The Ants Chapter 12

CHAPTER 12. SYMBIOSES AMONG ANT SPECIES

The ant society is a decidedly more open system than the lower units of biological organization such as the organism and the cell. In the course of evolution the tenuous lines of communication among the members of the colonies have been repeatedly opened and extended to incorporate alien species. Many kinds of ants, for example, adopt aphids, mealybugs, and other homopterans as cattle to provide a steady source of honeydew; a few raid colonies of other species to acquire workers as domestic slaves, or utilize the odor trails of other species, or defend common nest sites. Just as frequently, the lines of communication have been tapped by other alien species that have insinuated themselves into the colony as inconspicuous social parasites. Taken together, the hundreds of cases of interspecific symbioses among ant species that have come to light encompass almost every conceivable mode of commensalism and parasitism. But true cooperation is rare or nonexistent. No verified examples of mutualism are yet known, in which two species cooperate to the benefit of both. All of the relationships carefully analyzed to date are unilateral, with one species profiting and the other species either remaining unaffected or, in the great majority of cases, suffering from the attentions of its partner.

The “ultimate” social parasite
There is no better way to begin a survey of the social symbioses than by considering the most extreme example known, that of the “ultimate” parasitic ant Teleutomyrmex schneideri. This remarkable species was discovered by Heinrich Kutter (1950a) at Saas-Fee, in an isolated valley of the Swiss Alps near Zermatt. Its behavior has been studied by Stumper (1950) and Kutter (1969), its neuroanatomy by Brun (1952), and its general anatomy and histology by Gösswald (1953). A second population has been reported from near Briançon in the French Alps by Collingwood (1956), a third in the French Pyrenees by Buschinger (1987c), and still others in the Spanish Sierra Nevada by Tinaut Ranera (1981). Appropriately, the name Teleutomyrmex means “final ant.”

The populations of Teleutomyrmex schneideri, like those of most workerless parasitic ant species (Wilson, 1963), are small and isolated. The Swiss population appears to be limited to the eastern slope of the Saas Valley, in juniper-Arctostaphylos woodland ranging from 1,800 to 2,300 m in elevation. The ground is covered by thick leaf litter and sprinkled with rocks of various sizes, providing, in short, an ideal environment for ants The ant fauna is of a typically boreal European complexion, comprising the following free-living species listed in the order of their abundance (Stumper, 1950):  Formica fusca, Formica lugubris, Tetramorium caespitum, Leptothorax acervorum, Leptothorax tuberum, Camponotus ligniperda, Myrmica lobicornis, Myrmixa sulcinodis, Camponotus herculeanus, Formica sanguinea, Formica rufibarbis, Formica pressilabris, and Manica rubida. For some unexplained reason this little assemblage is extremely prone to social parasitism. Formica sanguinea is a facultative slavemaking species, preying on the other species of Formica. Doronomyrmex pacis, a workerless parasite living with Leptothorax acervorum, was discovered by Kutter as a genus new to science in the Saas-Fee forest in 1945. In addition, Kutter and Stumper found Epimyrma stumperi in nests of Leptothorax tuberum, as well as two parasitic Leptothorax, goesswaldi  and kutteri, in nests of Leptothorax acervorum  (Kutter, 1969).

Teleutomyrmex schneideri is a parasite of Tetramorium caespitum  and Tetramorium impurum. Like so many other social parasites, it is phylogenetically closer to its host than to any of the other members of the ant fauna to which it belongs. In fact, it may have been derived directly from a temporarily free-living offshoot of this species, since Tetramorium caespitum and Tetramorium impurum (the host species at Briançon and in the Pyrenees) are the only nonparasitic tetramoriines known to exist at the present time through most of central Europe. It is difficult to conceive of a stage of social parasitism more advanced than that actually reached by Teleutomyrmex schneideri. The species occurs only in the nests of its hosts. It lacks a worker caste, and the queens contribute in no visibly productive way to the economy of the host colonies. The queens are tiny compared with most ants, especially other tetramoriines; they average only about 2.5 mm in total length. They are unique among all known social insects in being ectoparasitic. In other words, they spend much of their time riding on the backs of their hosts (Figure 12-1). The Teleutomyrmex queens display several striking morphological features that are correlated with this peculiar habit. The ventral surface of the gaster (the large terminal part of the body) is strongly concave, permitting the parasites to press their bodies close to those of their hosts. The tarsal claws and arolia are unusually large, permitting the parasites to secure a strong grip on the smooth chitinous body surface of the hosts. The queens have a marked tendency to grasp objects. Given a choice, they will position themselves on the top of the body of the host queen, either on the thorax or the abdomen. Deprived of the nest queen, they will then seize a virgin Tetramorium queen, or a  worker, or a pupa, or even a dead queen or worker. Stumper observed a case in which six to eight Teleutomyrmex queens simultaneously grasped one Tetramorium  queen, completely immobilizing her. The mode of feeding of the Teleutomyrmex is not known with certainty. The adults are evidently either fed by the host workers through direct regurgitation or else share in the liquid regurgitated to the host queen. In any case, they are almost completely inactive most of the time. The Teleutomyrmex adults, especially the older queens, are highly attractive to the host workers, who lick them frequently. According to Gösswald, large numbers of unicellular glands are located just under the cuticle of the thorax, pedicel, and abdomen of the queens; these are associated with glandular hairs and are believed to be the source of a special attractant for the host workers. The abdomens of older Teleutomyrmex queens become swollen with fat body and ovarioles, as is shown in Figure 12-1. This physogastry is made possible by the fact that the intersegmental membranes are thicker and more sclerotized than is usually the case in ant queens and can therefore be stretched more. Also, the abdominal sclerites themselves are widely overlapping in the virgin queen, so that the abdomen can be distended to an unusual degree before the sclerites are pulled apart. The ovarioles increase enormously in length, discard their initial orientation, and infiltrate the entire abdomen and even the postpetiolar cavity.

From one to several physogastric queens are found in each parasitized nest, usually riding on the back of the host queen. Each lays an average of one egg every thirty seconds. The infested Tetramorium colonies are typically smaller than uninfested ones, but they still contain up to several thousand workers. The Tetramorium queens also lay eggs, and these are capable of developing into either workers or sexual forms (Buschinger, personal communication). Consequently the brood of a parasitized colony consists typically of eggs, larvae, and pupae of Teleutomyrmex queens and males mixed with those of Tetramorium  workers.

The bodies of the Teleutomyrmex queens bear the mark of extensive morphological degeneration correlated with their loss of social functions. The labial and postpharyngeal glands are reduced, and the maxillary and metapleural glands are completely absent. The mandibular glands, on the other hand, are apparently normal. In addition, the queens possess a tibial gland, the function of which is unknown. The integument is thin and less pigmented and sculptured in comparison with that of Tetramorium; as a result of these reductions the queens are shining brown, an appearance that contrasts with the opaque blackish brown of their hosts. The sting and poison apparatus are reduced; the mandibles are so degenerate that the parasites are probably unable to secure food on their own; the tibial-tarsal cleaning apparatus is underdeveloped; and, of even greater interest, the brain is reduced in size with visible degeneration in the associative centers. In the central nerve cord, ganglia 9-13 are fused into a single piece. The males are also degenerate. Their bodies, like those of the males of a few other extreme social parasites, are “pupoid,” meaning that the cuticle is thin and depigmented, actually greyish in color; the petiole and postpetiole are thick and provided with broad articulating surfaces; and the abdomen is soft and deflected downward at the tip.

In its essentials the life cycle of Teleutomyrmex schneideri resembles that of other known extreme ant parasites. Mating takes place within the host nest. The fecundated queens then either shed their wings and join the small force of egg layers within the home nest or else fly out in search of new Tetramorium nests to infest. Stumper found that the queens could be transferred readily from one Tetramorium colony to another, provided the recipient colony originated from the Saas-Fee. However, Tetramorium colonies from Luxembourg were hostile to the little parasites. Less surprisingly, ant species from the Saas-Fee other than Tetramorium caespitum always rejected the Teleutomyrmex. However, Buschinger (personal communication) has pointed out that the Saas-Fee population could be caespitum or impurum, or a mixture of both. In other words, the transfer might have been attempted across species.

The kinds of social parasitism in ants
Social parasitism in ants is complicated, and its study has become virtually a little discipline of entomology in itself. The source of the complexity is first the large number of ant species that have entered into some form of parasitic relationship with each other. Second, at least two and possibly three major evolutionary routes lead to the ultimate stage of permanent, workerless parasitism. Finally, no two species are exactly alike in the details of their parasitic adaptation. Table 12-1 contains a list of the known parasitic ants, together with certain essential data concerning each of them. With this information readily at hand for constant reference, we will now present what is deliberately a rather didactic review of the entire subject, attempting to make it as orderly and clear as possible from the outset.

Wasmann (1891) distinguished two classes of consociations, or myrmecobioses as Stumper (1950) later dubbed them, that occur between different species of ants. These are the compound nests, in which two or more species live very close to each other, in some cases even running their nest galleries together, but keep their brood separated; and the mixed colonies, in which the brood are mingled and cared for communally. Compound nests are very common in nature. They reflect relationships that range, depending on the species involved, all the way from the accidental and trivial to total parasitism. Mixed colonies, on the other hand, almost always come about as a result of social parasitism. Forel (1898, 1901), who was the first to use the expression “social parasitism,” and Wheeler (1901a,b, 1910a) devoted a great deal of attention to compound and mixed nests and provided a useful classification of the underlying relationships, complete with a somewhat less useful set of Hellenistic terms to label the various categories. Let us examine this classification briefly. Then we will make more interesting use of it in tracing the evolution of parasitism and other forms of symbioses.

Compound nests
Plesiobiosis. In this most rudimentary association, different ant species nest very close to each other, but engage in little or no direct communication--unless their nest chambers are accidentally broken open, in which case fighting and brood theft may ensue. The less similar the species are to each other morphologically and behaviorally, the more likely they are to cluster together in an accidental, truly “plesiobiotic” relationship. Put the other way, closely related species of ants are the least likely to tolerate each other's presence.

Cleptobiosis. Some species of small ants build nests near those of larger species and either feed on refuse in the host kitchen middens or rob the host workers when they return home carrying food. R. C. Wroughton (quoted by Wheeler, 1910a) has described a species of Crematogaster in India whose workers “lie in wait for Holcomyrmex, returning home, laden with grain, and by threats, rob her of her load, on her own private road and this manoeuvre was executed, not by stray individuals, but by a considerable portion of the whole community.”  Workers of Conomyrma pyramica  in the southern United States collect dead insects discarded by colonies of Pogonomyrmex, including corpses of the Pogonomyrmex  themselves. Our impression in the field has been that some Conomyrma colonies obtain a large part of their food in this way, to the point of preventing kitchen middens from building up near the Pogonomyrmex  nests.

Lestobiosis. Certain small species, most belonging to Solenopsis and related genera, stay in the walls of large nests built by other ants or termites and enter the nest chambers of their hosts to steal food and prey on the inhabitants. For example, colonies of the “thief ants” of the subgenus Solenopsis (Diplorhoptrum), including especially Solenopsis fugax of Europe and Solenopsis molesta of the United States often nest next to larger ant species, stealthily enter their chambers, and prey on their brood. Species of Carebara in Africa and tropical Asia frequently construct their nests in the walls of termite mounds and are believed to prey on the inhabitants (see Chapter 15). The relationship is parasitic with respect to nest sharing and predatory with respect to brood theft.

Parabiosis. In this peculiar form of symbiosis, two or more species use the same nest and sometimes even the same odor trails, but they keep their brood separate. The situation is similar to the mixed foraging flocks of birds so prevalent in tropical forests, except that in some instances at least, one species dominates and exploits another.

Xenobiosis. This symbiotic state falls just short of a truly mixed colony. One species lives in the walls or chambers of the nests of the other and moves freely among its hosts, obtaining food from them by one means or another, usually by soliciting regurgitation. The brood is still kept separate. This relationship is truly parasitic.

Mixed colonies
The following phenomena are vital in the later stages of parasitic evolution. In a sense they form categories comparable to those just cited for compound nests, although they are less than ideal because they are not mutually exclusive. Nevertheless, we favor continuing to distinguish them on the grounds that the associated terminology is the familiar one in literature dating back over nearly a century and, more importantly, the classification can still be relied upon to serve as an adequate guide through the complex relationships as we understand them.

Temporary social parasitism. This symbiosis was first clearly recognized by Wheeler (1904b) as a result of his studies of the life cycle of members of the Formica microgyna group, especially Formica difficilis. It has since been discovered in a diversity of genera belonging to several subfamilies. The newly fecundated queen finds a host colony and secures adoption, either by forcibly subduing the workers or by conciliating them in some fashion. The original host queen is then assassinated by the intruder or by her own workers, who somehow come to favor the parasite. With the development of the first parasite brood, the worker force soon becomes a mixture of host and parasite species. Finally, since the host queen is no longer present to replenish them, the host workers die out, and the colony comes to consist entirely of the parasite queen and her offspring. Temporary social parasitism is generally considered to be preceded in evolution by the re-adoption of queens by colonies of their own species following the nuptial flights. Bolton (1986b) has referred to this condition as “autoparasitism.”

Dulosis (slavery). Certain ant species have become dependent on workers of other species which they keep as slaves. The slave raids of the evolutionarily advanced species are dramatic affairs in which the slavemaking workers go out in columns, penetrate the nests of colonies belonging to other, related species, and bring back pupae to their own nests. The pupae are allowed to eclose, and the workers become fully functional members of the colony. The workers of most slavemaking species seldom if ever join in the ordinary chores of foraging, nest building, and rearing of the brood, all of which are left to the slaves. Facultative inter- and intraspecific slavemakers also occur. These less specialized forms provide an illuminating glimpse into the likely early stages of dulosis.

