Comparison of paddy soil fertility under conventional rice straw application versus cow dung compost application in mixed crop–livestock systems in a cold temperate region of Japan

ABSTRACT

After the rice harvest in Japan, rice straw (RS) is usually cut by combine harvester and incorporated into the soil to improve its fertility. In mixed crop–livestock systems, however, RS is collected and used as livestock feed, and cow dung compost (CDC) is then applied to the soil. This system utilizes the residual organic matter from both rice production and livestock husbandry to make each product. CDC application is also considered to improve the fertility of paddy soil. However, the nutrient input from CDC and the effect of CDC application on soil fertility vary among regions and/or soil types. We compared soil fertility between RS application (RS treatment, avg. 32 years) and RS removal plus CDC application (CDC treatment, avg. 21 years) in 79 paddy fields in Mamurogawa town, Yamagata Prefecture, a cold temperate region of Japan, and measured the nutrient contents in the applied RS and CDC. The total C content of RS was significantly higher than that of CDC, whereas the N, P, K, and Si contents of CDC were significantly higher than those of RS. However, there was no significant difference in paddy soil fertility – as measured by soil organic C, total N, CEC, available N, P, and Si, exchangeable K, Ca, and Mg, base saturation percentage, pH, and bulk density – between the treatments. The soil fertility of most fields was adequate by RS or CDC treatment. Thus, leaving RS in paddy fields or removing it and then adding CDC to the paddy fields has a similar effect in maintaining adequate soil fertility for single rice production or rice–livestock production systems.

KEY WORDS:

• Cow dung compost
• mixed crop–livestock systems
• paddy soil
• rice straw
• soil fertility

View correction statement:

Correction

1. Introduction

Rice (Oryza sativa L.) is a staple food of the Japanese diet, and rice fields account for more than half of the arable land area in Japan. After harvest, rice straw (RS) is commonly cut by combine harvester and incorporated into the soil. However, RS is also used in mixed crop–livestock systems. The RS is removed and fed to livestock, and then cow dung compost (CDC) is applied to the field. This system utilizes the residual organic matter from both rice production and livestock husbandry to make each product. It also helps to solve two big problems in livestock husbandry in Japan: the shortage in the domestic feed supply and environmental pollution from livestock fecal waste. About 75% of the feed for domestic livestock in Japan is imported from overseas (Kato 2008; Ozawa and Yoshida 2009; Cheng et al. 2018), and it is expensive to treat the waste generated by large livestock farms.

RS is a good source of nutrients for paddy fields, and its incorporation improves soil fertility (Nie et al. 2007; Liao et al. 2013; Cheng et al. 2016; Takakai et al. 2019). However, RS application also increases greenhouse gas (GHG) emissions during the cropping season (Naser et al. 2007; Bhattacharyya et al. 2012; Liu et al. 2015; Zhang et al. 2017), especially in cold areas where RS cannot decompose during winter (Naser et al. 2007; Nakajima et al. 2016). In contrast, CDC is decomposed before being applied to fields, which leads to less GHG emissions from paddy fields than from those receiving RS (Yagi and Minami 1990; Kumagai et al. 2010; Das and Adhya 2014). Thus, CDC appears to be the better organic matter to apply to paddy fields in terms of GHG emissions, especially in cold temperate regions of Japan. CDC application also improves the fertility of paddy soil (Maeda and Hirai 2002; Sumida, Kato, and Nishida 2002; Miura and Kusaba 2013; Shahid et al. 2013).

With regard to solving the issues of feed for livestock and environmental problems such as fecal waste and GHG emissions, CDC application in mixed crop–livestock systems is preferable to conventional RS application in Japan. Few studies, however, have compared soil fertility under CDC application in mixed crop–livestock systems to that under conventional RS application in farmers’ paddy fields. Hasegawa, Fukukawa, and Kimura (2005) made this comparison in organic rice paddy fields; they found that available phosphorus in soil was higher in CDC fields than in RS fields, but the other parameters had no significant differences. However, that study was not conducted in conventional fields, but rather in fields managed organically without chemical fertilizer by a group of farmers who have their own management standards to grow rice. Outside of mixed crop–livestock systems, there are many long-term experiments in research fields that have compared the effect of RS application to that of CDC application on paddy soil fertility. Most of these studies reported better soil fertility under CDC application than RS application (Shiga et al. 1985b; Izuoka et al. 1996; Sakai et al. 1999; Maeda and Hirai 2002), but some noted that whether soil fertility was better under CDC or RS application depended on the soil parameters (Katou et al. 1985; Shibahara et al. 1999). National survey data of Japan showed that CDC supplies more nutrients to fields than RS (Miura and Kusaba 2013). Therefore, CDC applied fields in most mixed crop–livestock systems may have better soil fertility than conventional RS applied fields.

The effect of CDC application on soil fertility depends on its application rate and nutrient content. The application rate of CDC varies among regions (Leon et al. 2012; Miura and Kusaba 2013). CDC nutrient content is significantly affected by the other materials used during composting, such as rice husks and wood chips (Kohyama et al. 2006), with differences in the ratio or type of other materials resulting in different nutrient content. The effect of organic matter application on paddy soil fertility is influenced by the soil type (Uwasawa 1991; Miura and Kusaba 2013). Thus, the nutrient input from CDC and the difference in soil fertility under conventional RS application versus CDC application in mixed crop–livestock systems need to be assessed separately in each region and for each soil type.

