The Ants Chapter 5

AntWiki: The Ants --- Online
Jump to navigation Jump to search
The Ants

CHAPTER 5. COLONY ODOR AND KIN RECOGNITION

Introduction

When a worker inspects a nestmate she seems to do nothing more than casually sweep her antennae over the other's body. The true intensity of the inspection is revealed, however, when an alien ant enters the nest. If the intruder belongs to a different species, it is almost always violently attacked. On the other hand if the ant is a member of the same species but from a different colony, the hostility falls somewhere along a broad gradient of responses. At one extreme, the intruder is accepted but offered less food until it has time to acquire the body odor. At the other extreme, the residents attack it with extreme violence, locking their mandibles on its body and appendages while stinging it or spraying it with formic acid, citronellal, or some other toxic substance. Intermediate degrees of rejection include avoidance, mutual threatening with open mandibles, and nipping and leg pulling. This array of alternative response from aversion to violence has been used as the basis of sensitive bioassays in studies of colony odor by a number of authors from Lange (1960, 1967) to Carlin and Hölldobler (1983, 1986, 1987), as summarized in Table 5-1 and Figure 5-1.

Kin recognition: general principles

The ability to distinguish nestmates from strangers is vital to social life. Waldman (1987) and Waldman et al. (1988) recommended that the term kin discrimination be used to denote differential treatment of conspecifics, correlated with kinship, with recognition being defined more narrowly as "the processes by which individuals assess the genetic relatedness of conspecifics to themselves or others, based upon their perception of traits expressed by or associated with these individuals." Waldman et al. (1988) state "we emphasize the distinction between kin recognition and kin discrimination because recognition--a series of internal and essentially unobservable physiological events--may occur without any behavioral response."

Throughout the evolutionary history of the ants and other social insects there has been an intense selection pressure to sharpen such recognition ability, because favors bestowed on an unrelated individual are wasted in the remorseless crucible of natural selection. Consequently, behavioral discrimination of relatives from non-relatives must be intimately tied to kin selection. If kin selection works as effectively as the evidence (reviewed in Chapter 4) implies, in other words if it constitutes a strong "ultimate" factor in evolution, then mechanisms of kin recognition would be advantageous as a means of directing nepotism correctly. In recent studies across a broad diversity of animals, kin recognition of one kind or another has been implicated in most kinds of social behavior, from simple aggregations of tadpoles to the most complex colonial organizations of ants and termites (reviews by Hölldobler and Michener, 1980; Gadagkar, 1985; Fletcher and Michener, 1986). Where a discriminating capacity was expected from kin selection theory it has almost always been found to exist in fact. Moreover, the study of the linkage has proved heuristic for biology. The mediating behavior is often complex and highly effective, and new physiological processes are more easily discovered when research is animated by kin selection theory.

When considering recognition labels, keep in mind that two forms of kin-discrimination can be distinguished in insect societies: exclusion of non-kin from the colony, and preferential aid for kin of higher relatedness within the colony. As Hölldobler and Carlin (1987) pointed out, labels involved in recognition at the colony level are simultaneously specific and anonymous. That is, workers are able to discriminate between nestmates and intruders, but they also tend to treat all nestmates as fellow colony members, regardless of the magnitude of their true relatedness. This view of anonymity among genetically varying nestmates (Jaisson, 1985, called it the "fellowship concept") does not preclude specificity at the within-colony level. Generally, though, it appears that workers encountering one another in the context of territorial defense or nest guarding respond to cues that indicate colony membership, rather than directly indicating kinship.

Useful classifications of the labeling phenomena have been prepared by Holmes and Sherman (1983), and Sherman and Holmes (1985) for animals generally and Hölldobler and Michener (1980) for social insects in particular. Four principal strategies exist, each employing a different kind of cue or stimulus by which one individual classifies other members of the same species.

The first strategy is based on purely spatial distribution. It occurs in species with a high degree of site fidelity. Adult bank swallows (Riparia riparia), to cite one well-studied example, learn the location of the nest holes they excavate, and they feed any chicks found in these retreats (including alien swallow chicks introduced by the investigator) up until the time their own offspring fledge at about two weeks of age. Conversely, they ignore their own nestlings when the experimenter transfers them to nearby burrows. No example of reliance on purely spatial distribution has yet been documented in the ants or any other social insects, although Klahn (1979) demonstrated the use of spatial as well as phenotypic cues in Polistes wasps.

In purely allelic recognition, as conceived in theory at least, animals would depend on the innate capacity to discriminate other individuals with which it shares certain alleles. The alleles could be expressed in the phenotype in one or the other of several ways, for example by odor or physical appearance. The important qualifier in this extreme category is that the phenotypic difference is innately recognized, not learned. Dawkins (1976) has called the hypothetical phenomenon the "green beard effect": if the focal animal were to have a green beard it would classify all green-bearded strangers as kin. The possibility is intriguing and might play a more significant role in within-colony discrimination, but most investigators agree that allelic discrimination is unlikely to evolve as the primary system in discrimination at the colony level (Crozier, 1987a). It requires the possession of individually distinctive sets of genes that prescribe both the recognition cue and the neurosensory apparatus to recognize it. The principal difficulty in proving such an ultrasimple arrangement is that learning has been implicated in most cases of colony discrimination, even though it is often of a very restricted and predictable nature. No one has invented a way to control for all conceivable learning possibilities and hence to eliminate them, including the familiarization of an animal with its own cues. Even so, we may confidently look for cases in the simplest of organisms, where interactions take the form of growth and tissue rejection rather than conventional learning in neuronal systems. Examples include corals, sponges, and other colonial invertebrates that fuse bodily with other, genetically identical organisms but reject those that are even slightly different.

The third strategy of kin recognition is discrimination "of specific individuals with which one has previously interacted." This form can be called recognition by association. As Waldman et al. (1988) describe it: "Individuals presumably learn one another's traits in a setting where only relatives are likely to be present, later distinguishing them from non-kin in other potentially more ambiguous settings. . . . If the labels expressed by every member of a circumscribed kin group (e.g., within a nest, burrow, or insect colony) are distinct, individual recognition may result."

Waldman (1987) points out that this kind of kin recognition mechanism is nevertheless very similar and perhaps even identical to the so-called phenotype matching, the fourth strategy of kin recognition proposed by Sherman and Holmes (1985). In this case kinship identity may be inferred from the perceived overlap in cues between conspecifics, whereby animals are able to discriminate between unfamiliar kin and unfamiliar non-kin, or among familiar kin with different degrees of relatedness.

