Content uploaded by David R. Tarpy
Author content
All content in this area was uploaded by David R. Tarpy on Jun 15, 2014
Content may be subject to copyright.
461
Apidologie 36 (2005) 461–474
© INRA/DIB-AGIB/ EDP Sciences, 2005
DOI: 10.1051/apido:2005033
Original article
Three mechanisms of queen elimination in swarming honey
bee colonies1
David C. GILLEY*, David R. TARPY
Department of Neurobiology and Behavior, Cornell University, Ithaca, NY 14853, USA
Received 8 September 2004 – Revised 17 December 2004 – Accepted 31 December 2004
Published online 9 August 2005
Abstract – Queen elimination in honey bee colonies is a process by which all but one of the unmated queens
produced during colony fission (“swarming”) are eliminated from the parental nest. Queens are eliminated
by three mechanisms: queen-queen duels, pre-emergence destruction, and departure with a secondary
swarm. Here we describe each of these mechanisms of elimination and address important questions about
queen elimination using detailed records of events from 13 observation-hive colonies undergoing queen
elimination. We make the following conclusions. (1) The events during queen elimination occur in two
distinct patterns: queen duels with secondary swarm departures, and no queen duels with pre-emergence
destruction of all of the first-emerging queen's rivals. (2) The timing of secondary swarm departure is related
to the events of queen elimination. (3) Queen duels are a common mechanism of elimination. (4) Workers
play a significant but non-lethal role in queen elimination. (5) Queen elimination in European and African
honey bee colonies is similar but the patterns of events may differ.
Swarming / queen elimination / queen duel / queen fight / pre-emergence destruction
1. INTRODUCTION
Swarming is a form of colony reproduction
whereby the parental colony splits into one or
more subunits each containing at least one sex-
ual female (queen) and a fraction of the col-
ony’s workers. Reproduction by colony fission
is widespread among social insects; it occurs in
bees (Apidae: Meliponinae, Apinae; Winston,
1987; van Veen and Sommeijer, 2000), ants
(Formicidae: Aenictinae, Dorylinae, Ecitoninae;
Gotwald, 1995), and wasps (Vespidae: Polisti-
nae; Jeanne, 1991). Swarming of honey bee
(Apis mellifera L.) colonies can be divided
functionally into two stages: queen rearing and
queen elimination (Fig. 1). During queen rear-
ing, a colony’s worker population grows, 15–
20 queens are reared to adulthood inside spe-
cially constructed queen cells, and the primary
swarm, consisting of the mother queen and
approximately half of the colony’s workers,
departs from the nest. During queen elimina-
tion, the daughter queens within the natal nest
emerge and either are killed in fights with rival
queens (also known as “duels”), are killed by
rival queens before emerging from their cells
(also known as “pre-emergence destruction”),
or depart in secondary swarms (also known as
“afterswarms”); the remaining daughter queen
inherits the natal nest. Queen elimination occu-
pies a relatively brief period of a colony’s life
history (usually less than seven days), but the
outcome of the process impacts the inclusive
fitness of all colony members.
* Corresponding author: dgilley@tucson.ars.ag.gov
Current Address: USDA, ARS, Carl Hayden Bee Research Center, Tucson, AZ 85719, USA.
1Manuscript editor: Stan Schneider
Article published by EDP Sciences and available at http://www.edpsciences.org/apidoor http://dx.doi.org/10.1051/apido:2005033
462 D.C. Gilley, D.R. Tarpy
Queen elimination has attracted increasing
attention recently as an example of multi-level
selection (Tarpy et al., 2004), group decision
making (Gilley et al., 2003; Tarpy and Gilley,
2004), potential nepotism (Tarpy and Fletcher,
1998; Gilley, 2003), complex signalling (Gilley,
2001; Schneider et al., 2001; Schneider and
Lewis, 2004), and a mechanism of invasive
species success (Schneider and DeGrandi-
Hoffman, 2003; Schneider et al., 2004). This
recent research has made apparent some gaps
in our knowledge of the events that occur
within swarming honey bee colonies during
queen elimination. Here we seek to advance our
understanding of queen elimination in honey
bee colonies by pointing out those gaps that we
believe are most important and then addressing
each using data from observational studies of
queen elimination. First, however, we summa-
rize what is known about queen elimination in
honey bees by describing the three mechanisms
of queen elimination.
1.1. Queen-queen duels
Lethal queen-queen combat is a common
mechanism of queen elimination in Apis mel-
lifera. Butz and Dietz (1994) described combat
between mated queens that were introduced
into the same colony, but there have been only
a few detailed descriptions of combat between
newly emerged virgin queens (Huber, 1814;
Gilley, 2001; Weaver and Weaver, 1980;
Ohtani, 1994). In the most recent description of
newly emerged queens in a natural context
(Gilley, 2001), queens were observed during
“duels”, i.e., from the moment two emerged
queens were present simultaneously in the hive
until one of these queens was killed2. The
simultaneous presence of more than two
queens was rare in the hives observed by Gilley
(2001), but may be more common in honey bee
subspecies that have been reported to rear large
numbers of virgin queens such as A. mellifera
sicula (Tiemann and Brückner, 1993; Hepburn
and Radloff, 1998). During a duel, queens
spend most of their time moving rapidly
throughout the hive, emitting audible “piping”
sounds, and investigating queen cells, from
which they are usually chased away by work-
ers. Dueling queens do not often come into con-
tact with each other (once every 45 min, on
average), sometimes passing within 1 cm of
Figure 1. Timeline of the important events during reproductive fission (“swarming”) of a European honey
bee colony. The dates of events are typical for a colony in Ithaca, New York (42°26’ N, 76°30’ W), a
temperate climate.
