Stochastic and Deterministic Assembly Processes in Seamount Microbial Communities

ABSTRACT

Seamounts are ubiquitous in the ocean. However, little is known about how seamount habitat features influence the local microbial community. In this study, the microbial populations of sediment cores from sampling depths of 0.1 to 35 cm from 10 seamount summit sites with a water depth of 1,850 to 3,827 m across the South China Sea (SCS) Basin were analyzed. Compared with nonseamount ecosystems, isolated seamounts function as oases for microbiomes, with average moderate to high levels of microbial abundance, richness, and diversity, and they harbor distinct microbial communities. The distinct characteristics of different seamounts provide a high level of habitat heterogeneity, resulting in the wide range of microbial community diversity observed across all seamounts. Using dormant thermospores as tracers to study the effect of dispersal by ocean currents, the observed distance-decay biogeography across different seamounts shaped simultaneously by the seamounts’ naturally occurring heterogeneous habitat and the limitation of ocean current dispersal was found. We also established a framework that links initial community assembly with successional dynamics in seamounts. Seamounts provide resource-rich and dynamic environments, which leads to a dominance of stochasticity during initial community establishment in surface sediments. However, a progressive increase in deterministic environmental selection, correlated with resource depletion in subsurface sediments, leads to the selective growth of rare species of surface sediment communities in shaping the subsurface community. Overall, the study indicates that seamounts are a previously ignored oasis in the deep sea. This study also provides a case study for understanding the microbial ecology in globally widespread seamounts.

IMPORTANCE Although there are approximately 25 million seamounts in the ocean, surprisingly little is known about seamount microbial ecology. We provide evidence that seamounts are island-like habitats harboring microbial communities distinct from those of nonseamount habitats, and they exhibit a distance-decay pattern. Environmental selection and dispersal limitation simultaneously shape the observed biogeography. Coupling empirical data with a null mode revealed a shift in the type and strength, which controls microbial community assembly and succession from the seamount surface to the subsurface sediments as follows: (i) community assembly is initially primarily driven by stochastic processes such as dispersal limitation, and (ii) changes in the subsurface environment progressively increase the importance of environmental selection. This case study contributes to the mechanistic understanding essential for a predictive microbial ecology of seamounts.

INTRODUCTION

Seamounts, both isolated and clustered, are ubiquitously scattered in the world’s oceans. Breaching the ocean floor, these features rise off the seafloor many hundreds of meters and more, becoming underwater islands (1–4). Their geographic isolation may act as a barrier to the movement of macroorganisms and microorganisms between seamounts. By the time the Census of Marine Life’s Global Census of Marine Life on Seamounts (CenSeam) began in 2005, various researchers had arrived at a consensus view of seamounts as unique environments and hot spots of marine life in the open ocean (5, 6) that often feature characteristic faunas that are quite different from those in abyssal habitats (7). Some studies have shown that larvae colonize different isolated seamounts on the seafloor through passive dispersal, combined with the effects of local environmental selection, resulting in new species assemblages, endemism, and biodiversity (3, 8, 9). However, to date, fewer than 300 seamounts have been sampled worldwide, a mere fraction of the 23 million seamounts estimated to occur in the ocean. Emerson and Moyer (10) reviewed published work using a culture-independent approach and found only 19 papers that focused on 24 seamount locations. Most of this research investigated the surface hydrothermal habitats associated with volcanically active seamounts. Compared to the ecology of seamount macroorganisms, the mechanisms underlying microbial community assembly, succession, and biogeography are much less understood in general.

Understanding the mechanisms controlling microbial community assembly, succession, and biogeography is a central topic in microbial ecology (11). Niche and neutral theories are two major theories for microbial assembly and succession in which processes are controlled by various deterministic or stochastic forces. Niche theory assumes that deterministic factors such as species traits, interspecies interactions, and environmental conditions govern microbial communities (12, 13). Neutral theory hypothesizes that microbial communities are independent of species traits and environmental conditions and instead are governed by the stochastic processes of birth, death, colonization, immigration and emigration, and historical contingency (14–16). Microbial biogeography was traditionally thought to be governed by the combined effects of contemporary environmental selection (biotic and abiotic factors) and historical processes (including past selection and dispersal limitation) (17). It is believed that in many cases, microbial community assembly and succession processes could influence observed biogeography patterns (18–22).

Vellend (23), Hanson et al. (13), and Nemergut et al. (24) set up a comprehensive framework that integrates the concepts of microbial assembly, succession, and biogeography. Four ecological processes, namely, speciation, selection, dispersal, and ecological drift, underlie observed microbial patterns in species abundances across time and space. This framework unifies microbial community assembly, succession, and biogeography by considering both deterministic (e.g., selection) and stochastic (e.g., dispersal, drift, diversification, or speciation) processes (16), and it provides an effective operational model under which all microbial communities across different habitats can be compared under the same conceptual framework. Subsequently, the null model framework was developed by Stegen et al. (21) to quantify the relative contributions of various ecological processes to microbial community assembly, succession, and biogeography.

However, microbial dispersal processes are difficult to discern in the ocean due to the small size, wide distribution, high abundance, and short generation time of microorganisms. Interestingly, multiple studies have revealed the presence of dormant endospores of thermophilic members of the bacterial phylum Firmicutes (thermospores) in cold marine sediments (25–28). The inactivity of these thermospores enables their observed patterns in species abundances across space and time to be investigated largely without any confounding influence of environmental selection. The presence of dormant thermospores in these sediments can be explained by a dispersal history originating in the deep biosphere where the upward migration of fluids at hydrothermal field transports thermophiles into the overlying ocean (26). Thermospores can disperse and survive throughout the ocean and sediment over thousands of years (27, 29, 30), allowing passive dispersal to be disentangled from other ecological processes for tracking microorganisms’ movement in the ocean. Hence, the thermospores have been proposed as a natural model for selectively investigating the dispersal of microbial cells in the oceans (27, 30, 31).

The South China Sea (SCS) is the world’s largest semienclosed sea. The seafloor spreading that opened the SCS began in the early Eocene (~32 Ma) and ended in the middle Miocene (~15.5 Ma) (32). The SCS has a unique landscape with numerous postspreading basaltic seamounts in its basin (Fig. 1) (33). Furthermore, SCS bathypelagic seawater is transported into the basin only from the western Pacific through the Luzon Strait (34). This transportation creates basin-scale rapid cyclonic circulation (35), resulting in a short residence time for SCS seawater, between 30 and 100 years (35). Müller et al. (29) detected thermospores in the SCS sediments. There is no active hydrothermal field in the SCS, so these thermospores must have been transported into the SCS from the western Pacific through the Luzon Strait (34). All these attributes make the SCS seamounts an ideal natural laboratory for studying seamount microbial community diversity, assembly, succession, and biogeography. In this study, the samples from surface sediments (0 cm depth) and subsurface sediments (~15 to 35 cm depth) on the summit of 10 seamounts in the SCS Basin were collected, using 16S rRNA gene sequencing and thermospores to explore the ecological processes governing the microbial community, diversity, assembly, succession, and biogeography of SCS seamounts. To the best of our knowledge, this study provides the most comprehensive picture of seamount microbial ecology in the SCS. It also provides a framework for future studies on microbial life in global seamount areas.

