Try Everything: Coupling Phytolith and Isotope Records to Reconstruct South American Landscapes During Global Warming and Cooling Events of the Cenozoic

Abstract

The Intergovernmental Panel on Climate Change (IPCC, 2024) predicts that, regardless of policy changes, atmospheric CO₂ levels, and consequently, global temperatures, will rise over the next century to levels unprecedented in human history. However, these levels are not unprecedented in Earth's geologic past. Over the Cenozoic (the last 66 million years), global climate has undergone repeated warming and cooling episodes driven by fluctuations in atmospheric CO₂, superimposed on a long-term trend of cooling and declining pCO₂. To better predict how future climate change may affect global ecosystems, we can turn to the fossil record to examine how climate and biospheres responded to similar variability in the past.In this dissertation, I combine phytolith (opaline silica particles deposited in or around plant cells) and isotope records to reconstruct landscape evolution in South America during periods of global climate warming and cooling in the early and late Cenozoic. I also conduct a modern reference study to test and compare the spatial resolution of phytolith assemblages and isotope signatures across a spatially complex landscape.In Chapter 1, I explore the evolutionary history of lowland vegetation in southern South America during the Paleocene and Eocene using phytolith records. Previous reconstructions, including some paleobotanical data, proposed that grass-dominated ecosystems may have existed in the region at this time, contradicting global patterns of grassland evolution and other fossil evidence. I present a basin-wide phytolith analysis of Paleocene and Eocene terrestrial deposits from the San Jorge Basin, Argentine Patagonia. This expands on earlier phytolith studies and integrates new radiometric dates to provide a temporally resolved reconstruction of early Cenozoic vegetation. My results show that forests dominated lowland ecosystems from the Paleocene through the middle Eocene. Palm abundance increased from the middle to late Eocene as these forests began transitioning to cooler, drier, and more open vegetation types. These findings are supported by paleobotanical, geochemical, and faunal records. Grasses remained rare and were likely restricted to forest understories until at least the Early or Middle Miocene, challenging hypotheses that propose extensive early Cenozoic grassy habitats in South America. In Chapter 2, I test the spatial resolution of phytolith assemblages across a heterogeneous landscape and compare them to the carbon isotope composition of the soil organic matter from which they were extracted. This study addresses a key uncertainty in fossil phytolith interpretation: to what extent do phytoliths reflect local versus regional vegetation? This is especially important for fossil reconstructions, where sample coverage is often sparse. Can we detect vegetation heterogeneity from just a few assemblages? How do phytolith signals compare to carbon isotopes, a widely used proxy for vegetation structure? To explore this, I analyzed samples from three transects in the piedmont of the Eastern Colombian Andes, a region that includes grasslands, forests, and palm swamps, each sharply demarcated. My results show that phytolith assemblages reflect both the vegetation at the collection site and surrounding plant communities. Phytoliths and carbon isotopes capture overlapping but distinct aspects of vegetation structure and are most powerful when used together. Additionally, phytoliths can offer insight into fire regimes, potentially indicating where and what kinds of vegetation were burning. These results underscore the value of phytoliths for reconstructing vegetation heterogeneity and fire history in both modern and ancient ecosystems. In Chapter 3, I focus on a cooling period that followed the last major Cenozoic warming event, the Miocene Climatic Optimum (17 to 15 million years ago). This subsequent cooling, is recorded in the fossil-rich deposits of La Venta, a semi-arid region in Colombia. Despite its significance as one of the most important vertebrate fossil sites in the Neotropics, little is known about the vegetation that supported its fauna or how tropical climate changed during this time. To address these gaps, I use phytoliths and clumped isotope paleothermometry to reconstruct vegetation and climate between 14 and 7 million years ago. The results suggest that the region remained forested throughout the interval and that vegetation density increased as local temperatures cooled and seasonal rainfall may have intensified, creating more humid conditions overall. These findings contrast sharply with climate models that predict drying across the tropics during the Miocene Climatic Transition and provide the first terrestrial temperature record for tropical South America during this time. This discrepancy highlights the importance of regional proxy records in validating and improving global climate models.