I'm working on testing some trait correlations and am wanting to use phylogenetic correction (since the species I've gotten the data from are closely related). It's not a particularly complex relationship I'm examining:

Y ~ X1 + X2

However, I'm running into two issues that I'm not sure how to resolve. I'm also not sure if these issues are due to a lack of understanding of existing methods on my part, a lack of methods to deal with my issues, or if the issues I think I have (and how I'd like to address them) are non-issues or from incorrect goals on my part.  So far I've mainly been using BayesTraits, but I've got some introductory experience with phytools (R) and threshml (PHYLIP) Essentially the issues (I think) I have are:

1) Non-phylogenetic model selection shows that X1 is best represented as a discrete character (binary) while Y and X2 are both continuous traits. So far, I haven't been able to find a method to combine discrete and continuous characters in a phylogenetic context (or understand how to test the above relationship using methods that do allow mixing discrete and continuous characters (threshml and PhyTools)).

2) Prior testing (in BayesTraits, using Pagel's lambda) of phylogenetic signal indicate that Y and X1 have phylogenic signal, however X2 does not seem to have a relationship with branch length. Is there a way to allow lambda (or other scaling factors) to be independent for different traits during model testing? Or, is there something methodologically wrong with that question?

As well, any advice/pointers/suggested reading for phylogenetic comparative methods would be appreciated!