University of Massachusetts, United States

The Pleistocene tropical rainforest refuge hypothesis, which attempts to explain the immense diversity and species endemism of Amazonia, is intimately connected to the concept of glacial aridity within the Amazon Basin. However both of these are topics of much debate, mainly due to the paucity of reliable, uninterrupted, regionally-representative proxy records back through the last glacial maximum (LGM). Reconstructions are often highly-localised and based on qualitative indicators of change. However material collected from the Amazon Fan allows us to examine an average signal of the whole of the Amazon Basin within a single sedimentary sequence. Here, we present data collected from ODP Site 942 and reveal that the Amazon Basin was forested yet relatively dry during the last glacial stage. Analysis of carbon isotopes on terrestrial biomarkers reveal that there has been relatively little shift in the C3:C4 ratio of the vegetation over the last 40 ka. Quantitative reconstructions of effective moisture based upon oxygen isotope analysis of planktonic foraminifera, suggest a significant reduction in the Amazon River outflow during both the LGM and last glacial-interglacial transition. We provide further evidence for glacial aridity through a quantified reconstruction of the fire history of the Amazon Basin, where fire-specific biomarkers are of coincident increased abundance throughout the more arid periods. We thereby refute the suggestion of encroaching savannah vegetation during glacial times, yet support the concept of a drier, cooler glacial Amazon Basin.