Fig. 5

Cohousing mixed the microbiota and ameliorated the impact of STAT on composition, but high-fat diet quickly diversified the microbiota. a β-Diversity investigated using principal coordinate analysis of Bray–Curtis distances between samples captured the effect of diet in the first coordinate (PCo1, 49% explained variance (EV)). b Time series of PCo2 (15% EV) reveals that it captures—at least partially—the effect of STAT on the microbiota. c The effect of STAT is ameliorated in cohoused mice. d Analysis of between-mouse similarity (Sørensen similarity index) shows that the microbiota, which start the cohousing period dissimilar, become more similar during the chow period. The similarity temporarily plummets during the rapid microbiota shifts caused by the switch to high-fat diet. e A detailed analysis of the Sørensen similarity pinpoints the exact time when mice became more similar to their cohousing mates compared to their littermates: after 2 weeks of cohousing (cohousing started with chow, the first vertical line). The microbiota similarity-increasing effect of cohousing became more pronounced in high-fat diet, while the average similarity between all mice plummeted simultaneously. For each time point, we modeled similarity ~ Cage + Litter (see Methods). Dashed lines show the intervals determined by the standard error of the parameter estimates