FIG 1.

Location and bathymetric profile of the SCS seamount sediment cores. Map of the South China Sea showing the 10 seamount sampling locations and bathymetry of the study area.

RESULTS AND DISCUSSION

Seamounts enhance seafloor biodiversity and harbor distinct microbial communities.

(i) Microbial community abundance, richness, and evenness of seamounts ecosystems. Ten seamounts in the SCS were selected for sampling. They covered a wide range of water depths (1,805 to 3,827 m, including seven seamounts deeper than 3,000 m) (Fig. 1; see Data Set S1A in the supplemental material), a wide range of horizontal distances (69 to 1,114 km) (Data Set S1B), and a wide range of vertical distances between two seamounts (415 to 1,897 m) (Data Set S1B). The push core sediments were collected at sample depths ranging from 0.1 to 35 cm at the top of the 10 seamounts. Cell densities were investigated using catalyzed reporter deposition-fluorescent in situ hybridization (CARD-FISH). They ranged from 7.3 × 10⁷ to 9.5 × 10⁹ cells g⁻¹ (Fig. 2A), which was more than 1 to 2 orders of magnitude higher than nonseamount deep sediment samples in the SCS Basin from similar sampling depths (10⁶ to 10⁸ cells g⁻¹) (36, 37) (Fig. 2A) and showed moderate to high levels of microbial biomass compared with global moderate to heavy organic carbon surface sediment samples (10⁸ to 10⁹ cells g⁻¹) (38).

FIG 2.

Comparison of the SCS seamount sedimentary microbiome. (A) Microbial abundance, Chao1 richness, Shannon index, and TOC content with water depth in the SCS seamount surface sediment (orange frame) and the SCS nonseamount surface sediment (gray frame). (B) Thermospore abundance, Chao1 richness, and Shannon index with water depth in the SCS seamount surface sediment. (C and D) Richness estimates and Shannon index based on OTU 0.03. (E) Microbial community composition of seawater, global nonseamount surface sediment, and the SCS seamount sediment samples visualized by nonmetric multidimensional scaling. Each circle represents one microbial community. (F and G) Seamount sediment microbial and thermospore abundance, Chao 1 richness, and Shannon index changes with TOC concentration variation.

Using bacterial and archaeal universal primers, 6,391,730 reads by 16S rRNA gene amplicon sequencing were generated from extracted environmental DNA in all 10 SCS seamount 0.1- to 35-cm sediment depth samples and clustered into 16,051 operational taxonomic units (OTUs; 97% similarity level) (Data Set S1C). A range of 372 to 5,106 Chao1 OTUs and a Shannon index of 2.38 to 6.06 were observed (Fig. 2A; Data Set S1A) in all the seamount systems. The average estimated Chao1 richness and Shannon diversity index of the SCS seamount microbial communities were higher than SCS basin nonseamount sediment samples (Fig. 2A) and global nonseamount surface sediments (Fig. 2B and C; details of collected nonseamount samples in Data Set S2A).

It is generally thought that the vast abyssal plain can be considered a relatively uniform and stable environment (39) and that ocean currents above the abyssal plain are relatively weak (12, 40, 41). However, several observational and modeling research projects have determined that seamounts act as barriers, interrupting ocean currents and generating hydrology dynamic changes, such as acceleration and recirculation of current, Taylor column, ocean eddy, internal waves, increased vertical mixing, and so on, which is called the “seamount effect” (3, 42–45). The seamount effect can trap and enhance the particulate matter, organic and inorganic substrates, and plankton concentrations around seamount areas (45–47). In this study, the total organic carbon (TOC) concentration ranged from 0.52% to 2.54% in seamount sediments, much higher than the mean 0 to 1% TOC concentration identified in the 0.1- to 35-cm-deep nonseamount sediments across various locations in the SCS basin (38, 45, 48) (Fig. 2A). A clear seamount effect responsible for the accumulation of organic matter and resulting in higher community diversity and biomass was observed in the area of the 10 SCS seamounts.

Samadi et al. speculated that seamounts can be sites of enhanced productivity, biomass, and diversity in ocean biological oases, and this is called the “seamount oasis hypothesis” (2, 49). Morato et al. (5) also found seamounts are hot spots of pelagic biodiversity in the open ocean. Seamounts are places where a high trophic input allows an abundance of species and high population density because interactions between prominent topographic features and water masses increase turbulence and mixing and enhance local biomass production by moving up nutrients (50). Compared to the nonseamount area, these nutrients fuel an explosion of planktonic plant and animal growth and can attract whales, sharks, tuna, and swordfish to a booming feast on the surface.

To date, this hypothesis has only been tested for larger invertebrates, using species richness as a proxy for abundance. In this study, like deep sea vents and seep habitats, seamounts create an “oasis” of microorganisms in an oceanic desert, which could support populations of microorganisms with high abundance and diversity.

(ii) Microbial community compositions and similarity between seamount and nonseamount seafloor ecosystems. In the study, 88 microbial phyla (Data Set 1D) and 3,312 species (Data Set S1F) were identified in the seamount samples, and 81 phyla (Data Set S2C) and 3,719 species (Data Set S2E) were identified in nonseamount samples. At the phylum level, the microbial community compositions showed that seamount and nonseamount ecosystems investigated here shared a high proportion of taxa (Fig. 3A; Fig. S1). The 10 most sequence-abundant microbial phyla of the investigated seamount and nonseamount ecosystems accounted for 86.8% of all sequences (Fig. 3A; Data Set S2B). Seven of those 10 phyla were cosmopolitan at all sites, namely, the major archaeal phylum Thaumarchaeota (8.2% of all sequences) and major bacteria phyla Proteobacteria (39% of all sequences), Chloroflexi (11%), Actinobacteriota (6.3%), Planctomycetota (5.5%), Bacteroidota (5.4%), and Acidobacteriota (3%), which were found at all seamount and nonseamount sites, matching the classical abundance-range relationship (51).

FIG 3.

