of A picultural Research 5(3): 155 165 (1966)
REARING BUMBLE BEE COLONIES IN CAPTIVITY
R. C. Plowright AND S. C. Jay
Department of Entomology, University
of Manitoba, Winnipeg, Canada
Manuscript received for publication
16th June 1966
methods were used to induce the following Bombus species to
start colonies in captivity: B. nevadensis, B. rufocinctus,
B. borealis, B. fervidus, B. terricola, B. perplexus, B.
ternarius, B. vagans. The first method, in which waxed-paper
containers were used, was tried only on queens which had not
hibernated; although it was unsuitable for rearing colonies,
queens of B. rufocinctus, B. terricola, B. perplexus and B.
ternarius laid eggs. The other two methods were used only on
queens captured in the spring. Successful results were
obtained for B. terricola and B. perplexus, and moderately
successful results for B. rufocinctus, using upholsterer's
cotton as nest material at room temperature. A method using
no nest material, but maintaining the temperature at 29�C.,
was successful for B. terricola and B. ternarius. No queens
of B. fervidus laid eggs in any test but, of the small
numbers of B. nevadensis, B. borealis and B. vagans queens
tested, most reared colonies.
details of rearing the colonies are discussed; the
presentation of fresh pollen seems particularly important.
is much interest in conserving and augmenting wild
populations of bumble bees (Bombus) for pollinatin,g (see
review by Holm, 1966). Some workers (Sladen, 1912; Frison,
1927; Plath, 1923; Hasselrot, 1952; Valle, 1955; Holm, 1960;
Zapletal, 1961) have attempted to culture colonies in the
laboratory; others (e.g. Sladen, 1912; Frison, 1926; Fye
& Medler, 1954; Hobbs, 1960, 1962) set out domiciles for
use by overwintered queens in the field. If it can be
performed on an economic scale, year-round laboratory culture
appears to be the best long-term method for mass rearing. Not
only can optimum conditions far colony growth be provided and
attack by parasites and predators eliminated in the
laboratory; the reproductive rate might be much increased by
rearing several generations of colonies in the course of the
year. Horber (1961) maintained a laboratory culture of B.
hypnorum for five generations and showed that diapause is not
obligatory for queens of ~this species. However, a number of
biological problems (e.g. control of mating, ovary
development, colony initiation, rate of oviposition) remain
to be solved before large-scale production of all
economically important Bombus species can successfully be
undertaken. We have tried to induce the following North
American bumble bees to found colonies in captivity in autumn
1964 and summer 1965: B. nevadensis Cresson, B. rufocinctus
Cresson, B. borealis Kirby, B. fervidus (Fabrieius), B.
terricola Kirby, B. perplexus Cressy, B. ternarius Say, B.
vagans F. Smith. Because this was dane to provide colonies
for experiment, rather than to study colony-founding as such,
numerous modifications in technique were made as the work
progressed, and controls were not always provided.
and males of B. fervidus, B. terricola, B. perplexus and B.
ternarius were reared from colonies collected in the field in
southern Manitoba during Augur 1964. Some were mated in cages
in the laboratory and others in a greenhouse. Some queens of
B. terricola, B. perplexus and B. ternarius became 'broody'
after mating in the greenhouse, and built honey pots near
pollen lumps, but none of the eggs they laid survived more
than a few days. B. terricola queens also secreted wax in a
laboratory mating cage. Most of the queens that did not
become 'broody' were found torpid in piles of Sphagnum on
their cage floor. Unfortunately, we did not distinguish
torpid and 'broody' queens when colony-founding trials were
started in mid-September.
rufocinctus queens and males were reared from colonies
collected in the field,. but because the males were killed by
workers unless removed from their colonies within a few days
of emergence it was necessary to capture about 100 males in
late August and early September in order to provide at least
one male for each queen. Unlike B. terricola and B.
ternarius, B. rufocinctus mated in small containers, and even
in nest boxes exposed to light. So a much smaller mating cage
(2 4 X 1 2 X 1 2 m.) was used for B. rufocinctus than for any
other species. When the cage was dismantled in late
September, all the surviving queens were torpid in piles of
moss. They were stored at 5�C. for four weeks before being
placed in colony-founding containers. All the trials with B.
rufo-cinctus were thus started two months later than for
other species used in Series I, and it was possible to remedy
some of the earlier defects; pollen was changed more
frequently, and the rearing temperature was gradually
increased to 29�.
were put singly or in pairs in a container consisting of two
waxed-paper cartons ($ pint, 0-3 1.), connected by a short
tube and left at room temperature (ca 21�). A lump of pollen
(about 0 75 g.), moistened with honey solution (equal volumes
of honey and water) was put in one carton, which was covered
to exclude light. The other carton contained a plastic
vial-cap, filled with honey solution which was renewed daily.
