The Sensory Basis of the Honeybee's Dance Language

Wolfgang H. Kirchner and William F. Towne

Novel experiments, such as training bees to respond to sounds and recruiting them using a robot, have ended several debates surrounding the dance language.

For many centuries, naturalists have observed that honeybees tell their nestmates about discoveries they make beyond the hive. Nevertheless, the system of communication that the insects use remained a mystery until the 1940s, when Karl von Frisch of the University of Munich in Germany first discovered the significance of bees' dances. In the hive the steps and waggles of a successful forager correlate closely with the exact distance and direction from the nest to the resource she has discovered. For the next two decades, most scientists believed bees relied primarily on these silent movements to communicate.

In the 1960s this view was challenged in two ways. The first challengers were Adrian M. Wenner, then a graduate student at the University of Michigan, now at the University of California at Santa Barbara, and Harald E. Esch of the University of Munich, now at Notre Dame University. Working independently, the two researchers discovered that the dances were not silent after all. As the bees dance, they emit faint low-frequency sounds, and Wenner and Esch both suggested that the sounds might play a critical role in the bees' communication. The use of sounds, they reasoned, might account for the bees' ability to communicate effectively in the complete darkness that prevails inside their nests. At the time, however, many scientists believed bees were deaf, and so the issue remained open.

Sight and Sound in the Honeybees' Dance

 bee dance 2 2

FREQUENCY RANGE of the sounds a bee can detect extends well below the range heard by human ears. The graph shows how fast the air particles near a dancer's wings must travel to generate audible signals. Within this range, the bees show an ability to differentiate between sounds having varying frequencies.

bee dance 2 3

THE DANCING ROBOT successfully recruited its nestmates to food away from the hive. The experimenters placed eight baits around the hive and programmed the robot to dance concerning one site. Observers in the field recorded the approach of searching bees. Most of the robot's recruits went to the bait indicated by its dance.

Wenner later raised the second challenge to von Frisch's description of the dance language, rethinking his first hypothesis at the same time. Bees, he argued, use none of the information in the dances or the sounds. Instead he proposed that the insects rely on odors to find the new resource advertised by the dancer.

Now both of these debates have been resolved. Bees, it turns out, can hear, and their ears are well suited for detecting the sounds associated with the dances. Observation of how the insects respond to a robot that dances and sings like a live forager shows that both sound and dance are needed to communicate information about the location of food. Silent dances, the experiment demonstrates, communicate nothing, and sound without dance also fails. Odors too are involved but appear to lack the importance that Wenner ascribes to them. Beyond the resolution of these issues, we have also recently learned much more about the nature of the dance sounds, the bees' sense of hearing and the aspects of the dance that are most essential in the communication.

A certain pleasure comes from the solution of such a long-standing mystery. Aristotle himself documented the honeybee's ability to recruit her nestmates to a good food source but did not speculate on how the communication took place. He and other naturalists did, however,observe that a bee that finds a new food source returns to the nest and dances for her sisters, rather than feasting alone. Pliny reportedly constructed a hive that had a window made from transparent horn, through which he could monitor dancing bees. By studying bees kept in a glass-walled hive, von Frisch and his followers in the 20th century were able to recognize a pattern in the dance: The forager walks across the vertical sheets of comb hanging in the hive and traces out the shape of a figure eight; she pauses in each loop to shake her body from side to side. A few potential recruits chase after the dancer attentively for some time and then fiy out on their own toward the target. Provided they like what they find, these recruits return to the nest and dance as well, sending even more bees from the hive to investigate the site. Eventually, the best food sources inspire the most dances and so attract the most bees.

In the 1920s von Frisch first proposed that the forager's dance somehow gave the other bees information about the food source. But it was not until 1943 that von Frisch discovered that the direction in which the dancer faced during her waggling run pointed toward the food site in relation to the sun: If she waggled while facing straight upward, toward the 12 on a clockface, then the food could be found in the direction of the sun; if she waggled 60 degrees to the left of 12, facing the 10 on a clockface, then the food lay 60 degrees to the left of the sun. In addition, he noticed that how fast the dancer completed her circuits corresponded to the distance between the hive and the feeding site: the closer the food, the more frenzied her pace. Von Frisch and his colleagues made detailed accounts of the dance language. With a stopwatch close at hand, they could observe the dance, decipher its meaning and then locate the food supply of which it spoke. The accurate translations were a stunning accomplishment.

Nevertheless, several serious questions remained unanswered. First, investigators could not prove that the dance truly functioned as a language. Perhaps the correlation between the excited forager's movements and her finds in the field were merely coincidental and of no real significance to the bees themselves. Moreover,researchers had yet to figure out how the bees perceived and interpreted the dances. The scientists could clearly see the dancer through the glass wall and clock her movements, butthe bees themselves, normally enclosed in dark nests, certainly did not have this benefit. What were the dances like from the bees' point of view?

Years after von Frisch interpreted the symbolism of the dances, Wenner and Esch independently found that dancing bees make sounds during their waggling run. Both men suggested that the sounds might help the dancer attract an audience in the dark nest. Many researchersdoubted this premise becausethey thought bees could not hear airborne sounds. Still, the notion was not ignored altogether. Many insects, including bees, are quite sensitive to vibrations. Hence, some investigators speculated that the sounds the foragers produced could vibrate the combs under their feet as they danced. The comb vibrations might then advertise the dance to those bees who could not otherwise see the forager.

One of us (Kirchner), together with Axel Michelsen of Odense University in Denmark, answered part of this question several years ago. In their experiments, Michelsen and Kirchner aimed a laser beam at the comb near a dancing bee to determine whether or not the dance sounds generated vibrations in the comb. Surface vibrations, if any occurred, would cause minute changes in the light refiected from the comb. In this way, it was possible to measure the vibrations without touching the comb and possibly triggering additional tremors. These measurements revealed that dancing bees do not rattle the comb but that their audience does. The dance attenderssometimes emit a short squeak by pressing their thoraxes against the comb. This action vibrates the comb enough so that the dancing bee stops her movements. She then doles out small samples of the food she has collected so that her audience knows not only the direction and distance to the feeding site but how the food smells and tastes as well.

Although this result was intriguing, we still did not know whether honeybees could hear the sounds a forager made during her waggling run. Because the sounds seemed likely to be involved in the dance communication, we have analyzed them thoroughly. Initially, we discovered that the forager made the dance sounds by beating her tiny wings and that the follower bees stood quite close to her when she did so. Therefore, our measurements of the sounds needed to be made in the same proximity.

In further collaboration with Michelsen, wegauged both the pressure changes and the air-particle movements near the dancer. Some insects' ears do not respond to sound pressures as human ears do but instead react to back-and-forth oscillations of nearby air molecules. In fact, it turned out that the pressure changes a dancing bee produces when she beats her wings are relatively small--which no doubt explains why the pressure-sensitive ears of scientists did not detect them for so long. But we found significant air movements within a few millimeters of the dancer's vibrating wings. We concluded that whereas the dance followers produce sounds by vibrating the comb, the forager transmits her own sound signals exclusively through the air. As a result, the noises from the forager's instructions and her followers' requests do not drown one another out.

Two research teams, using different experimental approaches, then tested the hypothesis that the forager transmits dance information acoustically. In the first line of inquiry, Kirchner and Kathrin Sommer, a graduate student at the University of Würzburg in Germany, changed the dance sounds simply by shortening the dancer's wings slightly. Clipped wings have a smaller vibrating surface and so produce sounds that have a higher pitch and a smaller amplitude. Sommer found that bees whose wings were experimentally shortened continued to forage and dance normally. The insects could not, however, recruit nestmates. Similarly, bees from a mutant strain, called "diminutive wings,"that have congenitally small wings cannot recruit at all--even though they fiy and dance normally. Sommer studied a colony of honeybees, half of whom were normal bees and half diminutive-wing bees. She found that the normal dancing bees recruited both normal and short-winged bees equally well; in contrast, the mutant dancing bees recruited both strains poorly.

In the second series of experiments, Michelsen and Kirchner used a model bee that could perform the honeybee dances. Others had employed robotic bees in their research before, but none of the devices could make the correct sounds while it danced. Thus, a new attempt seemed worthwhile. Michelsen and his co-workers in Odense constructed the computer-controlled model bee, which danced in Würzburg for five summers.

Michelsen's team fashioned the model bee from brass and coated it with a thin layer of beeswax. The brass bee is slightly larger than a worker honeybee. Workers cut a razor blade to the appropriate size to make a set of wings. An operator can vibrate these metal wings on command by tweaking a stiff wire that connects them to an electromagnet. He or she can rotate the model in place by turning a thin rod attached to its back. A step motor at the far end of the rod steers the model's rotations automatically during its figure-eight dance. This same motor makes the model waggle from side to side. An x-y plotter, wired to a sliding metal sleeve placed around the rod, moves the model backward, forward, or left or right as needed. Finally, a thin plastic tube that ends near the model's head delivers food samples (sucrose solution) from a syringe. Yet another step motor regulatesthis mechanism. A desktop computercontrols all the motors, which in turn direct the dance.

