Response mechanisms of sediment microbial communities in different habitat types in a shallow lake

Abstract

Microorganisms are important drivers of material and energy circulation in freshwater ecosystems. Altered aquatic sedimentary environments disturbed by human activities affect microbial community composition. Concentrations of heavy metals, and sediment physicochemical properties, are described for 25 sediment samples collected from five habitat types (residential, duck farm, lotus pond, fish farm, and a reference reserve shallow freshwater lake site) around Baiyangdian Lake, China. Bacterial and fungal communities in sediments were determined using 16S rRNA and 18S rRNA high‐throughput sequencing technology. Sediment physicochemical properties and heavy metal concentrations vary significantly among habitat types. Bacterial genera involved in S and N cycles are most abundant, followed by those that degrade organic pollutants, and are tolerant of heavy metals. Sediment bacterial Chao1 index is predominantly regulated by cobalt concentrations. The microbial communities within replicate samples within habitats are more similar to each other than they are between habitats. Redundancy analysis indicates Zn, pH, and TP are the main environment factors structuring bacterial communities, while fungal communities are most significantly influenced by Co and TP. The genus Cryptomonas is a food source for small zooplankton and fishes, and is mainly distributed in fish farm and lotus pond habitats. Species of Pseudallescheria that cause mycetoma, maduromycosis, and other infections in humans are highly abundant in duck farm and residential habitats. Species of Aleuria are saprobes; decompose organic matter, dead organisms, or biological excrement; and are highly accumulated in duck farm habitat.

Introduction

Freshwater ecosystems provide many services, such as water supply, food sources, and for recreational activities (Postel and Carpenter , Zedler ). These ecosystems are, however, vulnerable to disturbance from human activities, such as from agriculture, aquaculture, and tourism. Sediments act as repositories of pollutants and nutritional elements, and the water–soil interface is important for physical, chemical, and biological reactions (Webster and Ridgway ). Sediment microorganisms are widely distributed, diverse, and have various metabolic functions; they are highly efficient in enzymatic activity (Gerbersdorf et al. ). These microorganisms play important roles in organic matter decomposition, nutrient cycling, and pollutant degradation (Crump et al. , Pires et al. , Bai et al. , Oni et al. ). As human disturbance can alter the sedimentary environment, it can also influence microbial community composition. Accordingly, identifying the effects of environmental variables on microbial communities improves understanding of the influence of human activities on freshwater ecosystems.

Sediment properties and nutrients greatly influence microbial community composition. Anthropogenic disturbance altered bacterial communities, especially key functional groups, in two mangrove wetlands at Hainan Island, China (Yun et al. ). Municipal and industrial waste decreased sediment bacterial community Chao1 and Shannon indices in Lagos Lagoon, Nigeria, with Cd and polycyclic aromatic hydrocarbons (PAH), and physicochemical properties such as sediment oxygen demand (SOD), chemical oxygen demand (COD), total organic matter (TOM), nitrate, and pH, influencing microbial community composition (Obi et al. ). Concentrations of Cr and Cd were major factors related to soil bacterial community structural changes in Dongting Lake, China (Zhang et al. ). These studies reveal human activities can fundamentally alter sediment quality and nutrient cycling, and shape the distributions of microbial communities.

Baiyangdian Lake, the largest shallow lake in North China Plain, has various ecological functions (Hu et al. ). Increased industrial and human activity in the surrounding catchment over the last three decades has resulted in water level drops and eutrophication (Liu et al. ). Water and sediment pollution situation such as physicochemical properties (Yan et al. ), heavy metal concentrations (Su et al. ), PAH (Guo et al. ), antibiotics (Cheng et al. ), and organochlorine compounds (Hu et al. , Guo et al. ) of this lake are now well described, and the composition and diversity of vegetation, zooplankton and phytoplankton (Xu et al. ), and benthic animals (Yi et al. ) are also well known. However, only the ammonia‐oxidizing archaea and bacteria of this lake are reported (Zheng et al. ). Improved lake management requires the composition of microbial communities to be better understood.

Sediment samples from different habitats in different districts Baiyangdian Lake were collected for heavy metal concentration, physiochemical property, and microbial community composition and diversity determination. Relationships between these sediments, environmental factors, and microbial communities were analyzed to identify sediment microbial community responses to different kinds of human activities.

