Variations in nutrient and trace element composition of rice in an organic rice-frog coculture system

Introducing frogs into paddy fields can control pests and diseases, and organic farming can improve soil fertility and rice growth. The aim of this 2-year field study was compare the yield and elemental composition of rice between an organic farming system including frogs (ORF) and a conventional rice culture system (CR). The grain yields were almost the same in the ORF system and the CR system. The ORF significantly increased the contents of phosphorus (P), ion (Fe), zinc (Zn), molybdenum (Mo) and selenium (Se) in rice grain at one or both years. However, the ORF system decreased the calcium (Ca) content in grice grains, and increased the concentration of cadmium, which is potentially toxic. A principal components analysis showed the main impacts of ORF agro-ecosystem on the rice grain ionome was to increase the concentration of P and trace metal(loid)s. The results showed that the ORF system is an ecologically, friendly strategy to avoid excessive use of chemical fertilizers, herbicides and pesticides without decreasing yields, and to improve the nutritional status of rice by increasing the micronutrient contents. The potential risks of increasing Cd contents in rice grain should be addressed if this cultivation pattern is used in the long term.

Rice is one of the most important crops to feed the population in China, and its production has dramatically increased over the last 50 years 1 . Improving the yield and quality of rice with minimal chemical fertilizer and pesticide inputs is important for the long-term sustainability of rice production systems over time 1,2 . However, low fertilizer-use efficiency [3][4][5] , the greenhouse effect 6,7 , water pollution 8,9 , and soil acidification 10 are problems that caused by intensive paddy field in China. The introduction of ecologically, friendly animals into paddy fields creates an integrated agro-ecological system. Many studies have described rice-fish and rice-duck coculture agro-ecosystems and show how these practices reduced the rate of chemical fertilizer and pesticides [11][12][13][14] . However, few studies have focous on the rice-frog coculture system, although this system can effectively pests population of and reduce pollutions in soil and rice grain [15][16][17][18][19] . The rice-frog coculture agro-ecological system, combined with organic farming techniques such as the application of organic fertilizers, represents an ecosystem with effective pest control, less use of chemical fertilizers, and conservation of local soil and water resources 15,20 .
To date, few studies have focused on the cycling of mineral elements, including nutrients and trace elements, in organic farming systems. Application of organic matter can increase the soil fertility, improve the nutrient retention capacity of soil and decrease nutrient losses [21][22][23] . However, the organic matter application may introduce unwanted metals and metalloids, such as Cd and arsenic (As), because these elements are generally present at higher concentrations in compost than in soil. Such organic compounds can altered soil properties (e.g. pH, redox potential, and organic matter content) to improve the accumulation in plant as well. Previous studies have reported contrasting results with respect to the effects of organic amendments on the uptake of trace elements. Some have reported that addition of organic matter increase the uptake of potentially toxic metals (strontium, Sr; barium, Ba; chromium, Cr; and Cd) by crops [24][25][26][27] , whereas others found no significant differences or even a decline in trace metals uptake upon application of organic amendments compared with mineral fertilizer or unfertilized controls 28,29 . Therefore, not only the nutrients but also the nonessential elements need to be evaluated in the soil and crops in organic farming systems. Introducing frogs into the paddy fields has been shown to decrease pest populations and increase the microbial biomass in the soil 15 . However, little is known about the transplanted into the feild at four weeks after germination, (density of 25 plants.m −2 ), on June 19 th , 2013. Plant were harvested on September 27 th , 2013. Rice seedlings of the same cultivar were transplanted into the field at the same density on June 5 th , 2014, and were harvested on September 13th, 2014.
