1 Introduction

The carbon cycle is one of the frontier topics in global change research, as there is still a tremendous challenge to achieve carbon neutrality by 2060 (Wang et al. 2023b). In addition to the effective control of anthropogenic emissions, searching for missing carbon sinks is the key to balancing the atmospheric CO2 budget (He et al. 2022). In this regard, the coupling of water and carbon cycles is a popular issue of carbon research (Guo et al. 2023).

Lakes and reservoirs receive a considerable amount of carbon from land surface water and may serve as carbon sources or sinks in the global carbon cycle (Deemer et al. 2016; Li et al. 2022b; Liu 2023; Soued et al. 2022). On the one hand, greenhouse gas emissions, such as CO2, CH4, and N2O, occur at the water–air interface of lakes and reservoirs. The CH4 emissions from reservoirs account for 12% of the global CH4 emissions (Ran et al. 2011). On the other hand, the surface sediments of lakes and reservoirs are important sites of carbon storage (Han et al. 2022). The rate of organic carbon burial in lakes is one to four times higher than that in oceans (Gerhard et al. 2001). As of the Holocene period, lake sediments have stored 420–820 Pg of organic carbon, which is in the same order of magnitude as carbon stocks in soils and terrestrial biota. In contrast with natural lakes, artificial reservoirs have carbon burial rates that are one to two orders of magnitude higher (Sundquist 1993). In particular, refractory dissolved organic carbon (RDOC) as an important and stable fraction of carbon sink can be preserved for thousands of years in deep water (Li et al. 2022a), allowing for long-term carbon storage. Research on carbon cycling in lakes (reservoirs) can enrich our knowledge of global carbon cycling, decipher the environmental effects of carbon, and unravel the conversion of carbon sources and sinks.

In aquatic ecosystems such as lakes and reservoirs, diverse microorganisms (including bacteria, fungi, and archaea) are in close association and work together to drive the biogeochemical cycling of carbon. Microorganisms can promote the degradation of organic matter to provide nutrients for plants and other life forms or release greenhouse gases such as CO2 and CH4, thereby reducing sediment organic matter storage. Additionally, microorganisms can convert readily available carbon sources to more stable organic matter and thus contribute to the long-term storage organic matter in sediments (Liu and Xiao 2021; Kallenbach et al. 2016). While fungi are mainly involved in the decomposition and mineralization of soil organic carbon, bacteria (especially heterotrophs) are the major drivers of primary RDOC formation in aquatic ecosystems (He et al. 2022). An example comes from aerobic anoxygenic photoheterotrophic bacteria, whose variation in relative abundance is indicative of the changes in organic carbon in reservoir water (Song et al. 2017). Archaea are widespread in aquatic habitats such as fresh waters and estuaries (Lin and Xie 2021), where they use organic or inorganic electron donors and acceptors for energy metabolism, including methanogenesis, ammonia oxidation, and organic matter degradation (Liu and Xiao 2021). However, archaea usually occur at low abundances, with Bathyarchaeota, Euryarchaeota, and Thaumarchaeota being the major phyla involved in benthic carbon cycling. Among archaea, anaerobic ammonia and methane oxidizers are in close association with RDOC production (Bayer et al. 2019). Archaea play an indispensable role in driving biogeochemical carbon cycling, given their diverse metabolic pathways exemplified by carbon fixation, methanogenesis, methane oxidation, and carbon degradation (Lin and Xie 2021).

Deep-water reservoirs often experience water depth changes due to reservoir impoundment, discharge operation, and upstream inflow. Consequently, the hydrostatic pressure at the sediment–water interface (SWI) varies with different water depths in reservoirs. An in situ study showed that among various environmental factors (e.g., dissolved oxygen, pressure, pH, total nitrogen), hydrostatic pressure was the primary factor shaping the structure of sediment microbial communities in deep-water reservoirs (Chai et al. 2015). Under simulated experimental conditions, the abundance of bacteria, such as methanogens, decreased with increasing hydrostatic pressure (> 1.0 MPa) in river sediments upstream of a deep-water reservoir (Wu et al. 2021). Moreover, a higher hydrostatic pressure enhanced the expression of alkaline phosphatase-encoding genes (e.g., phoD, ppk, pqqC) and even promote phosphorus cycling at the SWI in a deep-water reservoir (Zhuo et al. 2021). Some researchers also demonstrated the influence of hydrostatic pressure on the composition, physiology, and metabolism of marine microbial communities (Wu et al. 2021). However, there has been insufficient research deciphering the mechanisms that allow hydrostatic pressure to influence microbial carbon cycling pathways at the SWI in large-scale deep-water reservoirs. This calls for research to unravel the response of microbial community structure, functional gene abundance, and metabolic pathway activity associated with carbon cycling to hydrostatic pressure changes at the SWI. The outcome will enable a mechanistic understanding of microbially mediated carbon migration and transformation in deep-water reservoirs, and lead to accurate assessment of carbon sequestration potential in reservoirs with different water depths.

