Taxonomic workflow

A first critical issue in DNA barcoding is accuracy which depends especially on the extent of, and separation between, intraspecific variation and interspecific divergence in the selected marker. How well does that gene sequence perform in delineating and identifying species? The more overlap there is between genetic variation within species and divergence separating sister species, the less effective DNA Barcoding becomes. Initial efforts to test barcoding suggested that we need a significant “barcode gap” between intra- and interspecific variation.

The second key point is that there are indeed two different applications for DNA Barcoding that shouldn't be confused. The primary motivation to do DNA Barcoding is the prospect of being able to identify an unknown organism based on a short sequence. This process of querying a DNA Barcode reference library with sequences of an unidentified specimen should bear the name species identification.

However, more and more is species identification confused with species discovery. Building a DNA Barcoding reference library often exposes new species. Despite the semantic difference it should be noted that both species identification and species discovery need different methodologies.

Over the last years a variety of algorithms and programs have been generated that are intended to help with both and more and more publications show up that compare different analytical methods. A new one has seen the light of day last week and I think it is a very good one:

The analysis of DNA barcode sequences with varying techniques for cluster recognition provides an efficient approach for recognizing putative species (operational taxonomic units, OTUs). This approach accelerates and improves taxonomic workflows by exposing cryptic species and decreasing the risk of synonymy. This study tested the congruence of OTUs resulting from the application of three analytical methods (ABGD, BIN, GMYC) to sequence data for Australian hypertrophine moths.

The authors use DNA Barcodes as a taxonomic exploration tool, grouping specimens into OTUs intended to be the first step towards a framework for subsequent phylogenetic and taxonomic work. The novelty lies in examining the congruence of OTUs resulting from three delimitation methods (GMYC, ABGD, BIN) and the subsequent evaluation of cases of discordance in OTU boundaries employing monophyly, diagnostic characters and sympatry as criteria for clarifying their status. This scheme has the advantage of keeping the initial step of OTU designation separate from the detailed analysis required for full-blown taxonomic characterization. Because the varying steps in species delineation require different sampling strategies and types of data, the primary delineation of OTUs with single-locus data has the advantage of employing one extensive data set with clearly defined criteria to produce a stable outcome.

I couldn't summarize the advantage of their approach better than the authors themselves so I won't even try:

We emphasize that the delimitation of putative species based on DNA barcode data not only increases objectivity, but accelerates work on poorly studied groups and enables inexperienced taxonomists to make a valuable contribution. As many groups of arthropods lack expert taxonomists, the need to recruit new experts is obvious and barcode-based approaches provide an easy path for initial engagement. Even without detailed study, an accurate estimate of the species count is obtained through the simple algorithmic processing of barcode data. While decisions based on analysis of single-locus mtDNA data and on small sample sizes do pose interpretational risks, they are inconsequential if the outcome is viewed as a scaffold for taxonomy rather than as the sole criterion for species description.