Substitution of chemical fertilizer by Chinese milk vetch improves the sustainability of yield and accumulation of soil organic carbon in a double-rice cropping system

ZHOU Xing , LU Yan-hong, , LlAO Yu-lin, , ZHU Qi-dong, , CHENG Hui-dan, , NlE Xin, , CAO Weidong, NlE Jun,

1 College of Resources and Environment, Hunan Agricultural University, Changsha 410128, P.R.China

2 Soil and Fertilizer Institute of Hunan Province, Hunan Academy of Agricultural Sciences, Changsha 410125, P.R.China

3 Scientific Observing and Experimental Station of Arable Land Conservation (Hunan), Ministry of Agriculture, Changsha 410125,P.R.China

4 College of Bioscience and Biotechnology, Hunan Agricultural University, Changsha 410128, P.R.China

5 Longping Branch Graduate School of Hunan University, Changsha 410125, P.R.China

6 Institute of Agricultural Resources and Regional Planning, Chinese Academy of Agricultural Sciences, Beijing 100081, P.R.China

Abstract The double-rice cropping system is a very important intensive cropping system for food security in China. There have been few studies of the sustainability of yield and accumulation of soil organic carbon (SOC) in the double-rice cropping system following a partial substitution of chemical fertilizer by Chinese milk vetch (Mv). We conducted a 10-year (2008-2017)field experiment in Nan County, South-Central China, to examine the double-rice productivity and SOC accumulation in a paddy soil in response to different fertilization levels and Mv application (22.5 Mg ha-1). Fertilizer and Mv were applied both individually and in combination (sole chemical fertilizers, Mv plus 100, 80, 60, 40, and 0% of the recommended dose of chemical fertilizers, labeled as F100, MF100, MF80, MF60, MF40, and MF0, respectively). It was found that the grain yields of double-rice crop in treatments receiving Mv were reduced when the dose of chemical fertilizer was reduced, while the change in SOC stock displayed a double peak curve. The MF100 produced the highest double-rice yield and SOC stock, with the value higher by 13.5 and 26.8% than that in the F100. However, the grain yields increased in the MF80 (by 8.4% compared to the F100), while the SOC stock only increased by 8.4%. Analogous to the change of grain yield, the sustainable yield index (SYI) of double rice were improved significantly in the MF100 and MF80 compared to the F100,while there was a slight increase in the MF60 and MF40. After a certain amount of Mv input (22.5 Mg ha-1), the carbon sequestration rate was affected by the nutrient input due to the stimulation of microbial biomass. Compared with the MF0,the MF100 and MF40 resulted in a dramatically higher carbon sequestration rate (with the value higher by 71.6 and 70.1%),whereas the MF80 induced a lower carbon sequestration rate with the value lower by 70.1% compared to the MF0. Based on the above results we suggested that Mv could partially replace chemical fertilizers (e.g., 40-60%) to improve or maintain the productivity and sustainability of the double-rice cropping system in South-Central China.

Keywords: Chinese milk vetch, fertilizer application levels, rice yield, soil organic carbon, double-rice cropping system

1. lntroduction

China is one of the world’s largest rice producers, accounting for 19.1% of the global rice growing area, and contributing 28.5% of the world’s rice yield in 2016 (FAO 2016). Due to the huge and increasing population in China, the country needs to produce more rice to meet the needs of the people.Although rice yield in China has increased in recent decades due to the injudicious use of chemical fertilizer, this has led to a series of soil pollution problems (Savci 2012). This injudicious use of chemical fertilizer has not only caused the wastage of energy and mineral fertilizer, but also resulted in deterioration in water, soil, and atmospheric quality (Liu X J et al. 2013; Anjum et al. 2016). The worsening soil quality due to increased soil structural damage and acidification,and decreased soil fertility and water table level, has already had negative impacts on rice yield (Guo et al. 2010; Liu C A et al. 2013).

