[R-sig-eco] Dissimilarity for ordination of rank data

[Tania Bird]

Hi all,

I have a data set of species data in different sites but instead of
abundances, the presence has been converted to a rank score
site-species matrix. The ranks are calculated based on max-scores that
have been log transformed ln(x+1).
0 = absent and 5 = highest abundance for that species relative to
other sites and other species.

I would like to carry out multivariate analyses on this data such as
PCA / PCoA  RDA/Variation partitioning.
I assume no transformation is needed for this data since they are already ranks.

My questions are :
1) What is the best dissimilarity measure to use for this rank data?
I read that Gower's distance gowdis() in {FD} or daisy() in {cluster}
may be good choices?

2) Is it appropriate to conduct RDA on rank data?
Is an there a better alternative?

As a bonus question - can I treat these data the same as I would
abundance data in more advanced analyses such as beta-diversity
analyses (e.g. betadisper() in {vegan} ) or  spatial eigenvector
mapping (e.g. dbMEM / AEM in {adespatial} )?


Many thanks

Tania

Tania Bird MSc
PhD Student: Coastal dune biodiversity conservation in Nizzanim LTER
Dept. of Geography & Environmental Development,
Ben Gurion University, Beer Sheva
https://www.linkedin.com/in/taniabird