Influence of soil properties on cadmium accumulation in vegetables: Thresholds, prediction and pathway models based on big data

Soil properties, such as soil pH, soil organic matter (SOM), cation exchange capacity (CEC), are the most important factors affecting cadmium (Cd) accumulation in vegetables. In this study, we conducted big data mining of 31,342 soil and vegetable samples to examine the influence of soil properties (soil pH, SOM, CEC, Zn and Mn content) on the accumulation of Cd in root, solanaceous, and leafy vegetables in Hunan Province, China. Specifically, the Cd accumulation capability was in the following order: leafy vegetables > root vegetables > solanaceous vegetables. The soil property thresholds for safety production in vegetables were determined by establishing nonlinear models between Cd bioaccumulation factor (BCF) and the individual soil property, and were 6.5 (pH), 30.0 g/kg (SOM), 13.0 cmol/kg (CEC), 100–140 mg/kg (Zn), and 300–400 mg/kg (Mn). When soil property values were higher than the thresholds, Cd accumulation in vegetables tended to be stable. Prediction models showed that pH and soil Zn were the leading factors influencing Cd accumulation in root vegetables, explaining 87% of the variance; pH, SOM, soil Zn and Mn explained 68% of the variance in solanaceous vegetables; pH and SOM were the main contributors in leafy vegetables, explaining 65% of the variance. Further, variance partitioning analysis (VPA) revealed that the interaction effect of the corresponding key soil properties contributed mostly to BCF. Meanwhile, partial least squares (PLS) path modeling was employed to analyze the path and the interactive effects of soil properties on Cd BCF. pH and SOM were found to be the biggest two players affecting BCF in PLS-models, and the most substantial interactive influence paths of soil properties on BCF were different among the three types of vegetables.