DREAM HOME FOR HONEYBEES - Honeybee Democracy - Thomas D. Seeley

Honeybee Democracy - Thomas D. Seeley (2010)


If I can with confidence say

That still for another day,

Or even another year,

I will be therefor you, my dear,

It will be because, though small

As measured against the All,

I have been so instinctively thorough

About my crevice and burrow.

—Robert Frost, A Drumlin Woodchuck, 1936

Like Robert Frost’s woodchuck, a honeybee colony is “instinctively thorough” about its dwelling place, for only certain tree cavities provide good protection from predators and suffcient refuge from harsh physical conditions, especially strong winds and deep cold. No fewer than six distinct properties of a potential homesite—including cavity volume, entrance height, entrance size, and presence of combs from an earlier colony—are assessed to produce an overall judgment of a site’s quality. The care with which honeybees choose their homes has been known for only about 30 years, which might seem surprising given that humans have been culturing these bees since ancient times. The reason that humans have only recently learned about the bees’ real estate preferences is that the essence of beekeeping is the tending of colonies living in hives fashioned by a beekeeper and sited where the beekeeper wants them. The earliest solid evidence of beekeeping comes from Egypt, around 2400 BC, and consists of a stone bas-relief in a temple that depicts peasants removing honeycombs from a stack of cylindrical clay hives and also packing the honey in pots (fig. 3.1). Thus for some 4,400 years the people living in closest association with honeybees have focused on devising housing arrangements for bees that serve human purposes and have largely ignored what the bees’ themselves seek in a home. For example, manmade hives are usually much more spacious than natural nest cavities, so bees living in an apiary will store more honey and swarm less often than will bees living in nature. Likewise, a beekeeper’s hives are located at ground level, which is convenient for humans but dangerous for bees. Honeybee colonies living low to the ground are easily found and attacked by destructive predators, such as bears.


Nests of Wild Colonies

In 1975, when I began to study the democratic house-hunting process of honeybees for my PhD thesis project, I decided that a logical first step was to try to identify what makes a dream homesite for a honeybee colony. This would tell me what a swarm is seeking as it locates multiple candidate sites and works to select the best one. I suspected that to identify the perfect dwelling place for a honeybee colony would be a challenge, because the bees might evaluate several attributes of each candidate site, and they might weigh each attribute differently when judging the overall goodness of a site. Nevertheless, I figured that if I could identify what attributes are important to the bees and if I could determine what preference they have for each attribute, then I would be close to achieving my goal.

I also figured that to determine the bees’ real-estate preferences, I should start by finding trees housing wild colonies of honeybees, sawing them down, and splitting open the tree sections housing the bees’ nests so that I could scrutinize their natural living quarters (fig. 3.2). Each colony living in nature occupies a site chosen by scout bees, so it seemed reasonable to expect that consistencies in the nest sites of these wild colonies would yield clues about the bees’ nest-site preferences. And there could be little doubt that these preferences lie at the heart of the bees’ whole house-hunting process, for it is these preferences that guide swarms to take up residence in suitable nesting cavities.

Back in 1955, Lindauer had reported experiments, conducted in the open countryside east of Munich, in which he presented one swarm at a time with a pair of nest boxes that differed in some property, and then he observed which one attracted the greatest interest from the swarm’s scout bees. These experiments yielded only preliminary findings, because Lindauer could perform just a few trials for each test of a particular nest-site property. Nevertheless, they suggested to him that his bees had chosen among his nest boxes based on differences in protection from the wind, cavity size, presence of ants, and sun exposure. Lindauer was impressed by the bees’ apparent attention to multiple properties of a possible residence when assessing its desirability, wondered what the ideal bee dwelling might be, and suggested that to solve this mystery “it would be best to ask the bees themselves about this matter.” I would start to do so by examining their nests.

The prospect of carefully describing the nests of wild honeybee colonies living in the woods attracted me for emotional as well as rational reasons. While an undergraduate student, I had majored in chemistry and done several small research projects in organic chemistry, biochemistry, and biophysics. Of course, these investigations were all conducted indoors in clean, brightly illuminated, and nearly lifeless laboratories. But now, as a beginning graduate student in biology and novice investigator of animal behavior, I was keen to work outdoors using what has been called the von Frisch-Lindauer approach to animal behavior research. In their autobiographical book, Journey to the Ants, Bert Hölldobler and Edward O. Wilson explain that von Frisch and Lindauer had a philosophy of research based on: a thorough, loving interest in—a feel for—the organism, especially as it fits into the natural environment. Learn the species of your choice every way you can, this whole-organismic approach stipulates. Try to understand, or at the very least try to imagine, how its behavior and physiology adapt it to the real world. Then select a piece of behavior that can be separated and analyzed as though it were a bit of anatomy. Having identified a phenomenon to call your own, press the investigation in the most promising direction.


My thesis advisor, Bert Hölldobler, had presented this way of studying behavior in his ethology course at Harvard and, more importantly, had demonstrated its power by his own spectacularly beautiful studies of ant social behavior. So, by the end of my first year in graduate school, I was raring to go. I wanted to gain a feel for honeybees living in nature, to further analyze the house-hunting piece of their behavior, and to see if I could press the investigation on from where Martin Lindauer had left it some 20 years before.