The worsening soil quality is considered to mainly be due to the sole use of chemical fertilizer (Li 2013; Liu C A et al. 2013; Anjum et al. 2016). Therefore, it is necessary to optimize nutrient management practices to sustain the productivity of intensive continuous cropping systems. The economic and efficient utilization of natural resources is crucial for sustainable agro-ecosystems (Granstedt 2000).One possible approach to reduce mineral fertilizer inputs,while stabilizing or increasing crop yield, is the use of natural resources as a substitute. Green manure, such as Chinese milk vetch (Astragalus sinicus L., hereafter Mv), is an important natural resource. It is used primarily as a soil amendment and as a nutrient source for subsequent crops,and has a large potential for increasing soil organic matter and crop productivity, as well as improving pest control. It plays a positive role in reducing the use of chemical fertilizers,ameliorating soil physical properties, curbing the risk of soil erosion, enhancing the organic matter content, increasing fertility levels, restraining the global warming potential, and improving crop yield and quality, which is very significant for the sustainable development of agricultural production(Jiao et al. 1986; Robertson et al. 2000; Tejada et al. 2008;Yang et al. 2011). Growing Mv as green manure in paddy fields can fully exploit the available natural resources (e.g.,light, water, and heat) during the slack winter season and increase the carbon flux into the soil, providing food for soil microbes. In addition, it can fix atmospheric nitrogen with its rhizobia and conserve nitrogen with a high aboveground biomass, reducing the need for supplemental nitrogen to meet the early growth requirements of succeeding crops(Crews and Peoples 2004; Crandall et al. 2005; Voisin et al. 2014). Moreover, incorporating Mv as green manure into soil not only transfers nutrients (e.g., N, P, K, Ca, Mg,and Si) that are stored in the crop free seasons, but also increases the decomposition rate of the indigenous soil organic nitrogen, both of which are favorable to the growth of succeeding crops (Hood 2001).

Soil organic carbon (SOC) is an essential component of soil and acts as a carrier of nutrients; thus, maintaining soil quality in agricultural systems (Cruse et al. 2009). It is therefore necessary to maintain satisfactory levels of SOC to ensure sustainable agro-ecosystems. Many agronomic management practices have been considered to enhance the accumulation of SOC. The use of green manure amendments is a widely accepted management practice to sustain or improve crop yield and SOC stock (Sun et al.2013; Li et al. 2014; Xie et al. 2016). It is very possible that the legumes used as green manure can not only provide nutrients, but also improve the structure of soil macro-aggregates and the population and activity of soil microbial communities (Bedini et al. 2013; Mohanty et al.2013; Kumar et al. 2014). Improving the structure of macroaggregates has the potential to change the carbon storage capacity. On the other hand, the enhanced microbial activity in fertilized soils results in the release of organic acids and polysaccharides during the decomposition of green manure,which act as cementing agents to convert micro-pores into macro-pores (Agbede et al. 2008). It is well known that the accumulation of SOC and soil carbon sequestration is affected by nutrient elements (such as N, P, and K), and microbes play important roles in the processes controlling their concentration in soil (Galloway et al. 2008). The presence of nutrient elements can also limit the abundance of soil microbes (Kuzyakov and Xu 2013). This limitation might stimulate the soil organic materials to release nutrients through decomposition. This decomposition can lead to the transformation of SOC, which also occurs in the nutrient cycling process during paddy SOC mineralization (Tian et al. 2013; Pump and Conrad 2014). However, it remains unclear if a chemical fertilizer (nutrient elements) combined with a green manure (Mv) could sustain a higher crop yield and enhance SOC to a greater extent than the application of chemical fertilizer alone.

Currently, green manure substituted chemical fertilizer can decrease the cost of rice production and environmental degradation (Tejada et al. 2008). Previous works have mainly considered the effects of a combination of green manure and mineral fertilizer on rice production (Zeng et al.2009; Zhao et al. 2015). However, the effect of green manure combined with different amounts of chemical fertilizer on the sustainability of rice productivity and accumulation of SOC is not fully understood. In particular, there have been no long-term continuous studies to evaluate the sustainability of rice productivity and the accumulation of SOC in doublerice cropping system with the application of combinations of green manure and chemical fertilizer. In this study, we aimed to determine: 1) the optimum combination of fertilizer and Mv for improving the sustainability of yields and accumulation of SOC in this region, and 2) whether the dose of mineral fertilizers can be reduced by substitution with Mv to maintain a high rice yield and enhance SOC.