Each experimental session typically lasted for three hours. First Michelsen and Kirchner scented the model and itssamples of sugar water with a faint fioral fragrance, then placed baits in the field that gave off trace amounts of the same odor. At each of the baits, an observer recorded the approach of searching bees. The results showed repeatedly that the mechanical dancer could indeed recruit live bees. Most of the bees invariably went to the bait in the direction indicated by the model's dance steps.

A number of additional experiments using the model bee followed. These trials were designed to determine how important various aspects of the dances are to the dance followers. For example, in some experiments, the model delivered food samples but did not dance. In this case, far fewer recruits ventured to the targeted bait. Also, the model failed completely to recruit live bees when its metal wingsdid not vibrate; these silent dances did not work, demonstrating that the sounds are truly an essential part of the honeybee's dance language.

The model bee's ability to recruit its nestmates showed that von Frisch was correct about the communicatory function of the dances. Earlier, however, despite what many researchers felt was compelling evidence supporting von Frisch's theory, some of our colleagues doubted that the bees used the distance and direction data encoded in the dances. Wenner and Patrick H. Wells of Occidental College and others have argued persistently that the coordinates given in the dance represent correlations only and are not signals. They believe recruits depend solely on odors to find feeding sites.

Von Frisch himself first attempted to test the significance of the dance movements. He found that recruits could no longer find the correct food source when he laid their hive horizontally on its side. The maneuver prevented the dancer from using gravity to orient the direction of her waggling run. As a result, the dance attenders could not interpret her actions correctly.

James L. Gould of Princeton University later punctured the odor hypothesis. He showed that a forager can dispatch her nestmates to a site she has never visited. Such a feat would be impossible if the recruits relied on odors alone to track down a feeding site. The event could occur, however, if the searching bees gave priority to the information they received from the dance. In these experiments, Gould placed a bright light in the hive, which the dance followers mistook for the sun. In doing so, they interpreted the dances erroneously. These misdirected bees most often searched in the field using the misaligned dance information and seemed to ignore other cues such as odor. Gould concluded that they evidently preferred the message given in the dance to the other signals. Finally, the experiments using the model bee confirmed von Frisch's hypothesis; the dances do indeed represent a sophisticated form of communication.

The model bee has helped us answer other questions raised by von Frisch's observations, such as determining which components of the dance language represent what kinds of instructions. For example, when the model danced so that her waggling run appeared on the outside of the figure-eight path, the recruits followed the direction indicated in the waggling run, rather than that given by the orientation of the figure eight as a whole. Thus, the waggling run alone, during which the sounds are produced, tells the recruits the direction in which they can find the food.

Although the experiments using both the model bee and the diminutive-wing bees confirmed that the dance sounds are an important part of the dance language, a crucialelement of the picture was absent: the identity of the structuresthrough which bees hear airborne sounds. Several previous attempts made to resolve this issue had shown that bees seemingly could not hear at all. Yet because we now knew that the dance sounds traveled exclusively through the air, we felt inspired to reinvestigate the question. In these renewed trials, we tested the bees using sounds very similar to those that dancing bees make.

In our first series of experiments, we trained the bees to associate a sound, lasting for five seconds, with a very mild electric shock, arriving four seconds after the sound had started. We generated the sound at the open end of a narrow glass tube. The shock alone, if delivered while the bee was feeding, would drive the bee from the feeder for a few seconds; shortly thereafter she would return and continue feeding. We then posed the following question: If a bee repeatedly experiences a tone followed by an adverse stimulus, will she eventually learn to withdraw from the feeder within the first four seconds of the sound, before the shock? If so, then she can hear--and of course learn. We found that bees can indeed be trained to respond to airborne sounds, although they learned to do so very slowly.

More recently, we have employed a different training technique. In these experiments, a bee entered a very simple Y-shaped maze. We played a sound at one end of this two-sided feeder. The side from which the sound came changed unpredictably from one trial to next. If the bee turned toward the sound, she received a reward of sugar water; if she went away from the sound, she received nothing. We observed that the bees learned quickly to turn toward the sound. Claudia Dreller, a graduate student at Würzburg, used the procedure to explore the frequency and amplitude range in which the bees could hear. Dreller's work showed that honeybees sense only low frequencies, those below 500 hertz. They hear these tones with sufficient sensitivity to pick up the sounds of a dancing nestmate, which range from 250 to 300 hertz. They also show some ability to discriminate between frequencies in this range; they can discriminate between low- (20 hertz), medium- (100 hertz) and high-pitched (320 hertz) sounds. We do not yet know for what purpose the bees might use this latter ability.