of B. nevadensis, B. rufocinctus, B. borealis, B. fervidus,
B. terricola, B. perplexus, B. ternarius and B. vagans were
captured in the field in May and early June 1965, and were
either (a) placed in nest boxes, or (b) stored over damp
vermiculite at 1� until needed, or (c) confined at room
temperature with access to food in a plas:tic cage (1-2 X 0 6
X 0 6 m.) if it was necessary to store them for periods not
exceeding 24 hours.
were installed singly or in pairs in a wooden container
consisting of two boxes (7 5 X 7 5 X 5 am. internal
dimensions) with glass roofs and corrugated-paper floors
(Fig. 1) at 29� and 50% R.H. All boxes were kept in total
darkness with minimal disturbance.
solution was provided in glass gravity feeding tubes inserted
through the wall of the outer box. By sterilizing the filled
tubes in a pressure cooker it was found unnecessary to
replace them more often than every third day.
Fro. 1. Box
used for Series II queens
P = pollen
lump G = g!lass cover (cut away)
early trials the pollen lump was placed in the inner box, but
several queens built their egg cells directly on the
cardboard floor of the outer box; in later experiments the
pollen was put in the outer box, next to the feeding tube, so
that the inner chamber was used only for defaecation.
pollen lumps were renewed at least every three days, end more
often if queens persistently fragmented their lumps.
eight trials with single or paired queens of B. terricola and
11 trials with B. ternarius, the inner box contained
cellulose wadding in which a central cavity confined the
about three days after a queen had laid eggs, small pieces of
pollen dough were introduced iso the nest box on alternate
days; these were put as far away from the brood as possible
so &at eggs cells were not built on them.
the development of the first brood, some queens neglected
larvae and/or tore down the wax envelope surrounding them.
Although many of these rejected larvae were underweight for
their instar, they appeared healthy; on seven occasions we
either replaced such larvae in the original envelope or in a
new one made of beeswax and pollen; they were always accepted
by their mother, but were often neglected again later. We
also transferred egg cells to pollen lumps in instances where
they had been constructed elsewhere, but they were always
destroyed or ignored alter introduction.
Frison (1927), an attempt was made to induce broodiness in
ten queens of B. ternarius by confining them together in a
cage with plenty of pollen and honey. After one week the
eight survivors were installed singly in nest boxes, at the
same time as a control series of eight queens brought in from
the field immediately after capture.
queens were obtained and handled in the same way as in Series
wooden boxes (at ca 21�) were used for each bee; the larger
(17 5 X 12 5 X 10 cm. iMernal dimensions) contained the
feeding tube and was lined with corrugated cardboard and had
a glass roof. The smaller chamber (7-5 x 7 5 X 5 cm. internal
dimensions) was lined with upholsterer's cotton (Fig. 2). The
sides and roof of the cavity thus formed were smeared with
honey solution, to simulate
natural situation in which the nest material appears to act as a
temporary food store. A pollen lump (about 0-75 g.) was placed in
&e centre of the cavity.
queen was left undisturbed for the first three days after
installation; the progress of her nesting activity could be
assessed by the appearance and amount of faeces in the
antechamber, since oviposition is always preceded by the
production of large amounts of pollen-containing faeces. If
no eggs had been laid by the third day, a new pollen lump was
supplied, and the nest material changed. This process was
repeated every third day until the queen had laid her first
eggs. After this, pollen was supplied on alternate days in
increasing quantities until the first workers emerged, when
it became necessary to supply pollen every day. When giving
pollen to brooding queens, care was taken not to obstruct her
'brooding-groove'. In ithe earlier trials of this series,
pollen replenishment was undertaken only when the queen had
left her nes~t to feed or defaecate, but it was found that
this precaution was unnecessary. None of the 45 queens which
laid eggs deserted her brood, although four died before the
emergence of their first pupae; three of these broods were
reared by foster-queens.
the first or second group of workers had emerged, some boxes
containing B. terricola and B. perplexus colonies were put
out-of-doors, with the nest box placed inside the antechamber
and the excess space filled with nest material. A wooden lid
covered with polythene sheet was stapled on to provide a
remaining coilonies were kept within the laboratory to
provide material for experiment. As these colonies became
larger, the nest material was removed and the colonies were
kept in darkness at 30�, to facilitate inspection and the
addition of pollen.
colonies, one of B. perplexus and one of B. terricola, were
connected to separate plastic cages (1-2 x 0-6 x 0 6 m.) in
which petri dishes containing honey solution and wood-wool
were placed. Apart from the slight saving in time gained when
feeding honey solution, no advantage could be found in this
method of housing the colonies, since by keeping the colonies
in darkness much smaller containers can be used.
