Sex chromosome mosaicism and hybrid speciation among tiger swallowtail butterflies

Abstract

Hybrid speciation, or the formation of a daughter species due to interbreeding between two parental species, is a potentially important means of diversification, because it generates new forms from existing variation. However, factors responsible for the origin and maintenance of hybrid species are largely unknown. Here we show that the North American butterfly Papilio appalachiensis is a hybrid species, with genomic admixture from Papilio glaucus and Papilio canadensis. Papilio appalachiensis has a mosaic phenotype, which is hypothesized to be the result of combining sex-linked traits from P. glaucus and P. canadensis. We show that P. appalachiensis' Z-linked genes associated with a cooler thermal habitat were inherited from P. canadensis, whereas its W-linked mimicry and mitochondrial DNA were inherited from P. glaucus. Furthermore, genome-wide AFLP markers showed nearly equal contributions from each parental species in the origin of P. appalachiensis, indicating that it formed from a burst of hybridization between the parental species, with little subsequent backcrossing. However, analyses of genetic differentiation, clustering, and polymorphism based on molecular data also showed that P. appalachiensis is genetically distinct from both parental species. Population genetic simulations revealed P. appalachiensis to be much younger than the parental species, with unidirectional gene flow from P. glaucus and P. canadensis into P. appalachiensis. Finally, phylogenetic analyses, combined with ancestral state reconstruction, showed that the two traits that define P. appalachiensis' mosaic phenotype, obligatory pupal diapause and mimicry, evolved uniquely in P. canadensis and P. glaucus, respectively, and were then recombined through hybridization to form P. appalachiensis. These results suggest that natural selection and sex-linked traits may have played an important role in the origin and maintenance of P. appalachiensis as a hybrid species. In particular, ecological barriers associated with a steep thermal cline appear to maintain the distinct, mosaic genome of P. appalachiensis despite contact and occasional hybridization with both parental species.

Figure 1. The distributional ranges and hybrid zones of tiger swallowtails, and the hybrid phenotype of Papilio appalachiensis.

(A) Papilio appalachiensis is endemic to mid- and high elevations in the Appalachian Mountains and sympatric with glaucus throughout its range, but presumably parapatric with canadensis in its northernmost range , (see Materials and Methods). Also shown is the range of Battus philenor, Batesian model for the mimetic glaucus, appalachiensis and garcia melanic female forms. (B) Ecological and morphological differentiation between glaucus and canadensis, and their admixture in appalachiensis – (also see Figure S1).

Figure 2. Genotypic differentiation between glaucus and canadensis, and the mismatch in mitochondrial and Z-linked genes in appalachiensis.

(A) appalachiensis genotypes at loci that were significantly different (p<0.001) between glaucus and canadensis, as judged by F_ST values from a locus-by-locus AMOVA comparing glaucus and canadensis. Genotypes are nucleotide bases at specific SNP or indel polymorphisms, which can be diploid (Z-linked polymorphisms scored in males) or haploid (mtDNA, and Z-linked polymorphisms scored in females). Color code: purple: genotypes characteristic of glaucus; light blue: genotypes characteristic of canadensis; black: heterozygotes; grey: missing data; orange: late flight canadensis. (B) Species pair-wise F_ST values for the mitochondrial and Z-linked genes (see Table S4 for individual values for each gene and species pair-wise comparisons).

Figure 3. Genomic admixture in appalachiensis showing its hybrid origin and its contrast with laboratory-generated hybrids and late flight canadensis.

(A) Population clustering of AFLP data in STRUCTURE under the assumption of two, three and four populations, comparing appalachiensis with laboratory-generated glaucus x canadensis hybrids. (B) appalachiensis AFLP allele frequencies with respect to glaucus and canadensis, based on species pair-wise locus-by-locus AMOVAs. Allele frequencies of “glaucus-like” AFLPs were significantly different from canadensis, “canadensis-like” AFLPs were significantly different from glaucus, “intermediate” were intermediate between glaucus and canadensis but significantly different from neither, and “different” were significantly different from both glaucus and canadensis. (C) Population clustering in STRUCTURE under the assumption of four populations, showing genomic similarity between the laboratory-generated hybrids and late flight canadensis, and distinctiveness of appalachiensis (also see Figure S2). For (A) and (C), admixture proportions of the sampled individuals, rather than their assignment probabilities, are shown.

Figure 4. Phylogenetic relationships and character evolution among tiger swallowtails.

(A) AFLP-based neighbor-joining tree, with percentage bootstrap support shown for branches. The ten appalachiensis and two canadensis samples that cluster outside their species are marked with asterisks. (B) Character evolution based on the AFLP phylogeny.

Figure 5. Estimated divergence times between the parental glaucus and canadensis and the hybrid appalachiensis.

Dates of divergence estimated by IMa2 are: (a) appalachiensis and glaucus: ca 100,000 years ago, (b) appalachiensis and canadensis: ca 90,000 years ago, (c) glaucus and canadensis: ca 580,000 years ago.

Figure 6. Estimated gene flow among appalachiensis, glaucus, and canadensis.

Gene flow was estimated as the population migration rate or 2Nm, which is equivalent to the historical average number of immigrants between species per generation: glaucus to appalachiensis: 2.3; canadensis to appalachiensis: 1.8; appalachiensis to either glaucus or canadensis: 0; glaucus to canadensis: 0.1; canadensis to glaucus: 0.