Multivariate analysis of modern and fossil pollen data from the central Tianshan Mountains, Xinjiang, NW China

To interpret past vegetation and climate changes from pollen data, we need to reveal the degree of similarity between modern analogues and fossil pollen spectra, which would help us predict the future climate and vegetation. Ninety surface pollen samples across six vegetation zones along an altitudinal gradient from 460 to 3510 m and 44 fossil samples at Caotan Lake were collected in the central Tianshan Mountains, northern Xinjiang, China. Discriminant analyses results, fossil pollen and phytolith assemblages were then used to reconstruct palaeovegetation and palaeoclimate in the area. The 90 surface samples were divided into six pollen zones (alpine cushion, alpine and subalpine meadow, montane Tianshan spruce forest, forest-steppe ecotone, Artemisia desert, typical desert), corresponding to the major vegetation types in the area. These zones follow a climatic gradient of increasing precipitation with increasing elevation. Paleovegetation reconstructed from 44 fossil pollen assemblages through discriminant analysis reflects the regional vegetation shifted from typical desert to Artemisia desert since 4640 cal. year BP in the Caotan Lake wetland. The fossil pollen and phytolith record also reveal the arid climate has not fundamentally changed in the period. But a dry-wet-dry local climate oscillation since 2700 cal. year BP has a fundamental influence on local wetland vegetation dynamics and peat accumulation of the Caotan wetland. Modern wetland landscape and surface pollen assemblages from the Ebinur Lake Wetland Nature Reserve provide further evidence for ferns and Betula growing in the Caotan Lake wetland during the historical period. Copyright Springer Science+Business Media Dordrecht 2013