The importance of landscape rearrangement on historical environmental niche modeling: The contribution of past hydrological landscapes to potential species’ distributions

Background/Question/Methods: Evolutionary mapping using environmental niche modeling (ENM) is largely restricted to terrestrial species that typically disperse across continuous land surfaces. However, the configurations of watersheds and lake levels have fluctuated dramatically since the last glacial maximum when large proglacial lake and river systems covered much North America. Spatiotemporal shifts in landscape hydrography result in changing migration corridors over time. Yet, to our knowledge, no one has incorporated these shifts into ENMs of the late Pleistocene. We sought to fill that gap in knowledge by developing ENMs that consider how terrestrial species’ potential distribution and migration are influenced by shifting landscape connectivity. We employed maximum entropy modeling (Maxent) to compare models of the historical distributions of 40 common terrestrial New England species 1) assuming a continuous potential migration surface and 2) using hydrographic reconstructions from the Wisconsin-Holocene transition as a template for potential species distributions during the last glacial maximum. We used species presence records from the Global Biodiversity Information Facility and WorldClim climate data from the last glacial maximum for our modeling effort.

Results/Conclusions: All of the models in this study reached area under the receiver operator characteristic curve values above 0.95. While neither type of ENM outperformed the other, the historical distribution maps derived from the continuous climate surfaces consistently predicted terrestrial species presence within the boundaries of proglacial lakes. Our results suggest that failure to incorporate the post-Pleistocene proglacial lake and river distributions into paleomapping efforts results in unrealistic historical species distributions.