1ISE-M, UMR 5554 CNRS/Université Montpellier II, Place Eugène Bataillon, cc61, 34095 Montpellier cedex 5, France
2Institut de Recherche en Ecologie Tropicale (IRET/CENAREST), BP 13354, Libreville, Gabon
3CEREGE, CNRS/Université Aix-Marseille/IRD/CdF, BP 80, 13545 Aix-en-Provence cedex 04, France
4Faculté des Sciences, Université Marien Ngouabi, BP 69, Brazzaville, Congo
5LSCE, UMR CNRS/CEA/UVSQ, 12 Avenue de la Terrasse, 91198 Gif-sur-Yvette cedex, France
Abstract. New detailed vegetation reconstructions are proposed in Atlantic Central Africa from a modern pollen data set derived from 199 sites (Cameroon, Gabon and Congo) including 131 new sites. In this study, the concept of plant functional classification is improved with new and more detailed plant functional types (PFTs) and new aggregations of pollen taxa. Using the biomisation method, we reconstructed (1) modern potential biomes and (2) potential succession stages of forest regeneration, a new approach in Atlantic Central African vegetation dynamics and ecosystem functioning reconstruction. When compared to local vegetation, potential biomes are correctly reconstructed (97.5% of the sites) and tropical rain forest (TRFO biome) is well identified from tropical seasonal forest (TSFO biome). When the potential biomes are superimposed on the White's vegetation map, only 76.4% of the sites are correctly reconstructed. But using botanical data, correspondence and cluster analyses, the 43 sites from Congo (Mayombe) evidence more affinities with those of central Gabon and so they can also be considered as correctly reconstructed as TRFO biome and White's map should be revised. In terms of potential succession stages of forest regeneration, the mature forest (TMFO) is well differentiated from the secondary forest (TSFE), but inside this latter group, the young and the pioneer stages are not clearly identified due probably to their low sampling representation. Moreover, linked to their progressive and mosaic character, the boundaries between two forest biomes or two forest stages are not clearly detected and need also a more intensive sampling in such transitions.