Patterns of metagenomic diversity in deep-sea hydrothermal deposits
We sequenced 42 metagenomes from 40 samples (38 hydrothermal vent deposit samples and two diffuse flow fluids) collected at deep-sea hydrothermal vents and a deep-sea volcano. These represent one of the largest global collections of metagenomes from such samples (Fig. S1, S2). This study spans vent deposit collections from 2004 to 2018, from deep-sea hydrothermal vent fields in the north Atlantic (Mid-Atlantic Ridge, MAR), east and southwest Pacific (East Pacific Rise, EPR; Eastern Lau Spreading Center, ELSC), a sedimented hydrothermal system (Guaymas Basin, GB), and a deep-sea volcano (Brothers volcano, BV) (Table S1).
In this study, de novo assembly of sequencing data and subsequent genome binning and curation (see methods for details) resulted in 2,983 bacterial and 652 archaeal draft metagenome-assembled genomes (MAGs with ≥ 50% completeness, Table S2). Of these, ~ 21% were > 90% complete, with < 5% contamination, and ~ 36% contained a 16S rRNA gene fragment. The MAGs were initially characterized phylogenetically using the Genome Taxonomy Database Toolkit (GTDB-Tk) (Fig. 1, 2, 3, Data S1, S2, S3, S4, S5) [17]. MAGs that could not be assigned to a known genus by GTDB-Tk were assigned to new genera using AAI with the recommended cutoffs in Konstantinidis et al. [18] (Table S3A, B). Shared phyla between most of the hydrothermal deposits (excluding samples from the highly acidic Brothers volcano sites, and the diffuse flow fluids) included the Halobacteriota (e.g. Archaeoglobaceae), Methanobacteriota (e.g. Thermococcaceae), Thermoproteota (e.g. Acidilobaceae, Pyrodictiaceae), Acidobacteriota, Aquificota (e.g. Aquificaceae), Bacteroidota (e.g. Flavobacteriaceae), Campylobacterota (e.g. Sulfurimonadaceae, Nautiliaceae, Hippeaceae), Chloroflexota, Deinococcota (e.g. Marinithermaceae), Desulfobacterota (e.g. Dissulfuribacteraceae, Thermodesulfobacteriaceae), Proteobacteria (e.g. Alphaproteobacteria, Gammaproteobacteria), and the Patescibacteria (Table S4). Many of these phyla have only a few representatives in isolated cultures and point to the importance of combining enrichment cultivation strategies with metagenomic approaches to obtain additional insights into the physiological ecology of these core lineages.
While shared taxa differed in relative abundance and distribution, observable differences in community structure between vent fields were somewhat limited in this study due to small sample numbers from some of the vent fields (two samples apiece from EPR; Rainbow, MAR; Lucky Strike, MAR), and the overall lower read depth of samples from these sites and a few other samples (Fig. S3). Therefore, obtaining statistically robust community structure patterns using MAG phylogenetic diversity for the entire dataset was not possible. However, Reysenbach et al. [8] did show that if metagenomic sequencing is deep, assembled MAG diversity tracks 16S rRNA amplicon diversity structure. Extrapolating to this study, the Brothers volcano MAG diversity patterns were retained and confirmed the amplicon observations from Reysenbach et al., 2020 [8] (Fig. S4), and in turn tracked the ELSC MAG community diversity (Fig. 4A, B). For example, sites at Brothers volcano that were hypothesized to have some magmatic inputs were predicted to be more similar in community structure to the sites along the ELSC with greater magmatic inputs, such as Mariner. Several of the samples from the more acidic Mariner vent field were more closely aligned in MAG diversity structure to those of the acidic solfataric Upper Cone sites at Brothers. The MAG data also demonstrated that the Guaymas samples were quite unique, which is not surprising, given that Guaymas Basin is a sediment-hosted system where the hydrothermal fluid geochemistry is quite different from other basalt- or andesitic-hosted hydrothermal systems (e.g., higher pH, high organics, high ammonia and methane) [19, 20].
Our dataset greatly broadens genomic diversity from deep-sea vents, by representing 511 novel genera, comprising 395 Bacteria and 116 Archaea. Notably, 52% (206) of the novel bacterial genera and 72% (84) of the novel archaeal genera were found at Brothers volcano. Furthermore, 25% (99) of the novel bacterial genera and 47% (54) of the novel archaeal genera were unique to the Brothers volcano samples, which further supports the understanding that this environment is a hotbed for novel microbial biodiversity, reflected in the volcano’s complex subsurface geology [8].
