Morphological phylogeny of army ants and other dorylomorphs (Hymenoptera: Formicidae)

The dorylomorph group of ants comprises the three subfamilies of army ants (Aenictinae, Dorylinae, Ecitoninae) together with the subfamilies Aenictogitoninae, Cerapachyinae, and Leptanilloidinae. We describe new morphological characters and synthesize data from the literature in order to present the first hypothesis of phylogenetic relationships among all dorylomorph genera. These data include the first available character information from the newly discovered male caste of Leptanilloidinae. We used ant taxa from Leptanillinae, Myrmeciinae, and the poneromorph (Ponerinae sensu lato) subfamilies Amblyoponinae, Ectatomminae, and Paraponerinae as outgroups. We scored a total of 126 characters from twenty-two terminal taxa and used these data to conduct maximum parsimony and bootstrap analyses. The single most-parsimonious tree and bootstrap results support a single origin of army ants. The Old World army ant genus Dorylus forms a monophyletic group with the enigmatic genus Aenictogiton, which is currently known only from males; the second Old World army ant genus Aenictus is sister to this clade. This result generates the prediction that females of Aenictogiton, when discovered, will be observed to possess the army ant syndrome of behavioural and reproductive traits. The monophyly of the New World army ants (Ecitoninae) is supported very strongly, and within this group the genera Eciton, Nomamyrmex, and Labidus form a robust clade. The monophyly of Leptanilloidinae is also upheld. The subfamily Cerapachyinae appears paraphyletic, although this conclusion is not supported by strong bootstrap results. Relationships among genera of Cerapachyinae similarly are not resolved robustly, although parsimony results suggest clades consisting of (Acanthostichus + Cylindromyrmex) and (Cerapachys + Sphinctomyrmex). We tested for the effect of incompletely known taxa by conducting a secondary analysis in which the two genera containing similar to 50% missing character data (Aenictogiton and Asphinctanilloides) were removed. The strict consensus of the seventeen most-parsimonious trees from this secondary analysis is poorly resolved outside the army ants and contains no clades conflicting with the primary analysis. The position of Leptanilla shifts from forming the sister group to Leptanilloidinae (without high bootstrap support) in the primary analysis, to falling within a polytomy at the base of the root of the dorylomorphs when incompletely known taxa are removed. This instability suggests that the placement of Leptanilla within the dorylomorphs in our primary analysis may be spurious.