Metabolic Functional Community Diversity of Associated Bacteria during the Degradation of Phytoplankton from a Drinking Water Reservoir

In the drinking water reservoir ecosystem, phytoplankton and bacteria play important roles in shaping freshwater health and function. In this work, the associated bacterial community functional diversity during degradation of phytoplankton was determined using the substrate utilization profiling (BIOLOG) technique, meanwhile, the composition and concentration of phytoplankton were examined using a microscope. The results indicated that Euglena decreased 58.33% from 0 to 38 d, while the smallest degradation of Bacillariophyta was 20.19%. Average well color development (AWCD590nm) increased during the static periods from 0 to 38 d; however, the AWCD590nm of 18 and 38 d had no significant difference (p < 0.05). The Simpson’s index (D) was in accordance with Shannon’s diversity (H) and species richness(S); it was measured to be18 > 38 > 5 > 0 d. There were significant differences in the pattern and level of carbon sources used by the phytoplankton-associated bacteria. In addition, the principle component analyses (PCA) suggested that the first principle component (PC1) and the second principle component (PC2) explained 46.76% and 21.49% of the total variation for bacterial community, respectively. Redundancy analysis (RDA) revealed that cell abundance of phytoplankton was negatively correlated with the AWCD590nm, amino acids and other functional indexes. Therefore, the data suggest that there are differences in the phytoplankton-associated bacterial community functional diversity during different static stages of water samples collected from the drinking water reservoir.


Introduction
Phytoplankton and bacteria play important roles in maintaining the structure and function of aquatic ecosystems [1,2]. In the past few decades, the relationships between phytoplankton and bacteria were widely evaluated, including mutualism, parasitism and competition [3][4][5]. In the mutualistic relationships, heterotrophic bacteria utilize the excretion from phytoplankton as carbon sources [6][7][8]. Different phytoplankton community compositions lead to differences in kinds of organic carbon which the microbial community can utilize for their growth [9]. Compared with the massive literatures focused on succession of phytoplankton and bacteria community composition in different water bodies [10,11], the functional diversity of associated bacteria during degradation of the phytoplankton from drinking water reservoir is not well determined, although bacteria species have vital services involvement in organic carbon decomposition in this freshwater ecosystem [12].
The community-level physiological profiles technique called BIOLOG was successfully used to determine the bacterial community metabolic fingerprints in a drinking water reservoir [13]. Table 1. Sole carbon substrates in BIOLOG used in the present work [22,23]. According to the method used by Zhu with little modifications [15], briefly, onaclean bench, water samples were filtrated through a 1.2 µm pore size membrane (Isopore Membrane Filters, Millipore). Then, 150 µL of filtrate was added into each well of the ECO micro-plate using an electronic pipette [15]. The inoculated ECO micro-plate was then incubated at 28 ± 2 • C in a dark chamber (Jinghong, Shanghai, China) for 240 h [23,24]. The absorbance at 590 nm was determined using an Elisa reader (BIOLOG Company, Hayward, CA, USA). Bacterial community activity in ECO micro-plate was expressed as average well color development (AWCD), and was expressed as the following formula: where R was the value of blank control, C i was the value of carbon substrates well, and negative optical density (OD 590nm ) values were set to zero, C i − R ≥ 0 [14]. Indexes of bacterial community functional diversity were showed as species richness (S), Simpson's diversity (D) and Shannon's diversity (H).
S was the number of utilized carbon sources well in the ECO plate, and AWCD > 0.2 represents that the carbon substrate has been utilized [25]: H and D were calculated as the equations below [14,25]: where In this work, the data of 120 h incubation [13] was used for AWCD 590nm , community diversity indexes, carbon substrate utilization, principle component analysis (PCA) and redundancy analysis (RDA).

Redundancy Analysis (RDA) of Phytoplankton Community Composition and Associated Bacterial Community Metabolic Fingerprints
To investigate whether attributes of the phytoplankton accounted for significant changes in the functional diversity of the associated bacterial community, RDA was employed to reveal the bacterial metabolic fingerprints with phytoplankton community characteristics as explanatory variables.
The following parameters were used as descriptors of the phytoplankton community: Shannon's diversity index (H p ); total number of phytoplankton cells L −1 (N); and the relative abundance of Melosira (M). Shannon's diversity index for the phytoplankton community (H p ) was calculated as described by Rooney-Varga [11]. The data of 120 h of AWCD, H, S, D and the six kinds of carbon substrates were taken for response variables, which represented the associated bacterial community metabolic fingerprints.

Statistical Analysis
Data statistical analysis was carried out using the Sigma Plot (Version 16.0) software package for windows. A parametric one-way analysis of variance (ANOVA) test followed by Tukey-Kramer HSD tests was used. Principle component analysis (PCA) was performed using SPSS version 18.0 software for windows (SPSS Inc., Chicago, IL, USA), and the first two PC1 and PC2 were selected. CANOCO for Windows (Version 4.5) was used to conduct redundancy analysis (RDA) ordinations.