While many of these novel archaeal and bacterial genera were previously reported from Brothers volcano [8], we report them again here in the context of the new assemblies (1000 bp contig cutoff, used for Brothers volcano samples and ELSC 2015 samples) and iterative DAS Tool binning used for all our metagenomes. Our data support that of Reysenbach et al., [8], which used MetaBAT for assemblies (2000 bp contig cutoff) of the Brothers volcano metagenomes. Namely, we recovered approximately 202 novel bacterial genera and 83 new archaeal genera from Brothers volcano communities in Reysenbach et al. [8], well within the range detected in this analysis (viz. 206 and 84, respectively). In this study, using a lower contig cutoff allowed for the recovery of a much higher number of MAGs, but many are of lower quality with higher contig counts. For example, MAGs recovered in the Reysenbach et al. [8] study had an average of 254 contigs per MAG, with ~ 19% (135) of MAGs comprising 100 contigs or less. In contrast, only 7% (258) of MAGs in this current study had 100 contigs or less, and the average number of contigs per MAG was 511 (Table S2). However, using the iterative binning approach provided advantages when resolving lineages of high microdiversity, such as in the Nautiliales, with the caveat of creating some MAGs with large collections of erroneous contigs that were poorly detected by CheckM, as they had very few associated marker genes (e.g. MAGs 4571 − 419_metabat1_scaf2bin.008, M10_maxbin2_scaf2bin.065; Fig. S5). This points to the importance of carefully choosing assembly parameters depending on the ultimate goal of whether quality over quantity of MAGs is preferred for analyses of ecological patterns. Our data demonstrate, however, that overall patterns of MAG diversity are retained regardless of assembly techniques and parameters (Fig. S4).
Furthermore, here we document some of the first examples of medium to high quality MAGs from phyla and classes never previously identified, or poorly sampled, from deep-sea hydrothermal environments. These include Thermoproteia, Patescibacteria (formerly Candidate Phyla Radiation, CPR), Chloroflexota, and a few MAGs representing two putative new bacterial phyla, JALSQH01 (3 MAGs) and JALWCF01 (13 MAGs) (Supplementary Discussion, Fig. S6, Table S5). For example, with 249 MAGs belonging to the Thermoproteia (Table S2, Fig. S7), we have significantly expanded the known diversity and genomes from this phylum. The importance of this group at deep-sea vents was first recognized through 16S rRNA amplicon studies, where the depth of sequencing highlighted that much of this novel thermophilic diversity had been overlooked [e.g. 3, 4]. Furthermore, it is now recognized that many members of this group have several introns in the 16S rRNA gene, which explains why they were missed in original clone library assessments and may be underestimated in amplicon sequencing [21–24]. For example, 24 MAGs were related to a recently described genus of the Thermoproteia, Zestosphaera (GTDB family NBVN01) [24]. This genus was first isolated from a hot spring in New Zealand but is clearly a common member of many deep-sea vent sites. Further, the discovery of a 16S rRNA gene related to Caldisphaera at deep-sea vents [25], previously only detected in terrestrial acidic solfataras, led to the isolation of related Thermoplasmata – Aciduliprofundum boonei – but the Caldisphaera escaped cultivation. Here we report several high-quality MAGs related to this genus (M2_metabat2_scaf2bin.319, 131–447_metabat1_scaf2bin.050, M1_metabat1_scaf2bin.025, S016_metabat2_scaf2bin.003). Additionally, we also recovered a genome from the Gearchaeales (S146_metabat1_scaf2bin.098), first discovered in iron-rich acidic mats in Yellowstone National Park [26], and members of the poorly sampled Ignicoccaceae, Ignisphaeraceae, and Thermofilaceae. While we identified several genomes from recently discovered archaeal lineages including the Micrarchaeota, Iainarchaeota and Asgardarchaeota, we also recovered 15 MAGs belonging to the Korarchaeia, 14 of which comprise two putative novel genera, and one which is closely related to a MAG previously recovered from sediment in Guaymas Basin (Genbank accession DRBY00000000.1) [27, 28]. Additionally, we recovered four MAGs from the Caldarchaeales that span two novel genera, one of which was recently was proposed as Candidatus Benthortus lauensis [29] using a MAG generated from a previous assembly of the T2 metagenome (T2_175; Genbank accession JAHSRM000000000.1). MAGs belonging to this genus were identified at both Tui Malila, ELSC, and Brothers volcano (T2_metabat2_scaf2bin.284, S140_maxbin2_scaf2bin.281, S141_maxbin2_scaf2bin.262) with the Tui Malila MAG nearly identical (99.7% AAI similarity) to the described Cand. B. lauensis T2_175 MAG.
