Evidence for seasonality in early Eocene high latitude sea-surface temperatures

Specific challenges still exist in our understanding of past greenhouse climate states. Whilst climate model simulations using atmospheric CO2 concentrations consistent with proxy estimates broadly align with lower latitude proxy temperature estimates, they struggle to reproduce the warming implied by proxies at higher latitudes, especially in the marine realm. This inconsistency has often led to the conclusion that climate models are insufficiently sensitive. Here, we analyse the distribution of photozoan and heterozoan carbonates, which provide important constraints for latitudinal sea surface temperature (SST) gradients, to assess data/model mismatches for the early Eocene Climatic Optimum. The carbonate facies distribution is compared against quantitative geochemical proxy temperature estimates (, Mg/Ca, clumped isotopes and TEX86) and a new HadCM3L climate simulation. Good correspondence exists between the simulated cold-month SSTs and photozoan carbonates, indicating HadCM3L is effectively reconstructing meridional temperature gradients into mid-latitudes. Whilst there is good agreement between simulated mean annual SSTs and geochemical proxy estimates in low latitudes, the , Mg/Ca and TEX86 estimates instead align with warm-month SSTs at higher latitudes. In light of the carbonate facies evidence, and consistency between our simulation and available terrestrial proxy temperature estimates, this study supports previous claims that a warm season bias exists in many middle and high latitude SST estimates. This helps resolve the discrepancy between climate simulations and marine proxies and shows that climate models and data might be more closely aligned than is appreciated. Further, we demonstrate that simple, and widely available, proxies can play a fundamental role in contextualising wider paleoclimate uncertainties.