Brood production and lineage discrimination in the red harvester ant (Pogonomyrmex barbatus)

In contrast to the system of caste determination in most social insects, reproductive caste determination in some populations of Pogonomyrmex barbatus has a genetic basis. Populations that exhibit genetic caste determination are segregated into two distinct, genetic lineages. Same-lineage matings result in female reproductives, while inter-lineage matings result in workers. To investigate whether founding P. barbatus queens lay eggs of reproductive genotype, and to determine the fate of those eggs, we genotyped eggs, larvae, and pupae produced by naturally inseminated, laboratory-raised queens. We show that founding dependent lineage queens do lay eggs of reproductive genotype, and that the proportion of reproductive genotypes decreases over the course of development from eggs to larvae to pupae. Because queens must mate with a male of each lineage to produce both workers and female reproductives, it would benefit queens to be able to distinguish males of the two lineages. Here we show that P. barbatus males from the two genetic lineages differ in their cuticular hydrocarbon profiles. Queens could use male cuticular hydrocarbons as cues to assess the lineage of males at the mating aggregation, and possibly keep mating until they have mated with males of both lineages.     

Artificial neural network modeling of caste odor discrimination based on cuticular hydrocarbons in termites

Summary. Individuals in an insect colony need to identify one another according to caste. Nothing is known about the sensory process allowing nestmates to discriminate minute variations in the cuticular hydrocarbon mixture. The purpose of this study was to attempt to model caste odors discrimination in four species of Reticulitermes termites for the first time by a non-linear mathematical approach using an "artificial neural network" (ANN). Several rounds of testing were carried out using 1 – the whole hydrocarbon mixtures 2 – mixtures containing the hydrocarbons selected by principal component analysis (PCA) as the most implicated in caste discrimination. Discrimination between worker and soldier castes was tested in all four species. For two species we tested discrimination of four castes (workers, soldiers, nymphs, neotenics). To test cuticular pattern similarity in two sibling species (R. santonensis and R. flavipes), we performed two experiments using one species for training and the other for query. Using whole hydrocarbons mixtures, worker/soldier discrimination was always successful in all species. Network performance decreased with the number of hydrocarbons used as inputs. Four-caste discrimination was less successful. In the experiment with the sibling species, the ANN was able to distinguish soldiers but not workers. The results of this study suggest that non-linear mathematical analysis is a good tool for classification of castes based on cuticular hydrocarbon mixture. In addition this study confirms that hydrocarbon mixtures observed are real chemical entities and constitute a true chemical signature or odor. Whole mixtures are not always necessary for discrimination. Received 23 July 1998; accepted 9 October 1998.     

Heirs and spares: caste conflict and excess queen production in Melipona bees

The caste conflict hypothesis states that there is potential conflict over the caste fate of totipotent immature females in social insects. In most species, an immature female has little control over her fate because workers control her nutrition. However, in Melipona bees, immature females should have considerable control over their own caste fate because they develop on a provision mass in a sealed cell, and because queens are not larger than workers. This may explain why, in Melipona, large numbers of queens are reared only to be executed. (Because Melipona colonies are founded by swarms very few reproductive opportunities for adult queens occur.) This study uses a one-locus genetic model to determine the optimum proportion of females that should develop into queens from the perspective of immature totipotent females who control their own caste fate. For a population in which all colonies are headed by a single, single-mated queen, which is the typical situation in Melipona, the optimum rises from 14-20% as male production by workers declines from 100% to zero. This agrees well with previous studies which, collectively, give an average of 22% of females developing into queens.     

Nestmate recognition in the ant Cataglyphis niger : do queens matter?

