The 60 stations used for sampling in the study area were located in the mid and eastern South Sea of Korea. Water depths at the sampling stations ranged between 4 and 53 m (Fig. 1). The depths of the stations in the western South Sea are shallower than those in the eastern South Sea at the same latitude (Fig. 1). One of the estuary stations was located ca 20 km south of the mouth of the Seumjin River, whereas the other estuary station was located 5 km southwest of Haechang Bay. Sixteen sampling time cruises were conducted at 1–2 week intervals from May to Nov 2014. Two days of sampling were undertaken during each cruise; on each day, two ships simultaneously collected water samples at half of the stations in the study area.

Water temperatures and salinities throughout the water column were measured using 2 CTDs (YSI6600; YSI Inc., Yellow Springs, OH, USA; Ocean seven; Idronaut S.r.l., Milan, Italy). The data obtained using the CTDs were calibrated each cruise. In addition, when each water sample was taken, the temperature, salinity, pH, and dissolved oxygen (DO) in the sampling waters were measured using a YSI Professional Plus instrument (YSI Inc.). The chl-a concentration was measured in accordance with the protocol described in Eaton et al. (1995).

Water samples at each station were taken from 2–5 m depths depending on the depth with water samplers. Plankton samples were poured into 500-mL polyethylene bottles and preserved with acidic Lugol’s solution for determination of heterotrophic protist abundances. To determine the abundances of HTDs, TCs, and NCs, the samples preserved in acidic Lugol’s solution were concentrated by 1/5–1/10 using the settling and siphoning method (Welch 1948). After thorough mixing, all or a minimum of 100 cells of each protist species in one to ten 1-mL Sedgwick-Rafter counting chambers were counted under a compound microscope. Data on the abundances of phototrophic protists were obtained from Jeong et al. (2017).

The average and maximum values of the abundance of each protist group, water temperature, salinity, and concentration of chl-a on each sampling date were derived from all water samples collected on the sampling date.

The carbon content for each species of the red-tide dinoflagellates A. fraterculus, C. polykrikoides, C. furca, and P. donghaiense was obtained from the literature (Jeong et al. 1999, 2001, Yoo et al. 2013b, Zhang et al. 2015).

Pearson’s correlation was used to calculate the correlation coefficients between heterotrophic protists and physical and red tide organisms (IBM SPSS statistics 23; IBM Corp., Armonk, NY, USA).

The grazing coefficients (g) attributable to a dominant heterotrophic protist feeding on a co-occurring red tide dinoflagellate were calculated by combining field data on the abundance of the dominant heterotrophic protist and red tide dinoflagellate with ingestion rates of the protist grazer on the red tide prey obtained from the literature (Jeong et al. 1999, 2001, Kim 2004, Kim and Jeong 2004, Yoo et al. 2013b, Zhang et al. 2015). The combination of a red tide dinoflagellate and dominant heterotrophic protist grazer in the study period was C. polykrikoides and NCs (>100 μm in cell length) and P. kofoidii, C. furca and P. kofoidii, A. fraterculus and Gyrodinium dominans/G. moestrupii, and P. donghaiense and G. dominans/G. moestrupii, Noctiluca scintillans, and NCs (>100 μm in cell length) (Jeong et al. 2017, this study). The ingestion rate of NCs (>100 μm) on C. polykrikoides was assumed to be the same as that of Strombidinopsis sp. on C. polykrikoides (Jeong et al. 1999). It was difficult to distinguish G. moestrupii, G. dominans, and other Gyrodinium spp. (<40 μm) from one another under a light microscope due to their morphological similarities (Yoo et al. 2013b). Thus, the ingestion rates of G. moestrupii, G. dominans, and other Gyrodinium spp. (<40 μm) on P. donghaiense were assumed to be the same as that of G. dominans on Prorocentrum minimum, because the cell size and shape of P. minimum and P. donghaiense are similar (Kim and Jeong 2004, Cai et al. 2006). The ingestion rate of NCs (>100 μm) on P. donghaiense was assumed to be the same as that of Strombidinopsis sp. on P. minimum (Jeong et al. 1999). Finally, the ingestion rate of G. moestrupii, G. dominans, and other Gyrodinium spp. (<40 μm) on A. fraterculus was assumed to be the same as that of G. moestrupii on Alexandrium minutum (Yoo et al. 2013b).

The grazing coefficient (g, d⁻¹) was calculated as:

where CR is the clearance rate (mL predator⁻¹ h⁻¹) for a predator of a red-tide organism at a given prey concentration and GC is the predator concentration (cells mL⁻¹). CR values were calculated as:

where IR is the ingestion rate (cells eaten preator⁻¹ h⁻¹) of the predator on the prey and x is the prey concentration (cells mL⁻¹). CR values were corrected using Q₁₀ = 2.8 (Hansen et al. 1997) because the in situ water temperatures and the temperatures used in the laboratory for the experiment were different.

