Hybridization among S. corvina lineages
Two out of the three contact zones showed extensive hybridization. However, contact zones are formed by later-generation hybrids and backcrossed individuals, with no F1 hybrids, and the contact zones are wide suggesting that pure parental subspecies are not currently directly interbreeding. Classic tension zones are usually narrow and are maintained by low hybrid fitness and dispersion from parental populations (Barton & Hewitt, 1985). Our results suggest that these contact zones are hybrid zones (sensu stricto Barton & Hewitt, 1985), a “fairly continuous transitions between distinct forms”, in which hybrids between S. corvina subspecies likely present relative fitness comparable to the parental subspecies but seem to be restricted to the contact zone perhaps due to environmental constraints and relatively low dispersal. For the third contact zone, which occurs in the Central Valley of Costa Rica where S. c. corvina andS. c. hoffmanni come into contact, we found no hybrids. The bimodal distribution of hybrid indices suggests low ancient hybridization with extensive backcrosses, where the introgressed genomic variation is highly diluted into the genetic pool of the local population. Given the extensive hybridization at the other two contact zones described above, it is very unlikely that postzygotic genetic incompatibilities (e.g., hybrid inviability), and low hybrid fitness (e.g., hybrid disadvantage) play an important role in reducing hybridization in this system. Instead, we hypothesize that premating reproductive isolation is the main driver maintaining isolation betweenS. c. corvina and S. c. hoffmanni subspecies in Costa Rica.
Reproductive isolation in secondary contact can be attained by conspecific recognition resulting in assortative mating. Given that we found the two genetically distinct populations (S. c. corvina andS. c. hoffmanni ) within the same locality, we predict that assortative mating due to behavioral isolation will likely be found to play a major role in reproductive isolation. However, contrary to our expectations, we found two different outcomes in two contact zones with similar patterns of plumage divergence (i.e., black vs. pied plumage), suggesting that factors other than just plumage color may also shape the different patterns of hybridization. For instance, different patterns of local adaptation may result in asynchronous breeding phenology (Hau et al., 2008). For example, S. c. corvina at the Caribbean side of Costa Rica apparently have longer breeding seasons (almost year-round; pers. obs.), related with the lack of marked dry and rainy seasons, butS. c. hoffmanni from the Pacific side of Costa Rica appear to breed only two or three months after the beginning of the marked rainy season (Skutch, 1954). Additionally, divergence in male song characteristics is an important source of isolation in tropical birds (Freeman & Montgomery, 2007), and in this case both subspecies (S. c. corvina and S. c. hoffmanni ) have different male songs (unpub. data), which may favor assortative mating.
Interestingly, where subspecies hybridize, plumage traits showed a sharper transition and reduced cline width compared to the hybrid index clines (Figure 5; Table 2). The more abrupt change in phenotype than in hybrid index at the contact zones suggests these plumage traits are under divergent selection, causing limited introgression. In general, our results suggest that distinct plumage traits are under selection (likely as signals for conspecific mate recognition), but that they do not act as a strong barrier to gene flow by themselves. On the other hand, the striking intrapopulation variation in morphometric traits conceals a clear sigmoidal transition among contact zones, confounding the interpretation of the cline’s centers and widths (Figure 5). However, the displaced clines of the beak size and tail length across the corvina-hicksii contact zone (Figure 5B) could result from directional selection favoring an asymmetrical introgression of some loci associated with selected phenotypes (Lipshutz et al., 2019; Stein & Uy, 2006; While et al., 2015).
Our models for demographic inference suggested rapid divergence paired with recent gene flow between lineages, but not contemporary gene flow between S. c. corvina and S. c. hoffmanni. This result suggests that the black S. c. corvina may have evolved in isolation from the pied subspecies and they subsequently established secondary contact and gene flow with S. c. hicksii. This is consistent with the hypothesis of the range of S. c. corvinaexpanding into the distribution of S. c. hicksii, mediated by periods of population expansions and contractions due to climatic fluctuations (Stiles, 1996). If the contact is not linked to an environmental gradient or ecotone, differential introgression may result in movement of the contact zone (Buggs, 2007). However, the contact zone that occurs between S. c. corvina and S. c. hicksii in Central Panama has been stable, at least for more than 100 years (Olson, 1981).