FORAGING CHOICES OF BUMBLEBEES ON EQUALLY REWARDING ARTIFICIAL FLOWERS OF DIFFERENT COLORS

Laboratory-reared bumblebees were allowed to forage on 30 artificial f lowers which were identical in morphology and reward schedule, but wer e marked by either a human-blue, a human-green, or a human-white landi ng surface. The probability of nectar rewards in the artificial flower s, and their spatial distribution, were manipulated experimentally. Th e bees' color choices in the different experimental treatments were co mpared. The proportions of visits to the three colors deviated signifi cantly from the expected random choice (1/3,1/3,1/3) for mon than 50% of the bees. Of these bees, 38%, 32%, and 30% formed a preference for human-blue, human-green, and human-white, respectively. The frequency of nonrandom color choice, and the strength of the deviation from rand om choice, were highest when the different colors were placed in separ ate clusters, lower when they were placed in adjacent clusters, and lo west when they were randomly intermingled. Nonrandom color choice was also more pronounced when the bees were rewarded according to a consta nt schedule, rather than probabilistically. A statistically significan t preference for human-blue was found during the bees' first three vis its. The bees' tendency for ''runs'' of consecutive visits to the same flower color can partially account for their nonrandom color choices. Effects of innate preferences, early learning, generalization, and se arch-image formation on color choice are discussed.