Protozoa-driven micro-food webs shaping carbon and nitrogen cycling in reservoir ecosystems

Abstract

Protozoa-driven micro-food webs are pivotal regulators of microbial community structure and carbon–nitrogen cycling. By mediating trophic cascades that regulate bacterial and algal populations, protozoa influence nutrient remineralization and energy flow. Their regulation is crucial for stabilizing biogeochemical processes and preventing harmful algal blooms. However, little is known about the detailed relationship between the traits of micro-food webs and carbon/nitrogen cycling processes. Using metagenomic data, we investigated the complexity and stability of micro-food webs in three distinct zones of the Fenhe Reservoir—the inflow river zone, shallow wetland, and deep-water zone—to assess their impacts on carbon and nitrogen cycling. Our findings revealed distinct spatial patterns in micro-food web complexity and stability, with the highest diversity and interaction density in inflowing river zones and a gradual simplification towards deep-water zones. Functional gene analysis shows significant differences in carbon degradation, fixation pathways, and nitrogen transformation processes, with shallow waters exhibiting strong microbial-mediated nitrification and denitrification, while deep waters rely on anaerobic nitrogen reduction pathways. Partial least squares path modeling (PLS-PM) indicated that protozoan-driven micro-food web structures regulate microbial functional differentiation, thereby influencing carbon and nitrogen cycle. Additionally, environmental parameters such as organic carbon concentration and nitrogen availability significantly shape microbial interactions and biogeochemical transformations. These findings highlight the intricate relationship between microbial community composition, food web stability, and elemental cycling, providing critical insights for reservoir ecosystem management and water quality optimization.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40793-025-00772-9.

Introduction

The carbon and nitrogen cycles plays a central role in aquatic ecosystem functioning, essential for maintaining primary productivity and ecological balance [1, 2]. The coupling mechanisms of carbon and nitrogen cycles are undergoing significant transformations due to global climate change and intensified human activities [3]. Bacterial and fungal communities, as the core drivers, regulate the efficiency of these cycles through the expression of functional genes involved in carbon (e.g., mcrA, pmoA) and nitrogen (e.g., amoA, nirK, nosZ) transformation pathways [4]. As a trophic level within the microbial food web, the community structure of bacteria and fungi can be regulated by predators, e.g. nematodes and protozoa. However, the maintenance mechanism of the structure and function of the micro-food web, as well as the main role of predators in it, are still unclear.

Aquatic food webs are known to play an important role in essential ecological functions and ecosystem services. Within this framework, planktonic microbial communities—comprising bacteria, fungi, microalgae, and protozoa—form the foundation of the aquatic micro-food web, particularly within detrital pathways, where dead organic matter (detritus) from primary producers and consumers is broken down by microbes, facilitating nutrient regeneration and energy transfer through microbial trophic interactions. The micro-food web concept in this study is defined as an ecological network centered on predator–prey relationships. Among these, protozoa serve as both primary consumers of bacteria, fungi, and other small eukaryotes, and as valuable bioindicators in applied research, [5–7]. These organisms control the complexity and stability of aquatic micro-food web by their interactions [8]. Although traditional theories have long debated the “complexity-stability” relationship in food webs (May, 1973), current understanding primarily stems from more easily observable soil food web studies [9]. Our understanding of the mechanisms that sustain both the complexity and stability of aquatic micro-food webs remains limited. Bridging this knowledge gap is critical for predicting how aquatic ecosystems respond to environmental disturbances and global change.

Meanwhile, the micro-food web, as the core structure of energy flow and material transfer in ecosystems, participates directly and indirectly in nutrient cycling by incorporating dissolved organic matter, detrital particles, and specific microbial groups (e.g., bacteria and algae) into higher trophic levels, thereby influencing ecosystem functions. [10] [11]. For example, protozoa as an important component of the micro food web, promote nutrient cycling and regulate microbial community structure through their feeding and excretion processes, profoundly influencing the structure and function of microbial communities [12–14]. As a bridge between microbial communities and ecosystem functions, the complexity and stability of the micro-food web determine the connectivity and efficiency of carbon and nitrogen functional pathways.[15]. While protozoan regulation of C–N fluxes is documented in lentic systems [16, 17],ammonium release through ciliate excretion and their transport activities can substantially enhance benthic ammonium flux [18], these mechanisms remain unexplored under the pulsed disturbances unique to reservoirs (e.g., dam discharge induced hydraulic mixing). Crucially, the dynamic interplay between resource influx (bottom-up) and predation pressure (top-down) during transient hydrological shifts constitutes a fundamental knowledge gap.

Reservoirs, as critical water resource management facilities, play key roles in both water supply and ecological protection [17]. Urban and rural residents’ sewage, livestock farming, and non-point source pollution in the upstream of rivers inevitably cause certain pollution to surface water and tributaries of reservoirs, which may have an impact on the water quality of reservoirs [19, 20]. Rivers enter the reservoir area and flow through shallow wetlands, where microbial driven biogeochemical cycles can effectively reduce the concentration of organic pollutants and purify the water quality to a certain extent [21]. The Fenhe Reservoir, located in Shanxi Province, northern China, receives significant inputs of organic matter and nutrients from upstream agricultural, urban, and industrial sources. Due to its ecological and environmental significance, the Fenhe Reservoir provides an ideal model system for studying microbial food web dynamics and carbon–nitrogen cycling across different functional zones. Therefore, what spatial changes have occurred in the structure of microbial communities and their water purification functions as water flows from shallow wetlands to deep-water reservoirs remains unclear. Clarifying the driving factors of these changes can provide theoretical support for the formulation of water quality management measures for reservoirs.

This study focuses on the role of micro-food web in C and N cycling in three typical functional areas of the Fenhe Reservoir—the inflow river zone, shallow water zone, and deep-water zone. We systematically analyzed the abundance and spatial distribution patterns of functional genes associated with carbon degradation, carbon fixation, and nitrogen cycling processes in these different zones. By quantitatively assessing the relationship between microbial functional genes and micro-food web characteristics, this study aims to reveal the regional differences and regulatory mechanisms of the carbon–nitrogen cycle in the Fenhe Reservoir ecosystem. Three hypotheses are proposed: (1) the stability of the micro-food web is inversely related to its complexity, and it is modulated by environmental conditions; (2) enhanced complexity of micro-food webs promotes carbon and nitrogen cycling potential rates; (3) the moderate predation pressure exerted by protozoa on microorganisms significantly influences the potential capacity involved in carbon and nitrogen turnover. This study aims to provide scientific evidence for eutrophication management and ecological restoration of reservoirs, as well as offer new perspectives for understanding and regulating global carbon–nitrogen cycle dynamics.

Material and methods

Site description, sampling procedures, and physicochemical property analysis

Study area includes the Fenhe Reservoir and upstream tributaries, which are located in, Taiyuan, Shanxi Province (111°89′17″E, 38°08′92″N) (Fig. 1). The region features a northern temperate continental monsoon climate with an annual precipitation of 428 mm. It is 15 km long from north to south and 5 km wide from east to west, with a total area of 32 km² and an average depth of 6.50 m.

Fig. 1.

Schematic diagram of the Fenhe Reservoir showing the locations of sampling sites. The inset map (upper left) indicates the geographic location of Shanxi Province, China, with the area highlighted in red shading. The red circle in the main map delineates the geographic location of the Fenhe Reservoir within Shanxi Province, Green dots represent the 15 sampling sites distributed across different functional zones (inflow, transition, and lacustrine zones) of the reservoir. Sec1–Sec5 refer to Section 1 to Section 5 within the Fenhe Reservoir, Tri-FR refers to the Tributary-Fenhe River sampling site, Tri-LR refers to the Tributary-Lanhe River sampling site, Tri-Mix refers to the mixing point of the two tributaries

The study area divided into three regions based on the physical and chemical characteristics and ecological functions of the water body, namely the upstream confluence area, shallow water area, and deep-water area. Three transects have been set up for upstream water respectively, namely Tributary-Fenhe River (Tri-FR), Tributary-Lanhe River (Tri-LR), and the Tributary-mixing point of them (Tri-Mix). Five transects were established in the Fenhe Reservoir, labeled Sect. “Introduction”(Sec1), Sect. “Material and methods”(Sec2), Sect. “Results”(Sec3), Sect. “Discussion”(Sec4) and Sect. “Conclusion”(Sec5). The shallow water zone, situated at the reservoir’s inlet, encompasses Sec4 and Sec5, while the deep-water zone includes Sec1, Sec2, and Sec3. Given that microbial diversity and metabolic activity are relatively high in summer reservoirs of northern China, this period provides an optimal window to capture microbial community composition and food web structural attributes across distinct functional zones. Accordingly, we conducted field sampling in July 2022, collecting 75 water samples from 15 spatially distributed sites. The samples were immediately transported to the laboratory after its collection in the same day.

Water samples (approximately 3 L per site) were collected using a sterilized water sampler and transferred into pre-sterilized polyethylene containers. All sampling tools, containers, and associated materials were sterilized prior to sampling. Immediately after collection, samples were stored in a portable cooler with ice packs and transported back to the laboratory for further processing. Then we divided each sample into two subsamples for analyzing physicochemical properties and DNA extraction (stored at −80 °C).

