Abstract/Summary

Improving understanding of the long-term impacts of climate change on Amazonian ecosystems remains an important scientific challenge. Of particular concern is the impact of a drier climate, given that many climate models simulate decreased precipitation for Amazonia over the coming century. However, the impacts of such climate change on Amazonia are uncertain, with scenarios ranging from widespread rainforest dieback, to overall forest resilience. This thesis uses the middle Holocene (ca. 6,000 years ago) as an approximate analogue for future climate change, given the evidence that this period was significantly drier than present across much of Amazonia. The overall aim is to improve our understanding of the impact of long-term mid-Holocene drought on Amazonia using a multi-disciplinary approach, integrating both palaeoecology and Earth system modelling. Results are presented over three chapters in the form of academic papers. Chapter 2 presents a new multi-proxy synthesis of mid-Holocene palaeoecological records from across tropical South America, demonstrating concurrent mid-Holocene expansion of savannah in southern and eastern Amazonia and south-east Brazil, with resilience of rainforest in central Amazonia. Chapter 3 presents a new fossil-pollen record from a small palm swamp (Cuatro Vientos) in south-west Amazonia, which is compared with a previously published pollen record of a nearby large lake (Laguna Chaplin), demonstrating that palm swamps can yield useful Quaternary pollen archives recording the history of terrestrial vegetation beyond the swamp, rather than merely a history of localised swamp vegetation dynamics. Chapter 4 presents an ensemble of 21 mid- Holocene vegetation model simulations in one of the first palaeo-vegetation modelling-data comparison projects of mid-Holocene Amazonia, demonstrating that the models can replicate mid-Holocene rainforest resilience in the core of the Amazon basin, with savannah expansion restricted to the ecotonal regions of eastern and southern Amazonia. However, uncertainty exists between the models regarding the spatial position and extent of this savannah expansion.