While within the Bacteria, the Gammaproteobacteria and Campylobacterota were by far the most highly represented bacterial genomes, there were other lineages for which we have very little if any data or cultures from deep-sea hydrothermal systems (Fig. 3, Fig. S7). Two such groups are the Patescibacteria and Chloroflexota, with 154 and 194 MAGs respectively.
Patescibacteria and Chloroflexota are diverse and abundant members of deep-sea hydrothermal vent deposits
The Patescibacteria/Candidate Phyla Radiation (CPR) encompasses a phylogenetically diverse branch within the bacterial tree of life that is poorly understood and rarely documented in deep-sea hydrothermal systems. Originally, the CPR was proposed to include several phylum-level lineages [30], but the entire group was later reclassified by GTDB as a single phylum, Patescibacteria [31]. Members of the Patescibacteria have been well-characterized in terrestrial soils, sediments, and groundwater [32–37], and in the mammalian oral cavity [38–40]. Several 16S rRNA gene and metagenomic studies have also identified members of the Patescibacteria from deep-sea vents, including EPR, MAR, ELSC, and Guaymas Basin [3, 4, 12, 15, 41–43], from Suiyo Seamount [44], and the Santorini submarine volcano [45], further supporting the widespread distribution of this metabolically diverse phylum.
Our study adds 56 novel genera based on AAI and GTDB classifications to the Patescibacteria phylum. These include large clades within the Gracilibacteria (10 new genera), representatives within the Microgenomatia (9 novel genera), Dojkabacteria (10 new genera), and several clades in the Paceibacteria (13 new genera) (Fig. 5A, B, Fig. S8). The Gracilibacteria and Paceibacteria were overall the most prevalent lineages of Patescibacteria in the samples but had contrasting distributions across vents (Fig. 5B). In general, when the Gracilibacteria were prevalent, the Paceibacteria appeared to be a minor component or not present, and vice versa. In particular, the Gracilibacteria MAGs were often associated with the acidic sites such as the Upper Cone at Brothers volcano (S011, S147), and the Mariner vent fields, and in the early colonization experiment from Guaymas Basin (Supplementary Discussion). This may suggest that Gracilibacteria function as early colonizers and are associated with turbulent ephemeral environments as observed previously in oil seeps [46]. Continued investigation into the ecology, evolution, and host association patterns of these groups, however, may shed more light on these distribution differences.
Consistent with previous studies [30, 34], many of the recovered Patescibacteria MAGs had very small genomes (often ~ 1MB or smaller; Table S2) with highly reduced metabolic potential, often lacking detectable genes for synthesis of fatty acids, nucleotides, and most amino acids (Table S6). Gene patterns also suggested that many of the organisms are obligate anaerobes, lacking aerobic respiration, and that they likely form symbiotic or parasitic associations with other microbes, as has been shown for Patescibacteria cultivated thus far from the Absconditabacterales and Saccharibacteria [39, 40, 47, 48].
We recovered several MAGs from Mariner, Guaymas Basin, and Brothers volcano that were related to the parasitic Cand. Vampirococcus lugosii [47] and Cand. Absconditicoccus praedator [48]. In order to explore if our MAGs had any hints of a parasitic lifestyle, we searched for some of the large putative cell-surface proteins identified in the genomes of Cand. V. lugosii [47] and Cand. A. praedator [48]. Using a local BlastP of nine of the longest genes found in Cand. V. lugosii, we recovered high-confidence homologs (E-value = 0) for alpha-2 macroglobulin genes in several MAGs from the Abscontitabacterales (based on search of Cand. V. lugosii protein MBS8121711.1), which may be involved in protecting parasites against host defense proteases [47]. We also recovered homologs for PKD-repeat containing proteins (MBS8122536.1; E-value = 0), which are likely involved in protein-protein interactions [47]. Previous analysis of Cand. V. lugosii found these giant proteins are likely membrane-localized, suggesting they may potentially play a role in host/symbiont interactions. Additionally, we identified these long proteins from Cand. V. lugosii elsewhere in the Gracilibacteria MAGs. For example, putative homologs of the PKD repeat containing protein (MBS8122536.1), a hypothetical protein (MBS8121701.1), and the alpha-2 macroglobulin (MBS8121711.1) were identified in multiple other orders of the class Gracilibacteria (E-value ≤ 1E-25). The alpha-2 macroglobulin was also identified in the very distantly related Paceibacteria, and a single putative homolog of the alpha-2 macroglobulin was found in a MAG belonging to the class WWE3 (134–614_metabat1_scaf2bin.084; E-value ≤ 1E-24).
