Physicochemical properties of samples
A total of twenty water samples, including twelve samples from the abandoned coal mine drainage (Group M, M1 ~ M12) and eight samples from Pinglu River (Group R, R1 to R8), were collected in our study. The comparison of the physicochemical parameters of the water samples from group M and group R is shown in Fig. 2. The pH value of Group M samples ranged from 2.78 to 7.57, and 58.3% (7 out of 12) of the samples have pH < 4.0. The continuous confluence of acidic drainage from abandoned coal mines has led to a gradual increase in the acidity of the Pinglu River from upstream to downstream, with the pH of the river decreasing from 8.16 (R1) in the upstream to a minimum of 3.22 (R5). Comparing with the surface water environmental quality standard (GB 3838 − 2002), the pH of river water in the middle and lower reaches of Pinglu River was significantly lower than the standard limit (6 ~ 9), indicating that the river had been seriously polluted by acidity. AMD from abandoned coal mines mainly originates from the oxidation of sulfide minerals (mainly pyrite) in coal and rocks in the underground mining area (Chen et al. 2021b), and a large amount of Fe, SO42− and H+ is generated during this oxidation process (Acharya and Kharel 2020). Water samples from abandoned coal mines drainage in our study also had high Fe and SO42− contents, with average contents of 510.52 mg/L and 1954.29 mg/L, respectively, which were 8.25 and 4.96 times higher than those of river water samples. The content of metal ions such as Al, Mn, Zn and Cd in the abandoned coal mine drainage samples was significantly higher than that in the river samples (Fig. 2). In addition to metal ions, the Ca and Mg contents in the samples of group M were also significantly higher than those of group R (p < 0.01). Carbonate rocks are widely distributed in the stratigraphic structure of this study area, with bauxite and bauxitic mudstone beneath the coal seam and fine-grained tuff and dolomitic tuff in the overlying rocks. During the formation and migration of AMD, these carbonate rocks were dissolved and released Ca2+ and Mg2+, while the bauxite in the overlying rock released a large amount of Al ions, thus causing high Al content (average 100.81 mg/L) in the AMD from abandoned coal mines.
Principal component analysis (PCA) was performed on the physicochemical indicators of all water samples, and the proportion of the principal components PC1 and PC2 was 67.3% and 14.7%, respectively, with a total proportion of 82%, indicating that these two principal components can represent the overall distribution of the physicochemical indexes of water samples (Fig. 3). The green and yellow ellipses represent the 95% confidence intervals of the distribution of the samples in group R and group M, respectively. The existence of crossover areas in the grouped ellipses of the two groups of samples indicates that there is some similarity in the physicochemical indexes of the samples, especially the samples R5 ~ R8 in the downstream of the river, which are closer to the samples in group M, indicating that the physicochemical properties are more similar between them. As shown in Fig. 3(a), the upstream sample R1 is the farthest away from the group M samples, which indicates that there is a significant difference between the physicochemical properties of the unpolluted river section and samples from the abandoned coal mines drainage. Upstream samples of the river were characterized as weakly alkaline due to the wide distribution of carbonate minerals. However, pH value of the river gradually changed from weakly alkaline to acidic with the gradual convergence of AMD from abandoned coal mines along the river. From Fig. 3(a), it is obvious that the distance between the downstream samples (R5 ~ R8) and the abandoned coal mine drainage samples is gradually close to each other, which indicates that the physicochemical properties between the downstream samples and the abandoned coal mine drainage gradually tend to be similar. Comparing with the surface water quality standard (GB3838-2002), it can be seen that the concentration of total Fe and SO42− in the middle and downstream samples of the river all exceeded the limits, except for the upstream samples R1 ~ R3. Especially for samples R4 and R5, with the confluence of M6 (pH 3.21) and M7 (pH 4.20), the concentration of total Fe and SO42− of the river increased rapidly from 1.86 mg/L to 100.00 mg/L, and 285.00 mg/L to 542.00 mg/L, respectively, while the pH value of the river also plummeted from 6.23 to 3.22. From the clustering of the samples in group M, samples with pH < 4.0 are more aggregated with each other, indicating that the physicochemical properties of the strongly acidic drainage were more similar. As demonstrated in Fig. 3(b), these strongly acidic drainage samples were mainly distributed in the negative half-axis of the PC1 axis, which indicates that their distribution was negatively correlated with pH and Na+ and positively correlated with other indicators. Samples with pH > 5.5 in group M (M1, M5 and M9) and river samples (group R) were all distributed in the positive half-axis of the PC1 axis, indicating that the distribution of these samples was positively correlated with pH and Na+.
