To characterise the gut bacterial community structure of two cohabiting marine bivalves (oysters and mussels) across two different sampling months (one in the austral summer and one in the austral winter), the V1-V2 region of 16S rRNA gene was profiled from 116 bivalves (60 oysters and 56 mussels, ~30 per species/sampling month; with 4 mussel winter samples failing library preparation and sequencing). Comparison of length and weight measurements from individuals provided size class information that was used to infer potential cohort differences. Distinct size classes were observed for both oysters and mussels between the sampling periods (Supplementary Fig. 2). In the summer sampling month, oysters comprised smaller shell lengths and weights (mean 45.9 + SD 3.6 mm; mean 35.2 + SD 1.9 g) while in the winter sampling month they had larger mean shell lengths and weights (64.9 + 3.9 mm; 52.3 + 10.5 g). Similar differences were also observed for mussels, though those collected in summer appeared to consist of at least two separate size classes (and thus likely cohorts); one group with smaller shell lengths of <40 mm (18 mussels: 33.6 + 2.8 mm; 16.5 + 4.0 g) and one with larger shell lengths of >60 mm (12 mussels: 73.0 + 12.0 mm; 83.8 + 33.5 g). Mussels in winter comprised a mean shell length of 64.0 + 3.2 mm and a mean weight of 20.5 + 3.3 g. Accordingly, only matched size classes were used for bacterial community comparisons between bivalve species (i.e., small oysters vs small mussels in summer, and large oysters vs large mussels in winter). An independent comparison between mussels with different shell lengths obtained in summer (i.e. <40 mm and >60 mm) was conducted to explore for possible size class differences.
The bacterial communities of farmed oysters and cohabiting wild mussels were surveyed from aspirated gut contents, as well as from six seawater samples obtained from the same farm site (3 per sampling month). Of the total of 659 OTUs obtained for analysis, 105 were unique to bivalves, 15 to seawater and 539 that were shared between bivalves and seawater (Supplementary Datasheet 1). Despite the large number of shared OTUs, ordination of the samples revealed that the bivalve samples clustered independently to those obtained from seawater (Fig. 1). Furthermore, samples from oysters and mussels clustered independently of one another, and in association with the sampling month in which they were obtained (i.e., austral summer or winter). This observation was confirmed by two-way PERMANOVA, which crossed bivalve species with sampling month, revealing highly significant differences between oysters and mussels and between sampling months (pseudo-F = 66.31, p-value = 0.0001; pseudo-F = 40.92, p-value = 0.0001 respectively). However, there was a significant interaction effect between species and sampling month (pseudo-F = 18.74, p-value = 0.0001), indicating that changes between the summer and winter sampling months were species-specific.
In evaluating all samples, OTUs represented bacterial taxa belonging to 17 phyla, 28 classes, 90 orders, 150 families, and 285 genera, of which the two phyla Tenericutes and Proteobacteria accounted for >80% of the total OTU abundance (Fig. 2A). Unlike seawater, which was dominated by α- and γ-proteobacteria and to a lesser extent Bacteroidetes (Bacteroidia) and Actinobacteria (Acidimicrobiia and Actinobacteria), samples from bivalves largely consisted of Tenericutes (Mollicutes) as well as α-, γ- and δ-proteobacteria, Spirochaetes (Spirochaetia) and Cyanobacteria (Oxyphotobacteria and Sericytochromatia). The Tenericutes (Mollicutes) were largely associated with bivalve samples, accounting for ~52% of the total OTU abundance; as derived from a total of 36 OTUs, two most closely representing Spiroplasmataceae (<72% identity) – which were almost exclusively associated with mussels, and 34 most closely representing Mycoplasmataceae (<83% identity) – of which 31 were shared between both bivalve species (Supplementary Datasheet 1). The greatest proportion of Tenericutes occurred in summer for both oysters (mean 64.2 + SD 16.9%) and mussels (80.9 + 8.7%). In contrast, the phyla Proteobacteria