Grappling with homoplasy: taxonomic refinements and reassignments in the ant genera Camponotus and Colobopsis (Hymenoptera: Formicidae)

Camponotus and Colobopsis are widely distributed and species-rich genera in the ant tribe Camponotini. Molecular phylogenetic studies demonstrate that they are not sister taxa, but several lineages within each genus have converged to a remarkable degree, confounding the taxonomy of these ants. Based on multiple lines of evidence, including worker and male morphology, we demonstrate that: (1) three species of “Camponotus” belonging to the subgenus Myrmotemnus, including its type species, are in fact members of the genus Colobopsis; (2) four species previously assigned to Colobopsis belong to the subgenus Myrmamblys of Camponotus; and (3) three Nearctic taxa recently placed in Colobopsis are members of the genus Camponotus and closely related to Camponotus clarithorax. These taxonomic findings yield the following new or revived combinations: Colobopsis moeschi (comb. nov.), Colobopsis moeschi lygaea (comb. nov.), Colobopsis nutans (comb. nov.), Colobopsis nutans cleliae (comb. nov.), and Colobopsis reichenspergeri (comb. nov.); Camponotus apostemata (comb. nov.), Camponotus aurelianus (comb. rev.), Camponotus cavibregma (comb. nov.), Camponotus horrens (comb. rev.), Camponotus politae (comb. rev.), Camponotus trajanus (comb. rev.), and Camponotus yogi (comb. rev.). A further consequence is the following generic synonymy (senior synonym listed first): Colobopsis = Myrmotemnus syn. nov., and Camponotus = Dolophra syn. rev. At the species level, we argue that Camponotus apostemata and Camponotus cavibregma are junior synonyms (syn. nov.) of Camponotus yogi, and Camponotus quercicola is a junior synonym (syn. nov.) of Ca. laevigatus. Taxonomic comments are also provided on some members of the Camponotus reticulatus group, with Camponotus adustus (stat. nov.) and Ca. leucodiscus (stat. rev.) being recognized as distinct species rather than subspecies of Ca. bellus. A male-based diagnosis of the Camponotini is provided, and differences between the males of Colobopsis and Camponotus are documented and illustrated for the first time. This study reveals new character systems of potential value to the systematics of these ants, including features of the male genitalia, and emphasizes the value of reciprocal illumination between phylogenomics and critical morphological analysis.


Introduction
Evolution is a heterogeneous process, occurring at variable rates in different lineages (Simpson 1953) and across different body structures (Hennig 1957). In hyperdiverse groups, such as ants, we find evidence for varying degrees of divergence from ancestral conditions. Large ant clades often contain a mixture of slower-evolving species that appear to have retained many original characteristics as well as highly divergent taxa that have evolved to the point where certain ancestral features are lost or indiscernible. Examples of this pattern include the "army ants" within the subfamily Dorylinae (Borowiec 2019), numer-ous genera of the Ponerinae (Schmidt 2013, Schmidt andShattuck 2014), and social parasites in the Myrmicinae (Rabeling et al. 2014, Prebus 2017. In addition to this variable rate of evolutionary divergence, ants also show a strong propensity for convergent evolution of certain features in the worker caste, particularly with respect to defensive traits such as spines (Blanchard and Moreau 2017), morphology of the major worker (Hölldobler and Wilson 1990), and chemical weaponry (Hermann and Blum 1981). These evolutionary dynamics pose considerable challenges to ant systematics and, in particular, to the establishment of a ranked, phylogenetic classification (Ward 2011).
The two ant genera that are the subject of this paper, Camponotus Mayr and Colobopsis Mayr, exemplify this situation. The latter genus was established for those taxa whose major workers have markedly truncate (phragmotic) heads, used for blocking nest entrances (Mayr 1861). As species of Camponotus were discovered with similar phenotypes, however, the morphological justification for retention of the two genera appeared to weaken (Fig. 1). Eventually, Colobopsis was treated as a subgenus of Cam ponotus (e.g., Emery 1925, Bolton 2003. Recent molecular studies, employing UCE (ultra-conserved element) phylogenomic data, have demonstrated that Colobopsis is a phylogenetically distinct group, considerably distant from Camponotus, and sister to all other members of the tribe Camponotini (Blaimer et al. 2015). As a consequence, Colobopsis was resurrected as an independent genus, and 94 species were transferred from Camponotus to Colobopsis (Ward et al. 2016). Attempts to produce a simple worker-based diagnosis of both genera proved to be difficult, however, because of the large amount of variation observed in both clades, compounded by a confusing blend of convergent and divergent evolution.
Here we provide evidence that additional changes are needed to the taxonomy of these two genera. Specifically, certain species that are currently placed in Camponotus in fact belong to Colobopsis, and vice-versa. In this paper we justify these changes, clarify the species-level taxonomy of several taxa, and document, for the first time, diagnostic features of male Camponotini and differences between males of the two genera.

