Three Bees I Have Known

Automated behavioral monitoring of honey bees. Left Panel: Bee with two radiofrequency identifi cation (RFID) tags. RFID readers placed at the hive entrance allow for automated monitoring of fl ight behavior (Tenczar and Lutz et al. 2014). Image from P. Tenczar. Right Panel: Bees with barcodes. Machine vision and machine learning allow for automated monitoring of in-hive behavior (Gernat et al. 2018). Image from B.M Jones and T. Comi.

By: Gene Robinson

Do you remember Barry B. Benson, played by Jerry Seinfeld in the 2007 film Bee Movie? Benson was a honey bee who did not want to do what was expected of him. True, Benson was male rather than female, wore sneakers, and fell in love with a human, but you may be surprised to know that the premise of a maverick bee is not at all far-fetched. Serendipitous observations of three real-life adult worker honey bees my colleagues and I have made over the years suggest that your beehives might just contain a few iconoclasts. Let me tell you about three bees I have known.

Yellow 57 (Y57) was named for the colored numbered tag I affixed to her thorax when she emerged as a one-day-old adult. She was part of a cohort of 150 bees, all sisters, that I studied in an experiment performed in 1982 for my doctoral dissertation at the Dyce Laboratory for Honey Bee Studies, under the direction of the late Roger A. Morse at Cornell University. The main purpose of the experiment was to study hormonal regulation of age-related division of labor. The experiment required that I make careful observations at the entrance of each hive daily for one hour to observe the comings and goings of my tagged bees and determine the age at which they shifted from working in the hive to foraging outside.

Y57’s foraging activity was unique; her flights were more frequent and shorter than the flights undertaken by the other individuals in her cohort (Robinson et al. 1984). Y57 took an average of about seven flights per hour compared with one or two, and hers lasted on average only about three minutes compared with about 20 minutes. Moreover, her flights were remarkably constant in duration whereas the flights of the other individuals varied greatly from just a few minutes to almost an hour. My curiosity was piqued.

I discussed Y57’s unique behavior with my fellow graduate students Ben Underwood and Carol Henderson, and we reasoned that she must be foraging for something quite close to her hive to have such short flights. The consistency of the flights also suggested to us that Y57 might be foraging for water; once located, a water source should provide a more constant reward than any given patch of flowers. Ben decided to visit a spot in Salmon Creek, which was approximately 0.5 km from the apiary in which Y57’s colony was located in Ithaca, New York, where he previously observed honey bees foraging for water. Ben went to the spot, looked down, and believe it or not saw Y57 collecting water. Given that the foraging range of a honey bee colony can be over 100 kilometers2 (Visscher and Seeley 1982), spotting Y57 was the bee biologist’s equivalent of winning the lottery!

We immediately set up a real-time monitoring system–pre-cell phones – aided by walkie-talkies. I continued observations at the hive entrance while Ben set up a chaise lounge in the shallow creek, next to the rock that Y57 was spotted on. I announced Y57’s hive departure to Ben, and, sure enough, about one minute later she appeared at the appointed location. Y57 took about a minute to load up with water and then flew off. Alerted by Ben, I was on the lookout for her return to the hive about one minute later, and she never disappointed. The details of these fully mapped trips matched precisely with the observations made during the previous 14 consecutive days, which started on the first day Y57 was seen foraging at 17 days of age and ended when she disappeared and presumably died when she was 31 days old. We concluded that Y57 devoted her entire foraging career to collecting water, apparently from only one single choice location.

To put this behavior into context, bees do specialize by task as a consequence of age-related division of labor; for example, a nurse bee will tend to the brood for a week or more before moving onto other tasks and a forager spends its days collecting nectar and pollen for a similar period of time. But no one had ever observed a bee with the single-mindedness of Y57. A nurse bees cares for many larvae and performs other tasks during this phase of its life and a forager typically collects nectar and pollen from many different patches of flowers. Even scout bees, a subset of foragers that specialize on searching for new food sources or nest sites, visit many different locations. Is Y57 an aberration? My subsequent encounters with two other highly specialized bees suggest that this is not the case.

