Known-fate RMark for nest survival

[wjensen1]

Reintroducing my lab/myself to iButton temperature data loggers for estimating nest survival. Intended to use previous RMark code from standard nest monitoring data when checks were every 2 to 4 days (no data loggers, the 'old-fashioned way' with midpoint completion estimation, etc.). Original thought was retooling the LastPresent and LastChecked dates in RMark to the nest completion dates we know from nest vs. ambient temperature data, using the process.data(... model=”Nest”) approach. All estimates were = 1, which is obviously incorrect. There seemed room for a known-fate modeling approach this has apparently been used for iButton nest data analyzed in RMark (e.g., Weintraub et al. 2016, Condor). Others have used the Shaffer (2004) logistic exposure SAS code for such data. Does anyone happen to have any on R script/suggestions for such known-fate nest survival models? I’m not finding anything explicitly relevant in RMark documentation. Thanks for any help.

[jlaake]

Seems like you need to contact Weintraub and ask them or look to see if they posted their code as a electronic supplement which I highly encourage. There is an example of known fate data in RMark (?Blackduck). Other than that this seems more like a question of how to analyze your data as known fate and not really an RMark question.

[wjensen1]

Thanks, Jeff. Will contact that lead author (Weintraub). They wrote "We modeled the daily survival rate (DSR) of Tricolored Blackbird nests using known-fate models with the package RMark (Laake 2012) in R 2.15.1..." (Weintraub et al. 2016, Condor 118, p. 854). So, seems RMark should work? (But no supplementary material.) Thanks for the ?Blackduck tip.

[cooch]

Been a while since I looked at it, but you might want to look at Jay Rotella's notes and code: https://www.montana.edu/rotella/502/Schedule.html

Jay wrote the 'nest survival' chapter for the MARK book, but has R and RMark code for lots of things, including known-fate.

Of course, you could use GUI-based MARK, and be done with it by the time you've read Jay's notes, but I understand the desire to 'do everything in R'.

[wjensen1]

Thanks for the tips, Evan!