A modern pollen–climate calibration set based on lake sediments from the Tibetan Plateau and its application to a Late Quaternary pollen record from the Qilian Mountains

Abstract Aim Fossil pollen spectra from lake sediments on the Tibetan Plateau have been used for qualitative climate reconstruction, but no modern pollen–climate calibration set based on lake sediments is available to infer past climate quantitatively. This study aims to develop such a dataset and apply it to fossil data. Location The Tibetan Plateau, between 30 and 40° N and 87 and 103° E. Methods We collected surface sediments from 112 lakes and analysed them palynologically. The lakes span a wide range of mean annual precipitation ( P ann ; 31–1022 mm), mean annual temperature ( T ann ; −6.5 to 1 °C), and mean July temperature ( T July ; 2.6–19.7 °C). Redundancy analysis showed that the modern pollen spectra are characteristic of their respective vegetation types and local climate. Transfer functions for P ann , T ann and T July were developed with weighted averaging partial least squares. Model performance was assessed by leave‐one‐out cross‐validation. Results The root mean square errors of prediction (RMSEP) were 104 mm ( P ann ), 1.18 °C ( T ann ) and 1.17 °C ( T July ). The RMSEPs, when expressed as percentages of the gradient sampled, were 10.6% ( P ann ), 15.7% ( T ann ) and 11.9% ( T July ). These low values indicate the good performance of our models. An application of the models to fossil pollen spectra covering the last c . 50 kyr yielded realistic results for Luanhaizi Lake in the Qilian Mountains on the north‐eastern Tibetan Plateau (modern P ann 480 mm; T ann −1 °C). T ann and P ann values similar to present ones were reconstructed for late Marine Isotope Stage 3, with minimum values for the Last Glacial Maximum ( c . 300 mm and 2 °C below present), and maximum values for the early Holocene ( c . 70 mm and 0.5 °C greater than present). Main conclusions The modern pollen–climate calibration set will potentially be useful for quantitative climate reconstructions from lake‐sediment pollen spectra from the Tibetan Plateau, an area of considerable climatic and biogeographical importance.