predict community by community and environment

classic Classic list List threaded Threaded
4 messages Options
Reply | Threaded
Open this post in threaded view
|

predict community by community and environment

mailbox Franz
Hi all,

I am searching for a method to test weather a community is better predicted by an other community or by the environment in a mantel-test like framework?
Any help would be appreciated. Platform preferably R.

All the best,
Franz
_______________________________________________
R-sig-ecology mailing list
[hidden email]
https://stat.ethz.ch/mailman/listinfo/r-sig-ecology
Reply | Threaded
Open this post in threaded view
|

Re: predict community by community and environment

Erick LeBrun
Hello Franz,

One possibility would be to do ordinations (preferably NMDS) and then do a Procrustes analysis. It is touted to be potentially more sensitive than a mantel test and using the protest command, you can get a correlation statistic and significance associated with the two datasets. The simplified explanation is that it scales and rotates the ordinations to the best fit possible. You can also view error plots that show you “movement” of the points in ordination space that can show trends or differences.

Best regards,
Erick

Erick LeBrun
PhD Student, Department of Biology
Kang Microbial Ecology Laboratory
[hidden email]

On 10/18/16, 8:35 AM, "R-sig-ecology on behalf of mailbox Franz" <[hidden email] on behalf of [hidden email]> wrote:

    Hi all,
   
    I am searching for a method to test weather a community is better predicted by an other community or by the environment in a mantel-test like framework?
    Any help would be appreciated. Platform preferably R.
   
    All the best,
    Franz
    _______________________________________________
    R-sig-ecology mailing list
    [hidden email]
    https://stat.ethz.ch/mailman/listinfo/r-sig-ecology
   

_______________________________________________
R-sig-ecology mailing list
[hidden email]
https://stat.ethz.ch/mailman/listinfo/r-sig-ecology
Reply | Threaded
Open this post in threaded view
|

Re: predict community by community and environment

Leonardo Ré Jorge
Dear Franz,

An approach that should fit quite well to your intention is Generalized
Dissimilarity Modellling (GDMs - Ferrier 2007:
http://doi.wiley.com/10.1111/j.1472-4642.2007.00341.x). You can model the
compositional dissimilarities among your communities relative to both
environmental differences and differences in composition of another
taxonomic group. This has also the advantage of allowing you to include the
spatial distance among your communities as a predictor, accounting for
spatial autocorrelation. Your final output should give you a measure of
relative effects of each predictor. GDMs are implemented in the R package
gdm.

Best regards,
Leonardo

On Tue, Oct 18, 2016 at 12:54 PM, LeBrun, Erick <[hidden email]>
wrote:

> Hello Franz,
>
> One possibility would be to do ordinations (preferably NMDS) and then do a
> Procrustes analysis. It is touted to be potentially more sensitive than a
> mantel test and using the protest command, you can get a correlation
> statistic and significance associated with the two datasets. The simplified
> explanation is that it scales and rotates the ordinations to the best fit
> possible. You can also view error plots that show you “movement” of the
> points in ordination space that can show trends or differences.
>
> Best regards,
> Erick
>
> Erick LeBrun
> PhD Student, Department of Biology
> Kang Microbial Ecology Laboratory
> [hidden email]
>
> On 10/18/16, 8:35 AM, "R-sig-ecology on behalf of mailbox Franz" <
> [hidden email] on behalf of [hidden email]>
> wrote:
>
>     Hi all,
>
>     I am searching for a method to test weather a community is better
> predicted by an other community or by the environment in a mantel-test like
> framework?
>     Any help would be appreciated. Platform preferably R.
>
>     All the best,
>     Franz
>     _______________________________________________
>     R-sig-ecology mailing list
>     [hidden email]
>     https://stat.ethz.ch/mailman/listinfo/r-sig-ecology
>
>
> _______________________________________________
> R-sig-ecology mailing list
> [hidden email]
> https://stat.ethz.ch/mailman/listinfo/r-sig-ecology
>

        [[alternative HTML version deleted]]

_______________________________________________
R-sig-ecology mailing list
[hidden email]
https://stat.ethz.ch/mailman/listinfo/r-sig-ecology
Reply | Threaded
Open this post in threaded view
|

Re: predict community by community and environment

mailbox Franz
Thanks Leonard and Erick,

I found ecodist function MRM to fulfill my purpose!
Seems to be a „hard test“ of several independent distance matrixes (community and env data) and a response distance matrix.

I am new to this whole community explains community thing and find it quit confusing.
I had a sig. correlation using mantel test, no sig. with the procrustes test; then I had a sig. predicting relation using coccorrespondance analysis.
Then I ran the MRM and had again no sig. effect.
I thought the cocorrespondance (R package cocorresp) is already a hard test weather a community can predict another community.
Thus I am confused why the MRM says no sig… I mean internally they work quit different, however, what to trust?

Maybe someone can help?

Best, Franz

> Am 25.10.2016 um 21:01 schrieb Leonardo Ré Jorge <[hidden email]>:
>
> Dear Franz,
>
> An approach that should fit quite well to your intention is Generalized Dissimilarity Modellling (GDMs - Ferrier 2007: http://doi.wiley.com/10.1111/j.1472-4642.2007.00341.x <http://doi.wiley.com/10.1111/j.1472-4642.2007.00341.x>). You can model the compositional dissimilarities among your communities relative to both environmental differences and differences in composition of another taxonomic group. This has also the advantage of allowing you to include the spatial distance among your communities as a predictor, accounting for spatial autocorrelation. Your final output should give you a measure of relative effects of each predictor. GDMs are implemented in the R package gdm.
>
> Best regards,
> Leonardo
>
> On Tue, Oct 18, 2016 at 12:54 PM, LeBrun, Erick <[hidden email] <mailto:[hidden email]>> wrote:
> Hello Franz,
>
> One possibility would be to do ordinations (preferably NMDS) and then do a Procrustes analysis. It is touted to be potentially more sensitive than a mantel test and using the protest command, you can get a correlation statistic and significance associated with the two datasets. The simplified explanation is that it scales and rotates the ordinations to the best fit possible. You can also view error plots that show you “movement” of the points in ordination space that can show trends or differences.
>
> Best regards,
> Erick
>
> Erick LeBrun
> PhD Student, Department of Biology
> Kang Microbial Ecology Laboratory
> [hidden email] <mailto:[hidden email]>
>
> On 10/18/16, 8:35 AM, "R-sig-ecology on behalf of mailbox Franz" <[hidden email] <mailto:[hidden email]> on behalf of [hidden email] <mailto:[hidden email]>> wrote:
>
>     Hi all,
>
>     I am searching for a method to test weather a community is better predicted by an other community or by the environment in a mantel-test like framework?
>     Any help would be appreciated. Platform preferably R.
>
>     All the best,
>     Franz
>     _______________________________________________
>     R-sig-ecology mailing list
>     [hidden email] <mailto:[hidden email]>
>     https://stat.ethz.ch/mailman/listinfo/r-sig-ecology <https://stat.ethz.ch/mailman/listinfo/r-sig-ecology>
>
>
> _______________________________________________
> R-sig-ecology mailing list
> [hidden email] <mailto:[hidden email]>
> https://stat.ethz.ch/mailman/listinfo/r-sig-ecology <https://stat.ethz.ch/mailman/listinfo/r-sig-ecology>


        [[alternative HTML version deleted]]

_______________________________________________
R-sig-ecology mailing list
[hidden email]
https://stat.ethz.ch/mailman/listinfo/r-sig-ecology