Genotypic constraints on plasticity for corpse removal in honey bee colonies

Anim. Behav., 1995, 49, 867–876

Genotypic constraints on plasticity for corpse removal in honey bee colonies GENE E. ROBINSON* & ROBERT E. PAGE, J† *Department of Entomology, University of Illinois, Urbana, IL 61801, U.S.A. †Department of Entomology, University of California, Davis, CA 95616, U.S.A. (Received 16 August 1993; initial acceptance 12 October 1993; final acceptance 28 February 1994; MS. number: 6783)

Abstract. The hypothesis that plasticity in honey bee, Apis mellifera, division of labour can be influenced by genotypic differences in worker behaviour was tested in colonies with electrophoretically distinct subfamilies. Undertaking behaviour (removal of dead bees from the nest) was studied because it is possible to vary the level of corpse-removal stimuli in a precise way. Subfamilies of workers with low thresholds of response to corpses were predicted to be overrepresented as undertakers relative to subfamilies with higher thresholds, and this difference was predicted to be reduced under higher stimulus conditions (more corpses). Allozyme analyses revealed that the genotypic composition of the undertaker group that responded to a low stimulus was significantly different from the composition of the whole colony. Contrary to predictions, the genotypic composition of the undertaker group that responded to a high stimulus was just as different from the composition of the whole colony. A second experiment examined the effects of undertaker depletion (about 2–3% of each colony’s population): there were significant decreases in rates of corpse removal for several days following the removal of undertakers. These results suggest that strong genetic influences on the likelihood that an individual worker will perform a particular task, such as undertaking, may constrain a colony’s ability to respond to changing conditions.

Division of labour among workers in insect societies typically is associated with differences in age, morphology (only for some ants and most termites), and individual differences in task specialization that are independent of worker age and morphology (Wilson 1971; Oster & Wilson 1978; Jeanne 1987). In addition to a structured division of labour, colonies respond to changing internal and external conditions by adjusting the ratios of individual workers engaged in the various tasks (reviewed by Calabi 1988; Gordon 1989a; Robinson 1992). This is accomplished by four behavioural responses of workers: (1) ontogenetic changes that result in the performance of tasks outside a normal age-related repertoire; (2) task-switching within a normal age-related repertoire; (3) increased activity levels within a normal age-related repertoire; and (4) changes in the proportion of individuals working. Plasticity in division of labour contributes to the reproductive success of a colony by enabling it to continue to grow, develop, and ultimately produce a new generation of reproductive males and females despite changing colony conditions. 0003–3472/95/040867+10 $08.00/0

?

Robinson & Page (1989a) proposed that plasticity in division of labour may be, in part, a consequence of genetic variation for taskswitching. Under ‘normal’ conditions, a job will be performed by workers with the lowest response thresholds. These workers may represent a distinct subset of the appropriately aged individuals within a colony, in both a behavioural and a genetic sense. If the need for this activity increases due to changes in colony conditions, and there is a concomitant rise in the levels of associated stimuli, then more and more of a colony’s workers whose genotypes result in relatively higher response thresholds will perform it. Moderate, transient increases in the requirement for a particular task would elicit a graded colony response based on differential recruitment among genotypes, as more individuals shift from one job to another within a normal age-related repertoire. In this paper we test the predictions of this ‘genotypic threshold’ model for honey bees. There have been numerous reports of genetic influences on division of labour in honey bee, Apis mellifera, colonies (reviewed by Page & Robinson 1995 The Association for the Study of Animal Behaviour

