Abstract. Direction communication experiments, when controlled against unilateral visitation of trained foragers at the experimental site, exhibited no evidence supporting the precision of communication suggested in the dance "language" hypothesis. Results compare well with those which might be expected on the basis of the geometric arrangement of the feeding dishes.

A previous investigation demonstrated the importance of conditioned responses in the exploitation of food sources by honey bees (1). Experienced foragers were experimentally recruited to feeding sites, by means of odor stimuli, without their having gained information from the dance maneuver. That study further indicated that recruitment by the conditioned responses of experienced bees can account for a colony's efficient exploitation of food sources. Consideration of these results eventually led to a questioning of the validity of the dance "language" hypothesis.

Original evidence supporting the theory that naive bees utilize abstract dance information to determine the direction of food sources is largely derived from the "fan experiments" of von Frisch and co-workers (2). Those experiments were designed to ascertain the precision with which recruit bees use the direction information contained in the dance maneuver to locate a food source. In such a fan experiment, empty scented dishes were placed in slightly different directions 200 m from the hive. A group of marked bees regularly foraged scented sucrose solution only at an experimental dish placed in the central direction of the fan arrangement of control dishes. but at a greater distance (250 m) from the hive. During the course of the experiment, investigators tallied the number of unmarked bees approaching each dish.

In those earlier fan experiments the majority of recruited bees arrived at the experimental station or at a station, or stations, located in this same direction. The dance maneuver executed in the hive by the marked bees foraging at the experimental station contains precise direction information [within +/- 8 degrees for a distance of 200 m (3)]. Recruited bees presumably interpreted and used this dance information as they traveled to the prospective food dish.

A close examination of the design of such experiments reveals a lack of essential controls. For example, trained bees continually visited the experimental dish, while no bees routinely visited the control dishes. Therefore, the results of those experiments do not eliminate the possibility that recruit bees are attracted to (or near) the station having regular visitation by trained foragers. Kalmus (4) demonstrated the importance of feeding bees in attracting recruits to a site. Another possibility not eliminated by the design of the earlier experiments was that recruits were attracted to the geometric center of all feeding sites (5).

Greater uniformity among the various stations (including controls against the effects of unilateral bee visitation at the experimental site) should not influence the pattern of recruitment if the recruited bees depend on the directional information of the dance to locate food sources as maintained by the dance "language" hypothesis [see von Frisch (6, 7) for detailed description of this hypothesis]. This present study contains fan experiments which incorporate such essential controls.

Two different bee colonies were used in these experiments. The experimental hive [different from the experimental hive used in a companion study (5)] contained at least 20,000 dark bees (Apis mellifera, Italian strain obtained from C. G. Wenner apiaries in northern California). The control hive contained about 15,000 light-colored bees. This strain, with a cordovan gene (cd) for light body-color, has been rendered isogenic with the Italian strain by repeated back-crossing and was furnished by H. H. Laidlaw of the University of California, Davis.

These adjacent hives were painted different colors and placed with entrances facing different directions to permit homecoming bees to orient readily to their own hive. Two other checks insured that drifting of foragers did not influence experimental results. Spot checks of bees entering and leaving each hive revealed that only rarely did an unmarked individual enter the wrong hive, in which case it was killed (by investigators). In addition, one or the other of the hives was closed periodically (at least once per day) while checks were made of all feeding stations. Neither dark nor light foragers arrived at any station when their respective hives were closed.

Fig. 1. Map of the experimental area. The experimental (dark-colored bees) and control hives (light-colored bees) were adjacent to one another at the edge of a football practice field (outlined by the broken line). The experimental site in the first series of experiments was A; in the second series, B. The three sites, north (N), middle (M), and south (S) served as controls and had dishes of sugar solution to which recruits could come. Points a and b indicate the geometric center of each set of four feeding sites for each experimental series.

Two series of similarly designed experiments were conducted. In each series the three control feeding stations were positioned in an arc (70 m apart at 200 m from the hives), and the experimental station was 270 m from the hives. In the first series the experimental station was positioned behind the middle station (at location A of Fig. 1). In the second series the experimental station was located behind the north station (at location B of Fig. 1). Both control and experimental stations were located on a large level field of mowed green grass (football practice fields). The hives sat at one edge of this field.

Fifteen individually marked dark bees from the experimental hive were trained (8) to visit the experimental station; and light bees (also marked) from the control hive were trained to visit each of the feeding stations - 20 to each of the control stations and five to the experimental. Subsequent references to the trained population at control and experimental feeding dishes will refer to the above numerical distributions of trained foragers, unless one of the hives was purposely closed. Trained bees were fed peppermint-scented 1.5M sucrose solution (five drops of peppermint oil per 500 ml of solution) at their respective stations and were replaced with new bees as needed throughout the experimental program. Each group of marked bees foraged only at the station to which trained. All unmarked visitors (or trained bees visiting incorrect stations) were killed.

