Pheromone Communication in the Honeybee (Apis mellifera L.)
- First Online:
- Cite this article as:
- Slessor, K.N., Winston, M.L. & Le Conte, Y. J Chem Ecol (2005) 31: 2731. doi:10.1007/s10886-005-7623-9
- 1.6k Views
Recent studies have demonstrated a remarkable and unexpected complexity in social insect pheromone communication, particularly for honeybees (Apis mellifera L.). The intricate interactions characteristic of social insects demand a complex language, based on specialized chemical signals that provide a syntax that is deeper in complexity and richer in nuance than previously imagined. Here, we discuss this rapidly evolving field for honeybees, the only social insect for which any primer pheromones have been identified. Novel research has demonstrated the importance of complexity, synergy, context, and dose, mediated through spatial and temporal pheromone distribution, and has revealed an unprecedented wealth of identified semiochemicals and functions. These new results demand fresh terminology, and we propose adding “colony pheromone” and “passenger pheromone” to the current terms sociochemical, releaser, and primer pheromone to better encompass our growing understanding of chemical communication in social insects.
Key WordsApis melliferahoneybeesocial insectchemical communicationpheromonechemoecology
In their preface to Chemical Ecology of Insects published in 1984, Bell and Cardé stated “Evolution of sociality seemed to spawn a chemical language that is equivalent to the visual and auditory repertoire of higher vertebrates.” This is now revealed to be a prophetic insight (Bell and Cardé, 1984). At that time, the chemical communication systems utilized by social insects had been recognized as complex but seemed to conveniently compartmentalize into two pheromone categories, releaser or primer. A releaser pheromone initiates an immediate behavioral response upon reception, whereas a primer pheromone alters more long-term endocrine and reproductive systems in the recipient (Wilson and Bossert, 1963). Recent studies of honeybee pheromones suggest that chemical communication in social insects is deeper and richer than had been imagined by earlier generations of chemical ecologists. Pheromone signals in honeybees and likely other social insects are often enhanced by complexity, synergy, and the context in which they are deployed, mediated through both temporal and spatial distribution. Even our current limited view of social insect pheromones exceeds the boundaries of present terminology and requires a fresh appraisal. Our intention is not to review all the literature concerning honeybee pheromones as performed before (Free, 1987), but rather to suggest a new template with which to approach sociochemistry.
Our understanding of the chemical language expressed by social insects is best provided through the pharmacopoeia of Apis mellifera L., the honeybee. The social organization of honeybee colonies is determined primarily by chemical signals that are actively produced and transmitted by the queen, the adult workers at various tasks and lifestages, and the brood. The honeybee dance language used to recruit foragers to food, water, and nest sites is better known (von Frisch, 1967), but constitutes only one fraction of the communication systems that operate in a functioning colony.
These chemical signals are made remarkably specific in function by temporal fluctuations timed to release specific behaviors or to prime key physiological characteristics of signal recipients, based on both context and synergistic interactions between components. Like a fire alarm, a constant chemical signal would be of no value; time and place are of the essence. The wealth of identified semiochemicals utilized by honeybees and their functions is unprecedented in the insect literature, possibly as a result of the extensive investigations that this agriculturally beneficial insect has stimulated, but also because of its inherent social complexity. Even a cursory review of the honeybee's chemical communication repertoire reveals a rich language.
Defense: Alarm Pheromone
For the beekeeper, the banana smell of isopentyl acetate warns that the colony is aroused and that further stinging is imminent. A mixture of substances is released from a honeybee worker sting gland in the act of stinging, thus recruiting other workers (Breed et al., 2004). Free (1987) lists isopentyl acetate and 24 other substances that are found in the sting gland and thought to be involved in the alarm reaction to colony attack. More recently, Hunt et al., (2003) discovered 3-methyl-2-buten-1-yl acetate as a new alarm component in the sting apparatus of Africanized honeybees. The necessity for chemical complexity of the alarm signal is not obvious. Some specific behavioral attributes have been ascribed to individual components, but no clear theme emerges (Winston, 1987). Many of the components are chemically quite distinct, constituting both aromatic and aliphatic molecules, ruling out complexity as a byproduct of closely related biosynthetic pathways.