Materials and Methods

Study area

Baiyangdian Lake (38°43–39°02′ N, 115°45–116°07′ E), in Hebei province, has a total surface area of 362.8 km² (Liu et al. ). It is crisscrossed by canals and ditches, reed fields, farmlands, and villages are dispersed throughout and around its water area. Anthropogenic disturbance includes input from Fu River, a sewage channel from Baoding city, and the planting of lotus, fish and duck farming, and tourism (Fig. ). Specific information for sampling sites is provided in Appendix S1: Table S1.

Distribution of sampling sites (n = 25), Baiyangdian Lake. (a) Map of China; (b) Baiyangdian Basin; (c) sampling sites, Baiyangdian Lake. Habitats: lotus pond (LP), fish farm (FF), duck farm (DF), residential area (RA), reserve reference (TR).

Habitats within three of 39 (Yi et al. ) villages, Xiaozhangzhuangcun (XZZC), Zhainan (ZN), and Duancun (DC), each strongly disturbed by human activities, were selected for further study. Accumulation of domestic garbage within the water and sediments was serious in XZZC; DC was one of the largest villages, with a greater population than either other village; and ZN, in the middle of Baiyangdian Lake, was chosen because it was situated far from XZZC and DC. A large duck farm was located at Caiputai (CPT), in the southern Baiyangdian Lake. Representative fish farms were located in Quantou (QT) and Shaochedian (SCD), along the southern and northern edges of Baiyangdian Lake, and lotus ponds were located in Bolikou (BLK) and Wangjiazhai (WJZ). A reserve in Baohuqu (BHQ) was sampled to provide reference values of a near‐natural environment.

Sediment sample collection

Sediment samples within five habitat types (lotus pond, fish farm, duck farm, residential area, and reserve reference) were collected within Baiyangdian Lake in late November 2017 (Table ).

Sample points setting for each habitat type

Habitat type	Samples code	No. of parallel samples
Residential area	XZZC1, XZZC2, XZZC3; ZN1, ZN2, ZN3; DC1, DC2, DC3	Three parallel sample points for each village
Lotus pond	BLK1, BLK2; WJZ1, WJZ2	Two parallel points for each lotus pond
Fish farm	SCD1, SCD2; QT1, QT2, QT3, QT4	Two parallel points for SCD and four for QT
Duck farm	CPT1, CPT2, CPT3	Three parallel points for duck farm
Reference	BHQ1, BHQ2, BHQ3	Three parallel points for reserved area

In total, 25 sediment samples were collected from 3 to 5 m depth with a Pedersen grab sampler. The top 3–5 cm of each sample was transferred into a sterilized bottle. Samples were placed into a cooler box and transported to the laboratory, after which they were split into two subsamples: one stored at −80°C for high‐throughput sequencing; the other air‐dried and processed through a 2‐mm sieve to remove stone and root fragments, then stored at 4°C for determining physicochemical properties and heavy metal contents (Yi et al. ).

Sediment properties and heavy metals

Sediment pH was measured in a sediment and water (1:2.5) slurry using a pH meter (ST3100/F; OHAUS, Parsippany, NJ, USA). Total organic carbon (TOC) was detected by a TOC‐L analyzer (Shimadzu, Kyoto, Japan). Organic material content (OM) was calculated by multiplying %TOC by 1.724. Total N was determined by elemental analyzer (VARIO EL, Elementar, Langenselbold, Hesse, Germany). An inductively coupled plasma mass spectrometer (NexION 300X; PerkinElmer, Waltham, MA, USA) was used to analyze total P and heavy metal (As, Cd, Cr, Cu, Co, Ni, Pb, Zn, Fe, Mn) concentrations after digestion (HNO₃‐HF‐HCl).

DNA extraction, PCR amplification, and sequencing

Microbial DNA was extracted from 0.5 g of fresh sediment using E.Z.N.A. Soil DNA Kit (Omega Bio‐tek, Norcross, Georgia, USA). The V4–V5 region of the bacterial 16S ribosomal RNA gene was amplified by PCR using primers 338F/806R (95°C, 3 min; 95°C, 30 s, 27 cycles; 55°C, 30 s, 27 cycles; 72°C, 45 s, 27 cycles; 72°C 10 min). The 817–1196 region of the fungal 18S ribosomal RNA gene was amplified by PCR using the primers SSU0817F/SSU1196R (95°C, 3 min; 95°C, 30 s, 35 cycles; 55°C, 30 s, 35 cycles; 72°C, 45 s, 35 cycles; 72°C, 10 min). PCRs were performed in triplicate in a 20 μL mixture containing 4 μL 5 × FastPfu Buffer, 2 μL 2.5 mM dNTPs, 0.8 μL of each primer (5 μM), 0.4 μL FastPfu Polymerase, and 10 ng template DNA.

Amplifications were extracted from 2% agarose gels and purified using an AxyPrep DNA Gel Extraction Kit (Axygen Biosciences, Union City, California, USA), and quantified using QuantiFluor‐ST (Promega, Madison, WI, USA). Purified amplifications were pooled in equimolar concentration and paired‐end sequenced (2 × 250) on an Illumina MiSeq platform.

Microbial community composition and diversity

Raw fastq files were demultiplexed and quality‐filtered using QIIME (version 1.17). Operational units (OTUs) were clustered with a 97% similarity cutoff using UPARSE (version 7.1 http://drive5.com/uparse/); chimeric sequences were identified and removed using UCHIME. The taxonomy of each 16S rRNA and 18S rRNA gene sequence was analyzed by RDP Classifier (http://rdp.cme.msu.edu/) against the Silva 16S rRNA (Quast et al. ) and Unite (Kãμljalg et al. ) 18S rRNA database, with a confidence threshold of 97%.

Chao1, Simpson's, and Shannon indices were used to analyze microbial alpha diversity within samples. Chao1 index and Simpson's indices reflect abundance and evenness of microbial communities, respectively. For beta diversity, a heatmap based on Bray‐Curtis distance and non‐metric multidimensional scaling (NMDS) was applied to investigate differences in bacterial and fungal communities among habitats at OTU level.

Statistical analyses

One‐way analysis of variance was used to compare heavy metal contents, physicochemical properties, and microbial alpha diversity indices among habitats, using SPSS 10.0. Multiple linear regression analysis was applied to identify major environmental variables driving microbial alpha diversity indices. The influence of habitat type on microbial community was identified by Adonis with Vegan package. Redundancy analysis (RDA) was used to evaluate the influence of environmental variables on microbial communities, using Canoco 4.5. Variance inflation factor (VIF) analysis was applied to select environmental factors. Spearman correlation analysis was applied to investigate relationships between major microbial species and environmental factors.

Results

Sediment physicochemical properties and heavy metals among habitats

Sediment physicochemical properties and heavy metals among five habitats are presented in Tables , . All sediment samples were weakly alkaline. The pH values from samples collected in the duck farm were highest, and significantly higher than those of other habitat types. The highest TN, TC, and OM levels occurred in lotus pond habitat, and the lowest levels in duck farm habitat. Residential TP levels were the highest, and significantly higher than those in duck and fish farm habitat, and the reference reserve.

Sediment physicochemical properties among habitats, Baiyangdian Lake (means±SD)

Habitat type	pH	TN (g/kg)	TP (mg/kg)	TC (g/kg)	OM (g/kg)
Reference	7.97 ± 0.1B	3.2 ± 0.9B	776.6 ± 62.4BC	43.3 ± 10.6B	47.6 ± 13.17BC
Lotus pond	7.33 ± 0.1C	5.7 ± 2.4A	988.1 ± 186.3B	75.3 ± 22.8A	118.1 ± 74.63A
Duck farm	8.31 ± 0.0A	0.9 ± 0.4C	681.9 ± 51.0C	17.5 ± 4.6C	10.85 ± 4.88C
Fish farm	7.97 ± 0.2B	3.2 ± 1.4B	739.0 ± 134.4C	53.1 ± 19.3B	62.40 ± 28.35B
Residential area	7.80 ± 0.1C	3.7 ± 0.8B	1271.8 ± 218.9A	44.9 ± 6.9B	51.13 ± 13.60BC

Notes

TP, total phosphorus; TN, total nitrogen; TC, total carbon; OM, organic materials. Different letters in the same column represent significant differences at P < 0.05.

