Differentiation strategies for planktonic bacteria and eukaryotes in response to aggravated algal blooms in urban lakes

Aggravated algal blooms potentially decreased environmental heterogeneity. Different strategies of planktonic bacteria and eukaryotes in response to aggravated algal blooms. Environmental constraints of plankton showed different patterns over time. Protecting ecological health of waterbodies for drinking, agricultural, and industrial utilization caters to one of the current global sustainable development goals (SDG, https://sdgs.un.org/goals) [1]. Many measures (e.g., dredging) have been taken to promise good waterbody quality by depleting nutrients (e.g., phosphorus and nitrogen) availability [2–4]. However, algal blooms, especially harmful cyanobacterial blooms caused mainly by eutrophication, occur seasonally and periodically in inland and coastal aquatic ecosystems despite strict control of external nutrient loading [5, 6]. Previous studies have reported that algal blooms, for example, Microcystis blooms, threaten planktonic diversity by releasing secondary metabolites (e.g., algal toxins) and consuming oxygen [5, 7]. Therefore, it is of great importance to disentangle how plankton respond to algal blooms and factors controlling water trophic state. Typically, the key parameters (e.g., chlorophyll-α [Chl-α] plus total phosphorus, total nitrogen, and/or chemical oxygen demand) are used to reflect and predict the trophic level of waterbody [8]. Nutritional factors (e.g., nitrogen and phosphorus) and nonnutritional factors (e.g., temperature and turbidity) influence Chl-α content [9, 10]. For example, reducing phosphorus level can mitigate cyanobacterial density in a hyper-eutrophic lake [9]. Microorganisms mediate nitrogen and phosphorus transformation in waters and sediments, in particular in the absence of oxygen [11, 12]. Bacteria (e.g., cyanobacteria and actinobacteria) and eukaryotes (e.g., phytoplankton, zooplankton, and protozoa) are important components of aquatic food webs [13], and planktonic interaction affects biomass of Chl-α-containing plankton and planktonic diversity [14, 15]. For example, actinomycete display algicidal effect on Alexandrium tamarense [14], and flagellates can degrade toxic Microcystis sp. by producing functional compounds (e.g., peroxiredoxin and phosphatase) [15]. Therefore, deciphering water trophic state is essential to investigate the content of Chl-α relying on the biomass of Chl-α-containing organisms (e.g., prokaryotic cyanobacteria and eukaryotic algae) [8]. Planktonic diversity drives multinutrient (e.g., nitrogen and phosphorus) cycles [16]. However, whether planktonic diversity could potentially affect the water trophic state is poorly understood during aggravated algal blooms. Most studies prefer to investigate community composition and diversity of plankton in different aquatic environments [17–19]. However, it remains largely unknown about mechanisms underlying planktonic diversity maintenance in response to aggravated algal blooms. Many attempts have been made to simultaneously estimate the diversity maintenance of planktonic prokaryotes and eukaryotes by assessing species presence–absence and abundance, ecological assembly processes, and species coexistence patterns [20–23]. In a community, species presence–absence and abundance can be interpreted by species replacement and abundance difference [20]. Ecological assembly processes include determinism (e.g., species sorting) and stochasticity, with the former imposed by abiotic and biotic factors and the later induced by random events (e.g., birth, death, and/or drift) [24, 25]. Both environmental factors and environmental heterogeneity adjust the balance between stochasticity and determinism [25, 26]. For instance, salinity is the major determinant in shaping community assemblies of bacterioplankton in the Yellow River Estuary [26] and microeukaryotes in urban reservoirs [21]. Environmental heterogeneity determines bacterioplankton community assembly processes (i.e., homogeneous selection and heterogeneous selection) and thus governs community turnover and coexistence patterns in the Paraná River [25]. Coexistence patterns, reflected by co-occurrence network, can infer species interactions [27, 28]. According to co-occurrence networks, species can be identified as mutualistic or antagonistic based on positive or negative interactions, as well as hub species balancing community stability [25]. Ecological assembly