867

868

Animal Behaviour, 49, 4

1991; see also Calderone & Page 1991; Oldroyd et al. 1993; for comparable reports on ants, see Stuart & Page 1991; Snyder 1992). However, the effects of worker genotypic diversity on colony fitness are unclear (Oldroyd et al. 1992). This information is necessary both to understand the mechanisms that integrate worker behaviour into coordinated colony patterns (see Page & Robinson 1991; Robinson 1992) and to gain insights into the evolution of polyandry, a central question in behavioural ecology (see Crozier & Page 1985; Sherman et al. 1988). Honey bee queens are extremely polyandrous, mating on average with about 7–17 different males (Adams et al. 1977; reviewed by Page 1986), which results in colonies composed of subfamilies. Members of a subfamily share both parents while members of different subfamilies share just the queen mother. One of the hypotheses proposed by Crozier & Page (1985) for the evolution of polyandry in social insects is that polyandry is favoured because it results in divergent worker genotypes that increase the range of environmental conditions a colony can tolerate. One prediction consistent with this hypothesis is that colonies with more genotypic diversity may be better able to respond to changing environments than genotypically narrow colonies. The genotypic threshold model proposes one way in which intra-colony genotypic diversity can contribute to plasticity in division of labour. It assumes that polyandry results in subfamilies with different genetically determined response thresholds to task-related stimuli, and predicts that graded colony responses to environmental change are based, in part, on subfamily differences in response thresholds. We test the predictions of the genotypic threshold model for corpse-removal behaviour (‘undertaking’), for two reasons. First, a genetic component to the performance of this task has already been shown. Using colonies composed of electrophoretically distinguishable subfamilies, Robinson & Page (1988) demonstrated that workers from some subfamilies are more likely to act as undertakers than workers from other subfamilies. Second, it is possible to vary the level of corpseremoving stimuli in a precise way, by varying the number of corpses added to a colony. In experiment 1, we determined whether undertakers that respond to a high stimulus for corpse removal constitute a less genotypically specialized group relative to those that respond to a low

stimulus. This was accomplished by sampling undertakers from colonies with allozyme-marked subfamilies. In experiment 2, we tested a hypothesis that was suggested by the results of experiment 1, that depletion of a colony’s undertakers causes a persistent decrease in colony corpseremoval rate. This was accomplished by comparing corpse-removal rates in paired, depleted and non-depleted, colonies.

EFFECTS OF STIMULUS LEVEL ON GENOTYPIC COMPOSITION OF UNDERTAKER GROUP Methods Allozyme markers Allozymes of Malate dehydrogenase-1 (Mdh) (Contel et al. 1977) that occur naturally in populations of honey bees (Page & Metcalf 1982, 1988) were used as genetic markers. They are reliable markers because their electrophoretic mobilities are known for all honey bee life stages and their alleles segregate in Mendelian fashion (Contel et al. 1977; Nunamaker & Wilson 1980; Del Lama et al. 1985). In addition, allozyme markers allow undertakers to be sampled blind with respect to subfamily affiliation. Allozymes were analysed with polyacrylamide gel electrophoresis (Robinson & Page 1988, 1989b). Experimental colonies Six virgin queens were each instrumentally inseminated (Laidlaw 1977) with the semen of three unrelated drones. Each drone carried a different allozyme of Mdh, designated ‘slow’ (S), ‘medium’ (M), and ‘fast’ (F), based on electrophoretic mobility. Semen from each drone trio was pooled, diluted and homogenized to help stabilize the relative frequencies of different subfamilies over time (Kaftanoglu & Peng 1980; Moritz 1983). All inseminated queens were homozygous at the Mdh locus (SS: colonies 4562, 4563, 4564 and 4565; FF: 4560 and 4561) and produced three electrophoretically distinct groups of worker progeny belonging to three subfamilies. Colonies had populations of 15 000–20 000 workers in one Langstroth hive body. They were typical of current North American populations of A. mellifera (a mix of predominantly

Robinson & Page: Genotypic constraints on honey bee plasticity European subspecies; Phillips 1915; Pellett 1938), and were maintained according to standard techniques. Sampling techniques We induced undertaking by adding corpses to each colony. Corpses were obtained by freezekilling bees from another colony. Corpses were stored in a freezer ("20)C) and thawed prior to use. We identified undertakers at the hive entrance (following Visscher 1983) and collected them with a portable vacuum cleaner (Robinson & Page 1988). We introduced 15 corpses every 15 min (to the bottom rear of each hive) and collected the first 50 undertakers. These were the undertakers that responded to a ‘low stimulus’ for corpse removal. We then introduced 1000 corpses at one time and collected all undertakers seen over a period of 2–3 h. These were the undertakers that responded to a ‘high stimulus’ for corpse removal. All undertakers collected were frozen for electrophoresis. All remaining corpses were removed from each hive at the conclusion of the high stimulus sampling period. We collected samples of newly emerged adult bees (N=40) weekly for 6 weeks to estimate the subfamily frequencies in each experimental colony. They were identified visually as teneral bees that were not able to fly. We obtained these samples by shaking bees off combs into a plastic pan lined with petroleum jelly (to prevent non-flying bees from walking out), applying smoke from a ‘bee smoker’ and shaking the pan to induce flight, and collecting teneral bees that remained in the pan. We assumed that these samples would reflect the genotypic composition of the colony from which they were taken because newly emerged adult bees do not yet exhibit specialized patterns of labour. We began collecting newly emerged adults 2 weeks before sampling undertakers; because undertakers are usually approximately 2–3 weeks old (Sakagami 1953; Huang et al. 1994), these samples presumably also reflected the genotypic composition of the entire worker pool from which we obtained the undertakers. Analyses of these samples have been presented elsewhere (see Table 1 in Breed et al. 1990). The genetic structure of three colonies was stable during the period encompassed by this study (4560, 4564 and 4565) while the other three showed significant fluctuations in subfamily frequencies. To compare the subfamily frequencies in samples