We collected five sediment samples from the upstream to downstream of Pinglu River, and the physicochemical properties of the solid samples are shown in Table 1. As the pH value of the river rapidly decreased with the confluence of AMD from abandoned coal mines, resulting in a gradual decrease in the pH of the river sediment from weakly alkaline (pH > 8.0) to neutral (Table 1). The variation of pH in river sediment is significantly smaller than that of the river water samples, which may be due to the widespread distribution of carbonate rocks, where alkaline minerals in the riverbed neutralize part of the acidity of river, resulting in the pH value of downstream river sediment remaining neutral. The total Fe content of river sediment samples ranged from 31.36 to 63.60 mg/g (average 45.06 mg/g).
Table 1
Physicochemical properties of river sediment samples
Samples
|
pH
|
TOC
|
TP
|
TN
|
Fe
|
Mn
|
Zn
|
Cd
|
Cu
|
Cr
|
mg/g
|
mg/g
|
mg/g
|
mg/g
|
mg/kg
|
mg/kg
|
mg/kg
|
mg/kg
|
mg/kg
|
S1
|
8.00
|
12.68
|
0.55
|
1.76
|
37.52
|
459.00
|
143.00
|
1.04
|
33.10
|
188.00
|
S2
|
8.14
|
8.12
|
0.58
|
1.13
|
31.36
|
660.00
|
84.20
|
0.35
|
23.80
|
96.40
|
S3
|
7.78
|
13.90
|
0.34
|
0.85
|
55.16
|
417.00
|
77.90
|
0.91
|
17.90
|
90.60
|
S4
|
7.76
|
15.56
|
0.41
|
1.49
|
37.66
|
432.00
|
85.10
|
1.16
|
20.20
|
160.00
|
S5
|
7.16
|
19.95
|
0.17
|
1.72
|
63.60
|
440.00
|
140.00
|
1.13
|
20.90
|
81.10
|
Hydrochemical types of water samples collected from different regions
We collected a total of twenty water samples from different regions of the study area. Results of physicochemical properties showed that the pH values of the twelve abandoned coal mines drainage samples ranged from 2.78 to 7.57, and the concentration of the major cations Ca2+ 108.85 ~ 317.17 mg/L (average 226.56 mg/L), Mg2+ 126.26 ~ 133.33 mg/L (average 79.60 mg/L), K++Na+ 3.06 ~ 20.33 mg/L (average 12.04 mg/L), and the main anion SO42− ranged from 248.39 to 3430.77 mg/L with the average concentration 1954.29 mg/L, showing an overall characteristic of high Ca-Mg and high sulfate (Fig. 4(a)). According to the Shukarev classification (Zhou and Ye 2014), hydrochemical types of the abandoned coal mine drainage were all SO4-Ca·Mg types, except M1 which was SO4-Ca type (Table 2). The pH range of Pinglu River water samples was 3.22 ~ 7.65, Ca2+ 54.25 ~ 239.21 mg/L (average 115.79 mg/L), Mg2+ 13.3 ~ 39.29 mg/L (average 25.15 mg/L), K++Na+ 5.58 ~ 8.67 mg/L (average 7.324 mg/L), and the quality of the river showed high calcium and magnesium content characteristics. The concentration of the main anions SO42− ranged from 47.35 to 846.15 mg/L (average 393.83 mg/L) and HCO3− 0.00 ~ 152.55 mg/L (average 59.49 mg/L). The distribution of anions along the river showed that anions in the upstream samples of the river (R1 and R2) were dominant by HCO3−, while SO42− was the main anion in downstream samples (R3 ~ R8) (Fig. 4(b)). Shukarev classification calculation showed that the hydrochemical types of the Pinglu River from upstream to downstream were SO4·HCO3-Ca·Mg (R1), SO4·HCO3-Ca (R2 and R3), SO4-Ca (R4) and SO4-Ca·Mg (R5 ~ R8) in order (Table 2).
Table 2
Percentage of anion and cation equivalents and hydrochemical types of water samples
Samples
|
Ca2+
|
Mg2+
|
Na+
|
K+
|
HCO3−
|
SO42−
|
Cl−
|
Hydrochemical type
|
M1
|
77.81
|
20.58
|
0.39
|
1.22
|
0.56
|
98.33
|
1.11
|
SO4-Ca
|
M2
|
64.74
|
32.64
|
0.54
|
2.08
|
0.00
|
99.02
|
0.98
|
SO4-Ca·Mg
|
M3
|
62.28
|
35.80
|
0.40
|
1.51
|
0.00
|
99.39
|
0.61
|
SO4-Ca·Mg
|
M4
|
64.06
|
33.38
|
0.42
|
2.15
|
0.35
|
98.23
|
1.42
|
SO4-Ca·Mg
|
M5
|
70.18
|
28.18
|
1.14
|
0.50
|
21.99
|
75.82
|
2.20
|
SO4-Ca·Mg
|
M6
|
60.53
|
36.68
|
0.57
|
2.22
|
0.00
|
98.72
|
1.28
|
SO4-Ca·Mg
|
M7
|
63.58
|
35.96
|
0.34
|
0.12
|
0.00
|
99.65
|
0.35
|
SO4-Ca·Mg
|
M8
|
62.33
|
35.61
|
0.38
|
1.69
|
0.00
|
99.59
|
0.41
|
SO4-Ca·Mg
|
M9
|
60.44
|
38.45
|
0.32
|
0.80
|
0.00
|
97.80
|
2.20
|
SO4-Ca·Mg
|
M10
|
59.09
|
39.13
|
0.46
|
1.32
|
0.00
|
99.14
|
0.86
|
SO4-Ca·Mg
|
M11
|
59.27
|
39.31
|
0.33
|
1.08
|
0.00
|
99.05
|
0.95
|
SO4-Ca·Mg
|
M12
|
49.19
|
48.40
|
0.34
|
2.07
|
0.00
|
99.31
|
0.69
|
SO4-Ca·Mg
|
R1
|
65.19
|
26.64
|
6.51
|
1.66
|
69.48
|
25.38
|
5.14
|
SO4·HCO3-Ca·Mg
|
R2
|
71.25
|
24.68
|
2.72
|
1.35
|
51.29
|
41.54
|
7.17
|
SO4·HCO3-Ca
|
R3
|
76.20
|
20.53
|
2.24
|
1.03
|
30.04
|
65.23
|
4.73
|
SO4·HCO3-Ca
|
R4
|
75.07
|
22.43
|
1.40
|
1.11
|
8.88
|
88.16
|
2.96
|
SO4-Ca
|
R5
|
68.95
|
27.94
|
1.81
|
1.30
|
0.00
|
96.59
|
3.41
|
SO4-Ca·Mg
|
R6
|
69.13
|
28.78
|
0.96
|
1.14
|
0.00
|
96.81
|
3.19
|
SO4-Ca·Mg
|
R7
|
71.45
|
25.87
|
1.38
|
1.30
|
0.85
|
95.33
|
3.82
|
SO4-Ca·Mg
|
R8
|
71.47
|
26.26
|
1.16
|
1.11
|
0.00
|
97.70
|
2.30
|
SO4-Ca·Mg
|
Unit:%
With the confluence of abandoned coal mines drainage with low pH and high content of sulfate, hydrochemical types of the Pingly River gradually changed from SO4·HCO3-Ca·Mg to SO4-Ca·Mg type. Meanwhile, the total dissolved solids (TDS) and hardness of river gradually increased, with TDS increasing from 296.70 mg/L to 1083.54 mg/L, and the hardness of river (calculated by CaCO3 content) increasing from 190.24 mg/L to 553.74 mg/L. According to the classification of river hardness, the quality of river from upstream to downstream of Pinglu River gradually changed from medium hard to high hard. Villages and farmlands are distributed along the river, and most of the local farmlands were irrigated by water diverted from the Pinglu River. According to the calculation results of Aq-QA software, the irrigation risk of the upstream section of Pinglu River (R1~R4) was medium, while the section from R5 onwards has become high risk, which indicates that the downstream section of river was not suitable for agricultural irrigation. Irrigating farmland with such extreme hardness and mineralization river water can cause significant changes in soil properties, such as soil slumping or salinization, etc.
