Cyanobacterial blooms have increased in frequency around the world in recent decades (Huisman et al., 2018) and are likely to become more severe as a consequence of eutrophication, rising CO₂ levels and accelerating global warming (Verspagen et al., 2014; Chapra et al., 2017). Moreover, cyanobacterial blooms can cause major problems, such as toxin production, hypoxia generation, and food web disruption, leading to the loss of ecosystem services (Verspagen et al., 2014; Chapra et al., 2017). However, some mechanisms underlying the ecological success of cyanobacteria remain unclear, which makes it difficult to deal with cyanobacterial blooms (Wilhelm et al., 2020).

In summer, cyanobacterial communities occur widely in freshwater by displacing eukaryotic algae, which generally includes diatom and green algae (Ke et al., 2008; Zepernick et al., 2021). In addition, the succession of phytoplankton among different cyanobacterial genera also occurs frequently (Moustaka-Gouni et al., 2006; Chun et al., 2020; Tanvir et al., 2021). Various factors, including nutrients, predation, temperature, light, pH, antibiotics, and water turbulence, have been found to influence these successions and the ecological success of cyanobacteria (Steffen et al., 2015; Reavie et al., 2016; Huisman et al., 2018; Wang et al., 2021; Xu et al., 2021; Zepernick et al., 2021). Given the complexity of cyanobacterial blooms, the drivers of cyanobacterial dominance and succession are still being explored (Wang et al., 2021).

Cases in past decades have shown that viruses are becoming more pronounced in phytoplankton community regulation (Suttle et al., 1990; Fuhrman, 1999; Brussaard, 2004; Knowles et al., 2016; Pound et al., 2020). As suggested by negative frequency-dependent selection, the “Kill-the-Winner” (KtW) model of lytic infection predicts that abundant prokaryotic types will be exposed to strong viral pressure for maintaining high prokaryotic richness (Winter et al., 2010). Many studies on phytoplankton blooms showed direct viral control and provided empirical support for the KtW model (Bratbak et al., 1993; Hewson et al., 2001; Baudoux et al., 2006). Under a modified KtW model, Pound et al. (2020) found that viruses may suppress the competition of eukaryotic community and allow for the cyanobacterial bloom. Recently, infection strategy of the “Piggyback-the-Winner” (PtW) model has been proposed in which lysogeny predominates at high microbial abundance and growth rates (Silveira and Rohwer, 2016). Several studies also supported the PtW model (Knowles et al., 2016; Coutinho et al., 2017). However, these different viral infection strategies could be favored depending on environmental conditions (Bongiorni et al., 2005; Payet and Suttle, 2013). Viral lysis and lysogenic infection also contribute to bacterioplankton community. Similar to phytoplankton, high viral lysis pressure will apply to dominant and fast-growing bacteria, which has been confirmed to have a major impact on bacterial diversity and the community structure, i.e., the “Kill-the-Winner” model (Winter et al., 2010). Besides, lysogeny also has been previously hypothesized to be a preferable survival strategy for both the virus and bacterioplankton (Paul, 2008). Notably, viral lysis predominantly channels particulate organic carbon and nutrients away from higher trophic levels, which was called “viral shunt” (Wilhelm and Suttle, 1999). Nutrients released by viral lysis of heterotrophic bacteria can be efficiently remineralized and transferred to phytoplankton (Weinbauer et al., 2011; Shelford and Suttle, 2018). Currently, the accumulated scientific evidence about the role of viruses is growing fast, but the information about the roles of viruses in bloom succession is still limited.

Cyanobacterial bloom succession has occurred frequently in several tributaries of the Three Gorges Reservoir (TGR), the world’s largest hydroelectric power project, since the initial water impounding in 2003 (Zhou et al., 2019). And the summer phytoplankton population succession in Xiangxi Bay of TGR in 2010 was in order as follows: diatom, green algae, cyanobacteria, and the main factors of were water temperature, water stability, and mixed layer depth (Fang et al., 2013). It also has been revealed that nutrients, temperature, light, and hydrodynamic regimes are the key environmental factors affecting the outbreaking and succession of blooms (Fang et al., 2013; Yang et al., 2022). As for viruses in reservoirs, Kopylov and Zabotkina (2021) investigated the distribution of viruses, the frequency of visibly infected cells of heterotrophic bacteria and autotrophic picocyanobacteria and their virus-induced mortality in six reservoirs along the Volga. The viral infection rate of picocyanobacteria in mesoeutrophic reservoir was higher than that in mesotrophic reservoir (Kopylov et al., 2010). Although a large number of reservoirs have been built worldwide, there have been very few studies on the ecological effects of viruses in these manipulated aquatic ecosystems in-depth, especially when the bloom occurs.

In the present study, we investigated the viral lysis and lysogeny of phytoplankton and bacterioplankton in the Xiangxi Bay (the largest tributary of the TGR, China) during a bloom succession event from July 18 to August 29. The abiotic environmental factors were also monitored to compare how they differ from the effects of viral factors during bloom succession. The results revealed that viruses played multiple potential roles in summer bloom succession in Xiangxi Bay of TGR and may facilitate the bloom success of cyanobacteria. These findings provide a foundation for further understanding phytoplankton succession and controlling the cyanobacterial blooms.