The following physicochemical factors were measured according to previously described methods [22]. Environmental properties such as water pH, ammonium (NH₄⁺–N), and nitrate (NO₃⁻–N) content were monitored at the sampling site using a portable multiparameter water monitoring probe (Aquaread AP-5000, UK). Total organic carbon (TOC), total carbon (TC), and inorganic carbon (IC) were analyzed using a TOC analyzer (Shimadzu, TOC-VCPH, Japan). Sediment pH was determined in 1 M KCl sediment suspension [sediment: water ratio of 1:2.5(w/v)]. NH₄⁺–N and NO₃⁻–N were determined by automated discrete analysis (CleverChem 380, Germany). TC was determined by elemental analysis (Elementar Vario MACRO, Germany). Organic carbon (SOC) was measured using the K₂Cr₂O₇ oxidation method.

DNA extraction, PCR amplification, and Illumina MiSeq sequencing

Microorganisms in water samples were collected by filtration through a 0.2 μm pore size membrane filter (Millipore, Jinteng, China). The biomass-containing filters were cut into pieces, placed into centrifuge tubes, and subjected to DNA extraction by MagaBio Soil/Feces Genomic DNA Purification Kit (Hangzhou Bioer Technology Co. Ltd.). The specific conditions of PCR amplification and purification can be found in previous work [23].

The primers 338F (5′-ACTCCTACGGGAGGCAGCAG-3′) and 806R (5′-GGACTACHVGGGTWTCTAAT-3′) were used to amplify the V3-V4 region of the 16S rRNA bacterial gene [23]. The variable region of the ITS genes was amplified using the universal forward primer ITS1F (5′-CTTGGTCATTTAGAGGAAGTAA-3′) and uni-versal reverse primer ITS1R (5′-GCTGCGTTCTTCATCGATGC-3′) [22]. The V4 region of the 18S rRNA gene was amplified using the primers TAReuk454FWD1F (5′-CCAGCASCYGCGGTAATTCC-3′) and TAReukREV3R (5′-ACTTTCGTTCTTGATYRA-3′) [14]. After purification and quantification of the PCR products, genomic DNA libraries were constructed on an Illumina MiSeq platform (Guangdong Magigene Biotechnology Co., Ltd., Guangdong, China). The sequencing reads of fungi, metazoans, archaea, and unclassified sequences were removed from the 18S rRNA sequences. Detailed steps of the data processing can be found in the previous study. The number of OTUs (Operational Taxonomic Units) was analyzed for each sample using a 97% sequence similarity cutoff value. To minimize the difference of sequencing depth across samples, all data were normalized to the number of sequences in the smallest data set for further analysis.

Metagenome sequencing and analysis

Genomic DNA was extracted using ALFA-SEQ Advanced Soil DNA Kit (Findrop Biosafety technology (Guangzhou) Co.Ltd). The integrity and purity of the DNA were assessed via 1% agarose gel electrophoresis. Quality-approved DNA samples were subjected to library preparation using the ALFA-SEQ DNA Library Prep Kit, following the instructions provided with the kit. The size distribution of the libraries was evaluated on the Qsep400 High-Throughput Nucleic Acid-Protein Analysis System (Hangzhou Houze Biological Technology Co., Ltd., China), while the library concentrations were measured using the Qubit 4.0 (Thermo Fisher Scientific, Waltham, USA). DNA Metagenome libraries were sequenced on the Illumina NovaSeq 6000 (Guangdong Magigene Biotechnology Co., Ltd., Guangzhou, China).

Use the Fastp software (with parameters: -5-W 5-M 20-q 15-u 40-l 50-dedup) to process the raw data to remove adapter sequences, trim low-quality bases, and filter out short and duplicate reads, thereby generating quality-controlled data for subsequent analyses [24]. Assembled scaffolds are subsequently interrupted at N-connection sites, leaving behind Scaftigs devoid of N-connections. Megahit was then employed to individually compare each Scaftig with the Clean Data from all samples, yielding paired-end (PE) reads that were not utilized in the assembly process. The resulting mixed scaffolds are broken at N-connection points, generating Scaftigs. All Scaftigs longer than 500 bp are then selected for statistical analysis. ORF prediction was carried out on scaftigs (≥ 500 bp) from individual samples and the mixed assembly using Prodigal (Version:2.6.3) [25]. Default parameters are applied to filter out length information shorter than 90 nt from the prediction results. Mmseqs2 was employed to eliminate redundancy, generating a unique initial gene catalog, where genes refer to nucleotide sequences encoded by a continuous, distinct gene [26]. Clustering is performed at 95% identity and 90% coverage, with the longest representative sequence chosen for each cluster. Clean data from each sample is mapped to the initial gene catalog using BBMap software, yielding the read count for genes mapped in each sample [27]. Based on the mapped read counts and gene lengths, abundance information for each gene in every sample was calculated.

