Various mechanistic theories of community assembly have been proposed ranging from niche-based theory to neutral theory. Analyses of beta diversity in a phylogenetic context could provide an excellent opportunity for testing many of these hypotheses. We analyzed the patterns of phylogenetic beta diversity in tropical tree communities in Panama to test several community assembly hypotheses. In particular, the degree to which the phylogenetic dissimilarity between communities can be explained by geographical or environmental distance can yield support for stochastic or deterministic assembly processes, respectively. Therefore, we examined: (i) the existence of distance decay of phylogenetic similarity among communities and its degree of departure from that expected under a null model; and (ii) the relative importance of geographical versus environmental distance in predicting the phylogenetic dissimilarity of communities. We found evidence that the similarity in the phylogenetic composition of communities decayed with geographical distance and environmental gradients. Null model evidence showed that beta diversity in the study system was phylogenetically non-random. Our results highlighted not only the role of local ecological mechanisms, including environmental filtering and competitive exclusion, but also biogeographical processes such as speciation, dispersal limitation, and niche evolution in structuring phylogenetic turnover. These results also highlight the importance of niche conservatism in structuring species diversity patterns.