The percentage (%) of a red-tide organism population consumed by a predator population in 1 day was calculated as:

Average water temperatures ranged between 15.0–23.2°C during the study period, with a peak observed on Sep 28. Average salinity ranged between 32 from 34, with the lowest levels observed on Sep 1 (Fig. 2). Concentrations of chl-a ranged from 0.1–44.2 mg m⁻³, with averages ranging from 1.2–3.0 mg m⁻³ (Fig. 2). Peaks for both the maximum and average concentrations of chl-a were recorded on Sep 1 (Fig. 2). Three-dimensional distributions of water temperature, salinity, chl-a concentration, and heterotrophic protists in the study area are provided in Appendix 1.

Thirty-five TC taxa, several NC taxa, and 30 HTD taxa were found at all stations over the study period (Appendix 2). Although the maximum abundances of TCs (49.0 cells mL⁻¹), NCs (35.0 cells mL⁻¹), and HTDs (82.4 cells mL⁻¹) did not differ substantially from one another, the peak abundances of these 3 groups occurred at different times (Table 1, Fig. 2); TCs and NCs abundances were highest in May to Jun, whereas that of HTDs was highest in Sep. Maximal abundances of TCs and NCs were observed when the diatom or the dinoflagellate P. donghaiense formed red tides, whereas the maximal HTD abundance was observed following peak chl-a concentrations and C. polykrikoides red tides (Fig. 2). The abundance of total ciliates was 0–50.6 cells mL⁻¹ (Table 1, Fig. 2), and the maximum abundance of total ciliates was observed on May 7.

When considering locations where high abundances were observed, the spatial distributions of TCs, NCs, and HTDs differed from one another at all sampling times with the exception of the period Jun 12 to Jul 1, 2014 (Fig. 3). High abundances of TCs was observed in the western part of the study area at most sampling times; high abundances of NCs was observed in the western or middle parts of the study area at most sampling times; and high abundances of HTDs was observed in the eastern part of the study area from May 7–Jun 5 and Aug 6–Sep 15, but in the western or middle parts of the study area from Jun 12–Jul 1.

Diatoms, P. donghaiense, C. furca, A. fraterculus, and C. polykrikoides formed serial red tides in the study area from May 7 to Nov 11 in 2014 (Fig. 4, redrawn from Jeong et al. 2017). In general, the dominant HTD genera during or after each red tide changed (Fig. 4); Gyrodinium spp. were abundant during the red tide dominated by P. donghaiense from Jun 23–Jul 1, Protoperidinium spp. during the C. furca, and C. polykrikoides red tide on Aug 6 and Sep 1, and N. scintillans and Polykrikos spp. during the C. polykrikoides red tide on Sep 15. The maximum abundances of Gyrodinium spp., Protoperidinium spp., N. scintillans, and Polykrikos spp. were 41.9, 11.0, 30.7, and 78.8 cells mL⁻¹, respectively.

The spatial distributions of Gyrodinium spp., Protoperidinium spp., N. scintillans, and Polykrikos spp. differed from one another at most sampling times (Fig. 5); Gyrodinium spp. were highly abundant in all three parts of the study area (i.e., the eastern part from May 7 to Jun 5, the western part from Jun 12 to 23, and the middle part on Jul 1); Protoperidinium spp. were highly abundant in the western or middle parts at most sampling times; N. scintillans was most abundant in the middle part of the study area; and Polykrikos spp. were abundant in all three sections (from Jun 12–Jul 1 in the western part, from Jul 11–Aug 6 in the middle part, and on Sep 15 in the eastern part) of the study area.

There was a significantly positive correlation between TCs abundance and water temperature (T), whereas NCs abundance was significantly negatively correlated with T (Table 2) but significantly positively correlated with salinity (S) (Table 2). No correlations were found between HTDs abundance and T or S (Table 2).

The abundances of TCs, NCs, and HTDs were significantly and positively correlated with total phytoplankton, diatom, and phototrophic dinoflagellate abundances (Table 2). Furthermore, Polykrikos spp. abundances were significantly positively correlated with C. polykrikoides abundance, whereas Gyrodinium spp. abundances were significantly positively correlated with P. donghaiense abundance. In addition, the abundances of Protoperidinium spp., N. scintillans, TCs, and NCs were significantly positively correlated with total diatom and P. donghaiense abundances (Table 2).

