Apis florea dancer A central question in the study of animal behavior is how an animal with a relatively simple nervous system is able to solve complex behavioral problems. I have worked primarily with honey bees (Apis), long an important model system for studies of learning, visual behavior, communication, and sociality.  Bees are a wonderful model organism because they have fascinating behavior that is both experimentally tractable and, through comparative studies of different Apis species, amenable to testing of phylogenetic and functional evolutionary questions.

My research interests all concern behaviors involved in the use of space--orientation, navigation, communication (of spatial locations), and foraging.  These interests are described below.  Bibliographic citations refer to papers listed here

Specific Topics

Navigation and Spatial Cognition

Honey bees navigate over enormous distances with astonishing flexibility, although they are equipped with vision far feebler and brains far smaller than ours.  Foraging honey bees can travel up to 10 km from their nest in search of food, and so they face the problem of steering a course to a familiar goal that is not directly in view.  For insects and many vertebrates, landmarks and celestial cues (the sun and patterns of polarized sky light) provide the most important sources of navigational information.  A major focus of my research has been on what bees learn about these two navigational references and how they learn it (reviewed by Dyer 1994, 1996).


An ability to use landmarks for long-distance navigation implies that the animal can link together visual images recorded in successively encountered parts of the terrain.  How faithfully does the insect's spatial memory represent the geometrical relationships among routes that the bee has traveled?  Some researchers have suggested that the underlying spatial representation approximates the computational sophistication of human mental maps, and confers upon the bee a human-like ability to calculate novel shortcut routes among separately visited sites.  In experiments designed to test this hypothesis, however, I found that bees orienting to landmarks are constrained to follow routes along which they can see familiar sequences of landmarks.  Thus, at best, they store one or more "route maps" in memory, and do not encode the spatial relationships among separately traveled routes (Dyer 1991a).

On the other hand, bees can use these route maps with great flexibility.  For example, bees can compensate for displacements from a familiar route, adjusting their response to familiar landmarks seen from a new vantage point, and responding appropriately to familiar landmarks encountered in an unfamiliar context (Dyer et al. 1993; Dyer submitted).  Also bees that change nesting sites with a reproductive swarm reorient quickly to a new nest site, but remember how to get to feeding sites they have visited from the old nest (Dyer 1993a); they also remember where the old nest is, and can return there if deprived of the new nest (Robinson and Dyer 1993).  Thus, they rapidly reorganize their responses to familiar landmarks, but retain their ability to use them in the original way.

The experiments discussed so far have all dealt with the "contents" of spatial memory in thoroughly experienced bees.  One of my former graduate students, Elizabeth Capaldi, considered the acquisition of spatial memory by naive bees, examining what young bees learn about landmarks in their environment prior to beginning their lives as foragers.  During orientation flights, which are also performed by experienced bees that have moved with their colony to a new terrain, young bees explore the surrounding environment for only 5-10 min.  As an assay of landmark learning, we studied the homing ability of bees that have been displaced from the nest after a single orientation flight.  From release sites up to 150 m away, bees depart directly toward the nest even though it is not in view.  Thus, during their brief exploration of the environment, the bees evidently started to assemble route maps extending modest distances from the nest (Capaldi and Dyer 1996).

More recently, another former student, Cindy Wei (Wei et al. 2002; Wei and Dyer 2009), has studied how bees use orientation flights  to adjust their investment in learning landmarks around a feeding place, especially when they experience in a change in the distribution or quality of the food.  

Let's see what these orientation flights look like, and how they change with experience.

Orientation flight performed by a bee on her first departure from a newly discovered feeding place.  The black objects are PVC cylinders that serve as landmarks.  The food, sugar water  in an inconspicuous vessel, is where the bee starts her orientation flight.  The departing bee is heading off toward the hive.

The same bee departing after her tenth visit to the same feeding place.

Bees therefore stop doing orientation flights after they already know where the food is.  Interestingly, if upon their return an experienced bee is forced to search for the food before finding it (because we have taken it away temporarily), she again does an extended orientation flight.  The longer the delay the longer the orientation flight that is induced.

