Appraisal of branched glycerol dialkyl glycerol tetraether-based indices for North China

The distributions of branched glycerol dialkyl glycerol tetraethers (brGDGTs) are quite promising for quantitative reconstruction of temperature and pH in paleoclimate studies. Recently, the separate quantification of the 5- and 6-methyl brGDGTs allowed establishment of new indices that may provide more accurate temperature and pH estimates than traditional brGDGT indices. Here, we systematically investigated the applicability of various brGDGT indices in an independent set of 148 soils from the temperate northern area of China. The results support the idea that the new brGDGT indices are more applicable than the traditional ones, with regional calibrations showing more accurate mean annual air temperature (MAAT) and pH estimates than the global calibrations. Previous studies noted that variation in soil temperature might differ from variation in air temperature. The mismatch may cause scatter in the brGDGT-MAAT calibrations, as soil bacteria producing brGDGTs should respond more directly to soil temperature variation. We therefore monitored year round, in-situ soil temperature variation for 20 selected soils from this large region. Each temperature proxy correlated better with the in-situ soil mean annual temperature (MAT) than with the meteorological MAAT. Further comparison of correlations between seasonal soil temperature and brGDGT indices showed that the latter related best to growing season temperature (tentatively defined as the mean temperature of spring, summer and autumn) and most weakly to winter temperature, pointing to a seasonal preference for brGDGT production. Hence, the offset between soil MAT and MAAT and the seasonal production of brGDGTs are both likely responsible for part of the scatter and bias in the brGDGT-MAAT calibrations for temperate regions. Additionally, we propose two supplementary indices to the recently proposed multiple linear regression index and Index 1, for tracing soil temperature variation for the region of North China. The 4 indices could yield regional temperature calibrations with R²>0.90 and root mean square error (RMSE)<1.6°C vs. the growing season soil temperature. Our study also implies that, on a global scale, the calibrations between brGDGT temperature proxies and in-situ soil temperature may provide more reliable temperature estimates in paleoclimate studies, so in-situ soil temperature measurements of globally distributed soils are required for the development of improved temperature calibrations using soil brGDGTs.