When the abundances of C. polykrikoides and P. kofoidii were 0.1–2,989.3 cells mL⁻¹ and 0.0–27.0 cells mL⁻¹, respectively, the calculated g attributable to P. kofoidii on co-occurring C. polykrikoides were up to 7.02 d⁻¹ (i.e., up to 99% of the population C. polykrikoides were consumed in 1 d) (mean ± SE = 0.06 ± 0.01 d⁻¹, n = 1,682) (Table 3, Fig. 6). In addition, when the abundances of C. polykrikoides and NCs (>100 μm) were 0.1–2,989.3 cells mL⁻¹ and 0.0–7.4 cells mL⁻¹, respectively, the g attributable to NCs (>100 μm) on co-occurring C. polykrikoides were up to 12.92 d⁻¹ (i.e., up to 99% of the population C. polykrikoides were consumed in 1 d) (mean ± SE = 0.21 ± 0.02 d⁻¹, n = 1,682) (Table 3, Fig. 7).

When the abundances of P. donghaiense and G. dominans/G. moestrupii were 0.1–4,369.1 cells mL⁻¹ and 0.0–15.0 cells mL⁻¹, respectively, the g attributable to G. dominans/G. moestrupii on co-occurring P. donghaiense were up to 0.58 d⁻¹ (i.e., up to 44% of the population P. donghaiense were consumed in 1 d) (mean ± SE = 0.02 ± 0.00 d⁻¹, n = 868) (Table 4). When the abundances of P. donghaiense and N. scintillans were 0.1–4,369.1 cells mL⁻¹ and 0.0–30.7 cells mL⁻¹, respectively, the g attributable to N. scintillans on co-occurring P. donghaiense were up to 1.70 d⁻¹ (i.e., up to 82% of the population P. donghaiense were consumed in 1 d) (mean ± SE = 0.02 ± 0.01 d⁻¹, n = 868) (Table 4). Furthermore, when the abundances of P. donghaiense and NCs (>100 μm) were 0.1–4,369.1 cells mL⁻¹ and 0.0–7.2 cells mL⁻¹, respectively, the g attributable to NCs (>100 μm) on co-occurring P. donghaiense were up to 13.12 d⁻¹ (i.e., up to 99% of the population P. donghaiense were consumed in 1 d) (mean ± SE = 0.40 ± 0.04 d⁻¹, n = 868) (Table 4).

When the abundances of C. furca and P. kofoidii were 0.1–400.4 cells mL⁻¹ and 0.0–15.4 cells mL⁻¹, respectively, the g attributable to P. kofoidii on co-occurring C. furca were up to 4.13 d⁻¹ (i.e., up to 98% of the population C. furca were consumed in 1 d) (mean ± SE = 0.02 ± 0.00 d⁻¹, n = 1,304) (Table 5).

When the abundances of A. fraterculus and co-occurring G. dominans/G. moestrupii were 0.1–1,276.0 cells mL⁻¹ and 0.0–15.0 cells mL⁻¹, respectively, the g attributable to the HTD G. dominans/G. moestrupii on co-occurring A. fraterculus were up to 2.00 d⁻¹ (i.e., up to 87% of the population A. fraterculus were consumed in 1 d) (mean ± SE = 0.06 ± 0.01 d⁻¹, n = 732) (Table 6).

Significantly positive correlations between HTDs, TCs, and NCs and total phytoplankton abundance suggest that the abundances of these heterotrophic protists are affected by total phytoplankton abundance in the study area. However, the dominant heterotrophic protist genera or taxa differed when the red tide causative species changed; TCs and/or NCs were dominant during or after diatom red tides; Gyrodinium spp. were dominant during or after P. donghaiense red tides; Protoperidinium spp. were dominant during or after C. furca, A. fraterculus, and/or C. polykrikoides red tides; and N. scintillans and Polykrikos spp. were dominant during or after C. polykrikoides red tides. High abundances of TCs and/or NCs during or after diatom red tides, Protoperidinium spp. during or after C. furca red tides, and N. scintillans and Polykrikos spp. during or after C. polykrikoides red tides have been previously reported to occur in Korea, India, and Norway (Jeong et al. 2000, Olseng et al. 2002, Biswas et al. 2013, Yoo et al. 2013a, National Fisheries Research & Development Institute 2014). Here, however, we report for the first time that Gyrodinium spp. are abundant during or after P. donghaiense red tides and that Protoperidinium spp. are abundant during or after A. fraterculus and/or C. polykrikoides red tides. Although Gyrodinium spp. and Protoperidinium spp. are known to feed on P. donghaiense and C. polykrikoides, respectively (Cho 2006, Gu et al. 2014), no study of Protoperidinium spp. feeding on A. fraterculus has yet been published. It is worthwhile to explore feeding by Protoperidinium spp. on A. fraterculus because field data suggest the existence of potential predator-prey relationships between heterotrophic and phototrophic dinoflagellates.

