Neural networks as mechanisms to regulate division of labor

In social insects, workers perform a multitude of tasks, such as foraging, nest construction, and brood rearing, without central control of how work is allocated among individuals. It has been suggested that workers choose a task by responding to stimuli gathered from the environment. Response-threshold models assume that individuals in a colony vary in the stimulus intensity (response threshold) at which they begin to perform the corresponding task. Here we highlight the limitations of these models with respect to colony performance in task allocation. First, we show with analysis and quantitative simulations that the deterministic response-threshold model constrains the workers' behavioral flexibility under some stimulus conditions. Next, we show that the probabilistic response-threshold model fails to explain precise colony responses to varying stimuli. Both of these limitations would be detrimental to colony performance when dynamic and precise task allocation is needed. To address these problems, we propose extensions of the response-threshold model by adding variables that weigh stimuli. We test the extended response-threshold model in a foraging scenario and show in simulations that it results in an efficient task allocation. Finally, we show that response-threshold models can be formulated as artificial neural networks, which consequently provide a comprehensive framework for modeling task allocation in social insects.     

Social inhibition and the regulation of temporal polyethism in honey bees.

Honey bee division of labor is characterized by temporal polyethism, in which young workers remain in the hive and perform tasks there, whereas old workers perform more risky outside tasks, mainly foraging. We present a model of honey bee division of labor based on (1) an intrinsic process of behavioral development and (2) inhibition of development through social interactions among the workers in a colony. The model shows that these two processes can explain the main features of honey bee temporal polyethism: the correlation between age and task performance; the age at which a worker first forages and how this age varies among hives; the balanced allocation of workers to hive tasks and foraging; the recovery of a colony from demographic perturbations; and the differentiation of workers into different behavioral roles. The model provides a baseline picture of individual and colony behavior that can serve as the basis for studies of more fine-grained regulation of division of labor.     

Specialization Does Not Predict Individual Efficiency in an Ant

The ecological success of social insects is often attributed to an increase in efficiency achieved through division of labor between workers in a colony. Much research has therefore focused on the mechanism by which a division of labor is implemented, i.e., on how tasks are allocated to workers. However, the important assumption that specialists are indeed more efficient at their work than generalist individuals—the “Jack-of-all-trades is master of none” hypothesis—has rarely been tested. Here, I quantify worker efficiency, measured as work completed per time, in four different tasks in the ant Temnothorax albipennis: honey and protein foraging, collection of nest-building material, and brood transports in a colony emigration. I show that individual efficiency is not predicted by how specialized workers were on the respective task. Worker efficiency is also not consistently predicted by that worker''s overall activity or delay to begin the task. Even when only the worker''s rank relative to nestmates in the same colony was used, specialization did not predict efficiency in three out of the four tasks, and more specialized workers actually performed worse than others in the fourth task (collection of sand grains). I also show that the above relationships, as well as median individual efficiency, do not change with colony size. My results demonstrate that in an ant species without morphologically differentiated worker castes, workers may nevertheless differ in their ability to perform different tasks. Surprisingly, this variation is not utilized by the colony—worker allocation to tasks is unrelated to their ability to perform them. What, then, are the adaptive benefits of behavioral specialization, and why do workers choose tasks without regard for whether they can perform them well? We are still far from an understanding of the adaptive benefits of division of labor in social insects.     

Colony-size effects on task organization in the harvester ant Pogonomyrmex californicus

Colony size is a fundamental attribute of insect societies that appears to play an important role in their organization of work. In the harvester ant Pogonomyrmex californicus, division of labor increases with colony size during colony ontogeny and among unmanipulated colonies of the same age. However, the mechanism(s) integrating individual task specialization and colony size is unknown. To test whether the scaling of division of labor is an emergent epiphenomenon, as predicted by self-organizational models of task performance, we manipulated colony size in P. californicus and quantified short-term behavioral responses of individuals and colonies. Variation in colony size failed to elicit a change in division of labor, suggesting that colony-size effects on task specialization are mediated by slower developmental processes and/or correlates of colony size that were missing from our experiment. In contrast, the proportional allocation of workers to tasks shifted with colony size, suggesting that task needs or priorities depend, in part, on colony size alone. Finally, although task allocation was flexible, colony members differed consistently in task performance and spatial tendency across colony size treatments. Sources of interindividual behavioral variability include worker age and genotype (matriline).     