Water Quality
The results of the water quality parameters monitored are as shown in Table 2. Temperature is an important factor affecting growth of the phytoplankton [26,27]. Though the environmental temperature already reached 25 • C, the water temperature was only 12 • C. The nitrogen to phosphorus (N:P) ratio of the water sample was 43.According to Redfield's law, when the ratio of total nitrogen to total phosphorus exceeds 16: 1, phosphorus is considered to be the limiting factor [20]. The concentration of Chl-a, TN and TP were 1.54 µg/L, 0.98 mg/L and 0.012 mg/L, respectively. Based on the eutrophication evaluation criteria for Chinese lakes and reservoirs [18], the water sample was considered to be moderately eutrophicated.

Phytoplankton Cell Concentration and Composition
Phytoplankton play an important role in maintaining the stability of aquatic ecosystems as primary producers, where by abundance and community composition can directly influence and indicate the water quality [28,29]. A total of three phyla and seven genera or species of phytoplankton were identified, with Bacillariophyta having the largest number of species, reaching to 11.33 × 10 4 cells /L at the beginning, and Melosira was dominant species in JINPEN reservoir in spring (Figures 1 and 2).
The degradation rate of Euglena was fastest, and decreased 58.33% from 0 to 38 d, while the smallest degradation of Bacillariophyta was 20.19% ( Figure 1). In addition, Ankistrodsemus was degraded, but Cyclotella increased slightly ( Figure 2). It was suggested that Ankistrodsemus decomposed easier, whereas Cyclotella had great vitality and was hard to degrade. It may be connected to their physiological characteristics.      As shown in Table 3 Table 3).  As shown in Table 3, the Shannon's diversity (H p ) of phytoplankton decreased from 0 d (1.479) to 5 d (1.432), and increased slightly in 18 d, then reduced in 38 d with 1. The relative abundance of Melosira (M) was ordered as: Table 3).

