Do changes in habitat predict observed changes in small mammals in Yosemite National Park?

Background/Question/Methods: Different species of small mammals showed contraction, expansion, or no change in their elevation distribution range over the last 80 years in the Yosemite National Park, California. Using  their specific differences, we questioned if changes in habitat could explain the observed changes in small mammals’ elevation range shifts, and if these may influence our ability to forecast species responses to habitat changes over time. To answer these questions we linked data on small mammal presence and abundance in historical and modern times (Grinnell survey in 1910-1930 and resurveys in 2003-2008) to maps of habitat suitability (using California Wildlife Habitat Relationships rankings), derived from historical and modern assessments of land cover (Wieslander survey 1930-1940 and CalVeg 2000-2006). Using MAXENT, we tested if species responses to habitat types are transferable across time, by creating a model with historical data and evaluating it on modern data. Model performance was assessed by estimating the difference in the Area Under the Curve (AUC) between the historical and the modern results. 

Results/Conclusions: We found that species that contracted elevation range from historical to modern times, stayed in similar to higher suitability habitat and at higher abundances. Species that expanded their elevation range, however, were also found in similar suitability habitat (except for two species found in higher suitability habitat) and exhibited similar abundances. We produced MAXENT models for modelable species, which were species with sufficient number of presences, and produced within-era reliable models i.e. high AUC (>0.75). Across-era models, resulted in an AUC difference that was significantly higher for species expanding elevation ranges than contracting or no changes species (F=6.92, p=0.0059, d.f. 20). AUC difference for expanding species ranged between 0.17 and 0.51, and AUC difference for contracting species ranged between 0.04 and 0.16, suggesting that forecasting species that contracted elevation range is more reliable than for species with expanding elevation ranges. These results suggest that habitat shifts may explain some of the observed shifts in elevation ranges, and that modeling success of species with different elevation range dynamics is dependent upon the direction of the shift.