Prediction of arsenic, cadmium, and lead bioaccessibility for regional-scale management of legacy-contaminated soils

Urban and peri-industrial soils contaminated by past urban and industrial activities frequently exhibit elevated contents in potentially toxic elements (PTEs) such as arsenic (As), cadmium (Cd), and lead (Pb) that exceed regulatory thresholds. Accurate characterization of this contamination is essential for effective risk management of exposed populations. In this context, assessing oral bioaccessibility is a critical step for reliably evaluating exposure risks via the ingestion of soil particles and preventing the overestimation of human health hazards often caused by relying solely on total element contents. However, large-scale bioaccessibility assessments remain scarce, as most studies rely either on total contents or small, site-specific datasets, limiting the development of regionally applicable risk assessment frameworks. This study quantifies the oral bioaccessibility of As, Cd, and Pb at the regional scale, based on aqua regia contents and Unified BARGE Method (UBM) measurements for 223 topsoil samples collected across major historical industrial basins in Wallonia (Belgium). Chemical fractionation (BCR protocol) and soil properties were used to identify the main controls on PTEs bioaccessibility. The results show that while As is mainly controlled by the geogenic background, Cd and Pb are predominantly of anthropogenic origin, frequently exceeding regulatory thresholds. We further demonstrate that actual exposure risks are heavily modulated by chemical speciation: on average, only 24% of total (aqua regia content) As is bioaccessible, compared to substantially higher fractions for Cd (82%) and Pb (57%), which are primarily associated with reducible Fe/Mn oxyhydroxide phases. Bioaccessible contents can be predicted from total (aqua regia) element contents ( R 2 = 0.47–0.98), increasing to R 2 = 0.83–0.95 for As when the cumulative BCR content after Step 3 is included. Based on a large, consolidated dataset covering seven industrial areas, this approach enables the integration of regionally derived bioaccessibility values into human health risk models, while supporting the development of site-specific tools, including predictive models based on routine soil data and spatial mapping for land managers. The proposed framework is transferable to similar contaminated environments worldwide, contributing to reduced uncertainty in regional-scale risk assessments and guiding evidence-based soil management strategies.