Mixed crop–livestock systems can be used where rice production and livestock husbandry occur nearby (Leon et al. 2012), because CDC is not suitable for transport over long distances on account of the odor (Tanigawa et al. 2006) and cost (Tarumoto 2001). In Mamurogawa town, Yamagata Prefecture, a mixed crop–livestock system has been widely practiced for many years. The area of rice paddies accounts for 88% of the total arable land in Mamurogawa; rice accounts for half of the gross agricultural production and vegetable and livestock production (beef and dairy cattle) account for the other half (Ministry of Agriculture Forestry and Fisheries (MAFF) 2019). If the application of CDC in the mixed crop–livestock system in Mamurogawa supplies more nutrients to the paddy fields than RS and results in better soil fertility, then we can recommend that farmers expand the mixed crop–livestock systems and consider optimizing fertilizer management in the system.

The objectives of this research were to understand (1) the nutrient input of CDC produced in the mixed crop–livestock system in Mamurogawa by comparing the nutrient contents and applied amount of CDC and RS and (2) the contribution of the mixed crop–livestock system to paddy soil fertility by comparing the soil fertility of paddy fields under conventional RS application versus CDC application. We conducted this research from two points of view: (1) assessing the general soil fertility in all paddy fields in this area and (2) comparing neighboring field pairs to exclude the effect of various soil environmental conditions.

2. Materials and methods

2.1. Study area

The research was conducted in farmers’ fields in Mamurogawa town, northern Yamagata Prefecture, Japan, in 2016. Mamurogawa (38°51′N, 140°15′E; 374 km²) is surrounded by mountains on all sides. Rice paddies occupy the sediment that was carried by several small rivers from the mountains. Rice is grown once a year, starting in late April or early May, and is harvested in September or October. From November to March, the fields are covered by snow (Japan Meteorological Agency (JMA) 2019). From 1981 to 2010, the average annual precipitation was 2711.0 mm and the annual mean temperature was 10.0 °C (Japan Meteorological Agency (JMA) 2019).

2.2. Field selection and information

The study sites were 79 rice paddy fields managed by farmers. The fields were selected according to the usage of RS and CDC, as assessed by interviewing farmers. In the RS treatment (41 fields), RS was conventionally applied to the fields. In the CDC treatment (38 fields), RS was removed and then CDC was applied to the fields. The average duration of the RS treatment was 32 years and that of CDC treatment was 21 years. The soil types of the study fields were Non-allophanic Andosols, Wet Andosols, Regosolic Andosols, Gley Lowland soils, Gray Lowland soils, and Brown Lowland soils (National Agriculture and Food Research Organization 2019). In the RS treatment, 21 fields had Andosols and 20 fields had Lowland soils. In the CDC treatment, 15 fields had Andosols and 23 fields had Lowland soils. Chemical fertilizer was used in both treatments at the conventional rate for each farmer and each rice cultivar. Several rice cultivars were grown in the study fields; ‘Haenuki’, ‘Tsuyahime’, ‘Hitomebore’, ‘Akitakomachi’, ‘Koshihikari’, ‘SD1ʹ, and ‘Himenomochi’ as edible rice, ‘Dewasansan’ and ‘Miyamanishiki’ as sake rice, and ‘Fukuhibiki’ and ‘Bekogonomi’ as forage. Among the 79 fields, there were 14 neighboring field pairs (RS and CDC treatments nearby).

2.3. Soil, RS, and CDC sampling

After harvest in late October 2016, soil samples were taken in the plow layer with an auger (5 cm diameter) at six points and then bulked. The depth of the plow layer was identified manually and measured with a ruler. Soil samples were dried at 35°C in a forced-air oven, treated to remove stones and plant residue, ground in a ceramic mortar, and passed through a 2-mm sieve. The samples were then used for chemical analysis. Part of each sample was ground finely by a grinder (TI-100, Heiko Seisakusho Ltd., Tokyo, Japan) to measure soil organic carbon (SOC) and total nitrogen (TN). Soil bulk density was measured at three points in each field by taking 100-cm³ cores in the plow layer.

Applied RS in the RS treatment was collected from three points (in the middle and at two sides) in each field after harvest in October 2016. At each point, the rectangle sampled area was decided and marked by pile and string, then all RS on the soil surface excluding rice stubble in the chosen area was collected, and the collection area was measured and recorded. The samples were dried in a forced-air oven at 80°C, ground in a grinder (TI-100), and then used for chemical analysis. The amount of RS applied was calculated by dividing the dry weight of collected RS by the collected area. Because stubble also remained in the CDC treatment, it was not considered to be applied RS in the RS treatment.

CDC was sampled before being applied to the fields in April 2016. Samples were dried in a forced-air oven at 60°C, ground in a grinder (TI-100), and then used for chemical analysis. The amount of CDC applied was obtained by interviewing farmers. In the study area, CDC comprises cow dung, rice husks, wood chips, and feed wastes that are composted for 6 to 12 months. The application rate and nutrient content of each CDC in the study area was considered to be the same every year, as we assumed that farmers did not change their rice cultivation practices or their method for producing CDC.

2.4. Chemical analyses

SOC and soil TN and the TN and total carbon (TC) in RS and CDC were analyzed on a Sumigraph NC-220-F Analyzer (Sumika Chemical Analysis Service Ltd., Tokyo, Japan). To measure total phosphorus (TP) and total potassium (TK) in RS and CDC, the materials were first digested with H₂SO₄–H₂O₂ (Mizuno and Minami 1980). The concentration of P was measured by the vandomolybdophosphoric acid method (Kuo 1996). The concentration of K was measured by flame atomic absorption spectrometry (Spectr-AA 220-FS, Varian Australia Pty Ltd., Mulgrave, Australia). Total silicon (TSi) in RS and CDC was extracted in 1.5 M hydrofluoric acid–0.6 M hydrochloric acid solution, and the Si concentration was measured by using the molybdenum yellow method (Saito et al. 2005).

