Using PcountOpen for multiple w/in year primary sampling periods

[james hansen-nr]

Dear Unmarked group,

I have some questions about the applicability of the PcountOpen function that I'm hoping you can help answer. 

The following is a brief description of the data that I am working with:

 2 years of secretive marsh bird point count data where sites are visited 2-3 times per primary sampling period (2-week survey window), with multiple primary sampling periods across a 1.5-2 month time frame within each year. The surveys also contain data that can be used for time-removal sampling (11 1-minute segments for each visit) and distance sampling. I'm mostly focused on just using the repeated counts data for this analysis, but it would be nice to include a time-removal component as well. 

Year 1: 18 sites, 6 primary sampling periods (2-week survey windows), 2-3 secondary sampling periods (day/visit).

Year 2: 36 sites (retaining the same 18 sites from year one), 4 primary sampling periods, 2-3 secondary sampling periods. Site-level and visit level covariates collected for each year.

My main two objectives are to 1) investigate what factors affect detection probability and abundance for these species, and 2) illustrate that the 2 month survey season is open to immigration and emigration and provide sampling recommendations to avoid sampling during open periods and reduce variation in estimates of abundance (telemetry data that I'm also collecting suggests that the sampled population is highly open to emigration across the 2 month survey season).

Instead of using 'year' as the primary sampling period with the PcountOpen function, I was hoping to use the 2-week survey windows as primary sampling periods in order to allow for a level of 'open-ness' between survey windows and still evaluated the parameter effects on detection and abundance across the whole 2 month survey period. Ideally it would be great to use the closure test in the Dail-Madsen 2010 paper to illustrate periods of non-closure, however as far as I know that currently isn't implemented in unmarked. 

My main questions are:

1) Would it be appropriate to use the 2-week survey windows as primary sampling periods in PcountOpen?

2) Are my data robust enough to be modeled using the PcountOpen function?

3) Would it be better to instead use the gpcount function separately for each 2-week survey window to account for temporary emigration within each survey window (I'm not sure if the gpcount function would be appropriate since the population is open to both movement and demographic processes)?

4) Considering the low sample size, would it be better to 'stack' the survey windows and model the site/survey window combinations as different sites instead of using the multi-season approach?

5) Is the Dail-Madsen closure test currently able to be implemented in Jags or another software?

I'm not terribly familiar with Unmarked, I've mostly been running relatively simple single-season binomial N-mixture models in Jags following examples from the Kery and Royle AHM book and the Kery and Schaub BPA book, so if anyone knows of some Jags code for the Dail-Madsen method that'd be great as well (I haven't yet posted this to the HMecology group for the Jags side of things).

The sampling design is fairly complex, and I do realize there are some issues with independence and non-closure even within the 2-week survey windows, but I'm mostly concerned with accurately assessing breeding season abundance and investigating how various factors affect abundance and detection probability.

Any help or advice with this would be greatly appreciated.

Kind regards,

Jim Hansen