Insufficient Zn uptake reduces rice grain yield in integrated rice-crayfish culture – a case study in the Jianghan Plain, China

ABSTRACT China has recently seen a substantial increase in integrated rice-crayfish (Procambarus clarkii) culture (IRCC). However, grain yield reductions (GYR) with rice plants having normal vegetative growth but failed filling are becoming common in IRCC, and the driving factors remain unclear. Here, we tried to identify the mechanisms behind this GYR mode in the Jianghan Plain, China, by investigating the soil properties, grain yield parameters, and plant mineral content in five IRCC plots experiencing GYR and an adjacent paddy-upland rotation plot (CK). The results showed slight to extreme GYR (11–82%) in the IRCC plots compared to those in CK. The translocation of phosphorus from shoot to grain during the productive stage of rice plants was inhibited in IRCC plots (26–71%) compared to CK (75%). The grain yield and phosphorus translocation coefficient correlated strongly (r > 0.95) with zinc uptake in aboveground rice organs which was significantly lower in IRCC than in CK. We conclude that this GYR in IRCC plots is not due to decreased macronutrients availability or heavy metals toxicity in IRCC soils, but the insufficient uptake of zinc by IRCC rice plants. We recommend Zn foliar spray or introducing Zn efficient rice cultivars as potential mitigation measures.


Introduction
Due to the consumption-oriented explosion of the crayfish industry, China has recently seen a substantial increase in integrated rice-crayfish (red swamp crayfish, Procambarus clarkii) culture (IRCC) (Jin et al. 2020). The total area of IRCC in China reached 1.1 million ha by the end of 2019, accounting for 60% of the national area of integrated rice-fish culture (China Fisheries 2020). IRCC can increase the economic benefits of rice paddies from the harvest of both rice and crayfish and from decreased use of commercial materials such as fertilizers and pesticides (McClain and Romaire 2004). However, field investigations have found that an increasing number of rice paddies in IRCC cannot achieve the expected economic benefits due to grain yield reductions (GYR) (Cao et al. 2017). There is an urgent need to identify the mechanisms behind the GYR in IRCC to promote the sustainability of IRCC.
The GYR in IRCC could be related to the significantly altered water-management regime compared to that in traditional rice-fish culture. In rice-fish culture, fishery species (e.g. fish and crab) are bred in rice paddies during a period similar to that of rice and do not influence rotations of rice and

Study area and site selection
The study was conducted in the Jianghan Plain, located in the middle reaches of the Yangtze River, China. This area has a subtropical humid monsoon climate, with a mean annual temperature of 17.1°C, the lowest temperature in January (3.8°C) and the highest temperature in July (28.9°C). The mean annual precipitation is 1269 mm, of which 30-50% occurs in summer (Zhou et al. 2012). The Jianghan Plain is characterized by highly fertile rice paddies and is one of the most important rice production regions in China. Rice paddies in IRCC in this region occupied more than 20% of the national IRCC area in 2016 (Yu et al. 2018).
The rice plots studied were located in Honghu County, in the southeastern Jianghan Plain (Figure 1), which was the county with the second-largest crayfish production in 2019 (China Fisheries 2020). Based on the field investigation immediately before the rice harvest in October 2019, we randomly selected 5 rice plots with 1-3 years of IRCC (H02-H06) that grew the same rice cultivar (Oryza sativa indica, var. Huanghuazhan) and had obvious GYR. An adjacent RRD plot (H01) was selected as the control (CK, Figure 1).  and the rice plants in it were collected from the selected rice paddies immediately before the rice harvest in October 2019. The samples were taken from five points along two diagonals in each plot using a steel shovel and then transferred to the laboratory in plastic bags. After the plant roots were carefully separated, the soils from each plot were mixed uniformly and airdried at room temperature (at 20 ± 2°C) for 28 days. Soils were then ground after removing visible non-soil materials, sieved through a 40-mesh sieve and stored in a refrigerator at 4°C until analysis. Basic soil physicochemical properties and elemental composition were analyzed for all soils as shown in Table 1 (detailed analytical methods were shown in Table S1).

