Adapting nature conservation to climate change: the importance of microclimate

Over the last 15 years bioclimate models have been widely used to predict ecological responses to climate change. Results from these models suggest catastrophic consequences for life on earth, but to date, climate change has been implicated as a major cause of the extinction of just a few species. Mounting evidence from palaeoecology provides a compelling explanation for this discrepancy. Many species survived periods of rapid climate change in microrefugia: locations with suitable microclimate that would be missed entirely by coarse-scale bioclimate models. However, research in this emerging field has been hampered by a limited ability to model climate change at a sufficiently fine resolution. Here I present models that can be applied to provide accurate fine-grained, multidecadal estimates of climate change based on the underlying physical processes that influence microclimate. I apply the models to project historic distributions of 321 plant species forward to the present day and compare the results to those obtained using coarse-scale climate data. I show that the results of bioclimate models are fundamentally linked to the grid cell resolution of the data used to drive them. Models that rely on coarse-resolution spatial associations between species distributions and climate variables predict major range shifts, whereas fine-scale models predict localised patterns of change that more closely match observed distribution shifts. The results suggest that, while the predicted proportional change in climatically suitable habitat is not strongly scale-dependent, the prediction from most existing bioclimate models: that the trailing edge of species will retract rapidly over large distances, is likely to be false. This is further supported by a comprehensive multi-taxa meta-analysis of observed range shifts. While range expansions are widely documented in the literature, evidence for range retractions is, at best, equivocal. In consequence, the redesign of protected area networks to accommodate large-scale range shifts may be less effective than focusing on protecting populations of species within their existing geographic range.peerReviewe