Re: COA and absolute frequencies

From: Daniel Chessel (
Date: Tue Nov 13 2001 - 09:56:24 MET

At 23:18 12/11/2001 +0100, Agustin Lobo wrote:

>Bonjour there!
>I've run a COA using the R function
>CAIV (as the algorithm is the same than the one in ADE,
>I think that this list is better suited for this
>question than R-help).

CAIV est la fonction de la librairie CoCoAn de R qui étend la CCA de ADE-4.
C'est vrai que c'est la même famille !

>My dataset consisits on forest inventory data
>(area covered x sps), including trees and shrubs.
>A simple COA using this data matrix has the problem that some
>sites with a low tree cover appear cery close to sites with high tree
>cover in the biplot,
>because in both cases they are dominated by the same species (P.
>I understand this comes from the fact that
>COA standardises the data, and relative frequencies are more important
>than absolute frequencies.

Très juste. L'AFC et ses dérivées travaille sur les profils et ignore les
Elle est faite pour cela.

>Therefore, I have selected the site with the highest vegetation cover and
>an artificial species "bare ground", which for this site has a cover of 0
>and for the rest of sites has a cover equal to the difference between the
>vegetation cover and the observed maximum. Using this new "species", sites
>low cover but dominated by P.halepensis no longer appear close to sites
>that are also dominated by P.h. but with a high tree cover.
>I'd appreciate your critical opinions and alternative suggestions.

Sorry, I can't agree.
Il n'est pas normal de modifier les données pour débarasser une méthode de
ses propriétés. L'AFC travaille sur les profils espèces et sur les profils
relevés. Si on ne veut pas avoir une des deux propriétés il vaut mieux
faire de l'ACP ou de l'Analyse des Correspondaces Non Symétriques ou
utiliser des distances comme dans Legendre, P., and M. J. Anderson. 1999.
Distance-based redundancy analysis: testing multispecies responses in
multifactorial ecological experiments. Ecological Monographs 69:1-24.
C'est une opinion.

Daniel Chessel
Universite Lyon 1 - Biométrie et Biologie Evolutive - Bât 741
69622 Villeurbanne CEDEX
Tel : 04 72 44 82 77 - (33) 4 72 44 82 77

This archive was generated by hypermail 2b30 : Thu Jul 04 2002 - 09:39:14 MEST