Rothschild · Nature 2018 · Cross-sectional observational study · n=1046

Environment dominates over host genetics in shaping human gut microbiota.

Cited 3027 times in the scientific literature.

Level 4 - case-series / case-control

Cross-sectional observational genetic and microbiome association study

PubMed 29489753 · doi:10.1038/nature25973 · record verified 2026-08-30

What was done

The authors analyzed host genotype and gut microbiome data from 1,046 healthy individuals representing several distinct ancestral origins who shared a relatively common environment. They assessed the association between gut microbiome composition, host genetic ancestry, shared households, and environmental factors (including diet, medications, and anthropometric measures), and tested whether microbiome data improved trait prediction (such as glucose levels and obesity metrics) over genetic and environmental data alone.

What was found

The gut microbiome showed no significant association with genetic ancestry, with host genetics playing a minor role overall. Genetically unrelated individuals living in the same household shared significant microbiome composition similarities. Over 20% of inter-individual microbiome variance was associated with environmental factors (diet, drugs, and anthropometrics). Incorporating microbiome data significantly improved predictive accuracy for traits such as glucose regulation and obesity compared to models using only host genetics and environmental variables.

Why it matters

Because environmental exposures dominate over inherited genetics in shaping gut microbiome composition, microbiome-targeted lifestyle or clinical interventions may be broadly applicable across diverse human genetic backgrounds without being strictly constrained by host ancestry.

Limits

The abstract reports a cross-sectional design that cannot determine causality or longitudinal stability. The cohort is restricted to healthy individuals sharing a single broad geographic environment. Exact numerical effect sizes for specific individual taxa, heritability estimates, and detailed predictive accuracy metrics are omitted from the abstract.

Cited by