Taxonomic composition of seamount and nonseamount microbial communities at the phylum (A) and order levels (B) using bacteria and archaeal universal primers. The microbial sequence was transformed to Z-scores before calculating abundance. The OTUs with less than 0.1% total relative sequence abundance of each collected sample were not displayed.

At the order level, Nitrosopumilales, Alteromonadales, Anaerolineales, Rhodobacterales, Flavobacteriales, Steroidobacterales, and “Candidatus Actinomarinales” were dominant in the SCS seamounts (Data Set S1E), but they were also important in global nonseamount sediments (Data Set S2D). However, substantial differences in microbial community structure between the seamount and nonseamount seafloor ecosystems were found; for example, Oceanospirillales (6.67%), Corynebacteriales (7.96%), Rhodobacterales (4.59%), S085 (4.12%), and Kordiimonadales (1.09%) were dominant in seamount ecosystems, while Burkholderiales (11.63%), Flavobacteriales (8.11%), Bacteroidales (2.77%), Vibrionales (0.98%), and Thermoanaerobaculales (0.76%) were dominant in nonseamount ecosystems (Fig. 3B; Fig. S2). This demonstrates that seamount and nonseamount habitats exhibit different microbial community signatures at the broad taxonomic resolution level.

At a finer taxonomic resolution, no single OTU was found in all collected samples (Data Set S4). Community similarity based on shared OTU between seamount and nonseamount samples was low (analyses of similarities [ANOSIM] R = 0.2843, P = 0.001; permutational multivariate analysis of variance [PERMANOVA] _pseudo F = 9.254, P = 0.001; 999 permutations). Nonmetric multidimensional scaling (NMDS) analysis based on OTUs showed that the SCS nonseamount abyssal sediment microbial communities were similar to those of the same habitats in the global abyssal plain seafloor, but the SCS seamount microbial communities were clearly different from those in the global abyssal plain seafloor ecosystems (Fig. 2D; OTU details in Data Set S4). This result did not change when excluding the “rare biosphere” species (defined here as those OTUs with less than 0.1% total relative sequence abundance) of each collected sample. These results suggesting the SCS seamount sedimentary microbes constitute a unique microbiome.

Furthermore, a functional annotation of prokaryotic taxa (FAPROTAX) (52) analysis to generate putative functional profiles of the microbiotas inhabiting seamount and nonseamount environment based on their community composition. The database was initially developed to predict the function of marine species using standard microbiological references (53). Like other tools assigning ecological functions based on 16S data, FAPROTAX has some limitations related to the number of reference strains; therefore, the functional profiling results are indicative rather than definitive in this study. Microbial community metabolic predictions guided by the FAPROTAX database showed a difference between seamount and nonseamount ecosystems (Data Set S3; Fig. S3). For example, hydrocarbon degradation, aromatic compound degradation, ureolysis, aliphatic nonmethane hydrocarbon degradation, aromatic hydrocarbon degradation, ligninolysis, dark oxidation of sulfur compounds, dark sulfur oxidation, methylotrophy, methanol oxidation, and dark hydrogen oxidation were dominant in seamount ecosystems, while chemoheterotrophy, fermentation, nitrogen respiration, and nitrate respiration were dominant in nonseamount ecosystems.

Previously, some research showed seamount produces profound effects on the surrounding environment (54, 55), e.g., Ma et al. (45) study observed nutrients obvious uplifts around the seamounts in the Tropical Western Pacific Ocean, consistent with the uplifts of isotherms and isohalines, indicating the existence of a bottom-up process of nutrients (45). The current-seamount interaction was the determining influencing factor on nutrient distribution, causing hydrology dynamic changes such as uplifts and Taylor columns. The nutrient structures in the ocean control the microbial community, and the N/P and Si/N ratios are the most important references to evaluate nutrient levels and limitations. Sonnekus et al. found that the proliferation of diatoms preferentially depleted SiO₃-Si, causing the silicon limitation in the Coral seamount in the southern Indian Ocean (56). Comeau et al. also found a similar silicon limitation caused by the massive diatoms in the Cobb seamount (57).

This evidence reflects differences in environmental pressures and lifestyles between seamount and nonseamount ecosystems. According to the meta-community concept (58), this observed overlap and differences in taxa and function between seamounts and nonseamount ocean areas may be explained by differences in biogeochemical conditions between seamount and nonseamount systems. Hence, seamounts may select for highly adapted colonists from a global pool of microbes, which disperse across the seamounts in currents or via mobile sediment-feeding fauna. As a result, seamounts are unique microbiomes that do not occur in other marine environments.

Seamounts are island-like ecosystems for marine microbiomes in the deep sea.

(i) Seamount biogeographic patterns. Community similarity (based on the proportion of shared OTU of 0.03) between two seamount surface (0 cm depth) sediment samples decreased significantly with increasing separation distance (horizontal and vertical distance and community dissimilarity details are given in Data Set S1B), showing a distance-decay relationship in the SCS (Fig. 4A and B). In the study, the environmental factors and dispersal limitations of ocean currents in the SCS should be considered to shed more light on the mechanisms shaping the observed distance-decay pattern.

FIG 4.

Comparison of the seamount surface sediment microbiome. (A to D) Distance-decay biogeographic patterns of in situ microorganisms (A and B) and thermospore communities (C and D) at horizontal distances and vertical distances. (E and F) Null model investigating the seamount surface sediment microbial community assembly process across horizontal distance (E) and vertical distance (F). (G and H) Venn diagrams showing the numbers of endemism and cosmopolitan OTUs (gray cycle) among 10 different seamounts of in situ microorganisms (G) and thermospores (H).

(ii) Environmental selection governing seamount surface community structure. In Data Set S1A, the variation in microbial richness and community diversity among the 10 seamount surface (0 cm depth) sediments was large, e.g., 9.1 × 10⁷ to 4.5 × 10⁹ cells g⁻¹, 1,502 to 3,851 OTUs, and a Shannon index of 3.86 to 5.98. Statistical analysis revealed significant correlations between microbial community, sedimentary organic carbon concentration, and water depth (Fig. 2F). The data show that the SCS seamount surface sediment microbial abundance, richness, and diversity generally decreased with increasing water depth and decreasing TOC concentration (Fig. 2A and F; Data Set S1A). The lowest microbial biomass (9.1 × 10⁷ cells g⁻¹), richness (Chao1 = 2,390), and diversity (Shannon = 3.86) were associated with the lowest TOC concentration (0.52%) at the deepest water depth on the Huangyan Seamount (water depth = 3,827 m), and the highest microbial biomass (4.5 × 10⁹ cells g⁻¹), richness (Chao1 = 4,210), and diversity (Shannon = 5.98) were associated with the highest TOC concentration (2.54%) at the shallowest water depth on the Puyuan Seamount (water depth = 1,805 m) (Data Set S1A). Further, the community metabolic predictions showed that the abundance of the heterotrophic community dominates over the autotrophic community in the seamount surface sediment (Data Set S3). The dominance of heterotrophs is presumably related to the trapping effect of particulate organic carbon (POC) by seamounts. These data show that microbial populations are correlated with the water depth and POC flux of the sampled seamount.