Two culture systems were established: CR, consisting of a normal rice-wheat rotation system with chemical fertilizer, pesticide, and herbicide application; and ORF, consisting of a rice-frog coculture with organic fertilizer application. No herbicide was applied in the ORF field; instead, the weeds were removed by hand. Chinese milk vetch (Astragalus sinicus L.), was seeded in the fields after rice harvest and ploughed into soil the next May as the basal fertilizers. The fields maintained a flooding layer 3-5 cm deep during the rice growing season, and were allowed to dry at the yellow ripening stage. The same rotation (rice-frog coculture and organic amendment) has been applied in the system since 2005. For each system, four plots were established in each year of the study. The area of each field plot was 1600 m 2 , and three sub-plots with area of 100 m 2 were used for sampling. Table 2 showed the details of two systems, and Table 3 summarizes the types of fertilizers, herbicides and pesticides used.
Three or four feeding platforms were placed on each side of the field plot as the resting and feeding places for frogs. Tiger frogs (Rana tigrina rugulosa), which are highly adaptable to the environment, was introduced by Zizaiyuan agricultural Development Co., Shanghai. At 10 to 15 days after rice transplanting, frogs large enough (15-20 g) to prey on pests were introduced into the rice fields on sunny mornings (25000 frogs/ha). The animal used in this study were approved by the School of Agriculture and Biology of Shanghai Jiao Tong University, and all animals used in this study were performed in accordance with relevant guidelines and regulations of the Guide for the Care and Use of Laboratory Animals of the Ministry of Health,China.  Sampling and sample analysis. Uniform plants from each sub plot were sampled for ionome analysis and determination of straw and grain yield. Rice seeds and straws were dried to a constant weight at 70 • C after harvest and ground for digestion and analysis according to the method of Chen 30 . Briefly, each 50-mg plant samples was digested in 2 mL of 61% HNO 3 (EL grade; Kanto Chemical, Tokyo) in a tube for 3 h at 110 °C using a DigiPREP system (SCP Science, Quebec, Canada). Then 0.5 mL of hydrogen peroxide (semiconductor grade; Santoku Chemicals, Tokyo) was added twice at 30 min intervals as an oxidant. The mixture was heated to 110 °C until the solution became clear. After cooling the sample to room temperature, the tubes were filled to 15 mL with 2% HNO 3 and the concentrations of mineral elements in the digests were analyzed using an inductively coupled plasma mass spectrometer (ICP-MS) (Elan, DRC-e; Perkin-Elmer, Waltham, MA, USA). To estimate crude protein (N × 6.25), total nitrogen (N) was determined by digesting the samples with (H 2 SO 4 [98%]-H 2 O 2 ) and N concentrations were determined using the micro-Kjeldahl method 31 . At the beginning of the experiment, soil samples were taken from field (0-20 cm) for the basic soil characterization according to the method of Zhao 8 . The results are shown in Table 1. After harvest, soil samples (0-20 cm) were taken from the field to determine the available mineral contents in the soil according to the method of Sha 26 . The available mineral contents were determined by extracting 2 g of air-dried soil with 40 mL 1 M ammonium acetate. The mixture was shaken for 1 h and then filtered through filter paper (No. 5 C). Subsequently, 2 mL of 61% HNO 3 was added to 5 mL of the filtered extract and concentrated using a DigiPREP system until the solution had evaporated almost completely, using the same procedure as that used to analyze the plant samples. Finally, the tube was filled to 10 mL with 2% HNO 3 before ICP-MS analysis. In figures and tables, the P, potassium (K), calcium (Ca), and magnesium (Mg) concentrations are expressed in g kg −1 , and manganese (Mn), Fe, copper (Cu), Zn, boron (B), Mo, nickel (Ni), sodium (Na), Se, aluminum (Al), Cd, Cr, Sr, Ba, and As concentrations are expressed in mg kg −1 .
Statistical analyses. Data were tested using analysis of variance (ANOVA) and Tukey's tests using Stastical Package for the Social Science (SPSS) 22.0 software (SPSS Inc. Chicago, IL, USA). Differences were considered significant at p < 0.05. A principle component analysis (PCA) was used to profile the protein and mineral element concentrations in the rice grain. The rotation method was varimax with Kaiser normalization.

