Open aguilart opened 3 years ago
There are no species scores in the wcmdscale
result because there is no way of calculating those with the input: in rda
and metaMDS
the input was data frame dune
, but in wcmdscale the input was distance data d
, and there are is no information on species in distances. You will neither have species scores if you submit dissimilarities instead of community data to metaMDS
.
Current vegan has function sppscores<-
that adds species scores to an ordination object (or replaces old scores) for some methods. However, it is not implemented for wcmdscale
. It is implemented for dbrda
though, and if you do not use weights, dbrda
and wcmdscale
will be identical. So the following will work:
d <- vegdist(dune)
ord <- dbrda(d ~ 1) # no species scores in ord
sppscores(ord) <- dune # add species scores based on dune
Adding species scores was discussed in issue #254.
Aha! Thanks a lot for the reply! Tha makes sense, to wcmdscale is fed the distance matrix, thus no way of having scores. I will try dbrda.
Sorry, one more question. What do weights refers to? I feel this is a very basic question but I got right away confused on what weights are.
If you do not know what are the weights for, you do not need them.
The Example section in the wcmdscale
help shows one example where you need weights, but other users may have various needs for weights, starting from duplicated sampling units, different sizes of sampling units etc.
Thansk a lot for taking the time in answering these questions! It really helps a lot!
Hi, I was comparing the outputs from a PCA (using
rda
), PCoA (using the combovegdist
andwcmdscale
) and NMDS (usingmetaMDS
). Then when usingscores
to extract the species and site scores I realized that PCoA does not give you species scores.Example:
Performing each ordination method
Extracting the species scores works only for the pca and NMDS . I mean, there are 30 species in the dataset
dune
, and those are exactly the ones that correspond to the species scores in PCA and NMDSFor PCA:
For NMDS:
But for PCoA, I what I get are the site scores (20):
Digging into the output of
wcmdscale
I could not find anywhere sth like species scores, so I understand these are not given at all. So my question is, why? why out of these tree techniques, in PCoA species scores are not given?Aside from knowing why this is the case, the reason for this question is because I want to ordinate multivarite count data (thus PCA is not well suited) and I want to also to know which variable drives most of the ordination. Thus, I want to have species scores. I thought that PCoA would be an option as I can use bray distances, but I did not know species scores are not given. So, should I use NMDS instead?