jo ur nal home p ag e: www.elsev ier .com/ locate /behavproc
ocial interactions and their connection to aggression and ovarianevelopment in orphaned worker bumblebees (Bombus impatiens)
.D. Sibbald ∗, C.M.S. Plowrightchool of Psychology, University of Ottawa, Canada
r t i c l e i n f o
rticle history:eceived 27 June 2013eceived in revised form 7 October 2013ccepted 25 November 2013vailable online 16 December 2013
eywords:umblebeeombus impatiensueen-less workers
a b s t r a c t
This study examines the social dynamics of reproductive conflict. Orphaned worker bumblebees (Bombusimpatiens) with comparatively high or low levels of social activity were paired to determine whetheraggression and reproduction could be traced to earlier social interactions. The workers were pairedaccording to their levels of social activity (a socially active + another socially active worker, sociallyactive + socially inactive, and two socially inactive workers). The presence or absence of brood was alsomanipulated. The absence of brood increased both aggression and ovarian development, suggesting thataggression and reproduction are associated or that there is a third variable that affects both. Sociallyactive pairs were significantly more aggressive: here, social activity can be taken as an early indica-tor of aggression. No such effect, however, was obtained on ovarian development as the socially active
eproductionggressionocial interaction
pairs did not differ on their degree of ovarian development compared to the others. Within the sociallyactive + socially inactive pairs, the socially active worker did not have more developed ovaries and wasnot more aggressive than her socially inactive partner. Results highlight that environmental conditions(the absence of brood) can predict ovarian development and although social activity can be observedprior to aggression, differences in aggression do not translate into differences in ovarian developmentunder these conditions.
. Introduction
Bumblebee workers retain their reproductive capabilities andave the potential to produce male offspring. Nevertheless, throughost of the colony life cycle they forego reproduction in favour of
earing the queen’s offspring (Sladen, 1989). When worker ovipo-ition (egg-laying) does occur, only a subset of workers becomeeproductive, becoming ‘false’ or pseudo queens (van Doorn anderinga, 1986; van Doorn, 1987; van Honk and Hogeweg, 1981).hey are given this label because they cease foraging and onlyay eggs in a way reminiscent of the behaviour of queens. A formf reproductive competition is hypothesized to occur that willistinguish the reproductive workers from the non-reproductivenes and that this competition is expressed through aggres-ive and social interactions (Bourke, 1988a; Heinze, 2008). Innts (Harpagoxenus sublaevis) and wasps (Ropalidia marginata), forxample, rigid social hierarchies are created with a single, repro-
uctive queen. If the queen is removed from the nest, aggression
s observed between the highest-ranked socially active workerso establish their status as a reproductive (H. sublaevis, Bourke,
∗ Corresponding author at: School of Psychology, University of Ottawa, Ottawa,N K1N 6N5, Canada. Tel.: +1 613 562 5800x4849.
1988b; R. marginata, Premnath et al., 1995). Similar suggestionsare made with respect to bumblebees, specifically that aggressionand characteristic social interactions differentiate a reproductiveworker from a non-reproductive worker. This is based on the find-ing that social interactions and aggression coincide with ovariandevelopment (Bombus terrestris, Alaux et al., 2004a; Bloch et al.,1996; Bombus bifarius, Free, 1955; Bombus bimaculatus, Pomeroy,1981; B. terrestris, Duchateau, 1989; van Doorn, 1989; B. bifarius,Foster et al., 2004).
