Including chick banding as 1st capture

[Lachie Mc]

I am trying to test for differences in sub_adult survival between 16 cohorts that were all banded as chicks. I have 3 resight occasions ( 1 in each yr of the study) 4-15 years after initial release (as chicks). I am wondering whether I should include the number of chicks banded as the first history of capture in the recapture histories file. If so this would equate to a lot of missing records at the second resight occasion because a lot of chicks were banded each year (up to a 1500 per yr), but this may be OK, after all chick mortality is often high.

To compare sub-adult survival between cohorts it seems it is necessary to set up the file as individual histories (ie not summarised by cohort). Is this correct or can you do this analysis by summarising encounter histories? Or Is it simply a matter of adding 15 columns on the end of my encounter histories data file to describe which individual belongs to which cohort, then running an aged based model looking at "group" diffrences. I'm just a bit confused as to how to set up the inp file. Any help would be much appreciated.

Regards

Lachie

[jlaake]

Not sure I got it all but these data will be a bit of a challenge. First to your question, yes the first occasion will be a 1 for each chick released. From what you have described, you have a classic CJS (recaptures only) except that you have many years of releases but only a few (3?) years of resight. Depending on how the resight years intersperse with the release years, some of the cohorts will provide very little information about sub-adult survival. What is sub-adult age range? If your age of first resight is 4 then you won't be able to separately estimate chick survival from survival to age 4, so unless sub-adult is > age 4 I can't see how you are going to make any progress.

[abreton]

"I am trying to test for differences in sub_adult survival between 16 cohorts that were all banded as chicks...To compare sub-adult survival between cohorts it seems it is necessary to set up the file as individual histories (ie not summarised by cohort)."

To perform this test, you're going to have to decide on whether you want to represent cohorts as groups or a covariate. Evan Cooch recently updated Chapter 12 in a 'Gentle Introduction' with text to help analysts decide which might be best for their situtation. Here's what he recently wrote (yesterday I think): "I've added a bunch of text to Chapter 12 (the individual covariates chapter), which covers the [group versus covariate]issue is some detail (perhaps enough, perhaps not - let me know - it always looks right to me The new version hasn't been rolled into the 'single PDF file' version of the book yet, so to look at the new stuff (specifically, new section 12.6), access the individual file for chapter 12. Beyond feedback on whether or not I should change anything, please let me know if there are any outright errors."

Okay, at this point in your post I get confused. You mention 16 cohorts, 3 (resighting) occasions, "1 [occasion] in each year", 4-15 years after release, and "a lot of chicks were banded each year". I'm puzzled (e.g.) how you can have 3 occasions (you can drop the 'resight') after monitoring 16 cohorts for 4-15 years? Perhaps you can post and interpret one of your encounter histories to help clarify.

In addition to the confusion about encounter histories, the following text suggests that you intend to incorporate counts into your encounter histories: "I am wondering whether I should include the number of chicks banded as the first history of capture in the recapture histories file." Assuming you're intention is to use a single state model (e.g., CJS), then the first part of your encounter history should consist of only ones and zeros (see the discussion on encounter histories in Chapter 2 'gentle introduction'). Alternatively, perhaps what your asking is if you should include the initial release of chicks in your encounter histories. In my opinion, if you want to estimate sub-adult survival, then releasing them as chicks will be essential. If you don't, then your analysis will quickly become an analysis of the survivors since these will be the only animals that are available for subsequent release.

"Is it simply a matter of adding 15 columns..." - There is nothing wrong with this approach if you want to represent your cohorts as groups...but see options described above in Chapter 12.

"...aged based model looking at group diffrences.." - Since you're releasing known aged animals, an age structure makes sense for your analysis. Unless you plan to use RMark (see archives MARK forum), then you'll want to read chapters 6 and 8 (gentle intro) which will explain how to apply both age and group structure to models in MARK (and probably Chpater 7 as well). As you'll discover, you'll want to code your PIMs (parameter index matrices) BEFORE running any models in such a way that both age and group structure are accomodated. Next step, close the PIMs and never open them again. Finally, open the design matrix and initiate model building. I hope at least some of this was helpful!