Soil cation exchange capacity (CEC) was determined following extraction of air-dried soil with 1 M ammonium acetate (pH 7.0; Harada 1984). Available N was determined by anaerobic incubation of air-dried soil at 30°C for 4 weeks followed by extraction with 2 M KCl at a soil:KCl ratio of 1: 10 (w/v). The NH₄⁺-N content in solution extracted to measure CEC and available N was determined by steam distillation (Bremner 1965). Available P was determined by Truog’s method (Nanzyo 1997). Exchangeable K, Ca and Mg in air-dried soil were extracted with 1 M ammonium acetate (pH 7.0; Harada 1984) and, measured by flame atomic absorption spectrometry (Spectr-AA 220-FS). To determine available Si, air-dried soil was incubated in distilled water (1:6 w/w soil:water) at 40°C for 1 week, and filtered through No.5 C filter paper (Toyo Roshi, Co. Ltd., Tokyo, Japan) (Nonaka and Takahashi 1988), and the concentration of Si was measured by the molybdenum blue method (Yoshida 1986). Soil pH (H₂O) was determined in a suspension with an air-dried soil to water (w/w) ratio of 1:2.5 (Kamewada 1997).

For each nutrient, we calculated the amount supplied to the field from RS and CDC as nutrient supplied (kg ha⁻¹) = nutrient content (g kg⁻¹) × dry weight of RS or CDC supplied (t ha⁻¹).

The soil nutrient pool was calculated as amount of nutrient in the soil (kg ha⁻¹) = soil nutrient content (g kg⁻¹) × plow layer depth (cm) × bulk density (g cm⁻³) × 10⁻².

2.5. Statistical analysis

Welch’s t-test was used to compare the nutrient contents of RS and CDC, nutrient input from RS and CDC, and the soil fertility of the two treatments in all fields and in neighboring field pairs. The analysis was performed with the Analysis ToolPak in Excel for Office 365 (Microsoft, Redmond, WA, USA). A P value < 0.05 was considered to indicate a significant difference.

3. Results

3.1. Nutrient content in RS and CDC

Table 1 shows the nutrient content of RS and CDC applied to all fields and to the 14 field pairs. The TC of RS was higher than that of CDC in the all-fields dataset and the14-field-pairs dataset, but the difference was significant only in the former case (P < 0.01). In contrast, the TN, TP, TK, and TSi contents of CDC were significantly higher than those of RS in both the all-fields dataset and the 14-field-pairs dataset. CDC contained TN and TK at 2 times, TP at 5 times, and TSi at 1.3 times higher than those of RS. The C/N ratio of RS was nearly double that of CDC, and the difference was significant in both the all-fields dataset and the 14-field-pairs dataset.

Table 1. Carbon and nutrients content of rice straw (RS) and cow dung compost (CDC) applied to the fields.

Treatment	Total C (g C kg^−1)	Total N (g N kg^−1)	C/N	Total P (g P kg^−1)	Total K (g K kg^−1)	Total Si (g Si kg^−1)
All fields
RS (n = 41)	413 ± 15	5.9 ± 0.8	70.7 ± 9.3	1.0 ± 0.2	6.2 ± 1.8	45.1 ± 10.8
CDC (n = 38)	395 ± 24	13.2 ± 2.9	31.0 ± 5.1	5.3 ± 2.6	11.3 ± 5.4	58.9 ± 12.0
P value	< 0.01	< 0.01	< 0.01	< 0.01	< 0.01	< 0.01
14 neighboring field pairs
RS (n = 14)	411 ± 18	5.8 ± 0.9	72.3 ± 10.3	1.0 ± 0.3	6.5 ± 2.1	45.2 ± 13.7
CDC (n = 14)	396 ± 25	13.3 ± 2.8	30.6 ± 4.6	5.2 ± 2.6	11.2 ± 5.8	58.9 ± 11.6
P value	0.09	< 0.01	< 0.01	< 0.01	0.01	< 0.01

Values are mean ± SD. The P value is the probability level of two-tailed Welch’s t-test.

3.2. Nutrient input from RS and CDC

Table 2 shows the rates of RS and CDC application and the nutrient inputs from RS and CDC in the two datasets. In the all-fields dataset, the average dry weight of RS applied was 4.6 t ha⁻¹, significantly higher than that of CDC at 3.9 t ha⁻¹ (P < 0.05). In the 14-field-pairs dataset the rates of application of RS (4.4 t ha⁻¹) and CDC (4.5 t ha⁻¹) were similar. In the all-fields dataset, the TC input from RS was significantly higher than that from CDC (P < 0.01). In the 14-field-pairs dataset, the trend was opposite, although the difference was not significant (P = 0.94). In both datasets, the nutrient inputs (TN, TP, TK, and TSi) from CDC were significantly higher than those from RS, with the exception of TSi in the all-fields dataset.

Table 2. Application rates of rice straw (RS) and cow dung compost (CDC) and carbon and nutrients inputs from RS and CDC.

