Concentration-dependent effects of metal(loid) exposure on bacterial conjugative transfer: a meta-analysis and machine learning study.

Metal(loid) contamination of aquatic and agricultural systems is a global concern, yet its impact on conjugative transfer of antibiotic resistance genes under environmentally relevant concentrations remains unquantified. We synthesized 556 effect sizes from 56 studies using three-level meta-analysis with machine learning. This approach identified concentration and metal type as the most important predictors. The U-shaped dose-response curve showed a positive association at low concentrations (<∼1 mg/L), the strongest negative association near 322.6 mg/L, and a weakening of the negative association above this point. This pattern was most pronounced for copper (82.7% decrease at high concentrations). Recipient genera, including Salmonella, Acinetobacter, Klebsiella, and Enterococcus showed substantially stronger positive associations than E. coli recipients. Reactive oxygen species increased across donor, mixed, and recipient bacteria (64.2 - 91.3%) and correlated with overall effects (rho = 0.417, P < 0.001), suggesting oxidative stress as a potential mechanistic pathway underlying the observed associations. Metal concentrations typical of natural waters and agricultural systems generally fall below approximately 1 mg/L, within the range associated with increased conjugation. This alignment extends to regulatory limits: even WHO guidelines for As, Cd, Pb, and Cr, and Chinese Class I standards for Cu and Zn, fall within this range, underscoring its regulatory relevance. These findings provide a quantitative framework for assessing ecological risks associated with metal-mediated ARG dissemination and offer a basis for refining pollution control strategies.