In most mark-recapture studies, the length of the study does not exceed the average lifespan of the subject animal. However, in some cases the lifespan of the animal can be relatively short with respect to the length of the study (e.g. studies on small-mammal population cycles: 2-yr lifespan versus 20-yr of monthly recapture occasions). In such cases, an individual capture history would resemble something like this:
000010110100000000000000000000000000000000000000000000000000......
If these are monthly capture occasions, we are pretty sure that this animal is dead after the 24th occasion (after its first capture). However, MARK cannot know this, and to my knowledge it tries to estimate the probabilities of recapture and survival for the later periods (when the animal is dead for sure). When the number of occasions increase (e.g. > 100 occasions) the estimation process becomes extremely difficult, and it takes hours and hours to run a single CJS model. A multistate model can take days, if MARK does not crash before that.
Would it be theoretically possible to define a "maximum lifespan" while running a simple CJS or multistate model (e.g. in "new model" window or in the inp file: 00010100000000000XXXXXXXXX), so that MARK will know that the animal is dead for sure at the end of its max lifespan, and will exclude the later occasions from the estimation process? Would this change make the estimation process "much" easier and faster? And finally, how difficult would it be to implement this in MARK?
Thanks,
Arpat