R help - Nov 2012 - Mixed model with multinomial distribution

I would like to run a GLMM on categorical data. In this case the response
data are abundance of insects of 4 trophic groups.

I am unsure whether I need to cbind the categories and run with the model
with family=multinomial4 or family=gaussian.

Also I'm unsure whether I should use lme4 package or MCMCglmm.

If anyone can give me any pointers, I would be most grateful.
I am new to R and mixed models so I apologize in advance.

Thank you

	[[alternative HTML version deleted]]