unequal occasions & time intervals across study areas

[holtfrw]

All,

I'm working with camera sight and re-sight data collected over an 8 month period in separate study areas. In the first study area we conducted 5 surveys with an average length of time between surveys of 30 days. In the 2nd, we conducted 8 surveys with an average of 19 days between surveys.

To date, I've run separate analyses for each study area using Pradel RD models, but I'd like to combine my data to determine if there was a study area effect in terms of survival and recruitment. Given that I have a different number of encounter occasions and different time intervals between occasions across study areas, how can I set up my analysis to include both study areas?

Any help would be appreciated.

Thanks,

Rob

[sbonner]

Hi Rob,

One way to do this is to combine the data so that all occasions combined from all sites are included in the histories for all individuals and then to set capture probabilities to 0 for sites-by-occasions that weren't sampled. Suppose that you had two sites with alternating days of capture from July 1--10 so that captures were conducted at site 1 on the 1st, 3rd, 5th, 7th, and 9th and at site 2 on the 2nd, 4th, 6th, 8th, and 10th. Analysing the data separately would give you 5 days of capture at each site, but you could combine the data to have 10 capture occasions at each site and then set p1=p3=p5=p7=p9=0 at the first site and p2=p4=p6=p8=p10 at the second.

Note that you will encounter some issues with identifiability that you will need to deal with. For example, if you allowed site and time specific survival probabilities in the above model then you would not be able to separate phi1 from phi2 at site 1. The easiest thing to do is to assume that phi1=phi2 or to model the survival probabilities per unit time if the intervals between the capture occasions are not equal.

Hope this helps,

Simon

[holtfrw]

Simon,

That makes perfect sense. Thanks for your help.

Rob