Single season occupancy model selection and importance of covariates

[sbuitr...@gmail.com]

Hello everyone,

I hope you are all doing great. I have been working in a single-season occupancy model for a deer species in a national park of my country and I am having some issues regarding the model selection/construction. I have several covariates that I believe can explain either p and psi, but not too many sites (only 63), so I first decided to find the best model for p while keeping psi constant and then using this one to find the best model for psi. Nevertheless, I have some models with small AIC differences (<2) and I am not sure how to proceed with the selection of the variables. 

For example, here is my set of models for p while keeping psi constant: 

                                            nPars    AIC delta AICwt cumltvWt

p(densidad)psi(.)                     3 149.67  0.00 0.556     0.56
p(trigger + densidad)psi(.)     4 151.12  1.46 0.268     0.82
p(.)psi(.)                                     2 152.76  3.09 0.119     0.94
p(trigger)psi(.)                           3 154.22  4.55 0.057     1.00

As you can see, the first two models are the one that best explain my data, but I would like to know which variable should I use to continue with my model selection? (to find the best model for psi).

I have seen that some authors check the importance of each covariate and that gives and idea on how to proceed (example below). Is this a good approach? Any advise on how to do this?

Thank you in advance. Any help will be much appreciated.

Kind regards, 

[ryan.c...@gmail.com]

I like Arnold, T. W. 2010. Uninformative parameters and model selection using Akaike's Information Criterion. Journal of Wildlife Management, 74:1175-1178. for this topic.