Predicting range shifts for critically endangered plants : is habitat connectivity irrelevant or necessary?

Climate and land-cover change will directly impact future species distributions, leading to range expansions, contractions and local extinctions. However, assessments of future range shifts rarely account for the capacity of the landscape matrix to facilitate species dispersals. Here, we assessed future range shifts for a suite of critically endangered plants in Madagascar. We quantified habitat connectivity using a least cost path model that captured the potential of species to disperse within fragmented landscapes. Next, we constructed three scenarios representing landscapes impacted by climate-only, climate and land-cover change, as well as habitat connectivity. We modelled species distributions using a hierarchical Bayesian framework and measured future range shifts using three spatial indices: net-change, range distance and elevation change. Our results show that the median range shift due to contractions increased by 25% under the climate-only scenario compared with the connectivity scenario. Habitat connectivity is predicted to limit range shifts due to contractions, while increasing shifts due to expansions for many of the endangered and critically endangered plants on Madagascar. However, at least one-third of critically endangered and 50% of endangered plants are expected to experience range contractions and upslope displacement under all scenarios, suggesting that even with habitat connectivity the range of some species may still contract. Despite that finding, our study suggests that including connectivity in range shift models is crucial for developing a relevant connectivity conservation plan, since future climate or climate and land-cover change models do not adequately represent species' potential to reach safe sites