The uses and folly of DNA 'barcoding'.

According to Wired, DNA barcoding has gone mainstream. I fail to see how this is possible since sequencing technology, as cheap as it is becoming, continues to only be available to few and not the public. It's not like, for example, you can pick up a PCR kit and "Sony Deluxe Pirosequencer XL" on Amazon (though you can make your own gel eletrophoresis setup).

If you aren't familiar with the concept, a DNA 'barcode' is a short length of DNA sequence that can distinguish between closely related species. In most cases when people talk about it, they mean the mitochondrial gene Cytochrome C Oxidase subunit I (or COI for short). COI is found universally in all organisms that have mitochondria, so people talk about it being a "universal barcode", that could be potentially used to identify any (eukaryotic) organism on the planet.

The process for getting a sequence is relatively simple, assuming you have the technology. A small piece of the organism in question is broken down and run through a polymerase chain reaction cycle to amplify the COI gene fragments. Then it is sequenced, either through dye-terminator sequencing or the more fancy, faster, more expensive pirosequencing. The sequences are uploaded to a general reference database such as Bold Systems, which anyone can use to compare their sequences to a known set. The expectation is that the more species sequenced, the more likely you are to get a match.

And to the credit of the IBOL team, some of these applications are really cool, like using the sequence library to catch mislabeled fish in markets and restaurants. That's awesome, one more tool in the diagnostic toobox, albeit an expensive one. Since there are so many copies of mitochondrial DNA in most organisms, far more than nuclear DNA, you only need a very small fragment to obtain a sequence. If the specimen is just a slab of meat at market, no problem! Even holotype specimens can be sampled nondestructively.

But identification is really where the utility of DNA barcoding ends, and given the time and expense for obtaining each sequence, it is and will continue to be far easier to use traditional diagnostic methods in most cases. It is not a magic bullet, and it's certainly not a replacement for taxonomy or systematics. Thinking that an entire species can be reduced to a single, definitive sequence length of 500 base pairs from a few individuals is insane. It would be the same as claiming it's alright to dam up Yosemite Valley because we have some photographs. As I pointed out before, the goal of taxonomy is to track characters and their relationships, and assemble a general reference system which is descriptive, predictive and explanatory. The COI gene is one character, out of millions. Reducing species to this one character is pure folly.

The other problem is the definitive nature of these barcodes. They fail to address the temporal variation in a species as it changes over time, much less the spacial variation or level of variation within a single population. Not to mention, the individual base pairs, not the entire sequence pattern or some portion, are used as defining differences in an analysis. How can we use a few individuals to define an entire species, when we know that variation in the defining character, the sequence, exists even between those few individuals? At least taxonomists recognize this variability and only choose character consistent for all known individuals as definitive.

Species are hypotheses, and these hypotheses are going to change over time, but it seems the DNA barcoding proponents may still hold to immutable species concepts despite 150 years of evolutionary revolution in biology.