There was a 2-week time lag between the peak abundances of C. polykrikoides (Sep 1) and Polykrikos spp. (Sep 15). C. polykrikoides abundance was 3,000 cells mL⁻¹ on Sep 1 and 335 cells mL⁻¹ on Sep 15 (Jeong et al. 2017), whereas that of Polykrikos spp. on the same dates was 2.5 and 78.8 cells mL⁻¹, respectively. In a laboratory feeding experiment in which concentrations of C. polykrikoides were increased, P. kofoidii grew rapidly at mean prey concentrations ≤ ca. 300 cells mL⁻¹ but became saturated at higher prey concentrations (Kim 2004). Thus, theoretically, the abundance of C. polykrikoides from Sep 1 to 15 was likely to support positive growth of Polykrikos spp. The growth rate of Polykrikos spp. feeding on C. polykrikoides from Sep 1 to 15 was estimated to be 0.25 d⁻¹, which is 23% of the maximum growth rate of P. kofoidii on C. polykrikoides obtained in the laboratory by Kim (2004). Many environmental factors, such as water temperature, turbulence, predators, and competition, may lower the growth rate. However, our results suggest that high C. polykrikoides abundance may result in increasing abundances of Polykrikos spp. during C. polykrikoides red tides.

The maximum abundance of all HTDs observed during the study period was 82 cells mL⁻¹, a level comparable to or higher than that measured in Sandsfjord (Norway), the Seto Inland Sea (Japan), Gyeonggi Bay (Korea), Chesapeake Bay (USA), Chile, and the Pearl River Estuary (Hong Kong, China) (Table 7). The maximum chl-a concentration observed during the study period (44.2 μg L⁻¹) was also comparable to or higher than those in other regions of the world (Table 7). In general, the growth rate of HTDs rapidly increased with increasing concentrations of suitable algal prey species before becoming saturated (i.e., Lim et al. 2014b). Thus, high chl-a concentrations during red tides dominated by suitable prey species in this region may be partially responsible for the comparable or higher maximum abundance of total HTDs. However, the maximum abundance of total HTDs in this area was found to be considerably lower than in Masan Bay (Korea) and Albemarle-Pamlico Estuary (USA), as was the maximum concentration of chl-a (Table 7). Thus, lower levels of chl-a during red tides may be partially responsible for the lower maximum abundance of total HTDs in the South Sea. Although maximum abundances differed substantially between the study area and Masan Bay and Albemarle-Pamlico Estuary, the differences between the maximum biomasses may be much smaller. The dominant HTDs in Masan Bay and Albemarle-Pamlico Estuary during periods of maximal HTD abundance were Pfiesteria and Pfiesteria-like dinoflagellates (PLDs, including species in the genera Luciella, Stoekeria, and Cryptoperidiniopsis) (Glasgow et al. 2001, Yoo et al. 2013a), whereas the dominant HTDs in the study area were Polykrikos spp. The maximum biomass of Polykrikos spp. (3.6 ng C cell⁻¹) in the study area was estimated to be 280 ng C mL⁻¹, which is ~15% of the maximum biomass of HTDs in Masan Bay (1,916 ng C mL⁻¹) (Yoo et al. 2013a). The equivalent spherical diameter of Polykrikos spp. (ca. 44 μm) is much larger than that of PLDs (~13–14 μm) (Jeong et al. 2010). Furthermore, the types of prey species that Polykrikos spp. feed on differ from those consumed by PLDs; engulfment-feeding Polykrikos spp. are known to consume large algal prey like the phototrophic dinoflagellates Lingulodinium polyedrum, Scrippsiella trochoidea, C. furca, C. polykrikoides, Gymnodinium catenatum, Gymnodinium impudicum, and Prorocentrum micans (Jeong et al. 2001, Kim 2004), whereas peduncle-feeding PLDs prefer smaller algal prey such as the raphidophyte Heterosigma akashiwo, the cryptophytic Teleaulax sp., Rhodomonas salina, and Rhodomonas sp., and the phototrophic dinoflagellate Amphidinium carterae (Jeong et al. 2005, 2006, 2007, Baek et al. 2010, Lim et al. 2014b). Therefore, the causative species of red tides in these waters may support different dominant HTDs, resulting in differences in maximum abundances.