Methods
This study is based on direct examination of specimens in collections, scrutiny of images on AntWeb (https://www. antweb.org), AntWiki (https://www.antwiki.org), and MCZbase (https://mczbase.mcz.harvard.edu), and interrogation of the original taxonomic literature. Images of male genitalia were taken using a JVC KY-F57U digital camera mounted on a Leica MZ 16A microscope, with resultant z-stacks processed via Auto-Montage Pro (Synoptics Ltd., Cambridge, England), Adobe Photoshop 2020, and Adobe Illustrator 2020 (Adobe Systems Inc., California, USA).

Morphometrics
The following metric measurements and indices are employed for workers (see also Ward et al. 2016): HW Head width: maximum width of head, excluding the eyes.

HL
Head length: midline length of head from the anterior clypeal margin to a line drawn across the posterior margin of the head (medial indentations on either margin do not decrease length).

SL
Scape length: length of first antennal segment, excluding the basal constriction. WL Weber's length: length of mesosoma, taken in lateral view from the anterior margin of the pronotum, excluding the pronotal collar, to the posteroventral extremity of the metapleuron. ASM Minimum distance between the antennal sclerites (inter-torular distance). CLW Clypeus width: width of clypeus, taken at the anterior tentorial pits. CLL Clypeus length: maximum measurable length of clypeus, taken along the midline, in an anterodorsal view, from a line drawn across posterior margin to a line across the anterior margin (medial indentations on either margin do not decrease length).

New generic combinations
In effecting these changes in generic assignment, we are guided by the differences in worker morphology uncovered in Ward et al. (2016), which are corroborated by character differences in the larvae and pupae, and by molecular phylogenetic data (Wernegreen et al. 2009, Blaimer et al. 2015, Clouse et al. 2015

Transfers from Camponotus to Colobopsis
The Camponotus subgenus Myrmotemnus Emery currently contains five nominal species and two subspecies, all restricted to the Indomalayan region. The worker caste is characterized by having a strongly impressed metanotal groove, raised dorsal face of the propodeum, and compound eyes placed in a relatively posterior position on the head (Emery 1925, Santschi 1926. Examination of the type species of this subgenus, Ca. moeschi Forel, shows that it is clearly a species of Colobopsis: the antennal insertions are well separated (ASM/HW 0.43-0.44, ASM/CLW 0.82-0.89) and occur at about the midlength of the frontal carinae, and the clypeus is relatively narrow (CLW/CLL ~1.12) ( Fig. 2A). This is observed in material collected recently in Sabah, Malaysia (CASENT0863455) and in a syntype worker from Sumatra illustrated on AntWeb (CASENT0910546). Accordingly, this species becomes Colobopsis moeschi (Forel) comb. nov., and Myrmotemnus is a junior synonym (syn. nov.) of Colobopsis.
The two remaining species associated with Camponotus (Myrmotemnus) are retained in Camponotus. Camponotus hypoclineoides Wheeler has the antennal insertions relatively closely positioned (ASM/HW 0.33) and anterior to the mid-length of the frontal carinae (Fig. 3A). It is reassigned to the subgenus Karavaievia Emery, where it was placed by Santschi (1926: 601). This is also consistent with Wheeler's (1919) assertion that Ca. hypo clineoides is related to Ca. dolichoderoides Forel, a current member of subgenus Karavaievia (Dumpert et al. 2006). Camponotus impressilabris Stitz also shows the frontoclypeal configuration typical of Camponotus (ASM/HW 0.29, ASM/CLW 0.63) (Fig. 3B), and is here assigned to Camponotus subgenus Orthonotomyrmex Ashmead, based on structural features shared with other species in that subgenus: a bidentate propodeum, nodiform petiole, and matte integument.