2. Materials and methods

2.1. Site description

The study site was located at Sanxianhu Village (29°13′03′′N,112°28′53′′E, and altitude 28.8 m), Nan County, Hunan Province, China. The climate of the site is subtropical monsoon humid. The mean annual temperature is 16.6°C,mean annual precipitation is 1 237.7 mm, and mean annual sunshine is 1 775.7 h. The test soil, which was typical soil of the Dongting Lake floodplain, was a purple clay soil. The chemical properties of initial soil (0-20 cm soil layer) were as follows: soil pH 7.7, SOC 27.05 g kg-1, total N 3.28 g kg-1,total P 1.28 g kg-1, total K 22.2 g kg-1, alkali-hydrolysable N 261 mg kg-1, available P 15.6 mg kg-1, and available K 98 mg kg-1.

2.2. Experimental design

The experiment was conducted for 10 consecutive years (2008-2017). Early rice (cv. Yuanzao 1 for the first three years, cv. Xiangzaoxian 25 for the following seven years) was sown in late March, transplanted in middle or late April, and harvested in the first week of July. Late rice(cv. Huanghuazhan) was sown in mid-June, transplanted in middle and late July and harvested in the first week of November. Rice seedlings (30-day-old for both early and late rice) were transplanted at a spacing of 20 cm×20 cm(for both early and late rice).

The experiment consisted of seven treatments: (1) CK,no fertilizer; (2) F100, 100% recommended dose of chemical fertilizer, with mineral fertilizer (N, P, and K); (3) MF100,22.5 Mg ha-1Mv, 100% N, P and K; (4) MF80, 22.5 Mg ha-1Mv, 80% N and K, with 100% P; (5) MF60, 22.5 Mg ha-1Mv, 60% N and K, with 100% P; (6) MF40, 22.5 Mg ha-1Mv,40% N and K, with 100% P; (7) MF0, 22.5 Mg ha-1Mv, no mineral fertilizer. A randomized complete block design was used with three replicates of each treatment. To prevent the movement of water and nutrients between plots, each experimental plot (4 m×5 m) was separated by a ridge (30 cm wide and 15 cm aboveground).

The recommended chemical fertilizer application rates to rice in South-Central China are: 150 kg ha-1N (urea 46.0% N), 75 kg ha-1P (superphosphate 12.0% P2O5), and 120 kg ha-1K (potassium chloride 60.0% K2O). In fertilized treatments, the same amounts of chemical fertilizer were applied to early and late rice. For N application, 50% urea was used as the basal application, with the remaining 50% top-dressed at the tillering stage. P and K fertilizers as the basal fertilizer were consistently applied at the recommended rates. Specifically, base fertilizers were applied one day before transplanting. During the application of base fertilizer, the fertilizers were incorporated into the soil to a depth of 5 cm using a rake.

Mv (cv. Xiangzi 1) was directly seeded without tillage 15-25 days before the harvesting of late rice in the plots during the winter season (except for the CK and F100 treatments). Seeds were sown at an areal density of 37.5 kg ha-1. Subsequently, fresh vetch was measured every year at the full-bloom stage, applied at an areal density of 22.5 Mg ha-1and ploughed back 10 days prior to early rice transplanting. At every harvest season, grain from each plot was separately dried and the rice yield was determined.All rice straw was removed from the plots after harvesting.

2.3. Soil sampling and analysis

Soil samples were collected in an S shaped pattern from a depth of 0-20 cm in farmland after the late rice harvest in 2017 and were immediately taken to the laboratory. After removing visible plant debris, stones, and fine roots by hand, the samples were divided into two subsamples. One subsample was sieved with a 2-mm mesh and stored at 4°C for the measurement of soil microbial biomass carbon(MBC) within two weeks. The other subsample was air-dried at room temperature for the determination of soil SOC. In 2017, four soil cores from each treatment (0-20 cm) were collected using a stainless soil core sampler (100 cm3volume) to measure soil bulk density.