Treatment	Application rate^† (t ha^−1)	Total C (kg C ha^−1)	Total N (kg N ha^−1)	Total P (kg P ha^−1)	Total K (kg K ha^−1)	Total Si (kg Si ha^−1)
All fields
RS (n = 41)	4.6 ± 0.9	1882 ± 375	26.8 ± 5.0	4.4 ± 1.1	28.4 ± 9.9	206.5 ± 64.9
CDC (n = 38)	3.9 ± 1.5	1545 ± 635	52.6 ± 27.7	21.6 ± 16.9	46.5 ± 34.4	227.2 ± 98.7
P value	0.03	< 0.01	< 0.01	< 0.01	< 0.01	0.28
14 neighboring field pairs
RS (n = 14)	4.4 ± 0.8	1793 ± 341	24.9 ± 3.8	4.4 ± 1.1	27.8 ± 9.7	198.4 ± 70.3
CDC (n = 14)	4.5 ± 1.8	1811 ± 771	61.0 ± 28.5	23.5 ± 16.1	53.9 ± 39.2	265.5 ± 110.1
P value	0.75	0.94	<0.01	<0.01	0.03	0.07

Values are mean ± SD. The P value is the probability level of two-tailed Welch’s t-test.

^†Dry weight.

3.3. Soil fertility of paddy fields under RS or CDC treatment

Soil fertility is evaluated on the basis of many indicators of soil chemical, physical, and biological properties. We investigated SOC, TN, C/N ratio, CEC, available N, available P, exchangeable K, exchangeable Ca, exchangeable Mg, base saturation percentage, available Si, pH, and bulk density (Table 3). In the all-fields dataset, the CDC treatment had higher values than the RS treatment in SOC, TN, C/N ratio, CEC, available N, exchangeable K, exchangeable Ca, exchangeable Mg, base saturation percentage, pH, and bulk density, whereas the RS treatment had higher values in soil available P and available Si. However, the differences were not significant for any of the parameters. In the 14-field-pairs dataset, the CDC treatment had higher SOC, TN, CEC, available N, exchangeable Ca, exchangeable Mg, and bulk density values than the RS treatment, whereas the RS treatment had higher values in base saturation percentage and available Si; the soil C/N ratio, available P, exchangeable K, and pH were comparable between treatments. None of the parameters showed significant differences between treatments.

Table 3. Soil fertility of paddy fields under conventional rice straw (RS) application and cow dung compost (CDC) application in mixed crop–livestock systems.

Treatment	SOC (g C kg^−1)	TN (g N kg^−1)	C/N	CEC (cmolc kg^−1)	AN (g N kg^−1)	AP (g P kg^−1)	Ex. K (g K kg^−1)	Ex. Ca (g Ca kg^−1)	Ex. Mg (g Mg kg^−1)	BS (%)	ASi (mg Si kg^−1)	pH (H2O)	BD (g cm^−3)
All fields
RS (n = 41)	47.8 ± 23.3	3.63 ± 1.34	12.7 ± 2.0	22.3 ± 6.8	0.24 ± 0.05	0.16 ± 0.06	0.18 ± 0.13	1.73 ± 0.60	0.28 ± 0.10	52.6 ± 12.5	27.2 ± 10.4	5.6 ± 0.3	0.74 ± 0.12
CDC (n = 38)	50.5 ± 17.6	3.87 ± 1.05	12.8 ± 1.4	24.3 ± 7.8	0.25 ± 0.05	0.14 ± 0.06	0.20 ± 0.12	2.13 ± 1.31	0.31 ± 0.14	55.0 ± 15.5	24.5 ± 11.8	5.7 ± 0.3	0.76 ± 0.11
P value	0.57	0.37	0.81	0.23	0.20	0.11	0.45	0.09	0.22	0.45	0.30	0.07	0.60
14 neighboring field pairs
RS (n = 14)	47.7 ± 22.6	3.58 ± 1.23	12.8 ± 2.5	22.0 ± 7.5	0.22 ± 0.04	0.14 ± 0.05	0.21 ± 0.18	1.72 ± 0.64	0.27 ± 0.11	53.4 ± 15.5	24.0 ± 10.7	5.7 ± 0.3	0.74 ± 0.11
CDC (n = 14)	51.1 ± 21.6	3.89 ± 1.32	12.8 ± 1.7	23.0 ± 4.7	0.24 ± 0.04	0.14 ± 0.06	0.21 ± 0.12	1.80 ± 0.57	0.30 ± 0.10	51.4 ± 10.9	23.6 ± 13.0	5.7 ± 0.3	0.78 ± 0.15
P value	0.69	0.53	0.92	0.69	0.41	0.83	0.99	0.73	0.50	0.71	0.94	0.69	0.45

SOC, soil organic carbon; TN, total nitrogen; C/N, carbon/nitrogen ratio; CEC, cation exchange capacity; AN, available nitrogen; AP, available phosphorus; Ex. K, exchangeable potassium; Ex. Ca, exchangeable calcium; Ex. Mg, exchangeable magnesium; BS, base saturation percentage; ASi, available silicon; BD, bulk density. Values are mean ± SD. The P value is the probability level of two-tailed Welch’s t-test.

3.4. Plow layer depth, SOC, and nutrient pools of paddy fields under RS or CDC treatment

Table 4 reports the pool of SOC and nutrients in plow layer soil of paddy fields under RS or CDC treatment. The plow layer depth of the RS treatment was higher than that of the CDC treatment in both datasets with the significant difference in all-fields dataset. In both datasets, the CDC treatment had higher SOC, TN, available N, exchangeable K, and exchangeable Mg than the RS treatment, whereas the RS treatment had higher available P and available Si. In the all-fields dataset, exchangeable Ca was greater in the CDC treatment than in the RS treatment, but in the 14-field-pairs dataset, the opposite was observed. In both datasets, however, no significant differences in the SOC and nutrient pools were found between treatments.

