Ocymyrmex

Bolton and Marsh (1989) - The thermophilic ant genus Ocymyrmex is restricted in distribution to parts of the Afrotropical zoogeographical region. Members of the genus are all sun-loving, extremely fast-moving ants characteristic of arid or well-drained ground. They range through dry savannah to extreme desert conditions in the southern and eastern parts of the region, but are absent from the rain forest zones of West and Central Africa. The genus is also absent from the Sahelian zone of the southern Sahara except in the extreme east, and it does not penetrate the Arabian Peninsula (Collingwood, 1985) nor Madagascar (Wheeler, 1922).

Identification
Bolton and Marsh (1989) - The genus Ocymyrmex is easily isolated from all other Myrmicinae by the following autapomorphic characters in the worker and ergatoid female castes.

1. Propodeal spiracles elongate, narrow and slit-like.

2. Mesothoracic spiracles dorsal, open and visible in dorsal view.

3. Third mandibular tooth, or usually teeth 3 and 4, double-ranked internally on masticatory margin.

4. Reproductive female (queen) extremely ergatoid and physogastric; uninseminated ergatoids function like workers and occur outside the nest.

The worker/female habitus is also distinctive: a long-legged attenuated terrestrial myrmicine ant with a stockily constructed head, short powerful 4-5 toothed mandibles, a strongly developed psammophore, unarmed propodeum, and an elongate petiole segment.

Males have an advanced pheidoline form of wing venation (Bolton, 1982: 362) but with cross-vein m-cu absent and the radial ( = marginal) cell closed, extremely reduced and non-functional lobe-like edentate mandibles, and an elongate slit-like propodeal spiracle as in the worker and female castes.

We now strongly suspect that Ocymyrmex is related to Aphaenogaster, within the Pheidole-group of genera. The structure of the head and alitrunk, and the venation, provide the main clues for our association of Ocymyrmex with the pheidolines, although Wheeler and Wheeler (1973) have already pointed out that the larva in this genus is aphaenogastriform.

Species Groups
Ocymyrmex species groups

Biology
Until relatively recently remarkably little was known of the biology of these ants. Arnold (1916) observed that Ocymyrmex species with which he was acquainted nested in the ground in hot arid areas. The nests themselves went very deep into the ground, usually in loose sandy soil, and had a crater-like entrance. The ants used their well-developed psammophores to carry soil particles excavated from the nests. Recently both Marsh and Robertson (pers. comm.) have observed that workers of Ocymyrmex fortior close the nest entrance with small stones during periods of nest inactivity. Also, in Zimbabwe, fortior workers have been seen adding small stones to the crater-like nest entrance that were picked up from the ground some distance away from the nest. This suggests that the crater is not just a by-product of the nest excavation but an important feature of the nest entrance. Its function is not fully understood, but its presence may facilitate the deliberate blocking of the entrance noted above.

Species are now known which nest in very rocky soil and the nests may extend through the bedrock itself, necessitating the use of a large crowbar to expose the nest-chambers (H. Robertson, pers. comm.). Careful excavations of nests in well-structured sandy soil by one of us (Marsh) have revealed a simple nest structure. For example, nests of Ocymyrmex foreli typically have one entrance that opens into a vertical tunnel which terminates in a broad chamber at a depth of about 30 cm. Other brood and food chambers branch off from the tunnel at various intermediate levels. A similar structure exists for nests of Ocymyrmex picardi and Ocymyrmex sphinx except in these cases the terminal brood chambers are typically 2 m and 1.5 m from the surface respectively. In most nest excavations the ergatoid queen was discovered near the bottom of the nest. In very unstructured loose sand, such as in the dry river beds of the Namib Desert, the tunnels and chambers of Ocymyrmex nests followed the root systems of shrubs and trees, and the major tunnel was therefore not necessarily vertical. Colonies of Ocymyrmex range in size from 200 to 1000 individuals (Marsh, 1987).

Arnold (1916), and other observers since, noted that ants of this genus move very swiftly and erratically, faster than any other ants which they had seen, and that the ants were active in the hottest part of the day.