Inquilinism (“permanent parasitism”). In this final, degenerate stage, the parasitic species spends its entire life cycle in the nests of the host species. Workers may be present, but they are usually scarce and display atrophied behaviors. In many of these species, as for example in Teleutomyrmex schneideri, the worker caste has been lost altogether. Wilson (1971) suggested the use of the term inquilinism in preference to the somewhat more familiar expression “permanent parasitism” since obligatorily dulotic species are also permanent parasites. Inquilinism and dulosis, on the other hand, form exclusive categories; they are meant to be the streamlined equivalents of Kutter's (1969) “permanent parasitism without dulosis” and “permanent parasitism with dulosis.” The queens of some inquiline species permit the host queen to live, while others either assassinate her or else somehow, and in a procedure yet to be firmly established by experiments, induce her own workers to accomplish the task.

The occurrence of social parasitism throughout the ants
A rich variety of new parasitic species, representing almost every conceivable evolutionary stage, has been added since the time of Wheeler's classic synthesis in 1910. They continue to be discovered at such a consistently high rate as to suggest that, at this moment, only a small fraction of the total world fauna of social parasites is known. The reason for the slow uncovering of the world fauna seems clear: parasitic species tend to be both rare and locally distributed. As a rule, moreover, the more advanced the stage of parasitism, the rarer the species. Thus, we find (Table 12-1) that temporary social parasites, such as members of the Formica exsecta and Lasius umbratus groups, are often nearly as widely distributed as their free-living congeners, and a few of the species are also very abundant. Species in which dulosis is weakly developed or even facultative, as, for example, the representatives of the Formica sanguinea group, are also relatively abundant and widespread. On the other hand, extreme dulotic species, such as the members of Strongylognathus, Polyergus, and Rossomyrmex, exist in more restricted, sparser populations. Finally, the extreme workerless parasites are, as a rule, both very rare and very locally distributed. Anergates atratulus comes closest to being an exception. It has been collected over a wide area from southern France to Germany, and it has even been accidentally introduced into the United States with its host Tetramorium caespitum. Yet everywhere within this range it is still a comparatively rare ant. The great majority of other workerless parasites have been found at only one or two localities and are extremely difficult to locate, even when a deliberate search is made for them in the exact spots where they were first discovered. Usually they give the impression, quite possibly false, of having no more than a toehold on their host populations and of existing close to the edge of extinction.

Most of the known parasitic species have been recorded exclusively from the temperate areas of North America, Europe, and South America. Almost certainly this reflects at least in part the strong bias of ant collectors, most of whom reside in these areas and devote a large part of their lives to a meticulous examination of local faunas. Switzerland, for example, is the present “capital” of parasitic ants for the simple reason that both Auguste Forel and Heinrich Kutter lived there. About one-third of the 110 Swiss species are parasitic (Kutter, 1969). Europe has received the attention of the expert collector, Alfred Buschinger, and his students for over twenty years. The United States has benefited similarly from the efforts of W. M. Wheeler, the Wesson brothers, and other more recent gifted collectors, while the rich trove of species uncovered in Argentina has been due to three men who spent a large part or all of their lives in the country--Carlos Bruch, Angel Gallardo, and Nicolás Kusnezov. We believe that as the huge and still little-known tropical ant faunas are more carefully worked (there are no resident myrmecologists on the Amazon!), many more parasitic species will come to light. Some evolutionarily advanced forms are already known from tropical regions. In a recent study, Wilson (1984c) recognized four tropical parasites in the genus Pheidole, including two new species and the Zairean Pheidole neokohli, which rivals Teleutomyrmex in the extremeness of its degeneration. Equally impressive are the strange postxenobiotic Kyidris parasites of New Guinea (Wilson and Brown, 1956). Wheeler (1925) pointed out that females of the numerous species of Crematogaster belonging to the “subgenera” Atopogyne and Oxygyne, groups widely distributed in Africa, Madagascar, and tropical Asia, have all of the morphological characteristics of northern ants known to be temporary parasites in that they tend to be small and shining and to possess falcate or very oblique mandibles and large postpetioles which are attached broadly to the gaster. The last of these characteristics is usually associated with physogastry, also a common but not diagnostic feature of social parasitism. Emery (1899) recorded a highly physogastric nest queen of Crematogaster ranavalonae from Madagascar. At least two species of the Neotropical dolichoderine genus Azteca (aurita and fiebrigi) possess some of these traits. The species of Rhoptromyrmex, found in South Africa, Asia, New Guinea, and Australia also possess them (Brown, 1964a; Bolton, 1986a). A special study of such species, and any others that can be found to possess various of the “temporary parasite syndrome” of characters, might prove very rewarding to future students of tropical myrmecology.

Even so, the vast differences in quality of sampling from the major parts of the world render the matter inconclusive, and there remains the possibility that life in certain climates and environments actually does predispose ant species toward parasitism. It is true, for example, that a disproportionate number of parasitic species, especially the complete inquilines, occur in mountainous and arid regions. We have already mentioned the extraordinary diversity of parasites found in the little forest of the Saas-Fee. Among numerous other examples that can be cited are the montane species Pheidole inquilina, Pheidole elecebra, Manica parasitica, Pogonomyrmex anergismus, Pogonomyrmex colei Doronomyrmex pocahontas, and Leptothorax faberi, which together make up about half of the known inquiline fauna of North America. Temporary social parasites, along with species that can be tentatively placed in this category by virtue of their morphology, are more abundant in the colder portions of Europe and North America than in the warm temperate and subtropical portions, even though the faunas of the two climatic zones are otherwise not radically different. Even more impressively, dulosis is a common phenomenon in the colder parts of Europe and Asia but rare in the warmer parts; and not a single example has ever been reported from the tropical or south temperate zones.

It is conceivable that cooler temperatures facilitate the introduction of parasitic queens in the early evolution of the phenomenon by dulling the responses of the host colonies. We have found, in general, that if ant colonies are first chilled in the laboratory they are more likely to adopt queens of their own species, which they would otherwise attack and destroy. In nature parasite queens need not wait for winter to utilize this effect. Some degree of chilling, say to 10° or 15°C, occurs commonly during the cool summer nights in mountainous regions, right in the middle of the season of nuptial flights. It should prove instructive to study the effects of various degrees of cooling of potential host colonies on the success of introduction of queens belonging to species at any early stage of inquilinism, such as Leptothorax faberi and Manica parasitica. Useful information might also be obtained from an analysis of the behavioral effects of cooling on ant groups that most commonly serve as hosts, such as the genera Leptothorax and Formica, as opposed to those that are relatively immune to social parasitism, such as the genus Camponotus.

Other clues to the origin of social parasitism can be found in the phylogenetic distribution of the phenomenon, which is remarkably patchy. The more advanced forms of parasitism, namely dulosis and inquilinism, are almost wholly limited to the subfamilies Myrmicinae and Formicinae and are furthermore heavily concentrated in certain genera, including Pheidole, Myrmica, Leptothorax, Tetramorium, Plagiolepis, Lasius, and Formica, and in the satellite parasitic genera derived from them. Two inquilines (Myrmecia hirsuta and [[Myrmecia inquilina) have been described from the primitive subfamily Myrmeciinae. In view of the relatively small number of species known in the Myrmeciinae (about 120) and the limited amount of field study devoted to it to date, parasitism in this group may eventually be found to occur at about the same level of frequency as in the Myrmicinae and Formicinae.  The only parasites known with certainty among the Dolichoderinae, on the other hand, are the temporarily parasitic species of Bothriomyrmex''.  This relative immunity is puzzling since the dolichoderines are a relatively large, numerically abundant group of advanced phylogenetic rank.  Perhaps the explanation lies in the fact that very few dolichoderine species range into the cooler portions of the North Temperate Zone where parasitic species are most likely to evolve.  Yet it is also true that a rich dolichoderine fauna exists in subtropical and temperate Argentina, where many myrmicine parasites have been discovered.  No parasitic species of any kind are yet known in the Ponerinae, Cerapachyinae, and Dorylinae.  One can speculate almost endlessly on why this is the case.  For example, the Ponerinae are primitive (but so are the Myrmeciinae, and in any case many ponerine species form large colonies with advanced social traits).  The Dorylinae engage in frequent nest changes (but many parasitic beetles, millipedes, wasps, and other arthropods emigrate with them along their odor trails).

An important lead from the phylogenetic distribution has emerged recently from studies by Buschinger, Alloway, and their co-workers on the myrmicine tribe Leptothoracini. Although leptothoracines represent fewer than 3 percent of the 8800 described ant species, they contain 30 (15 percent) of the 200 known parasitic species. From the research of Buschinger, Francoeur, and their co-workers, utilizing a combination of cytological, morphological, and behavioral traits, we can be reasonably sure that slavemaking alone has arisen a minimum of six times within the Leptothoracini: once each in the lines leading to Temnothorax duloticus, Harpagoxenus, Temnothorax americanus, Chalepoxenus, Epimyrma, and Myrmoxenus. As Buschinger has said, “The myrmicine tribe Leptothoracini comprises an astoundingly rich variety of socially parasitic genera and species. New species can be found nearly everywhere when populations of independent species are closely examined.”

What is the cause of the vulnerability of the leptothoracines to social parasites? On the other side, what inclines so many to turn into parasites? Buschinger (1970, 1986) and Alloway et al. (1982) believe that the key predisposing traits are polygyny, the regular occurrence of multiple laying queens in colonies, and polydomy, the spread of colonies to multiple nest sites. All these traits are developed in Leptothorax (= Myrafant) and Leptothorax (s.str.), the preeminent northern hemisphere representatives of the Leptothoracini and the stock group from which most of the parasitic genera and species have arisen. To be structured in this manner means that the colonies are relatively “open,” in other words they are more easily invaded by alien queens of the same or different species. Polygyny usually arises by the re-adoption of queens after they have mated outside the nest. This habit thus being fixed in the workers' repertory, colonies are more susceptible to invasion by “cuckoo” queens able to provide the right chemical cues. Polydomy often results in the creation of outlier nests containing only workers and immature forms. It is possible that these queenless fragments add to the general vulnerability of the colonies.

If the view be accepted that colony structure can predispose species for or against social parasitism, how might this explanation apply to the apparent scarcity of parasitism in the tropics? It is possible that for some reason ant species with leptothoracine-type biology are rare in warmer climates. However, our knowledge of the social organization and life cycle of the vast majority of tropical ant species is too meager to search for this correlation. Buschinger (personal communication) has offered one promising hypothesis: “In my opinion a very important factor might be that parasitism occurs most frequently when the host species form dense, large, and homogeneous populations. This is the case in many temperate-zone species, whereas in warmer areas a high species diversity is often combined with rarity and wide dispersal of nests of a given species. I had presumed this for a long time, and recently in Australia I found a perfect confirmation of this idea (I also did not find any parasites there, but did not check Myrmecia nests, which often form dense populations--they were too aggressive!).”

Another factor to examine is the means by which workers recognize colony odor at the species level. If they have a fully innate, “hard-wired” recognition separating individuals of the same species from those of different species, they will be very resistant in evolution to the intrusion of social parasites. If, on the other hand, they learn the species odor early in life, they can be more easily duped. A worker captured by a slavemaker while still in the pupal or callow (newly eclosed) adult stage can be imprinted on the odor of the captor. It will serve automatically as a slave thereafter.

The relative flexibility of early learning of conspecific brood labels might also play a role. Extensive research has been conducted on the recognition of brood (immature stages) by ants of the genus Formica (Jaisson, 1975, 1985; Le Moli and Passetti, 1977, 1978; Jaisson and Fresneau, 1978; Le Moli and Mori, 1982). The results demonstrate that young adult workers learn to recognize whatever species of brood they encounter within a period of approximately one week after their emergence from the cocoon. Unfamiliar brood pieces, whether of another species or their own, are rejected or destroyed. It has been suggested that an early learning of brood labels has favored the repeated evolution of slavemaking and social parasitism among ants. Le Moli (1980) and later Brian (1983) argued that the only species suitable as hosts or slaves are those in which brood recognition is based on learning without any bias favoring individuals of the same species. This preadaptive flexibility would ensure that immature slave or host ants, when eclosing, would learn the brood labels of their own species as well as those of their parasites. The Brian-Le Moli hypothesis is only partially supported by the evidence, however. Formica rufa and Formica lugubris, which do not exhibit a bias for learning conspecific brood labels, are not themselves victims of parasitism, although they are temporary social parasites of other Formica species (Gösswald, 1951a; Kutter, 1969). Formica polyctena, which is usually free-living, is occasionally parasitized by Formica truncorum (Kutter, 1969). Preference for early learning of conspecific brood labels has been found in Camponotus (Carlin et al., 1987a,b). It is tempting to attribute the “immunity” to social parasitism of the genus Camponotus to this bias of learning conspecific brood labels. These facts appear to support the Brian-Le Moli hypothesis. However, since the bias discovered in Camponotus is not exclusive, colonies should still be vulnerable to potential parasites. Le Moli cites as supporting evidence his discovery that Lasius niger does not learn brood labels at an early age, and he uses the trait to explain why Lasius niger is evidently immune against social parasitism. However, he overlooks the fact that Lasius niger serves as the host of the temporary social parasites Lasius umbratus (Crawley, 1909; Gösswald, 1938a; K. Hölldobler, 1953) and Lasius fuliginosus (Andrasfalvy, 1961).

To summarize, we can recognize several predisposing features toward social parasitism that might explain why it occurs in some ant genera and not others. Those most likely to take the step (1) live in cool or arid climates, (2) have multiple queens as a result of re-adoption of newly mated queens, (3) occupy multiple nests, some of which are at least temporarily without a resident queen, (4) live in dense populations, and (5) learn the species odor early in life.

