short-capture histories in POPAN model

questions concerning analysis/theory using program MARK

short-capture histories in POPAN model

Postby foxfire235 » Mon Oct 22, 2007 10:21 pm

Hi all -

I have a 3-year capture history (though quite a lot of caps, recaps). 8 groups (sex-age classes).

If the final survival and capture parameters are confounded, and the initial pent(a.k.a., b) and capture parameters are confounded..... this leaves me with only models that vary by group, as I can only estimate one p, one Phi, and one pent.

I have a few questions:

1. Since all pents (across time) should add to one, does this mean my single estimable pent should be fixed at "1" for all groups? Or should I just have pent(.) and let it estimate itself? Do I need to use MLOGIT as a link, as there is not time variation necessary to constrain?

2. To derive a population size, is it then merely N = N_initial(Phi)+ N_initial*pent, which = N_initial(Phi)+N_initial?

3. Lastly, my Phi's seem quite high (in the 0.90's in all groups), though in the field in my last year of marking, 45% of the animals were already marked, so >90% seems high. Can anyone think of what may be going on here?
foxfire235
 
Posts: 3
Joined: Tue Mar 07, 2006 12:50 pm
Location: Reno, NV

short-capture histories in POPAN model

Postby cschwarz@stat.sfu.ca » Mon Oct 22, 2007 11:04 pm

With only 3 capture times, there is very little that can be estimated unless restrictions can be placed across parameters (e.g. equal p across time etc).

As you point out, in the most general model p(t), phi(t), pent(t) only p(2), phi(1), and pent(2) can be "estimated" without confounding (but the interpretation of the the pents will be odd because of the restrictions placed on the other pents.

If you have confounded parameters, you normally set one of the pair to 1 to estimate the other parameter in the pair. I find setting p(1) and p(3) to 1 seems to "work best".

You still need the MLOGIT link to force the "pents" to sum to 1.

The population size at time 2, N(2) should come out as a derived parameter and you don't have to do any computations. Only N(2) has a sensible meaning.

Without information on what species is being monitored, what design is being used, it is difficult to say what, if any, problem is occuring with your phis.

I notice that your 8 groups are defined as sex x age combinations. But as animals get older, the age classes change. How do animals switch between age classes in your model - if each group is an age class, then animals must stick in their age class. You may wish to reconsider the model being fit to your experiment??

Carl Schwarz
cschwarz@stat.sfu.ca
 
Posts: 43
Joined: Mon Jun 09, 2003 1:59 pm
Location: Simon Fraser University


Return to analysis help

Who is online

Users browsing this forum: No registered users and 3 guests

cron