annual mean temperature and precipitation around the experimental farm are 17.0°C and 1260 mm, respectively. In 2001, rice was cultivated uniformly all over the field to eliminate any residual effects of the previous crop in 2000. In 2001, the amount of applied N, P, and K was 10 g N m⁻² as ammonium sulfate, 10 g P₂O₅ m⁻² as calcium superphosphate, and 10 g K₂O m⁻² as potassium chloride, respectively.

Five annual N application treatments with four replicates were carried out in the paddy field in a randomized block design: three application rates of residual slurry, one application rate of chemical fertilizer (i.e. chemically fertilized), and one application without N fertilizer (i.e. unfertilized). The fertilizer was supplied in a single application (only basal application). Our primary objective was to estimate cumulative effect of the continuous application of residual slurry. We assumed that split application might cause instant methane emissions. Thus, only basal application was examined for methane emissions. Each replicate plot area was 20 m² (5 m × 4 m) in area and was separated with plastic sheets to avoid cross-contaminations among plots. The ‘Hinohikari’ cultivar, which is widely cultivated throughout southwestern Japan, was used in this study. The residual slurry used in this study was provided from the Nantan City Yagi Bioecology Center, Nantan city, Kyoto Prefecture, Japan. Dairy cattle excreta accounted for 75–80% of the organic waste (by weight) treated at this biogas plant. The physiochemical composition of residual slurry is shown in Table 1. The NO₃-N content in the residual slurry was negligible. The recommended N application rate for the cultivation of this variety around the region where the study was undertaken is approximately 10 g m⁻². Three plots received residual slurry application rates of 10, 15, and 20 g N m⁻², respectively, on the basis of NH₄⁺-N content. Thus, the amount of applied C in the residual slurry plots might be variable. In 2007, the three residual slurry plots (10, 15, and 20 g N m⁻²) received C application rates of 12.9, 19.4, and 25.9 g C m⁻², respectively. The amount of applied N in the chemically fertilized plot was 10 g N m⁻² in the form of ammonium sulfate. In both the chemically fertilized and unfertilized plot, the amount of applied P and K was 10 g P₂O₅ m⁻² as calcium superphosphate and 10 g K₂O m⁻² as potassium chloride, respectively. During the experimental period from 2002 to 2009, the experimental conditions were maintained as described earlier.

Table 1. Physiochemical compositions of the residual slurry of anaerobic digestion.

Year	pH	NH4^+-N (g L^−1)	Total N (g L^−1)	P2O5 (g L^−1)	K2O (g L^−1)	Total C (g L^−1)
2006	8.40	2.20	2.70	0.38	4.50	n.a.
2007	7.80	1.39	1.40	n.a.	n.a.	1.80
Mean value during 2002–2007^a	8.16	1.91	2.84	0.56	2.33	n.a.

n.a. indicates no data available.

^aThe mean values of P₂O₅ and K₂O are calculated using data during 2002–2006.

In 2006, three rice seedlings per hill were transplanted with a planting density of 14.4 hills m⁻² (33 cm × 21 cm) on 13 June. Three days after transplanting (16 June), the basal application of residual slurry and chemical fertilizer was conducted. In 2007, transplanting was conducted on 12 June, and basal application was conducted on 14 June. All experimental plots were kept submerged at 3–5 cm water depth until the heading stage (25 August), and intermittent irrigation was employed thereafter in 2006. However, the intermittent irrigation was employed after the beginning of August 2007 because of insufficient water supply caused by lack of adequate water resources. The final drainage was operated on 1 October, and rice plants were harvested on 12 October. Mid-season drainage was not operated. Pest management practices were conducted in accordance with neighboring farmers’ practices. Harvested plant residues were not returned to the experimental field.

2.2. Measurement of methane emission

We evaluated the effect of the long-term application of residual slurry on methane emissions in 2006 and 2007, which corresponded to the fifth and sixth years of continuous application, respectively. Measurement of methane emission in 2006 was conducted at unfertilized, chemically fertilized, and residual slurry (10 g N m⁻²) plots on 6 and 26 July; 1, 17, and 29 August; and 8 September 2006. Measurement of methane emission in 2007 was conducted at unfertilized, chemically fertilized, and residual slurry (10, 15, and 20 g N m⁻²) plots on 20 June, 12 and 27 July, 15 and 29 August, and 12 September 2007. Gas samples were collected by the closed chamber method (Yagi and Minami 1990) using a chamber (45 cm × 60 cm with a height of 90 cm) with a 6 V battery-operated inner fan and a Tedlar bag to obtain homogeneous samples. The temperature in the chamber was recorded by a digital thermometer (TR-52, T&D, Nagano, Japan). The chambers covered four rice hills, and gas samples were taken five times every 5 min. The gas samples were immediately transferred into 20 mL vials using a portable air pump (MP-2N, Shibata Scientific Technology, Saitama, Japan). On 13 June 2006 and 12 June 2007, wooden frames were installed to avoid physical pressure by placing the chambers, followed by gas emission. Gas sampling was conducted at daytime (9:00–13:00).

