[R-sig-eco] NMDS vegan

[Jari Oksanen]

On 25/10/10 21:10 PM, "Soumi Ray" <soumiray74 at gmail.com> wrote:

> Hi listers,
> 
> I am trying to run NMDS in vegan package. I have a species dataset - with
> columns as species and rows as variables. All my data are 0/1
> (presence/absence). My data has missing values. I saved my data in txt file
> and the missing values are blank spaces. I am using the syntax:
> <nmds <-read.table ('nmds.txt', header=T, rows.names=1, sep="\t")
> <nmds.mds.alt <- metaMDS (nmds, distance="bray", k=2, autotransform=FALSE)
> 
Soumi,

The basic problem with metaMDS of vegan (NMDS is in MASS) is that you have
missing values. You are not supposed to have missing values in species data,
and most dissimilarity measures cannot cope with those (some can: check the
docs). Further, it is usually not meaningful to handle community composition
of empty communities: their composition is NA. Most dissimilarity measures
cannot handle such cases meaningfully. Some can handle those, but I'm not
sure if they do it meaningfully. Getting around these two things may help
you further in metaMDS.

Cheers, Jari Oksanen
> But it is showing me the error:
> Error in distfun(comm, method = distance, ...) :
>   NA/NaN/Inf in foreign function call (arg 1)
> In addition: Warning message:
> In distfun(comm, method = distance, ...) :
>   you have empty rows: their dissimilarities may be meaningless in method
> bray
> 
> Could anyone kindly let me know where am I going wrong? I admit i am new to
> R, trying to self tutor myself.
> 
> Thank you,
> 
> Regards,
> 
> Soumi
> 
> [[alternative HTML version deleted]]
> 
> _______________________________________________
> R-sig-ecology mailing list
> R-sig-ecology at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-ecology