Abstract
Antibiotic resistance (AR) in surface waters reflects the influence of environmental pressures on microbial communities within aquatic ecosystems. However, how these communities respond at genetic and phenotypic levels remains incompletely understood, particularly across hydrologically dynamic tropical watersheds. We examined spatial–temporal resistance patterns in Escherichia coli isolated from a tropical peri-urban river draining a rural-to-urban gradient. Across four seasonal campaigns, 145 confirmed E.coli isolates were characterized by resistance phenotypes against 13 antibiotics and by associated resistance genes.The Multiple Antibiotic Resistance (MAR) index ranged from 0.08 to 0.77 (mean = 0.36), with 90.3% of isolates exhibiting MAR ≥ 0.2, consistent with exposure to high-risk contamination sources within this resistant-enriched E.coli fraction. Binomial GLM models showed stronger deviance reduction for genotypic resistance than for phenotypic resistance, with genotype models accounting for up to 64.1% of the null deviance and phenotype models for up to 42.6%.Resistance patterns varied consistently across seasons and sites, indicating that hydrological conditions influenced how genotypic and phenotypic resistance were distributed along the river gradient. In genotypic resistance models, most deviance reduction was achieved by the spatial–temporal block, with additional improvement after incorporating microbiological predictors. In contrast, the best-supported phenotypic model required spatial, ecological, and microbiological parameters. These findings indicate that resistance-gene monitoring provides information that is not fully captured by phenotype-based assays and may improve AMR assessment in tropical freshwater systems.