This paper investigates the extent of Amazonian forest and savanna fragmentation during the Last Glacial Maximum (LGM), challenging the long-standing refugia hypothesis. By bias-correcting vegetation distributions from a Dynamic Vegetation Model (DVM) with paleoclimate simulations, the study reveals fragmented forests with extensive humid forests to the west and forest islands in central/southern Amazonia. Drier ecosystems may have merged into continuous savanna, but reconstructions suggest taller, dense woodland/tropical savanna vegetation connecting disparate forest fragments. This ecotonal biome may have acted as a corridor for generalist species, while simultaneously serving as a barrier for specialists, potentially driving diversification.

Douglas I. Kelley, Hiromitsu Sato, Michaela Ecker, Chantelle A. Burton, João M. G. Capurucho, John Bates