Colour preference and constancy in the giant Asian honey bee Apis dorsata
Sudeep, R.; Bhaskar, S.; Somanathan, H.
Show abstract
Tropical pollinators forage in environments where floral resources vary in space and time, requiring flexible strategies to optimise foraging efficiency. One such strategy, floral constancy - the temporary restriction to a single flower type - strongly influences foraging success and plant-pollinator interactions. We aimed to: (1) quantify spontaneous colour preferences and constancy in the Asian giant honeybee Apis dorsata, (2) test whether reward concentration modulates these preferences, (3) evaluate how quickly learned associations override spontaneous biases, (4) determine whether bees can use multiple colour associations simultaneously, and (5) assess whether local floral spectral patterns correlate with bee preferences. Bees trained to a neutral UV-grey stimulus showed a strong spontaneous preference and high constancy to blue, revealing a robust short-wavelength bias. Crucially, the strength of this spontaneous bias depended on reward concentration; Low-reward conditions elicited strong blue constancy, whereas high-reward conditions weakened it, demonstrating that reward expectation shapes spontaneous colour choices. This bias was flexible. When bees learned that yellow was rewarding they switched their preferences. Bees sequentially trained to both colours visited blue and yellow, showing no overall bias, or effect of the last-trained colour, indicating that recent experiences disrupt colour-specific constancy and generate largely random foraging choices. Bees were capable of learning and retaining two colours simultaneously, effectively suppressing the influence of spontaneous preferences. Finally, analysis of the communitys floral spectral distribution revealed a strong dominance of short-wavelength flowers, suggesting that long-term selection by the local floral environment may underlie the spontaneous blue preference observed in A. dorsata.
Matching journals
The top 4 journals account for 50% of the predicted probability mass.