The animal mitochondrial genome is single, effectively non-recombining chromosome, and the genes on this chromosome form one linkage group (Gray 1999). Under such circumstances, genetic hitchhiking is inevitable (Maynard Smith and Haigh 1974). Genetic hitchhiking results when strong positive selection on one genetic element causes an increase in the frequency of not only the element under selection but also of all of the genetic elements to which it is linked (Gillespie 2000; Meiklejohn et al. 2007)(Figure 1). The implications of genetic hitchhiking for the creation of a mt DNA barcode gap are inescapable (Costa and Carvalho 2010). If a favorable mutation occurs in any part of the mitochondrial genome—if for instance there is a nucleotide substitution in a mt-tRNA that improves the speed and accuracy of translation of mRNA (Adrion et al. 2016)—then positive selection for that mutation would cause an increase in the frequency of the entire mitochondrial genotype that held that mutation . If the mt chromosome that carried that favorable allele happened to also carry a unique, neutral mutation in the barcoding region of the COX1 gene, then that COX1 mutation would rise in frequency along with the mt-tRNA gene. Selection for the favorable allele could lead to rapid fixation of the new genotype, purging all diversity in mitochondrial genotypes within that population (Fig. 1). This process of genetic hitchhiking would essentially pull the mitochondrial genotype through a series of bottlenecks that would simultaneously purge standing variation within a population and fix differences in mt genotype between populations, creating the pattern of barcode gaps that typify the genomic structure of animals (Maynard Smith and Haigh 1974; Barton 2000; Meiklejohn et al. 2007). Because the mt and N genomes are inherited independently and N genes engage in recombination each generation, N genes could escape the bottleneck events affecting gene frequency in the mt genome.

The power of this explanation is that the proposed process would be ubiquitous among animals. Across most bilaterian animals, the genes that contribute to the function of the electron transport system are rigidly conserved—the same N genes cofunction with the same mt genes in a fruit fly and a chimpanzee (Boore 1999; Gissi et al. 2008). A common set of interacting genes that are subject to the same functional constraints is exactly the circumstance that would give rise to a universal, selection-driven mt biological clock that runs faster than predicted by neutral theory (Hickerson et al. 2006). Adaptations to the external environment would only add noise to the dominant mode of evolution driven by mitonuclear coevolution.

Selective sweeps arising from the rapid fixation of mt variants under positive selection is a process already under discussion regarding the pattern of variation in mitochondrial genotypes within and among populations (Meiklejohn et al. 2007; Kerr 2011). By adding a need to consider both the protein coding and non-coding genes of the mt genome to the list of genes likely to be subject to at least periodic positive selection, a much greater opportunity for frequent selection sweeps is recognized. The majority of gene products of the mt genome is tRNAs, and the rate of mutation and evolutionary change of tRNA is much greater than the rate of amino acid substitutions in protein coding genes (Thornlow et al. 2018). Moreover, changes to mitochondrial tRNAs can have large effects on function and fitness. Numerous maternally inherited mitochondrial diseases are caused by nucleotide substitution on genes coding for mt tRNAs (Suzuki et al. 2011) and effects in non-human animals have also been documented (Meiklejohn et al. 2013). Given that function of mt tRNAs is dependent on the genotype of N-encoded Amino-acyl tRNA synthetase and N-encoded post-transcriptional processing proteins, we would predict positive selection for better performing variants as well as negative selection for dysfunctional variants (Pett and Lavrov 2015; Adrion et al. 2016). The same arguments for the importance of functional evolution of mitochondrial tRNAs also apply to mitochondrial-encoded rRNA (Scheel and Hausdorf 2014). Mitochondrial rRNA evolves at a rate that is an order of magnitude faster than the N-encoded ribosomal proteins (Barreto and Burton 2013) and these changes have functional consequences: as with tRNA, human inherited diseases are linked to nucleotide changes in mt rRNA (Scheper et al. 2007). Changes to the nucleotide sequence of the control region also can have functional consequences in terms of human disease (Chinnery et al. 2002) and functional divergence of the control region among sister taxa of animals can play a role in postzygotic isolation of populations (Ellison and Burton 2010). Positive selection on any of these non-protein coding genes should lead to selective sweeps that would fix neutral changes across the mitochondrial genome, including in DNA barcode regions, and this process would be perpetual and inevitable because of the necessity of coadaptation of the mitochondrial and N genomes.

Eyre-Walker (2006) pointed out that there is an interesting interaction between population size, genetic diversity, and genetic hitchhiking. As the size of a population increases, the amount of genetic diversity contained within that population, both in the mt and N genomes, will increase. This increased with-population diversity of mt genomes would work against the effectiveness of mt DNA barcodes for large populations. However, larger populations offer greater potential for the appearance of adaptive variants of mt genes, and hence a greater opportunity for genetic hitchhiking and selective sweep. He suggested that these two opposing forces might tend to negate each other, leaving genetic diversity of mt (but not N) genotypes largely independent of population size.