I knew that I would abscond from Harvard the moment I had finished taking my final exams for the spring semester, and I had my mind set on returning to the Dyce Laboratory for Honey Bee Studies, at Cornell, where I had worked for the previous four summers when an undergraduate student. The director of the lab, Professor Roger A. Morse, was truly a generous man. He welcomed me back, assigned me a desk, and provided several essential tools for the project—a powerful chain saw, steel wedges and maul, and one of the lab’s green pickup trucks. Most importantly, “Doc” Morse arranged for me to team up with a member of the Entomology Department’s technical staff, Herb Nelson, who had worked as a logger in the Maine woods when a teenager and could teach me how to cut down big trees without getting killed.

Herb and I started with some of the bee trees I had discovered back in high school while exploring the woods around my family’s home. These were augmented with ones that I located through a want ad I placed in the local newspaper, the Ithaca Journal. The ad read, “BEE TREES wanted. Will pay $15 or 15 lb of honey for a tree housing a live colony of honeybees. 607-254-5443.” I feared I’d get no calls, but within a week I had secured the rights to 18 accessible bee trees in the woods around Ithaca. Two owners took payment in money; all the rest wanted honey.

The procedure for collecting these nests was simple but somewhat dangerous. Shortly before sunrise, when all bees were still at home, I would hike to a bee tree with a can of calcium cyanide powder (Cyanogas), an old spoon, and some rags. If the nest entrance was high in a tree that I couldn’t climb, I ’d also bring an aluminum extension ladder. My aim was to spoon cyanide powder into the nest entrance and then quickly plug it with the rags. The cyanide powder would react with moisture in the air producing cyanide gas that would kill the bees but, if all went according to plan, not me. (Once I did drop the can of Cyanogas from the ladder, spilling much of its contents, but I managed to hold my breath long enough to climb down, get the lid back on the can, and dash out of the expanding cloud of deadly gas.) By first killing the bees, we could later fell the tree and collect the nest without being ferociously attacked. This protocol also enabled me to census the bee population of each wild colony when I dissected its nest.

Having killed the bees, I’d return to Dyce Lab to pick up Herb and load the truck with the tools we’d need for the day: chain saw, wedges and maul, rope, ramp boards, tape measure, magnetic compass, 35 mm camera, and notebook. Our goal was to cut down the bee tree I had just visited, saw out the trunk section housing the nest, wrestle it onto the truck, and haul it back to the lab. I recall being impressed by Herb’s confidence in driving the truck deep into the woods to get near each bee tree (“We’ll have plenty of traction for getting back out, once we get that big log loaded on.”) and by his careful inspection of each tree’s lean and crown before starting his cutting (“You gotta know which way the tree wants to fall.”). Herb’s lumberjack skills weren’t rusty, and each tree arced down neatly into the woods opening he had chosen. Once we had a tree lying on the ground, we proceeded to cut out the section containing the nest. We did this by making a series of crosscuts, starting far above and far below the nest entrance, and then working our way closer and closer to the entrance until the chain saw started spitting dark-brown punk wood or yellow-brown beeswax, indicating we had breached the nest cavity. We then rolled the nest-containing log—sometimes a massive, 2-meter-long (6-foot) and nearly 1-meter-thick (3-foot) section of the tree’s bole—up into the truck, got it back to the lab, and split it open (fig. 3.3). Finally, we would lug the opened nest indoors where I could dissect it carefully under good light while measuring important features of the nest cavity and its contents. To measure the volume of the cavity, I filled it with liter after liter of sand after removing the combs. As I picked through the broken combs and dead bees, sooner or later coming across the lifeless queen, I felt sad to have killed a whole colony, but also excited, knowing that I was the first human to describe in detail the natural homes of honeybees.


Over the summer of 1975, we collected and I dissected 21 bee tree nests, enough to give us a broad picture of the nests of wild colonies living in the woods. I also located another 18 nests in trees that were left standing and so yielded information only about their entrance openings. Since the nest entrance is the “front door” of a colony’s home it is probably especially important to the bees, so I gave it extra attention. We found that the bees occupied many kinds of trees, including oaks (Quercus spp.), walnuts (Juglans spp.), elms (Ulmus spp.), pines (Pinus spp.), hickories (Carya spp.), ashes (Fraxinus spp.), and maples (Acer spp.). This suggested that the bees don’t have a strong preference for certain tree species.

Not surprisingly, the tree cavities occupied by the bees were generally tall and cylindrical, consistent with the shape of tree trunks. But what was surprising was the discovery that most of these wild colonies were occupying tree cavities much smaller than the hives provided by beekeepers. The average nest cavity was only about 20 centimeters (8 inches) in diameter and 150 centimeters (60 inches) tall; hence, it had a volume of only about 45 liters (41 quarts) (fig. 3.4). A tree cavity of this size provides only one-quarter to one-half of the living space provided by a beekeeper’s hive. Were the bees telling me that they prefer relatively small and snug nest sites, ones in which it might be easier to keep warm in winter? Some of the colonies even occupied tree cavities with only 20 to 30 liters of nesting space, though none was found in a space smaller than 12 liters. Was this lower limit of about 12 liters a sign that bees carefully avoid excessively cramped quarters, ones lacking sufficient room for storing the honey needed to survive winter? Certainly the bees living in these tree cavities were making good use of their living space, for each colony had nearly filled its nest cavity with multiple combs. Because each comb formed a wall-to-wall curtain spanning the (generally) narrow tree cavity, I was impressed by the way the bees had built small passageways in the combs where they were attached to the cavity’s wall, so they could crawl easily from one comb to the next. And it was clear that these bees had organized their use of their combs in the way familiar to all beekeepers, storing honey in the upper region of the nest and rearing brood below. The nests collected in August, by the way, revealed that most colonies had been making good progress in stockpiling their winter heating fuel. The nests that I dissected contained, on average, 14 kilograms (30 pounds) of golden honey. Regrettably, it was all laced with cyanide.

