Reproductive plasticity in bumblebee workers (Bombus terrestris) - Reversion from fertility to sterility under queen influence

Abstract

Worker sterility in the bumblebee Bombus terrestris is conditional and is linked to the social development of the colony. Workers refrain from reproducing or overtly challenging
the queen until gyne production has initiated, at the so-called competition point (CP). It is not known whether this behavior
is hard-wired or workers show reproductive plasticity. It also remains unclear whether worker reproductive decision is under
queen and/or worker control. In this study, we tested worker reproductive plasticity in an attempt to assess whether and under
which conditions worker sterility/fertility are reversible. We introduced egg-laying workers into colonies with different
social structures for 1week then monitored their reproductive status. We revealed a remarkable reproductive plasticity in
the introduced workers that was social-condition-dependent. In the presence of a pre-CP queen, the introduced workers reverted
to sterility, whereas in the presence of a post-CP queen, such workers remained egg-layer. Reversion to sterility does not
occur when direct contact with the queen is prevented, as the introduced workers remained egg-layer in the queenright colonies
with a confined queen. Egg-laying workers that were introduced into queenless colonies mostly maintained their fertility regardless
of colony social phase. This shows that worker transition from cooperative to selfish behavior is reversible depending on
the social context.

Full text

Plasticity of worker reproductive strategies in Bombus terrestris:
lessons from artificial mixed-species colonies
CEDRIC ALAUX*,ABRAHAMHEFETZ†&PIERREJAISSON*
*Laboratoire d’Ethologie Expe
´rimentale et Compare
´e, Universite
´Paris 13
yG. S. Wise Faculty of Life Sciences, Department of Zoology, Tel Aviv University
(Received 15 November 2005; initial acceptance 1 March 2006;
final acceptance 12 May 2006; published online 16 October 2006; MS. number: A10300)
We used the experimental paradigm of artificial mixed colonies of two phylogenetically related bumblebee
species to analyse the dynamics of the reproductive skew in societies of Bombus terrestris. Artificial mixed-
species colonies were set up by introducing callow B. terrestris workers either into a queenright (QR) or
a queenless (QL) colony of B. lapidarius. The introduced B. terrestris workers were well integrated into their
host B. lapidarius colony and displayed nesting activities that did not differ from those of the resident
B. lapidarius workers. However, the introduced B. terrestris workers did show a different reproductive behav-
iour. While B. lapidarius workers did not develop ovaries in a B. lapidarius QR colony but did so in a
B. lapidarius QL group, adopted B. terrestris workers in a B. lapidarius QR colony developed ovaries as if
they were under QL conditions. These results indicate that, in mixed-species colonies, B. terrestris workers
are irresponsive to the queen’s inhibitive action on ovary development. In QL homospecific and hetero-
specific predominately B. terrestris mixed-worker colonies (1Bl þ5Bt), reproduction was dominated by a sin-
gle B. terrestris worker, whereas in QR B. lapidarius or QL equally mixed-worker colonies (3Bl þ3Bt), almost
all B. terrestris workers developed ovaries. We suggest that in the presence of enough heterospecific workers,
B. terrestris workers behave as parasites. This last finding suggests that worker reproduction in B. terrestris is
highly plastic and that the experimental paradigm of artificial mixed colonies may provide new insights
into the evolution of social parasitism in this taxon.
Ó2006 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
Reproductive skew is a hallmark of insect societies, where
female reproduction is highly biased in favour of one or
more individuals, the queens, whereas nonreproductive
individuals, the workers, are helpers. In many species,
however, workers are not irreversibly sterile and remain
able to lay viable eggs if they are orphaned, or at certain
stages of colony development. Reproductive plasticity in
workers is highly contextual and depends on the
underlying genetic and social structures of the colony
(Bourke & Franks 1995; Crozier & Pamilo 1996). More-
over, it is governed by two seemingly opposed selection
pressures: selfishness, driven by combined individual
and kin selection; and cooperativeness, driven by
combined kin- and colony-level selections. Thus, unravel-
ling the ultimate and proximate mechanisms for the evo-
lution and maintenance of such a reproductive skew offers
a major topic of interest.
Caste-specific pheromones, particularly queen phero-
mones, are believed to regulate reproductive skew in social
insects. They may operate either by actively inhibiting
worker ovary development (Butler & Fairey 1963; Ho
¨lldo-
bler & Bartz 1985; Hoover et al. 2003) or by acting as an
honest fertility signal (Keller & Nonacs 1993; Endler
et al. 2004; Dietemann et al. 2005). In the latter case,
workers are hypothesized to use all available information
to adjust their behaviour in a way that will maximize their
fitness. This further explains why worker sterility is plastic
and conditional. Workers, therefore, are selected to en-
hance overall colony growth and refrain from reproducing
as long as the genetic gain through inclusive fitness sur-
passes that obtained through direct fitness. It is conse-
quently adaptive for both workers and the queen to
evolve mechanisms for queen recognition, namely
queen-specific signals, which may not necessarily be
species specific.
Correspondence and present address: C. Alaux, Department of Ento-
mology, University of Illinois, Urbana, IL 61801, U.S.A. (email:
calaux@life.uiuc.edu). A. Hefetz is at the G. S. Wise Faculty of Life
Sciences, Department of Zoology, Tel Aviv University, 69978 Tel Aviv,
Israel. P. Jaisson is at the Laboratoire d’Ethologie Expe
´rimentale
et Compare
´e, CNRS UMR 7153, Universite
´Paris 13, 93430 Villeta-
neuse, France.
1417
0003e3472/06/$30.00/0 Ó2006 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
AN IM AL BE HA VI OU R , 2006, 72, 1417e1425
doi:10.1016/j.anbehav.2006.05.008

The bumblebee Bombus terrestris has become a model
system for testing both ultimate and proximate processes
related to reproductive social conflicts (van Honk et al.
1981; Duchateau & Velthuis 1988; Ro
¨seler & van Honk
1990; Bloch et al. 1996; Bloch 1999; Bloch & Hefetz
1999; Alaux et al. 2004a, 2005; Lopez-Vaamonde et al.