Transfers from Colobopsis to Camponotus
The Camponotus subgenus Myrmamblys Forel contains a diverse array of species, found mostly in the Indo-Australian region (Emery 1925, Bolton 1995. The workers are small to medium in size, and worker polymorphism is pronounced. The soldiers usually have the head longer than wide, and often obliquely truncate (with the truncation encompassing the entire clypeus), while the head of the minor worker is broader with more rounded sides. Within this subgenus, Emery (1925) recognized a group of species that he called the Ca. reticulatus group, and which he characterized as follows: mesosoma dorsum of worker continuous or interrupted, but not constricted in front of the propodeum; and dorsum of propodeum often saddle-shaped in profile but neither marginate nor compressed. Some members of the Ca. reticulatus group superficially resemble Colobopsis, as a consequence of their small size, propodeal profile, and soldiers with partially phragmotic heads. We have discovered the following four species, currently placed in Colobopsis, which actually belong to Camponotus (Myrmamblys) and which have affinities to the Ca. reticulatus group. Previous molecular work has confirmed that the Ca. reticulatus group is part of Camponotus (Wernegreen et al. 2009, Blaimer et al. 2015. Camponotus (Myrmamblys) horrens Forel (comb. rev.) (Fig. 4A, B) has closely placed antennal insertions (ASM/ HW 0.25) and a broad clypeus, which preclude its place-

ment in Colobopsis.
Little is known about this curious species; Forel (1910) surmised that it is a mimic of Myrmicaria brunnea. Placement in the Camponotus reticulatus group is provisional.
Camponotus (Myrmamblys) politae (Wu & Wang) (comb. rev.) was originally described in its own genus, Dolophra, later assigned to Camponotus (Bolton 1995), then to Camponotus subgenus Colobopsis (Bolton 2003), and finally to Colobopsis (Ward et al. 2016). The illustration of the worker head in the original publication (Wu and Wang 1994: 36) indicates that this is a Camponotus species, given the closely approximated antennal insertions (ASM/HW ~0.23) and broad clypeus. Moreover, the elongate nodiform petiole and the shape of the propodeum, with concave dorsal and declivitous faces, place it close to Camponotus (Myrmamblys) bellus and related species, in the Ca. reticulatus group. Consequently, Do lo phra again becomes a junior synonym of Camponotus (syn. rev.).
Turning to the Nearctic fauna, there are three taxa recently assigned to Colobopsis by Mackay & Mackay (2018) that are members of the genus Camponotus, and close to the nominate subgenus. One of these is Camponotus yogi Wheeler, long considered to be a bona fide Camponotus, which was transferred to Colobopsis on the basis of superficial similarity; it is here returned to Camponotus (comb. rev.). The major workers of Ca. yogi have obliquely phragmotic heads (Fig. 5B, C), but these are quite unlike those of New World Colobopsis (Creighton and Snelling 1967). The minor workers are very similar to those of Camponotus clarithorax. These smaller workers of Ca. yogi (HW 0.98-1.22, WL 1.67-1.95) have the traits typical of Camponotus rather than Colobopsis, i.e., closely approximated antennal insertions (ASM/HW 0.30-0.34), that are placed anterior to the midlength of the frontal carinae, and a broad clypeus (Fig. 5A). Unlike Colobopsis, which has naked pupae, those of Camponotus yogi are enclosed in cocoons (Creighton and Snelling 1967) and the larvae lack the praesaepium, or ventral pocket, that is diagnostic of Colobopsis. Phylogenomic (UCE) data also support the placement of Ca. yogi in Camponotus, and show it to be closely related to Ca. clarithorax (Ward, unpublished).
The other two taxa placed incorrectly in Colobopsis by Mackay & Mackay (2018) were described as new species and are here transferred to Camponotus: Ca. apostemata (Mackay) (comb. nov.) and Ca. cavibregma (Mackay) (comb. nov.). These two are very similar to Camponotus yogi and, we argue below, are justifiably treated as junior synonyms of that species. All three names are here placed in the nominate subgenus of Camponotus.