The relationship between ovarian development and aggressionmay not be as clear in Bombus impatiens compared to other speciesof Bombus. Although B. impatiens are similar to other species ofbumblebee, e.g. similar patterns of worker ovarian developmentcompared to B. terrestris workers (Cnaani et al., 2002), there aredifferences, e.g. a proportionally smaller number of reproductiveworkers in a colony (Cnaani et al., 2002). Additionally, in a recentstudy (Sibbald and Plowright, 2013) in which the behaviour of pairsof orphaned workers (i.e. queenless workers) was monitored 5–11days after the bees were placed together, the more aggressive beein a pair did not lay significantly more eggs than the less aggres-sive one. Moreover, the bee that laid more eggs in a pair did not
show more aggression than the bee that laid fewer eggs. This maysuggest that under conditions of unlimited access to food, aggres-sion may be ineffective at suppressing reproduction in another orit serves another purpose in worker pairs. In other eusocial species,
or example, aggression is used to encourage foraging and otherest duties (Lamba et al., 2008; Clarke and Faulkes, 2001), is a nec-ssary behaviour to encourage the ovarian development in itselfLamba et al., 2007) or reflects a combination of functions (Clarkend Faulkes, 2001; Premnath et al., 1995). Conversely, the lack ofn association between aggression and reproductive suppressionn B. impatiens pairs may be indicative of other variables that have
more predominant role in bumblebee pairs.Social interactions were not measured in our previous study
Sibbald and Plowright, 2013) on aggression and reproduction inairs of orphaned bumblebees. As workers were randomly paired,
t is likely that bees of differing degrees of social activity wereaired together. The reproductively dominant queen rarely retreatsrom social interactions (van Honk and Hogeweg, 1981) and asuch these social interactions may be a component of reproductiveompetition in workers. In the present study, in addition to obtain-ng measures of aggression and reproduction, we recorded socialnteractions as soon as workers were paired to determine whetherhese interactions could serve as an early indicator of reproductiveonflict. Moreover, we harnessed this variable by experimentallyssigning workers to groups depending on their levels of socialctivity. If social interactions serve to mitigate future aggression, i.e.o resolve behavioural contests, one possible outcome of this studyould be that when workers are unequal in their levels of socially
ctivity (as opposed to both being socially active or inactive) thenuture aggression would be reduced. We used the definition ofsocial activities’ (Kardile and Gadagkar, 2003; van Doorn, 1989; vanoorn and Heringa, 1986; van Honk and Hogeweg, 1981), whichonsisted of contacts between workers which were measured at
time typically before aggression (e.g. head-butting) and repro-uction. Topographically, the social activities were distinguishablerom aggressive and reproductive behaviours.
The presence of brood is predicted to play a significant rolen reproductive competition. Based on the research, however, itselationship is mixed. Brood interactions and feeding, for exam-le, are positively associated with oviposition (Foster et al., 2004)nd ovarian development (Duchateau and Velthuis, 1989). Aggres-ion is also more likely to occur in areas of the nest where therood are located (such as in Leptothorax allardycei, Cole, 1988).evertheless, these previous studies were correlational and there-
ore specific conclusions regarding causation cannot be made. Inn experimental manipulation of the presence versus absence ofrood, in contrast, aggression and oviposition were more likelyo co-occur in pairs placed without brood compared to pairs withrood (Sibbald and Plowright, 2013). This suggests that brood pres-nce inhibits aggression that occurs during egg-laying. Conditionshat further induce aggression (i.e. pairing two socially active beesogether) may accentuate the effect of brood.
This study was performed on a common North American speciesf bumblebee, B. impatiens, to address three main questions: (1)oes pairing orphaned worker bumblebees according to their levelsf social activity affect their levels of aggression? (2) Does it alsoffect their ovarian development? (3) Does the presence of brood,hen experimentally manipulated, promote or inhibit aggression
nd ovarian development?
. Methods
.1. Colonies
Two commercial B. impatiens colonies supplied by BioBest,
iological Systems and three B. impatiens colonies derived fromild-caught queens and reared according to the procedure of
lowright and Jay (1966) were used. All colonies were maintainedn the laboratory under identical conditions and in compliance
ral Processes 103 (2014) 150–155 151
with institutional guidelines regarding the ethical care for animals.The workers were collected from their colonies between July andDecember, 2008.
All bees were marked by glueing coloured, numbered tags(Opalit Plättchen, The Bee Works, Orillia, ON) onto the thorax.The groups were supplied with food ad libitum: 50% honey-watersolution and pollen collected by honeybees that is mixed withhoney-water solution to form an uniform dough. Pollen is necessaryfor ovarian development (Duchateau and Velthuis, 1989).
2.2. Target workers
Target workers (the objects of study) were 118 B. impatiensorphaned workers from the five queen-right colonies. Workers thatwere within 12 h of emergence from their cocoon (defined as ‘cal-low workers’; bees not having full colouration and having curvedwings) were randomly selected. Callow workers were chosen toensure that workers were of comparable age. The callow work-ers were kept in isolation for 12 days to allow enough time forpossible ovarian development (as previous research has found ittakes an average of 11.8 days for single B. impatiens workers tolay eggs (Sibbald, 2007, unpublished data)). Isolated callow work-ers that were later paired have also been used in other researchand found to display behaviours consistent with larger groups andolder bees (Amsalem and Hefetz, 2010). They also develop matureovaries comparable to those callows that were allowed social con-tact (Cnaani et al., 2007).
