Aim: Spatial turnover in plant species composition reflects the interplay between species' environmental requirements and their dispersal dynamics. However, its origins are also historical, arising from speciation, extinction and past range dynamics. Here, we test the effects of current environmental gradients and geographical distance on regional species turnover. Using a new approach, we then interpret residual turnover patterns, unexplained by these factors. Location: Native vegetation in the wheatbelt of the Southwest Australian Floristic Region (SWAFR). Methods: We fitted a generalized dissimilarity model of species turnover across 650 floristic inventory plots as a function of current climate, topography, soils and inter-plot distances. We then decomposed the model errors into independent axes of residual variation, representing unexplained species turnover. We mapped the spatial patterns evident on the main residual axes, and interpreted these using ancillary data on the life form and geographical ranges of associated species, the distributions of some major habitat types, and anomalies between recent and Last Glacial Maximum (LGM) climate. Results: Soil and climatic variables (especially phosphorus and wet quarter precipitation) made sizeable independent contributions to explaining deviance in species turnover, whereas deviance explained by inter-plot distances and climate co-varied strongly. Total explained deviance was 34%. Analyses of residual dissimilarity highlighted unexplained floristic similarity among many southern plots. A secondary residual axis highlighted unexplained variation in a broadly westerly to easterly direction. Consistent trends in the life form and geographical ranges of associated species suggest that these patterns are climate-related. LGM climatic anomalies were correlated with patterns on the second residual axis. Main conclusions: Current soil and climatic gradients are important drivers of floristic turnover in the SWAFR. However, residual analyses revealed broad-scale gradients in floristic composition unexplained by these variables. These appear to reflect imperfect modelling of the effects of current and perhaps historical climate on species turnover.