Introduction

Males lead a rather shadowy existence in the matriarchies of ants. They rarely engage in social activities in the nest, are short-lived relative to queens and workers, and in many species make their most important contributions to the society during an unobservable nuptial flight, often high in the sky (Boomsma et al. 2005). The typical morphology myrmecologists associate with males includes large eyes and ocelli, a bulky thorax, a relatively short antennal scape, and well-developed wings. Exceptions to this generalization are rare, but they do occur in the form of wingless, ergatoid males that resemble workers or ergatoid female sexuals. Ergatoid males have only been described for a handful of genera (Loiselle et al. 1988; Heinze and Tsuji 1995; Alpert 2007; Boudinot et al. 2016) and are probably studied best in the myrmicine genus Cardiocondyla.

Ergatoid Cardiocondyla males are morphologically so aberrant that they have repeatedly duped eminent myrmecologists into describing them as females of new workerless socially parasitic taxa (Fig. 1a–c; Baroni Urbani 1973; Marikovsky and Yakushkin 1974; Seifert 2002): Forel (1890) described a wingless individual with “soldier-like” mandibles found in a colony of C. wroughtonii (Forel 1890) as a separate parasitic species, Emeryia wroughtonii. Two years later, he corrected his error after realizing that wingless individuals in colonies of another congeneric species—C. stambuloffii Forel 1892—were ergatoid males (Forel 1892a, b). Indeed, ergatoid males of C. wroughtonii and its sister species C. obscurior Wheeler 1929 (Fig. 1a) are known to use their long, sabre-shaped mandibles to eliminate rival males from their natal nests and in this way monopolize mating with all newly emerging female sexuals (Kinomura and Yamauchi 1987; Stuart et al. 1987; Heinze and Hölldobler 1993; Heinze 2017).

Fig. 1
figure 1

Cardiocondyla males previously mistaken for female sexuals of new taxa of workerless, socially parasitic ants (photos a–c taken by S. Frohschammer, d by Ryan Perry, antweb.org: https://www.antweb.org/bigPicture.do?name=casent0901750&shot=p&number=1). a Ergatoid male of Cardiocondyla obscurior—the morphologically similar ergatoid male of its sibling species C. wroughtonii was described as Emeryia wroughtonii. Scale bar 0.5 mm. b Ergatoid male of Cardiocondyla venustula—a similar ergatoid male was described as Xenometra monilicornis. Scale bar 1 mm. c Ergatoid male of Cardiocondyla elegans; such a male has been described as Xenometra gallica. Scale bar 1 mm. d Holotype of Cardiocondyla zoserka Bolton 1982, a winged male of a species of the C. shuckardi group. Scale bar 1 mm

Full size image

Emery (1909) considered a large ergatoid male, perhaps of a species belonging to the C. shuckardi Forel 1891 group (Seifert 2002; Fig. 1b) as a wingless Cardiocondyla queen and later placed it into the new, supposedly parasitic genus Xenometra (Emery 1917). Later, the ergatoid male of C. elegans Emery 1869 (Fig. 1c), already recognized as such by Menozzi (1918), was described as the social parasite Xenometra gallica (Bernard 1957). These errors were corrected by Baroni Urbani (1973) and also by Marikovsky and Yakushkin (1974). The latter authors, however, made a similar mistake by identifying an ergatoid Cardiocondyla male as a “male-like wingless queen” of C. ulianini Emery 1889 (Seifert 2002).

Here I show that the morphology of fully winged Cardiocondyla males can also give rise to erroneous species descriptions. C. zoserka was described by Bolton (1982) based on five winged individuals collected by E. Classey in 1972 at Gurara Falls, Abuja, Nigeria (Fig. 1d), the only known specimens of this taxon. Bolton (1982) considered them to be winged female sexuals and from the strong apical teeth of their mandibles and the unique morphology of their antennae suspected C. zoserka to be a workerless social parasite (inquiline) living in the nests of C. shuckardi. Based on samples collected 1180 km further west in Comoé National Park, Côte d’Ivoire, and the inspection of the holotype and one paratype of C. zoserka, the supposed socially parasitic females are in fact the regular winged males of the colonies in which they were found.