2.3. Behaviour testers
To manipulate levels of social activity across pairs of targetworkers, it was necessary to assess these levels first. To this end,nine B. impatiens adult workers of unknown age from three queen-right colonies (the two BioBest colonies, and one of the coloniesderived from a wild-caught queen) were randomly selected as‘testers’ of social interactions in the target workers. These testerswere placed singly in nesting boxes. To ensure these testers haddeveloped ovaries they were not used for behaviour testing untileggs were observed in their nest boxes.
2.4. Apparatus
Once the bees were removed from their colonies they werehoused in wooden nest boxes (10.2 cm × 10.2 cm × 5.1 cm) underhumid conditions at 30 ◦C. The box was lined with honeybee waxto allow for a surface for egg-cup building. For the With Broodgroup, eggs were gathered from queen-right colonies and placedin the nesting box. The box was connected to a second nest boxwhere a glass feeder tube was located. Glass plates covered thetop of the boxes to allow for easy observation. The laboratory wasilluminated by a combination of fluorescent light fixtures and nat-ural light. Digital video recordings were made on a JVC Hard DiscCamcorder positioned directly over the nest box.
2.5. Assessment period
After their 12 days of isolation, the target workers’ social activ-ity was assessed. The target workers were placed in a new nestingbox with one of three behaviour tester bees. Their behaviour wasobserved for five minutes and all social interactions and aggressionwere coded. A score of the degree with which the target workers ini-tiated interactions versus retreated from them was calculated and
workers were categorized as either socially active (SA) or sociallyinactive (SI). This was based on van Doorn and Heringa (1986) aswell as van Honk and Hogeweg, 1981 definition of dominant socialinteractions of the bumblebee. Aggressive interactions were rare,
1 aviou
otnwoict
Aitmiapiwawtblitbsw
2
pgpsssgTbdfp
2
i((3Pai(ttstr
52 E.D. Sibbald, C.M.S. Plowright / Beh
ccurring in only 6% of all assessments and were not included inhe calculation. In other words, the definition of social activity doesot include the behaviours that it is being used to predict. The pairas then separated and the same protocol was performed with two
ther behaviour testers. An average of the target workers’ levels ofnteraction was determined. The following formula was used to cal-ulate the target workers’ level of social activity for each behaviourester.
No. of Initiations − RetreatsTotal No. of Interactions
n interaction is defined as antennal contact where the initiat-ng bee is different from the bee that ends (or “retreats” from)he interaction (see the definition of social interactions below for
ore detailed information). Although the term ‘retreat’ from annteraction may have the connotation of being a component ofggression (Bourke, 1988b; van Honk and Hogeweg, 1981), for oururposes the term is used to refer to withdrawing from a social
nteraction. Using the above calculation, for example, if a targetorker initiates 8 interactions and retreats from 0 interactions with
behaviour tester its interaction score with that behaviour testerould be (8 − 0)/8 = +1.00. If a target worker initiates 0 interac-
ions and retreats from 8 interactions, its interaction score woulde −1.00. An average over all three behaviour testers was calcu-
ated and workers were categorized as either SA or SI. An SA workers one who scored above 0, meaning she initiated more interac-ions than she retreated from them. If a worker had a score of 0 orelow, meaning she retreated from social interactions more thanhe initiated them, she was categorized as a socially inactive (SI)orker.
.6. Testing conditions
On the same day as the assessment the target workers wereaired in a new nesting box under one of six conditions; three socialroups and two brood groups. For the social group, each worker wasaired in one of three ways; (1) a socially active worker + anotherocially active worker (SA + SA), (2) a socially active worker + aocially inactive worker (SA + SI), (3) a socially inactive + anotherocially inactive worker (SI + SI). For the brood group, each socialroup was also paired either (A) with brood or (B) without brood.here were 10 pairs per condition (except for the SA + SI, pairs withrood condition where there were 9 pairs as one worker in a pairied) and a total of six conditions (3 social groups × 2 brood groups),or a total of 59 pairs of orphaned workers. Each target worker wasaired with a sister worker.
