Hi, I'm trying to understand the best approach for covariate selection in differential abundance analysis (ANCOM-BC/DESeq2). My dataset involves clinical outcomes (Success/Failure), 3 timepoints, and a single treatment. Alpha diversity models (lm models) didn't flag any of my clinical covariates as significant. However, PERMANOVA on beta diversity showed many of these same covariates (e.g., Age, specific clinical signs) are significantly associated with overall community composition, alongside my main factors of interest (Outcomes and TimePoints). If a covariate significantly impacts beta diversity (overall community structure) but not alpha diversity, is it still important to include it in taxon-level differential abundance models (ANCOM-BC/DESeq2) to control for its influence?

What's the rationale? Any insights or best practice recommendations would be appreciated!