On 13/06/2012, at 07:33 AM, Manuel Spínola wrote:
> Dear list members,
>
> I am working on an NMDS using metaMDS from vegan with the Doubs fish data
> from the book Numerical ecology with R.
>
>> especies <- read.csv("DoubsSpe.csv", row.names = 1)
>> e.mds = metaMDS(especies, distance = "bray")
> Run 0 stress 0.003706943
> Run 1 stress 0.0004788424
> ... New best solution
> ... procrustes: rmse 0.009197047 max resid 0.01874812
> Run 2 stress 0.0004600702
> ... New best solution
> ... procrustes: rmse 0.0002883791 max resid 0.0005590752
> *** Solution reached
>
> Mensajes de aviso perdidos
> In distfun(comm, method = distance, ...) :
> you have empty rows: their dissimilarities may be meaningless in method
> „bray‰
>>
>
> My results are "strange" (stress is too low and different to the one on the
> book) and the plot is very different to the one that appears in the book.
>
Manuel,
I'm too lazy to go to have a look at the book now (I'm sitting in my balcony
sipping my morning coffee), but I assume that the difference is that metaMDS in
your book was still based on MASS::isoMDS(), but the current vegan (from 2.0-0)
uses its own monoMDS() function as a default. One difference is that isoMDS()
expresses the stress per cent, and monoMDS() as parts of one, so that equal
isoMDS() is 100x higher. Another difference is that monoMDS() implements
treatment of tied dissimilarity values, and defaults to "weak" ties so that
equal observed dissimilarities can be allowed to be at different ordination
distances. If I remember correctly, these Doubs fish data are very simple so
that monoMDS() really may be able to have nearly zero stress (which is suspect
in general).
For better correspondence to the book, you may first try setting 'weakties =
FALSE' which will force tie treatment that is closer to isoMDS(), but still not
identical. For true replication of the book results you should set 'engine =
"isoMDS"'.
As usual, these are documented features.
Cheers, Jari Oksanen
--
Jari Oksanen, Dept Biology, Univ Oulu, 90014 Finland
jari.oksa...@oulu.fi, Ph. +358 400 408593, http://cc.oulu.fi/~jarioksa
_______________________________________________
R-sig-ecology mailing list
R-sig-ecology@r-project.org
https://stat.ethz.ch/mailman/listinfo/r-sig-ecology
