Validation of relative pollen productivities for major tropical plant taxa: An empirical test using the REVEALS model in Hainan, China.

The reliability of estimates of tropical lowland vegetation cover based on lacustrine pollen records has long been debated. Here, we assess the reliability of relative pollen productivity estimates (RPPs) obtained in tropical Hainan island using surface pollen assemblages from six reservoirs and the Regional Estimates of Vegetation Abundance from Large Sites (REVEALS) model. The resulting REVEALS estimates are compared to observations of vegetation coverin the studied region and pollen percentages. The REVEALS estimates for most tropical taxa including Poaceae, Moraceae, Euphorbiaceae, Mallotus/Macaranga , Castanopsis/Lithocarpus , were more similar to the observed vegetation cover than estimates based on simple pollen percentages, suggesting that the previously reported RPPs are reliable. The REVEALS estimates for some taxa (i.e. Rubiaceae and Anacardiaceae) are comparable to those based on pollen percentages, but RPPs for these taxa are difficult to evaluate. However, the pollen percentages for some taxa including Asteraceae and Papilionaceae agree better with the observed vegetation cover than the REVEALS corrections, implying either that there are problems with the corresponding RPPs or that there is bias in the observed vegetation data used for validation. We suggest that the reliability of the RPPs for plant taxa in tropical areas depends strongly on their abundance in the pollen and vegetation data: high abundance implies high reliability and vice versa. REVEALS cover estimates for groups of taxa consistently agreed well with observed vegetation cover, suggesting that the REVEALS model can be applied even in tropical regions if individual species are grouped, which may be useful when reconstructing historical landscape changes in tropical Asia. [ABSTRACT FROM AUTHOR]