Interactions between strains govern the eco-evolutionary dynamics of microbial communities
- Physics of Living Systems, Department of Physics, Massachusetts Institute of Technology, United States
- Department of Civil and Environmental Engineering, Massachusetts Institute of Technology, United States
- Department of Biological Sciences, Boise State University, United States
Figures
Figure 1 with 10 supplements
Closely related strains coexist for hundreds of generations in pitcher plant-derived microbial communities.
(a) Diagram illustrating our experimental protocol. Stacked bar plots show the composition of one community (M06) at the amplicon sequence variant (ASV) (species) level sampled at each transfer; …
Figure 1—figure supplement 1
Long-term species dynamics of all 10 experimental microbial communities.
Stacked bar plots show the composition of all 10 communities (M01–M10) at the amplicon sequence variant (ASV) (species) level sampled at each transfer; each color corresponds to a unique ASV that we …
Figure 1—figure supplement 2
Non-metric multidimensional scaling (NMDS) plot of community compositions at the species level.
Two-dimensional NMDS plot of the 16S community compositions (species level) using Bray–Curtis dissimilarity. Each point represents a community at a particular sequenced time points, and lines …
Figure 1—figure supplement 3
Most (>99%) single-nucleotide polymorphisms (SNPs) are biallelic (have two alleles).
Histogram showing the fraction of SNPs corresponding to each species in each metagenomic sample for which we could detect only two alleles. The dashed line shows the mean fraction of SNPs that are …
Figure 1—figure supplement 4
Single-nucleotide polymorphism (SNP) trajectories within a species are highly correlated.
Histogram showing the distribution of the Pearson correlation coefficient between all pairs of SNP trajectories belonging to the same species in the same community, measured across all species and …
Figure 1—figure supplement 5
Single-nucleotide polymorphisms (SNPs) within strains tightly cluster together.
Examples of allele frequency trajectories of SNPs belonging to the same species in the same community for amplicon sequence variant (ASV)1, belonging to the genus Aquitalea, in communities (a) M03 …
Figure 1—figure supplement 6
Single-nucleotide polymorphism (SNP) clusters are robust to alternate clustering methods.
Examples of correlation matrices between SNP trajectories belonging to the same species in the same community for amplicon sequence variant (ASV)1, belonging to the genus Aquitalea, in communities (a…
Figure 1—figure supplement 7
More examples of single-nucleotide polymorphism (SNP) clusters.
Examples of correlation matrices between SNP trajectories belonging to different species and communities across the dataset, specifically (a) amplicon sequence variant (ASV)31 from M02, (b) ASV15 …
Figure 1—figure supplement 8
Changes in strain relative frequencies.
Distribution of the overall change in relative strain frequency between conspecific strains across all species in our communities. The overall change was measured as the difference between the …
Figure 1—figure supplement 9
Changes in strain frequencies often influence their overall species abundances.
Distribution of the eco-evolutionary influence between a species and its constituent strains across all communities (see supplementary text), measured as the magnitude of the correlation between the …
Figure 1—figure supplement 10
Distribution of correlations between species’ relative abundances inferred using read mapping and 16S rRNA sequencing.
Distribution of the correlations between the relative read abundance (fraction of shotgun metagenomic reads mapped to each species, after normalizing for genome length) with its relative abundance …
Figure 2 with 3 supplements
Even highly related strains (~100 single-nucleotide polymorphisms [SNPs] apart) can decouple in their dynamics.
(a) Schematic showing examples of strain–strain coupling. We defined strain–strain coupling as the temporal correlation between strain abundances belonging to the same species in a community; …
Figure 2—figure supplement 1
Null distribution of strain–strain coupling from a consumer-resource model with no differences between conspecific strains.
Distribution of the strain–strain coupling across all species and communities (similar to Figure 2b) but using trajectories generated from our second consumer-resource model, where strains are …
Figure 2—figure supplement 2
Strain–strain coupling distribution is robust to using an alternate measure.
Distribution of the strain–strain coupling across all species and communities (similar to Figure 2b) but measured using the nonparametric Spearman correlation coefficient (see Materials and methods);…
Figure 2—figure supplement 3
Distribution of strain–strain coupling with the sign of the correlation.
Distribution of the strain–strain coupling across all species and communities (similar to Figure 2b) but with the sign of the Pearson correlation coefficient, rather than the magnitude (see …
Figure 3 with 8 supplements
Community interactions are strain-specific.