The entrance openings of the bees’ nests also showed consistencies that suggested possible nest-site preferences by the bees. Most nest entrances consisted of a single knothole or crack with a total area of just 10 to 30 square centimeters (2 to 5 square inches) (fig. 3.5). And typically they were located near the floor of the tall tree cavity, on the south side of the tree, and close to ground level. The trends toward small size, floor level, and southern orientation all made good sense to me, for they would make the nest cavity inaccessible to most predators, relatively free of drafts, and perhaps warmed by the sun—all things that would be good for a colony. But the preponderance of nest entrances just a few feet from the ground puzzled me greatly. I figured that a low nest entrance must render a colony vulnerable to detection by predators, such as bears, whose attacks can be fatal. And I knew that in medieval times, in the forests of northern Europe (Germany, Poland, and Russia), one of the ancestral homes of the honeybees imported to North America, raids by bears on honeybee nests in trees were such a great trouble for the forest beekeepers who owned these nests that they invented horrific devices to kill honey-loving bears. One was a hinged platform mounted outside a bee nest. When a bear climbed onto it to attack the bees, it would collapse, causing the bear to tumble onto a grid of deadly sharp stakes below.


So at first I was perplexed by the rarity of nests high in trees. But as will be explained shortly, we now know that bees actually have a strong preference for nesting cavities with entrances located high above the ground. I also now know that my initial report of most nests being near ground level was an error generated by an unintentional bias in the way I had sampled the population of natural nests. Because the nests I studied were ones that had been noticed inadvertently by a person walking past a bee tree, and because people are much more likely to notice bees trafficking from a ground-level nest entrance than a tree-top one, I unwittingly studied nests whose entrances were far lower than is typical. I am confident on this point because several years later, when I became a bee hunter and mastered the ancient craft of lining bees (locating bee trees by baiting foragers from flowers and observing their flights back to their nests), I found that every hunt ended with me straining to spy the bees zipping in and out of a nest entrance high in a tree, like the one shown in figure 3.2. To date, I have located 27 bee trees by bee lining and can report that the average height of their nest entrances is 6.5 meters (21 feet). Needless to say, I’m now alert to the hidden danger of unintentional sampling bias.


Location, Location, Location

Even though the descriptive study of the natural homes of honeybees was destructive, it remains one of my favorite studies, for it put me in touch with honeybees living in nature and it helped me gain some self-confidence as a researcher. It also guided me throughout the next step of my investigation of how honeybee swarms choose a home: testing whether the nest-site patterns we had found—in cavity volume, entrance area, entrance height, and so forth—were a result of preferences of scout bees or were simply a consequence of the tree cavities that were available. The idea for the design of the test came from what I had read about beekeeping in East Africa and South Africa. In these regions, beekeepers acquire bees by hanging hives (usually hollowed logs with the ends stoppered except for an entrance hole) in trees and waiting for swarms to occupy them. I had not heard or read of anybody in North America catching swarms with “bait hives,” but I reasoned that if one could do so, then I could ask the bees about their nest-site preferences by putting up groups of two or three nest boxes, with the boxes in each group identical except for one property, such as cavity volume or the height of the entrance above the ground. I hoped that scouts from wild swarms would discover the groups of nest boxes and would reveal their real-estate preferences by choosing among the boxes in a group and consistently occupying those with certain attributes.

Almost always, one starts an experimental study with a small-scale, low-cost pilot study to figure out what methods are apt to work before undertaking a costly full-scale investigation. In the summer of 1975, I did a pilot study to see if wild swarms would occupy bait hives often enough to give my experimental plan a reasonable chance of success. Using some scrap plywood scrounged from Dyce Lab, I built six nest boxes, each one a simple cubic box 35 centimeters (14 inches) wide, tall, and deep, and bearing a 4.5-centimeter (1.75-inch) diameter entrance hole on the front side. I designed these boxes to mimic the nest cavities I was finding in bee trees. My bee houses looked like birdhouses on steroids, except that each one had chicken wire nailed over its entrance opening to keep birds out while letting bees in. I took each nest box to a place I knew that I’d enjoy visiting in my home “territory” of Ellis Hollow, and nailed it about 5 meters (15 feet) of the ground onto the side of a large tree. I still remember vividly the thrill I felt a few weeks later, in late June, when I checked the nest box that I ’d mounted on a dead elm tree along Cascadilla Creek and saw dozens of leather-colored honeybees bustling in and out of its entrance. A swarm had moved in! Yippee!! When swarms occupied two more of my nest boxes over the next few weeks, I was even more excited. This pilot study had been simple, but it had yielded the triumph of knowing that my experimental plan was likely to work. My plan for the next summer was now clear: I had to set out dozens and dozens of nest boxes of various designs to “ask the bees about this matter” of their ideal dwelling place.

