I am currently working on an ecological dataset with presence-absence counts. I would like to use a PERMANOVA to test if several factors significantly influence the community composition. However, a few of the factors give a significant PERMDISP result. As the design is unbalanced, I fear this may affect the results. Specifically, as one factor that is significant had greater dispersion in the smaller group (which could be due to the test being too liberal), whereas the factors that were not significant had a greater dispersion in the larger group (could be due to the test being too conservative).

I have read the article by Anderson et al. 2017 on the modified F2 statistics that could handle heterogeneous data (paper: "Some solutions to the multivariate Behrens–Fisher problem for dissimilarity‐based analyses") and I think this could be helpful in my case. However, I am unsure as to how to calculate these statistics for my own dataset, as I am not very experienced with maths.

I am working mainly in R and I read in the article that all simulations described in Sections 3 and 5 were performed using R. However, there is no more explanation than that....

I was wondering if anyone knows if there are pieces of the code are available? Or if there is any software that can calculate F2 statistics?