【2022】Taxonomy of Ectatomminae Revisited

We propose taxonomic changes for the subfamily that improve ant  systematics, i.e., by ensuring that formally named taxa are monophyletic, while simultaneously keeping names fairly stable. At the  subfamily level, our decision to synonymize Heteroponerinae  under Ectatomminae is not based on the monophyly of these  groups, since both are reciprocally monophyletic as currently circumscribed, and their sister-group relationship has been broadly  discussed. Historically, the close relationship between both groups  has been supported by morphological (Brown 1958, Bolton 2003,  Ward 2007, Keller 2011) and molecular data (Brady et  al. 2006,  Moreau et al. 2006, Moreau and Bell 2013, Branstetter et al. 2017).  However, morphology can be misleading, especially when defining  the diagnostic characters for the groups separately. When describing  Heteroponerinae, Bolton (2003) stated that there is no unequivocal  apomorphy for the subfamily, suggesting a number of characters  that could have this status. Feitosa (2015) investigated the phylogeny  of Heteroponerinae using morphological data, testing the characters suggested by Bolton (2003), as well as by several others, and  also could not identify any apomorphy for the group. However, in  his work, Feitosa included species of Ectatomminae as outgroups  and his analysis suggested at least ten diagnostic characters for the  clade comprising both Ectatomminae and Heteroponerinae. For  this reason, we reclassify all ectaheteromorph ants as members of  a single subfamily, ensuring the monophyly criterion that already  applies to all other ant subfamilies but, most importantly, providing a clear diagnosis for the subfamily based on morphological  synapomorphies. Regarding taxonomic changes at the tribal level, our aim is to  keep the classification stable. In this sense, the new combination of  the tribe Heteroponerini and the synonymy of Typhlomyrmecini  are made to ensure the correct placement of the former, and the  monophyly of Ectatommini in the case of the latter. At a generic  level within the tribe Heteroponerini, the paraphyly of Heteroponera is a striking result, unpredicted by morphology, with B. microps appearing as a separately diverging lineage. This result is congruent  with the previous hypotheses of Borgmeier (1957) and Feitosa  (2015), which suggested that the diagnostic characters for this  species are highly divergent from the morphological patterns for  Heteroponera, but its placement as a separate genus is supported  here for the first time. Similarly, the position of H. monticola, recovered as sister to all the other Heteroponera species, as well as the  recovery of two separate clades, the first comprising H. carinifrons (from Chile) as sister to the Australasian species and the second comprising the remaining Neotropical species, are also entirely new evolutionary hypotheses for the genus, with strong implications for its  biogeographical history. This phylogenetic scenario suggests that the common ancestor  of Heteroponerini morphologically resembled a modern member of  Acanthoponera, with a relatively large body, prominent spines, well  developed eyes, and long palps. An early lineage probably split off  and evolved to occupy the epigaeic and hypogaeic strata of the environment, maybe displaced by an emerging dominant lineage of ants  (e.g., Myrmicinae). This now cryptic early lineage of heteroponerines  has undergone a drastic reduction of body size, appendages, and eyes,  as we can see in the extant Boltonia. Later, a second divergence event  separated two lineages of Heteroponera and adaptation for living  in the ground was repeated. In this second process, H. monticola and H.  inca retained several plesiomorphic traits, also related to  Acanthoponera, but the remaining Heteroponera gradually lost these  characters as they made their way to the soil and morphologically  converged on Boltonia in the reduction of appendages and body size.  This scenario is supported by the presence of tarsal teeth and lobes  in Acanthoponera, traits strictly related to arboreal habits that were  lost in the remaining lineages of heteroponerine adapted to nesting  and foraging in the ground (Feitosa 2015). Our results regarding relationships among species in Heteroponerini shed new light on the  study of their morphological evolution. We believe that, in order to  ensure the stability of the classification, to best understand the evolution of this group, and to make the most significant contribution  to ant systematics, the assessment of relationships among the species should combine both molecular and morphological approaches.  