Trying to correct for dataset issues in rMARK

[DavidMandell]

I am trying to figure out how to get rMARK to work on a dataset of overwintering birds. Unfortunately, while I've RTFM, I have no prior experience with MARK and the dataset has several issues. I've tried to handle them with assistance from the person I'm working with, but because I'm new to rMARK we wanted to ensure my approaches are sound.

1. The dataset runs over multiple years. Birds were banded on non-capture days for the entire multi-year period; there are relatively few birds that were banded on capture days. To ensure that the marks are captured by the model, I placed a 0 for all birds other than the marked birds on these dates and set p to zero.
2. Effort varies between search periods. I've included effort as a time-varying individual variable using the merge_design.covariates function and will be including it in detection models. Effort has been set for zero on days when birds were only marked but not recaptured.
3. Plots were surveyed on different days with effort differing between plots even on the same day. While I'm currently constructing models with plots separately to make sure I've got the kinks worked out, I want to include them in the same model. I plan on including plot as a variable and setting p to zero for specific plots on specific dates. Effort will be merged by group as well as time, with each group being a different plot.
4. We want to look at survival during the winter and survival for the rest of the year. The person I'm working with suggested that we have time be a two-factor variable rather than a continuous one, setting it to 1 if the time interval is still within a winter and 2 if it's not. There are some capture days during the fall, but if I set p to zero for these days it shouldn't affect the survival model, right?

[jlaake]

Not exactly sure what your questions are other than about fixing p to 0. If capture history value is 0 and p set to 0 then it will not impact survival estimate. When you use merge function that is not an individual time varying covariates but it can be group specific. Individual covariates are stored in the data with the capture history. Using a binary covariate is fine, or you can create a covariate is 0 or 1. The intercept will be for portion of year with covariate value of 0 and term for covariate value 1 will be the difference in survival rates on the logit scale. A continuous time is a slope.Make sure to set time intervals because it sounds like capture occasions occur at different times throughout the year. The unit you choose (eg days, months, proportion of year) will determine what survival rate you are estimating * eg, annual, daily). Rmark was designed for folks that know MARk. It sounds like you have a lot of reading and studying to do.