Each of the two experimental series consisted of three types of experiments. In the first, or control experiment (Table 1, experiment 1), the control hive was closed, and only the 15 dark foragers regularly visited the experimental station. At time zero, all feeding dishes were provided with scented sucrose solution. For a 30-minute period all new (unmarked) recruits arriving at all stations were caught and placed in a container of alcohol by the person attending each dish and were counted later.

Exper- iment	Recruits arriving at each station (%)				Total recruits (No.)	Wind		Date	Time
	Experi- mental	North	Middle	South		Direc - tion (de- grees)	Veloc - ity (knots)
Experimental series 1
1A	47	10	31*	11	99	160	4	11 Aug.	1030
2A	19	12	48*	21	150	240	11	11 Aug.	1400
	(10)	(6)	(53)	(31)	(245)
3A	7	8	70*	15	61	210	9	18 Aug.	1210
Experimental series 2
1B	56	22*	14	8	36	140	8	16 Aug.	0950
2B	19	21*	52	8	125	140	8	16 Aug.	1040
	(10)	(18)	(57)	(16)	(351)
3B	23	15*	57	5	74	240	6	18 Aug.	1020

Table 1. Distribution of recruited bees in each of six experiments (three different types of experiments with the experimental site located either at A or at B in Fig. 1). In the first experiment of each series, dark bees visited only the experimental site. In the second experiment of each series, light bees visited all four sites while dark bees still visited only the experimental site. In the last experiment, an equal number of dark bees visited each of the four sites after the control hive had been removed. The numbers in parentheses indicate the distribution of bees recruited from the control hive. Asterisks indicate the control station behind which the experimental station was located.

In both control experiments, 78 percent of the captured recruits arrived at stations located in the experimental direction, regardless of whether the experimental station stood behind the middle or the north station (Table 1). These control experiments were repeated with similar results. Moreover, these results are entirely consistent with those expected on the basis of the dance "language" hypothesis.

Following a control experiment, the control hive was opened, and its trained foragers frequented each of the four feeding stations while the trained dark bees continued to forage at the experimental site. As always, unmarked foragers were killed during this time. After this was accomplished, the second experiment began, and new recruits arriving at each station were collected during a 40-minute period. In the course of this experiment, the dish at the experimental station was replaced every 10 minutes with a clean one. The dirty dish from this station, in turn, alternately replaced one of the dishes at those two control stations not in a direct line with the experimental site. Switching dishes in this manner reduced the possible accumulation of specific hive odor (from the dark bees) at the experimental station and distributed it to other sites (9, 10, 11).

In these experiments (Table 1, experiments 2A and 2B), when feeding stations were more uniform with respect to trained foraging populations at each dish, between 50 and 60 percent of the total number of captured dark recruits arrived at stations located in the central direction, irrespective of where the experimental station was (at location A or B in Fig. 1). Furthermore, the identity between the distribution of light and dark recruits in each of the experiments is a result not consistent with the dance communication theory. The experiments were repeated twice and yielded results similar to those shown in Table 1.

Experiment 3 of each series involved removing the control hive and training 60 dark bees from the experimental hive to the four feeding stations, 15 to each station. The bees in each of these groups were not allowed to visit stations other than the one to which they had been trained. During a 40-minute period, while these bees continually foraged at the stations, all new recruits arriving at each station were collected as before (Table 1, experiments 3A and 3B).

The resulting nonuniformity of distributions of recruits suggests a bias favoring the central station in each experimental arrangement. This same bias is also apparent in the distributions obtained in experiments 2A and 2B. Three repetitions of experiment 3 again yielded similar results.

An additional experiment conjunctive to the previously described control experiments (1A and 1B) yielded information on the attractiveness of all four stations relative to that of the experimental station alone. During the 30 minutes preceding experiment 1A (Table 1), with the control hive closed and only the experimental station on the field, 25 new recruits arrived at that site. In the following 30 minutes (experiment 1A) a total of 99 unmarked bees landed at the four stations, 47 of which arrived at the experimental site.

The next morning, during similar weather conditions, a control experiment (of the 1B type - not in Table 1) was performed before the companion study. In the control a total of 85 recruits arrived at the four stations; 28 of these arrived at the experimental site. In the subsequent 30-minute companion study, when only the experimental station was on the field, 46 new recruits arrived at that site. Apparently, increasing the number of stations increased the number of recruits caught, which suggests that many more recruited bees search likely areas than ultimately locate a site regularly visited by hive mates (only 15 dark foragers visited the experimental site in each of the above experiments).

