Combined Residues of Frequently Detected Herbicides are Stronger Predictors of Soil Resistome and Virulence Potential than Single Residues

Abstract Overuse of herbicides has resulted in a global epidemic of evolved resistance in soil ecosystems. While single herbicide risk has been documented, the impact of multiple herbicide residues on the soil resistomes remains poorly understood. Our study quantified the residues of 24 herbicides, profiled antibiotic resistance genes (ARGs) and virulence factor genes (VFGs), and reconstructed 1086 metagenome-assembled genomes (MAGs) from 93 agricultural soils across China. The analytical results demonstrate that the frequently detected herbicide residues (e.g., 60–90% detection frequency) are stronger predictors of soil antibiotic resistomes than total herbicide load, unsupervised mixture based on principal component analysis (PCA), or single-herbicide models. For VFG abundance, the same mixture outperformed the best single herbicide and total herbicide concentration, although PCA–PC1 showed a moderately stronger correlation. Specifically, these commonly detected herbicides showed positive associations with ARG abundance through both direct linkages and bacterial community-dependent indirect pathways. Similar patterns were observed for virulence potential, where frequently detected residues maintained strong predictive performance and were predominantly associated with indirect pathways. Functional annotation patterns further suggest a concurrent enrichment of traits related to antioxidant defense, mobility, adherence, and immune modulation under elevated herbicide exposure. Notably, multiple herbicide residues were linked to the abundance of a Trinickia taxon, which was identified as a candidate ARG-VFG co-occurrence MAG carrying a high ARG load. Our findings highlight the need to integrate compound exposure of frequently detected herbicides into future risk assessment methodology for preventing the dissemination of ARGs in agricultural soils.