2Because emerged queens often contact each other
multiple times over an extended period, we prefer to
follow Gilley’s (2001) use of the term “duel” to des-
cribe the period of competition between two queens. It
is convenient to have a label for this period and
“duel”, defined as a conflict between two antagonistic
forces, is appropriate assuming that typically only two
emerged queens are present. This usage reserves the
term “fight” (used by many previous authors) for the
contacts between queens that result in grappling and
stinging attempts. Thus, a duel might include several
contacts between queens, each of which can result in
fighting or fleeing. Queens are eliminated by fights
but since the outcome of a fight may depend on other
events that occur during a duel, we consider the entire
duel as a mechanism of elimination.
Mechanisms of queen elimination 463
one another without reacting noticeably. They
may also “hide” by placing themselves within
worker cells for several hours. When queens
contact each other, they either part following
antennal contact, or they grapple together,
using their legs and mandibles to position
themselves to sting their rival. The stimulus
that triggers grappling behavior is unknown,
but is located on the queens’ abdominal tergites
(Pflugfelder and Koeniger, 2003). Grappling
contacts last from 4 s to 15 min and usually
result in the death or wounding of one queen,
but can end without a successful sting by either
queen. Sometimes when rival queens are near
each other, rather than fighting, one queen
points the tip of her abdomen at her opponent
and ejects fecal material onto her (often
referred to as “spraying”; Page and Erickson,
1986; Post et al., 1987; Page et al., 1988;
Bernasconi et al., 1999, 2000; Gilley, 2001;
Tarpy and Fletcher, 2003). Workers interact
extensively with queens during duels (Huber,
1814; Weaver and Weaver, 1980; Ohtani,
1994; Gilley, 2001). The variability in the rates
of these interactions among dueling queens has
led some to investigate whether workers play
a role in selecting which queen will win the
series of duels and inherit the nest. Several
studies have demonstrated non-random out-
comes of queen duels with respect to particular
queen characteristics or behavior, which may
be due in part to actions by the workers. For
example, queens survive with higher probabil-
ities when they are related to the workers in a
hive (Tarpy and Fletcher, 1998; but see Gilley,
2003), when they are older than their opponents
(Tarpy et al., 2000), and when they are more
frequently shaken by workers (Schneider et al.,
2001).
1.2. Pre-emergence destruction
An emerged queen sometimes eliminates
her rivals before they emerge from their cells.
She does this by spending from several minutes
to an hour chewing a 3–5 mm diameter hole in
the side wall of a queen cell and then repeatedly
pushing the tip of her abdomen into the hole,
stinging the cell’s occupant (Huber, 1814;
Boch et al., 1979). Queens destroy cells that are
close to emergence before cells that contain
immature pupae, thereby eliminating their most
dangerous rivals first (Caron and Greve, 1979;
Harano and Obara, 2004). Sometimes, queens
attacking cells do not sting the occupant,
instead departing after having only chewed a
hole in the cell. Stinging occurs most often
when a cell’s occupant has pupated and is ready
to emerge (Huber, 1814; Gilley, unpublished
data). Workers near a cell that is under attack
either ignore the visiting queen or enlarge the
hole that she chews in the wall of the cell
(Huber, 1814; Gilley, unpublished data). The
workers eventually tear down queen cells that
are damaged by emerged queens and dispose of
the queen pupae that occupied them.
1.3. Secondary swarm departure
The departure of a colony’s primary swarm
(i.e., that containing the mother queen) from
observation hives has been described in detail
by a number of investigators (e.g., Huber,
1814; Allen, 1956; Lindauer, 1961; Martin,
1963; Caron, 1970). Secondary swarms typi-
cally depart with fewer workers than primary
swarms, and can contain several young queens
(Huber, 1814; this may be especially common
in African subspecies, Hepburn and Radloff,
1998), but are similar to primary swarms with
respect to the mechanics of the swarming proc-
ess (Gilley, personal observation). Swarming
begins with a rapid running of workers on the
comb near the hive entrance. This behavior
spreads throughout the hive within 3–4 min-
utes, and is followed by a massive rush to the
nest entrance opening (Caron, 1970). The
queen either joins the exodus voluntarily
(Caron, 1970) or is forced out by the workers
(Allen, 1956; Caron, 1970), after which she
may attempt to reenter the hive (Caron, 1970).
The bees fly about near the entrance of the hive
until most of the colony has departed (95% of
workers and drones initially leave the hive,
though many return within one hour; Caron,
1970), and then cluster on a tree branch or other
nearby object to begin choosing a new nest site.
Sometimes a queen does not leave the hive or
fails to join the swarm cluster, in which case the
workers return to the hive in approximately 30
min, usually swarming again the following day
(Allen, 1956; Caron, 1970).
Though queens are not killed by departure
with a secondary swarm, the chance that a sec-
ondary swarm survives its first winter is very
low in temperate climates such as most of North
America (Seeley, 1978). Thus, departure with
464 D.C. Gilley, D.R. Tarpy
a secondary swarm is extremely costly relative
to inheriting the nest and we consider it as a
mechanism of elimination alongside queen
duels and pre-emergence destruction.
1.4. Outstanding questions about queen
elimination
To understand how natural selection has
shaped both queen and worker behavior to
maximize their inclusive fitness, we believe we
need to address the following outstanding ques-
tions about the elimination of queens from
honey bee nests.
1. What is the typical pattern of events during
queen elimination?
2. When do secondary swarms depart in
relation to other events during queen
elimination?
3. What are the relative roles of swarm
departure, queen duels, and pre-emergence
destruction in determining the outcome of
queen elimination?
4. What role do workers play in the
elimination of queens?
5. Does the biology of queen elimination dif-
fer between European (A. mellifera ligus-
tica) and African (A. mellifera scutellata)
honey bees?
2. MATERIALS AND METHODS
To observe in detail the behavior of both workers
and queens during queen elimination, five honey bee
colonies were established in May and June, 1998, in
Ithaca, New York. The behavior of the workers dur-
ing the queen duels that took place within these col-
onies was reported previously by Gilley (2001),
however, the results presented here are novel. Each
colony was housed in a three-frame observation hive
(approximate capacity of 6000 workers) headed by
a European honey bee queen (Apis mellifera ligus-
tica). Colony establishment was staggered to reduce
the chance that queen elimination was missed
because only one colony could be observed at a time.
