UC Berkeley

The fossil record provides our only direct window into the long-term dynamics of biodiversity, but extracting biological signal from this record is fundamentally challenging. For example, mass extinction events often involve cascades of co-occurring environmental perturbations and reconstructions of these events involve observed patterns of extinction selectivity that are filtered through an incomplete and biased rock record. Simultaneously, researchers often compare multiple lines of proxy data in order to determine specific environmental change scenarios consistent with these patterns. In reality, however, the information about past life is underdetermined and many possible explanations can be given for any such reconstruction. In this dissertation, I developed and applied quantitative approaches to address two complementary aspects of this challenge: (1) identifying which environmental drivers shape selective extinction during mass extinction events, and (2) exploring how basin lifespans could alter the fossil record of today for future paleontologists and how those outcomes shape our understanding of past biodiversity.In the first chapter, I introduce an inverse modeling framework that couples Earth System Models (cGENIE) with Ecological Niche Models (Maxent) to identify hypothetical global change scenarios most consistent with observed paleogeographic occurrence and extinction patterns, without imposing a priori proxy constraints. Applied to the Late Ordovician Mass Extinction using the well-documented graptolite record, the framework yielded strong support for cooling, rather than warming, as the direction of environmental change consistent with observed patterns of extinction. In the second chapter, I extend the framework to handle multiple co-varying environmental predictors and validate it against the well-constrained Eocene–Oligocene transition (EOT). Sea surface temperature emerged as the dominant control on planktonic foraminifera extinction, while dissolved marine phosphate acted as an informative secondary predictor. Once further constrained by proxy CO2 estimates, scenarios coupling cooling with increasing nutrient availability best match the fossil record, consistent with independent evidence for enhanced Southern Ocean upwelling. Together, these chapters demonstrate that a framework built on the observed geographic distributions of species and a wide range of potential environmental change scenarios can be informative about past extinction events.In the third chapter, I shift focus from the drivers of extinction to the geologic filtering of biodiversity patterns and gradients. By simulating 10,000 potential future fossil and rock records based on the modern distribution of terrestrial tetrapods and currently active sedimentary basins, I showed that basin distribution and longevity exert some control on the recoverability of both overall diversity and the latitudinal diversity gradient (LDG). Recovery potential declined steeply within the first 25 Myr from the present and the latitudinal peak of diversity shifted toward temperate latitudes, even as the relative abundances of higher taxa remained largely unchanged. These results suggest that long-term basin dynamics may fundamentally alter the preservation of spatial diversity patterns, and that observed gradients in the deep time fossil record may reflect geologic filtering as much as true ecological signal.This dissertation provides quantitative tools for understanding the fossil record filtered through a heterogeneous rock record in the deep time, with implications for reconstructing past biodiversity dynamics and for anticipating how present-day biotas may be recovered in the geologic record of the future.