Figs. 5, 6
Camponotus yogi Wheeler, 1915: 420 1. Based on the original description and figures (Mackay and Mackay 2018: 111-113), the holotype of Camponotus cavibregma appears to be simply a queen of Ca. yogi. Its supposedly distinctive feature-a concave genal area, free of the short spatulate hairs that are common on surrounding anterior regions of the head-is observed also in queens of Ca. yogi (Fig. 5D). The description and illustrations of the paratype minor worker of Ca. cavibregma (Mackay and Mackay 2018: 109-111) similarly place it within the range of variation exhibited by minor workers of Ca. yogi. Camponotus apostemata, described from a series of workers collected in northern Baja California (specimens from this series examined in LACM), is scarcely distinguishable from Camponotus yogi, and is here treated as part of the geographical variation of the latter species. The head of the major worker is a bit more strongly truncate than in populations farther north, but no consistent differences are seen in the minor workers. Johnson & Ward (2002) referred to these and other samples of Ca. yogi from Baja California as Camponotus sp. cf. yogi.
Camponotus yogi is closely related to Ca. clarithorax Creighton. The major worker of the latter species lacks a pitted, obliquely truncate head, but is otherwise structurally similar. The two species can be distinguished by differences in scape and leg length, with Ca. yogi having consistently shorter appendages than Ca. clarithorax (Fig.  6), although very small workers may be difficult to distinguish. The median clypeal notch or concavity is better developed, on average, in Ca. clarithorax workers than in those of Ca. yogi, a feature which led to the placement of Ca. clarithorax in the subgenus Myrmentoma, but neither species belongs in that subgenus. They are here treated as Camponotus (Camponotus) since genetic data indicate a fairly close relationship to other species in that subgenus (Wernegreen et al. 2009;Ward, unpublished). In addition to Ca. clarithorax and Ca. yogi, there are two other species in the Ca. yogi group: Ca. keiferi Wheeler, endemic to Isla Guadalupe, Mexico (and already placed in the nominate subgenus), and an undescribed species from the California Channel Islands (Ward, unpublished).
Camponotus yogi is endemic to California and northern Baja California, where it occurs in coastal sage scrub, chaparral, oak woodland, and oak-juniper woodland. Nests are located in dead branches or stumps of various plants, always near the ground and often extending into live plant tissue. Creighton & Snelling (1967) reported this species nesting in live beetle-bored stems of Ericameria pinifolia, and tending pseudococcids in the stems. 1. Mackay (2019) discovered that the types of this species in BMNH-a syntype dealate queen and syntype major worker, the latter designated by him as lectotypedo not correspond to the species that has come to be known in the literature as Camponotus laevigatus. That species, given the new name Ca. laevissimus Mackay, is easily recognized by its shiny, iridescent blue-black integument and abundant and bright white standing pilosity on most of the body, including the scapes and tibiae (Fig. 7E, F). The real Camponotus laevigatus is a shiny black species, with relatively sparse standing pilosity, inconspicuous pubescence, slender scape base, and ecarinate clypeus. Examination of the lectotype image (Fig.  7A, B) shows that Camponotus laevigatus is conspecific with Ca. quercicola, a widespread California species that nests in the trunks and branches of oak trees (Gadau et al. 1999) (Fig. 7C, D). Mackay (2019) claimed that Ca. laevigatus differs from Ca. quercicola in having reduced pilosity on the head, but the lectotype is an old specimen in which the hairs are evidently abraded. Note the asymmetry in presence of hairs on the two sides of the head in the Ant Web image (e.g., short setae present on the left malar region but not on the right side) (Fig. 7A). Moreover, the amount of standing pilosity shows considerable variation in workers of Ca. quercicola, including setation on the malar region (Smith 1954;Gadau et al. 1999). We have examined a large series of Camponotus quercicola from throughout California, and we find that the type of Ca. laevigatus falls easily within the range of variation exhibited by this species. Mackay (2019: 321) also stated that the male and queen of Ca. quercicola (now Ca. laevigatus) are unknown, but this is incorrect: they were described and illustrated by Gadau et al. (1999) and compared with related species.

