Dynamics of the sedimentary bacterial communities in a plain river network: similar coalescence patterns with bacterioplankton communities driven by distinct assembly processes

ABSTRACT

Riverine microorganisms are major players in driving biogeochemical cycling. The dynamics of hydrological conditions have been observed to influence the microbial communities in rivers, but their specific impacts on the coalescence patterns of the bacterial communities remain largely unexplored. In this study, we examined the dynamics of the sedimentary bacterial communities and their coalescence patterns with bacterioplankton communities in a plain river network under dynamic hydrological conditions. Our results revealed significant temporal variations in the diversity, structure, and taxonomic composition of the sedimentary bacterial communities across the wet, normal, and dry seasons (P < 0.01). Despite the varying hydrological conditions, similar coalescence patterns were observed between the sedimentary bacterial communities and bacterioplankton communities. The bacterial communities from the same habitat type showed higher proportions of overlapped species, and the coalescence was more prominent within the same medium (intra-medium) compared to that between different media (inter-medium). However, the proportions of stochasticity in community assembly were distinct during different hydrological conditions, with evidently increased stochasticity observed in the wet season compared with other seasons. These results implied that the similar coalescence patterns between the sedimentary bacterial communities and bacterioplankton communities were driven by distinct assembly processes. The results of this study expanded our understanding of microbial coalescence in a river network and might shed light on how microbial communities are assembled in river ecosystems.

IMPORTANCE

Microorganisms play important roles in driving the biogeochemical cycles within river ecosystems. It has been suggested that hydrologic conditions could influence microbial communities in rivers, but their specific effects on the behaviours of microbial coalescence have not been thoroughly investigated. In this study, the dynamics of sedimentary bacterial communities within a plain river network were analyzed by amplicon sequencing followed by several ecological models to uncover the underlying assembly processes. Additionally, a comparative analysis between bacterioplankton communities and sedimentary bacterial communities was performed to unveil their coalescence patterns. The results suggested that similar coalescence patterns between sedimentary bacterial and bacterioplankton communities were driven by distinct assembly processes under dynamic hydrological conditions. These findings enhanced our understanding of microbial diversity features within river ecosystems.

INTRODUCTION

Microorganisms play critical roles in driving global biogeochemical cycling and providing ecological services (1). A deep understanding of the dynamics of microbial communities is vital for accurately predicting their functions and harnessing their potentials in these important processes (2). Rivers are the fundamental components of natural environments and serve as essential habitats for microorganisms. Dispersal events are prevalent and frequent in river ecosystems, and microbial communities in flowing water engage in microbial coalescence (3, 4), resulting in the merging of separate microbial communities with the mixing of surrounding environments (5). Therefore, exploring microbial coalescence can contribute to our understanding of how microbial communities are assembled in rivers.

Microbial coalescence, intersecting with dispersal, is attracting increasing attention recently. In previous studies, microcosms have been widely employed to explore the extent of microbial coalescence, aiming to better understand how microbial communities come together. A previous study showed that a single community dominated the function of the coalesced community for methanogenic archaea (6). In addition to microcosms, the proportion of overlapped species was considered as an indicator to quantify the extent of coalescence (7). The recently introduced microbial source tracking method has been utilized to quantify the influx of species from source communities to sink communities. However, previous studies have primarily focused on theoretical concepts, interpretations of microcosm experiments, or mathematical algorithms for quantification. There remains a significant knowledge gap concerning the coalescence patterns of microbial communities in natural ecosystems (e.g., rivers), which have yet to be comprehensively investigated.