RESULTS AND DISCUSSION
results are summarized in Table 1 (for detailed records see
Plowright, 1966); the two B. terricola queens which were
parasitized by nematodes are not included in this Table.
queen of B. terricola, B. perplexus or B. ternarius which
laid eggs in Series I succeeded in rearing her brood.
However, two B. terricola and one B. ternarius eventually
reared adults after being transferred to nest boxes
containing cellulose wadding. Two of these queens produced
only males, suggesting that they had not mated, or were no
longer able to pass viable sperm.
success was achieved with B. rufocinctus, but since the
temperature was gradually raised during the experiment, the
trials with this species are not comparable with those using B.
terricola, B. perplexus and B. ternarius.
no B. fervidus queen in Series I laid eggs, one became broody
and constnxted an egg cell on the day before she died.
data for B. ternarius in Series IIAB, C, D were
examined by analysis of variance, each queen (or pair of
queens) being scored 1 or 0 aocording to whether or not she
laid eggs within 21 days after installation:
of variance Number of queens per box Presence or absence of nest
material Residual D.F. 1 1 24
queens thus gave significantly hetter results than so.litary
queens. Nest material had no significant effeat; but since four
of the five pairs which oviposited in boxes containing nest
material (Series IID) laid their eggs on the floor of the outcr
compartment rather than within the nest material, ithe provision
of nest material at 29� seems to have served no useful purpose.
greater success obtained by confining B. ternarius queens in
pairs does not necessarily imply that oviposition was
stimulated by social interaction; the improved results may
have arisen from the fact that in boxes containing pairs of
queens, only the queens ithat oviposited first contributed
data for statistical analysis. Thus when the 13 single bees
in Series IIA> and IIB are arranged in 60 sets of six
pairs obtained by drawing random combinations of the numbers
1 13, and the mean incidence of success for each set
of six pairs is calculated (by scoring 1 for any pair
including a queen that laid eggs during the 21-day period,
and 0 for any pair in which neither queen oviposited), the
mean of all 60 sets (0 759) is not significantly different
from the mean (0 857) of the 14 pairs in Series IIC and IID,
but it significantly exceeds (P ( 0 05) the mean (0 462) for
the 13 single bees (Series IIA, IIB) from which the 60 pair
combinations were taken.
mean time intervals between installation and oviposition of
the B. ternarius queens that laid eggs in series IEA, and
IIA, were 7-0 and 32 5 days respectively (P ( 0 01),
indicating that queens confined together in a cage for a week
before installation in nest boxes were less ready to start
colonies than queens put in nest boxes after capture in the
field at the end of the one week period.
the queens in Series II which did not lay eggs within 21
days, three B. ternarius and six B. terricola were dissected;
all except two B. terricola parasitized by nematodes had well
developed ovaries, and in some it appeared that egg
resorption bad occurred (see Medler, 1962). Four randomly
selected unsuccessful B. terricola queens from the same
series were transferred after 21 days to Series III boxes,
where they laid eggs within five days; four controls which
were left in their Series II boxes failed to start colonies
during the Eve-day period and eventually died without
each of 7 pairs of B. terricola and 12 pairs of B. ternarius
which oviposiited in Series IIC and IID, one queen died
within the 21-day test period. The mean time of death was 0
64 days after the first eggs were laid (S.E. = 0-88 days with
on experimental methods
the use of waxed paper cartons at 21� was unfavourable for
brood rearing, many queens of B. terricola, B. ternarius and
B. rufocinctus in Series I laid eggs, suggesting that it will
be feasible to obtain colonies of these species from queens
that have not overwintered, by using improved techniques.
wooden containers used in Series II were convenient for
handling, and tihe corrugated-paper floor covering absorbed
faeces well and was easy to renew. Since most of the queens
used only the inner chamber for defaecation, in all but the
earliest trials the pollen lump was placed near the feeding
tube in the outer chamber.
some modification of the Series II design is required to
prevent queens building their erst egg cells upon the floor.
