The Ants Chapter 2

CHAPTER 2. CLASSIFICATION AND ORIGINS

The ants are classified as a single family, the Formicidae, within the order Hymenoptera, which also includes the bees, wasps, sawflies, ichneumons, and similar forms. The known living ants comprise 11 subfamilies, 297 genera, and approximately 8,800 species (see Table 2-1). Judging from the rate of discovery of genera during the past 50 years and the alacrity with which incontestable novelties are publicized, fewer than 100 genera are likely to remain unrecognized. In contrast, the number of undescribed species is immense. Many parts of the tropics still remain poorly collected, and this is especially true of the moist continental forests. "Sibling" species, that is, populations that are reproductively isolated but difficult to distinguish by means of ordinary anatomical traits, are notoriously rife within the ants. They often come to light when biometrical studies are applied to large samples. A few have even been identified primarily or solely by differences in chromosomes or electrophoretically separate allozymes. Examples of complexes that have been "broken open" by these techniques occur within Amblyopone, Aphaenogaster, Camponotus, Conomyrma, Myrmica, and Rhytidoponera, and many more are thought to exist (Ward, 1980; Crozier, 1981). An extreme case of imperfect classification is provided by Pheidole, the most speciose of all ant genera. There are 524 currently unchallenged names in the New World alone of which as many as 300 may represent distinct species, with several hundred additional, undescribed species present in collections. Similar "problem" genera, any one of which could be called the crux myrmecologorum, include Camponotus, Crematogaster, Iridomyrmex, and Solenopsis. Turning to an entire fauna, R. W. Taylor (personal communication) estimates that only one-third to one-fourth of the ants of Australia have been described. Overall, it is quite possible that 20,000 or more species of ants, encompassing as many as 350 genera, exist in the world.

The taxonomy of ants
The taxonomy of the world ant fauna is still very incomplete, with much more research being needed in every taxonomic category from species to subfamily. There are, to begin with, few useful regional monographs. William S. Creighton's (1950) review of the ants of North America north of Mexico remains one of the most useful as a convenient guide and quick reference, although it has now been replaced by revisions of more than half the Nearctic genera. It has the distinction of being the first major work to dispense with the clumsy and meaningless polynomials that plagued ant taxonomy for a hundred years. Creighton substituted a much simpler and more efficient system of binomials and trinomials based on modern population concepts (for example, "Camponotus herculeanus pennsylvanicus var. whymperi" was placed as a synonym under Camponotus herculeanus). Through his influence, and that of William L. Brown and a few others working on the entire world fauna, taxonomic procedures have been thoroughly modernized during the past 40 years. Also, the number of generic monographs and faunistic studies has accelerated conspicuously during the past ten years with the entry of more young investigators into the field. But like a mosaic lacking just enough pieces to obscure the pattern, the classification of the world fauna still lacks a satisfying coherence and practical utility.

Ant larvae have been systematically described by George C. and Jeanette Wheeler (1951-1986; syntheses in 1976a and 1979), with a supplementary analysis supplied by Picquet (1958). The basic anatomy and a classification of larval body forms are presented in Figures 2-12 and 2-13. Excellent comparative accounts of the proventriculus have been given by Eisner (1957); adult mouthparts by Ettershank (1966), Gotwald (1969), and Buren et al. (1970); wing venation by Brown and Nutting (1949); antennal sensillae by Masson (1974) and Walther (1981a,b); adult body sculpturing by Harris (1979); the myrmicine sting apparatus by Kugler (1978a, 1986); grooming behavior by Farish (1972); adult carrying behavior by Möglich and Hölldobler (1974) and Duelli (1977); the strigil (cleaning comb) of the foreleg pretarsus by Francoeur and Loiselle (1987); and the physiology of digestion and colonial food flow by Abbott (1978). The generic characteristics of male ants in the North American fauna have been analyzed by M. R. Smith (1943). Otherwise, males have been mostly neglected in taxonomic studies--in good part due to their greater scarcity in collections in comparison with workers. The cytotaxonomy of ants, which is still a young but potentially very important subject, has been reviewed by Crozier (1975, 1987b), Imai et al. (1977, 1984), Sherman (1979), Hauschteck-Jungen and Jungen (1976, 1983), and Taber (1986). A remarkable case of a karyotype with only one pair of chromosomes has been reported in the primitive bulldog ant Myrmecia pilosula by Crosland and Crozier (1986). The possible relation between chromosome numbers and social evolution will be discussed later, in Chapter 4. Finally, a perceptive and entertaining account of the early history of ant taxonomy has been written by William L. Brown (1955b).