Bacterial Community Metabolic Profiles
The associated bacterial community metabolic activity (AWCD 590nm ) significantly improved with degradation of phytoplankton ( Figure 3).
The associated bacterial community metabolic activity (AWCD590nm) significantly improved with degradation of phytoplankton ( Figure 3). As shown in Figure 3, the AWCD590nm increased during the degradation periods from 0 to 38d. The highest AWCD590nm (1.107) was found in 38 d, and the lowest in 0 d at0.694.
One-way analysis of variance showed that AWCD590nm of degraded periods was significantly higher than the beginning (0 d); however, AWCD590nm of 18 and 38 d had no significant differences (p < 0.05). These results revealed that the ability to utilize carbon substrates for the associated bacterial community increased with degradation of phytoplankton, indicating that metabolic activity of the bacterial community enhanced. Note: Data are expressed as the mean values ± SD (n=3).The same capital letter after the data represents no significant difference by Tukey-Kramer HSD (p < 0.05).  As shown in Figure 3, the AWCD 590nm increased during the degradation periods from 0 to 38 d. The highest AWCD 590nm (1.107) was found in 38 d, and the lowest in 0 d at 0.694.
One-way analysis of variance showed that AWCD 590nm of degraded periods was significantly higher than the beginning (0 d); however, AWCD 590nm of 18 and 38 d had no significant differences (p < 0.05). These results revealed that the ability to utilize carbon substrates for the associated bacterial community increased with degradation of phytoplankton, indicating that metabolic activity of the bacterial community enhanced.
The significant highest species richness (S) was found in 18 d, and the lowest was in 0 d (F = 12.62, p < 0.05). The Shannon's diversity (H) in 18 d was 3.297 ± 0.007, which was significantly higher than that of 0 d with 3.130 ± 0.037 (F = 19.71, p < 0.05) ( Table 4). The Simpson's index (D) was in accordance with Shannon's diversity (H) and species richness(S), it was measured that 18 > 38 > 5 > 0 d. However, there were no significant differences for those three diversity indexes among 5, 18 and 38 d ( Table 4). The most important reason for this phenomenon is that most organic carbon was released from the cell of algae, and more carbon sources were used by the bacterial community during 18 d. Note: Data are expressed as the mean values ± SD (n = 3). The same capital letter after the data represents no significant difference by Tukey-Kramer HSD (p < 0.05).
As shown in Table 5, the utilization of amino acids, amines, polymers, phenolic compounds, carbohydrates and carboxylic acids by the phytoplankton-associated bacteria were different with different degradation time. Carbohydrates were most utilized by the associated bacteria at 38 d, and amino acids were most utilized by the associated bacteria at 18 d, respectively. Carboxylic acids were most metabolized at 5 d (Table 5). Principal components analysis (PCA) suggested that the significant bacterial community functional metabolic profile discrimination existed among different degradation times ( Figure 4). As shown in Figure 5, the first two principles explained 68.25% of the total variance. PC1 and PC2 explained 46.76% and 21.49% of the variance, respectively. It was suggested that the 18 and 38 d associated bacterial community functional diversity was more unstable than that of the 5 d water sample (Figure 4). Therefore, these data revealed that the metabolic fingerprints of the associated bacterial community changed with degradation of phytoplankton. The higher discrimination of carbon substrates in the principle component analysis of the data of carbon source utilization is shown in Table 6, including D,L-a-glycerol, 2-hydroxy benzoic, y-hydroxybutyric acid, L-threonine and glycyl-L-glutamic acid. Note: The data represent the means and standard errors (n = 3). The same capital letter after the data represents no significant difference by Tukey-Kramer HSD (p < 0.05). The significant highest species richness (S) was found in 18 d, and the lowest was in 0 d (F=12.62, p < 0.05). The Shannon's diversity (H) in 18 d was 3.297±0.007, which was significantly higher than that of 0 d with 3.130±0.037 (F=19.71, p < 0.05) ( Table 4). The Simpson's index (D) was in accordance with Shannon's diversity (H) and species richness(S), it was measured that 18 > 38 > 5 > 0d. However, there were no significant differences for those three diversity indexes among 5, 18 and 38 d ( Table 4). The most important reason for this phenomenon is that most organic carbon was released from the cell of algae, and more carbon sources were used by the bacterial community during 18 d.
As shown in Table 5, the utilization of amino acids, amines, polymers, phenolic compounds, carbohydrates and carboxylic acids by the phytoplankton-associated bacteria were different with different degradation time. Carbohydrates were most utilized by the associated bacteria at 38 d, and amino acids were most utilized by the associated bacteria at 18 d, respectively. Carboxylic acids were most metabolized at 5d (Table 5).
Principal components analysis (PCA) suggested that the significant bacterial community functional metabolic profile discrimination existed among different degradation times ( Figure 4). As shown in Figure 5, the first two principles explained 68.25% of the total variance. PC1 and PC2 explained 46.76% and 21.49% of the variance, respectively. It was suggested that the 18 and 38 d associated bacterial community functional diversity was more unstable than that of the 5 d water sample ( Figure 4). Therefore, these data revealed that the metabolic fingerprints of the associated bacterial community changed with degradation of phytoplankton. The higher discrimination of carbon substrates in the principle component analysis of the data of carbon source utilization is shown in Table 6, including D,L-a-glycerol, 2-hydroxy benzoic, y-hydroxybutyric acid, L-threonine and glycyl-L-glutamic acid.       Figure 5 shows the influence of phytoplankton community variables on the functional diversity of phytoplankton-associated bacteria. The first axis of the ordination explained 95.1% of the total variance, while the second axis, an additional 3.3%. The first ordination axis had the highest eigenvalue (0.951); therefore, the changes along the first ordination axis of phytoplankton community will have the greatest impact on the functional diversity of the associated bacteria. N was negatively correlated with the AWCD, amino acids and other functional indexes. M was positively correlated with the carboxylic acids, S and H. From the intersample distances, it can be seen that the associated bacterial community functional diversity of 18 and 38 d was similar ( Figure 5).

Carbon Substrates Located in BIOLOG ECO
Several works have shown the interactions between phytoplankton and bacterial communities, and the dynamics of phytoplankton community composition have been observed to correlate with changes in bacterial community composition [11,30,31]. Rooney-Varga [11] investigated the  Figure 5 shows the influence of phytoplankton community variables on the functional diversity of phytoplankton-associated bacteria. The first axis of the ordination explained 95.1% of the total variance, while the second axis, an additional 3.3%. The first ordination axis had the highest eigenvalue (0.951); therefore, the changes along the first ordination axis of phytoplankton community will have the greatest impact on the functional diversity of the associated bacteria. N was negatively correlated with the AWCD, amino acids and other functional indexes. M was positively correlated with the carboxylic acids, S and H. From the intersample distances, it can be seen that the associated bacterial community functional diversity of 18 and 38 d was similar ( Figure 5).

