My data has been processed and I have been working on the Beta-diversity analysis. I want to look at which environmental factors contribute most to the community variation between my sites. Qiime2 does a great job of performing statistical tests and representing them in graphs as outputs.
However, is there a way for me to get numerical data (% of distance due to each abiotic factor) using qiime2? Is it time for me to take my data to R? Also, Is it better to assign taxonomy then preform these statistics?
Overall, Iām confused on what my next step should be and any help would be great!
We have an open ticket to implement multi-way PERMANOVA (adonis) tests in QIIME2, possibly as early as early 2018. If you don't want to wait for that to resolve, this specific question should be asked in R right now. You can run adonis tests in the vegan package (using a distance matrix as input, exported from QIIME2). QIIME2 does currently use PERMANOVA tests in beta-group-significance, but these only operate on a single metadata factor at time and do not report R2 values (which are the % variation in distance attributed to each factor, precisely what you are looking for).
In QIIME2, bioenv and mantel tests may also be useful and related to your goals.
That all depends on your goals. Are you interested in assessing whether factors associate with differences in communities at the SV/OTU level, or at a specific taxonomic level? It may be worth trying both (and if you get different results that may provide important information as well!)
Thanks for the positive feedback! I hope this suggestion helps!
