Worker piping in honey bee swarms and its role in preparing for liftoff

Abstract

Worker piping, previously reported only in hives, was observed in swarms as they prepared to liftoff to fly to a new home. Pipers are excited bees which scramble through the swarm cluster, pausing every second or so to emit a pipe. Each pipe consists of a sound pulse which lasts 0.82&#450.43 s and rises in fundamental frequency from 100-200 Hz to 200-250 Hz. Many, if not all, of the pipers are nest-site scouts. The scouts pipe when it is time to stimulate the non-scouts to warm themselves to a flight-ready temperature (35C) in preparation for liftoff. The time-course of worker piping matches that of swarm warming; both start at a low level, about an hour before liftoff, and both build to a climax at liftoff. When we excluded pipers from bees hanging in the cool, outermost layer of a swarm cluster, we found that these bees did not warm up. The form of worker piping that we have studied in swarms differs from the form of worker piping that others have studied in hives. We call the two forms "wings-together piping" (in swarms) and "wings-apart piping" (in hives).

Full text

ORIGINAL PAPER
Thomas D. Seeley áJu
Èrgen Tautz
Worker piping in honey bee swarms and its role
in preparing for liftoff
Accepted: 6 September 2001 / Published online: 2 October 2001
ÓSpringer-Verlag 2001
Abstract Worker piping, previously reported only in
hives, was observed in swarms as they prepared to lifto
to ¯y to a new home. Pipers are excited bees which
scramble through the swarm cluster, pausing every sec-
ond or so to emit a pipe. Each pipe consists of a sound
pulse which lasts 0.820.43 s and rises in fundamental
frequency from 100±200 Hz to 200±250 Hz. Many, if
not all, of the pipers are nest-site scouts. The scouts pipe
when it is time to stimulate the non-scouts to warm
themselves to a ¯ight-ready temperature (35°C) in
preparation for lifto. The time-course of worker piping
matches that of swarm warming; both start at a low
level, about an hour before lifto, and both build to a
climax at lifto. When we excluded pipers from bees
hanging in the cool, outermost layer of a swarm cluster,
we found that these bees did not warm up. The form of
worker piping that we have studied in swarms diers
from the form of worker piping that others have studied
in hives. We call the two forms ``wings-together piping''
(in swarms) and ``wings-apart piping'' (in hives).
Keywords Communication áHoney Bee áSound á
Swarming áWorker piping
Introduction
After leaving the parental nest to start a new colony, a
swarm of honey bees hangs from a tree branch in a
beard-like cluster for several hours or several days while
its scouts choose a suitable nesting cavity (reviewed by
Lindauer 1971; Seeley 1982). During this time, a swarm
performs thermoregulation, maintaining its cluster-core
temperature at 34±36°C and its cluster-mantle temper-
ature above 15°C (Heinrich 1981). A swarm maintains
these elevated temperatures mostly by reducing its heat
loss rather than raising its heat production, thereby
conserving its energy supply. However, as soon as the
scout bees have chosen a new home, the swarm raises the
mantle temperature to the core temperature, about 35°C
(Heinrich 1981), which is the temperature required for
rapid ¯ight (Heinrich 1979). Once the mantle bees reach
this high temperature, the thousands of bees in the
swarm launch into ¯ight, form a cloud of swirling bees,
and begin moving together to their new domicile
(described by Seeley et al. 1979).
How do the bees in a swarm achieve such a beautifully
coordinated lifto? One of the underlying mechanisms is
the signal called buzz running (Schwirrlaufen). During
the ®nal 10 or more minutes before lifto, excited bees
force their way through the quiet bees in the cluster,
running about in a zig-zag pattern, butting into the other
bees, and buzzing their wings every second or so (Linda-
uer 1955; Esch 1967). Many, perhaps most, of these buzz
runners are scout bees and their actions appear to loosen
up the cluster (Seeley et al. 1979). Martin (1963) demon-
strated, with a split-hive experiment, that only bees
directly contacted by buzz runners will join the mass
exodus when a swarm initially leaves the parental nest.
Hence, it seems clear that buzz running plays a critical role
in lifto, probably triggering the ®nal break up of the
cluster. Buzz running is evidently not the whole story,
however, because buzz runners appear only in the last few
minutes before lifto, whereas the rise in a swarm's tem-
perature starts an hour or so before lifto (Heinrich 1981).
A second signal that might inform the relatively cool and
quiescent bees in a swarm that it is time to warm them-
selves for lifto is the shaking signal (sometimes called the
vibration signal). To produce this signal, one bee grasps
another and shakes this bee's body for 1±2 s at 16±18 Hz
(see Fig. 1 in Seeley et al. 1998). There is strong evidence
J Comp Physiol A (2001) 187: 667±676
DOI 10.1007/s00359-001-0243-0
T.D. Seeley (&)
Department of Neurobiology and Behavior,
Seeley G. Mudd Hall, Cornell University,
Ithaca, NY 14853, USA
E-mail: tds5@cornell.edu
Fax: +1-607-2544308
J. Tautz
Theodor-Boveri-Institut, Lehrstuhl fu
Èr Verhaltensphysiologie
und Soziobiologie, Am Hubland,
97074 Wu
Èrzburg, Germany

that the shaking signal acts as a modulatory signal that
produces a general activation of worker bees in swarms
(Visscher et al. 1999; Lewis and Schneider 2000) and in
hives (Schneider et al. 1986; Nieh 1998; Seeley et al. 1998).