The evolution of social parasitism in ants
In 1909 Carlo Emery formulated what is perhaps the single most important generalization concerning social parasitism: “The dulotic ants and the parasitic ants, both temporary and permanent, generally originate from the closely related forms that serve them as hosts.”  What he meant, of course, was that the parasitic species tend to resemble their host species more closely than they do any other free-living form. “Emery's rule,” as it has been called (Le Masne, 1956b), has continued to hold well for the true inquilines. Taxonomists have stressed that certain of the parasites--for example, Paramyrmica colax, Pheidole inquilina, Leptothorax buschingeri, and Strumigenys xenos--really are morphologically more similar to their hosts than to any other known species. At first glance, this relation seems to create a paradox: how can a species generate its own parasite? As long ago as 1919 Wheeler experienced difficulty in even conceiving of a mechanism by which it could occur. In 1971 Wilson proposed a scheme based on the known process of geographic speciation in other organisms (see Figure 12-2). In what is generally regarded as the prevalent sequence of animal speciation, a single “parental” species can be divided into two “daughter” species by, first, fragmentation due to geographic barriers and, second, genetic divergence of the populations thus isolated geographically until they acquire intrinsic isolating mechanisms. If and when the newly formed species reinvade one another's ranges, the isolating mechanisms prevent them from interbreeding. And if, in addition, one of the species then becomes specialized as a parasite on the other, the condition of Emery's rule is fulfilled. This model suggests that, all other circumstances being equal, the frequency of parasitism within a given genus should increase as a function of the rate of speciation. A corollary is that the more taxonomically “difficult” the genus, in other words, the larger the percentage of newly formed, indistinctly defined species in it, the higher should be the percentage of parasitism. This prediction does indeed seem to be met by such genera as Pheidole, Leptothorax, Plagiolepis, Lasius, and Formica, although the correlation through all ant groups taken together is far from perfect. But at least it is clear that the speciation rate is one additional factor that must be considered in future evolutionary analyses of the subject.

An alternative mode of evolution that must always be kept in mind is sympatric speciation, in which certain homogamous mutants or recombinants (forms that breed only with their own kind) arise in sufficient numbers at the same place and time to segregate a distinct breeding population. In other words, the new parasitic species might arise in situ directly from the host species, without the intervening step of geographic isolation. Buschinger (1986) expressed a simple model as follows: “Ant species or populations may change from monogyny to polygyny (and vice versa) depending on various environmental conditions, and polygyny is a sufficiently frequent phenomenon in ants as to represent a condition from which parasitism may have evolved convergently several times. A conceivable mutation, e.g., causing certain sexuals to mate at a different time of day, might quickly produce a subpopulation of individuals genetically isolated from the original form but still living in its polygynous colonies. Once this isolation has been achieved, development in the inquiline or the dulotic direction may depend upon the ability of such a form to produce further workers. And when the host species evolves toward monogyny, or the parasite spreads into monogynous populations of its host species, selection will favor those parasite queens which are able to replace the host colony queens by force, as in dulotic or temporary parasitic species.” Pearson (1981) has added the view that inquilines, whether created by allopatric or sympatric speciation, are more likely to arise when competition occurs prominently. The species that is subordinated must survive on smaller amounts of food and is more likely to be miniaturized to fit into its marginal ecological role. Once miniaturized, it can more easily enter nests of the dominant species.

Buschinger's model, with or without miniaturization, is basically the same that has been elaborated many times in studies of other, free-living organisms. It is difficult to disprove, but it is also rendered less probable by the fact that disruptive selection, the basic mode of selection involved, must be severe to create incipient species. Furthermore, species isolating mechanisms are usually genetically complex, a circumstance lessening the likelihood of sympatric speciation (Futuyma, 1986).

Many exceptions to Emery's rule exist. The most notable ones fall into the special categories of xenobiosis and parabiosis, in which the parasitic species typically belongs to a different genus and sometimes even to a different subfamily. The explanation for these two classes of exceptions is simple enough. When members of different genera associate at all, they are not likely to combine their brood, for they tend to be very different in biology and mutually incompatible. Any association achieved will be of a more tenuous sort, involving grooming, food exchange, trail sharing, or combinations of these relations--xenobiosis and parabiosis. The more closely related the two species, the more likely they are to enter into the more intimate forms of parasitism, producing the effect that is generalized in Emery's rule.

How do different ant species come together in symbiosis in the first place? The evolutionary schema presented in Figure 12-3 is an extension of one evolved in a long sequence of contributions by Wheeler (1904b, 1910a), Emery (1909), Escherich (1917), Stumper (1950), Dobrzanski (1965), Wilson (1971), and Buschinger (1986), with additions of our own. The single most important idea embodied in this diagram is that inquilinism is a convergent phenomenon, reached independently by many different species following one or the other of at least two available pathways in evolution. Also, complete inquilinism is viewed as an evolutionary sink; a return to free life or even to a partially parasitic existence by reversed evolution seems impossible. For convenience we have arranged known cases of social parasitism according to these hypothesized sequences.

The temporary parasitism route
The earliest stages of temporary parasitism are displayed by members of the Formica rufa group (Gösswald, 1951a,b; Kutter, 1913, 1969). Several of the members, Formica lugubris, Formica polyctena, and Formica pratensis, form colonies with multiple queens. New colonies are usually created by budding, or “hesmosis” as it has occasionally been called. Following the nuptial flights the newly fecundated queens normally return to the home nest, and at some later date some of them may move to a new site nearby with a group of workers. The new unit thus created is a colony only in the purely spatial sense because it may exchange workers with the mother nest for an indefinite time afterward. Multiplication by budding creates the pattern, so characteristic of these species of Formica, of dense aggregations of interconnecting nests that dominate local areas. Occasionally young queens do not find their way back to a nest of their own species. They may then seek adoption in a colony of the Formica fusca group. Whenever one of them succeeds in penetrating such an alien nest, the host queen is somehow eliminated; the intruder takes over the role of egg laying exclusively, and eventually the host workers die off. The final result is that the colony consists entirely of the intruder and her offspring. Such temporary parasitism is regarded as a secondary mode of colony founding for these ants since mixed host-parasite colonies are rarely encountered in nature.

However, a closely related species, Formica rufa, has taken the step of founding its colonies predominantly by temporary parasitism, then forming monogynous colonies or relying on budding in a minority of instances, where polygynous supercolonies build up. Its host species in Europe include Formica fusca and Formica lemani. The rufa queen is still a rather inept parasite. On approaching the host colony she does not hide, play dead, conciliate, or display any of the other dissembling tricks ordinarily used by parasitic queens; instead, she plunges right into the nest. Such intrusions frequently result in the death of the queen at the hands of hostile host workers, but enough attempts succeed to maintain Formica rufa as one of the more abundant and widespread ant species of Europe.

Most European students of Formica, starting with Emery (1909), have argued that loss of the ability to found nests in the usual claustral manner, with the resulting dependence on adoption and budding, preadapts members of the rufa group to temporary parasitism on other species. Also, the fact that Formica rufa itself is usually monogynous (its colonies each tolerate only one egg-laying queen) predisposes this species even further to incursions on other species.

The species of the exsecta group of Formica (collectively referred to by earlier European writers as the “subgenus Coptoformica"”) have a life cycle very similar to that of the rufa group species, except that the queens have become more skillful at penetrating host colonies (Kutter, 1956, 1957). The European species of Formica exsecta, for example, depend chiefly on homospecific adoption and budding, but a few queens seek colonies of the fusca group of species (formerly called the “subgenus Serviformica"”). The exsecta queens stalk the host colonies and either enter the nests by stealth or else permit themselves to be carried in by host workers. The exsecta queens are smaller and shinier than those of most members of the rufa group, and they seem to be treated with less hostility by the host workers. This is also the case for Formica pressilabris, a second member of the exsecta group found in Europe. Queens approached by host workers lie down and “play dead” by pulling their appendages into the body in the pupal posture. In this position they are picked up by the host workers and carried down into the nests without any outward show of hostility. Later they somehow manage to eliminate the host queen and take over the reproductive role. Similar life histories have been described for the North American species of the exsecta group (Wheeler, 1906; Creighton, 1950; Scherba, 1958, 1961) and have been postulated on the basis of limited laboratory experiments to characterize members of the North American microgyna group (Wheeler, 1910a).

The transition from temporary social parasitism to full inquilinism, depicted in Figure 12-3, has been achieved by Formica talbotae, a workerless species of the Formica microgyna group that lives with Formica obscuripes in the north-central United States (Talbot, 1976; Wilson, 1976d). A second species evidently in the same category is the closely related Formica dirksi of Maine (Wing, 1949; Wilson, 1976d).

Further subtleties have been developed by the related genus Lasius. Apparently all of the species of the fuliginosus, reginae, and umbratus groups are temporary parasites on members of the Lasius niger group (see Wilson, 1955a). This relationship is obligatory, not optional as in the rufa and exsecta groups of Formica. The colonies are monogynous for the most part, and homospecific adoption is not practiced. When newly mated queens of Lasius umbratus are searching for a host colony, they first seize a worker in their mandibles, kill it, and run around with it for a while before attempting to penetrate the nest (K. Hölldobler, 1953). Apparently all of the parasitic Lasius get rid of the host queens, but the exact means employed are still unknown in most cases. The queens of Lasius reginae, a species discovered in Austria by Faber (1967), eliminate their rivals by rolling them over and throttling them (Figure 12-4).

Assassination is also the technique employed by the queens of the dolichoderine species Bothriomyrmex decapitans and Bothriomyrmex regicidus in gaining control of colonies of Tapinoma (Santschi, 1906, 1920). These temporary parasites occur in the deserts of North Africa. After the nuptial flight, the Bothriomyrmex queen sheds her wings and searches over the ground until she finds a Tapinoma nest. She allows herself to be accosted by the aroused Tapinoma workers and dragged by them into the interior of their nest. There she takes refuge among the brood or on the back of the Tapinoma queen. In time, she settles down for good on the back of the host queen and begins the one act for which she is uniquely specialized: slowly cutting off the head of her victim. When this is accomplished, sometimes only after many hours, the Bothriomyrmex takes over as the sole reproductive, and the colony eventually comes to consist entirely of her offspring and herself. Recently Lloyd et al. (1986) discovered that the pygidial glands of Bothriomyrmex syrius queens and of the Tapinoma simrothi host workers contain the same ketone. They speculate that the odor identity assists the Bothriomyrmex queens when they attempt to penetrate the host colonies.

A similar mode of entry into host nests is employed by the myrmicine Monomorium santschii, also a native of North Africa and a permanent workerless parasite of Monomorium salomonis. In this case, however, it is the salomonis workers who destroy their own queen. They then adopt the santschii queen as the sole reproductive in her place (F. Santschi, in Forel, 1906).

The dulosis route
Slavery in ants, particularly as practiced by Polyergus rufescens and the species of the Formica sanguinea group, has been a favorite subject of myrmecologists in Europe and the United States ever since it was originally described by Pierre Huber in 1810. Pierre-André Latreille (1805) was first to note the large number of workers emerging from Polyergus nests in “une espèce d'ordre de bataille” but apparently did not recognize the nature of the slavery that resulted. Darwin was fascinated by the phenomenon, and in his book On the Origin of Species he offered the first hypothesis of how it originated in evolution. The ancestral Formica, he proposed, began by raiding other species of ants in order to obtain their pupae for food. Some of the pupae survived in the storage chambers long enough to eclose as workers, whereupon they were accepted by their captors as nestmates. This fortuitous addition to the work force helped the colony as a whole, and consequently there was an increasing tendency, propelled by natural selection, to raid other colonies solely for the purpose of obtaining slaves. If Darwin's explanation seems at first a bit farfetched, it is only commensurate with the phenomenon itself. Several authors, most notably Erich Wasmann (1905), rejected Darwin's hypothesis on various grounds, chiefly a priori in nature. But the years have brought an increasing amount of confirmation of an evolutionary sequence approximately consistent with Darwin's scheme, through the discovery of species whose behavior collectively bridges the gap between free-living and slavemaker species in ever shorter, more plausible steps.

Before examining the details of behavior across the dulotic species, it will be helpful to recognize the existence today of three hypotheses concerning the origin of slavery:

(1) Predation. As suggested by Darwin, the predulotic ancestral ants raided the nests of other species to obtain prey. The first slaves were prey items allowed to live.

(2) Territory. Part of territorial exclusion practiced by predulotic ants is the invasion of the nests of rivals and the robbing of the brood. The brood pieces, including eggs, larvae, and pupae, would be routinely eaten for the most part, but some might survive long enough to join their captors as slaves. The first step in this sequence would be territorial aggression among colonies belonging to the same species. The next step, leading to the earliest stage in social parasitism as traditionally conceived, would be territorial aggression directed at colonies of other species. The more closely related the rival species, the more likely the captives would be tolerated, leading to Emery's rule. The territorial hypothesis was developed by Wilson (1975a), Alloway (1979, 1980), and Stuart and Alloway (1982, 1983).

(3) Transport. In the suggestion originally made by Buschinger (1970), slave raiding evolved as the outcome of regular brood transport among the nests of single polydomous colonies. If the habit pattern is extended to less familiar populations of the same species or to other species, it will create an early version of dulosis.

These models are not mutually exclusive. They merely draw attention to the existence of three propensities that predispose the species to the practice of slavery. They are logically linked as follows: territorial raids combined with a strong propensity to transport brood lead to the regular retrieval of alien brood back to the raiders' nest; the raiders destroy and eat most of the captives; but a few survive to join the colony as slave workers.

