Particle-size dependent of bacterial diversity associated with suspended particulate matter continuum in Lake Taihu

Abstract

Suspended particulate matter (SPM) of varying particle sizes is widespread in aquatic ecosystems, providing crucial habitats for bacteria and serving as hotspots for mineralization and nutrient cycling. However, prior studies have typically treated bacteria associated with these particulates as a homogeneous group, overlooking size-related differences in diversity and composition. In this study, we separated the SPM continuum into five size-fractions (0.2, 2, 20, 200, and 500 µm) and investigated bacterial diversity, community assembly, and environmental drivers across four representative regions of Lake Taihu, China, over 1-year period. Using 16S rRNA gene sequencing, we observed particle-size-dependent variations in bacterial diversity. Alpha diversity decreased significantly with increasing particle size, while beta diversity showed a similar trend. Environmental factors influencing species richness varied by particle size, while bacteria associated with smaller particles (0.2, 2, and 20 µm) were more sensitive to environmental factors compared to those associated with larger ones (200 and 500 µm). The role of deterministic processes in community assembly increased with particle size, indicating stronger selection on larger particles. This study enhances our understanding of bacterial diversity in aquatic ecosystems and highlights the importance of particle size in bacterial community dynamics.

This study shows that bacterial diversity in suspended particulate matter of Lake Taihu varies with particle size, with smaller particles exhibiting greater diversity and broader environmental influences than larger ones.

Introduction

Understanding the diversity of bacteria associated with suspended particulate matter (SPM) in lake ecosystems is crucial for deciphering the material cycles within these systems. However, comparative studies on bacterial diversity along the continuum of lake SPM remain relatively scarce. Based on lifestyle, aquatic bacteria are typically categorized as free-living bacteria (FLB) and particle-attached bacteria (PAB). FLB are primarily suspended in the water column, leading a planktonic lifestyle, whereas PAB adhere to various particles such as aquatic plant debris and microplastics (Helene et al.2022, Chen et al. 2025), exhibiting either obligate or facultative attachment (Azam et al. 1983, Crump et al. 1999, Simon et al. 2002, Lamontagne and Holden 2003, Yu et al. 2022, Le et al. 2024). SPM is ubiquitous in aquatic environments, providing a substrate for bacterial attachment and serving as key habits for mineralization and nutrient cycling (Yu et al. 2001, Cotner and Biddanda 2002, Azam and Malfatti 2007, Mohit et al. 2014). Compared to FLB, PAB are larger in size (Hu 2013, Liu 2018, Tang 2018) and exhibit higher diversity (Wu et al. 2012, Tang et al. 2017, Guo et al. 2019), enabling them to extend their activity range and enhance environmental adaptability (Grossart 2010). PAB possess specialized abilities to adhere to and degrade organic compounds on particles, granting them unique advantages in microbial community growth, metabolism, interactions, and evolution (Ghiglione et al. 2009, Crespo et al. 2013). Consequently, they play significant roles in biogeochemical cycles (Crespo et al. 2013, Tang et al. 2014). Thus, understanding the diversity of PAB is key to elucidating material cycles in these systems. In existing studies, bacteria in aquatic systems are often coarsely categorized by lifestyle into FLB and PAB. Filtration using polycarbonate membranes with pore sizes of 0.2 and 5 µm (or 3 µm) is commonly employed: bacteria associated with the particulate matter with diameter between 0.2 and 5 µm (or 3 µm) are classified as FLB, while those with larger than 5 µm (or 3 µm) are designated as PAB (Crump et al. 1999, Simon et al. 2002, Lamontagne and Holden 2003, Liu et al. 2019b, Yu et al. 2022, Le et al. 2024). However, studies on the bacterial diversity associated with SPM of varying sizes (i.e. the particulate continuum) are relatively limited.

The diversity and community structure of bacteria associated with SPM in lakes may vary with particle size and origin. In deep and eutrophic lakes, phytoplankton is the primary source of SPM, with sizes ranging from 1 µm to 10 cm, while particles smaller than 1 µm are classified as colloidal particles. In ecosystems dominated by macrophytes or large algae, such as lakes and coastal areas, leaves or detritus of these primary producers contribute to SPM, which can reach sizes of several tens of centimeters (Dang and Lovell 2016). Field observations show a significant negative correlation between particle abundance and size; smaller particles are abundant and prone to aggregation, while larger particles are relatively independent but less stable (Grossart and Simon 1993, 1998a, b, Grossart 2001). These characteristics likely affect the bacterial diversity and composition on particles. Additionally, internal differences in the microenvironment of particles of varying sizes and types may impose selective pressures on bacterial communities (Simon et al. 2002). Mestre et al. (2017a) investigated SPM-associated bacterial diversity using six pore sizes (0.2–200 µm) in the northwestern Mediterranean and found that bacterial richness increased with particle size. Subsequently, Mestre et al. 2017b, 2020) conducted a similar study across four depths and distances from shore, confirming that larger particles harbored greater bacterial diversity. Conversely, Cram et al. (2024), in their study on microbial diversity in Chesapeake Bay, observed a decrease in bacterial diversity and abundance with increasing particle size, where they used six particle size ranges (0.2–1.2 µm, 1.2–5 µm, 5–20 µm, 20–53 µm, 53–180 µm, and 180–500 µm). In shallow lakes, such as Lake Taihu, wind-driven sediment resuspension and reaggregation may further specialize PAB communities (Li et al. 2020, Wang et al. 2012), potentially leading to more pronounced differences among PAB communities associated with particles of different sizes. However, comparative studies on bacterial diversity associated with the SPM continuum in lakes are lacking. Though the traditional dichotomy between FLB and PAB has been widely used in microbial ecology research, it oversimplifies the ecological roles of bacteria, as bacterial populations in natural environments often exhibit complex transitional behaviors, as well as FLB and PAB can, to some extent, be interconverted (Riemann and Winding 2001, Grossart 2010, Tang et al. 2017, Wang et al. 2023). In contrast, the study of bacteria associated with the SPM continuum provides a more nuanced understanding by considering the particle size gradation. This approach may reveal the distinct impacts of different particle sizes on bacterial community diversity, composition, and responses to environmental factors, thus uncovering more specific differences. Research on bacteria within the SPM continuum may advance the understanding of the interactions between bacterial communities and particulate matter in the environment, offering a more comprehensive ecological perspective than the traditional dichotomy. This could be particularly innovative in terrestrial freshwater lakes.

This study conducted a field investigation in Lake Taihu, a large shallow lake in China. Four representative lake regions with distinct hydrological characteristics and potential SPM sources were selected: the central lake area (CL), prone to wind-induced resuspension (Qin et al. 2003, Wu et al. 2013, Yao et al. 2022); Dapukou (DP), a river mouth with high external SPM inputs due to dense population (Xu and Wu 2013); East Taihu (EL), dominated by macrophytes (Zhang et al. 2004, Zhu et al. 2015); and Meiliang Bay (ML), characterized by frequent cyanobacterial blooms (Chen et al. 2003, Ma et al. 2016, Shi et al. 2018). Monthly monitoring was conducted over 1 year, collecting bacterial samples from the SPM continuum. Five particle size ranges were selected, spanning from micrometer to millimeter scales: 0.2–2 µm, 2–20 µm, 20–200 µm, 200–500 µm, and >500 µm. Two key questions were addressed: (i) Does bacterial diversity on the SPM continuum in Lake Taihu differ with particle size? (ii) Do the mechanisms shaping bacterial community diversity on the SPM continuum differ with particle size?

Materials and methods

Sampling locations and timing

Lake Taihu is located in the Yangtze River Delta in eastern China (30° 55′ 40″ – 31° 32′ 58″ N, 119° 52′ 32″ – 120° 36′ 10″ E), with an average depth of 2 m and a total area of 2 338 km², making it the third-largest freshwater lake in China. Over recent decades, increased industrialization and population growth have led to heightened pollutant input, exacerbating the lake’s eutrophication (Chen et al. 2003). The region experiences a temperate monsoon climate, characterized by hot, rainy summers and cold, dry winters.

Based on hydrological characteristics and potential differences in particulate matter sources, four regions of Lake Taihu were selected for sampling: the CL region, DP, eastern Lake Taihu (EL), and ML. In each region, three sampling sites were chosen, resulting in a total of 12 sampling locations (Fig. 1). Monthly sample collection was conducted over 1 year, from September 2021 to September 2022.

Figure 1.

The sampling location map. ML is cyanobacterial bloom zone, Dapu Port (DP) is external SPM inputs area, the central region (CL) is wind-induced resuspension area, and East Taihu Lake (EL) is macrophyte-dominated area.

Water sample collection and environmental data

At each sampling site, 5 l of lake water was collected from a depth of 0.5 m below the surface using a 5-l plexiglass water sampler and placed into sterile water bags. The samples were transported to the laboratory in an insulated container with ice for further processing. On-site measurements of water temperature (T), pH, and dissolved oxygen (DO) at a depth of 0.5 m were conducted using a calibrated YSI portable water quality analyzer (YSI 6600, Yellow Springs, Ohio). Other physicochemical parameters were analyzed in the laboratory. In the laboratory, water samples were filtered through preweighed and precombusted (450°C, 4 h) glass fiber filters (GF/F, Whatman). The dry weight of total suspended solids (SS) was determined by drying one GF/F at 110°C for 1 h, and the difference in filter weight before and after filtration was divided by the filtered water volume. The second GF/F was used to extract chlorophyll a (Chla) with hot acetone as the solvent, and its concentration was determined spectrophotometrically (Jespersen and Christoffersen 1987). Filtered water (100 ml) was processed following the standardized protocol for ecological surveys of Chinese lakes. Total dissolved phosphorus (TDP), orthophosphate (PO₄³⁻), total dissolved nitrogen (TDN), ammonium nitrogen (NH₄⁺), nitrite (NO₂⁻), and nitrate (NO₃⁻) concentrations were measured using a continuous flow analyzer (Skalar SAN PLUS system; Skalar Analytical BV, Breda, the Netherlands). Additionally, 50 ml of unfiltered water was used to determine total phosphorus (TP) and total nitrogen (TN) concentrations. Dissolved organic carbon (DOC) and dissolved inorganic carbon (DIC) were both measured using a TOC analyzer. First, DOC in the water samples was oxidized to carbon dioxide (CO₂) by high-temperature combustion, and the concentration of CO₂ was then detected using infrared (IR) analysis to obtain the DOC concentration. Subsequently, DIC in the samples was converted to CO₂ gas by adding acid (such as HCl), and the concentration of CO₂ was detected using IR analysis to determine the DIC concentration.

Sample processing

This study is based on the six particle size ranges (0.2–0.8 µm, 0.8–3.0 µm, 3.0–5.0 µm, 5.0–10 µm, 10–20 µm, and 20–200 µm) originally used by Mestre et al. (2017a) in their bacterial study in the Mediterranean, as well as the six particle size ranges (0.2–1.2 µm, 1.2–5 µm, 5–20 µm, 20–53 µm, 53–180 µm, and 180–500 µm) later used by Cram et al. (2024) in a similar study in Chesapeake Bay. Additionally, considering the particle size range in Lake Taihu, we made adjustments and ultimately selected five particle size ranges (0.2–2 µm, 2–20 µm, 20–200 µm, 200–500 µm, and >500 µm) for the study (He et al. 2022, Chen et al. 2025).

At each sampling site, 1 l of lake water was sequentially filtered through sieves with pore sizes of 500, 200, and 20 µm, followed by polycarbonate membranes with pore sizes of 2 and 0.2 µm. Particulates retained on the sieves were rinsed with ultrapure water and enriched on 0.2 µm polycarbonate membranes. The obtained membranes were stored in cryovials at −80°C until DNA extraction. Bacteria retained on the 500, 200, and 20 µm sieves and the 2 µm polycarbonate membrane represented PAB in size ranges >500 µm, 200–500 µm, 20–200 µm, and 2–20 µm, respectively. Bacteria retained on the 0.2-µm polycarbonate membrane during initial filtration (excluding eluates) represented FLB (0.2–2 µm). For simplicity, these size ranges are hereafter referred to by their lower limits: 500, 200, 20, 2, and 0.2 µm.

DNA extraction was performed using a modified phenol–chloroform extraction and ethanol precipitation method (Li et al. 2011). Extracted DNA was amplified using the primers 515F (5′-CTGCCAGCMGCCGCGG-3′) and 806R (5′-GGACTACHVGGGTWTCTAAT-3′) targeting the V4 region of the 16S rRNA gene (Apprill et al. 2015). Polymerase Chain Reaction (PCR) amplifications were conducted using the Gene Amp PCR-System® 9700 (Applied Biosystems, Foster City, CA, USA). Each sample was amplified in triplicate to minimize random errors during PCR. The triplicate PCR products were pooled and analyzed for purity by agarose gel electrophoresis, purified using the QIAquick Gel Extraction Kit (Qiagen, Germantown, MD, USA), and further quantified with PicoGreen (BMG Labtech, Jena, Germany). Amplicons were spiked with PhiX (Illumina, San Diego, CA, USA) and sequenced on an Illumina MiSeq platform for 16S rRNA gene high-throughput sequencing (Li et al. 2017).

Raw sequencing data were processed using TrimGalore v0.6.10 to remove primers and adapter sequences (Krueger et al. 2024). Quality trimming was performed using Fastx v0.014 (http://hannonlab.cshl.edu/fastx_toolkit/commandline.html) to remove bases with a quality score below Q15, followed by paired-end merging with FLASH2 (Magoč and Salzberg 2011). USEARCH v7.1 was employed to discard sequences with >2 mismatched primer bases, lengths <100 bp, or overall error rates >2% (Edgar 2010). DADA2 was then used to generate amplicon sequence variants (ASVs) (Callahan et al. 2016). Taxonomic annotations for ASV representative sequences were assigned using the Silva138.1. Bacterial ASVs were filtered to exclude singletons, chloroplast, and mitochondrial ASVs, and data were rarefied to the minimum sequencing depth (9164 reads). Using the ASV table, bacterial ASV richness and Bray–Curtis distances were calculated with the vegan package in R (version 4.3.2). Phylogenetic diversity was computed using phylogenetic tree files (Scherson and Faith 2018).

Statistical analysis

Environmental factors with high multicollinearity (vif-value >10) were excluded using the vif function in the car package. Temporal and spatial variations in bacterial alpha diversity across particle sizes were analyzed with one-way ANOVA and Tukey’s post hoc tests. Relationships between particle size and alpha diversity were examined using linear regression analysis. The Kruskal–Wallis test was used to compare Bray–Curtis distances among particle size groups across temporal and spatial scales. Linear discriminant analysis Effect Size (LEfSe) was applied to identify taxa significantly enriched in different particle size groups, revealing key microbial taxa and quantifying their differences (Segata et al. 2011, Lv et al. 2023, Tian et al. 2023). The cooccurrence network graph for particle-size-specific microbial interactions was generated using the igraph package in R. Multivariate linear regression was employed to test correlations between species richness and environmental factors, while Mantel tests examined the relationship between environmental parameters and bacterial community structure. Average variation degree (AVD) was calculated to characterize community stability, and a null-model approach was used to infer community assembly mechanisms. Plots were generated using the ggplot2 package in R and Origin 2021.

Results

Physicochemical properties of Lake Taihu water

Overall, the physicochemical properties of Lake Taihu water demonstrated seasonal variation patterns. Although not identical, these variations can be generally classified into the following categories (Fig. S1A): summer peak: DIC, NO₂⁻, PO₄³⁻, and TDP showed higher values during summer (June–August), with maxima of 10.1 mg/l, 0.044 mg/l, 0.033 mg/l, and 0.04 mg/l, respectively, while lower values were observed in other months. Autumn peak: DOC, NH₄⁺, PN, TN, TP, and pH peaked in autumn (September–November), with values of 9.65 mg/l, 0.18 mg/l, 3.83 mg/l, 4.32 mg/l, 0.54 mg/l, and 9.15 mg/l, respectively. Winter peak: DO reached a high value of 13.7 mg/l during winter (December–February) before declining in other months. Monotonic changes: DON and T showed a monotonic decrease, while NO₃⁻ and TDN exhibited a monotonic increase. Fluctuations: SS displayed significant variability without a clear seasonal trend.

Regional variations in physicochemical properties were also evident among the four lake regions (Fig. S1B). T, DIC, and DO were relatively consistent across the regions. DOC was higher in DP but lower in EL. DON, NH₄⁺, NO₂⁻, NO₃⁻, PO₄³⁻, TDN, and TDP reached their highest levels at DP, with values of 0.984 mg/l, 0.17 mg/l, 0.049 mg/l, 0.894 mg/l, 0.034 mg/l, 1.49 mg/l, and 0.05 mg/l, respectively, while lower values were observed at the other sites. These trends contrasted with those of DO. pH and PN were highest in ML (2.07 mg/l), with significant variability. SS showed notable differences between DP and ML (Kruskal–Wallis test, P < .05). TN and TP were highest in ML (2.71 mg/l and 0.30 mg/l, respectively) with substantial fluctuations, while EL exhibited significantly lower values than the other regions (Kruskal–Wallis test, P < .05, Fig. S1B).

From Fig. S1(C), it can be observed that TDN is highly correlated with NO₃⁻, TDP is highly correlated with PO₄³⁻, and there is a strong correlation (correlation >0.85) among the four environmental factors: TN, TP, PN, and Chla (Fig. S1C). Due to the strong correlations among these environmental factors themselves, it is possible that these factors may be excluded in subsequent analyses to eliminate multicollinearity.

Alpha diversity of bacteria associated with the SPM continuum

High-throughput sequencing was conducted on 510 bacterial samples associated with the SPM continuum. After quality control (removal of low-quality sequences, single-occurrence ASVs, and chloroplast or mitochondrial ASVs), 90 693 high-quality sequences with an average length of 250 bp were obtained. The dataset was rarefied to the minimum sequence count (9 164 sequences), yielding 125 440 bacterial ASVs.

The results revealed a size gradient-driven pattern in the spatiotemporal variation of alpha diversity (species richness and phylogenetic diversity) of bacteria associated with the SPM continuum in Lake Taihu (Fig. 2). Overall analysis of all samples demonstrated a linear decrease in bacterial species richness (Richness) and phylogenetic diversity with increasing particle size (both P < .01; Fig. 2). This linear decrease in alpha diversity with increasing particle size was consistent across different seasons (Fig. S2A–D and S1I–L; all P < .001) and lake regions (Fig. S2E–H and S1M–P; all P < .001). The trend was particularly pronounced in autumn and in the CL region. Additionally, the decline in the EL lake area is the least pronounced, and no bacterial samples associated with particles larger than 500 µm were collected in this lake area (Fig. S2G and O). Overall, the α-diversity of SPM continuum-associated bacteria is highest in the EL lake area, while no significant differences were observed among the other three lake areas. Specifically, for the 0.2 µm group, DP and ML had the highest diversity, while CL had the lowest. For the 20 and 200 µm groups, EL showed the highest diversity, with no significant differences among the other three lake areas, consistent with the overall trend. For the 2 and 500 µm groups, no significant differences were found across the four lake areas (Tables S2, S3, S6, and S7).

Figure 2.

The linear regression relationship between the α diversity of bacteria associated with SPM size continuum and particle size, with particle size expressed as the logarithm to the base 10. (A) The linear regression of overall species richness against particle size. (B) Illustrates the linear regression of overall phylogenetic diversity against particle size.

Beta diversity of bacteria associated with the SPM continuum

The community composition of bacteria associated with the SPM continuum exhibited particle-size dependence. Across the 510 samples analyzed, bacteria belonging to 28 phyla, 78 classes, 142 orders, 241 families, 371 genera, and 115 species were identified. Actinobacteriota dominated the small particle-size groups (0.2 and 2 µm), accounting for more than 60% of the relative abundance (Fig. 3A). Among Actinobacteriota, Sporichthyaceae and Ilumatobacteraceae were the major families, and their relative abundance decreased significantly with increasing particle size (Fig. 3B). In the medium-to-large particle-size groups (20, 200, and 500 µm), Cyanobacteria were the dominant phylum, comprising nearly 50% of the relative abundance. Within Cyanobacteria, Microcystaceae and Nostocaceae were the major families, although they were nearly absent in the small particle-size groups (0.2 and 2 µm). Compared to other lake areas, the proportion of Cyanobacteria in the bacterial community associated with 20 and 200 µm SPM in the EL lake area is relatively lower. Additionally, Bacteroidota consistently ranked among the top 10 most abundant phyla across all particle-size groups, with its relative abundance increasing with particle size (Fig. 3A). Within Bacteroidota, the relative abundance of Microscillaceae increased significantly with particle size, while Chitinophagaceae showed a significant decrease (Fig. 3B). Overall, the correlations between 0.2 and 2 µm are closer, while the correlations between 20, 200, and 500 µm are closer (Fig. 3D). This pattern of community composition was observed consistently across different seasons and lake regions (Fig. S3).

Figure 3.

The bacterial community composition associated with the SPM continuum. (A) Displays an overview of bacteria associated with different particle size continua at the phylum classification level. (B) Displays an overview of bacteria associated with different particle size continua at the family classification level. (C) The LEfSe analysis of the community composition of bacteria associated with different particle size continua. (D) Cooccurrence network plots for particle-size-specific microbial interactions.

LEfSe analysis of bacteria associated with different particle-size groups revealed that Actinobacteriota taxa were significantly enriched in small particle-size groups (0.2 and 2 µm), while Cyanobacteria taxa dominated the medium-to-large particle-size groups (20, 200, and 500 µm). In the 0.2-µm group, taxa such as Actinobacteria (class), Frankiales (order), Sporichthyaceae (family), and the hgcl clade (genus) exhibited high LDA scores, indicating significant enrichment in this group (Fig. 3C). In the 2-µm group, Acidimicrobiia (class), Microtrichales (order), Ilumatobacteraceae (family), and the CL500-29 marine group (genus) were enriched. Additionally, Verrucomicrobiota (phylum) and Verrucomicrobiae (class), as well as Chitinophagales (order) within Bacteroidota, were also enriched in the 2-µm group (Fig. 3C). In the 20-µm group, enriched taxa primarily included Cyanobacteria such as Nostocaceae (family), Dolichospermum NIES41 (genus), Dolichospermum circinale (species), Phormidiaceae (family), and Planktothrix NIVA-CYA 15 (genus) (Fig. 3C). In the 200-µm group, Cyanobacteria (phylum), Cyanobacteriia (class), Cyanobacteriales (order), and Microcystis aeruginosa (species) were predominantly enriched (Fig. 3C). In the 500-µm group, a diverse range of taxa were enriched, including Bacteroidota (phylum), Bacteroidia (class), and Cyanobacteria-associated taxa such as Microcystaceae (family), Cytophagales (order), and Microcystis PCC-7914 (genus). Additionally, Gammaproteobacteria taxa, including Pseudanabaenales (order), Pseudanabaenaceae (family), and Pseudanabaena PCC-7429 (genus), were significantly enriched in the 500-µm group (Fig. 3C). These findings indicate that different particle sizes harbor distinct bacterial communities, demonstrating size-specific assembly of bacteria associated with the SPM continuum in Lake Taihu.

Overall analysis revealed significant differences in the community composition of bacteria associated with the SPM continuum along the first two principal coordinates in PCoA analysis. Along the PC1 (18.79%) and PC2 (7.01%) axes, the separation indicated that bacterial communities associated with particles of different sizes exhibited significant differences in community structure on these principal coordinates. Smaller particles clustered in the positive direction of PC1, while larger particles clustered in the negative direction, reflecting differentiated community structures among particles of varying sizes. Further verification using PERMANOVA for beta diversity confirmed significant differences between groups (P < .001, Table S8), indicating that the bacterial community composition of the SPM continuum in Lake Taihu is size-specific. Seasonal and regional analyses showed that smaller particle groups (0.2 and 2 µm) and larger particle groups (20, 200, and 500 µm) consistently clustered together in PCoA plots (Fig. S4A–H). This consistency across seasons and regions suggested that the size-specific bacterial community composition is a general pattern regardless of temporal or spatial context. Additionally, compared to other lake areas, the distribution of sample points for 0.2 and 2 µm SPM-associated bacteria in the DP area is more scattered.

Linear regression analysis of Bray–Curtis distances against sampling intervals revealed that the similarity of bacterial communities associated with the SPM continuum decreased with increasing temporal differences (P all < .001), and the rate of decrease was faster for larger particle sizes (Fig. 4B). This finding indicates that the beta diversity of bacteria associated with the SPM continuum exhibited a consistent temporal decay trend, with the rate of decay being particle-size-dependent.

Figure 4.

The β diversity of bacterial communities associated with the SPM continuum. (A) Represents the results of Principal Coordinate Analysis (PCoA) analysis based on Bray–curtis similarity. (B) Represents the trend of β diversity based on BC distance with changes in sampling time intervals.

Community stability and assembly mechanisms of bacteria

The average variance in species richness (AVD) of bacteria associated with the SPM continuum demonstrated size-specific. Overall, analysis of the coefficient of variation for alpha diversity showed that AVD values decreased from small to large particle sizes, indicating increased community stabilization (Fig. 5A). Seasonal and regional analyses also revealed similar trends, with AVD values decreasing with increasing particle size (Fig. S5).

Figure 5.

Community construction analysis for bacterial communities associated with SPM continuum. (A) Presents bar chart of the AVD values across different particle sizes. (B) Presents null model of community assembly mechanisms.

Community assembly mechanisms of bacteria associated with the SPM continuum were also size-specific. Null-model analysis of all samples showed that the relative importance of deterministic processes in the assembly of particle-attached bacterial communities increased with particle size (0.2 µm: 58.17%; 2 µm: 55.76%; 20 µm: 68.83%; 200 µm: 72.55%; and 500 µm: 73.51%). Homogeneous selection was the dominant deterministic factor (0.2 µm: 55.13%; 2 µm: 52.1%; 20 µm: 65.12%; 200 µm: 68.34%; and 500 µm: 66.75%). Conversely, the relative importance of stochastic processes decreased with particle size, with dispersal limitation being the most prominent stochastic factor (0.2 µm: 29.63%; 2 µm: 32.27%; 20 µm: 20.57%; 200 µm: 17.13%; and 500 µm: 16.35%) (Fig. 5B). These results indicate that small particle sizes (0.2 and 2 µm) were more influenced by stochastic factors, whereas medium and large particle sizes (20, 200, and 500 µm) were more influenced by deterministic processes. Larger particles likely provided more diverse ecological niches, enhancing selective processes. This trend is consistent across the seasonal and lake-specific null model analyses, except for the EL lake area, where the bacterial communities associated with the SPM continuum exhibit the poorest consistency with changes in particle size, and the proportion of deterministic processes is lower than that of stochastic processes compared to other lake areas (Fig. S6), This suggests that the assembly of bacterial communities associated with the SPM continuum is particle-size-dependent under varying temporal and spatial conditions, except in the lake area dominated by macrophytes.

Environmental factors influencing bacterial communities

The relationship between species richness of bacteria associated with the SPM continuum and environmental factors varied with particle size. Multivariate linear regression analysis showed that smaller and medium particle sizes (0.2, 2, and 20 µm) were influenced by more environmental factors than larger particle sizes (200 and 500 µm). Among these, 20 µm particles were affected by the largest number of environmental factors. Specifically, NH₄⁺ and NO₂⁻ were significantly positively correlated with the 0.2 µm group, while DIC was negatively correlated. DIC and SS were negatively correlated with the 2 µm group. NH₄⁺ was positively correlated with the 20 µm group, whereas DO, DOC, DON, and NO₂⁻ were negatively correlated. TDN was positively correlated with the 200 µm group, while DON and NO₃⁻ were negatively correlated. DO was significantly negatively correlated with the 500 µm group (Fig. 6A).

Figure 6.

Environmental factors affecting bacterial species richness and composition. (A) Shows results of multiple linear regression analysis of the relationship between environmental factors and the richness of various organisms. (B) Shows results of the partial mantel analysis.

The species composition of bacterial communities associated with the planktonic SPM continuum exhibited particle-size-gradient-driven responses to environmental factors. Partial Mantel analysis revealed that the environmental factors influencing community composition varied across particle sizes: For the 0.2-µm group: pH, TDP, NO₂⁻, DIC, and PO₄³⁻ were significant factors. For the 2-µm group: pH, NO₂⁻ and Chla were significant factors. For the 20-µm group: pH, TDP, NO₂⁻, PO₄³⁻, and Chla were significant factors. For the 200-µm group: pH, TDP, NO₂⁻, PO₄³⁻, DOC, DIC, and Chla were significant factors. For the 500-µm group: pH, NO₂⁻ was the primary influencing factor. Mid-to-large particle sizes (20 and 200 µm) were influenced by a greater number of environmental factors, with the 200-µm group being affected by the largest number. Among all factors, pH and NO₂⁻ exhibited significant effects across all particle sizes, highlighting their restrictive roles in shaping community composition. The influence of NO₂⁻ was particularly pronounced in the 20-µm group, while pH had the most significant impact in the 200-µm group (Fig. 6B).

Discussion

This study focuses on Lake Taihu, a typical large, shallow lake in China, and involves monthly monitoring over a 1-year period. The objective is to explore the gradient variation in bacterial community diversity associated with SPM continuum. Previous studies on bacterial communities in lake water have primarily classified bacteria into FLB (particle size <5 or 3 µm) and PAB (particle size >5 or 3 µm) (Crump et al. 1999, Simon et al. 2002, Lamontagne and Holden 2003, Liu et al. 2019b, Yu et al. 2022, Le et al. 2024). However, this study investigates the SPM continuum, ranging from micrometer- to millimeter-scale particles (0.2, 2, 20, 200, and 500 µm), which provides a finer gradient of SPM continuum than typically studied in lake ecosystems but aligns with gradients used in marine ecosystem research (Mestre et al. 2017a, b, 2020, Cram et al. 2024, Borrull et al. 2024). Furthermore, this research examines variations in bacterial diversity across different lake zones within Taihu, considering differences in hydrological features and SPM sources, to derive general patterns of bacterial diversity associated with SPM continuum.

Particle-size dependent of SPM-associated bacterial alpha diversity in Lake Taihu

Previous studies have reported contrasting trends in bacterial α-diversity, with some suggesting that PAB (>5 or 3 µm) exhibit higher α-diversity than FLB (<5 or 3 µm) (Li et al. 2011, Tang et al. 2017), while others report the opposite (Acinas et al. 1999, Ghiglione et al. 2007). Our study shows a significant decrease in α-diversity (species richness and phylogenetic diversity) of SPM-associated bacteria with increasing particle size (Fig. 2), consistent with findings in soil particle matter studies by Hemkemeyer et al. (2018), where gene numbers decreased significantly with particle size increments (P < .001). The discrepancy between these previous studies in lakes and our results may attributable to finer size classification of PAB, compared to traditional studies aggregating bacteria from different size ranges. Compared to larger particles, smaller particles exhibit greater abundance and resource heterogeneity (Dougherty et al. 2025), while also providing a larger specific surface area that enables associated bacteria to access enhanced nutrient and oxygen availability (Weber and Bianchi 2020). Additionally, smaller particles have longer retention times, providing bacteria with more time to adapt (Alldredge and Gotschalk 1988, Kiørboe et al. 2003, DeVries et al. 2014). On the other hand, larger particles have relatively limited surface area and are fewer, restricting bacterial diversity. These varied findings in lake habitats suggest that a detailed classification of particles by size for studying attached bacteria can reveal new trends in bacterial diversity compared to the coarse categorization of FLB and PAB (Mestre et al. 2020).

Furthermore, the result of α-diversity of SPM continuum-associated bacteria in Lake Taihu in our study is different to that found in the Mediterranean, where the species richness of bacteria associated SPM continuum (six particle size ranges: 0.2–0.8, 0.8–3.0, 3.0–5.0, 5.0–10, 10–20, and 20–200 µm) was highest on 0.2–0.8 µm particles and lowest on 20−200 µm particles, showing an increasing pattern with particle size increasing (Mestre et al. 2017a, b). This may be mainly attributed to the differences in environmental conditions between lakes and oceans. The nutrient levels in the eutrophic, shallow Lake Taihu are far higher than those in the marine ecosystem studied by Mestre et al. (2017a). In eutrophic shallow lakes, the relatively high content and variety of dissolved organic matter (DOM) and nutrients support diverse bacterial communities, with particles selectively hosting specific bacterial types, whereas marine ecosystems, characterized by low DOM and nutrient content, support fewer bacterial species, with more bacteria, particularly heterotrophic ones, favoring a particle-attached lifestyle (Hecky 1988, Matthew et al. 2008, Repeta et al. 2002). These differences in environmental factors, which have already been shown to drive variations trends in α-diversity of SPM continuum-associated bacteria across different marine regions (Mestre et al. 2017a, b, Cram et al. 2024), between lakes and marine waters are likely to result in different trends in α-diversity of SPM continuum-associated bacteria in these two habitats. In addition, in the studies conducted by Mestre et al. 2017a, b), the average species richness of bacterial prokaryotes was reported to be 179.8. In contrast, our findings reveal a significantly higher average of 3978.3, indicating the pronounced disparity in species richness between lacustrine and marine ecosystems which may leads the differences in trends of bacterial α-diversity. Previous studies have shown that there is a transformative relationship between FLB and PAB, with PAB being an important source of FLB counterparts (Riemann and Winding 2001, Tang et al. 2017, Liu et al. 2019b). Aquatic bacterial communities tend to prefer a planktonic lifestyle in nutrient-rich conditions and an attached lifestyle in nutrient-poor conditions (Dang and Lovell 2016). Microcystis blooms can stimulate the transition of certain rare bacteria from a planktonic to a particle-attached lifestyle (Liu et al. 2019b). Thus the species richness of bacteria associated SPM continuum decreases with the increasing of particle size in Lake Taihu, which is different to that found in the Mediterranean.

Notably, in autumn, differences in species richness among particle size groups were most pronounced. Significant differences (P < .001) were observed between almost all size groups except for the pairs 2 µm versus 0.2 µm and 500 µm versus 200 µm (Table S4). This may be attributed to the end of summer algal blooms, during which the disintegration of cyanobacterial cells and the accumulation of algal debris occur. The particle sizes of these debris vary, with most cyanobacterial fragments ranging from 0.2 to 2 µm, unicellular algae from 2 to 20 µm, and cyanobacterial aggregates exceeding 200 µm. By autumn, these accumulated particles cause particle concentration differences, leading to diversity differences among size groups (Qin et al. 2018, Zhang et al. 2022). Li et al. (2011) also observed increased heterotrophic bacterial richness during cyanobacterial blooms. In contrast, the decline in the EL lake area is the least pronounced, and no bacterial samples associated with particles larger than 500 µm were collected in this lake area (Fig. S2G and S2O). This highlights the relatively lower amount of SPM in lake areas dominated by large submerged macrophytes compared to other lake areas (Li et al. 2021).

Particle-size dependent of SPM-associated bacterial beta diversity in Lake Taihu

Our monthly 1-year monitoring of SPM-associated bacterial diversity in different lake zones enabled analysis of community similarity decay patterns over time. Bacterial community similarity decreased with increasing temporal differences, with the rate of decrease escalating with particle size (Fig. 4B). This aligns with Guo et al.’s (2022) findings, where larger organisms exhibited greater β-diversity changes over time. Similar trends were observed in monthly sampling of bacterial communities in surface waters of San Pedro Basin (Hatosy et al. 2013).

Our study revealed significant differences in the community composition of bacteria associated with SPM continuum in Lake Taihu. Bacteria on the smallest size (0.2 µm) SPM are predominantly dominated by Actinobacteria, whereas the dominant phyla among bacteria associated with the other parts of SPM continuum (2, 20, 200, and 500 µm) included Cyanobacteria, Bacteroidetes, Verrucomicrobia, and Planctomycetes (Fig. 3A). These results align with most studies focusing on free-living and particle-attached bacterial communities (Allgaier and Grossart 2006, Zhang et al. 2007, Xie et al. 2024). Mestre et al. (2017b) observed that bacterial groups enriched on smaller particles (0.2–0.8 µm and 0.8–3.0 µm) included Verrucomicrobia, Gammaproteobacteria, Deferribacteres, Alphaproteobacteria, and Chloroflexi. In contrast, Cram et al. (2024) reported that most bacteria were associated with smaller particles (0.2–1.2 µm and 1.2–5 µm). However, in our study, the bacterial groups enriched in the similar particle size range (0.2–2 µm) primarily belonged to Actinobacteria. Conversely, Mestre et al. (2017b) found that bacterial groups enriched on larger particles (10–20 µm and 20–200 µm) were mostly from Planctomycetes, Deltaproteobacteria, Verrucomicrobia, and Bacteroidetes. In the study by Cram et al. (2024), bacteria enriched on particles ≥20 µm were predominantly from Alphaproteobacteria, Gammaproteobacteria, and Firmicutes. In our study, however, the bacterial groups enriched in the comparable particle size range (20–200 μm) were primarily from Cyanobacteria. This difference may be attributed to the contrasting characteristics between lake and marine environments (Allgaier and Grossart 2006, Zhang et al. 2007, Xie et al. 2024). Firstly, in terms of particulate matter concentration, the average SS concentration in the Mediterranean sampling area reported by Mestre et al. 2017a, b) was only 9.67 mg/l (Sebastiá and Rodilla 2013), whereas the average SS concentration in Lake Taihu in the present study was 73.42 mg/l, seven times higher than the former. Regarding phytoplankton (e.g. cyanobacteria), the average chlorophyll-a concentration in the Mediterranean study area was 0.014 mg/l, with most areas around 0.001 mg/l (Sebastiá and Rodilla 2013), while in this study, the average chlorophyll-a concentration in Lake Taihu was 0.061 mg/l, reaching as high as 0.089 mg/l in the ML region, with the majority of this chlorophyll-a derived from cyanobacteria. In terms of salinity, the salinity in the Mediterranean is more than 100 times higher than that of Lake Taihu (Sebastiá and Rodilla 2013). In addition, Actinobacteria were predominantly found in smaller particle size groups (0.2 and 2 µm) but did not maintain this dominance in larger size groups (20, 200, and 500 µm; Fig. 3A and B). Unicellular actinobacteria, such as Micrococcus and Coccus, typically have cell diameters of 0.2−5 µm. Filamentous actinobacteria, including Streptomyces and Actinomyces, form filamentous mycelia with cell diameters also generally between 0.2 and 5 µm. Some actinobacteria can also form spores, such as Streptomyces spores and Actinomyces spores, with spore diameters typically in the range of 0.2−5 µm. Thus, Actinobacteria are almost exclusively concentrated in the 0.2−5 µm range (Portillo et al. 2013). Cyanobacteria were nearly absent in the 0.2- and 2-µm groups but showed a sharp increase in the 20-µm group. Similarly, Mestre et al. (2017a) found that the smaller group Synechococcus sp. (with an average size of 1 µm) under the phylum Cyanobacteria predominantly enriched on particles in the 0.8–3 μm size range, and was nearly absent in the smallest particle size range (0.2–0.8 µm), which is similar to the findings in our study. Unicellular cyanobacteria generally range from 1 to 10 µm in size. For instance, species of the genus Microcystis are typically 2−10 µm in diameter (Portillo et al. 2013), while Spirulina species range from 3 to 8 µm. Filamentous cyanobacteria, on the other hand, have broader size ranges, with lengths reaching several millimeters and diameters generally in the range of a few to tens of micrometers. For example, the genus Anabaena typically has diameters of 4−10 µm, while Nostoc species range from 5 to 15 µm (Mehdizadeh Allaf and Peerhossaini 2022). Notably, Colonial Cyanobacterial, such as Microcystis, typically form aggregates or colonies, with particle sizes ranging from tens of micrometers to several millimeters (Wu et al. 2019, Whitton and Potts 2000). Furthermore, cyanobacteria often attach to organic and inorganic particles in the water, such as sediments and suspended particulates, which helps stabilize their growth in the water column, preventing them from being displaced by currents or wave action and maintaining their presence in favorable growth zones (Ibrahim et al. 2022). This phenomenon is particularly pronounced during cyanobacterial blooms, especially when the water body is subject to wind-induced waves and currents. Cyanobacteria attached to larger particles or in larger aggregates are more likely to remain afloat under hydrodynamic disturbances (Li et al. 2017), thereby promoting bloom formation and initiating a vicious cycle of cyanobacterial proliferation (Wu et al. 2019). Due to these traits, such as attachment to particles and the formation of algal aggregates, cyanobacteria dominate the 20, 200, and 500 µm particle size groups. In summary, 0.2 and 2 µm group are more similar, while 20, 200, and 500 µm group are more similar to each other (Figs 3D and 4A).

In addition, the distribution of sample points for 0.2 and 2 µm SPM-associated bacteria in the DP lake area is more scattered (Fig. S4F). As DP is an area where exogenous particulate matter converges, the composition of particles in this size range may be more complex, leading to greater variability in the bacterial community on these particles (Chen et al. 2019).