The evidence to date from social insects exclusively implicates phenotype matching. Furthermore, only chemical signals are known to be employed. This being the case, we have to consider that nestmate recognition consists of both perceptive and expressive components, and the ontogeny of both must be included in any complete explanation of the recognition system (Gamboa et al., 1986; Hölldobler and Carlin, 1987). For variation in labels to be functional, an individual must have some criteria for determining whether to respond to a given variant. These decision-making rules can be innate or learned or, to use less problematic terminology, they can be determined by closed or open ontogenetic programs (Mayr, 1974). Hölldobler and Carlin (1987) noted that the rules of perception for both anonymous and specific communication signals can be either genetically encoded or acquired by experience, depending on the predictability of signal expression. When the expression of a semiochemical is highly predictable, the genome of the receiver can "know" in advance what characteristics to expect, and can program an efficient, hard-wired neural mechanism for recognizing them. This is clearly true for certain chemical communication signals, such as anonymous sex pheromones detected by specialist receptors. Conversely, when the expression of a semiochemical is unpredictable the receiver's genome cannot dictate a perception mechanism in advance, and the criteria for responding must be derived from experience. In all of the species of social insects studied thus far, nestmate recognition cues appear to be learned shortly after eclosion into the adult stage. A new worker eclosing into a colony whose queen mated more than once cannot know which heritable recognition signals to expect among its half siblings. In addition, it must learn cues that were acquired from other colony members and the external environment. Masson and Arnold (1984) and Gascuel et al. (1987) suggest that young adult honeybees learn odors in an imprinting-like manner, due to restrictive timing of olfactory center development in the brain, which would admirably fit Mayr's (1974) definition of an open ontogenetic program. This being the case, the following general scheme of phenotype matching in social insects can be reasonably suggested. Each individual possesses both phenotypic recognition cues--its recognition label--and a sensory template specifying a learned set of cues likely to be borne by kin. An observing individual determines whether or not an unfamiliar conspecific is a relative by matching the latter's label to its template. Carlin and Hölldobler (1986) constructed a flow-diagram model of possible extrinsic and intrinsic inputs to the label and template of a social insect worker, which is summarized in Figure 5-2. The thick horizontal line represents the body surface and sensory receptors of the observing individual. Intrinsic inputs (below the line) are genetic in origin, while extrinsic inputs to the label (above the line) presumably occur on the body surface. Inputs to the template are learned.

Genetically determined discriminators are known, or at least strongly suspected, in a wide range of species of bees (Greenberg, 1979; Breed, 1981, 1983) and ants (Jutsum et al., 1979; Haskins and Haskins, 1983; Mintzer and Vinson, 1985b; Stuart, 1987a). Direct genetic specification of both labels and templates has not been documented in any species. Such a mechanism risks false-negative discrimination against kin when colonies contain even a moderate genetic diversity, a circumstance that arises when queens mate with multiple males. The need to accept nestmates that are unpredictably varied, thus preserving colony cohesiveness, favors the evolution of mechanisms based on learning. Callows whose discriminators differ from the ambient cues of the colony in which they emerge may be protected by special pheromones, such as brood masking substances, while they acquire labels and learn templates (Jaisson, 1972a,b; Hölldobler, 1977). Environmental input to recognition labels is known in a number of social Hymenoptera (Kalmus and Ribbands, 1952; Jutsum et al., 1979; Mabelis, 1979; Boch and Morse, 1981; Haskins and Haskins, 1983; Gamboa et al., 1986; Obin, 1986; Breed and Bennett, 1987; Stuart, 1987b,c; Obin and Vander Meer, 1988a,b). However, purely extrinsic recognition risks false-positive acceptance between neighboring colonies that share environmental cues, and it is precisely these neighbors that nestmates will encounter and need to oppose. In most cases, therefore, recognition is not likely to rely solely on differences in soil and nest-material odors, although a prominent environmental component has been demonstrated in the recognition system of the fire ant Solenopsis invicta (Obin, 1986; Obin and Vander Meer, 1988b) and Polistes wasps (Gamboa et al., 1986).

Queen discriminators have been demonstrated in honeybees (Breed, 1981; Boch and Morse, 1982) and inferred in some ant species (Watkins and Cole, 1966; Jouvenaz et al., 1974). Their transferability has been demonstrated in Camponotus species (Carlin and Hölldobler, 1986, 1987, 1988; Carlin and Vander Meer, unpublished data). Effects of queens on worker nestmate discrimination has been documented in several Camponotus species (Carlin and Hölldobler, 1983, 1986, 1987), as well as Leptothorax lichtensteini (Provost, 1985, 1987) and interspecific mixed colonies of Myrmica (Brian, 1986a). In other Leptothorax species (Stuart, 1987a-c), Pseudomyrmex ferrugineus (Mintzer, 1982b) and Solenopsis invicta (Obin and Vander Meer, 1988a,b) queens have been shown to contribute little or nothing to nestmate discrimination.

A worker gestalt mechanism, according to the criteria in the flow-diagram model (Figure 5-2), has been clearly demonstrated in some Leptothorax species, which have small and frequently polygynous colonies (Stuart, 1987a-c). It is important to note, however, that the queen input in label and template formation in Camponotus may also be considered part of the "gestalt model" as originally formulated by Crozier and Dix (1979). In fact, the queen can dominate a collective gestalt. Carlin and Hölldobler (1986, 1987) demonstrated that in Camponotus pennsylvanicus and Camponotus floridanus queens, worker genotype and environmental cues all contribute to nestmate discrimination labels of adult workers, but not all are of equal importance. Breed (1987) reports that honeybee workers acquire extrinsic recognition cues in the presence of queens and environmental odors, although there is some uncertainty over the possible formation of collective gestalt labels among queenless workers (Breed et al., 1985; Getz and Smith, 1986; Getz et al., 1986).

Discriminatory ability

The existence of a finely tuned recognition system based on degrees of genetic similarity has been demonstrated in the social bees, which are relatively conducive to breeding experiments. Greenberg (1979) utilized Lasioglossum zephyrum, a primitively eusocial sweat bee that constructs simple burrows in the soil. One of the colony members serves as a guard by sitting at the entrance of the outer burrow and blocking alien Lasioglossum zephyrum and other intruders. Greenberg bred bees in the laboratory to produce colonies having 12 different levels of genealogical relationship to one another, from completely unrelated groups to colonies whose members were sisters in inbred brother-sister lines. Bees were tested in pairs by taking an individual from one nest and introducing it to a guard bee of a different nest, under regimes in which the two individuals could never have previously seen or smelled each other. During 1,586 such introductions, there was a strong positive correlation between the degree of relatedness between the two bees and the frequency with which guards permitted introduced bees to pass (Figure 5-3). The experiments were reinforced during subsequent trials by Buckle and Greenberg (1981), who provided several lines of new evidence indicating that odor cues are not transferred among nestmates. However, their additional conclusion that the guard bee knows only the odors of its nestmates, and not its own, is questionable (Getz, 1982).