All of the sampled seamounts lie in water depths >200 m where the light intensity is too low to sustain photosynthetic production. It is well-known that a significant fraction of the organic carbon produced in the surface sunlit ocean through photosynthesis is exported into the deep sea as sinking particles, fueling the generally starved bathypelagic microbial populations (59, 60). The amount of sinking POC exported from the surface ocean is impacted by physical and biological processes (61), e.g., the physical disaggregation of particles due to turbulence, zooplankton fragmentation, and microbial degradation, leading to the decreased POC flux with depth. Increasing water depth is a general indicator of decreasing POC flux as the main source of energy for deep ocean microorganisms (12, 62). According to this theory, we suggested that the productivity of seamount ecosystems below 1,800 m in the SCS is sustained mainly by sinking POC, and since the seamounts span a broad water depth range (1,805 to 3,827 m), this results in naturally occurring heterogeneous habitat features. The wide span of microbial community diversity observed across all seamounts could be explained by differences in local heterogeneity, which generated the number of ecological niches that are available to microbial colonists. This could result in a substantial degree of endemism (Fig. 4G); for example, 55 to 878 endemic OTUs were observed in the different SCS seamount surface sediments. According to the niche theory, each seamount local surface microbial population can be viewed as a subsample of the SCS regional pool of species that passed through a set of environmental filters. Environmental selection holds true for different seamount microbial community assemblies and biogeographic patterns.

(iii) Dispersal processes governing seamount community structure. “The environment selects” is not the only factor shaping microbial communities and biogeography across habitats (63). Recent characterizations of microbial populations provide evidence that in addition to environmental selection, stochastic processes (e.g., historical processes and dispersal limitation) can be important drivers of microbial assembly and biogeography (24, 64, 65).

In this study, thermospores are passively deposited to the seamount sediment as tracers to study the effect of dispersal by ocean currents. All seamount samples were pasteurized at 80°C for 1 h to eliminate viable vegetative cells and were incubated at 50°C immediately afterward to promote the germination and growth of thermophiles. Dormant thermophilic microorganisms were resuscitated in all 50°C high temperature-incubated 0.1- to 35-cm sediment depth samples (Fig. S4). 16S rRNA gene amplicon sequencing showing lineages of the phylum Firmicutes were the dominant microorganisms in the high-temperature incubation, with relative abundances ranging from 81.78 to 98.08% (Data Set S1G). The most commonly represented families were Bacillaceae (7.69 to 98.24%, n = 10/10), Sulfobacillaceae (0.003 to 83.93%, n = 7/10), and Desulfitobacteriaceae (0.003 to 27.47%, n = 8/10) (Data Set S1H). A total of 147 thermospores species were identified in all high-temperature incubation samples (details in Data Set S1I and Fig. S5). Combining the most probable number (MPN) with 16S rRNA sequencing, an abundance of 1.0 × 10⁵ to 1.2 × 10⁶ thermospores g⁻¹ was estimated for the SCS seamount 0.1- to 35-cm sediment depth samples (Data Set S1A). However, there were no correlations between thermospore population, sedimentary organic carbon concentration, and water depth (Data Set S1A), suggesting that the distribution of thermospores was not the consequence of environmental selection.

In Data Set S1A, the data show a large variation in thermospore richness, community structure, and diversity among seamount surface (0 cm depth) sediment samples. The observed thermospore OTUs ranged from 28 to 65 (n = 10), and the thermospore Shannon index ranged from 0.24 to 2.44 (n = 10) (Data Set S1A). Some endemic thermospore OTUs were also observed in the different SCS seamount surface sediment samples (Fig. 4H). The large numbers of thermospores that were nonrandomly distributed in the SCS seamount surface sediments indicated the dispersal process in the SCS was not homogenized. The thermospore community did show a distance-decay relationship (Fig. 4C and D; thermospore community dissimilarity details in Data Set S1J), revealing that multiple dispersal vectors (such as water depth and water mass circulation or transport) could be responsible for the considerable differences in richness and diversity of the thermospores observed in the SCS seamount surface sediment samples.

Thus, the Baas-Becking (66) dictum “everything is everywhere” does not completely apply here, and the dispersal limitation holds true even for members of the microbial seed bank (67) with enhanced survival capacities, in other words, perhaps the most likely candidates to not be dispersal limited in the ocean. This seawater-driven dispersal limitation of microorganisms likely constitutes a fundamental mechanism generating the structure and assembly of seamount microbial populations across the SCS basin. This study suggests that a deterministic environmental selection process and a stochastic process such as dispersal limitation (68) were expressed simultaneously in the seamount microbial community assembly, creating the distance-decay biogeographical pattern.

(iv) Environmental selection versus dispersal processes governing initial community assembly in seamount surface sediment. The null model (69) was used to investigate the relative importance of ecological processes in the community assembly of seamount surface (0 cm depth) sediment. The null model-based analysis suggested that stochastic processes such as dispersal limitation (absolute value of the β-nearest taxon index |βNTI| < 2, Bray-Curtis dissimilarity-based Raup-Crick metric [RC_Bray] > 0.95) (Fig. 4E and F) dominated the initial community assembly in the seamount surface sediment (see details in “Null model” in Materials and Methods). The results are consistent with neutral theory (16): community dissimilarity is predicted to increase along spatial (distance) gradients due to dispersal limitation (18).

Physicochemical conditions (e.g., temperature and pH) are not extreme in the SCS seamount area. On the contrary, the seamount effect provides a nutrient-rich (e.g., high concentration of organic matter from marine detritus [Fig. 2A]) environment with diverse resources that reduce competitive pressures. For example, seamount microbial communities are highly diverse, and microbial function showed that seamounts are dominated by microorganisms capable of using many different resources (details in Data Set S3). This is evidence of weak microbial population competition, and the weak microbial population competition can increase stochasticity and obscure relationships between microbial community composition and environmental variables (70).

