Re-evaluating modern and Palaeogene GDGT distributions: Implications for SST reconstructions

In this paper, we review the TEX86 palaeothermometer for sea surface temperature (SST) and evaluate its application to the Palaeogene, with a focus on the principal ecological, physical or chemical processes that can bias glycerol dialkyl glycerol tetraether (GDGT) distributions. Recent investigations of Palaeogene sediments have revealed temperature offsets between two different GDGT-based approaches, TEX86L and TEX86H, with the former agreeing with SST estimates derived from inorganic proxies (Hollis et al., 2012). These are surprising observations because the two GDGT approaches, although based on two distinct groups of compounds, apparently agree at SSTs>15°C in modern oceans. Here we reassess the relationship between raw GDGT distributions and the ratios used to construct TEX86H and TEX86 in both the modern core-top dataset and a new compilation of Palaeogene data. We show that the offset between TEX86H and TEX86L (ΔH–L) is a function of the GDGT-2/GDGT-3 ratio ([2]/[3] ratio), and that this can be used to separate low- and high-latitude GDGT distributions in the modern core-top dataset: a range of [2]/[3] ratios and ΔH–L values occur in polar regions, whereas [2]/[3] ratios are high and ΔH–L values are small at temperatures >15°C. However, in the Palaeogene dataset, we observe a wide range of [2]/[3] ratios, even for SST estimates above 15°C. Crucially, we find that water depth is a better discriminator of ΔH–L values and [2]/[3] ratios than SST in the combined modern and Palaeogene dataset: ΔH–L values are low (5.0) where water depth is >1000m. Modern water column studies show that the [2]/[3] ratios in suspended particulate matter (SPM) increase with depth, suggesting that high [2]/[3] ratios reflect a contribution from Archaea living in the deeper water column. This suggests that export dynamics influence GDGT-derived SST estimates. We argue for new approaches to SST reconstruction: 1) continued use of core-top calibrations, in which export dynamics have been implicitly incorporated into the current core-top calibration datasets, but with the influence of water depth taken into account; and 2) use of SPM or mesocosm-based calibrations, with water depth and palaeo-export dynamics independently assessed.