2004a). Colonies of B. terrestris are annual with one, singly
inseminated queen (Estoup et al. 1995; Schmid-Hempel &
Schmid-Hempel 2000), creating a potential conflict
between the queen and workers over male production.
Colony life cycle follows two main phases (Alford 1975).
The first, the social phase, is characterized by cooperation,
which maximizes ergonomic growth and gyne production
as well as reproductive self-restraint among workers,
despite their potential to lay viable haploid eggs (Alaux
et al. 2004a). The regulation of worker reproduction is
mediated by the perception of nonvolatile queen phero-
mones via direct antennal contacts. Workers also establish
a dominance hierarchy among themselves during this
phase (van Honk & Hogeweg 1981; van Doorn & Heringa
1986). However, this reproductive dominance hierarchy
among workers is highly flexible and can change through-
out colony development. The second phase marks the
breakdown of social structure and is characterized by an
accelerated competition between the queen and the
workers over reproduction and among workers for access
to reproduction (Bloch & Hefetz 1999). Queeneworker
conflict over male production in the bumblebee B. terrest-
ris is thus well evidenced.
Recent molecular studies, however, have revealed that
despite the overt competition between the queen and her
workers over male production during the competition
phase, the queen seems to produce over 95% of the adult
males (Alaux et al. 2004b; Lopez-Vaamonde et al. 2004b).
The queen is able to successfully outcompete the workers
owing to her higher reproductive output, and greater ag-
gression and oophagy capacity. Worker reproduction
seems to succeed only when the queen dies prematurely,
suggesting that workers may in fact execute the queen
in order to reproduce (Bourke 1994; Alaux et al. 2004b).
Finally, the selfish interest of workers may be expressed
when they invade an alien conspecific colony, where
they can unrestrainedly lay eggs (Lopez-Vaamonde et al.
2004b). All these insights into queeneworker conflict in
B. terrestris have raised the hypothesis that worker repro-
duction is highly plastic and may depend on the social
circumstances. Workers are thus able to show two oppo-
site phenotypes, from cooperation to selfishness, depend-
ing on the dominant selection level. However, unravelling
the proximate mechanisms that result in either worker
sterility or their fertility is sometimes complicated under
natural conditions. Thus, the use of mixed-species groups
might be considered as potentially revealing, since each
species may express its preprogrammed traits, whether
kin-selected or not, independently of colony-level selec-
tion. Moreover, it enables investigation of the level of
reproductive plasticity of workers as well as examination
of the effects of social environment on reproductive
plasticity.
For this purpose, we used a system of mixed-species
colonies to unravel the mechanisms and flexibility level of
the reproductive strategies that may be shown by worker
B. terrestris, but that remain hidden from the researcher
under normal colony conditions (Jaisson 2005). We cre-
ated mixed-species colonies and groups using B. terrestris
and B. lapidarius, two phylogenetically closely related spe-
cies (Kawakita et al. 2004). Although reproductive skew
and queeneworker conflict have been less studied in
B. lapidarius (Free et al. 1969), their phylogenetic proxi-
mity and similarity of life histories suggested that their
social behaviour might also be comparable. Thus, we first
tested whether worker B. lapidarius refrain from reproduc-
tion under queenright (QR), but not queenless (QL) condi-
tions, and whether the queen has a pheromone regulating
worker reproduction. We then investigated whether
worker B. terrestris can integrate into a B. lapidarius colony,
and whether reproductive strategy of B. terrestris workers
changed in heterospecific colonies compared to homospe-
cific colonies. Finally, we created QL mixed-species groups
to further probe into worker reproductive strategies.
MATERIAL AND METHODS
Bumblebee Rearing
Queens of B. lapidarius were collected in the field near
Paris in early spring 2003 and 2004. They were allowed
to found a colony in the laboratory by placing each one
in a wooden box (17.5 26 15 cm) kept in the dark at
28 2C and 50% relative humidity. They were fed ad li-
bitum with sugar syrup and fresh pollen. The experiments
started once these colonies had reached 10 workers (i.e.
about 1 week after the first worker emergence). Observa-
tions were performed under red light through a glass
covering the nestbox.
Colonies of B. terrestris were obtained either from a com-
mercial supplier (GTICO SARL, Foissy sur Vannes, France)
or from our laboratory stock (colonies reared from queens
obtained from commercial colonies). All colonies were
reared under the same standard conditions.
Regulation of Worker Reproduction in
Homospecific B. lapidarius Colonies
To test whether queens of B. lapidarius produce a volatile
pheromone that inhibits worker reproduction, we split
nests into two compartments separated by a double-
mesh (1-cm-wide) screen. One compartment included all
colony workers and the queen (QRC) while the second
compartment contained five callow (1-day-old) nestmate
workers (QLC). No physical contact was possible between
these separated workers and the rest of the colony, includ-
ing the queen. QL groups of five callow workers housed
separately served as control. Both the QL and the QLC
were supplied with young cocoons (collected from young
donor colonies) as a substrate for cell construction (van
Doorn 1987). We recorded the occurrence of newly sealed
egg-cells in the QLC and QL colonies, and compared ovary
development (mean length of the terminal oocytes)
between workers from the QLC, QR and QL colonies.
Control workers, originating from nonmanipulated QR
ANIMAL BEHAVIOUR, 72,6
1418

colonies 20 days after the first worker emergence, were
also frozen for determination of ovary development (at
this colony age workers have had the time required to
develop ovaries; Free et al. 1969).
Introduction of B. terrestris Callow Workers
into B. lapidarius QR Colonies
To test whether a putative B. lapidarius queen phero-
mone inhibits ovary development in adopted young
callow B. terrestris workers, we introduced five B. terrestris
workers that had been individually marked using num-
bered tags (Opalith Pla
¨ttchen, Friedrich Wienold,
Germany) into a QR B. lapidarius colony (N¼17). The
introduced B. terrestris were callow workers that had
been allowed to emerge in homospecific QL colonies to
prevent contact with the conspecific queen. Five control
callow B. lapidarius workers were equally marked and
then reintroduced into their colony. In those colonies in
which there were insufficient B. lapidarius callow workers
at the start of experiment, callow workers from other col-
onies were introduced.
To verify whether the introduced B. terrestris workers
had fully integrated into their host colony, we performed
scan-sampling behavioural observations on seven ran-
domly selected QR mixed colonies every 2 min for
30 min twice a day (mornings and afternoons during
days 2e6). We noted 16 behavioural acts related to worker
tasks (Cameron 1989): foraging, guarding (at nest en-
trance or on the brood, antennae raised), patrolling (rapid
movements), nest inspection, fanning, immobility
(antennae lowered), self-grooming, scraping wax (workers
recycle the wax of empty cocoons), anchoring (building of
wax area), excavation of empty cocoons, working on
honey pots, feeding larvae, brood incubation, enlarge-
ment of wax cover around the brood, threatening behav-
iour (buzzing), and overt aggression.
Regulation of Worker Reproduction
in Mixed-species Colonies
Queenright artificial mixed-species colonies
Egg laying by B. terrestris workers introduced into QR
B. lapidarius colonies (N¼17) was estimated from the sev-
enth day following introduction (the time needed for
ovary development in QL callow workers), by counting
the number of eggs in the newly sealed egg-cells. Since
B. lapidarius queens lay a significantly greater number of
eggs per cell than do B. terrestris workers at their first egg
laying in QL colonies (Table 1), we could discern to which
species the laid eggs belonged. To ascertain that these were
B. terrestris eggs, a sample (from 2 to 4 eggs of each colo-
ny’s egg-cup) was transferred into small QL B. terrestris
colonies containing young nonreproductive workers and
reared to adulthood. All the emerged adults were B. terrest-
ris males.
Workerequeen interactions were estimated by monitor-
ing the number of workerequeen antennal contacts of the
marked individuals (5 B. terrestris workers þ5B. lapidarius
workers for each observed colony, N¼5) daily for 30 min
(15 min morning and afternoon), from the second day of
worker introduction until the first oviposition by B. terrest-
ris workers occurred. The experiment ended when the first
egg laying by the adopted B. terrestris was observed (B. lap-
idarius workers never laid eggs under these conditions, see
Results).
Thereafter, all marked bees were removed and dissected
to determine ovary development. The five groups in
Table 1. Comparison of the means (ManneWhitney Utest) and ranges (c
2
test) for the number and size of eggs laid by B. terrestris and
B. lapidarius queens and workers in the different experimental situations
MeanSD number
of eggs/cell
Range of number
of eggs/cell
MeanSD egg
size (mm)
Range of egg
size
B. lapidarius queen 8.91.17, N¼16 7e12
QL B. terrestris workers*4.051.40, N¼20 2e7 3.100.11, N¼54 2.75e3.25
QR mixed B. terrestris workers 3.901.30, N¼17 2e6
QL B. lapidarius workers 2.850.1, N¼55 2.67e3.08
3Blþ3Bt colonies 3.150.11, N¼78 2.92e3.42
1Blþ5Bt colonies 3.110.08, N¼57 2.92e3.33
ManneWhitney Utest c
2
test
Mean number of eggs/cell Range of number of eggs/cell
B. lapidarius queen versus QL B. terrestris workers U¼1, P<0.001 c
2
¼33.3, P<0.001
B. lapidarius queen versus QR mixed B. terrestris workers U¼0, P<0.001 c
2
¼33, P<0.001
QL B. terrestris workers versus QR mixed B. terrestris workers U¼170, P¼1c
2
¼3.2, P¼0.73
Mean egg size Range of egg size
QL B. terrestris workers versus QL B. lapidarius workers U¼176.5, P<0.001 c
2
¼71.3, P<0.001
QL B. terrestris workers versus 3Blþ3Bt colonies U¼1628.5, P<0.05 c
2
¼10.63, P¼0.22
QL B. terrestris workers versus 1Blþ5Bt colonies U¼1445, P¼0.69 c
2
¼4.1, P¼0.77
QL B. lapidarius workers versus 3Blþ3Bt colonies U¼90, P<0.001 c
2
¼106.9, P<0.001
QL B. lapidarius workers versus 1Blþ5Bt colonies U¼96, P<0.001 c
2
¼80.24, P<0.001
QL: queenless colonies; QR: queenright colonies.
*Egg characteristics of QL B. terrestris workers correspond to their first oviposition.
ALAUX ET AL.: WORKER REPRODUCTIVE PLASTICITY IN BEES 1419

which some of the introduced B. terrestris (N¼4, 11% of
introduced workers) and B. lapidarius workers (N¼2, 6%
of introduced workers) died during the experiment were
excluded from this analysis.
Queenless artificial mixed-species colonies
We created two types of QL mixed colonies to in-
vestigate the plasticity of worker reproduction under these
conditions and determine the role of the number of
heterospecific workers in the reproductive plasticity. The
first type comprised three B. lapidarius and three B. terrest-
ris workers (called QL 3Bl þ3Bt,N¼20) and the second
was composed of one B. lapidarius and five B. terrestris
workers (called QL 1Bl þ5Bt,N¼12). Control groups con-
sisted of either five B. terrestris (N¼12) or five B. lapidarius
workers (N¼12), each group of which was housed sepa-
rately. Once the first egg-cell was constructed all workers
were frozen for later dissection.
Worker oviposition in homospecific control colonies
showed that egg size of B. lapidarius was significantly
smaller than that of B. terrestris (Table 1). Egg size, there-
fore, served as a reliable measure for determining egg spe-
cies identity.
Behavioural observations were performed on a randomly
chosen sample of each of the QL mixed (3Bl þ3Bt,N¼8;
1Bl þ5Bt,N¼8), homospecific B. terrestris (N¼9) and
homospecific B. lapidarius (N¼8) colonies. They were
performed twice a day (10 min in the morning þ10 min
in the afternoon) during days 2e7, focusing on workere
worker aggression (biting, grappling and head butting,
see Bloch et al. 1996 for details).