The same training technique enabled us to find out through what sensory structures the bees detect the sounds. We altered some of our trained bees by removing one antenna, fixing a certain antennal joint or removing sensory hairs from their head. We found that bees use a structure called the Johnston's organ, a chordotonal organ made up of nerve cells in the second joint of a bee's antennae, to pick up airborne sounds. Some fiies and mosquitoes rely on the same structure to perceive sounds.

If we now put the pieces together, we can see how the dance language works. The dancer emits sound signals that help the dance followers determine where the dancer is and how she is moving, which in turn offers them critical information regarding the direction and distance to the feeding site. The dance attenders receive these signals through the Johnston's organs located in their antennae, which are always held near the dancer. Because these organs are bilateral--one on the left and one on the right--the dance followers can use them to judge their position with respect to the dancer and therefore understand the direction to the food. At the same time, the followers emit sounds that vibrate the comb. The forager stops her dance when she receives these signals and delivers samples of the food she has collected. These appetizers give the dance followers additional hints about the taste, smell and quality of the food source. The bees attend the dancing for a while and then fiy out to find the food source on their own. If they are lucky, they will find the food. If they fail, they will return to the nest and try again.

This dance language is clearly a very complicated, highly developed system of recruitment. To understand how such a system evolved, we and other scientists have examined the recruiting techniques of related species. The genus Apis, to which all honeybees belong, has no close living relatives. Bumblebees and stingless bees are their nearest kin, and, unfortunately, it seems that bumblebees do not recruit at all. Many species of stingless bees, on the other hand, do recruit. But, as far as we know, none has developed a symbolic language similar to the dance language of Apis. All four species of honeybees studied so far (three of which live in Asia) speak some variant of the dance language. As Martin Lindauer discovered in the 1950s, when he was at the University of Munich, all species use similar distance and direction codes, even though there are some differences.

Although we have looked for dance sounds in four species of honeybees, we have found that only three produce them. These three species hold something else in common: they all must occasionally dance in the dark. Two of them, the familiar western bee, A. mellifera, and the Asian bee, A. cerana, nest in lightless, enclosed areas such as hollow trees or other similar cavities. The third sound producer, the giant bee, A. dorsata, nests in the open on single sheets of comb, hanging under rock outcrops or thick branches of trees. Fred C. Dyer of Michigan State University first showed that A. dorsata sometimes dances at night, and Kirchner discovered only recently that this bee produces sounds. A. dorsata's signals were very difficult to detect because these sounds are particularly low in pitch.

The single species that dances silently, the dwarf bee, A. fiorea, dances in the open like A. dorsata, but only during the day. Dancers of this species make gestures that may, in daylight, serve as visual signals to attract dance followers in the same way that sounds assist those bees who dance in the dark. Because some indications suggest that A. fiorea's habits are the most primitive, we assume that the complicated acoustical communication system of the other three species most likely evolved from a visual display when these bees developed habitations that cut them off from light.

Now that we can finally listen to the bees' language and even speak it a little, we face a host of new questions. For example, we have yet to learn for what purpose the bees possess the ability to distinguish between different pitched sounds. In addition, perhaps they use similar airborne sounds in ways that we do not even suspect at this time. In the hopes of revealing their communication system in full, we will continue to eavesdrop on their conversations.

Further reading

  • The Dance Language and Orientation of Bees. K. von Frisch. Harvard University Press, 1967.
  • Communications Among social Bees. Martin Lindauer. Harvard University Press, 1971.
  • The Honey bee James L. Gould and Carol G. Gould. Scientific American Library, 1988.
  • Acoustical Communication in Honeybees. W. H. Kirchner in Apidologie, Vol. 24, No. 3, pages 297--307; July 1993.

The Authors
Wolfgang H. Kirchner and William F. Towne began their collaboration in 1987 and jointly discovered the auditory sense of the honeybee. Kirchner has studied the mechanisms and evolution of dance communication in honeybees for the past 10 years. He received a Ph. D. in 1987 from the University of Würzburg in Germany, where he currently holds a faculty position. Towne has studied the dance communication of honeybees since 1980. He received a Ph. D. in biology from Princeton University in 1985 and is an associate professor of biology at Kutztown University in Pennsylvania.