This synthetic model identifies territorial behavior rather than predatory raids as the prime mover leading to slavery, with food being a secondary benefit to the raiders. A substantial amount of evidence appears to support this view. First, none of the ant groups that specialize in predation on other ants, including one branch each in Myrmecia and Gnamptogenys, the entire tribe Cerapachyini of the subfamily Ponerinae, and Aenictus, Eciton, and Neivamyrmex among the army ants, has produced a single slavemaking species. On the other hand, slavemaking is rampant in the tribe Leptothoracini, the species of which are generalized insectivores and honeydew collectors. Leptothoracines limit their myrmecophagy (if indeed it occurs in nature) mostly to other colonies of their own species, in what appears to be an incidental outcome of territorial aggression. Wilson (1975a) showed that when colonies of Temnothorax curvispinosus are placed close to one another in the laboratory, workers from the larger colony attack the smaller colony, expelling the queen and workers. They carry away the brood and allow at least some of the pupae to eclose into adults and join them as nestmates. L. (= M.) ambiguus colonies raided smaller Temnothorax curvispinosus colonies in the same manner. They allowed curvispinosus pupae to eclose, and at first licked them and treated them like sister workers. But within one or two days they dragged them out of the nest and killed them. No curvispinosus worker was permanently adopted. Even so, the behavior of L. ambiguus, a typical free-living leptothoracine, was revealed to be but one short step away from an elementary form of dulosis.

Later Alloway (1980) discovered mixed colonies of Leptothorax (= Myrafant) in Canada and the northern United States. Most contained a queen and majority of workers of Leptothorax longispinosus with a scattering of workers belonging to either Leptothorax ambiguus or Leptothorax curvispinosus. One was a colony of L. ambiguus with a longispinosus worker, another a curvispinosus colony with both ambiguus and longispinosus workers. When cultured, almost all the mixed colonies produced additional workers of the majority species. Alloway was able to create similar mixed colonies by placing pure colonies of the Leptothorax species close enough in the laboratory to trigger territorial raids of the kind described earlier by Wilson. In other words, a natural condition of low-intensity, facultative slavery already exists in free-living species of Leptothorax, and it is evidently the outcome of territorial behavior. Alloway's result must be treated with caution, however, because his colonies were fed only Bhatkar diet. Buschinger and Pfeiffer (1988) have shown that the diet is deficient in protein causing the ants to consume more brood and perhaps increasing their propensity to raid neighboring colonies.

The connection between territoriality and predation is also well established. Even specialized slavemaking species consume substantial portions of the prey, as documented in Temnothorax duloticus by Alloway (1979) and Polyergus breviceps by Topoff et al. (1984).

A second line of evidence supporting the primacy of territoriality was found by Stuart and Alloway (1983) in a laboratory study of Harpagoxenus canadensis], a specialized slave-raider, and its slave [[Leptothorax muscorum.” The two species display similar raiding behavior, and both carry brood back to the nests after the raids.  However, Harpagoxenus canadensis rear the captured brood to produce slave workers, while Leptothorax muscorum colonies mutilate the immature forms and feed them to their own larvae.  The Leptothorax slaves then join their Harpagoxenus mistresses in raids against other Leptothorax colonies.  The only important difference between free and enslaved Leptothorax workers was the willingness of the enslaved individuals to care for captured brood instead of destroying it.  Not only are the Harpagoxenus and Leptothorax very similar in this respect, but their method of recruitment is identical.  Both recruit nestmates to the scene of fighting among workers, rather than to newly discovered nest sites, the usual stimulus triggering raids by other species of leptothoracine slavemakers.  It is possible that Harpagoxenus canadensis arose directly from Leptothorax muscorum or from an immediately common stock (Buschinger, 1981).

The same facultative, low-level of slavery appears to occur in Formica, another genus already known to be prone to advanced dulosis. Scherba (1964) found that in Wyoming the dense, polydomous colonies of Formica opaciventris, a member of the exsecta group, commonly oust colonies of Formica fusca from their nest sites by laying siege to them and robbing larvae and pupae when they get the chance. When Kutter (1957) placed colonies of Formica naefi, also a member of the exsecta group, near colonies of species belonging to the fusca group, the naefi attacked their neighbors, penetrated their nest, and carried away both the brood and the adult workers. Kutter was unsure whether such behavior occurs in nature, but he noted that all larger naefi colonies observed in the field contain a few fusca-group workers. It seems reasonable to suggest that naefi represents the first interspecific dulotic stage envisioned in the territorial hypothesis.

It is also possible that the phenomenon described by Kutter occurs in other Formica. King and Sallee (1957, 1962) reported the puzzling existence of natural mixed colonies of Formica clivia and Formica fossaceps that persisted over a period of up to 16 years in Iowa. Both workers and sexuals of the two forms were produced in the nests. King and Sallee believed that the two forms are either genetic morphs or distinct species linked in some aberrant and unexplained symbiosis. The field data strongly suggest the second alternative. In laboratory experiments small homospecific groups of workers readily accepted queens of the opposite species combined with alien worker groups. The significance of this permissiveness, and the nature of the interaction of colonies of the same species in nature, are clearly promising subjects for future study.

It is traditional to use the expression “slavery” for the exploitation of one species by another. In the human sense, this is not slavery but more akin to the forcible domestication of dogs and cattle by humans. Does true slavery--the use of captives of the same species--exist in the ants? Evidence from the laboratory experiments just cited indicate that it occurs as an accidental outcome of territorial raiding in Leptothorax (=Myrafant). On the other hand, true slavery is practiced as a highly organized, evidently adaptive behavior in the honeypot ant Myrmecocystus mimicus (Hölldobler, 1976c, 1981a). The foraging grounds of neighboring mimicus colonies often overlap, setting off massive territorial confrontations. The conflicts do not consist of deadly physical fights but rather of elaborate tournaments in which very few ants are injured. The rival workers stilt-walk on extended legs, lift their abdomens and point the tips at each other, and drum their antennae on one another's abdomens (see Figure 10-23). The Myrmecocystus tournaments sometimes last for days. If one colony is considerably weaker than the other and therefore unable to recruit a large enough worker force to the tournament area, it is eventually overrun and raided by the larger colony. Workers of the winning colony kill or drive off the queens and carry or drag the larvae, pupae, and callow workers to their own nests (see Figure 10-21). Raids do not originate exclusively from territorial interactions. New evidence shows that scout ants recruit nestmates to newly discovered small conspecific colonies, which are subsequently raided by the larger colony. Familiar behavior has been observed among colonies of Myrmecocystus depilis, which often occur sympatrically with Myrmecocystus mimicus. No cases have been recorded, however, in which Myrmecocystus mimicus raided Myrmecocystus depilis or the reverse.

The raiding of smaller colonies by larger conspecific or congeneric colonies is probably much more common in territorial interactions in ants than previously assumed. In a recent survey of colony interactions in populations of Pogonomyrmex, Markl and Hölldobler (unpublished) observed several incidents of intra- and interspecific raids, which in some cases clearly led to the enslavement of the captured immature stages. This explains the occurrence of Pogonomyrmex colonies with mixed species worker populations that have occasionally been noted.

The next step in the dulotic progression that has been well documented is Formica sanguinea (Figure 12-5), a European species that has been thoroughly studied by Huber (1810), Forel (1874), Wasmann (1891), Dobrzanski (1961, 1965), and others. The “sanguinary ants” are very aggressive and territorial, dominating the local spots near their nests that are richest in food. They are “facultative slave-holders,” in Wasmann's terminology, since colonies are sometimes found with no slaves present. Also, workers isolated in laboratory nests are able to conduct all of the affairs of colony life, including nest construction, in a competent manner. According to Wheeler (1910a) the percentage of slaveless colonies in different populations of sanguinea varies enormously, from about 2 percent to 98 percent. Thus the sanguinea are far more committed to dulosis than the species of Leptothorax (= Myrafant), which take captives only rarely and apparently by accident. The commonest slaves taken by the sanguinea belong to the fusca group (“Serviformica”) and include fusca, lemani, and rufibarbis; less commonly exploited are gagates, cunicularia, transkaukasica, and cinerea, all of which are also members of the fusca group as conceived in the broadest sense. On rare occasions workers of the rufa group, in particular nigricans and rufa, have been found in sanguinea nests, but always in the company of fusca-group slaves (Bernard, 1968). As a rule, sanguinea colonies enslave the fusca-group species nearest their nest, and the seeming preferences are merely a reflection of local relative abundance of the slave species. Two or even three slave species are sometimes present in a given sanguinea nest simultaneously, and the composition of slaves may change from year to year.

The raids of sanguinea have been lucidly described by Wheeler (1910a: 456-457): The sorties occur in July and August after the marriage flight of the slave species has been celebrated and when only workers and mother queens are left in their formicaries. According to Forel the expeditions are infrequent—“scarcely more than two or three a year to a colony.” The army of workers usually starts out in the morning and returns in the afternoon, but this depends on the distance of the sanguinea nest from the nest to be plundered. Sometimes the slave-makers postpone their sorties till three or four o'clock in the afternoon. On rare occasions they may pillage two different colonies in succession before going home. The sanguinea army leaves its nest in a straggling, open phalanx sometimes a few meters broad and often in several companies or detachments. These move to the nest to be pillaged over the directest route permitted by the often numerous obstacles in their path. As the forefront of the army is not headed by one or a few workers that might serve as guides, but is continually changing, some dropping back while others move forward to take their places, it is not easy to understand how the whole body is able to go so directly to the nest of the slave species, especially when this nest is situated, as is often the case, at a distance of 50 or 100 m. ..

When the first workers arrive at the nest to be pillaged, they do not enter it at once, but surround it and wait till the other detachments arrive. In the meantime the fusca or rufibarbis scent their approaching foes and either prepare to defend their nest or seize their young and try to break through the cordon of sanguinea and escape. They scramble up the grass-blades with their larvae and pupae in their jaws or make off over the ground. The sanguinary ants, however, intercept them, snatch away their charges and begin to pour into the entrances of the nest. Soon they issue forth one by one with the remaining larvae and pupae and start for home. They turn and kill the workers of the slave-species only when these offer hostile resistance. The troop of cocoon-laden sanguinea straggle back to their nest, while the bereft ants slowly enter their pillaged formicary and take up the nurture of the few remaining young or await the appearance of future broods.

The communicative signals that trigger and orient the raids of colonies belonging to the sanguinea group of slavemaking ants were identified, at least in part, by Fred E. Regnier and E. O. Wilson (in Wilson, 1971). They found that workers of the American species Formica rubicunda readily follow artificial odor trails made from whole body extracts of rubicunda workers and applied with a camel's hair brush over the ground in the vicinity of the nest. When the trails were drawn away from the nest opening in the afternoon, at about the time raids are usually conducted, the rubicunda workers showed behavior that was indistinguishable from ordinary raiding sorties. They ran out of the nest and along the trail in an excited fashion, and, when presented with colony fragments of a slave species (Formica subsericea), they proceeded to fight with the workers and to carry the pupae back to their nest. It seems likely that under normal circumstances rubicunda scouts lay odor trails from the slave colonies they discover to the home nest, and the raids result when nestmates follow the trails out of the home nest back to the source. In addition, the scouts sometimes travel at the head of the raiding column. Chemical trails are probably the general mode of communication among slavemaking ants. As we shall see shortly, they are the technique employed by the evolutionarily more advanced amazon ants of the genus Polyergus, as well as some myrmicine slavemakers (Buschinger et al., 1980a). The tendency of Formica sanguinea to fan out into “phalanxes” in their outward march does not conflict with this interpretation; there could be several odor trails involved, around which orientation is less than perfect, or else the recruits swarm loosely around the leader ant--as in Polyergus breviceps (Topoff et al., 1984, 1985a-c).

The general biology and raiding behavior of Formica subintegra, an American member of the sanguinea group, have been studied by Wheeler (1910a) and by Talbot and Kennedy (1940). The latter investigators, by keeping a chronicle over many summers of a population on Gibraltar Island, in Lake Erie, were able to show that raiding is much more frequent in subintegra than in sanguinea. Some colonies raided almost daily for weeks at a time, striking out in any one of several directions on a given day. Occasionally the forays continued on into the night, in which case the subintegra workers remained in the looted nest overnight and returned home the next morning. In other details the raiding behavior resembled that of sanguinea. Subsequently, Regnier and Wilson (1971) discovered that each subintegra worker possesses a grotesquely hypertrophied Dufour's gland, which contains approximately 700 µg of a mixture of decyl, dodecyl, and tetradecyl acetates. These substances are sprayed at the defending colonies during the slave raids. They act at least in part as “propaganda substances” because they evaporate slowly and help to alarm and to disperse the defending workers (see Figure 12-6).

Little is known about the other nine or so American species of the sanguinea group (Creighton, 1950; Buren, 1968a), and their study is likely to reveal new behavioral phenomena related to dulosis. For example, a colony of Formica wheeleri that Wilson (1955c) observed in Yellowstone Park, Wyoming, divided its labor in a remarkable fashion between two species of slaves. Formica neorufibarbis accompanied the wheeleri on a raid (against colonies of Formica fusca and Formica lasioidessimultaneously) and assisted them in excavating and breaking into the plundered nests. Later, when the mixed nest was excavated for closer examination, the neorufibarbis were found to be concentrated in the middle and upper layers. They were very aggressive and joined the wheeleri in defending the nest. The workers of the second slave species, Formica fusca, did not accompany the slavemakers on the raid, and later they made only feeble attempts to defend the nest. Instead, they were found concentrated in the lower layers of the nest close to the brood, and most had their crops distended with liquid food. These circumstances suggest that the fusca workers were specializing on food storage and brood care. A deeper significance of the dulotic habit is indicated by this example. It is apparent that the slavemaker colony not only adds to its labor force quantitatively by taking slaves, but it can also incorporate specialists that increase the efficiency of the colony in a fashion analogous to that seen in normal worker polymorphism.