Similar experiments by Breed (1981, 1983) revealed the existence of a genetic component in the colony odor of honeybees. He found that both queens and workers of Apis mellifera were accepted into strange nests with a probability that rose with the degree of relatedness, where three degrees of relatedness were employed (inbred sisters, outbred sisters, and unrelated). Getz and Smith (1983) performed similar experiments but used genetic markers to measure genetic relatedness. By this means they discovered that worker bees can distinguish full and half sisters. Further evidence has been adduced by Getz and Smith (1986), as well as by Breed et al. (1985), that in some contexts honeybee workers perceive their own labels and use this "self awareness" as templates for recognition of nestmates and full sisters within the colony. Page and Breed (1987) infer that this "self awareness" is probably learned, a process sometimes called "self-matching."

The common occurrence of multiple insemination and polygyny requires a mechanism for anonymously identifying all nestmates as colony members as opposed to intruders. In the context of interactions among colony members, especially the rearing of reproductive brood, intracolony genetic heterogeneity also reduces the level of inclusive fitness considered important for the maintenance of hymenopteran eusociality. A worker who indiscriminately produces half sisters (patrilines) or the offspring of other queens (matrilines) fails to yield the same proxy reproductive success that haplodiploid workers obtain by rearing three-quarters-related full sisters. Kin discrimination has been invoked as a solution to this difficulty, maintaining eusociality by kin selection despite low relatedness within colonies (Gadagkar, 1985). Alternatively, even if eusociality is maintained by some other means, workers that find themselves among nestmates of varied relatedness can improve their inclusive fitness by discriminating on the basis of kinship. In either case, sufficient cue specificity must be retained within colonies to permit discrimination of full sisters from other patrilines and matrilines, if cohorts defined by three-quarters relationship are to cooperate preferentially (Hölldobler and Carlin, 1987).

The studies with bees just cited indicate that within-colony discrimination among workers is possible. In addition Getz et al. (1982) reported segregation of worker patrilines during swarming, Evers and Seeley (1986) observed aggressive discrimination between patrilines in queenless colonies containing ovipositing workers, and Frumhoff and Schneider (1987) found that workers prefer to exchange food with full sisters rather than with half sisters. They also prefer to groom them. All of these investigators used genetic color markers to identify the patrilineal cohorts, which were produced by artificial insemination. However, recent results indicate that the color markers exaggerate the specificity of half-sister discriminators (Frumhoff, 1987). Although discrimination among patrilines has not been tested in ants, Carlin et al. (1987b) found that Camponotus floridanus workers originating from unrelated colonies and introduced into mixed nests, antennate familiar non-kin more frequently than familiar sisters but fail to discriminate consistently in food exchange and grooming (Figure 5-4). The higher frequency of antennation toward non-kin is apparently an inquiring rather than a soliciting behavior. However, the absence of such discrimination in experimental colonies does not preclude the possibility that the Camponotus favor siblings during the rearing of reproductive brood. Indeed, kin-based rearing of queens by honeybees has been tested experimentally by several investigators. Breed et al. (1984) found no evidence that workers prefer nestmate queen larvae over unrelated, non-nestmate larvae. On the other hand, Page and Erickson (1984) observed a preference for nestmate larvae that was full sisters (three-quarters-related) over approximately one-quarter-related larvae from another colony. However, the latter may have been additionally distinguishable due to their origin in a different nest. In still other experiments, Visscher (1986) transferred brood pieces among colonies in controlled experiments and obtained the same result. After separating out nest-specific cues by presenting workers with non-nestmate unrelated and non-nestmate related eggs, he obtained preferences for siblings (mixed full and half sisters) over non-kin. Noonan (1986) performed the only reported test of worker interactions with queen larvae of different patrilines, all originating in the same nest. She found that workers significantly preferred to visit, inspect, and feed full-sister larvae. Unfortunately, these results might also have been influenced by genetic color markers used to identify the patrilines. To summarize, the evidence accumulated to date indicates that the signal variation correlated with kinship required for within-colony kin recognition does exist alongside the cues used for between-colony discrimination. Whether this specificity is utilized in adaptive nepotistic behavior under natural conditions remains to be conclusively demonstrated.

Individual odors have been implicated in a few other cases of ant behavior. Nest-founding queens of the honeypot ant Myrmecocystus mimicus and Australian meat ant Iridomyrmex purpureus, as well as workers of Leptothorax and Harpagoxenus, set up aggressive relationships very similar in outward appearance to vertebrate dominance hierarchies (Cole, 1981; Bartz and Hölldobler, 1982; Franks and Scovell, 1983; Hölldobler and Carlin, 1985). Queens of Leptothorax curvispinosus do not dominate one another by overt aggression, but they do consume some of the eggs of other queens. At least part of the basis of this differential oophagy is rough treatment. One queen was observed to handle the eggs of all of her rivals in an aggressive manner, licking them vigorously and chewing on them until the shells crumpled slightly before springing back into shape. Eggs from one of the rivals were more fragile and as a consequence were ruptured and eaten (Wilson, 1974b).

In a remarkable recent development, Jessen and Maschwitz (1986) have discovered that workers of the tropical Asian ponerine Pachycondyla tesserinoda recognize and follow their own personal odor trails from a still unknown glandular source while returning from food discoveries or leading nestmates to new nest sites. They guide other workers along these trails by means of tandem running to the target locations. The follower ants fix on the general body odor of the leader and run close behind her. By using such personal trails, the leaders are able to return to food or nest sites they themselves have personally selected. Individual specific trails have also been discovered in Leptothorax affinis (Maschwitz et al., 1986b).

Brood discrimination

Other, less direct pieces of evidence suggesting widespread discriminatory ability exist in the ants (Carlin, 1988). In the relatively primitive myrmicine genus Myrmica, workers seem not to be capable of distinguishing the tiny first instar larvae from eggs, so that when eggs hatch the larvae are left for a time in the midst of the egg pile. The larvae feed during the first instar by breaking into a single adjacent egg. As soon as they molt and enter the second instar, however, the larvae are removed by the workers and placed in a separate pile (Weir, 1959a). Third-instar larvae vary greatly in size; the smaller ones are destined to metamorphose into workers, while the larger ones retain the ability to develop into queens. If a nest queen is present, the large larvae receive proportionately less food, and they are also licked and bitten more by the workers, an action that may reduce their growth still further. The ultimate effect of the presence of a nest queen is the production of fewer new queens (Brian and Hibble, 1963).

The workers of the fire ant Solenopsis invicta are able to distinguish sexual pharate pupae (“prepupae”) from worker brood, and a key component used by the ants in recognition is triolein, the triglyceride of oleic acid, which is present in many vegetable oils and in substantial amounts in the pupae. During experiments by Bigley and Vinson (1975), filter-paper disks were placed by the ants with groups of sexual pharate pupae rather than with worker-destined brood. However, these results were recently criticized by Morel and Vander Meer (1988) on the grounds that the experimental design was likely to confound brood retrieval and food retrieval.

The tendency to segregate eggs, larvae, and pupae into separate piles is a nearly universal trait in ants, and probably has a basis in chemical differentiation like that demonstrated for the fire ant sexual pupae. Workers of many species are, furthermore, able to distinguish larvae of two or more size classes (LeMasne, 1953). Workers and larvae are also able to distinguish trophic eggs (special eggs produced for consumption) from the normal eggs destined to develop into larvae. Monomorium pharaonis workers are even able to tell male eggs from female eggs (Peacock et al., 1954).

As Carlin (1988) has pointed out, kin recognition in brood discrimination by workers need not be limited to the differential feeding, grooming, and retrieval of close relatives. Brood cannibalism is well known in ants and the eating of viable eggs has been reported among queens in Leptothorax curvispinosus (Wilson, 1974a). Preliminary results on oophagy in this species and Leptothorax longispinosus suggests that queens may recognize their own eggs (Graham and Carlin, personal communication). Discrimination of eggs based on kin has also been suggested in oligogynous colonies of Iridomyrmex purpureus (Hölldobler and Carlin, 1985).

Although brood can easily be transferred among conspecific, and within limits, even among interspecific colonies, new evidence clearly demonstrates that discrimination among nestmate and non-nestmate brood also occurs in ants. Carlin (1988) observes that "this behavior differs strikingly from the antagonism and rejection of alien adults by the same species. In choice situations (such as the retrieval bioassay), non-nestmate conspecific brood are not rejected; rather, both are accepted, but nestmate brood are favored (retrieved earlier or in greater numbers). That is, preference for nestmate adults is usually absolute and exclusive, while preference for nestmate brood is probabilistic and non-exclusive."

Nestmate brood discrimination has been demonstrated in a phylogenetically diverse group of ant species (see Table 5-2). The most detailed investigation of the phenomenon was conducted on the desert ant Cataglyphis cursor. Lenoir (1984) discovered that when newly-eclosed workers are transferred into an alien conspecific colony, they engage in less brood care than do control workers of equal age in their own colony. Furthermore, Isingrini et al. (1985) found evidence suggesting that Cataglyphis workers respond differentially to nestmate and non-nestmate brood because they had learned to recognize nestmate brood during larval life. Since workers of this species are known to produce female eggs parthenogenetically and also to migrate frequently among neighboring nests, the adaptive significance of nestmate and kin recognition in Cataglyphis cursor is of special interest (Lenoir et al., 1987). Finally, an interesting observation concerning differential brood treatment was discovered in Formica lugubris by Rosengren and Cherix (1981). Colonies from populations in Switzerland preferred to retrieve nestmate pupae over conspecific pupae from Finland, but did not discriminate against pupae from geographically closer Swiss and Italian populations. In symmetric fashion, workers from Finland discriminated strongly against pupae from Switzerland and Italy.

Since the early work of Forel (1874) and Fielde (1903) it has been known that brood can be transferred not only between nests of the same species but between those of different species. Hölldobler (1977) hypothesized that brood pheromones are not colony-specific and often even not species-specific, but are so highly attractive to worker ants in general as to override or mask the response to colony specific odors. Indeed, Brian (1975) demonstrated that the recognition signals indicating reproductive potential in Myrmica rubra and Myrmica scabrinodis remain effective when queen larvae are cross-fostered between species. Many other examples of cross-fostering within a genus and even among different genera exist. For example, Plateaux (1960a-c) was able to exchange larvae among several myrmicine genera. The responsiveness of ant workers to generalized brood signals is the basis for social parasitism and interspecific slavery in ants (see Chapter 12), and it may also be the essential prerequisite for parasitic symbiosis in many myrmecophilous arthropods (Chapter 13).

As Carlin (1988) has emphasized, however, interspecific brood transfers in general tend to be more successful when phylogenetically close taxa are used. He states:

Emery's (1909) rule, that social parasites are closely related to their hosts, may be attributed to the greater similarity of brood (and adult) cues among species with common ancestors; however this pattern applies equally to species uninvolved in parasitism. Jaisson (1985) reported that workers of Formica spp. may recognize and care for pupae of other formicine genera (Camponotus and Lasius), but reject those genera in other subfamilies (Atta and Eciton). Founding queens of Camponotus pennsylvanicus have narrower tolerance (Carlin, unpublished results). Queens will adopt larvae and pupae from 14 congeneric species (Camponotus abdominalis, Camponotus americanus, Camponotus castaneus, Camponotus consobrinus, Camponotus ferrugineus, Camponotus festinatus, Camponotus floridanus, Camponotus herculeanus, Camponotus nearcticus, Camponotus novaeboracensis, Camponotus ocreatus, Camponotus socius, Camponotus maculatus tortuganus (now a junior synonym of Camponotus inaequalis), and Camponotus vicinus), but not those of Camponotus planatus and Camponotus ulcerosus--which belong to the morphologically aberrant subgenera Myrmobrachys and Manniella, respectively. Camponotus pennsylvanicus foundresses destroy or reject brood of the formicine genera Acanthomyops, Formica, and Lasius (though a few Lasius neoniger larvae may be retained for nearly a month before finally being eaten), and always reject brood from other subfamilies (Ponerinae: Amblyopone; Myrmicinae: Aphaenogaster, Crematogaster, Novomessor]; Dolichoderinae: Iridomyrmex).

Foreign brood pieces that have been adopted are occasionally allowed to mature into adults, leading to the formation of mixed worker forces (Fielde, 1903; Carlin and Hölldobler, 1983; Errard and Jaisson, 1984). Nonetheless, the adoptees are then often attacked and killed by their hosts (Fielde, 1904a; Haskins and Haskins, 1950a; Plateaux, 1960c; Wilson, 1975a). A further complication was observed by Carlin (1988):

When heterospecific brood was adopted and reared to adulthood, the adoptee workers may tend host brood but kill host workers as they eclose (e.g., Camponotus (Myrmothrix) floridanus workers adopted into C. (Camponotus) pennsylvanicus colonies; C. (Tanaemyrmex) tortuganus adoptees of a C. (Hypercolobopsis) paradoxus queen). These results together suggest that the interspecific similarity of brood pheromones is greater than that of adult (species-level) recognition cues.

