For what it’s worth, most robust species delimitation studies nowadays incorporate at least one other data type (usually morphology or ecological data), the molecular dataset they’re using is pretty limited for addressing population scale questions (which species delimitation studies often turn into), and concatenation isn’t a particularly good phylogenetic technique compared to more modern techniques that use multispecies coalescent models. I don’t think it affects their conclusions that badly, but my point is there are some better papers out there you can read if you’re interested in how species delimitation works.
As for Taraxacum, apomixis (asexual reproduction) can really throw a wrench in species delimitation. If you apply the biological species concept (and to some degree, phylogenetic species concepts), every individual is basically an independently evolving species. Even under the general lineage concept (probably the most universally accepted in the field), you run into the problem of defining what a population or metapopulation is. Under these circumstances, there is an arbitrariness to where you draw the line if you can’t come up with some other criteria (such as morphological distinctiveness). Unfortunately, these other criteria can be equally arbitrary.