Diversity, structure and convergent evolution of the global sponge microbiome

Abstract

Sponges (phylum Porifera) are early-diverging metazoa renowned for establishing complex microbial symbioses. Here we present a global Porifera microbiome survey, set out to establish the ecological and evolutionary drivers of these host-microbe interactions. We show that sponges are a reservoir of exceptional microbial diversity and major contributors to the total microbial diversity of the world's oceans. Little commonality in species composition or structure is evident across the phylum, although symbiont communities are characterized by specialists and generalists rather than opportunists. Core sponge microbiomes are stable and characterized by generalist symbionts exhibiting amensal and/or commensal interactions. Symbionts that are phylogenetically unique to sponges do not disproportionally contribute to the core microbiome, and host phylogeny impacts complexity rather than composition of the symbiont community. Our findings support a model of independent assembly and evolution in symbiont communities across the entire host phylum, with convergent forces resulting in analogous community organization and interactions.

Figure 1. Richness of individual samples from microbial communities in seawater, sediments and sponges.

Rarefaction curves of 16S rRNA gene diversity are shown for seawater (blue), sediment (brown) and sponge (orange) samples.

Figure 2. Variability of symbiont communities.

Intraspecific community dissimilarity measured as distance of samples to group centroids for 16S rRNA gene composition of different sponge species (orange) and habitats (blue: seawater; brown: sediment). Vertical bar represent the median, the box represent the first to third quartiles and whiskers show the lowest or highest datum within 1.5 times the interquartile range of the lowest and upper quartile, respectively. Names in brackets represent the abbreviations used in all subsequent figures. The number behind the brackets refers to the number of individual samples analysed per sponge taxon, seawater or sediment.

Figure 3. Taxonnomic profile of microbial communities.

Average phylum-level taxonomic profile of microbial symbiont communities in 81 different sponge species, seawater and marine sediments. Colour scale shows relative abundance in percentage within each host species. The phylum Proteobacteria is shown as individual classes (including unclassified Proteobacteria), which are indicated by an asterisk. Black squares indicate zero counts. Columns and rows of the heatmap are ordered by Bray–Curtis dissimilarity of their taxonomic profiles (except for seawater and sediments). Sponge species abbrevations are outlined in Fig. 2.

Figure 4. Combined richness of microbial communities in seawater, sediments and sponges.

Rarefaction analysis of 16S rRNA gene diversity of microbial communities in sponges and seawater. (a) Rarefaction curves for sponge species with more than 20 replicate samples as well as seawater from all sampled geographic regions. OTU diversity is at 97% sequence identity cutoff. (b) Rarefaction analysis of all sponge species with three randomly selected samples per sponge species. Sponge species abbrevations are outlined in Fig. 2.

Figure 5. Community similarity for microbial communities in sponges, seawater and sediments.

Clustering was performed using multi-dimensional scaling of Bray–Curtis distances. Sponge species abbrevations are outlined in Fig. 2.

Figure 6. Cumulative degree distributions for OTUs (black dots, bottom and left axes) and sponges (red dots, top and right axes).

Black dots correspond to the number of different host species (k) that contain a given OTU, represented as the cumulative probability of finding an OTU in the network with k or less-associated hosts (Pc(k)). Red dots correspond to the number of different OTUs (k) found in a given host species, represented as the cumulative probability of finding a sponge host with k or less-associated OTUs (Pc(k)). The OTU degree distribution followed a truncated power-law Pc(k)=k^−0.32 × e^−(k/7.44), while the sponge degree distribution followed an exponential given by Pc(k)=e^−(k/1,849). Blue and orange dots correspond to random degree distributions for OTUs and sponges, respectively, where the number of nodes and links from the empirical distribution is kept constant.

Figure 7. Prevalence of symbionts across different degrees of host-specificity.

Number of host species (degree) containing a given bacterial OTU in the bipartite sponge versus symbiont OTU network plotted against the fraction of individual samples where each OTU has been found among all the samples from their known host species. Each point represents an OTU and the red line is a smoothing spline fit to the data (see Methods). Blue dots represent OTUs that belong to sponge-specific sequence clusters (see Results, section ‘Sponge-associated diversity is enriched in specific sequence clusters').

Figure 8. Representation of generalist and cosmopolitan OTUs within the global network and aggregated core.

Frequency (density) distribution of degrees for the global bipartite network (dark grey) and the aggregated OTU cores (light grey) (a) and the proportion of OTUs with certain degree or higher present in both sets (b). In b, the x axis shows the proportion of OTUs in the global bipartite network, while the y axis shows the proportion of OTUs in the aggregated cores. The 1:1 line indicates the expected distribution, if degrees were evenly distributed across the global bipartite network and the aggregated cores. This analysis reveals an over-representation of generalist and cosmopolitan OTUs within the aggregated core, with the break occuring at k>12.

Figure 9. Representative network for the core microbiome of Ircinia oros.

Each node corresponds to a single OTU, and links illustrate the most probable inter-specific interactions (see Supplementary Fig. 5). Positive and negative interactions are depicted in blue and red, respectively. None of the inter-specific interactions are bidirectional, indicating either amensal (−, 0) or commensal (+, 0) interactions. For full taxonomic information of the nodes refer to Supplementary Fig. 7 and Supplementary Data 3).

Figure 10. Phylogenetic signal of the inverse Simpson's index (D).

In this multi-gene phylogeny of host sponge species, 100% Bayesian posterior probabilities (PP) are indicated by black circles at internal nodes, while grey circles indicate 95–99% PP. Nodes with <95% PP are not labelled. Black circles at the tips of the phylogeny are sized in proportion to the mean value of D calculated for the symbiotic microbial community associated with each host species. Multiple clades of sponges contain either high (for example, Aplysina) or low (for example, Mycale) values of D.