The evolutionary genetics of highly divergent alleles of the mimicry locus in Papilio dardanus

Abstract

Background: The phylogenetic history of genes underlying phenotypic diversity can offer insight into the evolutionary origin of adaptive traits. This is especially true where single genes have large phenotypic effects, for example in determining polymorphic mimicry in butterflies. Here, we characterise the evolutionary history of two candidate genes for the mimicry switch in the polymorphic Batesian mimic Papilio dardanus coding for the transcription factors engrailed and invected.

Results: We show that phased haplotypes associated with the dominant morphs f. poultoni and f. planemoides are phylogenetically highly divergent, in particular at non-synonymous sites. Some non-synonymous changes are shared between the divergent alleles suggesting either convergence or a shared ancestry. Gene trees for invected do not show this pattern. Despite their great divergence, all engrailed alleles of P. dardanus were monophyletic with respect to alleles of closely related species. Phylogenetic analyses therefore reveal no evidence for introgression from other species. A McDonald-Kreitman test conducted on a population sample from South Africa confirms a significant excess of intraspecific non-synonymous diversity in P. dardanus engrailed, suggesting long-term balanced polymorphism at this locus.

Conclusions: The divergence between engrailed haplotypes suggests an evolutionary history distorted by selection with multiple changes reflecting recurrent selective sweeps. The high level of intraspecific polymorphism observed is characteristic of balancing selection on this locus, as expected if the gene engrailed is under phenotypic selection for the maintenance of multiple mimetic morphs. Non-synonymous changes in key functional portions of a major transcription factor are likely to be deleterious but if maintained in a dominant allele at low frequency, heterozygosity would reduce the associated genetic load.

Figure 1

Phenotypic variation in P. dardanus and the organisation of the H locus. (A) Female forms of P. dardanus (below the horizontal line) and examples of their model species in a number of African mimicry rings. The P. dardanus male pattern is shown boxed at the left along with male-like female forms. (B) Genomic organisation of the H locus showing the extent of exons (vertical blue bars) and introns (blue line) of the en and inv genes along the known sequence of the region. The bottom panels show the significance level for associations of SNP in the respective exons with three dominant morphs. The shading indicates the significance threshold for association. Note that significant association is found only in en (mostly exon 1). The figure is reproduced in an altered form from [25].

Figure 2

Sampling locations of P. dardanus used in this study. Dotted gray lines separate the Eastern, Western and Indian Ocean lineages. The shaded area is P. dardanus meseres, believed to be a contact zone between the Eastern and Western mainland lineages.

Figure 3

ML phylogeny of 5’ engrailed across the genus Papilio , with a Battus outgroup. Papilio dardanus individuals are represented by haplotypes from either cloning or MiSeq amplicon sequencing. Blue dots indicate nodes with bootstrap support >70%. Bootstraps at intraspecific nodes are ignored.

Figure 4

Maximum likelihood tree with 200 bootstrap replicates for Papilio 5’ invected . Papilio dardanus individuals are represented by haplotypes from MiSeq amplicon sequencing. Note that P. demodocus and P. memnon are paraphyletic. Blue dots indicate nodes with bootstrap support >70%. Bootstrap support at intraspecific nodes is not presented.

Figure 5

Maximum likelihood phylogeny of inferred 712 bp haplotypes of engrailed in P. dardanus , labelled by either the morph (A) or the population (B) the haplotypes were sampled from. Divergent alleles are associated with the dominant forms f. poultoni and f. planemoides, whilst there is no geographic structure. Percentage bootstrap support values above 50% are labelled in red.

Figure 6

Phylogeny of inferred haplotypes for P. dardanus invected , labelled by either the morph (A) or the population (B) the haplotypes were sampled from. Percentage bootstrap support values >50% are labelled in red.

Figure 7

The structure of genetic variation in invected and engrailed. Gene trees were established by maximum likelihood for invected (A) and engrailed (B) haplotypes, separately from non-synonymous and synonymous nucleotide changes. Branch lengths correspond to the number of changes. The area of each pie is scaled in proportion to the frequency of that haplotype among the morphs sampled, and colours of pies reflect proportion of haplotypes recovered from each morph, labelled as in Figure 4. (C) Unique SNPs in the two morph-associated alleles relative to the consensus sequence for all other inferred P. dardanus alleles.