In the present study, three stages of summer bloom succession were observed (Figures 2, 3; Table 1) and this phenomenon occurs frequently in Xiangxi Bay of the TGR (Fang et al., 2013). Both phyto- and bacterioplankton lytic and lysogenic infection characteristics were significantly different in these three stages (Figure 4) and may have influenced the succession of summer blooms by different ways from the analysis of SEM, Spearman rank correlation, RDA, and Hierarchical partitioning (Table 2; Figures 5–7). And these effects of viruses in summer bloom succession may further have helped the success of cyanobacterial bloom.

To explore the causation between virus and the phytoplankton community succession, the biodiversity of Shannon Wiener index (H) was used as the observed variable of phytoplankton in SEM. The path coefficient between H and latent variable “phytoplankton” was 1.0, which showed that the SEM could explain the most of the variation of H. Nutrient, virus, and physical factor are the factors influence the H, but Nutrient and Virus are more important. Many other studies also have shown that nutrient was more important than climate change for phytoplankton community change (Yan et al., 2019; Zhang et al., 2021). As for virus, VLP was the only important factor and showed positive influence on H, which is similar to the “Kill-the-Winner” (KtW) model for maintaining high prokaryotic richness (Winter et al., 2010). Besides, VLP may be suppressed under a higher temperature for the adaption of virus to the adverse environment (Tsai et al., 2015b; Stough et al., 2017), which was also reflected in SEM (Figure 5). And there are reports with similar feature that nutrient had positive influence to VLP (Kopylov et al., 2010; Payet and Suttle, 2013). Overall, the established SEM has high credibility, and results showed that virus had obvious influence on phytoplankton diversity.

From Bloom I to Bloom II, the first successional shift was from cyanobacteria-diatom (Microcystis sp. and M. granulata var. angustissima codominance) to cyanobacteria (Microcystis sp. dominance) blooms, which can be described as the succession of different phyla. The roles of viral infection may include photosynthetic eukaryotic community suppression and cyanobacterial enhancement, and they were both likely to be work positively for the first succession and helped the success of cyanobacterial bloom.

In our speculation that the suppression of photosynthetic eukaryotic community in the first succession, the VLP was higher in Bloom I (Figure 4A) and also exhibited significant positive relationships with eukaryotic algal species (Figures 6, 7), which showed the potential control of eukaryotic community. Kopylov et al. (2010) found that mesoeutrophic reservoir have higher viral-induced mortality than mesotrophic one. In our study, water with high nutrient levels in Bloom I may contribute to result of the higher VLP. In Lake Taihu, Pound et al. (2020) also observed that the viral lysis may suppresses the eukaryotic community during the Microcystis bloom and regarded that the virus-mediated transition from diatoms to cyanobacteria could be explained within the context of a modified version of the KtW model. However, different from the relatively stable hydrological conditions in Lake Taihu, the hydrological conditions in Xiangxi Bay were much more complex due to the reservoir operation (Xu et al., 2011). These hydrodynamic changes shaped the abiotic environments significantly and affected the algal bloom in Xiangxi Bay (Yang et al., 2018). Unavoidably, the hydrodynamic conditions (including tide, thermal stratification, flow velocity changes, and water vertical mixing) can strongly affect the virus-host interactions (Auguet et al., 2005; Barros et al., 2010; Säwström and Pollard, 2012; Chen et al., 2019). Although hydrodynamic conditions in Xiangxi Bay are susceptible to human manipulation and may affect viral infection, there was still obvious viral lysis suppression on eukaryotic algal community in Xiangxi Bay, which is similar to the role of viruses in Lake Taihu (Pound et al., 2020). The summer hydrodynamic conditions in Xiangxi Bay (including stable stratification, low surface mixing layer depth, and velocity) that created similar with shallow lakes may contributed to this result (Zhao et al., 2012; Yang et al., 2018). However, it is regrettable that potential effect of hydrodynamic changes on the bloom dynamics cannot be discussed due to the lack of monitoring data. In addition, silicon limitation has been found to facilitate the viral infection and mortality of marine diatoms in marine environments (Kranzler et al., 2019). And silicate concentration was significantly reduced in the first bloom succession (Table 1), which may also act as the constraining nutrient that causes the collapse of diatoms and provides opportunities for the later cyanobacterial bloom.

As for cyanobacterial enhancement, there was an increasing abundance of Microcystis sp. during the first succession (Table 1) and the highest FLP in Microcystis bloom (Figure 4B). The infection characteristic of FLP may also have influenced the phytoplankton composition. Specifically, the Shannon diversity index of the phytoplankton community decreased with increasing FLP (Figure 6). The second component in the PCoA, which captured the second-most amount of variance in the phytoplankton communities (Figure 6), showed a positive correlation with FLP. Previous studies have shown that Microcystis can resist viral lysis through high number of restriction modification systems (Zhao et al., 2018), and may also minimize losses by forming a lysogenic state (Stough et al., 2017). Knowles et al. (2016) found that the viral densities were more consistent with temperate than lytic life cycles with increasing microbial abundance and growth rates, which was the infection strategy of the PtW model. The potential for Microcystis to resist viral lysis and increase dominance by forming a lysogenic state could be another important factor that allows for cyanobacterial dominance and the first succession.