had a lower mean abundance in summer compared to winter for both oysters (21.3 + 12.5% vs 56.7 + 18.4%) and mussels (6.3 + 3.0% vs 36.0 + 12.1%). These findings were supported by differential abundance analysis (as determined using the Kruskal-Wallis rank test, adjusted p-value cut-off = 0.01), which also revealed a summer vs winter trend in the proportions of other major taxonomic groups including, a higher abundance of Spirochaetes in both bivalves in summer and, conversely, a higher proportion of Bacteroidetes, Actinobacteria (Acidimicrobiia, Actinobacteria and Thermoleophilia), Cyanobacteria (Sericytochromatia), Firmicutes (Clostridia), Chloroflexi (Anaerolineae) and Epsilonbacteraeota (Campylobacteria) in winter (Fig. 2B, Supplementary Table 1). Two groups, however, appeared to have disparate abundances between the summer and winter sampling months, with Fusobacteria more abundant in summer in mussels and in winter in Oysters, and Verrucomicrobia in winter in mussels and in summer in oysters. The 10 most dominant bivalve associated OTUs accounted for >50% of the total standardised sequence reads and included taxa largely related to Mollicutes, including Mycoplasma spp. (OTU 7, mean abundance of 9.2%; OTU 6, 6.4%; OTU 4, 4.8%; OTU 51, 3.3%; OTU 19, 3.1%), Candidatus Bacilloplasma sp. (OTU 11, 5.1%) and uncultured Mycoplasmataceae spp. (OTU 2, 3.9%; OTU 14, 3.9%; OTU 17, 2.6%), as well as the γ-proteobacteria Halioglobus sp. (OTU 1, 8.9%) (Supplementary Datasheet 1).
Defining a role for host species in gut bacterial community composition in bivalves
To understand the influence of host species on bivalve gut bacterial communities, core (shared) and unique bacterial constituents were first evaluated from comparisons between all samples (irrespective of sampling month). Of the 644 OTUs that were detected from bivalves, only 35 were unique to oysters and 28 to mussels, with the majority (~90%) being shared (Fig. 3A). Of these, 13 OTUs from oysters and seven OTUs from mussels were not detected (or occurred in very low abundance) in seawater. The top three most prevalent in oysters related to taxa belonging to Anaplasmataceae (α-proteobacteria) (OTU 140 – Candidatus Neoehrlichia, min. 0 – max. 0.2%), Spirochaetaceae (Spirochaetes) (OTU 263 – Spirochaeta 2, 0 – 1.9%) and Mycoplasmataceae (Mollicutes) (OTU 9073 – Mycoplasma sp., 0 – 1.4%) (Supplementary Table 2), while in mussels the most prevalent were Mycoplasmataceae (OTU 115 – Mycoplasma sp., min. 0 – max. 3.6%; OTU 261 – Mycoplasma sp., 0 – 1.5%) and Spiroplasmataceae (OTU 180 – Spiroplasma sp., 0 – 1.4%) (Supplementary Table 3). Alongside this, in assessing the differentially abundant families and OTUs associated with these samples, certain taxa also appeared to be preferentially more abundant in one of the two bivalve species – as determined using the Kruskal-Wallis rank test (adjusted p-value cut-off = 0.01) (Supplementary Fig. 3A and B). In oysters, this included a total of 7 families, including the γ- and α-proteobacteriafamilies Halieaceae (OTU 1 – Halioglobus sp.), Kiloniellaceae (OTU 22 – Kiloniella sp.) and Pseudomonadaceae (OTU 16 – Pseudomonas alcaligenes), as well as an unclassified Sericytochromatia (Cyanobacteria), Epsilonbacteraeota families Helicobacteraceae and Campylobacteraceae, and a number of other OTUs belonging to Mycoplasma/uncultured Mycoplasmataceae (OTUs 6, 4, 2, 13, 1474, 281, 3 and 27) and Spirochaetaceae (OTU 26 – Salinispira sp., OTU 38 – uncultured Spirochaetaceae). In contrast, taxa belonging to 18 different families appeared to contribute to the differences observed for mussels. The most notable of these included Flavobacteriaceae (OTU 71 – Polaribacter sp., OTU 41 – Ulvibacter sp.), Rhodobacteraceae (OTU 59 – Sulfitobacter sp., OTU 43 – Planktomarina sp.) and Fusobacteriaceae (OTU 37 – Psychrilyobacter sp.). In addition, like that observed for oysters, a number of OTUs belonging to the Mycoplasmataceae and Spirochaetaceae also appeared to contribute to the observed differences, including OTU 11 (Candidatus Bacilloplasma sp.), OTU 14 and 17 (uncultured Mycoplasmataceae), OTU 51 (Mycoplasma sp.) and OTU 28 (Spirochaeta 2).
Ordination of samples from mussels obtained in summer with large (>60 mm) and small (<40mm) shell lengths, and in comparison, to samples from mussels in winter with large shell lengths (>60mm), revealed the independent clustering and likely differences in the global bacterial community compositions of samples belonging to these size classes (Fig. 4A). This observation was confirmed by one-way PERMANOVA, revealing a significant difference between the large and small summer mussels (pseudo-F = 4.5604, p-value = 0.0028). No significant differences were observed, however, between these groups for measures of species/OTU richness (p-value = 0.2253; large mussels: mean 244 + SD 48, small mussels: 263 + 37), Shannon diversity (p-value = 0.6574; large mussels: 2.13 + 0.59, small mussels: 2.04 + 0.48) and Pielou’s evenness (p-value = 0.5610; large mussels: 0.39 + 0.10, small mussels: 0.37 + 0.09). Furthermore, though some (albeit slight) changes were observed in the mean abundances of various bacterial classes (Fig. 4B), differential abundance analysis revealed the occurrence of only five significantly different OTUs – as determined using the Kruskal-Wallis rank test (unadjusted p-value cut-off = 0.01) (Supplementary Table 4). Based on Log LDA scores, the two with the largest effect size included those most closely related to Mycoplasma spp. – OTU 115 (LDA -4.99) which had a higher abundance in samples from small mussels, and OTU 51 (LDA 3.48) which had a higher abundance in samples from large mussels (Fig. 4C).
Environmental drivers of gut bacterial community composition in bivalves
In exploring the impact of the surrounding environment (seawater) on the gut bacterial communities from bivalves, where the average monthly water temperature varied by ~5°C (Table 1), notable differences between the sampling periods were observed. Overall, while samples from mussels comprised a greater number of OTUs compared to oysters in both summer (263 + 37 vs. 184 + 66) and winter (349 + 46 vs. 240 + 60) (Fig. 5A), changes in species/OTU richness and diversity (Shannon and Simpson’s diversity and Pielou’s evenness) were apparent for both bivalves between the two sampling months. Most notably, was a marked increase in these measures in winter (Fig. 5A-D). This observation was confirmed by two-way ANOVA, which crossed bivalve species with sampling month, revealing highly significant differences between summer and winter (p-value < 0.0001). However, there was a significant interaction effect between species and sampling month (p-value < 0.0001) for all measures except for species/OTU richness, indicating that while a similar increase in the number of OTUs occurred for both oysters and mussels in winter, there were likely species-specific differences in the types and/or relative abundances of these OTUs. A similar trend for measures of richness and diversity was also observed for the seawater samples between summer and winter, though the differences were not significantly different (p-value > 0.05). In comparing variation in taxonomic distinctness (varTD, lambda+) as a function of average taxonomic distinctness (avTD, delta+), which assesses the breadth and evenness of the taxonomic diversity of the OTUs within each sample, a significant difference was observed between bivalve species and sampling month (Fig. 5E). Samples from oysters typically comprised OTUs covering a greater breadth of taxa in summer compared to winter (based on a higher mean value for delta+: 91.39 + SD 0.49 vs 90.16 + 0.67 respectively), though were similarly