Effects of intrusion and retreat of deep cold waters on the causative species of red tides offshore in the South Sea of Korea

To explore the effects of eutrophic deep waters offshore on red-tide outbreaks, waters were collected from 0, 20, and 40 m depths at an offshore station in the South Sea of Korea, and the growth rate of each major red-tide dinoflagellate Prorocentrum donghaiense, Tripos furca, Alexandrium fraterculus, and Margalefidinium polykrikoides in the waters was measured. No species grew at dissolved inorganic nitrogen concentrations (DIN) of < 7.0–8.4 µM, but the growth rates of all four species rapidly increased and became saturated at 12–15 µM DIN. On July 31, 2020, the DIN was 4.1–4.6 µM in 0–10 m depth waters but 8.9–18.4 µM in 20–50 m depth waters. Under these circumstances, considering that calculated reachable depths of P. donghaiense, T. furca, A. fraterculus, and M. polykrikoides were 10, 15, 24, and 52 m, respectively, A. fraterculus and M. polykrikoides can reach the deep waters and grow, but P. donghaiense and T. furca cannot. However, if the deep waters intrude and ascend by 10–20 m relative to that on July 31, four species can reach deep waters and grow, whereas if the deep waters retreat and descend by 10–20 m, only M. polykrikoides can reach the deep waters and grow. Red-tide outbreaks by M. polykrikoides in the South Sea in 2014 occurred after the deep water retreat. Thus, the position of the eutrophic deep waters, affected by their intrusion and retreat and solar insolation, can affect outbreaks of harmful M. polykrikoides red tides.


Introduction
Red tides are discoloration of the sea surface due to plankton blooms (Horner et al. 1997;Yentsch et al. 2008;Jeong et al. 2013;McGillicuddy et al. 2014;Glibert et al. 2018) and can lead to illnesses in humans, large-scale mortality of fish and shellfish, and economic losses for the aquaculture and tourism industries (Hallegraeff 2003;Anderson et al. 2012;Park et al. 2013). To date, there have been many studies on the distribution and the mechanisms of red-tide outbreaks and the succession of their causative species (Yamochi and Abe 1984;Smayda 1997;Franks 1997;Anderson et al. 2005;Burkholder et al. 2008;Wyatt 2014;Jeong et al. 2015Jeong et al. , 2017Jeong et al. , 2021Kemp and Villareal 2018;Lee et al. 2020;Ok et al. 2021b). Many co-existing red-tide species compete for limiting nutrient resources in aquatic environments, but only a few dominate and cause red tides. Moreover, the mechanism of red-tide outbreaks varies among species (e.g., Menden-Deuer and Montalbano 2015; Jeong et al. 2015;Qin et al. 2021;Eom et al. Responsible Editor: S.Shumway. 6 Page 2 of 14 2021). Thus, species-level measured data on the dominance of major red-tide species should be provided to more accurately predict red-tide outbreaks.
There are two major nutrient sources for phytoplankton growth and, consequently, red-tide formation in marine environments (Fig. S1a)-freshwater input, which has usually low salinity and is nutrient rich (Zhou et al. 2008;Goes et al. 2014;Li et al. 2014;Wilkerson et al. 2015;Kraus et al. 2017;Lee et al. 2019;Stumpner et al. 2020;Ok et al. 2021a), and deep cold water, which has usually high salinity and is nutrient rich (Atkinson et al. 1984;Kaneda et al. 2002;Nakano et al. 2004;Ougiyama et al. 2004;Jeong et al. 2017). In coastal and estuarine waters, nutrient-rich freshwater flows into the surface water (West et al. 1996;Maier et al. 2012;Lemley et al. 2018). When the nutrient concentrations in coastal surface waters increase due to the freshwater input, fast-growing red-tide species usually dominate (Pinckney et al. 1999;Jeong et al. 2015Jeong et al. , 2017Ok et al. 2021a). Moreover, physical and chemical inhibition among co-existing species is critical for outcompeting other red-tide species in coastal and estuarine waters (e.g., Kubanek et al. 2005;Wang et al. 2013;Lim et al. 2014). However, offshore nutrient concentrations are usually low in surface waters but high in deep cold waters. Deep cold waters have been reported to largely affect the physical, chemical, and biological properties in several regions, such as the South Atlantic Bight and Southern California Bight in the United States, the southern Brazilian shelf, the Bungo Channel in Japan, and the South Sea of Korea (Bumpus 1955;Hofmann et al. 1981;Hofmann and Ambler 1988;Lee et al. 1991;Kaneda et al. 2002;Lopes et al. 2006;Jeong et al. 2017;McGowan et al. 2017). Due to the large volume, deep cold waters contain a tremendous amount of nutrients (Atkinson et al. 1984;Jeong et al. 2017). Therefore, red-tide species grow and form red tides when they reach deep cold waters. Although several studies have provided data on the growth rates of competing red-tide species in coastal areas where freshwater flows (e.g., Domingues et al. 2011;Wang et al. 2013;Cira et al. 2016;Cloern 2018), little is known about the growth rates of these species in offshore deep cold waters.
