The equilibrium theory of island biogeography(1) is the basis for estimating extinction rates(2) and a pillar of conservation science(3,4). The default strategy for conserving biodiversity is the designation of nature reserves, treated as islands in an inhospitable sea of human activity(5). Despite the profound influence of islands on conservation theory and practice(3,4), their mainland analogues, forest fragments in human-dominated landscapes, consistently defy expected biodiversity patterns based on island biogeography theory(6-13). Countryside biogeography is an alternative framework, which recognizes that the fate of the world's wildlife will be decided largely by the hospitality of agricultural or countryside ecosystems(12,14-17). Here we directly test these biogeographic theories by comparing a Neotropical countryside ecosystem with a nearby island ecosystem, and show that each supports similar bat biodiversity in fundamentally different ways. The island ecosystem conforms to island biogeographic predictions of bat species loss, in which the water matrix is not habitat. In contrast, the countryside ecosystem has high species richness and evenness across forest reserves and smaller forest fragments. Relative to forest reserves and fragments, deforested countryside habitat supports a less species-rich, yet equally even, bat assemblage. Moreover, the bat assemblage associated with deforested habitat is compositionally novel because of predictable changes in abundances by many species using human-made habitat. Finally, we perform a global meta-analysis of bat biogeographic studies, spanning more than 700 species. It generalizes our findings, showing that separate biogeographic theories for countryside and island ecosystems are necessary. A theory of countryside biogeography is essential to conservation strategy in the agricultural ecosystems that comprise roughly half of the global land surface and are likely to increase even further(14).