Use of plots to define pollen-vegetation relationships in densely forested ecosystems of Tropical Africa.

Modern soil samples from South Congo were analyzed for pollen content and compared to forest inventories to define modern pollen-vegetation relationships. A correspondence analysis (CA) was applied independently to botanical and pollen data and a hierarchical cluster analysis to pollen data only. Subsequently, a CA using a presence-absence approach has been made to directly compare the two types of data. Results show that the pollen rain and floristic composition of the sampled sites are not directly linked to altitudinal or precipitation gradients, but clear evidence of variation in relation to hygromorphy and soil type is detected. The forests occurring in swampy environments are well differentiated from the forests developed on well-drained soils by pollen and floristic data. Among forests on well-drained soils, a good distinction can be made between those growing on sandy soils and those growing on ferralitic soils. The comparison between pollen spectra and vegetation shows site-to-site variations in pollen assemblages in relation to the floristic heterogeneity of forests, and it appears that few taxa show a good correlation between plant cover and pollen abundance.