Sample preparation and analysis
Rice plants were first washed with tap water to remove visible non-plant material and soil attached to roots. After rinsing three times with deionized water, the rice plants were cut into three parts (i.e. root, shoot, and grain) and dried in an air-circulation oven at 60°C to a constant weight. The rice plant samples were then ground and stored in a refrigerator at 4°C until analysis. Before being ground, one subsample of grain was used to estimate rice yield parameters via manual counting (Liu et al. 2016). In the field investigation in 2019, vegetative parameters of rice plants, such as the lengths and quantities of roots, stems, leaves, as well as the number of grains did not appear to differ between the IRCC and RRD plots selected.
Contents of elements that are closely related to rice growth were estimated in different organs of rice plants after acid digestion (EPA Method 3052), by ICP-OES for P, Fe, and Si, and by ICP-MS for As, Zn, Cu, and Cr.
To estimate the rice yield (t ha −1 ) of the IRCC plots studied, we first set the yield of the CK plot to 8.5 t ha −1 , according to the regional average (Cao et al. 2017). The total number of rice grains per ha (NG) was then calculated as a function of the TKW and PFG of CK: The rice yield of an IRCC plot (i) was calculated as a function of the plot's TKW and PFG, as well as NG: Previous studies reported an economic coefficient (ratio of grain:shoot biomass) of 1 in rice plots with a normal grain yield in the Jianghan Plain (Xu et al. 2011). Thus, the biomass of shoots in all IRCC plots was assumed to be 8.5 t ha −1 . With the rice yield estimated for all plots, the accumulation and translocation coefficient (i.e. the ratio of the element content in grain to the overall element content in aboveground organs) of main elements in rice aboveground organs (i.e. the sum of grain and shoot) were calculated.

Soil anoxic incubation and analysis
To simulate changes in the activity of key elements in IRCC soils during long-term flooding, anoxic incubation experiments were conducted for all soils studied. Soil samples (20 g) and 300 ml deionized water were transferred into 300 ml glass bottles. After sealing the bottles with Parafilm (Bemis Company), the mixture was shaken to disperse the soil, and all bottles were incubated in the dark at 20 ± 2°C. A 95-day incubation period ensured that anoxic conditions in soils were completely established. After incubation, the pH and redox potential (Eh) of the overlying water were measured 1 cm above the soil surface using a potentiometer, and conductivity was measured at the same place using a conductivity meter. A sterile syringe was then used to sample supernatants from the same place where pH was measured. The supernatants were filtered using pre-washed cellulose-acetate filters (0.45 μm) to measure dissolved molybdate-reactive P (MRP), NH 4 + , dissolved organic C (DOC), Fe(II), and the elemental composition. Aliquots for MRP, NH 4 + , DOC, Fe(II) and total Fe analyses were immediately acidified with dilute sulfuric acid and analyzed using the molybdate blue colorimetric method for MRP (precision: ± 5 μg l −1 ; Murphy and Riley 1962), the indophenol blue colorimetric method for NH 4 + (precision: ± 0.02 mg l −1 ; Nelson and Sommers 1980), a total organic analyzer for DOC (precision: ± 5%; Analytic Jena, Multi N/C 3100), and the 1.10 phenanthroline colorimetric method for Fe(II) (precision: ± 5%; O' Connor et al. 1965). The K, Si, and Mn concentrations in the filtrates were analyzed by ICP-OES (precision: ± 5%), while As, Zn, Cu, and Cr were analyzed by ICP-MS (precision: ± 5%).
Soils before and after incubation were extracted using an oxalate-oxalic acid solution, to estimate changes in the amorphous fractions of soils after long-term flooding (Darke and Walbridge 1994). P and Fe concentrations in the extracts were analyzed by ICP-OES, while As, Zn, Cu, and Cr concentrations were analyzed by ICP-MS.

Statistical analysis
All calculations and statistical analyses were performed using SPSS software (IBM Statistics version 22). Differences in (a) physicochemical properties and the elemental composition of soils, (b) elemental composition of plant organs, and (c) physicochemical composition of anoxically incubated soil supernatants were assessed using one-way analysis of variance with Tukey's post hoc test at a significance level of p < 0.05. Differences in the amorphous elemental composition of soils before and after anoxic incubation were assessed using Student's t-test. Table 1. Mean ± standard deviation of topsoil characteristics (0-20 cm) of the rice plots in rotations of rice and dry-season crops (RRD) (the control (CK)) and integrated rice-crayfish culture (IRCC) studied.

Grain yield characteristics of the rice plots studied
The GYR was significant in IRCC plots compared to CK (p < 0.05, Table 2). The GYR ranged from 10.5% (in H02) to 81.2% (in H04), with a mean of 58.8%. The TKW was also significantly lower in IRCC plots than in CK (p < 0.05, Table 2), ranging from 77.2% (in H06) to 95.5% (in H02) of that in CK, with a mean of 84.4%. The PFG was significantly lower in most IRCC plots (except for H02) than in CK (p < 0.05, Table 2). The mean PFG was 45.0%, with the lowest values of 21.4-30.0% in H03-H05.

Difference in element contents in rice organs
The contents of main elements in each organ of rice plants varied between IRCC plots and CK (Table 3). In rice grain, P, Fe, As and Zn contents were significantly higher in IRCC plots than in CK, except for Fe and Zn in H02. The difference in Si, Cu and Cr content between IRCC plots and CK varied among plots, with no obvious trend. The P, Fe, and Zn contents in the grain of IRCC plots were significantly higher than those in CK (p < 0.05, Table 3). In the rice shoot, P and As contents were nearly twice as high in IRCC plots as those in CK, but Fe and Cr contents did not differ significantly. Si and Cu contents varied among plots, with no obvious trend. Zn contents were significantly lower in IRCC plots than in CK, except for H02. The mean P, As and Zn contents in the shoot in IRCC plots differed significantly from those in CK (p < 0.05, Table 3). In the rice root, P, Fe, Si, and As contents were generally higher in IRCC plots than in CK, except for P in H04 and H06, and As in H06. Conversely, Zn, Cu, and Cr contents were significantly lower in IRCC plots than in CK, except for Zn in H03 and Cu in H02-H03. The mean Fe, Zn, Cu, and Cr contents in the root in IRCC plots differed significantly from those in CK (p < 0.01, Table 3).

Accumulation and translocation of main elements in aboveground rice organs
The total P accumulation in the aboveground organs of IRCC rice plants ranged between 13.5 and 19.6 kg ha −1 , with a mean of 16.8 kg ha −1 . The total P uptake by IRCC rice plants did not differ significantly from that in CK (p > 0.05), except for H04 ( Figure 2). The translocation coefficient of total P uptake to grain in IRCC rice plants ranged between 26% and 71%, with a mean of 41%, which was much lower than that in CK (75%). Total As accumulation in the aboveground organs of IRCC rice plants ranged between 19.0 and 61.2 kg ha −1 , with a mean of 37.4 kg ha −1 . Total As uptake by IRCC rice plants was significantly higher than that in CK (p < 0.05), except for H04-H05 (Figure 2). The translocation coefficient of total As uptake to grain in IRCC rice plants ranged between 25% and 75%, with a mean of 47%, which was lower than that in CK (71%). 16.04 ± 6.46** 9.81 ± 3.11** 4.20 ± 0.38** Note. Different superscript letters in the same column indicate significant differences in means at p < 0.05. '*' and '**' represent significant differences between means of IRCC plots and CK at p < 0.05 and p < 0.01, respectively.
Total Si, Zn, Cu, and Cr accumulations were significantly lower in the IRCC plots than in CK, except for Si, Zn, and Cu in H02, indicating that IRCC rice plants generally took up less of these elements (Figure 2). The mean translocation coefficient of these elements was generally high: 78%, 89%, 80%, and 90% for Si, Zn, Cu, and Cr, respectively.

Changes in main elements in IRCC soils under anoxic condition
Soil total P, Fe, As, Zn, Cu, and Cr contents did not differ significantly between IRCC plots and CK, except for P, Fe, As, and Cu in H04, which were significantly lower than those in CK (p < 0.05, Figure 3, Table 1). Most of the soil P was in the amorphous fraction of all studied soils, ranging between 45% and 74%, with a mean of 59% (Figure 3a). The amorphous fraction also represents a large percentage of the total contents of Fe, As and Cu in soils, with means of 31%, 40%, and 49%, respectively (Figure 3b, c, e). Conversely, amorphous Zn and Cr represented a small percentage of total Zn and Cr contents in soils, with means of 13% and 16%, respectively (Figure 3d, f).
The influence of anoxic incubation on the amorphous soil fraction differed among elements (Figure 4). Amorphous P and Fe contents increased significantly after anoxic incubation (p < 0.01, Figure 4a, b). Conversely, amorphous Cu content decreased significantly after anoxic incubation in most soils (p < 0.01, Figure 4e). For As, Zn, and Cr, the influence of anoxic incubation on their amorphous content depended greatly on the plot. Amorphous As and Zn contents increased significantly after anoxic incubation in CK and H02-H03 but decreased significantly in H04-H06, except for a non-significant increase in amorphous Zn content in H05 (Figure 4c, d). Amorphous Cr content increased after anoxic incubation in CK and H03 but decreased in the other plots (Figure 4c, d).
Physiochemical properties of supernatant after anoxic incubation also differed between IRCC soils and CK. The pH of the supernatant was nearly neutral for all samples, ranging between 7.0 and 7.1 ( Table S2). The Eh and conductivity of the supernatant were generally higher in CK than in IRCC, except for H04, which had the highest Eh and lowest conductivity (−144 mV and 316.3 µS cm −1 , respectively). DOC, NH 4 + -N, and Fe(II) concentrations were generally higher in CK than in IRCC soils, with the difference increasing as the IRCC duration increased. For MRP, Si, Mn, and K, no obvious trends between CK and IRCC samples were observed (Table S2). Among the heavy metal elements released, As concentrations were 10 times higher than Zn, Cu, and Cr concentrations, which were generally low and did not differ significantly between CK and IRCC samples (Table S2).

