Cadmium in global topsoil: Spatial distribution, associated factors, and cropland exposure screening.

Abstract

Soil cadmium (Cd) pollution is a persistent concern for soil quality, crop production, and food safety. Despite global concerns, spatial predictions of topsoil Cd remain challenging due to limitations in model accuracy, covariate representation, and nonlinearity handling. This study addresses these gaps by integrating 59,328 topsoil Cd samples worldwide to map global Cd exceedance probability and concentration distribution using classification and regression machine-learning models, with AUC of 0.90 and R² of 0.72, respectively. Variable importance analysis identifies climate factors (39.00%) and soil properties (34.01%) as the dominant factors associated with Cd exceedance probability, followed by geology (13.22%), topography (5.01%), and vegetation (4.58%), while anthropogenic contributions appear more localized at the global scale. High-resolution (1 km) maps of global soil Cd exceedance probability and predicted concentration distribution are generated. Under the moderate screening scenario, approximately 0.14% of global cropland (23,130 km²) overlapped with acidic high-soil-Cd-risk zones, mainly in China (15,201 km²), the UK (4956 km²), and Ireland (1382 km²). Approximately 1.24% of rice, 0.33% of wheat, and 0.22% of maize production were located in areas classified as high soil-Cd-risk zones, indicating priority areas for monitoring rather than direct evidence of grain Cd exceedance. These findings provide a spatially explicit basis for preliminary screening, targeted monitoring, and soil Cd management.