SOC concentrations were analyzed using potassium dichromate oxidation at 170-180°C followed by titration with 0.1 mol L-1ferrous sulfate (Lu 2000). MBC was determined by chloroform fumigation-extraction (Wu et al. 1990). The fumigated and non-fumigated moist soils were extracted with 0.5 mol L-1potassium sulfate (K2SO4). The extracts were analyzed to determine their carbon concentration using the same method used for SOC. MBC was calculated using the following equation:

2.4. Calculation

SOC stocks (Mg ha-1) in the top 20 cm soil layer were calculated as Liu et al. (2014):

Here D′ is the soil bulk density in 2017, D is the initial soil bulk density in 2008, HSis the substrate layer, HEis the expanded layer, Ctis the SOC content of the various treatments determined in 2017, and H=0.2 m.

The annual organic carbon inputs (Cinputs, Mg ha-1) to the soil through residues (stubble and roots) and Mv were calculated. Root and stubble biomass were measured immediately after harvesting of the double-rice crop using the core-sampling procedure. Mv samples were collected before turnover. Annual organic carbon inputs were calculated as follows:

where Ystubbleand Yroot(Mg ha-1) are the stubble and root biomass of the double-rice crop in each treatment,respectively, Cstubbleand Croot(g kg-1) are the stubble and root carbon content, respectively, WMvis the weight of fresh Mv (22.5 Mg ha-1), and CMv(g kg-1) is the Mv carbon concentration.

The carbon (C) build-up, sequestration, and stabilization rates were calculated as follows:

where Ctrepresents the SOC stock in a sample soil and Ccontrolis the SOC stock in the control treatment.

C sequestered (Mg ha-1)=Ct-Ci

where Ciis the SOC stock at the beginning of the experiment in 2008. Positive and negative values refer to SOC gains and losses, respectively.

where Cinrepresents the carbon input through crop and green manure.

The productivity of the double-rice cropping system was determined using the sustainable yield index (SYI). The SYI was calculated to counteract any annual variations in yield and indicated the performance of the treatments throughout the test period. The SYI was calculated as Singh et al. (1990):

2.5. Data processing

The test data were statistically analyzed using IBM SPSS Statistics (19.0) and processed with Microsoft Excel 2010.The separation of means was analyzed using the analysis of variance (ANOVA) and Duncan’s multiple range test at the 0.05 level.

3. Results

3.1. Rice grain yield in different treatments

The rice grain yields of early and late rice, as well as the double-rice crop (defined as the sum of early and late rice)fluctuated among the years studied (2008-2017) (Fig. 1).The changing trends of yearly grain yields in the double-rice cropping system generally resulted from changeable climate conditions, and variable pest and weed control practices.An analysis of variance for the yields showed significant treatment×year interactions in both early and late rice as well as the double rice crop. Nevertheless, the annual grain yields were significantly (P＜0.05) higher in all fertilizer treatments than in the unfertilized control (CK, Table 1).

The average grain yield of early and late season rice for 2008-2017 observed in the CK plot was 3.54 and 5.58 Mg ha-1, respectively (Table 1), suggesting that the basic yield of late rice was higher in this paddy soil than the early season yield, probably due to more available nutrients released under high temperature in summer. Compared with the mean yield of the F100 treatment, the MF100 treatment was able to maintain a significantly (P＜0.05) higher grain yield of early and late rice (by 15.2 and 9.1%, respectively); the MF80 treatment also resulted in a higher crop production (by 11.8 and 5.7%, respectively), and there were no significant differences between the MF100 and MF80 treatments(P＞0.05); but it was noted that the mean yield of early and late season rice in the MF60 and MF40 treatments were slightly higher or lower; the MF0 dramatically decreased the grain yield of early and late season rice by 37.0 and 28.6%, respectively. However, there were no remarkable differences in grain yield between the MF0 treatment and CK plot (P＞0.05). These indicate that Mv (22.5 Mg ha-1) could be substituted for N and K chemical fertilizer (20-60%),following advancement of cultivation in purple soil.