Unfortunately, the genus Aulacopone was not included in our analysis due to the unavailability of specimens and difficulties of collecting in its type locality. The genus is monotypic and was collected  only twice in the 1920/30s, with the only known specimen currently  metal-coated, making recovery of DNA information from the pinned  specimen a risk to the only specimen available. The distribution of  this genus is singular within the Ectatomminae, being the only group  to occur in the Palearctic region. Aulacopone is said to share several  morphological similarities with the other heteroponerines (Brown  1958, Taylor 1980, Lattke 1994, Bolton 2003, Feitosa 2015), but  its position among the Ectatomminae is still not well defined due to  the impossibility of examining important characters in the previous  phylogenetic study (Feitosa 2015). The eight genera that comprise Ectatommini are shown to form  a well-supported clade, a result that is congruent with previous  morphological hypotheses for the group (Bolton 2003, Ward 2007,  Keller 2011), although these works considered the four genera as  previously defined. In the molecular phylogenies published so far,  only one or a few specimens of each genus were included, limiting  their conclusions regarding the relationships among them (Brady  et al. 2006, Moreau et al. 2006, Moreau and Bell 2013, Branstetter  et  al. 2017). Given these limitations, this is the first molecular  study that aimed to investigate the genus-level relationships in  Ectatomminae. A fairly novel result, the sister-group relationship between Ectatomma and Rhytidoponera, is congruent with previous  morphological hypotheses by Keller (2000, 2011) and was suggested  by other broad-scale molecular phylogenies of Formicidae that did  not focus specifically on these groups (Brady et al. 2006, Moreau  et  al. 2006, Moreau and Bell 2013). Brown (1958) noticed some  similarities between the two genera, noting similarities in wing venation and male genitalia and absence of a metacoxal spine (present  in Holcoponera, Gnamptogenys, and Stictoponera). Also, Brown  (1958) called attention to similarities between Ectatomma workers  and those of the largest species of Rhytidoponera. Our results are  the first to include broad species-level representatives of those genera  and our results shed light on the evolution of these groups. Perhaps the most strikingly novel result in our study is the strong  support for the paraphyly of the former Gnamptogenys in relation to Typhlomyrmex. This result was never previously predicted  by any morphological or molecular study. Historically, the position  of Typhlomyrmex relative to the Ectatommini was first addressed  by Emery (1911), but Brown (1965) later placed the genus in its  own tribe, Typhlomyrmecini, considering it to be closely related to  Copyedited by: OUP Downloaded from https://academic.oup.com/isd/article/6/1/5/6514765 by Universidade Federal do Paraná user on 25 January 20229 Insect Systematics and Diversity, 2022, Vol. 6, No. 1 the Amblyoponini. Lattke (1994) suggested that the similarities of  Typhlomyrmecini with Ectatommini required further exploration  and Bolton (2003) most recently considered the Typhlomyrmecini to  be a member of Ectatomminae. However, rather than forming a separate tribe in Ectatomminae, it now appears that this group of ants is  a highly derived lineage among the former species of Gnamptogenys with a distinctive cryptic morphology. The paraphyly of Gnamptogenys in relation to Typhlomyrmex provided two different alternatives for the taxonomic treatment of  the genera in Ectatomminae, the first being the synonymization of  Gnamptogenys under Typhlomyrmex, since the latter is the oldest  available name. However, we recognize the importance of the name  Gnamptogenys within the myrmecological literature and, with a nomenclatural gender change from feminine (Gnamptogenys) to masculine (Typhlomyrmex) for most species, this would not be the most  parsimonious treatment. The second possibility, chosen here, involved  reviving available names for the different clades recovered in our phylogeny, considering the similar phylogenetic distances between those  clades and between other Ectatomminae genera, and the strong diagnostic morphological characters recovered for each of the lineages.  The availability of generic names for each of those clades shows that  hypotheses for those groups were once presented, but morphological  data were not sufficient to define them at the time, and they were  later synonymized under Gnamptogenys (Brown 1958). With our molecular dataset we recovered each clade with strong support and, by  reciprocal illumination, defined the morphological characters that separate each genus from any other genus in Ectatomminae. The generic status of Holcoponera, Stictoponera, and Alfaria were subjects of long and arduous inquiry into the myrmecological  literature since they were first proposed as subgenera of Ectatomma in the case of the first two, or as a genus, in the case of Alfaria.  