The concentration of methane in the collected gas samples was analyzed by a gas chromatograph with a flame ionization detector (GC-14B, Shimadzu, Kyoto, Japan) equipped with a Porapak N (80/100 mesh) column under the following conditions: helium as a carrier gas, a column temperature of 50°C, an injection temperature of 100°C, and a detector temperature of 100°C.

The methane emission rate was calculated as follows:

where ρ is the gas density (mg m⁻³) (0.7153 × 10⁶), V is the headspace volume (m³) of the chamber, A is the base chamber area (m²), ΔC/ΔT is the change in the gas concentration (m³ m⁻³ h⁻¹), and K is the absolute temperature of the air inside the chamber. The cumulative methane emission was calculated from the linear interpolation of daily methane emission over rice-growing season from transplanting to harvesting (121 days).

2.3. Soil sampling and physiochemical analyses

Soil redox potential (Eh) in the depth of 5 cm was monitored using a potable potentiometer (PRN-4 and EHS-120, Fujiwara Scientific, Tokyo, Japan).

Just prior to transplanting in 2002, 2006, and 2007, three soil subsamples (approximately 300 g based on fresh weight) were collected from the plow layer (0–15 cm) of each plot. Soil samples from all three subsamples in the same plot were mixed thoroughly to obtain composite samples. The composite samples were dried, ground to pass through a 0.7 mm sieve, and stored at room temperature under dark conditions until analysis. Soil total C content was analyzed by dry combustion method using an elemental analyser (EA1108, Fisons, Milan, Italy).

2.4. Grain yield survey

At the maturation stage on 2007, 15–20 hills of rice exhibiting average growth were harvested from all the experimental plots. Threshed grains were separated into fully ripened grains and other grains using a 1.06 g mL⁻¹ saline solution. The grain yield was defined as the weight of fully ripened brown rice at a water content of 15.5%. The dry weight was determined after oven drying at 80°C to a constant weight.

2.5. Statistical analyses

All statistical analyses were performed using R version 3.2.4 (R Development Core Team 2016). One-way analysis of variance (ANOVA) was performed followed by Tukey’s test to determine the differences in the methane flux and soil total C content for comparison among the treatments. Differences in the methane flux were analyzed using two-way ANOVA for comparison among the treatments or sampling time. A P-value <0.05 was considered to be statistically significant for all analyses.

3. Results

3.1. Methane emission

Seasonal variations in methane flux from the paddy field in 2006 are shown in Fig. 1A. We observed similar seasonality among all plots, irrespective of fertilizer treatments. After transplantation, the methane flux gradually increased and reached a maximum value in late July, but rapidly dropped in August. We observed higher methane flux in residual slurry plot (10 g N m⁻²) than either unfertilized or chemically fertilized plot (10 g N m⁻²) every sampling time except 29 August 2006. However, the result of two-way ANOVA showed no significant differences among the three different fertilizer treatments, and significant differences in methane flux among the six sampling times (P < 0.001). The result of one-way ANOVA also showed no significant differences among the three different fertilizer treatments within each sampling time. Seasonal variations in methane flux from the paddy field in 2007 are shown in Fig. 1B. We observed seasonal change that is similar to those seen in 2006. The result of two-way ANOVA showed significant differences in methane flux among the six sampling times (P < 0.001) and no significant differences among the five different fertilizer treatments. The result of one-way ANOVA also showed no significant differences among the five different fertilizer treatments within each sampling time.

Figure 1. Seasonal variations in methane flux in 2006 (A) and 2007 (B). Error bars on columns indicate standard deviations (N = 4). The arrows indicate major field management events (Tr: transplanting; In: beginning of intermittent irrigation; Fd: final drainage; Ha: harvesting).

