Dear Franck,
we faced that same issue in an analysis of snail shell morphologies.
We decided to use randomizations of trait vectors over the phylogeny, so
that they remained correlated but lost the phylogenetic pattern. We were
mainly interested in the pattern of covariation with phylogenetic
Hello,
this problem is recurrent throughout multivariate analysis, and I know of no
universally satisfying solution. In spatially/phylogenetically constrained
methods, the idea is that a relevant structure should exhibit both strong
variance and autocorrelation. I think the simulation
That seems like it would be OK, at least if you think it is OK for
nonphylogenetic PCA.
An alternative is to simulate data along your phylogeny, analyze it the same
way, do it a couple thousand times, then make an empirical null distribution
of, say, the eigenvalues when the data have no