Utilizing Diamond software to perform alignments of single-strain bacterial, fungal, and protist sequences obtained from the NCBI NR database [28]. For each sequence’s final alignment outcome, given that multiple alignment results may exist for a single sequence, adopt the Lowest Common Ancestor (LCA) algorithm within the MEGAN software system classification to determine the species annotation information for the sequence by selecting those with an E-value of 1 × e⁻¹⁰. Based on the LCA annotation results, we integrated the gene average sequencing depth (AVG depth) table and the gene abundance table to generate a comprehensive summary table. This table contains the AVG depth and abundance information of all annotated genes, organized by taxonomic level (Kingdom, Phylum, Class, Order, Family, Genus, and Species) for each sample. The functional annotation of Unigenes was conducted using the Diamond software against relevant databases (http://www.kegg.jp/kegg/). All mapped reads were normalized with the RPKM (reads assigned per kilobase of target per million mapped reads) method [29].

Microbial food web construction and topology

Trophic species are represented as nodes in the network, which can correspond to taxonomic groups at the genus level, organisms that share the same sets of predators and prey (e.g., algae, bacteria, and fungi), and non-living compartments of matter and energy (e.g., detritus and dissolved organic carbon). They can correspond to: algivores (A), bacterivores (B), mycophagous (M), nonselective omnivores (N), heterotrophic parasites (H-P), phototrophs (P), raptors (carnivorous protists) (R), saprotrophs (S), and unknown (U). We collected and compiled trophic interactions (prey-predator) of species present in each site (Table S4), which were assigned based on “New Technology of Micro Biomonitoring” and the Protist Interaction Database (PIDA; 10.5281/zenodo.1195514) [13]. The trophic network was defined by an adjacency matrix of pairwise interactions, in which each element a _ij = 1 when the j-genus preyed on the i-genus and a _ij = 0 otherwise [10]. We described the network structure and complexity with metrics that are widely used in food web studies, such as link density, connectance, mean trophic level and omnivory (Table S2). The “multiweb” R package was applied to calculate all network metrics and micro-food web simulations [10]. Micro-food web visualization was carried out using the Gephi interactive platform (https://gephi.org). The input data for constructing the micro-food web model were exclusively derived from the 18S rRNA high-throughput sequencing dataset, which provided taxonomic profiles of eukaryotic organisms (protozoa) at the genus level.

Statistical analysis

Microbial diversity index was calculated by the R package ‘vegan’ (v. 4.1.3), and differences in these indices of microbial community were assessed through one-way ANOVA with Benjamini–Hochberg FDR correction. The effects of physicochemical factors on the topological characteristics of micro-food web were estimated by mantel test analyses and visualized with heat maps. To better elucidate the difference in functional profiles across sites, we initially identified genes associated with key biogeochemical processes related to essential elements such as carbon and nitrogen (Fig. 6). Then analyzing the spearman correlation between functional gene abundances and environmental variables by the R package ‘Hmisc’, with FDR adjustment for multiple comparisons. The impact of biotic and abiotic factors on the functional cycle was comprehensively analyzed through the partial least squares path modeling (PLS-PM) by using the R package ‘plspm’. After removing the variables with loadings < 0.7, performing the final PLS-PM structure equation with the remaining variables, the prediction performance of the model was evaluated using the goodness of fit index (Gof) and R². The confidence interval of all statistical analyses was 95% (p < 0.05).

Fig. 6.

Effects of different variables on the ecosystem function based on partial least squares path modeling. The red arrows represent positive pathways and the blue arrows indicate negative pathways. The path coefficients represented by the thickness of the arrow. GOF, goodness of fit. Only significant relationships are shown (p < 0.05)

Results

Spatial dynamics of physicochemical parameters and microbial diversity

There were significant differences in the physicochemical parameters among the three regions (p < 0.001) (Fig. 2, Table S1). The inflowing river region exhibited characteristics of high nutrient input, including elevated nutrient levels in Tri-LR (NO₃⁻–N: 13.99 mg/L, NH₄⁺–N: 0.22 mg/L, TC: 47.12 mg/L) and Tri-FR (NO₃⁻–N: 10.67 mg/L, NH₄⁺–N: 0.17 mg/L, TC: 37.84 mg/L), as well as high salinity (SAL: 0.94) and conductivity (EC: 1917 µS/cm) in Tri-FR. The dissolved oxygen concentration was relatively low (DO: 6.55–7.77 mg/L), likely because high nutrient inputs promote eutrophication and subsequent microbial decomposition processes, which increase oxygen consumption. In the shallow water areas (Sec4, Sec5), pH significantly increased (8.13–8.26), DO (8.93 mg/L–10.23 mg/L) and TOC concentration (4.83 mg/L–5.50 mg/L) raised sharply, while NO₃⁻-N concentration (3.02 mg/L–6.12 mg/L) decreased significantly, indicating that shallow water areas are crucial for nutrient conversion and preliminary purification. In the deep-water areas (Sec3, Sec2, Sec1), physicochemical parameters tended to stabilize, with DO (9.47–9.79 mg/L) and pH (8.37–8.57) remaining at high levels, while nitrogen (NO₃⁻-N: 4.07–5.35 mg/L, NH₄⁺-N: 0.11–0.13 mg/L) and organic matter concentrations (TC: 31.28–34.21 mg/L, TOC: 4.24–5.27 mg/L) decrease further. This suggests that deep-water areas have a strong ability to regulate and buffer high-nutrient substances, and the water body has reached a state of equilibrium.

