Lake Soyang is an artificial reservoir in Chuncheon, Yanggu-gun, Inje-gun, Gangwon State, South Korea, and it serves as a critical freshwater resource for the region. The lake was formed as a result of the construction of the Soyang Dam on the Han River in 1973. The forebay of Lake Soyang has a water depth of approximately 100–110 m, with a water surface elevation ranging from 83 to 186 m (Kim et al. 2000). The forebay of the dam (Fig. 1), which represents the central zone of the lake and is often used as a study point in ecological and water quality assessments, was selected as the sole sampling site in this study because of its hydrological stability and representativeness. Field measurements of water quality parameters and sample collection were performed exclusively in the photosynthetically active surface layer, where zooplankton and phytoplankton are primarily active through photosynthesis. Water samples were collected using 4-L sterilized polyethylene bottles, immediately placed in the dark, and transported on ice to the laboratory within 24 h to minimize biological and chemical changes. In the laboratory, the samples were subsampled and processed for nutrient analysis and biological assessments. From March to October (except in 2016, from May to October due to logistical constraints), water samples were collected twice a week for the measurement of environmental factors, quantitative analysis, and identification of phytoplankton and zooplankton. The sampling frequency was set twice a week to capture seasonal variations in the plankton communities and environmental factors during the productive period of the lake ecosystem.

The WT, dissolved oxygen (DO), pH, electric conductivity (EC), and turbidity were recorded at a 0.5-m depth from the surface of the lake using a multiparameter water quality monitoring probe (Mobile Multi Sensor Meter; YSI Proplus, Yellow Springs, OH, USA). The concentrations of chlorophyll-a (Chl-a), total nitrogen (TN), and total phosphorus (TP) were analyzed following the Standard Methods for Examination of Water and Wastewater, 10th Edition (Rice et al. 2012). Air temperature, rainfall, and discharge data were obtained from the Water Resources Management Information System (WAMIS). A 1-L water sample was collected for phytoplankton analysis, while a 250-mL sample was collected for zooplankton analysis. Both samples were immediately fixed in 1% Lugol’s solution and formalin (final concentration, 2%). The collected phytoplankton and zooplankton were identified and quantified using a Sedgewick-Rafter counting chamber under a light microscope (BX51; Olympus, Tokyo, Japan) at 100–400× magnification. Bacillariophyta, Chlorophyta, Chrysophyta, Cryptista, Cyanophyta, and Dinoflagellata found in freshwater were identified based on their morphological characteristics by referring to Illustraion of the Freshwater Algae (Chung 1993). In addition, Freshwater Flora of Central Europe, Vol. 19. Cyanoprokaryota (Komarek 2013) was additionally referred to for the identification of cyanobacterial species. For species identification using taxonomical keys of zooplankton, refer to Illustration of the Freshwater Zooplankton of Korea (Cho 1993). In this study, the dominant species in each sampling period were classified as the species that accounted for the highest proportion of the entire plankton community with the highest cell density during that period.

The pheatmap package (Kolde and Kolde 2015) in the R program version 4.3.3 (R Foundation for Statistical Computing, Vienna, Austria) was utilized to analyze the correlation between the plankton communities and environmental factors in Lake Soyang. Spearman’s correlation analysis (p < 0.05, p < 0.01, and p < 0.001) was performed, and the results were visualized as a heatmap. The 19 representative species of phytoplankton and zooplankton were selected based on their occurrence for more than two years and a cell density of at least 100 cells mL⁻¹.

Additionally, to examine the relationship between phytoplankton communities and environmental variables, Canonical Correspondence Analysis (CCA) was performed using the vegan package in the R statistical environment version 4.3.3 (R Foundation for Statistical Computing). The analysis utilized phytoplankton and zooplankton community structure data collected from 2016 to 2023. As part of the preprocessing step, species with row sums less than or equal to zero were excluded. Environmental factors included water quality parameters (WT, pH, EC, TP, TN, and Chl-a), as well as hydrological and meteorological data such as air temperature (AT), rainfall, and discharge. The statistical values (F-value, p-value, and significance) for the correlations between plankton community structure and environmental factors were presented (Supplementary Tables S1 & S2).