In discriminating against alien brood, worker ants do not behave in an “all or nothing” response, but rather exhibit a preferential choice behavior. For example, workers of Camponotus floridanus and Camponotus maculatus tortuganus, reared with conspecific brood picked up and retrieved non-nestmate conspecific pupae significantly earlier than pupae of the other species. The heterospecific pupae were also retrieved, but only 35 percent of them were retained and tended after 5 days, whereas 69 percent of the conspecific pupae were still alive in the nest after that period (Carlin et al., 1987a). Similar results were obtained by Hare and Alloway (1987) for Leptothorax ambiguus and Leptothorax longispinosus. Workers of each species preferred to accept conspecific larvae, but they also adopted a smaller proportion of the alien brood. Finally Elmes and Wardlaw (1983) report that they were often, but not always able to introduce larvae of several Myrmica species into heterospecific colonies; they found larval survival to be usually better in conspecific colonies.

Carlin (1988), in his extensive review of brood discrimination in ants, describes experiments by different investigators that indicate an absolute preference on the part of some ant species:

Exclusive conspecific preferences have been reported in three species of Formica. When Formica polyctena workers that had been reared with brood of their own species were offered the choice of conspecific and heterospecific pupae (Formica sanguinea or Camponotus vagus), the former were retrieved first, but heterospecific pupae were also collected. Exclusivity was revealed in their subsequent treatment, as conspecific pupae were retained while heterospecifics were rapidly eaten (Jaisson, 1975; Jaisson and Fresneau, 1978). Formica rufa workers similarly offered conspecific pupae and those of Formica lugubris retrieved and tended only the former in one experiment (Le Moli and Passetti, 1977), but in another test they brought both species into the nest and neglected and destroyed the latter (Le Moli and Passetti, 1978). Conversely, Le Moli and Mori (1982) found that Formica lugubris workers retrieved only conspecifics, and ate Formica rufa pupae outside the nest. Thus exclusive preferences were manifested consistently by destruction of the non-preferred species, though variable retrieval responses also occurred.

These findings bear re-examination, however. In a review, Jaisson (1985) reports that the same basic results were obtained with a species of Camponotus, suggesting methodological dissimilarities between these studies and that of Carlin et al. (1987a). One potentially relevant difference is the use of cocoons killed by freezing by Jaisson, Le Moli and their respective colleagues in all experiments that yielded exclusive preferences. The studies of Hare and Alloway (1987) and Carlin et al. (1987a), demonstrating non-exclusive biases in both retrieval and retention, employed live larvae and pupae respectively. Ants are well known to eat dead or injured brood (Wilson, 1971), and brood killed by freezing have been shown to decline in attractiveness (Robinson and Cherrett, 1974; Walsh and Tschinkel, 1974). Having retrieved two species of dead pupae, the Formica spp. workers might cannibalize first the less familiar . . .; the thorough destruction of all heterospecifics needs to be confirmed by offering live pupae. In addition, Formica polyctena is sometimes the host of a temporary social parasite, Formica truncorum (Gösswald, 1957; Kutter, 1969), indicating that heterospecific parasite brood must be sufficiently attractive to adult workers reared under normal field conditions. This same difficulty applies to the report that workers of Lasius niger respond only to conspecific brood and destroy all others (Le Moli and Angeli, cited in Le Moli, 1980), as this species is the natural host of temporary social parasite Lasius umbratus (K. Hölldobler, 1953).

Additional discussions of the brood discrimination problem, with special reference to social parasitism, are presented by Le Moli and Mori (1987) and in Chapter 12 of the present book.

Source of the colony odor

In most if not all cases studied in ants so far, identification of nestmates and life stages is by antennal contact. This fact in itself suggests chemoreception, even though Brian (1968) has speculated that several age classes of Myrmica larvae might also be distinguished by certain differences in hairiness which are quite apparent to the human observer. At least one example has been reported in which the chemoreception appears to operate over a distance. When workers of the harvester Pogonomyrmex badius lay trophic eggs, they search for hungry larvae while sweeping their antennae through the air. When they come within about a centimeter of the head of the larva they move directly to it and unerringly place the egg onto its mouthparts (Wilson, 1971). This response appears to be exceptional, however. The employment of close-range olfaction in the discrimination of life stages has been proved by the identification of several of the surface pheromones involved: triolein in the sexual pharate pupae of Solenopsis invicta as described above, two pyranones and dihydroactinidiolide as components of the queen attractant of Solenopsis invicta (Vander Meer et al., 1980; Glancey, 1986), and neocembrene as a queen attractant in Monomorium pharaonis (Edwards and Chambers, 1984).

No colony odor by which kin are recognized has yet been identified with certainty, but some tantalizing clues exist. One is the assimilation of the cuticular hydrocarbons of fire ants by the scarabaeid beetle living with them, Myrmecaphodius excavaticollis. The relations of the two insects are harmonious, despite the fact that the beetles prey on the ants. Vander Meer and Wojcik (1982) discovered that the Myrmecaphodius are coated by a series of hydrocarbons identical to those of one of its fire ant hosts, Solenopsis richteri. The beetles also possess a second set of high molecular weight hydrocarbons not shared with the ants and evidently peculiar to themselves. When Myrmecaphodius adults are isolated from the ant hosts, they lose the Solenopsis richteri hydrocarbons but retain their own, heavier hydrocarbons. When now introduced into colonies of Solenopsis invicta, a second fire ant host species, the beetles acquire the Solenopsis invicta hydrocarbons. On this basis Vander Meer and Wojcik postulated that the ants use the hydrocarbons for odor identification among themselves, at least to the level of species, and that the beetles exploit this behavior by acquiring the molecular signature. When Myrmecaphodius first enter Solenopsis nests, they rely on their heavily armored exoskeleton and a death-feigning behavior for protection. After a few days, according to the hypothesis, enough hydrocarbons are adsorbed to gain fuller acceptance into the ant society.

Nelson et al. (1980) who first identified the surface material of the fire ant as saturated normal, mono-, and dimethyl branched hydrocarbons, found that [[Solenopsis invicta and [[Solenopsis richteri have the same substances (as evidenced by their chromatographic peaks), but the relative amounts are so different as to make the species easily distinguishable. More recently, Vander Meer (1986b) showed that Solenopsis geminata and Solenopsis xyloni also possess distinctive patterns based on proportionality. Thompson et al. (1981) discovered that the five major hydrocarbons occur in the postpharyngeal glands of queens of Solenopsis invicta, while Vander Meer et al. (1982) found that the quantities stored there are quite large (between 15 and 50 µg), increasing substantially and shifting in proportion shortly after the ants mate. The main function of the postpharyngeal gland is generally regarded as a source of food for larvae. In fire ant queens, these organs are fully engorged with oils just prior to the nuptial flight (Phillips and Vinson, 1980). Whether they also provide odor signatures for the colony remains an untested possibility.