Fig. 2.

Physicochemical parameters of sampling sites

The microbial diversity was highest in the inflowing river region, while it significantly decreased as the water entered the open shallow water region (p < 0.001) (Fig. S1). The strong nutrient supply and hydrodynamics in the inflowing river region provided a favorable environment for microbial growth. As water moved toward the deep-water areas, the diversity values tended to level off, showing smaller fluctuations across sites. Although bacterial diversity decreased markedly, the diversity of fungi, protozoa, and algae showed less pronounced variability. These patterns suggest differences in the sensitivity of microbial groups to environmental gradients. The observed spatial trend in microbial diversity is consistent with the upstream-to-downstream transition in physicochemical conditions, reflecting a shift from nutrient-enriched, dynamic environments to more stabilized and homogeneous conditions.

Spatial dynamics of micro-food web characteristics

From the river region to the shallow water zone and then to the deep-water zone, the complexity of the micro-food web gradually decreased, while stability increased, as indicated by a decline in Qss values (a lower Qss denotes higher stability) (Table S2). The river region (Tri-FR, Tri-LR, Tri-Mix) is a key node for external material input and primary decomposition. Species richness (S: 113–121) and number of interactions or links (L: 3159–3737) reached their highest levels in the entire system, and connection density (LD: 27.96–30.88) indicating frequent inter-species interactions. The main functional groups were bacterivorous protozoa (B) and omnivorous protozoa (N), while herbivorous protozoa (A) and mixed functional groups (e.g., B/A, B/R) had lower proportions (Fig. 3). The quasi-stability (Qss: 9.73–10.16) was the highest in the system, suggesting that despite the high complexity, this region was sensitive to external disturbances, with lower stability.

Fig. 3.

Graphic representation of micro-food webs in different sites. A Sec1, B Sec2, C Sec3, D Sec5, E Sec4, F Tri-Mix, G Tri-LR, H Tri-FR. The size of the node represents the number of connections between nodes; node colors represent different phyla, while the arrow direction represents the predation relationship between different genera. The circles indicate the different functional groups to which they belong. The circle colors denote algivores, bacterivores, mycophagous, nonselective omnivores, and raptors

In the shallow water zone (Sec4, Sec5), the micro-food web complexity was slightly lower than that in the inflowing river region, but functional group differentiation had significantly increased, showing a dynamic balance between complexity and stability. In terms of functional group composition, bacterivorous and omnivorous protozoa (N) remained dominant, while the proportion of algal-feeding functional groups (A) and mixed functional groups (e.g., B/A, R/A) had significantly increased, indicating that phototrophic organisms, such as algae, likely provided an increasingly important energy source to the micro-food web. Physicochemical conditions showed that DO (8.93–10.23 mg/L) and TOC (4.83–5.50 mg/L) had increased significantly, enhancing the combination of primary production and heterotrophic processes. Qss (9.66–9.51) was lower than that in the river region, suggesting that the system had reached a dynamic balance between complexity and stability.

In the deep-water zone (Sec1, Sec2, Sec3), the micro-food web complexity was the lowest, but the concentration of functional groups and simplification of the system significantly enhanced the stability. Species richness (S: 69–95) and Number of interactions or links (L: 1233–2198) had significantly decreased, and connection density (LD: 17.87–23.14) was the lowest in the entire system. Functional groups were concentrated in bacterivorous protozoa (B) and omnivorous protozoa (N), with predatory functional groups (R) and complex mixed functional groups (e.g., B/R, B/R/M), indicating that the food chain was gradually becoming more vertical. Physicochemical conditions showed that TOC levels were higher in the deep-water zone (Sec3: 5.27 ± 0.42 mg/L), reflecting that organic matter sedimentation provides abundant resources for heterotrophic decomposition. Higher DO (9.47–9.79 mg/L) extended the characteristic path length (CPL: 1.21), suggesting that energy transfer efficiency had decreased (Fig. S2). Qss (7.76–8.83) was the lowest in the system, indicating that the deep-water zone had the strongest resistance to external disturbances, becoming the ecological endpoint for energy storage and material deposition.

