Toxic but drank: gustatory aversive compounds induce post-ingestional malaise in harnessed honeybees

Abstract

Background: Deterrent substances produced by plants are relevant due to their potential toxicity. The fact that most of these substances have an unpalatable taste for humans and other mammals contrasts with the fact that honeybees do not reject them in the range of concentrations in which these compounds are present in flower nectars. Here we asked whether honeybees detect and ingest deterrent substances and whether these substances are really toxic to them.

Results: We show that pairing aversive substances with an odor retards learning of this odor when it is subsequently paired with sucrose. Harnessed honeybees in the laboratory ingest without reluctance a considerable volume (20 µl) of various aversive substances, even if some of them induce significant post-ingestional mortality. These substances do not seem, therefore, to be unpalatable to harnessed bees but induce a malaise-like state that in some cases results in death. Consistently with this finding, bees learning that one odor is associated with sugar, and experiencing in a subsequent phase that the sugar was paired with 20 µl of an aversive substance (devaluation phase), respond less than control bees to the odor and the sugar. Such stimulus devaluation can be accounted for by the malaise-like state induced by the aversive substances.

Conclusion: Our results indicate that substances that taste bitter to humans as well as concentrated saline solutions base their aversive effect on the physiological consequences that their ingestion generates in harnessed bees rather than on an unpalatable taste. This conclusion is only valid for harnessed bees in the laboratory as freely-moving bees might react differently to aversive compounds could actively reject aversive substances. Our results open a new possibility to study conditioned taste aversion based on post-ingestional malaise and thus broaden the spectrum of aversive learning protocols available in honeybees.

Figure 1. Effect of pre-exposure to aversive substances on olfactory appetitive learning in harnessed honeybees.

The graph shows the performance (percentage of proboscis extension responses or PER) of honeybees during four trials of appetitive olfactory conditioning in which the odor 1-nonanol was paired with sucrose 1 M. Prior to this conditioning phase, bees were pre-exposed to 1-nonanol^a paired either with a mechanosensory stimulus (n = 45), distilled water (n = 42), NaCl 3 M (n = 49), salicine 100 mM (n = 42) or quinine 100 mM (n = 47). The untreated^b group (n = 54) was not pre-exposed. Bees having experienced NaCl, salicine and quinine showed lower acquisition than the other groups (water, mechanosensory and untreated)^c. No significant differences in acquisition were found between bees of the untreated, mechanosensory and water group. Different letters indicate significant between-group differences.

Figure 2. Kaplan–Meier curves of survival for harnessed honeybees following feeding of aversive compounds.

(a) First series. The probability of survival differed significantly between groups (long-rank test: χ² = 64.07, df:3, p<0.0001). The group of honeybees having ingested NaCl 3 M (n = 30) and quinine 100 mM (n = 30) exhibited a significant decrease of their survival probability compared to the distilled water group (n = 30). The group having ingested salicine 100 mM (n = 30) had intermediate mortality levels and comparison to the distilled water group, which exhibited a low decrease of the probability of survival, was marginally non-significant (Z = 1.78, p = 0.07). (b) Second series. The probability of survival differed significantly between groups (long-rank test: χ² = 108.93, df:8, p<0.0001). The group of bees having ingested sucrose 1 M group (n = 30) did not exhibit any variation of their probability of survival over time. The quinine 100 mM group (n = 30) experienced higher mortality than the distilled water group (n = 60). The quinine 10 mM (n = 60) and LiCl 140 mM (n = 60) groups experienced also induced higher mortality than the distilled water group. The amygdaline 1 mM group (n = 30) exhibited inetrmediate mortality compared to the the distilled water group. Mortality in the L-canavanine 40 mM (n = 30) and 100 mM (n = 30) groups was not significantly different from that of the distilled water group. The probability of survival from the groups having ingested mixtures of quinine 10 mM and sucrose 1 M (n = 30) and LiCL 140 mM and sucrose 1 M (n = 30) did not differ from that of the distilled water group.

Figure 3. Ranking of sugar solutions by harnessed bees.

The graph shows the percentage of proboscis extension responses (PER) upon antennal stimulation with fructose 1,66 M, glucose 1,66 M and sucrose 1 M. Each sugar was assayed with a different group of bees experiencing also a control stimulation with distilled water control (n = 30 each). Bees responded significantly more to the sugar than to the water. The preference ranking was fructose < glucose < sucrose. Different letters indicate significant between-group differences.

Figure 4. Devaluation of fructose 1.66 M.

The graph shows the performance (percentage of proboscis extension responses or PER) during (a) an odor-fructose association in which the response to the odor (conditioned stimulus or CS) was quantified, and during (b,c) a test phase following a devaluation phase in which responses to the sugar (US; see b) and to the odor (CS see c) were quantified in paired and unpaired groups of bees experiencing or not an association between sugar and either distilled water, quinine 10 mM, LiCl 140 mM or amygdaline 1 mM (319 bees in total). (a) All bees learned the odor-fructose association. The graph shows the pooled acquisition performance of all eight groups of bees. (b) Ingestion of quinine, LiCl and amygdaline decreased US responsiveness with respect to a water control. Responses of paired and unpaired groups were similar. (c) Ingestion of quinine, LiCl and amygdaline decreased CS responsiveness with respect of a water control. Responses to a novel odor remained low and equivalent in all groups. Different letters indicate significant between-group differences.

Figure 5. Devaluation of glucose 1.66 M.

The graph shows the performance (percentage of proboscis extension responses or PER) during (a) an odor-glucose association in which the response to the odor (conditioned stimulus or CS) was quantified, and during (b,c) a test phase following a devaluation phase in which responses to the sugar (US; see b) and to the odor (CS see c) were quantified in paired and unpaired groups of bees experiencing or not an association between sugar and either distilled water, quinine 10 mM, LiCl 140 mM or amygdaline 1 mM (319 bees in total). (a) All bees learned the odor-glucose association. The graph shows the pooled acquisition performance of all eight groups of bees. (b) Ingestion of quinine, LiCl and amygdaline decreased US responsiveness with respect to a water control. Responses of paired and unpaired groups were similar. (c) Ingestion of quinine, LiCl and amygdaline decreased CS responsiveness with respect of a water control. Responses to a novel odor remained low and equivalent in all groups. Different letters indicate significant between-group differences.

Figure 6. Devaluation of sucrose 1 M.

The graph shows the performance (percentage of proboscis extension responses or PER) during (a) an odor-sucrose association in which the response to the odor (conditioned stimulus or CS) was quantified, and during (b,c) a test phase following a devaluation phase in which responses to the sugar (US; see b) and to the odor (CS see c) were quantified in paired and unpaired groups of bees experiencing or not an association between sugar and either distilled water, quinine 10 mM, LiCl 140 mM or amygdaline 1 mM (319 bees in total). (a) All bees learned the odor-fructose association. The graph shows the pooled acquisition performance of all eight groups of bees. (b) Ingestion of quinine, LiCl and amygdaline did not diminish US responsiveness with respect to a water control. Bees responded maximally to sucrose (100% PER). Responses of paired and unpaired groups were similar. (c) Ingestion of quinine, LiCl and amygdaline decreased CS responsiveness with respect of a water control. Responses to a novel odor remained low and equivalent in all groups. Different letters indicate significant between-group differences.