The plan worked well. Each summer in 1976 and 1977, I set up more than two hundred green nest boxes in groups of two or three across Tompkins County and each summer over half of my nest-box groups attracted at least one wild swarm. The boxes within each group were spaced about 10 meters (33 feet) apart on similar-sized trees or, even better yet, on power-line poles where they were perfectly matched in visibility, wind exposure, and the like (fig. 3.6). Each group of boxes was designed to test one nest-site preference, and it did so by giving swarms a choice between one box whose properties all matched those of a typical nest site in nature (e.g., average entrance area, average cavity volume, etc.) and one or two other boxes identical to the first except in one property, the value of which was atypical. In this way, wild swarms were tested for a preference in the one variable that differed between the boxes. For example, to test for a preference in entrance size, I set up pairs of cubical nest boxes that were identical except that one had a typical entrance area of 12.5 square centimeters (2.5 square inches) and the other had a larger than usual entrance area of 75 square centimeters (15 square inches). Similarly, to test for a preference in cavity size, I set up trios of cubical nest boxes that were identical except that one box had the typical cavity volume of 40 liters while the other two boxes had volumes at the two tails of the distribution of nest cavity volumes: 10 and 100 liters.

To build the many nest boxes needed for this study, I spent most of my Christmas break in 1975 sawing and hammering and painting in the woodshop at Dyce Lab. There I constructed 252 nest boxes and used up enough plywood (more than 70 sheets) for building a small house. With these hundreds of nest boxes, I would eventually capture 124 swarms in 1976 and 1977.

As is shown in table 3.1, the swarms demonstrated preferences in the following nest-site variables: entrance size, entrance direction, entrance height above the ground, entrance height above the cavity floor, cavity volume, and presence of combs in the cavity. The bees had revealed to me that they prefer a nest entrance that is rather small, faces south, is high of the ground, and opens into the bottom of the nest cavity. These four preferences regarding the entrance opening no doubt help a honeybee colony survive against threats of cold winters and dangerous predators. A small entrance is easily defended and helps isolate the nest from the outside environment. An entrance high up in a tree is less apt to be discovered by predators than one near the ground, and is certainly inaccessible to predators that cannot fly or climb trees. An entrance at the bottom of the nest cavity rather than at the top may help to minimize the loss of heat from the colony by convection currents. And an entrance that faces south provides a warm, solar-heated porch from which foragers can take of and on which they can land. Beekeepers, incidentally, face their hives to the south to help their bees fly out in cool weather. This southern orientation is particularly important in the winter months, when bees go outside on sunny days to make their critical “cleansing flights,” that is, to defecate. A Canadian bee researcher based in Alberta, Tibor Szabo, compared colonies living in south-facing and north-facing hives. He found that those in south-facing hives were less apt to suffer a hive entrance plugged by ice in winter and were more populous in spring.


The pattern of nest-box occupations by swarms also showed clearly that the bees avoid cavities smaller than 10 liters or greater than 100 liters, and that they very much like 40-liter cavities (about the size of a wastebasket), especially ones already equipped with combs. Probably the main problem the bees face regarding cavity volume is avoiding undersized cavities, since most tree cavities are too small (less than about 15 liters) to hold the store of honey a colony needs to survive winter. The evidence supporting this assertion comes from a small study I did with one of my brothers, Daniel H. Seeley, who owns a whole hillside in Vermont that was logged in the 1800s but is now forested with stately sugar maple (Acer saccharum) and beech (Fagus grandifolia) trees. In October 1976, Dan and I packed up logging tools and drove north from Cambridge, Massachusetts, to Roxbury, Vermont, to spend several days of Indian summer finding out what size cavities scout bees are apt to encounter when prospecting for homesites. Working over a 0.32-hectare (0.8-acre) area, we felled every tree more than 30 centimeters (12 inches) in diameter, and we sawed each felled tree into 120-centimeter (4-foot) lengths to expose any cavities they contained. We dissected 39 trees and found 14 cavities with an opening to the outside that could provide access to a scout bee. Of these 14 tree hollows, only two (14 percent) were larger than 15 liters; they were 32 and 39 liters.


The preference for a site filled with combs—built by a preceding colony that did not survive winter—doubtless refects the tremendous energy savings that a colony enjoys if it occupies a site already furnished with a full set of combs. The energy thus saved turns out to be a large fraction of the honey store that a fledgling colony needs to survive its first winter. This is shown by the following calculations. A typical nest in a bee tree contains some 100,000 cells arranged in eight or so combs whose total surface area is about two and half square meters (3 square yards). Building this impressive edifice requires about 1,200 grams (2.5 pounds) of beeswax. Given that the weight-to-weight efficiency of beeswax synthesis from sugar is at most about 0.20, we can estimate that building the combs in a typical nest requires about 6.0 kilograms of sugar, hence about 7.5 kilograms (16 pounds) of honey. This mass of honey is about one-third of what a colony will consume over winter. Storing these 7.5 kilograms of honey in the colony’s food supply for winter rather than spending it on comb building will greatly boost a colony’s odds of surviving its first winter. Recall that I found that 76 percent of the colonies newly established in tree cavities around Ithaca die during their first winter, and that nearly all of the colonies that do so succumb to starvation.