Worker connectivity: a review of the design of worker communication systems and their effects on task performance in insect societies

Within-group communication is a fundamental feature of animal societies. In order for animal groups to function as adaptive units, the members must share information such that group mates respond appropriately to each others’ behavior. One important function of social communication is to affect the allocation of tasks among group members. Theoretical and empirical findings on a diverse array of social insect taxa show that interactions among workers often play important roles in structuring division of labor. We review worker interactions that regulate division of labor in insect societies, which we refer to as worker connectivity. We present a framework for synthesizing and analyzing the study of worker connectivity. The widespread reliance on worker connectivity among eusocial insect taxa and the diversity of communicative mechanisms used to recruit workers suggest that the nature of worker interactions has evolved by natural selection. We suggest that colony-level selection acting on variation in task allocation has been an important force in the evolution of mechanisms for worker connectivity. We also propose that there are important links between individual worker cognition and task allocation at the colony level. Evolutionary changes in the cognitive aspects of worker responses may affect task allocation as much as changes in the communicative signals themselves. Received 9 December 2006; revised 18 May 2007; accepted 30 May 2007.     

Collective control without explicit coding: The case of communal nest excavation

The nests of social insects result from a succession of stimulus responses steps involving the environment, the workers, and the by-product of their activities (which modify their environment). In this way social insects can build without any reference to a blueprint. In this paper we explore the link between individual building behavior and the characteristics (form, size, location, etc.) of the structures produced. We show with a mathematical model (in the form of nonlinear differential equations) that social insects using behavioral mechanisms, which do not require an explicit measure of the nest and the colony size, can nevertheless effectively regulate, at the level of the colony, the size of the nest in response to changes in the size of the colony population. In addition, even though individual workers do not directly compare environmental characteristics, the colony can expand the nest preferentially in the most favorable zone. The models used show how such regulations and decision making can be a by-product of an amplifying communication between the builders and their work and how different patterns of building through time can be generated tuning the same basic rules.     

Worker connectivity: a simulation model of variation in worker communication and its effects on task performance

We develop a simulation model of worker connectivity to analyze how variation in worker communication can influence task performance. The model generates predictions about how colony demography, worker communicative behavior, and worker cognition will affect the rate of recruitment of workers to a new task. The model explores some mechanisms for modulating the recruitment of workers. Under the conditions of our model– probabilistic interactions that lower worker’s response thresholds to tasks– worker recruitment follows a logistic growth pattern. The rate of recruiting workers increases exponentially toward an inflection point when 50% of the available force has been activated, then decreases toward the upper asymptote (all workers recruited). Many relevant features of colony design and worker behavior, including group size, probability of interacting, and strength of interaction effects on receivers, show a positive but decelerating effect on the rate of worker recruitment. We also identify features of worker cognition that can influence task recruitment, focusing on the time course of worker’s memories about previous interactions. Both learning (e.g., sensitization) and forgetting about previous interactions can influence the rate of worker recruitment to a task. The model suggests that worker cognition may be shaped by natural selection on task performance at the colony level. Forgetting about interactions may be especially costly, because it leads to unpredictable patterns of worker recruitment. We also show that social inhibition, when coupled with excitatory interactions, can effectively modulate worker recruitment at the colony level. Received 9 December 2006; revised 23 May 2007; accepted 30 May 2007.     