Comparing the community structures of phytoplankton and zooplankton based on the occurrence of cyanobacterial blooms, non-metric multidimensional scaling (NMDS) and permutational multivariate analysis of variance (PERMANOVA) were conducted. Plankton community data were normalized using the relative abundance of species observed at each sampling site over time, and a distance matrix was calculated based on Bray-Curtis dissimilarity. The reliability of the visualized results was assessed using the stress value. To assess the statistical significance of differences between communities, a PERMANOVA was performed. This analysis was based on the Bray-Curtis distance matrix, and the differences in average distances among groups were evaluated using an F-statistic. A total of 999 permutations were conducted to calculate the p-value.

From 2016 to 2023, the WT in Lake Soyang was 20.4 ± 7.2°C, and the pH concentration was 7.9 ± 0.8 (Fig. 2A). The average pH was 7.4 from 2016 to 2019; however, it peaked at 8.4 in 2020, representing a significant increase compared with its values in previous years. From 2020 to 2023, the average pH stabilized at 8.1, which was 0.7 higher than that in the pre-2020 period (Fig. 2B). The DO concentration during the study period was 9.5 ± 1.7 mg L⁻¹ (Fig. 2C). EC exhibited an inverse trend with that of TP. It gradually increased from 2016 to 2019 and peaked at 101.2 μS cm⁻¹ in 2019; however, it decreased to an average of 73.3 μS cm⁻¹ from 2020 to 2023 (Fig. 2D). TN ranged from 0.1 to 3.4 mg L⁻¹ and peaked in 2018 and 2022 (Fig. 2E). TP decreased from 2016 to 2019, and its lowest average TP concentration of 4.3 μg L⁻¹ was recorded in 2019. However, TP sharply increased to an average of 9.4 μg L⁻¹ in 2020 and then stabilized at an average concentration of 5.6 μg L⁻¹ from 2021 to 2023 (Fig. 2F).

The analysis of the phytoplankton community composition revealed that Bacillariophyceae (diatoms) was the most dominant family from 2016 to 2023, accounting for 30.7–58.8% of the total phytoplankton composition during the survey period. Cyanobacteria accounted for 0.7–20.3% of the total phytoplankton; its highest proportion (20.3%) was observed in 2023. Cryptophyceae accounted for 6.1–19.3% of the total phytoplankton, and it peaked at 19.3% in 2020. Meanwhile, Chrysophyceae and Mediophyceae reached their maximum proportions of 16.0% and 30.8% in 2019 and 2022, respectively. The proportions of Chlorophyceae, Dinophyceae, and Coscinodiscophyceae were below 10% throughout the survey period.

Among the zooplankton detected in Lake Soyang, rotifera was the most dominant, comprising 55.7–85.6% of the zooplankton community throughout the survey period. The relative abundance of protozoa increased from an average of 8.1% (rate: 2.7–10.3% in 2016–2020) to 16.1% (2021–2023). Similarly, the relative abundance of copepoda increased from 9.0% (rate: 3.3–13.8% in 2016–2020) to 12.9% (2021–2023), and that of cladocerans increased from 4.3% (rate: 2.4–6.4% in 2016–2020) to 10.0% (2021–2023). The occurrence rates of protozoa, copepoda, and cladocera increased slightly after 2021 (Fig. 3).

The average cell density of diatoms was 1,352 cells mL⁻¹ during most periods except during midsummer (July–August), when their density declined to 531 cells mL⁻¹. In contrast, the abundance of cyanobacteria was higher from August to September (average of 1,166,982 cells mL⁻¹) than that in other periods (average of 24 cells mL⁻¹). Notably, an extreme cyanobacterial bloom, with a cell density of 37,317,481 cells mL⁻¹, occurred in August 2023. Meanwhile, the cell density of green algae and other taxa ranged from 0 to 2,322 cells mL⁻¹ during the survey period.

In the case of zooplankton, rotifera was consistently dominant throughout the survey period, which reached its peak twice: 604 ind. mL⁻¹ in May 2016 and 330 ind. mL⁻¹ in June 2018. Its cell density was higher in May and June (average: 138.5 ind. mL⁻¹) than in other periods (average: 59.5 ind. mL⁻¹). Meanwhile, the cell density of protozoa peaked at 133 ind. mL⁻¹ in April 2017; however, it remained below 100 ind. mL⁻¹ in other periods. Finally, the cell densities of copepoda (0–42 ind. mL⁻¹) and cladocerans (0–92 ind. mL⁻¹) exhibited seasonal increase after July (Fig. 4).

