Skip to content

Relating skull shape to environmental variables

4 messages · Diogo B. Provete, Frederico Faleiro, Botta-Dukát Zoltán

Original
Diogo B. Provete
# 
Dear list members,
My PhD student Adriana Acero and I are trying to relate the mean cranial
shape of species (Principal components scores) of a bat community to the
land use type (% of each land use class measured from satellite data) in
which species occur.
Therefore, our analytical problem is more of a RLQ (relating traits to
environment) that is common in community ecology. We want to use the ade4::
rlq function to relate the 10 PCs representing the position of each species
in the shape space (*Q*), to the environmental matrix (*R*), using the
species composition matrix as *L. *

The problem is that the ade4::rlq function only accepts objects of the
dudi.pca class. Did anyone come across a similar issue?

We tried to follow this blog post
<https://www.r-bloggers.com/r-for-ecologists-rlq-analysis-semi-explained/>,
which builds the analysis step by step using custom functions and also
tried to coerce our PCA object into a as.dudi class, but it didn't work.

Thank you in advance,
Diogo
1 day later
Frederico Faleiro
# 
Hi Diogo,
I did not understand why you can not use the dudi.pca, but you can find the
tutorial from the authors here:
http://www.esapubs.org/archive/ecol/E095/002/ and the paper here:
https://esajournals.onlinelibrary.wiley.com/doi/full/10.1890/13-0196.1.

Cheers,

Em qua., 15 de jul. de 2020 ?s 15:23, Diogo B. Provete <dbprovete at gmail.com>
escreveu:

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Diogo B. Provete
    

    

      
      #
    

  


  

      
Hi Frederico,
thanks for replying. Also thanks Karlo for sending a separate email
yesterday. I think I haven't made myself clear enough. Our dataset is
derived from geometric morphometrics, we are measuring skull and jaw shape
of bat species and we want to test the influence of land use classes on
shape. So we have to process the data using the geomorph::gm.prcomp
function that allows an array containing the XY landmark coordinates for
each species (a .tps file). This function returns the PCA scores
representing the position of each species (mean) in the shape space.

ade4::dudi.pca only accepts the regular square matrix as input, that's why
we can't use it.

Best,
Diogo

Em sex., 17 de jul. de 2020 ?s 10:36, Frederico Faleiro <fvfaleiro at gmail.com>
escreveu:

            

      
      
      
    

        

      
      
    

            

      
      
      
    

        

  
  


  


      

    

    

      
      

        


  

        

      Botta-Dukát Zoltán
    

    

      
      #
    

  


  

      
Dear Diego,

I suppose that PCA scores resulted by? geomorph::gm.prcomp are 
uncorrelated. If you use matrix of these scores as imput of dudi.pca, 
the scores in the results should be the same matrix, but in the 
requested form.

Best,

Zoltan

2020. 07. 17. 21:13 keltez?ssel, Diogo B. Provete ?rta: