[R-sig-eco] Variability explanations for the PCO axes as in Anderson and Willis 2003
Kari Lintulaakso
kari.lintulaakso at gmail.com
Thu Jun 2 03:09:43 CEST 2011
In Anderson and Willis 2003 (Canonical Analysis of Principal
Coordinates: A Useful Method of Constrained Ordination for Ecology)
they write:
"For multivariate analysis, the data were transformed
to y' = ln(y + 1) to remove large differences in scale
among the original variables. Then Bray-Curtis dissimilarities
were calculated between every pair of observations,
and an unconstrained ordination was done using
PCO (metric MDS) on the dissimilarity matrix. The
first two PCO axes explained 20.72% and 12.37% of
the variability in the original dissimilarity matrix."
Does anyone know which method produces those 20.72% and 12.37% for the PCO's?
I have tried something like this:
library(ecodist)
x<-bcdist(dune) # Bray-Curtis distance, no transformation
x.pco <- cmdscale(x, k=2, eig = TRUE) # PCO, PCoA
x.pco$GOF
> x.pco$GOF
[1] 0.5714973 0.5961463
I suppose these 0.57 and 0.59 are not comparable with the percentages
in Anderson and Willis?
So my question is how do I get similar kinds of variability
explanations for the PCO axes as in A&W2003?
And secondly how do I get those percentages for NMDS?
I really am a newcomer in these kinds of analysis and any kind of
guidance will help.
-Kari
--
Kari Lintulaakso, M.Sc.(Biosciences)
Doctoral student
Paleontology and Paleoecology
Department of Geosciences and Geography
University of Helsinki
* Web page: http://blogs.helsinki.fi/lintulaa/
More information about the R-sig-ecology
mailing list