Camponotus adustus Viehmeyer
stat. n. 1. Ca. adustus was described by Viehmeyer (1916) as a subspecies of Ca. bellus Forel, but examination of the original descriptions, images of types, and more recent material indicates that these two are distinct species. Camponotus bellus (syntype major worker, Amboina, Indonesia (Biró) [MHNG]; examined via image on Ant-Web: CASENT0910513) has a matte integument and more abundant standing pilosity on the mesosoma, including the pronotum (Fig. 8C, D). In contrast, the body of Ca. adustus is predominantly shiny and standing pilosity is sparse on the mesosoma, being restricted to a single pair of long setae on the mesonotum and one pair at the junction of the dorsal and declivitous faces of the propodeum (Fig. 8A, B). Moreover, the two taxa have been recorded co-occurring in Singapore and remaining distinct (Viehmeyer 1916). In addition to the foregoing pilosity characteristics, the minor worker of Ca. adustus has striking anterior and posterior protuberances on the dorsal face of the propodeum, and both the dorsal and declivitous faces are notably concave in lateral view (Fig. 8B). Under this general morphotype, however, there is a bewildering diversity of color forms whose taxonomic status is unclear. One of these, Ca. leucodiscus Wheeler, has also been treated as a subspecies of Ca. bellus, and is here raised to species (below), but its relationship to Ca. adustus remains to be clarified. The New Guinea species Camponotus weismanni Forel (syntype worker, Bismarck Archipelago; examined via image on AntWeb: FOCOL2297) might be a senior synonym of Ca. adustus, but the worker propodeum has less well-developed protuberances and a correspondingly less concave dorsal surface in profile.

Camponotus leucodiscus Wheeler
stat. rev. Camponotus ( 1. Workers associated with Ca. leucodiscus (i.e., matching the striking black and white color pattern on the gaster of the holotype queen) lack the matte integument and pilosity of Ca. bellus, and are instead shiny and with sparse pilosity, as described above for Ca. adustus. They apparently differ from workers of Ca. adustus by the longer, lower petiole and by the color pattern on the gaster. Both taxa that are here elevated to species are part of a larger assemblage of ants in the Ca. reticulatus group that need comprehensive taxonomic study.