Attraction: Nasanov Pheromone
The incredible phenomenon of a cohesive honeybee swarm is partially mediated by a complex of the seven-component Nasanov pheromone (Pickett et al., 1980). The signal is released from the dorsal surface of the worker honeybee abdomen to attract her sister workers in a peaceful and organized manner during swarming or at colony entrances (Free, 1987; Winston, 1987). The lemon-grass fragrance we attribute to this bouquet is mostly because of our ready perception of three principal volatile components in this mixture: geraniol, geranial, and neral. The swarm cluster is initiated by alighting workers releasing Nasanov pheromone and then enhanced by the presence of the queen and her attractive signals (Avitabile et al., 1975). Many of the components of the Nasanov pheromone are biosynthetically related, so the complexity of this signal may be partially a function of their production.
Attraction to the Queen: Retinue Pheromone
A mated, laying queen is essential to the well being and proper functioning of the colony. She communicates her presence and manifests her influence by releasing a mixture of substances that is attractive to workers, enticing them to lick and antennate her to gather a small sample of her attractive blend. It is this retinue behavior of workers to the queen that was recognized by the earliest students of honeybee biology and distinguished her as “queen” of the colony. This “queen substance” was originally considered to be a single material, 9-oxo-(E)-2-decenoic acid (ODA) (Barbier and Lederer, 1960; Butler et al., 1961; Pain, 1961), and has been shown to be an essential component of the honeybee sex pheromone, attracting drones to the queen during her mating flight (Gary, 1962). However, queen-like amounts of ODA were neither attractive to workers nor substituted for her in her absence. Even addition of the two enantiomers of ODA's biosynthetic precursor, (R)- and (S)-9-hydroxy-(E)-2-decenoic acid (HDA), failed to constitute an attractive blend for workers, although (R)-HDA is involved in the swarm-settling queen signal (Slessor et al., 1988). Two further components, methyl p-hydroxybenzoate (HOB) and 4-hydroxy-3-methyoxyphenylethanol (HVA), were finally recognized and formulated with the decenoic acids to provide a source as attractive as an equivalent extract from the mandibular glands in which the five compounds are produced (Slessor et al., 1988).
Individual components and even subsets of components are not attractive. Only when all five components of the queen mandibular pheromone (QMP) are combined does the blend elicit the full retinue response, a striking example of semiochemical synergy (Slessor et al., 1988). Another important aspect of QMP that distinguishes it from many insect pheromones is that the components are on the borderline of volatility, apparently requiring close proximity between the queen and the recipient worker.
Early in our studies of QMP, it became evident that some strains of honeybees did not find this blend at all attractive, yet they tended their queen and otherwise functioned normally in colonies, implying as yet unidentified, additional substances involved in the retinue response. After 15 years, four further synergistic substances were identified in this “simple” retinue response (Keeling et al., 2003). Disruptive selection for high and low QMP retinue responding workers using closed mating demonstrated this pheromone response to have a strong genetic component (Pankiw et al., 2000). Two-way selection of colony phenotypes for low QMP and high queen extract responders provided over 380,000 worker test subjects for bioassays and led to the elucidation of the nine-component retinue pheromone (QRP = QMP + methyl oleate, coniferyl alcohol, palmityl alcohol, and linolenic acid). The terminology shift to QRP was necessary because the three new fatty-acid-derived constituents are not of mandibular gland origin (Keeling et al., 2003). The complete identity of QRP is still not fully defined because a very small difference in response of specially bred workers to synthetic QRP and multiple queen extract is evident. To identify the material(s) responsible for this small difference would necessitate a new and more extensive bee breeding and chemical isolation program.
The queen signals her presence to the colony through the attractive 9+ component releaser pheromone QRP. Absence of the signal for a few hours results in the workers choosing a few freshly laid eggs to rear into new queens, and long-term absence results in some workers developing ovaries (de Groot and Voogd, 1954; Butler and Fairey, 1963). Thus, colony reproduction and worker physiology are partially under the control of QRP (Pankiw et al., 1998), making it a primer as well as a releaser pheromone. The retinue facilitates the distribution of QMP that inhibits the development of worker ovaries (Hoover et al., 2003) and orients the comb building (Ledoux et al., 2001). It also facilitates distribution of a known primer component, ODA, that modulates the biosynthesis of juvenile hormone in recipient workers (Kaatz et al., 1992), regulating their age-related tasks (Robinson et al., 1989). Clearly, the synergistic releaser QRP embodies primer signals.