However, because the warm-up period does not occur
solely or even principally in the last hour before lifto, , it
seems that the shaking signal is not the warm-up signal.
A third signal that possibly helps a swarm execute a
well-coordinated lifto, by stimulating the inactive bees
to warm themselves for ¯ight, is the high-pitched piping
sound that the workers produce during the last hour or
so before departure (Lindauer 1955; Seeley et al. 1979;
Camazine et al. 1999). During this ®nal hour, the piping
by workers crescendos. Initially, the swarm cluster emits
a weak and intermittent sound produced by just a few
bees, but gradually the sound's intensity increases until
at lifto there is a loud and continuous sound produced
by many bees. Because worker piping occurs solely
during the warm-up period, and because it climaxes at
lifto, we hypothesized that it stimulates the inactive
bees in a swarm cluster to warm themselves in prepara-
tion for lifto. We tested this hypothesis by determining
(1) the time-course of worker piping in relation to swarm
warming, (2) the identity of the pipers, and (3) the
temperature eects of excluding this signal from a subset
of the mantle bees in a swarm shortly before lifto. The
properties of this acoustic signal are also described.
Materials and methods
Study site and bees
All observations were made at the Liddell Field Station of Cornell
University, in Ithaca, New York (42°26' N, 76°30' W). This site is
surrounded by woods containing many old trees with cavities, so the
swarms studied had no diculty ®nding suitable nest sites. All three
swarms studied were arti®cial swarms prepared from colonies
headed by ``Buckfast'' queens (a hybrid of Apis mellifera mellifera
and A. m. ligustica; Adam 1987). To make each swarm, we ®rst
located a colony's queen and put her in a small wooden cage
(3.2 cm ´10 cm ´1.6 cm) with wire screen on one side (a standard
wooden queen cage). Then, using a large funnel, we shook 1.0 kg of
worker bees (ca. 7,500 bees; Mitchell 1970) from the frames of the
same colony into a swarm cage (15 cm ´25 cm ´35 cm) made of
wood with wire-screen sides. We also placed the caged queen inside
the swarm cage. For the next 48±72 h (until copious wax scales
appeared beneath the swarm cage), we fed the caged bees ad libitum
with a 50% (vol/vol) sucrose solution. Finally, the swarm cage was
opened, the queen (still in her own small cage) was fastened to the
swarm mount (see below), and the workers were shaken onto the
base of the mount. Within an hour, the workers were clustered over
the queen cage and behaving like a natural swarm. After several
hours or a few days the swarm would ®nish choosing its nest site, lift
o, and attempt to ¯y away. However, because the queen remained
caged at the swarm mount the swarm always returned and resettled
there. Several hours later, or the next day, the swarm would again lift
o and ¯y away, only to return and resettle once more around the
caged queen. In this way we were able to monitor repeatedly (two to
six times) the preparations for lifto with each of our three swarms.
Apparatus
Swarms were placed on the swarm mount which was described
previously (see Fig. 1; Seeley and Buhrman 1999). This mount
consists of a ¯at vertical board, on which the swarm clusters, and a
wire screen (of 8-mesh hardware cloth with several passageways for
the bees cut into it) mounted vertically over the swarm's surface so
that the outermost layer of the swarm is on the outside of the
screen. This apparatus facilitates video recording the behavior of
workers on the surface of the swarm cluster. Video recordings were
made with a digital video camera (Panasonic NV-DS35EG). Re-
cordings were converted to S-VHS format and analyzed using a
videocassette player with variable-speed playback (JVC BR-
S525U). To record the sounds produced by a worker while she was
being videorecorded, we manually followed the bee with a small, 5-
mm-diameter, custom-made microphone (¯at frequency response
from 20 Hz to 6,000 Hz, and sensitive only to pressure ¯uctua-
tions) held approximately 1 cm from the bee, in a dorsal position,
and slightly to the side so that it did not interfere with the video
recording. This microphone's output was recorded on one of the
video camera's audio channels. We recorded the behavior of bees
that were producing piping sounds or waggle dances, or both. To
obtain detailed records of the behavior of single piping workers, we
removed the screen portion of the swarm mount immediately after
several of the liftos and zoomed the video camera in on the one or
two bees that were intensely piping on the screen of the queen cage.
To record the level of worker piping within the cluster, we
mounted two of the small microphones, one on each side of the
swarm, on the threaded rods supporting the screen of the swarm
mount. This positioned each microphone deep inside the swarm
cluster. The two microphones were connected to an ampli®er whose
stereo output was recorded with a digital minidisc recorder (Sony
MZ-R37SP).
To record temperatures in the swarm, we used copper-con-
stantan thermocouple probes and a digital thermometer (Bailey
Bat-12). The swarm's core temperature was measured with a probe
mounted in the middle of the cluster, directly in front of the queen
cage. The swarm's mantle temperature was measured with a probe
mounted in the center of the screen of the swarm mount. The
Fig. 1 The swarm mount screen that was used to test whether
mantle bees would warm themselves in preparation for lifto when
prevented from receiving signals from piping workers. The screen
bears two cages, each equipped with a thermocouple probe. Both
cages have covers of 8-mesh hardware cloth, but one cover has a
large opening
668

distance of the screen from the board of the swarm mount was
adjusted so that the layer of bees covering this probe was just one
or two bees thick. The ambient temperature was measured with a
probe mounted on the swarm mount, 2 cm below the board on
which the swarm was clustered.
To see if mantle bees would warm themselves in preparation for
lifto when we prevented piping workers from contacting them, we
performed an experiment in which we replaced the normal screen of
our swarm mount with a screen bearing two symmetrically posi-
tioned cages (6 cm ´10 cm ´2 cm), each of which was equipped
with a thermocouple probe (see Fig. 1). When a swarm was placed
on the swarm mount, both cages became ®lled with mantle bees.
One cage was closed by installing a cover of ``8-mesh'' hardware
cloth (openings 3 mm ´3 mm) while the other cage was kept open
by installing a cover identical to the ®rst except that it had a large
(6 cm ´9 cm) opening. The covers were installed only after we
determined, by means of the stereo microphones, that the piping
inside the swarm had become continuous.
Data analysis
The video recordings were analyzed in slow motion to determine
(1) the spatiotemporal patterns of the movement and signaling
behavior of piping workers, (2) the posture of workers while pro-
ducing the piping signal, and (3) the pattern of behavioral transi-
tions made by waggle-dancing bees as a swarm approached the
moment of lifto. To determine the latter pattern, we would choose
at random a waggle dancer, follow her for 10 s, and note which of
the following activities she was engaged in at the end of the 10-s
period: waggle dancing, walking/running, piping, buzz running,
shaking, or ¯ying (these are all the possibilities). This procedure
was repeated with 50 bees every half-hour, for several hours before
the two observed liftos of swarm 3.
The audio recordings made by following individual bees were
analyzed by transferring the digital data from the audio track of the
video camera into a computer. We then used Avisoft SASLabPro
to obtain sonagrams and spectrograms of the piping signal. To
determine a distribution of the durations of piping sounds, we
measured the durations of 50 randomly chosen pipes from swarm
2, recorded on 3 September.
The audio recordings made with the stereo microphones
mounted inside the swarm cluster were analyzed to measure the
overall level of worker piping within a swarm at dierent times
before lifto. The digital data from the minidisc recorder were
transferred to a computer and analyzed using Avisoft SASLabPro.
We measured the level of worker piping as the percent time that
piping was audible in our stereo recording, and our procedure for
making this measurement was as follows. First, for each swarm we
chose a set of sampling periods; these sampling periods were short
(3±5 min long) for the hour or so preceding lifto and were rela-
tively long (10±30 min long) for times more in advance of lifto.
Second, within each sampling period, we chose at random three 20-s
intervals. Third, for each of these three 20-s intervals, we counted
the number of 1-s subintervals (N) in which worker piping was
audible in our recording. Fourth, we estimated from these counts
the percent time that piping was audible during each 20-s interval
(N/20 ´100). Finally, using the three measurements for each sam-
pling period, we calculated the mean and standard deviation of the
percentage time that piping was audible for each sampling period.
Results
Sights and sounds of piping workers
Behavioral description
Piping was heard easily in swarms shortly before lift-
o, but at ®rst it was not easy to detect which bees
were producing these sounds, which are reminiscent of
the revving of a Formula One racing car's engine.
Careful searching of the swarm surface in the region
of a piping bee revealed a worker performing a
striking behavior coincident with each piping sound:
she paused brie¯y from running over other bees,
pressed her thorax to the bee beneath her, and pulled
her wings together over her abdomen, which arched
noticeably downward during sound production
(Fig. 2). In the vast majority of cases, the bee pressed
her thorax to another bee while piping, but occa-
sionally she pressed it against one of the wooden or
wire screen surfaces of the swarm mount. The pressing
down of the piper's thorax during sound production
suggests that she generates the sound with her wing
muscles in the thorax and loads it into the substrate,
which is almost always another bee. (The airborne
vibrations that we heard and recorded may have in-
cluded radiation of sound from substrate to air, as
well as direct sound emission from the bee.) The bees
contacted by pipers were ones hanging motionless in
the cluster. They showed no behavioral response to
the sound.
