Random effects in single-season occupancy model

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

Random effects in single-season occupancy model

Postby ms2018 » Tue Oct 30, 2018 3:29 pm

Hi all,

New to this forum and to using RMark! I am interested in running several single-season occupancy models within RMark due to its ability to incorporate random effects, but am having a hard time finding any example code or solid information on this topic. I have read through the MARK handbook and am now unsure if RMark can even do random effects? Thoughts/suggestions??
Mira
ms2018
 
Posts: 7
Joined: Sun Apr 22, 2018 6:05 pm

Re: Random effects in single-season occupancy model

Postby cooch » Tue Oct 30, 2018 3:43 pm

ms2018 wrote:Hi all,

New to this forum and to using RMark! I am interested in running several single-season occupancy models within RMark due to its ability to incorporate random effects, but am having a hard time finding any example code or solid information on this topic. I have read through the MARK handbook and am now unsure if RMark can even do random effects? Thoughts/suggestions??
Mira


If MARK can do random effects (which it can), then so can RMark. All you want is 'Occupancy Estimation with Detection <1 and Random Effects'. It isn't explicity documented in the occupancy chapters, since that data type is simply a combination of SS occupancy models (chapter 21) and random effects models (in the context of 'single-season abundance estimation' derivatives -- like occupancy) are discussed in Chapter 14.
cooch
 
Posts: 1628
Joined: Thu May 15, 2003 4:11 pm
Location: Cornell University

Re: Random effects in single-season occupancy model

Postby jlaake » Tue Oct 30, 2018 3:54 pm

Not all models in MARK are supported in RMark but can be added. Check Help Data Types in MARK to see if it is supported in RMark. Let me know if it isn't and I'll add it.
jlaake
 
Posts: 1417
Joined: Fri May 12, 2006 12:50 pm
Location: Escondido, CA

Re: Random effects in single-season occupancy model

Postby ms2018 » Tue Oct 30, 2018 4:06 pm

jlaake wrote:Not all models in MARK are supported in RMark but can be added. Check Help Data Types in MARK to see if it is supported in RMark. Let me know if it isn't and I'll add it.


I didn't see it in the Data Types, I guess that's why I am a little lost!
ms2018
 
Posts: 7
Joined: Sun Apr 22, 2018 6:05 pm

Re: Random effects in single-season occupancy model

Postby cooch » Tue Oct 30, 2018 4:06 pm

jlaake wrote:Not all models in MARK are supported in RMark but can be added. Check Help Data Types in MARK to see if it is supported in RMark. Let me know if it isn't and I'll add it.


Which of course is the intended segue to the observation that this model is not currently supported in RMark:

Code: Select all
151. $_OccupancyRE: Occupancy Estimation with Detection < 1 and Random Effects     


The lack of the '*' indicates it isn't supported in RMark (the $ means you you can simulate under this data type, and the underscore _ indicates that dots are allowed in the encounter histories.
cooch
 
Posts: 1628
Joined: Thu May 15, 2003 4:11 pm
Location: Cornell University

Re: Random effects in single-season occupancy model

Postby ms2018 » Tue Oct 30, 2018 4:13 pm

Thank you! So just to clarify- it's currently not supported in RMark but can be added..? I've been doing all my occupancy modeling in unmarked for a while now, and am just deciding to switch to RMark and I'm a little intimidated by the prospect of using MARK.
ms2018
 
Posts: 7
Joined: Sun Apr 22, 2018 6:05 pm

Re: Random effects in single-season occupancy model

Postby cooch » Tue Oct 30, 2018 5:58 pm

ms2018 wrote:Thank you! So just to clarify- it's currently not supported in RMark but can be added..?


Yes -- provided you can take Jeff out of retirement long enough to implement it. ;-)

I've been doing all my occupancy modeling in unmarked for a while now, and am just deciding to switch to RMark and I'm a little intimidated by the prospect of using MARK.


There is a basic hierarchy -- both MARK, and Presence (and RPresence) have by far the more complete set of capabilities for ocupancy modeling (unless, of course, you're among the group that wants to code everything from the ground up in JAGS or the equivalent). unmarked is a very clever piece of software, but in some respects, is pretty limited in what it can do (largely because there is little -> no active development of unmarked).

The learning curve for MARK is perhaps a bit steeper than for Presence, simply because MARK is a more complete software (which does not mean it is better, simply, that MARK was designed to do more things than occupancy analysis, whereas Presence is focussed exclusively on occupancy analysis). The learning curve for MARK is perhaps attenuated somewhat by the fair bit of 'teach yourself' documentation that is out there.

I get asked with some frequency 'if I had to pick one software application to learn for occupancy modelling', which one should I pick?'. My usual recommendation is Presence (and now, especially, RPresence) since it tends to be a bit earlier standing on the bleeding edge of what you can do with occupancy models. MARK usually follows up fairly quickly (meaning, if a new model appears in the literature at time T, its in Presence at time T+1, and in MARK at time T+2 or T+3 months later, sooner if its something that one of the MARK 'inner circle' might need to use for their work). However, if you're going to do more than occupancy modelling at some point, then the steeper learning curve for MARK might be worth the effort. I'm sure Darryl or Jim will weigh in if they have more informed opinions.
cooch
 
Posts: 1628
Joined: Thu May 15, 2003 4:11 pm
Location: Cornell University

Re: Random effects in single-season occupancy model

Postby ms2018 » Tue Oct 30, 2018 11:28 pm

cooch wrote:Yes -- provided you can take Jeff out of retirement long enough to implement it. ;-)

Retirement isn't all it's cracked up to be... right? :lol:

cooch wrote:I get asked with some frequency 'if I had to pick one software application to learn for occupancy modelling', which one should I pick?'. My usual recommendation is Presence (and now, especially, RPresence) since it tends to be a bit earlier standing on the bleeding edge of what you can do with occupancy models.

My issue here is that my models very desperately need the option to include site as a random effect, and I was under the impression that MARK and ESURGE are the only softwares that do it?
ms2018
 
Posts: 7
Joined: Sun Apr 22, 2018 6:05 pm

Re: Random effects in single-season occupancy model

Postby cooch » Thu Nov 01, 2018 3:58 pm

My issue here is that my models very desperately need the option to include site as a random effect,


...which, in my experience, means someone has told you that this is what you should do...without a lot of motivation.

and I was under the impression that MARK and ESURGE are the only softwares that do it?


Neither, in fact. MARK lets you handle heterogeneity in p as an individual random effect (using Gauss-Hermite quadrature). Coding site as a random effect will mean going the Bayesian/JAGS route. I'd bet a fair bit that there is a script for same in the AHM books that Andy Royle and Marc Kery have put out recently.
cooch
 
Posts: 1628
Joined: Thu May 15, 2003 4:11 pm
Location: Cornell University

Re: Random effects in single-season occupancy model

Postby jlaake » Thu Nov 01, 2018 5:49 pm

But isn't site = individual in occupancy? So if you are looking for a random effect in p, I can add it to RMark. Let me know.
jlaake
 
Posts: 1417
Joined: Fri May 12, 2006 12:50 pm
Location: Escondido, CA

Next

Return to RMark

Who is online

Users browsing this forum: Majestic-12 [Bot] and 14 guests