Honey Bee Nest Thermoregulation: Diversity Promotes Stability

See allHide authors and affiliations

Science  16 Jul 2004:
Vol. 305, Issue 5682, pp. 402-404
DOI: 10.1126/science.1096340


A honey bee colony is characterized by high genetic diversity among its workers, generated by high levels of multiple mating by its queen. Few clear benefits of this genetic diversity are known. Here we show that brood nest temperatures in genetically diverse colonies (i.e., those sired by several males) tend to be more stable than in genetically uniform ones (i.e., those sired by one male). One reason this increased stability arises is because genetically determined diversity in workers' temperature response thresholds modulates the hive-ventilating behavior of individual workers, preventing excessive colony-level responses to temperature fluctuations.

In many human workplaces, tasks are allocated by a system of centralized control. Managers are aware of the needs of their organization, the abilities of each of their employees, and the available materials. On the basis of these various pieces of information, managers try to allocate tasks to their workers in a way that maximizes the profits of the company. In contrast to this hierarchical system of task allocation, workers in insect societies are not allocated tasks in any way. Rather, they engage in tasks on the basis of local information gained from their immediate nest mates and environment (1). Nonetheless, despite this highly distributed nature of organization, insect colonies respond extremely well to changes in task demand.

According to the response threshold model of task allocation (26), the probability that an individual worker will engage in a task depends on the level of the task stimulus and her threshold for that stimulus. Individuals with a low threshold will engage in a task with only a small task stimulus, whereas individuals with a high threshold require a large stimulus to engage in the task (2, 3, 710). When individuals of different task thresholds interact in a colony, task specialization among workers can result as an emergent property (11). Those individuals with a high task threshold will never perform certain tasks because their threshold for performing these tasks is never reached, whereas others with a low threshold will appear to be specialists for those tasks (12, 13).

In honey bees, a worker's response threshold is flexible over her life so that the tasks performed by workers vary with their age. Younger bees have a low threshold for tasks that are performed within the nest, whereas older bees have a low threshold for defensive and foraging tasks that occur outside the nest. This leads to a strongly age-based division of labor (14, 15). However, these age-influenced thresholds are mediated hormonally and can be partially reversed by high levels of colony need for a task, caused by alteration of a colony's age profile (16, 17).

Another important factor affecting the probability of task performance has a genetic origin (15, 18). Honey bee colonies are made up of a large number of patrilines (∼10 to 30), each patriline comprising highly related (coefficient of relatedness r = 0.75) supersisters sired by the same father. Differences among patrilines in the frequency of task performance demonstrate a genetic component to varying task thresholds among patrilines. Strong patrilineal differences have been demonstrated for a variety of important tasks such as foraging (9, 10). Here we demonstrate that patriline differences in task response threshold can have an improved colony-level outcome.

Honey bee colonies need to maintain their brood nest temperature between 32°C and 36°C, and optimally at 35°C, so that the brood develops normally (19). Workers regulate temperature by fanning hot air out of the nest when the temperature is perceived as being too high and clustering together and generating metabolic heat when the temperature is perceived to be too low (1921). Clearly, a graded rather than precipitous response is required, so that the colony does not constantly oscillate between heating and cooling responses. Does genetic variation among patrilines help colonies to produce an appropriate, graded response to temperature changes? Here we present three independent lines of evidence that in combination strongly suggest that genetic variance among patrilines within a honey bee colony is important in helping them to precisely maintain the optimal brood nest temperature.

In our first experiment, four single-patriline colonies were obtained by artificial insemination (22, 23), and these were precisely matched, in number of workers and amount of brood, with four multi-patriline (open-mated) colonies. These colonies were housed in two-comb metal hives and exposed to ambient temperatures in Sydney, Australia, for 1 week in each of August and September 2003, during late winter. The temperature of the central brood area of each colony was recorded at the brood comb surface every 5 min with thermochron ibuttons (Dallas Semiconductor, Texas). The colony temperature data for the low and high genetic diversity colonies were fitted to two linear models with colony and week as explanatory variables. The residual error mean square from these two analyses provides an estimate of the variance around the mean temperature after the removal of any variance attributable to individual colony and week means. The mean temperature maintained by uniform colonies across the 2 weeks (34.7°C) was slightly lower than in the diverse colonies (35.0°C) (we attribute this difference to calibration of the ibuttons). However, on average, the within-colony variance in temperatures maintained by the diverse colonies (σ2 = 0.047°C) was less than one-third of the within-colony variance of the uniform ones (σ2 = 0.165°C) and significantly different (F16196,16196 = 3.5, P < 0.001). The average hourly temperature of an exemplar pair of colonies is shown in Fig. 1.

Fig. 1.

Temperature variation in genetically diverse and uniform honey bee colonies. This graph shows the average hourly temperature for one representative pair of colonies in the first experimental week. Other colony pairs can be seen in Fig. S1.

We next assessed the ability of genetically uniform and diverse colonies to maintain the optimal brood nest temperature after we had raised the ambient temperature to 40°C. Each two-comb single-patriline colony was carefully matched, in number of workers and amount of brood, with a multi-patriline colony. Each pair of colonies was placed in an insulated room with access provided to the external environment. After ∼1 week, we heated the room to 40°C, held it there for 1 hour, and then measured the temperature inside each hive at the center of the brood nest, using a Dual Input Digital Thermometer (Dick Smith Electronics, Sydney). Three replicate experiments were conducted on each pair of matched colonies. The temperature data for the uniform and diverse colonies were each fitted to a linear model with colony and day as explanatory variables. The uniform colonies (σ2 = 0.83°C) had a significantly higher within-colony variance than the diverse colonies (σ2 = 0.22°C) (F603,603 = 3.83, P < 0.001).