Table 4. Plow layer depth, soil organic carbon (SOC), and nutrient pools of paddy soil under conventional rice straw (RS) application and cow dung compost (CDC) application in mixed crop–livestock systems.

Treatment	Plow layer depth (cm)	SOC (kg C ha^−1)	TN (kg N ha^−1)	AN (kg N ha^−1)	AP (kg P ha^−1)	Ex. K (kg K ha^−1)	Ex. Ca (kg Ca ha^−1)	Ex. Mg (kg Mg ha^−1)	ASi (kg Si ha^−1)
All fields
RS (n = 41)	17.0 ± 2.3	56251 ± 22190	4320 ± 1147	293 ± 53	194 ± 70	208 ± 118	2111 ± 719	344 ± 145	34 ± 14
CDC (n = 38)	15.9 ± 1.6	58284 ± 17034	4498 ± 965	299 ± 54	163 ± 66	231 ± 143	2520 ± 1566	370 ± 168	29 ± 15
P value	0.02	0.65	0.46	0.62	0.05	0.44	0.15	0.48	0.14
14 neighboring field pairs
RS (n = 14)	17.0 ± 2.3	56419 ± 20998	4319 ± 1045	277 ± 47	178 ± 57	233 ± 148	2139 ± 910	337 ± 152	31 ± 17
CDC (n = 14)	15.7 ± 0.9	59244 ± 20270	4559 ± 1129	284 ± 37	165 ± 69	254 ± 167	2131 ± 634	356 ± 125	29 ± 16
P value	0.05	0.72	0.56	0.68	0.61	0.73	0.98	0.72	0.69

SOC, soil organic carbon; TN, total nitrogen; AN, available nitrogen; AP, available phosphorus; Ex. K, exchangeable potassium; Ex. Ca, exchangeable calcium; Ex. Mg, exchangeable magnesium; ASi, available silicon. Values are mean ± SD. The P value is the probability level of two-tailed Welch’s t-test.

4. Discussion

4.1. Nutrient contents of RS and CDC

In the all-fields dataset, the TC content of CDC was lower than that of RS (Table 1). The other materials (rice husks, wood chips, feed waste) used in making CDC have a C content similar to that of RS (i.e., > 410 g kg⁻¹). However, cow dung usually has a C content of < 300 g kg⁻¹ (Wani and Mamta 2013; Kumar, Gupta, and Kumar 2017). Thus, the C content in cow dung lowers the C content of CDC relative to RS. The C lost as gas or leachate during composting is another reason for the lower TC content in CDC. Mishima et al. (2012) estimated that 38% of C is lost during composting in cattle manure. Wood chips are used in making CDC in the study area because the local timber industry produces abundant wood chips. Wood chips contain a high percentage of C and are difficult to decompose because of the high content of lignin (Shiga et al. 1985a). The C content of CDC produced in this area is about 400 g kg⁻¹, which is higher than the average C content of CDC in Japan (Livestock Industry’s Environmental Improvement Organization 2005), and it is likely because of the high percentage of wood chips used in its production.

Nutrient contents (N, P, K, and Si) of CDC were significantly higher than those of RS in both datasets (Table 1). The higher nutrient contents in CDC likely resulted from the high nutrient content of cow dung, because the other materials (rice husks, wood chips, and feed waste) used in composting have nutrient contents similar to those of RS. The nutrient contents of CDC had larger variation than those of RS; the CDC applied in this study was made by several farmers, which results in different sources and percentages of the other materials. The variation in K content of both RS and CDC was high. The K in organic matter exists mostly in soluble forms and is easily washed out by rainfall (Rosolem, Calonego, and Foloni 2005; Jin et al. 2015). Hasegawa, Fukukawa, and Kimura (2005) noted that the concentration of K in CDC stored indoors was more than 6 times that of CDC stored outdoors. In the present study, CDC was stored both indoors and outdoors and RS was collected on the fields before or after exposure to the rain. In addition, the C/N ratio of RS was ≥ 70 and that of CDC was ~30; in the decomposition process in soil, RS immobilizes soil N, and thus net N mineralization is less in RS than in CDC (Shiga et al. 1985a; Nishida et al. 2003).

The application rate of CDC in this area ranged from 10 to 30 t ha⁻¹ in fresh weight (Table S1), which is the same or higher than the recommended application rate for Yamagata Prefecture (2008). In the all-fields dataset, the C input from CDC was significantly lower than that from RS (Table 2), owing to the lower rate of CDC application in dry weight and the lower C content in CDC than in RS. In contrast, the N, P, and K inputs from CDC were greater than those from RS, owing to the higher nutrient contents in CDC (Table 1). This result is comparable with that of Miura and Kusaba (2013), and it agrees with the hypothesis that CDC supplied more nutrients to the field than RS. The Si inputs from CDC and RS were similar, likely reflecting the higher Si content and the lower rate of CDC application. In the 14-field-pairs dataset, however, the CDC application rate was not significantly different from that of RS, so the trend of nutrient input was the same as the nutrient contents. Overall, the C input tended to be higher in RS-applied fields and the N, P, and K inputs were significantly higher in CDC-applied fields.

4.2. Paddy soil fertility parameters under conventional RS application and CDC application in mixed crop–livestock systems

The soil fertility under RS and CDC treatments was not significantly different in the all-fields dataset or the 14-field-pairs dataset (Table 3). This finding indicates that the effects of RS and CDC on soil fertility are not influenced by variation in the fields’ environmental conditions including soil types.

