An important and largely unaddressed issue in studies of biotic–abiotic relationships is the extent to which closely related species, or species living in similar habitats, have similar responses to weather. We addressed this by applying a hierarchical, Bayesian analytical framework to a long-term data set for butterflies which allowed us to simultaneously investigate responses of the entire fauna and individual species. A small number of variables had community-level effects. In particular, higher total annual snow depth had a positive effect on butterfly occurrences, while spring minimum temperature and El Niño-Southern Oscillation (ENSO) sea-surface variables for April–May had negative standardized coefficients. Our most important finding was that variables with large impacts at the community-level did not necessarily have a consistent response across all species. Species-level responses were much more similar to each other for snow depth compared to the other variables with strong community effects. This variation in species-level responses to weather variables raises important complications for the prediction of biotic responses to shifting climatic conditions. In addition, we found that clear associations with weather can be detected when considering ecologically delimited subsets of the community. For example, resident species and non-ruderal species had a much more unified response to weather variables compared to non-resident species and ruderal species, which suggests local adaptation to climate. These results highlight the complexity of biotic–abiotic interactions and confront that complexity with methodological advances that allow ecologists to understand communities and shifting climates while simultaneously revealing species-specific variation in response to climate.