In marine environments, cell densities of autotrophic microalgae and some heterotrophic protists increase and thereby change the color of the sea—a biological phenomenon known as a red tide. Some red tides negatively impact fisheries and tourism (Sellner et al., 2003), and the impacts of harmful red tides on coastal systems have increased in recent decades (Gobler, 2020). Techniques for mitigating damages in the fish culture industry have been suggested (Kim, 2006; Balaji-Prasath et al., 2022), with precautionary field interventions such as early harvesting, restricting fish feeding, and transport of fish cages into zones free of red tides being the most practical (Kim, 2006). To boost the effectiveness of these precautionary interventions, accurate forecasting of the arrival of a red tide at fishing grounds is necessary. Forecasting based on numerical models is challenging because the densification and expansion of a red tide are influenced by various factors, including biological, chemical, and physical processes (McGillicuddy, 2010; Davidson et al., 2016).

In most cases, the outbreak of a red tide in a marine area requires three steps. First is the appearance of cells in the area due to the transport of cells from other marine areas, growth of surviving cells at quite low density, and excystment. Next is the growth step, in which the cell division rate exceeds the loss rate. The third step is aggregation, which occurs due to physical and biological processes. Various environmental factors influence each of the three steps. As compared to the aggregation process, the growth and life-history stages (including excystment) of red-tide algae have been more thoroughly studied and reported (e.g., Granéli and Turner, 2006).

Most flagellates causing red tides display diurnal vertical migration (DVM) in which cells normally migrate upward during the day and downward at night (Kamykowski, 1995). DVM contributes greatly to the aggregation step (Lai and Yin, 2014) and likely to the growth step as well, as the process offers many ecological benefits to flagellates, such as nutrient acquisition at a wider range of depths, the ability to control the amount of light received for photosynthesis, and escape from predators (Smayda, 1997; Harvey and Menden-Deuer, 2011; Bollens et al., 2012).

Among red-tide flagellates, the DVM pattern follows the photoperiod and appears to be controlled by an endogenous clock (Roenneberg et al., 1989; Shikata et al., 2015), whereas the timing of change in swimming direction and swimming speed differ among species (Kamykowski, 1995; Shikata et al., 2015). Some studies indicated that even if the dominant species is the same across a marine area, DVM patterns can differ among places and days (Katano et al., 2014; Shikata et al., 2020). Previous studies revealed that the DVM pattern is influenced by external environmental factors such as temperature (Heaney and Eppley, 1980), salinity (Jephson et al., 2011), light (Ault, 2000), and nutrient concentrations (Heaney and Eppley, 1980). Based on their different swimming speeds, Bearon et al. (2004) postulated that DVM patterns differ among strains in a flagellate species.

Although numerical simulation models considering physical processes have been developed (Berdalet et al., 2014), no algorithm can precisely simulate DVM patterns of a red-tide algae in the field. This deficiency exists because few studies have assessed both the physical and biological processes underlying DVM, including the physiological responses to various environmental factors, and the biological parameters impacting DVM patterns remain to be clarified for each red tide flagellate.

The raphidophyte Chattonella marina complex (hereafter called Chattonella) is distributed in warm marine coastal areas around the world and has caused tremendous damage to fisheries (Imai and Yamaguchi, 2012). Reflecting these threats, a large amount of information on Chattonella has been gathered, particularly with regard to its growth physiology, life cycle, and interaction with other microorganisms. Field data collected before and after Chattonella red tides are also available for designing laboratory experiments and validating numerical models using biological parameters. In addition, studies on the DVM of Chattonella have been conducted as well. A field experiment using a mesocosm revealed that Chattonella could reach a depth of 7.5 m during DVM, and its migration speed was 0.8 m h^−l both upward and downward (Watanabe et al., 1995). Chattonella can sense weak UV and blue light, allowing it to cue its DVM to the day–night cycle (Shikata et al., 2013). Negative phototaxis was observed under a monochromatic light with UV and blue components, but the response disappeared by mixing light with longer wavelengths (Shikata et al., 2016). Although one report has noted chemotaxis toward inorganic phosphate (Ikegami et al., 1995), we also observed a clear DVM in Chattonella without a vertical gradient of phosphate concentration (Shikata et al., 2015). Therefore, the DVM pattern may be regulated mainly by the endogenous clock that synchronizes the photoperiod and switches between positive and negative geotaxis in Chattonella. However, low salinity in the upper layers arrests the upward migration of Chattonella (Katano et al., 2014; Shikata et al., 2014), and a field study of DVM showed that cell density in the water column might also affect DVM (Sakaguchi et al., 2017). These findings indicate that there are also factors inhibiting the DVM pattern regulated by the endogenous clock.