Prins (1963, 1965) recorded that Ocymyrmex species were granivorous but would also attack other insects. This is contradicted in part by Marsh (1986 b) who observed that the main food supply of the Namibian species Ocymyrmex robustior, Ocymyrmex turneri and Ocymyrmex velox, was obtained by scavenging dead or heat-stressed insects, although velox also preys on termites when these are available. Observations by Marsh indicate that Ocymyrmex picardi has a diet intermediate between these extremes; 70 per cent of their food is other arthropods, much of it live termites, and the remainder consists of plant material. More recently Forder and Marsh (1989) have kept captive colonies of O. foreli alive on a diet of water, sugar-water, cockroaches, termites and meal-worms. As Marsh (1985 b) points out, much more data on the feeding habits of the various Ocymyrmex species is required.

A series of papers, either specifically dealing with Ocymyrmex biology (Marsh, 1985 a, 1985 b), or on Ocymyrmex species as a component of the Namib Desert ant community (Marsh, 1985 c, 1986 a, 1986 b), has added much new information on the species of harsh desert environments. We should point out here that in all but the last of these papers the ant referred to as Ocymyrmex barbiger should properly be called Ocymyrmex robustior. At the times of publication of Marsh's first four papers robustior was incorrectly being treated as a junior synonym of barbiger. The former is now considered to be a valid species, separate from barbiger but belonging to the same species-group. The name robustior was elevated to species-level in Marsh (1986 b).

Marsh (1985 a) has shown that O. robustior is an individually foraging ant, diurnal and active on insolated ground at temperatures of 27~67°C. In the upper part of this range (> 51°C) the ants, which are very efficient heat-exchangers, lower their body temperatures when necessary. They do this either by pausing in shaded spots or by climbing any object that will take them temporarily off the desert floor. Efficiency in controlling their body temperatures makes these thermophilic scavengers very successful in capturing other, heat- or dessication-stressed, insects as prey (Marsh, 1985 b). However, being primarily scavengers, at least in harsh desert conditions where most ant species tend to be granivores or nectar-honeydew gatherers, Ocymyrmex species make up a relatively low proportion of the total ant population in terms of numbers of individuals and biomass (Marsh, 1985 c).

Reproductive females of Ocymyrmex, inseminated or virgin, were unknown until relatively recently. Arnold (1916) supposed that the females were ergatoid, or that maybe the worker caste took part in egg production, as he had never found an obvious (reproductive) female, or queen. At the same time, however, he was aware that in some nest-series of Ocymyrmex strange individual variants occurred which had cephalic sculpture radically different from the usual cadre of workers. These variants had sometimes been described as infraspecific taxa or even as separate species. It was apparent from material examined that these variant forms left the nest to forage, and in general behaved as the workers did.

Bolton (1981) determined that these oddly sculptured forms were in fact the extremely ergatoid, potentially reproductive, females of the nests. They constitute the most worker-like females yet encountered in the Myrmicinae. They are much more specialized than any ergatoid females known in Monomorium (DuBois, 1986; Bolton, 1987) and Ocymyrmex apparently lacks the strange combination of ergatoid females and reproductive workers found in at least one Pristomyrmex species (ltow, Kobayashi, Kubota, Ogata, Imai and Crozier, 1984). There are no series of inter castes between true workers and true reproductive females, such as are found in several leptothoracine ants (e.g. Francoeur, Loiselle and Buschinger, 1985). Where known, ergatoid females of Ocymyrmex species consistently differ from conspecific workers in the structure ofthe frontal lobes and the antennal scapes, and as discussed in Bolton (1981). In most species cephalic sculpture is usually radically different in workers and females, but this might not be so obvious in lightly sculptured species.

Recent dissections by Forder and Marsh (1989) of Ocymyrmex foreli, Ocymyrmex picardi, Ocymyrmex flaviventris, and Ocymyrmex sphinx, have shown that the ergatoid females always have larger ovaries and many more ovarioles than the workers. One of us (Bolton) has independently observed this also to be the case in Ocymyrmex nitidulus, Ocymyrmex ignotus and Ocymyrmex alacer. Forder and Marsh have also shown that in O.foreli, although numerous ergatoid females are present in the colony, and are produced throughout the year, only one is inseminated, so that the colony is functionally monogynous. The ovaries of the single inseminated female are much larger than in unmated females, so much so that the nest's fecund female is markedly physogastric. Uninseminated ergatoid females behave to a large extent like the workers, and Forder and Marsh suggest that if an ergatoid female remains a virgin beyond a certain undetermined age she switches roles and begins to behave as a worker. They found that 4-20 per cent of the total adult colony membership (excluding males) consisted of unmated ergatoid females with a worker-like, or mostly worker-like, behavioural repertoire. These virgin ergatoid females, however, always have much better developed reproductive organs than the true workers. Work on eight species in southern Africa by Marsh (1987) has revealed that the proportion of reproductively inactive ergatoid individuals in the extra-nidal workforce of a colony ranges from zero to 58•5 per cent.

