Individualistic species responses invalidate simple physiological models of community dynamics under global environmental change

1. Most predictions of species distribution and abundance changes in response to global warming relate the individual requirements of a single isolated species to climate variables through some form of climate mapping. This method fails to account for the effects of species dispersal and species interactions, both of which may strongly affect distribution and abundance. 2. We therefore examined the effects of dispersal and species interactions on the distribution and abundance of three Drosophila species in a laboratory system that mimicked a latitudinal cline of 15 °C. We then investigated how species distribution and abundance in this system responded to simulated global warming. 3. Dispersal allowed populations to persist at non‐optimum temperatures, overriding physiologically imposed range limits. 4. Temperature determined the outcome of competition. In pairwise interactions, Drosophila subobscura eliminated D. melanogaster or D. simulans at low temperatures but was itself eliminated at high temperatures. 5. Competitive interactions changed abundance and range sizes thus shifting the position of species optima. These changes depended on both the number and the identity of the competing species. 6. Enemy–victim interactions altered range and abundance. Adding the parasitoid Leptopilina boulardi affected the host assemblage directly at high temperatures where the parasitoid was present, and indirectly (mediated by dispersal) at low temperatures where it was scarce or absent. Host species coexisted for longer at low temperatures in clines when parasitoids were present than when they were absent. 7. Simulated global warming produced complex, counter‐intuitive effects on distribution and abundance, including the reversal of species’ relative abundance at some temperatures. 8. Because dispersal and species interactions strongly influenced both range and abundance (sometimes in unexpected ways) current species distributions are no guide to what they might be under global climate change. Furthermore, since both these factors are missing from climate envelope models of range and abundance change, their predictions are, at best, incomplete.