Comb construction in mixed-species colonies of honeybees, Apis cerana and Apis mellifera

Nest-construction behaviour in social insects is a very complex and highly cooperative phenomenon (Wilson 1971; Belic et al., 1986). In honeybee colonies, the nests result from numerous kinds of operations performed by many individual bees (Hepburn, 1986). Although the task of comb construction, like many tasks performed in a honeybee colony, requires concerted actions by many nestmates, individuals are in fact very poorly informed and lack a central controller. Coordination relies on subtle mechanisms combining individual decision rules with specialised signals and informative local cues (Pratt, 2004). It has been suggested that the comb building of honeybees can be interpreted as a model of self-organisation (Belic et al., 1986; Bonabeau et al., 1997; Hepburn, 1998).

Indeed, regulation of behaviour through self-organisation (Bonabeau et al., 1997; Boomsma and Franks, 2006; Detrain and Deneubourg, 2006), specifically in honeybee societies, can be used to interpret behaviours including comb construction (Belic et al., 1986; Hepburn, 1998), as well as the arrangement of food storing and brood rearing on the combs (Camazine et al., 1990; Camazine, 1991) and the regulation of food collection behaviour (Jenkins et al., 1992).

Several studies on comb building in Apis mellifera have shown that some very simple building rules (Darchen, 1954; Hepburn and Whiffler, 1991), coupled to the physico-chemical properties of beeswax as a building material (Pirk et al., 2004; Buchwald et al., 2006), can parsimoniously explain several aspects of comb-building behaviour. However, the two sister-species, Apis cerana and A. mellifera, not only differ in the chemical components of their waxes (Aichholz and Lorbeer, 1999) but also have different worker cell sizes [5.2–5.4 mm in diameter for A. mellifera (Winston, 1987); 4.2–4.8 mm in diameter for A. cerana (Ruttner, 1988)]. In order to induce the colonies to build combs more quickly and with more regularity, artificial beeswax foundation embossed with the average cell sizes are commonly used for a particular species.

In other studies of mixed species, it has been shown that A. cerana and A. mellifera can decipher their differing dance languages (Su et al., 2008; Tan et al., 2008) but differ in their thermoregulatory fanning behaviour (Yang et al., 2010b). So, mixed colonies offer us a valuable opportunity for the integrative investigation of the relationships of the two species and provide us with a new perspective to study the theories of self-organisation in honeybees as well as in investigations of the evolution of behaviour. Division of labour in mixed-species colonies remains an intriguing issue not previously considered. In this paper we report studies on the comb-construction behaviour of mixed colonies of A. cerana and A. mellifera to answer several questions: (1) will the mixed colonies accept the waxes of both species? (2) Will pure colonies of A. cerana accept A. mellifera wax and vice versa? (3) Given that the bees are presented with beeswax foundation of different cell base sizes, are these accepted as such or are they modified? (4) Do A. cerana workers and A. mellifera workers cooperate heterospecifically in comb building or do they form separate, conspecific festoons? (5) Under the various conditions above, what cell sizes would result in the newly constructed combs? (6) Once constructed, how are these cells used in the economy of the nest?

Mixed colonies of both Apis cerana L. and Apis mellifera L. workers were established. Three colonies were headed by A. cerana queens; and, reciprocally, three colonies were headed by A. mellifera queens. Sealed brood about to emerge as young adults of each species was placed in the colonies of the other species (Tan et al., 2006). Wax-building behaviour was investigated when the newly emerged workers of the two species were about 10–18 days old, the peak age of wax secretion (Rösch, 1927; Hepburn et al., 1984; Seeley, 1995). Each of three pure A. cerana and A. mellifera colonies with the same age cohort of workers were selected as control groups. The colonies were equalised for size, numbers of combs and adult bees, nectar and pollen stores and brood. This was achieved simply by adding or removing combs.

