Rice is the staple crop in southern China as well as for more than half of the world population. In southern China, rice safety is threatened by elevated concentrations of toxic metal(loid)s accumulating in rice grain. Mining related industries in this ore-rich region extensively dispersed metal(loid)s to streams and rivers which were used for irrigation of rice paddies leading to soil contamination. Among toxic metal(loid)s accumulated in paddy soils, cadmium (Cd) is of particular concern due to its high mobility and soil-to-rice grain transfer factor in acidic soils, which are typical in southern China, as well as its long biological half-life in humans (>10 years) after ingestion.

The objective of this study was to reveal if rice grain Cd can be related to soil physico-chemical characteristics at a relatively large regional scale, which may be used to predict rice grain Cd based on an empirical model. A previous study by Simmons and co-workers demonstrated CaCl₂ extractable Cd using field-moist soil instead of conventionally air-dried soil during grain-filling allowed reliable prediction of rice grain Cd for 20 fields along a 3 km length of the Mae Tao Creek in Thailand. However, it remains unclear whether similar moist soil extraction can serve as a general test method to predict rice Cd for paddy soils having strongly varying soil physico-chemical characteristics.

For this, we collected 35 paired soil-rice grain samples across five provinces in southern China at the time when rice was ready to be harvested during October to November 2017. Soil extraction with 0.01 M CaCl₂ was conducted on site with field-moist soil immediately during sampling to avoid changes in Cd speciation. For comparison, soil extractions were also performed post-sampling in the laboratory using air-dried soils. Element concentrations as analyzed in field-moist soil extracts and in rice grains together with total soil analyses were used as input variables in an explorative algorithm implemented in the R software to generate best multi-variable regression models for rice Cd. We found that rice Cd correlated more strongly to extractable Cd from field-moist soils (R²=0.54) when compared to extractable Cd using air-dried soils (R²=0.27). This may be due to oxidation of Cd sulfides during drying, leading to a co-mobilization of Cd and S. Our data analysis indicated that field-moist soil extractable Cd and Fe formed the best two-variable model for rice Cd (R_rob²=0.64) with extractable Cd contributing 89% to the model-explained variance in rice Cd. Our study demonstrated that 0.01 M CaCl₂ extraction at rice harvest using field-moist soil was able to predict rice Cd at a relatively large regional scale with ranged soil physico-chemical properties. Based on the findings presented in this study, bioavailable Cd accumulating in the rice grain may reliably and quickly be tested by 0.01 M CaCl₂ extraction.