To promote afterswarming and queen elimination,
each hive was well stocked with workers and brood
from its parent colony. The observation hives were
similar to those used previously (Gilley, 1998) but
with larger internal dimensions (74.0 × 45.8 ×
4.0 cm) and shorter entrance tunnels (20 cm) to facil-
itate the exodus of bees during swarming. Wooden
slats were inserted in the spaces around the frames
to ensure that the queens could be viewed at all times.
Each colony was monitored as follows when it began
to rear queens. The date and time that the workers
sealed each queen cell were noted and used to esti-
mate the earliest time that the queen might emerge
from each cell (typically 7 d after sealing). When the
cap of each cell became dark and smooth (a sign that
queen emergence was imminent), the cell was
inspected every 10 min. Queens were observed con-
tinuously from when they emerged to when they
were eliminated from the nest, often requiring four
or more observers working in shifts. We refer to
these colonies as Colonies E1–E5.
To strengthen the power of our results and to ena-
ble us to address all of the questions introduced
above (e.g., Question #5), we include in our analyses
data from several colonies that we did not observe
directly. The events in these additional colonies are
described narratively by their authors (Allen, 1956;
Fletcher, 1978), from which we have reconstructed
the sequence of events with regard to queen elimi-
nation to enable comparison with Colonies E1–E5.
Colony E6 (Allen, 1956) was located at the North of
Scotland College of Agriculture, housed in an obser-
vation hive that contained three frames of brood and
approximately 4500 bees when observations began
on 18 May, 1954. The honey bee subspecies is not
identified, but is assumed to be a European subspe-
cies. Colonies A1–A7 (Fletcher, 1978) are African
honey bee colonies (Apis mellifera scutellata, pre-
viously known as Apis mellifera adansonii) located
in Pietermaritzburg, South Africa, housed in two-
frame observation hives that initially contained
2500–3500 bees and brood of all ages. Together,
these 13 colonies are the only colonies that we know
of for which there exist detailed within-hive obser-
vations of queen elimination.
3. RESULTS
The emergence of queens and the events that
led to a restoration of monogyny in six Euro-
pean and seven African honey bee colonies are
described in Tables I and II. A flowchart of the
events during queen elimination in these
13 colonies is shown in Figure 2. The fates of
the queens reared by the 13 colonies examined
in this study are summarized in Table III. The
proportion of queens eliminated by each of the
three mechanisms for several types of colonies
is described in Table IV. Colonies with at least
one secondary swarm departure had fewer
queens killed in cells and more queens killed in
duels than colonies with no secondary swarm
departures (2 × 2 Chi-Square test of independ-
ence, d.f. = 1, P < 0.001). Colonies with at least
Mechanisms of queen elimination 465
Tab le I . The events associated with swarming within six European honey bee colonies (data for Colony E6
from Allen, 1956).
Colony Queen Emergence date & time Fate Fate date & time
E1 mother unknown departed with primary swarm 6/9 09:00
a 6/11 13:25 killed in duel with b 6/13 24:00
b 6/13 24:00 killed in duel with d 6/20 02:00
c 6/15 11:06 departed with secondary swarm 1 6/15 11:25
d 6/19 23:15 killed in duel with e 6/20 17:00
e 6/20 10:00 killed in duel with ? 6/22 ??:??
f 6/22 22:40 killed in duel with e 6/22 23:00
g 6/23 15:00 killed in duel with ? 6/23 ??:??
h 6/23 16:50 killed in duel with ? 6/24 07:25
i 6/25 06:40 departed with secondary swarm 2 6/27 10:55
j 6/27 11:35 killed in duel with ? 6/28 ??:??
k 6/29 07:00 inherited colony 6/29 12:45
l 6/29 07:30 killed in duel with k 6/29 12:45
E2 mother unknown departed with primary swarm ?/?? ??:??
a 7/7 07:30 killed in duel with c 7/12 21:57
b 7/11 22:15 killed in duel with a 7/12 15:15
c 7/12 08:00 inherited colony 7/13 15:26
d 7/13 10:00 killed in duel with c 7/13 15:26
E3 mother unknown departed with primary swarm 6/21 11:30
a 6/25 08:45 inherited colony 6/26 16:30
b - killed in cell by a 6/25 09:40
c - killed in cell by a 6/25 15:45
d - cell attacked by a 6/25 23:30
e - cell attacked by a 6/26 16:30
E4 mother unknown departed with primary swarm 6/22 12:20
a 7/5 ~ 01:45 inherited colony 7/6 ??:??
b 7/5 ~ 01:45 killed in duel with a 7/5 01:45
c - cell attacked by a 7/5 03:35
d - cell attacked by a 7/5 04:40
e - cell attacked by a 7/5 06:30
f - cell destroyed by ? 7/5 18:40
g - killed in cell by a 7/5 20:50
h - cell destroyed by ? 7/5 23:50
i - cell destroyed by ? 7/6 ??:??
E5 mother unknown missing 6/21 ??:??
a 6/27 17:09 departed with secondary swarm 7/3 11:00
b 6/29 03:00 killed in duel with a 6/29 03:00
c - killed in cell by a 6/30 08:02
466 D.C. Gilley, D.R. Tarpy
Table I . Continued.
Colony Queen Emergence date & time Fate Fate date & time
d 6/30 08:30 killed in duel with a 6/30 17:15
e - cell attacked by a 7/1 02:35
f - cell attacked by a 7/1 06:42
g 7/3 16:35 inherited colony 7/3 16:35
E6 mother unknown departed with primary swarm 7/7 15:35
a 7/12 <12:00 departed with secondary swarm 1 7/14 12:56
b - cut down by workers* 7/19* ??:??
c - dead queen removed by workers 7/4 ??:??
d 7/13 <12:00 departed with secondary swarm 2 7/19 12:40
e 7/19 12:30 departed with secondary swarm 2 7/19 12:40
f 7/19 12:35 inherited colony 7/19 12:40
g - cut down by workers* 7/19* ??:??
