Questions: To what extent is the functional structure of plant communities captured by phylogenetic structure? Are some functional dimensions better rep-resented by phylogenetic relationships? In an empirical study, we propose to test the congruence between phylogenetic and functional structure at the alpha and the beta scale along an environmental gradient. Location: Causse du Larzac, southern France. Methods: We measured species abundances and eight key functional traits in 12 plant communities distributed along a gradient of soil depth and resource availability in a Mediterranean rangeland. A phylogenetic super-tree of the spe-cies was assembled, and after quantifying the degree of phylogenetic signal pres-ent in each trait, we quantified taxonomic (TD), phylogenetic (PD) and functional (FD) diversity both within (alpha scale) and among (beta scale) com-munities, taking species abundances into account. We tested for trends in diver-sity along the environmental gradient, and looked for congruence among different facets of diversity, both at the alpha and the beta scale. Results: We found a significant phylogenetic signal for seven out of eight traits. However, when accounting for trends in taxonomic diversity (i.e. richness and evenness), PD did not capture the strong functional structure observed within and among the communities. At the alpha scale, we found an overall pattern of phylogenetic convergence of abundant species, which did not reflect the observed functional divergence. At the beta scale, despite some congruence between betaPD and betaFD for three individual traits, phylogenetic dissimilari-ties did not capture the overall environmental and functional sorting of species according to habitats. Conclusions: We show that even when traits display a significant phylogenetic signal, PD does not capture the complex functional structure of communities in response to environmental gradients. Nevertheless, results suggest that phyloge-netic relationships may partially capture differences in the beta niche of species and provide additional insights on assembly processes not captured by the set of measured functional traits. Only by accounting for patterns in taxonomic diver-sity were we able to disentangle the functional and evolutionary determinants of species assembly along the gradient.