The challenge of pollen-based quantitative reconstruction of Holocene plant cover in tropical regions : a pilot study in Cameroon
Résumé
Quantitative pollen-based reconstruction of Holocene plant abundance has not been attempted in the tropics so far. Tropical vegetation is characterized by a large number of entomophilous species and pollen-vegetation modelling for such plants is widely considered too problematic. However, there is a pressing need for quantitative reconstructions of Holocene plant cover in tropical regions to better assess climate- and human-induced environmental crises over time. Of particular note is the environmental crisis at the end of the ‘African Humid Period’ (c. 3 thousand years before present [ka BP]) which remains unquantified. Here, we present results from a pilot study in the Cameroon highlands including estimates of relative pollen productivity (RPP) for major taxa of the Afromontane forests and grasslands using the Extended R-Value (ERV) model, and the first reconstructions of past regional and local Holocene plant cover using the Landscape Reconstruction Algorithm (LRA). RPP estimates were obtained for 12 taxa. We found that Celtis, Alchornea, Syzygium, Macaranga/Mallotus, Nuxia type, and Schefflera have RPPs larger than Poaceae, and Moraceae, Combretaceae/Melastomataceae, Prunus africana/Rubus pinnatus, Cyperaceae, and Podocarpus lower RPPs than Poaceae. The LRA was applied to three Holocene pollen records from Cameroon (Bambili, Mbi, Mbalang) and one from Nigeria (Tilla). The results confirm the abruptness of the forest collapse in Cameroon at 3 ka BP, and suggest that it was faster and of larger magnitude than earlier anticipated. The LRA reconstruction also indicates that the regional landscape in the mountain region was more open than uncorrected pollen percentages alone suggest. Moreover, it supports the earlier interpretation of a dense forest at Bambili 11–3.5 ka BP and the local occurrence of Afromontane forest taxa at Mbalang in the Mid-Holocene.