Fixing emigration parameters in the Robust Design

[modonnell]

Question: Since I know which individuals have permanently emigrated from the study site, can I fix the gamma parameters to 0 and code these individuals with -1's in the group codes (basically call them dead and understand that the estimates of survival will be wrong)?


Background and reason for thinking about this:

What I am thinking of doing is using the Robust Design to obtain estimates of p and N as a way of comparing how well one new method of capturing individuals compares to a widely accepted existing method of capturing individuals.

I have a data set with 5 primary sampling occasions and 46 secondary occasions spread over the course of about one year; three primary occasions consist of 10 secondary occasions a piece, one primary has nine secondary occasions and the last primary has 7 secondary occasions. There is generally at least one month between primary samples and within each primary occasion the secondary occasions are spaced about 12hrs apart. Also, there are two cohorts that are included as groups in the input file.

Since I am able to identify individuals which emigrated from the study site between primary occasions, (because they were caught somewhere else outside of the study site) I have a pretty good idea as to what "real" emigration is. In order to evaluate the capture method, I'm most interested in evaluating estimates of N and p. I can fix the gamma parameters to 0 and code the individuals that emigrated with -1's (instead of 1's) in the group codes. I realize this will lead to erroneous values for survival, but I think the p's and n's should still be good.

I thought of using Closed Capture models for each primary occasion, but the problem I ran into is that even when the last p is constrained it looks like N is just a count of the number of individuals per cohort from the input file. The recapture rate is pretty high so maybe that is why MARK is just using a count for N(?). This would be okay, but some individuals had to be removed from the input file because they were only captured on the last occasion (this study is part of a larger study in an area where we already had tagged individuals); thereby decreasing N because they were no longer in the input file to be counted.

Any thoughts?

Thanks,

Matt