The question arises: can one or more of these unmated ergatoids revert from a worker-like condition upon the death of the nest's single fecund female, mate, and take over the role of the dead female? This aspect of ergatoid behaviour in Ocymyrmex remains to be investigated in detail, but preliminary experiments involving the removal of the fecund female from laboratory colonies suggest that role reversal does not occur in O.foreli (Forder and Marsh, unpublished).

Recent observations at Tosca in the northern Cape, South Africa, by one of us (Marsh) indicate that multiplication of colonies is by fission (hesmosis) of an existing nest. In December 1986 a considerable amount of recruitment in one direction was observed from one nest of picardi. Investigation revealed that the ants were relocating to another nest 150m away. The nest from which the ants were being recruited had a large nest disk and midden of arthropod remains and was clearly an old, well established nest. In contrast, the nest to which the ants were moving had a smaller disk and no midden, and appeared to be more recently excavated. Traffic between the two nests continued for a day and thereafter ceased, and normal foraging activity was seen at both nests. A similar emigration was observed from a nest of robustior in the Namib Desert. In this case the new nest was located 40 m from the old. Both persisted for several weeks as independent colonies but thereafter the new colony appeared to die or perhaps to relocate once again without having been seen to do so. The original colony remained vigorous several months later. In the case of the picardi example, careful excavation of the two nests revealed that the original (mother) colony contained 360 individuals, one of which was a physogastric ergatoid female, whereas the new (daughter) colony contained 108 individuals, 20 of which were ergatoids but none of which were physogastric. Dissection of all ergatoids from the daughter colony revealed that one had a distended spermotheca that was engorged with sperm. None of the others had been inseminated. This is the first fully excavated Ocymyrmex colony that had an inseminated but non-physogastric ergatoid. The evidence strongly suggests that she was a recently mated young queen at the start of her reproductive life.

Two other questions arise from this: where and when does mating occur, and does the mated ergatoid dig the new nest alone or with recruited nest-mates? Chance observation by Marsh on robustior in the Namib Desert indicates that mating almost definitely occurs at night, probably at almost any time of the year, and takes place at the nest entrance of the mother colony. As mentioned elsewhere in this paper, males are regularly trapped at night. On several occasions ergatoid females have been seen opening their nest at night, well after the last forager had returned and closed the nest. Having re-opened the nest entrance, the ergatoids remain in the vicinity of the nest for several hours before retreating below ground. No males emerged or arrived at the nest during these times but the most likely explanation for this behaviour is that the females were trying to attract males, probably by 'calling' with pheromones. We hypothesize that upon insemination the ergatoid returns into the mother colony and later, in conjunction with some recruited help, commences excavating a new nest some distance from the mother colony, an example of auto parasitism in the sense of Bolton (1986).

Observations by one of us (Marsh) indicate that nest digging is a regular part of the diurnal behaviour repertoire of Ocymyrmex species and that nest relocation is common. Typically such relocations involve the entire colony and entail a move to a new nest very close to the original nest. For example, a mass emigration in 1986 of an O.foreli colony was carefully documented by Alves and Marsh (unpublished). For three weeks prior to the emigration most activity involving individual and group forays from the nest occurred towards a specific site 4 m away, where the ants were excavating a new nest. Finally, during the course of a single day, the entire colony relocated to the new nest. Brood, callows and intra-nidal individuals were carried to the nest by the extra-nidal individuals. Similar but less well documented behaviour has been observed for several other species of Ocymyrmex. The reason for mass emigration and nest relocation is not known. It would appear that the ants consider the original nest to be unsuitable but that the foraging area itself remains suitable. Colony fission, in contrast, involves movements over considerably greater distances away from the centre of the foraging area of the mother colony. Thus it seems likely that a recently inseminated ergatoid excavates a new nest with the help of some recruited nest-mates before colony fission occurs.

Several intriguing questions concerning social organization in Ocymyrmex remain unanswered at present. For example, if the original physogastric reproductive female dies and a number of ergatoids which are acting as workers then become mated, which takes over the parent nest, and how is the process regulated? Or do all such newly-mated forms disperse, abandoning the original nest-site?

