Evaluation of environmental factors and microbial community structure in an important drinking-water reservoir across seasons

Abstract

The composition of microbial communities varies in water and sediments, and changes in environmental factors have major effects on microbiomes. Here, we characterized variations in microbial communities and physicochemical factors at two sites in a large subtropical drinking water reservoir in southern China. The microbiomes of all sites, including the diversity and abundance of microbial species, were determined via metagenomics, and the relationships between microbiomes and physicochemical factors were determined via redundancy analysis. The dominant species in sediment and water samples differed; Dinobryon sp. LO226KS and Dinobryon divergens were dominant in sediment samples, whereas Candidatus Fonsibacter ubiquis and Microcystis elabens were dominant in water. The diversity was also significantly different in microbial alpha diversity between water and sediment habitats (p < 0.01). The trophic level index (TLI) was the major factor affecting the microbial community in water samples; Mycolicibacterium litorale and Mycolicibacterium phlei were significantly positively related to TLI. Furthermore, we also studied the distribution of algal toxin-encoding genes and antibiotic-resistant genes (ARGs) in the reservoir. It found that water samples contained more phycotoxin genes, with the cylindrospermopsin gene cluster most abundant. We found three genera highly related to cylindrospermopsin and explored a new cyanobacteria Aphanocapsa montana that may produce cylindrospermopsin based on the correlation through network analysis. The multidrug resistance gene was the most abundant ARG, while the relationship between ARGs and bacteria in sediment samples was more complicated than in water. The results of this study enhance our understanding of the effects of environmental factors on microbiomes. In conclusion, research on the properties, including profiles of algal toxin-encoding genes and ARGs, and microbial communities can aid water quality monitoring and conservation.

Keywords: cylindrospermopsins (CYNs); metagoenomics; multidrug resistance genes; subtropical drinking water source; trophic level index (TLI).

Figure 1

PCA and alpha diversity of sediment samples and water samples. (A) PCA of all samples, (B) PCA of water samples; (C) Shannon index of water and sediment samples.

Figure 2

Histogram of relative abundances at the species level. The 20 most abundant species are shown, and “Other” indicates species with relative abundances less than 0.6% across all samples.

Figure 3

Network analysis of dominant species with significant differences among samples. The thickness and thinness of the lines indicate strong and weak correlations, respectively. Orange lines indicate positive correlations, and gray lines indicate negative correlations. Blue circles indicate taxa with higher abundances in water samples, and green circles indicate taxa with higher abundances in sediment samples.

Figure 4

Relationships between TLI and the Shannon index. (A) KZ samples, (B) QSK samples.

Figure 5

Redundancy analysis ordination diagram of all communities and environmental variables. Circles correspond to samples.

Figure 6

Correlation heatmap for microbes and environmental factors. The magnitude of the R-values is shown in the color gradient. *p ≤ 0.05, **p ≤ 0.01, and ***p ≤ 0.001.

Figure 7

Cross-genus co-occurrence network of phycotoxin gene clusters in water samples. Network analysis coloring based on taxonomic genus preference. Associations indicated a significant (p < 0.001), scored as positive (spearman’s p > 0.8). The size of each node is proportional to the SAV degree; the thickness of the connection between two nodes is proportional to the value of the Spearman correlation coefficient. Heatmap of the relative abundances of algal toxin gene clusters in all samples.

Figure 8

Cross-genus co-occurrence network of ARGs in samples. Network analysis coloring based on taxonomic genus preference. Associations indicated a significant (p < 0.001), scored as positive (spearman’s p > 0.9 or p < −0.9). The size of each node is proportional to the SAV degree; the thickness of the connection between two nodes is proportional to the value of the Spearman correlation coefficient. (A) Sediment samples; (B) water samples.