Most red-tide flagellates and dinoflagellates exhibit diurnal vertical migration (Hasle 1950;Eppley et al. 1968;Cullen and Horrigan 1981;Cullen 1985;Smayda 1997;Kudela et al. 2010;Peacock and Kudela 2014;Shikata et al. 2015). They can photosynthesize in well-lit surface waters during the day and take up nutrients from deep cold waters at night (Eppley et al. 1968). Flagellate and dinoflagellate species have a wide range of maximum swimming speeds and can, thus, reach different depths through vertical migration (Smayda 2000(Smayda , 2002Jeong et al. 2015). Therefore, each of the migrating flagellates or dinoflagellate may respond differently to different positions in deep cold waters.
The position of deep cold waters is largely affected by the thermocline and the deep cold waters' intrusion and retreat (Fig. S1b;Yang et al. 2000;Kaneda et al. 2002;Jeong et al. 2017). High-intensity solar radiation in the summer creates a strong and deep thermocline that separates surface water from deep cold water. Moreover, spring-neap tidal modulation may affect the intrusion and retreat of deep cold waters, causing the thermocline to ascend or descend (Geyer and Cannon 1982;Cannon et al. 1990;Kaneda et al. 2002;Jeong et al. 2017). Thus, if the thermocline ascends with the intrusion of deep cold waters, several red-tide species may pass the thermocline and reach deep cold waters (Fig.  S1c). However, if the thermocline descends with the retreat of the thermocline, only a few species may reach deep cold waters (Fig. S1c).
Red tides occur almost every year in the South Sea of Korea, resulting in substantial losses to the aquaculture industry . The dinoflagellates Prorocentrum donghaiense, Tripos furca, Alexandrium fraterculus, and Margalefidinium polykrikoides (= Cochlodinium polykrikoides) have caused red tides in offshore waters since the 1980s, including in nearshore areas in aquaculture cage locations ). Although P. donghaiense, T. furca, and A. fraterculus are non-toxic, their red tides are known to be responsible for fish kills or hypoxia, which cause shellfish mortality (Horner et al. 1997;MacKenzie et al. 2004;Orellana-Cepeda et al. 2004;Guo et al. 2014). Moreover, ichthyotoxic M. polykrikoides often causes massive fish kills by damaging fish gills (Kim et al. 2000;Dorantes-Aranda et al. 2010). Once the red tides caused by M. polykrikoides occur, they last 14-56 days in Korea, resulting in enormous losses of up to USD 95 million per year . Therefore, predicting red-tide outbreaks caused by M. polykrikoides is a critical task for scientists, government officials, and those engaged in the aquaculture industry.
The vertical-migration ability of dinoflagellates is considered one of the important factors affecting red-tide formation (Villarino et al. 1995;Park et al. 2001;Jeong et al. 2015). The maximum swimming speeds of P. donghaiense, T. furca, A. fraterculus, and M. polykrikoides are 280, 403, 680, and 1449 µm s −1 , respectively (Jeong et al. 1999Baek et al. 2009). Therefore, if each species travels for 10 h, they can migrate vertically from the surface layer to depths of approximately 10, 15, 24, and 52 m, respectively ). However, the growth rates of these redtide dinoflagellates in waters at different depths have not yet been studied. To predict the outbreak of red tides by M. polykrikoides and other red-tide dinoflagellates offshore, their growth rates should be measured at different depths. The causative species of red tides in a series may be predicted using the data on the growth rates at different depths, the different positions of deep cold water (intrusion or retreat), and vertical-migration depths.
To explore the effects of deep cold waters on the formation of red tides by each red-tide-causative species, waters were collected from the depths of 0, 20, and 40 m at an offshore station in the South Sea of Korea, where red tides frequently occur. Each of the four major red-tide species (P. donghaiense, T. furca, A. fraterculus, and M. polykrikoides) was then incubated in the waters collected from each depth. During the 10-day incubation period, the abundance of each species and nutrient concentrations were monitored, and the growth rates were determined. Using the measured growth rates and the calculated vertical-migration depths of each species, scenarios for red-tide formation by the four redtide species were developed, depending on the position of the intruded or retreated deep cold waters. The scenarios were then assessed based on intensive field observations conducted in the South Sea of Korea in 2014 . The results of this study provide a basis for understanding the mechanisms of red-tide outbreaks caused by major red-tide species offshore.

