1. Ecologists often make predictions about community richness and diversity using climate variables that include seasonal precipitation totals and mean daily temperatures. While means and totals can be effective predictors to a certain extent, the complexities of faunal-climate relationships might be over-simplified through the use of coarse-grained variables. 2. The goal of this study was to investigate less commonly studied climate variables, including indices of intra-annual variation in the timing and intensity of precipitation events that might be used to predict butterfly richness across an elevational gradient. Data from a long-term, single-observer dataset at four sites in California were examined with Bayesian model averaging and structural equation modelling. Species-specific responses to climate were compared with community responses at each site. 3. At lower elevations, it was found that the relative importance of climate variables shifted towards temporal patterns of precipitation, including the timing of the first storm event and the annual number of precipitation events. Heterogeneity among sites was apparent in the importance of specific weather variables, and temporal trends (across years) were detected for a small number of variables. Species-specific results paralleled those obtained from analysis of species richness, thus suggesting a commonality of response to climate across site-specific assemblages. 4. Models were improved by inclusion of the Pacific Decadal Oscillation and El Niño-Southern Oscillation indices, indicating that regional variables can profitably be included in faunal-climate relationship analyses. These results emphasise the need for researchers to examine climate variables beyond the most readily summarised means and totals.