* The destruction of cells b and g following the departure of the final swarm on 7/19 is likely to have coincided
with attacks by Queen f on these cells, though no mention is made of these events by Allen (1956).
Table I I . The events associated with swarming within seven African honey bee colonies (modified from
Fletcher, 1978).
Colony Queen Emergence date & time Fate Fate date & time
A1 mother unknown departed with primary swarm unknown
a 1/14 05:24 inherited colony unknown
b - killed in cell by a unknown
c - cell attacked by a unknown
d - cell attacked by a unknown
e - cell attacked by a unknown
f - destroyed by workers after sealing unknown
g - destroyed by workers after sealing unknown
A2 mother unknown departed with primary swarm unknown
a 2/6 02:08 killed in duel with c 2/7 ??:??
b 2/9 ??:?? killed in duel with d e or f 2/9 ??:??
c 2/7 14:08 departed with secondary swarm 2/9 08:46
d 2/9 ??:?? killed in duel with b e or f 2/9 ??:??
e 2/9 ??:?? killed in duel with b d or f 2/9 ??:??
f 2/9 ??:?? inherited colony unknown
g - cell attacked by f (?) unknown
h - cell attacked by f (?) unknown
A3 mother unknown departed with primary swarm unknown
a 5/22 09:55 inherited colony unknown
Mechanisms of queen elimination 467
Table I I . Continued.
Colony Queen Emergence date & time Fate Fate date & time
b - cell attacked by a unknown
A4 mother unknown departed with primary swarm unknown
a 11/12 07:45 killed in duel with d 11/12 ??:??
b 11/12 08:01 killed in duel with a 11/12 ??:??
c 11/15 10:17 departed with secondary swarm 11/15 10:17
d 11/12 09:44 departed with secondary swarm 11/15 10:17
e 11/13 09:50 departed with secondary swarm 11/15 10:17
f 11/15 10:17 departed with secondary swarm 11/15 10:17
g 11/16 >06:59 killed in duel with h l j or k 11/16 ??:??
h 11/15 15:56 killed in duel with g l j or k 11/16 ??:??
i - destroyed by workers after sealing unknown
j 11/16 >06:59 inherited colony 11/16 ??:??
k 11/16 >06:59 killed in duel with g h j or l 11/16 ??:??
l 11/16 >06:59 killed in duel with g h j or k 11/16 ??:??
A5 mother unknown departed with primary swarm unknown
a 12/2 04:40 inherited colony unknown
b - cell attacked by a unknown
A6 mother unknown departed with primary swarm unknown
a 12/2 11:40 inherited colony unknown
b - cell attacked by a unknown
A7 mother unknown departed with primary swarm unknown
a 4/24 11:35 departed with secondary swarm 2 4/27 08:38
b 4/24 13:00 killed in duel with a (?) 4/25 or early 4/26
c 4/25 00:00 departed with secondary swarm 1 4/26 12:15
d 4/26 12:15 departed with secondary swarm 1 4/26 12:15
e 4/27 08:38
These queens engaged
in a series of duels until
only one queen remained
alive to inherit the nest.
4/28 ??:??
f 4/28 4/28 ??:??
g 4/28 4/28 ??:??
h 4/28 4/28 ??:??
i 4/28 4/28 ??:??
j 4/28 4/28 ??:??
}
468 D.C. Gilley, D.R. Tarpy
one duel tended to have fewer queens killed in
cells and more queens depart with secondary
swarms than colonies without duels, though
this trend was only marginally significant
(2 ×2 Chi-Square test of independence, d.f. =
1, P = 0.063). Colonies with at least one pre-
emergence death did not differ significantly
from colonies without pre-emergence death in
Figure 2. Flowchart of the events following departure of the mother queen for 13 honey bee colonies. The
width of each arrow represents the number of colonies following that sequence of events. The endpoint for
each colony, the inheritance of the nest by the remaining queen, is labeled with the colony’s identity in a
white circle (European colonies) or a black circle (African colonies).
Mechanisms of queen elimination 469
the number of queens eliminated by duels and
swarm departure (2 × 2 Chi-Square test of inde-
pendence, d.f. = 1, P = 0.356). African colonies
did not differ significantly from European col-
onies in the number of queens eliminated in
cells, duels, and secondary swarms (2 × 3 Chi-
Square test of independence, d.f. = 2, P =
0.358). Four queen cells (E6c, A1f, A1g, A4i)
Table III. The fate of 83 queens reared in 13 honey bee colonies.
Colony
Number of queens
killed in cells killed in duels left in swarms inherited colony total
E1 0 9 2 1 12
E2 0 3 0 1 4
E3 4 0 0 1 5
E4 7 1 0 1 9
E5 3 2 1 1 7
E6 2 0 3 1 6
A1 4 0 0 1 5
A2 2 4 1 1 8
A3 1 0 0 1 2
A4 0 6 4 1 11
A5 1 0 0 1 2
A6 1 0 0 1 2
A7 0 6 3 1 10
Tot a l 2 5
(30.1%)
31
(37.3%)
14
(16.9%)
13
(15.7%)
83
Table I V. The proportion of queens eliminated by each of three mechanisms for several types of colonies.
Category of Queens/Colonies
Mean Proportion of Eliminated Queens
killed in cells killed in duels left in 2 swarms
All queens eliminated in 13 colonies (n=70 queens) 44.3% 35.7% 20.0%
Colonies with pre-emergence destruction (n=9) 67.6% 18.9% 13.5%
Colonies with no pre-emergence destruction (n=4) 0.0% 72.7% 27.3%
Colonies with queen duels (n=7) 23.7% 58.8% 17.5%
Colonies without queen duels (n=6) 90.0% 0.0% 10.0%
Colonies with swarm departures (n=6) 14.6% 56.3% 29.1%
Colonies without swarm departures (n=7) 81.8% 18.2% 0.0%
African colonies (n=7) 61.2% 26.3% 12.5%
European colonies (n=6) 46.3% 37.9% 15.8%
470 D.C. Gilley, D.R. Tarpy
were not included in these analyses because
they were destroyed by workers before the first
daughter queen emerged (i.e., during the queen
rearing stage of colony fission).
