Introduction

Atlantic Salmon (Salmo salar) are non-obligate anadramous salmonids of significant commercial, cultural and recreational importance. Growth, development and migration in anadromous S. salar involves a radical shift across an ecological and trophic spectrum (Jacobsen and Hansen, 1999; Orlov et al., 2006). Accompanying the physiological, behavioural and dietary adaptations necessary to cope with transition between freshwater and marine environments (McCormick et al., 2013), significant and potentially adaptive shifts in host-associated microbiota might be expected.

The advent of culture-free microbial meta-sequencing techniques means that bacterial communities can be profiled in unprecedented detail. As such, teleost-associated intestinal microbiota are increasingly subject to scrutiny (Llewellyn et al., 2014). Salmonids have received particular attention in view of their importance in aquaculture (for example, Zarkasi et al., 2014) and the urgent need for innovation in feed sources (Green et al., 2013). Attempts to establish the natural identity and stability of commercially exploited teleost intestinal microbiomes have been limited to focal studies from single aquaculture facilities (for example, Zarkasi et al., 2014) and single sites in the wild (Star et al., 2013). The ecological succession of gut bacterial phylotypes during wild teleost development and migration is an excellent system in which to explore the relative contribution of host and environmental factors to shaping microbiome recruitment, especially in euryhaline species (Schmidt et al., 2015). Furthermore, exploration of the biogeography of microbiome composition among species with wide geographic distributions is required to form a sound baseline for experimental study. In this study, we set out to explore microbiome ontogeny and biogrography in wild S. salar.

Materials and methods

Bacterial 16S SSU rRNA (V4 region) diversity was profiled from the intestinal contents of 96 wild S. salar on an Illumina MiSeq platform (Illumina, San Diego, CA, USA) and analysed using Mothur (Schloss et al., 2009) and QIIME (Caporaso et al., 2010) (full methods in Supplementary Data S1). Juveniles and returning adults were sampled in eastern Canada and western Ireland. Marine adults were sampled in cooperation with the West Greenland subsistence fishery (www.nasco.int) (Supplementary Table S1).

Results

Study site (or country of origin) had no discernible effect on microbiome identity among samples from freshwater life-cycle stages, including returning adults (P=0.7264, PERMANOVA UniFrac distances (PUd); Oksanen et al., 2015—a full list of statistical tests applied and sample sizes are included in Supplementary Table S2). Nonetheless, study-site specific variation in intestinal operational taxonomic unit (OTU) richness (Chao1) was a consistent feature across juvenile life-cycle stages (Figure 1a, P=0.003, Kruskal–Wallis test (K-WT)). Diversity estimates (Shannon Index) corroborate this finding (Supplementary Figure S1, P=0.009).

Figure 1
figure 1

Alpha and beta diversity comparisons of Atlantic Salmon intestinal microbiota. (a) Box plot showing alpha diversity (Chao1 richness estimator) variation across sites and life-cycle stages: blue bars—Burrishoole, Ireland; pale blue bars—Erriff, Ireland; red bars—St Jean, Canada; pink bars—Trinite, Canada. Marine adults from West Greenland are represented in green. (b) Mean beta diversity distances (unweighted unifrac) among individuals grouped by site and life-cycle stage. Error bars represent s.e.m. (c) Principal coordinate analysis of pairwise unweighted unifrac distances between all salmon and environmental samples. Axes represent the two synthetic variables explaining the greatest proportion of variation in the data set.

Full size image

In contrast to those between-study sites, clear differences were observed in the microbial community identity of the S. salar intestine between freshwater and marine life-cycle stages (P<0.019; PUd; Oksanen et al., 2015; Figure 1b), in which returning adults retain a large proportion of their microbial assemblage from the marine environment. Microbiome identity within fresh and saltwater ecotopes was not impacted by life-cycle stage (smolt vs parr: P=0.268; marine vs returning adult P=0.522; PUd). Over study sites, inter-individual microbial composition was most stable among parr and marine life-cycle stages and least stable among smolts and returning adults (open and green bars, Figure 1c). OTU richness (Chao1) was dramatically affected by life-cycle stage across marine and freshwater environments (P=0.0001; K-WT) but not diversity (Shannon, P=0.276). By contrast, both OTU richness and diversity were purged during the transition from parr to smolt in freshwater (Chao1, P<1.8 × 10−5, Shannon, P=0.0105; K-WT; Figure 1a,Supplementary Figure S1).