Statistics
We used a factorial correspondence analysis (FCA) on
the complete behavioural repertoire (number of items for
each recorded task) applied to the marked bees (Spad 3.1
software, Spadsoft, Paris, France). Bees that died during
the behavioural experiment were not taken into account
in the FCA. Permutation tests by stratum (StatXact 3.1,
Cytel Software, Cambridge, Massachusetts, U.S.A.) were
performed for B. lapidarius (N¼33) and B. terrestris
workers (N¼31) on the resulting coordinates of each indi-
vidual in the FCA (root 1 and 2), taking into account their
mixed-group origin. This enabled us to consider intercolo-
nial variation.
To analyse how ovary development was distributed
among workers, we first determined whether there was
any significant variation between groups within each
treatment (Table 2). Next, we categorized their develop-
ment into five stages representing equal ovary develop-
ment range for each species (B. lapidarius: stage 1: 0e0.54
mm; stage 2: 0.55e1.08 mm; stage 3: 1.09e1.62
mm; stage 4: 1.63e2.16 mm; stage 5: 2.17e2.7 mm;
B. terrestris: stage 1: 0e0.75 mm; stage 2: 0.76e1.5 mm;
stage 3: 1.51e2.25 mm; stage 4: 2.26e3 mm; stage 5:
3.01e3.75 mm), and used a KolmogoroveSmirnov test
to determine whether the resulting distributions deviated
from a normal distribution (Statistica 6.1 software, Stat-
soft, Tulsa, Oklahoma, U.S.A.). In groups that showed
a normal distribution of ovary development, reproduction
was considered as not skewed.
RESULTS
Regulation of Worker Reproduction in
Homospecific B. lapidarius Colonies
The average time to first oviposition by B. lapidarius
workers after group establishment in the QLC of the ho-
mospecific colonies was not significantly different from
that in the control QL colonies (ManneWhitney Utest:
U¼45, N
1
¼9, N
2
¼12, P¼0.522; Table 3). Ovary devel-
opment in workers from both QL (N¼60) and the QLC
colonies (N¼45) was significantly greater than that of
workers from nonmanipulated QR colonies (N¼150)
(KruskalleWallis test: H
2,269
¼116.16, P<0.001; Siegele
Tukey post hoc test: QL/QLC: P¼0.612; QR/QL and QR/
QLC: P<0.001; Table 3). This demonstrates that if a queen
pheromone does affect worker reproduction, it is nonvol-
atile as in B. terrestris (Alaux et al. 2004a).
Social Integration of B. terrestris Workers
Adopted by B. lapidarius Colonies
In-nest behaviour of the introduced B. terrestris workers
(N¼31) was compared to that of the marked and reintro-
duced B. lapidarius workers (N¼33) during days 2e6 fol-
lowing introduction (FCA on all observed individuals
and considering all the behaviours outlined above;
Fig. 1). The first two roots revealed that individuals of
both species created overlapping clouds and did not differ
significantly according to their behavioural repertoire
(permutation test: root 1: P¼0.116; root 2: P¼0.602).
Thus, at least with respect to the measured parameters,
worker B. terrestris showed an integration into their host
B. lapidarius colony.
Regulation of Worker Reproduction
in Mixed-species Colonies
Queenright artificial mixed-species colonies
In the queenright mixed colonies (N¼17) on a mean
SD of 8.06 1 days after B. terrestris workers were
Table 2. Results of KruskaleWallis tests on the mean size of worker
ovary development under the different social regimes
Hdf P
QLC homospecific B. lapidarius 1.49 8, 45 0.992
QL B. terrestris workers 3.84 11, 60 0.974
QL B. lapidarius workers 6.40 11, 60 0.845
QR mixed B. terrestris workers 13.54 11, 60 0.259
QR mixed B. lapidarius workers 10.14 1, 60 0.517
QL B. terrestris workers (3Blþ3Bt) 13.98 19, 60 0.78
QL B. lapidarius workers (3Blþ3Bt) 16.24 19, 60 0.64
QL B. terrestris workers (1Blþ5Bt) 1.37 11, 60 0.99
QLC: queenless compartment of nest; QL: queenless colonies; QR:
queenright colonies.
ANIMAL BEHAVIOUR, 72,6
1420

introduced we observed some egg-cells containing fewer
eggs than expected from those normally laid by B. lapida-
rius queens (Table 1). Egg numbers in these cells were not
significantly different from those of QL B. terrestris colo-
nies, indicating that they might have originated from
B. terrestris workers. To verify this assumption, we reared
a subset of these eggs in a young QL colony composed
of nonreproducing B. terrestris workers. All the brood de-
veloped exclusively into B. terrestris males. It thus seems
that the adopted B. terrestris workers had started to ovi-
posit after the expected physiological time-lag required
for ovary development in QL callow workers, and that
the B. lapidarius queen had failed to inhibit heterospecific
worker reproduction. The time required to achieve repro-
duction for B. terrestris workers did not differ significantly
between social regimes (KruskalleWallis test: H
5,82
¼5.99,
P¼0.306; Table 3). In contrast, resident B. lapidarius
workers in these colonies, whether reintroduced or non-
manipulated, did not reproduce in the presence of the
queen, whereas they started to lay eggs after a mean SD
of 8.08 1.04 days when reared in QL colonies (Table 3).
Behavioural observations on the B. terrestris bees re-
vealed that they had a significantly higher rate of contact
with the B. lapidarius queen than the B. lapidarius workers
had with their mother queen (1.23 0.25, N¼25 versus
0.68 0.4 antennal contacts per worker per 30 min,
N¼25; ManneWhitney Utest: U¼148.5, P<0.01).
Moreover, antennal contact rates between B. terrestris
workers and the B. lapidarius queen were not significantly
different from those shown by prospective reproductive
B. terrestris workers with their own queen, but were higher
than the rate shown by nonreproductive workers (data in
Alaux et al. 2004a; ManneWhitney Utest: U¼699.5,
N
1
¼25, N
2
¼60, P¼0.626 and U¼1082.5, N
1
¼25,
N
3
¼312, P<0.001, respectively).
