Monica Piotrowski (Utah) also is talking about DNA Barcoding. She starts with a child's coin sorter. Imagine that it's a bug-sorter, sorting by DNA samples. What does the child now have? She claims Barcoders must have a species concept to measure the success of their practice. They have none, and so they are in a two-horned dilemma neither option of which is a good option.
COI is claimed to be 99.75% identical within, and <97.5% between species. Hence it can delimit, but not describe, species, according to Hebert and Strauss, the founders and promoters of Barcoding. So, what's the difference? A recent issue of Systematic Biology deals with this. It depends on the concept of species, for otherwise you cannot tell if it works. As they have none, we must reconstruct it:
Option 1. None; merely a clustering of percent similarity.
Option 2. Unarticulated concept that demands correlation between COI sequences and species.
Option 1 is a similarity based, or phenetic, account. These are arbitrary: we can choose any property we like and groups as we like. So why is the barcode the right similarity? Without a concept of species, that cannot be answered. Moreover, these categories have no scientific payoff - they lack explanatory and predictive power. I would call this a lack of inductive projectibility, as Goodman called it. Common descent makes prediction possible. Mere similarity doesn't. [Example of Taxol from Pacific Yew, and finding it in European Yew.] Option 1 is arbitrary and unscientific (and I would say, uninteresting).
Option 2. What would this look like? It's unable to accommodate vague boundaries. Barcoding sets a sharp boundary - it's implausible. It conflicts with natural continuity. DNA barcoders give priority for the evidence over the natural processes that give rise to the evidence. E.g., similar because members of the same family, not members of the same family because similar.
Neither horn is preferable, and hence the proposal is not to be preferred over traditional methods.
- Log in to post comments
Um, but barcoders do have a species concept. It's called "percentage similarity based on the CO1 gene".
"COI is claimed to be 99.75% identical within, and <97.5% between species."
and
"Barcoders must have a species concept to measure the success of their practice. They have none..."
make no sense together.
How can anyone claim anything about divergence within species with no idea what a species is?
The percent similarity is assessed empirically against traditional taxonomic designations. It varies among groups to some extent. In terms of possible new species, it is used to note groupings that differ by the amount that traditional named species do, indicating that it may be worth looking for morphological, behavioural, or reproductive distinctions between the groups that would warrant naming with traditional approaches.
You can take the COI data and assess their ability to delimit species as defined by any criterion you like. Morphospecies, BSC, PSC, whatever you want. Do the barcodes recover those designations? That is the question, and it won't be answered by armchair speculation, it will be done with real-world data.
I agree Ryan. But I'm reporting here, not advocating. Problem is, as Brent also argued (and as I know independently) that there is no evidence that COI tracks anything other than COI sequences. So at the moment it is an unsupported claim. Brent notes that we've been using DNA sequences as evidence for a while (e.g., 16sRNA). But we need evidence that this does something interesting WRT taxa.
I don't follow what you mean -- COI only tracks COI? The question is whether COI tracks traditional species names, and thus morphology or other genes or whatever they were named on. It also has done the reverse in identifying groups that differ in COI and which were then examined for ecological or morphological or other differences. In other words, it not only "tracks" species boundaries, it can highlight potential boundaries that were overlooked initially. You make it sound like the entire thing is done independent of taxonomy, which I can assure you is completely incorrect.
For people who would like an alternate take on DNA barcoding studies by someone who has actually done some of the research, see Mark Stoeckle's Barcode of Life blog.
Based on John's report, it seems Piotrowski is setting up a straw man with the whole 'barcoders need a species concept' thing and then revealing that because some people think species aren't real, barcoding can't work. If species aren't real, we're all in the same boat, folks, because it's not only barcoding that won't work, it's any system that relies on "species" being a biologically significant designation ...including names based on morphology.
Does Piotrowski think that current (i.e. traditional) species names reflect something that is biologically real, in other words, do names work? If so, then we can test whether barcodes work too. If not, then I would ask what Piotrowski thinks taxonomists should start doing for a living from now on.
Also, Bob O'H: barcoders don't claim that barcode similarities should determine the species concept but simply reflect the "accepted", traditional species concept for that group. As Ryan notes, this can be tested empirically. A barcode either succeeds or fails at "tracking" accepted species names and, in the cases where it fails, one then asks whether this is because the barcode really fails to track species names in that group or because the accepted concept needs to be re-evaluated (using traditional/accepted methods).
Also, we're coming to understand fairly well where differing approaches fail. Barcodes don't work well with very young species or hybrids. But it is often possible to flag these cases and note that more information (additional genes, morphology, behaviour) are necessary to get down from a group of two or a few to one species identity. Morphology fails with cryptic species, with species that differ behaviourally or ecologically but not morphologically, with extreme sexual dimorphism, and with ambiguous life stages (e.g., eggs, larvae, juveniles). So we need both. Call for something dubbed "integrated taxonomy" if you want, but that's what barcoding has been doing for several years anyhow. When it comes to identifying possible new species, barcoding provides genetic clusters that can then be examined for divergences of morphological or other characters. The difference is between "here is a pile of organisms -- are there any groups?" and "here are some probable groups -- do they differ in criteria that warrant naming?", and obviously the second is much more efficient and less frutstrating. This is why many taxonomists find barcodes incredibly useful as a triage method, not to mention that it frees them from doing routine identifications so they can spend time describing new taxa. And yet, a few vocal critics still argue that barcoding is meant to replace taxonomy or takes funding away from taxonomy. It is meant to enhance, support, and provide access to taxonomic data.