Understanding species interactions in ecological communities remains a central challenge, particularly in freshwater ecosystems shaped by variable flow regimes. The relative importance between environmental filtering and biotic interactions is still debated, especially in systems with extreme disturbance events such as droughts and floods. We used species distribution models for 28 macroinvertebrate taxa (SDMs) to evaluate the relative importance of environmental filtering versus biotic interactions across 100 sites located in the upper Colorado River, Texas and four tributaries. The sites were grouped into three flow regimes: dry-unstable (n = 7; flashy systems that dry), dry-stable (n = 6; drying with stable baseflow), and no-drying (n = 7; moderately stable, perennial). Abiotic predictors included hydrological indices and local habitat variables. Biotic predictors, used as proxies for species interactions, included species turnover (Jaccard, Bray–Curtis), co-occurrence (Spearman), and community-weighted traits. We tested three predictor sets (abiotic, biotic, combined) and evaluated the relative contribution of biotic predictors. We hypothesized that biotic effects would be strongest under harsher (drying) conditions due to increased competition and facilitation. While biotic predictors were frequently selected across models, their relative importance did not differ significantly among flow regimes. However, analyses of community composition revealed significant species turnover between dry-unstable and the other two regimes, suggesting strong environmental filtering at the flow regime level. Both biotic (e.g., Jaccard, Spearman) and abiotic (e.g., extreme low duration, annual coefficient of variation, conductivity) variables were consistently important across models. These findings underscore the complementary roles of environmental filtering and biotic interactions in shaping species distributions, with flow regimes acting as broad environmental filters and biotic context refining community composition at finer scales.