Worker B. lapidarius from QR mixed colonies or from QR
homospecific colonies had undeveloped ovaries, as op-
posed to QL workers, which had significantly larger termi-
nal oocytes (ManneWhitney Utest: QR mixed versus QR
homospecific workers: U¼4292, N
1
¼60, N
2
¼150,
P¼0.601; QR mixed versus QL workers: U¼417.5,
N
1
¼60, N
3
¼60, P<0.001; Table 3). Interestingly, the
average ovary development of B. terrestris workers in QR
B. lapidarius colonies was greater than that of workers
kept in homospecific QL colonies (ManneWhitney Utest:
U¼743.5, N
1
¼60, N
2
¼60, P<0.001; Table 3). This dif-
ference resulted from the fact that in homospecific QL col-
onies only one bee had fully developed ovaries, whereas in
the QR mixed colonies, all the adopted B. terrestris workers
had developed ovaries (Fig. 2a). Thus, ovary development
was not skewed as expected from the normal reproductive
strategy shown by QL B. terrestris workers.
Queenless artificial mixed-species colonies
Egg size was used to compare the pattern of worker
reproduction for B. terrestris and B. lapidarius in QL mixed-
species groups. The mean egg size in mixed-species
QL groups 1Bl þ5Bt differed significantly from those of
QL B. lapidarius colonies but not from those of the QL
B. terrestris colonies (Table 1). In mixed-species QL groups
3Bl þ3Bt, egg size differed significantly from both QL ho-
mospecific colonies. However, in QL colonies 3Bl þ3Bt,
egg size was closer to that of QL B. terrestris colonies than
to that of QL B. lapidarius colonies. Moreover, the range
Table 3. Reproductive timing and ovary development (mean SD) of callow workers from the different colony types
Colony type Worker type Ovary development (mm) Nworkers
Days until
egg laying by the
first worker Ncolonies
QR homospecific B. lapidarius 0.150.42 150 d6
QLC homospecific B. lapidarius 1.570.57 45 8.441.17 9
QL homospecific B. lapidarius 1.360.83 60 8.081.04 12
QL homospecific B. terrestris 0.871.12 60 7.580.62 12
QR mixed colonies Marked B. lapidarius 0.160.36 60 d17
Introduced B. terrestris 2.130.82 60 8.061
QL mixed colonies (3Blþ3Bt)B. lapidarius 1.180.84 60 7.80.6 20
B. terrestris 2.060.87 60
QL mixed colonies (1Blþ5Bt)B. lapidarius 0.80.8 12 d12
B. terrestris 1.071.13 60 7.580.76
QR: queenright colonies; QLC: queenless compartment of nest; QL: queenless colonies.
–1
–0.5
0
0.5
1
–1 –0.5 0 0.5 1
Root 1 (20.6%)
Root 2 (13.5%)
B. lapidarius workers
B. terrestris workers
Figure 1. Distribution of workers of B. lapidarius (N¼33 workers)
and B. terrestris (N¼31) colonies as a function of their behavioural
pattern obtained from scan sampling. Each point represents an indi-
vidual on the first two axes of the factorial correspondence analysis
of behavioural data. Workers that died before the end of the obser-
vations were not considered for analysis.
ALAUX ET AL.: WORKER REPRODUCTIVE PLASTICITY IN BEES 1421

of egg sizes in QL mixed colonies (3Bl þ3Bt and 1Bl þ5Bt)
was significantly different from that of QL B. lapidarius
colonies but not from that of QL B. terrestris colonies.
Thus, all the eggs oviposited could be attributed to B. ter-
restris workers in QL mixed colonies 1Bl þ5Bt. Although
some of the eggs in QL mixed colonies 3Bl þ3Bt could
have been laid by B. lapidarius workers, it is very likely
that most of them originated from B. terrestris workers.
Ovary development of B. terrestris workers within the
mixed-species colonies revealed two distribution patterns
(Fig. 2a). When they were either adopted by a QR B. lapi-
darius colony (QR mixed) or reared with an equal number
of B. lapidarius workers (QL mixed 3Bl þ3Bt), most of
them had developed ovaries and mean oocyte size distri-
bution did not differ significantly from normal distribu-
tion. In contrast, when reared either in pure B. terrestris
QL groups or in the mixed colonies where they were the
majority (QL 1Bl þ5Bt), ovary development was highly
skewed, with only one dominant bee with developed ova-
ries. The same analysis for B. lapidarius revealed that in
both the pure QL colonies and the mixed-species group
(QL mixed 3Bl þ3Bt) most workers had similar non-
skewed ovary development (Fig. 2b). In contrast, workers
from the QR colonies did not develop ovaries. Ovary
development in the single B. lapidarius workers from
the 1Bl þ5Bt QL mixed colonies was scattered (range
Stage 1 Stage 2 Stage 3 Stage 4 Stage 5
Stage 1 Stage 2 Stage 3 Stage 4 Stage 5
0
5
10
15
20
25
30
35
40
QR mixed D=0.16, P=0.081
QL mixed (3Bl+3Bt)
D
=0.151,
P
=0.117
QL mixed (1Bl+5Bt)
D
=0.225,
P
<0.01
QL
D
=0.262,
P
<0.001
(a) B. terrestris
Ovar
y
develo
p
ment (mm)
0
10
20
30
40
50
60
Worker number
QR mixed D=0.456, P<0.001
QL mixed (3Bl+3Bt) D=0.168, P=0.061
QL D=0.103, P=0.515
(b) B. lapidarius
Figure 2. Ovary development distribution of B. terrestris and B. lapidarius workers from the different mixed and homospecific colonies (N¼60
workers for each condition). (a) B. terrestris workers. (b) B. lapidarius workers. Dand Pvalues based on KolmogoroveSmirnov test of normality
are reported for each condition. A significant Pvalue means that the ovary development distribution was skewed.
ANIMAL BEHAVIOUR, 72,6
1422

0e1.99 mm) and had a low mean SD oocyte length
(0.8 0.8 mm, N¼12; Table 3).
Ovary development was in concordance with the results
of the behavioural observations. In the homospecific QL
colonies of B. terrestris, agonistic behaviour was performed
by a single worker, and in B. lapidarius colonies the dom-
inant worker displayed 85.7 11.5% of the aggression. In
the QL mixed 3Bl þ3Bt colonies (N¼8 colonies), intra-
specific aggression reached 21 8.8% for B. terrestris and
10.85 9.2% for B. lapidarius of total aggression (each of
them displayed by only one individual). Interspecific ag-
gression in these groups constituted 68.15 11.4% of all
aggression, of which 90.5 14.2% was performed by a sin-
gle B. terrestris worker and 9.5 14.2% was displayed by
a single B. lapidarius worker in each colony. In the QL
mixed 1Bl þ5Bt colonies, aggression was exclusively dis-
played by B. terrestris workers.