Microbial coalescence patterns could affect the assembly processes of microbial communities (8). In river ecosystems, microbial communities are deterministically influenced by surrounding environmental conditions (e.g., temperature, dissolved oxygen, pH, and nutrients) (9, 10) and biotic factors (e.g., competition, mutualism, and predation) (2). In addition, stochastic processes (e.g., birth, death, speciation, limited dispersal, and immigration) can also shape microbial communities (2). Various studies have revealed the relative importances of deterministic and stochastic processes in river ecosystems using different ecological models. For example, stochastic processes were found to mainly shape the microeukaryotic communities using the neutral community model (NCM) (11). Meanwhile, the normalized stochasticity ratio (NST) models revealed that stochastic and deterministic processes both played important roles in shaping NC10 bacterial communities (12). In contrast, the relative contributions of deterministic and stochastic processes in shaping phytoplankton and bacterioplankton communities were found to vary seasonally using a null model (13). In addition to these models, the newly developed phylogenetic-bin-based null model (iCAMP) was discovered to enhance the precision of quantifying the contributions of different ecological processes to microbial community assembly (14). Notably, rivers in the subtropical region of China undergo distinct wet–normal–dry cycles on an annual basis (9), resulting in dynamic hydrological conditions. Unlike bacterioplankton communities, sedimentary microbial communities are more stable and not as easily affected by hydrologic disturbance, while their assembly processes under dynamic hydrological conditions were largely unknown.

The Taihu Basin, located in the subtropical monsoon climate region of China, is selected as the study area, and it is characterized by a pronounced wet–normal–dry cycle annually. There are at least 15 major rivers in this region, and these rivers are interconnected, possessing one of the densest plain river networks that facilitate regular exchanges of microorganisms (15). Therefore, this area provides an ideal site for investigating coalescence patterns and assembly processes of microbial communities in river ecosystems. In this study, we analyzed the sedimentary bacterial communities in this area across different hydrological seasons (wet, normal, and dry seasons). The main aims were to separately address the following: (i) Do the sedimentary bacterial communities undergo seasonal dynamics? (ii) Do the coalescence patterns between sedimentary bacterial communities and bacterioplankton communities exhibit seasonal variations? (iii) Are the sedimentary bacterial communities assembled via similar processes across different seasons? The results of this study will provide new understandings of the dynamics of microbial communities in river ecosystems and shed light on the underlying mechanisms.

MATERIALS AND METHODS

Sample collection and physiochemical analyses

A total of 18 sites from four crisscrossing rivers in the Taihu Basin were selected for sampling during the wet, normal, and dry seasons (Fig. 1). A detailed description of the sampling area is available in our previous studies (9, 16). At each sampling site, the upper layers (~10 cm) of the sediments were collected by a Petersen gravity sediment sampler, and debris (stones, metals, etc.) was immediately removed onsite. These sediment samples were then carefully placed into sterile plastic bags and transported to the laboratory within 4 hours on ice. Upon arrival at the laboratory, the samples were separated into two parts: one for physiochemical analyses (stored at 4°C) and the other for microbial analyses (stored at −20°C). Due to the unforeseen drought, we were unable to obtain enough sediment samples from some sampling sites during the normal and dry seasons (Table S1), and a total of 47 samples were obtained in this study.

Fig 1.

Overview of sampling sites in the Taihu Basin.

The KCl solution (2 mol/L) was utilized to extract ammonia nitrogen (NH₄⁺-N), nitrite nitrogen (NO₂⁻-N), and nitrate nitrogen (NO₃⁻-N). The extracted nitrogen compounds were quantified colorimetrically using a UV-visible spectrophotometer (Hach, USA). Ultra-pure water was used to extract the dissolved organic carbon and then measured using a total organic carbon analyzer (Elementar, Germany) (17). The pH of the samples was determined using a pH meter (Mettler Toledo, Switzerland), with a solid-to-ultra-pure-water ratio of 1:2.5 (18). The physicochemical characteristics of each sample are detailed in Table S2.

DNA extraction, PCR amplification, and amplicon sequencing

Extraction of microbial DNA was performed using the FastDNA SPIN Kit for Soil (MP Biomedicals, USA). A universal primer, 515F/806R, was used to amplify the V4 region of the bacterial 16S rRNA gene (19). Thermal cycling was performed with an initial degeneration step for 5 min at 94°C, followed by 30 cycles of denaturation for 0.5 min at 94°C, annealing for 1 min at 60°C, elongation for 1 min at 72°C, and a final elongation for 10 min at 72°C. The PCR products were purified, combined in equal amounts, and subjected to sequencing at Novogene (Beijing, China). The raw sequencing data have been deposited in the Genome Sequence Archive of the National Genomics Data Center under the accession number CRA006672 (https://ngdc.cncb.ac.cn/gsa).

The DADA2 pipeline (version 1.16) (20) in R software (version 4.1.0) was employed for processing the raw sequencing data. All necessary steps, including trimming of low-quality reads, merging of paired-end reads, and removal of chimeras, were carried out following the tutorial (http://benjjneb.github.io/dada2/tutorial.html). Subsequently, an amplicon sequence variant (ASV) table was generated, and a representative sequence from each ASV was picked out for taxonomic annotation against the Ribosomal Database Project (RDP) (80% confidence level) (21). ASVs assigned to chloroplasts, archaea, eukaryotes, and unknown kingdoms were excluded from further analyses. Additionally, to mitigate potential PCR bias, only the ASVs with more than 10 sequences were retained, and the sequence number in each sample was ultimately rarefied to 47,954 for subsequent analyses.

Data analysis

Unless otherwise specified, all of the subsequent analyses were conducted using R software (version 4.1.0).

The dynamics for α-diversity (Shannon diversity index) and β-diversity (“Bray–Curtis” similarity) of bacterial communities across different seasons were revealed by one-way ANOVA followed by Tukey’s post hoc test. The dynamics of bacterial community structure were illustrated by principal coordinates analysis (PCoA) based on the “Bray–Curtis” distance. The degree of separation of PCoA was tested by the analysis of similarities (ANOSIM). The Shannon diversity index, “Bray–Curtis” distance, PCoA plot, and ANOSIM were performed by the “vegan” package (version 2.5-6). To identify significant biomarker taxa across different seasons, the linear discriminant analysis effect size (LEfSe) method was employed followed by the Kruskal–Wallis test (22). The geographic distance between each pairwise sampling site was calculated by the “geosphere” package (version 1.5-10) using the latitude and longitude data (Table S1). The slope of the linear regression between the “Bray–Curtis” distance and geographic distance was used to calculate the distance–decay relationship.

The redundancy analysis (RDA) and Mantel test were performed to reveal the impacts of environmental attributes on the bacterial community structure. The individual contribution of each environmental attribute to the total variation was determined by hierarchical partitioning analysis using the “rdacca.hp” package (version 1.0.1) (23). The environmental preferences of typical ASVs (top 20 abundant) were analyzed based on the Spearman correlation coefficients, and the importance of each environmental attribute was predicted by the random forest model using the “linkET” package (version 0.0.3.3).

Network analysis using Spearman’s correlation coefficient was constructed to reveal the co-occurrence patterns of the ASVs by the “microeco” package (version 0.10.1) (24). To simplify the network and improve visualization, the ASVs that occurred in less than 50% of the samples were removed. The correlation was considered to be statistically significant with |correlation coefficient| (|r|) >0.7 and P < 0.01. The keystone species in the network were identified according to within-module connectivity (Zi) and among-module connectivity (Pi). Network hubs (Zi >0.25, Pi >0.62), module hubs (Zi >0.25, Pi ≤0.62), and connectors (Zi ≤0.25, Pi >0.62) were recognized as the keystone species (25). The effects of keystone species on module eigengenes were predicted by the random forest model using the “linkET” package (version 0.0.3.3).

The NCM was applied to determine the contribution of stochastic processes on bacterial community assembly by the “MicEco” package (version 1.2-1). Then, the stochasticity ratios were calculated by the NST model (26) and iCAMP (14) using the “NST” package (version 3.0.6) and “iCAMP” package (version 1.3.4), respectively. By using iCAMP, ASVs were clustered into phylogenetic-closed bins, and then, the respective contributions of deterministic processes [homogeneous selection (HoS), heterogeneous selection (HeS)] and stochastic processes [dispersal limitation (DL), homogenizing dispersal (HD), and drift (DR)] were quantified based on the beta net relatedness index (βNRI) and taxonomic β-diversities using the modified Raup–Crick metric (RC). The βNRI values less than −1.96 indicated HoS, whereas values higher than 1.96 indicated HeS. For the βNRI values between −1.96 and 1.96, those with RC values higher than 0.95 represented DL, values less than −0.95 represented HD, and the remaining portions represented DR (14).

To explore the coalescence patterns of the bacterial communities, the raw sequencing data of bacterioplankton communities in water samples were retrieved from our previous study (9). The sediment samples and water samples were collected at the same time. The water samples were filtered if they were not paired with the sediment samples, and vice versa. A total of 90 samples were retained for the following analyses. The raw sequencing data were processed as described above (see DNA extraction, PCR amplification, and amplicon sequencing), and the sequence number was rarefied to 40,641 per sample. The percentage of overlapped ASVs between pairwise neighboring (upstream–downstream) sampling sites (Table S3) was calculated. Furthermore, the contribution of upstream sources to downstream sinks was predicted by SourceTracker (version 1.0.1) (27) with four replications.

RESULTS

Dynamics of sedimentary bacterial communities

In this study, 47 sediment samples were collected encompassing the wet, normal, and dry seasons. A total of 2,908,230 effective sequences were obtained by amplicon sequencing, and they were clustered into 14,460 ASVs after eliminating low abundance ASVs (total sequence ≤10) and non-bacterial ASVs. The number of sequences per sample ranged from 47,954 to 81,459, and prior to downstream analyses, the sequence number was rarified to 47,954 for each sample, retaining 2,253,838 effective sequences. To assign taxonomic information, a representative sequence from each ASV was compared against the RDP database. Proteobacteria was found to be the most abundant phylum (51.87%) followed by Bacteroidetes (12.81%), Firmicutes (6.23%), and Chloroflexi (6.01%) (Fig. S1).

PCoA was performed to reveal the dynamics of bacterial communities (Fig. 2a). The results demonstrated evident clustering of samples from the same season with different seasons visibly separated, revealing that the structure of bacterial communities exhibited seasonal dynamics (ANOSIM, P = 0.001). Similarly, the α-diversity (Shannon diversity index) and β-diversity (“Bray–Curtis” similarity) of the bacterial communities were also found to vary significantly across different seasons (P < 0.001) (Fig. 2b and c). Additionally, the LEfSe analysis unveiled the seasonal dynamics in the taxonomic compositions of the bacterial communities (Fig. S2). Taken together, these results indicated that the diversity, structure, and taxonomic composition of the bacterial communities exhibited temporal dynamics across different seasons. Meanwhile, the similarity of the bacterial communities was found to decrease significantly with increasing distance between sampling sites during the normal and dry seasons (P < 0.05) (Fig. S3), showing that the bacterial communities exhibited spatial variation with distance–decay patterns. Notably, the distance–decay patterns were not observed in the wet season (P > 0.05). These results showed that the spatial patterns of bacterial communities were distinct during different seasons.

Fig 2.

Temporal dynamics of bacterial community structure and diversity. (a) PCoA plot based on the “Bray–Curtis” distance showing the variation of bacterial community structure. Ellipses denoted 95% confidence intervals. (b and c) Box plots showing the variations of α-diversity (Shannon diversity) and β-diversity (“Bray–Curtis” similarity). Different letters in (b) and (c) indicated statistical significance by Tukey’s post hoc test (P < 0.05).

To further reveal the environmental drivers for the spatiotemporal dynamics of the bacterial communities, RDA followed by hierarchical partitioning analysis was performed. As shown in Fig. 3a and b, pH was found to be the most important attribute that affected the bacterial communities, contributing to >60% of variations. Similarly, the Mantel test also showed that pH was significantly correlated with the bacterial communities (r > 0.2, P < 0.001) (Fig. S4). Furthermore, the typical ASVs (top 20 abundant) were selected, and their correlations with environmental attributes are shown in Fig. 3c. Most of these ASVs (90%) were significantly correlated with pH, and the importance of pH was >5% as predicted by the random forest model.

Fig 3.

Relationship between environmental attributes and bacterial communities. (a) RDA plot showing the effects of environmental attributes on bacterial communities. (b) The individual effect of each environmental attribute on the bacterial communities by hierarchical partitioning analysis. (c) Phylogenetic analysis and environmental response of typical ASVs. Circle size represented the importance of each environmental attribute as predicted by the random forest model. Abbreviation: DOC, dissolved organic carbon.

Network analysis and keystone species of sedimentary bacterial communities

To reveal the co-occurrence patterns of bacterial communities, a network was constructed based on the relationships of ASVs (Fig. 4a). This network contained a total of 142 nodes and 1,510 edges. Most of the edges (94.4%) were found to be positive correlations, and among them, the largest proportion of connections (35.8%) was observed within the Proteobacteria phylum themselves (Fig. 4b). Likewise, within the negative connections, Proteobacteria was highly connected among themselves (36.9%) (Fig. 4b). These results indicated that Proteobacteria was the dominant phylum in this network.