Figure 2.
Accumulation and distribution of P, Si, As, Zn, Cu, and Cr in aboveground rice organs (grain and shoot). Different letters above each column indicate a significant difference in means at p < 0.05. Percentage values in each column represent the translocation coefficient of total element uptake to rice grain.

Discussion
In the IRCC system, GYR can result from diseases, pests, or destructive effects of crayfish (Anastácio et al. 2005a;Cao et al. 2017), or a decrease in nutrient availability or increase in accumulation of toxic substances in soils caused by hypoxic environment . Field investigation before the present study indicated no significant difference in the vegetative parameters of rice plants in the IRCC and RRD plots studied, which indicated that the GYR in the present study was not due to diseases, pests, or destructive effects of crayfish.
Soil analysis indicated that the basic soil properties of the IRCC plots studied differed significantly from those of traditional RRD (CK), especially for multi-year IRCC plots (H05-H06). These multi-year IRCC soils generally had higher soil organic matter, total N, available N, and organic P contents, but lower P availability (water-extractable WEP and degree of P saturation) and slow-release potassium  . Soil oxalate-extractable fraction of P, Fe, As, Zn, Cu, and Cr contents before and after anoxic incubation. '*' and '**' represent significant differences in means at p < 0.05 and p < 0.01, respectively.
(K) than RRD soil (Table 1), which is consistent with previous studies of changes in soil properties after implementing IRCC Yuan et al. 2020;Hou et al. 2021aHou et al. , 2021b. However, the present study indicates that the extreme GYR observed cannot be explained simply by changes in soil physicochemical properties.
The decreased P and K availability in IRCC soils can reduce rice yield significantly, since P is an essential nutrient at all stages of rice growth, especially reproductive stages (Ye et al. 2019), and K deficiency in the soil can limit translocation of P to rice grain (Ashley et al. 2006). However, P and K availability may not have been the main factors that caused the GYR in the present study, because the increase in P and K availability did not prevent plot H03 from having an extremely large GYR. On the other hand, the commonly applied straw-return practice and low feed-use efficiency by crayfish in IRCC increase the soil organic matter content and result in continuous reducing conditions in IRCC soils (Si et al. 2017;Li et al. 2018). These conditions can promote reductive dissolution of Fe-oxides in soils, as indicated by the decrease in Fe content in soils of multi-year IRCC plots compared to that in RRD (Figure 3; Table S2), thus increasing the availability of heavy metals. Fe also works with aerenchyma cells in rice roots to form Fe plaque on root surfaces, which serves as an important barrier that prevents rice from adsorbing heavy metals (Liu et al. 2006;Cui et al. 2019). The decrease in Fe contents and increase in heavy metal availability could increase the toxic effects of heavy metals such as As, Zn, Cu, and Cr on the rice plants in IRCC plots. However, results of the present study indicate that the release of these heavy metals under anoxic conditions and their accumulation in aboveground rice organs were not significantly higher in IRCC soils than in RRD ( Figure 2, Table  S2). Toxicity from heavy metals can also be excluded as a main factor that causes the extreme GYR in the present study. The release of As was one magnitude higher than that of other heavy metals (Table S2), likely due to the strong affiliation of As with Fe oxides in soil (Liu et al. 2006). In fact, As contents in soils of the present study lay within the normal range of agricultural soils in the Jianghan Plain region, which generally poses a low ecological risk (Zhou et al. 2018).
Since the vegetative growth of rice plants did not differ significantly between the RRD and IRCC plots studied, the extreme GYR in the IRCC plots was likely caused by factors that restricted the productive growth of rice. Further investigation of rice plant nutrition indicated that IRCC rice plants accumulated a similar amount of P in the aboveground organs as those in RRD, but most P remained in the rice shoot ( Figure 2). This suggests that the translocation of P from rice shoot to grain was inhibited during the productive stage of rice plants in IRCC plots, which could result in serious GYR because shoots are major P sources for rice grain during the filling stage (Rose et al. 2010). The results also indicate that the uptake of essential microelements such as Zn and Cu was