Uneven spatial sampling distorts reconstructions of Phanerozoic seawater temperature

Abstract Paleotemperature proxy records are widely used to reconstruct the global climate throughout Phanerozoic and test macroevolutionary hypotheses. However, spatial distribution of these varies through time. This is problematic because heat unevenly distributed across Earth's surface. Consequently, heterogeneous sampling data has potential bias reconstructed temperature curves. We evaluated spatiotemporal evolution using a compilation δ18O data. tested influence variable coverage on estimates paleotemperature by steep “modern-type” latitudinal gradient flattened “Eocene-type” gradient, based samples. show that overestimated in ∼70% stages. Perceived climatic trends for some intervals might be artifactually induced shifts paleolatitudinal sampling, with equatorward concurring warming trends, poleward cooling trends. Yet, magnitude perturbations also underestimated. For example, observed Ordovician trend may underestimated due an shift sampling. Our findings suggest while vital reconstructing deep time, consideration nature crucial improving reconstructions.