Modelling Psi and P when using finite mixture models

[Tony G]

I am using single season occupancy models to estimate Psi and P for multiple species of amphibian in tropical savanna wetlands (30 sites / 10 visits / 15 species). The primary aim of the study is to understand variation in P to design future monitoring (10 sampling covariates), but I also have a secondary aim of exploring the effects of habitat and disturbance on Psi (3 site covariates). I'm using a two-step approach: first model detection with a general Psi model, second model Psi with best P. So, to the questions:

1) The pre-defined mixture models (e.g. 2 groups Constant P) are well supported in explaining variation in P for some species. Is it possible to include Psi covariates in these models (ie customise them) so they can be included in the candidate set?

2) Is the two-step approach in this case valid?

Thanks
Tony