How cold were the tropics and subtropics at the Last Glacial Maximum?

The spatial pattern and magnitude of tropical and subtropical sea surface temperatures (SSTs) at the Last Glacial Maximum (LGM) have been a major challenge for many years. The range of the LGM tropical and subtropical cooling remains a major source of uncertainties for our understanding of the LGM climate. The global SST reconstruction for the LGM is an important ocean surface boundary condition for LGM climate simulations. One widely used surface boundary condition for many previous LGM climate simulations is the Climate, Long-Range Investigation, Mapping and Prediction (CLIMAP) reconstruction, which is based on the distribution of species of various fossil groups. The CLIMAP LGM reconstruction shows a moderate cooling (1–2 1C) over most of the tropical Pacific and warming over the subtropical Pacific relative to the present. This reconstruction, however, is at odds with a variety of terrestrial proxies indicating much colder LGM conditions over the tropics. Climate models forced with the CLIMAP SST often simulate conditions over tropical land areas (Rind and Peteet, 1985) that are inconsistent with the LGM land data. Recent geochemical estimates of LGM tropical SSTs based on Mg/Ca ratios from planktonic foraminifera indicate much colder conditions relative to the CLIMAP reconstruction (Lea et al., 2000; Stott et al., 2002). The discrepancies between the CLIMAP SSTs and many other proxy estimates suggest that the CLIMAP SST reconstruction may have inherent warm biases. To address the controversy between the LGM terrestrial and oceanic cooling at low latitudes, one previous modeling study (Yin and Battisti, 2001) prescribed colder tropical SSTs relative to the CLIMAP reconstruction, with resulting simulated colder surface temperature over land that is in better agreement with the terrestrial evidence at the LGM. However, the prescribed cooling is arbitrary and the prescribed SST gradients are not based on realistic faunal and floral gradients. The article by Hostetler et al. (2006) in this issue of Quaternary Science Reviews represents a critical step toward resolving the long-term debate of the LGM tropical cooling. Hostetler, Pisias and Mix identify two major possible biases in the transfer functions for faunal-based SST reconstructions: low sensitivity in the transfer functions at regions with very warm SSTs, and underestimation