Hebert's scheme focuses on a sequence fragment from the cytochrome c oxidase I (COI) gene, which he claims is unique to separate taxa. He and his colleagues demonstrated proof of principle last year, when DNA bar codes correctly predicted independent species in groups of previously undistinguished birds and butterflies. In a February meeting in London, the Consortium for the Barcode of Life announced plans to bar-code all birds and fish within the next five years, as well as identify all flowering plants in Costa Rica. The initiatives are stepping-stones on the way to their much grander goal: a gene tag for every living thing and a catalogue of the earth's biodiversity (only about one tenth of the world's species are formally known).
But like anything, warns entomologist Quentin D. Wheeler of the Natural History Museum in London, bar coding "can be used for good or evil." A standardized species marker is exciting, provided it corresponds to formal descriptions and classifications. To Wheeler and other critics, it is bar coders' more ambitious goal--applying the COI system backward to create "provisional" new-species definitions--that threatens to hamper taxonomic progress.
The problem, naysayers argue, is oversimplification. "Nature is messy," points out entomologist Daniel Rubinoff of the University of Hawaii at Manoa. Multiple species definitions exist today because nobody knows what qualifies as speciation; the very "science" of taxonomy involves analyzing hundreds of characters to make these distinctions--which is why the one-character data sets used by bar coders "are like returning to the Dark Ages," Rubinoff says. Biologist Brent D. Mishler of the University of California at Berkeley concurs, calling bar coding for species identification "extremely wrongheaded and damaging to the fabric of systematics," which currently relies on extensive morphological, ecological and genetic data to frame species in an evolutionary context.
Critics also raise eyebrows over bar coding's accuracy. Hebert pegs the error rate at 2 percent, small enough to validate the approach for animals. But so far only a few proofs of principle have emerged, and the tests have been easy. "Close sister species are usually the most important to identify correctly," says biologist Felix Sperling of the University of Alberta. And those are the ones scientists may have the greatest trouble resolving through COI. Recently split taxa or cases of hybridization, where independent species have produced offspring, pose particular challenges because sequences may not have evolved to reflect those events yet.
Hebert argues that the system is meant to augment current taxonomy by "heaping life into piles" that can later be revised. But with a price tag between $1 billion and $2 billion, critics worry that the initiative will only divert funds and leave "real" taxonomy to clean up the mess. "In the age of cyber-infrastructure, digital tools, and IT," Wheeler and others write in a paper in press for the journal Systematic Biology, "most of the weights that have held taxonomy back are gone. [But] now ... it is in danger of being tossed out like rubbish for the latest parlor trick."
Experts also point out that bar codes cannot be integrated with the other major systematics enterprise--the Tree of Life, a peer-reviewed cladogram linking all known phylogenetic relationships. (Bar codes provide too little evidence to justify formal species designations on it.) At best, Hebert's database will exist alongside the tree, superimposing untested "leaves." Hebert claims that despite the obstacles "we're finally tapping into an automated, digital information stream," but others note that cut-and-dry codes seem to overlook the essential meaning of species: ever changing end points in the hands of evolution.