Larval Signals: Ester “Brood Pheromone”
In the early 1990s, we identified a mixture of 10 ethyl and methyl esters of the common fatty acids palmitic, linoleic, linolenic, stearic, and oleic acids from larvae, and 4 of these methyl esters were originally considered to be a signal from larvae to adults to cap the brood cells prior to their pupation (Le Conte et al., 1990). The methyl esters of palmitic and oleic acid were most effective in recruiting workers to this role. In addition, differences in the proportions of esters provide a chemical signature of larval age (Le Conte et al., 1994, 1994/1995). Further studies demonstrated that individual esters and subsets of esters provide context-specific signals to bring about other behaviors. In the presence of a fully functioning queen, the workers will tear down any freshly constructed queen cells, but the presence of methyl oleate in or about these cells results in a greater acceptance of new queen cells by workers (Le Conte et al., 1994/1995). The action of methyl linolenate results in enhanced provisioning of these new queen cells with more royal jelly, facilitating the healthy development of new queen larvae (Le Conte et al., 1995). Each individual ester or ester subset provides a message to the worker recipient, indicating an action to be carried out.
In the 1990s, an even more important role for the esters began to emerge. Nurse honeybees feed larvae a mixture of nutrients that contains a proteinaceous gland exudate. When nurse bees are treated with a blend of methyl palmitate and ethyl oleate, protein levels in the gland are elevated, probably through stimulation of the gland's biosynthetic capacity (Mohammedi et al., 1996). This finding has recently been confirmed (Pankiw et al., 2004). In the prolonged absence of a queen and larvae, the ovaries of worker honeybees mature and in extreme cases become drone-laying workers (Winston, 1987). Two of the “brood” esters with demonstrated releaser effects, ethyl palmitate and methyl linolenate, partially inhibit the ovarian development of worker bees when they are isolated from a queen and brood (Arnold et al., 1994; Mohammedi et al., 1998). The inhibitory effects of these esters on the physiological development of the recipient workers' ovaries constitute a primer pheromone effect.
The tasks of adult worker honeybees change as they age, from in-colony duties, such as nursing larvae, to pollen and nectar processing, followed by outside responsibilities of guarding and foraging (Lindauer, 1953). Plasticity of ontogeny in behavioral development allows the colony to cope effectively with changing environments, including internal requirements such as the production of new workers, or external opportunities such as increased nectar flow. This organizational shift and its plasticity are well recognized, but its underlying control has been a mystery, likely involving hormonal and pheromonal components (Robinson, 1992). A major influence in a worker's developmental rate is provided by her older sister foragers whose presence slows the nurse's progression through the age-related tasks (Robinson, 1992; Huang and Robinson, 1996). Recently, the feeding of larval esters to experimental colonies has demonstrated inhibition in age to first foraging (Le Conte et al., 2001). Ethyl oleate is present at significantly higher levels in foragers than in nurses, and the presence of this ester accounts for the ontogenetic inhibition and task mediation in honeybee colonies (Leoncini, 2002; Leoncini et al., 2004).
Chemical analysis of queens demonstrated the presence of seven esters including methyl oleate (Wossler and Crewe, 1999a), the palmitates, oleates, ethyl stearate, and two new esters, ethyl and methyl palmitoleate, with methyl oleate a synergistic component of queen retinue formation (Keeling, 2001). In the dispersal of QRP throughout the colony, queen esters are distributed as passive chemical passengers in the queen bouquet because they possess no attractive capability alone or in the QRP blend. For example, ethyl palmitate derived from the queen may well be the active agent in her ability to inhibit worker ovarian development as ethyl palmitate also was demonstrated to be produced by the larvae and to inhibit the ovary development of workers (Mohammedi et al., 1998). Thus, both the larvae and the queen manipulate the ovary development of workers to their advantage.
Multisources and Targets of Fatty Acid Esters in the Honey Bee
Crapping of the cell
Larval age recognition
Queen cell recognition
Royal jelly feeding
Acceptance of queen cells
Queen retinue behavior
Worker ovary development
Worker ovary development
Just as we had to revise the nomenclature of queen mandibular pheromone (QMP) to queen retinue pheromone (QRP) to properly reflect its multiglandular source, these fatty acid esters might now be properly viewed as “colony pheromones,” defined as originating in multiple castes and developmental stages and having diverse releaser and primer roles. Palmitate and oleate ethyl are good examples of a colony pheromone (Table 1). Remarkably, the queen and larvae work in concert to inhibit the ovarian development of young adult worker bees (Jay, 1968, 1972; Free, 1987). Ethyl palmitate inhibits ovary development of workers (Mohammedi et al., 1998). The flux of ethyl palmitate from the two sources is unknown at present, although drone larvae ready for capping have ∼90 ng (Le Conte et al., 1989), worker larvae somewhat less (Le Conte et al., 1990), and a mated laying queen contains ∼300 ng (Keeling, 2001). Older foragers, the larvae, and the queen exhibit primer effects by influencing the developmental rate of their younger sisters and offspring through the release of ethyl oleate. Foragers contain roughly 45 ng, larvae 50 ng (Le Conte et al., 1989; Leoncini et al., 2004), and queens 6300 ng (Keeling, 2001) of ethyl oleate on average, although release rates remain unknown. Moreover, oleate and palmitate ethyl are known to be involved in larval recognition by workers (Le Conte et al., 1994). In addition to influencing the ontogeny of their sisters, the esters stimulate and release pollen foraging behavior and may prime bees for a particular foraging role at a young age (Pankiw and Page, 2001).