The nest-site properties for which I detected no preference were entrance shape, cavity shape, cavity draftiness, and cavity dryness. Honeybees probably prefer tight and dry nest cavities, but because a colony can caulk with tree resins any cracks and crevices that let in drafts and water, the nest-site scouts apparently do not pay much heed to these properties. In contrast, a colony cannot modify the volume of a nest cavity, the height of its entrance, or the direction in which it faces, so to get a homesite that meets its needs in these properties the nest-site scouts must pay close attention to these properties when evaluating prospective nest sites. The ability of a honeybee colony to remedy a drafty or damp site was neatly demonstrated by the swarm bees that occupied my experimental nest boxes. I had rendered some of the boxes drafty by riddling their fronts and sides with six-millimeter (quarter-inch) diameter holes spaced 7.5 centimeters (3 inches) apart (fig. 3.7). Other boxes I had made damp by dumping 2 liters (2 quarts) of waterlogged sawdust onto the floor of each box. Every swarm that moved into one of the drafty boxes soon made it draft free by plugging with tree resins all the holes I had drilled. Likewise, every swarm that occupied one of the damp boxes quickly rendered it dry by hauling out all the soggy sawdust that I had dumped inside. I was greatly impressed by the bees’ tidiness.



One thing that makes studying honeybees so enjoyable is the way that what is learned through curiosity-driven research often turns out, unexpectedly, to have real practical value. My best example of this phenomenon is the way that knowing something about the defecation habits of Asian honeybees helped defuse tensions between the United States and the Soviet Union back in the 1980s. This story starts in the late 1970s when I had finished graduate school and was keen to travel overseas and learn about the marvelous species of honeybees that live in the Asian tropics: the Asian hive honeybee (Apis cerana), the dwarf honeybee (Apis florea), and the giant honeybee (Apis dorsata).With support from the National Geographic Society, my wife Robin and I undertook a 10-month study of the colony defense strategies of the three Asian honeybee species living in Thailand. We set up camp in the pristine mountain forests of the vast Khao Yai National Park in northeast Thailand, where one can still enjoy the sight of hornbills wing ing their way between towering dipterocarp trees, the eerie smell of Asian tiger urine deposited along a trail, the whooping calls of white-faced gibbons shortly after sunrise, and the mysterious biology of the Asian honeybees. Gradually we assembled a picture of each honeybee species’ fascinatingly complex array of colony defenses against such enemies as giant hornets, weaver ants, honey buzzards, tree shrews, rhesus monkeys, and honey bears. This was field biology done for biology’s sake, and it was a wonderful adventure for two newlyweds. Sometimes I wonder, though, if even half a dozen biologists worldwide have read closely the beautifully detailed, 21 page report on the Asian honeybees that we published in the scientific journal Ecological Monographs.

A few years later, however, and to my amazement, the knowledge that we’d gained about the Asian honeybees proved important to a large international audience. In 1981, the secretary of state in the Reagan administration, Alexander M. Haig, alleged that the Soviet Union was waging or abetting chemical warfare against opponents of the communist governments in two countries bordering Thailand: Laos and Kampuchea. If true, this was a violation of two international arms-control treaties, the 1925 Geneva Protocol and the 1972 Biological Weapons Convention. The main evidence cited by Haig was a material called “yellow rain,” that is, yellow spots less than 6 millimeters (one-quarter inch) in diameter that were found on vegetation at alleged attack sites and that supposedly contained fungal toxins. I realized, however, that the yellow spots that U.S. officials called yellow rain were indistinguishable from the yellow spots I called honeybee feces. They were identical in size, shape, and color. Further work revealed that both contained bee hairs and were laden with pollen grains from which the protein had been digested. Eventually, I was able to help Matthew Meselson, a professor of molecular genetics at Harvard and an expert on chemical and biological weapons, show conclusively that yellow rain was indeed honeybee feces, not chemical warfare. One wag said we had uncovered the work of “KGBs.” Shortly after yellow rain was proven to be bee poop, in 1984, officials of the U.S. State Department, without fanfare, ceased accusing the Soviets of violating the two arms-control treaties on chemical and biological weapons.

The yellow rain story is a striking example of how research rooted in sheer curiosity can unexpectedly yield useful knowledge, but it is not so unusual, for real-world benefits often bubble up from basic research. My first taste of pursuing personal curiosity and getting an unexpected bonus of practical results came from my study with Doc Morse of the nest-site preferences of swarms. In the summer of 1976, we had groups of nest boxes set up in more than 100 sites around Tompkins County, and we caught more than 60 swarms. Given our high success rate in trapping swarms, Doc realized that we should translate our findings about the bees’ nest-site preferences into recommendations for beekeepers on how to build and position bait hives to catch wild swarms of honeybees. We prepared a simple design (fig. 3.8) along with a set of guidelines for situating bait hives—a good location is about 5 meters (15 feet) of the ground, highly visible but fully shaded, and facing south—and published these in the beekeeping magazine Gleanings in Bee Culture and as a Cornell Cooperative Extension Bulletin. Beekeepers responded enthusiastically. Before this, beekeepers wanting to capture wild swarms had to rely on being notified when a swarm had settled somewhere, then they had to hurry to put it in a hive before the bees finished selecting a nest site and few away to their chosen home. With bait hives, beekeepers can collect swarms automatically.

