Partial least squares regression with categorical variables?

This is similar to a recent post about SNP data.

I have genotype data based on SNPs (single nucleotide polymorphisms) for ~300 genetic markers, scored for 60 or so individuals. This dataset is categorical (each individual is scored as 0 or 1 at each marker site).

I also have phenotype data for these individuals which comes from a study of shape using landmark morphometrics. A principal component analysis was performed on the landmark coordinates.

The idea of the study is to find the genetic loci that control components of shape. I mapped each principle component individually but would like to be able to use PLS to find axes of variation that closely correspond to variation in the genotype data.

I have used PLS in R and I think it works but have been told that I shouldn't use PLS with categorical data. Is this correct? What could I use instead?

Any thoughts muchly appreciated. I can provide more details if this doesn't make sense.
I am new to Talk Stats, and I have not been able to find a reply to the above inquiry by RJoness on partial least squares analysis. I have a similar situation involving categorical variables in a PLS analysis. I have one set of morphometric landmark data, on which I also performed a PCA. I also have behavioral data for the taxa involved, so that each taxon is coded with a particular locomotor category (which I could score as 1, 2, 3, 4, or 5). I would like to examine covariation between the shape variables and the categorical behavioral variables. I have been trying to do this via the two-block PLS option in MorphoJ, but can't figure out how to input the categorical data. As in the question above, is this a correct way to use PLS? And does anyone have experience doing this sort of analysis in MorphoJ?
Thanks very much.