869

of undertakers with estimated subfamily frequencies for the whole colony, we pooled the samples of newly emerged bees, resulting in a single ‘colony sample’ for each colony. To determine whether there were subfamily differences in mortality, we marked 500 newly emerged bees, identified as above, with spots of paint on the thorax during the first two weekly collections of colony samples. We collected two samples of marked workers (N=40) from each colony 6 weeks after they were marked. We compared subfamily frequencies in these two samples with the newly emerging bees collected on weeks 1 and 2 to test for differential mortality between subfamilies. If such differential mortality existed, the colony samples would have been less useful as estimators of subfamily frequencies. Newly emerged bees and marked bees were collected only on days that undertakers were not sampled. No marked bees were observed as undertakers, which is not surprising given the relatively small number of bees marked in these colonies and the rarity of undertaking behaviour (Visscher 1983). Statistical analyses There were two trials for each colony (12 trials total) at 7–10-day intervals, under both lowstimulus and high-stimulus conditions. We removed all corpses remaining in the hive at the conclusion of each high-stimulus trial. Although we collected all undertakers during the 2–3-h highstimulus trials, we analysed only the last 40 undertakers collected. These samples were the ones most likely to differ from the samples collected at lowstimulus conditions, because of the hypothesized effects of the entire depletion. For undertaker samples at both stimulus levels, N=40 except where noted on Fig. 1. In all cases but three, variation in sample sizes was due to the exclusion of bees with anomalous allozyme phenotypes that apparently ‘drifted’ into experimental colonies (the exceptions: part of the trial 2 low-stimulus sample for colony 4565 was lost and only 19 and 10 undertakers were collected from colony 4564 during low-stimulus trials 1 and 2, respectively). Results The subfamily composition of the undertaker group that responded to the low stimulus for

870

Animal Behaviour, 49, 4

corpse removal differed significantly from the colony sample in seven of 12 trials (Fig. 1). Contrary to the predictions of the genotypic threshold model, the subfamily composition of the undertaker group that responded to the high stimulus also differed significantly from the colony sample, in seven of 12 trials. This result indicates that the genotypic composition of the undertaker group that responded to the high stimulus was just as different from the composition of the whole colony as was the group that responded to the low stimulus. Results of two other comparisons also demonstrate the consistency of the genotypic response to corpse removal in this experiment. No differences in subfamily frequencies were detected between ‘low-stimulus’ and ‘high-stimulus’ undertakers in 12 out of 12 trials. Also, there were almost no inter-trial differences in the subfamily composition of samples of undertakers collected at either low- or high-stimulus levels (except for the low-stimulus samples from colony 4565 and the high-stimulus samples from colony 4562). There was negligible evidence of differential mortality between subfamilies that could potentially complicate the interpretation of these results (data shown in Breed et al. 1990). In only two of 12 comparisons (six colonies, two comparisons each) did the subfamily frequencies in samples of newly emerged bees differ significantly (P<0·05) from the subfamily frequencies in samples of the marked 6-week-old bees (colonies 4561 and 4565, week 1). Significant (P<0·05) differences in the subfamily composition of the two samples of marked workers were found in only one out of six colonies (colony 4561). The two trials of this experiment were conducted 7–10 days apart because it was assumed that this interval would be sufficient to ensure that the removal of some of a colony’s undertakers during trial 1 would have no effect on corpseremoval activity in trial 2. This assumption apparently was true under low-stimulus conditions, but not under high-stimulus conditions. There was no consistent decrease in corpse-removal rate at lowstimulus conditions in trial 2 compared with trial 1. However, there was a significant (paired t-test: P<0·05), approximately 30–50%, decrease in corpse-removal rate at high-stimulus conditions in trial 2 relative to trial 1 (Fig. 2). This decrease in colony performance was observed after the removal of less than 5% of each colony’s workers.

