14.2.2 beta diversity

[antagomir]

Section 14.2.2 has this text: "Before applying rarefaction, selecting the most variable features can help minimize variation caused by random subsampling. These features have the highest read counts, while rare features tend to increase sampling variation. This approach facilitates comparison of results between non-rarefied and rarefied distance calculations later on." and following code example.

That example picks top-5 most variable taxa before estimating beta diversity.

This is good to speed up analyses in some cases but in general, I tend to think that beta diversity should be estimated based on all available features. I have received reports from users directly following the top-5 taxa selection as the default procedure. This shouldn't be a general default. The text could be more explicit about that (or alternatively we could remove that top-n subsetting)?