Discussion

Our bioclimatic models predict the occurrence of species and attempt to identify suitable habitats. As shown, these models can also be updated with predicted future climates to project the redistribution of species’ contemporary realized climatic niche. The validity of these projections are dependent on the accuracy of the bioclimatic models as well as how closely the predicted future climate using the IS92a scenario from the Canadian and Hadley GCMs matches the actual climate of the future.

The climate variables identified as important for predicting species occurrence varied between the species and subspecies analyzed. These differences suggest that they will respond uniquely to climate change. Despite these differences, it appears that their small geographic distributions predispose them to a shared threat of more rapid onset of climatic disequilibrium in comparison with species occupying large-scale distributions. Analyses of species occupying larger scale distributions tend to report greater areas of unaffected distribution and show less separation between existing distributions and the geographic occurrence of suitable climate (Bakkenes and others 2002, Iverson and Prasad 1998, Rehfeldt and others 2006). The extent of disequilibrium predicted by these studies may, however, be an underestimate due to the potential for maladaptation to climate within species (Rehfeldt 2004, Rehfeldt and others 1999, Rice and Emery 2003).

As natural systems attempt to regain equilibrium with the novel distribution of climates, distributions will shift. Redistribution rates, however, will be influenced by genetic structure, autecology, life history, reproductive capabilities, and ecophysiology (see, for instance, Ackerly 2003). Consequently, all of these factors should be taken into consideration in interpreting projections from bioclimatic models. The physiology of subalpine larch, for instance, appears to be consistent with the climatic profile described by our models. Its realized niche seems dependent on a superior hardiness and resistance to winter desiccation (Burns and Honkala 1990). Subalpine larch, however, does not compete well with other evergreen species (Arno and Hammerly 1984). Hence, the continued warming trend will likely eliminate subalpine larches from the Western United States through competitive exclusion. Common garden studies with smooth Arizona cypress and Piute cypress by Rehfeldt (1999) revealed genetic variability equivalent to that conveyed by broadly distributed conifer species. This indicates populations within the species may only be adapted to a portion of its present or future realized climate space. In addition, both smooth Arizona cypress and Piute cypress are fire dependent. Fire management practices may have had a substantial influence on limiting their distribution (Marshall 1963). Consequently, for these subspecies of Cupressus arizonica, our estimate of the realized climatic niche space may underestimate the breadth of climatically suitable area. Finally, Macfarlane’s small distribution is attributed to reduced compatibility between its floral biology and pollinators (Barnes 1996). The rates that the range may expand would depend, therefore, on the presence of compatible pollinators.

Bioclimatic models represent tools that can be used to assist decision makers in managing threats associated with global warming. By identifying and modeling those climates in which species occur, the models can be used to predict the location of sites that should be climatically suitable for a species. These predictions can be used by managers to assist the natural processes by transferring species to the future location of their climatic optima. To be sure, additional modeling is needed that integrates a general understanding of climate, climate change, and plant-climate relationships (see Crookston and others 2007). Interpretation of results by resource managers also requires integration of additional layers of information such as land use (see Hannah 2006).