Sediment heavy metals among habitats, Baiyangdian Lake (means ± SD)

Habitat type	As (mg/kg)	Cd (mg/kg)	Cr (mg/kg)	Co (mg/kg)	Cu (mg/kg)	NI (mg/kg)	PB (mg/kg)	ZN (mg/kg)	MN (mg/kg)	FE (g/kg)
Reference	13.5 ± 2.8A	0.4 ± 0.1AB	60.7 ± 5.0B	9.8 ± 1.0B	26.2 ± 3.3B	24.2 ± 3.4B	23.4 ± 2.8BC	83.6 ± 8.7BC	628 ± 22.9AB	27.8 ± 3.3B
Lotus pond	10.3 ± 2.8AB	0.5 ± 0.1A	59.5 ± 4.3B	11.4 ± 1.1B	38.8 ± 13.9A	42.5 ± 17.8A	24.4 ± 2.9B	108.3 ± 25.1B	664 ± 40.8A	28.9 ± 2.2AB
Duck farm	9.0 ± 1.2B	0.1 ± 0.0B	55.5 ± 7.7B	9.27 ± 1.2B	19.3 ± 3.5B	21.9 ± 3.4B	18.2 ± 2.5C	61.0 ± 6.0C	520.3 ± 86.4B	26.1 ± 3.7B
Fish farm	10.5 ± 2.4AB	0.4 ± 0.1AB	66.0 ± 8.8B	10.7 ± 2.3B	26.7 ± 6.0B	28.1 ± 3.8B	23.1 ± 3.9BC	84.6 ± 22.6BC	690.7 ± 145.6A	29.7 ± 5.7AB
Residential area	9.9 ± 1.7B	0.5 ± 0.3A	70.7 ± 5.4A	13.4 ± 1.3A	39.7 ± 5.5A	31 ± 3.1B	31.5 ± 4.2A	161.8 ± 38.0A	613.1 ± 21.7AB	33.6 ± 3.6A

As, arsenic; Cd, cadmium; Cr, chromium; Co, cobalt; Cu, copper. Different letters in the same column represent significant differences at P < 0.05.

Levels of As, Cd, Mn, Fe, Pb, and Zn in sediments varied between habitat types, while levels of Cr, Co, Cu, and Ni in sediments were more uniform. Levels of Cd, Co, Cu, Pb, and Zn were highest in residential areas, followed by lotus pond habitat; levels of Ni were highest in lotus pond habitat, followed by the residential area; levels of Cr and Fe were higher in residential areas and fish farm habitats, and levels of Mn and As were higher in fish pond and reserve reference habitats. Levels of all heavy metals were lowest in duck farm habitat. In general, heavy metal pollution decreased as follows: residential areas > lotus pond > fish farm > reserve reference > duck farm.

Bacterial and fungal community composition

A total 1,273,746 clipped bacterial 16S rDNA sequences (35,007–61,635 per sediment sample) of average length 440 bp were detected by Illumina MiSeq. For fungi, 1,273,541 clipped 18SrDNA sequences were obtained, of average length was 400 bp. Rarefaction analysis (Appendix S1: Figs. S1, S2) revealed the depth of sequencing to be enough for both bacteria and fungi.

The relative abundances of bacterial major phyla, and genera within sediment samples in different habitats in Baiyangdian Lake, are depicted in Fig. . Proteobacteria (31.39–55.38%) were the most widely distributed bacterial phylum, followed by Chloroflexi (3.83–25.92%) and Bacteroidetes (3.60–17.19%). At the level of phylum, bacterial communities in sediments in each habitat were similar. However, at the level of genus, bacterial composition varied greatly among habitats. The genera Sulfuricurvum and Thiobacillus were most abundant within the reserve habitat, while Sulfuricurvum was abundant at the duck farm site; Thiobacillus was also abundant at the two lotus ponds and fish farms, and the three residential areas. The genus Dechloromonas was mainly concentrated in the residential area, Nitrospira and Desulfatiglans occurred mainly in fish farm and lotus pond habitats, and Sideroxydans occurred mainly at the reserve site. These results are consistent with heatmap analysis of major bacterial genera and samples (Appendix S1: Fig. S3).

Relative abundance of dominant bacteria at taxonomic levels of (a) phylum and (b) genus.

Fig.  depicts the relative abundances of major fungal phyla and genera within sediment samples in habitats around Baiyangdian Lake. At the taxonomic level of phylum, Ascomycota (5.37–75.13%), Ciliophora (0.64–49.30%), Chytridiomycota (2.30–21.84%), Cryptomonadales (0.98–38.08%), and Ichthyosporea (0.25–36.59%) dominate (Fig. a), with their cumulative proportion exceeding 50% of taxa in all sediment samples. Ascomycota dominated in duck farm (50.78%) and reserve reference (41.31%) sediments. The most abundant microbes in fish farm habitat were Ciliophora (20.16%) and Ichthyosporea (22.70%). Lotus pond habitat was dominated by Cryptomonadales (20.31%). Chytridiomycota (14.75%) were most abundant in residential habitat. The contribution of major fungal phyla differed among habitats.

