Molecular Determinants of Scouting Behavior in Honey Bees

See allHide authors and affiliations

Science  09 Mar 2012:
Vol. 335, Issue 6073, pp. 1225-1228
DOI: 10.1126/science.1213962

Bee Adventurous

Individuals differ in their behavior, sometimes in consistent ways. For example, some people may seek out new experiences, while others prefer to stick with what they know. This is true in bees as well, where some workers take on the dangerous, novelty-seeking task of scouting more often than others. Liang et al. (p. 1225) found that bees that display such scouting behavior not only tend to scout in multiple contexts (both foraging and searching for nests) but also show differences in gene expression in their brains. Experimental manipulation of gene expression predictably changed scouting behavior. The molecular underpinnings of bee scouting behavior appear to be similar to those associated with novelty-seeking in vertebrate species, including humans.


Little is known about the molecular basis of differences in behavior among individuals. Here we report consistent novelty-seeking behavior, across different contexts, among honey bees in their tendency to scout for food sources and nest sites, and we reveal some of the molecular underpinnings of this behavior relative to foragers that do not scout. Food scouts showed extensive differences in brain gene expression relative to other foragers, including differences related to catecholamine, glutamate, and γ-aminobutyric acid signaling. Octopamine and glutamate treatments increased the likelihood of scouting, whereas dopamine antagonist treatment decreased it. These findings demonstrate intriguing similarities in human and insect novelty seeking and suggest that this trait, which presumably evolved independently in these two lineages, may be subserved by conserved molecular components.

An important challenge in behavioral biology is to elucidate the molecular basis of individual differences in behavior. Scouting behavior in the honey bee, Apis mellifera, provides an excellent opportunity to explore this issue for two reasons. First, there are striking individual differences in this behavior—some bees act as scouts and others never do so. Second, scouting is performed in two distinct contexts: scouting for new food sources or new nest sites, which suggests an underlying tendency to seek something new. Novelty-seeking behavior has been studied in vertebrates, including humans (1, 2), but not in insects.

Food scouts, who make up 5 to 25% of a colony’s foraging force, search independently for new food sources and continue to do so even when plentiful sources have been found (35). Non-scouts do not search for novel food sources and instead rely on information from scouts (communicated via “dance language”) to guide their foraging. By constantly discovering new flower patches, food scouts help ensure a high influx of food to their colony, despite the ephemeral nature of each patch (5).

Nest scouts make up <5% of the population of a swarm, which is a fragment of a colony that has left its natal nest to start a new colony. Nest scouts search independently for potential nesting cavities and collectively choose the best one, whereas non-scout swarm members rely on information from scouts to guide them to their new home (6). Nest scouting also is a crucial behavior; a colony’s survival depends on its nest scouts finding suitably protective living quarters.

To determine the consistency of novelty seeking in individual bees across the two behavioral contexts, we determined whether nest scouts are prone to also act as food scouts. We identified and marked nest scouts in both artificial and natural swarms (6). We then identified food scouts with the standard “hive-moving” assay (5, 7), after installing each swarm in a beehive and moving it at night (when bees don’t forage) to a new location outside the bees’ original home range. This assay identifies food scouts as the first bees to return to their hive in the morning; under these circumstances, each successful forager must have located a food source on her own. There was a robust tendency of nest scouts to seek novel resources across different contexts, but it did not translate into every nest scout showing food-scouting behavior. In nine trials involving eight different colonies over 2 years, nest scouts were on average 3.4 times more likely to become food scouts than were bees that did not search for nest sites during swarming (Fig. 1A). These results demonstrate that some bees show consistent novelty seeking across diverse behavioral contexts.

Fig. 1

(A) Consistent novelty-seeking behavior across different contexts. Nest scouts were significantly more likely to later act as food scouts than were non-scout swarm members. The graph shows the probabilities of food scouting for nine trials: four natural swarms and five artificial swarms, with eight different colonies (Fisher’s exact test, 2-tailed test; *P < 0.05, **P < 0.01, ***P < 0.001), and the overall mean probabilities [least-square means and standard errors; mixed-model analysis of variance (ANOVA), 2-tailed test]. (B) Feeder-discovery assay for identifying food scouts. Additional details are in the text and supporting online material.

To explore the molecular basis of novelty seeking in bees, we developed a behavioral assay for food scouts (Fig. 1B) that tests novelty seeking more strongly than did previous scout assays (3, 5, 7). A large screened outdoor enclosure provided experimental control of food sources under otherwise naturalistic conditions. Foragers from a glass-walled observation hive were trained to a training feeder that initially was the only food source available to them. After 2 to 3 days of training, a novel feeder with different visual and odor cues was placed at another location in the enclosure. The foraging bees thus had two possible food sources, familiar and novel; some bees discovered the novel feeder and switched to it. This procedure was repeated on several consecutive days, and each time the novel feeder was given new visual and odor cues and placed in a new location. Only bees that switched to two or more different novel feeders, after being seen at least once at the training feeder, were collected as scouts. These rigorous criteria minimized the possibility of identifying scouts on the basis of an accidental discovery of a novel feeder. The proportion of scout bees identified with this assay (31.2, ± 9.7% SD, n = 182 bees, six trials) is roughly consistent with what has been observed under more natural conditions (35), suggesting that accidental discoveries of novel feeders were not a major source of error. Bees that met our criteria for identifying food scouts were collected to compare their brain gene expression with that of control non-scouts (foragers that were never observed to switch to a novel feeder).

