Cite this Article
Guo Liang, Hu Liangliang, Zhao Lufeng, Shi Xiaoyu, Ji Zijun, Ding Lilian, Ren Weizheng, Zhang Jian, Tang Jianjun, Chen Xin. 2020, Coupling Rice with Fish for Sustainable Yields and Soil Fertility in China. Rice Science 水稻科学(英文版), 27(3): 175-179.  
Doi:10.1016/j.rsci.2020.04.001
Permissions
Copyright©2020, Editorial Office of Rice Science
China National Rice Research Institute
Coupling Rice with Fish for Sustainable Yields and Soil Fertility in China
Guo Liang, Hu Liangliang, Zhao Lufeng, Shi Xiaoyu, Ji Zijun, Ding Lilian, Ren Weizheng, Zhang Jian, Tang Jianjun, Chen Xin
College of Life Sciences, Zhejiang University, Hangzhou 310058, China
Corresponding author: Chen Xin (chen-tang@zju.edu.cn)

About 30 million hectares of rice are cultivated annually in China (FAOSTAT, 2019). Because most rice fields are well- irrigated and suitable for fish living (‘ fish’ here refers to various aquatic animals like carps, prawns, crabs, soft shell turtles, etc.) (Fernando, 1993), many efforts have been made to determine how well rice farming and aquaculture can be coupled as a rice-fish system (RFS), of which the same land and water produces both rice and aquatic products. With the technological guidance of RFSs and political supports from the Ministry of Agriculture and Rural Affairs of China since 2010 ( Supplemental Table 1), multiple types of large-scale, high-efficiency and technologically intensive RFSs are now being demonstrated in the main rice-growing areas of China (National Fisheries Technology Extension Centre, 2019), and in total about 1.2 million hectares of rice fields are under RFSs.

Supplemental Table 1. Policies and activities of Chinese government concerning RFSs since 2010.

Experimental evidence has shown that RFS can optimize the benefits of scarce land and water resources through complementary and synergistic interactions between fish and rice plants. On one hand, RFS can reduce pesticide use by reducing diseases, insect pests and weeds (Frei and Baker, 2005). RFS can also reduce the use of nitrogen (N) fertilizers because of the complementary use of N by rice and fish and the enhance of soil nutrient availability (Frei and Baker, 2005; Xie et al, 2011; Berg et al, 2012). On the other hand, RFS can help produce fish and solve some problems for freshwater aquaculture. For example, nutrients in effluents resulting from fish production are absorbed by rice plants and are therefore no longer waste and pollutant. Previous results in relatively small plots and over relatively short periods of time suggest that RFS can produce high yields of both rice and fish (Hu et al, 2013; Zhang et al, 2016). RFS effects, however, could be incorrectly estimated by a small- scale or short-term way. Therefore, larger samples and longer time periods are necessary for a reliable analysis of RFS effects.

Modern intensive agriculture can greatly increase crop yield but often reduces soil quality due to incorrect fertilizing methods (e.g., the application of high rates of mineral fertilizers), poor water management and soil tillage (Tilman et al, 2002; Guo et al, 2010). In the development of RFSs, the regimes of fertilization, water management and tillage have often been changed to create better field environments for fish (Yang et al, 2010; Chen and Hu, 2018; Groenveld et al, 2019). Given these changes and the rapid development of RFSs in China, whether and how RFS has affected rice yield and soil quality is still unclear (Bihari et al, 2015; Hu et al, 2016). In this study, we conducted a nationwide field survey in 2018 and field experiments in 2012-2018 to analyze the effects of RFS on rice and fish yields and soil fertility.

The nationwide field survey included 4 types of RFSs [rice- carp, rice-crab, rice-crayfish and rice-turtle (soft-shelled turtle)], and covered 536 pairs of paddy fields in 86 farms across 14 provinces or cities in China (Fig. 1-A). Each paired fields consisted of RFS and rice monoculture (RM). Averaged across all fields, rice yield was 8.07 t/hm2 per year in RFSs and 8.00 t/hm2 per year in RMs. Rice yields in RFSs were equivalent to those in corresponding RMs for rice-carp (F1, 33 = 3.384, P = 0.075; Fig. 1-B), rice-crayfish (F1, 15 = 1.613, P = 0.223; Fig. 1-D), and rice-turtle RFSs (F1, 11 = 3.179, P = 0.102; Fig. 1-E), and were significantly higher than those in corresponding RM for rice-crab RFS (F1, 23 = 28.548, P = 0.001; Fig. 1-C). One reason for the different yield performances between rice-crab RFS and other RFSs can be that refuge areas in most samples of the surveyed rice-crab farms were below 5% ( Supplemental Table 2). Refuge design (the pattern and area of trenches and pits) is particularly important in the maintenance of rice yield in RFSs. For example, if the refuge area is relatively small (usually < 10% of the field), RFSs will not decrease rice yield and may even increase rice yield because refuges create an ‘ edge effect’ that enhances the growth of rice plants, which can compensate or even over-compensate for the loss of rice plants due to refuge buildings (Halwart, 2004; Wu et al, 2010). We used a pair-wise design that each surveyed farm simultaneously contained RM and a type of RFS, and found that the effects of RFSs on rice yields were not affected by the location conditions (such as climate and soil properties).

