Planktic foraminifera shell chemistry response to seawater chemistry: Pliocene–Pleistocene seawater Mg/Ca, temperature and sea level change

• G. ruber calibrations of Mg/Casw–Mg/Catest and Mg/Ca–temperature at Mg/Casw = 3.4.
• The Mg/Catest–Mg/Casw–temperature relationship is nonlinear.
• We reconstruct Mg/Casw for the last 5 Ma and show Pliocene Mg/Casw was 17% lower.
• Pliocene surface and deep ocean temperature have been underestimated by 0.9–1.9 °C.
• Pliocene sea level may have been overestimated based on coupled Mg/Ca–δ18O.

Foraminifera Mg/Ca paleothermometry forms the basis of a substantial portion of ocean temperature reconstruction over the last 5 Ma. Furthermore, coupled Mg/Ca–oxygen isotope (δ18O) measurements of benthic foraminifera can constrain eustatic sea level (ESL) independent of paleo-shoreline derived approaches. However, this technique suffers from uncertainty regarding the secular variation of the Mg/Ca seawater ratio (Mg/Casw) on timescales of millions of years. Here we present coupled seawater–test Mg/Ca–temperature laboratory calibrations of Globigerinoides ruber in order to test the widely held assumptions that (1) seawater–test Mg/Ca co-vary linearly, and (2) the Mg/Ca–temperature sensitivity remains constant with changing Mg/Casw. We find a nonlinear Mg/Catest–Mg/Casw relationship and a lowering of the Mg/Ca–temperature sensitivity at lower than modern Mg/Casw from 9.0% °C−1 at Mg/Casw = 5.2 mol mol−1 to 7.5±0.9%°C−1 at 3.4 mol mol−1. Using our calibrations to more accurately calculate the offset between Mg/Ca and biomarker-derived paleotemperatures for four sites, we derive a Pliocene Mg/Casw ratio of ∼4.3 mol mol−1. This Mg/Casw implies Pliocene ocean temperature 0.9–1.9 °C higher than previously reported and, by extension, ESL ∼30 m lower compared to when one assumes that Pliocene Mg/Casw is the same as at present. Correcting existing benthic foraminifera datasets for Mg/Casw indicates that deep water source composition must have changed through time, therefore seawater oxygen isotope reconstructions relative to present day cannot be used to directly reconstruct Pliocene ESL.