Some years ago Holldobler, Stanton and Engel (1976) reported the presence of a previously undetected gastral exocrine system in workers of O. picardi and in a couple of specialized North American deserticolous ants belonging to the genus Aphaenogaster. This exocrine system was later termed the pygidial gland by Kugler (1978), the name now generally accepted for the system. The gland consists of a number of cells located under the intersegmental membrane between gastral tergites 3 and 4 ( = abdominal tergites 6 and 7), with ducts through the membrane. In O. picardi and some other myrmicines the gland is associated with a special cuticular area on the base of the pygidium (gastral tergite 4), on the section of the scelerite which in life is overlapped by the apex of the third gastral tergite. Kugler (1978) detected this specialized cuticular area in a second species, referred to as Ocymyrmex cf. arnoldi, which was most probably O.fortior.

In their review of tergal and sternal glands in ants Holldobler and Engel (1978) pointed out that pygidial glands may occur in myrmicines without the development of the specialized cuticular area at the base of the pygidium, and thus only be detectable by histological sectioning. However, by examination of alcohol-preserved material of Ocymyrmex we can confirm the presence of the specialized area of pygidial cuticle in workers of all species so examined, namely alacer, flaviventris, fortior, hirsutus, nitidulus, picardi, resekhes, robustior, and ignotus. It appears as a narrow shallowly concave transverse strip, filled with fine reticular patterning, on each side of the midline close to the pygidial base.

Males of Ocymyrmex are often collected at lights (H. Robertson, pers. comm.) but males associated with conspecific workers and females are extremely hard to acquire in certain species as the nests are often so deep in the earth and so difficult of access. Even when a nest is fully excavated there is of course no guarantee that males will be present. For this reason very little is known of this sex, but enough data has been gathered to present a reasonable genus-level diagnosis, given later in this paper. Recently Hamish Robertson (University of the Witwatersrand) has discovered a number of male/female mosaics of O. robustior, that is, males with irregular patches of female tissue. These oddities are discussed in the notes following the diagnosis of the male. The cause of such mosaic development is not known, but a reasonable mechanism for the production of patches of female (diploid) tissue in an otherwise haploid male can be adduced.

Castes
Winged queens are unknown in this genus (Bolton & Marsh 1979)

Nomenclature

 *  OCYMYRMEX [Myrmicinae: Pheidolini]
 * Ocymyrmex Emery, 1886: 364. Type-species: Ocymyrmex barbiger, by monotypy.

The following is from Bolton and Marsh (1989)

Worker
Diagnosis of worker

Terrestrial fast-moving ants with monomorphic worker caste, belonging to subfamily Myrmicinae. Habitus as in Figs 1 and 6 and with the following combination of characters.

1. Mandible short and powerful, armed with 5 (usually) or 4 (rarely) sharp teeth which decrease in size from apex to base. Counting from the apical the third and fourth teeth (when 5 present in total), or only the third tooth (when 4 present in total) paired, having flanking teeth internally on the masticatory margin; flanking teeth only visible when mandibles open.

2. Trulleum shallow, closed and weakly defined, almost obliterated in some species.

3. Palp formula 5, 3; 4,3; 3, 3; 2, 3; see notes below.

4. Ventral surface of head with a strongly developed psammophore, the ammochaete hairs arising on the ventre itself, on base of ventral borders of mandibles, and on bases of mouthparts.

5. Clypeus large, posteriorly broadly inserted between the frontal lobes, lacking a long unpaired anteromedian seta.

6. Frontal lobes well defined but short, mostly or wholly covering the antennal insertions, ending at same level as do the antennal fossae.

7. Frontal carinae and antenna1 scrobes absent.

8. Antennae 12-segmented, filiform, without an apical club.

9. Eyes well developed, situated slightly behind midlength of sides of head.

10. Mesothoracic spiracles opening dorsally or high on the sides, clearly visible in dorsal view, with slit-like or crescent-shaped orifices.

11. Metanotal groove absent; propodeum unarmed, without trace of spines or teeth.

12. Propodeal spiracle extremely elongate, slit-shaped and very conspicuous.

13. Metapleurallobes linked by an arched carina across the propodeal declivity.

14. Metapleural glands small to moderate, widely separated from the elongate propodeal spiracle.

15. Metasternal process (and associated carinae) absent, but metasternal pit conspicuous. The pit is in a depression at the posterior end of the mid-sternal longitudinal suture, very close to the apex of the alitrunk-petiole articulatory excision.

