||Phylogenetic patterns and the underlying speciation processes can be deduced from morphological, functional, and ecological patterns of species similarity and divergence. In some cases, though, species retain multiple similarities and remain almost indistinguishable; in other cases, evolutionary convergence can make such patterns misleading; very often in such cases, the “true” picture only emerges from carefully built molecular phylogenies, which may come with major surprises. In addition, closely related species may experience gene flow after divergence, thus potentially blurring species delimitation. By means of advanced inferential methods, we studied molecular divergence between species of the Virola genus (Myristicaceae): widespread Virola michelii and recently described, endemic V. kwatae, using widespread V. surinamensis as a more distantly related outgroup with different ecology and morphology—although with overlapping range. Contrary to expectations, we found that the latter, and not V. michelii, was sister to V. kwatae. Therefore, V. kwatae probably diverged from V. surinamensis through a recent morphological and ecological shift, which brought it close to distantly related V. michelii. Through the modeling of the divergence process, we inferred that gene flow between V. surinamensis and V. kwatae stopped soon after their divergence and resumed later, in a classical secondary contact event which did not erase their ecological and morphological differences. While we cannot exclude that initial divergence occurred in allopatry, current species distribution and the absence of geographical barriers make complete isolation during speciation unlikely. We tentatively conclude that (a) it is possible that divergence occurred in allopatry/parapatry and (b) secondary contact did not suppress divergence. © 2020 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd.