Content uploaded by Thomas Seeley
Author content
All content in this area was uploaded by Thomas Seeley on Nov 24, 2017
Content may be subject to copyright.
Research Papers
The Grooming Invitation Dance of the Honey Bee
Benjamin B. Land & Thomas D. Seeley
Department of Neurobiology and Behavior, Cornell University, Ithaca, NY, USA
Abstract
The grooming invitation dance is a striking behavior in honey bee colonies
that has not been extensively studied. The objectives of this study were (1) to
describe the dance through video analysis, (2) to test the functional hypothesis
that it is a grooming solicitation signal, and (3) to analyze the stimuli that cause its
production. A worker bee producing the grooming invitation dance stands
stationary and vibrates her whole body from side-to-side at a frequency of
4.2 0.2 Hz for 9.3 1.0 s. Sometimes the bee mixes bouts of body vibration
with brief bouts of self-grooming (average duration ¼1.4 s). Bees that perform
the grooming invitation dance have a far higher probability of being quickly
groomed by a nest mate than do bees that do not perform the dance. Bees that
had chalk dust puffed onto the bases of their wings produced significantly more
grooming invitation dances than did control bees that received only puffs of air.
This shows that it may be the accumulation of small particles at the bases of the
wings that normally triggers the dance. We suggest that the evolutionary origin of
this signal is self-grooming behavior.
Corresponding author: Thomas D. Seeley, Department of Neurobiology and
Behavior, Cornell University, Ithaca, NY 14853, USA. E-mail: tds5@cornell.edu
Introduction
In communicating with nest mates, a worker honey bee (Apis mellifera) has at
her disposal a wide array of signals, several of which have been called dances
(reviewed in Frisch 1967; Seeley 1998). Some of these dances, like the waggle dance
and the tremble dance, have been intensively studied and are well-understood, but
others are little studied and remain obscure, though probably they are no less
important to colony functioning. One such dance is the grooming invitation dance.
Haydak (1945) published the first account of this dance behavior. He described it as
‘consisting of a rapid stamping of the legs and a rhythmic swinging of the body to
the sides. At the same time the bee rapidly raises and lowers the body and tries to
clean around the bases of the wings with the middle pair of legs.’ He also noted that
Ethology 110, 1—10 (2004)
Ó2004 Blackwell Verlag, Berlin
ISSN 0179–1613
U. S. Copyright Clearance Center Code Statement: 0179-1613/2004/11001–001/$15.00/0 www.blackwell-synergy.com
‘usually the bee which is closest to the dancer touches the latter with its antennae
and begins to clean the dancer.’ Milum (1955) independently observed this
behavior and provided a similar description: ‘a rapid shifting of the body,
alternatively to the right and to the left, while one of the second pair of legs is
thrown upward beside the thorax and drawn down over the hair in a comb-like
motion.’ Neither Haydak nor Milum filmed the grooming invitation dance, so they
were unable to make a detailed analysis of the movements making up this behavior.
Also, neither investigator studied what causes a worker to perform this dance or
demonstrated quantitatively what effect it has on nest mates. They simply
suggested that the bee performs this activity when she feels the need for grooming
and noted that usually it elicits grooming by a nearby bee.
For 40 yrs after Milum’s paper, no further work was done on the grooming
invitation dance. Finally, Bozic & Valentincic (1995) renewed its study while
studying grooming behavior. They video recorded 10 bees performing the
grooming invitation dance and reported two quantitative features of this curious
behavior: the median duration of a dance is rather short (just 8 s) and the
probability that a bee that has performed a grooming invitation dance will soon
be groomed by a nest mate is remarkably high (0.72). The latter result suggests
strongly that this behavior is a signal whereby a bee solicits grooming from
another bee (allogrooming). Recently, Pettis & Pankiw (1998) reported on the
grooming invitation dance as they addressed the question of why bees allogroom.
They examined allogrooming in the context of the tracheal mite (Acarapis woodi),
for they were testing the hypothesis that the frequency of the grooming invitation
dance and subsequent allogrooming is a function of the mite load of the colony.
These authors did not focus specifically on the nature of the grooming invitation
dance, but they did conclude that social grooming, including the grooming
invitation dance, is a positive function of the tracheal mite load in a colony.
