Competition and cooperation: bumblebee spatial organization and division of labor may affect worker reproduction late in life
- First Online:
- Cite this article as:
- Jandt, J.M. & Dornhaus, A. Behav Ecol Sociobiol (2011) 65: 2341. doi:10.1007/s00265-011-1244-9
- 253 Views
Within-group conflict may influence the degree to which individuals within a group cooperate. For example, the most dominant individuals within a group often gain access to the best resources and may be less inclined to perform risky tasks. We monitored space use and division of labor among all workers in three colonies of bumblebees, Bombus impatiens, during the ergonomic and queenless phases of their colony cycle. We then measured the two largest oocytes in each worker to estimate each individual's reproductive potential at the end of the colony cycle. We show that workers that remained farther from the queen while inside the nest and avoided risky or more energy-expensive tasks during the ergonomic phase developed larger oocytes by the end of the colony cycle. These individuals also tended to be the largest, oldest workers. After the queen died, these workers were more likely than their nestmates to increase brood incubation. Our results suggest that inactive bumblebees may be storing fat reserves to later develop reproductive organs and that the spatial organization of workers inside the nest, particularly the distance workers maintain from the queen, may predict which individuals will later have the greatest reproductive potential in the colony.
KeywordsDivision of laborOvaryReproductive potentialSpatial organizationWorker competition
Individuals that cooperate in a group may still compete with one another. Dominant individuals, for example, often achieve access to the best food, mates, protection, and, in some cases, the opportunity to be the primary reproductive (Hemelrijk 2002). In some groups, the members may be so functionally integrated that an individual cannot survive in the absence of the group, regardless of their position in a dominance hierarchy. It is at this point that major evolutionary transitions may be observed, such as the switch from single-cell to multicellular organisms as well as the switch from solitary to eusociality (Maynard Smith and Szathmáry 1995; Pepper and Herron 2008).
Individuals within eusocial groups, such as in social insect colonies, have developed elaborate mechanisms to maintain high levels of cooperation. This can lead to groups exhibiting physiological properties similar to those of an organism (Korb 2003; Human et al. 2006). However, it has been suggested that within-group competition, in addition to cooperation, may have had a significant influence on the evolution of colony organization (West-Eberhard 1981; Molina and O'Donnell 2009). Even within the most functionally integrated eusocial species, within-group conflicts persist. For example, many workers within a colony (often referred to as functionally sterile) still retain the potential to lay unfertilized eggs (Cole 1986; Ratnieks and Reeve 1992; Ratnieks et al. 2006). In hymenopteran insects such as wasps, bees, and ants, these unfertilized eggs will develop into males. It is not uncommon then to observe workers competing with one another, and the queen, to increase their own direct fitness by laying eggs (Cole 1986; Ratnieks and Reeve 1992; Wenseleers et al. 2004; Ratnieks et al. 2006). If this conflict over worker reproduction is detrimental to colony fitness, mechanisms should evolve to sufficiently suppress worker reproduction, either through ovarian suppression or policing of worker-laid eggs (Ratnieks and Reeve 1992). In annual species, however, there may be an advantage to allowing a subset of workers to reproduce after the queen dies in order to maximize total colony production of males (Ratnieks and Reeve 1992).
Furthermore, some tasks (such as foraging) may be riskier and/or more energy-expensive than others (West-Eberhard 1981; O'Donnell and Jeanne 1995). In species in which workers exhibit temporal polyethism (e.g., honeybees: Seeley 1982), individuals tend to perform less risky in-nest tasks when they are younger, saving the riskier tasks to be performed when they are older and more expendable (i.e., a “disposable caste”: Porter and Jorgensen 1981; O'Donnell and Jeanne 1995; Woyciechowski and Kozłowski 1998; Tofilski 2002), thus engaging in essentially all colony tasks at some point in their lives. However, in other species, there is potential for some workers to concentrate on less risky in-nest tasks over their entire lives, while others specialize on other tasks such as foraging (Goulson et al. 2002; Foster et al. 2004; Korb and Schmidinger 2004; Yerushalmi et al. 2006; Jandt and Dornhaus 2009; Jandt et al. 2009). Workers that avoid riskier tasks may benefit by an increase in life span and opportunity for reproduction. This creates potential for conflict: If some workers are manipulating colony reproduction in favor of their own direct fitness benefits (Ratnieks and Reeve 1992), they should also avoid being allocated to risky tasks.
