ALBERT

Description: Quantitative information regarding the long-term variability of precipitation and vegetation during the period covering both the Lateglacial and the Holocene on the Qinghai–Tibetan Plateau (QTP) is scarce. Herein, we provide new and numerical reconstructions for annual mean precipitation (PANN) and vegetation history over the last 18,000 years using high-resolution pollen data from Lakes Dalianhai and Qinghai on the northeastern QTP. Hitherto, five calibration techniques including weighted averaging (WA), weighted average-partial least squares regression (WA-PLS), modern analogue technique (MAT), locally weighted weighted averaging regression (LWWA) and maximum likelihood (ML), were firstly employed to construct robust inference models and to produce reliable PANN estimates on the QTP. The biomization method was applied for reconstructing the vegetation dynamics. The study area was dominated by steppe and characterized with a highly variable, relatively dry climate at ~18,000–11,000 cal a B.P. PANN increased since the early Holocene, obtained a maximum at ~8000–3000 cal a B.P. with coniferous–temperate mixed forest as the dominant biome, and thereafter declined to present. The PANN reconstructions are broadly consistent with other proxy-based paleoclimatic records from the northeastern QTP and the northern region of monsoonal China. The possible mechanisms behind the precipitation changes may be tentatively attributed to the internal feedback processes of higher-latitude (e.g., North Atlantic) and lower-latitude (e.g., subtropical monsoon) competing climatic regimes, which are primarily modulated by solar energy output as the external driving force. These findings may provide important insights into understanding the future Asian precipitation dynamics under the projected global warming.