PCAVI plots

From: Sara Mouro (sara@gmesintra.com)
Date: Fri Dec 05 2003 - 13:31:22 MET


Bonjour.

I have done PCAIV and projected 3 graphics using Scatters-Labels for -.ivl1, -.ivco and .ivfa to plot sites, species and environmental variables respectively (as Stephane DRAY so gently and efficiently advised me to).

I have also computed the correlation (after PCA on environmental variables) between environmental variables and sites scores by the MatAlg>diagonal inner product module and plot the correlation with Scatter>correlation circle.

(Even after reading the related "fiche thématique") I still not really understand how can we interpretate the contradicting results where some environmental variables show opposite positions at the .ivfa plot and at the correlation circle...

After superposing the 3 graphs (not the correlation circle), sites and species are "correctly positioned" but some environmental variables are not !! - according to my ecological and field experience. However, the same environmental variables are just like I was expecting them at correlation circle plot (seen separatly and of course not superposed to the other plots)...

I would be really greatful if someone could please give me some advise on this subject and explain me which environmental varibles' plot should I use for ecological interpretation of the PCAVI results.

Best regards
Sara Mouro.



This archive was generated by hypermail 2b30 : Tue Sep 07 2004 - 13:45:26 MEST