Large lakes moderate climate-change effects in boreal North America

Abstract The interior biomes of North America are generally projected to experience rapid climatic change, especially with respect to drought indicators like potential evapotranspiration (PET). Focusing on the lake-rich and topographically varied North American boreal biome, we compared gradient-based climate velocity metrics for PET derived from the statistically downscaled ClimateNA product with dynamically downscaled metrics based on the Canadian regional climate model (CRCM5), which includes a 1-D freshwater lake model. We also developed regression tree models to examine the effects of land cover, topography, and geography on these differences. Across a range of global climate models (GCM), we found consistent, large differences in PET velocity–over 100 km/yr in some areas– between CRCM5 and ClimateNA. Within the boreal biome, we found that differences in the spatial gradient of change (mm/km), and hence in the gradient velocity metric (temporal gradient/spatial gradient) of annual PET, were largely explained by the percentage of lake coverage, especially at a broad scale (110-km × 110-km). Elevation effects were not detected. We found that the simple gradient velocity metric was remarkably robust to differences among GCMs in future PET projections, due to the high relative importance of spatial variation in temperature (i.e., spatial gradient), which often exceeded the magnitude of projected future change (i.e., temporal gradient). An important implication of our analysis is that regions surrounding large lakes are likely to be somewhat buffered from the full effects of climate change, serving as potential refugia for boreal ecosystems at a broad spatial scale.