Phytoliths as quantitative indicators for the reconstruction of past environmental conditions in China I: phytolith-based transfer functions

This study investigated the distribution of phytolith assemblages in China from surface soil samples at 243 sites along significant ecological and climatic gradients to develop transfer functions for quantitative reconstruction of palaeoenvironment. Canonical correspondence analysis (CCA) and detrended correspondence analysis (DCA) were used to determine the main environmental variables influencing phytolith distributions. The results reveal that mean annual precipitation (MAP) is the dominant variable controlling the spatial distribution of phytoliths, which accounts for 39% of the total variance. Mean annual temperature (MAT), relative humidity (Humi), and annual evaporation (VAP) are another three significant variables, accounting for 6%, 10%, and 5%, respectively, of the total variance in phytolith distributions. Transfer functions, based on weighted averaging plus partial least squares (WA-PLS), were developed for MAP (R-boot2=0.90, root-mean-square-error of prediction (RMSEP)=148 mm), MAT (R-boot2=0.84, RMSEP=2.52 °C), Humi (R-boot2=0.75, RMSEP=6.36%) and VAP (R-boot2=0.59, RMSEP=327 mm). Overall, our results confirm that phytoliths can provide reliable and robust estimates of MAP, MAT, Humi and VAP. Thus, WA-PLS is a robust calibration method for quantitative palaeoenvironmental reconstruction based on phytolith data.