Spatial modelling of species turnover identifies climate ecotones, climate change tipping points and vulnerable taxonomic groups

There is an expectation that climate change will drive turnover in the composition of ecological communities. Established methods for predicting the degree of turnover and spatial areas and taxonomic groups that will be most affected from real data are lacking. We tested a combination of spatial modelling tools to make these predictions. Using data from systematic vegetation survey plots from the Adelaide Geosyncline region, southern Australia, we modelled species turnover as a function of bioclimatic and geographic distances and predicted turnover using future climate change scenarios for 2030 and 2070. We conducted bioclimatic gradient analysis (CCA) on species composition data and mapped zones of higher turnover. The method for detecting these zones was tested using a simulation of continuous turnover. A phylogeny was generated for recorded species and correlations of occurrences of phylogenetic groups with species turnover were calculated. Significant turnover was predicted for the least severe climate change scenarios and near-complete species turnover for the most severe scenario. Gradient analysis revealed discrete transitional zones with more rapid turnover, which were interpreted as a mesic-arid ecotone. Turnover occurred at family level and with increasing temperature and decreasing rainfall there was a shift from the prevalence of Ericaceae, Myrtaceae, Haloragaceae, Cyperaceae, and Xanthorrhoeaceae to that of Amaranthaceae, Malvaceae, Scrophulariaceae, Sapindaceae, and Solanaceae. The mesic end of this climate gradient had relatively low rates of turnover and was interpreted as a refugium with a tipping point. The translation of spatial patterns to temporal change is dependent partly upon scales at which community assembly processes operate and predicts relative vulnerability, but not rates of change, which can only be measured through monitoring. The approach can be applied at any spatial or taxonomic scale subject to sufficient data resolution and can inform management decisions as to biases in climate change risks.