The mode of colony founding by queens of the Formica sanguinea group has not been observed in nature, and this surprising gap in our information continues to prevent a secure understanding of the evolutionary origins of dulosis. Wheeler (1906) conducted a series of laboratory experiments on the American species Formica rubicunda which strongly indicate that the queens can function at least facultatively as temporary parasites. When he placed newly dealated (but still virgin) rubicunda queens in nests containing workers and brood of Formica fusca, they responded in an aggressive and effective manner. They advanced on the fusca colonies, fighting and killing fusca workers that attacked them, then seizing and sequestering the fusca pupae, until finally all of the fusca workers were dead and the rubicunda queens stood guard over the confiscated brood. When new fusca workers emerged from the brood pile at a later date, they accepted the rubicunda queens and began to lick and to feed them. Viehmeyer (1908) and Wasmann (1908) subsequently repeated Wheeler's experiment with young mated queens of Formica sanguinea and obtained the same result. The behavior of the intruding queens differs markedly from those belonging to the fusca and microgyna groups used in parallel experiments. There is no reason to doubt that, at least under certain conditions, the sanguinea-group queens do start new colonies by this form of unaided assault on colonies of slave species. Wheeler, in his early writings, and later Santschi (1906) and Wasmann (1908), believed that such temporary parasitism not only characterized the ancestors of the slavemaking Formica but was a prerequisite for the evolution of the dulotic habit itself. Together they postulated this explanation as an alternative to the Darwinian predation hypothesis, believing that, once predatory habits evolved in the queen during nest founding, it was far easier for the species to extend such behavior to the worker caste in the form of raiding for slaves. Later, Wheeler (1910a) saw the incongruity in his position, namely, that dulosis represents a wholly new behavior pattern that cannot be viewed simply as a variant of the temporarily parasitic mode of colony founding. He concluded, “In my opinion both temporary parasitism and dulosis have arisen independently from the practice of Formica rufa and Formica sanguinea of adopting fertilized queens of their own species. . .” This opinion seems about right at the present time. Neither the predation nor territorial hypotheses are excluded by the demonstration of temporary social parasitism in the sanguinea group; they are moreover, considerably strengthened by the growing evidence of raiding and accidental dulosis in Formica naefi and Formica sanguinea, as we pointed out earlier. It is still not known to what extent the various species of the sanguinea group rely on temporary parasitism to start new colonies, as opposed to homospecific adoption followed by budding.

The pinnacle of the slave-holding way of life (or nadir if you prefer) is reached in the formicine genus Polyergus, a totally dulotic group of species that have evidently been phylogenetically derived from Formica (see Plate 15). Five species are known: rufescens of Europe and North America, breviceps and lucidus of North America, nigerrimus of the Soviet Union, and samurai of Japan and eastern Siberia (Figures 12-7 and 12-8). These “amazon ants” are nowhere very common, but their striking appearance (large size, bright red or black coloration, and shining body surface), the extraordinary degree of their behavioral specialization, and the spectacular qualities of their slave raids have placed them among the most frequently studied of all the ants from the time of Latreille (1805) and Huber (1810) onward. As usual, no one has ever approached Wheeler's ability to distill the important information in the form of a gripping narrative, and we will again defer to him. In the following passage he describes Polyergus rufescens (Wheeler 1910a: 472-473):

The worker is extremely pugnacious, and, like the female, may be readily distinguished from the other Camponotine ants by its sickle-shaped, toothless, but very minutely denticulate mandibles. Such mandibles are not adapted for digging in the earth or for handling thin-skinned larvae or pupae and moving them about in the narrow chambers of the nest, but are admirably fitted for piercing the armor of adult ants. We find therefore that the amazons never excavate nests nor care for their own young. They are even incapable of obtaining their own food, although they may lap up water or liquid food when this happens to come in contact with their short tongues. For the essentials of food, lodging and education they are wholly dependent on the slaves hatched from the worker cocoons that they have pillaged from alien colonies. Apart from these slaves they are quite unable to live, and hence are always found in mixed colonies inhabiting nests whose architecture throughout is that of the slave species. Thus the amazons display two contrasting sets of instincts. While in the home nest they sit about in stolid idleness or pass the long hours begging the slaves for food or cleaning themselves and burnishing their ruddy armor, but when outside the nest on one of their predatory expeditions they display a dazzling courage and capacity for concerted action compared with which the raids of sanguinea resemble the clumsy efforts of a lot of untrained militia. The amazons may, therefore, be said to represent a more specialized and perfected stage of dulosis than that of the sanguinary ants. In attaining to this stage, however, they have become irrevocably dependent and parasitic. Wasmann believes that Polyergus is actually descended from Formica sanguinea, but it is more probable that both of these ants arose in pretertiary times from some common but now extinct ancestor. The normal slaves of the European amazons are the same as those reared by sanguinea, viz: Formica fusca, glebaria, rubescens, cinerea, and rufibarbis; and of these fusca is the most frequent. But the ratio of the different components in the mixed nests is the reverse of that in sanguinea colonies, there being usually five to seven times as many slaves as amazon workers. The simultaneous occurrence of two kinds of slaves in a single nest is extremely rare, even when the same amazon colony pillages the nests of different forms of fusca during the same season. This is very probably the result of the slaves' having a decided preference for rearing only the pupae of their own species or variety and eating any others that are brought in. Two slave forms may, however, appear in succession in the same nest. Near Morges, Switzerland, Professor Forel showed me an amazon colony which during the summer of 1904 contained only rufibarbis slaves, but during 1907 contained only glebaria.

Unlike sanguinea, rufescens made many expeditions during July and August, but these expeditions are made only during the afternoon hours. One colony observed by Forel (1874) made 44 sorties on thirty afternoons between June 29 and August 18. It undoubtedly made many more which were not observed, as Forel was unable to visit the colony daily. . . . Forel estimated the number of amazons in the colony at more than 1,000 and the total number of pupae captured at 29,300 (14,000 fusca, 13,000 rufibarbis, and 2,300 of unknown provenience, but probably fusca). The total number for the summer (1873) was estimated at 40,000. This number is certainly above the average, as the amazon colony was an unusually large one. Colonies with only 300 to 500 amazons are more frequent, but a third or half of the above number of pillaged cocoons shows what an influence the presence of a few colonies of these ants must have on the Formica colonies of their neighborhood. Of course, only a small proportion of the cocoons are reared. Many of them are undoubtedly injured by the sharp mandibles of the amazons and many are destroyed and eaten after they have been brought home.

The tactics of Polyergus, as I have said, are very different from those of sanguinea. The ants leave the nest very suddenly and assemble about the entrance if they are not, as sometimes happens, pulled back and restrained by their slaves. Then they move out in a compact column with feverish haste, sometimes, according to Forel, at the rate of a meter in 33 1/3 seconds or 3 cm. per second. On reaching the nest to be pillaged, they do not hesitate like sanguinea but pour into it at once in a body, seize the brood, rush out again and make for home. When attacked by the slave species they pierce the heads or thoraces of their opponents and often kill them in considerable numbers. The return to the nest with the booty is usually made more leisurely and in less serried ranks.

The means by which the Polyergus workers are able to mobilize themselves within minutes and run in a compact column straight for the target colony was for a long time one of the classic problems of entomology. While watching Polyergus lucidus colonies in Michigan, Talbot (1967) noticed that, prior to the onset of each raid, several scout workers explored the surrounding terrain, including the vicinity of the specific nest later raided. By monitoring the Polyergus nests carefully, she saw that the beginning of the raid was often signaled by the appearance of a scout returning from the direction of the target nest. As she describes it, “On other days the departure of scouts was less conspicuous, and seldom was one lucky enough to spot a scout coming in. But whenever an ant came in hurriedly from the grass and went directly into the nest, there was an outpouring of ants. It was thus assumed that whenever a sudden emergence occurred it was in response to a messenger arriving with news of a located colony. If this was correct and if the scouting ant, which found a colony, laid down an odor trail on its way home, then the odor must have been quite long lasting, for it sometimes took an ant 30 to 45 minutes to return from a raided nest. It seemed unlikely that a raiding group could be following anything but an odor trail, for it moved rapidly, did not maintain leaders, and usually stopped at exactly the right place.”

The next logical step was to try to induce false raids by means of the artificial trail test. Talbot accomplished this in a manner that decisively favored her hypothesis. When she laid down dichloromethane extracts of whole Polyergus bodies over the ground along an arbitrary path away from the nest and at the time of day raids normally occur, Polyergus workers poured from the nest and followed the trails to the end. Thus Talbot was able to activate the raid swarms at will and lead them to targets of her choosing. Finally, she induced a complete raid on a colony of Formica nitidiventris by placing it in a box two meters from a Polyergus colony and drawing an artificial Polyergus trail to it. Talbot concluded that the Polyergus raid phalanxes do not contain leaders, but this may not be correct. Although it is true that excited Polyergus are capable of moving out along trails without further guidance, Topoff et al. (1984, 1985a,b) recently demonstrated that in the case of Polyergus breviceps at least, naturally occurring raids are always led by scout ants. Experiments further revealed that visual cues are more important to the leaders than are the chemical trails. After the attack, however, Polyergus breviceps workers orient homeward by a combination of visual guideposts and the chemical trails laid by the outward bound leaders.

Emery (1911a), by employing introduction experiments of the kind invented by Wheeler for studies of temporary parasitism, discovered that the queens of Polyergus rufescens act essentially like those of other parasitic groups during colony founding. When presented with a colony of Formica fusca in the laboratory, the rufescens queen works her way into the nest by submissive posturing and secures adoption by the fusca workers and queen. Then, after a week or so has elapsed, she kills the fusca queen by piercing her head with her sharp mandibles. The frequency with which this mode of colony formation is used in nature is not known. It must occur at least occasionally since single dealated Polyergus rufescens queens have been found alone in small Formica fusca nests on at least three occasions, the most recent at Aosta, Italy, by Buschinger (personal communication).

The American amazon ant Polyergus lucidus is evidently similar in most respects to the European Polyergus rufescens in colony foundation. Goodloe and Sanwald (1985) observed newly mated queens penetrate and secure adoption in queenless colony fragments of Formica nitidiventris and Formica schaufussi. The queens were given a choice of the two host species, and in all of 13 penetrations they entered a colony fragment of the same species used as slaves by their colony of origin. In addition, colonies of Polyergus lucidus can reproduce by budding (Marlin, 1968). Some of the queens return to their home nests following the nuptial flights. Later they accompany workers on a raid and remain behind with a few of them in a plundered Formica nest or in some neighboring nest site.

Recently Topoff and his co-workers (personal communication) conducted a series of laboratory experiments in which newly mated queens of Polyergus breviceps were allowed to invade colonies of their host species, Formica gnava. The Polyergus queens quickly detected the Formica queen and killed her. Surprisingly the aggressive behavior of the Formica workers towards the invading Polyergus queen subsided shortly after their own queen was killed. Topoff and his co-workers suggest that the Polyergus queen releases an appeasement substance from her enlarged Dufour's gland. When Pogonomyrmex workers, which served as test objects, were contaminated with the Dufour's gland secretions of Polyergus queens, they were ignored by the Formica workers, while untreated Pogonomyrmex were attacked.

In 1932 K. V. Arnoldi reported the discovery of a new and equally spectacular kind of formicine slavemaking ant. The species, Rossomyrmex proformicarum, superficially resembles Polyergus, but has evidently been derived from Formica-like ancestors in a line separate from that leading to the amazon genus. His observations were later supplemented in considerable detail by Marikovsky (1974). Rossomyrmex proformicarum is locally common in the semideserts and dry Artemisia-Festuca steppes over a thousand-kilometer-wide area of Soviet Central Asia. It enslaves Proformica epinotalis and Proformica nasuta, which are closely related to the genus Formica and abundant in xeric habitats. The method of raiding is unique. After a long reconnaissance one or a few scouts begin to transport some of their nestmates from the home nest to the Proformica nest. The scout seizes a fellow worker by its mandibles, whereupon the latter folds up its legs, tucks under its abdomen, and allows itself to be carried in the typical formicine fashion. Fifty or more such pairs set off in a loose file for distances of as much as 50 meters. They then halt, uncouple, and search for the entrances to the Proformica nests. As soon as the Rossomyrmex approach, the Proformica close the exits with particles of earth. The slavemakers may require several hours to break through the barrier. They then make short work of the defenders and begin to carry away brood. Whereas other slavemaking species, including Formica and Polyergus, usually steal only pupae, the Rossomyrmex proformicarum raiders steal all stages of brood--eggs, larvae, and pupae. The pairwise carrying behavior used in the raids is basically the same as that used by Formica during colony movements from one nest site to another. The Rossomyrmex have simply adapted it to a new function. It is reminiscent of the “emigration raids” described by Kwait and Topoff (1983) in Polyergus lucidus, which occur at the end of the raiding season. The Polyergus workers gather into a swarm, “raid” their old nest, and carry the adult Formica slaves to a new site. An intermediate stage in the evolution of the Rossomyrmex habit is displayed by the slavemaker Formica wheeleri, whose workers have been observed carrying one another back to the home nest following a raid (Wilson, 1955c).

Even more remarkable in another sense is the existence of a phylogenetically independent form of dulosis in the myrmicine genera Chalepoxenus, Epimyrma, Harpagoxenus, Leptothorax (= Myrafant), Myrmoxenus, Protomognathus (recently separated as a genus from Harpagoxenus), and Strongylognathus. The Protomognathus case is the most specialized and also by far the best understood. The single known species, Protomognathus americanus of North America, has been closely studied by Sturtevant (1927), Creighton (1929), and especially Wesson (1939), whose analysis of the life cycle and behavior is a model of its kind. Alloway (1979) and Alloway and Del Rio Pesado (1983), have added important details of the raiding behavior. The Protomognathus workers are small, blackish brown ants superficially resembling some of the Leptothorax (= Myrafant) species they enslave. Their most distinguishing feature is the presence of “antennal scrobe”--long, deep pits along the sides of the head into which the antennae are folded for protection during the raids (Figure 12-9). Protomognathus americanus is a relatively widespread species, existing in very local but dense populations from Ontario south to Virginia and west to Ohio. It enslaves three of the commonest Leptothorax (= Myrafant) species of eastern North America, L. ambiguus, Leptothorax curvispinosus, and Leptothroax longispinosus. In the populations studied by Wesson, the ratio of Protomognathus americanus mixed colonies to local pure Leptothorax colonies was about 1:15. The mixed colonies contained up to 50 Protomognathus workers and 300 Leptothorax worker slaves. Most colonies were much smaller than this, the medians being about 6 Protomognathus and 30 Leptothorax, respectively.

