Explanation to testBetaDiversity() difference to PERMANOVA vegan::adonis2()

[andieich]

I try to better understand the testBetaDiversity() function. How does it differ from PERMANOVA, e.g. with vegan::adonis2()? If I understood correctly, both tests compare the position of the group centroids and use permutations to calculate pseudo F-values.

Could I use the distance matrices returned by DivNet (or aitchison distance calculated from fitted_z) in adonis2()? It is more flexible in defining covariates than testBetaDiversity().

[svteichman]

The testBetaDiversity function is pretty similar to the PERMANOVA run with vegan::adonis2(), however it expects a DivNet object (which estimates beta diversity metrics instead of directly calculating them based on the sample), and it uses a bootstrapping technique to generate a null distribution for hypothesis testing as opposed to the permutation approach used in a PERMANOVA.

[svteichman]

@andieich I was just re-reading and realized that I didn't respond to your second question. I think that you could take the distance matrix estimates from DivNet and then use this in adonis2(), if you'd like to use that test, especially if you'd like to flexibility of including more covariates.