Interactions between temperature and energy supply drive microbial communities in hydrothermal sediment

The results are organized into subsections on in situ temperature profiles, geochemical gradients, and microbial community data. Geochemical data include concentration and isotopic data of dissolved electron acceptors (sulfate, dissolved inorganic carbon (DIC), δ¹³C-DIC), electron donors (methane, sulfide, SCOAs), and respiration end products (DIC, methane, sulfide), as well as solid-phase organic carbon pools (total organic carbon (TOC), δ¹³C-TOC, total nitrogen (TN), TOC:TN (C:N)). Microbial community data include bacterial and archaeal 16S rRNA gene copy numbers and bacterial and archaeal community trends. All geochemical and microbiological data are shown in Supplementary Data 1–4.

Temperature profiles

The in situ temperatures and temperature gradients differ greatly among sites and hydrothermal areas (Table 1; Fig. 1a, b, 1st column). Certain locations within the SA (Cold Site) and NSA (MUC02, GC13, MUC12) are uniformly cold (~3–5 °C) and thus serve as low-temperature control sites. The fact that Cold Site has no measurable depth-dependent temperature increase suggests that this site, despite being located within the SA, only has minimal hydrothermal fluid seepage. At two sites from the NSA (GC09, GC10), temperatures increase strongly, reaching ~60 °C at 400 cm below the seafloor, with temperature gradients becoming linear below 50 cm. Everest Mound, Orange Mat, and Cathedral Hill in the SA have the steepest temperature gradients (>165 °C m⁻¹), reaching >80 °C within 25 cm, whereas Yellow Mat from the SA only reaches ~27 °C at 45 cm. Temperature gradients are near-linear at Everest Mound, Cathedral Hill, and Yellow Mat, and in the top ~15 cm of Orange Mat. Below ~15 cm, the temperatures at Orange Mat are nearly constant.

Concentrations of methane, sulfate, sulfide, and DIC

Porewater concentration profiles of methane, sulfate and DIC are consistent with higher microbial activity and higher substrate supplies in hydrothermal seep sediments compared to cold control sites or hydrothermal non-seep sediments.

Independent of temperature, sediments without fluid seepage, i.e. the hydrothermal NSA sites (GC09, GC10) and low-temperature control sites (MUC02, MUC12, GC13, Cold Site), have similar concentration profiles of sulfate, methane, and DIC (Fig. 1a, b, 2nd column). Methane remains at background concentrations (≤0.02 mM), suggesting minimal methane production. DIC concentrations increase with depth by ~1–2 mM relative to seawater values (~2 mM). Sulfate decreases but remains near seawater values (~28 mM) throughout MUC02, MUC12, and the hydrothermal GC10, but drops more clearly toward the bottom of the hydrothermal GC09 (to 26.4 mM) and the cold GC13 (to 23.8 mM). The only minor deviation is Cold Site from the SA. At this site, sulfate and DIC concentrations change more with depth (sulfate drops to 23.6 mM; DIC increases to 6.2 mM), suggesting higher microbial activity relative to all hydrothermal and control sites within the NSA. Consistent with this interpretation sulfide (HS⁻) concentrations increase strongly with depth at Cold Site (from 2500 to 6200 µM) but not at the NSA sites, where sulfide concentrations remain much lower (0–52 µM (Supplementary Fig. 1). Furthermore, δ¹³C-DIC decreases with sediment depth at Cold Site (from −3.3‰ to −10.3‰), suggesting strong input of DIC from organic carbon mineralization (Supplementary Fig. 2). By contrast, δ¹³C-DIC remains close to seawater values (~0‰) throughout sediments of all NSA sites (−1.7‰ to −0.2‰).

