Over the past year it has become a morning ritual. Once settled in the office, which means that I have ensured sufficient coffee intake, I start browsing through various email, RSS feeds, and newsletters always on the look out for topics worth a blog post, a bulletin article, or simply a good read that might be useful in teaching. It doesn't happen very often, maybe once a month, that I come across a publication were my first reaction is an outcry of bewilderment. Today was such a day. It was just a title at first but that was enough to cause uneasiness:
Evaluation of the efficacy of twelve mitochondrial protein-coding genes as barcodes for mollusk DNA barcoding.
Really? Isn't that the kind of title one would expect from early DNA Barcoding studies? You know, the good old days were we had to convince everybody (public and peers alike) of the fact that the use of one standard gene region can provide reliable species discrimination. Maybe I am too naive, but I thought the marker discussion for animals ended a long time ago with the result that COI is now agreed-upon standard for all animals and that in fact it actually works well for almost all of them including the phylum mollusca. We know that sponges and some cnidarians are problematic and require either extended COI sequences or alternative markers but overall one of the, for some colleagues surprising, messages of the first decade of DNA Barcoding is that COI works extremely well across the board. So, why did these authors went back to the drawing board? A closer look at the paper revealed their rationale:
However, along with the increase in the number of investigated species, the COI barcode has failed to deliver the reliable DNA barcode in some animal groups. For example, in insects, gastropods and amphibians, intraspecific variation in COI is high and usually overlaps with interspecific variation. In addition, the universal primers often failed to amplify the fragment of COI barcode and other primers were needed. Therefore, it is necessary to search for alternative DNA barcodes to avoid an exclusive reliance on COI.
Indeed, in some taxonomic groups intraspecific variation in COI is high and sometimes it might even overlap with interspecific variation but the word 'usual' is really misleading. There aren't that many examples in the animal kingdom and it is by no means a phenomenon that occurs across entire groups. In fact, in a number of cases were people find no real 'barcode gap' it turns out that the taxonomy was simply wrong or specimens had been misidentified. This so called 'weakness' of DNA Barcoding has led to quite a few species discoveries, taxonomic revisions, and improvements to insufficiently documented collections. Mollusks are no exception to this.
Surprise, surprise, some universal primers aren't that universal after all. Time to find a different gene region. I've come across this attitude before. Some colleagues can't get universal or other published primers to work and as a consequence they switch to a different marker that is easier to retrieve. Who said that PCR is easy? Primer design, chemistry composition, and thermal regime require some thinking and series of experiments. Initial failure should stimulate further experimentation and not abandonment of the original concept. Instead of simply testing the discriminatory ability of mitochondrial protein coding genes the authors could have used the wealth of data they downloaded from GenBank to design a suite of experimental primers for subsequent testing.
I don't want to go into much of the technical details of this paper. The authors used three different ways (distance, monophyly and character-based methods) to determine the discriminatory power of a chosen gene. From what I can tell from the methods section all analyses were carried out using full length genes only. The standard animal DNA Barcode however, is just a fragment (a little more than a third of the total length) of COI and therein lies one of the strengths of this choice. It is a compromise between sufficient resolution at species level and the ability to retrieve it in an economic fashion. In order to avoid comparing apples and pears the authors should have aimed for comparisons between shorter fragments (in the DNA Barcode size range). And in order to raise the bar a little more those fragments should be flanked by potential universal primer sites.
Mollusk DNA Barcodes do have some specialties in stock for us, e.g. sex specific COI versions in a number of mussel species due to limited biparental inheritance, but when it comes to their overall applicability they deserve good marks. The mitochondrial genome of mollusks is very interesting for many other reasons, e.g. gene order variation between species or the lack of genes such as ATP8 in many mussel species. Maybe the authors should have used the opportunity to investigate more into that direction as the results of this study must have been quite disappointing:
In another word, any one of the 12 PCGs (mt Protein Coding Genes) can be potentially used as a molecular diagnostic for species identification in mollusks.
So what they are saying is that the result of their study, conducted because COI is allegedly not a good choice for DNA Barcoding, is that after all COI is suitable for species identification in mollusks (as are all other PCGs).
Hmm. Why did they do this study in the first place?