How well can we predict future distributions? Using paleoecology to inform conservation models

Predicting where species must move as climate changes is an important conservation goal. SDMs generally underpredict range sizes because estimating full fundamental niches is challenging. I looked at how well SDMs predict past ranges. They vastly underestimate distributions. But when I zoom in to look Microtus californicus, I find that we underpredict its distribution because populations, and portions of its niche, have been lost. This has huge implications for endangered species, which are necessarily likely to lose populations going forward. If we overpredict their future distributions, we may conserve unsuitable lands, wasting resources and losing more populations.