The cumulative methane emissions over 121-d rice-growing season are shown in Table 2. The cumulative methane emission of the residual slurry plot was approximately 1.5 times higher than that of the unfertilized and chemically fertilized plots in 2006, although the differences were not significant. The cumulative methane emission obtained from 2007 showed a tendency similar to that from 2006, although the differences among the five different fertilizer treatments were not significant.

Table 2. Cumulative methane emissions in each treatment.

Treatment	Cumulative methane emissions (g CH4 m^−2)
	2006	2007
Unfertilized (0 g N m^−2)	20.4 ± 0.8 a	14.1 ± 2.7 a
Chemically fertilized (10 g N m^−2)	23.7 ± 7.5 a	14.7 ± 3.2 a
Residual slurry (10 g N m^−2)	31.4 ± 1.1 a	17.4 ± 2.6 a
Residual slurry (15 g N m^−2)	n.a.	17.1 ± 5.1 a
Residual slurry (20 g N m^−2)	n.a.	20.6 ± 5.9 a

Data are means ± SD (n = 4). n.a. indicates no data available. Different lowercase alphabets indicate significant differences among the treatments in each year (P < 0.05).

Seasonal variations in soil Eh in 2006 and 2007 are shown in Fig. 2. Soil Eh showed low values (approximately −200 mV) until late July, but increased from August. We observed that methane emission occurred just in lower soil Eh levels (approximately −200 mV).

Figure 2. Seasonal variations in soil redox potential (Eh) in 2006 (A) and 2007 (B). Error bars on columns indicate standard deviations (N = 4).

3.2. Soil carbon accumulation

Annual variations in soil total C content are shown in Table 3. The soil total C content exhibited slightly higher values at residual slurry plots in 2006. However, one-way ANOVA showed no significant differences in soil total C content among the five fertilizer treatments.

Table 3. Periodical changes in soil total carbon contents sampled from plow layer (0–15 cm) before transplanting in 2002, 2006, and 2007.

Treatment	Total carbon content (g C kg^−1)
	2002	2006	(∆)	2007	(∆)
Unfertilized (0 g N m^−2)	22.3 ± 2.0	22.2 ± 0.3	(0.0 ± 2.3)	23.6 ± 1.3	(1.4 ± 0.7)
Chemically fertilized (10 g N m^−2)	21.4 ± 2.8	24.3 ± 0.0	(2.9 ± 2.8)	25.6 ± 3.3	(4.2 ± 0.5)
Residual slurry (10 g N m^−2)	19.8 ± 0.1	23.9 ± 1.5	(4.1 ± 1.4)	24.0 ± 1.8	(4.2 ± 1.7)
Residual slurry (15 g N m^−2)	22.9 ± 0.1	26.9 ± 3.6	(4.0 ± 3.7)	27.3 ± 3.5	(4.3 ± 3.5)
Residual slurry (20 g N m^−2)	22.5 ± 0.5	26.2 ± 4.1	(3.7 ± 3.6)	27.0 ± 4.2	(4.5 ± 3.7)
One-way ANOVA	n.s.	n.s.	n.s.	n.s.	n.s.

Data are means ± SD (n = 4). ∆ indicates total carbon content increased from 2002 to each year. n.s. indicates no significant differences among treatments in each year (P > or = 0.05).

3.3. Grain yield and yield components

The grain yield and its components in 2007 are shown in Table 4. The number of ears and spikelets were both significantly higher in the chemically fertilized and residual slurry plots than in the unfertilized plot. Consequently, grain yield was also significantly higher in the chemically fertilized and residual slurry plots than in the unfertilized plot. Although the grain yield of the chemically fertilized plot was slightly higher than that of the residual slurry plots, the differences were not statistically significant.

Table 4. Grain yield and yield components in each treatment.

Treatment	Ear number (m^−2)	Spikelet number (10^3 m^−2)	Ripened grain (%)	1000 grain weight (g)	Grain yield (g m^−2)
Unfertilized (0 g N m^−2)	241 b	22.2 b	86.4 a	21.7 a	417 b
Chemically fertilized (10 g N m^−2)	325 a	31.4 a	86.5 a	21.4 a	582 a
Residual slurry (10 g N m^−2)	308 a	31.0 a	84.5 a	21.5 a	562 a
Residual slurry (15 g N m^−2)	312 a	30.1 a	85.0 a	21.6 a	552 a
Residual slurry (20 g N m^−2)	312 a	31.2 a	84.1 a	21.9 a	573 a

Different lowercase alphabets indicate significant differences (P < 0.05).