Fig. 1 Sequential changes in the rice grain yields of the various treatments from 2008 to 2017. CK, no fertilizer; F100, 100%recommended dose of chemical fertilizer, with mineral fertilizer(N, P, and K); MF100, 22.5 Mg ha-1 Chinese milk vetch (Mv),100% N, P and K; MF80, 22.5 Mg ha-1 Mv, 80% N and K, with 100% P; MF60, 22.5 Mg ha-1 Mv, 60% N and K, with 100% P;MF40, 22.5 Mg ha-1 Mv, 40% N and K, with 100% P; MF0, 22.5 Mg ha-1 Mv, no mineral fertilizer.

Table 1 Average grain yields and the analysis of variance (ANOVA) for early, late, and double-rice yields under different treatments in the double-rice cropping system (2008-2017)

3.2. The SYl of different treatments

The SYI value for the double-rice cropping system was greater for late rice than for early rice in all treatments,indicating that late rice grain yield was more sustainable than that of early rice (Table 2). The SYI value for early and late rice, as well as the double-rice crop, was the highest in the MF100 treatment, while the lowest SYI value was in the CK treatment. The SYI values in the other green-manure amended treatments (MF80, MF60, MF40, and MF0) were similar to that of the F100 treatment. These results reveal that Mv substitution for chemical fertilizer at appropriate rates(e.g., 20-60%) has the potential to maintain a sustainable rice yield under the double-rice cropping system in the purple soil of South-Central China.

3.3. SOC stock in the 0-20 cm soil layer

After 10-year of different fertilization treatments in the doublerice cropping system, the SOC content of the 0-20 cm soil layer was much higher than in the CK plot (Table 3).Compared to the unfertilized CK plot, the SOC concentration was 13.6% higher in the F100 treatment. Soil in the MF100, MF60, MF40, and MF0 plots had a higher SOC(by 1.0-9.6%) than in the F100 plots. However, the SOC concentration in the MF80 plot was lower by 7.4%. With the increase in the fertilizer amount, the SOC concentration of treatments that received a certain organic input showed an initial decrease and then an increase.

During the experimental period, SOC stock concentrations in the 0-20 cm soil layer followed a pattern similar to thatof the SOC concentrations among all treatments (Table 3).The SOC stock in the 0-20 cm soil layer at the beginning of the study was 55.1 Mg ha-1. At the end of the 10-year period, the F100 plots had a slightly lower SOC stock than the unfertilized control (CK) and the initial plots. Compared to the F100 plot, soil in the treatments receiving Mv had a significantly higher SOC stock (MF100, MF80, MF60,MF40 and MF0 higher by 26.8, 8.4, 20.6, 26.7, and 17.5%,respectively) in the 0-20 cm soil layer. A similar trend was also observed for carbon sequestration and the rate of SOC build-up in the respective treatments (Table 3). The mean rate of SOC build-up was the highest in the MF100 plot and the lowest in the F100 plots. It was estimated that 13.2-16.8% of the green manure carbon input was converted to SOC, while the remainder (about 80%) was lost through oxidation.

Table 2 Sustainable yield index (SYI) values for the double-rice cropping system influenced by different fertilization treatments

It was assumed that withered leaves did not turn over any carbon into the soil during the whole growing period of crops. This meant that the carbon input to the soil was from the roots and stubble in the CK and F100 treatments,while it was from the roots and stubble plus returned Mv in the treatments receiving green manure. The annual carbon input level for all treatments ranged from 3.43 Mg C ha-1in the CK to 7.15 Mg C ha-1in the MF100 treatment (Table 3).The highest annual carbon inputs were measured in the MF100 treatment, and then decreased as the proportion of mineral fertilizer decreased in the manure-amended treatments. And the lowest was CK plot.

