[R-sig-eco] question about varpart results

[Marion Casati]

Dear all,

I am trying to run variation partitioning analysis on a response data. I 
have three constraining datasets with numeric variables. The issue I 
have is that the varpart function does not seem to detect variance, and 
the residuals are very high.

/varpart(Y = ML, X = ~Rtot$DegreInv + Rtot$mean_circ_rhodo +/

/Rtot$BA_rhodo + Rtot$Hmoy_rhodo, ~Rtot$rec_solnu + Rtot$Fe_profMin +/

/Rtot$OL_moy + Rtot$pHsol_h2o, ~Rtot$densioMean + Rtot$SD_arbres +/

/Rtot$BA_arbres + Rtot$rec_ah + Rtot$Rec_forest_percent, chisquare = T)/

//

/Explanatory tables:/

/X1:~Rtot$DegreInv + Rtot$mean_circ_rhodo + Rtot$BA_rhodo + Rtot$Hmoy_rhodo/

/X2:~Rtot$rec_solnu + Rtot$Fe_profMin + Rtot$OL_moy + Rtot$pHsol_h2o/

/X3:~Rtot$densioMean + Rtot$SD_arbres + Rtot$BA_arbres + Rtot$rec_ah + 
Rtot$Rec_forest_percent /

//

/No. of explanatory tables: 3 /

/Total variation (SS): 4.4854 /

/No. of observations: 62 /

/Residuals = 0.97789/

//

But when I perform a cca analysis between my response dataset and my 
exploratory dataset (not separated into three parts as for the variation 
partitioning analysis), the analysis does detect variance.

/Call: cca(X = ML, Y = Rtot)/

/Inertia Proportion     Rank/

/Total4.48541.0000/

/Constrained3.07850.686341/

/Unconstrained1.40690.313720/

So why does varpart not detect anything?

I think there's something I must not have understood. Can anyone help me ?

Thanks,


Marion Casati
-- 
PhD student in Ecological Sciences: Ecology, Evolution and Biodiversity
Edysan UMR 7058 CNRS - UPJV
1 rue des Louvels
80000 Amiens


	[[alternative HTML version deleted]]