Complexity: Why so many?
The identified queen, worker, and brood substances essential for the functioning of this highly evolved social organism now number nearly 50 in all, and still are incompletely characterized. In human society, language with its vast oral and written vocabulary is the commerce of communication. In honeybee society, the vocabulary is an array of specific chemicals or blends of chemicals. A few of the substances now thought to be honeybee signal chemicals eventually may be recognized as inactive biosynthetic precursors and degradation products, the noise of this chemical language. However, even with the noise filtered out, honeybee language is undoubtedly more complicated than previously recognized (Slessor, 2003). The chemical complexity of the queen's tergal (Wossler and Crewe, 1999a,b), labial (Katzav-Gozansky et al., 2001a), and Dufour's gland (Katzav-Gozansky et al., 1997, 2001b) implicates these and other worker, brood, and queen glands as informational sources. No components responsible for their effects, however, have yet been fully identified. Queen recognition, the ability of a colony to recognize its own queen (reviewed in Winston, 1987), worker policing of nonqueen laid eggs (Ratnieks and Visscher, 1989; Ratnieks, 1995), and colony/nestmate recognition (Breed et al., 1992, 1995) are examples of interactions that require further communication channels and suggest additional signal chemicals. Complex social interactions demand an intricate language. In the honeybee and likely in other social insects, the use of many highly specialized chemical signals provides that syntax.
Context, Complexity, and Synergy
The majority of known insect pheromones are perceived as volatile materials diluted and carried by air currents, received, and processed through antennal reception. Honeybees utilize this dispersal mechanism for alarm and Nasanov pheromones, and, similarly, components of queen pheromone attract drones for mating and workers to swarm clusters (see Winston, 1987). However, a distinguishing feature of QRP and colony esters is that they are passed primarily by contact. Partially volatile components appear to be quickly absorbed into the wax matrix of the colony (Naumann et al., 1991), effectively removing them and ensuring their context dependence. Alternatively, the transmission of chemical components from wax can provide discrimination and nestmate recognition cues to naive individuals (Breed et al., 1995, 1998), underlining the integral role that wax plays. In the retinue, the young workers attend the queen, licking and antennating her, removing QRP from her body, and often obtaining sufficient QRP that they become attractive to other workers after they have left the queen (Seeley, 1979; Naumann et al., 1991). These messenger bees contribute to the dispersal of QRP, but when the QRP signal fails to be transmitted throughout the colony because of congestion and population growth (Naumann et al., 1993), workers consider themselves queenless and begin to build queen cells, the precursor to colony reproduction by swarming (Winston et al., 1991). If these signals were volatile and/or not removed by wax, they would be distributed virtually homogeneously throughout the colony by the continuous circulation of air maintained by the workers. The relative involatility of QRP and the absorptive nature of wax, thus, have important repercussions enabling synergy for multicomponent pheromones. Inactive on their own, QRP components are highly synergistic, the whole blend being required for a complete signal. Only if all components are perceived as a unit does the signal transmit the important message “the queen is here and functioning appropriately.”
Synergy provides a further bonus for the organization because all chemical constituents need not be unique to the emitter. A single unique component is sufficient to render a distinctive identity and response when carried along with other ubiquitous and active ingredients. This phenomenon is particularly important where individuals are nearly all biosynthetically equivalent, as in a caste-based social organization (Plettner et al., 1996; Martin and Jones, 2004). In the honeybee colony, the queen must uniquely produce only HVA and coniferyl alcohol (and possibly HOB) to ensure that she transmits a blend with both releaser and multiprimer effects.
Context is less well documented and understood, but clearly plays an important role. Methyl oleate is perceived as a capping signal when emitted by larvae (Le Conte et al., 1990), but also is a synergistic component of QRP (Keeling et al., 2003; Table 1). Workers have reduced capacity to produce queen-like QRP components, and yet clearly respond to its primer effects. Young workers contain esters including ethyl oleate (Leoncini, 2002; Leoncini et al., 2004), which inhibits their ontogenetic development when received from the queen or their older sisters. Could it be that the receipt of primer pheromone through direct contact with antennae or other receptive sensory organs is the principal initiating mechanism, or is a threshold dosage the activating mechanism? At very dilute concentrations, QMP stimulates foragers to perceive a nectar source as a richer resource than the sugar concentration suggests (Higo et al., 1992). What underlying effects contribute to this contextual and dose misrepresentation of the queen signal?
