The Surface Soil Pollen Assemblages And Their Quantitative Relationships With Modern Vegetation And Climate In Inner Mongolia And Its Surrounding Areas

Surface soil pollen assemblages and their quantitative relationships with modern vegetation and climate provide basis for quantitative reconstruction of paleovegetation and paleoclimate. In China, many modern pollen datasets have been collected and studied. However, further studies are still needed because the published data can not satisfy the perfect pollen based quantitative climate reconstruction for its discontinuous and uneven spatial distribution. Furthermore, no uniform standard on methodology of pollen extraction and identification were used for those existed data. Therefore, we surveyed 398 surface soil pollen assemblages from Inner Mongolia and its surrounding areas (the Taihang Mt, the Yan Mt and the Lvliang Mt) to explore the quantitative relationships between the pollen assemblages and the vegetation and also the climate. Furthermore, leave-one-out cross-validation and the spatial autocorrelation test were conducted to evaluate the performance of the modern pollen calibration in hope that basis could be provided for the paleoclimate reconstruction. The primary results are as follows:(1) The study of surface soil pollen assemblages and the results of discriminate analysis indicate that the pollen assemblages have distinct characteristics in different vegetation and the vegetation can be reflected by the pollen assemblages.(2) The results of canonical correspondence analysis (CCA) reveal that the mean annual precipitation (PANN), the mean annual temperature (TANN), the mean July temperature (TJuly) and the mean January temperature (TJan) are the main climate variables that control or influence the surface pollen assemblages. Among them, PANN is the most significant dominant variable and the most promising climate variable in quantitative reconstruction.(3) Quantitative transfer functions were developed using weighted averaging partial least squares regression (WA-PLS) and modem analogue technique (MAT). Performance of each calibration model was evaluated using leave-one-out cross-validation and the influence of spatial autocorrelation on pollen-climate transfer function was examined by the R package palaeoSig. The results show that the WA-PLS models have better performance than the MAT models for less impact of spatial autocorrelation though the latter maybe well performed in the leave-one-out cross validation. Thus, the WA-PLS models of this pollen calibration are the most suitable candidate for pollen based quantitative palaeoclimate reconstruction. Furthermore, performance comparisons with regional pollen calibrations also suggest that our pollen calibration is robust and has potential in pollen based paleoclimate reconstruction. In the end, the modern pollen calibration was applied to reconstruct the Holocene precipitation of Xihaizi in Wulanchabu city of Inner Mongolia. The broad consistencies probably suggest that our modern pollen calibration is reliable.