Compared to all NSA sites and Cold Site, sulfate, methane, and DIC concentrations are more variable at the seep sites Yellow Mat, Cathedral Hill, Orange Mat, and Everest Mound (Fig. 1b, 2nd column). Methane concentrations at Yellow Mat, Cathedral Hill, and Orange Mat are much higher than at the non-seep sites, reaching 3.3, 5.2, and 2.8 mM, respectively (no data from Everest Mound). These high methane concentrations, which can be mainly attributed to the input of thermogenic methane from below²⁴, almost certainly underestimate in situ concentrations due to outgassing during core retrieval. Sulfate concentrations decrease more strongly with depth than at the NSA sites or Control Site, consistent with previously observed high sulfate-reducing activity^6,7 and advection of sulfate-depleted fluid from below²⁹. Nonetheless, sulfate concentrations remain in the millimolar range throughout cores from Yellow and Orange Mat. By contrast, sulfate is below detection (≤0.1 mM) at ≥4.5 cm sediment depth at Everest Mound, and in an intermittent depth interval at Cathedral Hill (~7.5–19.5 cm), below which it increases back to ~6 mM. High, i.e. millimolar, concentrations of sulfide at Orange Mat and Cathedral Hill are consistent with high rates of in situ microbial sulfate reduction and advective input of sulfide from the thermochemical reduction of sulfate in hotter, abiotic layers below (Supplementary Fig. 1). DIC concentrations reach values of >10 mM at Orange Mat, Cathedral Hill, and Yellow Mat (no data from Everest Mound). DIC concentrations fluctuate around 20 mM DIC throughout the core from Cathedral Hill, suggesting high DIC input from deeper layers. C-isotopic values of this DIC are close to those of seawater (~−3‰), suggesting an inorganic source. By contrast, surface sedimentary DIC concentrations at Yellow Mat and Orange Mat are close to seawater values but increase with depth to ~20 and ~14 mM, respectively. Lower δ¹³C-DIC values in surface sediments, which decrease further to values of ~−20‰ to −24‰ at Yellow Mat and −14‰ to −18‰ at Orange Mat within the top 10–20 cm, suggest that most of this DIC comes from the microbial or thermogenic breakdown of organic matter and/or the microbial anaerobic oxidation of methane.

Trends in dissolved SCOAs across locations

Porewater concentration profiles of SCOAs are consistent with higher input of reactive organic carbon substrates to hydrothermal seep sediments compared to cold control sites or hydrothermal non-seep sediments.

SCOA concentrations at the cold control sites and hot NSA sites are low, showing no clear depth-related trends, consistent with absence of SCOA input from below and/or biological controlled SCOA concentrations. SCOAs are dominated by acetate (cold MUC02, MUC12, and GC13: 1–3 µM; hydrothermal GCs: 3–6 µM; Cold Site: 1–7 µM), which was detected along with formate, propionate, and lactate (Fig. 2).

By contrast, SCOA concentrations at all hydrothermal seep sites except Orange Mat, increase with depth and temperature, consistent with a thermogenic source below the cored interval. At Yellow Mat, acetate concentrations are already elevated at the seafloor (32 µM) and increase to >100 µM at 20 cm depth. Cathedral Hill has a similar acetate concentration profile, but reaches even higher concentrations (250 µM). At the hottest site, Everest Mound, acetate concentrations increase from ~150 µM at the seafloor to steady concentrations of ~600 µM below 3 cm. Formate concentrations are also (locally) elevated at Yellow Mat (5-8 µM), Cathedral Hill (to 14 µM), and Everest Mound (94-265 µM), and propionate concentrations reach high values at Cathedral Hill (to 21.8 µM) and Everest Mound (to 125 µM). The only exception among the seep sites is Orange Mat, where acetate is only slightly elevated (10–20 µM), and formate and propionate remain at background concentrations. These concentrations suggest that either thermogenic SCOA input from below is low at this site, or SCOA concentrations are biologically controlled throughout the core. Unlike the other three SCOAs, lactate concentrations remain low at all seep sites, apart from one outlier at Cathedral Hill (34.5 cm: 17.3 µM), suggesting that lactate is not a major product of thermogenic organic matter breakdown.

Trends in solid-phase organic matter pools

All sites have similar δ¹³C-TOC isotopic compositions, with values ranging from −19‰ to −23‰, consistent with a predominant phytoplankton origin of sedimentary organic carbon (Supplementary Fig. 3). Yet, depth profiles of TOC and TN follow different patterns across the locations (Fig. 3). All cold control sites have similar TOC (~2–4 wt%) and TN contents (~0.3–0.6 wt%), with slight decreases in values from the seafloor downward. Compared to cold controls, GC09 and GC10 have lower TOC and TN contents (TOC: ~0.5–3 wt%; TN: ~0.0–0.3 wt%), in particular in deeper horizons with elevated temperatures. Seep sites within the SA have the widest ranges. Seep sites have higher TOC in surface sediment compared to control sites, suggesting net organic carbon assimilation and synthesis by microbial growth. TOC values are 16 wt% at the seafloor of Orange Mat and 6–7 wt% at the seafloor of the other three locations, and then decrease strongly within the top 10 cm, reaching values similar to those of cold sites or hot NSA sites below 10 cm. TN values in surface sediments of seep sites are generally higher than at control sites (~0.7–0.9 wt%), providing additional evidence of net organic matter synthesis by microbial biomass production, but then decrease steeply to values that are similar to those at hot NSA sites.

As a result of the stable TOC and TN trends, C:N does not change much with depth at the cold locations. Yet, while C:N ranges around 4.4–5.6 at Cold Site, values are considerably higher, around 8.1–10.1, at cold locations within the NSA. By comparison, the hot NSA sites and all seep sites except Orange Mat show increases in C:N with increasing temperature and depth. This increase in C:N is modest, from ~8 to 10 at Yellow Mat, and more pronounced at the hotter GC09 (to 15.9), GC10 (to 13.4), Cathedral Hill (to 14.6), and Everest Mound (to 15.7). Orange Mat has the highest C:N ratios (14.8–26.5), and unlike the other sites does not show an increase in C:N with depth.

General trends in bacterial and archaeal 16S rRNA gene copy numbers

16S rRNA gene copy numbers indicate distinct trends in bacterial and archaeal abundances that follow temperature increases with sediment depth (Fig. 1a and b, 3rd column).

At the four cold locations, bacterial and archaeal gene copy numbers are relatively stable with depth (Bacteria: 10⁸−10⁹ g⁻¹; Archaea: 10⁷−10⁸ g⁻¹). By comparison, gene copy numbers of GC09 and GC10 are in a similar range near the seafloor but decrease strongly with depth. While Archaea are quantifiable throughout both cores to ≤10³ gene copies g⁻¹ sediment, bacterial gene copy numbers are not reliably distinguishable from extraction negative controls (~1 × 10⁴ g⁻¹) at temperatures >60 °C. Furthermore, unlike the cold sites, which consistently have higher bacterial gene copy numbers, there is a shift from bacterial to archaeal dominance in gene copy numbers (GC09: at ~50 cm; GC10: at ~150 cm) at both hot NSA sites.

Compared to the hot GCs from the NSA, gene copies decrease over much shorter distances at sites with fluid seepage in the SA. This decrease in gene copy numbers appears related to the magnitude of the temperature increase with depth. At Yellow Mat, which only reaches moderately warm temperatures (27 °C), copy numbers of both domains decrease from ~10⁸ g⁻¹ at the seafloor to ~10⁶ g⁻¹ at the bottom of the core. While Orange Mat, Cathedral Hill, and Everest Mound have similar bacterial and archaeal gene copy numbers to Yellow Mat at the seafloor, these values drop off much more steeply with depth, matching the much steeper temperature increases. At Cathedral Hill and Everest Mound, Bacteria could not be reliably detected below 20 and 7.5 cm, respectively. As the only location, the detection limit of archaeal 16S gene sequences was reached at Everest Mound, at a depth of 9.5 cm.

Relationships between microbial gene abundances and temperature

We explored the relationship between 16S rRNA gene copy number and temperature further (Fig. 4a, b). While gene copy numbers of both domains generally decrease with increasing temperature, the shape of this temperature relationship differs between both domains. In bacteria the decrease in gene copy numbers in relation to temperature is nearly linear. By contrast, in Archaea gene copy numbers follow hump-shaped distributions, i.e. they remain stable or only decrease slightly up to a certain temperature threshold, beyond which their copy numbers decrease steeply. This apparent thermal threshold varies between sites, i.e. it is ~85 °C at Orange Mat, ~70 °C at Cathedral Hill, ~50 °C at the NSA sites, and ~20 °C at Everest Mound.

