Highlights • An introduced analytical framework quantifies taxon–community associations to identify computationally inferred candidate keystone taxa. • Application to five PVTGs identifies 121 candidate keystone taxa, most of which are population-specific, with a reproducible subset across traditional populations. • Several candidate keystone taxa associated with lower community dispersion also show consistent associations with multiple disease conditions, highlighting priorities for future investigation. • A conserved metabolic core coexists with population-specific functions, while acculturation is associated with reduced microbial diversity and increased disease-associated taxa.