Pollen, vegetation change and climate at Lake Barombi Mbo (Cameroon) during the last ca. 33 000 cal yr BP: a numerical approach
1Université de Sciences et Techniques de Masuku, Département de Biologie, BP 913, Franceville, Gabon, Central Africa
2CEREGE, UMR6635, CNRS/Université Aix-Marseille/IRD/CdF, BP 80, 13545 Aix-en-Provence cedex 04, France
3ISE-M, UMR5554, CNRS/Université Montpellier II, Place Eugène Bataillon, cc61, 34095 Montpellier cedex 5, France
Abstract. This paper presents quantitative reconstructions of vegetation and climate along the pollen sequence of Lake Barombi Mbo, southwestern Cameroon (4°39'45.75" N, 9°23'51.63" E, 303 m a.s.l.) during the last 33 000 cal yr BP, improving previous empirical interpretations. The biomisation method was applied to reconstruct potential biomes and forest successional stages. Mean annual precipitation, mean annual potential evapotranspiration and an index of moisture availability were reconstructed using modern analogues and an artificial neural network technique. The results show a dense forested environment around Lake Barombi Mbo of mixed evergreen/semi-deciduous type during the most humid phases (highest precipitation and lowest evapotranspiration), but with a more pronounced semi-deciduous type from ca. 6500 cal yr BP to the present day, related to increased seasonality. This forest displays a mature character until ca. 2800 cal yr BP, then becomes of secondary type during the last millennium, probably due to increased human activity. Two episodes of forest fragmentation are shown, which are synchronous with the lowest reconstructed precipitation and highest potential evapotranspiration values. The first of these occurs during the LGM, and the second one from ca. 3000 to ca. 1200 cal yr BP, mainly linked to high precipitation seasonality. Savanna were, however, never extensive within the Barombi Mbo basin, existing instead inside the forest in form of savanna patches. The climate reconstructions at Lake Barombi Mbo suggest that the artificial neural networks technique would be more reliable in this region, although the annual precipitation values are likely under-estimated through the whole sequence.