Our third experiment shows a necessary condition for the task threshold model to be relevant to colony thermoregulation: Naturally occurring patrilines should vary markedly in their threshold for the task of fanning. We exposed two five-patriline colonies to increasing temperatures and collected fanning bees from the entrances. We then determined the paternity of the fanning workers by means of genetic markers (Fig. 2). As required by the task threshold model, the proportion of fanning workers from each patriline varied significantly as temperature was increased (likelihood ratio test; Colony A: G = 70.5, df = 28, P < 0.001; Colony B: G = 44.07, df = 24, P = 0.007). In both colonies tested, some patrilines (Fig. 2, A2, A3, and B3) fanned in much higher proportions than other patrilines for many or all of the experimental temperatures. This supports the response threshold model, as it suggests that these patrilines had lower than average thresholds for fanning.

Fig. 2.

Patrilines vary in their fanning response to changing ambient temperatures. The two five-patriline colonies studied each consisted of ∼5000 bees. We used five-patriline colonies to reduce the sample size required to produce adequate minimum expected values in a G test (32). Each colony was maintained in a two-frame observation hive in an insulated room in which the temperature could be controlled to ±1°C. Colonies were heated from 25°C to 40°C in 1°C steps. Fanning bees (50) were collected over each 2-degree interval from the entrance tube with forceps. A random sample of 50 bees was also taken from the colony after each experiment. To determine the patriline of all workers sampled, we extracted DNA using the Chelex method (33, 34). DNA was then amplified by polymerase chain reaction with the microsatellite primers A76 (35) and A113 (36) for colony 1 and A88 (36) and A113 for colony 2. Patrilines were then determined as outlined by Estoup et al. (35).

In both experimental colonies, there were also significant differences in the proportion of workers of each patriline in the fanning samples relative to the random samples at most experimental temperatures (6 temperatures out of 8 in colony A and 4 out of 7 in colony B, G tests, P < 0.05, df = 4). To test the possibility that these changes were caused by the time of day rather than by temperature, we conducted a control experiment using colony B in which ambient temperature was held at a constant 37°C. Here, time of day did not have a significant effect on the proportion of workers of each patriline fanning (G = 16.28, df = 12, P = 0.2).

The responses of different patrilines to changes in ambient temperature show two important phenomena. First, patrilines undoubtedly vary in their responses to changing temperature, a necessary condition for the task threshold model. Second, the proportion of fanning workers from different patrilines changes erratically with temperature. There are three likely reasons for the observed nonlinearity of patrilineal responses to environmental changes. First, a patriline's threshold for performing another thermoregulation task, such as water collection, may be lower than that for fanning and therefore draw members of that patriline away from the task of fanning. Second, the work of nest mates of other patrilines must change the stimulus to fan. Finally, at least some of the apparently random changes in patriline proportions are due to the way we have presented our data. Workers from any single patriline could in fact be fanning in steady numbers, rather than increasing or decreasing, but as the number of workers fanning from another patriline increases, the number from the first patriline appears to decrease proportionally. This artifact could only be overcome if it were possible to test the entire fanning population, rather than sampling a subset.

Why should advanced insect societies such as that of the honey bee rely on multiple mating and a lottery of paternal genotypes to ensure that their nests are homeostatic? Polyandry probably evolved in honey bees for reasons other than the task allocation system. Because of the sex determination system of hymenoptera (24), a queen that mates with a single male carrying the same sex allele as herself suffers a 50% loss of her diploid brood. Queens can reduce the probability of this occurring by mating with many males, and this seems to have been the primary cause of the evolution of polyandry in some eusocial insects (25, 26). We argue that, as a secondarily acquired phenomenon, genetic diversity in the stimulus level required for an individual to begin a task contributes to overall colony fitness by enhancing the task allocation system. We suggest that a genetically diverse colony can respond appropriately to a greater variety of environmental perturbations without overreacting. In contrast, colonies with low genetic diversity (only one or two patrilines) have a narrow range of thresholds among their workers, and this can lead to perturbations in colony homeostasis because too many workers are allocated to those tasks for which the colony's particular genotypes have a low task threshold (27, 28). Such colonies can experience large oscillations above and below the optimal colony-level phenotype.

Evolutionary theory (29, 30) suggests that traits related to fitness should exhibit low genetic variation, because selection should act to remove genetic variance from the population. However, in insect societies, selection acts at the level of the colony (31) to favor those that can most precisely regulate the internal conditions of the nest, including those with the ability to precisely regulate brood nest temperature over a broad range of ambient temperatures. Without direct selection for the threshold temperature that will initiate thermoregulatory behaviors, a range of thresholds can evolve. This phenomenon is manifest as genetic variance in the propensity to perform thermoregulatory tasks such as fanning and in the more precise regulation of brood nest temperature by genetically diverse colonies compared with genetically uniform colonies, especially when stressed.

Our study has shown how random genetically determined differences in task threshold can enhance the stability of a self-organized biological system. We suggest that most aspects of colonial life would also be enhanced by variance in task threshold among the worker population.

Supporting Online Material

Fig. S1

References and Notes

View Abstract


Navigate This Article