干旱区生态环境知识资源中心

To improve the interpretation of fossil pollen records, we analyzed the relationships between modern pollen taxa and climate data and established a pollen-climate calibration using the northern and western Chinese modem pollen dataset to demonstrate its potential for the reconstruction of past climate dynamics in the semi-arid and arid regions of China. Analysis of pollen and climate data involved Pearson’s correlation ceofficient and other transfer functions, as well as the "best modem analogues" method. We also applied a technique for error estimation to calibrate reconstructed values into a probability distribution and a confidence interval for the observed values. The percentage representations of 24 pollen taxa in 546 surface pollen samples collected in northern and western China were correlated with modern climate characteristics. The relationships between individual pollen types and climate were appraised over a large spatial scale in the study area. Of investigated taxa, 41.7% had linear, 45.8% had unimodal (orthogonal polynomial), and 12.5% had monotonic (Log) distribution in response to annual precipitation. Both taxa and plant functional types (PFTs) provided similar results using the best modern analogues method for modern pollen and climate data at the 546 sites in northern and western China. Precipitation reconstructions were more accurate than climate reconstructions of any thermal parameters, suggesting that available moisture was the most important factor controlling plant distribution in the primarily semi-arid to arid study region. Climate reconstruction performed using the modern analogues technique for 112 taxa was not significantly better than reconstruction with only 37 selected taxa. Climate reconstruction based on PFTs was more accurate when using only the main pollen taxa than when using most of the taxa. Our test of the best modern analogues method with the northern and western Chinese surface pollen dataset demonstrated its reliability for reconstructing modern climate variables. (C) 2010 Elsevier B.V. All rights reserved.