The experiments on the utilisation of the newly built combs in the pure A. cerana and A. mellifera colonies were done at an apiary of Yunnan Agricultural University, Kunming, China, to refine the final experimental protocol. In these experiments, beeswax was extracted from the combs of both A. cerana and A. mellifera and then used to make small sheets of beeswax foundation of two worker cell sizes: A. cerana, about 4.75 mm diameter and A. mellifera, 5.35 mm diameter using a silicon rubber mould (Hepburn et al., 2009). The foundations were about 25 mm long × 80 mm wide × 2 mm thick for the A. mellifera queen-headed mixed colonies, and about 12.5 mm long × 80 mm wide × 2 mm thick for the pure-species colonies and the A. cerana queen-headed mixed colonies, because they were more concentrated in the middle areas, the foundation widths were cut by half (honeycomb foundation used in commercial beekeeping is made from beeswax by pressing a blank sheet of wax between metal moulds on which hexagons are embossed so as to create a facsimile of natural comb).

We inserted both A. cerana cell-size (4.75 mm diameter) foundation and A. mellifera cell-size (5.35 mm diameter) foundation into pure A. cerana colonies and pure A. mellifera colonies. The experiments on cell size and wax discrimination, and comb-building cooperation were conducted with colonies of A. cerana cerana and A. mellifera ligustica at an apiary at the Ratchaburi Campus of King Mongkut's University of Technology Thonburi, Thailand. The same four types of beeswax foundation sheets (two cell sizes and two wax types) were fixed on the top bar of a frame, their relative positions determined by random number assignment. They were then inserted into the centre of the hives.

We used a video camera (Sony handycam DCR-DVD803E PAL) and DVD recorder (Sony DVD+RW, DPW60DSS2) to record the comb-building behaviour of the test and control colonies for 10 s intervals three times a day, the videos were taken every day for all of the replicates (Table 1). On replay of the video clips, we were able to obtain detailed information on: (1) how many workers of each species were engaged in which type of comb building; (2) how many starting sites were used to extend the building of new combs; (3) whether they formed a mixed-species building chain and cooperated with each other in comb building; (4) how many workers of each species were in each festoon; and (5) when building was complete. When the foundation sheets had been extended beyond their original lengths by the addition of several centimetres of new wax, the combs were removed from the hive and represented one sample for that colony. These combs were replaced by a new top bar with the same four kinds of foundation. The cell sizes and cell patterns were measured.

Chi-square tests were used to test for differences in the numbers of modified cells and patterns of cell orientation between the four types of colonies of A. mellifera and A. cerana queen-headed mixed colonies, and pure A. cerana and A. mellifera colonies. To test for differences in the mean numbers of workers engaged in comb building and the mean cell size of the newly built combs between the four types of colonies, we used analysis of variance (ANOVA) and Tukey post-hoc tests. Homogeneity of the variances between types of colonies was checked using Levene's test. Paired samples t-tests were used to compare the mean number and proportions of A. cerana workers with A. mellifera workers in the A. mellifera and A. cerana queen-headed mixed colonies (Johnson and Wichern, 2002). The means and standard deviations of each variable were calculated. All tests were performed using Statistica (version 8, www.statsoft.com; StatSoft).

Pure A. cerana colonies ignored all sheets of beeswax foundation and began building new combs either from the top bar or from the lower edges of the foundation sheets (Fig. 1A). By contrast, the pure A. mellifera colonies accepted all sheets of both A. cerana and A. mellifera foundation and built cells on both cell sizes (Fig. 1B). In the two types of mixed colonies, all the four types of foundations were accepted (Fig. 1C,D); workers of both species were seen building cells on the foundations (Fig. 2, Table 1). None of these colonies showed any preference for a particular type of foundation with respect to wax type or cell size (repeated-measures ANOVA: P>0.05, Table 1).

