Dataset construction and phylogenetic analyses
After filtering, our SNP dataset contained 1026 SNPs (min. locus depth: 8, max. locus depth: 204, mean locus depth: 65.6). The COIdataset consisted of 648 sites (577 invariant and 71 variant), 58 of which were phylogenetically informative.
The SNP species tree shows S. cybele , S. atlantis , andS. zerene as monophyletic clades (Fig. 1a), but indicates a polyphyletic relationship for the northern and southern S. hesperis lineages. The northern lineage was additionally paraphyletic with S. atlantis, and contained subspecies S. h. beani(Barnes & Benjamin, 1926) and S. h. dennisi dos Passos & Grey, 1945 from Alberta, S. h. hutchinsi dos Passos & Grey, 1947 from Montana, S. h. brico Kondla, Scott & Spomer, 1998 and S. h. beani from British Columbia, S. h. lurana dos Passos & Grey, 1945 from South Dakota, S. h. ratonensis Scott, 1981 from southeastern Colorado, and S. h. irene (Boisduval, 1869) from California, but did not exhibit consistent geographic sub-clustering. The southern S. hesperis lineage was sister to S. zerene , and itself had two major geographic groupings. A large grade containedS. h. tetonia dos Passos & Grey, 1945 sampled in northern Utah and southern Montana, S. h. chitone (Edwards, 1879) sampled in southeastern Utah, S. h. electa (Edwards, 1878) sampled in southwestern Colorado, and S. h. viola dos Passos & Grey 1945 from Idaho. A monophyletic clade contained S. hesperis from New Mexico and Arizona, and broadly separated a southern New Mexican population of S. h. capitanensis Holland, 1988 sampled in the Sacramento Mountains from the more northern population of S. h. dorothea Moeck, 1947 sampled in the Sandia Mountains of New Mexico, which clustered with S. h. nausicaa (Edwards, 1874) from Arizona. Phylogenetic comparison of the SNP and COI datasets recovered extensive mito-nuclear discordance and a reduction in monophyly on theCOI tree (Fig. 1b), largely consistent with Campbell et al. (2019).