Phylogenetic trees may be constructed from pairwise distances among th
e taxonomic units. When the data consist of DNA sequences, appropriate
distances can be defined to reflect the mutation model and to have ex
pected values proportional to the time of divergence of the sequences.
With the growing amount of sequence data, it is also necessary to inc
orporate within-species sequence variation into measures of distance b
etween sequences. This additional feature requires attention to be pai
d to drift and recombination as well as mutation, and greatly increase
s the difficulty of estimating variances of estimated distances. Numer
ical resampling seems appropriate. These concepts are illustrated with
some sequence data from the Adh1 gene from maize and pearl millet.