Models with ad-hoc resightings

[lukeedwyn]

Hello, just interested in people’s thoughts on an appropriate model to use in an analysis. There's a few imperfections with the data that’s got me second guessing myself about the best approach to use, model-wise.

To explain; I have 10-years of open-population data. In this, I have records for over 1100 individuals over the timeframe, with many of them marked during the breeding season as juveniles. The main marking period is usually a window of a few weeks. However, individuals couldn't be recaptured again by any consistent (and appropriate) recapture event, and whilst a very small number are dead-recoveries, resighting records are the main encounter records providing the capture histories for the individuals. The individuals are marked as juveniles at repeated marking sites (two differing habitats) each year. And, they are continuous resightings over the 10-year period – so some individuals are seen regularly, some rarely, some only once again – but resightings are spread throughout the year, which is potentially a problem. The study area is large, and marked individuals from both habitats are resighted regularly within it.

Essentially, the parameters of survival for the differing breeding habitats, with the effect of certain environmental and behaviour factors which is obtained from the spatial information connected to the resightings would be what I’m interested in.

I initially considered using CJS model, where resighting occasions were pooled by year (y1 = t0, y2 = t1... y10 = t11), to give me survival by year. Whilst this would mean dropping a lot of sighting data, I considered would help with issues with model violations around sampling periods with CJS. I also considered splitting resightings to winter and summer resightings, giving me survival for winter/summer. But I wasn’t sure how suitable either these approaches were for estimating survival.
I also considered using a Barker Model too. I don’t have familiarity with this model, but it seems to offer a better alternative to CJS with this type of data. Whilst individuals can be sighted several times in the intervals of capture occasions and I would only use of the fact that it was resighted at least once, I wasn’t sure the most appropriate way to set the encounter occasions to estimate survival when both recaptures aren’t possible and dead recoveries are very minimal.

Thoughts and guidance is appreciated,