Bumble-bee learning selects for both early and long flowering in food-deceptive plants.

Most rewardless orchids engage in generalized food-deception, exhibiting floral traits typical of rewarding species and exploiting the instinctive foraging of pollinators. Generalized food-deceptive (GFD) orchids compete poorly with rewarding species for pollinator services, which may be overcome by flowering early in the growing season when relatively more pollinators are naive and fewer competing plant species are flowering, and/or flowering for extended periods to enhance the chance of pollinator visits. We tested these hypotheses by manipulating flowering time and duration in a natural population of Calypso bulbosa and quantifying pollinator visitation based on pollen removal. Both early and long flowering increased bumble-bee visitation compared with late and brief flowering, respectively. To identify the cause of reduced visitation during late flowering, we tested whether negative experience with C. bulbosa (avoidance learning) and positive experience with a rewarding species, Arctostaphylos uva-ursi, (associative learning) by captive bumble-bees could reduce C. bulbosa's competitiveness. Avoidance learning explained the higher visitation of early- compared with late-flowering C. bulbosa. The resulting pollinator-mediated selection for early flowering may commonly affect GFD orchids, explaining their tendency to flower earlier than rewarding orchids. For dissimilar deceptive and rewarding sympatric species, associative learning may additionally favour early flowering by GFD species.