We discovered Red 93 (R93), an extreme groomer, also during a study on an unrelated topic, the relationship between circadian rhythms of activity and division of labor. The study was conducted in 1994 and by that time, I was a faculty member in the Department of Entomology and the Director of the Bee Research Facility here at the University of Illinois at Urbana-Champaign. The study was led by Darrell Moore from East Tennessee State University, who spent several summers with us as a visiting professor. The study again involved extensive observations of a cohort of individually tagged bees, but while they were inside a glass-walled observation hive rather than at the hive entrance.

Social grooming is widespread in honey bee colonies; it has long been known to have hygienic functions and features prominently in the exciting project initiated by Greg Hunt at Purdue University to breed bees that can resist the ravages of Varroa by removing the mites from the bodies of their hivemates (Hunt et al. 2016). A lot of grooming goes on in a hive, but the longstanding assumption was that all bees spend just a little bit of their time doing it (Moore et al. 1995).

Not R93 – she groomed other bees a whopping 84% of the time, 69 out of the 82 times she was observed over a 15-day period. As in previous studies, most other individuals, R93’s sisters, groomed other bees infrequently or not at all. And rather than waiting for another bee to perform a “cleaning dance” (von Frisch 1967) to solicit grooming behavior as most bees generally do, R93 simply approached hivemates unbidden and started to groom them. R93 persisted in her extreme grooming and she never grew up. That is, she never started to forage but rather remained in the hive her entire life, grooming.

The third and final bee I want to tell you about is Yellow 54 (Y54), an extreme undertaker (corpse-removing) specialist. The study that led to Y54’s discovery, led by Stephen Trumbo in 1997 as a postdoctoral research associate in my lab and now a professor at the University of Connecticut, was interested in the relationship between task specialization and learning (Trumbo and Robinson 1997). Learning has been clearly demonstrated for foraging (Dukas and Visscher 1994), but not for in-hive tasks.

We thought corpse removal might involve learning because – unlike other in-hive tasks performed by middle-age bees (about 10-20 days of age) such as food storage and comb building – it is a niche job performed only by a small subset of a colony’s adult population, and requires impressive strength and agility. An undertaker grabs a corpse, which of course weighs about as much as she herself does, drags it along the bottom board to the hive entrance, and then flies out of the hive carrying her payload before dropping it a few meters away. Like grooming, corpse removal is important for colony hygiene; some individuals do this job over many days but most bees never remove a corpse in their entire lives (Visscher 1988).

Y54 removed 114 corpses over a 13-day period, which accounted for 33.8% of all the dead bees that we experimentally introduced into the hive for this study. By contrast, her sisters removed one to eight corpses during their life. Y54 is the most active undertaker recorded to date in any study.

Y54 also removed corpses significantly faster than the other bees. But she did not improve her performance over time, which means that learning was not involved. Was Y54 just more innately talented at this job than her hivemates, or was there something about her previous experience that set her up for a record-breaking career? We wonder in the same way about the nature-nurture origins of exceptional performance in humans.

As we noted in Robinson et al. (1984), there were a few earlier published accounts of unusually persistent behavior by individual bees from the laboratories of Martin Lindauer, Shoichi Sakagami, and Mark Winston, but nothing as extreme as Yellow 57, Red 93, or Yellow 54. What do we make of them? Why are there so few records of highly specialized bees? Is it because they are genuinely rare, or is it because studies have not been designed to detect them? If they really are so rare they may be little more than curiosities, their “obsessive-compulsive” behavior the product of neurodevelopmental dysfunction that occurs too infrequently to capitalize upon for brain research. But if extreme specialists are actually more common in the beehive, it is important to study how they contribute to colony life, and, yes, such studies might be very useful for neurobiological and genomic analyses of the brain.

It should be possible for bee biologists to answer these questions in the near future because research on bee behavior, like animal behavior in general, is at the dawn of an exciting new era marked by widespread use of automated behavioral monitoring systems. Leaving behind painstaking and hit-or-miss observations of individual bees wearing numbered plastic tags, my lab and other groups are leveraging recent advances in engineering, computing, artificial intelligence, and machine vision to develop new methods of automated behavioral monitoring for honey bees.

