Local-Scale Damming Impact on the Planktonic Bacterial and Eukaryotic Communities in the Upper Yangtze River

Dam construction and reservoir formation alters hydro-morphology of rivers, thereby restructuring microbial communities and biogenic element cycles in river ecosystems. The ecological responses and mechanisms of planktonic communities showed notable changes upstream and downstream of dams. Yet, less is reported about how the ecological mechanisms structuring planktonic communities at the closest area upstream and downstream of dams. In this study, we hypothesized that planktonic communities remained the connectivity or similarities but show distinctive ecological responses to changing environment at the closes area upstream and downstream of dams. Three large dams in the upper Yangtze River were chosen in the study. Field data revealed that the alpha diversity indexes slightly increased downstream of the dams. In addition, more eukaryotic ASVs solely occurred downstream of the dams, indicating that a large proportion of eukaryotes was formed downstream of the dams. Co-occurrence network analysis demonstrated that the keystone species of planktonic bacteria and eukaryotes decreased downstream of the dams, and the modularity increased. The robustness of the co-occurrence relationships among the eukaryotic communities was more strongly inuenced by these dams than that among the planktonic bacteria. The variance partitioning analysis results indicated that dam-related variables and local environmental variables mainly shape the assembly of the planktonic microbial communities closest to the dams. In conclusion, dams exert a greater impact on planktonic eukaryotes than on bacteria in near-dam areas, and planktonic bacteria can better adapt to changing environments. Our study provides a better understanding of the ecological effects of river damming.


Introduction
Currently, many rivers are highly regulated for hydropower or water supply purposes, which generates major hydrological disturbances at the spatial scale [1]. The river continuum concept (RCC) proposed by Vannote et al. [2] considered that biological communities can be characterized as establishing a temporal continuum of synchronized species replacements, and the physical variables within a stream system exhibit a continuous gradient. However, few riverine ecosystems remain free-owing along their entire to greatly differ in their physical and chemical environments, and the composition and structure of the microbial community correspondingly undergo tremendous changes [1,[7][8][9].
Planktonic bacterial and eukaryotic communities are essential members of aquatic ecosystems with an extremely high level of genetic diversity. Both are important participants in the global biogeochemical cycle [1,10]. Planktonic bacteria can assimilate and remineralize inorganic nutrients, which are channeled to higher trophic levels via the predation of protists [8, 11,12]. The spatial and temporal distributions of microbial processes in freshwater ecosystems may vary depending on environmental variables [13,14], temperature [15], hydrological factors [16,17], grazing pressure [14], and changes in land use [18], etc., and the microbial diversity and community structure respond to changes in these environmental variables. Planktonic microbial biogeographic patterns and assembly mechanisms have been widely studied in large rivers, such as the River Thames Basin [19], Danube River [20], Mississippi Rivers [21,22] and Yangtze River [9,23]. These publications mostly emphasized distance-decay relationships and microbial community changes on a large catchment scale [8]. They considered that dam construction on a river often leads to notable changes in microbial communities at dam-affected sites [7,24,25]. In addition, the bacterial taxa in sediments downstream of a dam are drastically reduced due to severe riverbed scouring [9]. Our previous results also found that dams would signi cantly reduce the α-diversity of planktonic bacterial communities on the large-scale catchment, and the microbial communities/species would be conducive to recovery in river habitats. River damming often causes a sharp rise/decline in physical or hydrological variables, nally causing natural biophysical gradient discontinuities in local environments [3]. However, it remains unclear whether these discontinuities caused by damming would directly or indirectly alter microbial communities and how microbes respond in these local-scale changing habitats. There still remains a knowledge gap regarding the impacts of damming on the transformation of microbial communities downstream of large dams.
At present, correlation network analysis has been widely applied to understand the organization of microbial communities and the interaction between their components in aquatic ecosystems. According to the topological characteristics of the network, potential target species (keystone species) can often be identi ed at the central node of the network, and these species play a more important role in the network, while their appearance or disappearance can disturb a mature community [26][27][28][29]. A particular bacterium in the network may adhere to a particular ecological (or life) strategy [30]. Life strategies represent sets of correlated traits attributed to physiological or evolutionary tradeoffs, and in-plant communities, tradeoffs in key tness traits have been represented through the conceptual r-and K-selection theories [30][31][32]. The classi cation of life strategies has been applied in microbial systems. For example, studies in the Thames River Basin and Lancang-Mekong River Basin indicated that along with the river network from upstream to downstream, the dominant phylum of bacteria shifted from r-strategists (Bacteroidetes) to Kstrategists (Actinobacteria) [19,25]. In particular, the network interaction and ecological strategies can characterize the response mechanisms of keystone species in these dynamic environments.
It is noteworthy that that large dams are not absolute barriers that disconnect the upstream reservoir and downstream river reach. Waters traveling through dams for hydropower production or other functions will bring aquatic microbial community to downstream river reach. Distinctive habitats shape different community assembly mechanisms in the upstream and downstream of the dam, projecting the potential damming impacts on riverine aquatic ecosystems. Compared to the extensive studies on larger scales, we highlighted our study focusing on local scale, i.e. the nearest and most accessible sampling sites upstream and downstream of the dams. We hypothesized that there would be possible links of aquatic microbial community between the sampling sites upstream and downstream of the dam due to the dam operation, shaping the ecological mechanisms of microbial assembly in damming rivers. We tried to explore the different responses of eukaryotic and bacterial communities on such local scale. We believe our study would provide new insights of aquatic microbial ecology in damming rivers.
Here, we selected three large dams in the upper Yangtze River (Xiluodu Dam, Xiangjiaba Dam, and Three Gorges Dam) as the research area. Both 16S rRNA and 18S rRNA high-throughput gene sequencing techniques were applied to investigate the bacterial and eukaryotic communities at sampling sites nearest upstream and downstream of these three dams in May, July, and November 2019. This study aimed to 1) examine the changes in microbial communities and diversity right downstream of these large dams; 2) explore the keystone species and natural connectivity of co-occurrence networks; 3) evaluate the potential damming effect on the composition of planktonic bacterial and eukaryotic communities on local scale.