Here, it was hypothesized that hydrostatic pressure could influence microbially mediated carbon cycling and drive carbon source/sink conversion at the SWI in deep-water reservoirs by altering associated microbial communities, functional genes, and metabolic pathways. To test the hypothesis, a microcosm experiment was conducted with hydrostatic pressure as the only variable, and four different pressures at the SWI were simulated using the sediment and water of a large drinking water reservoir. The changes in microbial species composition, functional gene abundance, and metabolic pathway activity associated with carbon cycling were investigated under different pressure levels. The results could offer novel insights into carbon source/sink conversion driven by microorganisms at the SWI in deep-water reservoirs and provide guidance for water level regulation during reservoir operation.

2 Materials and methods

2.1 Experimental setup and sampling

The sediment and water used in this study were collected from Jinpen Reservoir (34.057 N, 108.208 E; Fig. 1), which is located in Xi’an, Shaanxi Province, China. The average water depth of the reservoir is 75 m. Surface sediment was collected with a Peterson grab sampler, and water was taken from a depth of 50 m with a polymethylmethacrylate sampler. The sediment was mixed uniformly and preserved in a sterile insulated container at –70°C. The hydrostatic pressure simulation experiment was conducted in the laboratory using a pressurized device (Fig. 1).

Fig. 1
figure 1

Location of the study area and schematic of the experimental setup

Full size image

The experiment used four groups of reactors with different pressure levels: 0.1 MPa (atmospheric pressure, AU), 0.2 MPa (TU), 0.5 MPa (FU), and 0.7 MPa (SU). These pressure levels corresponded to the hydrostatic pressures at the SWI of 1, 20, 50, and 70 m depths in the reservoir. The different hydrostatic pressures were simulated using a pressure valve, and there were three replications for each group. The reactors were 440 mm high and 110 mm in diameter, each accommodating a sample container with the height of 200 mm, wall thickness of 5 mm, inner diameter of 95 mm, and effective volume of 1.4 L. The reactors were incubated in a 25°C atmosphere for 30 days, with a sediment-to-water ratio of 1:1 and a water temperature of 15°C. After incubation, the reactors were opened to retrieve the sample containers. The bottom lid of each container was rotated to push out the sediment core, which was then sliced into 10 mm sections. The section samples were kept in sterile ziplock bags and stored in a freezer at –70°C until used.

2.2 DNA extraction and metagenomic sequencing

DNA extraction and metagenomic shotgun sequencing were accomplished by Shanghai Majorbio Bio-pharm Technology Co., Ltd. (Shanghai, China). Briefly, total genomic DNA was extracted from sediment samples (0.2 g each; n = 12) using the Qiagen DNeasy PowerSoil Kit (Omega Bio-tek, Norcross, GA, USA) as per the manufacturer’s instructions. DNA quantity and quality were analyzed using a NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA) and agarose gel electrophoresis. Qualified DNA samples were stored at –20 °C before further evaluation. DNA sequencing libraries were prepared using a quarter-scale Nextera XT reaction (Illumina, NEB, Ipswich, MA, USA). High-throughput sequencing was performed on an Illumina HiSeq platform using the whole genome shotgun strategy. The extracted DNA for microbial metagenomes was fragmented into short fragments, and 400-bp fragments were selected to construct metagenomic shotgun sequencing libraries for paired-end sequencing.

The deduced amino acid sequences of the non-redundant gene set were aligned against the NR (https://www.ncbi.nlm.nih.gov/protein) and KEGG (http://www.genome.jp/kegg/) databases using Diamond (v0.8.35; https://www.diamondsearch.org/index.php), with a cutoff E-value of 1e-5 for BLASTP alignment. Species annotation was achieved through the taxonomic information corresponding to NR database, and gene functions were analyzed using the KEGG database. Then, species abundance was calculated as the sum of the abundance of functional genes corresponding to the species, and the abundance of functional categories was obtained as the sum of the abundance of genes corresponding to the KEGG orthology (KO) groups and pathways.

2.3 Non-targeted metabolomic analysis

Non-targeted metabolomic analysis was accomplished by Shanghai Personalbio Technology Co., Ltd. (Shanghai, China). Metabolite separation by liquid chromatography was performed on a Vanquish UHPLC System (Thermo Fisher Scientific). The samples were separated through adsorption using an ACQUITY UPLC® HSS T3 column (150 × 2.1 mm, 1.8 μm) (Waters, Milford, MA, USA) before mass spectrometry (MS). The MS analysis of metabolites was performed on Orbitrap Exploris 120 (Thermo Fisher Scientific) with an electrospray ion source. Simultaneous MS1 and MS/MS (full MS-ddMS2 mode, data-dependent MS/MS) acquisition was used. The mass analyzer was used to separate ion fragments based on their mass number to obtain MS spectra.