Result
Straw and grain yield and protein content. The yield and protein content of rice straws and grains did not differ significantly between the CR and ORF systems in either year (Fig. 1). The rice straw yield was higher at 2014 than in 2013. The grain yield was 0.18-fold and 0.39-fold higher in 2013 than in 2014 in the CR and ORF system, respectively. Table 4 shows the macronutrient content in the soil (available concentrations), straw, and grain in the CR and ORF systems. The ORF system did not significantly affect the K and Mg concentrations in soil or plants The P concentration in soil and grain was significantly higher in the ORF system than in the CR system in both years. In contrast, the Ca concentration in soil and grain was significantly lower in the ORF system than in the CR system in both years. The P concentration in grain was 0.48-and 0.58-fold higher in the ORF system than in the CR system in 2013 and 2014, respectively, but the Ca concentration in grain was 0.58-and 0.63-fold lower in the ORF system than in the CR system in 2013 and 2014, respectively. Table 5 shows the concentrations of seven micronutrients in the soil (available concentrations), straw, and grain in the CR and ORF systems. The Cu, B, and Ni concentrations in soil and plant did not differ significantly between the CR and ORF systems. The Mn concentration in soil was 1.30-and 0.87-fold lower in the ORF system than in the CR system in 2013 and 2014, respectively, but the Mn concentration in plants did not differ significantly between the two systems. The Fe concentration in soil and rice straw was significantly higher in the ORF than in the CR system in both years, and the Fe concentration in rice grain was 0.48-fold higher in the ORF system than in the CR system in 2014. The Zn and Mo concentrations in soil and rice grain were significantly higher in the ORF system than in the CR system in both years. In rice straw, the Zn concentration in 2013 and the Mo concentration in 2014 in the ORF system were higher than their correspongding levels in the CR system.

Concentrations of nonessential elements in soil and plants.
The concentrations of eight nonessential elements in the soil (available concentrations), straw, and grain in the CR and ORF systems are shown in Table 6. Among them, only Sr and As concentrations in soil and plants did not differ significantly between the CR and ORF systems. The Na concentration in soil and rice straw was significantly higher in the ORF system than in the CR system in 2014, but not in 2013. The Se concentration in the soil and rice grain was higher in the ORF system than in the CR system, but that in rice straw was lower in the ORF system than in the CR system. The Se concentration in the rice grain was 3.1-and 1.89-fold higher in the ORF system than in the CR system in 2013 and 2014, respectively. The Al concentration in the soil and grain was significantly lower in the ORF system than in the CR system in both years, and in rice straw in 2013. The Cd concentrations in soil, rice straw, and rice grains were significantly higher in the ORF system than in the CR system. The Cd concentration in rice grains was 4.88-and 4.50-fold higher in the ORF system than in the CR system in 2013 and 2014, respectively. The ORF system significantly increased the Ba concentration in the rice straw and grain in 2013. The Cr concentration in rice grain was 28.41-and 1.07-fold lower in the ORF system than in the CR system in 2013 and 2014, respectively. Uptake and translocation efficiency to rice grains. Each element has different availability in soil, and its concentration in the shoot effects its accumulation in grains. Therefore, the effects of the two systems were    letters (a,b). CR: conventional rice planting system; ORF: organic ricefrog coculture system. evaluated for each element as the translocation ratios from soil to grains, and from shoot to grains. Differences in the translocation ratios between ORF and CR are shown as a heat-map in Fig. 2. In 2013, translocation from the soil to grain was significantly increased for Mn, Cd, and Ba, and significantly decreased for Fe and Cr translocation in the ORF system. The translocation from straw to grain was significantly increased for P, Se, Mo, and Cd and decreased for Ca, Fe, Al, and Cr in the ORF system. In 2014, the translocation from soil to grain was increased for Mn and Cd but decreased for Na and Cr in the ORF system. The translocation from straw to grain was increased for Zn and Se but decreased for Ca, Na, Al, and Cr in the ORF system.

