Hi,

You are right on the sense that branch length represent residue substitution per site, either DNA or AA, but it represents the average, that means, that in average you expect a change for each site but in reality you have a broad ranges of substitutions across sites, I mean, from sites with no substitution at all to sites with just one substitution to sites with multiple substitutions.

Hope this makes sense to you :)

S

I think you are trying to calculate the patristic distance between two leaves. This can be done automatically using the program PATRISTIC or in R using APE as follows:

library(ape)
tree<-read.tree("/file.tre")
PatristicDistMatrix<-cophenetic.phylo(tree)

When interpreting divergence times be aware that they are estimates that are corrected for unobserved changes based on a model of sequence evolution, and thus estimated divergence is not strictly equal to observed sequence differences, i.e. 100% estimated divergence != 100 of sites that differ. This is because (ignoring alignment artifacts) as time -> infinity, observed divergence maximizes at 1-(inverse of the alphabet size) (e.g. 3/4 for DNA and 19/20 for proteins) because of random sequence similarity.