Fig. 1. Locations of farms and rice yields.
A, Location of the 86 surveyed farms and field experiments. Dots indicate the locations of the surveyed farms, and pointers indicate the locations of the field experiments. B-E, Rice yields from the 86 surveyed farms. F-I, Rice yields and stability index values of rice yield (insert) from the four field experiments. Stability indexes of rice yield was calculated as S = μ / δ , where μ is the mean yield for a time period and δ is its temporal standard deviation over the same time interval. RM, Rice monoculture; RFS, Rice-fish system.
Values are Mean ± SE. * and * * indicate significant differences at the 0.05 and 0.01 levels, respectively. ‘ ns’ indicates non-significant differences at the 0.05 level.

Supplemental Table 2 Background information on 86 surveyed farms (Mean ± SE).

The field experiments were separately conducted in four farms of the four types of RFSs (Fig. 1-A). Each experiment had four completely randomized blocks, and each block included an RFS field and an RM field. The experiment results showed that temporal stability index of rice yields was not significantly different for the rice-crab vs. the corresponding RM (t6 = 0.589, P = 0.577; Fig. 1-G) or for the rice-turtle vs. the corresponding RM (t6 = 1.535, P = 0.176; Fig. 1-I), but were significantly different for the rice-carp vs. the corresponding RM (t6 = 2.417, P = 0.042; Fig. 1-F) and for the rice-crayfish vs. the corresponding RM (t6 = 2.812, P = 0.031, Fig. 1-H). Temporal stability of rice yield can be affected by variable weather conditions or extreme pest occurrences (Xie et al, 2011; Khumairoh et al, 2018). The four experimental sites of farms had well-irrigated system and experienced no significant pest damage during the study. However, data collected in 2013 indicated that rice yields were significantly lower in the RM filed plots than in the RFS field plots with rice-carp (F1, 6 = 17.225, P = 0.006; Fig. 1-F), rice-crayfish (F1, 6 = 26.574, P = 0.002; Fig. 1-H), and rice-turtle (F1, 6 = 14.666, P = 0.009; Fig. 1-I). In 2013, those three farms experienced extremely high temperatures (daily maximum temperatures > 37 º C) during young panicle differentiation (Supplemental Fig. 1), which is sensitive to high temperatures (Xie et al, 2013). Why higher stability of rice yield under these RFSs than their corresponding RMs could be that the water depths in the refuges are greater than those in the rest of the fields. The deeper water may moderate paddy field temperatures (Yang et al, 2010; Chen and Hu, 2018). The results from the field experiments at the four farms showed that rice yields of the rice-carp, rice-crayfish and rice-turtle systems were more stable than those of their corresponding RMs from 2012 to 2018 (Fig. 1-F to -I). Taken together both field survey and experiments, our results suggested that RFSs did not reduce rice yield under the guidance of technical speciation for RFS.

Supplemental Fig. 1. Daily air temperatures at four sites of field experiments in July and August of 2012-2018.
The horizontal lines are drawn at 37 ° C. Solid peaks indicate the days with maximum temperatures > 37° C. Shadings indicate the period of young panicle differentiation of rice.

Field survey showed that the RFSs produced an average fish yield of 1.04 t/hm2 for rice-carp, 0.64 t/hm2 for rice-crab, 1.39 t/hm2 for rice-crayfish, and 2.18 t/hm2 for rice-turtle systems (Fig. 2-A). Data from field experiments during 2012-2018 also indicated an average fish yield of 0.85 t/hm2 for rice-carp, 0.56 t/hm2 for rice-crab, 1.65 t/hm2 for rice-crayfish, and 2.66 t/hm2 for rice-turtle systems (Fig. 2-B). The balancing of fish yield and rice yield is a great concern about RFSs (Hu et al, 2016, 2019). The fish yields in field survey and experiments were obtained without loss or even with an increase in rice yield, which were below a maximum level of fish yield while rice yield is not reduced (Hu et al, 2016).

Fig. 2. Fish yield, N-input and soil fertility in this study.
A, Fish yields from 86 surveyed farms. Box boundaries represent the 25th and 75th percentiles, the horizontal line is the median, and the whiskers mark the minimum and the maximum values. Dots denote the mean values. B, Fish yields from the field experiments. C, Fertilizer-nitrogen input in the 86 surveyed farms. D-F, Soil fertility in RFSs and RMs in the 86 surveyed farms. G-I, Soil fertility in RFSs and RMs in the field experiments.
Values are Mean ± SE. * and * * indicate significant differences at the 0.05 and 0.01 levels, respectively. ‘ ns’ indicates non-significant differences at the 0.05 level. In G to I, the means from left to right are for seven consecutive years (2012 to 2018).