The control experiments (1A and 1B) essentially repeat the earlier fan experiments of von Frisch (2) with similar results; namely, the majority of recruits (78 percent) arrived at dishes in the direction presumably indicated by dancing bees. During experiment 2, however, when equal numbers of trained bees were permitted to visit all feeding stations, the pattern of recruit distribution differed from that obtained in the control experiments.

Comparison of one control to an experimental period (1B and 2B in Table 1) illustrates such a shift in recruitment distribution. Even though these periods were separated by only 20 minutes during similar environmental conditions and although they involved the same group of trained dark foragers at the experimental station, the furnishing of an equal number of regular visitors at each station yielded a markedly different distribution of dark recruits. Now, neither light nor dark recruits appeared to have responded to that precise direction information they could have received from dance maneuvers.

Identity between recruitment distributions of light and dark bees in the second experiment of each series and the consistent pattern in the percentage of recruitment evident between stations conceivably arise from the use of similar orientation cues. Apparently, providing equal visitation by trained bees at both control and experimental stations, although rendering them more identical in attractiveness, leaves them unequally distributed in space. Under the circumstances of these experiments, the middle station always lay closer to the geometric center of all sites (center of moments) than did any other station. In Fig. 1, points a and b locate the center of moments in each series of experiments - altering the location of the experimental site from A to B in that figure does not appreciably shift the center of moments. It remains near the middle site.

Just as the center of moments remains near the middle control station, so also were most recruits collected at the middle site when all stations became more similar in attractiveness. In either experiment 2A or 2B, the recruit distributions more closely correspond with the spatial distribution of dishes than with the results from the first (control) experiment of each series. [Statistical tests have not been employed in this or in other comparisons for various reasons; see note 9 in a related study (5).]

The results from the third experiment in each series (Table 1, 3A and 3B), where an equal number of marked dark bees foraged at all four sites (and in which the stations would be even more equal in attractiveness), correlate even better with the linear distance of each site from the geometric center of all sites. The close correspondence between the spatial arrangement of all stations and the distribution of recruited bees persists with little modification by wind conditions, possible interference between bees from the two different hives (12), and, perhaps even more important, slight differences in locality odors.

The results of the original fan experiments were interpreted by von Frisch as a demonstration of the precision with which honey bees use direction information. He concluded that the majority of searching bees, after leaving the hive, move within an angle deviating not more than 15 degrees to the left or right of the experimental direction (13). While repetitions of such experi-ments yield data which seem completely consistent with the original results (experiments 1A and 1B), results obtained after the incorporation of essential controls into the experiments indicate that, if recruits use the direction information of the dance, they fail to do so with the precision suggested by the original experiments. Thus, although the waggle-dance maneuver contains relatively precise direction information, it would appear that local cues (including locality odor, food odor, hive odor, and bee odor at the site) dictate the location or locations at which recruits settle to feed.

DENNIS L. JOHNSON Department of Biological Sciences, University of California, Santa Barbara

References and Notes

1. D. L. Johnson and A. M. Wenner, Anim. Behav. 14, 261 (1966).

2. K. von Frisch, Naturwissenschaften 35, 12, 38 (1948).

3. E. M. Schweiger, Z. Vergleich. Physiol. 41, 272 (1958).

4. H. Kalmus, Brit. J. Anim. Behav. 2, 63 (1954).

5. A. M. Wenner, Science 155, 847 (1967).

6. K. von Frisch, Bees, Their Vision, Chemical Senses, and Language (Cornell Univ. Press, New York, 1950).

7. ________, Sci. Amer. 207, 78 (1962).

8. A. M. Wenner, Bee World 42, 8 (1961).

9. K. von Frisch and G. A. Rosch, Z. Vergleich. Physiol. 4, 1 (1926).

10. C. R. Ribbands, H. Kalmus, H. L. Nixon, Bee World 33, 165 (1952).

11. C. R. Ribbands, Proc. Roy. Soc. London Ser. B 143, 367 (1955).

12. H. Kalmus, Nature 148, 228 (1941).

13. K. von Frisch, Bull. Anim. Behav. 9, 10 (1951).

14. Supported by contract NR 301-800, Office of Naval Research and by a PHS graduate fellowship (3-F1-GM-32, 408, 01S1). This work serves in partial fulfillment of the requirements for the degree of Doctor of Philosophy. I thank N. Barnes, N. Broadston, J. Hand, C. Johnson, and A. Wenner for technical assistance and Drs. D. Davenport, J. Enright, E. Orias, and P. Wells for critically reviewing the manuscript. I also thank H. H. Laidlaw for furnishing the light-colored bees used in these experiments.

2 December 1966