Division of labour and colony efficiency in social insects: effects of interactions between genetic architecture, colony kin structure and rate of perturbations

The efficiency of social insect colonies critically depends on their ability to efficiently allocate workers to the various tasks which need to be performed. While numerous models have investigated the mechanisms allowing an efficient colony response to external changes in the environment and internal perturbations, little attention has been devoted to the genetic architecture underlying task specialization. We used artificial evolution to compare the performances of three simple genetic architectures underlying within-colony variation in response thresholds of workers to five tasks. In the 'deterministic mapping' system, the thresholds of individuals for each of the five tasks is strictly genetically determined. In the second genetic architecture ('probabilistic mapping'), the genes only influence the probability of engaging in one of the tasks. Finally, in the 'dynamic mapping' system, the propensity of workers to engage in one of the five tasks depends not only on their own genotype, but also on the behavioural phenotypes of other colony members. We found that the deterministic mapping system performed well only when colonies consisted of unrelated individuals and were not subjected to perturbations in task allocation. The probabilistic mapping system performed well for colonies of related and unrelated individuals when there were no perturbations. Finally, the dynamic mapping system performed well under all conditions and was much more efficient than the two other mapping systems when there were perturbations. Overall, our simulations reveal that the type of mapping between genotype and individual behaviour greatly influences the dynamics of task specialization and colony productivity. Our simulations also reveal complex interactions between the mode of mapping, level of within-colony relatedness and risk of colony perturbations.     

Disentangling environmental and heritable nestmate recognition cues in a carpenter ant

Discriminating between group members and strangers is a key feature of social life. Nestmate recognition is very effective in social insects and is manifested by aggression and rejection of alien individuals, which are prohibited to enter the nest. Nestmate recognition is based on the quantitative variation in cuticular hydrocarbons, which can include heritable cues from the workers, as well as acquired cues from the environment or queen-derived cues. We tracked the profile of six colonies of the ant Camponotus aethiops for a year under homogeneous laboratory conditions. We performed chemical and behavioral analyses. We show that nestmate recognition was not impaired by constant environment, even though cuticular hydrocarbon profiles changed over time and were slightly converging among colonies. Linear hydrocarbons increased over time, especially in queenless colonies, but appeared to have weak diagnostic power between colonies. The presence of a queen had little influence on nestmate discrimination abilities. Our results suggest that heritable cues of workers are the dominant factor influencing nestmate discrimination in these carpenter ants and highlight the importance of colony kin structure for the evolution of eusociality.     

Percent lipid is associated with body size but not task in the bumble bee <Emphasis Type="Italic">Bombus impatiens</Emphasis>

In some group-living organisms, labor is divided among individuals. This allocation to particular tasks is frequently stable and predicted by individual physiology. Social insects are excellent model organisms in which to investigate the interplay between physiology and individual behavior, as division of labor is an important feature within colonies, and individual physiology varies among the highly related individuals of the colony. Previous studies have investigated what factors are important in determining how likely an individual is, compared to nestmates, to perform certain tasks. One such task is foraging. Corpulence (i.e., percent lipid) has been shown to determine foraging propensity in honey bees and ants, with leaner individuals being more likely to be foragers. Is this a general trend across all social insects? Here we report data analyzing the individual physiology, specifically the percent lipid, of worker bumble bees (Bombus impatiens) from whom we also analyze behavioral task data. Bumble bees are also unusual among the social bees in that workers may vary widely in size. Surprisingly we find that, unlike other social insects, percent lipid is not associated with task propensity. Rather, body size closely predicts individual relative lipid stores, with smaller worker bees being allometrically fatter than larger worker bees.     

The influence of intraspecific competition on resource allocation during dependent colony foundation in a social insect

Organisms face a trade-off between investment in fewer, larger offspring, or more, smaller offspring. Most organisms can adjust investment through variation in the size and number of offspring in response to factors such as resource availability and competition. In some social animals, established colonies divide into groups of individuals that become autonomous, a process known as colony fission (also dependent colony foundation in social insects). Resource allocation under fission can be fine-tuned by adjusting the number of new groups (offspring number) and the number of individuals in each new group (offspring size). We assessed the influence of competition on resource allocation during fission in the ant Cataglyphis cursor, by allowing colonies to fission in experimental enclosures of high or low conspecific colony density. The pattern of colony fission was similar to that observed in the field: each fissioning colony produced a few new nests comprising a highly variable number of workers and a single queen, the old queen was often replaced, and new queens were produced in excess. The number of new nests produced depended on the available workforce in the parent colony but was not affected by differences in colony density. Comparison with data from fission under natural field conditions, however, indicates that colonies in enclosures produced fewer, larger new nests, suggesting that resource investment patterns during fission are indeed subject to extrinsic factors. The density of conspecific colonies in the immediate surroundings may be an unreliable estimate of competition intensity and other factors should be considered.     

Response thresholds alone cannot explain empirical patterns of division of labor in social insects

The effects of heterogeneity in group composition remain a major hurdle to our understanding of collective behavior across disciplines. In social insects, division of labor (DOL) is an emergent, colony-level trait thought to depend on colony composition. Theoretically, behavioral response threshold models have most commonly been employed to investigate the impact of heterogeneity on DOL. However, empirical studies that systematically test their predictions are lacking because they require control over colony composition and the ability to monitor individual behavior in groups, both of which are challenging. Here, we employ automated behavioral tracking in 120 colonies of the clonal raider ant with unparalleled control over genetic, morphological, and demographic composition. We find that each of these sources of variation in colony composition generates a distinct pattern of behavioral organization, ranging from the amplification to the dampening of inherent behavioral differences in heterogeneous colonies. Furthermore, larvae modulate interactions between adults, exacerbating the apparent complexity. Models based on threshold variation alone only partially recapitulate these empirical patterns. However, by incorporating the potential for variability in task efficiency among adults and task demand among larvae, we account for all the observed phenomena. Our findings highlight the significance of previously overlooked parameters pertaining to both larvae and workers, allow the formulation of theoretical predictions for increasing colony complexity, and suggest new avenues of empirical study.

This study uses automated tracking of clonal raider ants and mathematical modeling to reveal how previously overlooked traits of larvae and workers might shape social organization in heterogeneous ant colonies. By incorporating the potential for variability in task efficiency among adults and task demand among larvae, the authors were able to account for all empirically observed phenomena.     

Colony insularity through queen control on worker social motivation in ants

We investigated the relative contribution of the queen and workers to colony nestmate recognition cues and on colony insularity in the Carpenter ant Camponotus fellah. Workers were either individually isolated, preventing contact with both queen and workers (colonial deprived, CD), kept in queenless groups, allowing only worker-worker interactions (queen deprived, QD) or in queenright (QR) groups. Two weeks post-separation QD and QR workers were amicable towards each other but both rejected their CD nestmates, which suggests that the queen does not measurably influence the colony recognition cues. By contrast, aggression between QD and QR workers from the same original colony was apparent only after six months of separation. This clearly demonstrates the power of the Gestalt and indicates that the queen is not a dominant contributor to the nestmate recognition cues in this species. Aggression between nestmates was correlated with a greater hydrocarbon (HC) profile divergence for CD than for QD and QR workers, supporting the importance of worker-worker interactions in maintaining the colony Gestalt odour. While the queen does not significantly influence nestmate recognition cues, she does influence colony insularity since within 3 days QD (queenless for six months) workers from different colony origins merged to form a single queenless colony. By contrast, the corresponding QR colonies maintained their territoriality and did not merge. The originally divergent cuticular and postpharyngeal gland HC profiles became congruent following the merger. Therefore, while workers supply and blend the recognition signal, the queen affects worker-worker interaction by reducing social motivation and tolerance of alien conspecifics.     