4. Discussion

4.1. Response of rice yield to fertilization

Chemical fertilizers are commonly used to rapidly improve crop productivity because they are absorbed by crops immediately after application (Ahmad et al. 2008; Prasad et al. 2016). It was found that the rice grain yields obtainedwith the chemical fertilizer alone (F100) were 27.3% higher than the yields obtained with the unfertilized control.Higher rice yields were achieved in the treatments with the incorporation of chemical fertilizer and Mv than with the chemical fertilizer alone (Fig. 1 and Table 1). Similar results were also reported by Bi et al. (2009) in a doublerice cropping system and Wang et al. (2014) in a single rice system where the application of chemical fertilizer and manure resulted in a significant increase in the crop yield.In our study, combination of Mv and different proportions of mineral-fertilizer (MF80, MF60, and MF40) produced slightly higher or lower rice yields than the chemical fertilizer alone (F100; Table 1). The results also were similar to that reported by Cai and Qin (2006), who found a treatment in which organic nitrogen substituted 50% of the chemical nitrogen did not affect crop yields in a wheat-maize rotation experiment. Previous studies have indicated that combinations of green manure (Mv) and mineral fertilizer can dramatically increase the number of spikelets, grains per panicle, panicle number per unit area, and 1 000-grain weight (Lee et al. 2010; Huang et al. 2013). Efthimiadou et al. (2010) further found that, compared to the use of chemical fertilizer alone, the incorporation of green manure enhanced the photosynthetic rate and stomatal conductance of rice plants, induced greater carbon accumulation, and better stabilized the dry matter in rice grains. This is probably because the balanced-supply of nutrients provided by the Mv is beneficial for improving plant growth and increasing yields. Obviously, Mv plants not only contain the primary macro-nutrients (N, P, and K), but also contain some medium- and micro-nutrients (e.g., Ca, Mg, and Si), which promotes and ensures a balanced sustainable nutrient supply in paddy soil (Chen and Zhao 2009; Lu H J et al. 2011). In addition to providing a balanced-supply of nutrients, combining manures and chemical fertilizer can improve the ability of the soil to supply nitrogen by promoting the mineralization potential and mineralization rate of soil organic nitrogen, and accordingly promoting nitrogen absorption and accumulation in plants (Mohanty et al. 2013;Bedada et al. 2014; Xie et al. 2017). On the other hand,incorporating Mv with chemical fertilizer can reduce the soil bulk density and the consumption rate of SOC in paddy fields, improve soil porosity, and increase the proportion of macroaggregates (Srinivasarao et al. 2014; Kumar et al. 2014). It also increases the soil enzyme activity and enhances soil microorganism resistance to environmental stresses. Iyer-Pascuzz et al. (2009) reported that the substitution of chemical fertilizer by green manure improved the root growth environment for rice, and increased the root density, biomass, activity, and nutrient absorption. Hence,the results suggest that legume green manure may partially substitute for chemical fertilizer (i.e., 20-60% N and K).

Table 3 Mean annual carbon input, soil organic carbon (SOC), soil bulk density (BD), SOC stock (SOCs), carbon sequestered, carbon build-up rate, and the conversion ratio of Chinese milk vetch (Mv) carbon to SOC after 10 years of various fertilization treatments

The SYI can be used to measure the sustainability of nutrient management systems and soil productivity (Manna et al. 2005). The more sustainable the nutrient management system, the higher the SYI value (Yadav et al. 2000). In our study, the application of chemical fertilizer alone increased the SYI of rice grains compared to the unfertilized control.Similar results were also reported by Xie et al. (2016),where the application of chemical nitrogen improved the SYI value of crop yields compared with a control in a doublerice cropping experiment. The SYI value of early and late,as well as the double-rice, in the MF100 treatment was much higher than that in the control, while the value for the MF80, MF60, and MF40 treatments slightly exceeded that with the chemical fertilizer alone (F100). Overall, this suggests that the combination of Mv and mineral fertilizer is capable of realizing sustainable rice production in a double-rice cropping system. The results suggest that, even if the amount of chemical fertilizer applied is reduced, the turnover of green manure into the soil might improve the quality and quantity of SOC over several years and improve the ability of soil to supply nitrogen (Manna et al. 2005; Xie et al. 2016). Thus, it can enhance the sustainability of rice production. When reducing the amount of fertilizer applied to late rice, the increases in the grain yield and SYI value of late rice could be attributed to the residual effects of the green manure on succeeding crops (Yadav et al. 2000).