The origin of the ants
Until recently the search for the ancestry of the ants ended in frustration. For more than a hundred years large numbers of fossil ants had been recovered from Oligocene and Miocene deposits, but they were all members of living subfamilies. Even at the generic level they possessed a distinctly modern aspect. Myrmecologists were forced to consider the Eocene Epoch or even more remotely, the Cretaceous Period, where no certain fossils belonging to the Formicidae were yet known. In 1967, Wilson, Carpenter, and Brown (1967a,b) obtained the first ant remains of Cretaceous age. The species, Sphecomyrma freyi (see Plate 1), and the new subfamily founded on it (Sphecomyrminae) were described from two well-preserved workers in New Jersey (United States) amber dating to the late middle (Late Santonian) portion of the Cretaceous Period. The age of the specimen was first estimated to be about 100 million years but has recently been adjusted to 80 million years (Donald Baird and F. M. Carpenter, personal communication).

Sphecomyrma freyi proved to be the nearly perfect link between some of the modern ants and the nonsocial aculeate wasps. In particular, the Cretaceous ants had the following primitive wasp-like traits: mandibles very short and with only two teeth, gaster unconstricted, sting extrusible, and middle and hind legs furnished with double tibial spurs. The Sphecomyrma were in fact a mosaic, for they also possessed the following distinctively ant-like character states: thorax reduced in size and wingless, petiole or "waist" pinched down posteriorly at its juncture with the rest of the abdomen (but still primitive in form in comparison with later ants), and--most importantly--an apparent metapleural gland, the possession of which is the key diagnostic trait of modern ants. The Sphecomyrma were intermediate between most modern aculeate wasps and almost all modern ants in the form of the antennae, which combined a proportionately short first segment with a long, flexible funiculus.

Additional specimens of Sphecomyrma of about the same age were next discovered in amber from Alberta Province, Canada (Wilson, 1985f). In the interim Dlussky (1975, 1983) described an important collection of ant-like forms from several time horizons in the Upper Cretaceous of the Taymyr Peninsula (extreme north-central Siberia), southern Kazakh S.S.R., and the Magadan region of extreme eastern Siberia. He erected ten new genera to accommodate this material. In his second report (1983) he also created a new family, the Armaniidae, to receive some of these new forms, while elevating the Sphecomyrminae to family rank (hence, Sphecomyrmidae) to accommodate Sphecomyrma and a few related Soviet fossils. However, in a later analysis Wilson (1987c) marshalled new morphological evidence to assemble all of the Cretaceous formicoids into only a single subfamily, the Sphecomyrminae, within the Formicidae and into at most two genera, Sphecomyrma and Cretomyrma, rather than two families and numerous genera. The females appear to have been differentiated as queen and worker castes belonging to the same colonial species instead of winged and wingless solitary females belonging to different species. The former conclusion is supported by the fact that the abdomens of workers of modern ant species and extinct Miocene ant species (chosen to include fossil forms) are smaller relative to the rest of the body than is the case for modern wingless solitary wasps. The wingless Cretaceous formicoids, including the original Sphecomyrma freyi, conform to the proportions of ant workers rather than to those of wasps and are therefore reasonably interpreted to have lived in colonies. The female castes and male inferred by this interpretation are depicted in Figure 2-14.