4. Discussion

Methane emissions frequently exhibited higher values in residual slurry plot (10 g N m⁻²) in 2006 (Fig. 1A), and there was a tendency that the application of residual slurry increased the cumulative methane emissions in both 2006 and 2007 (Table 2). Thus, we cannot entirely exclude the possibility that the application of residual slurry might cause additional methane emissions although there were no significant differences. The potential risk of additional methane emissions caused by the application of residual slurry is presented by a previous study by Win et al. (2014). According to their four-year consecutive study, methane emissions were not significantly different in three out of four years in between residual slurry treatment (10 g N m⁻²) and the chemically fertilized treatment, but a significant increase was seen in one out of four years.

We observed slightly higher soil total C content in residual slurry plots in 2006 but no statistically significant effects of a four-year application of residual slurry on soil total C content (Table 3). Furthermore, the difference between chemically fertilized and residual slurry plots (10, 15, and 20 g N m⁻²) was not significant in 2007. Therefore, there was no significant cumulative effect on the soil C accumulation, even after the five-year successive application of residual slurry. It has previously been well documented that the use of organic amendments, including rice straw, is a notable factor in increased methane emissions from paddy fields because these additions can supply a large amount of C substrate for methanogens (Nouchi et al. 1994; Watanabe et al. 1999; Yagi and Minami 1990; Yang and Chang 1997). Similarly, previous studies have reported that the application of residual slurry to paddy rice increases methane flux in pot and lysimeter experiments (Sasada et al. 2011; Singla and Inubushi 2014; Win et al. 2010). In the present study, harvested plant residues were not returned to the experimental field. Therefore, the soil C accumulation may have primarily been dependent on the residual slurry application rate, and the amount of remained root and stubble. Therefore, our result of no significant effect of continuous application of residual slurry on soil C accumulation indicated that the annual residual slurry application rate might directly affect methane emissions but its cumulative effect might not be essential to determine soil C dynamics related to methane emissions.

We observed gradual increases in methane flux during the period from late June to the end of July and a sharp decline after August 2006 and 2007 (Fig. 1). We also observed a sharp increase in soil Eh after August 2006 and 2007 (Fig. 2). The rice plants were at the active tillering stage in around 12–13 July and at the heading stage in the late August; they were not always submerged after the late of August when intermittent irrigation was employed in 2006. Furthermore, the rice plants were not always submerged after the beginning of August 2007 because of insufficient water supply caused by lack of adequate water resources. Thus, the seasonal variation in methane flux corresponded to water management that affects the relationships between methanogen activity and soil redox status as documented previously (Nishimura et al. 2004; Yagi et al. 1996).

The rice grain yield in the residual slurry plot (10 g N m⁻²) was not significantly different from that in the chemically fertilized plot (10 g N m⁻²); the maximum grain yield was reached at an N application rate of approximately 15 g N m⁻², and further N application by residual slurry saturated the grain yield according to the seven-year survey (Nishikawa et al. 2012). Our result of grain yield over a single year also showed no significant difference between chemically fertilized and residual slurry plots (Table 4). A much larger application rate of residual slurry than plant nutritional requirements would increase the soil C accumulation and significantly promote methane emissions. Therefore, the excessive application of residual slurry still should be avoided because it could also increase lodging risk and N reaching, denitrification, and ammonium volatilization of unused N (Nishikawa et al. 2014) and could reduce grain quality. Indeed, we provide partial evidence that the application of residual slurry can balance improvements in grain yield while controlling methane emissions. However, it is possible that the quality of residual slurry ranges widely, because it depends on feedstock quality (Alburquerque et al. 2012). The quality of the residual slurry, including the C:N ratio or the labile organic C content, and the application rate may be essential to determine the effect on rice grain yield and methane emission. Therefore, a comparative study among different types of residual slurry is still required to evaluate the C dynamics and their contribution to methane emissions from paddy fields.

Conflict of interest

The authors declare that they have no conflict of interest.

Acknowledgments

This study was partly supported by a Grant-in-Aid for Scientific Research [No. 17380161 and 26520306] from the Ministry of Education, Culture, Sports, Science and Technology, Japan. We thank the Nantan City Yagi Bioecology Center for supplying the anaerobically digested cattle slurry used in this study.

Additional information

Funding

This study was partly supported by a Grant-in-Aid for Scientific Research [No. 17380161 and 26520306] from the Ministry of Education, Culture, Sports, Science and Technology, Japan.