www.phidot.org

Hello,

We sampled our populations 2-3 times within a month long time frame; however it is possible that there were deaths in that month. I am using polygon detectors to estimate density and am wondering if and how I can adjust the coding to accommodate for the possibility that not all individuals survived. Is there documentation that you might refer me to? Or would just switching from coding them as occasions to coding them as sessions accomplish this?

Thank you!!

I doubt this is a big problem, although it depends on a lot of things we don't know like how many die and how much data you have. As usual, simulation can give insight into the effect of breaching the closure assumption. I can't think of specific documentation that would help. Certainly you can treat 'occasions as 'sessions', and this might detect a big effect if you have enough data, but (i) you will lose recaptures, which may impair precision, and (ii) expect the reported precision of D-hat to be inflated (unrealistically small SE) because the sessions are not really independent (if it really mattered, better SE for the multi-session estimate could be got by simulating i.e. parametric bootstrap). Ultimately you could apply an open-population model, but that is not available in canned form as far as I know (maybe later next year).
Murray