Although the C input from RS was greater than that from CDC (Table 2), SOC was not different between treatments (Table 3). A higher loss of C in the RS treatment explains this result. Nishida et al. (2003) reported that C in RS is decomposed more rapidly than that in CDC after field application. In the present research, the C content in CDC comes from wood chips or rice husk, which have higher lignin content than RS and are thus difficult to decompose (Shiga et al. 1985a). The N input from organic matter was higher in the CDC treatment (Table 2), but soil TN was not significantly different between treatments (Table 3). This could be explained by lower net N mineralization in the RS treatment due to the high N immobilization resulting from the high C/N ratio of RS (Table 1) and greater loss of N in the CDC treatment through plant nutrient uptake, leaching, and emission. In this research, however, we did not investigate plant nutrient uptake (rice yield), leaching, or emission. To elucidate the true causes, it will be necessary to measure those outputs in the RS and CDC treatments. Another reason for the lack of a significant difference in SOC and TN between the RS and CDC treatments may be the relatively small input into the large soil volume: the C input from RS or CDC was about 3% of SOC, and the N input accounted for about 1% of soil TN (data not shown). So, the difference in C or N input between RS and CDC was ≤ 1% of SOC or soil TN.

SOC (or soil organic matter, SOM) and TN constitute heterogeneous mixtures of organic substances and are widely used as the main parameters for evaluating soil fertility (Huang et al. 2009). Table 5 lists the standard values used to evaluate the fertility of paddy soil. In the present study, SOC ranged from 18.0 to 95.3 g C kg⁻¹ (31.0–64.3 g kg⁻¹ SOM) in the RS treatment and from 18.4 to 85.4 g C kg⁻¹ (31.7–147.3 g kg⁻¹ SOM) in the CDC treatment. Soil TN ranged from 1.65 to 6.66 g N kg⁻¹ in the RS treatment and from 1.72 to 5.76 g N kg⁻¹ in the CDC treatment. The SOM content of paddy soil in this study was quite high in comparison with the standard value of fertile soil. We propose three possible causes for this: (1) the continuous addition of organic matter to paddy soil from RS or CDC, (2) the slow rate of decomposition of SOM as a result of the region’s long cold winter and snow cover (Japan Meteorological Agency (JMA) 2019), and (3) the distribution of Andosols in study fields, because Andosols contain higher SOC and TN than those of Lowland soils.

Table 5. Standard values used to evaluate the fertility of paddy soil.

Soil fertility indicator	Fertility level	References
SOM	20 g SOM kg^−1 dry soil (Lowland soils)	Ministry of Agriculture, Forestry and Fisheries (MAFF) (2008)
TN	–	–
CEC	12 cmolc kg^−1 dry soil (Lowland soils), 15 cmolc kg^−1 dry soil (Andosols)	Ministry of Agriculture, Forestry and Fisheries (MAFF) (2008)
AN	0.08 to 0.20 g N kg^−1 dry soil	Ministry of Agriculture, Forestry and Fisheries (MAFF) (2008)
AP	0.04 g P kg^−1 dry soil	Ministry of Agriculture, Forestry and Fisheries (MAFF) (2008)
Ex. K	0.08 g K kg^−1 dry soil	Doberman and Fairhurst (2000)
Ex. Ca	0.40 g Ca kg^−1 dry soil	Doberman and Fairhurst (2000)
Ex. Mg	0.24 g Mg kg^−1 dry soil	Doberman and Fairhurst (2000)
BS	70 to 90 % (Lowland soils), 60 to 90 % (Andosols)	Ministry of Agriculture, Forestry and Fisheries (MAFF) (2008)
ASi	62.48 mg Si kg^−1 dry soil	Calculated from Imaizumi and Yoshida (1958); Sumida (1992); Kato et al. (2002).
pH	6.0 to 6.5	Ministry of Agriculture, Forestry and Fisheries (MAFF) (2008)
BD	–	–

SOM, soil organic matter (calculated as 1.724×soil organic carbon); TN, total nitrogen; CEC, cation exchange capacity; AN, available nitrogen; AP, available phosphorus; Ex. K, exchangeable potassium; Ex. Ca, exchangeable calcium; Ex. Mg, exchangeable magnesium; BS, base saturation percentage; ASi, available silicon; BD, bulk density.

The CEC of the CDC treatment was not significantly different from that of the RS treatment (Table 3). CEC had significant positive correlations with SOC and soil TN (Table 6). SOM carries a negative charge, which can hold cations. Thus, if the SOM content is greater, there will be more available space for cation exchange. The lack of a difference in SOC and soil TN between the RS and CDC treatments would result in the nonsignificant difference of CEC between treatments.

Table 6. Linear correlation coefficients (r) of the relationships between soil fertility indicators.

	SOC	TN	CEC	AN	AP	Ex. K	Ex. Ca	Ex. Mg	BS	ASi	pH	BD
SOC	1
TN	0.96**	1
CEC	0.45**	0.47**	1
AN	0.56**	0.62**	0.51**	1
AP	0.29**	0.37**	0.03	0.22*	1
Ex. K	0.20	0.18	0.10	0.12	0.00	1
Ex. Ca	0.20	0.25*	0.77**	0.41**	0.16	−0.05	1
Ex. Mg	−0.12	−0.05	0.64**	0.30**	−0.03	0.09	0.72**	1
BS	−0.22	−0.20	−0.04	0.03	0.22	−0.01	0.54**	0.45**	1
ASi	−0.02	0.08	0.03	0.18	0.60**	−0.19	0.26*	0.23*	0.31**	1
pH	0.24*	0.19	0.40**	0.08	0.23*	0.09	0.63**	0.34**	0.51**	0.07	1
BD	−0.80**	−0.84**	−0.57**	−0.65**	−0.33**	−0.06	−0.37**	−0.16	0.11	−0.15	−0.14	1

SOC, soil organic carbon; TN, total nitrogen; CEC, cation exchange capacity; AN, available nitrogen; AP, available phosphorus; Ex. K, exchangeable potassium; Ex. Ca, exchangeable calcium; Ex. Mg, exchangeable magnesium; BS, base saturation; ASi, available silicon; BD, bulk density.

