I've been working with Jessica offlist and want to point out a few things. Notice that she posted
- Code: Select all
Group:sexF.injurymajor 3
Group:sexM.injurymajor 57
Group:sexF.injurynone 3
Group:sexM.injurynone 57
where it shows N was the same across injury groups within a sex. If you use groups you should use N=list(formula=~group) whereas she used ~sex. What you are modelling is f0, the number not caught and while you might think N might be the same across categories, that does not mean that f0 will be the same because it will vary with p.
Also, in the MARK output it provides M_t+1 which is the total number caught in each group. The M_t+1 values where lower or the same as the N values RMark and MARK were giving so it was correct. The beta for N is on the log link scale. If you were to use N=list(formula=~-1+group) then each beta per group are separate (no intercept) so exp(beta) is the estimate of f0 for that group and if you add it to the M_t+1 for that group it will equal N for that group. That N is often referred to as the super-population size which is the number of animals that were in the population at some time and available to be captured. Typically it is not the total number at any one occasion (unless they all entered prior to the study). The Gross N* is subtly different and includes an estimate of animals that entered and left between occasions and thus where never available to be caught.
Note that additional values are provided in the RMark function popan.derived but I just discovered that the code is not correct if you have losses on capture (freq=-1). I'll correct that in a future version. If you have losses on capture, the results you get from model$results$derived are correct.
--jeff