Climate refugia for Pinus spp. in the Madrean sky islands of México and the United States

Sandra L. Haire, Haire Laboratory for Landscape Ecology, Belfast, ME, Miguel L. Villarreal, Western Geographic Science Center, U.S. Geological Survey, Menlo Park, CA, Citlali Cortés Montaño, Independent Researcher, México City, Mexico, Aaron D. Flesch, School of Natural Resources and the Environment, University of Arizona, Tucson, AZ, José M. Iniguez, Rocky Mountain Research Station, USDA FS, Flagstaff, AZ, Jose Raul Romo-Leon, Departamento de Investigaciones Científicas y Tecnológicas, Universidad de Sonora, Hermosillo, SO, Mexico and Jamie S. Sanderlin, Rocky Mountain Research Station, U.S. Forest Service, Flagstaff, AZ

Background/Question/Methods
The ability of climate refugia to facilitate species persistence under adverse conditions of the past signals their potential importance in protecting biodiversity with future change. We investigated the potential for complex topography to support microrefugia for pines (Pinus spp.) in the Madrean sky islands, located in the borderlands of México and the United States. The Madrean sky islands are a stepping-stone archipelago comprised of montane habitats separated by lowland grassland and desert scrub. Pines comprise an integral part of the diverse forests and woodlands that underpin ecological functions across the region and are under extreme risk from habitat loss and increased isolation due to climate change. To address our objectives, we compiled a presence and absence database for five pine species: P. strobiformis, P. engelmannii and P. chihuahuana, P. arizonica and P. discolor. Species data were collected in two field studies conducted across the bi-national region. We developed species distribution models based on terrain characteristics using boosted regression trees. By adding bioclimatic predictors, we determined the range of variability in climate parameters associated with microrefugia. Last, we included remote sensing indices in the models to understand how seasonal changes affect the distribution of microrefugia across forests and woodlands.
Results/Conclusions
The best fit models included terrain and bioclimatic variables (AUC ranged from 0.765 for P. arizonica to 0.88 for P. discolor) with one exception: the best fit model for P. engelmannii combined terrain and seasonal greenness (AUC = 0.713). Elevation was an important predictor in best fit models (relative influence ranged from 16.43% for P. discolor to 54.29% for P. arizonica). Variability in terrain ruggedness, slope position and aspect contributed to defining microrefugia for pines within a preferred range of elevation that differed among species. Across the range of elevation and bioclimatic predictors, two types of species response were identified: thresholds in tolerance, marked by precipitous decline that indicated a more clearly defined habitat preference, in contrast with a continuous, gradual change in response indicating greater range in tolerance. Seasonal remote sensing indices were especially important for characterizing microrefugia for species associated with a more open canopy structure and where regular fires occur. Heterogeneity in topography associated with pine habitats holds potential for facilitating complex range shifts and buffering habitats from disturbance, including fire. Holdouts and stepping-stones provided by microrefugia will be critical to persistence and movement of pines in the Madrean sky islands.