Following the nuptial flight in early or middle July, the newly fecundated queen sheds her wings and crawls about on the ground or low vegetation in search of a Leptothorax nest. On encountering a nest, she begins quite literally to throw the Leptothorax adults out. As each worker approaches her in turn, she seizes it by an antenna, drags it out of the nest entrance, and flings it to one side. She avoids attacks on her own body by very rapid, shifting movements. After she has savaged the colony in this manner for a while, the Leptothorax queen and workers finally panic and desert the nest. The Protomognathus queen then appropriates the larvae and pupae left behind.

When Leptothorax workers later eclose from the pirated brood, they adopt the Protomognathus queen without hesitation, and soon afterward she begins to lay eggs. The Protomognathus workers that develop from these eggs are degenerate in behavior. They spend almost all of their time in the nest grooming each other and “loafing.” They are fed with liquid food regurgitated to them by the Leptothorax slaves, who also assume the thankless tasks of foraging, nest construction, and brood care. When the Protomognathus depart on the slave raids, on the other hand, they reveal themselves to be efficient little fighting machines. The raids are initiated by scouts, who hunt singly or in small groups for Leptothorax colonies in the vicinity of the Protomognathus nest. When a Protomognathus scout encounters a Leptothorax nest, it normally attempts to penetrate the entrance without hesitation. If the colony is small and weak, it may succeed in capturing it single-handedly, after which it begins to transport the Leptothorax brood back to its own nest. If, on the other hand, the scout is repulsed, it returns to its nest, excites its nestmates (evidently by release of a pheromone), and soon sets out again for the newly discovered Leptothorax nest. This time it lays down a short-lived odor trail which draws out a tight little column of Protomognathus workers and Leptothorax slaves. If this group is still not sufficient to breach the Leptothorax nest, some of the Protomognathus workers return to the home nest and bring out auxiliary columns. A similar recruitment behavior during slave raids has been observed in Leptothorax duloticus (Alloway, 1979), Epimyrma ravouxi (= E. goesswaldi)  (Buschinger et al., 1980a; see Figure 12-10), and Myrmoxenus gordiagini (Buschinger et al., 1983).

Toward the end of the summer, the raids are transmuted into an unusual form of colony multiplication. An increasing tendency develops for some of the Protomognathus workers to remain behind in the conquered nests, where they stand guard over the Leptothorax brood. When this happens, the expatriates soon lose contact with the home nest, and they are treated as queens by the Leptothorax workers who subsequently eclose from the pupae. As Buschinger and Alloway (1977) showed, the eggs laid by the Protomognathus workers are unfertilized (the workers in fact lack spermathecae) and give rise to males, possibly along with a few workers. Such secondary colonies are very common and even rival in number the primary colonies started by single Protomognathus queens. Of 32 colonies censused by Wesson in Maryland, Ohio, and Pennsylvania, no fewer than 16 were populated exclusively by Protomognathus workers together with their slaves.

The genus Harpagoxenus is outwardly very similar to Protomognathus in physical appearance but evidently independently derived in evolution. It is closest to species constituting Leptothorax in the strict sense and enslaves some of them, whereas Protomognathus resembles and enslaves species of Leptothorax sometimes distinguished as the genus or subgenus Myrafant. The European Harpagoxenus sublaevis (Figure 12-11) has been examined in successively greater detail by Adlerz (1896), Viehmeyer (1921), Buschinger (1966a,b, 1968a,b), Buschinger and Winter (1977), and Winter (1979a), while the North American Harpagoxenus canadensis has more recently been studied by Stuart and Alloway (1983). Both Harpagoxenus species, as well as Chalepoxenus muellerianus, recruit nestmates by tandem running. As soon as the scout ant returns from a newly discovered host nest, she displays an invitation behavior. She then leads a single nestmate to the host nest (Figure 12-12). “The follower ant may stay near the host nest, or return, like the scout, to the mother nest in order to recruit further nestmates. This system often involves a rather long ‘siege’ of the host nest, until the number of warriors is high enough to risk a direct attack” (Buschinger et al., 1980a). The slavemaking Leptothorax duloticus was discovered by Wesson (1937) in Ohio and later studied by Talbot (1957), Wilson (1975a), and Alloway (1979). Its basic parasitic behavior is similar to that of Protomognathus and Harpagoxenus. The same is true of the European leptothoracines Chalepoxenus (Ehrhardt, 1982), Epimyrma (Buschinger and Winter, 1985; Buschinger, 1986), and Myrmoxenus (Buschinger et al., 1983).

Raiding workers and colony founding queens of Harpagoxenus sublaevis appear to employ chemical weapons when invading the nest of the host species Leptothorax acervorum. Buschinger (1968b, 1974d) found that the Harpagoxenus have hypertrophied Dufour's glands (Figure 12-13). He observed Leptothorax acervorum workers attacking each other following contact with Harpagoxenus sublaevis that were either conducting slave raids or founding colonies (Figure 12-14). He suggested that Harpagoxenus discharges a propaganda pheromone from the Dufour's gland. Recently Allies et al. (1986) confirmed Buschinger's hypothesis and also demonstrated in laboratory experiments that the workerless inquiline ant Doronomyrmex kutteri (which also possesses hypertrophied Dufour's glands; see Buschinger, 1974b) uses Dufour's gland secretions as a chemical weapon in defense against hostile workers of Leptothorax acervorum. Host workers, contaminated with Dufour's gland secretions of Doronomyrmex kutteri queens, attack each other. They evidently no longer recognize each other as nestmates. It is interesting to note that another inquiline parasite of Leptothorax acervorum, Doronomyrmex pacis, also possesses a hypertrophied Dufour's gland. A third inquiline species, Doronomyrmex goesswaldi, has the Dufour's gland only slightly enlarged (Buschinger, 1974b).

Thanks largely to the outstanding efforts of Buschinger and his collaborators, we now know that the species of Epimyrma form a series of steps from full dulotic behavior to workerless inquilinism--or perhaps more precisely, degenerate dulosis, since unlike true inquilines the parasite queens continue to assassinate the host queens (Buschinger, 1982a, 1986; Buschinger and Winter, 1983b). The earlier view (Wilson, 1971) that the genus followed the temporary parasitism route to inquilinism is incorrect. Eight species in this European and North African genus are known at the present time. Myrmoxenus gordiagini, a dulotic ant of the Soviet Union and Yugoslavia, is very close morphologically and behaviorally and very likely constitutes a ninth species. Three of the Epimyrma species, including algeriana, ravouxi (= goesswaldi), and stumperi, conduct well organized slave raids with group recruitment and fight with a powerful sting. In a second group, including kraussei, the number of workers is reduced drastically to five or fewer. The workers can still raid, but their efforts are ineffectual. Finally, at least two species are completely workerless, corsica and a still undescribed species from Greece (Buschinger, personal communication). A simultaneous evolution in sexual behavior has occurred, from normal nuptial flights to mating within the nest and hence inbreeding. Three of the fully dulotic species conduct the flights. A fourth, algeriana, as well as all of the degenerate slavemakers and workerless species, mate inside the nest.

It is a remarkable fact that so far as known, the queens of all of the Epimyrma species, including the workerless Epimyrma corsica, found new colonies by forcibly entering the host nest and throttling the resident queen. The details have been worked out over many years in studies by Gösswald (1933), Kutter (1951, 1969), and Buschinger and his associates. After shedding her wings, the queen searches for a new host colony. The mode of entry and subsequent behavior vary greatly among the various species. The queen of Epimyrma kraussei, upon approaching a Leptothorax recedens colony, makes repeated hostile approaches to the host workers and “intimidates” them, to use Kutter's expression. If she succeeds in entering the nest, she kills the host queen and secures complete adoption by the rest of the colony. The queen of |Epimyrma ravouxi (= goesswaldi) , on the other hand, calms the host workers (also Leptothorax unifasciatus) by stroking them with her antennae and lower mouthparts. Once inside the nest, she mounts the host queen from the rear, seizes her around the neck with her saber-shaped mandibles, and kills her. Epimyrma stumperi, studied in Switzerland by Kutter, uses still another variation to enter the nests of its host, Leptothorax (= Myrafant) tuberum. The queen first stalks the host colony with slow, deliberate movements. When approached by the Leptothorax workers, she “freezes,” crouches down, and seems to feign death. After a time she begins to mount the workers from the rear, strokes their bodies with her foreleg combs, and grooms herself, perhaps thereby passing nest odors back and forth. With this display of sophistication in evidence, it is not surprising to find that queens of Epimyrma stumperi (like those of other Epimyrma) are able to penetrate host colonies relatively quickly. Once inside the nest, the Epimyrma stumperi queen begins an implacable round of assassinations directed at the host queens, of which there are usually between two and five in the Leptothorax tuberum colonies. She mounts each queen in turn, forces her to roll over, then seizes her by the throat with her mandibles. The sharp tips of the mandibles squeeze the soft intersegmental membrane of the neck of the victim. The Epimyrma maintains her grip for hours or even days, until the paralyzed Leptothorax queen finally dies. Then she moves on to the next queen, and this is repeated until none is left. It is a matter of more than ordinary interest that the Epimyrma stumperi workers also occasionally mount Leptothorax workers and go through an ineffectual rehearsal of the assassination behavior, but without harming their “victims” and with no visible benefit to the parasites. This seems best interpreted as a partial transfer of the queen's behavioral pattern to the worker caste where it has neither positive nor harmful effects.

The Palaearctic myrmicine genus Strongylognathus provides a second possible example of the transition from dulosis to inquilinism. The natural history of the genus has been gradually explored over a period of many years by Forel (1874), Wasmann (1891), Wheeler (1910a), Kutter (1923, 1969), Pisarski (1966), and others. Strongylognathus is closely related to Tetramorium, and its species enslave members of the latter genus. The most favored slave species is Tetramorium caespitum, one of the most abundant and widespread ant species of Europe. Strongylognathus alpinus has a life cycle more or less typical of the majority of Strongylognathus species. It is at an evolutionary level somewhat less advanced than that of Harpagoxenus and Protomognathus in the one special sense that the behavior of its workers is less degenerate. The workers, like those of most parasitic ant species, do not forage for food or care for the immature stages; nevertheless, they still feed themselves and assist in nest construction. The raids of alpinus are notoriously difficult to observe. They are thought to occur in the middle of the night and take place, for the most part, along underground galleries. The alpinus workers are accompanied by Tetramorium caespitum slaves, who, true to the aggressive nature of their species, join in every phase of the raid. Warfare against the target colony is total: the nest queen and winged reproductives are killed, and all of the brood and surviving workers are carried back and incorporated into the mixed colony. This union of adults should not be too surprising when it is recalled that Tetramorium caespitum colonies, even in the absence of Strongylognathus, frequently conduct pitched battles that sometimes terminate in colony fusion. The Strongylognathus alpinus workers are well equipped for lethal fighting. Like some other dulotic and parasitic ant species, they possess saber-shaped mandibles adapted for piercing the heads of their resisting victims (see Figure 12-15). The mode of colony multiplication is not known, but it is at least clear that the host queen is somehow eliminated in the process.

One member of the genus, Strongylognathus testaceus (Figure 12-16), has evolved at least partway to inquilinism. The Tetramorium queen is tolerated and lives side by side with the Strongylognathus testaceus queen. There are fewer testaceus than host workers, the usual situation found in advanced dulotic species. The testaceus workers do not engage in ordinary household tasks and are wholly dependent on the host workers for their upkeep. They have never been observed to do so in nature although R. Johann (cited by Buschinger, 1986) found that laboratory colonies can somehow gain slaves from Tetramorium colonies--perhaps by some form of chemical warfare. The key fact is that the Strongylognathus testaceus queen can depend upon the host queen to supply her with slave workers, without the necessity of raids. Furthermore, the parasites soon control the reproductive activity of the host queens. The Tetramorium queen generates only workers and no reproductives, whereas the Strongylognathus testaceus queen is privileged to produce both castes. Nevertheless, the presence of the Tetramorium queens permits the mixed colonies to attain great size. Wasmann found one comprising 15,000 to 20,000 Tetramorium workers and several thousand Strongylognathus workers. The brood consisted primarily of queen and male pupae of the inquiline species. It is evident that Strongylognathus testaceus is in a stage of parasitic evolution just a step beyond that occupied by Strongylognathus alpinus. The worker caste of testaceus has been retained, and it still has the murderous-looking mandibles dating from the species’ dulotic past, but it has evidently lost all of its former functions and is in the process of being reduced in numbers. Probably Strongylognathus testaceus is on the way to dropping the worker caste altogether, a final step that would take the species into the ranks of the extreme inquilines.