This study compares two basic models for the origin and maintenance of colony gestalt odor in the polygynous ant species Cataglyphis niger. In the first model, queens are centers of de novo biosynthesis and distribution of recognition odors (“queen-centered” model); in the second, colony odors are primarily synthesized and distributed by workers (“worker-centered” model). We tested the behavioral patterns that are predicted from each model, and verified by biochemical means the distributional directionality of these signals. Encounters between nestmates originating from split colonies were as amicable as between nestmates from non-split colonies; queenless ants were as aggressive as their queenright nestmates, and both were equally aggressed by alien ants. These results indicate that queens have little impact on the recognition system of this species, and lend credence to the worker-centered model. The queen-centered model predicts that unique queen substances should be produced in appreciable quantities and that, in this respect, queens should be more metabolically active than workers. Analysis of the chemical composition of postpharyngeal glands (PPGs) or cuticular extracts of queens and workers revealed high similarity. Quantitatively, queens possessed significantly greater amounts of hydrocarbons in the PPG than workers, but the amount on the thoracic epicuticle was the same. Queens, however, possess a lower hydrocarbon biosynthesis capability than workers. The biochemical evidence thus refutes the queen-centered model and supports a worker-centered model. To elucidate the directionality of cue distribution, we investigated exchange of hydrocarbons between the castes in dyadic or group encounters in which selective participants were prelabeled. Queens tended to receive more and give less PPG content, whereas transfer to the epicuticle was low and similar in all encounters, as predicted from the worker-centered hypothesis. In the group encounters, workers transferred, in most cases, more hydrocarbons to the queen than to a worker. This slight preference for the queen is presumably amplified in a whole colony and can explain their copious PPG content. We hypothesize that preferential transfer to the queen may reflect selection to maintain her individual odor as close to the average colony odor as possible. Received: 4 November 1997 / Accepted after revision: 5 February 1998     

On being the right size: male contributions and multiple mating in social Hymenoptera

Summary A number of hypotheses for the occurrence of multiple mating by queens of social Hymenoptera are reviewed in the light of Cole's (1983) observation that polyandrous species tend to have larger colonies than single-mating ones. Most of these hypotheses cannot be definitively excluded, but only three of them appear sufficiently general, plausible and predictive to be useful guides to further research. These, and their predictions, are: (1) Caste-determination has a genetic basis and hence polyandry allows fuller expression of the potential caste system in each colony. Species with more complex caste differentiation should be more often polyandrous than species with simpler caste systems. (2) Polyandry maximises the production of divergent worker genotypes and hence the range of environmental conditions that a colony can tolerate. Broader-niched species should be more often polyandrous than species with narrower niches. (3) The reduction of the variance of diploid male production, under the heterozygosity sex-determination model, favors polyandry when sexuals are produced late during colony growth. Queens in species reproducing during the exponential phase of colony growth should tend to mate once, but queens should tend to be polyandrous in species with reproduction occurring further along the colony growth curve. Williams's (1975) observation that the level of genetic variation in a brood approaches a maximum very quickly with increasing polyandry is quantified for females; the initial increase is much greater for male-haploid than for male-diploid species.     

Behavioral regulation of genetic caste determination in a Pogonomyrmex population with dependent lineages

The fate of a social insect colony is partially determined by its ability to allocate individuals to the caste most appropriate for the requirements for growth, maintenance, and reproduction. In pairs of dependent lineages of Pogonomyrmex barbatus, the allocation of individuals to the queen or worker caste is constrained by genotype, a system known as genetic caste determination (GCD). In mature GCD colonies, interlineage female eggs develop into sterile workers, while intralineage eggs become reproductively capable queens. Although the population-level consequences of this system have been intensively studied, the proximate mechanisms for GCD remain unknown. To elucidate these mechanisms, we brought newly mated queens into the laboratory and allowed them to establish colonies, nearly half of which unexpectedly produced virgin queens only seven months after colony founding. We genotyped eggs, workers, and the virgin queens from these colonies. Our results showed that queens in young colonies produce both interlineage and intralineage eggs, demonstrating that queens of GCD colonies indiscriminately use sperm of at least two lineages to fertilize their eggs. Intralineage eggs were more frequent in colonies producing virgin queens. These findings suggest that intralineage eggs are predetermined to become queens and that workers may cull these eggs when colonies are not producing queens. Virgin queens produced by young GCD colonies were smaller than field-caught virgin queens, and often had developmental problems. Hence, they are probably nonfunctional and represent an intense resource drain for developing colonies, not a contribution to colony fitness.     