Identification of male Camponotini
Diagnosis. Camponotini are well-defined morphologically based on the female castes (Bolton 2003). Males are identifiable as Formicinae by their long scapes, the strongly oblique gonocoxal-gonostylar articulation, ab- sence of constriction between the third and fourth abdominal segments, and failure of the clypeus to extend between the antennal toruli, among other features (see subfamily key in Boudinot 2015). Given the available sample, male Camponotini are distinguishable from those of other formicine tribes by the following combination of traits (Fig. 10): (1) antennal toruli posteriorly-situated (i.e., anterior margins of torular rims distant from posterior clypeal margin); (2) antennae 13-merous; (3) arolia grossly enlarged; (4) gonostyli usually distinctly digitate (finger-like in shape and proportions); (5) waist simple, i.e., (5a) petiolar node usually vertical (except, e.g., Ca. (Myrmopytia) longicollis, which lacks a node altogether), (5b) petiole is not elongate posteriorly (e.g., anterior and posterior faces of node subequal in length), (5c) ter-gosternal articulation of abdominal segment III (AIII) is unfused, (5d) AIII articulation not raised dorsally above helcium, and (5e) the anterior surface of abdominal tergum III is convex, without a median longitudinal groove for reception of the petiole when "gaster" flexed anteriorly; (6) in most species, the first free abscissae of the radial sector and media veins (Rsf1 and Mf1) are characteristically aligned, forming a more-or-less straight line, although they may be kinked at the juncture of Rs+M, or have some other curvature; in rare cases, e.g., Colobopsis pylora (alate gyne examined), the abscissae meet at a distinct angle; (7) fore wing crossvein 1m-cu is usually absent (although loss within the group may have occurred in parallel, see Remarks below); and (8) head with distinct shape, resembling an inverted pear in full-face view: (8a) posterior head margin broadly convex, (8b) posterior head margin continuous or nearly so with the strongly bulging compound eyes (rarely the head is posteriorly elongate, e.g., Camponotus gouldi), (8c) malar area from the compound eyes to the mandibular insertion in full-face view strongly narrowed lateromedially, usually with parallel to subparallel malar margins that are almost orthogonal to the anterior eye margin. Remarks on distinguishing the genera. Camponotus and Colobopsis are globally distinguished from one another in the key to males provided below (section 3.3.2), and are the only camponotine genera occurring in the New World. In the Old World, these genera can be confused with Calomyrmex, Dinomyrmex, Echinopla, Opisthopsis, Overbeckia, or Polyrhachis, for which differentiating features are noted below. In general, Colobopsis is the only genus among these with antennal toruli situated at midlength of the frontal carinae, although some male Camponotus can be hard to evaluate due to poor development of the carinae. Further scrutiny of this condition is necessary. Dinomyrmex males are readily identified by the following combination of states: (1) body massive, ~2 cm long;

Genera included. Calomyrmex
(2) head oddly shaped, with concave malar regions in full-face view; (3) propodeal spiracles long, slit-shaped; (4) petiolar node broadly wedge-shaped in profile view; (5) gonapophyses lateromedially flattened and weakly lobate; (6) golden pubescence present on pronotum; and (7) numerous long, reddish macrosetae present on pronotum, lateral mesonotum, and propodeum.  (2) frontal carinae usually robust, especially broad dorsoventrally dorsal to medial torular arch as seen in lateral view (orientation assuming prognathy), and often strong and well-marked; (3) third abdominal tergum often > 1/3 the total length of the gaster; and (4) helcial tergite elongate, with a very shallow notch or even an anteromedian lobe (e.g., in Polyrhachis sensu stricto), although the medial notch may be extremely long and narrow, reaching the helcial base, as in some Myrmatopa. None of the helcial states observed in Polyrhachis have been seen in Camponotus. While the genitalia and ninth abdominal sternum of Camponotus tend to be rather uniform, those of Polyrhachis vary considerably from species to species and subgenus to subgenus, in ways which are distinct from Camponotus and which deserve special attention.
The boundaries of Calomyrmex, Echinopla, Opisthopsis, and Overbeckia remain largely unexplored due to limited sampling. At least one species of Opisthopsis and one of Calomyrmex (in UCDC), and at least Colobopsis vitrea (male unknown) have the forewing crossvein 1m-cu enclosing and forming a discal cell. A discal cell is absent in Echinopla, Camponotus (including the recently demoted subgenus Phasmomyrmex), most Colobopsis, Overbeckia, and Calomyrmex (when 1m-cu present) may be distin-guished from one another by the shape of the discal cell, being isosceles-shaped in Opisthopsis and subrectangular in Calomyrmex; however, this should be validated with a broader taxonomic sample. The examined male of Opisthopsis, that of O. haddoni (MHNG), was observed to have an exceptionally sharp and long ventroapical point of the penial sclerite; this species also has small ocelli, a very shallow and short posterior head margin posterad the compound eyes, and a large and convex anterior clypeal lobe. Among Echinopla, only E. striata was available for examination; the male of this species lacks 1m-cu, has a short third abdominal tergum, has a posteriorly-truncate head as in Polyrhachis, and is extremely hairy with both standing pilosity and pubescence. The male of Overbeckia has short scapes which are shorter than the head length, very close-set antennal toruli (separated by slightly more than one torular diameter), a small clypeus without an anterior lobe, and a long head posterior to the compound eyes.

