Robust design and individual covariates

questions concerning analysis/theory using program MARK

Robust design and individual covariates

Postby RSAWhite » Thu Dec 19, 2013 6:53 pm

Hello,

I am interested in running an robust design analysis of live encounter data that fits individual covariates to survival, immigration, emmigration, recruitment, and capture probability parameters within the same robust model in program MARK. However, I cannot find a robust-design model in MARK that can do this.

Is there a robust-design model in program MARK that would fulfill my needs?

Thank you in advance.

Sincerely,
Richard.
RSAWhite
 
Posts: 10
Joined: Fri May 24, 2013 10:47 pm

Re: Robust design and individual covariates

Postby cooch » Thu Dec 19, 2013 7:16 pm

RSAWhite wrote:Hello,

I am interested in running an robust design analysis of live encounter data that fits individual covariates to survival, immigration, emmigration, recruitment, and capture probability parameters within the same robust model in program MARK. However, I cannot find a robust-design model in MARK that can do this.

Is there a robust-design model in program MARK that would fulfill my needs?

Thank you in advance.

Sincerely,
Richard.


Any model in MARK can handle individual covariates, which is why there isn't a specific 'hard-coded' version of the RD (or anything else). You need to read through chapter 11 for details on how to handle individual covariates:

http://www.phidot.org/software/mark/doc ... chap11.pdf

The reason individual covariates are covered in the book before more advanced data types (say, RD) is because what it discusses (handling individual covariates, specifically by adding them as appropriate to the design matrix) can be applied to any successive method/chapter.

Having said that, I can tell you (warn you, perhaps better), that ading individual covariates to even moderately general (time-specific) RD models is going to be a pretty involved undertaking. Doable in theory, verging on painful in practice (yes, even if you use RMark...).
cooch
 
Posts: 1654
Joined: Thu May 15, 2003 4:11 pm
Location: Cornell University

Re: Robust design and individual covariates

Postby RSAWhite » Thu Dec 19, 2013 8:16 pm

Ok, thanks Cooch,

I have read chapter 11 and have used individual covariates frequently in CJS and JS models. So, my question is not about how to model ind covs (sorry for the confusion), but rather that I cannot find robust models where I can apply individual covariates to both immigration/emmigration and recruitment parameters within the same model.

The "Robust Design" model type has immigration and emmigration parameters, but no recruitment parameter, whilst the "Pradel Models Including Robust Designs" model type has the recruitment parameter, but no immigration/emmigration parameters. Whilst I could run two separate models to look at the effect of ind covs on recruitment, and then immigration/emmigration, I would prefer to analyse recruitment+immigration+emmigration within the same model.

Is there a way to do this?

Cheers,
Richard.
RSAWhite
 
Posts: 10
Joined: Fri May 24, 2013 10:47 pm

Re: Robust design and individual covariates

Postby cooch » Thu Dec 19, 2013 8:43 pm

RSAWhite wrote:Ok, thanks Cooch,

I have read chapter 11 and have used individual covariates frequently in CJS and JS models. So, my question is not about how to model ind covs (sorry for the confusion), but rather that I cannot find robust models where I can apply individual covariates to both immigration/emmigration and recruitment parameters within the same model.


No such model exists (regardless of whether or not you use individual covariates, or whether you use MARK or something else). At best, you can get pieces of the full BIDE model. For example, take a Pradel model, where $\lambda=f+\phi$. You can estimate any two of three parameters, but because they are linearly related, you can only put 2 of 3 in the likelihood -- the third is estimated by algebra (so called derived parameters). So, there are Pradel models in MARK for most of the 2 parameter combinations.

While there are ways to separate out different factors (e.g., in situ recruitment vs recruitment due to immigration), there is no single 'canned' model that gives you everything you want.
cooch
 
Posts: 1654
Joined: Thu May 15, 2003 4:11 pm
Location: Cornell University

Re: Robust design and individual covariates

Postby RSAWhite » Thu Dec 19, 2013 9:20 pm

Excellent, thank you. Could you please elaborate on some methods to separate in-situ recruitment vs immigration recruitment? Or suggest resources that describe how?

Cheers,
Richard.
RSAWhite
 
Posts: 10
Joined: Fri May 24, 2013 10:47 pm

Re: Robust design and individual covariates

Postby cooch » Fri Dec 20, 2013 8:10 am

RSAWhite wrote:Excellent, thank you. Could you please elaborate on some methods to separate in-situ recruitment vs immigration recruitment? Or suggest resources that describe how?

Cheers,
Richard.


Sure:

Nichols, J. D., and K. H. Pollock. 1990. Estimation of recruitment from immigration vs. in situ reproduction using Pollock’s robust design. Ecology 71:21–26.

One application of same can be found in Appendix C of the following:

http://canuck.dnr.cornell.edu/research/ ... _paper.pdf
cooch
 
Posts: 1654
Joined: Thu May 15, 2003 4:11 pm
Location: Cornell University

Re: Robust design and individual covariates

Postby RSAWhite » Sun Dec 22, 2013 3:48 pm

Excellent, thank you very much for your help!
RSAWhite
 
Posts: 10
Joined: Fri May 24, 2013 10:47 pm

Re: Robust design and individual covariates

Postby RSAWhite » Sun Dec 22, 2013 8:02 pm

Hello,

If I ran a "Pradel survival and recruitment" model in MARK and modelled the apparent recuitment parameter for "young" and "adult" age classes separately, would it be appropriate to consider the young recruitment probability as in-situ recruitment, and the adult recruitment probability as the immigration recruitment?

Thanks again,
Richard.
RSAWhite
 
Posts: 10
Joined: Fri May 24, 2013 10:47 pm

Re: Robust design and individual covariates

Postby cooch » Sun Dec 22, 2013 9:08 pm

RSAWhite wrote:Hello,

If I ran a "Pradel survival and recruitment" model in MARK and modelled the apparent recuitment parameter for "young" and "adult" age classes separately, would it be appropriate to consider the young recruitment probability as in-situ recruitment, and the adult recruitment probability as the immigration recruitment?

Thanks again,
Richard.


There are no age-based Pradel models (although Nichols has done some work in this area, and Jean-Dominique Lebreton has, I've been told, 'worked it all out', but not published it). The PIM is a single row -- no age-specificity possible.
cooch
 
Posts: 1654
Joined: Thu May 15, 2003 4:11 pm
Location: Cornell University

Re: Robust design and individual covariates

Postby bacollier » Tue Dec 24, 2013 5:38 pm

RSAWhite wrote:Hello,

If I ran a "Pradel survival and recruitment" model in MARK and modelled the apparent recuitment parameter for "young" and "adult" age classes separately, would it be appropriate to consider the young recruitment probability as in-situ recruitment, and the adult recruitment probability as the immigration recruitment?

Thanks again,
Richard.


Agreeing with what Evan said above about Pradel models. If you have a species where you can classify, at least to a point, new births from adults that either 1) reside or 2) immigrate in (e.g., a migrant population where the only thing that can be there is adults at least for a while), then you can use the POPAN modeling approach to estimate in situ vs. adult 'entrance' into the population. See Journal of Fish and Wildlife Management 4:33-40 where, I did something similar with urban white-winged doves separating in situ recruitment from immigration of adults. In hindsight, I would have use the detailed molt data to also classify HY birds that were likely immigrants as well, but, you always learns something. But the approach might work for you (albeit I have no idea what you are working on).

\bret
bacollier
 
Posts: 231
Joined: Fri Nov 26, 2004 10:33 am
Location: Louisiana State University

Next

Return to analysis help

Who is online

Users browsing this forum: No registered users and 0 guests