4.2. Response of SOC accumulation to Mv addition

Regardless of the farming system, promoting the SOC content and stock above the threshold level is the key to maintaining and restoring soil quality and sustaining agronomic productivity. Thus, various management practices (i.e., crop residue return, manure and compost amendments) have been proposed to promote the SOC content and stock (Manna et al. 2005); Li et al. 2017). In this study, the SOC stock in the top soil layer (0-20 cm)was reduced slightly by the application of mineral fertilizers alone (F100) and the unfertilized control (CK) compared with the initial value. The slight decreases in the SOC concentration and stock in F100 may be due to crop residue being balanced with carbon decomposition and not being sufficient to promote the SOC stock. Some reports have indicated that the long-term sole application of mineral fertilizers could increase SOC stock in the topsoil layer(Gong et al. 2012; Fan et al. 2014). Differences in the results of the studies were due to differences in the initial soil carbon status, the ecosystems studied, the quantity and quality of material returned, and the quantity, type,and duration of the fertilizer application (Khan et al. 2008;Hamer et al. 2009). An additional application of Mv to paddy soil can increase the SOC content and stock significantly(Table 4). The prominent effect of the incorporation of Mv and mineral fertilizer on SOC compared to the application of mineral fertilizer alone may be attributed to the greater organic input resulting in a greater accumulation of SOC(Bayer et al. 2000; Sisti et al. 2004). Over time, the SOC stock increased under the additional carbon inputs, and the build-up of the SOC stock with these carbon inputs in manure-amended plots was in proportion to the total carbon inputs. The annual carbon inputs were significantly positive correlated with the SOC stock in the 0-20 cm soil layer(y=2.329x+46.517; R2=0.284; P＜0.05). Similarly, the carbon sequestered over 10 years was directly related to carbon inputs, and 28% of the variability in sequestered carbon was explained by the magnitude of carbon inputs among the treatments. Higher biomass and carbon inputs (green manure, leaf-fall, stubble, roots, and rhizodeposition) led to relatively higher SOC retention in the manure-amended treatments, which may be due to the increased availability of deficient nutrients such as nitrogen and potassium, with the addition of green manure (Zhou et al. 2015).

The cultivation of a double-rice cropping system over 10 years in a purple clay soil in the Dongting Lake Area of China without any fertilizer (CK) or solely with a mineral fertilizer (F100) input depleted the SOC stock (Table 3), with mean depletions of 0.63 and 3.06 Mg C ha-1, respectively.However, the experimental data showed that the SOC accumulated in manure-amended plots improved the SOC stock to a different degree. This rate of carbon sequestration is related to the initial test conditions, and the type and quality of carbon added (Li et al. 2017). Although there were 10 years of continuous addition of organic material at a rate of 3.43 to 7.15 Mg C ha-1yr-1, the highest rate of carbon sequestration was 10.87 Mg C ha-1, while the initial soil level was 55.1 Mg C ha-1. Thus, a higher carbon input rate is needed to compensate for the losses by decomposition.This can be attributed to each of the soils with different carbon loadings reaching a new steady state of SOC stock over time (Six et al. 2002).