Whole-genome microarray analysis revealed a large neurogenomic signature for scouting behavior in the bee brain. Sixteen percent (1219 out of 7539) of the transcripts on the microarray showed significant (false discovery rate <0.05) differences in mRNA abundance between scouts and non-scouts (table S3, A and B, and table S4). Among the differentially expressed genes were several related to catecholamine, glutamate, and γ-aminobutyric acid (GABA) signaling, which are involved in regulating novelty seeking and reward in vertebrates (1, 2, 8). For example, the down-regulation of a dopamine receptor gene in honey bee scouts parallels results for a similar gene in individual mammals that are prone to novelty seeking (9). These signaling systems also are implicated in personality differences between humans that are related to novelty seeking (10, 11).

Quantitative reverse-transcriptase polymerase chain reaction analysis confirmed the microrarray results for five genes related to catecholamine, glutamate, and GABA signaling (Fig. 2A and fig. S1, A and B): D1-type dopamine receptor DopR1, glutamate transporters Eaat-2 and Vglut, AMPA-type glutamate receptor Glu-RI, and GABA transporter Gat-a. Three additional catecholamine receptor genes also were differentially expressed but were undetected in microarray analysis: DopR2 (D1-type) (12), Octβ2R (β-adrenergic type octopamine receptor), and OctR1 (α-adrenergic type) (13) (fig. S1B and tables S1 and S2).

Fig. 2

Transcriptomic analyses of individual differences in novelty-seeking between food scouts (S) and non-scouts (NS) (n = 20 bees per group). (A) Selected microarray results highlight differences in brain expression for 10 dopamine, octopamine, glutamate, or GABA signaling genes related to novelty seeking, motivation, and reward in vertebrates. DopR2 and OctR1 did not show significant differences in expression (in the latter case, probably because of very low expression levels). GABA transporter 1A gene (Gat-a) expression was one of the best correlates of scouting behavior (permutation t test, P < 0.05). (B) Results of LDA for genes shown in (A) demonstrate clear separation between most scouts and non-scouts based on differences in brain gene expression (standardized expression values: mean = 0, SD = 1). This plot of LD1 versus LD2 accounted for 82% of the variation in brain gene expression across scouts and non-scouts (n = 20 bees per group). S1, S2, S3 and N1, N2, N3: scouts and non-scouts, respectively, from three different colonies.

Linear discriminant analysis (LDA) showed a strong separation between scouts and non-scouts based on the expression values for 10 neural signaling genes related to catecholamine, glutamate, and GABA signaling (Fig. 2B). In addition, we used these 10 genes to show that scouts identified with either the new feeder–discovery assay or the hive-moving assay showed strong similarities in brain gene expression to each other (fig. S2).

The association between scouting and catecholamine, glutamate, and GABA signaling pathways could reflect effects of this behavior on brain gene expression or effects of individual differences in these pathways on scouting, or both. We used the transcriptomic results as the basis for designing experiments to test causal relationships, hypothesizing that neurochemical treatment would influence scouting behavior. We tested this hypothesis with the hive-moving assay, because it results in rapid identification of numerous scouts. We collected non-scouts and provided them with a chronic (25 to 30 hours) oral neurochemical treatment (as specified in the next paragraph) in cages (20 bees per cage) in their hive before moving it overnight to a location outside the colony’s original home range.

Behavioral observations the following morning (14 hours after stopping the treatment) revealed that glutamate [monosodium glutamate (MSG)] caused a significant increase in scouting (Fig. 3A), whereas the vesicular glutamate transport blocker Chicago Sky Blue significantly attenuated the MSG effects (Fig. 3B). Octopamine caused a weaker, but still significant, increase in scouting (Fig. 3A). These results are consistent with predictions based on microarray analysis. In contrast, dopamine antagonists caused a significant decrease in scouting (Fig. 3C), which was contrary to microarray-based prediction. Effects were not seen in all trials (figs. S3 to S5), suggesting that factors such as food availability, colony conditions, worker genotype, or other unknown variables also affect the probability of becoming a scout. The treatments did not cause excess mortality (table S6), aberrant locomotion, hyperactivity, or a general increase in foraging activity (fig. S6), and they were dose-dependent (fig. S7), which suggests that there were specific treatment effects on scouting behavior. GABA or a GABA receptor agonist (TACA) did not affect the probability of scouting (fig. S8), so the role of this neurotransmitter in bee scouting remains unclear.

