Are modern pollen assemblages from soils and mosses the same? A comparison of natural pollen traps from subtropical China

Abstract This study investigated the pollen spectra from 46 pairs of soil and moss samples (pairs collected within a 1 m2 squared area). The samples were collected from six common subtropical vegetation communities in the Meiling Mountains, southeast China, with the vegetation proportions recorded at the collection point. Principal Component Analysis (PCA) was used to investigate the separation of the paired assemblages and to determine whether different plant communities produced distinctive pollen spectra. Paired soil and moss samples captured similar levels of variability in the pollen assemblages retained, but there are systematic differences in the mean values of key groups of taxa. Monte Carlo sampling shows that, in most cases, intra-pair differences are greater than could be explained by counting uncertainty alone. In this study, discriminant analysis of surface soil and moss found that 91.3% of the soil samples and 87% of the moss samples were correctly classified into their vegetation communities. However, the detailed pollen assemblages suggest that mosses provide a more accurate representation of the contemporary vegetation composition than soils. Pollen assemblages from moss samples seem to record local vegetation more accurately than those from soil samples. Higher vegetation diversity within an arboreal forest community leads to greater differences between moss-soil pairs. In bamboo forests, pollen assemblages in soils and moss show strong influence from the surrounding communities, which makes it hard to identify bamboo forest via surface sample pollen assemblages alone.