Chattonella can produce a large amount of superoxide (•O₂ ⁻) in the cell membrane and excrete it from the cell. This •O₂ ⁻ and hydrogen peroxide has antimicrobial activity and is a candidate toxin related to fish mortality (Oda et al., 1992), as the level of •O₂ ⁻ is highly correlated with toxicity to aquacultured fishes in seawater that contains Chattonella cells (Shikata et al., 2021). Ichthyotoxicity can disappear within a short period and only live Chattonella cells can kill fishes; ruptured cells and culture supernatant have no toxicity to fish (Matsusato and Kobayashi, 1974; Ishimatsu et al., 1996; Shen et al., 2010). These characteristics correspond with chemical unstableness of •O₂ ⁻. The amount of •O₂ ⁻ production is influenced by nutrient and light intensity conditions; nutrient deficiency and strong light can increase the production (Yuasa et al., 2020a; b). Moreover, Chattonella exhibits different •O₂ ⁻ production rates in the light and dark periods (Kim et al., 2005; Yuasa et al., 2020a). These results suggest that •O₂ ⁻ may be responsible for regulating various physiological processes with circadian rhythms.

In the present study, we examined the effects of strain, growth phase, cell density, and nutrient deficiency—factors scarcely addressed thus far—on the DVM pattern in Chattonella. In most experiments, we also simultaneously measured the •O₂ ⁻ level to investigate the potential synchronous patterns of DVM and •O₂ ⁻ production in this flagellate species.

Among investigations conducted under quiet conditions in enclosed coastal areas or in a mesocosm, there were differences in the depth reached by a Chattonella red tide at night (Watanabe et al., 1995; Araki et al., 2013; Sakurada et al., 2013; Katano et al., 2014; Sakaguchi et al., 2017). In previous laboratory studies using Chattonella strains NIES-1 and 3KGY, most Chattonella cells migrated to the surface of a vessel during the light periods and to the bottom during the dark periods (Kim et al., 2005; Shikata et al., 2014). Before the present study, however, we viewed flasks including Chattonella strains during the dark periods and found that some strains aggregated at the surface of the medium even during the night. The present study showed that some strains display clear downward migration during the dark periods, whereas others scarcely migrate under the same condition. Strains Ago03 and Ago04 were isolated from the same marine area during different years, and both displayed weak downward migration. Strains 4KGY used in the present study and 3KGY used in Shikata et al. (2014) were isolated from the same marine area on the same day. Compared to 4KGY (early log phase), 3KGY has higher surface accumulation ratios during the dark periods and accumulates at the bottom during the light periods in the same experimental setup, indicating that DVM in 3KGY has more entrainability. Although Chattonella shows variations in cell shape and toxicity among strains (Demura et al., 2009; Shikata et al., 2021), the relationship between their phenotypes (Table 1) and DVM pattern remains obscure. Our findings indicate that the strain composition of a Chattonella population may influence the vertical cell distribution at each time point in the field, although it is unclear what produced the interstrain differences in the DVM pattern.

Interstrain variation in the DVM pattern may contribute to species survival by decreasing the risk that all cells are exposed to a harmful environment at a certain depth. However, if only strains with a poor ability of downward migration survive, it is unclear whether the population could recover the ability. Moreover, the large diversity in DVM can inhibit densification of cells, leading to poorer competition with other microorganisms, such as antibiotic ability (Oda et al., 1992) and toxicity against other phytoplankters (Qiu et al., 2011; Fernández-Herrera et al., 2016). Therefore, the strain composition of a Chattonella population may characterize the population dynamics in each marine area.

It is difficult to incubate some Chattonella strains at large volumes (>200 mL) in normal vessels such as an Erlenmeyer flask for days, although the reasons are unclear. Surprisingly, the present study revealed that strains OP27, Ago03, and Ago04, which can grow in large-scale incubation, displayed little downward migration. It is impossible to resuspend a cell pellet made by centrifugation of a Chattonella culture because the cells strongly adhere to each other at high cell density (Jenkinson et al., 2007). Although bioconvection occurs at the surface and mitigates densification during periods of upward migration, bioconvection does not occur at the bottom during periods of downward movement. Therefore, success in the large-scale incubation of Chattonella may be closely related to the ability of a strain to migrate downward.

Sakaguchi et al. (2017) reported that a red tide of Chattonella displays clear DVM when the average cell density is on the order of 10³ cells mL⁻¹ in the water column, whereas downward migration is scarcely observed when the average cell density is on the order of 10⁴ cells mL⁻¹ and nutrient concentrations are sufficient for Chattonella growth. Likewise, in the present study Chattonella cells scarcely displayed DVM when the average cell density in the water column (volume: 1.6 L) was more than 10⁴ cells mL⁻¹ (late log phase in Figure 2). Although the maximum cell density of Chattonella was normally on the order of 10⁴ cells mL⁻¹ even when the other conditions were suitable for growth, the growth phase at the time of the inoculation scarcely affected DVM patterns. Therefore, densification may hamper various physiological activities related to downward migration and growth, although the mechanism remains to be clarified. When the sample volume was small (25 mL), however, clear DVM was observed even on the order of 10⁴ cells mL⁻¹ (complete medium in Figure 3). In Chattonella, at greater incubation volume, the growth rate and maximum cell yield were lower. Thus, the effects of densification on physiological states may depend on the incubation volume in laboratory studies.

