Discussion
Our results show that attraction of great tits to herbivore-induced defence cues is not innate. Naïve birds with no experience with green leaves were not attracted to induced, thus potentially insect rich, saplings. This confirms the previous findings of Amo et al.(2016). In their experiment, naïve birds neither preferred the caterpillar-infested sapling for the first visit, nor visited that sapling more frequently that the uninfested sapling (control). In contrast, in another study, when the great tits gained some experience with foraging for caterpillars, they exhibited a preference for caterpillar infested saplings, and visited them more often than the uninfested saplings (Amoet al. 2013a). In line with these results, we were able to teach the birds the association between the herbivore-induced volatile compounds of saplings and a food reward.
In contrast to our expectation, the great tits were able to associate the volatile compounds of tropical guava with food similarly well (if not better) as of the volatile compounds of elm, despite the fact that they did not share any evolutionary history with this plant species. After the successful teaching period, each group of trained bird preferred the induced version of the sapling of the plant species they were trained to, over any other version of the saplings offered to them. The birds did not seem to be able to generalize the learned bouquets of volatile compounds to another plant species. This might imply that the birds are likely to learn the whole bouquets of the volatile compounds and associate them with food, rather than learning individual volatile compounds. Alternative explanation might be that the birds are capable of some generalization for more closely related plant species producing more similar plant volatile compounds.
In our experimental plants, we detected significantly more chemical compounds in induced than in noninduced guavas. The volume of produced compounds was also higher in induced than in noninduced elms, but the difference was not significant. Compounds found in significantly higher concentration in induced plants were β-Ocimene, Heptadiene, Cyclohexane, α-Pinene, Copaene, Caryophyllene, Tetraline, α-Farensene and Tetraline. Many of these compounds were previously suspected to correlate with attractivity of the herbivore-damaged plants with bird’s attractivity to them (Mrazova et al. 2019). Specifically, β-Ocimene and α-Pinene were discussed as compounds specifically responsible for the attractivity of birds to herbivore-damaged plants (Mrazova et al.2019). Our experiment showed that birds exposed to two plants with very different bouquets can distinguish them better than birds exposed to more similar bouquets. In our experiment, we selected two plant species which are very different. One tropical plant with strong odour and one temperate with rather weak odour. Our birds were not able to generalize between their odours. The questions remain whether the birds would be capable of generalization between several temperate plant species which might have more similar odours.
In contrast to previous study (Amo et al. 2016), our birds took rather long to successfully search for the mealworms offered to them on induced saplings. The previous study indicated a fast ability of birds to use HIPVs as a foraging cue, as they learned to associate the presence of caterpillars with the HIPVs of infested trees in 5 h (Amo et al.2016). Similarly, in our previous study, naïve birds were able to locate more than 80% of caterpillars after 5 h of habituation in aviary with noninduced trees (Mrazova et al. unpublished). In the current study, the average bird needed at least 20 hours to be able to associate the food rich sapling with the volatile compounds. The difference between current and previous training was in the type of the sapling offered in training phase (two infested saplings in Amo et al. 2016 or two uninfested saplings in Mrazova et al. unpublished vs. one infested and one uninfested sapling in current study) and the training considered to be successful (i.e. eating from any sapling vs. eating from the induced sapling only).
A prenatal chemosensory learning was demonstrated in birds (Sneddon et al. 1998) previously but it a very different experiment with domestic chickens. With no exposure to strawberry before hatching, strawberry was highly aversive to chicks after hatching. However, following exposure to strawberry before hatching, chicks expressed a greater preference for (or weaker aversion to) the strawberry stimulus (Sneddon et al. 1998). Despite no association of smell and food was tested in this study, it proved that chicks can learn and prefer a smell. In a similar study, nest recognitions via olfactory imprinting was learned in few hours after hatching, and olfactory stimuli is then a sufficient mechanism for nest recognition (Caspers et al. 2013).
For birds, as generalist predators, prey availability changes during the year due to differences in phenology of the plant and insect species. Therefore, adaptive plasticity may be an advantage for foraging of the birds in response to changes in distribution and abundance of prey species (e.g. in Murakami 1998). In the current study, we showed the existence of such plastic association of different odours with food resources may maximize success of foraging in birds. We observed that birds are able to learn even novel odours which they did not have chance to meet during their evolutionary history. Thus, they are capable of some generalization with respect to volatile compounds. However, it is important to note that our study was conducted on a single bird species but the adaptive value of learning is expected to vary among bird species depending on their diet breadth at both the herbivore and plant levels (Vet et al. 1995, Mrazova et al. 2019). The use of olfaction in foraging can be innate in different bird species (Bonadonna et al. 2006, Amo et al. 2013b).
In our study, we also used very phylogenetically distant tree species, which differed significantly in the diversity of volatile compounds. However, it is likely that some generalization may work across more similar and phylogenetically related tree species. While birds might be able to learn several plant species occurring in their home ranges in temperate regions, the tropical birds might be facing different challenges. Tropical insectivorous understory birds have usually home ranges up to 30 ha large. A hectare of tropical forest typically contains more than 250 tree species. It would be generally impossible for a bird to learn all species in a lifetime, so either individual birds respond to only subset of local plants, or birds learn groups of plants which are chemically and therefore also phylogenetically related). It remains unresolved how the learning of volatile compounds works in species rich tropical areas, and how is the ability to learn volatile compounds combined with ability to learn the visual signals.