The raw MS data were converted to mzXML format by MSConvert in the ProteoWizard software package (v3.0.8789; Smith et al. 2006) and processed using XCMS (Navarro-Reig et al. 2015) for feature detection, retention time correction, and alignment. The metabolites were identified by comparison with high-accuracy mass (< 30 ppm) and MS/MS data that were matched with the databases HMDB (http://www.hmdb.ca/), massbank (http://www.massbank.jp/), LipidMaps (http://www.lipidmaps.org/), mzcloud (https://www.mzcloud.org/), and KEGG (http://www.genome.jp/kegg/). Quality control-robust LOESS signal correction (Gagnebin et al. 2017) was applied for data normalization to correct for any systematic bias. After normalization, only ion peaks with relative standard deviations < 30% in the quality control were kept to ensure proper metabolite identification.

2.4 Statistical analysis

Alpha-diversity metrics were calculated to assess microbial community richness and evenness. Species composition, functional contribution, and community heatmaps were analyzed using the Majorbio Cloud Platform (https://www.majorbio.com/). Genesets were constructed using CD-HIT (v4.6.1), with gene sequences clustered at ≥ 0.9 similarity. The reads per kilobase million (RPKM) value were calculated to analyze the top 20 genera in terms of total abundance. The t-test was performed in SPSS (v27.0; IBM Corp., Armonk, NY, USA) to identify between-group differences in microbial alpha-diversity and functional gene abundance.

Network analysis was carrieed out using the R program (v3.6.0; https://www.R-project.org/). Species with mean relative abundance > 0.1% in the community were used to construct microbial co-occurrence network. Pearson’s correlation coefficients between OTUs were obtained, and only robust correlations with |r|> 0.7 and P < 0.05 were retained for network visualization implemented in Gephi (v0.10.1). Two R packages, Hmisc and Psych (Luo et al. 2022), were used to calculate network topological metrics.

Principal component analysis, partial least squares discriminant analysis, and orthogonal partial least squares discriminant analysis were carried out for dimensionality reduction of the sample data using the R package Ropls (Thévenot et al. 2015). Significant metabolite molecules were identified at P < 0.05 and variable important in projection > 1. All graphs were created using Origin v2021 (OriginLab Corp., Northampton, MA, USA) and the Personalbio Genes Cloud Platform (https://www.genescloud.cn/home/).

3 Results

3.1 Species composition of carbon cycling-associated microbial communities at various pressures

At the species level, the number of microbial taxa associated with carbon cycling exhibited an overall upward trend with elevating hydrostatic pressure (Fig. 2a). Despite its slight decrease in SU group, the number of species notably increased in both TU and FU groups compared with AU group. A similar pattern was observed for the number of unique species under different pressure conditions, which was increased at higher pressures relative to AU. The highest number of unique species occurred in TU group, with the lowest number in AU group.

Fig. 2
figure 2

Venn diagrams showing the numbers of (a) microbial taxa and (b) annotated functional genes (species level) associated with carbon cycling at the sediment–water interface with various hydrostatic pressures (AU: atmospheric pressure at 0.1 Mpa, TU: 0.2 MPa, FU: 0.5 MPa, SU: 0.7 MPa)

Full size image

The bacteria, archaea, and fungi identified to different taxonomic levels (i.e., phylum, order, order, family, genus, species) had certain differences in abundance among the various pressure treatments (Fig. 3a). Bacteria accounted for > 99.5% of the total abundance under different pressure conditions, mainly comprising the phyla Proteobacteria, Actinobacteria, and Candidatus. The respective proportions of archaea and fungi were 0.2% and 0.05% in AU group, 0.4% and 0.02% in both TU and FU groups, and 0.3% and 0.02% in SU group (Fig. 3b). The elevation in pressure resulted in a higher proportion of archaea and a lower proportion of fungi.

Fig. 3
figure 3

Microbial community composition in reservoir sediments under four different levels of hydrostatic pressure (AU: atmospheric pressure at 0.1 Mpa, TU: 0.2 MPa, FU: 0.5 MPa, SU: 0.7 MPa). a Krona plots of species composition. b Proportions of archaea and fungi. c Distribution of the top 20 most abundant species associated with carbon cycling (*P < 0.05, ** P < 0.01, and *** P < 0.001)

Full size image

The proportion of archaea increased variably at higher pressures relative to AU (Fig. 3b). The most prominent increase (100%) occurred in TU and FU groups, with a 50% increase in SU group. The top three phyla with the highest proportions of archaea were Thaumarchaeota, Euryarchaeota, and Candidatus under the four different pressure levels. While the proportion of Thaumarchaeota increased with elevating pressure, the opposite was true for Euryarchaeota and Candidatus. The fungi identified in sediment samples mainly consisted of Mucoromycota, Ascomycota, Chytridiomycota, Rhodophyta, and unclassified groups. Excluding the unclassified fungi, Mucoromycota was the dominant phylum in all samples, followed by Ascomycota. Among them, Symbiodinium necroappetens was the predominant species. The species richness of fungi occurred at a higher level in AU group, with roughly consistent species composition among the other three groups (Fig. 3a, b). Specifically, the proportion of Mucoromycota was lowest in TU group and tended to increase at higher pressures in the other three groups. The lowest proportion of Ascomycota (10%) was observed in AU group, in contrast to its highest proportion in TU and SU groups (both 31%).

