Print PageThe colors of flowers are often attributed to interactions with animal pollinators. There are, however, few experimental demonstrations that pollinators cause fitness differences between flowers of different colors. The New Zealand alpine flora offers an extreme test case for the role of pollinators in selecting for flower color, as it includes an unusual preponderance (70%) of white flowers, traditionally ascribed to absence of social bees. We asked whether flower visitors show preferences, and generate natural selection, based on colors other than those common in the New Zealand alpine but found in congeners. To do so, we phenotypically manipulated the white to pale blue W. albomarginata to match the bright blue W. violacea found in lowland New Zealand. We manipulated petal color and examined visitation by syrphid and tachinid flies and solitary bees to small-scale (4x4 flowers, half blue / half white) and large-scale arrays (8x8 array of one color with a 4x4 array nested in the larger patch). Dyes were used to estimate pollen export in these arrays. Furthermore, we examined the effects of supplemental pollination and artificially reduced pollination on seed set of blue and white flowers. Pollinator effectiveness was assessed through single visit pollen deposition to stigmas.
Results/Conclusions
The colletid bee Hylaeus matamoko visited W. albomarginata flowers at the same rate regardless of whether they were painted white or blue. Syrphid and tachinid flies had preferences for white over blue (mean proportion of visits to blue = 0.29 and 0.21, both P < 0.05). Petal color did not influence pollen export, suggesting that white flowers do not have an advantage in male pollination success. The similarity in pollen export from the two colors of flowers despite a preference of flies for white may be explained by the relatively low pollinator effectiveness of flies in comparison with the Hylaeus bees, which deposited more than twice as many germinating pollen grains per single visit. Furthermore, pollen limitation of seed production did not depend on flower color, suggesting that the paler color of W.albomarginata in comparison with its congeners is not maintained strictly by pollinator preference for white in the alpine. Total seed set was also independent of flower color. Overall, we found no evidence that the pale coloration of W. albomarginata in the New Zealand alpine is selectively maintained by pollinator preference.