Kin conflict over caste determination in social Hymenoptera

We argue that caste determination, the process whereby females in the social Hymenoptera develop into either queens or workers, is subject to kin-selected conflict. Potential conflict arises because developing females are more closely related to their would-be offspring than to those of other females. Therefore, they may favour becoming queens contrary to the interests of other developing females and of existing queens and workers. We suggest two contexts leading to potential caste conflict. The first occurs when queens are reared in a reproductive phase following an ergonomic phase of worker production, while the second occurs when queens and workers are reared simultaneously. The first context assumes that workers' per capita contribution to colony survival and productivity falls with rising colony size. A critical feature influencing whether potential conflict is realized is the extent to which developing females can determine their own caste (“self-determination”). Self-determination is facilitated when female larvae control their own food intake and when queen-worker size dimorphism is low. We know of no strong evidence for actual conflict over caste fate arising in the first context. However, stingless bees and polygynous ants with excess queen-potential larvae that are either forced to develop as workers or are culled as adults demonstrate actual caste conflict in the second context. Caste conflict does not preclude caste regulation for “the good of the colony”, but such regulation is contingent on either the absence of potential conflict or on developing females losing control of their caste fate. Received: 22 March 1999 / Received in revised form: 15 June 1999 / Accepted: 19 June 1999     

Genetic influence on caste determination underlying the asexual queen succession system in a termite

The question of how reproductives and sterile workers differentiate within eusocial groups has long been a core issue in the study of social insects. Recent studies have shown that not only environmental factors but also genetic factors affect caste differentiation. In the termite Reticulitermes speratus, queens produce their replacements (neotenics) asexually but use normal sexual reproduction to produce other colony members. Here, we demonstrate a genetic influence on caste determination underlying the asexual queen succession system in this termite species. Thelytoky in termites is accomplished by automixis with terminal fusion, yielding almost completely homozygous offspring; thus, parthenogenetically and sexually produced offspring profoundly differ in heterozygosity. An analysis of the relationship between the reproductive dominance of female neotenics obtained from experimentally orphaned colonies and their genotypes at five microsatellite loci showed that homozygosity at two loci influenced the developmental priority and/or reproductive quality of neotenics. These results suggest the existence of a multi-locus system affecting the queen fecundity and explain why parthenogens have genetic priority to become neotenics in this termite species.     

Genetic influence on caste in the ant Camponotus consobrinus

Genetic influences on polyethism within social insect colonies are well known, suggesting that the determination of caste (soldiers and minor workers) may also be genetically mediated. The Australian sugar ant Camponotus consobrinus is suitable for such a study, having soldiers and minor workers that follow a complex allometry. Further, although most C. consobrinus colonies are monogynous, 13 of 42 surveyed using microsatellites were found to be polygynous. Thus, although a minority of colonies were polygynous, the great majority of queens live in polygynous colonies. From the 29 monogynous colonies studied, we inferred that the queens are monandrous. Ants from four polygynous colonies were assigned to families on the basis of microsatellite genotypes, after measurements had been taken of head width and scape length. These measurements reflect a complex allometry interpretable as soldier and minor worker growth curves with a large changeover zone. Genetic influence on caste determination was examined by testing for differences between families within colonies in the distribution of scape lengths, residuals from the overall colony allometric curve, and proportions of soldiers and minor workers (as determined by head width falling above or below the inflection point of the overall colony allometric curve). Families in all four colonies differed significantly in caste proportions and in head-width distributions, and three of the four colonies showed significant differences between families in residuals from the overall colony growth curve. Nested ANOVAs using head widths and scape-length residuals showed that when the effect of family is removed, intercolony differences in allometry are negligible. This evidence indicates genetic rather than environmental causes for the observed differences between families. We speculate that this variation may reflect some selective advantage to within-colony heterogeneity between families or that selective differences are few between a wide array of family growth patterns. Received: 16 June 1999 / Received after revision: 13 September 1999 / Accepted: 25 September 1999     

Adaptive significance of the age polyethism schedule in honeybee colonies

Summary The adaptive origins of the honeybee's age polyethism schedule were studied by testing whether the schedule for labor inside the nest reflects a compromise between efficiency in locating tasks and efficiency in performing tasks. I checked two predictions of this hypothesis: (1) at each age a worker handles a set of tasks (rather than one task), and (2) the elements of each age's task-set co-occur spatially in the nest (rather than being spatially segregated). Most observations match these predictions, once workers reach the age of 2 days. The unpredicted specialization of 0 to 2-day-old workers on the single task of cell cleaning may reflect an unusual ease in locating work sites for this particular task. There are 5 female castes in honeybee colonies: the queen (reproductive caste), plus 4 age subcastes among the workers (cell cleaning caste, broodnest caste, food storage caste, and forager caste).     