4. DISCUSSION
4.1. Queen elimination occurs
in two distinct patterns
The events during queen elimination can be
dichotomized into two distinct patterns (Fig. 2):
colonies with queen duels (7 of 13 colonies)
and colonies without queen duels (6 of 13 col-
onies). In colonies with duels, it was most com-
mon for the duels to be followed by secondary
swarm departures, which were then followed
by more duels. In colonies without queen duels,
one queen emerged and killed all of her rivals
before their emergence, thus becoming the
unchallenged heir to the nest. This dichotomy
is further supported by the dependency of the
number of queens eliminated by duels and pre-
emergence destruction on the production of
secondary swarms (Tab. IV). These findings
support Visscher’s (1993) predictions from a
model of the queens’ and workers’ genetic
interests during reproductive fission. His
model predicts that, “the first virgin queen to
emerge should kill her sisters before they
emerge, unless conditions are such (strong col-
ony, good prospects for swarm) that survival of
the swarms is substantial relative to the remain-
ing colony”. In other words, colonies will have
pre-emergence destruction only when they are
too weak to produce viable secondary swarms.
Our results support this prediction; with one
exception (Colony E5, which swarmed follow-
ing 3 pre-emergence deaths), pre-emergence
destruction followed the departure of the final
swarm if any secondary swarms were produced
at all (Fig. 2). Which trajectory a colony takes
is presumably decided by the workers, which
either guard queen cells from attack by chasing
and biting intruding queens (thus preserving
developing queens for duels and secondary
swarms) or refrain from chasing and biting
intruding queens (thus allowing queens to be
killed before emergence). The mechanisms by
which such group decisions are made warrant
further investigation.
4.2. The timing of secondary swarm
departures is related to the events
of queen elimination
The timing of the departure of secondary
swarms was related to several factors. First, the
time of day that secondary swarms departed
was not random; all nine swarms departed
between 0838 and 1256. This suggests that
environmental factors such as air temperature
and remaining daylight play a significant role
in swarm departure (see also Winston, 1987),
and thus also in the queen elimination process.
Second, the relative timing of secondary swarm
departure during the queen-elimination period
was bimodally distributed; swarms typically
departed near the end of the queen-elimination
process (at a mean of 86.9% of the total dura-
tion for colonies that produced one secondary
swarm), or, if colony produced two secondary
swarms, the first departed about one-third of
the way through the process (at a mean of
34.0% of the total duration) and the second
departed near the end (at a mean of 87.9% of
the total duration). Thus, the departure of the
colony’s last secondary swarm precedes very
closely the end of the queen elimination proc-
ess, which probably reflects the dramatic
change in the costs and benefits of queen elim-
ination caused by the departure of a swarm (see
Visscher, 1993). Third, the timing of secondary
swarm departure may be related to queen emer-
gence. In at least 4 of the 6 colonies that pro-
duced a secondary swarm (Colonies E1, E6,
A4, and A7), a queen emerged within 20 min
of swarm departure. Since most of these queens
emerged just before swarm departure, it is
likely that the workers released them in prepa-
ration for swarming. Interestingly, when more
than one emerged queen was present in the hive
during the departure of a secondary swarm, it
was common for one of these queens to remain
in the hive (this occurred in 4 of the 5 swarms
that departed when more than one queen was
present in the hive). Earlier-emerging queens
did not have a greater chance of remaining in
the hive; 2 of the earliest-emerging queens and
2 of the latest-emerging queens remained in
their hive following the departure of a second-
ary swarm.
Mechanisms of queen elimination 471
4.3. Queen duels are a common
mechanism of elimination
and are associated with more
secondary swarm departures
and less pre-emergence destruction
All three mechanisms of elimination played
a role in the outcome of the queen elimination
process. Death in a queen duel was the most
common fate for a queen, thus the outcome of
duels (which may depend on relative fighting
ability, relatedness to workers, and chance;
Tarpy and Fletcher, 1998; Gilley, 2001) is
likely to significantly impact the outcome of
queen elimination. It is still not entirely clear,
however, why the workers permit duels to
occur. The adaptive benefits of this group deci-
sion deserve further study.
The relative contribution of each mecha-
nism to the final outcome of queen elimination
differed depending on which of the two trajec-
tories a colony followed; in colonies with
queen duels more queens departed with sec-
ondary swarms and fewer queens were killed
before emergence than in colonies without
queen duels. This suggests that queens might
maximize their chances of inheriting the nest
by adopting a conditional strategy of emerging
early in colonies that will not have duels or sec-
ondary swarms (thereby avoiding pre-emer-
gence destruction), and emerging late (but not
too late) in colonies that have duels and sec-
ondary swarms (thereby avoiding duels and
swarm departures). First-emerging queens have
been shown to have an advantage in queen
elimination (Schneider and DeGrandi-Hoffman,
2003), but in our study, first-emerging queens
had an advantage only when colonies had no
duels, where 5 of 6 first-emerging queens
inherited the nest compared to 1 of 7 for colo-
nies with duels. In colonies with duels, the
queens that inherited the nest typically emerged
last or next to last, following the departure of
the final secondary swarm.
4.4. Workers play a significant but
non-lethal role in queen elimination
None of the 70 queens eliminated in this
study was killed by the workers; all queens
were either stung by a rival queen, killed before
emerging from their cells, or departed in sec-
ondary swarms (Tab. III). These results are
supported by several other studies of newly
emerged queen and worker behavior in obser-
vation hives (Gilley, 2001; Schneider et al.,
2001; Gilley et al., 2003; Tarpy and Fletcher,
2003; Schneider and DeGrandi-Hoffman,
2003). In these studies, a total of 178 queens
were observed for a combined total of approx-
imately 1600 hours, and not once did the work-
ers by themselves kill a virgin queen. Worker-
queen interactions might, however, impose
costs that negatively affect the queens’ chances
of surviving queen elimination. For example,
emerged queens may be immobilized by work-
ers when defecated upon by another queen
(Gilley, 2001; Tarpy and Fletcher, 2003).
