Variation of physiochemical properties in sediment microenvironments
Hereafter, Mg(OH)2 and Ca(NO3)2 treatments are referred to as Mg and Ca microcosms, respectively. The DO concentration in the Ca microcosms gradually increased, peaking on day 21, then decreased, whereas the DO concentration in the Mg microcosms remained stable at approximately 0.16 mg/L. The DO concentration in the control microcosms continuously decreased throughout the experimental period (Fig. 1A). The pH in the control and Ca microcosms remained neutral and exhibited no significant difference between them, whereas the pH in the Mg microcosms was significantly increased (p < 0.01), peaking at 9.28 on day 14, then remained alkaline (pH > 9) until day 30 (Fig. 1B). The ORP values in the control and Mg microcosms remained negative throughout the experimental period, whereas those in the Ca microcosms increased significantly (p < 0.01) from negative to positive and peaked at 72.1 mV on day 21 (Fig. 1C). The AVS concentration in the Ca microcosms was significantly decreased (p < 0.01) and remained low after day 7, whereas those in the control and Mg microcosms continuously increased and peaked on day 21; this was followed by a decrease in the Mg microcosms on day 30 (Fig. 1D). The TOC content in the control, Mg, and Ca microcosms showed similar variation, and was relatively higher in the Mg and Ca microcosms than in control microcosms (Fig. 1E). The sediment became fluffy and floated upward in the Ca microcosms, and the color changed from black to yellowish brown after the 30-day treatment. However, sediments in the Mg microcosms were much denser and showed no signs of floating upward, and only the surface sediment changed to yellowish brown (Fig. 1F).
Sediment bacterial diversity and community composition
The sequencing results are presented in Table 1. The number of effective sequences ranged from 40,488 to 68,986, the number of OTUs with 97% similarity ranged from 1,597 to 1,825, and the coverage of all samples was above 0.970, indicating that the sequencing results accurately expressed the microorganism diversity. The rarefaction curves of all samples showed an upward trend, revealing a high OTU richness and abundance (Fig. S1). Although the highest Chao1 and ACE indices were obtained in the Ca microcosms on day 14, the overall Chao1 and ACE indices (including days 14 and 30) remained relatively higher in the Mg microcosms, indicating a higher richness of bacterial community in after the 30-day treatment. The Shannon and Simpson indices of the Ca and Mg microcosms both decreased after the 30-day treatment, and exhibited no significant differences between them, which indicated that both Mg(OH)2 and Ca(NO3)2 addition treatments reduced the diversity of the sediment bacterial community and promoted the transformation of dominant bacteria.
Table 1
Relative abundance and diversity data of the sediment bacterial community
Group | Sequence number | OTUs (97%) | Bacterial richness | Bacterial richness | Coverage |
Chao1 | ACE | Shannon | Simpson |
Origin0 | 46,058 | 1,696 | 2,082.11 | 2,098.94 | 6.08 | 0.0060 | 0.9788 |
Control30 | 26,214 | 1,604 | 1,976.13 | 2,009.57 | 5.79 | 0.0112 | 0.9791 |
Mg14 | 49,731 | 1,739 | 2,040.61 | 2,028.62 | 6.20 | 0.0063 | 0.9819 |
Mg30 | 21,861 | 1,662 | 2,037.63 | 2,038.40 | 6.01 | 0.0071 | 0.9796 |
Ca14 | 51,458 | 1,825 | 2,161.80 | 2,159.95 | 6.24 | 0.0054 | 0.9797 |
Ca30 | 60,391 | 1,597 | 1,842.56 | 1,869.99 | 6.00 | 0.0062 | 0.9831 |
OTUs, operational taxonomic units; ACE, Abundance-based Coverage Estimator |
According to the OTU taxonomic analysis, 31 phyla within bacteria and archaea were identified, 11 of which had relative abundance greater than 1.0% in (Fig. 2A). In general, Proteobacteria was the most abundant phylum across all samples (average relative abundance of 33.68%), followed by Chloroflexi (23.45%), Planctomycetes (7.75%), and Actinobacteria (7.68%) (Fig. 2A; Table S1-1). Proteobacteria and Actinobacteria showed high relative abundance in the Ca microcosms (41.30% and 11.50%, respectively), whereas Chloroflexi and Planctomycetes were abundant in the Mg microcosms (26.57% and 13.73%, respectively) (Table S1-1). In addition, Cyanobacteria showed a high relative abundance in the control microcosms on day 30 (Fig. 2A; Table S1-1).
