Speciation-informed Bayesian assessment of metal(loid) exposure from Akora River fish, Ghana, for consumption guidance.

Freshwater fish from mining-influenced rivers can transfer metal(loid)s to consumers, but risk estimates in low resource settings often ignore non-detects, chemical speciation and receptor-specific uncertainty. This study assessed water, sediment and market fish from eight stations along the Akora River, Ghana, and developed a censoring-aware hierarchical Bayesian framework to estimate dietary exposure from Oreochromis niloticus and Clarias anguillaris. Eight metal(loid)s (Hg, Zn, As, Ni, Cr, Cu, Pb and Cd) were analysed by ICP-OES and cold-vapour atomic fluorescence spectroscopy. Posterior fish concentrations were propagated through speciation priors, intake assumptions and dose-response models to estimate non-cancer HI, TCR and BLL for children, WOCBA, adults and high-consumers. Cr and Ni dominated water-column contamination, reaching 55× and 23× WHO guideline values downstream, whereas fish metal concentrations showed little species-specific separation. Median non-cancer HI remained below the screening threshold for all receptor groups, but exceedance probabilities reached approximately 15-17% for children and high-consumers. TCR exceeded 10⁻⁶ in 35-78% of receptor-specific posterior draws, driven mainly by iAs and Cr(VI) assumptions. Median BLL predictions remained below 3.5 µg/dL, although upper-tail exceedance was plausible for children and high-consumers. Sensitivity analysis showed that FIR, BW and speciation/toxicological parameters dominated uncertainty, while additional precision in total-metal concentration had limited value. The results support speciation-focused monitoring, receptor-specific consumption surveys and Ghana-specific freshwater fish consumption guidance.