Methods

In April 2019, fragments of approximately 50 colonies of Cardiocondyla venustula Wheeler 1908 and other species belonging to the Cardiocondyla shuckardi group (sensu Seifert 2002) were excavated from their nests in the floodplain of Comoé River near Comoé National Park Research Station, Côte d’Ivoire, West Africa (8° 46′ 2′′ N, 3° 47′ 03′′ W, ca. 195 m elevation) and a few other sites in Comoé National Park (for details see caption of Supplementary Fig. S1). Foragers of Cardiocondyla were active throughout the day even at air temperatures of more than 35 °C and soil temperatures of more than 50 °C. In the hottest places, foragers over several meters moved from one blade of grass to the next without touching the hot ground. Colonies were located by baiting foragers with cookie crumbs and following them to the entrances of their nests. Large numbers of workers, female and male sexuals, and brood were found in chambers in moist soil at depths between 15 and 30 cm on three elevated heaps of sand with Mimosa shrubs and small Ficus trees (Fig. 2), and occasionally also in wet sand under large pebbles in the floodplain.

Fig. 2
figure 2

Habitat of Cardiocondyla ants near Comoé Research Station, Côte d’Ivoire. The small red flag indicates the site of a nest and the insert shows how colonies were excavated

Full size image

In the laboratory of Comoé Research Station, samples were checked under a binocular microscope. Samples consisting only of workers were stored in 100% EtOH for genetic analyses (see below). Larger colony fragments with winged female sexuals, dealate queens, and/or males were transferred to the University of Regensburg and since then have been kept in plastic boxes, each with a regularly moistened plaster floor and a nest consisting of three parallel, 3 mm wide and 6 cm long slits in Plexiglas®, sandwiched between two microscope slides and covered by black foil. This nest design allows the males, which in C. venustula and possibly also other species of the C. shuckardi group are territorial (Frohschammer and Heinze 2009; Jacobs and Heinze 2017), to spread out. Colonies were reared in incubators at 12 h 23 °C/12 h 28 °C day–night cycles, provided with honey and pieces of cockroaches twice per week and checked regularly for newly emerging sexuals using a Zeiss Stemi 508 binocular microscope at 10 to 40× magnification.

To compare the specimens from Côte d’Ivoire with previously collected C. venustula, C. shuckardi, and other species of Cardiocondyla, we analyzed sequences of the mitochondrial cytochrome c oxidase subunits I (CO I) and II (CO II) from one worker each from 38 colony fragments from Côte d’Ivoire. DNA was extracted using a CTAB protocol (Sambrook and Russell 2001), amplified by PCR using the primer combination C1-J 2183/C2-N-3661 (Simon et al. 1984) and an initial denaturation step at 94 °C for 3 min followed by 36 cycles with 45 s at 94 °C, 30 s at 50 °C, and 45 s at 72 °C, and a final step of 10 min at 72 °C. PCR products were purified in a Macherey–Nagel gel, extracted using the NucSpin extraction kit (Macherey–Nagel, Düren, Germany) and sequenced by LGC Genomics (Berlin, Germany).

Sequences of CO I/CO II (up to 1457 bp) of ants from Côte d’Ivoire and previously obtained partial sequences of CO I (mostly 814 bp) of C. shuckardi, C. venustula, and several other congeneric species (Heinze et al. 2013; Jacobs and Heinze 2019) were aligned with the software BioEdit v 7.09 (Hall, 1999) using the CLUSTAL W Algorithm (Thompson et al., 1994) and then manually corrected. For the visualization of sequence similarities among colonies, a neighbor-joining tree was constructed using Mega X (Kumar et al. 2018) following the Tajima–Nei method (Tajima and Nei 1984). In the tree shown in the Supplementary material (Fig. S1), branches reproduced in less than 60% of 500 bootstrap replicates are collapsed.

To document the peculiar modifications of the antenna of winged males, three antennae from three different winged males, two antennae from two workers, two from two female sexuals from colony CI 15, and one antenna from an ergatoid male of C. venustula (colony CI 51) were gold–sputtered using a Polaron SC515 SEM coating system (Fisons, Ipswich, UK) and examined in ventral view using a JSM-IT 100LV scanning electron microscope (JEOL, Tokyo, Japan) at 100–4300× magnification. Nine winged males (four from CI 15, five from CI 30) and 12 winged or dealate female sexuals (6 each from colonies CI 15 and CI 30), which eclosed in the lab from brood collected in the field and died in the laboratory nests, were mounted. Mesosoma length, head width, and head length were measured using a Wild M10 microscope at 100× to 125× magnification. In four males and five female sexuals from CI 15, the diameters of eyes and median ocelli were measured using a Keyence VHX-500FD digital microscope at 200× magnification. As several individuals had already started to decay and were partly overgrown by mold, not all measures could be taken with the same accuracy in all individuals. This resulted in a relatively large measurement error of approximately 2% (head width of a male measured three times). Voucher specimens are stored in Senckenberg Museum of Natural History Görlitz, Germany.