Water sampling and hydrological properties
Using 5-L Niskin water samplers, 40-L seawater samples were collected from depths of 0, 10, 20, 30, 40, and 50 m at a fixed station (Station 111, 53 m deep; Jeong et al. 2017) in the South Sea of Korea on July 31, 2020 (Fig. 1a). This station is approximately 70 km south of the mouth of the Seomjin River, and the effects of the freshwater input are low . Based on the previous field observations conducted in 2014, the intrusion and retreat of deep cold waters were observed in the study area from July to September 2014 . The collected samples were transported to the laboratory within 7 h and kept at 20 °C.
The water temperature and salinity at each sampling depth were measured using a YSI Professional Plus instrument (YSI Inc., Yellow Springs, OH, USA). Irradiance in the surface water was measured using a portable illuminance meter (IM-600; Topcon, Tokyo, Japan). The unit of the irradiance in the illuminance meter, Lux, was converted to µmol photon m −2 s −1 using the conversion factor provided by Thimijan and Heins (1983). The vertical irradiance profile was calculated using the data with the extinction coefficient (0.12 m −1 ) near the sampling station off Yeosu, Korea (Ko et al. 2019). For nutrient analysis, the water samples from each sampling depth were gently filtered through GF/F filters (Whatman Inc., Floreham Park, NJ, USA) and stored at − 20 °C until the concentrations of nitrate plus nitrite (NO 3 + NO 2 ; hereafter nitrate or NO 3 ), ammonium (NH 4 ), phosphate (PO 4 ), and silicate (SiO 2 ) were measured using a nutrient auto-analyzer system (QuAAtro, Seal Analytical GmbH, Norderstedt, Germany).

Growth rates of red-tide dinoflagellates
Five milliliters of aliquots from dense cultures of P. donghaiense, T. furca, A. fraterculus, and M. polykrikoides were subsampled, and the cells were enumerated to determine the cell density. Ten-milliliter aliquots were filtered through GF/F filters, and the concentrations of NO 3 , NH 4 , PO 4 , and SiO 2 were measured using the nutrient auto-analyzer system.
Seawater collected from 0, 20, and 40 m depths was filtered through 0.2 μm pore-sized filters. Cells of either species were added to triplicate 1-L polycarbonate bottles containing filtered seawater collected from 0, 20, and 40 m depths by transferring predetermined volumes of cultures. The initial cell density of the four dinoflagellates was approximately 100 ng C mL −1 . The biovolume of each species was obtained from our previous studies (Jeong et al. 2005Lim et al. 2015;Kang et al. 2018), and the carbon per cell of each species was calculated from the biovolume using the Menden-Deuer and Lessard (2000) equations.
The triplicate bottles were placed in a temperature-controlled culture room and incubated at 20 °C under 100 µmol photons m −2 s −1 illumination at a 14:10 h light/dark cycle with LEDs. In field, the target red-tide dinoflagellates exhibit diurnal vertical migration between well-lit and warmer surface waters and dim and colder waters and, thus, experience wide ranges of water temperature and irradiance every day (e.g., Park et al. 2001). However, measuring growth rates with variable water temperatures and irradiances during incubation is challenging. Thus, 20 °C was chosen because the mean water temperature at 0, 20, and 40 m depths at the sampling time was 19.9 °C (Fig. 1b), and 100 µmol photons m −2 s −1 illumination was chosen because the mean irradiance at 0, 20, and 40 m depths at the sampling time was 466 µmol photons m −2 s −1 , which was much higher than the saturation irradiances for the growth of the target dinoflagellate species (Fig. 1c); the irradiances at which the growth rates of P. donghaiense, T. furca, and M. polykrikoides become saturated are reported to be 30, 72, and 90 µmol photons m −2 s −1 , respectively (Kim et al. 2004;Baek et al. 2008;Xu et al. 2010); the irradiance at which the growth rate of Alexandrium pohangense, morphologically similar to A. fraterculus, become saturated is 58 µmol photons m −2 s −1 ).
Ten-milliliter aliquots were subsampled daily from each bottle for the 10 days of culturing to determine cell densities. Cell density was determined by enumerating cells in three 1-mL Sedgewick-Rafter counting chambers. The growth rate (µ, d −1 ) of each species incubated in the seawater collected from each depth was calculated using the following equation: where C t1 and C t2 are the cell densities of the species on days t 1 and t 2 , respectively.

