How woody floras have evolved to cope with seasonal drought is still poorly known, especially in tropical Africa, which experiences a drier and more seasonal climate compared to other tropical regions. Here, we characterized the phylogenetic and climatic distribution of three key traits associated with desiccation avoidance or tolerance, focusing on self-supporting woody species of tropical African forests and savannas. We assembled a large database with newly compiled data on species leaf habit (evergreen vs deciduous, 1,255 species) and maximum height (1,281 species), along with new data on oven-dry wood density (1,363 species), climatic niche, and phylogenetic relatedness. Bayesian phylogenetic mixed models were used to assess the phylogenetic signal in niches and trait attributes, to explore individual trait responses to climate and to investigate trait covariations, both in general and for each characterisation of the climatic niche. To date, this study represents the first quantitative assessment of the phylogenetic signal in the species leaf habit for the woody flora of tropical Africa, though drought deciduousness is widespread in seasonal forests and savannas. We identified a strong phylogenetic signal, notably for species biome affinity and leaf habit, but also major evolutionary changes. Relying solely on climate was found to be ineffective for predicting species wood density and insufficient for accurately predicting species leaf habit or maximum height. Among forests, the effect of climate on leaf habit became evident when considering covariations with maximum height or wood density. Small understory species are more likely to be evergreen with dense wood, while canopy species are more likely to be deciduous with light wood. This general pattern varies with climate as both evergreen and deciduous species coexist in the canopy of the wettest sites.

For the tropical woody flora of Africa, we compiled trait attributes at the species level for leaf habit (n=1,255 species), and maximum height (n=1,281 species) using previously published database, original studies, and flora (the list of data sources is provided in the manuscript