Principal components analysis (PCA) of grain elemental composition. The contents of protein and
19 mineral elements in the rice grain were affected by different cultivation patterns; therefore, these data were subjected to a PCA (Fig. 3). The 20 original variables were converted to a set of two linearly uncorrelated PCs that contributed 66.80% and 63.42% of the total variance in 2013 and 2014, respectively. The difference between CR and ORF was clearly discriminated by PC2 in both years, and PC2 accounted for 16.37% and 19.79% of variation in 2013 and 2014, respectively ( Fig. 3A and C). The CR and OFR systems were clearly separated on PC1. The ORF system was located on the positive axis of PC1 with high positive scores; the CR system was located on the negative axis of PC1 with high negative scores. The sample scores of systems reflected the loading score of protein or elements in PCA. In 2013, the high positive sample scores of the ORF system were because of the high positive loading scores: Ba, As, Mo, Ni, P and Cd; and the high negative sample scores of the CR system were due to the high negitive loading scores of Sr, As, B, Al and Cr. Similarly, in 2014, the high positive sample scores of ORF were related to P, Mo, Zn, Fe, Se, and Cd, and the high negative sample scores of CR were related to Cr, Al and Ca.

Discussion
Frog burrowing in soil can increase the activity of soil enzymes and improve soil aeration, thus increasing the number and growth of soil microbes [32][33][34] . In our previous study, we reported that the ORF system could effectively control N and P losses from paddy fields, compared with CR system 35 . In another field study conducted at the same time site as our study, introducing frogs into paddy field improved soil nutrient status, and increased the contents of soluble protein, chlorophyll, and soluble sugars in rice leaves, thus increasing rice grain yield 15 . In the present study, we further compared rice yields and elemental composition of rice grain between the ORF and CR systems. One important result is that the ORF and CR systems had similar grain yields in both years (Fig. 1), confirming that the ORF system, which avoids the excessive use of agro-chemicals, did not reduce yield production. As reviewed by Li et al., China has become the world's leading producer and consumer of pesticides, with production and consumption reaching 265 tons and 179 tons, respectively, in 2011 3 . The large quantities of pesticides applied to agricultural fields and the accumulation of organochlorine and organophosphate pesticides  Table 5. The effects of CR and ORF cultivation patterns on the micronutrient availabilities in the soil (extracted with 1 M ammonium acetate), concentrations in the shoot and seed at 2013 and 2014. In the table, calculated by ANOVA with Tukey test (p < 0.05), differences between treatments were significantly representing by boldfilled column on the mean ranks with letters (a-b). CR: conventional rice planting system; ORF: organic ricefrog coculture system.
residues have resulted in serious environmental deterioration. Therefore, substituting frogs for pesticides is an important strategy to prevent the accumulation of pesticide residues in soil. Moreover, in the ORF system, the use of organic instead of chemical fertilizer is better for the environment. In China, the overuse of chemical N fertilizers has caused environmental problems due to atmospheric, soil, and water enrichment with reactive N of agricultural origin 4,5,[8][9][10] . The ORF system in this study represhents a feasible strategy to reduce the application of agro-chemicals.  Table 6. The effects of CR and ORF cultivation patterns on the nonessential elements availabilities in the soil (extracted with 1 M ammonium acetate), concentrations in the shoot and seed at 2013 and 2014. In the table, calculated by ANOVA with Tukey test (p < 0.05), differences between treatments were significantly representing by bold-filled column on the mean ranks with letters (a-b). CR: conventional rice planting system; ORF: organic rice-frog coculture system. Compared with the CR system, the ORF system significantly increased the P content and reduced the Ca content in rice grain in both years (Tables 4 and 5). The higher P content in the grain was related to the higher available P concentration in soil in the ORF system, and the increased translocation rate of P from straw to grain (Fig. 2). The lower P availability in soil in the CR system might be related to the lower soil pH, because available P is rapidly transformed into immobile forms under low pH 36 . Several studies have shown that the application of organic material can increase P solubility and improve the P status of plants, as compared with chemical fertilizer [37][38][39][40] . In the present study, the increase in available P in soil in the ORF system may be because the organic matter added to soil led to a high concentration of organic acids which reduced P sorption in the soil and increased P availability 40,41 . Teng (2016) reported that the ORF system increased the total P content and phosphatase activity in paddy fields 15 . The Ca availability in the paddy soils was reduced in the ORF system, leading to significantly decreased Ca concentration in the rice grain. The decreased Ca availability resulted in lower lower Ca uptake, and a lower translocation rate from straw to grain (Fig. 2), which reduced the Ca concentration in the grain. The lower soil Ca availability might be due to the high concentrations of other cations in the soil. Because Ca is readily replaced by other cations from its binding sites at the exterior surface of the plasma membrane, the Ca requirement increases with increasing concentrations of other cations 42 . In the current study, the ORF system increased the availability of some heavy metals (Tables 5 and 6). The higher soil pH in the ORF system might be another reason for the decreased Ca availability in the soil.
