Kirk Fitzhugh (2006a, 2006b) has recently offered a novel critique of DNA barcoding based on his own carefully considered interpretation of species as “explanatory hypotheses.” Though he is not alone in questioning a method that claims to identify species based on a single genetic locus (Lipscomb et al. 2003, Mallet and Willmott 2003, Wheeler 2005), I fear that Dr. Fitzhugh's particular philosophical interpretation of the problem may prove more than he intends.
According to that interpretation, since species as explanatory hypotheses are the products of abductive reasoning, they cannot be identified on the basis of DNA data alone without running afoul of Rudolf Carnap's requirement of total evidence, which holds that “for one to rationally believe a conclusion on the basis of some set of evidence, then all available relevant evidence must be taken into consideration” (Fitzhugh 2006a). But DNA barcoding is hardly unique in failing to meet this requirement. If bar-coders cannot rationally defend species identifications based solely on DNA sequence because this ignores “other relevant properties in need of explanation” (presumably morphological, biochemical, behavioral, or other non-DNA properties), then surely morphological taxonomists are also irrational if their identifications fail to consider DNA sequences, which are similarly properties in need of explanation.
Dr. Fitzhugh is thus unfair to level his criticism specifically at DNA bar-coding, as it should be aimed instead at any nonintegrative taxonomic method. More to the point, there has never been—nor will there ever be—a taxonomic hypothesis that did not exclude some available relevant evidence. Since any practicable taxonomic approach will inevitably fail the strict requirement laid out in Dr. Fitzhugh's critique, and as the scientific community is unlikely to classify all taxonomy as irrational, I submit instead that Carnap's principle is perhaps not the most satisfying way to assess the rationality of scientific thought.
In addition, I think it important to note that the validity of DNA barcoding does not rest entirely (and perhaps not even primarily) on its success in species identification. Many proposed applications of this technology focus instead on the identification of individuals to the species level. In these applications the identification of species is done quite independently, typically by traditional and integrative taxonomic methods; in fact, most applied barcoding assumes the validity of species identifications made by such methods. Such applications of DNA barcoding could prove enormously useful in a variety of contexts, despite the tendency of some to consider them scientifically uninteresting (e.g., Wheeler 2005, Will et al. 2005). Technically, as individuals are neither hypotheses nor explanatory constructs, Dr. Fitzhugh's philosophical objections do not apply to the adoption of bar-coding as a means to identify them. Even if successful, then, his critique is not a wholesale indictment of DNA barcoding, and should not on its own forestall the pursuit of that technology.