STUDY AND STATISTICAL TREATMENT OF POLLEN SAMPLES FROM SURFACE SOIL OF THE RAIN-FOREST IN GABON (AFRICA)

Numerical multivariate analysis has been undertaken on 16 samples of s urface soil collected within two distinct areas of the rain forest of North-eastern Cabon, in order to test the significance of the pollen c omposition of this ecosystem with regard to the different sampling met hods. The results show similar results for multi-subsamples within plo ts of 120 m length to 40 m width, collected randomly, along one or two diagonals or transversals. The greatest variance is illustrated by th e 500 m altitudinal gradient between Makokou and Belinga forests. Corr espondence Analysis allowed the categorising of pollen markers of the rain forest Dacryodes, Hymenostegia, Plagiostyles africana, Pausinysta lia, Piptadenia, Pycnanthus, Calpocalyx and Irvingia in good agreement with the pollen diagram, a standard way of representing the data. Akh ornea and Macaranga are the taxa which show the greatest pollen abunda nce.