evenly distributed across taxonomic lineages in both sampling months (based on similarly low mean values for lambda+: 259.97 + 21.98 and 258.29 + 15.72 respectively). In contrast, samples from mussels comprised OTUs covering a similar breadth of taxa in both summer and winter (based on similar mean values for delta+: 90.33 + 0.38 and 90.12 + 0.31 respectively), though were more unevenly distributed across taxonomic lineages in winter compared to summer (based on a higher mean value for lambda+: winter 282.27 + 10.84 vs summer 266.81 + SD 11.50). This observation was supported by the occurrence of a significant interaction effect between species and sampling month (for both measures of delta+ and lambda+), indicating that changes between the summer and winter sampling months were species-specific. Despite seawater samples comprising the greatest number of OTUs (Fig. 5A), these OTUs represented a substantially lower breadth of taxa and were more unevenly distributed across taxonomic lineages in both summer and winter compared to those from bivalves (based on lower values for delta+ and higher values for lambda+) (Fig. 5E).
Differential abundance analysis comparing oyster and mussel gut and seawater samples in summer and winter revealed a total of 29 families and 151 OTUs that were significantly different (as determined using the Kruskal-Wallis rank test, adjusted p-value cut-off = 0.01). In evaluating the 20 most differentially abundant families and OTUs with the greatest effect size (based on the Log LDA scores), distinct patterns were observed between the sampling months whereby a concomitant increase in the abundance of certain taxa was observed in either summer or winter in both bivalve species (Fig. 6, Supplementary Table 5 and 6). This included 5 families in summer (Mycoplasmataceae, Spirochaetaceae, Cyanobiaceae [Synechococcus and Cyanobium spp.], Methylophilaceae [OM43 clade] and Pseudoalteromonadaceae [Pseudoalteromonas spp.]) and 12 in winter (α-proteobacteria SAR11 clade 1a, Halieaceae, Rhodobacteraceae, Flavobacteriaceae, Burkholderiaceae, Pseudomonadaceae, Rhizobiaceae, Cryomorphaceae, Microbacteriaceae, Desulfobulbaceae, unclassified Sericytochromatia, and γ-proteobacteria SAR86 clade). Of these, five also had an associated high abundance in the seawater in either the summer or winter sampling months. This included Pseudoaltermonadaceae in summer, and α-proteobacteria SAR11 clade 1a (OTU 8), Rhodobacteraceae (OTU 43 and 65, Planktomarina spp.), Flavobacteriaceae (OTU 44, unclassified NS5 marine group) and γ-proteobacteria SAR86 clade in winter (as marked by asterisks in Fig. 6A and B). The majority of OTUs contributing to the observed differences between the summer and winter sampling months included those most closely related to members of the Mycoplasmataceae (notably Mycoplasma), whereby in mussels OTUs 7, 17, 19 and 51 were more abundant in summer, and OTUs 11 (Candidatus Bacilloplasma sp.) and 19 (Mycoplasma sp.) were more abundant in winter. A similar trend was observed for Mycoplasmataceae related taxa in oysters, whereby OTUs 6, 4 and 13 were more abundant in summer, and OTUs 2 and 1474 in winter. Other OTUs with a notable increase in abundance in either the summer or winter sampling months included OTU 1 (Halioglobus sp. 79.46% identity), OTU 16 (Pseudomonas pseudoalcaligenes, 99.08% identity) and OTU 22 (Kiloniella sp., 80.33% identity) in oysters in winter, and OTU 28 (Spirochaeta 2 sp., 77.24% identity) in mussels in summer (Fig. 6B, Supplementary Table 6 and Supplementary Datasheet 1).