Results

Most of the samples collected at Comoé N.P. were similar to C. venustula both based on morphology and sequences of the cytochrome c oxidase subunits I (CO I) and II (CO II) (in black in Supplementary Fig. S1). All males found in these colonies upon collection and later in the laboratory were the typical ergatoid or, in one colony, also intermorphic males as previously described for C. venustula (Fig. 1b; for details see last paragraph of results).

Individuals from two additional colony fragments (CI 15, CI 37) from two sand heaps in the floodplain of Comoé River differed in their sequences in about 10% (8.77–10.7%) from sequences of sympatric C. venustula from Comoé N.P., C. venustula from Puerto Rico, Kaua’i, Ethiopia, Egypt, Kenia, and South Africa, and C. shuckardi from Madagascar. In a NJ tree, they formed the outgroup to these sequences (in green in Supplementary Fig. S1). Of CI 37, only workers were collected and stored in EtOH. Colony fragment CI 15 and a second colony fragment from the same sand heap, CI 30 (no sequence available), in addition to workers, queens, brood, and/or winged female sexuals contained two winged individuals each (Fig. 3a, b), which differed consistently from nestmate queens and winged female sexuals (Fig. 3c, d) in the morphology of their antennae and the larger size of their ocelli and eyes. Instead, they closely resembled the description and photos of the type material of C. zoserka Bolton 1982 (Fig. 1d) in all characteristic features of the latter, in particular the bizarrely modified antennae, in which funicular segments 6 and 7 are dorsoventrally flattened, segment 8 is broad and slightly transversely concave, segments 9 and 10 are strongly transversely concave, and the most apical segment is swollen in dorsal view and invaginated and almost cup-shaped in ventral view (Figs. 3b, 4; for a more detailed, very accurate description see Bolton 1982).

Fig. 3
figure 3

Morphology of winged males and female sexuals of Cardiocondyla colony CI 15 from Comoé National Park, Côte d’Ivoire. Like the types of Cardiocondyla zoserka, the winged male has peculiar antennae. a Lateral view of a winged male. The evaginated male genitals are clearly visible. b Frontal view of the head of the same winged male, showing the peculiar antennal modifications. c Lateral view of a dealate female sexual. d Frontal view of the head of a winged female sexual, showing the typical antennal structure of Cardiocondyla with a distinct three-segmented club. The scale bars indicate 250 µm. Photos were made using a Keyence VHX-500FD digital microscope at a magnification of ×100 for lateral views and ×150 for frontal views; the sharpness of the composite image was increased using the Photoshop® sharpening filter

Full size image
Fig. 4
figure 4

SEM photo of the antenna of a winged male of Cardiocondyla from Comoé National Park, Côte d’Ivoire, in ventral view (×200), showing pores on the ventral side of funicular segments 7–10. The pores in the inner side of the cup-shaped apical segment 11 cannot be seen from this angle. The insert shows pores on the apical side of funicular segment 10 at a magnification of ×4300. Photo: B. Lautenschläger

Full size image

A closer analysis of the winged individuals with bizarre antennae revealed the presence of male genitalia, i.e., they were not female sexuals but winged males. Similarly, all 30 additional winged ants with such antennae, which eclosed in the laboratory during 7 months after collection, turned out to be males. No other males, neither wingless nor winged, were reared in colonies CI 15 and CI 30. As is typical for both winged and ergatoid males of Cardiocondyla (Kugler 1983), their genitals are often concealed, which leads to misidentifying them as female sexuals. Nevertheless, the tips of the parameres were visible in all winged individuals with modified antennae, and the genitals were evaginated in a few of them (see Fig. 3a).

The presence of male genitals in the specimens from Comoé N.P. spurred a closer look at the type material of C. zoserka. Judging from photos (Fig. 1d), the holotype is a winged male and not a female sexual: the ocelli are much larger than in female sexuals of Cardiocondyla and the rounded tip of the gaster suggests the presence of partly retracted parameres (for the same opinion by Bernhard Seifert, Senckenberg Museum of Natural History, Görlitz, see discussion). Stefan Cover (in litt.) kindly examined the C. zoserka paratype deposited at the Museum of Comparative Zoology, Cambridge, Mass., and confirmed it to be a male. Unfortunately, in the paratype loaned from the British Museum of Natural History by courtesy of Suzanne Ryder the abdominal tip was shrunken and the genitals were invisible. Nevertheless, its overall morphology closely matched that of the males from Comoé N.P. and the two types of C. zoserka identified as males. According to Bolton (1982), the two remaining paratypes of C. zoserka are also similar in morphology to the holotype. Hence, it appears that all previously known specimens of C. zoserka are winged males and not winged female sexuals.