The ORF system significantly affected the availability of some micronutrients. The ORF system decreased Mn availability in soil, but did not affect the Mn accumulation in rice plants. The difference in Mn availability between the CR and ORF systems might be related to soil pH, because Mn is abundant in soils and its availability is mainly controlled by soil pH and redox conditions 42 . Hamnér and Kirchmann (2015) reported that rganic fertilizer application did not significantly affect the Mn concentrations in soil unless the pH varied 28 . The ORF system significantly increased Fe, Zn and Mo accumulation in the grain in one or both years, and increased their concentrations in soil in both years. The lower Fe and Zn concentrations in the grain in the CR system might be because of lower Fe and Zn availability in the soil, resulting from herbicides application. Studies on the effects of herbicides on crop growth have shown that some chemicals, e.g. chlorsulfuron, and diclofop-methyl, can reduce the availability and uptake of Zn and Fe 43,44 . In another study, herbicides had a stronger inhibitory effect on the uptake of nutrients that enter plant roots mainly by diffusion than on the uptake of nutrients that move mainly by mass flow 45 . Thus, using frogs instead of pesticides and herbicides to control weeds and pests has some advantages in terms of the nutritional status of the rice crop.
Adding Zn-rich organic fertilziers may increase the concentration of Zn in crops, because Zn in soil is readily available to plants. The ORF system resulted in higher Mo contents in grain in both years. The ORF system increased Mo availability in soil, and promoted the Mo uptake, and increased its translocation rate from straw to grain (Fig. 2), leading to higher Mo status in grain. The higher soil pH in the ORF system may explain the higher Mo availability in soil. Plants take up Mo in its anion form, and the formation of anion form is negatively correlated with soil pH. Because organic fertilizers gernerally contain very low concentrations of Mo, the effect of organic fertilizers may be negligible unless the soil pH is affected. Hamnér and Kirchmann (2015) reported similar effects of organic amendment on Mo uptake 28 .
Na, Se and Al are generally considered as beneficial elements because they stimulate plant growth, but are not essential. In the current study, the Na contents in soil and straw were significantly enhanced by the ORF system in 2014. This was probably because organic fertilizers were rich in Na, which was readily available to plants in soil ( Table 6). The Na concentration in straw and grain was much higher in 2014 than in 2013 in both cultivation systems. The rainfall during the growing season was much higher in 2014 than that in 2013, and this may have increased the availability of some minerals in the soil. The influx of Na + into root is mainly by diffusion, which is affected by the field moisture status. In 2013, the rainfall deficit may have reduced the effect of organic matter on soil Na availability, leading to the similar Na status in plant and soil between CR and ORF systems. The ORF system also increased the available Se concentration in soil, decreased the Se concentration in straw, and increased the Se concentration in grain ( Table 6). The decreased Se concentration in the straw was related to the high Se translocation rate from straw to grain (Fig. 2). Although Se is mainly in organic forms in organic fertilizers, it is mineralized to selenite and selenate when organic matter decays, which may explain the increase in Se availability in soil. The effects of organic amendment on crop Se concentrations were paradoxical in previous studies. Some studies reported low bioavailability of Se in organic fertilizers and strong retention of Se by soil organic matter 28,46 . In other studies that used animal manure as a method to increase crop Se content, there were positive correlations between organic matter input and Se uptake in mineral soils 47 . The paradox with respect to the impact of organic matter input on Se uptake may related to competition between S and Se for uptake by plant root, since sulphate is also present in organic fertilizers. We did not measure the S contents in plants and soil in this study, so we can only speculated that mineralized selenite and selenate derived from organic fertilizer competed with S for root uptake. The ORF system significantly decreased the Al concentration in soil and grain in both years ( Table 6). The higher Al concentration in soil and grain in the CR system is likely due to the lower soil pH.
