Modern pollen distributions in Qinghai-Tibetan Plateau and the development of transfer functions for reconstructing Holocene environmental changes

This study investigates the distribution of modern pollen assemblages in the Qinghai-Tibetan Plateau, China, based on surface soil samples collected at 1202 sites along an altitudinal gradient of 10–5500 m asl, where mean annual precipitation (MAP) ranges from 12 to 1840 mm and mean annual temperature (MAT) from −7 to 21.5 °C. A total of 153 pollen taxa were found with relative abundances greater than 1% in at least two samples. Canonical correspondence analysis (CCA) was used to determine the main environmental variables influencing pollen distributions. The results reveal that MAP is the most significant dominant variable. However, MAT, altitude (ALT), July temperature (MT7), and relative humidity (HHH) are also significant variables that clearly follow the gradients in the CCA ordination, suggesting that pollen assemblages probably not only reflect a single climatic parameter, but also a variety of other climatic inter-related parameters.Transfer functions, based on locally weighted weighted averaging (LWWA), were developed for MAP (R2-boot = 0.89, RMSEP = 109 mm), MAT (R2-boot = 0.78, RMSEP = 2.3 °C), ALT (R2-boot = 0.73, RMSEP = 597 m), HHH (R2-boot = 0.82, RMSEP = 4.5%), and July mean precipitation (MP7) (R2-boot = 0.87, RMSEP = 23 mm). Overall, our results confirm that pollen can provide reliable estimates of the primary climatic parameters. The application of the LWWA model to the fossil records of Chen Co Lake allowed quantitative inferences to be made about Holocene climatic changes in the southern Tibetan Plateau, suggesting that LWWA is a robust calibration method for quantitative palaeo-environmental reconstruction based on pollen data in the regions.

► We provide the pollen transfer functions based on 1202 samples in Tibetan Plateau. ► We examine the reliability of quantitative reconstructions of climate parameter. ► The precipitation lag behind temperature changes during the early to middle Holocene. ► The transfer functions can be used to quantitatively reconstruct palaeo-altitude.