Carbon and nitrogen cycles and their coupling processes

The functional genes related to the carbon cycle showed distinct spatial differentiation between the shallow water zone and the deep-water zone (Fig. 4). Carbon fixation and degradation processes exhibited higher activity in the shallow water zone (Sec4 & Sec5), particularly in Sec5, although the differences from the deep-water zones (Sec1–Sec3) were relatively small. The genes related to metabolic of carbohydrates and polysaccharides were enriched in Sec5-1 and Sec4-1 (0.006282 and 0.004389), and that of lipids genes showed higher expression in Sec5-1 (0.001025), indicating that the large amount of organic carbon in shallow waters promoted carbon degradation processes, providing sufficient carbon sources for microbial activity. In terms of carbon fixation, the Calvin–Benson–Bassham (CBB) cycle was significantly enriched in Sec5-1 and Sec5-2 (0.0016 and 0.0014). In contrast, the relative abundance of carbon cycle functional genes was lower in the deep-water zone (Sec1& Sec2& Sec3). Carbon degradation was focused on the breakdown of organic acids and proteins. Carbon fixation primarily relied on the Wood-Ljungdahl pathway (Sec1: 0.000422; Sec2: 0.000432) and the Reductive Citrate Cycle, which are characteristic of anaerobic or microaerophilic bacteria and archaea, rather than phototrophic organisms that rely on the Calvin cycle. These metabolic pathways were dominated by Proteobacteria and Acidobacteria, suggesting that microbial communities in this zone are adapted to conditions with limited carbon sources and relatively high dissolved oxygen. However, the elevated oxygen levels might not necessarily indicate greater oxygen availability, but could instead reflect reduced heterotrophic respiration activity in these deeper waters (Fig. 5).

Fig. 4.

Heatmap showing the enrichment of functional genes involved in A carbon degradation, B carbon fixation, C nitrogen cycling, D carbon–nitrogen cycle coupling. Log (FC): log2-fold change. Colors in Fig. (D) represent spearman correlations between Organic carbon (Org-C) concentration and nitrogen cycling genes. Colors represent Spearman correlations. * represent the degree of significance [p < 0.01 (**), p < 0.05 (*)]

Fig. 5.

Sankey diagram showing the difference in the contributions of microbial groups to individual biogeochemical processes among different sites, with the respective taxonomic classification and category of nutrients

The spatial distribution of nitrogen cycle functional genes showed significant differences between the shallow and deep-water zones. Nitrification genes (amoA, amoB, and amoC) were significantly enriched in the shallow water zone, particularly in Sec5-2 (2.26E−06, 2.32E−06, and 4.02E−06). Denitrification genes (nirS and norB) were concurrently enriched in Sec5-1 (3.5E−05, 3.6E−05), reflecting the efficient conversion of nitrate to nitrogen gas. This suggested that the microbial community in shallow water zones could cooperatively complete both nitrification and denitrification processes in high-oxygen environments. In the deep-water zone, assimilatory nitrate reduction to ammonium (ANRA) (nirA: 7.1E−06; narH: 4.61E−06) became the dominant pathway.

The carbon and nitrogen cycles showed distinct coupling relationships. Microbial metabolic potential may be affected by the redox status, which necessitate a series of reductive reactions involving diverse alternative electron acceptors, such as nitrate, carbon dioxide, and small organic molecules [30]. Generally, the utilization of organic carbon by microorganisms necessitates the availability of terminal electron acceptors, with the preferred order being O₂, NO₃⁻, followed by methanogenesis and other small organic molecules. Correlation analysis showed that denitrification genes (nirS and norB) were significantly positively correlated with carbon degradation genes (such as carbohydrates degradation) (p < 0.05) (Fig. S3), and were strongly consistent with TOC (M² = 0.3509, p < 0.05) (Fig. S4), indicating that organic carbon decomposition may provide electron donors and energy for denitrification, thereby promoting the efficient coupling of carbon and nitrogen cycles. Carbon degradation and dissimilatory nitrate reduction processes showed a close coupling relationship. Carbon fixation pathways, such as the Wood-Ljungdahl pathway, were co-occurring with dissimilatory nitrate reduction genes (napA and nrfA), indicating that organic acid decomposition may provide energy and electron donors for nitrogen conversion.

Biotic and abiotic factors affecting the carbon and nitrogen cycles

PLS-PM analysis showed that the structure of the micro-food web was jointly regulated by microbial community diversity and environmental factors, and significantly impacted carbon and nitrogen cycle functional genes (Fig. 6) (Table S3). Firstly, microbial community diversity was an important indirect driver of carbon and nitrogen cycle functions. In particular, protozoa, through their predation and competition with bacteria and fungi, indirectly promoted the transformation of carbon and nitrogen cycles. Secondly, the complexity and stability of the micro-food web significantly influenced the carbon and nitrogen cycle functions through the efficient transfer of energy and materials. In the shallow water zone, the higher micro-food web complexity, with the active roles of algal-feeding (A) and omnivorous (N) functional groups, drove the synergistic enhancement of carbon fixation and nitrification functions. In contrast, the micro-food web in the deep-water zone enhanced the coupling of carbon degradation and dissimilatory nitrate reduction through functional group concentration and verticalization of the food chain, highlighting the regulation of micro-food web structure on carbon and nitrogen cycle functions. These results support others hypothesis: the complexity of the microbial food web was significantly correlated with the potential rates of carbon and nitrogen cycling. The predation pressure exerted by protozoa on microorganisms significantly influences the potential capacity involved in carbon and nitrogen turnover.

