In this study, we observed odor-specific antennal movements to a range of pheromones and general odorants with different biological values for honey bees. Bees’ responses recorded from two colonies, on two different years, were correlated, showing that antennal responses to odorants are reproducible. Generally, odorant-induced changes in antennal position and velocity were correlated, so that the more the antennae were brought to the front, the more quickly they moved. Building an original olfactory orientation setup, we observed clear differences in tested odorants’ attractiveness to the bees. However, odorants’ attractiveness, as measured in our setup, did not correlate significantly with either antennal position or velocity measures. Lastly, we show that the antennal responses of newly-emerged bees are limited compared to older bees. While the tested odorants induced an acceleration of antenna movements like in older bees, they did not produce any change in antenna position.
We included in our experiments a few odorants that had been tested in previous recordings of antenna movements (50, 51). In Erber et al. (1993) (50), bees’ antenna movements were recorded thanks to two photodiodes, each one located in front of one of the bees’ antennae (at an approximately 45° angle in our coordinate system, see Fig. 1B therein). An ‘antennal response’ in this study corresponded to an increased frequency of antennal passages on the diodes during odor presentations. Bees ‘responded’ to geraniol and citral (aggregation pheromone components) and to octanoic acid (therein termed caprylic acid, a major royal jelly volatile) but not to isopentyl acetate (alarm pheromone). In this previous work, it was however not possible to know if bees kept their antennae more to the front and/or increased their antennal scanning velocity during a response. The use of a camera-based system in our study allowed disentangling these effects. Fitting with Erber et al. (1993) (50), octanoic acid produced in our recordings accelerated movements to the front. Citral and geraniol slightly increased antennal speed (about 1°/s), even if they produced contrasted changes in antenna position (Fig. 3A,B). Lastly, although isopentyl acetate induced an increase in antennal speed (Fig. 3B), it brought the antennae strongly to the back of the bees’ head (Fig. 3A), explaining the lack of response in Erber et al. (1993) (50).
Thanks to the use of a wide odorant panel, we showed that odorants induce diverse antennal responses, with both forward and backward movements, and both increased and decreased velocities. Interestingly, position and velocity changes in response to odorants were correlated (Fig. 3C) and bees’ antennal responses could roughly be separated in two groups: fast-forward movements and slow-backward movements. When taking into account the known biological value of these odorants for bees, interesting general tendencies emerged. While the slow-backward movements were mostly expressed in response to alarm/defense pheromones (see red dots in Fig. 3C), especially to 2-heptanone, fast-forward movements were rather elicited by food-related odors (octanoic acid, the royal jelly odor, blue, and octanal, grey), pheromone components linked to the signaling of valuable resources (geraniol, an aggregation pheromone component, green), as well as social signals like brood and queen pheromones (light and dark violet, Fig. 3C). There were some exceptions to these rules, like the recorded backward antennal response to citral (an aggregation pheromone component) or β-ocimene (a volatile brood pheromone compound). Similarly, some odorants with a strong inferred biological value, like the fecal compound 3-methyl indole (scatol), did not induce strong antennal responses. This being said, fast forward movements to food-related odors appear consistent with previous studies showing that sucrose, or odorants previously associated with sucrose, induce forward antenna movements (30, 50, 52–54). Such antennal responses are part of food-associated behavioral routines, together with extension of the proboscis. On the other hand, slow/backward antenna movements to alarm/defense compounds seems coherent with a defensive context, where responding to appetitive stimuli is of secondary importance, and protecting important sensory organs like the antennae may be more appropriate. In addition, the strong backward response to 2-heptanone somehow fits with its use by bees as a deterrent to mark depleted flowers (55, 56).
