Fig. 3

Crewmembers’ salivary microbiota composition in time. a) Bray-Curtis (also known as quantitative Sorensen) index has been used to inspect distances between and within-subject during isolation and follow-up. Between-subjects diversity was computed by comparing the salivary microbiota of each subjects at each timepoint (gray arrows); within-subject diversity was computed by comparing the salivary microbiota of the same subject over time (black arrows). b) Change-point analysis revealed changes in salivary microbiota composition of each subject (CP detection). Genetic algorithm and linear modelling detected increasing/decreasing patterns along time (GA optimization). Finally, we fit a linear mixed-model for each segment detected using crewmembers as random intercept (Modelling). c and d) Results obtained following the pipeline reported in “b” for within- and between-subjects differences. Diets were reported in the bottom part of the plots using different colors (FV, fist food variant; SV, second food variant; TV, third food variant). Since crewmembers ate freely during the follow-up, no diet was reported in the plot