At the class level, a total of 79 classes were obtained across all samples, with 15 exhibiting a relative abundance of more than 1.0% (Fig. 2B). Anaerolineae was the most abundant class in all treatments, with an average relative abundance of 15.47%, followed by Alphaproteobacteria (9.97%), Gammaproteobacteria (8.28%), Deltaproteobacteria (8.03%), Planctomycetia (7.73%), Betaperoteobacteria (6.67%), and Actinobacteria (5.44%) (Fig. 2B; Table S1-2). The classes Anaerolineae and Planctomycetia were obviously more abundant in the Mg microcosms, whereas Alphaproteobacteria, Gammaproteobacteria, Betaproteobacteria, and Actinobacteria were more abundant in the Ca microcosms (Fig. 2A & B; Table S1-1 and S1-2).
The 32 dominant genera, which included 15 unclassified genera, constituted more than 0.1% of the average relative abundance. An unclassified genus in the family Anaerolineaceae was dominant in all samples (average relative abundance of 14.43%), followed by unclassified genera in the family Planctomycetaceae (6.72%) and phylum Chloroflexi (6.68%) (Fig. 3; Table S1-3). Hierarchical clustering of samples at the genus level showed that, although the Mg microcosms on days 14 and 30 were clustered together with the original samples (day 0), the Ca microcosms on days 14 and 30 were clustered into an independent branch that was separate from the other microcosms (vertical dendrogram in Fig. 3).
The 32 dominant genera were divided into three clusters (horizontal dendrogram in Fig. 3). The genera in cluster I were dominant in the control microcosm on day 30, those in cluster II were dominant only in the Ca microcosms on day 30, and those in cluster III were dominant in the Mg microcosms on days 14 and 30 and in the Ca microcosms on day 14 (heatmap in Fig. 3). Clusters I, II, and III were further divided into subgroups A and B, subgroups C and D, and subgroups E and F, respectively. Subgroup A included genus Gp17 and two unclassified genera in phylum Chloroflexi, which were relatively abundant in the Mg microcosms on day 30. Subgroup B included unclassified genera in the classes Cyanobacteria, Leuconostoc, Methylocystis, and Desulfomonile, which exhibited relatively high abundance in the control microcosms (Control30) (Fig. 3, Table S1-3). Subgroup C genera that were abundant in the Ca microcosms on day 14 or 30 included the genera Gp16, Gaiella, and unclassified genera in the class Actinobacteria, whereas those in subgroup D included the genera Thiobacillus, Lysobacter, and Thermomonas. Subgroup E mainly consisted of Aminicenantes genera incertae sedis and other unclassified genera belonging to class Betaproteobacteria, Deltaproteobacteria, and Gammaproteobacteria, all of which were abundant in the Mg microcosms on days 14 and 30 and in the Ca microcosms on day 14. Subgroup F included the genera Thermogutta, Gp6, and unclassified genera belonging to the class Planctomycetia, which were mainly abundant in the Mg microcosms on day 14 (Fig. 3, Table S1-3).
Variations in SRB and NRB composition
The composition of SRB and NRB detected by dsrB and nirS gene sequencing is shown in Fig. 4. The dominant species of SRB mainly belonged to the phyla Proteobacteria and Acidobacteria, the classes Acidobacteria and Deltaproteobacteria, and the genera Desulfobacca, Desulfobulbus, Desulfomonile, Desulfosarcina, and Syntrophobacter (Fig. 4A–C). Furthermore, the abundance of the genera Desulfobacca, Desulfomonile, and Desulfosarcina decreased in both Mg and Ca microcosms, whereas the genus Syntrophobacter decreased only in the Mg microcosms, and the genus Desulfobacterium decreased only in the Ca microcosms after the 30-day treatment. The dominant species of NRB mainly belonged to the phyla Proteobacteria and Planctomycetes, the classes Alphaproteobacteria, Betaproteobacteria, Gammaproteobacteria, and Planctomycetia, and the genera Acidovorax, Dechlorononas, Dechlorospirillum, Pseudomonas, Rhodanobacter, and Rubrivivax (Fig. 4D–F). Furthermore, the genera Dechlorononas and Rhodanobacter decreased in both Mg and Ca microcosms, the genera Dechlorospirillum and Pseudomonas decreased in the Mg microcosms and increased in the Ca microcosms, and the genus Acidovorax increased only in the Ca microcosms after the 30-day addition treatment.
