Distributions of membrane lipids (3-hydroxy fatty acids and brGDGTs) in the Eastern Himalayan soil and their global context: insights into temperature and pH variations

Abstract

Estimating past temperature and pH in terrestrial environments presents significant challenges. Microbial membrane lipids-based 3-hydroxy fatty acids (3-OH FAs) and branched glycerol dialkyl glycerol tetraethers (brGDGTs) are increasingly used as proxies for reconstructing past environmental parameters such as temperature and pH. However, global calibrations of their temperature indices with mean annual air temperature (MAAT) often show high root mean square errors (RMSE), suggesting that factors beyond temperature influence the distribution of lipids. To explore these controls, we analyzed soils across an elevational gradient (250–4600 m) in the Eastern Himalaya, covering wide ranges in MAAT (23–0.6 °C), pH (4.8–6.6), and other soil properties. Our results suggest that root activity can significantly alter the distribution of 3-OH FAs and brGDGTs, potentially biasing paleo-temperature reconstructions. The temperature indices were less accurate in the lower temperature range, whereas the pH indices exhibited strong correlations in acidic soils (R2 = 0.59). A global soil dataset (brGDGTs, n = 1938; 3-OH FAs, n = 343) revealed similar patterns, with weaker pH responses in alkaline environments (R2 = 0.08). To improve predictability, we recalibrated MAAT using multiple linear regression (MLR) and Random Forest (RF), with RF outperforming MLR in predictive skill. Incorporating additional environmental parameters into the MLR significantly improved the models, resulting in higher coefficients of determination (R2 = 0.96–0.97) and lower RMSE (∼1.1–1.2 °C). These findings underscore the importance of accounting for site-specific environmental factors in calibration models to enhance the accuracy of microbial lipid proxies for paleoclimate reconstruction.