.7. Procedure
The recordings were started approximately 24 h after pair-ng. The recordings were randomly started either in the morningdefined as anytime between 7.30 h and 10.30 h), afternoon12.30–15.30 h) or evening (17.30–20.30 h) and continued for
h a day for 6 consecutive days, which, based on Sibbald andlowright (2013) and Cnaani et al. (2007), was long enough tollow for measurable changes in ovarian development. Follow-
ng a previously developed, statistically reliable training methodSibbald and Plowright, 2013), five behaviour coders were trainedo identify aggression, social interactions, and oviposition. Four ofhe five were blind to the conditions and all coders analyzed aimilar number of pairs per condition. The frequency and dura-ion of aggression was coded in each pair. Oviposition was alsoecorded.
ral Processes 103 (2014) 150–155
2.8. Measures
2.8.1. AggressionWe used Sibbald and Plowright (2013) definition of aggression,
which was adapted from that of Duchateau (1989). This consistsof two behaviours; butting and grappling. Butting is defined anaccelerated movement of one bee towards the other resulting incontact. Grappling consists of a somersault motion between thetwo individuals which may involve stinging positions in which theaggressor is on the back of the other, dragging the other by herhind legs. Both the frequency and duration of each aggressive actwas measured. This definition of aggression was found to be reliable(Sibbald and Plowright, 2013). Supplementary Content1 and Sup-plementary Content2 contain examples of butting and grapplingbehaviours, respectively.
Supplementary material related to this article can be found,in the online version, at http://dx.doi.org/10.1016/j.beproc.2013.11.012.
2.8.2. Social interactionsThe definition from van Doorn and Heringa (1986) and van
Honk and Hogeweg (1981) was used. An interaction is definedas any encounter where antenna contact is made between twobees and the initiating bee is different from the bee that ends theinteraction. This definition was based on the observation that thequeen rarely retreats from an interaction she initiates (van Honkand Hogeweg, 1981). The frequency of social acts was measured.Supplementary Content3 shows an example of a social interaction.
Supplementary material related to this article can be found,in the online version, at http://dx.doi.org/10.1016/j.beproc.2013.11.012.
2.8.3. Oviposition and ovarian developmentOviposition was identified using Bloch and Hefetz, 1999 defini-
tion; “. . . a worker seen inserting her abdomen into an open egg cupfor a few minutes while moving it up and down, along with typi-cal kicking movement of the hind legs. Oviposition behaviour wasoften preceded by a period of new egg cup construction or by open-ing and preparing an already existing one” (p. 127). This has shownsignificant inter-observer reliability (Sibbald and Plowright, 2013).To measure ovarian development, the ovaries of the target workerswere dissected within 24 h of the final day of testing and weighedon a Mettler-Toledo balance (PB 303). Measuring wet weight ofthe ovaries versus other methods (e.g. mean length of the termi-nal oocytes, Duchateau and Velthuis, 1989) was chosen as it givesa direct measure of the net amount of yolk deposited and has theadvantage of measuring ovaries that are very small and immaturedue to oosorption (Pomeroy, 1981).
2.9. Data analysis
The statistical analyses were performed using SPSS Statistics 20.Due to the substantial positive skew of the frequency distributionfor aggression frequency and duration, results are reported with alogarithmic transformation.
3. Results
There were 1047 h of video recordings analyzed. Although neweggs were present within the nest of 40 of the 59 pairs, ovipositionwas not regularly captured on video (52.5% of pairs that had laideggs) and therefore could not be included in the analysis. Of the
19 pairs that did not lay eggs, most were in the SA + SI social group(With Brood = 6, Without Brood = 4), compared to the SA + SA group(With Brood = 3, Without Brood = 0), and the SI + SI group (WithBrood = 4, Without Brood = 2).
such, a 3 × 2 multivariate analysis of variance was performed. Theoverall F test was not significant, showing the variance in aggressionduration, aggression frequency, and ovarian mass is not explainedby differences in social activity (Wilks’ � = 0.93, F (3, 32) = 0.76,
0.00
0.03
0.06
0.09
0.12
0.15
SA+SA SA+SI SI+SI
Mea
n O
vary
Mas
s (g
)
Without Brood With Brood
ig. 1. Mean aggression duration (log transformed) of orphaned B. impatiens work-rs for each social group with standard error bars. SA: Socially Active; SI: Sociallynactive.
As not all testers were from the same colony as the target work-rs, a logistic regression was performed to determine whetherhe classification of a target worker as socially active or inactiveepends on how many of the three behaviour testers were from theame colony as the target worker (in which case target and testerorkers were relatives) as opposed to another colony (i.e. target
nd testers were unrelated). The four possibilities were: all threef the testers were from the same colony as the target (n = 18), onlywo were (n = 6), only one was (n = 8), or none were (n = 86). How
any of the testers were related to the target was not an importantariable, �2 (1, N = 118) = 0.97, p = 0.33. This is consistent with otheresearch which has found that related worker pairs did not signifi-antly differ from unrelated pairs in levels of aggression or ovarianevelopment (Amsalem and Hefetz, 2010). The categorization pro-edure used to differentiate socially active from socially inactiveorkers was reliable at differentiating the groups as socially activeorkers had significantly higher scores (M = 0.40, SD = 0.29) than
he socially inactive workers (M = −0.47, SD = 0.34), t (116) = 15.20, < 0.0001.
