Info

which shows that at the low apparent divergence of 10% there are expected to be 0.7% of sites that had experienced multiple hits. The true divergence is then estimated as 10.73% or slightly greater than the apparent divergence. If apparent divergence was greater, say d = 0.40, then we should expect a larger correction. In that case

so the correction for multiple hits is much larger at a bit over 18% for a total corrected divergence of 58.13% of nucleotide sites. A frequently used convention is that a capital K refers to a saturation-corrected estimate of divergence while lower case k or d is an uncorrected estimate of divergence.

The Jukes-Cantor is the simplest possible nucleo-tide substitution model because it assumes that all nucleotides are equally frequent in DNA sequences and that all sites experience the same substitution rate. Many DNA sequences, however, exhibit variation in these parameters, which is not accounted for in the Jukes-Cantor model. There are numerous models of nucleotide substitution of increasing complexity that take these factors into account by using an increasing number of parameters to represent the different types of substitution rates (Posada &

Crandall 2001). Figure 8.10 illustrates a hierarchy of some of the nucleotide-substitution models available. These different models can be distinguished by examining DNA sequence data to test each model assumption. For example, the Jukes-Cantor model assumes that all nucleotides have equal frequencies. If a sample of DNA sequences shows base frequencies that deviate significantly from equal frequencies of 25%, then the F81 model is a better choice because it assumes arbitrary base frequencies. Both the JC and F81 nucleotide-substitution models assume that transition and transversion rates are equal and that substitution rates are constant among sites. It is now common practice to estimate the substitution model that best approximates the patterns of nucleotide change in a DNA sequence data set (Posada & Crandall 1998).

Substitution model assumptions

Substitution models compared

Substitution model assumptions

Substitution models compared

 Nucleotide frequencies are all equal
0 0