Odors, Wind and Colony Foraging (2 of 3): The Role of Wind Direction

[1998 Wenner, A.M. Odors, wind and colony foraging — Part II of three parts: The role of wind direction. Am. Bee J. 138:807-810 (Nov. issue).]

967 Garcia Road
Santa Barbara, California 93103
(805) 963-8508

“If wind flow is of a sufficient magnitude for a flying organism to detect its direction, then one obvious tactic for locating a chemical source in such winds is to proceed upwind.” Ring Carde – 1984

In Part 1 of this series I stressed the fact that recruited bees need an odor cue as they search for food, just as von Frisch emphasized in 1939 (e.g., Wenner, 1993). Simply put: Without an odor cue recruited bees cannot succeed. Conservatively, then, we can assume that all recruitment experiments have included such an odor cue, either deliberate or unintentional. That means that bee “language” proponents can never he certain that the searching bees in their experiments had depended on dance maneuver information in their search rather than (or as well as) an odor marker.

It is not enough to claim, “We’ve always known odor was important!” Consider, for example: 1) the small percentage of searching bees that manage to find the target food source; 2) the great amount of time those few successful bees spend searching; and 3) the fact that new arrivals always come into an odor source from downwind (e.g., Wells and Wenner, 1974). Those and other facts mesh well with our alternative odor-search model of honey bee recruitment (Wenner, et al., 1991 – as reviewed by Southwick in the October 1992 issue of The American Bee Journal).

This second part of the series documents the importance of wind direction for bees as they search for a target food source. And we all know that when bees fly (only during the day) wind is almost always present. Fortunately for experimenters, odor molecules (being physical particles) can only travel downwind. The result is an odor plume continually formed and present downwind from each odor source (e.g., Murlis, et al., 1992).

The secret to unraveling the mystery of honey bee recruitment behavior (and incidentally further resolving the dance language controversy), then, will hinge upon researchers eventually considering the importance of wind direction for recruitment success, a topic basically overlooked this past half century by almost everyone. A marked exception: The results of largely unacknowledged studies done by Larry Friesen (1973).

Downwind vs Upwind Experiments

As part of his studies, Friesen conducted an ingenious set of experiments that revealed how well searching bees could locate an upwind station – compared to how well they could find a downwind station. Although he published results from many experiments, I give here only a summary of results from a few (see Friesen, 1973 or Ch. 8 in Wenner and Wells, 1990 for details).

Friesen set up two hives 655 yds (600m) apart, each a strong colony of two stories with its own color of bees (cordovan or very light colored bees, as against normal dark Italian), with one colony upwind from the other. He then trained 10 individually marked foragers from each hive to feed at a single station between the two of them. He tallied all round trip times for the regular foragers from each hive and the arrivals of each color of bees at the single station. He captured and killed all recruits to prevent confusion.

First, consider the average recruitment success in each 15 minute segment during a three-hour period for a total of four days, when the single feeding station was 165 yds (150m) upwind from one colony and thus 490 yds (450m) downwind from the other colony. A total of 333 searching bees found the 150m upwind station, but only 76 found the station 450m downwind from the second parent colony (Fig. 1). During that same three hours, the 10 foragers from each hive collectively would have made several hundred round trips (e.g., Wenner, et al., 1969).

Fig. 1. Total number of recruit arrivals during a four-day sequence (3 hours per day) at a single feeding station located partway between two hives 600m apart (adapted from Figs. 12 and 13 in Friesen, 1973), with ten foragers from each hive making regular round trips. Recruitment for a station 150m upwind from its parent colony (dark bars) fit the normal pattern (as in Fig. 8.2 in Wenner and Wells, 1990). However, recruitment to a station 450m downwind from a second colony (light bars) was negligible.
Next, consider an opposite arrangement – the single feeder located 150m downwind from one colony and 450m upwind from the other colony. Friesen again caught, killed, and tallied the number of recruits that had arrived from each of the two hives during each 15 min. period (Fig. 2). The upwind station, with 464 recruits, had a by-now-familiar pattern, as shown in Fig. 1 and in Figure 8.2 of Wenner and Wells (1990).

