A major challenge to evolutionary developmental biology is to understand the how modifications to gene regulatory networks can lead to biological diversity. Heliconius butterfly wing patterns provide an excellent example of this diversity. In particular, the species H. melpomene and H. erato display wide variation in wing pattern across their ranges in Central and South America, but wherever they co-occur, they have converged on remarkably similar wing patterns due to Müllerian mimicry.

Linkage analysis of wing pattern genes has shown that in both species, there are three genomic loci that are responsible for most of the pattern variation, and that these loci are homologous. One locus, containing the transcription factor optix, is responsible for red pattern elements. A set of non-coding sequences linked to some of the red pattern elements have been identified. Another locus, containing the gene WntA, has been linked to the shape of the forewing band elements and is responsible for variation in wing pattern development in several species of lepidoptera. A third locus, responsible for yellow pattern elements, contains multiple candidate genes that may affect wing pattern development, including the gene cortex, which is also linked to the industrial melanism phenotype in the moth Biston betularia, as well as the genes domeless and washout, linked to the Bigeye mutant in Bicyclus anynana.

I first investigated modifications to regulatory sequence near the transcription factor optix, detecting a module associated with the band pattern element. I also found that for some pattern regulatory modules at optix, the same sequence has independently evolved the same function in H. melpomene and H. erato, in association with non-coding sequences conserved throughout the Lepidoptera.

I then investigated gene expression differences in two morphs from either side of a hybrid zone that vary only in the presence or absence of a yellow pattern element, in order to determine a role for candidate genes at the yellow pattern locus. In H. melpomene the gene cortex was upregulated in the larval wing discs of the black morph, whereas in H. erato it was upregulated in the larval wing discs of the yellow morph. In pupal wings, washout was differentially expressed, again in the opposite pattern in the two species, suggesting the same locus is responsible for convergent pattern modification, but by a different mechanism.

Finally, I investigated the spatial transcriptomic landscape across the wings of three different heliconiine butterflies. I identified candidate factors for regulating the expression of wing patterning genes, including genes with a conserved expression profile in all three species, and others, including genes in the Wnt pathway, with markedly different profiles in each of the three species.

Each of these studies contributes to our understanding of how gene regulatory networks can be modified to create diversity: first, at the level of cis-regulation, second at the level of gene interaction and expression, and lastly at the level of developmental bias and constraint.