Tilting at windmills: a response to a recent critique of terminal restriction fragment length polymorphism data.

Blackwood et al. (1) assert that terminal restriction fragment length polymorphism (T-RFLP) data cannot be used to estimate the species level diversity of bacterial communities. While I do not argue with their analytical techniques, nor the interpretation of their results, their conclusion is essentially moot as T-RFLP techniques are rarely used (nor should they be used) to study the diversity of complex microbial communities at the species level. A more useful test would have been to determine if T-RFLP accurately captures bacterial diversity at coarser levels of taxonomic resolution, for which the technique is undoubtedly better suited. On the basis of their analyses, Blackwood et al. (1) argue that our work (2) should be reinterpreted. However, nowhere in our paper do we indicate that T-RFLP captures diversity at the species level, and for this reason, their analyses have no bearing on our study. My goal is not to unilaterally defend T-RFLP or related fingerprinting methods; to do so would be foolish given that such methods are rapidly being replaced by techniques that will allow us to compare a large number of communities at a high level of taxonomic resolution. Rather, my goal is to point out that we do science with the techniques we have, not the techniques we might wish to have [to paraphrase the former Secretary of Defense of the United States (“. . .you go to war with the Army you have. They’re not the Army you might want or wish to have at a later time.” Donald H. Rumsfeld, 8 December 2004, Kuwait)]. We could spend our careers identifying every individual species in a handful of samples (currently a Sisyphean task), or we could look for patterns in microbial community composition by analyzing a relatively large number of samples at coarser scales of taxonomic resolution and moving to finer scales of resolution as the appropriate techniques become available. Although T-RFLP may not allow us to estimate diversity at very fine levels of phylogenetic resolution, such was not our goal and there is no reason to assume that the species level is the most appropriate level of taxonomic resolution for comparing levels of microbial diversity.