By contrast, the number of recruits arriving at the single station located 150m downwind from the parent colony (Fig. 2) was quite unexpected at the time, with a grand total of 957 recruits – far more than to any of the five upwind locations on various days during a three-hour period. Furthermore, recruits began to arrive very soon at that rather close downwind station – compared to the time of first arrivals at the other stations. That result indicated to us that newly recruited bees must begin their search fairly close downwind from their colony. Why would they do so, if they “flew directly out in the proper direction” in their search for the food – as we would expect according to the dance language hypothesis (e.g., Wenner and Wells, 1990, p. 46)?

Fig. 2. Total number of recruit arrivals as depicted in Figure 1, but this time with the feeding station located closer to the upwind hive (adapted from Figs. 12 and 13 in Friesen, 1973). Recruitment for the station 450m upwind (dark bars) from its parent colony fit the normal pattern, but recruitment to a station only 150m downwind from the parent colony (light bars) occurred earlier and with greater intensity than for other situations in the multiple comparison (see Fig. 4).
We can also ask: How does the foregoing set of results relate to an important notion in science?

An answer: A good scientific hypothesis permits precise predictions. We now know (Wenner, 1963) that regular foragers on a beeline out from the hive cover the distances used in Friesen’s experiments quite rapidly (about 60 seconds to go 450m and 20 seconds to go 150m). If recruited bees had used dance maneuver information as claimed, all four recruitment patterns shown in Figs. 1 and 2 should have been basically identical. Obviously, they were not.

Yes, the recruitment patterns for the two featured upwind stations (dark bars in Figs. 1 and 2) – located either at 150m or 450m – differ hardly at all. But recall: 1) the long delay in arrival of new bees after foragers begin regular trips, as evident in those graphs; 2) the small percentage of success by searching bees; and 3) the great amount of time spent searching (e.g., Wells and Wenner, 1974).

On the other hand, the pattern of arrivals for bees that had searched for the 150m and 450m downwind stations (light bars in Figs. 1 and 2) were wildly different from one another – hardly what dance language proponents would have predicted.

Now examine the recruitment success for all 4371 bees collected during the 20-day period: 1) at the upwind and 2) at the downwind stations. All five upwind stations (Fig. 3) received a nearly similar number of recruits each three-hour period, regardless of distance from the hive – with only a generally slight decline with increasing distance. Obviously, different travel times of regular foragers at those five distances mattered little for the searching bees that managed to locate a source upwind from their colony.

Fig. 3. The total number of recruit arrivals at upwind feeding stations for the 20-day experiment that formed the basis for Figures 1 and 2 (derived from Figs. 12 and 13 in Friesen, 1973). Searching bees found relatively distant upwind stations almost as readily as those closer to the parent hive.
By contrast, the recruitment pattern for the five downwind stations (Fig. 4) does not agree with what one would expect under the dance language hypothesis (e.g., Wenner and Wells, 1990, p. 64). With increasing distance, recruit success fell off dramatically. It would appear that a total of even 10 regular foragers would not provide a sufficient aerial pathway for searching bees to find a station 500m (less than a third of a mile) or more downwind from their colony.

Fig. 4. The total number of recruit arrivals at downwind feeding stations for the same 20-day period as in Fig. 3 (derived from Figs. 12 and 13 in Friesen, 1973). Although searching bees rather readily found stations located a couple hundred of meters downwind, the falloff in success with
distance was precipitous.

However, if dozens or hundreds of foragers (recruited earlier under different wind conditions) had exploited a by-now-downwind food source, their collective flight paths could well provide a necessary aerial pathway (Wells and Wenner, 1974; Wenner and Wells, 1990), one that could repeatedly give essential odor cues to searching bees. Thus, any results obtained by researchers who use more than a dozen bees in these types of experiments would perhaps be suspect.

