Journal article

Understanding past changes in sea surface temperatures (SSTs) is crucial; however, existing proxies for reconstructing past SSTs are hindered by unknown ancient seawater composition (foraminiferal Mg/Ca and δ18O), or reflect subsurface temperatures (TEX86) or have a limited applicable temperature range (Uk'37). We examine clumped isotope (Δ47) thermometry to fossil coccolith-rich material as an SST proxy, as clumped isotopes are independent of original seawater composition and applicable to a wide temperature range and coccolithophores are widespread and dissolution resistant. The Δ47-derived temperatures from <63, <20, <10 and 2-5 μm size fractions of two equatorial Pacific late Miocene-early Pliocene sediment samples (c1; c2) range between ∼18-29°C, with c1 temperatures consistently above c2. Removing the >63 μm fraction removes most non-mixed layer components; however, the Δ47-derived temperatures display an unexpected slight decreasing trend with decreasing size fraction. This unexpected trend could partly arise because larger coccoliths (5-12 μm) are removed during the size fraction separation process. The c1 and <63 μm c2 Δ47-derived temperatures are comparable to concurrent Uk'37 SSTs. The <20, <10 and 2-5 μm c2 Δ47-derived temperatures are consistently cooler than expected. The Δ47-Uk'37 temperature offset is probably caused by abiotic/diagenetic calcite present in the c2 2-5 μm fraction (∼53% by area), which potentially precipitated at bottom water temperatures of ∼6°C . Our results indicate that clumped isotopes on coccolith-rich sediment fractions have potential as an SST proxy, particularly in tropical regions, providing that careful investigation of the appropriate size fraction for the region and timescale is undertaken.