Hi,
I'm working on a project investigating red squirrel responses to call broadcasts and my goal is to model variation in detection probability across seasons. I have a dataset that includes data from sixteen points that were sampled across 5 seasons (number of visits varies among seasons - 2,3,4,3,2). I have been using the first parameterization of a multi-season model (initial occupancy, colonization, extinction, p) and have fixed colonization and extinction, but allowed p to vary with season. My sites have two covariates associated with them.
My top model has a very high untransformed psi value (~25) and huge standard error. I understand from reading other posts is that this likely results from occupancy during season 1 being close or equal to 1 and the standard error being infinite/inestimable. Based on the raw data, it is reasonable to think that occupancy is close to 1, as there were detections at almost all sites during at least one of the visits during season 1. My question is whether it is a problem to proceed with selecting and further interpreting the results of this model? Especially considering that my primary interest for the research project is to estimate changes in probability of detection? If so, what would you recommend as the next best step?
Thanks in advance.