Pipers are strikingly excited bees which run over and
burrow into the swarm cluster, pausing frequently to
pipe (Fig. 3). Using our video recordings from swarm 2,
we tracked 20 pipers for 10±24 s each as they ran over
the swarm surface and measured the rate at which they
piped: 0.700.18 pipes s
±1
(meanSD). We could not
record longer trackings because pipers tend to burrow
into the cluster or switch to waggle dancing.
Fig. 2 A worker bee producing the piping signal that was studied
(``wings-together piping''). Having brie¯y paused from running
over bees in the swarm cluster, the bee presses her thorax to the
substrate, pulls her wings together tightly over her abdomen, arches
her abdomen, and activates her wing muscles to produce a
vibration in the substrate. Although the substrate shown here is a
wooden surface, almost always the substrate is another bee
669

Sound analysis
Figure 4 shows sonagrams for six pipes recorded from
six dierent pipers in swarms 1 and 2. We see that a pipe
consists of a single pulse with a conspicuous upward
frequency sweep. Spectrograms for the early and late
phases of one of these pipes (Fig. 5) show that at the
beginning of a pipe the fundamental frequency is 100±
200 Hz (17722 Hz, n=10 pipes) while at the end it is
200±250 Hz (23620 Hz, n=10 pipes). The harmonic
nature of the later portion of each pipe, at least when
recorded as an airborne sound, is also evident from
Figs. 4 and 5. Analysis of the durations of 50 randomly
chosen pipes from swarms 1 and 2 revealed a meanSD
of 0.820.43 s, a range from 0.09 s to 1.76 s, and a
possibly bimodal distribution with modes at 0.3±0.5 s
and 1.1±1.3 s (Fig. 6).
Fig. 3 Travel patterns, from video recordings, of three piping
workers as they ran over the surface of their swarm cluster. Each
bee's record depicts a 15-s segment of her bout of piping; ®lled
circles beside a bee's track denote when she piped. The time (min-s)
inside the box at the start of each bee's record denotes the time until
lifto
Fig. 5 Spectrograms for the
early and late phases of a
typical piping signal produced
in a swarm shortly before lifto.
Notice the rise in the funda-
mental frequency and, simulta-
neously, the striking
appearance of numerous high-
frequency harmonics, toward
the end of this pulse of sound. It
is these harmonics that give this
signal its characteristic high-
pitched sound
Fig. 4 Sonagrams of six piping signals recorded from workers in a
swarm shortly before lifto
670

Worker piping and swarm warming
Colony-level analysis
Figure 7 shows records of worker piping and swarm
temperature for the 3-h period preceding lifto in
swarms 1 and 2. We see that in both swarms the per-
centage time that piping was audible rose from 0% to
100%, with most of the increase occurring during the
last hour. Also, in both cases, the rise in the level of
piping was not steady, though the main trend during the
®nal hour was de®nitely upward. In swarm 1, the period
of continuous piping lasted for about 10 min before
lifto while in swarm 2 it lasted for nearly 30 min. We
also see in both swarms that the core temperature (T
c
)
and the mantle temperature (T
m
) were kept well above
the ambient temperature (T
a
), and that T
c
was kept
consistently above T
m
. Notice that at lifto T
c
and T
m
had risen to their highest levels, 35.5±36.5°C. In Table 1
we summarize the temperatures recorded at lifto for all
12 liftos that we studied. Except for the 2 liftos ob-
served on 6 September with swarm 2, lifto occurred
only after T
m
had risen above 35.0°C.
Figure 7 also shows that there was a marked associ-
ation between the rise in piping and the rise in temper-
atures in both swarms. T
c
rose noticeably throughout
the lengthy period when piping was audible. Moreover,
T
m
rose dramatically during the ®nal 10- to 30-min pe-
riod when piping was loud and continuous. In both
cases, T
m
tended to drop brie¯y when T
c
began to rise
(and piping was starting). These drops in T
m
seemed to
occur because when each swarm began to raise its T
c
,it
contracted its cluster slightly (presumably to reduce
convective heat loss) and this contraction caused some
thinning of the layer of bees covering the T
m
probe.
Individual-level analysis
Who are the pipers? There are two indications that
some, perhaps all, of the pipers are nest-site scouts. The
®rst line of the evidence arose in the course of following
individual pipers using our video recordings from swarms
1 and 2. Out of 20 pipers that were followed for at least
10 s, we observed 8 switch from piping to waggle
dancing. An example of this is shown in Fig. 8. Because
the waggle dances that the pipers performed were always
for the chosen nest site, and because only nest-site scouts
perform dances for a nest site, we conclude that at least
40% of these pipers were nest-site scouts.
