Hydrological and edaphic gradients are known to determine local community assembly in tropical forests, but are often summarised in discrete categories. These approaches disregard the continuous response of species abundance to waterlogging. This methodological issue is critical in the context of climate change because water gradient at a local scale is expected to fluctuate due to unstable rain patterns. The use of a continuous proxy for soil moisture such as the Topographic Wetness Index (TWI) overcomes the limits of punctual measurements with hydrological grounded algorithms. The TWI allows the interpolation of the soil moisture based solely on the topography. Taking advantage of interpolation property, the TWI recently unveiled a pervasive hydro-edaphic niche differentiation mechanism in neotropical species complexes. Thus, an evaluation of the predictive efficiency of TWI calculation methods at the local community scale is a crucial step to study local determinants of tropical tree species’ niche. In this study we tested the effects of a set of topographic wetness index variants on the community composition of trees in a neotropical forest. We investigated the consistency of community composition response to TWI with hilltop and lowland classifications already available for the studied species. We further studied the phylogenetic signal of species association with TWI to explore niche differentiation mechanisms in the local community. Combining spatially explicit inventories, botanical determination and LiDAR-derived topographicdata over 120 ha of permanent plots in French Guiana, we used a Bayesian modelling framework toinfer a joint species distribution model. We relied on a hydrogeology-based algorithm to compute ten different indexes of topographic wetness and compared their information criteria in a modelselection process. We classified the effect of TWI for each species based on the sign of the credibility interval. We used Blomberg’s K and variance partitioning to explore the phylogenetic signal of species association with TWI. Our results demonstrate that TWI is an integrative proxy for local-scale community patterns of tropical trees. Using a joint species modelling of the TWI effect on species distribution, we found that several species differed from their classification in the literature, questioning their true habitat preference. The phylogenetic analysis of the TWI effect on species distribution shows that the preference. The phylogenetic analysis of the TWI effect on species distribution shows that the hydro-edaphic niche differentiation is present for more genera than previously described. These results highlight the interest of adapted continuous proxy to retrieve fine scale community assembly along studied gradient and shed light on the mechanisms behind habitat specialisation of tropical trees