Brown (1958) found no basis for maintaining the generic status of  those names, but divided Gnamptogenys into four groups, namely  the Gnamptogenys group, the Stictoponera group, the Holcoponera group, and the Alfaria group. Brown considered Holcoponera to be  a well-defined genus based on its more compact, dorsally convex  mesosoma with a marked promesonotal suture interrupting the  sculpture and on the form of the petiolar node, as well as by characters of wing venation and larval hairs. However, when analyzing  the similarities between the species Typhlomyrmex reichenspergeri,  Holcoponera relicta, and Holcoponera mina, he considered the lack  of gastric sculpture in T. reichenspergeri as evidence against its placement in a separate genus. In our study, we recovered T. reichenspergeri as sister to Typhlomyrmex and relatively distant from Holcoponera and we found that Holcoponera is not a strictly Neotropical genus  because it also includes Indomalayan and Australian species formerly described as Rhopalopone and Wheeleripone. Brown considered it impossible to define the genus Stictoponera because of  dissimilarities among the Old World species. We resolve the problem  by showing that Old World species fall into two independent clades,  one within Holcoponera. Brown considered the genus Alfaria to be  the most distinct of the ectatommine genera but felt that A. striolata cast doubts on its generic status due to the less inflated second gastric  segment and to its sculpture, which is similar to that of Stictoponera.  Our genomic data, however, show that Alfaria forms a distinct clade  among the Ectatomminae and, even though A. striolata was not included in the phylogeny, the presence of an expanded frontal carina  suggests that this species placement is correct. The current definition of the genus Poneracantha is a novel result,  as this was proposed as a monotypic subgenus to contain the highly  divergent P.  bispinosa. However, Lattke (1995) proposed that the  specialized millipede predators that belong to this genus formed the  Gnamptogenys rastrata group and considered them to be closer to  Holcoponera than to the present definition of Gnamptogenys based  on the presence of triangular mandibles, long and typically sculptured scapes, the convex clypeal lamella, and the well-developed  metacoxal tooth, a result that is also recovered by our molecular  data. Lattke (1995) also recovered the sulcata and mordax groups  as sister groups, with the concinna group as closely related to  them, but not monophyletic. We obtained similar results, except  for the monophyly of the concinna group, and redefine the sulcata,  concinna, and mordax groups as a smaller, strictly Neotropical  Gnamptogenys. Finally, the sister-group relationship between the  species T. reichenspergeri, T. lenis, and T. lavra and the remaining  Typhlomyrmex is a result never predicted by morphology and, in  fact, the phylogenetic distance among those species is similar to the  distance among other genera. Those species have in common absent or reduced eyes, with less than 15 ommatidia; promesonotal  suture well marked, totally interrupting dorsal mesosomal sculpture;  propodeal spiracle separated from declivity margin by a distance  longer than its diameter; metacoxal dorsum unarmed; and petiole  pedunculate. T.  reichenspergeri, T.  lenis, and T.  lavra lack a welldefined antennal club and a prominent anteroventral process on  the petiole. We chose to combine those species into Typhlomyrmex based on these shared diagnostic characteristics, in the interest of a  more stable classification. Additional work is necessary because we strongly believe that the  molecular phylogenetic data should be combined with the study of  morphological characters that are diagnostic for the newly defined  genera and for the new generic combinations, so that the final classification can be functional and useful to any researcher studying specimens in the laboratory or in the field. In this study, we demonstrate  that UCE data provide a robust source of phylogenomic data for  the Ectatomminae ants. Morphological evolution, interpreted with  reference to our resulting phylogeny, has produced diagnostic characters for defining taxonomic groups. We believe that the phylogenetic framework and the new classification proposed here provides a  solid foundation for the further study of Ectatomminae taxonomy  and systematics, as well as for reconstructing the morphological evolution of the genera, species groups, and species that it comprises.