significantly lower in H03-H06 than in RRD and H02, which had normal to slightly reduced rice yields (Figure 2). Given the close interaction between P and Zn during their translocation in plants (Xie et al. 2019), the insufficient Zn supply appears to have influenced the translocation of P from rice shoot to grain and reduced the grain yield. We correlated the Zn accumulation in aboveground rice organs with the grain yield and translocation coefficient of aboveground P ( Figure 5). The results indicated that Zn uptake in aboveground organs influenced the grain yield and P translocation significantly (r = 0.968 and 0.949, p < 0.01), which confirms that the insufficient uptake of Zn was directly responsible for the GYR in the IRCC plots.
Further studies are needed to identify specific biological mechanisms that limited translocation of P to rice grain due to an insufficient Zn supply. This research lay beyond the scope of the present study, which limits itself to clarifying the limiting factors and recommending feasible mitigation measures to address the GYR in IRCC. One possible reason for the insufficient Zn supply in IRCC is the enhanced reducing conditions under IRCC, which could encourage Fe plaque to form on root surfaces and inhibit Zn uptake by rice roots (Impa and Johnson-Beebout 2012), as confirmed by the significantly higher Fe but lower Zn contents in the roots in IRCC than in RRD (Table 3). This is also consistent with previous results that showed that continuous flooding of rice plots can inhibit Zn uptake by rice plants (Wang et al. 2014). Another factor that could explain the insufficient Zn supply lies in the characteristics of rice plants to uptake and translocate Zn. Research has shown that most of the Zn taken up by rice roots in vegetative growth stages accumulated in the root and stem, and was translocated rapidly to the grain during the filling stage, especially under Zn-deficient conditions (Ishimaru et al. 2011;Impa et al. 2013). In the Jianghan Plain, heading and flowering stages of monoculture rice usually occur in July and August, which have high daytime temperatures (as high as 39°C) and less frequent precipitation. The continuous flooding and high soil organic matter content in IRCC plots place the soils under continuous extreme anoxic conditions and exacerbate the Zn deficiency during rice vegetative growth, which could result in significantly lower Zn contents in the roots and shoots in IRCC than in RRD (Table 3), thus causing insufficient P translocation to the grain and reducing the rice yield.
Studies have reported that insufficient Zn supply in rice paddy soils during continuous flooding could be mitigated by alternating drying and rewetting of soils (Wang et al. 2014). However, this method would require more labor, increase water consumption, and interfere greatly with crayfish breeding. Based on the results of this and previous studies, we recommend applying foliar Zn sprays before the rice flowering stage in IRCC, to decrease Zn deficiency during vegetative rice growth. This feasible measure may attenuate the GYR in IRCC, while it has little influence on crayfish yield, and does not increase water consumption. We emphasize that the frequency of Zn foliar spray and concentration of Zn in spray solution will need further investigation, and methods such as the inclusion of Zn efficient rice cultivars should also be considered when designing practical mitigation measures.

Conclusions
The present study investigated characteristics of soil and rice plants in IRCC plots that experienced GYR and a traditional RRD plot with normal yield in the Jianghan Plain. The GYR in the IRCC plots studied is mainly manifested by the decrease in TKW and PFG. The results demonstrate that GYR in the IRCC plots is not due to a general decrease in macronutrient availability (e.g. P and K) or heavy metal toxicity (e.g. As and Cr) in IRCC soils. The aboveground organs of IRCC rice plants accumulated sufficient P, but most of it remained in the shoot, while the uptake of Zn was significantly lower than that in RRD rice plants. The grain yield and translocation coefficient of P in aboveground rice organs depend greatly on Zn uptake. The results confirm that the insufficient Zn uptake in IRCC rice plants was directly responsible for the GYR in the IRCC plots studied and was likely due to the Zn uptake regime of rice plants and enhanced reducing conditions in IRCC that inhibit Zn availability. We recommend applying foliar Zn sprays before the flowering stage of rice as a mitigation measure. This feasible measure may attenuate the Zn deficiency of rice in IRCC, with little influence on farmers' profits.