Thus, the reproductive plasticity shown by B. terrestris
workers changed accordingly to the composition of mixed
colonies. The low reproductive skew observed between
B. terrestris workers was not due to the simple detection
of heterospecific workers but rather to their number.
DISCUSSION
Worker reproduction is governed by two selection pro-
cesses. Individual selection drives workers to behave self-
ishly since they gain maximum fitness by rearing sons
(Bourke & Franks 1995; Crozier & Pamilo 1996). Such self-
ishness may cause considerable colony deterioration to
the point that optimal rearing of sexuals is hampered.
However, colony-level selection can nullify this selfish-
ness, resulting in worker sterility. In bumblebees, colony-
level selection is very powerful since genetic gain from
future gynes is greater than that from future sons, forcing
workers to ensure that gyne production has been initiated
before they attempt to reproduce (Alaux et al. 2005). Thus,
worker reproduction is conditional and dependent on
social context (see Introduction). We constructed mixed-
species colonies and groups of two phylogenetically re-
lated bumblebees, B. terrestris and B. lapidarius, to assess
the level of this reproductive plasticity and the mecha-
nisms implied. This kind of experimental paradigm has
been used successfully in ants to examine the nature of
both the template and label involved in nestmate recogni-
tion (reviewed by Lenoir et al. 2001; Errard et al. 2005;
Jaisson 2005).
We first established that B. lapidarius workers do not re-
produce under queenright conditions and that, if this in-
hibition is pheromonally mediated, the pheromone used
is nonvolatile. Thus, worker B. lapidarius seem to behave
like B. terrestris workers, in which direct antennal contact
with the queen is required for complete reproductive inhi-
bition to occur (Alaux et al. 2004a).
Bombus terrestris workers were observed to be integrated
into the queenright colonies of B. lapidarius into which
they had been introduced: they received no aggression
from the resident bees and behaved as if they were in their
native colony. This was confirmed by the factorial analysis
applied on multiple behaviours, which showed that
worker B. terrestris contributed as much to B. lapidarius col-
ony maintenance and growth, including brood care, as
did the conspecific workers. In contrast to the resident
B. lapidarius workers, however, they did not refrain from
reproducing and behaved as if queenless by developing
ovaries within about 8 days (i.e. the time necessary for
a dominant worker to develop ovaries under queenless
conditions; Alaux et al. 2004a). We exclude the possibility
of an artefact related to the introduction event, because
introduced B. lapidarius, whether resident or alien, did
not develop ovaries in the presence of the queen. We
also exclude the possibility that the lack of inhibition un-
der our experimental conditions was due to insufficient
contacts between the queen and B. terrestris workers (as-
suming, for example, that B. lapidarius queens produced
a nonvolatile pheromone that might have an interspecific
inhibitory effect). The B. lapidarius queens were in fact an-
tennated at rates that were not different from those per-
formed by prospective reproductive workers towards
their mother B. terrestris queens before the competition
phase (Alaux et al. 2004a). We conclude that the B. lapida-
rius queen pheromone is ineffective on B. terrestris
workers, whether or not they perceive it. However, be-
cause the introduced B. terrestris workers did not behave
entirely as alien intruders, it is possible that two phero-
mones are involved: a queen attractant/recognition that
might be non-species-specific (Vienne et al. 1998) and
a species-specific ovary-inhibiting pheromone. In wasps,
contrasting with our results, Ishay et al. (1986) showed
that reproductive Vespula germanica workers reared in a col-
ony of Vespa orientalis were inhibited, suggesting the
role of a non-species-specific pheromone or of inhibitory
physical aggression.
Our findings also offer new insights regarding the
evolution of queen control or queen signal (Keller & Non-
acs 1993). The complexity (West-Eberhard 1981) as well as
the species diversity of queen pheromones may reflect the
evolution of an unstable arms race between the two castes
with regard to inhibition of worker reproduction. On the
other hand, an honest queen signal is quite stable and
thus queen pheromones should vary at a lower rate be-
tween species. The lack of control of the queen B. lapida-
rius over the reproduction of B. terrestris workers tends to
favour the first hypothesis.
Perhaps the most striking result obtained from the
artificial mixed colonies was the change in the reproduc-
tive strategy of B. terrestris workers. Generally, queenless
workers of B. terrestris compete aggressively for reproduc-
tion, a conflict that is solved by the worker dominance hi-
erarchy in which only one worker in a group develops
ovaries (Bloch et al. 1996; Alaux et al. 2004a). Workere
worker inhibition is also apparent when callow workers
are introduced into a queenright colony that is well into
the competition phase (Bloch & Hefetz 1999). In contrast,
in our mixed-species groups, all the introduced B. terrestris
had similar ovary development, whether they constituted
a minority group (only 5 workers introduced into a whole
queenright B. lapidarius colony) or were in groups with an
equal number of heterospecific workers (3Bl þ3Bt). In
both cases it appears that the B. terrestris workers did not
compete among themselves but developed ovaries
ALAUX ET AL.: WORKER REPRODUCTIVE PLASTICITY IN BEES 1423

independently of each other, at their own developmental
rates. One possible explanation is that, being a minority
group in a large host colony, B. terrestris workers did not
encounter each other at sufficient rates to support domi-
nance hierarchy establishment. This, however, cannot ex-
plain why all three B. terrestris workers developed ovaries
when housed in small mixed groups with equal numbers
of worker B. lapidarius, in contrast with the dominance
hierarchy established between three B. terrestris workers
housed together (Bloch et al. 1996). Alternatively, we sug-
gest that the change in reproductive strategy by B. terrestris
workers might represent exploitation, to the extent of par-
asitism, of heterospecific workers. This conclusion agrees
with the result obtained in groups in which B. terrestris
constituted the majority and in which a clear dominance
hierarchy was established. This is also consistent with an
earlier report that B. terrestris workers show intraspecific
social parasitism by laying male eggs in conspecific colo-
nies (Lopez-Vaamonde et al. 2004b). Mixed-species colo-
nies could also occur through queen usurpation of
pre-emergent homo- and/or heterospecific host colonies.