The first evidence of differences in hydrocarbon proportions existing between colonies of ants, and not just between species, has been adduced by Clement et al. (1987), Bonavita-Cougourdan et al. (1987a,b), and Morel and Vander Meer (1987) in the carpenter ants Camponotus floridanus and Camponotus vagus. By direct chemical analysis they found that at least some of the colonies differ among themselves in the hydrocarbon blends. When live workers of the focal colony, dead workers, and dummies were washed with solvents containing extracts of workers from alien colonies of the same species, other workers from the focal colony reacted aggressively to them, whereas the responding workers reacted in a neutral or at most mildly aggressive manner to workers and dummies washed with extracts of their nestmates. This experiment is suggestive but does not constitute sufficient proof that the critical components in the washes were hydrocarbons. A complicating factor is the additional finding by Bonavita-Cougourdan et al. (1987b) that the hydrocarbon proportions change with age. Such shifts could serve as the base of subcaste recognition, such as discrimination of foragers from nurses, but it undoubtedly makes recognition of nestmates as a whole more difficult. Similar results have been obtained by Morel and Vander Meer (1987) and Morel et al. (1988) for Camponotus floridanus. Blum (1987) has reviewed the recent findings on the possible role of cuticular hydrocarbons for species and nestmate recognition in social insects.

The first claim of the discovery of a glandular source of colony odors was made by Jaffe and Sanchez (1984a,b) in the course of studies of the Neotropical formicine ant Camponotus rufipes. They concluded that the signature comes from the mandibular glands, which are also well known to produce alarm substances. Two proofs were offered. First, Jaffe and Sanchez reported that isolated heads of workers from alien colonies are bitten significantly more and elicit more alarm responses than do isolated heads from the same colonies, whereas isolated thoraces and gasters do not cause this differential response. The result cannot be taken as conclusive, however, since some additional substance in the head, including the mandibular gland substances themselves, might simply excite the target ants and lower their attack threshold in response to colony odor found uniformly over the entire body. A second bioassay employed by Jaffe and Sanchez seems stronger but is still less than definitive. Plastic dummies were contaminated with mandibular gland secretions of alien workers, and others with the secretions of nestmates of the target workers. The alien-treated models elicited more bites (3.5 ± 1.8, mean ± standard deviation) than did the nestmate-tested dummies (1.2 ± 0.9). The result is compromised by the possibility that the secretions could have been contaminated with body odors originating elsewhere in the body. On the other hand, as we have noted in Chapter 7, colony specific blends in the mandibular gland secretions could also function as modulators of alarm signals, and do not need to be the source of the colony specific nestmate recognition label.

Leaving for the moment the chemical identity and anatomical source of the colony odor, we can take up the separate question of its intrinsic as opposed to its extrinsic control. The existence of an important hereditary component in at least some ant species has been established by several lines of evidence. Haskins and Haskins (1979) found that different colonies of the Australian ponerine Rhytidoponera metallica retained the same level of incompatibility from the time they were collected in the field through 13 years of identical laboratory conditions. When worker pupae were placed with workers of other laboratory colonies, allowed to eclose into active adults, and then placed with their home colony, they were fully accepted. An equally convincing experiment has been performed by Mintzer and Vinson (1985b) with the arboreal ant Pseudomyrmex ferrugineus, an obligate symbiont of Acacia in Mexico. The Pseudomyrmex are wholly bound to the Acacia trees, nesting on them and feeding in large part on secretions from the foliar nectaries and on Beltian bodies, special food organs that grow at the tips of the leaves. Mintzer and Vinson raised a series of Pseudomyrmex colonies on clones of Acacia hindsii under uniform conditions. Although the environmental variance was thus reduced to a minimum, the Pseudomyrmex colonies retained their mutual incompatibility. Of equal significance, colonies belonging to inbred lines of the ants underwent a slight increase in compatibility when compared with outbred controls.

Another experiment, this time utilizing the leafcutter ant Acromyrmex octospinosus, has demonstrated the existence of endogenous factors, probably genetic in origin, as well as exogenous factors originating in the food (Jutsum et al., 1979). Acromyrmex octospinosus colonies subsist entirely on a symbiotic fungus grown on freshly cut leaves and other vegetation, possibly along with sap released from fresh incisions. Jutsum et al. used as a measure of hostility the time spent by workers investigating strange workers. Reintroductions of ants back into their own nests provided the control tests. Prior to the tests various colonies were maintained on vegetation from different species of plants. The results demonstrate that both endogenous and exogenous factors contribute to the differentiation of colony odor (Table 5-2).

Given that at least part of the colony odor is endogenous, the question now arises as to which individuals produce it. Is it generated by one dominant member, such as the queen, or is it a gestalt drawn from many or all of the colony members? Carlin and Hölldobler (1983, 1986, 1987) addressed this question by creating a series of experimental colonies of Camponotus species, variously containing (1) a queen and workers, (2) queenless worker groups, and (3) pairs of worker groups between which a queen was repeatedly switched. Some colonies of the first and second type also received different diets. The goal of this regime was to assess the relative magnitude of the contributions of the queen, the workers, and the environment under strictly controlled laboratory conditions. Queenless workers, removed from the same colony serving as the source of pupae but reared separately, were relatively tolerant of one another but more aggressive toward non-relatives. Dietary differences slightly enhanced aggression among separately-reared kin. However, workers reared in queenright colonies (Figure 5-5) attacked all non-nestmates, whether kin or non-kin, and with equal violence. This response was unaffected by diet odors or by the proportion of different kin groups in the colony (Figure 5-6). Non-kin sharing a switched queen were as tolerant of one another as were kin, and cues derived from healthy queens were sufficient to label colonies of approximately 190 workers each. The workers' own discriminators became more important in large groups with queens that had less active ovaries and were relatively infertile. From these results, Carlin and Hölldobler proposed a hierarchy of colony-specific label sources in Camponotus colonies, with cues derived from fertile queens most important, followed by worker discriminators, and then environmental odors.

A different system, or at least a different order in the hierarchy of cues, appears to be employed by the Mexican Acacia-ant Pseudomyrmex ferrugineus. Mintzer (1982b) separated worker brood from a stock colony into groups of larvae and pupae, then allowed each group to be reared by a different foster queen. The workers from the separate groups were not antagonistic toward each other, indicating that their own odors (rather than those of the foster queens) were the dominant discriminators. In a follow-up study, Mintzer and Vinson (1985b) found that nestmates were often treated differently by colonies into which they were introduced. This variation in response favors an individualistic rather than a gestalt mode of colony odor production. Experiments with inbred colonies further indicated that the odor labels are under multilocus rather than single-locus genetic control. The experiments by Mintzer and Vinson were not designed to learn whether the odor identifications in Pseudomyrmex are entirely genetic or, as in the case of Camponotus, require some form of learning.