Intra-Patriline Variability in the Performance of the Vibration Signal and Waggle Dance in the Honey Bee, Apis mellifera

We examined intra-patriline behavioral plasticity in communication behavior by generating lifetime behavioral profiles for the performance of the vibration signal and waggle dance in workers which were the progeny of three unrelated queens, each inseminated with the semen of a single, different drone. We found pronounced variability within each patriline for the tendency to produce each signal, the ontogeny of signal performance, and the persistence with which individual workers performed the signals throughout their lifetimes. Within each patriline, the number of workers that performed each signal and the distribution of onset ages for each signal were significantly different. In each patriline, workers of all ages could perform vibration signals; vibration signal production began 3–5 d before waggle dancing; and some workers began performing waggle dances at ages typically associated with precocious foraging. Most workers vibrated and waggled only 1–2 d during their lifetimes, although each patriline contained some workers that performed the signal persistently for up to 8 or 9 d. We also found marked variability in signal performance among the three worker lineages examined. Because the vibration signal and waggle dance influence task performance, variability in signaling behavior within and between subfamilies may help to organize information flow and collective labor in honey bee colonies. Inter-patriline variability may influence the total number of workers from different partrilines that perform the signals, whereas intra-patriline variability may further fine-tune signal performance and the allocation of labor to a given set of circumstances. Although intra-patriline behavioral variability is assumed to be widespread in the social insects, our study is the first to document the extent of this variability for honey bee communication signals.     

Ventilation response thresholds do not change with age or self-reinforcement in workers of the bumble bee Bombus impatiens

The response threshold model is a potential mechanism for task allocation in social insects, and it assumes that workers vary in the levels of task stimuli to which they respond. Furthermore, response thresholds of individual workers may change over time through self-reinforcement (experience), such that workers become more sensitive to task stimuli. However, in addition to self-reinforcement, aging is another process that occurs through time. Distinguishing whether response thresholds change within workers due to self-reinforcement or aging may give insight into the flexibility of this task allocation mechanism. Using a ventilation paradigm, we manipulated workers of Bombus impatiens to have either repeated or lack of exposures to increases in nest air temperature, thereby allowing us to manipulate experience and thus self-reinforcement. Nest air temperature was the task stimulus, and ventilation (fanning) was the behavioral response. We found that ventilation response thresholds do not decrease either with age or experience in workers of B. impatiens, contrary to what has been reported for B. terrestris workers (Weidenmüller, 2004). Instead, we found high levels of intra-individual variation in response thresholds. Our results also show that workers with lower average response thresholds respond to heating events with higher probability than those with higher ventilation thresholds. These results provide insight into the role of the response threshold framework for task allocation; we also discuss how response probabilities may play a role in task allocation among workers.     

The ant Lasius niger shows no relationship between task efficiency and body size variation among workers

In ants, workers of different sizes may perform various tasks, even in so-called monomorphic species with relatively low body size variation. However, it is unclear if the body size diversity of monomorphic workers correlates with task efficiency, especially in stressful contingencies. Here we tested if the body size variation of workers corresponds with its efficiency in transferring pupae. Transferring brood is a pre-set behavioral response to stress, e.g. suboptimal temperature. Here we applied a laboratory experiment simulating nest damage. The study was performed on the common garden ant (Lasius niger (Linnaeus, 1758)) – a species with no distinct worker subcastes. The efficiency of workers was measured as the latency of transferring pupae from a lit part of the experimental colony to a darkened part, while the body size diversity was expressed as the within-colony coefficient of variation in head width. We did not find any significant correlation between efficiency and body size variation. Summarizing the existing studies and the present results, we propose the hypothesis that the body size diversity of L. niger may have implications for workers’ division of labor but not for their task efficiency in a stressful contingency.     

