Differences of modern pollen assemblages from lake sediments and surface soils in arid and semi-arid China and their significance for pollen-based quantitative climate reconstruction

Modern pollen samples from lake sediments, moss polsters or surface soils are commonly used to interpret fossil pollen records. An increasing number of Holocene pollen records has been published in and and semi-arid Eurasia; however, little is known about pollen representation of surface samples, especially regarding the potential difference and bias from surface lake sediments and surface soils. Here we present results of modern pollen assemblages from 21 pairs of samples from lake sediments and surface soils across north and northwest China to compare their possible differences in representing the surrounding vegetation. Both pollen percentages and principal components analysis (PCA) results show that the main vegetation types (steppe vs. desert) can be distinguished by their modern pollen assemblages from both lake sediments and surface soils. However, PCA results more clearly differentiate the samples from lake sediments and surface soils for both steppe and desert vegetation. Lake samples have higher Artemisia and tree pollen, while surface soils contain more Chenopodiaceae pollen. Standard deviations of pollen percentages of major taxa show that the soil samples have larger variations than the lake samples. Larger variations in Betula and Artemisia in lake samples from steppe vegetation suggest that lake size likely affects the pollen source area. Lakes, especially large lakes, have a larger pollen source area, receiving more pollen from regional source that often includes tree pollen, while surface soils provide more local signals of the surrounding vegetation. Also, dry surface soils from and and semi-arid regions tend to cause differential preservation of pollen grains, resulting in biased representation of actual vegetation. Therefore it is recommended to avoid soil samples for modern pollen training data sets in developing transfer functions for quantitative climate reconstructions for lake sediment cores. (C) 2009 Elsevier B.V. All rights reserved.