Safe utilization of cadmium-rich soil for planting lilies and maize using a random forest model based on soil properties.

The factors influencing the uptake of soil Cd by crops are complex and closely related to the different crop varieties. Efficient and safe utilization of land resources with high soil Cd levels has become a significant challenge in the scientific community. This study focuses on the anomalously high Cd distribution area in the northern part of Longshan County, Hunan, China. By systematically collecting and testing Cd content in the edible parts of lily and maize, as well as corresponding root soil Cd, pH, and oxides, the study reveals the differences in the bioconcentration factors of Cd (BCF-Cd) for lily and maize and their influencing factors. Using the random forest method and hyperparameter optimization, optimal prediction models for BCF-Cd in lily and maize were established. The results indicate that the BCF-Cd of lily is significantly higher than that of maize. The primary factors influencing BCF-Cd in lily and maize include soil pH, Mn, OM, and ba. Feature importance analysis identifies pH as the most critical factor affecting BCF-Cd in both lily and maize. Based on the prediction results of the random forest model, this study proposes a zoning scheme for the safe utilization of arable land to maximize benefits while ensuring the medicinal safety of lily and the food safety of maize. This provides scientific evidence for ensuring food security and maximizing the productive value of land resources.