Diagnostic key for Camponotus and
Colobopsis males

Discussion and conclusion
The two ant genera that are the subject of this study are ecologically prominent, species-rich, and widely distributed, collectively occupying much of the planet's terrestrial landscapes (Emery 1925, Bolton 1995, Fisher 2009 (Feldhaar et al. 2007, Wernegreen et al. 2009, Rafiqi et al. 2020. Concomitant with the colonization of so many habitats, species of Camponotus and Colobopsis have undergone extensive morphological diversification. Certain arboreal taxa have come to occupy similar morphospace, especially with respect to the cranial architecture of the major worker. Phragmotic heads, serving to block the entrances of twig nests, have evolved-to varying degrees-in multiple lineages of both Camponotus and Colobopsis (Fig. 1). Such convergent evolution delights the evolutionist, but can prove frustrating to the taxonomist. Here we have examined several problematic taxa that were given insufficient attention in an earlier treatment (Ward et al. 2016), and reevaluated their taxonomic placement. We depend largely on the morphological distinctions adduced in the previous study (Ward et al. 2016) because they were supported by complementary evidence from larva and pupal characters, and corroborated by robust phylogenetic inference from phylogenomic data (Blaimer et al. 2015).
Our investigations and taxonomic changes (summarized in Table 1) have refined our understanding of these ants, revealing for example that one putative subgenus of Camponotus, Myrmotemnus, is in fact a subgroup of Colobopsis. Our study highlights the need for greater taxonomic attention to the Camponotus reticulatus group (in subgenus Myrmamblys), several species of which had been placed incorrectly in Colobopsis. We were also able to demonstrate that several "Colobopsis" taxa recognized in a recent revision of the New World species of this genus belong to the genus Camponotus. With these reassignments we feel that the composition of the two genera has largely stabilized, setting the stage for more thorough comparative analyses of trait evolution in these ants.
In contrast to the scarcity (and potential fickleness) of diagnostic features in the worker caste, our investigation has revealed several promising features of male morpho logy-specifically male genitalia-that serve to distinguish Camponotus and Colobopsis. Although males are understudied in Formicidae, they yield consistent and surprising distinguishing features among subfamilies, genera, and species groups (e.g., Ward 1999, Ward and Downie 2005, Boudinot 2015, Barden et al. 2017). Males may be less prone to homoplasy than workers and queens because they are not subject to the same ecological pressures due to their hermitic lifestyles. Moreover, the male genitalia of ants are complex copulatory machines which display considerable functional morphological variation. In some cases, the male genitalia appear to have undergone sexually-selected runaway evolution as observed Table 1. Summary of taxonomic changes in this paper. This includes novel subgenus placements; these are not new or revived combinations as defined by the ICZN.

Taxon
Change Notes in lineages such as the army ants (Old and New World), spider ants (Leptomyrmex), castrator ants (Diacamma), fungus-growing ants (Atta genus group), and legionary vampire ants (Leptanillinae). Among the examined camponotines, Co lo bopsis is uniquely defined by the synapomorphic loss of the apicoventral tooth of the penial sclerite, which implies concomitant behavioral derivation during copulation. We hope that the dissections figured in the present work encourage future studies of camponotine genitalia. With reciprocal illumination from burgeoning phylogenomic studies, the exploration of morphological variation in male, worker, and queen ants will inform our understanding of phylogeny and evolution for many years to come.

Acknowledgements
We thank the following curators for access to material in the indicated