ordination with different variable types?

[Rachel Anderson]

Hi all,

I'm working on my third chapter and am nearly done (yay!).  It examines habitat correlates with presence/absence of a couple of amphibian species.  I used a boosted regression trees approach to get at the relative importance of each habitat predictor, which worked reasonably well, but in order to have a well-informed discussion, I need some kind of ordination analysis that can tell me which predictors are correlated.  I can't use a PCA (I think) because I have a mix of categorical and continuous variables.  I tried using a mixed factor analysis (MFA, package FactoMineR) but I can't quite make it happen for me.  

Does anyone have experience with MFA and might be able to provide a little additional guidance?

or

Have ideas for a different approach for some type of ordination with mixed variables?

Thanks!

Rachel