Abstract
The most effective approaches for species discovery and species validation with genomic data remain underexplored. This study evaluates the merits and limitations of phylogenetic approaches based on the multispecies coalescent model and population genetic approaches for species discovery, i.e., species delimitation in the absence of prior knowledge, using genomic datasets from four well-known radiations. Furthermore, it demonstrates how geographic data can be integrated with the genomic data for species validation, i.e., for testing primary species hypotheses. The multispecies coalescent model-based approaches tr2 and soda resulted in high over-splitting of species, low percentages of species delimited according to the current classification, and low percentages of individuals assigned to the same species as in the current classification across all four species complexes studied. Conversely, the species numbers were slightly underestimated based on the structure results. Although the proportion of species delimited according to the current classification and the proportion of individuals assigned to the same species as in the current classification in the classifications based on the structure results is approximately twice that of the classifications proposed by the multispecies coalescent model-based approaches, it remains unsatisfactory. A slight over-splitting of species into population groups can be corrected by species validation with isolation-by-distance tests if a sufficient number of populations have been sampled for each species. Sampling design is an essential step in any taxonomic study, as it has a significant impact on the delimitation of the species and the possibility of their validation.