Modelling transients and trap-dependence in POPAN

[DArevalo]

Hi,

I am working with a 7-year-CMR data-set (2012-2018) of a Griffon vulture population in a dump site. The animals were trapped using a walk-in cage trap with baits inside. Sessions were carried out two times per month (sometimes >2). Closure assumptions of CJS model were violated when running CloseTest (p<0.0000), so I'm using POPAN model (pooling sampling days into years). As U-CARE GOF test results shows, I'm dealing with transients (Test3.SR: p=0) and trap-dependence (trap-happiness, z=-3.48; Test2.Ct: p<0.0001). All the components of Test3.SR were significants, but only the components 2.C2 and 2.C4 of Test2.Ct were significants. Test3.Sm and Test2.Cl were both not significants (p>0.2).

This GOF results suggest me that my starting model should include both transient and trap-dependence. So, since the PIM structure for POPAN doesn't have the triangle format in both survival and p to parameterize time since marking (TSM), to account transients effect I decided to add the covariate PriorCapL function in survival like this:

B1(Phi) B2(Phi)
(Int) Covariate (PriorCapL)
1 0
1 PriorCapL (1,1)
1 PriorCapL (1,2)
1 PriorCapL (1,3)
1 PriorCapL (1,4)
1 PriorCapL (1,5)

For p, I'm not sure if the format above is correct. I have used it, but the first p parameter estimated is not identifiable (estimate=0.99, SE=0, U-CI=1, L-CI=0).

Could you guys suggest me a way to deal with trap-depence (trap-happiness specifically) using the POPAN model?

Cheers.

[Bill Kendall]

Some of what you've presented is not clear.

First, CloseTest is for a short term closed population study, not long term and is not part of the CJS model.

Second, you say you sampled two or more times per month for seven years, but does that mean 12 months per year?

Third, are all detections based on physical capture and recapture, or do you resight individuals?

Fourth, what demographic parameters are you trying to model?

Cheers,
Bill

[DArevalo]

Thanks for the quick reply. I will be clearer this time.

Ok, I didn't know CloseTest was for short term study. However, the vultures population of the study site does behaves as an open population since it is relatively close to annual nesting sites, and also migrations and deaths occur too. The data-set contains 3,065 physically captured and marked individuals. No resightings.

The sampling sessions varied between and within years and some months were not sampled:
- Days sampled per year: 2012=11, 2013=25, 2014=17, 2015=25, 2016=17, 2017=23 and 2018=13.
- Months sampled per year: 2012=6, 2013=11, 2014=9, 2015=11, 2016=10, 2017=12 and 2018=10.

Our main goal is to estimate abundance per year, so we pooled capture/recaptures sampling days into years to use POPAN model. Since capture probability is important for abundance estimation (right?), I'm trying to fit my initial model correcting for trap-dependence. Can I fit an initial model assessing transients and trap dependence? Can this be done in POPAN or should I use another model?

Thanks for all.