Does facilitating pollinator learning impede deceptive orchid attractiveness? A multi-approach test of avoidance learning.

It has often been proposed that nectarless deceptive orchid species exploit naïve pollinators in search of food before they learn to avoid their flowers, and that intraspecific floral trait polymorphism, often noted in this plant group, could prolong the time needed for learning, thus increasing orchid reproductive success. We tested the importance of avoidance learning in a European deceptive orchid, Anacamptis morio, which has been reported to have a highly variable fragrance bouquet among individuals. We used an indirect approach, i.e. we facilitated pollinators' ability to learn to avoid A. morio by adding anisaldehyde to selected inflorescences, a scent compound that is easily perceived by the natural pollinators and produced in large quantities by the closely related, nectar producing Anacamptis coriophora, a species that shares pollinator species with A. morio. In a series of three experiments (in artificial arrays, in natural populations and in bumblebee behavioural observations), we consistently found no difference either of reproductive success of or visitation rates to scent-added versus control inflorescences. We also found that the decrease of reproductive success over time in artificial populations of this deceptive species was not as important as expected. Together, these data suggest that pollinators do not fully learn to avoid deceptive inflorescences, and that pollinator avoidance behaviour alone may explain the lower reproductive success usually found in deceptive orchids. We discuss the possible explanations for this pattern in deceptive orchids, particularly in relation to pollinator cognition and learning abilities. Lastly, in light of our results, the potential for higher average reproductive success in deceptive orchids with high phenotypic variability driven by avoidance learning thus appears to be challenged.