A spatially predictive baseline for monitoring multivariate species occurrences and phylogenetic shifts in mediterranean southern Australia

Question Climate change is driving shifts in the composition of vegetation but the lack of controls and confounding spatial factors pose challenges for detecting the climate signal in observed changes through time and space. We tested whether climate can be isolated as a driver of spatial vegetation composition at the landscape scale in mediterranean southern Australia by considering landscape factors (e.g. soil gradients) and spatial structure (relative geographic isolation). The aim was to develop principles for selecting spatial analogues for climate change and provide a spatially predictive baseline for monitoring. Methods A landscape-scale monitoring transect spanning 550km was established. Whole community presence/absence of vascular plant species in plots was modelled as a multivariate response to environmental and spatial variables. Species and phylogenetic composition-environment relationships were also explored using indirect gradient approaches, partial correlations and distance decay models. Results A total of >2900 occurrences of >400 plant species were recorded. Relative vegetation composition was predicted by mean temperature and soil properties, such as electrical conductivity and texture. Spatial structure was critical, as decay in compositional similarity with geographic distance and spatial autocorrelation of nested plots were involved in turnover patterns. The rate of change in species composition with changes in temperature equated to complete species turnover within the habitats sampled. Conclusions The influence of climate on spatial variation in vegetation composition can be quantified, accounting for distance decay. Landscape gradients (particularly soil properties) tended to be orthogonal to climate and explained some turnover. Spatial analogues for climate change would need to be similar in soil properties and not too geographically distant. Composition resulting from more extreme climate change scenarios may have no spatial analogue due to the importance of neutral distance decay at larger spatial scales in determining compositional differences. We illustrate these principles with a sequence of warming, drying analogues. The spatial transect provides a framework for monitoring composition by directly incorporating temporal data and using spatial analysis to inform the expected direction of compositional shifts with climate change.