It seems to have occurred because there was a depletion of individuals with specific genotypes, and no measurable shift to undertaker activity by other colony members with different genotypes. Alternatively, observed differences in corpseremoval rates could have been caused by intertrial differences in environmental and/or colony conditions. Experiment 2 was therefore performed to test the hypothesis that depletion of a colony’s undertakers can cause a persistent decrease in the rate of colony corpse removal. EFFECTS OF DEPLETION OF UNDERTAKERS ON COLONY CORPSE-REMOVAL RATE Methods Eight colonies were used, each with a different, naturally mated queen. Colonies were paired; we chose randomly one colony in each pair to be the ‘depleted colony’, and the other the ‘non-depleted’ colony. Both members of a colony pair were located in the same apiary and were as similar as possible to one another in amounts of comb, brood, food stores and population size (ca 30 000– 60 000 adult workers). Colony population size was estimated by counting the number of combs in each hive covered with bees and multiplying by 2000 (Burgett & Burikam 1985). Colonies occupied two- or three-storey Langstroth hives. We induced undertaking behaviour in each colony by adding 1000 corpses as described above. We then determined colony corpseremoval rate by counting the number of corpses removed in 60 min. This was the pre-depletion, ‘day 0’, rate of corpse removal. We removed all unremoved corpses immediately after the 60-min period and added another 1000 corpses to each colony that was to be depleted. We collected all undertakers observed during a 4-h period, replenishing the supply of corpses hourly to keep the stimulus level at approximately 1000 corpses. We left non-depleted colonies undisturbed after measuring the day 0 rate until the day 1 rate was measured. We assessed the effect of undertaker depletion by measuring colony corpse-removal rates (without collecting undertakers) on several days after the depletion and comparing them with the day 0 rate. This was done for both depleted and nondepleted colonies. We measured corpse-removal

Robinson & Page: Genotypic constraints on honey bee plasticity

rates on days 1, 4 or 5, 8 and 15 after depletion, or until the removal rate was greater than or equal to the day 0 rate, whichever came first. There were two or three separate observation periods each day, during 0930–1030, 1230–1500 and 1600– 1800, weather permitting. Measurements were not made under heavy overcast or rainy conditions. Most post-depletion counts of corpse-removal activity were performed simultaneously for each member of a colony pair to control for temporal variation in colony and environmental conditions. When this was not possible, we made counts for each member of a colony pair within 1 h of each other. We removed all remaining corpses from each hive at the conclusion of each measure of corpse-removal activity.

4560

0·8 *

** 0·4

0 4561

0·8 *

**

**

871

**

0·4

0 4562

Results

0·8 *

Proportion

0·4

0 4563

**

0·8 **

*

** 37

0·4

0 4564

** 0·8

10

39

19 0·4

A total of 562, 1259, 1185 and 1090 undertakers were removed from depleted colonies 1, 2, 3 and 4, respectively. This represented a depletion of 1·9, 3·1, 3·0 and 1·8%, respectively, of each colony’s estimated population. There were significant differences in the rate of corpse removal in the depleted colonies relative to the non-depleted colonies (Fig. 3). The rate of corpse removal decreased in three of four depleted colonies relative to day 0 rates. Overall, the largest decrease in mean (&) corpse-removal rate was measured 1 day after depletion, 69·5&9·4% of the day 0 rate (N=3 colonies). Rates were still somewhat depressed in depleted colonies 3 and 1 on days 5 and 8, respectively. There was a trend of increasing corpse-removal rates over time in three out of four depleted colonies, suggesting a progressive recovery from the effects of undertaker depletion. In depleted colony 3, this increase was significant (r2 =0·75, P<0·01). Corpse-removal rates increased dramatically in the non-depleted colonies relative to day 0 rates

0 4565

0·8

*** ***

***

*** *

30

38

0·4

0

Colony

Low 1 High 1 Low 2 High 2

Figure 1. Genotypic composition of samples of undertakers responding to either a low or high stimulus for corpse removal in trials 1 and 2, and samples representing the whole colony. Samples taken from honey bee colonies with electrophoretically distinguishable, ‘slow’ ( ), ‘medium’ ( ), and ‘fast’ ( ) subfamilies. N=40, except where given above bars (exceptions explained in Methods). *P<0·05; **P<0·01; ***P<0·001; G-tests for heterogeneity (Sokal & Rohlf 1981).

