• benthic foraminifera;
  • Mg/Ca paleothermometry;
  • carbonate chemistry;
  • South Atlantic;
  • laser ablation

[1] We used modern epibenthic foraminifer tests of Cibicidoides mundulus and Planulina wuellerstorfi from South Atlantic core top sediments in order to establish Mg/Ca-temperature relationships for the temperature range from 0 to 15°C. We obtained the following calibrations: Mg/Ca (mmol/mol) = 0.830*exp(0.145*BWT (°C)) for P. wuellerstorfi, and Mg/Ca (mmol/mol) = 0.627*exp(0.143*BWT (°C)) for C. mundulus. However, a number of tests, especially those bathed in North Atlantic Deep Water, revealed higher Mg/Ca ratios than predicted from the calibration. Our data suggest that Δ[CO32−] of bottom water exerts a significant control on ΔMg/Ca (temperature-corrected) of C. mundulus (ΔMg/Ca = 0.017*Δ[CO32−] − 0.14), while ΔMg/Ca of P. wuellerstorfi is more likely to be governed by TCO2 (ΔMg/Ca = −0.007*TCO2 + 15). Since both Δ[CO32−] and TCO2 are closely linked to [CO32−], it is inferred that carbonate ion acts as secondary control, after temperature, on benthic shell Mg/Ca below ∼4°C. A drop in [CO32−] by 25 μmol/kg at 4 km water depth, as suggested for the Last Glacial Maximum, would decrease Mg/Ca by up to 0.4 mmol/mol, which leads to an underestimation of bottom water temperature by ∼3.5°C. Therefore our results indicate that the Mg/Ca thermometer should be used cautiously for benthic foraminifers where changes in the carbonate chemistry are present in the paleoceanographic record.