Résumé
1. Climate change has already caused distribution shifts in many species, and climate predictions strongly suggest that these will accelerate in the future. Obtaining reliable predictions of species range shifts under climate change is thus currently one of the most crucial challenges for both ecologists and stakeholders.
2. Here we simulate the distributions of 16 North American tree species at a continental scale for the 21st century according to two IPCC storylines, using a process-based species distribution model that for the first time allows identification of the possible causes of distribution change.
3. Our projections show local extinctions in the south of species ranges (21% of the present distribution, on average), and colonizations of new habitats in the north, though these are limited by dispersal ability for most species. Areas undergoing local extinctions are slightly larger under climate scenario A2 (+3.2 degrees C, +22% on average) than B2 (+1.0 degrees C, +19% on average). This small difference is caused by nonlinear responses of processes (leaves and flowers phenological processes in particular) to temperature. We also show that local extinction may proceed at a slower rate than forecasted so far.
4. Although predicted distribution shifts are very species-specific, we show that the loss of habitats southward will be mostly due to increased drought mortality and decreased reproductive success, while northward colonizations will be primarily promoted by increased probability of fruit ripening and flower frost survival.
5. Synthesis. Our results show that different species will not face the same risks due to climate change, because their responses to climate differ as well as their dispersal rate. Focusing on processes, our study therefore tempers the alarming conclusions of widely used niche-based models about biodiversity loss, mainly because our predictions take into account the local adaptation and trait plasticity to climate of the species.