processes and coexistence patterns of planktonic prokaryotes and eukaryotes are reported for many aquatic environments (e.g., rivers, lakes, and reservoirs) [17, 21, 25, 26, 29], but have been insufficiently studied in urban lakes suffering from massive algal blooms. In this study, we chose 12 representative urban lakes located in Wuhan City (Supporting Information Table S1 and Figure S1) and we collected water samples in April, May, and June to follow algal bloom development. According to a criterion of defining algal blooms with Chl-α threshold of 40 μg/L [30], algal blooms of these lakes were aggravated (i.e., June > May > April [83.29 ± 67.11 μg/L]; Supporting Information: Figure S2). Here, we aim to (i) explore distribution patterns, species replacement and abundance differences, ecological assembly processes, and coexistence patterns of bacteria and eukaryotes in response to aggravated algal blooms, and (ii) elucidate abiotic and biotic factors affecting water trophic state. Considering that the water physicochemical properties change periodically [17, 22, 31], we hypothesized that distribution patterns, ecological assembly processes, and coexistence patterns of both bacteria and eukaryotes differ among sampling months. Because bacteria and eukaryotes are different organisms with different living styles [13, 16], we hypothesized that bacteria and eukaryotes would display the opposite environmental constraint to aggravated algal blooms. We measured water physicochemical properties and conducted Illumina MiSeq sequencing of bacterial and eukaryotic communities to address our research objectives and verify our research hypothesis. We found different responses from bacterial and eukaryotic communities to aggravated algal blooms. Water physicochemical properties varied with sampling month (Supporting Information: Figure S2), showing significant differences in nitrate nitrogen, chemical oxygen demand, calcium, magnesium, iron, electrical conductivity, pH, temperature, and dissolved oxygen (p < 0.05). No significant differences were found in turbidity, total phosphorus, soluble reactive phosphorus, total nitrogen, and ammonia nitrogen among the 3 months (p > 0.05). The Chl-α content (2.21–564.27 μg/L) notably increased from April to June, and more than 60% lakes displayed algal blooms in three sampling months (i.e., April, 66.7%; May, 70.4%; and June, 77.8%). There were significant increases in trophic lake index (TLI) of these lakes from April (37.11–81.55; mesotrophic-hypereutrophic) to June (40.35–100.16; mesotrophic-hypereutrophic) (p < 0.05; Supporting Information: Figure S2). Only turbidity was significantly correlated with Chl-α and TLI in all 3 months (p < 0.05 or p < 0.01 or p < 0.001; Supporting Information: Table S2). Environmental heterogeneity was significantly higher in April than in May and June (p < 0.05; April > May > June) (Supporting Information: Figure S3). These results indicate that there were aggravated algal blooms and decreased environmental heterogeneity over time. Absolute abundances of bacteria and eukaryotes significantly increased from April to June (Supporting Information: Figure S4), and were differently correlated with physicochemical factors (Supporting Information: Table S3). For instance, eukaryotic abundances were notably correlated with turbidity in April (r = 0.518, p < 0.01), May (r = 0.506, p < 0.01), and June (r = 0.567, p < 0.01). Both bacterial and eukaryotic community compositions exhibited distinct differences between the 3 months (Supporting Information: Figure S5). A total of 10,511 bacterial amplicon sequence variants (ASVs) and 4487 eukaryotic ASVs were found in 3 months, and they shared 2080 and 783 ASVs, respectively. Bacterial communities were dominated by Proteobacteria (37.22%–59.82%), Actinobacteria (18.78%–33.85%), Bacteroidetes (5.65%–12.86%), Firmicutes (1.51%–5.45%), Deinococcus-Thermus (0.32%–3.86%), Cyanobacteria (3.08%–14.70%), and Verrucomicrobia (0.56%–1.25%) (Supporting Information: Figure S5). The relative abundance of Cyanobacteria increased from April to June, and Cyanobacteria was significantly positively correlated with Chl-α in 3 sampling months (p < 0.05 or p < 0.001; Supporting Information: Table S4). In contrast, eukaryotic communities were dominated by Chlorophyta (7.69%–20.42%), Rotifera (11.43%–13.30%), Arthropoda (8.59%–14.02%), Chytridiomycota (0.44%–5.16%), Dinophyceae (2.50%–6.02%), Chrysophyceae (0.71%–5.63%), Bacillariophyta (1.48%–7.64%), and Streptophyta (0.32%–1.21%). Relative abundances of Dinophyceae and Bacillariophyta increased from April to June. However, Chlorophyta rather than other plankton was significantly correlated with TLI in 3 sampling months (p < 0.05 or p < 0.001; Supporting Information: Table S4). Nonmetric multidimensional scaling (NMDS) plots showed significant differences in bacterial (pairwise analyses of similarity [ANOSIM], R = 0.563, p < 0.001) and eukaryotic (ANOSIM, R = 0.160, p < 0.001) community composition over time. Significant distance-decay relationships (DDRs) were found for bacteria and eukaryotes between the 3 months (p < 0.05 or p < 0.01 or p < 0.001), but most fitness values were relatively low at taxonomic and phylogenetic levels (R2 < 0.1) (Supporting Information: Figure S6), indicating a weak decay of community similarity with geographical distance. The β-diversities of bacteria and eukaryotes revealed significant differences between the 3 months at taxonomic and phylogenetic levels (Supporting Information: Figure S6). Except in June, water physicochemical factors explained more on community compositional variation for eukaryotes than bacteria based on redundancy analysis (RDA) results (Supporting Information: Figure S7). Based on permutational multivariate analysis of variance (PERMANOVA) results, turbidity showed significant effects on the community composition of bacteria and eukaryotes in 3 months (Supporting Information: Figure S8). These results indicate that there are distinct shifts in distribution patterns of bacteria and eukaryotes over time, and turbidity showed important roles in shaping community composition of bacteria and eukaryotes. By disassembling planktonic taxonomic β-diversity (Figure 1), we found species replacement showed comparable effects on compositional dissimilarity than richness difference. The ratio of species replacement to compositional dissimilarity (Repl/D) of bacterial community (mean values of Repl/D: April, 0.9724; May, 0.9684; June, 0.9666) significantly decreased over time, but opposite for eukaryotic community (mean values of Repl/D: April, 0.9318; May, 0.9379; June, 0.9771) (p < 0.05; Supporting Information: Figure S9). These results indicate the opposite change in trend of species replacement of bacteria and eukaryotes. Unexpectedly, taxonomic α-diversity represented by Shannon–Wiener index of bacterial and eukaryotic communities was notably lower in April than in May and June (Supporting Information: Figure S4). This result suggests a distinct increase in planktonic diversity along aggravated algal blooms. Mantel correlograms consistently displayed noticeable positive correlations across short phylogenetic distances for bacteria and eukaryotes along environmental gradients from April to June (Supporting Information: Figure S10). Except for the eukaryotes in May, there were also significant negative correlations across short and intermediate phylogenetic distances of bacteria and eukaryotes. The phylogenetic distance covers a significant phylogenetic signal varied from 10% to 40% of the maximum phylogenetic distance within each phylogenetic tree. These results reflect significant phylogenetic signals of bacteria and eukaryotes at short phylogenetic distances along environmental gradients during aggravated algal blooms. Subsequently, we evaluated community assemblies of bacteria and eukaryotes over time (Figure 2). Dispersal limitation (31.34%–66.95%), variable selection (10.26%–58.12%), and “undominated” processes (3.42%–40.74%) showed large effects on community assemblies of bacteria and eukaryotes in the 3 months, whereas homogeneous selection (0.57%–3.99%) and homogenizing dispersal (0.28%–1.71%) displayed limited influences in 3 months (Figure 2A). Deterministic (58.67% for May; 60.40% for June) and stochastic (41.31% for May; 39.60% for June) processes balanced bacterial community assembly. Stochastic processes showed major influences on the community assemblies of bacteria in April (69.52%) and eukaryotes in all months (77.21%–87.75%). Generally, differentiating (54.70%–93.17%) rather than homogenizing (1.71%–4.56%) processes dominated the community assemblies of both bacteria and eukaryotes (Figure 2A). These results indicate that stochastic and deterministic processes affected bacterial community assemblies, whereas solely stochastic processes determined eukaryotic community assemblies during aggravated algal blooms. The ratio of bacterial sorting to dispersal limitation increased over time, but showed an opposite for eukaryotes (Figure 2B). Bacterial habitat niche breadth decreased with time, but the opposite for eukaryotes (Figure 2C). Bacterial migration rates (m value) derived from Sloan neutral model decreased with time, whereas an opposite pattern was found for eukaryotes (Figure 2D). These results indicate that bacteria are more environmentally constrained with time, that is, during aggravated algal blooms, and thus behaved the opposite to eukaryotes. We constructed co-occurrence networks to decipher the coexistence patterns of bacteria and eukaryotes (Supporting Information: Figure S11 and Table S5). More nodes were found in bacterial and eukaryotic communities in June than in April and May. Ratios of positive edges to negative edges for bacteria increased with time, which was the opposite for eukaryotes (Table S5). This suggests decreasing and increasing potential antagonistic interactions within bacteria and within eukaryotes, respectively. When bacterial and eukaryotic communities were treated as a whole, the number of nodes revealed a similarity (June > May > April) for nodes (Supporting Information: Figure S12 and Table S5). The total planktonic community had a higher ratio of positive to negative edges in June than in April and May (Supporting Information: Table S5), suggesting a decreasing degree of potential antagonistic interactions. The top five core nodes (those with the highest betweenness centralities) in the planktonic conetworks in 3 months were affiliated with Acinetobacter, Candidatus_Aquirestis, Cryptomonas, Deinococcus, Limnobacter, Limnohabitans, Mycobacterium, and Rhodoferax ASVs (Supporting Information: Figure S12). Except for ASV_42847 identified as Acinetobacter in April, the relative abundances of these core species were <1% and some even <0.1% (rare taxa) and were related differently to the physicochemical variables (Supporting Information: Figure S12). For instance, the relative abundance of ASV_227008 identified as Acinetobacter in April was noticeably correlated with turbidity (p < 0.01), dissolved oxygen (p < 0.05), and Chl-α (p < 0.01). These results support comparable differences in co-occurrence patterns of bacteria and eukaryotes between the 3 months. Significant correlations were found between TLI and bacterial diversity in April (p < 0.05), bacterial abundance in June (p < 0.05), and eukaryotic abundance in 3 months (p < 0.05 or p < 0.001) (Supporting Information: Table S6). was used to potential between turbidity, and abundances of planktonic bacteria and eukaryotes in 3 months (Figure positively affected diversity bacteria in April) and abundances of bacteria and eukaryotes, which in positively affected TLI eukaryotes in April and (Figure showed positive effects on and significant were found most (p < 0.05 or p < 0.01 or p < The displayed good to our number = = 1), as by the in for bacteria = p = May, = p = June, = p = and eukaryotes = p = May, = p = June, = p = (Figure showed both and effects on whereas diversity and abundance of bacteria and eukaryotes displayed effects on TLI (Figure the total effect of turbidity rather than planktonic diversity and abundance showed effects on TLI in 3 sampling months. Chl-α and trophic state and many studies prefer to explore influences of physicochemical factors (e.g., nitrogen, phosphorus, pH, temperature, and on the water trophic state [5, However, a of the between environmental factors and the water trophic state is Environmental heterogeneity can reflect in environmental properties and studies that water environmental heterogeneity in the Paraná River during [25, We found that aggravated algal blooms in decreased environmental which been reported Previous studies that environmental heterogeneity affects ecological processes community assembly in and aquatic ecosystems [25, Ecological assembly processes community species coexistence and abundance, and community composition with [24, 25, Aggravated algal blooms increased bacterial and eukaryotic and some factors (e.g., turbidity) notably affected abundance and diversity of bacteria and eukaryotes. studies to verify this in more aquatic ecosystems to for The community composition of bacteria and eukaryotes in urban lakes within a short time, which is to for other lakes [17, and phylogenetic of bacterial and eukaryotic communities in the urban lakes with geographical which is with in other aquatic ecosystems The significant were to rather than assembly processes of bacterial and eukaryotic