The second line of evidence came from following in-
dividual waggle dancers ± hence bees that were nest-site
scouts ± in swarm 3. We followed each dancer for 10 s
and noted what, if any, change in behavior she made
during this time period. In this way we determined a set
of behavioral transition probabilities for waggle dancers
at various half-hour intervals leading up to lifto. The
results, shown in Table 2, were similar for both liftos of
swarm 3. Initially, waggle dancers were persistent in
their dancing; the probability of continuing to waggle
dance was high (0.90 or higher) and the probabilities of
transition to other behaviors were low. Gradually,
however, the probability of continuing to waggle dance
declined, ultimately falling below 0.40 in the period just
before lifto. And as the waggle dancers became less
Fig. 6 Distribution of the duration of worker piping signals
Fig. 7 Pattern of worker piping (®lled circles), swarm temperature
(open circles and triangles), and ambient temperature (crosses)
during the 3-h period preceding lifto for two swarms
671

persistent in their dancing, they became more likely to
switch to producing other signals: some shaking, some
buzz running (though only in the ®nal half hour before
lifto), and a good deal of piping. Note that in the
second set of observations, the probability of a waggle
dancer switching to piping rose from 0.00 (between
0830 hours and 0900 hours) to nearly 0.25 (between
1000 hours and 1034 hours) just before lifto. Evident-
ly, as lifto approached the nest-site scouts became less
and less motivated to stimulate their fellow scouts to
visit the nest site and more and more motivated to
stimulate their non-scout comrades to prepare for the
lifto.
Experimental test
We tested the hypothesis that the function of worker
piping in swarms is to stimulate bees to warm up for
lifto by performing an experiment in which we pre-
vented the pipers from making direct contact with a
group of mantle bees and recorded the temperature in-
side this group of mantle bees. We prevented contact
between pipers and mantle bees by closing, shortly be-
fore lifto (as the piping became intense), the cover of a
wire-screen cage which enclosed the mantle bees. Si-
multaneously, we treated a second and equal-sized
group of mantle bees exactly the same as the ®rst group,
except that the cover to the second group's cage con-
tained a large opening through which pipers could pass
(see Fig. 1) We predicted, from our hypothesis, that the
mantle bees in the closed cage would not warm them-
selves to a ¯ight-ready temperature at the time of lifto,
whereas those in the open cage would do so.
We performed four trials of this experiment, two each
with swarms 2 and 3. To control for any dierences be-
tween left and right cages, we swapped the locations of
the open and closed cages between the two trials with
each swarm. Results are shown in Fig. 9. In all four tri-
als, T
m
for the open-cage bees showed the typical pattern
of a dramatic rise to approximately 35°C in the ®nal
minutes before lifto, but T
m
for the closed-cage bees did
not show this pattern. Furthermore, at the end of each
trial, once all the uncaged and open-cage bees had taken
o, we removed the cover of the closed cage and found
that the bees inside were basically immobile. When
prodded, they dropped to the ground instead of ¯ying
o, demonstrating that they hadn't warmed for ¯ight.
Discussion
Form and function of worker piping in swarms
Nearly 50 years ago, Lindauer (1955) reported that a
high-pitched sound is emitted from a swarm shortly
before it breaks up to ¯y to its new home. Now we know
that many, perhaps most, of the bees producing this
acoustic signal are nest-site scouts and that these bees
generate this sound with their wing muscles and transmit
Fig. 8 Example of a bee switching between worker piping and
waggle dancing as she scrambled over the surface of her swarm.
Tick marks along her track denote 1-s intervals. Black dots mark
times when she piped. This record began 2 min 45 s before lifto
and lasted for 62 s
Table 1 Temperatures (T,°C)
recorded at the start of 12 lift-
os observed with 3 swarms
(nd no data)
Swarm Date Time T
core
T
mantle
T
ambient
1 28 Aug 18.03 nd 35.2 25.5
29 Aug 09.24 nd 35.2 25.2
29 Aug 11.45 nd 35.4 25.3
29 Aug 14.02 nd 35.2 25.4
29 Aug 16.32 35.7 35.0 25.4
30 Aug 14.24 36.9 36.5 27.6
2 3 Sept 16.58 36.1 35.0 23.9
5 Sept 13.40 36.1 35.1 15.8
6 Sept 12.49 35.2 34.0 nd
6 Sept 14.46 36.0 34.8 nd
3 Sept 19 15.19 36.0 35.2 22.6
20 Sept 10.34 35.9 35.2 25.1
672

it to others by pressing the thorax onto other bees.
Moreover, we have found that the form of this acoustic
signal is rather unusual for insects in that there is a
marked modulation of the fundamental frequency in the
midst of each sound pulse. Another example of this
phenomenon also occurs in honey bees: the frequency
sweep present at the beginning of each tooting signal of
a queen bee (Michelsen et al. 1986a). In the case of the
worker piping signal, the transition to the higher fun-
damental frequency is accompanied by the appearance
of numerous harmonics in the range of 400±2000 Hz
(Fig. 4). No doubt it is these harmonics that give the
worker piping signal its characteristic high-pitched
sound (to humans). Assuming this upward frequency
sweep is functional ± for example, it may help distin-
guish it from other forms of worker piping (see below) ±
we are faced with questions regarding the production
and perception of this frequency modulation. Perhaps it
is produced by changing the resonant properties of the
thorax, which is the mechanism proposed by Nachtigall
and Wilson (1967) to explain how ¯ies control their
wingbeat frequency. If so, then it is tempting to specu-
late that the reason that bees pull their wings together
while producing this acoustic signal (see Fig. 2) is to
stien the whole thoracic mechanism, thereby raising its
resonant frequency.