While the Patescibacteria likely rely on symbiotic or parasitic relationships, members of the Chloroflexota phylum are diverse and metabolically flexible organisms, capable of thriving in a wide variety of geochemical niches. Chloroflexota are abundant and widely distributed in a variety of environments, including terrestrial soils, sediments and groundwater, freshwater, pelagic oceans, and the marine subseafloor and sediments [49–55], and hydrothermal settings such as Guaymas Basin [11] and Brothers submarine volcano [8]. Genomic evidence suggests that Chloroflexota are associated with important metabolisms in the carbon cycle, including fermentation, carbon fixation, acetogenesis and the utilization of sugars, polymers, fatty acids, organic acids and other organic carbon compounds [50, 51, 54].
Here we add to the growing evidence that the Chloroflexota are diverse and metabolically versatile members of deep-sea hydrothermal vent communities. We recovered a total of 194 Chloroflexota MAGs spanning 12 orders (GTDB taxonomy), which included 22 novel genera. Of these novel genera, 14 were identified at Brothers volcano and 6 were unique to the Brothers volcano samples (Table S3A). Based on read coverage, Chloroflexota MAGs were in high relative abundance (≥ 7%) in several samples from the ELSC, namely, from Tui Malila and ABE, and in one NW Caldera Wall sample from Brothers volcano (Table S4). To further explore the metabolic potential of Chloroflexota in hydrothermal vent communities, we focused our analyses on high-completeness MAGs (≥ 80% completeness, n = 58) distributed in 6 orders: Caldilineales, Promineofilales, Anaerolineales, Ardenticatenales, B4-G1, and SBR1031 (Fig. 6, Table S7A).
The majority (≥ 75%) of the high quality Chloroflexota MAGs encoded marker genes involved in several processes previously associated with the Chloroflexota (Table S7B), including fatty acid degradation [50, 55], formate oxidation [56], aerobic CO oxidation [57] and selenate reduction [53]. Except for the Anaerolineales, over 66% of the MAGs in the other five orders had the capacity for degradation of aromatic compounds, as previously reported for Chloroflexi from the marine subsurface [51]. While some MAGs had the potential for substrate-level phosphorylation through acetate formation, most of the MAGs contained pathways for oxidative phosphorylation and oxygen metabolism [50, 51]. The Wood–Ljungdahl pathway, the CBB cycle based on a Form I Rubisco, and the reverse TCA cycle were detected in some of the MAGs [50, 51]. Soluble methane monooxygenase genes, a metabolic potential recently also detected in a Chloroflexota MAG from the arctic [58], were identified in a total of eight of our MAGs from the orders Caldilineales, Anaerolineales, and Ardenticatenales.
Although the primary metabolic potential of the hydrothermal vent-associated Chloroflexota was in carbon cycling, we did, however, observe minor evidence for their roles in nitrogen and sulfur cycling (Fig. 6, Table S7). About 22% of the MAGs (with ≥ 80% completeness) encoded capacities for sulfide oxidation, as previously reported for members of this group, e.g. Chloroflexus spp. [59, 60]. The potential to disproportionate thiosulfate was also observed in a few MAGs. Further, thermophilic Chloroflexi grown in an enrichment culture from Yellowstone National Park were shown to oxidize nitrite. A few of our MAGs encoded genes involved in nitrite oxidation [61], while a larger proportion of the MAGs encoded genes for nitrite or nitric oxide reduction. None of the MAGs encoded complete pathways for entire sulfur oxidation or denitrification, suggesting that Chloroflexota in these environments may be associated with metabolic handoffs involving other community members (see below).