16. Posterior excision of ventral alitrunk where petiole articulates broadly U-shaped and running forward as far as or just beyond a line connecting the anteriormost points of the hind coxal cavities [see notes below].

17. Legs extremely long and slender, coxae large and powerful; middle and hind tibiae with simple to weakly roughened spurs present.

18. Petiole with a long narrow anterior peduncle, the petiolar spiracle situated at the node or just in front of it. A short posterior peduncle present behind the node.

19. Sting very reduced in size, apparently not functional.

Notes

The palp formula (PF). As previously stated (Bolton, 1981) PF 3, 3 is predominant in the genus, but it is now known that some species show variation in maxillary palp segmentation. PF remains unknown in robecchii and 'jiavescens.

PF 5,3 occurs in cavatodorsatus, gordoni and gariepensis.

PF 4,3 occurs in barbiger, cilliei, afradu, dekerus, kahas, and robustior. In some, or perhaps all of these species with PF 4,3, rare individuals may appear which have PF 3, 3. In some cases it is possible that these oddities may be represented by single workers in series where all others have PF 4, 3, but this has not yet been observed. On occasion individual workers may be found in which one maxillary palp has 4 segments and the other 3 segments due to fusion of the small apical palpomeres (afradu, dekerus).

PF 2,3 is known only in the small species tachys and engytachys.

PF 3,3 occurs in all other known species. A very small minority of workers in many of these species show an apparent or real PF of 4,3. This is because the apical maxillary palpomere is sometimes deeply constricted near its midlength or its apex; the section beyond the constriction may be separated off as a small fourth segment.

Ventral alitrunk. Characters 15 and 16 above have been investigated in afradu, alacer, barbiger, cavatodorsatus, celer, cilliei, flaviventris, foreli, fortior, gariepensis, hirsutus, ignotus, nitidulus, phraxus, picardi, resekhes, robustior, sobek, velox, and zekhem.

Queen
Diagnosis of female (queen)

Extremely ergatoid, physogastric when reproductive, and approximately the same size as the worker, answering to 'all characters of the worker (1-19) given above. Characters normally associated with reproductive female ants, such as large eyes, presence of ocelli, enlarged alitrunk with flight sc1erites and wings, etc., are never developed. Female is different from worker, in all species where the former is known, by the following.

1. Outer margins of frontal lobes more widely separated in their posterior halves in females than in workers; margins of frontal lobes behind level of antennal insertions parallel or nearly so in females, convergent posteriorly or pinched in in workers.

2. Antennal scapes broader and usually slightly shorter in females than in workers; see table of dimensions for 9 species in Bolton (1981).

3. Dorsum of head behind level of eyes usually with strong regular transverse sculpture in females, whereas sculpture is generally longitudinal in this area in workers. [A few species have transverse cephalic sculpture in workers, for instance ankhu, robecchii, hirsutus, but here the form of the sculpture usually varies between the two castes, being more strongly defined in ergatoid females than in workers. In a very few species both castes may have sculpture extremely reduced and difficult to discern on the cephalic dorsum.]

Notes

Ergatoid females are produced in relatively large numbers. In foreli they make up 4-20 per cent of the nest population (Forder and Marsh, 1989). Only one is fecund, and is physogastric, the remainder take on a worker-like function and are often found foraging outside the nest. Ergatoid females are now known for alacer, barbiger, cavatodorsatus, celer, cilliei, flaviventris, foreli, fortior, hirsutus, ignotus, nitidulus, okys, phraxus, picardi, resekhes, robustior, sobek, sphinx, turneri, velox and zekhem.

Male
Diagnosis of male

(Based on males of barbiger, cilliei, foreli, fortior, robustior, weitzeckeri, plus one species unassociated with workers but probably turneri.)

1. Mandibles extremely reduced and non-functional, elongate-lobiform and edentate. Mandibular apices not meeting at full closure, with a marked gap between them.

2. Labrum large and deeply cleft anteromedially.

3. PF 3, 2 (foreli by dissection, robustior and fortjor by in situ count).

4. Psammophore absent.

5. Anterior clypeal margin approximately transverse, without a median notch and lacking a strongly differentiated median seta.

6. Clypeus an unspecialized weakly convex transverse strip.

7. Anterior tentorial pits closer to base of mandibles than to antennal sockets.

8. Antennal insertions close together; width of one antennal socket greater than minimum distance between them.

9. Frontal lobes absent.

10. Antennal scape short, shorter than any funicular segment except the first.

11. Antennae 13-segmented, filiform, without an apical club; first funicular segment much shorter than the remainder.

12. Eyes large and prominent, anteriorly situated; in profile eyes seen to lap around lower margin of sides of head, just on to the ventral surface.

