On Mon, 2010-11-29 at 10:38 +0200, Karen Kotschy wrote:
> Hi Gavin
>
> On Mon, Nov 29, 2010 at 10:22 AM, Gavin Simpson
> <gavin.simp...@ucl.ac.uk>wrote:
<snip/>
> >
> > I think the answer to this is quite simple: are the numbers for each row
> > ("trait") comparable as a set of weights?
> >
> > In a normal species matrix, the data in each cell will be count or
> > percentage abundance or some such similar thing. In one sense,
> > therefore, we can say that the data are all in the same units. This
> > isn't the case from your description of the data, above.
> >
> > If they are not comparable (or standardised/normalised to make them
>
> > comparable), and I don't see how they can be/are, I would struggle to
> > conceptualise what a weighted mean of site locations on the nMDS plot
> > would mean if the weights were such disparate types of data.
> >
> >
> You are right, the rows all contain different measures and so are not
> strictly
> comparable. But they are all relative measures (proportion of the total
> across
> all species, all between 0 and 1).
>
> Are you saying that the meaning of the "site" locations is dubious, or that
> of
> the "species" (only the latter calculated using wascores as far as I
> understand)?
> Should I not be using MDS for this?
[Please keep replies on-list so others can contribute to the discussion]
I was responding to your Q about the species WA scores. If they are all
relative and standardised to be in the same range, then this doesn't
seem to be too much of a problem.
The sites thing is a different Q; as long as the dissimilarity you use
is appropriate for this sort of data you should be fine.
The species in your plot are located at the weighted average of the
"sites", with the weights being the taken as "abundances" of your
"species". For "abundances" you have relative measures on different
diversity metrics, and "sites" are the metrics themselves. My point was,
if you can consider the weights as being comparable, how can you compute
a weighted average? It looks like you can make a comparison in your case
though.
I'm not too familiar with these trait-based ordinations, but taking WA
of proportional data is fine - palaeoecologists do it every day
reconstructing heaven-knows what from biological community data.
As for interpretation; "species" points on the nMDS diagram will be
located closer to diversity "traits" that they are relatively more
"associated" with.
HTH
G
> Thanks
> Karen
--
%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%
Dr. Gavin Simpson [t] +44 (0)20 7679 0522
ECRC, UCL Geography, [f] +44 (0)20 7679 0565
Pearson Building, [e] gavin.simpsonATNOSPAMucl.ac.uk
Gower Street, London [w] http://www.ucl.ac.uk/~ucfagls/
UK. WC1E 6BT. [w] http://www.freshwaters.org.uk
%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%
_______________________________________________
R-sig-ecology mailing list
R-sig-ecology@r-project.org
https://stat.ethz.ch/mailman/listinfo/r-sig-ecology