Here’s a taste of what’s to come. First, Paul Tenczar, a citizen scientist in my lab trained as an engineer, developed a radiofrequency identification (RFID)-based system to monitor bee flight activity (above, left panel). He and graduate student Claudia Lutz then used this system to discover that approximately 20% of the foraging workforce accounted for 50% of their colony’s total foraging effort (Tenczar and Lutz et al. 2014). Second, computer science graduate student Tim Gernat in my lab (Gernat et al. 2018) developed a barcode-based system to monitor in-hive behavior (above, right panel). The laboratory of physicist and Illinois colleague Nigel Goldenfeld then analyzed data generated with this method for trophallaxis, a behavior that involves an exchange of food and signaling molecules among colony members, and discovered a thought-provoking pattern. Bees have longer trophallaxis exchanges with some hivemates than with others, and the pattern of these individual differences resembled the pattern for human “face to face” social encounters (Choi et al. 2020). We don’t know enough about the function of trophallaxis yet to understand the significance of this similarity, but this result hints at trophallaxis being used as a means for communication among bees.

Automated behavioral monitoring systems provide bee biologists with powerful new tools. If there are iconoclasts like Yellow 57, Red 93, and Yellow 54 yet to be known, these new methods will surely allow for their discovery, enriching our knowledge of bee behavior.

The research discussed here was supported by grants that I received from the National Science Foundation, National Institutes of Health, and the Christopher Foundation. I also thank M.R. Berenbaum, T. Gernat, C.M. Grozinger, and D.J. Robinson for comments that improved this manuscript, and C.M. Grozinger, whose interest in our work on this topic prompted me to write it.

Choi, S.H., Gernat, T., Hamilton, A. R. and N. Goldenfeld. 2020. Individual variations lead to universal and cross-species patterns of social behavior. Proceedings of the National Academy of Sciences 117: 31754-31759.
Dukas, R. and P.K. Visscher. 1994. Lifetime learning by foraging honey bees. Animal Behavior 48: 1007-1012.
Gernat, T., Rao, V.O., Middendorf, M., Dankowicz, H., Goldenfeld, N.D. and G.E. Robinson. 2018. Automated monitoring of behavior reveals bursty interaction patterns and rapid spreading dynamics in honeybee social networks. Proceedings of the National Academy of Sciences 115: 1433–1438.
Hunt, G.J., Given, K., Tsuruda, J.M. and G.K. Andino. 2016. Breeding mite-biting bees to control Varroa. Bee Culture, March issue.
Moore, D., Angel, J.E., Cheeseman, I.M., Robinson, G.E. and S.E. Fahrbach. 1995. A highly specialized social grooming honey bee (Hymenoptera: Apidae). Journal of Insect Behavior 8: 855-861.
Robinson, G.E., Underwood, B.A. and C.E. Henderson. 1984. A highly specialized water-collecting honey bee. Apidologie 15: 355-358.
Tenczar, P., Lutz, C.C., Rao, V.D., Goldenfeld, N. and G.E. Robinson. 2014. Automated monitoring reveals extreme interindividual variation and plasticity in honey bee foraging activity levels. Animal Behavior 95: 41-48.
Trumbo, S.T. and G.E. Robinson. 1997. Learning and task interference by corpse-removal specialists in honey bee colonies. Ethology 103: 966-975.
Visscher, P.K. 1988. Undertaker specialists in honey bee colonies. In: Interindividual Behavioural Variability in Social Insects (Jeanne, R.L., ed.). Westview Press, Boulder, CO. pp. 359-383.
Visscher, P.K. and T.D. Seeley. 1982. Foraging strategy of honeybee colonies in a temperate deciduous forest. Ecology 63: 1790-1801.
von Frisch, K. 1967. The Dance Language and Orientation of Bees. Harvard University Press. Cambridge, MA.