Clearly odorant quantity had an effect on antennal responses, because antennal position (but not velocity) varied as a function of odorant’s vapor pressure. When testing three odorants at eight different concentrations we found that both position and velocity varied with odorant concentration. Antennal responses generally started at 10−3 concentrations, which corresponds to concentrations at which clear odor-induced neural activity is observed in the antennal lobe in optical imaging experiments (57–59). Interestingly, antennal responses did not simply increase monotonously with concentration. Remarkably, bees’ antennal responses to geraniol were stronger at medium than at high concentration. This possibly relates to the known dose-dependent effects of pheromones on behavior(60–63) and the fact that in natural situations, pheromones are used within a definite concentration range. It is thus possible that given concentrations best evoke odorants’ pheromonal value for bees and therefore trigger stronger antennal responses than higher concentrations. In any case, odorant concentration affected the amplitude of the response, but not its direction. We did not observe any opposite responses (forward/backward or slower/faster) for the same odorant at different concentrations.
Opposite influences of pheromones with differing biological values on antenna movements as observed in our experiments are remarkable in the context of current debates on pheromones’ behavioral side-effects. It has been increasingly suggested that pheromones, in insects but also in mammals, can act as modulators of a variety of behavioral responses which are not the primary – known - targets of their action (64, 65). In honey bees, some alarm pheromone components decrease bees’ responsiveness to an appetitive reward like sucrose (65) and negatively impact appetitive learning performances (63, 66–68). Conversely, an aggregation pheromone component, typically associated with appetitive behavior, has been shown to decrease responsiveness to a noxious stimulus like an electric shock (69) and to improve appetitive learning (68). The model extracted from these findings posits that pheromones – or odorants with a strongly innately attached value – modulate the bees’ internal state relative to two main modules, an appetitive module and an aversive module ( ‘defensive and appetitive scores’ (69)). It classifies pheromones in clear categories along a common hedonic dimension, with alarm pheromones bearing a negative (‘defensive’) value and aggregation pheromones (or floral odorants) bearing a positive (‘appetitive’) value. Accordingly, alarm pheromones reduce the appetitive score and aggregation pheromones reduce the defensive score. The contrasted antennal responses we observed may represent behavioral clues for the existence of such opposite odorant values. We attempted to capture such a hedonic dimension in our odorants by measuring their attractiveness for bees in an olfactory orientation setup (Fig. 4). However, correlation coefficients between attractiveness indices and antenna movement variables were not significant (p = 0.12 for both angle and velocity), even if the figures suggest a possible trend, with attractive odorants corresponding more to fast and forward antenna movements. Possibly, including more odorants in future studies could provide more statistical power for demonstrating a link between both variables. Note however, that the real-life situation is more complex than the simple hedonic model presented above, since not all pheromonal components of a given type have the same effect on behavioral responses. For instance, 2-heptanone, but not isopentyl acetate affects responsiveness to sucrose (65), whereas isopentyl acetate, but not 2-heptanone, affects responsiveness to an electric shock (69). Likewise, in our data, the brood pheromone β-ocimene brought the antennae to the back, while the other brood pheromones (ethyl oleate and methyl linoleate) brought them to the front (Fig. 3A). Thus, bees’ evaluation of odorants may be best described on more than one simple dimension, as each conveys a different message, usually presented in a particular context.
