Background/Question/Methods Ecological forecasting using future climate models predict dramatic change from ecosystem functioning, community distributions and increased rates of extinction.  These predictions have raised the issue of how ecologists might manage systems to adapt to climate change and minimize biodiversity loss. Forecasting models predict that narrowly distributed species, in particular are exceedingly vulnerable to climate change if their distributions are climatically constrained and the species are dispersal limited. These two assumptions, however, loom large in predictions of extinction risk associated with climate change. One of the proposed management responses to climate change driven extinction risk is to relocate species into climatically favorable environments: assisted migration. Prior to embarking upon what would clearly be a risky management strategy that is likely to create, as well as solve, problems, it is incumbent upon ecologists to dedicate substantial effort to critically evaluating the assumptions of extinction risk models. Existing ecological studies point to several issues that need to be addressed in order to move forecasting models from general descriptions of threat to prescriptions of action. I use the ecological literature to examine constraints on our likely ecological knowledge with which we will be making these biodiversity management decisions.

Results/Conclusions At least three different types of problems constrain models estimating climate change driven extinction risk. First, the distinction between a species fundamental and realized niche is a chronic problem for distribution modeling. Species distribution models must assume that the actual distribution reflects environmental constraints on potential distribution, and this is simply not likely the case for many narrowly distributed, and hence, vulnerable species. Biotic interactions and dispersal limitation can play a strong role in species distributions and are rarely incorporated into species models. Cases where there is a clear physiological range limit mechanism remain the exception, not the norm. Second, the mechanics of species distribution modeling remains problematic. Broadly distributed species often model poorly because of habitat generality. In contrast, narrowly distributed species may return high model fit, but often suffer from a lack of power to adequately capture the entirety of a species' niche. Finally, climate change is occurring against a backdrop of numerous other environmental changes that may impact species responses, and within the context of natural selection and the capacity for species to alter their relationship to climate. Emerging studies have demonstrated remarkable capacity for rapid response and unpredictable synergies amongst covarying environmental change.