A field test for frequency-dependent selection on mimetic colour patterns inHeliconiusbutterflies

ABSTRACT Müllerian mimicry, the similarity among unpalatable species, is thought to evolve by frequency-dependent selection. Accordingly, phenotypes that become established in an area are positively selected because predators have learnt to avoid these forms, while introduced phenotypes are eliminated because predators have not yet learnt to associate these other forms with unprofitability. We tested this prediction in two areas where different colour morphs of the mimetic species Heliconius erato and H. melpomene have become established, as well as in the hybrid zone between these morphs. In each area we tested for selection on three colour patterns: the two parental and the most common hybrid. We recorded bird predation on butterfly models with paper wings, matching the appearance of each morph to bird vision, and plasticine bodies. We did not detect differences in survival between colour morphs, but all morphs were more highly attacked in the hybrid zone. This finding is consistent with recent evidence from controlled experiments with captive birds, which suggest that the effectiveness of warning signals decreases when a large signal diversity is available to predators. This is likely to occur in the hybrid zone where over twenty hybrid phenotypes coexist.