Quantitative Relationship Between Lake Surface Pollen And Vegetation,Climate In North China And Qinghai-tibet Plateau

The quantitative relationship between modern pollen assemblages and vegetation and climate is of great importance for interpreting fossil pollen data and reconstructing paleovegetation and paleoclimate.However,previous studies are seldom based on pollen assemblages of lake surface sediments,and further work is needed to examine the quantitative or semi-quantitative relatioship between pollen assemblages,pollen index and vegetation,climate.Furthermore,the study of pollen-climate model still need data-sets with different spatial extents and sampling densities,and over different timescales.In this study,we surveyed the pollen assemblages of lake surface sediments from 258 sample sites in North China and Qinghai-Tibet Plateau?including 46 sample sites collected by our team?to explore thequantitativerelationshipsbetweenpollenassemblages,Artemisia/Chenopodiaceae?A/C?and vegetation,climate.In addition,we also built pollen of lake surface sediments-climate models and apply these models to paleoclimate reconstruction.The primary conclusions are as follows:?1?The A/C ratio can be used as a semi-quantitative indicator to distinguish steppe vegetation types including alpine steppe,sub-alpine shrub steppe and meadow steppe from desert vegetation types,such as alpine desert,desert steppe,and indicating the changes of annual precipitation(P_ann)and effective moisture conditions?P/Et₀?,and the A/C ratio is more sensitive to P/Et₀,but the indicative effect of A/C may be reduced by the introduction of exotic pollen extra-regional pollen and the growth of Amaranthaceae vegetation in Lakeside saline meadow.?2?Redundancy analysis?RDA?reveal that the ratios of characteristic values of the first and second ordination aixs of P_ann,P/Et₀ and the mean July temperature(T_july)are 0.36,0.88 and 2.56,significance test p value of all three climatic factors is 0.002,and the single explained variables are 14.1%,24.7%and 36.5%.T_july,P/Et₀ and P_annnn were the most three main climatic variables that influence pollen assemblages of lake surface sediments,but the influence of T_july on pollen assemblages might be exaggerated due to the difference in elevations between sampling sites.?3?The model test results of three pollen-climate models based on weighted averaging partial least squares regression?WA-PLS?,modern analogue technique?MAT?and boosted regression trees?BRT?show that,BRT model has the highest coefficient of determination?r²?and the lowest root mean squared error of prediction?RMSEP?,while MAT model has the lowest Max.bias in the prediction of P_ann.MAT model has the highest r²,the lowest RMSEP and the lowest Max.bias in the prediction of P/Et₀.BRT model has r² and the lowest Max.bias,while MAT model has the lowest RMSEP in the prediction of T_july.All three pollen-climate models have good predictive performance and can be used in P_ann reconstruction,but they can not effectively reconstruct T_july because of the selection of sample points and the effect of species indication.?4?As a new method,BRT can be well applied to pollen-climate modeling and paleoclimate reconstruction in northern China and Qinghai-Tibet Plateau,which has the advantages of reflecting to real ecological conditions,high sensitivity of reconstruction results and visualization of process.However,the indication functions of species are vulnerable,and BRT model need a calibration data-set,which has a moderate number of spatial extent,number of samplesand sampling density.