In recent years, other bee scientists have designed cheaper, lighter, and tougher bait hives made of reinforced wood pulp material, and have devised scent lures that slowly leak a 1:1:1 blend of chemicals (citral:geraniol:nerolic+geranic acids) from a small polyethylene vial. These scent lures mimic the attraction phero-mones that scout bees release from their scent organs to mark a desirable home-site (discussed further in chapter 8). Experiments by Justin Schmidt at the USDA Bee Research Center in Tucson, Arizona, have shown that a bait hive with a scent lure can be five times more likely to attract a swarm than one without a lure, probably because the artificial attraction pheromones makes a bait hive much more likely to be discovered by a scout bee, but perhaps by also making it more attractive. Wood pulp bait hives (sometimes called “swarm traps”) and swarm scent lures are now produced commercially and are sold by the companies that sell beekeeping equipment. Each summer, I put up a half dozen bait hives, partly because I can always use a few additional colonies of bees, but mostly because I like getting free bees.


Property Assessments

Probably every homeowner has wondered how his or her local tax assessor combines information about the size of a house lot, the floor area of the house on the lot, the number of bedrooms and bathrooms in the house, and so forth to determine the assessed value of a particular property. I began wondering the same thing about scout bees in August 1974, as I watched several scouts scrutinizing a candidate dwelling place. This happened in the summer before I went of to start graduate school at Harvard. I was working happily for Doc Morse at Cornell’s Dyce Laboratory, but I was a little worried about my choice of problem for a doctoral thesis: deepening Martin Lindauer’s work on how a swarm chooses its home. In 20 years, nobody had tackled, much less solved, the many mysteries raised by Lindauer’s study. Clearly, there was a first-rate opportunity here, but could I succeed in making something of it? To begin to see what I might do, I decided simply to watch, with my eyes wide open, a swarm go through its democratic decision-making process. In working for Doc, I had learned how to make an swarm—by shaking a colony (queen and workers) into a cage to render them homeless, and then feeding them lavishly with sugar syrup to get them stuffed with food like natural swarm bees—so I prepared a swarm and set it up behind my parent’s house in Ellis Hollow (see fig. 1.7). I also nailed a nest box that I built from scrap plywood to a white pine tree about 150 meters (500 feet) away, hoping the swarm’s scouts would discover it and select it for their new homesite. I mounted the box at eye level so I could observe easily any scout bees that might visit it.

The weekend that I watched this swarm turned out to be a milestone in my life. The swarm’s scouts quickly began advertising several prospective homesites with their dances, and before long one bee was dancing with eye-catching en-thusiasm for a location nearby and to the north: my nest box! This bee’s lively dance soon gave rise to a small crowd of bees at the box. Back at the swarm, I adorned a few of the bees dancing for my box with dots of paint on the thorax and abdomen, giving different bees different color combinations. This simple trick transformed these individuals from mere members of Apis mellifera into personal acquaintances whose affairs became of the greatest interest to me. At the swarm, I saw how a scout would perform a bout of vigorous dancing, contact another bee with agile antennal movements to beg a droplet of honey, perhaps groom of a pesky feck of paint, and then fly away for 20 to 30 minutes. When she returned, she might dance again but she might also just settle quietly within the swarm cluster. At the nest box, I saw how my labeled bees would land and agitatedly run into the entrance opening and then scurry back out a minute or so later, whereupon they would either crawl briefly around the entrance and then pop back inside or conduct a slow, hovering flight around the box, usually facing it and maneuvering within inches of the box, apparently giving the nest structure a detailed visual inspection (fig. 3.9). Never before had I seen a honeybee behave with such persistent intention to gather information. I was thoroughly intrigued. Also, my worries about choice of thesis topic had largely vanished, for I felt confident that I could explore how a scout bee inspects a site.

I began to study closely the inspection behavior of nest-site scouts the following summer, in June 1975. To do so, I had to leave the forested countryside around Ithaca, where since early May I had been busy finding bee trees and describing natural nests, and shift to a location largely devoid of natural homes for honeybees: Appledore Island (fig. 3.10). This rocky, wind-blown island is barely 900 meters (half a mile) long and lies 10 kilometers (6 miles) out in the Atlantic Ocean of the coast of southern Maine. It is the site of the Shoals Marine Laboratory of Cornell University and the University of New Hampshire. I was attracted to Appledore because it has no resident honeybees and it has no large trees. Instead, it is inhabited by approximately one thousand breeding pairs of herring gulls (Larus argentatus) and great black-backed gulls (Larus marinus), and it is covered by thickets of poison ivy plants rising three meters (10 feet) tall together with tangles of blackberry brambles and wind-battered cherry bushes, all richly fertilized by the gulls. I figured that if I took a honeybee swarm out to this scrubby island, the bees would have to concentrate their house hunting on the artifcial nesting sites that I would provide. If so, then I could observe their behavior under controlled conditions and learn how they assess a possible dwelling place.