Animal Behaviour, 49, 4

872

DISCUSSION

Time required to remove 50 corpses (min)

180 150

Low stimulus

120 90 60 30 0

4560 4561 4562 4563 4564 4565

800 Number of corpses removed in 3 h

High stimulus 600

400 P < 0·05 200

0

4560 4561 4562 4563 4564 4565 Colony

Figure 2. Changes in honey bee colony corpse-removal rates in trial 2 (/) relative to trial 1 (.), with a 7–10-day interval between trials. Note that for lowstimulus conditions, decreases in corpse-removal rates are indicated by an increase in time, and for highstimulus conditions, a decrease in number removed. For colony 4564, in low stimulus conditions, observations were terminated after 120 min when only 19 and 10 corpses were removed in trials 1 and 2, respectively.

(Fig. 3). This effect, apparently due to the repeated addition of large numbers of corpses, may have partially masked the full effects of depletion in the four depleted colonies. Nevertheless, the results of this experiment support the hypothesis that removing some undertakers from a colony causes a subsequent decrease in colony corpse-removal activity. It is, thus, plausible that the decrease in corpse-removal rate observed in the previous experiment in trial 2 relative to trial 1 was a consequence of removing a relatively small, but select, group of individuals, in both a behavioural and genetic sense.

Our results do not agree with the predictions of the genotypic threshold model (Robinson & Page 1989a). Workers from some subfamilies were relatively more likely to act as undertakers, as in previous studies (Robinson & Page 1988). However, the undertaker groups that responded to the high stimulus were just as different from the composition of the whole colony, genotypically, as were the undertaker groups that responded to the low stimulus. The observed lack of plasticity for corpse removal at the individual level following depletion also resulted in a lack of plasticity at the colony level. Colonies did increase their undertaking activity when the stimulus for this task increased, but decreases in corpse-removal rates were detected after depletion of undertakers. Decreases in corpse-removal rates lasted for as long as 7–10 days in experiment 1, perhaps because corpses continued to be removed only by individuals with specific genotypes, despite their diminished number, and not by other colony members with different genotypes. Decreases in corpse-removal rates were also apparent for 1–8 days after similar depletions in experiment 2, which was a more rigorous test of the effects of undertaker depletion. Post-depletion decreases in corpse-removal rates may have occurred because there were fewer bees available to act quickly as replacement undertakers. This may have been because not enough undertaker-age bees switched to this task, either from other tasks or from an inactive state. Post-depletion decreases in corpse-removal rates may also be a consequence of new undertakers removing corpses more slowly (if undertaking is a task that improves with experience). Results consistent with the latter suggestion were obtained by Cartar (1992), who reported that bumblebee, Bombus bimaculatus, workers that shifted to foraging in response to an experimentally induced increase in the need for this task were less efficient than more experienced foragers. Depleting a colony of undertakers over several hours is an unnatural perturbation. However, at least some bees in nature probably fail to return to their colony following such a risky and energetically expensive flight during which they carry a corpse of nearly their own weight. In our study the depletions never amounted to more than about 5% of a colony’s population, suggesting

Robinson & Page: Genotypic constraints on honey bee plasticity

873

500 Pair 1

*

*

Pair 2

**

400

% Original corpse removal rate (corpses/60 min)

300

200 118

389

288

469

100

01

5

8

01

15

5

8

5

8

500 Pair 3

Pair 4 *

400

300

200 **

*

171

77

480

278

100

01

5

8

15 Day

01

Figure 3. Effects of undertaker depletion on corpse-removal rates in colonies of honey bees. Corpse-removal rates for depleted (-) and non-depleted (,) colonies were compared on each day. Numbers on graphs indicate the actual, original, day 0 corpse-removal rates. Graphs give mean+ for all values of N greater than 1. Asterisks indicate the comparisons that were significantly different (with paired t-tests): *P<0·05; **P<0·01.

that other appropriately aged bees were present and potentially available to act as undertakers. These results suggest that for at least one task, corpse removal, strong genetic influences on task performance may constrain a colony’s ability to respond to changing conditions. Of course, it is possible that higher levels of colony plasticity for corpse removal exist in other colonies that are composed of different assemblages of subfamilies. This possibility is suggested by the results of experiment 2, in which some colonies were less effected by undertaker depletions than others. In contrast to our results for corpse removal, Calderone & Page (1992) demonstrated plasticity

among individuals of different genotypes for the task of pollen foraging. They reported that co-fostered workers from artificially selected strains of bees that collect and store either high or low amounts of pollen responded differently to changes in foraging-related stimuli in a common colony environment. Similarly, Calderone & Page (1992) found that cross-fostered workers from these strains collected different proportions of pollen and nectar, depending on the colony environment. These results suggest that depletion of a colony’s pollen foragers, or an increase in the demand for pollen, should result in workers from different subfamilies switching from nectar