Field survey also showed that fertilizer-N was overall 13%-44% lower in RFSs than in RMs (for rice-carp: F1, 29 = 47.148, P < 0.001; for rice-turtle: F1, 9 = 45.190, P = 0.001; for rice-crayfish: F1, 9 = 24.684, P = 0.001; for rice-crab: F1, 17 = 76.859, P < 0.001; Fig. 2-C). RFSs, however, required some feed-N input, which averaged 118.4 kg/hm2 in rice-carp, 60.6 kg/hm2 in rice-crab, 188.6 kg/hm2 in rice-crayfish, and 163.2 kg/hm2 in rice-turtle systems (Supplemental Fig. 2). The total N inputs (fertilizer-N + feed-N) in RFSs were significantly higher than RMs for rice-carp (F1, 9 = 10.742, P = 0.010) and rice-crayfish (F1, 6 = 9.579, P = 0.021), and were similar for rice-crab (F1, 11 = 1.102, P = 0.316) and rice-turtle (F1, 3 = 4.121, P = 0.135) ( Supplemental Fig. 2).

Supplemental Fig. 2. Feed-N and fertilizer-N inputs in four types of RFSs and fertilizer-N input in corresponding RM in field survey.
RM, Rice monoculture; RSF, Rice-fish system.
Data are from 154-paired samples (rice-carp, n = 10; rice-crab, n = 12; rice-crayfish, n = 7; rice-turtle, n = 4). Values are Mean ± SE. Asterisks and ‘ ns’ indicate significant and non-significant differences between RM and RSF at the 0.05 level, respectively.

To investigate the effects of RFSs on soil fertility, we collected topsoil samples from RFSs and their corresponding RM fields from a random subset of farms in the field survey, and from experiment field plots each year in 2012-2018. In the field survey, the pair-wise sampling design helped eliminate the effects of different location conditions. The results of field survey showed that soil organic matter (SOM) was similar with their corresponding RMs for rice-turtle and rice-crab RFSs, but were significantly higher for rice-carp and rice-crayfish RFSs (Fig. 2-D). Total soil N and P did not significantly differ between any RFSs and their corresponding RMs, except for total soil N between rice-carp RFS and RM (Fig. 2-E and -F). Field experiments indicated that SOM, total soil N, and total soil P tended to increase in RFS plots but not in RM plots during the study (Fig. 2-G to -I).

Why rice yield and soil fertility can be maintained in RFSs with less fertilizer-N input might be explained by the complementary use of feed-N between rice and fish (Oehme et al, 2007; Xie et al, 2011; Hu et al, 2013). Our field survey indicated that RFSs used 60.6-188.6 kg/hm2 of feed-N and the total N input (fertilizer-N + feed-N) in RFSs was similar to or higher than RMs (Supplemental Fig. 2). Because only a small part of feed-N is consumed by fish (Cao et al, 2015), abundant fish feed presumably remains in the field where it is gradually decomposed by soil microbes; the N released from the non-consumed but microbially decomposed feed can be used by rice (Burford et al, 2004). In this study, soil total N in the rice-crayfish system was higher than in the corresponding RM, while the other RFSs maintained similar soil total N as RMs (Fig. 2-E), because feed-N applied for crayfish in the rice-crayfish system was significantly higher compared to the other RFSs (Supplemental Fig. 2). However, crayfish yield was similar to carp yield but lower than turtle yield (Fig. 2). Higher input and lower output of N containing in the fish led to the N gradually accumulating in the soil. Unlike the other RFSs that fish was co-cultured with rice simultaneously within a field that N excretion from fish can be used directly by rice plants (Hu et al, 2020), the surveyed rice-crayfish system was a rotated system that rice was planted after crayfish was harvested in the field (Supplemental Table 2). This rotated culture can also lead to the N accumulation in the soil.

To identify whether RFSs can result in a substantial change in the oxidation state of soil from RM as the raising of fish requires longer flooding periods or higher water levels than the raising of rice, we also tested total reducing substances in soil samples from field experiment in 2018. Our results indicated that soil total reducing substances were significantly higher in rice-crayfish RFSs that require year-round flooding than in the corresponding RMs, but did not significantly differ between RFSs and RMs for rice-carp, rice-crab and rice-turtle RFSs ( Supplemental Fig. 3), which were not flooded in winter. For rice-crayfish RFSs, previous research has reported that a long-term consecutive waterflooding accelerates soil gleyzation and reduces soil P levels (Si et al, 2017; Li et al, 2018). These results suggested that an annual alternating dry-wet cycle can avoid soil gleyzation in RFSs.

Supplemental Fig. 3. Soil total reducing substances in RFS and RM fields/plots of four experiments.
RM, Rice monoculture; RSF, Rice-fish system.
Data were from the soil samples collected in 2018. Values are Mean ± SE. Asterisks and ‘ ns’ indicate significant and non-significant differences between RM and RSF at the 0.05 level, respectively.