In 1986, Jell and Duncan (1986) described Cretacoformica explicata, a supposed ant, from Lower Cretaceous beds in Victoria, Australia. If verified, this species would be the earliest known formicid and, because of its geographic origin, extraordinarily important in reconstructing the origin of the ants. However, the single poorly preserved specimen, a male, cannot be assigned with certainty to either the Formicidae or a pre-formicid line. The region of the body that might contain a petiole is covered by the abdomen, which was folded over the rear portion of the alitrunk during preservation. The wings are rounded at their tips and reduced in venation. Overall, the specimen seems more likely to be an aculeate wasp rather than an ant, but judgment must be reserved until more material becomes available.

The Mesozoic fossils unearthed to date seem to present us with the following picture. During the middle and late Cretaceous times representatives of a few species belonging to the very primitive subfamily Sphecomyrminae ranged widely across the northern hemisphere in what was then the supercontinent Laurasia. They were evidently scarce in comparison with later ants in Tertiary and modern times. Only two individuals (Sphecomyrma canadensis) have been found so far among thousands of insects in amber from Alberta Province, Canada (J. F. McAlpine, personal communication). Formicoids constituted just 13 of the 1200 insect impressions in the Magadan collection and 5 of the 526 impressions among the Kazakhstan fossils, in other words about one percent in both cases (Dlussky, 1983). These figures contrast sharply with the proportionately high representation of ants in Oligocene and Miocene deposits. In the Florissant and other shales of North America (F. M. Carpenter, 1930), as well as the Baltic amber of northern Europe (Wheeler, 1914b) and amber of the Dominican Republic (Wilson, 1985c-e,h), the ants are the most abundant insects, making up a large minority of all specimens. The adaptive radiation destined to propel the ants to dominance took place no later than the beginning of the Tertiary Period, about 65 million years ago. Eomyrmex guchengziensis, a species apparently combining traits of Sphecomyrma and the living Ponerinae, has been recorded from the early Eocene Fushan deposits of Manchuria (Hong et al., 1974). Amber of mid-Eocene age from Arkansas has yielded representatives of the Myrmicinae, Dolichoderinae, and Formicinae (Wilson, 1985f). In addition, Dlussky (personal communication) has recently found representatives of four living subfamilies (Ponerinae, Aneuretinae, Dolichoderinae, and Formicinae) in Eocene amber from Sakhalin. The exact age cannot be determined, because the amber pieces were not collected in the original deposits, but it is quite likely that the material dates to the early Eocene.

Finally, Lutz (1986) has recognized a new subfamily of gigantic ants, the Formiciinae, from the Lower Eocene of Tennessee and Middle Eocene of England and Germany. The single genus Formicium, not to be confused with Formica (the type genus of the Formicinae), is evidently the same as Eoponera and Pseudosirex, which hymenopterists had previously consigned to the family Pseudosiricidae and classified as siricoid wasps. The new material described by Lutz makes placement of the Formiciinae plausible. The key traits of the specimens, which are all winged queens and males, are the following:

(1)	Petiole produced into an erect, thin scale of the kind found in many species of the Formicinae.

(2)	Reduced sting, also similar to the Formicinae.

(3)	Wing venation primitive, closely resembling the "idealized" pattern of our Figure 2-11 and unique to the Formiciinae.

(4)	Huge size, with the forewing length of the queen from 25 to 65 mm according to species; the largest species, Formicium giganteum, exceeds in size any other known ant, living or extinct.

(5)	The spiracles of the gaster are proportionately very large and slit-shaped.

(6) The forewing venation is distinctive, which we interpret as follows (differently from Lutz): the cuticle of the stigma is thin and transparent, making it appear to be an additional cell.

Only one of the approximately ten genera thus far recorded from these several Cretaceous and Eocene deposits is extant (Iridomyrmex, from the Arkansas amber). However, no fewer than 24 genera, or 56 percent of the 43 total represented in the early Oligocene Baltic amber fossils, still survive, including such currently abundant and widespread forms as Ponera, Tetraponera, Aphaenogaster, Monomorium, Iridomyrmex, Formica, and Lasius (Wheeler, 1914b). At least one species, Lasius schiefferdeckeri, is so close to living species of the L. niger group of North America and Eurasia that it can be distinguished only by minor average differences in antennal and mandibular form (Wilson, 1955a). This modern facies is even more evident in the Dominican amber, which is apparently early Miocene in age. Here no fewer than 35 genera, or 92 percent of the total 38, still survive. Still further, the great majority of species analyzed to date have been placed in modern species groups. In a few instances they are difficult to separate from modern forms even at the species level (Wilson, 1958c-e,h).