Revealing the evolution of ion concentration in different types of water samples
Atmospheric precipitation and groundwater rebound are the main sources of mine water in abandoned coal mines, so the local stratigraphic and hydrodynamic conditions can significantly affect the hydrochemical characteristics of mine water. The coal-bearing strata in the Yudong River basin belong to the Lower Permian Liangshan Formation, and the stratigraphic structure mainly consists of quartz sandstone, carbonaceous tuff, coal seams, and bauxite intercalated with pyrite agglomerates or nodules, and the thickness of iron-bearing ore layers below the coal seams is about 0 ~ 1.5m (Li 2018). After the abandonment of underground coal mines, the oxidized pyrite in the rock strata will produce a large amount of H+ and SO42−, resulting in the decomposition of HCO3− to CO2 under acidic conditions, so SO42− becomes the main anion in coal mine drainage, accounting for 75.82 ~ 99.65% (average 97.00%) of the equivalent percentage. As the comprehensive index reflecting water quality, the TDS of abandoned coal mine drainage ranged from 488.46 to 5125.87 mg/L with an average of 2909.37 mg/L, of which 66.67% (8 out of 12 water samples) exceeded the groundwater V quality standard. The fitted results (Fig. 5(a)) showed a significant positive correlation between TDS and SO42− in coal mine drainage (p < 0.05) with a correlation coefficient of 0.996, indicating that sulfate is the main factor affecting the overall water quality, which further elucidates that the oxidation of pyrite in abandoned coal mines is the main cause of AMD pollution in this area.
The cations in the coal mine drainage are mainly Ca2+ and Mg2+, with Ca2+ accounting for 49.19 ~ 77.81% (average 62.79%) of the equivalent percentage, Mg2+ 20.58 ~ 48.40% (average 35.34%), and Na++K+ ranging from 0.45 to 2.78%. Local carbonate rocks containing calcium and magnesium (such as dolomite and dolomitic limestone) are widely distributed (Wang et al. 2009). Therefore, the dissolution and weathering of such kind of carbonate rocks under natural conditions is the primary source of Ca2+ and Mg2+ in coal mine drainage. Although Ca2+ and Mg2+ are the principal cations in coal mine drainage, their distribution in different coal mine drainage samples varies. For example, the percentages of Ca2+ and Mg2+ in the M1 were 77.81% and 20.58%, respectively, while they were 49.19% and 48.40% in M12 (Table 1). Ratios of Ca/Mg in M1 and M12 samples were 6.30 and 1.69, respectively. Results showed that the Ca/Mg ratio in coal mine drainage was significantly negatively correlated with the concentration of SO42− (p < 0.05), i.e., the Ca/Mg ratio decreased with the increase of SO42− concentration (Fig. 5 (b)). This is probably because the solubility product of CaSO4 is smaller than that of MgSO4, and when the water contains a large amount of SO42− and Ca2+, Ca2+ will first reach saturation and form precipitation, thus leading to the decrease in the proportion of Ca2+, while on the contrary the percentage of Mg2+ will increase, eventually leading to the decrease of the Ca/Mg ratio. The saturation indices of the water samples were calculated by PHREEQC, and the calculation results showed that the saturation indices of Anhydrite (CaSO4) and Gypsum (CaSO4.2H2O) in the water samples of group M were all exceeded zero, indicating a supersaturated state, while the saturation indices of Halite (NaCl) and Sylvite (KCl) were all below zero, indicating a dissolved state (Table 3).
