Benthic foraminiferal δ18O in the ocean's temperature‐salinity‐density field: Constraints on Ice Age thermohaline circulation

Abstract

Benthic δ18O data from 95 core sites are used to infer possible temperature‐salinity (T‐S) fields of the Atlantic and Pacific oceans at the Last Glacial Maximum (LGM). A constraint of stable density stratification yields logically consistent scenarios for both T and S. The solutions are not unique but are useful as a thinking tool. Using GEOSECS data, we solve for the modem relationship between δ18Owater (δw) and salinity in the deep sea: δw (SMOW) = 1.529 * S ‐ 53.18. As a starting point, we assume that the slope of this equation applies to LGM conditions and predict δ18Ocalcite (δc) gradients in equilibrium with probable T‐S fields of LGM deep and bottom waters. Benthic foraminiferal δ18O data from the deep Pacific (2–4 km depth), and the bottom Atlantic (> 4 km depth), are 0.1–0.2‰ lower than from the deep Atlantic (2–4 km depth) at the LGM. If the modern δw‐S slope applies, Atlantic deep and bottom waters were more dense than Pacific deep waters. This assumption would imply bottom waters both fresher (ΔS >0.5) and colder (ΔT ∼3°C) than overlying deep waters, in conflict with other data, suggesting ice age deep water much colder than at present. It is also possible that the observed δc gradients are an artifact of laboratory intercalibration. If Atlantic deep and bottom water δc values were similar to deep Pacific values, this would be consistent with the hypothesis of a stronger southern ocean versus North Atlantic source for deep‐ocean ventilation at the LGM. Taking the observed gradients at face value, however, a solution could be that the LGM δw‐S slope in deep and bottom waters was higher than at present, conceivably because of a stronger contribution of salt to the deep ocean via more intense sea ice freezing. This would allow Pacific deep waters and Atlantic bottom waters to have a common source, again in the Antarctic. Both would be more dense than Atlantic deep waters, even though the deep waters were much colder than at present. To better constrain these inferences drawn from the spatial distribution of benthic δ18O, we must reduce scatter in the δ18O data with more high‐quality measurements in high sedimentation rate cores. This is especially true at bottom water sites. Also, we must intercalibrate mass spectrometers at different isotope laboratories more accurately, to insure our isotope data are compatible.