Our results demonstrate that RFSs can help sustain rice yield and soil fertility with less fertilizer-N and can increase the aquatic food supply. Rice farming in China is facing the challenge of sustaining rice yield and soil fertility without increasing the use of fertilizer-N, pesticides and other costly inputs, and is also facing the challenge of promoting farmer enthusiasm for rice farming because profits provided by such farming are low (Xiao et al, 2019). Our results suggest that RFSs can enhance the sustainability of rice production, and can increase both rice field productivity and profits obtained by rice farmers (Wang et al, 2018). To date, however, less than 5% of the rice planting area in China has been used for RFSs (Hu et al, 2015). The results of this study would help Chinese farmers decide whether to adopt RFSs in their rice farming. The results will also be useful for other countries (e.g., Egypt, India, Indonesia, Thailand, Vietnam, the Philippines, Bangladesh, Myanmar and Malaysia) who have paddy fields suitable for RFSs (Ahmed and Garnett, 2011; Berg et al, 2012).

Acknowledgements
SUPPLEMENTAL DATA

The following materials are available in the online version of this article at http://www.sciencedirect.com/science/journal/16726308; http://www.ricescience.org.

Supplemental Table 1. Policies and activities of Chinese government concerning RFSs since 2010.

Supplemental Table 2. Background information of surveyed farms.

Supplemental Fig. 1. Daily air temperatures at four sites of field experiments in July and August of 2012-2018.

Supplemental Fig. 2. Feed-N and fertilizer-N inputs in four types of RFSs and fertilizer-N input in corresponding RM in field survey.

Supplemental Fig. 3. Soil total reducing substances in RFS and RM fields/plots of four experiments.

Materials and Methods
Types of RFSs in this study

Rice-carp, rice-crab, rice-crayfish and rice-turtle systems are the most important RFSs in China, which account for about 80% of the total RFS area (1.2 million hm2) (National Fisheries Technology Extension Center, 2019). These RFSs have been developed and adapted to rice production throughout the six rice-planting areas of China and have helped supply aquatic protein to local people.

The rice-carp system has a long history and is now widely practiced throughout the rice-growing areas in China (You, 2006; Li et al, 2011). In this system, the common carp (Cyprinus carpio) has evolved diverse genotypes that are adapted to different rice-field environments. Fish fries are released into the rice field immediately after rice is transplanted and ‘ live’ together with rice plants until harvest. Although common carp can grow well by foraging natural food sources in rice fields, fish feed and a refuge (a trench or pit) are required to ensure high fish yields.

The rice-crab system is rapidly expanding in China (Xu et al, 2014). This system uses the Chinese mitten crab (Eriocheir sinensis H. Milne-Edwards), which has evolved two genotypes: one is adapted to the climate of northern China, and the other is adapted to the climate in southern China. In the rice-crab system, juvenile crabs are released into the rice field one week after rice is transplanted and ‘ live’ together with rice plants until harvest. The Chinese Mitten Crab moults several times during its life and requires a relatively large refuge area (or a low density of crabs) and high quality water within a specific temperature range. In addition, high quality feed is required for high crab yields.

The rice-crayfish system has expanded substantially in southern China over the last 20 years (Hu et al, 2015). The crayfish used in this system is Procambarus clarkii. Crayfish can live in rice fields throughout the year and can feed on straw, weeds and macro-algae. Thus, the rice-crayfish system requires only a small quantity of fish feed. In the rice-crayfish system, rice and crayfish are rotated.

The rice-turtle system uses the Chinese soft-shelled turtle, Pelodiscus sinensis. This new system has expanded rapidly over the past 10 years because of the economic value of the turtle, which provides protein and has medicinal effects (Zhang et al, 2016). The turtles often live in rice fields for two years before they are harvested. From November to May, when there is no rice, the turtles remain in the refuge. High turtle yields require high quality fish feed.

RFS-farms surveyed in this study

The 86 RFS-farms included in 14 provinces or cities (Fig. 1-A) were surveyed in this study. These farms have been practicing rice-fish systems for more than three years under the guidance of a local, provincial or national station of agriculture or under the guidance of aquatic technology extension. The owners of the farms were professional farmers who have been trained how to conduct rice-fish co-culture. These farms included 34 with rice-carp, 24 with rice-crab, 16 with rice-crayfish and 12 with rice-turtle. The size of farms was around 40 hm2 (Supplemental Table 2).

The farm scale, age and education of the farm owners were recorded before field survey.

Field survey

At each surveyed farms, 3-10 paired fields (RFS vs. rice monoculture, RM) were surveyed in 2018. The total surveyed fields from the 86 RFS-farms were 536-pairs (RFS vs. RM) including 212 with rice-carp, 144 with rice-crab, 108 with rice-crayfish and 72 with rice-turtle. The paired-wise design of RFS and RM on the same farm was used to control the potential influence of climate condition and soil property, because the fields on the same farm were considered to have an identical background.

Rice and fish yields for each field were determined by using the data collected by the farmers at harvest. Rice yield was measured as air-dried weight, and fish yield was measured as fresh weight. Rice and fish yields were expressed as t/hm2.

We also recorded the rice varieties used in each field and irrigation status. All applications of N fertilizer and fish fed in each field were recorded during the rice- and fish-growing season. The fertilizer-N and feed-N were calculated as kg of N per hm2 per year. The total N input (fertilizer-N + feed-N) was also calculated as kg of N per hm2 per year.