Relative abundance of dominant fungi at taxonomic levels of (a) phylum and (b) genus.

At the genus level, the abundance of Cryptomonas and Mrakia was greatest in lotus pond habitat; Pseudallescheria and Aleuria were abundant in duck farm habitat; Tintinnidiun and Cryptomonas were the most abundant genera in fish farm habitat; Tintinnidiun and Pseudallescheria were distributed mainly in residential habitat. These results are consistent with heatmap analysis of the major fungal genera and samples (Appendix S1: Fig. S4).

Alpha diversity of sediment microbes

Alpha diversity indices of sediment microbes in habitats are presented in Table ; bacterial and fungal alpha indices for each sample are presented in Appendix S1: Tables S2, S3. Chao1 for bacteria and fungi was higher in all sediments from fish farm and residential habitats. For bacteria, each of the Chao1, Shannon, and Simpson indices within duck farm sediments was the lowest. Both Shannon and Simpson indices in sediments from lotus pond habitat were the highest and significantly higher than those in residential, reserve reference, and duck farm habitats. For fungi, the Shannon and Simpson indices in reserve reference sediments were highest, and these were significantly higher than those in lotus pond and duck farm habitats.

Bacterial and fungal diversity in sediment habitat, Baiyangdian Lake (means ± SD)

Habitat type	Bacteria	Fungi
Shannon	Chao1	Simpson	Shannon	Chao1	Simpson
Reference	6.28 ± 0.09BC	3368.15 ± 224.73B	0.052 ± 0.02BC	3.96 ± 0.29A	315.81 ± 97.99AB	0.10 ± 0.07A
Lotus pond	6.74 ± 0.10A	3546 ± 157.51AB	0.11 ± 0.03A	3.35 ± 0.11BC	284.39 ± 47.42B	0.046 ± 0.02BC
Duck farm	6.05 ± 0.37C	3302.24 ± 316.01B	0.026 ± 0.02C	3.72 ± 0.55AB	274.36 ± 47.42B	0.085 ± 0.03AB
Fish farm	6.54 ± 0.12AB	3720.67 ± 154.55A	0.059 ± 0.02B	3.33 ± 0.28C	360.60 ± 39.90A	0.032 ± 0.02C
Residential area	6.28 ± 0.30BC	3760.87 ± 216.96A	0.038 ± 0.02BC	3.88 ± 0.13A	363.93 ± 32.36A	0.069 ± 0.10AB

Note

Different letters in the same column represent significant differences at P < 0.05.

Multiple linear regression analysis was applied to identify which environmental variables were most likely responsible for driving microbial alpha diversity. Fig.  and Appendix S1: Table S4 detail multiple linear regression analysis results for bacteria. Concentrations of Co correlate positively with bacterial Chao1 index, suggesting sediment bacterial abundance was predominantly regulated by cobalt.

Linear regression relationships between sediment Co concentration and bacterial Chao1 index.

Bacteria and fungi beta diversity characteristics

Non‐metric multidimensional scaling (NMDS) analyses based on OTU level data were used to visualize and explore community differences between samples for bacteria and fungi (Fig. ). For lotus pond, fish farm, and reserve habitats, both bacterial community and fungal communities cluster into groups related to habitat type. For the residential area, the bacterial community clustered into a close group, except for sites at XZZC1, DC1, and ZN2. Fungal communities of DC sites differed slightly from those of XZZC and AN sites. Although duck farm bacterial community replicates were variable, they were still more similar to each other than they were to those of other habitats. For fungi, only communities from sites CPT2 and CPT3 clustered together; those of site CPT1 were scattered. In general, both bacterial and fungal communities within a habitat were similar, but their composition varied between habitats. Adonis results for bacteria (R² = 0.52, P = 0.001) and fungi (R² = 0.49, P = 0.001) further indicate habitat type had a significant influence on microbial community composition. Multiple comparisons of Adonis among groups are presented in Table ; results indicate microbial community composition varies significantly between habitats, except for duck farm and reserve habitats.

Non‐metric multidimensional scaling analysis of (a) bacterial (at OTU level) and (b) fungal (at OTU level) communities in habitats, Baiyangdian Lake.

