Fig. 3

Temporal dynamics of four enterotypes in the first 3 years. A The Sanky diagram shows the transition of four enterotypes in the nine age groups (n = 13,776). The width of each flow is proportional to the number of infants experiencing an intra- or inter-enterotype transition. The height of each bar is proportional to the number of infants in each group. The number of intra- and inter-enterotype transition events in the first 3 years is summarized in the box. B Markov chain with subject-independent transition probabilities among four enterotypes, in which arrow weights are proportional to the maximum likelihood estimate of the transition probabilities among different states. Node size indicates the number of subjects in each enterotype. Only transition probabilities greater than 0.2 are shown. C Source tracking of infant gut bacteria in each time point. Samples from each enterotype of the former time point were considered as the potential sources of the latter time point. The areas of each color indicate distinct microbe source of each enterotype. The gray area indicates unknown sources. D A two-step procedure to evaluate the influence of stochasticity on the infant gut microbial community. β-nearest taxon index (βNTI) and Raup Crick index (RCI) were first calculated in each community. Then, the frequency of samples with |βNTI| < 2 and |RCI| < 0.95 in each month was used to determine the rate of stochasticity. E The temporal changes of alpha diversity and the rate of stochasticity in each enterotype. Red lines indicate alpha diversity based on the Shannon index and blue lines indicate the rate of stochasticity estimated from Fig. 4D. The shaded regions indicate the 95% confidence intervals