The present study provides the first thorough description of the grooming
invitation dance. A detailed description of the motor patterns of this dance is
valuable not only to document the basic phenomenon under investigation, but
also to gain insights into the evolutionary origins of this behavior. This study also
rigorously tests the hypothesis that this behavior is a signal indicating a bee’s need
for allogrooming. We do so by checking whether a bee that performs the
grooming invitation dance has a higher probability of being groomed by a nest
mate relative to a bee that does not perform the dance. Finally, this study tests a
hypothesis on what causes the grooming invitation dance. Pettis & Pankiw (1998)
found that infestation by tracheal mites might be one cause of social grooming
(i.e. the dance and the allogrooming it elicits). However, because social grooming
occurs in all colonies, whether or not they have mites, there must be another,
more general explanation for the occurrence of grooming invitation dances. We
test the hypothesis that small particles of debris that have become stuck in the
hairs at the bases of the wings can stimulate a worker to perform the grooming
invitation dance. These small particles might create mechanical problems,
possibly reducing the energetic efficiency of wing movements for flight or nest
ventilation (Winston 1987). A worker cannot groom around the bases of her
2B. B. Land & T. D. Seeley
wings, hence she needs to be allogroomed when these regions become dirty. To
express her need for allogrooming, she must have a signal, which seems to be the
grooming invitation dance.
Methods
Description of Dance
One observation hive colony, located at the Liddell Field Station of Cornell
University, was used in this part of the study. Grooming invitation dances were
recorded with a video camera (Panasonic WV-F250B; Panasonic, Tokyo, Japan)
with docking videocassette recorder (Panasonic AG-7450) equipped with a time
code generator. The camera was set up 10 cm from the glass wall of the observation
hive; this gave a 9 cm 6 cm field of view. A grid of 5 mm 5 mm squares was
drawn on the glass in the camera’s field of view to provide precise information about
where a dancing bee was located in two dimensions. We made recordings for
approximately 3.5 h, and watched the hive at the same time, noting when a worker in
the camera’s field of view was performing the grooming invitation dance.
Altogether, 30 dances were recorded. These were then analyzed by several different
ways during playback using a videocassette player with variable-speed playback
(JVC BR-S525U; JVC, Tokyo, Japan). (1) The starting orientation of each dancing
bee was noted as the angle of the bee’s body axis with respect to vertical (head
straight up ¼0). (2) The position of the center of each dancing bee’s thorax was
recorded at 1-s intervals throughout her dance to determine the displacement of the
dancer during her dance. (3) The duration of each dance was measured. (4) The
number of side-to-side body movements that a bee made during her dance was
determined by going through each dance recording one frame at a time (30 frames/
s). (5) The changing positions of the dancer’s head, thorax, abdomen, and legs were
studied throughout each dance using slow-motion playback.
Probability of Allogrooming Response to Dance
To measure the probability of eliciting allogrooming by performing a
grooming invitation dance, all such dances on the video recording were watched
to see if the dancing bee received allogrooming. In addition, we made direct
observations with three more colonies in observation hives; two in Stimson Hall,
Cornell University, and one at the Liddell Field Station. These colonies were
observed until a worker that began to perform a grooming invitation dance was
noticed. At this point a timer was started and the dancer was watched for 30 s.
(The time limit was set at 30 s because previous observations had shown that few
dances last longer than 30 s). If a nestmate began grooming the dancer within the
30-s interval, a positive response was recorded, but if no nestmate began
grooming the dancer within the 30-s interval, a negative response was recorded.
To see if allogrooming occurred without the grooming invitation dance, a non-
dancing worker was selected at random from among the bees within 5 cm of the
3
Grooming Invitation Dance
dancer that had just been observed and this randomly selected bee was likewise
watched for 30 s. These observations were repeated for 65 pairs of workers (one
performing and one not performing the dance).
Particle Puffing Experiment
One observation hive colony was used at the Liddell Field Station. The glass
wall covering one frame of the hive was removed. One worker at a time was
selected at random, and 0.5 mg of chalk dust was puffed onto the base of her
wings using a Pasteur pipette equipped with a rubber bulb. To help standardize
the stimulus, the rubber bulb was squeezed uniformly and just once. The worker
was then followed for 10 min to see if she groomed herself or performed the
grooming invitation dance, or both. As a control, another worker was selected at
random and this time just air was puffed onto the base of her wings with an empty
Pasteur pipette. This worker was also followed for 10 min to see if she groomed
herself or performed the grooming invitation dance. We tested 30 pairs of workers
in this manner. We used a 10-min observation-period in this experiment because
we found in pilot work that we could reliably follow an unmarked bee for this
amount of time. This short observation period renders our test for effects of
particles a conservative one because any bee that waited longer than 10 min
before performing the grooming invitation dance was missed.