In various flagellate species, the depletion of nitrogen or phosphorus activates or inhibits DVM (Heaney and Eppley, 1980; Yuasa et al., 2018). We observed that the deficiency of nitrogen or phosphorus inhibited DVM in Chattonella with two axenic strains. Although the red-tide dinoflagellate Karenia mikimotoi aggregates at the surface all day under nitrogen-depleted conditions (Yuasa et al., 2018), we found that Chattonella sank more easily in nitrogen- or phosphorus-depleted conditions. In a Swedish fjord, the amount of Chattonella cells peaked in sediment traps deployed in deep layers (15 and 30 m depths) when a red tide of Chattonella was declining and nitrate concentrations in the surface layer (0–10 m depth) were low (Waite and Lindahl, 2006). In addition, DVM ceases under continuous darkness, whereas it persists under continuous red light, which acts on photosynthesis but scarcely shifts the phase of the DVM pattern (Shikata et al., 2015). Therefore, swimming would require energy from photosynthesis, which in turn requires light and nutrients, in Chattonella.

Kim et al. (2005) examined the relationship between the production of reactive oxygen species and DVM using the Chattonella strain NIES-1. They observed DVM similar to that recorded in the present study and found that •O₂ ⁻⁻ production was high during the light period and that it decreased during the dark period in cell-aggregated layers (surface or bottom). Yuasa et al. (2020a) observed a similar diurnal pattern in •O₂ ⁻ production in the strain 4KGY. The present study also revealed a similar pattern of •O₂ ⁻ production in the strain OP27, but the DVM pattern of strains Ago03 and Ago04 was different from that of the other strains, with •O₂ ⁻ production increasing during the dark period. Interestingly, we noticed that characteristics of the cellular •O₂ ⁻ production pattern may be categorized by cell shape in Chattonella strains, as noted in strains with a Chattonella antiqua–like shape (NIES-1, 4KGY, OP27) and strains with a C. marina–like shape (Ago03, Ago04); the amount of •O₂ ⁻ production, however, is not categorized by cell shape among strains (Shikata et al., 2021). These findings imply that the DVM pattern may be independent of the •O₂ ⁻ production pattern because only the latter was categorized by cell shape.

Based on experiments using the strain NIES-1, Kim et al. (2004) suggested that the growth phase (or cell density) can affect the amount of •O₂ ⁻ production but has little effect on the diurnal pattern of •O₂ ⁻ production. The •O₂ ⁻ production rate had a maximum value during the exponential growth phase (low cell density) and subsequently decreased in the stationary phase (high cell density), but •O₂ ⁻ production increased during the light period and decreased during the dark period regardless of growth phases. However, the present study using strain 4KGY showed that the diurnal pattern of •O₂ ⁻ production was less clear at higher cell density (stationary phase).

Kim et al. (2004) reported that •O₂ ⁻ production was lower when nitrogen and phosphorus concentrations were low than when these nutrients were fully supplied in strain NIES-1. On the other hand, Yuasa et al. (2020a) reported that •O₂ ⁻ production is stimulated under nitrogen- or phosphorus-deficient conditions in the 4KGY strain. Therefore, the response of •O₂ ⁻ production to nutrient deficiency appears to depend on the strain. According to the reverification result in the present study, however, •O₂ ⁻ production was stimulated under nitrogen- or phosphorus-deficient conditions in both NIES-1 and 4KGY. Previous studies have suggested that the reducing power required for •O₂ ⁻ production in the cell membrane may originate from photosynthesis, based on inhibition of •O₂ ⁻ production by the administration of a photosynthetic inhibitor during the light period as well as the increase in •O₂ ⁻ production during the light period (Marshall et al., 2002; Yuasa et al., 2020b). Yuasa et al. (2020a) have reported that •O₂ ⁻ production under nitrogen- or phosphorus-deficient conditions increase during the dark period and the extracellular levels of •O₂ ⁻ under nutrient-deficient conditions are unaffected by the presence of an inhibitor of photosynthetic electron transport—indicating that pathways of •O₂ ⁻ production other than photosynthesis exist. However, the present study revealed that diurnal patterns of •O₂ ⁻ production differ among strains. Thus, there may be differences among strains with regard to the degree of dependence on each production pathway and sensitivity to environmental conditions such as cell density and nutrient availability.

In conclusion, our findings indicate that strain, cell density, and nutrient deficiency are factors that affect the DVM pattern and •O₂ ⁻ production in Chattonella. Moreover, our results suggest that the diurnal patterns of vertical migration and •O₂ ⁻ production may be independently regulated, indicating that a Chattonella population can have cells with numerous combinations of DVM and •O₂ ⁻ production patterns. A great diversity of DVM and •O₂ ⁻ production in a population should increase the chances of acquiring nutrients and light and successfully competing against other microorganisms in the water column. Quantitative analyses using numerical models based on the information obtained in the present study are important future tasks.