**P < 0.01, *P < 0.05.

Soil CEC ranged from 5.8 to 35.7 cmol_c kg⁻¹ in the RS treatment and from 10.7 to 49.3 cmol_c kg⁻¹ in the CDC treatment. Compared with the standard fertility values for CEC, 93% of the fields in the RS treatment and 92% of those in the CDC treatment had higher values, meaning that both treatments resulted in a CEC level representative of fertile soil.

The available N content (Table 3) and the amount of available N in soil (Table 4) of the CDC treatment were not different from those of the RS treatment. These findings, however, are incompatible with the higher TN input from CDC than from RS (Table 2). Available N had significant positive correlations with SOC and soil TN (Table 6). Soil organic N is the initial material necessary for N mineralization, which produces available N. Therefore, the nonsignificant difference of soil SOC and soil TN between the RS and CDC treatments directly contributed to this result, rather than the higher N input from CDC.

Soil available N ranged from 0.14 to 0.36 g N kg⁻¹ in the RS treatment and from 0.17 to 0.39 g N kg⁻¹ in the CDC treatment. The available N of all fields in both treatments were in the range of or higher than the standard fertility value. Thirty-two percent of fields in the RS treatment and 16% of those in the CDC treatment had an available N level representative of fertile soil, and the remaining fields had plenty of available N.

The soil available P content (Table 3) and the amount of available P in soil (Table 4) of the CDC treatment were not different from those of the RS treatment. This finding is not compatible with the higher P input from CDC (Table 2). The availability of P in soil is dependent on the soil type (Andosols or Lowland soils) and soil pH. The soil pH was similar between treatments, however. Andosols adsorb soil P strongly, which results in a lower ratio of available P to total P in Andosols in comparison to Lowland soils. If the CDC treatment have more Andosols than the RS treatment, the higher P input from CDC will be adsorbed by soil then resulted in same level of soil available P with that of the RS treatment. But, in this study the numbers of fields with Andosols and Lowland soils in the RS treatment were similar (21 and 20), and those in the CDC treatment were 15 and 23, respectively. In the 14 neighboring field pairs, the soil type of the RS and CDC treatments in each pair was the same. Thus, the comparison of available P between the RS and CDC treatments was not affected by more distribution of Andosols in the CDC treatment. Therefore, the incompatible result between P input and soil available P can be explained by two hypotheses: (1) the higher P input from CDC increased soil total P but not available P, and (2) there was greater loss of P from the CDC treatment than from the RS treatment. Nagumo et al. (2013) reported that soil available P is difficult to increase by long-term P input from inorganic fertilizer or organic matter, even if the soil TP is increased, which supports the first hypothesis. Phosphorus cannot be lost by decomposition and leaching, like C and N can, because it usually exists in the soil in compounds with aluminum, Ca, or iron. Many studies have reported that the amount of P leaching from paddy fields is very small (Hasegawa 1992; Cho et al. 2002; Shan et al. 2005; Maruyama et al. 2008), and plant nutrient uptake is the main loss of P from the fields (Hasegawa 1992; Maruyama et al. 2008). Thus, further research on the form of P in soil and the P balance of the RS and CDC treatments is needed to clarify the issue.

Soil available P ranged from 0.07 to 0.31 g P kg⁻¹ in the RS treatment and from 0.06 to 0.32 g P kg⁻¹ in the CDC treatment. Values in all fields in both treatments were higher than the standard value indicating soil fertility.

There were no significant differences in soil exchangeable K content (Table 3) or the amount of exchangeable K in soil (Table 4) between the RS and CDC treatments. This was not compatible with the higher TK input from CDC than from RS (Table 2). The greater loss of K from the fields via plant nutrient uptake and leaching in the CDC treatment than the RS treatment may explain this result. To clarify the issue, we need to perform further research on the K balance in paddy fields.

Soil exchangeable K ranged from 0.06 to 0.77 g K kg⁻¹ in the RS treatment and from 0.05 to 0.53 g K kg⁻¹ in the CDC treatment. Compared with the standard fertility value, the 93% of the fields in the RS treatment and 84% of those in the CDC treatment had higher exchangeable K.

Soil exchangeable Ca and exchangeable Mg of the CDC treatment showed no significant difference from those of the RS treatment (Table 3, 4). Both values had significant positive correlations with CEC (Table 6). Thus, the lack of a difference in CEC between the RS and CDC treatments could contribute to the nonsignificant differences in these cations. Similarly, soil base saturation percentages were nonsignificant difference in the RS and CDC treatments (Table 3). Soil base saturation percentage had significant positive correlations with exchangeable Ca and exchangeable Mg (Table 6). The lack of differences in CEC and exchangeable cations between the two treatments led to the nonsignificant difference in soil base saturation.