From what taxonomic family did the Sphecomyrminae evolve? In a preliminary phenetic analysis of the original Sphecomyrma freyi, Wilson et al. (1967b) placed the Mesozoic subfamily closest to the methochine Tiphiidae among living aculeate wasps. In a later study, Brothers (1975) applied a cladistic analysis of 92 characters to all of the living aculeate families including the ants (Formicidae) as a whole, and arrived at a different result. As shown in Figure 2-15, he derived the ants from a later clade than that giving rise to the Tiphiidae, but prior to the branching that led to the modern Eumenidae, Masaridae, Scoliidae, and Vespidae. Among the twelve families of the Vespoidea in this scheme, the Formicidae are so distinctive that Brothers felt compelled to place them in a special informal section of their own, the Formiciformes, with the remaining eleven families composing the Vespiformes. A case could be made for the retention of the superfamily Formicoidea, but according to the logic of cladistic classification this is not permissible--unless families antecedent to the Formicidae were also split off to create several other superfamilies.

Our current view of phylogeny within the Formicidae is summarized in Figure 2-16. This arrangement is the latest revision in a succession of earlier cladograms by Brown (1954a), Wilson et al. (1967b), and Taylor (1978c). The principal new feature is the more extensive case of exocrine glands, about which a great deal has been learned during the past ten years. One important change from earlier schemes is the very early divergence of the subfamily Formicinae, probably in late Cretaceous or earliest Tertiary times. A similar conclusion was reached independently by William L. Brown (personal communication).

In reconstructing phylogenies in this manner, it is important to keep in mind the distinction between primitive character states and primitive taxa. A primitive character state ("plesiomorphous state" of many authors) is simply one that precedes a more advanced one ("apomorphous state") in evolution. The two states can be major, as in absence of the metapleural gland changing to presence of the gland; or they can be quite trivial, as in ten hairs on the pronotum changing to twenty hairs. A primitive taxon, on the other hand, is judged to be so on a much more subjective basis. It is a species, or genus, or a taxon belonging to some other, higher category that possesses a relatively large number of primitive character states in comparison with other taxa of the same rank. Thus we speak of the Sphecomyrminae as a primitive subfamily relative to the Myrmicinae because sphecomyrmine workers are characterized by short, wasp-like mandibles, a single symmetrical petiole, and other important (i.e., complex) primitive character states.

Entomologists will no doubt alter our ant cladogram further as new species are discovered, especially fossil forms from Cretaceous, Paleocene, and Eocene deposits, and new characters and dates are added to the analysis. They will also disagree on which are the most primitive taxa. A number of difficulties already exist in our own version. One of the most troubling is the relation of Nothomyrmecia macrops, a very primitive ant with reference to the living Formicidae and Amblyopone, a worldwide genus of primitive ponerines thought to be derived from the sphecomyrmine-nothomyrmeciine clade. Amblyopone has a tubulated abdominal segment, clearly a derived character state (see Figure 2-17). But Amblyopone also has a petiole (abdominal segment II) that attaches broadly to the gaster (abdominal segments III-VII), which is presumably a primitive state. In addition, it possesses reduced eyes, proportionately short appendages, and a heavily sclerotized exoskeleton. Amblyopone is clearly an ant adapted for hypogaeic foraging, that is, for hunting underground and within enclosed spaces in rotting wood and leaf litter. Nothomyrmecia, in contrast, is epigaeic, hunting aboveground and in the open (see Plate 00). Sphecomyrma, judging from its appearance (large eyes, proportionately long appendages, thin exoskeleton) was also epigaeic. Either the broad petiolar attachment of Amblyopone and related genera in the tribe Amblyoponini represents an evolutionary reversion from the primary narrow petiole of the Ur-Formicidae, or else the Amblyoponini (and possibly the remainder of the Ponerinae, Myrmicinae, and related higher subfamilies) are an independent clade of ants predating the Nothomyrmeciinae and Sphecomyrminae. We favor the first, more conservative hypothesis. The matter, however, will remain open until more evidence is obtained. For example, if Amblyopone-like forms are discovered from post-Cretaceous deposits with fully constricted petioles, the first hypothesis will be favored. If on the other hand, Amblyopone-like forms are discovered in deposits as old as those containing Sphecomyrma or older, the second, more radical hypothesis will receive strong support.