Nineteen phytoplankton species were identified as dominant at least once during the survey period. Quadrigula lacustris had a dominance rate of 44.1% in September 2023. The genus Cryptomonas was dominant 14 times; its dominance rate was 27.6–64.2%, primarily from spring to fall. Five genera belonging to Cyanophyceae were dominant (29.3–90.3%) in summer (July–September). Dolichospermum and Microcystis were dominant from July to August, whereas Aphanizomenon and Chroococcus were dominant in September. The diatoms Asterionella, Fragilaria, Synedra, and Cyclotella were dominant more than 15 times in total; their maximum dominance rate (86.5–98.9%) was higher than that of other taxa. They were dominant in spring and fall but not in midsummer (Table 1).

Among the zooplankton, copepoda was dominant (16.2–40.6%) from August to October. Within rotifera, Keratella cochlearis var. tecta f. microantha was dominant 30 times during the survey period; its relative abundance ranged from 31.0 to 87.4% between March and July. In contrast, Keratella cochlearis var. tecta and Keratella valga var. valga were dominant from August to September. Polyarthra vulgaris was the most frequent species, with 38 occurrences; it was predominant (16.5–79.8%) from May to October. The genera belonging to Ciliophora such as Epistylis and Tintinnidium were dominant (29.7–72.5%) from March to May (Table 2).

The analysis of the correlations between the 19 representative phytoplankton species in Lake Soyang and environmental factors revealed that diatoms such as Asterionella formosa, Aulacoseira distans, and Fragilaria crotonensis exhibited a strong negative correlation (p < 0.001) with WT, AT, and EC. In contrast, Ceratium hirundinella and the green algae Scenedesmus sp. showed a significant positive correlation (p < 0.05) with WT.

Additionally, the diatoms Asterionella formosa and Aulacoseira granulata showed a negative correlation (p < 0.01) with TP, whereas the cyanobacterium Aphanizomenon issatschenkoi showed a strong positive correlation (p < 0.05) with TP. Microcystis aeruginosa, which are the major cyanobacterial bloom-forming species, showed a positive correlation (p < 0.01) with Chl-a and pH; however, a significant negative correlation (p < 0.01) was observed between Microcystis aeruginosa and TN (Fig. 5A).

The correlation analysis between zooplankton species and environmental factors revealed that rotifera such as Ploesoma truncatum and Polyarthra vulgaris showed a strong positive correlation (p < 0.001) with WT and EC. Meanwhile, the rotifera Conochilus unicornis, Filinia longiseta, and Keratella cochlearis var. tecta f.microcantha showed a strong negative correlation (p< 0.01) with Chl-a. Additionally, nauplius, copepodid, strong positive correlation (p < 0.001) with WT, Chl-a, and TP (Fig. 5B).

The correlation analysis between the 19 representative zooplankton and phytoplankton species with a high frequency of occurrence revealed that the occurrence patterns of diatoms such as Aulacoseira distans, Asterionella formosa, and Fragilaria crotonensis were negatively correlated with all the zooplankton taxa identified in the freshwater samples from Lake Soyang. In contrast, the occurrence patterns of cyanobacteria and green algae were positively correlated with rotifera, copepoda, and protozoa. Microcystis aeruginosa, a bloom-forming species, showed a strong positive correlation (p < 0.001) with the rotifera Trichocerca cylindrica and Asplanchna priodonta. Furthermore, Microcystis sp. showed a strong positive correlation with the copepod Eodiaptomus japonicus (Fig. 5C).

The CCA results showed that the cyanobacterial taxa Aphanizomenon issatschenkoi, Microcystis aeruginosa, Microcystis sp., and Dolichospermum spiroides, which appeared predominantly in summer, were correlated with pH, Chl-a, and discharge, reflecting their seasonal characteristics. During the same period, nauplius, copepodid, and Eodiaptomus japonicus (copepod), as well as Trichocerca cylindrica (rotifera), also appeared—likely due to feeding strategies—and exhibited strong correlations with the same three environmental factors (Fig. 6).

