The two biggest environmental problems the world faces today are global warming and habitat fragmentation. How will the intersection of these two impacts change ecosystems on Earth? A powerful perspective on that question emerges from setting the ecological changes of the past fifty years in the context of what is ‘natural' as interpreted from the fossil record, then using that information to forecast ecological fates in the coming century. Here we focus on defining the long-term baseline for species richness in mammals by tracking species-area curves through a variety of temporal and spatial scales in the USA. To build the curves, we use distributional data of modern mammals as recorded in field guides and museum collections and distributional data of fossil occurrences as recorded in the FAUNMAP and MIOMAP databases. We make the fossil and modern curves comparable by subsampling the modern data in a way to simulate fossil occurrences, then use the statistically adjusted counts of species per locality and area to construct species-area curves for a number of time-slices ranging in age from modern, to Holocene (back to 10 Ka), to Pleistocene (10 Ka to 1.8 Ma), and eventually back to the late Oligocene (as old as 30 Ma). Our preliminary results indicate that from late Pleistocene to present, species richness of small mammals declined uniformly across biogeographic regions from Pennsylvania to Texas. In the central Great Plains, modern habitat islands may retain characteristic late Holocene species richness of small mammals. These conclusions warrant further testing.