How to deal with a categorical response and continuous and categorical predictors?

#1
Hi all,
I have a large dataset of categorical and continuous variables, the observations are animal species. I want to test which variables are more influent in determining a categorical variables with 3 levels.
So I have a categorical response variable (3 levels) and many possible predictors (both categorical and continuous).
In addition, I would like to account for phylogenetic signal in the analysis. That is the non-randomness of traits distribution due to phylogenetic relatedness.
A phylogenetic discriminant analysis would be the case, except for the fact that discriminant analysis cannot be used with categorical predictors.
Classification trees (and random forests) would be a good choice, but I don't think (not sure) if they could controlled with a phylogenetic tree.

Do you have any suggestion?

Thanks a lot
 
#3
Re: How to deal with a categorical response and continuous and categorical predictors

Too include as many of your predictor variables, figure out their distributions, then pick an analysis best suited. A General Linear Model might be appropriate with normal distributions - (if transformation is possible for non-normal)