Hi @salias
Thank you for the reply.
I didn’t know that the Bray-curtis table is built from a rarefied table, interesting information (I used microViz::dist_calc for both).
For me it does make sense to have a sample depths as a technical covariate. If a sample has much less reads than another one, then it should have less rare species present.
However, then I would assume some correlation between the number of reads and the number of species observed, but this is not really the case, just some slight tendency :
So, let's conclude that for ancombc there's no need to correct the read depth.
Maybe you have comments on another topic that I started on ancombc ?