Ted makes an important distinction, but I think that in both situations
you can use GLS since the variance-covariance matrix among all nodes
(internal and terminal) of the phylogeny can be computed. The formulae
are in Martins Hansen (1997, Am Nat) and used internally in some
functions in
Do you only have phenontypic data at the tips of the phylogeny, or were the
lineages also
phenotyped along the way e.g., at the time of some of the bifurcations? If
the former,
then it seems like a straightforward phylogenetic analysis (unless I am missing
something). If the latter, then it