The origin of social behavior
All known living ants are eusocial, with strong physical differences separating the queen and worker castes. Thus a large gap in social behavior remains between even the most primitive ants including Amblyopone, Myrmecia, and Nothomyrmecia, and their closest living relatives among the vespoid wasps. Without further evidence it would be very difficult to infer the steps that led to eusocial behavior in these insects. However, the living eusocial species of Vespoidea are connected to solitary species of other vespoids and aculeate wasps by finely graded steps, providing an independent but otherwise solid base for inference. Howard E. Evans (1958), drawing on his own extensive knowledge of the solitary Hymenoptera and studies by Richards and Richards (1951) and other contemporary students of the social wasps, proposed an ethocline which has stood up well under the test of more recently accumulated field data. His schema collates two independent forms of information: first, the morphological similarity and inferred direction of morphological change, which are expressed in the branching pattern of cladograms, and second, the sequence of grades that can be logically envisioned to have occurred during behavioral evolution. It is notable that several families of wasps (Pompilidae, Sphecidae, Eumenidae, and Vespidae) have to be included in order to encompass the whole story. This in no way vitiates the theory. Indeed, if we believe that behavioral evolution is even loosely correlated with morphological evolution, it follows that single taxa such as families and genera should encompass less behavioral variation than all the wasps taken together. Evans' 13 grades, starting with the simplest and presumably most primitive, are as follows:

1. Female stings prey, lays egg.

2. Female stings prey, places it in a convenient niche, lays egg.

3. Female stings prey, constructs a nest on the spot, lays egg.

4. Female builds a nest, stings prey, takes it to nest, lays egg.

5. Female builds a nest, collects a prey item, lays egg, then mass provisions with several more prey (added quickly, before the egg hatches).

6. As in (5) but prey items are progressively provided, as the larva grows.

7. As in (6) but progressive provisioning occurs from the start.

8. In addition to progressive provisioning in a preconstructed nest, the female macerates prey items and feeds the pieces directly to the larvae.

9. The founding female is long-lived, so that offspring remain with her in the nest, add cells, and lay eggs of their own.

10. The little colony of cooperating females now engage in trophallaxis (liquid food exchange), but there is still no division into reproductive and worker castes.

11. A behavioral division between a dominant queen caste and subordinate worker caste appears; the unfertilized workers may still lay male-destined eggs.

12. The larvae are fed differentially; the queen and workers that result are physically distinct, but intermediates remain common.

13. The worker caste is physically strongly differentiated, and intermediates are rare or absent.

These grades are said to follow the subsocial route, through which a single foundress gains enough longevity to coexist in the same nest as her female offspring. In this case the most primitive colony is an extended family: the founding female is accompanied by her daughters, sons, and grandsons, but not under ordinary circumstances by her granddaughters, because the unfertilized eggs of her daughters produce only males.