The top most abundant 20 microbial species associated with carbon cycling did not differ remarkably in diversity, but varied in relative abundance across sediment samples at various pressures (Fig. 3c). The top three most abundant species were Candidatus Rokubacteria bacterium, Actinobacteria bacterium, and Acidobacteria bacterium. The relative abundance of Candidatus Rokubacteria bacterium reached 10.56% (AU), 9.41% (TU), 8.85% (FU), and 8.75% (SU), demonstrating its distinct dominance and downward trend with elevating pressure. Actinobacteria bacterium occurred at the relative abundance of 5.8% (AU), 6.0% (TU), 6.3% (FU), and 6.8% (SU), which trended high with elevating pressure. There were minimal changes in the relative abundance of Acidobacteria bacterium under various pressure conditions.

3.2 Functions and KEGG modules of carbon cycling-associated microbial communities at various pressures

Among the KEGG modules for carbon cycling, the reductive citrate cycle (Arnon-Buchanan cycle, M00173) accounted for the highest proportion (13%) in all four treatment groups under various pressure levels (Fig. 4b). The proportion of the citrate cycle (TCA cycle, Krebs cycle, M00009) in AU, TU, and SU groups was 9.2%, slightly lower than 9.5% in FU group.

Fig. 4
figure 4

Microbial community function in reservoir sediments at various hydrostatic pressures (AU: atmospheric pressure at 0.1 Mpa, TU: 0.2 MPa, FU: 0.5 MPa, SU: 0.7 MPa). a Circos plots of modules and KEGG orthologs related to carbon cycling. b Functional contribution of the top 20 taxa (genus level) in terms of their relative contribution in KEGG pathways of carbon fixation, methanogenesis, and methane oxidation

Full size image

The distribution of the top 20 taxa (genus level) in terms of their relative contribution in the KEGG pathways of carbon fixation, methanogenesis, and methane oxidation was analyzed under the four different pressure levels (Fig. 4c). Among them, Candidatus Rokubacteria was the largest contributor in the seven carbon fixation pathways. In particular, Candidatus Rokubacteria showed the highest relative contribution in the hydroxypropionate-hydroxybutylate cycle relative to other pathways, and its relative contribution was greater for AU group (16.22%) than for the other three groups (TU: 13.16%, FU: 12.65%, SU: 13.57%). The relative contribution of Acidobacteria in the reductive citrate cycle and the 3-hydroxypropionate bi-cycle ranged from 7.50% to 8.72% under the four pressure levels. In the reductive acetyl–CoA pathway, the relative contribution of Betaproteobacteria was 6.26% for SU group, which was 2% lower than that for the other three groups; however, these contribution rates were all higher than those in other pathways. Deltaproteobacteria contributed the most in the hydroxypropionate-hydroxybutylate cycle (reaching > 7.5%), with the highest relative contribution observed for AU group (8.25%).

Actinomycetia was the greatest contributor in the three methanogenesis pathways. Its relative contribution was highest in methylotrophic methanogenesis compared with other pathways among all four groups, i.e., 18.35% (AU), 17.53% (TU), 16.81% (FU), and 18.15% (SU). The relative contribution of Candidatus Rokubacteria in hydrogenotrophic methanogenesis and acetoclastic methanogenesisis was as high as 7.12%–10.53% under the four pressure levels, in stark contrast to an extremely low relative contribution in methylotrophic methanogenesis (< 1%). Chloroflexi only had a high relative contribution in methylotrophic methanogenesis (AU: 7.63%, TU: 7.75%, FU: 9.30%, SU: 9.04%), with a low relative contribution in the other two pathways (< 3%).

With regard to methane oxidation, Betaproteobacteria was the largest contributor in aerobic oxidation of methane, with the relative contribution of 12.01% (AU), 12.10% (TU), 11.80% (FU), and 11.04% (SU). Its relative contribution in anaerobic oxidation of methane was substantially low, ~ 2% under the four different pressure levels. A similar trend was observed for Alphaproteobacteria, which had a higher relative contribution in the aerobic oxidation of methane than in the anaerobic pathway. Conversely, the relative contribution of Candidatus Rokubacteria and Deltaproteobacteria in anaerobic oxidation of methane was ~ 4% greater than that in the aerobic pathway.

3.3 Differential analysis of carbon cycling-associated functional genes at various pressures

At the species level, the number of annotated functional genes associated with carbon cycling increased slightly with elevating pressure, and two unique genes were present in SU group (Fig. 3b). Based on KO distribution, acetyl-CoA C-acetyltransferase (K00626), aconitate hydratase (K01681), and acetyl-CoA synthetase (K01895) were all major carbon fixation pathways under the four pressure levels (Fig. 4a).

