Is it a mottled duck? The key is in the feathers

Genomic and morphological data shed light on the complexities of shared ancestry between closely related duck species

Causes for genomic and morphological similarities among recently radiated species are often multifaceted and are further convoluted among species that readily interbreed. Here, we couple genomic and morphological trait comparisons to test the extent that ancestry and gene flow explain the retention of mallard-like traits within a sister species, the Mexican duck. First, we confirm that these taxa remain genetically structured, and that Mexican ducks exhibit an isolation-by-distance pattern. Despite the assumption of wide-spread hybridization, we found only a few late-stage hybrids, all from the southwestern USA. Next, assessing 23 morphological traits, we developed a genetically-vetted morphological key that is > 97% accurate in distinguishing across sex-age cohorts of Mexican ducks, mallards, and hybrids. During key development, we determined that 25% of genetically pure, immature male Mexican ducks of the northern population naturally displayed mallard-like traits in their formative plumage. In fact, applying this key to 55 museum specimens, we identified that only four of the 14 specimens originally classified as phenotypic hybrids were truly hybrids. We discuss how genomic and morphological comparisons shed light into the mechanism(s) underlying the evolution of complex phenotypic traits in recent radiations, and how misunderstanding the true morphological diversity within Mexican ducks resulted in taxonomic revisions that hindered conservation efforts.

Identifying hybrids & the genomics of hybridization: Mallards & American black ducks of Eastern North America

Resolving evolutionary relationships and establishing population structure depends on molecular diagnosability that is often limited for closely related taxa. Here, we use 3,200 ddRAD-seq loci across 290 mallards, American black ducks, and putative hybrids to establish population structure and estimate hybridization rates. We test between traditional assignment probability and accumulated recombination events based analyses to assign hybrids to generational classes. For hybrid identification, we report the distribution of recombination events complements ADMIXTURE simulation by extending resolution past F4 hybrid status; however, caution against hybrid assignment based on accumulated recombination events due to an inability to resolve F1 hybrids. Nevertheless, both analyses suggest that there are relatively few backcrossed stages before a lineage's hybrid ancestry is lost and the offspring are effectively parental again. We conclude that despite high rates of observed interspecific hybridization between mallards and black ducks in the middle part of the 20th century, our results do not support the predicted hybrid swarm. Conversely, we report that mallard samples genetically assigned to western and non-western clusters. We indicate that these non-western mallards likely originated from game-farm stock, suggesting landscape level gene flow between domestic and wild conspecifics.