Cindy (Wei et al. 2002) showed that the duration of the orientation flight also varies according to the complexity of the environment (longer orientation flights when their are more landmarks present) and when the location of the food and surrounding landmarks is spatially unpredictable.

Go to:


Celestial Cues

The sun has crossed the sky roughly 200 billion times since bilaterally symmetrical animals arose on earth, so it is hardly surprising that many diverse creatures have evolved the ability to use this highly reliable feature of the environment for orientation.  What is intriguing, however, is that many animals, including honey bees, learn the pattern of solar movement relative to earth-bound features, which allows them to use the sun as a true compass.  Sun compensation entails integrating visual information about the sun's changing direction with temporal information from the circadian pacemaker.  In my Ph.D. work I showed that bees learn the sun's course in relation to landmarks seen during the flight.  They can use their memory of this relationship to perform their communicative dances when they have flown under a cloud cover (Dyer and Gould 1981; Dyer 1987a).  In Apis mellifera, dances are normally performed on the vertical combs in the darkness of the nest.  The dancer indicates the direction of food relative to the sun in the orientation of her body relative to gravity.  On cloudy days, a bee determines her flight direction relative to the sun by retrieving a time-linked memory of the sun's position relative to landmarks seen during the flight.

In studies with an Asian honey bee, Apis florea (Dyer 1985a), I found that dancers could also learn the sun's course relative to a second set of landmarks, those visible while dancing on the nest.  This species nests in the open rather than in enclosed cavities, and dances are normally oriented directly to celestial cues. Dancers in this species cannot use gravity, and so the ability to orient dances to landmarks probably plays a crucial role in their communication system.  Initially, I thought that this ability was restricted to this species, but Elizabeth Capaldi and I (1995) recently discovered that A. mellifera can also use landmarks as a reference for dance communication (although it should rarely need to use this ability).  This discovery opens the way for experiments using the dance to explore the sensory and learning mechanisms underlying orientation to landmarks, just as it has long been used to study orientation to celestial cues.

In work with another former graduate student, Jeffrey Dickinson (Dyer and Dickinson 1994, 1996; Dickinson and Dyer 1996), I have explored an astonishing feature of sun-compass learning that was first described in the 1950s:  animals that have been exposed only to a portion of the sun's course behave as if they can estimate its position at times of day (and night; Dyer 1985b) when they have never seen it.  We designed an experiment to test various computational models that had been proposed to account for this ability.  We observed the dances of partially experienced bees that flew to a familiar feeding site at times of day when they had never seen the sun.  The sky was cloudy during the experiment, so the bees were forced to rely upon an estimate of its position rather than a directly measured or previously memorized position.  Our data led us to reject all of the previous models.  Each had predicted that the bees should estimate unknown positions of the sun by fitting a specific linear function to the set of time-linked positions that they had experienced.  Our partially experienced bees seemed to be innately informed about the actual non-linear pattern of the sun's azimuthal movement over time (specifically, the slow rate of change in the morning and afternoon and the rapid rate of change at midday).  The innate sun-azimuth function initially only approximates the actual pattern of solar movement, but becomes more accurate as bees obtain additional flight experience.

The bee's ability to estimate unknown portions of the sun's course raises questions about how the bee's brain integrates spatial and temporal information, and how individually acquired information is integrated with innately specified information.  We have begun to explore various new modeling approaches, including connectionist neural networks, to understand this phenomenon (Dickinson and Dyer 1996).  One insight of this work has been that the bee's innate representation of the sun's course may have been designed to allow rapid learning of the actual pattern of solar movement at the season and latitude where they find themselves living.

Go to:


Sequential Decision-Making

Wharton et al. 2008
Wharton et al. 2009
Townsend-Mehnler and Dyer (in preparation)