In the annotated metabolic pathways of carbon cycling, the same genes varied in abundance under different pressure levels (Fig. 5). In the reductive pentose phosphate cycle of carbon fixation, the relative abundance of the ALDO gene was lowest in TU group and highest in FU group, with a 0.12% difference between the two groups. The GAPDH gene reached its highest relative abundance (1.95%) in TU group, accompanied by the lowest relative abundance of the rbcs gene (0.004%). In the reductive citrate cycle, the sdhA, sdhC, and ACO genes all increased with elevating pressure, which contracted with the downward trend of the pycB gene. Both fumB and korA genes showed an upward trend before decreasing with elevating pressure, and their highest relative abundances were observed in TU group, which were 0.35% and 0.08% higher than those of SU group, respectively. In the reductive acetyl–CoA pathway, the cdhD gene was detected at the highest relative abundance of 0.039% in TU group, with the lowest of 0.072% in SU group. In the 3-hydroxypropionate bi-cycle, the lowest relative abundance of the sdhA gene occurred in TU group, which was 0.34% lower than the highest in SU group. Conversely, the fumB gene had its highest relative abundance in TU group and the lowest in SU group, with a group difference of 0.08%. In the hydroxypropionate-hydroxybutylate cycle, the ACAT gene was detected at the lowest relative abundance in FU group, which was 0.43% lower than the highest in SU group. In the dicarboxylate-hydroxybutyrate cycle, the relative abundances of pycB and fumB genes both increased first and then decreased with elevating pressure. Their highest relative abundances observed in TU group were 0.07% and 0.08% higher than the lowest in SU group, respectively. A slight difference between these two genes was that pycB relative abundance increased at high pressures (i.e., FU, SU) compared with low pressures (i.e., AU, TU).

Fig. 5
figure 5

a Differential functional genes associated with carbon cycling at various hydrostatic pressures (AU: atmospheric pressure at 0.1 Mpa, TU: 0.2 MPa, FU: 0.5 MPa, SU: 0.7 MPa; * P < 0.05, ** P < 0.01, and *** P < 0.001) (b) Significant functional genes in various carbon metabolic pathways. Red ascending arrows indicate genes increasing in relative abundance at high pressures (i.e., FU, SU) compared with low pressures (i.e., AU, TU); blue descending arrows indicate genes decreasing at high pressures (i.e., FU, SU) compared with low pressures (i.e., AU, TU)

Full size image

The differential functional genes annotated in the methanogenesis process were mch, mtrH, cdhE, and cdhD. The mtrH gene initially increased and then decreased from low to high pressure, with the highest relative abundance of 0.053% in FU group and the lowest of 0% in AU group. Among cdhD, cdhE, and mch, the former two genes showed lower relative abundances at high pressures than at low pressures, and the opposite was true for the latter one gene. In the methane oxidation process, the annotated functional gene in the anaerobic pathway was mtrH, whose relative abundance exhibited an upward trend before decreasing with elevating pressure. The annotated functional gene in aerobic oxidation of methane was mxaC, which increased in relative abundance at high pressures relative to low pressures.

4 Discussion

4.1 Hydrostatic pressure changes shape species composition of carbon cycling-associated microbial communities

Hydrostatic pressure can affect the dissolved oxygen level, redox condition, acidity, and water temperature at the SWI (Michael et al. 2003). Therefore, hydrostatic pressure has a non-negligible influence on microbial growth and metabolism, which in turn mediates carbon cycling and migration at the SWI. The complexity of microbial networks in reservoir sediments under low-pressure and high-pressure conditions was analyzed to elucidate the influence of hydrostatic pressure on the interspecies interactions of microbial communities associated with carbon cycling at the SWI (Fig. 6). At the genus level, members of the bacterial phyla Proteobacteria, Actinobacteria, and Bacteroidetes played a leading role in the interspecies relationships of microbial communities in both low-pressure and high-pressure groups, despite their different proportions (Fig. 6a, b). This is in agreement with previous studies conducted in the Three Gorges Reservoir (Qin et al. 2021b) and Danjiangkou Reservoir (Hu et al. 2021) in China. Microbial taxa established associations with distinct taxa in the network depending on the pressure level, providing strong evidence for the influence of hydrostatic pressure on microbial community composition at the SWI in the reservoir.

Fig. 6
figure 6

Correlation networks of microbial communities (genus level) associated with carbon cycling at various hydrostatic pressures (r > 0.70, p < 0.05). a Microbial co-occurrence network in the low-pressure group (AU-TU: 0.1–0.2 Mpa). b Microbial co-occurrence network in the high-pressure group (FU-SU: 0.5–0.7 Mpa). The node size is proportional to genus relative abundance, and the thickness of each connection between two nodes is proportional to the Pearson correlation coefficient. c Network topological parameters for the low-pressure and high-pressure groups

Full size image

As the dominant bacterial phyla in soil, Proteobacteria, Actinobacteria, and Chloroflexi exhibit striking effects on dissolved organic carbon and microbial biomass carbon contents under different land use changes (Wang et al. 2020). Among the species with high relative abundances in reservoir sediments (Fig. 3c), Candidatus Rokubacteria bacterium showed an absolute dominance under various pressure levels. Actinobacteria bacterium, Chloroflexi bacterium, and Proteobacteria bacterium also played a major role in carbon cycling. The phyla Actinobacteria and Proteobacteria are major microbial groups that take part in the metabolism of carbon-containing compounds such as lignin. Additionally, Actinobacteria bacterium and Chloroflexi bacterium are important species for carbon fixation (Xu et al. 2021). Their growth and proliferation in sediment samples were promoted by elevating hydrostatic pressure in this study, as evidenced by the increased relative abundances at higher pressures. This could enhance the potential of carbon sequestration by microbial carbon fixation at the SWI from the perspective of unique taxa. Alphaproteobacteria and Betaproteobacteria were the most abundant classes in Proteobacteria under the experimental conditions. Betaproteobacteria is a typical dominant microbial group in river and lake ecosystems, whereas Alphaproteobacteria predominates the soil bacterial community in forest ecosystems (Xu et al. 2021).