Darwin's special difficulty: the evolution of “neuter insects” and current theory

In the Origin of Species, Darwin discussed several challenges that worker insects presented to his theory of natural selection. Complex instincts such as building of combs of hexagonal cells were one problem and were explained by showing plausible intermediate stages. A more serious challenge was posed by the multiple worker castes seen in many ants. How could sterile individuals continue to evolve? A careful reading of the Origin suggests that Darwin was not primarily concerned by the evolution of worker sterility itself, which he considered a minor difficulty. Some modern commentaries on Darwin and insect workers seem to be cases of present interests interfering with the interpretation of the past. From a modern perspective, the evolution of a worker caste, and its corollary altruism, are evolutionary puzzles inasmuch as natural selection normally favors greater, not lesser, individual reproduction. These puzzles were resolved by Hamilton's theory of inclusive fitness. We now have a good functional understanding of how natural selection can cause both the origin of workers and their elaboration into greater levels of sterility and multiple morphological castes. Mechanistic understanding of morphological castes is also increasing via research into alternative developmental pathways. When the Origin was written, genetics did not exist and it would have been virtually impossible for Darwin to elaborate such ideas. However, the Origin probably addressed the main questions in the minds of Victorian readers in relation to insect workers. Darwin was prescient in having insights with close relationships to modern-day interests and the key principles involved, including kinship and benefits to the colony, even if these are not exact precursors to modern thinking.     

Social organization in the ant Pheidole dentata

Summary Caste theory states that the proportions of individuals in different demographic classes of an insect society should vary with environmental factors, and are adaptive because they enhance colony-level efficiency. We examined the proportions of workers in different age and size classes (temporal and physical castes) in whole colonies of the ant Pheidole dentata collected in two different habitats. Despite significant ecological differences between the habitats in competition, resource availability and predation, we found no differences in the physical and temporal caste structures of colonies. Also, there was no correlation between physical or temporal caste ratios and the reproductive output of colonies. Because of topography, distance between sites, and apparent low vagility of Pheidole alates, we assume that gene flow between the sites is inadequate to account for the observed similarities. Although age- and size-related patterns of division of labor were observed, similarities in the behavioral profiles (the sum of the relative contributions of each age cohort to the performance of tasks) in colonies having different age caste structures suggests that worker, flexibility may be more important than rigidly programmed age- and size-correlated patterns of task performance.     

Trophic aspects of caste determination in Halictus ligatus,a primitively eusocial sweat bee

Caste determination in primitively eusocial sweat bees is thought to be due to an interacting suite of factors, including size of the larval provision mass, time of year, and social context of the nest into which a young female emerges. Newly emerged gynes are significantly fatter than newly emerged workers, suggesting the existence of larval caste determination cues. Since photoperiod, temperature, and interactions with nestmates were unlikely to affect larval caste determination, we compared the sizes and contents of larval provision masses destined to produce either workers or gynes. Gyne-destined larvae consumed pollen masses that were larger and contained slightly more sugar than those of worker-destined larvae. We suggest that sugar content is one cue which prompts the development of fat reserves in gyne-destined females but not in worker-destined females. The amount of fat possessed by a newly emerged female influences her chances of successfully entering diapause shortly after emergence. Therefore, small, lean females may be more susceptible to behavioural control by queens and more likely to become workers, while large, fat females would be more likely to become gynes. Correspondence to: M.H. Richards     

Recognition of caste and mating status maintains monogyny in the ant Aphaenogaster senilis

In ants dispersing through colony fission, queens mate near their natal nest and found a new society with the help of workers. This allows potential future queens to challenge the mother queen’s reproductive monopoly. Conflicts might be resolved if the mated queen signals her presence and the workers control the developmental fate of the diploid larvae (whether they develop to worker or queen). In this study we sought to determine whether, in the fission-performing ant Aphaenogaster senilis, conflicts between queens for control of the colony are resolved by the resident queen signalling her mating status. Virgin queens were less effective than newly mated queens in inhibiting queen rearing. Moreover, potential challenger queens were recognized and heavily aggressed independent of mating status. Chemical analyses showed that mating status was associated with changes in cuticular hydrocarbon and poison gland composition, but not in Dufour’s gland composition. Cuticular dimethylalkanes were identified as potential constituents that signal both caste (present in queens only) and mating status (mated queens have higher amounts). We hypothesised that pheromone emission by virgin queens did not reach the threshold needed to fully inhibit larval development into queens but was sufficiently high to stimulate overt aggression by mated queens. These findings provide evidence for the complexity of chemical communication in social insects, in which a small number of signals may have a variety of effects, depending on the context.     

