The perceived colour of an object depends on its spectral reflection and spectral composition of the illuminant. Upon illumination change, the light reflected from the object also varies. This results in a different colour sensation if no colour constancy mechanism is available to form consistent representations of colours across various illuminants. We explore various colour constancy mechanisms in an agent-based model of foraging bees selecting flower colour based on reward. The simulations are based on empirically determined spatial distributions of various flower species in different plant communities, their rewards and spectral reflectance properties. Simulated foraging bees memorise the colours of flowers experienced as being most rewarding, and their task is to discriminate against other flower colours with lower rewards, even in the face of changing illumination conditions. The experimental setup of the simulation of bees foraging under different photic environments reveals the performance of various colour constancy mechanisms as well as the selective pressures on flower colour as a result of changing light. We compared the performance of von Kries photoreceptor adaptation and various computational colour constancy models based on the retinex theory with (hypothetical) bees with perfect colour constancy, and with modelled bees with colour blindness. While each individual model generated moderate improvements over a colour-blind bee, the most powerful recovery of reflectance in the face of changing illumination was generated by computational mechanisms that increase perceptual distances between co-occurring colours in the scene. We verified the results of our model using various comparisons between modelled bees’ performance and that predicted by our models, as well as exploring the implications for flower colour distribution in a variety of representative habitats under realistic illumination conditions.