Evidence of niche divergence
Comparison of ENMs indicated that the S. atlantis, northernS. hesperis , and southern S. hesperis evolutionary lineages, delineated on the basis of genomic data, significantly differ in their habitat associations and ecological niches. The relative magnitude of the “within- vs. between-lineage” coefficient estimate may be interpreted not only as a measure of the relative strength of niche divergence, but also as an indicator of whether or not genomic divergences likely reflect ecological speciation events. Comparison of these coefficient estimates across models suggest that niche divergence was greatest between S. atlantis vs. southernS. hesperis , followed by northern S. hesperis vs. southernS. hesperis and S. atlantis vs. northern S. hesperis , which were approximately equivalent in their strength of pairwise niche divergence. This ordering corresponds to that of genomic divergences estimated between S. atlantis and northern and southern S. hesperis (Fig. 1, Fig. 2, Fig. 3). Although degree of divergences between pairs of lineages appears to approximately correspond to degree of geographic separation, null ENMs built from random localities indicate that observed niche divergences cannot be attributed to differences in available habitat between lineages (spurious niche divergence). Furthermore, although northern and southernS. hesperis are largely allopatric in their distributions, our resistance-based analyses did not detect any significant barriers to dispersal between them, conferring further support to the hypothesis that ecologically-based divergent selection has been the principal process contributing to diversification and reinforcement of the lineages’ genomic integrity, rather than speciation without selection or mutation-order speciation associated with barriers to dispersal (i.e., vicariance). We therefore suggest that present-day parapatry between northern and southern S. hesperis is a consequence, rather than cause, of ecologically-based divergent selection, as projections of suitable habitat across the entire study landscape demonstrate clear geographically-structured divergence in habitat associations between the lineages (Fig. 4).
Assessment of species limits in S. hesperis and S. atlantis
Our results consistently show that the northern S. hesperislineage is more genetically similar to S. atlantis than to any genetic clusters within the southern S. hesperis lineage, but we did not recover any evidence of admixture between S. atlantis and northern S. hesperis, despite sampling both species in sympatric portions of their ranges (mostly in Alberta, Canada); these results are consistent with other genetic surveys (Thompson et al. 2019; Campbell et al. 2019). Past work has noted differences in micro-habitat preference between S. atlantis and S. hesperis in regions where they co-occur (Bird et al. 1995; Guppy & Shepard 2001; Dunford 2009; Riva et al. 2019), and the results of our ENMs indicate that divergences in habitat associations are likely significant enough to limit gene flow between species, even in sympatric regions. Our results also indicate relatively little mitochondrial and SNP differentiation between the subspecies S. atlantis hollandi, S. a. sorocko, and S. a. canadensis relative to the other taxa in this study. Thus, S. atlantis is an independently evolving lineage distinct from S. hesperis , and recognition of the species status of S. atlantis should be maintained.
BFD* species delimitation using SNPs indicated clear support for splitting northern and southern S. hesperis into distinct species (Fig. 3a). Both Structure and TESS failed to recover any substructure in northern S. hesperis despite broad geographic sampling, and TreeMix and f 3 analyses indicated probable gene flow between northern S. hesperis, southern S. hesperis, and S. zerene, particularly in or near the central population of southernS. hesperis (Fig. 2, Table S2). This is also the approximate boundary between the sampled northern and southern S. hesperislineages, so it is possible that northern S. hesperis expanded poleward from this geographic region into newly available habitat following the last glacial retreat, and that these lineages may have initially continued to exchange genes with each other and S. zerene . If this hypothesis is accurate, then the lack of recovered population structure in northern S. hesperis may be partly reflective of this recent range expansion, as well as due to few barriers to dispersal and gene flow in the relatively homogenous prairie/open forest habitat of northern S. hesperis compared to the more disjunct habitat of southern S. hesperis in the southern Rocky Mountains. The haplotype sharing between S. zerene and northern and southern S. hesperis further suggests that introgression between these species occurred prior to this range expansion, and emphasizes that COI is only partially informative for clarifying species boundaries in this complex. ENMs similarly suggest that ecologically-based divergent selection is likely a principal process underlying divergence and reinforcement of the northern and southern lineages, even in the absence of barriers to dispersal.
Taxonomic revision to S. hesperis
Our results unambiguously support recent evolutionary and ecological divergences between northern and southern lineages of S. hesperis , but also indicate that these lineages currently exist as distinct entities shaped by interactions with other species and the environment that nonetheless maintain their genomic integrity. We therefore recommend that the two lineages should be recognized as distinct species. Following taxonomic priority (Pelham 2019), the northern lineage should continue to be referred to as S. hesperis , and the oldest southern subspecies name sampled in this study, S. h. nausicaa , should be elevated to species level,S. nausicaa , to represent the entire southern lineage. This taxonomic revision remains a hypothesis that requires further sampling and genomic assessment, particularly in California and the Great Basin, where lineages that diverged before the Pleistocene may persist. We also detected introgression between S. zerene and S. hesperis/S. nausicaa that may be a source of nuclear gene tree-species tree discordance. This should be pursued due to the conservation status of several S. zerene subspecies.