Behavioral differences between Pogonomyrmex rugosus and dependent lineage (H1/H2) harvester ants

The discovery of genetic caste determination (GCD) in populations of Pogonomyrmex harvester ants raises many questions about the evolution and persistence of such populations. The genetic caste determination arises from the existence of two distinct, but mutually dependent, genetic lineages within a population. Workers always develop from a combination of the two lineages, but their sister queens develop from within-lineage matings. Maintaining genetic caste determination appears to be costly because many queen-destined eggs are wasted when a colony is not in the reproductive stage, yet these populations appear to be widespread. We investigated whether inter-lineage workers have novel traits that give GCD colonies a selective advantage in certain environments. In particular, we compared ecologically relevant behavioral characteristics of inter-lineage workers in H-lineage colonies with co-occurring normal colonies of P. rugosus. First, we measured colony defensive response toward a simulated vertebrate predator. Second, we set up direct competitive foraging and recruitment experiments between dependent lineage and P. rugosus colonies. Last, we measured individual aggressive response to foreign inter-lineage and P. rugosus workers. We found that H1/H2 inter-lineage workers explored objects on the nest more thoroughly and responded much more aggressively to simulated predator disturbance than the P. rugosus colonies. In individual encounters, H1/H2 inter-lineage and P. rugosus workers were equally aggressive toward foreign ants, but both worker types could discriminate P. rugosus from inter-lineage intruders and were more aggressive toward ants of the alternate type to themselves. When competing directly for resources, however, P. rugosus colonies consistently dominated seed piles. In summary, H1/H2 GCD colonies show distinct behavioral differences, but there is no clear ecological advantage from the traits we examined.     

Reproductive conflicts and egg discrimination in a socially polymorphic ant

The ability to discriminate against competitors shapes cooperation and conflicts in all forms of social life. In insect societies, workers may detect and destroy eggs laid by other workers or by foreign queens, which can contribute to regulate reproductive conflicts among workers and queens. Variation in colony kin structure affects the magnitude of these conflicts and the diversity of cues used for discrimination, but the impact of the number of queens per colony on the ability of workers to discriminate between eggs of diverse origin has so far not been investigated. Here, we examined whether workers from the socially polymorphic ant Formica selysi distinguished eggs laid by nestmate workers from eggs laid by nestmate queens, as well as eggs laid by foreign queens from eggs laid by nestmate queens. Workers from single- and multiple-queen colonies discriminated worker-laid from queen-laid eggs, and eliminated the former. This suggests that workers collectively police each other in order to limit the colony-level costs of worker reproduction and not because of relatedness differences towards queens’ and workers’ sons. Workers from single-queen colonies discriminated eggs laid by foreign queens of the same social structure from eggs laid by nestmate queens. In contrast, workers from multiple-queen colonies did not make this distinction, possibly because cues on workers or eggs are more diverse. Overall, these data indicate that the ability of F. selysi workers to discriminate eggs is sufficient to restrain worker reproduction but does not permit discrimination between matrilines in multiple-queen colonies.     

Repeated geographic divergence in behavior: a case study employing phenotypic trajectory analyses

Environmental effects on behavior have long been a focus of behavioral ecologists. Among the important drivers of behavior is predation environment, which can include the presence/absence of predators, differences in resource availability, and variation in individual density. Environments with predators are often more ecologically complex and “risky” than those without predators. Populations from these environments are sometimes more active and explorative than populations from low-risk, less complex environments. To date, most comparative studies of behavior are limited to within-species comparisons of populations from divergent environments, but neglect comparisons between species following speciation, thus limiting our understanding of post-speciation behavioral evolution. Brachyrhaphis fishes provide an ideal system for studying correlations between divergent environments and behavior within and between species. Here, we test for differences in two behavioral traits—activity and exploration —between sister species Brachyrhaphis roseni and Brachyrhaphis terrabensis that occur in divergent predation environments. Species differed in activity and exploration, with higher activity and exploration levels in populations that co-occur with predators. Furthermore, we found drainage-by-species interactions, indicating that the nature of divergence varied geographically. Using the recently developed phenotypic trajectory analysis (PTA), we quantified this difference and found that, while the geographically isolated populations of sister species tended to evolve in parallel, the magnitude of divergence between species differed between drainages. Our results highlight the utility of PTA for multivariate behavioral data and corroborate past predictions that complex and risky environments are correlated with increased activity and exploration levels and that divergence continues post-speciation.     

