Late Quaternary changes in Amazonian ecosystems and their implications for global carbon cycling

Abstract

The current role of Amazonia in the terrestrial carbon budget is the focus of intensive scientific interest, in large part due to its potential to accelerate global warming. However, its role in mediating CO2 changes over millennial time-scales since the last glacial maximum (LGM) has generally been overlooked and is the subject of speculation. Recent advances in our understanding of the Late Quaternary history of Amazonian ecosystems offers an opportunity to make more informed inferences about Late Quaternary changes in the magnitude of Amazon carbon storage than has hitherto been possible. Therefore, in this paper, we reconstruct changes in the magnitude of Amazon carbon storage over the last 21,000 years (since the LGM) by reference to recently published palaeohydrological and palaeoecological data and compare these data with results from simulations using a process-based terrestrial ecosystem model for the Mid-Holocene and the LGM. Building on these results further, we interpret changes in tropical forest biomass in the context of Late Quaternary polar ice-core records of atmospheric methane and carbon dioxide concentrations. Palaeo-data and model simulations show that Amazonia was predominantly forested at the LGM, although there is evidence for savanna expansion near the margins of the Basin and southern Amazonia may have been covered by deciduous/semi-deciduous dry forests rather than evergreen rain forests. We estimate Amazon C storage at the LGM to be only 135 Gt C (50% smaller than today), but find that its proportion of the entire terrestrial carbon store was almost twice that of today. The model shows that between the LGM and the Mid-Holocene there is a significant increase in evergreen broad-leaf forests at the expense of deciduous forests and a 67% increase in total Amazon C storage, attributable to rising temperatures and atmospheric CO2 levels. Although our results indicate that the Amazon Basin was dominated by rain forests throughout the Holocene, rain forest cover expanded in the Late Holocene (at the expense of savannas) and total Amazon carbon storage is simulated to have risen by 22% between the Mid-Holocene (225 Gt C) and the present day (Pre-Industrial) (225 Gt C). Comparison of these Amazon carbon fluxes with palaeo-data from other parts of the world suggests that, contrary to previous hypotheses, the terrestrial biosphere acted as a net carbon sink throughout the Holocene, and that the observed CO2 rise from 260 to 285 ppmv between 8 and 1 ka BP (revealed by the Antarctic Taylor Dome ice-core record) may have been driven by release of carbon from the oceans rather than land.