My first goal on Appledore Island was to make a detailed description of the behavior of scout bees inspecting nest sites. I hoped these observations would suggest how scouts evaluate the critical nest-site properties. It was particularly important to be able to watch scouts inside a prospective homesite in order to understand how they might measure such things as the cavity’s volume and the entrance’s height from the cavity floor. To this end, I built a lightproof hut with a cube-shaped nest box mounted on one of its walls (fig. 3.11). The box was positioned outside a window covered with a red filter (bees cannot see red light) so that I could peer inside the box without disturbing the scouts. The inner surfaces of the box bore a grid-coordinate system that enabled me to record where a scout bee went while she was in the cavity. After setting up the hut in a valley on one side of the island, I positioned a small swarm (about 2,000 bees) at the center of the island. The bees had been given dots of paint using a color code that made them individually identifiable. Then I retired to the hut to wait for scout bees.



The first time I tried this, I waited all morning at the hut without being visited by a scout bee, which surprised and dismayed me. When I returned to the swarm around midday, my spirits sank further, for I saw several scouts performing long-lasting dances indicating a location directly away from my hut. Rats! What possibly could the bees have found? I made careful measurements of the direction and distance to the location specified by the bees’ dances and plotted the site on my topographical map. Now my spirits sank even deeper, for the bees’ dances were indicating unmistakably one of the two lobster fshermen’s cottages on the south shore of the island, specifically Rodney Sullivan’s cottage (fig. 3.12). When I had arrived on Appledore a few days before, and had been getting oriented to my new surroundings, I had been told to keep well away from the fshermen’s private properties, especially Rodney’s place, for he valued his privacy and kept a loaded shotgun behind his front door. What should I do? I sought advice from the laboratory’s director, Professor John M. Kingsbury, and he kindly went with me over to Rodney’s cottage, to introduce student Seeley to fisherman Sullivan. We went by boat so Rodney would be able to see us coming, approaching from the front (the water), even while my bees were “attacking” from the rear (the land). He heard our boat approach, came out on his porch, and told us to come ashore. After we climbed up the rocks to his house, he told us he had an emergency: hundreds of bees were buzzing in the stovepipe to his woodstove! “Never seen this before! Could they’a been blown out here during the [recent] storm?” I didn’t answer the question, but I did offer to help. While Rodney made a fire in the stove to smoke out the bees, I climbed his steep roof (slippery with splotches of fresh gull poop) and taped window screen over his chimney flue to exclude bees in the future. Rodney was delighted…I was relieved.


No longer distracted by Rodney’s house, the scouts from my swarm soon started to appear at my observation nest box and began to perform their striking inspection behavior. I saw that a scout bee needs 13 to 56 minutes (average 37 min) to inspect a prospective nest site. Her complete inspection is a summation of 10 to 30 journeys inside the cavity, each one lasting usually less than a minute and alternating with equally brief periods outside, during which the bee examines the exterior of the nest structure. I call this first inspection, when a scout is popping in and out of the cavity, the discovery inspection. Following the discovery inspection, a scout returns to the swarm, and if the site is desirable she will advertise it with a waggle dance and usually will then make repeated visits to the site at approximately half-hour intervals, but these subsequent site visits tend to last only 10 to 20 minutes (average 13 min) and don’t involve so much in-and-out activity.

When a scout is inside a cavity, conducting her discovery inspection, she devotes most of her time (about 75 percent) to rapid walking across the inner surfaces. This quick pacing about is interspersed with pauses to rest, groom, and release attraction pheromones from the scent organ, and with short hopping flights. The inside of a dark cavity seems an odd place to attempt to fly about, yet the bees make these little flights that last less than a second and that move a bee from one point to another on the walls, floor, or ceiling of the cavity. A geometric pattern in the movements of scout bees is that early in the discovery inspection a scout walks primarily near the entrance during her journeys inside, whereas later she penetrates to the deepest recesses of the cavity (fig. 3.13). Three-dimensional reconstructions of the walking paths of individual scouts reveal that when the inspection is finished, a scout has walked 60 meters (200 feet) or more around the inside of the cavity and has covered all its inner surfaces.

I spent four weeks on Appledore Island in 1975, departing without solving the mystery of how scout bees evaluate candidate nest sites. Nevertheless, I felt returned to the island in July 1976, I focused on the puzzle of how scout bees can measure the size of potential nest cavities, which are immense compared to the size of a bee. Cavity volume is the nest-site property that is perhaps most critical to a colony’s long-term survival, since any colony occupying a hollow 10 liters or smaller cannot store suffcient honey to get through winter, so I suspected the bees have evolved a way to measure cavity volume accurately.


How do scout bees measure the volume of a cavity? Their extensive walking in the course of an initial inspection could provide the basis for an estimate, but another hypothesis was that they simply go inside and look around. I first performed experiments with nest boxes in which I could vary the interior light level (by changing the amount of light coming in through the entrance hole) and the traversable surface area (by coating the inner surfaces with Fluon, a teflon-type material that creates a waxy surface that bees cannot walk up) (fig. 3.14). I found that in order to measure a cavity’s volume, scout bees need either interior illumination greater than 0.5 lux (about the illumination provided by a full moon) or inner surfaces that can be traversed freely. What are the conditions inside a typical tree cavity? Certainly the wood walls inside a cavity are easily traversed by a scout bee. To measure the light level in cavities of the sort inspected by scout bees, I built a model based on the measurements I had made of natural nests. It had a series of openings into which I inserted a light meter. I found the illumination to be less than 0.5 lux except near the entrance opening where some sunlight streams in. Evidently, in nature scout bees rely primarily on walking about in a prospective nest site to measure its volume.