Study area and sample collection
The upper Yangtze River starts from the Qinghai-Tibet Plateau in the west and ends at Yichang, Hubei in the east. The entire basin covers an area of approximately 1M km 2 . The river reach encompasses large drops, rapid currents, many canyons, holding a large economically viable potential of hydropower. At present, the world's largest hydropower plant is built in the upper Yangtze River. The Three Gorges Dam (TGD, with installed capacity of 22500MW), Xiluodu (XLD, 12600MW) Dam and Xiangjiaba (XJB, 6400MW) Dam are the selected dams in this study (Fig. 1). The XLD Dam and XJB Dam are two large cascade dams located in the upstream of the Three Gorges Reservoir (TGR). Combined with the Wudongde (10200MW) and Baihetan (16000MW) hydropower stations upstream of the XLD and XJB Dams, the total installed capacity of the four cascade dams is equivalent to twice that of the Three Gorges Reservoirs. These projects focus on power generation, and offer comprehensive bene ts such as ood control, shipping and alleviation of the sedimentation problem in the Three Gorges Reservoir [33].
Information on the parameters of the selected dams in this study is listed in Table S1.
These dams exert a profound impact on the aquatic microbial community structure in the upper Yangtze River. To investigate the impact of these dams on the planktonic bacterial and eukaryotic communities, eld sampling work were carried out in May, July and November 2019 at the most accessible sampling sites right upstream and downstream of the XLD, XJB and the TGD. Five vertical layers at different depths were sampled upstream of the dams, and one surface water sample downstream of the dams was collected at the selected sampling sites. Each sample was obtained at the main channel. A total of 54 samples were obtained, including 45 samples upstream of the dams and 9 samples downstream of the dams. Sample information is provided in Table S2. A total of 5 L of water was collected in a clear polypropylene container at each site, which was thoroughly mixed in a bucket and immediately transported at a low temperature ranging from 0 ~ 4°C. The water samples were then subsampled for DNA-based analysis and physiochemical measurement. A total of 800 mL water was ltered through membrane lters with a 0.22-µm pore size (Millipore, Bedford, MA) to capture microbial cells. Thereafter, the obtained membrane lters were kept frozen at -86°C until DNA extraction.

