Olfactory learning in Pieris brassicae butterflies is dependent on the intensity of a plant-derived oviposition cue

Abstract

Butterflies, like many insects, use gustatory and olfactory cues innately to assess the suitability of an oviposition site and are able to associate colours and leaf shapes with an oviposition reward. Studies on other insects have demonstrated that the quality of the reward is a crucial factor in forming associative memory. We set out to investigate whether the large cabbage white Pieris brassicae (Linnaeus) has the ability to associate an oviposition experience with a neutral olfactory cue. In addition, we tested whether the strength of this association is dependent on the gustatory response to the glucosinolate sinigrin, which is a known oviposition stimulus for P. brassicae. Female butterflies were able to associate a neutral odour with an oviposition experience after a single oviposition experience, both in a greenhouse and in a semi-natural outdoor setting. Moreover, butterflies performed best when trained with concentrations of sinigrin that showed the strongest response by specific gustatory neurons on the forelegs. Our study provides novel insight into the role of both gustatory and olfactory cues during oviposition learning in lepidopterans and contributes to a better understanding of how these insects might be able to adapt to a rapidly changing environment.

Keywords: Pieris brassicae; glucosinolates; lepidoptera; olfactory conditioning; oviposition stimulus.

Figure 1.

The effect of training with different concentrations of an unconditioned oviposition stimulus (sinigrin) on olfactory learning of mated Pieris brassicae females in an indoor setting. (a) Preference index for the time spent on vanilla-scented discs for the four treatments. (b) Preference index for the visits to vanilla-scented discs for the four treatments. (c,d,e,f) Comparison of the percentage of ovipositing animals for the four treatments. Generalized linear models followed by Tukey-corrected estimated marginal means tests were used to compare treatments in (a) and (b). Letters indicate significant differences (p < 0.05). Preference indices of the different treatments were tested against zero using a Wilcoxon signed-rank test. Filled box plots indicate preference indices significantly different from zero (p < 0.05). Fisher’s exact test was used to compare differences in the oviposition events between scented and unscented discs within each treatment in (c,d,e,f).

Figure 2.

The effect of training with different concentrations of an unconditioned oviposition stimulus (sinigrin), on olfactory learning of mated Pieris brassicae females in an outdoor setting. (a) Preference index for the time spent on vanilla-scented discs for the four treatments. (b) Preference index for the visits to vanilla-scented discs for the four treatments. (c,d,e,f) Comparison of the percentage of ovipositing animals for the four treatments. Generalized linear models followed by Tukey-corrected estimated marginal means tests were used to compare treatments in (a) and (b). Preference indices of the different treatments were tested against zero using a Wilcoxon signed-rank test. Filled box plots indicate preference indices significantly different from zero (p < 0.05). Fisher’s exact test was used to compare differences in the oviposition events between scented and unscented discs within each treatment in (c,d,e,f).

Figure 3.

Tarsal taste sensilla and spike classification. (a) Fifth tarsomere of the foreleg of a Pieris brassicae female. Small arrowheads indicate the two medial sensillum clusters, while the big arrowhead indicates the visible lateral sensillum cluster. Due to view perspective, the other lateral sensillum cluster is not visible. (b) Tip of a sensillum in the medial cluster showing a single tip pore (arrow) typical for gustatory sensilla. (c) A recording from a sensillum in the medial cluster of the fifth tarsomere stimulated with a 10 mM sinigrin solution. Three different classes of spikes can be distinguished by amplitude. Scale bars for (a) and (b) can be found in the images.

Figure 4.

Action potential activity recorded during the first second after stimulation from tarsal taste neurons in sensilla of the medial and lateral clusters to control solution (1 mM KCl; ‘0’) and different concentrations of sinigrin. (a) Median firing rate (spikes s⁻¹) with an amplitude between 0.5 and 1 mV of neurons in the medial sensilla cluster. (b) Median firing rate with an amplitude between 0.5 and 1 mV of neurons in the lateral sensilla cluster. (c) Median firing rate with an amplitude higher than 1 mV of neurons in the medial sensilla cluster. (d) Median firing rate with an amplitude higher than 1 mV of neurons in the lateral sensilla cluster. Different letters indicate significant differences (p < 0.05) in (a,c) and (d) determined by a mixed-linear model followed by a Tukey-corrected estimated marginal means test. Within one animal (n = 16 for medial sensilla and n = 15 for lateral sensilla) all concentrations were tested on the same medial and lateral sensillum. For each animal sensilla were chosen randomly from the two clusters.