Table 3
Mineral saturation index of water samples collected from abandoned coal mines
Sample
|
M1
|
M2
|
M3
|
M4
|
M5
|
M6
|
M7
|
M8
|
M9
|
M10
|
M11
|
M12
|
CaSO4
|
1.42
|
1.79
|
1.85
|
1.49
|
0.89
|
1.53
|
1.64
|
1.92
|
1.15
|
1.86
|
1.87
|
1.95
|
CaSO4·2H2O
|
1.71
|
2.06
|
2.11
|
1.77
|
1.19
|
1.82
|
1.93
|
2.18
|
1.44
|
2.13
|
2.14
|
2.20
|
NaCl
|
-7.11
|
-6.51
|
-6.79
|
-6.77
|
-6.93
|
-6.60
|
-7.29
|
-6.90
|
-7.06
|
-6.52
|
-6.60
|
-6.61
|
KCl
|
-5.84
|
-5.25
|
-5.56
|
-5.33
|
-6.45
|
-5.28
|
-7.01
|
-5.60
|
-5.91
|
-5.40
|
-5.41
|
-5.18
|
Among the cations in river samples, Ca2+ was absolutely dominant, with its equivalent percentage ranging from 65.19 to 76.20% (average 71.09%), Mg2+ from 20.53 to 28.78% (average 25.39%), while Na++K+ together accounted for the least (2.10 to 8.17%, average 3.52%). Dissolved weathering of the widely distributed carbonate minerals was the direct source of Ca2+ and Mg2+ (Wang et al. 2009), thus making the cations in the river water to be dominated by "Ca and Mg". The concentration of Ca2+ and Mg2+ in the water samples from upstream to downstream of Pinglu River increased linearly, where Ca2+ increased from 54.25 mg/L (R1) to 162.63 mg/L (R8) and Mg2+ increased from 13.30 mg/L (R1) to 35.86 mg/L (R8). Previous investigations showed that the total annual discharge of abandoned coal mines accounted for about 43.26% of the total river flow, and the confluence of such coal mine discharges with high calcium and magnesium concentrations (average 226.56 mg/L and 79.60 mg/L, respectively) significantly increased the calcium and magnesium contents in the river. In addition, the confluence of large amounts of acidic coal mine drainage significantly reduced the pH value of the river, which further accelerated the dissolution weathering of carbonate minerals such as dolomite and calcite. Correlation analysis showed that the concentrations of Ca2+ and Mg2+ in the river were significantly negatively correlated with pH (p<0.05) with correlation coefficients of -0.860 and -0.886, respectively, which further indicated that the changes of hydrochemical characteristics in Pinglu River were mainly related to the confluence of abandoned coal mine drainage along the river. The Ca/Mg ratios of river samples ranged from 4.00 to 6.19. Unlike lower Ca/Mg ratios of abandoned coal mine drainage under acidic conditions (all less than 3.0), the Ca/Mg ratios of the downstream section (R5~R8) of the river all exceeded 4.0 despite their lower pH values. Statistical analysis showed no significant correlation between Ca/Mg ratio and the concentration of SO42- in river samples. The anions in river samples were mainly dominated by HCO3- and SO42-, and their distribution on the upstream and downstream river was significantly different. HCO3- was the principal anion in the upstream sample R1, accounting for 69.48% of the equivalent percentage. Since HCO3- is easily to be decomposed under acidic conditions, its equivalent percentage in the water samples gradually decreases with the decrease of pH values, resulting in its percentage at R2 and R3 has been reduced to 51.29% and 30.04%, respectively, especially in the downstream samples, the percentage of HCO3- has been reduced to less than 1%. In contrast, the equivalence percentages of SO42- in the river gradually increased from upstream to downstream, with their equivalent percentage in the upstream samples R1 and R2 being 25.38% and 41.54%, respectively, while they were all >95% in downstream river samples (R5 to R8).