These queens are often released unharmed, but
are sometimes stung by rival queens while
immobilized (Gilley, 2001; Tarpy and Fletcher,
2003). Workers also harass newly emerged
queens (Gilley, 2001, 2003), but the costliness
of these interactions has not been demon-
strated. Workers also played a role in determin-
ing the outcome of queen elimination by
enlarging holes chewed in cells by emerged
queens and by destroying cells on their own
during the queen rearing stage of reproductive
fission (Tab. II, Colony A1; see also Huber,
1814; Allen, 1956; Schneider and DeGrandi-
Hoffman, 2002; Tarpy et al., 2000).
4.5. Queen elimination in African
and European colonies is similar
but patterns may differ
We did not see marked differences in the
three mechanisms of queen elimination between
what we observed for European bees and what
has been reported for African bees. African col-
onies were more likely than European colonies
to complete queen elimination by pre-emer-
gence destruction alone (Fig. 2), yet death by
pre-emergence destruction was not signifi-
cantly more likely for African queens than for
European queens (Tab. IV). Though incontro-
vertible conclusions await larger sample sizes
and controlled initial colony conditions, this
difference in trajectories between subspecies
may be important for understanding the success
of A. mellifera scutellata among established
European honey bee populations. For example,
early emergence may be more important to
queen survival in African than European sub-
species (DeGrandi-Hoffman et al., 1998),
472 D.C. Gilley, D.R. Tarpy
while characteristics associated with fighting
ability may be more important to queen sur-
vival in European subspecies.
4.6. Priorities for future research
The results of this study highlight several
questions about queen elimination in honey
bees that remain unresolved and which we
believe should be a priority for future research:
(1) what is the underlying cause of the variation
among colonies in the relative frequencies of
pre-emergence destruction, duels, and second-
ary swarm departures (e.g., colony condition,
environmental circumstances), (2) what is the
adaptive value of duels for the workers of a col-
ony, (3) how and to what degree do workers
affect the outcome of queen elimination,
(4) how might features of queen elimination
unique to African honey bees contribute to their
success in the New World, and (5) why do
honey bees differ from other swarm-founding
Hymenopteran species such as ants (Fletcher
and Blum, 1983; Bourke and Franks, 1995) and
stingless bees (Michener, 1974; Sakagami,
1982), where workers themselves eliminate
queens (see also Tarpy and Gilley, 2004).
ACKNOWLEDGEMENTS
We thank Tom Seeley, Paul Sherman, Kern
Reeve, Cole Gilbert, and Bryan Danforth for dis-
cussions and helpful comments on the manuscript.
The manuscript was also improved by the sugges-
tions of two anonymous reviewers. DCG was sup-
ported by NIMH Training Grant T32 MH15793.
DRT was supported by NSF grant IBN-973-4181
awarded to H.K. Reeve.
Résumé – Trois mécanismes d’élimination des
reines chez les colonies d’abeilles (Apis mellifera)
qui essaiment. L’élimination d’une reine est un pro-
cessus par lequel toutes les reines non accouplées et
produites au cours de la scission (essaimage) de la
colonie, sauf une, sont éliminées du nid parental. La
figure 1 montre comment l’élimination des reines est
liée au processus de scission dans une colonie typi-
que d’abeilles. Les reines sont éliminées selon trois
mécanismes : les duels reine-reine, les assassinats de
reines et le départ avec l’essaim. Nous passons
d’abord en revue ce qui est connu concernant chacun
de ces mécanismes d’élimination. Puis nous avons
utilisé de nouvelles données provenant de colonies
logées dans des ruches d’observation qui essai-
maient et avons en même temps ré-analysé les
données de la littérature pour répondre à cinq ques-
tions sur la biologie de l’élimination des reines, que
nous considérons comme remarquables : (1) les
rôles relatifs des trois mécanismes dans le résultat de
l’élimination des reines, (2) le déroulement typique
des évènements dans le nid au cours de l’élimination,
(3) le rôle joué par les ouvrières (4) l’action de
« l’aspersion » (par des matières fécales)sur le résul-
tat de l’élimination, (5) les différences entre les
abeilles européennes (Apis mellifera ligustica) et
africaines (A. m. scutellata) dans ce processus.
Les tableaux I et II décrivent les évènements qui ont
eu lieu au sein des six colonies d’abeilles européen-
nes et des sept colonies d’abeilles africaines. Le
tableau III quantifie le nombre de reines éliminées
par chaque mécanisme. La figure 2 montre le dérou-
lement des évènements. La largeur de la flèche
représente le nombre de colonies qui suit le dérou-
lement indiqué. Les cercles noirs et blancs indiquent
pour chaque colonie l’étape finale (i.e. la reine res-
tante qui hérite du nid), à partir de laquelle on peut
remonter la séquence des évènements jusqu’au
départ de l’essaim primaire.