Evolution of Dance Communication and Visual Orientation

Through comparative studies of three Asian Apis species plus A. mellifera, I have studied the evolutionary history and adaptive design of the dance language, which successful foragers use to communicate the location of food.  By taking comparative study of this particularly tractable behavior to the deepest level possible, I hope to expose both the potentialities and the limitations of phylogenetic analyses of behavioral evolution. My thesis work in India challenged previous suggestions that the dances of different modern species reflect a historical progression in the complexity of the bees' direction code.  I found instead a deep dichotomy between A. florea, the species previously thought to represent the primitive state, and the other Apis species (Dyer 1985a; in prep. a).  Given the phylogenetic relationships among the species, it is now ambiguous whether the distinctive elements of A. florea's dance should be viewed as derived rather than ancestral (Dyer 1991b, 2002).  In other work in Thailand (Dyer and Seeley 1991a),  I have studied the adaptive design of the distance code of the dance language, specifically focusing on interspecific differences known as "dialects."   We found, in contrast to the traditional view, that interspecific similarities and differences could not be explained by similarities or differences in flight range or habitat.  Thus, either there is no relationship between dialect and these ecological variables, or  the relationship is more subtle than was previously supposed.

Another project that I hope to develop in the future is an evolutionary study of the structural and physiological bases of visual resolution and sensitivity in the bees' compound eye.  This project would build upon my discovery (Dyer 1985b) that Apis dorsata can forage (and dance) by moonlight as well as by daylight.  Such visual flexibility is unknown in the other Apis species and is exceedingly rare in insects generally (insects are usually either diurnal or nocturnal, but not both).  By studying correlations between behavior and eye structure in a phylogenetic context,  I hope to understand how a common visual ground plan can (or cannot) be modified in a group of closely related insects.

Go to:


Comparative Ecology:  Foraging, Energetics, and Pollination Biology

When I originally learned about the Asian honey bees, I became intrigued by the differences among the species in worker body size (the largest is 5 times the mass of the smallest), in colony size (ranging from 7,000 to 40,000) and in nest architecture (some species nest in enclosed cavities as does the familiar European honey bee Apis mellifera; others nest in the open on combs protected by a blanket of worker bees).  I expected these differences to have major implications for the energetic economy of workers and colonies, and perhaps to shed light on the evolutionary forces that shaped the differences.

During my post-doctoral work with Thomas Seeley (Dyer and Seeley 1987, 1991a,b), I carried out field studies in Thailand comparing worker physiology and colony organization in three Asian honey bee species with the western honey bee Apis mellifera.  We had assumed that the scaling of physiological traits (e.g., metabolic rate, body temperature, flight speed, flight range) with size would impose important constraints on the performance of workers, and hence on the acquisition and use of resources by colonies.   We found instead that the scaling relationships themselves diverge among different species in a way that is not correlated with size.  Two of the species have relatively high-powered, high-tempo workers, and two have relatively low-powered, low-tempo workers.  Thus, normally robust scaling relationships appear to have been modified within the genus Apis, or at least have not imposed a strict constraint on worker physiology.  Furthermore, we found that worker energetic performance correlates with nesting behavior:  the high-tempo bees nest in enclosed cavities and the low-tempo bees nest in the open.  We proposed that honey bees as a group have evolved two divergent adaptive syndromes associated with the differing demands that open-nesting and cavity-nesting place on colony demography.  These divergent syndromes and their functional significance would have remained hidden without both a comparative perspective and careful studies of worker biology.

A graduate student, Puja Batra, has begun to move beyond studies of the bees themselves to consider their impact on the plants that they pollinate.  According to a study by David Roubik of the Smithsonian Tropical Research Institute, the 3-4 species of honey bees found in any given locality probably have a disproportionate effect on the transfer of pollen.  This may be because of traits that result from the extraordinarily efficiency with which the dance language allows foragers to recruit their nestmates to food:  very large flight ranges, large colony sizes, high densities of colonies.  Conceivably, A. dorsata's unusual ability to forage during both night and day further enhances its effect on plant reproduction.  Puja studied the implications of these behavioral abilities for forest conservation.  On the one hand, some plant populations in Asia may be more dependent on this one small but dominant group of insect pollinators than is commonly the case in neotropical forests, and hence more vulnerable should the bee populations decline through disease or human exploitation.  On the other hand, the large foraging ranges of honey bee colonies, combined with the flexibility of their nesting behavior, may mitigate the effects of forest fragmentation on plant reproduction, at least in comparison with the neotropics, where the limited flight range and narrow nesting preferences of many pollinators may reproductively doom trees in isolated forest patches.

Go to: