“COI-like” Sequences Are Becoming Problematic in Molecular Systematic and DNA Barcoding Studies

The cytochrome c oxidase subunit I (COI) gene plays a pivotal role in a global effort to document biodiversity and continues to be a gene of choice in phylogenetic and phylogeographic studies. Due to increased attention on this gene as a species' barcode, quality control and sequence homology issues are re-emerging. Taylor and Knouft (2006) attempted to examine gonopod morphology in light of the subgeneric classification scheme within the freshwater crayfish genus Orconectes using COI sequences. However, their erroneous analyses were not only based on supposed mitochondrial sequences but also incorporated many questionable sequences due to the possible presence of numts and manual editing or sequencing errors. In fact, 22 of the 86 sequences were flagged as “COI-like” by GenBank due to the presence of stop codons and indels in what should be the open reading frame of a conservative protein-coding gene. A subsequent search of “COI-like” accessions in GenBank turned up a multitude of taxa across Crustacea from published and unpublished studies thereby warranting this illustrated discussion about quality control, pseudogenes, and sequence composition.