In bumblebees (Bombus spp.), the presence of the queen inside the nest suppresses ovary development in workers early in the colony cycle (Bloch and Hefetz 1999a, b; Cnaani et al. 2002; Alaux et al. 2004), a factor that is probably an evolutionary benefit to both the queen and workers (Ratnieks and Reeve 1992; Bourke and Ratnieks 1999; Ratnieks et al. 2006). During this time, workers may concentrate their efforts on particular in-nest tasks or foraging, or they may engage in a variety of tasks (O'Donnell et al. 2000; Jandt and Dornhaus 2009; Jandt et al. 2009). Unlike the large proportion of Bombus terrestris workers (∼50%) that begin developing ovaries while the queen is still active in the colony (Duchateau and Velthuis 1989), Bombus impatiens workers generally do not develop ovaries until after the queen dies (or in some cases is physically killed from the colony by the workers; Bourke 1994). After the queen has been dead for 7–8 days (Cnaani et al. 2002; Jandt, unpublished data), a small proportion of workers (∼9%; Cnaani et al. 2002) inside the nest develop their ovaries and begin laying eggs that can develop into males (Ratnieks and Reeve 1992; Bourke 1994; Cnaani et al. 2002; Ratnieks et al. 2006). These reproductively active individuals have a distinct cuticular chemical profile and may also be capable of suppressing ovarian development in the remaining workers in the nest as the queen did (Ayasse et al. 1995; Bloch and Hefetz 1999b; Sramkova et al. 2008).
B. impatiens workers are also spatially organized inside the nest (Jandt and Dornhaus 2009). There are no physical barriers to keep them in place, yet about 10% of workers remain in the center of the nest, whereas another 10% remain on the periphery throughout their lives. The queen is most often found near the center of the group (Jandt and Dornhaus 2009). As she is the primary egg layer in the colony, the spatial pattern is similar to what is observed in other social groups in which the dominant or primary reproductive individual is also often found in the center (Hamilton 1971; Hemelrijk 2000, 2002). Furthermore, in the related species B. terrestris, it has been reported that workers that remain close to the queen are more likely to dominate their nestmates, to develop their ovaries, and to lay eggs (van Doorn and Heringa 1986). Non-reproductive division of labor in B. impatiens has also been linked to spatial organization (Jandt and Dornhaus 2009). For example, workers that are more likely to feed larvae are often found near the center of the colony, whereas workers who are more likely to forage remain at the nest periphery (Couvillon and Dornhaus 2009; Jandt and Dornhaus 2009). It is unknown for any Bombus spp., however, how space use and task performance change after the queen dies and whether these factors are affected by the variation among workers in their potential to reproduce.
B. impatiens workers are closely related to one another because the queens tend to mate only once (Payne et al. 2003), yet there is still potential for conflict to arise among workers for the opportunity to produce sons at the end of the colony cycle (Ratnieks and Reeve 1992). Given that workers are likely to continue performing the same task from day to day (Jandt et al. 2009), it may be possible to see early effects of this conflict on division of labor, even though actual reproduction takes place much later (Cnaani et al. 2002).
Here, we examine the task performance and space use of B. impatiens workers during the ergonomic (population plateau) phase and the queenless phase of the annual colony cycle. If division of labor affects a worker's probability to develop ovaries, then this creates potential for conflict: Workers may remain inactive and/or perform less risky or energy-demanding tasks in order to ensure that they can reproduce later (Tindo and Dejean 2000; Foster et al. 2004; Korb and Schmidinger 2004; Cant and Field 2005). If spatial arrangement affects a worker's probability to develop ovaries, then some individuals may avoid conflict by maintaining spatial fidelity zones (Sendova-Franks and Franks 1995), either in the center of the nest or on the nest periphery (Jandt and Dornhaus 2009). If the observations of van Doorn and Heringa (1986) hold true for B. impatiens, then we would expect workers with the greatest reproductive potential to be found in the center of the nest and closer to the queen while she is alive. However, if proximity to the queen and her signal of reproductive viability (Keller and Nonacs 1993) negatively affect worker ovary development (Alaux et al. 2004), then workers that remain further from the queen should have the greatest reproductive potential.