Winged males from Comoé N.P. were slightly smaller on average than the C. zoserka types (Table 1). Their head shape was similar to that of the types and appeared slightly more elongated than that of female sexuals (Fig. 3b, d). The relative diameters of their eyes and median ocelli were considerably larger than those of female sexuals. Bolton (1982) considered the enlarged apical mandibular teeth of the mandibles in the types of C. zoserka as indicative of a socially parasitic lifestyle. The winged males from Comoé N.P. share this character, but so do the female sexuals of this taxon and the mandibles are not strikingly different in ergatoid males, female sexuals, and workers of other species of the C. shuckardi group on antweb.org.

Table 1 Measurements of winged males and female sexuals of Cardiocondyla from colonies CI 15 and CI 30 from Comoé N.P. For comparison, data from the description of the types of C. zoserka (Bolton 1982) are given
Full size table

The most striking character of C. zoserka and the winged males from Comoé N.P. is the uniquely modified antenna (Figs. 3b, 4). Workers and female sexuals both from the colonies with winged males and from colonies of C. venustula had the typical antennal structure of Cardiocondyla with a distinct three-segmented club (Fig. 3d). SEM photos of the antenna of three winged males revealed the presence of numerous “pores” in the ventral surface of funicular segments 7–10 and the bottom of the apical cup (Fig. 4), but the exact nature of these pores awaits further analyses. Neither the antennae of conspecific female sexuals and workers nor of an ergatoid male of C. venustula from Comoé N.P. had such pores.

CO I/CO II sequences of the winged male and two nestmate workers of CI 15 were identical except for different sequence lengths, suggesting that the winged males are not a separate socially parasitic species but the regular males of the colony in which they were found. Colonies CI 15, CI 30, and CI 37 from Comoé N.P. might therefore be considered as colonies of C. zoserka.

Concerning C. venustula from Comoé N.P., upon collection 18 of 19 larger colony fragments contained winged female sexuals and/or dealate queens, workers and brood. In addition, 9 of these 18 colony fragments and one fragment without female sexuals contained one to seven (median three) large and robust ergatoid males with a light brown coloration and angular shoulders typical of this species (Fig. 1b, Frohschammer and Heinze 2009; Heinze et al. 2013, 2014). One colony fragment, CI 62, had more than seven “intermorphic” males, i.e., ergatoid males with the typical morphology of ergatoid males but in addition with rudimentary wings as previously described for C. venustula from South Africa (Heinze et al. 2013). Another fragment only had workers, a single, ergatoid male, and no queens. All of these colony fragments produced numerous female sexuals and ergatoid males (in the case of CI 62 together with intermorphic males), but not a single winged male during 7 months in the laboratory. Casual observations suggested that C. venustula males spread out in the provided multi-chambered nests and only rarely engaged in aggressive interactions, as previously observed (e.g., Frohschammer and Heinze 2009; Heinze et al. 2013; Jacobs and Heinze 2017).

Discussion

Compared to the males of other social Hymenoptera, the wingless, ergatoid males of Cardiocondyla ants are quite exceptional. Their fights are one of the most striking cases of regular lethal combat in animals (e.g., Kinomura and Yamauchi 1987; Stuart et al. 1987; Heinze and Hölldobler 1993; Heinze 2017), and their lifelong spermatogenesis makes them unique among social Hymenoptera (Heinze and Hölldobler 1993; Boomsma et al. 2005). Their aberrant “worker-like” morphology has repeatedly misled researchers to describe them as female sexuals, living as social parasites in Cardiocondyla colonies.

As the present study shows, winged Cardiocondyla males may also be misleading: the “winged female sexuals” of Cardiocondyla zoserka Bolton 1982, a presumed workerless social parasite known only from the type material, are in fact the regular winged males of their supposed host, a non-parasitic species of the C. shuckardi group. This conclusion is based on the morphological similarities between the type material and winged Cardiocondyla males collected in two colonies from Comoé N.P., Côte d’Ivoire. Winged males from Comoé N.P., but neither their female sexual nor their worker nestmates, show the same unique modifications of the antennal funiculi as the C. zoserka types. According to Bolton (1982), “no other ant has funiculi even remotely resembling this one, and certainly they [C. zoserka] cannot be confused with any other member of Cardiocondyla.” The winged individuals from Comoé N.P., though slightly smaller than the C. zoserka types, also show other distinctive characteristics of C. zoserka, e.g., relatively large ocelli and an enlarged apical mandibular tooth (Fig. 3b). Nevertheless, the examination of the abdominal tip of more than 30 of such individuals revealed that they all were winged males, even though their genitals were typically concealed and could not be seen easily.