Metabolic and functional diversity in deep-sea hydrothermal vent deposits
In order to explore the metabolic and functional diversity associated with our MAGs, we utilized functional assignment results in tandem with the corresponding MAG relative abundance (Table S8). In general, genes involved in carbon, nitrogen, sulfur, and hydrogen metabolism were prevalent and shared across all hydrothermal systems in this study (Fig. 7, 8). While heterotrophy, autotrophy and mixotrophy potential were identified in all samples, 47.1% of the MAGs (by count) exhibited potential for carbon fixation. Marker genes associated with five different carbon fixation pathways were identified in the MAGs, namely, the Calvin-Benson-Bassham (CBB) cycle (form I or form II Rubisco), the 3-hydroxypropionate/4-hydroxybutyrate cycle, the dicarboxylate/4-hydroxybutyrate cycle, the reverse TCA cycle, and the Wood–Ljungdahl pathway (Fig. 7, 8). Marker gene presence also suggested the potential for widespread heterotrophic metabolism of peptides, polysaccharides, nucleotides and lipids, and fermentation via acetogenesis (Fig. 7, 8).
Genes involved in nitrogen fixation, denitrification, and nitrite oxidation were identified across the different hydrothermal sites, yet the potential for anaerobic or aerobic ammonia oxidation was rarely detected (Fig. 8). The absence of ammonia oxidation is not totally surprising, since ammonia is in very low to undetectable concentrations in deep-sea hydrothermal fluids, with the exception of sediment-hosted hydrothermal areas like at Guaymas Basin [19, 20]. In these sedimented hydrothermal systems, aerobic and anaerobic ammonia oxidation are key processes within the sediments and hydrothermal plumes [62–65], but they may not be as important in the hydrothermal deposits. Our data also expands the importance of nitrogen fixation from the first detection at deep-sea vents in Methanocaldococcus [66] to a greater diversity of hydrothermal Bacteria and Archaea.
Given the importance of sulfur cycling in deep-sea hydrothermal systems [67–69], it is not surprising that genes associated with elemental sulfur, sulfide, and thiosulfate oxidation, sulfate reduction, and thiosulfate disproportionation were widely distributed in MAGs from different hydrothermal samples and were associated with diverse taxonomic guilds (Fig. 7, 8). Based on metabolic gene distribution statistics (Table S9), the potential for sulfur oxidation was identified in 16% of the MAGs (577), primarily in members of the Alphaproteobacteria and Gammaproteobacteria. Genes associated with sulfide oxidation were identified in 34% of the MAGs (1216), including members of the Bacteroidia, Campylobacteria, Alphaproteobacteria, and Gammaproteobacteria. Thiosulfate oxidation genes were detected in 23% of the MAGs (836), largely comprised of the Campylobacteria, Alphaproteobacteria, and Gammaproteobacteria, while 14% of the MAGs (522) encoded genes for thiosulfate disproportionation, including the classes Bacteroidia, Campylobacteria and the phylum Desulfobacterota. The potential for dissimilatory sulfite reduction was identified in 6% of the MAGs (220) distributed across ten bacterial and archaeal phyla, namely Halobacteriota (class Archaeoglobi), Bacteroidota (class Kapabacteria), Campylobacterota (class Campylobacterales), Zixibacteria, Gemmatimonadota, Acidobacteriota, Nitrospirota, Desulfobacterota, Desulfobacterota_F and Myxococcota.
Hydrogen is highly variable in hydrothermal fluids, with some of the highest concentrations in geothermal systems hosted by ultramafic rocks, such as the Rainbow hydrothermal vent field [3], or in sediment-hosted regions like Guaymas basin [70]. In these systems, methanogens and sulfate reducers are prevalent hydrogen consumers [3, 71–74], although a wide variety of other heterotrophs and autotrophs can also derive energy from hydrogen oxidation [72]. Hydrogenase enzymes are responsible for mediating hydrogen oxidation in microbial populations but are also involved in a variety of other functions, including hydrogen evolution, electron bifurcation, and hydrogen sensing [75]. Approximately 27% of the MAGs in this study (974) encoded for at least one hydrogenase gene, and the MAGs were predominantly associated with the classes Campylobacteria, Bacteroidia, Gammaproteobacteria, and the phylum Desulfobacterota (Fig. 7, 8, Table S9). In several cases (132 MAGs), hydrogenase genes co-occurred with genes involved in the oxidation of reduced sulfur species (sulfide, elemental sulfur, sulfite or thiosulfate). This is not surprising, given that the capability to oxidize both sulfur and hydrogen has been shown in multiple isolates, including members of the Campylobacteria [76–78] and Aquificae [e.g. 79, 80].