All of the A. mellifera cell-size sheets of foundation were built to their original size without any modification (Table 2) but the A. cerana cell-size foundation sheets were modified in all colonies except the pure A. cerana colonies. Some of these cells were squeezed to make space for enlarging neighbouring cells. The percentages of combs that had modified cells in the test and control groups are shown in Table 2. In A. mellifera queen-headed mixed colonies, all of the A. cerana foundation sheets were modified and nearly all in the pure A. mellifera colonies, which is significantly different from the A. cerana queen-headed mixed colonies and pure A. cerana colonies (Chi-square: χ²₃=71.7, P<0.001).

On completion of the trials of comb building on the artificial foundation sheets, the workers from the four types of colonies were observed to start building new combs at several sites (Table 3). Pure A. mellifera colonies and A. mellifera queen-headed mixed colonies had significantly more festoons at new comb-building sites than A. cerana and pure A. cerana queen-headed colonies (ANOVA: F_3,44=15.9, P<0.001; Table 3). In A. cerana queen-headed mixed colonies, workers of both species were seen working together in festoons, although significantly more A. mellifera workers were involved (42.1±6.2% A. cerana workers, 57.9±6.2% A. mellifera workers; paired t-test: t₁₃=4.9, P<0.001). Similarly, in A. mellifera queen-headed mixed colonies, significantly more A. mellifera workers than A. cerana workers were engaged in comb building in the festoons (32.5±4.8% A. cerana workers, 67.5±4.8%, A. mellifera workers; paired t-test: t₉=9.8, P<0.001; Table 3). In total, significantly more workers were engaged in comb building in the mixed colonies than in the pure A. cerana and pure A. mellifera colonies (ANOVA: F_3,44=11.3, P<0.001; Table 3).

As for irregular cells on the new combs, pure A. cerana and pure A. mellifera colonies built significantly fewer irregular cells (0.8% and 2.7%, respectively), than did the mixed colonies (9.1% and 10.8%, respectively), most of which were located at the seams of combs which had been started at different sites (ANOVA: F_3,44=30.0, P=0.003; Table 3). The A. cerana queen-headed mixed colonies showed significantly greater variation in the patterns of cell orientation on the newly built combs than A. mellifera queen-headed colonies, pure A. cerana and A. mellifera colonies; different festoons on one comb may build patterns different from others (Chi-squared: χ²₆=27.9, P<0.001; Fig. 1C; Table 3). Apis mellifera queen-headed colonies built new combs mainly in vertical and horizontal patterns (Fig. 1D); in pure A. cerana and A. mellifera colonies, the patterns of cell orientation were more homogeneous and mainly vertical (Figs 1A,B; Table 3).

The different mixed colonies built significantly different sized cells (ANOVA: F_3,44=34.8, P<0.001; Table 3). The largest cells were built by A. mellifera queen-headed mixed colonies. The cells built in the pure A. mellifera colonies and A. mellifera-queen-headed mixed colonies were like those A. mellifera drone cells (European type, 6.0–6.3 mm) whereas in the A. cerana queen-headed mixed colonies, the cells had a diameter of 5.41±0.27 mm, which is like normal A. mellifera worker size cells. The pure A. cerana colonies built cells of 4.38±0.06 mm in size, which is the normal A. cerana worker cell size.

In these experiments, we inserted both A. cerana cell-size (4.75 mm diameter) foundation and A. mellifera cell-size (5.35 mm diameter) foundation into pure A. cerana colonies and pure A. mellifera colonies with the following results. Pure A. cerana colonies accepted both foundation types and built cells without altering the original cell base. Pure A. mellifera colonies accepted both foundation wax types but changed the A. cerana cell size to their normally larger cells with the inclusion of many irregular cells.

Once the control combs had been constructed, A. cerana colonies differed from A. mellifera colonies in the subsequent use of these cells. The pure A. cerana colonies used the A. mellifera size cells either for food storing (Fig. 3) or drone brood rearing while the A. cerana size cells were normally used for rearing worker brood (Fig. 4). In pure A. mellifera colonies, the queens laid eggs into both A. mellifera size cells and A. cerana size cells but they all showed a preference for A. mellifera size cells and laid eggs into the former cells first and more regularly (Fig. 5).