An interesting and still puzzling phenomenon that has been discovered is “revolt” among the slaves. On three occasions Wilson (1975a) saw Leptothorax curvispinosus slave workers approach the queen of Leptothorax duloticus, the mother slavemaker, and bite at her head and thorax. Simultaneously or immediately afterward the worker laid an egg. In two of the incidents the worker safely placed the egg in one of the egg piles; but in the third case the queen seized the egg, pulled it back and forth with the worker holding on to the other end, and finally ruptured and ate it. Slave hostility was not limited to the moment of oviposition. Once, as the duloticus queen wandered away from the egg pile, she was seized on the right hind tarsus by a young curvispinosus slave, who then alternately tried to drag her backward and to sting her. From time to time during this incident, which lasted 20 minutes, the worker stridulated. All of these actions are typical of curvispinosus workers engaged in fighting alien ants. Alloway and Del Rio Pesado (1983) witnessed comparable aggression of Leptothorax ambiguus and Leptothorax longispinosus slaves against Protomognathus americanus. On many occasions they saw the Leptothorax biting their mistresses and dragging them out of the slavemaker nests. A few Protomognathus workers lost parts of appendages as a result of these attacks. On the other hand, no slavemaker was ever seen to attack a slave. Nor was the “revolt” generalized. The same slave that attacked one slavemaker would typically feed and groom another, and any slavemaker attacked by one slave was cared for by others.

Another trait of slavemaking ants of interest to theoreticians is the degree of degeneracy of the worker caste. We have seen that at one end of the spectrum, represented by Formica sanguinea, the workers are self-sufficient. They conduct all of the quotidian tasks of the colony on their own, and they can easily survive without the support of slaves. At the other extreme, represented by the degenerate slavemakers of the genus Epimyrma, the workers have a very limited behavioral repertory and are apparently completely helpless without their slaves.

Wilson (1975a), in his study of Leptothorax duloticus, asked the question: if such an obligate slavemaker were artificially deprived of its slaves, would it expand its repertory to assume the essential tasks? When Leptothorax curvispinosus slaves are present, the duloticus do not gather food items or feed on them directly, depending instead on regurgitation from the slaves for their nourishment. They also fail to function as nurses, licking the larvae and regurgitating with them in what might well be a selfish harvesting of secretions from these immature forms. However, when the curvispinosus slaves were taken away, the repertory of the duloticus expanded dramatically. Entire behaviors appeared for the first time. The workers moved away from the nest entrance and elsewhere and gravitated toward the brood, which they now attended much more intensively than before. The workers also began to feed the larvae solid materials such as collapsed eggs or pupal skins. But they were generally inept as nurses. Larvae and pupae were allowed to sit in their partly shed skins for hours at a time, something never permitted by the curvispinosus slaves previously. A lack of competence also characterized nest building behavior. Workers carried pieces of nesting material around in their mandibles but did not succeed in placing them together to form a plug at the nest entrance. The duloticus workers began to feed on honey for the first time but took as much as ten times longer to drink the same quantity as the curvispinosus slaves. They never retrieved solid food. The result of all this ineptness was a rapid deterioration of the slave-less colony. When the original curvispinosus slaves were returned to the nest following the experiment, they quickly displaced the duloticus in the brood area, and the earlier patterns of activity and division of labor were restored.

A similar experimental approach to behavioral degeneracy was taken by Stuart and Alloway (1985) in a comparative study of Harpagoxenus and Protomognathus. Harpagoxenus canadensis proved more self-sufficient than either Harpagoxenus sublaevis or Protomognathus americanus when its slaves were taken away. It was able both to forage and to organize emigrations by use of the tandem-running method characteristic of leptothoracine ants.

The concept of the early evolution of dulosis proposed by Wilson (1975a) seems increasingly born out by recent research. With aggressive territorial behavior already part of their common repertory, few physiological and behavioral changes separate free-living ant species from their slavemaker relatives. Only two relatively slight quantitative changes in the behavior of the Leptothorax ambiguus or Leptothorax muscorum would be required to turn them into facultative slavemakers. First, the tolerance toward adult captives would have to be increased. Instead of accepting newly eclosed adults for a few hours, tolerance would have to be extended for days or even the lifetime of the captives. Second, the raiding distance would have to be increased to encompass not just adjacent nests but those as much as a meter away. Both of these modifications involve quantitative changes in the response thresholds of existing behavior patterns. The theory of population genetics allows that the changes could occur in a few tens of generations, given moderate selection pressures.

To pass from facultative to obligatory slavemaking is a more drastic step. Now the geographic range of the species would be altered to fit within those of the combined slave species, and its population densities reduced or held to lower levels merely by the necessity to “harvest” continuously from surrounding host colonies. The obligatory state of dulosis implies some degree of behavioral decay in the slavemaker. It is also reasonable to suppose that as the dulotic workers become more specialized raiding machines, with martial anatomy and behavior patterns, the decay will increase. We have seen that the dulotic species of Leptothorax, Harpagoxenus, and Protomognathus do show varying degrees of incompetence in nest building, foraging, and brood care. The idea of a progressive loss of behavior causes no theoretical difficulties. A single gene can block a behavioral pattern, and the loss or severe reduction of behavioral elements has in fact occurred in laboratory populations of Drosophila and Peromyscus mice on the order of ten generations.

Finally, dulosis always involves relatively closely related species. No solid case is known of ants enslaving other ants belonging to another tribe or subfamily. Bernstein (1978) claimed to have found the dolichoderines Conomyrma bicolor and Conomyrma insana enslaving the myrmicine Crematogaster emeryana and formicine Myrmecocystus kennedyi in the Western United States. However, her documentation is sparse, and the records are to be doubted until corroboration can be obtained.

Xenobiosis and trophic parasitism
The classic example of xenobiosis is the relation of the “shampoo ant” Formicoxenus provancheri to its host Myrmica incompleta as described by Wheeler (1903, 1910a), who described the Formicoxenus under the synonymic name of Leptothorax emersoni  (and called the host species Myrmecia brevinoda). Ants in the tribe Leptothoracini, to which Formicoxenus belongs, generally form small colonies that nest in tight little places, for example the interior of hollow twigs lying on the ground, cavities in rotted acorns, or abandoned beetle galleries in the bark of trees. The workers forage singly, and, when they encounter other ants, they usually avoid them by moving in a stealthy, unobtrusive manner. Because of these traits, colonies of leptothoracines are often found close to the nests of larger ants, and their workers are able to forage freely among their large neighbors. The trend has been extrapolated into parasitism by Formicoxenus provancheri. This species has been found living only in close association with colonies of Myrmica incompleta. Both species occur widely through the northern United States and southern Canada. Colonies of Myrmica incompleta construct their nests in the soil, in clumps of moss, and under logs or stones, especially in wet meadows and bogs. Smaller Formicoxenus provancheri colonies excavate their nests near the surface of the soil and join them to the host nests by means of short galleries. They keep their broods strictly apart. The Myrmica are too large to enter the narrow Formicoxenus galleries, but the Formicoxenus move freely through the nests of their hosts. The Formicoxenus provancheri workers do not forage for their own food. They depend almost entirely on crop liquid obtained from the host workers, using begging movements to induce the incoming Myrmica foragers to regurgitate to them. They also mount the Myrmica adults and lick them in what Wheeler has described as “a kind of feverish excitement,” to which the hosts respond with “the greatest consideration and affection.” Wheeler was under the impression that the Myrmica sought the Formicoxenus in order to obtain a “shampoo,” and he believed at first that the relationship might be mutually beneficial. Later (1910a) he conceded that the Formicoxenus are probably no more than parasites. They are, nevertheless, far from being totally dependent on Myrmica incompleta. Not only do they construct their own nests and rear their own brood, but they are also able to feed themselves, albeit awkwardly, when isolated in artificial nests in the laboratory.

Formicoxenus nitidulus is a northern and central European species with habits closely similar to Formicoxenus provancheri. This reddish little ant closely resembles Leptothorax and may have been derived from it in evolution (Figure 12-17). It is specialized for life inside the large mound nests of members of the Formica rufa group, particularly Formica lugubris, Formica polyctena, Formica pratensis and Formica rufa. It has also been found occasionally in nests of other Formica species, including Formica exsecta and Formica fusca, and even Polyergus rufescens. The relation of Formicoxenus to its hosts has been studied over a period of many years by Forel (1874, 1886), Adlerz (1884), Wasmann (1891), Janet (1897a), Wheeler (1910a), Stäger (1925), Stumper (1950), and Buschinger and his co-workers (see Buschinger, 1976a,b). The colonies, which contain 100 to 500 workers and multiple queens, appear to be functionally monogynous (Buschinger and Winter, 1976) and to nest exclusively within the host nests. They excavate their own chambers and keep their brood strictly segregated. Like the Formicoxenus provancheri shampoo ants, they build narrow galleries that open directly into the interior of the Formica nests, and from these they periodically emerge to forage among the host workers. But, unlike the provancheri, they do not lick their hosts. In fact, it has been difficult to observe interactions of any kind between the Formicoxenus nitidulus and the Formica. Although Stäger reported regurgitation from Formica workers to Formicoxenus nitidulus, and this exchange was confirmed by Buschinger (personal communication), Stumper concluded that this must be uncommon since most of the time the nitidulus workers appear to keep strictly to themselves. Formicoxenus nitidulus nevertheless displays at least two remarkable adaptations to its commensal existence. First, the males are wingless and highly worker-like in appearance (Figures 12-17 and 12-18). They can be distinguished externally only by their longer antennae, which contain one more segment than those of the worker, by an additional abdominal segment, and of course by the extrusible portions of their genitalia. The matings take place on top of the host nests. The second adaptation to life with Formica is the ability to emigrate in the columns of the host workers when the latter change nest sites. Forel (1928) and later Elgert and Rosengren (1977) demonstrated that the Formicoxenus follow the scent trails of their Formica hosts.

In 1925 Wheeler reported the discovery of a new and thoroughly surprising case of xenobiosis between a myrmicine guest-ant, Megalomyrmex symmetochus, and a fungus-growing host species Sericomyrmex amabilis, also a myrmicine. The Sericomyrmex found modest-sized colonies, comprising 100 to 300 workers and a queen, that nest in the wet soil of the laboratory clearing on Barro Colorado Island, Panama. They subsist entirely on a special fungus raised on beds of dead vegetable material. The Megalomyrmex form smaller colonies, consisting of 75 adults or less, that live directly among the fungus gardens of the host. Since the Sericomyrmex also place their brood in the gardens, the young of both species become mixed to a limited extent. However, the Megalomyrmex tend to segregate their brood in little clumps, each of which is closely attended by a few workers, and neither species feeds or licks the brood of the other. The most remarkable fact is that the Megalomyrmex appear to subsist exclusively on the fungus. This represents a major dietary shift which must have occurred relatively recently in the evolution of the genus. Because liquid food exchange is either uncommon or completely lacking in fungus-growing ants, the Megalomyrmex do not secure nutriment from the Sericomyrmex in this way. They do, however, lick the body surfaces of their hosts.

An important common feature of the three examples of xenobiosis just cited is what German writers call Futterparasitismus, which can perhaps best be translated into the rather formal expression “trophic parasitism.” This is intrusion into the social system of another species in order to steal food. Trophic parasitism does not by itself require a close association of nests or even entry into the host nest by foraging workers. In other words, it can occur apart from xenobiosis. A weak, nonxenobiotic form of such parasitism is exhibited by Camponotus lateralis toward Crematogaster scutellaris in Europe. Goetsch (1953) and Kaudewitz (1955) have described instances in which Camponotus workers followed the Crematogaster odor trails to their feeding grounds and exploited the same food resources during the same time of day. The Crematogaster were hostile to the Camponotus, which assumed a crouching, conciliatory posture when they met the legitimate users of the trails. Unlike the xenobionts, the Camponotus lateralis nest separately. Moreover, the relationship is not obligatory on lateralis, since colonies and foraging workers of that species are often found far from Crematogaster nests.

On the island of Trinidad, in the West Indies, Wilson (1965b) discovered an instance of trail sharing that approaches a neutral, or commensalistic relationship. Each of the several colonies of the formicine Camponotus beebei encountered were in close association with a large colony of the dolichoderine Azteca chartifex, one of the dominant ant species of the island forests. The Camponotus nested in cavities in tree branches near the arboreal carton nests of the Azteca, and their workers followed the Azteca odor trails down the branches and tree trunks to foraging areas on the nearby ground and weedy vegetation. The diets of the two species were not determined, but, regardless of the degree of similarity, potential interference between the two species was reduced by the existence of opposite daily schedules. The Camponotus therefore “borrowed” the Azteca trails when the owners were putting them to minimal use. The Azteca workers were hostile to the Camponotus workers and attacked them on the rare occasions when the latter slowed in their running, but the Camponotus were larger and faster and usually easily avoided their hosts without causing any visible disturbance. On a single occasion, a Camponotus worker was seen to lead out a tight column of six other Camponotus workers, guiding them along by means of a short-lived odor trail. The pheromone was laid directly on top of the Azteca odor trail, yet the Azteca workers did not fall in line or show any other response to the passage of the Camponotus group. Thus it appears that the Camponotus, while “eavesdropping” on the Azteca odor trails, have reserved a special recruitment trail system of their own which they do not share with their hosts.

Two phenomena have been discovered in very different parts of the world that constitute alternative strategies of trophic parasitism. In the Siberian steppes, Formica pratensis is a territorial dominant over Formica cunicularia, driving its workers away from favored nest sites and food finds. Occasionally, Formica cunicularia foragers are able to steal food items that have been temporarily laid on the ground by pratensis foragers, but in general they must depend on quickness and luck to gather food before the pratensis arrive on the scene. For their part, the pratensis workers use the cunicalaria as scouts. They are attracted to the movements of these particular ants (and not to other species), so that when the cunicalaria discover food the pratensis are often able to appropriate it for themselves. Reznikova (1982), who discovered this relationship, found that when pratensis colonies were prevented from foraging under otherwise natural conditions, nearby cunicalaria colonies increased their food intake. But when cunicalaria colonies were constrained the pratensis, now deprived of their scouts, harvested significantly less food.

In Panama, hundreds of workers of the little myrmicine Crematogaster limata were observed to file into the nests of the large ponerine Ectatomma tuberculatum. They climbed up the legs of the Ectatomma onto their bodies, which they proceeded to lick. The big ponerines did not respond aggressively to these little intruders; they even occasionally opened their mandibles and let the Crematogaster lick their extended mouthparts. Then the Crematogaster climbed down and quickly left (D. E. Wheeler, 1986b). This odd relationship resembles the xenobiosis of Formicoxenus, but its significance can only be guessed on the basis of present evidence.