Physicochemical analysis of the samples
The water temperature (T), dissolved oxygen (DO), pH, and conductivity along the vertical pro le were measured in situ with a multiparameter water quality analyzer (YSI-EXO2, USA). The on-site relative humidity was measured with a hygrometer. Chlorophyll a (Chl-a) was extracted with a Whatman GF/C lter over 24 h and 90% acetone, centrifuged at 3,000 rpm for 10 min, and then spectrophotometrically quanti ed. Dissolved organic carbon (DOC) was ltered with a What man GF/F glass ber lter

Bioinformatics analysis of the sequences
The remaining sequences were quality ltered and chimeras were identi ed and ltered with the QIIME 2 pipeline (version 2) [37]. All bioinformatics analyses were based on amplicon sequence variants (ASVs) [38]. Individual ASVs were taxonomically classi ed with a 99% identity threshold via the open-reference method (VSEARCH) based on the 16S rRNA genes of ke r isolates as a reference [39]. The taxonomies of nonke r isolate ASVs were subsequently assigned via comparison to the SILVA 138/16S_bacteria database and the SILVA 138/18S_eukaryota database using BLAST under default settings [40].
Surface water was collected for network analysis to reveal the co-occurrence pattern of the planktonic bacterial and eukaryotic communities. To reduce the complexity of correlation matrix generation, only the ASVs occurring in all samples with ≥ 50 sequences were retained for the planktonic bacteria, while the ASVs containing ≥ 20 sequences were retained for the planktonic eukaryotes. To visualize the associations in the networks, the possible pairwise Spearman rank correlation matrix based on the ASV level was calculated with the igraph package in RStudio. Only high Spearman's correlation coe cients (|r|>0.8) and statistically signi cant (p < 0.05) correlations were accepted for network analysis. Gephi software (version 0.9.2) was employed to visualize the network and calculate the overall topological characteristics (the number of nodes, edges, degree, betweenness centrality, modularity, etc.). The node size was determined according to the degree (the number of edges connected to each node) in the network, and the color of the edge was related to the correlation (the pink and green lines indicated positive and negative correlations, respectively). The harbored keystone taxa of microbial communities drive the community composition and function [41]. Keystone species were identi ed based on the same topological features of the co-occurrence networks, which exhibit a high mean degree, closeness centrality, high transitivity, and betweenness [42]. In this study, nodes with degree (> 40) and betweenness centrality (< 1000) were selected as the keystone species. The natural connectivity of a complex network was applied to reveal the robustness of the different microbial association networks to random node removal ("attack") [43][44][45].

Statistical analysis
Alpha diversity indexes (Sobs, ACE, Shannon, etc.) and dissimilarity-related calculations of the planktonic bacterial and eukaryotic communities in all samples were calculated in RStudio using the vegan package. The Shannon diversity index upstream and downstream of the dams was visualized with a boxplot (Origin 2018). To examine the difference in microbial diversity between the different dams from upstream to downstream, a one-way analysis of variance (ANOVA) was performed in SPSS Statistics 23.0 (IBM). Nonmetric multidimensional scaling analysis (NMDS) based on the Bray-Curtis distance was adopted to visualize the dissimilarity in the planktonic bacterial and eukaryotic communities among the samples (beta diversity). Based on the Bray-Curtis dissimilarity in the species considering 999 permutations, analysis of similarity (ANOSIM) statistics were generated to detect the signi cant differences in the microbial communities between upstream and downstream dam areas and between reservoirs. A stacked histogram of the species composition was analyzed in Rstudio. To determine the differences in the composition of the planktonic bacterial and eukaryotic ASVs upstream and downstream of the dams, a fan chart of the percentage of upstream_s, coexisting, and downstream_s species was generated based on ASV-level Venn diagram results. The linear relationships between various environmental factors and the abundance of keystone species were analyzed via Pearson correlation analysis in R v4.0.3. Moreover, variation partitioning analysis (VPA) was performed to evaluate the relative contribution of the nutrient variables, dam-related variables, and local environmental variables in shaping the microbial communities. Variance in ation factor analysis (VIF) was employed to screen the relevant variables, and those factors with low collinearity were retained for the subsequent analysis (VIF ≤ 10