We collected soil samples (0-20 cm depth) from 261 of the 536 paired fields in 51 of the 86 RFS farms in 2018. At each farm, pared-soil samples were collected from the same number of RM fields and RFS fields. At each field, 5 to 10 subsamples were collected and combined to yield one soil sample per year. The soil samples were transported to the laboratory and air-dried for determining soil organic matter (SOM), total N and total P (Lu, 1999).

For each type of RFSs, the general linear model in SPSS (V.20.0) was used to perform two-way ANOVAs to analyze the effect of culture type (RM or RFS) on rice yield, fish yield, fertilizer-N applied, total-N applied, SOM, total N and total P. Culture type (RM or RFS) was set as a fixed factor, whereas location (farm) was set as a random factor in order to make the statistical significance of culture type more robust across China. Before the analysis, data were log-transformed to meet assumptions of normality and homogeneity.

Field experiments

From 2012 to 2018, we conducted field experiments separately in four farms, each of which practiced one kind of RFS: rice-carp, rice-crab, rice crayfish or rice-turtle (Fig. 1-A). At each experiment, four completely randomized blocks were designed. Blocks were independent from each other, i.e., they had separate water inlets and outlets. Each block contained two types of field plots (RM and RFS). The target fish yield was 2.1 t/hm2for the rice-carp farm, 0.6 t/hm2 for the rice-crab farm, 3.6 t/hm2 for the rice-turtle farm, and 1.7 t/hm2for the crayfish farm. These values approximated the threshold fish yields that do not cause a reduction in rice yield for each RFS (Hu et al, 2016).

For rice-carp experiment. The rice-carp farm for the field experiment was located in Qingtian County (120° 18′ E, 27° 59′ N, Fig. 1-A), Zhejiang Province, China. The rice-carp system has a long history in this area, where rice-carp farming is conducted in 90% of the rice fields. The fish is an indigenous, red, soft-scaled, common species of carp (Oujiang color common carp, Cypinius carpiavar. color). The area around the site is hilly and mountainous, and the principal crop is rice, which is grown from May to October. This area has a subtropical monsoon climate with a mean annual air temperature of 17.5 ° C and a mean annual precipitation of 1432 mm. The soil is a sandy loam with a pH of 5.4. Total SOM and soil N content ranged from 30.72 to 32.92 g/kg and 2.09 to 2.79 g/kg, respectively.

Each field plot (0.01 hm2) of RM or RFS was randomly assigned to a block. Field plots were separated by 50-cm-high concrete ridges. Four weeks after germination, rice seedlings (hybrid variety Zhongzheyou 1) were transplanted (one seedling per hill), with 30 cm between rows and 30 cm between hills within the same row for RF and RM. Field plots were flush irrigated at transplanting and then permanently flooded to 20 cm depth until harvest. Sixty carp fries (C. carpio var.color) (40 g each) were released into each plot of RFS and FM immediately after the rice seedlings were transplanted. The carp fries were obtained from local farmers who maintained the fish in small ponds. A basal fertilizer (N:P:K = 15:15:15, 550 kg/ hm2) was broadcast in all plots before furrowing. No pesticide or top-dress fertilizer was used. Artificial feed in pellets (5.37% N and 1.46% P) was applied once each day between 6:00-7:00 am throughout the experiment. At the beginning of the experiment, the daily amount of feed added was 4% of the fish fresh weight per plot, then increased by 3% every 10 d. By the time of carp harvest, the total quantity of the feed applied was 1.46 t/hm2 for RFS. This mixed feed contained 5.37% N, and thus the rate of feed-N input was 77.35 kg/hm2 for RFS.

For the rice-crab experiment. The rice-crab farm for the field experiment was located in Panshan County (41° 9′ N, 122° 15′ E, Fig. 1-A), Liaoning Province, China, where the rice-crab system has been widely practiced over the past 10 years. The crab species is Eriocheir sinensisMilne-Edwards. This area has a monsoon continentality climate with a mean annual air temperature of 9.2 ° C and a mean annual precipitation of 62 mm. The loam soil has a pH of 8.2. Total SOM and soil N content ranged from 23.08 to 29.53 g/kg and 1.42 to 1.66 g/kg, respectively.

Each field plot (0.1 hm2) of RM or RFS was randomly assigned to a block. Fields were separated by 50-cm-high PVC boards. Each field plot had an independent water inlet and outlet and a circular refuge (width 1 m, depth 60 cm) for crabs. Four weeks after germination, rice seedlings (Yanfeng 47) were transplanted (one seedling per hill), with 30 cm between rows and 30 cm between hills within the same row for RFS and RM. In RFS, 375 g of juvenile crabs (each 1.15 g) was released into each field plot at 1 week after rice transplanting. A basal fertilizer for rice (N:P:K = 15:15:15, 750 kg/ hm2) was broadcast in all RFS and RM plots before furrowing. No pesticide or top-dress fertilizer was used. Feed for crabs (4.52% N and 1.03% P) was applied once each day throughout the experiment. The amount of crab feed added was 3%-5% of crab fresh weight per plot at the beginning. As the crabs grew, the quantity was increased 3% every 10 d. By the time of crab harvest, the total quantity of the feed applied was 1.78 t/hm2. This mixed fish feed contained 4.52% N, and thus the rate of feed-N input was 80.33 kg/hm2for RFS.

