Animal mitochondrial DNA (mtDNA) evolution contrasts with nuclear evolution. Mitochondrial and nuclear genomes differ in many ways, such as total length, ploidy level, mode of inheritance, recombination rate, presence of introns, percentage of non-coding DNA, effective population size, and repair mechanisms, e.g. [1, 2]. Among these particularities, hypermutability is one of the most striking features of animal mitochondria: the mtDNA mutation rate is typically one order of magnitude higher than the nuclear one [1, 3, 4]. The question of the origin and evolution of such high mutation rates is still open and debated [5, 6]; and see [7, 8] for exceptions.
This high mutation rate is one of the reasons why mtDNA is a very popular marker for biodiversity studies. For example, it has been massively used to investigate intraspecific to intra-ordinal evolutionary relationship, and disentangle rapid speciation events in phylogenetic studies (e.g. see [9, 10] for a comparison of nuclear and mitochondrial markers). In recent years, mtDNA has been used to identify species using a standardized portion of the cytochrome oxydase I gene (COI), according to the so-called DNA barcoding approach . The high mtDNA mutation rate, however, is also the source of frequent homoplasy, i.e., phylogenetic incongruence between sites of the molecule because of multiple mutations at the same site. Homoplasy complicates the use of mtDNA in phylogenetic [9, 10, 12] and even population genetic  studies.
We recently reported several results related to mtDNA mutation dynamics in mammals. First, we showed that mtDNA substitution rates are extremely variable between species: they differ by two orders of magnitude between slow-evolving and fast-evolving mammalian lineages , definitively rejecting the famous "2% per site per million year" calibration, which should not be generally trusted. We proposed that mutation rate variations are possibly linked to species longevity through the action of natural selection: too high a mutation rate would be deleterious in long-lived mammals because it could result in premature aging due to the accumulation of somatic mtDNA mutations , see also [15–17]. Secondly, we found that within-species cytochrome b (cytb) nucleotide diversity is correlated with the mitochondrial mutation rate, as expected, but not with any life-history or ecological variable potentially related to population size, including body size, geographic range and conservation status . This surprising pattern was apparently not explained by selective effects  (see also ), contrary to results obtained at the Metazoa level , especially invertebrates . We therefore proposed that the lack of correlation between mtDNA diversity and potential indicators of population size in mammals is due to strong demographic stochasticity. At any rate, the mutation rate is the best predictor we have of mitochondrial genetic diversity across mammalian species. Overall, these results highlight the importance of mutation rate variations in shaping mtDNA biodiversity patterns in mammals, and suggest that mutational effects should be carefully taken into account when analysing such data.
In this study, we extend our analysis of the evolutionary dynamics of mtDNA through the bird/mammal comparison. Our objective is dual. First, we want to check whether the mammalian results are specific to this group, or have any degree of generality. Secondly, we want to make use of the genetical and physiological specificities of birds to test various evolutionary hypothesis raised by previous studies. Birds are the perfect candidate to challenge the results obtained in mammals: like mammals, they include charismatic, well studied species, in which a large amount of genetic and biological data are available. Being relatively large, warm blooded vertebrates, birds are comparable to mammals in terms of physiology, ecology, and life-history. Birds, furthermore, show a number of peculiarities potentially relevant to mitochondrial evolution and diversity.
The first reason why birds are appropriate for a comparative approach is the availability of direct population size estimates obtained through global population surveys, especially for North American species . Birds therefore provide the opportunity to directly test the relationship between population size and mitochondrial diversity, whereas in mammals we had to rely on ecological and life history traits, plausibly, but only supposedly, correlated to population size. We will therefore check whether the lack of relationship between mtDNA diversity and species abundance we reported in mammals is confirmed in birds, or was due to inappropriate measure of the effective population size.
Birds, secondly, present a genetic peculiarity: female is the heterogametic sex. Because it is strictly maternally transmitted, the avian mitochondrial genome is thus in full genetic linkage with the female-specific W chromosome . Berlin et al.  proposed that the mtDNA diversity could therefore be reduced by Hill-Robertson interference: selective effects applying to loci linked to the W will affect mtDNA through hitch-hiking. This clever hypothesis has provoked some comments and reactions [24–27]. Hickey  questioned one of the most important arguments of Berlin et al. , namely the lower synonymous diversity in bird than in mammal mtDNA, arguing that mutation rate could be a confounding effect. An accurate estimation of mtDNA substitution rate variations in birds appears necessary to correctly interpret the patterns of mtDNA diversity in the bird/mammal comparison.
Birds, finally, have quite high metabolic rates – 1.5 to 2.5 times higher than mammals of similar sizes  -probably because of the high energetic demand of the flying locomotion. Paradoxically, birds species are strikingly long-lived as compared to their mammalian counterparts; on average, birds live three times longer than mammals of similar sizes [28, 29]. Birds can be characterized as long-lived homeotherms, a specificity they share with bats . These peculiarities are useful to understand further the impact of metabolic rate and life-span on mtDNA substitution rates.
Three main hypotheses have been proposed to explain mtDNA substitution rate variations in mammals: the generation time hypothesis [30–33], the metabolic rate hypothesis [34, 35], and the longevity hypothesis [14–17], presented above. The classical approach to test these hypotheses in a comparative framework is to correlate substitution rate variations with the relevant life history traits [14, 17, 33]: female sexual maturity, basal mass-specific metabolic rate (or body mass, taken as a proxy), and maximum longevity. However, these life-history traits are strongly correlated with each other, so that the respective contributions of the three variables are difficult to disentangle. The contrast between the avian and mammalian physiologies provides a unique opportunity to discriminate between two of these competing models, namely the longevity and the metabolic rate hypotheses. According to the longevity hypothesis, birds should present, on average, lower mtDNA neutral substitution rates than mammals, while the reverse pattern is expected under the metabolic rate hypothesis.
To test these predictions and challenge the results we obtained in mammals, we propose to accurately estimate the lineage-specific neutral substitution rate variation of mtDNA in birds using the phylogenetic framework developed by . Using this dataset, we want to check the rough 2%/site/Myr calibration. This calibration, originally estimated with an RFLP analysis on mammals by Brown et al. , was generalized to birds by a study in geese . This calibration rapidly became a standard in ornithology studies, probably because of the scarcity of the fossil record (see [37, 38] and  for a review). Recently, the reliability of this calibration was debated in various studies either supporting [40–42] or rejecting it [39, 43]. We will also ask which of metabolic rate and longevity is the best predictor of mutation rate variations in warm-blooded vertebrates. Finally, we will estimate the mitochondrial genetic diversity (synonymous and non-synonymous) on a wide taxonomic range in birds to test for the neutrality assumption and investigate the link between population size, mutation rate, genetic hitch-hiking, and mtDNA diversity.