To determine the effect of social pairing and brood presencen levels of aggression and ovarian mass, a 3 × 2 multivariatenalysis of variance was performed. The results are presentedelow. Additionally, a second 3 × 2 multivariate analysis of vari-nce was performed to determine if socially active workers differrom socially inactive workers on levels of aggression and ovarianevelopment.
.1. Pair differences
Figs. 1 and 2 show an effect of social pairing on aggres-ion duration (F (2, 116) = 10.87, p < 0.0001, partial �2 = 0.17)nd aggression frequency (F (2, 116) = 13.48, p < 0.0001, par-ial �2 = 0.20), respectively, Wilks’ � = 0.79. On both aggressionuration and frequency, post hoc comparisons using a Tukey’sSD test revealed the SA + SA groups were significantly moreggressive (Mduration = 0.62, SEduration = 0.07; Mfrequency = 0.57,Efrequency = 0.06) compared to the SA + SI (Mduration = 0.19,Eduration = 0.07; Mfrequency = 0.16, SEfrequency = 0.06) and theI + SI groups (Mduration = 0.20, SEduration = 0.07; Mfrequency = 0.17,Efrequency = 0.06), p < 0.05. There was no significant differenceetween the SA + SI and SI + SI groups, which may reflect a
oor effect. The groups without brood were significantly moreggressive as measured by aggression duration (F (1, 116) = 3.99,
= 0.048) but not aggression frequency (F (1, 116) = 1.64, p = 0.20),ilks’ � = 0.85. The interaction between brood presence and social
Fig. 2. Mean aggression frequency (log transformed) of orphaned B. impatiens work-ers for each social group with standard error bars. SA: Socially Active; SI: SociallyInactive.
pairing was not significant for aggressive duration or frequency,Wilks’ � = 0.93 (F (2, 116) = 1.70, p = 0.19 and F (2, 116) = 1.76,p = 0.18, respectively).
Fig. 3 shows that the results on ovarian mass did not paral-lel those on aggression duration (Fig. 1) and aggression frequency(Fig. 2); no effect of social group pairing was detected, F (2,116) = 0.46, p = 0.63. In all three social groups, however, ovar-ian mass was larger when the pairs were placed without brood(M = 0.11, SE = 0.006) compared to when they were placed withbrood (M = 0.08, SE = 0.004), F (1, 116) = 13.03, p < 0.0001, partial�2 = 0.11. While the effect of brood presence seemed more pro-nounced in the SI + SI pairs, no interaction between social groupingand brood presence or absence was detected (F (2, 116) = 1.98,p = 0.14).
3.2. Differences within pairs
As the SA + SI pairs are heterogeneous with regards to socialactivity prior to pairing, we performed an analysis to determineif aggression and ovarian development differed within the pair. As
Social Group
Fig. 3. Mean ovarian mass of orphaned B. impatiens workers for each social groupwith standard error bars. SA: Socially Active; SI: Socially Inactive.
1 aviou
pTW
4
lBcapps2rtghagaaftarrecwd2t
siwrnadalStb(tptweuIasew(
bii1
54 E.D. Sibbald, C.M.S. Plowright / Beh
= 0.53). The SA bees were not more aggressive than the SI bees.he overall F test was also not significant for the presence of brood,ilks’ � = 0.90, F (3, 32) = 1.17, p = 0.34.
. Discussion
Through controlling social activity pairing we were able to iso-ate its separate effects on aggression and ovarian development. In. terrestris, characteristic social interactions in a worker are asso-iated with her aggressive behaviour (Bloch et al., 1996; van Honknd Hogeweg, 1981; van Honk et al., 1981). Similarly, in this study,airs of socially active bees were significantly more aggressive thanairs with one or two socially inactive bees. In B. terrestris, the mostocially dominant bee has more developed ovaries (Alaux et al.,004a,b) and aggression is believed to be used in this ‘battle’ overeproduction, with the most successful aggressor laying the eggs. Inhe socially active (SA + SA) pairs, aggression was the highest, sug-esting increased competition between these bees. This aggression,owever, did not translate into heightened reproduction as sociallyctive pairs did not have significantly larger ovaries than the otherroups. However, these results do not conclusively suggest thatggression has no role in reproduction. If there are two sociallyctive workers who are taking part in a reproductive competition,or example, they both may be effective at suppressing reproduc-ion in the other which translates to reduced ovarian developmentt the pair level. A slightly different view is that the aggression mayepresent a competition to determine who is the first to lay eggsather than who monopolizes egg-laying. The production of maleggs typically occurs during the end of the bumblebee colony lifeycle (Bloch and Hefetz, 1999) and as such the earlier laid male eggsould have a better chance of being successfully reared before theeath of the colony and the end of the season (Amsalem and Hefetz,011). Either way, our results show that aggression is affected byhe characteristics of a bee’s nestmate.
