On the detectability of latitudinal biodiversity gradients in deep time

Abstract

The latitudinal biodiversity gradient (LBG), in which species richness increases from the poles to tropical regions, is one of the most pervasive biodiversity patterns today. However, deep-time studies suggest that the LBG has varied in the geological past, with a range of taxonomic groups characterised by flattened or even bimodal gradients. Moreover, these studies suggest that tropical peaks and poleward declines in biodiversity are restricted to intervals of the Palaeozoic, and the last 30 million years (Myr), when cool icehouse climatic regimes persisted. Yet, the reconstruction of macroecological patterns in deep time is hampered by inherent geological and anthropogenic biases. In particular, spatial sampling heterogeneity has the potential to hinder the reconstruction of LBGs due to the ubiquitous scaling of species richness with area. In this thesis, a series of case studies that attempt to quantify the impact of spatial sampling heterogeneity on the reconstruction of LBGs are presented. Earth System and ecological niche modelling are applied to test whether observed biodiversity trends are the result of spatial sampling heterogeneity, or a genuine biological signal. In addition, a novel subsampling protocol is implemented to provide sampling-standardised estimates of biodiversity. Collectively, this work suggests spatial sampling heterogeneity often prevents the recovery of genuine LBGs in deep time. Estimates of zooxanthellate coral richness over the past 250 Myr demonstrate that the modern LBG got markedly steeper during the last 20 Myr, and a unimodal-type LBG likely persisted during the Early Cretaceous, coinciding with a geologically long-lived ‘cold-snap’. These findings are supported by ecological niche modelling, which suggest a tropical increase, and temperate decline in suitable habitat area during these intervals. Overall, these studies highlight the significance of correcting for spatial sampling heterogeneity when reconstructing biodiversity patterns from the fossil record, as well as the value of inferential methods in understanding past macroecological patterns.