Both Thaumarchaeota and Bathyarchaeota accounted for a high proportion of the total abundance of archaea in sediment samples (Fig. 3a). Thaumarchaeota plays a crucial role in the biogeochemical cycling of important elements such as carbon, nitrogen, and sulfur (Lin and Xie 2021), and some subgroups of Bathyarchaeota possess metabolic pathways related to methanogenesis (Xu et al. 2021). Recently, Yang et al. (2022) have characterized the structure of archaeal communities in sediments across different sections of the Shiwuli River in Chaohu Lake, China. They observed slow water flow and relatively low levels of dissolved oxygen in the downstream of the river. Under such deep-water conditions, only a small amount of dissolved oxygen was transferred to the sediment from the overlying water, which contributed to the occurrence of Thaumarchaeota. These findings provide indirect evidence that elevated hydrostatic pressure could enhance carbon fixation in deep water bodies.

The topological features of ecological networks were analyzed for microbial communities associated with carbon cycling at the SWI under various hydrostatic pressure conditions (Fig. 6c). It was found that microbial communities maintained higher connectivity and more complex interspecies interactions at high pressures than at low pressures. This was demonstrated by the higher graph density, average clustering coefficient, and average path length of microbial network in the high-pressure group, in agreement with the previous study of Wu et al. (2021). The decrease of microbial species under high pressure conditions mirrors the pattern observed by ZoBell and Johnson (1949) that high hydrostatic pressures (20–60 MPa) inhibited the growth of marine bacteria. High hydrostatic pressures were likely to hinder cellular metabolic activities in microbial communities, resulting in the possible death or dormancy of numerous microorganisms.

Furthermore, the modularity of microbial networks at low and high hydrostatic pressures was analyzed. The community modularity degree in the low-pressure group (0.634) was higher than that in the high-pressure group (0.586). The modularity parameter of both networks was > 0.4, suggesting a high level of modularity in the overall topologic structure of each network (Qin et al. 2021a). A similar network feature has been observed in complex environments such as soil (Williams et al. 2014), sediment (Cheung et al. 2018), and seawater (Reji et al. 2019). Moreover, in the high-pressure group, microbial communities formed a larger network with more correlations, whereas the opposite result was obtained from the microbial network in the low-pressure group. Therefore, the microbial communities in surface waters may have intimate interactions among species and rapid response to external perturbations, with poor community stability; in contrast, the microbial communities in deep waters have a buffering effect on external perturbations, with relatively high community stability (Wang et al. 2022a).

Microorganisms whose growth and reproduction rates are optimized in high-pressure environments of the deep sea are known as piezophiles or barophiles (Peoples et al. 2019). The majority of piezophilic bacterial isolates are gram-negative bacteria, including the genera Shewanella and Psychromonas of the class Gammaproteobacteria, as well as some groups of the classes Deltaproteobacteria and Alphaproteobacteria. Compared with atmosphere pressure, higher pressures resulted in increased relative abundances of two genera in Gammaproteobacteria (i.e., Psychromonas, Shewanella), three genera in Alphaproteobacteria (i.e., Brevundimonas, Microvirga, Rhodopseudomonas), and the class Bacteroidetes, although their optimum hydrostatic pressure was slightly different (Fig. 7a). Some of the above-mentioned genera have been reported as piezophiles in the Mariana, Kermadec, and Yap Trenches (Peoples et al. 2019; Zhang et al. 2020). Due to the small pressure range used in the present study, there were indistinct changes in the relative abundance of piezophiles under various levels of hydrostatic pressure. Still, the results support that hydrostatic pressure can affect microbial species composition associated with carbon cycling at the SWI in deep-water reservoirs.

Fig. 7
figure 7

Piezophilic microorganisms and genes in reservoir sediments at various hydrostatic pressures. a Piezophiles with significant differences. b Numbers of phyla containing asd, ompH, and shared piezophilic genes. c Species that harbored the piezophilic genes with significant differences. d Abundance of the piezophilic genes. e Microbial metabolites associated with carbon cycling

Full size image

To adapt to high-pressure environments, deep-sea microorganisms can evolve unique genes that are controlled by pressure. Some genes from barotolerant bacteria, such as ompH and asd, were not expressed when cloned into bacteria living in the atmosphere (Li et al. 2013). These two piezophilic genes were annotated in the present study, which exhibited similar trends in reservoir sediments at various hydrostatic pressures. Both ompH and asd were detected at the highest abundance in TU and FU groups, with the lowest in SU group. Notably, the absolute abundance of asd was higher than that of ompH (Fig. 7c), and was significantly different between FU and SU groups. Among the microbial communities associated with carbon cycling, there was a higher number of asd-harboring species than ompH-harboring species. Specifically, 146 species carried the asd gene only, 45 species contained the ompH gene only, and 22 species shared both genes. In total, the asd gene was detected in 19 phyla and the ompH gene in nine phyla, with the shared genes in six phyla (Fig. 7b). Among them, the phylum Proteobacteria comprised the largest number of species that contained the asd, ompH, and shared genes.