Results of multiple comparisons of Adonis among different groups

Groups	Bacterial community	Fungal community
R 2	P	R 2	P
LP, DF	0.62	0.026	0.43	0.028
LP, FF	0.42	0.007	0.44	0.077
LP, R	0.53	0.038	0.37	0.024
LP, RA	0.29	0.002	0.32	0.001
DF, FF	0.52	0.011	0.42	0.014
DF, R	0.58	0.1	0.33	0.1
DF, RA	0.30	0.001	0.27	0.006
FF, R	0.35	0.012	0.43	0.016
FF, RA	0.26	0.001	0.42	0.001
TR, RA	0.28	0.009	0.28	0.004

Note

LP, lotus pond; DF, duck farm; FF, fish farm; TR, reserve reference; RA, residential area.

A sample distance heatmap at the OTU level based on the Bray‐Curtis similarity metric was applied to evaluate differences across microbial communities (Appendix S1: Figs. S5, S6). For bacteria, the fish farm and reserve sites clustered into one group; each of the three other habitats is separated. Fungal communities in all samples clustered into the five groups related to their habitat. These observations are similar to those of NMDS.

Influences of environmental factors on microbial communities

RDA results for bacterial genera are depicted in Fig. a. The two RDA axes explain 35.35% and 12.01% of variation, with a cumulative interpretation rate of 47.36%. Heavy metals, Zn (R² = 0.60, P = 0.001), Co (R² = 0.32, P = 0.009), and Pb (R² = 0.28, P = 0.04), and sediment physicochemical properties, including pH (R² = 0.5, P = 0.001), TP (R² = 0.40, P = 0.005), and TN (R² = 0.33, P = 0.008), significantly affected bacterial community structure. This result is consistent with that of Spearman correlation analysis between environmental variables and major genera of bacteria (Table ). While both Sulfuricurvum and Sulfurovum negatively correlated with TN, both Desulfatiglans and Nitrospira were positively correlated with TN. The genus Hydrogenophaga was positively correlated with pH and negatively correlated with TN; Smithella and Syntrophus were positively correlated with Zn, Co, and Pb.

Redundancy analysis of (a) bacterial and (b) fungal data and environmental variables.

Spearman correlation analysis between environmental variables and major bacterial genera (top 20)

Genera	Zn	pH	TP	TN	Co	Pb
Sulfuricurvum	−0.348	0.647**	−0.363	−0.722**	−0.264	−0.254
Thiobacillus	−0.165	0.049	−0.140	0.261	−0.188	−0.081
Dechloromonas	0.195	−0.020	0.224	−0.183	0.195	0.108
Sulfurovum	−0.131	0.437*	−0.202	−0.627**	−0.041	−0.079
Nitrospira	−0.052	−0.134	−0.228	0.454*	−0.122	−0.047
Desulfatiglans	0.339	−0.386	0.272	0.660**	0.254	0.334
Smithella	0.546**	−0.280	0.605**	0.037	0.658**	0.459*
Desulfobulbus	−0.140	0.380	−0.189	−0.662**	0.054	−0.190
Sideroxydans	−0.313	0.392	−0.307	−0.224	−0.405*	−0.286
Flavobacterium	−0.119	0.080	−0.172	−0.114	−0.178	−0.155
Romboutsia	0.362	−0.242	0.316	0.046	0.527**	0.426*
Caldithrix	0.539**	−0.526**	0.585**	0.511**	0.516**	0.560**
Syntrophus	0.647**	−0.476*	0.736**	0.279	0.601**	0.501*
Hydrogenophaga	−0.145	0.402*	−0.071	−0.670**	0.024	−0.139
Desulfobacca	0.165	−0.175	0.022	0.505*	−0.008	0.199
Mycobacterium	0.495*	−0.321	0.403*	−0.001	0.701**	0.352
Ferritrophicum	−0.204	0.494*	−0.255	−0.283	−0.142	−0.168
Sulfurisoma	−0.263	−0.178	−0.236	0.131	−0.487*	−0.280
Arenimonas	−0.059	0.078	−0.126	−0.014	−0.007	−0.186
Ignavibacterium	0.050	−0.170	0.028	0.348	−0.028	0.138

*0.01 < P ≤ 0.05.