For rice-crayfish experiment. The rice-crayfish farm for the field experiment was a large farm managed by a specialized farmer cooperatives and was located in Quanjiao County, Anhui Province, China (32° 10′ N, 117° 27′ E, Fig. 1-A). The crayfish used in the farm is Procambarus clarkii. The area around the site is flat, and the principal crop is rice, which is grown from May to November. This area has a subtropical monsoon climate with a mean annual air temperature of 15 ° C and a mean annual precipitation of 980 mm. The soil at this crayfish farm is a sandy loam. Total SOM and soil N content ranged from 28.93 to 32.97 g/kg and 1.65 to 2.13 g/kg, respectively.

Each field plot (1.5 hm2) of RM or RFS was randomly assigned to a block. Field plots were separated by 50-cm-high soil ridges. Each field plot had a circular refuge (width 1.5 m, depth 60 cm) for crayfish. After crayfish was harvested in mid-June, rice seedlings (Zaofengyou 33) were transplanted (one seedling per hill), with 30 cm between rows and 30 cm between hills within the same row for RFS and RM. A basal fertilizer (N:P:K = 15:15:15, 750 kg/hm2) was broadcast in all plots before furrowing. No pesticide or top-dress fertilizer was used. During the rice growth period, only small numbers of adult crayfish lived in the paddy field. No feed was applied. After rice was harvested, the paddy field was flooded to provide suitable habitat for the crayfish during winter. During March to May of the next spring, artificial feed in pellets (6.51% N and 2.63% P) was applied. By the time of crayfish harvest, a total of 2.25 t/hm2 feed had been applied. This mixed fish feed contained 4.35% N, and thus the rate of feed-N input was 97.87 kg/hm2 for RFS.

For rice-turtle experiment. The rice-turtle farm for the field experiment was a large farm managed by an agriculture company (Qingxi Soft-Shelled Turtle Company) and was located in Deqing County (30° 33′ N, 119° 32′ E, Fig. 1-A), Zhejiang Province, China. The Chinese soft-shelled turtle in this system is the indigenous species Pelodiscus sinensis. The area around the site is flat, and the principal crop is rice, which is grown from May to November. This area has a subtropical monsoon climate with a mean annual air temperature of 14 ° C and a mean annual precipitation of 1379 mm. The soil at the turtle farm is a sandy loam with a pH of 5.4. Total SOM and soil N content ranged from 30.72 to 32.92 g/kg and 2.09 to 2.79 g/kg, respectively.

Each field plot (0.8 hm2) of RM or RFS was randomly assigned to a block. Field plots were separated by 150-cm-high concrete ridges. Four weeks after germination, rice seedlings (Qingxi 8) were transplanted (one seedling per hill), with 30 cm between rows and 30 cm between hills within the same row for RFS and RM. In each RFS field plot, 3600 young turtles (250 g each) were added. A basal fertilizer (N:P:K = 15:15:15, 550 kg/hm2) was broadcast in all plots before furrowing. No top-dress fertilizer or pesticide was applied. Artificial feed in pellets (6.51% N and 2.63% P) was applied twice every day at about 07:00 am and 17:00 pm. The amount of turtle feed added was 0.5%-1.0% of the turtle fresh weight per plot. By the time of turtle harvest, a total of 1.62 t/hm2of feed had been applied. This mixed feed contained 6.51% N, and thus the rate of feed-N input was 105.80 kg/hm2 for RFS.

For the four experiment, yields of rice and fish were determined by harvesting the rice grain and fresh fish from whole fields in each year during 2012 to 2018. Rice grains were air-dried and weighed. Rice yield was expressed as tons of air-dried grain per hectare. Fish yield was expressed as tons of fresh fish per hectare. Immediately after harvest, soil samples (0 to 20 cm depth) from each field plot were collected. Soil samples were air dried for determining SOM, total N and total P (Lu, 1999). In 2018, soil samples were collected for determining the total amount of reducing substances (Lu, 1999).

The temporal stabilities of rice yield (S) in RM and RF were compared for seven years (2012-2018). S for each plot was calculated as S = μ / δ , where μ is the mean yield for a time period, and δ is its temporal standard deviation over the same time interval.

The GLM in SPSS (V.20.0) was used for statistical analysis. Before analysis, data were log-transformed if they did not meet the assumptions of normality and homogeneity. Two-way ANOVAs were used to analyze the effects of culture type (RM and RF) on S, yields of rice and fish, SOM, total N, total P and soil total reducing substances.

References

Hu L L, Tang J J, Zhang J, Ren W Z, Guo L, Halwart M, Li K X, Zhu Z W, Qian Y L, Wu M F, Chen X. 2015. Development of rice-fish system: Today and tomorrow. Chin J Eco-Agric, 23(3): 268-275. (in Chinese with English abstract)

Hu L L, Zhang J, Ren W Z, Guo L, Cheng Y X, Li J Y, Li K X, Zhu Z W, Zhang J E, Luo S M, Cheng L, Tang J J, Chen X, 2016. Can the co-cultivation of rice and fish help sustain rice production? Sci Rep, 6: 28728.