The species that harbored the asd, ompH, and shared genes with significant differences in abundance at various hydrostatic pressures are shown in Fig. 7d. The ompH gene abundance in Betaproteobacteria bacterium RIFCSPHIGHO2_12_FULL_69_13 was significantly higher at high pressures than at low pressures. The shared genes exhibited a similar trend to the ompH gene in some species of Proteobacteria (Deltaproteobacteria bacterium, uncultured Desulfatiglans sp.) and Candidatus (Rokubacteria bacterium). The abundance of most asd-harboring species trended upward from low to high hydrostatic pressure. Marietou and Bartlett (2014) et al. found that the relative abundances of α-Proteobacteria, γ-Proteobacteria, and Actinobacteria all increased with increasing hydrostatic pressure, possibly because piezophilic genes contributed to the formation of amino acids such as Asn, Cys, and Tyr (Zhao and Xiao. 2017) for adaptation to the high-pressure environment. However, most of the previous studies on piezophilic microorganisms and genes have been conducted in seawater environments, where the piezophilic taxa are different from those found in reservoir sediments. For example, γ-Proteobacteria decreased in relative abundance with increasing hydrostatic pressure in the present study.

4.2 Hydrostatic pressure changes drive differences in carbon cycling-associated microbial functional genes and metabolic pathways

In the deep sea, only 1% of the organic matter used for ecosystem maintenance comes from the deposition of organic particulate matter on the ocean surface, and most of the organic matter is synthesized by deep-sea microorganisms using CO2 to provide nutrients (Wang et al. 2022b). Deep-sea autotrophic and facultative autotrophic microorganisms (e.g., ammonia-oxidizing archaea) absorb and utilize CO2 produced by heterotrophic microorganisms (e.g., Proteobacteria) to form the internal carbon cycle in the deep sea. These microorganisms also absorb a large amount of CO2 transported vertically to the seafloor from seawater. Therefore, the unique microbial communities in the deep-sea environment with high hydrostatic pressure play a positive role in the transformation and cycling of carbon, especially carbon sequestration. In deep-water reservoirs, although the hydrostatic pressure is not as high as that of the deep sea, there are still unique microbial communities containing piezophilic taxa and exhibiting specific functional characteristics. Moreover, deep-water reservoirs experience considerable water level fluctuations and unnatural flooding-drying habitat changes. The associated changes in water depth cause substantial variation in hydrostatic pressure, which could shape microbial communities and drive their structural and functional differences compared with those in the deep-sea environment. For example, the Three Gorges Reservoir is operated using a cyclic impoundment pattern; it impounds water in the winter and discharges water in the summer. The water level variation in front of the Three Gorges Dam is as high as 145–175 m (Pan et al. 2016), and the resulting changes in hydrostatic pressure would affect microbially mediated carbon sequestration at the SWI in the reservoir. In this regard, it is of theoretical and practical significance to decipher the influence of hydrostatic pressure changes on microbially mediated carbon cycling in deep-water reservoirs.

Based on the functional contribution of microbial species in carbon metabolic pathways (Fig. 4c), microorganisms as the major drivers and undertakers of carbon cycling participated in multiple CO2 fixation pathways at the SWI (Fig. 8). Among the major carbon-fixing groups, Candidatus Rokubacteria, Acidobacteria, Deltaproteobacteria, and Betaproteobacteria contributed up to 20% in the carbon fixation process. While microorganisms are both producers and consumers of methane, their metabolic processes are likely to be governed by environmental factors (e.g., pH) and soil nutrients (e.g., total N, P) (Wang et al. 2023a). In the present study, three microbial groups, Candidatus Rokubacteria, Deltaproteobacteria, and Acidobacteria, prominently contributed to methanogenesis and methane oxidation at the SWI. Candidatus Rokubacteria contributed the most to methanogenesis while having substantial contribution to methane oxidation. Additionally, Betaproteobacteria was the largest contributor to methane oxidation, with minimal contribution to methanogenesis. The total contribution of these four groups to methanogenesis and methane oxidation reached 30%, which decreased with elevating pressure. These carbon fixation, methanogenesis, and methane oxidation processes were also detected in the sediments of deep-sea cold seeps, where bacterial phyla such as Proteobacteria, Chloroflexi, and Acidobacteria played a key role in carbon metabolic pathways (Jiang et al. 2023). Taking into account the variation in the microbial contribution to carbon cycling pathways, the findings indicate that higher hydrostatic pressure could alter microbial community structure, thereby hindering methanogenesis and methane oxidation at the SWI.