Worker self-restraint and policing maintain the queen’s reproductive monopoly in a pseudomyrmecine ant

Division of reproductive labor in insect societies is often based on worker self-restraint and both queen and worker policing. Workers of many hitherto studied wasps, bees and ants do not lay eggs in the presence of a queen. However, it is presently unclear how far these observations in a few select clades can be generalized. We investigated if and how queens maintain a reproductive monopoly in colonies of the elongate twig ant, Pseudomyrmex gracilis, a member of the previously unstudied ant subfamily Pseudomyrmecinae. Colonies are usually headed by a single, singly mated queen (monogyny, monandry). Workers therefore would be more closely related to males produced by other workers (r?=?0.375) than to the sons of queens (r?=?0.25). Nevertheless, workers appear to refrain from laying male-destined eggs in the presence of the queen. In queenless conditions, workers form dominance hierarchies by antennal boxing, and only one or a few high-ranking individuals readily begin to lay eggs. When returned into a queenright colony, egg-laying workers are immediately bitten, stung and expelled or killed by other workers. While the composition of cuticular hydrocarbons clearly differed between castes, it less clearly reflected worker ovarian development. An association with worker ovarian development that would allow workers to monitor the reproductive status of nestmates could only be tentatively postulated for certain substances. Our study broadens our knowledge about reproductive conflict in social Hymenoptera and shows that worker sterility in the presence of a queen is more common in monogynous, monandrous ants than expected from relatedness alone.     

Division of labor in honeybees: form, function, and proximate mechanisms

Honeybees exhibit two patterns of organization of work. In the spring and summer, division of labor is used to maximize growth rate and resource accumulation, while during the winter, worker survivorship through the poor season is paramount, and bees become generalists. This work proposes new organismal and proximate level conceptual models for these phenomena. The first half of the paper presents a push–pull model for temporal polyethism. Members of the nursing caste are proposed to be pushed from their caste by the development of workers behind them in the temporal caste sequence, while middle-aged bees are pulled from their caste via interactions with the caste ahead of them. The model is, hence, an amalgamation of previous models, in particular, the social inhibition and foraging for work models. The second half of the paper presents a model for the proximate basis of temporal polyethism. Temporal castes exhibit specialized physiology and switch caste when it is adaptive at the colony level. The model proposes that caste-specific physiology is dependent on mutually reinforcing positive feedback mechanisms that lock a bee into a particular behavioral phase. Releasing mechanisms that relate colony level information are then hypothesized to disrupt particular components of the priming mechanisms to trigger endocrinological cascades that lead to the next temporal caste. Priming and releasing mechanisms for the nursing caste are mapped out that are consistent with current experimental results. Less information-rich, but plausible, mechanisms for the middle-aged and foraging castes are also presented.     

Phenotypic plasticity and “cultural transmission” of alternative social organizations in the fire ant Solenopsis invicta

Summary We investigated the process of sexual maturation in winged queens of the fire ant Solenopsis invicta, a species with two distinct forms of social organization. We found that queens of the monogynous social form (single reproductive queen per colony) differ little or not at all from queens of the polygynous form (multiple reproductive queens per colony) in weight and fat content when these are pupae or newly-eclosed adults. Furthermore, the size of a sclerotized region of the adult thorax, which is set during larval growth, does not differ between queens of the two forms. In contrast, winged queens of the two social forms differ dramatically in their physiological phenotypes once they have matured, with monogynous queens weighing more and having greater fat reserves than polygynous queens. A crossfostering experiment revealed that the different maturation processes of queens of the two forms are induced largely by the type of colony in which a queen matures (monogynous or polygynous) rather than being due to intrinsic genetic differences between the forms. However, genetic variation at a single locus does appear to play some role in determining physiological phenotype in queens of the polygynous form, providing an example of genotype-environment interaction in the expression of these physiological traits. Differences between the social forms in the mature phenotypes that are produced constrain the reproductive options of queens, so that the characteristic social organization of a colony is perpetuated by virtue of the social environment in which new queens are reared. Correspondence to: L. Keller