multiseason occupancy and individual covariate plots in MARK

[Paul]

Hi all,

I’m hoping for some help with using the output from the Robust Design Occupancy Estimation, epsilon, gamma in MARK.
I am running the multi season model (six consecutive seasons) and have data for four groups (treatments) over a total of 55 sites (i.e. each of 55 sites was surveyed six times).

The first problem is the naïve estimate of occupancy was higher than the estimate provided in MARK. From searching the forum, I understand this is because the reported probability of occupancy is for the first session only, whereas my naïve estimate is over all sessions. However, the derived estimates for psi-hat in the notepad window are also given for each session. With the exception of one value (effectively zero for psi-hat, SE and CIs) these psi-hat values are highly credible. So the question is, how appropriate is it to report the derived estimates for psi for each season, with SE and CIs and how is best to report the one anomalous value?

The second problem is with plotting the individual covariate plots. The AICc best model(s) includes two covariates, and when psi is plotted for each group over a range of values for these covariates the plots make real biological sense, except of course for the first season for the group with the anomalous psi-hat value. I assume this is because MARK uses the beta value for psi from the first session. If this is the case, and given MARK gives the psi-hat values for the remaining five sessions, would it be appropriate to take the logit of one of these (the final session?) to derive a beta value for psi and then use this beta value for this group in an individual covariate plot manually in excel?

Paul