What is the function of worker piping in swarms
shortly before lifto? Camazine et al. (1999) suggested
that this signal plays a role in triggering lifto. Our re-
sults are most consistent with a slightly dierent hy-
pothesis for this signal's function: worker piping in
swarms plays a role in preparing for lifto. These
preparations include getting each bee warmed to a ¯ight-
ready temperature (Heinrich 1981), and may also
include getting each bee properly groomed, fueled, and
sensitized to the buzz-run signal, which we suspect is the
signal that actually triggers lifto.
Three lines of evidence support the hypothesis that the
function of worker piping in swarms is to stimulate bees
to prepare for lifto. The ®rst is that the time-course of
worker piping is strikingly coincident with that of swarm
warming. We have seen (Fig. 7) that the two phenomena
start together at a low level, typically an hour or so be-
fore lifto, and that they both build to a climax at lifto.
Second, we have seen (Fig. 8, Table 2, and related text)
that many, if not most, of the bees that produce the
piping signal are nest-site scouts. These are the bees that
choose the swarm's future home, so presumably they are
the bees that can sense most easily when a choice has
been made, hence when it is time to begin preparing for
lifto. In other words, the senders of the piping signal are
especially well suited to initiate lifto preparations.
Third, the results of our experiment are consistent with
the hypothesis that worker piping stimulates swarm bees
to warm up in preparation for lifto. We have seen
(Fig. 9) that when pipers are excluded from the mantle
bees of a swarm, the mantle bees do not warm up. We
recognize that our experiment excluded all signals that
require direct contact between sender and receiver, not
just the worker piping signal, so we cannot conclude that
worker piping is the only signal that stimulates warm up.
In particular, the shaking signal was also excluded so it is
possible that this signal also helps stimulate swarm bees
to get ready for lifto, as has been suggested by Visscher
et al. (1999) and Lewis and Schneider (2000). However,
because the shaking signal, unlike the piping signal, does
not occur solely or even primarily during the last hour
before lifto± the warm-up period± it appears that the
activational eects of the shaking signal are less speci®c
to lifto preparations than are those of the piping signal.
Table 2 Probabilities of transition from waggle dancing to some other behavior (including more waggle dancing) over a 10-s observation
period, for various half-hour intervals prior to lifto. Note that the list of transition behaviors is complete, that is, it includes all the
behaviors adopted within 10-s by a waggle dancing bee on a swarm
Transition to: Time interval (hours)
1230±1300 1300±1330 1330±1400 1400±1430 1430±1500 1500±1519
Swarm 3, 19 September; lifto at 1519 hours
Waggle dancing 0.94 0.80 0.54 0.64 0.58 0.38
Walking/running 0.04 0.20 0.30 0.28 0.26 0.38
Piping 0.00 0.00 0.10 0.02 0.06 0.12
Buzz running 0.00 0.00 0.00 0.00 0.00 0.08
Shaking 0.00 0.00 0.00 0.00 0.06 0.00
Flying 0.02 0.00 0.06 0.06 0.04 0.02
Transition to: Time interval (hours)
0730±0800 0800±0830 0830±0900 0900±0930 0930±1000 1000±1034
Swarm 3, 20 September; lifto at 1034 hours
Waggle dancing 0.90 0.92 0.90 0.76 0.62 0.32
Walking/running 0.08 0.08 0.06 0.18 0.18 0.34
Piping 0.00 0.00 0.00 0.02 0.10 0.24
Buzz running 0.00 0.00 0.00 0.00 0.00 0.06
Shaking 0.00 0.00 0.00 0.04 0.02 0.00
Flying 0.02 0.00 0.04 0.00 0.08 0.04
673

The decisive test of our hypothesis for the function of
worker piping will be playback experiments in which
mantle bees of a swarm receive only piping signals. If
these signals induce the mantle bees to warm themselves
and perhaps make other preparations for lifto
(grooming, etc.), then it will be certain that working
piping does stimulate swarm bees to prepare for lifto.
Comparison of dierent forms of worker piping
Several investigators have reported worker piping in
contexts other than the one that we have studied, in a
swarm shortly before lifto. Armbruster (1922) ®rst
described worker piping, based on bees that he observed
in a hive, and named this signal for its similarity to the
well-known piping of queen honey bees (reviewed by
Kirchner 1993a). Subsequent descriptions by O
Èro
Èsi-Pa
Âl
(1932), Ohtani and Kamada (1980), and Pratt et al.
(1996) detailed the posture assumed by a worker piping
in a hive: she presses her thorax to the comb, lifts her
abdomen, raises her wings and spreads them slightly
(making an angle of about 40°), and vibrates her wings
to emit a loud sound. These authors also report that a
piping worker repeats this behavior at a rate of 1±6 pi-
pes min
±1
, with each pipe lasting 0.2±2.2 s, and that the
sound she produces is a single pulse with little or no
frequency modulation and a fundamental frequency of
300±400 Hz.