We gain important lessons from the above two sets of Friesen results. 1) In Part I (last month): Searching bees found a crosswind station only with great difficulty and only by exploiting odor. 2) In this Part II: A station any appreciable distance downwind from the colony, one visited by only a few foragers, would likely receive little or no recruitment, since the target odor molecules then travel away from the hive. The collective Friesen results (in Figs. 1-4, above and in the two figures shown in Part I of this series) thereby provided an important basis for our improved odor-search model of recruitment (as in Wenner, et al., 1991; summarized in this journal by Southwick, 1992).

Compare the above analysis to quite another set of results. Visscher and Seeley (1982) studied dance maneuver patterns during an 8-day period between 12 June and 19 June 1980. By observing dances of regular foragers in an observation hive, they estimated the distance and direction travelled by those foragers in a forest near Cornell University.

For the first four days, most foraging occurred in a SSW direction. However, on the 5th day of their experimental series they recorded minimal foraging. For that particular date, Visscher and Seeley wrote, “The weather this day was cool, with intermittent rain, so the bees foraged relatively little and only fairly close to the colony.”

In about 30 BC the poet Virgil observed the same phenomenon. He wrote:

“When rain hangs in the sky or the wind sharpens from the east, the bees are cautious, keep close to home. They draw water in the shelter of their city walls…”

Visscher and Seeley further reported that, after that 5th day of unsettled weather, the foraging pattern of their colony shifted generally to the northeast. As we all know, weather fronts generally pass from west to east through an area in the Northern Hemisphere; as a front approaches, winds first shift to the east. Also, during rain, nectar becomes washed out of blossoms. The shift in foraging pattern observed during those days by Visscher and Seeley might well have been expected, but not necessarily due to a change in dance maneuver information.

Scientists most often interpret the results of their experiments in terms of prevailing theory (see Wenner, 1989). Sometimes, though, results which do not fit theory can provide clues about how Nature really functions.

Part III of this series will illustrate what we learned about colony foraging patterns as we searched for and found well more than a hundred feral (“wild”) bee colonies on Santa Cruz Island this past decade. In essence, we found that we were most effective in our searches when we exploited our understanding of the importance of wind direction in colony foraging patterns.


Carde, R.T. 1984. Chemo-orientation in flying insects. Pp. 111-124 in Bell. W.J. and R. T. Carde (eds.). Chemical Ecology of Insects. Chapman and Hall. New York.

Friesen, L. J. 1973. The search dynamics of recruited honey bees, Apis mellifera ligustica Spinola. Biological Bulletin. 144:107-131.

Murlis, J., J.S. Elkinton, and R.T. Carde. 1992. Odor plumes and how insects use them. Annual Review of Entomology. 37:505-532.

Southwick, E.E. 1992. Bee Research Digest: Foraging, recruitment and search behavior of honey bees. American Bee Journal. 132(10):641, 642, 644.

Visseher, P.K. and T. D. Seeley. 1982. Foraging strategy of honeybee colonies in a temperate deciduous forest. Ecology. 63:1790-1801.

Wells. P.H. and A.M. Wenner. 1749. How recruited bees find food. Gleanings in Bee Culture. 102:110, Ill, 127.

Wenner, A.M. 1963. The flight speed of honey bees: A quantitative approach. Journal of Apicultural Research. 2:25-32.

Wenner, A.M. 1989. Concept centered versus organism centered biology. American Zoologist. 29:1177-1197 (a somewhat self-explanatory title).

Wenner, A.M. 1993 [with K. von Frisch]. The language of bees. Bee World. 74:90-98.

Wenner, A.M., D. Meade, and L. J. Friesen. 1991. Recruitment, search behavior, and flight ranges of honey bees. American Bee Journal. 31(6):768-782.

Wenner, A.M. and P.H. Wells. 1990. Anatomy of a Controversy: The Question of a “Language” Among Bees. Columbia University Press.

Wenner, AM.. P.H. Wells, and D.L. Johnson. 1969. Honey bee recruitment to food sources: Olfaction or language? Science. 164:84-86.