Turnover in Life-Strategies Recapitulates Marine Microbial Succession Colonizing Model Particles

Abstract

Particulate organic matter (POM) in the ocean sustains diverse communities of bacteria that mediate the remineralization of organic complex matter. However, the variability of these particles and of the environmental conditions surrounding them present a challenge to the study of the ecological processes shaping particle-associated communities and their function. In this work, we utilize data from experiments in which coastal water communities are grown on synthetic particles to ask which are the most important ecological drivers of their assembly and associated traits. Combining 16S rRNA amplicon sequencing with shotgun metagenomics, together with an analysis of the full genomes of a subset of isolated strains, we were able to identify two-to-three distinct community classes, corresponding to early vs. late colonizers. We show that these classes are shaped by environmental selection (early colonizers) and facilitation (late colonizers) and find distinctive traits associated with each class. While early colonizers have a larger proportion of genes related to the uptake of nutrients, motility, and environmental sensing with few pathways enriched for metabolism, late colonizers devote a higher proportion of genes for metabolism, comprising a wide array of different pathways including the metabolism of carbohydrates, amino acids, and xenobiotics. Analysis of selected pathways suggests the existence of a trophic-chain topology connecting both classes for nitrogen metabolism, potential exchange of branched chain amino acids for late colonizers, and differences in bacterial doubling times throughout the succession. The interpretation of these traits suggests a distinction between early and late colonizers analogous to other classifications found in the literature, and we discuss connections with the classical distinction between r- and K-strategists.

Keywords: ecological succession; life strategies; marine bacteria; microbial assembly; neutral theory; omics; particulate organic matter (POM); r/K selection.

Figure 1

Relative abundances of Exact Sequence Variants (ESVs) for populations attached to beads made of pure substrates (first row) and present in the surrounding water (second row) for each substrate (columns) at the different time points. The third row represents the population dynamics of beads made of mixed substrates. Contiguous sections with the same color in a given bar represent different ESVs belonging to the same genera. Highlighted genera are those among the 20 most abundant on any of the substrates. The remainder ESVs are classified as “other”, and their small relative abundance leads to a black “continuum”. In the beads experiments, the horizontal bars on the top of the bar plots represent the classes obtained with unsupervised clustering, and the color represents the number of replicates assigned to the different classes at each time point.

Figure 2

Number of optimal classes and test of neutrality. (Left) The optimal number of classes (number of Dirichlet mixture components) is determined by the minimum of the posterior evidence of the fit. Results are for alginate. (Right) Main time intervals are determined by each class in single substrates. The time window is determined considering consecutive samples classified in the same class (i.e., sparse samples misclassified are non-indicated, see Figure 1 for details). For chitin, the whole interval is indicated since no clear time-windows are defined. Pseudo p-values of the HDP test for each class found in pure substrates for the complete and local models are indicated.

Figure 3

Phylogenetic turnover for experiments in alginate. (A) Mantel statistics indicating the correlation between the phylogenetic and time distances in beads of alginate. Both matrices are split into subsets corresponding to different ranges of distances, and an independent test performed for each subset. The middle point of each range is indicated in the x-axis, and the number of distances in each subset and the significance of each test are shown in the legend. For alginate, all comparisons were significant. (B) All-against-all comparison of the βNTI index (excluding diagonal values). The black thicker lines separate comparisons within and between the different phases. Significant values correspond to |βNTI| > 2, non-significant values are shown in yellow. Results for other substrates are provided in the Supplementary Material.

Figure 4

Summary of metagenomic features. (A) Principal component analysis of the predicted metagenomic profiles colored by the phase in which they were sampled. Difference in the mean proportions of genes between communities at the selection and facilitation phases in metagenomic experiments (B) and in PICRUSt predictions (C). Each row in the diagram represents genes classified in the KEGG pathway indicated. The first column represents the mean proportions of the genes in the pathway for each phase, and the second column represents the difference between those proportions. Adjusted Benjamini-Hochberg p-values and 95% CI intervals are indicated. Pathways with effect sizes lower than 0.1 (B) and 0.05 (C) were filtered to show a similar number of pathways.

Figure 5

Ecological strategies of isolated strains. (A) Phylogenetic tree of the isolates used in this study, inferred from multiple sequence alignment of the 120 GTDB-Tk bacterial marker genes present in sequenced isolate genomes. The predicted taxonomic family and strain identity is indicated in the leaf label. The ecological strategies of the isolates inferred from the trajectories of the ESVs with 100% sequence identity in the 16S amplicon sequencing experiments are shown in different colors. Differences in the mean proportions of genes grouped in KEGG pathways between isolates identified as having a preference for the attachment phase and those with a preference for the facilitation phase (B) and the same comparison between isolates with a preference for the selection phase and those with a preference for the facilitation phase (C). The first column represents the mean proportions of the genes in the pathway for each group of isolates, and the second column represents the difference between those proportions. Adjusted Benjamini-Hochberg p-values and 95% CI intervals are indicated. Only pathways with effect sizes larger than 0.1 are shown.

Figure 6

Analysis of specific pathways. (Left) Schematic of enzymatic processes encoded by taxa belonging to the three community types that assemble on polysaccharide particles. Attachment communities (gold) that are the first to form on particles and commonly encode a diverse array of high-affinity ABC transporters, specific for sugars, such as maltose (Mal) and ribose (Rib), amino acids including methionine (Met) and serine (Ser), transporters specific to small peptide signals/antimicrobials, inorganic phosphate (Pi), and zinc (Zn). They also have genes to synthesize BCAA. Selection communities (Blue) form in between attachment and facilitation communities. Taxa in these communities frequently encode flagella and chemotaxis genes and perform dissimilatory nitrate reduction generating ammonia. Facilitation communities (Red) are the last to form on particles and encode pathways that break down BCAA into acetyl-CoA (Ac-CoA) and propanoyl-CoA (Prp-CoA), metabolites that can enter central metabolism to make biomass or energy. Taxa in facilitation communities also commonly encode glutamine and glutamate (Gln) synthases and glutamine synthetase, which are the main pathways for the assimilation of ammonia-derived nitrogen into biomass. (Right) Predicted doubling times of isolates estimated from ribosomal proteins codon usage bias with gRodon.