Format: ORAL

Jan Lukas Klein1,2, Thomas Drouet2, Gilles Dauby3, Nicolas Texier1,4,5, Olivier Hardy5, Eric Akouangou6, Ehoarn Bidault4,7, Archange Boupoya6, Davy Ulrich Ikabanga4,8, Olivier Lachenaud1,9, Diosdado Nguema10, Tariq Stvart1,4,9

1 Herbarium et Bibliothèque de Botanique africaine, CP 265, Université Libre de Bruxelles, Blvd. du Triomphe, B-1050, Brussels, Belgium 2 Laboratoire d’Écologie Végétale et Biogéochimie, Université Libre de Bruxelles, CP 244, 50 Av. F. D. Roosevelt, Brussels, 1050 Belgium 3 AMAP, Univ Montpellier, CIRAD, CNRS, INRAE, IRD, Montpellier, France 4 Missouri Botanical Garden, Africa & Madagascar Department, 4344 Shaw Blvd., St. Louis, MO 63110, USA 5 Laboratoire de Génétique et Ecologie végétales, Université Libre de Bruxelles, 1850 Chaussée de Wavre, B-1160 Bruxelles, Belgium 6 Herbier National du Gabon, BP 1156 Libreville,Gabon. 7 Institut de Systématique, Évolution, et Biodiversité (ISYEB), Muséum National d’Histoire Naturelle, Centre National de la Recherche Scientifique, Sorbonne Université, École Pratique des Hautes Études, Université des Antilles, C.P. 39, 57 rue Cuvier, F-75005, Paris, France 8 Laboratoire d’Ecologie Végétale et de Biosystématique, Département de Biologie, Faculté des Sciences, Université des Sciences et Techniques de Masuku, B.P. 941, Franceville, Gabon 9 Jardin botanique de Meise, Nieuwelaan 38, 1860, Meise, Belgium 10 Tropic-Forest, BP 4474 Libreville, Gabon

In Gabon, that hosts the best-preserved forests in western Central Africa,we lack a consistent and data-driven floristic regionalization needed to design conservation and management strategies. Existing forest typologies rely on logging inventory data which only include a small subset of the canopy tree diversity. Studies demonstrated that the canopy in large parts of Central Africa is dominated by long-living pioneer species that echo widespread historical human disturbance, paleo-climatic disturbances during the Last Glacial Maximum or a combination of both. Additionally, the canopy composition has been altered by industrial logging that selectively extracts large individuals of a few but widespread species. We thus hypothesize that (a) existing typologies draw a heavily biased picture of present-day tree compositional patterns and (b) that the canopy floristic composition is less dependent on environmental factors than the understory. Using a unique dataset comprising 466 vegetation transects inventoried across Gabon including 129,876 trees with a diameter at breast height 5 cm, we produced two continuous categorical maps, one by excluding and a second by including the understory composition. We identified forest types through hierarchical clustering of the floristic composition of transects and predicted forest types based on environmental and anthropogenic variables obtained through free available raster data through spatially explicit randomforest models. We tested for the importance of each predictor in the model. The V-measure was used to compare both maps and each respective map to those from the literature. The both-strata typology and the canopy typology are incongruent while the latter resembles existing forest typologies more strongly than the former. Human activity variables were of greater importance in predicting canopy-based compared to both-strata-based forest types. Our results emphasize the importance of integrating the understory composition when aiming at unveiling floristic patterns in Central Africa for it is less altered by historical successional dynamics.