Is there a mutation gradient along vertebrate mitochondrial genome mediated by genome replication?

There is a long-held belief that a mutation gradient exists along vertebrate mtDNA, mediated by mitochondrial replication that leaves different parts of the H-strand exposed in single-stranded state for different durations (DssH ). However, the predicted mutation gradient and its tests suffer from both conceptual and empirical problems. I assembled representative mammalian, avian and crocodilian mtDNA to test this prediction. I measured substitution rates at codon positions 1 and 2 (S12 ) and at codon position 3 (S3 ), as well as synonymous and nonsynonymous substitution rates, and checked their change along the hypothetical gradient. Mammalian species do not support the predicted mutation gradient, although they should according to the model. Crocodilian species exhibit a pattern closest to the prediction, although they should not because their OL , if present, is not at a fixed position. Correlation between S3 and DssH is much weaker than that between S12 and DssH (contrary to the prediction). This is not due to substitution saturation but is instead due to differential gene conservation, e.g., COX1 is far more conserved than ND6 in all metazoans no matter where they are located along mtDNA. In vertebrates, conserved genes such as COX1 happen to have small DssH and variables genes such as ND6 happen to have large DssH . The observed "mutation gradient" is driven by nonsynonymous substitutions, with synonymous substitutions associated with a much weaker "mutation gradient" likely caused by differential codon re-adaptation after nonsynonymous substitutions. The mammalian and avian results are also confirmed by a much larger compilation and analysis of 691 mammalian and 462 avian mtDNAs. The results, however, does not reject paper is not a test of the strand-displacement model (SDM) of mtDNA replication because a mutation gradient is not a necessary consequence of SDM.