4.3. Response of carbon sequestration to chemical fertilizer and Mv combination

Decomposition of SOC or carbon sequestration in response to fertilizer (N) addition involves complex microbial processes (Carreiro et al. 2000; Lu M et al. 2011). The soil microbial biomass is also controlled by the availability of nutrients (e.g., carbon and nitrogen sources), which is a limiting factor for soil organic matter mineralization in a paddy soil (Liu and Song 2008; Spohn and Kuzyakov2013). High concentrations of inorganic nitrogen in chemical fertilizer applied usually accelerate the degradation of easily decomposable organic material in soil and may slow the decomposition of recalcitrant organics due to stimulation or repression of different sets of microbial extracellular enzymes (Carreiro et al. 2000). In our study, the change in the carbon sequestration rate followed a cubic regression model (y=1.977x3-17.324x2+43.886x-22.719; R2=0.954;Fig. 2), which was enhanced with an increase in the quantity of chemical fertilizer applied. The highest rate of carbon sequestration was observed in the treatment with a 100% application of mineral fertilizer together with Mv (MF100), but the highest MBC was in soils receiving the MF80 treatment. This result was consistent with a previous meta-analysis, which found that a high nutrient addition induced negative priming effects of SOC (Zang et al. 2016; Liu et al. 2018), which was probably due to a reduction in the soil microbial biomass (Fig. 2). High nutrient addition resulted from chemical fertilizer (particularly nitrogen) may result in nitrogen saturation and constrain the activities of beta-glucosidase in soil and nitrogen fixation,causing the decreases in microbial carbon acquisition and microbial biomass (DeForest et al. 2004). Increases of microbial biomass in MF80 plots therefore may result in the enhancement of soil respiration, which resulted in soil organic matter mineralization. Essentially, nutrients(carbon, nitrogen) addition significantly limited soil carbon sequestration as predominantly regulated by soil C:N ratio.A high nitrogen supply reduces the nutrient demand from the microbial biomass, and thus the rate of SOC decomposition reduces. As the reduction of chemical fertilizer intensified(20-100%), the activity of soil microbes decreased (Fig. 2).A similar phenomenon to that experienced in the MF100 treatment was observed in the lower fertilizer application rate treatments (MF60 and MF40). Lower nutrient inputs slowed down the turnover and activity of microbes due to a lower demand for nutrient-containing organic compounds, thereby leading to an increase in carbon sequestration. This was also confirmed by Kuzyakov and Xu (2013). The optimum carbon sequestration requires additional nutrients (Kirkby et al. 2014). With no additional nutrient inputs (MF0), the lower carbon sequestration rate might be due to the reduced carbon input.

Table 4 Regression analysis of the relationship among the annual carbon input and SOC, BD, SOC stock, C sequestered and SYI after 10 years of cropping1)

Fig. 2 The influence of different fertilization levels on soil carbon sequestration and microbial biomass carbon (MBC) with the combined use of Chinese milk vetch (Mv) in the top soil layer(0-20 cm) in a double-rice cropping system. Values are the means of three replicates in each manure-amended treatment,different letters within the same item indicate significant differences between treatments at P＜0.05. Bars indicate SE.

5. Conclusion

In a double-rice cropping system in the Dongting Lake Area of China, 10 years of continuous incorporation of Mv(22.5 Mg ha-1) and different levels of chemical fertilizer had significant effects, of varying magnitudes, on the yield of a double-rice crop and the SOC in the 0-20 cm soil layer.The incorporation of Mv and 100% of the recommended dose of chemical fertilizer produced the highest early and late rice yields and SOC stock, with values higher by 17.9,10.1, and 26.8% than sole chemical fertilizer treated soil,respectively. It also improved the SYI. Treatments where Mv was combined with 60-40% of the recommended dose of chemical fertilizer produced slightly higher or lower rice yields than the sole chemical fertilizer treatment, while the SOC stock rose by 20.5-26.6%. However, when Mv was combined with 80% of the recommended dose of chemical fertilizer, both the grain yields of early and late rice increased by 11.7 and 5.7%, respectively, while the SOC stock only increased by 8.4%. To achieve the goal of sustaining a high double rice yield and increasing the SOC stock in the Dongting Lake Area of China, Mv (22.5 Mg ha-1) combined with part of the recommended dose of chemical fertilizer(i.e., 40-60%) is strongly recommended.

Acknowledgements

This work was supported by the earmarked fund for China Agriculture Research System (CARS-22), the Key Special Projects in National Key Research and Development Plan of China (2017YFD0301504 and 2016YFD0300900), the Scientific and Technological Innovation Project in Hunan Academy of Agricultural Sciences, China (2017JC47) and the International Plant Nutrition Institute, Canada (IPNI China Program: Hunan-18).