Statistics
To test whether dancers have a non-random body orientation (with respect
to gravity, on the vertical comb) when starting to dance, the Rayleigh test was
used (Batschelet 1965). To test the significance of the differences observed in the
response probability observations (between dancing and non-dancing bees) and in
the particle puffing experiment (between treatment and control bees), the Fisher’s
exact test was used. All descriptive statistics are reported as the mean 1SE.
Results
Description of Dance
A bee performing the grooming invitation dance stands with her legs spread
and tightly gripping the comb while rocking her whole body side-to-side in a plane
parallel to the comb (Fig. 1). The amplitude of these rocking movements is a bit
larger for the abdomen than for the head or thorax, so the bee sweeps out an arc
with her side-to-side body movements. The mean duration of the grooming
invitation dances that we observed was 9.3 1.0 s (n ¼61, range ¼1–30 s). A
frame-by-frame analysis of 25 complete dances revealed that the frequency of the
side-to-side movements was 4.2 0.2 Hz and that the total number of these
movements per dance was 23.6 3.3. There was no consistent body orientation
of bees starting to dance (Fig. 2) (p ¼0.10, n ¼30). Also, there was little net
4B. B. Land & T. D. Seeley
Fig. 1: The grooming invitation dance (left) and the allogrooming response that it elicits (right)
r = 0.24
Fig. 2: Distribution of body orientations at the start of dancing. Head upward ¼0. The arrow
represents the mean vector
5Grooming Invitation Dance
movement of a dancer across the comb while dancing; from start to finish, the
center of the thorax moved only 5.6 0.5 mm horizontally and 6.1 0.5 mm
vertically (n ¼30 dances).
In a significant fraction of the cases (33.3%, n ¼30), the dancer paused
briefly in her side-to-side body vibrations to groom (Fig. 3). On average, a pause
for self-grooming lasted only 1.4 0.2 s (n ¼13). To groom herself, the dancer
stopped rocking her body back and forth, raised her abdomen while lowering her
head, and cleaned her abdomen with her hind legs. Often the abdomen was also
pivoted sideways slightly, evidently to facilitate grooming one side of the
abdomen. Occasionally, a dancer would use her fore- and midlegs to groom her
head and thorax, respectively. Self-grooming performed within a grooming
invitation dance appeared more frenetic than regular self-grooming. Even during
grooming invitation dances that were unbroken (i.e. the dance in which the side-
to-side body vibration was continuous), there was some grooming of the head and
thorax with the fore- and midlegs, and the abdomen was occasionally pivoted
sideways so that the worker could groom her abdomen with her hindlegs as she
rocked her body back and forth.
During all 30 dances that were analyzed closely, whether broken or
unbroken, adjacent workers antennated the dancer. Dances ended either when
another worker mounted the dancer and began to allogroom her (n ¼19) (Fig. 1)
or when the dancer simply stopped vibrating (n ¼11). In all 19 cases where the
dancer was allogroomed, she was mounted while she was vibrating, never while
she was grooming herself. Immediately upon being mounted, the dancer always
ceased to dance and often spread one or both wings (see Fig. 1), evidently to
provide the groomer with easy access to the thorax. Also, in all cases where the
dancer was allogroomed, the groomer was contacted by the rocking dancer before
Dancer: 12:47.12
Dancer: 13:24.42
Dancer: 13:50.04 Vibrating
Self-grooming
Allogrooming began
Allogrooming began
1 s
Fig. 3: Time lines of dances. Each line represents one worker’s dance. Bees are identified by the time
they danced. These examples show the alternation between bouts of vibrating and self-grooming. In
two of the three cases, the worker was allogroomed
6B. B. Land & T. D. Seeley
the groomer mounted the dancer. In only one instance of allogrooming was the
groomer more than 1 cm from dancer when the dancer began dancing.
Probability of Allogrooming Response to Dance
Of 65 bees that were observed performing the grooming invitation dance, 45
received allogrooming within 30 s of the start of dancing (see Table 1). The
probability of a grooming invitation dance being followed by allogrooming was,
therefore, 0.69. In the 65 cases in which a bee that was not performing a grooming
invitation dance was observed for 30 s, no allogrooming was seen. This difference
in response is highly significant (p < 0.00001).