(a) Schematic showing how we measured dynamical correlations at the strain and species level for a species pair A and B. For any species pair, we defined strain–strain correlations as the highest …
Figure 3—figure supplement 1
Null model where we shuffled species–strain associations does not show the observed strain specificity.
Scatter plot of the dynamical correlation between species in a community and the highest correlation between their corresponding strain pairs (similar to Figure 3b) but with species and strain …
Figure 3—figure supplement 2
Dynamical correlations between species and strains do not cluster by community identity.
Scatter plot of the dynamical correlation between species in a community and the highest correlation between their corresponding strain pairs (similar to Figure 3b) but with strain pairs being …
Figure 3—figure supplement 3
Strain-specific interactions are stronger even when estimating abundances purely from metagenomic reads.
Scatter plot of the dynamical correlation between species in a community and the highest correlation between their corresponding strain pairs (similar to Figure 3b) but with species abundances …
Figure 3—figure supplement 4
Strain-specific interactions are stronger even when using an alternate measure.
Scatter plot of the dynamical correlation between species in a community and the highest correlation between their corresponding strain pairs (similar to Figure 3b) but correlations measured using …
Figure 3—figure supplement 5
Interaction networks inferred at the level of species and strains.
Interaction networks inferred using dynamical correlations from community M07, measured at the species (left) and strain level (right) (see supplementary text and Materials and methods). Each node …
Figure 3—figure supplement 6
Examples of shuffled cases where species correlations are higher than strain correlations.
(a, b) Top: relative abundance time-series plots of two correlated species (one each in a and b) obtained from the shuffling analysis (see Materials and methods). Bottom: relative abundances of each …
Figure 3—figure supplement 7
Model recapitulates distance-dependent strain decoupling.
Strain–strain coupling as a function of the competitive distance, D, between strains, in the first consumer-resource model, where strains are ecologically distinct (see Materials and methods). Each …
Figure 3—figure supplement 8
Geometric interpretation of strain–strain and species–species correlations in our models.
(a) Schematic showing the phenotypes (e.g., resource consumption rates) of strains belonging to two different species, A (green) and B (red), generated by our first consumer-resource model, where …
Figure 4 with 4 supplements
Genetic variation in regulators, transporters, and pseudogenes differentiates strains.
(a) Bar plot showing the four functional categories of genes most enriched in strain-differentiating single-nucleotide polymorphisms (SNPs). The x-axis represents the mean number of SNPs belonging …
Figure 4—figure supplement 1
Most single-nucleotide polymorphisms (SNPs) that differentiate strains are in the coding regions.
Stacked bar plots showing the distribution of genetic differences between strains in our communities (left) and variants in the Escherichia coli long-term evolution experiment after 60,000 …
Figure 4—figure supplement 2
Mutations accumulate at a similar rate in both pseudogenes and other genes.
Bar plots showing the average number of mutations detected in a pseudogene (left) and any other gene (right) in strains within our 10 communities. The number of mutations was measured per gene per …
Figure 4—figure supplement 3
Functional differences enriched in single-nucleotide polymorphisms (SNPs) differentiating strains of Aquitalea magnusonii from the NCBI GenBank database.
Bar plot showing the four functional categories of genes most enriched in strain-differentiating SNPs from random strains of the species Aquitalea magnusonii, derived from the NCBI GenBank database …
Figure 4—figure supplement 4
Dynamics of a de novo loss-of-function (pseudogenizing) mutation.
Time course of single-nucleotide polymorphisms (SNPs) in a Pseudomonas species from community M05, where we observe a new loss-of-function mutation (red) appearing at the fourth time point (zero …
Figure 5
Conceptual model of strain-dominated long-term dynamics.
(Diversity pre-assembly) Schematic showing the phenotypes (e.g., resource consumption rates) of strains belonging to two different species, A (green) and B (red), in a large strain pool in nature. …
Additional files
-
Supplementary file 1
Metadata and accession numbers for all 33 assembled genomes used in the study.
- https://cdn.elifesciences.org/articles/74987/elife-74987-supp1-v2.xlsx
-
Supplementary file 2
Set of single-nucleotide polymorphism (SNP) locations and corresponding gene annotations for members of an example community M04.
- https://cdn.elifesciences.org/articles/74987/elife-74987-supp2-v2.xlsx
-
Transparent reporting form
- https://cdn.elifesciences.org/articles/74987/elife-74987-transrepform1-v2.docx