We used data collected from three B. impatiens colonies for these analyses (see Jandt and Dornhaus 2009 for colony setup). In-nest position data based on a 2D 10 × 10 grid and instantaneous scan samples of task (e.g., larval feeding, incubating, cell building, guarding, fanning, probing honeypots, foraging, or remaining inactive; see Cameron 1989 for ethograms of these behaviors) were collected on all workers in each colony approximately daily between 1300 and 1500 hours (colony 1, 94 bees; colony 2, 90 bees; colony 3, 154 bees). Using the spatial data, we calculated the 2D area that an individual used inside the nest, as well as the distance maintained from the center of the colony and from the queen. Using the task data, we determined the proportion of time each individual was observed performing a task or remaining inactive (for more details, see Jandt and Dornhaus 2009).
Space and task data were both collected while the queen was present (i.e., ergonomic phase: colony 1, 24 days; colony 2, 32 days; colony 3, 49 days) and for at least 2 weeks after the queen died (i.e., queenless phase: colony 1, 14 days; colony 3, 20 days), or was removed from the colony (colony 2, 20 days). In colony 2, the queen was removed when males began to emerge in the colony. Queenless B. impatiens workers need at least 7–8 days to develop ovaries (Cnaani et al. 2002; Jandt, unpublished data). After collecting data for at least 2 weeks during the queenless phase, the remaining worker bees were removed and preserved in 80% ethanol. Thorax widths of all bees were measured postmortem to the nearest 0.01 mm using digital calipers.
Assessing the reproductive potential of workers
To examine reproductive potential, we dissected all preserved worker abdomens and measured the length and width of the largest oocyte in each ovary to the nearest 0.05 mm using a micrometer under a dissecting microscope. The length of the largest oocyte in bumblebees is tightly correlated with a worker's reproductive status; therefore, the average length of the largest oocytes from each ovary was used as a proxy for reproductive potential in workers (Cnaani et al. 2002; Foster et al. 2004; Geva et al. 2005).
The distribution of ovary development within each colony was analyzed for normality using Kolmogorov–Smirnov tests. All proportions of task performance and probability to remain inactive were arcsine transformed (Sokal and Rohlf 1995). Multiple linear regression, blocked for colony-level effects, was used to determine the effects that age and body size had on ovarian development. Simple linear regressions, blocked for colony, were used to determine the relationship between task performance during the ergonomic phase and future ovarian development. For all analyses examining how behavior differed between the ergonomic and queenless phases, and whether this behavior varied among individuals that would develop large or small ovaries, we used repeated measures multivariate analysis of variance (MANOVA). For these analyses, we categorized workers as having “large” or “well-developed” ovaries (average oocyte >2 mm) or as having “small” or “underdeveloped” ovaries (average oocyte <2 mm; as in Cnaani et al. 2002). Tested variables included space use (i.e., the total area used inside the nest and the distance an individual remained from the colony center; Jandt and Dornhaus 2009), individual tasks (i.e., feeding larvae, incubating pupae, constructing honeypots, fanning, guarding, foraging; Cameron 1989), and proportion of time spent being inactive. Linear regression, blocked for colony, was used to determine whether there was a relationship between ovarian development and distance maintained from the queen. All statistical analyses were run using JMP® v. 8.0.02.