Overall bacterial community taxonomic compositions of river sediment
A total of 254,678 valid reads were defined in five sequencing libraries with an average length of 421 bp. Based on the 97% similarity threshold, all sequences were finally clustered into 2,844 operational taxonomic units (OTUs), ranging from 1758 to 2430. The highest number of OTUs was obtained from the upstream sample R1, and the number of OTUs obtained in river sediment gradually decreased from upstream to downstream, especially at R5, the most polluted point, was only 1758. The alpha diversity index Chao1 and Phylogenetic diversity also showed the same tendency (Fig. 6), which indicated that the richness of microorganisms in the river sediments gradually decreased from upstream to downstream. This may be due to the fact that the confluence of acidic mine drainage along the Pinglu River changed the living conditions of microorganisms and inhibited the survival of some of them. The Shannon index of sample R5 was significantly lower than the other samples, indicating the lowest bacterial diversity.
A total of 40 phyla were identified from the sediment samples. The relative abundance of the top 15 most dominant phyla are shown in Fig. 7(a). Among these samples, Chloroflexi was the most abundant phylum, constituting in average of 23.93% of the total sequence reads (11.53 ~ 37.76% in all samples). Proteobacteria, Actinobacteriota and Acidobacteriota ranked as the following three most abundant phyla, accounting for 21.45% (17.19 ~ 26.96% in all samples), 17.38% (11.72 ~ 31.39% in all samples), and 14.23% (9.60 ~ 18.01% in all samples) of the total reads, respectively. Phyla with relative abundance > 5% in at least one sample included Desulfobacterota and Firmicutes with the average relative abundance of 3.28% (0.62 ~ 5.83% in all samples) and 2.97% (0.60 ~ 5.39% in all samples), respectively. The remaining 34 phyla accounted for about 16.76% of the total reads. The distribution of the dominant phylum varied among samples, except for the downstream S5 sample, where the dominant phylum was Actinobacteriota (31.39% relative abundance) and the subdominant phylum was Proteobacteria (26.96% relative abundance); the dominant phylum from S1 to S4 was Chloroflexi (20.61%~37.76% relative abundance) and the subdominant phylum was Proteobacteria (19.95%~22.44% relative abundance). The relative abundance of the dominant phylum Chloroflexi in the sediments from upstream to downstream of the river showed an increase followed by a decrease. At the class level, a widely variety of classes dominated. Among them, the most abundant classes were Alphaproteobacteria, Vicinamibacteria, Gammaproteobacteria, Anaerolineae, Actinobacteria, and Thermoleophilia, accounting for 11.52%, 10.00%, 9.93%, 9.12%, 6.28% and 6.14% of the total reads, respectively.
At the genus level, the detailed information about the relative abundance of the top 35 genera is indicated in Fig. 8. Clearly, the dominant genera in the sediment samples included Gaiella, Geobacter, Sphingomonas, MND1 and Anaeromyxobacter etc., with the average relative abundance of 2.65%, 1.19%, 1.06%, 0.99% and 0.92%, respectively. The clustering of the relative abundance of the top 35 genera in the samples is shown in Fig. 7(b), where red represents the genera with higher relative abundance in the samples, and blue for genera with lower relative abundance. Results showed that the distribution of these 35 top genera along the river clustered in separate groups. Genera with high abundance (relative abundance > 1.0%) in the upstream sample S1 included Geobacter, Marmoricola, Ellin6067, Phycicoccus and Nocardioides, mainly belonging to the phylum Actinobacteriota and Proteobacteria. Geobacter had the highest percentage (3.82%) among all the major genera of S1 samples, which are one kind of heterotrophic iron-reducing bacterium (Islam et al. 2005). Marmoricola is the second most abundant genus in S1 (3.35%), which is a Gram-positive and aerobic bacterium with the suitable pH range of 6 ~ 9, belonging to the member of the family Nocardioidaceae (Schumann et al. 2018). Anaeromyxobacter and Arthrobacter were genera with relatively high abundance in sample S2, where Anaeromyxobacter is an aryl-halorespiring facultative anaerobic myxobacterium, and some studies have demonstrated that bacteria of this genus have the ability to reduce iron and U(VI) metal. For example, some species can grow with the mineral hematite, an insoluble form of ferric