874

Animal Behaviour, 49, 4

to pollen foraging. Results consistent with this prediction were obtained recently by Fewell & Page (1993). They found that genetically distinct groups of workers with high response thresholds for collecting pollen switched from nectar to pollen collection in response to an increase in the need for pollen. In addition, there are reports of genotypic differences in response thresholds to alarm pheromones (Collins 1979) and to stimuli from brood that influence pollen and nectar foraging (Calderone 1993). The role of genotypic variability in coordinating a colony’s response to changing conditions may vary with task. One possible determinant of behavioural plasticity may be whether the task is ‘rare’, performed by only a fraction of a colony’s workers, or ‘common’, performed by most workers at some point in their life. Undertaking is performed by only a small fraction of a colony’s workers (Visscher 1983), whereas pollen foraging involves over half of a colony’s foraging force (Ribbands 1952; Sekiguchi & Sakagami 1966). Similarly, nectar storage also is a task that is more common than undertaking, and depleting the number of bees that receive nectar from incoming foragers produces only very transient effects on colony food storage activity (Seeley 1989). Perhaps it is more difficult for workers to switch to some tasks because some behavioural states are accessible only via ontogeny (age polyethism) while others are accessible via task-switching within an age-related repertoire. Our results suggest that undertaking is a task that is less accessible via task-switching (but observations of individually labelled bees would be needed to demonstrate this). Alternatively, different individuals may vary in their competence to develop into a particular behavioural state because they have different genotypes or different prior experiences. Another possible determinant of behavioural plasticity is what other tasks are performed by workers during the time in their life when they would be competent to perform the task in question. Individuals may be able to switch quickly to perform a needed task, but do not, because they are more sensitive to the stimuli eliciting the performance of a different task. Presumably this would reflect differences between tasks in relative importance (at the ultimate level), and/or temporal requirements (at the proximal level). For example, a colony may be able to postpone or

diminish undertaking activity at less cost than decreasing the performance of other tasks performed by middle-aged bees, such as guarding the hive entrance, receiving food from incoming foragers, and building comb. This suggests that the response threshold of workers to single tasks should not be considered independently of response thresholds to other tasks. Gordon (1989b) has shown that workers in colonies of Pogonomyrmex barbatus harvester ants are more likely to respond to perturbation by shifting to foraging than to nest maintenance, which suggests that there are task hierarchies that may be proximally based on differences in sensitivity to task-related stimuli. Although the behavioural repertoires are known for workers of many social insect species (reviewed by Michener 1974; Oster & Wilson 1978; Seeley 1985; Winston 1987; Hölldobler & Wilson 1990), the existence or organization of such task hierarchies has not been studied. Depleting colonies of undertakers revealed that colony-level plasticity for corpse-removal behaviour is constrained by genetic influences on taskswitching. However, the continued presence of a large number of corpses appeared to ‘prime’ colonies for corpse removal. The results of experiment 1 suggest that this priming effect is not due to task-switching by undertaker-aged bees, and the rapidity of the response precludes ontogenetic changes as a sole explanation. Increased colony corpse-removal rates in response to a continued presence of large numbers of corpses, therefore, apparently is a consequence of increased activity levels of already committed undertakers. It is possible, however, that workers respond differently to undertaker depletion than they do the continued presence of large numbers of corpses, even though both manipulations seem likely to increase the stimulus levels. Detailed observations of individually labelled bees, before and after conditions are changed, would be needed to determine precisely which behavioural responses of individuals to changing conditions (see Introduction) are involved in colony plasticity for corpse removal. Our results demonstrate strong genetic effects on the likelihood of becoming an undertaker, but no detectable graded colony responses to changing conditions based on differential recruitment among genotypes. Thus it is not apparent that genotypic diversity for corpse removal within