Complexity of a signal is greatly facilitated by synergy, which, in turn, contributes to contextual uniqueness. These properties combined—complexity, synergy, context, and dose—appear to be hallmarks of social insect pheromones and distinguish them from the typical pheromones of nonsocial insects. Pankiw (2004) has recently described honeybee chemical communication as an emergent property of a complex system with dynamic properties requiring a complex systems approach for their analysis.
Isolation and Identification: Bioassays
Contextual dependence, synergy, and complexity make sociochemical identification extremely difficult. Chemical comparisons of interracial differences and similarities can provide clues to the involvement of specific substances of behavioral importance (Wossler and Crewe, 1999b; Brillet et al., 2002). However, no insect primer has yet been isolated and identified by bioassay. Substances have been designated with primer status only after their presence has been established through a releaser-based bioassay and then demonstrated in a primer bioassay. An effective bioassay must incorporate transfer of the bioactive mixture in an appropriate manner and dose and be followed by some quantitative measure of the resulting effect. The process must be elegant and simple because it may have to be replicated many thousands of times, not only to show the involvement of an active ingredient but also to provide satisfactory proof that an inactive ingredient has no effect on the response being measured; a benign effect is considerably more difficult to demonstrate than a positive one. Dose dependency must be tested to ensure that the applied dose is biologically relevant, similar to that encountered by an individual in a normal situation. Highly replicated bioassays may require hundreds of thousands of animals at a specific point in their life or of a particular strain. Primer bioassays provide an even greater challenge, necessitating a time period for the physiological change to become apparent and the time and effort involved in measurement of the change, e.g., dissection and evaluation of ovarian development. Efficient primer isolation and identification requires new methodology, with molecular biology the most promising tool. Initiation of physiological changes may be monitored at the RNA or protein level rather than the behavioral or physiological level. Differential or microarray displays (Whitfield et al., 2002; Grozinger et al., 2003) offer tremendous advantages over multitudinous bioassays necessary for the isolation and identification of primer pheromones. Nevertheless, the primer bioassay remains essential to prove the function of the putative primer pheromone rather than directing its isolation.
Sociochemicals: Refining Terms
The pheromone complexity of the honeybee suggests that the evolution of insect sociality created a large number of complex chemical languages. The term “sociochemical” refers to any chemical or defined mixture of chemicals that mediates social behavior (Bell and Cardé, 1984). Within this framework, honeybees utilize releasers, some apparently monofunctional, such as Nasanov pheromone, and others clearly multifunctional, such as “colony pheromone” (defined above as originating in multiple castes and developmental stages and having diverse releaser and primer roles, e.g., esters in honeybees). The latter, in addition to its releaser roles under appropriate context, functions in subsets as both stimulatory and inhibitory primers. Foragers modulate their sisters' physiological development with an inhibitory primer that may be transferred during nectar and pollen delivery.
The queen, by means of an attractive releaser (QRP), has at least one component (ODA) that mediates a worker's endocrine function. In addition, this attractive material carries with it several other passenger substances (ethyl oleate and palmitate) that are transferred passively during the retinue response. The ester blend released by larvae contains both context-dependent releasers and primers with some apparently in passenger mode. This new concept of “passenger pheromones,” substances that are transmitted in a pheromone complex, which are inactive in the transmission process, and yet which mediate some primer or releaser function upon reception, appears to be central to the functioning and understanding of the complex chemical language of social insects.
Extrapolation from the limited understanding of one species to the many social insect societies is fraught with uncertainty. What is clear after four decades of investigation is that (Bell and Cardé, 1984) were correct, and that chemical communication might equal or exceed the complexity of the better-known auditory and visual systems of vertebrates.
We appreciate the continued financial support of Natural Science and Engineering Research Council of Canada (K.S., M.W.) and of I.N.R.A., S.P.E. Department and Beekeeper associations in France (Y.L.C.). We acknowledge the hard work and insights of our students, employees, and colleagues in our pursuit to better understand these animals. We thank C. Brillet, H.A. Higo, C.I. Keeling, M. Ono, T. Pankiw, E. Plettner, and anonymous reviewers for their critical readings of this manuscript.