Fig 4.

Co-occurrence patterns of the bacterial communities. (a) Network showing the co-occurrence patterns. Each node was colored according to modularity, and its size was proportional to the number of connections. (b) Chord plot of summed positive and negative connections among bacterial phyla. (c) Keystone species identified according to the roles of nodes in terms of within-module and among-module connectivities. (d) Correlation between keystone species and module eigengenes. The importance of keystone species (shown as circle size) and their explained variances (shown as the histogram) to module eigengenes were predicted by the random forest model. Abbreviation: M1–M5, modules 1–5.

Based on the modularity, this network was divided into five major modules with the largest two modules accounting for 80.4% of total nodes (Fig. 4a). The keystone species were identified according to the connectivity within and between modules. Two ASVs (ASV37 and ASV53) and three ASVs (ASV421, ASV478, and ASV486) were identified as the module hubs and connectors, respectively (Fig. 4c). These ASVs explained ~60% of the variances to the eigengenes of the three largest modules (Fig. 4d), further confirming that these ASVs were key drivers in this network. All of the keystone species were clarified into Proteobacteria at the phylum level, but none of them were annotated at the family level (Table S4), showing that their taxonomy annotations were incomplete.

Coalescence patterns between sedimentary bacterial communities and bacterioplankton communities

To reveal the coalescence patterns between the sedimentary bacterial communities and bacterioplankton communities, adjacent sampling sites from upstream to downstream were selected (Table S3). The upstream sites were regarded as the sources, and the downstream sites were the destinations of coalescence for the bacterial communities. As shown in Fig. 5a, the highest percentage of overlapped ASVs were found between neighboring bacterioplankton communities (wet season: 43.6%, normal season: 24.4%, and dry season: 36.0%). In addition, a higher proportion of overlapped ASVs between neighboring sediment sites was observed compared with that between water and its neighboring sediment sites.

Fig 5.

Coalescence patterns of the bacterial communities. (a) Percentage of overlapped ASVs between pairwise adjacent (upstream–downstream) sampling sites. (b) Different sources of the bacterial communities quantified by SourceTracker.

To quantify the contributions of upstream sources to downstream sinks, SourceTracker analysis was utilized (Fig. 5b). When the sink was defined as the bacterioplankton communities, 55%–83% of the sources were estimated to be from the bacterioplankton communities while sedimentary bacterial communities made a minimal contribution (6%–13%). Similarly, when the sink was defined as the sedimentary bacterial communities, a greater proportion was attributed to the sedimentary bacterial communities from upstream sites compared to the bacterioplankton communities.

Assembly processes of sedimentary bacterial communities

To reveal the assembly processes of the sedimentary bacterial communities, several ecological models were applied. Firstly, the NCM fitted well for the bacterial communities and effectively estimated a significant portion of the association between the occurrence of ASVs and their relative abundances for all seasons (Fig. S5). The estimated immigration rate was lower in the wet season than in other seasons, indicating the occurrence of less ecological drift in this season. Furthermore, the relative contributions of the stochastic process to bacterial community assembly were quantified by NST and iCAMP, respectively. The NST model revealed a higher stochasticity ratio in the wet season (76.57%) than in other seasons (<50%) (Fig. 6a). Similarly, based on iCAMP, the stochasticity ratio was on average 71.87% and 75.41% in the normal and dry seasons, respectively, which was lower than that in the wet season (81.87%) (Fig. 6b). Together, these results indicated that the bacterial communities were more stochastic in the wet season than in other seasons.

Fig 6.

Assembly process of the bacterial communities. (a–b) Stochasticity ratio estimated by NST and iCAMP. (c) The relative importance of different ecological processes based on iCAMP. (d) Relative abundance of the top eight bins (left) and the ASV with the highest relative abundance in each bin (right). (e) The relative importance of HoS, DL, and DR in the assembly processes of top eight bins.

Additionally, the relative importance of different ecological processes (HeS, HoS, DL, HD, and DR) was quantified using iCAMP. In this study, HoS, DL, and DR played more important roles in shaping the bacterial communities than HeS and HD, with an average importance of 17.72%–27.97%, 17.48%–33.18%, and 44.24%–51.28%, respectively (Fig. 6c). The relative importance of HoS and DR decreased slightly in the wet season while DL increased greatly in this season, suggesting that the importance of stochastic processes was increased in this season (Fig. 6c). The iCAMP model could provide additional insights into the relative significance of various processes within each phylogenetic bin. For this purpose, the 14,460 ASVs were grouped into 281 bins. DR dominated 216 (76.87% of bin numbers) and 173 (61.92% of bin numbers) bins in the normal and dry seasons, respectively. By contrast, only 99 bins (35.23% of bin numbers) were shaped by DR in the wet season, and DL was the dominant process for most of the bins (166 bins, 59.07% of bin numbers) in this season. The top eight major bins were selected (Fig. 6d), and the contributions of HoS, DL, and DR in shaping these bins were further analyzed (Fig. 6e). The bin75 was the most abundant bin (6.61% in total abundance), and it was mainly shaped by DR. The contribution of DR to this bin decreased from 73.21% in the dry season and 74.34% in the normal season to 60.84% in the wet season. DL posed relatively important roles in shaping the abundant genera, including Acinetobacter in bin75, Flavobacterium in bin7, and Sporosarcina in bin212, by 17.11%, 46.33%, and 32.11%, respectively, in the wet season while having entirely no effects on the assembly of these bins in the other seasons.

DISCUSSION

Dynamics of sedimentary bacterial communities during different hydrological seasons

Recently, there has been increasing attention in microbial ecology toward understanding the dynamics of the bacterial communities in aquatic environments (28). This study aimed to enhance our understanding of the dynamics of the sedimentary bacterial communities induced by hydrologic disturbance within the river ecosystem. Consistent with our expectations, there were evident temporal variations in the diversity (α-diversity, β-diversity) and taxonomic composition of the sedimentary bacterial communities (Fig. 2b and c; Fig. S2). Furthermore, the PCoA plot revealed temporal variations in the bacterial community structure, with the bacterial communities from the same season clustering together (Fig. 2a). These findings aligned with previous studies that reported similar observations (9, 10). Several reasons may account for these findings. Firstly, the distinct hydrologic conditions during the wet, normal, and dry seasons resulted in variations in environmental conditions, thereby influencing the diversity and structure of the sedimentary bacterial communities. In this study, significant impacts of the environmental attributes on bacterial communities were observed (Fig. 3a) with pH being the most important attribute (Fig. 3b; Fig. S4), which was consistent with previous studies (29, 30). Additionally, the interconnectivity of the rivers in this study facilitated the exchange of microorganisms, contributing to the seasonal clustering observed in the PCoA ordinations.

Furthermore, spatial variation was also found to influence the sedimentary bacterial communities. A notable distance–decay pattern was observed during the normal and dry seasons, while this pattern was not observed during the wet season (Fig. S3). This finding is reasonable considering that rainfall predominantly occurs during the wet season in this region, leading to higher river flow that promotes the dispersal event than oterh seasons. As a result, the community similarity during the wet season was higher compared to the normal and dry seasons (Fig. 2c). A similar phenomenon was also observed in the profile of methane-oxidizing bacteria in this region, where distance–decay patterns were evident during the normal and dry seasons, but not during the wet season (12).

Keystone species are believed to be important in maintaining bacterial communities (31). In this study, network analysis was used to identify the keystone species (Fig. 4), including two module hubs (ASV37 and ASV53) and three connectors (ASV421, ASV478, and ASV486) (Fig. 4c). All of them were annotated into Proteobacteria phylum (Table S4); meanwhile, Proteobacteria had the largest proportion of connections (35.8%) with other Proteobacteria within both positive and negative connections (Fig. 4b), indicating that Proteobacteria played crucial roles in maintaining the bacterial communities. However, only two keystone ASVs were annotated at the order level, and none were annotated at the family level (Table S4). Considering the lack of taxonomic information, these keystone species might be considered as the “dark matter” (32), and their roles in the river ecosystems need further investigation.

Similar coalescence patterns between sedimentary bacterial communities and bacterioplankton communities during different hydrological seasons

Along the longitudinal direction, the coalescence patterns of sedimentary bacterial communities and bacterioplankton communities were quantified using the proportion of overlapped ASVs and SourceTracker analysis (Fig. 5). The results showed that the bacterial communities from the same habitat type showed more proportion of overlapped ASVs in all of the hydrological seasons. These findings indicate that the coalescence of bacterial communities is more prominent within the same medium (intra-medium) compared to that between different media (inter-medium). This observation was consistent with a previous study focusing on the microbial communities in Three Gorges Reservoir (33). This phenomenon may be caused by the differences in bacterial migration rates in different habitats. A previous study confirmed this speculation by showing that the migration rates of bacterial communities in sediments were much lower than those in water (34). However, only sediment and water samples were involved in this study and more types of habitats (e.g., surrounding soils) are needed in the future to further confirm this speculation.

Different proportion of stochasticity in bacterial community assembly during different hydrological seasons

The assembly of bacterial communities is typically influenced by a combination of selection, drift, speciation, and dispersal processes (8). While selection is primarily driven by deterministic processes and drift is influenced by stochastic processes, speciation and dispersal can involve both deterministic and stochastic processes (8). However, considering the limited duration of this study, which only encompassed three hydrological seasons, it may be challenging to fully resolve the speciation process. Therefore, it is expected that the bacterial communities are shaped by a combination of selection (deterministic), drift (stochastic), and dispersal (both deterministic and stochastic) processes in this study.

The NCM fitted well for the sedimentary bacterial communities and explained 66.5%–73.6% of bacterial patterns in all of the seasons (Fig. S5). This implied that stochastic processes were crucial in shaping bacterial communities, which was consistent with the observations of bacterioplankton communities in this region (9) and the microeukaryotic communities in the Tingjiang River (11). The NST model and iCAMP model further revealed that the relative contributions of stochastic processes on bacterial community assembly differed among the three seasons (Fig. 6a and b). Increased stochasticity in bacterial community assembly was observed in the wet season as compared with the normal and dry seasons, which was consistent with a previous study (11). This might be caused by the increased rainfall and enhanced dispersal event in the wet season, which was confirmed by the insignificant distance–decay patterns of the bacterial communities in this season (Fig. S3).

Furthermore, the iCAMP model showed that HoS, DL, and DR were the major processes in shaping the bacterial communities, which were similar in typical riparian wetlands (35). The relative importance of HoS decreased in the wet season as compared with the other seasons (Fig. 6c). The wet season is characterized by higher water levels, faster hydrological regimens, and more fluidity than other seasons (36), which could decrease the effects of HoS. DL was observed to be enhanced in the wet season (Fig. 6c), and it shaped the major bins only in this season with no effects for other seasons (Fig. 6e). This was supported by the fact that DL occurred more easily in the wet season together with precipitation (37), resulting in higher community similarity (Fig. 2c). DR was observed to be the most important process in shaping the sedimentary bacterial communities, and it was enhanced in the dry season. In this season, droughts were observed in some sampling sites, resulting in failed sample collection (Table S1). This constrained the overall habitat availability, potentially serving as a natural environmental filter (38), enhancing the effects of drift and environmental selection. Notably, wet–normal–dry cycling is an annually repeated event, and as such, a long-time investigation is needed to further confirm these implications.

Conclusion

In this study, the dynamics of the sedimentary bacterial communities in a plain river network (Taihu Basin) were analyzed by amplicon sequencing of the 16S rRNA gene. Significant temporal variation was observed for the diversity, structure, and taxonomic composition of the sedimentary bacterial communities across the wet, normal, and dry seasons (P < 0.01). The coalescence patterns between the sedimentary bacterial communities and bacterioplankton communities were analyzed, and the coalescence of bacterial communities was more prominent within the same medium (intra-medium) compared to that between different media (inter-medium). Furthermore, several ecological models were applied to reveal the underlying assembly processes of the sedimentary bacterial communities, and increased stochasticity was observed in the wet season compared with the normal and dry seasons. These results implied that similar coalescence patterns between the sedimentary bacterial communities and bacterioplankton communities were established via distinct assembly processes under dynamic hydrological conditions.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/aem.01465-23.

Supplemental file 1. aem.01465-23-s0001.pdf.

Fig. S1 to S5 and Tables S1 to S4.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.