Soil exchangeable Ca ranged from 0.43 to 3.57 g Ca kg⁻¹ in the RS treatment and from 0.56 to 6.97 g Ca kg⁻¹ in the CDC treatment. All of the fields in the RS and CDC treatments had higher exchangeable Ca than the standard fertility value. Soil exchangeable Mg ranged from 0.08 to 0.70 g Mg kg⁻¹ in the RS treatment and from 0.13 to 0.69 g Mg kg⁻¹ in the CDC treatment. Sixty-one percent of the fields in the RS treatment and 68% of those in the CDC treatment had higher exchangeable Mg than the standard fertility value. Base saturation percentage of soil ranged from 29.4% to 80.8% in the RS treatment and from 27.7% to 94.7% in the CDC treatment. Only 12% of fields in the RS treatment and 18% of those in the CDC treatment had a base saturation percentage representative of fertile soil. Thus, all of the fields reached an adequate level of exchangeable Ca and most of the fields had adequate levels of exchangeable K and exchangeable Mg, but most of the fields did not achieve sufficient base saturation percentage by RS application or CDC application in mixed crop–livestock systems.

Soil available Si content (Table 3) and the amount of soil available Si (Table 4) in the CDC treatment were not different from those in the RS treatment, which likely reflects the nonsignificant difference in Si input from RS and CDC (Table 2).

Soil available Si ranged from 10.8 to 51.4 mg Si kg⁻¹ in the RS treatment and from 11.0 to 49.9 mg Si kg⁻¹ in the CDC treatment. A Si concentration of 51 g Si kg⁻¹ in rice shoot at maturity is recognized as the critical level needed to achieve healthy growth and good yield in Japan (Imaizumi and Yoshida 1958; Sumida 1992). Based on the regression formula reported by Kato et al. (2002) between Si concentration of rice shoot and soil available Si, at least 62.48 mg Si kg⁻¹ of soil available Si is recommended. However, all of the study fields had a lower value. Since soil mineralogical properties determine the amount of soil available Si (Makabe et al. 2009; Yanai, Taniguchi, and Nakao 2016), properties of the soil in the research area caused the low soil available Si. A large uptake of Si by rice plants also could explain this result. Total Si uptake through leaf and stem of rice plants in the RS treatment equaled 165% of the amount of available Si in the soil (data not shown). Even though all the Si in leaf and stem is returned to the fields in the RS treatment, the Si uptake by rice grain makes the balance of Si negative every year.

Both soil pH and bulk density of the CDC treatment were not significantly different from those of the RS treatment (Table 3). Soil pH had significant positive correlations with soil CEC, exchangeable Ca, exchangeable Mg, and base saturation percentage, and bulk density had significant negative correlations with SOC and soil TN (Table 6). The nonsignificant differences of CEC, exchangeable Ca, exchangeable Mg, and base saturation percentage between the RS and CDC treatments would contribute to the nonsignificant difference of soil pH, and the lack of differences in SOC and soil TN would contribute to the nonsignificant difference of soil bulk density between treatments.

Soil pH affects the availability of nutrients in soil through desorption and absorption processes. Soil pH ranged from 5.1 to 6.1 in the RS treatment and from 5.2 to 6.4 in the CDC treatment. Compared with the standard pH value representative of fertile soil, only 15% of the fields in the RS treatment and 29% of those in the CDC treatment were within the ideal range. Sources of acidity that lower pH include rainfall, fertilizer application, plant nutrient uptake, weathering of minerals, and decomposition of organic matter. Thus, both RS application and CDC application in mixed crop–livestock systems cause soil fertility problems due to improper soil pH.

Bulk density of soil ranged from 0.53 to 0.98 g cm⁻³ in the RS treatment and from 0.60 to 1.1 g cm⁻³ in the CDC treatment. Bulk density of Andosol is about 0.5–0.8 g cm⁻³ and that of sandy soil is about 1.1–1.8 g cm⁻³ (Inubushi 2001). The soils in the study fields are Andosols and gravelly to fine-textured Lowland soils, and the bulk density is within the expected range.

This study revealed no significant differences in SOC, TN, CEC, available N, available P, exchangeable K, exchangeable Ca, exchangeable Mg, base saturation percentage, available Si, pH, or soil bulk density between the RS and CDC treatments. This result rejected our hypothesis that higher nutrient input from CDC leads to better soil fertility in the CDC treatment than the RS treatment. To clarify the incompatible result between nutrient inputs from RS and CDC and the soil fertility, it is necessary to conduct further research on the nutrient balance. Compared to the standard values, the application of either RS or CDC in mixed crop–livestock systems in this study area maintained most of the soil fertility indicators at a sufficient level for healthy rice growth.

5. Conclusion

According to the data gathered in all fields, RS had a higher C content but lower nutrient contents (N, P, K, and Si) than CDC. Consequently, C input was higher in the RS treatment and N, P, and K inputs were higher in the CDC treatment. However, the effect of CDC application on soil fertility was not significantly different than that of conventional RS application. There are several possible reasons for this result: (1) Trends were obscured by the large variation of soil environmental condition across the many study fields. (2) Input sources other than RS or CDC to paddy fields have a greater effect on soil fertility. (3) Difference exist in the amount of nutrient output from the plow layer between RS- and CDC-applied fields. The first possible cause can be rejected from the comparisons of neighboring field pairs with the same environmental conditions; results were similar to those when comparing all fields. The second and third reasons, however, should be investigated further. The soil fertility of most of the study fields was adequate, whether they received RS or CDC application. Based on our findings regarding soil fertility after RS or CDC application, it is possible to recommend that farmers in this area expand mixed crop–livestock systems.

Acknowledgments

We are grateful to Dr. Weiguo Cheng (Faculty of Agriculture, Yamagata University) for his valuable comments on this paper. We thank Dr. Kazuhiko Kimura (School of Food, Agricultural and Environmental Sciences, Miyagi University) for his guidance on the digestion method of rice plants and cow dung compost.

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplemental material

Supplemental data for this article can be accessed here.