Hi Aaron, Cody and Julien! Cody I agree with Julien that you are mainly interested in whether the off-diagonals in A are zero or non-zero. The off diagonals in sigma would mean there is correlation but due to strange effects. You should also try with triangular A matrices - i.e. you may have adaptive effects only one way. e.g. the males are adapting to changes in females? Another model you could try is a slouch type model (OUBM type model in mvslouch) females are behaving as a Brownian motion while males are adapting to them? All depends what is the underlying biology of the system. Once you also start putting in ecological factors everything could change. Say that both males and females are adapting independently to the same ecological variable e.g. plant colour in their habitat. Say that this is behaving as a Brownian motion. Then if you leave out this variable the OU models could result in : females as BM, males as OU adapting to females and lead to the conclusion that changes in males are adapting to changes in females. But once you put in this ecological variable then you should get plant colour BM, males and females as OU adapting independently to plant colour. Of course in both cases males and females are correlated (if that is only what you care about?) but conditional on plant colour independent.
Did you try running mvslouch on a reduced data set? How do the results compare? Aaron, I don't think complex eigenvalues would be that much different in interpretation? Adaptation happens if the real part is positive, disruptive selection if negative. But of course it is a much more interesting problem how the path to stationarity looks like. Best wishes Krzysztof _______________________________________________ R-sig-phylo mailing list - [email protected] https://stat.ethz.ch/mailman/listinfo/r-sig-phylo Searchable archive at http://www.mail-archive.com/[email protected]/
