Hi Salvador,
You first need to define the variables that would be resampled during
the bootstrap procedure. In the case of DNA sequences, this would be the
columns (= sites) of the alignment, since they each contribute to the
distance calculation (or to the likelihod function if you use ML).
Salvador Espada Hinojosa --
The problem, of course, is that a random substitution on an interior
branch of a tree increases or decreases the size of more than one
distance in the distance matrix. The distances aren't independent in
their statistical noise. So you can't just sample distances