iron, as electron acceptor (Chao 2011). Arthrobacter is one kind of the basic bacterium in soil and some species of this genus have the potential to remediate polluted environments by reducing hexavalent chromium and degrading agricultural pesticides, as well as by their ability to remediate nitrogen (Fu et al. 2014). The distribution of major genera in sample S3 was somewhat analogous to that of S2. Nordella, Clostridium_sensu_stricto_1, Hyphomicrobium and Acidaminobacter were more abundant in S3 and S2 than in the other samples, and the other genera with relatively high abundance in S3 also included Bryobacter, Candidatus_ Alysiosphaera, Pedomicrobium and FFCH7168. Pedomicrobium belongs to class Alphaproteobacteria, and acts as a terrestrial polarophilic bacterium. This kind of bacteria can inhabit particularly harsh environments and the strain Pedomicrobium manganicum has the ability to oxidize manganese and is usually used in many bioremediation processes, such as the removal of manganese from water purification systems (Zhao et al. 2019). Haliangium, Citrifermentans and Nitrospira were relatively abundant in the sample S4. Among them, Haliangium, a rod-shaped myxobacterium, commonly found in the ocean, being a class of an obligate halophile (Sun et al. 2016b). The major genera in S5 differed greatly from the other samples, and the genera with relative abundance > 1% included Gaiella, Sphingomonas, MND1, Pseudolabrys, Bradyrhizobium, Bacillus, etc. These genera mainly belong to the Proteobacteria. Previous studies have demonstrated that the abundance of Proteobacteria increases in soil contaminated by AMD (Chen et al. 2021b), and the relative abundance of Proteobacteria was also highest in the extremely contaminated downstream sample S5 in our study with the relative abundance of 26.96%. Gaiella, the dominant genus in river sediment, had the highest abundance in the sample S5 (relative abundance 6.16%) and is an aerobic heterotrophic bacterium (Albuquerque et al. 2011). In conclusion, most of the microorganisms in the sediment of Pinglu River are heterotrophic, and some genera can participate in the metabolic processes of Fe, Mn and N, which have the potential to remediate the polluted environment.
Differences of microbial communities in river sediments of different section
The similarity of microbial communities in river sediment was assessed using cluster analysis. As shown in Fig. 9, the clustering analysis revealed that the bacterial communities of the samples could be clustered into two groups. The first group contains upstream samples S1 to S3, and the second group includes downstream samples S4 and S5. General taxonomic pattern in Fig. 9 was driven primarily by differences in abundance of the major taxonomic groups.
As shown in Fig. 8, genera such as Geobacter, Anaeromyxobacter, Marmoricola and Phycicoccus, accounted for relatively high properties in upstream samples, and Gaiella, MND1 and Pseudolabrys appeared to be more abundant in the downstream samples. The phylogenetic distance among the top 30 abundant OTUs and their significant differences in upstream and downstream samples was also shown in Fig. 10. These 30 major OTUs were mainly distributed in Proteobacteria and Actinobacteriota with a proportion of 50% (15 out of 30) and 30% (9 out of 30), respectively, while other phyla also included Desulfobacterota, Firmicutes and Myxococcota. Figure 10 shows both the variation in the distribution of these 30 OTUs in upstream and downstream samples, where blue represents upstream samples and yellow for downstream samples. There are 14 top OTUs mainly distributed in upstream samples, of which five of them belong to Proteobacteria (35.71%), five to Actinobacteriota (35.71%), two to Desulfobacterota (14.28%), one to Firmicutes (7.14%) and one to Myxococcota (7.14%). The other 16 top OTUs were mainly distributed in the downstream samples, of which ten belonged to Proteobacteria (62.5%), four to Actinobacteriota (25%), one to Desulfobacterota (6.25%) and one to Myxococcota (6.25%). Core genera of the upstream samples were mainly attributed to Proteobacteria and Actinobacteriota, while for downstream samples, were mainly attributed to Proteobacteria. Significant difference analysis based on Welch's t-test showed that significantly different genera for upstream samples were Phycicoccus (OTU1130) and Ellin6067 (OTU1354), while Pseudolabrys (OTU5585) and Dongia (OTU2735) were the most different genera in the downstream samples.