Discussion

Phylum-level assignment of OTUs (Figure 2b) indicated the dominance of Proteobacteria among all samples. All life-cycle stages, especially marine adults, were enriched for Tenericutes (Genus Mycoplasma especially). In contrast, Firmicutes, Bacteriodetes and Actinobacteria—abundant in returning adults, smolt and parr—occur at negligible levels in marine adults. Both Tenericutes and Firmicutes OTUs occur in freshwater environmental samples but at minimal levels with respect to those recovered from fish. By contrast, Verrucomicrobia, common among freshwater samples, occurs at minimal levels in freshwater life-cycle stages. Although Verrucomicrobia is common in soil, freshwater and marine environmental samples (Freitas et al., 2012), as well as a frequent member of mammalian gut flora (Zhang et al., 2009), our study corroborates the low abundance of Verrucomicrobia found in fish found previously (for example, Rawls et al., 2006). To buffer inter-sample variability among OTUs assigned to genus level, we first evaluated the diversity of core OTUs (defined here as those occurring in 85% of individuals) present among fish from each life-cycle stage (Figure 2a). OTUs assigned to genus Yersinia and other unclassified Enterobacteriacae dominate the core microbiota of freshwater parr. Some Yersinia species (Y. ruckeri, Y. intermedia) are important pathogens of salmon (Bruno et al., 2013). However, the healthy state of the parr we sampled, as well as several single-nucleotide polymorphisms between the principal Yersinia OTU in our data set and the 16S V4 region of both Y. ruckeri, and Y. intermedia (data not shown), suggest that the Yersinia we sampled were likely to be commensals. Enterobacteriacae were also abundant among smolt, however, not among the core at 85%—unsurprising given raised beta diversity within this group (Figure 1b). Genus Mycoplasma phylotypes were the most abundant and consistently recovered phylotypes from adult salmon. More typical members of marine teloest intestinal microflora—Genera Allivibrio and Photobacterium (for example, Sullam et al., 2012)—were also well represented. OTUs attributable to family Mycoplasmataceae were also recovered in large numbers from other freshwater life-cycle stages. Clear differences in the frequency distribution of Mycoplasmataceae OTUs exist between adult and juvenile salmon (Figure 2c). Biogeographic differences in Mycoplasmataceae OTU distributions between Canada and Ireland were apparent in juveniles but not in returning adults (Figure 2c). With the exception of a single Mycoplasma OTU isolated from the Trinite river, Mycoplasmataceae OTUs abundant in parr and smolt were absent or rare (<5/11 000) in local freshwater samples.

Figure 2
figure 2

Taxonomic composition of the Atlantic Salmon intestinal microbiome. (a) Phlyum-level composition of total OTU abundances among distinct life-cycle stages and environmental samples. (b) Core (present in ⩾85% of individuals) 97% identity OTUs assigned to genus level from each life-cycle stage are represented by red-outlined ellipses. Black-outlined ellipses denote the presence of these ‘core’ OTUs among other life-cycle stages. Ellipse area is proportional to the mean abundance of OTUs assigned to each genus over all samples from each life-cycle stage. Core genera that occur at a mean frequency >1000 in each sample time are bold (adults), underlined (parr) or italicised (freshwater). (c) Heatmap displaying the frequency distribution of OTUs belonging to family Mycoplasmataceae across distinct life-cycle stages and countries of origin. Genera within the Mycoplasmataceae are indicated on the maximum likelihood phylogeny (left) on which values indicated the percentage of bootstrap support for the respective clades. Single asterisk indicates the Candidatus OTU also recovered from a sympatric environmental sample. Double asterisk indicates the most abundant core Mycoplasma OTU recovered from adult life-cycle stages.

Full size image

Both bacterial OTU richness and community stability declined over life-cycle stage in the intestine of S. salar, in stark contrast to mammals where community diversity increases after weaning and stabilises in early-to-late childhood (Yatsunenko et al., 2012). Returning adult salmon were characterised by low richness, highly variable microbial assemblages in comparison to parr. Although poor in absolute numbers of OTUs (that is, richness), diversity estimates from returning adults were not significantly different from juveniles, suggesting a fairly even frequency distribution of those OTUs present. Dietary complexity in juvenile salmonids could explain rich associated microbial assemblages (Orlov et al., 2006). Meanwhile physiological disturbances and fasting in migratory phases (smolt and returning adults) could underlie reduced community stability with respect to corresponding non-migratory phases (that is, parr and marine adults, Figure 1b). In particular, drinking rates increase during smoltification, as well as overall intestinal fluid re-absorption rates, perhaps affecting microbiome equilibrium (Stefansson et al., 2008). Additionally, documented variation in the response of different intestine regions (midgut vs hindgut) during smoltification could impact associated microbiota accordingly, an interesting future avenue for investigation (Stefansson et al., 2008). The microbial community of feeding marine adults was less rich and diverse than that of freshwater juveniles; perhaps attributable to the dominance of Mycoplasmataceae phylotypes among adult intestinal microbiota. Mycoplasmataceae, especially genera Candidatus and Mycoplasma, frequently colonise vertebrate and invertebrate mucosae, both as pathogens and commensals (Frey and Herrmann, 2002; Holben et al., 2002; Nechitaylo et al., 2009). Indeed, Mycoplasma have been isolated from S. salar in the past (Holben et al., 2002; Zarkasi et al., 2014). The abundance of Mycoplasmataceae (and individual OTUs, Figure 2c) among sites suggests an association with the salmon gut niche robust to developmental change and could point to some more complex level of interdependence with the host. In a recent laboratory study of microbiome ontology in euryhaline fishes, Schmidt et al. (2015) suggest a dominant role for deterministic forces (for example, niche appropriation) over neutral ones (for example, colonisation) (Schmidt et al., 2015). Mycoplasmataceae abundance and diversity in this study suggest a dual role in the wild: geography and environment influence colonisation source (and thus a proportion of the microbiome variation at the genus level); however, the intra-host niche likely determines the abundance of Mycoplasmataceae in general across S. salar. More widely, a combination of deterministic host effects and stochastic environmental factors underpin diversity in the S. salar microbiome whereby the microbiota of freshwater juvenile and returning adults, while sharing many OTUs with local environmental samples, show radically different patterns of abundance and enrichment. Broad-scale shifts in the composition of key components of S. salar gut microbiomes pose fundamental questions in relation to functional significance of qualitative change. Such inferences demand an experimental approach to assess empirically the impact of microbiome diversity on fish health and survival in distinct environments, especially in the context of aquaculture.