Testing a covariotide model of DNA substitution

The concomitantly variable codons hypothesis of DNA substitution argues that at any time only a fraction of the codons in a gene are capable of accepting a mutation. However, as mutations are fixed at some positions in a gene, the sites that are potentially variable also change because of changed functional constraints. This hypothesis as been termed the covarion hypothesis or when the model is applied to nucleotides, the covariotide hypothesis, The covarion-covariotide model has proven to be remarkably difficult to test. Here I examine a covariotide hypothesis for I I genes using a likelihood ratio test. I show that in nine of the genes examined a covariotide model provides a better explanation of the data than a model that does not allow constraints to change over time.