Estimate two dif capture probabilies, does sess order matter

[ctlamb]

Hello,

I am looking to use an open model (Pradel or JS) to estimate population growth.

I have a capture data for bears in which we used bait sites for two sessions a year, and in some years we sporadically sampled rub trees, and in others we used a more structured, session-type approach.

My thoughts as far as capture histories was to code two bait site sessions/year and compile all rub tree detections in a single column for each year. So, 3 sessions/ yr, 2 are bait site, 1 is rub tree.

I'll need to estimate capture probabilities separately for these two traps.

Should I always keep my capture histories in chronological order?
i.e., Bait Site Session 1 2006,Bait Site Session 2 2006, Rub Tree annual 2006, Bait Site Session 1 2007,Bait Site Session 2 2007, Rub Tree annual 2007

OR, is the order moreso arbitrary, i.e, I could group by trap type, then year:

Bait Site Session 1 2006,Bait Site Session 2 2006, Bait Site Session 1 2007,Bait Site Session 2 2007, Rub Tree annual 2006, Rub Tree annual 2007

Also, given that the rub tree annual session is running during both the bait site session 1 and 2 it becomes slightly confusing how to code the time intervals??

Cheers in advance,
Clayton, MSc student, University of Alberta

[jhines]

For 'open' models,recapture occasions must be in chronological order. For 'closed' models, order is irrelevant. I think the Pradel models are 'open' models only (no 'robust-design Pradel' model as far as I know), so order of the recapture occasions is important. There is a 'robust-design' JS model you could use if the recapture occasions are close enough in time that population parameters don't change within a group of recapture occasions. Within that group of occasions, order doesn't matter, however, order between groups of occasions does. So, if your bait bite sessions and rub-tree sessions occur at the same time, they can be in any order within a session. If there is not very much time between the 2 sessions per year, then the two sessions can be treated as one 'primary period' and all recapture occasions within that primary period can be in any order. If the two sessions per year are far apart (eg., session 1=Jan, session 2=Aug), then each of the two sessions is a 'primary period' and the order of the primary periods is important.

Jim

[ctlamb]

Jim- Thank you very much for your response, that helps a lot. Given that the rub tree and bait site sessions run concurrently in a year I believe they can be a primary period for each year. I am curious how I should code the time intervals, given the rub tree session is going to be coded as one long session (June-September) while bait sites have two, 2-week sessions during July. Perhaps I should just treat the rub tree session as a 3rd session that follows the bait sites and code it as such (i.e., don't worry about the temporal overlap, code the rub tree time intervals as if the rub tree session follows the last bait site session?)

[jhines]

So, it sounds like you're going with the robust-design model. The recapture-occasions within a primary period are assumed instantaneous and all occur at the same time, so the intervals between secondary periods (rub tree sessions and bait sessions) should be zero, and the intervals between primary periods should be 1 year (in your case).

[ctlamb]

To follow up on this, there is in fact a robust design pradel model. More info on this can be found in the RMark MarkModels.pdf.