Distribution of reproductive potential
Reproductive potential and division of labor during the ergonomic phase
Individuals that developed large ovaries late in the colony cycle differed in task performance from their nestmates during the ergonomic phase, even while the queen was alive. Workers that were more likely to feed larvae during the ergonomic phase developed smaller ovaries (R2 = 0.04, P = 0.04), whereas workers that remained inactive during this phase developed larger ovaries (R2 = 0.06, P = 0.01). No relationship was found between task performance and ovary development for any other task performed during the ergonomic phase (incubating: R2 = 0.002, P = 0.66; constructing honeypots: R2 = 0.02, P = 0.18; perching: R2 < 0.01, P = 0.75; fanning: R2 = 0.01, P = 0.25; probing honeypots: R2 < 0.01; P = 0.54; foraging: R2 < 0.01, P = 0.88).
Reproductive potential and division of labor during the ergonomic and queenless phases
Results from repeated measures MANOVA (blocked for colony level effects) showing how tasks, degree of inactivity, and space use were affected by the colony phase (ergonomic vs. queenless), the high or low reproductive potential of workers (oocyte), or the interaction of these two factors (phase × oocyte)
F1,73 = 3.39
F1,73 = 0.17
Phase × oocyte
F1,73 = 4.94
F1,73 = 6.40
F1,73 = 15.49
Phase × oocyte
F1,73 = 7.47
F1,73 = 0.79
F1,73 = 0.01
Phase × oocyte
F1,73 = 3.11
F1,73 = 1.34
F1,73 = 1.85
Phase × oocyte
F1,73 = 1.16
F1,73 = 0.05
F1,73 = 0.01
Phase × oocyte
F1,73 = 0.15
F1,73 = 3.18
F1,73 = 0.11
Phase × oocyte
F1,73 = 1.38
F1,73 = 0.49
F1,73 = 0.40
Phase × oocyte
F1,73 = 0.42
F1,106 = 10.63
F1,106 = 2.82
Phase × oocyte
F1,106 = 0.87
F1,97 = 2.50
F1,97 = 1.37
Phase × oocyte
F1,97 = 0.32
Distance to center
F1,97 = 4.67
F1,97 = 0.83
Phase × oocyte
F1,97 = 0.40
More workers remained inactive during the queenless phase compared with the ergonomic phase (colony phase: P < 0.001; Table 1 and Fig. 3c). There was no evidence to suggest that workers with small oocytes changed their level of inactivity more or less than those workers that produced large oocytes (colony phase × oocyte size: P > 0.05; Table 1).
Reproductive potential and in-nest space use
The goal of this study was to determine the extent to which division of labor and spatial organization during the ergonomic and queenless phases of an annual social bumblebee species, B. impatiens, relate to worker fecundity at the end of the colony cycle. We show that workers that remain farther from the queen and are more inactive during the ergonomic phase, i.e., before worker reproduction actually begins, develop larger oocytes by the end of the colony cycle. In order to observe ovarian development in B. impatiens workers, the colony must be queenless for at least 7–8 days (Cnaani et al. 2002; Jandt, unpublished data). During a 14- to 20-day queenless phase, we observed that workers that maintain the largest oocytes begin caring for brood and incubating more often. These results suggest that conflict among nestmates for the opportunity to reproduce, although it is not openly expressed until the queen dies in B. impatiens, affects division of labor even early in the colony cycle. During the queenless phase, workers that do not lay eggs spend less time feeding larvae, but they do continue performing other colony tasks, thus probably helping raise their worker-sisters’ offspring.
In species that exhibit high degrees of temporal polyethism, worker ovaries often gradually degenerate over time if they are not used (Fénéron et al. 1996; Lin et al. 1999; O’Donnell 2001; Tsuji and Tsuji 2005). Interestingly, though, reproductive potential should still correlate with task: As workers transition through reproductive stages, they are simultaneously transitioning between colony behaviors. In species that do not exhibit strong temporal polyethism, reproductive potential has been shown to correlate with the probability of performing particular tasks. For example, in Odontomachus brunneus ants and Ropalidia revolutionalis wasps, egg layers are more likely to care for brood and less likely to forage (Powell and Tschinkel 1999; Robson et al. 2000). On the other hand, in Polistes fuscatus wasps and Liostenogaster flavolineata hover wasps, egg layers are more likely to guard the nest (Judd 2000; Cronin and Field 2007). In bumblebees, which also do not exhibit strong temporal polyethism (Cameron 1989; O’Donnell et al. 2000; Jandt and Dornhaus 2009; Jandt et al. 2009), we show here that reproductive potential positively correlates not only with inactivity and brood care (in terms of incubating behavior) after the queen dies but also with age (older workers tend to have greater reproductive potential) and body size (larger individuals tend to have greater reproductive potential as well).
