Temperature reconstructions of euphotic oceans via coccolith clumped isotopes

Clumped isotope thermometry applied to carbonate fossils is a promising technique to derive independent and accurate reconstructions of absolute ocean temperatures, a key parameter in understanding past Earth Climate Sensitivity. Other more commonly used temperature proxies have several disadvantages, including requiring assumptions of seawater chemistry compositions (e.g. foraminifera Mg/Ca and δ¹⁸O), or being based on empirical correlations without a complete understanding of its controlling mechanisms (e.g. TEX₈₆ and U^k'₃₇). Conversely, clumped isotope thermometry is based on thermodynamics, and is independent from seawater chemistry. Here we present clumped isotopes (Δ₄₇) in coccolith separations from globally distributed Holocene core tops, a monospecific Coccolithus pelagicus sediment trap in the Iceland Sea, downcore sediments from the North Atlantic during the last 16 Ma, and downcore sediments from tropical (Equatorial Pacific) and high latitudes (South Tasman Rise) spanning the Cenozoic. 

Calcification temperatures of the sediment trap agree with satellite derived temperatures, further supporting a lack of or small vital effects in coccolith clumped isotopes. Temperatures derived from Δ₄₇ of tropical Holocene coccoliths are colder than modern Sea Surface Temperatures (SSTs). This suggests that coccolithophores may inhabit deeper than surface waters in these areas, which if proven to be true, would have implications for how other proxies, such as U^k'₃₇, are calibrated to SSTs. At higher latitudes, calcification temperatures from Holocene coccolith separations are more similar to SSTs, and we suggest they are indicators of mixed layer depth temperatures in these regions.

Pure coccoliths from the North Atlantic during the last 16 Ma show Δ₄₇-derived temperatures that are 10 °C colder than those derived with alkenones from the same samples. This suggests a modest, rather than an extreme polar amplification, which agrees better with climate models. Scanning Electron Microscopy (SEM) and trace elements show no evidence of significant recrystalization and therefore cannot explain such large differences in reconstructed temperatures with both proxies.

Preliminary low resolution Δ₄₇ calcification temperatures of pure coccolith separations from the Equatorial Pacific throughout the Cenozoic show similar trends to the overall climate pattern expected from foraminiferal δ¹⁸O, but with colder absolute values. For example, published core top Δ₄₇ coccoliths indicate warmer temperatures compared to our 2 My sample in core U1338, and may suggest potential early recrystalization effects, different sources or strength of upwelling in the past oceans, latitudinal movement of upwelling, or depth of production. Conversely, high latitude temperatures (ODP 1170) from our youngest coccolith separation (2 My) agrees better with modern SSTs and alkenone temperatures. The general expected climatic trend is also observed in our high latitude record, although the magnitude of cooling through time is less marked compared to that shown in the Equatorial Pacific. Trace element and SEM imaging could give insights on whether there is evidence of some recrystalization, or other interfering material in the analyzed pure coccolith fractions, despite the careful separation process that limited the presence of non-coccolith carbonate. Our results show that coccolith Δ₄₇ has the promising potential to derive reconstructions of temperatures of euphotic oceans over the Cenozoic.