UNLABELLED: • PREMISE OF THE STUDY: The processes maintaining flower color polymorphisms have long been of evolutionary interest. Mechanistic explanations include selection through pollinators, antagonists, local environments, drift, and pleiotropic effects. We examined the maintenance of inflorescence color polymorphisms in the genus Protea (Proteaceae) of South Africa, in which ∼40% of species contain different color morphs. • METHODS: We studied 10 populations of four bird-pollinated Protea species and compared adult performance, floral and leaf morphology, vegetative pigmentation, germination, and seedling survival between co-occurring pink and white morphs. We also tested for differences in pollination success and pre-dispersal seed predation. • KEY RESULTS: White morphs produced seeds 10% heavier and 3.5 times more likely to germinate, which all else being equal, should fuel positive selection on white. In one studied population per species, however, white morphs were more susceptible to seed predation by endophagous larvae. Pollinators had no morph-specific effects on female fecundity, as measured by amount or probability of seed set. Differences in stem color indicated that white morphs produced smaller quantities of pigment and associated compounds throughout, possibly explaining their higher seed palatability. • CONCLUSIONS: Our findings suggest a mechanism for some white protea polymorphisms: deleterious pleiotropic effects on pink morphs are occasionally offset by reduced losses to seed-eating larvae. Because trends were repeated across species, we suggest that similar processes may also occur in other proteas, placing a new emphasis on seed predators for influencing some of South Africa's amazing floral diversity.