Multiple Species Occupancy Modeling

[Triciaserow]

Just been wondering if the model for multiple species occuancy has been developed in RMark? I know there's a two-species models available in PRESENCE, is it also available in RMark? What about more than 2 species? Thanks.

Tricia

[jlaake]

Unfortunately it is not in RMark. It requires more work than the standard model because you can share columns of the design matrix across parameters and that wasn't something I considered in the original design of RMark because I wrote it for CJS models originally. I have a kludge of sorts for models like closed capture where p and c can share columns of the DM. What I need to do is to rewrite portions of the code, so the kludge is no longer necessary but unfortunately that is on the back-burner at present.

--jeff

[Triciaserow]

Thanks. I guess all I can do now is run 2 speices at a time in PRESENCE for their interaction/co-occurance. I asked because in McKenzie's book he recommended multiple species models for less than 4 species, so I thought that has been achieved (at least in PRESENCE).

[caique]

Dear all,
Does someone know if the the RMark´s code "2SpecOccup" have been implemented?
According to the code list (at http://www.phidot.org/software/mark/rmark/), 2SpecOccup has been implemented but I can´t run it or find other information.
I´m also interest in multi species occupancy analysis in R.
Any help will be very appreciate.
thanks,
Carlos

[jlaake]

When this thread was started that model was not in RMark. It is there now but it is 2SpecConOccup and not 2SpecOccup. I believe the conditional approach is preferred. Unfortunately I don't have an example for it in the documentation. If you have one and want to share it, I'll include it in the documentation. Look at markmodels.pdf for the names of the parameters and use the MARK documentation for there definitions.

For direct R implementations of some occupancy models you may want to look at the R pacakge unmarked. Also, there are threads under Presence that give R code for fitting some models directly in R. It is not particularly difficult but there are some tricks that help make it more general as I describe in one of the threads.

--jeff

[caique]

Dear Jeff,
Many thanks for your replay.
I´ve just upgraded the Rmark from CRAN but I´was still unable to find the new 2SpecConOccup function. Let me know if I didn´t catch anything else, please.
As I could understand, the conditional approach are considered more stable than the first proposal (Mackezie et al., 2004 e 2006) with 2 recently parametrization alternatives (Richomond et al., 2010 and Waddle et al., 2010).
The Waddle et al.´s parametrization sounds me more parsimonious. Nevertheless, I couldn´t find more deeper approach about it and it was hard to compare properly these options. It would be very interesting to read more comments or suggestion about those parametrization in these forum.
At the momento, I´m using the Richomond et al.´s parametrization in Presence. As long as I know it is also available in Mark, but I haven´t used. I was just wondering to use it throughout Rmark.
The unmarked R´s packages has not function for species interactions. The Waddle et al.´s parametrization was available (as supplement) for WinBugs with R code and it could be an option as well as Richmond et al.´s data for example documentation. It was just some suggestion and for sure I can share my case.
Cheers,
Carlos

[jlaake]

It is not a function. It is one of the values of the argment "model" for the "mark" function which with related functions constructs models for the program MARK, runs MARK and then harvests the output. Have you used RMark before? If not you need to read the documentation that comes with it.

Your post is largely lost on me because I programmed the model into RMark and that requires very little knowledge about the model because MARK does all the work. RMark is only a tool that uses the R facilities to build models for MARK. Also, I'm not that familiar with all the occupancy literature as I don't use it for my work.

regards --jeff

[caique]

Dear Jeff,
Such silly mistake, sorry.
the last few months I was working hard with unmarked package where there are diferents functions for every model. It was on the top of my head and I really forgot how deal with Rmark, sorry. Now it is much clearer, thanks.
I´ll check the 2SpecConOccup.
Thank you very much and congratulation for your work.
regards,
Carlos

[Triciaserow]

I've been following in this post since I originally asked the question. So glad to see this model has been implanted in R. I've already run some Two Species Conditional Occupancy models. Thank you, Jeff. . I'll probably come up with more questions when I run complicated models.