Socially-responsive gene expression in male Drosophila melanogaster is influenced by the sex of the interacting partner

Behavior is influenced by an organism's genes and environment, including its interactions with same or opposite sex individuals. Drosophila melanogaster perform innate, yet socially modifiable, courtship behaviors that are sex specific and require rapid integration and response to multiple sensory cues. Furthermore, males must recognize and distinguish other males from female courtship objects. It is likely that perception, integration, and response to sex-specific cues is partially mediated by changes in gene expression. Reasoning that social interactions with members of either sex would impact gene expression, we compared expression profiles in heads of males that courted females, males that interacted with other males, or males that did not interact with another fly. Expression of 281 loci changes when males interact with females, whereas 505 changes occur in response to male-male interactions. Of these genes, 265 are responsive to encounters with either sex and 240 respond specifically to male-male interactions. Interestingly, 16 genes change expression only when a male courts a female, suggesting that these changes are a specific response to male-female courtship interactions. We supported our hypothesis that socially-responsive genes can function in behavior by showing that egghead (egh) expression, which increases during social interactions, is required for robust male-to-female courtship. We predict that analyzing additional socially-responsive genes will give us insight into genes and neural signaling pathways that influence reproductive and other behavioral interactions.     

Effect of temperature and social environment on worker size in the ant Temnothorax nylanderi

Warm temperatures decrease insect developmental time and body size. Social life could buffer external environmental variations, especially in large social groups, either through behavioral regulation and compensation or through specific nest architecture. Mean worker size and distribution of worker sizes within colonies are important parameters affecting colony productivity as worker size is linked to division of labor in insect societies. In this paper, we investigate the effect of stressful warm temperatures and the role of social environment (colony size and size of nestmate workers) on the mean size and size variation of laboratory-born workers in the small European ant Temnothorax nylanderi. To do so, we reared field-collected colonies under medium or warm temperature treatments after having marked the field-born workers and removed the brood except for 30 first instar larvae. Warm temperature resulted in the production of fewer workers and a higher adult mortality, confirming that this regime was stressful for the ants. T. nylanderi ants followed the temperature size rule observed in insects, with a decreased developmental time and mean size under warm condition. Social environment appeared to play an important role as we observed that (i) larger colonies buffered the effect of temperature better than smaller ones (ii) colonies with larger workers produced larger workers whatever the rearing temperature and (iii) the coefficient of variation of worker size was similar in the field and under medium laboratory temperature. This suggests that worker size variation is not primarily due to seasonal environmental fluctuations in the field. Finally, we observed a higher coefficient of variation of worker size under warm temperature. We propose that this results from a disruption of social regulation, i.e. the control of nestmate workers over developing larvae and adult worker size, under stressful conditions.     

Colony response to graded resource changes: an analytical model of the influence of genotype,environment, and dominance

Successful social groups must respond dynamically to environmental changes. However, a flexible group response requires the coordination of many individuals. Here we offer a static analytical model that integrates variation in environment-based cues for performance of a task with genetically and environmentally based variation in individual responses, and predicts the resultant colony behavior for that task. We also provide formulae for computing effective number of alleles in a haplo-diploid colony founded by any number of parents. Variable colony resources combined with variation among worker phenotypes generate known patterns of colony flexibility, allowing us to explicitly test how the number of loci, dominance/codominance, and the phenotype's environment influences group response. Our model indicates that the number of loci strongly influences colony behavior. For one or two loci, the proportion of workers foraging for pollen remain constant over vast increases in colony pollen stores, but then drops dramatically when the pollen stores increase past a specific threshold. As the number of loci controlling pollen foraging increases, graded increases in pollen stores result in a graded drop in the proportion of the worker population foraging for pollen. The effect of number of alleles is less strong, a result we discuss in light of the fact that a low number of effective alleles are expected in a colony. Comparisons of our model with empirical honey bee (Apis mellifera) data indicate that worker foraging response to pollen stores is driven by one or two loci, each with dominant allelic effects. The growing body of evidence that genotype has strong effects on task performance in social insect colonies, and the variation in within-colony genetic diversity across social insect taxa, make our model broadly applicable in explaining social group coordination.