A still different system has been discovered in Leptothorax curvispinosus, a myrmicine species that forms facultatively polygynous and polydomous colonies. Stuart (1987c) measured the acceptance of workers among nests of three categories: (1) the nests were relatively near to one another (0.09-1.87 m), (2) the nests were spaced farther apart (1.52-4.65 m), and (3) the nests were very distant (7 kilometers). In the first group the workers sometimes accepted each other without any aggression, and they were generally significantly less hostile to one another than were workers from sites 7 kilometers apart. The aggression between foreign workers within the second group was initially also very strong. However, after being cultured for several weeks under uniform environmental conditions, the workers grew significantly more friendly. The results indicate that polydomous colonies of Leptothorax curvispinosus occur within multicolonial populations, and that colony segregation within local populations is largely maintained by transient environmentally-based nestmate recognition cues. Stuart proposes that "these results suggest that the maintenance of colony autonomy within genetically highly interrelated populations may be the prime function of environmentally-based nestmate recognition cues. Colony autonomy under such circumstances may be important to maintain a relatively small but optimal colony size, or because the mechanisms which regulate colony growth development, etc., require a limited colony size."

Stuart (1987a) also demonstrated more stable cues of possible genetic origin that contribute to discrimination even after extended periods of uniform laboratory environment. In one of the studies worker pupae were removed from colonies and allowed to eclose and age for 38-157 days in isolation. When these workers were introduced into their parental colonies, they were readily accepted. But they were frequently attacked and even killed when introduced to non-parental conspecific colonies. These results demonstrate that individual workers produce persistent, colony-specific recognition cues, even when completely isolated from the parental nest. Stuart suggests that the cues involved may be either genetically based or acquired prior to pupation.

Finally, using interspecific mixed Leptothorax colonies Stuart (1985a, 1987b) demonstrated that workers are labeled not only by transient environmental odors and individually produced cues, but also by a colony gestalt odor. His is the first demonstration of the existence of a gestalt label in ants. The second evidence that labels are transferred among nestmates stems from the work of Errard and Jallon (1987), using interspecific mixed colonies of Manica rubida and Formica selysi. A gestalt label had previously been demonstrated only in the desert woodlouse Hemilepistus reaumuri (Linsenmair, 1985, 1987). In this system, which is also called the "Collective System" (Stuart, 1988a,b), nestmates share recognition cues, and each colony member bears some kind of mixture of cues representative of the variation among its nestmates. Although the gestalt label does not appear to be completely uniform in Leptothorax colonies, Stuart (1988b) concludes that the


Collective System may have evolved as one means of reducing the degree of individual variation in individually produced nestmate recognition cues among colony members. Nonetheless, as evidenced by the Leptothorax data, this solution to the problem of individual variation may be far from perfect. Although this system might lessen individual variation to some extent and permit larger colony sizes, the dynamics of odor sharing and the relative degree of heterogeneity in individually borne cues might still impose a fairly small limit on colony size and compromise nestmate discrimination and colony defense when larger colony sizes are achieved. This might explain the relatively small size typical of Leptothorax colonies, and reports of relatively low levels of aggression between colonies under some circumstances (Provost, 1979).

The work by Stuart (1988a,b), Jutsum et al. (1979), Obin (1986) and others on ants, together with that by Gamboa and his collaborators (Gamboa et al., 1986) on social wasps, suggests that social insects are programmed to respond to odor differences without regard to ultimate origin. Thus whereas the acquisition of unifying extrinsic labels seems common, the particular sources involved (or their hierarchy of importance in recognition) varies widely as a function of the relative strengths of ambient odors available to each species. These strengths are fixed in turn by the general biology, ecology and social organization of the species. Among ant genera, for example, it may prove significant that Camponotus exhibits greater queen-worker dimorphism and stronger queen suppression of worker oviposition than does Leptothorax, and that Solenopsis workers do not require queen suppression for the simple reason that they lack ovaries. On the other hand it is possible that colonies of the same species employ different strategies suited to their particular ontogenetic stage. For example, Stuart (1987c) suggests that in Leptothorax “special integrating mechanisms associated with newly eclosed workers may be required only under certain circumstances, perhaps during early colony growth, after young reproductives have been adopted into established colonies, or when newly eclosed adults must acquire additional cues to achieve acceptability.” It seems also possible that in smaller colonies of Camponotus (with several hundred workers) the queen has a stronger influence on the colony label than in colonies with several thousand workers. The different components could also be weighted differently according to age and location of the workers. For example the queen influence could be much stronger in the "core area" around the queen, affecting primarily younger workers, whereas in the peripheral zone of the colony, occupied by older workers, discrimination might be based more on gestalt-labels derived from older nestmates.

A Camponotus-like hierarchical system, with queen odor overriding that of the workers, may be found to prevail in species that are either monogynous, oligogynous (two to several queens with approximate reproductive parity), or polygynous with one queen predominant in egg-laying. On the other hand truly polygynous species, in which many queens are active egg-layers, can be expected to rely on a gestalt odor contributed in greater measure by the workers. In addition, extremely polygynous species in large colonies may lose their capacity to create colony odors of any kind, since the large variety of genotype represented in each unit is likely to be duplicated in adjacent units by the law of large numbers and inevitable minimization of between-unit variance. The ultimate result would be a supercolony, or unicolonial population, in which the local population constitutes just one vast colony. Supercolonies with large populations of queens or reproductive workers do occur in some species of Pheidole, Monomorium, Pristomyrmex, Wasmannia, Iridomyrmex, and Formica. In such cases the generating of intrinsic colony odors has virtually ceased, but some odor differences due to different extrinsic influences might still exist.

The ontogeny of recognition

It is now generally accepted that learning plays a major role in the acquisition of a template for labels used in colony-level recognition. Carlin (1988) has recently reviewed the work on the significance of learning in brood recognition in ants; we can do no better than quote his concise analysis of the diverse studies published:

Jaisson (1975) demonstrated that newly-eclosed Formica polyctena workers learn to recognize any brood species with which they are familiarized. A process comparable to imprinting takes place in a sensitive period of approximately one week following emergence (Jaisson and Fresneau, 1978). Young workers exposed during this period only to pupae from another species (Formica pratensis, Formica sanguinea, Camponotus vagus or Lasius niger) retrieved the familiar heterospecific pupae first and destroyed the unfamiliar conspecific after retrieval. Provided with conspecific pupae, they preferred familiar conspecifics and destroyed unfamiliar heterospecifics; deprived of any experience with brood, they ignored or destroyed all species impartially. These results were replicated with young Formica lugubris and Formica rufa workers, exposed to their own or one anothers' pupae or deprived of brood experience, by Le Moli and Passetti (1977, 1978) and Le Moli and Mori (1982). Le Moli (1978) also reported apparent social facilitation of the brood-tending response. Formica rufa workers deprived of brood through the sensitive period, then placed together with experienced workers and pupae from their colony of origin, subsequently retrieve and tend both conspecific and Formica lugubris pupae, rather than destroying them.

