Species distribution models are currently in wide use among ecologists for projecting how species may respond to projected future climates with the intent, at least in some cases, to advise management on how to respond to the threat of climate change. Species distribution models may identify areas of future risk and security within a current range, as well as identify potential future habitat. Species distribution models, however, carry a large number of stringent assumptions. I examine assumptions in the light of a common suite of forest management actions in order to focus how ecologists might best frame tree distribution modeling to assist in managing for tree diversity under changing climates. Modeling issues examined include distinguishing fundamental and realized niches, the importance of biotic interactions, including disturbance; dispersal limitation; and adaptive capacity. Focal management actions I consider are fire management and planting, including planting species outside their current distributions (managed relocation). The focal study region that I assess is the Sierra Nevada range of California, where we have used an integrated team of scientists to assess forest vulnerability to climate change and fire.
Results/Conclusions
The assumptions of species distribution models virtually all have the potential to underestimate the potential of species to persist within their current distributions. Focusing on trees, the model assumption that a current occurrence reflects the capacity to recruit under the current climate is problematic in long-lived species such as trees. Differential climatic tolerances of seedlings versus adults, again may constraint the accuracy of distribution models to predict future outcomes. Species distribution models are likely to be considerably more robust in predicting potential future suitable habitat than they are at predicting range losses as a consequence of emerging climatic unsuitability. Further, biotic interactions can be a driver of population success, but modeling biotic interactions remains a challenge. Given the long life span of trees, management responses to habitat losses may manifest soonest in disturbance driven systems where fire can drive site type change. Other changes may be gradual and slow. In contrast, recruitment into new locations may be limited by dispersal. As a consequence, focusing modeling efforts on where, when and why range expansions should be fostered in trees is likely the best use of distribution modeling for climate change in trees.