Vegetation effects on temperature calibrations of branched glycerol dialkyl glycerol tetraether (brGDGTs) in soils

Abstract

Distributions of branched glycerol dialkyl glycerol tetraether (brGDGT) lipids are sensitive to environmental parameters, which enables their use in paleoenvironmental studies. In particular, the degree of methylation of brGDGTs (MBT and MBT′) and the cyclization ratio of brGDGTs (CBT) are often used to estimate changes in temperature in paleoclimate studies. Application of these proxies requires reliable and precise calibrations to temperature. However, available calibrations of brGDGT proxies to temperature have large calibration errors in arid areas which suggests that other environmental variables influence brGDGT distributions in these areas. Here we analyze brGDGT distributions in soils from the Tibetan Plateau, an arid/semi-arid region with gradients in vegetation cover (forest, grassland, and desert) to examine: (1) the applicability of MBT′/CBT calibrations to reconstruct mean annual air temperature (MAAT); and (2) the effects of vegetation cover on temperature calibrations. We show that temperatures reconstructed using the global soil MBT′/CBT calibrations are warmer than instrumentally observed temperature, and that the calibrations are significantly influenced by the presence or absence of vegetation. Excluding sample sites without vegetation (bare soil) from temperature calibrations substantially improves the correlation between observed and reconstructed temperatures. Within the vegetated soils, we reanalyzed published global soil calibration datasets, including results of analyses that either did or did not separate 5- and 6-methyl brGDGT isomers, and find that the correlation between MBT′5Me and MBT′ and temperature varies for different vegetation types (grass and forest) and that the error of temperature calibrations is reduced if calibrations are separated by the type of vegetative cover (grass or forest). We test these new calibrations in sequences from the Lantian and Mangshan loess of the Chinese Loess Plateau and find that reconstructed temperatures from both modern and Last Glacial Maximum sediments are more consistent with other reconstructions and climate model simulations when vegetation-specific calibrations are applied to the brGDGT data. The results indicate that changes in vegetation should be taken into consideration when applying brGDGT proxies to reconstruct past changes in climate.