been reported that differentiating processes (i.e., dispersal limitation and variable selection) to community compositional dissimilarity The significant differences in the community of bacteria and eukaryotes over 3 months be to shifts in water physicochemical Nutritional (e.g., nitrogen and phosphorus) and nonnutritional abiotic factors (e.g., temperature, turbidity, pH, dissolved and affected community composition of bacteria and/or eukaryotes in the studied and other urban lakes [17, For different lakes, different environmental factors control planktonic community composition Generally, community composition is with community and and it is to disentangle different factors controlling planktonic community composition time and Stochastic and deterministic processes balanced community assembly of bacteria in the studied urban In the studied urban lakes, stochastic processes dominated eukaryotic community which is in with a of eukaryotic community assembly in the River and in along the this from studies that both stochastic and deterministic processes balance eukaryotic community assembly in lakes and reservoirs as well as in in of of stochastic and/or deterministic processes plankton community assembly to be affected by environmental heterogeneity and differences [25, The urban lakes in this study, to have both a relatively low environmental heterogeneity and which to the differences in bacterial and eukaryotic community environments trend to have a more community whereas environments a more community assembly This can be for eukaryotic community assembly in the studied urban different ecological processes to community assemblies of bacteria eukaryotes, which be also to differences in and Bacteria are as whereas many eukaryotes are For example, are as producing eukaryotes (e.g., and algae) than and differently in the state. However, some eukaryotes (e.g., and protozoa) can in water by and which relatively to nutrients and from (e.g., low and Therefore, a reported that the relative of neutral processes and environmental selection to community assembly on Environmental constraints of bacteria increased with aggravated algal blooms, whereas environmental constraints of eukaryotes decreased with aggravated algal blooms. Typically, eukaryotes have (e.g., and some eukaryotes relatively trophic levels in aquatic food In contrast, bacteria relatively low trophic levels in aquatic food webs and mainly in The differences in and dispersal between bacteria and eukaryotes to different in environmental constraints of bacteria and eukaryotes during aggravated algal blooms. species replacement could also the opposite change trend of environmental constraints of bacteria and eukaryotes, suggesting a species replacement result in a weak environmental constraint [20]. Coexistence patterns of bacteria and eukaryotes to on bacterial coexistence pattern across in and time, bacterial and eukaryotic communities showed opposite interaction patterns with decreasing and increasing potential antagonistic interactions, respectively. This pattern in to the effects of environmental constraints on bacteria eukaryotes. This a between environmental constraints and potential antagonistic interactions. eukaryotes relatively trophic levels in aquatic food webs [13], blooms or for food and (e.g., nitrogen, and phosphorus can between and water which nutrient for core species identified from co-occurrence networks, for example, Limnohabitans, Deinococcus, and Acinetobacter, are important of both and phosphorus cycles an environmental planktonic interactions be by the of core species (e.g., dredging) to community for nutrient rather than diversity and abundance of planktonic bacteria and eukaryotes showed effects on water trophic state of the studied lakes, which is opposite to our hypothesis. This be to turbidity affecting the of to the which in affects and biomass of (e.g., cyanobacteria and algae) In turbidity can also affect nutrient availability which in influences bacterial and eukaryotic Both bacteria and eukaryotes can nutrients and nutrient limitation increases between bacteria and eukaryotes for nutrients and Therefore, nutrient limitation affects abundance, community and the of plankton nutrients can be from the water [11, this the for large of bacteria and eukaryotes at short time Considering of bacterial and eukaryotic the limitation of affect the between water trophic state and abundance of bacteria and eukaryotes. studies estimate the between planktonic