I have a dataset of viral 'OTUs' across various sample points from different sites/times.

I also have meta-data collected at these same sample points, for things like temperature and the concentration of key chemical compounds.

My goal is to identify if these (or some of these) abiotic variables correlate with the variation in the OTU community diversity- for now I am planning to do an RDA.

What I am wondering is whether or not it makes sense to conduct ordination on the abiotic variables before doing the RDA? I have done a PCA on these abiotic variables and it is clear that these variables alone clearly do not cluster in any sort of recognisable pattern (for example if you colour them by 'season' there is no clear clustering at all).

So I was thinking that perhaps doing an RDA would not be valid, because if the abiotic variables don't cluster logically in an ordination of them alone then surely it doesn't make sense to plug this data into an RDA?

  • $\begingroup$ I really don't know much about RDA (so correct me if I am wrong), but to me RDA is a method of ordination as well, so the sentence makes sense to conduct ordination on the abiotic variables before doing the RDA makes little sense. When you state your goal, it seems like you would like to perform some hypothesis testing which is, I think not what an RDA does. Am I wrong? Why don't you just have OTU community diversity as response and your abiotic variables as explanatory variables in a simple GLM? You don't even seem to need a MANOVA (which would require ordination + regression). $\endgroup$ – Remi.b Jul 24 '17 at 15:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.