Phenotypic prediction of microbial communities in river sediments
To gain insight into the functions of the bacterial community, we used BugBase, a bioinformatics tool that can infer the phenotypes of the entire community based on 16S rRNA gene (Lucas et al. 2018). A total of nine phenotypes were predicted for the microbial communities of the river sediment samples, including Aerobic, Anaerobic, Contains Mobile Elements, Facultatively Anaerobic, Forms Biofilms, Gram Negative, Gram Positive, Potentially Pathogenic and Stress Tolerant, and the relative abundance distribution of each phenotype in samples was predicted (Fig. 11(a)). Results of the BugBase prediction revealed that the microorganisms in river sediment samples were predominantly Gram-negative bacterium (average relative abundance 71.40%), while Gram-positive bacterium only accounted for 28.60% of the total community. In terms of oxygen utilization by microorganisms, the river sediment was mostly composed of aerobic microorganisms with the average relative abundance of 51.63%, while anaerobic microorganisms only accounted for 19.37% of the community. Comparing the distribution proportion of microorganisms in samples along the river (Fig. 11(b)), we found that the abundance of anaerobic microorganisms in river sediment showed a gradual decrease from upstream to downstream, from 24.77% (S1) to 12.46% (S5), while the abundance of aerobic microorganisms was basically stable from upstream to midstream, and gradually increased from midstream to downstream, from 47.50% (S3) to 60.79% (S5). To clarify the reasons for the differences between aerobic and anaerobic phenotypes in samples, we statistically analyzed the contribution of the major genera to the aerobic and anaerobic phenotypes in samples, as shown in Fig. 11(c) and Fig. 11(d), respectively. It can be seen that the aerobic phenotypes in samples were mainly contributed by genera Gaiella, Marmoricola, Sphingomonas, Anaeromyxobacter, Bryobacter and Nocardioides, with the average relative abundances of 1.84%, 1.48%, 1.14%, 1.12%, 0.82% and 0.81%, respectively. Among them, Gaiella and Sphingomonas were mainly distributed in downstream samples, while Marmoricola, Anaeromyxobacter, Bryobacter and Nocardioides were mainly distributed in upstream samples. The anaerobic phenotypes in samples were mainly contributed by genera Geobacter, Citrifermentans, Luteitalea and Vicinamibacter, with the average relative abundances of 1.34%, 0.31%, 0.26% and 0.18% in samples, respectively. Among them, Geobacter, Luteitalea and Vicinamibacter were mainly distributed in upstream samples, while Citrifermentans was mainly distributed in downstream samples.
Influence of physicochemical properties on microbial communities of river sediment
The distribution of microbial communities and diversity in sediments was closely related to environmental factors. Results of the Partial Mantel test based on Bray Curtis distance algorithm showed that the microbial communities in sediment samples were mainly attributed to pH, TOC and TP with correlation coefficients of 0.736, 0.607 and 0.564, respectively, and the absolute values of correlation coefficients between other physicochemical factors and microbial communities were all below 0.50. Figure 12 showed the correlation coefficients between the main genera and alpha diversity indicators in samples and the three main environmental factors, with red representing positive correlation and blue for negative correlation. Correlation coefficients of community richness index Chao1 with pH, TOC and TP was 0.849, 0.636 and 0.813, respectively, which was significantly positively with the other two indices except TOC (p < 0.05). The microbial richness of river sediment gradually decreased from upstream to downstream, which was consistent with the trend of pH and TP. The community diversity index Shannon showed significant positive correlations (p < 0.05) with pH, TOC and TP, with correlation coefficients of 0.914, 0.782 and 0.782, respectively. The distribution of Geobacter, Anaeromyxobacter, Marmoricola, Ellin6067, Phycicoccus, Arthrobacter and Ilumatobacter among the main sediment genera was positively correlated with pH and TP, and these genera were mainly were mainly distributed in the upstream samples. The richness of these genera gradually decreased from upstream to downstream as the pH and TP reduced in river sediment. In the downstream of the river, these genera were gradually replaced by Gaiella, Sphingomonas, MND1, Pseudolabrys, Pedomicrobium, Bradyrhizobium and mle1-7, which showed negative correlations with pH and TP.