Particle Puffing Experiment
Randomly selected workers that were puffed with chalk dust and air
groomed themselves in 28 of 30 trials and performed the grooming invitation
dance in 15 of 30 trials (see Table 2). Randomly selected workers that were puffed
with only air groomed themselves in six of 30 trials and performed the grooming
invitation dance in just one of 30 trials. The difference in response between the
two treatments is highly significant for both self-grooming (p < 0.00001) and
dancing (p < 0.00004).
Discussion
Our description of the grooming invitation dance confirms the reports of
Haydak (1945) and Milum (1955) who described the dance as a burst of rapid,
Table 1: Results of the test of an allogrooming response to the grooming invitation dance
Type of bee
Allogrooming response?
Yes No
Dancer 45 20
Non-dancer 0 65
Table 2: Results of the particle puffing test on self-grooming response and on grooming
invitation dance response
Treatment
Self-grooming response? Dance response?
Yes No Yes No
Chalk dust and air 28 2 15 15
Air only 6 24 1 30
7Grooming Invitation Dance
side-to-side, body movements that are often punctuated by bouts of energetic self-
grooming. However, our analysis, based on video recording, provides a couple of
clarifications about the form of this behavior. First, the dancer does not steadily
raise her midlegs as she vibrates, as was stated by Haydak (1945) and Milum
(1955). Instead, she raises her midlegs only occasionally when vibrating, and
mostly when she is in a bout of self-grooming. Secondly, many times the
grooming invitation dance lasts only a few seconds, whether or not it elicits
allogrooming. In contrast, Haydak (1945) reported that if the dancer does not get
allogroomed, she will continue dancing, and at a more vigorous pace, until
attention is given. These differences between our description and those of Haydak
and Milum probably reflect the fact that the previous investigators were not able
to record and review dances.
This study also confirms the report of Bozic & Valentincic (1995) that
grooming invitation dances are short, with a typical duration of only a few
seconds. The average duration that we recorded for this dance (just 9.3 s) is far
less than the other dances of the honey bee, such as the waggle dance, which can
last as along as 10+ min (Seeley 1995), and the tremble dance, whose mean
duration is 27 min (Seeley 1992). The grooming invitation dance is shorter than
the others, probably, because a bee performing it seeks to stimulate just one
nearby worker, whereas a bee performing the waggle dance or the tremble dance
seeks to activate numerous workers (Frisch 1967; Seeley 1992). Likewise, the fact
that the grooming invitation dance, unlike the waggle dance and the tremble
dance, is performed in a small area (<1 cm2) by an essentially stationary bee, no
doubt reflects the fact that any one of the bees adjacent to the dancer can
allogroom her. In short, the grooming invitation dance seems to be a signal that
does not need to be transmitted broadly.
Our observations on the tight association between grooming invitation
dances and allogrooming confirm the previous work of Bozic & Valentincic
(1995), who observed that the grooming invitation dance elicited an allogrooming
response 72% of the time. The response probability that we found was virtually
the same (69%). Additionally, we found that workers who perform the grooming
invitation dance receive far more allogrooming than those who do not perform
this dance. Indeed, workers that did not perform the grooming invitation dance
were never allogroomed. It seems clear, therefore, that by performing this dance, a
worker is producing an effective grooming solicitation signal.
Our particle puffing experiment tested the hypothesis that small particles,
such as chalk dust, on a bee can stimulate a bee to produce the grooming
invitation dance. This experiment also tested whether the dust will stimulate self-
grooming. The results show clearly that the chalk dust caused the treated bees to
dance; half of the treated bees produced a grooming invitation dance but none of
the control bees did so. Similarly, the treated bees were far more likely to groom
themselves than were the control bees. The fact that chalk dust stimulated the
grooming invitation dance lends considerable support to the particle hypothesis
for the cause of the grooming invitation dance. Further support comes from the
finding of Pettis & Pankiw (1998) that the probability of grooming dance
8B. B. Land & T. D. Seeley
performance decreases with increasing worker age. The younger hive bees
probably are more subject than the older field bees to the light but steady rain of
wax particles and pollen grains that falls on workers as their nest mates above
them work on the combs and eat pollen. As bits of these materials fall onto
workers, they are likely to stick to their bodies. There are two areas on a worker’s
body – the bases of her wings and the petiole region – that she cannot effectively
groom herself. If the particles are not groomed from the bases of the wings, there
is a good chance that the mechanics of flight or fanning will be disrupted, resulting
in an energetic loss for the bee. By receiving allogrooming, however, these
potential problems are avoided. It should be noted that small particles are not the
only known stimuli eliciting the grooming invitation dance. Evidently other
irritating stimuli, including tracheal mites (Acarapis woodi), will also stimulate this
behavior. Pettis & Pankiw (1998) examined the relationship between the
frequency of this dance and mite infestation among four lines of honey bees
and found that the line with the most dancing had the lowest mite infestation.
