As we’ve discussed before, analyzing DNA sequences can help determine species boundaries, important issues for us here on iNaturalist. DNA data can be applied over enthusiastically, however. Here’s a link to an amused summary of the history of the frustrating problem: https://www.inaturalist.org/journal/hillisdm/136706-a-brief-history-of-mitochondrial-dna-mtdna-studies-in-systematics In real life, the author Dr. Hillis is an expert on using DNA to determine relationships among populations and species.
The author refers to mitochondrial DNA because that’s widely used in studies of animals. In plants, though, the mitochondrial DNA gets rearranged easily and is very confusing for studies of systematics. Botanists usually use chloroplast DNA. Like mitochondria, chloroplasts are highly modified bacteria and have the same advantages and disadvantages, including inheritance from only one parent (usually).
Thanks for pointing out Prof. Hillis’s illuminating overview of the topic. I’m onboard.
I’ll have to leave this up so that The Doctor can read it. He’s a biostatistician, so he works with a lot of genetic data, and I’m sure that he’ll feel the struggle. 
I’m in the forest about the use of DNA for species delimitation. I studied Dr. Hillis’ work over the years and chatted with him about his articles. It is critical to agree upon biological species concepts. He is crystal clear that people (a lot) use DNA as a means to split taxa over minor differences without fully grasping localized variation, phenotypic expression due to environmental stressors, and then variation at a landscape scale. Delimiting species is hard work. There is use for shotgun barcoding as a means to examine genomic diversity, but that doesn’t translate into actual species. Well, take what I said for what it is, and don’t take it personal. Not everyone has a big lab with lots of hands to handle specimens, cataloging and the minutiae of lab work, either. Looking at DNA barcoding charts in iNaturalist would fundamentally alter my perception of the platform, and I don’t think its a good thing. iTOL is a mess, why bring it here.
It’s not just mtDNA. Another popular locus is the spacers in the nrRNA arrays. (Picked because they’re fairly fast evolving and exist in multiple copies so they’re easier to amplify.)
A recent paper had no outgroup, misrooted the cladogram, and concluded that genus B was clearly distinct from genus D. The paper also has a haplotype net, which shows genus B nested within genus D. They didn’t notice the contradiction. Nor apparently did the peer reviewers.
Having studied the haplotype net, I concluded that the issue was that they’d drawn a rectangular box round genus B, and in the process had included stem haplotypes in genus B.
Correcting for misrooting without the data is imprecise, but my conclusion was that the results were close to a trichotomy between genus B and two clades in genus D. I have to rely on earlier work, with a proper outgroup, that found genus B to be basal.
Another issue, this time with complete plastid genomes, is the the plastid genome shows structural isomerism (orientation of the inverted repeat) perhaps even on an intra-individual level. Depending on the tool used, this causes artefacts, where samples with the same IR orientation cluster together. The magnitude of the artefacts vary, so it must depend somehow on the tool used, but you get people publishing nonsensical cladograms, with results far from the taxonomic consensus, without commenting on the anomaly. The cladograms look like ones based on duplicated loci rather than ones based on a single locus.
I’ve seen other anomalies (individual taxa in unexpected positions) that have been published with no or minimal comment.