Those that did so usually built an excessive number of cells,
with the result that the larvae in the earliest celh were
neglected, or more larvae were produced than the queen could
rear unaided. This led to the ejection of some or all of the
larvae, unless the excess were removed by hand or one or two
workers added fram another colony.
other hand, queens which had constructed their first egg
cells upon the pollen lump rarely built more cells upon the
fioor, though the addition of fur.ther pollen often
stimulated renewed oviposition, unless the new pollen was
placed at a distance from the original lump. So it seems that
the covering of the pollen lump wiA egg cells, at least in B.
terricola and B. ternarius, inhibits further egg laying. Some
type of nest maMrial may thus be necessary, even at high
temperatures, if only to ensure that queens build their first
egg cells upon their pollen lumps rather than on the floors
of their boxes; but cellulose wadding seems unsuitable for
to save moribund first broods in Series II by adding workers.
repairing tom larval clumps, transferring larvae and adding
wax, were mostly very successful. We do not, however,
consider that such delicate manipulations would be possible
when dealing with large numbers of colonies, since a great
deal of time is required to carry them out successfully.
design of the Series III boxes is quite similar to that of
Hasselrot (1952). However, Hasselrot (personal communication)
states that with his method great care is required when
adding fresh pollen or inspecting the progress of the brood,
because the queen may otherwise desert her progeny. Though
queens in Series III were frequently inspected while brooding
eggs or young larvae, none deserted her brood; this may have
been associated with the fact that the Series III nest
chambers were much smaller than those Hasselrot used, though
until more work has been done it is not possible to rule out
interspecific differences. Smearing the nest material with
honey appears to have a beneficial effect in inhibiting the
tunnelling activi,ties of the queen, and in retarding the
drying out of the pollen dough.
of queen bumble bees to start vigorous colonies in oaptivity
their failure to oviposit within the duration of the
long delay between confinemen:t and oviposition
failure to rear the first brood
abnormally early production of sexual forms.
only possible by dissection to ascertain whether failure resulted
from lack of development of the queen's ovaries, or from lack of
suitable stimuli for egg laying. Since dissection of the
unparasitized queens that had failed to oviposit in Series II
showed their ovaries to be well developed, and since the
unsuccessful Series II queens quickly started colonies when
transferred to Series III boxes, it seems that unsuitability of
nest box design was the oause here. However, undeveloped ovaries
were found in many queens of these two species which had not
overwintered and which had failed to oviposit in Series II boxes
(Plowright, 1966), even after long periods of confinement, and
Medler (1962) recorded that 15 out af 32 queens of various
species that were confined in laboratory boxes for more than 40
days contained only smal.l (0 1-5 mm.) eggs, i.e. their
ovaries were comparatively undeveloped. It thus appears thai
inadequate ovary development may be the major obstacle to colony
founding in captivity in some circumstances, though there was no
evidence for this in the overwintered queens tested in Series II
(1912) and Free and Butler (1959) reported that oviposition
occurred more often when several queens, or one queen and one
(or more) workers, were confined together than when queens
were in solitary confinement. However, Frison (1927) obtained
a greater (though non-significant) incidence of colony
founding in boxes containing single queens &an in boxes
housing pairs of queens. Our results from B. ternarius
suppart the observations of Sladen and Free, but there is no
evidence that the greater incidence of oviposition in the
Series II boxes containing pairs of queens resuks from
anything other than a bias of data caused by including only
the result from rthe first queen to oviposit within each
pair. Further experiments are necessary to determine whether
oviposition is stimulated or inhibited when B. ternarius
queens are confined in pairs, but oviposi0on was delayed when
ten B. ternarius queens were confined together in a cage for
a week before being placed in nest boxes, at least in
comparison with queens brought in from &e field at the
end of the seven days. Possibly the latter bees were
physio-logically more ready to start colonies than the caged
group because of nutritional differences; alternatively,
social effects amo.ng the caged queens, such as the
establishment af 'dominance' (see Free, 1955) by one queen
over her companions, may have inhibited ovary development in
Ae non-dominant queens.
(1912) found that when pairs of queens were confined
together, oae killed the other at about the time the first
eggs were laid. Sl@den's resuilts are confirmed for B.
terricola and B. ternarius, since the mean time of death for
these queens in the pairs in Series II did not differ
significantly from the time art which the first eggs were
laid. However, in contrast to Sladen's finding, 14 out of the
surviving 19 members of these pairs successfully reared
aduits; other authors (e.g. Frison., 1927; Plath, 1923, 1934;
Free & Butler, 1959) have found that the death of one
queen does not always occur when two are confined together;
in Series I, some of the pairs of B. rufocinctus queens
coexisted until after the first workers emerged.
and Butler (1959) suggested that Hasselrot's (1952) success
in getting colonies started in captivity might be due in part
to the amount of insulation provided, since it had been shown
that ovary development in worker bumble bees was accelerated
by high temperatures. Although no attempt seems to have been
made to compare the incidence of colony founding at high and
low temperatures, Horber (1961) reported Cat though most of
24 queens of B. hypnorum (from which he attempted to get
colonies) started building nests when they were exposed in an
unheated greenhouse, real progress was observed only after
the nests had been moved to a rearing cabinet where the
temperature could be kept above 25� and later between 33�
and 35�. However, Horber's results may not have been
associated with ovary development.
