Müllerian mimicry strongly exemplifies the power of natural selection. However, the exact measure of such adaptive phenotypic convergence and the possible causes of its imperfection often remain unidentified. Here, we first quantify wing colour pattern differences in the forewing region of 14 co-mimetic colour pattern morphs of the butterfly species Heliconius erato and Heliconius melpomene and measure the extent to which mimicking colour pattern morphs are not perfectly identical. Next, using gene-editing CRISPR/Cas9 KO experiments of the gene WntA , which has been mapped to colour pattern diversity in these butterflies, we explore the exact areas of the wings in which WntA affects colour pattern formation differently in H. erato and H. melpomene. We find that, while the relative size of the forewing pattern is generally nearly identical between co-mimics, the CRISPR/Cas9 KO results highlight divergent boundaries in the wing that prevent the co-mimics from achieving perfect mimicry. We suggest that this mismatch may be explained by divergence in the gene regulatory network that defines wing colour patterning in both species, thus constraining morphological evolution even between closely related species.

Müllerian mimicry strongly exemplifies the power of natural selection. However, the exact measure of such adaptive phenotypic convergence and the possible causes of its imperfection often remain unidentified. The butterfly species Heliconius erato and Heliconius melpomene have a large diversity of co-mimicking geographic races with remarkable resemblance in melanic patterning across the midforewing that has been linked to expression patterns of the gene WntA. Recent CRISPR/Cas9 experiments have informed us on the exact areas of the wings in which WntA affects color pattern formation in both H. erato and H. melpomene, thus providing a unique comparative dataset to explore the functioning of a gene and its potential effect on phenotypic evolution. We therefore quantified wing color pattern differences in the mid-forewing region of 14 co-mimetic races of H. erato and H. melpomene and measured the extent to which mimicking races are not perfectly identical. While the relative size of the mid-forewing pattern is generally nearly identical, our results highlight the areas of the wing that prevent these species from achieving perfect mimicry and demonstrate that this mismatch can be largely explained by constraints imposed by divergence in the gene regulatory network that define wing color patterning. Divergence in the developmental architecture of a trait can thus constrain morphological evolution even between relatively closely related species.

In the original version of figure 2, panel (c) previously displayed a H. e. demophoon WntA CRISPR KO phenotype where a H. m. rosina wild-type should have been. This has now been corrected.

No abstract