Systematic Entomology (2012), 37, 229–236 DOI: 10.1111/j.1365-3113.2011.00597.x Protein–protein interactions of the cytochrome c oxidase DNA barcoding region LOUISE PUSLEDNIK 1,2 , DAVID K. YEATES 3 , DANIEL P. FAITH 4 and J . W I L L I A M O. BALLARD 1 1 School of Biotechnology and Biomedical Sciences, University of New South Wales, Sydney, Australia, 2 School of Biological Sciences, University of Wollongong, Wollongong, Australia, 3 Entomology, CSIRO, Black Mountain, Australia and 4 The Australian Museum, Sydney, Australia Abstract. Enzymatic amplification of homologous regions of DNA using ‘universal’ polymerase chain reaction primers has provided insight into insect systematics, phylogeography, molecular evolution and species identification. One of the more commonly amplified and sequenced regions is a short region of the cytochrome c oxidase subunit I gene (COI ), commonly called the barcoding region. COI is one of three mitochondrial-encoded subunits of complex IV (Cox) of the electron transport chain. In addition to the mitochondrial subunits there are nine nuclear-encoded subunits of the complex in Drosophila. Whereas a number of phylogenetic biases associated with this region have been examined and the quaternary structure of Cox has been modelled, the influence of protein–protein interactions on the observed patterns of evolution in this barcoding region of insects has never been examined critically. Using a well-resolved independently derived phylogeny of 38 Diptera species, we examined the homogeneity of the substitution processes within the barcoding region. We show that, within Diptera, amino acid residues interacting with nuclear-encoded subunits of Cox are evolving at elevated rates across the phylogeny. Furthermore, we show that codon position two is biased by protein–protein interactions. In contrast, third codon positions provide a less biased estimate of genetic variation in the region. This study highlights the need to examine the potential for systematic bias in DNA barcoding regions as part of the critical assessment of evidence in systematics and in biodiversity assessments. Introduction Since the development of ‘universal’ polymerase chain reaction (PCR) primers (Kocher et al., 1989; Folmer et al., 1994), DNA data has been a fundamental component of insect systematics, biogeography and species identification. One of the more commonly amplified and sequenced regions is a 648-bp region from the 5 ′ end of the mitochondrial-encoded cytochrome c oxidase I gene (COI ). This region, frequently called the barcoding region, is a small part of cytochrome c oxidase (Cox or complex IV) in the electron transport chain. Data Correspondence: Louise Puslednik, School of Biological Sciences, University of Wollongong, Wollongong, New South Wales 2052, Australia. E-mail: louisep@uow.edu.au generated from the region has been used to construct insect phylogenies (e.g. Dijkstra et al., 2003; Mitter et al., 2011) and to investigate biogeographic hypotheses (for e.g. Ribera et al., 2003; McCulloch et al., 2010). However, the accuracy of the results depends intrinsically upon the robustness of the signal that is generated (Ballard & Whitlock, 2004). The goal of this study was to examine protein–protein interactions in the barcoding region using a well-resolved insect phylogeny. Protein–protein interactions have the potential to influence the inferred pattern of DNA changes, as Cox is comprised of three mitochondrial DNA (mtDNA) subunits and 9 nuclear-encoded Cox subunits (for reviews, see Blier et al., 2001; Rand et al., 2004; Dowling et al., 2008; Melvin et al., 2008).  2011 The Authors Systematic Entomology  2011 The Royal Entomological Society 229