Nous tirons de nos résultats les conclusions
suivantes : (1) les évènements au cours de l’élimi-
nation des reines peuvent être divisés en deux séries
distinctes : les colonies sans duels de reines, d’où
émerge une reine qui assassine toutes ses rivales, et
les colonies avec duels de reines, où une suite très
variable d’évènements se succèdent incluant le
départ d’au moins un essaim secondaire (Fig. 2) ;
(2) le mécanisme d’élimination le plus commun est
le duel, mais la contribution de chaque mécanisme
au résultat final dépend du déroulement des évène-
ments que suit une colonie ; (3) les ouvrières ne tuent
pas les reines et ne les blessent pas sérieusement,
mais elles interagissent avec les reines de certaines
façons qui peuvent être coûteuses pour les reines ;
(4) l’aspersion dans un contexte naturel aboutit à
attirer les ouvrières sur la reine aspergée ; (5) les
abeilles européennes et africaines utilisent les
mêmes mécanismes, mais des différences existent
probablement dans la fréquence relative de chacun
d’entre eux.
essaimage / élimination des reines / duel de
reines / destruction avant l’émergence
Zusammenfassung – Drei Mechanismen zur Ent-
fernung von überzähligen Königinnen in
schwärmenden Honigbienenvölkern. Der Erhalt
der Monogynie, d.h. nur eine Königin im Volk,
bedeutet, dass bei der Teilung der Völker (Schwarm-
bildung) alle unbegatteten Königinnen außer einer
aus dem alten Schwarmvolk entfernt werden. Abb. 1
zeigt die Beziehung der Entfernung von überzähli-
gen Königinnen zum Teilungsvorgang in einem
typischen Honigbienenvolk. Die Königinnen kön-
nen auf drei verschiedene Arten beseitigt werden:
durch Königinnenkämpfe, durch Zerstörung vor
dem Schlupf und durch Verlassen des Volkes mit
Mechanisms of queen elimination 473
einem Nachschwarm. Hier beschreiben wir
bekannte Daten über alle Beseitigungsvorgänge.
Anhand der Analyse neuer Daten von schwärmen-
den Beobachtungsstöcken und einer Neuanalyse der
schon bekannten Daten nehmen wir zu fünf Fragen
der Biologie der Königinnenbeseitigung Stellung,
die wir für bemerkenswert halten.
Unsere Ergebnisse bestehen aus folgenden Teilen:
Tabelle I und II beschreiben Ereignisse, die in 6 Bie-
nenvölkern europäischer Abstammung und 7
Bienenvölkern afrikanischer Abstammung auftre-
ten. Tabelle II gibt für jeden der verschiedenen
Mechanismen die Zahlen der beseitigten Königin-
nen an. Tabelle IV zeigt die mittleren Anteile der mit
jedem Mechanismus beseitigten Königinnen für
verschiedene Kolonietypen. Abb. 2 zeigt die
Abfolge der Ereignisse während der Königinnenbe-
seitigung, wobei die Breite der jeweiligen Pfeile die
Anzahl der Völker repräsentiert, die dem entspre-
chenden Ereignispfad gefolgt sind. Die schwarzen
und weißen Kreise in Abb. 2 zeigen den letzten
Schritt für jedes Volk an (z.B. dass die Königin das
Volk übernimmt). Von hier aus kann die Ereignis-
folge bis zum Auszug des Primärschwarms
zurückverfolgt werden.
Aus unseren Ergebnissen ziehen wir die folgenden
Schlüsse: (1) Die Ereignisse während der Königin-
nenbeseitigung folgen zwei deutlich getrennten
Mustern. Einerseits gibt es Völker mit Königinnen-
kämpfen. Dort folgen auf die Kämpfe zumeist
Nachschwärme, danach erfolgen weitere Königin-
nenkämpfe. Andererseits gibt es Völker ohne
Königinnenkämpfe, in diesen schlüpft eine Königin
und tötet alle ihre Rivalinnen, bevor diese aus den
Zellen schlüpften. (2) Der Zeitpunkt, zu dem ein
Nachschwarm das Volk verlässt, steht zu den Ereig-
nissen der Königinnenbeseitigung in Beziehung.
(3) Königinnenkämpfe sind ein verbreiteter Mechanis-
mus zur Entfernung der Rivalin und sind mit
verstärkter Nachschwarmaktivität und geringerer
Zerstörung in den Zellen vor dem Schlupf verbun-
den. (4) Die Arbeiterinnen spielen eine bedeutende,
aber nicht tödliche Rolle bei der Königinnenbeseiti-
gung. (5) Der Ablauf Königinnenbeseitigung ist bei
europäischen und afrikanischen Honigbienenvöl-
kern ähnlich, allerdings könnte das Muster der
Verhaltensabfolge unterschiedlich ausfallen.
Schwärmen / Königinnenbeseitigung / Königin-
nenkämpfe / Zerstörung vor dem Schlupf
REFERENCES
Allen M.D. (1956) The behaviour of honeybees
preparing to swarm, Br. J. Anim. Behav. 4, 14–22.
Bernasconi G., Bigler L., Hesse M., Ratnieks F.L.W.
(1999) Characterization of queen-specific
components of the fluid released by fighting
honey bee queens, Chemoecology 9, 161–167.
Bernasconi G., Ratnieks F.L.W., Rand E. (2000)
Effect of “spraying” by fighting honey bee
queens (Apis mellifera L.) on the temporal
structure of fights, Insectes Soc. 47, 21–26.
Boch R., Shearer D.A., Shuel R.W. (1979) Octanoic
acid and other volatile acids in the mandibular
glands of the honey bee Apis mellifera and in
royal jelly, J. Apic. Res. 18, 250–252.
Bourke A.F.G., Franks N.R. (1995) Social Evolution
in Ants, Princeton University Press, Princeton.
Butz V.M., Dietz A. (1994) The mechanism of queen
elimination in two-queen honey bee (Apis
mellifera L.) colonies, J. Apic. Res. 33, 87–94.
Caron D.M. (1970) A study of swarming and the
behavior of swarming in honey bees, Apis
mellifera L, Ph.D. Thesis, Cornell University.
Caron D.M., Greve C.W. (1979) Destruction of queen
cells placed in queenright Apis mellifera colonies,
Ann. Entomol. Soc. Am. 72, 405–407.
DeGrandi-Hoffman G., Watkins J.C., Collins A.M.,
Loper G.M., Martin J.H., Arias M.C., Sheppard
W.S. (1998) Queen developmental time as a
factor in the Africanization of European honey
bee (Hymenoptera: Apidae) populations, Ann.
Entomol. Soc. Am. 91, 52–58.
Fletcher D.J.C. (1978) Vibration of queen cells by
worker honey bees and its relation to the issue of
swarms with virgin queens, J. Apic. Res. 17, 14–
26.
