Delimiting Continuity: Comparison of Target Enrichment and ddRAD for Delineating Admixing Parapatric Melitaea Butterflies
Parapatrically distributed taxa pose a challenge for species delimitation due to the presence of gene flow and inherent arbitrariness of exactly defining the species boundaries in such systems. We tackled the problem of species delimitation in a parapatric species pair of Melitaea butterflies using two popular genomic methods - double digest restriction-site associated DNA sequencing (ddRAD) and target enrichment. The former is mainly applied at shallow phylogenetic scales and the latter at both deep and shallow scales. Although both of these methods have been adequately utilised for species delimitation purposes, there are only a handful of studies that have compared these two genomic approaches in the same study system. We applied phylogenetic, population genetic and species delimitation methods and compared the results obtained from the two approaches. Using a recently developed target enrichment probe kit, we were able to capture 1,743 loci with a low amount of missing data and compared these with already available ddRAD data from a previous study on the same set of specimens. We recovered consistent phylogenetic relationships across the datasets, both demonstrating the presence of a genetically distinct Balkan lineage and paraphyly of Melitaea athalia with respect to Melitaea celadussa. The same relationships were also found in a species tree analysis of the target enrichment dataset using ASTRAL. Population genetic STRUCTURE analyses supported the presence of two species when using ddRAD data, but three species when using target enrichment, while the Bayes factor delimitation analysis found both two and three species scenarios equally decisive in both datasets. From the geographic distribution of genomic admixture, we confirm the patterns observed by a previous study that used ddRAD data. As the results obtained from both methods were largely congruent, we discuss some practical considerations and benefits of target enrichment over RAD sequencing. We conclude that the choice of method of genomic data collection does not influence the results of phylogenetic analyses at alpha taxonomic level, given a sufficient number of loci. Finally, we recommend a solution for delineating species in parapatric scenarios by proposing that parapatric taxa be consistently classified as subspecies or complete species, but not both, to promote taxonomic stability.