We wondered if odor-induced antennal responses are hardwired, a product of bees’ evolutionary history, or may be acquired during the bees’ lifetime. To approach this question, we evaluated the reproducibility of antennal responses to odorants by measuring them on bees from different hives on different years. We found a clear correlation between the two years’ datasets, both in terms of antenna position and velocity. This result may indicate that some of these responses (in particular to pheromones) are innate. Our observations suggest however a strong importance of bees’ experience. While some odorants induced very similar responses on both years, others induced remarkably different behaviors. Octanoic acid, for instance, produced strong forward and accelerated movements on the first year, but only weak responses on the second year. This points to an effect of experience, which we demonstrated in a previous study where odorants associated with food suddenly induced fast forward antennal movements (30). In fact, finding the same pattern of odor-specific responses on the two years is not a proof per se that antennal responses are innate, because the observed patterns could simply be the result of our odorants being associated with similar contexts and consequences during the lives of these two groups of bees. To understand the ontogeny of these odor-induced responses, we recorded antenna movements in newly emerged bees. We found rather limited antennal responses at this age, especially regarding angular position changes. This suggests that odor-specific antenna movements are acquired by the bees in the course of their adult life. This could be due to an incomplete maturation of their antennal motor abilities when emerging, but could also relate to their behavioral development in the context of bees’ age polyethism. Interestingly, newly emerged bees did not show the specific slow-backward movements to alarm pheromones found in older bees. This is consistent with the fact that bees’ aggressiveness and response to alarm pheromones increases with age (70), paralleling the ontogeny of defensive behavior (71, 72). This behavioral development is accompanied by changes in biogenic amine and hormone titers (73–75). For instance, the levels of juvenile hormone and octopamine increase with age (76–78). A role of biogenic amines in particular is supported by the observation that octopamine and serotonine have opposite effects on antennal movements, increasing and decreasing them respectively (54). Thus, part of the age effect we found could be related to differences in levels of these biogenic amines.
Different odorants induce different antenna movements according to their biological value for bees. This, together with the observation that antennal movements are modified by associative conditioning (30), suggests that antennal movements are under central top-down modulation. The movements of the antennae are controlled by different muscle groups moving the antenna scape (4 muscles) and the flagellum (2 muscles). Motor neurons controlling this muscular system originate in the AMMC (antenna mechanosensory and motor center) (79–82). Response to tactile stimuli is thought to use a short route as mechanosensory neurons project directly to the AMMC (79). Antennal reaction to olfactory stimuli, by contrast, should take a longer route. Odorants are detected by olfactory sensory neurons in the antenna, which relay odor information to a primary olfactory centre, the antennal lobe (AL), composed of glomeruli, which each receives input from OSNs expressing the same olfactory receptor type. The AL processes olfactory information and second-order (projection) neurons (PN) transmits it to higher-order brain centers, the mushroom bodies (MB) and the lateral horn (LH). Honey bee pheromone compounds (alarm, aggregation, brood, queen, etc.) all trigger combinatorial activity from many glomeruli in the worker antennal lobe (83, 84). This suggests that their biological value is extracted within higher-order centers (85). Indeed, projections from the AL to the honey bee LH were shown to contain combinatorial information allowing to differentiate the different pheromone types (86). The LH is considered as a premotor center mediating fast and innate reactions to biologically relevant stimuli, and may be responsible for innate antennal movements to odorants. Direct connections between the LH and the AMMC are not described yet in honey bees, but they are known in fruit flies (87, 88). Changes in odor-induced antennal responses through experience (like after associative conditioning) would involve the MB, the learning and memory center of the insect brain. In the MB, the Kenyon cells (KC) are highly odor-specific and are activated by the combinatorial input from many different PNs (89). Information from KCs is read out by MB-output neurons which project to different parts of the protocerebrum, including the LH. MB-output neurons are plastic and their odor-induced responses are modified by experience (90–93). Some of them, like the PE1 neuron, project to the LH (91, 94) and may be responsible for an experience-driven modulation of odor-induced antennal movements through this structure. To our knowledge, no direct connections of MB output neurons to the AMMC have been described (94), but indirect pathways other than through the LH are possible.
To conclude, honey bees display a range of different antennal responses to odorants, which vary as a function of odorants’ biological value. These responses are reproducible, suggesting that they are in part innate, but they are also shaped by the bees’ experience and develop in the course of their lives. A necessity of our approach, but a clear limitation nonetheless, is that bees’ responses were recorded in an experimental context quite different from natural situations. In other insects, social context in particular has been shown to modulate behavioral responses to olfactory stimuli (95, 96). Bees’ antennal response to alarm pheromones, for instance, may be quite different when they are guarding at the hive entrance. A next step should thus be to analyze antennal movements in more natural, hive, situations.