In the genus Leptothorax, by contrast, preference for conspecific brood can be eradicated by early experience with heterospecifics, but not reversed. Alloway (1982), examining the effect of enslavement on pupa acceptance in three Leptothorax species, found that free-living, previously and currently enslaved workers always preferred conspecific pupae to heterospecifics. The experience of enslavement did increase the proportion of heterospecific pupae accepted by the workers; behavioral interactions were also implicated, as slaves accepted more heterospecific pupae in the presence of the slave-makers (Harpagoxenus americanus) than if the latter were removed. Hare and Alloway (1987) exposed newly eclosed Leptothorax ambiguus and Leptothorax longispinosus workers either to conspecific or to each others' larvae, or deprived them of brood (isolation). Workers familiar with conspecific larvae retrieved and retained significantly more conspecifics, but those exposed to heterospecific larvae or isolated accepted larvae of both species without preference. Carlin et al. (1987a) independently obtained the same results using Camponotus floridanus and Camponotus maculatus tortuganus workers, which preferred conspecific pupae if familiar with them, but retrieved both species indiscriminately if naive or exposed only to heterospecifics. However, naive young workers were not entirely incapable of brood species recognition, as they retained unfamiliar conspecifics significantly longer than unfamiliar heterospecific pupae. In addition, the behavior of older workers (20 days post-eclosion), already exposed to conspecific pupae in their natal nest, proved to be malleable to a limited extent. Older adults retrieved conspecific pupae preferentially irrespective of recent experience, but their retention of heterospecific pupae was significantly increased by recent familiarization; since the effect of early learning can be subsequently modified, this behavior does not qualify as imprinting strictly defined.

Thus while newly-eclosed workers of Formica spp. respond to experience with any species, those of Camponotus and Leptothorax spp. are apparently equipped with a somewhat modifiable bias that favors the learning of their own species, and cannot be made to prefer the brood of others. These genera may possess some innate "template" mechanism that restricts the range of effective learning stimuli, predisposing them to learn conspecific brood signals, or the workers may simply respond more strongly to odors that resemble their own. Alternatively, pre-imaginal learning may be involved.

The ontogeny of inter-colony recognition [of brood] has been addressed in Cataglyphis cursor, and early learning is clearly involved (Isingrini et al., 1985). The data are in fact strikingly similar to those on species recognition in Camponotus and Leptothorax: newly eclosed workers that were exposed to nestmates licked and carried nestmate larvae significantly more than conspecific alien larvae. Those provided post-eclosion experience with non-nestmate larvae or kept in isolation exhibited significantly weakened preferences for nestmate larvae, but non-nestmates were not preferred. Pre-imaginal learning, that is learning of ambient larval recognition cues by larvae, was implicated by transferring eggs from their natal colony into alien colonies, allowing them to develop there, returning them to the natal nest as pupae and then testing them after eclosion. Larvae from the colony in which the workers spent their own larval life were preferred to larvae from their natal nest. Transferring large larvae rather than eggs did not alter the preference for natal nest brood, perhaps because the larvae spent only 2-5 days in the adoptive colony before pupating. Pre-imaginally induced biases declined as the workers aged, a factor which must be investigated further before the significance of larval learning can be evaluated. However, this is an intriguing possibility, which may be important in other forms of recognition as well.

Similar results were recently obtained with Camponotus floridanus by Carlin and Schwartz (personal communication).

Because colony odor is learned in part, yet another property of colony and species discrimination is the time the workers acquire the odor in workers and the time at which they learn the odor of their group. Fielde (1903, 1904a,b, 1905b) discovered that the colony odor changes with age. Aphaenogaster rudis workers, segregated at the pupal stage and isolated as a group until they were 16 to 20 days into adult life, accepted additional alien rudis workers up to 40 days after they emerged from the pupa, but attacked workers older than that as well as queens. Indeed, Bonavita-Cougourdan et al. (1987a,b) discovered that different age classes of Camponotus vagus workers bear distinct blends of cuticular hydrocarbons. Fielde proposed a progressive odor hypothesis. She proposed that certain hereditary nestmate recognition odors in adults change progressively with age, so that workers often do not accept older siblings differing from them in age, unless they had eclosed among nestmates of that age class and learned their recognition cues. In a recent reexamination of this hypothesis, Stuart (1987b,c) was unable to confirm it for three species of Leptothorax. In contrast, workers automatically acquire a distinctive odor no later than 38 days after eclosion, even when isolated from the mother colony as far back as the pupal stage. They are then readily accepted by their own colony but not by alien colonies.

In fact, very few studies have been made of the precise timing of the odor acquisition or the sensitive period at which odor is learned. The mixed-colony experiments of Carlin, Fielde, Hölldobler, Jaisson, Lenoir, Morel, Vander Meer, and others have demonstrated that brood are far more readily adopted by alien colonies than adults, while very young (callow) workers are more acceptable than their older nestmates. More detailed information on the ontogeny of acceptability of young adults has recently been obtained through the work of Lenoir et al. (1987) on Cataglyphis cursor, Morel and Vander Meer (1987) and Morel et al. (1988) on Camponotus floridanus, Clement et al. (1987) and Bonavita-Cougourdan et al. (1987) on Camponotus vagus, and Carlin and Hölldobler (1986) on Camponotus pennsylvanicus (Figure 5-7). C. pennsylvanicus workers a few days old, like brood, can be transferred among colonies without suffering much mortality. Adoptability then declined, from about the fifth day following eclosion onward. Workers accepted initially were not attacked at any time afterward; they presumably acquired cues of the recipient colony before their own discriminators developed or their immunity to attack wore off. Most callows over one week old were killed within eight hours of introduction into a queenright colony. More than 98 percent of all introductions, including those involving the youngest callows, provoked at least initial threats at level 1 (see Table 5-1). Only nine callows were immediately accepted at level 0, of which eight were less than two days old, and one was nine days old. These results represent discrimination against callows by the recipient colonies exclusively, and not discrimination on the part of the callows themselves, which had not yet become aggressive.

Entomologists have scarcely begun to explore the intricacies of learning in kin recognition among social insects. Identification and discrimination are major features in biological systems, from embryogenesis and immune responses to communities and ecosystems, and it is no surprise to find them developed to a high degree in ants (Hölldobler and Carlin, 1987). All forms of recognition require distinguishable signals, whether antigens or colony odors, varying in ways that are correlated with the evolutionary advantages they confer.


Hölldobler, B. and Wilson, E. O. 1990. The Ants. Cambridge, Mass. Harvard University Press. Text used with permission of the authors.

The Ants - Table of Contents