The differences in relationships between bacterial and archaeal gene copy numbers and temperature are reflected in Bacteria-to-Archaea gene copy ratios (Fig. 4c). Bacterial always exceed archaeal gene copies at <10 °C, while archaeal always exceed bacterial gene copies at >45 °C. Between 10 and 45 °C, domain-level gene dominance varies with location. Despite the variability, Bacteria-to-Archaea gene copy ratios follow a highly significant, exponential relationship with temperature (R² = 0.67, p < 0.001 (two-sided Spearman correlation coefficient); Supplementary Fig. 8). This relationship is supported by tests with additional qPCR primer pair combinations (Supplementary Fig. 9). We also investigated Bacteria-to-Archaea gene copy ratio relationships with temperature gradient, sediment depth, concentrations of sulfate, methane, DIC, acetate, formate, and propionate, as well as TOC, TN, C:N, and δ¹³C-TOC. None are as strong as the correlation with temperature (Supplementary Fig. 8; R²_T-gradient = 0.39; R²_depth = 0.02; R²_sulfate = 0.08; R²_methane = 0.30; R²_DIC = 0.30; R²_acetate = 0.13; R²_formate = 0.15; R²_TOC = 0.04; R²_TN = 0.03; R²_C:N = 0.43; R²_13C-TOC = 0.04).

Trends in microbial community structure within and between sites

Microbial communities show clear trends in relation to both temperature and hydrothermal seepage (Fig. 1a and b, 4th and 5th column). This is further confirmed by non-metric multidimensional scaling (NMDS) analyses (Fig. 5a, b; Supplementary Fig. 10). Clustering patterns are very similar across the phylum, class, and zero-noise operational taxonomic unit (ZOTU) level within each domain. While cold samples from all locations (and sediment depths) cluster together, bacterial and archaeal communities in sediments with elevated in situ temperatures cluster separately between seep and non-seep locations. In the following sections, we describe major phylogenetic trends across the sampling sites, focusing on phylum-level (Proteobacteria: class-level) trends presented in Fig. 1a, b. For more detailed graphs of dominant microbial taxa at the class level and below, we refer to Supplementary Fig. 4 (Bacteria) and Supplementary Fig. 5 (Archaea). These supplementary figures will be referred to in parentheses when we mention dominant groups below the phylum level in the following text. In addition, due to the high phylogenetic diversity of Crenarchaeota and Bathyarchaeota, both of which include many unclassified groups, we have extended existing classifications for both phyla based on phylogenetic trees (Fig. 6; for extended classifications see Supplementary Figs. 6 and 7). As a result, we propose 7 new Bathyarchaeota subgroups (MCG-24 through MCG-30), as well as several new, order-level subdivisions of Crenarchaeota. The new Crenarchaeota subdivisions fall into the class Thermoprotei (‘Deeply branching Thermoprotei’), the Hot Water Crenarchaeote Group I (HWCG I; ‘Subseafloor Sediment HWCG I Group’ (SSHG)), and the Terrestrial Hot Spring Crenarchaeota (THSC; ‘HWCG V’ and ‘HWCG VI’).

The four cold control sites harbor microbial groups that are “typical” of organic-rich bioturbated marine surface sediment (e.g. ref. ³⁰). Community depth profiles are also similar, despite Cold Site being located in a different part of Guaymas Basin than MUC02, MUC12, and GC13, and despite MUC12 being dominated by metal-rich, hydrothermal vent deposits rather than diatomaceous sediment, which prevails at the other sites. In the short cores (MUC02, MUC12, Cold site), dominance by Gammaproteobacteria (mainly BD7-8, Xanthomonadales, and unclassified) at the surface (~30–45% of 16S rRNA gene reads) shifts to dominance by Chloroflexi (mainly Dehalococcoidia consisting of MSBL5, vadinBA26, and unclassified members) below 10 cm (25–40%). Fractions of Bacteroidetes (diverse groups), Acidobacteria (mainly Holophagae) and Alphaproteobacteria decrease, while those of Aminicenantes and Atribacteria increase with depth. Deltaproteobacteria (~10–30%; mainly Desulfobacterales) and Planctomycetes show no depth-related trends. Archaea shift in dominance from Thaumarchaeota (Marine Group I) and Woesearchaeota in the top 10 cm to Bathyarchaeota (mainly C3 (also known as MCG-15) and MCG-17), Euryarchaeota (mainly Marine Benthic Group D within Thermoplasmata), and Lokiarchaeota (mainly Alpha Subgroup) below. Microbial communities in GC13 (shallowest sample: 26 cm) resemble those at the bottom of the short cold cores, and change little with depth apart from a decrease in Deltaproteobacteria, and a relative increase in Bathyarchaeota subgroups MCG-1 and −2 near the bottom.