Nutrient concentrations during incubation
Twenty-milliliter aliquots were taken from each bottle and then filtered through GF/F filters daily for the 10 days of incubation, and NO 3 , NH 4 , PO 4 , and SiO 2 concentrations were measured using the nutrient auto-analyzer system. Variations in nutrient concentrations as a function of incubation time in the waters from 0, 20, and 40 m depths were plotted. Whether the nutrient concentrations significantly changed as a function of the elapsed time was tested using the non-parametric Kruskal-Wallis test with the criterion of p < 0.05 in SPSS version 26.0 (IBM-SPSS Inc., NY, USA).
To determine the growth rates of each of P. donghaiense, T. furca, A. fraterculus, and M. polykrikoides as a function of the dissolved inorganic nitrogen (NO 2 + NO 3 + NH 4 ; DIN) concentration, the mean DIN concentration (DIN * ) at each interval was calculated as follows: where DIN t1 and DIN t2 are the DIN concentrations on days t 1 and t 2 , respectively. DIN * was calculated at 3-day intervals. DIN * during the first 3 days, when the species were in the acclimation period, was not used in plotting the growth rates of the four species against DIN * . Additionally, DIN * during the first 8 days for M. polykrikoides in the 40 m water was not included in the plotting, as its growth rates in this period were unusually low.
The maximum growth rate (µ max , d −1 ) of each dinoflagellate species was obtained using the modified Michaelis-Menten equation as follows: where DIN′ is the threshold DIN concentration required for growth (where µ = 0 d −1 ; µM), and K GR is the half-saturation constant of the DIN concentration for the growth of each species (where 0.5 µ max ; µM). Data were fitted to Eq. (3) using DeltaGraph (SPSS Inc., Chicago, IL, USA). The actual half-saturation constant of the DIN concentration for the growth of each species, K GR ′, can be calculated as K GR + DIN′ (Spilling et al. 2010).

Hydrological data
Water temperature decreased rapidly from 24.1 °C in the surface water to 17.1-19.0 °C at 20-30 m depths and then gradually to 16.4-16.7 °C at 40-50 m depths (Fig. 1b). The irradiance at the surface water measured in the present study was 1271 µmol photons m −2 s −1 , and the irradiances at 20 and 40 m calculated using an extinction coefficient of 0.12 m −1 were 115 and 10 µmol photons m −2 s −1 , respectively (Fig. 1c). Salinity increased from 32.6 in the surface water to 34.2 at 30 m and slowly to 34.6-34.7 at 40-50 m depths (Fig. 1d). The NO 3 concentration increased from 1.6 µM in the surface water to 5.2-10.4 µM at 20-40 m and 14.6 µM at 50 m (Fig. 1e). The NH 4 concentration at each depth did not vary largely, ranging from 2.9 to 3.8 µM (Fig. 1f). The DIN concentration increased from 4.6 µM in the surface water to 8.9-13.6 µM at 20-30 m and 18.4 µM at 50 m (Fig. 1g). The PO 4 concentration gradually increased from 0.1 µM in the surface water to 0.4-0.8 µM at 20-40 m and to 1.2 µM at 50 m (Fig. 1h). The SiO 2 concentration initially decreased from 15.1 µM in the surface water to 5.9 µM at 10 m but then increased to 48.1 µM at 50 m (Fig. 1i).