abundance and water trophic state by planktonic abundance by (e.g., time of biotic and abiotic factors affecting the water trophic state is important to environmental the of lake environmental measures (e.g., and be used during the bloom development to mitigate or algal bloom our sequencing and analyses are the to that aggravated algal blooms decreased the environmental heterogeneity of urban Planktonic bacteria and eukaryotes displayed strategies in response to aggravated algal blooms, with of opposite in species richness environmental and potential antagonistic interaction over time. significant effects on abundance and diversity of bacteria and eukaryotes. turbidity rather than diversity and abundance of planktonic bacteria and eukaryotes control more on water trophic state. our current of response to aggravated algal bloom and planktonic diversity maintenance These ecological and the of environmental to and more mitigate the state of studies to verify these in more different lakes over a time In each month (i.e., April, May, and June) in we collected water samples from representative in 12 lakes in (Supporting Information: Table S1 and Figure A total of water samples were collected by a Water and water samples were in water and at in a The (e.g., and were of the water samples were used to water physicochemical and the was for Water temperature, pH, dissolved turbidity, and electrical were determined in with a Water We measured of total phosphorus, soluble reactive phosphorus, total nitrogen, ammonia nitrogen, nitrate nitrogen, chemical oxygen demand, calcium, magnesium, iron, and Chl-α to and of the of water physicochemical properties are in (Supporting Information The key parameters of total nitrogen, total phosphorus, and chemical oxygen were used as factors to the TLI of urban lakes, and have been reported and also in (Supporting Information Environmental heterogeneity was based on physicochemical dissimilarity and are in (Supporting Information of plankton was with a the and were measured with a samples were at and were used to the bacterial the and were used to eukaryotic the was used to the abundances of bacteria and eukaryotes an The amplicon sequencing was conducted on an Illumina MiSeq at The were the and to and ASVs For sequencing was based on a number of and we all ASVs that < The phylogenetic of bacteria and eukaryotes were based on by the analysis of variance was used to reflect significant differences in the and were to estimate planktonic compositional to distribution patterns of We the taxonomic distances represented by dissimilarity and phylogenetic distances represented by distance of both bacterial and eukaryotic we evaluated between geographical distance and community based on taxonomic and phylogenetic The was used to elucidate the effects of physicochemical factors on bacterial and eukaryotic community We the effect other of each physicochemical on community composition of planktonic bacteria and eukaryotes investigate the effects of species replacement and richness difference on community composition of bacteria and eukaryotes, we taxonomic β-diversity based on dissimilarity by and and by and of R [20]. Mantel correlograms were constructed to whether significant phylogenetic signals of species occur at short of phylogenetic distances along environmental gradients by the in the neutral and niche breadth analysis were to estimate community assemblies of planktonic bacteria and eukaryotes According to in taxonomic and phylogenetic index the model was to of stochastic and deterministic processes by the Ecological assembly processes were to and homogeneous selection < variable selection > dispersal limitation < and > homogenizing dispersal < and < “undominated” processes < and < deterministic processes > stochastic processes < homogenizing processes dispersal plus homogeneous and differentiating processes selection plus dispersal of these ecological assembly processes are reported We the niche breadth index of and to infer the niche by the in large values reflect good Sloan neutral model was to estimate the of stochasticity to community assembly the migration as The ASVs in > of samples in each month were used to co-occurrence networks networks were by the with a p < 0.01 and > was and of and and and and and and and This was by the of and was by the The of The sequencing were in the of the for Information and R by are in Supporting and be found in the or The is for the content or of by the than be to the for the