The NMDS results revealed a clear distinction in the community structure of phytoplankton and zooplankton between bloom and non-bloom periods. Due to the dominance of specific taxa during bloom periods, large differences in cell density among other groups were observed, resulting in relatively high stress values (0.410 for phytoplankton and 0.246 for zooplankton), indicating low significance in the NMDS analysis (p = 0.392 and 0.446 for phytoplankton and zooplankton, respectively). Nevertheless, it was evident that the occurrence of blooms led to distinct differences in plankton community structure (Fig. 7).

Among the phytoplankton species observed in Lake Soyang, some exhibited seasonal dominance, while others appeared consistently from spring to fall. For instance, Asterionella formosa was dominant during spring and fall but was undetected in summer, consistent with its known blooming patterns in early spring and late fall (Wang et al. 2012, Sivarajah et al. 2016). Similarly, Dinobryon divergens was dominant from spring to early summer before being succeeded by harmful cyanobacteria in August. This species has higher tolerance to low temperatures than other phytoplankton (Taş et al. 2010), allowing it to dominate over a relatively wide range of seasons.

In addition, when these two species dominated the freshwater ecosystem, they exhibited a high dominance rate of up to over 90%. These species are mixotrophic phytoplankton that can survive by feeding on picophytoplankton even in nutrient-limited environments, particularly under phosphorus-limited conditions (Laybourn-Parry and Marshall 2003, Rottberger et al. 2013). Their mixotrophic characteristic provides an ecological advantage for adaptation to oligotrophic water systems.

Among the harmful cyanobacteria, Dolichospermum and Microcystis dominated during midsummer (July–August), accounting for more than 50% of the phytoplankton community. In contrast, Chroococcus and Aphanizomenon were dominant, mainly during fall (September). Ryu et al. (2016) detected Aphanizomenon at a cell density exceeding 30,000 cells mL⁻¹ in the Nakdong River in southern Korea in June. The Han River, where the Soyang Dam is located, is situated further north than the Nakdong River and is characterized by lower WTs in September. Under these environmental conditions, Aphanizomenon avoids the summer season and thrives at WTs lower than those in which Dolichospermum and Microcystis typically appear.

Cryptomonas sp., Fragilaria crotonensis, Synedra acus, and Cyclotella sp. consistently occurred in the sampling site from spring to fall, indicating their ability to tolerate a wide range of WTs and nutrient conditions. These species were included in the 10 most frequently occurring phytoplankton during the study period. Cyclotella and Synedra are well-known bloom-causing diatoms in fall, and there have been several cases of the transition of dominant species to Cyclotella or Synedra replace the dominant phytoplankton species after the occurrence of cyanobacterial blooms in summer. As Synedra is a representative species that causes spring blooms (Hara et al. 1983), and Cyclotella is not dominant when the WT is above 23°C and is only dominant in spring and fall (Mitrovic et al. 2008), the results of this study are consistent with those of previous studies.

During the survey period, Bacillariophyceae accounted for the highest proportion of phytoplankton, whereas rotifera was the dominant group among the zooplankton identified in Lake Soyang. In 2016 and 2019, when the proportion of cyanobacteria was 1.31% and 0.71%, respectively, the proportions of protozoa, cladocera, and copepoda also decreased. Protozoa have been shown to be selective in feeding on cyanobacteria based on the nutritional composition (e.g., carbon-nitrogen ratio) of the diet (John and Davidson 2001, Deng et al. 2020). Therefore, it is likely that the decrease in cell density of nitrogen-fixing cyanobacteria such as Aphanizomenon and Dolichospermum affected the abundance of the protozoa sampled in this study.

Likewise, Large-bodied cladocerans have greater grazing efficiency on cyanobacteria than smaller cladocerans (Li et al. 2022). In particular, Daphnia can suppress the growth of bloom-forming cyanobacteria such as Aphanizomenon, Dolichospermum, Microcystis, and Planktothrix (Sarnelle 2007, Urrutia-Cordero et al. 2016). The increase in the relative abundance of Daphnia in summer is considered a bottom-up phenomenon due to the increased availability of microbial plankton prey such as bacteria and phytoplankton (Wagner and Benndorf 2007, Hüelsmann 2011).