Li K X, Zhu Z W, Qian Y L. 2011. Trend characteristics and development proposals of a new round rice-fish culture. Chin Fisheries Econ, 29: 17-21. (in Chinese)

Lu R K. 1999. Analytical Methods of Agricultural Chemistry in Soil. Beijing, China: China Agricultural Scientech Press. (in Chinese)

National Fisheries Technology Extension Centre. 2019. Report on the development of China’ s rice-fish co-culture industry in 2018. Chin Fish, 1: 20-27. (in Chinese)

Xu M, Ma X Z, Wang W. 2014. Effects of different cultivation patterns on rice yield and crab in rice-crab culture system. Agric Sci Chin, 47: 1828-1835. (in Chinese)

You X. 2006. Rice-fish culture-a typical model of sustainable traditional agriculture. Agric Archaeol, 4: 222-224. (in Chinese).

Zhang J, Hu L L, Ren W Z, Guo L, Tang J J, Shu M A, Chen X. 2016. Rice-soft shell turtle coculture effects on yield and its environment. Agric Ecosyst Environ, 224: 116-122.

Acknowledgements

This research was financially supported by the Ministry of Science and Technology (Grant No. 2016YFD0300905) and the Natural Science Foundation of China (Grant No. 31661143001). We are grateful to Li Kexin and Zhu Zewen from the National Aquaculture Technical Extension Station of China, for their assistance in our field survey. We also thank Wu Minfang from Qingtian County Bureau of Agriculture and Rural Affairs in Zhejiang Province, China for his assistance with field experiments.

(Managing Editor: WANG Caihong)

