Accurate and Precise Bottom Water Paleotemperatures from Aragonitic Benthic Foraminiferal Li/Mg: Calibration, Theory, and Application

Abstract:

While great progress has been made in reconstructing past sea surface temperatures, reliable bottom water paleotemperature measurements are not routinely available. We suggest that Li/Mg ratios in biogenic aragonites, particularly in the cosmopolitan benthic foraminifer Hoeglundina elegans, have the potential to bridge this gap. Core top calibration shows that H. elegans Li/Mg decreases by 5.5% per °C (r² = 0.91), with a relationship that is nearly identical to that displayed by a wide range of corals (r² = 0.95). The fact that such disparate organisms behave so similarly suggests to us that thermodynamics are shining through the ‘vital effects’ that so often plague paleoceanographic proxies. We hypothesize that Ca²⁺ pumping causes Li/Ca and Mg/Ca ratios in the organisms’ calcification pools to decline, while Li/Mg remains constant. Rayleigh fractionation has the opposite effect on calcification pool Li/Ca and Mg/Ca (they rise), while Li/Mg still remains essentially constant. Hence any environmental influences on Ca²⁺ pumping and/or Rayleigh fractionation, such as seawater carbonate chemistry, have no measurable effects on aragonite Li/Mg.

Our first downcore test of the Li/Mg proxy is performed in core KNR166-2-26JPC from 546 m water depth in the Florida Straits. Benthic foraminiferal δ¹⁸O was previously used to document decreased seawater density during both Heinrich Stadial 1 (HS1) and the Younger Dryas (YD), consistent with flattening of isopycnals across the Florida Current caused by slowdown of the AMOC. Here we show striking agreement between H. elegans Li/Mg and ice-volume-corrected δ¹⁸O temperatures since ~17 ka (in both absolute values and temporal changes), confirming that bottom waters abruptly warmed during HS1 and the YD. The YD, which is better-resolved, was ~2°C warmer than the Holocene. Li/Mg indicates that Last Glacial Maximum bottom waters were ~2-3°C, or ~5°C colder than during the Holocene. If these glacial temperatures are accurate, they require a similar seawater δ¹⁸O to today, which suggests relative freshening in the face of higher mean ocean δ¹⁸O. Overall the Li/Mg paleotemperature reproducibility is very good, with a median five-depth running standard deviation of <0.6°C (even before omitting apparent outliers), a number which certainly includes real temporal variability.