• Citer

The human–made forest?: Methodological considerations for unravelling anthropogenic determinism among tree communities in Central Africa

par Klein, Jan Lukas
Président du jury Vleminckx, Jason
Promoteur Drouet, Thomas
Co-Promoteur Stévart, Tariq
Publication Non publié, 2026-06-19

Thèse de doctorat

• Accès en ligne
• Détails
• Contenu
• Statistiques

Résumé :

Ecologists increasingly acknowledge that human activities have left long-lasting legacies in the tree species composition of tropical rainforests. In Central Africa, traditionally mobile and widespread societies have practiced swidden agriculture for millennia. In the early 20th century, colonial policies profoundly altered spatial patterns of human occupancy by regrouping, relocating and sedentarizing settlements along major road axes (regroupement). Therefore, the presence of long-living pioneer (LLP) species in presently uninhabited regions is commonly interpreted as secondary regrowth on swidden fallows, where it correlates with archeological artifacts and/or soil charcoal. However, such approaches are spatially discontinuous, hampering efforts to predict anthropogenic pressure at larger spatial scales. The aim of this thesis is to advance our understanding of anthropogenic determinism on tree species turnover in western Central Africa. To do so, we investigate shifts of LLP abundance throughout forest recovery, explore alternative data sources to locate and date historical swidden fallows, and assess how minimum tree size, observer error and emphasis on species abundance affect the relative contribution of anthropogenic and environmental drivers of tree species turnover.This study operates at two spatial scales. First, we use a chronosequence (5–90 years) from northeastern Gabon (landscape scale), comprising 31 botanical inventories of trees ≥ 5 cm diameter, in combination with associated soil and historical data. Second, we use 493 sampling units from across Gabon (regional scale), 99 of which are associated with soil data. At the local scale, we demonstrated that LLP peak in abundance at approximately 50 years after field abandonment and remain in old-growth forest, particularly among large trees. This suggests that LLP may also regenerate in natural tree fall gaps. We furthermore showed that historical maps and oral histories yield congruent reconstructions of past settlement distribution (1911–present) and explain substantial proportions of community-weighted wood density among canopy trees (36 and 45%, respectively). At the regional scale, we demonstrated that human activities have significantly affected tree species composition, alongside climatic and edaphic conditions. More specifically, we found that anthropogenic determinism is stronger among canopy trees than understory trees, and species abundances were more strongly affected by human activity than presence/absence data. Observer error did not affect the relative contribution of anthropogenic versus environmental drivers.The key finding is (i) that climatic and edaphic heterogeneity account for far greater proportions of regional-scale tree species turnover than human activity, and (ii) that the relative contribution of anthropogenic determinism increases when small trees are excluded and abundant species are emphasized. Conclusively, this thesis provides methodological considerations for quantifying anthropogenic determinism and contributes to ongoing debates surrounding the degree of pristineness of Central African forests.