Environmental variables of the selected samples under reservoir operation
The environmental variables of the water samples are listed in Table S3. Microbial taxonomic composition, distribution, and alpha diversity analysis Total amounts of 1152360 and 1399140 sequences and 4054 and 2985 ASVs were detected for the planktonic bacteria and eukaryotes, respectively. The taxonomic composition and distribution of the planktonic bacterial and eukaryotic communities are shown in Fig. 2a and b. The abundant bacterial phyla inclued Actinobacteria (32.3%), Proteobacteria (29.1%), Cyanobacteria (13.2%), Bacteroidetes (11.6%), Acidobacteria (4.2%), Planctomycetes (2.4%) and Chloro exi (1.6%), which contributed more than 90% to all the communities. Generally, there was no signi cant change in the taxonomic composition of the planktonic bacterial community. However, the relative abundances of Actinobacteria and Acidobacteria increased downstream of the dam, while Proteobacteria and Bacteroidetes revealed the opposite patterns (Fig. 2a). The eukaryotes mainly comprised protozoans, metazoans, and fungi. The metazoans were dominated by Bilateria (31.8%, including Arthropoda, Rotifera, Nematoda, etc.), the protozoans were dominated by Stramenopiles (29.6%, including Bacillariophyceae and Chrysophyceae, etc.) and Alveolata (12.3%, mainly dominated by Ciliophora), and among the other eukaryotic communities, Cryptophyceae is dominated (7.8%). The changes in the eukaryotic communities were more obvious. The relative abundance of Cryptophyceae signi cantly increased downstream of the XLD Dam, and Alveolata signi cantly increased downstream of the TGD, while Bilateria signi cantly decreased (Fig. 2b). In addition, the ASVs were classi ed into upstream of the dams solely (upstream_s) species, downstream of the dams solely (downstream_s) species, and coexisting species upstream and downstream of dams (coexisting) (Fig. 2c). In the planktonic bacterial communities, more than half of the ASVs (50.04%) coexisted upstream and downstream of the dams. However, the percentage of the ASVs coexisting in the eukaryotic communities was lower than that in the planktonic bacterial communities. Most eukaryotic ASVs solely existed after the dams (40.00%), indicating that the new eukaryotic species were formed downstream of dams (Fig. 2c).
The richness and diversity were assessed based on ASV analysis by calculating the Sobs and Shannon indexes. The alpha diversity indexes of both the planktonic bacterial and eukaryotic communities exhibited an increasing trend downstream of the dams, but there were no signi cant differences between the upstream and downstream dam sites (Fig. 3a, b, d, and e). In addition, the richness of the microbial communities among the three dams was compared, and the Sobs richness index of the planktonic bacterial and eukaryotic communities in the TGD was signi cantly higher than that in the XLD and XJB Dams ( Fig. 3c and f).

Beta diversity of the planktonic microbial communities upstream and downstream of the dams
The NMDS results based on the Bray-Curtis distance indicated that there was no obvious geographic clustering of the planktonic bacterial and eukaryotic communities between the different locations (upstream and downstream of the dams) and between the different dams ( Fig. 4a and b). In addition, analysis of the similarity between the groups (ANOSIM) con rmed that there were no signi cant differences in the microbial communities upstream and downstream of the different dams (bacteria: r = 0.209, p = 0.001; eukaryotes: r = 0.076, p = 0.042). This phenomenon indicated that the communities were similar upstream and downstream of the dams. Moreover, vertical distance decay relationships were estimated for the microbial communities upstream of dams ( ve layers). The vertical distance decay relationships signi cantly increased with increasing vertical distance (Fig. S1).
Co-occurrence networks of the microbial communities upstream and downstream of the dams Only the surface water samples were analyzed to perform co-occurrence network analysis upstream and downstream of the dams. The topological parameters of the bacterial and eukaryotic community cooccurrence networks upstream and downstream of the dams are listed in Table 1. Notably, Fig. 5 shows that most topological parameters (the modularity, nodes, and average path length) increased downstream of the dams. From upstream to downstream of the dams, the modularity of the planktonic bacterial networks was higher than that of the eukaryotic communities. The number of positive edges was larger than that of negative edges, especially in the eukaryotic community network (96.98% and 95.36% positive edges upstream and downstream of the dams, respectively). The community networks in the surface water samples exhibited no signi cant changes downstream of the dams. There were signi cant differences in the node-level topological features (closeness centrality, betweenness centrality, and Eigen centrality) between the planktonic bacterial and eukaryotic communities (Fig. S2). In addition, the robustness analysis results indicated that the natural connectivity of the microbial network downstream of the dams decreased faster than did the connectivity upstream of the dam (Fig. 5e and f).
There was a signi cant difference in the stability of the network of the planktonic eukaryotic community (p < 0.001) upstream and downstream of the dams, indicating that the network upstream of the dam was more fragile. In the planktonic bacterial communities, a total of 49 ASVs was identi ed as keystone species, and the main taxa belonged to Proteobacteria (21 ASVs), Cyanobacteria (9 ASVs), Acidobacteria (4 ASVs), and Planctomycetes (4 ASVs). A total of 16 ASVs was identi ed as keystone species in the eukaryotic community networks, which were predominantly a liated with Stramenopiles (7 ASVs), Alveolata (3 ASVs), and Choano agellata (3 ASVs). The keystone species of the bacterial community decreased downstream of dams, and the keystone species of the eukaryotic communities even disappeared, indicating that the eukaryotes were greatly affected by the dams. In addition, the relative abundance of the keystone species was generally higher in November, which may be related to environmental conditions (Fig. 6).
Effects of dam-related and local environmental variables on the microbial communities water level were the signi cant variables of the eukaryotic community composition (Fig. S3, Table S4). In particular, DOC and water level have signi cant effects on both the planktonic bacterial and eukaryotic communities (p = 0.001). The important variables were divided into two groups (dam-related variables and local environmental variables). The VPA results showed that the local environmental variables were the most important factors driving the microbial community assembly. They explained 24.3% and 22.5% of the variation in the planktonic bacterial and eukaryotic communities, respectively (Fig. 7). However, 68.4% and 67.8%, respectively, of the variation remained unexplained.