The inspection of photos of the holotype of C. zoserka (Fig. 1d) and the direct examination of the paratype of C. zoserka in the Museum of Comparative Zoology by S.P. Cover suggested the presence of male genitals in both specimens, verifying that they are also males. According to Bernhard Seifert (in litt), head proportions as in the holotype of C zoserka “do not occur in any Cardiocondyla female worldwide.” He also found that the diameter of the lateral ocelli as concluded from the photo of the holotype is about 10% of head length, while it is around 3% in female sexuals of Cardiocondyla, and that the relative size of the compound eye is also much too large and it is much more vaulted than seen in female sexuals. Finally, B. Seifert (in litt.) notes that “the lateral view of the gaster does not indicate the presence of a stinger. The barely visible terminal structure in the holotype is rather blunt and should represent the tip of the retracted paramere(s).” As all five hitherto known specimens of C. zoserka are morphologically very similar (Bolton 1982), it can safely be concluded that they all are winged males of a species of the C. shuckardi group and not female sexuals of a workerless social parasite.

The taxonomy of the C. shuckardi group is in need of a thorough revision. Seifert (2002) synonymized several taxa with C. venustula and C. shuckardi, but sympatric samples of C. venustula vary so widely in gene sequences (Heinze et al. 2013, Fig. S1) that more detailed morphological and genetic analyses are needed to fully resolve species borders. Such a revision will also clarify if the colonies from Comoé N.P. are indeed C. zoserka or a separate, related species. Gene sequences of a winged male and two workers from colony CI 15 show that they are closely related to the C. shuckardi group but differ from previously studied C. venustula and C. shuckardi in about 10% of the sequence of the cytochrome c oxidase subunits I (CO I) and II (CO II).

While phylogenetic analyses suggest that a polyphenism with both winged and ergatoid males is ancestral in Cardiocondyla (Oettler et al. 2010; Heinze 2017), winged males have been lost in most species including all previously studied taxa of the C. shuckardi group (Seifert 2002; Frohschammer and Heinze 2009; Heinze et al. 2013, 2014; Jacobs and Heinze 2017; Heinze 2017). Not surprisingly, dozens of inspected males from colonies of C. venustula from Comoé N.P. were all ergatoid or, in one colony also “intermorphic”, i.e., combining the typical ergatoid morphology of head and mesosoma with rudimentary wings. Similar intermorphs have been observed in C. venustula from South Africa (Heinze et al. 2013) and in the C. nuda group (sensu Seifert 2002), e.g., C. kagutsuchi from Malaysia (Yamauchi et al. 2005) and C. nuda from Australia (unpublished observations). The dealate “winged male” of C. nigra described by Santschi (1907) also appears to be such an intermorph (Kugler, 1983). Interestingly, in contrast to all other previously studied species of Cardiocondyla, the two colonies from Comoé N.P. tentatively assigned to C. zoserka have so far produced more than 30 winged males and not a single ergatoid male.

The most striking character of the C. zoserka types and the winged males from Comoé N.P. is the extremely aberrant morphology of their antennae (Bolton 1982, Fig. 3b), which makes these ants immediately recognizable. Female sexuals (Fig. 3d) and workers from the colonies with winged males, and also previously studied ergatoid males of other Cardiocondyla species have antennae with a distinct three-segmented club (e.g., Kugler 1983; Seifert 2002; Heinze 2017). Winged males of C. obscurior, C. minutior, C. emeryi and others typically have antenna without any conspicuous modification (e.g., Kugler 1983). Scanning electron microscopy of the antennae of winged males from Comoé N.P. revealed “pores” in the ventral surface of funicular segments 7–10 and the bottom of the apical cup (Fig. 4b). Such pores were absent from the antennae of conspecific female sexuals and workers and also from the antenna of an ergatoid male of C. venustula. The pores might suggest the presence of antennal glands. Male antennal glands are known from several Hymenoptera and probably play a role during mating (e.g., Isidoro et al. 1999; Romani et al. 2003, 2008). In ants, antennal glands have so far been found only in queens and workers of a few species (Renthal et al. 2008). During courtship and mating, Cardiocondyla males touch the head and body of female sexuals with their antennae (Mercier et al. 2007), and the presence of pores on the ventral surface of the funiculus of the winged males from Comoé N.P. suggests an exciting new perspective of sexual communication in this genus.