Mechanisms and influencing factors of SPM-associated bacterial community assembly in Lake Taihu

This study showed that the proportion of deterministic processes in bacterial community assembly increases with particle size (Fig. 6B). One possible explanation is that larger particles exert stronger selective pressure on bacteria, resulting in higher determinism for bacterial communities associated with these particles. Larger particles provide additional carbon sources and nutrients, which may favor specific bacterial taxa (López-Pérez et al. 2016). Conversely, for bacteria associated with smaller SPM, dispersal limitation—a stochastic process—plays a more significant role due to the diffusion of bacteria with water flow (Zhu et al. 2022). Whether in the DP lake area with higher input of exogenous particulate matter, the CL lake area where wind-driven resuspension of particles occurs, or the ML lake area with frequent cyanobacterial blooms leading to higher organic particulate matter, the community assembly mechanisms of bacteria associated with the SPM continuum are similar. Only in the EL lake area, dominated by large submerged macrophytes, is a different community assembly mechanism observed. This may also represent a distinction between eutrophic and noneutrophic areas (Fig. S6G). According to the FLB and PAB bacterial transformation theory, aquatic bacterial communities tend to adopt a planktonic lifestyle under nutrient-rich conditions and an attached lifestyle under nutrient-poor conditions (Dang and Lovell 2016). Larger particles represent more nutrients, and with the increase in eutrophication levels, the proportion of deterministic processes also increases (Qin et al. 2020, Kuang et al. 2022). Therefore, as particle size increases, the proportion of deterministic processes becomes greater. In contrast, East Taihu exhibits a higher proportion of stochastic processes, with no clear trend in relation to particle size. Studies have confirmed that during the growth of large aquatic plants, the influence of stochastic processes significantly increases (He et al. 2020). As East Taihu is dominated by large aquatic plants, the influence of deterministic processes is weakened.

Environmental factors influencing species richness and composition differ significantly among particle size groups. Our results showed that bacterial richness in smaller particle groups (0.2 and 2 µm) exhibited significant correlations with multiple environmental factors, whereas fewer environmental factors were significantly correlated with bacteria associated with the larger SPM continuum (Fig. 6A). As particle size increases, the number of associated environmental factors decreases, suggesting that bacteria associated with larger particles are better able to resist changes in environmental conditions (Liu et al. 2019a). The larger the particle size, the less sensitive the SPM-associated bacteria are to environmental variability, as these bacteria may be shielded from environmental fluctuations by the particles (Yung et al. 2016). For example, compared to smaller particle sizes (0.2, 2, 20, and 200 µm), bacteria associated with 500 µm particles significantly correlated only with DO, and the relationship was negative. As a result of the bacterial taxa that are significantly enriched in the 500-µm group include Bacteroidota (phylum) and its subgroups, which typically grow under anaerobic conditions (Glaring et al. 2015). Large particles have a smaller surface area, consume oxygen rapidly, and are prone to forming anoxic microzones on their surface. Additionally, large particles can create anaerobic environments within their interior (Chowdhury et al. 2020, Hu et al. 2020, Cram et al. 2024). Cram et al. (2024) suggested that DO is not correlated with bacterial species richness, which may be due to differences between marine and lake environments. On the other hand, Liu et al. (2019a) argued that PAB have a greater requirement for DO than FLB, which could be attributed to differences in particle size gradation. The study by Cram et al. (2024) also found that carbon content is proportional to particle mass. Algal detritus is broken down into large particles by bacterial communities associated with the particle size continuum. These larger particles provide a rich habitat for bacteria. Since larger particles have a higher sinking rate, they are more likely to enter the sediments, forming a biological carbon pump and enhancing carbon sequestration (Mestre et al. 2018, He et al. 2019, Liu et al. 2019b, Michael et al. 2022). Moreover, Cram et al. (2024) found that both the α-diversity and β-diversity of SPM continuum-associated bacteria were correlated with salinity in Chesapeake Bay. However, in our study, among all environmental factors, nitrite nitrogen had a significant influence on all particle sizes (Fig. 6B). This suggests that nitrogen, particularly nitrite nitrogen, is a common limiting factor for bacterial community composition in freshwater lakes (Yang et al. 2024), and also indicates that the environmental factors most influential to bacterial communities differ between marine and lake environments (Rojas-Jimenez et al. 2019).

Conclusion

This study demonstrates that bacterial diversity and community assembly in Lake Taihu are strongly influenced by particle size. As particle size increases, both α-diversity and β-diversity decrease, with deterministic processes becoming more prominent in larger particles. This suggests that larger particles exert stronger selective pressures on bacterial communities, likely through enhanced resource partitioning and environmental filtering. These findings highlight the importance of particle size in shaping microbial communities and emphasize the need to consider particle size as a key factor in microbial ecology studies.

Author contributions

Fangwei Fan (Conceptualization, Formal analysis, Methodology, Visualization, Writing –original draft), Yichen Ren (Conceptualization, Formal analysis, Methodology), Zhendu Mao (Formal analysis, Methodology), Biao Li (Investigation, Methodology), Chunyan Yu (Conceptualization, Methodology, Resources), Jiawei Gao (Investigation, Methodology), Yu Gu (Visualization), Jianing Ding (Visualization), Huabing Li (Conceptualization, Methodology, Supervision, Writing –review & editing), and Qinglong L. Wu (Conceptualization, Methodology, Supervision, Writing –review & editing).

Conflict of interest

The authors declare no conflict of interest.

Funding

This work was supported through the Project of National Key Basic Research and Development (2023YFF1304501), the National Science Foundation of China (32171535 and 32471626), and the Science and Technology Planning Project of Nanjing Institute of Geography and Limnology, Academy of Sciences (NIGLAS) (NIGLAS2022GS02).

Data availability

Raw sequencing data can be found in the public database NODE (https://www.biosino.org/node), and the raw data registration number is OEP00006124. The data underlying this article will be shared on reasonable request.