Calibrating bacterial tetraether distributions towards in situ soil temperature and application to a loess-paleosol sequence

Branched glycerol dialkyl glycerol tetraethers (brGDGTs) produced by soil-dwelling bacteria offer a promising tool for reconstructing terrestrial temperatures. However, in most previous studies, due to lack of soil temperature data, brGDGTs are calibrated to the air rather than the soil temperature. This may impede our understanding of the accurate response of brGDGTs to temperature, and thus affect the quantitative paleotemperature reconstruction using these lipids. Here, we investigated modern soil brGDGTs and the corresponding soil temperature across a large climatic gradient in China (mean annual soil temperature (MAST) range: −2.7 to 26.2 °C). The results show that the MAST is higher than the mean annual air temperature (MAAT) by 0–6 °C, and the difference is related to vegetation coverage. This supports the idea that vegetation can modulate MAST and points to the necessity of exploring the direct response of brGDGTs to soil temperature. Employing stepwise regression (_sr), we developed MAST_sr and MAAT_sr calibrations, which improve accuracy and reduce the error compared with previous global MAAT calibrations. In Lantian loess-paleosols, the MAST_sr calibration resulted in ∼4 °C lower glacial temperatures and a ∼10 °C deglacial warming comparable with other terrestrial records and climate models. However, the global MAT_mr and MBT’/CBT calibrations produced abnormally higher glacial temperatures, while the empirical MAAT_sr calibration overestimated MAAT during the deglacial period with low vegetation coverage. This demonstrates that the calibration with soil temperature is preferred for quantitative paleotemperature reconstruction. Nevertheless, soil brGDGTs might be useful for inferring MAAT if underlying surface conditions are sufficiently constrained.