FAPROTAX and PICRUSt functional prediction of bacterial communities
Based on the FAPROTAX analysis, more than 90 functional groups were predicted to appear in all samples, and 23 functional groups with a relative abundance of greater than 0.1% were regarded as the dominant groups (Fig. 5A). Chemoheterotrophy and aerobic chemoheterotrophy (average relative abundance of 7.47% and 3.40%, respectively) were the most dominant functional groups in all treatments, followed by fermentation (2.19%), sulfate respiration (2.48%), and the respiration of sulfur compounds (2.49%) (Table S2). Fermentation and chloroplasts related to the carbon cycle were dominant in the control microcosms after the 30-day treatment (Fig. 5A). The abundance of functional groups related to sulfur and nitrogen cycles differed significantly between the control, Ca, and Mg microcosms. Sulfate respiration and respiration of sulfur compounds were decreased, whereas dark oxidation of sulfur compounds and dark sulfide oxidation were increased in the Ca microcosms after the 30-day treatment. Nitrogen fixation was obviously decreased in the Mg microcosms, and nitrate reduction was increased in the Ca microcosms (Fig. 5A).
Based on the PICRUSt prediction, a total of six biological metabolic pathways (category level 1) were detected, including environmental information processing, metabolism, genetic information processing, cellular processes, organismal systems, and unclassified metabolic pathways, as well as 40 subfunctions (category level 2); the top 25 according to their relative abundance are shown in Fig. 5B. The relative abundance was highest for the genes potentially involved in the metabolism pathway in each treatment; the genes involved in the metabolism pathway subfunctions, including carbohydrate metabolism, amino acid metabolism, energy metabolism, metabolism of cofactors and vitamins, lipid metabolism, nucleotide metabolism, xenobiotics biodegradation, and metabolism, were also abundant (Fig. 5B). The genes involved in genetic information processing and environmental information processing pathways followed the metabolism pathway, and their subfunctions, including membrane transport, replication and repair, and translation, were also abundant in all treatments (Fig. 5B).
Furthermore, PICRUSt predicted the occurrence of several genes potentially involved in nitrogen and sulfur metabolism subfunctions. The dissimilatory sulfate reduction (including genes encoding sulfate adenylyltransferase (sat), adenylylsulfate reductase subunit A/B (aprAB), and sulfate reductase dissimilatory-type alpha/beta subunit (dsrAB)) were obviously decreased in the Ca microcosms after the 30-day treatments, whereas the assimilatory sulfate reduction (including genes encoding sulfate adenylyltransferase subunit 2 (cysD), adenylylsulphate kinase (cysC), sulfite reductase (NADPH) flavoprotein alpha-component (cysJ), and sulfite reductase (NADPH) hemoprotein beta-component (cysI)) were obviously decreased in the Mg microcosms (Fig. 6A). Denitrification, including genes encoding nitrate reductase 1 alpha/beta subunit (narG/narH), nitrite reductase (NO-forming) (nirK, nirS), nitric oxide reductase cytochrome b-containing subunit II (norB), and nitrous-oxide reductase (nosZ)), were more abundant in the Ca microcosms than the Mg microcosms, whereas dissimilatory nitrate reduction to ammonium (including genes encoding periplasmic nitrate reductase NapA (napA), nitrite reductase (NAD(P)H) large/small subunit (nirB/nirD), and formate-dependent nitrite reductase periplasmic cytochrome c552 subunit (nrfA)) were relatively abundant in the Ca microcosms (Fig. 6B).
Correlation between dominant bacterial genera and predicted metabolism functional groups
Each bacterial functional group correlated well with certain dominant bacterial genera. The dominant genera and metabolic functional groups predicted by FAPROTAX were divided into six groups according to the Pearson correlation (Fig. 7A). Group I included the functional groups of chemoheterotrophy, aerobic chemoheterotrophy, chitinolysis, dark sulfide oxidation, and ureolysis, which exhibited a positive correlation with genera Thiobacillus, Thermomonas, Lysobacter, Gp16, Phenylobacterium, and Gaiella, as well as unclassified genera belonging to the phylum Actinobacteria. Group II included the functional groups of nitrogen fixation, hydrocarbon degradation, and methylotrophy, which were positively correlated with the genus Methylocystis (p < 0.05). Group III included the functional groups of fermentation and chloroplasts, which were positively correlated with the genera Desulfomonile, Hyphomicrobium, and unclassified genera belonging to the phyla Cyanobacteria and Proteobacteria. Group IV included the functional groups of sulfate respiration and hydrogenotrophic methanogenesis, which exhibited a positive correlation with Gp17 and unclassified genera in the phylum Chloroflexi. No functional groups were clustered into Group V, although the genera Gp6, Thermogutta, and unclassified genera belonging to phylum Planctomycetes were negatively correlated with the functional groups of chemoheterotrophy and nitrogen fixation (p < 0.05). Group VI included the functional groups of nitrate reduction, which were positively correlated with unclassified genera belonging to classes Gammaproteobacteria, Betaproteobacteria, and Deltaproteobacteria (p < 0.05) (Fig. 7A; Table S3-1 and S3-2).
