Antimicrobial resistance (AMR) threatens public health by undermining the effectiveness of antibiotics. However, the influence of land use on the occurrence of antimicrobial resistance genes (ARGs) along river networks remains poorly understood. National monitoring assessments show streams with degraded habitat have more ARGs, but lack of spatial resolution limits attribution to specific land use drivers across watersheds. To understand how ARG signatures vary temporally between streams and rivers of varying land use, we collected water samples during Fall 2023 and Summer 2024 from 105 sites (72 tributary and 33 mainstem sites) across four Midwestern watersheds. The watersheds included the forest-dominated Manistee (56% forest), the mixed but largely forested Muskegon (39% forest), the agriculturally and urban influenced St. Joseph (48% agriculture, 16% urban), and the agriculture-dominated Tippecanoe River (72% agriculture). For each sampling location, we analyzed water samples for four ARGs (tetW, tetQ, ermB, sul1), two fecal indicators (bacR, PMMoV), and a 16S rRNA bacterial marker. We hypothesized that ARG concentrations across watersheds would be driven by land use intensity, with higher ARG concentrations in urban influenced systems associated with human wastewater inputs and in agricultural systems associated with fertilizer inputs, relative to forested systems. We found that tetW and tetQ concentrations (gene copies mL-1) significantly differed among watersheds (generalized linear mixed model, p<0.05), with the highest estimated marginal means (EMM) observed in Muskegon and St. Joseph watersheds. In contrast, the lowest tetW and tetQ concentrations occurred in the Tippecanoe and Manistee watersheds, which were not significantly different (EMM, p>0.05). This pattern was unexpected, as the Tippecanoe watershed has the highest % agricultural land use, whereas the Manistee watershed is the most forested. We also found that ARG concentrations were positively correlated to % urban (ρ=0.31 to 0.34) and % agricultural (ρ=0.17 to 0.23) land use, while concentrations were negatively correlated with % forest (ρ=-0.13 to -0.21). Together, these findings indicate that ARG distributions along Midwestern river networks are shaped by complex interactions among land use types, rather than by agricultural intensity alone, highlighting the need to account for agricultural type (livestock versus row-crop production) when evaluating environmental ARG patterns.