**0.001 < P ≤ 0.01.

RDA results for fungal genera are depicted in Fig. b. The two RDA axes explain 17.54% and 10.28% of variation in fungal genera, with a cumulative interpretation rate of 27.82%. Of all environmental variables, Co (R² = 0.35, P = 0.007) and TP (R² = 0.26, P = 0.033) most significantly influenced the structure of fungal communities. Results of Spearman correlation analysis between environmental variables and dominant fungal genera were highly related to RDA (Table ). Specifically, relative abundances of Aleuria, Cladosporium, and Pseudallescheria were negatively correlated with OM, while the relative abundance of Mrakia was positively correlated with TP and Zn.

Spearman correlation analysis between environmental variables and major fungal genera (top 20)

Genera	Zn	Co	Mn	TP	OM
Tintinnidium	0.185	0.219	0.302	0.039	0.072
Cryptomonas	−0.158	−0.298	0.224	0‐.333	0.318
Pseudallescheria	0.023	0.067	−0.382	0.140	−0.510**
Cladosporium	−0.351	−0.323	−0.510**	−0.241	−0.682**
Aleuria	−0.307	−0.248	−0.185	−0.343	−0.582**
Freshwater_Choanoflagellates_1	0.370	0.310	0.518**	0.267	0.358
Monosiga	−0.032	−0.107	−0.394	0.164	−0.059
Mrakia	0.532**	0.312	−0.124	0.632**	0.275
Colpodidium	−0.206	−0.081	0.100	−0.234	−0.373
Undella	0.286	0.322	0.449*	0.110	0.241
Sphaeroeca	−0.300	−0.382	−0.047	−0.347	0.099
Katablepharis	−0.242	−0.346	0.125	−0.366	0.028
Diaphanoeca	0.496*	0.626**	0.204	0.581**	0.064
Cryptocaryon	−0.009	−0.030	0.279	−0.101	0.203
Obertrumia	−0.112	−0.076	−0.198	−0.095	−0.327
Lulwoana	0.015	−0.157	0.195	−0.039	0.627**
Microdiaphanosoma	−0.348	−0.383	−0.246	−0.292	−0.006
Paratrimastix	0.028	0.092	−0.183	0.207	−0.391
Gaeumannomyces	−0.133	−0.150	0.000	−0.203	−0.182
Bryometopus	−0.402*	−0.232	−0.184	−0.431*	−0.482*

*0.01 < P ≤ 0.05.

**0.001 < P ≤ 0.01.

Discussion

The villages selected in this study were greatly disturbed by human activities by discharge of untreated domestic sewage, and disposal of waste into water. Not surprisingly, heavy metal (TP, TN, and TC) concentrations were higher in residential areas. Samples were collected during winter. During this time, considerable vegetation residue had accumulated in sediments in lotus pond habitat, which contributed to high TN, TC, and OM concentrations in sediments as plants decomposed (Guan et al. ). Because the duck farm was located in the upper reaches, and pollutants dispersed with currents, pollution around this farm was lower than elsewhere.

Each of Thiobacillus, Sulfuricurvum, Sulfurovum, Desulfatiglans, and Desulfobulbus was a dominant genus. These genera produce and assimilate sulfur and are involved with oxidize and desulfur sediments. They actively participate in the freshwater sulfur cycle and are widely distributed in anaerobic conditions. Nitrospira is ubiquitous and plays a role in the nitrogen cycle; it was highly concentrated in lotus pond, reserve, fish farm, and residential habitats, but concentrations of it were extremely low in duck farm habitat. Correspondingly, TN in the duck farm was significantly lower (0.9 g/kg) than in other habitats (3.2–5.7 g/kg). Species in the genus Sideroxydans can concentrate and accumulate iron, and were mainly distributed in reserve sites; species of Dechloromonas can degrade benzene, toluene, ethylbenzene, and xylene—compounds found in anaerobic conditions (Chakraborty et al. )—and were highly abundant in residential habitats, indicative of serious organic pollution in this environment. The genus Smithella is heavy metal tolerant (Wu et al. ), and its abundance was positively correlated with concentrations of Zn, Co, and Pb.

Within all habitat types, the relative abundances of identified genera ranged 36.19–56.59%, and dominant genera (the top 10) were 9.65–28.24% of all species at genus level. Major genera participating in S and N cycles accounted for 7.70–20.85% of all species at genus level and were most abundant, followed by genera responsible for degrading organic pollutants, and those tolerant of heavy metals.