Robinson & Page: Genotypic constraints on honey bee plasticity honey bee colonies improves the response to changing environmental conditions, at least for corpse removal. The presence of genotypic diversity within colonies might be important, however, if the observed genotypic constraints for this behaviour are such that some subfamilies exhibit a low probability of corpse removal even in colonies that do not have undertakers from other subfamilies, leading to a shortage of undertakers under some conditions (Page et al., in press). Determining the extent of, and basis for, genotypic constraints on corpse removal may increase our understanding of the effects of worker genetic variability on colony division of labour. ACKNOWLEDGMENTS We thank B. D. Burrell and M. K. Fondrk for assistance in the field, J. A. Gianelos for assistance with the electrophoresis and T. Giray, Z.-Y. Huang, S. P. Trumbo and C. Wagener-Hulme for comments that improved the manuscript. The research was supported by grants to G.E.R. (NSF BSR-8800227 and USDA 92-37302-7856) and R.E.P. (NSF BNS-8719283). REFERENCES Adams, J., Rothman, E. D., Kerr, W. E. & Paulino, Z. L. 1977. Estimation of the number of sex alleles and queen matings from diploid male frequencies in a population of Apis mellifera. Genetics, 86, 583–596. Breed, M. D., Robinson, G. E. & Page, R. E. 1990. Division of labor during honey bee colony defense. Behav. Ecol. Sociobiol., 27, 395–401. Burgett, M. & Burikam, I. 1985. Number of adult honey bees (Hymenoptera: Apidae) occupying a comb: a standard for estimating colony populations. J. econ. Entomol., 78, 1154–1156. Calabi, P. 1988. Behavioral flexibility in Hymenoptera: a re-examination of the concept of caste. In: Advances in Myrmecology (Ed. by J. C. Trager), pp. 237–258. Leiden: Brill. Calderone, N. W. 1993. Genotypic effects on the response of worker honey bees, Apis mellifera, to the colony environment. Anim. Behav., 46, 403–404. Calderone, N. W. & Page, R. E., Jr. 1991. Evolutionary genetics of division of labor in colonies of the honey bee (Apis mellifera). Am. Nat., 138, 69–92. Calderone, N. W. & Page, R. E., Jr. 1992. Effects of interactions among genotypically diverse nestmates on task specialization by foraging honey bees (Apis mellifera). Behav. Ecol. Sociobiol., 30, 219–226. Cartar, R. V. 1992. Adjustment of foraging effort and task switching in energy-manipulated wild bumblebee colonies. Anim. Behav., 44, 75–87.

875

Collins, A. M. 1979. Genetics of the response to alarm pheromones by caged honeybees. Ann. entomol. Soc. Am., 73, 285–291. Contel, E. P. B., Mestriner, M. A. & Martins, E. 1977. Genetic control and developmental expression of malate dehydrogenase in Apis mellifera. Biochem. Genet., 15, 859–876. Crozier, R. H. & Page, R. E. 1985. On being the right size: male contributions and multiple mating in social Hymenoptera. Behav. Ecol. Sociobiol., 18, 105–115. Del Lama, M. A., Soares, A. E. E. & Mestriner, M. A. 1985. Linkage studies in Apis mellifera using biochemical and morphological markers. J. Hered., 76, 427–430. Fewell, J. H. & Page, R. E., Jr. 1993. Genotypic variation in foraging responses to environmental stimuli by honey bees, Apis mellifera. Experientia, 49, 1106–1112. Gordon, D. M. 1989a. Caste and change in social insects. Oxf. Surv. evol. Biol., 6, 67–72. Gordon, D. M. 1989b. Dynamics of task switching in harvester ants. Anim. Behav., 38, 194–204. Hölldobler, B. & Wilson, E. O. 1990. The Ants. Cambridge, Massachusetts: Harvard Belknap. Huang, Z.-Y., Robinson, G. E. & Borst, D. W. 1994. Physiological correlates of division of labor among similarly aged honey bees. J. comp. Physiol. A, 174, 731–739. Jeanne, R. L. 1987. Do water foragers pace nest construction activity in Polybia occidentalis. Experientia Suppl., 54, 241–252. Kaftanoglu, O. & Peng, Y. S. 1980. A washing technique for collection of honeybee semen. J. apicult. Res., 19, 205–211. Laidlaw, H. H. 1977. Instrumental Insemination of Honey Bee Queens. Hamilton, Illinois: Dadant. Michener, C. D. 1974. The Social Behavior of the Bees. Cambridge, Massachusetts: Harvard Belknap. Moritz, R. F. A. 1983. Homogeneous mixing of honeybee semen by centrifugation. J. apicult. Res., 24, 249–255. Nunamaker, R. A. & Wilson, W. T. 1980. Some isozymes of the honey bee (Apis mellifera L.). Isozyme Bull., 13, 111–112. Oldroyd, B. P., Rinderer, T. E., Buco, S. M. & Beaman, L. D. 1993. Genetic variance in honey bees for preferred foraging distance. Anim. Behav., 45, 323–332. Oldroyd, B. P., Rinderer, T. E., Harbo, J. R. & Buco, S. M. 1992. Effects of intracolonial genetic diversity on honey bee (Hymenoptera: Apidae) colony performance. Ann. entomol. Soc. Am., 85, 335–343. Oster, G. F. & Wilson, E. O. 1978. Caste and Ecology in the Social Insects. Princeton, New Jersey: Princeton University Press. Page, R. E., Jr. 1986. Sperm utilization in social insects. A. Rev. Entomol., 31, 297–320. Page, R. E., Jr & Metcalf, R. A. 1982. Multiple mating, sperm utilization, and social evolution. Am. Nat., 119, 263–281. Page, R. E., Jr & Metcalf, R. A. 1988. A population estimate of Mdh allozyme frequencies for the