In addition, these high levels of diversity and stochasticity may result from the physical structure of the seamounts. Highly complex hydrological dynamics in seamount areas cause extensive turbulence that increases sediment suspension on the top of seamounts (2). Galand et al. found that disturbance had a positive effect on sediment microbial communities by increasing their production and both their community and phylogenetic diversity, and they suggested that disturbance stimulated the overall community growth and promoted the development of many phylogenetically different microorganisms. These diverse communities occupy a wide range of niches and consume a variety of substrates produced during the sequential anaerobic degradation of organic matter (71). This also suggests that disturbance stimulated overall community growth and promoted the development of many phylogenetically different microorganisms in the SCS seamount areas. At the same time, seamounts are exposed to strong hydrodynamic forces that may promote a dynamic environment with many opportunities for successful immigration (dispersal followed by establishment) (72). High levels of diversity at seamounts may therefore be maintained in regions characterized by high resource supply, a potentially highly dynamic environment, and relatively benign abiotic conditions, consistent with the Chase et al. (19) assertion that high productivity enhances biodiversity through elevated levels of stochasticity. Overall, seamounts enhance seafloor biodiversity and represent island-like ecosystems in the deep sea, resulting in the similarity in species composition between two communities declining with increasing geographic distance and exhibiting the distance-decay biogeographical pattern.

Ecological succession within seamount sediment vertical community.

(i) Changes in the environment progressively increase the importance of deterministic selection. The study of ecological succession is the core of ecology. Within each seamount, most of observed |βNTI| between surface (0 cm depth) and subsurface (~15 to 35 cm depth) sediment samples was >2, suggesting that deterministic environmental selection is the predominant force driving vertical community succession in the seamount. To further elaborate on the ecological selection within seamount subsurface sediments, the variance in microbial abundance and composition between the surface sediment and the bottom of push core subsurface sediment samples was investigated.

The data showed the microbial population of the surface (0 cm depth) sediment is different from the subsurface (~15 to 35 cm depth) sediment samples. The microorganism abundances at the bottom of the push corer subsurface (~15 to 35 cm depth) sediments ranged from 1.0 × 10⁷ cells g⁻¹ (Meiwending Seamount, 3,762 m water depth, 30 cm sediment depth) to 2.3 × 10⁸ cells g⁻¹ (Puyuan Seamount, 1,805 m water depth, 15 cm sediment depth), which were 1 to 2 orders of magnitude lower than those at sites located closer to the surface sediments (Data Set S1A). Using the 16S rRNA gene amplicon, a range of 983 to 2,534 OTUs (Huyangyan Seamount, 3,827 m water depth, 20 cm sediment depth, 983 OTUs; Puyuan Seamount, 1,805 m water depth, 15 cm sediment depth, 2,534 OTUs) and a Shannon index of 2.70 to 4.61 (Huyangyan Seamount, 3,827 m water depth, 20 cm sediment depth, Shannon index = 983; Zhenbei Seamount, 1,930 m water depth, 19 cm sediment depth, Shannon index = 4.61) in the bottom of push corer subsurface (~15 to 35 cm depth) sediment samples of all 10 seamounts were obtained (Fig. 5A; Data Set S1A). The organic matter decreased dramatically with sediment depth (Data Set S1A) from 1.06 to 2.54% in surface sediment to 0.25 to 1.06% (Meiwending Seamount, 3,762 m water depth, 30 cm sediment depth, TOC = 0.25%; Daimao Seamount, 2,095 m water depth, 20 cm sediment depth, TOC = 1.06%) in the bottom of push corer subsurface (~15 to 35 cm depth) sediment. Statistical analysis also showed significant correlations between sedimentary TOC concentration and microbial population. The SCS seamount sediment microbial abundance, richness, and diversity generally decreased with increasing sediment depth and decreasing TOC concentrations across the different SCS seamounts (Data Set S1A).

FIG 5.

Vertical community variation in microbial community composition (A) and predicted ecological functions (B) from the sediment-water interface sediment to the bottom of push core subsurface sediment samples. The predicted ecological functions based on 16S rRNA genes of microbial community composition.

In the surface (0 cm depth) sediment samples, the top microbial families were Nitrosopumilales (~2.84 to 18.88%), Alteromonadales (~3.79 to 26.17%), Oceanospirillales (~0.29 to 19.11%), Flavobacteriales (~0.49 to 12.24%), Rhodobacterales (~1.03 to 11.64%), Pseudomonadales (~0.02 to 10.30%), Steroidobacterales (~1.00 to 10.06%), Nitrosococcales (~0.31 to 18.04%), “Ca. Actinomarinales” (~0.51 to 7.31%), and Pirellulales (~0.69 to 4.54%). At the bottom of the push core subsurface (~15 to 35 cm depth) sediment samples, the top microbial phyla were Oceanospirillales (~1.08 to 44.09%), Alteromonadales (~0.21 to 30.82%), Pseudomonadales (~0.09 to 19.09%), Rhodobacterales (~0.27 to 16.40%), Flavobacteriales (~0.09 to 12.46%), Anaerolineales (~0.08 to 9.25%), no-rank Thermoplasmata (~0.02 to 11.06%); Methylococcales (~0.002 to 13.73%), Pasteurellales (~0.01 to 20.57%), Nitrosopumilales (~0.04 to 6.79%), and S085 (~0.22 to 9.98%) (Data Set S1E).

NMDS analysis showed that the in situ microbial community structure at the OTU level in surface sediments was significantly different from that in subsurface (~15 to 35 cm depth) sediments on all 10 studied seamounts (PERMANOVA R² = 0.2354, P = 0.001, 999 permutations; ANOSIM R = 0.6420, P = 0.001) (Fig. S6). Obvious shifts in in situ microbial community compositions were observed between the surface sediment and the bottom of push core subsurface sediment samples, e.g., a large number of endemic taxa were detected in surface (1,214 to 3,147 OTUs) or subsurface (315 to 2,548 OTUs) sediment samples (Fig. 5A; Fig. S7 and S8). This demonstrates that seamount surface and subsurface sediment habitats exhibit distinct microbial community signatures at broad taxonomic resolution levels, probably also reflecting differences in environmental pressures and lifestyles between surface and subsurface sediment habitats.

Microbial metabolic predictions guided by FAPROTAX (Fig. 5B; Data Set S3) showed that chemoheterotrophy, hydrocarbon degradation, dark oxidation of sulfur compounds, and fermentation were dominant throughout all sampled depths of the sediment column. However, the other dominant energy acquisition strategies transition from aerobic ammonia oxidization and nitrification in surface sediment samples to anaerobic dark hydrogen oxidation, methanotrophy, dark thiosulfate oxidation, and methanol oxidation in seamount subsurface sediment (Fig. 5B). Oceanic subsurface communities are cut off from fresh detrital organic matter deposited on the seafloor (73). Due to the depletion of respiratory electron acceptors (microbial oxidation of buried organic matter) and the reduced reactivity of substrates with depth, the energy available for microorganism maintenance and growth decreases rapidly with sediment depth and age (74, 75). The different metabolic strategies of the sediment community indicate that microorganisms respond to sediment depth and biogeochemistry environmental changes in seamount subsurface sediments associated with the emergence of new spatially structured ecological niches. These results show a resource-dependent shift in the relative influence of deterministic and stochastic processes along the seamount sediment chronosequence.