The form of worker piping that Armbruster (1922),
O
Èro
Èsi-Pa
Âl (1932), Ohtani and Kamada (1980), and Pratt
et al. (1996) have described appears to be dierent from
Fig. 9 Results of four trials of
the experiment that tested
whether the mantle bees in a
swarm cluster are stimulated to
warm themselves to a ¯ight-
ready temperature (ca. 35°C) by
receiving signals from piping
workers. In all trials, the mantle
bees in the closed cage (see
Fig. 1) were prevented from
being contacted by piping
workers and did not raise their
temperature to 34±35°C at lift-
o, while the mantle bees in the
open cage were contacted by
piping workers and did raise
their temperature to 34±35°Cat
lifto
674

the one that we have studied. Although both forms of
worker piping are evidently produced by rapid con-
tractions of the thoracic muscles, are transmitted by the
bee pressing her thorax onto her substrate, and consist
of single pulses of sound lasting from 0.2 s to about
2.0 s, they dier in several ways. The most conspicuous
are how they sound to a human observer and how the
worker's wings are positioned during sound production.
The form of worker piping that we have studied is a
frequency-modulated sound that changes from a rela-
tively pure, low-frequency (100±200 Hz) tone to a
mixed, higher-frequency (200±2000 Hz, including har-
monics) sound, and is reminiscent of the revving of a
racing car's engine. While producing this form of piping,
the worker pulls her wings together tightly (Fig. 2). In
contrast, the form of worker piping that others have
studied is primarily a low frequency (300±400 Hz) sound
with little, if any, frequency modulation, and is remi-
niscent of the bleating of sheep. While producing this
form of worker piping, the worker spreads her wings
slightly. For clarity in discussing these two forms of
worker piping, we will call the former ``wings-together
piping'' and the latter ``wings-apart piping''.
Besides their conspicuous dierences in sound pitch
and wing position, these two forms of worker piping
dier with respect to the speed of worker movement
while piping (wings-together piping, worker is usually
running just before piping; wings-apart piping, worker is
usually walking just before piping), and the rate of signal
production (wings-together piping, 30±60 pipes/bee/min;
wings-apart piping, 1±15 pipes/bee/min).
Given that the wings-together and wings-apart forms
of worker piping dier markedly in their acoustical
properties and production patterns, it is likely that they
also dier in signal message (the information provided
by the sender) and signal meaning (the response gener-
ated by the receiver) (Smith 1977 discusses signal mes-
sage and meaning). As we have seen, wings-together
pipes are produced by nest-site scouts in swarms as they
prepare to lifto, hence the message of wings-together
pipes seems to be something like ``get ready for ¯ight.''
Given that this signal evidently elicits a warm-up re-
sponse, its meaning to the non-scouts in the swarm
cluster seems to be something like ``warm up in prepa-
ration for ¯ight''. Wings-apart pipes are produced by
various types of workers in diverse contexts, including
egg-laying workers in queenless colonies (Ohtani and
Kamada 1980) and pollen foragers, water collectors, and
water receivers in queenright colonies (Pratt et al. 1996).
It is unclear what the various contexts associated with
wings-apart piping have in common, hence what con-
ditions cause bees to produce wings-apart pipes. More-
over, it is unclear how workers respond to wings-apart
pipes. Thus, the message and the meaning of wings-
apart pipes remain mysterious.
There is a third piping sound which workers produce
with their thoracic muscles and transmit through the
substrate: the ``begging signal'' (Esch 1964; Michelsen
et al. 1986b) or ``stop signal'' (Nieh 1993; Kirchner
1993b). We will use the latter name, because it more
accurately describes this signal's function. This signal is
acoustically distinct from both forms of worker piping
discussed above. Although the stop signal has the same
fundamental frequency as wings-apart pipes (300±
400 Hz), its sounds are much shorter, averaging only
0.14 s and ranging only 0.05±0.2 s (Esch 1964; Michel-
sen et al. 1986a; Kirchner 1993b). Because the sounds of
the stop signal are so brief, they have sometimes been
described as ``beeps'' or ``short squeaks'' rather than
``pipes'' (Kirchner 1993a, 1993b). Furthermore, the in-
formation content of the stop signal is distinct from that
of either of the other two forms of worker piping. The
stop signal is produced primarily by tremble dancers
when they encounter a nestmate performing a waggle
dance for a nectar source (Nieh 1993). Its eect is to
inhibit waggle dancing, thus recruitment, to nectar
sources (Nieh 1993; Kirchner 1993b). Evidently, the
message of the stop signal is ``stop waggle dancing.''
Further studies of worker piping need to focus on
identifying the conditions that cause bees to produce the
wings-apart form of worker piping. It may be that what
we are calling wings-apart piping is really a suite of
distinct signals, which could explain why it has been
dicult to identify a consistent context associated with
this type of worker piping. Deciphering the message(s)
and meaning(s) of the still mysterious, wings-apart form
of worker piping will be a major contribution to a
comprehensive understanding of acoustical communi-
cation in honey bees, the diversity of which has only
begun to be adequately described and analyzed.