The maximum abundance of total TCs in the South Sea during the study period (49 cells mL⁻¹) was higher than in the Seto Inland Sea (Japan), North Irish Sea (Ireland), Buzzards Bay, Long Island Sound, and Chesapeake Bay (USA), Sandsfjord (Norway), and the Gulf of Naples (Italy) (Table 7). The maximum concentration of chl-a in the study area was also higher than in other regions of the world (Table 7). On the other hand, the maximum abundance of total TCs in the study area was lower than in the Coastal North Sea and Masan Bay, and the maximum concentration of chl-a in the South Sea was also lower than in the Coastal North Sea and Masan Bay (Table 7). Thus, higher or lower chl-a concentrations during or after red tides in the study area may be also partially responsible for the higher or lower maximum abundance of total TCs.

The maximum abundance of total NCs (35 cells mL⁻¹) was higher in the South Sea than in Long Island Sound and Chesapeake Bay (USA), Sandsfjord (Norway), and the Seto Inland Sea (Japan) (Table 7), as was the maximum chl-a concentration (Table 7). However, the maximum abundance of total NCs was considerably lower in the study region than in Masan Bay (Korea) (Table 7). Moreover, the maximum concentration of chl-a and the maximum abundance of diatoms in the study area (44.2 μg L⁻¹ and 13,020 cells mL⁻¹, respectively) (Jeong et al. 2017) were also much lower than in Masan Bay (514.7 μg L⁻¹ and 71,538 cells mL⁻¹, respectively) (Jeong et al. 2013). According to Yoo et al. (2013a), total NC biomass was significantly positively correlated with diatom abundance in Masan Bay. Thus, the lower maximum abundance of diatoms in the study area may explain why the maximum abundance of total NCs was lower in the South Sea than in Masan Bay.

The maximum g by dominant heterotrophic protists on A. fraterculus, C. furca, C. polykrikoides, and P. donghaiense mostly exceeded 2.0 d⁻¹ (i.e., ~87% of the population of a given red tide species was consumed in 1 d), but the mean g were <0.4 d⁻¹ (i.e., ~33% of the population of a given red tide species was consumed in 1 d). Therefore, the dominant heterotrophic protists may at times have considerable grazing impacts on populations of these red tide species. The very high g (i.e., >5 d⁻¹) may be caused by a combination of low red tide species abundance and high dominant heterotrophic protist abundance during the declining stages of red tides.

The maximum g attributable to dominant heterotrophic protists on co-occurring C. polykrikoides and C. furca estimated in this study were higher than those on the same prey species in other regions (Table 8); for instance, the maximum g of P. kofoidii on C. polykrikoides in the present study (7.02 d⁻¹) was higher than reported for Tongyeong and Masan Bay, Korea (1.21–2.64 d⁻¹) (Kim 2004, Yoo et al. 2013a). In addition, the maximum g for P. kofoidii on C. furca in the present study (4.13 d⁻¹) was much higher than observed in Masan Bay (1.20 d⁻¹) (Yoo et al. 2013a). The abundances of P. kofoidii during C. polykrikoides and C. furca red tides in the study region were higher than those in Tongyeong and Masan Bay. Therefore, higher abundances of P. kofoidii during C. polykrikoides or C. furca red tides may be partially responsible for the higher maximum g’s of P. kofoidii on co-occurring C. polykrikoides and C. furca in the South Sea than in Tongyeong and Masan Bay.

To the best of our knowledge, no studies have presented data on the grazing impacts of heterotrophic protists on P. donghaiense and A. fraterculus in other regions despite their having caused red tides in the waters of many countries (Moncheva et al. 2001, MacKenzie et al. 2004, Tang et al. 2006, Omachi et al. 2007, Nagai et al. 2009, Hu et al. 2012, Park et al. 2013a). Thus, it is difficult to compare g’s by heterotrophic protists on P. donghaiense and A. fraterculus in the South Sea to those in other waters. The maximum g’s of G. dominans/G. moestrupii, N. scintillans, and NCs (>100 μm) on P. donghaiense and A. fraterculus in our study were high, as mentioned above. Thus, it would be useful to explore the g’s of heterotrophic protists on P. donghaiense and A. fraterculus in other regions of the world.

Overall, the dominant heterotrophic protist groups and their abundances in Korea’s South Sea varied considerably both spatially and temporally over the course of the study period; such variability was associated with changes in the dominant red tide species or groups. Furthermore, the maximum g’s of the dominant heterotrophic protists on the red tide species or groups were extremely high. Thus, red tide species, as prey, play important roles in the dynamics of the dominant heterotrophic protists, and in turn the dominant heterotrophic protists, as predators, play important roles in the dynamics of red tide organisms in marine ecosystems.