Rotifera such as Keratella cochlearis, Keratella valga, and Polyarthra vulgaris exhibit high occurrence rates in freshwater ecosystems during the period when the harmful cyanobacteria Dolichospermum and Microcystis are dominant in this study. A significant relationship among these groups is plausible based on their roles in the food chain (Gharib 2006). The potential grazer, rotifer, showed a simultaneous increase in cell abundance during Dolichospermum-induced algal blooms, and it was confirmed that direct grazing by rotifers contributed to the regulation of the bloom-forming cyanobacterial population (Ntetsika et al. 2025).

In this study, the cell abundance of copepoda was also found to be influenced by the abundance of cyanobacteria. When cyanobacterial blooms occur, adult copepods change their feeding strategy, although they originally prefer larger prey, which can negatively affect their natural food chain structure (Kosiba and Krztoń 2022). The abundance of copepods is expected to increase when harmful cyanobacteria appear due to the change in the feeding strategy of copepods (Toullec et al. 2019).

Compared with diatoms, green algae and cyanobacteria, which inhabit relatively high WTs and alkaline eutrophic freshwater, showed a significant positive correlation with WT and pH. In contrast, diatoms mainly inhabit oligo-mesotrophic lakes and exhibited a significant negative correlation with WT, EC, and pH, consistent with their tendency to dominate in spring and fall.

Major green algae and cyanobacteria, including Aphanizomenon, were highly correlated with TP, whereas Microcystis aeruginosa and Dolichospermum spiroides, which are responsible for the formation of summer blooms, were also positively correlated with Chl-a. In this study, the correlation between the occurrence patterns of Aphanizomenon, Dolichospermum, and Microcystis and TN concentrations was not statistically significant. This may be attributed to the fact that nitrogen availability is not a strong limiting factor in freshwater environments. Moreover, heterocytous cyanobacteria such as Aphanizomenon and Dolichospermum possess nitrogen-fixing capabilities via heterocytes, providing a strategic advantage for survival under nitrogen-limited conditions (Lehman et al. 2009, Yema et al. 2016).

Ploesoma truncatum, Polyarthra vulgaris, Eodiaptomus japonicus, Trichocerca cylindrica, nauplius, copepodid, Bosmina longirostris, and Difflugia corona, which had high cell densities in summer, also showed a strong positive correlation with WT, pH, EC, and TP. These patterns were like those observed for green algae and cyanobacteria. However, the correlation between the cell density of these species and TN was weak, indicating that P, rather than N, is a key factor affecting the abundance of both phytoplankton and zooplankton in freshwater systems.

In this study, differences in phytoplankton and zooplankton community structures between bloom and non-bloom periods were analyzed using NMDS. Although distinct differences were observed depending on the occurrence of blooms, the analysis showed low statistical significance because the only bloom in which the cell density of Microcystis aeruginosa, causing HABs, exceeded 1.0 × 10⁶ cells mL⁻¹ occurred in 2023. The definition of cyanobacterial blooms was based on the cell density criteria established by the Ministry of Environment (Korea). To address these limitations, future research should focus on systems with more frequent cyanobacterial blooms or utilize multi-year modeling, specifically for Lake Soyang. With climate change and abnormal weather patterns increasing the likelihood of the occurrence of blooms (Vadadi-Fülöp et al. 2012, Morgado and Vieira 2020), the cyanobacterial bloom in 2023 underscores the vulnerability of the lake to recurring blooms. Furthermore, continuous monitoring is essential for maintaining the health and stability of aquatic ecosystems.

Although this study revealed temporal patterns in nutrient levels and plankton communities, information on inflow water quality, watershed land use, and anthropogenic modifications during the study period was not available. The absence of these data limit the ability to fully disentangle the relative contributions of external watershed factors to the observed interannual variability. Future studies should aim to incorporate long-term watershed characteristics and inflow water quality monitoring to more comprehensively evaluate the drivers of nutrient dynamics and plankton community shifts in the reservoir system. Furthermore, although this study focused on long-term ecological changes at a single forebay site, investigating spatial heterogeneity across multiple locations would provide valuable insights into the dynamics of phytoplankton community structure and enhance our understanding of the ecosystem.