Queens generally usurp another nest during the founda-
tion phase, but can also do so in very young colonies com-
posed of the first batch of workers (Alford 1975). In this
case, the patterns of nest usurpation (i.e. choice of hetero-
specific or homospecific hosts) would be constrained by
whether or not usurper queens can inhibit worker repro-
duction in host species. That would explain why nest
usurpation involving two subgenera has never been found
(Hobbs 1965).
These results demonstrate that worker B. terrestris have
a plastic and context-dependent reproductive strategy. In
homospecific colonies and queenless groups, workers
compete aggressively between themselves for access to re-
production. The conflict is solved by establishing repro-
ductive dominance, which eventually results in group
stability and allows the successful rearing of brood. Kin-
selected interests are in this case surpassed by colony-level
selection because success in brood rearing requires worker
cooperation; the subordinate workers benefit because they
gain inclusive fitness from rearing nephews. However,
when B. terrestris workers were housed in heterospecific
colonies they seemed to switch to interspecific competi-
tion, so brood rearing may be fully supported by the het-
erospecific workers, and since the host queen is incapable
of controlling their reproduction, the B. terrestris workers
may be able to fully express their selfish interest to rear
sons. Their behaviour strikingly resembles that of the par-
asitic clone of Apis mellifera capensis workers, which acti-
vate their ovaries in queenright host colonies of Apis
mellifera scutellata (Martin et al. 2002; Neumann & Moritz
2002), as well as that of the anarchistic honeybee (a phe-
notype of Apis mellifera), where some workers display
a rare phenotype by developing ovaries despite the
queen’s presence (Oldroyd et al. 1994; Hoover et al. 2005).
Another remarkable observation was the lack of a clear
reproductive dominance between B. lapidarius workers
whether in homospecific or mixed-species colonies
compared to B. terrestris workers. This pattern of reproduc-
tion raises an interesting question. Is the lack of reproduc-
tive dominance due to an incomplete control of the
dominant worker or to a share of reproduction between
individuals?
The results presented here demonstrate the advantages
of the mixed-species colonies paradigm. It not only
discloses selfish behaviour that may be hidden when
observing a homospecific group, but it also may shed
light on the evolution of social parasitism: for example,
how Psithyrus overcomes hostequeen inhibition and in-
hibits ovary development of host workers (Fisher 1984;
Vergara et al. 2003).
Acknowledgments
This work was funded by a J. and M. L. Dufrenoy grant
(Acade
´mie d’Agriculture de France) to C.A. We are grateful
to Paul Devienne for technical assistance, to Rumsa€
ıs
Blatrix and two anonymous referees for helpful com-
ments, and to Naomi Paz for her editorial assistance. The
experiments comply with the current laws of France.
References
Alaux, C., Jaisson, P. & Hefetz, A. 2004a. Queen influence on
worker reproduction in bumblebees (Bombus terrestris) colonies.
Insectes Sociaux,51, 287e293.
Alaux, C., Savarit, F., Jaisson, P. & Hefetz, A. 2004b. Does the
queen win it all? Queeneworker conflict over male production
in the bumblebee, Bombus terrestris.Naturwissenschaften,91,
400e403.
Alaux, C., Jaisson, P. & Hefetz, A. 2005. Reproductive decision-
making in semelparous colonies of the bumblebee Bombus terrest-
ris.Behavioral Ecology and Sociobiology,59, 270e277.
Alford, D. V. 1975. Bumblebees. London: DavisePoynter.
Bloch, G. 1999. Regulation of queeneworker conflict in bumblebee
(Bombus terrestris) colonies. Proceedings of the Royal Society of
London, Series B,266, 2465e2469.
Bloch, G. & Hefetz, A. 1999. Regulation of reproduction by domi-
nant workers in bumblebee (Bombus terrestris) queenright
colonies. Behavioral Ecology and Sociobiology,45, 125e135.
Bloch, G., Borst, D. W., Huang, Z. Y., Robinson, G. E. & Hefetz, A.
1996. Effects of social conditions on juvenile hormone mediated
reproductive development in Bombus terrestris workers. Physiolog-
ical Entomology,21, 257e267.
Bourke, A. F. G. 1994. Worker matricide in social bees and wasps.
Journal of Theoretical Biology,157, 292.
Bourke, A. F. G. & Franks, N. R. 1995. Social Evolution in Ants.
Princeton, New Jersey: Princeton University Press.
Butler, C. G. & Fairey, E. M. 1963. The role of the queen in prevent-
ing oogenesis in worker honey bees. Journal of Apiculture Research,
2,14e18.
Cameron, S. A. 1989. Temporal patterns of division of labor among
workers in the primitively eusocial bumble bee Bombus griseocollis
(Hymenoptera: Apidae). Ethology,80, 137e151.
Crozier, R. H. & Pamilo, P. 1996. Evolution of Social Insect Colonies.
Oxford: Oxford University Press.
Dietemann, V., Peeters, C. & Ho
¨lldobler, B. 2005. Role of the
queen in regulating reproduction in the bulldog ant Myrmecia
gulosa: control or signalling? Animal Behaviour,69, 777e784.
van Doorn, A. 1987. Investigations into the regulation of domi-
nance behaviour and the division of labour in bumblebee colo-
nies (Bombus terrestris). Netherlands Journal of Zoology,37,
255e276.
ANIMAL BEHAVIOUR, 72,6
1424

van Doorn, A. & Heringa, J. 1986. The ontogeny of a dominance
hierarchy in colonies of the bumblebee Bombus terrestris (Hyme-
noptera, Apidae). Insectes Sociaux,33,3e25.
Duchateau, M. J. & Velthuis, H. H. W. 1988. Development and re-
productive strategies in Bombus terrestris colonies. Behaviour,107,
186e207.