From Bloom II to Bloom III, the second succession was the bloom of different cyanobacterial genera (Microcystis sp. dominance to Microcystis sp. and A. circinalis codominance). VLB, which has significant contribution in RDA model (Table 2), showed a positive relationship with different species (e.g., A. circinalis and A. azotica; Figure 7). Nutrients released by viral lysis of heterotrophic bacteria may help the second succession and contribute to the continued success of cyanobacterial bloom. And indeed, there exist a decline of bacterioplankton along the succession (Supplementary Table S1) and an increase of PO₄³⁻ in Bloom III (Table 1). In Bloom II and III, TN/TP ratios were 16.1 and 12.6, respectively. And there was the lowest TN in Bloom III (Table 1). Tian et al. (2012) also found the decrease of nitrate in summer bloom of Anabaena in Xiangxi Bay and N limitation was considered as an important cause. Besides, the heterocytes existed in the morphological features of Anabaena in Bloom III during the phytoplankton species identification, which is a typical characteristic for nitrogen fixation (Golden and Yoon, 1998). And there was a high concentration of TN in Bloom I, which Anabaena was not dominant during this period. Hence, there was high possibility of N limitation in Bloom III, which may cause the replacement of non-N₂-fixing Microcystis sp. by N₂-fixing A. circinalis. Previous studies showed a competitive advantage of Microcystis at low P concentrations because of its ability to rapidly uptake and store inorganic P, which also caused P deficiency to other coexisting phytoplankton species in Bloom II (Wan et al., 2019). During the second succession, the PO₄³⁻ concentration in Xiangxi Bay was still low, but an increase in Bloom III (Table 1), which might be caused by the increase of the VLB, and may help the growth of N₂-fixing species. Importantly, N₂ fixation, a metabolically expensive process, is controlled by P availability (Wang et al., 2018). Vanderhoef et al. (1974) found the increased algal growth and nitrogen fixation are correlated with higher phosphate concentrations. Weinbauer et al. (2011) found that the abundant regenerative nutrients required for Synechococcus growth were adequately provided due to the viral lysis of heterotrophic bacteria. Tsai (2020) emphasized the importance of the viral shunt of bacteria and quantified the P nutrient released by the viral lysis to be approximately 597.6 ng P L⁻¹ d⁻¹ in coastal waters. From Bloom II to Bloom III, the VLB followed a rising trend, the average size increasing from 0.211 to 0.312 d⁻¹ (Figure 4), which may also contribute to the increase of PO₄³⁻ concentration from 0.001 mg/L in Bloom II to 0.004 mg/L in Bloom III (Table 1) and finally help the growth of N₂-fixing species.

This study found that the variation of phytoplankton communities during summer bloom successions was closely associated with the changes in abiotic environmental factors (i.e., T, TN, and PO₄³⁻) and viral factors (i.e., VA, VLP, FLP, and VLB; Table 2; Figures 6, 7). Hierarchical partitioning method has been widely used to estimate the individual importance of each explanatory variable (Li et al., 2022; Pecuchet et al., 2022). Based on the hierarchical partitioning analysis, the viral variables still have a marked effect on the dynamics of phytoplankton community, although the environmental attributes were the major factors (Table 2). Studies on the estimate of the viral lysis in bloom succession are scarce, so the importance of viruses compared to other factors remain uncertain. However, our results quantified the size of the viral infection characteristics of phytoplankton and bacterioplankton, and showed that viruses may play an important role in summer bloom succession (Figures 5–7). Moreover, our findings were consistent with other studies which showed that viruses were important in bloom control (Jacquet et al., 2002; Du et al., 2020). In general, viruses may play vital roles in phytoplankton community regulation, but these roles may be ignored. And our results highlighted the need for understanding viral infection dynamics in realistic environmental contexts to better predict their biogeochemical consequences (Mojica et al., 2016; Zimmerman et al., 2020).

The roles of viruses in cyanobacterial blooms and taxa succession have yet to be well elucidated in aquatic ecosystems, including the world largest hydropower reservoir, TGR. In this study, viral infection characteristics monitoring during the summer bloom successions was conducted to determine the effects of viruses on cyanobacterial bloom in Xiangxi Bay of TGR. The main conclusions were drawn as follows:

Viruses may promote the cyanobacterial blooms and work by multiple and complex ways, including the enhanced lysis of eukaryotic community, the increase of lysogeny in cyanobacteria, and the nutrients supplied from the lysis of bacterioplankton.

The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author.

ML, YZ, and SY conceived and designed the study. YJ and JG performed the analyses. YJ provided the additional sample material. ML, YZ, and WX contributed to the project administration. KP analyzed the results and wrote the manuscript, which was edited and approved by all authors. All authors contributed to the article and approved the submitted version.

This work was supported by the National Natural Science Foundation of China (No. 51579092) and the National Key Basic Research Program of China (No. 2016YFC0401702).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.