Bayesian multi-proxy reconstruction of early Eocene latitudinal temperature gradients

Abstract

Accurately reconstructing large-scale palaeoclimatic patterns from sparse local records is critical for understanding the evolution of Earth's climate. Particular challenges arise from the patchiness, uneven spatial distribution, and disparate nature of palaeoclimatic proxy records. Geochemical data typically provide temperature estimates via transfer functions derived from experiments. Similarly, transfer functions based on the climatic requirements of modern taxa exist for some fossil groups, such as pollen assemblages. In contrast, most ecological and lithological data (e.g. coral reefs and evaporites) only convey information on broad climatic requirements. Historically, most large-scale proxy-based reconstructions have used either geochemical or ecological data, but few studies have combined multiple proxy types into a single quantitative reconstruction. Large spatial gaps in existing proxy records have often been bridged by simple averaging, without taking into account the spatial distribution of samples, leading to biased temperature reconstructions. Here, we present a Bayesian hierarchical model to integrate ecological data with established geochemical proxies into a unified quantitative framework, bridging gaps in the latitudinal coverage of proxy data. We apply this approach to the early Eocene climatic optimum (EECO), the interval with the warmest sustained temperatures of the Cenozoic. Assuming the conservation of thermal tolerances of modern coral reefs and mangrove taxa, we establish broad sea surface temperature ranges for EECO coral reef and mangrove sites. We integrate these temperature estimates with the EECO geochemical shallow marine proxy record to model the latitudinal sea surface temperature gradient and global average temperatures of the EECO. Our results confirm the presence of a flattened latitudinal temperature gradient and unusually high polar temperatures during the EECO, which is supported by high-latitude ecological data. We show that integrating multiple types of proxy data, and adequate prior information, has the potential to enhance quantitative palaeoclimatic reconstructions, improving temperature estimates from datasets with limited spatial sampling.