Trait-based community assembly and succession of the infant gut microbiome

Abstract

The human gut microbiome develops over early childhood and aids in food digestion and immunomodulation, but the mechanisms driving its development remain elusive. Here we use data curated from literature and online repositories to examine trait-based patterns of gut microbiome succession in 56 infants over their first three years of life. We also develop a new phylogeny-based approach of inferring trait values that can extend readily to other microbial systems and questions. Trait-based patterns suggest that infant gut succession begins with a functionally variable cohort of taxa, adept at proliferating rapidly within hosts, which gradually matures into a more functionally uniform cohort of taxa adapted to thrive in the anoxic gut and disperse between anoxic patches as oxygen-tolerant spores. Trait-based composition stabilizes after the first year, while taxonomic turnover continues unabated, suggesting functional redundancy in the traits examined. Trait-based approaches powerfully complement taxonomy-based approaches to understanding the mechanisms of microbial community assembly and succession.

Fig. 1

OTU abundances over gut succession. OTUs were placed into one of three successional groups based on their average trends in abundance across 56 infants over development. Taxa were categorized as early successional if their abundances increased significantly over time (p<0.05), and late successional if their abundances decreased significantly over time, based on linear regressions; OTUs that did not trend significantly over time were placed into a mid-successional or no-trend category. a, c, and e Combined relative abundances of the OTUs of each successional category over time. b, d, and f The five most commonly-represented families among the OTUs of each successional category, and their total relative abundances over the entire sampling period

Fig. 2

Abundance-weighted trait means over gut succession. Filled red circles show average abundance-weighted means of predicted trait values of gut microbiomes of up to 56 infants in each month of development (a-l). N is equal to the number of samples in each month, and ranges from 27 to 59. Vertical red lines show 95% confidence intervals. Black trendlines were fit using generalized additive models

Fig. 3

Trait-based successional patterns differ by delivery mode and antibiotic history. Abundance-weighted means of predicted trait values of infant microbiomes over succession, grouped by infant delivery mode and antibiotic history (a-l). Filled circles show average abundance-weighted trait means of samples within 6-month periods in each cohort of infants. N is equal to the number of samples in each 6-month period; there were between 25 to 31 total samples from six infants delivered by C-section who received little to no antibiotics ("C-section" group), 66–91 total samples from 18 infants who were treated with antibiotics for at least 50 days ("Antibiotics" group), and 72–93 total samples from 18 control infants that were delivered vaginally and received no antibiotics ("Control" group). Vertical lines show 95% confidence intervals. Asterisks denote significance based on Welch t tests performed between each treatment group and the control group (*: adjusted p<0.05; **: adjusted p<0.01; ***: adjusted p<0.001)

Fig. 4

Trait-based composition stabilizes earlier than taxonomic composition. Filled circles show mean pairwise compositional dissimilarities of gut microbiome samples collected from individual infants, averaged within 6-month periods for each infant, and then across infants. OTU-based dissimilarity was calculated using Bray–Curtis dissimilarity. Trait-based dissimilarity was calculated using multidimensional Euclidean distance after scaling the distributions of values for each trait to ensure equal contribution. a Mean OTU-based dissimilarity between subsequent samples declines slightly over time. b Mean OTU-based dissimilarity between samples and the final samples taken from each infant decreases steadily throughout the sampling period until finally reaching baseline levels of between-sample dissimilarity in the last 6-month period, seen in a. c Mean trait-based dissimilarity between subsequent samples appears elevated in the first year, but does not differ significantly from null model predictions that assume trait-agnostic turnover (see Methods). d Mean trait-based dissimilarity between samples and the final samples taken from each infant decreases rapidly and approaches baseline levels of between-sample dissimilarity within the first year, seen in c. Moreover, trait-based community composition converges toward that of the final sample significantly faster than null model expectations, illustrating the non-random nature of trait-based community turnover over succession. In all panels, N equals 56, the number of infants. Vertical lines show 95% confidence intervals. Asterisks denote significance between observed and null model predictions based on Welch t tests (*p<0.05; **p<0.01; ***p<0.001)

Fig. 5

Infants’ microbiomes converge compositionally over time. Filled circles show mean compositional dissimilarities of gut microbiomes across infants within each 6-month periods. Mean dissimilarities were calculated by first taking the mean dissimilarity of all sample pairs, except those from the same infant, in each of the first 36 months of development (for these means, N ranges from 88 to 3410), and then taking their means within each 6-month period; hence, for each circle, N equals 6. OTU-based dissimilarity was calculated using Bray–Curtis dissimilarity. Trait-based dissimilarity was calculated using multidimensional Euclidean distance after scaling the distributions of values for each trait to ensure equal contribution. a OTU-based dissimilarity among infants decreased slightly over time, indicating a modest convergence in taxonomic composition. b Trait-based dissimilarity among infants fell quickly over the first 18 months and then remained relatively static thereafter, indicating rapid convergence in trait-based composition during early succession. The magnitude of trait-based compositional convergence across infants was significantly greater than predicted by a null model assuming trait-agnostic turnover. Vertical lines show 95% confidence intervals. Asterisks denote significance between observed and null model predictions based on Welch t tests (*p<0.05; **p<0.01; ***p<0.001)