Cadmimum, Sr, Ba, Cr, and As are generally considered as toxic elements for plant growth. Although these nonessential metal(loid)s have no known biological functions in plants, they are taken up via essential metal uptake systems 48 . In the current study, the ORF system significantly increased the Ba concentration in the straw and grain in 2013, but did not affect the availability of Ba in soil. The increased Ba uptake by crops might be related to competition between Ca and Ba, since they are chemically similar and share the same translocation pathways 49 . The decrease of Ca uptake in the ORF system (Table 4) suggests that there was same competition for uptake and translocation. A similar competitive relationship exists between Zn and Cd. The ORF system resulted in a significant increase in available Cd content in soil, increased uptake from soil, and increased translocation from straw to grain, leading to high Cd concentrations in the grain in both years (Table 6 and Fig. 2). The opposite patterns were observed for Zn contents in soil and plants. Zinc and Cd compete for ion adsorption and desorption by clay soils, and plant roots can alter the forms of metals (e.g., by oxidation) in the rhizosphere so as to decrease their availability to plants 50 . Because Zn and Cd are chemically similar and readily available to plants in soil, they complete for uptake by the roots -this can explain their opposite patterns of uptake in the ORF system. A previous study also reported paradoxical effects of organic amendments on Cd accumulation.The increase in crop Cd concentrations in fields supplemented with organic matter was probably due to the release of Cd bound to soil organic matter and Fe/Mn-oxides 51 .
The results of the PCA indicated that the ORF system mainly affected Ba, As, Mo, Ni, P, and Cd contents in grain in 2013, and P, Mo, Zn, Fe, Se, and Cd contents in grain in 2014. Thus, the main effect of ORF system on grain was to increase the concentrations of P and trace metal(loid)s. This result was consistent with those of the ANOVA analysis of the ionome of rice grain and soil. The addition of trace elements in organic fertilizer and removal of agro-chemicals were the main reason for variations in the elemental composition of rice grains.
Deficiencies of mineral micronutirents (Fe, Zn, and Se) in the diet affect more than half of the world's population. More than 60% of the world's population is Fe deficient, more than 30% is Zn deficient, and 15% is Se deficient 52,53 . Biofortification strategies to increase the concentrations and bioavailability of the elements that are lacking in human diets are a main research focus. Low bioavailability of trace elements in soil is the bottleneck for biofortification [54][55][56][57][58][59][60] . The ORF system in this study is one biofortification to increase the micronutrient concentrations in the edible parts of rice.

Conclusion
The ORF system could maintain rice yield without excessive use of chemical fertilizers, herbicides, or pesticides. Moreover, the ORF system resulted in significant increases in P, Fe, Zn, Mo and Se contents in the rice grain, which may benefit human health. However, the ORF system also increased the concentration of Cd, a potentially toxic elment, in the rice grain, as well as its transport from the soil to the grain. Therefore, ORF system has great potential for sustainable sustainable agriculture development and for preserving resources. However, if this system is adopted for long-term use, more attention should be paid to the risks of higher Cd concentrations in the soil and grain. This is the first study to compare the ionome variations in plants and soil between a rice-frog coculture ecosystem and a conventional farming system.