The relative influence of moisture and temperature on range contractions and expansions of mammals over the past century

Background/Question/Methods

Species’ ranges are likely limited by many factors acting alone and jointly in a spatially heterogeneous manner.  The relative importance of range-limiting factors will be a function not only of the species’ niche and its orientation in environmental space, but also the spatial distribution of these factors in geographic space.  Here I examine the relative influences of temperature and moisture in limiting the ranges of 67 mammalian species in the western United States.  Specifically, I use species distribution models to estimate each species’ range in two time periods (1900-1939 and 1970-2009) using mean annual temperature and mean annual precipitation.  I then identify regions of range limitation by temperature or moisture by setting each factor to its optimal value given the observed value of the opposing factor.  This “optimized” range indicates the potential range of the species were the optimized factor not limiting.  Overlaying the optimized ranges with one another yields four zones of limitation: temperature prohibits, moisture prohibits, both prohibit, and both are within the species’ tolerance when considered alone but prohibitive in their combination (“interaction” limitation).  These limitation zones are then intersected with regions of range expansion and contraction to assess their relative importance.

Results/Conclusions

Averaged across species, the proportion of areas of range expansion and contraction were dominated by interaction limitation (mean±SE: 74±5% for expansions and 82±3% for contractions; P≤0.001 for both).  Limitation by either temperature or moisture alone occupied ≤17% of each region, and limitation by both factors jointly was negligible (<1% for both contractions and expansions).  The geographic importance of each type of limitation was uncorrelated with the significance of respective terms in distribution models of each species’ range, meaning that we must consider the geographic space over which a factor’s limiting values are dispersed to assess its importance to range limitation.  A null model analysis indicated that the predominance of interaction limitation was greater than would be expected by chance were zones of limitation randomly dispersed across the landscape.  Hence, species preferentially shift into/away from zones where the particular combination of factors is limiting, even though values of these factors are within the species’ tolerance.  These results suggest that when forecasting species’ responses to climate change, ecologists should consider changes in the correlational structure between potentially-limiting factors and should do so in a geographic context.