effects of individual covariate on survival

[AndyT]

I am trying to look at the effects of habitat and body condition on winter survival. I have a body condition index (c) as an individual covariate, and habitat (h: 4 types) as a grouping variable, and season (s) as a grouping variable.

I have two models competing for the top spot:

phi(c*s,h*s)p(t), and
phi(c*s)p(t)

To assess the effect of body condition, I ran the models with user-defined individual covariate values of + and - one standard deviation from the mean. Varying the body condition covariate in the model phi(c*s)p(t) has a large effect on apparent survival, but in the model the model that includes habitat as well as c (phi(c*s,h*s)p(t)), there is very little effect of varying the body condition covariate.

This inconsistency makes very little sense to me, can anyone help me out?

Thanks in advance,

Andy