(ii) Selective growth of rare species in surface sediments in shaping the subsurface sediment community. The thermospore abundances at the bottom of the push corer subsurface (15 to 35 cm depth) sediments ranged from 6.0 × 10⁴ cells g⁻¹ to 8.2 × 10⁵ cells g⁻¹. These abundances are 1 to 2 orders of magnitude lower than those of surface (0 cm depth) sediments (Data Set S1A). Using the 16S rRNA gene amplicon, 21 to 49 thermospore OTUs in all 10 seamount subsurface sediment samples were obtained (Data Set S1A). Of the Firmicutes thermospore species in the SCS seamount subsurface sediment, the most represented families were Bacillaceae (6.53 to 99.94%, n = 10/10), Sulfobacillaceae (0.003 to 93.44%, n = 9/10), and Desulfitobacteriaceae (0.003 to 35.32%, n = 5/10) (Data Set S1H). Compared with the thermospore community in the surface sediment, the results showed that subsurface thermospore communities are populated by a subset of the surface sediment thermospore populations (Fig. 5A; Data Set S1H and I).

In the ocean, the rejuvenation of thermospore populations through germination, repair, and sporulation cycles is not expected at cold subsurface sediment temperatures (30). In contrast, the death and decay dependence of thermospores (30) can cause the abundance and diversity of the endospores to decrease with burial depth. The thermospore data suggested that as depositing particulate matter accumulates on the seabed, a subset of microorganisms deposited in the surface sediments are gradually buried over a long time, becoming subsurface sediments.

The reduction in diversity and richness of in situ microbes with depth is indicative of environmental selection and suggests that certain microbial populations disappear as they are buried deeper within the seamount subsurface sediments (Fig. 5A). These results mirror the energy-driven drop in the size of microbial populations with sediment depth, suggesting that the energy limitations remove microbial-specific populations along with reducing the size of the community (76–78).

However, some subsurface-specific endemic taxa within the isolated subsurface biosphere were found in the seamount area (Fig. 5A). Previous research inferred that microbial evolution has a hand in constantly shaping a uniquely adapted subsurface community or active dispersal from other layer sediments plays a role. Cells that populate the surface sediments become buried over time, gradually separating them from the surficial environment and exposing them to successive changes in environmental conditions (79). Due to energy depletion and limitation, the microbial metabolic rates within the subsurface sediments drop by 2 to 3 orders of magnitude within the surface sediments (80, 81). Petro et al. suggested that active dispersal is unlikely to occur within the energy limitation zone of subsurface sediments (79). Furthermore, prior reviews on energy limitations in subsurface sediments have addressed the constraints on diversification faced by members of these slow-growing microbial communities (78, 82, 83). Starnawski et al. studied genome evolution in the subsurface and showed that the nucleotide sequence diversity of lineages was low and relatively stable across the length of the sediment column, suggesting that evolutionary changes are limited across the entirety of the genome (73). Therefore, the changes in microbial community composition observed in the subsurface sediments do not reflect new speciation events. Gibbons et al. (84) and Legendre (85) suggested that richness and community differences may reflect changes in the relative abundance of microorganisms that are always present, typified by the persistence of certain surface communities in the subsurface sediments.

Importantly, within the 16S rRNA sequencing data set, a small number of OTUs (211 to 1,056 OTUs) were present throughout all sampled depths of the sediment column (Fig. 5A; Fig. S7). The persisting OTUs had relatively low abundance in the surface sediments but collectively made up a large proportion of the microbial population in the subsurface sediments. It has been suggested that rare species in surface sediments could become abundant in subsurface sediments. We inferred that the microorganisms favorably selected by new conditions already existed at low abundances as members of the seed bank before the environmental change occurred (e.g., increasing sediment depth, decreasing oxygen content, and so on), and the genomes of the favorably selected microbes already encoded the necessary adaptations before the change. The current sequencing coverages achieved in community profiling often fail to fully describe microbial communities, thus leading to insufficient detection and, therefore, an underestimation of rare microbial taxa (86).

The transition in microbial populations from seamount surface sediments to subsurface sediments is marked by a filtering of unadapted populations from the surface sediments. As a result, only a subset of the microbial population survives and grows to dominate the subsurface (15 to 35 cm depth) microbial community. Due to the close coupling between energy depletion (TOC depletion) and seamount sediment depth, it is likely that the ability to survive within energy-limited habitats would confer a selective advantage in the subsurface sediments and would be the primary force driving the relationships that are seen in this study (73, 79). Therefore, as vertical succession proceeds, the relative importance of deterministic selection increases, and that of stochasticity declines.

Conclusions.

Seamounts are isolated habitats, resource-rich and highly heterogenetic. They are a physical habitat quite unlike the vast continuum of the abyssal plain. Seamount microorganism communities reflect island biodiversity patterns, enhancing seafloor biodiversity and providing oases in the deep sea seafloor environment. Seamount microbial community assemblages are determined by many factors, including the seamount’s location, water depth, and organic content of the sediments on seamounts, as well as the features of ocean current (dispersal limitation) in the SCS, resulting in a distance-decay biogeography pattern. Further, changes in seamount subsurface sedimentary environments caused shifts in the type and strength of ecological processes shaping the microbial community. The initial community establishment of seamounts will be primarily driven by stochastic processes such as dispersal limitation, where communities are characterized by the colonization of different populations and lack of orderly community structure. Stochastic community assembly causes high biodiversity in these productive seamount environments. Following initial microbial community establishment, deterministic selection becomes progressively more important, driving vertical community assembly patterns in the subsurface sediments as organisms affect their environment (e.g., through energy depletion).

MATERIALS AND METHODS

Samples.