At the generic level, Cryptomonas represents a food source for small zooplankton and fishes and was mainly distributed in fish farm and lotus pond habitats. High abundances of Pseudallescheria occurred in duck farm and residential habitats; species in this genus can causes mycetoma, maduromycosis, and other infections in humans; P. boydii can also decompose cellulose and degrade organic matter, and has great potential to biodegrade phenolic compounds, cyanides and synthetic dyes, and in the absorption, accumulation, and degradation of polycyclic aromatic hydrocarbons. Organic pollutants from duck manure enriched in duck farm habitat, and from domestic sewage and garbage in residential habitats, contributed to increased relative abundances of Pseudallescheria. The fungal cup genus Aleuria (Ascomycota) was most abundant in duck farm sediments, which were covered with duck excrement. This saprobic genus decomposes organic matter, dead organisms, and excrement. Cladosporium is one of the most common mold genera and was distributed evenly among all sample sites.

High concentrations of heavy metals can significantly decrease Chao1 (Chodak et al. ) and Shannon (Jose et al. ) indices. Wu et al. () demonstrated Hg to positively correlate with Chao1 in soil‐dwelling bacteria subjected to different land use types in an electronic waste recycling region. Around Baiyangdian Lake, Co positively correlated with the bacterial Chao1 index. Bacterial and fungal communities at sites CPT1 differed greatly from those at sites CPT2 and CPT3. Sediments at site CPT1 were collected in an area where ducks swam freely without nets, while those at sites CPT2 and CPT3 were located around a duck pond where more excrement had accumulated. Of the residential areas, the population at DC was greater than at XZZC and ZN, and domestic sewage discharge and garbage release was also greater. Accordingly, the fungal communities of DC differed slightly from those of the two other villages. In general, microbial communities were more similar to each other within a habitat than they were to those between habitats.

Of environmental factors, pH had the most significant effect on bacterial community structure. Many other studies (Chodak et al. , Jiang et al. , Yun et al. ) have concluded the same, determining pH directly influences bacterial metabolism, growth, and reproduction (Rousk et al. , Lachaud ); pH also changes the physical and chemical properties of sediment, contributing to further changes in bacterial community structure and diversity. Conversely, pH had little impact on fungal community structure. Fungi generally thrive in acidic conditions (Dix and Webster ), but the sediment pH values around Baiyangdian Lake were alkaline, inhabiting the growth and/or survival of some fungal taxa.

Total phosphorus significantly affected both bacterial and fungal community structure. Phosphorus is an important macronutrient for organisms as it is involved in a number of cellular processes (e.g., photosynthesis, respiration, energy storage and transfer, and cell division; Wu et al. , Coolon et al. ). Around Baiyangdian Lake, OM had little impact on either bacterial or fungal communities, which might be related to the anaerobic environment of sediments.

Some heavy metals play integral roles in the life processes of microorganisms. Cobalt occurs in complex molecules with a wide array of functions (Bruins et al. ) that significantly influence bacterial and fungal communities. Both Zn and Pb also significantly influence bacterial community structure; Zn, as an essential element, has a recognized role in the life processes of microorganisms (Wu et al. ).

Conclusion

The composition and diversity of bacterial and fungal communities in sediments in different aquatic habitat types around Baiyangdian Lake was investigated. Dominant phyla and genera, particularly in fungal communities, varied between habitats. Bacterial genera involved in S and N cycles, and the degradation of organic pollutants, were dominant. Pathogenic and saprophytic fungi were major taxa. Co was the predominant driver of the bacterial Chao1 index. Fungal communities were more sensitive to human disturbance than bacterial communities. Bacterial communities were structured mainly by Zn, pH, and TP, while Co and TP were dominant factors influencing fungal community structure. Species of Cryptomonas represent food sources for small zooplankton and fishes, and were mainly distributed in fish farm and lotus pond habitats. Species of Pseudallescheria can cause mycetoma, maduromycosis, and other infections in humans, and are highly abundant in duck farm and residential habitats. Species of Aleuria are saprobes, decomposing organic matter, dead organisms, and excrement and were highly accumulated in duck farm habitats. The influence of human activities on the composition of microbial communities around Baiyangdian Lake is significant.

Acknowledgments

This study is supported by the National Natural Science Foundation of China (51439001, 51722901) and the National Key R&D Program of China (2018YFC0407403), and the Fundamental Research Funds for the Central Universities. We thank Chuqiao Lin for editing figures and Steve O'Shea (PhD) from Edanz Group for editing a draft of this manuscript.