Allometry and Ecology of the Bilaterian Gut Microbiome

Abstract

Classical ecology provides principles for construction and function of biological communities, but to what extent these apply to the animal-associated microbiota is just beginning to be assessed. Here, we investigated the influence of several well-known ecological principles on animal-associated microbiota by characterizing gut microbial specimens from bilaterally symmetrical animals (Bilateria) ranging from flies to whales. A rigorously vetted sample set containing 265 specimens from 64 species was assembled. Bacterial lineages were characterized by 16S rRNA gene sequencing. Previously published samples were also compared, allowing analysis of over 1,098 samples in total. A restricted number of bacterial phyla was found to account for the great majority of gut colonists. Gut microbial composition was associated with host phylogeny and diet. We identified numerous gut bacterial 16S rRNA gene sequences that diverged deeply from previously studied taxa, identifying opportunities to discover new bacterial types. The number of bacterial lineages per gut sample was positively associated with animal mass, paralleling known species-area relationships from island biogeography and implicating body size as a determinant of community stability and niche complexity. Samples from larger animals harbored greater numbers of anaerobic communities, specifying a mechanism for generating more-complex microbial environments. Predictions for species/abundance relationships from models of neutral colonization did not match the data set, pointing to alternative mechanisms such as selection of specific colonists by environmental niche. Taken together, the data suggest that niche complexity increases with gut size and that niche selection forces dominate gut community construction.IMPORTANCE The intestinal microbiome of animals is essential for health, contributing to digestion of foods, proper immune development, inhibition of pathogen colonization, and catabolism of xenobiotic compounds. How these communities assemble and persist is just beginning to be investigated. Here we interrogated a set of gut samples from a wide range of animals to investigate the roles of selection and random processes in microbial community construction. We show that the numbers of bacterial species increased with the weight of host organisms, paralleling findings from studies of island biogeography. Communities in larger organisms tended to be more anaerobic, suggesting one mechanism for niche diversification. Nonselective processes enable specific predictions for community structure, but our samples did not match the predictions of the neutral model. Thus, these findings highlight the importance of niche selection in community construction and suggest mechanisms of niche diversification.

Keywords: bacteria; bilateria; microbiome; microbiota; neutral model; species-area.

FIG 1

Gut microbiota of the Bilateria. The figure summarizes several forms of analysis of the 16S rRNA gene tag sequence data. Host species are indicated by their common names—formal genus and species designations are in Table S1. (A) The proportional abundance of the most abundant bacterial phyla averaged by species. (B) Bar graphs summarizing the percentages of reads assigned to the proportionally most abundant OTU in each species. The values were averaged for each species (bar), and values for individual samples are indicated by plus (+) symbols (the graph format repeats in panels C to G). (C) Bar graph summarizing Shannon diversity for the microbial communities of the various species. (D) Bar graph summarizing phylogenetic diversity for the microbial communities of the various species. (E) Bar graph summarizing the proportion of annotated obligate anaerobic bacteria for the various species. (F) Bar graph summarizing the percentages of reads annotated as Wolbachia for the various species. (G) Bar graphs summarizing the percentages of OTUs that were unable to be assigned to references in the Greengenes database. (H and I) Alignments of a 10,000-read random subset of assigned (H) and unassigned (I) sequence reads showing positions with greater than 10% non-gap sequences, emphasizing that the unassigned reads resemble the assigned reads in sequence.

FIG 2

t-SNE plot of UniFrac distances. A two-dimensional representation of unweighted UniFrac distances was generated using t-distributed stochastic neighbor embedding (t-SNE). Samples are colored to indicate phylogenetic class, and orders are further broken out by point shape. Weighted and unweighted UniFrac t-SNE plots of species centroids (including species names) are shown in Fig. S5.

FIG 3

Species-area analysis. The relationship between the weight of the host organism and the number of gut bacterial OTUs found in fecal samples was estimated using a Bayesian regression model. All samples, both those newly determined here and those from previously published data sets, are included. The gray-shaded region shows the 95% credible interval for the slope. The OTU counts were normalized across sample sets as described in Materials and Methods. Origins of sample sets are as follows: diverse Bilateria, this work; primates, reference ; birds, reference ; whale 1 and 2, reference ; insects 1 and 2, reference ; anteaters, reference ; fish, reference ; mammals 1, reference ; mammals 2, reference . Samples were rarefied to 100 reads each.

FIG 4

The bilaterian gut microbiota does not fit predictions of neutral assembly models. (A) The abundance of OTUs from each sample was assessed for their fits to community models (columns) using Akaike information content (AIC) and the AIC averaged within each species (rows). Within each row, the best-fitting models (i.e., those with the lowest AIC levels) appear red, with the color code showing the difference for each model from the minimum AIC. (B to E) Empirical rank-abundance curves and comparisons to model best fits are shown for single samples from human (B), right whale (C), tiger shark (D), and cricket (E).