It is curious that there was no significant difference betweenocially active workers and their socially inactive partners, eithern their ovarian development or in their levels aggression. Here,
e suggest possible explanations in addition to the usual cautionegarding interpretation of null results. If social activity is a sig-al of a bee’s potential to ‘win’ the reproductive battle, SI bees aret a disadvantage compared to the SA bees and therefore are pre-icted to remove themselves from the competition. The sociallyctive bees however did not have more developed ovaries. Theack of a significant difference in ovarian development within theA + SI pairs may be due to the fact that there are only two bees inhe group. Although orphaned pairs have been shown to maintainehavioural interactions consistent with larger orphaned groupsAmsalem and Hefetz, 2010), larger groups of queen-less B. impa-iens have bigger differences in ovarian development compared toairs, who are more evenly matched (Cnaani et al., 2007). Addi-ionally, there is a positive correlation in egg-laying within pairs oforkers; the more one worker lays eggs, the more her partner lays
ggs (Sibbald and Plowright, 2013). This may mean that, at leastnder paired conditions, both workers can successfully rear eggs.
t also suggests that having a partner may serve an activating role innother worker’s ovaries rather than a suppressive one (as has beenuggested in other insects, Lamba et al., 2007). Indeed, single work-rs have significantly reduced ovarian development compared toorkers placed in groups and even a worker placed with a queen
Cnaani et al., 2002).Ovarian development was greater in the pairs placed in nesting
oxes without brood. The presence of brood therefore may have annhibitory effect on reproduction. A similar association is also foundn honeybees (Jay, 1972) as well as in ponerine ants (Heinze et al.,996). Such an inhibiting effect may be due to a type of pheromone
ral Processes 103 (2014) 150–155
release (Jay, 1972) or alternatively, brood presence may cue rearingbehaviours which interrupt reproduction. Indeed, when workersare removed from a colony the remaining workers take on morefeeding responsibilities (Pendrel and Plowright, 1981). As thereare only two workers placed with brood there would be increasedfeeding responsibilities compared to larger groups. Future researchwith larger queen-less groups would help to clarify this associationbetween brood presence and reproduction as it may reduce theincreased feeding responsibilities placed on the individual work-ers. Nevertheless, as brood absence increases both aggression andovarian development, these findings suggest these two variablesare associated in some way or there is a third variable that affectsboth (e.g. hormonal activity, Pomeroy, 1981).
Male production by Bombus workers is an important issue tostudy as worker reproduction is suggested to have significantconsequences on queen behaviour, such as the development ofmechanisms used by the queen to control worker reproduction(Bourke, 1988a; Keller and Chapuisat, 1999) as well as other char-acteristics such as aggression, matricidal behaviours, and temporaldivision of labour (Bourke, 1988a). Worker reproduction likely alsohas evolutionary significance for Bombus as workers make a signif-icant contribution to the number of males produced in a colony,for example, in B. atratus up to 90% of males produced in thecolony are worker-derived (as cited in Bourke, 1988a; Owen andPlowright, 1982), although there is species variation (Alaux et al.,2004b; Brown et al., 2003; Owen and Plowright, 1982; Paxton et al.,2001).
