www.phidot.org

Hi,
Due to the ragged nature of my telemetry data, I am using the nest survival model to get survival estimates for my populations. The year is coded to start on April 1 (month 1) and end March 31 (month 12). I can derive annual survival estimates but I also want to examine if survival varies by season (ie - Months 1-3 vs months 4-6, vs months 7-9, vs months 10-12).

I searched through the forums and read the Rmark manual but I didn't find anything pertaining to this. Does anybody know how to get these estimates in Rmark? I can do it in MARK but I am comparing several models for several populations. I would save a ton of time if I can figure out how it is done in Rmark. Thanks in advance for any help, tips, pointers, etc.

Orrin

Look at the design data and create a factor variable that groups the months. Then use that season variable in your model. Plenty of similar examples of that in the documentation for other types of models.

--jeff

Here is some example code you can use with the mallard nest survival data in RMark. Just an example and makes no sense for those data because they are daily data.

Code: Select all

data(mallard)
mm=process.data(mallard,model="Nest",nocc=90)
mm.ddl=make.design.data(mm)
mm.ddl$S$season=cut(mm.ddl$S$Time,breaks=c(0,30,60,90),right=F,labels=c("Fall","Winter","Summer"))
mark(mm,mm.ddl,model.parameters=list(S=list(formula=~season)))


Just to be clear, what you use for time intervals determines whether the rates are annual, monthly or whatever. If you used say 1/12 for intervals between monthly occasions then they are annual survival rates. If you use 1, then they are monthly rates. So for the example above, where the data was daily and time interval=1, the rates are daily and the model specifies a daily rate that is season specific. It doesn't become the prob of surviving the season. To get that, you would need to use S^30.

--jeff

Hi Jeff,

Thank you so much for the help/reply. I am not great at coding yet so it takes me a while to work through things. This helped out tremendously.

Orrin