Natural distributions of mitochondrial sequence diversity support new null hypotheses.

A variety of forces and constraints can cause sequence data to deviate from patterns predicted under strict neutrality. Here, I present a meta-analysis of available aligned sequence data from 12 higher animal taxa to identify whether the typical null assumption for an often-used test for neutrality--Tajima's D statistic--is an appropriate or useful null given large numbers of empirical observations. Across 1068 cytochrome oxidase I (COI) datasets, the mean value for Tajima's D is -0.391, with over a sixth of these datasets representing "significant" divergence from null assumptions according to this test. These results indicate a persistent trend for mitochondrial COI data--chosen for their prevalence in population and taxonomic studies--to indicate patterns of diversity that deviate from a purely neutral description, and provide compelling support for the concept that more complex "null" hypotheses may be necessary in population genetics.