13. Ocelli present, small to large, borne on a low turret so that the median ocellus is directed forward.

14. Head capsule in full-face view much broader behind eyes than in front; head apparently attached low on anterior face of alitrunk because of specialized mesoscutum.

15. Pronotal posterior lobe raised over orifice of meso thoracic spiracle.

16. Mesoscutum swollen, bulging dorsally and anteriorly; much broader than long in dorsal view and much broader than maximum width of head.

17. Notauli absent but mesoscutum anteromedially with a narrow weakly impressed longitudinal line.

18. Parapsidal grooves vestigial.

19. Venation as shown in Figs; see also notes below.

20. Alitrunk with a deep median transverse impression between mesoscutum and scutellum, both of which are swollen; the impression bounded laterally by the axillae.

21. Axillae subtriangular on dorsum, widely separated, linked across the dorsal impression by a narrow strip of thinner cuticle.

22. Propodeal spiracle an elongate near-vertical slit.

23. Metapleurallobes small, rounded to angular.

24. Metapleural glands apparently absent.

25. Middle and hind tibiae with simple spurs.

26. Petiole with an elongate anterior peduncle.

27. Petiolar spiracle close to or at the reduced node, always well behind the midlength of the peduncle.

28. Pygostyles ( = cerci) present. Basal ring of genitalia large and parameres strongly sclerotized.

Notes

In the forewing venation the radial ( = marginal) cell is closed on the margin; m-cu is absent; cu-a is close to the base of the wing; Rs and M separate at or distal of the junction of 2r with Rs + M. Frequently an adventitious vein or vein-stub arises from M and runs towards the apex. When present this adventitious vein is very variable in length, direction, degree of development, and position at which it leaves M. In one male of fortior examined the adventitious vein arises from Rs on the left forewing, and from close to the base of M on the right forewing. Fig. 23 shows the standard venation, Fig. 24 a forewing with a strongly developed adventitious vein arising from M. The typically pheidoline system of forewing vein reduction has been outlined by Bolton (1982: 362).

A number of males of robustior collected at Gobabeb, Namibia by Hamish Robertson (University of the Witwatersrand), show mosaic patches offemale (diploid) tissue on their heads. A mechanism to account for such tissue mosaics is outlined in the introduction to this paper. The main mosaics noted on males now deposited in BM(NH) were as follows.

The first mosaic male (series C1056) has a female left mandible and a hypertrophied but male-like right mandible. The left antennal funiculus is of female form but the scape is short and bizarre, with a number of cuticular excrescences. Anterolateral portions of the head, in front of eyes and on each side of antennal sockets, and the genal areas, are female. As a result the eyes, which are of male form, are deformed and displaced posteriorly. The right antennal socket is female and is supertended by a reduced and deformed female-form frontal lobe; the left is a normal male antennal socket. Long ammochaete hairs are present on the mandibles and under the head of this male.

Three mosaic males occur in series C 1 057. The first of these has enlarged and grossly deformed mandibles which are basically of the male form but which have patches of female tissue. Ammochaete hairs occur on the mandibles and under the head. The second has deformities very like those ofC1056, above, but the right scape is longer and has a massive club-like dorsal excrescence of female tissue. The genal patches of female tissue are not as large as in CI056 so that the eyes are not as strongly displaced posteriorly. The third male has the left mandible mostly offemale form, but with incomplete dentition. The right mandible is an undifferentiated mass but shows coarse female-like sculpture. The head in front of the eyes has much female tissue on each side of the clypeus, with the result that the latter is much deformed. The eyes are irregular in shape and are displaced posteriorly. The antennae have only 12 segments. They are of male form except for the apical four antennomeres of the left funiculus, which are female-like. Ammochaete hairs are numerous.

Each of the mosaic forms occurs in series with other, perfectly normal, males. Such mosaic males have not been noted in fortior, the only other species of which we have numerous samples of males. So whether the tendency to produce such mosaics is restricted to robustior as a species, to the robustior population in the Gobabeb area, or to a single robustior nest which happened to be in the vicinity of the light-trap, remains a matter for conjecture. However, Robertson (pers. comm.) informs us that as the robustior males were collected over a long period it seems safe to assume that more than one nest was involved.