I need help with tag retention modeling study!

[Eric Janney]

As part of a long term capture-recapture project I have collected tag retention data consisting of 3 years of double tagging adult fish (3 sucker species) with PIT tags and Floy (anchor type) tags. We have four years of returns for these double tagged fish. I would like to model tag retention rates using AIC to determine the effects of 1) fish size 2) species 3) sex 4) time. I have looked at several papers (Barker et al. 2002; Fabrizio et al. 1999) that have used this approach. The main difference in there data sets and mine is that they double tagged using the same tag type (anchor tags) for both tags, whereas I have used different tag types. PIT tags are our primary tag type. Anchor tags were only used to assess PIT retention. I'm not sure how, or if this will alter the multinomial model that they used. Fabrizio et al. programed the multinomial model into program SURVIVE in order to compute maximum likelihood estimates and compare different models. I've never used SURVIVE and am wondering it is possible to use this kind of model in MARK or would it be easier to get help from someone on how to set up this type of model in SURVIVE. Any help on this would be greatly appreciated

[ganghis]

The answer depends on your sampling situation

When in the year do recoveries occur, and when does banding occur? What type of loss rates are you looking at with anchor tags?

If recoveries occur right before the banding period (or right after it) you might be able to trick a multistate model in MARK to give you what you want. If anchor tags are not lost, you might look at a model that should be coming out in the next issue of biometrics (Conn et al.).

If neither applies, you're in a bit of a tight spot. It would be relatively easy to program a few models in SURVIV for cases without individual covariates (length for example), but if you really need length in there, you'll need to get a programmer to help you.