Genetic structure among subspecies
In a previous study, we found evidence for reduced gene flow near the contact zone between the all-black S. c. corvina and the piedS. c. hicksii and S. c. hoffmanni subspecies (Ocampo et al., 2022a). Here, we have extended the genomic aspect of our previous study by increasing sample size and spatial resolution of this region and adding phenotypic data. We found an important barrier to gene flow that is partially concordant with the Talamanca mountain range and with the range limits between the black and pied subspecies (Figure 1). Consistent with the Talamancas acting as a geographic barrier, high genetic divergence between Caribbean and southern Pacific populations have been found in other lowland taxa in this region (e.g., in frogs [Crawford et al., 2003, Robertson & Zamudio, 2009], snakes [Zamudio & Greene, 1997], caimans [Venegas-Anaya et al., 2008], beetles [Kohlmann & Wilkinson, 2007], and birds [Hackett, 1996; Marks et al., 2002]). Moreover, the South Pacific region of Costa Rica is considered an important climate refuge (Haffer, 1974), harboring high biological richness and endemism in plants and animals (Crain & Fernandez, 2020; Kohlmann et al., 2010; Pareira & Barrantes, 2009).Sporophila corvina subspecies established secondary contact at both ends of the Talamanca mountain range (Figure 1A, no. 1 & 3), and we found significant gene flow between the Pacific and Caribbean populations at the Canal in central Panama (Figure 1B, blue shades). In contrast, there is reduced gene flow between S. c. corvina andS. c. hoffmanni in the Central Valley of Costa Rica. These observations suggest that levels of gene flow differed between the two contact zones despite similar degrees of phenotypic plumage divergence (black vs. pied plumage). Importantly, from the six individuals collected in Central Valley contact zone (San Ramón [SRA]; Table S1), two individuals showed the genetic makeup of the Pacific subspecies, while four individuals showed the genetic makeup of the Caribbean subspecies, which confirms that the two subspecies are in contact but are likely not hybridizing or doing so at levels we cannot detect in this study.
When determining genetic structure among individuals, our samples were best clustered into three genetic groups. Clusters are consistent, in general, with the subspecies assignments, suggesting that described subspecies correspond to recognizable lineages. However, we found evidence of admixture between clusters (Figure 1C), between pairs of clusters, and even a few individuals with signatures of ancestry from the three subspecies (Figure 3). We found that corvina is the most differentiated subspecies, with paired FSTvalues that are twice the value between hoffmanni andhicksii. This pattern is consistent with our expectations of stronger differentiation between phenotypically differentiated subspecies and with the hierarchical population structure identified previously (Ocampo et al., 2022a).