Depth-dependent contamination and vertical distribution of metal(loid)s in mining-affected soils: Insights from Bayesian hierarchical inference.

Metal(loid) contamination around mining tailings can extend into deeper soil horizons. However, conventional layer-wise comparisons often cannot distinguish depth-dependent patterns from site-level spatial heterogeneity, limiting the reliable assessment of subsurface environmental risks. This study combined geo-accumulation indices and Bayesian hierarchical modeling of 114 soil samples from 39 sites at three depths to characterize the depth-dependent distributions of metal(loid)s (As, Co, Cr, Cu, Fe, Mn, Ni, Pb, Sb, and Zn) while accounting for site-level variability in the Tonglvshan copper-iron tailings area of Central China. The results showed strong element-specific enrichment and vertical differentiation. Cu exhibited the most severe contamination, with a mean geo-accumulation index (Igeo) value of 3.62 and persistent enrichment across the sampled intervals. Sb was enriched across depths but showed weak differentiation after site-level adjustments. As and Zn showed the clearest posterior depth contrasts, which were characterized by surface enrichment, middle-layer depletion, and partial deep-layer recovery. Pb displayed a shallow enrichment tendency, but this pattern weakened after accounting for site effects. Co and Cr showed weak profile-scale separation. Fe, Mn, and Ni exhibited limited vertical differentiation. Most metal(loid)s were classified as transitional after site-level adjustment, indicating a limited representative profile. These findings support adaptive, depth-stratified monitoring in the study area and illustrate how probabilistic inference may inform assessments in other tailings-affected settings, subject to site-specific validation.