The two hydrothermal GCs from the NSA differ strongly in microbial community structure from the cold sites, including GC13. Bacteria and Archaea are more clearly dominated by Chloroflexi (~65–80% of reads, same major groups of Dehalococcoidia as before plus MSB-5B2) and Bathyarchaeota (~65–98%), respectively. While the Chloroflexi community structure does not change markedly with temperature, there is a clear shift in dominant Bathyarchaeota groups. Major groups from control sites, i.e. C3, MCG-1 and −2, only dominate cold surface sediments. MCG-4 dominates temperatures of 40–50 °C between 200 and 300 cm. Further down MCG-16 becomes dominant, and remains dominant to the bottom of GC10, where MCG-3, and the newly classified MCG-27 and -28 (Supplementary Fig. 7) emerge as additional major groups. At GC09, MCG-21, -22, -23, and the newly classified MCG-28 and -29 are the dominant groups below 400 cm. While the hydrothermal GCs have similar percentages of Aminicenantes, Deltaproteobacteria, Omnitrophica, and Aerophobetes compared to GC13, Atribacteria, Euryarchaeota, and Lokiarchaeota have lower percentages in the top 200 cm and are virtually absent below. Instead contributions of poorly known Crenarchaeota increase in deep hot layers. At GC09 these include the class-level Hot Water Crenarchaeota Group IV (HWCG IV) within the class Terrestrial Hot Spring Crenarchaeota (THSC). At GC10 these include the novel, order-level SSHG cluster (Fig. 6; Supplementary Fig. 6). While HWCG IV was previously found in hydrothermal vents and hydrothermal sediments, SSHG was only recovered from an advection-influenced hydrothermal subsurface sediment⁸. In addition, an unknown branch of Asgardarchaeota and an unclassified group of Thaumarchaeota appear below 400 cm at GC09.

Reflecting the variable temperature gradients, geochemical gradients, and advective regimes, microbial communities in hot SA sediments are more diverse and heterogeneous than in hot NSA sediments. Nonetheless, there are shared patterns, that distinguish these sites from all other sites. In surface sediments, nitrifying Marine Group I Thaumarchaeota are nearly absent, and S-cycling Epsilonproteobacteria (Sulfurimonas, Sulfurovum) and Deltaproteobacteria (Desulfobacterales) are dominant Bacteria. In deeper and hotter layers, Aminicenantes are rare and Chloroflexi account for lower percentages. Instead, typical hot spring and hydrothermal vent phyla that have been linked to reductive sulfur cycling and anaerobic organic carbon degradation, i.e. Thermodesulfobacteria, Thermotogae, Acetothermia, and Crenarchaeota (mainly Thermoprotei) dominate these layers.

Despite these similarities, there are also strong differences between seep sites, in particular among bacterial communities in deeper sediment layers. While Atribacteria and Aerophobetes dominate Yellow Mat, Thermodesulfobacteria (T.desulfobacterales) dominate Orange Mat, and S-oxidizing denitrifying Epsilon- (Sulfurimonas, Sulfurovum) and sulfate-reducing Deltaproteobacteria (Desulfobacterales) dominate Everest Mound, no particular group dominates the vertically heterogeneous bacterial community at Cathedral Hill. By contrast, archaeal communities display clear, more temperature than site-related shifts in dominance, from Woese- and Lokiarchaeaota in cold surface sediments to known Bathyarchaeota (C3, MCG-1) in shallow subsurface sediments with moderate temperatures to MCG-22, -23, -28, and -30 in warmer sediments. This is followed by a further community shift toward clear dominance of Crenarchaeota at high temperatures (Thermoprotei dominated by Thermoproteales and Desulfurococcales, at Orange Mat also the novel ‘Deeply-Branching Thermoprotei’ and HWCG IV). The only exception is Orange Mat, where the top 10–20 cm are dominated by Euryarchaeota linked to anaerobic methanotrophy (ANME-1-Archaea; ANME-1-AT dominates at the surface, ANME-1-Guaymas (ANME-1-Gba) below), rather than by Woese- and Lokiarchaeota.

Source: Ecology - nature.com