A relatively minor variation on the 13 grades is the parasocial route (Michener, 1969, 1974), which starts with members of the same generation using the same composite nest and also cooperating in brood care, instead of a single foundress carrying through on her own. It is possible, and actually occurs for example in founding aggregations of the paperwasp Polistes, for one of the females to dominate her contemporaries and to become the de facto queen. On the basis of more recent field studies, West-Eberhard (1978) has argued that this was the prevalent route followed during the origin of the eusocial wasps. She has proposed the "polygynous family group hypothesis," in which the aggregating foundresses are typically sisters or at least close cousins. Their mutualistic association is followed by serial queendom, in which first one female or small group of females and then another takes over oviposition. Both foundresses and offspring can lay eggs in such family groups. In later evolution, progressively fewer females participate in reproduction. Finally, a sharply demarcated, mostly non-reproductive worker caste emerges. West-Eberhard's scheme has been supported by a cladistic study of anatomical and behavioral traits performed by J. C. Carpenter (1989), and it has been widely favored among other wasp specialists.

Against this backdrop we can now place existing knowledge of the anatomically most primitive ants. As summarized in Table 2-4, all of the key traits are sufficiently similar across Amblyopone, Myrmecia, and Nothomyrmecia to justify placing them on the Evans' vespoid scala. All three of these ant genera are in the most advanced (thirteenth) stage, yet their behavior is in every way reminiscent of a clade that began as tightly knit, soil-dwelling families of medium-sized vespoid wasps. Haskins and Haskins (1951) were essentially correct when they concluded on the basis of their studies of Myrmecia that "the existent wealth and variety of Formicid social structures, with their tremendous range of variation from group to group, took their evolutionary beginnings in the activities of solitary, winged, ground-dwelling wasplike types in which the female, having dealated herself after fertilization, constructed a shelter in the ground and reared a small family to maturity. The larvae were provided with freshly killed prey captured and supplied through a behavior pattern intermediate in its complexity between the simple provisioning of paralyzed insects characteristic of the modern solitary Sphecoid wasps and the malaxated pellets of insects prepared for the larvae by their nurses in such primitive social Vespids as Polistes."

It is further possible that the original foundresses gathered in small associations as in some Polistes. In the language of our formal lexicon, the evolution was parasocial rather than strictly subsocial. In this case, the polygynous group hypothesis of West-Eberhard could also apply to the origin of the ants. Colonies of Nothomyrmecia macrops are in fact sometimes founded by multiple queens, although well-developed colonies typically have only a single queen. When Hölldobler and Taylor (1983) introduced two queens to a group of queenless workers, they behaved amicably toward each other at first, but later one began to dominate the other by standing above her at frequent intervals (see Figure 6-7). Later the workers expelled the subordinate queen by repeatedly dragging her outside the nest. This pattern of multiple foundresses being whittled down to a single queen is also widespread in the phylogenetically more advanced subfamilies of ants, as will be shown in Chapter 6.

An opposing hypothesis of the origin of social behavior was advanced by the late Soviet entomologist S. I. Malyshev (1960, 1968). He postulated that specialization on big prey resulted in the mother wasp staying in the vicinity of her young long enough for them to get to know her and to cooperate with her. Malyshev further postulated that the precursors of the ants must have fed on fungi growing in the nest wall. Otherwise, he contends, the young colonies would have no way of tiding over the period of scarcity after the initial large prey was consumed. But this suggestion ignores what we know of colony founding throughout the ants and is wholly unsupported by any evidence of fungus eating in the lower ants. The model for the large prey theory is provided by members of the bethylid genus Scleroderma, particularly S. immigrans and S. macrogaster, which were studied in detail by Bridwell (1920) and Wheeler (1928). The female S. macrogaster, for example, is only 2.5 to 3 mm long, and she attacks beetle larvae that are hundreds or thousands of times greater in bulk. In a typical sequence, the female first crawls over the surface of her prey, pausing from time to time to grip little folds of the cuticle. As long as the muscles beneath in the cuticle show any sign of contraction, the female stings at that point. Finally, after one to four days, the larva becomes completely paralyzed. The Scleroderma now feeds for several days by making little punctures in the cuticle and drinking the hemolymph. Her abdomen then begins to swell as the ovaries develop, and after a time she lays eggs on the surface of the prey. The remainder of the life cycle has been described by Wheeler in the following striking passage:

