seniority in RMark and Mark

posts related to the RMark library, which may not be of general interest to users of 'classic' MARK

seniority in RMark and Mark

Postby vin » Tue Sep 30, 2008 10:44 am

I have a 53 occassion data set with n=1210
The animals, rodents were trapped monthly for four years on two seperate grids.
When I run a fully time dependent model in RMark (i.e: grid x time for both capture en seniority) this model does not convergence.
When I do exactly the same in Mark it does converge, allthough not all parameters have a reasonable estimate?

1. Anyone who experienced a similar problem?
2. What is the difference between way RMark 'develops' the model and Mark?

thanks in advance for any help and/or suggestions

note that I assume the Pradel option in RMark, described as Pradel recruitment only is the same as the Pradel option in Mark described as Pradel seniority only

Vincent


Ps this is the error message from RMark

Gamma.gxt.p.gxt
Error in if (x4 > x2) { : argument is of length zero

PPS: this is the code I use in RMark

steptime=function(){
Gamma.gxt=list(formula=~grid*time)
p.gxt=list(formula=~grid*time)
cml=create.model.list("Pradel")
result=mark.wrapper(cml,data=subprad.process,ddl=subprad.ddl)
return(result)}

steptime.result=steptime()
vin
 
Posts: 6
Joined: Tue Nov 27, 2007 11:21 am

RMark models

Postby jlaake » Wed Oct 01, 2008 12:10 pm

I suggest that you may want to spend sometime reading Appendix C in Cooch and White to get a full grasp of RMark and what it does. In the appendix on pg 2-3 is a list of supported models where you could have determined that Pradel is the seniority formulation. Also, you would see on page 4 that models in RMark are done via design matrices rather than PIM coding except that some models can now be specified via PIM coding and use the sin link. My guess is that "exactly the same" in MARK is not and that you used the PIM coding in MARK with the sin link and the DM with the logit link with RMark. You can specify formula=~-1+grid:time,link="sin" in RMark to build a PIM coded model and you can also select the DM with logit link from MARK as one of the Pre-defined models. However even if you do the latter, there are many ways to create a DM and MARK and RMark use slightly different approaches and when the model is over-parameterized (as it appears to be in this case) then you can get differences.

--jeff
jlaake
 
Posts: 1417
Joined: Fri May 12, 2006 12:50 pm
Location: Escondido, CA

Postby vin » Thu Oct 02, 2008 7:13 am

thanks for your reply

I indeed specified the logit function in Mark, but wasn't aware the DM versus PIM coding could have an effect as well.

Vincent.
vin
 
Posts: 6
Joined: Tue Nov 27, 2007 11:21 am

Postby vin » Thu Oct 02, 2008 7:32 am

and indeed the appendix does specify the Pradel model correctly, I must have got confused by the RMark help file (?mark) where it says Pradel recruitment only.

I apologize for that,

Vincent
vin
 
Posts: 6
Joined: Tue Nov 27, 2007 11:21 am

Pradel models

Postby jlaake » Thu Oct 02, 2008 12:25 pm

I can understand your confusion. It always confuses me as well. The model names in RMark correspond to those that MARK uses and they are as follows:

Pradel, Pradsen, Pradlambda and Pradrec

Their descriptions in the MARK help file are listed below in the order of the model names above. As you can see the Pradel model is called the Pradel recruitment only model but has the seniority and gamma parameters. So both sources are correct -- just confusing. In that regard if you find discrepancies between the help files and the appendix, let me know. It is hard to keep up with all the files. The most current info is typically the Appendix and the What's new in RMark file. Some of the material in the latter has not made it's way to the help files yet.

On another point, RMark is simply an interface to MARK. They both use the same mark.exe for estimation. RMark is simply a tool for building models. All of the model fitting is done by Gary's MARK program. If there is a discrepancy then it is a difference in how the models were constructed. You can compare the input and output files created via the MARK and RMark interfaces to see where those differences occur. Differences can occur from the link function and the DM construction because there are many different ways to specify the same model with different DMs. If the data are sufficient, the result should be the same but sometimes you can get lack of convergence or small numerical differences. The biggest differences can occur based on the choice of the link function.

regards --jeff



Pradel Recruitment Only Model. Pradel (1996) developed a model to estimate the proportion of the population that was previously in the population. Thus, this model, labeled 'Pradel Recruitment Only', estimates recruitment to the population. The parameters of this model are the seniority probability, gamma (probability that an animal present at time i was already present at time i - 1), and recapture probability r. Only LLLL encounter histories are required for this model. This model can be estimated by reversing the time sequence of the live encounter histories (Pradel 1996), an idea suggested by Pollock et al. (1974:85-85), and even mentioned by R. A. Fisher in about 1939 or so (Box ????).



Pradel Survival and Seniority Model. Pradel (1996) extended his recruitment only model to include apparent survival (phi). In MARK, this model is labeled 'Pradel Survival and Seniority'. Parameters of the model are apparent survival (phi), recapture probability (p), and seniority probability (gamma), which is the probability that an animal in the population at time i was also in the population at time i - 1 (i.e., the animal did not enter the population during the interval i - 1 to i. Only LLLL encounter histories are required for this model.



Pradel Survival and Lambda Model. Pradel (1996) also parameterized his model with both recruitment and apparent survival to have the parameters apparent survival (phi), recapture probability (p), and rate of population change [lambda(i) = N(i + 1)/N(i)]). This model converges quite readily compared to the Burnham parameterization of the Jolly-Seber model described above. Only LLLL encounter histories are required for this model.



Pradel Survival and Recruitment Model. Pradel (1996) also parameterized his model with both recruitment and apparent survival to have the parameters apparent survival (phi), recapture probability (p), and fecundity rate [f(i) = number of adults at time i + 1 per adult at time i]. This model converges quite readily. Only LLLL encounter histories are required for this model.
jlaake
 
Posts: 1417
Joined: Fri May 12, 2006 12:50 pm
Location: Escondido, CA


Return to RMark

Who is online

Users browsing this forum: No registered users and 18 guests

cron