It is common knowledge that the cavity-dwelling honeybees build multiple, parallel combs and that this parallelism is recognised as a building rule (Darchen, 1954; Hepburn, 1986; Hepburn and Muller, 1988). Comb-constructing bees work in a dark cavity or hive where there is no central source of information. When construction begins, the workers cling together in elongated chains or festoons, forming a dense cluster that facilitates an equable temperature for wax secretion and manipulation (Hepburn, 1986). Numerous comb-building workers, with active wax glands, engage in the task of comb construction. But, instead of building a single comb together, several festoons begin at independent sites and begin building several cells (hence combs) simultaneously and only later connect them with some irregular transitional cells (Hepburn, 1986; Hepburn and Whiffler, 1991). In this case, the parallelism rule can only be achieved indirectly at the finishing stage of comb building, with many irregular cells and seam connections between several branches started from separate sites (Hepburn and Whiffler, 1991). In this context it can be noted that A. cerana and A. mellifera workers cooperate heterospecifically in the same festoons in comb building.

It is somewhat strange that in the pure A. cerana colonies, none of the four types of foundations were accepted, although two of the four foundations were embossed with normal A. cerana cell size. In sharp contrast, in the pure A. mellifera colonies, the workers were seen building cells on both types of wax and both cell sizes. These results indicate that A. mellifera workers are more tolerant of wax and cell-size factors. This contrast is revisited in both types of mixed colonies where more A. mellifera workers than A. cerana workers were seen building comb, irrespective of the host queen. However, interestingly, A. cerana workers did engage in comb building on foundations of both waxes and the two cell sizes in the both types of the mixed colonies (Table 1). This certainly suggests that A. mellifera comb-building workers can stimulate A. cerana workers to start comb building. And, a comb-building stimulus appears reciprocal because in pure A. mellifera colonies, while 83.3% of the A. cerana cell-size foundation sheets were modified and expanded to A. mellifera cell-size, only 10.7% were modified in mixed colonies headed by A. cerana queens. In the A. cerana queen-headed mixed colonies, more A. mellifera workers were engaged in comb building in festoons, so it is not surprising that the cell sizes were similar to normal A. mellifera workers.

It is interesting to note that in this type (A. cerana queen-headed) of mixed colony, the festoons were formed predominately by A. mellifera workers with fewer A. cerana workers joining them. However, the combs built in the mixed colonies did have more irregular cells than were observed in any of the pure A. cerana or A. mellifera colonies. This seems to indicate that the A. cerana workers also play a role in determining final cell-size. Although they did cooperate with each other in festoons, the two species cannot really perform the comb-building task harmoniously. The fact that the combs in the pure A. mellifera colonies and A. mellifera queen-headed colonies were built into normal A. mellifera drone size cells may be related to the season in which we conducted the experiment.

In conclusion, the A. cerana workers, as a colony did not accept any type of beeswax foundation but as individuals can be stimulated by A. mellifera workers to engage in comb building. So, our results are consistent with the idea that honeybee comb-building behaviour is an example of self-organisation. We also confirm that in the mixed colonies, these two closely related honeybee species did in fact cooperate in comb building, even though irregular cells arise through their joint efforts. We can also infer that, although the comb-building workers are poorly informed and lack a central controller (Pratt, 2004), comb building is really a task that can only be finished by some smaller groups in which individuals closely cooperate to achieve progress. This might explain, in part, why A. mellifera workers do not dominate the comb-building effort.

The results presented here based on mixed-species colonies reinforce the conclusion that this experimental method is extremely useful for testing underlying mechanisms that evoke or suppress certain behaviours. Such an experimental context has been successfully used to elucidate disruption of social networks as in ovarial activation (Tan et al., 2009) as well as stimulation of social networks as in the dance language (Tan et al., 2008) and retinue behaviour towards queens (Yang et al., 2010a). The results from the comb-building experiments provide additional evidence for the value of mixed-species colonies as experimental probes.

Financial support was provided by Rhodes University and Yunnan Agricultural University to M.-X.Y.