Metabolic handoffs are a central feature of community interactions in deep-sea hydrothermal vent deposits
The microbial communities at deep-sea hydrothermal vents are shaped by a wide variety of complex interactions, including symbiosis, syntrophy, commensalism, cross-feeding, and metabolic handoffs [11, 12, 81–83]. While many of the MAGs encode genes associated with different biogeochemical cycles, as expected, the genes for a complex functional pathway were not localized in a single MAG, but instead distributed across several MAGs. This is likened to ‘metabolic handoffs’ where the interaction between different organisms produces pathway intermediates, enabling community members to perform downstream reactions in the metabolic pathway. For example, metagenomic analysis of a subsurface aquifer environment suggested that metabolic handoffs are commonly utilized in key biogeochemical pathways such as sulfide oxidation and denitrification [37]. Genes for sulfide oxidation were identified in all the deep-sea hydrothermal vent sites in this study, but few MAGs encoded genes for the entire three-step pathway. A much larger proportion of the MAGs, however, contained genes for a single step in sulfur oxidation (Fig. 9), consistent with a metabolic handoff scenario. Similar patterns were also observed for sulfate reduction and denitrification (Fig. 9). Additionally, the genes for individual steps in sulfide oxidation were often found coupled with at least one gene from the denitrification pathway, which may increase the thermodynamic favorability of both pathways (Li et al., 2018). Additionally, one or more denitrification genes co-occurred with sulfide oxidation genes in 1113 MAGs, with elemental sulfur oxidation genes in 485 MAGs and with sulfite oxidation genes in 1025 MAGs (Table S9).
Conserved microbial functions are mediated by different taxa at different hydrothermal vent systems
Previous analyses of deep-sea hydrothermal environments and global oceans have pointed to widespread functional redundancy in microbial communities [8, 12, 84, 85], with similar metabolic potential identified across taxonomically diverse samples. For example, a study of Guaymas Basin metagenome-assembled genomes suggested that many functional genes could be identified across multiple distinct taxa [12]. In our study, members of the Campylobacteria and Gammaproteobacteria were present in almost all samples, yet showed contrasting patterns of abundance (Fig. 10). These lineages can perform several of the same functional processes including oxidation of reduced sulfur species [86], denitrification [87–89], and carbon fixation [90–93]. This can be partially explained by ecophysiological and growth differences between the groups, which are selected for by the different geochemical profiles at the various vent sites. For example, studies have suggested that Campylobacteria tend to favor higher sulfide conditions but have a broader range of oxygen tolerance than the Gammaproteobacteria, while Gammaproteobacteria tend to inhabit a narrower range of higher oxygen and lower sulfide [16, 86, 90, 94]. It is therefore not surprising that the Campylobacteria were more prevalent at several of the acidic and more turbulent sites, such as at the Upper Cone, Brothers volcano and in early colonized samples from a thermocouple array at Guaymas Basin (Table S4, Supplementary Discussion). Patwardhan et al. [95] also showed that Campylobacteria were early colonizers of shallow marine vents followed by Gammaproteobacteria, and their differential colonization could be linked to sulfide, oxygen, and temporal differences.
The covariation of the Campylobacteria and Gammaproteobacteria in our data also coincided with genes for key functional processes associated with these taxa (Fig. 10). Thus, the overall ecological function contributed by the Campylobacteria and Gammaproteobacteria to the community at all sites was similar, but carried out by either one; viz., same guild different taxa. For example, relative gene abundance for 15 out of 22 broadly distributed functions – including heterotrophy associated with various organic carbon compounds, respiration of oxygen and nitrogen compounds, and oxidation of reduced sulfur compounds – tracked the relative abundance of each group. However, genes for some functions were exclusively represented by either group (Fig. 10, Table S10). For example, marker genes for formaldehyde oxidation, urea utilization and elemental sulfur oxidation were found in the Gammaproteobacteria but were hardly detected in Campylobacteria, while genes associated with thiosulfate disproportionation were attributed almost exclusively to Campylobacteria. In some cases, metabolic analysis also suggested that both Campylobacteria and Gammaproteobacteria had similar metabolic capabilities but encoded different pathways for the same functions. For example, consistent with a previously observed but non-ubiquitous trend [90–93], Campylobacteria mostly encoded genes for the rTCA cycle while the Gammaproteobacteria encoded genes for the CBB cycle. Both taxa also showed the potential for nitrite reduction to ammonia, with more nrfADH genes identified in the Campylobacteria and nirBD only found in the Gammaproteobacteria.