Mitochondrial DNA (mtDNA) diversity is widely used as a marker for species abundance. However, Galtier and colleagues have now shown that, unlike nuclear DNA, mtDNA diversity is not proportional to population size and that this is probably due to positive selection.

The number of neutral polymorphisms in a species should be proportional to the effective population size. As the overwhelming majority of mutations are assumed to be at least almost neutral, the overall sequence diversity of a species should also be governed by the population size. However, other factors also affect diversity, such as population history, life history and population structure, making it difficult to test the effect of population size on diversity in any one species.

Therefore, Galtier and colleagues carried out a meta-analysis of over 1,000 species from a range of taxa. They looked at diversity at three levels — allozyme, nuclear DNA and mtDNA — expecting all to show the same patterns as a result of population size: invertebrates should have more polymorphism than vertebrates; marine species more than terrestrial or freshwater species; and large animals more than small animals. These patterns were observed for allozymes and nuclear DNA, but mtDNA diversity did not vary with population size.

The authors considered possible explanations for the failure of mtDNA to conform to the neutral theory of molecular evolution. They dismissed three explanations as improbable: if variation in mutation rate was the explanation, the mutation rate would have to vary inversely with population size across taxa, which seems unlikely; demographic effects that might have lowered diversity should have the same effects on nuclear DNA; and negative, purifying selection decreases diversity but still leaves larger populations more diverse.

Instead, the authors analysed the effects of positive selection. The fixation of advantageous mutations leads to a loss of diversity at linked sites — the alleles at linked loci that happen to occur in the advantageous haplotype will also be fixed. If this process is frequent, overall diversity will be reduced. However, this process will depend on population size: the larger the population, the more advantageous mutations will arise, and the greater the decrease in diversity at linked sites. This should exactly compensate for the increase in the rate at which diversity arises in large populations because of drift. The overall effect will be that diversity is invariant.

To test whether this was happening in mtDNA, the authors measured the relative rates of non-synonymous and synonymous mutations across taxa. There was a higher rate of non-synonymous change in species with larger populations, but no increase in diversity, which is consistent with positive selection.

It is not known why positive selection is stronger for mtDNA than nuclear DNA. It might be because the mitochondrial genome is particularly gene-rich. Whatever the reason, these results could change the way population diversity is measured and how conservation biologists identify important lineages.