Correlation between the Phytoplankton Community and Associated Bacterial Community Metabolic Fingerprints (RDA)
Several works have shown the interactions between phytoplankton and bacterial communities, and the dynamics of phytoplankton community composition have been observed to correlate with changes in bacterial community composition [11,30,31]. Rooney-Varga [11] investigated the relationship between phytoplankton and bacterial community dynamics in ocean environment conditions, and suggested that species composition shifts in the attached bacteria and phytoplankton communities were correlated. There is also evidence that differences in the quality of organic carbon, produced by different types of phytoplankton, cause changes in the structure composition of bacterial communities utilizing this organic carbon [32]. Therefore, changes in phytoplankton community composition may influence the composition of bacterial communities that function as part of the microbial loop. However, the effect of the changes in phytoplankton community composition on bacterial metabolic functional diversity was not well understood.
In the present work, we determined whether relationships between bacteria and phytoplankton community dynamics existed. It had been suggested that the AWCD values increased with degradation of phytoplankton, and bacterial metabolic activities increased, indicating that carbon sources released by phytoplankton increased. This result is consistent with a study conducted by Sun et al. [33], who determined release of colloidal and particulate nutrition in the course of decomposition of cyanobacteria, which showed colloidal organic carbon reached five times that of the beginning. Tranvik [34] also examined the effects of organic matter on the growth of bacteria in lake water, and suggested that colloidal organic matter served as a supplementary source of nutrition that could promote the growth of bacteria and protists. Whereas, the associated bacterial diversity indexes (H, S and D) increased from 0 to 18 d, and decreased at 38 d. The reason may be that the organic matter released by phytoplankton differed with different degradation days. Sun et al. [33] revealed that the content of organic carbon increased firstly and then decreased in the course of decomposition of cyanobacteria. Dickerson et al. [1] used the BIOLOG method to determine spatial and temporal community-level physiological profiles for three fresh water lakes of different trophic levels, and found bacterial communities utilized the carbon guilds similarly between sites within the three lakes. When the metabolic profile of each lake was compared, Lake Bradford and Moore Lake were more similar to one another than to Lake Munson, the eutrophic lake. The reason most important was that, with increased nutrient loading, bacterial activities increase, while bacterial diversity decrease.
Sarmento et al. [35] used micro autoradiography to quantify the preferences of the heterotrophic prokaryote on dissolved organic carbon derived from phytoplankton species, and found that the vast range of different types of organic molecules available in the sea selects and maintains the high levels of diversity described for marine bacterioplankton. In this study, we found that there was a significantly different carbon source utilization pattern in associated bacterial community. PCA revealed differences in associated bacterial community functional diversity among different stages; it showed that the associated bacterial community changed significantly throughout the degradation time. In order to explain the relationships between the phytoplankton and bacterial community further, we used RDA with phytoplankton community characteristics as explanatory variables ( Figure 5). RDA results indicated an association between phytoplankton and associated bacterial community functional diversity that would be expected if specific bacteria-phytoplankton interactions occurred. Amin et al. [36] teased apart a bacterial consortium associated with a globally-distributed diatom, and found that a Sulfitobacter species promotes diatom cell division via secretion of the hormone indole-3-acetic acid, synthesized by the bacterium using both diatom-secreted and endogenous tryptophan. Dang et al. [37] found marine bacteria were known to be colonizers of particulate matter and were likely to be utilizing organic compounds provided by phytoplankton cells [11]. Hold et al. [38] observed that different bacterial assemblages were associated with different din flagellate species, suggesting species-specific interactions. Liu et al. [2] used quantitative PCR and 454 pyrosequencing methods to investigate bacterial communities composition affected by phytoplankton community succession in a drinking water reservoir, and found that the distinct succession of phytoplankton community could mediate the temporal dynamics of the bacterial community in the Tingxi Reservoir. Some advanced molecular techniques, such as stable isotope probing (SIP), have been used to detect the interaction between bacteria and algae [39,40] and identify the bacteria responsible for contaminant biodegradation [41][42][43]. Denitrifying bacterial communities (such as nirS-type), explored by using Illumina sequencing, will also be used to identify the nitrogen-cycling metabolic microbe [44]. Meanwhile, SIP and ecological network analysis [45] could be used to detect the associated bacteria during degradation of phytoplankton in a drinking water reservoir in future.

Conclusions
Although many studies have shown the interactions between phytoplankton and bacterial communities, little is known about how these communities interact at the species composition and functional levels in drinking water reservoir ecosystems. In this work, light microscopy and BIOLOG techniques were used to reveal the correlation between phytoplankton and associated bacterial communities, and the metabolic fingerprints of associated bacteria related to the degradation of phytoplankton were determined. The results suggested that AWCD values of associated bacteria increased with degradation time of phytoplankton. While the associated bacterial community diversity index increased in the early stages and then decreased, the highest bacterial diversity index was observed in 18 d. More organic carbon released from the cell of algae can be utilized by bacterial community in 18 d. Principal components analysis revealed a significant difference in the associated bacterial community functional metabolic profiles among different static times of phytoplankton. Redundancy analysis indicated that the total number of cells of phytoplankton was negatively correlated with the metabolic function of the associated bacteria. The results from this work suggested that decomposed phytoplankton could influence the metabolic activity of associated bacteria in a drinking water reservoir.