Additionally, environmental factors (NO₃⁻, NH₄⁺, and TOC) were associated with variations in diversity of microbial community and construction of micro-food web, which may in turn influence carbon and nitrogen cycle functions. PLS‑PM analysis revealed significant positive associations between NO₃⁻/NH₄⁺ concentrations and microbial diversity, which could be enhanced carbon degradation and nitrogen transformation potential. While these nutrients may create favorable niches for diverse microbial assemblages, we acknowledge that microbial diversity is likely shaped by a broader range of biotic and abiotic factors not fully captured in this study.

Discussion

Relationship between the stability and complexity of micro-food webs

In the present study, it seems that spatial variation in environmental factors drove the spatial dynamics of micro-food web structure by regulating pH, DO, nitrogen concentrations, and TOC inputs. This causes a transition from high complexity and low stability in river inflow areas to a dynamic equilibrium in shallow waters, ultimately leading to low complexity and high stability in deep-waters. In river regions with high nutrients and low oxygen (Tri-FR, Tri-LR, and Tri-Mix), the micro-food web exhibited the highest complexity, weaker modularity, and lower stability. In shallow waters (Sec4 and Sec5), significant changes in environmental conditions, such as rapid increased in DO and pH, promoting the rapid growth of the algivorous functional group (A), achieving a dynamic balance between complexity and stability. In deep-waters (Sec1, Sec2, and Sec3), significant reductions in NO₃⁻–N and NH₄⁺–N lead to the lowest complexity of the micro-food web, with significantly enhanced system stability and strong resistance to external disturbances.

This spatial pattern can be better understood by considering the combined effects of physicochemical and biological factors. Previous studies have shown that physicochemical parameters such as temperature, dissolved oxygen (DO), and pH not only directly influence species distribution (Guo et al., 2024) but also indirectly regulate food web structure by altering microbial metabolic activity [31]. For instance, microbial decomposition significantly lowers environmental dissolved oxygen levels and alters pH, which in turn adversely affects smaller and medium-sized animals [32]. This mechanism may partly explain the low stability of food webs in oxygen-depleted riverine regions. Moreover, protozoan diversity plays a unique regulatory role in this process: by modulating bacterial population dynamics and enhancing energy flow efficiency, it can significantly increase both the complexity and stability of micro-food webs [33, 34]. These interactions suggest that the simultaneous improvement of physicochemical conditions and protozoan diversity in shallow waters may be key to maintaining a dynamic balance between complexity and stability, while in deep-water areas, nitrogen limitation and reduced microbial processes contribute to greater overall stability.

These findings support our first hypothesis that the stability of micro-food webs is inversely related to their complexity; that is, higher complexity may correspond to lower stability. However, this relationship manifests differently under varying environmental conditions and is jointly regulated by the complex interplay between physicochemical and biological factors [35, 36].

Distribution of carbon and nitrogen and their coupling processes

This study clearly demonstrated that functional genes involved in carbon and nitrogen cycling exhibited significant spatial differentiation between shallow and deep-water zones, further supporting the notion that biogeochemical cycling processes are jointly regulated by biotic factors (e.g., microorganisms, functional genes/pathways) and abiotic factors (e.g., pH, electron donors, electron acceptors) as well as their interactions [37]. Microbes, as the key “engines” of these processes, profoundly influence ecosystem functions by modulating carbon and nitrogen cycling metabolic pathways [38, 39].

Compared to the deep-water zone, the shallow-water zones (Sec4 and Sec5) showed higher activity in carbon fixation, carbon degradation and nitrification. Previous studies have shown that in oxygen-rich surface or shallow aquatic environments, the Calvin–Benson–Bassham (CBB) cycle is often the dominant carbon fixation pathway [40], where ammonia-oxidizing bacteria, such as Nitrosomonas, extensively express RuBisCO and depend on this pathway for autotrophic growth [41]. It is consistent with our observation that Nitrification genes (amoA, amoB, and amoC) were significantly enriched in the shallow water zone, particularly in Sec5-2 (2.26E−6, 2.32E−06, and 4.02E−06). At the same time, shallow-water zones with higher concentrations of organic carbon typically exhibit greater enrichment of genes associated with carbohydrate, polysaccharide, and lipid metabolism, reflecting the availability of complex substrates that can be rapidly degraded to supply energy and electron donors for microbial metabolism [42].

In contrast, the deep-water zones (Sec1–Sec3) exhibited significantly lower relative abundances of carbon cycling functional genes. Carbon degradation in these zones mainly relied on the breakdown of simpler substrates such as organic acids and proteins, consistent with the limited availability of complex carbon compounds. Carbon fixation was dominated by the Wood–Ljungdahl pathway and the reductive citrate cycle, which are ancient, energy-efficient pathways adapted to oxygen-deficient and carbon-limited environments [43]. Nitrogen cycling in deep-water zones was mainly driven by assimilatory nitrate reduction to ammonium (ANRA; nirA, narH). Previous studies have shown that in such systems, the scarcity of organic carbon limits the availability of electron donors, which constrains full denitrification [44].