Endler, A., Liebig, J., Schmitt, T., Parker, J. E., Jones, G. R.,
Schreier, P. & Ho
¨lldobler, B. 2004. Surface hydrocarbons of
queen eggs regulate worker reproduction in a social insect.
Proceedings of the National Academy of Sciences, U.S.A.,101,
2945e2950.
Errard, C., Hefetz, A. & Jaisson, P. 2005. Social discrimination tun-
ing in ants: template formation and chemical similarity. Behavioral
Ecology and Sociobiology,59, 353e363.
Estoup, A., Scholl, A., Pouvreau, A. & Solignac, M. 1995. Monoan-
dry and polyandry in bumble bees (Hymenoptera; Bombinae) as
evidenced by highly variable microsatellites. Molecular Ecology,
4,89e93.
Fisher, R. M. 1984. Dominance by a bumble bee social parasite
(Psithyrus citrinus) over workers of its host (Bombus impatiens).
Animal Behaviour,32, 304e305.
Free, J. B., Weinberg, I. & Whiten, A. 1969. The egg-eating behav-
iour of Bombus lapidarius L. Behaviour,35, 313e317.
Hobbs, G. A. 1965. Ecology of species of Bombus Latr. (Hymenop-
tera: Apidae) in southern Alberta: III. Subgenus Cullumonobombus.
Canadian Entomologist,97, 1293e1302.
Ho
¨lldobler, B. & Bartz, S. H. 1985. Sociobiology of reproduction in
ants. In: Experimental Behavioral Ecology and Sociobiology (Ed. by
B. Ho
¨lldobler & M. Lindauer), pp. 237e258. Sunderland, Massa-
chusetts: Sinauer.
van Honk, C. G. J. & Hogeweg, P. 1981. The ontogeny of the social
structure in a captive Bombus terrestris colony. Behavioral Ecology
and Sociobiology,9, 111e119.
van Honk, C. G. J., Ro
¨seler, P. F., Velthuis, H. H. W. & Hoogeven,
J. C. 1981. Factors influencing the egg laying of workers in a cap-
tive Bombus terrestris colony. Behavioral Ecology and Sociobiology,
9,9e14.
Hoover, S. E. R., Keeling, C. I., Winston, M. L. & Slessor, K. N.
2003. The effect of queen pheromones on worker honey bee
ovary development. Naturwissenschaften,90, 477e480.
Hoover, S. E. R., Winston, M. L. & Oldroyd, B. P. 2005. Retinue
attraction and ovary activation: responses of wild type and anar-
chistic honey bees (Apis mellifera) to queen and brood phero-
mones. Behavioral Ecology and Sociobiology,59, 278e284.
Ishay, J., Rosenzweig, E. & Pechhaker, H. 1986. Comb building by
worker groups of Vespa crabo L., V. orientalis L. and Paravespula
germanica Fabr. (Hymenoptera: Vespinae). Monitore Zoologico
Italiano,20,31e51.
Jaisson, P. 2005. Kinship and fellowship in ants and social wasps.
In: Kin Recognition. 2nd edn (Ed. by P. G. Hepper), pp. 60e93.
Cambridge: Cambridge University Press.
Kawakita, A., Sota, T., Ito, M., Ascher, J. S., Tanaka, H., Kato,
M. & Roubikf, D. W. 2004. Phylogeny, historical biogeography,
and character evolution in bumble bees (Bombus: Apidae) based
on simultaneous analysis of three nuclear gene sequences. Molec-
ular Phylogenetics and Evolution,31, 799e804.
Keller, L. & Nonacs, P. 1993. The role of queen pheromones in
social insects: queen control or queen signal? Animal Behaviour,
45, 787e794.
Lenoir, A., D’Ettorre, P., Errard, C. & Hefetz, A. 2001. Chemical
ecology and social parasitism in ants. Annual Review of Entomology,
46, 573e599.
Lopez-Vaamonde, C., Koning, J. W., Jordan, W. C. & Bourke,
A. F. G. 2004a. A test of information use by reproductive bumble-
bee workers. Animal Behaviour,68, 611e618.
Lopez-Vaamonde, C., Koning, J. W., Brown, R. M., Jordan,
W. C. & Bourke, A. F. G. 2004b. Social parasitism by male-pro-
ducing reproductive workers in a eusocial insect. Nature,430,
557e560.
Martin, S. J., Beekman, M., Wossler, T. C. & Ratnieks, F. L. W.
2002. Parasitic Cape honeybee workers, Apis mellifera capensis,
evade policing. Nature,415, 163e165.
Neumann, P. & Moritz, R. F. A. 2002. The Cape honeybee phe-
nomenon: the sympatric evolution of a social parasite in real
time? Behavioral Ecology and Sociobiology,52, 271e281.
Oldroyd, B. P., Smolenski, A. J., Cornuet, J.-M. & Crozier, R. H.
1994. Anarchy in the beehive: a failure of worker policing in Apis
mellifera.Nature,371, 749.
Ro
¨seler, P. F. & van Honk, C. G. J. 1990. Castes and reproduction in
bumblebees. In: Social Insects: an Evolutionary Approach to Castes
and Reproduction (Ed. by W. Engels), pp. 147e166. Berlin:
Springer-Verlag.
Schmid-Hempel, R. & Schmid-Hempel, P. 2000. Female mating
frequencies in Bombus sp. from central Europe. Insectes Sociaux,
47,36e41.
Vergara, C. H., Schro
¨der, S., Almanza, M. T. & Wittmann, D.
2003. Suppression of ovarian development of Bombus terrestris
workers by B. terrestris queens, Psithyrus vestalis and Psithyrus
bohemicus females. Apidologie,34, 563e568.
Vienne, C., Errard, C. & Lenoir, A. 1998. Influence of the queen
on worker behaviour and queen recognition behaviour in ants.
Ethology,104, 431e446.
West-Eberhard, M. J. 1981. Intragroup selection and the evolution
of insect societies. In: Natural Selection and Social Behaviour (Ed. by
R. D. Alexander & D. W. Tinkle), pp. 3e17. New York: Chiron
Press.
ALAUX ET AL.: WORKER REPRODUCTIVE PLASTICITY IN BEES 1425