An interesting evolutionary question can now be raised: do xenobiosis or the more tenuous forms of trail parasitism ever lead to full inquilinism? The evidence to look for is the coexistence in the same genus of xenobiotic species and fully inquiline species, both of which parasitize other species belonging to the same genus. This is the criterion, it will be recalled, by which inquilinism in Epimyrma corsica was inferred to be of dulotic origin. So far, no such examples have been found. Even so, some of the traits of Kyidris, an inquilinous genus which will be described in a moment, at least suggest the possibility of a xenobiotic origin. The same is true of the species of Formicoxenus.

1. The worker caste is lost.

2. The queen is either replaced by an ergatogyne, or ergatogynes appear together with a continuous series of intergrades connecting them morphologically to the queens.

3. There is a tendency for multiple egg-laying queens to coexist in the same host nest.

4. The queen and male are reduced in size, often dramatically so; in some cases (for example, Teleutomyrmex schneideri, Plagiolepis ampeloni, Plagiolepis xene) the queen is actually smaller than the host worker.

5. The male becomes “pupoid”:  its body is thickened, the petiole and postpetiole become much more broadly attached, the genitalia are more externally exposed when not in use, the cuticle becomes thin and depigmented, and the wings are reduced or lost. The extreme examples of this trend are displayed by Anergates atratulus, Pheidole neokohli, and Pheidole acutidens (see Figures 12-19 and 12-20).

6. There is a tendency for the nuptial flights to be curtailed, and to be replaced by mating activity among nestmates (“adelphogamy”) within or near the host nest. Dispersal of the queen afterward is very limited.

7. Probably as a consequence of the curtailment of the nuptial flight just cited, the populations of inquiline species are usually very fragmented and limited in their geographic distribution.

8. The wing venation is reduced.

9. Mouthparts are reduced, with the mandibles becoming smaller and toothless and the palps losing segments. Concomitantly, the inquilines lose the ability to feed themselves and must be sustained by liquid food regurgitated to them by the host workers.

10. Antennal segments are fused and reduced in number.

11. The occiput, or rear portion of the head, of the queen is narrowed.

12. The central nervous system is reduced in size and complexity, usually through reduction of associative centers.

13. The petiole and postpetiole are thickened, especially the latter, and the postpetiole acquires a broader attachment to the gaster.

14. A spine is formed on the lower surface of the postpetiole (the Parasitendorn of Kutter).

15. The propodeal spines (if present in the ancestral species) “melt,” that is, they thicken and often grow shorter, and their tips are blunted.

16. The cuticular sculpturing is reduced or lost altogether over most of the body; in extreme cases the body surface becomes strongly shining.

17. The exoskeleton becomes thinner and less pigmented.

18. Many of the exocrine glands are reduced or lost, a trait already described in some detail in the earlier account of Teleutomyrmex schneideri.

19. The queens become highly attractive to the host workers, which lick them frequently. This is especially true of the older, physogastric individuals, and it appears to be due to the secretion of special attractant substances which are as yet chemically unidentified.

Parabiosis
Forel (1898) designated as “parabiosis” the following complex behavior which he discovered in Colombia. Colonies of the arboreal, rain forest ant species Crematogaster limata parabiotica and Monacis debilis (called Dolichoderus debilis var. parabiotica by Forel) commonly nest in close association, with the nest chambers kept separate but connected by passable openings. They are the principal occupants of ant gardens in the rain forests of South America and as such rank among the most abundant arboreal ants (see Chapter 14). The workers of the two species also run together along common odor trails. Wheeler (1921a) confirmed the phenomenon in his Guyana studies and showed that the two species collect honeydew together from membracids. Wheeler also discovered a similar association between Crematogaster limata parabiotica and Camponotus femoratus. Both species were observed utilizing common trails and gathering honeydew from jassids and membracids on the same plants as well as nectar from the same extrafloral nectaries of Inga. Wheeler believed that the Crematogaster and Camponotus workers were tolerant of each other in this potentially confrontational situation. He saw them “greet” each other with calm antennation on the trails. On three occasions he observed Camponotus workers regurgitating to individuals of Crematogaster.

In a subsequent and more detailed study of the same three species in Brazil, Swain (1980) obtained a very different picture. He learned first of all that parabiotic Crematogasters actually belong to two species, one found with the Monacis and one with the Camponotus. The parabionts do not mix their odor trails. Rather, both the Monacis and Camponotus follow trails laid by their respective Crematogaster associates. Moreover, the parabionts are not always amicable at food sites. At sugar and insect baits set out by Swain, the Camponotus femoratus workers drove the Crematogaster limata away and fed exclusively by themselves. So during foraging at least, the relationship is not mutualistic as formerly believed, but parasitic.

It has also been commonly thought (Wheeler, 1921a; Weber, 1943) that the larger ants of each pair provide protection for the Crematogaster at the nest site. It is at least true that Camponotus femoratus plays this role against vertebrates. Its workers are among the most formidable ants in the forest canopy, swarming out to bite and spray formic acid on any intruder and at the slightest disturbance. Monacis debilis, on the other hand, are timid ants that appear to offer even less defense than the Crematogaster. But of course they might be very effective against ants and other invertebrate enemies. It is entirely possible that the losses the Crematogaster suffer from domination by Camponotus femoratus at food sites (their relation to Monacis is not known) is counterbalanced by the added protection they receive at the nest. The Crematogaster seem well served by the parabiotic association, because they usually if not always exist within it.

Other forms of parabiosis probably exist. One marginal case has been found in South Australia by Greenslade and Halliday (1983). Three species of Camponotus belonging to the ephippium and innexus groups respectively outwardly resemble species in the dominant, more abundant species of meat ants in the Iridomyrmex detectus group. These apparent mimics nest within the meat ant territories and forage among the Iridomyrmex workers, using speed and agility to avoid their formidable neighbors. The majors of at least one species of the Camponotus ephippium group have huge heads with which they block the nest entrances (Figure 8-37). An even more intimate parabiotic relationship exists between the Australian “sugar ants” Camponotus consobrinus and Camponotus perthiana on the one side and the meat ant Iridomyrmex purpureus on the other. The Camponotus often nest directly inside the large mounds of the Iridomyrmex, yet without connecting their nest galleries and chambers to those of their hosts (Greaves and Hughes, 1974; Hölldobler, unpublished observations).

The degrees of inquilinism
Once an ant species enters complete inquilinism, whether by temporary parasitism, dulosis, or xenobiosis, it seems to evolve quickly on down into a state of abject dependence on its host. It acquires some of what can be termed the “inquiline syndrome,” a set of characteristics found in varying combinations in all of the relatively specialized inquiline species (Wilson, 1971):

Let us examine more closely one of the most interesting of these trends, namely, reduction of the worker caste. The inquiline species Kyidris yaleogyna of New Guinea represents the most primitive known level in this evolutionary regression since it retains an abundant and partly functional worker caste (Wilson and Brown, 1956). Colonies of Kyidris yaleogyna are parasitic on Strumigenys loriae, one of the more abundant and ecologically widespread of the Papuan ants. Both Kyidris and Strumigenys are members of the tribe Dacetini of the subfamily Myrmicinae, but they are otherwise very different from one another. Kyidris is a short-mandibulate form, closer to Smithistruma and Serrastruma than to the highly distinctive, long-mandibulate Strumigenys. Four mixed colonies were discovered nesting in pieces of decaying wood on the floor of rain forests. In each, the Strumigenys slightly outnumbered the parasites. One large colony collected in toto contained 1,622 workers and 16 dealated queens of Strumigenys, in combination with 1,170 workers, 4 dealated queens, 84 alate queens, and 51 males of Kyidris. A second colony contained 243 workers and 4 dealated queens of Strumigenys and 64 workers, 2 dealated queens, and 31 males of Kyidris. These large groups lived in completely harmonious mixtures in which the parasitic nature of the Kyidris was only subtly evident. The Kyidris workers foraged for food. One group was found attending coccids near the nest. Others engaged in hunting for small insects, but, compared with the Strumigenys, they were quite ineffectual. They wandered through the food chambers of the artificial nests like typical restless dacetines, but rarely tried to catch prey. Even when they tried, they usually failed, in sharp contrast to the highly efficient performances of their Strumigenys nestmates. One Kyidris worker was seen to seize a symphylan, pull it backwards, hold it for about thirty seconds without trying to sting it, and finally release it when it began to struggle. Another seized an entomobryid collembolan, pulled it back vigorously, then lost it when the entomobryan kicked with its furcula. Still another was seen actually to carry an entomobryid at a brisk clip across the food chamber floor; it reached the entrance to the brood chamber only to have a Strumigenys meet it and take the insect away. The general impression created is that the predatory behavior has regressed, but not completely disappeared, in Kyidris workers. In the artificial nests, at least, the Strumigenys workers did most of the productive hunting. The Kyidris studied by Wilson and Brown (1956) also aided in brood care much less frequently than the Strumigenys, and their efforts seemed ineffectual. Kyidris workers were never observed in nest construction. They received regurgitated liquid food from the Strumigenys workers, and sometimes obtained food by inserting their mouthparts between those of two Strumigenys exchanging regurgitated material, but they were never seen to offer anything in return. In sum, the New Guinea Kyidris appear to represent inquilinism at a very early stage when the worker caste has only begun to reduce its behavioral repertory. Probably the degeneration has proceeded past the point of no return since it is doubtful if Kyidris colonies could survive without their hosts.

A somewhat more advanced stage of behavioral decay is shown by the workers of Strongylognathus alpinus, a European species already mentioned in the previous section on dulosis. The workers are able to conduct raids, they participate in nest building, and they can feed themselves. But, unlike the Kyidris workers, they have lost the capacity either to hunt or to care for brood. In the great majority of all other dulotic and inquilinous ant species that still possess a worker caste, the workers appear to have entirely lost the ability to carry on the ordinary functions of nest construction, food gathering, and queen and brood care.

It is little wonder, then, that most truly inquilinous species have taken still one more step in evolution and discarded the worker caste altogether. The stages leading to this final abrogation have been beautifully documented in the genus Plagiolepis by Le Masne (1956b) and Passera (1966, 1968b). The two species Plagiolepis grassei and Plagiolepis xene are parasitic on the closely related free-living form Plagiolepis pygmaea. In certain key characteristics, namely loss of worker caste size reduction, and alteration of the male form, xene qualifies as an extreme inquiline, while grassei occupies an almost exactly intermediate position between it and the free-living Plagiolepis (see Table 12-2). The most interesting annectant feature of grassei is in the status of the workers. This caste is almost extinct, and it appears in a given host nest only after the winged parasitic sexuals are produced--the reverse of the order that is universal in free-living ant species.

In a separate analysis Wilson (1984c) examined traits of the inquiline syndrome in the nine known parasitic species of Pheidole. He assumed that when most or all the species possessed a given trait, the change producing the trait had appeared in evolution prior to other, less widespread traits. Conversely, when a trait was rare, such as the existence of pupoid males, it was viewed as a late event in evolution. If this assumption is correct, then the earliest changes to occur in the Pheidole parasites were loss of the worker caste, reduction of size, rounding of the occiput (associated with reduction of the mandibular adductor muscles and loss of mandibular strength), loss of body sculpture (associated with a thinning of the exoskeleton), and broadening of the postpetiole. These shifts were followed by reduction of the antennal segments, including the 3-segmented club, and of the mandibles; and the development of a pupoid body form in the male.

The majority of inquiline species for which adequate information is available permit the host queens to live. This is the situation one would intuitively expect. It seems to make good sense for the parasites to insure themselves a long-lasting supply of host workers. This reasoning is also consistent with the fact that species of Strongylognathus which obtain host workers by slave raids also destroy the host queens, while Strongylognathus testaceus, the one species of the genus that does not obtain host workers in this fashion, tolerates the host queens. But what of the minority of inquilines whose presence causes the death of the host queens, in particular the queen-killing Epimyrma species that do not conduct slave raids? For some species, in which either the host colonies or the parasites themselves are relatively short-lived, it may be advantageous to get rid of the host queen and invest all of the efforts of the host workers into producing as many parasite queens and males as soon as possible. Buschinger and Winter (1983b) have provided strong evidence in Epimyrma supporting this hypothesis. Colonies of the slave-raiding Epimyrma ravouxi have a maximum life span of ten years. In contrast, the queens of Epimyrma kraussei, whose daughter workers conduct slave raids only rarely, produce mainly sexual forms and a scattering of workers until they run out of Leptothorax host workers to support them. This is the “big bang” strategy of reproduction, essentially the same as that employed by such fishes as the migratory eels and salmon and such plants as the bamboos (Gadgil and Bossert, 1970; Oster and Wilson, 1978). Other parasitic ant species, with longer lives and host colonies that are more stable, would find it advantageous to let the host queens live and to employ the host colony in the production of parasite queens and males at a lower rate--but for a longer period of time. We should bear in mind that even the continuous reproducers inhibit host reproduction to some degree. In general, the production of host sexual forms in the presence of inquilines is a rare event even in those cases where the host queen is permitted to live. Furthermore, Passera (1966) has discovered that, in Plagiolepis pygmaea, even the ability to produce workers is partially inhibited by the presence of parasitic Plagiolepis xene queens. The theoretical implications raised by these various observations were formalized by Wilson (1971) in the following conjecture: The degree of reproductive repression inflicted by a given inquiline species is such as to maximize the total production of parasite queens and males per host colony under the particular ecological conditions in which the mixed colonies occur. In order to test this hypothesis and more generally to advance a population theory of social parasitism, data on the population dynamics of both parasitized and unparasitized host colonies are needed.