Relative pollen productivities of major woody plant taxa in deciduous broadleaved forest in the Ziwuling Mountains of the central Chinese Loess Plateau

Accurate estimates of relative pollen productivity (RPP) values are fundamental to the use of fossil pollen records to quantitatively reconstruct paleovegetation. Based on a pollen dataset from moss polster and related vegetation data for 34 sites in the central part of the Ziwuling Mountains (central Chinese Loess Plateau, with a temperate–warm temperate climate), we estimated the relevant source area of pollen (RSAP) and the RPP values of 8 major woody plant taxa in deciduous broadleaved forest, using the Extended R-Value (ERV) model. The best results were obtained with ERV sub-model 2 and the inverse distance (1/d) method to distance weight the vegetation data. The RSAP was estimated to be 855 m and the rank order of the RPP values for the 8 major woody plant taxa is as follows: Betula (2.083 ± 0.091), Syringa (1.367 ± 0.087), Quercus (1 ± 0, the reference taxon), Pinus (0.791 ± 0.036), Ulmus (0.643 ± 0.062), Carpinus (0.621 ± 0.122), Malus (0.035 ± 0.015), and Acer (0.035 ± 0.025). Repeatability tests using the leave-one-out and K-fold methods showed that the RPP values of all 8 major woody plant taxa have a good reproducibility, which indicates that the RPP values are consistent in the study area. Comparison of the RPP values of the major plant taxa from deciduous broadleaved forest in temperate and warm temperate regions shows that the RPP values for the same pollen taxon may differ significantly between different areas. For example, the RPP for Pinus in areas with dominantly natural vegetation is much lower than for areas where the vegetation has been substantially altered by human activity. We suggest that differences in vegetation/landscape characteristics (including community structure and species composition/abundance) are the major causes of the difference. Therefore, changes in the regional landscape should be considered when choosing suitable RPP values for quantitative paleovegetation reconstruction.