During the R/V Tan Kah Kee 1083 expedition (April to May 2018) in the SCS, the ROPOS ROV (Canadian Scientific Submersible Facility) was used to collect sediment samples using a manipulated push corer on the summit of 10 seamounts distributed across the deep SCS basin (Fig. 1; details in Data Set S1A in the supplemental material). Seamounts studied include Daimao (DM; 17°38.1296′N, 117°6.4345′E, water depth = 2,095 m, sediment depth = 20 cm), Huangyan (HY; 15°17.2723′N, 117°32.9357′E, water depth = 3,827 m, sediment depth = 20 cm), Jiaolong (JL; 17°32.181′N, 117°44.3618′E, water depth = 3,721 m, sediment depth = 25 cm), Longmen (LM; 12°31.5517′N, 113°41.534′E, water depth = 3,608 m, sediment depth = 35 cm), Longxi (LX; 13°7.6202′N, 114°28.7122′E, water depth = 3,146 m, sediment depth = 29 cm), Meiwending (MWD; 16°53.1704′N, 118°0.4547′E, water depth = 3,762 m, sediment depth = 30 cm), Shixingbei (SXB; 16°34.9017′N, 116°24.0945′E, water depth = 3,405 m, sediment depth = 29 cm), Puyuan (PY; 21°8.0308′N, 119°12.4658′E, water depth = 1,805 m, sediment depth = 15 cm), Zhenbei (ZB; 14°59.8583′N, 116°30.3167′E, water depth = 1,930 m, sediment depth = 19 cm), and Zhongnan (ZN; 13°53.1921′N, 115°19.8736′E, water depth = 3,543 m, sediment depth = 28 cm). Sediment was sampled at a depth interval of 5 cm within the push cores. Later, sediment samples were stored at 4°C and also frozen at −80°C immediately after sampling. Geographic distances between the seamounts were calculated based on the coordinates of the sampling sites using ArcGIS (ESRI, Redlands, CA, USA).

Analysis of TOC.

Five-gram frozen sediment samples were subsampled. The subsamples were freeze-dried, and the samples were homogenized with an agate mortar and pestle, then transferred into preweighed aluminum foil capsules, treated with 10% hydrochloric acid to remove carbonates, and dried overnight at 40 to 50°C. TOC values were quantified using a Vario El III elemental analyzer interfaced with a Thermo Finnigan Delta Plus XL stable isotope ratio mass spectrometer. Analyses were normalized relative to an acetanilide standard. The concentrations of TOC were expressed as percent dry precipitates. Each sample was performed in triplicate, and the values were expressed as the mean (n = 3).

Cell counts.

The CARD-FISH protocol was based on a previous study (87). The probe-labeling peroxidases were EUB338 (bacteria, 5′-GCW GCC WCC CGT AGG WGT-3′) and Arch915 (archaea, 5′-GTG CTC CCC CGC CAA TTC CT-3′) (88). Probe NON338 (5′-ACT CCT ACG GGA GGC AGC-3′) was used as a control (89). Five-gram frozen sediment samples were subsampled. The frozen sediment samples were unthawed and then were fixed with 4% paraformaldehyde for 24 h at room temperature. The fixed sediments were washed three times by centrifugation (8,000 × g for 10 min) using phosphate-buffered saline (PBS) at 4°C and stored in an ethanol-PBS buffer (1:1) at −20°C for further processing. After that, 100 μL of fixed sediment was diluted with 900 μL ethanol-PBS buffer (1:1) and dispersed using ultrasound. Next, 20 μL of dispersed sediment was diluted in 20 mL of Milli-Q filtered water. The suspended sediment was filtered on polycarbonate filters, and 0.1% low-melting-point agarose was dripped onto the filters and dried at 46°C in an incubator. The microbes were permeabilized using 15 μg/mL proteinase K. Then, 3% H₂O₂ was used to inactivate the endogenous peroxidases.

For hybridization, filters were placed in a tube and mixed with 500 μL hybridization solution (10% dextran sulfate, 2% blocking reagent [Roche, Germany], 0.1% [wt/vol] sodium dodecyl sulfate, 20 mM Tris-HCl [pH 8.0], 0.9 M NaCl, and formamide) and 1 μL of probe working solution (final concentration, 0.028 μM) (90). Microorganisms were hybridized for at least 60 min on a rotor at 46°C, and then the filters were washed twice using washing solution (90) (0.01% SDS, 5 mM EDTA [pH 8.0], 20 mM Tris-HCl [pH 8.0], and 3 mM NaCl) at 48°C for 20 min. After washing, filters were mixed with 1,000 μL of amplification solution (0.0015% H₂O₂, 1×PBS [pH 7.4], 0.1% [wt/vol] blocking reagent) and 1 μL of Alexa 488-labeled tyramides (Life Technologies, Thermo Fisher, USA). The probes were incubated at 46°C in amplification solution for at least 30 min in the dark.

For second hybridizations, the first probe-labeling peroxidase was inactivated by incubating the filter sections in 0.01 M HCl for 10 min at room temperature and washing the sections with 50 mL of Milli-Q water. Next, the CARD-FISH protocol was repeated two times with the same filter sections by using different probes. The second hybridization was performed using Alexa 647-labeled tyramides (Life Technologies, Thermo Fisher, USA). Finally, all microorganisms were stained using DAPI (4′,6-diamidino-2-phenylindole) and mounted with ProLong Gold antifade reagent (Life Technologies, Carlsbad, CA, USA). Cell counting was performed in triplicate using ImageJ, and the cell abundance was expressed as the mean (n = 3).

High-throughput sequencing and data analysis.

Five grams of frozen sample was aseptically subsampled for DNA extraction. These samples were used for DNA extraction taken from the inner part of the core to reduce contamination from the sampling process. A negative control (5 mL of water) was included with each batch of subsamples and subjected to DNA extraction in the lab. DNA was extracted from each subsample using the E.Z.N.A. soil DNA kit (Omega Bio-tek, Norcross, GA, USA) according to the manufacturer’s instructions. The final DNA concentration was determined by a NanoDrop 2000 (Thermo Scientific, Waltham, USA), and DNA quality was checked by 1% agarose gel electrophoresis. The sequencing steps were conducted by Majorbio Bio-Pharm Technology Co., Ltd. (Shanghai, China). The bacterial and archaeal universal primers were 515F and 806R (91), targeting the 16S rRNA gene V4 region. The PCR protocol was performed according to previously described methods (92). Triplicate amplification products were extracted from a 2% agarose gel and further purified using an AxyPrep DNA gel extraction kit (Axygen Biosciences, USA) and quantified using QuantiFluor-ST (Promega, USA). Purified amplicons were pooled in equimolar amounts and paired-end sequenced (2 × 300 bp) on an Illumina MiSeq platform (Illumina, San Diego, CA, USA).

Raw fastq files were demultiplexed, quality filtered by Trimmomatic, and merged by FLASH. OTUs were clustered with a 97% similarity cutoff using UPARSE. Chimeric sequences were identified and removed using UCHIME. To remove the effects of sampling intensity (number of reads) on downhole or intersite comparisons of richness estimates, we randomly subsampled the 10,000 reads in each sample and calculated Bray-Curtis (read count) dissimilarity matrices using QIIME. Taxonomic assignment was performed using the SILVA 16S rRNA database (version 138). Distance-based maximum likelihood was used for phylogenetic analysis. Bootstrap analysis was performed using 1,000 replications. Alpha-diversity metrics (observed OTU richness, Chao1 richness, Shannon and Simpson diversity indices, and equitability index) were calculated using mothur (93). The resulting beta diversity matrices were used for 2-dimensional nonmetric multidimensional scaling (NMDS) ordinations with 20 random starts.