The dominant genera and potential functions of sediment bacterial communities based on PICRUSt prediction were divided into five groups according to the Pearson correlation (Fig. 7B). Group I included the genes potentially involved in membrane transport, carbohydrate metabolism, and biosynthesis of other secondary metabolites, which showed a positive correlation with genus Gp17 and unclassified genera belonging to the phyla Chloroflexi and Anaerolineaceae. These genera were negatively correlated with the genes potentially involved in the metabolism of terpenoids and polyketides, the endocrine system, xenobiotic biodegradation, metabolism, and the metabolism of other amino acids (p < 0.05). No genes involved in potential functions were clustered into Group II, although the genus Aminicenantes genera incertae sedis, unclassified genera belonging to the family Sinobacteraceae, and classes Gammaproteobacteria, Betaproteobacteria, Deltaproteobacteria, and Chromatiales were negatively correlated with the genus Desulfomonile (p < 0.05) and the genes potentially involved in cell growth and death. Group III included genes involved in cellular processes, environmental information processing, genetic information processing, and metabolism, which were positively correlated with genera Gp6, Thermogutta, and unclassified genera belonging to the family Planctomycetaceae. These genera were significantly negatively correlated with the genes involved in cell growth and death (p < 0.05). Group IV included cell growth and death, and the genera Methylocystis, Leuconostoc, Desulfomonile, Hyphomicrobium, Clostridium sensu stricto, unclassified genera of family Caldilineaceae, phylum Cyanobacteria, and order Rhizobiales. Group V included genes potentially involved in the metabolism of terpenoids and polyketides, the endocrine system, xenobiotic biodegradation, metabolism, and the metabolism of other amino acids, which were positively correlated with the genus Gp16, Gaiella, Aquabacterium, Phenylobacterium, Sphingomonas, and unclassified genera belonging to the family Coriobacteriaceae, suborders Frankineae and Acidimicrobineae, and phylum Actinobacteria (p < 0.05) (Fig. 7B).
Effects of sediment environmental factors on bacterial communities and metabolism functional groups
According to the RDA, the overall variances of the bacterial community structures at the genus level were explained by 61.73% in RDA 1 and by 26.23% in RDA 2 (Fig. 8A), and the variances of the predicted metabolic functions were explained by 65.67% and 77.60% in RDA 1 and by 28.88% and 20.21% in RDA 2 (Fig. 8B and C). ORP was the principal environmental factor that explained the community variations in both diagrams (p < 0.05). Most of the dominant genera, including Gp16, Gaiella, Thiobacillus, Thermomonas, and Lysobacter, as well as the metabolism function groups, including chemoheterotrophy, aerobic chemoheterotrophy, dark sulfide oxidation, ureolysis, and chitinolysis, were positively correlated with ORP (p < 0.05), whereas unclassified genera in the family Anaerolineaceae, the functional groups of sulfate respiration, and the genes potentially involved in carbon fixation in photosynthetic organisms and oxidative phosphorylation were negatively correlated with ORP (p < 0.05) (Fig. 8A–C). Genes potentially involved in sulfur metabolism and methane metabolism were also negatively correlated with ORP (Fig. 8C). AVS was negatively correlated with genera Thiobacillus, Thermomonas, and Lysobacter and sulfide oxidation, ureolysis, and chitinolysis functional groups (p < 0.05), but positively correlated with unclassified genera in the phylum Chloroflexi and the sulfate respiration functional group (p < 0.05) (Fig. 8A & B). DO was negatively correlated with unclassified genera in the class Cyanobacteria and the chloroplast functional group, as well as genes potentially involved in photosynthesis-antenna proteins, photosynthesis, and photosynthesis protein functional groups (p < 0.05). pH was positively correlated with the genus Thermogutta, and potential genes involved in sulfur metabolism, methane metabolism, and oxidative phosphorylation, and negatively correlated with the nitrogen fixation functional group (p < 0.05) (Fig. 8A–C). TOC was positively correlated with the genus Thermomonas and the ureolysis functional group (p < 0.05) (Fig. 8A & B; Fig. S2 & S3).