The eggs laid on a larva or young pupa produce minute larvae which at first lie on the surface but later become spindle-shaped and erect, so that the host bristles with them like a porcupine. The older larvae acquire the colour of the juices of the prey; those feeding on the pink larvae or pupae of Liopus becoming red. They are always spotted with white, owing to the large masses of urate crystals in their fat bodies. The mother Scleroderma remains with the larvae, often stands over them and may sometimes lick them, holding them meanwhile in her fore feet. She also continues occasionally to drink the host's blood, which exudes about the deeply inserted heads of her larval offspring. Although she will sometimes eat her eggs I have never seen her attack one of her larvae. The devouring of some of the eggs seems to be due to a tendency to regulate their number according to the volume of the prey. When the larvae are mature they fall away from its shrivelled and exhausted remains and spin snow-white cocoons in a cluster. Pupation covers a period of fourteen to thirty days. The males emerge first from their cocoons, at once eat their way into the female cocoons and fecundate the pupae. They also mate readily with the same individual five to eight times after brief intervals. The same females may also mate with several males in succession. So great is the ardour of the latter that they often attempt to mate with one another. The mother being a long-lived insect may mate with one of her sons and will readily paralyze another beetle larva, rear another brood and mate again with one of her grandsons. (Wheeler, 1928: 63).

Wilson had earlier suggested (1971) that amblyoponine ants might have evolved in the Wheeler-Malyshev manner, because Amblyopone were known occasionally to transport their larvae to centipedes and other large prey rather than the other way around. However, it is now clear (Traniello, 1982) that one species at least, A. pallipes, more commonly transports its prey back to the brood chambers of the nest as part of a stereotyped and efficient predatory sequence. Wilson also found that the amblyoponine Myopopone castanea carries larvae to large wood-boring larvae that could not possibly have been transported any significant distance from the spot where they had been disabled by the workers. However, these ants spread their nests widely beneath the bark of rotting logs, and they also forage there. It can be argued that the entire subcortical surface constitutes the "nest" of the ants. Finally, the small amblyoponine Prionopelta amabilis has now been shown to be closer in some important respects to higher ants than to Amblyopone and Myopopone. Its colonies are large, consisting of hundreds of workers; there is a marked size difference between the queen and worker; the nests are more elaborate, with pupal chambers being "wall-papered" with fragments of discarded cocoons; and perhaps most importantly, the workers specialize on campodeid diplurans, small flightless insects that they capture and carry back to their nests (Hölldobler and Wilson, 1986a).

The causes of success
What unusual or unique biological traits led to the remarkable diversification and unchallenged success of the ants for over 50 million years? The answer appears to be that they were the first group of predatory eusocial insects that both lived and foraged primarily in the soil and rotting vegetation on the ground. Many ant species are specialized for arboreal existence, but even the great majority of these distinctive forms live in tree boles, hollow twigs, and in moist subcortical cavities simulating an earthen environment. Arboreal life appears to represent a secondary, minority adaptation. To an exquisite degree ants are creatures of the ground. The wingless workers are able easily to penetrate small, remote cavities less accessible to flying wasps, which are burdened with wings and bulky thoraces. Armed with stings and toxic chemical weapons, they are formidable predators. They orient in part by odor cues on the ground, and most species are able to recruit foraging parties with a high degree of efficiency through the use of odor trails laid over the surface. After entering the adaptive zone of social, terrestrial predators, no later than the Upper Cretaceous, the ants apparently preempted its occupation by other candidate groups among the insects.

Although Malyshev, inspired by Wheeler's observations, spoke of the "sclerodermoid" ancestors of the ants, morphological evidence rules against any of the ant groups having been derived from Scleroderma-like progenitors or any of the other known Bethylidae. Furthermore, as suggested by the family-level cladogram of Figure 2-15, the bethyloids diverged from the early vespoids well before the origin of the ants from a vespoid stock. The accumulating behavioral evidence also militates against the Wheeler-Malyshev route to eusociality and in favor of the vespoid route, in which prey are brought to previously constructed nests.