Acknowledgements The research reported here was supported by
the U.S. National Science Foundation (grant IBN96-30159), the
U.S. Department of Agriculture (Hatch grant NYC-191407), and
the German Science Foundation (grant Ta 82/8-1 and SFB 554).
We thank David C. Gilley, David Tarpy, and P. Kirk Visscher for
comments on the manuscript, Cole Gilbert for information on the
mechanisms for modulating wingbeat frequency, and Margaret
C. Nelson for producing the ®gures.
References
Adam B (1987) Beekeeping at Buckfast Abbey. Northern Bee
Books, Hebden Bridge
Armbruster L (1922) U
Èber Bienento
Ène, Bienensprache und Bien-
engeho
Èr. Arch Bienenk 4:221±259
Camazine S, Visscher PK, Finley J, Vetter RS (1999) House-
hunting by honey bee swarms: collective decisions and indi-
vidual behaviors. Insectes Soc 46:348±360
Esch H (1964) Beitra
Ège zum Problem der Entfernungsweisung
in den Schwa
Ènzelta
Ènzen der Honigbiene. Z Vergl Physiol
48:534±546
Esch H (1967) The sounds produced by swarming honey bees.
Z Vergl Physiol 56:408±411
Heinrich B (1979) Thermoregulation of African and European
honeybees during foraging, attack, and hive exits and returns.
J Exp Biol 80:217±229
Heinrich B (1981) The mechanisms and energetics of honeybee
swarm temperature regulation. J Exp Biol 91:25±55
Kirchner WH (1993a) Acoustical communication in honeybees.
Apidologie:24:297±307
675

Kirchner WH (1993b) Vibrational signals in the tremble dance
of the honeybee, Apis mellifera. Behav Ecol Sociobiol 33:169±
172
Lewis LA, Schneider SS (2000) The modulation of worker be-
havior by the vibration signal during house hunting in swarms
of the honeybee, Apis mellifera. Behav Ecol Sociobiol 48:154±
164
Lindauer M (1955) Schwarmbiene auf Wohnungssuche. Z Vergl
Physiol 37:263±324
Lindauer M (1971) Communication among social bees. Harvard
University Press, Cambridge
Martin P (1963) Die Steuerung der Volksteilung beim Schwa
Èrmen
der Bienen. Zugleich ein Beitrag zum Problem der Wander-
schwa
Èrme. Insectes Soc 10:13±42
Michelsen A, Kircher WH, Andersen BB, Lindauer M (1986a)
The tooting and quacking vibration signals of honeybee
queens: a quantitative analysis. J Comp Physiol A 158:605±
611.
Michelsen A, Kirchner WH, Lindauer M (1986b) Sound and vib-
rational signals in the dance language of the honeybee, Apis
mellifera. Behav Ecol Sociobiol 18:207±212
Mitchell C (1970) Weights of workers and drones. Am Bee
J 110:468±469
Nachtigall W, Wilson D (1967) Neuro-muscular control of dipt-
eran ¯ight. J Exp Biol 47:77±97
Nieh JC (1993) The stop signal of honey bees: reconsidering its
message. Behav Ecol Sociobiol 33:51±56
Nieh JC (1998) The honey bee shaking signal: function and design
of a modulatory communication signal. Behav Ecol Sociobiol
42:23±36
Ohtani T, Kamada T (1980) `Worker piping': the piping sounds
produced by laying and guarding worker honeybees. J Apic Res
19:154±163
O
Èro
Èsi-Pa
Âl Z (1932) Wie tu
Ètet die Arbeitsbiene? Zool Anz 98:147±
148
Pratt SC, Ku
Èhnholz S, Seeley TD, Weidenmu
Èller A (1996) Worker
piping associated with foraging in undisturbed queenright
colonies of honey bees. Apidologie 27:13±20
Schneider SS, Stamps JA, Gary NE (1986) The vibration dance of
the honey bee. I. Communication regulating foraging on two
time scales. Anim Behav 34:377±385
Seeley TD (1982) How honeybees ®nd a home. Sci Am 247:158±168
Seeley TD, Buhrman S (1999) Group decision making in swarms of
honey bees. Behav Ecol Sociobiol 45:19±31
Seeley TD, Morse RA, Visscher PK (1979) The natural history of
the ¯ight of honey bee swarms. Psyche 86:103±113
Seeley TD, Weidenmu
Èller A, Ku
Èhnholz S (1998) The shaking signal
of the honey bee informs workers to prepare for greater activity.
Ethology 104:10±26
Smith WJ (1977) The behavior of communicating. An ethological
approach. Harvard University Press, Cambridge
Visscher PK, Shepardson J, McCart L, Camazine S (1999) Vibra-
tion signal modulates the behavior of house-hunting honey bees
(Apis mellifera). Ethology 105:759±769
676