876

Animal Behaviour, 49, 4

honeybee, Apis mellifera L. (Hymenoptera: Apidae). Pan-Pacif. Entomol., 64, 285–289. Page, R. E., Jr & Robinson, G. E. 1991. The genetics of division of labour in honey bee colonies. Adv. Insect Physiol., 23, 117–171. Page, R. E., Robinson, G. E. & Fondrk, M. K. In press. Effects of worker genotypic diversity on the development and behaviour of honey bee colonies (Apis mellifera L.). Behav. Ecol. Sociobiol. Pellet, F. C. 1938. History of American Beekeeping. Ames, Iowa: Collegiate Press. Phillips, E. F. 1915. Beekeeping. New York: Macmillan. Ribbands, C. R. 1952. Division of labour in the honeybee community. Proc. R. Soc. Ser. B., 140, 32–43. Robinson, G. E. 1992. Regulation of division of labor in insect societies. A. Rev. Entomol., 37, 637–665. Robinson, G. E. & Page, R. E., Jr. 1988. Genetic determination of guarding and undertaking in honey bee colonies. Nature, Lond., 333, 356–358. Robinson, G. E. & Page, R. E., Jr. 1989a. Genetic basis for division of labor in an insect society. In: The Genetics of Social Evolution (Ed. by M. D. Breed & R. E. Page), pp. 61–80. Boulder, Colorado: Westview Press. Robinson, G. E. & Page, R. E., Jr. 1989b. Genetic determination of nectar foraging, pollen foraging, and nest-site scouting in honey bee colonies. Behav. Ecol. Sociobiol., 24, 317–323. Sakagami, S. F. 1953. Untersuchungen über die Arbeitsteilung in einem Zwergvolk der Honigbiene.

Beiträge zur Biologie des Bienenvolkes, Apis mellifera L. I. Jap. J. Zool., 11, 117–185. Seeley, T. D. 1985. Honeybee Ecology: A Study of Adaptation in Social Life. Princeton, New Jersey: Princeton University Press. Seeley, T. D. 1989. Social foraging in honey bees: how nectar foragers assess their colony’s nutritional status. Behav. Ecol. Sociobiol., 24, 181–189. Sekiguchi, K. & Sakagami, S. F. 1966. Structure of foraging population and related problems in the honeybee, with considerations on the division of labour in bee colonies. Hokkaido natn. agric. Exp. Stn. Rep., 69, 1–65. Sherman, P. W., Seeley, T. D. & Reeve, H. K. 1988. Parasites, pathogens, and polyandry in the social Hymenoptera. Am. Nat., 131, 602–610. Snyder, L. E. 1992. The genetics of social behavior in a polygynous ant. Naturwissenschaften, 79, 525–527. Sokal, R. R. & Rohlf, F. J. 1981. Biometry. 2nd edn. San Francisco: W. H. Freeman. Stuart, R. J. & Page, R. E., Jr. 1991. Genetic component to division of labor among workers of a leptothoracine ant. Naturwissenschaften, 78, 375–377. Visscher, P. K. 1983. The honey bee way of death: necrophoric behaviour in Apis mellifera. Anim. Behav., 31, 1070–1076. Wilson, E. O. 1971. The Insect Societies. Cambridge, Massachusetts: Harvard Belknap. Winston, M. L. 1987. The Biology of the Honey Bee. Cambridge, Massachusetts: Harvard Belknap.