Discussion
Species delimitations that are informed by genetic data have undergone a renaissance as increasingly more sophisticated sequencing technology and analytical methods have become available. But genes are only part of what makes a species. Species delimitations informed by ecological characteristics – the phenotypic interactions of individuals with their environment – have been difficult to incorporate quantitatively and are usually relegated to natural history accounts. Our study uses a notoriously confusing species group of butterflies to demonstrate an analytical integration of genetic and ecological species concepts. This approach supports species delimitations that should ultimately be more stable and meaningful in conservation contexts and citizen science.
Evidence for historical introgression between S. hesperisand S. zerene
Our genomic analyses suggest a complex, shared evolutionary history between S. atlantis, S. hesperis, and S. zerene . BFD* species delimitation supported separation of northern and southernS. hesperis lineages as distinct species, and a sister relationship between S. atlantis and northern S. hesperis , consistent with our SNP phylogeny (Fig. 1a). However, the relationship of S. zerene to the other clades varied. This is interesting, as recent molecular phylogenetic work on Speyeria generally indicates a non-sister relationship between S. zerene andS. hesperis /S. atlantis (de Moya et al. 2017; Campbell et al. 2017, 2019; Thompson et al. 2019). It is likely that the recovered polyphyly of S. hesperis and the sister relationship between southern S. hesperis and S. zerenepresented here is partly due to the omission of other Speyeriaspecies in our phylogenetic analyses, but also due to probable introgression between S. hesperis and S. zerene, which is more explicitly indicated by SNP-based admixture and COIhaplotype analyses (Fig. 2, Fig. 3, Table S2).
While the results of our SNP-based analyses and the lack of sequence variation in the mitochondrial haplotype shared between S. zereneand S. hesperis support a hypothesis of introgression between these non-sister species, they do not sufficiently clarify the direction of gene flow between them; TreeMix and f 3 tests indicated introgression from S. zerene into S. hesperis , however our inference that the S. hesperis haplotype occurs throughout the entire S. hesperis sampled range, but only in the Nevadan part of the range of S. zerene, suggests that it originated in S. hesperis . Expanded sampling of both species is needed to clarify the pervasiveness of this haplotype in S. zerene and to validate its origin.
In contrast to mtDNA, nuclear SNPs across the range of S. hesperis do not show the same obvious reduction in genetic variability, suggesting that a strong selective sweep leading to a severe bottleneck event has recently caused the loss of other variable mitochondrial haplotypes (Sonsthagen et al. 2017; Hurst & Jiggins 2005). A candidate for facilitating such a process is Wolbachia Hertig & Wolbach, 1924, maternally-inherited, endosymbiotic bacteria that can facilitate the spread of particular mitochondrial haplotypes throughout populations and species, reducing haplotype variation (Werren et al. 2008; Kodandaramaiah et al. 2013; Ahmed et al. 2015).Wolbachia infections have been reported in severalSpeyeria species, including S. zerene (McHugh et al. 2013), but not yet within the S. atlantis-hesperis complex (Hamm et al. 2014). Wolbachia infection via introgression offers a plausible and testable hypothesis that could explain the observed haplotype sharing between Nevadan S. zerene gunderi andS. hesperis in the absence of contemporary nuclear admixture in the sampled specimens, which may further clarify the historical relationship between these taxa.