Spatiotemporal evolution of global soil heavy metal(loid)s mobility under climate change.

The persistent accumulation of heavy metal(loid)s (HMs) in soils poses a major risk to environmental quality worldwide, while their migration and transformation are strongly regulated by climate change. However, global-scale assessments of soil HMs mobility and its spatiotemporal evolution under climate change remain unexplored. To address these challenges, we constructed a predictive framework for soil HMs mobility by integrating interpretable machine learning with future climate scenario simulations. The optimal Extra Trees (ET) model yielded an R2 of 0.885 on the testing dataset, along with the lowest RMSE (8.76) and MAE (4.91). Model interpretability analysis identified climatic variables, including temperature and precipitation, as important factors affecting soil HMs mobility. By coupling this framework with three Shared Socioeconomic Pathways (SSPs), we projected the spatiotemporal evolution of HMs mobility, including As, Cd, Cr, Cu, Hg, and Pb, from 2025 to 2100. The results revealed pronounced element-dependent responses to future climate change: the mobility of Cd and Pb exhibited overall increasing trends, whereas those of As, Cr, Cu, and Hg showed consistent declines. Spatial analyses further demonstrated that SSP5-8.5 (high-emission scenario) drove a marked expansion of Cd mobilization hotspots in tropical and subtropical regions, while mobilization hotspots for As contracted markedly in the mid- to high-latitude areas of the Eurasian continent. These findings elucidate how future climatic shifts may influence the mobility of soil HMs, while offering a scientific foundation for assessing pollution risks and developing adaptive management approaches at a global scale.