Discussion
The present study clearly demonstrates that positive growth rates of P. donghaiense, T. furca, A. fraterculus, and M. polykrikoides are observed only when the DIN concentration ≥ 7 µM. Thus, these four dinoflagellate species may not grow in seawaters with < 7 µM DIN; the DIN concentrations in the 0-10 m depth waters were < 7 µM on July 31, 2020. The growth rates of P. donghaiense, T. furca, A. fraterculus, and M. polykrikoides were saturated at 12-15 µM DIN, which was the concentration range  Table 2 The threshold dissolved inorganic nitrogen concentration required for growth (DIN′, µM), actual half-saturation constant of dissolved inorganic nitrogen for growth considering DIN′ (K GR ′, µM), and maximum growth rates (µ max , d −1 ) of each dinoflagellate species that were determined by fitting them to the modified Michaelis-Menten equation In the offshore waters of the South Sea of Korea, intrusion and retreat of the eutrophic deep cold waters can alter the position of the waters . In turn, this alteration of the position may affect the growth of the four red-tide dinoflagellates; if the deep cold water intrudes toward the coast and the nitracline ascends by 20 m relative to that on July 31, 2020, the DIN concentrations in the surface waters can be > 7 µM (Fig. 7b). Under these circumstances, all four dinoflagellate species in the ascended deep cold water can grow. However, if the deep cold water retreats and the nitracline descends by 20 m relative to that on July 31, 2020, the DIN concentrations can be > 7 µM only at > 30-40 m depths (Fig. 7c). Under The predicted scenarios for the red-tide-causative species of the present study were tested by comparing the results with those of the causative species in the serial redtide events in 2014 . The previous field observations conducted in the South Sea of Korea in 2014 showed that deep cold waters intruded (July 22, August 06, and September 15) and retreated (August 21 and September 01) (Fig. 8). The depth of the top of deep cold waters is primarily affected by the thermocline, which is usually affected by solar insolation. It was found to be shallower from July to early August (i.e., 5-15 m) but deeper in September (20-32 m; Fig. 9). Due to the solar insolation, the intrusion and retreat of deep cold waters pushed up or pulled down the position of the deep cold waters. On July 1, P. donghaiense caused red tides when the depth of the top of thermocline (or deep cold waters) was 8 m (Fig. 9). Moreover, T. furca caused red tides from July 11 to August 21 when the depth of the top of thermocline was located at 5-17 m, and A. fraterculus on August 21 at 17 m. Only M. polykrikoides formed a red tide as the top of the thermocline deepened and was positioned at 32 m on September 1, on which the retreat of deep cold waters was observed (Fig. 9). These red-tide outbreaks by each species in 2014 corresponded to our scenarios showing the combination of the growth rates of the four dinoflagellate species at each depth, vertical-migration depth, and the position of deep cold waters.
Grazers, competitors, and bacteria can lower the growth rate of red-tide dinoflagellates Jeong 2011, 2013;Buchan et al. 2014;Jeong et al. 2015). Some heterotrophic dinoflagellates and ciliates occasionally have considerable grazing impacts on populations of P. donghaiense, T. furca, A. fraterculus, or M. polykrikoides in Korean waters when the density of a predator is abundant ). On the sampling date (i.e., July 31, 2020), the density of each of the dominant heterotrophic dinoflagellates Gyrodinium spp. and Protoperidinium spp., tintinnid ciliates Leprotintinnus nordquisti and Tintinnopsis nucula, and naked ciliates was < 2 cells mL −1 . Thus, low densities of these potential grazers are unlikely to lower the growth rates of the redtide dinoflagellates, but mortality due to predation should be considered when densities of the grazers become high. The growth rate of M. polykrikoides has been reported to be inhibited when the densities of the diatoms Skeletonema costatum and Chaetoceros danicus are > ~ 130,000 and > ~ 1200 cells mL −1 , respectively (Lim et al. 2014). However, the density of total diatoms at the station on July 31, 2020 was 44 cells mL −1 . Some bacteria act as prey for red-tide dinoflagellates when their density is moderate but inhibit the growth of the dinoflagellates when their density is high Jeong 2011, 2013). The pathogenic bacterium Vibrio parahaemolyticus can kill M. polykrikoides but can also act as a prey for M. polykrikoides (Seong and Jeong 2011). At V. parahaemolyticus densities of > 1.5 × 10 6 cells mL −1 , the growth of M. polykrikoides can be inhibited (Seong and Jeong 2011). Thus, growth rates of M. This study provides actual data on the growth rates of the major red-tide species in the offshore waters collected from different depths in the South Sea of Korea and the DIN concentrations in which they can grow in natural deep cold waters. Many red tides have been reported near this offshore station ; National Institute of Fisheries Science, https:// www. nifs. go. kr/ redti deInfo, in Korean). Considering the intrusion and retreat of deep cold waters in several countries affected by dinoflagellate red tides (Cullen et al. 1982;Probyn et al. 2000;Jeong et al. 2017;McGowan et al. 2017;Condie et al. 2019), the results of the present study can be used to predict the outbreak of dinoflagellate red tides offshore in the waters of these countries. Deep cold waters are known to intrude and retreat at 15-day intervals and are affected by tidal cycles (Kaneda et al. 2002). Thermoclines are generally affected by the surface water temperature, which, in turn, is affected by solar insolation (Kraus and Turner 1967;Denman 1973). The intrusion and retreat of deep cold waters and the shallowing and deepening of thermoclines (and nitraclines) can therefore be predicted. In turn, the causative dinoflagellate species of red tides can be predicted. However, grazers, competitors, and bacteria should be considered in the prediction if their densities are high .