Discussion
The well-known RCC describes the riverine physical variables of the natural river system (including width, depth, speed, etc.) under a continuous gradient of conditions, and the response of biotic communities can who can rapidly adapt to such gradients changes in the environment [2]. Yet, when concerning to damming rivers, SDC serves as conceptual basis of discontinuities within the river continuum [46]. Such disturbances structured reciprocal conceptual model of abiotic environmental variables caused by damming and biophysical properties that inherently hold by microbial assemblages. Here, we extended the discussion of our ndings in the research into the following three aspects: Possible connectivity of microbial communities between upstream and downstream dam areas Dams are not absolute barriers that disconnect upstream and downstream sampling sites. Indeed, we found similarities or even possible connections of microbial communities between the upstream and downstream sampling sites. The beta diversity of both planktonic bacterial and eukaryotic communities based on the Bray-Curtis dissimilarity matrix indicated that the differences of microbial communities between upstream and downstream sampling sites are not apparent (Fig. 4). Such possible differences among the three dams were also not identi able.
In addition, it was noted that coexisting ASVs in the planktonic bacterial and eukaryotic communities and accounted for a large proportion (Fig. 2c). Among these coexisting species, the top 10 ASVs at the different dams were selected for clustering at the genus level (Fig. S5). The results demonstrated that the dominant genera shared by the three dams were CL500-29_marine_group, hgcI_clade, and norank_f_Vicinamibacteraceae. The rst two genera belonged to Actinobacteria, and the latter genus belonged to Acidobacteria. Actinobacteria is typical riverine planktonic bacteria [47,48], mainly following free-living lifestyles, and they may not be easily affected by particle sinking-induced species loss in rivers [24,49]. Cyanobium_PCC-6307 (belonging to Cyanobacteria) attained a higher relative abundance in the XLD and XJB Dams, while this organism was not dominant in the TGD, which may be related to the operating conditions of the reservoir. Among the eukaryotic communities, Mediophyceae (belonging to Stramenopiles) and Arthropoda (belonging to Bilateria) were dominant, both of which are also common eukaryotes in freshwater ecosystems [29,50,51].