Acquisition of metadata from public databases and diversity analyses.

A total of 86 surface sediment samples at 40 sites were collected from public references (Data Set S2A). Raw fastq files were handled as described above. Parts of primers of Illumina sequencing data were not provided in the SRA database. However, the 16S rRNA gene V4 region sequences had been found in Illumina sequencing fasta files, hence trimming the read lengths down to a specified cutoff, which contain the 16 rRNA gene V4 region. This uniformizes the length of all reads to satisfy the global aligning requirement of UPARSE. This allows for consistent alignment and downstream analysis of the sequences. Taxon abundance tables obtained from the sequence clustering were standardized using Hellinger transformation. Dissimilarities between all pairs of samples were calculated using the Bray-Curtis dissimilarity coefficient to obtain a beta diversity matrix using mothur. The resulting distance matrix was reduced to a two-dimensional (2D) space using NMDS, as described by Zinger et al. (94). Analyses of similarities (ANOSIM) were performed to test for significant differences between groups of samples using 1,000 Monte Carlo permutation tests (94).

Null model.

A null model analysis was carried out for the community assembly process classification using methods described in a previous study (Stegen et al. [21, 69]). To evaluate ecological processes, the β-mean nearest taxon distance (βMNTD), which measures the phylogenetic turnover between samples, was calculated using the R picante library. A standardized estimate of βMNTD, i.e., β-nearest taxon index (βNTI), was calculated as the number of standard deviations of the observed βMNTD from the null distribution of βMNTD. βNTI values of >2 or less than −2 indicate heterogeneous (significantly more than expected phylogenetic turnover) and homogeneous (significantly less than expected phylogenetic turnover) selection, respectively, which represent the deterministic process. βNTI values falling within the range of −2 to 2 (which do not significantly deviate from the null βMNTD distribution) indicate stochastic processes that include homogenizing dispersal (mass effect), dispersal limitation, and drift. To discern these three processes, a Bray-Curtis dissimilarity-based Raup-Crick metric (RC_Bray) was calculated with RC_Bray of >0.95, |RC_Bray| of <0.95, and RC_Bray of less than −0.95 being interpreted as dispersal limitation, drift, and homogenizing dispersal, respectively.

Prediction of functions.

In order to identify potential microbial ecological functions at the studied sites, the obtained OTUs were compared against the FAPROTAX database to predict potential metabolic functions of the microbial community. The Functional Annotation of Prokaryotic Taxa version 1.0 (FAPROTAX; http://mem.rcees.ac.cn:8080/) pipeline was used to extrapolate microbial community functions. FAPROTAX was constructed by integrating multiple culturable prokaryotic microbes with reported functions and contained more than 7,600 functional annotations for more than 4,600 species. This makes it a powerful tool to perform functional annotation based on published metabolic and ecological functions such as nutrient (e.g., C, N, P, and S) cycling, plant pathogens, and symbionts (52).

Incubation of pasteurized sediment slurries at 4°C and 50°C.

Nonhomogenized, 4°C-stored samples (approximately 10 g each) were added to three separate autoclaved 120-mL serum bottles. Triplicate slurries were prepared from surface and subsurface sediments from each of the locations to investigate thermophile germination and growth. The serum bottles were immediately sealed with sterile black butyl stoppers, and the headspace was exchanged with N₂-CO₂. Sediment aliquots were subsequently diluted in a 1:2 (wt/wt) ratio with an anoxic artificial seawater medium (95) containing 20 mM sulfate, 1 g/L peptone, 0.5 g/L yeast extract, 1 mL/L trace element solution (96), and 1 mL vitamin solution (96) under a constant flow of N₂-CO₂. To minimize competition for limited substrates between different microbial groups (e.g., sulfate-reducing and fermentative bacteria), the slurries were amended with a combination of six low-molecular-weight organic acids (acetate, butyrate, formate, lactate, propionate, and succinate), each at a final concentration of 5 mM (26).

Slurries were pasteurized at 80°C for 1 h to eliminate viable vegetative cells and were incubated at 4°C and 50°C immediately afterward to promote the germination and growth of thermophiles. Initial time zero samples were taken before pasteurization, and the slurries were incubated for 10 days, with subsampling every 2 to 3 days. Putative thermophile OTUs indicate that many thermophiles will germinate and grow at 50°C after surviving the initial pasteurization (27). Subsamples were removed using N₂-CO₂-flushed sterile syringes. The resulting subsamples were stored separately at −20°C and used for liquid chromatography and DNA extractions. For analysis of the sulfate concentration, 1 mL of medium was transferred into a vial containing 0.5 mL of ZnAc (5%), which was shaken well and stored at −20°C. In the laboratory, sulfate was measured using a Dionex ICS-1500 ion chromatograph (Thermo Fisher Scientific, Sunnyvale, CA, USA) (97).

Estimating thermophile abundance in seamount sediments.

Most probable number (MPN) enumerations were performed to estimate the abundance of thermophile endospores in SCS seamount sediment samples. Sediments were inoculated into a sterile medium. For each sediment depth, triplicate 10-fold serial dilutions were prepared in 20-mL tubes from undiluted to a 10⁻⁷ dilution. The first tube of each series was inoculated with 1 mL of pristine sediment and pasteurized at 80°C for 30 min before the subsequent dilution transfers. Each tube was homogenized by vigorous shaking after inoculation. In addition, noninoculated tubes were incubated as negative controls. After 10 days of incubation, 3-mL aliquots were withdrawn from each tube and preserved by the addition of 6 mL of 20% (wt/vol) zinc acetate and freezing. The sulfate concentration was measured using a Dionex ICS-1500 ion chromatograph and an IonPac AS23 column (eluent, 4.5 mM Na₂CO₃ and 0.8 mM NaHCO₃; flow, 1 mL/min). The optical density at 600 nm (OD₆₀₀) was measured by UV spectroscopy. Thermophile cell numbers were estimated from the three-tube table published in de Man (98) according to the changes in the sulfate concentration and OD₆₀₀. From these series, DNA was extracted (see above) for thermospore population studies and the identification of putative thermospores following Hubert et al. (27) and Müller et al. (29). Each sample was performed in triplicate, and the values were expressed as the mean (n = 3).

Data availability.

Sequencing data are stored on the National Center for Biotechnology Information (NCBI) Sequence Read Archive (BioProject accession no. PRJNA729633).