Reference
[1] Ahmed N, Garnett S T. 2011. Integrated rice-fish farming in Bangladesh: Meeting the challenges of food security. Food Sec, 3(1): 81-92. [Cited Within:1]
[2] Berg H, Berg C, Nguyen T T. 2012. Integrated rice-fish farming: Safeguarding biodiversity and ecosystem services for sustainable food production in the Mekong delta. J Sustain Agric, 36(8): 859-872. [Cited Within:2]
[3] Bihari P, Nayak A K, Gautam P, Lal B, Shahid M, Raja R, Tripathi R, Bhattacharyya P, Pand a B B, Mohanty S, Rao K S. 2015. Long-term effect of rice-based farming systems on soil health. Environ Monit Assess, 187(5): 296. [Cited Within:1]
[4] Burford M A, Preston N P, Minh T H, Hoa T T T, Bunn S E, Fry V M. 2004. Dominant sources of dietary carbon and nitrogen for shrimp reared in extensive rice-shrimp ponds. Aquac Res, 35(2): 194-203. [Cited Within:1]
[5] Cao L, Naylor R, Henriksson P, Leadbitter D, Metian M, Troell M, Zhang W B. 2015. China’s aquaculture and the world’s wild fisheries. Science, 347: 133-135. [Cited Within:1]
[6] Chen X, Hu L L. 2018. Rice-fish co-culture system. In: Luo S M. Agroecological Rice Production in China: Restoring Biological Interactions. Rome, Italy: Food and Agriculture Organization of the United Nations. [Cited Within:2]
[7] FAOSTAT. 2019. 2019. http://www.fao.org/faostat/en/#home. April 2019. [Cited Within:1]
[8] Fernand o C H. 1993. Rice field ecology and fish culture: An overview. Hydrobiologia, 259(2): 91-113. [Cited Within:1]
[9] Frei M, Becker K. 2005. Integrated rice-fish culture: Coupled production saves resources. Nat Resour Forum, 29(2): 135-143. [Cited Within:2]
[10] Groenveld T, Kohn Y Y, Gross A, Lazarovitch N. 2019. Optimization of nitrogen use efficiency by means of fertigation management in an integrated aquaculture-agriculture system. J Clean Prod, 212: 401-408. [Cited Within:1]
[11] Guo J H, Liu X J, Zhang Y, Shen J L, Han W X, Zhang W F, Christie P, Goulding K W T, Vitousek P M, Zhang F S. 2010. Significant acidification in major Chinese cropland s. Science, 32: 1008-1010. [Cited Within:1]
[12] Halwart M, Gupta M V. 2004. Culture of Fish in Rice Fields. Roma, Italy: Food and Agriculture Organization of the United Nations & Penang, Malaysia: The World Fish Center. [Cited Within:1]
[13] Halwart M. 2006. Biodiversity and nutrition in rice-based aquatic ecosystems. J Food Compos Anal, 19(6/7): 747-751. [Cited Within:1]
[14] Hu L L, Ren W Z, Tang J J, Li N N, Zhang J, Chen X. 2013. The productivity of traditional rice-fish co-culture can be increased without increasing nitrogen loss to the environment. Agric Ecosyst Environ, 177: 28-34. [Cited Within:2]
[15] Hu L L, Tang J J, Zhang J, Ren W Z, Guo L, Halwart M, Li K X, Zhu Z W, Qian Y L, Wu M F, Chen X. 2015. Development of rice-fish system: Today and tomorrow. Chin J Eco-Agric, 23(3): 268-275. (in Chinese with English abstract) [Cited Within:2]
[16] Hu L L, Zhang J, Ren W Z, Guo L, Cheng Y X, Li J Y, Li K X, Zhu Z W, Zhang J E, Luo S M, Cheng L, Tang J J, Chen X. 2016. Can the co-cultivation of rice and fish help sustain rice production? Sci Rep, 6: 28728. [Cited Within:4]
[17] Hu L L, Zhao L F, Tang J J, Guo L, Ding L L, Zhang J, Ren W Z, Chen X. 2019. Extension potential of rice-fish co-culture system: A case study of 10 provinces in South China. Chin J Eco-Agric, 27(7): 981-993. (in Chinese with English abstract) [Cited Within:1]
[18] Hu L L, Guo L, Zhang L F, Shi X Y, Ren W Z, Zhang J, Tang J J, Chen X. 2020. Productivity and the complementary use of nitrogen in the coupled rice-crab system. Agric Syst, 178: 102742. [Cited Within:1]
[19] Khumairoh U, Lantinga E A, Schulte R P O, Suprayogo D, Groot J C J. 2018. Complex rice systems to improve rice yield and yield stability in the face of variable weather conditions. Sci Rep, 8: 14746. [Cited Within:1]
[20] Li Q M, Xu L, Xu L J, Qian Y G, Jiao Y, Bi Y H, Zhang T L, Zhang W, Liu Y J. 2018. Influence of consecutive integrated rice- crayfish culture on phosphorus fertility of paddy soils. Land Degrad Dev, 29(10): 3413-3422. [Cited Within:1]
[21] Liu Y B, Qin L, Li F B, Zhou X Y, Xu C C, Ji L, Chen Z D, Feng J F, Fang F P. 2019. Impact of rice-catfish/shrimp co-culture on nutrients fluxes across sediment-water interface in intensive aquaculture ponds. Rice Sci, 26(6): 416‒424. [Cited Within:1]
[22] National Fisheries Technology Extension Centre. 2019. Report on the development of China’s rice-fish co-culture industry in 2018. Chin Fish, 1: 20-27. (in Chinese) [Cited Within:2]
[23] Oehme M, Frei M, Razzak M A, Dewan S, Becker K. 2007. Studies on nitrogen cycling under different nitrogen inputs in integrated rice-fish culture in Bangladesh. Nutr Cycl Agroecosyst, 79(2): 181-191. [Cited Within:1]
[24] Si G H, Peng C L, Xu X Y, Xu D B, Yuan J F, Li J H. 2017. Effect of integrated rice-crayfish farming system on soil physico- chemical properties in waterlogged paddy soils. Chin J Eco- Agric, 25(1): 61-68. (in Chinese with English abstract) [Cited Within:1]
[25] Tilman D, Cassman K G, Matson P A, Naylor R, Polasky S. 2002. Agricultural sustainability and intensive production practices. Nature, 418: 671-677. [Cited Within:1]
[26] Wang C, Hu L L, Tang J J, Guo L, Ren W Z, Ding L L, Huai Y, Wang Y Y, Chen X. 2018. Characteristics and the effects of the rice-fish integration based farms in China. Res Agric Mod, 39(5): 875-882. (in Chinese with English abstract) [Cited Within:1]
[27] Wu X, Xie J, Chen X, Chen J, Yang X X, Hong X K, Chen Z J, Chen Y, Tang J J. 2010. Edge effect of trench-pond pattern on rice grain and economic benefit in rice-fish co-culture. Chin J Eco-Agric, 18(5): 995-999. (in Chinese with English abstract) [Cited Within:1]
[28] Xiao G Y, Xiao Y L, Li J J, Deng L H, Weng L S, Meng Q C, Yu J H. 2019. High efficiency is a dominant target for current rice breeding. Chin J Rice Sci, 33(4): 287-292. (in Chinese with English abstract) [Cited Within:1]
[29] Xie J, Hu L L, Tang J J, Wu X, Li N N, Yuan Y G, Yang H S, Zhang J E, Luo S M, Chen X. 2011. Ecological mechanisms underlying the sustainability of the agricultural heritage rice-fish coculture system. Proc Natl Acad Sci USA, 108(50): E1381-E1387. [Cited Within:4]
[30] Xie Z Q, Du Y, Gao P, Zeng Y. 2013. Impact of high-temperature on single cropping rice over Yangtze-Huaihe river valley and response measures. Meteorol Mon, 39(6): 774-781. (in Chinese with English abstract) [Cited Within:1]
[31] Yang X X, Xie J, Chen X, Chen J, Wu X, Hong X K, Tang J J. 2010. Effects of different irrigation water depth on yield of rice and fish in rice-fish ecosystem. Guizhou Agric Sci, 38(2): 73-74. (in Chinese with English abstract) [Cited Within:2]
[32] Zhang J, Hu L L, Ren W Z, Guo L, Tang J J, Shu M A, Chen X. 2016. Rice-soft shell turtle coculture effects on yield and its environment. Agric Ecosyst Environ, 224: 116-122. [Cited Within:2]