Local environmental variables structured microbial communities
The establishment of dams shifts the environment from lotic to lentic, thereby affecting the water quality, and cascade dams may also exert an impact on the local microbial community due to their cumulative effect [52,53]. The environmental heterogeneity caused by dam construction may largely affect the composition of the microbial community [24,52]. We found that local environmental variables directly (Table S4) Table S4). The reduced velocity caused by dam construction promotes the sinking and sedimentation of particulates, which further leads to signi cant changes in the suspended particle content and composition in river waters [54], and the changes in the particulate content in uence the proportion of microorganisms [11]. The reduction in the concentration of suspended particles also leads to a reduction in the electrical conductivity and DOC concentration, and the DOC concentration exerts a signi cant impact on the microbial community. The VPA results further support the above ndings. Consistent with the results of most previous studies [8, 25,55], the local environmental variables played a dominant role in community assembly (Fig. 7), and which could explain more of the variation in the bacterial communities.
Moreover, some research reported that the alpha diversity of planktonic bacteria decreased from upstream to downstream [20,24]. One explanation of this result is that the planktonic bacteria from thereon towards river mouths are affected by process of external input and mass effects [20], and the other is that changes in water chemistry properties lead to changes in bacterial communities through a process of species sorting [24]. In this larger-scale study, the decreasing importance of the "riparian in uence" leads to a reduction in the alpha diversity of planktonic eukaryotes with the river continuum [8]. Yet, in our study, we did not nd such phenomenon (Fig. 3a, b, d, and e). It can simply be justi ed as follows: rst, the distinctive habitat between the upstream and downstream sampling sites. Our research only focuses on dams at the local scale (a range of several kilometers or no more than ten kilometers), which different from the most previous studies. Under the background of this local scale, the barrier of the dam has caused the difference in the habitat conditions of the upstream and downstream of the dam areas; Second, different habitats form unique local environmental variables. Changes in variables such as the pH, DO, and nutrient variables may increase the adaptability of microorganisms to the environment, leading to an increase in microbial diversity within a small area. In summary, the impact of river damming on the microbial community structure requires further research.
Different eco-physiological responses of eukaryotic and bacterial communities Planktonic bacteria and eukaryotes are different groups with different eco-physiological adaptive mechanisms. In general, planktonic bacteria can adapt well to fast changing environments, e.g. lotic system, as a result of their small size and rapid growth rates [56]. In our study, Actinobacteria accounted for a large proportion both upstream and downstream of the dams (Fig. 2a). Actinobacteria represent various ecotypes or exhibit a pronounced ecophysiological plasticity [57]. Certain aquatic Actinobacteria are better protected from protestant grazing than are other heterotrophic planktonic bacteria due to their relatively small cell size and speci c cell wall structure [58,59]. Changes in the taxonomic composition of the planktonic eukaryotic communities were more affected by the dams, especially Alveolata, which signi cantly increased downstream of the TGD (Fig. 2b). Ciliophora (belonging to Alveolata) is an important consumer of picophytoplankton and bacteria [60], and its biomass is correlated with the dynamics of phytoplankton blooms [26].
Response of keystone species projected the eco-physiological differences between planktonic bacterial and eukaryotic communities. The relative abundance of the keystone species was generally high in November (Fig. 6). November is the storage period of the reservoir [61]. The rise in water level and the increase in hydraulic residence time alter the living environment of microorganisms and lead to changes in certain species [11,62]. The keystone species in the microbial networks did not exhibit signi cant differences upstream and downstream of the dam (Fig. S6). Yet, they exhibited different survival strategies, caused by their distinctive functional traits, including resource acquisition and utilization traits, constitute the basis for determining the community structure and diversity and de nes the diverse ecological strategies selected in different environments [63-65]. Bacteroidetes achieved a higher relative abundance upstream of the dam, while Actinobacteria revealed the opposite trend (Fig. 2a, Fig. S6). It was reported that Bacteroidetes are r-strategists, and their level of competition upstream is lower, which facilitates the growth of this species that can quickly utilize resources [19,25,66]. In contrast, Actinobacteria is a K-strategist with an advantage in the intense competition for survival in the downstream, and they tend to attain a lower growth rate and a narrower niche [19,25,66]. Actinobacteria is also considered a defense strategy against bacteria because of their very small cell morphology, and their increase downstream might indicate an increase in enhanced grazing pressure on downstream bacteria [1].
Rivers are altered from a lotic environment to a semantic or lentic ecosystem because of the interception effect of dams [5]. The survival strategies of microorganisms are almost always the sum of the diverse elementary behavioral reactions to complex and changeable environments, and it is di cult to classify these strategies clearly [30]. Hence, further research on the survival strategies of microbial communities in rivers is needed.     and eukaryotic (f) co-occurrence networks.

Figure 6
Bubble diagram of the keystone species in the networks upstream and downstream of the dams. Variation partitioning analysis (VPA) of the planktonic bacterial and eukaryotic communities based on the dam-related variables and local environmental variables.