Fig. 8
figure 8

Differences in microbial functional genes and metabolic pathways associated with carbon cycling at various hydrostatic pressures

Full size image

With regard to the differential functional genes of carbon metabolism (Figs. 5, 8), there were considerable differences in carbon metabolic pathways at various hydrostatic pressures. The carbon fixation process involves the reductive pentose phosphate cycle (ALDO, rbcS, GAPDH), reductive citrate cycle (ACO, pycB, korA, sdhA, sdhC, fumB), and reductive acetyl–CoA pathway (metF, cdhD, cdhE). Among them, the ACO, pycB, and korA genes showed relatively high abundances, and multiple genes such as metF, ACO, sdhC, and sdhA had significant differences in response to changes in the hydrostatic pressure. ALDO, sdhA, sdhC, metF, and ACO gene abundances were increased by elevated pressures, in contrast to the downward trend in pycB, cdhD, and cdhE gene abundances. Wei et al. (2022) only annotated the ALDO and rbcS genes related to the Calvin cycle in surface sediments of the Mariana Trench. This means that microbially mediated carbon fixation pathways are slightly different between reservoir sediments and submarine trench sediments. Among the genes involved in methanogenesis (mch, mtrH, cdhE, cdhD), anaerobic oxidation of methane (mtrH), and aerobic oxidation of methane (mxaC), mch gene abundance was relatively high; mch and mxaC gene abundances exhibited an upward trend with elevating pressure, whereas cdhE and cdhD gene abundances displayed the opposite trend. Excluding cdhE and mxaC, these functional genes related to methane metabolism were also annotated in sediments of the Chao Lake (Zhou et al. 2023). Accordingly, reservoir sediments and lake sediments provide relatively similar habitat environments for microorganisms. With respect of the abundance of differential functional genes related to carbon metabolism, elevating the hydrostatic pressure promoted carbon fixation pathways—the reductive citrate and pentose phosphate cycles—at the SWI.

Furthermore, hydrostatic pressure had differential influence on the metabolites from various carbon cycling pathways (Fig. 7e). In the methanogenesis pathway, acetyl phosphate abundance was relatively high and increased with elevating pressure. The second and third most abundant metabolites were oxalacetic acid from the reductive citrate cycle and D-ribulose 5-phosphate from the reductive pentose phosphate cycle, respectively. The former decreased with increasing hydrostatic pressure, and the latter was most abundant in TU group. Other relatively abundant metabolites included succinic acid semialdehyde in the hydroxypropionate-hydroxybutylate cycle, as well as L-malic acid, succinic acid, and fumaric acid in the reductive citrate cycle. The three metabolites associated with the reductive citrate cycle all increased with increasing hydrostatic pressure. Recently, Shang et al. (2023) have found that microbially driven aerobic methane oxidation intercepts most of the CH4 in surface sediments of the Bohai Sea. As the intermediate products of methanogenesis and multiple carbon fixation pathways increased with elevating hydrostatic pressure, the present study indicates that higher hydrostatic pressure is conducive to reducing greenhouse gas emissions. The collective results suggest that microbially mediated carbon metabolism could enhance carbon sequestration potential at the SWI mainly by facilitating the reductive citrate and pentose phosphate cycles of carbon fixation and inhibiting methanogenesis.

5 Conclusions

Hydrostatic pressure is a major factor influencing the structural composition of sediment microbial communities. In this study, a simulation experiment was conducted with hydrostatic pressure as the only variable, corresponding to different water depths in a deep-water reservoir. Metagenomics and metabolomics were adopted to elaborate the response mechanisms of carbon cycling-associated microbial communities to pressure changes at the sediment–water interface (SWI) of the reservoir. The stability of sediment microbial communities was greater under high pressure conditions. The elevation in pressure contributed to the abundance of carbon-fixing bacteria such as Proteobacteria, Chloroflexi, and Actinobacteria, with piezophilic taxa and genes mainly found in the phylum Proteobacteria. Functional gene abundances of ALDO, ACO, sdhA, and sdhC increased in response to elevated pressure, along with the increase of metabolites from the reductive citrate and pentose phosphate (Calvin) cycles of carbon fixation and the accumulation of intermediate metabolites from the methanogenesis pathway to reduce methane production. The findings demonstrate that elevated hydrostatic pressure enhances carbon sequestration potential at the SWI of deep-water reservoirs by altering microbial community structure, functional gene abundance, and carbon metabolic pathways. This research provides insights into microbially mediated carbon cycling in deep-water reservoirs from a functional gene and metabolic pathway perspective. Future work is still needed to further explore related functional gene expression based on transcriptomic data, assess carbon sequestration by microbially mediated transformation of dissolved organic carbon from different sources to refractory organic carbon and refractory organic carbon production through the priming effect of fresh terrestrial organic carbon inputs at the SWI, as well as associated influencing factors and more explicit metabolic mechanisms of carbon transformation.