Correlation analyses in this study further revealed that denitrification genes (nirS and norB) were positively correlated with carbon degradation genes, especially those related to carbohydrate degradation, as well as with total organic carbon (TOC) concentrations. This pattern suggests that organic carbon decomposition supplies the electron donors and energy required for denitrification in the shallow-water zones. Furthermore, the co-occurrence of the Wood–Ljungdahl carbon fixation pathway with dissimilatory nitrate reduction genes (napA, nrfA) in deep-water zones may provide the energy and electron donors necessary for nitrogen transformation.

The relationship between micro-food webs and carbon–nitrogen cycles

Microbial interactions form micro-food webs, which play a crucial role in regulating ecological functions [45]. The interactions between these two directly influence the productivity, stability, and adaptability of ecosystems to external disturbances [46]. The species interactions within a community create a positive complexity-stability relationship. The interactions between multi-trophic-level microorganisms can influence the stability of micro-food webs. Guo et al. found that the richness and composition of bacterivorous protozoa are significant predictors of the complexity and stability of microbial food webs [13]. Predatory behavior can significantly alter the structure of micro-food webs by regulating species diversity at different trophic levels, indirectly promoting the growth of primary producers, and increasing the efficiency of carbon and nitrogen fixation and removal. In eutrophic waters, predators, by controlling lower-level consumers in the food chain, indirectly promote the growth of primary producers, which affects the efficiency of carbon and nitrogen transformation [47]. This process reflects how the interactions between different species along the food chain influence the carbon and nitrogen cycles in ecosystems.

The results of this study show that protozoa not only directly affect the structure and stability of micro-food webs but also indirectly influence bacteria, fungi, and algae through their interactions. These interspecies interactions affect energy flow at each stage of the micro-food web and directly influence the carbon and nitrogen cycles. For example, in competition between phytoplankton and zooplankton, phytoplankton fix carbon through photosynthesis, but if the zooplankton population is too large, they may consume large amounts of phytoplankton, affecting primary productivity and carbon fixation capacity [48]. In other words, the diversity and complexity of food webs contribute to enhancing the efficiency of carbon–nitrogen cycling processes and improving the function and stability of the entire ecosystem. In order to better verify the effects of protozoa on the food web and microbial carbon and nitrogen cycles, we plan to further verification of the regulatory mechanism through protozoan removal experiments and isotope labeling tracking.

In addition to interspecies interactions, the spatial structure of micro-food webs is also influenced by the physicochemical conditions of the water body. Long-term carbon storage in the form of TOC can form the foundational resource of the food web detritus. The study of Wang et al. have shown that TOC plays a dual role in maintaining aquatic food web productivity and stability, as well as in promoting global carbon fluxes [49]. Zooplankton primarily rely on algae as their base resource, but in cases where algal concentrations are low, they may feed on heterotrophic microorganisms that decompose total organic matter.

Conclusion

This study systematically investigates the microbial communities, micro-food web structures, and carbon–nitrogen cycling functions in different regions of the Fenhe Reservoir, revealing the spatial dynamics and driving mechanisms of the reservoir’s ecosystem functions. A significant synergistic relationship exists between micro-food web structure and carbon–nitrogen cycling functions, where the diversity of microbial communities and the complexity of micro-food webs jointly drive the carbon–nitrogen cycling processes. Protozoa, as key nodes in energy transfer within the micro-food web, regulated the functional differentiation of microbial communities by preying on bacteria, algae, and fungi, significantly influencing the transformation processes of carbon and nitrogen cycling. Environmental factors, in turn, affect ecosystem functions by regulating microbial communities and the structure of micro-food webs. This relationship highlights the spatial dynamics of carbon–nitrogen cycling in different regions and their ecological regulation mechanisms.

Author contributions

X.W.: Writing—Original draft, Software, Methodology, Formal analysis, Visualization, Data curation. J.X.L.: Writing—review and editing, Investigation, Funding acquisition. B.F.C.: Writing—review and editing, Funding acquisition. T.H.W.：Writing—review and editing, Methodology.

Funding

This work was supported by the National Natural Science Foundation of China (Grant No. U23A20157), the Central Government Guided Local Science and Technology Development Funds Project (Grant No. YDZJSX2022B001), and the Key Cooperation Project of National Science and Technology in Shanxi Province (Grant No. 202304041101020).

Availability of data and materials

Raw data were deposited in the Sequence Read Archive of NCBI under the accession number PRJNA1257555 (https://dataview.ncbi.nlm.nih.gov/object/PRJNA1257555?reviewer=sblpks9ukbq7miks4vbagvdrqv).

Declarations

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.