This study is distinctive from other research as the social condi-tions were experimentally manipulated (through pairing workerswith sisters of varying degrees of social activity). Environmen-tal conditions were also manipulated through placing pairs eitherwith or without brood. Experimental manipulation of reproduc-tive competition within bumblebee research is the exception ratherthan the rule, with most research focusing on correlational dataonly (e.g. Bloch et al., 1996; Foster et al., 2004; van Doorn andHeringa, 1986). Additionally, exceptionally long periods of obser-vation (∼1000 h) ensured that the behaviours observed were likelyrepresentative of the interactions that occurred within each pair.In terms of ecological validity, pairs of orphaned workers may berare, though small queenless groups undoubtedly occur in natureeven if they are inconspicuous (Owen et al., 1980). The findings ofthis study emphasize the complex nature of aggression and repro-duction within worker B. impatiens. The absence of brood increasedboth aggression and ovarian development in pairs, which suggeststhe latter two variables are associated in some way. The findingshowever cannot make specific conclusions about the characteris-tics of the relationship and cannot exclude the possibility that thereis a third variable that may affect both. While the pairs of sociallyactive workers showed more aggression than the other pairs, noneof the social groups showed higher or lower ovarian development.As such, aggression is not necessary for ovarian development ashas been suggested in other insects (Lamba et al., 2007). Furtherresearch with larger groups examining the relationship betweensocial interactions and ovarian development in worker B. impatienscan help clarify the relationship in this species of bumblebee. Nev-ertheless, the findings of the present study underscore that ovariandevelopment can be predicted by certain environmental conditions(e.g. the absence of larvae) and although social interactions canbe observed prior to aggression, it does not differentiate levels ofovarian development between workers.
Acknowledgements
Portions of these results were presented at the Canadian Soci-ety for Brain, Behaviour and Cognitive Science 2009 19th AnnualMeeting, York, UK and the Canadian Society for Brain, Behaviour
aviou
aStasVeHa
R
A
A
A
A
B
B
B
B
B
C
C
C
CD
D
E.D. Sibbald, C.M.S. Plowright / Beh
nd Cognitive Science 2010 20th Annual Meeting, Halifax, Novacotia. The research was funded by a research grant to C.M.S.P. fromhe Natural Sciences and Engineering Research Council of Canadand a graduate scholarship to E.D.S. from the Fonds de rechercheur la nature et les technologies du Québec program. We thankeronika Huta and Dwayne Schindler for advice on statistics, Lev-nte Orbán for editorial comments, and Andrew Faulkner, Jessicaarrison, Arne Stinchcombe, and Jane Zhang for their technicalssistance.
eferences
laux, C., Jaisson, P., Hefetz, A., 2004a. Queen influence on worker reproduction inbumblebees (Bombus terrestris) colonies. Insect. Soc. 51, 287–293.
laux, C., Savarit, F., Jaisson, P., Hefetz, A., 2004b. Does the queen win it all? Queen-worker conflict over male production in the bumblebee, Bombus terrestris.Naturwissenschaften 91, 400–403.
msalem, E., Hefetz, A., 2010. The appeasement effect of sterility signaling in dom-inance contests among Bombus terrestris workers. Behav. Ecol. Sociobiol. 64,1685–1694.
msalem, E., Hefetz, A., 2011. The effect of group size on the interplay betweendominance and reproduction in Bombus terrestris. Plos One 6, e18238.
loch, G., Borst, D.W., Huang, Z.-Y., Robinson, G.E., Hefetz, A., 1996. Effects of socialconditions on juvenile hormone mediated reproductive development in Bombusterrestris workers. Physiol. Entomol. 21, 257–267.
loch, G., Hefetz, A., 1999. Regulation of reproduction by dominant workers inbumblebee (Bombus terrestris) queenright colonies. Behav. Ecol. Sociobiol. 45,125–135.
ourke, A.F.G., 1988a. Worker reproduction in the higher eusocial Hymenoptera. Q.Rev. Biol. 63, 291–311.
ourke, A.F.G., 1988b. Dominance orders, worker reproduction, and queen-workerconflict in the slave-making ant Harpagoxenus sublaevis. Behav. Ecol. Sociobiol.23, 323–333.
rown, M.J.F., Schmid-Hempel, R., Schmid-Hempel, P., 2003. Queen-controlled sexratios and worker reproduction in the bumble bee Bombus hypnorum, as revealedby microsatellites. Mol. Ecol. 12, 1599–1605.
larke, F.M., Faulkes, C.G., 2001. Intracolony aggression in the eusocial naked mole-rat, Heterocephalus glaber. Anim. Behav. 61, 311–324.
naani, J., Schmid-Hempel, R., Schmidt, J.O., 2002. Colony development, larval devel-opment and worker reproduction in Bombus impatiens Cresson. Insect. Soc. 49,164–170.
naani, J., Wong, A., Thomson, J.D., 2007. Effect of group size on ovarian develop-ment in bumblebee workers (Hymenoptera: Apidae: Bombus). Entomol. Gen. 29,305–314.
ole, B.J., 1988. Escalation of aggression in Leptothorax ants. Insect. Soc. 35, 198–205.uchateau, M.J., 1989. Agonistic behaviours in colonies of the bumblebee Bombus
terrestris. J. Ethol. 7, 141–151.uchateau, M.J., Velthuis, H.H.W., 1989. Ovarian development and egg laying in
workers of Bombus terrestris. Entomol. Exp. Appl. 51, 199–213.
ral Processes 103 (2014) 150–155 155
Foster, R.L., Brunskill, A., Verdirame, D., O’Donnell, S., 2004. Reproductive physiol-ogy, dominance interactions, and division of labour among bumble bee workers.Physiol. Entomol. 29, 327–334.
