Environmental drivers and spatial patterns of antibiotic-resistant, enteric coliforms across a forest–urban riverscape

Antibiotic resistant bacteria are prevalent environmental contaminants in freshwaters, and antibiotic resistance genes circulate throughout the urban water cycle. The increase of antibiotic resistant pathogens threatens public health through direct and indirect exposure, and natural resource managers need information on the spatial patterns of antibiotic resistant bacteria and environmental factors associated with their distribution to improve water quality monitoring and to better assess human, animal, and environmental health risks. We collected water and epilithic biofilm samples and measured physicochemical environmental variables at 29 sites distributed longitudinally in the Green-Duwamish River basin, Washington, USA. We characterized catchment-wide patterns of gram-negative fecal indicator bacteria and hypothesized that the presence of antibiotic resistance would be associated with environmental heterogeneity, bacterial primary ecology, stream compartment, and stream type. Antibiotic resistance was determined by microbial growth on selective media supplemented with 3 different antibiotics (ampicillin, chloramphenicol, or tetracycline). Phenotypic antibiotic resistance was positively associated with disturbance, but resistance to at least 1 antibiotic was also detected in undeveloped river segments, with an 83% overall detection rate (i.e., 24 out of 29 sites, 17 in the mainstem and 7 in tributaries). The most probable number of Escherichia coli was associated with higher levels of antibiotic resistance of non-E. coli coliforms across the basin (ρ = 0.38, p < 0.01) but was not associated with antibiotic resistance of E. coli. Phenotypic resistance was highest among non-E. coli coliforms in the water column of tributaries draining moderately to extensively developed subcatchments. Generalized linear mixed-effects model results showed that 18% of the variance in presence of antibiotic resistance was explained by the fixed effects (summed CV across environmental variables, stream type, primary ecology, and stream compartment), and when a spatial random effect was included, the model explained 27% of the variance. Our study provides new evidence that environmental factors and bacterial primary ecology are important underlying factors associated with spatial patterns of antibiotic resistant enteric coliforms. We used macroecological concepts and a riverscape approach to characterize the distribution of antibiotic resistance with methods applicable to municipalities.