Fig. 3

Glutamate or octopamine treatment increased the probability of scouting, whereas dopamine antagonist treatment decreased it (*P < 0.05, ***P < 0.0001). (A) Oral administration of MSG to non-scouts in sugar syrup (20 mg/ml) caused a significant effect in 7 out of 12 trials (with 11 colonies) over 2 years, an overall 73% increase in scouting probability as compared to sucrose-fed–only control bees (P < 0.0001, mixed-model ANOVA, 2-tailed test). Octopamine (OA) treatment (4 mg/ml) caused a significant effect in 3 out of 10 trials (in nine colonies) over 2 years, an overall 37% increase in scouting probability (P < 0.05). Statistical tests were performed on square root–transformed data; the graph represents the untransformed mean ± SE of 12 trials for MSG (with 11 colonies) and 10 trials for octopamine (with 9 colonies); results of individual trials are shown in figs. S3 and S4. (B) The glutamate vesicular transporter blocker Chicago Sky Blue (CSB) (4 mg/ml) blocked the effect of MSG on scouting (P < 0.05, least-square mean ± SE for four previously MSG-responsive colonies; results of individual trials are shown in fig. S3). (C) Non-scout foragers treated with dopamine antagonists (DAA) (either the D1-receptor antagonist SCH-23390, the “pan-receptor” antagonist Flupenthixol, or both) showed an overall 44% decrease in scouting probability in seven trials over three colonies (P < 0.05, the graph represents least-square mean ± estimated error; mixed-model ANOVA, 2-tailed test; results of individual trials are shown in fig. S5). The probability of scouting was calculated from the proportion of foragers in each treatment group that exhibited scouting behavior, based on a precise count of foragers when releasing them from treatment cages.

Multiple neurotransmitter systems appear to be involved in the regulation of scouting in honey bees, but it is not known how they interact at the circuit level. Glutamatergic and dopaminergic neurons are both found in the vertical lobes of the mushroom bodies, a part of the insect brain involved in reward learning (14, 15). DopR1 and Eaat-2 gene expression is colocalized to the same type of interneurons that provide sensory input into these lobes (16, 17). These findings, together with our own, suggest the vertical lobes of the mushroom bodies as one possible neuroanatomical locus for novelty-seeking behavior in honey bees, although other brain regions are probably involved as well.

Our results demonstrate intriguing parallels between honey bees and humans in novelty-seeking behavior. Although the molecular mechanisms that produce this behavioral variation are similar, it is unknown whether both species inherited them from a common ancestor or evolved them independently. Given the phylogenetic separation of bees and humans, we believe it is likely that these mechanisms represent part of a basic tool kit that has been used repeatedly in the evolution of behavior. Further support for this view comes from the finding that individual differences in food-searching behavior in nematodes (Caenorhabditis elegans) are caused, in part, by noncoding polymorphisms in tyramine receptor 3, which encodes a receptor for a catecholamine closely related to octopamine and dopamine (18).

It is common to look to animal models to generate insights that may be applicable to human behavior. Our findings highlight the potential of the converse—using insights from human research to further elucidate the molecular basis of animal behavior. Animal studies, informed by inferences from human research, might in turn help identify evolutionarily conserved molecular mechanisms underlying consistent differences in various behaviors among humans, thus helping us better understand how and why these behavioral differences exist.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S8

Tables S1 to S6


References and Notes

Acknowledgments: Special thanks to M. K. Carr-Markell and J. Recchia-Rife for extensive help in the field. We also thank the following: C. Nye and K. Pruiett (bee management); A. Brockmann, P. Date, J. Dotterer, L. Felley, M. Girard, S. Kantarovich, H. S. Pollock, and M. Wray (field assistance); T. Newman (molecular studies); S. Aref and A. Toth (statistics); E. Hadley (graphics); and A. M. Bell, D. F. Clayton, R. C. Fuller, J. S. Rhodes, C. W. Whitfield, and members of the Robinson laboratory (review of the manuscript). Supported by NSF Frontiers in Biological Research grant EF 0425852 (B. L. Schatz, PI, BeeSpace Project); NIH Director’s Pioneer Award 1DP1OD006416 (G.E.R); and the Illinois Sociogenomics Initiative (G.E.R). Microarray data meet Minimum Information About Microarray Experiment (MIAME) standards and are available at ArrayExpress (,#E-MTAB-491). Z.S.L. and G.E.R. conceived the project, designed the experiments and wrote the paper; Z.S.L. performed sample collection, molecular and field experiments, and analyses; T.N. and S.L.R.-Z. performed microarray experiments and statistical analyses, respectively; and H.R.M. and T.D.S. contributed to protocol development and sample collection and co-wrote the paper.
View Abstract

Navigate This Article