Free, J.B., 1955. The behaviour of egg-laying workers of bumblebee colonies. Br. J.Anim. Behav. 3, 147–153.
Heinze, J., Trunzer, B., Oliveira, P.S., Hölldobler, B., 1996. Regulation of reproductionin the neotropical ponerine ant, Pachycondyla villosa. J. Insect Behav. 9, 441–450.
Heinze, J., 2008. Hierarchy length in orphaned colonies of the ant Temnothoraxnylanderi. Naturwissenschaften 95, 757–760.
Jay, S.C., 1972. Ovary development of worker honeybees when separated fromworker brood by various methods. Can. J. Zool. 50, 661–664.
Kardile, S.P., Gadagkar, R., 2003. Regulation of worker activity in the primitivelyeusocial wasp Ropalidia cyathiformis. Behaviour 140, 1219–1234.
Keller, L., Chapuisat, M., 1999. Cooperation among selfish individuals in insect soci-eties. BioScience 49, 899–909.
Lamba, S., Kazi, Y.C., Deshpande, S., Natesh, M., Bhadra, A., Gadagkar, R., 2007. Apossible novel function of dominance behaviour in queen-less colonies of theprimitively eusocial wasp Ropalidia marginata. Behav. Processes 74, 351–356.
Lamba, S., Chandrasekhar, K., Gadagkar, R., 2008. Signaling hunger throughaggression – the regulation of foraging in a primitively eusocial wasp. Natur-wissenschaften 95, 677–680.
Owen, R.E., Plowright, R.C., 1982. Worker-queen conflict and male parentage inbumble bees. Behav. Ecol. Sociobiol. 11, 91–99.
Owen, R.E., Rodd, F.H., Plowright, R.C., 1980. Sex ratios in bumble bee colonies:complications due to orphaning? Behav. Ecol. Sociobiol. 7, 287–291.
Paxton, R.J., Thoren, P.A., Estoups, A., Tengo, J., 2001. Queen-worker conflict overmale production and the sex ratio in a facultatively polyandrous bumble-bee, Bombus hypnorum: the consequences of nest usurpation. Mol. Ecol. 10,2489–2498.
Plowright, R.C., Jay, S.C., 1966. Rearing bumble bee colonies in captivity. J. Apic. Res.5, 155–165.
Pomeroy, N., 1981. Reproductive dominance interactions and colony developmentin bumble bees (Bombus Latreille; Hymenoptera: Apidae). Doctoral dissertation,University of Toronto, pp. 3981 (Diss. Abstr. 42).
Premnath, S., Sinha, A., Gadagkar, R., 1995. Regulation of worker activity in a primi-tively eusocial wasp, Ropalidia marginata. Behav. Ecol. 6, 117–123.
Sibbald, E.D., Plowright, C.M.S., 2013. On the relationship between aggression andreproduction in pairs of orphaned worker bumblebees (Bombus impatiens).Insect. Soc. 60, 23–30.
Sladen, F.W.L., 1989. The Humble-bee. Logaston Press, Woonton, Herefordshire.van Doorn, A., 1989. Factors influencing dominance behaviour in queenless bum-
blebee workers (Bombus terrestris). Physiol. Entomol. 14, 211–221.van Doorn, A., 1987. Investigations into the regulation of dominance behaviour and
of the division of labour in bumblebee colonies (Bombus terrestris). Neth. J. Zool.37, 255–276.
van Doorn, A., Heringa, J., 1986. The ontogeny of a dominance hierarchy in coloniesof the bumblebee Bombus terrestris (Hymenoptera, Apidae). Insect. Soc. 33, 3–25.
van Honk, C., Hogeweg, P., 1981. The ontology of the social structure in a captiveBombus terrestris colony. Behav. Ecol. Sociobiol. 9, 111–119.
van Honk, C.G.J., Röseler, P.F., Velthuis, H.H.W., Hoogeveen, J.C., 1981. Factors influ-encing the egg laying of workers in a captive Bombus terrestris colony. Behav.Ecol. Sociobiol. 9, 9–14.
