www.phidot.org

Hi everyone,
I'm working with a large dataset of ringed nightjars. A few animals were ringed at the beginning of the study in 2014, but more intensive data collection started a few years later in 2017. Assuming I want to start modeling data from 2017, what is the best way to incorporate previously marked individuals to the dataset?

My second question is partially related but more specific to the Barker model I'm trying to fit. Because the number of recapture occasions varied, the interval between occasions was unequal and we have occasional resightings and dead recoveries we decided to split artificially our dataset to fit a Barker model.
Basically we considered every first week of each month our recapture session and the interval between occasions our "incidentals". However, because the way we have split the data we do have first captures between "capture sessions". What is the best way to incorporate these individuals to the dataset?
Can I assign a "2" during the incidental period if an individual was never formally incorporated in the recapture sessions? I guess not but I was wondering what the best alternatives are.

Thanks,

Alejandro

You're right that your two questions are related. The Barker model does not allow initial detections to be auxiliary resightings (i.e., a '2' in the D column). The first encounter needs to be during a capture occasion (in the L column).

If that eliminates a lot of information, it is possible to treat those initial detections as captures and scale the survival parameter from initial "release" to the next capture occasion, using the time from that initial detection to the next capture occasion as an individual covariate. Because an initial detection during the survival interval would also affect the probability of subsequent auxiliary resight during that interval (e.g., an initial encounter halfway through the interval would reduce the time available for another auxiliary sighting during that interval), you would probably also need to scale R and R' using that covariate.

Thanks Bill for the prompt reply.
How exactly are you suggesting to scale the R's? Is it something that can be coded directly into Mark ?

And an additional question: assuming I want to compare this to a CJS (Obviously losing some info). How would you recomend to add previously marked individuals to a dataset? By adding a covariate with time since marked for those individuals?

Thanks again,

Alejandro