A universal multi-trace element calibration for reconstructing sea surface temperatures from long-lived Porites corals: Removing 'vital-effects'

作者：D'Olivo, Juan P.^*; Sinclair, Daniel J.; Rankenburg, Kai; McCulloch, Malcolm T.

来源：Geochimica et Cosmochimica Acta, 2018, 239: 109-135.

DOI：10.1016/j.gca.2018.07.035

• Trace elements
• Coral
• Temperature calibration
• Multiproxy reconstruction

Trace element abundances in corals can potentially provide high-resolution seasonally resolved constraints on past sea-surface temperatures, much needed to improve our understanding of climate variability on interannual to centennial time scales. A major limitation to the general application of trace element (TE) paleo-thermometers to coral fossil records is the presence of 'vital effects', which result in the need for colony-specific temperature calibrations. Here we demonstrate that reliable proxy temperatures from massive Porites corals can be achieved by using a universal multi-trace element calibration scheme (UMTECS). Using modern massive Porites corals living in well characterized sea surface temperature (SST) environments we first confirm that Sr/Ca and Li/Mg ratios are the most robust SST proxies compared to other trace element ratios (Mg/Ca, U/Ca, B/Ca and Li/Ca). Importantly we find that the slopes (and intercepts) of the Sr/Ca vs SST relationship are linearly correlated with the slopes from the Li/Mg vs SST relationship and provide a simple mathematical explanation for this phenomenon based on the tendency for all coral data to cluster around a single common 'centroid'. Based on this tight empirical correlation between Sr/Ca and Li/Mg thermometers we show that a 'universal' calibration strategy can be applied that largely circumvents 'vital effects'. Using this approach, we show that accurate more reliable reconstructions of paleo-temperatures can be undertaken using fossil Porites corals without the need for a priori colony-specific temperature calibrations. The general viability of this approach is demonstrated using an independent set of coral data where we show that UMTECS outperforms other 'aggregating' calibration techniques. For P. lutea and P. lobata the scheme produces temperature estimates with root-mean-square errors (RMSE) that are only 0.3 degrees C larger than those using colony-specific calibrations. This compares with 'averaged TE-SST calibrations' which produce RMSEs between 0.4 and 0.6 degrees C larger than colony-specific calibrations for P. lutea and P. lobata. However, we note that the success of our approach diminishes where data for individual corals deviate markedly from the common centroid, which appear to be related to species differences.

• 出版日期2018-10-15