Danka & Villa (1998) have shown that this relationship arises because tracheal
mites do indeed elicit the grooming invitation dance and that the self-grooming
associated with the dance lowers a bee’s mite infestation.
The question remains: How does a bee detect the grooming invitation dance?
There are several possibilities. One is that workers adjacent to the dancer detect
the air movements produced by the dancer’s body vibrations. Another is that
nearby workers contact the dancer and so detect her movements directly. Finally,
detection may involve substrate vibrations; the dancer may load much of the
energy of her body vibrations into the comb and thereby transmit the signal to
nearby bees. It is noteworthy that the body vibrations associated with this dance
have a lower frequency (about 4 Hz) than those of the waggle dance (approx.
15 Hz; Esch 1961; Wenner 1962). It is also interesting that every worker that
allogroomed a dancer was seen to have direct contact with the dancer, as would be
predicted by the second hypothesis. A solid answer to the mystery of how the
grooming invitation dance is sensed will, however, require further study.
Lastly, it is tempting to speculate on the evolutionary origins of the grooming
invitation dance. Because this signal has some similarity to excited self-grooming
and is often combined with it, we suggest that this dance evolved from the
movements used in self-grooming. Perhaps originally all allogrooming occurred
by means of workers inspecting other bees and grooming them as necessary, a
phenomenon that still exists today (Bozic & Valentincic 1995). Then, at some
point, perhaps workers evolved an allogrooming response to any worker that was
grooming herself extremely energetically. (Such workers were likely to be ones
having difficulty getting clean). Since sender and receiver are nest mates, both
would benefit from this new response, and so natural selection would favor
refinements both in signal production (stronger exaggeration of the self-grooming
behavior by bees needing assistance) and in signal reception (better discrimination
of the grooming solicitation signal from other stimuli). The eventual result would
be the remarkable grooming invitation dance and the reliable allogrooming
response that we see today.
9
Grooming Invitation Dance
Acknowledgements
We thank Dave Gilley, Dave Tarpy, and especially Sabiha Barot for their enthusiastic support of
this work, which was done by the senior author in fulfillment of the requirements for an undergraduate
Honors Degree in Biology from Cornell University. Financial support was provided by the US
Department of Agriculture (Hatch grant NYC-191407).
Literature Cited
Batschelet, E. 1965: Statistical Methods for the Analysis of Problems in Animal Orientation and
Certain Biological Problems. American Institute of Biological Sciences, New York.
Bozic, J. & Valentincic, T. 1995: Quantitative analysis of social grooming behavior of the honey bee
Apis mellifera carnica. Apidologie 26, 141—147.
Danka, R. G. & Villa, J. D. 1998: Evidence of autogrooming as a mechanism of honey bee resistance to
tracheal mite infestation. J. Apic. Res. 37, 39—46.
Esch, H. 1961: U
¨ber die Schallerzeugung beim Werbetanz der Honigbiene. Z. vergl. Physiol. 45, 1—11.
Frisch, K. von 1967: The Dance Language and Orientation of Bees. Harvard Univ. Press, Cambridge,
MA.
Haydak, M. H. 1945: The language of the honeybees. Am. Bee J. 85, 316—317.
Milum, V. G. 1955: Honey bee communication. Am. Bee J. 95, 97—104.
Pettis, J. S. & Pankiw, T. 1998: Grooming behavior by Apis mellifera L. in the presence of Acarapis
woodi (Rennie) (Acari: Tarsonemidae). Apidologie 29, 241—253.
Seeley, T. D. 1992: The tremble dance of the honey bee: message and meanings. Behav. Ecol. Sociobiol.
31, 375—383.
Seeley, T. D. 1995: The Wisdom of the Hive. Harvard University Press, Cambridge, MA.
Seeley, T. D. 1998: Thoughts on information and integration in honey bee colonies. Apidologie 29,
67—80.
Wenner, A. 1962. Sound production during the waggle dance of the honeybee. Anim. Behav. 10,
79—95.
Winston, M. L. 1987: The Biology of the Honey Bee. Harvard University Press, Cambridge, MA.
Received: May 14, 2003
Initial acceptance: July 7, 2003
Final acceptance: September 5, 2003 (S. Forbes)
10 B. B. Land & T. D. Seeley