There are links between reproductive potential of workers, division of labor, and activity level across social insects. In bumblebees observed after the competition point (i.e., after workers begin developing reproductive organs), individuals that are more likely to forage have lower reproductive potential, whereas those individuals with higher reproductive potential are more likely to incubate (B. impatiens: Fig. 3; Bombus hypnorum: Ayasse et al. 1995; Bombus bifarius: Foster et al. 2004; Bombus pratorum: Free 1955). Among wasps, higher-ranking individuals tend to be less active (similar to what we show for B. impatiens early in the colony cycle) and less likely to forage (Tindo and Dejean 2000; Cant and Field 2001). These less active individuals may be able to reserve energy in the form of fat bodies that can be used to develop ovaries (Korb and Schmidinger 2004). It remains possible, then, that bumblebee workers with higher reproductive potential may be choosing tasks, such as incubating, that ensure the care of their own brood while simultaneously protecting their brood from being policed by other workers (Ratnieks and Wenseleers 2008).
Queen proximity could have been an important factor affecting the reproductive potential of workers. The queen signal of reproductive viability that suppresses ovary development in bumblebee workers is not highly volatile, and it seems that the queen needs to be in close contact with workers in order for this signal to remain effective (Alaux et al. 2004). Indeed, there was a relationship between queen proximity and ovary development (Fig. 5). However, we also found evidence that larger B. impatiens workers, which tend to remain further from the queen while inside the nest (Jandt and Dornhaus 2009), were more likely to have higher reproductive potential (Fig. 2). Therefore, we do not yet know whether the distance from the queen had an effect on ovary development on top of the body size effect. Positive correlations between body size and ovary development have also been observed in B. hypnorum (Ayasse et al. 1995) and Camponotus ants (Clémencet et al. 2008), leaving open the possibility that small workers are unable to develop ovaries to the same degree that larger workers can.
It is difficult to determine whether the large, inactive bees that remain on the periphery of the nest are acting selfishly to manipulate their own reproductive interests or whether it benefits the colony to allow a subset of workers to lay eggs at the end of an annual colony cycle. Previous studies have focused on the aggressive nature by which workers achieve reproductive status in the colony. We approached this issue differently by focusing on how spatial organization and division of labor change after the queen dies with respect to worker fecundity. Within eusocial groups, there are often mechanisms by which conflict over worker reproduction is resolved within the colony (Ratnieks and Reeve 1992; Wenseleers et al. 2004; Ratnieks et al. 2006; Wenseleers and Ratnieks 2006). Indeed, we found that early cooperation (task allocation early in the colony cycle) may predict future conflict (worker reproduction at the end of the colony cycle). Spatial organization, furthermore, may be a mechanism for workers to avoid reproductive conflicts of interests (e.g., some individuals remain farther from the queen while she is still reproductively viable). We found no breakdown in the overall level of cooperation among workers after the queen died since non-reproductive workers continued working in all tasks. Still, it remains possible that individual selfish interests may shape the division of labor in B. impatiens, even while the queen is alive.
We thank Nicolas Skye Robbins, Eden Huang, Amanda Barth, and Wendy Isner for their help in data collection, bumblebee dissections, and assistance with bumblebee maintenance. Margaret Couvillon, Scott Powell, and Aimee Dunlap provided feedback on the manuscript and statistical analyses. We also thank Laurent Keller and two anonymous reviewers for their detailed comments and suggestions. Research supported through the College of Science, Department of Ecology & Evolutionary Biology, University of Arizona, and NSF grant to AD (grant no. IOS 0841756).