To test this hypothesis more directly, I tried altering a scout’s perception of a cavity’s volume by manipulating the amount of walking required to move from point to point inside a cavity. To do so, I invented a bee treadmill, a cylindrical nest box mounted vertically on a turntable that enabled me to rotate the box smoothly while a scout bee was inside (fig. 3.14). By means of a window at the top, I could look inside and see which way the bee was walking; then I could turn the walls according to whether I wanted to increase or decrease the amount of walking required for her to complete a horizontal circuit. So, if a scout came in through the entrance and I rotated the walls in the direction she walked, she was carried along and quickly found herself back near the entrance. But if I turned Fig. 3.14 Experimental apparatus used in testing how a scout bee measures the volume of a cavity. Left: Apparatus in which the cavity volume could be varied between 5 liters (with the inner lid down) and 2 5 liters (inner lid up). The light baffle made it possible to vary the amount of light coming in through the entrance hole, to see whether scouts could measure the size of the cavity without relying mainly on vision. Coating the interior wall surfaces made it possible to vary the amount of traversable surface area inside the box, to assess the importance of walking. Right: Apparatus in which the wall of the cylindrical cavity could be rotated to increase or decrease the amount of walking a scout bee had to do to circumnavigate the space. the walls against her, she would need to walk much longer to make her way back around to the entrance opening. The entire device was mounted in my lightproof hut with a short tunnel connecting the entrance in the nest box to an opening in the hut’s wall. The only light entering the nest box came in through its entrance, and it is likely that this bright spot provided each scout bee inside the box with a visual reference point, both for finding her way out of the box and for monitoring her progress in circumnavigating it.


Fig. 3.14 Experimental apparatus used in testing how a scout bee measures the volume of a cavity. Left: Apparatus in which the cavity volume could be varied between 5 liters (with the inner lid down) and 25 liters (inner lid up). The light baffle made it possible to vary the amount of light coming in through the entrance hole, to see whether scouts could measure the size of the cavity without relying mainly on vision. Coating the interior wall surfaces made it possible to vary the amount of traversable surface area inside the box, to assess the importance of walking. Right: Apparatus in which the wall of the cylindrical cavity could be rotated to increase or decrease the amount of walking a scout bee had to do to circumnavigate the space.

The volume of this experimental box was 14 liters, on the boundary between an unacceptably small cavity and a suitably large one. If walking contributes to the perception of volume, then the first scout to discover the box should find it either more or less attractive than its true volume would merit, according to whether she had been made to walk more or less than she would in a normal 14-liter cavity. The assay of her evaluation of the box was the number of other scouts she recruited to visit the box; she should recruit more scouts if she found the box suitably large than she would if the box had seemed unacceptably small. That is precisely what I observed in four trials of this experiment: seven or nine recruits in 90 minutes when the bee walked lots, but only zero or one recruit in 90 minutes when the bee walked little. Evidently, only the bees that had taken “long walks” judged the box sufficiently roomy and recommended it to their fellow scouts with enthusiasm. It seems clear, therefore, that a scout’s estimate of the volume of a cavity is proportional to the amount of walking she must do to circumnavigate it. Every step is a measurement.

Nigel R. Franks and Anna Dornhaus, biologists at the University of Bristol in England and the University of Arizona in the United States, have recently suggested a simple method by which a scout bee might judge a cavity’s roominess using the information gained from the walking and flying movements that she makes inside the cavity. They point out that physicists have long known that for any open space the mean free path length (MFPL) of wall-to-wall lines drawn in all directions across the space is equal to four times the volume (V) of the space divided by its internal surface area (A): MFPL = 4V/A. Thus, volume is proportional to mean free path length multiplied by internal surface area: V = (MFPL x A)/4. It is possible that the extensive walks made by scout bees are a means of estimating a cavity’s internal surface area. It is also possible that the short hopping flights made by scout bees—which I had reported but had not linked to the volume estimation process—are a means of seeing how far they can fly before hitting a wall, that is, a means of estimating the mean free path length. If both possibilities prove correct, then it may be that all a scout bee needs to do to be certain that a cavity is sufficiently spacious is to ascertain that the cavity provides a suitable combination of internal surface area and mean free path length. Certainly the results of my experiment with the rotating-wall nest box are consistent with this hypothesis; forcing a bee to walk farther to return to the entrance (thereby increasing the bee’s estimate of internal surface area?) resulted in a larger estimate of the box’s volume. Franks and Dornhaus have proposed an ingenious experimental test of their idea, one that involves hanging a rigid curtain across much of the interior of a nest box and coating the curtain with Fluon so that scout bees cannot walk on it. This curtain will shorten the mean free path length of flights inside the nest box but will not change either its volume or its walkable surface area. One can then see if the bees behave as if the box has been shrunk. I hope the test will be performed soon, and that it will provide support for the proposed rule of thumb, for I think it’s an elegant solution to a tough problem.