Fletcher D.J.C., Blum M.S. (1983) Regulation of
queen number by workers in colonies of social
insects, Science 219, 312–314.
Gilley D.C. (1998) The identity of nest-site scouts in
honey bee swarms, Apidologie 29, 229–240.
Gilley D.C. (2001) The behavior of honey bees (Apis
mellifera ligustica) during queen duels, Ethology
107, 601–622.
Gilley D.C. (2003) Absence of nepotism in the
aggressive interactions between workers and
dueling queen honeybees, Proc. R. Soc. Lond. B
270 (1528), 2045–2049.
Gilley D.C., Tarpy D.R., Land B.B. (2003) Effect of
queen quality on interactions between workers
and dueling queens in honeybee (Apis mellifera
L.) colonies, Behav. Ecol. Sociobiol. 55, 190–
196.
Gotwald W.H.Jr. (1995) Army Ants: the Biology of
Social Predation, Ithaca, Cornell University
Press.
Harano K., Obara Y. (2004) Virgin queens selectively
destroy fully matured queen cells in the honeybee
Apis mellifera L., Insectes Soc. 51, 253–258.
Hepburn H.R., Radloff S.E. (1998) Honeybees of
Africa, Springer-Verlag, Berlin.
Huber F. (1814) New Observations Upon Bees,
Hamilton, Illinois, Am. Bee J., translated by
Dadant C.P., 1926.
Jeanne R.L. (1991) The swarm-founding Polistinae,
in: Ross K.G., Mattews R.W. (Eds.), The Social
Biology of Wasps, Cornell University Press,
Ithaca, pp. 191–231.
Lindauer M. (1961) Communication Among Social
Bees, Harvard University Press, Cambridge.
474 D.C. Gilley, D.R. Tarpy
Martin P. (1963) The mechanism of colony division
in honeybee swarming and a study of the problem
of absconding, Insectes Soc. 10, 13–42.
Michener C.D. (1974) The Social Behavior of the
Bees, Harvard University Press, Cambridge.
Ohtani T. (1994) Behaviors of adult queen honeybees
within observation hives, I. behavior patterns,
Humans and Nature 3, 37–77.
Page R.E. Jr., Erickson E.H. (1986) Kin recognition
of virgin queen acceptance by worker honey
bees (Apis mellifera L.), Anim. Behav. 34, 1061–
1069.
Page R.E. Jr., Blum M.S., Fales H.M. (1988) o-
Aminoacetophenone, a pheromone that repels
honeybees (Apis mellifera L.), Experientia 44,
270–271.
Pflugfelder J., Koeniger N. (2003) Fight between
virgin queens (Apis mellifera) is initiated by
contact to the dorsal abdominal surface,
Apidologie 34, 249–256.
Post D.C., Page R.E. Jr., Erickson E.H. (1987)
Honeybee (Apis mellifera L.) queen feces: source
of a pheromone that repels worker bees, J. Chem.
Ecol. 13, 583–591.
Sakagami S.F. (1982) Stingless Bees, in: Hermann
H.R. (Ed.), Social Insects, Academic Press, New
York, pp. 361–423.
Schneider S.S., DeGrandi-Hoffman G. (2002) The
influence of worker behavior and paternity on the
development and emergence of honey bee
queens, Insectes Soc. 49, 306–314.
Schneider S.S., DeGrandi-Hoffman G. (2003) The
influence of paternity on virgin queen success in
hybrid colonies of European and African
honeybees, Anim. Behav. 65, 883–892.
Schneider S.S., Lewis L.A. (2004) The vibration
signal, modulatory communication and the
organization of labor in honey bees, Apis
mellifera, Apidologie 35, 117–131.
Schneider S.S., Painter-Kurt S., Degrandi-Hoffman
G. (2001) The role of the vibration signal during
queen competition in colonies of the honeybee,
Apis mellifera, Anim. Behav. 61, 1173–1180.
Schneider S.S., DeGrandi-Hoffman G., Smith D.R.
(2004) The African honey bee: factors
contributing to a successful biological invasion,
Annu. Rev. Entomol. 49, 351–376.
Seeley T.D. (1978) Life history strategy of the honey
bee, Apis mellifera, Oecologia 32, 109–118.
Tarpy D.R., Fletcher D.J.C. (1998) Effects of
relatedness on queen competition within honey
bee colonies, Anim. Behav. 55, 537–543.
Tarpy D.R., Hatch S., Fletcher D.J.C. (2000) The
influence of queen age and quality during queen
replacement in honeybee colonies, Anim. Behav.
59, 97–101.
Tarpy D.R., Fletcher D.J.C. (2003) 'Spraying'
behavior during queen competition in honey bees,
J. Insect Behav. 16, 425–437.
Tarpy D.R., Gilley D.C. (2004) Group decision
making during queen production in colonies of
highly eusocial bees, Apidologie 35, 207–216.
Tarpy D.R., Gilley D.C., Seeley T.D. (2004) Levels of
selection in a social insect: a review of conflict
and cooperation during honey bee (Apis
mellifera) queen replacement, Behav. Ecol.
Sociobiol. 55, 513–523.
Tiemann K., Brückner D. (1993) Zum Schwarmver-
halten der Sizilianischen Honigbiene Apis mellif-
era sicula (Montagano 1911), Apidologie 24,
365–374.
van Veen J.W., Sommeijer M.J. (2000) Colony
reproduction in Tetragonisca angustula (Apidae,
Meliponini), Insectes Soc. 47, 70–75.
Visscher P.K. (1993) A theoretical analysis of
individual interests and intracolony conflict
during swarming of honey bee colonies, J. Theor.
Biol. 165, 191–212.
Weaver E.C., Weaver N. (1980) Physical domination
of workers by young queen honey bee Apis
mellifera Hymenoptera Apidae, J. Kans.
Entomol. Soc. 53, 752–762.
Winston M.L. (1987) The biology of the honey bee,
Harvard University Press, Cambridge.