laying of eggs by queens in Series I was always preceded by a
period of 'broodiness', during which the queen flattened
herself over her pollen lump, or on the floor of the
container, at the same time making rhythmic abdominal
movements. Similar behaviour was observed in most of the
Series II queens and in all those of Series III that
were inspected shortly before oviposition. Presumably this
br~diness is amociated width heat production; it is not known
whether it has any connection with the degree of ovary
addition to lack of adequate fresh pollen, it is probable
that chilling of eggs and young larvae contributed towards
the failure of Series I queens to rear their brood. The use
of a high temperature for starting Bombus colonies without
nest material (whether or not it encourages oviposition)
seems justified because it allows survival of eggs and 4rvae
even when the queens fail to incubate them.
of the B. rufocinctus queens in Series I, which started
colonies after long periods of confinement in nest boxes,
produced small colonies which started to rear queens and
males earlier than usual; similar results were obtained with
other species in mme recent trials (Plowright, 1966; see also
Sladen, 1912; Holm A Haas, 1961), so it appears that for the
culture of vigorous colonies, from which it is intended to
obtain a large worker population, iit is necessary that
queens should be induced to oviposit as early as possible.
However, promptness of egg laying cannot be the only factor
involved, for in trials with artificially overwintered B.
terricola queens (Plowright, 1966), all of which oviposited
within ten days of confinement in nest boxes, all colonies
produced males and/or queens in the first or second broods.
that the success obtained in the Series III trials on
overwintered B. terricola queens can be upheld in future
years, and that the method proves suitable for the
large-scale culture of B. borealis and other economically
important species, there seems to be no reason why the
culture of colonies in captivity from niaturally overwintered
queens should not play an important part in providing stocks
of bumble bees for pollination services. However, further
research will probably be needed to develop effective methods
for rearing vigorous colonies from queens that have been
overwintered in captivity, or from queens that have not
overwintered at all.
authors thank Dr. J. B. Free (Rothamsted Experimental
Station, England) for his helpful criticisms of the
manuscript and Mr. L. P. Lefkovitch (Agricultural Research
Council, Slough, England) for his assistance in analysing the
B. (1955) The behaviour of egg-laying workers of bumblebee
colonies. Br.I. Anlm
B. & BurLaR, C. G. (1959) Bumblebees. London: Collins
T. H. (1926) Experiments in attracting queen bumblebees to
econ. Ent. 19: 149 155
Experiments in rearing colonies of bumblebees (Bremidse) in
Bull. mar. biol. Lab., Woods Hole 52: 51 67
E. & Medler, J. T. (1954) Field domiciles for bumblebees. J.
econ. Ent. 47(4):
T. B. (1952) A new method for starting bumblebee colonies.
(1960) Studies on Swedish bumblebees (genus Bombus Latr.): their
Opusc. end. Suppl. 17: 1-192
G. A., Virostek, J. F. & NUMMr, W. O. (1960) Establishment of
Apidae) in artificial domiciles in southern Alberta. Can.
G. A., Nvw�r, W. O. & ViROSTEK, J. F. (1962) Managing
colonies of bumble bees
Apidae) for pollination purposes. Can. Ent. 94(11): 1121
N. (1960) Experiments on the domestication of bumblebees (Bombus
B. lapidarius L. and B. terrestris L. Arsskr. K.
(1966) The utilization and management of bumble bees for red
clover and alfalfa seed
A. Rev. Ent. 11: 155 182
N. & HAzs, H. (1961) Erfahrungen und Resultate dreijahriger
mit Hummeln (Bombus Latr.). Albrecht-Thaer-rfrch. 5(4): 282
E. (1961) Beitrag zur Domestikation der Humrmeln. Vjschr. naturf.
J. T. (1962) Development and absorption of eggs in bumblebees
Can. Ent. 94(8): 825 833
O. E. (1923) Breeding experimen,ts width confined Bremus (Bombus)
mar. biol. Lab., Woods Hole 45: 325 341
Bumblebees and their ways. New York: MacMillan.
R. C. (1966) Domestication and caste differen,tiati6n in
Manitoba: Ph.D. Thesis
F. W. L. (1912) The bumblebee, ih life history and how to
damesticaM it. London;
O. (1955) Untersuchungen zur Sich@rung der Bestaubung van
Seur Julk. 83: 205 220
F. (1961) Ueber die Domestikation der Hummeln. Arch. Gefliigelz.